SOME BASIC PROBLEMS OF THE

MATHEMATICAL THEORY OF
ELASTICITY
Fundamental Equations Plane Theory of Elasticity Torsion
and Bending

Edited by

N. I. Muskhelishvili

、 Springer-Science+
Business Media, B.V.
SOME BASIC PROBLEMS OF THE MATHEMATICAL
THEORY OF ELASTICITY
SOME BASIC PROBLEMS OF THE
MATHEMATICAL THEORY OF
ELASTICITY
FUNDAMENTAL EQUATIONS
PLANE THEORY OF ELASTICITY
TORSION AND BENDING

BY

N.I. MUSKHELISHVILI

FOURTH, CORRECTED AND AUGMENTED EDITION

MOSCOW 1954

TRANSLATED FROM THE RUSSIAN

BY

J. R. M. RADOK

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

 © 1977 Springer Science+Business Media Dordrecht
Originally published by Noordhoff International Publishing in 1977

4〃rights reserved. No 力口力 。/ this may be reproduced, stored 加

a retrieval system, or transmitted, £% any form by any means, electronic,
mechanical, -photocopying, recording otherwise without the prior permission
”
o/ the copyright owner.
ISBN 978-90-481-8245-9 ISBN 978-94-017-3034-1 (eBook)
DOI 10.1007/978-94-017-3034-1

Reprint of the second English edition

AUTHOR'S FOREWORD TO THE SECOND ENGLISH EDITION

I was pleased to learn that the need has arisen for a second edition of
my book, which will be based on the fourth Russian edition of 1954.
It is for me a plesant opportunity to express my deep gratitude to
Dr. J. R. M. Radok, on whose initiative the English edition of this book
has been published, for his beautiful translation which considerably
assisted in the spreading of the methods and results studied in this book
beyond the boundaries of the USSR.
I likewise wish to thank Professor I. S. Sokolnikoff of Los Angeles,
U.S.A, who even before the publication of the English edition of my book
studied a number of these results in his widely known course of the
theory of elasticity which also greatly helped foreign readers to come
into contact with this work.
Finally, I wish to express my heartfelt thanks to the publishing house
of P. Noordhoff for the magnificent appearance of the book.

Tbilisi, March !962. N. Muskhelishvili

 FOREWORD TO THE FIRST ENGLISH EDITION.

The appearance of an English translation of Muskhelishvili's prize¬

winning monograph on the mathematical theory of elasticity is certain
to be greeted with enthusiasm by Western European scholars. Although
a version of this monograph was first published in Russian some twenty
years ago, the general methods of solution of two-dimensional elastostatic
problems developed by Muskhelishvili and his co-workers still remain
relatively unknown outside Russia.
Mr. J. R. M. Radok and his Publishers, P. Noordhoff of Groningen,
have performed an extremely valuable service by making available to
Western scholars a translation of one of the most significant contributions
to the theory of elasticity since the publication of Love's celebrated
Treatise 0% the Mathematical Theory 。/ Elasticity.

Los Angeles, California, 1952. I. S. Sokolnikoff

 TRANSLATORS PREFACE
TO THE FIRST ENGLISH EDITION.

In preparing this translation, I have taken the liberty of including

footnotes in the main text or inserting them in small type at the appropriate
places. I have also corrected minor misprints without special mention. The
Chapters and Sections of the original text have been called Parts and
Chapters respectively, where the latter have been numbered consecutively.
The subject index was not contained in the Russian original and the
authors' index represents an extension of the original list of references.
In this way the reader should be able to find quickly the pages on which
any one reference is discussed. The transliteration problem has been
overcome by printing the names of Russian authors and journals also in
Russian type.
While preparing this translation in the first place for my own informa¬
tion, the knowledge that it would also become accessible to a large
circle of readers has made the effort doubly worthwhile. I feel sure that
the reader will share with me in my admiration for the simplicity and
lucidity of presentation.
I should like to thank Professor E. Sternberg of Illinois Institute
of Technology at Chicago for loaning me his copy of Muskhelishvili's
book through the good offices of Professor F. S. Shaw of the Brooklyn
Polytechnic Institute, New York, Professor W. S. Hemp of the College
of Aeronautics, Cranfield, England, for reading my translation in manu¬
script and Professor I. S. Sokolnikoff of the University of California for
reading the second proof and for fulfilling my wish by writing his foreword
to the book. Finally, I must express my gratitude to the publishers for
making the printing of this book possible and for their constant willingness
to meet all special requests.

Zurich, November 1952. J. R. M. Radok

 EXTRACT FROM THE AUTHORS PREFACE TO THE
FIRST RUSSIAN EDITION.

This book reproduces, in a considerably revised and enlarged form,

the contents of a course of lectures, delivered by me in Spring 1931 at
the invitation of the Seismological Institute of the Academy of Sciences
of the U.S.S.R. before the scientific workers of the Institute, and of
lectures delivered in 1932 before post-graduate students of the Physico-
Mathematical Institute of Mathematics and Mechanics at the University
of Leningrad. The lectures were intended for persons acquainted with the
principles of the theory of elasticity and were to be devoted to separate
fundamental questions the choice of which was largely left to me; I
naturally dwelt on subject matter in which I had been working myself.
Thus, this book deals only with a few chapters of the theory of elasticity
each of which receives fairly complete treatment. I shall not touch here
on the subject matter of the book an idea of which may be gained from
the list of contents, but I consider it necessary to make the following
comments.
Seeing that the problems considered in this book may prove of interest
to a wider circle of people, in particular to those whose work requires
application of the theory of elasticity, I have tried to make the exposition
as far as possible intelligible for readers who are only familiar with the
fundamentals of differential and integral calculus and the elementary
theory of functions of a complex variable. Thus, for example, problems
involving integral equations are relegated to separate sections which
may be passed over without impairing the understanding of the re¬
mainder ; Part I which deals with the foundations of the mathematical
theory of elasticity (it contains even more than is required) is intended for
readers not specializing in the theory of elasticity. In order to make the
text more accessible, I refrained from employing tensor calculus which
I used in my lectures at the Seismological Institute ; an elementary
introduction to tensors is given in Appendix 1. Appendices 2 and 3 are
devoted to some aspects of elementary mathematics which are necessary
for the understanding of the subject matter of the book and which, as a
rule, are insufficiently elucidated in elementary courses on analysis.

Leningrad, Spring 1933. N. Muskhelishvili

 PREFACE TO THE THIRD RUSSIAN EDITION.

The second edition of this book, which was published in 1935 almost
immediately after the first (which appeared in 1933), has been out of
print for a long time, but, as I was engaged on other work, I have only
now been able to start the preparation of a new edition. The warm re¬
ception given to my book and the high distinction with which it was
favoured made it imperative to treat its reissue with special attention.
To this was added the circumstance, highly gratifying to me, that soon
after the appearance of the first editions many papers were published in
which the methods expounded by me were applied to different concrete
problems and also substantially amplified and generalized. It is natural
that the new edition should at least reflect the main results of these
papers as well as some results obtained by me. I have tried to accomplish
this, but I am afraid that some papers may still have escaped my notice
for which omission I tender my apologies to their authors.

The general design of exposition in this edition has remained the

same as before. However, the text of the book, with the exception of the
first two and last Parts *, has been thoroughly rewritten and considerably
enlarged. Two new Parts have been added, namely the fourth and the
sixth. The contents of the fourth Part are only to a negligible extent taken
from the preceding edition ; in Part VI, results are given which have been
obtained by me and also by other authors since the publication of the
preceding edition, ** if no account is taken of the few problems whose
solution had been given in the previous edition, but by means of other
methods.

* However, Chapter 25 has been considerably enlarged and in this way the
theory of extension and flexure of compound bars has attained an aspect of com¬
pleteness.
** A large part of these results were introduced by me into my book "Singular
Integral Equations" (Moscow-Leningrad 1946), but now I find that their natural
place is in the present book. They will accordingly be omitted from, the following
edition of "Singular Integral Equations". Here, the exposition of these results
has been rearranged in order to make it independent of the above-mentioned book.
X PREFACE

Although I do not think it possible or necessary to indicate all the

changes and additions in the text of the earlier editions, I must draw the
reader's attention to the new arrangement of Chapters 14—16 (Chapter
17 is new). Compared with the previous editions, the results contained in
these chapters are not new, but the method by which they have been
obtained has been replaced by a different one which seems tome to be more
germane to the matter under consideration. However, I should like
to mention that the new method (which was influenced by the work of
J. Plemelj on the theory of functions of a complex variable, published
long before the first edition of my book, but unfortunately unknown to
me at the time) leads to the same calculations as the former method.
For this reason and also because of the, say, greater complexity of the
new method, I am not certain that I acted rightly in introducing this
change. Be that as it may, a comparison of the new and the old approach
may prove useful.
In conclusion, I should like to add that as far as possible I have care¬
fully quoted the authors of any results which I have used, just as I have
done with regard to some of my own results, at times even of minor im¬
portance and adduced merely as examples. I have followed this practice,
not because I attach exaggerated importance to these results, but only
to avoid puzzling the reader who might not be acquainted with the
previous editions of my book and who might have encountered material
taken from it without clear indication of its original source in some other
publications (mostly non-Russian) .
To simplify reference, the quoted works have been listed at the end of
the book in alphabetical order. In references, the author is named and the
number of his publication, according to this list, is given in square brackets.
The first edition was greatly assisted to its success by the preface of the
late Academician Alexei Nikolaevich Krylov whose outstanding scientific
and public merits are well known to all and for whom I shall always
entertain feelings of profound gratitude and respect. Krylov's preface
is reproduced below without any changes. I have not been able to fulfil
in this edition the wish expressed by Krylov at the end of his preface
concerning the development of numerical methods of solution. While
realizing the importance of his request, I felt that I would be unable to
meet it sufficiently well. Nevertheless, Krylov's wish has been fulfilled
by other authors, referred to in the text of the book.

Tiflis, November 1948. N. Muskhelishvili

 PREFACE TO THE FOURTH RUSSIAN EDITION.

This edition differs from the preceding one by numerous additions and
changes in the text most of which are small, but at times rather essential.
In this respect the author had in mind the reader concerned mainly with
practical applications as well as the reader interested in the mathemati¬
cal aspects of the subject.
I wish to use this opportunity to express my sincere gratitude to
G. F. Mandjavidze who read with great interest the proofs of the entire
book, noted a number of slips and made several critical comments which
led to improvements of the study.
I also wish to thank A. Ia. Gorgidze who read the proofs of the first
and seventh chapters and performed certain computations.

Tbilisi, December 1953. N. Mukshelishvili

 PREFACE OF ACADEMICIAN A. N. KRYLOV TO THE
FIRST EDITION *

Elasticity is a fundamental property of all substances occurring . in

Nature. This property must be ascribed even to that imaginary ether
whose very existence is now acknowledged, now disputed by Physics, and
which at the same time is so widely used in practice.
Thus, it may be said that the elasticity of bodies has been unwittingly
used since prehistoric times for all kinds of buildings from the huts of
savages to the Coliseum and majestic palaces and temples, for all tools
and instruments of everyday life and for all kinds of weapons. As primeval
man or a contemporary Papuan would use the elasticity of his material
when making a bow for his flint or bone-headed arrow, so Vickers &
Armstrong rely on the very same property for their 150-ton-16-inch guns
for the battleships Nelson and Rodney or for the fortifications of Dover.
The only difference is that the Papuan makes his bow by methods arrived
at over countless generations as if by natural selection whereas Vickers &
Armstrong design their guns by methods based on mathematical
calculations developed by our own Academician Gadolin only 60 years
ago. If therefore man's utilization of the property of elasticity has gone
on from time immemorial, yet the first attempt at its scientific foun¬
dations only dates back 295 years and it was undertaken by Galilei who
described it in his famous Discourses, published in 1638.
Forty years later, the Englishman Hooke gave the solution to his
anagram "ceiiinosssttuv" : ut tensio sic vis, which contained the funda¬
mental law of the elasticity of materials.
After another 60 years, the works of Daniel Bernoulli and Euler
appeared in the Transactions of the St. Petersburg Academy of Sciences,
which gave the equations still used in most practical calculations.
Yet 80 years later, the French engineer Navier adopted quite a differ¬
ent approach to the study of the elasticity of matter by giving its general
theory and setting up the equations of equilibrium and of motion. This
theory was further developed by the famous mathematicians Cauchy
and Poisson who applied it first to the study of the propagation of waves
1

* The preface by Krylov has been reprinted without alterations from the first
edition. Since Krylov analyses the contents of the book Part by Part, it
should be noted that Part III of the first edition, considerably enlarged, constitutes
Part V, while Part IV of the first edition corresponds to Part VII of the present
edition.
XIV PREFACE

in elastic media, i.e., of sound, since they did not admit at that time the
undulatory theory of light ; in his course of mechanics which Poisson
gave at the Ecole Polytechnique he developed the theory of Bernoulli and
Euler.
At the same time, at the Institute of Communication Engineers
founded a few years earlier at St. Petersburg, there were two French
engineers, Lame and Clapeyron, who as professors held the rank of
lieutenant-colonel in the Russian Army. They turned towards developing
Navier's theory for the purpose of applying it to building and con¬
struction ; however, soon after his accession, Nicholas I dismissed them
from the Russian service and sent them home.
Here, in 1852, Lame also published the first treatise on the theory of
elasticity under the title "Legons sur la theorie mathematique de
l'61asticit6 des corps solides" which has become a classic and has not
lost its significance even today.
Together with the development of railway construction, of steel
railway bridges, of huge buildings, with the development of machine
construction and shipbuilding, with the setting up of laboratories for the
scientific investigation of the elastic properties of materials and for their
testing on qualities stipulated by contract went the rapid development
of the study of the elastic properties of materials and their proper utili¬
zation in building practice.
This development followed two trends: on the one hand, the mathe¬
matical theory of elasticitity was worked out in the form in which it had
been created by Navier and first set out by Lame, on the other hand, a
simplified theory was developed which was given the name of the Theory
of Strength of Materials and which was based on methods, the foun¬
dations of which were laid by Bernoulli and Euler.
It is remarkable that the formulae and conclusions of Lame had their
first practical application on the initiative of Gadolin in the project of
steel guns reinforced by rings which were first produced by Krupp. It
was only then that the Obukhovsky plant was founded. Frequently,
practice itself has required the examination of the conclusions of the
simplified theory by the more severe and exact theory, so that it might
approximate more closely to reality and to ensure that the simplifications
did not lead too far away from the truth and that, on their account, the
very point of the matter was not disregarded.
In many cases, research has been stimulated by notable collapses of
gigantic railway bridges, correctly constructed, one would have thought,
 PREFACE XV

and calculated according to all the rules and norms, or by cracks and
rents in the deck plates of ocean steamers or by cases of loss of fast liners
which, after collision with small vessels, broke in two and sank. In the
last case, facts about the fractures could be clearly judged by the different
inclinations of the masts before they finally submerged.
The theory of elasticity would often help and give an exact quantitative
explanation of the causes of the accident, and once the causes were
known it was not difficult to find the means to prevent the harmful
consequences.
About thirty years ago reinforced concrete began to be used for
building purposes ; this new application rapidly developed and has now
attained vast importance. At that juncture, new problems arose and for
the solution of many of them the simplified methods of Bernoulli and
Euler were ineffective, so that the theory of elasticity found a number of
new applications.
Hence it is clear how important the study of this subject is at present
with our tremendous building activity in its countless different forms,
and therefore N. I. Muskhelishvili's book fills an essential need.
In his preface, the author refers the reader, wishing to acquaint himself
with the subject matter of the book, to its list of contents, but the dry
list gives an insufficient idea not only of the method of exposition, but
even of its very contents. I shall therefore try to give a brief introduction
to the nature of the exposition, noting, above all, its originality, thanks
to which a conciseness has been obtained which enabled the author to
present in his small book such vast, and to a great extent, new material
and at the same time to preserve full clarity of treatment.

Professor Muskhelishvili's book contains four Parts:

Part I. Fundamental equations of the mechanics
of elastic bodies.
Here, in 75 pages, all the general principles of the theory of elasticity
are set out, viz., a) Analysis of stress, b) Analysis of strain, c) Relation
between stress and strain, 〃) The equilibrium equations of an elastic body
and the two fundamental problems : 1) to determine the state of stress of
a body when the forces acting on it are given, and 2) to determine the
state of stress of a body when the displacements of points of the boundary
of the body are given.
XVI PREFACE

Part II Plane Problem. General formulae and

elementary applications
These 100 pages deal with the plane problem and show the main
methods for its solution. The solution is achieved by the help of the stress
function and its complex representation ; first, an account of the general
theory of methods is given, and then the methods are practically de¬
veloped in a number of examples. Of these examples we shall note a)
tension of a plate weakened by a circular hole, 6) effect of a concentrated
force applied at a point of an unbounded plane, c) effect of a concentrated
couple, d) examination, of the state of stress in a ring under the action of
given forces, e) flexure of a circular beam, /) general theory of thermal
strains and stresses.
Part III Application of conformal mapping and
complex integration to the plane problem
These 108 pages contain the theory and examples of conformal mapping
and of its application to the transformation of equations of the plane
problem and of its boundary conditions. After that, the general method
of solution of the fundamental problems is developed and the solution of
these problems is illustrated by the example of the continuous ellipse.
Next, the author develops the theory of Cauchy integrals and gives the
generalized formulae which he employs further on.
Having derived the general solution of the fundamental problems for
regions bounded by one contour and having reduced this solution to that
of Fredholm equations, the author gives the solution of many examples
for regions of different form; in particular, he examines the solution of
the fundamental problems for the plane with elliptic perforations and
with nuclei ; after that, he gives the solution of both fundamental problems
for the half-plane and for regions of a more general character.
All these solutions are obtained, not by fortuitous particular methods,
but by the application of a common method based on properties of
Cauchy integrals discovered by the author.
When reading this chapter, I inadvertently remembered the Spring
meeting of the Society of Shipbuilding Engineers in 1898. At this meeting,
Professor Hele-Shaw showed, for the first time, the equipment by which he
projected on a screen, with marvellous distinctness, the flow of a liquid
and showed how the streamlines bypassed different kinds of obstacles.
In the following year, I was again in London at the Spring meeting of the
 PREFACE XVII

Society. Among other papers, the Norwegian engineer Bruhn read a paper
on the influence of perforations and apertures in the deck on the overall
strength of vessels, and since, not long before, the huge steamer "City of
Rome" broke up after a collision with a sailing ship a few miles off New
York, and since this catastrophe was still fresh in everybody's mind,
Bruhn's paper was received with special interest. To study this influence,
he took an oblong sheet of rubber, ruled it with straight lines running
parallel and perpendicular to its long side thus forming squares and made
perforations in it of various forms ; having stretched the sheet lengthwise,
he traced the form of the curves into which the originally straight lines
on the sheet had been transformed. From these lines he obtained a picture
of the strain distribution, and consequently of the stress distribution.
By use of similar models, Bruhn proposed to study the solutions of the
plane problems which N. I. Muskhelishvili has solved so brilliantly
analytically.
Accidentally, one of Bruhn's perforations had the same shape as one of
the obstacles in Hele-Shaw's experiments, demonstrated the year before.
There was almost complete identity between Bruhn's curves and Hele-
Shaw's streamlines. Since nobody drew attention to this fact, I obtained
the Reports of the Society for the preceding year, requested leave to be
heard and explained that the coincidence was by no means accidental,
that Bruhn's method was the mechanical and Hele-Shaw's method the
hydrodynamical solution of the same generalized problem of Dirichlet and
that there was no need for Bruhn's complicated models and measurements
nor for the tracing of the curves obtained ; in fact, one had only to insert an
obstacle of a corresponding form into Hele-Shaw's apparatus and to take a
photograph by projecting them on ordinary sensitive paper, in order to
obtain a map of the strain distribution. Such a j uxtaposition of phenomena
from quite different fields came altogether unexpectedly to the meeting
and its old chairman, the highly experienced engineer Benjamin Martel,
made special acknowledgement of my remarks in his final address.
The flow of gases and liquids is now studied by many methods which
are modifications of that of Hele-Shaw as well as of others, because
of its considerable significance in aerodynamics. It might possibly be
a worthwhile task to compare such an "aerodynamic spectrum” with the
solutions given by N. I. Muskhelishvili, and hence to elaborate an experi¬
mental method for solving the plane problems of the theory of elasticity.
Another method which has been so splendidly developed by Acade¬
mician N. N. Pavlovsky happens to come to my mind, namely the electri-
XVIII PREFACE

cal solution of the hydrodynamic problem of determining equipotential

and stream lines on a conducting plate with perforations of given shape.
All this represents a realisation of conformal transformations, so
elegantly developed in analytic form by Muskhelishvili, and shows the
interrelationship by way of common differential equations between
apparently quite disconnected branches of physics.
In this context, one might usefully remember that the stress function
itself was introduced into the theory of elasticity by the famous Astrono¬
mer Royal Sir James Biddel Airy who, I believe, was director of Green¬
wich Observatory for more than 50 years. At the beginning of the 1860's,
he built for the Observatory a new large meridian line with a telescope
having an 8-inch object lens. He had to count with the flexure of the
telescope under the weight of the lens and ocular and of other devices,
a fact which had caused errors up to 2 arc seconds at the Paris Observato¬
ry, errors which are inadmissible in such accurate observations with
significant measurements in decimal seconds.

Part. IV Torsion and flexure o f homogeneous and

compound bars
This Part provides an excellent account of Saint-Venanfs problem
and also studies compound bars, so important for reinforced concrete
construction, by entirely new methods developed by the author.

Even at a cursory reading of the book, the originality of the solutions

is evident not only as far as the quite new problems proposed by the author
are concerned, but also with regard to problems that have long ago been
solved by other authors. Thus, in Part I, which by the nature of its contents
deals with well established results, the author elucidates many questions
more fully and distinctly than has been done before ; for example, the
new deduction of the compatibility conditions of Saint-Venant and
Beltrami-Michell belongs to the author. In Part II, the author offers a
new strict deduction of G. V. Kolosov's formulae and of a number of
other formulae ; the detailed study of the analytic character of the solu¬
tions of the plane problem for multiply connected and infinite regions
enables us to discover fallacious conclusions committed by other investi¬
gators.
Also of significance is the author's general investigation of the effect of
concentrated forces and the presentation of their most general ex-
 PREFACE XIX

pressions ; this again helped to reveal mistakes in other papers. The

establishment of the connection between thermal stresses and multi-valued
displacements also belongs to Muskhelishvili.
All examples in this Part have either been solved by the author for
the first time or, if solved before, more complicated methods had been
applied for their solution.
Part III belongs entirely to the author in the originality and in the
generality of the problems solved in it as well as in the originality of the
methods applied. How important this method is may be seen from the
fact that in § 68 (§ 82 of the 3rd edition) the author gives the general
solution of the second fundamental problem for an infinite plate with an
elliptic hole on two pages of large print. A particular case of this problem
was solved by L. FOppl in the Zeitschrift fiir angewandte Mathematik
und Mechanik, his solution occupying five large pages of small print
which, if set up in our academic type, would fill about twenty pages ; in
§ 69 (§ 82a of the 3rd edition) , an example is solved in a few lines the
simplest particular case of which agrees with that of Foppl.
In Part IV, as previously mentioned, all matter relating to non-
homogeneous bodies, beginning with the very statement of the problem,
belongs to Muskhelishvili.
From this short sketch may be seen the rich content as well as the
variety and importance of the problems covered in this book and the
originality and generality of the methods applied for their solution.
There only remains to express the wish that in future editions, which
without doubt will be required, the author illustrate the general de¬
ductions and formulae by numerical examples, by diagrams and by
indications as to the number of ordinates or subdivisions required for
approximate integration in order to ensure accuracy within, say, -%%. He
will thereby render a great service to engineers and make his excellent
book more accessible to those people who will apply its deductions to the
solution of the purely practical problems of the building industry.
Academician A. Krylov
A
 CONTENTS

Part I - Fundamental equations of the mechanics of an elastic body. 1

Chapter 1. Analysis of stress.

§ 1. Body forces 5
§ 2. Stress 6
§ 3. Components of stress. Dependence of stress on the
orientation of the plane 7
§ 4. Equations, relating components of stress 10
§ 5. Transformation of coordinates. Invariant quadratic
form. Stress tensor 14
§ 6. Stress surface. Principal stresses 17
§ 7. Determination of principal stresses and axes .... 22
§ 8. Plane stress 23

Chapter 2. Analysis of strain.

§ 9. General remarks 28
§ 10. Affine transformation 29
§ 11. Infinitesimal affine transformation 31
§ 12. Decomposition of infinitesimal transformations into
pure deformation and rigid body motion 32
§ 13. The invariant quadratic form, connected with defor¬
mation. The strain surface, principal axes. Trans¬
formation of coordinates 38
§ 14. General deformation 41
§ 15. Determination of displacements from components of
strain. Saint -Venanfs condition of compatibility . . 44

Chapter 3. The fundamental law of the theory of elastici¬

ty; THE BASIC EQUATIONS.

§ 1 6. The fundamental law of the theory of elasticity (gener¬

alized Hooke's Law) 52
§ 17. Isotropic bodies ' 56
XXII CONTENTS

§ 18. The basic equations for the statics of an elastic iso¬

tropic body 60
§ 19. The simplest cases of elastic equilibrium. The basic
elastic constants 62
§ 20. The fundamental boundary value problems of static
elasticity. Uniqueness of solution 66
§ 21. Basic equations in terms of displacement components 73
§ 22. Equations in terms of stresses 74
§ 23. Remarks on the effective solution of the fundamental
problems. Saint-Venanfs principle 77
§ 24. Dynamic equations. The fundamental problems of the
dynamics of an elastic body 78
Part II - General formulae o/ the plane theory o/ elasticity .... 85
Chapter 4. Basic equations of the plane theory of elas¬
ticity.

§ 25. Plane strain 89

§ 26. Deformation of a thin plate under forces acting in its
plane 92
§ 27. Basic equations of the plane theory of elasticity ... 96
§ 28. Reduction to the case of absence of body forces ... 101
Chapter 5. Stress function. Complex representation of
THE GENERAL SOLUTION OF THE EQUATIONS OF THE PLANE
THEORY OF ELASTICITY.

§ 29. Stress function 105

§ 30. Determination of displacements from the stress function 1 07
§ 31. Complex representation of biharmonic functions . . 1 10
§ 32. Complex representation of displacements and stresses 1 13
§ 33. The physical meaning of the function f. Expressions for
the resultant force and moment 1 16
§ 34. Arbitrariness in the definition of the introduced
functions 1 18
§ 35. General formulae for finite multiply connected regions 121
§ 36. Case of infinite regions 126
§ 37. Some properties following from the analytic character
of the solution. On analytic continuation across a
given contour 131
§ 38. Transformation of rectilinear coordinates 137
 CONTENTS XXIII

§ 39. Polar coordinates 140

§ 40. The fundamental boundary value problems. Uniqueness
of Solution 141
§ 41. Reduction of the fundamental problems to problems of
complex function theory 147
§ 41a Supplementary remarks 156
§ 42. Concept of the regular solution. Uniqueness of a regular
solution 158
§ 43. On concentrated forces, applied to the boundary ... 162
§ 44. Dependence of the state of stress on the elastic constants 1 64
Chapter 6. Multi-valued displacements. Thermal stresses.
§ 45. Multi-valued displacements. Dislocations 167
§ 46. Thermal stresses 170
Chapter 7. Transformation of the basic formulae for con¬
formal mapping.

§ 47. Conformal transformation 176

§ 48. Simple examples of conformal mapping.
1°. Bilinear function 180
2°. Pascal's limaqon 185
3°. Epitrochoids 186
4°. Hypotrochoids 187
5°. Elliptic rings 188
§ 49. Curvilinear coordinates, connected with conformal
transformations into circular regions 190
§ 50. Transformation of the formulae of the plane theory of
elasticity 192
§ 51. Boundary conditions in the image regions 194

Part III - Solution of several problems of the plane theory of elasticity

by means of power series 197
Chapter 8. On Fourier series.
§ 52. On Fourier series in complex form 199
§ 53. On the convergence of Fourier series 202
Chapter 9. Solution for regions, bounded by a circle.

§ 54. Solution of the first fundamental problem for the

circle 204
XXIV CONTENTS

§ 55. Solution of the second fundamental problem for the

circle 207
§ 56. Solution of the first fundamental problem for the
infinite plane with a circular hole 208
§ 56a. Examples.
1°. Uni-directional tension of a plate, weakened by a
circular hole 21 1
2°. Bi-axial tension 214
3°. Uniform normal pressure applied to the edge of a
circular hole 214
4°. A concentrated force, applied at a point of the
infinite plane 215
5°. Concentrated couple 216
§ 57. On the general problem of concentrated forces ... 217
§ 58. Some cases of equilibrium of infinite plates, containing
circular discs of different material 221
1°. Infinite plate with a circular hole into which an
elastic circular disc with an originally larger radius
has been inserted 222
2°. Stretching of plates with inserted or attached rigid
discs 224
3°. Stretching of plates with inserted or attached
elastic discs 226

Chapter 10. The circular ring.

§ 59. Solution of the first fundamental problem for the

circular ring 230
§ 59a. Examples.
1°. Tube subject to uniform external and internal
pressures 235
2°. Stress distribution in a ring, rotating about its
centre 236
§ 60. Multi-valued displacements in the case of a circular ring 237
§ 61. Supplement. Bending of a curved beam 242
§ 62. Thermal stresses in a hollow circular cylinder .... 246
 CONTENTS XXV

Chapter 11. Application of conformal mapping.

§ 63. Case of simply connected regions 250

§ 64. Example of application of mapping on to a circular ring.
Solution of the fundamental problems for a continuous
ellipse 257
Part IV - 。处 Cauchy integrals 265
Chapter 12. Fundamental properties of Cauchy integrals.

§ 65. Notation and terminology 267

§ 66. Cauchy integrals 270
§ 67. Values of Cauchy integrals on the path of integration.
Principal value 271
§ 68. Boundary values of Cauchy integrals. The Plemelj
formulae 276
§ 69. The derivatives of Cauchy integrals 279
§ 70. Some elementary formulae, facilitating the calculation
of Cauchy integrals 281
§ 71. On Cauchy integrals, taken along infinite straight lines 286
§ 72. On Cauchy integrals, taken along infinite straight lines
(continued) 296
Chapter 13. Boundary values of holomorphic functions.

§ 73. Some general propositions 298

§ 74. Generalization 301
§ 75. Harnack's theorem 301
§ 76. Some special formulae for the circle and the half-plane 303
§ 77. Simple applications: solutions of the fundamental
problems of potential theory for a circle and half-plane 308
Part V - Application o/ Cauchy integrals to the solution o/ boundary
problems of plane elasticity 315
Chapter 14. General solution of the fundamental pro¬
blems FOR REGIONS BOUNDED BY ONE CONTOUR.

§ 78. Reduction of the fundamental problems to functional

equations 317
§ 79. Reduction to Fredholm equations. Existence theorems 323
§ 79a. On some other applications of the preceding integral
equations 333
XXVI CONTENTS

Chapter 15. Solution

of the fundamental problems for
REGIONS MAPPED ON TO A CIRCLE BY RATIONAL FUNCTIONS.
Extension to approximate solution for regions of
GENERAL SHAPE.
§ 80. Solution of the first fundamental problem for the circle 334
§ 80a. Examples.
1°. Circular disc under concentrated forces, applied to
its boundary 338
2°. Disc under concentrated forces and couples acting
at internal points 342
3°. Rotating disc with attached discrete masses . . . 345
§ 81. Solution of the second fundamental problem for the
circle 346
§ 82. Solution of the first fundamental problem for the
infinite plane with an elliptic hole 347
§ 82a. Examples.
1°. Stretching of a plate with an elliptic hole ... 351
2°. Elliptic hole the edge of which is subject to uniform
pressure 353
3°. Elliptic hole the edge of which is subject to uniform
tangential stress T 354
4°. Elliptic hole (or straight cut) part of the edge of
which is subject to uniform pressure 354
5°. Approximate solution of the problem of bending
of a strip (beam) with an elliptic hole 358
§ 83. Solution of the second fundamental problem for the
infinite plane with an elliptic hole 361
§ 83a. Examples.
1°. Uni-directional tension of an infinite plate with a
rigid elliptic centre 363
2°. Case when the elliptic centre is not allowed to rotate 365
3°. Case when a couple with given moment acts on the
elliptic kernel 366
4°. Case when a force acts on the centre of the elliptic
kernel 366
§ 84. General solution of the fundamental problems for
regions, mapped on to the circle by the help of
polynomials 366
 CONTENTS XXVII

§ 85. Generalization to the case of transformations by means

of rational functions 374
§ 85. Solution of the second fundamental problem. On the
solution of the mixed fundamental problem .... 379
§ 87. Other methods of solution of the fundamental problems 379
§ 87a. Example. Solution of the first fundamental problem
for an infinite plane with a circular hole 380
§ 88. Further examples. Application to some other boundary
problems 384
§ 89. Application to the approximate solution of the general
case 385
Chapter 16. Solution of the fundamental problems for the
HALF-PLANE AND FOR SEMI-INFINITE REGIONS.
§ 90. General formulae and propositions for the half- plane 391
§ 91. The general formulae for semi-infinite regions .... 397
§ 92. Basic formulae, connected with conformal transfor¬
mation on to the half-plane 399
§ 93. Solution of the first fundamental problem for the
half-plane 402
§ 93a. Example 406
§ 94. Solution of the second fundamental problem .... 409
§ 95. Solution of the fundamental problems for regions,
mapped on to the half-plane by means of rational
functions. Case of a parabolic contour 41 1
Chapter 17. Some general methods of solution of boundary

VALUE PROBLEMS. GENERALIZATIONS.

§ 96. On the integral equations of S. G. Mikhlin 414
§ 97. On a general method of solution of problems for
multiply connected regions 416
§ 98. The integral equations, proposed by the Author . . . 417
§ 99. Application to contours with corners 427
§ 100. On the numerical solution of the integral equations of
the plane theory of elasticity 427
§ 101. The integral equations of D. I. Sherman-G. Lauricella 427
§ 102. Solution of the first and second fundamental problems
by the method of D. I. Sherman 431
xx vin CONTENTS

§ 103. On the solution of the mixed fundamental problem and

of certain other boundary problems by means of D. I.
Sherman's method 440
§ 104. Generalization to anisotropic bodies 441
§ 105. On other applications of the general representation
of solutions 441
Part VI - Solution o/ the boundary o/ the plane theory of elasticity
by reduction %。the problem of linear relationship 445

Chapter 18. The problem of linear relationship.

§ 106. Sectionally holomorphic functions 447

§ 107. The problem of linear relationship (the Hilbert problem) 448
§ 108. Determination of a sectionally holomorphic function
for a given discontinuity 449
§ 109. Application 452
§ 1 09a. Example 455
§110. Solution of the problem : F+ —
§111. Case of discontinuous coefficients
gF~ + / 456
468

Chapter 19. Solution of

the fundamental problems for the
HALF-PLANE AND FOR THE PLANE WITH STRAIGHT CUTS.

§112. Transformation of the general formulae for the half¬

plane 471
§113. Solution of the first and second fundamental problems
for the half-plane 476
§114. Solution of the mixed fundamental problem .... 478
§ 114a. Examples.
1°. Stamp with straight horizontal base 486
2°. Stamp with straight inclined base 488
3°. Effect of asymmetrically distributed forces . . . 492
§ 1 15. The problem of pressure of rigid stamps in the absence
of friction 492
§116. Application 496
§ 1 1 6a. Examples.
1°. Stamp with straight horizontal base 501
2°. Stamp with straight inclined base 501
3°. Stamp with curved base 502
 CONTENTS XXIX

§ 117. Equilibrium of a rigid stamp on the boundary of an

elastic half-plane in the presence of friction 504
§ 117a. Examples.
1°. Stamp with straight horizontal base 508
2°. Stamp with straight inclined base 509
§118. An alternative method for the solution of the boundary
problems for the half-plane 510
§ 119. Problem of contact of two elastic bodies (generalized
plane problem of Hertz) 510
§ 120. Boundary problems for the plane with straight cuts 515
Chapter 20. Solution of boundary problems for regions,
BOUNDED BY CIRCLES, AND FOR THE INFINITE PLANE, CUT
ALONG CIRCULAR ARCS.
§ 121. Transformation of the general formulae for regions,
bounded by a circle 525
§ 122. Solution of the first and second fundamental problems
for the region, bounded by a circle 529
§ 123. The mixed fundamental problem for a region, bounded
by a circle 531
§ 123a. Example 536
§ 1 24. Boundary problems for the plane, cut along circular arcs 538
§ 1 24a. Example . Extension of the plane, cut along a circular arc 542

Chapter 21. Solution of the boundary problems for regions,

MAPPED ON TO THE CIRCLE BY RATIONAL FUNCTIONS.

§ 125. Transformation of the general formulae 546

§ 1 26. Solution of the first and second fundamental problems 552
§ 127. Solution of the mixed fundamental problem .... 554
§ 1 27a. Example . Solution of the mixed fundamental problem
for the plane with an elliptic hole 558
§ 128. The problem of contact with a rigid stamp 560
§ 128a. Examples.
1°. Circular disc 568
2°. Infinite plane with a circular hole 571
3°. Infinite plane with an elliptic hole 574
Part VII - Extension, torsion and bending of homogeneous and
compound bars 579
XXX CONTENTS

Chapter 22. Torsion and bending of homogeneous bars

(problem of Saint-Venant).
§ 129. Statement of the problem 583
§ 130. Certain formulae 586
§ 131. General solution of the torsion problem 587
§ 132. Complex torsion function. Stress functions 594
§ 133. On the solution of the torsion problem for certain parti¬
cular cases 597
§ 134. Application of conformal mapping 599
§ 134a. Examples.
1°. Epitrochoidal section 602
2°. Booth's lemniscate 604
3°. The loop of Bernoulli's lemniscate 605
4°. Confocal ellipses. Eccentric circles 607
§ 135. Extension by longitudinal forces 607
§ 136. Bending by couples, applied to the ends 608
§ 137. Bending by transverse forces 612
§ 138. On the solution of problems of bending for different
cross-sections 618
§ 138a. Example. Bending of a circular cylinder or tube . . 619

Chapter 23. Torsion of bars consisting of different

MATERIALS.

§ 139. General formulae 621

§ 140. Solution by means of integral equations 626
§ 140a. Applications ..
1°. Torsion of a circular cylinder, reinforced by a
longitudinal round bar of a different material . . . 630
2°. Torsion of a rectangular bar, consisting of two
different rectangular parts 635

Chapter 24. Extension and bending of bars, consisting of

DIFFERENT MATERIALS WITH UNIFORM POISSON’S RATIO.

§ 141. Notation 640

§ 142. Extension 642
§ 143. Bending by a couple 642
§ 144. Bending by a transverse force 643
 CONTENTS XXXI

§ 144a. Example. Bending of a compound circular tube by a

transverse force, applied to one of its ends 647
Chapter 25. Extension and bending for different Pois-
son's ratios .
§ 145. An auxiliary problem of plane deformation .... 650
§ 146. The problem of extension and of bending by a couple 652
§ 147. Particular cases.
1°. Extension of a bar, having an axis of symmetry 662
2°. Bar with plane of symmetry, bent by a couple . . 663
§ 148. Principal axis of extension and principal planes of
bending 664
§ 149. Application of complex representation. Examples . . 670
§ 150. Problem of bending by a transverse force 675
Appendix 1. On the concept of a tensor 682
Appendix 2. On the determination of functions from their perfect
differentials in multiply connected regions 697
Appendix 3. Determination of a function of a complex variable from
its real part. Indefinite integrals of holomorphic
functions 708
Authors index and references 713
Supplementary references 727
Subject index 729
 PART I

FUNDAMENTAL EQUATIONS OF THE MECHANICS OF AN

ELASTIC BODY
A
 In this introductory Part the basic concepts of the mathematical theory
of elasticity will be recapitulated. A deduction of the complete system of
equations of the mechanics of an elastic isotropic body will be given and
some fundamental propositions about these equations proved.
It will be assumed that the reader has some knowledge of the physical
foundations of the theory of elasticity and little consideration will be
given to this side of the subject. A more detailed account of the physics
and also of a number of general theoretical and practical problems, not
treated in this book, may be found in the following text books on the
theory of elasticity :
—
A. E. H. Love [1] (This book, first published in 1892 1893, is in many
respects obsolete, but nevertheless is very useful for the abundance
of material it offers.)
P. F. Papkovicz [1] 1939
L. S. Leibenson [1] 1947
S. Timoshenko [1, 2] 1914, 1916
R. Grammel [1] 1928
P. Burgatti [1] 1931
I. S. Sokolnikoff 切 1946
Further, the following textbooks on theoretical mechanics should
be mentioned :
G. Kirchhoff [1] 1897
A. G. Webster [1] 1904
These last two books contain a study of the basic theory of elasticity ;
the first of these, although its first edition appeared more than 70 years
ago, is still of interest at the present time.
A brief, but rather detailed outline of the historical development of
the theory of elasticity is given at the beginning of the book by A. E. H.
Love [1], A very detailed history of the theory up to 1893 with a careful
analysis of the different papers and books was presented by I. Todhunter
and K. Pearson [1].
The first two chapters of this book deal with all types of bodies which
may, with sufficient approximation, be called "continuous" (i.e. fluids,
elastic and plastic bodies, etc.). It is only at the beginning of the third
chapter that assumptions are introduced which characterize the (ideal)
elastic body as such. Throughout the first Part orthogonal rectilinear
coordinates are used.

3
A
 Chapter 1

ANALYSIS OF STRESS

§ 1. Body forces. In the mechanics of continuous bodies a distinction

is made between two types of forces :
1. Body forces, acting on the elements of volume (or mass) of the body ;
2. Stresses, acting on surface elements inside or on the boundary of
the body.
In order to explain this distinction in detail, imagine that a volume V
of arbitrary shape, bounded by the surface S, has been detached from the
continuous body under consideration. It is seen that the sum of the ex¬
ternal forces acting on V may be conceived as consisting of body forces
(eg gravity) and surface forces (e.g. pressure) .
The body forces will be considered first. They act on the volume
elements of the body, or actually on the mass contained in these elements.
Assume that the forces, acting on the infinitely small volume element
-> ->
dV, have the form dV where is some finite vector ; any point (%,夕，z)
of the element dV may be chosen as point of application of the vector
The vector is called a body force, referred to unit volume. If p denotes
the density at a given point of a body (i.e., the quantity of mass contained
1 f
.
in a unit volume), the vector —P will be the body force per unit mass.
-
In the case of gravity forces the vector is directed vertically down¬
wards and is in magnitude equal to pg, where g is the acceleration due to
gravity. Speaking generally, the vector depends on the position of the
volume element inside the body, or, in other words, on the coordinates
%,% z of a point within the infinitely small volume element. In dynamics

the vector depends also on the time.

NOTE. The mathematical statement that a body force, acting on a

volume element dV, may be represented by a vector applied to

5
6 I. FUNDAMENTAL EQUATIONS §2

some point of the element dV, must be understood in the sense that the
resultant force vector 里，acting on any finite volume V of the body,
may be represented by a triple integral, i.e., by

T=
加而亚
v
=
v
5 dx dy dz, (1.1)

and similarly the resultant moments of these forces about the axes Ox,
Oy, Oz of an orthogonal, rectilinear system by

Mx = (‘Z — zY)dx dy 恕 My = (zX — dx dy dz,

V V

Mz
p
3y
— yk)"％的)( L2) 取

where X, Y, Z are the components of the vector

Components of a vector will always be scalar quantities. Many authors, e.g

Love [1], denote by X, Z the components of body forces, referred to unit mass.
In that case the components of the vector will be pX, pF, pZ, where p is the
density.

§ 2. Stress. Surface forces act on the elements of the surface S of a

volume V, detached from a body (cf. § 1). It will be assumed that the force
acting on the infinitely small surface element has the form F dS,
where F is some finite vector. Any point of the element dS may be assumed
as the point of application of the vector F. The precise mathematical
statement of this fact must be understood in the same way as was in¬
dicated in the Note at the end of § 1 with regard to body forces. The
force F dS will be called the traction exerted on the element dS, and the
—>■
—
>
vector F the traction per unit area or the stress. F will often also be called
the stress vector.
The traction F dS represents the force acting between the parts of the
continuous body adjacent to either side of the surface element dS. Thus
—>
F dS is the force with which the part outside V acts on the part within V ;
CHAP. 1 ANALYSIS OF STRESS 7

the force with which the part within V acts on the part lying outside is by
Newton's third law of motion equal to
— F "S.
In general, any area (i.e., surface element) , conceived inside a body, is
bounded by two parts of the body adjoining
the area on either side. In order to distinguish
between these two elements of the body draw
the normal n to the area in question and give
it a definite positive direction (Fig. 1).
The traction, acting o% an area, will always
be understood to be the force which the ^art lying
on the positive side o/ 痣 surface element exerts
o% the part lying on the negative side. (The
same is of course true for the stresses, i.e., the Fig. I-
tractions per unit area.)
For example, when considering the traction exerted by the sides of
the surrounding body on the surface S of a part V imagined detached from
the body, one has to use the normal to S which is outward with regard to V.
—>■
As in the case of body forces, the vector 尸 depends on the position of the
element S and (in dynamics) on the time. In addition, it depends on the
orientation of the area in the body, i.e., o% the direction of normal n.
Therefore, when it is necessary to point out that the stress F refers to a
"'V

plane with the normal n, this will be indicated by writing Fn. The compo¬
nents of this vector will be denoted by Xn> Yn, Zn.

§ 3. Components of stress. Dependence of stress on the ori¬

entation of the plane. In order to study the dependence of stress on
the orientation of the plane to which it refers, select any orthogonal,
rectilinear system of axes Oxyz. Let M be a given point contained in that
plane. It will be shown that it is sufficient to know the stresses acting on
three mutually perpendicular planes passing through M, in order to be
able to calculate the stress acting on a plane orientated in any direction
whatsoever (and passing through that point) .
For the above-mentioned three planes select those, which are per¬
pendicular to the coordinate axes Ox, Oy, Oz respectively, and as positive
directions of the normals to these planes take the positive directions o/ the
corresponding axes.
The following standard notation will be used throughout this book.
8 I. FUNDAMENTAL EQUATIONS §3

Denote by Xx, Yx, the components of the stress vector acting on the
plane normal to 0%; here the index % indicates that the plane under
consideration is normal to Ox. Xx is the normal stress component acting
on this plane, while Yx, Z* are the tangential 。夕 shear stress components.
Similarly denote the components of the stress vector acting on the plane
normal to Oy by Xy, Yy, and the stress components acting on the
plane normal to Oz by Xz> YZ)
It will be shown that the quantities
X&, Y x, Zr,
z”, (3.i)
y y
C z， 工 7
乙 z，

characterize completely the state of stress in the neighbourhood of the

point considered. Therefore they are called stress components (at a given
point, at a given instant of time).
These components are shown in Fig. 2. However, it must not be forgot¬
ten that they are, by definition, scalar quantities. For example, in Fig. 2.
the actual quantity Xx is not depicted, but rather the vector whose %-wise
component equals Xx.

In order to find the relations between the quantities (3.1) and the
components of the stress vector Fn acting on the plane with the normal
n, passing through the given point M, consider the following approach.
Through the point M draw three planes, parallel to the coordinate
planes, and in addition another plane having the normal n and lying a
CHAP. 1 ANALYSIS OF STRESS 9

distance h from M. These four planes form a tetrahedron, three faces

of which are parallel to the coordinate planes, while the fourth ABC is
the face to be considered (Fig. 3).
Here and in the sequel it will be assumed (unless stated otherwise) that
the body forces and stresses change continuously with the position of the
point to which they refer. Further, it will be assumed that they maintain
equilibrium. This means, by a known principle of statics, that the sum
of the external forces, acting on the considered tetrahedron, has a resultant
vector equal to zero. Having in mind the transition to the limit 力 一> 0
the size of the tetrahedron will be assumed infinitely small.
Consider the projection on the % axis of the resultant vector of all
external forces acting on the tetrahedron.
The arguments will be based upon the supposition that the segments
->
—
>• ―»
MA, MB, MC have the same directions as the axes Ox, Oy, Oz. The reader
will easily convince himself that the results will hold true in all other
cases.
The projection of the body force equals (X + where dV is the
volume of the tetrahedron. The value X refers to the point M and s is
an infinitely small quantity (on account of the continuity of X).
Further, the projection of the tractions, acting on the face ABC, is
(Xn + where a denotes the area of the triangle ABC and z is again
infinitely small; Xn) Yn, as will be remembered, are the components
of the stress vector acting on the plane through M with normal n.
Finally, the projection of the external forces acting on MBC, normal
— —
to 0%,is ( Xx + 叼)％ where 久 is the area of MBC. Here Xx has been
taken instead of 十 Xx> since one is dealing with a force acting on an area
from that side of the body which lies on the negative side of the surface
element MBC (remembering that, by definition, Xx was to be positive
when the normal has the same direction as the axis 0%). For the sides
—
MCA and MAB one obtains similarly ( Xy + e2)o2 and (— Xz £3)cr3 +
respectively. Here q, s2 and e3 denote again infinitesimal quantities.
Thus, noting that
dV = 妥加，j = cr cos {n, %), & = 0 cos 仇，y), % = e cos (%,z),
one has
(X + £)蓊b + {Xn 十 —
+ ( Xx + %)b cos (% 劝 十
—
十 (— Xy + £2)ct cos 仇, y) + ( ~ Xz 知)!？cos {n, z) = 0.
Dividing by o and taking the limit 力 f 0 one obtains the following for-
10 I. FUNDAMENTAL EQUATIONS §4

mulae the last two of which have been written by analogy with the first :
Xn = Xx cos {n, %) + Xy cos (将，y) + Xz cos (四，z),
Yn = Yx cos (将, ％) + Yy cos 仇, y) 十 Yz cos (w, z), (3.2)
Zn = Zx cos (疆, 先) + Zy cos (n, y) + Zz cos (n, z).
The relations (3.2), as well as those to be deduced in § 4, were first found by
—
A. L. Cauchy (1789 1857) in a memoir, presented to the Paris Academy in 1822;
—
the results of this memoir were published in parts in the years 1823 1828.

§ 4. Equations, relating components of stress. It is known from

elementary theoretical mechanics that the resultant force and moment
of all external forces, acting on any body in equilibrium, are equal to zero.
In the case of absolutely rigid bodies (i.e., bodies which do not deform)
this condition leads to a system of six equations completely specifying
the state of equilibrium. In the case of a deformable body, however, the
above condition, when applied to the body as a whole, does not, by any
means, completely define the state of equilibrium.
However, in this last case as well, equations may be derived from the
above condition which (together with a law, expressing the relationship
between the stresses and deformations, to be discussed later) will give
all the necessary relations. For this purpose it is necessary to apply the
above condition not only to the body as a whole, but to each part which
may be imagined detached from it.
In the sequel, unless stated otherwise, it will be assumed that the
components of stress are not only continuous, but also have continuous
partial derivatives of the first order in the entire region occupied by
the body.
Let Pbe an arbitrary part of the body under consideration (which, by as¬
sumption, is in equilibrium), bounded by a closed surface S. The condition
of equilibrium will again be expressed by saying that the resultant vector
of all external forces, acting on V, is zero.
The projection of the resultant vector of the body forces on the Ox
axis is equal to

and the projection of the resultant vector of the tractions, exerted on the
surface S, is equal to 广“
//
Xn dS.
s
CHAP. 1 ANALYSIS OF STRESS 11

Replacing in the last formula Xn by the expression given by (3.2) and

equating to zero the sum of the projections on the Ox axis of the body
and surface forces, one obtains

v
XdV
s
+ J / [X& cos (肛 ％) + X% cos (%,y) + Xz cos (%,z)] dS = 0,

where n denotes the outward normal.

But by Green's Theorem

[Xx cos (%,%) + Xy cos (%,y) + Xz cos (知, z)] dS =

— :

欣当 + 墨 + 誓所
Introducing this expression into the preceding formula one obtains
finally

似X + 今 + +有斯=0.
Remember now that this equation must hold true for any region V
in the body. This can only be so, however, if the function under the in¬
tegral signs is zero at each point of the body. Thus one obtains the
equations
温 dXy dXz X o,
a% + dy
+ &
十 =
* +&— dYz
+ y = 0, (4.1)
a% dy
% SZy dZz
+ + z = 0.
These equations, to which reference will often be made, will be called
equilibrium equations.
The last step leading to (4.1) is based on the following reasoning. If F{x, y, z)
is a function continuous in a given region and

F{x, 州 z) dP = 0

for any part V, contained in that region, then F{x, 夕，z) = 0 in the entire region.
12 I. FUNDAMENTAL EQUATIONS §4

In fact, let, for example, F{x, y, z) > 0 at some point (g, y0, z。).Then, on the basis
of the continuity of F, one will have around the point (%» yQ, zj some region V,
where F{x, y, z) > % e being a positive constant. Hence

flf F dV > eV > 0

V
which contradicts the original condition.

Next, use will be made of the condition that the moment of the external
forces about the origin of the coordinate system must be zero, or, what is
the same thing, that the resultant moments about the coordinate axes
must be zero.
Writing that the resultant moment about the Ox axis of the body forces
and stresses acting on the surface S containing the volume V is equal
to zero, one obtains

V
(yZ — zV)亚 +
"建〃
s
— zY^dS = 0. (a)

But by (3.2)

— zY^dS = “{(yZ” 一 zYj cos (肛 劝 +

s s
+ — zYy) cos (nt y) + (yZz — zYJ cos (%,z)} dS,
or transforming, using Green's Theorem,

(yZg — •

Introducing this expression into (盘) and using (4.1) one finally finds

IH(Zy-Yz)dV = 0.
v
Since the region V is arbitrary, it follows by the same reasoning used to、
obtain (4.1) that
E = Zy, Z, = Xz, X“=y#. (4.2)
The two last formulae may be obtained from the first by cyclic
permutation of the symbols (or by applying the above reasoning to the-
axes Oy and Oz).
CHAP. 1 ANALYSIS OF STRESS 13

Thus it is seen that in the table of the stress components

Xx Xy X； z

ZX Zy Zjj
the terms, symmetrical with respect to the principal diagonal (running
from the upper left-hand to the lower right-hand corner), are equal in
pairs ; in other words, table (A) is symmetric. Thus only six of the nine
terms of the table are distinct, i.e.,
Xj：，Y y, Zz,Yz = Z^，Zx Xg；Xy = y
—
Hence it may be said that the state of stress at a given point is charac¬
terized by six of the quantities (A).
The formulae (4.2) may be presented in the form of a proposition. Let
there be two planes, passing through one and the same point ; then the
projection o/ the stress, acting on the first plane, the normal 力。the second
plane is equal to the projection of the stress, acting o% the second j)lane, the
normal to the first plane. Actually, the formulae (4.2) prove this proposition
immediately only in the case when the planes are perpendicular to one
another, (i.e., parallel to two coordinate planes) . But it is easy to generalize
this result to the case of two arbitrary planes and thus to obtain the
proposition formulated above.
In fact, let 足, 印，丫' be the direction cosines of the normal n' to the first
plane, and 邛'，丫〃 those of the normal n,r to the second plane. Then
-A-

the components of the stress vector Fn^ acting on the first plane, are by
(3・2)
x
〃，= %£ + xe + Xzy', = w， + y© + yy,
Z = Z* 十 Z*' + Zzy'.
愕

Using now the relations (4.2), the projection of this stress on the
normal to the second plane will be given by
("%" -+ X〃，a〃+ 丫小〃 + Z”，
Yz
- + 丫0旷+ Z*丫〃 十
+ ry') + Z式y灰〃+ y"w) + Xy^ff + a〃3')， (4.3)
where ( )启，indicates projection on the direction nn.
It will be seen that the above expression is quite symmetrical in the
quantities 印，丫' and 0", 丫〃 and that hence the parts played by the
two planes may be interchanged ; but this proves the proposition.
NOTE ON NOTATIONS. The notation %,W etc., used here for the
14 I. FUNDAMENTAL EQUATIONS §5

stress components, was first introduced by F. Neumann (1841) and has

been widely used, e.g. in books by G. Kirchhoff [1], A. E. H. Love [1],
S. Timoshenko [1, 2] and others. Besides this notation certain others
have been used, but only the following will be mentioned:
X#，^yy == y, ZZ Z)
~~ %期 = Z y>

~
— Z — Xz, — %£ — X? — Y
企 Tey x,

which is as widely used (with one or the other unimportant modification) in

contemporary literature as the notation used here. It is very convenient
from many points of view, especially as it agrees with the modern tensor
notation. In many places one finds ^册, <sy, written instead of ^纪加 xyy, tzz.

§ 5. Transformation of coordinates. Invariant quadratic form.

Stress tensor. Formula (4.3) allows the calculation of the projection in
any direction of the stress vector, acting on a given plane. In particular,
this formula may be used to deduce the transformation formulae for the
transition from one rectangular system of axes Oxyz to another Q^y'z'.
Let the direction cosines of the axes of the "new” system Oxfyfzf with
regard to the axes of the "old" system Oxyz be given by the following
table :
% y z

%, al Pi Yi

y， a2 圆 Y2

z， a3 03 丫3

In this table, for example, 仇, denote the direction cosines of the

axis 0%' with regard to the old axes, i.e.,
叼 = cos (婷，％), Bi = cos 出， y), Yi = cos (%', z).
The stress components in the new system of axes will now be denoted
by X；，, Z；，, Yi, Z；，, X； and the formulae will be found which
express these "new" components in terms of the old Xx, Yyr . . Xy.
Formula (4.3) immediately gives the required expression. For example,
for X；，one obtains
一
CHAP. 1 ANALYSIS OF STRESS 15

where Fx, denotes the stress vector acting on the plane, normal to the
new axis Ox' . Consequently one has to put in (4.3)
， ，
a — 0 = 0" = %,丫，= 丫" = Yi
which leads to the first of the following formulae, the others being ob¬
tained in an analogous manner:
X】= X,说 + Yy傥 + ZzY；+ 2Y*1Y1 + 2Z,Y1% + 2^^^,
y；，= + y财储 + + 2丫2仇丫2+ 22^7^2 + 2X^2^,
；
Z = XaW + Y”傥 + ZzY? + 2yz83Y3 + 2Z#丫3(X3 + 2乂尹3饱，
；
y = X*a2a + 1,8283 + ZzY2Y3 + 丫%(。2丫 + 限丫2) +
3 3

+ Z/丫2a + 丫3a2) + X?(a2B3 + a?%), (5.1)

：
3

Z = k跳3% + V2伪瓦 + ZzY3Yi + Y2(风y1 + 瓦y3) +

+ 2式丫3al + 7^3) + X/asSi + 修饱)，
X ；， = Xm口2 + Yy仇％ + ZzYiY2 + Y?(阮丫2 + 仇丫1) +
+ ^x(Yla2 + Y2al) + Xw(0CiB2 + “201).
One important result follows from these formulae. Adding the first
three and using the well-known relations
w
说 十 器 + = 优 + 能 + 优 = y 十 十 y: = 1, ； yg
81Y1 + 82Y2 + 83Y3 = Yiai + Y2a 2 + Y3a3 = %% + a2^2 + = 0,

；,+ ； + z；, x
one finds
x y = 工十 匕十 2
This formula may be interpreted as follows. expression
®= 十 +己
is invariant with regard to transformation o/ {orthogonal, rectilinear)
coordinates, or, in other words, the sum of the normal stress components,
acting on three mutually perpendicular planes, does not depend on the
orientation of these planes.
Next (4.3) will be used to calculate the normal component of the stress
Fn, acting on a plane with normal n. Let N denote the unknown normal
component, i.e., N = (Fn) n. For N〉 0, the normal stress will be tensile,
for N < 0, compressive.
If a, 。，丫 are the direction cosines of the normal n, then one obtains by
(4.3) the simple and important result
N = X" 十 匕俨 + +
十 2匕8丫 2Z«y(x + 2X^8. (5.2)
16 I. FUNDAMENTAL EQUATIONS §5

Introduce the notation

2Q(E, 7], 9 = XH + 丫炉 + ZQ + 2丫尔+ 2Z,殳 + 2X舟.(5.3)
〃
The function Q(E, 0 is a homogeneous rational function of the second
degree in 之， 刁， J i.e., in other words, a quadratic form in the variables
W, 叮，* It has a very simple geometric meaning. Thus, let P = (己, Q
denote a vector, normal to the considered plane and acting in the same
vj,

direction as the positive normal %. [In general, (己，寸，0 will denote a vector
with components 己，刀，乙，but it may also at times refer to the point with
coordinates 己，叮， Then
e a p 已
=p
a =p e=广 Y
…A
where P is the length of the vector P, and, by (5.2),
N.P2 = 2Q(E,7), O (5.4)
Now the following will be noted. The quantity N, by definition, has
physical meaning and hence cannot depend on the particular choice of
coordinate axes. In the same way the quantity P2 (i.e., the square of the
length of the vector) does not depend on this choice. Consequently the
quadratic form Q(E, 叮，Q cannot depend on it, i.e., it must be invariant
to transformation of (orthogonal, rectilinear) coordinates. In other
words, if E', vf , 7 denote the components of the vector P relative to new
axes and Q'(E', vf , C') is the quadratic form, involving 彳， 7 and
X；，, y), . . X；，in the same manner as Q(E, 刀，Q involves E, 叮，J Xxt
-
Y y, . • , Xy> then
QM"? = Q&"), (5.5)
i.e.,
+ 匕* + z%2 十
2
+ 22/宫 + =
= X/ + 丫炉 + z<2 + 2yH + + 2羽切 (55)
This equality must become an identity, if on the left-hand side one
substitutes for X；，, , X； from (5.1) and on the right-hand side expresses
E, 4 C in terms of the new coordinates, using the following formulae known
from analytic geometry:
E= + <X2刀'+ 仪3匚
灯 = 仇e' + 佐斯' 际'， (5，6)
〈 = y/ + y2刀’ 丫36
That this is so, is easily checked directly. For this it is sufficient to in-
CHAP. 1 ANALYSIS OF STRESS 17

troduce on the right-hand side of (5.5') the expressions (5.6) and to

compare the coefficients of 己飞 刀汽 »不，已看，之％‘ on both sides. It is
then seen that one finds for X；，, Y；，etc. the expressions (5.1).
Thus, to deduce the formulae (5.1), one may use the above stated rule
which is very convenient in practice. Namely, it is sufficient to write
down (55), to transform on the right-hand side (or on the left-hand side,
if one wants to obtain the transformation formulae from the new to the
old components of stress) the variables 己，刁，已 into g', vf, C' (or g', 娟，已，
into 己，7], E) and to compare the coefficients of the squares and products
of E', 叭 (or 7), O
The property of invariance of the quadratic form Q(E, 刀，Q proves that
the stress components Xx, …，X are components of a (symmetric)
1y

tensor of second order which will be called the stress tensor.

In the main part of this book the reader will not be assumed to be conversant
with tensor calculus. For the understanding of certain remarks it will be suf¬
ficient to study Appendix 1 at the end of this book. The following will help to eluci¬
date the final paragraph of this section.
Let there be given a quadratic form
2Q(E, 7), 0 = %君2 + &刈2 4- + + 2&KE 十 2%话力
where E, 7), C are the components of some (arbitrary) vector and the coefficients
%， …， ^Xy are quantities independent of & 7), J but depending on the direction
of the axes of the orthogonal rectilinear coordinate system. If, for transition
from one system of axes to another, the coefficients …， rxy change in such
a way that the quadratic form Q remains invariant, one says that the set of
quantities rxx, .. . , (involving two subscripts) represents a symmetric second
order tensor. The quantities txx etc. are called the components of the tensor.
In the notation of § 5, t淑 = Xx etc. (cf. Note at the end of § 4). With regard to
the definition of non-symmetric tensors of second order see Appendix 1.

§ 6. Stress surface. Principal stresses. Consideration of the

quadratic form Q(E, 刀， introduced in § 5, admits of a very simple and
clear geometric representation of the dependence of the stress vector
on the orientation of the plane to which it refers. This representation is
concerned with planes, passing through any definite point of a body.
In order to save space let the origin of the coordinate system coincide
with the point under consideration. Formula (5.4), viz.,

allows calculation of the normal stress component acting on the plane

the normal to which has the direction of the vector P = (已，7), 0 ；the
length of this vector may be fixed quite arbitrarily.
18 I. FUNDAMENTAL EQUATIONS §6

In the sequel Q(E, % 0 will be assumed to be not identically zero,

because in that case there are no stresses at the given point.
Use of an arbitrary length for P will be introduced by putting
N •F = 土 where c is arbitrary, but constant and different from zero
(note that c2 has the dimension of a force). The case has not been
excluded, when for some orientation of the plane : N = 0. When N = 0
it will be assumed that P = 8.
Thus 土
N (6.1)
P2
where the sign with c2 will be chosen such that 土 d and N have the same
sign (or, in other words, 4- c2 will be used when dealing with tensile, and
~ c2 when dealing with compressive normal stresses).

Let one end of the vector P = OH be at the origin 0 of the coordinate

system. Then the end p, 9 of the vector P
>
—
will lie on the surface
RR, 刀，9 = 土 c2, (6.2)
i.e.,
Xd 十 L哦 + … + 2X高 = 土 滑 (6.3)
The sign on the right-hand side must be chosen in the manner stated above,
depending on the sign of M
The surface (6.2) or (6.3) is obviously a quadric with the centre at the
origin. It is called the stress surface (stress quadric of Cauchy) referring to
a given point of the body. It will be seen later that two cases may occur :
in the one, the sign on the right-hand side of (6.2) or (6.3) remains the
same for all possible orientations of the planes ; in the other, the sign will
change depending on the orientation of the planes. Thus, in the second
case, one will, in actual fact, be dealing not with one but with two second
order surfaces
2Q = + c2 and 2Q = ~ c2
which obviously have a common axis (cf. below). (One may also fix the
sign of ,2 once for all and hence always deal with only one surface. But in
that case one has to give consideration to imaginary surfaces.)
Once the stress surface has been constructed, the normal stress acting
on a given plane (passing through the origin of the coordinate system) may
be found without difficulty ; it is sufficient to find the intersection H
of the normal n to the plane with the surface (6.3) . [It will be seen later
that such an intersection always exists, provided a definite choice has
CHAP. 1 ANALYSIS OF STRESS 19

been made for the sign on the right-hand side.] Then the normal stress
is given by (6.1) with P = | OH
Further, it is likewise easy to obtain the direction of the stress vector,
acting on the plane. In fact, equation (3.2) may be written
1 1 2Q
Xg = + 苦，

—p加 =2_当
=_L型， zn~
p (6・4)

remembering that cos {n, %) —— etc.

These formulae show that the vector Fn is parallel to the normal to

the surface (6.2) at the point H& p, Q. Thus, in order to find the direction
of Fn, it is sufficient to construct the tangent plane to the stress surface
at the point H and to draw the perpen¬
dicular to this plane from the origin. The
vector Fn then lies along this perpendicular
(Fig. 4). Further, since the projection N
of Fn on to the normal n to the plane un¬
der consideration is already known, the
construction of Fn offers no difficulty.
The vector Fn will have the direction
of the normal n to the considered plane
only in the case when the radius vector OH is perpendicular to the
tangent plane at H. In that case only a normal stress will act on the
plane, and no shear stress.
As is known, the radius vector OH will be perpendicular to the tangent
plane at H only when OH, and hence the normal n to the plane, has the
direction of one of the principal axes of the surface (6.3) ; in that case the
plane will coincide with the principal plane, normal to this axis.
In the general case there are known to be three such principal axes
which are mutually perpendicular. Only when the stress surface is a
surface of rotation will there be an infinity of such axes: one of these
coinciding with the axis of rotation, while all the others are perpendicular
to it. Finally, if the stress surface is a sphere, each diameter will be a
principal axis.
20 I. FUNDAMENTAL EQUATIONS §6

A direction with the property that only a normal stress acts on the
plane normal to it will be called a principal direction of stress or a principal
axis of stress, while the corresponding normal stress will be referred to as a
principal stress.
As has just been seen, there are always three such directions (and in
the general case only three) which are mutually perpendicular; in special
cases there may be infinitely many, of which, however, one may always
select three perpendicular to one another.
If one selects the coordinate axes along the three principal axes of
stress, i.e., along the axes of the surface (6.3), then its equation is known
to have the form
N/ + %哦 + %芋 = ± 岸 (6.5)
(i.e., the products of the coordinates disappear), where N2, N3 denote
the values of the quantities Xx, Yvt Zz for the new coordinate axes.
It is seen from this equation (as likewise on the basis of the definition
of principal axes of stress) that relative to the new axes the components
Yz> ZX) Xy become zero, i.e., no shearing stresses act on the planes coin¬
ciding with the coordinate planes. It should again be noted that all the
time consideration is being given to planes passing through a given point
(i.e., in the present case the origin of coordinates) . In general, when passing
from one point of the body to another, the principal directions will alter.
By definition, the quantities Nv N2, N3 are the principal stresses. The
stress distribution around the point 0 depends on the signs of these quan¬
tities ; for the time being they will be assumed to be different from zero.
First the case will be considered when all the principal stresses are po¬
sitive
Ni > 0, M > 0, % > 0.
In that case one has obviously to take the positive sign on the right-hand
side of (6.5) which takes the form
N/ + + N3芋 = + c2. (6.50
The corresponding surface is an ellipsoid. By (6.1)
+ c2
whence it is seen that the normal stress components acting on any plane
through 0 are tensile.
Next consider the case when all principal stresses are negative
CHAP. 1 ANALYSIS OF STRESS 21

which becomes
N也2 + 必噌 + 乂芋 = • c2.
_
(% v 0, N2 V 0, N3 V 0). Then the negative sign has to be taken in (6.5)

(656)
The stress surface is again an ellipsoid, but the normal stresses are now
given by N = | OH , indicating that, in contrast to the preceding
|2
case, the stresses on all planes are compressive.
Finally consider the case when the principal stresses differ in sign, e.g.
$ > 0, N2> 0, N3 < 0.
Then (6.5) takes the form

or
Ng + M哦 一 I 乂 I ? = 十 * (6.5c)

M哦 — I M If = — c2. (6.5d)
The surface (6.5c) is a hyperboloid of one sheet and the surface (6.5〃) a
hyperboloid of two sheets. Both surfaces are sepa¬
rated by the common asymptotic cones
必仔十必哦一【此亿2=0
(see Fig. 5). If the normal to the plane lies outside
the asymptotic cone, it intersects the surface (6.5c) ;
hence the normal stress is given by

N + c2
[ OH [2
audit willbe tensile. If the normal is inside the cone,
it intersects (6.54), so that the normal stress which
is now compressive is given by

N ~ ~—

\OH\2'
Finally, if the normal to the plane is directed along
one of the generators of the asymptotic cone, j OH | = 00 and N = 0,
i.e., the corresponding plane is only subject to shear.
The case < 0, N2 < 0, N3> 0 differs from the preceding one only
in that the regions of tension and compression are interchanged. All
other cases differ from those considered above in the way that the parts,
played by the coordinate axes, are interchanged.
Previously the cases, when one or two of the quantities N】，N?，N3 are
22 I. FUNDAMENTAL EQUATIONS §7

zero, have been excluded. (When = N& = N? = 0, no stress whatsoever

occurs.) In the case when one of these quantities is zero, the stress surface
degenerates into a cylinder and the state of stress at that point is then
called plane. This case will be studied in detail in § 8. When two of the
quantities N】， N?, N3 vanish, the stress surface evidently degenerates
into two parallel planes.

§ 7. Determination of principal stresses and axes. The problem

of finding the principal stresses and the corresponding principal axes has
been seen to be linked with the problem of determining a system of co¬
ordinates for which the quadratic Q(W, 刀，C) reduces to its "canonical”
form
*
Ng 十 N + NQ.
This problem is equivalent to finding the principal planes of stress, i.e.,
to reducing the equation to the form
必守+32哦十明烂=土皮 (7.1)
Its solution may be found in any textbook on Analytic Geometry or
Higher Algebra. It is likewise given in Appendix 1 at the end of this
book. It will be solved in § 8 for the case of plane stress.
It will be remembered that the values of the coefficients N&, N3
of (7.1), i.e., the values of the principal stresses, are given by the roots of
the third order equation in N (cf . Appendix 1)
Xx N— Xy
Yx Y、一N工
Xz
=— N3 + ®N2 + 4N + B = 0, (7.2)
Z, Zz~N
where
= X, + 匕 + 容
@

4 = 灯 + Z3 十
—匕乙 一 ZK 一％工， (7.3)
x x*
B= 工
Zx
夕

Zy Z?
- + 2yzX, — x,巴 ya - -
Since the roots N» N&, % do not depend on the choice of the coordinate
system, the coefficients of (7.2), i.e., A, B, likewise cannot depend on it.
In other words, these quantities are invariant with respect to trans-
CHAP. 1 ANALYSIS OF STRESS 23

formation of orthogonal rectilinear systems of axes. The invariance of

the expression
@ = X, 十 匕
+ Zz
has already been proved above by independent reasoning. This result
is likewise obvious on the basis of the fact that the sum of the roots of
(7.2) must be equal to ®, from which follows
N] + % + % = X, + 匕 Zz. (7.4)

§ 8. Plane stress. The state of stress of a body is called plane,

parallel to the plane IT, if, taking for II the plane one has for all
points of the body
工=匕=么=0. (8.1)
Thus there will be only three non-zero components of stress
y v V
If (8.1) does not hold true throughout the body, but only at some given
point, one speaks of a plane state of stress at a given point.
The formulae (3.2) indicate that the vector components of stress
acting on any plane, passing through a given point, will in the present
case be given by:
Zk = 0，
Xn = Xx cos (舛, 劝 + Xv cos (处, y), (8.2)
Yn = yx cos (% 力) + Yy cos (如 y).
It follows from Z = 0 that for any orientation of the
处 plane the stress
acting on it will be parallel to the plane
In the present case the quadratic form 2Q& 叮，Q becomes
2Q(E, 7]) = X" + 2X^ 十 V炉 (8.3)
and the equation of the stress surface
X』2+ 2X^7] + 丫炉 = 土 滑 (8.4)
This is a cylindrical surface the intersection of which with the plane
Oxy is the second order curve (8.4) with the origin as centre.
Limiting consideration to planes, parallel to it is sufficient for
an investigation, similar to that in § 7, to deal only with the above curve
instead of with the entire surface.
24 I. FUNDAMENTAL EQUATIONS §8

Now the transformation formulae will be found for the transition from
the stress components

to the components
y' v*

referring to a new system of axes, obtained from the old system Oxy by
rotation through an angle a in its own plane.
The angle a will be measured from the old
axis Ox to the new 0%' in the positive direc¬
tion of rotation in the plane Oxy (i.e., anti¬
clockwise ; see Fig. 6.)
These transformation formulae may be
obtained from (5.1), but they will be deduced
here anew using the property of invariance
Fig. 6. of the quadratic form QR, 刀) (cf. end of §5).
Using the known formulae for the transfor¬
mation of a vector 信 叮) into (匕 噌, i.e.,

one obtains
X需2 +
7) = E'

+ Y；
.
sin a
_
——
己 = 己’cos a vf sin a,
十 7/ cos
and introducing them on the right-hand side of
a,

才 2 x/ 十 2X^7) + 丫炉，
(8.5)

X；£'2 + + 吗切'2 = x病飞os oc-V sin 欧 十

十 2Xy(F cos a
— sin a)(g' sin a + 娟 cos a) + 丫财(己' sin a + vf cos (x)2,
whence follows by comparison of the coefficients of 己丝, 刀‘2 anj 己飞’
X；* = Xx cos2 a + 丫夕 sin2 a + 2XV sin a cos a,
Y；，= Xx sin2 a + 匕 cos2 a 2Xy sin a cos a,
— (8.6)

_ xq
X；, (•一 Xx 丫J sin a cos a + X?(cos2 a sin2 a) .
—
After obvious transformations these formulae become

X； l 互产 8s
—
2a 十 X” sin 2%
Xq + Yy Xx Yy .
丫式 = cos 2a —— sm 2a, (8.7)

X；, = sin 2a + X, cos 2a.

CHAP. 1 ANALYSIS OF STRESS 25

A direct check shows that from (8.7) follows

£, + y；, = % 十 ％,
(8・8)
居，一 X；，+ 2iX'y> = (匕 一 X* + 2%Xy) 0a.
The first of these formulae has been known for a long time and it was
proved above for the more general case [cf. (5.2)]. The second,very im¬
portant and convenient formula was stated by J. H. Michell [3] and it
was found independently by G. V. Kolosov [1].
Introducing in this formula 济汝 = cos 2a 4-勿 sin 2a and separating
real and imaginary parts, one obtains expressions for V；, ~ X；, and
X〉 in terms of the old stress components. Combining these with the first
equation of (8.8) one obtains expressions for X；，,丫),X〉 which, as is easily
verified, agree with those given in (8.7). Finally, note yet another formula
obtained by subtracting the equations (8.8) from each other:
2(X；, — z'Xj,) = Xa+ Yv~(Yy — Xx + 2iX 以那± (88)
Returning to (8.7) it will be shown that these formulae offer a very
simple way of determining the principal axes of stress and the principal
stresses. In fact, if 0%', Oy' be the unknown principal axes (the third
principal axis obviously being the axis Oz), then X；, = 0, whence by
the last equation of (8.7)

tan 2a = Mb-——
—匕
— (8・9)

Here a denotes the angle, measured in the sense stated earlier, which
the principal axis 。婷 makes with Ox. Formula (8.9) gives two values for
a；if one of these is denoted by a0, the second will be a0 十 All other
possible values differ from these two by multiples of 兀，and obviously a
may take any of these values. Substituting this value in the first two
formulae of (8.7) one obtains the principal stresses N1} the first
formula giving N、corresponding to the angle a, the second N? correspond-
, 冗
ing to oc •
2
Next, if one takes for the original coordinate axes the principal axes,
then
= ，匕 = 乂，％ = 0
M
26 I. FUNDAMENTAL EQUATIONS

and the formulae (8.7) become even simpler :

2
(8.10)
x ；,”
=— hsm
2
滤.

These formulae show that the maximum absolute value of the shear
stress is given by
%—%
I ^-y' I max ― 2
i.e., it is equal to half the absolute value of the difference of the principal
stresses. This value is attained on two mutually perpendicular planes,
bisecting the angle between the principal directions Ox, Oy.
Finally, the formulae will be written down which give Xx> Yy, Xy, if
the principal stresses Nlt N? and the angle oc between the principal axis
corresponding to N、 and the Ox axis are known. They are obtained from
(8.10) by interchanging the parts played by the old and the new systems
and by replacing the angle a by 一 a. In this way
•

N「N2co UM Nr~N2 cos 2a,

Xx = ——+ N? 1
Ni v = ——
cos 2a, Yy

3迫二丝
2
sin 2a. (8・U)

The formulae (8.1 1) are equivalent to the following which likewise

result directly from (8.8.) :
必，兀 一 X* + 2iXy = 一 (Ni —％)产 上 (8.i2)

NOTE. It is easily seen that the transformation formulae for the

stress components Xx, Yy, Xy into X；，, 丫)，X；,, for rotation of the
system in its own plane, remain the same as those deduced above, in
the case of a more general state of stress (and not a plane one), provided
that the axis Oz is 。处e of 应e principal 龙％es at the point considered. In
fact, in that case
工 = 匕 = £, = y；，
=o
there. The identity (55) then takes the form
X胃 十 2X岛 + Y" + n奏 = X； 石 '2 十 十 乂翁
CHAP. 1 ANALYSIS OF STRESS 27

because, by assumption, the Oz axis remains unchanged and hence

C = 片 必 denotes the principal stress corresponding to Oz, i.e.,

The earlier equation («) follows from the preceding relation and the
transformation formulae could have been deduced from it.
 Chapter 2

ANALYSIS OF STRAIN

§ 9. General remarks. The term deformation, when applied to a

continuous body, will refer to changes in the position of the points of this
body such that their relative distances are altered.
Refer such a body to an orthogonal coordinate system Oxyz and
denote by %,y, z the coordinates of a point of the body before deformation
and by %*, y*, z* the coordinates of the same point afterwards. Let V be
the region occupied by the body before deformation. Each point of the
body, occupying before deformation the position (%,y, z) of the region V,
will afterwards occupy a unique position (力*,y*, z*). This is the basic
assumption of the present chapter. Thus the coordinates %*, y*, z* must
be definite functions of the coordinates %, y, z of the same point before
deformation of the body:
%* = /1(%，y，z), y* = /2(%，% z), Z* = /3(%, y, Z)・ (9.1)
The functions f1} f2, /3 will be assumed to be continuous in the region V
(ie, the deformation causes no cleavage of the body) . The points (%*,
y*, z*), corresponding to the points (%,y, z) of V, cover some region V*
occupied by the body after deformation. Conversely, it will be assumed
that the coordinates 先，y, z are definite functions of x*} y*, z* [in other
words, that the equations (9.1) can be solved uniquely for 力，y, z] and
that these functions are likewise continuous for 先*.y*, z* in V*.
From a geometrical point of view the formulae (9.1) represent a certain
transformation of V into V*. It will be noted that not each such trans¬
formation, i.e., not all relations of the form (9.1), represent a deformation
of the body in the above sense. In fact, if one displaces the considered
body as a rigid unit (such a displacement will be called rigid body motion},
then the coordinates %*, y*, z* of the new positions of the points of the
body will be definite functions of %,y, z ; however, this is not a deformation,
i.e., a displacement of the points of the body with respect to each other.
For the sequel it will be very important, once the equations (9.1) are given,
to separate the actual deformation from the rigid body motion ; in other

28
CHAP. 2 ANALYSIS OF STRAIN 29

words, it will be important to find the quantities characterizing deformation

as such.
§ 10. Affine transformation. A transformation of the form (9.1)
is called affine, if the coordinates 力*, y*, z* are linear functions of the
coordinates x, y, z, i.e” if (9.1) has the form
%* = (】+ + 旬2y + + G,
y* = + (1 + 々22)?4- 023% 4- b, (10.1)
Z* = 他/ + G32y + (1 + 陶3)Z + C,
where 旬i，〃谑，...，a, 6, c are constants (for reasons, which will become
clear in the sequel, the diagonal terms have been denoted by 1 + alv
1 劭2， 1 〃33instead of by an, a22, a33). With reference to § 9 it must
be assumed that these equations are soluble with regard to x, y, z, ie, that
1 + 013
D ^21 1 + ^22 以23 (10.2)
«31 女
32 1 + 仇
33

is different from zero.

The affine transformation possesses many simple important properties
of which only the following will be noted. First of all, it is obvious that
the inverse transformation will be affine, since, solving (10.1) for %, y, z}
one clearly obtains linear expressions in terms of %*, y*, z*, i.e.,
% = (1 + 如)先* + %铲 + + af,
b13z^
y = 砥P + (1 + 超W* + 姐z* + y, (10.3)
z= + 奴y* + (i + 姐)z* + cr,
where bllt blit 6', cf are constants.
Further, it is easily shown that points, lying before the transformation
in some plane II, will after the transformation lie in some plane II*.
In fact, let Ax By Cz D = Q be the equation of the plane II.
Substituting for %, y, z from (10.3) one sees that this equation is trans¬
formed into one which is again linear in %*, y*, z*, i.e., into an equation of
the form A*x* + + C*n* + Q* = 0 which is, of course, the
equation of the plane II*. The points which were previously in the
plane IT will now lie in the plane n*.
It may also be shown that the above, in. combination, with, the property of con¬
tinuity of the transformation (i.e., that points at a finite distance correspond to
points at a finite distance, and points infinitely close together correspond to points
infinitely close together), characterizes the transformation, so that every transfor¬
mation with these properties will be affine.
30 I. FUNDAMENTAL EQUATIONS §10

It follows from the above property that points lying before the trans¬
formation on some straight line will move to points likewise on a
straight line A*. In fact, the straight line A may be considered as the
intersection of some planes 11^ II2. After the transformation, the points
of the straight line A, i.e., the points common to the planes H] and II2>
become points common to two planes II* and II* which are the trans¬
formed planes IIi and IT?, and this proves the assertion.
It follows from this that any straight segment is transformed into a
f

straight segment, and any vector into a vector. Let the vector P = (己, 叮，已)，
as the result of the transformation, become a vector

Further, let (%0, y0, z0) and 仇, y,z) be respectively the starting and end
points of P so that
己= —
&, 刀 = y —
%, E = z
The vector P* will similarly have the components
—
E* = —
where, for example, by (10.1)
—
7]* = y* y*, C* = z* 君， —
%* = (】+ + 的2y + 的3Z + 见瑞 = (1 十 诙J% 旬2yo 4- 旬 a.
Subtracting these two equations one finds the first of the following
formulae ; the others can be obtained in an analogous manner:
= (1 + 011)已 + 的2制 + 通3
的也 设

他1之 十 他 寸 十 033二
The formulae (10.4) simply express (cf. Appendix 1.2) that the vector
(匕*, 刀*, C*) is a linear vector function of the vector (匕，p, Q. Consequently the
quantities 1 alv a12> ..., a33t or more briefly q仃 十 &力 are components of a
certain tensor. But since is a tensor,then also (a方) is a tensor obtained from the
former by subtraction of the tensor (峪).
It follows directly from (10.4) that two equal vectors (i.e., vectors
having identical components 己，刀，Q become after transformation two
equal vectors, and that two parallel vectors become two parallel vectors,
the ratio of their lengths remaining unchanged. (The ratio of the moduli of
non-parallel vectors, generally speaking, is altered by the transformation.
Cf. Appendix 1.2). It follows also from this first property that two iden¬
tical and identically orientated polygons (lying in different parts of space)
are also transformed into identical and identically orientated polygons.
CHAP. 2 ANALYSIS OF STRAIN 31

But since every geometric figure may be considered the limit of polygonal
figures, it follows that the above property is valid for all figures. This
means that all parts of a body, independent of their position, will deform
in an identical manner. Therefore, the deformation arising from an affine
transformation is often called homogeneotis.

NOTE. It will always be assumed that the coordinates are not only
rectilinear, but also orthogonal. However, all the above will also be true
for an oblique coordinate system.
It is almost obvious that the character, i.e., the linearity of the relations
(10.1) or (10.4) remains unchanged, if one rectilinear systems of co¬
ordinates is replaced by another. This follows directly from the linearity
of the transformation formulae.

§ 11. Infinitesimal affine transformation. A transformation of

the form ( 1 0. 1 ) will be called infinitesimal, if the a诒,a, b, c are infinitesimal
quantities, the squares and products of which may be neglected in
comparison with these same quantities. It follows then from (10.1) that
by this assumption the differences
%* — % = an% + @10 + 〃招？4- a, —
y* y = a21x + + b,
z* — z = 说31% + a32y + %3Z + c
between the coordinates of one and the same point before and after the
transformation will be infinitesimal quantities.
Consider the result of two consecutive infinitesimal transformations.
Let the first of these be the affine infinitesimal transformation
%* = (1 + 曲1)% 十 诙2y + 曲3Z + a,
y* = + (1 + + 方，
Z* — 贬
3/ 十 。32y 十 (1
曲3Z
+ 口33)Z 十 C, (11.1)
and apply to %*, y*, z* another infinitesimal transformation
1* J + Av1 * + 70 Rz** + « f
一一

为演

+ 707012 z* + cTO
由*

3
* + J - To /
2
十

3
* Au3 + +
/

33

These two infinitesimal transformations transform the point (力，y, z)

into the point (%**, y**, z**). One obtains the relations between the
coordinates of these points by substituting the expressions (11.1) in
(11.2). Neglecting products of the quantities a, b, c one finds
32 I. FUNDAMENTAL EQUATIONS §12

without great difficulty

%** = (1 + + c12y + +

where
夕* * =
z** =
Ql%
C31%
+ +
(1 ，22)y
G3Z
+ C23Z +匕
+ c32y + (1 + C33" + c",
“ (11.3)

C{j =而 b{j Q, 7 = 1 , 2, 3), &" = ii + K b" = 6 十 6', c" = c + c'. (11.4)

These formulae prove that the result of two affine transformations is
again an affine transformation. This property, as the reader will easily
verify for himself, is true for any affine transformation whatsoever
(and not only for infinitesimal transformations) .
But the two following properties, deduced directly from (11.3) and (11.4),
are, generally speaking, only true for infinitesimal transformations. They
are as follows: the resulting transformation does not depend on the order
in which the two transformations were applied ; the coefficients c协
a", b", c" are the sums of the corresponding coefficients of these trans¬
formations.
It will be said that the resulting transformation was obtained by
composition of two transformations. All the above may be directly
generalized to the case, when an arbitrary number of transformations
is to be composed.

§ 12. Decomposition of infinitesimal transformations into pure

deformation and rigid body motion. Since in the sequel interest will
be concentrated on the problem of deformations, one may limit con¬
sideration to the transformation formulae (10.4) for the components of
a vector. If these formulae are given, i.e., if the quantities as3 are
given, the formulae (10.1) for the transformation of the coordinates of
a 力。说力 will not actually be completely defined, ie, the quantities a, b, c
still remain undetermined. But then these quantities obviously do not
influence the deformations, but only the rigid translatory displacement
of the body.
The formulae (10.4) may be written
求 = %也 +0 2+旬
12
的 + &2P + 23c
点

买 a3也 a32寸 E33已

where
匏 已* — e，m = 刀* — 刀， 又 〈* — a (12.2)
CHAP. 2 ANALYSIS OF STRAIN 33

denote the components of the vector difference P* — P — 5P, i.e., of the

increment of the vector P, caused by the transformation.
Next consider what conditions must be satisfied by the quantities
%I，旬2，^13 J

口22， 设23, (12.3)

口31，口32，“33,

which will be called the coefficients of the transformation under consider¬

ation (as stated in § 10 these coefficients represent the components of a
second order tensor), in order that (12.1) does not involve deformation,
i.e., that it expresses only rigid body motion.
A necessary and sufficient condition for this is that the length P of
any vector P, or what is the same thing, that the square of its modulus
户= 烂 + T2 十 (2
remains unaltered by the transformation.
In the following, consideration will be restricted to infinitesimal
transformations. Calculate the increase 3P of the length P. The preceding
formula together with (12.1) gives, neglecting higher order quantities,
PSP = 我 + 瑜7] + 双 = 口 做2哦 3? +命

十 023+ 他2 优 (031 + 旬3 生 312 + a21)之不 (即.4)

In order that 占尸 0 for all possible values of 己，不 C, it is obviously
necessary and sufficient that
设
11—~ 口22 a33 ， 0, 女
23
~1
a32 设
31-卜 口13 " 血12 I ,
~ (1 2.5)
This is the required condition that the transformation (12.1) represents
rigid body motion. It may be written briefly as
= — %” (z，7 = 1，2, 3) (125)
[which expresses the fact that the tensor (徇) is antisymmetric (cf.

，，
Appendix 1.2)] ; in fact, for i 片 j one obtains the second group of the
formulae (12.5), while for = one finds 口沈 =
agrees with the first of the formulae (12.5).
—
aiit i.e., 能书 = 0 which

Thus (12.1) may in the present case be written

where
—
毙=k1 F 劭=吒 一 力J 次 0刀 一 死， (12.6)

0 “32 ~ 以23， g = «13 = 口3卜y = 工 仇 (12-7)

34 I. FUNDAMENTAL EQUATIONS § 12

These are the well-known kinematical formulae expressing rigid (in¬

finitesimal) body motion. The quantities p, q, r, are known to be the
infinitesimal angles of rotation about the coordinate axes and will be
called components 0/ rotation. [The set of the quantities (自 q, may be
considered as a vector (cf. Appendix 1.3); in fact, it is the (infinitesimal)
rotation vector commonly used in kinematics]. The terms which refer to
the translatory displacement are missing from these formulae, because
consideration is being given to the components of a vector which is not
altered by the translatory displacements.
In order to obtain the formulae of transformation for the coordinates
of a point, occupying before displacement the position M{x, y, z), it is
sufficient to apply the preceding formulae to the vector
A
MqM = (% — — — y y0, z Zo),

where M^xQ, y0, z。) is an arbitrary, but once and for all fixed point of
the body. Substituting in (12.6) ％。， y
obtains the well known formulae of kinematics —— —
y0, z — z0 for 之，刀，C one

近 =〃 + ?(z
z°) r(y
—— —— — %),
匆 = 6 + r{x %0) 夕(z
& = c + 附一 ％) — q(x
—— z0),
航)，
(12.8)

where
a = b = 3y0, c = Sz。；
in other words, the vector (a, b, c) describes the displacement of the
point (%。，y0, zj. If one uses the origin of the coordinate system for the
point Mq, then (12.8) is somewhat simplified ; in fact, it becomes
a
—
qz ry, = b rx — 6z, c py — (12.8') —
where the vector 依 b, c) refers to the displacement of the point which
before the transformation coincided with the origin.
Next consider (12.4). It indicates that the change in length of the
vector P is characterized by the quantities
仇 '22, 033，口32 + 向3 + 口312^21 + 血12F

for which the following notation will now be introduced :

仇
11 = a22 = 内炉 他3 = %力) +
表(口32 a23) = ~ e必 .
(12.9)
.

*313 + 他1)= %工 *('21 + '12) = 分颦 —

CHAP. 2 ANALYSIS OF STRAIN 35

Actually, deformation is characterized by variations in the distances

between points, i.e., by changes of length of vectors ; it is determined by
the six quantities exx, eyy, ezzt e”，勿工，^xy which will be called components
of strain.

Since the % are the components of some second order tensor, the quantities
%?, …，％y are the components of a symmetric second order tensor, as may be seen
from Appendix 1.3; a direct proof of this fact will be given below.
Similarly, the quantities *(a 仃 一 ％。are components of an anti-symmetric second
order tensor which may be represented by means of the vector (仇 g, y) (cf. Appendix
L)

Further, in agreement with what has just been stated, introduce the
notation
o
= 春(他2 ~~ 四23)， ? = i(^13 口31)， y = *(口21 (12.10)
In the above notation, obviously,
他2= eyz 十 0， 旬3= e* + % 仇21 = 九

o
/

g
J2 [ ])
a23 ~ e 少z ， 分0； 0， 旬2 = 分对 ，
which demonstrates the division of the tensor (&?) into the sum of symme¬
tric and anti-symmetric parts.
The formulae (12.1) may now be written
能= + 济 + exzE 十 K ' '

劭 — 上 疙 力C, (12.12)
淡 + ezX + 勿) 一 代。
These formulae show that the original affine transformation may be
divided into two transformations : one of the form
丽
剑
- %上 发
十 八沙刀
+ 分工
( 12. 13)
义
and another of the form (12.6) representing rigid body motion. The
transformation (12.13) which contains only components of deformation
will be called actual or pure homogeneous deformation (see later).
It is characteristic of the formulae (12.13) that the array of coefficients
^xz
^yx 6财VCyZ
勿勺
is symmetric.
36 I. FUNDAMENTAL EQUATIONS §12

Each of the components of strain has a very simple geometric meaning.

-
The latter may be deduced directly from the formula (12.4) which in
the new notation becomes
P^P %营 + 分储 + c/2 + 2%z求 2%殳 + 2e/斗 (124)
Consider some vector 0, 0) which before deformation is parallel
to the 0% axis. For this vector
PSP = &上t
or, taking into account that 皆 = 尸2,
6P
e#a：= p • (12. 14)

Thus exx represents the relative increase of the vector (or segment),
originally parallel to the axis Ox. The components evy and ezz have an
analogous meaning.
If all the components of deformation, except e皿 are zero and if one
considers pure deformation, i.e., if
力 = g = y = 0,
then (12.13) gives
匏 = %£ 询 = 次 = o.
Hence, in this case, all vectors parallel to the axis Ox are stretched in one
院
and the same manner (the proportional increase being 丁 — ；however,
vectors perpendicular to this axis do not change their direction nor
their length. Thus this case represents a simple and homogeneous extension
in the direction Ox. Similar results will be obtained in cases when either
eyy or ezz are the only non-zero components.
In order to explain the meaning of eyz, one has to determine the
->

change of the originally right angle between the two vectors 尸式0, %,0)
and 尸2(°，0, O which before deformation were directed along Oy and Oz.
Let the angle between these vectors after deformation be denoted by

2
£yz (i-e., > 0 if the angle decreases and eyz < 0 if it increases).
By a known formula the cosine of the angle between two vectors
(该i，小 + 谢i，次J and (能2, 初2, Q + 肥2)
is given by
…/兀
\2
、 ，溜
编编 版
+ m)2 + 混 .e寇
的1)的2 十次
湖 q + 端产
CHAP. 2 ANALYSIS OF STRAIN 37

But when zyz is small,

COS (2 Eyz，

neglecting infinitely small higher order terms. Omitting higher order

terms on the right-hand side of the above equation one obtains
万物2 + J第1 次1
。
的 2

5
But by (12.12), applying it to 号(0, 0) and P2(0, 0, ^2),
兴I = + 勿)1，问2 = 6”工2 〈 勿 2；
introducing these values in the preceding formula one finds
= %、+ ^yz = (12.15)
Thus the quantity 2e” represents the decrease of the angle between
two vectors having originally the (positive) directions of the axes Oy
and Oz. Similar interpretations may be found for 2ezx and 2exy.
Now consider pure deformation for which all components but eyz are
equal to zero. Let OB and OC be two vectors,
starting for simplicity from the origin and di¬
rected along the axes Oy and Oz, and let OBCK
be a rectangle constructed on these two vectors
(Fig. 7). After deformation the rectangle be¬
comes the parallelogram OB'C'K' (where it is
assumed that the origin is not displaced ; if this
assumption does not hold, one may bring the
origin back to its old position by means of a
translation) .
By (12.13) the point B is transformed into
the point B' on the straight line BK and the point C into C' on CK ;
further,
BB‘
— ezg.OB, CC' = eyz .OC.
Since, neglecting infinitely small higher order terms,
Z\ CC Z\ Z\
—
- tan BOB' = BOB' ,
OB = OC =
tan COCf = COCr ,
the preceding formulae give
Z\
BOB'
- 八
COC = 入
38 I. FUNDAMENTAL EQUATIONS §13

whence one obtains again

Z\ Z\
zyz = BOB' + COC' =
If, by means of rigid rotation about Ox, one causes the segment OB'
to coincide with OB (the difference in their lengths obviously being a
higher order quantity) , the parallelogram
takes the position OBKnC" (Fig. 8)
Z\ .
and the angle COC" is again equal to the
angle zyz (where it has been assumed that
C" lies on the straight line CK, since obvi¬
ously this will be so, neglecting higher order
terms) .
Thus the deformation represents a shearing
of planes, parallel to the plane Oxy in the
Fig. 8. direction of the axis 。夕， and the displace¬
ment of each layer is proportional to its
distance from the plane Oxy. The quantity
CC" measures the "absolute shear”, and
CC" •- TOTI 匕
G ~ C* /名
匕货 Z
0c
the "relative shear" or the angle 。/ shear. The considered deformation is
called simple (homogeneous) shear.

§ 13. The invariant quadratic form, connected with deformation.

The strain surface, principal axes. Transformation of coordinates.
The formula (12.4) may be written
P* = 2F(E«，0， (13.1)
where now
+ +
2万倡 r j) = 4斤 %哦 %* + 2以戒 + 2%总 2%高， (13.2) +
i.e., F is a quadratic form in the variables 匕，刀，G Since the left-hand side
of (13.1), i.e., P^P, has a definite meaning, independent of the choice of
coordinate axes, it follows that the quadratic form 刀，0 is invariant
with regard to transformation of coordinates. In other words, if %,疗,
…， ex,yf are the components of strain in a new coordinate system and

己'， 已’ are the components of the vector P in the new system, then
每/'2 十 分，鹏'2 + . . . + 2%， 忌%' %应2 + . . . + 2%患7], (13.3)
CHAP. 2 ANALYSIS OF STRAIN 39

which becomes an identity in 7, if on the right-hand side g, 刀，已

are expressed in terms of 己'，刁'，C'. This proves that the array of the
quantities
分8分ye a%
^yx ^yy
e? 财 ©zz

represents a symmetric second order tensor (cf. end of § 5) . In particular,

as in § 5, it follows that the components of strain in the new coordinate
system are related to the old ones by the same formulae (5.1) as the new
components of stress were related to the old ones (one only has to replace
in those formulae Xx by exx, Yz by eyz etc.).
Just as in § 6 the stress surface
2QR, 刀乂) = ± c2
was introduced for the study of stresses, one may here consider an
analogous surface.
The formula (13.1) may be written

P = 2F信 刀 j)
r
or
Pe = 2F怎 刀 j),

——
瓶
—
f

where e p- denotes the relative increase of the vector P (己，斗

the length P so that P2e = ± _

As is known, this quantity does not depend on the length of the vector P,
but only on its direction. Therefore one may for every direction choose
where c is an arbitrary fixed positive
constant which has the dimension of a length.
If one takes as the starting point of the vector P the origin of the co-
ordinate system, then the end point H of this vector will lie on the surface
2尸&布 Q = ± 岸， or 分竟2 + . . . + 2%冽 = ± c2 (13.4)
which is called the strain surface (Cauchy's strain quadric). Once this
surface has been constructed, one can immediately find the relative
increase in length e of any vector. For this purpose it is sufficient to
draw, parallel to the vector, from the origin the semi-axis OH to its
intersection H with the surface ; in order that such a point of intersection
will exist (i.e., that it is real) it is necessary to choose the sign of c2 on the
40 I. FUNDAMENTAL EQUATIONS § 13

right-hand side in a definite manner. The relative change in length of the

considered vector will be
c2
e= ± . —— —
All the above is quite analogous to what has been said in § 6 with
regard to the determination of the normal component of stress N and
therefore it is not necessary to repeat those details here.
If the coordinate axes are chosen in such a way that they coincide with
the principal axes of the surface (13.4), its equation takes the form
延2 十的哦+冬2 = 土* (134)
where % %，% denote the values of exx, eyy, ezz for the new system ; na¬
turally the components eyz, ezx> will be zero in that system. Con”
sequently the new system of axes has the property that the angles between
the axes after deformation remain right angles. This means, as there are
always three such mutually perpendicular axes, that the angles between
them remain unchanged by deformation. Those three axes are called
伊in滋al axes of strain. The quantities %,% % are referred to as principal
strains.
In the general case there exists only one such set of three axes. But
if the surface (13.4) is a surface of revolution (i.e., when two of the quan¬
tities are equal), there will be an infinity of such sets.
If one chooses the principal axes of strain as coordinate axes, the
formulae (12.13), expressing pure deformation, take the form
能 — 谢 = %刀， 又 气答
Consequently every pure deformation may be represented as the result
of three simple extensions in three mutually perpendicular directions
which are the directions of the principal axes of strain.
Finally note that the principal strains 4, %，S are the roots of a cubic
equation in e (cf. § 7)

%” e” —e = —济 + ^次 十 加 十 c = 0, (13.6)

where, in particular,
e = e + 财 + ezz.
显 (13.7)
Since the coefficients of (13.6) must be invariant (cf. §7), it is clear
CHAP. 2 ANALYSIS OF STRAIN 41

that 6 must be so. Obviously 6 represents the sum of the roots of (13.6),
i.e.,
9 = %工 + + %z = 4 + % + (13.8)
The quantity 6 has a very simple geometrical meaning. In fact, consider
a right parallelepiped, constructed on segments OA, OB and OC of the
principal axes and having the volume
rm I J 2^3〉
where
= OA, Z2 = OB, 4 = OC.
After deformation the considered parallelepiped will still be a right
parallelepiped with sides
"1(1 + %)， %2(1 + ©2)， %3(1 + /)，
and its volume will be
『= + %)(1 + ©2) (1 + /)= P(1 + 4 + ©2 + %),
广
__
neglecting higher order terms. Consequently
y
— 厂一二勿+/十% (13.9)

This formula shows that 6 is the relative expansion of the volume V or the
cubical dilatation.

§ 14. General deformation. Consider now the most general de¬

formation of a continuous body. Let the point M, having initially the
coordinates %, y, z, move as a consequence of deformation to the po¬
sition
y*, z*).
Write
%* = % + &, y* = y + », Z* = Z + M； (14.1)
>
u, v, m are the components of the vector MM* which expresses the
displacement of the point M as the result of deformation. This vector
will be caUed displacement vector or simply displacement, and 孙 厉 缈
displacement components. Since different points of a body, generally
speaking, will be displaced in a different manner, u, v, w will be functions
of the coordinates %, y, z of the original position of the point under
consideration
% = y, z), 0 = v(x, y, z), m = 如(％, y, z). (14.2)
42 I. FUNDAMENTAL EQUATIONS § 14

(Sometimes the d运placement may also be a function of the time ; in that

case one considers the state of deformation at some definite instant of
time.)
In the sequel, unless stated otherwise, it will be assumed that the
functions %〃缈 are not only single-valued and continuous, but also that
they have continuous derivatives up to and including the third order.
Select at some point M (力, y, z) of the body an infinitely small neigh¬
bouring volume and investigate its changes as a result of deformation.
For this it is sufficient to study the variations of infinitesimal vectors
having (before deformation) the point M as their starting point. Let
MN = 信力9
be such a vector. After deformation M will have moved to M*, and N to
—
>•
N*, so that the vector P becomes the vector P* = M*N*. Calculate the
f

vectorial increment
—A
6A
->
of the vector P*, i.e., — A
= P* —
f

P. The coor¬
dinates of M* are
+ u{x, y,z), y + 矶％, y, z), + w(x, y, z),
while those of N*, having before deformation the coordinates
% + & y + % z + J
will be
十 己 十 ％(% + 己, 夕 + 办 z + 9，y + 7) 十矶％ + & y + 7], Z + Q,
z + E 十 》(% + & y 十 斗 z + 0.
Therefore the components of the vector P* will be
之 + (力 + &y + 4z十3— y, z),

" v(% + ^, y 4- 7], z + Q

+ — 也，, y, z),
E + 切伊 十 E, y + 口 z + Q — »(%, yf z).

Finally, the components 8己， 6刀， of the vector will be

%(% + E, y +〃Z + 9
(％ + 己， 夕 +
切
— %(%, % z), v(x +
力 z + 9
y+

— M(%, y, z).
— W%, % z),

But by Taylor's Theorem

沏 a%
"(% + E, y +斗z + Q ~~
d% —
〃(％, y, z ) = ~ 已 + 彳一 刀
cy —
+ —cz 已 + 所
CHAP. 2 ANALYSIS OF STRAIN 43

where e is an infinitely small term of higher order than vj, Neglecting

w and proceeding analogously in the case of the other components, one
finds

说=后巴+ 可 72 +既还
加 加 加
访 - ""
E —I— '
» —[―
' ' ' ' (_ (14.3)
3% 勿 '

dw dw
次=后之+ 方刀 + dz

a%
in these formulae the values of etc. refer to the point (%, y z) and do
ox
not depend on 己，刀，J These formulae show that, apart from higher order
terms involving the linear dimensions of the considered body element,
the change of this element may be expressed by means of an affine
. . a% du
transformation with the coefficients = ox a12 = oy etc.
， ——
Hitherto no limiting assumptions have been introduced with regard
to the order of smallness of the displacement components u, v, w. It will
now be assumed {and this condition will always a少恒1y) that the components
of displacement 见 o, 沙 and also their derivatives with respect to %, y, z
are infinitely small quantities the squares and products of which may
be neglected in comparison with these quantities. Then (14.3) will be
an infinitesimal transformation and everything said in the preceding
sections will apply.
It was seen that pure deformation of the element under consideration
was expressed by the formulae (§ 12)
踞 — + ^x^>>
== e eE +
少; + eyzE， (14.4)
次= + 久此 + ^zX>
where exx, .. exv are the strain components, determined by the formulae
du 加 dw

(14.5)
44 I. FUNDAMENTAL EQUATIONS §15

Generally speaking, the pure deformation should still be combined

with the rigid displacement of the considered element with the in¬
finitesimal components of rotation
/ dw 加\ /3% dw\ J 加 加、
=改而 一法) 夕=《法一 为) 一可) (海6
/八

力
‘ 可
and the translatory displacement which is equal to the displacement of
the point M(x, y, z), i.e., its components will be the values of u, 0, w at
M(x, y, z).
The essential difference between the present deformation and the
homogeneous deformation of § 10 arises from the fact that here the
components of strain eXXJ ... etc. depend on the location of the con¬
sidered body element, i.e., on the coordinates %, y, z. In particular, the
directions of the principal axes of strain will now change from point
to point. Similarly, of course, the components of rotation will depend
on %,y, z.
Finally, it will be noted that the quantity
加 加 dw
+
/
8 == e” 十 "x I x I ~

3% cy oz
is invariant with regard to transformation of orthogonal coordinates
and represents the cubical dilatation. But since one is now dealing with
non-homogeneous deformation, it is of course clear that one can only
talk of the dilatation of a volume element in the neighbourhood of a
given point.
Most of the properties of deformation, studied above, were first deduced
by Cauchy in his memoir of 1822 (cf . § 3) .
§ 15. Determination of displacements from components of
strain. Saint-Venant*s conditions of compatibility. In § 14 formulae
have been deduced by which the components of deformation can be
calculated from the displacement components, given as functions of
%, y, z. Now the inverse problem will be considered : to determine the com¬
ponents of displacement 说，q w, if the strain components 矽e

given as functions 0/ %, y, z. Before solving this problem, several pre¬

liminary remarks will be made which will make it possible to predict
certain results.
The values of the strain components have been seen to determine
the change in shape of an infinitesimal element of the body near a given
point. Thus the strain components as functions of 先，y, z determine
CHAP. 2 ANALYSIS OF STRAIN 45

the change in shape of every infinitesimal element of the body. As a

result it is obvious that the deformation of the body as a whole is ef¬
fectively determined, i.e., the values of the displacements u, 0, 3 as
functions of %, y, z. It is likewise clear that %,巩 m may not be determined
uniquely. In fact, if displacements, corresponding to given strain com¬
ponents, have been found, then, by adding an arbitrary (infinitesimal)
displacement of the body as a rigid unit, one will obtain different values
for the displacements which will still correspond to the same components
of strain, because the rigid body motion has no effect on the deformation.
In order to make the problem unique, one may, for example, assume
in addition that the displacement of any arbitrary point MQ of the body
and also the components of rotation at this point are given.
The following may also be noted. By an earlier assumption, the
components u, o, w are single-valued and have continuous derivatives
up to and including the third order. Hence the given components of strain
exx> must likewise be single-valued and have continuous deriva¬
tives of the second order; this condition will be assumed to be satisfied.
However, it is easily seen beforehand that the quantities e 皿 …，％t
must still satisfy definite relations, in order that the problem will have a
solution. This follows already from the following rough considerations.
Let an infinitesimal element, e.g. a cube (which is not adjacent to the
boundary), be separated from the body. If one subjects every such cube
to a deformation with given components and then tries again to join
all the infinitesimal parallelepipeds obtained in this way so that their
boundaries, which were adjoining before deformation, again touch, then,
generally speaking, this will turn out to be impossible ; in the attempt
of joining the separate elements there may either appear gaps between
several of them, or boundaries of elements which should match may be
found to differ from each other in size, or finally some elements may be
too large for the space available. All this shows that the components
of strain must satisfy certain relations, in order to allow deformation with¬
out discontinuities. This will now be proved strictly by actually
solving the original problem.
Thus let it be required to find functions %,% w satisfying the conditions
a% 加 3加 • a
C
ox —
，‘
‘
， ।

匕ZJC,
oy —
C
q
匕
03 C
oz — 匕zz,
‘， ，

(15.1)
a 物 加 Zu aw 加 加
-
C
ay
十 c
. . 1 —

oz
一

—.
， ， JQ __
N'yc， c
cz
I
十
d% —
C
一

"zm
， .
i
/Q -

6%
__ _
c
_
十
I
C
dy
. xQ
一 /个女夕，
46 I. FUNDAMENTAL EQUATIONS § 15

where %工， are given single-valued functions of 沏 y, z having

continuous second“ order derivatives.
One has six equations for the determination of three unknown functions.
This again shows that the problem may not have a solution, if the given
functions exx> are not subject to certain additional conditions ;
these conditions will be found while solving the above problem.
Let V be the region originally occupied by the body ; this is the domain
of values of x, y, z for which the functions exy are given and for
which the functions u, ”，w must be found. For the present, V will be
assumed to be simply connected. It will be remembered that a region is
called simply connected, if it has the following property : every closed
contour, lying inside the region, may be shrunk into one point by means
of continuous changes which do not take the contour outside the region.
Such regions are, for example, represented by a sphere, a cube etc. (for
more details see Appendix 2.)
Let ％,^0) be any point of V, %。，％ % the values of the com¬
ponents of displacement there and %,%,夕 the corresponding values of
the components of rotation. Let %(看， °
月, 近) be any other point of V.
Consider the problem of determining the components of displacement
at the latter point.
Let MqM1 denote any curve which joins Mo and and lies in V. If the
m and du- were known throughout V, one
partial derivatives
加
3% cy —oz
could find the value of the function % at the point from the formula
脑 du \
+
t
d% 十 -y- dy dz J, (15.2)

where the integral must be taken over the curve But

a% 3%
exy ― 夕, = ezx + q, (15.3)
3% dz
where q and r are determined by the formulae (14.6).
Hence

的 = 十 / + exydy + exzdz) 十 / (qdz

M0M1
— rdy). (a)

The first integrand involves only given functions. Consider now the
CHAP, 2 ANALYSIS OF STRAIN 47

second integral. One has

^(qdz — rdy) =/{rd(y1~y)—qd(z1 — z)}}

MqMjl Mo%
whence, integrating by parts,

/
MqMx
(fdz— My) =%(%—%—々仇
一％)T MQMt
{仇 — 夕)办 — ki—z)〃办 仅)

In order to evaluate the last integral, one requires the values of dr, dq or,
what is the same thing, the values of the first order partial derivatives of
the functions r and q. But it may be verified directly that
He* dr devy dr
a% a% dy ' 如 — 2% dy
J
dz a% dy '
丝__ a% 的 次颦工 a%? 却 a% a%
dx - dz3% ' dy
一
a%
J
dz dz 3%
Substituting these expressions in
， dr dr
协 =
ax
办
+ oy
的+
dz
dz,

, 困， dq 却
%=标 族十可 为+ dz,

one obtains, using («) and (6), the first of the three formulae below (the
other two may be obtained from the first by cyclic interchange of symbols)
贫
31? %' Z, = 十
— z。) 一々仇
—
(刀 + Uy dy Uz dz),
也%1，%，
年) + %(%1 - 40) — %(4 一 70) +
/
+ (Vxdx+Vydy+Vzdz),
距0如

2,=
%， 切
0 十 九伉 — 端 — 一%)

+ J (印 + W.dy 十 Wzdz),
48 I. FUNDAMENTAL EQUATIONS § 15

where, for convenience,

加①0 \ 、/ a%
— 苇丁)+(2】 2)(1r
— . z 之

+ (% y)

口 + 仇 - 切 (>-»+(LZ)(M-T),(65)
f + 仇2(>-»+ (z】T 性-曾).
The formulae for Vx,匕， Vz and Wx,Wv, Wz are obtained from the above
by cyclic interchange of symbols (by simultaneously transposing the
symbols U, V, W and x, y, z).
The formulae (15.4) essentially agree with those found by V. Volterra
[1], p. 406, using transformation formulae given by G. Kirchhoff [1],
Vorles. XXVII, § 4. The deduction presented here is due to E. Cesaro
(Rendiconti d. R., Academia di Napoli, 1906; it is also quoted in V. Vol¬
—
terra [1], where it is reproduced on pp. 416 ~417, as due to Cesaro) who
gave Volterra^ formulae a more symmetrical form.
The formulae (15.4) determine the displacement components 的,巧，切i
at any point 加式叼,％,z, of the body, if the displacement
(附，％, and the rotation (%,%,r0) are given at some other point
Mq(x0, %,z°) which has been chosen once for all. The formulae for the
displacements contain integrals taken over some curve connecting
the points Mo and Mx. But u, 巩 w must be functions of %】，y1} and
should not depend on the path of integration MOMV This means, in order
that the problem may have a solution, it is necessary that the integrals
in (15.4) are independent of the path of integration.
It is easily seen that the necessary and sufficient conditions for the
integral

/…刀 …物

犯
勿
_
to be independent of the path MQMr are (cf. Appendix .2.)
犯
彻 ’
犯 肛
一
dx
里％ 巩
a% a, •

For the two other integrals one obtains analogous conditions by cyclic
rotation of symbols. These conditions must be satisfied at all points
(x, y, z) of V and for all values (%】，％,z, in that region.
CHAP. 2 ANALYSIS OF STRAIN 49

Performing the differentiations, it will be seen that these conditions

may be reduced to the following six:
a%% 3%名 a / a%%
+ —- Q
N — j
2G
+
，
朋 效2 dydz ' dydz a% \ a% 效
a ( ^ezx 3%
+ 答｝”6）
a% 洲% 严分工
+
缉
=2
a%? 盼 dzdx dy 卜 如 &
洛埒 1
十
关??
=2
洛明
，
o2ezz J
— (
( ^eXy
+ +
如2 a%2 dxdy dzdy a% 如 /
For example, the condition
犯 as
dy
gives, by (15.5),

(月 - y)
dxdz
i
lz
+(4 — z)
z)
'%
/(有 夕

砺
炉
、
j
\/济噎 光约、 （句
（九 一 加（_声 - 丽^ - （连阳 备豆
a（济/ 一 ^ezz 工

Since these relations must hold true for all 九， in a given region, one
must have
济分 y 洲公财
一
济分? 济内2 3分财济分 z炉力, 济分z
dydz dy2 dxdy 源 dxdz dzdy dxdy
These relations agree with the last two of the right-hand column of (15.6).
The others may be obtained by the same procedure. It should be noted
that the formulae in the second and third row of (15.6) may be deduced
from those in the first row by cyclic interchange of symbols.
The equations (15.6) are called conditions of compatibility of Barre de
—
Saint-Venant (1797 1886) , since they were first discovered by him
(in fact, he lectured about them to the Societe Philomathique in 1860 and
published the relations in 1861).
These conditions are the mathematical form of those relations which
must be satisfied by the components of strain in order that deformation
may take place without discontinuities (cf. the earlier part of this section) ,
and for this reason they are also sometimes called conditions 。/ con¬
tinuity.
50 I. FUNDAMENTAL EQUATIONS §15

Provided these conditions are fulfilled, the formulae (15.4) give com¬
pletely defined expressions for u, % w which do not depend on the choice
of the path of integration, and it is easily verified directly that dis¬
placements found in this way actually satisfy the equations (15.1).
Further, the constants
徇，％，缈o, %,%,夕o
remain quite arbitrary, as had been anticipated previously. As can be
seen from (12.8), variations in these constants will only cause rigid
displacement of the body as a whole. In particular, if
— Cyy = . . . = Cxy = 0
throughout a region, one obtains, putting for simplicity %。= = z() = 0
and omitting the subscripts of xlf 夕卜 z】，
= 价 + 夕冰 一 小％ o = % + %% %Z, 沙 = 十
i.e., only rigid body displacement.
— 一 0产,

Hitherto it had been assumed that the region V was simply-connected.

Consider now cases of multiply-connected regions, i.e., of regions inside
which there exist closed contours which cannot be shrunk into one point
without cutting them apart or taking them outside V. As an example
for a multiply-connected region
one may consider a torus, i.e., a
body obtained by revolving a circle
about an axis lying in its plane
but not intersecting it (Fig. 9).
A multiply-connected body
becomes a simply-connected one,
if one introduces suitable cuts
(for more detail cf. Appendix 2.) ;
for example, in the case of
the tore it is sufficient to cut
it at one of its meridional circles,
shown in Fig. 9. Everything said above will apply to the region cut
in this manner. In fact, provided the compatibility conditions are satisfied,
the components u, v, w, determined by (15.4), will be single-valued
functions of the coordinates of the point 凶式句,％,z, ; in addition, it must,
of course, be assumed that the path of integration does not leave
the cut region, i.e., that it does not intersect the cut. Further, when con¬
necting the point to any point of the cut, the quantities %,巩 m will,
CHAP. 2 ANALYSIS OF STRAIN 51

generally speaking, have different values depending on the side from

which the point on the cut is approached.
Let u+, o+, w+ and u~, v~, w~ be the values of u, v, w, when a point on
the cut is reached from one or the other side respectively. The condition
of compatibility of deformation for the body as a whole will only be satisfied,
if in addition to (15.6) the following conditions are satisfied on all
cuts, introduced in the body to make it simply-connected :
=比 = 0-, 彼+ = 徵一. (15.7)
When (15.7) is not satisfied, discontinuities will occur at the above-
mentioned cuts and even parts of the body may penetrate each other
in these places.
It is clear from what has been said that, if (15.7) is not satisfied and if
the functions u, 也 w are still determined by use of (15.4) in uncut regions,
i.e., if one admits intersection of the cuts by the path of integration, then
u, v, m will be multi-valued functions of xv y1} i.e., after travelling once
around certain closed contours the functions u, v, w will not revert to
their original values ; it is easily seen that this may only happen in the
case of contours which cannot be shrunk to a point by a continuous process
(cf. Appendix 2).
The first to comment on the above was J. H. Michell [1]. A. Timpe [1]
indicated for the case of the plane problem of the theory of elasticity
the possibility of a physical interpretation of multi-valued displacements.
For the general case of three dimensional problems the question of the
meaning of multi-valued displacements was studied in detail by V. Vol¬
terra in a number of publications ; a summary of this work has been given
by him in his paper [1], and a short study of Volterra's results is also
contained in A. E. H. Love [1] (appendices to chapters VIII and IX)
and in P. Burgatti [1]. For the case of plane elasticity this problem
will be studied in detail in Chapter 6.
 Chapter 3

THE FUNDAMENTAL LAW OF THE THEORY OF ELASTICITY.

THE BASIC EQUATIONS.

Everything said in the previous chapters may be applied to any con¬

tinuous body. In order to obtain equations characterizing a body which
will be called elastic (or more correctly ideally elastic) , it is still necessary
to have a law expressing the connection between the stressed state of
the body and the corresponding deformation.

§ 16. The fundamental law of the theory of elasticity (ge¬

neralized Hooke's Law). The first, very incomplete formulation of
―
the law relating stresses to strains was due to Robert Hooke (1635 1702).
In 1660 Hooke discovered this law which has been named after him;
he published it in the form of an anagram in 1676 and gave the solution
of the latter in 1678. Expressing the essentials, which Hooke stated in his
law, in contemporary language one may say: "The deformation of an
elastic body is proportional to the forces acting on it". This formulation
may only be given a definite interpretation in the case when the "force"
acting on the body and the deformation connected with it can be char¬
acterized by one quantity each.
For example, if one has a long thin cylindrical rod, stretched by
longitudinal forces applied to its ends, one may assume that the force
acting on the body is characterized by the given value F of the applied
traction and the deformation by the extension of the rod. In this case
—
Hooke's Law gives M C.F, where C is a constant depending only on
the original length I, the form of the cross-section and the material of the
rod. Actually, it will be shown later that C = 〃石S, where S is the area
of the cross-section and £ is a constant depending only on the material
of the rod. Many similar examples could be quoted here.
Experiments have verified that Hooke's Law agrees well with the
behaviour of many solid bodies, provided the deformations are sufficiently
small. For finite deformations the law of proportionality fails to give
even approximately correct results.

52
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 53

However, also in the case of small deformations, when the law of pro¬
portionality may be assumed to be valid, Hooke's Law as introduced
above may not give the complete picture of what actually takes place
in the deformed body. Indeed, it has been seen that the state of stress
and strain is characterized by six quantities each, and that these quantities
change from one point of the body to another, so that in actual fact one
is dealing with an infinite number of quantities characterizing the state
of the body as a whole.
For example, in the case quoted above "only” the tensile forces F
acting on the ends of the cylindrical rod have been considered. In actual
fact, the "force" F expresses only the resultant effect of the external
stresses applied near the ends of the rod. These stresses may be distributed
in any manner whatsoever, for example they may be spread over the
end-sections or over parts of the side surface in the neighbourhood of the
ends; the distribution may be uniform or non-uniform, etc.
It is clear that the distribution of stresses and strains inside the rod
depends largely on the distribution of those external stresses. It is only
in the case, when the dimensions of the cross-section of the rod are small
compared with its length, that the manner in which the external forces
are distributed near the ends has no great effect on the state of the
rod (and then only in parts away from the ends) . Under these circum¬
stances consideration may be limited to the resultant "force” F (cf.
also § 23).
Thus it is obvious that, if one does not want to limit oneself to a
crude and superficial investigation, one has to generalize Hooke's Law.
The most natural generalization of a law of simple proportionality of
two quantities will be a law of linear dependence between several quanti¬
ties. Hence consider as the generalization of the original law the following
fundamental law o/ the theory of elasticity or generalized Hooke's Law:
The components o/ stress at a given point of a body are linear and ho¬
mogeneous functions of the- components o/ strain 0/ the same point (and vice
versa).
Of course the above statement refers to small deformations. (As regards
the limits of applicability of Hooke's Law, cf., for example, R. Gram-
mel [1]). The generalized Hooke's Law in this form was first stated by
A. L. Cauchy in his memoir of 1822. In subsequent work, published in
1828, Cauchy deduced this law, basing it on molecular theory, under
a simple supposition referring to the interaction of forces between molecules
considered as material points. The same result was obtained by S. D. Pois-
54 I. FUNDAMENTAL EQUATIONS § 16

—
son (1781 1840) by an analogous method in a memoir delivered to the
Paris Academy in 1828 and published in 1829.
It is not proposed to present here the deduction due to Cauchy and
Poisson, the more so because it has been found to be insufficient (cf.
below), but the generalized Hooke's Law will be accepted as the foun¬
dation of the present theory, based on the fact that for small deformations
this law agrees sufficiently well with reality for very many materials.
Before going further the following remark should be made. Since
generally stresses and strains are different in different parts of a body,
it is only possible to discuss their components at a given point.
However, the expression "at a given point" will be interpreted in a dif¬
ferent manner according to whether it is applied to components of
strain or stress. For example, when stating that exx is a function of the
coordinates %, y, z, this will always refer to the position 夕，z) of the
point before, deformation. The same will be true with regard to the com¬
ponents of displacement 孙，v, w. On the other hand, when it is said that
Xx is a function of %, y, z, this will refer to the position (居 y, z) of the point
in the final (i.e., stressed and hence deformed) state of the body.
However, for the small deformations considered here this distinction is
not essential, since, for example, the values of Xx at (%,%,zj and (%,y, z),
where (居 y, z) is the position of the point (%,%,4) before deformation,
differ by an amount which is small compared with Xx. Thus the value
of Xx at a given point (巧，乃，zj of the deformed body may be replaced by
its value at (阳 y, z). In the sequel the values of all functions considered
will be taken at (geometric) points representing the original positions of
the points of the deformed body.
Accordingly, in the sequel, when speaking 。/ a region 『 occupied by a
body or of its boundary S we will always have in mind the region occupied
by the body before deformation and its boundary.
Now consider the generalization of Hooke's Law. It may be written in
the following manner. If Xx, Yy, Zz, Yz, Z#, Xy are the components of
stress at a given point of the body and exx, eyy, esz, eyz) ezx> exy the compo¬
nents of strain, then
X# = + '12%" + ，六分2 + 2ci心" + 2c境久上 + 2c场3力
y* = ^21^xx + + ^23^zz + 2c246,2 + ^25^za; + ^26^xv>
Zz = ， 31%工+‘32%!/ + ^33^zz + 2c34%? + 2c356?^, + 2c36%“， /

丫？ =‘41%工 +‘42%”+ ^43^23 + 2c44%z 十 2c^&zx + 2,C^&xyt

，
Zr — 51%工 十 ^52^w 十 ^53^zz + 2c54%? + + 2c56%@，
， ，
X T = 61分8+ 62分3/+ 弥%:i + + 2c65分0; + 2c66%少，
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 55

where the factor 2 has been introduced for the sake of convenience
(cf. below) .
Since on the basis of the adopted fundamental law the components
of strain must likewise be definite linear functions of the components of
stress, the preceding equations must be soluble with respect to exx, ..
分 力
i.e., the determinant of the coefficients c订 must be different from
zero.
The quantities c订 are constants characterizing the elastic properties
of the body at a given point. They are called elastic constants. The term
''constant” must be understood in the sense that these quantities do not
depend on the values of the components of strain and the corresponding
stresses at a given point. However, they may vary from point to point of
the body. If that is so, the body will be said to be non-homogeneous (as
regards its elastic properties) . On the other hand, if the elastic constants
are the same for all points of the body, it will be called homogeneous.
The formulae (16.1) are seen to contain 36 elastic constants. But by
considerations based on the law of conservation of energy and on a study
of the potential energy of deformation, it may be shown that the following
relations must hold between these constants :
G? = c* 0, 7 = 1,2, . . . , 6),
in other words, the array of coefficients is symmetric. Thus in the
most general case the number of elastic constants may be reduced to 21.
Application of these considerations and deduction of the stated result
was first given by G. Green in 1837 whose paper on the subject was
published in 1839. A more complete foundation for this result, based upon
the first and second law of thermodynamics, was presented by Lord
Kelvin (W. Thomson) in 1855. (For more detail see A. E. H. Love [1]).
It will be seen in the next section that in the case of the isotropic
body the number of elastic constants may be reduced to two.
By the old theory of Cauchy, based on the consideration of molecular
forces, the number of elastic constants in the most general case is equal
to 15, and not 21 ; in the case of the isotropic body one has by this theory
only one elastic constant (in the first of his memoirs, where Cauchy
did not rely on molecular theory, he obtained two constants for the iso¬
tropic body). Poisson arrived at the same results. However, this was not
confirmed by experiments. But it should not be thought that the mole¬
cular theory led to the wrong results and that it is impossible to obtain
from it the correct number of constants. The point is only that Cauchy
56 I. FUNDAMENTAL EQUATIONS §17

and Poisson applied molecular theory in an oversimplified form. Using

modern concepts of the structure of materials one can obtain the correct
result, i.e., 21 constants. This has been done recently by M. Born [1]
(cf. also A. E. H. Love [1], Note B at the end of his book).
It is not proposed to give here further details of these problems, since
in the sequel only isotropic bodies will be considered. In that case definite
formulae may be deduced by means of very simple considerations.

§ 17. Isotropic bodies. As mentioned earlier a body will be called

isotropic, if its properties are the same in all directions. In other words,
if one cuts a volume element of a definite shape (say a cube) from an
isotropic body, this element will not differ from any other element of the
same form (cut from the same part of the body) but orientated differently
from the first. For example, wood is not isotropic, since a beam cut in the
longitudinal direction (along the fibres) differs very much from a beam
cut across the grain. Likewise all crystalline bodies are anisotropic. In
nature there are no ideally isotropic bodies, but many materials, important
in industry, may with sufficient approximation be assumed to be isotropic.
Many such materials (e.g. metals) consist of small anisotropic parts
(crystals) arbitrarily placed with respect to each other. It is this random
distribution which is the reason that bodies of not too small dimensions
made from these materials may be considered to be isotropic.
A body will not only be called isotropic, but also homogeneous, if
the properties of volume elements cut from different parts of it are the
same. It should still be noted that a body which is isotropic and homo¬
geneous with regard to one property may be anisotropic or non-homo-
geneous with regard to others.
In the following only isotropic and homogeneous bodies will be considered,
where it must be understood that this isotropy and homogeneity refers
only to its elastic behaviour.
In mathematical language this fact may obviously be expressed in
the following manner: the coefficients c6Q in (16.1) do not depend
on the orientation of the coordinate axes with respect to the body nor o% the
position of the point under consideration in the body. Owing to this property
the above-mentioned formulae take a very simple form, as will now be
shown.
First of all it is easily proved that at every point of an isotropic body the
principal axes of strain and stress coincide. In fact, let the principal axes
of strain at a given point lie along the coordinate axes. Then
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 57

^yz = %* = exy = U.
By the generalized Hooke's Law one has, in particular,
z 力分① + Be?” + Cezz, (a)
where A, B, C are constants. Introduce now a new coordinate system
Ox'y'z' , obtained from the old system by a simple rotation of 180° about
the axis Oz. The axis Oz' of the new system will coincide with Oz, while
0%', Oy' will be in the opposite directions to Ox, Oy. Since the coefficients
A, B, C are not to depend on the choice of axes, one will have in the new
system
丫2‘= 乂侬少 + + Ccz，zr . (6)
But obviously
侬疗 = %*，%'a' = £”， = ^zz，匕：' = 丫2，

Comparing (a) with (b) one sees that one must have
月分工 十 月 Be” Cezz,
and hence *
A= B — C = 0.
way it can be proved that
But this means that Yz = 0. In the same
Z& = Xy = Ch
However, this shows that the coordinate axes are principal axes of
stress and the above statement is proved. Thus, in future, it will be un¬
necessary to distinguish between principal axes of strain and stress ;
they will simply be called principal axes.
Let it still be assumed that the coordinate axes coincide with the prin¬
cipal axes. By the generalized Hooke's Law one may, in particular, write
X = ^xx + 方%旷 + C^zz，
工

where a, b, c are constants. Let Oxfyfzr be a new system of axes obtained

from Oxyz by a rotation of 90° about the axis Ox. In the new system one
must again have
X8，= 夕分 十 加”, 十 ^ez'z'

* We use here the following argument which, we assume to be obvious (or

proved by experiment) : the strain components exx, ... , at a given point may
assume arbitrary (of course, sufficiently small) values for a suitable choice of
external forces acting on the body element under consideration.
58 I. FUNDAMENTAL EQUATIONS § 17

But obviously in the present case

= Xx, &X'x' = ©*， = ^zz， ^z'z' =
and hence
= 口40；
+ 加2 + 06期0。

Comparing this formula with the earlier one for Xx one sees that b = c.
Thus
— aexx + ，伉
Xa； y + %Z) = 十 eyy + 6zz) + 一 方"皿

Finally introduce the notation

方 = 入， a b — — 25z,
so that the preceding formula becomes
Xw = 入(％工 + 十 6zz) + = 入9+ 2及分m
where 8= 十明十
Because of isotropy one can obtain from the above formula for Xx
those for Yy> simply by replacing the letter % by y or by z. Conse¬
quently, one finally finds
N0 = 入6+ 235，N2 = 入0+ ^3 = 入6 十 23分。
In these formulae N2, N3 and %,%，/ denote the principal stresses and
strains. The corresponding coordinate axes will now be denoted by
0%', Oyr , Oz', where it should not be forgotten that they are principal axes.
In order to find now the formulae relating the stress components
Xg ... , to the strain components exx, .. exv in any coordinate
system Oxyz, it is sufficient to express the quantities Xx, . .. , by the
known transformation formulae for the transition from one system
of axes to another in terms of 7vp N2> N3. Using (c), this will give
expressions for Xx, ... , in terms of 勺，％, %. Expressing, finally,
勺，％， in terms of exx, one finds the required formulae. Actual
execution of this process leads to unwieldy calculations which may be
avoided in the following way.
One can replace the set of formulae (c) by a single equation which is ob¬
tained by multiplying the equations (c) by 之丐 »2, 不2respectively, where
—
>

V, 娟，烂 are the components of some arbitrary vector P in the system

0%'y'z' , and by adding them:
ME'2 + N/ + N*2=
= 入9(中 + 娟2 + d + 2"(第2 +
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 59

Now transform from the axes Ox^'z' to the axes Oxyz. It is known that
the quadratic form
Ne2 + 5 十
will then become the quadratic form (cf. § 5)
十 ... + 2X 高,
and
e百z + %城 2 + %已'2
the form (cf. § 13)
分发2 十 分" + + 2%z优 24或 + 2%后.
Here 已，叮，已 are the components of the vector P in the system Oxyz. But
obviously
可 +d+ = ? 十噌+ 0
As regards the quantity
e = 4 + % 十 ％,
its value in terms of the components for the new axes will be
9 = exx + e”1f + ezz
(cf. end of § 14). Hence in the new coordinate system equation (d) becomes
X + 匕哦 + z<2 + 2y冰 + 2ZJE + 2X舟 =
= 入职2 + 哦 + + s + e以2 + e/2 + 2公h 十 2%殳 十 2分舟).
But since this equation will be true for the components of any vector
1
>'

P, i.e., for all values of E, 刁，J the coefficients of ^2, . . . , 卫刀 on either side
of the equation must be equal, and hence
x h e
入 - 匕 = 入9 十 2sy, Zz = 入6+
v (]7 ])
, Z x = 2fle皿 Xy = 2^xv,
where 0 = CXX 十 Cyy + &za (17.2)
is the cubical dilatation.
Formulae (17.1) give the unknown relations between the components
stress and strain in an isotropic body. The quantities 入，出 are constants
characterizing the elastic behaviour of a given body *. This notation was
* The formulae (17. 1) obviously remain also true in the case when the body
under consideration, while isotropic, is non-homogeneous. Then the quantities 入
and will themselves be functions of the coordinates x, y, z of the point under
*
consideration. Certain of the formulae and propositions derived below likewise
remain in force for non-homogeneous bodies, as the reader will readily verify.
60 I. FUNDAMENTAL EQUATIONS § 18

—
introduced by G. Lame [1] (1795 1870) and for this reason they are called
the constants of Lame. They have to be determined for every material
by experiment, but in actual fact other quantities, in terms of which
these constants are easily expressed, are more suitable for direct measure¬
ments, and that is the procedure normally adopted.
By a condition, stated during the formulation of the generalized Hooke's
Law, the equations (17.1) must be soluble for eXX} .. Consider what
conditions must be satisfied by 入 and so that the above demand is
satisfied. For this purpose (17.1) will now be solved for the components
of strain. Adding the first three equations, one gets
X# + 匕十 Zz = (3入 十 2fi)9 = (3入 + 2jx) {exx + evy 十 ez^. (17.3)
This equation can be solved for exx 十 eyy 十 ezz only if 3入 + ¥ 0.
Further, solving the last three equations of (17.1) for eyzi ezx, exyt one finds
that one must have r # 0. It will be seen in § 19 that for all actual bodies
入 > 0, pi > 0. Assume now that these conditions are satisfied. Sub¬
stituting the value for 6, obtained from (17.3), in (17.1) one finds the
formulae
十
e = .x (Y 4 Z )
""十

_
[z(3入 2(z) ” 2以3 十 2a) 入

以3入
M y
2团" 2M
入
3入 十 2[i)
,
(Z石 4
十 X川
) (17.4)

入+ (1 X y内]
一 ?z I “十]
以3入 2(1) 2M 3入 十 2凶
1 1 1
% ^ZX ~ X 财,

2M 2“
expressing the components of strain in terms of the stress components.

§ 18. The basic equations for the statics of an elastic isotropic

body. It is now possible to write down the complete system of equations
for the statics of an elastic body. This system consists of the "equilibrium
equations", relating the stress components (§ 4), and of the equations
(17.1), relating stresses to strains. It will be shown in § 20 that the fol¬
lowing equations constitute a complete system :
常 叫 axz
-
ox —
5
砂
1 —oz [- X = 0,
立 八

也 ayy 阴
+y
蜷十城 “ = °， 八

三
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 61

以 以 琢
&
十
砂 + CZ
+ Z = o, (18.1〉

Xx 入9十 Yy —
X6 2以分+ Zz = 入8 + 2^22,
财，
(18.2)
y2 = 2及692, Zw = 2,^ZX) Xy ~ 2,]l&Xyt

(18.3)

where u, v, w are the components of displacement and

du 加 3m
9 = %工 + + =菽 +可+ (18.4)

These equations must still be supplemented by the formulae giving the

components of the stress vector acting on a plane with normal % (§ 3) :
—
Xn Xx cos (%,%) + Xy cos (笈，y) + Xz cos (知 z),
Yn = Yx cos (n, %) + Yy cos (肛 y) + Y2 cos (处, z), (18.5)
Zn = Zx cos (n, %) + Zy cos (见, y) + cos (处, z).
Next, a general remark will be made with reference to the sets of
equations (18.1) and (18.2). These equations are linear and homogeneous
in the displacement components u} % w, the stress components Xx, ... ,Xy
and the body forces X> Y, Z. Hence, if
优，％ 也 X；, …， X；and %", z", 如"，X：, …，X；
are two solutions of (18.1) and (18.2) corresponding to body forces X' ,
Y', Z' and X", Y", Z" respectively, then
〃= %' 十 ％", V == V* + Vr，, 3 = »' + 彼"，
x,= + x； x：, …， x” = K K
+
(18.6)

is a solution of the same system of equations for the body forces

X = X' + X", Y= r + Y", Z = Z' + Z". (18.7)
It will be said that the solution (18.6) has been obtained by super-
imposition of the two given solutions. Formulae (18.5) show that the
external stresses, applied to the surface of the body (for this purpose n
refers to the outward normal) and corresponding to the last solution,
62 I. FUNDAMENTAL EQUATIONS §19

are given by the sums of the surface tractions corresponding to the given
solutions. In particular, if %", "， X* …,X：is any solution when
there is no body force (X" = Y" = Z" = 0), then (18.6) will satisfy the
same equations with the same body forces as the solution %', % w',
y' y'

§ 19. The simplest cases of elastic equilibrium. The basic

elastic constants. Before going further, several very simple cases of
elastic equilibrium will be considered for the purpose of studying the
physical meaning of the constants characterizing the elastic properties
of a body.
First it will be noted that in the absence of body forces, i.e., if
X=Y = Z = 0, (19.1)
the static equations of the elastic body may be satisfied, in particular, by
assuming the strain components exx, to be (arbitrary) constants,
i.e., by assuming homogeneous deformation. In fact, by (18.2), the
stress components will likewise be constants and hence the equations
(18.1) will be identically satisfied (since by supposition X = Y = Z = 0).
Further, the compatibility conditions of St. Venant (§ 15) will be satisfied,
since one may always find displacements u, 内 w corresponding to the
given strain components. In this simple case the above may be proved
directly by finding expressions for the displacements ; namely, direct
substitution shows that the displacements
每 = exxx + exyy + exzz + qz —ry a,
。= 分/ + eyyy + %病 + rx — pz + b, (19.2)
切 = 分修 + ^zvy + + 6y 夕％ + c,
satisfy for constant exx, . . exy the relations (18.3). Here a, b> c, 夕，q, r
are arbitrary constants ; the corresponding terms express therefore
rigid body displacement [these formulae could also have been written
down immediately using (12.12)]. By § 15 the solution (19.2) is the only
possible one for the given exx, . . exy.
In exactly the same way it is obvious that the static equations may be
satisfied by putting the stress components equal to arbitrarily chosen
constants. In fact, (18.2) gives then definite constant values for the strain
components and one obtains again the preceding case.
Now certain very simple particular cases will be considered. First put
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 63

Xx = T = const, Yy = Zz= Y z == Zx ~ Xy = 0. (19.3)

Then, by (18.2) or by (17.4),

£工，
"
= = = T, (19.4)
；i(3入2r) 2M3入 + 2
eyz = 分6= %?.= (19*5)

It will now be assumed that the body under consideration is a prism

or a cylinder with generators parallel to 0% and with ends perpendicular
to this axis. Then it is obvious from (18.5) that on the side surfaces
= 0,i.e., they are free from surface tractions. On the end
facing in the positive direction of the axis Ox 丫勿 = = 0, Xn = T,
and on the other end
Yn = Zn = 0, Xn = — Xx = — T.
Consequently the external forces acting on the cylinder are uniformly
distributed over the ends and produce tension, if T > 0, and com¬
pression, if T < 0. The quantity T denotes the tensile or compressive
traction, exerted per unit area of the ends. Now the obvious assumption
(which may be based upon experimental evidence) will be made that for
these conditions and for T > 0 the cylinder extends in the longitudinal
and contracts in the transverse direction, i.e., for T > 0 one must have:
> °, ^yy V °, ©zz V
Then, by (19.4),
…
二入
入
“
+J
[X
> 0，
入
> o. (19.6)
lz(3 十 2ji) 2e(3入 2科)
Therefore, in particular, (入 +卜) #0；further, it follows from these in¬
equalities (dividing one by the other) that

2( 入 + 万
>
"
Introduce the notation
门 以3入 + 入
(19.7)
入 N 2（入
On the basis of the above the quantities E and a are positive for all ma¬
terials. The quantity E is called modulus of elasticity or Youngs modulus
(Th. Young 1773—1829) and o Poisson's ratio. The physical meaning of E
64 I. FUNDAMENTAL EQUATIONS § 19

is obtained from the first of the formulae (19.4) which gives

7= Eexx. (19.8)
Thus 石 is the ratio of the applied stress to the strain caused by it in the
longitudinal direction. The physical meaning of e follows also from
(19.4) which show that
I I _ I I
，■ ， ' >

,
k
u (19.9)
I I I %幻 I
i.e., the ratios of the transverse strains to the longitudinal strain are a
constant quantity which does not depend on the shape of the cross-section
of the rod nor on the magnitude of applied traction.
Next consider another particular case. Let
Yz = T = const, Xx = Yy — Zz ~ Zx = Xy = 0. (19. 10)
Then, by (18.2),

^yz =
2K
T, exx — Cyy = ezz = ezx = ezy = 0, (19.11)

i.e., the corresponding deformation is pure shear. If the body under

consideration is in the undeformed state a right parallelepiped with sides
parallel to the coordinate planes, then
it is easily seen from (19.10) that the
sides perpendicular to the axis Ox are
free from surface traction. The tractions
applied to the other sides lead to the
shearing forces shown for the case T > 0
in Fig. 10, where only a cross-section
in a plane parallel to Oyz is drawn. The
angle between the sides originally paral¬
lel to Oxy and Oyz is shown to differ
from a right one by zyz = 2eyz (cf. § 12).
Fig. 10. Hence, by (19.11),
T = %z・ (19.12)
Thus p, is the ratio of the shearing stress T and the corresponding angle
of shear. For this reason is called the shear modulus.
Finally consider the case
— —
Xx Yv = Z? = 0 = const, Yz = Zx = = 0. ( 19.13)
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 65

In this case (18.5) shows that the stress acting on any plane with normal
% is given by the formulae

Xg = — 少 cos (笈，幻， Yn ~ — /> cos (w, y), = — cos (扬 z),

expressing that the stress vector is parallel to the normal and that its
magnitude is ［力 Hence only a normal stress acts on any plane ; if one
assumes 夕 > 0, the stress will be compressive. The surface of any
part of the body under consideration will only be subjected to uniform
normal external pressure ("hydrostatic pressure").
Adding the first three formulae of (18.2) one finds
夕= ——（入 + 条2）6.
Since 0 is the cubical dilatation (and consequently
compression) , the quantity
— 0 is the cubical
儿 = 入+和 (19.14)
is called the 碑。&四2%$ of compression or bulk modulus. The obvious as¬
sumption will be made (which may be based on experimental evidence)
that for 力 > 0 a decrease in volume actually takes place, and hence
that > 0 for all materials.
In addition to 入 and the following constants have been introduced
*
in the above work : the modulus of elasticity E, Poisson's ratio a, the modu¬
lus of compression k. The quantities 入 and q may be expressed in terms of
any two of these constants. For example, solving the equations (19.7)
for 入 and 闷 one obtains
备 E
(19.15)
+
产
6 —26 2(1 十
and substituting these expressions in (19.14)

— 3(1 —2b) .
E
k ~ (19.16)

The last formula shows that one must have for all materials
o < |. (19.17)
The formulae (19.15) show that
入 > 0, fi > 0
which is now also obvious on physical grounds (cf. § 17).
Note that by the old theory of Cauchy and Poisson for all bodies
g= or, what is the same thing, 入 = 内 But this is not confirmed by
experiment. However, for many materials, a has approximately the same
value of J (and not |).
66 I. FUNDAMENTAL EQUATIONS §20

If one introduces in (17.4) the constants E, o instead of 入, 函 the formulae

take the simpler form

exx = 亏 [X# — 匕 + Z,],

o(

分 0=h[Y夕一(7区 十 X4)],
jC
(19.18)

—
1
= 石 [Zz “Xg + y/,

NOTE. In the literature one often finds the quantity m — — which

a
is called Poisson's coefficient (eg R. Grammel [1]). The shear modulus 出
is often denoted by G. Recently determined values of the above
constants for different materials may likewise be found, for example, in
GrammeFs book.

§ 20. The fundamental boundary value problems of static

elasticity. Uniqueness of solution. Consider now the basic equations
of the static elastic body (§ 18) which will be written in the form
% ax. ax,
+ X = 0,
十可 3z
也 1 ar. * + y = o.
+石
(
(20.1)
a% 效
az,
卜" C +17 十 z = 0,
加 Sv
X, = 入9+ 2以
c

拓
Yy = 入 十 6 2“ — oy
, Z? = 入6 十 2(1-—,
cz
(20.2)
为一院十
”
匕= ，

where
du 加 浏
H = H ~ -4- ' .
dx 砂 dz
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 67

The nine equations (20.1) and (20.2) contain just as many unknown
functions u, w, Xx) . . . , Xy. The system (20.1) and (20.2) has earlier
been called the complete system of static equations of the elastic body. In
order to prove this statement, it has to be shown that the system (20.1)
and (20.2) completely determines the elastic equilibrium of the body, if
the external forces to which it is subject and the "internal'' body forces
are known.
It has been assumed here that the elastic equilibrium of a body is
known, if the stress components or what is the same thing, thanks to the
equations (20.2) , if the strain components are known at every point of
the body. It should not be concluded that body forces are exclusively
external, since, for example, gravitational forces between parts of the
elastic body are "internaT' body forces. The external forces mentioned
above comprise, firstly, external body forces and, secondly, external
tractions applied to the boundaries of the body.
In connection with all this there arises the first fundamental
boundary value problem:
I. Find the elastic equilibrium of a body, if the external stresses acting o%
its boundaries are given. Here, as in all the following work, it will be as¬
sumed that the body forces are given once and for all.
In practice, this last point arises in the following manner : body forces acting on
a body element depend as a rule on the mass contained in it and on its position
with respect to other masses (eg gravity forces, centrifugal forces due to rotation,
etc.). Under deformation the position of the element as well as its density will
change, so that the body forces (X, Y, Z), referred to unit volume, generally speak¬
ing will also vary. But in view of the smallness of the deformations and displacements
these variations are insignificantly small and may be disregarded.

With (20.1) and (20.2) in mind, this problem leads to the following
one: Find functions %,巩 w, Xx, . . Xy, satisfying (20.1) and (20.2) in
the region V originally occupied by the body (cf. § 16), and, in addition,
satisfying on the surface (boundary) S of the body the following boundary
conditions [cf. (18.5)] :
Xx cos(% 幻 + Xy cos (% y) + 太名 cos 例，z) = flf
Yx cos (肛 幼 + Yy cos (%,y) 十 Yz cos (% z) = /2, (20.3)
cos (知， + Zy cos (n, y) + cos (见 z) = /3,
where n denotes the outward normal to S and 々，/2, /3 are functions, given
on the boundary (and representing the components of the known stress
vector acting on the surface of the body).
68 I. FUNDAMENTAL EQUATIONS §20

In addition to the first fundamental problem stated above the

second fundamental boundary value problem is
of considerable interest :
II. Find the elastic equilibrium of a body, if the displacements 0/ the points
0/ its boundary are given. Physically this corresponds to the case when, by
means of suitable tractions applied to the points of the surface, these
points are subjected to known displacements and the boundary is cor¬
respondingly deformed. In relation to the equations (20. 1 ) and (20.2)
this leads to the determination of solutions which satisfy on the surface
of the body the following boundary conditions :
% = 0 = g2， 捌 ~ g3， (20.4)
where g?, g3 are functions known on the boundary.
Finally, in many investigations an important part is played by the
mixed fundamental boundary value problem
which arises whenever displacements are known on one part, and external
stresses on the remaining part of the boundary.
In addition to the problems stated already a number of others may be
formulated which are no less important in applications ; some of these
will be considered later, when dealing with the plane case.
In the sequel, unless stated otherwise, it will be assumed that u, v,
w are single-valued functions having continuous derivatives up to and
including the third order inside the region occupied by the body. Under
these conditions the strain and stress components will also be single¬
valued and continuous functions having continuous second order de¬
rivatives inside the same region.
Further, we will assume that the displacement and stress components
are continuous along the boundary of the region V occupied by the body.
This assumption, although not stated explicitly, was implied above, for
example, when writing down the formulae (20.3) and (20.4).
As regards the surface S of the body which represents its boundary
we will assume that it satisfies the conditions which are usually imposed,
in order to ensure the validity of the known formulae of integral calculus
which will be used below.
Having in mind the need to prove a uniqueness the-oreni, , i.e., to prove
that the system of equations (20.1) and (20.2) has one and only one
solution for each of the fundamental problems, one important lemma will
first be deduced.
Consider the double integral, extended over the surface of the body,
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 69

J= (Xnu 十 Ynv + Znw)dS, (20.5)

s
where Xn) Yn, are determined by (18.5) and by n must be under¬
stood the outward normal to £ Substituting from (18.5) into (20.5) one
finds
J = [J [p cos (%,%) 十 0 cos (%,y) + K cos (肛 z)]dS,
where for convenience
P = X*% + Yxv + Z^w, Q = X/ + y w + z抻,
r= xzu +
y戒 + Zz步.
By Green's formula

0
+■1+ 鲁）私的通

But in the present case

a% 。加 加
+ X
+匕石 +北
1r
__ 3% dw \
(豆 马 一 (法 +蒜
/ dw 加、 /加 加、
Z, /
十 元
(- + —
十 X,

or, by (20.1),

where
dP
ex
dQ
] -~
cy— —
3R
-— =
cz — (X% + 丫。+ Zm) 十 2TV,

2W = 太晨现 + 丫必件 + Zzezz + 2Y zeyz 4- 2Z 4- t2Xyexy. (20.6)

Thus one finally obtains

Jj
s
(X 楔 Ynv + Znw) dS
v
(X% + Yd 十 Zw) dxdydz
err
—
— 2
Ijj W dx dy dz. (20.7)
70 I. FUNDAMENTAL EQUATIONS §20

The expression W in this formula represents, as will be proved in

§ 24, the potential 。夕 strain energy per unit volume ; but at the moment this
is of no importance.
Introducing on the right-hand side of (20.6) the expressions (18.2) for
the stress components in terms of the strains one finds
2W = 入(n + &夕 + %)2 + 2“(邑;+ + dz + 2% + 2dH 十 2瑶J, (206)
which proves that W is a positive quadratic form involving the components
of strain, and, in addition, that it is definite, i.e., that it becomes zero, if and
only if all the strain components are zero ; this follows from the fact that 入
and m have already been shown to be positive quantities.
Similarly, W may be expressed in terms of the stress components ;
obviously it will again be a positive definite form in these components.
Let it now be assumed that one of the earlier stated problems has
two solutions. Let 〃'， m', X；, . .. , X； be the components of dis¬
placement and stress corresponding to the first solution, and u", wn,
Xj, …，Xj be the analogous quantities of the second solution. Form
the "difference” of these two solutions, i.e., put
〃= = —
Obviously (cf. § 18) the functions ut v, w, Xx, ％ satisfy the same
equations (20.1), (20.2) in which one has only to put
X= y = Z = 0;
in other words, the "difference” solution satisfies the basic equilibrium
equations in the absence of body forces. Thus, by (20.7) for X= V = Z = 0,

(Xnu 丫弱 Z典)dS = 2 W dxdydz. (207)

s 厂
Now the following will be noted: In the case of Problem I the quantities
Xnt Ynf Z*, formed by subtracting the two solutions, will be zero on S,
since both solutions, by supposition, satisfy the conditions (20.3) for
the same functions flf /2, /3. Hence
Xn = Xx cos 依v) 十 Xv cos {n, y) + cos (%,z) = 0,
L = 0, = 0.
In the case of Problem II one will in the same way find % = = w = 0
”
on S. Finally, in the case of the mixed problem, 劭， w will be zero on one
part, and X”, Yn, Zn on the remaining part of the boundary. In all three
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 71

cases the expression Xnu + V洲 + Znw is zero on S. Hence (20.7')

becomes
W 4% dy dz = 0.

However, since 印》0, the above relation is only possible, if W = 0

at all points of V. It has been seen earlier that this condition implies
exx = eyy = ezz = e” = ©zu = %? = 0 throughout the body. But exx =
— ^xx etc., where 北，…，虞 and …，战 are the components of
strain, corresponding to the two solutions under consideration. This
means that both solutions give identical strain components, and con¬
sequently also identical stress components. Hence both solutions are
identical in the sense that they give an identical state of stress (and de-
fonnation). This proves the uniqueness theorem. [The theorem and the
proof given here is due to G. Kirchhoff ( 1 858) .j
However, it should be noted that the displacements may not be com¬
pletely identical. In fact, from the vanishing of exx, .. exy does not
—
follow that u v = w = Q, but only
—
= 口 + gz ry, v — —
b rx 力z, w = c 十 仑y 夕％, (20.8) —
(where a, b, c,自 q> r are constants) expressing rigid body motion. Thus,
when solving the first fundamental problem, one will always obtain the
same stresses (and strains), but one may find for the displacements
values, differing from each other by terms expressing rigid body motion.
This could, of course, have been predicted, because such displacements
do not affect the stresses and deformations. Such differences in the
solutions are, however, unimportant.
In the cases of the second and the mixed boundary value problems such
differences cannot occur, since the displacements are given beforehand for
the whole or part of the boundary.
Finally note the following proposition which is a particular case of the
uniqueness theorem proved above : If the body forces are zero and if, in
addition, either 疯) the external stresses or b) the displacements of points

The above-mentioned proofs of existence are given in a great number of original

publications. Reference will be made here to only a few of these. For the secon^
fundamental -problem : I. Fredholm [1], G. Lauricella [1, 2], A. Korn [1, 2], L. Lich¬
tenstein [1], D. I. Sherman [21]. For the first fundamental -problem : A. Korn [3],
H. Weyl [1]. Note that as a rule the first problem of this book is called in literature
the second boundary value problem and vice versa.
72 I. FUNDAMENTAL EQUATIONS §20

of the boundary or c) the external stresses on one part and displacements

on the remaining part of the boundary vanish, then the stresses throughout
the body are zero (and hence there is also no deformation) .
The above proof of uniqueness of solution holds true for simply as
well as for multiply connected bodies, because at no stage has the as¬
sumption of simple connectivity been introduced *. However, the hy¬
pothesis that the components of displacement are single-valued functions
of the coordinates is essential for the proof. As has already been stated,
in the case of multiply connected bodies one may admit also the existence
of displacements which are not single-valued. For such a generalized
study of the problem the above uniqueness proof loses its validity and
the theorem is no longer true. For a physical interpretation of this case
see Part II of this book.
Note that only the following has been proved : if the fundamental
boundary value problems of elasticity have a solution, then that solution
is unique. But this, of course, it not a proof of the existence of such
solutions. The proof of the existence of a solution is much more difficult
than the uniqueness proof and it requires application of the most powerful
methods of modern analysis. This explains the fact that the proofs of the
existence of solutions of the fundamental problems have only been found
comparatively recently.
The scope and character of the present book do not allow a general
treatment of these problems. Therefore it will only be stated here that
the existence of solutions of the first and second fundamental boundary
value problems has been proved recently with full mathematical rigour
under sufficiently general conditions. The proof for the plane case will
be given in Part V of this book.
For the existence of a solution of the first fundamental problem ob¬
viously the following condition must be satisfied : the resultant vector
and moment of the body forces and (known) external stresses applied
to the boundary must be equal to zero. This condition follows from the
fundamental principle of statics and may also be deduced from (20.1).
In fact, the projection of the resultant vector of these forces, for example
on the axis Ox, is

V
Xdxdydz
0
s
XndS.

* This proof is applicable without any changes in the case of non-homogeneous

bodies, i.e., when 入 and are functions of the coordinates x, y, z (cf. footnote p. 59).
“
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 73

But, as has been shown in § 4, this expression is equal to

ax,ax;、
十方上+守上+一) ，，
为
，
加 砂 oz /

this triple integral, however, is zero by (20.1).

Further, the resultant moment, for example about the axis Ox, is given
by

0
(yZ
— zY)dx dydz+ (yZ — zY^ dS. 整

This expression, as shown in § 4, is equal to

v
(Z®
— dx dy dz,

where again use has been made of (20.1); but since Z"= Yzt the last
triple integral vanishes.

§ 21. Basic equations in terms of displacement components.

The system of equations (20. 1 ) and (20.2) involves simultaneously the
components of stress and displacement. However, it is possible to obtain
systems containing only one or the other type of components. It is simplest
to deduce the system containing the components of displacement. For
this purpose it is sufficient to substitute from (20.2) in (20.1) which
gives, after some obvious simplifications,
加
(X + ii) jxA% + X = 0,
ox 1-
36
(入 + m)
可
十 s” + y = o, (2i.i)
、而
(入 十 M) 不
cz
F— rAw + Z = 0,
where again c

—ex— cy— ——
々个
CU CV CW
e= ~
oz
and A denotes the Laplace operator, i.e.,
d2u 洲〃 d2u
」 •
卜 …
" —— •
' ‘

砂2 a/
Starting from the representation of the elastic body as a system of
74 I. FUNDAMENTAL EQUATIONS §22

—
material points, Navier (1785 1836) obtained in his memoir, presented
to the Paris Academy in 1821 and published in 1827, the equations which
must be satisfied by the displacements of the points of an elastic body in
the dynamic as well as in the static cases. Navier's equations for the
latter case agree essentially with the equations (21.1), if one puts in
these 入 = & The discovery of these equations may be considered one of
the most important stages in the development of the theory of elasticity,
and therefore Navier is rightly ranked among the most important of its
founders.
The equations (21.1) are very convenient, because of their symmetry
and because they contain only three unknowns.
§ 22. Equations in terms of stresses. However, it is often more
convenient to deal with equations containing only stresses. It should
not be thought that for this purpose one may limit consideration to
the equations
ax, dxy 区
3% + + 加 + X = 0,
ay 工 a% %
8% dz
十 y= o. (22.1)

zzx 游夕 zzz _
—- H + + Z = 0,
八

which have been called "equilibrium equations”

In fact, if Xm . . . , satisfy these equations, this does not mean
that these quantities express some actually possible state of stress ;
it is also necessary, in addition, that displacements 体 也 m) can be found
which are related to these stresses by (20.2). For this, on the other hand,
it is necessary and sufficient (with certain reservations in the case of
multiply connected bodies ; cf. the end of this section) that the strain
components which will now be written [cf. (19.18)]
1十 1 +ay
E
上
3
_ 尸q
E
LG
下，

1 + b Zz (22.2)
石— E
1 + or
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 75

where
® = X# + Yy + Zz，
satisfy the compatibility equations of St. Venant [cf. (15.6)].
Substituting from (22.2) in (15.6) one obtains from the formulae in the
first row of (15.6)
32 yz
+ 济®]—
洲丫t Zz 炉 a ra2©
(22.3)
&2
十
两' 1十 ( 砂2 盼 J dy &

a2o _ a
针+ 得b
济％ b r ay2 AZ a 琢］ , 、
dy dz 1 + er 砂 dz 3% 1 加
a 川

by cyclic interchange of symbols one obtains four similar relations

corresponding to the remaining compatibility conditions. Equations
(22.3) and (22.4) may be somewhat simplified using (22.1). Thus, dif¬
ferentiating the second equation of (22.1) with respect to y and the third
with respect to z and adding them, one obtains

But, by the first of the equations (22.1),

a /ax, axa ax 眸
，
3% '砂 dx / a%?
substituting this last expression in the preceding formula gives
a2yz 济x, a2y 济z% /ax ay ax
dy dz 婚 砂2 箫 \ 3% dy / 3%
which, when introduced on the right hand-side of (22.3), leads to
a
（严® 洲®） ¥（％ +么） 阳匕 + ZJ
一
〈守 十 许屋 H
项
'

a?
=— /ax ay
az、 ax
（古十 行 + 法^十2元
Finally, noting that V# + ® — Xx, one obtains
1 320 ax ,、
1 +b
A® — 环
i +b a%? ex
76 I. FUNDAMENTAL EQUATIONS §22

Adding this equation to the two analogous equations, obtained from it

by cyclic transposition of symbols, one finds a formula which is important
in itself
1+b
A® (22.5)
1 —
Substitution of (22.5) in (a) finally gives
a2©i b
/ax ay az \ ax
AC _1..

”
，， ,，

1 十 o 3%2
，， ， ' 」
： ，

1 —a
… 一 1

\
' ，，

6%
，
'卜
如
— — dz /
~ ' 11
' I m
乙 " '"

dx
'■
(22.6)

This is one of the required formulae, the other two being obtainable by
cyclic transposition of symbols.
Now consider equation (22.4). Differentiating the second equation
of (22.1) with respect to z and the third with respect to y and adding,
one obtains
济 K 济吟 洲E
a% 加 效 加效
十 —1 +
Adding this equation to (22.4) which may be written
32X* 济匕 洲Zr a2o
—
b

沏 加 — dx dz
-I— - ,
0,
八

+
1
" "― * '

dy Hz 谬 1 a dy
one finds a20 (聆Z 3V、 1
+ 1 + a cycz 】 = — ('砂
——
—/). (22.7)
The other two equations of this type may be obtained by cyclic trans¬
position.
Thus it is seen that the stress components must satisfy nine equations,
i.e., (22.1), (22.6) and (22.7) with their analogous equations. The
equations (22.6) and (22.7) were obtained by J. H. Michell [1] (pp. 112
1 13) ; for the case of zero body forces, these equations were found earlier
—
by E. Beltrami (1892). Therefore the equations (22.6) and (22.7)
with their four analogues will be called conditions compatibility of
Beltrami-Michell.
It follows from the above that, if the six equations of the type (22.6)
and (22.7) are satisfied, the strain components corresponding to the
stress components, satisfying the equilibrium equations (22.1), will
fulfill the compatibility conditions of St. Venant. Thus, the equations
(22.1), (22.6) and (22.7) with their analogues are not only necessary,
but also sufficient.
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 77

Some reservations must be made only in the case of a multiply con¬

nected body, when the displacements, corresponding to the stress com¬
ponents satisfying all the above conditions, may be found to be multi¬
valued. In that case one has to introduce either an additional condition
of single-valuedness of the displacements or to admit the existence of
multi-valued displacements which, as has been mentioned earlier, may
be given a definite physical interpretation.

§ 23. Remarks on the effective solution of the fundamental

problems. St. Venant's Principle. Solution of the above-mentioned
fundamental boundary value problems for the general case presents in
practice great difficulties, if one has in mind effective calculations. The
so-called general methods give (in the general cases) only theoretical
solutions, i.e., in the end they only prove existence of the solution.
(These general methods are given e.g. in the papers quoted in § 20).
Solution of one or the other problem is often considerably simplified
by application of St. Venant's Principle which may be formulated as
follows: If one applies to a small part of the surface of the body a set of
forces which are statically equivalent to zero, then this system of forces
will not noticeably affect parts of the body lying away from the above
region. Alternatively : If a set of forces, acting on a small part of the
surface of a body, is replaced by a system of forces (acting on the same
part) which is statically equivalent to the former, then such replacement
does not cause a noticeable change in the elastic equilibrium of parts of
the body which do not lie too near to the above-mentioned region. Both
formulations of St. Venant's Principle are obviously equivalent.
The Principle was first pronounced in St. Venant's memoir [1] of 1855. It agrees
very well with reality. However, its mathematical foundation (which must consist
of an estimate of the influence of a system of forces which are statically equivalent
to zero) is rather difficult, at least in the general case. By a system of forces, which
is statically equivalent to zero, will be understood a system, equivalent to zero
from the point of view of the statics of absolutely rigid bodies, i.e., a system, the
resultant vector and moment of which are equal to zero. Systems are called statically
equivalent, if they have the same resultant vectors and moments.

Thus St. Venant's Principle offers the means of modifying (under the
definite conditions stated above) the given stress distribution on the
boundary, and thus of simplifying problems. The Principle will be widely
used in the later parts of this book.
78 I. FUNDAMENTAL EQUATIONS §24

§ 24. Dynamic equations. The fundamental problems of the

dynamics of an elastic body. Although this book deals only with
problems of static equilibrium, nevertheless the dynamic equations of an
elastic body will be deduced, the simplest fundamental problems for these
equations stated and the uniqueness of their solutions proved. In
passing, an expression will be obtained for the potential energy of a
deformed body.
The deduction of the dynamic equations of an elastic body does not
offer any difficulties. These equations may be obtained directly from the
static equations by use of D'Alembert's Principle. In fact, it is sufficient
for this purpose to write down the static equations and to add the inertia
forces to the body forces.
In the present case the components of displacement, strain and stress
will be functions of %, y, z as well as of the time t. The components of
acceleration of a point, occupying a position (%,y, z) in the undeformed
state of the body, will be
户说（％, y, z, 力） 那贝先, y, z, 力） 82M(力, y, z, t)

The components of the inertia force, applied to a volume element dV

containing mass d加, will be
32M
dm.
IF
But since〃加 = p dVt where p is the density*, the components of the inertia
force per unit volume become
羿% o2v 济必
—
获’ 「获'
Adding the inertia force to the body force and introducing these values
into (18.1), one finds

ax# । 工
ax,
加十加 十&十 '
* As has already been stated in § 16, the region V is here that occupied by the
body before deformation. Accordingly dV denotes the initial volume element of
the body containing the mass dm and p is the density of this element before its
deformation ; this density may depend on the coordinates %, y, z of the point con¬
sidered, but not on the time t.
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 79

工 82v
a% +
十 y= (24.1)

dZx 况 济飨
+ z=
a% 效
These equations now take the place of the "equilibrium equations”
i.e., of the equations (18.1). The equations relating stresses to strains
and expressing the generalized Hooke's Law remain unaltered, since the
body forces do not figure in them. In the case of an isotropic body these
equations are (18.2) and (18.3). The equations (18.5) remain likewise
unchanged.
In the present case it is convenient to use equations in terms of dis¬
placements which can be obtained in the same way as it was done in
§ 21 and which in the case of an isotropic body have the form

(入 + F) -j~ [zA我 十" X

—p
•
朋

、00 82v
（入 +
时豆 +
y—p
+ IF (24.2)

济必
（入+向17 4~ fiA + 侬 Z=p
»2
These equations differ from those obtained by Navier in 1821 (cf. § 21)
in that Navier's equations contained only one elastic constant, i.e., one gets
his equations from (24.2) by putting 入 = 内
Analogous to the fundamental boundary value problems, which were
formulated in § 20 for the static case, one may similarly state problems
with regard to the dynamic equations. An essential difference is that the
boundary conditions have to be augmented by "initial conditions"
i.e., given displacements and velocities of points of the body at some
"initial'' instant of time t0. Mathematically these problems may be
formulated as follows:
First fundamental problem. Find functions 勿(％, y, z, t),
y, zt /), m(%, y, z, 力) satisfying (24.2) and the following supplementary
conditions :
X n = /i， Yn = /2， Zg = /3 (24.3)

on the surface S of the body at all times, starting from t — 曲，and

80 I. FUNDAMENTAL EQUATIONS §24

加 .加 . ..
说 = %0, » = Vq, W = ct = %, ot = Vq,
~~
ct
=% (24.4)

in the region V occupied by the body at time t = t0.

In these formulae 九，/2, /3 are functions given on the surface S of the
body and depending, in general, also on the time. Further, 劭，％ wQ>
&>,而 o, 血) are known functions of %,y, z. The equations (24.3) are the bounda¬
ry, and (24.4) the initial conditions.
The second fundamental problem differs from the
first only in that the boundary condition (24.3) is replaced by
%
— gl， 0 = g2， M = g3 (24.5)

on S ; gi，g2, g3 are given functions on S depending, in general, also on the

time.
For the mixed problem the condition (24.3) will refer to one
part and (24.5) to the remaining part of the boundary S.
Apart from these problems there are a number of other important
problems which will, however, not be stated here.
In the above cases it has been assumed that the body forces are known
at all points of the body and at all instants of time (beginning with
力 = %). No consideration will be given here to the difficult question of
the mathematical proof of the existence of solutions of these problems,
and it will only be proved that, 可 o solution o/ the given problems exists,
then it is unique.
Before giving this proof, a formula will be deduced which is of con¬
siderable independent interest and which expresses the law of con¬
servation of energy, as applied to the case under consideration.
Consider any definite motion of a given elastic body and choose as the
initial instant % the moment when the body lies in a "naturaT' state of
equilibrium, i.e., when body forces and stresses, and consequently also
deformations, are absent. Let KR) denote the work done by the external
stresses and body forces between the starting time tQ and the instant t
under consideration. This work will now be determined and for this
purpose calculate the work dR done by these forces in the time interval
力力十 威，assuming dt infinitely small.
A point, occupying before deformation the position (%,y, z), will at
time t have the coordinates
%
+ "(%, y, z"), y + 矶％, y, z"), z 十 切(％, y, z, 力).
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 81

The displacement of this point in the time interval + dt has obviously

the following components:
力 港， 方 dt, w a力
where
. a%
% = —— etc.
济

The work of the external stresses, acting on the surface element dS of

the body, in the time interval is
(Xgd + y京 + Znw)dS dt,
and the work of the body forces, applied to a body element dV, is
(X& 十 Y方 Zw^Vdt.
Thus the work dR, done by all the above forces during the time interval
dt, is given by

鲁= + 丫京 + z^)ds +川即 + y" + zw)dv.

3 P

Replacing under the first of these integrals Yn> Zn by their expressions

(18.5), transforming the integral in the same way as the integral / was
transformed in § 20, one finds using (24. 1 )
dR err ,
+ + +
、巾
力力 ww)dV

+ [jj
V
(Xa；%C 十 Yy^yy 十 Z z &zz + 2 y乞" + 2Zxezx + 2Xyexy)dV. (b)

But

川 (沛 + p vv + ww)dV = jjj 卷 Jp ('族

'
源 血 2)
彳丁

dt
,

where

7=
皆加p(族 + 源-
卜 鹤汇 (24.6)
82 I. FUNDAMENTAL EQUATIONS §24

Obviously T is the kinetic energy of the elastic body, i.e., the sum of the
kinetic energies of its different elements
当加(源 +源 + 凝) |p(^2 十 游 十 向2)
Next transform the second term on the right-hand side of (6). Assume
now that the body under consideration is isotropic and introduce the
function
印 = i^(exx + + ez^ + m 层 + + 凰 + 2落 十 2原 十 2e&)； (24.7)
it is immediately seen that
一涧 v _
沛
？ —即
de 期期
(24.0)
aw ,
-
dw 即
一 个 — > 2X期
3%中
>

and hence that the expression under the second integral of (6) is equal to
aw
—— and
济

膘7加犷
V V

Thus 0) takes the form

dR d rrr
—
dt =—-^-—
dt dt J J J
7TZ
WdV. (24.9)

Integrating both sides of this equation from % to t and taking into

consideration that at the initial instant the body is in a natural state of
rest (i.e., T = W = 0 at 力 = one finds for the work R done by the
external stresses and body forces in the time interval 传°, 力)
R^T +U, (24.10)
where

(24.11)

Formula (24.7) shows that W depends solely on the state of deformation

at a given moment at a given point ; hence U depends on the state of
CHAP. 3 FUNDAMENTAL LAW. BASIC EQUATIONS 83

deformation of the considered body at a given instant t. The quantity

U is the potential energy of deformation of the body, i.e., the work which
must be done by the body forces and external stresses, in order to cause
a given state of deformation. In fact, if under the influence of these
forces the body changed from a ' 'natural" state of rest to a new, deformed
state of rest, then, by (24. 1 0) , R = U, because for a state of rest T = 0.
Formula (24.10) indicates that the work of the body forces and external
stresses is transformed into kinetic energy and strain energy ; it thus
expresses the law of conservation of energy.
The quantity W, defined by (24.7), is the strain energy per unit volume.
In fact, it follows from (24.1 1) that the amount of potential energy, be¬
longing to the body element dV, is WdV. The expression W had already
been introduced in § 20; it will be remembered that W is a positive de¬
finite quadratic form in terms of the strain components. This follows
directly from (24.7).
Next consider the question of the uniqueness of the solutions of the
fundamental problems. Let any one of them have two solutions for
identical boundary and initial conditions and identical body forces.
Form the "difference” of these solutions (cf. § 20). The new solution
(%, v, 削) will satisfy the same equations as the two former solutions,
but in the absence o/ body forces ; in addition, in the case of the first
problem, one will have
Xn = Y n = Zn = 0 on S, (243)
and in the case of the second problem,
% =o=m=0 on S; (24.5')
in the case of the mixed problem, condition (243) will hold on one part
of the surface and (245) on the remainder. In all cases one has
Xnu + Ynv + Znw = 0 on S.
In fact, in the case (243): Xn = Yn = Zn = 0; in the case (245):
u — —v w ~ 0 (on S) at all instants of time, beginning with t = 力°,
and hence
du 3。
_ dw
况 — dt
•*—
- ~~
加 —— 0 on S ;
similarly for the mixed problem.
Further, one obviously has at the initial instant
84 I. FUNDAMENTAL EQUATIONS §24

每 = i) = 必 = 说 = u = 必 = 0,
because both known solutions satisfy the same initial conditions.
It follows from the above that the work R for the solution 猊，v, w is zero,
and hence by (24.10) that
T + U = 0.
But obviously this is only possible, when T = 0, U = 0, and therefore
at all instants of time, starting with t = %,one will have
。
Ai
律
-
一 U ~ 70]
W
1
f
D, “
匕c£‘
一."

〃
“ 匕
yy —
，，
“ A 0
Cgz r2仁
‘ ‘ 〃
已 C夕
• fI
U・

The first set of these equations shows that the displacements do not
depend on the time, i.e., that one is dealing with a static problem. It follows
from the second set of conditions that all the strains are zero, i.e., the
solution %, can only represent rigid body motion. Finally, it follows
from the condition, that at the initial instant all displacements are zero,
that there can be no body motion. Thus one has for all points of the body
and at all times 每 = o = m = 0. It is seen from this that the two solutions
of the problem, mentioned earlier, must be identical and this proves the
assertion.

…
NOTE. It follows from (20.7) that the potential energy

u=ffjw
of the deformed body in equilibrium may be expressed by the formula

U=
去， (X* + Y小 + Znw)dS + 卷仙X% + Yv + Zw)dVf (24.12)
s' V

and, in the absence o/ body forces, by

o= 斗
4
s
因沉 丫小 + Znw)dS, (24.13)

where the double integral is taken over the entire surface 。/ the body. Note
that, by (24.11) and (24.7), U > 0 for all states of non-zero-deformation.
The formulae (24.12) and (24.13) are easily remembered ; they show
that the strain energy of a body is equal to half the work done by the
external stresses and body forces of the final equilibrium state, acting
through the displacements of the equilibrium state.
 PART II

GENERAL FORMULAE OF THE PLANE THEORY

OF ELASTICITY
A
 The considerable mathematical difficulties which arise during any
attempt to solve the fundamental problems of the theory of elasticity
necessitate the search for practical methods of solution in special classes
of particular cases. One of the most important of such classes is concerned
with the so called "plane theory of elasticity” or "the plane problems
of the theory of elasticity" to which are devoted Parts II VI of
this book.
—
The development of the theory will here be based on the complex
representation of the general solution of the equations of the plane
theory of elasticity which will be stated below. This complex represen¬
tation, originally introduced by G. V. Kolosov (cf. his papers [1],
[2] and his book [6]), has been found very useful for the effective
solution of the fundamental boundary value problems as well as for
investigations of a general character, as is shown by a large number of
important papers which have been published lately in Russia. Several
of these will be studied or referred to in this or later Parts *.

From time to time papers have appeared outside Russia in which complex repre¬
sentation of partly incomplete solutions has been used and results have been given
which are either contained in the work of Russian authors or which follow directly
from the results obtained by the latter. Among these are, for example, the papers by
A. C. Stevenson [1] and H. Poritsky [2] about which some remarks will be made
in § 32.

It will only be mentioned now that some of the methods, fundamental

to complex representation, may be successfully generalized to the case
* It should be noted here that among the manuscripts and notes found after
the death of S. A. Chaplygin was discovered his work on the theory of elasticity
originating around 1900; it contains several results obtained subsequently by
G. V. Kolosov as well as by other authors. Cf. Chaplygin [1], p. 420 (article of
N. V. Zvolinskii and D. lu. Panov).
of anisotropic bodies, regarding which a brief statement will be made in
§ 104. Important and interesting results in this direction have been
obtained by S. G. Lekhnitzky [1], S. G. Mikhlin [11], D. I. Sherman
[9, 19], G. N. Savin [3—6] and others. A systematic study of a number
of these results may be found in Lekhnitzky's book which contains some
of the results obtained by its author.
Finally, it should be noted that the fundamental nature of the results
—
of the plane theory of elasticity (Parts II VI), stated below, must of
course be seen not in the new deduction of Kolosov's and other formulae,
but rather in the application of these formulae to the solution of the fun¬
damental boundary value problems by systematic utilization of the pro¬
perties of Cauchy type integrals and conformal transformation.
In fact, Kolosov's formulae may be deduced in many ways some of which are
extraordinarily simple. The method chosen here requires somewhat lengthier
calculations than some of the others, because it is completely elementary ; but it has
been retained here, since it gives, as by-products, a number of formulae useful
in the sequel; it also guarantees complete generality of the obtained solutions and
does not assume beforehand that these solutions are analytic. Note also that
before Kolosov several authors (e.g. L. N. G. Filon) have obtained some complex
representations of solutions, but no one (or almost no one) has actually applied
them.
 Chapter 4

BASIC EQUATIONS OF THE PLANE THEORY

OF ELASTICITY

The equations of the plane theory of elasticity apply to two cases of

equilibrium of elastic bodies which are of considerable interest in practice,
namely : to the case of plane strain and to the case of the deformation
。/ a thin plate under forces applied to its boundary and acting in its plane *.
These two cases will be discussed in detail in the following two sections.

§ 25. Plane strain. A body will be said to be in the state of plane

strain, parallel to the plane if the displacement component w is zero and
if the components u, 方 depend only on % and 夕，but not on z. In this case

e= —F cy
and the formulae (20.2) give
3% ,
匕 = 入6+ 2p.
Sv
— 加 du 、
_
X# = X0 + Xv
八

r 十

Zz — X0, Xz — Yg = 0.
These formulae show that the stress components are likewise independent
。
of z (since %, and hence 6 do not depend on it).
Further, the first two of the equations (20.1) take the form
X 双 双 aL c
ex —
F oy [- X
八

—
= 0, f3% F cy— y = 0,
八

and the third becomes Z = 0, indicating that for plane deformation,

* The results associated with the equations of the plane theory of elasticity
apply likewise to the problem of the equilibrium of elastic plates loaded by forces
normal to their plane regarding which some remarks will be made in Part V.
Quite recently I. N. Vekua [6] has shown that these same results may also be
used for the effective solution of boundary value problems of the theory of elastic
shells.
89
90 IL PLANE THEORY OF ELASTICITY §25

parallel to the plane the component of the body force in the direction
perpendicular to the plane of deformation must vanish. The preceding
equations also show that the components X, Y of the body force do not
depend on z.
Thus, in the end, the static equations of an elastic body in the case
of plane strain, parallel to the plane reduce to the following :
ax# ax” 3Xy SVy
X 丁十1+1 = 0,
寸

c
_L
I
_1_
I ** *• 0, (25.1)
个

dy ex
du 加
一 /加 a说、 (25.2)
入9
+ 2^^， ％ M诟十顾
where all the quantities appearing in these equations are independent
of z; the component Zz (likewise independent of z) is given by Z = 入8
or, noting that by (25.2)

x# 十 丫财 = 2(入 + w)e, 6 = 2( 入 --(x= + yj

十内
—
by
Z
『入9=
竟下区十%)=。区+ (25.3)

where g is Poisson's ratio. The formula (25.3), determining Zz, has been
intentionally deduced, since solution of the system (25.1) and (25.2)
represents the fundamental problem, and Zz is determined from (25.3)
after its solution. There remains now to state
those cases when plane deformation takes
place.
It will be assumed that one is dealing
with cylindrical (prismatic) bodies, bounded
by surfaces parallel to the axis Oz (sides) and
by two plane faces normal to the generating
surface (ends) (Fig. 1 1). Further, assume
that the external stresses, acting on the
sides, are parallel to the plane Oxy and do
not depend on z and that the same condition
is satisfied by the body forces. The latter
as well as the external stresses will be as¬
sumed known.
Consider whether under these conditions
plane deformation of the cylinder is possible.
CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 91

For this it is necessary and sufficient that the equations (25.1) and (25.2)
have solutions %,巩 Xx> Yv, Xy, satisfying on the sides of the cylinder
the boundary conditions
Xr cos (制, 盼 + Xy cos (w, y) = Xn,
Yx cos (%,%) + Yy cos (%,y) Yn) —
where Xn, Yn are the known components of the external stress vector,
acting on the side surface, and 处 is the outward normal ; the condition
(25.4) is obtained from (3.2) which gave the stress vector acting on the plane
with normal n. (The third of these formulae is identically satisfied, since,
by hypothesis, Zn = 0, = 0 and cos (%,z) = 0 on the ends.)
One is thus led to a problem, completely analogous to the first funda¬
mental boundary value problem of the theory of elasticity in the general
case (§ 20) ; but one is dealing here with a simpler case, because the un¬
known functions ut v, Xx, Yy, Xy depend only on the two variables %
and y and, instead of considering the entire region occupied by the body,
one may restrict the investigation to one of its sections in a plane, parallel
to In other words, one is dealing with the two-dimensional analogue
of the problem of § 20.
Under certain general conditions, referring to the shape of the cross¬
section of the cylinder, it may be shown (cf. Part V) that the two-di¬
mensional problem has always a solution which is unique, provided the
resultant of the body forces and the stresses acting on the sides is static¬
ally equivalent to zero.
Let u, % X#, Yy, Xy be the solution of the two-dimensional problem.
Calculating Zz from (25.3) and assuming 切 = = Z = 0, one obtains
财

the solution satisfying all the conditions above. It is seen that the ends
of the cylinder are not free from stresses, but that they are subject to
normal stresses. In fact, the normal stress Zz acts on the upper and
—
( Zj acts on the lower end, where, for simplicity, the end facing in the
positive z direction has been called "upper". Application of these stresses
is seen to be necessary for the maintenance of plane deformation. As
has been stated, the given body forces and stresses, acting on the sides,
determine the functions * 勿, Xx, Yy, X” and hence also Zz. Thus the
choice of the longitudinal stress is not arbitrary.
At first sight, this fact seems to reduce the value of the study of plane
* The functions u, o are determined exactly apart from terms expressing an
arbitrary rigid body displacement parallel to the plane Oxy ; however, these terms
are not essential.
92 II. PLANE THEORY OF ELASTICITY §26

strain. But in practice this inconvenience is very easily removed in the

case of a long cylinder (the height of which is large compared with the
transverse dimensions) . In fact, in order to remove the above stresses on
the ends, it is sufficient to superimpose on the obtained solution the
solution of the problem of the equilibrium of the cylinder under the con¬
dition that there are no body forces, that the sides are free from external
stresses and the ends are subject to tractions equal in magnitude and
opposite in sign to those which are to be removed.
Consider these latter tractions, exerted on one of the ends; since
they are parallel to the axis Oz, their resultant is statically equivalent
to a force parallel to the same axis, acting, say, at the centroid of the end,
and a couple the plane of which is likewise parallel to Oz. The resultant
of the stresses, acting on the other end, is statically equivalent to a force
and couple, statically balancing the former. But the question of the
elastic equilibrium of a (long) cylinder under the influence of tractions,
applied to the ends and statically equivalent to a tensile force and
a bending couple, belongs to a number of very simple problems of the
theory of elasticity and can be solved by elementary methods (cf. Part
VII). Therefore one can always remove the tractions on the ends by very
simple means.
Thus, from the solution of the problem of plane strain of a cylinder
under tractions of the stated type, applied to the side surfaces, one obtains
the solution of the problem of equilibrium of a cylinder under the influence
of the same forces, but subject to the conditions that the ends are free
from stresses ; in this latter case, generally speaking, deformation will
no longer be plane.

§ 26. Deformation of a thin plate under forces acting in its

plane. The equations of the plane theory of elasticity apply also to
another case, yet more important in practice, namely to the case of thin
plates for definite types of loading.
By a plate will be understood a cylinder of very small height or thickness
Z 2%. The middle surface of the plate (i.e.,
the plane parallel to the ends and half
way between them) is taken as the plane
(Fig. 12). It will be assumed that the
faces are free from external stresses
and it will be postulated that the external
Xn Fig. 12 stresses acting on the edges are parallel
CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 93

to the faces and symmetrically distributed with respect to the middle

surface. The same will be assumed to hold true for the body forces. From
the practical point of view, it is sufficient, as far as the stresses acting on
the edges are concerned, to assume that the resultant of the stresses acting
on any element of the edge, included between the two faces, is statically
equivalent to a force, applied at the centre of the element and lying
in the middle surface ; in fact, by St. Venanfs Principle (§ 23), every such
•esultant may be replaced by a statically equivalent resultant satis¬
fying the earlier conditions.
For reasons of symmetry, it is obvious that the points of the middle
surface will remain in it * after deformation, that the displacement com¬
ponent w will be very small and that the variations of the components
u and 力 over the thickness of the plate will be insignificant. Therefore
it is clear that it is possible to obtain a completely satisfactory represen¬
tation of the elastic equilibrium of the plate by considering the mean
values of the quantities 劭 and 力 over the thickness of the plate ; these
mean values which will be denoted by %* and o* are defined by

%*(%, y)
J
"3, y) = 丸 M%, y, z)dz.
—h

By assumption, the functions Xz{x, y, z}, Yz(x, y, z} and Zz(x, y, z)

vanish on the ends, i.e., for z = 土九 (since the ends are free from external
stresses) . Therefore it follows from

% 区
a%
=0
that
0

for z = ± h. In fact, it follows from Zx(x, y, 土％) = that

az式%,y，土 例 一
°
dx '

and similarly that y, 土 力)

= 0.

* We disregard here the rigid displacement which may be common to the

entire body.
94 II. PLANE THEORY OF ELASTICITY §26

Thus the quantity y, z) is not only zero for z = ± 瓦 but also its
derivative with respect to z vanishes for these values. Therefore it is
obvious that Zz will be a very small quantity throughout the thickness of
the plate and one may assume, as a good approximation, that Z% = 0
everywhere.
Consider now the equations
a冗 ar.+y
ay, 十 ar,
y+l+ _+x“ = o,
ox
aXy
oy
ax?
u 3% oy cz
“ 0，
八

and take the mean values of both these equations, i.e., integrate them with
—
respect to z from 力 to + % and divide by 2勿 One has

_
fax,__ 1 1 f dY.
—
1I ,
—2 J —
1
2力 J —
——
力
/ Ldz
3z
T
=：
2 力
心
rxj+j = 0,， / az 2小 」f
z 抵
[Yz] “ =0,，
—A 一九

and hence the preceding equations become

阳 ax* ay： ay：
-ex
x —— —
]
效
… = o,
f x* ~
- — - y——~
—ex ：
~

cy
y* = o. (26.1)

Further, it follows from

—
/加 dv dw
入
\ d%—F
dy
F cz =
+2[i^— Zz =0
that /du
Sw 入
dz 入十 2以 1 加
十

Substituting this value of ——一 m

Hz
/加 加 du 彻 加 加
入底7 + 豆 + 十 2B 益，y产入 （蒜 额 + 22 而，
one obtains

入
2W ，初加
I
2 \ 34
以 cy/
j 、 2[1-—
加
ex
Yy ~ -
入
2入[1
T—
2q
，加
1
\ ex
F 加、
cy / + cy
「
加
2(1—

Taking the mean value of these two equations and of the equation
/ du 加
Xy
CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 95

one finally finds

.
X* = 入初*
+ 2{1丁，
d%
yj
"
.
- X*0* +
加*
cy
(26.2)

where

(26.3)

Comparison of (26.1) and (26.2) with (25. 1) and (25.2) shows that
the mean values of the displacement components v and the stress
components Xx) Yy) Xy satisfy the same equations which govern the case
of plane strain, the only difference being that one has to replace 入 by 入*
defined by (26.3).
Following A. E. H. Love [1] (§§ 94 and 146), the stressed state of a
plate, for which Z? = 0 everywhere and Xz, Yz vanish on its faces, will be
called "generalized plane stress". Such a state of stress was first con¬
sidered by L. N. G. Filon [1] (cf. also: Filon [2], E. G. Coker and Filon [1])
who established the above equations for the mean values. These equations
are, of course, applicable to plates of finite thickness. It has been seen
that for thin plates and under the conditions, stated above, the state of
stress may, with good approximation, be assumed to be one of generalized
plane stress. For further justification of the assumption that in the case of
a thin plate : Zz = 0, reference may be made to J. H. Michell [1] who fur¬
nished additional evidence with regard to this point.
Let ds be any line element in the plane Consider a rectangular
area of height 2%, perpendicular to Oxy, the trace of which in that plane
is ds (Fig. 12). The components of the mean stress, acting on this area,
in the directions Ox, Oy are
Y^ds,
where
X：= X：cos (%,劝 + X；cos (% y),
y：= y：cos 伊,幻 + yj cos (肛 力 (26.4)

and % is the positive normal. The projections of the forces acting on this
element are equal to 2hX* ds, 2hY* ds. We may call the quantities
96 IL PLANE THEORY OF ELASTICITY §27

X* Y* ds the projections of the mean values of the forces per unit

height (thickness) of the plate acting from the side of the normal n
to the element ds.
§ 27. Basic equations of the plane theory of elasticity. It has
been seen that in the two cases, considered in §25 and § 26, one is led to
the study of the following system of equations :
a%
o, a% + y =o, (27.1)
a” 砂 ' a% dy

Xh = 入e +2[z— — , 匕=冲
哈，Xy『(|^ + ||)'(27・2)
2

where
a% 加
6=标 十秀.
In the case of generalized plane stress (§ 26) the components of dis¬
placement and stress have to be replaced by their mean values over the
thickness of the plate and 入 by 入 .
2[2
入
Since all quantities depend only on x and y, consideration may be
limited to points of the plane which will be assumed to be the plane
of one of the normal sections of the cylinder under consideration, and
in the case of § 26 the middle plane. Therefore, when talking, for example,
of a region occupied by a body, one will have in mind a two-dimensional
region, i.e., the intersection of the considered body with the plane
further, instead of talking about tractions acting on areas perpendicular
to the plane Oxy, one will speak of tractions acting on line elements ds
of the cross-sections. Thus, in the case of § 25, one will say that a force
with components Xn 03, Y n ds is applied to the linear element ds,
where n is the normal to ds; but the fact is that they are the com¬
ponents in the Ox, Oy directions of the force applied to a rectangular
area, perpendicular to the plane Oxy, with base Gs and unit height
(its component in the z direction vanishes). In the case of § 26, X”
Yn ds will be understood to be the quantities which were denoted at
the end of § 26 by X* Y* ds.
As in Part I it will be assumed that the components of displacement are
single-valued continuous functions with continuous derivatives up to and
including the third order throughout the region occupied by the body.
Then, by (27.2), the stress components will be single-valued functions
CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 97

with continuous second order derivatives.

Just as in § 21, the system (27.1) and (27.2) may be replaced by one
involving only displacements. For this purpose one has only to put w = 0
in the equations of § 21, or simply to substitute from (27.2) in (27.1). In
either way one finds
89 油
(入 + w) + =X 0, (入十 q) + “Ad + 丫 =0, (27.4)
where
洲 济

Having found some solution of this system, the corresponding stresses

are obtained from (27.2) by differentiation.
It is likewise not difficult to form the equations which involve- stresses.
It is now seen that these equations comprise the equations (27.1) and
supplementary equation which replaces in the present case the six
conditions of compatibility of Beltrami-Michell. This additional equation
expresses the condition which must be fulfilled so that one may find,
corresponding to functions Xx, Yy, Xy satisfying (27.1), functions
u, o related to Xx> Yy, Xy by (27.2). This condition may, of course, be
obtained as a particular case of the general compatibility conditions,
but it will be deduced here independently in two ways.
The first method is based on St. Venant's conditions of compatibility,
as was the deduction of the conditions of Beltrami-Michell in the general
case. Thus in the case of plain strain, when exx, eyy and exy are independent
of z and eyz = ezx = ezz — 0, the conditions (15.6) obviously reduce to

Substituting here the expressions

2( 入 + J)
QQ + %)}
% —1
i
M 2( 入
入
(°B + 匕)}, (27.5)

式
—
e p= G X i

力
98 IL PLANE THEORY OF ELASTICITY § 27

deduced from (27.2), one easily obtains

济X, 济w 入 济瑞

效2
1-
加2 2(入 十 (X)
a3
+ VJ， v、
~~
c
2 —— = 0.
—3%—砂 八 小”八
(27.6)

This is the required condition. It may be considerably simplified by

taking into consideration that Xx> Yy, Xy satisfy (27.1). In fact, dif¬
ferentiating the first equation of (27.1 ) with respect to % and the second
with respect to y and adding, one finds
x 学 yy ax ay
—'—砂z—H —F
济 涉
—2
工

W = ~•
ex一歹
cy 伙

炉 Xy
Substituting this expression for — 2 in (27.6) one obtains, after
some obvious simplifications,

+匕)=一 美%以(穿十 簧). 团7)

The second method of deduction of (27.7) is based directly on

the equations (27.1) and (27.2) and it presents at the same time a
method for calculating displacements from given stress components
(or, what is the same thing, strain components) . It is more elementary and
more convenient in practice than the one given in § 15 for the general
case.
Since it is desired to find conditions which must be satisfied by the
stress components XXi Yyt Xy, so that there exist functions %,”, related
to the former by (27.2), it will now be attempted to actually calculate
u, v from (27.2), assuming that X%, Yy, Xv represent a given solution of
(27.1).
The first two equations of (27.2) may be written
彻 入
2u,
产加 —
— = Xx — 2 X
+ u
—(Xx
k + y期)， y)

c (275)
dv 入

如 2( 入 十
Let (a, 6) be an arbitrary point of the body. For the present, consideration
will be limited to points lying inside some rectangle with centre (a, 6)
which lies completely inside the body. Putting P = X# + Yy> one finds
from (275)
CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 99

(27.8)
XP ]
2[1 M%, y)
取不前八y+限%),
where /式夕)，％(%) are functions, at present unknown. The expressions
(27.8) satisfy (275), i.e., the first two relations of (27.2).
In order to satisfy the third equation of (27.2), substitute in it from
(27.8). Differentiating under the integral sign, one obtains
8 V
入 即
h入 dP]
a
十八苇T-而 b
平/x!为 -24 =
= -/；(，) 一/；(%)• (27.9)
This equation may only be satisfied, if the left-hand side can be conceived
as the sum of two functions one of which depends only on % and the other
only on y. For this to be so, it is necessary and sufficient that the second
a2 . .
of the left-hand side is identically zero *. Differ-
derivative
34 效
entiating the left hand side one time with respect to %, another time
with respect to y and equating the result to zero, one finds exactly
(27.6), whence follows, in turn, (27.7).
If (27.7) is fulfilled, then the left-hand side of (27.9) has the form
B(y) +
and (27.9) leads to the condition
F2(%) 十 为 3) = 一下式y) 一汇®)
which is only possible if both sides are equal to one and the same constant
to be denoted by 2M. Then, by the last equation,

d2F
* If F(x, y) = 玛(力) 十 FJy), then = 0 and conversely. For the ful-
加 织
fillment of the last condition one is readily seen to have
尸(力, 训 = F区 b) + F(a, y) — F& b).
100 II. PLANE THEORY OF ELASTICITY §27

九（夕） ~
-J Fl®)分
b
一 2眸y + 2皎,
(27.10)

/2(%) = —J + 2灯; + 2明
a
where a, § are arbitrary constants. Substituting from (27.10) into (27.8)
one finds expressions for〃and v which are definite apart from terms of
the form
% - -
—sy -J- (X, = e% (27.11)
where a, 8, e are arbitrary constants. These terms express only rigid body
displacement (in the plane 0%y) and they do not influence stresses and
strains. The constants a, e attain definite values, if one assumes as
given the values of the components of displacement u, » and of rotation
3%
= 2 (27.12)

at some point of the region under consideration, e.g. at (口，乃.

So far consideration has been limited to points (招 y) lying inside a
rectangle with centre (曲 6) which is entirely inside the region occupied
by the body. In order to find values of v at other points of the region,
one has to select some point (屋，6') inside that rectangle and near its
boundary and to construct a second rectangle with (屋，br) as centre.
This rectangle must again be chosen in such a way that it does not leave
the region occupied by the body, although it will extend beyond the bound¬
aries 。/ the first rectangle. In this way one may find the values of v
at all points of the second rectangle by the method presented above. In
order that the values of u, 叫 obtained in this manner, agree in those parts
common to both rectangles, one has to select the arbitrary constants,
entering into the formulae for the second rectangle, so that the values of
。
u, and 夕 at (屋，夕) coincide with those calculated for this point from
the formulae for the first rectangle. Hence it is seen that the formulae
for the second rectangle will not involve any new arbitrary constants.
By repeating this procedure sufficiently often one may calculate the
displacements for any point of the body. (This method may be compared
with the well known process of analytic continuation of functions of a
complex variable.)
 CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 101

However, there arises the following question. Let (叼，yj be some

point of the body different from the initial point (a, 6). In order to calculate
the values of u, d at (先小 ％), one has, by the above method, to construct a
set of rectangles, partly covering each other, the first of which is the
rectangle with centre («, b) and the last a rectangle containing (巧，％).
But there is an infinite number of such sets. The question is then whether
the particular choice of one of these sets will influence the values of u, o
at (衍，夕。；in other words, whether u, 力 will be single-v allied functions
of
This question is easily resolved by methods differing from those of
the present section, using formulae expressing the displacement compo¬
nents u, d in terms of the stress components Xx, Yy, X® by means of
curvilinear integrals taken along arbitrary curves linking the points
{a, b) and (先卜 ％). These formulae follow from (15.4) by putting there
w = eyz egx— — = 0 and by replacing the components of strain
eyV> exy by their expressions (27.5) in terms of the stress components
X#, Yv, Xy. Proceeding in quite an analogous manner as in § 15, it is easily
verified that u, o are necessarily single-valued functions, provided the
region occupied by the body is simply connected.
In the case of multiply connected regions the components u, v may
be found to be multi-valued functions, in spite of the fact that (27.7) is
satisfied. Therefore, in the case of multiply connected regions, (27.7) must
be supplemented by a condition of single-valuedness o/ displacements,
where, of course, it has been assumed that the stress components are
always single-valued functions. Later on this question will be considered
in greater detail.
The necessity of the condition (27.7) may also be inferred in the following manner :
Differentiating the first of the equations (27.4) with respect to % and the second with

、
respect to 夕 and adding, one obtains
/ ax ay
(入 十 2W)A8 十 — —)
1 - = 0.

Further, noting that by the first two relations of (27.2)

- 2( 入 凶
and substituting this value of 0 in the preceding equation, one obtains again (27.7).

§ 28. Reduction to the case of absence of body forces. The solution

of the equations of the plane theory of elasticity is considerably simplified
in the case of absence of body forces, i.e., when X = Y = 0. On the other
102 口. PLANE THEORY OF ELASTICITY §28

hand, the general case may always be reduced to the last: for this purpose
it is sufficient to find any particular solution of the system of equations
(27.1) and (27.2). Let X化 丫化 办①，淤) be such a particular solution.
Putting
X? = X黑 + etc., = 切⑴ + etc., (28.1)
it is seen that the functions X*), . . . , 力⑴ satisfy the same equations as
Xx> …，％ but for X = y =0.
The determination of particular solutions X霆，..
” will be limited
here to two cases which cover most practical applications : the case
of gravity and the case of centrifugal forces for rotation about an axis parallel
to Oz. However, the determination of a particular solution for arbitrarily
given body forces does not present any particular difficulties.
In order to find the particular solutions, one may, from a point of view
of convenience, either use the equations (27.1) and (27.7) which involve
stresses or the equations (27.4) in terms of displacements. The first set
of equations will here be used for the problem of gravity forces and the
second for the case of inertia forces.
Consider first the case- of gravity forces. Assuming that the axis Oy
is directed vertically upwards, one has X = 0, Y =
—
gp, where g
is the gravitational acceleration and p is the density which will be as¬
sumed constant.
Therefore (27.1) and (27.7) take the form
ax, aXy a匕 aw
喊十哥 =°' 管+引=p砥 A^+y^°-

Clearly these equations will be satisfied by putting, for example,

Xa；= Xy = 0, Y"=pgy. (28.2)
The displacements corresponding to this particular solution may be
calculated in the manner stated earlier. In fact, by (27.8),

= /一 2；+
J 2( 入 jjl)
— yig
2( +
入
孙 + /式力
// +入 2M
入十
2M = J w/ pg夕 如
心不， 吧 府（以
CHAP. 4 EQUATIONS OF THE PLANE THEORY OF ELASTICITY 103

Substituting these values in

one obtains the equation

which may be satisfied by putting, for example,

%（y）= 0, %㈤ = 石经
•

p
4（ + F）
入
娟

Thus, one has for the displacements

+ 2（z y
入 入 X
%=一双不请物 仁皈4请 + 频不针 33）
g%.

Next use （27.4） to solve the problem of inertia forces. If the body is
rotating uniformly about an axis, perpendicular to the plane Oxy and
passing through 0, the inertia （centrifugal） forces are given by
X = po）2%, V = ,32%
where 3 is the angular velocity. Hence （27.4） takes the form

（入 + © + pA% +「3?% = 0, （入 十 科） 十 p32y =

It is easily seen that these equations will be satisfied by expressions of the
form
% = a%3 + 力%y2, e — 冲
3
+ 6%2y.
In fact, substituting these values in the preceding equations, it is seen
that both will be satisfied, if
2（3« + ，）（入 + 2[i） + p3? = 0,
or
pco2
3d 十 b （28.4）
2（入 十 2前
Thus one of the constants a, b may be chosen arbitrarily. For example,
put
,32
d = b
8（入 + 2r）
104 II. PLANE THEORY OF ELASTICITY §28

in which case the displacement is purely radial, since then

pco2
——（先2 t
8（入十 2E ' +乙，
8（入 2# ' +
= （先 y » （28.5）
y2）y.
方 — 2
' 7
The corresponding stresses are given by

Xx = — 4（入 十十上
~一
2j1）-
、 yp3?（%2 t
' +词 OG）2^2,
'' 2（ 入 十 2囚 '
Yy = 一
4（入 2®
2（%2+ 娟 — o 入
段
2（ + 2® 上
pw2y2, （28.6）

flpGJ2
Ay = 孙
2（入 十 即T
 Chapter 5

STRESS FUNCTION. COMPLEX REPRESENTATION OF THE

GENERAL SOLUTION OF THE EQUATIONS OF THE PLANE
THEORY OF ELASTICITY

§ 29. Stress function. In the sequel (unless stated otherwise)

attention will be concentrated on the equations of the plane theory of
elasticity when no body forces are present. In that case the stresses may
be expressed by means of one single auxiliary function which is called
a stress function or Airy function and which plays an important part in
the plane theory of elasticity.
In fact, under the conditions considered, one has

三十三
3% 2y
=0, 四产
3% 3夕
0. (29.1)

The first of these equations represents the necessary and sufficient con¬
dition for the existence of some function 5(%,y) such that
as =X
—
/
= X入第， 18.
3夕
上 上

The second of the equations (29.1) is the necessary and sufficient condition
for the existence of some function A(%, y) such that
34 SA _
F
0”
—
= Y ~
如
= 2(. 旷

Comparison of the two expressions for Xy shows that one must have
az dB
a%
whence follows the existence of some function U(%, y) such that

Substituting these values for A and B in the preceding equations, it is

105
106 IL PLANE THEORY OF ELASTICITY §29

seen that (in the absence of body forces) there always exists some function
U(%,y) by the help of which the stresses may be expressed in the following
manner:
洲 u 洲 u 济 u
=菖’吃二一弱’匕 =声. （29.2）

This fact was first noticed by G. B. Airy (1862). The function U is called
a stress function or Airy function.
Since, by a hypothesis in § 27, the functions Xx, Yy, Xy are single¬
valued and continuous together with their second order derivatives,
the function U must have continuous derivatives up to and including the
fourth order and these derivatives, from the second order onwards, must
be single-valued functions throughout the region, occupied by the body.
Conversely, it is obvious that, if U has these properties, the functions
Xx, Yy, Xyt defined by (29.2), will satisfy (29.1). However, it is known
that this does not yet mean that these functions correspond to some
actual deformations. For this purpose also the condition (27.7) must be
satisfied which in the absence of body forces becomes
+ Yy) = 0； (29.3)
or, noting that
X* 十 匕 = AU,
one obtains the equation

AAU = 0 or
铲。
r 2
那0
L
__
济U
(29.4)
%2 %2 效&
Equation (29.4) is called biharmonic and its solutions biharmonic functions.
J. C. Maxwell was the first person to notice that the stress function must
satisfy (29.4).
However, in the sequel, biharmonic functions will be understood
to be only functions, which satisfy the biharmonic equation, the der¬
ivatives of which are continuous to and including the fourth order and
the derivatives of which, starting from the second order, are single-valued
throughout the region under consideration.
If the considered region is simply connected, single-valuedness of the
second derivatives implies that of the function itself. In multiply con¬
nected regions, however, this is not necessarily so, as will be shown later.
Thus it has been proved that the stress functions must be biharmonic.
It is known that this condition, which is nothing else but the condition
CHAP. 5 COMPLEX REPRESENTATION 107

(27.7), is also a sufficient condition that the corresponding stresses may be

produced by some actual deformation, if for the time being no impor¬
tance is attached to the fact that the corresponding displacements may
(in the case of multiply connected regions) turn out to be multi-valued.

NOTE. In § 27 some restrictions have been imposed on the con¬

sidered displacements and stresses. In fact, it has been agreed to assume
that the functions u, 力 are single-valued and have continuous derivatives
up to and including the third order ; the continuity and single-valuedness
of the stress components and their derivatives up to the second order
was a direct consequence of the relations

X8 = 0入 十 2(i—du
― , Yy 入6+ - 2p.
加
— , Xy = [1
/加
1
3说、
—- ) (29:5)

From the point of view of certain deductions of a general character

it is convenient to relax these conditions slightly. Thus, everything to
be said below will remain true, if from now onwards the following con¬
ditions are assumed to apply in the region S occupied by the body:
a) Conditions referring to stresses. The components Xx, Yy> Xy are
single-valued continuous functions having continuous derivatives up
to the second order and satisfying equations (29.1) and (29.3). A con¬
sequence of these conditions is that the function U is biharmonic (in the
sense stated above).
6) Conditions referring to displacements. The components % o are
single-valued, continuous functions having first order derivatives con¬
nected with the stress components by (29.5).
It will be seen below that the conditions a) ensure the existence of
derivatives of any order of the functions Xx> Yy, furthermore, it
will be seen that these functions are analytic (cf. § 32).
Similarly, the conditions 6) together with a) ensure the existence of
derivatives of any order of the functions 说，o (cf. § 32) (and even their
being analytic). Note that in many cases it is sufficient to adopt the
preceding conditions, omitting the condition of single-valuedness of the
functions u, v. For example, in the case of simply connected regions this
single-valuedness is a necessary consequence of the remaining con¬
ditions a) and b) ; this follows from the results of the next section.

§ 30. Determination of displacements from the stress function.

If a (biharmonic) stress function U be given, the corresponding stresses
108 II. PLANE THEORY OF ELASTICITY §30

follow from the formulae of § 29, viz.:

x,= 要，
砂 要，％
z 鲁;
cxcy
(30.1)

the displacements, corresponding to these stresses, may be found by the

methods of § 27. However, different formulae will be given here which
are more convenient than the former and which were first stated by
A. E. H. Love [1] who obtained them in a somewhat different manner.
Let the region S, occupied by the body, for the time being (up till § 35)
瓦 assumed to be simply connected (cf . § 1 5 and Appendix 2 for a definition
of connectivity) . The present problem is to find functions %,e from the

协 济U
入8
c
+
丁* — — — —ay2
2ii -
加
~~ 入8
八

+ 产砂
加
—— — 洲。
"a 'J
2炉
(30.2)

The first two of these equations, solved for

du 加
>
3% 砂
审ve
丝 =空 — —L— AO c
加 洲。 入
汲 2
砂 2（入 十 M） '效 3*2 2（入 十“）
Introducing the notation
KU = P, (30.3)
u
— 黑 and in
洲
replacing in the first of the above equations by
au—- by 炉。 说
the second —a%2 ) P —, one obtains
ay2
a% 洲 u 入
+ 2[z 加 入 2[x
a% 2( 入 + a)
P, 2“ —效 ~ P. (30.4)
砂 2
2（ 入 十 M）
From (30.3) the function P is seen to be harmonic, because
AP = AAU = 0.
Let Q be the harmonic function, conjugate to P, i.e., the function
CHAP. 5 COMPLEX REPRESENTATION 109

_
satisfying the Cauchy-Riemann conditions
dP 附 a? _ 30 .
3% '砂 dx
this function is determined for a given F apart from an arbitrary constant
term (cf. Appendix 3). Then the expression
/(z) = 尸 y) + 勿0(%,y) (30.5)
will represent a function of the complex variable z — x iy, holomorphic
in the region S occupied by the body.
Furthermore, put

— p + iq = i //k) "z.
_
4(z) (30.6)

Obviously
中 ，⑶ 孳 十 ，字 +仍 iQ),
8% ex
whence, noting that by the Cauchy-Riemann conditions

3' dq 3力 dq
3% dy dy 加’

one obtains

普=黑=杷
3% oy 票=
cy —票
d%
=—

Thus
效 dq
P= 4 ~~
加
= 4 —~
dy
and hence (30.4) may be written
加 济02( 入
+ 2(i) 效 加 济02( 入 十 2t1) 却
a% a%2 入 m % 如 的2入 q 效
Integrating, one obtains
au 2( 入 + 2jz)
2M =— a%
। .入十N
, 0 + /曲，
一、

au_ 2( 入 2凶1 ,、
2[iv =— 效
„

— p .入十出
, 9 十为3)
110 II. PLANE THEORY OF ELASTICITY §31

Substituting these expressions in the third of the equations (30.2) and

noting that
效 dq
—砂 a% = o,
one finds
+ 为(幻 = o.
and hence (cf. § 27) that the functions /式夕) and /2(%) have the form
/i = 2M(— ey + a), /2 = + 0),
where a, 0, s are arbitrary constants (the factor 2“ having been in¬
troduced for convenience). Omitting these terms, which only give rigid
body displacement, one obtains formulae coinciding essentially with
those of A. E. H. Love [1] :
au 2( 入 + 出)
2 3U 2( 入 + 2#
— 2^ = — 万 十「巴幺 (30・8)
c
2y =

Since the function <p(z), defined by (30.6), is obviously holomorphic

(cf. Appendix 3) in S (which, as will be remembered, was assumed simply
connected), the functions u and v will be found to be single-valued
throughout S.
Thus it is seen that every biharmonic function, subject to the con¬
ditions of § 29, determines some deformation satisfying all the required
conditions.

In conclusion, it will be noted that in omitting on the right-hand

sides of (30.8) terms expressing rigid body displacement no generality
is lost, because, as is readily seen, the functions on the right-hand sides
of (30.8) are determined by the given stress components exactly only
apart from certain terms which correspond to a rigid motion of the
body as a unit (cf. § 34) .

§ 31. Complex representation of biharmonic functions. It will

now be shown that every biharmonic function U{x, y) of the two variables
x, y may be represented in a very simple manner by the help of two
CHAP. 5 COMPLEX REPRESENTATION 111

functions of the complex variable z = 十彻. This fact is of greatest

importance for the theory of the biharmonic equation and, in particular,
for the plane theory of elasticity, since the properties of functions of a
complex variable are generally well known.
The function
<P(z) =p + iq
has already been introduced by (30.6). It is easily verified directly,
—
using (30.7), that the function U ~~ px qy is harmonic, i.e., that
A(U — px ~ gy) = 0.
Hence
U = 力，十 十 名，
"
where % is some function harmonic in the region S under consideration.
Now let %(z) denote the function of the complex variable z, the real
part of which is 力. (In order to find %(z), one has to calculate the harmonic
function %, conjugate to %.) If the region S is simply connected, the
function %(z) will be holomorphic there.

。
Obviously one may then write
= 沆{祁(z) + %(z)}, (31.1)
where 猊 denotes "the real part" and
z=x — iy;
in general, if A is some complex number a 十 ib, then A will denote
—
its conjugate complex value a ib, so that, for example,
= p — iq.

。
With this notation, (31.1) may be written
2 = 狗(z) +zg(z) 十 ％(z) 十词. (31.2)
This is the required expression. It was first given by E. Goursat [2] in a
somewhat different form, his method of deduction being likewise dif¬
ferent. However, in the sequel, no use will be made of this expression
for U, but of expressions for its partial derivatives, since these derivatives
have direct physical meaning.
The method of deduction used by Goursat is as follows. Let there be given the
equation
= 0.
112 II. PLANE THEORY OF ELASTICITY §31

Introduce instead of 力 and y the new variables z = % -Y iy and z — 力 一 £y：

then the preceding equation takes the form
济

822 苏2
= 0,

whence it follows directly that

U = 包⑵ + 乎20) + 次l(z) + 以
where <p1, ^2, 乂】， are ''arbitrary'' functions. This formal approach may be well
justified, if one assumes beforehand that U is analytic. If U is a real function, it is
easily seen that one must put

中2(宕) — 中1(Z)， X2(2) X1(Z)；

hence one obtains (31.2).
The proof, produced in the main text, was first given by the Author [4].

It is easily found that

au
2k
OX
勺(Z) — 中 — + x'(z) + —
+ — + zg'(z)
—
片(z),
¬

(31.3)
3U
2 3― = 无一9⑶
oy
—
狗'⑶ (z) -片(z)].

It is immediately seen from (31.3) that, instead of considering the

expressions for
dU ^U
效，
1
dx
it will be more convenient to deal with the expression for
au 1-
.
2
au
a% 效
which is by far simpler. In fact, one has
ao du
—— ——
/(%, y) = 后十 2 乎(Z) 十 少⑵， (31.4)
可=
where
“z)= 孚. (31.5)
CHAP. 5 COMPLEX REPRESENTATION 113

Returning to (31.2) it is noticed that, conversely, every expression of

〃
the form (31.2) represents a biharmonic function, if <p(z), z)are holo¬
morphic functions of z. In fact, differentiating the first of the equations
(31.3) with respect to % and the second with respect to y and adding, one
finds
= 2Q(z) + 而j = 4猊[心叫， (31.6)
and hence it follows that AU is a harmonic function. Consequently,
AAU = 0.
The formula (31.6) shows, in addition, that AU is completely determined
by the real part of the function 乎 '(z).

§ 32. Complex representation of displacements and stresses.

Multiplying the second formula of (30.8) by i and adding it to the first,
one obtains

2"3 + 谢) _（至十 ，迫）+空上也 中 (z),

dx \ 入
*
whence one finds, by (31.4), the very important and convenient formula
2Mw + 谢) = g⑵
乂 ~•的 '(z)
'
— 中团 (32.1)

which essentially agrees with a formula, first stated by G. V. Kolosov [1]

who obtained it in a different way; in (32. 1)

x — —入 ——N— — 3 — 4b. (32.2)

+
In the case of thin plates ("generalized plane stress", § 26) one has to
replace 乂 by x*, obtained from (32.2) by substituting 入* for 入. Thus, in
this case,

x* =
入*
+ 3 —— cr
= 1 b.
,
(322)
入* n +
Obviously x > 1, x* > 1.
Next consider the representation of the stress components by means
of the same functions 平 and 少. For this purpose an expression will be
found for the forces acting on an element of any shape lying in the plane
Oxy.
Consider some arc AB in the plane Oxy. Let its positive direction be
from A to B and draw the normal n to the right of the arc when looking
114 II. PLANE THEORY OF ELASTICITY §32

along it in the positive direction. In other

words, postulate that the positive direction
of the normal and the tangent are orientated
with respect to each other as the axes Ox and
Oy (Fig. 13).
As always, the force (X/$, 丫勿曲)，acting
on an element ds of the arc AB, will be
understood to be the force exerted on the
side of the positive normal. One has
济 U 济0
cos % %)、
,
Xn — Xx cos (%,幻 Xy cos (肛 y) =
3y 2 — ox cy cos (%,y),
洲0 ，、谭 U ,
Yn = Yx cos (如 切 + Yv cos (%,y) = cos [n9 先 + kv cos % y).
ex dy a%?
But
dy dx
cos (州, = cos (i = -3—,
y) cos (",y) = —— cos (L ％) = ds
,

where t is the positive direction of the tangent. Introducing these values

into the preceding formulae, one finds

(32.3)
ds \ a% /

or in complex form
. d (W , 2.
^n + \ dy
~ %不一
d%
Z I
ds \ 3%
, (32.4)

(X^ + iY^ds = 一 露 (善 + *) £ • (32・5)

Substituting from (31 .4) in (32.5), one obtains

(X* + iY^ds = — n中'(z) + W(z)}. (32.6)
First let the element ds have the direction of the axis 0y. Then
ds = dy, dz = idy , dz = — idy, Xn = Xx, Yn = Xy,
and (32.6) gives
X工 iXy = b(z) +西一 z西 一西. (32.7)
CHAP. 5 COMPLEX REPRESENTATION 115

Next let have the direction of 0%. Then

—
= dx, dz ~ dz dx, X n = — Xy,
and (32.6) gives, after multiplication by i,
Yn= —Y y,

The
—
y” iX® = ?'(z) +河 + 西
formulae (32.7) and
Z

(32.8) are the required expressions

(32.8)
for the
stress components. Adding and subtracting (32.7) and (32.8) and replacing
in the latter case E by —
i, one obtains the simpler formulae
X& + = 2B(z) + 工 4领d(z) = 4求①⑶ = 2[①⑶ 场] (32.9)
匕 一 X。 2iXv = 2[和''⑵ + y(明 = 2(5 '0 子师， (32.10)
where
中'⑶， h⑶ = u⑵. (32.11)
The very useful formulae (32.9) and (32.10) are likewise due to G. V. Ko¬
losov [1] who obtained them in a different manner without recourse to
the stress function.
The expressions, deduced here for the components of displacement
and stress, show that these components, under the earlier stated con¬
ditions, are analytic functions of the variables x, y inside the considered
region, because the functions 平⑵，@(z), 甲(z) possess this property.
A function of the real variables x, y is called analytic in a given region S> if at
each point (力°, 夕°) inside S it may be developed into a (double) series of non-
negative powers of (力 一 (夕 一 y。)， i.e., into a series of the form
工旬,强一&产自一九产
PM
(This definition may be extended to any number of variables) .
As is known, each function of the complex variable z = x 4- iy, holomorphic
in a given region, is analytic in the sense that it may be expanded into series of
non-negative powers of (z — 。
zj near any point z = #() 十 iy0 of that region. On
the other hand, it is easily shown (cf. for example E. Goursat [1]) that every analytic
function of z = % + is an analytic function of 力 and y.
Finally, a remark will be made with regard to Non-Russian work along the lines
of this section. In a recently published paper, A. C. Stevenson [1] deduced formulae
which, in essence, agree with those of G. V. Kolosov and also with some of those
obtained by the Author of this book, all of which have been published considerably
earlier (not only privately, but also in journals well known outside Russia) ;
however, no reference has been made there to this fact.
In a still later publication, H. Poritsky [2] uses formulae which differ only in
appearance from those deduced above ; in a rather vague reference the author
ascribes some of these formulae to the Author, quoting his paper [8] of 1933. How¬
ever, no mention is made of the Author's earlier work and of that by G. V. Ko¬
losov, although this work (which contained the formulae used and had been
published much earlier) is referred to in the quoted paper.
116 II. PLANE THEORY OF ELASTICITY §33

[The following statements were obtained by the translator from the two authors,
referred to above.
A. C. Stevenson wrote that at the time when he worked on the paper, quoted
—
by the Author (i.e. , 1 939 40), he was admittedly ignorant of prior work along these
lines. However, in a paper of later origin, published rather earlier and not quoted
in this book, he was equally clearly at pains to acknowledge the priority of Kolosov
and Muskhelishvili by referring to a total of six papers by G. V. Kolosov, dating
as far back as 1909, of four papers by N. I. Muskhelishvili the first of which appeared
in 1919 and to the combined paper by both authors, published in 1915.
H. Foritsky indicated that he deduced his formulae in 1931, although his paper
was not published until 1 945. By that time the Russian work had been given a fair
amount of publicity in the U.S.A, and he quoted one of Muskhelishvili 's papers
merely for the purpose of acknowledging that he had been anticipated.]

§ 33. The physical meaning of the function f. Expressions for

the resultant force and moment. 1°. The function

八刀
泳
1- z dy
= (p(z) + + W(z)
introduced in § 30 has a very simple mechanical interpretation which
will be elucidated by finding the expression for the resultant vector of
the forces applied to a given arc AB in the region S occupied by the
body. In this context one is concerned with the forces acting on an
element of the arc AB from the right, where the arc is to have the
direction from A to B, in other words, with the forces acting from the
side of the positive normal n, drawn as shown in Fig. 13.
Let (X, Y) be the resultant force. It follows from (32.5) and (32.6) that
r (Xn
—
. rdu duiB
.
— J —
.
X 十 iY = / + Qds = z Hz ~

+ w叫号，
/b
=一论卜) (33.1)
where [ will always denote the increase undergone by the expression
in the brackets as the point z passes along the arc from A to B.
If in this formula the point A is assumed to be fixed at all times,
the point B is permitted to move and its coordinate is denoted by
z = x iy, one finds
/(%, y) = 中⑶ + zd(z) + 殖 =
J
= 勿 {Xn + iY 消 ds + const. = “X 十 勿丫) const.,
CHAP. 5 COMPLEX REPRESENTATION 117

where (X, Y) represents the resultant vector of the forces applied from
the side of the positive normal to an arbitrary arc containing the fixed
point A and the variable point B.
It follows from the above as well as from its mechanical meaning
(cf. 3°) that this resultant vector does not depend on the shape of the
arc joining A and B, except, of course, that it must not leave the region S.
2°. Next, a formula will be obtained for the resultant moment about
the origin of the coordinate system. One has

M 一 yX^ds

which, by (32.3), becomes

M= —/ — F

integrating by parts, one finds

L A J k ex 砂
AB
and finally
方

(33 .2)

But

不 +y 詈*
3% cy KI \联—子
ox cy
and, by (31.4),
au-
— z
au
—— +
= 中⑶
-
zg'(z) + 3(z).
Further,
U = 混｛视⑶ + 明，
so that (33.2) becomes
M = 猊以⑵ 一 + z w0修 (33.3)
These formulae were first given in the Author's paper [11].
3°. Hitherto it has been assumed that the region S is simply connected,
and as a result the functions 中(z), @(z)，Z(z) will be single-valued in S.
1 18 II. PLANE THEORY OF ELASTICITY §34

Thus, if A and B coincide, i.e., if the considered curve is a contour, the

values of these functions will be the same at A and B, and hence one
finds
X= Y=M —
0, (33.4)
as was to be expected. The formulae (33.4) express the fact that the
sum of the external forces acting on a part of the body, contained inside
any contour, is statically equivalent to zero.

§ 34. Arbitrariness in the definition of the introduced functions.

The important question will now be studied as to how far the functions
Y, 少 define the state of stress or the displacements of points of the
body.
First consider the problem of the uniqueness of these functions for a
given state of stress. Expressed in greater detail the problem is as follows.
Let Xx, Yy, Xy be the components of stress for some given state of elastic
equilibrium of a body. As has been shown in § 32, there exist functions
(z), ¥(z) of the complex variable z which are related to X* Yy, Xy by
the formulae
Xx 十 匕 = 4猊① (z), (34.1)
居 一 X, 十 工 = 2［施*) 学 (34.2)
the questions are then: how completely are the functions (z), 乎⑶，
and also the functions

中⑶ = / dz, 3(z) = /x⑶放 (34.3)

determined by the components Xx, Yv, Xy and does there remain some
arbitrariness in their choice ? What is the degree of this arbitrariness ?
There is no difficulty in answering these questions. Let <p13 be
some other system of functions, related to the given components Xx,
Yyf Xy and to each other by the same equations (34.1) to (34.3), as were
the functions V, 少，i.e.,
X“ + % = 4fRQ(z), (34”

— 十 2近， = 2叵到 (z) + 巴(z)］, (342)

中式z) = / z)dz, 队 (z) =，％(z)az. (343)

 CHAP. 5 COMPLEX REPRESENTATION 119

Consider how the functions 力 V1? 鱼，％ may differ from the functions
乎，G 八 Comparing (34.1) with (34.1') it is seen that the functions
i(z) and (z) have identical real parts ; hence these functions may only
differ by an imaginary constant C勿 (cf. Appendix 3), so that
i(z) = + Ci, (34.4)
where C is a real constant.
It follows from (34.4), (34.3) and (343) that
中i = (p(z) 十 Ciz 丫， (34.5)
where 丫 = tx + 50 is an arbitrary complex constant. Further, noting
that by (34.4): = '(z), comparison of (34.2) and (342) obviously
gives
巴⑶ = 里(z), (34.6)
and finally, by (34.3) and (34*3'), one finds
匕⑶ = 少⑶ + Y', (34.7)

+
where 丫' = a' 滔 is an arbitrary complex constant. Thus one arrives
at the following result :
For a given state o/ stress the function 甲⑵ is completely defined, the
functions (z), 中⑵，少⑵ are defined apart from the terms CEz 十 丫,
丫' respectively, where C is a real and 丫，丫' are arbitrary complex constants.
Conversely, it is obvious that a state of stress is not altered, if one
replaces .
中 (z) by 中⑵ +
Ciz + y，
Mz) „ @(z) + y，，
where C is a real and 丫, 丫' are arbitrary complex constants. By this
substitution (z) = <p'(z) obviously becomes (z) + Ci and ¥(z) remains
unchanged.
Next investigate the question as to how far the arbitrariness of these
functions is removed, £/ components o/ the displacements u, e are given.
The components of displacement completely determine the stress
components. Therefore it is clear that, when the former are given, one
may not make substitutions different from those of the type (4). Consider
how these substitutions affect the components of displacement which
were seen in § 32 to be determined by the formula
120 II. PLANE THEORY OF ELASTICITY §34

+ 汨) = xg(z) —— Z?' —— 少(%).

Direct substitution shows that
2MM + 通) becomes 2a(0 十 说J,
where
2M("i + z勿J = 2从(方 拓) 十 (x + V)Ciz+ x了一 (34.8)
and hence, putting v = a + 型， 丫’= a '十 ,
”'i == o， ％
况
°’ + 火， (34.9)
二^

where

2分二
(y + 1)C y za - a/ , %= (x 十 1)0 % H +
6 一 上. (34.10)
卬 卬 2M Zy.

It is thus seen that the additional terms have the form

得 = 一 吵 十 a0) % = " + 3o， (34.11)
where

e =
(x -J- 1)C
, a0 二
xa
一~
—— oc，, xB +
— * ， (34.12)
2[i
and that they express pure rigid body motion. This result had, of course,
to be expected, since the displacements, corresponding to a given state
of stress, are uniquely determined apart from a term describing rigid
body displacement.
Formula (34.8) shows that a substitution of the form (A) will affect
the displacements, unless
C = 0, 乂丫
— 丫' = 0. (34.13)

Thus, for given displacements, it is impossible to select the constants

'1 Y，Y‘ arbitrarily ; if, for example, of the constants 丫，丫' has been
chosen, the other is determined by (34.13).
The arbitrary constants, entering into the above functions, may be
given one or the other value which may be convenient. Assuming, for
simplicity, that the origin lies within the region S, occupied by the body,
these constants will be chosen in the following manner (unless stated
otherwise).
When the stresses are given, the three constants C, 丫，丫' will be chosen
in such a way that
CHAP. 5 COMPLEX REPRESENTATION

cp(O) = 0, 8明o) = 0， 少(0) — 0， (34.14)

where§denotes the imaginary part.
The first of these conditions leads to a suitable choice of 丫，the second
to that of C and the third to that of 丫'. These conditions obviously
remove all arbitrariness as far as the functions 中 and 少 are concerned.
When the displacements are given, a suitable choice of 丫 or 丫' will be
assured by the conditions

中(0) — 0 or w(0) = 0 (34.15)

either of which will completely determine both the functions 中 and 少.
Note still the following fact. It is obvious that the expression
au
诙 •西
au = 6z)+ ——
卬(z) + Wz)
一
(34.16)

completely determines the state of stress of the body, since it determines

the quantities —— , , and hence the second derivatives of U which
specify the components of stress. Consider now the question as to what
conditions must be fulfilled by the constants C, 丫，丫'，so that the trans¬
formation (幺 ) does not only leave the state of stress, but also the expres¬
sion (34.16) unchanged.
It is easily verified that, applying (4), this expression becomes
au . au
3%
Fz —cy FY +Y •

Hence, if —3%-—— oy
i —
be given, one must have 丫『 + = 0. Thus the
constant C and one of the constants 丫 and 丫' may be chosen arbitrarily.
One may, for example, put
9(0) = °， 济6(0)} 0 (34.17)
and in this way completely determine the functions

§ 35. General formulae for finite multiply connected regions.

平 and
“
Consider now the case when the region S, occupied by the body, is
multiply connected. For simplicity assume that the region is bounded
by several simple closed contours Llt L2, . . Lm, Lm+1 (i.e., by contours
122 IL PLANE THEORY OF ELASTICITY §35

which do not intersect themselves ; for more detail see § 37) ; the last
of these contours is to contain all
the others, as is shown in Fig. 14
(e.g. a plate with holes). Further,
assume that these contours have no
points in common.
It will be remembered that, by
supposition, the components of stress
and displacement are to be single¬
valued functions. In spite of this fact
the functions 勺 and 少 may, in this
case, be found to be multi-valued.
However, it will be noted, on the basis
of the statements of the preceding
sections, that these functions will be
holomorphic and hence single-valued in any simply connected part of
the region S occupied by the body. Thus the functions cp and 少 are
analytic in S (cf. § 30).
The above circumstances will now be explained in detail. Let S' be some simply
connected part of S. One may define the functions % 少， corresponding to a given
state of elastic equilibrium of S', by (arbitrarily) fixing the undetermined constants
introduced in § 34. These functions have been shown there to be holomorphic in S'.
But if one continues these functions analytically beyond S' (remaining, of course,
in S), then, by describing a closed path and returning into S', one may not return to
the former values of 中 and 山. However, it is easily seen that the new values of these
functions can differ from the old ones only by terms of the form stated in § 34,
because both values correspond to one and the same state of elastic equilibrium.
This fact also follows from (35.10) and (35.11) below.
The type of multi-valuedness of the relevant functions will now be
studied. First of all, the formula
XX + 号 = 4沉①⑵
shows that the real part of is single-valued (since, by supposition,
the left-hand side of the equation is so). But this does not yet mean
that also its imaginary part is single-valued. In fact, for one circuit
(e.g. anti-clockwise) around some contour 匚，surrounding one of the
contours Lk, this imaginary part may undergo an increase Bki, where
is a real constant (cf. Appendix 3). Introduce, instead of the constants
Bk> other real constants Ak, defined by
k 加.
CHAP. 5 COMPLEX REPRESENTATION 123

Next consider the function

*(z) =<D(z) —e£41og(z —
=1
(35.1)

where z2> .. zm
denote fixed points, arbitrarily chosen inside the
contours Llt , Lm (i.e., outside S). Since log (z 年) undergoes
an increase 2兀％ when z passes once around Lk (anti-clockwise), the ex¬
—
—
pression A k log (z zQ increases by 2niA 加; the remaining terms under
the summation sign in (35.1) will revert to their former values. Hence
returns to its original value for a circuit around any closed contour
in S.
Thus one has
m
(z) = Z 4 log (z
k—1
— 勿) + *(z), (35.2)

where *(z) is holomorphic and hence single-valued in S. Further, one

obtains for 乎 (z)
z

q⑶ =
%
/ "z 十 const. =
W
m 广
= S A^ — 啕 log (z — 力) 一 (z — z/} +
J
/ *(z)dz + const.
%

z
where zQ is an arbitrarily fixed point of S. But the {z)dz repre-
%
sents itself a function of the complex variable z which for a circuit
of one of the contours Lk may undergo an increase of the form
2次比，
where ck is a constant which, in general, will be complex (and the
factor 2兀％ has been introduced for convenience). Hence, proceeding as
before, one can write

广 m
/ *(z)dz = S ck log (z — z/ +a single-valued function.
J
%

Introducing this expression into the preceding formula and combining

terms of the form A^k log (z — z?) and c 无 log (z %), one obtains —
124 H. PLANE THEORY OF ELASTICITY §35

m
—
打8

少⑶ = zHAk log (z + S 丫九 log (z — %) + 申*(z), (35.3)

where <p*(^) is a function, holomorphic in S, and 丫加 are constants (which,

in general, are complex).
Finally, it is seen from
匕一 十 2z% = 子(明

that ¥(?) is a holomorphic function. Whence it follows that the function

少⑵ = {z)dz,
J
as before, may be written

W(Z) = £ Y/；log (z
b=1
— 肛.) + 炉⑶， (35.4)

where 丫[ are certain (generally complex) constants and ^*(z) is a holo¬

morphic function.
In an analogous manner one has for the function

the expression
X(z)
=/5»
m 加？ .
y(z) = z 2 y[ log (z
左 =1
— z&) + k=1 E 丫[ log 3 — zj + /* (z) , (35.5)

where 丫怖 are (generally complex) constants and %*(z) is a holomorphic

function.
Hitherto no consideration has been given to the condition of siitgle-
ahtedness o/ displacements. By (32. 1 ) one has

2*(〃+ 谢) — 乂中 (z)
— z<p'(z) —— W(z).
Substituting in this formula the expressions found above for 平 (z) and
W(z), it is immediately seen that

+ 珂以 = 2拓{(y + 1)/庐 十 yy + Yfc}, 左 (35.6)

where [ ]以 denotes the increase undergone by the expression in brackets
for one anticlockwise circuit of the contour £] Hence it is necessary
and sufficient for the single-valuedness of displacements that in the
formulae (35.1) — (35.5)
CHAP. 5 COMPLEX REPRESENTATION 125

月乃 =0, 乂丫加
+ = 0, k ~ 1,2, . . .,m. (35.7)

It will now be shown that the quantities yk, 丫［ may be very simply
expressed in terms of Xkf Yk> where Y0 denotes the resultant
vector of the external forces, exerted on the contour J (力 = 1,2, …， 加)，
applied to Lk from the side of the external normal 处 to S (Fig. 14).
This resultant vector will now be actually computed. For this purpose
use may be made of (33.1) which is obviously valid also for the multiply
connected region S under consideration here, provided only the contour
AB referred to in this formula lies entirely inside the region S. This
formula may also be applied in the case when the contour AB belongs
partly or completely to the boundary of the region, assuming that
certain assumptions have been made with regard to the behaviour of
the functions <p(z), W(z) near the boundaries (cf. for more detail § 42).
At this stage no additional assumptions will be introduced ; instead
of considering the forces applied to Lk> the forces acting on some closed
contour L'k in S from a suitable side (namely from that side which faces
the side of the contour LQ will be studied, where L'k is to contain Lk
and no other contour of the boundary of the region (Fig. 14). The
resultant vector (Xk, of those forces does not depend on the choice
of the contour Lrk, provided this contour satisfies the above conditions.
This is obvious on physical grounds and also follows from the formula
(36.8) below. In the capacity of the contour L'k one may select any curve
which lies arbitrarily close to the contour Lk.
Starting from this it may be assumed, without imposing any con¬
ditions on the behaviour of the functions 中 (z), 少 (z) near the contour Lk>
that by definition the resultant vector of the external forces applied to
Lk is equal to the resultant vector of the forces applied to L'k from the
appropriate side.
An analogous definition may be established for the resultant moment.
This formula shows that the resultant vector (Xk, does not depend
on the choice of L'k as long as this contour contains only Lk and no
other contour of the boundary of S.
By (33.1), applying it to the contour L'k, one has now

Xk + iYk = — *(z) + z^(z) + 少(z)］ 以.

In this formula it has been assumed that the contour is traversed in the
direction for which the normal n points to the right. But in +he present
126 II. PLANE THEORY OF ELASTICITY §36

case the normal n must be directed outwards with respect to the region S
(Fig. 14), because one requires the resultant vector of the external forces.
Consequently, in the preceding formula, the contour Lk must be traversed
in the clockwise direction (assuming, of course, that the axes 0%,Oy are
right-handed ; see Fig. 14). Taking this fact into consideration, one easily
obtains, using (35.3) and (35.4),
X% 十 %丫九 = — 2 （y为 一 %）.
九 （35.8）
Formulae (35.7) and (35.8) give then
x + 论 , % (xk— (35.9)
2k ( 1 + x)
'' 2k(1 + x)

Using (35.9) (and also the fact that Ak = 0), formulae (35.3) and
(35.4) may finally be written
] m
中⑶ = 一 c兀
2 (1 十 F
“
£禺
x) 尼 =i + 2•片) log (z — %) + 平*(z), (35.10)

v m
W(z) =
… …
I K
2k(1 十 k) 及=1
S 陷 - log (z - 弘) + W*(z). (35.11)

§ 36. Case of infinite regions. From the point of view of application

the consideration of infinite regions is likewise of major interest. For the
present the investigation will be limited to the case when the region S
consists of the entire plane Oxy from which finite parts, bounded by
simple contours, have been removed (infinite plate with holes) . The boun¬
dary of such a region consists of several simple contours Llt L净…， Lg
which is the limiting case of the region considered in the preceding sec¬
tion, when the contour Lm+1 has entirely moved to infinity.
The formulae of § 35 hold, of course, for any finite part of S. There
only remains to study the behaviour of the functions 9 and 少 in the
neighbourhood of the point at infinity in the plane Oxy.
Draw about the origin as centre a circle Lr with radius R sufficiently
large so that all the contours Lk lie inside LR. For every point outside Lr
one obviously has
I z [ > I 次],
and hence
CHAP. 5 COMPLEX REPRESENTATION 127

log (Z — Z%) = log Z 十 log ( 1

= log z 十 a function, holomorphic outside LR.

Therefore, by (35.10) and (35.1 1),

? log
兀0+X)
*(2), W(z) = 兀
2 (1 十 守B log z+Q*(z), (36.1)

where
m 扰

X = EX 1c ,
y=EV无 (36.2)
=1 无 fc=l

and 中**(z), W**(z) are functions, holomorphic outside LR with the

possible exclusion of the point at infinity. Obviously X and Y are the
components of the resultant vector of all external forces acting on the
boundary of S, i.e., on the union of the contours Llt L2, . . Lm.
A function will be called holomorphic at the point z = co, if in the neighbourhood
of that point (i.e., for sufficiently large | z ]) it may be represented by a series of
the form
— 丁 ... •
«2
ao n। H।
z
।

By the theorem of Laurent, the functions 甲**(z) and may be re¬

presented outside Lr by the series
8 OO

乎** = £ anzn, W**(z) = £ (36.3)

*" 'OO *'' *'OO

which will converge uniformly for every finite region outside LR. The
above theorem is known to hold for a function, holomorphic inside a ring
bounded by two concentric circles Z】and L2, where Z] may be shrunk
into a single point and may become infinitely large.
This is all that may be said with regard to the functions 中 and 少，unless
additional conditions are introduced with respect to the distribution of
stresses in the neighbourhood of the point at infinity of the plane.
Introduce now the following condition : the components of stress are
bounded throughout the region S. Consider what must be the functions 平
and %, in order that this condition is satisfied.
By (32.9) and (32.10)
128 H. PLANE THEORY OF ELASTICITY §36

X, 十匕 = 2回(z)
+ <?'(“
% — Xr 十 2iXy = 2[狗〃⑵ + U(z)]・
Introduce into the first of these formulae the expression (36.1) for
中 (z), replacing 华**(z) by (36.3):

Xa；+
”丁
,
Yy =
(
2s-
I
X + iY 1
r
2兀(1 十 x) z
X —iY 1
2k(1 +% 2 +
襄/
£双
_8
+即
1
.

The only terms which may grow beyond all bounds with | z ] arise from
the series
8 OO

S %(距2位-1 + anzn-1)
71 2
— S nr^1?% =2
+ 7区8-1灯，

where z —
夕N*. Whence it follows that for

as 夕 f 8, one must have

Xx + Yy to remain finite,

a 7i 0 ( %* 2) •
‘‘ n

Assuming this condition to be satisfied, it is easily seen in an analogous

manner from (b) that it is neccesary and sufficient for the boundedness of

that
y. — + 2ixy
OO

S 找yLl a'仍7环
2

remains bounded, whence it follows that

a；= 0 (" 2).
Conversely, it is obvious that Xx, Yy, Xy will be bounded, if these
conditions are satisfied. Hence one has finally

<p(z) = — 2九(1+ 乂)
K logz 4- r? 十 优(z), (36.4)

少(Z) =
2兀 (1 + x)
+ F" + 少o(z), (36.5)

where

r =石 ic, r‘ = B' + icr (36.6)

CHAP. 5 COMPLEX REPRESENTATION 129

are constants, generally complex, and 的⑶，必⑶ are functions, holo¬

morphic outside Lr, including the point at infinity, i.e., for sufficiently
large | z | they may be expanded into series of the form
a。
f f
仇i a ,
—
, 、

z" +
〉

W
w-\
?o(z) = 曲 H 卜 ••• , 少o(z) = «0 H 1 z”晟+ ••• (36.7)
z z
(where, for convenience, a2 etc. have been written instead of 沈一】，«_2, etc.).
On the basis of § 34 the state of stress will not be altered by assuming
况 = 劭 = 0,
ie, °
<?0(8) = 30(8) = 0，
and, in addition,
C = 0.
The real constants B, Bf , C', introduced into (36.4) and (36.5) by means
of r and I5', have a very simple physical interpretation. In fact, it follows
directly from (0 and (6) that for z f 8
Jim (X* + y/ = 43, lim {Yy — %+ 2iXv) = 2 「= 2(夕 十 £(7), (36.8)
whence
X竺 = 28 —
y片 = 28 十 方， X俨 = C. (36.9)
This means that in the neighbourhood of the point at infinity the stresses
are uniformly distributed (or rather that their distribution differs from a
uniform one by infinitely small quantities) . Let N& be the values of
the principal stresses at infinity and a the angle made by the direction
of N] with the axis Ox. Comparing (36.8) with (8.12) one finds
猊 r = B = ±(M + N2),

T' 三 B 十 iC
-— MN1 — 现)歹2吗
The constant C, which does not affect the stresses, may be related to
( " )

the rotation of an infinitely remote part of the plane. The expression for
the rotation is

的/
,
3%
(cf. § 14, where it is denoted by 夕)， whence, by (30.8),
入 十 2M (Bq 净) 1 十 代 物 涉)
* 2r(入 [i) \ 3% 42 卜加 dy/
130 II. PLANE THEORY OF ELASTICITY §36

1 +乂 乎 '(z) — <p'(z) (36.11')

2；
since, remembering that
涉 纽 and，砂 z.因
~

效 加
—8% 1- dx- = (p (z), -
one has
变 — •国 =西，里 =2_r(2)— 西］.
It follows immediately from (36.11'), (36.4) and (36.5) that
1 +
-孙
X
~~7- (,>
%OO c

and hence
k
2mg
U = (36.12)
1 X +
It will be noted at the same time that the state of stress characterized
by the linear functions
平 (z) = (B + iC)z + const, {B' + iC')z— + const
is homogeneous : the stresses are uniformly distributed, i.e., the stress
components (and hence the strain components) are constant quantities.
In fact, the components of stress are expressed by (36.9), if the superscript
oo is omitted.
Next consider the behaviour of the displacements at infinity under the
assumed conditions in the general case. For this purpose use will be made
of (32.1) which by (36.4) and (36.5) becomes
x(X + zY)
2以〃 十 谢）= log (z5) + (乂 r — r)z — 可5 + (36.13)
2兀(1 十 x)
where terms remaining bounded for large values of | z | have been
omitted. It is easily seen from (36.13) that, generally sneaking, the dis¬
placements will not be bounded at infinity under the conditions introduced
so In order that they may be bounded, one obviously has to impose
the conditions
x = y = o,「= r = o. (36.14)
The first group of these conditions postulate that the resultant vector
of all external forces acting on the boundary of the region is zero, while
CHAP. 5 COMPLEX REPRESENTATION 131

the second group demand that the stresses at infinity vanish and, besides,
that infinitely remote parts of the plane do not undergo any rotation.
Note also that even in the case of rotation the stresses at infinity are
zero and that in the absence of rotation (C = 0) the displacements
increase like log(z2) = 2 log r, if the resultant vector (X, Y) is not zero.

NOTE. In (36.4) and (36.5) the functions 的⑵,少°(z) are holomorphic

outside any circle, enclosing all contours L1} L2, .. Lm. If there is
only one such contour Lx (plane with one hole), it is easily seen that <p0(z),
少o(z) will be holomorphic throughout the region S, provided only the origin
of coordinates is taken outside S (i.e., inside the hole). In fact, in this case
(35.10) and (35.1 1) coincide with (36.1), if one puts in the former
Zi =0
and replaces <p*(z), 少*(z) by respectively. But the functions
W*(z) are known to be holomorphic throughout S with the possible
exclusion of the point at infinity, which proves the assertion.

§ 37. Some properties following from the analytic character

of the solution. On analytic continuation across a given contour.
Some terms and propositions. In proceeding to the fundamental question
of the complex representation of the general solution of the equations of
the plane theory of elasticity certain terms will be defined which will be
used (and partly have already been used) and certain simple propositions
will be recalled.
1°. In the sequel, when speaking of lines (arcs, contours) , one will have
in mind (unless stated otherwise) simple, non-intersecting, open or closed
continuous curves. Further, it will be assumed without often stating it
specifically that the curve under consideration is smooth or, more gener¬
ally, sectionally smooth. It will be recalled that a line is called smooth,
if its tangent varies continuously; lines are called sectionally smooth, if
they consist of a finite number of smooth arcs. It will not be necessary
to give here more detailed definitions of these well-known concepts.
2°. It will be assumed (unless stated otherwise) that the region occu¬
pied by the body represents a connected, finite or infinite part 5 of the
plane bounded by one or several (simple, smooth or sectionally smooth)
closed contours. The boundary of the region S, if it is finite, thus consists
of a finite number of closed contours Llt L2, . . . , Lmr Lm+1 which have
no points in common and one of which, say Lm+1, encloses all the others ;
132 II. PLANE THEORY OF ELASTICITY §37

none of the other contours enclose each other (plate with holes, Fig. 14).
For 加 = 0, the boundary consists of a single, closed contour and the
region S is simply connected ; for w I, the region is multiply con¬
nected *, i.e., (加 l)-ply connected.
In the case of an infinite region, the outer contour Lm+1 is absent (or,
as will often be said, it lies at infinity) . In this case, the region S repre¬
sents the infinite plane with holes **.
In general, in what follows (unless stated otherwise) , a region will be
understood to be a (finite or infinite) region of the type described above.
The boundary L o/ the region S will not be included in S. If some
property is not only true for points of the region S, but also for points
of the boundary L or for points of some part Z/ of the boundary, it will
be said that such a property is valid for S + L or for S L' respectively.
A part 。/ 应 boundary will always be understood to be a part consisting
of one or several continuous arcs or contours.
3°. Let F(x, y) be some function defined in the region S (but not on
its boundary L) and continuous in S.
It will be said that the function F(%,y) can be continued continuously
onto the part Z/ of the boundary L (where Z/ may coincide with L),
if one can give the function F{x, y) such values on L' that the resulting
：
function is continuous in S + L Then it will often be simply said that
the function y) is continuous in S + or continuous in the region
S %夕 to L' ; this will mean that the function F(x, y) has been given
suitable values on
Let (%。, y0) be some point of the boundary L and let F(x, y) tend to
a definite limit as the point (%,y) approaches the point (%°, y°) from
inside S, but otherwise in an arbitrary manner. Frequently this means
that the point approaches the point (%0, yj along any line which remains
inside S. It is not necessary to assume for this that the ''path'' is a
continuous line; for example, it may consist of a number of individual
points. It will then be said that F(先，y) has (or assumes) a definite
boundary value at the point (力°, y0) or that F(%, y) continued continu¬
ously up to the point (&, %). The limit just defined will always be referred
to as boundary value.
* A definition of simply and multiply connected regions has been given in
§ 15; cf. also Appendix II.
** The infinite region S bounded by one (simple) closed contour (infinite plane
with a single hole) can be referred to equally well as a simply or doubly connected
region depending on whether the point at infinity is included in the region S or not.
CHAP. 5 COMPLEX REPRESENTATION 133

It is readily shown that, if a function F(%, y) can be continued con¬

tinuously at all points (g, %) of some part Lr of the boundary L (where
L' may be the same as L) and if F(%0, %) denotes the boundary value of
F(%, y) at the point (%0, %) ,then F(%o, %) will be a continuous function
of the point (%0, y0) on E'.
From the definition of continuity follows then that, if F (x, y) is
continued continuously at all points of the part Lf of the boundary L,
F(%, y) will be continuous in S + L', i.e., continuous in S up to L',
provided one understands by F(%, y) for (%,y) on L' the corresponding
boundary value.
In future, when saying that F(xQf is the boundary value of the
function F(x, y) or that F(x, y) assumes the boundary value F(xQ, y0), it
will always be implied that F(%(), %) is the limit approached by F(x, y)
as the point (%,y) tends to the point (%。,％) in an arbitrary manner,
provided only that it is to remain all the time inside S. In other words,
the term boundary value at a given point or o% a given part of a boundary
always means that the function under consideration is continued con¬
tinuously at the given point or on the given part of the boundary.
4°. Let the boundaries of two regions and S2 which do not cover
each other have the common part L representing a simple, smooth (or
sectionally smooth) arc or closed contour and let F^z), be functions
—
of the complex variable z x iy, holomorphic * in 工 and S2 re¬
spectively and continuous up to L. (Fig. 15). Further, let F“z) =
on L. Then the function F .defined by
F 1G ) for z in and on L
{ F^z) for z in S2 and on L'
* The term holomorphic1 is equivalent to the term "regular'' used in V. I.
<( ;

Smirnov [1], Vol. III. A function holomorphic in a given region (which is simply
or multiply connected) is always single-valued. In the sequel, an analytic function
of the complex variable z in a given region S will always be understood to be a
function which can be multi-valued, but each continuously varying branch of
which is holomorphic (and hence single-valued) in any finite simply connected
part of the region S.
The word "analytic" denotes that each, function (more correctly, each of its
branches) can be expanded near any point a of the region S in a series of the form
+ — —
a) 十
2(z 力 ...
Sometimes a function analytic S will be understood to
in
a) 立 十
be a function analytic
(in the above sense) in a region obtained from S by exclusion of certain individual
points ; however, this will always be stated specifically.
134 口. PLANE THEORY OF ELASTICITY §37

will be holomorphic in the region S, obtained by addition of the regions

Si and S2 including L. The proof of this theorem can be found in most
treatises on the theory of functions of a complex variable.
In particular, this theorem leads to the following conclusion. Let the
function 万⑶ be holomorphic in some regions S and let the function
F(z) have on some part L' of the boundary of this region a boundary
value which is equal to zero. Then F(z) = 0 in the entire region S.
In fact, let there be added to the region S some part S' of the plane
adjoining L' from the other side and let F(z) = 0 in S'. Then, on the
basis of the above statements, 尸(z) will be holomorphic in the region
S + S' and, since it is zero in S', it will vanish everywhere, because an
analytic function which vanishes in a part of a region vanishes every¬
where in the region., where part of a region means a subregion.
If in this terminology and under the same conditions as above the
boundary value of 尸(z) on Z/ has the constant value C which is not
necessarily equal to zero, then F(z) = C in the entire region S. This
result follows from the preceding proposition, if one applies it to the
—
function F(z) C.
5°. The functions 叫z), 少(力， ¥(z), occurring in the general solution,
are analytic functions of z in the entire region, occupied by the body,
even in the case when that region is multiply connected. This follows
from the expressions for these functions, deduced in the preceding
sections. The only difference from the case of a simply connected region
is that the functions 中(z) and @(z) may be found to be multi-valued, as
a consequence of the presence of logarithmic terms. (If multi-valued dis¬
placements are admitted, the function (z) may also turn out to be
multi-valued) . Since an analytic function of z = x iy leads at the
same time to analytic functions of the real variables %, y (cf. end of
§ 32), the components of stress Xx, Yv, Xy and displacement e are, as
in the case of simply connected regions, analytic functions of %, y through¬
out the region, occupied by the body.
From this property of the solution there follows immediately a propo¬
sition which at first sight may appear somewhat unexpected.
If some part of the body (Ze, a subregion which may even be arbitrarily small)
。
is in a ^natural" state, £.e., no stresses occur there, then the whole body is in
"
natural state or, in other words, no stresses occur anywhere.
In fact, if Xx = Yy = Xy ~ 0 in some part of S, this will be so in the
whole of S, because an analytic function cannot vanish in a part of a
region without being zero in the whole region.
CHAP. 5 COMPLEX REPRESENTATION 135

The proof will now be given of a simple and important proposition

concerning the analytic continuation of the solution across a given
contour. Let there be two regions S+ and S~ which do not cover each
other, but the boundaries of which have a part in common consisting
of a smooth line L (i.e., an arc or contour). Assume that the components
of displacement and stress satisfy in each of the regions S+ and S~
the conditions of § 27. In that case they will be analytic functions in
each of the separate regions S+ and
Consider the necessary and sufficient conditions for the components
of stress and displacement to be analytic in the region S, obtained by
joining 5+ and (including £). If uf〃Xx, Yy, Xy are analytic in the
whole of S, it is obvious that they will be continuous on L from S+ as
well as from and that their boundary values on L from either side will
be equal. Denoting the boundary values, obtained by going to the limit
—
-
from S+ and S~, by superscripts (+) and ( ) respectively, one finds the
necessary conditions
〃+ = 厂， = 符， 埒 = Y；on L, (37.1)
where (X*, Y*), 丫力 are the stress vectors, applied to an element
of L at the point 力 when that element is assumed to belong to S+ and S~
respectively, i.e.,
= Xj cos (%,冗) + X； cos (肛 y),
cos (观,％) +
Y* = X；
"
cos (%,y),
and similarly for X~, Y~, where n is the normal to L at the relevant
point, directed to a definite side (which may be chosen arbitrarily) .
It will be shown now that the conditions (37.1) are sufficient (assuming
the existence of the boundary values of the components of displacement
and stress from either side). These conditions simply express the fact that
the displacements remain continuous for a passage across L and that the
stresses acting on an element of this line from either side satisfy the law
of action and reaction. It follows from the first two conditions of (37.1)
that
—
2fz(w + 说) = xcp(^) 的'⑶ 一 W(z) (37.3)

is continuous on L from both sides and that the boundary values from
either side are equal. Further, it follows from the two latter conditions
of (37.1) that the same properties may be ascribed to
136 II. PLANE THEORY OF ELASTICITY §37

工
au
0%
F z
au
oy =
—— ——
<?(^) + zq/(z) 十 W(z) , (37.4)、

provided a proper choice has been made for the arbitrary constants
which must be added to one of the functions 9, W in the regions S+ and S-.
This is obvious since, by (33.1), one has for (37.4), in both S+ and S~,
the formula

__p iY^ds, (37.5)

where the integral is taken along an arbitrary line I, remaining all the
time (with the exception of the point 霖) in S+ or S_ and joining some
fixed point of 工 to a point z in S+ or by bringing the point z into
the neighbourhood of some point t of L, from S+ or S~, the line I may be
chosen so close to L (i.e., close in the sense of distance as well as of tangential
direction) that the integral on the right-hand side of (37.5) is approximated
as closely as one pleases by the integral
t t

E
a
“X + 味 iYi)ds = ij
a
阳 + iY^ds,

taken over an arc of L, connecting a and t.

Adding (37.3) and (37.4), it is immediately verified that the function
中(z) is also continuous on L from S+ and S~ and that its boundary values
from either side are equal. Hence, by what has been said above under 4°,
the function 乎⑵ will be analytic in S; but then cp'(z) will likewise be
analytic. It is then obvious that 少 (z) is continuous on L from both sides
and that its boundary values are equal. Thus %(z), like 平 (z), will be
analytic in S. This proves the proposition.
The following result is easily deduced from the preceding ones:
If 0% some part [however small) of the boundary of a body
Xn = Yn =〃= 0 = 0, (37.6)

then the stresses are zero throughout the body. This result is due to E. Al¬
mansi [3] who proved it in a different way for the general three-
dimensional case.
Let S be the region occupied by the body and U that part of the
boundary where (37.6) is fulfilled. Select some region S', adjacent to U
and outside S. By the above and by (37.6), the functions Xx, Yy) Xv>
CHAP. 5 COMPLEX REPRESENTATION 137

纵 方may be continued analytically from S into S' by simply putting

these functions equal to zero in S'. But then, by what has been said
earlier, one finds that Xx=Yy = Xy = u = v = 0, because these
functions, being analytic in the region obtained by adding S and
vanish in the part S'.
NOTE. The results on the analytic continuation through a given
contour, proved above, may be somewhat generalized. In fact, retaining
the condition that the components of displacement must be continuous
on L from S+ as well as from S~, one can replace the corresponding con¬
dition for the stress components by a weaker requirement which is
more natural from the physical point of view, namely that the ex¬
pression (37.4) must be continuous up to L. This condition is easily seen to
lead to the following. Select some (smooth) arc Z+ (or l~), entirely in S+
(or S~) and close to L, and suppose that this arc tends to some arc % of the
line L ; further, let (X, Y) be the resultant vector of the forces, applied
to Z+(or %-) from the sides, facing S+(or S) Then, as Z+(or 广) approaches I,
this resultant vector tends to (X+, Y+) [or Y-)] which, by sup¬
position, is the resultant vector of the forces, exerted from the sides
S+(or S~) on the arc I of the boundary of the body.
Provided the stated conditions are fulfilled, it is easily seen that (37.1)
can be replaced by
u+ = %-, e+ = v~, X+ + = 0, y+ -|- y- 0, — (37.1')
where (X+, Y+) and {X~, Y~) are the resultant vectors of the forces
applied from the sides S+ and S~ respectively to an arbitrary arc I of the
line L.
In the same manner (37.6) may be replaced by
X=Y = % = » = 0, (376)
where (X, Y) is the resultant vector of the forces, exerted on an arbitrary
arc of the boundary.
§ 38. Transformation of rectilinear coordinates. Consider now
how the various functions, corresponding 切 a given state of stress of a body,
change under transformation from one system of rectangular rectilinear
coordinates to another.
First investigate the effect of the translation of the origin to a new
point (&, y0). Let (%, y) and (%,%) be the coordinates of the same point
in the old and new systems and let
z = 十次， Z] = 巧 + E%.
138 II. PLANE THEORY OF ELASTICITY §38

Obviously
z = + zo， (38.1)
where “
= %g 十 个0.
Beginning with the formulae
X, 十 匕 = 4汲① (z), 7, —^ + 〈z)+ ¥(z)], (38.2)
= 2[淡
denote by 式无) and 甲式无) the functions playing in the new system the
same parts as (z) and X(z) in the old one. Since the stress components
are not altered by a translation, one has by the first equation of (38.2)

whence
求①(力=例)心)= 沆①
-
= i(z — 啕. (38.3)
One might have added on the right-hand side any purely imaginary
constant which would have no influence on the distribution of stress.
The second formula of (38.2) gives
加(z) 十 中⑶ = 蓝①; + k㈤ —啕 (z — 以 %(z Zo) = —
whence, by (38.3),
= 5 i(z ~

—。
z°) + ¥\(z — z()) 5 i(z — z°),

¥(z) = %(z — z/— 5°<X(z — zo). (38.4)

Integrating (38.3) and (38.4) with respect to z, one obtains
平 (z) = 中i(z — No), Mz) = “(z — Zo) — 50d (z — 漏)， (38.5)
where certain arbitrary constants which do not affect the stress distri¬
bution have been omitted.
It is seen that the function 中⑶ is not invariant for a translation of the
origin, i.e., the values for the old coordinates are not obtained by simply
replacing in “(zj the variable % by z
(p(z) is invariant.
—
zQ. In contrast, the function

Next consider the effect of rotating the axes, leaving the origin fixed.
If the axis is turned with respect to Ox by an angle a, then
% i cos a ——％ sin a,
y = % sin a 十 乃 cos a,
whence % + W = (%.+
CHAP, 5 COMPLEX REPRESENTATION 139

i.e.,
= z^, zr = ze-%
z (38.6)
In view of the invariance of Xx + Yy, one has, on the basis of the
first equality of (38.2),
= 猊①式zj = 泥① (ze*),

whence, omitting a purely imaginary constant term,

= 总产). (38.7)
Further, the expression corresponding to

but referring to the new system, is by (8.8) equal to

(匕一k+ 2区)碑巴
Thus, by a formula analogous to the second formula of (38.2),
之何㈤ + 巴㈤ = 却)3 + ¥(z)W啊
whence
2 (z) 十 里 [zeia^(ze~ia) 巴3-以生出吗
Further, noting that by (38.7) '(z) = , one finds
子⑵ = 于1体-汝)6-26a. (38.8)
Integrating (38.7) and (38.8) with respect to z and omitting arbitrary
constants which do not influence the stress distribution, one obtains
中⑵ = %(ze"a)#, Mz) = ^zeria)e-ix. (38.9)
Finally, integration of the second of the preceding formulae gives
四)=石四*), (38.10)
where again an arbitrary constant has been omitted.
NOTE. If the arbitrary constants had not been omitted, one would
have found, for example, instead of (38.9) and (38.10)
平⑵ = gi(zLa)* + eV2 +夕 十 通， 小⑶ = 队整6一巧歹也 + 夕’+ 步，
where C, a, b, a', are arbitrary real constants which do not affect the
stress distribution. In the earlier formulae
C = 口 = 6 = 屋 = 6' = 0.
140 II. PLANE THEORY OF PLASTICITY §39

Thanks to that choice of constants, not only the stresses, corresponding

to the new and the old functions, but also the displacements will be the
same-. (Otherwise the latter would have differed by a rigid body displa¬
cement.) Further, by omitting an arbitrary constant in (38.10), it has
been ensured that the stress functions U, formed by means of the new
and the old functions, will be identical. This would not have been the case
for a different choice of constants, since stress functions, corresponding
to one and the same state of stress, can differ from each other by an
arbitrary term of the form: Ax + By 十 C.

§ 39. Polar coordinates. In many cases it is convenient to express

stresses and displacements in polar coordinates. Let the origin 0 of the
system Oxy be the pole, and Ox the
polar axis. Then, if r and 眇 are the
polar coordinates of some point M(x, y),
one has, with an obvious choice of the
angle,
z= x iy ~ (39.1)
Draw through the point M two axes ;
being a prolongation of the radius
vector [on the side of increasing 打，(眇),
perpendicular to the first [to the side
of increasing A; Fig. 17].
Let %,% denote the projections of
the displacement at M on to the axes (/) and (分). These quantities are
called components of displacement in polar coordinates. By known
formulae of analytic geometry
u — vr cos A ~ sin A, v = vr sin 8 十 万分 cos 分，

where (〃，劝 are the components of the same displacement in the cartesian
coordinate system Oxy; thus
比
+谢= + 玲， ％ + = + 谢) 吗 (39.2)
whence by (32.1) '^{yr + = 歹叫宁⑵ 一 z否 一 丽]. (39.3)
This formula gives expressions for vr and »眇 in polar coordinates, if one
replaces on the right hand-side z by ye'* and separates real and imaginary
parts.
CHAP. 5 COMPLEX REPRESENTATION 141

The components of stress in polar coordinates are defined in quite the

same way as in cartesian coordinates, with the distinction that the part
of the axes Ox and Oy is now played by and (A) at the point M where
the stresses are to be studied. Denoting temporarily the axis (y) by Mx'
and (9) by My' t the above-mentioned components are
X；，， X；,.
In the sequel, the following notation which is widely employed in literature
will be used for these components :

〃= X], 98 = y；,, 心 — Xy,.

Thus〃denotes the projection on of the stress acting on the plane
normal to ; 分眇 is the projection on (8) of the stress acting on the plane
normalto (^). Finally, 歹9is the projection on (眇) of the stress acting on the
plane normal to or the projection on of the stress acting on the
plane normal to (9).
By (8.8)
方 + 藤 = 4沉① (z) = 2[① (z) + 函], 修用
薪一分 2^= 2历① ，(z) + 子(切於吗
These formulae enable one to calculate the components of stress in
polar coordinates.
By subtraction one obtains from (39.4) the useful formula
n~iri = (z) +跖一 十 可(z)], (39.5)

—
giving the stresses acting on an arc of the circle r const, from the side
opposite the centre.
These formulae are analogous to those given by G. V. Kolosov in a
somewhat different form.

§ 40. The fundamental boundary value problems. Uniqueness

of solution. The fundamental boundary value problems will be defined
in complete analogy with those formulated in § 20 for the three-dimen¬
sional case. As before, absence of body forces will be assumed.
First fundamental problem (Problem I) : To find the
state o/ elastic equilibrium for given external stresses allied to the boundary
L o/・ the region S.
 II. PLANE THEORY OF ELASTICITY §40

Second fundamental problem (Problem II): To find the

state of elastic equilibrium for given displacements of points of the boundary L.
By S will be understood a region of the form discussed in § 35 and
§ 36, by L the union of the contours Llt … •，工 乙仇+1 (if the region
阳，

is finite, cf. § 35), or of the contours Llt 工2，.…，上旭 (if the region is in¬
finite, cf. § 36). In the sequel, unless stated otherwise, it will be assumed
that all considered contours are smooth lines (i.e., that they have con¬
tinuously varying tangents). If S is infinite, it will be assumed that the
stresses in infinitely remote parts of the plane satisfy the conditions of
§ 36, i.e., that they remain bounded.
In addition, in the case of Problem I for infinite regions, it will be as¬
sumed that the values of the stresses at infinity are known ; by § 36
they will enter into the constants
猊・r= B, r = B' + %C'. (40.1)
Further, since the constant C (remembering that 「= 5 + EC) does not
influence the stress distribution, let C = 0.
In the case of Problem II for infinite regions, it will be assumed that the
quantities
r = B + ic, r‘ = 3' 十 ic, x, y (40.2)

are given, i.e., that not only the values of the stresses at infinity, but also
that of the rotation (§ 36) and, besides, the resultant vector (X, Y) of
all external forces, applied to the boundary, are given. At first sight, the
last condition seems to be unneccessary, but it can be shown that without
it the problem remains indefinite, i.e., that it has an infinite number of
solutions.
Apart from the stated problems, the fundamental mixed
problem plays an important part, i.e., the problem for which dis¬
placements are given for one part and stresses for the remaining part of
the boundary. In the case of the mixed problem for infinite regions, it
will be assumed, as in Problem II, that, in addition, the values of X,
Y, T, r' are given. In Part VI several problems of a different type will
be considered.
It will now be proved that, if the above problems have solutions, these
will be unique. For finite regions, the proof is completely analogous to
that presented earlier for the general case of three dimensions, while
for infinite regions (such regions not having been considered for the
three-dimensional case) certain additional considerations are required.
CHAP. 5 COMPLEX REPRESENTATION 143

For the proof it will be assumed that the components of the stresses and
the displacements, corresponding to the solution under consideration, are
continuous up to the boundary L (cf. below).
First consider the case of finite regions (simply or multiply connected).
Study the integral (cf. § 20)

/
/ = (xnu + y 洲)公，
L
where
Xn = Xh cos (处, x) + Xy cos (将, y),
(40.3)
Yn = Yx cos {n, x) + Yy cos (% y)
denote the stress components, applied to the boundary L, and n is the
outward normal to L.
By Green's theorem

/
=/ [(%；说 + y初 COS (外, ％) + {Xvu 十 ym COS (外, 朔面 =

十 x年 十 X, 停十 书 V 缜} dxdy.
ox \ ex 砂 cy )

But, by (29.1),
打工I _i ax,上 o aya- -I ak o
a% oy
= ——d% dy — •
八

also
初 加 加 a%
— = exx>
ex 砂
= %，， Wd% H cy一 = 2e '

and
— 入6 ~ y — 入6 i 2“©^?, y • 2^i,^
Hence, the above expression becomes

/
L
(X渣 + y/磔 =
S
J (入 * 2M落 4 + 2M)}标 dy. (40.4)

If u, Xn) Yn, exx, 分打 exy represent the “difference" of two solutions

of the first, the second or the mixed problem, the expression + 丫建
will be zero on the boundary L (cf. § 20). Hence, the double integral on
144 II. PLANE THEORY OF ELASTICITY §40

the right-hand side will be zero. But since the function under the integral
is a positive definite quadratic form, this can only be so, if

Hence also Xx, Yy, Xyt arising from the difference of the two solutions,
must be zero, i.e., both solutions are identical in the sense that they lead
to the same stresses and strains. However, the displacements may differ
from each other by terms of the form
—
= ey + a, % = e% + g,
corresponding to rigid body displacement in the plane Oxy. In the cases
of the second and the mixed problems, this difference does not occur,
since the displacements of both solutions must be the same on the
whole or part of the boundary.
Next consider the case of infinite regions. As before, let it be assumed
that any of the three fundamental problems possesses two solutions and
that u, v, Xnt Yn denote the "difference" of these solutions. For Problem
I : Xn = = 0 on the boundary, i.e., the resultant vector of all forces,
exerted on the boundary, is zero. However, for the second and the
mixed problems, this vector had been assumed given for both solutions,
and hence it will also vanish for the difference of the two solutions.
Thus in all the cases considered: X = Y = 0. In addition, the quantities
r,「，corresponding to the difference, will be zero, since they were
to be the same for both solutions if it is assumed for Problem I that the
imaginary part of T is zero which can always be done, since it does not
affect the stresses.
Now apply (40.4) to the finite region, bounded by the contours Llf . . . ,
Lm and the circle LK with radius R and centre at 0, which contains
all the contours Llt ... , Lm. It will be proved that

/ (X
Lr
渣 + Ynv)ds (40.5)

tends to zero as R f 8. In fact, by (36.4), (36.5) and (36.7), where one

has to put
x= y=p= = 0,
one has for \z\> R
= 题 + —— ••.,
z +
?(^) = 他 H , W(z)
z 1- •••
CHAP. 5 COMPLEX REPRESENTATION 145

= b(z) =-4 + •… 野) = 少'(z) =—§+・•••

The formula
2[z(仇 + w) = K<p(z) — z<p'(z) — W(z)
shows that under these conditions u, e remain bounded. Further, the re¬
lations

% 十 % = 2[①⑶ + 丽], W — X, + 2汉y= 2[2'(z) + ¥(明

1
indicate that Xx, Yy> Xy tend to zero with order | 1 /z2 (at the least)
as | z | f oo. Hence, the expression Xnu + Ynv will be of order
1 位2 the circle LR. On the other hand, the path of integration in (40.5)
is of length 2-kR and hence the integral (40.5) is of order 1 /R and tends to
zero as 7? f 8.
Applying (40.4) first to the region, contained between L and LR,
and then increasing R beyond all bounds, it is seen that the integral on
the left-hand side will tend to the integral taken over the boundary L ;
hence, the integral on the right-hand side will likewise tend to a
limit which, by conventional definition, will represent the integral
taken over the infinite region S. Thus (40.4) applies to infinite regions
S and, hence, the conclusions regarding uniqueness of solutions remain
true also for these cases.
NOTE. It has been assumed in the proofs above that the displacement and stress
components are continuous up to the boundary. This assumption can be replaced
by one which is considerably more general. Only a brief remark will here be made
on this subject.
The proof presented obviously remains in force, if it is assumed that the dis¬
placement and stress components corresponding to the differences of the solutions
under consideration are continued continuously at all points of the boundary with
the exception of a finite number of points ck near which they must behave in such
a manner that the integrals (X您仇 + Ynv)ds taken over arcs 丫允 of infinitesimal
circles belonging to S and having the points ck as centers tend to zero together
with the radii of these circles.
In § 42, the uniqueness theorems for the first and second fundamental problems
will be proved under assumptions which are somewhat different from those of
this section.

Regarding the question of the existence of solutions, the following re¬

marks will be made for the present. From a mathematical point of view.
146 II. PLANE THEORY OF ELASTICITY §40

the first fundamental problem is completely equivalent, at least

for regions bounded by one contour (for the case of several contours
cf. Note 1 at end of the next section), to the problem o/ equilibrium
。/应 thin elastic plate, clamped at the "ges, under the influence of
loads normal to its plane. This latter problem can be reduced to the
determination of a biharmonic function U for given values of its partial
derivatives
au au
on the boundary of the region.
This problem is discussed in any of the treatises, mentioned in 、helist given at
the beginning of Part I. As a rule, the problem leads to the determination of U for
dU
—an
given boundary values of U and of the normal derivative — . But obviously one
may in this case immediately determine the boundary values of
dU 8U
1
dx dy
since
8U
dx
= ——
CU
as —dU
d%
8U
cos (力 ％) H- - cos (%,力)， — — =
纱 ds
cos Q, 夕) +
dU
d%
cos 例，夕)，

where s denotes the coordinate along the boundary and t the direction of the
tangent. Thus one arrives at the problem, stated in the text.
The first fundamental problem will be reduced to just such a mathe¬
matical problem (cf. § 41). The problem of finding a biharmonic function
au au .
ex—
for given values of the derivatives and on a contour will
oy
be called the fundamental biharmonic problem. This problem (or its equi¬
valent problem of the equilibrium of a plate, clamped along the edges)
has been the subject of many investigations, especially since 1907
when the Paris Academy of Science declared it the object of a prize.
This prize was obtained by J. Hadamard [1], G. Lauricella [3], A. Korn [4]
and T. Boggio. The above authors completely solved the problem for
the case of finite regions, bounded by a simple contour and satisfying several
conditions of a general character. (In 1936 S. L. Sobolev [1], using
variational methods, gave a proof of the existence of solution of a boun¬
dary problem which represented a considerable generalization of the
fundamental biharmonic problem) .
Use of functions of a complex variable provided recently the means of
CHAP. 5 COMPLEX REPRESENTATION 147

obtaining the solution of the first as well as of the second fundamental

problem for regions, bounded by an arbitrary number of contours. It
also solved the fundamental mixed problem and a number of other im¬
portant general problems. Certain of the above-mentioned general results
will be studied below in Part V, while short statements will be given of others.
It will just be noted here that in the case of finite regions the first
fundamental problem has, of course, a solution only when the resultant
vector and moment of the given external forces, applied to the boundary
L of the region, are zero.
But in the case of infinite regions a solution exists only when this
condition is not satisfied, even if it is required that the stresses at in¬
finity vanish. This is explained by the fact that, if one considers part of
the body enclosed between a given contour L and a circle, containing
this contour, then, although the external stresses acting on the circle tend to
zero as the radius increases beyond all bounds, taken over the whole
boundary they may give a finite resultant vector and moment, because
they are distributed over a circle the length of the circumference of
which increases beyond all bounds. The resultant vector and moment of
the external forces, applied to the union of the given contour L and the
circle, is always zero.
Regarding the above-mentioned general solutions of the fundamental
problems it may be noted that, just because of their generality, these
solutions are often unsatisfactory from the point of view of application.
Therefore one is obliged to study special methods of solution offering
the possibility of practical analysis of more or less wide classes of regions,
——
important in applications. Parts III VI of this book are largely devoted
to such methods.

§ 41. Reduction of the fundamental problems to problems of

complex function theory. 1°. Since the state of stress and the dis¬
placements can be expressed by means of the two complex functions
甲 (z) and W(z), the problems formulated in the last section lead to the
determination of these functions under certain conditions which they
must satisfy on the boundary of the region occupied by the body.
As in the preceding section, we will assume that the displacement and
stress components are continuous up to the boundary L of the region S・
The points of the boundary L will usually be denoted by 力 so that
—
t x + iyf where % and y are the coordinates of the boundary point
under consideration.
148 II. PLANE THEORY OF ELASTICITY § 41

However, often when this will not be convenient, the boundary points
will be denoted also by z just as all other points of the plane.
Let /(力) be some (real or complex) function of the point 力 of the bound¬
ary L. Since the position of the point 力 on each of the contours composing
L is completely determined by the arc s measured in a definite direction
along the given contour from some fixed point, /(/) represents on each
of these contours a function of the real variable s; therefore it will
often be convenient to write /(s) instead of /(/) without introducing a
new symbol for /. Further, integrals of the form

J
tot
/($)曲

taken along some arc % of a contour will likewise be denoted by

t

J
to
W) ds

and interpreted as before.

2°. For greater clarity, a beginning will be made with the case of a
region S which is finite and bounded by a single simple closed contour L.
In the case of Pyoblem the boundary condition can be expressed
in the following manner:
叼 (z) — —
卬(z) 肌z) = 2(x(g1 + ig? on L, (41.1)

—
where g】 g£s) and g2 g?(s) are the given displacements of the points
=
of the contour L. They are functions of the point t of the contour S or
of the corresponding arc S which are known and by the strength of the
above conditions are continuous. The positive direction on L can be
chosen in an arbitrary manner.
In this context, the statement (41.1) must, of course, be understood
within certain conditions. In fact, the left-hand side of this equality
represents the boundary value of the expression

跑 一 z</(z) — W(z)
as z, while remaining inside S, tends to the point t of the contour L ;
this boundary value exists, since this expression is equal to 2(z(弘 + 说)
and u and v are, by definition, continuous up to the contour.
In the case of Problem I, the boundary condition can be expressed by
two different methods which should be used according to their convenience.
CHAP. 5 COMPLEX REPRESENTATION 149

Only one of these will be stated now, while the othei will be given at
the end of this section. Let X，%), 丫“&), or in the new notation Xn(s),
y<s), be the given values of the components of the external loading at
the point t, measured in the positive direction from some fixed point %
As positive direction of L take that which leaves the region S on the left.
On the basis of (33.1) one has
+ %) = /i + /2 + const., (41.2)
where

/1 + 分2- /1(力) + 272(力) — /1 十 2丁2($) =

£ s

ta
J
= % (Xn + iY^ds i j(Xn + 工)
0
2 而. (41.3)

The expression on the left-hand side of (41.2) must be understood as

the boundary value of
+ 少⑵
as z f £ on L. This boundary value is easily seen to exist under the
assumed conditions regarding the continuity of the stress components
up to L. It should still be noted that (41.2) was written using (33.1)
which was deduced under the assumption that the arc AB in § 33 lies
completely inside S. However, it is readily seen that in the present case
the last formula is also applicable when the arc AB belongs to L ; this
follows from the same condition of continuity of the stress components
up to the boundary.
Thus the boundary condition for Problem I is expressed by (41.2) in
the sense stated above. The functions /式力) and /2(Z) are then the given
real functions along L, defined by .3).
Now the following will be noted. It has been seen in the preceding
section that knowledge of X/s), Yn(s) completely determines the state
of stress of the body. But the functions 中 (z), W(z) will then not be com¬
pletely determined ; in fact, it has been found in § 34 that the substitutions
+ C^ + y for 平 (z),
中 (z)

%(z) + y， for Mz),

where C is a real and 丫 = a + 峪，丫' = a' 十 也' are complex constants,
do not alter the state of stress and, conversely, that any transformation,
which does not affect the stresses, must be of the form (4). Then (§ 34)
150 II. PLANE THEORY OF ELASTICITY §41

—au.au
au au
3%
F% —砂— is then replaced by
3%
F z —cy 一
F 丫 十 丫'. (B)

It thus follows that by a suitable choice of 丫 and 丫' the constant in (41.5)
can be given any arbitrary value.
In the case of Problem II, the boundary values completely determine
the displacements at all points of the body (§ 40). Therefore, by the results
of § 34 (assuming the origin to lie within the region S), only one of the

<p(0) —0 or 少(0) = 0 ; 3中'(0) = 0. (41.5)

In the case of Problem I, when the boundary conditions completely

determine the state of stress of the body (the displacements being de¬
termined apart from rigid body translation) , both quantities p(0), 少 (0)
may be fixed arbitrarily, in addition to the imaginary part of 中'(0). But,
if the constant on the right-hand side of (41.2) is fixed in a definite manner,
only one of the two quantities <p(0), 山(0) can be decided upon arbitrarily.
Therefore, in the case of Problem I, one may assume
乎⑼ = 0, §平 '(0) = 0. (41.7)
Regarding the last point, the following remark may be made. If 9 and @ are any
functions solving *Problem I, application of the transformation (A) gives functions
solving the same problem. In order that (41.5) may be fulfilled for a definite value
of the constant on the right-hand side, the quantity 丫『 + will be fixed, as can be
seen from (B). For example, if 丫 be given, the constant 丫' will be completely
determined.

The supplementary conditions (41 .4), (41.5) completely fix the functions
9(2), (p(z), if in the case of Problem I the constant on the right hand side
of .2) is fixed.
The following may still be said with regard to Problem I. This problem
is known to have a solution only if the resultant vector and moment of
the external forces acting on the boundary of S vanish (assuming the
region to be finite) .
The condition that the resultant vector is to vanish leads to

L
J (X〃+ iY^ds - 0. (41.6)

On the basis of (41.3), this condition is equivalent to the condition that

the function卜 /2 given on L is to be continuous. In fact, if (41.6) is
CHAP. 5 COMPLEX REPRESENTATION 151

fulfilled, the function & + /2 of the point t of L reverts obviously to its

original value as t executes a complete circuit, and vice versa.
Next, the condition that the resultant moment is to vanish will be
considered. Taking the moment about the origin, one finds

J
L
- yX^ds = 0. (41.7)

On the basis of (41.3), one has

after an integration by parts,
Xn ds = df&, Yn ds — 一 dflt and hence,

J
L
他 一 yX^ds = 一
J
L
",+ 夕 ''2) =
「
= - + 对2)乙 + I (/1 加 + /2 内).
L

If the resultant vector of the external forces vanishes, the functions

/2 are continuous on L, whence follows
［端 +
Consequently, the vanishing o/ the resultant moment 。/ the external forces
when the resultant vector of the external forces is zero is expressed by the
formula

J
L
(/i dx + /2 dy) =0. (41.8)

3°. The case of 々％ infinite region S bounded by a simple closed contour L

(infinite plane with a hole) will be considered next. In this case the
boundary conditions for Problems II and I assume the form (41.1) and
(41.2) respectively. Also, now the function + if2 in (41.2) is given by
(41.3), where it must be noted that this time on L the positive direction
is clockwise.

However, there is an essential difference between this case and those

studied earlier. In fact, for a finite simply connected region, the functions
中 (z) and W(z) are holomorphic (and consequently single -valued) in the
entire region, while in the present case this is, in general, no longer true.
On the basis of the formulae of § 36, assuming for definiteness that the
origin of coordinates lies inside L (i.e., outside S), one has
 152 II. PLANE THEORY OF ELASTICITY § 41

平⑵ =
X + iY i°g z + rz 十 平。⑶,
-
-vvrn
2tt(1 十 x)r
(4L9)
=
乂 (X — iY)
l°g z+Tz + 必(z)，
O兀 (1 \
2 乂)
where 中°, 少。are holomorphic in S, including the point at infinity.
It will be recalled that X and Y denote the components of the re¬
sultant vector of the external forces applied to L and T, T' are constants
(which are, in general, complex) determined by the stress distribution
as well by the rotation at infinity.
The constants X and Y must be assumed known : For Problem II,
their values are given in § 40, for Problem I they must be calculated
from the external loading by the formula

X 十 zY =
J
L
(X〃+ H 然 (41.10)

Further, by § 40, one can assume as given, for Problem II, the constants
r, T' and, for Problem I, the constants 况{r}, T'; the imaginary part of T
does not affect the stress distribution.
Using (41.9), the problem under consideration may be reduced to the
determination of the functions 中°(z), %(z), holomorphic (and hence single¬
valued) in S. In fact, the boundary condition of Problem II, on the basis
of (41.1) and (41.9), assumes the form

乂 伙)(£) 一 须而 — O) = 2^(g；+ z鸿) on L> (41. H)

where

2k(g；+ 密) = 2Mgi + *2) +

X iY
~
t
2九(1+ X)

— 3P ~
一
P)' 十 P*
_
log 侬) -
.
(41」2)
o (1
27V /1 ।
+ x)T T
t
The right-hand side of (41.1 1) is single-valued and continuous on L,
since these properties are possessed by all terms on the right-hand side
of (41.12) and, in particular, by log 历 = 2 log | 力
For Problem I, one has analogously
CHAP. 5 COMPLEX REPRESENTATION 153

处 '(力) + 少(力) /； 嘴 const, on L, (41.13)

where
。 。 X + iY _
/? + 喟 = /1 + 殖 + 刁2 门
(1 工 x)V
Mg … … - X log 0 +
兀
+
— iY _
-
X t
+ (r + r‘) 力 r% (41.14)
2 (1
加
+w T-
t
These formulae show that the right hand side of (14.13) is a single-
valued and continuous function of 力 on 工. In fact, the formula (41.3)
shows that, as t describes a complete circuit of L in the positive direction,
the expression fr + % acquires an increase given by

i (X^iY^ds^HX + iY);
L

however, exactly the same increase with opposite sign is acquired for
such a circuit by
X+丫 z 八
o「八 —f (kg 力一乂 log %),
and hence /{ + reverts to its original value.
In addition, it is seen that the right hand side of (41.14) effectively
does not contain the imaginary part of the constant (because T + F =
= 2猊{巧，as was to be expected, since it does not affect the stress
distribution.
For Problem II, one can assume arbitrarily
中o(8) = 0 or Wo(oo) =0, (41.15)
because it is known from § 34 that one may add to one of the functions
#(z), 少 (z) an arbitrary constant without changing the displacements.
For Problem I, for an arbitrarily fixed constant on the right-hand
side of (41.13), one may assume (in the case of a finite region)
平 o(oo) = 0 or %(8) = 0, 3{r} = 0. (41.16)
The supplementary conditions (41.15), (41.16) fix completely the
unknown functions cp0(z), 以⑵，if for Problem I the constant on the
right hand side of (41.13) is fixed.
4°. Next, the general case will be considered when the boundary
consists of several contours J, L*, .. . , Lm, Lm+1 (finite region) or
154 II. PLANE THEORY OF ELASTICITY §41

L2, . . Lm (infinite region) and a beginning will be made with the

case of the finite region.
As in the preceding case, the unknown functions 中 (z), 少⑶ are, in
general, multi-valued. In fact, by (35.10) and (35.1 1),
| m
中(z) = - o2 (1 + x) X (X
兀八 上
「
%=i
电 + log (z - zj + 伙 (z),
(41.17)
m
、' (X
比
Mz) =々 八 一 log (Z - zj + 媳z)，
+
允
2兀 (1 工
乂) 论=1

where 中°(z), ^0(z) are functions, holomorphic (and hence single-valued)

in the region S and the zk are arbitrarily fixed points lying inside the
contours Lk 他 = 1, 2, . . . , 加). The resultant vector of the external
forces applied to the contour Lk has been denoted by (X^, YJ.
For Problem I, the quantities Xk, Yk are known beforehand, since
they may be evaluated directly from the given loading.
For Problem II, these quantities are not known beforehand and
must bs determined together with 平o(z), %(z). The boundary condition
of this problem is given by (41.1), where one must understand by L
the contour〔 二 + + ••• + 工僧 + 二加+1
In the present case, the boundary condition for Problem I can ob¬
viously be written in the form

4㈤ + /而 + 丽 = 弘 Ck on Lk,
(41.18)
々 = 1, 2, . . . , 加+ 1,

where the Ck are certain constants and

t

/i + 浜 = J
tk
{Xn + iY^ds on J; (41.19)

in the last formula, tk denotes an arbitrarily fixed point of Lk and the

positive direction of Lk is assumed to be that' for which S lies to the left.
The constants Ck are not known beforehand, but one of them, say
Cm+1, can be fixed arbitrarily, since the expression 中 (z) + zq/(z) +
is determined for given stresses exactly apart from an arbitrary constant
term (cf. 2°). The remaining constants Cx, C2, .…， are unknown and
subject to determination together with the functions <p0(z), %(z).
The boundary condition (41.18) can be replaced by the following,
CHAP. 5 COMPLEX REPRESENTATION 155

obtained by transferring to the right hand side the terms corresponding

to the logarithmic terms of (41.17)

?oW + 力妣 Q) + (力) = # + 鹤 + J on Lk,

.20)
1 , 2, 加 一 1,

where /? + is a function, given on the contours Lls L2, . .. , Lm+1,

which is single-valued and continuous on each of the contours. For an
explicit expression for this function, see § 98. It should be noted that
the continuity and single-valuedness of /J + 那 on the external contour
Lm+1, is a consequence of the condition (assumed fulfilled) that the
resultant vector of all external forces applied to the boundary L is
equal to zero; this statement is to be verified by the reader.
The unknown functions <p0(2), 少o(z) may be fixed completely with the
aid of supplementary conditions which are analogous to those in 2°.
In fact, assuming that the origin of coordinates lies in the region S,
one may assume for Problem II

的(0) = 0 or 巾o(O) = 0, (41.21)

and for Problem I, with Cm+1 fixed,
(po(O) =0 or % = 0, 3{?o(O)} = 0. (41.22)
The case of infinite regions can be studied in an analogous manner
as in the preceding two sections and therefore it will not be done here.
5°. The boundary conditions of the fundamental mixed problem can
be readily written in an analogous form; in fact, one will have conditions
of the form (41.1), wherever the displacements are given, and conditions
of the form (41.2), wherever stresses are specified.
6°. Finally, the boundary condition for Problem I will be stated in a dif¬
ferent form. Let there be given the normal and the tangential components
N and T of the external stresses acting on the boundary L. The com¬
ponents N and T will be the projections of the stresses on the outward
normal n and on the tangent, pointing to the left of %. Then
— — —
2(N 汉) = Xa 十 Yy (y, X, + 2EX") 产a on 工，
where a is the angle between the normal n and the axis Ox, measured from
the latter. In order to obtain this formula, one only has to think of the
normal n as axis O'%' and of the tangent as axis 0y. Then
N = X；,, T = X；,,
156 H. PLANE THEORY OF ELASTICITY § 41a

and the above formula agrees with (88). Introducing into this formula
the expressions (32.9) and (32.10), one finds

(z) 十函一 呼气5 —

+ 空⑵｝= N 打 on L. (41.23)
The boundary condition, written in this form which is mainly used by
G. V. Kolosov [1, 2], is often more convenient than the form stated
above, since the functions (z) and ¥(z) are single-valued also in the
case of multiply connected regions. However, in certain cases, the forms
of the boundary conditions stated earlier have a great advantage. One
of their principal advantages is that in such representation the boundary
condition of the first fundamental problem is very similar to that of
the second fundamental problem as a consequence of which the methods
of solution for both problems are very much alike.
In addition, in the case of finite simply connected regions, the boundary
condition (41.2) in the same form also applies to the problem of a plate
with clamped edges (fundamental biharmonic problem). In the case of
multiply connected regions, there exists a certain difference between
these problems about which more will be said in the next section.

§ 41a. Supplementary remarks.

1°. In the case of a multiply connected region S, the following
difference exists between the fundamental biharmonic problem and the

。
first fundamental problem of the plane theory of elasticity : for the
fundamental biharmonic problem, the expression
2
后十
J
一2
即 =/i+醇
is given completely on each of the contours Lk, for the first fundamental
problem it is given apart from constants Ck on Lk (these constants being
unknown beforehand) and only one of the Ck may be fixed arbitrarily.
Further, there is a difference in the conditions imposed on the un¬
known function y): for the fundamental biharmonic problem, it is
usually required that the derivatives

3% ' dy
be single-valued in S, or even that U(%,y) be single-valued there
(e. g. when dealing with the equilibrium of plates clamped at the edges) ;
for the first fundamental problem, it is required from U(%,y) that the
CHAP. 5 COMPLEX REPRESENTATION 157

components of stress and displacement, corresponding to it, be single¬

valued. In the case of simply connected regions, both conditions lead to
single-valuedness of U(%,y) .
In the case of multiply connected regions, the two problems under
consideration also differ by the conditions which are usually imposed
on the behaviour of their solutions near the point at infinity.
2°. In spite of the fact that solution of the boundary value problems
in the general case offers very great practical difficulties, it is very
easy in certain particular cases to guess the solution from the form of
the boundary condition. For example, assume that the boundary of
the body* is subject to uniformly distributed normal tension P (for
P < 0 one would have compression). Let n be the outward normal.
Then

Xn+iYn = 尸[cos(% X ) +2 COS(w, y)]=~Pi

「d”
—+ i
dy ~| dt

whence, on the basis of (41.19), one has on each contour constituting

the boundary 归 i% = Pt + const. Therefore the boundary condition
(41.18) assumes the form

+ 抑'(力) + W(c) = Pt + j on l 改 = 1, 2, …，加十 1,

and it is seen directly that this condition can be satisfied by writing

+
W(Z) = Pz, =0， = ~
= 0；
on the basis of the uniqueness theorem all other solutions may differ
from this solution only by a rigid body motion.
The corresponding stress components are now determined by the
formulae of § 32 : Xx = Yy = P, Xy = 0.
Another curious case will still be noted when the boundary problem
can be solved directly almost without any calculations. Consider first
Problem I for the same body as above and assume that the function
/i— of the point t on L, obtained by going on both sides of (41.19)
to the conjugate values, coincides, apart from constant terms, on every¬
one of the contours Lk with the boundary values of some function
F(z), holomorphic in S. Then the boundary condition (41 .18), written in
conjugate form, becomes

+ 如'(£) + W(£) = + Ck on Lk,

* In general multiply connected.
158 IL PLANE THEORY OF ELASTICITY § 42

。
and one obtains obviously the solution of the problem by writing
乎 (z) — °， W(z) — F(z), C\ = 2= ♦…
= C»n+i =

It follows from the uniqueness theorem that the problem has no other
solutions, except for those differing by rigid body displacements. Quite
an analogous reasoning may be applied to Problem II, and the gener¬
alization to multiply connected regions does not present any difficulty.
Consider, as the simplest example, an arbitrary (simply or multiply
connected) body and suppose that F(z) = Qz, where 0 is a real constant.
This corresponds to the case when

Xn-iYn = Qi^= 0•[寰 + i^],

i.e.,
Xn = — Q cos (应 4)， yw = o cos (%, y).
Thus, uniformly distributed stresses, equal to Q, are applied to the boun¬
dary of the body; however, these stresses are not directed along the
outward normal, but in the direction of the reflection of the normal in
the axis Oy. In the present case, one has
cp(^) = 0, 小⑶ = 0z.
For the components of stress one finds from the formulae of § 32
Xx = Q, Yy = Q, Xy — 0.
If, for example, the considered region is a rectangle with sides parallel to
the coordinate axes, the above is the solution of the problem for the case,
when uniformly distributed tensile forces act on the sides, parallel to the
axis 0%, and similar compressive forces act on the sides parallel to Oy.

§ 42. Concept of the regular solution. Uniqueness of a regular

solution.
1°. In § 40, it has been assumed for the formulation of the funda¬
mental boundary value problems and for the proof of the uniqueness of
the solutions that the displacement components u, & and the stress
components Xx> 丫勿 Xy are continuous up to the boundary L of the
region S. The same assumptions were made in § 41.
CHAP. 5 COMPLEX REPRESENTATION 159

The condition of continuity up to the boundary imposed on the com¬

ponents of the displacements is equivalent to the condition of continuity
up to the boundary of the expression

+ 说) — 乂中 (z) — z<p'(z) — 少⑶* (42.1)

From the condition of continuity up to the boundary for the com¬
ponents of the stresses follows the continuity up to the boundary of
the expression
eu . au —
a% +
_ —
平⑶ + 卸'(z) + —— (42.2)

but not the converse ; in fact, the expression (42.2) may obviously be
continuous up to the boundary without this property also belonging to
the stress components. The formulation of the boundary problems in the
form in which this was done in § 41, 6° required only continuity of the
expressions (41.1) and (41.2) up to the boundary without enforcing the
continuity of the stress components.
Therefore it will be quite natural to replace the requirement of con¬
tinuity of the stress components by the less restrictive condition of the
continuity up to the boundary of the expression (42.2), in other words,
by the requirement that the expression (42.2) is to be continued con¬
tinuously at all points of the boundary L (§ 29). Such a formulation of
the problem is also more natural from the point of view of mechanics.
However, for the application of the methods of effective solution of
boundary problems to be used later, it will be expedient for a significant
simplification of the reasoning to impose on the unknown functions the
following more restrictive condition : the functions 中 (z), cp'(z) and 巾⑵
are continued continuously at all points of the boundary L 。/ the region S.
A solution which has this property will be called regular.
If a solution is regular only in the stated sense, the expressions (42.1)
and (42.2) are obviously continued continuously on L. Generally speaking,
the converse is not true: from the fact that (42. 1 ) and (42.2) are con¬
tinued continuously on L follows obviously the same property for the
function 中(z) and the sum 的'(z) + 山⑵ or, what is the same thing,
the sum 狗'⑵ W(z), but not for the functions 中'(z), %(z) separately.
It should be noted here that the usual condition of regularity of a
solution is much less restrictive than the condition of continuity up to
the boundary imposed on the stress components ; however, it is not a
consequence of the last, since it does not follow from the regularity of
160 口. PLANE THEORY OF ELASTICITY §42

the solution that the stresses are continuous up to the boundary.

the sequel {unless stated otherwise) it will be assumed that the solutions
under consideration are regular.
2°. In § 40, the theorems regarding the uniqueness of the solutions
of the fundamental problems have been proved under the assumption
that the components of the displacements and the stresses are continuous
up to the boundary. It is easy to prove the uniqueness theorems for the
first and second fundamental problems under the assumption that the
solutions under consideration are regular.
The proof to be given here was first presented in the Author's paper [11]. S.G.
Mikhlin [6] gave a proof of the uniqueness theorems for the basic biharmonic
problem for somewhat moge general conditions. The uniqueness theorem for the
mixed problem, was proved for analogous conditions by G. F. Mandjavidze [2].

It will be assumed for the proof that the region S under consideration
is finite, since the extension of the proof to infinite regions presents no
difficulties.
A beginning will be made with the first fundamental problem. Let
乎 (z) and W(z) denote the differences of the functions corresponding to
two assumed solutions. These functions will be holomorphic in the
entire region S, since the logarithmic terms in the formulae (41.17)
will cancel. The boundary conditions for these functions may be written
in the form
au
F2%
… e. au —=
+ foy— = 勺(f) + 3(£) + W) —— .
on Lk,
(42.3)
k — 1,2, + 1,

where, as always, 力 = % + Zy is a point of

the boundary and the Ck are
certain (initially unknown) constants. The quantities

石p 谈， 松， 她

—
7P
denote here the corresponding boundary values.
Consider now the integral

L
鲁 卷)"+ G 鲁 (424)

taken in the positive direction over the entire boundary of the region,
CHAP. 5 COMPLEX REPRESENTATION 161

where the symbols P, Q denote the real and imaginary parts of the
function 4中'(z), so that 4cp'(z) = P 十 E0; it will be recalled that P = AU
(cf. § 30, 1°). However, by (42.3),
HU 2U
—
Fox = a 九, —一 = Pt on 上方，
dy
丁

where afc) 瓦 are certain (real) constants. Therefore

1 广 m+1 广
/= e
左= 1
a JJ (0 获 + P w) 一 £ 员 (严放 一
J
k= 1
Ltc
0 的).
Ltc
Now
z z z

中 (z)
J
= (P + iQ)dz + const. = (P dx
Zo 00
— Q dy) i
Nd
dx + P 十

const.,
十

where the integrals are taken along any path in S joining the arbitrarily
fixed point zQ and the variable point z. Since the function 中 (z) is holo¬
morphic (and, hence, single-valued) in S, it is now easily seen that

Lk
j (尸 dx — 0 dy) =eJ (0 °% + P 为) = 。，
肥
左 = 1 , 2, . . . , 冽+ 1 ,

and hence / 0. —
On the other hand, transforming the line integral in (42.4) into a
double integral using the Green formula, one obtains

/= JJ
$
(A 17) 2 dx dy. (42.5)

Consequently, since / = 0, one finds P = AU = 0, and hence that

中'(z) — Ci, where C is a real constant. Thus 平 (z) = Ciz 丫， where 丫
is a constant. It follows then from (42.2) that

i
—— = 3团 + V.
Thus, the function 少⑶，holomorphic in S, assumes on the contours Lk
constant values which, on the basis of § 36, 4° is only possible if
W(z) = const, throughout S. Hence <p(z) = Ciz 丫， 少⑵ = 丫'， where
丫' is some constant. This proves that the difference of two assumed
162 II. PLANE THEORY OF ELASTICITY § 43

solutions leads only to a rigid displacement of the entire body, as was

to be shown.
The uniqueness of the regular solution of Problem II is proved in an
analogous manner with the aid of a study of the integral

J
/* = {(0〃 — Pv)dx + (Pu + qv)dy}, (42.6)

which is obtained by writing on the right hand side of (42.4) 见 》instead of

dx 砂，
of course, one is talking here again of the difference of two possible
solutions.
Transferring this integral into a double integral and using (30.8), it is
readily shown that

J JRP2 + (入 + 22)02] 的. (42.7)

However, by definition, % = o = 0 on L. Consequently, on the basis

of (42.6), /* = 0 and, by (42.7), P = 0 = 0 in S, whence follows that
平 (z)
— const. = 丫. The formula (42.1) then shows that
= — 少⑵ 十 xy. However, since
— 谢) =
v = 0 on L, the boundary value
—
of the function — 山(z) + xy is equal to zero on L. Thus, 中卜) + 乂〒 = 0
throughout S, and hence % = » = 0 in S.
Thus the uniqueness theorem can obviously be extended to more
general cases similar to those mentioned in § 40, 3°.

In a recently published paper R. Tiffen [1] gives certain (almost obvious)

generalizations of the uniqueness theorem presented here using the same method,
but a somewhat different notation, in that he uses the representation of A. C,
Stevenson [1].

§ 43. On concentrated forces applied to the boundary. In the

preceding sections, the solutions of the equations of the plane theory of
elasticity have been obtained under different conditions which ensured,
in particular, the continuity of the expression
su au — —— (43.1)
2%
CHAP. 5 COMPLEX REPRESENTATION 163

up to the boundary L of a region. It must not be imagined that the

requirement of the continuity of the expression (43.1) up to the boundary
belongs to the set of conditions which bear a purely mathematical
character, usually imposed in order to simplify the reasoning. This
condition has an essentially mechanical significance and can be ex¬
pressed as follows : the resultant vector of the forces applied from a
definite side to a given, arbitrarily placed arc tends to zero together
with the length of this arc. Next, consideration will be given to one of
the simplest cases when this condition is violated.
Let AB be some arc belonging to the boundary £ of a region and
let (43.1) be continued continuously at all points of the arc AB except
only at one point C. It is known from § 35, 3° that under those con¬
ditions the expression (43.1) will be continuous on the arc AB except,
may be, at the point C. It will be assumed for the sake ot simplicity
that the point C is a point of first order discontinuity of the boundary
value of the expression (43. 1 ) and that the function U(%, y) is continued
continuously at all points of AB including C. This condition for U(%，夕)
will certainly be fulfilled if (43.1), and hence also
au , au remain
3% dy
bounded in S near C. The last condition is completely natural from the
point of view of mechanics : it implies that the resultant vector of the
forces applied to an arc of finite length remains bounded, even when
the arc is near the boundary of the region.
Let

denote the jump in

du
i
w for a passage through C, if z
.
describes
ex
1- dy
AB in the positive direction (i.e., leaving S on the left).
Consider an infinitesimal part CrDCr, of the body and the resultant
vector of the external forces, acting on the arc C'QC" (Fig. 18) of the
boundary of this part. By (33.1), this resultant vector (X, Y) may be
written

Letting (7 and C" approach C, one obtains in the limit

164 II. PLANE THEORY OF ELASTICITY §44

(43.2)

The resultant moment (with respect to the origin) of the same forces is
easily calculated by (33.2), its limit being

where %, y are the coordinates of C. Hence, the forces applied to the

infinitesimal arc C'DC" are equivalent to one
single force (X, Y) applied at C, its compo¬
nents being given by (43.2).
Thus, the point of discontinuity C of the

expression
——
dx
z 一 on the contour (poss-

essing the properties stated above) should

be considered as the point of application of
the concentrated force (X, Y), defined by
(43.1).

§ 44. Dependence of the state of stress on the elastic constants.

An important property of the solution of the first fundamental problem
will now be discussed. First consider the case 。/ finite simply connected
regions. The unknown functions <p, 少 are then holomorphic in the region S.
Further, since the boundary condition (41.2) does not depend on the
elastic constants 入 and p., the functions 乎 and 少，giving the solution of
the first fundamental problem, will also solve this problem (for the
same given external stresses) for a body of the same shape, but made of
some other (homogeneous and isotropic) material.
Thus, for given external stresses, the state of stress o/ a simply connected
(finite) body depends only on its shape, but not (m its material. The dis¬
placements and strains will, of course, depend on the material, since
the constants 入 and q enter into the formulae, giving the displacements
in terms of the functions 勺 and
“
CHAP. 5 COMPLEX REPRESENTATION 165

In the case of the exact problem of plane strain this proposition, of course, only
holds with respect to the components Xx, 丫山 Xy, because Zz depends on 入 and p.
(or more correctly, on relations involving these quantities) . But in the case of thin
plates, ie, for "generalized plane stress" (§ 26), the proposition holds fully, because
then Z,
— 0.
The theorem on the independence of the state of stress on the elastic constants
(always with reference to the components Xx, Yy, Xy} is with little justification
called the theorem of M. Levy, for example by G. V. Kolosov [3, 4]. The truth is
that M. Levy [1] emphasizes the fact that the equations, to be satisfied by Xx,
Yy, Xy, do not involve the elastic constants. But it does not follow from this fact
in the general case that the stressed state does not depend on the elastic constants
(cf. later).

Next consider the case of multiply connected bodies. Also in this case
the constants 入 and q do not figure in the boundary conditions. But
they do appear (through x) in (35.10) and (35.1 1), viz.,

中⑵ = 一水
E谓产 + 以）log（一 %）十 洲哈
(44.1)
v m
"z）.
事=
而中^产一幅现 Gf 十

Assume that the first fundamental problem has been solved for a given
material, i.e., that the corresponding functions 乎，少 have been found.
Consider whether the same functions may give the solution of the same
problem for the same boundary stresses and for a body of the same shape,
but of different material with the constants 入'， instead of 入 and 内
Denote by 比' the corresponding value of 乂. The functions 乎，少 will, of
course, satisfy the given boundary conditions also for the second body,
because the elastic constants do not figure in these conditions. However,
the displacements, corresponding to these functions, may turn out
to be multi-valued. In fact, for single-valuedness of the displacements,
one has by (35.7), in which one has now to replace x by
乂 'y加 + = 0，
where, by (35.9),

YL — — 2Ml
X 充 + iYJC
十 乂)
， 注 一 _ x(Xfc
2ml
+ iYQ
十 x)
and hence
+ iYk
(x
— k')
2兀(1 + x)
166 IL PLANE THEORY OF ELASTICITY §44

—
But this will only be possible for 乂' 力 x, if Xk = Yk 0. Thus, the
same functions 中 and 少 will give the solution for bodies of different
materials (with different constants x) if, and only if, the resultant vectors
of the external forces applied 力。each of the contours Lk separately are zero;
then, and only then, the state of stress does not depend on the elastic
constants. Otherwise it depends on the value of x, i.e., on the value of
入
This result is due to J. H. Michell [1]. It is of considerable interest
for experiments involving models made of various materials which are
convenient for the purpose. It is seen that under the given conditions
the material does not affect the results. G. V. Kolosov [3, 4] gave formulae
elucidating the influence of the elastic constants also in the case, when
body forces are present the components of which are analytic functions
of the coordinates. However, the results of Kolosov require additional
study in the case of multiply connected regions
A more detailed statement of a practical nature with regard to the
influence of the choice of material constants of multiply connected bodies
can be found in the paper by L. N. G. Filon [3] and also in the book by
E. G. Coker and L. N. G. Filon [1]. It should be noted that the deduction
of all the results, obtained by Filon, can be considerably simplified, if
one starts from the above formulae.
 Chapter 6

MULTI-VALUED DISPLACEMENTS. THERMAL STRESSES

§ 45. Multi-valued displacements. Dislocations. The condition

of single-valuedness of the displacements, which hitherto has always been
assumed to be fulfilled, seems at first sight to be quite inevitable from a
physical point of view. However, it will be seen that a very simple physical
interpretation can be given to multi-valued displacements.
As before, it will be assumed that the components of stress, and hence
the components of strain, are single-valued functions in the region, oc¬
cupied by the body; more exactly, it will be assumed that all the conditions
stated at the end of § 29 are fulfilled with the exception of the condition o/
single-valuedness of the displacements.
It will be remembered that in the case of simply connected regions single-
valuedness of the displacement components remains the necessary con¬
sequence of the other conditions (cf. §§ 29, 30). Therefore only multiply
connected regions need be considered. As in § 35, suppose that the region
S, occupied by the body, is bounded by several simple contours Llt
L2, . . . , Lm) Lm+1 the last of which contains the others.
It will also be remembered that the deduction of the formulae (35.1)
(35.6) was not based on the condition of single-valuedness of the displace¬
—
ments ; this condition was only introduced starting with (35.7). Therefore,
in particular, (35.3) and (35.4) remain valid under the conditions to be
considered now.
In order to study the character of multi-valuedness of the components
of displacement, convert the region S into a simply connected one by
means of m cuts 旬&, , ambm, connecting Llt L2, . . Lm with the
outer contour Lm+1 and not intersecting each other (Fig. 19). (These
cuts may be produced in any manner whatsoever, e.g. by joining some
point of Lr with some point of some point of Z2 with some point of Lz
etc. and by reaching in this manner some point of Lm+1 ; but for the sake
of simplicity the above stated system of cuts has been adopted.)
In the region cut in the above manner the functions 勺, 少，and hence
also the displacements, will be single-valued. At each cut a distinction

167
168 II. PLANE THEORY OF ELASTICITY §45

must be made between the two

sides which will be denoted by
—
+ and ; these signs will be
allotted in such a way that, in
order to go from some point
—
(%,y) on the side ( ) of the
cut 口也 (and remaining in the
cut region) to the corresponding
point of the side (十) of the
same cut (i.e., to the point with
the same coordinates), one has
Fig. 19
to encircle the contour Lk in
an anti-clockwise direction. By (35.6), one has for such a circuit

+ 一 说一
湿
+ *»+ — 厂) = —{(x+ 1)月式％ 十切 +X丫h一 k}, (45.1)

where 幺丸 is the real, 丫兀 口而 i^k and 丫" a] + 温 are the complex

constants, appearing in (35.3) and (35.4) ; here 每+, »+ and 厂 are the

values of the components of displacement, corresponding to the points

on the sides (+) and (—) which coincide in the geometrical point (%, y).
The formula (45. 1 ) may be rewritten

where
四 +
—w = ~• g
'
d + 成， — 0
一 = e春 位， (45.2)

7T(X 十 1)4左 兀(一 比阮 氏) 也竺t吼 (45.3)

2
There is no difficulty in giving a physical interpretation of these
multi-valued displacements. (It will be remembered that only very small
deformations of the body are being considered ; hence also the quantities
年，琮，位 will be very small.) In fact, in order to explain those displace¬
ments, it is sufficient to suppose that along each cut akbk the two sides of
the body have been connected by removing from it, before deformation,
a (very narrow) strip the sides 或丛 and (Fig. 19) of which are con¬
gruent and placed in such a way that 磋立 results from by a rigid dis¬
placement, consisting of a rotation by an angle q about the origin and
a translation with components 城，麻 It has been implied here that for
a reunion the same points are to be combined which would correspond
CHAP. 6 MULTI-VALUED DISPLACEMENTS. THERMAL STRESSES. 169

to each other, but for the above rigid displacement. The notation has been
chosen in such a way that the lines 磷丛 and & will, after deformation,
—
become the sides ( ) and (+) of the cut a*.
The relations (45.2) have been obtained, in order to elucidate how the above
operation of reunion could be accomplished ; for example, let remain fixed
and let the side 以初£ move as a rigid unit until it meets 确磅. Then ur — ~ 0,
勿+ =
— e秒 + 琮, £/ 十 能 and hence (45.2) is fulfilled. If after this process
the body is left to its own devices and becomes, in addition, subject to some or¬
dinary deformation, the relations (45.2) will not be disturbed, because adjoining
points of the contacting sides will move like one point and no additional differences
between (〃+, »+) and (u~t l) will arise. Clearly the shape of the line akbk in the final
state -will, in general, differ from that of and

For simplicity, the above discussion only refers to the removal

of strips with sides 或& and a^. But for some values of 年，说，傥 it may
happen that (before deformation) the sides 航瓦 and 或% will overlap,
so that virtually strips have to be added rather than removed. Similarly,
it may also happen that 或立 and only partly overlap, in which case
material may have to be added in one place and removed in another.
However, in the sequel, for the sake of brevity, only "removal” of strips
will be mentioned. Likewise it is clear that, when joining the sides“立
and 媒&, their end points may not completely coincide with each other so
that after reunion they may form (small) steps on the boundaries of the
region ; but these will not be considered here.
The above interpretation of multi-valued displacements was first stated
by A. Timpe [1] for the particular case of a circular ring. (This case will be
treated as an example in § 60.) Somewhat later, V. Volterra obtained more
general results referring to multiply connected bodies of arbitrary shape.
Cf. V. Volterra [1] which contains a summary of his results, and also
his books [2] and [3]. The case of plane deformation has also been considered
in a paper by L. N. G. Filon [3] which presents interesting results referring
to the problem of the study of a state of stress by means of experiments
with models of different materials; cf. also. E. G. Coker and N. G. Filon [1].
Volterra uses for deformations of the body of the type described above
the term "distorsion". A. E. H. Love [1] proposed instead the term "dis-
location” which will be used here.
Note the following important property of dislocations, stated by
V. Volterra. If the cuts 况丸源 are to shift and change their shape, but
in such a way that the points ak and bk remain on the contours Lk and Lm+1
respectively and that the cuts never intersect each other, the quantities
170 II. PLANE THEORY OF ELASTICITY §46

ek, 城，优，determined by (45.3), obviously remain unchanged. In other

words, these quantities do not alter from one system of cuts to another,
as long as the latter are topologically equivalent.
It has been seen that, under the requirement of single-valuedness of
the displacements, the stresses inside a body are completely determined
by the external loads. This requirement is equivalent to the conditions
=尤= 优 = 0 他 = 1 , …，加).
It is easily shown that the stresses will likewise be fully determined by
given external loads and arbitrarily prescribed (small) quantities
琮，储；in fact, the "difference” between two solutions (if there exist two
of them) obviously gives a solution for which there are no external
stresses and for which
w 及 = 4 = 阮 = 0,
i.e., for which the displacements are single-valued. But under these
conditions the stresses are known to be zero everywhere. The quantities
q, S3 of which there are 3 m, will be called characteristics o/ dis¬
locations (they are the "'caracteristiques de la distorsion?, of V. Volterra).
NOTE 1. There arises the question : Why is there no possibility of
dislocations in a simply connected body? For example, a sector may be
cut from a circular disc in order to bring into contact and join free edges ;
thus, of course, stresses will arise in the disc and apparently the same case
will occur as for multiply connected bodies. But the difference here is
that in this case the stresses will not satisfy the conditions of continuity,
stated in § 29, because it has been seen that for a simply connected body
the displacements cannot be multi-valued, provided the conditions of
continuity are fulfilled.
2. A similar answer must be given to the question as to why one had
to restrict consideration to dislocations, caused by removal (or addition)
of strips with congruent sides and joined in a definite manner.

§ 46. Thermal stresses. There is a remarkable relation between the

dislocations considered above and the stresses caused in a body by a
non-uniform temperature distribution ; this will now be explained. But
first it is necessary to become acquainted with the law expressing the
effect of non-uniform temperatures in an elastic body. The equations of
the theory of elasticity, hitherto used, refer to the case when the temper¬
ature is the same throughout the body. On the basis of a law, enunciated
CHAP. 6 MULTI-VALUED DISPLACEMENTS. THERMAL STRESSES. 171

by J. M. C. Duhamel and F. Neumann (cf. A. E. H. Love [1], Chap. Ill),

the following relations hold in the case of non-uniform heating between
the components of strain and stress :
3% 加
X, = — 订十 入6 十2~丁， y, =
ox — vr + X6 + 2[i-^,
ay

Z = — 丫丁 十 入6 + 2[i— , (46.1)
cz

here T denotes the temperature at a given point, taking as "zero” of the

temperature scale the temperature of the body in its "natural" state ;
v is some positive constant depending on a property of the material of
the body. (This law is only strictly applicable for not too large temper¬
ature variations, because the coefficients 入，内 v change with the
temperature and these changes cannot otherwise be disregarded.)
The equations (46.1) replace, for the present, the generalized Hooke's
law and they only differ from those, expressing the latter, by the terms
— vT on the right-hand sides of the first three formulae.
The components of stress must, of course, satisfy the same equations

—
(18.1), since in their deduction no assumption regarding the temperature
distribution had to be made.
Consider now the case of plane strain of a cylindrical body, studied in
§25 侬 = 0, u,o independent of z) , and assume that T does not depend
on the coordinate z. Likewise let there be no body forces present. Then
元 = 乙 = 0,

j- 0, 尸 十 = o, (46.2)
3“ oy

—dv
沏
Xa；= — vT + 入® = — vT + 入6 + —,
<

"
I
八

oy
(46.3)

偿+给
彻 加
， e~ —H oy
ex
and
Z9 — 入 e — vT, (46.4)
172 II. PLANE THEORY OF ELASTICITY §46

or, noting that by (46.3)

x * y -—, 汀
x, 十兀 = —2vT + 2( 入十出)9, 6^-^^
2( 入 + N + 入 + Ji

2 = 名

M
—
T + 97. . r (Xw + Y)
入 十 2( 入
+ “) — ।(464)

Let it now be assumed that one is dealing with a steady state o/ heat flow,
so that the temperature T depends only on x, y and not on the time. Then
it is known that
AT = 0, (46.5)
i.e., T is a harmonic function of % and y. Denote by F(z) the function of
—
the complex variable z % iy (there being no danger of a confusion
with the coordinate z), having as a real part T{x, y), and put

比 *(%, y) + 初*(%, y) — F dz. (46*6)

Obviously,
加* 加* 加*
(46.7)
3% dy 3%
Further, let、
w*
% = %
2( 入
V v + 2（入 十 r）
(46.8)

where 力，vf are two new functions. Substituting from (46.8) into (46.3)
and using (46.7), it is easily verified that
加' 加' 。 加'\
Xx = 3%
匕=入6'+ 2俨-厂，
oy
X? = R + 砂 /
(46.9)

where
e，= -7初
—
加
h —~•.
八

加 效
Thus it is seen that the functions Xx) Yy, Xv, %', M satisfy the well known
equations of the 恒awe theory of elasticity, as if the body were uniformly
heated (in fact, as if T = 0), where vr play now the parts of the dis¬
placements. (This property was stated in the Author's paper [1] and in a
somewhat changed form in his paper a short study of the results
has likewise been given in a section of his paper [3]. A long time after¬
wards, H. Poritsky [1] published similar results).
CHAP. 6 MULTI-VALUED DISPLACEMENTS. THERMAL STRESSES. 173

Thus, the problem 0/ the study of stresses in a given cylindrical body,

caused by a steady flow of heat, in the case of plane strain is reduced to the
ordinary problem (i.e., for T = 0) for a body of the same shape with the
same external stresses acting on its sides. This latter problem (concerned
with Xx, Yv, Xy, %', »') will be called the auxiliary problem. The fact
that the stresses X*, Yy, Xy are the same in the original and in the auxiliary
problems is very remarkable.
First, consider the case of a simply connected body. Suppose that no
external forces are acting on its side surfaces. Then the auxiliary problem
is known to have only the following solution (omitting rigid body dis¬
placements) :
XX = Yy = Xy = 0, 仅，= M = 0.
Thus, in a simply connected cylinder, steady heat flow (which depends only
on % and y) does not cause stresses Xxt Yy, Xv. The displacements will be
given by the formulae, obtained from (46.8),

% — 〃* v
v=—
VD*
(46.10)
'
「，
2( 入 十凶 2(入十⑻
where %*, are determined by (46.6), using the temperature T (x} y) .
It must not be imagined that there are no stresses whatsoever present.
In fact, the component Zz will, in general, be different from zero and be
given by (464) (where one has now to put X* = 丫财 = 0) :

Zs = --~^T(x, y). (46.11)

入 十 w
It is seen that this direct stress Zz must be applied to the faces of the
cylinder as a necessary condition for the maintenance of plane strain.
If it is desired to find the solution when the faces are free from stress,
one may, in the case of a long cylinder, resort to the following method
(cf. § 25). The stresses applied, say, to the "upper" face, which are given
by (46.11), are statically equivalent to a force, directed parallel to the
generators of the cylinder, and a couple, the moment of which is parallel
to the face; in an application, one may, for example, place the force at
the centroid of the face. Similarly, the stresses acting on the "lower”
end may be replaced by a force and couple, opposite to the former. Next,
superimpose on the solution obtained above that of the problem of a
cylinder, subject to tension and bending by forces and couples opposite
to those above. (It will be shown in Part VII that the solution of this
174 II. PLANE THEORY OF ELASTICITY §46

problem for any (long) cylinder may be obtained by quite elementary

means.) This will give an (approximate) solution of the problem. Actually,
the stresses applied to the faces will now be statically equivalent to
zero. Thus, by Saint-Venanfs Principle (§ 23), they may be assumed to be,
in general, non-existent (provided the dimensions of the faces are small
compared with the length of the cylinder). Only near the ends, the solution
will differ appreciably from the exact one. It may still be added that, as
will be seen in Part VII, the components Xx> Yy, Xy will be zero for the
above-mentioned problem of tension and flexure of a cylinder. Thus,
one will have in the final solution, as before, = 丫的 = = 0 and
only Zz 力 0.
When the dimensions of the ends are not small compared with the
height of the cylinder, one has to look for a more exact solution which
does not only take account of the resultant forces and moments, applied
to the ends, but also of the actual stress distribution there.
Now the case of multiply connected regions of the type studied in the
preceding section will be considered. In this case the function F(z)
the real part of which is the (single-valued) function T(%,y) may be multi¬
valued. In fact, reasoning in the same way as for the function (z) in
§ 35, it is seen that
m
F(z) = E Bk log (z ― +a holomorphic function, (46.12)
k =l

where Bk (k = 1, . . . , 切) are real constants and 〃

are arbitrary fixed
points inside the contours Lk. Further [cf. the deduction of (35.3)]

/
m m
F dz = z 2 凡 log (z
k =1
— zj + Z (a* +
无 =1
log (z — zj 十
holomorphic function, (46.13)
十a
where 磷，位 are certain real constants. (The constants Bk, 琮，健 will be
known, if the temperature T(%, y) is given at each point.) For an (anti¬
clockwise) circuit of a contour, surrounding Lk) this expression undergoes
an increase (cf. the notation of § 45)
%*+
— %*— + ®*+
—
o*-) = 2izi[zBk + a* + 斗％). (46.14)
Let it be assumed that the body under consideration is not subject to
dislocations, i.e., that the displacements 。/ the original problem (每，。) 矽e
single-valued. Then, by (46.8),
0 = (w'+ —- %'一) + — vr-) + { (w*+ — 四*一) We* — 0*-)}———
2( 入 U)
CHAP. 6 MULTI-VALUED DISPLACEMENTS. THERMAL STRESSES 175

and hence, using (46.14),

(#+ — #-) +"M+ — M-) = — —入 +―
R
(&z +a* 十 暇). (46.15)
- 2

This formula proves that the displacements %', vf 。/ the ^auxiliary pro-
bZe加" are the same as if the body {which at uniform temper ature) were
subject to dislocations with the characteristics [cf. (45.2)]

E比 = —-+—N B
入
(46.16)
7TV 7W
器, 能 = 一
辗

入十科
Thus, the auxiliary problem is reduced to the determination of the
elastic equilibrium for a uniform temperature (T ~ 0) and for given
characteristics of dislocations.
〃 there are no external stresses acting o% the side surface, the stresses
Xx, Yyt Xy (in the auxiliary as well as in the original problem) are the
same, 霖$ if the body were subject to given dislocations in the absence of external
loading and for uniform temperature.
If the sides of the cylinder are loaded in an arbitrary manner, the
solution of the ordinary problem of the plane theory of elasticity for
given external stresses applied to the boundary must be superimposed.
As regards stresses applied to the ends, all that has been said with regard
to the case of simply connected regions remains in force with the only
exception that the stress will not be given by (46.1 1), but by the general
formula (46.4'), because now Xx + Yy will, generally speaking, be dif¬
ferent from zero.
 Chapter 7

TRANSFORMATION OF THE BASIC FORMULAE FOR

CONFORMAL MAPPING

§ 47. Conformal transformation. In this section the simplest

properties of conformal transformations will be recalled, without proofs
being given. An elementary study of the theory of conformal transfor¬
mation may be found in V. I. Smirnov [1], Vol. Ill, and in S. A. lan-
chevskii [1]. More detailed studies of the theoretical problems are given
in the books by I. I. Privalov [1] and A. I. Markushevich [1]. The
recently published book of M. A. Lavrent'ev and B. B. Shabat [1] should
also be recommended.
Let z and 已 be two complex variables such that z = co© (47.1)
where 3(0 is a single-valued analytic function in some region S in the
乙 plane. The equation (47.1 ) relates every point 乙 of E to some definite
point z in the z plane. These latter points will cover in the z plane some
region S. Conversely, let it be assumed that each point z of S, by (47.1),
corresponds to some definite point of 2. It will then be said that (47.1)
determines an invertible single-valued conformal transformation or con¬
formal mapping of the region S into the region S (or conversely). (In the
sequel, when speaking of conformal transformations, they will always
be assumed to be reversible and single-valued.)
The transformation is called conformal, because of the following
property which is characteristic for relations of the form (47.1), where
3(Q is a holomorphic function: If in S two linear elements be taken which
extend from some point 已 and form between them an angle a, the cor¬
responding elements in S will form the same angle a and the sense of the
angle will be maintained.
Unless stated otherwise, regions, considered in the sequel, will always
be assumed to be bounded by one or several simple contours, as was
stated in § 37. The regions S and S may be finite or infinite (and, in
particular, one of them may be finite, while the other is infinite). If, for
example, the region S is finite and S is infinite, the function 3(〈) must
become infinite at some point of S (as otherwise there would not be some

176
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 177

point of £ corresponding to the point at infinity in S). It is easily

proved that w(Q must have a simple pole at that point, i.e., assuming
for simplicity that z = oo corresponds to E = 0, then
c
3(0 = + a holomorphic function, (47.2)
《
where c is a constant and no other singularities can occur in S ; otherwise
the transformation would not be reversible and single-valued. If S and S
are both infinite and the points at infinity correspond to each other, the
function co(Q must for the same reason have the form
co(Q =魂+a holomorphic function, (47.2')
where 2? is a constant. It will be remembered that a function, holo¬
morphic in an infinite region, is understood to be one which is holomor¬
phic in any finite part of this region and which for sufficiently large ［ 乙|
may be represented by a series of the form
01 口2
ao + 石 +
W十
Further, it may be shown that the derivative 3‘(。cannot become zero
in E; otherwise the transformation would not be reversible and single-
valued.
Next there arises the following question : If two arbitrary regions
S and S be given, is it always possible to find a function 3(0 such
that (47.1) gives a transformation of S into E (and vice versa) ?
This problem has been solved in recent times with extremely wide
generality. Here only some general remarks will be made. First of all,
it is obviously impossible to obtain a (reversible and single-valued)
transformation of a simply connected region into a multiply connected
one.
Consider now the case when the two regions are simply connected and
bounded by simple contours. Then a relation of the form (47.1), mapping
the one region on to the other, can always be found and the function will
be continuous up to the contours. In addition, the function may
always be chosen so that an arbitrarily given point of E corresponds to
an arbitrarily given point / of S and that the directions of arbitrarily
chosen linear elements, passing through〈°and 面， correspond. These
supplementary conditions will fully determine the function 3(E).
For simplicity, suppose that S is the unit circle with its centre at the
origin. Denote the circumference of the circle by 丫，so that one has on 丫
178 IL PLANE THEORY OF ELASTICITY §47

|C ] = 1. Since the transformation is to be continuous up to the contours,

the function 3© will be continuous on 丫 from the left (taking the anti¬
clockwise direction as positive); let its boundary values be denoted by
3(a), where b = 浮 is a point of 丫.
In the sequel, the behaviour of the derivative near and on 丫
will be of interest ; in particular, the question has to be considered whether
3'(〈) vanishes at any point of the contour. This problem is resolved by
the following proposition. For the sake of simplicity it has here been
formulated for less general conditions than in the paper by V. I. Smir¬
nov [2] ; the same remark applies to the subsequent proposition regarding
the second derivative.
If the coordinates of the points of the contour of S have continuous
derivatives up to the second order along the arc (i.e., if the curvature of the
contour changes continuously) , the function 3‘(C) is continuous up to
丫 and, denoting its boundary values by 3'

3, = ; (47.3)
do
in addition,
3' ¥ 0 everywhere on 丫 (47.4)
(it being already known that 3'(C) #0 inside 丫). Further, if the coordi¬
nates of the points of the contour of S have also continuous derivatives up
to the third order, the second derivative 3"© will be continuous on 丫

-
from the left and its boundary value to"(a) is given by

3" 空” (473)

In the sequel, unless stated otherwise, it will assumed that one is dealing
with contours satisfying these conditions.
Note also that, once the region S has been mapped on to the unit circle,
it can always be transformed into the infinite plane with a circular hole.
For this purpose it is sufficient to make the substitution

in fact, when 已 covers the region | | < 1,

^ covers the infinite region
with a circular hole | Ci I > 1 , and hence, considering z as a function of
Ei，one obtains the required transformation. In the sequel, finite simply
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 179

connected regions will almost always be mapped on to the circle [ [ < 1,

and infinite simply connected regions on to the region R | > 1 . In both^
cases one could limit oneself to transformations into the circle j | < 1,
but the stated convention is somewhat more convenient in practical
applications.
Next, a few remarks will be made with respect to multiply connected
regions. Obviously only regions of equal connectivity may be mapped
on to each other. For example, a doubly connected region S (i.e., a region,
bounded by two contours, because regions of more general shape will not
be considered here) may always be mapped on to a circular ring. But, in
contrast to the case of simply connected regions, this ring may not be
chosen quite arbitrarily. The ratio of the radii of the inner and outer
circles will depend on the shape of S.
Two simple theorems will now be stated which are very useful in
practice:
I. Let £ 加 a finite oy infinite {connected) region in the 工 plane, bounded
by a simple contour 丫 (no other assumptions being required with respect
to the contour) , let s© be a function, holomorphic in £ (including
the point at infinity, if the region is infinite) and continuous %力 to the
contour. Further, let the points, defined by z = co(〈)，describe in the z plane
(moving always in and the same direction some simple contour 工,
when C describes 丫 (where it is assumed that different points of 丫 correspond
to different points of £). Then z = co(Q gives the conformal transformation
of the region S, contained inside L, on region S {and vice versa) (cf.
W. F. Osgood [1] p. 377 where a completely elementary proof is given,
assuming the contours 丫 and L to consist of a finite number of smooth
arcs) .
This theorem may be generalized to the case of multiply connected
regions in the following manner: (The proof of the generalized theorem
differs little from that given for the preceding one by Osgood).
II. Let 2 be a finite or infinite {connected) region, bounded by several
simple contours 丫卜 y2, .…，丫兀 [having joints in common) . Let be
a function, holomorphic in S and continuous 〃力 力。the boundary, and let the
point z, defined by z = co(Q, describe in the z plane the simple contours
工i， 工2，…， 耳 {not having common points), bounding $o冽e [connected^
region S, when 已 describes the contours 丫小 ...,丫卧 For this purpose it has
been assumed that, when 乙 describes the boundary o/ £ in the positive direction
Re, leaving S RZ the time o% the the corresponding j)oint z describes the
180 II. PLANE THEORY OF ELASTICITY §48

boundary 。/ S likewise in the positive direction. Under these conditions

z = 3(0 represents the conformal transformation o/ S on to {and vice
versa) .
These theorems are easily generalized in different directions (eg, for
the case, when the boundaries contain arcs) , but this will not be done here.

NOTE. It is easily seen that, if 2 and S are conformally transformed

into one another by a relation of the form (47.1), the point z will move in
the positive direction along the boundary of S, when 〈 describes the
boundary of S in the positive direction. This condition has not been in¬
troduced into the formulation of Theorem I, since it is not required in the
proof; the conditions, included there, are already sufficient to prove the
theorem. Thus, the direction in which S is described will without fail
be as stated above. But for the formulation of Theorem II this condition
is necessary ; otherwise the theorem may be found to be untrue.
The statement, referring to the directions in which the contours are described, may
be proved in the following manner. Let v be the normal to the boundary of S,
directed inward, and t the tangent in the positive direction of the boundary ;then
v will be pointing to the left of t. The same relation will exist between the corres¬
ponding directions of n and i at points of the boundary of S, in view of the fact that
conformal transformation does not only preserve the magnitudes of angles, but also
their sense. Here it has been assumed that the transformation is conformal up to
the boundary, but the above stated property is also easily proved for the general
case.

§ 48. Simple examples of conformal mapping.

1°. Bilinear function. Consider the case, when z is a. bilinear
function of〈
十6
z= (48」)
式十，
where a, 6, c, d are constants (in general, complex) and ad 6c W 0
(the latter condition having been introduced to exclude the case, when
—
the right-hand side of (48.1) does not depend on Q. Solving (48.1) forC, one
obtains the, likewise bilinear, inverse transformation
—— 十 6
(48.1)

Thus every point of the 已 plane corresponds to a definite point of the

z plane and vice versa. The point at infinity has not been excluded. In
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 181

fact, the point 乙 = corresponds to the point z = 8, and z = ——

to = oo. Hence (48.1) gives an invertible, single-valued relation between
the unbounded planes of z and ；
(It may be shown that the function
(48. 1 ) is the only one having the stated property.)
The bilinear transformation has the remarkable property that it preserves
〈 plane to a circle in the z plane
circles, i.e., that it relates any circle in the
and vice versa. For this purpose straight lines are to be considered par¬
ticular cases of circles. This is most simply proved in the following way.
The equation of any circle in the z plane is known to be of the form
A(x2 十 娟 3% Cy d —
where A, B, C, D are real constants (the case 4 = 0 corresponding to
straight lines). Since % =
Z

2
z
, y= —
Z'

2，
—
x2 + y2 = zz, this equation
may be written
Azz + Mz + Mz + D = 0,
where A and D are real and M, M are conjugate complex constants.
It is easily verified that, conversely, an equation of the preceding type
may always be reduced to the form (a). In order to obtain now the
equations of the lines corresponding in the 已 plane to the circles in the z
plane, it is sufficient to substitute in (方) from (48.1). After some simpli¬
fications, one finds
+
4a 跖^ + Ml 十 = o, a
where Xo, DQ are real, Mo, / are conjugate complex constants. Hence
one has again obtained the equation of circles, as was to be proved.
One of the simplest particular cases of (48.1) is

(48.2)
〈 z
where R is a real constant ; let it be assumed that 7? > 0. In order to give
a clear description of this transformation, the concept of the reflection
「be the circle with radius R and
of a point in a circle will be recalled. Let
with the origin as centre. Let z be some point in its plane. Construct
another point z', related to z in the following manner:
= R2. (48.3)
If % = then obviously z' = where 夕 = | z [ and f ~\zf \ are
182 II. PLANE THEORY OF ELASTICITY §48

the distances of z and z' from 0 which are connected by the relation
= R2. (483)
Thus, the points z and z' are located on the same ray through 0 and their
distances from that point are related by (483). The point z', related to
the point z in the above manner, is
called the reflection of z in T. Clearly
z is in the same sense the reflection
of z'. The transformation (48.3), re¬
lating z and z', is also called an in¬
version. The points z and z' are also
called conjugate points with respect
to the circle F. When one of the
points is given, the other is easily
constructed by the use of a compass
and ruler: If, for example, z be given
outside r, it is sufficient for the
construction of to draw the tangent
from z to T and from there the per¬
pendicular to the ray Oz (Fig. 20).
Obviously, for an inversion, the points of F correspond to themselves
and the point z = 8 corresponds to z' = 0 ; points outside T go over into
points inside, and vice versa.
Now consider the transformation (48.2). Imagine that the 〈 plane is
placed on top of the z plane in such a way that the origins and axes
of their coordinate systems coincide. The point & corresponding to the
point z = 夕e玲，will then be given by
= *e =z
Hence the point 已 may be found in the following manner : Reflect the
point z in the circle T and reflect its image, thus obtained, in the real
axis; the latter image will be the point 已 (Fig. 20).
Next, another bilinear transformation of the form
__ 〈 ，一 _ z
z
]
— 戏，
C
1 +©
will be studied, where a is a real positive constant. The points 〈= 0
(48.4)

〈 = 1/〃 correspond to the points z = 0 and z = 8; the point 已 = 8

and
corresponds to the point z =
—
1 /% Thus straight lines, passing through
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 183

E = 0 in the 已 plane, will correspond to circles, passing through the points

O(z = 0) and O'(z =
— 1

in the z plane (Fig.s 21a, 21b). Further, concentric circles with the centre
at C = 0 will correspond to circles
in the z plane which are orthogonal
to the circles, passing through the
points z = 0 and z = — 1 /g (as a
consequence of the fact that the
transformation is conformal) ; the
centres of these circles obviously
lie on the axis Ox.
Draw about the origin of the
E plane the circle 丫 with radius
p. The points
《= +p and 乙 —
-—p
will correspond in the z plane to
the points

1 —
— 上 _ Fig. 216.

—
b，= - , 6“ = VO (48.5)
ap 1十
184 II. PLANE THEORY OF ELASTICITY § 48

on the Ox axis. Thus the abscissa c of the centre of the circle L, correspond¬
ing in the z plane to the circle 丫，and its radius are given by

c = 爰e'+ 方") = 丁竺于

1 —— 口铝
吉(6' — 从) -1 —--
展俨
； (48.6)
—•

let it be assumed that 夕 v 0, if the point b' lies to the left of the point 力".
If p V 1 /讥, then 右'> 0 and r > 0. When g '[a, y and c increase beyond
f

all bounds and L becomes in the limit the straight line perpendicular to
the axis Ox and passing through the point K with abscissa 1/2口. If
p > 1 /g, the corresponding circle in the z plane lies on the other side of
—
this straight line.
、
Consider now two circles L and L2 in the z plane, corresponding to
two circles 丫〕and 丫？with radii px and p2 in the 〈 plane, and let px < p2 <
< 1/% Then, obviously, the transformation (48.4) gives the conformal
mapping of the region, contained between the two eccentric circles
and 工2,on the ring, bounded by 丫】and y2- Provided the elements, deter¬
mining the first region, be given, i.e., the radii
4， 夕2 %> 4)
of the circles J, % and the distance % between their centres (% < r2 ~ 々),
then it is easy to determine the quantity a, appearing in (48.4), and the
radii p2 of the circles 丫力 丫 【nfact, these quantities are given by the
formulae

7^需‘ 了二万醺 一]_*? = 4

=

from which one obtains

"
- -—
_—
立 *' ” '
— …一
一 .
'
"(W — 瑾)2 2户(理 建)
J (48.8)
V 1 + 4y初 2
+ 4理 B — ].
%=
. 1
j P2= " 1 2

行
The quantities a, plr p2 are easily constructed by the use of a compass and ruler.
It is obvious that the points z = 0 and z =
—
1/a are simultaneously conjugate
with respect to the two circles 工1and L2, and this property allows the immediate
construction of the above points.

By the same method, the infinite region, consisting of the points

outside two given circles J and L3 (Fig. 21a), may be mapped on the ring
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 185

bounded by the two concentric circles 丫^ and 丫3 with rad五 and p3.
In this case p3 > 1 /a.
2°. P a s c a Fs limacon
Let
Z = 3(0 = R(C + 加b), 衣 > 0, 0< w 1 (48.9)
Putting
z= % + iyt 已 = pe'*,
one finds
+ = 衣(P净 3p2g2巧，
whence
% = R(p cos 8 + 加p? cos 29), y = 衣 (p sin 眇 wp2 sin 20-). (48. 1 0)
When the point C describes the unit circle 丫，the point (x, y) describes in
the z plane the curve L the parametric representation of which is
冗 = R(cos 分 十 加 cos 28), 少 = K(sin 眇 + 加 sin 29). (48. 1 1)
This curve is called Pascal's limacon and it is a particular case of the
epitrochoids studied later on. If, as has been assumed,
0< m<
this curve does not intersect itself and, while 8 varies from 0 to 2兀， the
point z traces it out in one and the same direction. Thus, by what has been
stated at the end of the preceding section, (48.9) gives the conformal
transformation of the region inside Pascal's limagon on to the unit circle.
For 例 = 0, the limacon of Pascal becomes a circle and, for 加 = 扪
a cardioid. In the latter case, the curve has a cusp at the point, corres¬
ponding to = —
1, since 3‘(0 = 0 there. (The fact that 3'© becomes
zero on 丫 does not contradict the statements of § 47, since in the case of
the cardioid the boundary has a cusp.)
Circles with radii p<l in the 已 plane also correspond to limagons of Pascal
the parametric representation of which is obtained by putting in (48.10)
p = const.
〈 plane are transformed into curves in
The radii of the circles 丫 in the
the z plane, their parametric representation being found by putting
分 = const.
186 II. PLANE THEORY OF ELASTICITY §48

in (48.10) (p will now be the parameter, 0 < p 1) ; these curves are

easily verified to be parabolas. In fig. 22夕 are shown

Fig. 22a. Fig. 226.

the curves, corresponding to the circles p —

const, and the rays 令 = const,
of Fig. 22札 These curves are, of course, orthogonal.

3°. Epitrochoids.
Let z = co© = 衣依 + 加 K>0, Q<m<~,
n
(48.12)

where % is an integer larger than unity. Putting, as before, z = % iy

〈 = p鸿 one finds
and
% = R(p cos 0- + wpn cos y = sin 打 + 加p" sin%A). (48.13)
The circle | E | = p = 1 corresponds in the z plane to the curve L with the
parametric representation
= R(sin 眇 sin 处9). (48.14)
“ = 7?(cos 0* + w cos y 出

These curves are epitrochoids. In fact, if a circle of radius々rolls (in the z

plane) on the outside of a circle with radius &, then a point M, lying
at a fixed distance % from the centre of the moving circle and travelling
with it, describes the curve
*= +々) cos 秽 十 2 cos 笈趴 夕 =々 / sin 眇 十 Z sin 方眇， (48.1 4‘)
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 187

where a denotes the polar angle of the point of contact of the circles and
% = (4 + 夕2)/小 Putting

R n-一 1 _
yi =
n ——
, %=R
%
z = mR9 八

one finds that the curve (48.14) agrees with the curve (48.14). Since, by
assumption, m < 1 /%, one has
I < rv Hence, the point M lies
inside the rolling circle and the
curve does not intersect itself.
In the limiting case m
the point M lies on the
—
1/n
circumference of the rolling
circle and the curve becomes
an epicycloid having n 1
cusps. Fig. 23 shows the case
—
—
n 1 /m = 4. On the basis of
the theorem, stated in § 47, it
is concluded that (48.12) maps
the region inside the curve L
on the region | j < 1. The
circles p = const, of the 乙 plane
correspond in the z plane to
epitrochoids the parametric representation of which is given by (48.13).

4°. Hypotrochoids.
Let
\ ]
, R>0, OV 加 V— , (48.15)
b/ n
where 处 is a positive integer. In this case the curve L corresponding to
|^| = 1 is easily seen to be an hypotrochoid which does not intersect
itself. It is described by the point M of a circle of radius rls rolling on the
inside of a circle with radius & ; if % is the distance of M from the centre of
the moving circle, then

r 1
— n
, ，2 八

n
，
〃「八

It is easily seen that (48.15) maps the outside of L in the z plane on to

the region | j > 1. Circles ] | = p
^ —
const. > 1 in the 已 plane also
188 II. PLANE THEORY OF ELASTICITY §48

correspond to hypotrochoids, in the z plane. If % = 1, the contour L will

be an ellipse; this case will be considered in detail later on. For m = 1/%,
the curve L becomes an hypocycloid with n + 1 cusps.
When n — \jm = 2 or 处 = 1 /a = 3, the corresponding contours
have three or four cusps respectively, and they resemble in shape a triangle
or square. Circles with radii p > 1 in the 已 plane correspond in the z
plane to hypotrochoids which likewise for p near 1 resemble triangles or
squares with rounded corners. In Figs. 24 and 25 the cases % = 1 /加 = 2
and 处 = 1/m — 3 are illustrated.

Fig. 24. Fig. 25.

If in（48.1 5）C is replaced by 1/J one obtains the transformation of the

region outside the hypotrochoids on the unit circle ; in this case

z = 8（C） = R（歹 加b ）. （48.15'）

5°. Elliptic rings.

Let
z= co© = R 《 + 三）， 衣> 0, 加 0, （48.16）
i.e., in the above notation,

% =R （p cos A, y = K（p —竺）sin 认 7）

（48」

Circles with radii pi correspond to ellipses in the z plane, their parametric

CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 189

representation being
% = j? { Pi I cos 9, y 衣 ( Pi ) sin

If pi m, then the semi-axes of the ellipses will be

质 =五 Qi + g=五 (pi - (48.18)

and the point z describes an ellipse in the z plane in an anti-clockwise

direction, as the point 已 moves around the circle with radius pi in the C
plane, likewise in an anti-clockwise direction.
Thus, if one selects in the 已 plane two circles 丫卜 丫窑 with radii p2,

"
and if p2 > Pi 丽,then, by the theorem of § 47, (48.17) maps the region
between the ellipses and L2, corresponding
to these circles, on the ring between them.
The ellipses will be confocal, since by (48.18)
the distance c of the foci of the ellipse J
from the origin is given by c? = 瑞 一 —
= 4物五2, i e., it is independent of Circles
with radii p (p】V p V p2) will become ellipses,
lying between Z，! and% and confocal with the
latter. The rays 眇 = const, in the 〈 plane will
correspond to confocal hyperbolas, having the
same foci as the ellipses. These ellipses and Fig. 26.
hyperbolas are, of course, orthogonal.
If one lets p2 tend to infinity, one obtains in the z plane the infinite
region consisting of the points outside the ellipse Lr ; this region is trans¬
formed into the 〈 plane with the circular opening 丫「 In this case, the
circle px = 1 will always be used, and hence one will have m < 1. For
例 = 1, the ellipse becomes a straight slit. For m ~ 0, one obtains a circle.

If one replaces in (48. 1 6) 乙 by 1/J i.e., if one puts

z = 3(C) ~ — R > 0, 0 < m < 1 , (48.16')

one obtains the transformation of the plane with an elliptic hole into
the unit circle | I < I.
6. As has just been stated, the function

Zi = + Wi — 3 式0= R + 加已〉 R> 0, 0 < m 1 (48.19)

 190 II. PLANE THEORY OF ELASTICITY §49

transforms the infinite 句 plane with an elliptic hole into the circle
〈
［| < 1. The equation of the boundary of the opening will be

衣2(1
+ )2
加
+ 衣2(1 一琢 )2
• (48.20)
Let
1
z = ——， (48.21)
Z1
then, by (48.19),
z = 3© = 而二
R(1 f
腕2)
(4&22)

which maps the finite region bounded by the lemniscate of Booth on the
unit circle. When m is almost equal to unity, this
region differs little from that produced by two
contacting circles of equal radius. Fig. 27 shows
the curve corresponding to 加 = 0.8.
If one replaces the transformation (48.21) by
1
—
--
2 C = ,
Z1 c
where the point c is outside the ellipse (48.20),
then one is easily seen to obtain the transfor¬
mation of some region, which for m = 1 becomes
the infinite plane cut along the arc of a circle,
into the circle |^| < 1. (In fact, f or w = 1, the
ellipse becomes a straight slit, and hence it is
transformed into the arc of a circle, because
the bilinear transformation, of which the above
is a special case, maps straight lines into circles) .

§ 49. Curvilinear coordinates, connected with conformal trans -

formations into circular regions. In the sequel, use will be made of
conformal mapping of a given region S in the z plane on the region £ of
〈 plane, where the latter will either be a circle, a circular ring or the
the
infinite plane with a circular hole ; the origin 已 = 0 will always be chosen
as centre. In all these cases it is natural to introduce polar coordinates
? and 9 in the 已 plane by putting 〈 = pe论 Circles p = const, and radii
6 = const, of the 乙 plane will correspond to certain curves in the z plane
which will be denoted by p = const, and A = const.
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 191

If S is a finite region bounded by one contour L and £ the unit

circle with centre at 已 = 0, it can always be assumed that the points
z = 0 and 已 = 0 correspond to each other. Then the curves p = const in
the z plane will be simple contours, surrounding z = 0, while the curves
仍 = const, will pass through this point and end at the contour L which
will correspond to p = 1.
If S is an infinite region bounded by a simple contour L and S the in¬
finite plane with a circular hole, and if the points 工 = 8 and z = 8
correspond to each other (and it is known that this can always be arranged) ,
the curves p = const, will be contours surrounding L and the curves
0- = const, will start on L and go to infinity. Similar circumstances will
prevail when the infinite region S is mapped on the circle |乙 [ < 1.
Likewise it is easy to understand the distribution of the curves p = const,
and 眇 — const, in the case of a region S, bounded by two contours and
mapped on the circular ring S.
The quantities p and 8 may be considered as curvilinear coordinates of
the point (%, y) of the z plane. They are related to %,y by the equation
十 Ey = 3(0 = 3(pe诒); (49.1)
the lines p = const, and 9 ~ const, will be the coordinate lines which,
as a consequence of the conformal property of the transformation, will
be orthogonal.
Let there be given some point of [y (力
the z plane and draw through it the
relevant lines
p = const, and 9 = const.
Let (p) denote the tangent to the
line 卧 = const, drawn to the side
of increasing p. Let (9) be the tangent
—
to the line p const, drawn to the ~O|
| 詈 _X
side of increasing 8. These tangents
will be called the axes of the curvi- 28.
linear coordinates at the point (p, A).
The system of axes (p), (9) in the stated order is oriented as the system
of axes Ox, Oy, i.e., moving in the positive direction of the axis (p), the
axis (9) is directed to the left. This follows already from the fact that
a conformal transformation preserves the orientation of directions.
—>

Let A be some vector in the z plane, starting from the point z = 3(pe粉)
192 H. PLANE THEORY OF ELASTICITY §50

(Fig. 28). The projections of this vector on the axes Ox, Oy will be denoted
by Ax, Ay, and on the axes (p), (卧) by 月加 Obviously

4P + z/分 = ^{Ax + 2/) (49.2)

where a is the angle between the axes (p) and Ox, measured from the latter
anti-clockwise. If the point z be given a displacement dz in the tangential
direction (p), the corresponding point 口 will undergo a displacement 苑
in the radial direction. Hence
dz = 1 dz I, 更=产|公|,
whence

0
dz
—西 —西
3 熊鹿 _ 治

［一。 I 3名)| 一不7^

祝
Sa) E 化)
3

u•a i西 W 西
p 13m
Hence, by (49.2),

I 3，为I
4+ = -丁 •
(4 + 国) (49.4)
P ©

§ 50. Transformation of the formulae of the plane theory of

elasticity. In the sequel, expressions will be required for the quantities
au , au
—ox— —— (i.e. for the derivatives of the Airy function) the displacements
砂
, ,
and the stresses in terms of the new variable C, defined by
z = 3(0. (50.1)
Denote by
支⑶， 队 ％(叫 巴⑵
the functions which were earlier written as
中(z), Mz), ， 中(z)

and introduce the new notation

班) = 平i(z) = *1(3©), MO = 少i(z) = “(3(Q), (50.2)

峭=
*=% = 温*) (50.3)
CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 193

With this notation, the formulae (31.4) and (32.1) become

au +^--=9(0
—a% au
即
+3
3(0
—— + ——
中'(0 少@ (50.4)
式)
and
2r(% 十 ， 叫 — 乂中 LL b(0 —
—3'(0二 (50.5)

respectively. The components % % of the displacements in terms of

the curvilinear coordinates are, by (49.4),

and hence
- 上@_ (设 十 说)，
+ 勿％ = P —I 3名) (50.6)

如 西卜明一黑西一两｝.
Next, the components of stress will be found in the curvilinear coordi-
nate systems. Denote these components by pp, 89, pA so that, if the system
0'%'y' is placed in such a way that the axes O'%' and O'y' coincide with
(p) and (9) respectively, one has
PP = X；
, 眇分 = K，
， p8 = X；,
(cf. § 39). Then, by (8.8),

益 + 蕊 =羽+Y式， 菊一£ 十 2£熊=(w一X + 2/XJ*a. (50.8) 5c

By (32.9), (32.10) and (49.3), the last giving

已2 (3，(Q)2 仔 ®，(Q)2 芋 3，g
—P 2
13n)|2 — P2 3亿)函 p2
，
c7w
one easily finds
谪 + 藐 = 4猊①g = 2[0>(0 + 丽] (50.9)

99
— pp + 2z・p8 = —^｛3(0
p23«)
代) 十3'(0甲(5・ (50.10)

Subtracting (50.10) from (50.9), one obtains

蓝一 2 = (D(Q 十函 ——p2J(9｛词 3 名)里勤， (50.11)

194 II. PLANE THEORY OF ELASTICITY § 51

giving the stresses acting on the contour p = const, from the side where
p increases. The formulae (50.7) , (50.9) —
(50. 1 1 ) are analogous to those
given by G. V. Kolosov [1, 2].
Finally, a formula will be deduced which relates to the case when an in¬
finite region S is mapped on an infinite region S, so that the points 已 = oo,
z = 8 correspond to each other. Then for large | z by (36.4) and (36.5),

鱼⑶ = — -2k(1
+ —
x)
log % + rz + 叫⑶ ,
(50.12)
队(z) = —
(X ;y\
签不转丁 log z + Fz 十 此⑶ ，
where cpj(z), 做 (z) are functions holomorphic at z = 8. Further, for
sufficiently large j 已|and | z | [cf. (47.2')],

z = 39 =五"7 + 会+ 等十一. (50.13)

Hence, by (50.12),
X -V iY
律) 三 强©
7 (50」4)

临) =当=?题 C + r
「七+ %© (50.15)
2加(1+X)
where (p0(Q, %© are functions, holomorphic for = oo.

§ 51. Boundary conditions in the image regions. First consider

the case when the (finite or infinite) region S is bounded by one simple
contour L. Map this region on the unit circle or on the infinite region out¬
side this circle (there being really no difference, but, generally speaking,
it will be more convenient in practical problems to map finite and infinite
regions on similar types of regions).
The boundary condition of the first fundamental problem, i.e., when the
external stresses acting on the boundary are given, may be expressed
in two ways. Firstly, by starting from (41.5), which becomes in the new
notation
au au __ .
禧+ %可 = 汽⑵ + 2d + 队⑵ = 九 + 纵 + const, on L.
(z)

Introducing C by the relation z = 3(0 and denoting by a — 修心 points

CHAP. 7 TRANSFORMATION OF THE BASIC FORMULAE 195
of the circle 丫，corresponding to the contour L, this condition takes the
form [cf. (50.4)]

中⑹ + s' + 山⑹ = + 改 + const, on y. (51.1)

The expression fr + if2 on the right-hand side of this formula must,

of course, be understood now as a given function of the point o =
of the circle 丫 or, what is the same thing, as a function of the arc 分
of this circle. The function + 屯 is determined on 丫 in the following
manner.
By (41.3), the expression
t

/1 + 2% = % J
力
(X“ + zK)ds (51.2)

is a given function of the point £ on However, since one has between

the points t and o on L and 丫 respectively the single-valued invertible
relationship / = co (a), the quantity 力 牙2 is a definite function of the
point a; thus, this function may be assumed to be known.
The boundary condition of the first fundamental problem can also be
expressed in terms of the functions and ¥, if one makes use of (50.1 1)
which gives (for p = 1)

©⑹ + 0(c) — 3， 9 '⑹ + 3'(b)¥(b)} = pp — z'p》on 丫， (51.3)

where pp and 9s must be understood as known functions of the point c

or, what is the same thing, of the arc 分 of the contour 丫.
The boundary condition of the second fundamental problem may be
written, using (50.5),

*(p(b) —
"
co (b)
——
3‘(g)
1

中'3) —— 33) = 2Mgi 十 很2)on y,

，

(51.4)

where g? are the boundary values of the displacement components

说 and (referred to the 。奴 coordinate axes Ox, Oy), which are given
”
functions of a or of the arc 分 of the circle 丫.
An analogous procedure may be used in the case of a doubly connected
、
region, bounded by two simple contours L and L2, after mapping it on
a circular ring (cf. § 41).
A
 PART III

SOLUTION OF SEVERAL PROBLEMS OF THE PLANE THEORY

OF ELASTICITY BY MEANS OF POWER SERIES

Several simple boundary value problems of the plane theory of elasticity

will be solved by use of power series. This method of solution is directly
applicable to regions bounded by one or two concentric circles. However,
conformal transformation permits extension of the method to regions
of more general shape.
A
 Chapter 8

ON FOURIER SERIES

§ 52. On Fourier series in complex form. In the subsequent

sections use will be made of the expansion of given functions in Fourier
series and it will be more convenient to represent them in complex form ;
some remarks will now be made about this.
Let /(A) be a real function, given in an interval 0 2tc. Under
well-known, very general conditions, such a function may be represented
in the form of a Fourier series
oo
/(3) = + E (afc cos 论A + 阮 sin ^), (52.1)
where
27r 2冗

a* — —J/ /偿) cos 〃分，

兀
~
rc J
// sin 々ACA 伊 = 0, 1, 2, . . .). (52.2)
o o
In order that the function /(9) may be developed in a Fourier series, it is suf¬
ficient, for example, that it satisfy in the interval (0,2k) under consideration the
so-called Dirichlet condition which consists of the following : The function
is continuous in the interval, with the possible exclusion of a finite number
of first order discontinuities, and has a finite number of maxima and minima.
A discontinuity of first order is such that, if 9° be the point of discontinuity and
if the argument 心 tends to 距 from the left or from the right, the function /(9)
tends to (different) finite limits ("limit from the left" or '*limit from the right")
which are usually denoted by /(8°
—
0) and + 0). The Dirichlet condition
further assumes that, when 8 approaches the ends 0 and 2tc of the interval, the
function /(9) tends to definite limits which are denoted by /( + 0) and /(2h 0).
If the Dirichlet condition is satisfied, the Fourier series (52.1) converges at all
——
points of the interval (0,2兀). However, at points of discontinuity, it does not give
the value /(9°), but
_
/0 一 0) + /(% + 0) .
— ；
at the ends 0 and 2兀 of the interval the series gives
/(+0) 十 八2兀一0)
， - ' •

199
200 III. SOLUTION BY SERIES §52

If /(9) does not only satisfy the Dirichlet condition, but is also continuous
throughout the interval 0 2兀， and if further /(0) —
/(2tc), the Fourier series
gives the values of /(8) in the whole interval, including the ends; in this case the
series converges uniformly.
Finally note that functions, satisfying the Dirichlet condition, are particular
cases of so-called "functions with bounded variation". All that has been said here
and later on will remain true, if the Dirichlet condition is replaced by the less strict
requirement that the functions are of bounded variation.

Substituting in (52.1) the known expressions

cos e9 =
e 十
2
e
——,
.
sm =
e
— 2%
e

one finds the expansion

a论 '阮
| e-taa 「)
(52.

which, with
%
万=商 —% =
a*
一 以
如
— 服一尼, (522)

gives
8

/信） 劭十 £（以产心 产). (52.3)

7c = 1

This formula may, obviously, be written

十8
/(8) = S akeik\ (52.4)
— oo

where summation extends over all integers from oo to + co.

In order to deduce expressions for the coefficients ak, note that
2亦

Cm 片q
_= / ° ，if is an integer,
% 死 # 0,
° 谭
[ 2兀， if 热 = 0. (52.5)

Multiplying both sides of (52.4) by 尸斓，where n is any integer or zero,

and integrating with respect to 9 from 0 to 2兀，one obtains
23T 2冗
r +8 广
/ 厂油/⑻站 = E 做 / *6热2私
n
0
k—
—8
0
J
CHAP. 8 ON FOURIER SERIES 201

But, by (52.5), the only n on-zero term on the right-hand side is obtained
for k = n and it is equal to Hence
2"

% (52.6)
27V J
0

The result (52.6) holds also true when the function /(A) is an ordinary
Fourier series. In order to verify this, it is sufficient to note that (52.6)
may be obtained indirectly by replacing in (52.2') a& and 瓦 by their
expressions (52.2).
Consider now an expression of the form 十 仔2(分)， where 九 and
/2 are real functions which may be represented in the interval (0,2兀)
by ordinary Fourier series, and hence by series of the form (52.4). Adding
these series, after multiplying the second one by i, one obviously obtains
a series expansion of the form
九⑻ + 浜网 =卷即产， (52.7)
where
27t

% — /(/i +
o
仇 = 0, ±1, ± 2, . . (52.8)

The only difference from the preceding cases is that there the quantities
an, a_n are conjugate complex numbers, as follows from (52.2') or
(52.6), whereas here an and a_n will not, generally speaking, be conjugate.
NOTE. Separating real and imaginary parts, one may, conversely,
find from (52.7) the common Fourier series for the functions 九(眇)
and In fact, putting 仇加 = + 胃自先 (where a*, are real), one finds
十8

/1 + 仔2 = W (知 十 2’讥) (cos 粉 + i sin 网 =

~©o

4*8 十8

= S (afc cos 々眇 一 瓦 sin 力9) + 2 Z (讥 cos 狗分 + a% sin 万8) =

-—-OO QQ

= & + £{% + cos k以 — (阮 — 兄为) sin 々9}+

=1 左
8

+ 斗0+ zN{仇十 8—0 COS

上工 1
+ (a* — a_fc) sin 力分}.
202 III. SOLUTION BY SERIES §53

Hence
oo
十 2 (月％ cos 68 十 Bk sin 万台)，
儿=1
oo

/2 = 骡；+ w (4 cos 脱 十 及 sin 战)，

a=i
where
Mo = %, Ak = + a_ 元，
—
Bk = •阮 + 8_九,
物4；= 瓦， 屋 8 3— 加 %， 瓦 九 一 a_fc 他= 1,2,3,…).
Incidentally, it follows from the foregoing that the expansion of the form
(52.7) is unique, because this is known to be true for ordinary Fourier
series.

§ 53. On the convergence of Fourier series. If a function "(8)

is continuous and has in the interval 0 <9- 2k continuous derivatives
of order up to and including 丫 一 1 , and if, further, the derivative of order
v satisfies the Dirichlet condition in that interval, the coefficients a比， 0 先
of the Fourier series (52.1) satisfy inequalities of the form
C C
I I < 々v+i ， I 队 I <
力v+i
侬 = 1,2,..) (53.1)

where C is a positive constant.

The above statement that a function is continuous in 0 V 卧 V 2n will be
understood to mean that the function is not only continuous in this interval, but
also that its values at 加e end of 统e interval are equal to each other. The inequalities
(53.1) will also be true, if one assumes that the vth derivative is of bounded
variation.

It follows from (53.1) that the coefficients of the complex Fourier series
(52.7) satisfy inequalities of the form
c
\ ak\ <
I
1
I
« |v+i
(B = ± 1, ± 2, . . .), (53.2)

provided 九传) and 府(眇) satisfy the conditions stated above for /(今).
If v = 1, i.e., in the case, when the function has a first derivative
satisfying the Dirichlet condition, one will have
C C
I hi v I I Q，
阮 V

from which it follows that the Fourier series for /(8) will be uniformly
CHAP. 8 ON FOURIER SERIES 203

and absolutely convergent. (Uniform convergence actually ensures

continuity of /(3) and bounded variation, or, in particular, fulfilment
of the Dirichlet condition.) In fact, one has
2c
I 8k cos 々眇 十 氏 sin. 6分 I I oct|4-|P*|
thus the terms of the series (52.1) are less in absolute value than the
terms of the convergent series
8
2c 8i
= 2C 2 ,
六声 次
with positive terms, which do not depend on 品
 Chapter 9

SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE

§ 54. Solution of the first fundamental problem for the circle.

Solutions of this problem have been given by many authors. A
simpler, but less elementary solution is given in § 80.
Let the origin of coordinates be at the centre of the circle with radius
R. Let Xn, Yn be the known components of the external stresses, acting
on the circumference L of this circle. They will be assumed to be con¬
tinuous and single-valued on L and varying with the polar angle 也
measured like the arc coordinate $ from the positive Ox axis.
By (41.3),

/1 + 巩 = ( xn十 iY^ds = iRj (Xg + (54.1)

o o
It is known that for the existence of a regular solution the functions
九 and /2 must be continuous and single-valued on L (§ 41 , 2°), i.e., one
must have
27r

/(X + iY^ =
o
处 0 (54.2)

(which means that the resultant vector must vanish) . Further, the
condition of zero resultant moment (§ 41)

/仇勿 +
L
=o
here takes the form
2口

/
o
(— fi sin 8 + /2 cos 分)〃眇 — 0. (54.3)

The conditions (54.2) and (54.3) will be assumed satisfied.

204
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 205

The boundary condition (41.2) may be written (putting const. = 0)

中⑵ 十 z<p'(z) + W(z) = /i 仔2 0n E, +
(54.4)
where the symbol z is used for points inside the region as well as on L.
The expression % + 逼 may be represented by the series
+ oo
h+ (54.5)
- co

the coefficients of which may be calculated by the method of § 52;

hence these coefficients will be assumed known.
It is known that the functions W(z) must be holomorphic inside L
and that, by § 41, one may assume 中(0) = 0. Thus (p(z) and “z) may be
developed for | z ] v 五 in power series of the form
OO 8

中(z) — S akzkt
fc-1
W(z) = S a'必羯
次=0
(54.6)

where in the first series the constant term is absent, because of the con¬
dition <p(0) = 0. Further, one has
8 oo
福 = E 为源/T, 丽 = 2 疏亚 (546)
e =0
Assuming these series to converge, not only in the interior, but also
on L, and substituting them in (54.4), one finds
OO oo oo
£ o衣 十 z E ka^"1
k=l fc™ 1
+ £0不浮 = /i + 稿 on L.

But on L: z = R^, 2 — Re~^. Noting also that

8 8 OO

化=1
T
z.2 砺丸办 = S kakRker^k-^ =
左=1
仍
+ £=1侬
光
2)ak+2Rk+2
one finds from the preceding formula, using (54.5),
CO OO 8 +8
b a的湃s +
7c =1
+ £=0传+ 2)源+2
左
用+2 歹做
B 不0
S 说k 小殛 F 2 M丝
一OO

Comparing coefficients of 凌眇， one obtains

力
口工^/^ ~ ^4 工) 口工 ""I- 0工Q …
(54.7)
R
Similarly, one has for 次泡 (％ > 1 )
以九r = 办
An (符〉1), (547)
206 in. SOLUTION BY SERIES §54

Finally, e~in^ (% 0) gives

(% + 2)篇+2冗计2 + RM；
= A_n (% > 0). (54.8)*
The equality (54.7) is only possible, if Ar is real, since = 2an
where 巧 is the real part of 旬. Hence, in order that the problem may be
possible, one must have
&=a real quantity. (54.9)
The meaning of this condition will be explained below. If it is satisfied,
the real part 阳 of the coefficient 旬 is given by
力
沆旬 = a】= — (54.10)
2R
As was to be expected, the imaginary part of % remains indeterminate,
because it is the imaginary part of cp'(O) which may be fixed arbitrarily
(§41), for example, by putting it equal to zero.
Further, the coefficients an {n > 1) are given by (547) as
A

距=君 3>1)， (54」1)

and, finally, one obtains for a'n

仇 0) from (54.8) [replacing all quantities
by their conjugate complex values]
月 Z A

V— + 2)%+2a = 看 一 (％ + 2)
找 A*
(n>0). (54.12)

Thus all coefficients of (54.6) have been determined and the problem
could be considered solved, once it has been proved that the series
for 中(z) and actually satisfy the conditions of the problem. This
question will now be studied, but first the condition (54.9) will be
explained. One has (§ 52)
27t

2ka y
= (九+歹2)厂切8=
0 2k

0 0

i.e., (54.9) leads to (54.3) which expressed that the resultant moment of
the external forces vanishes.
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 207

As regards the question posed above with respect to the series for
<p(z) and W(z), consideration will be limited to the simple case, when not
only the functions Xn and Yn are continuous, but when also their first
order derivatives satisfy the Dirichlet condition. (Actually, it is not
difficult to prove the correctness of the solution for more general con¬
ditions, but this will not be done here.) It is easily shown that under
the above conditions the series
8 8 OO

cp(z) = £ 沈必， = £ ka^"1, W(z) = £ 无

k —1 Jc — 1 k= 1

are absolutely and uniformly convergent on the circle L, and hence

also inside L. Thus 勺， 少， will be continuous up to the boundary and
the solution is regular.
To prove the convergence of these series on L, consider the series,
formed by the moduli of the terms of the former when | z | = R,
Z］瞅 |¥, Zk\ak \Rk~\ S | 4 \Rk.
Since Xn, Yn have first order derivatives, satisfying the Dirichlet con¬
dition, the functions fr and /2 have derivatives of second order, having
the same property. Hence, by what has been said in § 53,

I 4% I v R
［月 v R
(k = 1,2, ..
where C is some constant and, by (54.11) and (54.12),
C C C"
底 /V 切即的
行 筋 声'
where C', are some other constants. From this follows immediately
the convergence of the series (々)， and consequently the uniform and
absolute convergence of the series for 中，乎' and 少.
NOTE. The problem has been solved using the boundary conditions
in the form (41.2). One could also have used the conditions in the form
(4 1 .23) . This alternative proof will be left to the reader (cf. § 56, where
an analogous problem is solved by this method) .

§ 55. Solution of the second fundamental problem for the

circle *). This solution is quite analogous to the preceding one. In fact,
the condition (41.1) gives
W(z) — z<p'(z) — W(z) = 2^ + 欣2) on L.
*) Another solution is given in § 81.
(55.1)
208 III. SOLUTION BY SERIES § 56

Developing the given expression 2Mgi + z'g?) into a complex Fourier

series
+ oo
2场1+ 稔2) = 2-oo (55.2)
-
and substituting the series (54.6) in (55.1), one finds, as before,
+8
e
OO 8 OO

X S g小场
证9
— 册及论 一 £ 3 + 为 =0
2)质+2用
k —0
凑五场一认9 =
-— 8

and hence
区(兑肉
— 藏) = 4, (55.3)
ytanRn = 4 仇 1), — (% + 2) 裔+2区%+2 — a^Rn = A_n 》0). (55.4)
All coefficients are determined by these formulae, i.e., in contrast to
the case of the last section, 曲 is also completely determined by (55.3),
as was to be expected, since in the present problem it is impossible to
fix arbitrarily the imaginary part of @'(0)・ In fact, equation (55.3) and
its conjugate equation give

KGi —— 巨i = 法一,
Jx
比夜1
—— 夕！=
Jx
and hence
41 + /.i
兀
的1 = 2—1
(比 冰
(remembering that always x > 1).
As in § 54, it is easily proved that these series actually satisfy the
conditions of the problem, if, for example, gx and g2 have second order
derivatives satisfying the Dirichlet condition.

§ 56. Solution of the first fundamental problem for the infinite

plane with a circular hole *). This problem may be solved by a method

_
quite similar to that of § 54. However, as demonstration, use will be made of
the boundary condition in the form (41.9). Let the origin of coordinates
be at the centre of the hole of radius R. One has then, in the notation of
§39,
「
〃 一 = N ~iT on the circle L, (56.1)
where N and T (cf. § 41) are the components of the external stresses
*) This problem will be solved by another method in § 82 for the more general
case of an elliptic hole. Cf. also § 87a.
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 209

acting on the circumference of L in the direction of the normal n, outward

with respect to the body (i.e., directed towards the origin), and of the
tangent t, directed to the left of the normal n.
The correctness of (56.1) is easily verified, i.e., the truth of the relations

= N, #9 = T.
The definitions of yy and 旷9 are given in § 39. It should not be overlooked that the
axes and (眇) of § 39 are now in the opposite directions of n and t, while〃and 夕8
refer to stresses, acting on the sides of elements opposite to the direction of n.

The condition (41.23) may be obtained directly from (39.5) which

gives
十函一 产3网)，⑶ 甲 = N iT on L. (56.2)—
[In (41.9) one should have 滔。
instead of 滔 where a = 卧 土兀 is the angle
between the normal n and the axis 0%. But 庚泠 — 浮热，since = 1]
Consider now the formulae (36.4), (36.5) and (36.7) and note that in
the present case the expansions (36.7) hold true in the entire region S,
i.e., outside the circle L (cf. Note at end of § 36.) Differentiating the above-
mentioned formulae, one finds for (z) = and 中(z) = ex¬
pansions of the form

—S晨
OO OQ

= £ akz~k, 乎 (z) 2-尢， (56.3)

化=0

where the notation for the coefficients is different from that of § 36.
In particular, the coefficients a0) av % in (56.3) have the values
讥
o = F = B, a；= r‘ = B + iC' , (56.4)
(remembering that it had been agreed in § 40 to assume C = 0),
_ x+ ，— qx — 钟
<
1
2tc(1 + x) 1
2k(1 + x) '
The formulae (56.5) are not necessary for the solution of the problem.
One has, of course, only to use the condition of single-valuedness of
the displacements which in the present case may be expressed as
乂的 +册=0 (56.6)
[cf. (35.7), where the quantities 丫；and refer to the contour 二出；but
—
here L Llf 丫丸 = alt 丫；— %.] Substituting (56.3) in (56.2) and assuming
210 III. SOLUTION BY SERIES §56

the series to converge on the circle L, one finds (cf. § 54)

1 +k 8〃 a
1 /9
— 即：+2 -ik^
比
以
内 q一流S
无6
i
十 4
y e少泳3 a
斯
'a2说'
6 -6 —
0
0 用 K
= N —汉 on L. (56.7)
Expand the function N — iT, given on L, in a complex Fourier series
十8

N — iT =— oo
(56.8)

and compare coefficients of d吗 after introducing (56.8) into (56.7).

Then one obtains from the constant term and from those involving e论
and 於施 respectively

2曲一
夕
2 /

声 =月0， 天 天 41‘ —— = Az.

j

(56.9)
f

For 泮眇 》3), one finds

含 =4 侬》3), (56.10)

讥$
while 》1) gives

展盘 (见》1)• (56.11)

From (56.10) one finds

dn — AnRn (外 3). (56.12)
Further, it is known that
他 = r, 劭 = r,
(56.4)
where F, T' are known quantities, specifying the stress distribution at
infinity. Hence, by the last of the formulae (56.9),
4
&=不宠2 +阳.(56.13)
In order to find expressions for 曲 and %,it is necessary to refer to the
condition (56.6) for single-valuedness of the displacements which in
combination with the second relation of (56.9) gives
, y^A^R
, 见 = (56.14)
1 + X 1+x
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 211

The first formula of (56.9) leads to

a；= 2VR2 — (56.15)
and, finally, (56.11) determines all coefficients 0；for 将》3:
晨= — 1)R2〃*2 — RS—n+2 仇 3). (56.16)
Thus the problem of determining the coefficients has been solved.
It is easily shown by elementary arguments of the type used in § 54
that, if N and T have second order derivatives satisfying the Dirichlet
condition, the series for '(z) and 乎(z) will be uniformly and ab¬
solutely convergent on L (and consequently also outside 上); it follows
from this that they are solutions of the problem.
NOTE. If one had started from the boundary condition (41.5) instead
of from (56.2), one would have obtained for 箪 (z), 少 (z) series, for which one
could have proved by the method of § 54 that they solve the problem,
provided Xn and Yn have first order derivatives satisfying the Dirichlet
condition. Thus, by applying the boundary condition (41.9) [i.e., con¬
dition (56.2)], one has been forced to impose more restrictive conditions
than would have been necessary with the condition (41.5). However,
it is easily seen that these additional limitations are not due to the problem,
but to the elementary method used in proving the correctness of the
solutions. In fact, it is almost obvious (and this is easily verified directly)
that, starting from (41.5), one would find for 中⑵，W(z) series which
could have been obtained by differentiating those found above for (z),
乎 (z). But as 中 (z), 3(z) satisfy the conditions of the problem, obviously
(z) = q'(z), 里⑶ — ¥'(z) will also solve the problem.

§ 56a. Examples.
1°. Uni-directional tension of a plate, weakened
by a circular hole.
Let the edges of the hole be free from external stresses and let at
infinity
X， = 自 可）= X俨 = 0,
where is a constant, (i.e., tension in the direction Ox which is equal to
0 at infinity). Then, as is shown by (36.10) (remembering that, by sup¬
position, C = 0),
r = —4 , r = — £
2
(56.⑷
212 m. SOLUTION BY SERIES §56a

Further, since on the contour N ~ iT = 0, one must put in the formulae

of § 56 4 = 0 for all k. Under these circumstances (56.12) and (56.16)
give
—
an 0 {n 3), a；
= 0 仇》5).
Also, from (56.4), (56.14), (56.13), (56.15) and (56.16),

, 0 »9 ，八 ， 3整4
以 0j 位工
工- 3
"***

and hence, finally,

-
力 力/ R2
(z) I1
昨)——式一/+ . (56

Next determine the corresponding components of stress in polar

coordinates. By (39.4), putting z — 夕*,

少十 分分 = 4沉 (z) = 0猊 1
/ 2用
八1 一^
—cos (56.3a)

9那 一" 2z•心 2[5 '(z) + 里0]浮 论

whence, separating real and imaginary parts and solving for rr,
and r^, one finds
力 /
-- 冲、 0/
+ TV
4R2
+
3K 4 、 cos 2分

g (1 +
卧卧
IV + 2
(56.4a)

2R2
sin 2^.
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 213

At the internal boundary (i.e., for r = R), as was to be expected, one

has
少 = = 0,
•*•*■***■■.

while the value of 眇眇 there is given by

分分=0(1 — 2 cos 29) on L.

The maximum value of 眇分 thus occurs for cos 2分 = 一 1, i.e., for

兀
9= —2 ,
where
分小w = 3自
so that the value o/ the tensile stress Es increased.
This problem was first solved by G. Kirsch (Zeitschrift des Ver. d. Ing., 1898)
in a quite different way. Cf. also the solution by G. V. Kolosov [1], pp. 20 24.
The solution for the case of an elliptic hole will be given in § 82a.
—
In order to find the displacements, calculate the functions

平(z) = / ^{z)dzt W(z) = / 中(z)dz.

One obtains, omitting unimportant constants,

平⑶
4 \ )， W(z) = — —2 (\
z/
d z
y )- (56.20
z'/

Then, by (39.3), one has

22(% + 谢9) = 0一叫xw⑶

「
力 、 2R2
'
~iZq/®
—2R2
醇z)} =
2R2 2R4
hx— l)r + —— 6司普
x
歹
浮出 十 2y歹2用
yr3.

whence,separating real and imaginary parts,

8V T (比 一 1W2 + 2R2 + 五 *— cos 2",

看阿一 ］)+ * + ——Ism 2

214 III. SOLUTION BY SERIES §56a

2°. B i~a x i a 1 tension.

The problem of bi-axial tension of a plate with a circular hole is solved
still more easily ; in that case one has at infinity

—
X俨 = = 夕， = 0. X%
By (36.8)
r = ——力2 , r = o,
and, similarly as before, one finds

while all other coefficients of the series for (z), 乎 (z) vanish. Hence
力 以?2
= 上， 甲0 = (56.5a)
2 严

and
众R2
—2
力
z, 山(z) —
，

z
. (56.5'女)

The stresses and displacements can be calculated, using, as before,

the formulae (39.4) and (39.3) which give

(1
rr =一少 2 )， 9分 一 0(1 + 2 ) * — °， (56.6g)

vr = 4
M
[仅 一 I)/2 + 2R2], % = 0. (56.6'a)

This solution could have been obtained directly from the solution of
problem 1° by superimposing two uni-directional stress distributions
along the axes Ox and Oy respectively.

3°. Uniform normal pressure, applied t o the

edge of a circular hole.
Consider now the case when the edge of the hole is subject to uniform
normal pressure P and when the stresses vanish at infinity. Then
n = — p, r = o, r = r = o.
In (56.8)
4 = —P 4= o
^
(6#o)；
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 215

hence, on the basis of the formulae of § 56, one finds that

a' = PR2
4

and that all other coefficients of the expansions for and 甲(z) are
zero. Thus
PR^ PR2

and
(z) = 0, X(z) = 2 > 中 ~
二

—
〃=
PR2
z

, —
W(z)

99 =
PR2
,
_
心
—
= 0,
z
(56.7a)

夕2 夕2

PR2 (56.7%)
VT = , % = 0・
2W
4°. A concentrated force, applied a t a point o f
the infinite plane.
Let the stresses at infinity be zero (F = T' = 0) and the stress, applied
to the edge of the circular hole, have constant magnitude and direction :
X
(56.8。)
2元R
where X, Y are constants. Obviously (X, Y) is the resultant vector
of the external forces.
Under these conditions the normal and tangential stresses N, T are
given by
N =— 工
2
(X cos 0- + y sin 池
九次2冗R
T=一 ，不(一 X sin 8 + Y cos 8),
whence

〃— = N — iT = 2kR
(X ~ iY)^ on the contour.

Hence only one of the coefficients in (56.8) does not vanish, i.e.,
,
Ai —
— — —— ,
ZY X

and, by (56.14) and (56.16) for 笈 = 3,

q =2 •
〃
X + Y,
—— —
2tc(1 乂)
L = 2
x(X iY)
兀(14- x)
— ,
口安 — 尺4
(1 十 M)
兀
+
while the remaining an and 夕; are zero.
216 III. SOLUTION BY SERIES §56a

Thus the problem is solved by the functions

叩z)= 暨二9
27r (1 + x) z
二二 x) 学:
2k(1 十乂) z 兀
+ (1 z3
Let it now be assumed that the radius of the hole tends to zero and that
the stress (Xn> Y^) increases beyond all bounds, so that the resultant
vector (X, Y) remains unchanged. Then the preceding formulae give
x + ay i
= 2k(1 十%)—z
华 %)=£*；•
(1 + X) z 2兀
•明

Under the stated circumstances it will be said that « concentrated

force (X, Y) acts at 0. The state of stress, caused by a concentrated
force, is determined by the functions T of (56.9«) . The determination
of the components of stress and displacement does not offer any dif¬
ficulties. For example, the stress components in polar coordinates are
given by
尸'
x+3 X cos 眇 十 Y sin A.
yy ••

2兀(y +1) y '

卜 —=
8 -—
乂

-——
一1
2加(k + 1)
X cos » 十 丫 sin A
-—
夕
— (56.9%)

y cos a
心
x
——十1 1) —
X sin 一
r ——
—
"

NOTE. When considering thin plates ("generalized plane stress"),

the constant 乂 in the preceding formulae must be replaced by
3- o
比*
1 +b

。
(cf. § 32), and the quantities X, Y by
。
X
~2h}
V

where X°, Y° are the components of the concentrated force, applied

to the plate of thickness 2为. In fact, it must not be forgotten that X
and 丫 are distributed over the thickness of the plate.

5°. Concentrated couple

Consider now the case, when a constant tangential force T is applied
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 217

to the edge of the hole. Let the stresses vanish at infinity. Then
〃= 0, » = T on
夕 the contour,
and only the coefficient 4° = — iT in the series (56.8) will be different
from zero.
The formulae of § 56 give
以 =— 2孤2 = "R2；
all other quantities an, a；vanish. Hence, putting

one has
iM 1
(z) = 0, 乎(z) = — 三一 — , (56.10a)

where M obviously denotes the resultant moment about the centre

of the external forces, applied to the boundary. These formulae remain
valid also in the limiting case, when R decreases and T increases in such a
way that M remains constant. Then (56.10«) leads to what will be called
the effect of a concentrated couple, with moment M about the origin,
on the infinite plane. The stress components are easily found to be

〃= 分9 = 0, 夕眇 = —. (56. Ila)
2m
(Cf. also the Note preceding this example) .

§ 57. On the general problem of concentrated forces. In § 56夕，4°,

expressions have been found for the functions and T, corresponding
to concentrated forces acting at the origin of coordinates on an un¬
bounded body. Now let the region S be arbitrary in shape and, in addition
to ordinary forces corresponding to the functions and T holomorphic
in S, let a concentrated force (X, Y) be applied to the body, say at the
point z = 0. The effect of this concentrated force may be superimposed
on that of the ordinary forces, and therefore the functions <I> and 空 will
have the form

=一^^*L+2k(1 十乂) z
。(以
2tc(1 + x) z
瑞团 尔⑴

near z = 0 [cf. (56.9〃)], where and 乎。are functions holomorphic

°
218 III. SOLUTION BY SERIES §57

in the neighbourhood of that point. If the concentrated force is applied

at some arbitrary point z = z0, instead of at z = 0, then, using as
origin of an auxiliary coordinate system, (57.1) will take the form

=
x十 i n
里击1)
- x(X — iY) 1 +理
-笳ETE + 3 2k(1 + X) 通
where 画 z
and (38.4)
— z°. Reverting to the old system, one finds from (38.3)

X ^iY 1
=
-
工

o.
2兀(1十 z 20
(57.2)
里⑵ = 2兀 —
qx 祖 1 2o(X + 27) 1
+n
(1 十 X) Z Zq 27t (1 + x) (z
—
z°)2
The index 0 on the symbol indicates that the function is holomorphic
near the point z = z°. Integrating one obtains for 中 and %

中 (z) ~ — 5 (j'wy l°g (z — z。) + ?0>

兀 比
(57.3)
K(X-zY) _ z。) 2o(x+z・y) 1
少⑵
八
际” 3
°g + 十

In an analogous manner one finds for a concentrated couple M, applied

at z = z°,
iM 1
(z) = 观(z), 甲⑶ = 一行一 + %⑵， (57.4)
ZK (Z
—— Z0)“
and
£M 1
中 = 平o(z), W(z) = —
2兀 ；
(z
—
V
z°) + 狐 (z) (57.5)
[cf. (56.10«)].
It is thus seen that the point of application of a concentrated force
or couple is an isolated singular point of the functions 中，少，0, Con¬
。 +
versely, every isolated singular point z = %。 iy^ of these functions
(if the existence of such points is admitted) may be considered the
point of application of concentrated forces or moments. In order to
determine the analytic character of the functions 9 and 中 near these
points, it is sufficient to apply the reasoning of § 35 by surrounding the
point Zq by a sufficiently small contour £0 and considering this contour
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 219

as one of the boundaries of S. Then, by § 35, one has near z =

X + EP
中 (z) = — 2 (1 + x)
兀
rl°g (Z Zo) + — (57.6)
乂 (X
少⑵ = o2兀 /1
(1
—+^y)、 log (z — 陶 + 少*(*
上 x)

where 平*, 少* are single-valued near z。， X and Y are the components
of the resultant vector of the external forces, applied to Lo (or to any
other contour, surrounding z0).
The functions 铲，少*, which are single-valued in the neighbourhood of
the isolated singular point z°, may be represented by the Laurent series
+8
中* = E d
QQ
+ E%) (Z — Zo) 力
+8
少* = E (a；+ 冏) (z—Zo产. (57.7)
— OO

Simple reasoning, based on (33.3), shows that the resultant moment

about the origin of the forces, applied to Lo from the inside, is given
by

"o， = 2印二 — 孙:
见
Mo- 2邓二 + 工 + %0Y— (57.8)
1 十比 1 + x
(provided the contour is infinitely small). Taking into consideration
that the resultant vector of these forces is (X, Y), one obtains for the
resultant moment M about the point z0

M — (％。V -夕㈤ = 2兀3二 - 1 十比

(57.9)

Thus, is the point of application of the concentrated force (X, Y) and

of the concentrated couple with moment M.
However, it is seen, that the knowledge of X, Y and M does not
In fact, the coefficients of
yet determine the singularities of 9 and
the negative powers of (z — “
zj in (57.7), which characterize the singu¬
larities of 9 and 巾，are arbitrary (within the limits of convergence), with
the exception of the imaginary part of the coefficient a」4- 裆二1 which
is determined by (57.9). Thus, the nature of the singularities, caused
by concentrated forces and couples, remains to a large measure un¬
decided, unless additional conditions are introduced. It was only possible
to obtain completely defined expressions for these singularities [cf.
—
(57.2) (57.5)], because the concentrated forces and couple were intro¬
duced by means of a definite limiting process.
220 in. SOLUTION BY SERIES §57

Exactly the same expressions would have been found by a number of

other limiting processes. One of the simplest examples, which more or
less accurately reproduces the special conditions of the application of
"concentrated” forces and couples, will now be stated. Imagine that
a rigid disc be introduced into a circular hole in an infinite plate,
and let this disc, which has the same radius as the hole, be joined to the
plate along its circumference. Further, let some force and couple (in the
plane of the plate) act on this disc. The solution of the problem of elastic
equilibrium of plates under these conditions will be given below (cf. § 83«,
examples 3°. and 4°., where the solution of the more general case of an
elliptic disc is given) . If one now allows the radius of the disc to tend to
zero, leaving the force and couple unchanged, the above-mentioned
solution gives in the limit a result which agrees exactly with those
obtained above.
In the sequel, when speaking of concentrated forces and couples
applied to internal points of bodies, it will be assumed that the cor¬
responding singularities are given by the formulae (57.2) (57.5).
NOTE. The above formulae may be applied to the determination of
—
particular solutions of the equations of the plane theory of elasticity in
the presence of body forces, which gives the possibility to reduce these
equations to the equations for the case when body forces are absent
(§ 28). For example, one may proceed for this purpose in the following
manner. Omitting in the formulae (57.2) and (57.3) the terms °, To,
@o，Wo，one obtains certain particular solutions (for the case of no body
forces) which correspond to the effect of a concentrated force (X, Y)
applied at the point z。= + 济.
The displacement components corresponding to this solution are given
by

2兀(1 +--- — zq)(z 一 切]

划 ——— 一 log〔(z
x)
(57.10)
• — — —
X iY z
—
] '' “ i ，
“
5
2k(1 乂) z z0

obtained from (31.2) and (57.2). The stress components may be evaluated
in a similar manner using (32.9), (31.10) and (57.2).
Choose now instead of X and Y the quantities X(x0, y^dS^ and
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 221

Y (x0, y^dSf,, where X (x0, y0) , Y (%0, y0) are certain (real) functions of
the point x0, y0 and dS0 = dxQ dy0 is an (infinitesimal) area element
surrounding this point. In that case one obtains the displacement and
stress components which correspond (approximately) to the effect of
body forces [with components X (%0, y0) , Y (%0, y0)] on the part of the
body corresponding to the element dS0. Summing these expressions for
all the elements dS0, one is easily seen to arrive at some particular
solution corresponding to the effect of body forces (X, Y) on the entire
body under consideration.
In particular, one finds for the displacement components

2*3 + 汨) = - -
2k(1 十 k) J J
(X + iY) log (z — 而)(5 — 而)〃 d% +
£ (57.11)

+ 乙人 (J L x)
人/
j
a/ a/
H) ；二年
u "0
的0•

The corresponding stress components can be evaluated either directly

in an analogous manner or with the help of formulae expressing the
stress components in terms of the displacements given by (57.H).
It is readily verified directly that for sufficiently general assumptions
regarding the functions X(x0, y0), Y (x0, y0) one can actually derive a
certain particular solution of the equations under consideration.

§ 58. Some cases of equilibrium of infinite plates containing

circular discs of different material *. By means of a simple modification
of the formulae of § 56仇， a number of problems are easily solved which
are important from the point of view of application. These problems
refer to the equilibrium of infinite plates with circular holes into which
discs of the same or other (likewise isotropic and homogeneous) material
have been inserted.
For the solution of some of these problems use will be made of the
solution of the problem of equilibrium of elastic circular (continuous)
discs under the influence of uniform normal pressure, applied to their
edges. This solution was already stated in § 41a for discs of arbitrary
* The problems considered here are particular cases of the problems of the
equilibrium of elastic homogeneous bodies made of different materials, bounded
by concentric circles. Certain references dealing with these problems will be given
in § 59a, 3°.
222 in. SOLUTION BY SERIES §58

shape ; in the present case it may, of course, be obtained directly from

the formulae of § 54, but it is simplest to utilize the fact that the con¬
ditions of the problem will obviously be fulfilled by putting
—
X* = P, Yy ~ P, Xy = 0 —
in the entire disc, where P denotes the magnitude of the constant
(58.1)

pressure, applied to the edge.

In fact, for such a stress distribution, the stress on an arbitrarily oriented
element reduces to the normal pressure P; this follows immediately from (8.8).
Thus, in particular, the edge will be subject to the normal pressure P. This will
likewise remain true in the case of discs of arbitrary shape.

It is easily verified that the functions (z), 乎(z), 中⑵，少 (z), corre¬
sponding to the state of stress (58.1), have the form

(z) = — 乎⑵ = 0, 中 (z) = W(z) =0 (58.2)

(omitting unessential arbitrary terms which only effect the rigid body
displacements) . The polar components of stress and displacement follow
then from (39.4) and (39.3) :

〃— = — P, 夕卧 = 0, (58.3)

vr = ~
4出 —, % = 0. (583)

Next, a number of selected problems will be solved.

1°. Infinite plate with a circular hole into which

a n elastic circular disc with a n originally larger
radius has been inserted.
It will be assumed that there is no friction between the disc and the
plate, so that the interaction of these bodies reduces to normal pressure
on the edges of the disc and the plate. In view of the complete symmetry,
this pressure will be constant along the boundaries. Therefore it is
obvious that the solution of this problem may be constructed from the
solution of problem 3° of § 56«
— ——for a plate with a hole and from
(58.3) and (583) for the disc -, if one wants to calculate the magnitude —
of the pressure P acting between the plate and the disc.
Let the radius of the undeformed disc be R 十 e, where R is the radius
of the hole in the plate before deformation (and s is, of course, assumed
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 223

to be small, i.e., of the same order as the admissible displacements).

All terms (elastic constants, components of stress etc.) referring to the
disc will be marked with an index 0. For example, 吸 will denote the radial
displacements of points of the disc, while vr will refer to those of points
of the surrounding plate.
It follows from the conditions of the problem that, after insertion of
the disc into the hole in the plate, one must have along the common
boundary of the disc and plate
%—
The radial displacement e? * = e. (58.4)
of a point of the rim of the disc may be considered
—
to consist of the radial displacement ( s), necessary to reduce the radius of the
disc to R, and of the displacement vr, which it undergoes together with the point of
the edge of the hole with which it is in contact. Thus
follows (58.4).
—
= s + vr, whence

But by (56.7%) and (583)

_
-
PR2 P(XO-1>
--
o_
% 一
4r 4网
Putting in these expressions r = R and substituting them in (58.4),
one finds
PR P% 1)衣
2M
। —
4网
whence
4e科科0
P= (58.5)
+ w(乂° — 1)]

and the problem is solved. (Note that in actual fact one should have
put 夕 = 天十 e for the points of the disc, but in view of the magnitude
of s this is of no importance) .
In the case of an absolutely rigid disc one will have, instead of (58.4),
(584)
and hence, proceeding as before,
2然
*一
= R (585)

The same value of P would have been obtained by putting in (58.5)

Mo = 8 and assuming x0 to be finite.
224 in. SOLUTION BY SERIES §58

In this case (56.7 'a) gives for the plate

夕夕 = 2
产 82 —, 99 =— -
一

(58.6)
sR
力丁 , % = 0.
r

2°. Stretching o f plates with inserted o r a t-

t a c h e d rigid discs.
In § 56笈 (example i°), the solution was obtained of the problem of a
plate with a circular hole of radius R under uni-directional tension. The
functions <p(z), ^(z), giving the solution of this problem, may be rewritten
少,0五2\ 0/ y五2
中⑶ =
/ \

2 (z H 〉
--—W(z) ~
— —— H H—- F — , (58.7)

where
0 = 2, 丫 =1, 力=
— 1. (58.8)
The stresses and displacements, corresponding to these functions 平 (z)
and 少⑵，whatever may be the real constants 因 丫，6, are easily calculated
on the basis of (39.4) and (39.3) which give (cf. § 56a)

(58.9)

% = 1)”+2丫k2+ ［做比+1冰2+2/+ 泞
cos 2
(58.10)
2减7
一 1)R2 + 2严 —} sin 2分.
82 J

If the constants 0,丫，6 have the values (58.8), one obtains the earlier
solutions of the problem of tension of a plate with a circular hole. By
allotting these constants other (real) values, one may solve some problems
which are of equal interest. Thus, for example, it is easy to deduce the
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 225

solution of the problem of the stretching of a plate with a circular opening,

cut before deformation and filled with a perfectly rigid disc of the same
radius R.
First, suppose that the rigid disc is joined to the surrounding
plate along its edge. It may be assumed that the rigid disc is not dis¬
placed during the stretching of the plate ; otherwise it would be sufficient to
subject the entire system to a rigid displacement, in order to return the
disc to its original position. Hence the conditions of the problem are
vr = 0, 修心 =0 for y = R. (58.11)
The problem will be solved, if one succeeds in choosing the constants
8, Y， figuring in (58.9) and (58.10), so that (58.1 1) is satisfied. By (58.10),
the conditions (58.11) give
乂
— 1 十 2丫 = 0,
whence one finds
(乂 1 )8 + 2 + 28 = 0, (x — 1)0 + 2 — 26 = 0,

2 x— 1
p=
x
, y = 一一
2 - (58.12)

or, remembering that 乂 = (入十 3“)/(入

2( 入 办 . 入十 (X
—— X + 3fx
科 受
Q
•，二 ‘‘
O (58.12')
入 4" 3fl 入十
(This problem can also be solved for the case when, in addition, ar¬
bitrarily given forces and couples act on the rigid disc; cf. §830.
Next consider the problem when the disc is not joined to the plate,
but only inserted into the opening, under the assumption that there
is no friction between the disc and the surrounding plate. Instead of
(58.11), one has now the conditions
vr = 0, 3= 0
夕 for y = R> (58.13)
since it may no longer be postulated that % = 0 at the edge of the plate,
because points of the plate there are free to slide on the rim of the disc.
As in the last example, it is easily verified that (58.13) will be satisfied, if
one puts in (58.9) and (58.10)

p —
- 3x
4
+ l
_
r
乂

2
1
'
4—
—- % 1
3x *4- 1
( SR 1 4A
1

— 1

or
_2( 入 4- 时
j
2入 + 5“
， Y1 =
L
入
+N
. o
2 入 + 5“ ' —
. (58.14) ——
226 HL SOLUTION BY SERIES §58

However, it must be noted that the first of the conditions (58.13)

assumes that the material of the plate is in close contact with the disc
all along the common boundary ; if this assumption is modified, the
problem becomes considerably more difficult *. It is easily verified that for
the values of 丫，3, given by (58.14) and determined under the above
assumption, the normal stress rr becomes positive over certain parts of
the common boundary, i.e., the disc does not press on the surrounding
material, but pulls it away. However, this is physically impossible,
because the disc and the plate are not joined to each other. In order to
make the problem physically possible, it is sufficient, for example, to
suppose that the radius of the rigid disc is somewhat larger than was the
radius of the opening before the stretching of the plate and before the
disc was inserted. The solution, corresponding to this supposition, is
obtained by superimposing the preceding solution on that given by
(58.6) . One has, of course, to take £ so large that the composite solution has
〃 《 0 along the common boundary.
3°. Stretching of plates with inserted o r a t-
t a c h e d elastic discs. The preceding results will now be gen¬
eralized to the case when the disc, inserted into the opening of the
plate, is also elastic, but not of the same material as the plate.
An attempt will be made to satisfy the conditions of the problem, as¬
suming that in the region occupied by the plate (i.e., for 夕〉R) the
elastic equilibrium is determined as before by the formulae (58.7), viz.,
y蔺M4
—— (
P /
*?(z) — £ Z "I- 一、
中 (z) = H [
4 / \ Z Z

and that the equilibrium in the region, occupied by the disc (i.e., for r<R)
is governed by
p
中o(z) = — 4
(58.15)

where 国 Y，覆 Bo, Yo, 诟 are real constants, subject to definition.

This method of solution of the problem displays (outwardly) an artificial character,
since the form of the solutions has been partially guessed beforehand. One could,
of course, have eliminated any artificiality by using infinite series instead of
(58.7) and (58.15). In that case one would have found for the solution of the problem
that all coefficients of the series, with the exception of those retained above, must
vanish. This observation also applies to the other problems treated in this section.
* Such, a problem has been considered in a recent paper by M. P. Sheremetjev [1].
CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 227

The stresses and displacements, corresponding to the functions 中⑶

and 少 (z), are given by (58.9) and (58.10). Those corresponding to 佻(z)
and 狐⑶ must be calculated from the formulae of § 39 which give
力
[!^0 + 瓦 cos 29],
380 —
〔
微 Bo+ 浮一
瓦》05 9], 2 (58.16)

就
° 春 理一
说)sin 2也
and

吟 在｛仇网
即。I
一 1) 『L过梦 代
产 2301 cos 23%J
」

_— W JI Yo(乂 + 3) 0
(58.17)

酝
Let it first be assumed that the disc has been welded into the hole and

_
that the radii of the disc and the hole were equal before deformation.
Then the following boundary conditions must be satisfied :
=〃， 心o 心， = 许， 噱 = 为 for y = R. (58.18)
Substituting in (58.18) the expressions (58.9), (58.10), (58.16) and (58.17),
one finds the following equations for the determination of 丫，5, 仇，丫。,
说：
。，
8o(乂
Ro = 1 - y， 瓦 1 — 253 — 35, 3丫0— 说 — — 1 p — 33,
o — 1)
Mo
~

— 1 + 2丫 YoRo 3) 十 2S0
比

k '
1

—Mo
乂十 1)8 + 2 + 2S _ — i

M
'

Yom + 3) — 2说 一
(比 一 • 1)8 ,+ 2 —— 2 成

Ro 2

Solving this system of equations, one obtains

8
2(国 一 k) N(比o 1 )_ No(乂 一 1 ) 1

— § =
Mo — W
+ Mo比 '
“
2囱+ fx(x0 1) '
— r + Un 比 '
(58.19)
内)(" 1) c 内 + 1)
饱 万 —
7 N，yo = °， *说 一
一^

以比0 1

1) k + 阿比
228 HI. SOLUTION BY SERIES §58

Consider now the circumstance that, because of the relation y0 = 0,

the functions g()(z) and %(z) characterizing the elastic equilibrium of
the disc are linear:
力 力
a o(z) = — 8庐， 小o(z) = ~

this means that the disc undergoes homogeneous deformation. In rectan¬

gular coordinates the stress components will be constant ; in fact, it
is easily verified that

x?=少智 还， 冲 = 少 当；@, 鹫 =0・ (58.20)

2 z

In the direction of the Ox axis the disc is subject to tension, while in the y
direction it suffers tension or compression, depending on the sign of 仇 一r 瓦.
In the limiting case
“° 。
= 8 (perfectly rigid disc) , one obtains for
8, 丫，8 the values (58. 1 2) ; in the limiting case &) = (no disc), one obtains
for these constants the values (58.8). Finally, if 廿 = 回， = x0, one is
dealing with a continuous homogeneous plate. In this case (58,19) ”
shows that § = 丫 = 5 = 丫。= 0, 仇 = 瓦 = 1 and that the functions
中 (z), 少 (z), characterizing the equilibrium of the plate as well as of the
disc, are given by

(58.15〃)

as had, of course, to be expected.

Next consider the case when the disc has been inserted into
the opening in the plate, assuming that the radii of the disc and of the
hole were the same before deformation and that no friction is present.
Obviously the boundary conditions have the form
〃° ~ rr, 心°= 0, 心 = 0, 吟 = vr for y = R. (58.21)
Substituting in (58.21) from (58.16), (58.17), (58.9) and (58.10), one finds

d =1一y，说 =1— 2B - 3次 3Y0 — 说=0, 1+3 + 36 = 0,

8o(y0 ——1) 2
一1 十丫 Yo3o —— + + + +
3) 2说 B(x 1) 2 28
—
乂

Mo M No 从

Solving these equations, one obtains

CHAP. 9 SOLUTION FOR REGIONS, BOUNDED BY A CIRCLE 229

= 2
十 3)
— 2Mo — 1 ) P-(xo (X - 1 )
一网

“(比o 十 3) 十 网(3% + 1) 2Mo + — 1)

—— N + + 1）
(Kq 3) 乂 一 1) R - 网（比
园― (58.22)
+ 3) 网(3乂 1) 工网十 山乂0—1）'
=
2网(比 1) >
6处 +D .
J(乂o 十 3) 十 “o(3比 十 1) “(比o 十 3) + 内)(3乂 1)
As is easily seen, the values of and rr will be positive on parts of
the common boundary, and this is physically impossible. The problem
can be made physically possible by superimposing on the present solution
that of problem 1° *.
By putting in (58.22) [i。= 0 or = oo, one finds for 8, 丫，5 the values
(58.8) or (58.14) respectively.

* The (more complicated) case when the disc may lag behind the surrounding
material was considered by M. P. Sheremetjev [1].
 Chapter 10

THE CIRCULAR RING

§ 59. Solution of the first fundamental problem for the circular
ring.
A solution, using definite integrals and differing from the one to be deduced here,
was published by G. V. Kolosov [5]. S. G. Mikhlin [8] (using power series) solved the
somewhat more general problem where the ring consists of two concentric rings
with, different elastic constants, under the supposition that they are joined along
the common boundary. In particular, the inner ring may be a continuous disc.

Consider the case when the region S occupied by the body is a circular
ring, bounded by two concentric circles and L2 with radii and R2
伍i < 尺2) and centre at the origin. Let the external stresses acting on

_—
—
and L2 be given, i.e., the values of rr 勿夕8 on Lr and L2 as functions
of the angle 3. Expanding this expression for Llt as well as for L2, in
complex Fourier series, one will have
4-8
rr — E 心 = E A,；
j

on Llf
—8

(59.1)
rr —
——
勿心 = £
8
A[ "心 on L2.
The boundary conditions may then be written (cf. § 56)
•
48
S 4尹3 for 夕 =
(z) +函一 外吨5*(z) 子 一°°
(59.2)
I S aqk" for n 扁.
By (35.2),
I
-OO

(z) = A log z + *(z),

where 幺 is a real constant and is holomorphic inside the ring, so
that it may be represented by a Laurent series. The function T(z) is
holomorphic in the considered region (§ 35) and hence may likewise be
expanded as a Laurent series.
230
CHAP. 10 THE CIRCULAR RING 231

Thus, inside S,
d-8 十8

(z) — A log z 十—Z «庐乃, 子⑵ E a概比. (59.3)

8 一
The requirement of single-valuedness of displacements is expressed by
(35.7) which, since there is only one internal boundary, becomes
月 = 0, 乂笈 _]
+ 巨—i = 0- (59.4)
However, this condition will not yet be imposed, since the more general
solution has many interesting interpretations.
It will be remembered that the quantities 丫允 and 丫力 of § 35 were the
coefficients of terms of the form (a log z) in the expansions of the functions

勺 (z) = a，
/ (z)dz
and

小⑶ = J 甲(z)dz.

In the present notation these terms are a：】 log z respectively.

j log z and.

Instead of (59.4), assume for the time being that A is an arbitrarily

given real constant. Substituting from (59.3) in (59.2), one finds

2A log y 一 A + S (1 •~
国口渡％
话
a+
— oo
' 4-00
X 月；.净a for n = &,
九 1

+ S akrk e~ik^ — S 或
一
_2 娱一2 /合 °° (592)
4- oo
OO —
*— OO »7 »

E for 夕 耳
、 一
Comparison of terms independent of 卧 gives
24 log & ― A '
2ao — 勿_2芯？ AQf — (59.5)
2月log x2 4 — + 2ao 说一 = 4()；
~~

here the assumption has been made that a0 = 巨 i.e., that a0 is real, which
° part of (z) has been
can always be done, since any constant imaginary
shown not to influence the stress distribution.
Comparison of terms involving for 为 = ± 1 , ± 2, .. . gives
(1 一 用以鹰 + 乙小L 一 矶2巡-2 = 4,
(59.6)
(1 -电即礴 十 汇然言 一 卷_2魔一2 晶•
232 m. solution by series § 59

Eliminating from (59.5), one finds

_ 解段一 4理 4 4段 党&一段 iogj) 小•八
2 一
工
曲一 十
2(建一 段) 马一招
Since 曲 is real, it follows that
(59.8)

[If a0 had not been assumed to be real, one would have had on the left¬
十
hand side of (59.7) 电 instead of a0 and (59.8) would still have
2
been valid.] A simple calculation shows that (59.8) expresses that the
resultant moment of the external stresses must be equal to zero.
Next the remaining coefficients will be determined. Dividing the
first equation of (59.6) by R铲，and the second by 璃"， and subtracting,
one obtains the first of the following formulae :
(1 一国 (凡一 ^)ak + (五/+2 一 区y+2)l= Bk) (59.9)

where
依飙 2
— R/2) 瞅 + (1 +南阳 一 段)乙 及九，
Bk = 4依产 2 一 幺短产2； (59.10)
the second equation (59.9) is obtained from the first by replacing k
by —々 and by going to the conjugate complex expression. (It will now
be sufficient to consider (59.9) only for 论 = 1 , 2, 3, . . since for / = 1> —
— 2,
former).
—3 one obtains a system of equations which is conjugate to the

For any given value of 万，the system of two equations (59.9) will de¬
termine ak and a_k, provided the determinant

一
(1 — 用 (周一 招) 区清+2 —应 2A;+ 2

一
Rlk+2~~Rlk+2 (1、+ 团 (周 一 4)
= (1 _收) 段一段)2 一饵能+2— 五y+2)(R产
does not vanish.
+2
— 后2E+2)(59.11)

The determinant D vanishes for 々 = 0, ± 1, and it is easily verified

that for all other values of k it is different from zero. The value 万 = 0
is of no interest. For 6 = 十 1, (59.9) gives
0 = 鸟， (段 — 瑞)的 2因 一 五源t 更 r(59.12)
CHAP. 10 THE CIRCULAR RING 233

For 万= —
I, one finds two equations, obtained from (59.12) by
transition to the conjugate complex values. Thus, for the problem to
be possible, one must have, in addition to (59.8),

Bi = 禺邑 一 W& = 0. (59.13)
A simple calculation shows that this condition gives no new information,
since it states that the resultant vector of all external forces must
vanish.

-
To verify the earlier statement regarding the values of the determinant (59.11),
consider
P 号城)，
where

e = (韵 > 1
and
/© =
It is easily verified that
(I — — 也 1)2 + 守+1 + —许+ 1.

/⑴ = 尸⑴ = 广⑴ = 〃'⑴ 0，

If ] 克
/"(E) = 伊+ 1)机 A 1)[(A 2肖Y
2, the last expression is positive for
— —
> 0. Thus, for
十 2纥i力.

> 1, one will have

广熊) > 0, /〃© > 0, > 0, 堆) > 0. q.e.d.
To show that (59.13) is the condition, that the resultant vector of the external
forces must vanish, consider first points of the outer circle. It is easily verified that

Xn iYn = 净，
—
Xn iYn = ("一 E y联 "i论
y〃)the resultant vector of the external forces, applied to L2,
on L2.
Denoting by
one has
2k 2n

X〃一 = / [Xn —iY^R^ = & / 方 c屈小初

0 0
一 9 =

by definition of Similarly, one has for the inner circle

¥ — W=— 2吟 q.e.d.
When (59.13) is satisfied, the system of equations (59.12) becomes
possible, although it does not permit calculation of both the coefficients
口 and 々t； thus one of them may be chosen arbitrarily, neglecting for
the time being the condition o/ single-valuedness o/ displacements.
234 HI. SOLUTION BY SERIES § 59

即
__
All the other coefficients 瞅 (后 = ± 2, ± 3, . . .) are found by solving
(59.9). For any given k, one determines simultaneously ak and d^. In
fact, (59.9) gives

(1
(1 +电 (尼一 尼) Bk -(r/ + 2 一 尺产，
论2)(尼
—
2)方j
—
段产 一 (五铃 2 号右十2)(扁2加+2 网2A+2) —
伊= ±2, ±3,•.) (59.14)
and 互—加 is obtained from this formula by replacing 为 by
—
k and by
transition to the conjugate complex value. Thus all coefficients ak have
been determined for 为卢 0, ±1.
Finally, the coefficients a'k may be found from one of the two formulae
(59.6), with the exception of which can be calculated from one of
the equations (59.5). Since all akt with the exception of 女卜 have already
been determined, all except for 球 and a%, can be calculated in this
way.
Now the condition o/ sin^le-valuedness of displacements will be in¬
troduced, i.e., condition (59.4). One then finds, by (59.7),
短期 一 4期
0-
2(段 一号)
The coefficients a_i， are determined by the second equation of (59.4)
and, for example, by the first equation (59.6) for 袤 = 4- 1, which gives

_
(59.15)
Solving (59.4) and (59.15) for and a」，one obtains

a~i = 1一 ।
， a-i ~ '
一!1―। x ； (59.16)
乂
+
finally, one finds from (59. 1 2)
瓦1 2月；&
曲 — 段一熠 U 十乂)(段 +R犷— (59.16')

The formulae (59.16) could have been written down immediately, using
(35.9). Thus all coefficients in the expansions for ¥ and have been
found ; in particular, a[3 can now be calculated from (59.6), because
% and j are known.
Note with regard to the convergence of the above series that the
series for ©(z), '(z) and 乎卜) will obviously be alsolutely and uni-
CHAP. 10 THE CIRCULAR RING 235

formly convergent in the ring (including the boundaries), if the following

series converge :
oo 00 co
上 I分 I 段， 「
^k\ak\ 璃 鹰，
无 =1 R=1
oo 8
然.⑺
Z ! 0一X I 扁 ^k\a_k\ RfT, 2 1 心|中
九=1 k=l

Convergence of the latter series will be ensured, if it is assumed that

the quantities rr and 夕眇，given on 工】and L2, have second order derivatives
with respect to 9, which satisfy the Dirichlet condition. In fact, the co¬
efficients A'k and 端 of the series (59.1) will then satisfy inequalities of
the form (§ 53)
C C
। (论 = ± 1, ± 2, …).
' '
V
那 |「
Hence it is easily concluded on the basis of (59.10), (59.14) and (59.6)
that the following inequalities will hold true for k = 1 , 2, 3 :
C C C C
I 做 I 璃 v与 R
•， I ak I 璃 v 后， i a-k I 用'<
a R
I a-u I 片' v
R

whence it follows immediately that the series (59.17) converge.

The second fundamental 'problem can be solved in a similar manner.
If one compares the solution deduced here with that obtained by
application of Airy's function, the advantage of the introduction of
functions of a complex variable becomes obvious.
With regard to the above considerations of convergence of the series, it may
be shown, as in § 56, that, if one uses the boundary conditions in the form (41.5),
it has to be assumed that Xn and Y n have on and L2 first order derivatives,
satisfying the Dirichlet condition.
The Airy function is used in A. Almansi [2], J. H. Michell [1] and A. Timpe [1]
for the solution, among others, of the first fundamental problem. Timpe has the
same expression for the boundary conditions which is represented by (59.2')
above, and he deals with it at great length. The coefficients are determined, eight
at a time, from systems of eight linear equations with eight unknowns.

§ 59a. Examples.
1°. Tube, subject t o uniform external and
internal pressures.
Let the internal and external circles be subjected to uniformly dis-
tributed normal pressures 力】and 力2, so that〃= — 力工 on Lv〃= —
236 III. SOLUTION BY SERIES §59a

on 工2>/^ = 0 on 2・In this case

上

4= 一九 月；=一%.
All other coefficients A] A‘[ vanish. The condition for the existence of
a solution is obviously satisfied. Formulae (59.7') and (59.5) give
孤周 一力君 段段
仇 一%）
2（此一号）'
J 一
用—皮 — (59. U)

一
All the other coefficients ak, ak are zero. Thus
%段 一％段 的一如尼段 1
彳（590）
.
0 一 一 2(段一 段)， 里仅) 镇一段
The polar components of stress are
密段 1
-少因 ， （％ 一 %）
段一段 段一号
%尼 一夕阳 物一 煽瑞尼 1 (59.3«)
忌一号 Rl~Rl 夕2,
心 0.
This problem was also solved by G. Lame.
2°. Stress distribution i n a ring, rotating about
its centre.
Let the ring rotate in its plane about 0 with a constant angular velocity
s and let no other external forces act on it. Let the system of axes
rotate together with the body and hence be fixed relative to it. Then the
problem reduces to a static one, the applied forces being centrifugal in
origin.
One of the particular solutions of the equations of equilibrium is given
by the formulae of § 28. The stresses, given by (28.6), are easily found to
have the following polar components :
2入 + 3r … 2入+
—
pi e
rr = pwM, = p3?夕2 夕9 = 0.
40 + 2内
•—

4(入 凶 2
If one wants to apply this solution to thin plates (§ 26), one has to replace
入 by 入*, so that
2入*
+ 3{2 3 + b 2入* + fi +
_ 1 3a .、
而 = 8—
. . 勾
「
1

4( 入* + 2 8 ' 4( 入*
CHAP. 10 THE CIRCULAR RING 237

The stresses (59.4a) do not satisfy the boundary conditions on the

edges of the plate. In fact, one has there 夕9= 0, but〃takes constant
values which will be denoted by % and %.
The solution of the present problem is obtained by superposition of
the stresses (59.4a) and (59.3a) for
2入 _ 2入 十 „ a

+U
力= —
9
夕 = 「 超.32 (59.6)
4( 入 + 2[1) 2
4(八入 工 2{1)y
The problem is thus solved. In the case of a thin plate, 入* replaces 入.
The solution, obtained in that manner, gives only mean values of the
stresses. For not very thin plates this will not be sufficient. (For more
complete solutions, cf. A. E. H. Love [1] § 102). When = 0, one
finds the solution for the case of a solid rotating disc.

3°. Certain generalizations.

From the point of view of engineering applications considerable
interest attaches to the solution of the more general problem when the
body under consideration consists of several concentric rings with
different elastic constants which are either welded together or are in
contact with each other along the interfaces, where the outer circle
may be extended to infinity and the internal ring may be reduced to
a solid disc.
Such problems can be solved with the help of series in exactly the
same manner as this has been done in the preceding section.
The case when the body consists of two concentric rings which are
welded together was solved by S. G. Mikhlin [8]. More general problems
of this type were studied in a sequence of recently published papers by
G. N. Savin, D. V. Vainberg and other authors. A description of large
sections of this work and a listing of the corresponding literature has
been given in the monographs by G. N. Savin [8] and D. V. Vainberg
[1], and this makes it possible not to dwell on this work here.

§ 60. Multi-valued displacements in the case of a circular ring.

Consider now the general case and study the results, given by the
formulae of § 59 when one omits the conditions of single-valuedness
of displacements which are expressed by (59.4), viz.,
4=0, + 无—i = 0 (60.1)
If these conditions no longer hold, the boundary conditions (59.2)
238 m. SOLUTION BY SERIES §60

are not sufficient for the complete determination of (z) and 甲(z);
some of the coefficients in the expansions for these functions remain
indeterminate and a known number of arbitrary constants is retained
about which more will be said below. By fixing these constants in an
arbitrary manner, one obtains definite expressions for and ¥ which
satisfy all the conditions of the problem, except the condition of single-
valuedness of displacements. In fact, if one describes a closed path Lf,
starting from some point z, passing around the inner circle in an anti¬
clockwise direction and reaching again z, one finds that the increase in
u + iv for one circuit of this path is given by
[w + 河 z/ = —
p,
{(x 十 l)4z 十 比立 1+ 万-J；
一
(60.2)
this follows from (35.6), using the notation of § 59.
It has been seen in § 45 that, in spite of the multi-valuedness of
the displacements, such a solution may be given a definite and very
simple physical interpretation.
Firstly, this solution makes sense in the ordinary way, if one does
not apply it to the complete ring, but to a part of it obtained by re¬
moving from the ring a strip, bounded by two lines a' 6 and 废'6〃 con¬
necting the inner and outer circles (in Fig. 29 the removed part has been
shaded). One then has a simply connected body, i.e., a "curved beam”,
〃
bounded by two circular arcs and the lines 屋 and 诡巧". In this simply
connected part the functions u, & are single-valued. The functions and
T correspond to some definite state of elastic equilibrium of the beam
for which the external stresses, applied to the circular boundary, have
known values, i.e., those which appear in the boundary conditions of
§ 59 for the solution of the problem
of the continuous ring. As regards
the external stresses, applied to
the ends a'b' and 相 they may
be calculated from the functions
and 甲 by the help of the previ¬
ously deduced formulae. The pro¬
blem of equilibrium of a curved
beam will be considered in § 61.
Turn now to the case of the
continuous ring. It has been seen
in § 45 that the solution considered
here, admitting multi-valued dis-
CHAP. 10 THE CIRCULAR RING 239

placements, corresponds to a particular mode of deformation, which is

called dislocation. This type of deformation will now be described in as
far as it applies to the present problem, and, in parts, this will involve
repetition of what has been said in § 45.
Produce a cut ab, joining the inner and the outer circle, and denote
—
the sides of this cut by (+) and ( ), as indicated in Fig. 29. Then, for
a circuit along a closed contour L' starting from some point (%, y), con¬
—
sidered to lie on the side ( ), and ending at the same point, but now
conceived to lie on the side (+), the components of displacement undergo,
by (60.2), the increases
%+ — %- = — ey + a, e+ — 厂 = (60.3)
where
e4(1 + x) , ni ,
e = a + 平 = 一 (乂讥T + 港J ； (60.4)

(%+,o+) and ("i, 厂) denote the displacements of the point (招 y), con¬
sidered to lie on the sides (+) and ( ~~ respectively.
In accord with the statements of § 45, the multi-valuedness of the
displacements in the present solution may be interpreted by means of
the hypothesis that before deformation a (small) transverse strip with
sides 0nb‘‘ (see Fig. 29) had been removed from the ring and that
the free edges had been joined. For this purpose it was assumed that
before deformation the ends a'br and were congruent and disposed
in such a way that a"b" is obtained from 诡6' by a rigid displacement,
consisting of a rotation e about the origin and a translation (a, 8). When
joining the free edges together, those points which would correspond to
each other, but for the above-mentioned rigid displacement, must be
combined.
Note that, as indicated in § 45, the quantities z, a, p do not depend
on the shape of the cut db nor on its location in the ring; in the present
case this follows immediately from (60.4). Thus, the transverse strip
which must be cut from the ring before deformation may be taken from
any part of it; one of its sides, for example 屋勿，may be given any shape
and location, and the position of the other side will be determined by the
quantities g, a, p.
It should be remembered here that it is only for the sake of convenience
that reference has been made to ''removal'' of strips, since in practice
one has often to "add” material in one part and remove it in another.
240 III. SOLUTION BY SERIES §60

The quantities s, a, 3 represent, in the terminology of § 45, the charac¬

teristics of the dislocation. According to § 45, knowledge of these quantities
and of the external stresses, applied to 工】 and L2, will completely de¬
termine the deformation of the body under consideration. In the present
case this fact is verified directly, because it is easily seen that, if those
quantities are known, all coefficients in the expansions for (z) and
甲⑶ are determined (except for the imaginary part of a0 which is un¬
important) ; in fact, these coefficients may be calculated as in § 59, the
only difference being that (59.4) is to be replaced by the more general
condition (60.4) for given values of % a, &
The coefficients in the expansions of the functions (z) and 甲(z)
will now be determined for the particular case, when no external forces
are present (i.e., N = T = 0 on J and 乙). Then all A'k and 力怖 vanish
and (60.4) gives
,
—-
3 [L
力 =— ... , 比
+ 立_] r (a + 用). (604)
九(1 十 凯) 九2

Together with the equation 左

second equation of (60.4') leads to
— = 0, obtained from (59.15), the

T
“(a + 小)E
m(1十比)'
, 邙
兀(1十M)
— m (60.5)

Further, (59.7) gives

他一
而下外
增 归段 logj
r(1 十
log 号)
尼一麴
，—m — (60.6)

and, finally, by the second equation of (59.12), by (59.5 and (59.6)

(for 々 = 1), one obtains
2M（a 一 巧）勿 , 尼尼 1 -^2
（1 + 乂）因 超）’
兀 兀（1 十 乂）（段 一 曙）
（60.7）
, 2N3 + 邙% 居尼
3
一.』t丁不用
All remaining coefficients are zero, and hence

-
/ / /

(z) = A log z 十 曲 + a-^z +

z — 乎⑵ = — 4 梳 H
z z z — — (60.8)

where the coefficients have the values stated above.

CHAP. 10 THE CIRCULAR RING 241

In the particular case, when e = a = 0 = 0 (i.e., when the displace¬

ments are single-valued), one has (z) = T(z) = 0, as was to be ex¬
pected, because it is known (§ 40) that, if the displacements are single¬
valued, no stresses occur in the absence of external forces.
If external forces are present, the corresponding solution may be
obtained by superimposing the solution just found on that of § 59 which
was deduced under the hypothesis of single-valuedness of displacements.
As has been stated in § 45, the interpretation of multi-valued dis¬
placements in the case of the circular ring was first given by A. Timpe [1],
who also found formulae equivalent to those deduced here.
Turning now to the dislocations, corresponding to (60.8), it is noted

。.
that one may distinguish between the following three simple cases :
1 e 于 0, a = B = 0. This dislocation is obtained, for example,
by cutting from the ring a radial sector with straight edges, forming
the angle s, and by joining the ends.
2°. s = 0. a 0, B
— 0. This dislocation is obtained, for example,
if one cuts the ring along the positive Ox axis, slides the lower against
the upper edge by a distance a and again joins the contacting parts.
The same dislocation is obtained, if one removes along the positive
Oy axis a strip of thickness a and rejoins the ring by displacing the
free edges parallel to the Ox axis. In the latter case, when a > 0, a strip
must be added.
3°. e = 0, a = 0, g W 0. This case follows from the preceding one
by interchanging the parts played by the axes 0% and Oy.
Thus, it will be sufficient to state the formulae referring, for example,
to the cases 1° and 3°. The expressions for the functions and T and
for the polar components of stress will be stated here. They agree with
those, obtained in a different manner by A. Timpe [1] ; the method used
here was taken from the Author's paper [1]. Note, however, that in the
following formulae x has been replaced by
入
+ ——
3M •

1°. （& W 0, a =
一^^
B = 0）:
同
叩（入 + k）（ 1
=2 （入 +面 Q
加
段 log&—
四 居 一
j 十
qx（入
2M 入 十杀
+ m）

）
中, ）
昨
部（入3瑞超 Rz 1 1

而语二商现瓦卡
242 III. SOLUTION BY SERIES §61

- W(入 + NLVog T, 1 超段 1log & m logj一瑞—

log & ]
-，
加而 声呼 F 尺 璃二用
声— w(入+ L R遇—loggg—
(60.10)
居一兀(入十 2
.
F°g” r^'Rl-Rr^^
R^ogR.-R^R, J
济 =0. 酸二用 」卜
3°. (e = 0, a = 0, 8 W 0):
M入 + [ 2z 1
I
(60.11)
5
Rz) 、= 肌 J1
h 上
招MI
./)'
1
苏宁布 Ef
- M(入+2"
兀(入 2囚 1
旷

R[ + Rl^ r
上
i 册或
居十 彪
11sq
74
加（入 时 3y 号用 .1] 1
(60.12)
兀（入 十 2 I 瑞 氏 十夕十 号+段 济 j

方 m 入 办[ 旷 1 r酒n sm. .
标下而 L 研小密 7 存F h "
心 上
一 .

In the case of a thin plate (§ 26) 入 must be replaced by 入*.

Hitherto it has been assumed that no external loading is present.
If arbitrary external stresses occur, these solutions have to be combined
with those of § 59 for single-valued displacements.

§ 61. Supplement. Bending of a curved beam by

forces, applied t o the ends, for arbitrary d i s-
tribution of external stresses o n the curved
boundaries.
Let it be assumed that one is dealing with a part of a ring, bounded by
two radii. First, let the curved boundaries be free from external loading.
The solutions, obtained in § 60, satisfy, of course, all equilibrium conditions
and give zero external stresses on the circular boundaries. The dis¬
placements will be single-valued in the region considered here. However,
the stresses, applied to the straight edges (ends) of the beam, will be dif-
CHAP. 10 THE CIRCULAR RING 243

ferent from zero and will depend on the three constants a, 0, s. Generally
speaking, it is impossible to choose these constants, so that one obtains
at the ends a given external stress distribution. But, as will now be
shown, one may always arrange that the stresses, applied to one of the ends,
will be statically equivalent to a given force and couple, i.e., that they
have a known resultant vector and moment. The forces, applied to the
other end, will then be statically equivalent to a force and couple, op¬
posite to the former.
If the length of the beam is large compared with its width, the given
resultant vector and moment of the forces, applied to an end, may be
replaced by a fictitious distribution of forces, using Saint-Venanfs Princi¬
ple (§ 23). In the sequel, when speaking of the force and couple applied
to an end of the beam, this will refer to application of external forces
which are statically equivalent to those given in the problem. For
example, it may be assumed that one of the ends is clamped ; then, due
to the clamping, reactions will occur which statically balance the force
and couple applied to the other end.
Let the part of the ring, to be considered here, correspond to values
of 9 in the interval %《分 %.Consider first the solution 1° of § 60.
The resultant vector of the forces applied to either of the ends will be
zero. In fact, it is easily seen that, if 分8 is determined from (60.10),

/ 分眇 dr = 0.

The resultant moment of the forces acting on the end (per unit
thickness of the beam in the direction perpendicular to the plane
about 0 is given by
2

标 (周一 M)2 — 44底

叩(入 +*)
M = 2为/ AS r dr — h
2(—
z,,
n
J
&
段 一瑜 冗(入
+ 2jl)
where 2h is the thickness of the beam in the direction perpendicular to
the Oxy plane.
The solution of the problem of flexure of a curved circular beam by
couples, applied to its ends, is thus obtained by substituting in the
formulae (60.10)
244 in. SOLUTION BY SERIES §61

部（入 十 N _1 4M阳一础
(61.2)
（入 20 h （段 一 R；）2 4超段
冗

—
It is easily seen that the denominator on the right-hand side of (61.2) is always
positive ; in fact, let
2
阳— 础 2
— 4瑶期 (log 第 = 冲㈤，
where
氏2
备 1, /(%) = (%— 1)2 — %(log*)2.
But
/⑴ 产⑴ 广⑴ /〃，⑴ =0,
and
2 log 力
广'㈤

Hence, /〃'(％) 0 for 力 1, whence follows that /〃(％) > 0, /'(%) > 0, /(%) >0
"
for % > 1, q.e.d.
Next consider the solution 3° of § 60 and assume that the direction
—
of the coordinate axis is such that 眇2 兀/2. On the end 眇 = 8? one has
88 = 0, as is shown by (60.12) . Thus the resultant vector of the external
forces, applied to this end, passes through 0, is parallel to the axis 0y
and its magnitude (i.e., its projection on the Oy axis) is

P=
长
2h I 7^- dr =
(段 + 必) log
不
% — 段+曙 0M + 入 W)
(61.3)
J 号 段 a(入 23
Hence one can solve the problem of bending of a beam by a transverse
force, applied to the end 分 = ^2, by substituting in (60.12)
M 入 + a) j_ _ P（号 + 礴
(61.4)
亮一段
加(入十
2© 2为
（唐 凡）log +段
It is easily verified that the denominator on the right-hand side of (61.4)
is always positive [cf. remarks following (61.2)].
The problem for the case of forces normal to the end of the beam
may be solved in the same manner. The solution can either be found
directly, as in the preceding case, or by adapting this solution. In fact,
CHAP. 10 THE CIRCULAR RING 245

consider the part of the ring included between the radii 8=0 and 9 = 兀/2.
The preceding solution gives on the end 分 = 兀/2 a system of forces,
statically equivalent to a single force, parallel to Oy and passing through 0.
Consequently, the forces applied to the section 眇 = 0 will be equal
in magnitude and opposite in direction to that force, i.e., they will be
equal to a force, normal to the straight end 9 = 0; the line of action of
this force passes through 0. By adding a suitable couple (using the so¬
lution already found for flexure by a couple), one can always obtain
a force the line of action of which passes through an arbitrary point.
The preceding solutions of the problem of bending of curved beams by forces
and couples applied to the ends (and likewise for other types of loading) were
found by Kh. Golovin [1] ; Golovin's paper remained unknown outside Russia and
his solutions were rediscovered independently by several other authors.
Thus the complete solution has been found for the case when the
curved sides of the beam are free from external stresses. Now suppose
that these sides are likewise loaded in an arbitrary manner. The so¬
lution may then the obtained by the following method.
Imagine that the beam, be extended into a complete ring and impose
arbitrary loads on the curved boundaries of the added part, in such a
way, however, that these loads, together with those given for the curved
sides of the original beam, are statically equivalent to zero ; then solve
the problem for the complete ring by the method of § 59.
Such a solution will satisfy on that part of the ring, which corresponds to
the original beam, the known conditions on the curved boundaries.
There only remains to select the solutions of the present section in such
a way that one obtains, by their superposition, at the straight ends
forces which give the known forces and couples (where, of course, the
latter must be such that they statically balance the forces given on
the curved boundaries).
Note that by varying the loads on the curved boundaries of the com¬
plementary part of the ring, different solutions may be obtained. This
does not contradict the uniqueness theorem, because only the resultant
vector and moment, and not the stress distribution at the straight ends
have been taken into consideration. All the different solutions, mentioned
above, will correspond to different distributions of the external stresses
at the ends (which, however, give the same resultant vectors and mo¬
ments) . All these solutions, by Saint-Venant Js Principle, will differ little
from each other in parts of the beam which are not too close to the ends,
provided the width of the beam is small compared with its length.
246 m. SOLUTION BY SERIES §62

NOTE. It will be recalled that in the case of plane stress (§ 25) the
plane faces of the beam, parallel to the plane are subject to normal
stresses which are not arbitrary.
When, the thickness of the beam (in the direction perpendicular to
the plane Oxy) is small, one can assume that one is dealing with gener¬
alized plane stress (§ 26) and then these faces are free from external
loading. However, it must not be forgotten that in this case 入 must be
replaced by 入*.

§ 62. Thermal stresses in a hollow circular cylinder. Since the

problem of dislocations in a circular ring has been solved (§ 60), the
problem of deformation of a hollow cylinder, the transverse sections of
which are circular rings and which is placed in a two-dimensional
axially symmetrical flow of heat, can likewise be solved on the basis of the
results of § 46. Consideration will be limited here to one simple application.
The notation of § 46 will be retained. In the present case the region S
is bounded by two concentric circles with radii R2 < A2) and
the origin as centre.

—
Suppose that the hollow cylinder under consideration is heated by
being placed in a heat flow and that T = for 夕 = & and T — T2
for 夕 = A?，where and T2 are constant and r is the distance of a
point (%,y) from the origin. Then, as is easily verified (cf. Note at end
of this section),
7 —
T2 f
—
log% log & ° ~ log 邑 一 ％ log &
—
log R2 log R±
Hence, denoting by F the same quantity as in § 46 and omitting
例• n

the imaginary part of an arbitrary constant, one obtains

F(z)
二诲" (622)

Thus, in the present case, one has by (46.6), omitting again a constant,

72 —
然* zo* = —log
—邑 — log
= Z
i
log Z +*
&
马(lQg%+ 1)—T2(晚&+ 1)
方， (62.3)
lo久二 log j
and it is seen that, since there is only one internal contour £】，(cf. § 46)
CHAP. 10 THE CIRCULAR RING 247

垃 =偌=0. (62.4)

This means that the solution of the "auxiliary” problem of § 46 is ob¬

tained by substituting in (60.9) and (60.10) [cf. (46.16)] the value
7W
一

Since the stresses Xx, Yy>

入十“
'
log；?2 — log 2?/
in the auxiliary problem are the same
(62.5)

as in the original one, these stresses are obtained directly from (60.10) by
substituting for s the above value. In this way a well known formula has
been obtained (cf. for example A. Foppl [1]). In order to calculate the
displacements, one has to find the displacements ur and o' of the auxiliary
problem. Then u and v will be given by (46.8) and (62.3).

NOTE. Several additional remarks will now be made with regard

to the present problem, i.e., to the case when the cross-section is a cir¬
cular ring.
If the temperature distribution T is not given, but only its values on
the circles L、and L2> it may be calculated in the following manner.
The problem of finding T is a particular case of the so-called first fundamental
problem of the theory of the logarithmic potential (Dirichlet problem) ; it consists
of the determination of a function (i.e., 丁)，harmonic in a region, for given values
on the boundary. It may be shown that this problem has always a unique solution
(under very general conditions) . The general solution of this problem for the case
when the region is a circular ring is given below.

For the determination of F(z) one has

2T = 尸⑶ 硝 (62.6)
and, by (46.12) putting z1 = 0,
4-oq
F(z) = 4 log z + Eooakzk, (62.7)
—
where X is a real constant.
The function F(z) must be determined for the boundary conditions
F(z) 十 F(z) = 2/i(9) for r — Rlt
F(z) 十丽 = 2^(9) for / = &，
248 in. SOLUTION BY SERIES §62

where 九(8), 为母) are the known values of T on and Let these
functions be represented by their complex Fourier series
8
十 +8
AW = £ 左尹a, % = s 月评吗 (62.8)
-
8

where by § 52, since the functions 九 and /2

—are real,
8

and, in particular,
，一万，
o一 a
々 Zig 。, _
— 万"
4°,
i.e., 月; and 4；are real.
The boundary conditions may now be written
+ 00
2 S 月渥怖 for = Rlt
+8 +8
——
—
8
2A log 夕 十 £ 秋*法碱 十 £ a^erik^
H-OO
—'OO — 00
2 E 晨尹3 for r
—8
—
Hence
24 log & + 曲 十 诙 = 2/j, 2A log & + 曲 + 篇 = 24；, (62.9)

次戏+匹s丁 =24， 收超 +工小萨 =24 他 字 0)；

(62.10)
the equations (62.9) determine A and aQ 匹,while each pair of equations
(62.10) gives ak and 无_兀, where it is sufficient to give k only positive values,
in order to find all the coefficients. The imaginary part of a0 remains
indeterminate, as was to be expected, and may be fixed arbitrarily.
For example, if T = 7\ for # = R】 and T
and T? are constants, one has
—T2 for r = R2, where

4 = 71， 4 = C，匈 = 左 = (论 N o),

and F(z) is given by (62.2).
Note the important fact that multi-valued terms in the function
/ F(z)dz can only originate from the term A logz and the term 口_卢-1
in the expansion (62.7). However, as shown by (62.9) and (62.10), the
constants A and *1 are determined solely by Aq, Aq, and 4；. Con¬
sequently, the characteristics of the dislocations for the auxiliary problem,
and therefore also the stresses Xx, Yy, Xy in the original problem, depend
solely on the quantities
CHAP. 10 THE CIRCULAR RING 249
2k

0 [
= 2k /2
J
0
2江 2TV

*；=去//1" 0
"机 题=之//式3%分, 0
i.e., on
2 2* 2n

0
力（眇）&8,
0
/
〃0
i（3）cos3dA,
〃 0
2 cos

2tc 2 n:

J /式眇）sin »«8, j 似心）sin»d*

0
0
 Chapter 11

APPLICATION OF CONFORMAL MAPPING

It has been seen in the earlier chapters of this Part that the use of
power series for the unknown functions leads to effective results in the
case of regions bounded by one or two concentric circles. By mapping
given simply or doubly connected regions on a circle or circular ring,
such expansions of the unknown functions will likewise secure effective
solutions. The present chapter deals briefly with this problem, while a
more satisfactory application of conformal mapping by other means
will be described in Parts V and VI.

§ 63. Case of simply connected regions. 1°. Consider first the case
of a finite region S bounded by a simple contour L which may be mapped
on the circle | | < 1 by the function z = 3(0 ; denote by 丫 the cir¬
^
cumference | | = 1 of that circle.
Since, in the notation of § 50, the functions 平 i(z) and S(z) are ho¬
lomorphic in S, the functions <p(Q and 巾(Q will be holomorphic in 丫.
Hence one will have inside 丫 the expansions
8
十 4-8 +OO
吸) = £ 喈匕 = E 加以 叫9 =E (63.1)
0 0 0

One can thus try to solve the fundamental boundary problems by substi¬
tuting these series (assuming them to converge on 丫，i.e., for 已 = a =
in (51.1) or (41.5) which give certain systems of equations for the de¬
termination of the coefficients ak and ak.
This procedure will be explained for the case of the first fundamental
problem. The boundary condition (51.1) will now be written

cp(b) 4~ 8(b) 十 山⑹ = A. + 仔2， (63.2)

s'3)
omitting an arbitrary constant on the right-hand side. It can always
be assumed that the point z = 0 corresponds to the point C = 0, i.e.,
that 3(0) = 0. It is also known that 乎式0) and the imaginary part of
250
CHAP. 1 1 APPLICATION OF CONFORMAL MAPPING 251

or, in terms of cp(Q, that 平 (0) = aQ and the imaginary part of

中'(°)
--~ i.e., §[沅 Jed'®)], may be fixed arbitrarily. Hence, in the following, 夕
s (0) °
will be put equal to zero, while the imaginary part of zJs'(O) will be
left indeterminate for the time being. Further, suppose that 侬⑹心'
may be expanded (for cr = in a series of the form
3( +8 +00
—
3，(b)
- = S 呢加舟 = £ 廉仙
— 00
—8
(63.3)

which will be assumed to be absolutely convergent. It is easily shown

that this condition will be met, if the contour L satisfies the conditions
of §47.
In fact, this follows from the well known theorem of S. N. Bernstein [1] which
states that, if a function /(-9-) satisfies the Holder condition for the index a > j
(cf. § 65 for a definition of this term), the series of Fourier coefficients of this
function is absolutely convergent. In the present case, 3 伍/⑹ has a continuous
first order derivative and hence satisfies the Holder condition for a = 1.

Developing /工 if2 in a complex Fourier series (assuming this to

be possible)
QO OO

/1 + "2 = £ 月解侬 = S 念M (63.4)

— 8
—OO

and substituting (63.1), (63.3) and (63.4) in (63.2), one obtains

OO 4-8 OO OO 十8

S 诙/ + £ 仇 b" E kdka~k+1 £ 戏 = £ 力庐£ (63.5)

k~l 2= — oo 左=1 —8

Multiplying out the series of the middle term on the left-hand side —
the operation being known to be permissible, if it is assumed, for example,
that the series for g'(b)”as well as the series (63.3) converge absolutely
and comparing coefficients of (物 = 1,2, ...), one finds
—
00

+ £= T阪无仍? n+左一1 月?n (加 == 1 > 2, . . .) , (63・6)

k

similarly, one obtains from 仃一憎 (加 = 0, "2, . . .)

，+ £ m
OO

化 =i
万互 初—m+比—i — 月一惘 (冽 = 0, 1,2, , . (63.7)

The equations (63.6) form an infinite system of equations for the

infinitely many unknowns 儆. Each of these equations provides two
252 m. SOLUTION BY SERIES §63

real equations for the quantities 3%, where

如 + ，瓦 = 外， a 元 一 z% = 薪.
If one succeeds in solving this system by one or the other method,
the function <p(Q will be determined. The coefficients in the expansion
for MQ can then be found from (63.7). Thus, the basic problem consists
of the solution of the system (63.6), i.e., of the determination of the
function (p(0 Further, if the series for 乎(已)， 中(0, 勺'(9，determined in
this manner, are found to be uniformly convergent for' | ] = 1 and if
the series for 9*(9 is，in addition, absolutely convergent, one can be^
sure that the conditions of the problem will be satisfied.
Obviously, uniform convergence for | |
| C | V 1 and, hence, continuity of ^ — I entails uniform convergence for
乎' and 少 up to the contour, i.e., the regularity
of the solutions (§ 42) .
Note that, having found 乎(？， the function ^(C) may be determined directly
without recourse to (63.7). In fact, if cp(Q is known, the boundary value of @(0
on | | = 1 is given by

WO = 力 一 ifz 一 —
[cf. (63.2)]. Hence, the function 少(匕) may be calculated directly by means of
Cauchy's formula.

In many cases, the actual solution of the system of equations (63.6) will
present no difficulties. An analogous system, obtained for one particular
case by D. M. Volkov and A. A. Nazarov [1,2], was solved by the
method of successive approximation. Still earlier, P. Sokolov [1] gave
the solutions of a number of particular problems, which are of practical
importance, by an analogous method. The present treatment will be
restricted to the following remarks of a general character with respect
to the system (63.6), and a start will be made with the simple case
when 3(Q is a polynomial
3© = + C2F + ... + cnb w
(% 0, J # 0). (63.8)
In this work use will be made of the following notation which is a par¬
ticular case of a somewhat more general notation explained in § 76 and
which will be widely used. If
/© = 诙 曲 《 + ...+ a
is some polynomial, then / (Q, where the bar extends only over f, will
be understood to be the polynomial obtained from by replacing the
CHAP. 11 APPLICATION OF CONFORMAL MAPPING 253

coefficients by their conjugate complex values, so that, by definition,

79 = 厮 十五k + …十 2y.
In this notation one will have (remembering that o = e诒，a = £一语 = a-1)

/(b)= + 万3+ . .. + =2 。十 + ... + 『

=
O
Applying this notation to the expression figuring in (63.2),
one finds
3(cr) + c2c2 + . . . + cnan
— —
3
一
33 \ g + 2法-1+ … + 加m5+1
“ = an” + … + "7
％ 十 C2b
*7 + 2c 丁2+ … +.%%
2

The right-hand side, considered as a function of the complex variable <r

in the entire plane (and it is, obviously, a rational function) , has no poles
outside the circle 丫, including its boundary, with the exception of the point
6 = 8, because 3‘(0 does not vanish inside and on 丫 (§ 47) and hence
石'(1/0 does not vanish on or outside 丫. In fact, suppose that at some
point
plex value, one has co'(l/^o) = 0 or 3‘(Q = 0, where
—
for | | 1, one has 石'(1/^) 0 ; taking the conjugate com¬
1 /Co and —
| [ 1 , which is impossible. Thus, one has an expansion of the form
3(b) oo

— 6庐" + 方"一1n-1 + --- + 瓦 e 6一%「乃

ifcu 1

which holds for||a 1 and, in particular, for b = e^. Hence in the

present case the series (63.3) contains only a finite number of terms
with positive powers of o, and, in fact, one finds
bk = 0 {k n 1). (63.9)
The equations (63.6) now reduce to the following:
am = Am w 4- 1) (63.6')
and
+ 应 + 2&力 + . . + 符巨2 •
应2 十 册b2+ 2互 演 2 一 1)服一1风
.… 外
— 42， (63.6")
an + Kn 月熊；
254 III. SOLUTION BY SERIES §63

one also finds from (63.7)

律 +1
晨十 £ 航次一仇+3 = Z—懒 (m = 0,1, . . .). (637)

Thus, one has for the determination of the coefficients 曲，...，口”

the n equations (63.6") which correspond to 2% real equations for the
determination of the 2% real coefficients afc, 伊= 1, where
afe + zpk = ak. If the equations (63.6") have a solution, the remaining
coefficients are determined by (636) and (63.7') and it is easily proved
directly that the series for 平 © and 少(已)， obtained in this manner, will
satisfy the conditions of the problem, provided the given functions 九
and /2 are sufficiently regular, i.e., for example, they have second order
derivatives with respect to 9, satisfying the Dirichlet condition.
In fact, if this last condition is satisfied, one will have inequalities of the form
C 、
I "m I V ~ ~ (秋 = 土 1, ± 2, ・) (㈤
Im产
whence it follows by (63・6‘) that the series for *《) and 中'(？will be absolutely and
uniformly convergent for R I
where
1- Further, (63. 7Z) shows that 口 =二 —
匕愕 + 月一

他十 1

' 克诙 +

But the series Z 月一 converges absolutely by (a) ; the series S cm is likewise ab¬
馆

solutely convergent, because its terms are found in a number of terms of an ab¬
solutely convergent series, obtained by multiplying out the absolutely convergent
series 2 kak and S bk. It follows directly from this that the series for W® is absolutely
and uniformly convergent for | |
^ 1.

Thus, the solution of the problem will be obtained, provided the system
(636') can be solved. However, it is clear that the system. (63.6") cannot
give definite values for all the coefficients 叼，阮，.. an, in fact,
it is known beforehand that the imaginary part of 诙/3‘(0)=(叼+ 不
remains always arbitrary. This means that the determinant of the
system (63.6〃) must vanish, and from this it is known to follow that, for
the existence of a solution, the quantities Alt . . An must satisfy a
certain additional condition which will be deduced by excluding un¬
knowns from (63.6"). This condition will obviously express the demand
that the resultant moment of the external forces must vanish (and this
、
condition had been allowed for by assuming f 十 if2 to be continuous
on the contour), because the present problem has a solution for this
CHAP. 1 1 APPLICATION OF CONFORMAL MAPPING 255

(and only this) sufficient condition (cf. the existence proof in Part V.).
It follows from the theorem of uniqueness of solutions that all coefficients
%,. . an are completely determined, with the exception of the first
only the real part of which「or,better, the real part of %/3‘(0)] will be
fixed. The system (63.6〃) will be discussed in greater detail in § 84.
2°. As an example, consider the case when E is P a s c a 1's L i m a-
9 o n. By § 48, 2° (writing q instead of m)
z = 3(Q = 2? > 0, 0《夕 v
One has
(T
+ = 2+ (1
的 2a2) o — 2硝一 2谈)
1 + 2赤
] H

= (] 2a% — 2以1 — 2凝）官（一 1产 左=0

and hence in the present case 舛 = 2,

演 = 说， % = 1 一 2G2, b_k (一 1 产+1(20产+1(1 2凝), —
(万 0,1,2, . . .).
The system (636') will now reduce to the following :
a】+ @i(l — —
2a2) j— 口2 H- 巨1口左 2・

Substituting a2 = — + 笈曲obtained from

幺2， the second of these
equations, into the first equation, one finds
一 =
诙 ― ——
41 2设N 2
旬
—— --j- )

1 2谈
which determines the real part of av Hence, in order that the problem
have a solution, one requires that the imaginary part of & 2aA2 be
zero. It is easily verified directly that this is the condition for the vanishing
—
of the resultant moment of the external forces.
Putting for definiteness 灯(％) = 0, one finds
A] — ^dA 2
， '2 = 力 2
_
应1笈 N2
以
1 = 2(] 2应2)
%见

m+3
~ 4m(夕处 》3), 左加 S 6薪6 一馆 皿 (加 》°)

and the problem is solved. The series obtained for 中(〈) and 少公) can be
256 III. SOLUTION BY SERIES §63

summed and expressed by means of Cauchy type integrals, but this

will not be done here, since the corresponding formulae are more easily
obtained by another method (cf. § 84).
3°. Now consider the general case when 3(〈) is not a polynomial. By
omitting from the expansion
3(9 = + C2芋 4- ... + + J+ib+1 + ...
all terms, beginning with Cg+<m+i,one obtains instead of a poly¬
nomial s/9 which does not map the region S, but a region Sn on to the
unit circle, where represents an approximation to S which improves
with increasing n. The solution of the problem for the region has been
seen to present essentially no difficulties. The simplification brought
about by replacement of co(O by the polynomial 3式？is equivalent to
omitting in (63.6) and (63.7) all terms involving 瓦 for 左 + 1.
It has been seen that in this case the problem reduces to the solution
of a finite number of linear equations with a finite number of unknown
coefficients, namely, to the solution of (63.6〃) and to the calculation of
the remaining coefficients by means of (63.6') and (637). This is one of
the methods of approximate solution of the infinite systems (63.6) and
(63.7), i.e., of approximate solution of the original problem. Letting now
% increase beyond all bounds, the regions Sn will tend to S and the
approximate solution will tend to the exact one, i.e., the functions 平 and
found for the regions Sn> will tend to definite functions giving the
“
exact solution for the region S. Under known general assumptions with
respect to the contour of the region 5 and to the functions 九 and /2
given on the contour, this statement may be proved rigorously (cf. § 89
for a more detailed discussion).
4°. Quite similar remarks refer to the method of solution of the second
fundamental problem. This problem is even simpler, since, when the
displacements on the contour are known, the coefficient is completely
determined and the boundary problem will not be subject to any ad¬
ditional conditions. Thus, the system analogous to (63.6〃) will always
have a unique solution.
5°. In the case of infinite regions, mapped on to the circle | | < 1 by
a function of the form ^
3(? = 看十 冬 + C2? + ... + c尸， (638)

results are obtained which are as simple as for finite regions. No further
detail will be given on this, since in the above (and likewise in more
 CHAP. 1 1 APPLICATION OF CONFORMAL MAPPING 257

general) cases effective solutions may be deduced by the method of

Part IV.
The method of solution of the fundamental problems, studied in this
section, is given in more detail and with certain additional interesting
extensions in Chapter VI of the book by L. V. Kantorovich and V. I.
Krylov [1] ; the presentation given here has been reproduced without
essential changes from the earlier editions of the present book.

§ 64. Example of application of mapping on to a circular ring.

Solution of the fundamental problems for a continuous ellipse.
It is natural to try to generalize the method of § 63 to the case of
doubly-connected regions by using transformations on to the circular
ring. However, even for regions of a very simple form, direct application
of this method does not lead to simple results. Mapping on to the circular
ring will be used here to solve the fundamental problems for 轨e continuous
ellipse.
The first fundamental problem for a region, bounded by two eccentric circles,
has been solved by G. B. Jeffery [1], using a method which is essentially close to
that of § 63. A solution for a region bounded by two confocal ellipses has been
given recently by M. P. Sheremetjev [2] and A. I. Kalandiia [5]. It should be
noted that an earlier solution of this problem is due to A. Timpe [2] which has
turned out to be incorrect, as will be shown below. In fact, Timpe tried to obtain
the solution ol the problem by expanding the corresponding Airy function as a
series in terms of a certain system of particular solutions of the biharmonic equation.
But it is not difficult to verify that his system of particular solutions is incomplete.
The complete system is easily constructed, starting from the complex representation
of the biharmonic functions and utilizing conformal transformation on the circular
ring. Further observations on some other work referring to problems of an analogous
type may be found in the monograph, of G. N. Savin [8].
The fundamental problems for the continuous ellipse were solved by O. Tedone [1]
and T. Boggio [3] by other, more complex, means. The solution, given in this section,
was first presented in the Author's paper [16] and it was also contained in the
earlier editions of this book. Recently D. I. Sherman [18] gave a solution, based
on the integral equations of G. Lauricella. The final formulae of Sherman agree
with the Author's earlier formulae, if the latter are somewhat transformed. In fact,
the Author originally gave only the equations (64.21) and (64.21') for the calculation
of the coefficients ck, appearing in (64.23). However, by taking into consideration
the coefficients of the expansion (64.24), formula (64.27) follows directly from those
mentioned earlier. By substituting the expression for ck on the right-hand side
of (64.19), one obtains a formula which agrees with that of D. I, Sherman.
It is a fact that the finite region, bounded by an ellipse, may, like
every region bounded by a single contour, be mapped on to a circle.
258 HL SOLUTION BY SERIES § 64

But the corresponding transforming function is complicated and in¬

convenient. That is the reason why another transformation will be used
here.
Imagine that the ellipse has been cut along the segment connecting
its foci. This cut may likewise be conceived as an ellipse which is confocal
with the original one and whose minor axis is zero. Thus, one has the
limiting case of regions, lying between two confocal ellipses. This region
may be mapped on the ring between two circles 丫】 and y2 in the 已 plane
bv putting (§ 48, 5°) / ]\
z= = 衣 >0. (64.1)

〈 plane will correspond to ellipses in the z plane,

Circles of radius p in the
where the parametric representation of the latter is

% =R (p + — cos 9, y = 7? (p L) sin 眇. (64.2)

The circle p = 1 of the C plane will correspond to the segment AB of

the Ox axis (of the z plane), enclosed between the points
% = — 2R and = 十 2R
When the point 乙 describes the circle p = 1, the corresponding point z
moves along the segment AB in accordance with the law

z=% — 2R cos ~ R (<j )

-j = e巧， (64.3)

so that the points b — and a = 尸诒 of the 已 plane correspond to one

and the same point of the segment AB.
Thus, one can take for 丫1the unit circle ; the radius of 丫2 will be denoted
by p0 (p0> 】)• The magnitude of p0 will be determined by the given ec¬
centricity 2R and the major semi-axis of the ellipse @ = K(po+ l/p0),
whence it follows that
a + Va2 — 4R2

— —
(since V^2 4R2 would give p0 < 1).
In the notation of § 50, the functions 平式z) and 4fl (z) must be holo¬
morphic inside the uncut ellipse. Furthermore, they must be holomorphic
in the ellipse, cut along AB. Hence 乎⑷ and must be holomorphic
CHAP. 11 APPLICATION OF CONFORMAL MAPPING 259

in the ring between 丫】 and 丫2 and they will have expansions of the form
+8 +OO
吸) = Z 瞅y, 30 = S “s (64.5)
which are convergent for 1
—
OQ

-OO

< |
^[ < p0.

Actually, the series (64.5) will even converge for 1/% v R [ v p0, since it can
easily be shown that <p(0 and W(C) may be analytically continued into the region
between 丫, and the circle of radius p' = l/p0. For this purpose it is sufficient
to consider the Riemann surface of two sheets, superimposed on the z plane, with
branch points at A and B. The relation (64. 1) gives the transformation of this
surface on the ring l/p0 < | [ < p0- and hence the above statements are justified.
For p
^
the functions (64.5) must satisfy the boundary condition
= p0,

m)+ 书中化) + 怵) = /i 一 (64.6)

——
where 力 if? is a known function of 分 [cf. (63.2) ; the conjugate complex
expression has been given here solely for the sake of convenience].
Further, one must have on yr
中⑹ = 印⑹， 少 = 响， (64.7)
because the points b and g correspond to one and the same point of the
segment AB in the z plane. Conversely, if this condition is satisfied, the
functions 中式%) and “(z) will take one and the same value, when the
point% approaches the segment 月B from either side, and hence they will
be analytic functions in the uncut ellipse.
It follows from (64.5) and (64.7) that
口一 口右 次_人“ (64.8)
Introducing the series (64.5) into (64.6), noting that for p = p0
铮、
=农(1-a)=五(1 p，
硝9

哈 rg +
9 =喉 1 却 .
and multiplying both sides of (64.6) by 1 — — one obtains ,

(1 一导)
\ po —
擀不 /
+(4
8 q + 4) WhaQf +
Po/ —
\ 8

+ £“？= (/】一也) 一卷)f8 p -

+8 /
(1- %, (64.9)
— OO \
260 in. SOLUTION BY SERIES §64

where use has been made of the notation

/ 1 \ / 1 \ +°° , +8 +8

1-77 O- 1-7T)/
^普 =£饱 一收 +2)y = £6Q, (64.10)
\ J" / \ 4,
— OO
— 8
— oo

so that
瓦; = & — 忌+2・ (64.11)
Expanding the right-hand side of (64.9) in the complex Fourier series
十8

(/1- 浜) ( 1 - P瑞 汾) = £ 4尹M -2 (64.12)

putting 已 = p/3 and comparing coefficients of e*吗 one finds

—8

p萨万一% —
Po*-4^-fc-2 + 3 + 2) 若 +2题+2 + 如净-2做 %若 4 加，
'

or, noting that by (64.8) a_k = ak, a^k_2 = &丸+2，

(万 + 2)p，+2即+2 — p/I薪 + 砧号一 瞅+2
2
+ ^fcPo ~ 2a (64. 13)
Replacing 万 by — k 2 and noting that by (64.8) and (64.1 1)
~

b—k—2 = d一e一2 区Tb = 次 + 2 族 = 6%， (64.8 )

one finds
3+2)f芦-4曲+2+9§+2薪+2~~的小以一Po-2^fc—6%pki2=Z_i_2。 (64.13')
，

Elimination of bk from (64. 1 3) and (64.13') finally leads to

3 + 2) (pQ — p常以+2 + (P萨+4 ~ 獭+2
—
(64 ]4)

where
^(Po
— — (Po* — p/") 诙= Bk,
—
Bk = Ak^ + (64.15)
The coefficients ak can be determined from the recurrence formula
(64.14), provided 曲 and 的 are known. [In actual fact, (64.14) gives for
each k two equations obtained by separating real and imaginary parts ;
instead of this, one may deduce a second equation by going to conjugate
complex quantities (see later) .] The coefficient 曲 may be fixed ar¬
bitrarily, since one can always add an arbitrary constant to <p(0. The
formulae (64.14) show that, as was to be expected, the coefficient «2 (and
hence also aQ …) does not depend on «0; in fact,, for k 0, the terms —
involving a0 cancel out. In order to calculate 旬 = j, put 万 =
(64.14) which gives
1 in
—
+&= (64・16)
旬 勿
2(pj — 心
Po )
=
Po — Po
CHAP. 1 1 APPLICATION OF CONFORMAL MAPPING 261

This relation permits calculation of the real part of % and at the same
time shows that one must have
力 —i = a real quantity, (64.17)
so that the problem can be solved. It is easily verified directly that
(64.17) expresses the condition for the vanishing of the resultant moment
of the external forces. (The vanishing of the resultant force vector has
already been taken care of by assuming 九 and /2 to be continuous on the
boundary.)
The imaginary part of 媒】 remains arbitrary, as was to be expected,
since one can always add to 平式2) a term of the form Ciz, where C is an
arbitrary real constant ; hence one can add to 平(V) a term of the form
Ciz = CiR^ 十 1/Q. It is easily seen that this imaginary part does not
affect 他，口5, etc. Thus, by giving arbitrary values to a0 and the imaginary
part of % and by determining successively all remaining coefficients
by means of (64.14), one obtains an expression for 平(？.
After this the coefficients bk can be found from one of the formulae
(64.13) or (64.13'). In this way one finds an expression for

1 一 TT
1 \

q /
WC)
-— 4-<x>
E 猴f = S b^k
8
oo

fc— o
oo
+ Si
or, remembering that b_ 1c =
1
c
、——
/
b— (i.e., in particular, b—】=
°°
加=(
，_工 = 0),

(64.18)

It will be shown below that for definite conditions the series for 少(？
and <p(Q converge in the relevant region. The right-hand side of (64.18)
vanishes for 工 = ± 1 and, consequently, the function 少(？，obtained by
—
dividing the right-hand side by 1 1/% will not be singular for 已 = ± 1
(see remarks following (63.7) with regard to direct determination of
W(0 by use of 平⑷ and the boundary condition.)
Thus the problem is solved. The second fundamental problem can be
solved in a similar manner.
Before turning to the question of the convergence of the above series,
it may be noted that calculation of the coefficients a% (B = 2, 3, .. . ) can
be simplified as follows. For convenience put
—
^(Po pr题 + (p学 一 f滑)蕊 = c 九， (64.19)
in which case (64.14) becomes
爆+2
— 爆 = (64.20)
262 III. SOLUTION BY SERIES §64

Substituting in this formula successively k 0,2, ... 2% 2, adding

the results thus obtained and noting that cQ = 0, one obtains
— —
州 =1 n—1
C2n = X 岛卜 = 2 (儿丽混
尼 =o B一o
+ Z—2lP/2). (64.21)

Similarly, substituting in (64.20) successively

and adding, one finds
为 = 1, 3, . . . , 2% —1
n n
C2n+L = 十£ = % + £ (^2fc-lPo2& +1 + 月-21浒”)' (64.21')

where, by (64.19) and (64.16),

ci = (Po ~~ Po') ("i 十 丘i) ~ 月tPo・ (64.22)

Thus, one has found closed expressions for the quantities ck. The
coefficients ak, however, can be expressed very simply in terms of the
%; in fact, writing down the conjugate complex equation of (64.19)
and solving for ak, one finds.

=
L
论(P；
— Pd^Cjj；
(pZ ~ Pb2A ~ •
灭
他 = 2,3, …). (64.23)
淤储 -p消2 一常 一 p萧尹
The expressions (64.21) and (64.21') for ck may be further simplified,
if one introduces instead of the coefficients Ak of (64.12) the coefficients
Ck of the expansion of the function 左 一 弘 as a complex Fourier series
十8

九一 = 吗 (64.24)

Comparing (64.24) with (64.12), one observes that

Ak = Ck-9fCk+2. (64.25)
The expression (64.22) for q takes then the form
% = C-ipo CrL (64.26)
Substituting (64.25) and (64.26) on the right-hand side of (64.21) and
(64.2T), one finally finds the simple formula
ck = C-力需 一 (6 = 1,2,...). (64.27)
Finally consider the question of the convergence of the series obtained
above and suppose that the functions 力 and /2 have second order deri¬
vatives, satisfying the Dirichlet condition (or more generally, being of
CHAP. 11 APPLICATION OF CONFORMAL MAPPING 263

bounded variation). Then one has for the coefficients Ck of (64.24)

inequalities of the form
I \ <7-7- 3=±1, ±2, …).
I 左 I3
On the basis of (64.27), (64.23) and (64.13) or (64.13') one easily finds
the inequalities
C C
1 做赭 ±1，±2, …)， (64.28)
a I IPo < n 3=
TT
from which follows immediately the absolute and uniform convergence
of the series for 甲(Q, (1
—
l/^2) 少(0 也 the interval

——
《 I E《
Po
I Po，
and hence the suitability of the solutions.
A
 PART IV

ON CAUCHY INTEGRALS

In the subsequent chapters wide use will be made of so-called Cauchy

integrals. A systematic study of the properties of these integrals may
be found in the Author's book [25] , but for the convenience of the reader,
who wants to limit himself to the information actually required for the
understanding of what follows, the essentials will be given in the present
Part. Some deductions will be stated without proofs ; these may be found
in the book mentioned above, or in I. I. Privalov's book [1]. On the
other hand, a number of elementary formulae and results will be given
here which are of practical value and which are not contained in those
books.
A
 Chapter 12

FUNDAMENTAL PROPERTIES OF CAUCHY INTEGRALS

§ 65. Notation and terminology.
1°. In the sequel, unless stated otherwise, L will be a simple smooth
contour, a simple finite smooth arc or the union of a finite number of
such disconnected arcs and contours in the plane Oxy (Fig. 30). In this
case L will be called a simple smooth
line, where the words "simple'' and
''smooth” will often be omitted.
Thus the line L may consist of parts
placed separately from each other. If L
contains arcs, their ends will be called
ends of the line L.
The line L will always be given a
definite positive direction ; in the case
when it consists of disconnected parts
a positive direction must be chosen on
Fig. 30.
each of these parts.
If one draws about any point of the (simple, smooth) line L, which
does not coincide with one of its ends, a circle of sufficiently small radius,
this circle will be divided by L into two parts one of which will
lie on the left and the other on the right of the line (looking in the
positive direction of L; cf. Fig. 30). In accordance with this, a distinction
may be made between "left” and "right" neighbourhoods of the point
t on L, other than one of its ends. For example, the left neighbourhood
of i consists of points, not on L and in the left section of a circle drawn
with sufficiently small radius about t.
In a similar manner one may distinguish between left and right
neighbourhoods of any part of a line L, the ends of which are not ends
of L. As before, a 夕矽/ of the line L will always be a part consisting of (a
finite number of) arcs or contours belonging to L. The left and right
neighbourhoods will be denoted by (+) and (—) respectively.
2°. The definitions of § 37 will now be recalled and partly extended.
267
268 IV. ON CAUCHY INTEGRALS §65

Let F(z) be some function, given in the neighbourhood of L, but not on

L, and assume that F(z) is continuous there. (As in § 37, the function
F(z) will not be assumed to be analytic). Also let t be some point of L
other than its ends (if the last exist).
The function F(z) will be said to be continuous at t from the left (or right)
f

if F (z) tends to a limit as z tends to 力 along any path remaining, however,

on the left (or right) of L [i.e.,zmay take any position in the left (right)
neighbourhood of 力. The limiting values of F(z) as z 力 from the left
今

or from the right will be denoted by

F+Q) or t1(力
respectively and they will be called the boundary values of F(z).
This notation, and the term "boundary value” will only be %sed in
such cases when the corresponding limiting values exist as z tends to t
along any path on the left or on the right of i.e., when F(z) is continuous
at t from the left or from the right.
Let 工' be some part of L the ends of which do not coincide with those
of L (if such exist). The function F(z) will be said to be continuous at
V from the left [or right], if the limiting value 尸+(/) [or 尸R)] exists
for all points t of U .
As mentioned in § 37, if F(z) is continuous on 工' from the left [or
right], the function F+«) [or 万一(切 will be continuous on V . Hence it
follows that, if the line U be added to the left [or right] neighbourhoods
of Lr and if the function F(z) be given its values F+«) [or (切 on
then F(z) will be continuous in the left [or right] neighbourhood, including
the line Lf.

3°. Let 附 be some, in general complex, function of the point £ of

this means that
力)=/m)+ 弘巴 (65.1)
where /式力) and %(力) are real functions of t on L.
—
In future, 力 will denote a point as well as its coordinates, i.e., t

、,
The function /(力) will be said to satisfy on L the Holder condition, or
just the H condition, if for every pair of points t t2 of L the following
inequality holds true:
I 他) 一/&)1< 月也一 (652)
where A and r are positive constants and 0 < p, 1 ; A is called the
Holder constant and pt the Holder index.
CHAP. 12 FUNDAMENTAL PROPERTIES 269

The condition (65.2) is easily seen to be equivalent to

"⑹ 一/㈤ I 啖， (65.3)
where B is a positive constant and g12 is the length of arc of L between
£ and %；if h and t2 lie on a contour forming part of L, the shorter of the
two arcs between 力】and % must be taken for <r12. If L consists of several
different parts, the condition (65.3) must be understood to be fulfilled
for any pair of points lying on one and the same part.
The equivalence of the conditions (65.2) and (65.3) follows from these propositions
which are easily proved :
1°. If (65.2) is satisfied for any pair of points whose distance does not exceed
some fixed number 8, it will be satisfied for the whole of L, provided, if this should
be necessary, A is replaced by a larger value.
2°. For any pair of points % whose distance does not exceed 8
— %犯I•VI,
——
|
A
a12
where A is a positive constant. Proofs of these propositions may be found, for
example, in the Author^ book [25].

If pt, > 1 in (65.2) or (65.3), it is easily seen that the derivative of

/(/) with respect to the arc sofL will be zero ; hence, in this case, /(/)= const,
on L or, 迁 L consists of different parts, on each of these parts. This case
is of no interest, and for this reason consideration will be restricted to
(X 1.

NOTE. If for a given point 曲 of L one has the inequality

I 力) 一限)1《4卜一记“
for all Z of £, sufficiently close to 曲，the function /(/) will be said to satisfy
the H condition on L 就 the given point 力 ; however, this does not mean
°
that /(£) satisfies the H condition in the neighbourhood of %,i.e., that (65.2)
holds true for any pair of points in the neighbourhood of % on L
4°. In the sequel the following well-known notation will sometimes
be used. Let be a variable quantity which runs through some set of
values and tends to 0 [or to 8〕. Then 0© will denote a quantity such
that O(E)/E remains bounded for sufficiently small [or sufficiently large]
values of|己 In other words, for those values of 己

where C is a finite constant. Further, 。(已) will denote a quantity such

270 IV. ON CAUCHY INTEGRALS § 66

that (the modulus of) 。(号/己 will be as small as desired, when ] | is

sufficiently small [or sufficiently large] , i.e., ^
I 喈) 【 c •化 I，
where c is a positive quantity which only depends on | | and which
tends to zero for 己 f 0 [E oo]. ^
For example, if /Q) satisfies the H condition in the neighbourhood of
the point t0, this condition may be written
"(幻 一 /(^1) I — 0(1 —42 ,
左 I")
for all points 左，％ which are sufficiently near to tQ.
One particular case of this notation should be noted. Consider the
expression 0( | 已『)，where a is a real number. By definition, 0(] E F)/| E [a
remains bounded when [ | -> 0 口己]—8]. In particular, for a = 0,
^
the expression 0(| 己 ]*) becomes 0(1). Thus 0(1) denotes a quantity which

—
remains bounded for sufficiently small [or sufficiently large] values
of | [. Similarly, o(l) denotes a quantity which tends to zero for [ | 0
^
[or | 己| co], i.e., | o(l) [ < s, where s only depends on | [ and lim s = Q
^
for | E | f 0 [or | W | f oo|.
For example, the condition that 超) is continuous on L may be written
"(%) — /&) I = 。
for | — 力i I f 0.

§ 66. Cauchy integrals. Let L be the same as in the preceding

section and let 力) = /式力) 十 if 购 be some, in general, complex function
given on L. Unless stated otherwise, it will always be assumed that /(/) is
finite and integrable in an ordinary (Riemann) sense.
The integral of the form
I
/
27c2’ J t
L
Z

taken over L with z some point in the plane of L, will be called an integral
of the Cauchy type or Cauchy integral ; the factor l/2m is not essential
and has only been introduced for the sake of convenience.
For the present it will be assumed that the point z does not lie on L.
In that case the integral (0 has a definite meaning and represents a
function of the complex variable z throughout the entire plane with the
CHAP. 12 FUNDAMENTAL PROPERTIES 271

exception of the points of L. This function will be denoted by F(z), so that

F (66.1)
L

It is easily seen that F is holomorphic in the whole plane, excluding

the points of L. Ii L consists of contours, as shown in Fig. 30, the preceding
statement must be understood in the sense that F(z) is holomorphic in¬

—
side all parts into which the plane is divided by L. (It must not be thought
that F(z) is analytically continued when z passes from one part to another ;
this will become clear from the following work.)
Further, it is easily seen that for z 8 F tends to zero, i.e.,
尸3)= 0. (66.2)

§ 67. Values of Cauchy integrals on the path of integration.

Principal value. Hitherto it has been assumed that the point z in (66.1)
does not lie on the line of integration L. Now let z coincide with some
point t0 of L. For the present write formally

(67.1)
27V2 J t —%
If / 0, the integrand becomes infinite like |力 一 j-1 as 力 f %.
Hence the integral has no meaning in the ordinary sense. However,
for certain conditions referring to /(/) , the integral (67.1) may be given a
definite interpretation. In fact, assume that 力 is not an end of L (if
such exist) and separate from L a sufficiently small° arc 々％ which contains
in such a way that
修一％1 = 1^2 ^!.— (67.2)
Denote the arc 友％ by % and the remaining part of L by £
the integral
— I and consider
1
［ 足成
书八不 • (67.3)
/67 3)

This integral is completely defined in the ordinary sense, since, as t travels

—
along the path of integration L I, |力 一 /0|> § where 3 is some positive
constant.
Next suppose that 片 and t2 tend to 力 in such a way that (67.2) remains
satisfied. If under these conditions the° integral (67.3) tends to a definite
272 IV. ON CAUCHY INTEGRALS §67

limit, this limit is called the principal value o/ the Cauchy integral (67.1).
Clearly, if (67.1) has a meaning in
b the ordinary (Riemann) sense, its
principal value will exist, but the
converse proposition is, generally
speaking, not true. (In this connection
the term ordinary means that 友 and
力2 tend to tQ in an arbitrary manner,
so that (67.2) is no longer fulfilled).
The principal value of an integral,
if it exists, will be denoted by the
same symbol as the ordinary integral,
i.e., by (67. 1 ) , where it must be under¬
stood that, if the integral has no
meaning in the ordinary sense, its principal value must be taken (provided
that it exists). Some authors use a special symbol for the principal value ;
for example, the integral sign is accented (') or the letters VP (Valeur
Principale) are put in front of it.
No consideration will be given to the problem of finding the most
general conditions for the existence of a principal value, but instead
one very important case (which is completely sufficient for the purpose
of this book) will be stated when this existence is definitely ensured. In
fact, the principal value o/ the integral (67.1) exists, if the function 力)
satisfies the H condition in the neighbourhood of the point 片，i.e., the con¬
dition (cf. § 65)
"㈤ 一/（幻 I 4a 也一 片产， 0<fx< 1. (67.4)
This proposition will now be proved by actually expressing the principal
value of the integral by an ordinary integral. For this purpose investigate
the integral (67.3) and consider first the case (cf. Fig. 31) when L consists
of a simple closed arc ab, i.e.3 consider the integral

(673)
ab-l
f
—— ％
where the positive direction is from a to b.
This integral may be written as follows :
1 「附近 1 ，m)一/⑷ / dt
2兀
—
z J t % T /
2m J t tQ — 力一
十 7
27rz
/
J t —° 力
• ( O/.5)
CHAP. 12 FUNDAMENTAL PROPERTIES 273

The first integral on the right-hand side tends to the limit

2tt勿 J
ab
4 一％

as 左 f 曲，t2 f 片，because it converges in the ordinary (Riemann) sense ;

in fact, by (67.4),

! / ~
- t0 [ 1 1 — % 11*
—
and since 1 v inequality ensures convergence
q 1 , this
by a well-known elementary convergence theorem.
of the integral

Next consider the second integral on the right-hand side of (67.5)

which is easily represented in the form

ru：
1 / d力

ab—l
=
1
[log (力 一戏 +
元
,1
[log 一 他
而 3
%

—
where by log (/ 幻 on the parts and t也 of the line ab must be under¬
stood any branches of this function which change continuously with t
on each of the parts %, separately. These branches may be chosen
arbitrarily on each of these parts, but for the sake of definiteness they
will be related by the following condition: the value log Q
£ = % is to be obtained from the value log (£ ％) for / = 左 by means of
for
—— —
a continuous change of log Q
—
, as the point t moves from 片 to t2
along a (infinitely small) semi-circle, lying to the left of L (cf. Fig. 31).
Under this condition the branch log Q 幻 on atx completely determines
the choice of the branch on 她 and, provided this choice has been made,
—
one may write
\
/C dt
——
1
y log —
I) — 曲丁tq + ] ,
log —% — —
M, .
(67.6)
2兀z J
ab-l
力
— 2冗2 口 — — %
where

log-
b ―i = e — 幻 一 log (以 — 4),
Zo log
•

log/ ~
= log & — 幻 一 log (力2 一 幻，
%1

%
274 IV. ON CAUCHY INTEGRALS §67

for the choice of logarithms, stated above. Further, since by supposition

I % l/l 力2 — Ml — 1,

log"?=
还一 M
汨，

where 6 is the angle, shown in Fig. 31. Obviously, as tQ, one

has: lim 8 = 兀. Hence, proceeding in (67.6) to the limit, one obtains

log 2=4
d —%
Thus, the integral (673) has a definite limiting value and this limit is,
by definition, the principal value of the integral
i r
it is given by the formula
27cz m t
- 力。

1 6
—
•

抓％) + kt/(幻 log

i”

F
ci
— 7

+而 J FZ^一 祝 (67.7)
ab

where the integral on the right-hand side has an ordinary (Riemann)

value.
The preceding statement implies that the condition (67.2) is satisfied. If this
condition were not fulfilled, one would have instead of (a)

— = log
立］
22 —— ％
log
力 产工
夕
2
F 2仇

where |
— % r2 = \ t2 — %
(b) would not tend to limit.
a
hence, if (67.2) is omitted, the expression

Now let L be an arbitrary line of the form discussed in § 65, 1°. Then,
selecting on L some arc ab containing 曲 (in such a way that 口 or 6 do
not coincide with 幻，one may rewrite the integral (67.3)

t 力o

Provided (67.2) is satisfied, the first integral on the right-hand side

CHAP. 12 FUNDAMENTAL PROPERTIES 275

has just been shown to tend to a definite limit, as 々 f tQ, /2 f %; the

second integral obviously does not depend on J and Hence (67.3) tends
to a definite limit which is, by definition, the principal value of (67.1) ;
on the basis of the preceding work, this principal value is given by

= i/M T(%) log*— +

%
—— to 方
N7T? Cl
— Zq
, 1 广力) 一 力o) (
1 ，力)业 (67.8)
2兀 VJ t — %
«&
' 2Tzi J
L-ab
t—

This formula shows little symmetry and it will not be used below ;
it has only been introduced here to show that the Cauchy principal value
exists under the conditions referring to /(/), stated earlier, and that it may
be expressed by means of ordinary integrals.

NOTE. 1. The formula (67.8) is considerably simplified, if E is a

simple contour. The relevant formula for this case may be deduced
from (67.7) by letting the end b of the arc ab tend to the end a, so that
one obtains in the limit the contour L. Assuming, for definiteness, that
the positive direction on this contour is chosen in such a way that the
finite part of the plane bounded by L lies to the left when looking along
L in the positive direction, one will have (for 万 = 口)

and (67.7) takes the form

牌任| 二
0，

1 广力城~~ 1 (即 一 fk)
1
।
— - 2八引十 不一;
/— -- 小
ai. (677)
L L
In fact, it is sufficient to assume that
NOTE. 2. Assuming that f satisfies the Holder condition for the
values of t under consideration, the following situation will be noted.
It is unnecessary to introduce the condition (67.2), in order to define
the principal value of the Cauchy integral.

lim Ul二M
—% I !
as % %，
f 力i%o, ie, that % = %
f
— Zo |, r2 =|片 — o ] are infinitesimal 力
quantities of equal order of magnitude. Obviously one will again have
276 IV. ON CAUCHY INTEGRALS §68

under this condition that

. 匕
i
lim —
一
=2
•
兀
o
(cf. remarks following 67.7), and hence all the preceding conclusions and
formulae will remain valid.
In particular, the condition (67.2) may be replaced by
j = C2» (672)
where % and & denote the lengths of the arcs 印 and t点 so that the
point 力o divides the arc °
into two equally long parts.
NOTE. 3. Obviously the formulae and conclusions of this section
will remain in force, if /(/) satisfies the H condition only at a point
(cf. §65, 3°), i.e., if
"(%)-/(幻
(for t sufficiently close to 力°); thus there is no need for the H condition
to be satisfied for any pair of points in the neighbourhood of 力. But in
that case the above work will only hold true for the given value t0.

§ 68. Boundary values of Cauchy integrals. The Plemelj for¬

mulae. As regards the values of the Cauchy integral

尸 gl)
27rz J t
L
—— Z

on the line of integration, considered in § 67, a distinction must be made

between its boundary values, i.e., between the limits of F@ as z tends
to tQ on L from the left or from the right. The following important pro¬
position holds true with regard to these boundary values.
If the function /(£), given on L, satisfies the H condition in the neighbour¬
hood of a point o/ L, other than of its ends, the integral F(z) Es c。％-
tinuous 威 L from the left and from the right, E.e., the boundary values F+{t^
and exist. (Naturally, the H condition is only satisfied for points of
L near tQ} since /0) is not given for other points of the plane. See also
§65, 2。).
These boundary values are given by the formulae

尹(公 = 时依) + 不下 (68.2)

zm
L
J t —— %
CHAP. 12 FUNDAMENTAL PROPERTIES 277

1 C
P一(幻 = 一娘6 + k ， (68.3)
27cz J t %
工

where the principal values of the integrals must be taken the right-hand
sides.
The formulae (68.2) and (68.3) may be replaced by the equivalent
expressions

+ F-(幻 = ，
F+(%)—FR°) = 力°),

/ 誓——£-*
E
L
J t %
(68.4)

(68.5)

Further, the following proposition holds true:

If the function /Q) satisfies H condition o% so^e part L' of L, the
boundary values F+(%) 尸一(曲) 瀛。satisfy the H condition o% L' , except
possibly in an arbitrarily small neighbourhood o/ the ends of L' (歹 the latter
exist) .
The above theorems were first given by J. Plemelj [1] and refined by
I. I. Privalov. The formulae (68.2) and (68.3) were likewise given by
J. Plemelj (in the same paper) and for this reason will be called Plemelj
formulae. The proofs of the formulae and theorems of the present
section may be found in the books by I. I. Privalov [1], A. I. Mar-
kushevich [1] and the Author [25].
NOTE. 1. The expression (68.4) follows from (68.2) and (68.3) which
have been obtained under the supposition that /(/) satisfies the H con¬
dition in the neighbourhood of the point %. But it may also be extended
(in a conventional way) to the case when /(£) is only continuous in the
neighbourhood of %. Draw through % some straight line A which does not
coincide with the tangent at % and select on this line two points and t" ，
in such a way that the segment 比" is bisected by % Then, if /(力) is con¬
tinuous (on L) near t0, the difference
f(u) — f&)
tends to the limit /(%) as 力' f tQ, t" -> tQ (provided t' and 力" are all the
time equidistant from %). Hence, denoting this limit by F+&)
one finds (68.4) to be still valid under the new conditions for /(/). —
This circumstance was likewise noted by J. Plemelj [1] ; the proof may
be found in the Author's book [25]. It can also be shown that
tends uniformly to the limit /&) (with regard to the position of 力 on
F&)—
°
278 IV. ON CAUCHY INTEGRALS §68

some sufficiently small part of L), if the non-obtuse angle between the
straight line A and the tangent to L at 琳 is not less than some fixed acute
angle (the proof of this proposition is given in the Author's book [25]).
NOTE. 2. The following result follows immediately from the
statements of the preceding Note: If the function /(力) is continuous
on L in the neighbourhood of t0 and if the boundary value 尸+(片) [or
F-(幻] exists, the boundary value 尸-(%) [or 尸+(%)] also exists and these
boundary values are related to each other by (68.4).
NOTE. 3. In contrast to what has been said in Note. 3 of § 67, it
is not sufficient for the existence of the boundary values 尸+(幻，尸㈤ to
assume that /Q) satisfies the H condition only at a given point tQ (cf.
§ 65, 3°.) and not in some (arbitrarily small) neighbourhood of (on E).
However, under this last supposition, there will exist limits of the function
F(z) as z f % from the left or from the right, if it is assumed that this
“
transition takes place along a definite path not tangential to L.
NOTE. 4. Let Z, be a simple closed arc the ends of which will be
denoted by a and b and the positive direction of which is from a to b.
The behaviour of the function F near the ends is easily determined. In
fact, let it first be assumed that f(a) = 0. Then, extending the line L
beyond the end a, for example by a segment of the tangent there, and
putting on the additional part /(£) = 0, one arrives at the case where
a is not an end. Hence, applying the earlier results, it is easily concluded
that _F(z) tends to a definite limit as z tends to a along any path. [By
(68.2) and (68.3), one will have F+ = because in the present
case /(%) = /(a) = 0.] If /(口) 乎 0, the formula (68.1) can be rewritten
in the form
1 1 (他
咽
力 (z) = 1 / 卜 . at =
2兀2 J t
ab — Z 27Tz J
t
ab — Z

一/⑷ b-Z 1 /超)一/0

ab
and it is easily seen, on the basis of the preceding remarks, that near a

尸 = 华
2
rlog 」一
九2 z a ~-
F*(z), (68.6)

where F*(z) tends to a definite limit as 2 f 原 Similarly, one has for the
CHAP. 12 FUNDAMENTAL PROPERTIES 279

end b
F(z) —27rz log 十 尸**(z), (68.7)
—
~
z b
where F**(z) tends to a definite limit as z 6.
These results can be immediately extended to the case when L contains
an arbitrary number of closed arcs a*.

§ 69. The derivatives of Cauchy integrals. 1°. As before, let

/心)今
F(z) (69.1)
L
t z — •

where 超) and L are the same as at the beginning of § 66 and z is a point

not on L. Derivatives of any order of F may be obtained by differ¬
entiating the integral on the right-hand side with respect to z, so that

(69.2)
Q — z)2
and, in general.

M(z) =
k\
/ 州)今
(69.3)
方, — z产
L

Now the question arises with regard to the behaviour of these deriva¬
tives when z approaches L from one or the other side. This question is
easily answered, if it is assumed that /(/), given on L, satisfies certain
conditions.
For example, suppose that /(/) has on some arc 尤 0fL a first derivative
with respect to 力 which satisfies on L the H condition. By the derivative
of /(0 with respect to t will, of course, be understood the limit

Hm 幽）二 B
as / in an arbitrary manner, remaining all the time on the arc ab ;
this derivative will be denoted, as usually, by /'(/) or 研©固.
Subdivide the integral on the right-hand side of (69.2) into two in¬
tegrals one of which is taken over the arc ab and the other over the re¬
mainder of L. Obviously, the second integral represents a function of z
which is holomorphic near the points of the arc ab, other than its ends.
280 IV. ON CAUCHY INTEGRALS §69

The first integral can be transformed by an integration by parts :

上「 附位 i i
2m J 优 一 z)2 -
一
2拓 J —z
ab ab

Since, by supposition, /'(/) satisfies the H condition on ab, it is clear

that, using the results of § 68, the right-hand side of the preceding formula
is continuous at ab from the left and from the right, if one excludes the
ends a and 力；hence the same will be true with regard to 尸(z).
Proceeding progressively to the higher order derivatives, it is easily
shown that the function F is continuous at ab from the left and
from the right, excluding the ends a and b, provided the function /(£)
has an 死-th order derivative with respect to t which satisfies on the
same arc ab of L the H condition.
Using the results of § 68, it may be shown that under the stated conditions
the boundary values [户乳)(切+ and [F (切一 satisfy the H condition on
优6, if (arbitrarily small) neighbourhoods of the ends a and b are excluded.

2°. Consider the first derivative F'(z). If the function is subjected

only to the H condition, it is already impossible to assert that the
derivative F'(z) is continued continuously on £; it may even turn out
to be unbounded near the boundary.
The following simple estimate of the modulus of this derivative is
often useful.
Let M be a point of the line L lying at a finite distance from its ends
(if such exist) and let 力) satisfy the H condition near this point. Then,
for points z sufficiently near to tQ>
const.
I F' I (69.4)

where 5 is. the shortest distance from z to E and a is a constant smaller

than unity. (In fact, if 及 is the Plemelj index for /(/), then, assuming
[x < I, which is always admissible, one may take a = 1
—
p).
This estimate follows directly from an estimate in the author's book
[25].
NOTE. If one determines the position of the point t on ab by means
of the arc s, measured from some fixed point (say a) of L in the positive
direction, one obviously has
a力) 〃、近 ,
长一
=片力) 也，孙 伯9.5)
CHAP. 12 FUNDAMENTAL PROPERTIES 281

where a is the angle between the positive tangent to 上 at 力 and the 0%

axis. Hence
讶(£) 曲 _.a df(t)
— 一
ds dt
“W .
— ds
, J *
(69.6)

By supposition, the line L was to be smooth, i.e., such that the angle a
changes continuously with t (or with s). It does not follow from this that
a satisfies the H condition. Therefore, if /'Q) satisfies the H condition,
this is not necessarily true for df(t)/ds.
If, in addition, it is assumed that a satisfies the H condition, then it
follows from the fact that /'(/) satisfies the H condition that also 财⑦ [ds
satisfies that condition, and vice versa.
Further, it does not follow from the existence of the second derivative
with respect to t

〃
that 2力)/曲2exists, even if it is assumed that a satisfies the H con¬
dition. But, if it is assumed that the derivative “a/ds (which is known to
represent the curvature of the line L at /) exists, then the derivative
涉/㈤ /ds? exists and it may be expressed by the formula
凝/(/) / dt \2 "2/ .
飞 厂=广防(石) +/储源 = 十沙石 /3， (69.7)
which follows from (69.4) ; this derivative will satisfy the H condition,
if that condition is satisfied by (力) and by d%]ds.
Similar reasoning may be applied to derivatives of higher order.

§ 70. Some elementary formulae, facilitating the calculation

of Cauchy integrals. A number of simple formulae will now be deduced
which facilitate calculations in many cases.
Let E be a simple smooth contour. Denote by S+ the finite part of the
plane bounded by L, and by the infinite part of the plane consisting
of the points lying outside L. The contour L will not be included with S+
or S~. The region, consisting of S+ and of the points of L, will, by an
obvious notation, be denoted by S+ 十 L, and the region, consisting of S~
and of the points of 上，by S~ + L. The positive direction of L will be
chosen so that the region S+ lies on the left.
Now the following well-known formulae will be recalled.
282 IV. ON CAUCHY INTEGRALS §70

1°. Let /(z) be a function, holomorphic in S+ and continuous in

S+ + L. Then
for z in S+, (70.1)
L

/幽 S =0
t Z
for z in S-; (70.2)
27c4 J
L
(70.1) is Cauchy's formula and (70.2) is a direct consequence of Cauchy's
theorem, because in this case the integrand /(/)/(/ —— z), considered as a
function of 力， is holomorphic in S+ and continuous in S+ + L.
2°. Let /(z) be a function, holomorphic in including the point at
infinity and continuous in S~ + L. (It will be remembered that this
means that for sufficiently large | z |

/(z) = % 4—z- H—z”£ + • • • ,

so that Co = /(oo).) Then

1
27cz
.吸 /(z) /(8)
J t — z = —— +
for z in S~, (70.1)

= /g) for z in S+; (702)

L
力
—— z
(70. T) will be called Cauchy's formula for the infinite region S~. The
signs on the right-hand sides of (70.1') and (702) must be inverted, if the
positive direction on L is chosen in such a way that (and not S+) lies
to the left.
Note how the formulae (70.1') and (70.2') may be deduced from
Cauchy's formula and theorem for finite regions. Assume for the time
being that /(oo) = 0. Let r be a circle with centre at the origin and
with so large a radius that the contour L and the point z lie inside T.
Then, assuming z to lie in the region between L and T, one has by Cauchy's
formula
__L /&£ =
__ L __ 伍演
，L/
2拓 £J —z 2拓 t z
l
J — 2高 t z
r
J —
CHAP. 12 FUNDAMENTAL PROPERTIES 283

where T = 上 denotes the union of the contours T and L and the positive
direction, on F is assumed to be clockwise ; the (—) sign on the right-hand
side follows from the fact that the region between T and L lies to the
right for motion along L and T in the positive direction.
It will now be shown that the integral
1
6 /(/)也z
2m J
r
力一
-
is zero. In fact, the value of I does not change, if the radius R of the circle P
is arbitrarily increased, since the function /(/) is holomorphic outside L.
On the other hand, as /(oo) = 0, one will have for sufficiently large | / 1

I /(91 <
I力I
where C is a positive constant. Hence, putting
t = R^, whence dt = \ dt \ ~ R\ \,
one has
2k 2B 27T
1
八他心 《
i [ CRdm c
2k J R I t — z |
|/|
0
〕£ — z I
0 0
R -— r
where 夕 = | z Thus, when R f 8, 7 f 0. But since I does not depend
on R, I 工0. Thus (70. 1 ') has been proved under the supposition /(00) = 0.
In order to prove (70.2') under the same condition, it will be assumed
that z lies in S+. Then
修)
t z' ——
considered as a function of t, is holomorphic in the region between L
and r. Therefore, by Cauchy's theorem,
i 1
0=
2冗〃 t
r+ 工
—z 2tc,
lr
t —z
|
2兀2 J t
—z
But the last integral, denoted earlier by 2, is zero, and hence the
required formula (70.2') follows for /(8) = 0.
If now = /(00) 卉 0, on applying the formulae just deduced to the
function /(z) — %, which vanishes at infinity, and noting that
284 IV. ON CAUCHY INTEGRALS §70

1 c祯t/ [0 for z in S-
2储 J t
工
z
— I % for z in S+ '
one obtains (70.1') and (70.2') for the general case.
With a view to the generalization of the preceding formulae, the following
terminology will now be introduced. Let a be some finite point of the
z plane and let the function /⑵ have the form
= G + %(z)
in the neighbourhood of this point, where
月1 月2 , “Z
G = z a FI (z a + +5 /z\

— —— b
•• •
产
倪

(Alf A2, . . 4% being constants). Then it will be said that /(z) has at
that point a 夕。岳 o/ order I with the principal part G(z) .
Similarly, if in the neighbourhood of the point z = oo, i.e., for suf¬
ficiently large | z the formula («) holds true, where now /。⑶ is ho¬
lomorphic near z = 8 and vanishes at that point, and where
G(z) = 4 + + A izl ... (c)

(4°, 月i，. . 4% being constants), then it will be said that /(z) has at
z = 8〃 pole of order I with the principal part G(z).
It will be noted that in the case of the ^point at infinity the constant Ao
has been added 力。the principal part.
A function 九⑶, holomorphic in the neighbourhood of the point a, may be
expanded in a series of the form
/o(z) = — a) + /(z — a” 十 ....
Even in the case when /(z) is holomorphic near the point at infinity, i.e., when
for sufficiently large | z |

/(z) = c() Z —/
1 - + • ,
••

the function /(z) will be said to have at z = 8 a pole of zero order with the principal
part cQ.

The following simple formulae will now be proved.

3°. Let /(z) be holomorphic in S+ and continuous in S+ + Z with the
possible exception of the points a1} a2, .. .,an of S+, where it may have
poles with the principal parts G/z). Then
CHAP. 12 FUNDAMENTAL PROPERTIES 285

—
1 / ' w)今
—— r /
27Vz J t — z
= /(z) G](z) — G2(z) ... — G*(z)
•~ for z in S+ (70.3)
L
and

— — Gz(z) — ... — Gn(z)

1
it z G](z) for z in S~. (70.4)
2tv^ J —
L

4°. Let /(z) be holomorphic in S~ and continuous in S~ + £ with

the possible exclusion of the finite points 旬，％，• an of S~ and also
the point z = co, where it may have poles with the principal parts
Gi(z), Gz(z), , Gn(z), Gg(z). Then

/丁 [胆竺 = 一/卜)
27r J t — z
2
Gi(z) + ... + G/) + Gg(z) forzinS- (703)
L

and

J“W
1
2大4 t—z
G](z) 十 … + G/z) + G^z) forzinS+. (704)
工

These formulae are easily established. Because of the similarity in the

proof, attention will be restricted to (70.3) and (703) ; the reader may
verify the other two formulae in a similar manner.
First consider (70.3). Applying Cauchy's formula to the function
/o(z) = /⑵ 一 Gi ~ ... — G£z)
which is holomorphic in S+, one finds (assuming z to lie in S+)

1 弟

2兀2/t —z，

But each of the functions G^z), k = 1, is holomorphic in S~

and vanishes at infinity, because these functions are of the form (b).
Hence, by (70.2'),
1
2K% J t
式
z /G-m
0, / 1,2,
'

%• .**,

工
286 IV. ON CAUCHY INTEGRALS §71

Thus

and so (70.3) follows. L

To prove (703), let T be a circle with centre at the origin and with
radius so large that L and the points z, %, an lie inside T. Ap¬
plying Cauchy's formula to the function
/o(z) ~ /(z) — Gi(z) — G2(z) — … 一 G〃(z)— Gg(z)
which is holomorphic in the region between L and T, one has (with the
former convention regarding the positive direction on T)
1 1 / f^dt
(力成
/o(z) = —
- -
~~ -
2在2J t -— Z 2九g J t z 2兀E J t z
l r
(assuming, of course, that z lies in But, by (70.2'), the last integral
vanishes, since %(z) is holomorphic outside T and vanishes at infinity.
Hence

…一 J
然） 60（力成
1
一五 =7
L
1 "(W
J t—
L
2

1 FG^dt 1 fG邓dt 1
… +三)不3 + 石J 厂二北
But all the integrals on the right-hand side containing • • •>
Gn(t), GgQ) vanish, since these functions are holomorphic in S+ and the
point z lies in S~. Hence

wj/州成
1
/。⑶ = — mL
and so (703) follows.

§71. On Cauchy integrals, taken along infinite straight lines.

Hitherto consideration has been restricted to integrals taken along
finite lines. There is no difficulty in extending the definition of Cauchy
integrals to the case where the line of integration goes to infinity ; it is
only necessary to study the question of convergence of such integrals
with infinite integration limits.
In the sequel only those cases of infinite lines of integration will be
CHAP. 12 FUNDAMENTAL PROPERTIES 287

considered which are straight lines. Without affecting generality it can

be assumed that the line of integration is the real axis. This case will be
considered below in detail.
Thus let L be the real axis and consider the Cauchy type integral

2兀vJ t —— z 2兀£ J. t
—— z
， '

L
- 8

in the present case 力 is a real variable, which assumes all real values, and
/(£) is a function (in general complex) of the real variable %:
超) - /1W + 巩肛
where /式力) and /2(^) are real functions. Unless stated otherwise, it will always
be assumed that /(/) is finite and integrable in the ordinary sense on every
finite segment of the straight line L.
For the present let it be assumed that z does not lie on L. The integral
(71.1) will converge uniformly, if, for sufficiently large [ t [, the inequality
B
"以 V77两 (7L2)
I立I
holds, where B and p, are positive constants. (This condition is, of couise,
sufficient, but not necessary.) In fact, in this case the integrand is of
order | t for large | t | and the above statement follows from a
known convergence criterion for integrals with infinite limits.
However, in the sequel the more general case will occur where /(/) f c
as|方|f oo, the limit c being the same for % 十 8 and for t
This limit will be denoted by /(8). It will now be assumed that for suf¬
oo. —
ficiently large [ 力

超) = c + (") = /(8) +。
(看｝ S
° 竹 > (7L3)

Then (7 1 . i ) will diverge, i.e.,

N"

J t—
N，
Z

will not tend to a limit as N' and tend independently of one another
to — oo and +oo respectively. In fact,
288 IV. ON CAUCHY INTEGRALS §71

Elementary reasoning shows that

N"
r dt y〃
/
工工 = ± + log 厂 ,
近
N，

where a (0 < a < tc) denotes the angle between the straight lines con¬
necting z with N' and Nf,
(Fig. 32) and /, y" the
distances of z from N' ,
The (+) sign refers
to the case when z lies in
the upper half-plane and
—
the ( ) sign to the case
when z lies in the lower
half-plane.
If N' and tend (in¬
dependently of one an- Fig. 32.
other) to — oo and + co
respectively, a tends to 兀，but log 夕 夕' does not have a limit. Hence the
preceding integral does not tend to a limit and the same may be said
with respect to the left-hand side of (a), because the first integral on
the right-hand side converges on the basis of (71.3). However, if Nf and N"
do not increase independently of each other, but if it is assumed that at
all times N' = — N", then log r"^ tends to 0 and
+N
lim dt 二［2 兀2c. (7L4)
Nf8 J t —z
—N
The expression on the left-hand side is called the Cauchy principal value
of the integral
-{-OO

力)夜 ［力)成
J — t Z J t —Z
— oo L
CHAP. 12 FUNDAMENTAL PROPERTIES 289

taken between infinite integration limits. future, when using integrals

with infinite limits, their principal values will be understood whenever these
integrals do not exist in the ordinary sense.
It has been seen that, provided (7 1 .3) is satisfied, the principal value
exists and
+8 十8

(7L5)

where on the left-hand side the principal value must be taken, while
the integral on the right-hand side exists in the ordinary sense ; the
—
signs (+) or ( ) must be chosen according to whether z is in the upper
or in the lower half-plane. (Note that for the definition of the principal
value it is not necessary to assume N' • ——
but it will be sufficient
if limM/N" = — 1.)
Thus the term "principal value” will be used in two different, but
analogous senses : when the integrand becomes infinite at some point
(as in the preceding sections) or when the integration limits are infinite.
Next suppose that the point z = % lies on the path of integration,
i.e., on the real axis E. Then the integral
+8

J
L
力一 M J— oo
f 一％

must be taken as principal value in both the senses stated above, i.e.,
its value will be defined as
4-oo

(71.6)
—
J
8
t
—
if that limit exists.
It is easily seen that

力 一，
= o.
—
8

The principal value (71.6) will clearly exist, if (7 1.3) is fulfilled and if /Q)
satisfies the H condition near %. It follows from (b) that the principal
value of (71.6) may then be expressed by either of the following formulae:
 _
290 IV. ON CAUCHY INTEGRALS §71

必
/津)
Jt (716)
—
—
8
%
or
4-00 十8

J
/2丝 ./
力 一 力0
一他)
― 0
(71.6〃)
— OO
— OO

where the principal values on the right-hand sides may be understood

only in one of the senses indicated above ; in the first case, it will be the
limit
8

at + /
J t —M J

since both integrals in the curly brackets converge ; in the second, it

will be the limit of the ordinary integral
+N
「 lim / 4 网 一他) 成, 小
Nf 8
—N
J 方 一 /0

since the integrand is now integrable in the ordinary sense.

Let /(/) satisfy (71.3) and, of course, the conditions of integrability
and finiteness imposed at the beginning of this section. Then F(z),
defined by
+8
i n(t)dt
-
(7L7)
2高J / z t —— z
— OO

will, obviously, be holomorphic in the upper as well as in the lower half¬

plane (but, generally speaking, not on L). Denote these half-planes by
S+ and respectively ; the boundary L will not be included with
either of these regions. The Plemelj formulae and the theorems on the
boundary values stated in § 68 are extended without difficulty to the
present case.
In fact, if 布 is a point on L (lying at a finite distance from the origin)
and if /(/) satisfies the H condition near this point,
CHAP. 12 FUNDAMENTAL PROPERTIES 291

F+（幻 = 豺（幻 + (71.8)

-4- co

f-（幻 = —豺
去/胃 —OO
(7L9)

F+(%) and F~(幻 denote here the limiting values of F(z) as z -> 布 along
any path on the left and right of L respectively, i.e., in S+ or S~. Further,
if /(/) satisfies the H condition on some segment of L, F+(%) and F-(/o)
satisfy the H condition there, except possibly near the ends of the
segment. The statements in the Notes at the end of § 68 will also remain
true in the present case.
In order to verify the correctness of (71.8) and (71.9) and of the sub¬
sequent statements, it is sufficient, for example, to represent the integral
(71.7) in the form (71.5) and to divide the integral on the right-hand side
into two integrals : the one to be taken over a finite segment, containing
tQ, the other over the remaining part of the straight line.
Hitherto, when speaking of the behaviour of the function F(z) near
a point of the boundary L and of its boundary values, points in the finite
part of the plane have always been implied. In order to study the behaviour
and the boundary values of F (z) near the point at infinity (which in the pres¬
ent case lies on L), one may, for example, proceed in the following
manner.
Introduce the coordinate transformation
z = — f
(71.10)

then the point 已 = 0 of the 乙 plane corresponds to the point z = 8

of the z plane and vice versa ; the real axis of the z plane becomes the
real axis of the 已 plane and upper and lower half-planes correspond to
—
one another; when the point z t travels along the real axis in the positive
direction from t —— 8 toi( = 十 oo, the corresponding point

1
a= (7L10‘)

of the C plane also travels along the real axis in the positive direction as
follows : from a = 0 to b = 十 8, from a —— oo to = 0 (since the
292 IV. ON CAUCHY INTEGRALS §71

—
points b = 8 and
E plane).
十 8 represent the same point C = 8 of the
工

Introducing the transformation (71.10) in (71.7), changing the inte¬

gration variable in accordance with (71.1O') and introducing the notation

F =F 尸
/©=/（—；）= 产⑹，（7LH）
one finds
+8

(71.12)
一OO

Assuming for the time being (in order to simplify the reasoning)
that /(/) satisfies the H condition at the point / = 0, it is easily seen
that one can rewrite the preceding formula in the form
+8 +8
f* do 1
(71.13)
cr一 E 2兀勿 J o '
©o

all these integrals must be taken as Cauchy principal values. Obviously

these will exist, if, as it has been assumed, /Q) satisfies the H condition
for all finite t and the condition (71.3) for large | 力. The second integral
on the right hand side of (71.13) is constant, and hence the study of the
function F(z) near z = 8 is reduced to that of the integral
+ ©o
i f f* g
(71.13')
2九E J —— 已
— OO
CT

near E = 0, i・e., to a problem discussed earlier.

In general, the study of the integral (71.1) may be reduced to that of an integral
of the same form, taken over a finite closed line, e.g., a circle. For this purpose it is
sufficient, for example, to introduce the transformation

z + £ =
_ *

Then the real axis L of the z plane becomes the circle Lof the C plane which is
〈=
tangential to the real axis and passes through the point
(71.1) takes the form
i, and the integral
—
2m J a—C 2脑 J o+ N
，
CHAP. 12 FUNDAMENTAL PROPERTIES 293

where

产⑹
/
— \

Note that (Z) transforms the half- plane S+ on to the region bounded by the
circle I.

In order to utilize immediately the earlier results, impose on /(/)

the condition that 产⑹ is to satisfy theH condition near o = 0, i.e., that
1产值) 一 产(q) | 4 B | 见 一 ％ F， 0 V M 4 1
This leads for /(/) to the condition

"(%) — /㈤ I
1 1
“ 0 < p. 1 (7L14)

for sufficiently large| 々 I, | % I ；(71.14) will be called the H condition for

the neighbourhood of the point at infinity. Note that (71.3) is obviously
a consequence of (71.14) for the neighbourhood of the point at infinity,
but that the converse statement is not true. (71.3) may be called the
H condition for the point z = oo (but not for its neighbourhood) .
Assuming that /(%) satisfies the H condition in the neighbourhood of
the point at infinity, i.e., the condition (71.14), it will now be shown that
the boundary values of 尸(z) exist when z tends to infinity along any
path which remains in either the upper or lower half-plane. These
boundary values will be denoted by F+g) and F~(oo) respectively
and (71.13) will be used to prove their existence and to calculate their
values.
If z f co, remaining in the upper or lower half-plane, then E f 0,
also remaining in the upper or lower half-plane. Hence, applying (7 1.8)
to the first integral on the right-hand side of (71.13), one obtains
+ oo +8

、
^+(oo) = 尸
77*+仆
*+(0) = 豺*(0) + F
1
2m J
/
a 2九，J
1
/
/7*(6加 ,
o
-
oo
-oo
and the first of the following formulae is deduced :
F+（8）= 豺（8）, 尸-（8）=
— 豺3）； （7L15）
the second formula may be proved in an analogous manner. Obviously,
one can now discard the assumption regarding the behaviour of /(/)
294 IV. ON CAUCHY INTEGRALS §71

near the point t = 0, adopted temporarily above.

The following property of the integral F as defined by (71.7),
will now be noted. Suppose that not only /(/) but also the product //(%)
satisfies the H condition near the point at infinity.
It is easily seen that, if //(£) satisfies the H condition near the point at infinity,
then also satisfies that condition ; in addition, obviously f(8) = 0, so that,
by (71.15), F+(8) = F-(8) = 0.

Under these conditions the product zF{z) tends to a definite limit as

z co along any path remaining in the upper or lower half-plane. In
fact, putting
施)=/他)， (71.16)
one has
-boo +8 +8
1 C fy{f)dt 1

oo
— z) J
2冗， t
-OO
z— 2冗乙
8

whence, by (71.15),

lim [zF(z)] = 土 豺式8)一

8
土豺工(8)
「
不 — J I M弭 (71.17)
—co

where the upper or lower sign must be chosen according to whether z

remains in the upper or lower half-plane.
This formula may also be written

in each half-plane, (71.18)

where 4 is a constant (which may have different values in the different

half-planes) and o(l/z) indicates, as always, that z.o(l/z) tends to zero
as | z | grows beyond all bounds.
Similarly, it may be shown that, if in addition to 00) also the product
句⑹ = %(£) (71.19)
satisfies the H condition in the neighbourhood of the point at infinity,
then
尸⑶
—— W + 。g) in each half-plane, (71.20)
CHAP. 12 FUNDAMENTAL PROPERTIES 295

where A is the same constant as in (71.18). It is readily seen that one

may limit the proof to the case when the derivative /'(力) is continuous
for all values of i, and not only at the point at infinity*.In fact,integrating
by parts, one obtains
-1-00 +8

-2 JQ 兀 — z)2 — 2m J t—z
— co co

whence, noting that

1 浮 1 £
±—Z ?2(力 一 Z)Z Z2 '
one easily deduces
+8 +8
z2F (z) =
——r/
2m J t — z
— OO —oo

-
Letting z -> 8,one finds by (71.15)
广
]
lim 团f⑶] 土 沙2(8) — — 牛份次
々一A ZTCZ J
— 8

and it is easily verified that the right-hand side agrees with the right¬
hand side of (71.17),taken with the opposite signs.[Using the substitution
£ = — 1/g, it is seen that /2(oo) = — ^(oo).]
It is just as easily shown that, if in addition to (71.16) and (71.19) the
following relation also holds true:
阴"(9 = %&)， (71.21)
where 加力) satisfies the H condition in the neighbourhood of the point at
infinity, then
2A
F〃(z) = (71.22)

where A is the same constant as before.

* In fact, otherwise one may replace the function /(£) by another function /0(Z)
with the stated property which differs from /(/) only on some finite interval
检 £ 机 The estimate for the difference of the corresponding integrals is quite
elementary.
296 IV. ON CAUCHY INTEGRALS §72

These results can be formulated briefly as follows : under the stated

conditions both sides of the equality (71.18) can be differentiated where
differentiation under the sign o is admissible.
The generalization of the above work to any order derivatives is
obvious ; however, only derivatives up to and including the second order
will be encountered in subsequent chapters.

§ 72. On Cauchy integrals, taken along infinite straight lines

(continued). A number of formulae, analogous to those of § 70, may
be deduced, in order to simplify the calculation of Cauchy integrals, taken
over an infinite straight line L. Consideration will be limited here to the
simplest of these formulae which may easily be generalized by the
reader.
1°. Let /(z) be a function, holomorphic in S+ and continuous in
S+ 十 L including the point at infinity, and let /(8) = a. Then

/ z =
/(z) —— for z in S+, (72.1)
2rcz J
z,
/
—
1 f 超)今
r / = 一 ia for z in S~. (72.2)
2m J t z ，

2°. Let /(z) be a function, holomorphic in and continuous in

十 E including the point at infinity, and let /(oo) = a. Then

=韶 for z in S+, (72. T)

27Tz J t Z
L

1 7 超减
-~~r
L
t
— z = — /⑵ + W for z in S-. (722)

The condition that /(z) is continuous in S+ 十 E [or in S~ + L] and

at z = co may be expressed as follows :
/(z)=/(8)+o =口+。 for Zf8 in S++L [or in S-十上]. (72.3)
(The notation o(l) denotes a quantity which tends uniformly to zero
as |z] -> oo; cf. §65, 4°).
The formulae (72.1) and (72.2') may be called Cauchy formulae for
the regions S+ and S~ respectively.
CHAP. 12 FUNDAMENTAL PROPERTIES 297

Formula (72.1) will now be proved. Draw about the origin as centre a
circle with sufficiently large radius R, so that the point z lies inside.
Consider the contour F, consisting of the segment AB of the real axis
contained in the circle and of the semi-circle, lying in S+; select the
positive direction on F in such a way that AB is in the direction Ox.
Since, by supposition, the point z is inside r, one has by Cauchy's formula
一
力)就 i r /(/)威
t
—z 2兀， J t—z
Y

where 丫 is the semi-circle, forming part of the path of integration.

The second integral on the right-hand side tends, thanks to (72.3),

—
to the limit
iti
a.
27n
7 = la
高 as R f 8 ；
the first term then tends to a definite limit, as R oo, and this limit
is given by
— ^a. But
+_R

i /他今 _r i "(9近
27rz J t— z
lim 7 / lim
— —
"

冗f oo 27Vz , t z Rf 8
NB ~R

is also, by definition, the principal value of the integral

1
2兀/
“阿
J t—z
and so (72.1) is proved. Note that also the existence of the principal
value of the preceding integral has been proved by this argument ; this
was not obvious beforehand, since in the present case /Q) is subject
to the condition : /Q) = a + o and not to the condition: 超) =
=a 4- 「吁 under which the existence of the principal value had
been proved earlier.
The other formulae of this section can be proved in an analogous
manner.
 Chapter 13

BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS

§ 73. Some general propositions. Let E be a simple contour,

S+ and the finite and infinite parts of the plane, bounded by L ; let the
positive direction on L be such that S+ remains on the left. The contour
L will not be included in S+ or S~. Further, let
用= /1W + 为4)
be a continuous function given on L.
Consider the question as to whether /(%) can be the boundary value of
some function F(z) = U(x, y) + iV(x, y), holomorphic in S+, where
reference here is, of course, to boundary values as z f 力 from S+.
It is easily seen that, in general, this cannot be the case, if the con¬
tinuous function /(%) is otherwise arbitrary. In fact, it is known that it is
sufficient to give the boundary value /式力) on L of a function U(xf y),
harmonic in S+, in order to completely determine this function ; but
then also its conjugate function V (%, y) will be completely determined,
neglecting an arbitrary constant term, and hence also the boundary
value 府心) of this function, if indeed it exists. Clearly the roles played
by /也) and /2(^) may be interchanged.
The problem of determining a harmonic function from its boundary values
represents the well-known Dirichlet problem. Also note that it does not follow from
the existence of the boundary value of U[x, ”) that those of its conjugate function
V (#, 夕) exist.

It follows from the above that only one of the two real functions
/1«), 府(£) may be given arbitrarily, if the function /Q) = f曲 十 牙2«)
required to be the boundary value of some function, holomorphic in
S+. Hence it will be of great interest to find the necessary and sufficient
condition that a continuous function /Q), given on L, represents the
boundary value of some function F(z), holomorphic in S+; an analogous
question will arise with regard to the region S~. The following theorems
answer these questions:
298
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 299

I. A necessary and sufficient condition for g continuous function

/(£), given on L, to be the boundary value of some function, holomorphic
S+,

-
0 for RZ z in (73.1)
t z
L

II. A necessary and sufficient condition for a continuous function

/($)，g谢e% 0% L, /o be the boundary value o/ some function, holomorphic
in S~ (including the point at infinity), is
1 f fM
——
2tw
r /
J t — z =a /
"z
for all •
&n
a
S+, (73.2)
L

where a is so/e constant which is equal to the value of the above-men¬

tioned holomorphic function at infinity.
These propositions are almost obvious on the basis of the results
of the preceding sections. In fact, if 超) is the boundary value of some
function holomorphic in S+, condition (73.1) holds true by (70.2) ;
hence (73.1) is necessary. It is also sufficient, for, assuming it to be
fulfilled, one may write
中 、
1 八超）北
(73.3)
L

Taking into consideration that F(z) = 0 for z in S^, and hence 尸一㈤ = 0
on L, one obtains from (68.4) and Note 2 of § 68 that
(幻 = Wo)，
i.e., if (73.1) is satisfied, /Q) represents the boundary value F+(力) of the
function 尸⑶，defined by (73.3).
The second theorem" may be proved in an analogous manner. If /(/)
is the boundary value of a function, holomorphic in S, (73.2) is neces¬
sary by (702) ; it is also sufficient, since, if it is satisfied, the function

J/2^
i
尸⑶ = 4' (73.4)
—
d
2兀z / •

is holomorphic in S~ and takes the boundary value 六(幻 = /&)；

the last conclusion follows from (73.2), (68.4) and from Note 2 of § 68.
Hitherto it has been assumed that the function /(/) is only continuous.
If, in addition, it is assumed that it satisfies on L the H condition (§ 65),
300 IV. ON CAUCHY INTEGRALS §73

then (73.1) and (73.2) may be given a new form which is in many respects
very convenient. In fact, denoting by tQ some point of L and performing
in (73.1) and (73.2) the limiting process z f 环 from S+ and respectively,
one obtains, on the basis of the Plemelj formulae (§ 68),
1 / /加 =0
2兀2J 卢
一时田 十 5V (731)
-%
% 丁
L
and
春超 —
1
•—
力核 7 . (73.2')
—
I d
2兀2 J t 给
工

respectively (for all on £). These conditions are equivalent to the con¬
ditions (73.1) and (73.2). In fact, (73.1') expresses that the boundary
value of the function
1 小心
声⑵ =
2rzi J t
L
z —
holomorphic in S~, is zero along the entire boundary L of S~; hence,
applying Cauchy's formula to or from §37, 2°, 尸⑵ =0 throughout
S~, which is the condition (73.1). Similar reasoning applies to (73.2) and
(732). The conditions (73.1') and (732) were stated by J. Plemelj [1].
So. far it has been assumed that L is a simple contour. Consider now
the case when L is an infinite straight line and let the real axis represent
this line. As in § 71, let S+ and represent the upper and the lower
half-planes respectively. The following theorems are easily proved in a
manner analogous to that used in the preceding proofs.
Let /0) be a function, continuous on L, for which for large 1 1 1
刖 = 々 十 0([力+) = /(oo) + 1 1^), (73.5)
where a and pc are constants and [j. > 0. Then
III. A necessary and sufficient condition for the function /修) to be
the boundary value of a function, holomorphic in S+ and continuous in
S+ + L {including the point z = oo), is
1 / 力)欣
—— — 妥a for all11 z m S".
•
= if S M
(73,6)
27Vz J/ t
—
r
z
工

IV. A necessary and sufficient condition for the function /(力) to be

the boundary value 夕 function, holomorphic in S~ and continuous
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 301

S- 十 L {including the point z = oo), B

/ = 初 for all z in S+. (73.7)
L

If, in addition, /(£) satisfies on L, including the point at infinity [cf.

(71.14)], the H condition, (73.6) and (73.7) may be replaced by
1 C f 出祝
- i/M、 + 5一 卢丁 (736)
—— 。
= 一细
2兀z j t 力
L

/⑴〃力
看顺+H 初
and 1
(73力
L

respectively, where tQ can be any point of L.

The proofs of these theorems will be left to the reader.

§ 74. Generalization. The formulae and theorems of the preceding

section, referring to the case of regions bounded by one simple contour,
can immediately be extended to the case when the boundary consists
of several such contours.
It is easily seen that the conditions (73.1), (73.2), (73.1') and (73.2')
remain valid, if S+ is a connected finite region bounded by simple con¬
tours Li，乙2 …， 几,Lm+1 which do not intersect each other and the last
of which surrounds all the others, if L is the union of these contours and,
finally, if is the part of the plane which is the complement of the
region S+ + L with regard to the entire plane. Thus the region S~ con¬
sists of the finite regions Sf, S* . . . , S~ , bounded by Llr L2, .. Lm
respectively, and of the infinite region +1, consisting of the points

outside Lm+1. The function F(z), holomorphic in S~, must then be con¬
ceived as the union of the functions, holomorphic in S[, S$, . . +1.

§ 75. Harnack's theorem. A theorem which is frequently used and

which is due to A. Harnack [1] follows almost immediately from the
results of the preceding sections.
Let L be a simple contour and let S+, S~ be the finite and infinite ^parts
into which the plane %s divided by L (which does not itself belong to S+ or
S-). /(/) be a real and continuous function on L. Then, if

2兀三
r
J
/ 力一 z =
0 for all z m S+, (75.1)
L
302 IV. ON CAUCHY INTEGRALS §75

f(t) = 0 everywhere o% L. Also,

1 C /团业
/J t z = 0 ，
for all z in S~, (75.2)
2兀2
L —
力) = const, o% L.
In fact, it follows from (75.1), on the basis of the results of § 73, that
f(t) is the boundary value of some function F(z) = U y) + iV(x, y),
holomorphic in S' i.e., /R) = U~ + iV~. But since /传) is a real function,
the boundary value V~ of the function V(x, y), harmonic in S~, is zero
everywhere on L. Hence V(%,y) = 0 everywhere in S 〕 Therefore U C -- —
— const, in S~, and hence /(力
in (75.1) and noting that
—
U~ = C on L. Substituting this value

_L /0— = c

it is verified that C = 0.
2m J
L
t
— z '

It may be shown in the same manner that it follows from (75.2) that
/(/) = C = const. ; however, in this case it is impossible to conclude that
C = 0, since, substituting /(/) = C in (75.2), one obtains the identity
0 = 0.
Thus the theorem is proved. It will be left to the reader to generalize
it to the case of the regions considered in § 74. In that case it follows
from (75.1) that /(/) = C k on LJ: (k = 1 , 2, , m), /(/) 0 on Lm+l)
and from (75.2) that /(/) = C on where C, Cv C2, . . . , are constants.
—
It is also easy to formulate a theorem, analogous to the preceding one,
for the case when L is an infinite straight line.

NOTE. 1. The following conclusion follows directly from Harnack's

theorem (having in mind the case when £ is a simple contour) .
Let /#), /2G) be two real continuous functions, given on L. Then, if

£ /谯 击/竺?
L
工

L
for all z in S+, (75.3)

/1«) = /2O on 二；also, if

1
/
2m J t~z
L
f
f^dt
= ——
1 ，
2tzz J t~z
L
,
. °
/ t^dt for all z m S~,
” (75.4)
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 303

/2(9 = /iW + const, on L. This result is verified by applying Harnack's

theorem to the function 力0) /式力
—
NOTE. 2. It is not difficult to show that the preceding theorem
remains true, if it is not assumed that /0) is continuous, but if it is allowed
to have a finite number of first order discontinuities. This case will not
be considered further here and it will only- be noted that the theorem, if
properly formulated, will hold for much more general conditions.

§ 76. Some special formulae for the circle and the half-plane.
When E is a circle or a straight line, the formulae of § 75 may be given
a form which is convenient for future applications.
1°. First some special notation will be introduced. Let
F(z) = U(%, y) + iV{x, y) (76.1)
be a function of the complex variable n, defined in some region of the
plane z. Then F{z) [where the bar only
extends over F] is to denote the function,
having the conjugate complex value of
F(z) at the point 5, which results from a
reflection of the point z in the real axis, i.e.,
which is simply the conjugate complex
value of z (Fig. 33).
Thus, by definition,
夙z) = 丽 (76.2)
or
凤z) = U(%, — y) — iV{x, — y).
For example, if F(z) is a polynomial
(76.2')
Fig. 33.

F(z) = a^n + 旬 十 ... + an> (76.3)

then obviously by (76.2)
凤z) = d* + 匹*t +… + 2， (763)
i.e., 歹(z) is obtained from F(z) by replacing the coefficients by their
conjugate complex values. Similarly, if F(z) is a rational function

F(z) =
々庐
笃
+ 诙 + ... + an (76.4)
bo2n + *7 + •.. + 廉
304 IV. ON CAUCHY INTEGRALS §76

then
祀
然一 、 都”
%一尸十 …十％
+ 十… +以
It is easily seen that, if 尸(z) is holomorphic in some region S, F(z)
is holomorphic in the region S, obtained from S by reflection in the real
axis (cf. Fig. 33).
In fact, putting
F = U^x, y) + iV1 {x, 夕)，
one has by (76.2')
U^x, y) = U(%, — y), 0式舟, 夕) =
— V{x,
—训 •
Hence, if U(x, y), V {x, y) satisfy the Cauchy
—Riemann conditions
2U 9V QU &V
一
粉 纱' 如一 Qx

in S, then 夕)， V^x, 川 will satisfy the Cauchy

—Riemann conditions
_ a匕
a% „

8” Zy ' Sy 2%
in the region S.

If the function F(z) is holomorphic in S, except at certain points where

it has poles, the function _F(z) will have the same properties in g and
its poles will be at points obtained from the poles of the function F(z)
by reflection in the real axis.

-
Note also that the function F(z), conjugate complex to F(z), may be
represented as
诟 立⑶ ； (76.5)
this follows from (76.2) by replacing z by 5.
Now suppose that the function F(z) is defined in one of the half¬
planes S+, S- into which the z plane is divided by the real axis, say,
in the region S+. Then the function 尸(z) will be defined in the region
S-. Further, if the boundary value F+(t) exists, where t is some point of
the real axis, it follows immediately from (76.2) that also the boundary
value exists and that
人) = 环② (76.6)
(since, if in (76.2) z f / from S^, z f t from S+).
Obviously the roles played by S+ and S~ may be interchanged ; in
that case one will have
户 = (76.6')
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 305

2°. Let 丫 be the unit circle with centre at the origin of the plane
of the complex variable〈；the points of 丫 will be denoted by a so that

一
ct = e吗 《 0 3 v 27r. (76.7)
Denote by 2+ and 2 the regions |^ | < 1 and [ | > 1 respectively and
^
choose the positive direction on 丫 so that the region S+ remains on the left.
Let 尸(0 be a function, defined in Z十[or £-]. Consider the function
F*(Q, defined in S_[or £+] in the following manner :

/9 = F (|) (76.8)

or, remembering the meaning of the symbol F,

尸 *9 = F (I). (768)

The last formula shows that 尸*(。may be defined as follows : the

function takes values, con¬
jugate complex to those of F(Q
at points which are reflections
of the point 已 in the circle 丫
(Fig. 34). [It will be remembered
(§ 48, 1°) that the reflection of
the point 已 in the circle 丫 is the
point 7 = 1R, because in the
present case the radius of the
circle is unity.]
It is easily seen that, if F©
is holomorphic in W+[or S_], the
function F*(3) is holomorphic in
S~[or E+], and vice versa. For
example, if 尸依) is holomorphic
in E+, it may be represented by the series
F© = 曲 + 咄 + 喈2 十 … (76.9)
which is absolutely convergent in E+, i.e., for [
F*(E) will then be represented by the series ^ ] < 1 ; the function

F* (9 - =魅 +令 咨十 …，

absolutely convergent in S", i.e., for | |

^ > 1.
306 IV. ON CAUCHY INTEGRALS §76

Now suppose that 万(0, defined in W+, has the boundary value F+((y)
for C f b, where a is a point on 丫. Then it is easily seen from (768) that
also the function F*(Q has the boundary value F~ (cr) , defined in and
K(o)=F丽， (76.10)
because, if in (76.8') C 今 o on 丫 remaining in S~, then U = 1/C tends to
1历 = b remaining in £+. Clearly the roles of E+ and may be inter-
changed ; instead of (76.10) one will then have
琦⑹ = 声问. (76.11)
3°. Using the fact that every function, holomorphic in S+ [or
corresponds to a function 歹(1/Q, holomorphic in [or-S+], one may,
in the case of circular boundaries, modify the formulation of the pro¬
positions I and II of § 73 which hold in the general case. In fact, the
following theorems are easily proved:
I. 幺 necessary and sufficient condition /oy the function /(a), continuous
on the circle 丫，/o be the boundary value 。/ some function, holomorphic inside

/ 而击 =
1 f
j for all C inside 丫， (76.12)
Y

where a is a constant which is equal to the value of the above-mentioned

function at C = 0.
IL A necessary and sufficient condition for the function /(b), continuous
o% the circle 丫，to be the boundary value of a function, holomorphic outside
Y，is

——r / —- ~
=0 for all C outside 丫. (76.13)
y

The conditions (76.12) and (76.13) follow directly from the conditions
(73.2) and (73.1) and from the statements of the present section. For
example, if /(a) is to be the boundary value 尸+(cr) of some function F©,
holomorphic inside 丫， the function /(a) must be the boundary value
尸京⑹ of the function F*(C) = 歹(1/〈)，holomorphic outside 丫；this follows
directly from (76.10). Hence, applying (73.2), one obtains immediately
—
(76.12), where a F*(8) = 尸⑼ = F(0).
The condition (76.13) may be proved in an analogous manner. However,
one special point must be noted : let (76.13) be fulfilled and let it be
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 307

required to find the function F(9，holomorphic outside 丫 and taking

the boundary value /(a) on 丫；if one wants to use for this purpose Cauchy's
formula for the infinite region (§ 70), viz.

碎）
G
-- C + （8）,
尸 (76.14)

one has to know F(oo). As is easily seen, this quantity is given by

1
F（8）= 2兀， (76.15)

Introducing (76.15) in (76.14), one may write

= - 2兀％ J
-
F(Q (76.14')
g(o Q
By (70.2'), for 二 inside 丫，
1 /⑹说
—
2k，
= F(oo),
j C7

whence,for E = 0,one obtains (76.15).Thus (76.15) may,obviously, be replaced by

"
F3)= j 等'
where 焉 is any point inside 丫.

Note the following formulae which will be used in the sequel. Let

9(q
- «0 + «1^ + • • = ?(0) + ^z(0) + vy ?"(0) + … (76.16)
•

be a function, holomorphic inside and continuous up to y. Then

占
2 z J
注
/ 四单
o -— G
匹y+ 藏y—i + ... +^/ 3=0,1,2, . . .) (76.17)

for all C inside 丫. In fact, c邓 is the boundary value of 已他 (1®, holo¬

morphic outside 丫，except at the point 已 = oo near which it has the form

C*中(《) (左° = C密 -{ — ~
+ •••
= 左oC + &上"1 + •• +蕊+
°
308 IV. ON CAUCHY INTEGRALS §77

and (76.17) follows immediately from (704). In particular, for 々 = 0,

one has for all 乙 inside 丫

— q = 中(°)。
(76.18)
7； /
—
"

2m j o
Y

This formula is the same as (76.12), but written somewhat differently.

4°. As before, using the fact that to every function F(z), holomorphic
in the upper [or lower] half-plane S+ [or S~], corresponds the function
尸(z), holomorphic in the lower [or upper] half-plane [or S+], one
deduces the following propositions from the conditions (73.7) and (73.6).
As in § 73, let /(/) denote a function given on the real axis L, where it
is continuous and such that for large [ t [
/(/) = 霖 + O(| 力*) = /(co) + OQ 「)， / m = const. > 0. (76.19)
Then
III. A necessary and s对 condition for the function /(/) to be the
boundary value of a function, holomorphic in S+, is

= 躯 for all z in S+. (76.20)

L

IV. A necessary and sufficient condition for the function /(/) to 方e the
boundary value of a function, holomorphic in S~, is

—— for all z in S~. (76.21)

§ 77. Simple applications: solutions of the fundamental pro¬

blems of potential theory for a circle and half -plane. As simple
applications of the preceding results the solutions of the fundamental
problems of the theory of the logarithmic potential will now be given
for the cases of a circle and a half-plane.
The first fundamental problem (Dirichlet problem) consists of determi¬
ning a function, harmonic in a region, when its boundary values are
given. (The solution of this problem for the circular ring by use of in¬
finite series was stated at the end of § 62.)
The second fundamental problem (Neumann problem) consists of
determining a function, harmonic in a given region, when the boundary
values of its normal derivative are given.
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 309

1°. First fundamental problem for a circle.

For simplicity, let the radius of the circle be unity and its centre at the
origin. As before, denote the circumference of the circle by 丫 and its
points by b = e谣；other points of the plane will be denoted by J Let
the unknown harmonic function be P and its conjugate complex function
Q. The latter function is known to be determined apart from an arbitrary
constant, if the function P is known. Finally, put
= p+ (77.i)
where must be holomorphic inside 丫.
By the condition of the problem, the unknown function P must take
the definite boundary value P+, as C tends to the point o of 丫 (from the
inside of 丫)，which must be equal to the real function /(b) or /(9), given
on y；it will be assumed that the given function /(A) is continuous on 丫.
Hence the boundary condition of the problem may be written
p = /⑻， (77.2)
where, for simplicity, P has been written for P+.

As a matter of fact, if it is assumed that P takes (definite, finite) boundary

values for points on 丫， then the given function /(9) must necessarily be
assumed to be continuous ; this follows from, the statements in § 37.

The problem will now be restricted by assuming that not only the
function P, but also its conjugate complex 0, and hence also the function
尸(0 take definite boundary values. (This condition is not necessary and
has only been introduced to simplify the reasoning.) Denoting F+(g)
by F(a), the boundary condition (77.2) may now be written
F(b) + 丽 = 2/⑻. (77.3)

Multiplying (77.3) by 二 -
2兀4 o Q
where 乙 is a point inside丫,and integrat-
ing around 丫，one finds
1
/ F⑹而 1 CF b 1 / /⑻说
2m J a —— 已
丫
1
2兀E J
丫
a —已 大匕 J
丫
a ― '

On the basis of Harnack's theorem (§75), this condition is completely

equivalent to the preceding one. (Cf. § 75, Note 1.)
By Cauchy's theorem, the first integral on the left-hand side is equal
 310 IV. ON CAUCHY INTEGRALS §77

to F© ; by (76.18), the second integral equals

f⑼ = 吃 一 洛,
where a0 and 仇 are real (for the present, unknown) constants. Thus

L
m J 且经
尸(q =
c
—G
Y
一气 +油. (77.5)

There remains still to determine a0

in (77.5) which gives
—— ^0. For this purpose put 已 = 0
2H
1 f /⑶而 1
f )
兀勿 J a tv J
y o
Thus a0 may be determined from this formula ; the quantity 3°, however,
remains quite arbitrary, as was to be expected, because the function
0, conjugate complex to P, was determined by P apart from an arbitrary
real constant, and hence must be determined apart from an imaginary
constant.
Introducing the value of a0 in (76.5), one finds

This last formula is the well known Schwarz formula ; the unknown harmonic
function P is obtained from it by separating real and imaginary parts

P = 猊 尸(已) ~ fR-—-
27cz J
/ /(8)
c
—— — —.
Q a
• (777)
Y

It has only been proved that, if the solution of the problem satisfying
all the imposed conditions exists, it is necessarily given by this formula.
There remains to prove that this formula actually gives the solution.
This will be done, assuming that /(A) satisfies the H condition. In this
case, on the basis of the statements of § 68, the function F(Q, determined
by (77.7), takes definite boundary values which satisfy (77.3); hence
尸 satisfies (77.2).
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 3口

The existence and uniqueness of the solution may be proved for much more
general conditions (it is sufficient, if /⑻ is continuous), but no space has been devoted
to this here, since this problem is considered in any textbook on complex function
theory or potential theory.
It follows from the equivalence of the conditions (77.3) and (77.4) that' the
boundary values of F© satisfy (77.3). That F(Q satisfies (77.3) may be verified
directly on the basis of the Plemelj formulae (68.2). In fact, denoting by a0 = &诒。
some point on 丫，one has

= /(%) + —
1 /⑻ do
八^一
1
I / /⑻说
…孙 =
——
F+"o) T
MJ b CT0 赢
Y Y
+
-
1 C , a da .
、
+ /
= 2m J
y
/(分)
ct
— b0 b
k 唱l

Writing under the integral sign a = = 次硼，one finds

F+(%) = 他) + 二
27rz
1
J/
/ /⑻ cot
————
9 瓦
-
2 + 泯 = 俭) 十 an imaginary quantity,
whence ° 沆F+&) = /(%)•
Substituting in (777)
a = 净， 已 = peR 说 — 渥黑舟，
one easily deduces Poissons formula

p=_L / (l—p2)/网粉
(77.8)
2兀 J 1—
o
2P cos (9 山) + p2 •

which solves the above problem without the use of complex variables.
2°. The second fundamental problem for a circle.
Let F(Q denote the same function as in 1°; it will be assumed that
the derivative F'(Q takes the definite boundary value F'(o).
One deduces from the equation

通212P = F(E) 十丽 = F(pe巧 砺巧

2 k = 法叩'(pe巧 + = 上 F'(。十 — 万 (77.9)
S P P
The boundary condition of the problem under consideration has the
form (denoting by n the outward normal)

= 0%
= /⑻ or — = /(9) on
dp
丫， (77.10)
312 IV. ON CAUCHY INTEGRALS §77

where /(3) is a given continuous function. By (77.9) this condition may

be written
oF'3) + == 2/(8) on 丫. (77.11)
In the same manner as in 1° one obtains
1 I f °F'
2九％
一 C + 京 J 工一石 —=
where C is an arbitrary point inside 丫；hence, applying Cauchy's formula
and (76.18) and noting that cF'(b) vanishes for g = 0, one finds

(77.12)
6
-C
This formula determines 尸'(0 and shows that the right-hand side must
vanish for = 0, if the problem is to have a solution. This means that,
in order for the problem to be possible, one must have
2k
/⑻说
CT
=o //
o
= 0. (77.13)

In contrast to the Dirichlet problem, the Neumann problem does not

always have a solution, but only when (77.13) is satisfied.
If the condition (77.13) is satisfied, the function determined by
(77.12), will be holomorphic also for E = 0. The function F(Q is deter¬
mined by integration

碎)=L/手(半4。陪
J C J —C b
修.⑷
Y

where const, denotes an arbitrary complex constant. The value of the

unknown function 尸 = 猊 F(0 is thus determined apart from an ar¬
bitrary real constant. This was to be expected, because, if Pis a solution
of the Neumann problem, P + const, will obviously be a solution of the
same problem.
It is easily seen (cf. the preceding problem) that the formulae, just
obtained, actually solve the stated boundary problem, if, for example,
the given function /(») satisfies the H condition.
CHAP. 13 BOUNDARY VALUES OF HOLOMORPHIC FUNCTIONS 313

Inverting the order of integration and evaluating the inner integral

on the right-hand side of (77.14), noting that
广 苑 1 1
/ = — log c log (b — Q + const.,
and using (77.13), one obtains the formula which was stated by T. Bog¬
gio [4]
1 c
/ /(8) log (b — 0
而
F(0 = ~ - F const. =
兀2 J G
Y 2k

—— //(8)log(b — + const. ; (77.15)

九 J
o
separating real and imaginary parts gives the formula of U. Dini [1]
2n
_
P= — JI
k
o
/(眇) log y 淄 + const., (77.16)

where y = | b
—
| and const, is an arbitrary real constant. However, in
most applications, it is convenient to use the formulae (77.12) and (77.14).
3°. The first and second fundamental problems
for the half-plane may be reduced to the corresponding problems
for the circle by means of conformal transformation (cf. § 71) or may
be solved directly by a method, analogous to that used in the earlier
problems for the circle. In view of the complete analogy with the above
work, only short remarks will be made here.
Let /(/) be a real continuous function, given on the real axis L, and let

—
it be required to find P{x, y), harmonic in the upper half-plane S+ and
taking the boundary value P+ = /(/) on L including the point at infinity,
so that for z oo (in S+ + L) P -> a, where 应 is the real constant

Introducing the function of a complex variable

F(z) = P + iQ,
口 = /3).

(77.18)
(77.17)

holomorphic in S+, and assuming that this function has a definite boun¬
dary value F+0) for all points of L, including the point at infinity, one
can write down the boundary condition of the problem
尸
+ = 2/(/) on 乙 (77.19)
314 IV. ON CAUCHY INTEGRALS §77

where F(£) has been written instead of ?+(/). Let

F(oo) =a + ib/
where a is the same as in (77.17) and b is some other real constant ; multiply
1 dt
both sides of (77.19) by —— ——
2兀2 6 z
, where z is an arbitrary point of S+,

and integrate along E:

1
27rz J
/ F(t)dt
t—z
j
1
2兀2
(丽
J —z
t
念二 ]
TV% J
刚威
t — Z
L L

Noting that F(£) is the boundary value of F(z) holomorphic in the upper
half-plane, F© is the boundary value of F(z) holomorphic in the lower
half-plane, F(8) = a + 仍 and #(oo)
— —
a ib, and applying (72.1)
and (72.1'), one concludes that the first integral on the left-hand side
equals F(z)
Hence
— + ib), while the second integral equals 仍).
—
万⑵ = 十 / F 仍, (77.20)
加2J / Z
工

where, as was to be expected, the quantity b remains arbitrary.

It is easily verified that (77.20) solves the problem, if the function /(/),
for example, satisfies the H condition on L (including the point at in¬
finity, cf. § 71.)
The second fundamental problem may be solved in an analogous manner.
 PART V

APPLICATION OF CAUCHY INTEGRALS TO THE SOLUTION OF

BOUNDARY PROBLEMS OF PLANE ELASTICITY

As mentioned earlier, the solution of the fundamental boundary

problems of the theory of elasticity for regions of general form presents
great practical difficulties. However, there are certain classes of regions
for which effective solutions may be obtained by simple means. In plane
elasticity, one such class comprises regions which may be mapped on to
a circle by rational functions (one particular case has already been
encountered in § 63). At first sight this class may appear to be too restrict¬
ed ; however, as will be explained in detail in § 89, regions of this type
may be used to approximate to any desired accuracy simply connected
regions of arbitrary shape.
This Part will be devoted almost entirely to the solution of boundary
problems for regions of this kind. However, at the beginning (§79), there
will be given the solutions of the first and second fundamental problems
for arbitrary regions bounded by one contour, using a method closely
connected with the method of solution for regions of the particular type
described above. Finally, after a short introduction to other methods,
the detailed solution of the above-mentioned problems will be given for
the case of regions, bounded by an arbitrary number of contours (§ 102).
This solution is due to D. I. Sherman. It will thus be seen that Cauchy
integrals present very convenient means for the theoretical solution of
general problems as well as for the effective deduction of practical
results.
315
A
 Chapter 14

GENERAL SOLUTION OF THE FUNDAMENTAL PROBLEMS FOR

REGIONS BOUNDED BY ONE CONTOUR

In this chapter a general method of solution of the first and second

fundamental problems will be studied for regions bounded by a simple
contour (§ 79). These solutions follow from integral equations which, for
their part, are obtained directly from the functional equations deduced
in § 78. These latter equations form the foundation for the practical
methods studied in the remaining chapters of this Part and they
may be investigated directly without recourse to integral equations. For
this reason § 79 may be omitted by any reader not acquainted with the
elements of the theory of integral equations, since the understanding
of the subsequent chapters, containing solutions of problems for par¬
ticular cases, does not require knowledge of that section.

§ 78. Reduction of the fundamental problems to functional

equations.
1°. Let S be a finite or infinite region of the z plane, bounded by one
simple contour L satisfying the conditions of § 47. Let S be mapped
on to the circle | j < 1 of the 口 plane by the function
z= (78.1)
and let the circumference of that circle be denoted by 丫.
In the case of finite regions S, it will be assumed that the point 已 = 0
corresponds to the point z = 0, while in the case of infinite regions the
point C = 0 is to correspond to the point z = 8. Thus, for finite regions
s(0) =0 (78.2)
and for infinite regions (cf. § 47)
c
s(Q = + a holomorphic function.
《
(78.3)

It should also be remembered that W 0 inside and on 丫 (§ 47).

317
318 V. SOLUTION OF BOUNDARY PROBLEMS § 78

Further, it will be assumed (for the time being) that for infinite regions
stresses as well as displacements remain bounded at infinity. This is
equivalent (§ 36) to the supposition that both the stresses at infinity and
the resultant vector of the external forces applied to the boundary
vanish (conditions which must always hold true for finite regions) .
Under these conditions and with the notation of § 50, the functions
(Pi(z) and %i(z) will be holomorphic in S (including the point z = 8 in
the case of infinite regions, cf. §36). Hence the functions 中(Q and 少(？
will be holomorphic inside the circle [ | < 1. It will be assumed that
(p(Q, cp'(C), are continuous up to the circumference 丫 of the circle under
consideration, i.e., that these functions have definite boundary values
as C approaches points of 丫 along arbitrary paths ; or, in other words,
it will be assumed that the solutions are regular (§ 42) and only such
solutions will be studied.
In addition, one may always assume (§ 4 1 ) 科(0) = 0 for finite regions and
M(8) = 0 for infinite regions, i.e., in both cases it may be assumed that*
少⑼ = 0. (78.4)
In the case of the first fundamental problem for finite regions the
imaginary part of i.e., of
心0)
may also be fixed arbitrarily. "
2°. The boundary condition of the first fundamental problem takes
the form (cf. §51)
+ =7晨W + 少——3) ——
CO ( b )
— /i 十 也 — f, (78.5)
3

where a = e彷 is an arbitrary point of 丫 and 9(a), 9/(a)> +3) must be

interpreted as boundary values for a from inside 丫. This condition
may be rewritten in conjugate complex form

3'
<P‘(b) + 少⑹ =/1 — 牙2 = A (78.6)

The quantity / = /1 + ^/2 is defined on L by the equation (§ 41)

$

/ = /1 + 2% =
d 0
(Xn + ，'匕磔 + const., (78.7)

* Instead of (78.4) one may introduce the condition 中(0) = 0, as was done in
the preceding editions of this book. However, the condition given here somewhat
simplifies the reasoning.
CHAP. 14 GENERAL SOLUTION 319

where s is the arc coordinate of L and the constant may be fixed ar¬
bitrarily. This expression will be a given function of 8 (because s is a
known function of 阳 or of b.
It will not only be assumed that / is single-valued and continuous, but
also that it has a continuous derivative with respect to 8, satisfying the
H condition (§ 65, 3°). For this it will obviously be sufficient, if the func¬
tions Xn and Yn satisfy the H condition.
It will be recalled that single-valuedness and continuity of / = + if2 would
be impossible, if the resultant vector {X, V) of the external forces did not vanish,
because in that case 十 if2 would undergo an increase i(X 十 iY) for every
complete circuit of L, ie, it would not revert to its original value.

The following will now be noted. Provided has been found in one
way or another, the function 少⑷ can be calculated directly from the
boundary condition. In fact, equation (78.6) gives the boundary value
山⑹ of W© which therefore is determined by

Introducing in this formula ^(a), as determined by (78.6), and remem¬

bering that by (76.18)

one obtains
1 ( 3 g '⑹而
MD (78.8)
2九，J cr —已 2九，J
Y
G C
There still remains the problem of finding For this purpose a
functional equation will be constructed which contains only 勺 and

___
which follows directly from the boundary condition. In fact, rewriting
(78.5) as follows:
. 3（b）
岭）= / — 平⑹ 一• 勺’ (78.9)

and denoting the right-hand side for the time being by F(a), it is seen
that the function 尸 must itself represent the boundary value of some
function holomorphic inside 丫 and vanishing for = 0. However,
320 V. SOLUTION OF BOUNDARY PROBLEMS §78

it is known (cf. § 76, 3°) that the necessary and sufficient condition for
this to be true is
1 I' F(b)db
= a for all 乙 inside 丫;
22 J —E
b

the constant a on the right-hand side of (76.12) is in this case equal

to zero, since 少(0) = 0.
Introducing into the preceding equation for F the right hand side
of (78.9), one finds

1 厂 M __1_ 中'⑹而
u
2花 J —
o
Y
C “° 2储
Y
—
O' ' X
二

or, finally,

续）+
8 do
+ 口 = 4(0 (78.10)

for all 已 inside 丫，where

参) L/的
J
2Hz
— Y
b •
(78.11)

In deducing (78.10) use has been made of the fact that

1 f 中⑹而_ = 建）.
京r
The expression (78.10) is the functional equation from which the
function 中 (C) must be determined. It will be seen in the next section that
this equation completely determines the unknown function, if, in the
case of finite regions, the imaginary part of 中'(0)/3‘(0) is fixed.

Hitherto it has been assumed that in the case of infinite regions the
resultant vector (X, Y) of the external forces, applied to the contour L,
and the stresses at infinity vanish ; it has likewise been assumed that the
rotation vanishes at infinity. This assumption will now be relaxed.
In that case the functions 鱼⑶，g(z) for infinite regions have the form
(§36)
CHAP. 14 GENERAL SOLUTION 321

—o X + iY、iogz+ rz
平i(z) =
M + 材⑵，
°
(78.12)
"⑶ = o
+
/1，.
2rc(l 乂)
"g z 、 +〃
r + 州(*
where cp；(z), are functions, holomorphic in S (including the point
z = oo), and 泥 T 「and are given quantities ; further, the imaginary
part of r may be fixed arbitrarily. The quantities X and Y can be cal¬
culated beforehand, since the external stresses, acting on the boundary,
are known.
By (78.3) these formulae may be written

中" = + 中。(0，
"
2冗(！十 不
(78.13)
…、
W(Q —— — %(x
2兀 (1
a)
+ z)
已
3
rz
—Q + "(Q，
where <p0(Q, 少虱。are functions, holomorphic inside and continuous up
*to 丫. [Cf. (50. 1 4) and (50.15), where it should not be forgotten that the
region S has there been mapped on to the region outside the circle, while
here it has been mapped on to the inside of 丫.]
In the sequel, when solving the first fundamental problem, it will
always be assumed that the imaginary part of T is zero, so that 「= T,
i.e., it will be assumed that there is no rotation 威 infinity.
Substituting (78.13) in (78.5), it is seen that ^o(Q must satisfy
the same condition (78.5) as the functions cp(Q, 少⑷，with the only dif¬
ference that f has now to be replaced by f0, where

, X + iY , rc —iY 1 rc]
百石石一 笆卜
3
%=/_
wyiog。一^一常
I
%(X +祖 ।
_ r?

or, noting that a = 1 /5

, X +丫
z rc 3⑹ {[ X—iY
— rcWJ} — r
,
/0=/—
3,
-• •
—
12兀(1十田
o' 'Eg. (78. 1 4)
'
322 V. SOLUTION OF BOUNDARY PROBLEMS §78

In this expression may be written instead of log a. It is easily seen

that /0 will be a single-valued, continuous function on 丫 the derivative
of which with respect to 眇 satisfies the H condition, provided the given
functions Xn and Yn satisfy (as it has been assumed) that condition.
The single-valuedness of /0 follows from the fact that / increases by
2(X + zY) and increases by the same quantity for every
complete circuit of 丫 (in anti-clockwise direction which corresponds to
a clockwise circuit of L, leaving the infinite region S on the left).
Thus and will be found from the same condition as 平©
and W(E). Hence the more general case can always be reduced to that
considered earlier.
3°. Next the second fundamental problem will be investigated. In
this case the boundary condition has the form (§51)

乂中(b) —
(b)
二=中'
"

~
—
—=
~
'Mb)
.
2M (gi + 吻2) = 2阳， (78.15)
s'
where g» g2 are the known boundary values of the displacements u, v.
The almost complete analogy with the first fundamental problem is
easily seen. Assuming at first that (in the case of infinite regions)
x = y = o, r = r = o,
i.e., that p© and 小(。are holomorphic, and proceeding as in the case of
the first fundamental problem, one obtains the equation (analogous to
78.10)

X 建) 一六 3®) b
— C
"时 (78.16)
Y
〈 inside 丫， where
for all
8(0 — J[-岁？
—— L
2m CT
彳 (78.17)

is a known function.
Equation (78.16) is the functional equation which completely de-
termines the function <p(Q ；this will be proved in the next section.
CHAP. 14 GENERAL SOLUTION 323

After the function has been found, the function may be

determined from the formula
3 中'⑹加
(78.18)
o —— C 2ni J s'3)
Y
b —已
The case when X, y, F, T' are not zero, but have arbitrarily fixed
values (referring, of course, to the case of infinite regions) may be re¬
duced to the preceding one in the same way as this was done for the
first fundamental problem.
4°. The functional equation for 中(〈) for the case of the mixed fun¬
damental problem, when the external forces are given for one part and
the displacements for the remaining part of the boundary, may be
constructed in an analogous manner. In this case the equation becomes
somewhat more complicated and no consideration will be given to it
here (cf. also end of § 79, 4°).

§ 79. Reduction to Fredholm equations. Existence theorems.

The proofs of the existence theorems, given in this section, have been taken,
without essential changes from the Author's paper [11]. However, several simpli¬
fications and corrections of two elementary, but annoying blunders in the Author's
reasoning in that paper have been introduced here. One of these had been
kindly brought to the Author's notice by S. G. Mikhlin and the correction was
already included in the first edition of this book. The other, despite its obvious
and elementary nature (or rather because of it), remained unnoticed and was only
discovered by the Author himself, while preparing the third edition. A short
account of these proofs was also given in the Author's papers [9, 10].

1°. The functional equations (78.10) and (78.16) represent a somewhat

unusual type of integral equations which are, however, easily reduced
to ordinary Fredholm equations of the second kind.
In the case of infinite regions, it will again be assumed that
x = y =「= r = o,
since the problems can always be reduced to this case (cf. § 78).
2°. A beginning will be made with the first fundamental problem. In
order to reduce (78.10) to a Fredholm equation, it will be rewritten

中©
『 / + 2兀叮 = —— o
<y(b)(b —
«9)诙 + 万3© + @ = 4(9，(79.1)
324 V. SOLUTION OF BOUNDARY PROBLEMS §79

where
幽. (79.2)

it is easily seen, on the basis of (76.18), that the left-hand side of (79.1)
is identical to the left-hand side of (78.10). In the case of infinite regions
3‘(0) = co and hence = 0.
Differentiating (79.1) with respect to 匕 one obtains

X。
Y
｛汽 瑞 + M9 = 40 09

whence, letting 已 tend to an arbitrary point n0 of 丫，one finds

+六
「 I— -=L 而
+ 为3‘(％) = 4'(咏. (79.4)
— Go
~~~
J Co0 I CT J3,
Y

It is readily seen that this transition to the limit is completely justified

under the conditions postulated earlier for the function / and the contour
L. In fact, it has been assumed that / has a first derivative satisfying the
H condition. Therefore 4'(0 will be a function, continuous inside and
up to 丫；the function 力十％) in (79.4) denotes the boundary value of

Further, the conditions assumed with regard to L ensure the con¬

tinuity of 3©, s"(C) up to 丫, with the exclusion of the point = 0
in the case of infinite regions, and also that # 0; hence it follows
that the function
i a 3 一 3© 3 一 3（0 —（…已）3亿）
6= (79.5)
比 —— 已 西(b 一 92
is continuous for all values of and 已 inside and on 丫，except for b = 0,
3 = 0 in the case of infinite regions.
In fact, by Taylor's formula with the remainder term in the form of a definite
integral, one has

3 一 3日 — 3亿) (b — /
Q = 3〃 (b
— 次

where the integral may be taken along the segment of the straight line connecting
CHAP. 14、 GENERAL SOLUTION 325
b and 口 Putting t = g
— (b — 已)， one obtains
入 •
1

3(<T) 3(0 3，(〈) (b

—- C) = (b Q2
0
3〃[(J 入(<J Q]入d入，

and so the continuity of K^,c) is proved.

Finally, by supposition (cf. §78), the functions <p(Q, <?'(?, W(C) are
continuous up to the boundary.
The equation (79.4) may also be deduced from the equations obtained
in a different way by V. A. Fok [1,2] who, however, restricted con¬
sideration to finite regions, (cf. V. A. Fok and N. I. Muskhelishvili [1]).
The preceding formulae refer to the cases of finite as well as of infinite
regions. However, in the latter case, they may be given a somewhat
different form which will be more convenient. First of all, in that case
= 0. Hence (79.1) becomes
中(Q+ 下T
1 / 3(o) - co (Q
b+ 0 = ——7rb (79.1')
兀”
(o
Y
((y —
Q)

Further, noting that in accordance with the imposed conditions

= + 3o(0 ,
3© 《
where 侬。© is holomorphic inside 丫，one has
3(b) ~~ 3©
— *
~~
—
3o(b) So(Q
•

—— o 已 ——'

- ，二 —
• …
" '
— '
C

O*乙

，—
b C
Substituting this expression in (79.1‘)，one finds
1 3。 ~ 3o©
o+ 2m J g/(g) (o Q
中'3）而 衣 = 4© , (79.1〃)

because, as is easily verified,

2汽

in fact, the expression conjugate complex to the last integral

2兀
326 V. SOLUTION OF BOUNDARY PROBLEMS §79

is zero, because <p'(b)/bs'3) is obviously the boundary value of a function,

—
holomorphic inside 丫.
Differentiating (79.1")

MQ 十三/全｛器七答｝两说 49 亿91〃)

，
Y

and taking, as before, the limit《 f a0, one obtains the integral equation

卫 。⑹ —
bg) +—L
2mJ/ 卜
法 I — 叫J w b %
= ©(啦 (794)
Y

Thus equation (79.4) which will now be written

9‘(bo) + / K(bQ, b) + 々3‘(b0) = 4 '(%) (79.6)

Y

and which is applicable to both cases may in the case of infinite regions
be replaced by (794), i.e.,

63o) + 2兀z J
r / KoM o) <p'(o) do = 4'3o), (79.6')
Y

where

础P= 上得
3’ 比 处上产
一已 6
• (795)

The integral equations (79.6) and (796) will now be investigated

and a beginning will be made with the case of infinite regions.
Substituting in (79、6‘)
中 '⑹ =
and
Ko(o(), 6 — K］十 2K2
and separating real and imaginary parts, one obtains two real Fredholm
equations ; this system may be reduced by ordinary means to a single
equation. This equation will not be written down, since it is sufficient to
know that (796) reduces to a single Fredholm equation (of the second
kind).
Suppose now that (796) has the (continuous) solution 中’ By
CHAP. 14 GENERAL SOLUTION 327

substituting this solution in the second term on the left-hand side of

(79.1〃') one finds some function which is seen to be holomorphic
inside 丫 and to have a definite boundary value on 丫. Further, it is also
seen that this boundary value coincides with the function 平 '(c), figuring
in the second term on the left-hand side of (79.1'"), or, in other words,
that the holomorphic function 中'(Q, thus determined, actually solves the
functional equation (79.1'〃). Finally, it is readily seen that one obtains
the solution cp(Q of the functional equation (79.1〃)，if one determines
this function by the same equality (79.1〃)，with the understanding that
one has to use as function <p'(a) appearing in the second term on the left
hand side the solution of the integral equation (79.6') under consideration.
Once has been found the function 少© is determined by (79.8),
viz.,

(p' 而
(79.7)
(二 a
Since 少鱼) and 中化) are continuous up to 丫， the functions 中(Q
and “Q give a regular solution of the problem. Thus a definite regular
solution of the problem corresponds to every (continuous) solution
平 '⑹ of the integral equation (796).

For the determination of the constant a, referred to above, it is sufficient to

put C = 0 in (79. i"), or, better still, in (78.10) which is equivalent to (79.1〃)；in
this way one obtains
a ~ A(0) —~ —丁二=" 中 .
2H2 J /⑹
0(1
y
The properties of stated above, have been discussed earlier and those of
中(0 are then obvious ; those of 少(？ follow from the same reasoning and from the
fact that (79.7) may be written

If/ 『击 ［ 函一市
1 函
2m J/ — — - ? (0・ (79.7 )
7，、
"
V, /70
MO = o . z
—C —
9
2m o 3 (c) (a Q 3 (Q
Y Y

It will now be shown that the integral equation (796) has always a
unique solution. It is known that for this purpose it will be sufficient to
prove that the corresponding homogeneous equation

63o) + JrJ/
2兀2
y
= 。 (79.6〃)
 328 V. SOLUTION OF BOUNDARY PROBLEMS § 79

has no non-zero solution. This is almost obvious on the basis of the earlier
remarks. In fact, if this equation had a solution different from zero,
one could by means of this solution obtain a solution of the fundamental
problem for the case / = 0 with 9'(9 力 However, this would mean
that a non-zero solution exists for the case when no external forces
act on the boundary and that the corresponding internal stresses in the
body differ from zero. The impossibility of this is proved by the uni-
queness theorems (cf. § 40, 3°, § 42, 2° ; also § 41, 3°). Thus the existence of
the solution of the first fundamental problem has been proved for
infinite regions.
The case 0/ finite regions will be considered next. For the time being
it will be assumed that the constant k in (79.1) has been fixed arbitrarily.
In order to remove the term 力 in this equation, introduce the trans¬
formation
端0, (79.8)
where is a new unknown holomorphic function. One thus obtains
from (79.1) the equation

平 o(Q + /告T7」
(c 空
靛5加 + d = 月(0
—
(79.9)
2兀z J 3 0
Y

from which follows, as before,

中 M)
2m J
[ K(& 6端祕 = 4(9 (79.10)
y

and, for 已 f

p/由 +-y-v
2H2 J
[ K(q, o) m⑹而 = 月'(00). (79.11)
Y

As in the case of (79.6'), this equation may be reduced to a system

of two Fredholm integral equations from which one finally obtains a
Fredholm equation (of the second kind). It will be shown below that
(79.1 1) always has a (unique) solution.
First, however, consider the construction of the solution of the original
problem, once any (continuous) solution 中 j(o) of (79.1 1) has been found.
Substituting this solution in the second term on the left-hand side of
(79.9), one obtains the function @£0 The function 中 (C) will then be
given by (79.8) and it will be the solution of the functional equation
CHAP. 14 GENERAL SOLUTION 329

(79.1) for the given value of k. In order that the function de¬
termined in this manner, will lead to the solution of the original problem, it
is, however, necessary and sufficient to select the value of k to satisfy
(79.2), ie,

疗⑼
or, using (79.8), ,
为 月=%2_. (79.12)
k(0)
This is obviously possible only if

―(0) = a real number.

3
(79.13)

Let it be assumed that (79.13) is fulfilled. Then (79.12) determines the

real part of k. By fixing the imaginary part of k arbitrarily a definite
expression will be obtained for the unknown function 平依)；after de¬
termining the corresponding function 少 © from (79.8) one finally finds
a certain regular solution of the original problem.
The meaning of the condition (79.13) is easily explained. In fact,
introducing the function %(。) by letting

?oW + *3) + 弘⑹ = /1 + ^2 + '。⑼ 侬⑹ (79.14)

8
3，(0)
or
‘ ‘ f

3(O) <Pn(0)
记 s(o)・
z
中o(o) H co 7ytp
(。)
q + Wo(o) = 九 一 牙2 4 co「
(79.15)

It is readily verified directly that by (79.9) the function %(b) represents

the boundary value of some function “(0, holomorphic in 丫 (and
vanishing for E = 0, the last fact has no significance here). In fact,
interpreting %(o) as the expression given by (79.14) and taking into
consideration (79.9), it is easily seen that
 V. SOLUTION OF BOUNDARY PROBLEMS §79

1 f 少o3)而
= (J
J
I
2m

which proves the statement on the basis of § 76, 3°, Theorem I.

甲o(b)ao = / 更的付油值)] = 0,

/ 3 = / d[<po co(b)] = 0,
Y

= / 少o(o)G/(<y)do = 0,
Y

=
L
J Zdz,
Y
(0 = /
L
zdz,

zdz ZaZ,
L

one obtains

0—2 T(/1^ + 椒夕) 十 zdz.

L
But
zdz = / (xdx 4- ydy) + i / — = — 2ES,
L L L

where S is the area of the region inside L. Hence

；(0) 媪0)
(/M% + ，2如) = vS (79.16)
3'(0)
工

The expression in the curly brackets differs from the imaginary part
of 90(0)/(070) only by a factor, and hence (79.13) is equivalent to

L
I (M% + My) = 0 (79.17)
CHAP. 14 GENERAL SOLUTION 331

which is the condition for the vanishing of the resttliant moment of the
external stresses applied to L.
Now equation (79.11) will be considered and it will be shown that it
has a unique solution. For this purpose the corresponding homogeneous
equation

ME) +
2兀2 J
[ K(^, cr)@；(b)
击 =0 (79.1 1')
Y

will be studied. This equation is obtained, if one wants to solve the first
fundamental problem in the above-stated manner in the absence of
external stresses, i.e., for % = % = on L. As under this condition
°
(79.17) will obviously be satisfied, the condition (79.13) will be fulfilled
for any solution 中;(o) of (79.1 1'). Selecting the real part of k in accordance
with (79.12) and fixing its imaginary part arbitrarily, the solution of the
first fundamental problem for /幺=0 may be constructed starting
from <pQ(a). If this function does not vanish everywhere on 丫，the solution
constructed in this manner will not correspond to the case of absence of
—
stresses. In fact, the function 中(乙) will be given by 中(0 = 后co© 十 平。©
and, in the absence of stresses, one should have + const.,
where C is a real constant. Hence, in this case, 干。(已) = 加3©, where m
is some constant. Substituting this expression in (79.9) with X(^)
one obviously finds 例3(〈) = const, which is only possible for 加 = 0,
— 0,

i.e., @0© = const., and hence 中6(Q = 0. Thus the presence of a

non-zero solution of (79.11') implies a solution of the first fundamental
problem, giving the state of stress in the absence of external forces,
which is impossible by the uniqueness theorem. In this way it has been
shown that the homogeneous equation, corresponding to (79.1 1), has no
solution which is not identically zero, and therefore (79.1 1) has one and
only one solution.
Solving (79.11) and assuming (79.17) to be satisfied, the function
<p(Q will be found from (79.8) with the real part of k chosen in agreement
with (79.12); the function 少© can then be determined by (78.8) and the
solution of the original problem obtained. The imaginary part of k
remains arbitrary, as was to be expected, since a term of the form
in the expression for (p(Q, where C is a real constant, does not affect
the stress distribution.
It will be remembered that (79.17) is the condition that the resultant
moment of the external forces must vanish. The condition for the vanish-
332 V. SOLUTION OF BOUNDARY PROBLEMS §79

ing of the resultant force vector is ensured by the continuity of the func¬
tions /1 and /2 on L ; that is the reason why it does not appear in explicit
form.
Thus the existence of the solution of the first fundamental problem
has also been proved for the case of finite regions. At the same time
(theoretical) methods of solution have been given for this problem for
the cases of finite as well as of infinite regions.
3°. Next consider the second fundamental problem. This problem has
been seen to reduce to the solution of the equation (78.16) which is
quite analogous to the equation obtained for the first fundamental
problem. The 'methods of solution of the first and second problems are
so alike that there is no point in repeating the reasoning.
A certain difference occurs only in the case of the problem for finite
regions; in fact, one will have now instead of (79.8)
k
建) =— 3(。 (79.18)

and k will have to be determined from the equation

k ——x = 丝上
^7(o)
(79.19)

which gives a definite value for k (remembering that x > 1) without

any additional condition for the existence of the solution.
Thus the existence of the solution of the second fundamental problem
has been proved and at the same time a (theoretical) method has been
obtained for its solution.
4°. The mixed fundamental problem may be solved by methods analogous
to those above. In this case the stated procedure does not lead directly
to a Fredholm equation, but to a so-called singular integral equation
which is then easily reduced to a Fredholm equation. The mixed problem
has been solved in this way by D. I. Sherman [10]. The solution may be
considerably simplified, if use is made of the general theory of singular
equations which has been developed recently.
5°. A more detailed study of the integral equations for the first and
second fundamental problems, obtained above, was likewise presented by
D. I. Sherman [7]. In fact, Sherman introduced into these equations a para¬
meter 入 (not to be confused with the Lame constant), similar to that
occurring in the general theory of Fredholm equations, and proved that
CHAP. 14 GENERAL SOLUTION 333

all the characteristic values of this parameter are real and distributed
inside the region — 1 X I. This fact is of practical value, since it
shows that the above integral equations may be solved by iteration
methods, i.e., that the Neumann series will converge for those values
of 入 to which these equations correspond ; in fact, the integral equations
for the first and second fundamental problems correspond to the values
入 = 1 and 入 =
— 1/x respectively (remembering that x > 1).
Apart from these results, Sherman deduced in the above paper a
number of other results which are of independent interest.
In the later chapters of this Part remarks will be made with regard to
the existence theorems for regions of more general shape and also with
regard to some other general methods of solution of the fundamental
problems.

§ 79a. On some other applications of the preceding integral

equations. The integral equations of § 79 may also be applied to certain
other important problems of the theory of elasticity, e.g. the (appro¬
ximate) theory of bending of plates, loaded by forces normal to their
plane. It has already been stated above that the case of plates clamped
along their edges may be reduced to the so-called fundamental bi-harmonic
problem, i.e., to the same boundary problem as the first fundamental
problem of plane elasticity.
The case of plates with free edges has been found to reduce to the same
boundary problem as the second fundamental problem of plane elasticity ;
the only difference is that the constant 尤 has to be replaced by some
other constant, likewise larger than unity. This has been proved by
S. G. Lekhnitzky [3[ and, later on and independently, by I. N. Vekua [3].
 Chapter 15

SOLUTION OF THE FUNDAMENTAL PROBLEMS FOR REGIONS

MAPPED ON TO A CIRCLE BY RATIONAL FUNCTIONS.
EXTENSION TO APPROXIMATE SOLUTION FOR REGIONS OF
GENERAL SHAPE

As stated earlier, Cauchy type integrals provide the means for ob¬
taining theoretical as well as practical solutions of the fundamental
problems for certain fairly wide classes of regions. The starting points for
this work are the formulae (78.10) or (78.16) or analogous formulae
to be stated below. The case for which, the mapping function 3(Q is
rational is particularly simple, since, as will be shown in this chapter,
the solution in this case is obtained by quite elementary means. However,
for the sake of clarity, a beginning will be made with the direct solution
of the problems for some very simple regions.
The major part of the results stated in this chapter were contained in
the Author's papers [4, 5, 7, 8.]
In their two papers [1, 2] D. M. Volkov and A. A. Nazarov gave a method which
apparently permits solution by elementary means in the case of a wider class of
regions. However, this class has not been specified by the authors with sufficient
exactness, so that it cannot be stated beforehand in what cases, in addition to
those stated by the Author here, a solution may be obtained by elementary means.
In fact, in order to state the cases, where one can certainly obtain elementary
solutions by applying completely definite methods, the Author, in those of his
papers which were devoted to elementary methods, has limited consideration to
cases where is a rational function. In addition, he should indicate that he does
not agree with Volkov and Nazarov in their claim that their method leads to
simpler calculations; cf. § 87a.

§ 80. Solution of the first fundamental problem for the circle.

It has already been stated in § 54 that many solutions of this problem
are known. Among those mention will be made only of the solutions by
G. V. Kolosov [1, 2], G. V. Kolosov and I. N. Muskhelishvili [1] and
G. V. Kolosov [5] which was also published in 1931. In § 54 this problem
was solved by the use of series. Cauchy type integrals achieve the object
more rapidly and give a solution which is more convenient in ap¬
plications.
334
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 335

-
Let R be the radius of the circle S with circumference L. In the present
case
= 3g 岚, (80.1)
where here and below the notation of § 78 will be used. In particular, 丫
will denote the unit circle [ C | = 1 , = e毋 a point on this circle.
The boundary condition now becomes
平⑹ 十 + 巾⑹ = /i 十 if* = / (80.2)
or

中3) + 的'⑻ + Mb) = /1 — ^/2 = £ (80.3)

Expressing the fact that the right-hand side of
山⑹ = f — 中⑹ 一 函)， (803)
must be the boundary value of some function W(Q, holomorphic inside 丫,
and vanishing for C = 0, one obtains by (76.12)

2兀匕 J cr
Y —— 已 J
Y
a — (
= 0,

which leads to the functional equation z = 3(0). Thus

, 1 / 四'⑹说
a = —— 14
1 如
哪) + 2兀2 J G
—C
丁 + ——J G
2位
-
(80.4)
Y Y

This equation is nothing else but the functional equation (78.10) for
the case s(Q = RE; the deduction given here repeats the derivation
of § 78 for the particular case under consideration.
In the case considered here the functional equation may be solved in
a simple manner without transition to an integral equation, since the
integral on the left-hand side can be immediately calculated in finite form.
In fact, consider the first three terms of the expansion for <p(Q
中(9 = 曲 + 口< +々Q +… (80.5)
from which follows
中 (C) 口1 + 2tl2c ~
… ,
and hence, by (76.17), 1 / o(p' do
2兀e J cr
— C — ] 2a 2 ,
Y
336 V. SOLUTION OF BOUNDARY PROBLEMS §80

(80.4) may thus be written

i
虱9 + 十 2质十五 = (80.6)
2九5
Y

This last relation determines g© apart from 左工 + 2^2, ie, the

unknown constants alf a2 have still to be found. For this purpose one
has to express the condition that %,a2 are coefficients of the expansion
(80.5), for C and 芋 respectively [If this condition is not satisfied, the
function 平(已) defined by (80.6) obviously will not fulfill (80.4)]. These
conditions will now be formulated. For this purpose one has only to
set 已 = 0 in the equations obtained from (80.6) by one and two differ¬
entiations with respect to 已 and take into account that, by definition,
%= 242 = This gives respectively
1 八 da
旬 十的=
元 JG'
Y
(80.7)

(80.8)

The last formula determines the constant a2.

The relations (80.7), (80.8) may also be obtained by substituting in (80.6) the ex¬
pansions

中(9 = G式 做？+ ...

and

and by comparing coefficients of ^2.

CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 337

Substituting / — /x + 歹 2, b = e* the formula (80.7) gives

2k

设
1十 五1 (A +枷 (80.9)
0
This condition can only be fulfilled when the right-hand side is real,
i.e., when
2k

/ — /i
o
( sin 9 十 府 cos 眇 = 0,

which expresses the necessity for the vanishing of the resultant moment
of the external forces [cf. (54.3)]. If it is satisfied, the real part of 的 is
completely determined by (80.7), while its imaginary part, as expected,
remains arbitrary ; putting, for definiteness, §(口) = 0, one obtains from
(80.7)

(80.10)
• Finally, one finds

(80.11)

where ax and 能 are given by (80.8) and (80.10).

Having found 中(Q, the function MQ may be immediately determined,
because its boundary value 少⑹ is given by (803). Determining 少©
from Cauchy's formula and taking into consideration that [cf. (76.18)
and (70.3)]
而
(p
= <p(0) — 0,

tjq/

one obtains
怀)
等+ 通. (80.12)
338 V. SOLUTION OF BOUNDARY PROBLEMS § 80a

It is easily seen that the solution obtained will be regular (in the sense
of § 42), if the function f given on L has a derivative satisfying the H
condition.
Thus the problem has been solved. It will be noted that the last terms
on the right-hand sides of (80.1 1) and (80.12) may be omitted, because
constant terms in the expressions for 勺(卫)， 少(？ do not influence the
stress distribution. These constants have only been calculated with the
"fundamental biharmonic problem" in mind, where the constant terms
are significant.
Omitting the above-mentioned constant terms, one has instead of
(80.1 1) and (80.12) the much simpler formulae
.、 1 do
建) = 不一 一 中， (80.U')
Y

少 (Q = —
i (80.12)
Y
3U dU
In this case the boundary value of +i may differ from
by a constant term. If one takes (80.1 1) and (80.12), the above boundary
value will be exactly equal to f.
The solution deduced above is very convenient for applications, as
will become apparent from the examples considered in the next section.

§ 80a. Examples.
1°. Circular disc under concentrated forces,
applied t o its boundary.
This problem was first solved by H. Hertz in 1883 and studied in detail by
J. H. Michell [2], using methods quite different from those used in this book.
Cf. also A. E. H. Love [1] § 155.
Let the concentrated forces
匕)， 阳，
区， 匕)，…， (^n，
act at the points
% ~ Re'% % = ReS, . . . , / =
(0 ax < a2 < ... < an < 2兀)
of the edge of the circular disc. The points
Qi = 越豌，..., b% = 法％
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 339

correspond to these points in the 已 plane.

Under these conditions the expression / = 占 + if? will be constant
on each of the arcs a2<T3, . . . , 0久61(because these arcs are free from
external forces) , but it will change discontinuously by i(Xk + EYJ for
a passage through the point 6fc (cf. § 43). For the derivation of the
formulae of the preceding section to be used here it has been assumed
that the function fr 勿％, given on the contour, is continuous (and
even has a derivative satisfying the H condition) . However, it is readily
verified by direct checking of the final results that these formulae lead
also to the solution in the case under consideration here.
For example, let / = 0 on Then one will have on the arcs
b2b3, etc. that /
—
i(X] + iYj, f = “Xi + 勿匕) i(X2 十 勿匕)， etc.
respectively. In order that f attain its original value 0 on %% after one
complete circuit, one must, obviously, have the conditions X】+ X? + …
+ X% = 0, + y2 + ... + — 0, i.e., the resultant vector of the
applied forces must vanish. This condition is necessary, because
3U.3U must be single-valued • -j • ,, x
+z 1 j
mside 丫， since the region under
加 效
consideration is simply-connected.
Further,

Xi + %二 4 —已 （占"k）+ 陷"k）,log 小一已

—三 * 不 7+ 2冗 啊 Q
——
(X1 + ，当) 十 …十 十 eh) —已
+…H log ％
2h bn —v
q
The last term is, of course, zero, but it has been written down for the
sake of symmetry. After some obvious manipulations one finds

L7
1 f da 1
q = {(为 + 271) log(%-0 + 区 + M) log 应一0十
2兀z jo
丫
— -不-
/k
... + 氏十 2K)题际一？}.
Similarly,

Y
—p
cr — C
~
2兀 —，匕) log (j—〈)+ •…
—
+(Xg - zY*) log(%
— 0}.
340 V. SOLUTION OF BOUNDARY PROBLEMS §80a

Finally, the constant 诙 has to be determined. For this purpose use

may be made of (80.10) ; however, it will be simpler to adopt the following
procedure. Since, by supposition, = (p'6), one finds from (80.1 1),
replacing the integrals on the right-hand side by the expressions above,
differentiating with respect to 已 and putting 已 = 0,

2 阳+ ，
1 Xk + iYk
-
、， 、_
2的 = E 一 =丁
/a fc= 1 N
兀 k—1

In order that ar may be real, the right-hand side of this expression must
be a real quantity. This condition is easily seen to lead to
n
z (力？万一 九Xit) = o,
fc— 1

where xk + iyk = z% = i.e., to the vanishing of the resultant moment.

If the above conditions (for the resultant force and moment) are
satisfied, the solution of the problem is given by (80.1 T) and (80.12')
1 n r n
建) =_ S 陷十 HMogE-D- 六 s 陷+EYM ， (80.⑷ 乃
2冗 加 =1 兀 4 尼

中(9 — e (X% - 订屋) log (ga — Q -

~ ~ ~ (80.2口)

It is easily established that the stress function U will be continuous

up to 丫， so that one will
actually have the concen¬
trated forces at the specified
points.
Consider, for example, the
case when two equal and
opposite forces (》，0) and
(— 0, 0), parallel to the Ox
axis, act on the disc at the
points 画 — R&a and z? =
=Re：ga)= —R-ia gig. 35).
Then, reverting to the old
variable z=R& one obtains
from (80.1a) and (80.2a), omitting inessential constant terms,
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 341

= —— W{log —— z) — log —— z) +
' '
的(z) (外 曲
Z

“(z) = 2{log(z】T—log(”z)
I(z) = 中;(z)

巴⑶ = U(z) = 一 2｛在一提7+ (4 —— Z)2 —— z)

The stresses will be given by the formulae
X 化 =4猊％(z),

匕 一 X, + = 2[2；(z) + 乎式叫.
Substituting for i and and noting that

/ — R&a, Z2 = — Re~ia>
4 Z =“一啊 — Z = - Q语 — -
(see Fig. 35, 瓦 and % being positive or negative, when the point 2 lies
above or below the line of action of the forces), one finds
COS % cos % cos a[
+ V, = + R J
p fcos 3% + cos % cos 3% + cos %
& —— 匕= —6 !

p ( sin 38] + sin 分】 sin 3% + sin %

2X” 兀 I % %
hence
Xh = + cos a,

cos
Yv =
sin2 3] cos 9]
1
sin2
4 7tK
cos a,

sin % cos2 % sin % cos? %

x# = —
342 V. SOLUTION OF BOUNDARY PROBLEMS §80a

The displacements u and w are also easily obtained, using the formula

2以四 + 谢) =
which gives 2M(〃 谢) =
- z*(z) — 心O
£卜 — z — 1) cos a z?.
Z2 z z Z 的一 z (k
+ log 上 + Zy
« 宏
log
— —z
二 h
z R
The values of the multi-valued functions, occuring in this formula, have
to be restricted to one definite branch. If another branch is used, the
resulting displacements will differ from the first by a rigid body dis¬
placement. Separating real and imaginary parts and replacing k by its
value (入 + 3“)/(入 “)，one finds
22) ， r2
—
力 ( 2( 入 cos a
u= { log H cos 2% cos
4Wt I 入十以 夕
1 入 十 fZ

v
- 力
知兀
f 2J
l—
从，

入十 m

+
击 + 2'
x

— sin 2%1 — fi
.
sm 2%2

In these formulae 力 一次 = z. In the last formula one may write y ~ I

instead of y, where I is the distance of the centre from the line of action
—— +
2ii cos a y 1
入
•

=
kJ} •

of the forces (obviously this amounts to the addition of a rigid body

displacement). In that case all points, lying on the line of action of the
forces, will remain there after deformation.
If one is not dealing with plane deformations, but with a thin disc,
one must use 入* instead of 入 and /> must be conceived as the quantity
F/2%, where F is the concentrated force and 2% is the thickness of the plate.
(Actually, in the above work, 力 denotes a force which does not act at
a point, but on a straight line, perpendicular to the plane, and which
is estimated per unit length of this line.)
A large number of examples of a similar kind may be treated which are
of interest for technical applications. In particular, it is very simple to
deduce solutions for all those cases which were considered by J. H.
Michell [2] using other, artificial methods (cf. G. V. Kolosov and I. N.
Muskhelishvili [1]).
2°. Disc under concentrated forces and couples
acting a t internal points. The solution of this problem is
likewise obtained with extraordinary simplicity from the general formulae
of § 80. For this purpose it is sufficient to introduce into the functions
中，少 definite singularities at the points of application of the concentrated
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 343

forces and couples, as stated in § 57. It will be left to the reader to find
the general solution. For the sake of brevity, consideration will be restrict¬
ed here to the example of two exactly opposite forces one of which acts
at the centre, while the other acts at an arbitrary point of the disc. Without
affecting generality, it may be assumed that the second force is applied at
a point of the Ox axis (and directed along it). Thus one has the two
—
concentrated forces ( % 0) acting at 0 and (+ 夕，0) at z。， where
z0 is real.
In the case under consideration the functions 鱼⑶，“(z) will have the
following forms (cf. § 57) :
夕
师⑶ = 八八
力
2tc(1 +,F
x)
z — 2兀(1 工一
人 十 然)
log (z — zo) + 望⑶，
"⑶ = —々
2兀(1 :l°g z + 2 (1 + x) l°g (z ~ % +
十 x) 兀
(80.3〃)

+ 57 U
27V (1 + X) Z Zq
F 记(0
or, in terms of〈 = z/R,
中(9 ~ o 1 十 X)
/矶
\
{log C ―‘Og (C — 3} + 少o(C)，

少(。= —o /1 \ {1°巨 已 ~ 'og (C

'

— Co)} + (80.4©

] 厂 rU i /y\

十说丁率不
【

where 勺o(9， are holomorphic inside 丫 and = z^R.

The boundary condition (assuming the edge of the disc to be free)
may be written
中⑹ 少⑹ 0
or, substituting from (80.4〃),
go 四; %⑻ (80.5a)
where
/o —
2k(1 + x)
log±=^ 2兀（1
比力
+ x）
log (1 — 焉。)
夕 b

I1
— Co b 一 炉）, (80.6a)
2兀（1 十 K） 备 0
344 V. SOLUTION OF BOUNDARY PROBLEMS §80a

and hence

豆 =前
Me 10g (1 ~ — 2兀(；；比) ng 个"
__i\
力 厂 1_k
(80.6%)
+ x) I cr — a2 J
*
2k( 1 <y 焉
The functions (Po(Q, 中亚) will be found from (809) and (80.10), where
one has to replace 平， 少，/ by <p0, %,/0. Calculation of the integrals oc¬
curring in these formulae presents no difficulty.
The choice of the branches of the multi-valued functions log (1 — 吃) and
log (1
— 式
o) is arbitrary. However, they must be chosen in such a way they
represent on conjugate quantities. For the first function, a branch will be chosen
丫
that

which is holomorphic outside 丫 and zero for〈 = 8； for the second function, a
branch will be taken which is holomorphic inside 丫 and zero for = 0.

Noting the properties of the branches of the functions log (1 ~

—
and log (1 CO, as chosen above, one has from the formulae of § 70
and by Cauchy's formula
1 / b
—0 — 而
=0,
1 C
log (1 —幅)
da
— - = log (1 一 泌).
—
一
2冗2J b b 『
q 2九2J a
Y Y

Also, by the same formulae,

i，八一q o j 而 已一得
2k2 J/ 11
I1 — 吟Jo 一 已
— 衣厂0 1
—六
源
；2,
气

Hence

2 元(：1 田 log 1 一调 十
丁告方{工^T}'
] 万而
2大VJ G
- C 2兀(1 十 x)
log (1 - /).
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 345

For the calculation of 旬 it will be noted that 2旬 equals the value of

the derivative of the last but one expression for
3-1)%
—
0 (cf. § 80). Thus,

优工
4兀 (1 + x)
and, by (80.11') and (80.12'), one finds

伙⑷ =
- ?冗(1+ T
2 x)
l°g ( 1 — 柒) + |
+ x)；
2兀 (1 I 1 — CoC
一
WJ —
(X 一 1)外《
4兀 (1 + x)

次比 一 1)迄 1
2兀(1 十田 一油)
11

既
~2k(1 +x) *
(1~W
A constant term has been omitted in the last expression. Finally, by
(80.4«),

平(Q = 屋二 Mg - 柒)
+
2tv(1 +「
x)
log-一
；
—
；o 2兀(1+田 °
，
+ I 1 一 焉已
2k(1 x) 4兀(1 十 x)

WQ =
- 2 (1 + X)、log J?
加 已一
- 十 万,々 、•Q 一 五 +
2兀(1+ X)
(80.7。)

」 (I T) _」•
(乂 一 1溜十 L
十 —
2碎 + x)
g
log
—
2 (1 十 6 1 —泳兀
—
0 1—谈
一
2兀(1 下可. 一已炉'
and the problem is solved. When dealing with a thin plate one has to
replace 比 by x*.
The problem for systems of arbitrarily distributed forces can be solved
just as simply.
3°. Rotating disc with attached discrete masses.
Let the thin elastic plate rotate about its centre with angular
velocity Q and let there be arbitrary discrete masses attached to points
346 V. SOLUTION OF BOUNDARY PROBLEMS § 81

of the plate. It is sufficient to find the solution for the case of one mass m,
because the solution of the general case may be obtained by super¬
position of several such solutions.
The effect of a concentrated mass obviously reduces to the action of
a concentrated centrifugal force in a radial direction and of magnitude
F = 加ZQ2, where I is the distance of the mass from the axis of rotation ;
a reaction, equal in magnitude and opposite in direction to the force F,
will act at the axis of rotation. Thus the solution of this problem will
be obtained by adding to the solution of the problem of a rotating disc
without discrete masses (cf . end of § 59口) the solution of the problem,
considered in the preceding example. In the present case 力 = F/2A
wZQ2/2瓦 where 2% is the thickness of the plate (because 力 is calculated per
—
unit thickness).
The solution of the problem of a disc rotating about an eccentric
axis may be obtained in a similar manner.

§ 81. Solution of the second fundamental problem for the circle.

In the notation of the preceding sections the boundary condition in
this case takes the form
x中⑹ 一 w '⑹ 一 w⑹ = 2M (gi 十 语2 2
=阳 (81.1)
or
m中 一 两 '⑹ — 少⑹ 2Mgi — 欣2) = 2函 .2)
where gi，g2 are the given displacements of points of the boundary.
In view of the complete analogy with the problem of § 80, only the
final solution will be recorded here:

兀2
.
Y
J o "裆十 2&十历 (81.3)

wo = TC, J b -
9 (0 + + x^o * (81.4)
Y
where .. 包『当，
些十
广
乂口0 = = ― r
J
1
o mx J a4
(81.5)
Y

(义2 ])曲 i 十 尼 gdo. (81.6)

兀4 J a2 冗2 J
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 347

诙 =— 匕
mx J
ga do. (8L7)
Y

This solution will be regular, provided the function g, given on the

boundary, has a derivative satisfying the H condition.

§ 82. Solution of the first fundamental problem for the infinite

plane with an elliptic hole *). Use will be made here of the transfor¬
mation of the region S under consideration on to the region [ | > 1,
i.e., on the infinite plane with a circular hole. (In the Author's earlier ^
work the transformation on to the circle has been used instead.)
The relevant transformation is (cf. § 48, 5°)

z= = + 菅) 你 >0, 0 w < 1). (82.1)

The circle | = 1 corresponds to the ellipse L with centre at the origin

and semi-axes
a = 7?(1 m), b = 冗(1 + m).
—
By suitable choice of R and m one may obtain ellipses of any dimension
and shape. If w = 0, the ellipse becomes a circle. In the limiting case
加 = 1, it is the segment of the Ox axis between the points % =
± 2R
and the region S is the infinite plane with a straight cut.
In the present case
3 1 Q2 +m 3(b) 1 + ma2
研彳 b 1
—m 户 3‘(b)
O
n2 —m
and the boundary condition takes the form
1 . . .

+ 中⑹ = /
i

中 (b) H 少'3) (82.2)

or
b 1 — md

+ W(b) = —/.
1 4- WG2
平⑹ + — a
a2 ~
tp'(b) (82.3)

First assume that

x=y = o, r = r‘ = o
[the notation being the same as in § 50; cf. (50.14) and (50.15)], i.e.,
*) N.I. Muskhelishvili [4].
348 V. SOLUTION OF BOUNDARY PROBLEMS §82

that the resultant vector of the external forces applied to the contour
is zero and that the stresses as well as the rotation vanish at infinity.
Then 中 (0 and 少(0 will be holomorphic outside 丫，including the point at
infinity. In addition, one may assume 9(00) = 0.
The statement that 山⑹ must represent on 丫 the boundary value of
some function 少⑷，holomorphic outside 丫， takes by (76.13) the form
1「 f do 1 / <p(b) 而 1 1 (T* —I— 野J
——
M(a)
-•

— = 0,
—已 —E
m
2兀/J a
丫
2九E J
丫
a 2tc/ .
丫
b 1—*

where is a point outside 丫；noting that by (70.1'), i.e., by Cauchy's

formula for infinite regions,
1 平⑹aG
（9 + 平（8） 一 P（0 ,
-C
一平
2那
y
J a

one obtains

一 中(Q +
1 b?
2兀z J/ a 71
/ 1 — ‘

r 9 9) - 工 = 一可
1 ( / do
g)
方丁
— —
~
加o* b —G 兀 2 2 J a C
Y Y

This equation, corresponding to the functional equation (78.10) for the

general case, can immediately be solved, because

1 a2 + —ttz
展•丁二Q中⑹
is the boundary value of the function
1 F +加

holomorphic inside 丫，as a result of which the integral on the left-hand

side of becomes zero.
Since (p(^) is holomorphic outside 丫 and <p(oo) = 0,

最+…
_
落
Hence
吆 2&
〃八
中 w — ~ … for H| > 1,
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 349

so that
= — 瓦芋 — 2诙守 ... for K|v 1,

which proves the above statement.

Thus one obtains the very simple formula

建） (82.4)
Y

which determines <p(Q. In this way the boundary value 小(。) of the
function 少停) will be known by (82.3), and therefore “0 is given by
Cauchy's formula [(70. T)]
1 / 蛔而
m + 3（8）；
Y

substituting for 少⑹ from (82.3) and noting that (see remarks below)
1 /o 1 +
J (o)
de
= Q
1 + Q (C),
2K2 J
Y
<y2 —例 E% — C 加

one finds finally, omitting the constant W(8) which does not influence
the stress distribution,
1 十 喇2
(82.5)
? —加
Y

It is easily seen that the formulae obtained give a regular solution of

the problem, provided / has a derivative satisfying the H condition.
For the deduction of the formulae preceding (82.5), it is sufficient to note
that 中⑹ is the boundary value of 歹(1“)，holomorphic inside 丫， and that
1 +M 3)

is the boundary value of a function, holomorphic outside 丫 and vanishing at

infinity {cf. also preceding note).
The constant W(8) may be determined by (76.15); in fact,

弧 8）=
1
/
2m J CT
y
350 V. SOLUTION OF BOUNDARY PROBLEMS §82

Substituting for 少(g) from (82.3) and taking into consideration that
1+
do = 0, 平 '⑹do = 0,
2E£ J
y
—
a2 加

one obtains

W(8) ~~
7

Next, the general case will be considered and it will be treated in

accordance with the rule of §78. By (50.14) and (50.15)
X + iY
鹤) = 段 一 兀
T log 已 + 佻© (82.6)
2 (1 +7T
x)

「段
怵) = ——
4—-^4
iY}
log 已 %© (82.7)
2 (1 十 X)
注

where (p0(Q and %(? are holomorphic for | | > 1 and where one can
assume ^
伙)(8) = 0；

in addition, as always for the solution of the first fundamental problem,

it will be assumed that there is no rotation at infinity, i.e., that T = T.
Substituting these expressions in (82.2), it is seen that %(? and 少虱已)
satisfy the same boundary condition (82.2), the only difference being
that / must be replaced by /0, where
a2 + w TfR
/0 = / 一 PR a(l w(j2) a
+

+…:k + ―
X iY 浸
+加 (82.8)
+
2k(1 尤) 1 — wa2

It will be remembered that /0 will be single-valued on 丫， because the

increase of / for a complete circuit of 丫 will be compensated by the in¬
crease of the logarithmic term.
The functions cp0(Q and 少 may be found by help of the formulae,
stated above.
平 °(Q (824)
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 351

Wo(0 =
] f fo 击 1 + 欣2 ,, (825)
2兀2 J a— C
Y
•

and the problem is completely solved.

§ 82a. Examples.
1°. Stretching o f a plate with a n elliptic hole.
Let the edge of the hole be free from external stresses and let the state
of stress at infinity be tension of magnitude 力 in a direction forming an
angle a with the Ox axis. Then X = Y 0 and by (36.10) [putting
h=么 w2 = 0]
「=『 =

Also in the present case

* 「=

/ = 0. Hence (82.8) gives

—尚产 (82.1 a)

少R那
&
m )
ct(1
— 2a

PR /1 1 4- 夕&毋％
<j2 — 加) 2

—
The function (? + %)/〈(] • 3c2) js holomorphic inside 丫， except at
_
C = 0, where it has a pole with the principal part 防 the function
«1 泓芋)/((2 .加) is holomorphic outside 丫，except at〈 = 8,where
it has the form 加已 十 0(1 Hence, by the formulae of § 70,
1 / ^2 + 加 db m
2兀£J <t( 1 — 物
—C
N)
"
o C '

__ _
Y

1+ ma2 da 1 + 侬芋 (1 严
—； - .解 q =— Q
3
F = (2 —m
Further, it is obvious that

/
1 而
Wj 而二^
1
一七
Y

Thus (824) and (825) give

mpR 以*2也 pR^.^ —— m)
中 。© —才十一不= 灭
352 V. SOLUTION OF BOUNDARY PROBLEMS §82a

PR 淖口 + + 喇2 价9
1
%© = 4^ 而2二F 工飞 —m
i

and, finally, by (82.6) and (82.7)

学卜 一重七十
(1 + 加2) (g2za
m
— 加) , (82.2乃

and the problem is solved.

A solution of this problem, by a quite different method, was given by C. E.

Inglis [1], and it was found again in 1921 by T. Poschl [1]. It has been seen that
this solution is a very particular case of the general solution of the first fundamental
problem for the infinite plane with an elliptic hole which was published by the
Author [4] in 1919; cf. also the Author's paper [7]. A particular case of this problem
(tension in the direction of the major axis of the hole) was solved in 1909 by G. V.
Kolosov [1].
In 1931, L. Foppl published a (very complicated) solution of the above-mentioned
particular case which he considered as an illustration of his method of solution of
problems by the help of conformal mapping. The general method of F6Ppi (studied
in the same paper) is very difficult to understand (at least the Author has not
succeeded in doing so).

The calculation of the components of stress and displacement is not

difficult. Only the sum
pp 十 99 = 4 沆①(0
will be determined here, where by the preceding formulae

4 =寸
个十加 一 2四a
=4
(p2g2玲+w — 2产a)(p2e-2 份一的
—
百 二力
一 …

j_加 b产9 —明干 厂配 二防)

The denominator of the last fraction is real and equal to
p4 — 2碑 p2 Cos 29 + w2.
Separating the real part in the numerator one finds

-— —
p4 — 2P 2 cos 2(8 a) 加2 + 2物 cos 2a
pp + 9分 — 力
— p4 2mp2 cos 29 源
On the boundary of the hole p — 1 and pp = 0. Hence the value of 9眇
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 353

along the edge of the hole is given by

~
‘
99 =
- 4
1 — • —I—
Z
1 ——
—
cos 2a
—
2 cos 2(卧
cos 2分 加2
_
— a)L；
this formula, apart from the notation, agrees with that given by
T. Poschl [1] (Note that his formula contains a misprint) .
In the case of bi~axial tension, when at infinity

n、= n2 = 自 r = —力 , r = o,
one obtains either directly or by superposition of the preceding solutions
for a = 0 and a — —2
建) =
PR / q - 例\
丁)
_
SQ =
力R(1 + 加沉 .
2°. Elliptic hole the edge o f which i s subject t o
uniform pressure. In this case
Xn = ~ P cos (%, %),
—
Yn = Pcos (% y),
where P is the magnitude of the pressure ; hence
(Xn + iY n)ds = ― P{dy •
— idx) = Pidz.
Therefore

— p五(<7 + ?),

“
/= ak+过/公=— Pz =

f = — PR ( H 加b).
Substituting these values in (82.4) and (82.5) (assuming the stresses to
vanish at infinity) one finds
PRm PR PRm 1 次2
s—
， 、
少 (Q = = 3 • 卞 ,

and the problem is solved.

The displacements and stresses will only be calculated here for the
limiting case m = 1 (i.e., the straight cut; cf. Fig. 36.) ; their calculation
in the general case is likewise not difficult.
 354 V. SOLUTION OF BOUNDARY PROBLEMS §82a

In fact, one finds by (50.7), (50.9) and (50.10)

P(p2— 1)3 (p2+ 1)
pp = P+
— —
(p4 2p2 cos 2^ + 1)2 '

—
P(p4 — 1)(p4(1 +2p2p2cos+29p4 — I)24P cos 29)
2
9» = _P -J-
— 2
十

2Pp 2(p2
—1)2 sin 29
(p4 — 2p 2 cos 2联 +1)2 '

+ x)p2 cos 29 十 1 — x — 2P2

- 一南
PR (1 •

”p =
Vp4 — 2P2 COS 29 + 1
PR? (1
— x)2 sin 29
2“ V p4 — 2P cos 23 + 1

Fig. 36. Fig. 37a.

3°. Elliptic hole the edge of which is subject t o

uniform tangential stress T.
In this case

/ = iTz

=
(Xn + iY^ds = Tdzt

— iTR (<j + —•),

TRmi
Wg =
TRi
j
_
_
/=

TRmi 1 十 加芋
二

4°. Elliptic hole (o r straight cut) part of the

edge o f which i s subject t o uniform pressure.
— iTR ( — + ao).
As in the preceding example, one obtains (assuming the stresses to
vanish at infinity)

Consider now the case when the uniform pressure P acts only on
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 355

the part zrMz2 of the boundary (Fig. 37口) and when the stresses, as
before, vanish at infinity.
In this case (cf. example 2°.) one may take (beginning the circuit at zj

/ = — Pz = — +— on the arc

/ = — P& on the arc

Following around the contour {in anti-clockwise diyec杭oH) and returning

to Z], the expression / undergoes the increase

- P(Z2 - zj = P(4 zj

By (82.8)

/0 = / -j
X + Elog b 十 X iY— Q2
+W
— ma2
一
茶 2碎 田 1 •

尸（为 一 Z2）10g b P01 一 52) N+加

2几％ 2m(1 乂) 1 — me2

because, for example by (78.7), X + = iP区 — z2) ;

it must not be
forgotten that in the above-mentioned formula the direction of the
circuit is that which leaves the region occupied by the body on the left,
i.e., in the present case this direction is clockwise.
The value of the multi-valued function log a may be fixed arbitrarily
—
at any point (e.g. at the point j el近 ，corresponding to the point zj ;
for a circuit along 丫 the function log a must vary continuously, so that
for a complete circuit (in clockwise direction) log c undergoes an increase
2拓 and /0 reverts to its original value. Hence /0 will be single-valued and
continuous on the entire contour.
If /0 had discontinuities, this would correspond to concentrated forces at the
locations of these jumps; however, by supposition, no concentrated forces are
to be present. Note that the derivative of /0 has discontinuities at the points,
corresponding to zx and z2, but it is easily verified that the formulae, deduced below,
give the solution of the problem.

There remains the determination of <p0(Q and from (824),

(82.5'). Denoting by & the point of 丫 which corresponds to z2 (Fig. 376),
356 V. SOLUTION OF BOUNDARY PROBLEMS §82a

one finds

P& — 万2) 1 f
•
a2 do
—E
*
2兀“1 + %) 2m J 1 ~ m a2 b
Y

But

a—E
6—C

where log ——
%
must be understood as the quantity /0, with 0 being the
angular distance of the points % and a2> measured from j in anti-clock¬
wise direction ; further,

1 // erz +物
2几2 1
——
J —
a炉 a
说

― 3 — 0'

the latter result following from the fact that the integrand is holo¬
morphic inside 丫.
There remains the calculation of the integral

/(0 =
Y

which is most easily achieved in the following manner. One has

打 1 f
log a , 1 八 1

，
j

况 2m J (b
Y
—
C)2 2兀2 J
Y
b 工已

1 一 log a - 0=01 1
+ J
2病 b
- — = q 2痘
Y
Q
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 357

But by Cauchy's formula for infinite regions

/ 1 诙- 1
君/ 二

and
『J
-.logo 2拓

- b
—— 已 」(j = q 6 ~~

_
because for a circuit of丫 the function log a increases by 2兀工 Hence

〃 1 1

consequently

°
= log (久 一 — log ( + const.
7(0
Thus, omitting constant terms, one has

勺£9 = 2兀z 【 C
log 欢 十 R
b工 L\ « 十
Q /
一 ^2llog
J
(电 一 D —

「R("
—
where

Z] = R( b] ), Z? R( ^2 +

，
The function %(Q may be obtained in the same manner ; one thus
finds finally

三)
P f mR
述) =w{— ~
&
+ R + 一名 2log zQ — 9 1

——
R
(( Y-) - 句] l°g
H— 1

一 0 ~~
X
+ i
log n,
PJ K(l+%2) a2 1+ 喇2
WQ =
而l
一-
Hogw +项久一的
I +勿庐
三2 Sg 应 0 + Z/Og 也一 Q ]华(
X +* 1 X +1 〈2 —m
358 V. SOLUTION OF BOUNDARY PROBLEMS §82a

If the entire contour is loaded,

b] — b?， l°g bl
27r2

and one obtains the simple formula deduced directly for this case (cf.
example 2°).
On the other hand, letting the arc 司加面 tend to zero and increasing
P in such a way that lim P | — zr \ ~ F is a finite quantity, one
obtains in the limit the case of a concentrated normal force applied to
the edge of the hole.
It is also easy to find directly the solution for the case of any number
of arbitrary concentrated forces applied to points of the contour or to
internal points of the body. (Cf. the analogous solution for the circular disc.)
5°. Approximate solution of the problem of
bending o f a strip (beam) with a n elliptic hole.
The stress function
U= 24y3
不
corresponds to the following state of stress :
Xx = — Ay, Yv = Xy = 0. (82.3a)
Hence, cutting from the body a strip bounded by the straight lines
y = 土口，the edges of this strip will be free from external stresses, while
purely normal forces Xx = ―Ay will act on any transverse (i.e., parallel
to Oy) section of the strip. These forces are statically equivalent to a
couple with moment

M= 2h] AyHy = ^Aha^; (82.4©

—a
the forces act per unit thickness of plate (normal to the plane Oxy). Thus
the above function solves the problem of bending of continuous strips
(beams) by couples, applied to their ends (Fig. 38). The functions 中式幻,
少i(z) corresponding to U are easily seen to be
4源4 加2
甲i(z) = —q一 ， 以
O ——
= 一qO~~ • (82.5a)

It will now be assumed that an elliptic hole with centre at the origin
has been cut out of the strip. The problem of bending of such a beam
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 359

will now be solved approximately, subject to the conditions that the edge
of the hole is free from external forces and that, at large distances from
the hole, the state of stress
tends to that given by (82.3a) ;
thus it will be assumed that
the dimensions of the hole are
small compared with the length
of the beam (see also the penul¬
timate paragraph of this sec¬
tion) and the problem will be Fig. 38.
solved as if the elliptic hole
were in an unbounded plate. Under these circumstances one must have
月源
少i(z)= 梆 +
(82.6a)
Aiz2
g(z) = 此

where 鸡 are functions, holomorphic outside the ellipse including the

point at infinity.
For simplicity, it will be assumed that the major axis of the ellipse is
directed along the axis of the beam. The solution of the general problem
would only be slightly more complicated. The solution for the particular
case when the major axis is perpendicular to the axis of the beam was
found by A. S. Lokshin [1] using different methods.
Introducing the variable 匚 one has in an obvious notation

平(9 — 9o(Q H
— (七 + 丁)， (82.7四)
岭 )=%廿一善五2("守
Substituting from (82.7〃) in the boundary conditions (82.2) or (82.3)
with / = 0, it is seen that 中o(0, 机(9 satisfy the same conditions, provided
one takes instead of / or /

― -—
4立?2(1 加 )2
/o =
(82.8a)
A) =
— m)2 / 1 \2
360 V. SOLUTION OF BOUNDARY PROBLEMS §82

respectively. Substituting these expressions in (824) and (825), noting

that the right-hand sides of the preceding formulae represent functions
holomorphic inside 丫 with the exception of the point o = 0, where they
have poles with the principal parts

respectively, and applying (70.4), one finds immediately

%(9 =
一石 J
Y
~~~
，

1 7 /n do 1十 ,
%(Q 二

一 P2—
2m Jo f
k
Y
C
T 泮
§ m
喇 q=
—
…
(82例

•
R2A(\ — m)2i
—•例 户
- 工2
.
] +加守
嬴'
and the problem is solved.
—
For m 0, one obtains the solution for the circular hole, while m = 1
gives that for the case of a straight cut; it is easily verified that in the
latter case 平虱已) = 焉© = 0, i.e., a longitudinal cut does not influence
the state of stress.
The problem of bending by transverse forces and other analogous
problems may be solved in the same manner. A number of such problems
for the case of circular, elliptic and some other types of holes (in fact,
holes bounded by hypotrochoids approximating to triangles and squares,
cf. § 48, 4°) were solved and studied in detail by M. I. Naiman [1] using
the methods of this book. Many problems, important from the point of
view of application, were solved by G. N. Savin [2] who reduced them
to numerically convenient formulae and gave a number of tables which
enabled comparison of the deduced results with experiments ; a detailed
study is given in Savin's monograph [8]. More will be said about Savin's
work in § 89. S. G. Lekhnitzky [2] studied several cases of bending of
beams with circular holes at a somewhat earlier stage. Even earlier
than this, Z. Tuzi [1] gave the solution of the problem of pure bending
of a beam with a circular hole (which can be obtained from the solution
given above by putting m = Q).
Experiments with models have shown that the solution remains
sufficiently exact from the practical point of view, when the dimensions
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 361

of the hole are not small compared with the width of the strip, provided
they are not larger than 3/5 of the width of the beam for circular holes
(Z. Tuzi) and 1/3 for square holes (G. N. Savin) .
All the above solutions are approximate and based on the consideration
of an infinite plane with the corresponding holes. There also exists a
(fairly complicated) exact solution of the first fundamental problem of
the theory of elasticity for an infinite strip (of finite width) with sym¬
metrically distributed circular holes which was given by R. C. I. Howland
and A. C. Stevenson [1].

§ 83. Solution of the second fundamental problem for the

infinite plane with an elliptic hole. In this case the boundary condition
has the form
] a2 p- ^*2* •…-
一 M。)= 2Mgi + 物) = 2留 (83.1)
可
展 厂就应5⑹
or

3㈤
‘ ‘‘

。1
千7 '⑹ 一 少同
一 少
—
2照】一 &）= 2国 （83.2）
where g】， g2 are the given components of displacement of points on the
ellipse.
At first, it will be assumed that the displacements remain bounded
at infinity (i.e., X = Y=r = r‘ = 0); one then obtains in the same
manner as in § 82

80 —空 H
g"
, (83.3)
2m J CT "V
" Y
—
w(0=巴z •g
说 _C 加
.
c s(0+ 3(8). (83.4)
兀 J b
—— ——
？ m
If one leaves the value of 少(8) arbitrary, the boundary condition will
be fulfilled apart from a constant term. In order to determine W(oo),
. 1 de
multiply both sides of (83.2) by ——；
, .
and integrate around 丫. This is
2k/ a
easily seen to give [cf. remarks following (82.5)]

W(oo) =: (83.5)
362 V. SOLUTION OF BOUNDARY PROBLEMS §83

and the problem is solved for the case when the displacements are to be
bounded at infinity.
In the general case, assuming as in § 82 that P = F, i.e., that the rotation
at infinity is zero, one has
X 4- iY
鹤) = T成 一
-7—-
(1
2兀 x)
log C + 中。©
十
(83.6)

8Q = + 室二弋
2 (1 十 x)
兀
log c + %© (83.7)

Substituting these expressions in (83.1), one sees that 中。© and 堀(0
satisfy the same boundary condition as 中(已) and 小(已)，except that 2Mg
must now be replaced by

2尔0 2pig — T7? xg

+
r'K X— iY
——1) (83.8)
b 2兀 (K + 1 — WG2
whence follows

2照0= 2{zg — YR (\ b
cr
d —~
m )+
/

+ PR?
X + iY 1 + 加 02
(83.9)
2tc(x 十 1) g2
— m
The values of 仇依) and %© will be obtained from (83.3) and (83.4)
%, g°, g0 respectively. Thus (cf. § 82«,
by replacing
example 1°)
<p,

“ g, g by

…
—，
2u, 1 C g da „ R 、
r/ + (「加+「)一 ，
「

甲o(9 = 歹 (83.10)
X 2兀2 J b C y已
Y

X + iY
2k (x + 1)
1十
芋一m
加2
c些作
where Wo(oo) is determined by the following formula, obtained from
… 3), (83.11)
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 363

(83.5) and (83.9),

心3)=— 7V2j 上/显也= —号/J 或十"F (83」2)

y
b TZt
y
Q 2兀(比十 1)

It is easily verified that the above solution will be regular, if the function
g, given on the contour, has a derivative satisfying the H condition.
In the limiting case 加 = L one obtains the solution of the second fun¬
damental problem for the infinite plane with a straight cut.

§ 83a. Examples.
1°. Uni-directional tension of a n infinite plate
with a rigid elliptic centre. Let the infinite plate with

Fig. 39.

a rigid elliptic centre, welded into it, be subject to uni-directional tension,

as in § 82〃，example 1° (Cf. Fig. 39). It will be assumed that no external
forces, apart from those exerted by the surrounding material, act on
the rigid kernel, and hence X = Y = 0. In the notation of § 82以,
example 1°, one will have

r =f = £, r = —劭产a.
364 V. SOLUTION OF BOUNDARY PROBLEMS § 83a

The applied tension may cause a (rigid) translation and rotation of the
kernel. Since the translation may be eliminated by a rigid displacement

_
of the entire system, it may be neglected and it may be assumed that
the kernel rotates about its centre by an (unknown) angle s. The boundary
values of the displacement components will thus be
gl = 引， g2 = + 8%， (83. IQ
so that
g = iz{x 十 iy) = izz = izR (o ) , g — — ieR (
•
P

Further, since

/1 g Go izR
而 J
Y
= "V

and, by (83.12), tp0(oo) = 0, one obtains from (83.6), (83.7), (83.10) and
(83.11), putting X = Y = 0,
R
_
9(9 = + +巧
「加 —

r，w , 2"放
VC (1 + 冽叭 \)
। — +
—
(un 252口)
叭Q = I MH 已 H iK I\ L m /

+ (2固72或 + 「) (2 ——加
Tw + —.
R1
儿(
. +就 2

--
There remains the determination of the angle e from the condition
that the resultant moment of the forces, acting on the elliptic centre from
the surrounding material, must vanish. This moment will be calculated
by the help of (33.3).
Since in the present case 中⑷，少(〈)，and hence %(z), “(z), are single¬
valued, the resultant moment Mo of the forces, acting on the side of
the rigid centre, will be equal to the increase of 沆灯⑶ for a complete
circuit of the ellipse (in clockwise direction). Thus it will be sufficient
to calculate the multi-valued term of

wiC) = 猊/ s⑶必 = 汲 / MQs,g理

CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 365

「= 5' 十 2c
The second formula of (83.2a) shows that, putting
this multi-valued term is

卜2时甯( 1 +
猊
X
)log z 一 (
iC'mR2 1 十 — log 已

Hence

Mq —— (i + — 2nmB2C, f 1 +— %
(83.3a)

The condition Mo = 0 thus gives

加 (1 + x)C' 乂) sin 2a
£
—— ，

2x(例2 + X) 十 x)
(83.甸

and the problem is solved. In the case of a circular centre (m 0)

the rotation is zero, while in the limiting case of the segment of a straight
—
line, i.e., for a straight rigid reinforcement (m 1), it is —
p sin 2a
(83.4%)

In the case of bi-axial tension, when

— 力
r = r=£,
2
r = o,
one obviously will have 宫 = 0 and

g 师（… 加、 （1 十加2）已 1 +例j m\

2°. Case when the elliptic centre is not allowed

t o rotate.
If under the conditions of the preceding example (uni-axial tension)
the rigid elliptic kernel is restrained in its original position by a couple,
then a = 0 and (83.2a) gives

吸) (rw
R
延
—ro
(83.6a)
+ r^ + r
v m /+
= — ) .
— m 也 —
366 V. SOLUTION OF BOUNDARY PROBLEMS §84

The moment MQ of the couple restraining the kernel is, by (83.3讥)，

Mq = — 27m R2Cf ) = ― 兀勿忘2 (］ J_) sin 2a.

十 (83.7a)

3°. Case when a couple with given moment acts

o n the elliptic kernel.
It will be assumed that the stresses vanish at infinity. Then (83.2a)
gives
2yMzRi 2tLzRi / 1 十 初？ \
上
63 = 厂， 峭) = 卜+ 加 (83.8a)

where by (83.3a)

47中区2 例 十 x)

4°. Case when a force acts o n the centre of the

elliptic kernel.
It will be assumed that the stresses vanish at infinity. It is easily
seen that the kernel does not rotate in this case. In fact, this is obvious
from symmetry considerations for the cases when the force acts along
one of the axes of the ellipse ; the general case is then obtained as a com¬
bination of these two cases. Further, it may be assumed that the kernel, in
general, remains in its original position (because this may always be
brought about by a rigid displacement of the entire system). Hence
one has in (83.10), (83.1 1) and (83.12) : g = 「 0, = r = 0, whence it
follows that
5 C 一八
X 十 2N i + 源 加(X
+ 2y)
2兀 +1) u ~~- w1
2兀(Y 十 1)
where (X,V) is the applied force ; therefore, by (83.6) and (83.7),
x +立
9=一国二西其'
/八
中

—
(X zY)
乂 X + iY 1+ 加2
1°8已+ 2兀
皿x + 汉)
“° = 2兀3十1) (x + 「守一
1) 加 2tt(x + 1)
§84. General solution of the fundamental problems for regions,
mapped on to the circle by the help of polynomials. It is not ac¬
cidental that the solutions for the regions considered in the preceding
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 367

—
sections (§ 80 83) have been so simple and elementary. Actually, it
has been shown that the solution of the fundamental problems may always
be obtained in elementary form and that, in fact, it may be expressed
by Cauchy type integrals, provided the mapping function is rational.

The solution of the fundamental, biharmonic problem for the case when
3© is a polynomial was first given by E. Almansi [1]. T. Boggio [1, 2] stated
a method of solution of the second fundamental problem for the case when
is a rational function. The present method is quite different from, the methods
employed by these authors and is, in the Author's opinion, much simpler. It was
first studied in the Author's paper [4] and given in greater detail in his paper [5].

A beginning will be made with the first fundamental -problem and, in

particular, with the case when s(〈)，mapping S on to the circle [ [ < 1,
is a polynomial
3(Q = . . . + c jg (q NO, 篇 片 0) (84.1)
4-

(i.e., the region S must be finite *) ; cr cannot be zero, because 3«) would
vanish inside the circle and the transformation would not be single-
valued and invertible. No generality is lost by omitting a constant term,
i.e., it may be assumed that = 0 corresponds to z = 0.
In this case the functional equation (78.10), viz.,

律) 月(Q, H !< 1, (84

2m J 3,
Y

which determines the function (p(Q, may be solved in an elementary

*and very simple manner. It will be remembered that the function 力 (Q
is given by

A® = (84.3)
2兀％ j o —— q

as in § 78, it will be assumed that /=九 仔2, given on 丫, has a derivative

satisfying the H condition.

In the present case 33)/s'(b) is the boundary value of the rational

* Regarding infinite regions, cf. the end of this section.

368 V. SOLUTION OF BOUNDARY PROBLEMS §84

function
c/" %+•••+ 渥7

o -
= (84'
n
入 十 2遍 T . 品L+1 年7+ . . . 十 脸’

holomorphic outside 丫 (cf. § 63), except at —

co where it has a pole
of order n. Hence this function may be represented outside 丫 in the form
oo
bJ” + bn^~^ 十 .• 十 十品 + 2 "京心
加=1
(84.5)

It should be noted that it is unnecessary for the deduction of the solution

up to the boundary to calculate all the coefficients of (84.5) : it is sufficient
力o determine only %
algebraic operations.
z
_____
, bn> and this is known /o require only elementary

As a result of the fact that co (a) /co (a) has the above stated form the
integral on the left-hand side of (84.2) can be calculated by elementary
methods. In fact, 平' is the boundary value of 岁'(l/C), holomorphic
outside 丫 (§ 76, 2°). Hence
(b) —•
=
3’
中'o

is the boundary value of

3(E)

holomorphic outside 丫 except at 已 = 8 where it has a pole of order

not greater than 处. Its principal part there will now be found. One has
建) = 砧 + 碳2 十 ... 十 距y 十 册 … (|^| < 1) (84.6)
{where only those terms which have been written down will, in fact, 加 re¬
quired} . Hence
q/(Q = 诙 + 2〃2c 十…+ + (% + 1 4~ •
and therefore
%af
— (w)
/1\ 2&
= 藏 十 y +… +
i

+
°’
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 369

Thus, for ]
^M)>
| 1,

+ … + K田 + 0 (《)， (84.7)
where denotes a function, holomorphic outside 丫 and vanishing
for = co, while

Ko = 猴品 + +...+(%- 1) 源一1g_2+ %左/『1 + (% + 1)2+13，

K1 — 万161+ 2藏为 十 1 + 二九匕
*…
1 九 州)

a2 五抄2 十 2说力3 + ， +(n — 1)园一也， (84.8)

^n—1 = 216也一1 十 2a2纵 F

，
K 胃 — 应抄於

Therefore, by (704) and (84.7),

= & + &"•..+ KQ, (84.9)

Y

and (84.2) gives directly

SQ 十巨十A 十 。 K< + … + 1 C f da
= 4(Q. (84.10)
27rz J a
Y ——
已
In this expression the constants

are, at present, still unknown ; they must be determined from the con¬

。,
ditions that the constants 说( a2, . . . , 口久+力 occurring in the expression
(84.10) for <p&) by means of K Kt, . . K^, must represent the coef¬
ficients of (84.6). These conditions will now be formulated. Noting that

1 1 已 仔
O 6
十
节十/ 十…，
one has
幺 (9 = z兀z
1
~

Y
j
/ f da
g
—，q = 0+
4月 工十 力 2c2 + • • •, (84.11)
370 V. SOLUTION OF BOUNDARY PROBLEMS § 84

where
27r

Ak = — —/ /
27n J
击 = -—/
2k
/
J
= 0, 1,2, .... (84.12)
y o

Comparing coefficients of G, , Ct+1 on both sides of (84. 10), one obtains

the (% + 1) equations
瞅 Kk = 力肚， k
十 — 1,2, . . . , n. (84.13)
an+i = "n+r
The last of these equations gives directly the value for 〃*+】：

。儿+1 = 力您+i = ~~

2兀z J
f /「"2 而. (84.14)
Y

The remaining formulae, by (84.8), may be rewritten

al + + 2a 2b2 + .… 十 (％ — ~
Alt
a2 + + 2而力3十 •.・ +(气 一 1 )^n--^n = 42,
(84.15)
%-1 + 左 + 2无2图= 4 力― 1，

册 十 册如 =4 炉

Thus, a system of linear algebraic equations has been obtained for the
determination of the constants 诙，他，.…，〃九； this system agrees with
the system (63.6〃) which was obtained when solving the problem by a
different method.
Putting 〃丸 = a 尢 + i^k and separating real and imaginary parts, one
obtains from (84.15) 2% linear equations with the 2% unknowns
a比，瓦， 万 = 1 , 2, . . . ,
〃
(Instead of this system of equations one may construct the conjugate
complex system in terms of 藏 )
This system must have a solution, provided the condition for the
vanishing of the resulting moment of the external forces is satisfied,
ie, the condition *
* The condition (84.16) is obviously equivalent to the condition

J
L
/就 +
JL
加z = 0, or
J J
+ %名廊 = 0;
hence, taking into consideration (84.1) and (84.12), one obtains

G4 + 2工222 +…+ EnAn 乳

- — 2c242 一 … — - 0・
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 371

+ 助= 。； (84.16)
L

in fact, it is known that under this condition the original problem has
a solution, and hence the system (84.15) cannot be incompatible. In
addition, it is clear on the basis of the uniqueness theorem that the system
under consideration completely determines the unknown constants,
if one fixes (arbitrarily) the imaginary part of the quantity
3(0) _ 的
+
，公
L — 3T0) 硝0)
Fixing this imaginary part (arbitrarily) , finding the quantities 旬，&,
satisfying the system (84.15), substituting them into the ex¬
pression for Kq, Kn together with the expression (84. 1 4) for
and introducing the values obtained into (84.10) *, one arrives at
the expression for cp(Q which satisfies the relation (84.2) identically.
After this the function 少⑷ may be calculated from (78.8), viz.,

*) = Lf 与—
2m J
Y
a —— C 2兀z J
y
平号一
—C3
丽 (o) cr 3.⑺

in the present case the second integral on the right-hand side is expressed
in finite form by In fact,
3

is the boundary value of

* The constant an+1 appears only in the expression for since this last
constant may be neglected because it does not change the stresses, it is actually
not necessary to compute an+1 and Ko.
Further, the substitution of %, a2, . . . , an in the expressions (84.8) for
...
Ki，K2, . Kn actually need not be performed, since the values of KQ,
can be evaluated by use of (84.13) from the formulae
Kn

比 — 为 a 论， 为 — 1,2,
Finally, it should be noted that the special form of the system of algebraic
equations (84.15) makes it possible to apply to its solution fast methods which
greatly ease the task of computation, as was shown by M. M. Kholmianskii [1],
372 V. SOLUTION OF BOUNDARY PROBLEMS §84

f f /Z\
8，
「
飞
holomorphic inside 丫，except at = 0 ; on the basis of (84.7), it is easily
seen that one has inside 丫 *

_ 1
_ 1 .
... + K
，(0 《+
= Ki 施可
十 a holomorphic function.

Hence, by (70.3),

1「 3 '⑹说
2兀打 33 3 — 0 b©
北） 一件 n

and (84.17) gives

求）
Y
c
/ do
— { 口严
a) + w
左 京
+ … + 清—丽 • (84.18)

The preceding results apply, with obvious insignificant modifications,

to the case of g” infinite region S mapped on to the circle | | < 1 by a
function of the form ^
= + + . . + cK". • (84.1')

* If one puts temporarily

艰) MC) = Q(C),

one has in the notation of § 76, 2°

3©
Q*&)
石'

and
Q(? = [Q*(C)]*
*
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 373

The problem must first be reduced to the case when cp(Q and 中«) are
holomorphic inside 丫 (§78). Afterwards, the procedure stated above
the system, analogous to (84.15), will always have a (unique) solution
without the supplementary condition (84」 6) .

NOTE. 1. The solution of the problem considered in this section can

also be obtained by the following process leading to essentially the same
computations as the above.
Noting that

is holomorphic outside 丫 and has at infinity a pole of order not higher

than n, i.e., that it may be represented in the form (84.7), where
K], . . . , Kn are constants which are not known beforehand, one
finds, as above, the formula (84.10). However, the constants Ko> Kv
. . . , Kn can be determined by direct substitution of the expression for
<p(Q obtained from (84.10) into the initial functional equation (84.2),
whence

- Kq — K工 — ….
— Kn丁 - r
~

2Bz J 硝引
f _U
{M + 2K 2d + . . - +
Y

州"巧 ——
b
-
—
已
—
2兀Z J
上 L
(g — Q
= 0. (84.19)
Y

The integrals appearing on the left-hand side are readily computed

in finite form, just as the integral of the formula (84.9) was evaluated *
and the left-hand side of (84.19) becomes an ”-th degree polynomial
the coefficients of which involve the quantities KQ, Kx, Kn, Kn
linearly ; setting these coefficients equal to zero, one finds a system of
linear equations in the stated quantities to the solution of which the
original problem is reduced.
NOTE. 2. For the study of the system (84.15) one relies not only
on the uniqueness theorem for the solution, but also on the less ele¬
mentary existence theorem, proved in § 79. However, one can proceed
* For this purpose one must use the expansion (84,11), or rather {n + 2) of
its first terms.
 374 V. SOLUTION OF BOUNDARY PROBLEMS §85
readily without application of the existence theorem, limiting oneself to
the uniqueness theorem and obtain thus an elementary proof of the
existence theorem for the particular case of interest here, when co(Q is a
polynomial.
In fact, it is readily verified directly (and this is left to the reader
as an exercise) that by (84. 1 6) , which must be assumed fulfilled, one of
the 2% equations obtained by separation of real and imaginary parts in
the equations (84.15) is a consequence of the remaining (2%
equations * . Therefore, if one adds to these (2” 一 1 ) equations an
1) —
equation expressing that the imaginary part of 引3‘(0) is equal to an
(arbitrary) given quantity a (for example, zero), one obtains a system
of 2钝 linear equations with 2% unknowns which is certainly soluble,
because its corresponding homogeneous system obtained for / = 0,
a = 0 has no nonzero solutions as a consequence of the uniqueness
theorem.

§ 85. Generalization to the case of transformations by means

of rational functions. The cases of regions, mapped on to the circle by
the help of polynomials and of functions of the form (84.1'), are particular
cases of regions mapped by means of rational functions of general form.
In this more general case the solution may be obtained by the same
method as above ; the only difference is that in this case one has, in
general, to calculate roots of a certain algebraic equation.
Consider again the functional equation (78.10), viz.,

—b . 1
*)十 —r- — 2九%
口 = 49；
(85.1)
6T
Y

although the notation used here is the same as in the preceding section,
the function (o(Q is now a rational function of general form which trans¬
forms conformally the given region S on to the circle | ] < 1. In the
case when S is infinite it will be assumed that the point z = 8 cor¬ ^
responds to the point 0. —
Also in the present case the integral on the left-hand side of (85.1)
* It then also turns out that, if the condition (84.16) is not fulfilled, the system
is insoluble. It will be noted that actually it is more convenient to operate instead
of with the above system with a system obtained by adding to the equations
(84.15) their conjugate complex equations, a fact which has been mentioned
earlier.
CHAP.15 SOLUTION FOR PARTICULAR REGIONS 375

may be evaluated by elementary means, as it will now be proved.

Consider the expression s(b)/3‘3) which is the boundary value of
the rational function

Since 3(〈) may now have outside 丫 poles other than at = oo, w(l/.Q
may have poles inside 丫，and not only at the point 已 = 0, as it was the
case in § 84. The function 石(1/C) cannot have poles outside and on 丫
except at〈 = 8, because s(? must be continuous inside and on 丫,
—
except at C 0 in the case when the region is infinite. Similarly, it will
be remembered that 3‘(已) cannot have zeros inside or on 丫.
Denote the poles of <o(Q, other than the pole = oo (if it exists) , by
叠乂2，…， Q；these poles are the roots of the algebraic equation
— 0 to which reference has been made earlier, and all of them will
lie outside 丫. Then the poles of the function 石(l/已)，other than the pole
已 = 0, will be
r ，一 ， _2_ r ，_ _L
Q]
乙 — s>2 J%

all of which lie inside 丫. These points and, generally speaking, the point
C = 0 will also be poles of the function &(l/Q/3‘(0 which lie inside 丫.
Hence this function may obviously be represented in the following
manner :
斌。 C, n 叫
Ckl
—a (Q =7 十2三 N E 十 R© (85Z
? fc-i 2=1 (C
— 嬴尸
where c0, . . . , q, ckl are known constants, is a rational function,
holomorphic inside and on 丫 and vanishing at 乙 = 0, and mQ, , wn
are the orders of the poles 0, 亮，…，以，respectively.
Consider now the product

&G)
3 (U
Obviously, this product represents a function, holomorphic inside 丫 with
the exclusion of the points 0, 乙；，…，& where it may have poles of
376 V. SOLUTION OF BOUNDARY PROBLEMS §85

order mQ> (but not higher) , and hence it may be written

in a form, analogous to (85.2),
的) C， n m比Ckl
Q(? = + E 招十工 £ + Q°(Q, (85.3)

where Co, Cu Ckl are constants and Q()(0 is a function, holomorphic

inside 丫 and vanishing for = 0.
It is easily seen that the following important statement is true: the
constants (% = 1, 2, . . 出。) are linear combinations (with known
constant coefficients) of the quantities
MO), b'(o),
and the constants C 端 are similar combinations of
MG 5'(G，…，平(叫G，不 = i, 2, …，气.

These relations are readily written down. Further, as is easily seen, the
。
constant C is a linear combination of the quantities (a) and of 6m
It will be seen below that for the solution of the problem under con¬
sideration here (i.e., the first fundamental problem) there is no need to
actually compute Co. When speaking later of the quantities Ct, this will
always imply that Z 1.

Now consider the expression

3 3) —
on the left-hand side of (85.1) which is the boundary value of

唳卜 拊《)

where Q()(
道
、
By (85.3), this may be written in the form (remembering that 瑟 = 1/晨)

G) = 0。+ 萍 泊 舍等 悬G) -
— )is holomorphic outside 丫 and vanishes at C = 8.
恪玛

The
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 377

constants Ckl are obviously linear combinations of quantities, con¬

jugate complex to and (6) ; the same is true for CQ.
Applying (70.4'), one immediately obtains
_
1 / 3(<7)Cp'((T)〃b = Co + S
犯)
— +2 n
2
左
Ck^
2*
y
J 诟J*? 2=1 k— 1 l—l 解一 索

It is easily seen that the expression

n
c一。+ z eg + S S
Z= 1 k= 1 Z= 1 &T)z
may be written in the form

Co + s 一
馆0 九 加比

z=i
+s s
而=1 z=i (— J)
where C[, CQ are constants, i.e., using the notation of § 70, in the form
Gg© + G*) + ... + G”©.

Substituting this expression in the left-hand side of (85.1), one finally

obtains
8Q + 左 + & +Z £&? + £ £ ，尸岁 = 月(9 • (85.5)
=1 =1 Z%1 (金 一 9
左

From this follows an expression for cp© as a function holomorphic

inside 丫 (and continuous up to 丫)，because the points Q lie outside 丫.
It remains to formulate the condition that the quantities (a) and (b),
linear combinations of which are the coefficients Ci and Ckl, actually
are the relevant derivatives of 勺(？，as determined by (85.5).
This condition is readily expressed by differentiating the equality
(85.5) a corresponding number of times and substituting for 已 the
corresponding value 0 or g, J . . . ,
For example, one must have
独
一 一
d(o) + & + E =
1

etc. Thus one deduces a system of linear equations (with constant co¬
efficients) in the unknown quantities (〃)， and their conjugate complex
values. This system (cf. § 84) will have a unique solution, if, in the case of
finite regions, the imaginary part of 中'(0)/a>'(0) is fixed arbitrarily and
if, in the same case, the following condition is satisfied:
378 V. SOLUTION OF BOUNDARY PROBLEMS §85

L
J (&&% + = 0， (85.6)

which is necessary for the existence of the solution of the problem.

Having found the quantities (© and (b), the values of C?, Ckl can also
be determined ; substituting their values into (85.5), one arrives at an
expression for 中(？ which is exact apart from a constant which may
be omitted, since it does not influence the stress distribution. However,
if this constant has nevertheless to be determined for some purpose, it
is sufficient to compute from (85.5) the value of 勺他"1)(0) and to
substitute it together with the values of the quantities (a) and (b) into
the linear combination expressing Co.
Having found 勺 (Q, the function 少 (Q may be found from (78.8), viz.,
1 / s(b) Mb)而 F (85・7)
而/ 西Y

3
•
whence, taking into •
consideration that ―汀丁中'⑹ is the boundary
3
value of the function Q(。，one obtains directly

电） 如
a s
2兀2
y
J a — ( 3'(Q k— 12=1

where the constant term —(p(0) may be omitted.

Thus the problem is solved. In the case of infinite regions, it is sometimes
more convenient (mainly for the sake ot clarity) to use the transformation
on to the region [ J > 1 ; but this is not always so, since the method
used here is equally applicable to finite and infinite regions. With obvious
minor modifications, the above reasoning will also apply to that method
of conformal mapping.
NOTE. With obvious modifications one may also apply the method
of solution stated in Note 1 at the end of § 84. Also the observations in
Note 2 of that section are readily extended to the case considered here.
The method of solution, discussed above, may be somewhat modi¬
fied, and in some concrete cases this may lead to considerable simpli¬
fications. For example, it will sometimes be profitable to multiply
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 379

beforehand both sides of the boundary condition by a conveniently

chosen polynomial. One such method was stated in the earlier editions
of this book ; generally speaking, it leads in the end to about the same
amount of computations as the present method, although in certain,
individual cases it simplifies the calculations.

§ 86. Solution of the second fundamental problem. On the

solution of the mixed fundamental problem. In the preceding
sections the first fundamental problem has been considered for the sake
of definiteness. However, comparing the boundary conditions of the
first and second fundamental problems in the form given in § 78, it
becomes clear that the above methods of solution may, almost without
any change, be transferred to the case of the second fundamental problem.
Hence there is no necessity to restate the method separately for appli¬
cations to the second fundamental problem.
The solution of the mixed fundamental problem is somewhat more
complicated ; however, in this case too, effective solution can be achieved
by elementary means when, as in the preceding sections, the transforming
function co© is rational. Such a solution was stated by D. I. Sherman
［10］. However, no time will be devoted to this problem here, since a
simpler method will be studied in the next Part.

§ 87. Other methods of solution of the fundamental problems.

Reverting to the first fundamental problem, attention will now be
drawn to the fact that in some cases it is practically more convenient
to start from the boundary condition (51.3) rather than from (51.1) ;
this condition may be written

［①⑹ + 6扁西'一d［画①Q) + j(b)里⑹］= R—z•两而j(87.1)

or
［丽 + 3' 3 酮］=自十， (87.2)

the left hand sides of these equalities must be understood as the corre¬
sponding boundary values the existence of which is assumed.
When co (9 is a rational function, the method of § 85 again leads to an
elementary solution. On the basis of the work of that section, its ap-
380 V. SOLUTION OF BOUNDARY PROBLEMS § 87a

plication is so obvious that no space will be wasted on details (cf. N. I.

Muskhelishvili [5] for a detailed study) ; however, a simple example will
be presented in the next section. It should only be noted that this method
is particularly convenient in the case when the region, occupied by the
body, is infinite, because in that case <p(Q and MO are not single-valued,
whereas (？and T(Q are holomorphic throughout the region under
consideration.
Similarly, the method of solution of the second fundamental problem
may be modified by replacing the boundary condition (78.15), which
may be rewritten
*勺⑹ —3 一 3⑹ = 2M(gi + 鹤2)， (87.3)
by a condition obtained by differentiation of (87.3) with respect to 9;
noting that o = e诒 and multiplying the differentiated equation by
—
( 小 , one thus finds
，
[m① (b) — 3 砥2心 ，⑹ 子⑹] —

「学
=2科 L_
do
十*] 如」
(87.4)

§87a. Example.
Solution of the first fundamental problem for
a n infinite plane with a circular hole*).
In this case let
z = 3© = 周， (87.1a)
where R is the radius of the hole, i.e., the region is mapped on [ | > 1 .
The boundary condition (87.1) then takes the form ^
瓯 一 。①'(0 — 炉子⑹ N — 汉， (87.20
where N and T are the normal and tangential external stresses with the
same sign convention as in § 56 (in fact, N is the projection of the external
stress on the normal n to the circle, directed towards the centre, while T
is the projection on the tangent, directed to the left when looking along 外).
For the sake of simplicity, it will be assumed that the stresses vanish
at infinity. Then (〈)，空(已) are not only holomorphic outside 丫，including
the point at infinity, but they also vanish at infinity, if it is assumed that

*) This problem has been solved by another method in § 56.

CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 381

the rotation there is zero. Thus, for large [ ],

©=告十0倍)， a)=0田，子(0=年+ 0(廿 (87.3乃

in addition, if the displacements are to be single-valued, the condition

然 1+ a；= 0 (87.3%)
must also be [cf. (56.6) ; % and
satisfied, a；
may be determined before¬
hand (cf. § 56), but this will not be done here.]
Calculations may be somewhat simplified by multiplying both sides
of (87.2夕) by 厂】 — this gives
1
一①⑹ + —
CT
1
b
'⑹ — G里⑹ = — (Xn — iY^ (87.4«)

or, going to the conjugate complex expression,

b
+ 画可 一币q 一 西诲 〃十 iY^t (87.5说)
because, as is easily verified,
N — iT = — 啰 期一 (87.6彷
It will be assumed that each of the functions (Q, '(Q, 里 (C) are
separately continuous up to the boundary.
Formulating now the condition that aT(a), as determined by (87.4a),
is the boundary value of holomorphic outside 丫，one obtains by
(76.13)
1 C G 1 喃加
b 1 C 说
2H J —— 已
丫
g
十
2九% J
丫
b E — • 2兀% J —已
丫
ct
十

I
十 ——r ，乙m
/ = 0, (87.7口)
2兀2 J
Y
b
—— G

where 已 is some point outside 丫，or

— @9 + 旬 + 2兀z j
[ 一
o —— q - 击 = o,
382 V. SOLUTION OF BOUNDARY PROBLEMS §87a

whence finally

(0 = /y [
2九北 J
4巴土-学-
b
— Q
而
+ 年.
； .时
Y

In transforming (87.7a), use has been made of the formulae of § 70 and of the
fact that g (cr) is the boundary value of C (C), holomorphic outside 丫 and equal
〈 = 8, and that o (o), '(o) are the boundary values of (1/Q, '(1/Q,
to % for ；
holomorphic inside 丫.

The constant 旬 is not determined by the functional equation (87.7g)

which will be satisfied by the expression found for ©(Q for any value
of 曲；in fact, expressing that on the right-hand side of (87.8a) is
the coefficient of in the expansion for one obtains the identity
However, this was to be expected, since, for the present, no
consideration has been given to the condition of single-valuedness of the
displacements.
Next, the function 灯(Q will be determined. Its boundary value is
known from (87.4a), if (Q is replaced by (87.8a). In order to find
from Cauchy's formula, one has to know its value for 乙 = 8； this
value may be found from (87.4b) by multiplying it by do/2兀力 and by
integrating around 丫 which gives
_ , 1 广
/ (^n — .而 「 、 1
2rc
X—iY , 、
々1 — ai — — — — — z- / (Xg — zY (口)
丫 o
where (X, Y) is the resultant vector of the external forces, applied to
the edge of the hole.

The following considerations lead to (a):上

o2
and
<y —
Oz(a) are the boundary
values of functions, holomorphic outside 丫 and vanishing at infinity like ^-3;
hence the corresponding integrals are zero. Further,
1-/1、
1 ~.
—— —
0(a) is the boundary value
,
of ~
)holomorphic inside
\F 丫， except for〈 = 0, where it has a simple pole
with the principal part 诙/：Finally, T(ct) is the boundary value of a function,
holomorphic outside 丫 and having, for large |
"
the form

三 + o 信)
The quantities 曲 and % may now be determined from (873a) and
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 383

x + wy
2的1 工-外
， 为" —
*(X iY)
2hR(1 十 x)
(87.9a)

Applying Cauchy's formula to determine from its boundary value

as given by (87.4讥)，one finds
1 fxn~ iYn （Q '9 a.
喔本 丁二三-丽 + 飞
J r+
》
(0 / （
(87.10a)
Y

Thus the problem is solved by (87.8。) and (87.10a), where aT and a；

have the values given by (87.9a).
As a simple example consider the case when a uniformly distributed
force, parallel to the Ox axis, acts on the right-hand half ( 旬《
2 《 —
A 兀/2)
of the hole, while the other half is free from stresses *. This problem was
solved directly in the above-mentioned paper [1] by D. M. Volkov and
A. N. Nazarov who did not deduce general formulae of the type (87.8«)
and (87.10办 TT rr
In this case 瑞= p 二 for … ,

箱= 0 二 for the remaining values of 况

二 0 for all 机

Hence

X= Rj Xnd肾 = r:Rp) y = 0 ,
and therefore
%1 = —, ai1 - —2(1
2(1 +x) + 乂)
Since Yn = 0, the integrals in (87.8a) and (87.10«) are equal to

where the last integral must be taken along the right-hand semi-circle in
* It is readily verified that the preceding formulae are applicable in spite of
the fact that Xn is discontinuous.
384 V. SOLUTION OF BOUNDARY PROBLEMS §88

the positive direction, so that

J
/ o
—Cy = 沙[log (C —— Mb = -w = 力 log —C—十~~~2;
Y

for a suitably chosen branch of the logarithmic function.

Substituting this value in (87.8〃) and (87.10^), closed expressions are
obtained for (0 and ¥(?, but it is unnecessary to write them down
here.

§ 88. Further examples. Application to some other boundary

problems.
1°. The method of solution of §§ 84—87 is applicable, in particular,
to all those simply-connect ed regions for which the conformal transfor¬
mations on to the circle have been stated in § 48.
The case of the infinite plane with an elliptic hole which is one of the
examples of § 48 has been considered in detail in §§ 82, 83.
The case of finite regions, bounded by Pascal's limacon, has been
studied in § 63 as an application of the method of series expansions ;
application of the method of § 84 leads much faster to the final results. It
will be left to the reader to solve the fundamental problems for this case
by use of the method of § 82.
The case of the infinite plane with hypo trochoidal holes (cf. § 48, 4°)
has been studied in detail by G. S. Shapiro [1] who applied the method
of § 82 to several practically important problems (cf. also reference in
§ 89 to the work of G. N. Savin).
The solution of the first fundamental problem for regions, bounded
by Booth's lemniscates (§ 48,6°), was obtained by G. N. Bukharinov [1]
by means of the method stated in § 85.
Several other examples which are of greater interest from the point
of view of application will be stated in the next section.
2°. The problem of frictionless contact between an elastic and a rigid
body under mutual pressure may also be easily solved by a method,
—
analogous to that of §§ 84 87, provided the region, occupied by the elastic
body, can be mapped on to the circle by means of a rational function. The
solution of the problem by such a method was given by the Author in
his paper [19] and it was studied in detail in the second edition of this
book. A simpler solution of this problem was given in the third edition.
It will be reproduced in the next chapter (§ 128).
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 385

3°. As has already been stated in § 79例 the problem of bending of

plates under the influence of lateral loads reduces, in the case when the
edges of the plate are clamped, to the fundamental biharmonic problem,
i.e., to the same boundary problem as the first fundamental problem of
the plane theory of elasticity, while in the case of free edges it leads to
the same boundary problem as the second fundamental problem. Recent¬
ly (and almost simultaneously) A. I. Kalandiia [1] and M. M. Fridman
[2] have shown that the case when the edge of the plate is supported
leads to a problem analogous to a certain problem of the plane theory of
elasticity ; in fact, it leads to the problem referred to under 2° above
(cf. also § 128).
Therefore, if the region occupied by the plate is mapped onto the
circle with the help of a rational function, the effective method of solution
studied above may be applied to it directly (or almost directly in the
third case). Regarding the third case, cf. A. I. Kalandiia [2]. (See also
the earlier papers of A. I. Lourie [1, 2] and M. M. Fridman [1].
In another paper, A. I. Lourie [2] gives the solution of the problem of
bending of circular plates for all the three above-mentioned edge con¬
ditions.
Note also the recently published note by M. M. Friedman [1] on the
bending of plates with curvilinear holes.
Finally, it should be mentioned that it has recently been shown by
L. A. Galin [3] that the method of complex representation in conjunction
with complex transformation also allows the effective solution of some
boundary problems in those cases where parts of the body are subject
to plastic deformation ; generally speaking, problems of this type are
very complicated, since the line of division of the elastic and plastic
regions is not known beforehand *.

§ 89. Application to the approximate solution of the general

case. The above method of solution may also be applied successfully
to the approximate solution of the fundamental problems for simply-
connected regions, bounded by practically arbitrary contours. It will now
be indicated how this can be done and it will be necessary for this purpose
to repeat some of the statements made at the end of § 63.

* Cf. likewise G. N. Savin 网.

386 v. SOLUTION OF BOUNDARY PROBLEMS § 89

Let S be a given region bounded by one simple contour L and let

z= co® (89.1)
map the region S on to the circle [ | < 1. It will first be assumed that
S is finite. Then ^
is holomorphic for | ^| < 1 , and hence it may be

-
expanded, for the stated values of J in a series of the form
3(Q 冬 + 媒2 + …； (89.2)
it has been assumed here that c0 = (o(O) = 0, but this is, of course, not
essential.
If one only retains the first n terms of (89.2), i.e., if one. takes instead
of s(Q the polynomial

熊) = 冬 + C2? + ... 4- (89.3)

then
Z = 熊)(89.4)
will map on to the circle ] | < 1 some region Sn, and not S. If one takes
^
n sufficiently large, the region Sn will be as close as one pleases to the
region S ; it has already been indicated in § 63 that this will be so for
known, very general conditions referring to the contour L. In practice,
it is usually sufficient to retain only a small number of terms in the
expansion (89.2), in order to obtain a region, sufficiently close to S.
In very many cases even a crude approximation is sufficient. For
example, if one is dealing with frequently occurring practical applications
of the equations of the theory of elasticity to bodies such as rock stratas
which are far from being homogeneous, it is clear that in such cases
great accuracy is unjustifiable. Thus one may practically solve a given
problem for the regions by retaining a number of terms in (89.2) which
will be sufficient for the stated purpose, and the solution will represent an
approximate solution for the original region S.
In the case of infinite regions, one has instead of (89.2) an expansion of
the form

3© =
* 十冬

(assuming c0 = 0) and instead of (89.3)

+ 媒2 + ... (892)
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 387

3熊)
=£+，＜ 嗟2 + ... + 盘丁 ， (893)

and the earlier statements will again apply.

Naturally, any other expansion in a series of rational functions can
be used for 3(C) instead of these power series. It may be proved that
under known, general assumptions with regard to the contour L and to
the selected method of expansion, the solutions for the regions Sn will
tend to the solution for the given region S, when % f 8； the proof is
given in the Author's paper [6] and, for more general conditions, in a paper
by D. I. Sherman [5].
It will only be noted that the method of approximate solution
gives good results even in cases when L is not smooth, but has corners
(angular points), e.g. when 7L is a polygon. In order to transform regions,
bounded by straight line segments, on to the circle, the known Schwarz-
Christoffel formula may be used.
The above method was applied successfully by G. N. Savin to the
solution of a number of practically important problems. Referring the
reader in the first place to the recent monograph of G. N. Savin [8]
as well as to his papers [1,2] and the paper by A. N. Dinnik, A. B.
Morgaevskii and G. N. Savin [1], consideration will here be restricted to
two examples taken from G. N. Savin [1] (cf. also G. N. Savin [8])
which illustrate the practical usefulness of the method (cf. also the
recent paper by C. A. M. Gray [1]).
As a first example consider the region, represented by the infinite
plane with a hole in the form of an equilateral triangle. In this case the
mapping function may be written in the form
y

= — A I (1 — 於)”7^—F const.,
i

where 4 is a real constant determining the dimensions of the triangle.

Expanding in a power series, one finds for a suitable choice of the arbitrary
constant

By retaining the first two or three terms of this expansion one obtains
388 V. SOLUTION OF BOUNDARY PROBLEMS §89

instead of the triangle the contours illustrated in Figs. 40 or 41 re¬

spectively.
As a second example consider the infinite plane with a square hole.
In this case

碎） 0 + /平我 + const.,

where A determines the dimensions of the square. Expanding in a

power series and choosing the appropriate value for the arbitrary constant,
one finds

The contours corresponding to the retention of the first 2, 3 and 4 terms

of this expansion are shown in Figs. 42, 43, 44.
It is seen that an approximation, sufficiently good for most purposes,
is given by three terms. By a slight modification of the coefficients of the
terms retained even better approximations may be obtained ; practical
methods for deducing better approximations have been evolved by
M. I. Naiman in a paper which has not yet been published. G. N. Savin
also considers in detail the case of holes with straight sides of different
lengths.
In order to avoid any later reference to this problem, it will be noted
here that the above methods may likewise be applied to the semi-infinite
regions considered in the next chapter.
CHAP. 15 SOLUTION FOR PARTICULAR REGIONS 389

Finally, it should be stated that these methods of approximate solution

can also be applied successfully to regions bounded by several contours,
if they are combined with the so-called "alternating method” ("Schwarz
algorithm") or the method of successive approximation, analogous to
that used by Schwarz in solving Dirichlet's problem. This method admits
reduction of a given boundary problem for regions, bounded by several
contours, to the successive solution of the same problem for several
regions, each bounded by a single contour, for successively varying
boundary conditions. An infinite number of such operations is required
for the exact solution, but practically useful approximate solutions may
be obtained after a finite number of steps. Each separate problem for
a region bounded by one contour may likewise be solved approximately,
using the above method.
It should be noted that the method of successive approximation has

—-
been developed by S. G. Mikhlin [5, 9, 13] and by D. I. Sherman [5] ;
/ X

J
Fig. 42.

Fig. 43.
1
6 56
390 V. SOLUTION OF BOUNDARY PROBLEMS § 90

a study of their results may be found in Mikhlin's book [13]. Note also the
work of A. Ya. Gorgidze [1, 2]. A convergence proof of the Schwarz algo¬
rithm for very general conditions has been given by S. L. Sobolev [2].
The method of successive approximation was applied by S. G. Mikhlin
[4] to the solution of the first fundamental problem for a half-plane with
an elliptic hole. This problem has been solved by D. I. Sherman [4]
using a different method.
Finally, the recent papers by D. I. Sherman [24—26] will be noted
which give new successful examples of the effective solution of certain
boundary value problems which are of significant practical interest.
 Chapter 16

SOLUTION OF THE FUNDAMENTAL PROBLEMS FOR THE

HALF-PLANE AND FOR SEMI-INFINITE REGIONS

Hitherto consideration has been restricted to regions bounded by

(finite) contours. The study of the case where the boundary is an open
line, extending to infinity in both directions ("semi-infinite region"),
does not meet with any essentially new difficulties. In certain cases it
is convenient to map the region under consideration on to the half¬
plane rather than on to the circle (there being, of course, no essential
difference between these two methods) . The general case will not be
considered in the present chapter; only the solution of the fundamental
problems for the half-plane and for certain definite semi-infinite regions
will be treated.
The general case of semi-infinite regions has been studied by S. G.
Mikhlin [7]. In recent papers, R. Tiffen [2, 3] gives what appear to the
Author to be much more complicated solutions of the problems considered
in this chapter in which is reproduced without essential changes the
study presented in the second (1935) and third editions of this book. In
particular, in R. Tiffen [3], the method of solution with the aid of con¬
formal mapping is set forth for the case of a parabolic contour as in the
Authors，own work (cf. § 95) .

§ 90. General formulae and propositions for the half -plane.

1°. Let the region S, occupied by the body, consist of the "lower'' half¬
plane (Fig. 45) bounded by the
Ox axis, i.e., of the points
y < 0. In §§ 90, 91 temporary
use will be made of the no¬
tation of Chap. 5, i.e.,
W(z), 子⑶
will again be written instead
of cpi(z), “(z) etc.
It will be assumed that the
391 Fig. 45
392 V. SOLUTION OF BOUNDARY PROBLEMS §90

stress components satisfy those conditions of continuity and

differentiability which apply throughout the preceding work. In addition,
it will be assumed that the stresses, and likewise the rotation (which is
inessential, but somewhat simplifies the study), tend 力。 zero, as z ap¬
proaches infinity along any path remaining in S. If the boundary of
the region S did not extend to infinity, but were instead a circle, it
would follow from these conditions that and T would have the forms

=
Z - -1— — 淤
里⑶ = —z —j-
z2 — .
for large ) 2 | (cf. § 36).

In the present case the condition will be imposed that the functions
and T may be represented for large | z | by

=工十 0 (—), 甲(z) =— —F 0

(90.1)
'(z)

where 丫 and 丫' are constants. (With regard to this choice, see also the
Note at the end of § 93.)
In addition, the functions (z) and 乎(z) will be holomorphic in every
finite region, contained in S.
The following conditions may be added to (90.1)：
<p(z) log z 十 0
=丫 + const., (90.2)
少⑵ = Y’log z + o(l) + const. ;
in these formulae one definite branch of the multi-valued function log z
must be selected, e.g. log | z | 十 £出 where 分 (argument of z) varies
from — 7：to 0.

It will be remembered that the symbols 0 (1/z) and o(l) denote quantities such
that
。G) Y
1 I 。 IV %

where e only depends on | z ] and e -> 0 as | z | f 00. The condition (90.2) would
follow from (90.1) by an integration, provided one had on the right-hand sides of
(90. 1) 0 (1/zl + w) instead of (1/z), where is a positive constant (which is arbi¬
trarily small) .
CHAP. 16 SEMI-INFINITE REGIONS 393

Finally, the further condition will be introduced : the resistant vector

of the external forces, applied to a segment AB of the Ox axis, tends to a
definite limit A and B move to infinity (月 towards the left and B
towards the right) . This condition will always be satisfied, if only a finite
part of the boundary is loaded.
This last condition will now be formulated mathematically. If Xf, Y'
are the components of the resultant vector of the external forces, applied
to AB, then one has by (33.1)

X' + zY = (90.3)

where

-
an
ox ——
F
au
cy
= 平 (z) 十 z</(z) 十 W(z) —— + ——
= 平 (z) 十 z W(z). (90.4)

In order to be justified in applying (33.1) to the segment AB of the

boundary, certain conditions must be imposed on the behaviour of (90.4)
near the boundary. However, one may avoid this without any additional
assumptions, if one replaces the evaluation of the resultant vector of the
forces applied to 4 R by that of the resultant vector of the forces applied
to any simple arc A'Br in S, the ends 4', B' of which lie infinitely close
to the points 力，B (cf. § 35). For example, one may take as the arc AB
an arc obtained from the semi-circle ACB (Fig. 45) by selecting on it
力 '，8' infinitely close to A, B.
In the sequel, in order to simplify the notation, the letters A, B will
be written instead of A' , B' and this should not cause any confusion.
If A and B lie sufficiently far away and on different sides of 0,
then (90.1), (90.2) and (90.4) give
dU~]B
+ £不一
效」 4
Y log + log
— + 丫高『 『位 + * (90.3')
where rf and y" are the distances of A and B from 0 and e is arbitrarily
small (and tends to zero as y" increase) . In order that the preceding
expression will remain finite for any abitrarily large 夕' and 广 (indepen¬
dent of each other), it is obviously necessary and sufficient that

丫十 Y‘ = 0. (90.5)
394 V. SOLUTION OF BOUNDARY PROBLEMS §90

Under this condition the vector (X, Y) of the external forces, applied
to the whole of the Ox axis, will be given by

X^iY = i +% — oo
(90.6)

It follows from (90.5) and (90.6) that

X + iY X —2kiY (90.7)
Hence one has finally for large ] z [ :

峭= -
X + iY
2^+。
〃 、 __
6
X 十 iY
= —
/ 1\

+
…
m 、

0
X — iY
G)，
(901)

x + ‘y , ,
log 2 十 o 、 十 const.,
27c
(902)
X~iY , ,、
少⑵ =
2P
log z 十o + const.
Note also that under the above conditions the stress components
X?, X”, 丫夕
will be of order O(l/z), while the displacements will have for large | z |
the form

2出侬
— y' log z — Y —z + o(l) + const. =
说) = 乂丫 log z

x(X + iY) , X + iY _ X — »Yz (90.8)

= 2冗
log
— Z log H
zk
Z =z + 2冗
0
小
const.
上

If X = 丫 = 0, then Xx, Yy> Xy will be of order o (1/z) and 勿 十 iv will be

bounded.
The same fundamental problem may be set for the region S as for the
regions considered in the earlier chapters of this Part. One has only to
give special consideration to the fact that the behaviour of those quanti-
CHAP. 16 SEMI-INFINITE REGIONS 395

ties which are given on the boundaries should be in agreement at infinity

with the conditions imposed above.
2°. In the first fundamental problem, the quantities Y y = N(E) and
—
Xy T(/) will be given on the axis Ox as functions of the abscissa t.
On the basis of (90.1'), it is easily found that for large | t \

T=o (：). (90.9)

In the second fundamental problem the functions u and 2 given on Ox,

must satisfy by (90.8) the conditions
I— ]
2从(％ 谢) =
2冗
( X + %Y) log 力 + c 十 o(l) for t > 0,

. + 1 .
— —2
乂
悌 谢) 人(X
H
+ log|t ] + c +
W)" (90.10)

+ 2k
(X 十 汉) 0(1) for t < 0,

where c is a constant.
The reader will easily establish analogous conditions for the mixed
fundamental problem^
the cases 0/ the second fundamental and of the mixed problems the
quantities X, Y will be assumed known.
3°. Only the solutions of the stated fundamental problems will be
given below (§§93, 94, 1 13, 114). Consideration will be restricted to
the uniqueness theorems.
The uniqueness theorems for the present case are readily proved by
a method which is completely analogous to that studied in § 40 for the
case of infinite regions. In the present problem, one must apply the
integral formula (40.4) to the region bounded by the segment AB of the
boundary and the semi-circle ACB (Fig. 45), and then go to the limit
when A and B move to infinity in opposite directions. This proof is
directly applicable to the case when the displacement and stress com¬
ponents are continuous up to the boundary without the point at infinity,
where they behave in accordance with the assumed conditions.
Thus, in this case, the proof of the uniqueness theorem is readily
carried out for the first and second fundamental problems of § 42 under
the assumption that the solutions under consideration are regular, i.e.,
396 V. SOLUTION OF BOUNDARY PROBLEMS § 90

that the corresponding functions cp(z), cp'(z), W(z) are continued con¬
tinuously at all finite points of the boundary.
It may still be added that by the existence theorems the solutions
of the second and mixed problems are determined completely and the
solution of the first fundamental problem apart form a rigid body
displacement (but only a translatory one) ; this follows from the fact
that by an assumed condition the rotation corresponding to the solution
under consideration vanishes at infinity.
NOTE. 1. The formulae (90.1), (90.2) or (901), (90.2') may be
replaced by others which are more convenient for the study of the
behaviour of the functions under consideration near the boundary.
For example, one may obviously write instead of (902)
x+
中(Z) = 2tc
log (z — Zo) + <p*(z) + const.,
(902')
———z
X iY
少 = log (Z — Zo) + %*(z) + const.,
where is an arbitrarily fixed point outside S (i.e., a point of the upper
half-plane) and tp*(z) are functions, holomorphic in S and of order
o for large [ z |.
NOTE. 2. O n concentrated forces applied t o the
boundary.
If one retains only the first terms in the formulae (90.2'), i.e., if one
writes

p(z) =
X + gy log 幺， 少⑵ —
X ~iY
log z (90.1 1)
27V 27t

and applies them to the whole half-plane S, then it is easily seen that they
correspond to the effect of a concentrated force (X, Y) applied to the
boundary at the origin. In fact, for a circuit along an infinitely small
semi-circle below O, the expression dU /讯十胃 increases by
and hence the resultant vector of the forces, applied (from above) to this
semi-circle, equals (X, Y) ; further, it may be shown that the resultant
moment of the same forces about the origin is zero.
The components of stress and displacement corresponding to these
functions 中 and 少， i.e., to the effect of concentrated forces, may be cal¬
culated by means of the general formulae of § 32 or § 39. For example,
CHAP. 16 SEMI-INFINITE REGIONS 397

——
one has by § 39 for the polar components of stress
X + iY 2
〃+99 = 4员q '(z) =— 4沉
2大夕
= — ——
•

加夕
cos 分十 y sin a),

—
99 〃+ 2力8 = 「
2 印'(z) + w'(z)W = —— cos 盼+ v sin 9),
whence

rr = (X cos 9 十 y sin 2), 83 = 0, 心 = 0. (90.12)

7zr

This solution of the problem of the effect of a concentrated force on

the boundary agrees in essence with the solution found by Flamant
(Cf. A. E. H. Love [1] §§ 149, 150). This problem is the two-dimensional
analogue of the problem of the effect of a concentrated force on the
boundary of a body, occupying the half-space (the boundary of which
is an unbounded plane), i.e., of the so-called problem of Boussinesq.

§ 91. The general formulae for semi -infinite regions. Next

consider semi-infinite regions and the generalization of the formulae
of the preceding section to this case. Let L be the boundary of the region
S which is a simple open line extending in both directions to infinity.
The line L divides the plane into two parts one of which is S ; the second
part will be denoted by S'. The positive direction on L will be chosen in
such a way that it leaves S on the left.
It will be assumed that the line L has the following property : if Mo
is some fixed point, the rays M0A, linking MQ to two points A and B
of L tend to definite positions as A and B move to infinity along L in
opposite directions.
Let II(M0) be the (signed) angle covered by the ray MqM as the point
M, moving along L in the positive direction, describes the entire line.
It will be said that II (Mo) is the angle subtended by L at Mo. It is easily
seen that the magnitude of II(Af0) will be the same for all points on one
side of L and that the angles IT, IT subtended by L at points lying
in S and S' respectively are related by the condition
n — IT = 2 兀. (91.1)
For example, if S is the lower half-plane, its boundary L is the Ox axis,
but the positive direction of L is the negative 0% direction (because S
398 V. SOLUTION OF BOUNDARY PROBLEMS §91

must lie on the left for motion in that direction). Then, for points Mo
of the lower half-plane, one has II —
k, while for points Mo of the upper
half-plane IT = 兀.—
In the present case of semi-infinite regions it will be assumed that
the functions (z), ¥(z), '⑵ are also subject to the condition that for
large | z | (cf. § 90)

- 。
f

Y 丫
(2) = — 0 里
z
1
0 '⑶ = ― ~

or, what is the same thing,

Z
—
- Zq + 0 (1)'
\ Z /
中⑶
f 二— (；)'
Z
0
\ Z /
(91.2)

where z0 is some (arbitrarily) fixed point of S' (i.e., not in S) ; in (91.2)

0(1/冲) are symbols for functions, holomorphic in S and having
for large { z [ the indicated order.
The following may still be added to these conditions (cf. § 90):
中 (z) =丫 —
log (z + o(l) + const., (91.3)
W = Y’log (z — + o + const.;
here 。(1) is the symbol for a function, holomorphic in S and tending to
zero as | z | 8.
As in the preceding section, it will also be assumed that the resultant
vector of the external forces, applied to an arc 4B of L, tends to the
definite limit (X, ¥) as the points A and B move to infinity in opposite
directions.
In the present case one finds instead of (903) the formula
B
+% = Y log f + y115 + 7(浮 B — 浮力 +
夕

+ log
『 — £ (913)

where one may move along L, since does not lie on this line, and where,
as above, IT is the angle subtended by L at points of S', while a, 8 are
the (signed) angles between the 0% axis and the limiting positions of
CHAP. 16 SEMI-INFINITE REGIONS 399

the rays and M^B, drawn from some fixed point MQt as A and B
move to infinity along L, the first in the negative and the second in the
positive direction ; 歹' and r" are the distances of % from A and B re¬
spectively and e is a quantity tending to zero as A and B move to infinity.
Clearly, one may assume
0 — a = rr. (9i.4)
As in the preceding section, the conclusion is drawn that one must have
Y +p=0 (91.5)
and that

X ^iY = i + /管］= (y 一 y) IT — "产B — 产力

or, using (91.5),
X十 = 2ITy — 耶
— 滔犷. (91.6)
In the analogous formulae of the preceding section it had been assumed that the
boundary of S was described in the negative direction ; hence one had to take

It will now be assumed that IT #0, i.e., that H # 2兀. Then, adding
to (91.6) the relation obtained by transition to the conjugate complex
expression and solving for 丫 and y, one finds
21r (X + iY) +“於币 — 产a)(X — iY) (9L7)
4(n‘2 — sidlT)

and, by (91.5),
2II'(X — iY) — — 歹2巧 (X + iY)
4(n‘2 — sid II')
(91.8)

In deducing these formulae it has been assumed that LT # 0. If

IT = 0, then (91.6) gives X = Y = 0. This shows that for IT = 0 it
is necessary for the existence of a solution under the above conditions
that the resultant vector of the external forces, applied to the boundary,
be zero.

§ 92. Basic formulae, connected with conformal transformation

on to the half -plane. It is convenient for the solution of problems of
400 V. SOLUTION OF BOUNDARY PROBLEMS §92

the theory of elasticity for semi-infinite regions to make use of trans¬

formations on the half-plane rather than on the circle. (As stated earlier,
there is no difference in principle between the two approaches.)
As for transformations on to the circle, it is advantageous to introduce
on the z plane of the elastic body curvilinear coordinates, related to the
transformation.
As before, denote the region under consideration by S and its boundary
by L. Let
z= [z x iy, C = E 十 河 (92.1)
be the function mapping S on to the lower half of the 已 plane, i.e., on to
the half-plane 刀 V 0, so that finite points correspond to finite points.

Straight lines 叮 = const., lying in this half-plane, obviously correspond

in the S region to some open lines which go to infinity at both ends;
these lines will be denoted by (E). Similarly, the semi-infinite straight
lines 己 = const, in the lower half of the
〈 plane correspond in S to lines
which begin on L and go to infinity (Figs. 46见 46s.

Since in S a completely definite point z = + %) of the z plane

corresponds to every pair (E, for 叮 V 0, the quantities 之 and 刀 may
*
be conceived as curvilinear coordinates in the z plane. The lines ©
and (刀) form an orthogonal net of coordinate lines.
Let z be some point of S. Draw at z the tangents to the lines (&), (7))
in the directions of increasing 已 and 7). These tangents which will likewise
be denoted by (E) and (力 will represent the axes of the curvilinear coor-
CHAP. 16 SEMI-INFINITE REGIONS 401

dinates at the point z (Fig. 46a) ; all this is quite analogous to the procedure
in§49. 今

Let A be some vector starting from the point z = (已 十 and let

Ax, Ay be its projections on the axes 0%,Oy and its projections
on the axes (? and (力. As in § 49,

己+ + 2/ J (92.2)
where a is the angle between the axes (E) and 0%, measured from 0%
in the positive direction.
In order to determine ^a, the point z will be given a displacement dz
in the direction (g) ; the corresponding point 已 will then undergo a dis¬
placement 延 > 0 in the direction 已 of the 已 plane. Obviously

dz^\dz *= |3«) 评延 血 = 3‘(0诙,

whence
3 亿） 3 亿)
T西 — JcoW (92.3)

Thus one obtains

(92・2‘)

Denote by 荻，％ the components of displacement and by 陵，

the components of stress in the (2), (刀) directions of the curvilinear
coordinate system. By (92.2'),

,I .» •、
s'© 依+划
.. (92.4)
吃十啊
]
<7 1 4
， ।Ai a

一丁诋丁
where u, & are the components of displacement with respect to 0%, Oy.
By § 8, the following relations hold between the stress components in
the two coordinate systems :

蓝十石 = 十 h， w 一近 + 高 2 = (匕 一羽 2汉川巴 (92.5)

where, by (92.3),
好)
一
3 亿) 3 9 bq
402 V. SOLUTION OF BOUNDARY PROBLEMS §93

The expressions for the components of displacement and stress in

terms of functions of the complex variable 已 may be found as in § 50;
it will now be agreed to denote by
女⑶，队⑵，6⑶，巴(z)
the functions
6% 山⑶，① (z), 里 (z)

of Chapter 5 (and likewise of § 90) . As in § 50, write

建) = 物心9］, 怵) = 娟
(92.6)
9 6心◎ = 笔，
3 (Q
中(Q = 巴心⑼ = 书
3 (0 4
Applying now the formulae of § 32, expressing Xx> Yy, Xy in terms
of % ", i，%,one finds by (92.5) and (923)

2+不 = 2［① (Q + 亚］= 4员①9 (92.7)

台一经十 2通 一一{词 a)+ 硝9》(0}. (92.8)

3‘任)
Using (92.4) and the formula

2以说 + 说) = wp(C) — ,
38 :中代)
3'(9 —— MQ, (92.9)

the expression for % + 勿％ is also easily deduced. Finally, adding (92.7)

and (92.8), one finds the useful relation

布 +岛 = +函 + L{碉①a) + 3名)中(0}. (92.10)

3'©
§93. Solution of the first fundamental problem for the half-
plane. Let the body S occupy the lower half-plane. By the conditions
of the problem
丫夕 = Ng Xy = T«) on Ox, (93.1)

where N(£) and T(/) are given functions of the abscissae t (which represent
the normal and tangential stresses).
By (32.8)
—
Yy iXy = (z) + 丽 + z函 +呵 (93.2)
CHAP. 16 SEMI-INFINITE REGIONS 403

Hence the boundary condition may be written

(力) +画 + 频J +西 = N — iT (93.3)

(%) (力 纯＞'(/) 子团 =N +汉. (93.4)

The left-hand side of (93.3) must be understood as the boundary
value of the right-hand side of (93.2), and an analogous statement holds
true with respect to (93.4). In what follows it will be assumed for the
sake of simplicity (unless stated otherwise) that there exist also boundary
values (力)， '(力)，甲 of the functions (z), '(z), 乎卜) separately.

The condition (93.3) could, of course, have also been deduced from (41.23) ; how¬
ever, it must not be overlooked that the quantity T of the present section is the
—
quantity ( T) of that formula. This follows from the fact that, when moving
along the 0% axis in the positive direction, the region S lies on the right, and not
on the left.

It will be assumed that N and T are continuous functions satisfying

the condition (90.9), i.e., that

N ~ o (93.5)

From the condition that ¥(%), defined by (93.4), is to be the boundary

value of some function 甲(z), holomorphic in the lower half-plane and
vanishing at infinity [as follows from (90.1)], one obtains by (76.21)
+8 +8 +8
N — zT 1 C Q)也 画业
1
/ 以
1
27r J勿
—
t z
OO
— 2兀£ J
—8
力
— z J
—CO
t —z
~

0,

where z is any point of the lower half-plane. But Q) is the boundary

value of (z), holomorphic in the lower half-plane and vanishing at
infinity, and Q), 炖， Q) are the boundary values of (z), z '(z), ho¬
lomorphic in the upper half-plane and likewise vanishing at infinity
[cf. (90.1)], and hence, by (722) and (72.2), the conclusion is drawn that
404 V. SOLUTION OF BOUNDARY PROBLEMS §93

the second integral in the preceding formula is equal to — (/), while

the last two integrals vanish. Thus

(z) =— (93.6)

Having found (z), the function 乎卜) may be determined from (72.2),
since the boundary value is given by (93.4). Thus one obtains,
using the formulae of § 72,

1「 N + iT
__
中(z) =— / 北 — 0 一 绅'⑶ =
—
8
十 十8
oo

r —
z「 N iT 二
— -- 7zJ
1
-I
—8
t
Tdt
—— Z
1 /
2兀/ J (/ —— Z)2
— OO
dt =

48 十8

— —
1
// N 十汉， 1 CN
I
iT
— — —
-
力威
7
dt -j (93.7)
2m J
—8
t z — 2兀zJ (力 一 z)2
— OO

It is readily verified on the basis of § 68, § 69, 1° and § 71 that, if

NQ), T«) and their first derivatives N'(t), T'(£) satisfy the H condition
at the finite points and the derivatives tNft), t2Tr(t) satisfy
the H condition likewise near the point at infinity, the expressions
obtained for (z) and 里 (z) satisfy all the postulated conditions. In
particular, the functions (z), '⑵，空⑵ are continuous up to the
boundary and have for large | z | the required form determined by
(90.1) and (90.5). Thus, the problem has been solved.
The above result agrees essentially with that obtained by G. V. Kolosov
[1] by another method. The same problem was also solved later (and
independently of G. V. Kolosov) by M. A. Sadovski [1, 2]. But neither
of these authors presented a strict investigation of the solution.

NOTE. 1 . It is readily verified directly on the basis of results in § 68

and § 69, 2° that, if N«) and T«) satisfy only the H condition (including
the point at infinity) , the displacement and stress components correspond¬
ing to the functions (z), 里⑵ and determined by (93.6), (93.7) will be
CHAP. 16、 SEMI-INFINITE REGIONS 405

continuous up to the boundary and that the boundary condition (93.1)

will be satisfied. But then the functions (z) and ¥(z) may not be
continued continuously on the boundary, which was not required by
the problem formulated ; the function (z) and the expression 乞① '(z) 十
中(z) will be continued continuously on the boundary.

NOTE. 2°. The right-hand sides of (93.6) and (93.7) obviously are
holomorphic functions in the lower as well as in the upper half-planes, but,
in general, they are not analytic on the common boundary Ox of the
half-planes. It is clear, however, that, if any part of the boundary remains
unloaded, the right-hand sides of (93.6) and (93.7) will also be analytic
on that part, and hence (z), 里(z) may be continued analytically through
this part from the lower into the upper half -plane.
This property of the solution is easily proved directly without re¬
ference to (93.6) and (93.7). In fact, let
Q(z) =
— (z) — z — X(z) ; (93.8)
since, by supposition, 乎⑶ are holomorphic in the lower half¬
(z) and
plane, Q(z) is likewise holomorphic there. Next consider the functions
币(z), C(z) = — & (z) — z@(z) —-甲(z)
which are holomorphic in the upper half-plane. By (93.3) and (93.4), one
has on any unloaded part of the Ox axis
= Q(z), 耻/) = QQ), (93.9)
where <!>(/), Q«) are the boundary values, assumed by the corresponding
functions for z / from the lower half -plane, while (%), Q(£) are those
assumed for z % from the upper half-plane.
It follows from the first equality (93.9) that holomorphic in the
upper half-plane, is the analytic continuation of (z) from the lower into
the upper half-plane, and hence the analytic continuity of (z) is proved.
Similarly, the second equality (93.9) leads to the conclusion that Q(z)
is analytically continued into the upper half-plane, where it takes the
value (z). Hence it follows by (93.8) that T(z) may also be analytically
continued, and the earlier proposition is proved.
In particular, it is now seen that, if only a finite segment of the boun¬
dary is loaded, the functions (z) and ¥(z) may be expanded for suf¬
ficiently large | | in Laurent's series.
406 V. SOLUTION OF BOUNDARY PROBLEMS § 93a

These results present simple means for studying the behaviour of

(z) and ¥(z) for large | z provided the behaviour of the stress com¬
ponents at infinity is known. It will be remembered that the behaviour
of (z) and 甲(z) at infinity had been postulated a priori in § 90 and that
this step seemed to be artificial in character. However, it will be easy
now to remove (or diminish) this artificiality. In fact, it may be readily
proved that (z) and 里(z) necessarily have the form, given by (90.1), if
only a finite part of the boundary is loaded and if the stresses are, for
example, subject to the following condition : the quantities
$陷 +h)
* v y
A”, “y *
~

oy
tend (uniformly) to zero as z moves to infinity (remaining, of course, in
the lower half -plane). No space will be devoted here to the proof which
will easily be provided by the reader.

§93a Example. As an application of the above results consider

the case when the segment
《
口 t 4?
《

of the 0% axis is subject to a uniform pressure 力，while the remaining

part of the boundary is free from external forces. Then T = 0 for all t,
N —— for N / M 十 a, N = 0 for the remaining values of t,
and (93.6), (93.7) give
+a -ra
力/也 力/业 z力/ dt
= 后 彳不
J
一a ia
7=7 ， 乎⑶ —
a
i
'

whence

H
力
[log (Z
— 切胃 =
力
log
z
—，
——
a
a

(93. 1 a)
paz
于⑶一 获言』.

Note that the results of § 93 have been applied, although in the present case the
given function N(Z) is discontinuous ; the correctness of the final result may be
verified directly.
CHAP. 16 SEMI-INFINITE REGIONS 407

In (93.1a) the term log (z — 幻 /(z + a) means the increase of the function
log (z — £) for a continuous change of t from
—
—
41to + 圆 For greater
clarity, write z t ~ 龄，where p = \z~t\ and 6 is the angle between
the vector starting from t and ending at 2 and the axis Ox which
will be assumed to lie between 0 and 兀 and to be measured from the
positive 0% axis in clockwise direction (Fig. 47). Then
.
log (z —— = log p —
.
log
Z
Z
' '

十〃
醇
log
Pi
p2
(Oj — %),
~ (93.2(2)

—
where & 02 is the angle subtended at z by the loaded segment of the
Ox axis.
The stress components will now be calculated. One has
2力
+ 兀 = 4猊① (z) = —— 他 一 ％), (93.3©

一 X, + 2iXy
- 2加
一二 2
- 2
2叵① '⑶

4力冲
中（明
7C2 22
4力冲（浮 一 洲
— ' (Z2

(93.甸
九啰 . 谈） — （步
九 . 谈）
（浮 — 四2）

Fig. 47.

whence, finally,
力 2力热贝炉 — — * .
X,= 一 装他一 员）
m 谟二帝不不广
408 V. SOLUTION OF BOUNDARY PROBLEMS §93a

y _ — (瓦 — ％)
2力冲 (%2 —夕2
—*
晨或 , (93.5a)
兀
齐钉 ，
H[(先z + 尸 一 a2)2 + 4a2y2]

砂g到2

—
X
" 兀[出 + 外 a2)2 + 4a2y2]
The solution of this problem was first given by J. H. Michell [3] and
was later obtained by G. V. Kolosov [1 , 2] by a different method. (However,
both papers by Kolosov contain a misprint in the expression for Xy,
where appears instead of d).
"
The law of the stress distribution becomes clearer, if one writes in
(93.4a) the term / — as
超 一 谟 = pm

then
y, _ X。 2iXy = —国丝
十 /他+限 (93.4%)
P1P2
which gives, in conjunction with (93.3口)，
y cos (0T + %)，
——P1P2——
力
% = —2 他―
、
2^— -- ,
H

cos + %) , a)、
—
力 、 歹 me 广,

Yy = (A】 ％) ~~
八

2夕g (93.5
兀 P1P2
v
Av =— 2%
供
夕 sin (% + %) .
P1P2
These formula demonstrate that the stress components are continuous
up to the boundary, provided the points 力 = ± 笈 are excluded. At these
points they cease to be continuous but remain bounded (as is seen
by noting that y = —
% sin 国 = — p? sin 如) K is likewise clear
that the boundary conditions are satisfied.
The components of displacement are also readily calculated and it
is verified that they remain continuous up to the entire boundary (in¬
cluding the points t = ± provided the point at infinity is excluded,
since they increase with | z | f 8 like log | z
The solution for a tangential stress, applied to a segment of the boun¬
dary, may be deduced just as simply.
CHAP. 16 SEMI-INFINITE REGIONS 409

§ 94. Solution of the second fundamental problem. In this case

the boundary condition may be written

W的 1 5Q)― 少(力) = 2M(gi 十 鹤

2)(94.1)
or
水网 一彳5(%) — = 2[1原 一 总). (94.2)
It will be assumed that the displacements remain bounded at infinity
which, by what has been said in § 90, is equivalent to the condition
X = Y = 0. (The more general case, considered in § 90, is easily reduced
to the preceding one by means of a method analogous to that of § 78.)
The condition (94.1) may be replaced by one obtained by differentiating (94.1)
with respect to £; in that case one has only to deal with (z), ¥(z), and the difficulty
arising from the presence of the logarithmic terms in cp(z), ^(z) is removed. The
problem will be solved by such methods in § 1 13, 2°.

Under this condition and those of § 90 the functions 山⑶，平 '⑶ =

= (z), W'(z) = 乎⑶，holomorphic in the lower half-plane, must satisfy
(90.1') and (90.2') with X = Y = 0. 0nly the following of these conditions
will be considered here:
= 。(1), W(z) = C + 。⑴， ＜p'(z) =0 (94.3)

where c is some constant which is not given beforehand (so that the
basis of the problem is somewhat more general in comparison with the
conditions of § 90) ; the constant term in the expression for (p(z) has been
omitted, as usually, without affecting generality.
By (90.10) one has, in addition, to assume that for large | t \ the
given functions satisfy the condition
gi + & = G 十 。⑴， (94.4)
where G is a constant which is, in general, complex. Further, it will now
be assumed that g】十 ig2 satisfies the H condition on the boundary,
including the point at infinity.
Expressing that the function 少(片)， determined by (94.2), must be the
boundary value of the function W(z), holomorphic in the lower half-plane,
one obtains by (76.21)
4-oo 4-oo +8
f/ glt + 很z 2 加上比
兀wJ
[ 中㈤北 1 f 也
TV?
—8
—— 2
—8
t '~

—8
410 V. SOLUTION OF BOUNDARY PROBLEMS §94

where 2 is an arbitrary point of the lower half-plane, or, applying the

formulae of § 72,
+8

—— n中⑺… = ——
f + ig2 四 i
J/ ——
r
7TZ t Z
—8

where, in particular, use has been made of the fact that 如 '(£) is the boun¬
dary value of z弓'(z), holomorphic in the upper half-plane and vanishing
at infinity. The value of c is obtained by letting z f 8 (in the lower
half-plane) ; then, using the second formula (71.15), one finds
= —— [iG.
Thus
十8

-及
y中 h
— J
包
— OO
t Z
1lG. (94.5)

The function 少⑵ is now easily determined from its boundary value,

given by (94.2); in fact, applying (72.2‘)，one obtains

+初
— OO
or, applying again the formulae of § 72 and substituting for c,

一 对 (z) — 国, (94.6)

where, in particular, use has been made of the fact that <p0) is the
boundary value of 0(z), holomorphic in the upper half-plane and vanishing
at infinity. It is easily seen, on the basis of the results of §71, that these
functions @(z) satisfy all the conditions of the problem, including
(94.3), if, for example, the expression g】 ig^ and its derivative 十 温 g；
with respect to t satisfy the H condition and if the expressions
+ ig2 — G), *(g；+ Eg；) satisfy that condition near the point at
infinity. Thus the problem is solved.
If it is not only required to satisfy (94.3) but also the conditions (90.1'), (90.2'),
it is sufficient to assume, in addition, that also 滑(g；+ Eg；) satisfies the H con¬
dition near the point at infinity.
CHAP. 16 SEMI-INFINITE REGIONS 411

The solution of this problem (by other means) was likewise given by
M. A. Sadovski [1, 2] who also made a careful study of the character
and of the conditions for the existence of the solution.

§ 95. Solution of the fundamental problems for regions, mapped

on to the half -plane by means of rational functions. Case of a
parabolic contour. When the given region S may be mapped on to the
half-plane by a rational function 3©, the fundamental problem may
be solved by elementary means, as in the analogous cases of § 84
et seq.
In view of the analogy with the earlier work, consideration will be
limited here to an explanation of the method of solution by the concrete
example, where the boundary L is a parabola and S is the part of the
plane, lying outside the parabola (i.e., not on the side of the focus).
Consider the transformation
z = 3(0 = 诋 一 诵产， (仇 0), (95.1)
ie,
% = — 2氢刀一内 y=1一①—
The real axis 刀 0 of the 已 plane corresponds in the plane to a line
with the parametric representation
% = 2就， y = 烂 一 谈，
i.e., to the line
炉 = 4/(y + 口2) ； (95.2)
this is the parabola L with parameter 2谈，its axis parallel to the axis
Oy and its vertex at the point (0, ~ 谈)；the origin is the focus of the
parabola.
When the point E moves along the 已 axis from the left to the right,
the corresponding point z moves along the parabola likewise from the
left to the right.
It is readily verified that (95.1) maps the region S, outside the parabola,
on to the half-plane 刀 V 0. The coordinate lines (? and (刀) are easily
seen to be confocal parabolas ; the axes of the parabolas (? and (刀)
are orientated in opposite directions. Fig. 46。of § 92 shows several
parabolae of the family (? [i.e., 刀 = const.] and those parts of some
parabolas 己 = const, which are included in S.
The angle, subtended by the parabola at points inside the parabola
412 V. SOLUTION OF BOUNDARY PROBLEMS §95

(i.e., outside S), is seen to be ( — 2兀)，so that one has to take

IT= —兀(95.3)
2
in the formulae of § 91. The solution of the fundamental problems for S
presents no difficulties. As an example, the first fundamental problem
will now be solved (the second fundamental problem can be solved in an
analogous manner) .
Let a denote points of the real axis of the 已 plane. Then, by (92.10),
the boundary condition may be written

二—一
2G Q Q

+ + 2 b
丁
十 ztz
甲⑹ =N + iT, (95.4)

where N and T are the known boundary values of the normal and
tangential stresses 硝 and 笈. The condition (95.4) will now be multiplied
by o 十必 which gives
(O 卜一
一 2
(o + ia)^) + (a + 词丽 + ' '•
一
— (<y —血)空⑹
一 一

F, (95.5)
2
where
(N + 过) 0 + 町 (95.6)
It has, of course, been assumed here that N and T are given iri such a way
that they do not violate the conditions imposed in § 90 with regard to the behaviour .
of the stresses at points, away from the origin.
The method of solution, to be used below, may, of course, be applied directly to
(95.4) ; in the earlier editions of this book the problem under consideration was
solved in that manner. The present method, however, leads more quickly to the

_
solution.

The conjugate complex form of (95.5) is

+
. 值 ——— 加
A
2
'(<T)

(b 十 淘丽 F. (95.7)

The unknown functions (0, 甲(0 which are holomorphic in the lower
half-plane satisfy on the basis of (90.1) and of (95.1) the conditions

9=O (V) yg=0&)' ' *(0 =0(卷) (95.8)

Expressing that the function (q — %tz)¥(o), determined by (95.5),

CHAP. 16 SEMI-INFINITE REGIONS 413

is the boundary value of the function (已 一 holomorphic in the

lower half-plane and vanishing at infinity, one obtains, applying (76.21),
T-8 +8 +8
F da (c — ia) 十工赢J
1 ( (o — id) <I)— 十
E

，
b* (
—8
— 8

8 十
f8

(b — 打户
1
十
2ni 2(—0 J = 0,
where〈 is a point of the lower half -plane ; noting that (a
the boundary value of 一 血)
is
holomorphic in the lower half¬
—
plane and vanishing at infinity, and that (a 打) —
and (b 加产 ，⑻ —
—
are the boundary values of (C 血) (C) and (已 一闻2 holomorphic
in the upper half-plane and vanishing at infinity, one finds, applying
the formulae of § 72,
十8

(— ? (Q = 0,
— OO
whence
4-oo
1 r 府
峭=
2兀 — ia) J
— OO
a —已
(95.9)

The function (乙 血)¥(Q is now easily determined from its boundary

~

，
value given by (95.5). One thus obtains
+8
I F da + ia , 已 收 + 血)2
子 9 = 丽C
— J
M)
一 co
o— ?+ 2(^ — ia)
(95.10)

It is readily seen that the above solution satisfies the imposed con¬
ditions, if the given function 尸 and its first derivative F' with respect to
g satisfy the H condition and if this condition is satisfied near the point

at infinity by the functions bF and 炉尸'.

Hence the problem is solved.
 Chapter 17

SOME GENERAL METHODS OF SOLUTION OF BOUNDARY

VALUE PROBLEMS. GENERALIZATIONS *)

One of the general methods of solution of the fundamental boundary

value problems of the plane theory of elasticity for simply-connected
regions has been studied in §§ 78, 79. The present chapter gives a short
introduction to several other methods (also applicable to multiply
connected regions) which are either generalizations of the methods of
the earlier chapters of this Part or closely related to them.
Only one new method, due to D. I. Sherman (§§101, 102), for the solution
of the first and second fundamental problems will be studied in detail
and justified with complete proofs.
At the end of this chapter (§ 104), several other general problems of the
theory of elasticity will be formulated to which analogous methods of
solution may be applied.

§ 96. On the integral equations of S. G. Mikhlin. The method of

reduction of the fundamental problems to integral equations which was
studied in § 79 cannot be applied directly to multiply connected regions,
since it relies on the conformal transformation of the region under
consideration on to the circle and such a transformation (simple and
invertible) is impossible, if the given region is multiply connected.
However, S. G. Mikhlin succeeded in modifying the above-mentioned
method so that it becomes applicable also to multiply connected regions.
The essentials of this modification will now be .summarized. It is known
from the theory of functions of a complex variable that the problem
of conformal transformation of a region S, bounded by one simple
contour L, on to the circle is equivalent to the determination of the so-
called Green function for this region, i.e., of the real function G(居 y)
which is defined in the following manner :
1°. G(招 y) is a regular harmonic function throughout S, except at a
*) This chapter is not necessary for the understanding of the later work.
414
CHAP. 17 SOME GENERAL METHODS 415

given point (%。，％) where it has a logarithmic singularity. Thus

G(招 夕) = log— + G°(%, y),

r
where r is the distance between the points (%, y) and (%0, y0) and G°(%, y)
is a regular harmonic function.
2°. The boundary value of G(%, y) on L is zero.
If y) is the harmonic function, conjugate to GQ(x, y), then the
analytic function of the complex variable

z ——
= log —
Af - F G°(%, y) + 汨 y)f
— %。
where % = %()+ iyQ, is called the complex Green function. As shown
by the preceding formula, the function is multi-valued because of
the presence of the logarithmic term. Since the complex Green function
depends on z as well as on it is more logical to denote it by M (z, z°)
rather than by M(z).
The fact that the problem of determination of the Green function is
equivalent to the problem of conformal transformation of a given region
on to the circle permits modification of the method of §§78, 79 so that
the use of conformal mapping may be replaced by a study of the function
M(z, %).
On the other hand, the concept of the Green function, whether real
or complex, may also be applied to multiply-connected regions, bounded
by several contours. Hence the above-mentioned method may be gen¬
eralized to the case of multiply-connected regions.
In this way S. G. Mikhlin reduced the first and second fundamental
problems of the plane theory of elasticity for multiply connected regions
to Fredholm integral equations which are somewhat more complicated
(as was to be expected) than the equations of § 79 (which apply only to
simply connected regions) , but which are quite useful for general in¬
vestigations. In particular, they have been used in a number of papers
by S. G. Mikhlin [1—3, 7, 9] to prove the existence theorems. The
reader's attention is drawn to these papers and likewise to the book
[13] by the same author in which he gives a sufficiently complete study
of the results.
Apart from the first and second fundamental problems for multiply
connected regions, S. G. Mikhlin also solved by his method other boundary
416 V. SOLUTION OF BOUNDARY PROBLEMS §97

problems which are of great interest ; for example, the problem of

elastic equilibrium of a body, composed in a definite manner of different
homogeneous parts having different elastic constants (restricted, of
course, to those bodies to which the solutions of plane elasticity may
be applied) ; this problem is treated in S. G. Mikhlin [10] and several
particular cases are considered by elementary means in his paper [8].

§ 97. On a general method of solution of problems for multiply

connected regions. One general method of solution of boundary value
problems, developed by D. I. Sherman [1,5] and S. G. Mikhlin, deserves
special consideration ; this method permits the construction of the
Fredholm equation for a given multiply connected region, if by some
means the general solution of the corresponding problem has been
deduced for simply connected regions each of which is bounded by
one of the simple contours, constituting the boundary of the given
multiply connected region. For this purpose these general solutions must
be presented in a definite manner, eg in the form given by the solutions
of the integral equations, stated in § 79.
Particularly simple and practically useful equations are obtained
in the case where the above-mentioned , separate, simply connected
regions are mapped on to the circle by rational functions for which
the effective methods of solution, studied above, maybe applied. An
example of such a case is the half-plane with elliptic holes, considered by
D. I. Sherman [4].
In turning to the general case, it will be noted that the integral equations,
obtained in the manner stated above, have the following, practically
useful properties: If these equations are solved by the known algorithm
of successive approximation (i.e., by expanding the solutions in so-called
Neumann's series), then this algorithm coincides, in essence, with the
algorithm, generalizing the algorithm of Schwarz for the problem of
Dirichlet ; with regard to this generalized algorithm of Schwarz com¬
ments have already been made in § 89.
A study of this method will also be found in S. G. Mikhlin's book [13] ;
Mikhlin's investigation differs somewhat from that by D. I. Sherman,
since the latter starts from integral equations, obtained by the Author
(§79), while Mikhlin starts from his own equations mentioned in the
preceding section.
CHAP. 17 SOME GENERAL METHODS 417

§ 98. The integral equations, proposed by the Author. The

integral equations, deduced in § 79, are quite useful for general in¬
vestigations and give effective, practically applicable results in a number
of important particular cases, but they suffer from one essential dis¬
advantage; namely, the transforming function is required for their
construction. The same disadvantage attaches to S. G. Mikhlin's equations
(§ 96), since the complex Green function M(z, z°) has to be determined.
Integral equations have long ago ceased to be useful only for general
theoretical investigations ; lately, rather effective methods have been
developed for their numerical solution, in particular, in those cases,
where they involve only simple (and not multiple) integrals, as is the
case with those which are of interest here.
It is therefore very desirable for direct practical applications to have
integral equations whose kernels are related directly and simply to the line
elements constituting the boundary of the region and which do not involve
elements the determination of which requires preliminary solution of
auxiliary boundary problems as Dirichlet's problem (or its equivalent)
for the determination of the functions co(Q or M{z, 窗).
Equations of this type, used by G. Lauricella—D. I. Sherman, will
be considered in detail in §§ 101 , 102; these equations are, in the Author's
opinion, the simplest and most suitable for the purpose of general in¬
vestigations.
However, the Author proposes to devote some space here to equations,
obtained in his papers [17, 18], since the trend of thought, leading to
these equations, is closely related to that which led to the results of the
preceding chapters of this Part, and since they are of interest in them¬
selves.
These equations are very similar to those of G. Lauricella (cf. § 101), but never¬
theless they differ significantly from Lauricella's. In the Author's opinion Lauri-
cella's equations had (at least outwardly) a rather complicated form, so that at
the time he did not notice the similarity and supposed that his equations were
considerably simpler.
Moreover, the Author's equations were the subject of a number of
investigations by other authors (in the first place D. I. Sherman) worthy
of reference, since the methods of investigation, developed by them, may
be successfully applied to the solution of analogous problems.
For the sake of clarity, a beginning will be made with the case of
finite regions S, bounded by one simple smooth contour L ; the positive
direction on L will again be such that it leaves S on the left.
418 V. SOLUTION OF BOUNDARY PROBLEMS §98

The first and second fundamental problems will be considered simul¬

taneously. The boundary conditions for these problems may be written as
硒力 + + = 而， (98.1)
where, in the notation of § 41, for the first fundamental problem 也 = 1,
s

力) = /1(0 + 分2 = zj (Xn + iY^ds + C, (98.2)

0

-
while for the second fundamental problem k = — x,
, 超) —
2Mgi + *2) ； (98.3)
勺 (£), 中'(力)， 少份 are, of course, the corresponding boundary values the
existence of which is thus assumed, i.e., the unknown solution is assumed
to be regular. The arbitrary constant on the right-hand side of (98.2)
may be fixed to suit convenience. .
Conditions will now be stated such that the right-hand side of the
equation
W) = 而— 频 — 如的， (98.4)
equivalent to (98.1), must be the boundary value of some function
少⑵，holomorphic in S. It is known from §73 that a necessary and
sufficient condition for this to be so is given by

for all z outside S, or

上［ 双」
2kV J t — z 2那J
j 里处 = 4⑶
z 力一
(98.5)
L L

for all z outside S, where

A
2H2 J 4^.
--
t z
(98.6)
L

In this way the functional equation has been deduced for the deter¬
mination of 勺 (z). Once one has succeeded in finding by some means the
CHAP. 17 SOME GENERAL METHODS 419

function 中(z), holomorphic in 5 and satisfying (98.5), the problem

will be solved, since %) can be determined from (98.4) by Cauchy's
formula

W(0 =
2兀〃 t
i

—z
_ i
2兀〃 t
r
—z
k ，— 华g就i
2Tti J t z
国② 就
2ni J t ~ z
(98.7)1
L L L L

(where, of course, z lies in S).

The functional equation (98.5) may be readily reduced to a Fredholm
equation in the following manner. [It would be of interest to study (98.5)
independently without reduction to a Fredholm equation. In all pro¬
bability this would offer the opportunity of finding new classes of regions
for which the fundamental problems may be solved effectively.] * In (98.5),
let z tend to some point tQ of L (remaining, of course, outside S). Then,
on the basis of the Plemelj formulae (cf. § 68), one obtains, assuming
平 伍)，q/0) and 伍) to satisfy on L the H condition,

1
—— f7
k
iaF
昴叭幻 i
t ―一 厂
一 赢中'（幻 H
4(幻'
27cz j 力 27c2 J t %
L °
where is the boundary value of 4(z) as Zo from outside S, i.e.,
4 (zo) —
— 豺Go) + 以幻 ib(t^; (98.8)

口(幻，.方 (%o) denote here real functions which will be assumed known.
Equation (a) which is obviously not a Fredholm equation may be
simplified as follows. Expressing that 中(9 and 乎'(%) must be the boundary
values of functions, holomorphic in S, one finds by (73.1')
w o
平

1_2
-
1 / g，(/成
— 树'(%) 十
2tw J t
L
% —
= 0;

the first of these conditions becomes in its conjugate complex form

* The truth of this statement in the preceding editions has been confirmed in
a known measure by interesting results obtained recently by D. I. Sherman
[25, 26] which have already been referred to in § 89.
420 V. SOLUTION OF BOUNDARY PROBLEMS §98

.
— 和(力0)
. . 1 ; 中传)近
7t T = 0，
广J
L
—
Multiplying
one obtains
and (c) by — 曲 and k respectively and adding them to
—— —
一 力中 (M)
为
丁^
27rz J
r
— log —
t
—
t
t
二
% w下
1
W)力
一％
就 = 4(%)，
z,

and finally, integrating the second integral on the left-hand side by

parts,

— g(%) log-—A-— /
t 玲 27cz J
— 中(/川7t i
挤=
Zq
4(幻.(98.9)

This is the integral equation which was mentioned earlier and which
was to be deduced.
It may still be written in a different way. In fact, if
t一 = 夕鸿 (98.10)
where = [力 一 Zo [ and 9 = 眇(％, 力) is the angle between the vector
夕

and the Ox axis measured in the positive direction, one has

log ，
t A% — 2~ 诵\ 厂2毋 = cos 29 — i sin 2 ;

hence (98.9) becomes

刎册) ~ ~
亦 J
{加(9 + =—A (989)
工
By writing
中 (9 = 力R) + 勾 (98.11)
where 力Q) and 夕修) are real functions, and by separating real and imagi¬
nary parts, (989) may be represented in the form of the two real equations

Q(%) SG) (々 cos 29-) + g(f) sin 2A}d8 = — 口隔)，

L
(98.9〃)
1 /
放 / {/>(/) sin 28 + 0(力)但 一- cos 28)}&分 — 旗％).
L

In these equations "分 ds,

-^―
Vv
CHAP. 17 SOME GENERAL METHODS 421

where s is the arc coordinate of the contour, corresponding to the point t.

It is easily seen that
盼 COS a
Hs r '

where a = a(Z0, t) is the angle between the outward normal at t and the
vector 州.
In order to verify the last relation, it is sufficient to remember that by the Cauchy
— Riemann equations
5 -9- a log t 1
8s r dn

——
because log r and 9 are the real and imaginary parts of the function log (£ ％) of
the complex variable t (for fixed %); n denotes here the normal which points to the
right as one moves in the positive direction of the tangent.

If it is assumed that the angle between the normal (or tangent) to L

at the point t and some fixed direction (considering this angle as function
of 力 or s) satisfies the H condition, then
cos a K(b 力)
，
7 洛

where pi is a constant such that 0 p. < 1 and K(%, £) is a function

continuous on L (which even satisfies the H condition) . (Cf., for example,
the Author's book [25]).
Thus the system (98.9") represents an ordinary system of Fredholm
equations. Correspondingly the equations (98.9) or (989), equivalent
to (98.9"), may be called Fredholm equations.
As the study of these integral equations in the case of simply connected
regions does not present any difficulties (cf. S. G. Mikhlin [13] for
the case 力 = 1 ) , the following results will be merely enunciated here and
the reader will later be given references, where the corresponding proofs
may be found. First of all, it will be noted that, as is easily seen, every
(continuous) solution 中(£) of (98.9') will satisfy the H condition every¬
where on L, on the basis of the conditions assumed above. But, in ad¬
dition, the solution was to be such that the derivative 平'(力) also satisfies
this condition on L, because this had been assumed in the deduction of
the equation. It is readily verified that fulfilment of the last condition is
ensured, if it is assumed that the curvature of the line L satisfies at every
422 V. SOLUTION OF BOUNDARY PROBLEMS §98

point the H condition and that the function /(/), given on L, has a
derivative with respect to % which satisfies the H condition.
Consider the first jundamental problem. In this case k = 1 and /(/)
is given by (98.2). Since, by supposition, the function /(/) is continuous
on L, the condition that the resultant vector of the external forces
is to vanish will be automatically satisfied ; however, the condition
of vanishing of the resultant moment is expressed by (cf. § 41)

y
L
(f^dx %"夕) — 0- (98.12)

It is easily verified that the homogeneous system, obtained from

(98.。〃) for 口(/) = 如) = 0, has the solution
》 (力) + 勾@ = izt 十 a 十制 (98.13)
where % a, g are real constants ; this follows from the fact that (cf. § 34)
the state of stress and the constant C will not be changed by adding to
甲(z) an expression of the form 企z + a 十 On the other hand, it
may also be verified directly that 力0),q(£), as given by (98.13), satisfy
(98.9〃).
The formula (98.13) involves linearly three arbitrary real constants
and it gives three linearly independent solutions of the homogeneous
system.

Writing 1 — + one may take as these three solutions

1) 夕 = — 笃， 4 = E， 2) 夕 = 1, 夕 = 0, 3) 夕 = 0, g= L

It may be shown that the homogeneous system has no other linearly

independent solutions. Hence, by the general theory of Fredholm
equations, the system (98.9〃) will only have solutions, if the right-hand
sides of these equations satisfy three conditions of a well known form.
However, a closer study shows that two of these conditions are auto¬
matically satisfied as a consequence of the fact that 口(力)，〃力) are not
arbitrary, but such that 〃(£) + 冽力) is the boundary value of a function,
holomorphic outside S and vanishing at infinity; the third condition,
as was to be expected, reduces to the condition (98.12).
Thus, if (98.12) is satisfied, the system (98.9〃) or, what is the same
thing, the equation (989) has a solution which is determined apart from
an expression of the form (98.13). In addition, it may be shown (and this
CHAP. 17 SOME GENERAL METHODS 423

is not obvious beforehand) that every solution 中(/) of (98.9') will be the
boundary value of a function, holomorphic in S; this function <p(z)
follows from 中(/) by the help of Cauchy's formula, and 小⑺ will then be
determined by (98.7) .The solution of the first fundamental problem
has thus been obtained.
In the case of the second fundamental problem, where k = 乂 and
/Q) is given by (98.3), quite analogous results may be found ; the only
——
difference is that the homogeneous system, corresponding to (989'),
has now only the two linearly independent solutions
= 夕(方) + 勾(£) = a + 小, (98.14)
where a and B are arbitrary real constants ; the system (98.9〃)，in spite
of the presence of the solutions of the corresponding homogeneous
system, is always soluble (as a consequence of the particular form of
the right-hand sides) and its solution gives the solution of the original
problem, as in the case of the first fundamental problem.
Hitherto, it has been assumed that the region S is finite and simply
connected. Suppose now that S is bounded by several simple contours
Llt 上2，.…，上彷，上馆+i the last of which contains all the preceding ones, as
in § 35 (cf. Fig. 14) ; the contour Lm+1 may be absent in which case S will
be infinite (i.e., the infinite plane with holes). It will be assumed that
the individual contours satisfy in a certain way the conditions of
smoothness, stated above. As always, let 二 = 工】+ L? + + 工馆 +.…

十 Lm+1 denote the complete boundary of S ; the positive direction of L will

be chosen in such a way that it leaves S on the left.
The only difference from the case of finite, simply connected regions
is that here the unknown functions 9(2) and 少卜) may be (and, in general,
will be) multi-valued. In fact, by (35.11) and (35.12),
1 m
平 (z) = 2k(1 + x)
s区
=1
工) log (z — z?) + *z),
(98.15)
m
M%) =
比

27r (1 + X)
£ (X,・ 一 2•匕) log (z — + WoC),
where (X, ，Yj) are the resultant vectors of the external forces, applied
to the contours Ljt are arbitrarily fixed points inside Lj (/ = 1 , . . . , w)
and (p0(^), Wo(z) are functions, holomorphic in S, if this region is finite
(i.e., if Em+i is present) ; if S is infinite (i.e., if 工切+iis absent), then (cf. § 36)
= Fz 十 少o(z) = F'z + /(z), (98.16)
424 V. SOLUTION OF BOUNDARY PROBLEMS §98

where 少*(z) are functions, holomorphic in S including the point

at infinity ; in the case of the first as well as of the second fundamental
problem it will be assumed (§ 40) that the constants T, T' are known
beforehand ; further, in the case of the second fundamental problem for
infinite regions, it will also be assumed that the quantities
m m
X = E Xj, y = E 匕,
7=1 ?=1

i.e., the components of the resultant vector of the external forces, applied
to the entire boundary L of S, are known.
For the sake of brevity, it will now be assumed that S is finite, i.e.,
that Lm+1 is present ; the case of infinite regions may be considered in
quite an analogous manner.
A beginning will be made with the first fundamental problem. In this
case the boundary condition may be written (as in the case of simply
connected regions)
甲(9 + S(£) + WQ) = /(/) + Cj on Lj (/ = 1, 2, …，碑，w + 1), (98.⑺
where now, instead of (98.2),
s

/(%) = /
?

0
(X" + + Cj on Lj (，= 1,2, . . w + 1), (98.18)

while the arc coordinate s is measured (in the positive direction) on

each of the contours from an arbitrarily fixed point of that contour,
and Cj is a constant having, in general, different values on different Lj；
these constants are not known beforehand, except for one, say, Cm+1
which may be fixed arbitrarily, and it will be assumed here that Cm+1 = 0.
Substituting from (98.15) into (98.17), one finds
(£) +& on % (，= 1,2, … ，碑，加 + 1), (98.19)
°
勺

where
1 m
/o(£) — /0) + 2°兀(1 十:—x)
~ E {Xj + iY {log (/ - zj
?・=i
— — x log (于 一 5夕)}+
t 馆区 — e匕
•
， 、
+ '7 /—-
(98.20)
2
…'

上 \
E ~ ~~ •

On the left-hand side of the boundary condition (98. 1 9) one has the
CHAP. 17 SOME GENERAL METHODS 425

boundary values of holomorphic (i.e., single-valued, analytic) functions ;

the right-hand side is likewise a single-valued function (assuming, of
course, the choice of definite branches on each contour E，，because for
a circuit in the positive direction (leaving S on the left) of the contour
0' = 1, 2, . . . , w) the function /(/) undergoes an increase i(Xj + iY^,
while the second term on the right-hand side of (98.20) shows the same
increase, but with opposite sign; similarly for Lm+1> under the condition
(which is implied) that the resultant vector (X, Y) of all external forces,
acting on L, is equal to zero.
Since in the case of the first fundamental problem the quantities
Xjt 匕 are known beforehand, the function /0(/) in (98.19) is determined
on every L$.
Applying to (98. 19) the same reasoning as in the case of a single
contour, one arrives at exactly the same equation (98.9) with = 1 or
at the equivalent equation (98.9') ; the only difference will be that cp0(/)
takes now the place of 甲 修)， while /(力) is replaced by /。(力). In addition,
the right-hand side now involves the initially unknown constants
C2, . . . , Cm which must be determined in the process of solving
the problem.
In the case of the second fundamental problem, proceeding in an ana¬
logous manner, one finds, in the former notation, the boundary con¬
dition
— z9o(^) + E(') + %(£) — < (98.21)
where this time

/oQ) = 一 2凶®+^g2) — ~7；―.— f 火 (X；+ ，匕) log

n (1 十%) 夕
~

" — z八 +

/ X*
m iY上
.
*

+ 2兀(1 十 K) 7 =1 t —
~~7
Zj
(98.22)

—
Thus one obtains the same equation (98.9) for 々 = x as in the case
of a single contour, if one replaces 中(力) by 勺°(力) and 力) by /()(/). In the
present case, the unknown constants Xjt 匕 appear on the right-hand side
and they must be determined at the same time as the function g").
It may be shown that in the case of multiply connected regions the
derived integral equations allow complete solution of the corresponding
boundary problems.
426 V. SOLUTION OF BOUNDARY PROBLEMS §98

A preliminary study of these integral equations was given in the

Author's notes [17, 18], quoted earlier, in which, for the sake of definite¬
ness, the first fundamental problem was considered ; this study was
based on the supposition that the existence theorems for multiply
connected regions had already been proved by some other method.
Soon after, D. I. Sherman [2, 3, 6, 1 1] gave a very complete study of
these equations in which he did not rely on other proofs of the existence
theorems, but, on the contrary, proved these theorems directly by means
of the equations under consideration.
D. I. Sherman also modified these equations so as to give them a form
more convenient for studies of a general character and for applications.
In particular, in his paper [11], he studied in detail the question of the
distribution of the eigenvalues of the integral equations, obtained by
a definite modification of the above equations, introducing some para¬
meter 入， as is done in the general Fredholm theory. This investigation
showed that for values of 入， corresponding to the first and second fun¬
damental problems, the solutions of the relevant integral equations may
be expanded in Neumann's series which, in general, can be obtained
by the method of successive approximation.
By means of the method of this section D. I. Sherman [6] also solved
one particular case of the mixed fundamental problem when the external
stress is given on one of the contours, bounding the region, while the
displacements are given on the others.
Further, D. I. Sherman [8] solved by a method, analogous to the
preceding one, the first and second fundamental problems for bodies,
consisting of different homogeneous parts ; as indicated in § 96, the same
problem was solved somewhat earlier by S. G. Mikhlin by use of another
method.
In later papers D. I. Sherman gave new solutions of the above as
well as of some other boundary problems by means of a method which
is a generalization of that of the present section ; this work will be dis¬
cussed below.
Finally, one more interesting problem will be mentioned which was
solved by G. N. Savin [7] * by a method, analogous to that above ; this
problem deals with the equilibrium of an elastic plane with an infinite
number of identical, equally spaced holes which are subject to the same
external forces. A study of this solution may likewise be found in S. G.
Mikhlin [13].
* Cf. also G. N. Savin [8].
CHAP. 17 SOME GENERAL METHODS 427

§ 99. Application to contours with corners. The form of the

equation (98.9') or of the system (98.9") suggests that, if the integrals
occurring there be considered as Stieltjes integrals, these equations
may be applied to regions, bounded by contours of a much more general
form than has been assumed for their deduction.
—
The investigations of L. G. Magnaradze [1- 3], based on known
results by J. Radon and partly by T. Carleman, show that this is actually
the case and that the above-mentioned equations, interpreted in a
suitable generalized sense, apply, for example, to the case where the
contours, bounding the region, have corners other than cusps. There
may even be infinitely many such corners ; it is sufficient if the boundary
of the region consists of contours having so-called "bounded rotation”
(according to J. Radon).
It may be noted that L. G. Magnaradze [4] succeeded in extending
these results also to one very general class of three-dimensional bodies
the surfaces of which may have polygonal edges (even an infinite, but
denumerable number of them) ; in this case one has, of course, to apply
the corresponding integral equations for three-dimensional bodies.

§ 100. On the numerical solution of the integral equations

of the plane theory of elasticity. Equation (989) or the equivalent
system (98.9") may, thanks to their simplicity, be used successfully for the
numerical solution of the corresponding boundary problems of the plane
theory of elasticity. One of the methods of numerical solution is outlined in
the Author's note [21] and it was studied in greater detail by A. Ya.
Gorgidze and A. K. Rukhadze [1] who applied this method to several ex¬
amples and gave also estimates of accuracy.
This method seems to give satisfactory results also in cases when the
boundaries have corners.

§ 101. The integral equations of D. I. Sherman -G. Lauricella.

—
Recently D. I. Sherman [15 17] succeeded in deducing integral
equations for the solution of the first and second as well as of the mixed
fundamental boundary value problems of the plane theory of elasticity
which deserve greater attention. Apparently the most natural way of
arriving at these equations is the following which is based on one simple
general idea, analogous to that used by I. Fredholm [1] for the de¬
duction of the integral equations for the second fundamental problem
in the three-dimensional case.
428 V. SOLUTION OF BOUNDARY PROBLEMS

D. I. Sherman begins directly from the formulae (101.3) and (101.4) without
indicating the means by which these were obtained (and he considers separately
the cases k —— x and k = 1).
I. Fredholm's idea consists, in principle, of the following. If one takes instead of
the body under consideration the half-space and writes down the known formulae
solving the corresponding boundary problem in closed form, using definite integrals
over the plane boundary of the half-space, then these formulae, when applied to the
given body (taking now the integrals over the surface of that body instead of over
the plane), do not, of course, solve the boundary problem in closed form; they lead,
however, to integral equations which under certain condition^ will be Fredholm
equations.

At first suppose that the region S under consideration is finite and

bounded by one simple contour which satisfies the same conditions as
in § 98. The boundary conditions of the first and second fundamental
problems will now be written, using the notation of § 98,

+ 师， 十 山(%) = f ②,
海 0) (101.1)
remembering that k = 1 and k — — x for the first and second problem
respectively.
Under the supposition that S is the upper half-plane, L is the real axis
and <p(z), 少⑵，z(p'(z) vanish at infinity, the solution of the boundary
problem (101.1) is given by

L L

The solution for 克 = — x can be found in § 94, where in the present case G = 0
and the difference in sign arises from the fact that the problem, has been solved
there for the lower half-plane. The solution for A = 1 is obtained in an analogous
manner. It may also be deduced from the solution of the first fundamental problem
for the half-plane, found in § 93.

Substituting in the second formula above for and introducing the

notation
1
3(%)=*附, (101.2)
one finds
1 / 3(9凝 (101.3)
L
CHAP. 17 SOME GENERAL METHODS 429

n o 11 •4

In the case when L is the real axis (as it has been assumed for the present) ,
t = 力 dt — dt', the reason why dt and t have been written instead of dt and t in
the second and third integrals of (101.4) respectively will become clear later on.

Integrating the last integral of (101.4) by parts, this formula may be

rewritten
W（2） (1014)

Now the case when S is not the half-plane will be considered and
an attempt will be made to find the solution of the boundary problem
(101.1) in the form (101.3), (101.4), where sQ) now denotes some function
o/ points of the contour L which initially unknown and has to be deter¬
mined. It will be assumed that coQ) has a derivative 3'«) which satisfies
the H condition. This is easily seen to ensure the continuity of the
functions 平团，甲'(/), WR) up to the boundary, i.e., the regularity of the
solution (in the sense of § 42).
Using the Plemelj formulae for boundary values of Cauchy integrals
and substituting into (101.1), the boundary values of the functions
cp 少 determined by (101.3), (1014), and likewise the function
1/ 3(力成 1 / s' 出dt

"
平 = 元/ 7^^ =
L E
KT
[where the latter expression is obtained by an integration by parts] ,
one finds the integral equation

+ 亏丁J/ -―r/ -*
3«)〃^力 一 10 = /Ro).
力3仇) log ~ (101.5)
t—M 2兀之J
l l

This is the integral equation, obtained by D. I. Sherman in the quoted

papers [15, 16]. It is seen to be very similar to the equation (98.9)
which, for the purpose of comparison, will now be written in its conjugate
complex form
丽(％) — ~

2兀2 J
L
中 (QdlogTt
—
~~
£ J
L
t
—
~~
^=—4(%).
tQ

However, it differs essentially from (101.5) by the sign of the first in-
430 V. SOLUTION OF BOUNDARY PROBLEMS § 101

tegral, by its right-hand side and, what is more important, by the character
of the conditions, imposed on the unknown functions. In fact, the un¬
known function of (101.5) is subject to no other condition except one
referring to its continuity, while the unknown function 申(/) of equation
(霖) must be the boundary value of a function, holomorphic in S. This
last condition, as already stated in § 98, is automatically satisfied in the
case of finite simply-connected regions which will now be considered ;
but in the general case it plays an essential part.
Equation (101.5) will now be considered. As in § 98, let t — 曲 = 川*;
(101.5) then becomes

力3（幻
+ (1015)

Further, writing
3(力) = 力 + 勾Q), ZW = /1(%) + (101.6)

；
one obtains the system of two Fredholm equations

面坎幻
/
+ {0(力) 但 一 cos 29) — ' sin 2a}a分 /式片)，

加(M)
J.
—— / {0(9 sin 23 — g(6)(九 cos 2分)}a8= /2W-
(101.5〃)
亦 J
The following should be noted with regard to the system (10L5").
For
— 1, i.e., for the first fundamental problem, this system reduces
to that, deduced by G. Lauricella [3[ for the solution of the fundamental
biharmonic problem which, as has been pointed out earlier, is equivalent
(with certain reservations in the case of multiply connected regions)
to the first fundamental problem of plane elasticity. For k — 乂,
i.e., for the second fundamental problem, the system (1015') corresponds
—
to the system, likewise deduced by G. Lauricella [1, 2] for the second
fundamental problem in the three-dimensional case.
However, Lauricella does not use Cauchy integrals and he presents the
connection between the functions, which directly occur in the correspond¬
ing problems (the biharmonic function U in the fundamental biharmonic
problem, the displacement components in the second fundamental
problem), and the auxiliary functions 力，夕 of points of the boundary E
in a (apparently) very complicated form. Lauricella^ integral equations
CHAP. 17 SOME GENERAL METHODS 431

likewise are not as simple as (101.5〃). This last circumstance is, of course,
of no significant importance, but the formulae (101.3) and (101.4),
expressing the relations between the functions <p(z), 少 (z) and 3«) =/>(/) +
匆(£), are of great value and so is the form (101.5) of the integral
equation which clearly demonstrates the connection with Cauchy integrals.
In fact, the discovery of this relationship considerably simplifies the
analysis, in particular in the case of multiply connected regions (with
regard to which more will be said later) , and, in addition, offers the
opportunity of deducing (relatively) simple solutions of a number of
other important boundary value problems. Therefore it seems only
just to call (101.5) or (1015) the equations of J). L Sherman—G. Lau-
ricella.
In the case of multiply connected regions, it is advisable, according
to D. I. Sherman, to somewhat modify (101.3), (101.4) and the integral
equations which follow from them, thus leading to (relatively) very
simple results ; this question will be treated in greater detail in the next
section.

§ 102. Solution of the first and second fundamental problems

by the method of D. I. Sherman x). Let the region S be bounded
by several, simple, non-intersecting contours 工卜 L2, . . . , Lmt Lm+1, the
last of which contains all the others, and let L = Lj + L2 + ... 二俏+i
denote the complete boundary of S. In addition, it will be assumed that
each of the contours L has a curvature, satisfying the H condition. The
finite regions, bounded by the L, (/ = 1, 2, . . 加)，will be denoted by
Sj, and the infinite region, bounded by Lm+1, by S +「 讥

TAe first fundamental problem will be solved first. Without affecting

generality, it may be assumed that the resultant vectors Y\) of the
external forces applied to the contours Lk (/ = i , 2, . . . , 加) are equal
to zero and that therefore the unknown functions 巾(z) are single¬
valued, since otherwise one can take away from them multi-valued
terms (which are known beforehand) and transfer them to the right
hand side of the equality expressing the boundary condition 2) which
then leads back to the preceding case. In addition, it must obviously
be assumed in order to make the problem possible that also the resultant
vector of the forces applied to the contour Lm+1 vanishes.
i) D. I. Sherman [15, 16]. These papers have been reproduced here with only
insignificant modifications.
2) Cf. (98.15), (98.19), (98.20).
432 V. SOLUTION OF BOUNDARY PROBLEMS § 102

The boundary condition is

@(£) + 力％'(%) + WR) = /(%) + Cj on E" = 1, 2, …，加十 1), (102.1)

where /(/) is a given function (which is single-valued and continuous

on every Lj and Cv C2, . . Cm, Cm+1 are initially unknown constants
only one of which may be fixed arbitrarily ; it will be assumed that
Cm+l — °，
Following D. I. Sherman, the solution will be written in the form
1 /* eoQ) 弟
1
m bj
中⑶ = V7 力 + y i 丁二l，
/ — —
£ (102・2)
2兀2 J Z Z — Zj
L

， (102.3)
2tUZ J/ \t
yW ~
2兀2 J t
kk 丁 /
27cz J (力
t,
Z)%
2
Zj
Z Z ?，= 1 Z
L L L

where 3(£) is a function of points of L, subject to definition, z夕 are ar¬

bitrarily fixed points of S), j = 1, . …，加 (so that they lie outside S)
and bj are real constants, related to sQ) in the following manner:
bj = i — 30)也｝, j = 1,2, . . .,m. (102.4)
%
The introduction of % leads to the modification of the integral equations which
was mentioned at the end of § 101. In the case of simply connected regions (m = 0) ,
the formulae (102.2), (102.3) become (101.3) and (101.4) for 左 = 1.
Only regular solutions of the initial problem will be sought (in the
sense of § 42). For this purpose it is sufficient to assume that the unknown
function co(/) has a derivative 3‘(@ satisfying the H condition.
Substituting in (102.1) the boundary values of the functions 平 (z),
9'^), W(z) as determined by (102.2) and (102.3), one obtains as in § 101

1 7一 方一片
yy J 3(^7^
t 给
+
L

-
— ~工一
— Ck = /(%) on Lk,
々= 1,2, ...,m
(102.5) 1.
It will be expedient to further modify this equation by adding to the
CHAP. 17 SOME GENERAL METHODS 433

left-hand side the term

where bm+1 is a purely imaginary constant, related to 3(%) by the formula

q｛
[cf. note following (102.4)]
k+i = /平
2兀2 JI*
L
dt +当
尸 成｝；
J
(102.6)

it will be assumed that the origin of coordinates lies in S.

Thus one obtains the equation
1「 t tn 1 C t %
3(幻 'h
27cz J
/ 3(/) log^t 27c2 J t +
L L

+《1 1 %
)= 1 — 盯
十
环一 、
工Ll (1 — Uj)｝
J
- 金 = 九) on
% — zi /
Lk>

where zm+1 = 0.
— 1, 2, .. w + 1, (1025)

In addition, the unknown constants Ck will be related to the unknown

C丸 =
_
function 3(/) by the formulae
. 3«)d$, 左 = 1,2, . . . , m, (102.7)
Lk
where ds is the element of arc of Lk.
If one now replaces the constants &, C?・ on the left-hand side of (102.5')
by the expressions (102.4), (102.6) and (102.7), then ( 102.5') becomes an
integral equation which involves no othey unknowns but 3Q). Separating
real and imaginary parts, as was done in § 101, one obtains a system of
two Fredholm equations, but since this system serves no purpose in
what follows it will not be written down here.
The integral equation (102.5') will be called the equation of D. I. Sher¬
man. In the case of simply connected regions, it differs from (101.5)
only by the term
员十也
片
(1一2).
/ 力o 、 M
It will be assumed below that the function 附 given on L has a de¬
rivative /'(/) satisfying the H condition. It is not difficult to show that
434 V. SOLUTION OF BOUNDARY PROBLEMS § 102

under this condition and under the conditions assumed for the boundary
L every (continuous) solution 3修) of the equation (1025) has a de¬
rivative 3‘(f) satisfying the H condition.
It will now be shown that, if (1025) has a solution, then necessarily
6 +1 = 0, provided the resultant moment of the external forces is zero;
淑

the vanishing of the resultant vector of the external forces is ensured by

the single-valuedness and continuity of the function /«). The condition
for the resultant moment may obviously be written [cf. § 41]

猊 / 工 0, (102.8)

while (102.1) may be presented in the form

+ 碎十丽 + bm+1〔l，一
力 』t
十
* J — g = /(/) on % (102 1 ')

if one interprets 平㈤，勺'(力) and 少心) as the boundary values of the expres¬
sions (102.2) and (102.3). Multiplying both sides of (102.1') by and
integrating over L, one finds after an integration by parts

L
/ {(pQ) 应 一 中传)位} 6 仙+1/ I—
L
- H— + 2 仍加+i =
h
Im
L
册.

Since the last term on the left-hand side of this equation is real and all
the other terms are purely imaginary, one must have bm+1 0, as was
to be proved.
—
Thus, in order to satisfy (102.8), every solution s(£) of (1025) is at
the same time a solution of the original equation (102.5), and hence
a solution of the boundary problem (102.1), and the constants Cj will
be determined by the formulae (102.7).
It will now be proved that the equation (1025) always has a solution.
[Equation (102.5) would only be soluble under the condition (102.8)].
For this purpose the homogeneous equation, obtained from (1025)
for f(t) = 0, will be considered and it will be shown that it has no non-zero
solutions. Let 3()Q) be any solution of this equation and 例⑶，少o(z),
C? be the corresponding values of <p(z), 少⑵ and the constants g, de¬
termined by (102.2), (102.3), (102.4) and (102.7) for 30) = 3o0)；in
particular, by (102.2) and (102.3),
CHAP. 17 SOME GENERAL METHODS 435

阴
1 / 5(9弟 加.

(102.9)

，
j= 1 z— 书

1 用⑴近 1 砺(力)应 设 学
- —z (102.10)
%(力
—
[ /」 一― ，•

—
।। * '，《■■'

2兀£J
13-"— ' ' * 1
“

2tcE J t z t /=i z ~~ z j
L L

where 殍 are the constants, given by (102.4) for s© — 如㈤，and the

expression for %(z) has been transformed by means of an integration
by parts. The functions ^Q{z), 弘⑶ satisfy the boundary condition
平 o(力) +瓯 少0«)— 今 0 on Ljf j = 1, . . ,,^+ 1,C^ + 1 =0, (102.11)
as may be seen from (102. 1 taking into consideration that in the present
— —
case /(£) 0 and also bm+i 0, since obviously (102.8) will be satisfied.
Hence 仇⑵，M(z) solve the first fundamental problem in the absence of
external forces, and therefore, by the uniqueness theorem,
= izz + c, (102.12)
中o(z)

where e is a real and c a complex constant ; thus, by (102,1 1), using the
fact that C*+i = 0,
Wo(z) = — (102.13)
and, obviously,
Cj =0, 7* = 1 , 2, . …，磔 + 1. (102.14)
It follows from (102.9), (102.10), (102.12) and (102.13) that

zez + c = 2 —2 J/ — —t — FW
—
, (1029)

，—
兀 Z /= 1 z子
L

-
C
1
—二 ]
1 题 近 誉
+ E— 6： . (102.10')
J
U
一」

2兀2 J t z 27cz t 2 z Zj
工 L

Introduce now the notation

£
'4*(*) = 30(%) 十f 工1 一•马 — — 力 (102.15)

一，
俏
68 ¬

辿*(') 3()(%) %()(/) + j Z F (102.16)

= l t — Zj
The equations (102.9% (102.10') may then be written
436 V. SOLUTION OF BOUNDARY PROBLEMS §102

1 / gW w*(力成
2戒 J t z
— — c
' t
- z — 0 for all z in S,
and hence (cf. § 74) 平*(力)，W*(力 are the boundary values of the functions
holomorphic in the regions Slf S2, ...,S +卜 and 惟

<p*(oo) = (p*(oo) = 0.
It will now be recalled that in the present case bm+1 = 0, where bm+1
is given by (102.6) for 3 = 川。©. Substituting in (102.6) with bm+1 = 0
for s(力) the expression (oo(^> obtained from (102.15), and taking into
consideration the previously stated property of 乎*(%), it is easily seen that
£ = 0.
Further, eliminating 询心) from (102.15) and (102.16), one finds
m ( 1 1 t 1
如* 储 少*(力) Z 呼 一h
2・=1 IE
--
2才 t — zi
1 7
(力 一 j — 22c
' '

on L.
Multiplying both sides of this equation by dt and integrating over the
contours 工式々 = 1,2, . . m), one obtains

and hence, since the b? are real,

破 = 0, k = 1,2, . . . , w. (102.17)
Therefore
g*(e) 十 如*'(/) + = — 2诟 on Lk, A = 1, 2, …，加+ 1.
Consequently 中*(z) solve the first fundamental problem for the
regions 5把， 也 = 1, 2, …，加 + 1, in the absence of external forces. Ap¬
plying the uniqueness theorem to the region Sm+1 and using the con¬
dition 中*(8) = = 0, one finds <p*(z) = W*(z) = 0 in Sm+1, and
hence c = 0. Further, the uniqueness theorem applied to the regions
Sk(k = 1, . . . , w) gives (remembering that c = 0)
= 於戒 + ck, 中*(或 = — ck on Lk, 左 = 1 , 2, . . . , 加，
whence, by (102.15)
— (102.17),
woW = — + ick on Lk, 为 =1,2, 加;
CHAP. 17 SOME GENERAL METHODS 437

。
in addition, since 中*(z) = tp*(z) = 0 in Sm+1,
3()(%) ~ 0n 二
Finally, using successively the equations (102.4), (102.17), (102.7) and
(102.14), it is easily verified that 年 = c 兀 = 0 for all 为，and hence 3°«) = 0
everywhere on L.
Thus the homogeneous equation, corresponding to (102.5'), has no
solutions, different from zero. Consequently the equation (102.5') has
one and only one solution 3(%). Substituting this solution 3® in (102.2)
and (102.3), one obtains the solution of the original problem, provided
the condition (102.8), expressing zero resultant moment of the external
forces, is satisfied (the vanishing of the resultant vector of the external
forces being ensured by the continuity of /修) on L). Thus the problem is
solved.

If (102.8) is not satisfied, 4P (z) and 少⑶ will not satisfy the boundary condition
(102.1), since in that case bm+1 声 0 and the solution of (102.5') will not be a solution
of (102.5).

Next consider the second fundamental problem. In this case the boundary
condition may be written

— ^'(0 — WQ) = g(9. (102.18)

where, as before,
gR) = 2Mgi + 超儿 (102.19)
Taking into consideration (101.3) and (101 .4) (for k ~ 乂) and likewise
the form of the functions cp(z), 少⑶，as determined by (35.10) and (35.1 1),
—
the solution of this problem will be sought in the form

9⑶ = £ 月；log (z ~ z?),
z F 炉 =1
(102.20)
27Vz J
L
t —
W(z) = +
2* J
L
力 一z 2兀勿
L
j t —z
• 2力 J
L
(力 — z)2

— S xAj— 10g(z — zj,

m
(102.21)
7=1

where the are constants. These constants will be related to the un¬
known function 3 修) by the formulae
438 V. SOLUTION OF BOUNDARY PROBLEMS § 102

Aj=l 3(%)〃S・ (102.22)

It is easily seen that the displacements, corresponding to the functions

中 (z), 少⑶， are single-valued in S.
As in the preceding case, a regular solution of the problem will be
sought. As above, it is for this purpose sufficient to assume that the
unknown function cog) has a derivative 3'«) satisfying the H condition.
, 、 乂 广 t n1
力 4 tc
+ —— / t — Eq + +
3(%)d log 厂- 亍- d7
J 2冗,
“
t — tq
' 1

力 L

+ 7z= 1｛log (%o

乂
— 》+
Z log (%
— 司｝/ 3(力)曲 = g&) on L,
J
(102.23)

where log (片 ~ 盯) log (琳 一 2) must be conceived as a single-valued

function which is equal to 2 log | tQ z；|. —
It will be assumed below that the given function g(£) has a derivative
g'Q) satisfying the H condition. Then, as in the case of (1025), every
(continuous) solution 3(/) of the equation (102.23) will have a derivative
3'(%) satisfying the H condition.
The integral equation (102.23) is found to be always soluble. In order
to see this, consider the homogeneous equation, obtained from it for
= 0. Let coo(Z) be any solution of this homogeneous equation and
中o(z), 少o(z) the corresponding expressions for the functions 中 (z), 少 (z).
Then
xcpoW + 却M) — +oW = o on E.
Hence, by the uniqueness theorem,
中0(z) = c, %(z) = K,
where c is a constant. As a consequence of the single-valuedness of
3o(z) which are simply constants, one obtains from (102.20) or
(102.21) that
= 0, / = 1,2, . . . , w, (102.24)
where the 4? are determined by (102.22) for s(£) = (oo(0.
Further, it follows from (102.20) and (102.21) that
CHAP. 17 SOME GENERAL METHODS 439

1
28晨
L

X
I =
—
>CC,
23 “
L
力 z 27rz J
L
t — z

whence it is easily concluded (cf. the case of the first fundamental prob¬
lem) that the functions 平*(力)，山* 修) determined by
匈 — c, — 他 *(£)
are the boundary values of some functions 乎*(z), 炉⑶，holomorphic in
= x(o0(^) + 必；(£) 十 xc (102.25)

the regions S1} S2> … Sm+1, while <p*(oo) = = 0.

Eliminating 3°«) from (102.25), one obtains
义勺* Q) — ~ W*Q) = — 2/xc on Lk, k — 1,2, . . . , w + 1. (102.26)
Applying the uniqueness theorem for the second fundamental problem
to each of the regions one finds
9*(^) =c 丸， ip*(z) ~ KCk + 2ixc
S* 左 = 1,2, . . . , w + 1,
in
and, since one has in Sm+1: 少*(8) = 平 (*8) = 0, obviously cm+1 = 0,
c = 0.

The functions 平*(z) = 0, = 2cMe clearly solve the second fundamental

problem, for with the boundary condition (102.26) ; by the uniqueness theorem,
the most general solution is obtained by adding to 中*(z) some constant ck and to
W*(z) the constant 比册 [cf. (34.13) and the remarks following it].

One thus has

中*(z) = ck, W*(z) = y^k in Sk(k = 1, … .，加)，
中*(z) = W*(z) = 0 in Sm+i・

But then, by ( 1 02.25) ,

3()(£) on. L*, k = 1, . . • , 加， 3()(£) - 0 on Lm^.
It follows from this by (102.24) and (102.22) that all ck = 0, and con¬
sequently 3(jQ) = 0.
Thus the homogeneous equation, corresponding to (102.23), has no
•solutions different from zero, and therefore the equation (102.23) has
always one and only one solution. Hence the problem is solved.
440 V. SOLUTION OF BOUNDARY PROBLEMS § 103

With obvious insignificant modifications, the above results likewise

apply to the case when the contour Lm+1 is absent and hence S is the
infinite plane with holes.

§ 103. On the solution of the mixed fundamental problem

and of certain other boundary problems by means of D. I. Sher-
man's method. The method of the preceding section may be success¬
fully extended to the solution of certain other important boundary-
value problems.
In the first place the mixed fundamental problem will be mentioned.
This problem was solved for regions of the same shape as in the preceding
section by D. I. Sherman [17] who used in this case the same representa¬
tion for the functions 少⑶ as in the case of the second fundamental
problem, i.e., (102.20), (102.21). However, this time the integral equation,
to which the stated representation leads directly, is already not a Fred¬
holm equation, but belongs to a class of singular integral equations
studied subsequently by G. F. Mandjavidze [1, 2]. D. I. Sherman himself
did not develop the general theory of this class of singular equations and
stated without a detailed investigation a method of solution based on
reduction to a Fredholm equation. (Another method of solution for
simply connected regions was also given by D. I. Sherman and has
already been mentioned in § 79). However, an ingenious complete study
of Sherman's integral equation with the aid of his general theory
mentioned earlier has been presented by G. F. Mandjavidze [2].
Using a method, analogous to the preceding one, D. I. Sherman [20]
gave a new and simpler solution than that by S. G, Mikhlin [10, 8] of
the first fundamental problem for bodies consisting in a definite manner
of homogeneous parts having different elastic constants ; this problem
(and likewise the corresponding second fundamental problem) has
already been mentioned in §§ 96, 98.
Finally, D. I. Sherman [22] gave (by means of a method, analogous
to the above) the general solution of the following problem. Let S be
a region of the same shape as in § 102, and let it be required to find the
elastic equilibrium of a (homogeneous) body, occupying S, if the normal
component of displacement vn and the tangential component of the
—
external stresses T on the boundary E of S are given. For T 0, this
problem reduces to that of the frictionless contact of the body under
consideration with rigid profiles at its boundary L. In the following
Part, this last problem will be solved for the case when the region S is
CHAP. 17 SOME GENERAL METHODS 441

simply connected and mapped on to the circle by means of rational

functions, as has already been mentioned in § 88, 2°.
Finally, it should be noted that analogous methods apply to the
problems of equilibrium of plates subject to normal loading for different
boundary conditions : when the edge of the plate is clamped, free or
supported, and likewise when on different parts of its boundary different
conditions are given which correspond only to the cases stated. Reference
will here only be made to the basic papers by Z. I. Khalilov [i], G. F.
Mandjavidze [2], A. I. Kalandiia [1, 2, 4], M. M. Fridman「2].
consideration with rigid profiles at its boundary L.
In the following Part, this last problem will be solved for the case when
the region S is simply connected and mapped on to the circle by means
of rational functions, as has already been mentioned in § 88, 2°.

§ 104. Generalization to anisotropic bodies. The methods of the

present Part may be successfully generalized to the case of homogeneous
anisotropic bodies. As shown by S. G. Lekhnitzki, complex representation
of the solution may also be given in this case, although it will, of course,
be more complicated than for isotropic bodies. By means of such a
representation and of a suitable generalization of the above methods
a number of general as well as particular problems have been solved.
The scope of this book does not permit a study of these questions. Therefore
only reference will be made to the literature on this subject; a detailed
statement on the literature is given in the survey by M. M. Fridman [3].
The interesting papers by S. G. Lekhnitzki will not be quoted here
in detail, since they are studied in his recently published book [1].
Among publications of a theoretical character, giving the general solution
of several fundamental boundary value problems, the following should
be mentioned : S. G. Mikhlin [1 1], G. N. Savin [3, 4], D. I. Sherman [9, 19],
I. N. Vekua [2].
The solution of many particular, but practically important problems
was given in the above-mentioned book by S. G. Lekhnitzki ; this book
does not only summarize the author's work, but also that of other in¬
vestigators. For this reason no detailed reference will be made here to
work, giving solutions of a particular character, as this can be found
in that book and in G. N. Savin [5, 6] (cf. also Savin [8]).

§ 105. On other applications of the general representation of

solutions. The methods of solution, studied in Parts IV and V (as well
442 V. SOLUTION OF BOUNDARY PROBLEMS § 105

as in Part VI), of the boundary problems of plane elasticity are based

on the general representation of the solutions of the corresponding dif¬
ferential equations by means of functions of a complex variable. Such
general representations of the solutions of partial differential equations
by means of "arbitrary" functions acquired exaggerated importance at
the outset of the development of mathematical physics, similar to that
given to the integration of ordinary differential equations by means of
quadrature. But it soon became clear that the determination of a
"general solution" by no means exhausted a problem and that for the
solution of the corresponding boundary problems such general solutions
were often next to useless.
This fact caused the usual reaction in such cases and led to other
extreme points of view which have been dominant until quite recent
times, i.e., that no benefit whatsoever may be derived from "general
solutions".
However, in actual fact, this is not so. The general solutions, if they
can be found and if they are used efficiently, are often extraordinarily
useful, particulary in practical problems. In a number of such cases they
permit the construction of a complete theory of a given problem in a
manner which is simpler and more thorough than would have been
possible by other, hitherto known methods ; the theory of plane elasticity
may serve here as an example.

In contrast to this, the hitherto known, general solutions of the equations of the
three-dimensional theory of elasticity do not permit the construction of a complete
general theory ; however, they have been, found to be useful for the solution of a
number of problems of a special character and have served as means for the solution
of several general problems.

Therefore it has been found very desirable to extend methods, analogous

to those studied above, to other sections of the theory of elasticity as
well as to a wider range of problems. There exist already results in this
direction which deserve more attention and further development.
Since there is no space to dwell at length on this range of problems,
reference will be made to the work of I. N. Vekua, where the method of
complex representation of solutions is generalized to a wider class of
differential equations of the elliptic type, to which the equations of
plane elasticity belong (in the static case) ; a full study of this work is
given in I. N. Vekua's book [1] and therefore only those of his papers
[4, 5] will be quoted here which refer directly to the theory of elasticity.
CHAP. 17 SOME GENERAL METHODS 443

The method of general representation has also successfully been

applied to several problems of elastic vibrations, but again no more can
be said about this here.
Finally, only a reference will be made to the general solutions of the
equations of the three-dimensional theory of elasticity, stated by Boussi¬
nesq, B. G. Galerkin, P. F. Papkovich and others (see also the earlier
remarks referring to this problem). Some information on this question
may be found in the text books by L. S. Leibenson [1], P. F. Papkovich
[1] and A. E. H. Love [1].
A
 PART VI

SOLUTION OF THE BOUNDARY PROBLEMS OF THE PLANE

THEORY OF ELASTICITY BY REDUCTION TO THE PROBLEM
OF LINEAR RELATIONSHIP

Many important problems of the theory of elasticity, including the

problems considered in Chapters 15 and 16, may be solved very simply
by reduction to a single boundary problem of complex function theory
which the Author calls the problem of linear relationship 0/ the boundary
values, or, briefly, the problem of linear relationship. The formulation of
this problem and its solution for several particular cases (which will be
required later on) is given in Chapter 18.
This problem has been called by many authors the Riemann problem.
It would have been more correct to call it the Hilbert problem, as has
been- done in the Author's book [25]. However, the Author proposes now to use the
above term, as an alternative.
A
 Chapter 18

THE PROBLEM OF LINEAR RELATIONSHIP

§ 106. Sectionally holomorphic functions. As in § 65, let L be the
union of a finite number of simple, non-intersecting arcs and contours
in the complex z plane ; these arcs and contours will always be assumed
to be smooth. As in § 65, L will be called a simple smooth line and it will
be assumed that it has a definite positive direction (i.e., on each arc or
contour which is a component of £). The ends of the arcs (if such exist)
will form part of L and will be called ends of the line L.
These closed arcs (i.e., including their end points) will often be denoted
by ab or, if there are several of them, by a/% 力 = 1,2, . . . , where the
symbols are to indicate that the positive direction is from 倪 to 方 or
from ak to 6m
As in § 65, a distinction will be made between "left'' and "right”
neighbourhoods of the points of L, other than its ends.
Denote by S' that part of the plane which contains all points not
belonging to 7L ; in other words, S' is the z plane cut along L. If L consists
only of arcs, then S' is a connected region, while, if L includes contours,
S' consists of several connected regions, bounded by these contours.
Let F(z) be some function, given in S' (but not on L) and satisfying
the following conditions:
1°. The function F(z) is holomorphic everywhere in S'.
2°. It is continuous from the left and from the right at all points of
X, other than the ends ak, bk.
3°. Near the ends ak, bk
4
|F(z) | <- 0 < pi < 1, (106.1)
I z
— c产
where c is anyone of the ends ak, bk} 4 is a positive constant and jz is
likewise a constant, subject to the stated condition.
Such a function F(z) will be called sectionally holomorphic in the entire
plane or, more simply, sectionally holomorphic. The line L will be called
the line of discontinuity of F (z) or the boundary.
447
448 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 107

As in § 65, denote by F+(£) and『(£) the boundary values of the

function F(z) at the point t of L from the left and right respectively.
On occasion, functions will be considered which satisfy the above
conditions everywhere, except at a finite number of points zv
not belonging to L, where F(z) may have poles (without any other
stipulations) . In such cases the function F(z) will be said to be sectionally
holomorphic everywhere, except at the points 卷，.... In particular,
functions will often be considered which are sectionally holomorphic
everywhere, except at the point at infinity, where they have a pole, i.e.,
functions which have for sufficiently large | z [ a series expansion of the form
C C
= Cmzm + 。惟_声 馆t
+ … + Co H —H +… (106.2)

Finally, the following condition will be introduced: when it is said

that F(z) vanishes at some point t0 of L which is not an end, this will
imply that F+(%) = =0； if 曲 is an end, then the vanishing of
F(z) at 片 will imply that f 0 as z f tQ.

NOTE. The definition of the concept of sectionally holomorphic

functions may, of course, be extended to the case, where the function is
not given in the entire cut plane Sf, but only in some part of it. For
example, let SQ be some connected region bounded by one or several
contours, the union of which will be denoted by Z。， and let the union
L of contours and arcs, considered above, be entirely contained in So.
If the function F(z), given in So (except at points of L), satisfies the
—
conditions 1° 3° and if, in addition, it takes definite boundary values
on the boundary Lo of SQ, then it may be called a function, sectionally
holomorphic in SQ. Such a function may be extended into a function,
sectionally holomorphic in the entire plane, by putting, for example,
F(z) = 0 outside S°.
In the sequel, unless stated otherwise, sectionally holomorphic functions
will be assumed to be given in the entire plane (except on the line of
discontinuity) .

§ 107. The problem of linear relationship (the Hilbert problem).

Let L be a given smooth line which satisfies the conditions of the
preceding section. The following problem will be considered :
To find the sectionally holomorphic function 尸(z) with the line of dis-
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 449

continuity L the boundary values of which from the left and from the right
satisfy the condition
F+(力 = G0)F-«) + 附 on L (107.1)
(except at the ends) , where G(£) and /(%) are functions, given o% L and
G(£) WO everywhere o% L.
In addition, it will be assumed that the functions GQ) and /(力) satisfy
the H condition.
Since the concept of boundary values from the left and from the right
is not defined for the ends of the line L, the reservation has been intro¬
duced that (107.1) is to be satisfied on L, except at the ends. In the sequel,
this stipulation will be omitted, although it will always be implied.
The above problem will be called the problem o/ linear relationship
。/ the boundary values or simply the problem o/ linear relationship or
the Hilbert problem [because the boundary values are connected (related)
by a linear expression (with, in general, variable coefficients)].
If /(£) = 0 everywhere on L, the problem will be called homogeneous.
The homogeneous problem was first considered by Hilbert for the case
where L is a simple contour ; the non-homogeneous problem (for the
same case) was proposed by I. I. Privalov (under somewhat more general
assumptions) . However, a complete, but very simple solution has only been
found recently. This solution and its literature is studied in the Author's
book [25].
Only the particular and very simple case when GQ) is a constant will
be studied here, because it is the case required in the later sections. For
the sake of clarity, the cases when GQ) = 1 and G«) = g, where g is an
arbitrary constant different from unity, will be considered separately.

§ 108. Determination of a sectionally holomorphic function for

a given discontinuity. The simplest case of the problem of § 107 occurs
when Gif) = 1. Then the problem reduces to the determination of a section-
ally holomorphic function F(z) for the given discontinuity /Q) so that
F+© —. 尸一㈤ = /(力 on L. (108.1)
The solution of this problem may be written down immediately.
In fact, consider the Cauchy integral
i r /w
及⑵ =
23 J %
— z
450 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §108

On the basis of the statements of § 68, FQ(z) is a sectionally holomorphic

function which vanishes at infinity and for which by (68.4)

— 尸大(力) = /W 0n L (except at the ends) . (a)

By Note 4 of § 68 the function F()(z) satisfies near any end c of 工 the condition
(106.1), viz.,
Z
i F#) |V z c
l 产 ——
even for arbitrarily small, positive 内

Hence 臬⑵ is one of the solutions of the problem. Next consider the

difference F(z)
By (108.1) and (a)
— = F*(z), where F(z) is an unknown solution.

Ft = 0 on L.
Thus, on the basis of a known property of functions of a complex variable
(§ 37, 4°), the values of F*(z) on the left and on the right of L continue
each other analytically. Therefore, if one prescribes for the function
F*(z) suitable values on L, this function will be holomorphic in the entire
plane, except possibly in the neighbourhoods of the ends ak, bk of L.
However, since in the neighbourhood of any end c, by (106.1),
A
I F*(z) I |z —~
c —r , 0 [i < 1, (b)

the point c is a removable singularity and it may be assumed that F*(z)

is holomorphic in the entire plane.
By the product (z — c)F*(z) remains bounded near c; hence this product function
has a removable singularity at that point (cf. for example, I, I. Privalov [1]).
—
Therefore (z c)F*(z) may be assumed to be holomorphic near c, i.e., (z c)F*(z)
where 声* *(z) is a holomorphic function. Thus F*(z) can only have a first
— —
order pole at c; but by (&) there is no such pole, because (z
—
c)F*(z) -> 0 as ? -> c.

Consequently, by the Liouville theorem, F*(z) = C = const, in the entire

plane and the general solution of the problem is given by
F(z) = 瓦(z) + C
or
(108.2)
2tc? J %
L — z
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 451

where C is an arbitrary constant. If it is required that F(oo) = 0, then

one has to take C = 0.
The solution of a somewhat more general problem will now be found.
In fact, it will be assumed that the unknown function F(z) is sectionally
holomorphic everywhere, except at the point at infinity where it may
have a pole of order not greater than m, i.e., it must have there the form
(106.2). It-is then easily seen (by application of the generalized Liouville
theorem) that
p = 不一 (108.3)
2冗2 J t
L — z

where P (z) is a polynomial of degree not higher than m, i.e.,

砒

尸/z) = Cmzm + 以_铲1 +… + Co； (108.4)

CG, Clt …，Cm are here arbitrary constants.
The generalized Liouville theorem consists of the following : If a function F(z)
is holomorphic in the whole plane, except at the point z = 8, and if for large | z [
F(z) = O(z多，
where m is a positive integer, then F(z) is a polynomial of degree not higher than

Finally, if the solution is permitted to have poles of order not greater

than 加 m at the given points 句，面，..., 画，oo, then
〃
1 /' /(/)弟
F(z) = 方r (108.5)
N兀z J t z
L —
where R(z) is an arbitrary rational function with poles of the type in¬
dicated, i.e.,

+ 7(Z— •十 ... + —
务严 +
?2

—^^7-
2
— —
…

j= l Z Zj (Z
品 C]Z 十 … + Cmzm} (108.6)
where the Ck are arbitrary constants.
NOTE. It follows from the above statements that every sectionally
holomorphic function F(z) may be represented in the form of a Cauchy
integral
F(z) = 方~丁 / C,
27rz J t
— z F
452 VI. APPLICATION OF THE PROBLEM OF "NEAR RELATIONSHIP § 109

where /R) denotes the discontinuity of 尸(z) on the line L, i.e.,

超) = F+(%)— 尸 (£), -
and C is a constant.
Further , if F(z) is a sectionally holomorphic function in some region
So which does not coincide with the whole plane (as in the Note at the
end of § 106), then this function may always be represented in the form
of the sum of a function holomorphic in S° and a Cauchy integral, i.e.,

F(z) =
2m J t / 卢一
—z
F*(z), (108.7)
L

where L is the line of the discontinuity /&) = F+«) — F-Q) inside So

and F*(z) is a function, holomorphic in S°. The expression (108.7) holds
true everywhere in So, except at points of L where F(z) is, in general, not
defined.
The truth of (108.7) follows from the fact that the difference
一、
1 广 力)惑 、
L

is a function, holomorphic in So except at points of L, where obviously

F*+Q) — F*-(£) = 0 on
and hence F*(z) is holomorphic everywhere in So, provided it is given
suitable values on L.
The function F*(z) may likewise be represented by a Cauchy integral
taken over the boundary 工°of the region So.

§ 109. Application. One interesting application of the formula (108.7)

will be stated here which is due to D. I. Sherman [14]. Let there be given
an elastic body, such as a plate with several holes, and let solid discs
of the same material be inserted into these holes ; however, let the contours
of these discs differ slightly, in the unstressed state, from those of the
holes. It will be supposed that the boundaries of the inserted discs and
of the corresponding holes are brought into contact without any gaps
and that they are welded together (or restrained by frictional forces
from sliding relatively to each other) .
Denote the body, obtained in this manner, by and its boundary
by Lo. It will be assumed that 工 is a simple contour.
°
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 453

The results of this section will also remain true (with obvious insignificant
modifications) , if it is assumed that 二 consists of several contours ; this corresponds
° by discs and where some of the inserted
to the case, where not all holes are filled
discs have holes.

Further, denote by L the union of the contours of the holes into

which discs have been inserted.
As before, let <p(z) and W(z) be the functions which determine the elastic
equilibrium of the body These functions are defined in each of the
regions into which the region So is divided by L and they are holo¬
morphic there ; however, they suffer discontinuous changes for a passage
through L.

This is obvious for regions, occupied by discs, because these functions are as¬
sumed to be single-valued. However, single-valuedness of <p(z), W(z) in the regions,
occupied by the material surrounding the discs, follows from the fact that the
resultant vector (and also the resultant moment) of the forces, applied to the edges
of the original holes by the discs, obviously vanish.

It will be assumed that the external forces Xn, Yn, acting on the
boundary of the body SQ, are given and that, in addition, the discon¬
tinuities in the displacements for a passage through L are determined by

— = 且式力)，
v~ = g2(^) on L,
什一 (109.1)
where the functions gW) and g2&) are likewise given ; they depend on
the shapes of the holes and inserted discs before deformation and on the
method by which the edges of the discs and of the surrounding plate
were brought into contact before welding occurred.
Under these conditions one has the following boundary conditions :

+ 他'(%) + 少修) = /0) on 二

<p(£) °, (109.2)

如'+4) + (力) 平 I/) + g'-(力 十 3一(%)on L, (109.3)

m中+0)
where
— — /(力) 比中一(f) — 即'i(f) — 小一(力) 2尔(%) on 上，(109.4)

“
/»

弗) (Xn + iY^ds on Lo> = g*) + 织?(/) onL (109.5)

o
are known functions.
The condition (109.2) expresses that the external stresses, acting
 454 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 109

on the boundary of the body So, are given. The condition (109.3)
indicates that the stresses, acting from either side on the element of
the line of division, balance each other. Finally, (109.4) shows that
the discontinuities in the displacements of the line of division are known.
As a matter of fact, (109.2) must only be satisfied exactly, apart from
an arbitrary constant ; similarly, (109.3) must be fulfilled, apart from
arbitrary constants on each of the contours, constituting L. However,
it is easily seen that the last constants may be included with the unknown
functions.
Adding (109.3) and (109.4), one finds
光幽
中+(力) 一 旷(力)
M
+ -1
on L. (109.6)

Further, taking the conjugate complex form of (109.3) and using (109.6),
one obtains

W+(f) —旷 = 产= on L,
K 十 1
(109.7)

where

碑) = — 丽一 g肛g'G) -华 ， 009.8)

i.e., h(t) is a function, known on L.

On the basis of the statements in the Note at the end of § 108, one
deduces that

6力= 中。
L
(109.9)
W(z) = Wo + 十 1) J —
t- Z

where 午o(z), K(z) are functions, holomorphic in S°.

For the sake of brevity, let
广 g©dt
中*3) ， (109.10)
“X + 1) LJ
九
-
t z + 1) LJ t~z

and (109.9) becomes

中 = 平o(z) + 中* W(z) = %(z) + W*(z); (109.11)
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 455

the holomorphic functions 乎o(z), %(z) are subject to definition, while

<p*(z), W*(z) are known, sectionally holomorphic functions, determined
by (109.10).
Substituting (109.1 1) in the boundary condition (109.2), one obtains

-
+ W) 泌 on Lq> (109.12)
where
fM /传) 一 中*R) 一 祠'一郎 (109.13)
is a function, known on L&
One has thus arrived at the usual first fundamental problem for
the body S°. A仕er having determined 仰⑵，必⑶，the functions <p(z), 少⑵
may be found from (109.9) or (109.1 1).
Consequently, it is seen that the problem under consideration reduces
directly to the czistomary Ryst fundamental problem for the same region S°.
If, instead of the stresses, displacements are given on L^, the problem
will reduce in the same manner to the customary second fundamental
problem.
If the discs and the surrounding body have different elastic properties,
then the above is no longer true; more will be said about the solution
of this case later on.

§ 109a. Example. Consider the simplest case of a circular ring with

outer radius 1 and inner radius r into which a circular disc of radius / + s
has been inserted, where e is a known quantity. Then Sq is the unit
circle, Lo the circumference of this circle and L a circle with radius
r < 1.
If it is assumed that the origin lies at the centre and that the positive
direction on L (as well as on %) is counter-clockwise, it is easily seen
that, with t = pe? one has
g^) = — e(cos 卧 十 a sin 8) — ~
诒
= r
on L
and
— _
—
2eE 2b
. gR) 皆(力) = — on L.

Hence (109.10) gives

2眸 0 for \ z\ <r
f z for \ z \ <r
中*(z) =｛ 旷 (x 十 1) ， 中*(r= 1
I 0 for
—z for | j >r
456 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 10

and the functions 中°(z), 狐⑵ are determined by the boundary condition

勺o(*) + 力中o(') + WoG) = /o(£) on 乙(),

where
4p,er
/o«) = 力) + x十1
The last expression follows from (109.13), since g*(/) = = 0 on 工。，because
<p*(z) = 0 for | z | > 夕 and

w*(/) = —— 比
4^,87
十 1
1
2
= —
-
4 peer
>c 十 1
on 工o・

Thus, in order to solve the original problem, one has to find the solution
of the customary first fundamental problem for the circle, after adding
to the stresses, actually acting on & and characterized by /(/), the fictitious
stresses corresponding to the second term in the expression for %(/),
obtained above .These fictitious stresses are easily seen to correspond to
a distribution of uniform normal tension of magnitude
4H e.
*+ 1
Thus the solution of the problem may be written down directly, using
the formulae of § 80 * .
§ 110. Solution of the problem : F+ = gF~ + f.
Consider the case G(力) ~ g, where g 片 1 is a given, in general complex
constant. The boundary condition in this case will have the form
F+&) — — /(/) on L, except at the ends. (1 10.1)
It will now be assumed that L consists of n simple smooth arcs Lk
3 = 1,2, which have no points in common ; these arcs, as in¬
dicated earlier, will be denoted by a 血却 where ak> bk are the ends of Lk
The case, where L consists only of contours, is easily seen to reduce directly to
the problem of § 109. For example, if L consists of one simple contour which divides
the plane into two regions S+ and , adjoining L on the left and right respectively,
one may consider, instead of F(z), the function defined as follows : F*(z) =
F in S+, F* 工 gF(z) in S~; then (1 10.1) takes the form
One may proceed in an analogous manner, when L consists of
(Z)
several
—
= 棒).
contours.
In the case where L consists of contours and arcs, the problem is likewise easily
solved.
* Certain other simple applications were given by N. D. Tarabasov [1, 2] and
A. G. Ugodchikov [1] ; a more complicated case has been studied by D. I. Sherman
[23].
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 457

and the positive direction is from ak to 匕比. (Fig. 48).

First, a solution will be studied which may have a pole of arbitrary order
at infinity, and a beginning will be made with the homogeneous problem
F+G) — gFR) = o. (iio.r)
A particular solution Xo(z) of this problem will be sought in the form

—
n
X°(z) = n (z — 能尸(z 娟Y-1, (110.2)
夕=1

*
where 丫 = a 十 is a constant.
The function Xo(^) is holomorphic in S[ i.e., in the plane cut along L,
if a definite branch of this function is selected, e.g. the branch for which

Fig. 48.

lim [z"Xo(z)] = 1 ; or, in other words, the branch which has for large [ z |

=
Z—A OO

the form
X°(z) =
尹 + zn~r …； (110.3)

in the sequel, unless stated otherwise, this branch will always be implied.
It is readily verified by an investigation of the variation in the ar¬
gument of z — ak or z — bk, when z describes a closed path beginning
at a point 力 of the arc a* and leading, without intersecting L, from
the left side of 纵既 around the end ak to the right side of the arc (as in
Fig. 48) or around the end bk (not shown in Fig. 48), that
为0)= 产"x底)， (110.4)
.
ie, X*)= e2mx渡).
458 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 10

By definition, the power with complex exponent

(z
— aJ" = = 5-YlIogi2-afc| + iai = ^-Ylog lz — a" 小丫眇，

where log [ — ak | is real, so that

— 心厂y = — 即 小吟
(Z -Y

where 3 = arg (z
— 纵) and |z — ak\^ is the uniquely defined value g~Ylog [z-a^^
When z goes from the left side of Lk around ak to the right side, as shown in
Fig. 48, then 9 increases by (十 2tc), and hence (—〃丫9) by (— 2 丫)，and therefore 拓
(z
—When z goes from
-y
a把) must be multiplied by
side of Lk around bk to
the left the right side, then
9 = arg
— %) increases(z by既)(— 2兀)
(z and

must bemultiplied by
—6-2^"丫—1)
y-1

= 一2^次，as
1=

6 first case,
『(y-DS
丫7
in the since e2ni = 】.
Hence Xo(^) will satisfy the boundary condition (1 10. T)» provided
於出丫 = g, ie,
, -o = bgg = loglgl , e
Y = a +，B + -%■
WT F—
(110.5)

where 6 denotes the argument of the constant g. This argument is de¬

termined, apart from an additive term where k is an integer ;
however, 9 will always be chosen in such a manner that
0 6 < 2k, (110.6)
by which condition 9 is completely determined. In particular, if g is
a real, positive number, then 9 = 0, while, if g is a real, negative number,
8 = 加.
It will now be investigated as to whether the inequality
/
XoOv^I z—c r 0<；1< 1 (110-7)

is fulfilled, where c is any end ak> 既；this condition must, by supposition,

be satisfied by any sectionally holomorphic function. By (1 10.6),
0 a < 1. (110.8)
If g is not a real positive number, then a # 0, 1 a < 1. Hence, ex¬
cluding the case when g is real and positive, the inequality (1 10.7) will
—
be fulfilled by taking = a for c = 瞅，浊 = 1 a for c
“ bk. — —
One has
_ e-(a+
3 一 纬)—Y = (z
— 纵)Ta+ 郊) = g-(a+?3) log(z-ak) 痢 (logr + 毋)，
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 459

where / = |z — 耿 [, 9 = arg (z ― a%) . Hence

•

(Z
— 即尸 = £-alogr.® =
where ® so that ] 0 J = When z is in the neighbourhood
of the point ak in the plane, cut along L, then 8 lies between finite limits (because
z cannot cross L) and therefore | 0 | is bounded and, in addition, | 0 | > a, where
a is a positive constant. Similar reasoning applies to the neighbourhood of the
point 近.

Thus, a particular solution X°(z) of the homogeneous problem has

been found (for a 7^ 0) ; it is given by (1 10.2) with 丫 determined by (1 10.5).
This particular solution does not vanish anywhere in the finite part of
the plane and it is unbounded like | z ak j-a and | z - bk
the ends 0 尢 and bk respectively.
— — near

The most general solution of the homogeneous problem will now be

found which has a pole at infinity. For this purpose it will be noted
that Xo(z), being a solution of the homogeneous problem, satisfies the
condition
=gX60) on L, (H0.9)
whence

且=篇
项传)
0nL (H°9)

Replacing in (110.1') g by its value (1109), one obtains

尸+Q) F-R)
on L,

or
=0 on L,
where 尸*(z) denotes the sectionally holomorphic function 尸⑵ /Xq(z).
It follows from the preceding relation that F*(z) is holomorphic in the
entire plane, except at the point z = 00, provided it is given suitable
values on L (cf. § 108). Further, since 尸*(z) can only have a pole at
infinity, it must, by the generalized Liouville theorem, be a polynomial.
Thus, the most general solution of the homogeneous problem is given
by
F = Xo(z)P(z), (110.10)
where P(z) is an arbitrary polynomial.
If it is desired to obtain a solution which is also holomorphic at
 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 110

infinity, it must be assumed that the degree of the polynomial P(z)

does not exceed n; this follows from the behaviour of Xo(^) at infinity,
as determined by (1 10.3). If one requires that 尸(8) = 0,then the degree
of P(z) may not be higher than n 1. —
In general, the solution (1 10.1 0) will not be bounded near the ends.
However, if it is desired to find the solution which is bounded near the
given ends c2, ..., cv> the polynomial 尸(z) must be chosen in such a
manner that it vanishes at these points, i.e., 尸⑵ — (z
(z•
~
— —
(z c2) ...
where 0(z) is a polynomial. In that case the solution F(z)
will not only be bounded near the ends ck, but it will vanish there. (It
is seen that a solution which is bounded near certain ends, but does not
vanish there, does not exist, assuming,of course, all the time that a W 0.)
Writing
驾⑵ (z — q)(z — C2) … (z — (HO.ll)
all solutions bounded near the ends q, c2, may be represented
in the form
F(z) = X#).0(z), (H0.12)
where 0(z) is an arbitrary polynomial.
Naturally, is itself a particular solution of the homogeneous
problem, similar to However, it is bounded near the given ends
and it vanishes at these ends in such a way that
Xp(z) = |z — 匕®, OVjzVl, (110.13)
® being a bounded quantity; in fact,
| ®
| > a = const > 0.[cf. remarks
following (110.8)].
Among the solutions X*z) the two following will be specially noted:
n
X = 。 " —
口 (z )(z —妃 口 (z — 为广丫(z — 娟y，
(110.14)

which is bounded near 成％ ends (where it actually vanishes), and

— a3qJ
X*(z)=Xo(z)II(z — %) = n , (110.15)
J =1 7 =1

which is bounded near the ends bjf j = 1,2,...,% (where it vanishes).

For large|z J these solutions have the forms

X{z) + + ... + z
(110.16)
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 461

X*(z) = l+ 子 十圣+ …. (110.17)

Next consider the non-homogeneous problem. Using (110.9'), the

boundary condition (1 10.1) may be written
F+(£) F-0) 超)

or

where F^z) = F(z)/X°(z),f* 修) = /例XQ).

Using the results of § 108, one finds

下⑵ = 兀 + X。 ， (H0.18)
2 2 J
L
{t
— Z)

where P(z) is an arbitrary polynomial. This is the general solution of

the problem, admitting a pole at infinity.
If it is desired to obtain the solution, holomorphic at z = co, it must
be assumed, in view of (110.3), that P(z) is a polynomial of degree not
higher than %:
尸⑶ = c* + +… + Cn^z + Cn, (1 10.19)
where Co, Clt . .. , are arbitrary constants. If, in addition, it is re¬
—
quired that F(oo) 0, one has to assume Co = 0.
In general, the solution _F(z) will not be bounded at the ends ak, bk.
However, by a suitable choice of the polynomial P(z), it may be ar¬
ranged that it is bounded at certain ends clf c2, . . c^. It is simple
to prove this directly by constructing the general solution, having that
property. For this it will be sufficient to repeat the reasoning leading to
(110.18), but using, instead of the particular solution Xq(z), the solution
Xp as determined by (1 10.1 1).
Also in that case one is led to the condition
time
(/)
— F不 = /*(£), where this

F* (z)
哲项
however, Xj)(z) vanishes now at the ends clt but since, by supposition,
the unknown function is bounded near these ends" F*(z)|vZ/| z 内 巴 0
— V 1,
and the former reasoning applies, provided one does not consider the fact that. “
462 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 10

in general, /* (£) is not bounded near the ends It is readily shown by a study of
the behaviour of /*(%) near these ends [cf. N. I. Muskhelishvili [25]) that this cir¬
cumstance is of no importance. As a matter of fact, it has been shown earlier that,
if a solution of the required type exists, it is given by (110.20) and that the first
term on the right-hand side of (1 10.20) actually remains bounded near the ends
‘2，
'1，

Thus the general solution, bounded near the ends q, c2, . . . , cP, is
given by

(110.20)
NTH J
L
R — 司

where P(n) is an arbitrary polynomial.

It follows from (110.11) and (110.3) that for large |z|
Xp(z) = T 十 • •. ； (110.21)
therefore X*z) is holomorphic at infinity only for 《 力 ％.
If 力《切+ 1, the first term on the right-hand side of (i 10.20) remains
bounded as z 8 and, in order to obtain a solution which is also ho¬
lomorphic for z = 8, one must assume that P is a polynomial of degree
—
not higher than n 夕；for 夕 = + 1, one has to assume P(z) — 0.
”
However, if 力〉％ + 1 , a solution which is holomorphic at infinity-
will only exist, provided certain conditions, to be stated now, are satisfied.
Since for large | z |

one has the expansion, likewise valid for large | z

1 / /(/)&£ & A2 (110.22)
2耘 J
L
X菰力)(£— z) z 弹

where

/
1 弟
1 , 2, . . .. (110.23)
密./ XjQ厂
h
L

Hence, if it is required that the solution (1 10.20) should be holomorphic

at infinity for 夕 > + 1 , then one has to put P(z) = 0 and, in ad¬
dition, /(/) must be subject to the conditions Ak = 0, 力 = 1, 2, . .
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 463

0 — —
% 1, i.e.,
1 /
而 ./ 号0厂
= 0, 为= 1,2, ...，勿 —— —— 1 .
％ (110.24)
L

Thus (1 10.24) must be satisfied, in order that for 力〉％ + 1 a solution,

holomorphic at infinity and bounded at the ends %,c2, . . . , may exist.
Further, if it be required that F(8) = 0, then in the preceding formula
k = 1,2, — %.
Hitherto the case where a = 0 has been excluded, i.e., the case where g
is a real, positive number. If a = 0 (assuming as before that g # 1 ) , one
may use the particular solution X*(z), determined by (1 10.15) : It is
readily seen that this solution remains bounded near all ends and does
not vanish anywhere when 丫 = 笫 胃 0. Applying the same reasoning as
before, one obtains the general solution of the problem
F(z) + X*(z)P(z), (110.25)
X 注力)Q-z)
where P(z) is an arbitrary polynomial ；in the present case the solution
is seen to be necessarily bounded near all ends *. If it is desired to find a
solution, holomorphic for z = 8, one has to put P(z) = C = const.,
as may be seen from (1 10.17) ; if, in addition, F(oo) = 0 is required,
then P(z) = 0.
The formula (1 10.25) is, of course, also applicable for a # 0, when
it becomes a particular case of (1 10.20) and gives all solutions, bounded
at the ends blf 为，.…, 办公
Finally, the general solution of the problem will be found under
the supposition that it may have poles of order not greater than
加2，• • • , 加 at the given points z】, , zlf oo. Reasoning as before
and taking into consideration (108.5), one obtains for the general solution
of the required form, applicable for a # 0,

确=学打药器^r。（纳佟），L
010.26）

where R[z) is a rational function of the form [cf. (108.6)]

距） =,上{3+ 十 …+ 金身 十 尸 （u。.27）
* It can be shown that the first term on the right-hand side of (1 10.25) remains
bounded.
464 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 10

P(z) being an arbitrary polynomial of degree not higher than + %.

The polynomial P(z) must be such that the pole of P {2)X^2} at infinity is not
加

1
of greater order than m; however, it is known that for large | z | : X°(z) =— 十
j-JLi
-i_
十 巴 4-
十 •
/+i
When a = 0, an analogous formula results which is obtained by
substituting X*(z) for XQ(z) ; in that case the degree of the polynomial
P(z) must not exceed m, because for large [ z |

川）*十子+ •…

In conclusion, note the particular case of the problem under consi¬

deration when g = —
1 ; the boundary condition takes then the form
F+团 + f— = /(札 (110.28)
and one has .log . ,
(—— 1)
—
. 1
2Klz ——
丫= a +空= r = . (1 10.29)
2
Hence, by (110.2),
X°(z) = £ (z — 勾尸 (z — 6尸 =
]
= …… 以… … (11 0.30)
"(z ~ 诙)(z
•
— &)...(% — an)(z — 办力)
and, by (110.14) and (1 10.15),
X(z) — 7(z 旬) (z '~• ^1) ... (^ 笈 ti)(z ~
' , (110.31)
— &)...(? — 鼠)
— V (z 质) ...(z ~
•
距)
(110.32)

The most general solution of the problem, admitting a pole at infinity,

is given by (1 10.18) with Xo(^) replaced by the value (1 10.30); using the
fact that now X()(z) = 1/X(z), one has for the general solution
i r x+e）超成 P团
F团 =
而编 J +谢 (110.33)
L
where X(z) is determined by (1 10.31) and P(z) is an arbitrary polynomial.
Assuming the degree of the polynomial not to exceed n, one obtains
the solution which remains holomorphic at infinity.
The solution, bounded near all ends, is obtained from a formula which
follows from the general formula (1 10.20) for /> = 2% and X*z) deter-
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 465

mined by (1 10.31) :

器J 亚+% —
2兀2
L
A+(t) (f Z)
十 X(z)尸(z). (110.34)

If the solution is also to be holomorphic at infinity, one must assume

P(z) = 0 and, in addition, subject the function f(t) to the following
conditions, which follow from (110.24) :

桃TfM
「 力 = 1,2, . . . , % — 1. (110.35)

If also F(8) = 0, then 6 = 1, 2, . . . , % in (1 10.35).

NOTE. 1. In many cases the integrals, occurring in the formulae

of the present section, may be easily evaluated in finite form. For
example, this is true in the case, which often occurs in practice, when
/(/) is a polynomial
第) = 4 加
产
+ + … 十 4°. (H0.36)
In fact, consider the integral

'⑵ =/席黑k
L
(,10-37)

which occurs in (1 10.20) ; in particular,

for 少=0, one obtains the integral of
(110.18).
Simultaneously with the integral /
consider another integral

("38)
A

where A is the union of the n con¬

tours A】, A2, ...» An, surrounding the
arcs J, %， …，Ln in clockwise direc¬ Fig. 49.
tion, as shown in Fig. 49, and assume
that in this case z remains outside these contours.
466 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 110

Using (1 10.21), it is concluded that for large |已|

p(q)
；
+ + ... + H q
H—q + • • , (110.39)

where 夕 = —
力 + 加 and the coefficients %, aQ_x, . . . , a0 (the others
not being required) are easily determined by elementary means ; for
夕 V 0, one has to assume that all of them are equal to zero.
In fact, is the product of binomial terms of the form
may be expanded as follows
—
• c) 大 which

By (110.21), the sum of the exponents 入 equals %

—— 夕，i.e. an integer (or zero).

Hence, by (703), one obtains

Q = V M — ao
(if 0 V 0, the polynomial on the right-hand side will vanish).
In (70.3') the integral is taken over one contour only, but obviously this is of no
importance. The difference in sign arises from the fact that the positive direction
of the path of integration in the present case is opposite to that in § 70.

On the other hand, letting the contours shrink into the arcs Lk
and noting that Xp(Q in (1 10.38) will then tend to or XgQ), de¬
pending on the position of 已 on A*, it is readily seen that
他)就 r 的祝
J
A*
x品g (二不 J 若而f
ajtb^
十
九伙
J x熊北二B
or, since one has X~(/) = 1/gX*(力) on Lk and must change the sign of
the integral when replacing the path. bkak by the path akbk,

J xg (厂才 -g)J
hence it is obvious that
他城
—
Xt(C) (C z)
(1 一 g)
/ X部）（—）
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 467

Consequently
2脑
/⑵ = Q(z) =
J
L
可用J
J
-
2戒
—
= j二
/(z)
1. (110.40)

The fact that the function 1/X»(z) may be unbounded near the ends is easily
seen not to influence the above reasoning, since near any end c
1 const,
匚不，"i，
and hence integrals, taken over a small circle surrounding the end, tend to zero
(cf. Fig. 49).

Integrals of the form

(110.41)
L

where m is an integer (positive, negative or zero), may likewise be

evaluated in finite form. In fact, proceeding as in the preceding case, it
is found that

力厂 J n就厂’
A
一

where A is the same as before. On the other hand, if a_TO denotes the
coefficient of z~m in (110.39), then, by the residue theorem,
他睡
— — 2兀以_帆
and, consequently,
睁一叶 (it 2ma_m
(110.42)
x*r=— i 二p

Analogous results may be obtained when /(/) is a rational function, and

not only a polynomial.
It must be borne in mind that, if /«) is represented by different po¬
lynomials (or different rational functions) on the different arcs con¬
stituting L, then the calculation of the preceding integrals may, in general,
not be as easily performed.
468 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §111

If L only consists of one arc and if /(%) is not a polynomial, then in the
majority of cases which are of practical interest one may replace /0) with
sufficient accuracy by a polynomial with a small number of terms, or by a
rational function.
NOTE. 2. When solving the boundary problems above, the selected
particular solutions X*z) of the homogeneous case have been used in
a definite manner. However, it is obvious that nothing will be changed,
if X*z) is replaced by CXv{z), where C is an arbitrary constant which
is not zero. It is only important in (1 10.20) and those formulae, connected
with it, that X* 心) on L is the value taken by the selected functions
on L from the left.
For example, let g = ~ 1 and £ be a segment ab of the real axis.
In agreement with the above condition, that branch of the function

— a) (z —
X(z) =
“ (z
in (1 1 0.33) must be taken which for large | z | is given by
6V 6
八务/
/-
V iz — —
a) (z
/
—z
= z(\1 — ) (\ 1 / z/
1 =z
々

2
(a— 6)2
8z
F…
This function takes on the segment ab purely imaginary values. But
sometimes it is more convenient to deal with a function, having real values
on this segment. Such a function is, for example,
“(z —— a) (b —— z) = 土 iV (z —— a) (z —— b) = 土 iX(z).
If one takes the lower sign, i.e.,

— “(z (b — z) = — £X(z),
a) • •

one obtains a function which is easily seen to take positive values on

the left side of the segment ab of the Ox axis.

§ 111. Case of discontinuous coefficients. It is also not difficult

to find the solution in the case, where the coefficient G(%) in the boundary
condition
—
F+(£) G 也 F-电 = /(0 on Z (111.1)
remains constant on different parts of L, but changes discontinuously
for transition from one part to another; in this connection the term "part”
refers to sections, into which the line L may be divided by a finite number
of points cj on it. However, in this case one has to admit that the function
CHAP. 18 THE PROBLEM OF LINEAR RELATIONSHIP 469

F (z) may be unbounded near the points of discontinuity ci as well as near

the ends of the line L, where it must satisfy the same condition as at
the ends ; in general the points ci will be analogous to ends.
It will be left to the reader to deduce the solution in the general case ;
only the following particular case will be studied here, which is the only
one required in the sequel.
Let Z, be a simple contour and let
this contour consist of the n arcs
L] — 应 , L?
* 口2^2， • , "^处 — 以丁,'
'
处

without common ends (Fig. 50) ; the

notation has been chosen such that
during a circuit of L in the positive
direction the points %,br, a2, b2> . .
an,bn are encountered in the stated
order.
Denote by L' the union of the arcs
Lk, 6 = 1, 2, …，％ and by Ln the
remaining part of L, i.e., the union of the arcs 公出，%他，•一 图旬，and
assume that
G(%) = g on Lf, G«) = 1 on L" , (111.2)
where g # 1 is, in general, a complex constant. Thus the boundary
condition (111.1) takes the form
—
F+修) gF— 0) = /伍) on L\ F+«) —
F-(£) = /(%) on (111.3)
It will be assumed that the given function /Q), satisfying the H con¬
dition on Z/ and L,f separately, may change discontinuously for passage
through the points ajt
The homogeneous problem will be considered first :
尸+0)— gF— (£) = 0 onE', 尸+(£)
— 尸一⑴ =0 on 上〃. (1113)
The second of these equations shows that the values of the unknown
function F(z) inside and outside L continue each other analytically
through the part Z/' of L, or, in other words, that Lr, is not effectively
a line of discontinuity.
Thus one arrives at the same homogeneous problem
F+伍) — gF-0) =0 ；
on ( 1 1 1 .3〃)
where F(z) is a function, holomorphic in the plane cut along Lr, as in
the preceding section. Hence, for example, the function deter-
470 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §111

mined by (1 10.2) and (110.5), is, for a # 0, a particular solution of

(1113), as may easily be verified directly ; for a 0, one has to take
X*(z), determined by (1 10.15), instead of
—
Now consider the non-homogeneous problem (1 1 1.3). Using the solution
X°(z) and noting that, by (H0.9') [remembering that here L' takes the
place of E in § 110], one finds that

g=
驾。
"
(111.4)
and that
_1 cn 厂， (111.5)

because the function Xo(^) is holomorphic everywhere, except at points

of Lr; consequently (1 1 1.3) may be written as one single formula
F+团 fr) ，，"
阚 J
一
一 一

X*) '

whence, as in § 108,

J,'- JLx工需
——
F(z) = ^~+XoC)F(z), (111.6)
2 兀2 丁伍 7N ) z)
工

where 尸⑶ is an arbitrary polynomial and F(z) may have a pole at in¬

finity.
The formula (1 1 1.6) does not appear to differ from (110.18); however, this is
due to the notation : in (1 10. 18), the integral is taken, in the notation of this section,
along U, and not along L. The formula (1 1 1.6) will actually agree with (110. 18),
if /(Z) = 0 on L" , and this is quite natural, since in that case one is dealing ef¬
fectively with the same problem.

If the function is to be holomorphic at z = oo, it must be assumed that

the degree of P(z) does not exceed %; if F(z) is to vanish at infinity, the
degree of P(z) must not exceed n — 1.
If g is a real, positive number, then one has to take in (1 1 1.6) X*(z)
instead of X()(z).
If F(z) may have poles at given points, not on L, the formula (1 11.6)
must be replaced by another one, analogous to (1 10.26).
Finally, it may be noted that the preceding results are easily extended
to the case, where the line L extends to infinity, e.g., where it is an infinite
straight line ; this extension is so obvious that no more need be said here.
 Chapter 19

SOLUTION OF THE FUNDAMENTAL PROBLEMS FOR THE

HALF-PLANE AND FOR THE PLANE WITH STRAIGHT CUTS

The results, studied in Chapter 18, offer the possibility of solving in

a very simple manner the principal boundary problems for those cases,
where the region under consideration is the half-plane or the plane with
straight cuts (along one and the
same straight line) , including the
first and second fundamental
problems for the half-plane which
have already been considered in
the preceding Part. Some of
these problems, and likewise
the important problem of con¬
tact of two elastic half -planes,
will be considered in this chapter (§ 119) 】) Fig. 51.
For this purpose certain of the results of the preceding chapter must
be applied also to the case when the line L is an unbounded straight
line. In spite of the fact that the case when the line L extends to infinity
has not been considered there, the extension of the corresponding results
to this case is so simple that no space will be devoted to it here apart
from some remarks in the relevant sections.

§ 112. Transformation of the general formulae for the half-

plane 2). 1°. It will be assumed that the elastic body occupies the lower
half-plane y < 0 which will be denoted by S~, so that the region S- is
to the right, if one moves in the positive direction along the Ox axis.
The upper half-plane will be denoted by S+ and the Ox axis by L. (Fig.
51).
First, general formulae will be recalled which will constantly be used:
i) For a number of the problems considered here see the recent monograph of
L. A. Galin [4] and likewise that of I. la. Staerman [4].
2) N. I. Muskhelishvili [22].
471
472 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 112

Xr + E, = 2[① (z) + 丽], (112.1)

匕'~ Xj；+ 2iXv = 2[z*(z) + 乎(z)], (1 12.2)
2^(% + 浏 = 乂中⑶ 一 —碉， (112.3)
where = 中侬，乎⑵ = V(z) are functions, holomorphic in S". As
in Chapter 16, it will be assumed that the resultant vector (X, Y) of
the external forces, acting on the boundary, is finite and that the stresses
and rotation vanish at infinity ; hence one has for large | 2 (, as in § 90,

%一留 +6) w)=中十。(2(”2.4)

+ (3)

+ iY
八

勺 (z)
— X
2n
小
log z 十 o(l) + const.,
(112.5)
Mz) =
X — iY、
log z + o(i) + const.
2H

The following formula follows from (1 12.1) and (112.2) :

Yy — iXy = 函 + z硝 + 函， (112.6)
while the formula
2M + 说1) = 比①⑶ 一南 一 —场 (1127)
is obtained from (1 12.3) by differentiation with respect to %, so that
加
u' 0，= 丁 ， (112.8)
3% ox
where this notation will be used in the sequel.
2°. These formulae will now be transformed by extending the definition
of the function to the upper half-plane. This extension may be
achieved in many ways, because (z) is, in general, not defined in the
upper half-plane. However, practically useful formulae are obtained by
executing the process by definite methods, one of which has been stated
earlier, while another analogous one will be indicated in § 1 13. In fact,
the function (z) has been defined in the upper half-plane in such a way
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 473

that its values there continue (z) analytically from the lower half-^plane
through the unloaded parts 。/ the boundary (if such exist).
A way of doing this is easily deduced from (1 12.6) [cf. Note in § 93].
Actually, this formula shows that on the unloaded parts of the boundary,
where obviously Y~ = X~ = 0,

—
-Q) +(/) = o,
provided that in the upper half-plane the function (z) is defined in
the following manner :

—
(z) = 币(z) —
This will now be explained in detail.
—¥(z) (for z in S+). (112.9)

Here use has been made of the notation, introduced earlier ; in fact,
the signs (+) and (—) indicate the boundary values from the left and
from the right of L (i.e., in the present case, from the upper and lower
half-planes), while _F(z) denotes the function obtained from F(z) in the
following manner (§ 76) :
户(z) = 晴； (112.10)
if F(z) is holomorphic in S~[or S+], then 尸⑵ will be holomorphic in
s+[s-].
It will also be remembered [(76.6), (766)] that, if 尸⑶ is defined, say,
in the lower half-plane and if FR) exists at a point t of the real axis, then
#+(/) exists and
立项 =下+(方)； (112.10")
similarly, interchanging the parts played by the upper and lower half¬
planes,
同)= 户一g). (H2」0〃)
It is clear from the above that the right-hand side of (112.9) represents
a function, holomorphic in the upper half-plane S+, and that on the un¬
loaded parts of the boundary the condition (a) holds true.
Replacing in (112.9) the variable z by z, assuming that z lies in
(and hence z in S+) and taking conjugate complex values on both sides
of (112.9), one finds
不⑶ = — (z) —z '⑶ 一 手⑵，
whence
474 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 112

甲(N) = —① (Z) — (Z) — z*(z); (H2.ll)

this formula expresses ¥(z) for z in in terms of (z) which has been
extended into the upper half-plane.
Without the extension of (z), the formula (1 12.1 1) does not make sense, because
it involves which is, by definition, equal to ; however, in order that (g)
for z in the lower half plane (i.e., for 5 in the upper half-plane) have a meaning, the
function (z) must be defined in the upper half-plane.

Introducing the value (1 12. 1 1) for 甲(z) in (112.2), one obtains the
following expressions for the stress components in terms of the single
function (z), defined in the upper as well as in the lower half-plane :
X* + 匕 = 2［① (z) + 不 (112.12)

whence
匕 一 工 十 2iXy = 2［( 万一 z) '⑵ 一 (z) — &⑵ ］, (112.13)

— + (z — 2)6 (112.14)
where the last formula could also have been obtained directly from (112.6).
Further, it follows from (1 12.7) that

2t1(优 + 加') = k① (z) + 一 (z — 团@(5). (112.15)

An expression for 2［i(^ + 说) is likewise easily deduced, if one also
extends 平 (z) into the upper half-plane under the condition 勺'(z) = (z)
in the upper half-plane as well. Noting that, by (1 12.9), in the upper
half-plane (z) = — + w⑵］', one obtains
中⑶ = — 一 g(z) + const. (forzinS+), (112.16)
and hence, as before,
j(z) = — 中 (z) — ztp'(z) + const, (for z in S~). (1 12.17)
With this value for 少⑶ the formula (1 12.3) becomes
2［±侬 + 说) = 乂勺 (z) 十 — (z — 5)0'C) + const. ; (1 12.18)
it is seen that for a given function @(z) the displacements are defined
apart from a rigid body translation, as was to be expected.
If the rotation does not vanish at infinity, the arbitrary constant which in¬
fluences the rigid body rotation enters into the function (z) and in this way
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 475

into <p(z) on the right-hand side of (112.18). However, under the present condition
by which the rotation is to vanish at infinity, the function (z) is uniquely de¬
termined for any given state of stress of the body.
Note still a formula which follows from (1 12.17) and (3 1 .4) :
eu
——F
--
au
丁一 =中 一 中(5) + (z — —— +
.
const. (112.19)

It will be noted that it follows from the definition of the function

(z) in the upper half-plane and from the conditions (1 12.4) and (1 12.4')
which, of course, refer to the lower half-plane that the conditions (1 12.4)
also hold for the upper half -plane.
3°. In the sequel, it will be assumed that (z) is continuous from the
left and from the right at all points t of the real axis, except possibly at a
finite number of points 友，％, ..., which will always be specially
mentioned (so that no such points will exist, unless stated otherwise)
and that
lim 夕①'(z) = lim y^r(t 十 初 = 0 (112.20)
期f 0 3/->0
for any point t of the real axis, except possibly for tlf t2, . . tk.
Further, it will be assumed that near the points 幻，mentioned above,

卬(z) |=| |
——
—- tj
Iz |
0 <a< 1, (U2.21)

in other words, extending in a natural manner the definition of § 106

to the case when the line of discontinuity extends to infinity, it will
be assumed that the function (z) is sectionally holomorphic with the
real axis as line of discontinuity.
The preceding conditions ensure that the stress components are
continued continuously at all points of the boundary except, may be,
at the points t^, . . .,tk (including the points introduced in order not
to lose solutions important from the point of view of applications) .
In addition, these conditions are readily seen to ensure that the dis¬
placement components are continued continuously as well as the ex¬
pression (1 12. 19) at all finite points of the boundary without exceptions.
It will be recalled that, if the expression (112.19) is continued con¬
tinuously on the boundary, this includes the presence of concentrated
forces applied to the boundary (§ 43).
Finally, note that the adopted conditions ensure that the derivatives
u' , are continued continuously at all points of the boundary except
476 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 113

at the points 九%，… •，力廿 For this it is readily seen that the derivatives
with respect to t of the boundary values of 勿，o are equal to the boundary
values of the derivatives ur , i.e.,

(112.22)

by (112.20).

§ 113. Solution of the first and second fundamental problems

for the half -plane. The solutions of the first and second fundamental
problems were given in §§ 93, 94; new solutions will be deduced here
on the basis of the formulae of § 1 12, thus exhibiting their simplest ap¬
plication.
1°. First fundamental problem. In this problem the
external stresses are given, i.e., the pressure 尸(£) = —
Y~ [denoted in
§ 93 by ~ and the tangential stress T0) = X~, applied to the
entire boundary Ox which will again be denoted by L. It will be assumed
that PR) and T«) satisfy the H condition on L, including the point at in¬
finity, and that they vanish for 力 = co.
By ( 11 2. 1 4) , the boundary condition takes the form

— — + 汀Q),
+(£) -Q) = P(£) (H3.1)
because for z, tending to t from the lower half-plane, the functions
R),
by (112.19).
-> +Q), while (z
—
5)*(2) = 2iy^(z) tends to zero

Strictly speaking, (112.19) only ensures the limit 夕① '(z) f 0, when zf £ along
the normal to the boundary. However, it is readily verified by means of a simple
estimate of the derivative of a Cauchy integral, stated in the Author's book [25],
that the final result gives a regular solution of the problem.

A solution of the boundary problem (113.1), vanishing at infinity,

may be written down on the basis of the results of § 108, where only the
case of a finite line L has been considered ; however, the applicability
of the relevant formulae to the present case is obvious. In fact,

(z) = —J/ 4力.IJ.Z)

2kz
L
t — z
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 477

The problem is thus solved, because (z) determines the components

of stress and displacement by the formulae of §112.
The solution derived here coincides with the solution obtained in
§93 (cf. Note 1° in that section) .Regarding the behaviour of at
infinity, see the remarks in § 93.

2°. Second fundamental problem. In this problem the

boundary values of the displacement components u, v are given, i.e.,
贫一 = gi«), 厂 = 取(£) on L. It will be assumed that the given functions

gi© and g2(力) have derivatives g；(£),g；0), satisfying the H condition on L

including the point at infinity, and that they vanish for £ = oo.
The boundary condition
%-
+ 说- = gi(£) + 讴2(看) on L (1 13.3)
gives, after differentiation with respect to t,
("-)' +〃厂y = g；R) + 温 Q)
or, by (112.22),
力-
+ 谢" = g；Q) + 漏的. (113.4)
By (112.15), this condition takes the form
+Q) 4- 乂①-Q) = 2证g；(E) + 温(切 on L. (H3.5)
The validity of the assumption u'~ = (〃-)', v'~ —
®—)' under known conditions
referring to the given functions g*), g2&) may be verified after the solution has
actually been constructed.

For the time being, denote by Q(n) the following sectionally ho¬
lomorphic function : Q(z) = in S+, Q(z)=— k in S~. Then
(1 13.5) takes the form

-
Q+0) QR) = on £, (H3.6)
and hence, as in the preceding case,

Q(z)= 上
m J
L
/1也士逊
t ——
Lz
丸 (113.7)

so that, finally,
for y > 0,
f Q(z)
(z)= 1 (113.8)
Q(z)foryvO,
478 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 14

where Q(z) is determined by (1 13.7). For the investigation of the be¬

haviour of this solution at infinity, cf. 1°.
The sectionally holomorphic function Q(z), introduced in the above manner,
may again be denoted by (z); one then obtains an extension of the original
function (z) into the upper half-plane, different from that stated in the preceding
section. This new method of extension is characterized by the fact that the values
of (z) in the upper half-plane analytically continue the values in the lower half¬
plane through those parts of the boundary (if such exist) where 0 —
The solution is also easily deduced when one begins with (112.18) ;
however, the above solution is more convenient, because it does not
require an additional investigation, arising from the fact that 中 (z) is
not holomorphic at infinity, but behaves there like log z, unless a further
condition, restricting generality, is imposed, as was done in § 94.

§ 114. Solution of the mixed fundamental problem.

If there is only one segment of the real axis on which displacements are given,
i.e., if in the notation of the main part of this section % = 1, the problem is fairly
simple ; however, no effective solution of this problem was obtained in the Author's
paper [20], published in 1935. Two years later, V. M. Abramov [1] gave a more
effective solution of this case, using Mellin's integrals. The solution of this section for
arbitrary n which was first presented by the Author in his paper [22] is incom¬
parably simpler. A somewhat more complicated solution which is in essence closely
related to that studied here was deduced soon afterwards by N. I. Glagolev [1,2]
who was not acquainted with the Author 飞 work [22].

1°. Let £' = £] + L? + ... + Ln be the union of a finite number

of segments akbk of the real axis 0%, numbered in such a way that they
are encountered in the order ar, 力】， a2, b2, . . . , ant bn when moving along
the real axis in the positive direction. Let the components of displacement
be given on L' and those of the external forces on the remaining part
of the real axis. As before, it will be assumed that the elastic body
occupies the lower half-plane its boundary, i.e., the real axis, will
be denoted by E = 27 十 L" .
Since one knows how to solve the first fundamental problem, the
mixed fundamental problem under consideration may obviously always
be reduced to the case where the components Yy> Xy of the external
stresses, given on L" , are zero. Hence it will be assumed that the part
L" of the boundary is free from external forces, i.e., that
= X, = 0 on L'r . (1 14. 1)
(The solution of the general case may also be deduced directly ; cf. the
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 479

Note at the end of this section).

In view of the considerable practical importance of the present problem,
the ordinary mixed problem in the form, formulated above and to be
called Problem A , will be studied side by side with a somewhat modified
problem, to be called Problem B.
In both these problems the boundary condition on has the form
u~ 谢一 — + c® on Lf, (1 14.2)
where gQ) = g*) 吆2«)is a function given on L' .
In Problem A, it will be assumed that c(£) = c = const, on Z/ and that,
in addition, the resultant vector (X, Y) of the external forces, applied
to Lf, is known. Without affecting generality, one may, for example,
put c = 0, because its value only influences the rigid body translation.
In Problem B, it will be assumed that c(力) = ck on Lk, where the c 把
are constants which are intially unknown and which are, in general,
different on different segments. In this case it will be assumed that the
resultant vectors (X左， YJ of the external forces, applied to the individual
segments" are known. Without affecting generality, one of the constants
may be fixed arbitrarily, i.e., one may, for example, put q = 0. For
处 = 1, the problems A and B coincide.

The physical meaning of these problems will now be explained. One

may imagine that rigid stamps with given profiles are placed on the
segments Lk = 左, that the points of the segments Lk of the
boundary
of the elastic body are brought in a definite manner into contact with
the points of the profiles of the rigid stamps and that the same points
remain (or are welded) together. Further, suppose that given forces be
applied to the stamps and that the stamps can only move vertically. The
problem of finding the equilibrium of the elastic half-plane under these
conditions is Problem A, provided the stamps are rigidly inter¬
connected; if the stamps can move vertically, independently of each
other, one arrives at Problem B.
For greater understanding, consider the following particular case.
Let there be only one stamp the profile of the base of which before contact
with the elastic half-plane is given by y = /(%), 口 《 b. Further,
assume that the stamp is pressed into the half-plane by a given force,
perpendicular to the boundary, and that the friction between the stamp
and the elastic body is so great that no slip can occur. Then, assuming that
the segment V = ab of the boundary of the body has come into contact
with the stamp, one obtains the boundary condition (1 14.2) in which
480 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 114

g«) = giR) + 很2々)，where g*) = 0, g2(t) = /(/) and c(力) = c is a constant

which may be put equal to zero. The case of several stamps, whether
they are interconnected or not, is quite analogous.
These problems represent idealizations of the problems of the pressure of founda¬
tions on the ground in the presence of sufficiently large friction (or, more correctly,
cohesion) , since slip and after-effects are excluded.
The above problem may be generalized in the way that the stamps
are not only permitted to move vertically, but also to rotate ; this case
will be discussed below in 4°.
It is easily shown that neither of the problems A and B can have more
than one solution, neglecting rigid body motion of the entire system
(elastic half-plane and stamps). In fact, the proof of § 40 (cf. also end
of § 90) may be repeated almost word for word, if it is noted that in the
present case the integral of the expression (Xnu + Ynv)ds along the
boundary of the region for the difference of two possible solutions is zero.
Actually, one has on the part Ln of the boundary : Xn = = 0. Further,
for Problem A, it may be assumed that the difference of the solutions
% =
“ = 0 on the boundary ; however, for Problem B (where for the
difference of the solutions u = const., 力 = const, on EQ, all the integrals

/ {uXn + vY^ds =
%/Xndt + /Yndtd

Lk
= 说羽 + °%

“ , Lk
vanish, since the resultant vectors (X 兀, Y%), applied to Lk, are zero for
the difference of two solutions.
In the sequel, the uniquess theorem will also be applied to cases, where
the stress components are continued continuously at the points akf bk.
In such cases, as already noted in § 40, 3°, the uniqueness theorem
remains in force, if the integrals of the expression (Xnu 十 Ynv)ds,
formed for the difference of solutions, when taken in S along infinitely
small semi-circles about the points ak, bk as centres, tend to zero together
with the radii of the semi-circles. In all the cases considered below this
condition will be fulfilled, as the reader may easily verify in each indi¬
vidual case.
It will be assumed below that the function (z) satisfies the conditions,
stated in § 112, 3°, and that the ends 町，b5 of the segments play here
the parts of the points 叩. In addition, the known function g«) is to have a
first derivative g'Q), satisfying the H condition on L'.
In both problems A and B the boundary condition (114.2) gives
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 481

(弘一)' + (说 — = g'Q) on L', whence by (1 12.15) [cf. (112.22)]

<!>+(£) + 乂①-(£) = on L' . (1 14.3)
However, the condition (114.1) is equivalent to the requirement that
+(力) 一 一(力) 0 on L" , i.e., that (z) should be holomorphic in the
entire plane, cut along Lf, and this will now be assumed.
The condition (114.3) differs essentially from (1 13.5) in that now U
is only part of the Ox axis.
A solution of the boundary problem (114.3), vanishing at infinity,
may immediately be written down on the basis of (1 10. 18). In the present
case, by (110.5),
log ( — x) log % 1

or
Y = 2
—— 码,
where 0 is the real quantity
logx
I _ _ _ __
(114.4)
2H
which was denoted by (— 0) in § 1 10.
Therefore, by (1 10.2),
n
xo(z)= n 3-外尸+诏k一狐尸饰.
k=l

In future, X^z) will be denoted by X(z) and X+(£) by X(%)・ Thus

n
X(z) = A;n
=1
(z — 瞅) (z — (114.5)

and
n
X(£) = X+«) = n (/ — 曲) Q — 狐厂A% (114.6)

where (114.6) refers to the value, taken by X(z) on the upper side of
the Ox axis (i.e., from the left of 0%); if the point t lies outside L' , i.e.,
on L'r , the values from the left and from the right coincide : XR)
X+Q) = X(力)； however, if t lies on U , one has, by the definition of the
—
function X(z), X+(t) 十 xX-修) = 0, whence
 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 14

X-Q) = — ^X+(力) = — Lx«).

比 比
(114.7)

Applying now (1 10.18), one obtains

(0 = 号'J 邛If片 L'

十 X(z)Pi(z), (114.8)

where Pn_^ is the polynomial of degree not higher than n —1

P*i(z) = CoZ"T + C1/-2 + …十 C*1， (114.9)
because, by supposition, the function 0(^) is to vanish at infinity.
The coefficients Co, C〔，..., of the polynomial have still
to be determined from the supplementary conditions of problems A and B.
2°. Problem B will be considered first;in this case the known resultant
vectors (Xk> 丫欠) serve as supplementary conditions. In order to express
these, the normal pressure P = —
Y~ and the tangential stress T = X~,
acting on the boundary of the half-plane at the stamps, i.e., on U, will
be calculated.

-
By (112.14), one has for the point t0 on Lf
P(幻 + "㈤ +(幻 一 幻 14.10)
or, by (114.3),

尸伉) + iT(Q=-^—^Q
尤
一^~

x
^g'(幻 on ZA (114.11)

Applying to the right-hand side of (114.8) the Plemelj formulae, one

easily finds

+(幻 = 网'(幻
从
J
兀沙 J
V
e⑺ 心
X(%)P*iRo)；

substituting this value of in (1 14.1 1) and writing

g'M
(114.12)
z
k J x(f) 一 幻'
one obtains
N(乂 一 1) x 4- 1
P&) + -—— X
lg隔)十二^%(幻+
X
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 483

比 —1~ ]
+ x
X(M)Pi 仇) on L. (114.13)
By expressing that

/ [p(6 + "(%)]a% = —
^ + 2n，狗=1,2,...,外,(114.14)
one deduces a system of n linear equations for the determination of
the n constants this system has a unique solution, as may be seen
from the uniqueness of the solution of the original problem.
3°. Next consider Problem A. Since the solution found above satisfies,
the condition 〃'一 + 谢 '一 = u~f 十 谢 = g'«) on L' (as is easily verified
by substitution) , one has on the segments Lk
，厂
+ 谢- = g(£) + ck,
where the ck are constants. One has now to formulate the condition
% = = . . . = c 驾； (1 14. 15)
having succeeded in satisfying this condition, one can also fulfill the
condition = c2 —... = % = 0 by adding an arbitrary constant to
the right-hand side of (112.18).
In order to express the condition (1 14.15), the value of u'~ 谢 '一 will
be determined on the unloaded part L" of the boundary. By (112.15),
one has for a point 片 of L"
2浊(心 + 谢 = (乂 + 1) (幻 (尤 1)/(%) + (x + l)X(%)Pi(%),
(114.16)
where Qo(/o) is given by (1 14.12) and 力 is now on Ox outside Z/ (i.e.,
on E"). Obviously (114.15) reduces to° the following conditions :

g(即+i) — = (力一 + 说'-)欣o, 万 1,2, . . 笈一 1. (114.17)

bk
Substituting in (1 14.17) for %'- + iv'" from (114.16), one obtains a
—
system of n 1 linear equations for the determination of the C,. One
more equation is given by the condition that the resultant vector of
the external forces (X, Y) applied to L' is known. This last condition
is most simply expressed, using the first of the formulae (112.4) which
gives
484 VL APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 114

lim Z^(Z)
X + iY (114.18)
4f8 2k
Applying (1 14.18) to (114.8), one deduces directly that the coefficient
Co of zn~x in the polynomial Pn-^ is given by
X + iY
(114.19)

Thus only the coefficients Clf C2, . . . , have still to be determined

—
from the above system of n 1 linear equations. This system will
always have a unique solution, as in the preceding case.
—
In the particular case g'(/) 0 (stamps with straight profiles, parallel
to the axis 0%), the formulae (114.8) and (1 14.13) become very simple,
since the integrals vanish.
4°. Hitherto it has been assumed that the stamps may only move verti¬
cally (i.e., at right angles to the boundary) . The case will now be con¬
sidered, where the stamps may rotate (of course, in their own plane) .
In Problem A (the stamps being rigidly interconnected), let e denote
the angle of rotation of the system of stamps, measured counter-clock¬
wise. Then in the boundary condition (1 1 4.2) the function must be
replaced by g«) + izt and hence in all the subsequent formulae g'(£)
by g'(£) 十 区. Correspondingly an additional term, involving e as a
multiplier, will appear in the expression for (z). (This additional term
may be calculated in finite form ; cf. example 2° of § 1 1 4«) .
The additional displacements (%,%) of a point t of the boundary which are
caused by the rotation of the stamps are given by w0 0, — = E, because, in
general, the displacements, arising from a rigid body rotation by an angle e about
the origin, are

The quantity s
=
—may not be given directly ; for example, one may be
吵，v0 ~ zx, while on the boundary 夕 = 0, # = %.

given instead the resultant moment M about the origin of the ex¬
ternal forces which act on the stamps.
These external forces do not, of course, coincide with those, applied by the
profiles of the stamps to the sides of the elastic body ; these last forces must balance
the external forces, applied to the stamps. Obviously, the resultant vector (X, V)
and the resultant moment M of the external forces is equal to the resultant vector
and moment of the forces, applied to the boundary of the elastic body by the faces
of the stamps.
One thus has for the determination of e the additional relation

M=
__ / /
0P (114.20)
L
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 485

In Problem B, the angles 泳岱 = 1,2, . . . , of the rotations of the

various stamps may differ ; if they are not given directly, but if, say, the
moments Mk oi the external forces, acting on the individual stamps, are
known, then one will have the n additional conditions for the deter¬
mination of the 宫九
加右 = 1

~~ (114.21)
Lk
It is easily shown that these conditions completely determine the
solution apart from a vertical rigid body displacement of the entire
system ; the proof is quite analogous to that stated above for the case
where the stamps may only move vertically.

NOTE. If the part Lr, of the boundary is loaded by given externa]

stresses, the boundary condition takes the form
+(%) — ' 々① i(£) = /(9 on L, (114.22)
where
力 = — 比 or //, = 1 on E", (1 14.23)
and /(/) is a given function :
/(£) = 2阳常) on Z/, /(0 = P(t) + 江 ©on (1 14.24)
where P(£) and 丁伍) denote, as in § 1 13, the pressure and the tangential
loading ; it will be assumed that 尸(力) and T(t) satisfy the H condition
on Ln including the point at infinity and that they vanish for / = 8.
One is thus led to the problem, considered in § 111. Applying (111.6) *,
one obtains in the notation of the present section

= 芈J A E (方 一 z)
+ X(z)尸一 (H4.25)
L

where the integral must now be taken over the entire boundary. The
coefficients of the polynomial Pn-S^) may be determined as before.
On the behaviour of (z) near the point at infinity cf. § 93 following
(93.7).

* This formula was derived under the assumption that L is a closed contour.
However, its applicability to the present case is obvious.
486 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 14a

§ 114a. Examples.
1°. Stamp with straight horizontal base.
Consider the case of one stamp (” = 1) with a straight line profile,
parallel to the axis Ox, which may only move vertically so that
= 0 on (1 14.⑷
as stated earlier, this problem was solved by V. M. Abramov [1], using
a quite different method. In addition, it will be assumed that the ex-
ternal forces, acting on the stamp, have a resultant in the downward
direction, so that
X = 0, Y = Pq, ―
(114.2a)
where is a given positive constant.
The segment L' of the boundary which is in contact with the stamp
will be assumed to lie symmetrically with regard to the origin ; its length
will be denoted by 2Z, so that for points of Z / Z.
Thus, in this notation and that of the preceding section,
—
X = (z + Z尸+律(z — /『IB, (1 1 4.3a)
and (114.8) gives

or, using (1 14.19),

-
iP ip
(z) =— 2H
(z) =—
2
- (z + Z)T+吗z
大
— %)+叱 (114.4a)

and the problem is solved.

The pressure PQ) and the tangential stress acting on the body
underneath the stamp, are given by (1 14.1 1) which becomes
J— 1
P(E) + ETQ) = —
比

x
+(£).

Hence, substituting from (1 14.4g), one obtains

P0 + iT 出 = =

=
_
2

2k
P
2k
v —I— 1

比
x

%尸+单 0 — (1 14.5〃)
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 487

where X(力 stands for X+(£), i.e., the value taken by X{z) on the left side
of the segment (— 2, 十 2), and it should be remembered that the branch
of X(z) is determined by the condition HmzX(z) = 1.
gf8

It is easily verified that under these conditions

_ 律

where the root V/2 —— t2 is positive and the logarithm real. Since
log x
8== e =S，
f

one may still write

"x 那 log
/八
= ,.二二_„^^6 "=
HP — 护

+ i sin
空｝(114.6g)
—
( Substituting this expression in (114.5© and separating
real and imaginary parts, one obtains

(11 4.7昉

(114.8a)

These formulae agree with those, obtained by V. M. Abramov [1].

It follows from (1 14.7a) that PR) changes its sign an infinite number
of times as / approaches the values —I and 十 2, so that effectively
tensile forces, instead of compressive forces, act on certain parts of the
boundary underneath the stamp. However, it is easily seen that these
parts lie in very small neighbourhoods of the ends of the segment
十 I. In fact, the point t at which PQ), positive for / = 0, vanishes for
—
I,
488 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 14a

the first time, when t approaches one of the values 土 ％, is determined

by the equation
2+/ 加
3 log-}4 r = ±方
J /

whence

% = ± Z tanh 2 log—x .
However, for all actual bodies, 1 < x < 3, since

-
therefore the smallest possible value of | t \ is obtained by putting in
(114.9g) k = 3 which gives
t 土 0.9997Z.
Thus a change in sign of PQ) only occurs in those places near which the
solution obtained does not, in general, describe the actual state of the
body, because, obviously, Hooke's law does not apply for the stresses
which must occur according to the formulae above.
2°. Stamp with straight inclined base.
Consider the case of the same stamp as in the preceding example,

O +1
Fig. 52.

but assume now that the resultant vector of the external forces, acting
on the stamp, is zero, while the base of the stamp forms an angle &
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 489

with the Ox axis, the angle being measured in the positive direction
(Fig. 52).
Thus, in the present case,
gW = gi + 电2 = 热， g'Q) = 说 (1 14.10a)
and
X=Y = 0. (114.1U)

Hence one has, by (114.8), in the notation of the preceding example

__
=
wX(z) / 近
(114.12a)
XQ) (/— z) '
J
-z
where X(z) is given by (1 14.3〃) and XQ) = X+Q).
On the basis of Note 1 at the end of § 1 10, the integral may be evaluated
in closed form. In the present case, for large | z
l \6华 / i\
g —— 2)"汨= z (1-R
~~ = (z + l+R =

(I — 不)—F Z
•
J 11
《
1

修 斗) —F
Z

and (1 10.40) gives

C 就
J 1
z -j- 2唱2
J %■(£)(£ — z)- = 乂 1 i X
-z
Hence (1 14.12«) becomes
眄
= X 1 +
口 — (z — 制)X
2 (114.13«)

and the problem is solved.

The value of P&) + iT(t) will now be calculated for a point t underneath
the stamp. One has
490 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 14a

产(力)
+ 2?g = +传) 一 一(f) = ——x 十 1 Q — 仅){X+(f) — X-e)} =

=一工(力一2制)9k+(力),
x 1 + 乂

-
i.e., in the present notation,

尸(力) + iT© — 空巴 — 2泓)X@.

- (i 14.14«)
x
Substituting for X«) from (114.6a) and separating real and imaginary
parts, one may obtain closed expressions for P(£) and TQ).
Hitherto it has been assumed that the angle s is known and, con¬
sequently, that the stamp sustains in the given position some couple of
the external forces which is not known beforehand. However, the problem
may be stated differently ; in fact, it may be assumed that the moment
M of the couple of the external forces, acting on the stamp, is given and
that it is required to find the corresponding angle of tilt e.
For this purpose the moment M,corresponding to a known angle %
will be calculated ; this moment is given by

M= — tP©dt /
-i
(114.15«)

(the positive direction of rotation being assumed to be counter-clockwise) ,

where P传) has the value, given by (1 14.14).
The integral ( 1 1 4. 1 5口) is the real part of the integral
i

7= — / "PQ) + iT^dt, (H4.16G

which is readily evaluated in closed form. In fact, substituting from

(114.140, one finds
i

7 = 3^
/
-i
即 一 2渺)X(f)然 (114.17«)

This integral may be calculated by the same method as the integral

in (1 10.42). Consider the integral
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 491

= 羹 — 2津)X(Q落
J
A
(1 14. 18仇)

taken in counter-clockwise direction over a contour A surrounding the

segment E'( - Z
one obtains
— ,,
Z). Shrinking the contour A into the segment Lf,
?

a = /行― 2z•初X+Q)近 抱 一 2淑)X-(%)也

-Z +Z
or, remembering that XR) = — (1/x) X+(力 = — (l/乂) X«),
+1
1 _l
x r
Iq = 力/ t[t —
T
and hence 2阳.
—
On the other hand, one has for large | C |
八T+印/ 八-Ib
w)=7(1+J (1-7)
1 /
=-

whence it follows that the coefficient of in the expansion for

进 一 2津)X(Q is equal to 庐(1 + 4伏)
'
2
Therefore, applying the residue theorem to the integral in (114.18g),
one finds
，o —
= *(1 + 4俨)卢,
and hence
2^(1 + 4俨)岸
X + 1
492 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 15

It is seen that I is real and thus M ~

= I, so that
2^（1 + 4俨）岸 （114.19«）
%+ 1

For a given moment M, the angle of tilt s of the stamp will be determined
by
x + i
27Tp.（1 十 4俨）庐
M. （114.20〃）
Substituting this value of s in （114.13〃）， one obtains the solution of
the problem of the equilibrium of a stamp, subject to a given couple.
3°. Effect of asymmetrically distributed forces
Let asymetrically distributed forces act vertically downward on a
stamp with a straight base which is not restrained to move vertically.
The effect of these forces is equivalent to that of the same forces, applied
symmetrically, and of a certain couple. Hence the solution of this problem
will be obtained by adding the solutions of the problems 1° and 2°,
treated above.

§ 115. The problem of pressure of rigid stamps in the absence

of friction.
This problem was first solved by M. A. Sadovski [1] for one particular case.
In the second edition of this book, the general solution for the case n = 1 was
given, which was generalized by A. I. Begiashvili [1] to the case of arbitrary n.
The much simpler solution, reproduced here, was given by the Author in his paper
[22] ; simultaneously（and independently）A. V. Bitzadze found a solution which is

Consider now the problem of pressure of one or several stamps on

the boundary of an elastic half-plane under the supposition that there
is friction. For greater clarity, it will first be assumed that there is
only one stamp with a given profile. Let y = /（%） be the equation of the
profile before it is pressed into the elastic half-plane. If it is assumed that
the stamp may only move vertically, i.e., in the direction normal to the
boundary, then its profile, after pressure has been applied, will have the
equation y /（%） 十 c, where c is a real constant. It will be assumed that
工

the segment ab of the boundary of the body comes into contact with the
stamp. Since a point of the elastic body, occupying before deformation
the position （力 0） and after deformation the position Q + u, v）, where
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 493

%,0 are the components of its displacement, must lie on the liney = /(%) + c,
one must have e = /(/ 十 ％) + c. Assuming, as always, that u, v are
small quantities and that /(%), /'(%) are likewise small (this supposition
being a consequence of the requirement of small deformations), one has,
omitting small quantities of higher order, o = /(/) + c (« Z b),
where v = v~ is the normal displacement of points of the boundary of
the elastic half -plane. The reasoning for the case of several stamps is
quite analogous.
Correspondingly, the boundary conditions for the problem of pressure
due to a system of stamps which may only move vertically and are
completely frictionless may be formulated similarly to the conditions
for the problems of § 1 14, the only difference being (using the notation
of §114) that now
X) = 0 everywhere on L, = 0 on L,f, (1 15.1)
while on only the normal component of displacement
Lf
厂 = /Q)
+ c(, on Lf (1 15.2)
is given; as before, L denotes here the entire real axis, L' the union of
segments Lk = a* 便 = 1, . . 加) and L" the remaining (unloaded)
part of L. The first of the conditions (115.1) applies to the whole boundary
L, since in the absence of friction the tangential stress at the boundary
is also zero underneath the stamps.
In (115.2), the function /0), given on L\ characterizes the profiles of
the stamps, and, in fact, y = /(%), where rv belongs to represents the
equation of the union of the profiles of the stamps before their displace¬
ment ; cQ) is determined as follows: either cQ) = c on 7/ (rigidly inter¬
connected stamps) orc(f) = ck on Lk = 阪6 后 (free stamps), where c and ck
are now real constants. Without affecting generality, one may assume
in the first case that c = 0, and in the second case, for example, & = 0;
the remaining constants will not be known beforehand.
In the first case, the resultant vector (0, Y) of the external forces,
pressing the system of stamps into the elastic body, will be given, and
in the second the resultant vectors (0, 匕) will be known separately
for each stamp.
It had been assumed that the stamps may only move vertically ;
the case, where they may also rotate, may be reduced to the preceding
one in quite the same manner as in §§ 1 14, 114〃. In that case one must,
in addition, be given either the angles of tilt of the stamps or the re¬
sultant moments of the external forces, acting on them.
494 vr. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § I 15

It is easily shown that the problems, as formulated above, can only

have one solution, neglecting vertical rigid body displacements of the
entire system ; the proof is analogous to that for the case of § 1 1 4.
It should still be noted that a translation of the stamps, parallel
to the boundary L, has no influence on the elastic equilibrium (within
the accuracy to which one is always restricted) .
It will be assumed that /'(/) satisfies the H condition on each of the
segments Lk = ag.
As in § 1 14, the boundary conditions (115.1) show that (z) is holo¬
morphic in the plane, cut along U. In addition, it is easily deduced
that the first of the conditions (115.1) gives by (112.14)
+Q) + +(3 = 一(9 + 回 everywhere on L,
whence it follows that the function (z) + (z) is holomorphic in the
entire plane ; further, since it vanishes at infinity, it must vanish every¬
where. Consequently
石 (z) = — (115.3)
By (112.14)
— = -(/) —① +&),
whence, going to the conjugate complex expression and remembering (§ 112) that
R) = +(力， e>+(/) =不一(e),
y
7+ = 3+(/)一 不一团；
subtracting these equations, one obtains

— 2iX~

and the above statement follows.

= -Q) 币-(/)
— +Q) —不

On the basis of this result, the boundary condition (1 15.2), which

will now be written [cf. remarks following (1 13.5)]
L =『(/) on L', (115.4)
gives by (U2」5)*
呐什加” 皿
x 1 十

Thus one has arrived at the boundary problem of § 1 10, and, in fact,
—
at the particular case, where g = 1.' Applying (1 10.31) and (1 10.33),
* Cf. also (112.22).
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 495

one finds
2M 「X©网dt iPn_g
= 而不承 —yzrl a/ L'
' 0 66)

where iPn^z) denotes an arbitrary polynomial of degree not higher than

—
n 1 and
X(z) = 7(z — 旬) (z — “) . . . (z — aj (z — bj; (115.7)
here X(z) is the branch, single-valued in the plane cut along L', for
which zf X(z) fl as z —a co. In future, X+(£) will be simply denoted by
X(£) so that, by definition,

X(方) = "(/ b鼠… 二点) — X+(力)； (115.8)

note also that
X- — X+(£) = —XQ). (115.9)
The condition (115.3), i.e., —
= O(^), has still to be satisfied.
It is readily verified that the first term on the right-hand side of (i 15.6)
satisfies this condition ; the second term will satisfy it if, and only if,
all the coefficients of the polynomial Pn_-^ are real.
The first of the preceding statements may be proved as follows. Denote, for
the time being, the first term, of (115.6) by o(z), i.e.,

2tx f X("
0(Z) —— 兀(乂
：— 八 y八 /J
L'
£ -~ z
*
，

remembering that, by definition, o(z) = o(2), one obtains

拓 2p. r 丽片力成
° + l)N(z)
L'
J t —— z *

because /'(/) is a real function. It is easily seen that X(z) = X(z), because by (1 15.7)
X(z) represents the same root as X(z) and doubt may only arise with regard to its
I

sign : X(z) = 土 X(z); it is seen from the behaviour of X(z) and X(z) at infinity
(both functions behaving for large | z [ like zn) that the upper sign must be chosen.
Finally, on the basis of these results and of (112.10〃)，

and hence
邱) = 'x+ = = x-(f) = — X(X),

o(z) = o(z).
496 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 16

The second statement follows because, as has just been shown,

文⑶ = X(z).
Thus the general solution of the original problem is given by (115.6),
where
+ D甲-2 十 ... + Dn_r (115.9)
must be a polynomial with real coefficients.

_
The pressure P(/), exerted by the stamps on the boundary of the half¬
plane, will now be determined. By (112.14),
P® = y- = o+W — 0-W， (H5.10)
whence follows, applying the Plemelj formula and remembering that
X—&)=_X+(%)=_XR),

(o) ( )
标TE瀛 J 丁 Fof—.
二

E，

The coefficients D$ of the polynomial Pn^z} will be determined from

the additional conditions, stated above when formulating the problems
in the same manner as in § 1 14.
For example, consider the case where the resultant vectors (0, Yfc)
of the forces applied to the stamps separately are given. Then

—Y 卜 = k = 1,2, . . . , n, ( 1 15. 12)

where P[t^ is given by (115.11). One thus obtains a system of n linear

equations in the unknowns Do> Dlf .. . , Dn_\. It is easily shown, based
on the uniqueness of the solution as a whole, that this system of linear
equations also has always a unique solution.
A method, completely analogous to that studied here, leads very simply to
the solution of a problem, connected with the investigation of the stresses in a
stratum above layers of coal; this problem was stated and solved (using more
complicated means) by S. G. Mikhlin [12].

Hitherto, it was assumed that the stamps may only move vertically.
As stated earlier, the case, where the stamps can rotate, is easily reduced
to the preceding one.

§ 116. Application. The solution, obtained in the preceding section,

will now be studied in somewhat greater detail. To simplify the in-
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 497

vestigation, the case of one stamp will be considered which comes into
contact with the axis Ox along one continuous segment 〃占；the general
case may be considered in an analogous manner.
1°. In this case one will have, instead of (1 15.6) and (1 15.1 1),
b

0⑶ - -
k(x + 1 )
3
M(z —— — J
a)(b —
z)
/
a
/ — z +
D , 、

and
0
(z
— a) 他 — z)

p⑷
_ 如
十
tc 低 + 一 a）仅 一 公 J f
2D
+ •7— =， (116.2)
a/ 数)仅 一 曲) —
where is a real constant. In these formulae, the function X(z)
Z) —
— —
= “(z a) (z ，) has been replaced by the function a/(z 夕) (6 — z)
[cf. Note 2 at the end of § 1 10) and, as a consequence, the expression for
—
PQ) becomes real. For a <t <b, the root a) {b ― 力 must be —
——
taken as a positive quantity, while a/(z g) (b z) must be taken as
the branch, holomorphic in the plane cut along ab and taking positive
——
values on the upper side of ab. This branch is easily seen (cf. Note 2
§ 1 10) to be characterized by the condition that for large ] z |

— a) (6 — z) = — iz + 0(1). (116.3)
The constant D is determined by the condition
b

/
a
PW = Po， (116.4)

where Po is the given magnitude of the forces, applied to the stamp,

and PQ) is given by (116.2). The constant D may be determined in
a simpler manner by noting that it follows from (116.1) and (1 16.3)
that for large | z [
=?+ 0
498 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 16

whence, comparing with (112.4),

D= 员. (116.5)
2k

In order that a solution may be physically possible, one must

obviously have P«) 0 (for 〃《力 b). Thus, after the solution has
been obtained, it must be verified whether this condition is satisfied.
It will be assumed for the solution of the problem that the segment
ab of contact between the
stamp and the elastic half¬
plane is given beforehand.
This corresponds, for exam¬
ple, to the case where the
stamp has the form shown
in Fig. 53 and where the
force, applied to the stamp,
is sufficiently large to ensure
that the corners A and B of
the stamp come into contact
with the elastic body. The
presence of the corners A,
B also explains the occurrence of infinitely large stresses at the points a, b
of the elastic body which coincide with the corners A, B of the stamp.
2°. Considerable interest attaches to the case, where the rigid profile.
pressed into the elastic half-plane,
has no corners (e.g. circular disc)
or where the force is not sufficient¬
ly large for the corners A and
B to come into contact with the
elastic body, as in Fig. 54. In
that case the ends a and b of
the region of contact are unknown
beforehand. However, the formu¬
lae obtained above also permit
solution of this problem. In fact,
Fig. 54. the general formula (116.2) for
the presure P&) underneath the
stamp now contains the two constants a, b which will not be known
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 499

beforehand. [The constant D is given by (116.5).] For the determination

of these constants one has the two relations
P[a) — 0, P(b) = 0, (116.6)

which express the condition that P(%) tends to zero continuously as t0

leaves the area of contact. This condition may be replaced by the more
general one (which, in addition, is physically more obvious) that P(t^)
remains bounded near the ends a, 6, provided these points are not corners
of the profile of the stamp. In fact, it is seen that the condition of boun¬
dedness of P(%) near a, 6 entails also the relations (116.6).
In order to express the condition of boundedness of the pressure
P%) in the neighbourhoods of the points a, b, introduce temporarily
the notation
0Q) = 伍 .a) (b /)
― —
and write (1 1 6.2) in the form
b
4fl /• 。（ Q）以 , 2。
（o）一
（m +
兀 1）“西l
a
M丽（力一 %） M西
b

兀依 十 1） J （%— %）
+
a

b
如 — 0(%) 片方)也 , 2D
'
K(X + 1) 瓯 aJ ％ VW)」
or, noting that
" 力一

0传）— 。㈤
力… 力0
— —t — 十〃十 6,

，
in the form

0（幻 /e威 必 B + 20
p八— ，
「仇2行灯/宾丽二0十 a
M函）
500 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 116

where
b

〃
4fx r /'(/)成
m 乂 十 1 竹
b b
如 rtf{t)dt
十
+S
4俨侬
兀 (乂十 i) J ，前
q
兀
3+ i)J
a
Vwj
The first term on the right-hand side of (a) is not only bounded near
the points a, b, but it also vanishes there, as may easily be shown using
the bounds for the value of a Cauchy integral near ends which are
given in the Author's book [25]. Hence it is necessary and sufficient for
the boundedness of P(幻 near a, b that 4=0, B 2D = 0, or, by
(1 16.5), that
b b

/ /'(M
/' tr^dt _K+lp 6八
J \ {t
—
，
— 力) J Mg — 力 女
― 4以
a a

If the conditions (116.7) are satisfied, the formula (倪) gives

&

42 V7% — 0(6 — %) /* / . •
(口6 8〉

as has been stated earlier, this expression vanishes for = 〃，％ = 瓦

It should still be noted that it is readily verified by transforming
(1 16.1) in the same manner as (1 16.2) that, under the conditions (1 16.7),
b

(z)
V
兀
—
= 22 V G © 二2)
十 1) J / —
V(t — tz) (6 —•%)(/■ - z)~~~
(116.9)
a

It may be added that (116.9) as well as (1 16.7) could have been ob¬
tained by seeking a solution (for the particular case n = 1) of the boun¬
dary problem (115.5), which remains bounded near the ends, and by
applying the relevant formulae of § 1 10.
In that approach the first condition of (116.7) coincides with the condition for
the existence of such a solution, while the second condition of (116.7) expresses
that the coefficient of zT in the expansion of (z) for large | z | must be equal to a
given quantity, determined by the magnitude of the force applied to the stamp.
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 501

Thus one has for the determination of a and b the two conditions
(1 16.7) which, in general, determine a, b uniquely, provided the con¬
dition P{t) 0 underneath the stamp has been observed (cf. § 116〃).
Hitherto the stamp was restrained to move vertically. The case,
where it may tilt, can be studied in an anologous manner (cf. § 116g).

§ 116a. Examples.
1°. Stamp with straight horizontal base
In this case /'Q) =0 and (116.1) gives, using (116.5) and writing
a —— I, b —I, where 2Z is the width of the base,

=
2兀
0_ z2 6.闻

One finds for the pressure P(t) underneath the stamp, using the
formula PR) = +(力) 一 ~(9，
PQ) = ——
a/岸 一声
(116.2a)
兀

This solution was obtained (by other means) by M. A. Sadovski [1].

2°. Stamp with straight inclined base
Let e be the angle of tilt (cf. § 114g, 2° and Fig. 52). In that case
f ) = is and, by (116.1) and (1 16.5) (assuming d = 2, 方 = —
岭) =
k(x + 1)V/2 — z2 3
J
-i
三里
— t z
—
2k VZ2 — 冲

The integral on the right-hand side may be calculated in closed form

by (110.40). In the present case, g = ~ 1, = VZ2 — z?; since,
by (116.3),
Vl2 ~ Z2 = — 贬 ( 1 B 1 ( + …) =
. 那 .

formula (110.40) gives

+z

t
-z"
一 dt = 兀"也2
— z? + iz).
502 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §1 16a

Hence
= ,、
2{i£Z 困2 Pc , 、
(z) L 二 -TTT：| — 厂； .：「.；“77^7 . ( 1 ]6.3①)

"
1
+ 1)V/2 — z2 乂
2k 庐 一 z2

Using the relation PQ) = +Q) ~ one obtains for the pressure
P(t) underneath the stamp

尸(£) =
R 空 . (116.4«)
兀
— /2 (x + 1)VZ2 — /2

The solution will be physically possible, if PR)

i.e., if
0 for
—Z 《力 Z,

4 2%
0》
产
x+1
兀
& (116.5a)

The resultant moment of the external forces restraining the stamp in

the given position is likewise easily calculated by the formula

M=
—j
* T

In fact, applying the same method as in § 114口，2° or evaluating the

integral by ordinary means, it is found that

M= •
2
7td— & (116.6a)
M 十 1

3°. Stamp with curved base

Let the stamp be represented by a strip, bounded by the vertical
—
straight lines x ~ I, x = + I and by an arc AB of a circle with radius
R and convex downwards (cf. Figs. 53 and 54). It will be assumed that
the radius R is very large. This assumption is necessary, because small
deformations are being considered. With the usual degree of approxim¬
ation, one may write
, 岸
/G) =
2宣

This implies replacement of the arc of the circle by that of a parabola,

having the same curvature at the vertex.
Then (1 16.1) and (116.5) give, under the supposition that the entire
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 503

arc AB is in contact with the elastic body,

+z
力"律 一 拦
(z) = I/* drt 十
R兀(比 十 1) V/2
— 22 J
-E
/ t z

The integral on the right-hand side may again be evaluated in closed

form ; in fact, since for large ] z |
渺 1/ \
zV /2 — / = — 十 0 I —)

one obtains, by (1 10.40),

—
+2
/ n/庐——庐 . f z- 啊
J t-z t
and hence

U 上— (116.7g)
衣(K
+ 1)v> — 外 衣 I) 2kV^2 — 好
By the formula PQ) = +修) 一 R), the pressure underneath the
stamp becomes
P(t) =
2Mp 2*)
-
,..... = "4
R(乂 十 1)“
P. .
庐—律
(1 ]6.8a)
7n/岸― /2

The solution will be physically possible, if P0)

i.e., if
0 for
—
KT—T~~iV (1 16.9«)
Pq ^2,

If Po does not satisfy the preceding condition, this means that the
force of magnitude J% is not sufficient to bring the arc ZB of the stamp
into complete contact with the elastic body. The arc 4 '5', which actually
engages the elastic body for some given Po < 27rM2/r(乂 + 1), will now
be found.
From symmetry, it is obvious that the segment W of the boundary
of the elastic half-plane which enters into contact has its centre at the
—
origin, so that one may write 屋 = V , b' = %', where 22' is the length
of the segment afbf. The function O(z) and the pressure P(/), correspond-
504 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 117

ing to a given will be obtained by replacing in (116.7口) and (116.8©

I by V . Expressing that PQ) = 0 for / = ± F, one obtains
VP^R(y. + 1)
(116.10^)
\/2印
It is sufficient to express that remains bounded for % = ± 厂；the result
would have been the same, as was to be expected on the basis of the statements
in § 116.

Alternatively, one may assume that %' is known and calculate the
magnitude PQ of the force necessary to make the length of the line of
contact equal to 2Z\ Corresponding to a given V , the functions (z) and
P(f) are determined by the formulae
2piv?2
五低 十
—1)z2 2"?
五 (X
、
+ 1)
PQ) =$•
K(y + 1)
不 上
… • (116.12a)

§117. Equilibrium of a rigid stamp on the boundary of an

elastic half -plane in the presence of friction.
This section reproduces, with minor modifications, the Author's paper [24].
At about the same time, N. I. Glagolev [1] published a paper in which he gave
the solution of the problem under consideration for the particular case of a stamp
with a straight base. Somewhat later, N. I. Glagolev gave the solution for the
case where the profile of the stamp is of arbitrary shape and where the friction may
depend on the area of contact. L. A. Galin [1] gave a somewhat different method
of solution (also applicable to an anisotropic body) ; cf. likewise L. A. Galin [4].

The problem of the equilibrium of a rigid stamp on the boundary

of an elastic half-plane has been solved in the preceding sections for
the two extreme cases, where the coefficient of friction is zero (§§ 1 15
1 16) or infinite (§ 1 14) ; in the latter case, a further condition had to be —
imposed, namely, that the elastic material could not leave the stamp and
that thus the presence of negative pressures, even arbitrarily large ones, is
admissible.
Using the method of the preceding sections, one may also solve the
problem for a finite coefficient of friction such as will occur in reality.
In this context, consideration will be limited to the case where the stamp
is on the verge of equilibrium ; obviously, the solution for this case will be
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 505

an approximation to the case, where the stamp slides slowly along the
boundary of the half-plane. More exactly, it will be assumed that T kP —
underneath the stamp on the boundary of the half-plane, where P and T
are respectively the pressure and the tangential stress, applied to points
of the boundary of the half-plane, and k is the coefficient of friction
which will be assumed constant.
Recently L. A. Galin [2] gave a clever solution of the problem, of the impression
made by a rigid stamp with a plane base under the supposition that the segment
of contact consists of three parts: a centre section with cohesion and two outer
sections on which slip occurs. In a simultaneously published paper, S. V. Falkovicz
[1] gave the solution of the same problem under the supposition that on the parts,
where slip occurs, friction is absent. Cf. likewise L. A. Galin [4].

As before, let the Ox axis be the boundary of the elastic half-plane and
the Oy axis perpendicular to it, so that the elastic body occupies the
lower half-plane y < 0. For this choice of axes, P = Y~, T = X~. —
Further, assume that the stamp engages the elastic half-plane along
one continuous segment U = ab. The result below is easily generalized
to the case, where the region of contact consists of a finite number of
individual segments (cf. preceding sections) .
In addition, it will be assumed that the stamp may only move vertically.
As in the preceding sections, the case where the stamp may tilt is readily
reduced to this case; cf. § 1 16«, 2°.
The boundary conditions of the present problem have the form
TQ) = 々P(£),(117.1)
厂 = /(力) 十 const. (117.2)
on L' , T(t) = PR) = 0 outside Lr on Ox. As before, t denotes here the
abscissa of a point on the Ox axis, is the projection of the displacement
on the Oy axis, /(/) is a given function for the profile of the stamp, i.e.,
y = /(%) is the equation of this profile. It will be assumed that 超) has a
derivative /'(%), satisfying the H condition.
In addition, it will be assumed that the quantity

Po=/ P 财, (117.3)

i.e., the total pressure exerted by the stamp on the half-plane, is known.
The total tangential stress will then obviously be TQ kP0. Thus
the resultant vector (X, Y) = (岂,一 Po) of the external forces, acting
—
 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 117

on the stamp and balanced by the reactions of the elastic half-plane,

will be given.
In the notation and under the general suppositions of § 1 12,the bounda¬
ry conditions (1 17.1) and (1 17.2) of the present problem may be written,
by (112.14) and (112.15) and using, as before, instead of (1 17.2) the
condition 厂，— /'(£),
(1 — 谢①+R) + (1 + 词①+(/) — —两① 词① (117.4)
n① —(力) +(力) '
(力 17.5)
on L' while the condition P(£) = TQ) = 0 on Ox outside L'is equivalent
to the condition that the function (z) must be holomorphic outside
the segment L' = 曲.
—
The formula (1 17.4) shows that the function (1 测①(z) 十 (1 认)
is holomorphic in the entire plane; since it must vanish at infinity,
one has
(1 — 淘①⑶ + (1 + 词⑤(z) = 0 (117.6)
in the whole plane. Expressing by the help of (117.6) the function (z)
in terms of (z) and substituting this value in (117.5), one obtains
the boundary condition for
+0) = g 一(力) 4- /oW on Lf, (M7.7)
where
x+一1 + 二冽* 一 1) ， _ 加- (1 词-
4
g— '

— 为 iT
—
一八
/ov)• 比一
一一
流( -- •

These last formulae may be simplified by the introduction of the constant

a, determined by the conditions (remembering that x > 1, k > 0)
x 1
tan Fa =k , 0 a <|. (1 17.8)
况 1
Then
(yt I j
+ 3一 1) = V(x + ])2 十讲 仅— l)2g =\ 乙
丁
1 土 土力a
流 ,
cos 冗a
and hence
则m
g = 一 於吗 加) = 7.9)
X
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 507

Applying now the method of § 1 10 to solve the problem (117.7),

noting that in the present case
logg
y 二后 +a
and that one may take for the function X°(z) in (1 10.18) the expression
(z— (力 ~
- yf. §110, Note 2), one obtains

_
b
2以1 + 识)产m cos ha
/ (方 — a)"a e — 坤-叶电dt +
= 兀 1) (Z — ©"a 他 — z)*-a J •z

丁
{z
— a)^ {b —。
「，
z)^
(117.10)
'
—
where Co is a constant and where (z ©务 0 z产 must be under¬ —
stood to be that branch which is holomorphic outside the segment ab
and which takes on the upper side of this segment the real, positive value
—
Q a)^+a (b — 9*~力 as is easily seen, this branch is characterized by
the fact that
产
lim"， (117.11)
2->ooZ

For the determination of 丫， itmust be remembered that by the condition, in¬

troduced in § 1 10, the value of the logarithm, must be taken for which

0 阮姐I < 1.
2m
The quantity, denoted in § 110 by a, is now denoted by|+ a, and the function
/(力 by

The constant is directly determined by (1 12.4) which gives

lim z (z) = 一
B+ 击。= 汨°(i +认) ,
gf 8 2加 27V —
whence, by (117.11),

Cq = +— 的产。
508 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 17a

and (117.10) becomes

b

=
2Ml +_i* 取 COS a 兀

J — /
口) "a
(办 一 /)Aa /'(£) 成
+
3
兀 十 1) (z 牙+a 也 一 Z产a
a
t — Z

Po(] +
十 (117.12)
2兀 (z — 为"式力 一 班-a
It is readily verified that all the conditions of the problem will be
satisfied, provided, as has been assumed, /'(力) satisfies the H condition on
Thus the problem is solved, because completely characterizes the
state of stress.
Naturally, the solution will be physically possible only in the case
where the pressure PR) at the points t underneath the stamp satisfies
the condition P© 0. The pressure is easily calculated on the basis of
(117.12). In fact, by (112.14),
P(幻 + 汉甸 = P&)(1 + 询 = +(幻 一 o).
This last difference, using the Plemelj formula, gives
4u. sin 7ta cos Ka „ 、
、
%°)=— — — /《)+

4fx cos — ^)i+a(6 — 力户巧'(力)〃

C (t 力
+ 7V(X 十 1) —^a产 “(力 J /
a
尸0cos Tta
(117.13)
冗饱 一存 广 a

For —
0 (when also a = 0), one obtains again the solution for the ideal¬
ized case without friction.

§ 117a Examples.
1°. Stamp with straight horizontal base.
In this case 广 (£) — 0 and (117.12), (1 17.13) give

=
产0(1
+ 国产 (117.1勾
2兀 (Z — a)"汽6一力
卜 a

_ cos 兀a
— 兀(4一 a产仅一尸
(117.2a)
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 509

2°. Stamp with straight inclined base.

Let the base of the stamp form a (small) angle s with the axis Ox. Then
/(/) = E 十 const., /'(力) = £. Substituting in (1 17.2) and putting, for
—°
simplicity, a = I, b = I, one obtains by (110.40)

=
(1 +洲浮 P0(x + 1) 8Gl ~>4邛wz — +
+ —z 1
ik)
(117.3«)
2兀3+ 亍 +
'
'

Q + z)"a("z户
In the present case, one has for large | z ]

X「(z) = (2 十 z)* + a Q — z)Aa = 一 幅足式2 + 2历) + 0(1b).

The pressure P(%) is given by
+（9 一（力 cos 兀a 产0（乂
+ 1）— 8 — 四民
―
兀
P(%) = ―（"7•的
”
7V(X + 1)
*

i十 声第 一尸
The solution is physically possible, i.e., P(£) 0 for —Z / Z,
when
(117.5a)
4冗p7(l ~
• 2a) 4兀心 1 十 2a)
The moment

M =

is likewise easily determined ; it is equal to the resultant moment of the

external forces, acting on the stamp；in fact, proceeding as in § 114«,
2°, one obtains

+ (117.6,)
x十1

This last formula determines w for given M and Po. In particular, if

跖 = 0, i.e., if the external forces, acting on the stamp, are equivalent to
a force applied to the centre of the base, then

w — (比
&
+ DB .
， (117.7
必1 — 4a

since 0 a < J, this value of w leads, by (1 17.5〃)，to a possible solution.

510 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 119

§ 118. An alternative method for the solution of the boundary

problems for the half -plane. In the preceding sections the solution
of the boundary problems for the lower half-plane has been reduced to
the search for a function (2), suitably extended into the upper half-plane.
However, it is obvious (cf. the methods of solution, studied in the
preceding Part) that these problems could be reduced directly to the
search for a function
=
J (z)dz,

likewise extended into the upper half -plane [cf. ( 1 1 2. 1 6)]. Some in¬
convenience would result from the fact that, in general, <p(^) is multi-valued.
However, this inconvenience may be removed by separating from <p(z)
the multi-valued part, which is very simply done. On the other hand, in¬
troduction of the function 少(?) instead of (z) has the advantage that
in constructing the boundary conditions, involving the boundary values
of the displacements, one is not obliged beforehand to differentiate
these values.

§ 119. Problem of contact of two elastic bodies (generalized plane

problem of Hertz) .
Consider two elastic bodies Slt S2 the shapes of which approximate to
half-planes and which are in contact along a segment ab of their bounda¬
ries (Fig. 55). The segment of contact ab will not be given beforehand.

but will be subject to determination. The shapes of the boundaries

(approximately straight lines) before deformation and the resultant
vector of the external forces, applied, for example, by the body S± to the
body S2，will be initially known. It will be assumed that there is no
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 511

friction and, in addition, that the stresses and rotations in Sr and S2

vanish at infinity.
This problem, which is of great independent interest, is even more
important, because it constitutes the problem of contact of two bodies of
arbitrary shape (i.e., having the two-dimensional case in mind), provided
the area of contact is very small in comparison with the dimensions of
the bodies ; in that case, if one is interested in the stresses and displace¬
ments near the region of contact, one can assume without appreciable
error that the bodies under consideration are in shape close to half¬
planes.
In three dimensions, the problem of contact of two elastic bodies was
first formulated and solved by Hertz under several restrictive assumptions ;
in particular, he assumed that the area of contact is very small and that
the equations of the undeformed surfaces near the region of contact
could be approximated sufficiently accurately by functions of the form
z — + 2Bxy + with a suitable choice of coordinate axes.
I. Ya. Shtaerman [2] reduced the three-dimensional Hertz problem, under
more general suppositions, to an integral equation.
Thus the problem of contact of two bodies, approximating to half-planes,
which was stated above is the two-dimensional analogue of Hertz'
problem, however in a somewhat generalized form, since it has not been
assumed that the region of contact is small and, correspondingly, no
assumption regarding the shapes of the boundaries has been made,
except the condition that they should approximate to straight lines (and
be sufficiently smooth).
In recent times, several authors have considered this problem. I. Ya.
Shtaerman [1, 3] (cf. also his recent book [4]) reduced it to a Fredholm
Equation of the first kind which in the present notation may be written
b

a
/尸㈤ log \ t — tQ \ dt — /(%) + const.,
where 户(£) is the unknown pressure, exerted by one body on the other
at a point 力 of the region of contact of the bodies, and /(/) is a given
function. The problem of pressure of a rigid stamp on an elastic half¬
plane, as treated in the earlier editions of this book (§ 87), led to
just that equation ; this equation is easily solved by quadrature for
given a and b (cf. the earlier editions of this book, § 88).
A. V. Bitzadze [1] reduced this problem to a singular integral equation
512 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 19

which is immediately solved in closed form. His equation may be ob¬

tained by differentiating equation (a).
The following solution of the problem is obtained by a method, com¬
pletely analogous to that used earlier in the case where one of the con¬
tacting bodies was absolutely rigid (§ 115).
It will be assumed that the body occupies the lower half-plane
and the body S2 the upper half-plane S+ and the corresponding stresses
and displacements, and likewise the constants 入， will be provided
with the subscripts 1 and 2.
Let i(z) be a sectionally holomorphic function which corresponds
to the body Sr and is defined as in § 1 12; let 2 3) be the analogous
function for the body S?. These functions are holomorphic throughout
the plane, except at the segment ab of the Ox axis, because outside ab
the boundaries of the bodies are free from external stresses. As, by
supposition, there is no friction, one will have [X/i = 0 on 0%; hence,
as in § 1 15, one may conclude that i(z) = —— 式句 ；similarly, one finds
that 2(勿 — — 2(z)・ Further, if P($) is the pressure exerted by one
body on the other at the point t, then, as in § 1 15,
p(方) =值(e)一 汕 (119.1)
similarly, one has
PR) = 一①义务 (119.2)
It is seen from these conditions that [Q + C>2]+ = @1 + 『，2 ie,
that the sum (z) 4- 乞⑶ is holomorphic in the entire plane ; further,
since it vanishes at infinity, one must have "z) + 2(z) — 0. Thus

If now
电(Z) = — l(z), 电⑵ = — 2(z), @2 =
— i(z). (H9.3)

y = /i(0> y = /2(*)
are the equations of the boundaries of the bodies and Sz before de¬
formation, where /式£), /2(Z) as well as their derivatives /；«), must
be small, one will have in the region of contact, after deformation,
/iW + 近(£) = /2W 十 说 Q)，
whence

— 球 — 力) on ab (1 19.4)
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 513

or [cf. (112.22)]
万二 ——球，= /'(/) on ab, (119.5)
where
力)— (H9.6)
It will be assumed that /'(/) satisfies the H condition.
With regard to the deduction of the condition (119.4), reference should be made
to § 115. Strictly speaking, one should have written +巧 =
/式力) +喏 = FQ + 〃扑，where y = F(f) is the equation, of the line of contact
after deformation ; however, within the accuracy considered here, it may be
assumed that F(t 4- 衍) = F(t + in which case one obtains the stated relation.

Expressing now the boundary condition (1 19.5) by means of (1 12. 15),

applied to Sx and S2 respectively, one finds, in view of (1 19.3),

+ 式力) = B，
A
(119.7)
where

+ 4M2 (119.8)
4出
One has thus arrived at the same mathematical boundary problem
as in the case of the problem of pressure of an absolutely rigid stamp on a
half-plane, i.e., at the problem, corresponding to the boundary condition
(115.5); the only difference is that in this formula 式名) and 1/K take
the place of (z) and 4俨/乂 + 1 respectively. In addition, in the present
case, the segment of contact is not known beforehand and, as in § 1 16,
2°, it is required to find the solution 】(z), vanishing at infinity and
bounded near the ends 〃，b.
Using the formulae of §116 or directly those of § 1 10, one arrives
at the following conclusions.
The function i(z) is given by [cf. (116.9)]
b
_ ，(z — a) (b — 港
2hK J/ a/ R — 四) —4 (* — z)
For the determination of a and b one has the two relations [cf. (1 16.7)]
b
/'(£) 今
0 (119.10)
V (t — g) 仅'—
514 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 119

and

[ 竺」， = k (H9.ll)
J W (% — a) 3 — %)
a
/

where is the magnitude of the resultant vector of the external forces

applied by the body S2 to Si (or S】to S2) which will be assumed known.
As in § 1 1 6, V (z — 讥) 仅 一 z) must be interpreted as a branch such
that for large | z j
—
M(z — 幻 (b z) = 近 + 。⑴， — (119.12)
and Mg
root.
— a) (b ~ for a < refers to the positive value of the

The pressure P(幻 = 1b — ©式幻 is given by

b
'(w — 切他 一 力) f r⑦ dt
J —V{t — a) — /)
n八 — (119.13)
片如 /-
一【 …
加R
a
If the function /(/) is even, i.e., if
/(—%)=/(%)， (119.14)
one may, from considerations of symmetry, write from the beginning
—
_
a= • %, 力 = 十 ％, where I is subject to determination. In this case the
condition (119.10) is automatically satisfied and I may be determined
from the equation

j KPqi (ii9」r)
J
0
. /2

The final formulae, obtained for (119.14) and

with those deduced by A. V. Bitzadze [1].
b, agree
—
As shown in §110, Note 2, the integrals in the preceding formulae
may be evaluated by elementary means, provided /'(/) is a rational
function or, in particular, a polynomial. For example, let
怨) = 4 个
where X is a constant and n a positive integer ; one then obtains im¬
mediately the solution, found by I. Ya. Shtaerman [1]. Putting

物=
2(高 高)
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 515

which corresponds to the case, where S】and S2 are bounded by circles

with radii and & (which are large as compared with the region of
contact), one obtains the solution, found by L. Foppl [1] by other means.
(As regards the calculations, this case is the same as in § 116a, 3°.)
Several other examples may be found in the paper [3] and the book [4]
by I. Ya. Shtaerman.
In the presence of friction between the bodies in contact, the problem
is considerably more complicated. Solution of several problems of contact
in the presence of friction which are of special practical interest has been
given by N. I. Glagolev；some of these results are published in his paper

§ 120. Boundary problems for the plane with straight cuts *).
The fundamental boundary problems, and likewise some other prob¬
lems, for the case where the region, occupied by the body, is the entire
plane with straight cuts, distributed along one and the same straight
line, are easily solved by methods, analogous to those used in the preceding
sections. Let the Ox axis be the locus of the cuts. A beginning will be made
with the deduction of several formulae, analogous to those of § 1 12.
1°. General formulae
Let the region S', occupied by the elastic body, be the entire plane,
cut along n segments Lk = 瞅% (也 = 1 , . . . , of the Ox axis; the union
of these segments will now be denoted by L.
It will not be assumed in this section that the stresses vanish, but
only that they are bounded at infinity.
Then and 乎(z) are holomorphic in S', including the point at
:infinity, and for large | z by (36.4) and (36.5),
X+过
27c(1 + 比)
(102.1)
乂 (X — Wx) z 十。6），
1
2兀(1 +
where (X,丫) is the resultant vector of the external forces, applied to
the edges of
P= r‘ = B' + 2c (120.2)
*) This section reproduces, almost without modifications, the contents of the
Author's paper [23].
516 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 120

are constants, determined by (§ 36)

8 +(跖 盟)， c= 1 十 M
rf
— (120.3)

N» N? being the values of the principal stresses at infinity, a the

angle between and the Ox axis and the magnitude of the rotation
at infinity.
In the usual notation, introduce the function
Q(z) = + + 于(z), (120.4)
which is also holomorphic in S' and has, by (120.1), for large | z | the form

Q(z) = r+r+ 聂
2k (1LT
*0 (I).
+ x) z \z2 /
(120.5)

Substituting in (120.4) z for z and taking the conjugate complex value,

one obtains
乎(z) = 0(z) — (z) — z*(z). (120.6)

Since the stress components are expressed in terms of the functions

¥(z), one may also express them in terms of <!>(z) and Q
In particular, one has by (32.8)
兀 一 iXy = + QR) + (z — 可 *(z). (120.7)
Similarly, one may express the components of displacement by in¬
troducing instead of 少 (z) the function

g)(z) — =z (z) + 少⑶ const. (120.8)

which, like the functions cp(^), 少(z), is determined by ©(z), 乎(z), apart
from an additive constant. Thus (32.1) takes the form
2^(% 十加) = 叫⑶ ~ co (5) — ~ ^)0(^) + const. (120.9)
It will be assumed in the sequel that (z), Q(z) are sectionally ho¬
lomorphic in the sense of the definition in § 106, so that, in particular,
near the ends ak> bk
A A
I I<
— (120.10)
Q
—
V
z
|In c \产 ' Iz c亚
1 '
where A, a are positive constants, 0 a < 1, and c denotes the cor-
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 517

responding end. Further, it will be assumed that for all 力 on L which do

not coincide with ends
+ iy) = 0. (120.11)
0

2°. First fundamental problem

Consider now the solution of the first fundamental problem, i.e., assume
the values of Yy , Xy, Y~, X~ on L to be given, where the (+) and ( )
signs, as always, refer to the boundary values on the upper and lower
—
edges of the cuts. A (less simple) solution of this problem was given by
D. I. Sherman [12].
In addition, it will be assumed that the constants = B and
F' + iC\ i.e., the values of the stresses at infinity, are known.
Since one is concerned with the stress distribution, one may, without
affecting generality, assume that C = 0, i.e., that
F= f =艮
By (120.7) and (120.11), the boundary conditions take the form
—
+(/) + QR) = Y； 凶， R) + Q+R) = 石一 近京 (120.12)
on L. Adding and subtracting, one obtains
[0(/) + Qd)]+ + [①(力) Q(切- = 2加)， (120.13)
[① 伍) - Q©]+ + [©(/) - = 2跑 (120.14)
on L, where 力(力)， 夕传) are the following functions, given on L:

的= J[母 + 叼 一搭m + 有，
(120.15)

It will be assumed that 力(力) and q© satisfy the H condition on L.

Since C>(oo) — Q(8) = 一•T', the general solution of the boundary
problem (120.14) is given by (§ 108)

m J/ — R.
一 Q(z) =3 力 —~
z
(120.16)
L
Further, writing

—
n
X = n=i (z
化
(z — (120.17)
518 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 120

and applying (1 10.33), one obtains the general solution of the boundary
problem (120.13), bounded at infinity [as follows from (120. 1) and
(120.5)],
1 2P“(z)
+ Q(z) = 卞小 (120.18)

where Pn{z) is the polynomial of degree not greater than %

P*z) = 。庐勿 + + ... + (120.19)

XQ) must be interpreted as the value, taken by X(z) on the upper (left)
side of L.
The formulae (120.16) and (120.18) give

where
(z) = o(z) +
工 一 转'， Q(z) = Qo(z) +
工 +转 (120.20)

小 小一
。㈠ —
__ ，理幽 艺 + _1_ /史丝 ， (120.21)
2脑X(z) J
t z — t~z 2m J
°
_
。㈠ —
__1
23X(z)
工

/ X 也阴祝
J t — z
__ L

2ni J t—z
(120.22)
L L

It is easily seen that under the present conditions with regard to

0(/) and q(t) the condition (120.1 1) is fulfilled on the basis of § 69, 2°.
The polynomial F£z) has still to be determined. It will be assumed
that X(z) is a branch which for large | z j has the form
，X(z) = 尹t
十 z处十 a*】 + .... (120.17')
The coefficient Co follows immediately from the first of the formulae
(120.20) and from the condition (8) = T which give
Co = r + ir. (120.23)
The remaining coefficients must be determined from the condition of
single-valuedness of the displacements. By (120.9), this condition implies
that the expression 乂平 (z) — 30) must revert to its original value as the
point z describes a contour A 丸, surrounding the segment 以初 把 — Lk.
By contracting the contour into the segment Lk, it is readily verified
that the following relations express the condition of single-valuedness,
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 519

where the differences 力怖 一 Q and Qj — Q. are easily determined by

the Plemelj formula,

2(乂 /W
+ 1) J ―A V)
Lk
1- x
工汇
J
八①力 一 ©dQW

尸
+

+
&
/ — Qj(切弘 =0, (120.24)

k = 1,

which give a system of linear equations for the constants C2, . . . , Cn.
”
This system has always a solution. In fact, the homogeneous system,
obtained in the case F = T' = 0, = X； ~ Y~ = X~ = 0, can

have no other solution except = C2 = ... = Cn = 0, because the

original problem, as is easily established by ordinary means, has in
this case only the trivial solution (z) = Q(z) =0. Therefore the non-
homogeneous system (120.24) always has a unique solution and the prob¬
lem is solved.
In the particular case, where the edges of the cuts are free from stresses
(problem of extension of plates weakened by cracks), o(z) = Qo(^) = 0 and
the solution takes the extraordinarily simple form

=
瑞- 衰 '， Q(z) = + 疗'， (120.25)

and the coefficients of Pn{z) are determined by the conditions

c0 -r+ jr; ^^=0, /=1,2, … ，忆 (120.26)

For n = 1 (single crack), letting 诙 = — 口，％ = 口， one obtains the

formulae

(z) = 竺±1竺 一犷， Q =回小十 隶.（w.27）

2Vz — 谈 2 2 Vz2 — 说
A solution of the (less simple) problem for the particular case n = 1
is effectively contained in § 82a, as a particular case of the problem
of the equilibrium of a plate with an elliptic hole under the influence
of external forces, applied to its boundary.
520 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 1 20

3°. Second fundamental problem.

Consider now the second fundamental problem, i.e., assume that the
values of the displacements %+(/), #(%) on the upper edges and
厂
@ on the lower edges of L are given; also, if 优(瞅)，办(阪) and 矶瓦)，
研6 及) denote the (given) displacements of the points ak, bk, assume that
铀) = %一(以) = %(距)， 什(w) = 厂(公) m
曲)， /sc C6
(IzU.zo)
(如 依) 勿伍)， 力 厂(如 川姐.
In addition, let also the constants T and P' (without assuming this
time C = 0) and the resultant vector (X, Y) of the external forces,
applied to L, be known.
In order to avoid having to consider directly the functions 勺(z), s(z)
which may be multi-valued, the boundary conditions will not be con¬
structed beginning from (120.9), but from a formula, obtained from (120.9)
by differentiation with respect to %, i.e., from

〃
2M + ') = 乂①⑶ Q0) — (z —
说 一 (120.29)
〃
where uf, are partial derivatives a切2%,
the
boundary conditions
dv/dx. Accordingly the
may be written
y +B) — = 2M3'+ + 说'+),
(120.30)
乂① -口) — Q+Q) = 2yL(u'- +
Adding and subtracting, one finds
仅① (2) —+ Q(切 + 十
[乂(DQ) - Q(切一 = 2/(/), (120.31)
[x0>(Z) Q(切+
—
[炖 + Q(切- = 2g(Z)
on L, where /(/), gQ) are the following functions, given on L:
(120.32)

力) = %") + %®+' + 厂')],

gQ) = 瓦(％+/ 一 ％1') + 厂')].
%®+'
—
It will be assumed that these functions satisfy the H condition on L.
020.33)

In the same way as in the preceding problem, the general solutions of

the boundary problems ( 1 20.32) and (120.31) are given by

乂①⑵ +Q / 改咬 + rz + xr + f, (120.34)
L

”迹
乂① (z)
— 。少—
Q(z)
1
= ―年ry fJ/ 一 1—&/
2Pn (z)
X(z)、
， (120.35)
km L
z) 一z
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 521

where X(z) is determined by (120.17) and，X(力) is its value on the left side
of L
The preceding formulae determine the unknown functions (z), ¥(z)
apart from an additive term, containing the polynomial
p/z) = c^n + + ...十
The first two coefficients Co and 。工 of this polynomial are immediately
determined by (120.35), if one takes into consideration that for large
{ 2 I， by (120.1) and (120.5),
，十 0 (口).
x©(z)
— Q(z) 三
— P —『' 一
3+
加 1) Z \0/
(120.36)

It is readily verified, on the basis of (120.28) and (120.30), that the

displacements %, % calculated from (120.9) using the functions @(z),
Q just found, will be single-valued. However, these displacements
will assume on the cuts Lk the given values, apart from some constant
terms which may be different on different cuts. Let the constants,
by which the expression 2以〃+ iv), calculated from Q(z), differs
on the cuts Llf L2, . . Ln from the given values, be denoted by q,
c2, . . . , cn. The functions (z) and Q(z) will satisfy the conditions of
the problem only when q = % = • • • =
These constants will be equal on the upper and lower edges of each cut, because,
as is easily seen from the conditions introduced earlier, the expression 2M(勿 谢)
tends to a definite limit as z approaches one of the ends a论，b无.If q = & = … = cn,
the condition % = 0 may be attained at the expense of an arbitrary constant
entering into the right-hand side of (120.9).

It is easily seen, by (120.29), that the conditions for the ck may be

expressed in the form

/[源＞©
b*
— QQ)]弟 = 2M{["(瞅+i) — 勿 仇)] *(即+i)
— ? 仇)]}, (120.37)

where the quantities on the

力 1,2,・ ・
一 ％~
• 1,

side are given [cf. (120.28)].

right-hand
Substituting for 比①(力) 一 Q(£) from (120.35), one obtains a system
—
of w 1 linear equations for the determination of the n — 1 coefficients
C2, …， Cn which were so far undefined ; similarly as before, it is easily
seen that this system always has a unique solution. Thus the problem
522 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 120

is solved. The solution for the particular case n

§ 83 by different means.
—
1 was obtained in

The problem, where the displacements are only given apart from con¬
stant terms which are different on different cuts, may be solved in an
analogous manner ; however, in that case, the resultant vectors of the
external forces, acting on the individual cuts separately, must also be
given.
4°. A certain mixed problem
In conclusion, a certain problem will be solved which was considered
by D. I. Sherman [13]. In this problem the external stresses, applied,
say, to the upper edges of the cuts, and the displacements on the lower
edges are given. D. I. Sherman solved this problem by rather complicated
means, reducing it to a system of singular integral equations (which
admittedly is simple), and there is one omission in his solution about
which more will be said later.
By (120.7) and (120.29), the boundary conditions may be written
+(/) —
+ Q-Q) = Yj iX^, 乂① -(£)
— Q+0) = 2M 力- 谢'-) (120.38)
on L. Multiplying the second of these equations first by —
iW 乂, then
by + £/标，and adding to the first (cf. D. I. Sherman [13], p. 333), one
obtains

「L0(0 + -y/x
Q(“+
J — 「
Hi L + 3
vx
Q0)]
」
= 2/式力， (120.39)

= 2/2G) (120.40)

on L, where /式力)， are functions, given on L; it will be assumed that

these functions satisfy the H condition on L.
Thus the functions
(z) H —Vx
尸'Q(z), (%) 尸 Q(z)
vx
determine the solutions of the boundary problems (120.39) and (120.40)
which are particular cases of the problem, solved in § 1 10. In the notation
of that section, one has for the problem (120.39): g = 勿“元 while for the
problem (120.40) : g = ~ V工
Solving these problems by the method of § 1 10 and taking into con-
CHAP. 19 BOUNDARY PROBLEMS FOR THE HALF-PLANE 523

sideration the behaviour of Q(z) at infinity, one finds

+个 Q =
q 十铲 + X】(z)Pg)(z),
年2rz「 (120.41)
Vx J [t)(t
L
Aj
— z)
0⑶一Q
•yx
= 鬻J/f需"
2m
——L
z)
+ xz
A g (t) * (120.42)

where
n n
X](z) = n (z-叫f (z —呢尸T, X2(z) = n (z—瞅产 (z
i
— 姐Y2-1, (120.43)
log (2Vx) x logx log (— logx 、
=—
Y】 "后 Y2 =I 十 3(120.44)
^一
For Xi(z),
2to
one must select branches, holomorphic in the plane
4tuz

cut along L.
By adding and subtracting (120.41) and (120.42), one may obtain
closed expressions for (z) and Q(z), but this will not be done here.
For the determination of the 2% 十 2 coefficients of the polynomials
P法) and one has the following conditions. Firstly, that the functions
and Q(z) must behave at infinity in accordance with (120.1) and
(120.5) ; in this connection it will be assumed that the constants F,
r‘，X, Y, entering into these formulae, are given.
The resultant vector of the forces, applied to the upper edges of the cuts, are
determined by the values , X+ on £ ; in addition, it is assumed that the resultant
vector of the forces, applied to the lower edges, is known. The sum of these vectors
is the vector (X, V).
Secondly, that the displacements must be single-valued, as in the case of
the first fundamental problem. Finally, that on the lower edges of the
cuts the expression 2M(怒 十 w) assumes given values, and not only apart
from certain constants ; as in 3°, it is sufficient for this purpose to express
that 2M(铝 + iv) assumes on the lower edges of the cuts the given values
apart from a constant which is the same for all cuts. In this way one
obtains a system of 2% + 2 linear equations, since the first group renders
four, the second n equations one of which is a consequence of all the others
by the strength of the equations of the first group, while the last group
——
contains n 1 equations. These equations determine the 2% 十 2 un¬
known coefficients and it is easily verified, on the basis of the uniqueness
theorem (which obviously holds under the given conditions), that this
system has always a unique solution.
524 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 120

D. I. Sherman [13] did not subject the unknown solution, to the conditions,
securing uniqueness, in contrast to what he did in his paper [12] in which he solved
the first fundamental problem. Therefore his solution of the problem, considered
just now, contains constants which cannot be determined without additional
conditions. Sherman gave no study of his solution and assumed these constants
to be arbitrary.

For example, consider the case where there is only one cut L = ab
the lower edge of which does not move ("— = »- = 0 on 上)，while the
upper edge is free from stresses (Yj = XJ = 0 on L), and where the
stresses and rotation vanish at infinity (F = F' = 0). Further, let it
be assumed that the vector of the external forces, applied to the lower
edge, is equal to (0, ― Po).
This problem may be interpreted as follows: a rigid straight strip has
been welded to the lower edge on which acts a symmetrically distributed
force of magnitude Po in the negative 夕 direction.
In the present case, % = 1, /工) — /2(^) = 0,
—
Xi(z) = (z a)-Yi (z — —
X2(z) = (z 幼一丫^ R 乃丫2-i —
「
and, since (z) Q(z) must vanish at
and infinity (because F = = 0),
one has, by (120.41) and (120.42),
= 加占⑵ 十 偌乙⑵， Q(z) = —
where Cr and C2 are constants. These constants are determined on the
basis of the conditions, following from (120.1) and (120.5), i.e., for large | z |

3 十U2/
2Ml 十 x) z
十
\^) 2兀(1十x) 7

whence, assuming that X】(z)and X2(^) refer to branches for which

lim zX^ = lim zX^z) = 1 ,
名f 8 名68

Cl + 7；?
0兀(1 「
2 + 乂)
— '“算°2 =
-o 兀 「
x)
ie,
—
‘工=一4碎 十比)
汨
， 品=
厂
击 十
疯1 十M)
…，
 Chapter 20

SOLUTION OF BOUNDARY PROBLEMS FOR REGIONS,

BOUNDED BY CIRCLES, AND FOR THE INFINITE PLANE, CUT
ALONG CIRCULAR ARCS

Important boundary problems for the circle and for the infinite plane
with circular holes may be easily solved in a manner analogous to that
used in the preceding chapter. The solutions of the first, second and
mixed problems for these cases, and likewise for a more general case
to be considered in Chap. 21, were given by I. N. Kartzivadze in his
dissertation parts of which have been published in his papers [1,2];
only finite regions are considered, since the case of infinite regions may
be solved by analogous means. Those results of Kartzivadze which refer
—
to regions bounded by circles will be studied in §§ 121 123.
B. L. Mintzberg [1] published recently a solution of the mixed problem
for an infinite region with circular holes ; he was apparently only ac¬
quainted with the first of the above papers by Kartzivadze.
In § 124, the solution of the fundamental problems for an infinite
region cut along circular arcs will be given.

§ 121. Transformation of the general formulae for regions,

bounded by a circle. Let L be the unit circle with centre at the origin
and let S+ be the inside of this circle and the remaining part of the
plane (excluding L).
Let the elastic body occupy one of the regions S~, S+. Introduce polar
coordinates r, 分 by the relation
z —x iy = 夕外"
and, as in § 39, denote by rr, 分分，夕9the stress components in polar co¬
ordinates. The formulae, expressing these components in terms of the
functions (z) and 里(z) (§ 39), will now be written as follows [(39.4), (39.5),
where in the latter 於龄 has been replaced by (z/万)]:
方 + 薪 = 2[① (z) + 函], (121.1)

525
526 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §121

// + ir^ = (z) + 跖一 2酝 ——z T(z). (121.2)

Further, the formula

2[i(% 十 加) — 叩(z) — z(p'(z) — 虱z) + const. (121.3)
will be recalled which expresses the displacement components u, v (in
rectangular coordinates) in terms of the functions 中 (z), W(z), related to
and 乎 (z) by the formulae (p'(z) (z), W'(z) = T ；differentiating
—
(121.3) with respect to 分，one finds the formula

2以%' + 谢') = 应 [y①(z) —函+ 三 m]，(121.4)

where now

—.
彻
e，=
du
u' — ,

The functions (z) and 乎(z) are holomorphic in the region under
consideration (S+ or S-). When this region is S~, these functions have
for large | z | the form

(12k5)

Rz)=「+

where in the former notation

； + (⑵・6)

F = B + %C, F = B' (121.7)

B= MN. + NJ C
- 1 + 比
T' = 一 MN】— N^. (121.8)

Using a previous notation (cf. § 76), the definition of (z), originally

defined in S+ [or in S~], will now be extended to the region S~[or S+]
by writing

= 一不 (；)+ ：齐(：)+ 3千(；) (121.9)

in S~[or S+], i.e., for | z | > 1 [or [ z ] v 1].

This extension has been selected in such a way that the values of from the
right and left of L continue each other analytically through the unloaded parts
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 527

of the boundary (cf. § 112); with this aim in mind, one readily arrives at (121.9),
remembering (121.2) and noting that on L

-z — —.
1

Replacing in (121.9) z by 1 jz, as one is, by supposition, justified to do

for 【 z | > 1 [or for | z | < 1], and assuming now \z\ = \z\ <1 [or
| 5 | = | z [ > 1], one obtains

= - o® + 而3 + /于
whence, taking conjugate values,

*⑶ — 方 (2) +0 G) '(z)・ (121.10)

Since the components of stress and displacement may be expressed

in terms of (z) and ¥(z), they may also, by use of (121.10), be expressed
in terms of (z) only, which is now defined throughout the plane (ex¬
cluding 乃.
When S+ is the region occupied by the body, the function is
holomorphic in S+ as well as in S~, including the point at infinity ; this
follows from (121.6) and (121.9). However, the behaviour of C>(^) at
infinity must be subject to several conditions, in order that the cor¬
responding function ¥(z) will be holomorphic in S+. In fact, let
二
4+ + 月23 +… (for|z|< 1),
8。 (121.11)
(z) = 当 H Bl
… 炉— …
n
! F (for|z|> 1).
Z

In order that 空⑶，determined by (121.10), will also be holomorphic

at z = 0, one easily deduces the condition
Aq + Bq = 0, B] — 0. (121.12)
In the sequel it will be assumed that these conditions are satisfied.
When S~ is the region occupied by the body, the function (z) is
holomorphic in S~ (including the point at infinity) as well as in S+, except
at the point z = 0 where it may have a pole. In fact, (121.9) together
with (121.5), (121.6) shows that near z = 0
F x(X + iY) 1
…
(z) =
Z? ——2 (1 + X) —
—— 兀
— 4- a holomorphic
2
function. (121.13) /
528 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 121

The stress components will be obtained in terms of the function (z)

from the formulae (121.1) and (121.2), if one understands by 乎⑶ in
these formulae the expression (121.10). In order to give (121.10) the
form which is most convenient for the present purpose, it may be noted
that by (121.10)

环=浮画一① (g)；
substituting this expression in (121.2), one finds

〃+ •—① (121.14)

where on the right-hand side one should understand by ¥(z) the expression
(121.10).
Similarly, one obtains from (121.4)

2M(力 源) = 数 h① (z) + (121.15)

where again 空(z) is given by (121.10) and, as before,

加 彻
% = , vr = — . (121.16)

It will now be assumed that (z) is continuous at L from S+ and from

S~, except possibly at a finite number of points ck of L near which

I
m,、
I V z
const. I
ck !—
《 a V 1； 。 (121.17)
•in addition, it will be assumed that for all points t on L, except possibly
the points ck
lim (1
Tf 1 - 力 *(z) =0, z = ”吗 (121.18)

It is easily seen, on the basis of (121.10), that

—
lim
1
(z\ — —z / 里⑵ — lim L* (r\

as shown by (121.14), +0) -Q)—

什 i

If L contains unloaded sections Lf, i.e., if on Lf: rr =

—
/ /
~ 0, then,
0 on U. Consequently, the values
of (z) inside and outside of L are analytic continuations through
空(z) = 0. (121.19)
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 529

the unloaded sections of the boundary, as in the case of the half-plane.

In fact, the definition (121.9) of (z) in S~[or S+] was chosen in order to
ensure this property.
A number of fundamental boundary problems for the circle are easily
solved by use of the preceding formulae in a manner, similar to that
used in Chap. 1 9 in the case of the half-plane.

§ 122. Solution of the first and second fundamental problems

for the region, bounded by a circle. These problems have been
previously solved by different methods. Their solution will be outlined
here as an illustration of the new method.
1°. First fundamental problem for the circle
In this case S+ is the region, occupied by the body, and the boundary
condition has the form
/ + z•而 = 阳方)
2T(机 (122.1)
where N and T are the normal and tangential stresses on L which will
be assumed known. By (121.14), this boundary condition takes the form
[cf. remarks following (113.1)]
+(/) 一 0-(^ = N© + 仃0). (122.2)
One has thus arrived at the problem, solved in § 108; in the present
case it is required to find the solution, bounded at infinity. Applying
(108.2), one finds

L/也产
z
27cz J
L
力 一

where Bq = (8) is a constant, at present unknown. In order to de¬

termine and also to ascertain whether the problem is possible, con¬
sider (121.12) which must, by supposition, be satisfied.
For this purpose the constants and Br of (121.11) will be calculated.
One has
27r

4= (0) = 上/J WQ) 4-

27Vz
疗
^-
t + BQ = 4~hN +
Ztz J
汀)游 + 筑，
L 0
2K

鸟 — lim [z
8
—— 2兀2
L
WB) + 汉㈤W = — 0
+ 汀)净
530 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP
§ 122

The conditions (121.12) give then

2n

2F J
一 iT)d^ + 绕 + 瓦 = 0,
o
27r
(122.4)

/(N
0
十 汉)理阴 = 0.

The first of these formulae shows that one must have

27t

，粉 =
o
0; (122.5)

if this condition is satisfied, then

沉风= (122.6)

The conditions (122.4) and (122.5) express that the resultant vector
and moment of the external forces vanish, which are necessary for the
existence of solutions.
The formula (122.6) determines the real part of the constant Bo; the
imaginary part of BQ remains undetermined, as was to be expected,
because it only influences the rigid body motion. Thus the problem is
solved.
2°. First fundamental problem for the plane
with a circular hole
This problem may be solved in the same way as the preceding one.
In this case
益 + E 标- = Nq) + 疗肛 (122.7)
where Ng) and T© are the given external normal and tangential stresses ;
as in § 87〃 and § 56, N is the projection on the normal, directed towards
the centre, while T is the projection on the tangent which points to
the left as one looks along the positive normal. It will be assumed that
N(t) and. TQ) satisfy the H condition.
On the basis of (121.14), this condition takes the form
+（/） 一 =一 + iT^~] (122.8)
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 531

which is of a similar form as (122.2). However, in the present case, one

has to find a solution which has at infinity the given value T and at z = 0
a pole with the principal part equal to
可 x(X + iY) 1 .
十
2k(1 + x) z
this follows from (121.5) and (121.13). Therefore, applying the results
of § 108, one finds immediately
1，加⑴+汉⑴用工t工乂k+亚） 口
- J
1
me—
= 石 一二7— +
位 r 十
多尸不 Z + 22
(122.9)
L

The quantities X, Y (i.e., the components of the resultant vector of

the external forces) may be calculated directly from the given data;
in fact,
2k

x + iY =
—/ (N + 汀)* d*
The constants T and「'，however, which determine the stresses and
the rotation at infinity must be assumed known. It is readily verified
that the displacements will be single-valued.
The problem is thus solved. It is easily seen that for T = T' = 0 the
present expression for (z) in S_ agrees with that obtained in § 87«
(where it had been assumed that the stresses and rotation vanish at
infinity) .
3°. The second fundamental problem for S+ and
S~ may be solved in an analogous manner, beginning with (121.15). This
will be left to the reader.

§ 123. The mixed fundamental problem for a region, bounded

by a circle. This problem has so far not been solved. Let Lk = akbk
侬 = 1,2, be given arcs of the circle L, denoted in such a way
that the ends are encountered in the order 诙, , an, bn, %, when
passing around the circle in the counter-clockwise direction. The union
of these arcs will be denoted by Lr , so that
L = L] + 乙？十 ••• + Eg,
and the remaining part of L by L'f.
532 Vl. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 123

Let the components of displacement and of stress be given on L'

and Ln respectively.
Since the problem of the case, where the external stresses are given
along the entire boundary, has already been solved, the mixed problem
under consideration may be reduced to the case where the displacements
are given on the segments Lk = 倪钞比， while the remaining part of the
boundary is free from external stresses. (The solution of the general
problem may also be obtained directly ; cf. the Note at the end of this
section.)
1°. Solution of the mixed problem for the circle
Consider first the case where S'+, i.e., the inside of the unit circle,
is the region occupied by the body. The boundary condition then takes
the form
勿+
+ 谢+ = gQ) on L', (123.1)

〃+ 十勿心+ =0 on L>f, (123.2)

：
where g«) is a function, given on L It will be assumed that the derivative
g'(9 satisfies the H condition.
On the basis of (121.15), one obtains from (123.1)
x0+(/) + 一(£) = 2必'0) on L', (123.3)

=*
where

=
一"条 (1234)

On the other hand, (123.2) leads, as noted earlier, to the condition +(力) —
— 一(力) 0 on U' which expresses that (z) is holomorphic in the entire
plane, cut along
Thus the problem of finding is reduced to the determination of
a solution of the problem, considered in § 1 10, which must be bounded
at infinity. In the present case the constant g of § 1 10 is equal to — 1/x
and

/(力) = 上 g'R)・
By (110.5),

y=六题(一，)=一2^=一字十番
ie,
Y 去 沮
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 533

where
8 =生
2H
(123.5)

Therefore, by (1 10.2),

。 = n (z -
n
X (z 一 姐T+吗 (123.6)
&=i

where one has to understand by 瑞⑵ that branch which for large [ z |

has the form
X 。 =机 守 + (123.7)

Applying (1 10.18) and remembering that (z) is to be bounded at

infinity, one finds
= 兀仅 / 需*
J XJ Q) 传 — z)
L'
+ X0(z)f (z), (123.8)

where is a polynomial of degree not higher than %:

P*Z) = C/ + C*T 十… + g. (123.9)

The constants Cn have still to be determined so that they
satisfy all the requirements of the original problem, i.e., the conditions
(121.12) and the boundary condition (123.1); it will not be sufficient to
satisfy only (123.3) which was obtained from (123.1) by differentiation
with respect to A. It should be noted that (123.1) has to be fulfilled
exactly, apart from a constant which must be the same for all Lk, because
then it may be satisfied exactly by a suitable choice of the arbitrary
constant on the right-hand side of (121.3). It is easily seen that this last
condition may be expressed by the relations

[
/
反①+(M) + Ro)] 耽 = 2HgE+J — g仇)], (123.10)
万 = 1,2,...,% (瞅+1 = %)，
where +(%) and °) must be obtained from (123.8). Since +(%) =
-(%) on the arcs bkak+v the conditions (123.10) give
(乂 + 1) [J + E— 0AgCj = 2fz[g(曲+i) — g仇)],
?
( 1 23.1 1 )
^kOjt+1
534 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 123

where

4(%) =
面特(6 / 夕团也
(123.12)
兀拉 J Xq (£) (%
L'
— %)
4% = (…小 • (123.13)

One has thus obtained n linear equations in Co, Clf …， Cn. The
conditions (121.12) have still to be satisfied. It is readily verified that
the second of these is a consequence of the conditions (123.1 1), obtained
above. In fact, it follows from (123.1 1), equivalent to (123.10), that

7
值 + = 0,
L
since, by (123.3),

似①+(外
*/
]编 2a[gd)一 g(瞅)].

However, since (z) is holomorphic in S+, the integral over the first
term vanishes ; hence

and this means that the coefficient Bx in the expansion for (z) in de¬
creasing powers of z near the point at infinity is equal to zero.
Thus there remains only the first condition of (121.12) which may
be written
0(0) + (8) = 0,
so that, by (123.8),

A (r) 与
当 X°(0)g + t J/碧
Mx qt
= 0・ (123.14)
L'

Consequently, one has finally the n 十 1 linear equations (123.1 1) and

(123.14) for the determination of the constants Co, Cv . . . , Cn, or,
more correctly, a system of 2% + 2 linear equations for the determination
of the real and imaginary parts of these constants.
CHAP. 20 REGIONS WITh CIRCULAR BOUNDARIES 535

It has still to be shown that this system has always a unique solution.
But for this purpose it is sufficient to verify that the homogeneous system,
obtained for g(力) = const., has no other solution except CQ Cr ~ ...
= 。处 = 0. However, this is a direct consequence of the uniqueness
—
theorem for the mixed problem.
2°. Solution of the mixed problem for the plane
with a circular hole
This may be treated in quite an analogous manner. As mentioned in
the introduction to this chapter, the solution of this problem was re¬
cently published by B. L. Mintzberg [1]; his solution (for the particular
case 制 = 1) is somewhat more complicated than that given here.
In the present case, the boundary condition has the form
-
u~ + 谢 = g(£) on Z/, (123.15)

+ i 心一 =0 on I/'; ( 1 23. 1 6)
it follows from (123.15), using (121.15), that
+(w) + y①-0) = 2阳'(9，
(123.17)
while (123.16) gives, as before, + 一
= 0 on—
It is now required to find a solution (z) which (as in the preceding
case) is bounded at infinity and has at the point z = 0 a pole of not higher
than second order [cf. (121.13)].
It will be assumed that the stress components at infinity, i.e., the
constants T, r in (121.5), (121.6), as well as the resultant vector (X, Y)
of the external forces, applied to L' , are given.
As before, the results of § 1 10 will be applied for the solution of the
problem (123.17). This time
他 = 2阁⑹，

Y
1
'
log (一 x)
“，

C
— 3—
— —
- 一 c ・
• ，，
_
logx
， “…“
„
十 2
, .A
，

i.e.,

where 。is the same as previously, viz.

Q
l°gX
2H
536 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 123a

In the present case

n
X°(z) = 口 (z
左=1
— 外尸+ (z — 瓦尸-吗
阴
(123.18)

where one has again to select a branch, satisfying (123.7); the general
solution of the problem (123.17), satisfying the conditions stated above,
is given by [cf. (1 1 0.26)]

粤 /时需y +{多十乡「同
=—
L'
十 (123.19)

where Pn{z) is a polynomial of degree not higher than n and Dlf D? are
constants. These constants are immediately determined from the con¬
dition [cf. (121.13)] that near z = 0

。 H+V=
X
v x(X + iY) 1
。⑴ • (3。)
Similarly, the coefficients and Cr of zn and zn~r in the polynomial
are determined by the conditions [cf. (121.5)] that for large | z [

= r— J汇+*L + (二)；
2兀(比 1) z
0
\ /
(123.21)

in particular, Co = r. The values of the remaining coefficients C2> • • Cn

are found from conditions, completely analogous to the conditions
(123.10). It is easily seen that the required single-valuedness of the
displacements will then be ensured.
NOTE. It has been assumed that the part Lr, of the boundary is free
from external stresses. However, the solution of the case, where Z/' is
subjected to arbitrarily given loads, is easily written down; for this
purpose it is sufficient to refer to the statements in § 1 1 1 (cf. also the
Note at the end of § 1 14).

§ 123a Example.
This example was presented in the paper [1] by B. L. Mintzberg who started
from his own general formulae which are more complicated than those deduced
here; he was therefore obliged to evaluate several integrals, in contrast to the
present method where the solution is obtained almost without any calculations.
Other problems, treated in Mintzberg's paper, may likewise be solved in this
way.
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 537

Let a rigid stamp be applied to the arc Lf —

ab of the circular hole
of radius 1 ; let the profile of this stamp be an arc of a circle of the same
radius and let it be rigidly attached to the elastic body. A normal force
of magnitude Po which is distributed symmetrically is applied through
the stamp. It will be assumed that the stresses vanish at infinity.
Further, let the centre of the arc ab lie on the positive part of the Oy
axis, so that X = 0, Y = Po. In the present case, = 1, g(±) = const.,
= 0, T = r = 0. Therefore, by (123.19), ”
(z)
— 卜庐 + 4—
Xo(z) ，

where
X°(z) = (z — 以尸+邙 (z -
with the supplementary condition lim zXQ(z)
Zf8

r=
— 1.

It follows from (123,20), since X + = iPg that

5 = 0, Q]X°(0) = 5Td
2tc(x 十 1)

and from (123.21), since F = 0 and, for large [ z X°(z) = 尸十 0(1),

that
济。
=0， ci 一
co 一 '
2k(x + 1)
It is also easily established by investigating the changes in the arguments
— —
of (% a) and (z &) as z moves along the Ox axis from infinity to the
point z = 0 that
X°(0) =
where co is the central angle subtended by the arc ab, and consequently
”3
— 2k(x 十 】)

With those values for the constants one finally obtains

and the problem is solved.

538 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 124

§ 124. Boundary problems for the plane, cut along circular

arcs *). Let an elastic body occupy the plane, cut along the arcs
上1= Lk = 以狐 of one and the same circle. As previously,
it will be assumed that the ends of the arcs are encountered in the order
alt an) bn, % as one moves around the circle in an anti-clockwise
direction. The union of these arcs will now be denoted by Lt so that
L = Ej. + 上2+ • • • +
Let the radius of the circle be of unit length and its centre coincide
with the origin.
The solution of the boundary problems for such a body may be obtained
in a similar manner as in the case of the plane with straight cuts (§ 120).
—
A beginning will be made with (121.1) (121.3) in which now <>(z) and
里 (z) are defined throughout the plane, cut along L, and instead of
*(z) a function Q(z) will be introduced which will be defined in the
following manner:
Q = &（；）-；&（；）- 2千（；） (124.1)
whence
子⑵ = 行 ~
（7）— 7 '（以 (124.2)

It follows from (124.1) that Q(z) is holomorphic everywhere in the

plane cut along L (including the point z = 8), except at z = 0 where
it has a pole of not higher than second order. In fact, it is easily seen
from (121.5) and (121.6) that near z = 0
比 (X + iY)
Q(z) [- a holomorphic function ; (124.3)
2兀(k + 1)
if this condition is satisfied, the function 乎 (z), defined by (124.2), will
satisfy the condition (121.6).
Further, in order that 里(z), as given by (124.2), will be holomorphic
near the point z = 0, the function Q(z) must satisfy certain conditions ;
in fact, if
— + + . . . (for | z | V 1),
(124.4)
8
Q =筑H z F … (for I z | > 1),
*) To the Author's knowledge, no solution of these problems has been published
previously.
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 539

then, for 中(z) to be holomorphic near z = 0, it is necessary and sufficient

that
Aq = Bq, = 0. (124.5)
It will also be recalled that for large | z |
X + iY 1
(z) = F — (124.6)
2九(y + 1) z
Since the components of stress and displacement may be expressed in
terms of (z), 里卜)，they may likewise be expressed in terms of (z) and
Q(z). In fact,

rr + 0 心 = (z) + Q
(2)+ 2(5 x (124.7)

2M弘' 谢') iz x① (z) —

Q
（9 — W—；）庄司’（⑵
where T(z) is now given by (124.2) and
，物 ，加

It will now be assumed that for all points t = ei& on L, except the
ends ak, bk) the functions (z) and Q(z) are continuous at L from the
left and from the right and that
lim(l —少F(z) = 0. (124.9)
1

In addition, it will be assumed that near any end c

I (z) I <
const.
覆， I Q(z) | <-
const.
~ , 0 a
, …
< 1. (124.10)

By means of the above formulae, all the boundary problems, solved in

§ 120 for the case of straight cuts, may now be solved for the present
type of cuts by methods, quite analogous to those used in § 120. In
view of this analogy, consideration will here be restricted to the solution
of the first f^mdamental problem, when the stresses on both sides of L are
given, i.e., when the values of 〃+ »9+ and of 少一 勿9一on L are
known.
In addition, it will be assumed that the values of the stresses at infinity,
i.e., the values of the constants B and T', as defined by (121.8), are known.
540 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 124

Let the rotation vanish at infinity, i.e., assume that C = 0 and therefore
r = f = B.
By (124.7) and (124.9),

+(%) 十 Q-Q) =〃+ + -(%) + Q+(/) = rr~ + ir^~ on L, ( 1 24. 1 1 )

whence, adding and subtracting, one finds
@@ + Q(切+ + [① 修) + Q(切- = 2g) (124.12)

—
on L,
[＜：＞(/) Q(/)]+ — [① (%) - Q(切- 2/) (124.13)
where 夕 Q) and 夕心) are the following functions, given on L:

力@ = 虹”+
+ ”-] + 授 [r^+ + 鹏，
(124.14)
q(t) = 妥[〃+ — 〃-]
It will be assumed that
2 -
卜8+ 心
these functions
-
satisfy the H condition on 上.
Taking into consideration that the function (z) Q(z) is bounded
at infinity and has, in accordance with (124.3), at z = 0 a pole with the
—
principal part P x(X + iY) 1
1
" 1

I ' "

毅 2k(x + 1) z
one obtains from (124.13), using (108.5),
+ 汉） i
一 Q(z)
=3/曹^+ A + （x
乂

2兀（比 十 1） Z
*
z2

where DQ is a constant.
Similarly, one obtains from (124.12), using (110.26),

+ Q(z) = J ~~T~ z +亚"卜志）+

L
where X(z) denotes one of the branches of
n
x（z）= n（z — 秋产 （z
化=1 — 瓦产 (124.15)
which is single-valued in the plane, cut along L, and X(/) has been
written for X+(£), i.e., the value of X(z) on the left side of L; further,
D] and D2 are constants and
「式方) = C庐" + + •• • + Cn
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 541

is a polynomial of degree not higher than n. [Note that, in applying

(110.26), use has been made of the fact that X 。—
Thus
1 1 / q©dt
=
2^X0 LJ t-z + 2^JT^7 +
工

1 J
H 2X(z) (124.17)
+
丁
z d j 2 4兀(m 1) z
1 /X（杨修）近 1 fq幽 十
Q
不法百 J
L
一 一
mJ E
1 J 、 A 4[ A 伞 iY) 1 r7 •
+ 0 j 2 4k(乂 十 1) z 声 （⑵网

The constants and D& may be determined immediately from ( 1 24.3)

which, by (124.18), takes the form
x(X + iY) 1
(124.19)
4冗(X 十 1) z

near the point z = 0. If and D2 satisfy this condition (by which they
are uniquely determined), the right-hand side of (124.17) will be holo¬
morphic near z = 0.
The remaining constants in the above formulae, i.e.,
A，CoC， …。， (124.20)
of which there are % + 2, are determined by the following conditions:
(8) = r, (124.5) and the single-valuedness of the displacements ;
this last condition (giving n equations) may be expressed in an analogous
manner as in the case of straight cuts (§ 120, 2°). It is easily shown on
the basis of the uniqueness theorem that these conditions determine
the constants (124.20) uniquely.
There are actually % 十 3 conditions for the determination of the % + 2 constants,
i.e., there is one more condition than there are unknown constants. This is due to
the fact that the quantity X + NV in (124.3) had been assumed known, i.e., it had
been calculated beforehand from the stresses, given on the boundary. However,
one may assume that it is initially unknown and find its value together with those
of the constants (124.20) from the above conditions. Similarly, one could have left
the coefficient of 2-2 in (124.3) indeterminate in which case one would have to
retain the condition T(oo) = T'.
542 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 124a

It will be left to the reader to construct the solutions of the second

and mixed fundamental problems, where in the latter case the external
stresses and the displacements are given on the left and right edges
respectively.

§ 124a Example
Extension o f the plane, cut along a circular
arc.
Let the elastic plane be cut along the circular arc ab, let the edges of
the cut be free from external stresses and let the stresses at infinity, i.e.,
the constants T and P, be given, while F = T (implying that the rotation
vanishes at infinity). The radius of the arc will be taken as unity and
its centre at the origin ; the axis Ox will be assumed to pass through the
midpoint of the arc ab which subtends at the origin an angle 26 so
that
媒 = £一吗 b e吗 — (124.1a)
In the present case, n= \ t 力 Q) = 夕伍) =0, X = V = 0. Hence one
finds from (124.17) and (124. 18)

D] AI A
+明十 ।
+ (124.2a)

2十 A r
+ C] + ? Nj 一方 声（⑵闻
where now

X(z) = V (z ~ a) (z ~ b) = V z2 ~ 2z cos 6+1. (124.4口)

It will be assumed, that 1 for 8. It is easily seen that
under this condition X(0) = — 1 and hence, near z = 0,

z
— — 0 + 2〃 + 百蒜 +
1+3 cos 29 Z?
= — ——
1 z cos 6
4 + .... (124.5〃)
CHAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 543

Hence for small | z |

D2 D] + Z)2 cos 8 Dr cos 6
~~
2
1 + 3 cos 29 Z)2
8 + •…， (124.6t?)
and (124.19) gives
z)2 = r, D] — — rz cos o. (124.7a)
For the determination of Do, CQ, Clf one may use the formula (124.6),

。
which now has the form

= F + ( f ) for large | z | , (124.8b)

and the conditions (124.5). Since for large | z ]

1 1 cos 6
十…
I (124.9«)
入⑶ Z 炉

one obtains, by (124.2口) and (124.8g),

Cq + Dq = 2F, C] + Co cos 0 = 0.
The second condition of (124.5) does not give any new information
(as it coincides with the second of the preceding conditions) . In order
to formulate the first condition of (1 24.5), it will be noted that, by (124.2<z) ,
(124.6g) and (124.7a),
Do
—+
C] D± cos 6 1 3 cos 29
022 H——2-
八

Zq = (°) =
___
"1

2X(0) 2 8
区 44 rz sin2一9
十 一
T 2
and that, by (124.3«),

筑 = Q（8）= 上广
Co —
therefore the first condition of (124.5) gives

Co — ％ + 击 sin2 8.
+ 力° 『 (b)
The relations and (Z>) determine all the unknown constants ; in
544 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §：124a

fact, one finds

-09
_ e 4r +(r + r) sin^-co^-
Co = *(r — p) sin^- + ——
； 冰 (124.10a)
2(52乡
Cr— — Co cos 6, A = 2P — Co. (124.1 la)

It is easily verified that the displacements are single-valued ; this

condition has not been used above, since another condition, ensuring
their single-valuedness, had been introduced into the preliminary analysis.
Thus the problem is solved.

Fig. 56.

In particular, if the stresses at infinity reduce to a tension in a direc¬

tion, forming an angle a with the Ox axis (Fig. 56), then

F=

* .产. (124.⑵)
二 HAP. 20 REGIONS WITH CIRCULAR BOUNDARIES 545

If one has uniform tension p in all directions, then

r = 2,
2
r = o. （124.13a）
In this case
力
Z）1 — = 0，
1 sin2 —

9 （124.14）
今 cos 9
八 分 …
sm2 ——
2
C]= — —V
, Dq = 屋
V
1 + —
sin2 1 sin2 —

_
2 2
and hence
t— 二中 e 一十

2（l+sin^）lV1-2'COS0 + "2
（124.间）
Q（z）=
cos 0 n
. o —\ I V1 — 2^ cos 6 +1 /
— 9
— sin2 —
2J
o A . sin2
211 + I

In particular, for a semi-circular cut（9 = 兀/2）, these formulae take the

simple form

= 号 Q
Ykrfq- 共 （⑵•&）
 Chapter 21

SOLUTION OF THE BOUNDARY PROBLEMS FOR REGIONS,

MAPPED ON TO THE CIRCLE BY RATIONAL FUNCTIONS

The methods of solution of the preceding chapters are easily extended

to the cases of regions, mapped on to the circle by rational functions.
It has already been seen in the preceding Part that the first and second
fundamental problems for such regions are easily solved in closed form.
The new method, studied below, leads to the same results, and about
the same amount of calculation is involved in obtaining the final so¬
lution as would be required, when using the method of the preceding Part.
However, the present method offers the possibility of solving also
the fundamental mixed problem, and likewise some other boundary
problems.

§ 125. Transformation of the general formulae *). Let S be a

finite or infinite region in the z plane, bounded by one simple smooth
contour L, and let
z = 3(0 (125.1)

be the function which maps S on to the circle j | < 1 of the 已 plane ;

^
the boundary of this circle will be denoted by 丫 and its positive direction
will be assumed to be counter-clockwise.
If the region S is finite, the function 3(已) is holomorphic inside 丫;
if S is infinite, s(C) is holomorphic everywhere inside 丫，except at the
point corresponding to z = 8, where it has a simple pole. Without
affecting generality, it may be assumed that this point is the centre of 丫,
ie, the point〈 = 0; under this supposition

z= 3(Q=W + 3o(Q, (125“

where (o0(Q is holomorphic inside 丫 and c is a constant which is not zero.

* The results of §§ 125-127 are due to I.N. Kartzivadze [2]; some simplifications
have been introduced by the Author.
546
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 547

_
in terms of the functions (Q and
(§ 50) :
_
The formulae will now be recalled which express the components of
stress and displacement in the corresponding curvilinear coordinates
of the complex variable 乙 = p 凌*

常 藤 = 2［① (Q + 画］, (125.2)

常+ 标 = (Q +函 — 方% 诋 西丽｝ (125.3)

2从 I 3化) I (% + m) 三函
P
仅建) 一 39丽 一 语｝, (125.4)

where 中(已)， 少 (Q are related to (已)，里(0 by the formulae

中位) = 化)，少 里93化).
(125.5)
In the majority of cases it is more convenient to use instead of (125.4)
the formula
2t + 汨) = - 3(Q函
-丽
which gives the displacement components 勿，o in rectangular coordinates.
(125.6)

If 平(9，少(已) are given, the functions (Q, ¥(z) are completely de¬
termined ; however, if (Q, are given, the functions <p(0，少(已) are
only determined apart from arbitrary constants. Hence one may in the
last case rewrite (125.6)

2俨(〃 汕 — xcp(Q — co© (0

— 少(C) + const., (1256)
stressing the presence of the arbitrary constant.
It will now be assumed that 3(C) is a rational function ; the definition
of (Q will be extended into the region | | > 1 by writing
^
6GW) = 一 b9币

for | | > 1. (125.7)

^
This extension has been chosen so that the values of (C) on the left and on. the
right of 丫 extend each other analytically through the unloaded parts of the boundary
L of the region S, i.e., through the segments on which pp = p-9- = 0; in this choice
one was guided by (125.3).
548 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §425

Replacing in (125.7) C by ?-1 (| ? | = | | < 1) and going to the conju¬

gate complex expression, one finds ^
=
守
s'
(《){①(Q + (《)} — s(《)@'(9・ (125.8)

This formula expresses T(Q for | | < 1 (this function is not defined
^
for other values) in terms of O(Q for | | < 1 as well as for | > 1, «
^
The definition of 平 (0 may likewise be extended to the region | | > 1
^
by imposing the condition that in this region

89 =
J 灯© 岚; (125.9)

integrating both sides of (125.7) with respect to J one easily obtains,

after omitting an arbitrary constant,

for | 1> 1, (1257)

whence, similarly as before,

少 (9 = — & (J — & g) (？ for| j < 1 . (1258)

Thus the components of stress and displacement may be expressed

in terms of the single function (Q, defined for| 乙 V 1 as well as for
|
|^|> 1.
1 1 •■*•*■*.

The expression (125.2) for pp + 89 remains unchanged, while (125.3)

is easily seen to take the form

pp = 一 (；)+ 不{关^ 黑)}*9

碉 ｛高rW 场 （如0）
where now T(Q is given by (125.8).
In order to deduce (125.10), one has to proceed as follows : add and. subtract the
function (1/W) on. the right-hand side of (125.3) and replace the added term by the
expression, obtained from (125.8) after going first to the conjugate complex value.

Replacing 少 (Q in (1256) by the expression (1258), one obtains for

CHAP. 21 APPLICATION OF CONFORMAL MAPPING 549

the displacement components u, &

2（1（贫 说） （Q
式中
+中 3(? - 3 (")}丽 const. (125.11)

。
3
+
a%
From (50.4) follows the analogous formula

£ + 哂）一 s (/)}阚 十 const. (125.12)

In the sequel also the expression for 勿' z" will be required, where
加
%
— , v
花
This expression will be deduced by differentiating both sides of (125.6)
with respect to 9 and by transforming the resulting expression in the
same manner as (125.3) above. One thus obtains

2从("' 加') = 30 +
松化) {题T) 3(?
}WT-
亿T) 8«)

西皮厂熹m. (即5」3)

Under the usual conditions the functions (0 and 乎 are holo¬

morphic inside 丫. The behaviour of ©, extended by (125.7) to the
region R | > 1, will now be studied outside 丫；it is sufficient for this
purpose to investigate the behaviour of (C)o/(0 = 平 '(C)・
Revert to the formula (125.7) which defines (6 for | | > 1. The
rational function 3© may have poles at a finite number of points ; all
these points lie outside 丫， except in the case, where S is infinite and
where, consequently, has a simple pole for 已 = 0.
Denote by the poles of 3© outside 丫，not counting the
point C = 8 which may likewise be a pole. If the orders of these poles
are — 1, m2
—
1, . . .,mr — 1, the function 3‘(Q will have at the
same points poles of order 加曾 w2, . . . , ; further, if co© has at infinity
a pole of order 加+ 1, s'(E) will have there a pole of order m.
Thus will have poles of order not greater than m1} m2, . . . , 加 「

at the points Q, Q, . . . , J ; these poles originate from the first two terms
on the right-hand side of (125.7), because, as is readily seen, the third
550 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 125

term represents a function, holomorphic outside 丫，including the point at

infinity. In addition, the point 已 = 8 may be a pole of order not greater
than m.
It should also be noted that in the case of an infinite region
may have inside 丫，and, in fact, at
second order.
—
0, a pole of not greater than the

Thus, all possible poles of the function (已)3‘隽) and the maximum
orders of these poles are known beforehand.
Finally, it should be stressed that there will not always be a function
甲(0，holomorphic inside 丫 as required by the present conditions, cor¬
responding to a given (C), defined inside as well as outside 丫 and having
poles of the stated type. In fact, formula (125.8) shows that cor¬
responding to a given (C), may have poles at the points

}一 _L

and also at the point 已 = 0, lying inside 丫.

Expressing that T(Q is also to remain holomorphic at the stated
points, one obtains a known (finite) number of linear equations, relating
a certain (finite) number of the first coefficients in the expansions of
(9 near the points 〈；，...，C；
= AkQ + 4a - Q + 4 破收 一 蕊产 十 = 1, 2, . ・ .，y
to the coefficients of. the principal parts of the poles of the function
(Q at the points 叠，...，以；likewise, a known number of. analogous
linear relations corresponds to the point = 0.
In order to verify the above, one has to keep in mind that the principal part
of the pole of

唉)&(")
at 已［ may be obtained directly from the principal part of the pole of the function
3'g哨

_
at the point J. In fact, if near 短
Bi B.
3‘仔) =
任一 jy
+… H
c 短
p a holomorphic function,
, 一
then near 翌 =

（一1尸瓦签七
+ a holomorphic function.
任一到
Similarly for the pole at the point 乙 = 0.
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 551

These relations will not be written down here, but will simply be
denoted by
F1 = 0, 居 = 0, • ••, Fn = 0. (125.14)

They are always easily constructed for any given region, i.e., for any
given function (o(Q.
The construction of these conditions is especially simple in the case,
where s© is a polynomial, i.e.,
a© = 次 + 次2 十 …+ j+曾+1, (125.15)
when S is finite, or

3(0 = 春 十次 + … + Cm+16+i, (125.16)

when S is infinite. The function (Qs'© may then only have poles at
the points〈 = 8 and 〈 = 0 (the last pole only being possible when S
is infinite) .
Note also that, if S is infinite, one has near = 0 [cf. (50.14), (50.15)]
FcX -ViY 1
0(0^0- --±----
-^ + 27V 1) 已 +
0(1), (125.⑺
3+
子(g 购 = 一
Q2 1萼；
差一 号2tc(x 1) C 十
(125.18)
01), 十

where c is the same constant as in (125.1') and, in the previous notation,

r = B + ic = MM + NJ + >3^
+1
2；
乂
(125.19)
T‘ = B'十 iC -
一 MM Nje-2", —
while X, Y are again the components of the resultant vector of the
external forces, applied to the boundary of S.
In future, it will be assumed that ©(C), defined for j | < 1 and for
^
[ | > 1, is continuous at all points a of the circle 丫 from the left as well
^
as from the right, except possibly at a finite number of points yk — e施*
near which
const
I < , 《o a v 1； (125.20)
IE —f

in addition, it will be assumed that for all points a = 潭 of the circle 丫,

552 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 126

except possibly for the same points 丫九 = e1%,

lim (1 一 p) '(pe巧 =0, C = ”吗 (125.21)

By (125.8), one will then also have

lim (1 - p)¥(p*) =0 (125.22)

with the same reservations as before.

NOTE. 1. It is easily seen, on the basis of (125.21) and (125.22),
that the last two terms on the right-hand sides of (125.10) and (125.13)
tend to zero as p f 1, except possibly for the values b = 丫炉
NOTE. 2. In the case of infinite regions S, it is sometimes more
convenient to make use of transformations on to the region 亿|> 1
rather than on to the circle | | < 1 ; however, this distinction is not
^
of great importance. The reader will easily introduce the necessary
modifications in some of the preceding formulae.

§ 126. Solution of the first and second fundamental problems.

These problems have already been solved for regions of the type under
consideration in Part V. The formulae of the preceding section offer the
opportunity of solving these problems in a very simple manner. Consider,
for example, the case of the first fundamental problem, where the
boundary condition has the form
肃 + 勿孤+ = N iT(o) ; (126.1)
N and T(cr) are given functions of the point a of the circle 丫，since the
normal and tangential stresses are given at the point 力 of the actual
boundary L which correspond to the point a. By (125.10), one then has
一 = N3 + 汉(今 (126.2)
Thus one has reached, for the determination of (9, the same boundary
problem which was obtained in the case, when the region S is a circle
(§ 122, 1°). The essential difference arises from the fact that the unknown
function C>(Q may now have poles outside y, and this circumstance must
be taken into consideration when constructing the general solution of
the boundary problem (126.2).
From the practical point of view it will be more convenient to some¬
what modify the condition (126.2) by writing it in the form
［①⑹ 3'(切+ —［①⑹3'(切- ［N 汉⑹］3'
 CHAP. 21 APPLICATION OF CONFORMAL MAPPING 553

and by choosing (03‘居) as the unknown function. The distribution of

poles of this function has been discussed in the preceding section ; it
will be recalled that in the case, where the region S is infinite, the function
(C)o/(Q may also have a pole (of not higher than second order) inside 丫,
and, in fact, at — 0.
The general solution of the problem (126.2') has the form

(QW© = —/ … , 」3' R© (126.3)

Y

where is a rational function of 已 for which the general expression

is easily written down, since all possible poles and their maximum
orders of the function (〈)s'(9 are known.
The arbitrary constants of K© may be determined from the following
supplementary conditions :
1. The function ¥(〈)，determined by (125.8), must be holomorphic
inside 丫.

2. In the case, where the region S is infinite, the stresses must have
given values at infinity and the displacements must be single-valued.
The first condition above is expressed by. the relations (125.14) which
give a set of linear algebraic equations involving the real and imaginary
parts of the unknown coefficients ; the second condition renders similar
equations. These equations completely determine the unknown constants,
except for one real constant, in agreement with the fact that © is
only determined apart from an additive term iC, where C is an arbitrary
real constant. In the case of finite regions, the above equations will only
be compatible, provided the resultant vector and moment of the external
forces vanishes.
The above statements are a direct consequence of the uniqueness
and existence theorems.
By means of a more detailed analysis, it may be shown that these assertions are
not based on the existence theorem (but only on the uniqueness theorem) ; this
was done in the paper by I. N. Kartzivadze quoted above (cf. Note 2, § 84).

The second fundamental problem may be solved in quite an analogous

manner ; on the basis of (125.13), this problem leads to the determination
of (Q from the boundary condition

+ @6)3«)]- = 2M (126.4)
 _
554 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 127

g， = 字= i 寨， (即6.5)
g(o) = gi + &, gi and g2 being the boundary values of the displacement
components %,v.
The amount of calculations, required for the solution of the first and
second fundamental problems by the present method, is approximately
equal to that required, when applying the method of Part V. Therefore
no more will be said about it here, particularly, since the first and second
fundamental problems are particular cases of the mixed fundamental
problem which will be considered in more detail in the next section.
NOTE 1. In the case of the first fundamental problem for an infinite
region S, the conditions 2 above are expressed by (125.17), (125.18),
where the real part B of the constant T and the constant「'，determined
by the stress components at infinity, must be assumed known. The
constants X and Y may remain undetermined, as their values will be
found from the remaining conditions, referred to above. However, they
may be calculated beforehand from the given boundary values of the
stresses ; then, when requiring (〈) and T(Q to satisfy the conditions
(125.17), (125.18), one will obtain additional equations which may be
used to replace some of the other, less simple relations between the
unknown quantities.
In the case of the second fundamental problem for infinite regions,
the constants X and Y as well as T, T' must be assumed known.
NOTE 2. In order to solve the first and second fundamental problems,
one may, of course, begin from (125.12) and (125.11) respectively. This
will be especially convenient in the case of a finite region, because the
unknown function will then be single-valued. However, in the case
of an infinite region, the multi-valuedness of the unknown function is
easily removed by separating the logarithmic term, just as it was done
in Part V.
§ 127. Solution of the mixed fundamental problem *.
Let = 口抄1，乙2 = &%， •• = 口/促 be the arcs of the boundary
L of the elastic body S, numbered in such a way that the ends are en¬
countered in the order %,%,. . an> bn, when passing around L in the
positive direction. Let ...
+ Ln and L" be the remaining
part of the boundary.
* A solution analogous (in the sense of the character of the result) to that
derived here, but more complicated, was first given by D. I, Sherman [10].
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 555

Let the displacements be given on 77 and the external stresses on Lf,.

Without affecting generality, it may be assumed that L,f is free from
external stresses. The general case is also easily solved directly (cf. Note
at the end of this section) .
Denote by afc, 0% the points of the circle 丫 which correspond to the
points akf bk of E, by 丫' the part of the circle, corresponding to L',
and by 丫" the remaining part of 丫. The points afc, 除 will play the parts of
the points 丫出，mentioned at the end of § 125.
On the basis of (125.10) and (125.13), the boundary conditions of
the present problem may be written

0>+3) - = 0〃，
on y (127.1)

［①⑹3, +
+ —［①(M3)］- = /⑹ on (127.2)

where, if 取，g2 are the given boundary values of the displacement com¬
ponents u, a on U,
^2)
/⑹ = — x I da 右 h (127.3)

it will be assumed that /(cr) satisfies the H condition on 丫'.

The condition ( 1 27. 1 ) shows that 丫" is not a line of discontinuity
of the function (〈)，i.e., that (0 is holomorphic in the plane, cut along
Y‘， except at a finite number of points, where it may have poles ; the
same is, of course, true with regard to the function
For the determination of this last function one has the condition
(127.2) which is exactly the same as the condition, obtained when dealing
with the mixed fundamental problem for the case, where 5 is a circle
(§ 123,1°); however, this time the unknown function may have poles at
predetermined points ^2, . . . , J, 8, the order of these poles not being
higher than the given numbers m2, . . . , mr, m (§ 125). When the
region S is infinite, there may also occur a pole of not higher than the
second order at the point 已 = 0.
As in § 123, 1°, let
logx
(127.4)
27t
n
X。©= n(c — 小)-MB (C - 阮)T+ 单, (127.5)
556 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 127

where X°(g represents the branch for which

limbX。© = + 1; (127.6)
乙一AOO

one then obtains, by (110,26),

s«) (Q = 竽~ 声署％77 十 X°(9R© (127.7)

Y，

where R(Q is a rational function of the form

尺 (9 = 名•十单 + G 十 C式+ …+ C 馆+展憎+驾

T mr m
%
也
十£ E 7 (127.8)
= 1 丸 =1 (J Qk)

(in the case of a finite region S : = B2 = 0).

The constants Djs Cj, Bkl in (127.8) have still to be determined on
the basis of the following conditions :
1. The function T(^), corresponding to and defined by (125.8),
must be holomorphic inside y. This condition is expressed by the re¬
lations (125.14).
2. In the case of an infinite region S, the components of stress and
rotation must have given values at infinity, the components of the re¬
sultant vector of the external forces, applied to L' , must likewise take
given values and the displacement components should be single-valued.
These last conditions are equivalent to (125.17), (125.18) for given values
of r, r, x, y.
3. Finally, the fact must be taken into consideration that, if all the
above conditions are satisfied, the displacement components 勿 v will
only take on the arcs a% the given values, apart from certain constants
ck 伊 = 1, 2, … because, when solving the problem, it was only
demanded that the derivatives of % and e with respect to 卧 have given
values on these arcs. Thus, one has, in addition to the above conditions,
q= = •• =品= 0 (127.9)
which may be replaced by the weaker conditions
G = %= ... — cnt (127.10)
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 557

because, when (127.10) is fulfilled, the condition (127.9) may be satisfied

by use of the arbitrary constant on the right-hand side of (125.6')・
The condition (127.10) may be expressed in a manner, quite analogous
to that followed in § 123, when S was a circle. Therefore it will be un¬
necessary to write out the corresponding formulae.
From the conditions 1 —3 above one obtains a known number of
linear algebraic equations for the determination of the unknown constants
which will in this way be. completely determined, as is easily seen on the
basis of the theorems of uniqueness and existence of the solution *.
The solution will be particularly simple, when the boundary L contains
only two arcs 旬方】= L' and = £", i.e., when 黑 = 1 ; in that case
the conditions ( 1 27. 1 0) are superfluous.
NOTE 1. When the part L" of the boundary L is not free from
external stresses, but subject to external loads, the problem is likewise
easily solved directly. In that case the boundary condition takes the
form
-|
[①⑹ = /(g) on
(127.11)
(o)3‘(b)] — [①⑹3‘(o)] — /⑹ on
where
/3) =
x I 4 / +4 力
an Ji on Y ,
(127.12)
/⑹ N bQ) 卜⑹ + 汉⑹] on y";
N and T(o) denote here the same as in § 126. It will be assumed that
/(a) satisfies the H condition on each of the parts 丫' and 丫〃 (but that it
may be discontinuous at the points afc, 阮).
Applying the results of §111, one obtains the formula, completely
analogous to (127.7),
击
6(0= 孚~
2 z J
兀 X丁 9—
…
元
Q
(127.13)
Y
* The uniqueness theorem has been proved in § 40. The existence of the solution
has been proved in the papers by D. I. Sherman [17] and G. F. Mandjavidze [2],
referred to earlier. A somewhat more complicated study of the system of linear
algebraic •equations which has been referred to in the text permits to show that
it is uniquely soluble without operating on the existence theorem, but only on
the uniqueness theorem (cf. Note 2, § 84).
558 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 127a

where X。©and K© are the same as before, i.e., as determined by

(127.5) and (127.8); however, the integral now extends over the entire
circle 丫 and /(b) is determined by (127.12). The remaining calculations
are the same as before.
NOTE 2. One may (and sometimes this is more convenient) solve
the present problem by beginning with the simpler formulae (125.11),
(125.12)[cf. Note 2 at the end of § 126].In that case it must not be over¬
looked that, when determining the function 平(Q from the corresponding
boundary condition, this function has to remain bounded near the
points a&, as follows easily from the conditions, imposed earlier on
the function ©.

§ 127a. Example. Solution o f the mixed fundamental

problem for the plane with a n elliptic hole

,，
In the notation of § 48,5°, one has in this case for the transformation

^
on to the circle| |< 1

3(C) =R ("十 加 R> Qf 0 m < 1.

One may also (and even somewhat more conveniently) use the transformation
(127.1a)

on to the region | | > 1 ; however, the transformation on to the circle will be used
^
here, in order to be able to make direct use of the formulae of the preceding section.
The formulae (125.7) and (125.8) then take the form

s'g (0 工 R
(百 一
三泰 乙) —
(127.2a)

硝通0 一 R
—卷)叫— R0 高不 一 —
一 r代+m)*(0for HI vl,(127.3a)
while (125.17) and (125.18) become
7?r X A- iY 1
已 +
3名) (Q = 1钎 7 01), (127.
2k(x

S化 )甲(0 ^7-—^ + 。⑴ • 027.50

CHAP. 21 APPLICATION OF CONFORMAL MAPPING 559

For simplicity, assume that % = 1, i.e., that the contour L is divided

into two parts 诙恤，&旬，the second of which is free from external
stresses, while the displacements are given on the first.
Denoting by 邑 and % the points of 丫 which correspond to the points
% and 恤 of the ellipse, one has
Xo(O = K- q)+曲 何 一 。尸+吗 8 = 粤二 • (127.6。)
Let
-
3
浮i, a2 他，％ 分2= % 十方， (127.70)
where 分 is the argument of the midpoint of the arc q% and 3 is the
°
angle subtended by this arc at the centre of the circle.
For large | | 10c
+ 7r 十 …，
=v (127.8g)
where
a = 咒土
2 + 不(％ — %) = ?^fcos
\ —2 +2{3 sin —2 k /
(127.9g)

Further, it is easily verified that

X()(0) = — ,邛3一调 。 (127.10a)
so that for small|已 ［
-I 一卜曲
"
i-

X。©= X°(0) I 一 I

= X°(0)U +M + (127.1 U)

_
where

a0 = 取京 = eT%(cos£—— 28 sing~) . (127.12^)

The formula (127.2〃) shows that 3式) (Q must be holomorphic

outside 丫， including the point at infinity ; further, since 依) is holo¬
morphic inside 丫，the function 3‘(已)©(9 has at —
0 a pole of not higher
than the second order. Therefore, in agreement with (127.7),

亿) (9 =
Xo(O / 侗db
3
2病 J 豆⑹ 9
0102
一 0
十

+ {c0 + —
-j 48, (127.13«)

where D 、D2 are constants, subject to determination.

 560 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128

The values of the constants D1 and 7)2 are determined directly from
(127.4〃)；in fact, by (127.13a) and (127.1 1«), the principal part of the
pole of the function (Qs'© at C = 0 is given by

X
陪+
whence, by comparison with (127.4a),

X°(O)02
- - RF, X°(0)(A + &R)
X iY
= 2k (x 1T
+ 1)
； (127. 14〃)
the values of Dr and D2 may be obtained from these formulae.
The coefficients Co and C1 may be determined by the help of the
conditions (127.5«). For this purpose the principal part of the pole at
〈 = 0 of the function 里 as determined by (127.3霖)，will now be
calculated. It is given by
& — 加。2X0(0) Co + 而r .
- 十
-
Comparison with (127.4a), taking into consideration (127.14a) and going
to the conjugate complex value, yields

G = — r(加r 十 r‘)， G + aCi= ——2tc(x + 1) (127.15«)

Thus all the constants have been determined and the problem is solved.
—
For w 0, one obtains the solutions for the infinite plane with a circular
hole. This case was considered independently in § 123, 2°.

§ 128. The problem of contact with a rigid stamp.

1°. Statement of the problem. Uniqueness o f
solution
In very many cases, occurring in practice, the boundary problems
arise from the contact of the surface of the elastic body under consider¬
ation with the surfaces of other bodies. Several particular cases of pro¬
blems of this type have been considered in §58, §§ 115—119.
The case will be studied here where the given elastic body is in contact
with an absolutely rigid body of given shape. It will be assumed that contact
occurs along the entire boundary of the elastic body and that the surfaces
of the bodies are perfectly smooth, so that there is no friction.
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 561

To the Author's knowledge this problem was first formulated and

solved by J. Hadamard [2] for the case of an elastic sphere.
The solution of the problem for plane regions, mapped on to the circle
by means of rational functions, was given in the Author's paper [19]
and it was reproduced, with some additions, in the preceding editions of
this book. In the second part of this section a solution will be deduced
which is somewhat different in appearance, but essentially the same.
In the later work, consideration will be restricted to the plane case and
it will be assumed that the boundary of the elastic body consists of one
simple contour; however, the body may be finite or infinite (infinite
plate with a hole) . Hence one will have to deal with one of the following
two cases:
A. Case of a finite region
An elastic disc is inserted into an opening of given shape in a rigid
body (plate) ; the boundary and position of the disc, before deformation,
differs slightly from the shape of the hole into which it is pressed (because,
as always, displacements have to be small) .
B. Case of a n infinite region
Into a hole in an infinite elastic body (plate) a rigid disc is inserted
whose boundary and known position, before deformation, differs some¬
what from that of the hole. In this case it will be assumed that the values
of the stresses and rotation at infinity are given (i.e., that the constants
T, ]?' are known) as well as the resultant vector (X, Y) of the external
forces, exerted by the disc on the surrounding material. This vector is
obviously equal to the resultant vector of the forces, applied from outside
the disc (the forces, exerted by the elastic body on the edges of the
disc, are not included here) .
The boundary conditions of these problems will now be constructed,
although they could have been written down simply on the basis of
the results of § 1 15 which hold for the particular case of a straight
boundary. However, they will be approached heje in a somewhat dif¬
ferent, possibly slightly clearer, manner and certain additional observa¬
tions will be made.
First of all, since there is no friction, one will have on the boundary
of the elastic body
T = 0,

where T is the tangential stress, acting on the boundary.

562 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP §128

Next, the condition of contact between the elastic and the rigid bodies
will be stated. As indicated earlier, it will be assumed in the sequel that
contact occurs along the entire boundary. For greater clarity, attention
will be concentrated, for the time being, on the case A. Let the elastic disc
originally lie on the hole in the rigid plate (like a lid), so that its edge
somewhat overlaps the edge of the hole. Further, let the points of the
boundary of the disc, as a result of suitable forces applied to this contour,
execute normal displacements vn of such magnitude that in the end the
boundaries of the disc and of the hole will coincide. The disc will then be
inserted into the hole. The disc will now be in some state of elastic equilibri¬
um which is to be determined. Since the points of the edge of the disc can
slip freely along the edge of the hole, the tangential displacements on the
boundary will be initially unknown. However, the normal displacements
vn will be given, since they will be determined by the position of the
boundary of the disc before deformation relative to the edge of the hole.
Thus the boundary conditions of the present problem are
T = 0, =/ on the boundary, (128.1)
where / is a given real function of the arc coordinate of the contour.
Consider now the following circumstance. The process of compressing
the disc until it has the dimensions of the hole (by means of normal
displacements v„) may be performed, beginning from different positions
of the disc before deformation ; all these positions may be obtained
from some fixed position by means of rigid body displacements of the
disc (as always, one is here only concerned with small displacements) . If
one begins from some position of the disc (before deformation), different
from that on which the second condition of (128.1) was based, the
quantity / there will have a value /' which differs from / by the normal
components of the rigid body displacement necessary to return the disc
to its original position ; the boundary conditions will now be
T = 0, % on the boundary. (128.1')
However, it is obvious that the solution of the problem (128.1') may
be obtained from that of (128.1) by superimposing on the latter the
above-mentioned rigid body displacement which is known not to affect
the stress distributions.
Next consider Problem B (of an infinite region). Repeating the above
reasoning almost word for word, one arrives again at ( 1 28. 1 ) which has
now to be supplemented by the conditions, stated earlier (i.e., that the
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 563

constants T, T', X, Y must be known) . It should still be mentioned that

the rigid body displacement of the elastic plate will be purely translatory,
because the constant C = g(F), characterizing the rotation at infinity,
is, by supposition, given beforehand.
It is easily shown that the problem, corresponding to the boundary
conditions (128.1), cannot have two different solutions. In fact, it will
be remembered that in the proofs of the uniqueness of the solutions of
the fundamental problems an important part was played by the fact
that the expression X核 + 了泗
for the "difference" of two solutions vanished on the boundary (§40).
However, this expression is the scalar product of the vector {Xn, Yn),
representing the stresses applied to the boundary, and the vector (u, v),
representing the displacements of the points of the boundary. Further,
since for the "difference” of two solutions satisfying the boundary
conditions (128.1) one has 丁
=
° » =0
the vectors (Xn, Yn), (铀 e) will be perpendicular to each other, and
hence their scalar product will vanish.
Therefore, by the same reasoning as in § 40, it may be verified that the
stress components in both solutions will be identical, and consequently
the displacements may only differ by rigid body displacements.
Further, it is obvious that, if one excludes the case when the boundary
is a circle, the displacements can likewise not be different. In the case of
a circular disc, solutions may clearly differ from each other by rigid
rotations about the centre of the circle ; in the case of an infinite plate
with a circular hole, one will again have complete definiteness, because
it has been assumed that the rotation at infinity is given.
It has been shown that solutions of the stated problems, if they exist,
are unique ; the existence was recently proved by D. I. Sherman [22].
No space will be devoted to it here, but instead an effective method of
solution will be stated for regions, mapped on to the circle by the help of
rational functions *.

2°. Solution for regions, mapped o n t o the circle

by rational functions
The method below is completely analogous to that, studied in detail
in § 126 for the cases of the first and second fundamental problems.
* Regarding the case when the boundary of the body is a rectilinear polygon
cf. G. N. Polojii [1] who also gives references to his other work.
564 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128

Therefore only general remarks will be made here and the application of
the method will be demonstrated by means of examples.
In the Author's paper [19] (and likewise in. the second edition of this book)
the solution was obtained by the method, applied in Part V to the solution of the
fundamental problems.

Let the region S be mapped on to the circle | | < 1 by the relation

Z = 3©
^ (128.2)
where, by supposition, 3(? is a rational function (one may also use the
transformation on to the region | | > 1) ; the circle|已 [ = 1 will again
^
be denoted by 丫 and the positive direction on it will be taken as counter¬
clockwise.
The boundary conditions, in the notation of § 50, take the form

= 0, 吟= on the boundary. (128.3)

The expressions for and vp in terms of complex functions may be
obtained from (50.11) and (50.7) respectively. In order to deduce that
for it is sufficient to subtract (50. 1 1 ) from its conjugate complex
expression in which case one finds on the left-hand side 2s3. In a similar
manner, 的 may be calculated from (50.7). With these expressions (128.3)
leads to
炉3‘ '（b） + 3‘（6里g）｝一 滔函｛3 正）
酮｝

where all terms are to be interpreted as the boundary values of the respect¬
ive functions as C f b from inside 丫；/(b) denotes a known real function
of b, which will be assumed to satisfy the H condition.
For the present, it will be assumed that in the case, where S is infinite,
the resultant vector (X, Y) of the external forces, applied to the edge of
the hole (i.e., to the boundary of S) is equal to zero. In addition, the stresses
are to vanish at infinity.
Under these conditions, 中⑷ and 巾⑷ as well as (0 and ¥© will be
holomorphic inside 丫.
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 565

Introduce now the following functions which are sectionally ho¬

lomorphic, except for a finite number of poles :

I \ J /
+ 3'(0子(0 forJ^Kl,
(128.4)

386'(")+&'00(")]for
一

(《) (Q — 3(w)中'9 乙
中 for 一 3，(0中(c) z
|| V 1 ,

Q2(Q = ， —农3‘(C)歹(《) + (《)+ (128.5)

Obviously, the preceding boundary conditions may now be written

Qf 一 Q*) = 0, (128.6)
^(a) 一 Q§(o) = 和|皿⑹ |/(o). (128.7)
As indicated above, the functions Q】©,Q?© are sectionally holo¬
morphic, except for a finite number of poles, i.e., they are holomorphic
in each of the regions | | < 1, | | > 1, except for a finite number of
^
points where they have poles. These poles and their maximum orders
will be known beforehand, since they arise from the poles of the rational
function and from the factors L】，乙 on the right-hand side of (128.5).
It is readily seen that to each pole J inside (outside) 丫 there corresponds
a pole 装 = 1鼠 of the same order outside (inside) 丫.
Applying now the results of § 108 to the solution of the boundary
problems (128.6), (128.7), one finds
Q1(O =&© (128.8)
3
Qz© = 如/J
HZ
而
b
— C
+瑞©, (128.9)
Y

where 区式乙) and 衣2(0 are rational functions with undetermined co¬
efficients which have at given points poles whose order is not greater
than a known limit. The general expressions are easily written down, but
this will not be done here and only the following observations will be
566 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128

made. By the definitions (128.4) and (128.5) of the functions %©

and Q2%), one must have

© (|) = Q© 虱 G)= — q2(q,

whence, by (128.8) and (128.9), it is found that the rational functions

7?x(g and &© must satisfy the following identities：

(128.10)

& g) = I J(b) I /⑹ (128.11)

in order to deduce the last condition, use has been made of the fact that,
if /⑹ is a real function and if

The relations (128.10), (128.11) impose definite conditions on the

coefficients of and 衣2 (9 ；these conditions, together with others to
be stated below, serve for the determination of the above coefficients.
Applying (128.8) and (128.9) to points inside 丫， one deduces from
(128.4) and (128.5)
守3'(0粒 (")*(0 + 亿浬©} 3 &(b (128.12)

— 已3 卜'(Q - 已3'(已)少9《3’(m)建)
(128.13)
CHAP. 21 APPLICATION 0? CONFORMAL MAPPING 567

Application of (128.8) and (128.9) to points outside 丫 renders no new

information and only leads to the conditions (128.10), (128.11) which
will be assumed to be fulfilled. For this reason consideration may be
restricted to the preceding equations.
Dividing these equations by 芋3〈() and respectively and noting
that
bg 9 = 中化)， kg子© = U◎
these same equations may be rewritten

一晚一 号 合、 两+ 啪 =即）， 晔

where G(。， H&) are known functions, containing linearly a certain

number of constants, as yet undetermined.
The function 少⑷ is easily eliminated between the last two equations.
In fact, differentiating the second equation and adding it to the first,
one finds, after certain reductions,
3 + 1) + xQ式刖？= G(Q + 牙 (Q, (128.16)
where

0(9=
嗔) (>28.17)

Thus the function <p(0 satisfies the linear, first order differential
equation
t«)+v 就r9=T9， (128.18)
where
“
眸19)
(…)Q(9
is a known function, containing linearly a certain number of undeter¬
mined constants, and
V = —十
比
—1 a
- <V < 1). (128.20)
568 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128a

Integrating (128.18), one obtains

69 = [Q [K 十 F(凯Q(9了照， (128.21)
where K is a constant.
Having found 勺 (Q, one may determine from (128.15). The un¬
known constants in the expressions for and 少(〈) may be determined
from the conditions (128.10), (128.11) and also from the requirement that
these functions are to be holomorphic inside 丫.
It has been assumed in the case, where the region S is infinite, that the
stresses vanish at infinity. This condition is not essential. If it is assumed
that the stresses have given finite values at infinity, the preceding reason¬
ing will remain valid. It must only be noted that in the case under
consideration the functions 勺 (Q and have first order poles at = 0
with known principal parts which can only affect the form of the rational
functions &(Q.
In addition, it has been assumed that in the case of infinite regions the
resultant vector (X, Y) is equal to zero. If the vector (X, V) is not zero,
the corresponding problem is easily reduced to the preceding one (cf.
the second example of the next section).
NOTE. In a quite analogous manner can be solved the somewhat
more complicated problem obtained by replacing T = 0 by the condition
that T is equal to a given function of t.
§ 128a. Examples.
1°. Circular disc. In this case
z= = KJ (128.1a)
where R is the radius of the disc and the boundary conditions (128.6)
(128.7), written explicitly, take the form (after dividing the first equation
by R2 and the second by 五)

+ ci2x(o)]+ — [—
/

(
一 H ^子( )]
一 " (128.2d)

The subsequent calculations will be somewhat simplified, if it is assumed

that (128.4a)
中(0) = 0,
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 569

and this may be done without affecting generality.

Solving the boundary problems (128.2媒) and (128.3〃) and taking into
consideration that the functions

*(")+*G)， G) + (9 +江 一春 守

are holomorphic for | | > 1 [where it follows from (128.4〃) that the se¬
^
cond of these functions is holomorphic for = oo], that the first of these
functions vanishes at infinity and that the functions
他的 + 可©
are holomorphic for | | < 1, one finds that inside 丫
^ 典化) +
_
y 建) 一 mq — 国(0

0, (128.5〃)
, , 乂 、 2[i C fMda
(128.6a)

where 龙 is a constant. The condition (128.10) is automatically satisfied,

while (128.11) gives
27r

以 十万 =— 7vz J
/⑹ a = — kJ
/⑹游. (128.7说)
y o

(The multiplier | s'3) | = 7? does not appear on the right-hand side,

since (128.3〃) had been divided by R)
Comparison of (128.50, (128.6a) with (128.14), (128.15) shows that
in the notation of § 128
40 4- a
G© = 0, H(0 = — , (128.80

where

瓶= 生/及
J —咚.
q
Tzi o
(128.明
Y

By (128.17) and (128.19), one has in the present case

Q©V， F(Q =
4(0 —149
X + K +1
570 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128a

and, by (128.21),

[ qrg — 4Q〕b +工 ,
—
2V
建) = 苑
X + 1J x 1

where K is a constant ; this last formula may still be written

69 = K 十 1J
0
/
[S'© — 49 + 4(0)]U2V岚 +
dC
——
40)
H (128.10«)
乂 一 1 x — - 1

The lower integration limit in the last formula is justified, because,

as it is easily seen, the expansion near the origin of the expression.
_ jr(。狞
[4© — 月(o) 2V

begins with a term multiplied by b2v+2, anj 江 is known that

2x
] V 2v = V 2.
x + 1
The constants K and a must be determined from the condition that
平 (Q is holomorphic inside 丫 and from (128.7a), because (128.4a) is satisfied.
It is obvious that 少(0 will be holomorphic if, and only if, K = 0,
because 2V is not an integer.

morphic function, because the multi-valued multiplier ，

The second term on the right-hand side of (128.10a) is easily seen to be a holo¬
2V under the integral
sign compensates the multi-valued factor ^2v outside the integral. Finally, it
should be realized that the branches of C2 V and L2 V must be chosen, such that
]
2V
) =
The condition (128.7e) determines the real part of 许 its imaginary
part remains arbitrary, as was to be expected (because it only affects
rigid body motion). Assuming this imaginary part to be zero and noting
that the right-hand side of (128.7〃) is equal to
况 = 痣
/ (0), one obtains
= — *4(0),
—
and (128.10g) finally gives

0
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 571

by (128.6a) one finds now

=
合 吸)— "o + 婴'
一 (128.⑵)

where it is readily verified that the right-hand side is holomorphic at

C = 0. Thus the problem is solved.
2°. Infinite plane with a circular hole
In this case the transformation on to the region | ] > 1 will be applied,
so that (128.⑷ remains true. ^
If / denotes the normal displacement, assumed positive when it is
directed inward, i.e., towards the centre, the boundary conditions take
the form (after division by R2 and R respectively)

[b '⑹ 十 里⑹厂
- 0, (128.13«)

(128.14)

For the present, it will be assumed that the resultant vector (X, Y) is
equal to zero and that the stresses and rotation vanish at infinity. Then
will be holomorphic for | | > 1, including the point at infinity,
and for large | 乙| ^
(0 =0 延， 子©
目；
=o
in addition, it may be assumed, without affecting generality, that
= 0.
Normally it has been assumed in such cases that 中(8) = 0; however, one may
put instead
convenient.
— 0. In the present case this last assumption is somewhat more

Taking into consideration the stated properties of the unknown func¬

tions and solving the boundary problems (128.13d) and (128.14a), one
finds for points of the region | | > 1
^
《）+守子g
典， =见 （128.150

*+ 一
1 =当/然 ' （128J6a）
572 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128a

where a, b are certain constants which may be determined in the following

manner. By (128.10^)
a —— 倪,
and by (128.1 la) (remembering that R had been eliminated)
2e

b+ 一丝卜 6 也 = 一生卜
E J a kJ
y o
In addition, letting 已 f co in (128.15。)，(128.1 60) and noting that

［守子(Qk =8= 一』［贝(须=8，

one finds
力 = Ra,
and hence
2k

b= - Ra = /0)於. (128.17〃)
o
The relation［芋乎(已)卜=8 = — ［®(C)k=8 above is obtained in the following
manner. Remembering that
日
以
3式)
0— MU)
R
，

so that, if for large|已|

WC) =
压 + o(摄)
one has

*(q=一焉+。假).
Comparing (128.15a), ( 1 28. 1 6a) with (128.14), (128.15), it is readily
verified that in the present case
40 力

-产
b
G(? = 即)
飞 X
where A(Q is given by (128.9); however, 已 lies now outside 丫.
Thus one arrives again at (128.18), where now
G(Q + H0 一 49
的
® —
一
比十 1)Q(9
一
一 ~~

r；
1
•
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 573

By (128.21), one has

g(Q
- XL 十 二+
X 1
［月9 一 s，9狞2位；
J
8

the choice of the lower integration limit is justified, because the integral
is easily seen to converge ; on the other hand, this limit may be chosen
arbitrarily.
It is obvious that <p(Q will only be holomorphic, if K = 0, because 2V
is not an integer (see above) . Hence
w
吸) x+1J
仙g - 2V
戏. (128.180
8

The function 少© may now be found from (128.16a) which gives

比 11b
怵) = 话 咪) 一 歹料0 4? + (128.19。)

where b is given by (128. 17办

Hitherto it has been assumed that the resultant vector of the forces
(pressures), applied to the plate from the sides of the disc, is equal to
zero. If it is finite (still assuming the stresses and rotation to vanish at
infinity), then, by the same method as in the analogous cases of the first
and second fundamental problems (cf. § 78), it is readily seen that the
solution is given by
述) 十 g)+ %©
where 平(已)， 小⑷ are the same functions as above, while

—— + iY 、
iY X
o= - o +
X ,
2&200
"
八

一，

—
如
2兀 (1 + X)
Iw 已 T
ZKX

必© =
乂 (X
-3 — L iOg “ 3(乂 一 l)(X
EY)|
— (1 畔( —
+ zY)L .
T 1 28.21。)
2加 (1 比) 4k + x

It is likewise easily verified directly that 勺。(已)， solve the present

The solution 9 + <p0> +

—+
boundary problem for / 0 and for a given resultant vector (X, Y).
corresponds to the case, where external
forces, the resultant of which is equivalent to the force (X, Y) applied
to its centre, act on the rigid disc.
574 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIP § 128a

If this force did not act at the centre, equilibrium of the disc would be impossible,
because, since T = 0, the resultant moment (about the centre) of the forces,
applied to the boundary of the disc, is equal to zero.

When the stresses do not vanish at infinity, but have given (finite)
values there, the corresponding problem is likewise easily solved.
3°. Infinite plane with elliptic hole
As in the case of the first and second fundamental problems, one might
use here the transformation on to the region | C | > 1. However, use of
the transformation on to the circle j | < 1 simplifies the calculations
somewhat. ^
Thus, let
z = 3(? = R c),
物 7? > 0, 0 V wz v 1. (128.22a)

Then

s'(C) =
— (1 — 加 石
(《)= (已 + 丁)五
'
(128.22®
苏 (J = K的 一 6.
It will again be assumed that the stresses and rotation vanish at
infinity and, in addition, that the resultant vector of the forces, applied
to the boundary of the hole, is equal to zero (the general case may be
reduced to this case.)
Under these conditions 中(Q and 3© will be holomorphic inside 丫 and,
in addition, near the origin

峭
=器=03 ”0=露= 。管）. 偿3

On the basis of these formulae it is readily verified that the function

Qi(0，defined by (128.4), is holomorphic inside as well as outside 丫
<
(including the point = 8).
However, the function Q?©，defined by (128.5), may have at = 0
a first order pole with the principal part
R
后 加中 (0) + 3(0)}.

The following observation will now be made which considerably

CHAP. 21 APPLICATION OF CONFORMAL MAPPING 575

simplifies the further analysis. It is known that addition of an arbitrary

complex constant a to g© together with addition of xa to WQ will not
alter the displacements (and, in consequence, not the stresses). It is easily
seen that this constant may always be chosen such that [remembering
that 加 # 1 (加 v 1)]
xw<p(0) + W(0) = 0. (128.24〃)
Thus, without affecting generality, it may be assumed that (128.24©
is fulfilled.
Then Q?© will be holomorphic for | | < 1 and, as is easily seen,
^
also for | | > 1, including the point at infinity. Therefore the functions
^
and 五2(0 亩 (128.8), (128.9) are simply constants which will be
denoted by a and b respectively and (128.12) and (128.13) may be written

(y) '© 位)于©}

?3位) 即 (128.25a)

一 五 (J 5(0 — 3名)怵) +^f 建) 月© 6 (128.260

^
(for [ ] > 1), where
/⑹|3’ [ do (128.27©
出Q

The conditions (128.10) and (128.1 1) give

dcr
a a , 占十万= 一当
7T2 J
|3‘(tr) | =
y

(128.28«)

Comparing (128.25仇) and (128.26a) with (128.14) and (128.15), it

is seen that, in the notation of § 128,

' ?—
G© = - (9 - -]
.

Finally, noting that in the present case

K G)
Q(Q=
w3亿)
=_ 纥 乙2
1 一砒2，
工
576 VI. APPLICATION OF THE PROBLEM OF LINEAR RELATIONSHIPS 128a

one obtains from （128.19）

明e） 2b
山一
五（乂 +1） 项比 1）例 一 的（1 3?） —
d b
+ N7~H —7一^?，（12829〃）
+ 1）的一 ？）
1T
where

m）= — 十要安4成 （128.300）

The function 平© is given by（128.21）which now becomes

次2 T
F（，）（f" （128.3
1 一
建）
\ m — / 3 ?
—V m

where K is a constant. The integrand has inside 丫 only two singular

points ± Vm, because m < 1, and it is easily seen that the integral
converges（remembering that v > 0）. Further, clearly the - second term
on the right-hand side of （128.31«）remains finite as C— — 5/丽. Thus,
for （p（Q to be holomorphic near 已 =
Hence
—
/薪，it is necessary that K = 0.

建）=（—）｝）
—^ （三讲
\m /
—7 m
J \1— 加 U /
• （2）

“
Further, for 中（？ to remain finite for 十 丽， it is obviously
necessary that
T-

/ （128.33〃）
J \ 1 — 加 w / 丝=0.
— Vm
If the condition（128.33a）is satisfied, the right-hand side of （128.32a）
is easily seen to be holomorphic inside 丫. Substituting from （128.32〃）in
（128.26〃）, assuming （128.33〃） to be satisfied, an expression is found for
W（0 which will clearly also be holomorphic inside 丫. It is likewise readily
verified that （128.24。） is fulfilled.
There only remains to determine the constants〃and b in the expres¬
sions for 中（Q and “（Q. For this purpose one has the relations（128.28a）
CHAP. 21 APPLICATION OF CONFORMAL MAPPING 577

and (128.33办 It may be assumed that the integral in (128.33口) is taken

—
“
over the segment 拓 “丽 of the real axis and that on the path
of integration
/ w y
\1
is positive.
The condition (1 28.33a) may be written
/ + (2& - K^b + Kza = 0, (128.340
where
+Vm

(3.350

(128.36a)

The constants & and K2 are real and it may be assumed that they
have been calculated once and for all for ellipses of any given eccentricity
(determined by 加). It is easily seen that K* < K】. The quantity / may
likewise be assumed known, since /⑹ will be given.
Equation (128.34。)，together with (128.28。)，determines a and b. In
fact, subtracting from (128.34a) its conjugate complex equation, one finds
7-7
b
— 24 员
—
b
一
(128.37a)

which, together with the second condition (128.28a), determines b. After

this the constant a may be found from (128.34。).
Thus the problem is solved. The present solution is easily generalized
to the case, where the stresses at infinity have given finite values and
the resultant vector (X, Y) is different from zero.
A
 PART VII

EXTENSION, TORSION AND BENDING OF HOMOGENEOUS AND

COMPOUND BARS
A
 —
The first three chapters (22 24) of this Part are reproduced here in the same
form as in the first (1933) and second (1935) editions of this book, apart
from minor editorial modifications. In Chapter 25 of the third edition, the，
study of the solutions of the problems of extension and bending by couples of bars,
consisting of different materials with, different Poisson ratios (§§ 146, 147,
149), has been greatly extended. Further, a section (§ 150) has been added which
deals with a solution, due to A. K. Rukhadze, of the problem of bending of such
bars by transverse forces. The present (fourth) edition presents the text of the
third edition without essential changes.

As there is no space even to touch upon the interesting results of A. Ya. Gorgidze
and A. K. Rukhadze, referring to the (approximate) solution of the problem, of
extension, bending and torsion of almost prismatic compound bars, as well as to
the calculation of "secondary effects" for prismatic compound bars, the relevant
papers will only be listed here: A. Ya. Gorgidze [3—10], A. K. Rukhadze [4 7],
A, Ya. Gorgidze and A. K. Rukhadze [2, 3].
—
In the present Part, the problems of extension, bending and torsion of
cylindrical (prismatic) bars will be considered, since they are of great
practical importance.
Chapter 22 is devoted to the classical results, referring to the problems
of torsion and bending of homogeneous bars (the solution of the problem
of extension being trivial in this case), which are, in principle, due to
Saint-Venant. Since these results are studied with sufficient completeness
in almost all text books on the theory of elasticity, only the basic theory
will be presented here; certain results which are due to the Author
and represent applications of complex function theory will be studied in
greater detail with examples.
The remaining chapters of this Part give results, referring to the
problems of extension, torsion and bending of compound bars which
arise in connection with certain problems of civil engineering, such as
those of reinforced concrete. In principle these results are due to the
Author.
581
A
 Chapter 22

TORSION AND BENDING OF HOMOGENEOUS BARS

(PROBLEM OF SAINT-VENANT)

§ 129. Statement of the problem. Consider a homogeneous isotropic

bar, bounded by a cylindrical (prismatic) surface ("side surface") and
two planes ("ends"), normal to the side surface. It will be assumed that
there are no body forces present, that the side surface of the bar is free
from external stresses and that given forces (satisfying, of course, the
equilibrium conditions of the body as a whole) are applied to its ends.
The Oz axis will be directed parallel to the generators of the side surface
and the plane Oxy chosen to coincide with the "lower" of the ends of the
bar. The "upper” end of the bar will then have the coordinate z ~ I,
where I is the length of the bar.
The complete problem of the elastic equilibrium of such a bar under
the stated conditions then leads to the following mathematical problem
(§ 20): To find the quantities XXf Zz, Yz, Xy> ut v, w which
satisfy in the region V, occupied by the bar, the equations

加 + 效
十 …
一 0,

用
a% +—
效
1
— 0* (129.1)

见 见 应
H = 0,
a% 效

—
加 加
Xx = 入9 + 2pt 淑
, Yy = X0 + 2p,~ —
效
, Z?
— XO + 2M
Sz
,
(129.2)

where
彻 彻 dw
加 + 妙 + Sz

583
584 VII. EXTENSION, TORSION, BENDING § 129

and, in addition, the following boundary conditions :

Xx cos (为, 劝 + Xy cos 例,夕) = 0 '

Yx cos (”，％) + Yy cos (修，夕) = 0 } on the side surface ; (129.3)

Zx cos (为，％) + Zy cos (”, 夕) = 0
finally,
Xzt Yzt Zz equal to given functions at the ends, (129.4)
i.e., for z = 0, z ~ I.
The problem, when formulated in the above manner, presents con¬
siderable mathematical difficulties, particularly, if one is not only
interested in its theoretical solution, but also in a solution which permits
effective calculations.
Fortunately, it has been found that in the majority of practical
cases it is unnecessary (and even senseless) to consider the problem in
such completeness. In fact, the actual distribution of the external stresses
at the ends of the bar is rarely known ; the resultant vector and moment
of these stresses will be known more or less exactly. In other words, the
union of forces and couples, statically equivalent to the resultant of the
forces applied to the relevant end, will be given.
On the other hand, by Saint-Venanfs principle (cf. §23), if one is
dealing with bars which are of great length in comparison with the
dimensions of the ends, one only needs to ensure that the resultant vector
and moment of the forces, applied to the ends, will have given values ;
the actual stress distribution at the ends, however, will have negligible
influence on those parts of the beam which are not close to the ends.
Thus there appears to be a rather wide choice of solutions. This
arbitrariness may be used to simplify the problem in the following
manner : one may prescribe beforehand the form of the solution which
must, however, be sufficiently general so that one can obtain on the ends
of the beam a stress resultant, statically equivalent to that given ("semi¬
direct method” of Saint-Venant).
In this connection it is only necessary to consider one of the ends.
In fact, having given the resultant vector and moment of the forces,
acting on one of the ends, these quantities will also be determined for
the other end, since the sum of the forces, applied to the ends, must be
statically equivalent to zero (i.e., it must satisfy the equilibrium conditions
of the body as a whole). On the other hand, each solution of the equations
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 585

(129.1) always gives a stress distribution on the surface of the body which
is statically equivalent to zero (cf. end of § 20).
The complete theoretical solution of the problem with the above simpli¬
fication and its application to a number of technically important cases
is mainly due to Saint-Venant.
Saint-Venant's results are studied in his two extensive memoirs
[1, 2] and in a number of other publications, in particular, in the lengthy
notes in the French translation of A. Clebsch's book [2].
A. Clebsch (1833—1872), who was considerably younger and died
earlier than his contemporary Saint-Venant, gave a very strict solution
of a problem which is of interest here (A. Clebsch [1,2]); he showed
that, if one introduces beforehand the condition
Xx=： Yy = Xv = 0 in the region V, (129.5)
there remains just sufficient arbitrariness to satisfy the conditions at
the ends and on the side surface, and that this condition leads to the
solution, obtained by Saint-Venant by another lengthier method. Clebsch
called the problem of the determination of the elastic equilibrium of a
cylinder (with unstressed side surface) under the supplementary con¬
ditions (129.5) the "problem of Saint-Venant>,.
The condition (129.5) obviously has the following physical meaning :
if one imagines the given cylinder to consist of a number of longitudinal
"fibres" (i.e., thin longitudinal prisms), these fibres exert neither direct
nor shear forces on each other in transverse directions (i.e., the fibres may
only exert on each other cohesive forces in the longitudinal direction) .
If (129.5) is satisfied, the conditions (129.3) on the side surface ob¬
viously reduce to
Zx cos (%,%) + Zy cos (肛 y) = 0, (1293)
because the first two conditions of (129.3) are automatically satisfied.
The method of Clebsch will not be studied here (it can be found, for
example, in A. G. Webster [1] and also in I. Todhunter and K. Pearson
[1]) ; a less strict, but simpler method will be applied instead which
agrees, in essence, with that used by A. E. H. Love [1], Chaps. XIV and XV.
It should still be noted that the results of Saint-Venant may be
obtained by beginning from the following formulation of the problem
which is due to W. Voigt (cf. A. E. H. Love [1], Chap. XVI): To find the
elastic equilibrium of the cylinder under consideration on the basis of
the supposition that the stress components are linear functions of z.
Consider, for definiteness, the forces applied to the upper end. The
586 VII. EXTENSION, TORSION, BENDING § 130

union of these forces is statically equivalent to a force, applied to some

(arbitrary) point O', and to a couple. The point of intersection of the Oz
axis with the upper face will be chosen as the point O'. The force may be
divided into two components : one in the Oz direction and the other at
right angles to the Oz axis. The couple may be decomposed in the same
manner : the moment of one of its components will be taken parallel
to the axis Oz ("twisting couple”)，while the moment of the other will
be in the plane of the end ("bending couple").
Correspondingly, the problem may now be divided into the following
four component problems :
笈) torsion by couples, acting in the plane of the ends;
b) extension (or compression) by longitudinal forces, applied to the
ends;
c) bending by couples the plane of which is perpendicular to that of
the ends;
〃) bending by transverse forces, applied to one of the ends and acting
in its plane (one must, of course, in consequence apply to the other end
a force which is equal in magnitude and opposite in direction to the
above ; the same is true with regard to the couple, in order that the entire
system of forces be in equilibrium).
It must be remembered that in the subsequent work one is not dealing
with concentrated forces and with pairs of concentrated forces, but
with forces and couples which are statically equivalent to certain stresses
distributed over the ends.

§ 130. Certain formulae. In order to facilitate future reference,

it will be recalled that the equations (129.2) may be replaced by the
following which are equivalent to them (§ 19) :
加 1
加
—= 1
bZ + ZJ], —= ~ + XJ],
1
加

oz —宕
已
[Zz — a(xx + yj],
(130.1)
加 加 2(1+b du dw 2(1 + er)
织丁加 E 物 讯 E
加 3%
1 dy ==
2(1 +
— er)
Xz
dx E
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 587

where E is Young's modulus and a Poisson's ratio which are related

to 入 and fi by the formulae (§ 19)
J（3入 + 2© 入
七
七 = or = (130.2)
入十E 2（入 十 R）
Eg E
(130.3)
+ + C)
尸
(1 b)(l - 2力 2(1
Finally, the compatibility equations of Beltrami-Michell (§ 22) will
be reproduced which in the present case, with absence of body forces,
have the form
82® 1
1 —
+ —— -
= o. ' = o,
G 1 +b 即彻
济® 1
十一 = 0, 十 = 0, (130.4)
1 + CT 1 + b 次讯

济®
以十圭
?2® 1
乌 AXy + 1+0 = 0,
豆L '
dxdy
where
® = X* + 匕 十 Zz. (130.5)
Each set of functions Xx, .. Xy, satisfying these conditions (which
will, in future, be simply called compatibility conditions) and the equa¬
tions (129.1), gives a certain possible stress distribution in the body (under
the requirement of single-valued displacements).
In the sequel, the set of equations ( 129.1 ) and (129.2) will be called static
equations 0/ elastic body, while the equations (129.1) will, as before,
be called equilibrium equations.

§ 131. General solution of the torsion problem. The solution of

the component problems will now be considered and a beginning will be
made with the problem of torsion.
Let the coordinate axes be as in § 129. The coordinate system will be
assumed to be right-handed. Let the forces, applied to the ends, be
statically equivalent to a twisting couple, i.e., to a couple with moment
vector at right angle to the plane of the ends. Let M be the (scalar)
moment of the couple, acting on the upper end (M > 0, when the couple
tends to twist counter-clockwise, looking upwards, since, by supposition,
the coordinate system is to be right-handed).
588 VII. EXTENSION, TORSION, BENDING § 131

The first idea to enter one's mind is that all transverse sections of the
cylinder will remain plane and that they will twist (in their own planes)
about the Oz axis by some angle e. If the lower end is restrained from
moving, it is natural to assume that the angle z is proportional to the
distance z of the section under consideration from the lower end, i.e.,
e = tz, (131.1)
where t is a constant which measures the angle of relative twist of
cross-sections, unit length apart. For this reason, t is called the relative
twist.
Under the present suppositions the displacement components will be
given by
%=
— —
zy = Tzy, o = xzx, w = 0
(since an infinitely small rigid rotation through an angle e in the 。町
。
plane about the origin leads to 四 = — 引， = e%). Calculating the
stress components from these displacements, it is easily seen that the
equations (129.1) will be satisfied ; however, it is also readily verified that
the conditions (129.3) may not be fulfilled, unless one is dealing with
a circular cylinder (this will become obvious on the basis of the later
work). It is therefore clear that a too restrictive hypothesis has been
introduced.
The following investigation will now be based on the assumption (which
will be found to be successful) that the cross-sections do not remain
plane, but that they warp (and that all cross-sections warp in an iden¬
tical manner).
This supposition obviously leads to the following expressions for the
displacement components :
我 =
— xzy, v = tzx, w = 叫)(％, y), (131.2)

where t is a constant (relative twist) and <p(^, y) is some function of

%,y, to be determined later (the factor t has been introduced into the
expression for w simply for the sake of convenience).
The formulae (129.2) give for the stress components, corresponding
to the displacements (131.2),
(131.3)
and
Xx = Yv = Z% = Xy = 0. (131.4)
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 589

Substituting these values in the equations (129.1), one sees that they
will be satisfied, provided

十 2
(131.5)
如
In other words, 聿 must be a harmonic function of the two variables x, y
in the region, occupied by the body; since 9 does not depend on z, it
is, of course, sufficient to consider any cross-section S of the cylinder.
Further, the condition (129.3') (expressing absence of external stresses
on the side surface) takes the form
\ /%、
— y} cos (%,%) 十 { ———b
x) cos (n, y) = 0 on L,
where L denotes the boundary of the region S and n the outward normal
to L (i.e., the normal, directed outwards from S). Noting that

兽 cos (%,%) 十兽 cos (曜, y) = 华 ，

ex oy on
one obtains finally the boundary condition in the following form:

= y cos (%,幻一％ cos (为，夕) on L. (131.6)

Thus the function 9 which is called the torsion function must satisfy
the following conditions : it has to be single-valued (because otherwise
必 would be multi-valued and such multi-valued displacements will not

be considered here) and harmonic in S, and on the boundary of this

region its normal derivative must take a previously given value, in fact,
the value
y cos (%,力) 一 ％ cos 仇，y).
The problem of finding cp is thus a particular case of one of the fun¬
damental problems of potential theory
— which has already been mentioned in § 77.
—
"the Neumann problem"

The Neumann Problem, i.e., the problem of the determination of a

function y), harmonic in a region S, from the boundary condition
的
'二^
，on T ,
L
an
where / is a continuous function given on L, has a solution if and only
590 VII. EXTENSION, TORSION, BENDING § 131

if the condition

is satisfied, where is the element of arc of the contour L ; if this con¬

dition is fulfilled, the solution is determined exactly apart from an
arbitrary constant term*. In § 140, the solution of a more general
problem will be given.
This constant is not essential, because replacement of by 中 const,
does not change the state of stress, as is readily seen from (131.3) ; it
only gives rise to a rigid body translation of the beam in the direction
of the Oz axis, as follows from (131.2).
It is readily shown that in the present case the condition for the
existence of a solution of the Neumann Problem is fulfilled.
In fact, assuming that one has chosen as positive direction of L that
which leaves the region S on the left and denoting by $ the arc measured
on L in this direction, one finds

cos (”，％) =
dy
cos (力 y) = —— ,
as
cos (% y) = — cos (力 ％) =
dx
—,
ds

where t is the positive tangent. Hence

L
f ds =
L
[y cos (% x) — x cos (n, = J
L
{y dy x dx) =

as was required to be shown. 工

= J 竭(炉 + 词 —. 0,

Thus, the function 平(力, y) can be determined by solving a Neumann

Problem.
Further, (131.3) and (131.4) show that the ends of the bar are only
subject to tangential stresses.
It is easily shown that, if 甲 satisfies the above conditions, the resultant
vector of the stresses is equal to zero, i.e.,

s
]]Xz 菽 dy = G, jj
s
丫工 d% 必 = C. (131.7)

* It goes without saying that the statement is true for several (very general)
conditions, imposed on the boundary L of the region S.
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 591

In fact, by (129. 1)

whence

j jxzdXdy = f j[xz dx dy =
8 s
X) d(xY ) ] 「
——
光
। " j dx dy = x{Xz cos (nx) + Yz cos (”y)}ds;
?
L

however, the last integral vanishes by (129.3'). This proves the first
formula (131.7). The second formula can be proved in the same manner.
The resultant moment of the external stresses, applied to the upper
end, will be given by

m= yy
L
— 以z川4 dy = w yy（婷
L
十川 + 得 兽） 的,
一夕 。％

i.e”
M= (131.8)*
where

『"6
D
+a琮 一夕鲁"s
犯 (131.9)

The formula (131.8) shows that the torsion moment is proportional to

the relative twist T. The coefficient of proportionality D is called the
torsional rigidity. It is seen to be the product of the shear modulus and
a quantity which only depends on the shape of the cross-section, and
not on the material.
“
Once the torsion function 干 has been determined, the constant D
may be calculated. It will be shown below that D is essentially positive.
Therefore the constant t will be given by (131.8) for a given couple, i.e.,
for a given value of M, and the problem is solved.
It only remains to prove that Z) > 0. This result is most simply de¬
duced by an investigation of the potential energy, stored in the twisted
bar. In fact, it is known that this energy is given by (cf. Note at the
end of § 24)
592 VII. EXTENSION, TORSION, BENDING § 131

°must* ff Wg% +over

be taken
y%" + z眼)as,
the entire surface of the bar.
where the integral
However, in the present case, the integrand vanishes on the side surface
and on the lower end, so that there only remains the integral over the
upper end, where (since z = 2)
% ==
—rly > v t/%, X” == X；?， 丫力 Vz， Zg === = 0;

〃
hence

" =务
s
ff (%工 +
s
= -y (%工 — yX^dx dy = 工 |巴
and, since in the presence of deformation U > 0, one finds D> 0 and the
above assertion is proved.
This result may also be proved directly. In fact, applying Green's
formula and using (131.6), one obtains

〃(r -嚏" =//(驾 誓)勿

S
砂
S
的

— J，勺{y cos (n, x) — x cos (为，y)}ds = — J/ 9 an ds.

L L

However, one has by a known formula for every harmonic function 9

/小
L S
初i
Thus one finds from above

°="{W一不+偿） 管）}
s
十 获孙

Multiplying this equation by 从 and adding it to (131.9), one obtains

D
dx
2
+ + %] ^dx ( ⑶ 9)
s
and hence the assertion follows.
If one had 0 = 0, this would imply
娜
= 一4
纱
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 593
throughout S ; however, this is impossible, because y dx — 力 dy is not a perfect
differential.
NOTE. 1. One may obviously add to u, 巩 w, respectively, terms
of the form
a+ ry, —
+ rx pz, 丫 + 力y qx, — —
expressing rigid body motion, without affecting the state of stress.
NOTE. 2. Since the above work was based on the formulae (131.2)
the first two of which express rigid rotation of the cross-section about
the axis Oz, it may be shown that a new solution of the problem is obtained
by replacing this axis by another one, parallel to it. In fact, let 0式《,b)
be the point of intersection of the new axis with the plane Oxy, then
— —
tz(y 6), 巧 = rz(x a), —
— 叫式％, y), (1312)
where ulf 缈 are the displacement components and
中】 is the torsion
function, corresponding to the new position of the axis. The corresponding
stresses will be given by

Xz = [XT — y 十 方)， Yz — “g + %― •
fl). (131 .3')

As above, it may be shown that 钓 is harmonic and that it satisfies

on L the condition
”华 i
—— = (y
dn — b) cos (%,%) — (% — a) cos (%,y) =
= y cos (笈，％) —— cos (n, y) — b cos (%,%) 十 0 cos (%,y)
％

which may obviously be rewritten

——(% bx ~ ay) — y cos (%,盼 一 ％ cos (w, y) .

Thus, the harmonic function <px + —

ay must satisfy the same con¬
ditions as the function 9, whence it follows that these two functions
may only differ by a constant, i.e.,
中1(%，夕) = 中(％，y) + ay — 汝十 const. (131.10)
Thus, by (131.2) and (1312),
= 十 =万 一 ^az, 1=
切 缈 十 ^ay
— 十 const. (131.1 1)
 594 VH. EXTENSION, TORSION, BENDING § 132

The terms by which (%,v, w) differs from (%,%,%) only express rigid
body motion and therefore do not affect the stresses ; this is likewise
readily verified directly from (1313) which give the same values as
(131.3).

§ 132. Complex torsion function. Stress functions. It is at times

more convenient to introduce instead of the torsion function 平(羯 y) its
conjugate harmonic function y) which is related to y) by the
Cauchy-Riemann equations
即 卸 即 邢
， ―， ，
—— (132.1)
dx 妙 如 dx

The boundary condition (131.6) is easily expressed in terms of the

function
“
For greater generality, it will be assumed that the bar under con¬
sideration may contain longitudinal (likewise cylindrical) cavities so
that the boundary L of the region can consist of several simple contours
工i，工2，.…，上 m+i the last of which surrounds all the others (cf. Fig. 14,
§35).
Let t denote the tangent to one of the contours Lk in the positive
direction (i.e., leaving S on the left). Then
dy dx
cos (%,%) = cos (力 y) — , cos (% y) = — cos (力 ％) = —
where s is the arc of 工论；it thus follows, by (132.1), that
那 4％ 邪
邮
—dn— = —
8x
，T
cos 例，％)〈 丁
卿
纱
cos (%' V)
—a% —I—
ds —町 dyds ds

出p
dn
= ds (132.2)

In addition,

y cos (%,y) — % cos (%,y) =% ———

dx
Hy
dy
丁
d
= 丁 i(%2 + y2). (132.3)
Hence (131.6) takes the form
今=小出十 阳
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 595

whence it follows that

小 = 廿炉 泻 十 Ck on
“
where Ck are constants which may have different values on the dif¬
ferent Lk.
(132.4)

A function 中， conjugate to a given single-valued harmonic function,

may, in general, be multi-valued (cf. Appendix 3.). However, in the
present case, this cannot happen, because by (132.4) the function 少
reverts to its original value for a circuit of any of the contours Lk.
It is known that the function 乎 is determined, apart from an ar¬
bitrary constant, and hence its derivatives will be fully determined ;
it follows from this that 少 is defined by (132.1), apart from an arbitrary
constant.
It is thus seen that the constants Clr . . . , Cm+1 in the boundary
conditions (132.4) may not be fixed arbitrarily. Only one of these constants
may be fixed in an arbitrary manner, i.e., one may, for example, put
Cm+1 = 0; all the remaining constants must then have completely
determined (initially unknown) values.
As one may dispose freely of one of the constants Ck, it is clear that one will be
justified in adding any arbitrary constant to the function 巾.

It will now be assumed that the constants Ck have been given some defi¬
nite values. In that case the problem of determining 少 coincides with the
problem of finding a harmonic function for given values on the boundary,
i.e., the "Dirichlet problem” which has already been discussed in § 62
(Note) and in § 77 and which is known to have always a unique solution.
Having found 中，the function 勺 may be determined from (132.1). How¬
ever, if the constants Ck are chosen in a haphazard manner, the function
may be found to be multi-valued. Thus, the constants must de¬
termined from the condition o/ single-valuedness o/ the junction y) ; as
stated earlier, one of these constants may be fixed arbitrarily.
In the case of multiply connected regions it is therefore, generally
speaking, more convenient to operate directly with the function 中 rather
that with
“
In the case of simply connected regions, bounded by one simple
contour L, single-valuedness of the function p will be automatically
ensured ; only one constant which may be fixed arbitrarily will enter
into the boundary condition. In this case it is often more convenient to
operate with the function
“
596 VII. EXTENSION, TORSION, BENDING § 132

It is often also very convenient to consider the function ?(§) of the

complex variable 3 = % + W，defined by
尸⑶ = 平 + 池 (132.5)
where 中 is the torsion function and 少 its conjugate function. The function
万(3) will be called the complex torsion function. It is obviously holo¬
morphic in the region S.
By (131.3)

、加 如 Ldx dx 力

whence, in the customary notation used earlier,

乙 一 iYz = m了 一 场｝. (132.6)
It is also convenient to use the so-called stress function, defined by
》(鸡 y) = 少(4,力 一 多(42 + 消； (132.7)
in terms of this function, the stress components are given by
aY ar 、
Xz=»~^, Yz= — [it . (132.8)

The function ¥ is not harmonic, since it obviously satisfies the equation

AT = - 2. (132.9)
It satisfies on the boundary the conditions
Y= on Lk 3= 1,2, …， w), (132.10)
where Ck are the same constants as in (132.4).
The curves, defined in the plane of the cross-section S by the equations
¥(%, y) = const., (132.1 1)
are the “lines of shear stress”，i.e., the lines whose tangents have at
every point the direction of the stress vector (X%, acting on the
corresponding element of S. This is a direct consequence of (132.8). The
boundaries of the region are always stress lines, as is, of course, obvious
a priori.
From the practical point of view, it is of great interest to find those
points of the cross-section, where the magnitude of the resultant shear
stress 、
T = VX^ + (132.12)
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 597

—
has a maximum, because it is at these points that failure of the material
will begin.
It is easily shown that these points lie on the boundary of the region.
In fact,

Using (132.9), it is readily seen that

/平于\2]
1F) (可)+2(矿)卜
中、 /a2T \2
2
十

and hence that 八(丁2) > 0 throughout the region. On the basis of a
well known theorem (cf. below), it then follows that the function T2
may only attain its maximum value on the boundary, as was to be
proved.
One may not have the equal sign in the above inequality, since, by (132.9), at
least one of the quantities ( , (02T/^2) must have a modulus not less than
unity. Incidentally, it may be shown by a simple reasoning that A(T2) 4“2T2.
If some function U, having continuous second order derivatives in a region S,
satisfies the inequality At7 > 0, this function can only attain its maximum value
on the boundary. In fact, let it be supposed that U has its highest value at some
internal point (力°, 夕 Describe, with this point as centre, a circle 丫 of sufficiently
small radius so that dU /dn < 0, where n is the outward normal to the circle. On the
other hand, by Green's formula,
CdU /
/J ——
dn
ds = / AU dxdy,
J
Y a
where a is the area inside 丫. Since AL7 > 0, one is led to a contradiction, and hence
the assertion is proved.

§ 133. On the solution of the torsion problem for certain par¬

ticular cases. It has been seen that the torsion problem may either
be reduced to the Neumann problem (for 9) or to the Dirichlet problem
(for 少；in the case of multiply-connected regions, the constants C * must still
be determined from the condition of single-valuedness of 9；cf. the
preceding section) . For this reason all the known, well developed methods
of solution for the Neumann and Dirichlet problems may be applied.
In addition, in view of the special simplicity of the boundary values
of 少 or d^fdn, one may fall back successfully on particular methods,
designed for the present problem.
598 VH. EXTENSION, TORSION, BENDING § 133

Saint-Venant himself solved and studied in detail (by constructing

tables and graphs) the torsion problem for a large number of cross¬
sections of different shapes which are of practical interest. He obtained
solutions for many cases (ellipse, equilateral triangle, etc.) by very
simple means. For the case of the rectangular section, he gave a solution
in the form of a rapidly converging series.
The reader should consult Saint-Venant [1] and also the small book
on the theory of elasticity by A. N. Dinnik [1] which is specially devoted
to the torsion problem, where the solutions for a large number of
cross-sections may be found. See also the book by I. Todhunter and
K. Pearson [1] and B. G. Galerkin [1,2]; in the first of his papers, Galerkin
gives the solution for a section, represented by an isosceles right-angled
triangle. Finally, one may also find remarks on approximate and ex¬
perimental solutions in the book by A. N. Dinnik [1].
Only the (almost obvious) solution for the case of a circular or annular
circular section will be quoted here. If the origin is placed at the centre,
—
one has clearly : y cos 仇，％) % cos (胃，y) = 0 on the boundary. Therefore
的 c

on the entire boundary. Hence = const., and one may take <p = 0.
The displacements and stresses are then given by
% =
—
tzy, o = tz%, 幽 = 0, (133.1)
Xz —— Yz = [ltx (133.2)
(the remaining stresses being zero). It is thus seen that in this case
transverse sections remain plane, unlike in the other cases.
By (131.9), the torsional rigidity is given by

D=
町J (% 2+ y^dx dy = 以, (133.3)
6
where I is the polar moment of inertia of the cross-section about the
centre. In the case of a circular section with radius R, one has thus
(133.4)
while in the case of the circular ring
7C
7 =万（尼一玛）， （133.5）
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 599

、
where R and & are the inner and the outer radii.
§ 134. Application of conformal mapping.
The results of this section were given in the Author's papers [12, 13]. A detailed
study of these results with several new applications is given in the book by I. S. So-
kolnikoff [1].
The torsion problem may be considered solved, if one has been able
to map the region S on to the circle (where, of course, S must be a simply
connected region) . In fact, let
=x iy ~ 3© (134.1)
be the function, mapping S on to the circle | j < 1 whose boundary, as
always, will be denoted by 丫.
If the complex torsion function expressed in terms of & is given
by
<p + = F(3) = /(0， (134.2)
the function /(Q will be holomorphic inside 丫. The real part 少 of the
function
4/(0 = 少 一 冲 (134.3)

will satisfy on 丫 the following boundary condition [cf. (132.4)] :

少
or, by (134.1),
= 3(炉 4 - 外) + const. — 入3宕
+ const.,

少= on 丫，

where b = e诒 denotes points of 丫； the arbitrary constant has been

justifiedly omitted.
However, in an earlier Part of this book, a formula has been deduced
which permits to find a function, holomorphic inside 丫，when the boundary
value of its real part is known. In fact, by (77.5),

r 仅) =r ；蓝加 const.,
2 m J 2(a
Y
Q
—
whence, finally,
i
2兀J r<±M^
cr— + J
constj 034.5)
600 VII. EXTENSION, TORSION, BENDING §134

and the problem is solved.

If 3© is a rational function, then s(b)3(b) = 3(a后(l/o) will likewise
be a rational function of a. The integral on the right-hand side of (134.5)
is, in this case, easily calculated by use of the residue theorem and it ob¬
viously leads to a rational function in & so that the solution will 加
expressed terms of elementary functions.
In general, if the expression co(b)G(l/b), considered as a function of o,
is an analytic function inside (or outside) 丫，continuous up to 丫 and has
inside (or outside) 丫 a finite number of poles, then the integral on the right¬
hand side of (134.5) may be directly evaluated by means of the residue
theorem.
The torsional rigidity D may be calculated from a simple formula
which will now be deduced (see the Author's papers [12,13]). One has
(§ 131)

D=N
£
(%2 + + jx
8
jj (得 金) —y dxdy = 以 + 以)o, (134.6)

where / is the polar moment of inertia of S about 0 and

(134.7)

Applying Green's formula, one obtains

L
+ = —j
L
中 .d(犷2), (134.8)

where L is the boundary of the region.

Noting that on the boundary 夕2 = = (o(b)3(b) and that
9=知 + /⑹],
one may rewrite (134.8)

00 = 一
J
Y
{/⑹ + 而w3(引而')}. (134.9)

If is a rational function, /(0 will likewise be a rational function

(see above) and hence also /(1/a), Z) (1/c) will be rational, so that the
preceding integral is easily evaluated in closed form, using the theorem
of residues.
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 601

In this case it is sometimes convenient to transform also the expression

for

〃
a ]
/ = (峭 + 4y = (%2?) + -y- (冲2) \dxdy
a% j
~

j %y(xdx ~ydy).
七
Noting that
r 5+3 3-3

it is readily seen that

(52- H) + 号石).
L
However,

(where the last relation is obtained by an integration by parts). Thus

/=
// =2/
L
523d3
Y
32(b) 3(<j)〃33). (134.10)

If 3(0 is a rational function, formula (134.10) permits calculation of /

in closed form by elementary means.
In the case of doubly connected regions, the torsion problem is likewise
easily solved, provided the function co© is known which maps this
region on to the circular ring. In fact, in this case the problem is reduced
to the determination of a function /(Q, holomorphic inside the ring and
satisfying the following boundary conditions :

沉 2偿）=*s（Q函 +的 on yi，
—
(134.11)
1
跟干他）=如（03（9十02 on y2,
I

where 丫】，丫？ are the circles, bounding the ring, and Cv C2 are two real
constants one of which may be fixed arbitrarily. One thus arrives at
the problem solved in § 62 (Note).
602 VIL EXTENSION, TORSION, BENDING § 134a

In the present case, the function ( 1 /i) /(Q plays the role of F(Q and in
the expansion (62.7) (where z must be replaced by Q one has to put
4=0, because would otherwise be multi-valued. For the functions
/i(A) and /2(9) of § 62 one has now
当3(P1曲正薄) + G， 题净)CO(P2净) + 02； (134.12)
Pi and p2 denote here the same quantities as Ri and R? in § 62. If one
writes C2 = 0, then the constant Cx will be determined by (62.9).
Having found the stresses may be calculated either in terms of
the old coordinates %, y or in terms of the curvilinear coordinates of
§ 49 which are related to the conformal transformation.
Let T denote the vector of the shear stress, acting at some point of
the cross-section S, and Xz, Yz its components in the Ox, Oy directions.
The projections of this vector on the axes (p), (A) of the curvilinear
coordinates will be given by (49.4) which has the conjugate complex form

— 2Ts 万 I ”) I (Xz 北)
p -

or, substituting for Xz — iYz from (132.6) and noting that

廿( 4
dF "/J 1
~ ，
d5~ d3' W(3)
one deduces finally the very simple and convenient formula

T. - iT, = P I 3 (0| 一 '函3'©}. (134.13)

On the boundary of the region Tp = 0, so that the preceding formula

determines directly the boundary value of the shear stress 丁分 and, in
particular, the maximum value of this quantity.

§ 134a. Examples. The method of the preceding section will now

be applied to certain particular cases.
1°. Epitrochoidal section. Let the section S be bounded
by the epitrochoids, considered in § 48,3° (Fig. 23). Then
3 = 3(9 = + 限) 例 an integer > 1, b > 0, a > n), (134.1^)
where, in the notation of § 48, 6 = Rm, m = 1/以
Formula (134.5) gives (on replacing g by 1/e)
 22

,
CHAP. TORSION AND BENDING OF HOMOGENEOUS BARS 603

, , / 1 口、 do
他） °
o2(cn + ao) -j — + const., —

whence, by (70.3),
=
仍2
2B”\
r
Y
"
/( 1 + 加 7+ b
d<s
—
—s + const.,

/K) = 淤戏"T (134.2a)

-
(the arbitrary constant having been omitted here), and the problem
is solved.
By (134.13) one obtains for the stress components Tp> T$
丁 丁 y
加 依 一 al一 酒的 成 (或1 一 G

or, putting 乙 pe拎 and separating real and imaginary parts,

M就2(% —
l)p"-2 (1 —
p2) sin 例— 1)眇
I 3化) I
~

T产… a^n~^\n — — (n +方丽

1 l)p2] cos 仇 — 1卅 — wp2^-1

where
] 3‘(C) [ = 旷3‘(03‘(0= 6]/%2p2e-2 2an^n^ cos 仇 — 1)9 + a2.
On the boundary (ie, for p = 1) one has Tp = 0 and
:

T= T8 =
n + 2a cos (% 12 + 谈
M炉 + 2an cos (n — 1)9 十 a2
—
H n < a, i.e., if the contour has no angular points, the maximum value
of T occurs at those points of the boundary where cos(% 1)S =
these points are closest to the centre. The maximum value of T there
1;
— —
is given by
々2一 2« 十 ％
F2一;也
If a f %, Tmax f oo, i.e., in the case of the boundary with angular
points, as shown in Fig. 23, the stress Tmax becomes infinite at those
points.
604 VII. EXTENSION, TORSION, BENDING § 134a

The torsional rigidity is easily obtained from (134.9) and (134.10)

which give
出7T方4
D= - (淤 + 4a2 + 功.
2°. B o o t h's lemniscate.
The solution of this problem (as well as of all the others, presented in this section)
were published by the Author in 1929 in the papers, quoted earlier, and reproduced
in the first edition of this book. Recently (1942), T. J. Higgins published the solution
of the same problem which he obtained by a more complicated method (cf. I. S. So-
kolnikoff [1], p. 184.).

The transformation function for the region, bounded by this curve,

was stated in § 46,6° (Fig. 27). Changing the notation slightly, one may
write
3(0 = 谈十 。
— (q > 1， / > 0) ；
one then obtains for the formula

的=
1
( 署<J2而
云/ ((/+〃) 十 加 2) (b — C)
'

The integrand, considered as a function of o, has outside 丫 two simple

poles : j = %“万 and 网 = — 勿“万； for large |b|, it is of order l/M.
Therefore, by the residue theorem,
/(C) —— + *2),
where Alf A2 are the residues, corresponding to the points%, c2 respect¬
ively. One has
r -
[3 — 尼2^2
4 "“仇) 十 a) (I 4- 刖2) 3 9 - E.w —
" — 2"1 — 谓) 每 '
similarly,
44 2
k^a
2^(1 一叫谭十？’
whence, finally,
(134.4©
期 一 i）（守 十 0
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 605

The stress components may be calculated as in the preceding example.

Only the value of T = on the boundary will be given here :

T
小双 1 + 谈)
(«2 — 1) V1 — 2a cos 外
+ 口2
The maximum value of T occurs for cos 外 = 1, i.e., at the ends of the
minor axis, and it is given by
+ 1)_.—
7n
/

—
'

°” (a 4-1)(^- 1)2
…一 -•I

One easily obtains for the torsional rigidity

=
_
即依(a4 + i)
j)“ _
2啰
3°. The loop o f Bernoulli's lemniscate. An example
will now be given of a simply connected region for which 3«) is not a
rational function.

^
Assuming | | < 1, let
3= 39 =(?"1+已 仅>0), (134.5")
and select that branch of the multi-valued function
is equal to unity for 已 = 0. In other words (Fig. 57a), "
1 十 已 which

IT
—— —
J

Fig.57a. C plane.

As C describes the unit circle y,

(— K <
606 VII. EXTENSION, TORSION, BENDING § 134a

and
卧
夕 = 2 cos —2 .
Hence

If R and 少 denote the modulus and argument of 3, then, by the

preceding formulae,
]/ 8 3
五 =口 2 cos , 少 = ，
彳 —
whence
R = 痣 5/2 cos 2^. (134.6a)
Thus 5 describes one loop of Bernoulli's lemniscate [Fig. 576] and (134.5)
maps the region inside this loop on to the circle | | < 1.
One finds for /(Q

(134.7a)

where one has to take that branch of the function (1 + 0/“短 which
就
is positive ony, i.e., one must take = e2
The integrand will be single-valued in the region, bounded by 丫 and
cut as shown in Fig. 58. Therefore (in the
notation shown in this figure, where, in
particular, 丫^ denotes an infinitely small
circle)

*[/+/+/+/]= a
丫 a B yi

where A is the residue at o = E which is

obviously equal to
1 +c Fig. 58.
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 607

By means of a simple transformation of the integrals, taken along a and

0 (the integral over being infinitely small) , one finds
1
/1+a 而 _ 1十 C 1
1

2痘 J
Y
Mg a — k
o
whence, omitting an arbitrary constant,
2,
•凝 1
/(Q = . = artan V^;
杆
德
in this formula one must take for
1 , 1 + iV^
artan = ——2?
log
i
_ iVK—
the branch which is defined by the series

The problem is thus solved.

4°. Confocal ellipses. Eccentric circles. When the
cross-section of the (complete) cylinder is bounded by two confocal
ellipses or two (eccentric) circles, the solution is likewise easily obtained
by transformation on to the circular ring. In particular, the solution
of the last case may be deduced directly from example 1° of § 140a.

§ 135. Extension by longitudinal forces. The solution of this

problem is quite elementary and has, in essence, already been deduced
in § 19. In fact, if
F
X— Yy=Yz = Z, = X“ = 0, (135.1)

where F is the magnitude of the given force, assumed positive in the case
of tension, and S is the area of the transverse section of the bar, then
all the required conditions will be satisfied. This solution corresponds
to normal stresses, distributed uniformly over the ends. The resultant
of the stresses, applied to the upper end, will be equivalent to a force F,
applied at its centre of area.
608 VII. EXTENSION, TORSION, BENDING §136

If the given force is not applied to the centre of area of the end,
it may be transferred to that point by adding a couple the plane of which
is perpendicular to the end (i.e., a bending couple) . Thus, the solution
of the problem of bending by a couple, stated in the next section, has
to be added in this case.
The displacements, corresponding to the stresses (135.1), are easily
verified to be given by
cF
SE
X, j = ，
"
SE
= y
F
y,
SE
(135.2)
W Z

where any rigid displacement of the bar as a whole may be added.

The quantity SE, which is the coefficient of proportionality between
the extending force F and the corresponding extension of the bar, may
be called the rigidity o/ extension {compression.

§ 136. Bending by couples, applied to the ends. In this case the

solution is also quite elementary.
In accordance with established custom, the bar will be placed ho¬
rizontally with the Oz axis running from the left to the right and the
Ox axis vertically downward, as shown in Fig. 59 (where the Oy axis
is not shown ; it is directed away from the reader, because the coordinate
system is to be right-handed).
The ends which were formerly called "lower'' and "upper” will now
be called "left" and "right'' respectively. In addition, it will be assumed
that the point 0 lies at the centre of area of the left end, so that Oz is
the "axis of centroids", i.e., the locus . of the centroids of the cross¬
sections.
An attempt will now be made to satisfy the conditions of the problem
by writing
Zz = 耽 A. = 丫缉 = Yz = Z# = = 0.
These values obviously satisfy the equations of equilibrium and of
compatibility (§ 130). It will now be investigated whether the stresses,
applied to any transverse section (from the right), are statically equi¬
valent to a bending couple.

Clearly, if the forces which are applied, say, to the right end are equivalent to
a couple, then the forces, applied from the right to any section, must be equivalent
to the same couple.
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 609

The resultant vector of these forces is equal to zero, since

jj dx dy = a 沟 = 0;
s s
the last integral vanishes, because the origin lies at the centroid of
the section S.

The resultant moment of the above-mentioned forces about an axis

through the centre of area of the section and parallel to the axis
Oy is
M ——
s
dxdy — js
— a［姨 〃先 为 =
— «1, (b)

where ? is the second moment of area of the section S about the axis Oy.
Finally, the resultant moment of the forces about the axis passing
through the centroid of the section and parallel to Ox is equal to

dxdy = a xydxdy. (c)

s s
If

yy
s
xy dx dy = 0,

i.e., if the coordinate axes Ox, Oy are principal axes of inertia of the
section S (with regard to its centroid), then the moment (c) is equal to
zero and the forces are equivalent to a couple with moment vector
parallel to Oy and determined by (6). For a given value of M the constant
a is determined by
M
a=
610 VII. EXTENSION, TORSION, BENDING § 136

Let it be assumed that the coordinate axes have been chosen in the
stated manner. In that case the solution has been obtained of the problem
of bending of a bar by couples, applied to the ends, whose moments
are parallel to one of the principal axes of inertia of the section with
regard to its centre of area.
The above results will now be summarized. Let a couple with vector
moment, parallel to one of the principal axes of inertia of the end with
regard to its centroid, act on the right end of the bar. If one takes as axes
0%, Oy the principal axes of inertia of any cross-section, e.g. of the left
end, and directs the axis Oy parallel to the moment of the couple, then
the solution of the problem of bending is given by

Xz= 匕= E = Z# = X© = 0. (136.1)

In these formulae I denotes the moment of inertia of the end-section about

the axis Oy and M is the magnitude of the moment of the couple (which
is positive, if the moment is directed along the axis Oy).
It is readily verified by direct substitution that the displacements,
corresponding to these stresses are:
M M M
(滤 + "2 — 砂2)， 2= 狈； (136.2)
JL £> 上 口工7
一三
terms, expressing rigid body displacements, may be added to these
formulae.
The plane % = 0 is a "neutral plane": fibres, lying in this plane,
will neither be stretched nor compressed. Fibres, lying to one side of
this plane, will be extended, while those on the other side will be com¬
pressed.
The normal stresses are distributed over the cross-section according 加
the linear law, expressed by (136.1) ; cf. also Fig. 59.
The points of the ' 'central fibre”， having before deformation the
coordinates (0, 0, z), will move to points with coordinates (&, 刀，C), where

4=0, E = z; (136.3)
2EI
this is easily seen from (136.2).
Thus the central fibre remains in the plane Oxz which is therefore
called the plane of bending. In the present case it is parallel to the plane
of the bending couple. The radius of curvature R of this line (after
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 611

deformation) is determined (apart from small, higher order terms) by

1 灌
— 而’
云=
(where it will be assumed that R v 0, if the curve is convex downward) ;
hence one obtains the important relation.
1 M
(136.4)

which expresses the so-called law of Bernoulli-Euler : the curvature

of the central fibre is proportional to the bending moment. The quantity
E? is called the flexural rigidity. Since a constant value had been obtained
for R, the central fibre in its deformed state will represent a circular
arc of radius R which must be assumed very large in view of the assumed
smallness of the deformations ; in fact, the quantity 1 jR must be of the
same order of smallness to which the deformations are restricted.
The relation 1/五 = 凝己必炉 used above is based on the following reasoning by a：
well known formula.
1 1
"
元 = (1+中部
~ 十 中产
where accents denote differentiation with respect to z. In view of the smallness of
the deformations, all but the first term of this expansion may be omitted and the
stated result is obtained.
In actual fact, the equation of the curve of the central fibre is given by

2R
which is a parabola ; however, the difference between this curve and the circle
with radius 五 is a second order quantity.

Points, lying before deformation on the normal to the section z —

c,
move as the result of the deformation to points (E, tj, Q, where, in particular,
by the last of the formulae (136.2)
M
X=c 切 =c =c
EI
Replacing on the right-hand side % by & which is justified in view of
the smallness of 1/Rt one finds
612 VII. EXTENSION, TORSION, BENDING § 137

this is the equation of the plane, perpendicular to the plane of bending.

Thus, normal sections remain plane.
If the moment of the bending couple is not directed along one of the
principal axes of the cross-section, this couple may always be decom¬
posed into two each of which satisfy this condition, and the solution
of the problem will be obtained by superimposing two solutions of
the stated form. In this general case the plane of bending does not co¬
incide with the plane of the couple ; however, in this case also it is per¬
pendicular to the neutral plane which will again exist. It will be left to
the reader to prove this simple property.

§ 137. Bending by transverse forces. Let the coordinate axes be the

same as in the preceding section, i.e., select as origin the centroid of
one of the ends ("left end”)，and let the Ox, Oy axes be parallel to the
principal axes of inertia of this end about its centroid. Let the forces,
acting on the right end, be equivalent to a force Wt applied at its
centre of area and directed parallel to the axis Ox (Figs. 60a, 606). The

resultant moment M of the forces, applied (from the right) to any section
which is a distance z from the left end, about the axis through the centroid
of this section and parallel to Oy is obviously given by
z), (137.1)
where I is the length of the bar.
If only a couple with moment M were acting on the section under
consideration, one might, on the basis of the results of § 136, write

— y x，
where ? is the moment of inertia of the section about the axis, parallel
to Oy and passing through its centroid.
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 613

An attempt will now be made to satisfy the conditions of the problem

by writing
M W{1 一 z)%
乙7z
—
. 一

I
% =
/
- (137.2)
-

However, it is clear that one may not now assume that all the remaining
stress components vanish, because in that case the forces, acting at
the cross-section, might not give the resultant vector W, acting in the
plane of the section. However, let it nevertheless be assumed that
Xx=Yy = Xy = 0. (137.3)
Substituting these values in the equations (129.1), one obtains

—dz —
dXz dYz
= 0, =0
and
% ayz w%
9%
( t
1 /
= c(J. (137.4)

It follows from the first two of these equations that Xz, Yz do not
depend on z. Equation (137.4) may be rewritten l
%
1T + = 0,
whence it follows that
v
阳
v
HQ Wxy
Yz = T- (137.5)
cy 9% /

where Q is some function of % and y. Substituting the expressions for

the stress components in the compatibility equations (130.4), one sees
that the first, second, third and sixth of these equations are satisfied
identically, while the two remaining ones give
HAQ 3AQ W
— = o,
加 —
即 (厂+M
whence
wy
ac
= ; L
(1 + b)/ —2 LIT (137.6)

where — 2叩 denotes some constant.

614 VII. EXTENSION, TORSION, BENDING § 137

Noting that
Af、 "+ 十（1一摄）基 一号 (%2 + 但｝
_
/\

I
*
" ~

2(1 b)/
Wy
一 - 2町
行
one may write
, 印
c
安
2(1 +加 十（1 -搭 （炉 4- 洲, (137.7)

where
If
“
is a harmonic function.
denotes the harmonic function, conjugate to 少】，i.e., the function
for which
沟1 那1 邠i
= ，
2% 勿 dy a%
then obviously (137.5) may be written
W
ex 5(i + 丞 十（1 — (137.8)

… +
郎(2
f=鲁+
b)
Sy 2(1 + 加冲
Finally, one can always write
W
5 = 即P - or 无 (137.9)
2(1 +
上 \了

where g is the torsion function, defined earlier (§ 131), and / is some

new harmonic function. Hence

Xz = "傍 y）— 2（i% 性 一 如“2 + a — 和）丹，

匕 户
儒 “）— +（2 +。
同 （⑼0）
W（l - z）
元一
The displacements corresponding to these stresses are easily calculated
(by the general formulae of § 15 or by simple elementary means, as
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 615

used eg in A. E. H. Love [1], Chap. XV) ; the reader will readily verify
that the following expressions satisfy (129.2) :
W
% = — TZy + ---
七/ — z) (%2 — y2) + — 就3},专岳2

w
v — + ——
EI
,
a(2 — z)%y, 、
(137.11)

w
W = — -―{力佐 一 专巧 + X +
土/
冲2}.

A rigid body displacement may again be added to these expressions.

The last of the formulae (137.11) shows that the function x must be
single-valued, since W and 中 are single-valued.
Substituting from (137.10) into the boundary condition on the side
surface
Xz cos (%, x) + cos (% y) = 0
and taking into consideration (131.6), to be satisfied by the torsion

_
function 中，one obtains

?=
dn
— 乌"2 十 (i 切)外] cos (笈, ％) — (2 + g) 冲 cos (%, y) (137.12)

on the boundary L of the region S.

Hence one has to solve the Neumann problem, just as in the case
of the torsion function, in order to find the function 左
It is easily seen that the condition, for existence of the solution of the Neumann
problem, i.e.,

L
{马32 + (1 — 如)力 cos (8, m十 十 b)即 COS (符，y^ds = 0,

is satisfied in the present case. In fact, applying Green's formula, this condition
becomes
2(1 + ct)
S
x dx dy — Q,

where the integral vanishes, since, by supposition, the centroid of S lies at the
origin.

The resultant vector of the external stresses, applied to the right

616 VII. EXTENSION, TORSION, BENDING §137

end, is easily seen to be parallel to 0% and equal in magnitude to W.

In fact, by (137.4), one has

s s
%（淑+妙十7 ） 卜为一
Js | j— —
~~ —— - j ———J J x
~~ ~ dx dy dy
s

=
L
^x[Xz cos 仇,％) + Yz cos (%, y)]ds W = W,

since by (讥) the last integral vanishes.

In a similar manner, one finds
/ ax. 3
JJ
6
Yz dx dy 匕
+夕"，+可+ 7)｝藤分= Wx\\ ,

2g
上
xy dx dy =
s

dx dy = 0,

because, by (©, the first integral on the right-hand side vanishes, while
the second is equal to zero as a result of the fact that Ox and Oy are
principal axes of inertia for the section S.
However, if g remains arbitrary, the forces applied to this end still give
a twisting couple. In fact, the terms containing g give a couple with
a moment determined by (131.8), while the terms with W give a couple
with moment

01 三 拉7 f] — 管 ° —如) — (2 + (137.13)
$

In order to remove the twisting couples, it is sufficient to give t a suitable

value so that the sum of the stated moments vanishes.
The terms involving W determine the bending of the bar. The plane
% = 0 is here the neutral plane and Oxz the plane of bending. The central
CHAP. 22 TORSION AND BENDING OF HOMOGENEOUS BARS 617

line (i.e., the line % = 0, y = 0, which passes through the centroids of

the cross-sections) becomes a curve in the Oxz plane, its radius of cur¬
vature (at a given point z) being determined by the Bernoulli-Euler
relation

where
z)
is the moment of the forces, applied to the transverse section (from the
right) at the given point, about an axis which lies in the plane of the
section in the direction of Oy.
In addition, the terms involving t cause torsion of the beam about the
axis Oz. Clearly, in the case of sections symmetrical with regard to the
Ox axis, one will have t = 0 and no torsion will occur.
Finally, if the force W is not parallel to one of the principal axes of
inertia of the cross-section and does not pass through its centre of area,
then its point of application may be moved to the centroid by introducing
a suitable couple and the force may be decomposed into two parts,
parallel to the principal axes of inertia. The unknown solution of the
problem is then obtained by solving the problem of torsion and two
problems of bending by forces, parallel to the principal axes.
Consider now again the above case. Instead of %,its conjugate function
%' will be introduced so that
艺=返 取 双'
0% 町' 砂 一 *

Then, using the relation [cf. (132.2)]

政
dn
= '

one obtains for %' the boundary condition

%' = F式s) + 觉 on J,%， …，工憎+i， (137.15)
where Lk denotes the contours forming the boundary L of the region,
Ck are constants and
618 VII. EXTENSION, TORSION, BENDING § 138

F式s) = — /[J"? + ( i — 务o)y2] cos (%,切〃s— (2 十 o)/冲 cos 色,y) 曲

= — 佳 + (1 — ^y^dy 十 (2 十 b)/xy dx
/
— — —— |(1
/
*cr)y3 + {(2 + ^xy
—— dy} + const.,

where the integral is taken along Lk from an arbitrary point of this

contour to the variable point (%, y). Noting that

^x2dy = x2y —2 dx + const.,

the preceding formula may be written

F式s) = — (1 — ct + 2(1 十 + const. (137.16)

Since the last integral, if taken along the entire contour, will not,
in general, vanish, the function %' will be multi-valued. However, in
the case of simply connected sections bounded by one contour L, the
integral, if taken around L, vanishes and %' will be single-valued, as
was to be expected.
By Green's formula

xy dx = 干 fj # dx dy,
Lk Sc

where Sk is the part of the plane surrounded by Lk. The upper sign must be
chosen for Lm+1> the lower for the remaining contours. The integral will only
vanish in the case, where the centroid of Sk lies on the Oy axis.

As in the case of torsion, it is sometimes convenient to consider the

complex function
G(» = 叉十仪'• (137.17)

§ 138. On the solution of problems of bending for different

cross -sections. In his major work on bending [2] as well as in other
papers Saint-Venant gave the solutions of bending problems for a
number of cross-sections, in particular for a rectangle. As in the case
CHAP. 23 TORSION AND BENDING OF HOMOGENEOUS BARS 619

of the torsion problem, Saint-Venant illustrated his solutions by de¬

tailed explanations, numerical examples and graphs. The reader should
consult his original work as well as the book by I. Todhunter and K.
Pearson [1] (cf. also the interesting paper by B. G. Galerkin [3]).
Also in the problems of bending, conformal mapping may be of great
help, just as it was for torsion. In particular, the results of § 134 (with
obvious modifications) are easily applied to the present problem and
the problem of bending may thus be solved for all the cases, considered
in § 134〃. But not much space will be devoted to it here and consideration
will be limited to the simple example, presented in the next section.
In a recently published paper, S. Ghosh [1] applied the method of
conformal transformation to bending problems ; evidently he was only
conversant with the Author's paper [12] and with the study of the
problem, contained in the book by I. S. Sokolnikoff [1].

§ 138a. Example. Bending of a circular cylinder o r

tube.
Consider a cross-section of the shape of a circular ring, bounded by
concentric circles 7a and 工公 with radii R】and A2 (R】 R?)，For the
ring (cf. § 62, Note)
+8
G(3) = X 十 取' = 4 log 3 + W (如 十 海)充 .⑷
—8

whence, putting 3 川心，

+8
x = h log 夕 十 E (瞅 cos 艇' —
—00
sin 为分)产. (138.2a)

In (137.12) take for n the normal, directed away from the centre ;
then obviously
cos (%,“) = cos S, cos (% y) = sin 8.
Further, noting that
傍 十 (1 — 如)夕2] cos 分 十 Q + cr)到 sin 9 =

and that
= (亲 + 刎)尸 cos 9 — Jr2 cos 39 (138.3a)

dn 勿'

one obtains the boundary conditions in the form

620 VIL EXTENSION, TORSION, BENDING § 138a

X +8

/ — oo
瞅 cos 6A — bk sin =
= — 信 十 那）评 cos a + 专理 cos 3a for r = 0, R2,
whence, comparing coefficients of cos々8and sin 砾
4=0, % = 0 （论 = ± 1, ± 2, …）， ak = Q # 0, ± 1, ± 3）,
的 一 匹2a-1= 一 （圣 + 加冰；， 一 芯％t — （4 加诉召，

3a3五； 3四一3五t
These equations give
—— —
3% 3口一3过4 — ——
= 一（普 + 同因 段）， 0T = —（叠十 和冰;心，
口3 == 专 >
—
q 3 === 0.
The constants 々o，bQ remain arbitrary, as was to be expected.
Finally, one finds for %（%, 夕）
2
1R22
X = 一 修十知
3 cos 39
+ const. （1 38.4©
夕

and the problem is solved.

For Ri = 0, one obtains the solution for the solid bar with circular
cross-section.
 Chapter 23

TORSION OF BARS CONSISTING OF

DIFFERENT MATERIALS *)

§ 139. General formulae. 1°. The problem of the torsion of bars

consisting of prismatic (cylindrical) components made of different
materials and joined along their side surfaces will be studied next.
Each component will be assumed homogeneous and isotropic.
The cross-section S of the bar will consist of several regions So,
S2，. . Sm, corresponding to different materials and bounded by curves
to be called dividing lines. In the sequel, when speaking of a part of the
bar, this will refer to such a region S>
Although the majority of the results to be stated below are true in the
most general case, the reasoning will be given only for the special basic
case, when the bar under consideration consists of a series of parallel solid
bars which do not touch each other and which are surrounded by elastic
material filling the space between them and the surrounding cylinder
which is parallel to the component bars.
The cross-section 5 of such a bar
will consist of a set of different
regions Si， S2, . . . , corre¬
sponding to the component bars,
and a region corresponding to
the surrounding material. Let
the boundaries of the regions
be denoted by 乙(， = 1 , 2, …， 加)；
the boundary of So will then
consist of the closed contours Lr,
. .,Lmt Lm+lt where the last
contains all the preceding ones
(Fig. 61).
Fig. 61.
2°. It will now be attempted to
satisfy the conditions of the torsion problem, by writing, as for the case
*) The results of the present chapter were first given in the Author's papers
[14, 15].
621
622 VII. EXTENSION, TORSION, BENDING §139

of a homogeneous bar,
u = ~ ^zy, v = rzx, w = ^(x} y),
• (139.1)
where the constant t and the function 中(%,夕) are subject to definition ;
the latter function will be called the torsion function.
On the basis of (129.2), one finds, as in the case of the homogeneous bar,
that in each region = 0, . . . , w)
= — y), Yz = + %)， (139.2)

where 固 denotes the shear modulus corresponding to the region S, ；the

remaining stress components are zero.
By substituting these expressions in (129.1), it is easily seen that these
equations, as in the case of the homogeneous bar, reduce to the Laplace
equation
工
.
— 八
0.
Thus, in the present case, the function 乎 must also be harmonic in each of the
regions Sj. The difference from the case of the homogeneous bar mani¬
fests itself only in the boundary conditions. These conditions express that:
痣) the external surface of the bar is free from external forces,
6) the forces acting on elements of the surfaces, separating the dif¬
ferent materials, are equal in magnitude and opposite in direction,
c) the displacements u, e, w remain continuous across the dividing
surfaces (because, by supposition, the various parts of the bar are
joined together) .
The condition a) obviously leads to
Z cos (”, ％) + Yz cos (% y) = 0 on Lm+1 (139.3)
on the boundary of S and 他) to
[Xz cos (n, 劝 + Yz cos 侬,朔 = [Xz cos (为,％) + Yz cos (%,y)]fc (139.4)
on the common boundary between S夕 and S^. Here n denotes the normal
to the corresponding contour, where it is implied in (139.4) that on both

，，
sides of the equality the normal points in one and the same direction.
The subscripts k indicate that the expression in the brackets is evalu¬
ated for the material in the regions and Sk respectively.
In the case which was termed above basic the conditions (139.3) and
(139.4) assume the following form (in the notation of 1 °)

and
X? cos (% %) + 匕 cos (% y) — 0 on Lm+1 (1393)
CHAP. 23 TORSION OF COMPOUND BARS 623

[Xz cos (%,幻 + Yz cos 例,y)]; = [X2 cos (%,劝 + Yz cos (% y)]0 (1394)
°n L” 二2， • •• , 'm+i，
where n is understood to be the outward normal of S°.
Substituting for Xz and Yz from (139.2), the conditions (139.3) and
(139.4) may be expressed by the single formula

兀
— (孚)
\ dn /
= (fXfc — 出) W cos (%, x)—x cos (如 朔 (139.5)

on the lines of subdivision, if one agrees to include among the lines of

subdivision also the contours corresponding to free surfaces, assuming
that 内 = 0 for a free surface.
In the basic case this condition assumes the form

Wo = (% — cos 例，劝一％ cos (%,朔 (1395)

on Lj, / = 1, . . . , 加
+1 with p,w+1 = 0.
The condition c) leads to the requirement that the function 勺 is to
remain continuous for the transition from one material to another. In
other words, the function 9 must be continuous in the entire region
S = S° + Si + + Smt ...
including the dividing lines.
It is readily shown that, if 9 satisfies the preceding condition, the
resultant vector (X, Y, Z) of the forces applied, to either of the faces,
say, the upper face vanishes.
In fact, first of all it is clear that Z = 0 (because Zz = 0 everywhere) .
Further, taking into consideration that in each of the regions
1, … ，加)
(/ 0, —
双 dYz 八

dx dy
one will have

X=
JJ Xz dx dy =
彻 dy J

=
； L
刀工)
彻
加
如
I ddx dy
}
J
修

7=0
624 VII. EXTENSION, TORSION, BENDING § 139

Transforming the last integrals using Green's formula, one obtains,

finally,
X=E
? J
Li
x[Xz cos (v, %) + Yz cos (v, 朔ds,
where L$ denotes the boundary of and v is the outward normal to S夕.
The integrations along the lines of subdivision must be performed twice,
since these lines belong to the boundaries of two regions. The expression
Xz cos (v, x) + Yz cos (v, y) assumes for these integrations by (139.4)
opposite signs without changing its absolute value; therefore the integrals
along the lines of subdivision cancel each other.
In (139.4), it was assumed that the positive normal n is taken to be one and
the same for sections contiguous to a line of subdivision from either side. On
the other hand, the normal v has for those sections exactly opposite directions.

Therefore the integrals along the lines constituting the boundary of S

are likewise zero by (139.3).
Thus X = 0. An analogous procedure shows that also Y = 0.
Consequently, the forces acting on the "ends'' produce pure couples.
The moment M of the couple acting on the "upper" end is obtained
by calculating the resultant moment of the above forces about the axis
Oz. Obviously
M = 2, (139.6)

D = 2
// 臼 (%2 + 2+
歹 —y dxdy. (1396)

The constant D is the torsional rigidity. As in § 131, it is quite easily

shown that D > 0. For a given moment M, the constant t is
obtained from (139.6). Thus, in the end, the problem is reduced to the
determination of the harmonic function @ which is continuous throughout
the normal derivative of which has on the boundary of S given values and
on the lines of subdivision of the sections given discontinuities ; these
values are expressed by (139.5) which must be interpreted in the sense
stated above.
In the so called basic case, the unknown harmonic function 9, continu¬
ous throughout S, must be determined from the conditions (1395).
3°. Restricting the consideration to the basic case, the next section
gives the solution of several, more general problems in which the con¬
ditions (139.5') are replaced by the conditions
CHAP. 23 TORSION OF COMPOUND BARS 625

No (萨) — &(/)= L on /= 1, 2, . . ., 侬 十 1, (139.7)

where the 力 are functions given on the contours 上力

For the problem of torsion
—
力 = (&) &) [y cos (n, x)—x cos 仇，朔. (139.8)
It is easily shown that the condition (139.7), apart from an arbitrary
constant, determines the unknown function tp uniquely. In fact,

"电
M+1

l 。(给
m+1
(F)]
/

J/
2 df = f
7=1 =i
夕= j \ / 0n \ a /」
% /
4 Lj

=网
〃热 0
〃s — 与 内/
/= 1 务
cp (?)z ds,
"笈 j
工 Lj
where L denotes the union of the contours Llt L2f . . . , Lm+V But by
a well known formula
/的
I cp --— ds
J an
L S

where tp is a function, harmonic in some region S bounded by a contour L

and d^/dn is the derivative in the direction of the outward normal.
Taking into consideration that for Slt . . Sm, in the present notation,
% represents the inward normal, one finds

2+(得)2]d%a第
m-f-1 / m I
2
i=i J
/ w力 = y2 %
=0 J,
(139.9)
Lj

whence it follows that, if on all contours Lj

力 = 0,
then
西 卿
17 =
一
可一
=0

in the entire region S, and hence <p = const.

If now and <p2 are two solutions of the problem, then
平 = @1 q>2

will likewise be a solution, corresponding to the case fj

tours. It follows from this that
— 0 on all con¬
626 VII. EXTENSION, TORSION, BENDING § 140

勺2 = const.,
which proves the above statement.
Next consider the question of the existence of a solution and of its de¬
termination. First of all, it is easily seen from (139.7) that

审//=
J
y=i 阿/*)
J \ an /0
Li i
而 T j=i J停"
J \ an /
Li
j

whence, using the fact that the integral over the normal derivative of a
function, harmonic in a region, taken over the boundary of the region is
zero, it follows that
th+i r
J"那
S 6 = 0. (139.10)
%
Consequently, the condition (139.10) is necessary for the existence of a
solution. It can also be shown to be sufficient, as will be seen later. This
condition is always satisfied in the torsion problem, because, if 力 has the
form (139.8),
fj ds = Q
J
Lj

for every Lj separately (cf. § 131). Thus condition (139. 10) is certainly
fulfilled in this case.

§ 140. Solution by means of integral equations.

It will be assumed that one is dealing with the basic case of § 131, 1°.
Taking into consideration the continuity of the unknown function cp in
the entire region S and the discontinuity of its normal derivative on the
contours 上力 it is natural to try to represent this function in the form of a
potential of a simple layer, distributed over these contours, since the po¬
tential of a simple layer has just the required properties. In this way one
is led to the generalization of the known problem of Robin Poincare
(as it has been called in J. Plemelj's book [2]).
—
Thus, let
1 r 1
8%,y) = / p($) log
J
L
—
T
ds — £
4
/ p(s)logb〃s,
J 夕
(140.1)
CHAP. 23 TORSION OF COMPOUND BARS 627

where / denotes the distance of the point (%,y) from the point s lying
on one of the contours Lj and p(s) [the density of the layer] is an unknown
continuous function of s. The symbol s will refer at the same time to
any of the contours Lj over which the integration is extended.
On the basis of the known properties of the potential function of a
simple layer, the function <p, defined in (140.1), will be continuous through¬
out the region. Its normal derivative will be discontinuous for a
passage through the 上力 In fact, the following relations will hold:

*Q) +
J P(s) G= i , 2, …， m
L
(140.2)
/ COS 小
兀P0) +J p(s) —广 ds,
L

where (d^ldn)Q and 飒口* relate to a point t on one of the contours

L” r denotes the distance between the points s and 力 and 山 is the
angle between the vector ts and the normal n at t (remembering that n is
always the normal outward with regard to the region So; see Fig. 61).
Using (140.2), the boundary condition (139.7) now becomes
C cos 山
7T(Mo + + Mo — 闯 J/ P 歹
= 力0)， (140.3)
L

where t denotes a point on Lj (j = 1, 1)・ In this way a system

of Fredholm equations has been obtained which may be reduced to one
single equation
p&) + J/ Kit, $)p ds = 附， (140.4)
L
where

K(*) = 金，力) = h(No for t on 乙 . (140.5)

+ 内) 夕
+ 闯
Next examine for what conditions (140.4) has a solution. The ho¬
mogeneous equation
p(£) 十
/k
L
(力 s)p($)ds = 0, (140.6)

obtained from (140.4) by putting /(/) = 0, ie,

628 VIL EXTENSION, TORSION, BENDING § MO

力(力) =0, (/ — 1, . . w 4- 1),

has only one linearly independent solution. In fact, the function 中,
defined by (140.1) where p(s) is the solution of the homogeneous equation
(140.6), will satisfy the boundary conditions (139.7) for = 0. As proved
in § 139, such a function <p is constant throughout the region S. But for
平 — const., (140.2) gives

27cp 修) = = 0 on Lj (，= 1,2, …，旬.

Thus, the solution p of the homogeneous equation (140.6) is the density
of a layer, distributed over the outside boundary of the region S
and giving a constant potential in this region ; for example, this is the
two-dimensional analogue of the "natural distribution" of electricity on
a conductor. As is known (cf. eg J. Plemelj [2], p. 63), the density of
such a distribution is determined uniquely apart from a constant mul¬
tiplier, and this proves the supposition.
By a known theorem of Fredholm, the adjoint homogeneous equation,
i.e.,

p(s) 十 KQ, s)p(£) 威 = 0, (140.7)

will likewise have a unique (linearly independent) solution.

p*(力) = + & (when £ is on ，，

In fact, it is easily verified that this solution is
7, = 1, …，加 + 1). (140.8)
Actually, if the point s is taken on (/ < w + 1),
(J K(l 加力 = 产二%
(r 闻 J
兀
/巴山 + 夕
弟 = 出二巴 ，
内) 十巴
工 七夕

[J K(t, s)〃力
—(回 + 阳) [
k J
C°—
夕
dt — 0, i j, E W 磔 1 ,
工S 工才

J
鼠(力,6成=旦 J k必 g
怒 =— 2

The above leads to the well known formula

CHAP. 23 TORSION OF COMPOUND BARS 629

c
/• cos
. 0, if s is outside Ls,
/
~
j 兀， if S is on

“ I 2k, if s is inside L§*

(For Lm+1 the sign has to be changed, since in that case, by the adopted
notation, n will be the outward normal with regard to that contour, and
not inward as it is with respect to all other 工力)
Using these formulae, one immediately establishes that p*, defined
by (140.8), satisfies (140.7), if the point $ is taken on <m + 1).
Finally, if $ is on Lm+1, then, similarly as above,

/ K(t, s)dt = 0 (/ v 加 + 1),

/ K{t, s)dt
—— 1,

whence it follows that (140.7) is also satisfied in this case.

Thus p*, defined, by (140.8), is one of the solutions of (140.7) ; the
remaining solutions may differ from it only by a constant factor.
According to a known theorem of Fredholm, the original integral
equation will have a solution if, and only if, the condition

J
/ p*/(s)ds = 0
L

is satified, i.e., by (140.8) and (140.5), 迁

〃
m+l r
S 内 =0. (140.9)

With the fulfillment of this condition the original integral equation

(140.4) has a solution, determined apart from an additive term of the form
Kp**, where K is an arbitrary constant and p** is the solution of the ho¬
mogeneous equation (140.6). The potential, corresponding to this term,
is a constant. Therefore the function 中 is uniquely determined apart
from an arbitrary constant.
In the case of the torsion problem, the condition for the existence of
a solution, i.e., (140.9), is always satisfied, as has already been shown
above. Consequently, the torsion problem always has a solution of the
stated form; the torsion function 9 is determined uniquely apart from
an arbitrary constant which has no influence on the stresses and de¬
formations.
630 VII. EXTENSION, TORSION, BENDING § 140a

The problem of the torsion of a bar, consisting of a number of hollow

cylinders, inserted one into another and joined together along the side
surfaces so that the curves, dividing the cross-section S of the bar into
regions corresponding to different materials, are themselves closed
contours, may be solved in quite an analogous manner (i.e., the problem
of "composite tubes"). The case when the various component bars have
longitudinal cylindrical cavities does not present any particular dif¬
ficulties.
In the preceding work, it has, of course, been assumed that the closed
contours L2, etc. satisfy a definite condition of regularity. For the
preceding work to be valid, it is sufficient to assume, for example, that
each of the considered contours has at every point a continuously changing
tangent and a bounded curvature.

§ 140a. Applications. In several particular cases it is, of course,

possible to obtain a solution of the problem without using integral equations.
Thus, use of conformal transformation may sometimes be preferable, as
will be shown in the first of the examples to be treated below.
1°. Torsion of a circular cylinder, reinforced by
a longitudinal round bar o f a different material.
Solution of this problem was obtained by I. N. Vekua and A. K. Rukhadze and
published in their paper [1]; part of this paper has been reproduced here almost
without any change. Solution of the case when there are cavities instead of the
reinforcing rods was given by H. M. Macdonald [1]. Another solution of the same
problem was obtained by E. Weinel [1]. See also a recent paper by R. C. F. Bar¬
tels [1],

Let the cross-section S of the bar consist of the region Slt bounded by
the circle £】，and the region S2, bounded by the same circle 工】 and a
circle enclosing the former. Let 内 and 出 be the moduli of rigidity
of S] and S2 respectively.
If is easily seen that, if Lr and are concentric circles and if the origin
of the coordinates is taken at their centre, the torsion function will be
constant, so that the ends of the inner rod and the surrounding cylinder
move as if they were not connected with one another and the torsional
rigidity of the composite bar were equal to the sum of the rigidities of
the component parts.
The case when 1a and L2 are not concentric is more complicated. The
notation of the first part of § 48 will be used with the exception that
CHAP. 23 TORSION OF COMPOUND BARS 631

z = % + 次 is replaced by 3. Let
r
力
+W =3= 1 •或
~ = 3(Q (140.1a)

be the relation mapping the 3 plane on to the E plane. The circles £1 and L2
will correspond to circles 丫【 and 丫2 in the 已 plane the radii of the latter
being and p2 (pi < p2)・ These radii and the constant a are related to
the radii 勺 and. r2 of the circles 工] and E? and the distance I between their
centres by the formulae (48.7) and (48.8). It should be remembered
(§ 48) that
0 Vpi V P2 V —
d
• (140.2a)

By (140.1«), the region Si will correspond to the circle 工|< pr and

S2 to the circular ring pi V | C | < p2. Let 9 be the torsion function ; its
values in S± and S2 will be denoted by 支 and (p2 respectively. Let 中
be the function, conjugate to 中 and defined separately in S】 and S2;
and S2 will be denoted by and 必 The functions 勺卜
its values in
平 2, Wi and 必 are harmonic in the respective regions.
“
The boundary conditions satisfied by 中】 and <p2 are (cf. § 139):

丽
= y cos (肛 盼 一 ％ cos 仇，y) on L2,
(140.3a)
<Pi = ?2 on Llt
"2 "1 = (“2 — M) [y cos (九 x)—x cos 3, y)] on J,
where n is now the normal directed away from the centre of the respective
circle. Also let s be the arc measured anti-clockwise.
Assuming that the first order partial derivatives of 中卜 队，例，R are
continuous up to the boundaries of the regions of their definition (a
fact which can be verified after the solution has been found) and taking
into consideration the relations
。支 科 电 叫1
(140.4«)
dn ds ds dn ds dn
following from the Cauchy-Riemann equations, one may replace the
conditions (140.3«) by
632 VII. EXTENSION, TORSION, BENDING §140a

我 = （%2 + 外）十 const, on

| L2,

——肉dn = 〃少dn on 、,
L
~
i 2
（140.3g） T

胆少 — Ml" = 去（“2 — ）（%2 + y2）+ const, on £>

2 内

Let F（§） = 勺 + 辿 be the complex torsion function and let

/（0 = 乎十训 （140.5〃）
be the same function, expressed in terms of Let f1 and /2 be the values
of this function in % and where % is the circle | | v p】and ct2 the ring
Pi v R I v p2・
Then one will have
oo
/1（9 = E
e =0 + 祝片 in % （140.6〃）
十8

力⑷ = E （或+ 泯厅 in % （140.7。）
左 — — oo

whence, putting 已 = pe叱

OO

认 = 讣 + 2（戒 sin 々8+ 4 cos （140.6屋）

念 =i
oo
我=弘 +fc=i
E [（p^~ p-^4）sin 研 + «叭 + p—S-2 COS^]. （140.7废）

Further, note that

But
十 y2）
存（%2 十打 —； 5gJ号
加 = 管存.
（1— ^）（1—^）- —
~~ ： 吉.
窘
（1 —岚）（1 —岚）
^― .

-71 ——
（1
1 — 吟2
z（1
倪
r\7]

4-（位
— 数q） 1 — 0cF + 1
= 1 十 ",

口2" + …） = 1 + 2聆 cos 9
或
—— 以 C 一+ 2gQ2 cos 29 +
一....
= 1 十（a（ +t 谈2yF2 +I . . .）
y

This series converges absolutely for p < l/«. Thus

p2 j °°
去（%2 + y2）= 厂章上十m忖 cos （140.8«）
Substitute（140.6屋），（140.7屋）and（140.8^）in（140.3a'）[where the middle
condition may be replaced by = 邢2/玷（for p = pj, since and
d^2/dn differ from these only by a factor] and compare coefficients of
CHAP. 23 TORSION OF COMPOUND BARS 633

cos 々9, sin 々A. One easily finds for k 1:

ak ak = a~k = °，
~

P涉广十 隐 = 7谈t浮，
(140.9a)
vp部； 十 vq凝-+产,
九 玩 一 P『％后,
where
9 2
眼 一 以1 0 M 砰$
V
出2 十
, C] —— 1
i

' 凝Pl
一65 〉
*

1 02P2
二 w (140.10^)

here q and c2 are the distances of the centres of and L2 from the
origin [cf. (48.6)].
The constants 6『，6；remain quite arbitrary, as was to be expected.
The condition of continuity of <p gives, as is easily seen, = 口;'，where
the value of « j and
*
is arbitrary. Therefore put
= bG = a。= 0.
Then, by (140.9口)，
依
丫

£= 凝t
+ 1 — vaAr K,
淤t,
i
= —氏 丁 (140.1 U)
1
—— v"九

源 =c0—】+ 加 /凝口
1
where
a = Pi/pt I = c2 — q. (140以12)
Substituting these values in (140.6a) and (140.7矶 one finally obtains
1 ~~~、心
M0 =~~y 十以 它 谈飞,
1 aq 左 =i 1 —— va 加
(140.13。)

1 -

z
口 e=1 1 VOt k— 1

These series and their derivatives are easily seen to converge absolutely
and uniformly in the relevant regions including the boundaries.
If v = 0, ie, 闷 = (z2) one finds for /工 and /2 one and the same ex¬
pression
634 VII. EXTENSION, TORSION, BENDING § 140a

i.e.,
f(3) = 近23 = 北式% 十 iy).
This is the complex torsion function for the homogeneous cylinder. If the
origin of coordinates is taken at the centre of the cylinder, one obtains
F(3) — const.
(cf. § 131, Note 2).
Thus it may be said that the function 为公)，determined by the second
of the formulae (140.13^), consists of two parts : one, corresponding to the
case when one deals with a continuous, homogeneous bar (first term), and
the other, expressing the "indignation” aroused by the presence of
the component bar.
Once the functions /式乙) and 为(9 have been found, the stress components
can be calculated by the formulae of § 134 (cf. I. N. Vekua and A. K.
Rukhadze [1]). The torsional rigidity D is likewise easily calculated
from a formula of § 134. One finds

o = M+(内一⑷r — 心 + 从2
co 心p k
— 9」
4)
可
where
I= + (140.15©

here I is the polar moment of inertia of the solid bar of radius r2 referred
to its centre, and 7' is the polar moment of inertia of the bar with radius
々 referred to the centre of the first bar.
If D' and denote the torsional rigidities of the component bars
with the moduli of rigidity and “叩 taken separately, then
= 臼① 一 制叼)，
oo a% (1 40. 16〃)
CHAP. 23 TORSION OF COMPOUND BARS 635

[where the latter follows from (140.14a) by putting

By (140.1甸 and (140.16a)
0]. —
D — (D‘ + D〃) = 4kMiM2
内, 十 N2
+ 眼戒 忧 左 =1 (1 一/扇/)2 —
岸蜡 2 2 S afc(l W+i) (140.17a)

whence it follows that

Dr + Df, < D, (140.18©
as may have been expected beforehand.
For the homogeneous cylinder (臼 — 眼)，one has instead of D
00 ~ 出 2?•
In the general case when々is small, one has approximately, neglecting
fourth and higher order terms in p】，

向 +眼
whence follows the approximate formula
D 2(% ―回 r (140.19a)
% +% /

This formula reduces for 臼 = 0 to that obtained by H. M. Macdonald [1]

for the hollow cylinder.
If the cylinder is reinforced not only by one but by several longitudinal
bars of the same material and if these rods are so thin and removed
from one another that the regions which are "affected” practically do
not overlap, then obviously the approximate formula (140.19a) may
also be used in this case, provided one interprets as the sum of the
moments of inertia of the component rods with regard to the centre of
the circle L2.
The torsion problem has also been solved in the case when 工工 and L2
are confocal ellipses (cf. I. N. Vekua and Rukhadze and when the
boundaries are epitrochoids disposed in a definite manner.
2°. Torsion o f a rectangular bar, consisting o f
two different rectangular parts. The problem of the
torsion of a rectangular homogeneous bar was solved by Saint-Venant
(cf. e.g. A. E. H. Love [1] § 221). It is often possible to obtain a solution
of the torsion problem in cases which are excluded from consideration by
the reduction of the problem to integral equations. For example, such
cases occur when the boundaries have corners, as in the problem to be
636 VII. EXTENSION, TORSION, BENDING § 140a

y treated here.
Consider a bar, consisting of
two parts the cross-sections of
which are rectangles with sides 诙,
b 26 and & and which meet along
the boundary of width 2b （Fig.
O
s. X 62）. Let the moduli of rigidity of
b $2 the component bars be % and %
respectively. Take the axis Oy
J a? along the dividing line of the
Fig. 62. regions ST and S?, corresponding
to the different materials, and let
the origin be at the midpoint of this line ; denote by 平1and（p2 the values of
the torsion function o in the regions S】 and S2.
Further, introduce the harmonic function = 中 + and denote its
values in and S2 by and 〃. It is immediately seen that the bounda¬
ry conditions are as follows :
3©"
—
刎
ox
= 2夕（％ = -~ %,
• 《力 y
《 办）， —
ox —
2y（% ：== 6《夕 —
《 6）, （«）
*

a a
内
云一眼 k=2他一 （％ =0，一'< '《幼

（％ — y 6）,
《

泄'
（y 土九 ~~质 0）? = 0（y b, 0 《 口2），（'）

The harmonic functions and will be determined in the form of the

series
oo
6 = Z（4 晨+i sinh mx 十 cosh mx）sin my,
n™ 0

= £（幺短 +i sinh mx + cosh 加％）sin my,

区N 0
where
m = （2%
1）7C
（/）
26
Each term of the two preceding series is obviously a harmonic function.
The coefficient m has been chosen so that the conditions（d）are satisfied ;
CHAP. 23 TORSION OF COMPOUND BARS 637

clearly also the condition (c) is fulfilled.

There remains to satisfy (a) and 仅). For this purpose it will be re¬
—
membered that in the interval ( b, + b) the function 2y may be re¬
presented in the form of a series :
8

2y = £ mA2n+1 sin my, (g)

0
91=

where
/2 '2 (— 1 产
八 L%)访早可，
i.e.,
例
2*1
3+1)3'
The series (g) is a Fourier series for the function /, defined in the interval

—
( 26, + 2b) in the following manner :

— + b),
/
/=
2y
助 一 2y „
in the interval
„ ,,
(—
" 2b),
f —— 4b + 2y ,, „ ,, (— b, — 2b).
On the basis of (g), the conditions will be satisfied, if

—
^2n+i cosh 初旬 B2n+1 sinh m% = X2n+P
^2n+i cosh ma2 + B2n+1 sinh ma2 = 力 2的+1，

and the condition (&), if

mH 2tt +1 一 以2月2”+1 = (3. — 从2)月2 期+1.

Solving the three preceding equations for ^2n+i> and B2n+1 and
substituting the values thus found in (e), one obtains after some obvious
transformations

= 4b 2 g (一 1产
四=0 (2w + I)3

-
[pt,2 4 (Mj-W2)coshwt&2j coshwt(% + %)+ 网 sinhma2sinhm^-[z1 cosh ma2 cosh mx
sin my,
Mi cosh sinh 楸％ + 出2 cosh mar sinh ma%

(m
\ / 「二'
E= (2% + 1尸 .
3
= 序
兀 n 0
.
— ——— sm my.
十( cosh ma^ cosh m 4- [Xj^sinh majSinh mx + pt,2 cosh cosh mx
.■ ―
Wi cosh. 例口2sinh 僻々i十 p,2 cosh sinh ma2 max ——
638 VII, EXTENSION, TORSION, BENDING § 140a

The form of the coefficients shows that the series obtained converge
rapidly (uniformly and absolutely) . Also, the use of differentiationduring
the process of deduction is justified.
The torsion functions are given by
= — xy in 工， 甲2 = '' 一 xy in ^2-
The torsional rigidity is obtained from (1396) which in the present
case has the form

D=
(炉 +外十 X^ - y dx +
2 . 肉2
dx

Substituting in this formula the expressions for 少i and %，one finds,

8
1 M
using the result that S
n— 0 (2% + 1)4 96 —
- (compare
、 x with the case of the
homogeneous bar, A. E. H. Love [1], § 225),
8
D = W ( M1g1+ 以202)》3+

+(2)5护 s 忧 cosh1加劭(“i+cosh
忌 cosh — (fXj+p-2) cosh 侬旬 coshm a2
碑

\ h/ m =0 (2% 十 产 sinh 加& cosh sinh

加诙 + p.2 加 ma^
卷 cosh mar + cosh ma2 cosh 加(诙一口?)一1
m=o (2%+ 1)5 cosh ma2 sinh
—
p,2 cosh ma^ sinh ma2)

If 旬 and a2 are large compared with b (in fact, if 曲，a2 > 56), one may
with sufficient accuracy put
sinh 加a sinh. ma2 1 1
】 ]
' cosh
〔
'
cosh 加旬 ma2 sinh ma】 cosh 加％

1
J. J J ,
_ -- 1
. _ , ''
1 ,
:

sinh 价々
2cosh 加恤

and obtain for D the approximate formula (compare with the analogous
expression for the case of the homogeneous bar, A. E. H. Love [1], § 225)

3 内 +眼 \ 兀/ ”=o (2% + 1)
CHAP. 23 TORSION OF COMPOUND BARS 639

8 谭十
= —（内向 + 2）" — 3.361 护 力_二约 .
网

3°. In conclusion, a recent paper by D. I. Sherman [27] will be noted

in which has been solved the problem of torsion of an elliptical cylinder
reinforced by a circular rod. The method of solution of this paper may
be applied with success to the approximate solution of problems of the
above type in a number of cases which are of practical interest.
 Chapter 24

EXTENSION AND BENDING OF BARS, CONSISTING OF

DIFFERENT MATERIALS WITH UNIFORM POISSONS RATIO*
The various cases of elastic equilibrium of bars, which were stated in
§ 129, will now be considered. It will be assumed, that the different
materials, constituting the bar, have the same Poisson's ratio but,
generally speaking, different Young's moduli (cf. § 139, 1°).
Remembering that o has almost the same value for many materials, it
becomes clear that this restriction is not very severe. On the other hand,
it considerably simplifies the solution.

According to Poisson's original theory, the quantity e was the same for all
materials and. its value was equal to However, this circumstance is not confirmed
by experiment. The variations in the values of a for different materials are con¬
siderably less than those for E. For example,
for copper : 1/b = 2.87, 石 = 1.25 X 106 [kg/cm2],
while for aluminium : 1/b = 2.92, E = 740 000 [kg/cm2]
(cf. Note 2 at the end of § 146).

In fact, as a consequence, one may consider separately the case when

Poisson's ratio is the same for all materials constituting the beam. The
general case will be considered in the next chapter.
In particular, for problems of extension (compression) and of bending
by couples, the present case is almost as simple as that for homogeneous
bars; this will be shown in §§142, 143.

§ 141. Notation.
The concepts of this section, as well as the corresponding formulae also apply
to the case, where the values of Poisson's ratios of the various materials are dif¬
ferent.

Consider the quantity

*) The contents of this chapter were taken from the Author's paper [15].

640
CHAP. 24 EXTENSION AND BENDING FOR UNIFORM b 641

=
J / Edx dy = E SjEj, (141.1)

where E denotes the modulus of elasticity at a given point of the cross¬

section which takes constant values Ej in different parts S? ，corresponding
to the materials constituting the elastic body; the areas of these parts
of the cross-section will likewise be denoted by
Further, the "reduced centre of gravity" of the cross-section will be
understood to be the centre of gravity which is obtained by ascribing the
various parts of the cross-section surface densities which are equal to
the corresponding moduli of elasticity ; thus, if the origin of the coordinate
system is placed at the reduced centre of gravity,

s
dx dy =
V
//
s
J
Ey dx dy = 0. (141.2)

The "reduced moment of inertia” will now be defined as the moment

of inertia, calculated under the same supposition with regard to the
densities of the different parts of the cross-section. In particular, the
reduced moment of inertia /右 about the axis Oy in the plane of the cross¬
section will then be given by

IE = jJ 石％2 dx dy ~ Ejlj, (141.3)

where lj is the customary moment of inertia of the area S?・ about the
same axis.
Finally, the principal axis of inertia of the cross-section, under the
same assumption with regard to the densities, will be called the "reduced
principal axes".
If the axes 0%,Oy coincide with the reduced principal axes of inertia,

〃
one will have
Exy dx dy = E E $ xy dx dy = 0. (141.4)
s g

Here, as well as in § 142, 143 and at the beginning of § 144, it is un-

neccessary to assume that one is dealing with "basic" cases (§ 139,1°);
it is sufficient to suppose that the bar consists of a number of homo-
642 VII. EXTENSION, TORSION, BENDING §143

geneous, isotropic, cylindrical bodies (fibres or strips), welded along their

side surfaces.

§ 142. Extension. It is easily seen that in the above notation the

problem of extension of bars by longitudinal forces, applied to the reduced
centre of gravity of the cross-section, is solved by the following formulae
[cf. (135.1), (135.2)]:
E-F
Zz 」— in the region S”
Se (142.1)

—z
gF eF F
u= %, o= y, m ~

% 3石

(the remaining stress components being zero) ; F denotes here the total
tensile force (F < 0 will correspond to compression).
The rigidity of the bar for extension (compression) is equal to SE
(see § 135).

§ 143. Bending by a couple. The problem of bending by a couple

whose moment lies in the plane of the ends is likewise very little different
from the same problem for the homogeneous bar (§ 136).
Let the origin lie at the reduced centre of gravity of the "left" end
and let the axes Ox, Oy coincide with the reduced principal axes of
inertia.
If the moment of the couple, acting on the "right” end, is parallel
to the axis Oy and if its magnitude is M, the solution is given by
ME$ % m .
“Zz = — Sj (143.1)

(the remaining stress components being equal to zero) and

M MM
= 啰 6炉
—
oy2), p
= —
— axy, 皿 = 广 (143.2)

Substitution of these expressions in the static equations of the elastic

body show that all the equations are satisfied ; the boundary conditions
are obviously fulfilled.
The resultant vector of the external stresses applied, say, to the right
end is equal to zero, since, by (141.2),
CHAP. 24 EXTENSION AND BENDING FOR UNIFORM b 643

[[Zz
$
dy = 0.

The moment of these stresses about the axis Oy is, by (141.3),

s
力Zz dxdy=~^
s
Ex^dx dy = M;

finally, the moment about the axis Ox is, by (141.4), equal to

J J yZzdx dy = —— j J Exydx dy = 0.
The above solution thus satisfies all the imposed conditions. It is easily
seen that in the case under consideration the Bernoulli-Euler law is
valid ; it is now expressed by
1 _ M
(143.3)
五
The flexural rigidity is equal to IE.
§ 144. Bending by a transverse force. The solution of the problem
of bending by a transverse force will now be considered. Let the origin
0 be at the reduced centre of gravity of the "left" end and let the axes
Ox, Oy coincide with the reduced principal axes of inertia.
This problem may always be reduced to the case, where the transverse
force, applied to the "right" end, acts through its reduced centre of
gravity and parallel to the axis Ox (cf. § 137).
Guided by the form of (137.10), (137.1 1) which refer to the homo¬
geneous bar, it will be assumed that the conditions of the problem may
be satisfied by expressions of the following form:
u= — cyz + A[lcr(Z — 2)(炉 - y2) + 妥厉 2
一 看明，
& = XXZ 月 6.— (144.1)
M =四 一 A\xilz — 我2) 十 又

where 9 is the torsion function of Chapter 23 and 工 = y) is some

function which has still to be defined ; I is the length of the bar and
t, A are constants.
Calculating the stress components, corresponding to these displace-
644 VH. EXTENSION, TORSION, BENDING §144

merits, one finds —

Xx Y y = Xy — 0 (as in the case of the homogeneous
bar) and, in the regions S#' = 0, 1, . . . , 加)，
~ ~ Bj + 知% 2
卜
-J- (1 — 知)
(144.2)
oy +
=“尹 \k 力 一岛 +（2十6）
到,
,

费 一 K# - 切，
where 石外 K$ are constants which may have different values in the
different regions Sj；in fact,
g
b = 4g = ，
，& = (144.3)
2(1 工、
十
1

Substituting the expressions (144.2) in the equilibrium equations, i.e.,

in (129.1), it is readily verified that the function 力 as well as the function
<p, must satisfy the Laplace equation in each of the subregions con¬
versely, the above equations will be satisfied under these conditions.
Now consider the boundary conditions. To ensure that the displacements
% v, 汕 will be continuous throughout the body, the function / must
obviously be continuous throughout the entire cross-section S (since the
torsion function cp is, by definition, continuous throughout S).
The boundary conditions with regard to the stresses lead, as in the
case of torsion, to the requirement that the expression
Xz cos (n, %) + Yz cos (% y) (口)
must vanish on the free side surface and that it must be continuous
for a passage through the surfaces separating the different materials.
The resultant vector and moment of the stresses, acting on the right
end, will now be calculated. First of all, it is clear that the component of
the resultant vector in the Oz direction is equal to zero. Its component
in the Ox direction is given by
X Xzdx dy.
s
Remembering that (144.2) satisfies the equilibrium equations and, in
particular, the equation
^XZ d dYzL %L 0,
+ 初 =
__
讯 如
one obtains, after substituting for from (144.2),
CHAP. 24 EXTENSION AND BENDING FOR UNIFORM b 645

ay
+ + AEj % = 0.
On the basis of this identity, one may write

宁〃偿⑻ +看（占）｝ dx dy + AI®

However, as has been shown in § 139, under the above conditions for
the expression (a)

辛卜 ，沟 .。
6,

consequently

A = ZE 玛
JJ
Sj
%2 dx dy = AIe, (144.4)

Further, by assumption, X = W, where W is a given force. This con¬

dition determines the constant A :

For the components of the resultant vector along the Oy axis one obtains
in an analogous manner

丫 = JJ
s
Y zdx dy =
JJ xy dx dy.

Hence follows, by (141.4), that Y = 0.

Finally, since Z% =0 for z = I, no bending couple will act on the
right end.
The moment of the twisting couple is given by
W a 双 双
AjT —— tD —2(1 + 6)/e7・=0
——— £ 石 / / { y — —
JJ I 9%
炉 %
dy
H
646 VII. EXTENSION, TORSION, BENDING § 144

+ (1 — 知)夕3 — (2 十 dx dy, (144.6)

where D is the torsional rigidity. The constant t must be determined

from the condition M —
0 which may always be done, once the functions
9 and % have been calculated.
The function cp can be determined, using the results of Chapter 23.
The function % has still to be found. Assuming, for definiteness, that
one is dealing with the "basic" case (§ 139,1° ; Fig. 61), it is readily verified
on the basis of (144.2) and (144.3) that the boundary conditions reduce,
in the notation of Chapter 23, to the following (remembering that, by
supposition, = 0) :

务) — 〃on Lj^ 1)， (144.7)

；
where

%= —
—| + (i
(No
_ 闯 知%2 y2 cos in, 4) 十

+ (2 + ct) 冲 cos (肛 y) ｝. (144.8)

Thus, one has arrived at exactly the same problem as in the case of
torsion, except that the functions 力，given on the contours, do not have
the same values.
It will now be investigated whether the condition (140.9) for the
existence of a solution is satisfied. One has
m+l 广
£ / Ms =
7 = 1 </

L
/
= 一 Mo ｛［知炉 + (1 — 扣)y2］cos (力, + (2 + b)%y cos(% y)｝曲 十
m r
+i J/ ｛［知 + (1 — 知)外］cos
£ & %) + (2 + o)%y cos (n, y)｝〃s
or, transforming the integrals by use of Green's formula,
c cr
//// 2(1
m
m+i
S
J
/ fjds = — JJ + a)^xdxdy — j=ij
S / / 2(1 + g)
J
内 ％ 4% dy =
4 So
CHAP. 24 EXTENSION AND BENDING FOR UNIFORM a 647

= — E% dy ;

however, the last integral vanishes, since, by supposition, the origin lies
at the reduced centre of gravity.
Thus, the existence condition is fulfilled and the present problem will
always have a definite solution which may be obtained by the help of
the same integral equation as in the preceding chapter, except that the
functions 力 are now determined by (144.8).
In particular, the remark in Chapter 23 with regard 加 the applicability
of the solutions for other shapes of cross-sections, e.g., for the case of a
compound tube, still remains true.
Finally, note that it follows from the formula for u, i.e., from the
first formula of (144.1), that the curvature of the central line (which is
the locus of the reduced centres of gravity) satisfies the relation

in other words, the Bernoulli-Eider law again holds true.

§ 144a. Example. Bending o f a compound circular

tube by a transverse force, applied, t o o n e of its ends.
Let the cross-section of the bar con¬
sist of two concentric circular rings
S], S2 the first of which surrounds the
second one, as shown in Fig. 63. The
inner, middle and outer radii will be
denoted by R2, Rlt Rq respectively,
and the moduli of elasticity, corre¬
sponding to Si and S2, by E» E2.
Let the transverse force act through
the centre of the circles in the direc¬
tion of the 0% axis. In view of the
complete symmetry, it is clear that
t = 0, i.e., no torsion takes place.

The function ^(x, y) will now be found

and its values in the regions S1} S2 will be denoted by and /2, respectively.
Let 九分 denote the polar coordinates in the plane. By (138.3g)
648 VII. EXTENSION, TORSION, BENDING § 144a

［加%2 — 如）夕4cos 9 十（2 十 b）%y sin 8

+ （1 —
—
= 1 cos 3那 十（基十 知）评 cos 8.
Correspondingly, the boundary conditions have the form

—— = kR^ cos 眇
+ 永R2 cos 3眇 for r =

一
Xl -- X2
玛“1 一 // =（g ''
（- 万渣 cos 分 + 普号 cos 3如｝
3 夕？ 夕

dr
= — k璀 cos A + 普段 cos 3 分for r &，
where
皆十知 = 不

and in the second condition the shear moduli 如 眼 have been replaced
by the moduli of elasticity Elf E2 which are proportional to them.
Expanding the harmonic functions 石，抬 出 series and substituting in
the preceding formulae, these functions may be determined. However,
it is easily shown that these conditions may be satisfied by writing （cf.
solution for the hollow homogeneous circular cylinder in § 138a）

Xi = 0/ + 包） cos 9 + 卷六 cos 39 （五i 48 五0）,

心
（% + 年）
= n cos 分 小浮 cos 3分 （及2 y &）•
Substituting these expressions in （i＞）, one immediately sees that all the
conditions will be satisfied, provided
— a；= —
ct^ 出 w = 万段， 就 出 一 一

2区； a2 五；口 口1 口1，

-
石式的凡 4）- E2（a^ — 2 = — 凤马 — 玛）段.
Hence one finds
Ei 牌一 号）（瑞 陶 五野 十码一 磁
/（瑞一 脸［（玛
玛因 出）璃 - 段） /阳 M）因 一 璃）
CHAP. 24 EXTENSION AND BENDING FOR UNIFORM b 649

弓阳 一 玛）[（尼 十 段严 一瑞 鼎] 十 玛因 一 段）阳 十 瑞）
能- 一 ，
当阳 礴的 — 号） 玛阳 十 周阳 一 磅
while
旬症 十 kR%, &周 十 万招.
Thus the problem is solved. The solution for the case where the circles
are not concentric was given by A. K. Rukhadze [1]. The solution for the
case of confocal ellipses was given by I. N. Vekua and A. K. Rukhadze [2].
In the paper [2] by A. K. Rukhadze the solution is given for the case of
epitrochoids.
The problem of bending of the rectangular bar, considered in § 140a,
is likewise easily solved.
 Chapter 25

EXTENSION AND BENDING FOR DIFFERENT POISSONS

RATIOS*

In the general case, when the Poisson's ratios of the various materials
may also differ, the problems of extension and bending become con¬
siderably more complicated. In fact, it will be found that it is now
impossible to assume X® = = Xy = 0, as was done in the case of
Saint-Venant^ problem as well as in the case where the Poisson's ratios
were uniform.
As a consequence, one has to give attention to a certain auxiliary
problem of plane deformation which will now be introduced.

§ 145. An auxiliary problem of plane deformation. The auxiliary

problem, mentioned in the introduction to this chapter, consists of the
following. It is required to find the elastic equilibrium of a beam, con¬
sisting of different materials in the same manner as described at the
beginning of § 139 and under the supposition that it is subject to plane
deformation parallel to the plane Oxy (i.e., that 如 = 0 and u, v depend
only on %, y and not on z), for the following conditions :
1. The external stresses, applied to the side of the bar, are equal to
zero, i.e., Xn = 0, Yn = 0, (145.1)
where, as always,
Xn = Xx cos (%,%) + Xy cos (% y), Yn = Yx cos (%,%) + Yy cos (笈，y)
and n denotes the normal to the side surface.
2. On the dividing surfaces of the different materials
(XJ = (Xja (Y j = (yn)&, (145.2)
where n is the normal to the (cylindrical) dividing surface in a definite
direction and the subscripts /, k indicate the values for the materials
* The problem of extension and of bending by a couple was solved by the
Author in his paper [15]. The present chapter presents a new, more detailed study
of the solution (cf. foreword to the third edition).
650
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 651

occupying the regions j, k, adjoining the dividing surface. The con¬

ditions (145.2) express that the stresses applied to elements of the dividing
surface from either side must balance each other.
3. The displacements undergo the following discontinuities on the
dividing surface :
物 一 ％% = g, 巧一 ％ =瓦 (145.3)
where (物， (孤，％) are the values of the displacements on either side
of the dividing surface and g, h are functions, given on these surfaces
(and not depending on z).
Since one is dealing with plane deformations and all the functions
under consideration are independent of z, one may restrict the investi¬
gation to any transverse cross-section of the bar, just as this was done in
the preceding chapters.
It is easily shown by ordinary means that the solution, if it exists, will
be unique (apart from rigid body displacement). Further, it may be
assumed to be physically obvious that a solution exists. In fact, the
present problem corresponds to the following physical problem which,
for the sake of brevity, will be formulated for the case, where there are
only two parts with transverse cross-sections and S2, divided by the
line L. Consider two bars which consist of the same materials as the given
one, but which have cross-sections S；, S；, different from Si，S2. In fact,
let it be assumed that the cross-section S；is obtained from by im¬
posing on the points of the line L the displacement (•—
— 力)，while
—
S；results from S2 by the displacement (— ^2, %) of the points of L;
further, let
的 一 说 2 = g， ”1 一 02 = 九
If the corresponding sides of these beams with cross-sections S；, S；are
now forced into contact, so that corresponding points touch each other,
and if they are then welded together without disturbing the plane
deformation, the compound bar, thus obtained, will exhibit exactly
the same stresses and displacements as must be expected in the case
of the above problem.
The existence of the solution (under certain ordinary suppositions
of a general character) may also be proved mathematically. This was
done most simply in the paper [20] by D. I. Sherman (mentioned already
in § 103) who considered the case which has been called basic in § 139,1°.
No space will be devoted here to this proof.
With a view to what follows, it will be recalled that for plane deform-
652 VII. EXTENSION, TORSION, BENDING §146

ation
Xz — Yz = 0 throughout the bar, (145.4)

Zz = = MX* + yj 0n Sj> (145.5)

where 入分 are the values of the Lame constant and Poisson's ratio in
the region S$.

§ 146. The problem of extension and of bending by a couple.

In the case of the compound bar for which, however, Poisson's ratio
was constant, the problem of extension and of bending by a couple was
solved very simply and it was found possible to investigate separately
the problem of a tensile force with its line of action along the axis Oz
and the problems of bending by couples in planes parallel to the planes
Oxz and Oyz. The possibility of such an independent study depended
on the particular choice of the system of axes in the plane of the
"left" ("lower") end (and, in fact, the origin 0 was placed at the reduced
centre of gravity, while the axes 0%,Oy were directed along the principal
reduced axes of inertia of this end).
It will be seen later that in the case of different Poisson's ratios such
a choice of coordinate axes does not, in general, offer the possibility
of solving the above-mentioned problems separately.
However, as will be shown in § 148, one may also in the present case find a
special system of axes which permits separate consideration of these problems ;
however, the determination of such a system requires the solution of several
auxiliary problems of plane deformation.
Therefore, in the present section, the system of axes Oxy will refer to
any (rectilinear) system in the plane of the left end S and it will not be
assumed that the plane of the bending couple is parallel to one of the
planes Oyz.
1°. Let My and Mx denote the projections of the moment vector of the
bending couple on to the axes Oy and 0%, and F the magnitude of the
tensile force with line of action along the axis Oz.
Guided by the form of the solution for the case of constant Poisson's
ratio, an attempt will be made to satisfy the conditions of the problem
by linear combinations of the following three solutions :
Zz = E$x>
Zz = Ejyt
% =
= —
— *02 +
&
一 叼y2),
= — 妥02+
& = _— 叼％2),
渺 = (146.1)
矽 = yz, (146.2)
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT a 653

= — = —
”
Zz = 芯， ％ 侬 = z, (146.3)
in the region S；(the remaining stress components being equal to
zero).
If all the Poisson's ratios were the same and if the coordinate axes
were chosen as indicated at the beginning of this section, these solutions,
multiplied by suitable constants (the same constants being used for
stresses and displacements) , would give those of the problems of bending
by a couple in the plane Oxz, of bending by a couple in the plane Oyz
and of extension by a force, directed along Oz.
In reality, however, solutions constructed in this manner do not
satisfy the conditions of the above-mentioned problems for the reason
that the corresponding displacements have discontinuities on the dividing
lines between the sections S力 Sk.
In order to remove these discontinuities, the solutions of three auxiliary
problems of plane deformation will be constructed which represent
particular cases of the problem, formulated in § 145; the functions g, h
in the formulae (145.3) will now be given the following values :
gl = Ms - M（炉 一 词， 友 （5• — 以）％夕， (146.U)
g2 = 一 %）冲， 海 — 以）（外 — 炉）， (146.2Q
g3 = — 为 — 5)y (146.3a)
on the dividing lines between the regions S力 Sk.
For the sake of brevity, these three problems will be denoted by
(146.U), (146.2«), (146.30 respectively and it will be assumed that they
have been solved.
The components of displacement and stress, corresponding to these
three auxiliary problems, will be denoted by superscripts (1), (3)

In particular, one will have in the region S$

(146.助
Z厦 = %・(X望 + 丫俨)， (146.2b)
Z罗 = %・(X纵 + 丫俨). (146.33
Superposition of the solutions (146.1), (146.2), (146.3), multiplied
by certain constants 旬， % respectively, and of the solutions of the
problems (146.1a), (146.2«), (146.3d), multiplied by the same correspond¬
ing constants, is easily seen to give the solution of the problems of bending
and extension of a bar for the following values of the moments MXf Mv of
654 VII. EXTENSION, TORSION, BENDING § 146

the bending couple and of the magnitude F of the tensile force :

— My = (Tn + a1)旬 + (712 + %2)02 + (713 十 %3)03，

— (221 十 + (722 十 42)以2 + (723 + K23)a3， (146.4)
F — (731 + 41)旬 (732 十 岛以02 十‘33+ K33)他，
where

- yy
S
e,㈤%的族 沟 上
为"x^x^dx
S炉
dy, (146.5)

衣丽 = Ex^Z^dx 玛 / / x^Z^dx dy, (146.6)

s s,
a, 8 = 1, 2, 3,
and %a) = %, (2)
力 — 夕，％(3) = i ；in more detail,

where S® = £ is the same as before and xQ> y0 are the coordinates

of the reduced centre of gravity of the end S; Ilt and /22 are reduced
moments of inertia of the end S with regard to the axes Oy and 0%,
and /12 = ，2i is the reduced product of inertia. Further,

--— dx dy, & %Z£2) dy, K13 xZ^dx

s 8

〃yy S
yZ，) dxdy, &2 =

z^dy, &
5
覆 dxdy, &3 =

徵也乂 dy, K33 =

s
//yZ^dxdy, (146.6')
v' a

dxdy\
s s
it will be assumed that these constants have been calculated.
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 655

The problem will be solved, when the unknown constants alf %

have been determined from the system (146.4) for given My, Mxr F.
The determinant of this system

7n + '12 + &2 ^13 + &3

A= 221 + X21 722 十 区22 /23 + K23
/31 + X31 /32 + X32 /33 + %3
(as will be proved below) is always different from zero', more exactly,
A > 0. Hence the system (146.4) determines the constants 劭，口3
uniquely and the problem may be considered solved.
2°. Before proceeding to the proof of the inequality A > 0, certain
formulae will be considered which are connected with the expression for
the potential energy of deformation and which will be required later on.
It will be recalled that the following expression was introduced in
§20:
2W(e) = 入⑸a；+ 勿初 + %z)2 + 2r(落 + 统 + 邑 十
+ 2雁 2鹰 2魔) (146.7)
which represents twice the potential energy per unit volume, corre¬
sponding to the strain components exxt . . exy; this deformation will be
denoted by (e), and accordingly %(e) has been written above instead
of simply W, as was done in § 20.
The expression W(e) represents a positive definite quadratic form
of the components of deformation (e) and it only vanishes for
(ie, for exx = %财 = = ^yz = ~ %期 = 。)・
0 —
It will be remembered that a quadratic form Q(力1，&，• •♦,%") of the variables
力2，.. ，*”is called -positive definite, if Q(力卜 ^2, ..., 力门) > 0 for all (real) values

of the variables, except when xr = x2 = .. . = = 0. The form is called positive

semidefinite, if for all values of the variables “
x2, . . ., 力仅)
exist some (real) values of xlt x2, . . xn, not all zero, so that 0(^, 力
0, i.e., if there
％*) = 0.

The stress components, corresponding to the deformation (e), are

given by the formulae

— 入 e 2^x63#, yy 入 e [ z? — 入e i
Yz = 2(1%, Zx — 2“％«, Xy = 2肥 抄 (146.8)
(9 ^xx 十 e” + &zz)
656 VII. EXTENSION, TORSION, BENDING § 146

and, accordingly, (146.7) may be written

2W(e) = X/w + Yyeyy + Zz^zz + 2Vz%z 十 ?Z 忠工* 十 2Xyexy. (146.9)
Consider now two different deformations (e‘) and (e") and indicate
the corresponding strains by one or two accents. The following expression,
analogous to (146.9), will now be introduced :
2W(d, J) = XX + 瑞胎 + NW + 2Y；总 + 2ZX + 2X；
虞=
= 兄点 + 骂媒 + /X + 2Y；
媒 + 2ZX + 2K质. (146.10)
If one interprets the stress components X；, etc. and X；, etc. as their
expressions in terms of the strain components 曦，etc. and e晨，etc., then
W(e\ e") reduces to a bilinear form in these last components. The equality
between the two expressions for W(e', e"), given in (146.10), may be
verified directly ; this proves that
印(d, e〃) = 印(J, d),
i.e., that the bilinear form W{er , e〃)is symmetrical.
If the deformations (e‘) and (e〃) are identical, i.e., (e‘)
then
= (e〃) — (e),
e) = W{^, (146.11)
where W{e) is the same as in (146.7) or (146.9).
It has been proved in § 20 that

E
(Xnu + Y/ + ZnM四 = 2 yyy
V
巾⑹勿 办 dz = 2氏 (146.12)

where S is the surface of the deformed body, n is the outward normal

and V is the region, occupied by the body; U denotes the potential
energy of strain of the entire body.
The reader will easily prove in an analogous manner the following
formulae:

2Ui2=
E
+ Y；W + 2；/)aw = 2 口［0(吟*次
0
的除

(146.13)
2力1=
〃
s
(X淑' =2/
〃 W(e,ftef)dxdydz.
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT a 657

〃
It follows from (146.10) that = Si，i.e.,
ff (X；“ + 丫材’+ Z>/')
s
涯 =
£
(XM + 丫北 十 (146.14)

which expresses the Reciprocal Theorem due to Betti (more correctly,

the theorem of Betti has a somewhat more general form which also in¬
volves body forces) .
The preceding formulae will only be applied here to the case of plane
—
deformations of a bar. In this case : Xz = Yz = w Q, and all the
functions under consideration are independent of z. Hence
W(* + y；
媒 + 2局媒 = + 驾媒 + 2X；e =现

= 入(%力 + ^yy) (^xx + + 2[1(60刀勺工 + ^yy^yy + 2%?勺y)‘ (146. 15)

W(e, e) = 印⑹ = 入&。+ + 2^(4 + % + 2场). (146.16)
In the case of plane strain, it is more convenient to apply the formulae
(146.12)—(146.14) not to the entire bar, but to a segment of unit length
included between two normal transverse cross-sections. Instead of

〃
(146.12), one then obviously obtains

/ (Xnu + Ynv)ds = 2 W^x dy = 2U, (146.17)

where U is now the potential energy per unit length of the bar, and,
instead of (146.13), (146.14),

2力2 = 2% = /
L e
/
+ Y#)ds = (X*' 十 丫蕾)康 =

=2〃
厂

W^^e^^dxdy. (146.18)
s
In these formulae S denotes the cross-section of the bar and L its boun¬
dary.
In the case where the displacement components, as in the auxiliary
problem of plane deformation of § 145, have discontinuities at the dividing
lines between the parts S” has to understand by L the union of the
boundaries o/ these regions, so that, if L$ is the boundary of S$, the integral
is to be taken along the whole of Ljt and those parts of L》which are com¬
mon to the regions Sj} S“ will be covered twice, once in the capacity of
658 VII EXTENSION, TORSION, BENDING § 146

boundary to and a second time in the capacity of boundary to S；

see (146.21) and (146.22) below.
3°. Consider now the proof of the inequality A > 0. Let (^(2)), (^(3>)
denote the deformations, corresponding to the auxiliary problems of plane
deformation (146.1a), (146.2^), (146.3a), and Ua^ (a, p = 1,2,3) the ex¬
pressions (146.18) with (&⑷)，(e taking the places of (e‘)，(e〃). It will be
proved that
-
2Ua^ a, 0=1, 2, 3, (146.19)
where 改即 are the constants, defined by (146.6) ; in particular, it will
follow from this that 区即 = 岛寸
For this purpose transform the formula

2。殖 = + Y^v^ds (146.20)

L
in the following manner. As stated earlier, L denotes here the union of
all contours bounding the parts S；of the region S. Hence

2% = e/ L}
[X^*) + Y^v^ds, (146.21)

where L§ is the boundary of S$ and 续), *) are the boundary values of

勿⑶，万的 on Lj from the direction of the region $•; n denotes the normal
to which is outward with respect to 5户
Noting now that, by supposition, Xn = 丫您 = 0 on the boundary of
S and that during the integration the dividing line Lkl between Sk, Sl
is covered twice, it is easily concluded that (146.21) may be rewritten

20邓 = E
AU J
八圣幻(说淤一 哦)) 十 y，
(磅) — 嫡加 工 (146.22)
工磊

where now the lines Lkl are only covered once and v is the normal, directed
from Sk into S”
The truth of (146.19) is easily proved by use of (146.22). Consider,
for example, the relation = ^21 = 2?712. By (146.22),
2 八xf )(唬)一 说2)) +
2% = AU (噌 一喂)您
J
工制

Noting that by (146.2«)

42) —说 公
= — 6)冲， 噌 — 吸) — 铲 — X2)
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 659

and substituting these expressions in the preceding formula, one obtains

24 = AUX M 一 内 J｛X^xy 十 强了心)廿 一 力2)心

Ljci

Applying to this expression the same transformation which was used

to deduce (146.22) from (146.21), but in the opposite direction, one
finds

2% = Z [X^xy 十 皆y料俨 一 环磔,

where and % are the same as in (146.21).

Further, noting that
cos (如 x) + cos (% y),
Y俨 = Yg) cos (n, 力) + YjD cos (舞, y),
and transforming by use of Green's formula, one obtains

2U】2 = 2%
J J y(X^ + 斗吟办；
Si
砂,

since
ax俨 y

沏
十
dy
0，
dx
/ ।
~T"
dy
—— _if*

— cv ,
L 1 •

hence, by ( 146.1g,

2U yZ^dx dy = K21.

In exactly the same manner, applying the formula

2% = [xe)% / 丫 r0 w$
L

for the calculation of 2U12 = 2t/2i> one finds 2U12 = Kl2. Hence
2U]2 = 及12= K21，as was to be proved.
The remaining formulae (146.19) may be proved in quite an analogous
manner ; this will be left to the reader.
On the basis of (146.19), the determinant A may now be considered as
660 VII. EXTENSION, TORSION, BENDING § 146

the discriminant of the following quadratic form in terms of %,«?，％：

2Q&, 闲 = 2Go&, &, 他）+ 2G（的,％ 他）， （146.23）
where
3 3
2G0（«1，羯 他）= £ £ （146.24）
a— 1 3~1

33 3 3
2G（口i，仰，的）= £ Z 衣邓％% = 2 S £ U^a^. （146.25）
— —
a 1 3 1 a— 1 0 1
—
It is easily seen that the quadratic form Go is positive definite, i.e.,
that Go（%,沏，a3） 0, unless 的 = & = 他 = 0. In fact, it follows from
the very definition of 7aB that

2Go（%,&，a3）= jj a2y + 陶）2〃％ dy,

s
which proves the statement.
It should be pointed out that it is readily verified by the help of
（146.12） that G0（alt itself represents the potential strain energy
per unit length of the bar, arising from the superposition of the deform¬
ations corresponding to the solutions （146.1）—（146.3） after they have
been multiplied by 旬，仇乞，& respectively（where it has been assumed that
the component parts of the bar deform independently of each other, i.e.,
that they are not welded together）.
It is likewise easily proved that the quadratic form G（旬， 口3） is
positive definite, unless all the Poisson's ratios have the same value.
（If this is the case, then obviously all K* vanish and the form G（%,〃2,〃3）
is identically zero.）
In fact, it may be shown that G（%, a2, 仇3） represents the potential
energy of deformation per unit length of the bar, arising from super¬
position of the solutions of the auxiliary problems （146.1a）, （146.2^）,
（146.30 after multiplying them by alt &，口3 respectively.
Indeed, let, as before, （e （e 伍⑻） be the （plane） deformations
corresponding to the problems （146.1口），（146.2〃），（146.3。） and let （e）
denote the deformation
（e） = %（第））+ 倪2（承）十 咏⑶），
i.e., the deformation with strain components
= 的嫩 &螺 他喀，…，分v 徇噬 被盘 33曙
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT CT 661

By (146.17), the strain energy per unit length of the bar is given by

2U = J/ w⑻热 为，

where W(e) is defined by (146.16). However, it is readily seen that

w⑻ = alW(e^) 谑印(*)) + alW(e^ 十 2的的印修⑵田⑶) 十

+ 2他的印(m), e 十 2%他印(0 e
and hence, taking into consideration the definition of U^, 2U =
= 2G(痣卜 @2,〃3)，as was to be proved.
If not all the Poisson's ratios have the same value and if at least one
of the quantities 曲，做，他 is different from zero, deformation necessarily
takes place and therefore U > 0. Thus the original proposition has been
proved.
If there is no deformation, one obviously has on the dividing lines 工狄 between
the parts S力 Sk

ui —
，
做 = 一 号秒 + 看%， 一 叼/ + B业，
where 匕沥 a? 自方左 are constants; on the other hand, one must have on these lines,
—
by (146.1a) (146.3a),
一 ％为工 一 合) 3a式斓 俨) 4- 做％夕 a解]，
一 ％ = (% —“)[4/y + 加2(俨 一 整) 4血 ；
it is easily verified by comparison of these expressions that, if % 声 cr无， one must
have 旬 = a? = % = °， 。乐 = a泳 = B汴 = 。・

The form Q(%, a2> 他)，being the sum of the two positive forms G。， G
of which the former is certainly positive definite, must also be positive
definite. However, it is known that the discriminant of such a form is
certainly positive ; therefore the assertion made at the end of subsection 1°
with regard to A is proved.
NOTE 1. The fact that Q(%, a2, %) is positive definite could have
been proved more simply without splitting it up into the forms G() and G.
Such a proof may be carried out, based on the fact that Q = G° + G
is the potential strain energy, corresponding to the earlier stated com¬
—
bination of the solutions (146.1)—(146.3) and (146.16?) (146.3a) ; this
statement is easily proved directly.
However, a different procedure has been followed here, because it
was desired to calculate the additional coefficients which characterize
662 VII. EXTENSION, TORSION, BENDING § 147

the influence of the different Poisson's ratios of the component materials.

NOTE 2. Generally speaking, the coefficients 衣邓 are very small,

if the Poisson's ratios of the different materials do not differ much from
each other ; in other words, they are of the same order as the squares and
products of the differences — ok. In fact, denoting temporarily by b汴
the differences 斗 — 网 which occur on the right-hand sides of (146.1a)
(146.3d) and considering %% as variables, it is readily verified that the
—
—
solutions of the auxiliary problems (146. 1 «) (1 46.3。) depend linearly
on %. Further, by (146.22) and due to the fact that X*, Y* depend
linearly on o汴，it is seen that depends linearly on the squares and
products of cr泳，as was to be proved.

§ 147. Particular cases. 1°. Extension o f a bar, having

a n axis o f symmetry. It will be assumed that the axis Oz is an
axis of symmetry of the bar, where the symmetry refers to geometrical
as well as elastic properties.
In that case the origin 0 is obviously the reduced centre of gravity
of the "left” end. Directing the axes Ox, Oy along the reduced axes
of inertia of this end, one has Z12 = 0. Further, on the basis of the sym¬
metry and of the form of the functions g3, 饱 in the formulae (146.3a),
it is easily seen that the solution of the corresponding auxiliary problem
will likewise be symmetrical about 0 and, in particular, that

Z£3)(—%, -y) y).

〃
It follows from this that

K31 = &3 = xZ^dx dy = 0, &2 = &3 = yZ^dx 办 = 0.

s s
Hence the equations (146.4) have the form (remembering that %0 = % = 0)
一 M = (^11 +
期 + &2以2，
A'* = 衣21旬 + (/22 + K22)02，
F = 十 K33)他.

If it is proposed to solve the problem of extension by a force of magni¬

tude F, directed along the axis of symmetry Oz, one must put in these
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 663

equations Mx = My = 0, i.e., ax = a2 = 0, and

F
“ (147.1)
S^；十 K33
If all the Poisson ratios of the different materials, constituting the
bar, have the same value, then K33 = 0, and one obtains the result,
deduced earlier. If not all the cr's are equal, then necessarily K33 > 0.
In the last case, the quadratic form 2G(%,a2, @3) is positive definite and therefore
all the coefficients Ru，K22, K33 are positive ; this follows from the fact that
= 2G (1 0, 0), etc.
j

Since a3 represents the relative lengthening of the bar as the result

of F, SE + K33 is the rigidity of extension and the preceding formula
shows that the dijference of Poisson's ratios {for constant S^) increases the
rigidity of extension, independently of the sign of the difference crfc. —
2°. Bar with plane of symmetry, bent by a couple.
Let Oxz be the plane of symmetry of the bar (as regards its geometry as
well as its elastic properties) . It may then be assumed that 0 coincides
with the reduced centre of gravity of the "left" end; the axes Ox, Oy
are again to be principal reduced axes of inertia of this end with regard
to 0.
—
For this choice of axes, one has in (146.4) : 713 /23 = /12 = 0. Further,
on the basis of the symmetry and of the form of the functions g】， h工
in (146.1g), it is easily concluded that the solution of the corresponding
problem of plane deformation is likewise symmetrical with regard to Ox;
in particular, Z^{x, y) =
= — y). Hence
— y). Similarly, one finds (尤，一 y) =

= ^21 = yZ^dx dy = 0, 玛3= X32 = Z产族 沟 = 0

s s
and the equations (146.4) take the form
— My ~+ Ki】)旬 + K13他，
(ZU
Mx = (% + &2)&， (147.2)
f = X31“1 十 (S_e + K33) 他.
If it is desired to solve the problem of bending by a couple whose -plane
perpendicular to the plane of symmetry, one has My = 0, F = 0;
664 VII. EXTENSION, TORSION, BENDING § 148

hence 旬 = 他 = 0 and
Mx (147.3)
‘22 + -^22
where K22 > 。，
unless all the Poisson's ratios have the same value.
However, if it is proposed to solve the problem of bending by 例 couple
whose plane parallel to the plane symmetry, one has Mx = 0, F = 0 ;
hence
^1-^31 My
sa 可' 一质=工7中 一 (147.4)

where
区；3 %&工 + K1K i3 一 Ab
Se — 43 Se 十
(147.5)

If not all 网 have the same value, then K > 0, because KUK33 一 > 0,
as this is the discriminant of the positive definite quadratic form in the
variables 丽，a3
2G(曲，0, ~ Ki%；+ 2Kl3al他 +
It is easily seen that in both the above cases the Bernoulli-Euler law
holds true and that in the first case the flexural rigidity is equal to
?22 + &2, (147.6)
while in the second case it is given by
/11 十 K； (147.7)
it must not be forgotten that Z22 and Iu are now reduced moments of
inertia about the axes 0%, Oy.
It is seen that in both cases the difference of the Poissons ratios in¬
creases 统e flexural rigidity (for constant 1^ and Z22), independently o/
the sign of 4 —
Some simple examples will be presented in § 149.

§ 148. Principal axis of extension and principal planes of

bending. The equations (146.4) may be considerably simplified, if the
arbitrary system of coordinate axes in the plane of the "left”
("lower”) end is replaced by another system 0'%'y' in the same plane,
where the new axis O'z' is given the same direction as the old axis Oz.
In fact, as will be seen below, this new coordinate system may be chosen
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT G 665

in such a way that on the right-hand sides of the equations (146.4)

all but the coefficients on the main diagonal vanish.
Let K% denote the constants for the system corresponding
to the constants 衣邓 for the system The relations expressing in
terms of Kg are easily found ; this will be left to the reader (cf. Note at
the end of this section), and so, in what follows, only those relations
will be deduced which are required in the later reasoning.
For greater clarity, the transit to the new axes will be carried out in
two steps, producing a translation of the origin and a rotation of the
coordinate axes.
Let the new system O'x'y' only differ from the old system Oxy by the
position of the origin and let a, b be the coordinates of the new origin in
the old system, so that

— 讥， y' = y — 6.
It is easily seen that in the present case K；
3 K33. In fact, in the auxi¬
H

liary problem corresponding to (146.3勿)， but in the new coordinates

Ofxfyr, one will have for the discontinuities in the displacements on the
dividing lines
Uj ——
uk = (与 ——
o ％，= (%•—— Ofc) (% ~ d) = (%■
•
——
crj% 十 const.,
盯 — ％= — = 一 &) (y — b) = (%・ — + const.,
and clearly the solution of this problem leads to the same stress distri¬
bution as the solution of the problem for the following discontinuities :
% ~ uk —
= (弓 bj%, Vj % = — — G比)y,

because the constants in the previous formulae may be removed by rigid

translations of some of the parts, constituting the bar. Thus, in particular,
the stress component will be the same for these auxiliary problems
in the systems Oxy and This means that

remains unchanged for the transit to the new system.

Next, the constants K% = and = KQ will be calculated.

〃
Using (146.6') and the above result regarding Z£), one has

居3 = dyf = - a)Z餐藤 dy,

666 VII. EXTENSION, TORSION, BENDING § 148

whence
K]3 = Ki3 — 口K33； (148. 1)
similarly
K& = K23 一 勿％3 (148.2)
Denoting by /品 the quantities, defined for O'£y' in the same way as

，
/邪 was defined for the coordinates a, b will be chosen such that
As + &3 ~ 十 居 3= 0 /23 十 岛3= 6由0+ k23 = 0，
or, since %；= 孙 一 。，媪 = — b, one obtains by the preceding formulae
。 + 区13 Sb% + &3力
_ ! Ao /

Se + K33 Se + K33
With these values of a and b, the formulae, corresponding to (146.4),
but for the new system of axes, acquire the simpler form

— Mx —
My
—
(Wjj + Ki])% + (^12 + Ki?)%
(121 十 区21)N1 十 222十 区22)02，
(148.4)
F = (Se + K33) 讥3’
where, for simplicity, accents have been omitted, i.e., My, Mx, ?婢，"磷
have been written instead of M 小 M,,, 〃，
K》. Accordingly, the new
system of axes Ofxfyf will now again be denoted by
The new axis Oz will be called principal axis of extension {compression) .
The principal axis of extension may also be determined in the following manner.
In (146.4), let 旬 = 映 = 0, % W 0. Then
= (6七％0 十 衣13)g， = (Sj5夕o + X23)03， F = 3e 十 及33)03，
Thus, in the present case, the forces applied to the "right" end are statically
equivalent to a tensile force of magnitude F 片 0, directed along the axis Oz, and
to a couple with moment perpendicular to the line of action of the force. However,
such a system of forces is known to be statically equivalent to a force of the same
direction and magnitude. The line of action of this last force is easily found and it
is the principal axis of extension, defined above.
This term is justified by the fact that, if tensile forces of magnitude
F be applied to the ends of the bar for which the line of action is the
principal axis of extension, the solution of the problem will be obtained
by putting 西 = & = 0,

Se + &3
so that the extension will not be accompanied by bending.
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 667

The preceding formulae show that the rigidity of extension is equal to

Se + (148.5)
Since for different Poisson's ratios K33 > 0, it is seen that the dif¬
ference of Poisson's ratios (for constant SE) increases the rigidity of
extension, independently of the sign of
—
“，a circumstance which
had been observed above for a case with axes of symmetry.
The formulae (148.4) may still be further simplified by means of a
rotation of the axes in their plane.
If the new system of axes O'x'y' is obtained from the old system by
rotation through an angle a, then
% —
cos a y' sin a, y = %' sin a + y' cos a, (148.6)

— x cos a + y sin a, y' — — sin a 十 y cos a. (148.7)

xf %

The quantities K), K% — KQ the new system will now be

小
expressed in terms of = K21, K22. For this purpose, the auxiliary
K12
problems corresponding to (146.1〃) and ( 1 46.2a) will be compared with
the auxiliary problems in the new system.
The discontinuities of the displacements for the above-mentioned
problems in the old system are given by
如 一以 = 强(%• — 取 ) (炉 — y2), 一 以 = (%■ — 以)到 (I)
for the problem (146.1〃)，
物 一瞅 = — 到， — 一 u)(外 一 力2) (H)
for the problem (146.2痣). Correspondingly, one has for the new system
得一短 = 看⑹ 一 内出2 旷2),0；一 球 — ^x'yf (V)
and
%；—k = *6 — 以)，歹， 心 一球 6 一 书 (y'2 炉2). (II') —
In order to compare these problems, the boundary conditions (I') and
(II') will be expressed in terms of the old coordinates %, y. In fact, by
(148.7), one obtains instead of (I')
%；—域 = — — y2) cos 2a + — sin 2a,
—媒 = (巧 一 cos 2a 一 务(％
— crfc)the(峭quantities
— y2) sin 2a.
Introducing now instead of 悌; 一 蝮 — %论，
叼 — vk, expressing the same discontinuities in terms of the old coordinates,
—
668 VIL EXTENSION, TORSION, BENDING § 148

and taking into consideration that

— %* = [Uj — %乃) cos ot (% — sin Gt,

啊一％ = (%；— %)sin a + — 心) cos a,
one obtains
物 一％ = —
去⑹ 一 “) (％2 y2) cos a 十 sin a, — (148.8)
% — % = (j — —
a^)xy cos a 十 丛j 以) (外 一 x2) sin a.
The deduction of (148.8) may be simplified by considering, instead of x, y and
u, v, the variables 3 = % 十 Ey and u 十 谢.

Thus it is seen that the solution of the problem, corresponding to (I'),

may be obtained by adding the solutions of the problems, corresponding
to (I) and (II), which must be multiplied beforehand by cos a and sin a
respectively. Hence, if denotes the component Zz, corresponding to
the problem (I'), and Z% as before, the stress components Z^,
corresponding to the problems (I), (II), then
Z； = Z?) cos a •+ Zg2) sin a. (148.9)
Similarly, one obtains for the problem (IF)
_
Z；(2) = z£D sin a + Z，) cos a. (148.10)
Using (148.9) and (148.10), the quantities K%, K%，K；2 are easily
expressed in terms of Kllf K12, K22. For example,

= yy
s
yfZ^dxf 沙 =

=//
s
(― ^五口^ 歹 cos °C (Zf cos a + Z^) sin dy,

whence
K；2 = &2 cos 2a — — &2) sin 2a. (148.11)
The expressions for and may be deduced by the reader (cf. Note
at the end of this section).
Now the expression will be deduced for the reduced product of inertia.
in the new system. One has

s
j Exfyfdxr dyr =
s
cos a 十 y sin a) ( — % sin a + y cosa)〃％"y,
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT a 669

whence
= G cos 2a — MG — %)sin 2a； (148.12)
the complete analogy with (148.1 1) is obvious (see the Note at the end of
this section).
The angle a will now be chosen in such a manner that
+ K” = +
『21 “21 = °， (148.13)
By (148.1 1) and (148.12), one obtains
(人2 + &2) cos 2a ― |(/ii + — /22 — &2) sin 2a — 0,

whence

tan 2a - 2(?12 + "12)

Mu. X22
Al 十 T
— —
622
- 1 4).
( 1 48.

Giving a one of the values, satisfying this condition (the other values
differing by integral multiples of a right angle) , one arrives at a system
of axes Ox'y' for which (148.4) assumes the very simple form, mentioned
at the beginning of the present section,
一 My — (/ll + -^11)^1»
^2/ ~ (/22 + X22)，2， F = 6e + *33)，3，
because, as is readily seen, one has in the new system 几 十 K% =
~ 5芯％0
+ e13 = 0， 723 十 r23 = ^Ey() + r23 = 0， and, in addition,
K33 = 及33・
The planes Ox'z and Oy'z will be called principal planes of bending.
It is seen that, if Oz is the principal axis of extension and if Oxfz, Oy'z
are the principal planes of bending, the problems of extension by forces
with the line of action Oz and of bending by couples with planes parallel
to Ox'z, Oyfz may be solved independently of each other.
Omitting the accents, the last equations may be rewritten as
一 Afy = (Ju + Ki，％, Af w h F = (S_g + 左33)&・ (148.15)
(/22 + 长22)口2，
It is seen that the law of Bernoulli-Euler is valid for bending by couples
with the planes Oxz, Oyz and that the respective flexural rigidities are
given by
&+ % + &2； (148.16)
the rigidity of extension is again equal to
Se + &3・
670 VII. EXTENSION, TORSION, BENDING §149

NOTE. It will be left to the reader to verify that for transition from
one system of axes Oxy to another 0'%'y' the quantities KaB 笈少 trans¬
formed in accordance with the same formulae as the quantities
Instead of a simple verification, this property may be deduced by investigating
the general expression for the strain energy of the deformed bar.

For example, for a translation of the origin 0 to a new position 。'他 b)

In =
s
Ex' dy' =
S'
(力 — d)2dx dy = — 2«713 + «2Z33 ;
in correspondence with this one has
Ku = Kji
_ 2口&3+ N 2K33.
It follows from this result that, from the point of view of simplifying
the notation, it would have been expedient not to consider the quantities
Kap，^pseparately, but to consider their sums = 7aB + 衣邪 which alone
occur in (146.4). This has not been done (cf. § 146, Note 1), because
it was desired to distinguish clearly the terms KaP which only occur in the
case where 与 一 & # 0.

§ 149. Application of complex representation. Examples. 1°. In

order to find the solutions of the auxiliary problems of plane deformation,
it is convenient, as in many other cases, to use functions of the complex
variable
X ly.
The general solution of the equations of plane elasticity for a ho¬
mogeneous isotropic body (§ 32) will now be written in the form

+ 谢 = 码(3) -
勿 ~ B丽， (149.1)

+ 丫“ = 猊中'(3), 丫旷 一 X。十 22Xy = 2叵中〃(3) + 4/(3)], (149.2)

where 9(3), 皿肯) are analytic functions of 3 in the region under con¬
sideration and
入
+ 3“ (3 — 4e)(l+b) 1
r
2“ 2“(入+“) E 2[z E
In the case of the auxiliary problem of § 145, the constants a, 8 have
different values 自
电， in the regions constituting the cross-section
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 671

of the bar, and the functions 中(3), 少(3) are holomorphic in each of these
regions (the multi-valued terms in the functions 9, 少 drop out in the
present case, because the resultant vectors of the forces, applied to the
boundaries of the regions S力 are all equal to zero) .
It will be recalled that the components Xn, Yn of the stress vector,
applied to the element of any contour from the positive direction
of the normal n, are given by
(X" + iY^ds = - + 3西)+ 砧］, (149.4)
where it has been assumed that the positive directions of the normal %
and of the element ds are orientated with respect to each other as the
axes Ox, Oy.
In correspondence with this, the condition (145.1) may now be written
甲(3) + 3彳因 + 丽 = const. (149.5)
on the boundary of the region S, while (145.2) becomes

m⑶ + 3而)+ 中⑶上 = h>(3) + 3w '⑶ + 少⑶h + const. (149.6)

on the dividing lines between the parts S“
Further, the condition (145.3) takes the form
团平⑶ - 83b 一 阿⑶上
on the dividing lines between S；,
- Sk>
［冲(3) — P3M3) — 时⑶卜
/
(149.7)
where denotes functions given on
these lines. In the cases (146.1a), (146.2a), (146.3^) respectively one will
have
/ = gi + % - 变⑹一“)资， (149.80

/ = g2 十 帆 = 一 g (叼一 吟3?, (149.82)

/ = g3 + '% = — (149.83)
2°. As an example, the case will be considered where the free surface is
a circular cylinder and the dividing surface between the two materials
is likewise a circular cylinder with the same axis. Let the region Sr be
bounded by a circle with radius and the region S2 by the same circle
and a circle with radius R2；the origin 0 will be placed at the centre of
these circles.
As a consequence of the symmetry, it is obvious that the axis Oz
will be the principal axis of extension and that the planes Oxz, Oyz will
be principal planes of bending.
672 VII. EXTENSION, TORSION, BENDING §149

In the present case, the solutions of the auxiliary problems are easily
found by expanding the functions 9 and 中 in the regions 邑 and S2 in
positive and in negative and positive powers of 3 respectively. Substitut¬
ingin (149.5), (149.6) and (149.7), the coefficients are immediately determin¬
ed; any arbitrary constants which may occur do not influence the stress
distribution (because of the uniqueness of the solutions of the problems).
However, the case to be considered here is somewhat simpler, as the
form of the solutions may be guessed immediately and, instead of infinite
series, only a few terms need be retained (see later).
3°. The problem of extension will be solved first. It is easily guessed
that it will be sufficient to write in this case
少i(3) = "13， 巾式3) = 。
in Si，

中2(3) = /'23， 少2 in S2，

3
where Alt B2 are real constants and the subscripts 1 and 2 with
the functions 9 and 中 indicate the relationship of the functions to the
regions Slt S2.
—
For / = (s &)3，the conditions (149.5), (149.6) and (149.7) respec-
tively give, omitting arbitrary constants,
243 十 =0 for I 3 I = &,

24# = 243 + for I 3 I = (149.9)

(/ 一 阮)力13 = (“2 — M^23 一 82323T + (% — ^3 for I 3 I =
Further, for 3 = 夕&吗 g — 必一吗 one obtains, after division by e*
2,4 2&
+& — 。， 2A1R1 = 24 2& +
&
(al — = (82 — 32那2& —
Hence
(5一曲(星—础
A1
(a2 — %)尺； 2取超
「

(ai ~ d)(段 一段
力2
璃 /)49.10)
31 — 81) (五2 一 尺；) (a2 —
'
+ 2p27?2
4 =
2( j — 泗底
b .
E 一 仇) 尼 — 玛) 32 — 陶段 2限周
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT (7 673

the quantities p!，a2, 82 are given by (149.3), where E and cr must be

given the corresponding subscripts ; in fact,

m 一d =31也j 二 &一 风=
2(1 + %) (1 — 2%)

31r
Since always % <:玄 these expressions are all positive.
Superimposing the above solution of the auxiliary problem, after
multiplication by as, on the solution (146.3) which must also be multiplied
by %, one obtains the solution of the original problem, provided 他 is
given the value
(149.11)
他二u瓦？
where F is the magnitude of the tensile force,
品 = 邑/ + S2E2 = 冗因G + 阳 一 R^, (149.12)

and
&3 =
〃 s
Zfdxdy

Z£) = + in $ = 1,2) in the notation of § 146.

In the present case
巾X£) + 丫俨) = 纲 脚;(3) = 4%4 in M (/=1,2).
Hence
K33 ― 1 + S/z月2） ~

4兀（s 国2 解一 础心 .
2"g
一
不

区 一 的（段 一 滥） （m - 限）史 '

'
as was to be expected, for 々 K33 > 0, as it only contains the
factor (ctj — a2)2.
4°. Next consider the problem 。/ bending by a couple, assuming its
plane to be parallel to Oxz. In this case the conditions of the auxiliary
problem of plane deformation, corresponding to (149.8。，may be satisfied
674 VIL EXTENSION, TORSION, BENDING § 149

by writing
= A.^^, Wi = 0 in S],
(149.14)
91 = ^z32> 少2 = o~ + C? in $2,
3
where Av A2, B2> C2 are real constants.
Substituting these expressions in (149.5), (149.6) and (149.17) with
/= — ct2)32 gives, as in the preceding example, four equations
for the determination of the constants Alf A2, C2 which are easily
solved and render the values of these constants. Only the expressions

_
for the first three of these constants will be given here, since C2 does not
influence the stress distribution :

二 I 伉 — %)(周 一段)
L 2
修(尼 一班) a刷十 仇思’ +
—看 结 F 孕
心
修保2 一 a1) + 82及 + B2K 1 2
(149.15)

q2 =
一
1
窃.
泗玛
(6 —
出(马 一段) +82段 十园段.

The stress Z，)，corresponding to this auxiliary problem, is given by

MXg) + 4% -
Hence, in the notation of § 146,
他 猊5(3) = 呢4% in S,・ G = 1, 2).

Kji ] j xZ^dx dy = dy + 8cr2^2 / / %?“％ 砂 =

：
s Si S2
== 4M + 2g2N2(尼 一 玛)
or, by (149.15),
九应 3)2(段 一 段闲
一
K (149.16)
(尼 — 号) a2段 62号
Thus, the flexural rigidity is equal to
Ie + (149.17)
(where IE has been written instead of ?口)；one has for IE the formula
大

0=4［玛玛+岛（玛一正扣， (149.18)
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT <7 675

As was to be ex] acted, for % W 衣口 > 0, as it only contains the

—
factor (q %)2.

§ 150. Problem of bending by a transverse force.

The solution, presented in this section, was given by A. K. Rukhadze [3] ;
however, not all statements in that paper are correct. They will only be so, if by
the system of axes Oxyz is understood the system which will be used below and
not that used by A. K. Rukhadze and if one (inessential) modification is introduced
into his reasoning.

Let the axis Oz be the principal axis of extension and the planes Oxz,
Oyz the principal planes of bending (§ 148). For such a system one has,
in the notation of § 146,
'13 + "13 + *13 = 0, ‘23 十 衣23= 十 灭23 °，
人2十R12=0， (150.1)
where xn, % denote the coordinates of the reduced centre of gravity
of the “left” end.
It will be assumed that the bending force of magnitude W is applied
at the point, where the axis Oz intersects the "right” ("upper") end, and
that it is directed parallel to Ox.
The solution for the general case will be obtained by combining the
solution of the above problem with the analogous solution, obtained by
interchanging the roles of 0%,Oy and the solution of the torsion problem
(§139).
Guided by the form of the solution, obtained in § 144 for the case of
constant Poisson's ratio, the solution of the present problem will be
sought in the form
= 一可Z + 4 及j(2 - z) (炉 一 y2) + 权源 — 看於],
力⑼ — (150.2)
论⑼ y)
— 月 y) + — 春力 十 町2]
in the regions S? ; in these formulae t, A are constants, subject to defi¬
nition, <p(%, y) is the torsion function, defined as in § 139, and x(%, y)
is some functions, continuous throughout S and subject to definition.
The stress components, corresponding to (150.2), are given in the
regions S$ by the formulae .
X猥 = 岑) = X俨 = 0, (150.3)
676 VII. EXTENSION, TORSION, BENDING §150

Xg°) = T臼 一 B + 知声2 +（1 — 知力外 ，

y（o） 平”
(告+") B 十（2 + (150.4)

Zg) = - &•(% — z)%,

where
Bj = K$ = AE$・ (150.5)
The displacements (150.2) cannot satisfy the conditions of the problem,
because ut » are not continuous for a passage through the dividing line
of the regions Sjf Sk; in fact, on these lines
一 说* = 以⑹一 珀 (Z
—
Z) (峭 词， -
可°) 一 吨°) 4 6 一 书 R z)xy. —
These discontinuities cannot be removed by finding a solution of the
problem of plane deformation, since they depend also on z.
However, a beginning will be made with the solution of the auxiliary
problem of plane deformation, formulated in § 145, for the following
discontinuities in the displacement components on the dividing lines :
物一以 = g = 妥(邑 — “) (％2 —
/), 巧—— ％ = 勿 = (① — (150.7)
this is the problem (146.1a)
As in § 146, denote the stress and displacement components, corre¬
sponding to this problem, by the relevant symbols with the superscript
and assume the auxiliary problem to have been solved; consider now
the spatial deformation, characterized by the following displacement
components:
〃* = (% — z)% v* (l — —
z)vwt w* 0. —
(150.8)
The corresponding stress components are given by
x^ =(z
Z$ = (%—
— z)z)x然 yj -
(Z - z)Y^\ Xj =(z
戏D = 与修 一 z) (Xg) + 斗D),
— Z)xy), (150.9)
(150.10)
xj = — y* = - 出0 (i5o.il)
in the regions S$.
Finally, the deformation will be written down which is obtained by
superposition of the deformations, corresponding to (150.2) and (150.8),
where the last is to be multiplied by
sponding to the displacements
—
2, i.e., the deformation, corre¬

% = %⑼ 一 月%*, o =
—
Av*, m = m (150.12)
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 677

The corresponding stress components are given by

% = X，) 一 力应， k = 40) 一 力冲， Zz = 理 一 /z 3 (⑻ 功
e =丫护一 2 程， z. = Z如一 AZ" X. = - XX*
Substituting these values in the equilibrium equations, it is readily
verified that they will be satisfied, provided the function y) satisfies
the equation
A/(%,y) = p(%, y) ( 1 50. 1 4)
in each of the regions S” where

、=
P(%, y)
入于 十 3
8(1), 9
a
。)।
加⑴ 物⑴ .
——
F oy m S/, fIS 1 C\
(150.15)
臼 8%
it may be assumed that this function is given, since the auxiliary problem
of plane deformation has been assumed to be solved.
Further, supposing for definiteness that one is dealing with the basic
case of a compound bar (§ 139,1°) and formulating the boundary conditions
on the free surface and on the dividing surfaces, one easily obtains, in
the former notation,

“。(条)。-内(给=力 gl (150J6)

(/ — 1, 2, …，碑+ 1, Rm+i = 0), where the functions

力= 一 {KWo 一 35)炉 +
十 产 一 ％说
D
出吟 D}cos (%,%) —
- {k(2 + %) - 出(2 + 坊即 — 阿球) 内吸} cos (如 y) (150.
are known on L§*
One has thus arrived at the familiar boundary problem (150.16),
except that the unknown function /(%, y) does not this time satisfy
the Laplace equation A/ = 0, but the somewhat more general Poisson
equation (150.14).
However, this problem is easily reduced to the case where the unknown
function satisfies Laplace's equation. In fact, let %()(%, y) be any particu¬
lar solution of (150.14); such a particular solution is always easily
found.
678 VIL EXTENSION, TORSION, BENDING §150

For example, it is known that the logarithmic potential

Xo（%，夕） = log V dg
s

—
is such a particular solution, where * = (% E” + (夕 一 刁户. in practice, however,
it is usually more convenient to find a particular solution by different elementary
methods.

Writing
%(%,y) = &(%,y) + %*(%, y)， (150.18)

where %*(%,y) is a new unknown function which obviously satisfies

the equation A%* = 0, one arrives at the boundary conditions

=疗011乙(,=1,2,. . .,用 1 ;以+1 = 0), (150.19)

where

(150.20)

It is known that the condition of solubility of the problem ( 150. 19) is

given by
m+l C
S / fjds = 0. (150.21)
— J
/ 1

This formula will be somewhat simplified. Substituting for f* its expres¬

sion (150.20), one obtains

or, transforming the last integrals by use of Green's formula,

?n+l
E
?=i J
%
.

桢s
— “° J /Sq
A斓％ dy —
工产 JJ =

?n+l /
=I J Z / 力ds 心Xq"% dy = 0
4 s
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT <j 679

or, remembering that A/o = p(%, y),

m+i r
S / /Us — j j dy = 0. (150.22)
2=1 J
Lj S
It will now be verified as to whether in the present case this condition
is satisfied. Substituting in (150.22) for 力 from (150.17) and transforming
the integrals by means of Green's formula, one readily finds that (150.22)
reduces in the present case to the following condition :

— yy
S
2{i( i 十 dy + yy*
S
成 “％ 沟 一 yy
$
册 。
or, since
23(1 十 b) E, pip = ( 入 + a)8 入6 = Zf),

〃
to the condition

— Exdx dy — =0
s s
or, finally, to

however, this last condition is imposed by (150.1).

Thus, in the present case, the boundary problem (150.16) is soluble ;
its solution is determined, apart from an arbitrary constant term which
does not influence the stress distribution.
If one substitutes this solution for %(%,y) in (150.2), the formulae
(150.12), (150.13) then determine the solution of the original problem
which satisfies all the required conditions on the side surface and on the
dividing surfaces.
It will now be shown that the constants A and t may always be chosen
in such a manner that the forces, applied to the "right” ("upper”) end,
likewise satisfy the required conditions. For this purpose the resultant
vector (X, Y, Z) and the resultant moment of these forces will be calcu¬
lated. Since for z = I: Z£) = Zj = 0, one obviously has Z = 0. Further,
X= 必. (150.23)
s
This formula will now be transformed. By the equilibrium equation,
az? az%-= o；
— 8v ——
dx
+― 8y
-h
680 VII. EXTENSION, TORSION, BENDING ”50
substituting for Zz =
and (150.10), one finds
— AZ* its value, given by (150.4), (150.5)
鲁 十春 A(旦夕 + 2黑) = 0 in S, (150.24)

Thus, one may write

2(%Xz) 0(先 Y\) .
v
Xz = — ——H — ——F 2(E声 十 Z£D) x m
j 7(i)\ •

Consequently (150.23) may be rewritten

I 丁 dx dy +A 的 ffxZ^dx
JJ
s
dx 纱
s s
It has been shown in § 139 that the first integral vanishes. Hence, in
the notation of § 1 46,
X = N(/n + &i). (150.25)

〃
Further, one finds by the same method

Y= Yzdxdy = A(I12 + K^)

s
whence it follows, by (150.1), that Y = 0.
Thus the resultant vector of the external forces, applied to the "right”
end, is parallel to the axis Ox.
It is also easily seen that the resultant moment of these forces about
the point of intersection between the axis Oz and the "right” end is
parallel to Oz and that its magnitude is given by

M= pD 十月 E &
; J J V ox
%
cy + (1 — 她)夕3—
Si

— (2 + |dx Ay 十 2 三内 ff^xvW — yu^dx dy,

$
(150.26)

where D is the torsional rigidity which is known to be always larger

than zero.
Consequently, all the conditions of the problem will be satisfied,
provided the constants A and t are chosen in such a manner that X W
and Af = 0 respectively. The first condition, taking into consideration
—
CHAP. 25 EXTENSION AND BENDING FOR DIFFERENT b 681

(150.25), gives

公
_JL—
in + %
， (150.27)

on the basis of (150.26) and (150.27), the second condition determines c,

since D ^0.
Thus the problem is solved. It is readily seen that also in the present
case the Bernoulli-Euler Law remains valid and that the flexural rigidity
is given by
+ &i， (150.28)
as in the case of bending by a couple.
When the dividing line and the external boundary of the region S are
concentric circles, as in the example of the preceding section, the problem
is readily solved in closed form.
 Appendix 1

ON THE CONCEPT OF A TENSOR

1. Tensor calculus has rapidly achieved recognition in pure as well

as in applied contemporary mathematics and is beginning to enter into
technical literature, in particular, into the literature dealing with the
theory of elasticity. For this reason it is considered necessary to give
here at least an elementary introduction to the concept of a tensor
which, for the sake of simplicity, will be confined exclusively to orthogonal
coordinates. It should, however, be noted that the principal advantage
of tensor calculus arises in its application to curvilinear coordinates of
the general type. In order to give the subsequent definition of a tensor
a more natural background, certain remarks will first be made with regard
to the concept of a vector (since a vector is a particular type of a tensor,
in fact, it is a first order tensor).
It will be assumed that the ordinary geometrical definition of a vector
as a straight segment which has direction is knjwn. Further, coordinate
axes will not be denoted by Ox, Oy, Oz, as in elementary analytical
geometry, but by 0xlt 0x2, 0x3. Correspondingly, the components of
a vector P will not be denoted by g, 办 as in the main part of this
book, but by 匕2,
Only the length and direction of the vector, and not the position of
its starting point, will be considered ; thus, a vector will be considered
completely known, if its components 短，纵，蕊 (ie，its projections on
the coordinate axes) are given. The vector P with components ^2, 蕊
will be denoted by (短，£， £) or，still more briefly, by (%); the index i
takes then the values 1, 2, 3.
Thus, a vector in space is characterized by three scalar quantities.
Many physical and geometrical quantities exist which for a given
choice o/ coordinate axes are likewise characterized by three scalars, for
example : velocity, force (applied to a given point), etc. However, not
every such quantity can be represented as a vector, as may, for example
be done with a velocity or a force. In fact, let El 匕2, £ be scalars character-

682
 APPENDIX 1 683

izing a given physical quantity for 笈 given choice of coordinate axes.

One may, of course, always construct a vector

P= 纵 以)
with components 获，纵 蕊 and claim that it represents the given physical
quantity for the given choice of coordinate axes. However, this relation
between the given quantity and the vector may be disturbed, if the
system of coordinate axes is replaced by another one. In fact, it may
happen that the scalars 已；， 己；， E；, characterizing the original physical
quantity in the new coordinate system, do not coincide with the com-
ponents of the vector P in the new system, i.e., the vector having
in the new coordinate system the components E；, E；, & may differ from P.
In order that the representation of a physical quantity be independent
of the choice of the coordinate system, it is obviously necessary that
the scalars 短，蕊，&3，characterizing it, transform in the transition from
one coordinate system to another according to the same law as the
components of a vector. It may only then be said that the given physical
quantity is represented by a vector, or that it is a vectorial quantity.
In future, vectorial quantities will often simply be called vectors, i.e.,
they will be identified with the vectors, representing them.
The law by which the components of vectors change during transition
from one coordinate system to another will now be recalled. The notation,
used in the main part of this book, will be somewhat modified. In fact,
the cosines of the angles between the old and new axes will now be
denoted by
X1 %2

%; % "12 "13
(A)
篇 ‘22 %23
%3 “31 “32 %33
The relations between the new components g, 虞，已；of a vector P and
its old components 蔡，&， 呢 may then be written
3 3
短 =e 源京， 亮 品. (LL1)

The following well known relations hold between the elements of

684 APPENDIX 1

table
3 3
2 = ^km> £ m - (1.1.2)
W=1 t=i
where
1for B = m,
{ 0 for k 手 m.
Consider now the two vectors
->
N = 仅i，能，口3)
and
f = 图,a, «•
Their scalar product is given by
今 3
Z ・P = + ^2^2 + 沆
3已3 = E〃匕.
£=1
By definition,
A.P — AP cos (4, P),
where A , P denote the lengths of the vectors A, P ; it is thus seen that
the scalar product does not depend on the choice of coordinate axes,
i.e., that
-J- 02a2 十 口3已3= % + B2E2 + 他已3。 (1.1.3)
The reader will easily verify this formula directly from (1.1.1) and (1.1.2)
Conversely, it will now be shown that, if 旬，a2, 他 are three scalars
which are related to the coordinate axes in such a manner that the
linear form
3
F = 口iEi +〃2&2+ 他 &3 = i X= l (1.1 .4)

where 疏，&，蕊 are the components of an arbitrary vector, is invariant

for the transition from one coordinate system to another, then the
triad of numbers Qi，a2> represents a vectorial quantity (i.e., a vector) .
In order to prove this statement, it is sufficient to verify that the quan¬
tities %,做，％ transform for the passage from one coordinate system to
another by the same law (1.1.1) as the components of a vector. In fact,
one has by supposition
S 咽 = S 数晶；
z =1 左 =1
 APPENDIX 1 685

substituting on the right-hand side for 短 the expression (1.1.1), one

obtains
3 3 3 3 3
£ Q W £ “f 思 = £ 亮 £ likak.
芯 = =1
6=1 检 匕=1 6 =1 6 =1

Since this equality must hold true for any values of 孩，E；, the co¬
efficients of 已；must be equal ; hence
, 3
仰 = £ Zf；/? 乃, (Ll.l)
a=i

and this formula agrees with the second formula of (1.1.1), if a is replaced
by 己. The proposition is thus proved. Therefore :
// the linear form 3
z 。岛
€=1

is invariant for coordinate transformation and 聂 are the components of

。笈 arbitrary vector, then 危 are likewise- the components o/ a vector.

2. Generalizing the concept of a vector, based solely on the above

stated property, one arrives by a natural process at the concept of a tensor.
In fact, instead of the linear form (1.1.4), consider the bilinear form
3 3 3
F =p £ 能孟加 = £ a仃旗w = 质工律1 + 旬2a谓2 + %3已r3 +
。2122刀1十 22E242+〃23之2刀3+
《 (L2.1)
%也3% + 他2已3刀2+ 他3之3刀3
which depends linearly on the components of two vectors

P = 图,纵 蕊)
and
0 = (%，刀 ％)•
2,

It will now be postulated that the coefficients a” of this form vary for
transformation of coordinates in such a way that the form F remains
invariant. Under this condition it will be said that the set of quantities
口必，depending on the two indices i, represents a tensor 0/ second order
(since there are two indices) ; 仰夕 are called the components of this tensor
(with respect to a given system of axes). This tensor will be denoted by
the symbol %).
On the basis of this definition, the transformation law of the tensor
components is easily found. Let 耳芯， be the components of the tensor
686 APPENDIX 1
-> ->
and of the vectors P, Q in the new coordinate system. By definition,
3 , , , 3

M= 1 k^m — 1
Substituting on the right-hand side the expressions
3 3
鼠 = 2 2或短，刀憎 = W

one finds
3 3 3
'尸'
'= * 匕i%'
v1 产' V-* t 7
2f * hkhmakm>
名n=1 k^m—l
whence, comparing the coefficients of the products 亮**
3
(1.2.2)

This is the required transformation formula.

A second order tensor is called symmetric, if 探 订 — aH. It is easily
seen from (1.2.2) that this property of symmetry is retained during
coordinate transformations.
In the case of a symmetrical tensor, one may use for its definition,
instead of the bilinear form (1.2.1), the quadratic form 2C(&i，凰, 己3)
which is obtained from F by putting & = 刀加 In this way one obtains
the definition, given in § 5 of the main part of this book. The transfor¬
mation formulae for the components of the stress tensor, given in § 5,
coincide with the formulae (1.2.2), if the last are rewritten in the notation
of that section.
The simplest symmetrical tensor is the tensor (3O) , defined by
for i —
JI 01 for i
(1.2.3)
丰

It is easily seen that is a tensor, since

3
上 &自力 = El% + 已2小 已3%
*7 =1

is obviously invariant (as it is the scalar product of the vectors P and .

The tensor (弟) is called the unit tensor.
A tensor is called anti-symmetrical, if 沈沙 = — q*. Since, in particular,
one must then have 包？ = —
aiit in an anti-symmetrical tensor
设口 — 他2 = 鹿33 = 0- Thus, an anti-symmetrical tensor is characterized
 APPENDIX 1 687

by three quantities %,%,自，such that

% = 032 = 一〃23，
02 = %3 = 一 031， ％ = ^21 ~ 一 的2，

It is easily seen from (1.2.2) that the property of anti-symmetry is

retained during transformation of coordinates.
3. Two tensors 仅仃) and (%) are said to be equal, if 痣灯 = 姐. The
tensor (c^ whose components are equal to the sums of the components
of two given tensors
c?? = 〃仃 十 姐
is called the sum of the tensors and (姐). It follows from
3 3 3
Z 0d之浦j 2 口仃已消 j + E 方仃即小
£J = 1 i>/= 1
that G?) is a tensor. Since the terms on the right-hand side are invariant,
the left-hand side is also invariant, and this proves the tensorial character
of the set of quantities c仃. The difference of two tensors may be defined
in an analogous manner.
If («“) is a tensor, the quantities 区寺 = 。藜 likewise determine some
tensor (礴)；this result follows likewise directly from the definition of
a tensor.
Every tensor (%) may (in a unique manner) be decomposed into
the sum of a symmetric tensor (e仃) and of an anti-symmetric tensor
(%?). In fact, let = e仃 + 仑夕. Interchanging the indices i, j and noting
that, by supposition, % = ejif 少" =
—
夕订， one finds a* =
In combination with the preceding equation, one finds
— 勿沙

e方 = 妥(d书 能) 少灯 — *(如 一 名) (131)

It is readily verified that the tensors (%，and (力订) satisfy the imposed
conditions.
Several examples of tensors will now be presented.
Let (%) and 仇) be two vectors. Write c 仃 = a^bj. The set of the quanti¬
ties (c 订) is a tensor. In fact, let (&), (刀, be two arbitrary vectors. One
has
3 3 3 3
7
2 a力总
工 c 茄洲2- = 口
产 工
油8— 2 的旗 2
5 5 产 产 工

i,j=l m =1 t=l 7=1

The right-hand side is invariant (as product of two invariant quantities).

Hence also the left-hand side is invariant, and this proves the assertion.
It is known that (嚼，where 唠 = = 包 is likewise a tensor.
688 APPENDIX 1

Hence, if one writes

力仃 ― 瑙 一 = %& — aibj, (132)
then (夕仃) is also a tensor which is obviously anti-symmetrical.
This tensor is called the vector product o/ the two given vectors. In vector
analysis, the vector product is considered as a vector, and not as a tensor.
In order to elucidate this, consider the following. Let
= 32 = 一 的除
*

力 力 —
句％ 《 (1.3.3)
It will be investigated whether the set of quantities (自，％,少3) is a
vector. For this purpose the criterion, formulated earlier, will be applied,
i.e., an arbitrary vector (短，^2, 蕊) will be introduced and it will be verified
whether
力晶 + 已2 十
is invariant. Obviously, one has
获 短
名El + 42a2 十 少 已3 说 (1.3.4)
外

However, it is known from analytic geometry that this determinant

represents the volume of the parallelepiped, constructed on the vectors
(£), (&),(&)；the sign of this volume depends on the orientation of the
coordinate axes: the volume changes sign as one changes from a left-
handed system of axes to a right-handed one. The sign will not vary
otherwise. Thus, the expression (1.3.4) will be invariant, if one uses
consistently a right-handed or a left-handed system, and in that case the
quantities %,/>2, 夕3 may be considered as representing a vector which
does not depend on the choice of coordinate axes.
It is easily shown that for transition from right-handed to left-handed systems
or conversely the vector (仇，％， defined by (1.3.2) and (1.3.3), inverts its
direction.
Finally, it will be shown that, when the above-mentioned restriction
of the choice of coordinate axes is imposed, every anti-symmetrical
second order tensor may be represented as a vector (having all the time
in mind three-dimensional space, since otherwise this assertion is not
true). In fact, let (力仃) be any anti-symmetric second order tensor.
 appendix 1 689

Construct the sum

3
e = —（0 + 少2c2 + 力已3），
3

where （虞），（勿） are two arbitrary vectors and where

名 = 032 = 一 少23， % = 413 = 一 031， 03 = 一 012，
已2刀3― 已3刀2，
乙2= 23刀1 - E1713, 已3= ^1^2 ― 已2刀1。 （1.3.5）
However, by the statements above,（%,电，Q）is a vector. On the other
hand, the left-hand side of （a） is invariant. Hence, also the right-hand
side is invariant and（Q,J， O is an arbitrary vector. This means that
（%,力2, %）is a vector, and the proposition is proved.
4. The concept of a tensor of any order n may be introduced in an
analogous manner. For this purpose it is sufficient to consider, instead of
a bi-linear form, an 切-linear form depending linearly on the components
of n arbitrary vectors.
For example, the set of coefficients aijk of the tri-linear form
3 3 3
S S S 苑怎电・Q,
i~ 1 / =1 1

where 烈，叮力 已 论 are the components of three arbitrary vectors, determines

the third order tensor（倒汴） with components aijk. In the same manner one
may define a tensor of order n. From this point of view, a vector must be
interpreted as a first order tensor, defined by the help of the linear form
3
s 说.
d=1

5. Consider again the second order tensor（0抄. Let（&）be some vector

and construct the expression
3
鸳=£应息. （151）
It is easily shown that （劈）—（岩，竟，匕力 is a vector. In fact, let 仇）
be an arbitrary vector. Then
3 3
£ 笃口 = S
i= l =l
is invariant, because the right-hand side is invariant on the basis of the
definition of a tensor.
It is obvious that also conversely, if 笑，磅，桀，defined by （1.5.1）,
690 Appendix 1

where 匕】， ^2> W3 are the components of an arbitrary vector, are the com¬
ponents of a vector, then 河 are the components of a tensor.
Thus the relation (1.5.1) relates to every vector (凰) a completely
defined vector (窈). For this reason the vector (鸳) is called the linear
vector junction of the vector (旗)，determined by the tensor (%，.
An example of such a vector function has been encountered in the
main part of this book. In fact, the relations (3.2) show that the stress
vector (Xn, Yn, Z^, acting on the plane with normal n, is a linear vector
function of the vector n, determined by the stress tensor. In this case, n
denotes a vector of unit length which has the direction of the normal n.
The case where the tensor (a 订) is symmetrical, i.e., where 口订 = ajit
is of particular interest. It will now be studied in detail. For this purpose
introduce the quadratic form

2Q(E1, 基 纵) = S 能e却 =
= ^11^1 十 做2程 + %3^3 + 2a23E2E3 十 203居3之1 + 2am已也?。 (1.5.2)
In this case one may rewrite (1.5.1)
HQ

The following important proposition will now be proved : By a suitable

choice of new coordinate axes (i.e., rectilinear, orthogonal) Ox[, 0%；, 0%；,
every quadratic form 2Q may be reduced to its canonical form
2Q = 入篇 2 + 入凰2 + 酬， (1.5.4)
where 卜 入 入？， 入3 口* constants (根乃纱6 it has been assumed that 仇订 are
real) . This proposition is equivalent to the following one. By a suitable
choice of coordinate axes, it may be ensured that the new components
〃春 of any symmetrical tensor (能 j, having different indices, vanish, i.e., that

(while the remaining, i.e., the "diagonal", components

'11 = ，
'22 ，
入1 = 入2 ~ 入3

will, in general, be different from zero).

If the form 2Q has the stated canonical form, the relations (1.5.3) in
the new coordinate system will reduce to the following :
 APPENDIX 1 691

These relations show that, if the vector （募） is directed along one of the
new coordinate axes, the corresponding vector （匕；*） will be parallel to it.
For example, the vector（亮），parallel to the axis 0%； has the components
E；, 0, 0 for 匕；# 0. Its corresponding vector （葛*） has the components
Hence, in order to reduce the form 2Q to the required type, one has
first of all to find the directions with the above stated property. Thus
there arises the following problem concerning the relations （1.5.1）: To
what direction of the finite vector （膜） corresponds a vector （^*） with
the same direction ? In order that the vectors（&）and（臂）will be parallel,
it is known to be necessary and sufficient that

where 入 is some parameter. Introducing here the expressions （1.5.1）

for 鸳，one obtains the system of equations

311 1

入）El + 式12E2 + ^13^3 =

口 21E1 十（022 1

入）之2+ 血23E3 = 0’ （1.5.6）

女3国 他2已2 +（N33 一 入）已 3 —

This system of linear homogeneous equations in 己力 &， E3 admits

non-zero solutions if, and only if, its determinant is equal to zero, i.e.,

^11 -- 入 龙
12 设
13

&2 -- 入 “23 ~ 0. （1.5.7）

口32 033 1
入

This equation is a third order polynomial in 入. It will be shown below

that all its roots are real. At the moment it will only be noted that this
equation has at least one real root, since it is a polynomial of odd degree.
This root will be denoted by 入3.
If one substitutes in （1.5.6） for 入 the value 入3, this system will give
a solution for which not all 却 are simultaneously zero. Let 羽，驾，羽 be
one such solution. The vector determines such a direction that for
any vector （£）, parallel to it, its corresponding vector （软） will likewise
be in the same direction.
Every such direction is called a princi-pal direction, corresponding to the
tensor
692 APPENDIX 1

The new system of axes Ox^, 。瑞 will now be chosen such that the
axis 0%；has the principal direction, found above. The two other axes
(perpendicular to this direction and between themselves) remain for
the time being arbitrary.
The components of tensors and vectors in the new system will be
denoted by the same symbols as before, but with two accents. In the new
system the equations (1.5.6), which may be written
r*
看 一一
了
a.
人

take then the form

where
3

thus, written explicitly, one finds

_
When 入 = 入3, these equations must have the solution (0, 0, 以) for
以 W 0. Hence
口13 = c
*
— *
5 °23 = 门 — )人3，
5 “33 =

so that the quadratic form 2Q becomes in the new system

2Q = 螳芯2 + 2a^ + g芯2 十 入芯2. (1.5.8)
In order to reduce 2Q to the required type, it is sufficient to rotate
the axes 0%；, 0%；in their plane, leaving 0%；unchanged, so that the
term with the product in (1.5.8) vanishes. This may always be
done. In fact, let the new axes be 0%；, 0%；, 0%；and let 0%；make an
angle a with 0%；. Then
C = E；cos a — 苓 sin a, 琰 = 孩 sin a + cos a, = &
Substituting these expressions in (1.5.8.), one obtains
2Q

where, in particular,
〃
= %芯2 + 2 2a 篙 + 呃召 十 入后2, (159)

%2 = — — 口;2) sin 00 cos a 十 a% (cos2 a — sin2 a) =

== — *('ii —〃22) sin 2a -p 旬2 cos 2a.
 APPENDIX 1 693

Hence a* = 。for

tan 2a = ———"〃
旬11
…方―
22
. (1.5.10)

If a0 is an angle which satisfies this condition, then

TV
% +e
as well as all angles

ao H—
will satisfy it, where k is an integer. Thus, two mutually perpendicular
directions have been found which satisfy the required conditions (both
of these being perpendicular to 0%；, which coincides with ; only
for — 。
0 and a；】 = 仇短 will there be an infinite number of such direc¬
tions, in which case 九 = 砧「 all values of a. Choosing one of the axes
for 0x^ (and its perpendicular for 。琐，the form 2Q is reduced to the
required type (1.5.4), where 入力 入2, 入3 are real numbers. One has thus not
only proved the possibility of the stated reduction, but also deduced an
effective method for its execution and for the determination of the
directions of the corresponding new axes.
It is known that 入3 is one of the roots of (1.5.7). It will now be shown
that 入1，入2 are the two other roots of the same equation. For this purpose
it will first be noted that the determinant
组2 %3
A= 倪
21 仇
22 023 (1.5.11)
龙
31 032 仅
33

is invariant, i.e., that it does not change for transformation of coordinates

(this determinant is called the discriminant of the quadratic form 2Q).
In fact, for transition to new axes 0%；,0%；, 0% j this determinant becomes
r f r
^11 ^12 "13
，
/ = 铀1%2 女23
f /

fr f
仇
31 〃32口33
where, by (1.2.2), 3 3 3
£ 泳沙 加加m =
a=1次==1 不 =1
694 APPENDIX 1

with
3

On the basis of the well known theorem on multiplication of deter¬

minants, one has
如 如 &3 "11 "12 "13
Dq — A. 方 21 622 方23 , where A — 221 %22 %23
台33 231 %32 %33
On the basis of the same theorem
如 612 如 诙1 旬2 "13
^21 %2 623 ——- zx * 口 21口22 口23 = A *Dq・
姐 %2 %33 倪
31 口32 口33

；
Hence D = 公?。。. However, on the basis of well known properties of
the direction cosines 必， = ± 1 , whence it follows that D = Dq, ；
as was to be proved.
Next consider the tensor with the components Aik = aik — 入t 注，
where 入 is an arbitrary number and（BQ the unit tensor. The deter¬
minant of the components of the tensor（4 加
旬1 一 入 旬2 旬3

2= 说
21 222 入 423
^31 凄
32 @33 一 入

will, on the basis of the earlier statements, not depend on the choice
of coordinates, i.e., is invariant.
Now choose new axes 0x[, 0%；, Ox^ in such a way that the new com¬
ponents of the tensor （口注）， which have different indices, are equal
to zero, so that the quadratic form 2Q is of the type
入1已 / 入晶2 + ^3^32-
The determinant Dx for this new system of axes will be
入i
— 入 0 0
0 入2 一入 0=（入1 一 入）（入2 - ）（入3
入 入）.
0 0 入3
- 1
入
 appendix 1 695

Hence one has the identity

all 一 入
i

为2 以
13

。入 = 的2
— 入 &3 —（% 一 入）（入2 - 入）（入3 1

入）,
〃3I «32 数
33 —入
and it is seen that the real numbers %,入2, 入3 are the roots of the equation
少入 = 0.
Thus, in passing, the important theorem of algebra has been proved
by which all the roots of (1.5.7), which is called the characteristic equation,
are real (under the essential supposition that «仃 are real and, in addition,

Now the linear vector function, defined by (1.5.1), will be considered,

retaining the supposition = ah. It has been seen that one may always
find at least one triad of mutually perpendicular principal directions and
that, if the coordinate axes are given these directions, the form 2Q be¬
comes
入1针
+ 入Q + 入3肾,
while the relations (1.5.1) have the form
岩=入1获， 食 =京2， 凌 二入晶 (L5.12)
(where now the accents have been omitted) .
Next, the question will be discussed whether there are any principal
directions other than the three, found above. If (蔡，己?，蕊) is a vector,
parallel to any principal direction, then, by definition, the vector (聋)
must be parallel to the vector (£), i.e.,
金= 呜， 食=延2， 绫=花3
Substituting from (1.5.12), one finds
(入1一入)法=0, (入2 —入总 =0，(入3 -入)之3 =0，
whence it follows that 入 can only have one of the three values 入1，入?，入3
(otherwise one would have to have 蔡 = & = E3 = 0)・
First suppose that 入】， 入2, 入3 are all different. Substituting 入 = 加 in
(1.5.13), it is seen that these equations are only satisfied by the following
—
values : £ an arbitrary quantity, 己？= £ = 0- Thus the vector,
corresponding to 入 = 入】， is parallel to the axis ; this gives one of the
possible principal directions (which is already known) . In an analogous
696 APPENDIX 1

manner it is verified that the values 入 = 入孕 入 = 入3correspond to the

directions of the axes 0%2, 0x3.
Thus, if the three roots of (1.5.7) are different, there are only three
principal directions which are mutually perpendicular.
Now let 入i = % W 入3・ In that case one obtains again for 入 =
only one direction, namely the direction of 0x3. However, for 入 = 入】 = 入
the solution of (1.5.13) will be: 获 arbitrary, & arbitrary, & = 0. Thus
all directions perpendicular to the axis Ox3 (and only these directions)
will be principal directions, corresponding to this value of 入. One may
always select among these directions an infinite number of pairs of
mutually perpendicular directions (which will also be perpendicular to
the axis 0%3).
Finally, it is obvious that, if N = % = 入3, then the equations (1.5.13)
will be satisfied for 入 = 入】 = 入？ = 入3by any values of 蔡，^2, 蕊. In
other words, in this case any direction is a principal direction.
 Appendix 2

ON THE DETERMINATION OF FUNCTIONS FROM THEIR

PERFECT DIFFERENTIALS IN MULTIPLY CONNECTED
REGIONS

1. The case of two dimensions will be considered first. Let S denote

some region of the plane. Only such connected regions which are
bounded by one or several simple contours will be studied. Such regions
may also be infinite (infinite plane with holes) , but, for the present,
consideration will be restricted to finite regions.
A region is called connected, if any two points in it may be joined by a simple
line which does not leave the region. A contour is called simple, if it does not
intersect itself.

A region 5 is said to be simply connected, if any cut joining any two

points of its boundary disturbs its connectivity, i.e., dividesit into separate
regions.
A region is said to be
multiply connected, if cuts
linking points of the bounda¬
ry may be introduced with¬
out dividing it into indi¬
vidual parts.
It is readily seen that a
region, bounded by one
simple contour, is simply
connected. In contrast, a
region, bounded by several
simple contours, is multiply
connected. In fact, let the
boundary of a region con-
Fig. 64.
sist of the contours 工1，工2,• • •
Lmt Lm+1, the last of which contains all the others inside it (Fig. 64). If
the region is cut along any line connecting a point 旬 of Lr with a point
697
698 APPENDIX 2

、
b of the outer contour Lm+1, such a cut does not affect the connectivity of
the region.
If, in addition to 旬跖 similar cuts 他方2， …， a 袍琳 are introduced,
which do not intersect one another, the connectivity is still not disturbed ;
however, as is easily seen, any further cut will affect the connectivity of
the region. Thus, the m cuts aj)lt 痉鬼匕 make the region under
壮

consideration simply connected.

If m cuts are required to convert a given region into a simply connected
one, it will be said that the region is (加 + l)-ply connected or that its
connectivity is equal to (w + 1).
It is seen that in this way the connectivity of a region is equal to
the number of contours bounding it. For example, the region between
two concentric circles is doubly connected.
A simply connected region differs from a doubly connected one in that
it has the following property. If one draws inside the simply connected
region a simple contour, the region inside this contour belongs entirely
to S ; this contour may be shrunk into a point by means of continuous
deformations during which it remains always in S.
In the case of multiply connected regions, there exist contours which
do not have this property. For example, in Fig. 64, L； is one such contour ;
it is impossible to contract it into one point without cutting it or without
the contour leaving S.
2. Let there be given the differential
y)dx 十 0(%, y)dy, (2.2.1)
where y) and Q(x, y) are single-valued and continuous functions
with continuous first order derivatives in some region S. The following
question will be asked: What conditions must be satisfied by the functions
P, Q, in order that (2.2.1) should be a perfect differential of some single-
valued function F(%, y), i.e., in order that there should exist a function
F(力, y) such that
dF = Pdx + Qdy (2.2.2)
or, what amounts to the same thing, that
dF 3F
f = foy一 =0(4,力？ (222)
讯

Although this problem is studied in all, even elementary textbooks

on calculus, it has nevertheless been considered necessary to dwell on
 APPENDIX 2 699

it here, in order to draw the reader's attention to certain circumstances

which are very essential for the purpose of this book.
First, suppose that the region S is simply connected. Inside S select
some fixed point M0(x0, %) and connect it with the variable point M(x, y)
by an arbitrary line M0M which does not leave S. If the function F(招 y) ,
satisfying (2.2.2), exists, one obtains by integrating both sides along
MM
F(%, y)
/
= (Pdx + Qdy) + C, (2.2.3)

where C = F(xQ, %) is a constant.

By supposition, F{x, y) is a single-valued function of %,y; hence its
value at the point M(%, y) must only depend on the position of M and not
on the path of integration M0M. Thus, if F(xt y) exists, the line integral

/(Pd% + Qdy)
cannot depend on the path of integration (which lies, of course, in the region
S) . This condition may be formulated as follows : The integral

(Pdx 十 Qdy)
J
L

taken along any contour L (entirely inside S) must be equal to zero. In

fact, by linking any two points A and B of some
contour L with the points Mo and M respect¬
ively (Fig. 65), one has, by supposition,

= 0;
MqADBM

however, since the integrals along M^A and BM

in both terms are equal, one must have

ADB AD'B L

as was to be proved. Fig. 65.

700 Appendix 2

On the basis of Green's formula,

(Pdx + Qdy) = (2.2.4)

L a

where b denotes the region, bounded by the contour 工. If follows from the
above that the integral on the right-hand side must vanish for every part
o of the region S. Thus the integrand must be zero at each point of S, i.e.,
one must have
丝 2P (2.2.5)
dx
throughout S. As has just been seen, (2.2.5) is the necessary condition for
the existence of the function 尸(招 y). It may also be proved that it is
sufficient.
In fact, if this condition is satisfied, the line integral

/ + 0沟)
does not depend on the path of integration, but only on the starting and
end points of this path. This follows directly from the above : If A and B
are any points of S, and ADB, AD'B are any two paths connecting these
two points, then

ADB ND'B
because (Fig. 65).

ADB AD'B L
and the last integral is zero by (2.2.4) and (2.2.5). For this purpose it
has been assumed that the lines ADB and AD'B do not intersect each
other, so that they form a simple contour. However, it is readily verified
that this condition is not essential ; if these lines intersect at one or several
points, the difference of the integrals along these paths may be reduced
to the sum of integrals along two or several contours.
In particular, the integral on the right-hand side of (2.2.3) represents,
for a fixed point M0(x0, a single-valued function of % and y, and hence
(2.2.3) determines the single-valued function F(x, y), provided C is
 APPENDIX 2 701

given an arbitrary (constant) value. Further, it is readily verified that

(2.2.2') actually holds true.
In fact, extending the path of integration in (2.2.3) by the straight
segment MM' , parallel to Ox, to the point Mr(x + y), one obviously
finds
r + Ah
八

Fix A%, y) = F{x, y) +

/ P(x, y)dx,
X

whence

8F
= hm + A%) — Fix) = lim 1— f/ P(x, y)dx — P(x, y),
0% 0 A% J
Ax~> 0
法

i.e., one obtains the first relation (2.2.2'); the second relation may
be proved in the same manner.
It has thus been seen that the condition (2.2.5) is necessary and suf¬
ficient for the existence of the single-valued function F(%, y), satisfying
the conditions (2.2.2) or (2.2.2'). If these conditions are satisfied, the
function 尸(％, y) will be determined by (2.2.3), apart from the constant C
which is quite arbitrary.
Hitherto, it has been assumed that the region S is simply connected.
Next consider what supplementary conditions are required in the case of
multiply connected regions.
The condition (2.2.5) is also in this case necessary ; its deduction differs
in no way from that for the case of a simply connected region. One has
only to select for the application of (2.2.4) such a contour L that the region
g, bounded by it, lies entirely in S (this condition will be automatically
satisfied in the case of a simply connected region). The question of the
sufficiency of this condition will now be investigated. It will be proved
that in the present case this condition secures the existence of the function
F(x, y), defined by (2.2.3), but that this function will, in general, be
multi-valued.
A beginning will be made with the following remark. Let the region S
be cut along
旬 , a 憎方馆，

as indicated in the preceding section ; one thus obtains a simply connected

region which will be denoted by S*.
It should be understood that two edges of the cut region adjoin to
702 Appendix 2

each line so that every point of this line must be considered as

two points belonging to the one or the other of the corresponding edges.
Hence a distinction must be made between the two edges at each cut which
will be denoted by (+) and (—).
Since the region S* is simply connected, the function F(x, y), defined
by (2.2.3) where the path of integration must not go outside S*, i.e.,
must not intersect a cut, will be single-valued in S* on the basis of the
earlier results.
However, this does not mean that at points, lying on different edges
of the same cut, the values of the function F will be identical (because
these points must be conceived as different points of 5*). For example,
select any point A of the cut 旬& and denote by F+, the values of
F at the points A+, A~ of the edges (+) and (— ) which coincide at the
geometrical point A. By (2.2.3),
F- =
/ {Pdx + Qdy) + C, 「+ = /(尸然 + 。沟) + C

where the first integral is taken along any line M0A~, lying in S* and
going from Mo to the point A, approaching it from the side (— ) ; the
second integral is taken along a path M0A+, likewise beginning from Mo,
but approaching A from the positive side (+) [Fig. 64, where the point
Mq and the paths of integration are not shown]. As path of integration
for the second integral one may take the path of integration MqA~
of the first integral supplementing it by the line L；which surrounds
the contour匕 once and leads from the edge (—) to the edge (十) without
leaving the cut region S*. Thus

F+= + / 十 C = F- + A，

where
+ 0勾)
刀1’

and 二；is a simple contour, going in S* from the edge (— ) to the edge
(+) of the cut a1b1 without intersecting another cut (Fig. 64). This
contour intersects the cut 旬如 crossing from the side (+) to the side (—).
、
It is readily seen that J does not depend on the choice of the contour
L；which is to surround the contour only once, going in S* from the
edge (— ) to the edge (+) of 口也.
 APPENDIX 2 703

In fact, let 工; be another such contour which intersects at some

point B. Consider the contour which does not leave S* and which consists
of the segment AB of the positive edge of the cut, of the path taken
in the negative direction [i.e., in 5* from the edge (+) to the edge (—)],
of the segment BA of the negative side of the cut and, finally, of the
contour L；. On has

J
八 = o,
where the integral is taken along the above-stated closed path in S*.
Further, since the integrals along AB and B4 cancel each other, one has

—
Lr
J (Pdx + Qdy) + j (Pdx 十 。勿) = 0,
Li
and this proves the assertion (the first integral has here been given a
minus sign, assuming 工; to denote the path in the positive direction).
Similarly, one obtains for any cut aM that
F+ = F一十
where

Jk = [(Pdz + Qdy) (2.2.6)

and L'k is any contour, surrounding Lk and only intersecting the one cut
a* in the direction from the side (+) to the side (— ).
The integral (2.2.6) may, in particular, be taken along the boundary
Lk itself, provided the functions P, Q are continuous up to the boundaries.
The nature of the function F (x, y), defined by (2.2.3), is easily seen, if one
considers it as defined in the uncut region, i.e., if one allows the path of
integration to intersect the cuts.
Let 尸o(%, y) denote the value given by (2.2.3) in the cut region, i.e.,
when the path of integration does not intersect a cut. Consider the
arbitrary path of integration MQM (Fig. 66) ; let it intersect the cuts
in, say, n points. Follow the path of integration from the point Mo to
the first intersection with one of the cuts akbk. On the subsequent part of
the path MQM, which lies between the first and second encounter with
a cut, select two consecutive points A and B and replace the segment
4gby the line AM^B which goes from A to Mo and returns to B without
704 Appendix 2

intersecting a cut. This procedure does not, of course, alter the value
of the integral (because the new segments belong to the cut region). The
original path from MQto M is thus replaced by the contour which
surrounds the contour once, and the
path MqBM, which now intersects the
cuts in only % —
1, and. not n points.
The integral taken along the contour
MqAMq is, by (2.2.6), equal to + Jk
—
or Jk, depending on whether its path
of integration intersects the cut akbk
from the edge (+) to the edge (— ) or
in the opposite direction. Thus, one
may omit from the (modified) path of
integration the closed part M0AM0
under the condition that a term 土/丸
be added to the final result.
Proceeding in the same manner, one may reduce the path of integration
to one which does not intersect any cuts. The integral, taken along this
path, must then be combined with the quantities 土 Jk\ each of these
terms must be added according to the number of times which the original
path of integration intersects the corresponding cut; the sign (十) will
apply, if the cut is crossed from the side (+) to the side (—), while the
sign (— ) must be taken in the opposite case.
Since the path of integration which does not intersect cuts gives the
value y), the final result may be written in the following form:
F(x,y) = y) + + %/2 十
-
… (2.2.7)
where nlt . . .,nm are integers which are positive or negative and which
are easily calculated, on the basis of the above results, as the number of
intersections of the path M0M with the cuts (where account must be
taken of the directions of intersection) . For example, in the case of
Fig. 66.,
F(x, 夕) = 瓦 y) 十/修 +
In order that the function F(%, y) will be single-valued, it is necessary
and sufficient that together with (2.2.5) the following condition be
satisfied:
/1 = = ••• =
The above results will likewise apply to the case, where the contour
 Appendix 2 705

上团+i is entirely at infinity, so that the region S becomes the infinite

plane with holes.
3. Quite analogous results will apply in the case of three dimensions.
In this case a distinction must also be made between simply and multiply
connected three-dimensional regions (bodies) . A region will be said to be
simply connected, if it has the property that every closed line inside it
may be shrunk into one point by means of continuous deformation
during which it does not leave the region (eg sphere, cube). Otherwise a
region will be multiply connected. As examples of multiply connected
regions, one may quote the torus (i.e., the body, obtained by rotating a
circle about an axis in its plane, but not intersecting it) or a cube with
one or more holes, drilled through it, etc.
The torus is doubly connected, because it may be made simply con¬
nected by the help of a single cut; however, in contrast to the case of
two dimensions, the cut will now not be a line, but a surface.
+
In general, a body will be (冽 l)-ply connected, if it may be made
simply connected by the help of m cuts. Attention is drawn to the fact
that a body, bounded by one closed surface, is not necessarily simply
connected (e.g. : torus) ; in contrast, a body may be bounded by several
closed surfaces and it may still be simply connected (e.g. the region
between two concentric spheres) .
Let there be given the differential expression
P(力, y, z)dx + 0(%, y, z)dy + R(x, y, (2.3.1)
where P, Q, R are single-valued and continuous functions which have
continuous first order derivatives in some simply connected region V.
As in the case of two dimensions, it may be shown that for the existence
of a single-valued function F(%, y, z), satisfying the condition
dF — Pdx + Qdy + (2.3.2)
it is neccesary and sufficient that

/(Pd% + gfy Rdz) =

十 0, (2.3.3)
L
where L is any contour in the region V. Under this condition, the function
F(%, y, z) will be determined by the formula

F(x, y, z) = + Qdy 十 Rdz) +C, (2.3.4)

706 APPENDIX 2

where C is an arbitrary constant and the integral is taken along any

path (in V) which links the fixed point with the variable point
y, z).
The condition (2.3.3) will now be transformed. For this purpose the
well known Stokes formula will be recalled
匆、
(Pdx —J COS 伊,切、+
L

) cos (%, z) } do, (2.3.5)

dx dy / J

where o is any (open) surface (inside V) with the boundary L and n is

the normal to b in a definite direction. Applying (2.3.5) to (2.3.3), one
finds

cos (肛 y) +
a

+(普 — 署)8s 例,z)} da = 0; (2.3.6)

this condition must be fulfilled for any surface a (in V). Selecting for a
any plane, normal to the axis Ox, one obtains, in particular,

/「(筝 —*)
o
匆办 =°‘

whence (in view of the arbitrariness of the plane a) one arrives at the
first of the following formulae :
2R 笛 8P ?R 绚 dP
'
(2.3.7)
dy dz bz a% a% Sy
(the remaining two being obtained by cyclic interchange of symbols).
Conversely, the conditions (2.3.7) are obviously sufficient for the
truth of (2.3.3) and, hence, for the existence of the single-valued function
F(力，y, z), determined by (2.3.4).
In the case of multiply connected regions, if (2.3.7) is satisfied, the
function F(x, y, z) determined by (2.3.4) may be multi-valued. In fact,
reasoning as in the case of two dimensions, one may establish the fol¬
lowing result. If one has introduced m cuts (partitions) which convert
the given (m 十 l)-ply connected region into a simply connected one and
 APPENDIX 2 707

if Fq{x, y, %) denotes the function, defined by (2.3.4) under the condition

that the path of integration does not intersect these partitions, then one
will have for an arbitrary path of integration
F(%, y, z) = y, z) 十 nJ、+ nj2 + ... + nmJm, (2.3.8)
where nlt . . .,nm are integers and Jz, J m are constants, cor¬
responding to integrals taken along closed paths. In fact,

Sk =
Lk
/ (Pdx + Qdy + Rdz), (2.3.9)

where L7c is the simple contour which intersects only the cut k from its
side (+) to the side (— ). The numbers nk are defined in the same way
as in the case of two dimensions.
In order that the function F will be single-valued, it is necessary and
sufficient that, in addition to (2.3.7), the following conditions be sa¬
tisfied :
/1 = /2 = •… = (2.3. 10)
 Appendix 3

DETERMINATION OF A FUNCTION OF A COMPLEX VARIABLE

FROM ITS REAL PART. INDEFINITE INTEGRALS OF
HOLOMORPHIC FUNCTIONS

1. Let
0(%,y) + 匆(%, y) = /(z) (3.1.1)
be a function of the complex variable z —
x iy which is holomorphic
in some region S of the z plane. Under these circumstances, the real
and imaginary parts 力 and q are known to be related by the Cauchy-
Riemann conditions
效 dq 坐
dx 效 '勿
―
-a%
(3.1.2)

Conversely, it is known from complex function theory that, if two

single-valued real functions 夕 and q which have continuous first order
derivatives are related by the equations (3.1.2), then 力十勾 is a holo¬
morphic function of the variable z in a given region.
It is known that holomorphic functions have derivatives of any order (and,
further, may be developed into Taylor series in the neighbourhood of any point).
Hence the functions 勿，q also possess this property.

The function q, related to a given function 勿 by the equations (3.1.2),

is said to be conjugate to 力.
Not every real function 力 can be the real part of a holomorphic function
of a complex variable. In fact, differentiating the equations (3. 1 .2) with
respect to % and y respectively and adding, one obtains
赞力 济力 、
行7" = 八夕 = ；
，
"FT + 0 (3.1.3)

hence the function 力 must be harmonic. In the same manner it may

be shown that the function q must be harmonic. In what follows, a
function will be understood to be harmonic, if it satisfies (in a given
708
 APPENDIX 3 709

region S) the equation (3. 1 .3) and if it has continuous second order
derivatives. In addition, it will be assumed that the function 夕 is single¬
valued.
It is easily shown that one may determine for any harmonic function
0 a function q, conjugate to it. In fact, by (3.1.2), one has for the de¬
termination of q
=：
劲 效

The condition for the existence of the function q in the present case
reduces to the following (cf. Appendix 2) :

一 = 讲’
谈
which is satisfied thanks to (3.1.3). Hence the function q will be de¬
termined, apart from an arbitrary constant, by the formula

夕(%，y)= /(一" 蒙的) 十 十 。’ (3.1.4)

where MQM is an arbitrary path which connects some (arbitrarily) fixed

point Mo with the variable point M(%,y) and which does not leave the
given region S
The formula (3.1.4) may be written in a somewhat simpler form.
Let t denote the tangent to the path of integration (in the direction Mo
to M) and n the normal towards the right of t (see Fig. 13, § 32), then
dx = ds.cos (勿 ％) = —〃$.cos (制，夕)， dy = ds.cos (力 y) = ds.cos (咒，
where ds is the arc element of the path of integration ; hence
效 效 [效 效 ] dp
— -—dx +
纱
—
^ —
= 1I cos (% y)
纱
+—
ox
cos (%, x)\ds = -- ds
J

and (3. 1 .4) may be rewritten

夕(%, y)= —— ds + C. (3.14)
J an
This formula could have been written down straight away by noting that always
dq dp
'
dn
This relation follows directly from the Cauchy-Riemann conditions.
710 APPENDIX 3

When S is simply connected, the function q determined by (3.1.4)

or (3.1.4') will be single-valued and, on the basis of the above statements,
the function
/(z) = 力 十 勾
will be holomorphic in S ; will be determined for g given 力 apart from a
purely imaginary arbitrary constant Ci.
In the case of a multiply connected region, the function /(z) = 力十 勾,
where q is determined by (3.1.4) or (3.14), will be holomorphic in
every simply connected subregion of S (and, in particular, in the cut
region S* ; see Appendix 2) . However, if the only restriction on the path
of integration is that it must remain in S, the function /(z) may turn
out to be multi-valued. In fact, for a circuit along a contour surrounding
one of the contours Lk (using the same notation as in Appendix 2), the
function q will increase by some constant Bk, and hence /(z) undergoes
the purely imaginary increase iBk.
The constants Bk are determined by the formulae

the integrals may be taken along the actual lines Lk, provided the partial
derivatives of 力 are continuous up to the boundary.
In order that the function /(z) will be single-valued in the multiply
connected region S, it is necessary and sufficient that all the constants
Bk are equal to zero.
2. In connection with the above, one remark will be made regarding
the indefinite integral of a function, holomorphic in some region S. By
the indefinite integral

will be understood the function

尸(z) = / f(z)dz + const., (3.2.1)

where the integral is taken along an arbitrary path which does not
leave S and connects the arbitrarily fixed point % with the variable point
z and "const." is an arbitrary (in general, complex) constant.
 APPENDIX 3 71 1

If S is a simply connected region, F(z) will be a single-valued function.

This follows from the fact that, by Cauchy's theorem, the integral

//(z) 次

if taken around a contour, is equal to zero, so that

z

does not depend on the path of integration (cf. the analogous reasoning
in Appendix 2).
However, if the region S is multiply connected (assuming that it
has the form indicated at the beginning of Appendix 2), the function
F(z) may turn out to be multivalued ; in fact, for a circuit around a
contour which surrounds Lk once (see Appendix 2), it undergoes
an increase
afc + 访左 = f{z)dz. (3.2.2)
工'a

In general, the integral on the right-hand side of (3.2.2) will be different

from zero, because the region contained inside L'k does not entirely
belong to £ The quantity 血 勿阮 does not depend on the choice of
the contour E；, except that it is to surround Lk once, that it may not
intersect any other cut beside a* and that it must be described in a
definite direction. This may be proved by the same method as the
analogous result for the function F(%,y), proved in Appendix 2. Reasoning
as in Appendix 2, it is easily established that the function F{z}, defined
by (3.2.1), may be represented in the form
F(z) = Fq(z) 十 %1(/ + 洛) + ... 十 时3播 十 记伏)， (3.2.3)
where FQ is a single-valued function, defined in the cut region S*, and
笈2，•
… are integers, defined as in Appendix 2.
A
 AUTHORS INDEX AND REFERENCES
The following journals contain the listed references and are quoted by the
relevant numbers in fat print. Italic numbers refer to pages of this book.

JOURNALS
1. Doklady, A. N. S.S.S.R. JXoKJia^Bi, A. H. C.C.C.P. (Comptes rendus de
l'acad6mie des sciences de 1'U.R.SS)
2. Rendic. del Circolo Mat. di Palermo.
3. Annali di Mat. pura ed applicata.
4. Rendic. d. R. Accad. dei Lincei 5th. ser., 6th ser.
5. Transactions of the American Math. Soc.
6. Soobshcheniya Charkovsk. Mat. obshch. Coo(5meHnH XapbKOBCK. MaT. o6m.
7. Soobshcheniya A. N. Gruz. S.S.R. CooSineHHH, A. H., Tpy3. C.C.P.
8. Atti Accad. di Torino.
9. Atti del R. 1st. Veneto.
10. Prikl. Mat. i Meeh. npnKJi. MaT. n Mex.
11. Phil. Trans. Roy. Soc. London, ser. A.
12. Quarterly Journal (Oxford series).
13. Zeitschr. f. Angew. Math. u. Meeh.
14. Comptes Rendus, Paris.
15. J.R.F.KH.O., cz Fiz. 次. P. .X.O., q. 中H3.
16. Arkiv for matematik, astronomi och fysik.
17. Izv. A. N. S.S.S.R. H3B. A. H. C.C.C.P.
18. Mess, of Math.
19. Bull, of the Calcutta Math. Soc.
20. Trudy Tbilissk. Mat. Inst. Tpy*>i T611nHeck. MaT. Hhct.
21. Ingenieur Archiv.
22. Soobshcheniya Gruz. Filiala A. N. S.S.S.R.
Coo^meHHH Tpy3. minajia A. H. C.C.C.P.
23. Bull, de la Soc. Math, de France.
24. M6m. des Savants Etrangers, Paris.
25. Ann. de L'Ecole Norm. Sup. 3-me s6r.
26. Math. Ann.
27. Trans. Inst, of Naval Architects, London.
28. Zeitschr. f. Math. u. Phys.
29. Izv. Electrotechn. Inst. Petrograd.
03B. JIeTpozpagCKOzo SjieKTpcrrexH. Hhct.
30. Ann. de la Fac. des Sci. de Toulouse, 2-me s6r.
31. Il Nuovo Cimento, ser. V.
32. Acta Math.

713
714 AUTHORS INDEX AND REFERENCES

33. Math. Zeitschr.

34. Izv. Leningr. Politechn. Inst. IIsb. JlEHHHrp.IIojiUTexH. Uhct.
35. Proc. Cambridge Phil. Soc.
36. Proc, of the London Mat. Soc.
37. Trudy Seismologiczesk. Inst. A. H. SSSR.
Tpyai>i CeficMOJiornqecK. Hhct. A. H. C.C.C.P.
38. Matematiczecki Sbornik. MaTeMaTnqecKnii CSopHUK.
39. Izv. Tifl. Univ. Hsb. Th Yhhb.
40. Izv. Tifl. Politechn. Inst. 03B. Th^ji. nojiHTexH. Hhct.
41. Trudy TZAGI. Tpy^bi 口AFU. Public, of the Central Aero-hydrodynamic
Institute, Moscow.
42. Monatshefte fiir Math. u. Phys.
43. Phil. Mag. ser. 7
44. Journ. de Math. (Liouville), 2me s6r.
45. Trudy Dnepropetrovsk. I njenerno-stroit . Inst.
Tpyabi AnenponeTpoBCK. Hhskohopho-ctpoht. Hhct.
46. Naukovi pratzi Inst. Bydivelju. Mekhaniki AN URSR.
HayKOBi npan;i Ihct. SyaiBejibH. MexaHiKn A. H. y. P.C.P.
47. Zbirnik pratzj Inst. Mat. 36ipHiK npaub Ihct. Man
48- Dopovidi AN URSR, viddil tekhn. nauk.
XConOBiffi, A. H., Y. P.C.P. BUwiJi Texn. Hayn.
49. Biull. n.-techn. Kom. UVMSRKKA.
Eiojiji, H.-TexH. Kom. y.B.M.C.P.K.A.
50. Proc. Roy. Soc. Ser. A.
51. Trudy Gruzinskogo politekhnich. inst.
Tpy^bi rpysuHCKoro UoJiiiTEXHiiq. JIhct.
52. Inzhenernyi Sbornik
hHJKEHEpHB谊 C'OpHHK.
53. Quart. Journal of Meeh, and Applied Math.

Abramov, V. M. A6paMOB, B. M. [1] The problem of the contact of an elastic

half-plane with an absolutely rigid base for the determination of the frictional

Airy, G. B. zo6.
—
force. 1. Vol. XVII, no. 4. (1937) pp. 173 178. 47^, 487, 488.

Almansi, E. [1] Sulle ricerca delle funzioni poli-armoniche in un' area piana sem-
plicemente connessa per date condizioni al contorno. 2. Vol. XIII (1899) pp.
225—262. 367.
[2] Sull' integrazione dell' equazione differenziale = 0. 3. Ser. Ill,
Vol. 2. (1899) pp. 1—51. 235.
[3] Un teorema sulle deformazioni elastiche dei solidi isotropi. 4. 5th ser.
Vol. XVI (1907) pp. 865—868. 工37・
Bartels, R. C. F. [1] Torsion of hollow cylinders. 5. Vol. 53. no. 1 (1943) pp.
—
1 13. 63。.
Begiashvili, A. I. EeruaniBHJiii, A. H. [1] Solution of the problem of the pressure
of a system of rigid profiles on the straight boundary of an elastic half-plane.
1. Vol. XXVII, no. 9. (1940) pp. 914—916. 49^-
 AUTHORS INDEX AND REFERENCES 715

Beltrami, E. 77.
Bernstein, S. N. BepHiiiTeHH, C. H. [1] On the absolute convergence of trigono¬

PR. 200—201. 251.

—
metric series. 6. 2nd ser. Vol. XIV, no. 3 (1914) pp. 139 144. no. 5. (1915)

Bitzadze, A. V. Snn;aA3e, A. B. [1] On local strains in elastic bodies in compres¬

—
sion. 7. Vol. V, no. 8 (1944) pp. 761 770. 492, ^n, 514
Boggio, T. [1] SuH'equilibrio delle membrane elastiche piane. 8. Vol. XXXV
(1900) pp. 219—239. 367.
[2] SulF equilibrio delle membrane elastiche piane. 9. Vol. LXI2 (1901/2)
pp. 619—636. 367.
[3] Integrazione delVequazione
LX? (1900/1) pp. 591—609. 257.
— 0 in un'area ellittica. 9. Vol.

[4] Sulle funzioni di variable complessa in un'area circolare. 8. Vol. 47

(1911/12) pp. 22—27. 146.
Born, M. [1] Dynamik der Kristalgitter, Leipzig & Berlin, 1915. (Russian trans¬
lation by Frenkel, Leningrad-Moscow, 1938). 55.
Boussinesq, J. 397, 442.
Bukharinov, G. N. ByxapuHOB, r. H. [1] Solution of the plane problems of the
theory of elasticity for regions, bounded by curvilinear contours of particular
shape. Section of ,, Experimental methods for the determination of stresses
etc." (Results of the Laboratory for optical methods of the Institute for Math,
—
and Meeh., University of Leningrad), Leningrad-Moscow, 1935, pp. 135 149. 384.
Burgatti, P. [1] Teoria matematica della elasticita. Bologna, 1931. 3, 5/.
Carleman, T. 427.
Cauchy, L. io, 18, 39, 44, 53, 54, 55, 65.
Cesaro, E. 48.
Clebsch, A. [1] Theorie der Elastizitat der festen Korper, Leipzig 1862. 563.
— [2] Th6orie de l'61asticit6 des corps solides (Translation of [1] by B. de
Saint-Venant and A. Flamant with numerous applications by Saint-Venant).
Paris 1883. 585.
Coker, E. G. and Filon, L. N. G. [1] A Treatise on Photo-Elasticity. Cambridge
1931. (Russian Translation, Leningrad-Moscow, 1 936) . 96, 167, /69
Dini, U. [1] Sulla integrazione della equazione = 0. 3. 2nd. ser. Vol. V.
(1871/3).
Dinnik, A. N. 1n，迈 hhhk, A. H. [1] Torsion. Theory and application. Moscow-Lenin¬
grad, 1938. 598.
Dinnik, A N., Morgaevski, A. B., Savin, G. N. 卬ihhhk, A. H. MopraeBCKHfi,
A. B. CaBHH, r. H. [1] Distribution of stresses around mine shafts. Pro¬
ceedings of the Conference on the character of underground pressure. Academy
of Sciences, U.S.S.R. (1938) pp. 7—55. 3^7-
Duhamel J. M. C. 17
Falkovicz, S. V. anbKOBH% C. B. [1] On the pressure of a rigid stamp on the
elastic half-plane in the presence of regions of adhesion and slip. 10. Vol. IX,
—
no. 5. (1945) pp. 425 432. 505.
Filon, L. N. G. [1] On an approximative solution for the bending of a beam of
―
rectangular cross-section etc. 11. Vol. 201. (1903) pp. 63 155. 88, 96
716 AUTHORS INDEX AND REFERENCES

— [2] On the relation between corresponding problems in plane stress

in generalized plane stress. 12. Vol. I. (1930) pp. 289—299. 95.
and

[3] On stresses in multiply-connected plates. British. Association for the

advancement of science. Report of the 89-th meeting (1921), London 1922, pp.
305—315. 工56, 工59・
Flamant, A. 378.
Foppl, A. [1] Vorlesungen iiber technische Mechanik. Vol. V. 4th edition. 1922.
234-
Foppl, L. [1] Konforme Abbildung ebener Spannungszustande. 13. Vol. XI (1931)
—
pp. 81 92. 338, 49务
Fok, V. A. ok, B. A.[l] Sur la reduction du probleme plan d'ClasticitG d, une
Equation intSgrale de Fredholm. 14. Vol. 182 (1926) p. 264. 311.
[2] Reduction of the plane problems of the theory of elasticity to a Fred¬
holm integral equation. 15. Vol. 58, no. 1. (1927) pp. 11—20. 311.
Fok, V. A., Muskhelishvili, N. I. ok, B. A. Mycxe^ninsiuin, H. H. [1]
Sur r Equivalence de deux methodes de reduction de probleme plan biharmo-
nique 为 une Equation int6grale. 14. Vol. 196. ( 1933) p. 1947. 311.
Fredholm, I. [1] Solution d'un probleme fondamental de la th6orie de l'&asticit6.

Fridman, M. M.
—
16. Vol. 2, no. 28 (1906) pp. 3 8. ，工、《。8,《。9，603.
M. M. [1] Bending of a thin isotropic strip with a
curvilinear hole. 10. Vol. IX, no. 4. (1945) pp. 334—338. 368.
Galerkin, B. G. PajiepKHH, B. r. [1] Torsion of a triangular prism. 17. 1919,
pp. 111—118. 423, 575-
—
[2] Torsion of parabolic prisms. 18. Vol. 54, (1924) pp. 97 110. 575.
[3] Solution of St. Venanfs problem of flexure for certain boundaries in
—
the case of the prism. 18. Vol. 57. (1928) pp. 155 168. 595.
Galin, L. A. TaJiHH, JI. A. [1] The mixed problem of the theory of elasticity with
—
frictional forces for the half-plane. 1. Vol. XXXIX, no. 3. (1943) pp. 88 93. 483.
[2] Imprint of a stamp in the presence of friction and adhesion. 10.
Vol. IX, no. 5. pp. 413—424. 484.
[3] The plane elasto-plastic problem. Plastic regions around circular holes
in plates and beams. 10. Vol. X, no. 3, pp. 367—386. 368.
Ghosh, S. [1] On the flexure of an isotropic elastic cylinder. 19. Vol. 39, no. 1.
(1947) pp. 1—14. 595.
Glagolev, N. I. rjiarojies, H. H. [1] Elastic stresses along the bases of dams. 1.
Vol. XXXIV, no. 7 (1942) pp. 204—208. 457, 483.
[2] Determination of the stresses for the pressure of systems of rigid
—
stamps. 10. Vol. VII, no. 5 (1943) pp. 383 388. 457, 483.
[3] Resistance to rotation of cylindrical bodies. 10. Vol. IX, no. 4. (1945)
—
pp. 318 333. 49务
Golovin, Kh. Fojiobuh, X. [1] A static problem of the elastic body. Minutes
—
of the Technological Institute, St. Petersburg, 1880 1881, St. Petersburg,
1882. 232.
Gorgidze, A. Ya. Fopra^se, A. [1] The method of successive approximations
as applied to a plane problem of the theory of elasticity. 1.. Vol. IV, no. 5—6
(1934) pp. 254—256. 372.
[2] On an application of the method of successive approximations in the
 AUTHORS INDEX AND REFERENCES 717

—
theory of elasticity. 20. Vol. IV, (1938) pp. 13 42. 372.
[3] Secondary effects in the problem of extension of a compound bar. 7.
Vol. IV, no. 2 (1943) pp. Ul™114. 559.
[4] Secondary effects in the torsion of compound bars. 7. Vol. VII. no. 8.
(1946) pp. 515—519. 559.
[5] Extension and torsion of compound bars, which are almost prismatic.
7. Vol. VIII, no. 9—10 (1947) pp. 605—612. 559.
[6] Torsion of an extended, prismatic compound bar. 7. Vol. IX, no. 3.
—
(1948) pp. 161 165. 559-
[7] Torsion and bending of compound bars which are almost prismatic.
—
20. Vol. XVI (1948) pp. 117 141. (Georgian with short Russian summary). 559.
Gorgidze, A. Ya., Rukhadze, A. K. 「opruaee,A. H. Pyxa^se, A. K. [1] On
the numerical solution of the integral equations of the plane theory of elasti¬
city. 22. Vol. I, no. 4. (1940) pp. 255—258. 408.
[2] On secondary effects in the torsion of a reinforced circular cylinder.
7. Vol. Ill, no. 8. (1942) pp. 759—766. 559.
[3] Secondary effects in the problem of extension and bending by a couple
—
of compound bars. 20. Vol. XII (1943) pp. 79 94. 559.
Goursat, E. [1] Cours d'analyse math6matique. Vol. Ill, 3-me 6d., Paris 1927.
in, 114.
[2] Sur liquation = 0, 23. Vol. 26. 1898, p. 236. no.
Grammel, R. [1] Mechanik der elastischen Kdrper. Bearbeitet von G. Angenheister,
A. Busemann, O. Foppl, J. W. Geckeller, A. Nadai, F. Pfeiffer, Th. PGschl, R.
Riekert, E. Treffts, redigiert von R. Grammel, Berlin 1928. (Partly translated
into Russian). 3, 66.
Green, G. 55.
Hadamard, J. [1] M6moire sur le probleme d'analyse relatif 古 l'6quilibre des pla¬
ques dlastiques encastr^es. 24. Vol. 33, no. 4. Paris. 1908. 143.
Sur l'6quilibre des plaques 61astiques circulaires libres ou appuyees et
celui de la sphere isotrope. 25. Vol. XVIII (1901) pp. 313—342. 540.
Harnack, A. [1] Beitrage zur Theorie des Cauchyschen Integrals. Ber. d. K.
—
Sachs. Ges. d. Wiss. 1885; Republished in 26. Vol. 35. (1899) pp. 1 18. 286.
Hertz, H. 324.
Higgins, T. J. 58 工.
Hooke, R. 52.
Howland, R. C. I., Stevenson, A. C. [1] Biharmonic analysis in a perforated
strip. 11. Vol. 232 (1933) pp. 155—222： 347.
Inglis, C. E. [1] Stresses in a plate due to the presence of cracks and sharp cor¬
ners. 27. Vol. LV. 1913, pp. 219—230. 338.
Jeffery, G. B. [lj Plane stress and plane strain in bipolar coordinates. 11. Vol.
221 (1921) pp. 265—293. 244.
Kantorowicz, L. B., Krylov, V. I. KaHTopoBnn, JI. B. Kphinos, B. H. [1]
Methods of approximate solution of partial differential equations. Leningrad-
Moscow, 1936. 244.
Kartzivadze, I. N. KapiiMBa^se, H. H. [1] The fundamental problems of the
—
theory of elasticity for the elastic circle. 20. Vol. XII (1943) pp. 95 104. (Geor¬
gian with, short Russian summary). 504.
718 AUTHORS INDEX AND REFERENCES

[2] Effective solution of the fundamental problems of the theory of elasti¬

city for several regions. 7. Vol. VII, no. 8. (1946) pp. 507—513. 504, 552, 532.
Keldysh, M. V. KeJifflbini, M. B. 47工.
Kirchhoff, G. [1] Vorlesungen iiber mathematische Physik, Vol. I. Mechanik,
4th. ed. Leipzig, 1897 (1st ed. 1876.) 3, 14, 48, 70, 142.
Kirsch, G. 203.
Kolosov, G. V. Kojiocob, T. B. [1] On an application of complex function theory
to a plane problem of the mathematical theory of elasticity, Yuriev, 1909.
25, 87, 88, ii2, 工工4, "5, 工38, 148, 184, 2。3,320, 33& 385, 388.

—— —
~ [2] Uber einige Eigenschaften des ebenen Problems der Elastizitatstheo-
rie. 28. Vol. 62. (1914) pp. 383—409. 87, 148, 184, 320, 388.
[3] The influence of the elastic constants on the stress distribution in a
—
plane problem of the theory of elasticity. 29. no. 17. (1931) pp. 85 88. 155, 156.
[4] Sur Vextension d'une th^orfeme de Maurice Levy. 14. Vol. 188. p. 1593.
155, 76・
[5] Sur une application des formules de M. Schwarz, de M. Villat et de
M. Dini au probleme plan de l'61asticit6. 14. Vol. 193 (1931) p. 389. 32。.
[6] Application of the complex variable to the theory elasticity.
of Moscow-
Leningrad. 1935. 87.
Kolosov, G. V., Muskhelishvili, N. I. Kojiocob, r. B., MycxejimnBHjm, H. H.
[1] On the equilibrium of elastic circular discs. 29. Vol. XII (1915) pp. 39—55.
320, 328.
—
Korn, A. [1] Sur les equations de 1'elasticity 25. Vol. 24. (1907) pp. 9 75. 7 工.
[2] Uber die Losung des Grundproblems der Elastizitatstheorie. 26. Vol.
75. (1914) pp. 497—544. 71.
— [3] Solution g6n6rale du probldme d'Gquilibre dans la thCorie de l'61asticit6
——
dans le cas ou les efforts sont donn6s 力 la surface. 30. Vol. X (1908) pp. 165 269.
7工
[4] Sur Cquilibre des plaques Glastiques encastrCes. 25. Vol. 25. ( 1908) pp.
529—583. 143.
Lame, G. [1] Lemons sur la th6orie math6matique de l'61asticit6 des corps solides.
Paris, 1852. 59.
Lauricella, G. [1] SulF integrazione delle equazioni dell' equilibrio dei corpi
—
elastici isotropi. 4. Vol. XV. (1906) pp. 426 432. 41, 244, 398.《iz.
[2] Alcune applicazioni della teoria delle equazioni funzionali alia fisica-
matematica. 31. Vol. XIII (1907) pp. 104—118, 155—174, 237—262, 501—518.
71，4ii.
[3] Sur r integration de liquation relative A I'Cquilibre des plaques elasti-
ques encastr6es. 32. Vol. 32. (1909) pp. 201—256. i«3, 411.
Lavrentev, M. A. JIaBpeHTbeB, M. A. [1] Conformal transformations with appli¬
cations to several problems of mechanics, Moscow-Leningrad 1946. 166, 16g.
Leibenson, L. S. JIe86eH3OH, V. C. [1] Treatise on the theory of elasticity.
2nd, ed. Moscow-Leningrad 1947. 3, 423.
Lekhnitzki, S. G. JlexHHiiKnii, C. T. [1] Anisotropic plates. Moscow-Leningrad .
1947. 87, 42 工.
[2] On the influence of a circular hole on the stress distribution in a beam.
 AUTHORS INDEX AND REFERENCES 719

„ Optical methods of investigation of stresses in machine parts." Moscow-Lenin¬

grad. 1935. 346.
[3] On some problems, related to the theory of bending of thin strips.
—
10. Vol. II, no. 2. (1938) pp. 181 210. 319.
Levy, M., [1] Sur la 16gitimit6 de la regie dite du trapeze dans Fetude de la resistan¬
ce des barrages en magonerie. 14. Vol. 126, (1898) p. 1235. JC55.
Lichtenstein, L. [1] Uber die erste Randwertaufgabe der Elastizitatstheorie.
33. Vol. 20. (1924) pp. 21—28. 71.
Lokshin, A. S. JIokihhh, A. C. [1] Sur [influence d'un trou elliptique dans la
poutre qui eprouve une flexion. 14. Vol. 190. (1930) p. 1178. 345.
Lourie, A. I. Jlypbe, A. H. [1] On the problem of the equilibrium of plates with
—
supported edges. 34. Vol. XXXI (1928) pp. 305 320. 368.
[2] Some problems of bending of circular plates. 10. Vol. IV, no. 1. (1940)
—
pp. 93 102. 368.
Love, A. E. H. [1] A treatise on the mathematical theory of elasticity. 4th. ed.
Cambridge. 1927. Russian translation Moscow-Leningrad 1935. 3, 4, 14, ^1,
55, 95, ”7, T。％工59, i6i, 225, 324, 378, 423, 563, 592, 595, 6zo, 613.
Macdonald, H. M. [1] On the torsional strength of a hollow shaft. 35. Vol. 8. (1893)
pp. 62—68. 605.
Magnaradze, L. G. Mamapa/xse, 1n. F. [1] The fundamental problems of the
plane theory of elasticity for contours with cusps. 1. Vol. XVI, no. 3. (1937) pp.
157—161. 408.
[2] On the solution of the fundamental problems of the plane theory of
—
elasticity for contours with cusps. 1. Vol. XIX, no. 9 (1938) pp. 673 676. 408.
[3] The fundamental problems of the plane theory of elasticity for con¬
—
tours with cusps. 20. Vol. IV. (1938) pp. 43 76. 408.
— —
[4] Some boundary problems of mathematical physics for surfaces with
angular lines. 20. Vol. VII ( 1 939) pp. 25 46. 408.
Michell, J. H. [1] On the direct determination of stress in an elastic solid, with

95, 汇6, 223.

—
applications to the theory of plates. 36. Vol. 31 ( 1900) pp. 100 124. 5i, 76,

—
[2] Elementary distributions of plane stress. 36. Vol. 32 (1901) pp. 35 61.
324, 328.
388.
[3] The inversion of plane stress. 36. Vol. 34 (1902) pp. 134
—142. 25,

Mikhlin, S. G. Muxjihh, C. r. [1] Le probleme fondamental biharmonique 为

deux dimensions. 14. Vol. 197. (1933) p. 608. 309, 395, 396, 398, 407.
[2] Reduction of the fundamental problems of the plane theory of elasti¬
city to Fredholm integral equations. 1. New. ser. Vol. I. (1934) p. 295. 396.
[3] La solution du probleme plan biharmonique et des problemes de la
thSorie statique d'61asticit6 为 deux dimensions. 37. No. 37 (1934). 396.
[4] On the stress distribution in the half-plane with an elliptic hole. 37.
No. 29 (1934). 372.
[5] The method of successive approximation in the biharmonic problem.
37. no. 39. (1934). 372.
[6] The uniqueness theorem for the fundamental biharmonic problem. 38.
—
Vol. 41, no. 2. (1934) pp. 284 291. 153.
720 AUTHORS INDEX AND REFERENCES

[7] Quelques remarques relatives 亢 la solution des problemes plans d'61as-

—
ticit6. 38. Vol. 41. no. 3 ( 1 934) pp. 408 420. 373, 396.
[8] Certain cases of the plane problem of the theory of elasticity for non-
homogeneous media. 10. Vol. II, no；1. (1934) pp. 82 90. 218, 397, 420.
[9] The plane problem of the theory of elasticity. 37. No. 65. (1935).
372, 396.
[10] The plane problem of the theory of elasticity for non-homogeneous
media. 37. No. 66. (1935). 397, 42。.
[11] Plane deformations in anisotropic media. 37. No. 76 (1936). 87, 42Z.
[12] On the stresses in strata below layers of coal. 17. OTN, no. 7—8.
(1942). pp・ 13~~28. 475.
[13] Application of integral equations to certain problems of mechanics,
mathematical physics and engineering. Moscow-Leningrad.. 1947. 372, 396,
397, 402, 407.
Mintzberg, B. L. MnHix6epr, B. JI. [1] The mixed boundary problem of the
theory of elasticity for the plane with circular holes. 10. Vol. XII, no. 4. (1948)
pp. 415—422. 504, 514, 5工5.
Muskhelishvili, N. I. MycxejiuniBHJiH, H. U. [1] On thermal stresses in the
—
plane problem of the theory of elasticity. 29. Vol. XIII (1916) pp. 23 37.
H62, 229, 338, 5熄，525, 55夕
— [2] Sur i'Cquilibre des corps Glastiques soumis 左 [action de la chaleur.
39. No. 3 (1923). 162.
— [3] Sulla deformazione piana di un cilindro elastic。isotropo. 4. Vol. XXXI
(1922) pp. 548—551. 162.
[4] Sur Fint^gration de Fequation biharmonique. 17. (1919) pp. 663 686.
32。，333, 338, 353-
—
[5] Applications des int6grales analogues 为 celles de Cauchy 左 quelques
problemes de la Physiques Math6matiques. Tiflis, Edition de rUniversit6, 1922.
32。，353, 363・
— [6] Sur Fint6gration approch6e de liquation biharmonique. 14. Vol. 185
(1927) p. 1184. 370.
[7] Sur la solution du probleme biharmonique pour Faire ext6rieure 为
—
une ellipse. 33. Vol, 26, ( 1 927) pp. 700 705. 32。.
[8] Praktische Losung der fundamentalen Randwertaufgaben der Elasti-
zitatstheorie in der Ebene fiir einige Berandungsf ormen . 13. Vol. 13 (1933) pp.
—
264 282. 工“，320.
[9] Nouvelle m6thode de reduction de probleme biharmonique fonda-
mental 为 une Equation de Fredholm. 14. Vol. 192. (1931) p. 77. 309.
[10] Thdoremes d'existence relatifs au probleme biharmonique et aux pro-
bl^mes d'61asticit6 a deux dimensions. 14. Vol. 192 (1931) p. 221. 309.
[11] Recherches sur les probUmes aux limites relatifs 亢 liquation bi-
harmoniques et aux Equations de r61asticit6 礼 deux dimensions. 26. Vol. 107.
—
(1932) pp. 282 312. n6, Z53, 309.
[12] Sur le probleme de torsion des cylindres, 6Iastiques isotropes. 4. Vol.
IX (1929) pp. 295—300. 576, 577, 5^, 595・
[13] Zum Problem der Torsion der homogenen isotropen Prismen. 40.
—
Vol. I (1929) pp. 1 20. 576, 577, 581.
 AUTHORS INDEX AND REFERENCES 721

[14] Sur le probleme de torsion des poutres Slastiques composSes. 14. Vol.
194. (1932) p. 1435. 597・
[15] On the problem of torsion and bending of elastic compound bars. 17.
no. 7 ( 1 932) pp. 907—945. 597, 6口，624.
[16] The solution of the plane problems of the theory of elasticity for the
—
continuous ellipse. 10. Vol. I. no. 1. (1933) pp. 5 12. 244.
[17] A new general method of solution of the fundamental boundary
problems of the plane theory of elasticity. 1.. Vol. Ill, no. 1 (1934) p. 7, 398, 407.
[18] A study of the new integral equations of the plane theory of elasticity.
1. Vol. Ill, no. 2. ( 1934) p. 73. 398, 407.
[19] On a new boundary problem of the theory of elasticity. 1. Vol. Ill,
no. 3. (1934), p. 141. 367, 54。，543.
[20] The solution of the fundamental mixed problem of the theory of
—
elasticity for the half -plane. 1. Vol. VIII, no. 2. (1935) pp. 51 54. 457.
[21] On the numerical solution of the plane problems of the theory of elastici¬
—
ty. 20. Vol. I. (1937) pp. 83 87. (Georgian with short Russian summary) . 408.
[22] The fundamental boundary problems of the theory of elasticity for
——
the half -plane. 7. Vol. II, no. 10 (1941) pp. 873 880. 451, 457, 471.
[23] The fundamental boundary problems of the theory of elasticity for
—
the plane with straight cuts. 7. Vol. Ill, no. 2. ( 1 942) pp. 103 1 10. 494.
[24] On the problem of the equilibrium of a rigid stamp on the boundary
of an elastic half-plane in the presence of friction. 1. Vol. Ill, no. 5. (1942) pp.
413—418. 483.
[25] Singular integral equations. Boundary problems of function theory
and their application to mathematical physics. Moscow-Leningrad 1946. (Trans¬
lation into English, edited by J. R. M. Radok, Groningen 1953). 251, 255, 263,
264, 425, 429, 442, 455, 479, 497.
Naiman, M. I. HaiiMaH, M. II. [1] The stresses in a bar with a curvilinear hole.
41. no. 313. Moscow 1937. 346, 371.
Navier, C. L. M. H. 73, 78.
Neumann, F. /4,
Osgood, W. F. [1] Lehrbuch der F unktionentheorie. Vol. I. Leipzig 1912. 169 .
Papkovicz, P. F. UanKOBHq, II. [1] Theory of elasticity. Moscow-Leningrad
1939. 3, 423.
Plemelj, J. [1] Ein Erganzungssatz zur Cauchyschen Integraldarstellung analy-
tischer Funktionen, Randwerte betreffend. 42. Vol. XIX (1908) pp. 205 210.
际，285.
—
[2] Potentialtheoretische Untersuchungen. Leipzig 191 1. 602.
Poincare, H. 601.
Poisson, S. D. 53, 54, 55, 65.
Poritsky, H. [1] Thermal stresses in cylindrical pipes. 43. Vol. 24. (no. 160) (1937)
pp. 209—223. 工工5，162.
[2] Application of analytic functions to two-dimensional biharmonic analy¬
—
sis. 5. Vol. 59. no. 2. (1946) pp. 248 279. 87, 114.
Poschl, T. [1] Uber eine partikulare Losung des biharmonischen Problems fiir
—
den Aussenraum der Ellipse. 33. Vol. XI. (1921) pp. 89 -96. 338, 339.
722 AUTHORS INDEX AND REFERENCES

Privalov, I. I. IIpMBajiOB, H. H. [1] Introduction to complex function theory.

8th. ed. Moscow-Leningrad 1948. r66, 251, 263, 429, 430.
Radon, J. 408.
Robin.
Rukhadze, A. K. Pyxajuse, A. K. [1] Bending by a transverse force of a circular

—
cylinder, reinforced by additional circular rods. 17. no. 9. (1933) pp. 1297 1308.
623.
[2] Torsion and bending of bars, consisting of two elastic materials,
—
bounded by epitrochoids. 20. Vol. I (1937) pp. 125 139. (Georgian with short
Russian summary) . 623.
[3] On the bending of elastic compound bars. 22. Vol. I, no. 2. (1940)
—
pp. 107 114. 649.
[4] Secondary effects in the problem of bending by a couple of a compound
bar. 7：Vol. IV no. 2. (1943) pp. 115—122. 559.
Saint-Venant, B. de. [1] MGmoire sur la torsion des prismes etc. 24. Vol. XIV
—
(1855) pp. 233 560. 44, 4% 76, 77, 559, 562, 575, 6io.

563, 595・
—
M6moire sur la flexion des prismes, etc. 44. Vol. I. (1856) pp. 89 189.

Sadovski, M. A. Ca^OBCKHii, M. A. [1] Zweidimensionale Probleme der Elastizi-

—
tatstheorie. 13. Vol. 8 (1928) pp. 107 121. 385, 3^i, 47 工, 480.
[2] Uber Randwertaufgaben fiir die elastische Halbebene und die ge-
—
schlitzte elastische Vollebene. 13. Vol. 10. (1930) pp. 77 81. 385, 3gi.
Savin, G. N. CaBHH, r. H. [1] Stress distribution in a plane field with several
holes. 45. Report 10 (1936). 370.
[2] Stress concentrations near small holes in a non-homogeneous plane
field. 45. Report 20 ( 1 937) . 346, 370.
[3] The fundamental plane static problem of the theory of elasticity for
anisotropic media. 46. No. 32 (1938). 87, 421.
[4] On a method of solution of the fundamental plane static problems of the
—
theory of elasticity of anisotropic media. 47. No. 3 (1939) pp. 123 139. 8^, 421.
[5] The pressure of an absolutely rigid stamp on an elastic anisotropic
—
medium. 48. No. 6 (1939) pp. 27 34. 87, 42T.
[6] Some problems of the theory of elasticity of anisotropic media. 1.
Vol. XXIH, no. 3 (1939) pp. 217—220. 87, 421.
[7] Stresses in the elastic plane with an infinite number of equal cuts.
1. Vol. XXIII, no. 6 (1939) pp. 515—518. 407.
Schwarz, H. A. 371.
Sedov, L. I. Cg^ob, JI. H. 47 工.
Shapiro, G. S. IHanHpo, r. C. [1] Stresses around a hole in an infinite wedge.
34. no. 3 (1941) pp. 184—199. 367.
Sherman, D. I. UlepMaH, 口. U. [1] On a method of solution of a static problem
of stresses in plane multiply connected regions. 1. Vol. I, no. 7 (1934) pp. 376
378. 3", 3工9， 395, 397> 398, 409, 413, 4工4 —
[2] On the solution of the second fundamental problem of the theory of
elasticity for plane multiply connected regions. 1. Vol. IV (IX) no. 3 (1935)
p. 119—122. 4”.
— [3] On the new method of N. I. Muskhelishvili for the plane problem of
 AUTHORS INDEX AND REFERENCES 723

—
the theory of elasticity. 1. Vol. I (X) no. 5 ( 1 936) pp. 201 206. 4”.
[4] Determination of stresses in the half-plane with an elliptic cut. 37.
No. 53 (1935). 372, 397.
[5] On a method of solution of the static plane problem of the theory of
elasticity for multiply connected regions. 37. no. 54 (1935). 370, 372, 397.
[6] The static plane problem of the theory of elasticity. 20. Vol. II (1937)
—
pp. 163 225. 4”.
[7] On the distribution of the eigenvalues of the integral equations of
the plane theory of elasticity. 37. No. 82 (1938). 318.
[8] The static plane problem of the theory of elasticity for isotropic lion-

— —
homogeneous media. 37. No. 86 (1938) pp. 1 50. 407.
[9] The plane problem of the theory of elasticity for anisotropic media.
37. No. 86 (1938) pp. 51—78. 87, 42r.
[10] The plane problem of the theory of elasticity with, mixed boundary-
conditions. 37. No. 88 (1938). 318, 362.
[11] On certain properties of the integral equations of the theory of elasti¬
city. 37. No. 100 (1940). 407.
[12] The elastic plane with straight cuts. 1. Vol. XXVI, no. 7 (1940) pp.
635—638. 503。
[13] The mixed problems of potential theory and of the theory of elastic¬
ity for the plane with a finite number of straight cuts. 1. Vol. XXVII, no. 4
(1940) pp. 330—334. 5”，503.
[14] On a problem of the theory of elasticity. 1.. Vol. XXVII, no. 9 ( 1 940)
—
pp. 907 910. 432.
[15] On the solution of the plane static problem of the theory of elasticity
for displacements, given on the boundary. 1. Vol. XXVII, no. 9 (1940) pp.
911—913. 408, 4r o, 412.
— 。，
F16] On the solution of the plane static problem of the theory of elasticity
—
for given external forces. 1.. Vol. XXVIII, no. 1 (1940) pp. 25' 28. 408, 4r
412.
―connected regions. Vol. XXVIII, no. 1 (1940) pp. 29—32. 408, 420.
— - [17] mixed problem of
The
1..
static theory elasticity for
the of multiply plane

[18] On the stresses in an elliptic plate. 1. Vol. XXXI, no. 4 (1941) pp.
—
309 310. 244.
[19] A new solution of the plane problem of the theory of elasticity for
anisotropic media. 1. Vol. XXXII, no. 5 (1941) pp. 314—315. 87, 42工.
[20] Plane strain in isotropic, non-homogeneous media. 10. Vol, VII (1943)
—
pp. 301 309. 625.
[21] The three-dimensional static problem of the theory of elasticity for
displacements given on the boundary. 10. Vol. VII (1943) pp. 341—360. yi.
[22] On a mixed problem of the theory of elasticity. 10. Vol. VII ( 1 943)
—
pp. 413 420. 42i, 542..
Shtaerman, I. Ya. IIlTaepMaH, H. H. [1] On Hertz' theory of local deformations in
—
elastic bodies in compression. 1.. Vol. XXV, no. 5 (1939) pp. 360 362. 490, 493.
Generalization of Hertz' theory of local deformation in elastic bodies
—
in compression. 1. Vol. XXIX, no. 3 (1940) pp. 179 181. 490.
724 AUTHORS INDEX AND REFERENCES

[3] Some special cases of the contact problem. 1.. Vol. XXXVIII, no. 7
(1943) pp. 220—224. 490, 494.
[4] The contact problems of the theory of elasticity. Moscow-Leningrad
1949. 490, 494.
Smirnov, V. I. Cmhphob, B. H. [1] Higher mathematics for engineers and phys¬
icists. Vol. II, Vol. III. Moscow-Leningrad.
[2] Uber die Randerzuordnung bei konformer Abbildung. 26. Vol. 104
(1932) pp. 313—323. /68.
Sobolev, S. L. CooojieB, C. JI. [1] On a regional problem for polyharmonic equa¬
tions. 38. Vol. 2 (44), no. 3 (1937) pp. 465—499. /43.
[2] Schwarz's algorithm in the theory of elasticity. 1. Vol. XIII (1936)
—
pp. 235 238. 372.
Sokolnikoff, I. S. [1] Mathematical theory of Elasticity. New York London
1946. 3, 576, 583 595-
—
Sokolov, P. Cokojiob, II. [1] Stress distribution in a plane strip with several
holes. 49. Vol. IV. 1930, pp. 39—71. 239.
Stevenson, A. C. [1] Complex potentials in two-dimensional elasticity. 50. Vol.
184, no. 997 (1945) pp. 129—179, pp. 218—229. 87, 114, 115.
Tedone, O. [1] Sui problemi di equilibrio elastico a dui dimensioni. 8. Vol. 41
(1905—1906) pp. 86—101. 244.
Thomson, W. (Lord Kelvin) . 55.
Timoshenko, S. P. Thmoihohko, S. II. [1] Theory of elasticity. Petrograd. Vol.
I, 1914, Vol. II, 1916. 3. 14-
[2] Theory of elasticity. New York—London. 1934. 3, 14.
Timpe, A. [1] Probleme der Spannungsverteilung in ebenen Systemen, einfach

—
gelGst mit Hilfe der Airyschen Funktion. 28. Vol. 52 (1905) pp. 348 383. 5工,
汇9，223, 228, 22$l.
[2] Die Airysche Funktion fiir den Ellipsenring. 33. Vol. 17 (1923) pp.
—
189 205. 244.
Todhunter, I., Pearson, K. [1] A history of the theory of elasticity and of the
strength of materials. Cambridge, Vol. I. 1886, Vols. 工耳，II2, 1893. 3, 563, 575,
595*
Tuzi, Z. [1] Effect of a circular hole on the stress distribution in a beam under
—
uniform bending moment. 43. Vol. 9, no. 56 (1930) pp. 210 224. 346.
Vekua, I. N. BeKya, H. H. [1] New methods of solution of elliptic equations.
Moscow-Leningrad 1948. 42Z.
[2] Application of the method of the Academician N. I. Muskhelishvili to
the solution of boundary problems of the plane theory of elasticity of anisotropic
media. 22. Vol. I, no. 10 (1940) pp. 719—742. 42r.
[3] On the bending of plates with free edges. 7. Vol. Ill, no. 7 (1942) pp.
641—648. 3H9 •
[4] Integration of the equations of spherical shells. 10. Vol. IX, no. 5 (1945)
—
pp. 368 388.
[5] On the theory of thin elastic shells. 10. Vol. XII, no. 1 (1948) pp.
69—74. 42r.
Vekua, I. N., Rukhadze, A. K. Beicya, 0. H. Pyxa«3e, A. K. [1] The problem
 AUTHORS INDEX AND REFERENCES 725

of torsion of a circular cylinder, reinforced by additional circular rods. 17. No.

3 (1933) pp. 373—386. 605, 608.
[2] Torsion and bending by a transverse force of a bar, consisting of two
—
materials bounded by confocal ellipses. 10. Vol. I, no. 2 (1933) pp. 167 178.
6io, 623.
Voigt, W. 563.
Volkov, D. M., Nazarov, A. A. Bo^kob, 口. M. Ha3apoB, A. A, [1] On a boundary
problem and its application to the plane theory of elasticity. 38. Vol. 40, no. 2
—
(1933) pp. 210 228. 239, 32。，366.
[2] The plane problem of the theory of elasticity in the case of simply
—
and doubly connected regions. 10. Vol. I, no. 2 (1933) pp. 209 227. 239, 32。.
Volterra, V. [1] Sur l'6qu"ibre des corps Gastiques multiplement connexes. 25.
Vol. 24 (1907) pp. 401—517. 48, 51, /59.
[2] Drei Vorlesungen iiber neuere Fortschritte der Mathematischen Physik,
gehalten im September 1909 an der Clark-U niversity, Deutsch von E. Lamia,
Leipzig-Berlin, 1914. 759.
— [3] Lemons sur V integration des Equations differentielles aux d6riv6es par-
tielles profess^es A Stockholm, Paris. 1912. /59.
Webster, A. G. [1] The dynamics of particles and of rigid, elastic and fluid bodies.
Leipzig 1904. (Russian translation of the second unchanged edition. Moscow-
Leningrad 1933). 3, 563.
Weinel, E. [1] Das Torsionsproblem fiir den exzentrischen Kreisring. 21. Vol.
Ill, no. 1 (1932) pp. 67—75. 605.
Weyl, H. [1] Das asymptotische Verteilungsgesetz der Eigenschwingungen eines
―
beliebig gestalteten elastischen Korpers. 2. Vol. XXXIX (1915) pp. 1 49. jt.
Young, Th. 63.
A
 SUPPLEMENTARY REFERENCES

Avazashvili, D. Z. ABasamBH.iH,. 口 3. [1] On applications of the theory of functions

of a complex varaible to problems of torsion and bending. 10, Vol. IV, no. 1,
1940, pp. 129—134.
Chaplygin, S. A. Han.ibiruH, C. A. [1] Collected works. Vol. Ill, Moscow-
Leningrad, 1950.
Fridman, M. M. mpnnMaH, M. M.「2]Solution of the general problem of bending
of thin isotropic elastic plates supported along the edges. 10, Vol. XVI,
1952, pp. 429—436.
Galin, L. A. Fa^nir, JI. A. [4」The contact problems of the theory of elasticity,
Moscow-Leningrad, 1953.
Gorgidze, A. la. Fopmase, A. H. [8] Bending by a couple of an extended pris¬

—
pp. 539 545.
——
matic beam consisting of different materials. 22, Vol. IX, no. 9 10, 1948,

[9] Bending by a couple of a clamped composite beam. 51, no. 17, 1948,
—
pp. 79 87.
[10] Extension and bending by a couple of naturally twisted beams. 22,
Vol. XIlL no. 2, 1952, pp. 73—80.
Gray, C. A, M. [1] Polynomial approximations in plane elastic problems 53,
Vol. IV, 1951, pp. 444—448.
Ianchevskii, S. A. Hii^encKUH, ( A. [1] Functions of a complex variable. Lenin¬
grad, 1934.
Kalandha, A. I. Kananana, A. H. [1] On a mixed problem of bending of an
elastic plate. 10, Vol. XVI, no. 3, 1952, pp. 271—282.
[2] The general mixed problem of bending of elastic plates. 10, Vol. XVI,
no. 5, 1952, pp. 513—532.
[3] Solution of certain problems of bending of elastic plates. 10, Vol.
XVII, no. 3, 1953, pp. 293—310.
[4] On the problem of equilibrium of an elastic plate with supported
edges. 20, Vol. IXX, 1953, pp. 193—210.
[5] The bending of an elastic plate in the form of an elliptic ring. 10,
Vol. XVII, no. 6, 1953, pp. 691—704.
Khalilov, Z. I. XaTin.noB, 3. H. [1] Solution of the general problem of bending
—
of supported elastic plates. 10, Vol. XIV, no. 4, 1950, pp. 405 414.
Kholmianskii, M. M. Xo^mhhckmh, M. M. [1] On the solution of the systems of
algebraic equations of the basic problems of the plane theory of elasticity. 10,
Vol. XV, no. 3, 1951, pp. 317—322
Lavrentiev, M. A., Shabat, B. V. JIaBpeHTbeB, M. A., IIIadaT, B, B. [1] Methods
of the theory of functions of a complex variable, Moscow-Leningrad, 1951.
Mandjavidze, G. F. MaHAJKaBM4,3e, T. 4>. [1] On a c均ss of singular integral equa¬
—
tions with discontinuous coefficients. 22, Vol. XI, no. 5, 1950, pp. 269 274.
[2] On a singular integral equation with discontinuous coefficients and
—
its application in the theory of elasticity. 10, Vol. XV, no. 3, 1951, pp. 279 296.
Markushevich, A. I. MapKymeBHq, A. H. [1] The theory of analytic functions.
Moscow-Leningrad, 1950.
727
728 SUPPLEMENTARY REFERENCES

Polo jii, G. N. IIoJiojKnii, r. H. [1] The solution of the third fundamental problem
of the plane theory of elasticity for an arbitrary, finite, convex polygon. 1,
Vol. LXXHI, no. 1, 1950, pp. 49—52.
Rukhadze, A. K. Pyxa^se, A. K. [5] On the problem of the deformation of
beams which are approximately prismatic and consist of different elastic ma¬
—
terials. 51, no. 23, 1951, pp. 23 37.
[6] The problem of extension by forces of naturally twisted beams con¬
—
sisting of different elastic materials. 22, Vol. XIII, no. 3, 1952, pp. 137 144.
[7] The problem of bending by a couple of naturally twisted prismatic
beams consisting of different elastic materials. 22, Vol. XIII, no. 5, 1952,
—
pp. 265 272.
Savin, G. N. CaBWH, T. H. [8] Stress concentrations around openings. Moscow-
Leningrad, 1951.
Sheremetiev, M. P. UlepeMeTbeR, M. A. [1] The elastic equilibrium of infinite
plates with inserted absolutely rigid or elastic discs. 10, Vol. XVI, no. 4, 1952,
—
pp. 437 448.
[2] The elastic equilibrium of an elliptic ring. 10, Vol. XVII, no. 1, 1953,
pp. 107—113.
Sherman, D. I. IIJepnaH, 1nl. H. [23] On the state of stress of several pressed
machine parts. 17, OTN, no. 9, 1948, pp. 1371 1388.
[24] On the stresses in a heavy half-plane with two circular openings.
10, Vol. XV, no. 3, 1951, pp. 297—316.
[25] On the stresses in a plane heavy medium with two identical sym¬
metrically distributed openings. 10, Vol. XV, no. 6, 1951 pp. 751—761.
[26] On the problem of the state of stress in mine shaft pillars. Elastic
heavy media with two openings of elliptic shape. 17, OTN, no. 6, 1952, pp. 840
857, no. 7, 1952, pp. 992—1010.
—
[27] Torsion of an elliptic cylinder reinforced by a circular rod. 52, Vol. X,
1951, pp. 81—108.
Sokhotskii, Iu. V., CoxoriKHfi, K). B. [1] On definite integrals and functions used
for expansions in series. Petrograd, 1873.
Tarabasov, N. D., TapaGacoB, H. 以. [1] Determination of stresses in plates
—
with several pressed in disc. 1, Vol. LXIII, no. 1, 1948, pp. 15 18.
[2] Analysis of the strength of press fits. 1, Vol. LXVII, no. 4, 1949,
―
pp. 615 618.
[3] Analysis of the strength of composite annular machine parts, 1,
Vol. LXX, no. 6, 1950, pp. 977—980.
Tiffen, R., [1] Uniqueness theorems of two-dimensional elasticity theory. 53,
Vol. V, 1952, pp. 237—252.
[2] Boundary value problems of the elastic half-plane. 53, Vol. V, 1952,
—
pp. 344 351.
[3] Solution of two-dimensional elastic problems by conformed mapping
—
on to a half-plane. 53, Vol. V, 1952, pp. 352 360.
Ugodchikov, A. G. yro^^HKOB, A. r. [1] Determination of the stresses for pressed
in circular discs in plates bounded by a curve of given shape. 1, Vol, LXXVII,
—
no. 2, 1951, pp. 213 216.
Vainberg, D. V. BaiiHbepr, 口. B. [1] The state of stress of composite discs and
plates. Kiev, 1952.
Vekua, I. N. BeKya, H. H. [6] On the solution of boundary value problems of
—
the theory of shells. 22, Vol. XV, no. 1, pp. 3 6.
 SUBJECT INDEX

Cauchy integrals 267—313

—
Absence of body forces 1 0 1
application of 316 443
Affine, transformation 29
calculation of 281
infinitesimal transformation 31
Airy function 105
derivative of 279
—
Analysis, of strain 28 51
principal value of 271
—
—
of stress 5 27
Analytic, continuation 131, 135
properties of 267 295
Centre of area 608
Characteristic of dislocation 170
—
function 109, 115
Angle of shear 38
Circular ring 184, 230 249, 455
disc 204, 338—345, 529, 568
Anisotropic 441
Arc 267
Compatibility conditions 44, 77
Complex representation 1 10, 1 13
Auxiliary problem, thermal 173
of plane deformation 650
Components of stress 7
Concentrated, couple 216
force 162, 215, 338
Conformal mapping 1 76~ 195,
B 250—264, 399
by polynomial 366

——
compound 621 681 —
Bars, homogeneous 583 620 by rational function 374, 411
Conjugate points 182
bending of 583 681 Conservation of energy 55
—
Basic, equations 60, 72, 89 104
case 621
Constants, elastic 61
Contact of bodies 510
Bending of curved beam 264 Continuity condition 49
Bernoulli-Euler law 607, 647, 664, 669, Coordinates, curvilinear 190
681 polar 140
Biharmonic function 1 10 rectilinear 137
Bilinear function 180 transformation of 14
Body, anisotropic 441 Cubical dilatation 41, 65
force 6, 101 Curve, simple smooth 131
homogeneous 55 with corners 427
isotropic 56
motion 32 D
non-homogeneous 55
Boundary, conditions 194 Deformation, general 41
part of 132 of a plate 92
value problems 66, 141 pure 32
Bulk modulus 65 Dilatation 41, 65

729
730 SUBJECT INDEX
Dirichlet, condition 199, 202 holomorphic at infinity 1 27, 284
problem 247, 298, 308, 389, 595, 597 of bounded variation 200
Discontinuity 449 principal part of 283
Discontinuous coefficients 468 sectionally holomorphic 447, 449
Dislocations 167, 237 stress 105
Displacement 41 Functional equation 317
assumptions 42 Fundamental, law 52
at infinity 130 problems 66, 78, 141, 147, 204—264,
components of 41 316, 588
general conditions 107
multi-valued 167, 237 G
single-valuedness of 125
General, formulae of plane elasticity
89—195
E
representation of solutions 441
Generalized Hooke's law 52
Elastic constants 61
Green's function 415, 417
Ellipse 257, 598, 607
Elliptic, hole 347
H
ring 188
Energy, kinetic 82 Half-plane 391, 471—524
potential 69, 83 Hamack's theorem 301
Epitrochoids 186
Heat flow 172
Equilibrium, dynamic 78 Hertz problem 510
static 61
Existence theorems 323 —
Hilbert problem, 447 578

—
homogeneous 449, 460
Extension, of bars 583 68 1
non-homogeneous 449, 462
simple, homogeneous 36
Holder (H) condition 268
at a point 269
at infinity 293

Force, body 5 708—711

—
Holomorphic function 298 313,

centrifugal 101 Hooke's Law 52

concentrated 162, 215, 338, 396 Hydrostatic pressure 64
external 10 Hypotrochoids 187
gravity 101
resultant 1 1 6
surface 5
—
Fourier series 199 203
Fredholm equation 317
Inversion 291

Function, analytic 109, 1 15

biharmonic 1 1 0
boundary value of 132, 268, 299 Lemniscate 190, 604, 605
continuous from left (right) 1 32, 268
continuous on boundary 268
Linear relationship, problem of 447
578 —
continuous up to 132 Line 267
—
holomorphic 109, 284, 298 313 end of 267
 SUBJECT INDEX 731

left (right) neighbourhood of 267 Polar coordinates 140

of discontinuity 447 Potential energy 69
positive direction of 267
simple, smooth. 267 Principal, axes 22, 38
—
Power series solutions 199 264

Liouville theorem 450 stress 17, 22

value of Cauchy integral 271
M
R
Modulus, bulk 65
of compression 65 Reciprocal theorem of Betti 657
of elasticity 63 Reduced, centre of gravity 642
Moment, resultant 1 16 principal axes 642
Multiply connected regions 121, 416, Reflection in a circle 181
697 Regions, infinite 126
multiply connected 50, 100, 121, 157,
N 416
semi-infinite 397
Natural state 134 simply connected 48, 50, 164, 250
Neumann, problem 308, 589, 597, 615 Regular solution 158
series 426 Relative twist 590
Normal, 7 Representation of functions 111, 441
direction of 7 Resultant, force 1 16
moment 1 1 6
O Riemann problem 445

O(x), o(x) 269 elliptic 188

—
Ring, circular 230 249

0(1), o(l) 270 rotating 236

Rotation, at infinity 129
components of 34
of axes 139
Pascal's limacon 185, 255 positive direction of 24
Plane, strain 89
stress 23
with circulair cuts 538
Plate, middle surface of 92 Saint-Venant * principle 76
supported 385 Schwarz, formula 310
with circular hole 208, 21 1, 222, 380, algorithm 389, 416
530, 571 Sectionally holomorphic 447, 449
with clamped edges 145, 333, 385 Shear modulus, 64, 622
with elliptic hole 351, 363, 558, 574 Solution, regular 158
with free edges 333, 385 Square 188
with holes 452 Stamp 486—504, 537
with straight cuts 515 Strain, analysis of 28—31
Plemelj formulae 276 components of 44
Poisson's, ratio 63, 65 plane 89
coefficient 66 principal axes of 38
formula 31 1 surface 38
732 SUBJECT INDEX

Stress, 6 affine 29
—
analysis of 5 27 conformal 399
components of 7
function 105
―
formulae 176 195
infinitesimal 31
generalized plane 96 polynomial 366
plane 23 rational 374, 411, 546—578
principal 17, 22 Torsion, function 594, 596, 622
surface 17 problem 597
tensor 14 Tube 235
thermal 170, 246
Strip with elliptic hole 358
U
T
Uniqueness, of solutions 66, 141, 560
Temperature distribution 247 assumptions 71

—
Tensor 682 696
Thermal stress 1 70, 246
of functions 1 18
Uniform tension 156

Torus 50 —
Torsion of bars 583 681
Y
Traction 6
Transformation, of coordinates 14, 38,
137 Young's modulus 64