Kunia Murasugi

Translated by Bohdan Kurpita

KNOT THEORY
and
ITS APPLICATIONS
Birkhauser
Boston Basel Berlin
Kunio Murasugi
Department of Mathematics
University of Toronto
Toronto, Ontario M5S tAl
Canada
Bohdan Kurpita
The Daiwa Anglo-Japanese
Foundation and Waseda University
Shinjuku-ku, Tokyo
Japan
96-16329
eIP
Library of Congress Cataloging-in-Publication Data
Murasugi, Kunio, 1929..
[Musubime riron to sono oyo. English]
Knot theory and its applications I Kunio Murasugi ; translated by
Bohdan Kurpita.
p. em.
Includes bibliographical references (p. -) and index.
ISBN 0-8176-3817-2 (alk. paper). -- ISBN 3-7643-3817-2 (alk.
paper)
1. Knot theory. I. Title.
QA612.2.M8613 1996
514'.224....dc20
Printed on acid-free paper
Published originally in 1993 in Japanese
1996 Birkhauser Boston
Birkhiiuser
Copyright is not claimed for works of U.S. Government employees.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronic, mechanical, photocopy-
ing, recording, or otherwise, without prior permission of the copyright owner.
Permission to photocopy for internal or personal use of specific clients is granted by
Birkhauser Boston for libraries and other users registered with the Copyright Clearance
Center (CCC), provided that the base fee of $6.00 per copy, plus $0.20 per page is paid
directly to CCC, 222 Rosewood Drive, Danvers, MA 01923, U.S.A. Special requests
should be addressed directly to Birkhauser Boston, 675 Massachusetts Avenue, Cam-
bridge, MA 02139, U.S.A.
ISBN 0-8176-3817..2
ISBN 3-7643-3817-2
Typeset in lEX by Bohdan Kurpita
Printed and bound hy Maple-Vail, York, PA
Printed in the U.S.A.
t) H 7 6 543 2
CtI"I,,,I,
Introduction 1
Fundamental Concepts of Knot Theory.............................. 5
1 The elementary knot moves 6
2 The equivalence o/knots (1) 7
3 The equivalence ofknots (II) 9
4 Links 14
5 Knot decomposition and the semi-group ofa knot .. oo 17
6 The cobordism group of knots 23
Knot Tables ....... 25
1 Regular diagrams and alternating knots 26
2 Knot tables 30
3 Knot graphs 34
Fundamental Problems of Knot Theory .. 40
1 Global problems 41
2 Local problems 43
Classical Knot Invariants . 47
1 The Reidemeister moves 48
2 The minimum number ofcrossing points oo oo 56
3 The bridge number 58
4 The unknotting number oo oo oo oo 61
5 The linking number 64
6 The colouring number ofaknot oo 69
Seifert Matrices ..... 75
1 The Seifert suiface 76
2 The genus ofa knot 80
3 The Seifert matrix 83
4 S-equi'Qtllence of Seifert matrices .oo oo oo oo 89
I
I
I
I r
Invariants from the Seifert matrix 104
1 The Alexander polynomial 105
2 The Alexander - Conway polynomial 108
3 Basic properties of the Alexander polynomial 116
4 The signature ofaknot 122
Torus Knots ......... 132
1 Torus knots 133
2 The classification of torus knots (1) ..................................... 137
3 The Seifert matrix ofa torus knot 141
4 The classification of torus knots (II) 143
5 Invariants of torus knots 148
Creating Manifolds from Knots . 152
1 Dehn surgery 154
2 Covering spaces 159
3 The cyclic covering space ofaknot oo 163
4 A theorem ofAlexander oo 166
Thngles and 2-Bridge Knots ........... 171
1 Tangles 172
2 Trivial tangles (rational tangles) 176
3 2-bridge knots (rational knots) 182
4 Oriented 2-bridge knots 194
The Theory of Braids .. 197
1 Braids.................................................................................... 198
2 The braid group 201
3 Knots and braids 209
4 The braid index 214
The Jones Revolution ....... 217
1 The lones polynomial : 219
2 The basic characteristics of the Jones polynomial 222
3 The skein invariants 231
4 The Kauffman polynomial 232
5 'fhe skein polynomials and classical knot invariants.
(Alternating knots and the Tait conjectures) 241
Knots via Statistical Mechanics 248
1 The 6-vertex model 249
2 The partition function for braids 255
3 An invariant ofknots 260
Knot Theory in Molecular Biology 267
1 DNA and knots 268
2 Site-speCific recombination Of to 271
3 A .model for site-specific recombination 273
4 Recombination due to the recombinase Tn3 Resolvase 276
Graph Theory Applied to Chemistry............................... 284
1 An invariant ofgraphs: the chromatic polynomial 286
2 Bing's conjecture and spatial graphs 289
3 The chirality ofspatial graphs Of 296
Vassiliev Invariants . 300
1 Singular knots 301
2 Vassiliev invariants 304
3 Some examples of Vassiliev invariants ff 308
4 Chord diagrams 313
5 Final remarks 321
Appendix . 325
Appendix (1): a table ofknots 326
Appendix (II): Alexander and Jones polynomials 327
Notes 329
Bibliography 333
Index: ..... 337
Knots and braids have been extremely beneficial through the ages
to our actual existence and progress. For example, in the primordial
ages of our existence, in order to construct an axe a piece of stone
was bound/knotted to a sturdy piece of wood. To make a net, vines
or creepers, animal hair, et cetera were bound/braided together. Also
it is known that the ancient Inca civilization developed a system of
characters that were formed from knotted pieces of string.
Although people have been making use of knots since the dawn of
our existence, the actual mathematical study of knots is relatively young,
closer to 100 years than 1000 years. In contrast, Euclidean geometry
and number theory, which have been studied over a considerable number
of years, germinated because of the cultural "pull" and the strong effect
that calculations and computations generated. It is still quite common
to see buildings with ornate knot or braid lattice-work. However, as
a starting point for a study of the mathematics of a knot, we need to
excoriate this aesthetic layer and concentrate on the shape of the knot.
Knot theory, in essence, is the study of the geometrical apects of these
shapes. Not only has knot theory developed and grown over the years
in its own right, but also the actual mathematics of knot theory has
been shown to have applications in various branches of the sciences, for
example, physics, molecular biology, chemistry, et cetera.
In this book, we aim to guide the reader over the multifarious as-
pects that make up this theory of knots. We shall, in a straightforward
manner, explain the various concepts that form this theory of knots.
Throughout t ~ s book, we shall concentrate on lucid exposition, and
the exercises that can be found liberally sprinkled within act as a con-
duit between the theory and the understanding of this theory by the
reader. Therefore, this book is not just another book for those who
work or intend to work within the confines of knot theory, but is also
for those engaged in other areas in which knot theory may be applied
even if they do not have a considerable background in mathematieH.
The general reader is also welcome, hopefully adding to the diversity
of knot theory. We shall cover what exactly knot theory is; what n.. ~
its motivations; its known results and applications; and what hUH I,oon
discovered but is not yet completely understood.
Knot theory is a branch of the geometry of 3 dimensions. Since
three dimensions is the limit of what is usually perceived intuitively, we
can calion this to help us explain concepts. To this end, in this book
we make extensive use of the numerous diagrams. Moreover, often the
intuitive approach is carried through into the actual text. However, the
proofs are still proven to the usual standard of mathematical rigour. In
certain cases, for the convenience of the reader, we have appended at the
end of this book several short, more detailed notes and commentaries.
Since we have tried as much as possible to avoid formal terminol-
ogy, i.e., we do not use concepts that are common in topology, such
as knot group and homology group, it has been necessary to leave out
several theorems and proofs. For the reader who is interested in a more
formal approach, a good guide is the book by Crowell and Fox [CF*].
For those interested in obtaining an even deeper understanding of knot
theory than that which may be garnered by reading this book only, we
recommend the research-level book edited by Kawauchi [K*]. Since the
purpose of the bibliography at the end of this book is to cite the theo-
rems that appear in the text, as a general bibliography for knot theory
it is inadequate. However, in Kawauchi [K*] and Burde and Zieschang
[BZ*] there can be found exhaustive bibliographies, so the inquisitive
reader who requires further references should consider consulting these
two bibliographies.
As a supplement, we include the knot diagrams of prime knots with
up to eight crossing points (35 in total), and a second table lists their
Alexander and Jones polynomials. Hopefully, this will prove of practical
use to the reader.
Finally, during the gestation period of this book I received the
valuable opinions of M. Sakuma, M. Saito and S. Yamada. Also, sev-
eral people kindly explained ideas to me that are outside my field of
speciality. The students of M. Sakuma and S. Suzuki provided many
additional, helpful comments about the original Japanese edition. To
all these individuals I express my deep gratitude. Furthermore, for the
Japanese language edition of this book I received much help from the
editorial staff of the publishers, especially from T. KameL For the En-
glish edition, the staff at Birkhauser in Boston, especially E. Beschler,
have been extremely helpful. To all these people I express my warmest
thanks.
Postscrij)taUll, even in the few years since the Japanese version was pub-
lished in 1993 have been interesting developments in knot theory. In this
iHh transhtt.ioll, we have incorporated some of these recent dev(,loptnellts.
With a reasonably long, say 30cm in length, piece of string or cord,
loosely bind a box as shown in Figure O.l(a). You should now be hold-
ing in your hands a simple type of knot. Now take the two ends and
glue them together so that it is not immediately noticeable that the
string/cord has been joined. This exercise should be performed in such
a way that the string does not come into contact with the box. The box
is more a prop than a necessity. When the exercise is completed, what
you should see before you is a single knotted loop, approximately 30cm
long, Figure O.l(b). In mathematics this loop is called a knot.
(a) (b) (e)
Figure 0.1
To be a bit more precise, this (slender) string should, ideally, be
thought of as a single curve, then a knot is a simple closed curve; in
space. If the reader is left-handed then the above knot will differ slightly
in appearance and will take the form shown in Figure 0.2.
(a)
Figure 0.2
(b)
Introduction 2
At first sight, the above two knots seem similar; however, with a
more careful perusal of these two figures, it is possible to see that they
differ in several places. These two types of knots are each called a trefoil
knot, or sometimes due to their resemblance to a clover-leaf, a clover-
leaf knot. Since the form of the trefoil knot differs in Figures O.l(c) and
O.2(b), we on occasions in order to distinguish them, refer to the knot
in Figure O.l(c) as the right-hand trefoil knot and that in Figure O.2(a)
as the left-hand trefoil knot.
To describe a knot it is not sufficient just to say a knot is the object
obtained when we bind a box or something similar with a piece of string.
We can make various types of knots that are independent of how we tie
the string. Let us create an extremely complicated knot by using a very
long piece and knotting it in the most muddled fashion we can imagine;
an example, due to Ochiai is given in Figure 0.3.
Figure 0.3
In contrast, just glue together the ends of a IOcm piece of string,
as illustrated in Figure 0.4.
Figure 0.4
This knot, which was made without knotting the string is called the
trivial knot or the unknotted knot. We invite readers to experiment and
create their own complicated knots.
Now, choose two knots from the various knots you have created in
this random manlier and call them A and B. The natural question to
3 Introduction
ask is, Can we change knot A into knot B? An approach we may try
is to determine how it is possible to change the knot in Figure O.I(b)
to that in Figure O.I(c). One condition we impose is that we cannot
cut the knot, we can only manipulate it by hand. If we can change A
into B by slowly changing the form of A, then these two knots are said
to be equivalent or equal. (This notion of equivalence is probably the
most intuitive; we give a more mathematical definition in Chapter 1,
Section 2.)
So, can we resolve the four knots in Figures O.I(b), O.2(a), 0.3,
and 0.4 into classes of equivalent ones? Before we attempt this, let us
further restrict our notion of a knot. Since what we are interested in
is the actual form of the knot, we do not need to worry about how
long or how thick the knot is. Let us mull a bit over the way we wish
to consider a knot. It is perfectly possible that a knot made from a
30cm piece of string is equivalent to a knot made from a 10cm piece of
string. Recall, we want to say that two knots A and B are equivalent
if we can manipulate, using our hands, the knot A into the knot B (to
visualize this it is best to suppose the strings have a certain amount
of elasticity so that we stretch and shrink them). So the crux of our
problem will lie in the form of the knot, not in how thick or long it is.
The reader familiar with topology will notice this discussion is just the
type of problem encountered in this branch of mathematics. To put it
plainly, knot theory, at the very least, may be considered to be a branch
of topology.
In fact, the knot in Figure 0.3 and the trivial knot depicted in
Figure 0.4 are equivalent. As our first "mathematical" exercise, we
leave it as a straightforward exercise for the reader to show by simple
manipulations that this indeed is the case. Emboldened by an early
success, we now ask ourselves, Is it true that the knots in Figures O.I(b)
and 0.4 are equivalent? The answer is "no", however, this will not be
our second "mathematical" exercise since the proof is far from simple.
It could be said that knot theory owes its present development to this
very question being asked towards the end of the 19
th
century. In other
words, one aspect of knot theory is to provide the ways and means, to
a standard of mathematical rigour, to determine whether two knots are
equivalent.
Let us try our luck again: are Figures O.1(b) and O.2(a) equivalent?
The answer unfortunately is "no", and the proof again requires Illore
than just a modicum of intuition, so we do not leave it as an
It is difficult to say who first showed a mathematical intorost in
Introduction 4
what we now call knot theory and when. However, in modern times it
is known that the famous C.F. Gauss (1777-1855) had some interest in
this field, but it was his student, Listing, who undertook research into
knot theory and gave not insignificant influence to its later development.
Originally, in honour of his accomplishments, the knot now known as
the figure 8 knot, Figure 0.5, was called the Listing knot.
Figure 0.5
The American mathematician J.W. Alexander (1888-1971) was the
first to show that knot theory is extremely important in the study of
3-dimensional topology. This was further underlined by the work of,
amongst others, the German mathematician H. Seifert from the late
1920s to the 1930s. In addition, in Germany at this time there was
already considerable activity in the study of the relationship between
algebraic .geometry and knot theory.
After the Second World War, in the 1950s, research into knot the-
ory progressed at a great pace in the United States. Under the influence
of this research, there was a great boom of research into knot theory in
Japan, which has continued to the present day. In the 1970s knot theory
was shown, among other things, to be connected to algebraic number
theory, by virtue of the solution of Smith's conjecture concerning peri-
odic mappings.
At the beginning of the 1980s, due to the discovery by V.F.R. Jones
of his epochal knot invariant, knot theory moved from the realm of topol-
ogy to mathematical physics. This was further underlined when it was
shown that knot theory is closely related to the solvable models of sta-
tistical mechanics. As knot theory grows and develops, its boundaries
continue to shift. Now, in addition, they overlap certain areas, of math-
ematical biology and chemistry. To expand briefly on this, in biology
certain types of DNA molecules have been experimentally seen to take
the form of certain types of knots. In the chapters that follow, we shall
introduce, 110I>cfully in a fairly easy and understandable manner, the
eontributioI1H of various mathematicians to knot theory and also the
relation/application of knot theory to other fields.
,,,,,tI.,,.,,,,., e,,,e,,', '1
/l",' 111""
A knot, succinctly, is an entwined circle. However, throughout this
book we shall think of a knot as an entwined polygon in 3-dimensional
space, Figure 1.0.1(a). The reason for this is that it allows us, with
recourse to combinatorial topology,
1
to exclude wild knots. For an
example of a wild knot, consider the knot in Figure 1.0.1(b). Close
to the point P, in a sense we may take this to be a "limit" point, t.he
knot starts to cluster together in a concertina fashion. in
tIle vicinity of such a point particular care needs to be taken with t.he
Chapter 1 6
nature of the knot. We shall not in this exposition apply or work within
the constraints of such (wild) knots. In fact, since wild knots axe not
that common, this will be the only reference to these kind of knots.
Therefore, in order to avoid the above peculiarity, we shall assume,
without exception, everything that follows is considered from the stand-
point of combinatorial As mentioned in the preface, our in-
tention is to avoid as ,much as possible mathematical argot and to con-
centrate on the substance and application of knot theory. Infrequently,
as above, it will be necessary, in order to underpin an assumption, to
introduce such a piece of mathematical argot. Again, as mentioned in
the preface, knowledge of such concepts will not usually be required to
be able to understand what follows.
p
ow
(a) (b)
Figure 1.0.1
This (first) chapter will be devoted to an explanation of the con-
cepts that form a foundation for the theory of knots.
1 The elementary knot moves
If we consider a knot to be polygonal in form, then since it is pos-
sible to think of it as being composed of an immense (but still finite)
number of edges, a knot is often depicted with smooth rather than polyg-
onal arcs. As the reader can see by flicking through the book, we shall
also follow this aesthetic criterion. However, mathematically, it remains
a collection of polygonal lines.
ContiIluing towards a precise (mathematical) interpretation of a
knot, it is readily obvious that we can make alterations to the shape of
a knot. For exaIIlple, it is possible to replace an edge, AB, in space of
a, knot K by two Ilew edges AC, CB. We can also perform the converse
rnplacement. Sueh replacements are called elementary knot moves. We
Hha.ll now define the possible moves/replacements.
7 Fundamental Concepts of Knot Theory
Definition 1.1.1. On a given knot K we may perform the following
four operations.
(1) We may divide an edge, AB, in space of K into two edges, AC,
CB, by placing a point C on the edge AB, Figure 1.1.1.
(1)' [The converse of (1)] If AC and CB are two adjacent edges of
K such that if C is erased AB becomes a straight line, then we
may remove the point C, Figure 1.1.1.
(2) Suppose C is a point in space that does not lie on K. If the
triangle ABC, formed by AB and C, does not intersect K, with
the exception of the edge AB, then we may remove AB and
add the two edges AC and CB, Figure 1.1.2.
(2)' [The converse of (2)] If there exists in space a triangle ABC
that contains two adjacent edges AC and CB of K, and this
triangle does not intersect K, except at the edges AC and CB,
then we may delete the two edges AC, CB and add the edge
AB, Figure 1.1.2.
\-J
A B
(1)
---.
+--
(I)'
Figure 1.1.1
(2)
---.
+--
(2)'
\-.J
A C B
Figure 1.1.2
These four operations (1), (1)', (2) and (2)' are called the elemen-
tary knot o v s ~ [However, since (1) and (I)' are not "moves" in the
usual understanding of this word, often only (2) and (2)' are referred
to as elementary knot moves. In this book we shall, by and large, also
use this interpretation.]
2 The equivalence of knots (I)
A knot is not l>erceptively changed if we apply only one elementary
knot move. However, if we repeat the process at different places, sovera.l
Chapter 1 8
times, then the resultant knot seems to be a completely different knot.
For example, let us look at the two knots K
1
and K
2
in Figure 1.2.1,
which may be called Perko's pair.
Figure 1.2.1
In appearance Perko's pair of knots looks completely different. In
fact, for the better part of 100 years, nobody thought otherwise. How-
ever, it is possible to change the knot K
1
into the knot K
2
by per-
forming the elementary knot moves a significant number of times. This
was only shown in 1970 by the American lawyer K.A. Perko.
Knots that can be changed from one to the other by applying the
elementary knot move are said to be equivalent or equal. Therefore, the
two knots in Figure 1.2.1 are equivalent.
Definition 1.2.1. A knot K is said to be equivalent (or equal) to a
knot K' if we can obtain K' from K by applying the elementary knot
moves a finite number of times.
If K is equivalent to K' , then since K' is also equivalent to K, we
say that the two knots K and K' are equivalent (or equal). We shall
denote this equivalence by K K'. Since in knot theory equivalent
knots are treated without distinction, we shall consider them to be the
same
2
knot. .
The elementary knot move (2) allows us to replace an edge AB
with the edges AC and CB. Since the points within the triangle ABC
do not intersect with the knot itself, intuitively we may rephrase Def-
inition 1.2.1 as follows: Two knots are equivalent if in space we can
alter one continuously, without causing any self-intersections, until it
becomes transformed into the other knot.
A knot has no starting point and no endpoint, i.e., it is a simple
closed curvo (to be precise a closed polygonal curve). Therefore, we can
assign an oriontation to the curve. As is the custom, we shall denote
the orientation of a knot by an arrow on the curve. It is immediately
ohviollR that any knot has two possible orientations, Figure 1.2.2.
9
(a)
Fundamental Concepts of Knot Theory
(b)
Figure 1.2.2
If two oriented knots K and K' can be altered with respect to
each other by means of oriented elementary knot moves, Figure 1.2.3,
then we say K and K' are equivalent with orientation preserved (or, for
brevity, with orientation), and we write K K'.
Figure 1.2.3
Two knots that are equivalent without an orientation assigned are
not necessarily equivalent (with orientation) when we assign an orienta-
tion to the knots.. The two knots in Figure 1.2.2 are certainly equivalent
without an orientation assigned; it is not, however, immediately obvious
whether they are equivalent with orientation.
Exercise 1.2.1. Show that, in fact, the two knots in Figure 1.2.2 are
equivalent with orientation.
3 The equivalence of knots (II)
The elementary knot moves on a knot are "local" moves or trans-
formations applied to only a small part of the knot itself. Instead of
such local modifications, we can redefine an equivalence of knots in
terms of "global" transformations/moves. These transformations move
the whole space in which the knot exists.. First, however, we need to
explain briefly a few concepts that can be found in most textbooks OIl
algebraic topology.
Let f be a map from a topological space X to a topological Hlluee
Chapter 1 10
Y. For our purposes we can restrict our attention to the cases where
X and Y are 3-dimensional Euclidean spaces or subspaces thereof. If
f is a map that is also onto and has a 1-1 correspondence, then we
may define the inverse map 1-1 : y ~ X. In addition, if both f and
/-1 are continuous maps, then the map / from X to Y is said to be
a homeomorphism, and the spaces X and Y are said to be homeomor-
phic. From the point of view of algebraic topology, spaces that are
homeomorphic may be thought as being exactly the same, i.e., without
any distinctions. In the case where X and Y have orientations assigned
to them, we say f is an orientation-preserving homeomorphism if the
original orientation of Y agrees with the orientation on Y that is the
effect of f on the orientation of X. Finally, a homeomorphism from X
to itself, Le., X == Y, is said to be an auto-homeomorphism.
3
Example 1.3.1. Suppose that both X and Yare R
2
Then the par-
allel translation along a line given by (x, y) = (x + a, y + b); a rota-
tion about some fixed point (for example, the origin) are examples of
orientation-preserving (auto-)homeomorphisms, Figure 1.3.1(a).
y
o
(a)
y
e(a,b)
f:
0
+
Z
I
.(a,-b)
(b)
Figure 1.3.1
However, the mirror "image" with respect to the x-axis given
by the homeomorphism I(x, y) = (x, -y) is a not an orientation-
preserving (auto)-homeomorphism, in fact, the orientation is reversed,
Figure 1.3.1(b) (in this figure, the effect of the map f on the y-axis is
to reverse its original orientation).
z z
~ __...
y
z
right-hand
Figllre 1.3.2
'Y
o ~ ...
left-hand
11 Fundamental Concepts of Knot Theory
There is also a natural way of assigning an orientation to R
3
, which
is done by means of the right-hand rule with regard to the xyz-axis,
Figure 1.3.2.
Definition 1.3.1. We say that two knots K
1
and K
2
are equivalent,
or K
1
is equivalent to K
2
, if there exists an orientation-preserving
homeomorphism of R3 that maps K
1
to K
2

Although we now have two definitions of equivalence, Defini-

tions 1.2.1 and 1.3.1, mathematically they are the same. A proof of this
"equivalence" is given in Kawauchi [K*]. (Nota bene, in their proof it
is not assumed that the necessary mapping is a PL-map.)
The (Euclidean) spaces in Example 1.3.1 are 2-dimensional, but it
is not hard to see that if we move up a dimension, then a rotation about
a fixed point (or fixed axis) and a parallel translation are examples of
auto-homeomorphisms of R3, which preserve the orientation of R
3

However, if we consider the mirror image, with regard to the xy-plane,

given by cp(x, y, z) = (x, y, -z), then this map reverses the orientation.
For suppose that the xy-plane is a mirror, then we may think of cp
as "reflecting" the point P in the mirror to a point P', Figure 1.3.3.
Similarly, the three axis with the right-hand rule are ''reflected'' to the
three axis with the left-hand rule. So cp does not preserve orientation.
%
Figure 1.3.3
In general, we say a map cp is a mirror image (or a symmetry)
with regard to a plane E if we can map an arbitrary point in R
3
to
its reflected point with regard to E. The image cp(K) , obtained from
the effect of the mirror image cp on the knot K, is said to be the mir-
ror image of K. If K has an orientation assigned, then we assign, in
the obvious manner, an orientation to the mirror image of K from t l ~
orientation of K.
Consider, now, two knots, K
1
and K
2
, both of which have orien-
tations assigned. If we can map K
1
to K
2
by means of an orientntjoll'"
Chapter 1 12
preserving auto-homeomorphism of R3, that does not alter the orien-
tation of either K
1
or K
2
, then K
1
and K
2
are said to be equivalent
with orientation.
The advantage of concentrating our attention on the definition of
equivalence in Definition 1.3.1, rather than the one in. Definition 1.2.1,
which, let us recall, depends on elementary knot moves, is that from
Definition 1.3.1 we may fairly immediately see, in a clear, intuitive way,
a number of knot equivalences.
Example 1.3.2. Consider cp(x, y, z) = (-x, -y, z), a 180
0
ro-
tation about the z-axis, which is an orientation-preserving auto-
homeomorphism, Figure 1.3.4. Since cp maps the oriented left-hand
trefoil knot K to K', these two knots are equivalent with orientation.
K' is the knot with the "reverse" orientation to K.
Figure 1.3.4
An auto-homeomorphism of R3 need not always move the whole
of R
3
, as was the case in our previous examples: a parallel translation
and a rotation about a fixed point (axis).
Example 1.3.3.
Figure 1.3.5
Let us fix n UIlit circle R, everything that lies "outside" this unit
13 Fundamental Concepts of Knot Theory
circle and the origin o. If we twist the "inside" of R about 0, then
this map is also an orientation-preserving auto-homeomorphism of R
2
,
Figure 1.3.5.
This type of continuous "movement" is called an isotopy. In the
above example a point P on the radius OR, rather like in a whirlpool,
has been dragged by the isotopy towards the centre.
Similarly, fix the unit sphere in R3 and everything that lies outside
the unit sphere. The map that twists the inside of the unit sphere about
the x-axis is also an orientation-preserving auto-homeomorphism of
R
3
. An example of a knot K' that has been obtained from K by such
a twist is shown in Figure 1.3.6.
z
y
Figure 1.3.6
For the same reasons as given above, these two knots are equivalent
(this map is called a twist of the ball, keeping the surface fixed).
Figure 1.3.7
Figure 1.3.7 gives us a way of seeing the 1-1 correspondence between
the 2-dimensional sphere, 8
2
, excluding the "North Pole" N, and the
whole of the plane. 80, if we add to the plane, R
2
, "a point at infinity,"
00, then R
2
U 00 and 8
2
become homeomorphic (spaces).4 Similurly,
Chapter 1 14
the 3-dimensional sphere, S3, is often thought of as R
3
with a point,
00, added at infinity. On some occasions it will be more convenient
for us to think of a knot lying in S3 rather than R
3
. (The necessary
adjustment to Definition 1.3.1 is to replace R3 by S3. We assume, for
obvious reasons, the knot K does not contain the point at infinity.)
Using this (re)definition, the following theorem can be seen to hold:
Theorem 1.3.1.
If two knots K
1
and K2 that lie in S3 are equivalent, then their
complements S3 - K
1
and S3 - K
2
are homeomorphic.
Exercise 1.3.1. Show first that the knot K, shown in Figure 1.3.8, is
the mirror image of K, K*, and second K and K* are equivalent with
orientation.
K*
Figure 1.3.8
Exercise 1.3.2. Show that the two knots K
1
and K2 in Figure 1.3.9
are equivalent. (Hint: Consider the part of the knot that lies within the
dotted circle; what happens if we twist this part?)
\ I
\ ,
, ,
, ,
, "
...... _-,
Figure 1.3.9
4 Links
So Cn,l" wo have only looked at a rathor specific, but interesting in
15 Fundamental Concepts of Knot Theory
its own right, set, namely, knots. In this section we shall look at a
generalization of this set, Le., what happens when we "link" a number
of knots together.
Definition 1.4.1. A link is a finite, ordered collection of knots that
do not intersect each other. Each knot K
i
is said to be a component of
the link.
Definition 1.4.2. Two links L == {K
1
, K
2
, , K
m
} and L' =
{Ki, ... are equivalent (or equal) if the following two con-
ditions hold:
(1) m = n, that is, L and L' each have the same number of
components;
(2) We can change L into L' by performing the elementary knot
moves a finite number of times. To be exact, using the elemen-
tary knot moves we can change K
1
to Ki, K
2
to ... ,
K
m
to (m == n). (We should emphasize that the triangle
of a given elementary knot move does not intersect with any
of the other components.)
We may replace (2) by the following (2)':
(2)' There exists an auto-homeomorphism, cp, that preserves
the orientation of R
3
and maps cp(K
1
) == Ki, cp(K
2
) =
K
2
,.. , cp(K
m
) ==
Strictly speaking, the equivalence of links should also be related to
how we order the components. In general, however, such a stringent
condition is not necessary, for we may suitably reorder the components.
Usually, therefore, (2)' is replaced by the following (2A):
(2A) There exists an auto-homeomorphism that preserves the ori-
entation of R
3
and maps the collection K
1
U ... UK
m
to the
collection Ki U ...
In this book we shall, on the whole make use of (2A) rather than
(2)'. If each component of the link is oriented, then the definitions of
are just an extension of the knot case.
Example 1.4.1. Since the two links Land L' in Figure 1.4.1 are
the same, they are equivalent. However, if we change the or<ler
of the components of L, then condition (2) of Definition 1.4.2 is not
satisfied and tIle links are not equivalent. But condition (2A) is satisfiod;
therefore, we shall eonsider them to be equivalent.
Chapter 1 16
Figure 1.4.1
Now, let us assign an orientation to Land L', Figure 1.4.2. The
addition of an orientation to the two links cause condition (2A) to no
longer hold, and hence these oriented links are not equivalent. (To prove
this we need to wait upon the definition of linking number in Chapter 4,
Section 5.)
Figure 1.4.2
Due to this, we need to take some care when considering the problem
of the equivalence of links, which is in direct contrast to the much less
delicate problem of the equivalence of knots.
As might be expected, we may extend the concept of the trivial
knot to the case of links, Le., the trivial n-component link. In the
extension to links, the relevant link consists of n disjoint trivial knots,
Figure 1.4.3.
00
0
Figure 1.4.3
There is only one trivial n-component link for each n (orientating the
trivial link has no bearing on this).
Propositions that hold with respect to knots and subsequently can
be extended to links are numerous. Therefore, in what follows, if the
examples, propositions, et cetera also hold for links, we will add "also
17 Fundamental Concepts of Knot Theory
holds in the case of links" at the end of the relevant statement or just
"knots (or links)."
Exercise 1.4.1. Showthat the two links in Figure 1.4.4 are equivalent.
This link is called the Whitehead link.
Figure 1.4.4
Exercise 1.4.2. Showthat the two links in Figure 1.4.5 are equivalent.
This link is called the Borromean rings.
Figure 1.4.5
5 Knot decomposition and the semi-group5 of a knot
We may define a sum or product operation on the set that comprises
all knots. If via such an operation the set becomes a group, then we
might be able to apply group theoretical techniques to knot theory.
Before we may try to investigate if this is possible, we must define such
operations. Therefore, in this section we shall look into how we can
obtain a single knot from two knots, called the sum (or connected sum)
of these two knots. However, since a detailed explanation is a touch
complicated, we shall delay a rigorous explanation until a bit later in
the book. To get an insight into this approach we shall concentrate on
the reverse operation, Le., we shall decompose a knot (or link) into two
Chapter 1 18
simpler knots.
In this section it will be more convenient for us to think of the knot
as lying in S3.
Let us, now, consider a sphere, S, in S3 (or R
3
) and the ball that
is bounded by S, B3, Le., the 3-dimensional ball whose boundary is S.
In the interior of B
3
take a simple curved line, a (in fact, a polygonal
line) whose endpoints A, B are on the surface S. If this curve, a,
intersects S only at the points A and B, it is called a (1, I)-tangle,
Figure 1.5.1. [We shall study generalized tangles, Le., (n, n)-tangles, in
greater detail in Chapter 9.] We should note that a (I,1)-tangle may
have disjoint simple closed curves.
We may apply the elementary knot moves to the segments of the
knotted (1, I)-tangle, a that lie in the interior of B
3
, and in so doing
suppose that we change a, having fixed A and B, to the (1, I)-tangle
shown in Figure 1.5.I(b). Such an a is called a trivial (1, I)-tangle.
Figures I.5.1(a),(b) are both examples of trivial (1, I)-tangles, while
Figure 1.5.1(c) is an example of a (1, I)-tangle which is not a trivial
(1, I)-tangle. [The ambitious reader might like to try to give an expla-
nation for why it is not a trivial (1, I)-tangle.]
(a) (b) (c)
Figure 1.5.1
Suppose K is a knot (or link) in S3. Further, let us suppose there
exists a 2-dimensional sphere, l;, that intersects (at right angles) K at
exactly two points A and B. We may perceive E to fulfill the role of
S described above. However, since K lies in S3, K is divided by E
into two (1, I)-tangles a and (3, one of which lies within E and the
other withollt, Figures I.5.2(a) and I.5.3(a). [Two (1, I)-tangles arise
because ~ is the boundary of two 3-dimensional balls, one formed from
~ and its intorior, and the other from E and its exterior. We may
think of 8
3
n ~ being made up of two (3-dimensional) balls that have
1)0<1I1 glued t ~ ~ t h e r along their l>ollndaries, namely, the 2-dimensional
19 Fundamental Concepts of Knot Theory
sphere.
6
This gluing process is more easily visualized if we drop down
a dimension. For if we take two disks and glue them along their bound-
aries, in this case a circle, we obtain the 2-dimensional sphere.] Let us
now connect A to B by means of a simple polygonal line, 8, that lies
on E. Then by joining 8 to a we obtain a knot K
1
, and by joining
s to {3, we obtain a knot K
2

(a) (b)
Figure 1.5.2
What we have shown is that a knot K can be decomposed into two
knots K
1
and K
2
, Figure 1.5.2. The choice of s is arbitrary because
if we connect A to B by means of some other simple polygonal line that
lies on E, 8', we shall once again decompose K into two knots say Ki
and K ~ It is reasonably straightforward to see that K
1
and Ki, and
K
2
and K ~ are equivalent (since we may apply the elementary knot
moves to 8 on E to change it into 8'). If one of a or {3, say, {3,
is the trivial (1, I)-tangle then K ~ is the trivial knot. In such cases
K
1
and K
2
are not, strictly speaking, a "true" decomposition of K, see
Figure 1.5.3.
", ..
, ..
/' a ~ \ A
# ,
,
.

,
,
,
~
~
....
...... t1','
...... _....
(a) (b)
Figure 1.5.3
In fact, K and K
1
are equivalent, and so we do not think of K
as being decomposed into simpler knots. When a true (non-trivial)
decomposition cannot be found for K, we say that K is a prime k11ot..
Chapter 1 20
In a sense this is equivalent to how we define a prime number, Le., a
natural number that cannot be decomposed into the product of two
natural numbers, neither of which is 1.
From the above discussion, a knot K is either a prime knot or can
be decomposed into at least two non-trivial knots. These non-trivial
knots are either themselves prime knots, or we may, again, decompose
one or the other, or both of them, into non-trivial knots. We continue
this process for the subsequent non-prime knots. The reader will be
heartened to know that this process does not continue ad infinitum.
In fact, not only is the process finite, but it also leads to a unique
decomposition of a knot into prime knots. Succinctly, this is expressed
in the following theorem.
Theorem 1.5.1. (The uniqueness and existence of a decomposition of
knots)
(1) Any knot can be decomposed into a finite number of prime
knots.
(2) This decomposition, excluding the order, is unique. That is to
say; suppose we can decompose K in two ways: K
1
,K
2
, . ,
K
m
and K i K ~ ... K ~ Then n = m, and, furthermore,
if we suitably choose the subscript numbering of K
1
, K
2
, ,
K
m
, then K
1
~ Ki,K
2
~ K ~ ... ,K
m
~ K ~
A proof of this theorem can be found in Schubert [ScI]. The above
theorem also holds in the case of links.
Let us now think about the composition that brings about the con-
verse of the above decomposition of knots. Essentially what we are
looking for is the sum of two knots. First, however, let us explain why
this "sum" is a touch more troublesome than the process of decompo-
sition. For example, if we take two links it is not at all clear which
component of these links we need to combine, so that we obtain a sin-
gle link that is their sum. Even in the case of knots, there is also a
hurdle to overcome. If we reverse the orientation on a knot we know
that it is not necessarily equivalent, via an orientation-preserving auto-
homeomorphism, to the knot with the original orientation. Therefore,
when we combine two knots their orientations become important.
We s}lall show how to overcome the hurdle for two oriented knots.
Suppose P is n point on an (oriented) knot K in S3. We may think of P
as the centro of n ball, B3, with a very small radius, Figure 1.5.4(a),(b),
that posseHSOH the following properties:
21 Fundamental Concepts of Knot Theory
(1) K intersects (at right angles) exactly two points on the surface
of boundary sphere of B3;
(2) In the interior of B3, the (1, I)-tangle, a, that is obtained
from K is a trivial tangle.
(a) (b)
Figure 1.5.4
(c)
Similarly, to some other knot K' in another 3-dimensional sphere
8
3
, we may choose a point P' and, as above, obtain from K' a trivial
(1, I)-tangle, (3, in some other ball B,3, Figure 1.5.4(b). We may in
a natural way assign orientations, from K and K' respectively, to the
(1, I)-tangles a and (3. Let 8
3
be the ball that is obtained by re-
moving from S3 the points inside of B3. Similarly, let 8,3 be the ball
that is obtained by removing from S3 the points inside of B,3. The
surface of each of these balls, i.e., 8
3
and 8,3, is a (2-dimensional)
sphere. If we now glue these two balls along this sphere, applying a
homeomorphism that reverses throughout the orientation of the sphere
of one of these balls, we obtain a 3-dimensional sphere, S3. In gluing
!)rocess the end (initial) point of a and the initial (end) point of f3
are joined. Therefore, in this 3-dimensional sphere, S3, a new, single,
oriented knot, K is formed, Figure 1.5.4(c). By construction, the ori-
entation of this Kwill not contradict the original orientation of either
K or K'.
The knot Kthat is formed in the above process is said to be the
sum of K and K' (or the connected sum), and is denoted by K#K'.
Moreover, this knot K#K' is independent of the .points P and P' that
were originally chosen. We can therefore say that K#K' is uniquely
determined by K and K'.
From the definition of the sum of knots, the next proposition f()l-
I<>ws readily.
Chapter 1 22
Proposition 1.5.2.
The sum of two knots is commutative, i.e., K
1
#K
2
K
2
#K
1
.
More concretely, K
1
#K
2
and K
2
#K
1
are equivalent with orientation.
Also, the associative law holds, K
1
#(K
2
#K
3
) (K
1
#K
2
)#K
3
.
The above definition of the sum of knots is defined on the set of
all (oriented) knots, A. However, this definition (of the sum of knots)
does not make A into a group. The reason for this is that although
the trivial knot, 0, is the unit element of A, A does not possess
inverse elements. For example, suppose K is the trefoil knot; for K it is
not possible to find a K' such that K#K' 0 (this will follow from
Theorem 6.3.5). Therefore, A is only a semi-group, not a group. This
semi-group is called the semi-group formed under the operation of the
sum of knots.
Exercise 1.5.1. Show that the two knots in Figure 1.5.5 are equiva-
lent.
Figure 1.5.5
Exercise 1.5.2. Show that the two knots in Figure 1.5.6 are not equiv-
alent. The knot in Figure 1.5.6(a) is called the square knot, while the
one in Figure 1.5.6(b) is called the granny knot.
(a)
Figure 1.5.6
(b)
23 Fundamental Concepts of Knot Theory
6 The cobordism group of knots
We know from the above discussion that the set of all (oriented)
knots is not a group under the most obvious operation, since inverse
elements do not exist. Therefore, in order to ameliorate this situation
and actually obtain a group, we may, for example, consider the following
two possibilities:
(1) a change of the definition of the sum of two knots;
or
(2) a change of the definition of the equivalence of knots (or make
it slightly weaker).
For example, the set of all integers under the action of multiplication
is only a semi-group, while under the action of addition it becomes a
group.
However, if we change the operation in such a manner for knots,
then the algebraic structure becomes changed. So, it is not really an
"improvement" of the semi-group obtained in the previous section. If,
on the other hand, we slightly weaken the definition of equality (of
knots), then perhaps the structure itself will not change considerably.
At the beginning of the 19508, J. Milnor introduced a new definition of
equivalence called corbordant. With respect to this definition, the set
of all knots does become a group under the action of the sum #. The
group itself is called the corbordism group of knots, and the knots that
become the unit element are called slice knots.
To explain the idea of a slice knot, consider S3 as the boundary of
a 4-dimensional ball :8
4
and take a knot K in S3. The knot K is called
a slice knot if it is the boundary of a disk, D, in B
4
that does not have
any singular ''points.'' To make this precise, an interior point P of a disk
I) is not a singular point if we can always choose a neighbourhood U
(llomeomorphic to a 4-dimensional ball) of P in B
4
in such a way that
the intersection 8U n D is a trivial knot in aU, a 3-sphere.
For example, by joining an interior point Qof B
4
with each point
()f a non-trivial knot K in S3, the boundary of B
4
, we can construct a
2-dimensional surface F in B4 whose boundary is K. However, Qis a
Hingular point of F, since for any neighbourhood V (homeomorphic to
a 4-dimensional ball) of Q, aVnF is equivalent to the original knot K
ill av. The trivial knot, obviously, is a slice knot, but there arn also
Illany non-trivial knots that are slice knots.
Chapter 1 24
Example 1.6.1. The square knot, shown in Figure 1.5.6(a), is a slice
knot, but the trefoil knot and the figure 8 knot are not slice knots.
At the time of writing, no methods have been found that will detect
exactly which knots are slice knots (see also Chapter 6, Section 4). It
follows from the above "definition" that the study of the cobordism
group is a problem that lies firmly in 4-dimensional topology, rather
than within the realm of 3-dimensions. Finally, let us mention, without
proof, a simple proposition concerning slice knots.
Proposition 1.6.1.
Suppose K is an oriented knot, and -K* is the mirror image of K,
with the orientation reversed. Then K# - K* is always a slice knot.
Exercise 1.6.1. Show that the knot in Figure 1.6.1 is a slice knot.
Exercise 1.6.2. Show by repeated use of the elementary knot moves in
R
4
on any knot in R
4
(to be precise a polygon in R
4
) that we can
transform any knot into the trivial knot. For this reason, our definition
of knot theory has no substance in 4-dimensions.
Figure 1.6.1
,l",f 1"61,,
Knot theory, in essence, began from the necessity to construct knot
tables. Towards the end of the 19
th
century, several mathematical ta-
bles of knots were published independently by Little and Tait in British
Rcience journals. They managed to compile tables that in total COI1-
sisted of around 800 knots, arranged in order from the simplest to
rnost "complicated." However, since these tables included, for exaUl-
pie, the two knots in Figure 1.2.1 as "distinct" knots, these tables
subsequently fOl1nd to be incomplete. However, considering tha.t
lists were cOlIlpilc(1 around 100 years ago, they are accurate to u. vPl'y
Chapter 2 26
high degree. In this chapter we shall explain two typical methods of
compiling knot tables.
1 Regular diagrams and alternating knots
Let us denote by p the map that projects the point P(x, y, z) in
R
3
onto the point P(x, y, 0) in the xy-plane, Figure 2.1.1.
p
A Y
K
Figure 2.1.1
IfK is a knot (or link), we shall say that p(K) = K is the projection
of K. Further, if K has an orientation assigned, then in a natural way K
inherits its orientation from the orientation of K. However, K is not a
simple closed curve lying on the plane, since Kpossesses several points
of intersection. But by performing several elementary knot moves on K
- intuitively this is akin to slightly shifting K in space - we can impose
the following conditions:
(1) K has at most a finite number of points of intersection.
(2) IT Q is a point of intersection of K, then the inverse image,
p-l(Q) n K, of Qin K has exactly two points. That is, Qis
a double point of K, Figure 2.1.2{a); it cannot be a multiple
point of the kind shown in Figure 2.1.2{b).
(3) A vertex of K (the knot considered now as a polygon) is never
mapped onto a double point of K. In the two examples in
Figure 2.1.2(c) and (d), a polygonal line projected from K
cornes into contact with a vertex point(s) of K, so both of
theHe cases are not permissible.
27
(a) (b) (c)
Knot Tables
(d)
Figure 2.1.2
A projection K that satisfies the above conditions is said to be a regular
projection.
Throughout this book we will work almost exclusively with regu-
lar projections, and to simplify matters, we shall refer to them just
as projections, we will draw a distinction only if some confusion might
otherwise arise. However, even if we restrict ourselves to (regular) pro-
jections, there are still a considerable number of them; secondly, and at
this juncture of quite some importance, is the ambiguity of the double
points. At a double point of a projection, it is not clear whether the
knot passes over or under itself. To remove this ambiguity, we slightly
change the projection close to the double points, drawing the projection
so that it appears to have been cut. Hopefully, this will give a trompe
l'oeil effect of a continuous knot passing over and under itself. Such an
altered projection is called a regular diagmm, Figure 2.1.3(a),(b).
(a) (b) (c)
Figure 2.1.3
A regular diagram gives us a sense of how the knot may in fact
lie in 3-dimensions, Le., it allows us to depict the knot as a spatial
diagram on the plane. FUrther, we can use the regular diagram to
recover information lost in the projection, for example, Figure 2.1.3(c)
is the projection of the two (non-equivalent) knots in Figure 2.1.3(u)
and (b).
Therefore, we need to be a bit more precise with regard to t l ~
exact nature of a regular diagram and its crossing (double) points, H i l l e ~
Chapter 2 28
from the above description a regular diagram has no double points. The
crossing points of a regular diagram are exactly the double points of its
projection, p(K), with an over- and under-crossing segment assigned to
them. Henceforth, we shall think of knots in 'terms of this diagrammatic
interpretation, since, as we shall see shortly, this approach gives us one
of the easiest ways of obtaining insight (and hence results) into the
nature of a knot.
For a particular knot (or link), K, the number of regular diagrams
is innumerable. To be more exact, there is only one regular diagram of
a knot, K, in R
3
. However, from our discussion in the previous chapter,
the knot K and a knot K' obtained from K by applying the elementary
knot moves are thought of as being the same knot. So, we can think of
the regular diagram of K' as being a regular diagram for K. Hence, it
follows that for K the number of regular diagrams is innumerable.
It is possible that a regular diagram may have crossing points of
the type shown in Figure 2.1.4(a) and/or (b).
(a) (b)
Figure 2.1.4
More generally, suppose two regular diagrams of two knots (or
links) are connected by a single twisted band; see, for example, Fig-
ure 2.1.5{a) or (b). We can, in fact, remove this "central" crossing
point by applying a twist, either to the left or right, to the knot, Fig-
ure 2.1.5(c) [in Chapter 1 Section 3 we explained how we can perform a
twist that keeps the (2-)sphere fixed]. A regular diagram that does not
possess any crossing points of this type is called a reduced regular dia-
gram.
(a) (b) (e)
Figure 2.1.5
Let us flOW take an arbitrary point P on a regular diagram D of
29 Knot Tables
a knot K and move it once round D. If P, at the crossing points of D,
is shown to, alternatively, move between a segment that passes over
and a segment that passes under, then we say the regular diagram is
an alternating (regular) diagram. Figures 1.5.5 and 1.5.6(b) are ex-
amples of alternating (regular) diagrams, while Figure 1.5.6(a) is an
example of a non-alternating diagram. A knot that possesses (at least
one) alternating diagram is called an alternating knot. These types of
knots have great importance in knot theory, since many of their char-
acteristics are known. (For a more detailed discussion, see Chapter 11,
Section 5.) The trivial knot is also an alternating knot (we leave it as
an exercise to explain why this is the case.) Many "simple" knots are
alternating knots. Therefore, that is to say, in the nascent years of knot
theory, all knots were thought to be alternating knots. The simplest
non-alternating knot, in fact, is a knot with 8 crossing points shown in
Figure 2.1.6. However, it is by no means trivial to prove that we can
never find an alternating diagram for this knot. (For further details, see
Chapter 7.)
Figure 2.1.6
Exercise 2.1.1. Show that a regular diagram for K
1
#K
2
can be
obtained by placing the regular diagrams of the oriented knots K
1
and
K
2
side by side, and connecting them by means of two parallel segments,
Figure 2.1.7.
Figure 2.1.7
Exercise 2.1.2. The definition of an alternating link follows directly
from the definition of an alternating knot. Divide the knots and IiUI<H
Chapter 2 30
that we have discussed so far into those with alternating diagrams and
those that have non-alternating diagrams.
Exercise 2.1.3. Figures 1.5.6(a) and 2.1.7 are non-alternating dia-
grams for their respective knots. However, both of these knots are al-
ternating knots. Show that they do possess alternating diagrams. (Hint:
They have 6 and 7 crossings, respectively.)
Exercise 2.1.4. (Taniyama) Let K
1
and K
2
be alternating knots.
Suppose that they have alternating diagrams with nt and n2 crossing
points, respectively. Show that the connected sum of K
1
and K2 has
an alternating diagram with exactly nl + n2 crossing points.
2 Knot tables
A table of reduced regular diagrams of knots may be thought of
as a knot table. So, let us think how we may ascribe some sort of
code/index system to these regular diagrams of knots. This aim is far
from new. Gauss, probably as a recreation, devised one such code.
Although other coding systems have been created, we shall describe,
with a slight enhancement, the system due to Gauss [DT].
Suppose that a regular projection K of a knot K has n crossing
points, {PI, P2, . . ,Pn}. Each crossing point Pi of Kis the projective
image of exactly two points and of K, Figure 2.2.1.

...............
:p!'
.'

Figure 2.2.1
Now, starting with an arbitrary point P of K, move around K in
a fixed direction (if K already has an orientation assigned, then follow
K along this orientation). When we first arrive at a point or
assign rUlIul>er 1 to this point. MovillE?; on from the point or
31 Knot Tables
, when we arrive at the next point Pj or P'J, assign the number
2 to this point (it is quite possible that we may assign the number 1
to and the number 2 to In this way, we may assign to the
2n crossing points of K, {Pi, ... the numbers 1 to
2n, Figure 2.2.2.
Figure 2.2.2
Due to this, we may assign two numbers to a point Pk of K, i.e.,
the projection of the points P
k
and P/:. From these sets of pairs, (i,ji)
for each point Pk of K, we obtain a collection of 2n pairs of numbers,
We shall rewrite these 2n pairs of numbers in the form of a permu-
tation, Le.,
2 3 ... 2n)
j2 j3 j2n
The sign + or - in front of ji obeys the following condition:
a "+" is assigned if the point of K that has the integer i
assigned to it is above the point that has the the integer
ji assigned to it. If it is below, then we assign" -. "
(2.2.1)
For example, the permutation corresponding to the knot in
ure 2.2.2 is
(
1 2 3 4 5 6)
4 -5 6 -1 2 -3
Chapter 2 32
Example 2.2.1. The permutations that are obtained from the regular
diagrams in Figure 2.2.3(a) and (b) are, respectively,
and


2 3 456 7
-7 6 -1 8 -3 2
2 3 4 5 6 7
7 -6 -1 8 3 -2
(a) (b)
Figure 2.2.3
If we look closely at the permutations, then the following observa-
tions are almost immediate. Firstly, in the pair (k, jk) one integer
is always even and the other odd. (Why is this the case? Hint: Con-
sider the Jordall curve theorem.
6
) Also, if the pair (k, jk) is part
of the permutation, then the inverse pair, (jk' =Fk), is also part of the
permutation. Therefore, if we know the pairs that have an odd k, au-
tomatically the pairs that have an even k are also known. So, it is
sufficient to consider only the permutations of odd-numbered pairs that
originally comprised exactly half of the permutation. This now allows
us to write down the following series, a row of n even numbers:
(jl, j3, js,, j2n-l).
This series is the code assigned to (a regular diagram of) K.
Example 2.2.1. (continued) The code for the knot Figure 2.2.3(a) is
(4,6,8,2), while for knot Figure 2.2.3(b) it is (4, -6,8, -2).
Exercise 2.2.1. Show that if all the sigrls in a given code agree, then
it is a code of nn alternating diagrfll11; show that the converse also llolds.
33 Knot Tables
In the link case, choose one of its components and assign numbers
to this component in the manner described above. (As in the knot
case, if the component is oriented, then follow its orientation; otherwise,
the choice of direction in which we traverse the component is left to
the reader's discretion.) Next, choose another component and repeat
the above process, and continue until all the components have been
traversed and numbers assigned. In fact, if the starting points on each
component are suitably chosen, we may assign to each crossing point an
even number and an odd number. (Why is it possible to choose such
a starting point?) Hence, we may write down for each component a
row of even numbers in a similar manner as in the knot case. So, each
component will have a code assigned to it, and the sequence
(jl, j3, .. ,j2k-l I ll, l3, ,l2m-l I ...)
is the code of a link (diagram). The sign in front of the ji, ii, ... is
determined in exactly the same way as in the case of knots. The symbol
I between the row of ji and li signifies that at this point the row of
even numbers for the first component comes to an end.
Example 2.2.2. The code for the (regular diagram of) Borromean
rings in Figure 2.2.4 is (-6, -81 -12, -10 I -2, -4).
Figure 2.2.4
Exercise 2.2.2. Determine the codes for the square knot, the granny
knot, and the Whitehead link. Their regular diagrams were given in
Chapter 1.
We now have a method of assigning a code to a given knot K.
However, we immediately encounter a couple of problems. Firstly, the
code depends on the starting point, and, secondly, a knot, K, has all
abundance of regular diagrams. Hence each K has an abundanen of
codes. Unfortunately, there is no known method to decide
Chapter 2 34
or not two codes correspond to equivalent knots. However, we can
determine whether or not a finite row of even integers is a code for some
knot [DT].
Exercise 2.2.3. Suppose a sequence (al' a2,' .. ,an) is a code of a
knot K. Show that the same sequence can be a code for the mirror
image of K.
Exercise 2.2.4. Find all knots or links that have the following codes:
(a) (4,8, -12,2,14,16, -6, 10)
(b) (6,8,22,20,4, -16, -26, -10, -24, -12,2, -14, -18)
(c) (6, 10,2, -12 14, -8)
Exercise 2.2.5. Show that there cannot exist a knot with the code
(8, 10, 2,4,6).
Exercise 2.2.6. Use the code of a knot to show that the number of
knots and links that have regular diagrams with n crossing points is at
most 2
n
n!
Let us now consider reduced regular diagrams with exactly n cross-
ing points. Suppose that A(n) is the number of prime (unoriented)
knots that have a regular diagram with n crossing points, but none
with fewer crossing points than n. If we do not distinguish between
a knot K and its mirror image K*, i.e., we count them as the same
knot, then we know from Exercise 2.2.6 that A(n) 2
n
n! In fact, A(n)
is quite a bit smaller than this upper bound. However, at the time of
writing, there is no known method to determine the exact value of A(n).
At present, the following values for A(n) are known:
n 0 1 2 3 4 5 6 7 8 9 10 11 12 13
A(n) 1 0 0 1 1 2 3 7 21 49 165 552 2176 9988
It is natural, of course, that as n increases, A(n) begins to increase
rather rapidly. Actually, it was only a few years ago that it was proven
that if n is large, then at the very least A(n) is bigger than n
2
[ES1].
Before this result was announced, basically all that could be said was
that A(n) 1 for large n!
3 Knot graphs
Let us first explain Tait's method for knots. Suppose that D is
a regular diuJ;!;rarrl for a knot K and K iR a projection of K. We can
35 Knot Tables
think of K as a graph on the plane. (We shall explain the concept of
a graph in a more detail in Chapter 14. For the present, the image we
wish to use is that of the usual plane graph, Le., composed of vertices
and edges on the plane.) The vertices of the graph correspond to the
crossing points of K.
In Figure 2.3.1 we have drawn a couple of plane graphs obtained
from the two Ks in that figure.
(a) (b)
Figure 2.3.1
As can be seen from the above figures, K divides the plane into
several domains. Starting with the outermost domain, we can colour
the domains either black or white. By definition, we shall colour the
outermost (unbounded) domain black. In fact, we can colour the do-
mains so that neighbouring domains are never the same colour, i.e., on
either side of an edge the colours never agree. (Why is it possible to
colour domains in this manner?) Next, choose a point in each white
domain; we shall call these points the centres of the white domains.
If two white domains Wand W' have the crossing points (of K),
Cl, C2,.'" Cl, in common, then we connect the centres of W and W'
by simple arcs that pass through Cl, C2,' .. ,Cl and lie in these two white
domains (other than at the centres of Wand W', these arcs do not
intersect each other). In this way, we obtain from Ka plane graph G.
The vertices of G are the centres of the white domains.
(a)
Figure 2.3.2
(b)
Chapter 2 36
Example 2.3.1. The plane graphs G of, respectively, Figure 2.3.2(a)
and (b), are those obtained by the above method for the knots in Fig-
ure 2.3.1(a) and (b).
However, in order for the plane graph to embody some of the char-
acteristics of the knot, we need to use the regular diagram rather than
the projection. So, we need to consider the under- and over-crossing at
a crossing. To this end, in Figure 2.3.3 is shown a way of assigning to
each edge of G either the sign + or
(a) (b)
Figure 2.3.3
A +sign is assigned to an edge e if the domains are coloured in the
manner of Figure 2.3.3(a), and a - sign if they are as in Figure 2.3.3(b).
A signed plane graph that has been formed by means of the above
process is said to be the graph ~ K. (To be precise, it is called the
graph that is formed from the regular diagram D of a knot K.)
Example 2.3.2. In Figure 2.3.4 we have drawn the signed plane
graphs that correspond to the respective regular diagrams in that figure.
(a)
Figure 2.3.4
(b)
37 Knot Tables
Conversely, we can construct from an arbitrary signed plane graph G a
knot (or link) diagram; see Figure 2.3.5.
To construct the subsequent knot, first place a small "x" at the
centre of each edge of G, Figure 2.3.5{b). From the endpoints of one of
these of "x," draw four lines that follow along the edges of G until they
reach the endpoints of a neighbouring "x." What should start to slowly
appear if this process is carried out at each "x" is a projection of a knot,
but with no information with regard to the nature of the crossing points,
Figure 2.3.5(c). Now, we can colour the planar domains (obtained from
the partition of the domain by the newly-formed projective diagram)
either black or white using the same method to decide which colour
to apply as discussed previously. We may ascertain, and hence draw
in, the relevant crossing point information from the signs of the original
graph. Obviously, the black and white colouring information disappears
once the crossing point information is added, and hence we obtain the
required knot diagram, Figure 2.3.5(d).
(a) (b) (c) (d)
Figure 2.3.5
Therefore, for each signed plane graph there exists a corresponding
(regular diagram of a) knot. However, it is not necessarily true that two
different plane graphs give rise, by means of the above process, to two
non-equivalent knots. At the time of writing, no method has yet been
found to determine whether or not the two knots are equivalent.
The above approach was originally one of the methods used to
construct a table of regular diagrams of all knots starting with graphs
with a relatively small number of edges and then increasing the nunlber
of edges. In this manner, Tait and Little produced an almost conlI>lete
table of regular diagrams of knots with up to 11 crossing POiIlt",S. In
recent years this table has been amended and increased to include kuots
Chapter 2 38
with up to 13 crossing points. [Appendix (I) is a complete table of all
(prime) knots with up to 8 crossings.]
In Figure 2.3.6 we have placed in juxtaposition the connected plane
graphs with up to 4 edges and their corresponding knots (and links).
The number of edges is equal to the number of crossing points of the
regular diagram of the knot. Since, for the sake of clarity, we have not
assigned signs to the edges, these figures are not regular diagrams of the
knots (or links) but rather their projections.
Graph Knot or Link Graph Knot or Link
o CD
o
e
Figure 2.3.6
Exercise 2.3.1. Show that a regular diagram that is also an alter-
nating diagram corresponds to a graph G with the same sign on all the
edges. Moreover, show that these are the only possible kind of graphs.
Exercise 2.3.2. Why may we not think of an edge, e, as shown in
Figure 2.3.7, to be an edge of a graph.
>----. or
Figure 2.3.7
39 Knot Tables
Exercise 2.3.3. List all the knots (and links) that correspond to con-
nected plane graphs that have 5 and 6 positive edges. Moreover, deter-
mine which of these knots are equivalent.
To create a knot (or link) table, it is sufficient to create a table
of prime knots (or links). A table of non-prime knots can be created
directly from the table of prime knots. However, there is no known
method that allows us to create a table of only prime knots. Moreover,
at the time of writing, there is no known method to determine whether
or not a given knot is prime (see also Chapter 3, Section 2).
,,,,,1111,,.',,'11' ",6',,,., "
,l"" 111""
The problems that arise when we study the theory of knots can
essentially be divided into two types. On the one hand, there are those
that we sllall call Global problems, while, in contrast, there are those
that we shall call Local problems.
Glohal l>foblems concern themselves with how the set of all knots
behaves. As the label implies, in contral>osition, Local problems are con-
cerned with the exact nature of a knot. As to the question which
is U10l'P illlportant, and Hhould coneentrate our att.ention
41 Fundamental Problems of Knot Theory
tention on, the unhelpful answer is that it is impossible to say. In order
to solve Global problems it is often necessary to find solutions to various
Local problems. Conversely, the determination of Local problems may
rely on how they fit within the Global problem.
In this chapter, we shall explain and give examples of these two
types of problems. Problems in the theory of knots are not just limited
to this bifurcation into Global and Local problems. However, in the past
the above dichotomy has formed the axis around which knot theory has
developed, and it is more than likely that this will substantially remain
the case in the foreseeable future.
1 Global problems
One of the typical classical Global problems is the classification
problem.
(1) The classification problem
The classification problem, at least in definition, is very straightfor-
ward, as the name suggests we would like to create a complete knot (or
link) table. What exactly we mean by a complete table is one in which,
firstly, no two knots are equivalent, and, secondly, a given arbitrary knot
is equivalent to some knot in this table.
At the time of writing, a complete table in the above strict sense
has been compiled only up to prime knots with 13 crossings. One future
problem is to steadily expand this table. Another (sub-)problem that
germinates directly from the original classification problem is to create
a complete table for only certain specific types of knots, for example, for
alternating knots. As we introduce other types of knots, this question
of whether we classify them completely will always be in the vanguard
of the questions that we will ask ourselves. In fact, in Chapters 7 and
9 we shall discuss two specific knot types that have been completely
classified.
(2) A fundamental conjecture
This conjecture can be immediately stated as follows:
If 8
3
- K
1
' and 8
3
- K
2
, which are usually called comple-
mentary spaces, for two knots K
1
and K
2
, respectively, /l["C
homeomorphic, then the knots are equivalent.
rrllis conjecture can readily be seen to be the converse of Theorerll 1.:J. 1.
Chapter 3 42
Figure 3.1.1
Example 3.1.1. Although the two links in Figure 3.1.1 are not equiv-
alent, their complementary spaces are homeomorphic.
7
In the late 19808 this conjecture was, in fact, proven by C. McA
Gordon and J. Luecke [GL]. As a consequence of this result, the problem
of knots in S3 transforms itself from what we may call a relative problem
which concerned itself with the shape of a knot in 8
3
, into an absolute
problem, which now concerns itself with the study of the complementary
spaces.
However, much to our dismay we cannot always transform a relative
problem into an absolute problem. The counterexample that immedi-
ately comes to hand is that, in fact, the above fundamental conjecture
is false in the case of links.

1

j
a
j.t

::\

.!:
l
.J;

'1

.i
J
::j


i

In general, results that hold for knots pass through fairly readily to i
hold for links as well. However, as the above example shows, we cannot
take this for granted.

1,
.tl
(3) Knot invariants ;}
:.
As a way of determining whether two knots are equivalent, the ]
,1
concept of the knot invariant plays a very important role. The types of :,
'M
knot invariants are not just limited to, say, numerical quantities. These
knot invariants can also depend on commonly used mathematical tools,
such as groups or rings.
Suppose that to each knot, K, we can assign a specific quantity ;:i
p(K). If for two equivalent knots the assigned quantities are always
equal, then we call such a quantity, p(K), a knot invariant. This con-
cept of assigning some mathematical quantity to an object under in- ,
vestigation is not limited just to knot theory, it can be found in many
branches of rnathematics. Probably the sinlplest analogous example oc-
curs in Kroul> theory. The of elenlcnts in a group, called the
43 Fundamental Problems of Knot Theory
order of the group, is a group invariant, since for isomorphic groups
their respective orders are equal.
We know that if a knot K and another knot K' are equivalent,
then it is possible to change K into K' by applying the elementary knot
moves to K a finite number of times. Therefore, for a quantity p(K)
to be a knot invariant, p{K) should not change as we apply the finite
number of elementary knot moves to the knot K. It follows from this,
for example, that the number of edges of a knot is not a knot invariant.
The reason is that the operations defined in Definition 1.1.1(1) and (1)'
either increase or decrease the number of edges. Similarly, if we consider
the operations in (2) and (2)' of the same definition, then it also follows
that the size of a knot is not a knot invariant.
A knot invariant, in general, is unidirectional, i.e.,
if two knots are equivalent their invariants are equal.
For many cases the reverse of this arrow does not hold. In contra-
position, if two knot invariants are different then the knots themselves
cannot be equivalent, and so a knot invariant gives us an extremely ef-
fective way to show whether two knots are non-equivalent. The history
of knot theory may be said to be an account of how the various knot
invariants were discovered and their subsequent application to various
problems. To find such knot invariants is by definition a Global problem.
On the other hand, to actually calculate many of these knot invariants,
which we shall discuss in Chapter 4, is quite difficult. Further, to find
a method to calculate these invariants is also a Global problem.
2 Local problems
To illustrate and explain the idea of a Local problem, we shall give
several examples.
(1) When are a knot K and its mirror image K* equivalent?
If K and K* are, in fact, equivalent, then we say that K is an
amphicheiral knot (sometimes also referred to as an achiral knot). For
example, since the right-hanq trefoil knot [Figure O.1{b)] and its mirror
image, the left-hand trefoil knot (Figure 0.2), are not equivalent, the tre-
foil knot is not amphicheiral. On the other hand, however, the figure 8
knot is amphicheiral (cf. Exercise 1.3.1). Due to the extremely special
nature of amphicheiral knots, there are, in relative terms, very few of
them. This (Local) problem has over the years been quite exteJ1Hivoly
Chapter 3 44
studied, and for particular types of knots many amphicheiral results
have been proven. (For further details see Chapter 7 and Chapter 9,
Section 3.)
(2) When is a given knot prime?
In the way described in Exercise 2.1.1 (Figure 2.1.7), a regular dia-
gram of K
1
#K
2
, the connected sum of K
1
and K
2
may be constructed
by placing the regular diagrams of K
1
and K
2
side by side and then
connecting them by means of two parallel segments. Therefore, if a knot
K can be decomposed into K
1
and K
2
, then K has a regular diagram of
the type shown in Figure 2.1.7. However, although theory predicts this
in practice, since most regular diagrams of non-prime knots are usually
not so nicely presented, we cannot deduce from the regular diagram
whether a knot is prime.
Example 3.2.1. The regular diagram of the knot, K, shown in Fig-
ure 3.2.1(a) is not of the form of Figure 2.1.7, but K is not a prime knot.
(a) (b)
Figure 3.2.1
Recently, this (Local) problem has been completely resolved in the
case of alternating knots (cf. Chapter 11, Section 5).
(3) When is a knot invertible?
We know that we can assign to a knot two different, opposite orien-
tations. Let us denote one of these knots by K and the other, with the
opposite orientation, - K. We would like to determine whether K and
- K are equivalent. When K and - K are, in fact, equivalent, then K is
said to be invertible. Knots with a relatively small number of crossing
points are in general invertible. It follows from Example 1.3.2 that the
left-hand trefoil knot is an example of an invertible knot.
That non-invertible knots do exist Wf1.. '; first shown by H.F. Trotter
in 1963.
8
'rhe knot in Fig\lre :1.2.2(n.) WUH tIle example that was given
45 Fundamental Problems. of Knot Theory
by Trotter; following this discovery, many other non-invertible knots
were soon found.
(a) (b)
Figure 3.2.2
In contrast to 1963, it is now fair to say that almost all knots are non-
invertible. We have drawn in Figure 3.2.2{b) the simplest non-invertible
knot.
(4) What is the period of a knot?
If we rotate the figure 8 knot, Figure 3.2.3(a), by an angle of 1r
about the Oz-axis, the figure will rotate to its original form. So, this
knot may be said to have period 2. The left-hand trefoil knot, Fig-
ure 3.2.3(b), if it is rotated by 2; about the Oz-axis, will also rotate
to its original shape. In general, if we can rotate a knot by an angle
2: about a certain axis so that it rotates to its original shape, then we
say that this knot has period n. In this case, the (Local) problem is to
determine all the periods for a given knot. This problem has, also, been
extensively studied and has been completely solved for particular types
of knots (cf. Chapter 7).
(a)
Figure 3.2.3
(b)
Chapter 3 46
(5) When is a knot a slice knot?
Of all the (Local) problems that we have so far discussed, this is
probably by far the most difficult. The present state of affairs is that
only several necessary conditions are known for a knot to be a slice
knot. Further, effective methods to determine slice knots are also not
known. Therefore, this (Local) problem seems at the moment to be
quite intractable.
The subsequent chapters will be an exposition of knot theory, which
will take their bearings from the bifurcation of knot theory problems
outlined in this chapter, namely, the Global and Local problems.
A knot (or link) invariant, by its very definition, as discussed in
the previous chapter, does not change its value if we apply one of the
elementary knot moves. As we have already seen, it is often useful to
project the knot onto the plane, and then study the knot via its regular
diagram. If we wish to pursue this line of thought, we must now ask
ourselves what happens to, what is the effect on, the regular diagralll
if we perform a single elementary knot move on it? This question wa.s
studied by K. Reidemeister in the 1920s. In the course of time, InallY
krlot invariants were defined from Reidemeister's seminal work. III t.hiH
Chapter 4 48
chapter, in addition to discussing these types of knot invariants, we shall
also look at knot invariants that follow naturally from what one might
say is mathematical experience.
1 The Reidemeister moves
A solitary elementary knot move, as might be expected, gives rise
to various changes in the regular diagram. However, it is possible to
restrict ourselves to just the four moves (strictly speaking, changes)
shown in Figure 4.1.1 and their inverse moves, Theorem 4.1.1.
J
J
J
--+
.--
--+
.--
--+
.--
>

>
Figure 4.1.1
That these moves may, in fact, be made is reasonably straightfor-
ward to understand. For example, 0
1
may be thought of as the move
that corresponds to an elementary knot move on a regular diagram,
which replaces AB by AC U CB, as shown in Figure 4.1.2.
--+
Fip;uro 4.1.2
49 Classical Knot Invariants
Exercise 4.1.1. Verify that, in fact, O
2
and 0
3
are possible (Le.,
they are a consequence of some finite sequence of elementary knot
moves).
Example 4.1.1. The sequence of diagrams in Figure 4.1.3
.J
\
",.,- .. ~ ~ ~
, ,
. ,
. .
, .
~ ~ J /
V"
Figure 4.1.3
shows that the deformation
can be obtained as a sequence of Reidemeister moves.
Exercise 4.1.2. Show that the four deformations, and their inverses,
in Figure 4.1.4 can be obtained as a sequence in the Reidemeister moves
no, n
1
, n
2
, 0
3
(and n ~ and their inverses.
0'
J
-..

2
+--
0'
,
>(
-..
~
3
/
+--
~ "
Oil
:k
-..
~
3
/ I ~
+--
/r
0
3
k
-..
~
'"
+--
~ ~
I
Figure 4.1.4
Chapter 4 50
An obvious deformation like no is one of the (plane) isotopic de-
formations defined in Chapter 1, Section 3, see Figures 1.3.5 and 4.1.5.
J
Figure 4.1.5
Under these isotopic deformations, which differ in their nature from
n
1
, O
2
, or 0
3
, D remains essentially unchanged. Therefore, we can
apply plane isotopic deformations quite freely to any place on the regular
diagram, as long as D has a sufficient number of segments. To make
sure that we have enough such segments, it may be necessary first to
add a number of vertices to D. The addition (or elimination) of vertices
on D corresponds exactly to the elementary knot move (1) [or (1)']
given in Definition 1.1.1. By the same reasoning, (1) and (1)' should
not really be considered as moves (or changes), so we shall not classify
them as an integral part of the moves. Keeping these remarks in mind,
we can now define an equivalence between two regular diagrams D and
D' of knots K and K'.
Definition 4.1.1. If we can change a regular diagram, D, to another
D' by performing, a finite number of times, the operations {ll, n
2
, {13
and/or their inverses, then D and D' are said to be equivalent. We
shall denote this equivalence by D D'.
These three moves n
1
, O
2
, {13 and/or their inverses are called the
Reidemeister moves. Due to the above, we may state the following
theorem.
Theorem 4.1.1.
Suppose that D and D' are regular diagrams of two knots (or links)
K and K', respectively. Then
K K' <==> D D
'
.
We may conclude, from the above theorem, that the problem of
equivalence of knots, in essence, is just a problem of the equivalence of
regular diap;rams. Therefore, a knot (or link) invariant may be thought
of as a quuntity that remains unchanged when we apply anyone of the
above Reidorneister moves to a regular diagram. In the following, we
shall often nood to perform locally a. finite number of times a composition
51 Classical Knot Invariants
of Reidemeister moves (or plane isotopic deformations), for simplicity
we shall call such a composition an R-move.
)
(at)
'---J

(bl)
or
(a2)
(b2)

- >or X> or X
-v'\ -V, / '\
(cl) (c2) (c3)
(al)
Figure 4.1.6
Lemma 4.1.2. The moves shown in Figure 4.1.6 are R-moves.
Proof
We prove two of the cases diagrammatically in Figure 4.1.7. The
other cases we will leave for the reader to prove along similar lines to
Figure 4.1.7.
I
Figure 4.1.7

Chapter 4 52
Proof of Theorem 4.1.1.
In order to simplify the notation we shall restrict the proof to the
case of a knot, K (in the link case the idea and sequence of the proof is
completely the same). Suppose that K' is obtained by replacing AB of
K by the two edges ACU CB of We may then assume that the
regular diagram D' of K' is obtained from the regular diagram, D, of
K (if necessary, we may need to shift the slightly). Therefore,
we need to consider the two cases in Figure 4.1.8(a) and (b). (Generally,
the internal part of will contain many line segments of D.) Here,
we shall only prove the first case and leave the second case as an exercise
for the reader.
(a)
b
(b)
Figure 4.1.8
For the sake of clarity, we shall denote by the projection,
of We shall now show that we can change W = ac U cb to the
segment ab by repeatedly using R-moves.
Let Do = D - abo It follows immediately that Do and D' - {acU
cb} are equal. It also follows readily that Do is a polygonal curve on
the plane that has as its endpoints a and b. In the trivial case, when
Do and the internal part of do not intersect, we may perform the
R-move n
o
1
to W to obtain the segment abo
Consider, now, the case when a part of the polygonal curve Do is
in the internal part of this is now no longer trivial and we need
to actually our hands slightly dirty. Suppose that the number of
crossing points of Do in the interna.l part of a is m.
First, hy iu(luction on these TTl, erossiIlg points, we will find, by
applying Il.-Illoves to W, .. Hitnplo polygonal curve W' on the
53 Classical Knot Invariants
plane. This (polygonal) curve W' will have as its endpoints the same a
and b. In addition, inside the polygon formed from W' and the segment
ab there will now be no crossing points of Do. (Further, the only points
where W' intersects with the segment ab is at the endpoints a and b.)
To find such a W' we need the following definition.
Definition 4.1.2. A (not necessarily simple) polygonal curve a in Do
that lies in the internal part of d (or, more generally, in the internal
part of a polygon E on the plane) can be divided into two types:
a enters the internal part of E by passing over (or in the second
type under) at a point P and then exits the internal part of E by
passing over (or in the second type under) at a point Q. Such an a (in
relation to E) is called an overlying (or in the second type, underlying)
polygonal curve, Figure 4.1.9.
9
"
underlying
Figure 4.1.9
Lemma 4.1.3. The polygonal curves of Do that are in the internal
part of d are necessarily eitber overlying or underlying, (that is to say;
there cannot exist a polygonal curve of Do that enters the internal part
of d passing over d and the exits by passing under d). Further,
at the intersection, in the internal parts of d, of an overlying polyg-
onal curve a and an underlying polygonal curve {3, a always passes
above (3.
Proof
Let us consider which, in fact, is a (infinitely long) trian-
gular prism T in R3, Figure 4.1.10. This triangular prism is divided
by dABC into an upper prism T1 and a lower prism T2.
So, the question now follows: In which of these two prisms T1 or
rr
2
does p-l(a) n K lie, where p-l(a) is the inverse image of the
polygonal curve a of Do that lies in the internal part of d? Note that
since K and the internal part of dABC do not intersect, p-l(a) n K
lie in both prisms. If p-l(a)n K lies in T
1
, then a is overlyinp;,
aIld if it lies in T
2
, it is underlying. This proves the first part of the
Chapter 4 54
lemma. The latter half of the lemma also follows from the above.
C

Figure 4.1.10




,.1\
;\;
h
.Suppos1e, now'fthAat NCo is a SinghIe crlOSSin
d
g
point of D
II
0 . thlat in
t e lnterna part 0 u. ext, on t e p ane raw a sma Circ e I.e., a
simple closed polygonal curve) M whose centre is CO' From W let us

take a point P and from M a point Q, and further let us also choose
a simple polygonal curve 1, whose endpoints are P and Q. These are
chosen in such a way that the following conditions are satisfied:
(1) P and Qare not vertices of either W or M, respectively;
(2) P and Q are not points of Do (i.e., P and Qare not crossing ,..
points of the regular diagrams D, D';
(3) l may intersect with Do at a finite number of points (at right
angles); however, it cannot pass through a crossing point of ii
Do;
(4) 1 does not intersect with the segment abo
We can always find points P and Q and a polygonal curve l, see .;.::

Figure 4.1.11. :Xi
l
':i1:

FiKIU"P 4.1.11
55 Classical Knot Invariants
....
I
In the obvious manner make 1 slightly wider, Le., we create a band
P/Q/Q"P". Our next move is to replace, by applying R-moves, the edge
P'P" by the other three edges of the band, II = pIQ' UQ' Q" UQ"P",
Figure 4.1.11. To achieve this, firstly, if say, ai, is the first overlying
(underlying) polygonal curve that intersects 1 and Do, then use the
R-move (bl) [or (b2)] of Figure 4.1.6 to change P'p" into P
/
Pi Pl"P",
Figure 4.1.12(a) [or (b)].
(a)
(b)
Figure 4.1.12
In a similar way, by repeating the process for the next intersection
of l with a polygonal curve say, a2, of Do, we can replace
by PIP2P2"Pl". In this manner, we shall, finally, replace by
PkQ/Q"Pk", and, so, we shall have completed the process of changing
p'p" into ,1, where k is the number of points of intersection between
Do and l. Next, replace Q/Q", which is a part of M, by M- Q'Q" (=
Q'SQ"). Depending on the type of polygonal curve, f31 and f32, of
Do that intersects with co, Figure 4.1.11, we shall need to perform the
changes as described below. If both {31 and {32 are overlying, then
apply the R-move (cl) shown in Lemma 4.1.2. If {31 is overlying and
{32 is underlying, then apply (c2). Finally, if {31 and {32 are both
underlying, then apply (c3).
The outcome of the above is that we replaced pIp", a part of W, by
another simple polygonal curve, ... ... Pl"P", which
we shall denote by W1. This W1 does not intersect with the segment
abo Therefore in the internal part of the polygon formed by WI and
ab, the number of crossing points of Do has decreased by 1 to m - 1.
Further, by the above operations, the polygonal curve of Do that lies
in the internal part of is divided into several polygonal curves. The
types of the polygonal curves that remain in the internal part of 1 are
all the same as the original types of the polygonal curves.
We now repeat the above operations on the crossing points of Do
that are in the internal part of At the end of this lengthy process,
we shall reach a simple polygonal curve Wm that has as its endl)()jllts
Chapter 4 56
a and b. In the polygon formed from ab and Wm, there are no
crossing points.
Figure 4.1.13
By the above remarks, since the type of polygonal curves of Do that
are in Em are either overlying or underlying, then will be like in
Figure 4.1.13.
In order to complete the proof of the theorem, we need to change
W
m
to the segment abo This is done by repeatedly using R-moves given
in Lemma 4.1.2. This final part we leave as an exercise to the reader.

In the next few sections, we shall explain several knot invariants

that have played a substantial role in research into knots.
2 The minimum number of crossing points
A regular diagram D of a knot (or link) K has at most a finite
number of crossing points. However, this number c(D) is not a knot
invariant. For example, the trivial knot has two regular diagrams D
'and D', which have a different number of crossing points, Figure 4.2.1.
c(D) = 0
Figure 4.2.1
C><)D
1
c(D') = 1
Consider, instead, all the regular of K, and let c(K) be the
minimUIn ulllnber of crossing poiuts of all the regular diagrams. This
c(K) is a knot invariant.
57
Theorem 4.2.1.
Classical Knot Invariants
c(K) = min c(D)/
'D
is a knot invariant, where V is the set of all regular diagrams, D, of K.
The above quantity is called the minimum number of crossing
points of K. A regular diagram of K that has exactly c(K) crossing
points is said to be the minimum regular diagram of K. For example, if
K is a trivial knot, then c(K) = O.
Proof of Theorem 4.2.1.
Suppose that Do is the minimum regular diagram of K. Let K'
be a knot that is equivalent to K, and suppose that D ~ is its minimum
regular diagram. Since we can think of D ~ as a regular diagram for K
(K and K' are equivalent), from the definition we have that c(D
o
) :5
c D ~ . However, since Do is a regular diagram of K', it again follows
from the definition that c D ~ :5 c(D
o
). Hence, combining these two
inequalities, we obtain c(D
o
) = c D ~ , Le., c(D
o
) is the minimum
number of crossing points for all knots equivalent to K. Consequently,
it is a knot invariant.

Exercise 4.2.1. Show that for c(D) = 0,1,2, the trivial knot is the
only knot that possesses a regular diagram D with one of the above
values.
Exercise 4.2.2. Show that the trefoil knot (either left-hand or right-
hand), K, has c(K) = 3. Further, show that among all knots and links
the trefoil knot is the only one with c(K) = 3.
Exercise 4.2.3. List all the knots and links with c(K) = 2, 3, 4, 5.
In general, there is no known method to determine c(K). Recently,
however, c(K) has been completely determined in the case of alternating
knots (or links), see Chapter 11, Section 5. For some specific types of
non-alternating knots (or links), c(K) has also been determined, see
Chapter 7. However, the following conjecture has yet to be resolved:
Conjecture Suppose that K
1
and K
2
are two arbitrary knots (or
links), then
Chapter 4 58
In the special case when both K
1
and K
2
are alternating knots (or
links), this conjecture has been shown to be true, see Chapter 11, Sec-
tion 5.
3 The bridge number
At each crossing point of a regular diagram, D, of a knot (or link)
K, let us remove (from D) a fairly small segment AB that passes over
the crossing point. The result of removing these segments is a collection
of disconnected (i.e., without any crossing points) polygonal curves, see
Figures 4.3.1{a) rv (c). We may think of the original regular diagram, D,
as the resulting diagram that occurs when we attach the segments AB,
... , (that pass over) to the endpoints of these disconnected polygonal
curves on the plane.
(a) (b)
Figure 4.3.1
(c)
Since these segments AB pass above the segments on the plane,
these segments AB are called bridges. For a given D the number of
bridges is called the bridge number. To be more exact, let us introduce
the following definition:
Definition 4.3.1. Suppose that D is a regular diagram of a knot (or
link) K. If we can divide up D into 2n polygonal curves aI, a2, ... ,an
and f31' {32, . ,(3n, i.e.,
D =al U (t2 U ... U an U f31 U f32 U . U f3n'
that satisfy the conditions given below, then the bridge number of D,
br(D), is said to be at most n.
(1) a1, (t2, ... ,an are mutually diHjoint, simple polygonal curves.
(2) {jt, lJ
2
, ,(3n are also luutuully disjoint, simple curves.
59 Classical Knot Invariants
(3) At the crossing points of D, al, a2, ... ,an are segments that
pass over the crossing points. While at the crossing points of
D, {31, (32, ,f3n are segments that pass under the crossing
points.
If br(D) :5 n but br(D) 1= n - 1, then we define br(D) = n.
Example 4.3.1. The bridge number of the regular diagrams shown
in Figure 4.3.2 are, respectively,
(a) (b)
Figure 4.3.2
(c)
D
a
is another addition to our set of familiar and named knots and
links; this link is called a Hop! link.
The bridge number of a regular diagram D is not a knot invariant
for a knot K. There exist knots that have regular diagrams with different
bridge numbers. In fact, Figure 4.3.2(a) and (b) are regular diagrams
for the right-hand trefoil knot. (Show that these two diagrams are
equivalent.) As in the previous section, if we consider all the regular
diagrams for a given K, then the minimum bridge number of all these
regular diagrams is an invariant for K.
Theorem 4.3.1.
For a knot (or link) K, br(K) = min br(D) is an invariant for K,
v
where V is the set of all regular diagrams ofK. This quantity is called
the bridge number (or the bridge index) oiK.
Exercise 4.3.1. By considering the proof of Theorem 4'.2.1, prove the
above theorem.
Exercise 4.3.2. Show that if br(K) = 1, then K is the trivial knot,
and the trivial knot is the only knot with bridge number equal to 1.
Chapter 4 60
Exercise 4.3.3. Show that if L is a n-component link then br(L)
n. Unlike Exercise 4.3.2, if br(L) = n then L need not be the trivial
link. For example, show that the Hopf link has br(L) = 2.
In the specific case of br(K) = 2, there are many knots with this
bridge number, including the trefoil knot and the figure 8 knot. These
knots, called for obvious reasons 2-bridge knots, have been extensively
studied, to the point that they have been completely classified. In gen-
eral, however, no method has yet been found to allow us to determine
br(K) for an arbitrary knot K. But the following theorem has been
proven in Schubert [Sc2].
Theorem 4.3.2.
Suppose K
1
and K
2
are two arbitrary knots (or links). Then
Therefore, there exist knots with arbitrary large bridge index. For
example, the connected sum of n copies of a trefoil knot has the bridge
index n + 1. But in comparison to c(K), br(K) is usually quite small.
The following conjecture, which has yet to be completely proven, signi-
fies that these two quantities are quite closely related:
Conjecture. If K is a knot, then
c(K) 2: 3(br(K) - 1),
WllfJre equality only holds when K is the trivial knot, the trefoil knot,
or the (connected) sum of trefoil knots.
It is possible to calculate the bridge number in a different way to
tllut described above. In order to redefine the bridge number so that
we can calculate using the alternative method, we shall assume that the
regular diagram of a knot (or link) K is a smooth (plane) curve D.
4.3.:1
61 Classical Knot Invariants
Denote by v(D) the number of local maxima of D, in relation to
the direction of a certain (plane) vector V, Figure 4.3.3.
As above, v(D) itself is not an invariant for K. However, if we
consider all the regular diagrams for K, then the minimum value of all
the number of local maxima, over all regular diagrams, is an invariant
for K. This knot invariant is equal to br(K).
Theorem 4.3.3.
If we consider all the regular diagrams, D, of a knot K, then
br{K) =min v{D).
'D
Exercise 4.3.4. Give a proof of the above theorem.
Exercise 4.3.5. Determine the bridge number of the knots in Fig-
ure 4.3.4.
Exercise 4.3.6. Find a regular diagram D of the square knot [Fig-
ure 1.5.6{a)] with br(D) =3.
(a)
Figure 4.3.4
4 The unknotting number
(b)
At one of the crossing points of a regular diagram, D, of a knot (or
link) K exchange, locally, the over- and under-crossing segments. Since
this type of alteration is not an elementary knot move, in general what
we obtain is a regular diagram of some other knot.
Example 4.4.1. In Figure 4.4.1(a), if we exchange the under- and
over-crossing segments within the small circle, the subsequent regular di-
a.gram can readily be seen to be that of the trivial knot, Figure 4.4.1 (1)).
Chapter 4
(a) (b)
Figure 4.4.1
62
Proposition 4.4.1.
We can change a regular diagram, D, of an arbitrary knot (or link)
to the regular diagram of the trivial knot (or link) by exchanging the
'over- and under-crossings segments at several crossing points of D (it
may also be necessary to use the Reidemeister moves).
Due to Proposition 4.4.1, the above operation, which exchanges
the over- and under-crossings segments at a crossing point, is called an
unknotting operation.
Proof
The proof is based on induction on the number of crossing points,
c(D), of D. In the trivial case c(D) = 0, since the knot can only be the
trivial knot, we have nothing to prove.
Therefore, suppose that the proposition holds for all regular di-
agrams D that have c(D) < m. Let us suppose that D is a regular
diagram with c(D) = m. Let P, an arbitrary point on D that is not a
crossing point, be what we might term a starting point. From P follow
the knot around, naturally, in the direction of its fixed orientation.
If at a crossing point of D, we move along a part that passes over
the crossing point, then do nothing just continue traversing the knot,
Figure 4.4.2(a). However, if we arrive at a crossing point and then move
along the part that passes under the crossing point, Figure 4.4.2(b), then
at this crossing point perform an unknotting operation, Figure 4.4.2(c).
"",.."
.' .." "
P
(8)
x
P
(b)
Fip;ure 4.4.2
63 Classical Knot Invariants
In this way, we will slowly create a regular diagram on which, start-
ing from P, we shall always pass over the crossing points of the knot. If
we continue traversing along D, repeating the above process, we shall
eventually arrive at a crossing point A that we have already passed
through, Figure 4.4.3(b). (If K is a link, then we may arrive back at
our starting point P.)

/X'
X
p )
(a)
Figure 4.4.3
(c)
Once the above process has been finished, what will have been
created is a loop that includes A, Figure 4.4.3(b). By applying Reide-
meister moves, we may remove this loop. The new regular diagram, D',
created by this process will have fewer crossing points than D. We may
now apply to D' the induction hypothesis, in so doing we complete the
proof.

As in our previous discussions, we define the unknotting number of

D as the minimum number of unknotting operations that are required
to change D into the regular diagram of the trivial knot (or link). We
will denote the unknotting number of D by u(D). As might be expected,
u(D) is not an invariant of K.
(a)
Figure 4.4.4
(b)
Chapter 4 64
Exercise 4.4.1. In Figure 4.4.4 we have drawn two regular diagrams
for a certain knot. Show that the regular diagram in Figure 4.4.4(a)
requires only one unknotting operation to change it into a regular di-
agram of the trivial knot, while the regular diagram in Figure 4.4.4{b)
requires two unknotting operations.
As before, consider all the regular diagrams for K, then the mini-
mum number of unknotting operations from all the regular diagrams is
a knot invariant.
Theorem 4.4.2.
If K is a knot (or link), then u(K) = min u(D) is an invariant of
v
K, where V is the set of all regular diagrams of K. We say that u(K)
is the unknotting number of K.
If we exclude the case when K is the trivial knot, then u(K) ~ 1.
However, to, actually, determine u(K) is a very hard problem. Even for
very specific types of knots, there are virtually no methods, as yet, to
determine u(K).
Exercise 4.4.2. Show that the unknotting number of both knots in
Figures 4.3.3 and 4.3.4(b) is 1, while the unknotting number of the knot
in Figure 4.3.4(a) is at most 2.
Exercise 4.4.3. Show that the knot in Figure 2.1.6 has an unknotting
number of at most 3.
Exercise 4.4.4. Show that it is possible to change an arbitrary reg-
ular diagram to an alternating regular diagram, which has the same
projection, by performing the unknotting operation a finite number of
times.
5 The linking number
The knot (or link) invariants that we have discussed thus far have
all been independent of the assigned orientation of the knot. In this
section we shall define the linking number, an important invariant for
oriented links.
First, let ns assign either +1 or -1 to each crossing point of a
regular d i a ~ r n n of an oriented }<J)ot. or link.
65 Classical Knot Invariants
Definition 4.5.1. At a crossing point, c, of an oriented regular dia-
gram, as shown in Figure 4.5.1, we have two possible configurations. In
case (a) we assign sign(c) = +1 to the crossing point, while in case (b)
we assign sign(c) == -1. The crossing point in (a) is said to be positive,
while that in (b) is said to be negative.
x
sign(c) = +1
(a)
x
sign(c) =-1
(b)
Figure 4.5.1
Suppose, now, that D is an oriented regular diagram of a 2-
component link L == {Kl, K2}. Further, suppose that the crossing
points of D at which the projections of K
1
and K
2
intersect are
Cl, C2, ... ,c
m
(We ignore the crossing points of the projections of K
1
and K
2
, which are self-intersections of the knot component.)
Then
+Sign(C2) +... +sign(c
m
)}
is called the linking number of K
1
and K
2
, which we will denote by
Ik(K
1
,K
2
).
Theorem 4.5.1.
The linking number Ik(K
1
, K
2
) is an invariant of L.
That is to say, if we consider another oriented regular diagram, D',
of L, then the value of the linking number is the same as for D. Therefore,
we shall call this number the linking number ofL, and denote it by Ik(L).
Further, the linking number is independent of the order of K
1
and K
2
,
Le., 1k(K
1
, K
2
) = Ik(K
2
, K
1
).
Before we give a proof of this theorem, we would like to consider a
couple of examples and exercises.
Example 4.5.1. Let us calculate the linking number of the links L
and L' in Figure 4.5.2(a) and (b), respectively.
(a) We need only calculate the signs at the 4 crossing points
Cl, C2, C3, and C4. Since the sign at each crossing point is
-1, we obtain that Ik(K
1
, K
2
) = -2.
Chapter 4 66
(b) Similarly, it is easy to show sign(cl) = sign(c4) = +1, while
sign(c2) = sign(cg) = -1. Therefore, Ik(Ki, K ~ = o.
Exercise 4.5.1. Show that the linking number is always an integer.
(Hint: Consider the Jordan curve theorem.
6
)
Exercise 4.5.2. Suppose that we reverse the orientation of K
2
, which
we will denote by - K
2
Show that
Therefore, the linking number of L is an invariant that depends on
the given orientation.
(a)
Figure 4.5.2
(b)
Proof of Theorem 4.5.1
Suppose that D' is another regular diagram of L. From our dis-
cussions thus far, we know that we may obtain D' by performing, if
necessary several times, the Reidemeister moves nr, n ~ and ngi=, Fig-
ure 4.1.1. Therefore, in order to prove this theorem it is sufficient to
show that the value of the linking number remains unchanged after each
of nt, n ~ and nt is performed on D. We shall only prove the theo-
rem for the case when n
i
(i = 1,2,3) is applied and leave the remaining
cases as exercises for the reader.
(1) At the crossing points of D at which we intend to apply
0,1, every section (edge) of such a crossing point belongs to the same
component. Therefore, applying a 0
1
does not affect the calculation
of the linking number.
(2) An application of O
2
on D only has an effect on the linking
number if A and B, see Figure 4.5.3(a), belong to different components.
67 Classical Knot Invariants
Since A and B can be assigned two different orientations, Figure 4.5.3(b)
and (c), it is necessary to consider these cases separately. However, in
both cases, since the newly created crossing points Cl and C2 have
opposite signs, we have sign(cl) + sign(c2) = 0; again, the linking
number is unaffected.
>
A B
f f
A B
f t
A B
(a)
(b)
/
A B
(c)
Figure 4.5.3
(3) Finally, let us consider the effect of {l3 on D, see Figure 4.5.4,
i.e., the effect on the signs of ci, and Cl, C2, C3, the crossing points
that are affected by {lg.
na

B C
(a)
--+
+--
(4.5.1)
Figure 4.5.4
Irrespective of how we assign the orientation on A, B, and C, the
following equations will always hold:
sign(cl) = sign(c2) =
sign(cg) =
If A, B, and C all belong to the same component, then, as before,
Chapter 4 68
the linking number is unaffected. So suppose, first, that A belongs to
a different component than Band C. Then the only parts that have
an effect on the linking number is the sum sign(C2) + sign(Cg) in Fig-
ure 4.5.4(a) and +sign(cs) in Figure 4.5.4(b). Due to (4.5.1),
these two are, in fact, equal, and therefore this does not cause any change
to the linking number. The other case, i.e., the various possibilities for
the components to which A, B, and C belong, can be treated in a sim-
ilar manner. Therefore, the linking number, Ik(L), remains unchanged
when we apply Ogo

Suppose, now, that L is a link with n components, K

1
, K
2
, , Kno
With regard to two components K
i
and K
j
(i < j), we may define as
an extension of the above the linking number lk( K
i
, K
j
), 1 i < j n.
(To calculate this linking number, we ignore all the components of L
except K
i
and K
j
.) This approach will give us, in all, n(n
2
-1) linking
numbers, and their sum,
L Ik(Ki,Kj) = 1k(L),

is called the total linking number of L.
Exercise 4.5.3. Show that, in fact, the total linking number of L is
an invariant of L.
So far we have ignored the crossing points of K
1
itself or K
2
We
shall now consider a definition in which they become significant.
Definition 4.5.2. Suppose that D is an oriented regular diagram of
an oriented knot (or link). Then, the sum w(D) of the signs of all the
crossing points of D is said to be the Tait number of D (or the writhe
of D).
The Tait number of D, w(D), is itself not an invariant of a knot
(or link). (Why is this the case?) As the name suggests, w(D) was first
considered by P.G. Tait at around the turn of the 20
th
century. He
thought that if D and D' are two minimum (in terms of the number
of crossing points) regular diagrams of a knot, K, then w(D) = w(D').
It was believo<l for a long time that indeed this was the case. For this
reason the in Figure 1.2.1 were thought to represent different
knots. In fnet, t.hey are two different rep;ular diagrams of the same knot!
69 Classical Knot Invariants
Exercise 4.5.4. Calculate the Tait number of the two knot diagrams
in Figure 1.2.1.
Exercise 4.5.5. Calculate the linking number of Figure 4.5.5(a)
and (b).
(a)
Figure 4.5.5
Exercise 4.5.6. Let L* be the mirror image of an oriented link L.
Show that Ik(L*) == -lk(L).
Figure 4.5.6
Exercise 4.5.7. Show that the (unoriented) link in Figure 4.5.6 is not
amphicheiral. [Hint: Give various orientations to each component of L
and compare lk(L) and lk(L*), L* is the mirror image of L.]
6 The colouring number of a-- knot
In this final section of this chapter we would like to and
define an invariant that is called the colouring number of a knot.
Suppose that the projection K of a knot (or link) K has n crossing
points Pl, P
2
, ,P
n
. Since each Pi is the projection of the points
and Pi" of K, Figure 2.2.1, we can, by means of these points, divide
Chapter 4 70
K into 2n segments (or polygonal curves) AI, A
2
, .. , A
2n
. To each of
these segments we may assign one of three colours red, blue, or yellow
in such a way that the following two conditions are satisfied:
(1) If A
k
and A, are as in Figure 4.6.1, then they have
the same colour assigned.
(2) A
k
(or Al), A
r
, As, Figure 4.6.1, either all have (14.0.1)
the same colour assigned or each, respectively, has a
different colour assigned to it.
Figure 4.6.1
A regular diagram, D, of K, which can have the three colours as-
signed throughout the diagram in the above fashion, is said to be 3-
colourable.
Example 4.6.1. (a) The regular diagram, D, of the trefoil knot, K,
as drawn in Figure 4.6.2, is 3-colourable. For example, we may assign
red to the segment AB, blue to the segment Be, and yellow to the
segment CA.
(b) The regular diagram of the figure 8 knot, Fig-
ure 0.5, is not 3-colourable.
Figure 4.6.2
Proposition 4.6.1.
If there exists a regular diagram D of a knot (or link) K that is 3-
colourable, then every regular diagram, D', of K is 3-colourable. Such
a knot, K, is sllid to be 3-colourable.
71 Classical Knot Invariants
Proof
Suppose that D is a 3-colourable regular diagram of K. If D' is
another regular diagram of K, then, as before, we can change D into D'
by applying, possibly several times, the Reidemeister moves 0
1
, O
2
, 0
3
and their inverses. Therefore, to prove this proposition it is sufficient to
show that each of the regular diagrams obtained after we have performed
one of the Reidemeister moves Or, n ~ and nt is 3-colourable. (Note
that the conditions for 3-colourability do not depend on the orientation
of the knot, so we will ignore orientations.)
(i)
A
(a)
(il)
)<J
(i)
A B
(b)
(i)
B'
Os
Ck
B"I
~
(c)
Figure 4.6.3
If. (ii)
"
>X'
~ B
A
(ii)
B'
C ~ ~
~ ~
(1) In the case of a 01-move, Figure 4.6.3(a), we may colour
each segment of Figure 4.6.3(a)(ii) with the same colour as A. The same
is true in the case of 011, which is left as an exercise for the reader.
(2) In the case of a {l2-move, Figure 4.6.3(b), if A and B are the
same colour, then A' and A" may also be assigned the same colour.
However, if A and B have different colours assigned to them, then assign
to A' the same colour as A, and to A" assign the colour that neither
A or B is coloured with, Le., the third colour. The same arguments hold
Chapter 4 72
in the case of
2
1
, which also is left as an exercise for the reader.
(3) In the case of a Os-move, Figure 4.6.3(c), if before we apply
this move, Figure 4.6.3(c)(i), A, B, C, ... , all have the same colour,
then after the 03-move is applied, Figure 4.6.3(c)(ii), we may assign
to them this same colour. In the general case, for clarity let us suppose
that the three colours are denoted by 0, f3, "I. Now, let us consider the
case shown in Figure 4.6.4(a) (i), where A and B are both assigned the
colour a and C is assigned the colour (3. Since A', B", and B' are
assigned the colours "I, 0, and " respectively, in Figure 4.6.4(a)(ii)
we may assign to B the colour ,. The cases where, firstly, A and C
are both assigned the colour and B is assigned the colour {3 and,
secondly, Band C are assigned the colour and A is assigned the colour
{3, can be dealt with in the same way as the preceding case. Finally,
let lis consider the case where A, B, and C each have a different colour
assigned to them, 0, (3, and " say, Figure 4.6.4(b). Since B", A', and
B' have the colours " (3, and " respectively, assigned to them, we
may assign to B, Figure 4.6.4(b)(ii), the colour o. The same argument
holds in the case of 0;1.
(i)
(ii)
(i) (ii)
PKY
~
':Y 'J
Yk
P
~
~
--+
~ >Y
;r
--+
~ >a
a I aC:
a I
p
a(
a a
(a)
(b)
Figure 4.6.4

Exercise 4.6.1. Show that if a knot (or link) K is 3-colourable, then

its mirror image K* is also 3-colourable.
In this section, so far, we have only considered 9-colourable knots
(or links). The reader may have already wondered: Is it possible to have
p-colourable knots (or links) where p is a prime number.
As above, consider AI, A
2
, . , A
2n
, which we created previously,
and assign to each segment (or polygonal line) Ai an integer Ai that
takes its value from the set of consecutive integers 0 to p - 1, inclusive.
These Ai are assigned so that the following conditions are satisfied,
with regard to Figure 4.6.1:
73 Classical Knot Invariants
(1) Ak = Al
(2) A
r
+As == Ak +Al (modp).10
It is always possible to find Ai for which the above conditions hold.
For example, in the extreme case we may assign Al = ... = A2n == o. In
the particular case when all the segments have the same colour assigned
to them, the colouring is said to be a trivial colouring of a regular
diagram, D. A regular diagram, D, is said to be p-colourable, if, at the
very least, two segments are assigned two different integers from the set
oto p - 1. The following proposition may be proven along similar lines
to Proposition 4.6.1:
Proposition 4.6.2.
If a knot (or link) K has at least one p-colourable regular diagraJIl,
then every regular diagram is p-colourable. .
A given knot (or link), K, may be p-colourable with regard to sev-
eral different ps. Therefore, the different number of colours with which ,"
K may be coloured is an invariant of K. This invariant is the colouring
number set of K. It is possible to determine the colouring number set
for K [F2].
Exercise 4.6.2. Show t ~ t a knot cannot be non-trivially 2-coloured,
but a link with n ~ 2) components can always be non-trivially
2-coloured.
Exercise 4.6.3. Show that the figure 8 knot is 5-colourable but not
3-colourable.
Exercise 4.6.4. The knots in Figure 4.6.5 are 3-colourable. Show that
this is the case. (Note: The method of colouring them is not unique.)
Figure 4.6.5
Exercise 4.6.5. For the case p = 3, show that the definition of
p-colourable agrees with our original definition of 3-colourable.
Chapter 4 74
Exercise 4.6.6. (1) Find two different 3-colourings of the diagram
of the right-hand trefoil knot [Figure O.l(c)] and show that one of the
3-colourings can be obtained from the other by simply permuting the 3
colours. (Such two colourings are called equivalent.)
(2) Find two non-equivalent 3-colourings of the (reg-
ular diagram of the) square knot, Figure 1.5.6(a). How many non-
equivalent 3-colourings does this knot (diagram) have?
In any science, in any discipline there are moments that can be
called turning points - they reinvigorate and deepen the understanding
of the subject at hand. What exactly is a turning point, even among
friends, is usually contested and debated feverishly. Knot theory also
has many turning points; however, there are two that are beyond debate:
the Alexander polynomial and the Jones polynomial.
The Alexander polynomial, discovered by J.W. Alexander in 1928,
has become one of the cornerstones of knot theory. Although the polyno-
mial carries the epithet Alexander, Reidemeister at essentially the same
Chapter 5 76
time announced something that he called the L-polynomial. These two
polynomials can be shown to be the same, even though Reidemeister's
approach is independent of the work of Alexander. (The "L" in the
L-polynomial is an abbreviation of Laurent, since in this polynomial
some of the terms may have negative exponents, for example, t-
2
.) We
will deal with the second turning point, or more accurately, the Jones
''revolution,'' in Chapter 11.
The study of the Alexander polynomial and determination of its
precise context in knot theory has been extensively carried out. It has
been found that the Alexander polynomial is very closely connected
with the topological properties of the knot, and this has had a very
profound impact on the theory of knots. Also of great significance is
that there are various methods by which we may calculate the Alexander
polynomial. In the mechanics of one of these methods, a concept called
the Seifert matrix is brought to light. This concept of the Seifert matrix
is itself extremely interesting and is, in fact, also one of the cornerstones
of classical knot theory. This chapter is devoted to an explanation of
the Seifert matrix and its properties. This will allow us, in the next
chapter, to define the Alexander polynomial via Seifert matrix.
1 The Seifert surface
Let us begin with the following theorem due to L. Pontrjagin and
F. Frankl:
Theorem 5.1.1.
Given an arbitrary oriented knot (or link) K, then there exists
in R
3
an orientable, connected surface, F, that has as its boundary
K. (That is to say, there exists an orientable connected surface that
spans K.)
The above theorem was proven in 1930; here, however, we shall
give a very neat proof due to Seifert.
Proof
Suppose that K is an oriented knot (or link) and D is a regular
diagram for K. Our intention is to decompose D into several simple
closed curves. The first step is to draw a small circle with one of the
crossing points of D as its centre. This circle intersects D at four points,
say, a, b, c, and d, Figure 5.1.1{a). As shown in Figure 5.1.1{b), let us
splice this crossing point and connect a and d, and band c.
77
(a)
Seifert Matrices
(b)
Figure 5.1.1
What we have done is to change the original segments ac and bd
into the new segments ad and bc. In this way we can remove the cross-
ing point of D that lies within the circle. This operation is called the
splicing of a knot K (along its orientation) at a crossing point of D. If
we perform this splicing operation at every crossing point of D, then we
shall remove all the crossing points from D. The end result is that D
becomes decomposed into several simple closed curves, Figure 5.1.2(b).
These curves are called Seifert curves. D, itself, has been transformed
into a regular diagram of a link on the plane that possesses no crossing
points (Le., the trivial link). Each of these simple closed curves may
now be spanned by a disk.
(a)
o
(b) (c)
(d) (e)
Figure 5.1.2
Chapter 5
78 "
In the case of Figure 5.1.2(b), by slicing we obtain three disks,
D
1
, D
2
, and Dg , Figure 5.1.2(c). The boundary of D
i
is the Seifert
curve C
i
In Figure 5.1.2(c), there is a possibility that D
2
may lie
on top of D
1
, or D
2
may be under D
I
. This ambiguity causes some
difficulties in Section 3 (see Example 5.3.4). However, these difficulties
will be resolved once we prove Theorem 5.4.1. Finally, in order to create
a single surface from the various disks, we need to attach to these disks
small bands that have been given a single twist. To do this, firstly, take a
square acbd and give it a single positive or negative twist, Figure 5.1.3(a)
and (b), respectively; these twisted squares are the required bands.
a
d
(a) (b)
Figure 5.1.3
If we attach positive (negative) bands at the places of D that
corresponded to positive (negative) crossing points before they were
spliced (see Figure 4.5.1), then we obtain a connected, orientable
surface F, Figure 5.1.2(d). (In the case of a link, K, if we alter
K in such a way that the projection of K is connected, then by
the above method we can also obtain a connected surface.) The
boundary of this surface, F, is plainly the original knot K. Further,
as noted above, F is also an orientable surface (see Exercise 5.1.1).

(a) (b)
Figure 5.1.4
As shown in Figure 5.1.4(a), by shading the front of the surface and
dotting the back of the surface, we may distinguish between the front
and the back of the surface. This allows us to assign an orientation to
79 Seifert Matrices
the surface. However, as in Figure 5.1.4(b), if one of the bands has a
double twist, then it is not possible for us to distinguish between the
front and the back.
Exercise 5.1.1. Show that the surface constructed in the proof of
Theorem 5.1.1 is an orientable surface.
In general, an orientable, connected surface that has as its boundary
an oriented knot (or link) K is called a Seifert surface of K. (Due to
its origins, maybe, in fact, we should call it a Pontrjagin-Frankl-Seifert
surface.) The orientation of F is induced naturally from the orientation
of the knot K that forms its boundary. The Seifert surface that was
constructed in the above proof depended on the regular diagram, D, of
K. Hence, it is more precise to say that is the Seifert surface formed
from D, a regular diagram of K.
Caveat lector, in the link case, even a seemingly innocuous change
of orientation of a component(s) may cause the Seifert surface to change
quite substantially.
Suppose, now, that a surface, F, is the Seifert surface of a knot
K obtained from the disks and bands as described above. If we shrink
(contract) each disk to a point, and at the same time the width of the
bands is shrunk, ideally, into quite narrow segments, then from these
points and segments a graph in space is formed. Such a graph is called
the Seifert graph (of a regular diagram D) of K. These graphs, in fact,
lie on the plane, i.e., they are plane graphs (cf. Exercise 5.1.2). Fig-
ure 5.1.2(e) is the Seifert graph of the figure 8 knot, Figure 5.1.2(a), with
vertices (segments) corresponding to, the disks (bands, respectively).
Exercise 5.1.2. (a) Show that a Seifert graph is a plane graph. Fur-
ther, show that it is also a bipartite plane graph. [A graph G (not
necessarily plane) is said to be bipartite if the set of vertices of G can
be divide into two non-empty disjoint subsets, VI and V2, such that
every edge of G has one end in V
l
and the other in V2.]
(b) Show that a graph G is bipartite if and only if
every closed path of G consists of an even number of edges, and, in
particular, show that a bipartite graph cannot have a loop.
Exercise 5.1.3. Construct Seifert surfaces (obtained from the regular
diagrams) for the knots and links in Figure 5.1.5. Also, determine their
Seifert graphs. Further, change the orientation of one of the components
in Figure 5.1.5(d) and once again determine its Seifert graph. Compare
this Seifert graph with the previous one.
Chapter 5 80
(a) (b) (c)
Figure 5.1..5
(d)
2 The genus of a knot
At this juncture, we ask the reader's indulgence as we now need
to consider a well-known, fundamental theorem!! in topology that is
concerned with the classification of surfaces. This theorem states that a
closed (i.e., one that is compact 'and without boundary) orientable sur-
face, F, is topologically equivalent (Le., homeomorphic) to the sphere
with several handles attached to its surface. The number of these han-
dles is called the genus of F, and is denoted by g(F).
(a) (b)
(c)
Figure 5.2.1
Example 5.2.1. The surface of genus 1, shown in Figure 5.2.1(a), is
called a torus, while the surface in Figure 5.2.1{b) has genus 2.
81 Seifert Matrices
Let us now consider how to calculate the genus of a Seifert surface,
F, of a knot. Since F has a boundary, then by the above theorem F is
homeomorphic to a sphere with several handles attached, and further-
more, with a hole on the sphere (with handles) for each component of
the link, Figure 5.2.1(c). Unfortunately, it is usually not that easy to
visualize the Seifert surface of a knot. For example, the Seifert surface
of the figure 8 knot shown in Figure 5.1.2(d) is, in fact, topologically
the same surface as in Figure 5.2.1(c).
Seifert, by the method we described in the previous section, re-
confirmed that such orientable surfaces do exist, and hence in theory,
anyway, we may consider the minimum genus of all Seifert surfaces for
a given knot K. This minimum genus is called the genus of K, denoted
by g(K). The genus is a knot invariant (the invariant is defined, as on
previous occasions, as the minimum one over such genera of a given K).
For an arbitrary knot there does exist an algorithm to actually calculate
its genus, but it is exceedingly difficult to implement. In truth, to calcu-
late the genus of an arbitrary knot is a difficult undertaking. However,
for certain types of knots the calculation of the genus is a relatively
straightforward matter (cf. Chapter 7 and Chapter 11, Section 5). Al-
though the determination of the genus of an arbitrary knot is difficult,
to determine the genus of "constructed" orientable surface is quite easy.
The calculation relies on a classical invariant, the Euler characteristic.
Theorem 5.2.1.
We may divide a closed orientable surface into 00 points, 01
edges, and 02 faces. Let
X(F) = Qo - Q1 +02;
then X(F) is an integer that is independent of how we have divided F;
i.e., it is only dependent on F. This integer is called the Euler charac-
teristic of F.
The Euler characteristic X(F) and the genus ofF, g(F), are related
by means of the following equation:
Therefore,
X(F) = 2 - 2g(F). (5.2.1)
g(F) = 2 - X(F) .
2
If F has a boundary, since the boundary is also composed of several
points and edges, the above formula becomes
X(F) = 2 - JL(F) - 2g(F), (5.2.2)
b the
~ b b d =
/o{ the link K.
/
-d+b,
-d+b.
Knot, since p,(K) = 1 it follows that
/' = I-d+b.
(b) (a)
Chapter 5
83 Seifert Matrices
(5.2.3) f -1 = 1- d+b,
,1S equal to the number of faces of this division of 8
2
,
face that contains the point at infinity, 00.
, 5.2.3. Calculate the genus of each Seifert surface of the
,.,nd links in Exercise 5.1.3. Further, with regard to the Seifert
obtained from these surfaces, verify that the above formula,
.2.3), holds.
For the rest of the section let us consider this number, i.e., 1-d+b.
Suppose r(D) is the Seifert graph constructed from the Seifert surface
in Figure 5.1.2{e). Since r(D) is a plane graph, r(D) divides 8
2
into
several domains. (We may think of the sphere 8
2
as R
2
with the
of the point at infinity.) In this partition of 8
2
, the number
, "'oints is d and the number of edges is b. Suppose that the number
'co; is f; then from Theorem 5.2.1 and Exercise 5.2.1 we obtain,
3 The Seifert matrix
Suppose that F is a Seifert surface created from the regular dia-
gram, D, of a knot (or link) K, and r{D) is its Seifert graph. We want
to create exactly 2g(F) +JL(K) - 1 closed curves
12
that lie on F.
When r(D) partitions S2, then we showed in the previous section
that 2g(F) +J.t(K) - 1 (= 1 - d + b = f - 1) is equal to the number
of domains (excluding the domain that contains 00). The boundary
of each of these domains (faces) is a closed curve of r(D). Therefore,
we can, from these closed curves, create the closed curves on the Seifert
surface.
(a) (b)
Figure 5.3.1
(c)
Chapter 5 82
where p,(F) is the number of closed curves that make up the boundary
ofF.
Example 5.2.2. We can divide the torus with a hole in the manner
shown in Figure 5.2.2, so that 00 = 7, 01 = 14, and 02 = 6. It follows I
from this that X(F) = -1, and therefore g(F) = 1.
Exercise 5.2.1. Show, by suitably dividing it, that the sphere S2 has
Euler characteristic 2.
Let us now apply the above Euler characteristic, (5.2.2), to the
Seifert surface that were previously constructed. We may think of the
disks and bands of F as a division of F. The points of F in this division
are the four vertices of each band. The edges of F are the polygonal
curves that constitute the edges of the bands and the boundaries of the
disks between the vertex points. The faces of F are the disks and the
bands.
Figure 5.2.2
Exercise 5.2.2. Show that if d is the number of disks and b the
number of bands, then 00 = 4b, 01 = 6b, and 02 = b+d.
Therefore, it follows from Exercise 5.2.2 that X(F) = 4b-6b+b+d =
d - b. Further, JL(K) is just the number of components of the link K.
So from (5.2.2) we obtain that
2g(F) = 2 - p,(K} - X(F) = 2 - p,(K} - d +b,
or equivalently,
2g(F) +p,(K} - 1 = 1 - d +b.
In the special case when K is a knot, since JL(K} = 1 it follows that
2g(F) = 1 - d +b.
83 Seifert Matrices
For the rest of the section let us consider this number, Le., 1-d+b.
Suppose r(D) is the Seifert graph constructed from the Seifert surface
in Figure 5.1.2(e). Since r(D) is a plane graph, r(D) divides S2 into
several domains. (We may think of the sphere S2 as R
2
with the
addition of the point at infinity.) In this partition of S2, the number
of points is d and the number of edges is b. Suppose that the number
of faces is f; then from Theorem 5.2.1 and Exercise 5.2.1 we obtain,
2 = X(S2) =d - b+f.
Therefore,
f -1 = 1- d +b, (5.2.3)
Le., 1 - d + b is equal to the number of faces of this division of S2,
excluding the face that contains the point at infinity, 00.
Exercise 5.2.3. Calculate the genus of each Seifert surface of the
knots and links in Exercise 5.1.3. Further, with regard to the Seifert
graphs obtained from these surfaces, verify that the above formula,
(5.2.3), holds.
3 The Seifert matrix
Suppose that F is a Seifert surface created from the regular dia-
gram, D, of a knot (or link) K, and r(D) is its Seifert graph. We want
to create exactly 2g(F) +J.t(K) - 1 closed curves
12
that lie on F.
When r(D) partitions 8
2
, then we showed in the previous section
that 2g(F) +J.t(K) - 1 (= 1 - d +b = f - 1) is equal to the number
of domains (excluding the domain that contains 00). The boundary
of each of these domains (faces) is a closed curve of r(D). Therefore,
we can, from these closed curves, create the closed curves on the Seifert
surface.
(8) (b)
Figure 5.3.1
(c)
Chapter 5 84
Example 5.3.1. The two closed curves 01 and 02, Figure 5.3.1(b)
on F, correspond to the boundaries of the two faces, excluding the
one that contains 00, 11 and 12 (on 8
2
), Figure 5.3.1(c), obtained
from r(D).
These 2g(F) +JL(K) - 1 (= m) closed curves, it would seem, are
nothing but a collection of very ordinary closed curves. However, they
will, with a bit of perspicacity, indicate certain characteristics of the
knot K that is the boundary of the surface F. [In this case, it is necessary
that the Seifert surface is in R
3
(or S3 ).] Individually, however, these
closed curves are of little interest, but as a collection of closed curves
they will provide us with a knot invariant. For example, the knot K
in Figure 5.3.1(a) has a Seifert surface of genus 1, from which we can
obtain two closed curves, at and a2. It is quite possible that at and
(}2 may have points of intersection; therefore, {at, a2} is not a link.
But if we lift a2 slightly above the surface, so that the new curve af
is ''parallel'' to curve a2, we can remove these points of intersection,
thus making {at, 02#} a link.
FX(l)
Fx(O)
a
(a)
a*
X
a
IXJx
a
(b)
Figure 5.3.2
In order to make this lift precise and slightly easier to understand,
let us consider the mathematical construction shown in Figure 5.3.2.
Firstly, we need to thicken F slightly; in other words, create F'x
[0,1], Figure 5.3.2. Some care needs to be taken during this thickening
process, so that both F and the segment [0,1] have the orientations that
obey the right-hand rule, Figure 5.3.3.
Figure 5.3.3
85 Seifert Matrices
The original surface F may be thought of as F x (0), and so we
may say that both 0,1 and 02 lie on F x (0). To be exact 0,1 and
02 should now be called 0,1 x (0) and 0,2 x (0), respectively. For the
sake of simplicity we shall retain the original notation, and also for this
purpose we shall denote 01 x (1) and a2 x (1) by 0,1# and 02#,
respectively.
We may assign an orientation to a1 and 0,2 in an arbitrary
fashion. These orientations induce, in a natural manner orientations
on af and af. This now allows us to calculate the linking num-
ber lk(a1' a2#). It is possible to similarly define the linking numbers
lk(0,2, a1#), lk(a1, a1#), and lk(02,02#). These four linking numbers
may be rearranged into the following 2 x 2 matrix form:
This matrix M is called the Seifert matrix of the knot K in Fig-
ure 5.3.1(a). Since the linking numbers themselves are integers, the
matrix M is an integer matrix. (It should be noted, however, that the
matrix M depends on the orientations of 0,1 and a2; therefore, the
matrix is not an invariant of K. )
If the genus of the Seifert surface, F, of a knot (or link) is g(F), then
on F there are 2g(F)+J-t(K)-1(= m) closed curves 01,02, ... , am- Ex-
panding the process outlined above, with arbitrary orientations assigned
to these closed curves, we can calculate their various linking numbers.
As above, we may formulate them in terms of the entries of a m x m
matrix,
Therefore, from the regular diagram, D, of K, we can obtain an
integer matrix. However, we should underline that this matrix depends
on the orientations of a1, 0,2, . .. ,am. This matrix is called the Seifert
matrix of K (constructed from a particular regular diagram of K). In
general, the linking numbers lk(ai, aj #) and lk(OJ, 0i #) are not equal,
so the matrix M is not a symmetric matrix.
Further, in the case when g(F) = 0, the Seifert matrix of K is
defined to be the empty matrix (K, as we have already mentioned, is
the trivial knot).
Let us now carefully consider several examples to illustrate how to
calculate, in practice, the Seifert matrix of a knot.
Chapter 5
~ o ~
; ~ J \
86
1
,;
Example 5.3.2. If we transform the regular diagram of the right-hand ~
trefoil knot to the one in Figure 5.3.4(a), then it is fairly straightforward ~ f
.'!tl
to see its Seifert surface is the one in Figure 5.3.4(b) and the subsequent
Seifert graph is as in Figure 5.3.4(c).
(a) (b)
r(D)
e
(c)
Figure 5.3.4
From Figure 5.3.4(b) it follows that there are two closed curves
a1 and 02 on the Seifert surface. The mutual relationships between
0:1, 1:r2, l:rl# and 1:r2# are shown in Figure 5.3.5(a) rv (d).
(a) (b) (e) (d)
Figure 5.3.5
From these four diagrams it follows that
Ik(O:I,O:I#) = -1, lk(0:2,0:1#) = 1, lk(0:2,0:2#) = -1,
with the linking number of the other case equal to O.
Therefore, the Seifert matrix for the right-hand trefoil knot is
[
-1 0]
M = 1 -1 .
Example 5.3.3. If in a similar manner we consider the Seifert matrix
of the left-hand trefoil knot, we obtain the following matrix:
[
1-1]
M= 0 1
87 Seifert Matrices
Example 5.3.4. Let us now consider the knot, K, in Figure 5.3.6{a).
(a) (b)
Figure 5.3.6
Various Seifert surfaces can be constructed from this diagram. In
particular, we shall consider the two cases in Figure 5.3.7{a) and (b).
(a)
Figure 5.3.7
(b)
Since the Seifert surface of K has genus 3, then its Seifert matrix is
tl. 6 x 6 matrix, and on the surface correspondingly there are 6 closed
~ u r v e s First, let us consider case (a). To try to avoid too much con-
fusion, we shall build up the linking number formulae pair upon pair.
Chapter 5 88
Let us begin with the pair a1 and Q2 on F1, which correspond to the
vertices a and b of the graph r(D). This pair of curves, if we look
at Figure 5.3.7(a), are formed on F
1
from the bands that connect the
top disk (which we may call the first disk) to the next disk below it
(i.e., the second disk), and of course from these two disks themselves;
see Figure 5.3.7.
In a similar way as in Figure 5.3.5, we can obtain the following
formulae:
lk(al,al#) = -1, lk(a2,al#) = 1, Ik(a2,a2#) = -1.
Let us now consider the pair a3 and 04, on F1, which correspond
to the simple closed curves in r(D) with vertices band c. This pair of
closed curves lies on the the second and third disks of F1 and the bands
that connect these two disks. We leave as an exercise the calculation j
of the linking numbers for this pair, namely between a3, at, a4, and
at The next step is for us to calculate the mutual linking numbers
between the two pairs of closed curves, i.e., aI, 02 and a3, a4, some
of the diagrams are shown in Figures 5.3.8{b) I'V (d). The subsequent
..j".
calculations yield
:.

(b}
(a)
We leave it as a straightforward exercise for the reader to show that all
the other linking numbers between these two pairs of closed curves are
1

J
Figure 5.3.8
If we continue in this vein, we next have to add the final pair
of closed curves and the relevant linking numbers. This is
89 Seifert Matrices
reasonably straightforward and so we shall, without direct computation,
give the subsequent matrix, the Seifert matrix of K, leaving it as an
exercise for the reader to check that the linking numbers are as printed.
-1 0 0 0 0 0
1 -1 0 0 0 0
M=
1 0 -1 0 0 0
-1 1 1 -1 0 0
0 0 1 0 -1 0
0 0 -1 1 1 -1
For the second case, Figure 5.3.7(b), similar calculations lead to
the following Seifert matrix M' from the Seifert surface F
2
:
-1 0 -1 1 0 0
1 -1 0 -1 0 0
M'=
0 0 -1 0 0 0
0 0 1 -1 0 0
0 0 1 0 -1 0
0 0 -1 1 1 -1
Exercise 5.3.1. Determine the Seifert matrix obtained from the reg-
ular diagram D of the knot in Example 5.3.1, with an arbitrary (Le., at
the reader's discretion) orientation assigned to al and a2.
Exercise 5.8.2. Determine the Seifert matrix for the knots in Fig-
ure 5.1.5(a) and (b). In particular, for the knot in Figure 5.1.5(b), find
two Seifert matrices by applying the method explained in Example 5.3.4.
As was noted in the above example, the Seifert matrix of a knot is
not unique. In fact, since we have fixed neither the orientation nor the
order of the closed curves aI, a2, ... , am, even by the seemingly minor
adjustment of changing the order, we can cause the Seifert matrix to
change. Therefore, in order to obtain an invariant of a knot from a
Seifert matrix, we need to examine the relationship between the Seifert
matrices of the same knot. The concept that is being alluded to in the
preceding sentence is the S-equivalence of two square matrices.
4 S-equivalence of Seifert matrices
The construction of the Seifert matrix outlined above depends on
the regular diagram we use. Hopefully the following statement is now
Chapter 5 90
second nature to the reader: We may transform one regular diagram
into another equivalent regular diagram by applying the Reidemeister
moves several times. Therefore, our next course of action, if we wish to
use the Seifert matrices to define knot invariants, is to carefully examine ,
the effect of the Reidemeister moves on the Seifert matrix.
Theorem 5.4.1.
Two Seifert matrices, obtained from two equivalent knots (or links), ,
can be changed from one to the other by applying, a finite number of ,
times, the following two operations, Al and A
2
, and their inverses:
where P is an invertible integer matrix, with det P = 1 (det P is
just the usual determinant ofP), and pT denotes the transpose matrix
ofP.
*
0 0 0
M
1
M
1
A
2
: M
1
--+- M
2
=
*
0
or
0 0
0 0 0 1
* *
0 0
0 0 0 0 0 0 1 0
where *
denotes an arbitrary integer.
The above mathematical argot is essential for the theorem to be
precise, but let us peel away some of this terminology and try to under-
stand exactly what effect the two operations will have on a matrix.
The operation A
l
either interchanges two rows, say i
th
and jth
rows, and then interchanges the i
th
and jth columns; or it adds k
times the i
th
row to the jth row, and then adds k times the i
th
col-
umn to the jth column. We shall call this operation an elementary
symmetric matrix operation. The operation has been defined in such a
way that it corresponds to the change of order or the change of orien-
tation of the closed curves mentioned above and others.
The operation A
2
, on the other hand, is a matrix operation that
is particular to knot theory. This operation has been defined so that
it corresponds to the change in the genus of the Seifert surface due to
a Reidemeister move, Le., it makes the Seifert matrix either smaller or
larger.
Exercise 5.4.1. In Example 5.3.2 reverse the orientation of Ql (and
hence, ar) and then ~ r m i n the new Seifert matrix M'. Compare
91 Seifert Matrices
M and M', and confirm that M' is obtained from M by multiplying
the first row and first column by (-1).
Definition 5.4.1. Two square matrices M, M' obtained one from
the other by applying the operations AI, A
2
and the inverse A
2
1
a
finite number of times, are said to be S-equivalent, and are denoted by
M M'. (The "S," of course, stands for Seifert.)
For convenience's sake, we say two matrices are Al -equivalent if
one is obtained from the other by applying the operation A
l
a finite
number of times.
Now, two Seifert matrices obtained from two equivalent knots (or
links) are, by Theorem 5.4.1, S-equivalent. Before we proceed with the
proof of Theorem 5.4.1, to avoid a situation as described in the next
paragraph from occurring, we would like to generalize and refine some
of our previous concepts.
A Seifert surface is, by definition, a connected surface. If a given
regular diagram D is not connected, we need to transform it into a con-
nected regular diagram, :5, by applying the Reidemeister move {l2.
Hence, a suitable connected surface can now be constructed. However,
it is possible that as we apply subsequent Reidemeister moves to trans-
form our original regular diagram to an equivalent regular diagram,
we shall encounter, within the intermediary regular diagrams, one that
is not connected, thus returning to our original problem. So to stop
chasing our tails, it is better to redefine the Seifert matrix, making it
independent of whether the Seifert surface is connected or disconnected.
Therefore, let D be a regular diagram with p connected compo-
nents D(l), D(2), ... D(p), (p 1). We can, using the methods al-
ready described, construct Seifert surface F( i ) for each D( i ) and sub-
sequently a Seifert matrix M( i) from F( i), i =1,2, ... ,p.
Definition 5.4.2. The Seifert matrix M of a disconnected (Seifert)
surface F(l) U F(2) U ... U F(p) is defined to be the direct sum of
M(l), M(2), ... M(p) and the zero matrix Op-l of order p - 1, Le.,
M(l)
M(2) 0
M=
o M(p)
Op-I
Next, we shall show the following proposition is a straightforward
Chapter 5 92
consequence of this definition.
Proposition 5.4.2.
Let F be the connected surface obtained from F(l) UF(2) U... U
F(p) by adding two bands with an opposite twist, see Figure 5.4.1,
between F(i) and F(i + 1), i = 1,2, ... ,p - 1. (So, F is a Seifert
surface constructed from a connected diagram D.) Then the matrix
M obtained from F is At-equivalent to the matrix defined in Defini-
.......8
tlon 5.4.2; hence, M""M.
F(l) F(2) F(p)
Figure 5.4.1
Proof
On a pair of these new bands, place a new simple closed curve ai,
i = 1,2, ... ,p - 1, see Figure 5.4.2.
F(i) F(i+l)
Figure 5.4.2
Since for each original simple closed curve aj,k constructed on
F(j), lk(aj,k' ar) =0 and lk(ai' atk) =0 [of course lk(ai' ar) =0],
we have that Mis AI-equivalent to M. (Nota bene, the only difference
between if and M is either a change of the numbering of the aj,k or
a change in the orientation.) I

We are now in a better position to proceed with the proof of The-

orem 5.4.1.
Proof of Theorem 5.4.1.
To prove the theorem, we need first to look at how the Seifert
surface changes when we. apply each of the Reidemeister moves, and
93 Seifert Matrices
secondly, as a consequence, to examine how the Seifert matrix changes.
In fact, due to the next proposition, we may restrict the proof to consider
only local changes of the surface.
Proposition 5.4.3.
Let D be a regular diagram and let D' be the regular diagram
obtained from D by applying only a single Reidemeister move on it.
Further, let F be a Seifert surface constructed from D. Similarly, we
can construct a F' from D', but this can be done so that F and F'
differ only at the parts that are affected by the Reidemeister move. (In
other words, F and F' are identical except at a few places.) Then the
Seifert matrices M and M' obtained from F and F', respectively, are
S-equivalent. (In fact, by the very construction of F', M' is identical
to M, except at a few rows and columns.)
The proof of Theorem 5.4.1 will be complete if, in addition to the
above proposition, we can prove that two different Seifert surfaces con-
structed from the same regular diagram have S-equivalent Seifert ma-
trices. This actually falls out from the next proposition.
Figure 5.4.3
Proposition 5.4.4.
Let F be one of the Seifert surfaces constructed from a regular
diagram D, and let MF the subsequent Seifert matrix ofF. Then there
exists a diagram Do such that
(1) Do is equivalent to D.
(2) Only one Seifert surface, F
o
, is constructed from Do; i.e., all
disks are on the same level, see Figure 5.4.3. (Such a surface
F
o
is sometimes called flat.) "
(3) The Seifert matrix of F
o
, MF
o
' is At-equivalent to MF, and
s
hence, MFo"-lMF.
Chapter 5
...~
"
94 t
~
\
f ~
:':j
Exercise 5.4.2. Show that Proposition 5.4.3 and 5.4.4 imply Theo-
rem 5.4.1.
Proof of Proposition 5.4.3.
The idea of the proof is quite straightforward, it requires only check- ;',
ing several possible cases. So we shall look at some of these cases and
leave the rest as exercises for the reader.
The single Reidemeister move applied is 0
1

With regard to this Reidemeister move, we shall consider two cases:

(i) We increase the number of bands and disks by only one of ,I
each, as in Figure 5.4.4(b}.
(a)
--+
(b) (c)
Figure 5.4.4
In this case the genus of the surface does not change, and the
corresponding Seifert graph only adds a single vertex and an edge, Fig-
ure 5.4.4(c). Therefore, since no new domain is created, the Seifert
matrix will remain unchanged.
(ii) The move is as shown in Figure 5.4.5.
(a)
--+
(b) (c)
Figure 5.4.5
However, this can be dealt with along similar lines to (i), and so
the Seifert matrix remains unchanged.
95 Seifert Matrices
The single Reidemeister move applied is !12.
Since the possibilities depend on how we assign the orientation to
each segment, there are several cases to consider. We shall only look at
some typical cases.
(i) In the case of Figure 5.4.6 the number of disks does not change,
but the number of bands increases by 2.
JC
(b)
(a)
(c)
Figure 5.4.6
If the original Seifert surface F is not connected, but the new surface
F' is connected, then it follows from Proposition 5.4.2 that M ~ M
So suppose that F is connected. Then the genus of F' is given by
g(F') = g(F) + 1. Therefore, F' compared with F has an extra 2 closed
curves, which we will denote by a' and a", Figure 5.4.7.
.... ----- ..
-- 41''''
",
a':'
(a)
Figure 5.4.7
,-_..... --- ... _- ......
, ,
, ,
a" ,
~ ~
a"*
(b)
We may suppose that the newly created two bands are the final
bands that connect the disks D' and D". (If necessary, we may change
the numbering of the closed curves.) The new matrix, M', in compari-
Chapter 5 96
son with M, has an extra two rows and columns, as shown below. Note
that lk(a', al/#) = 1, see Figure 5.4.7(b).
b
1
0
M
M'=
b
m
0
bi
b' b 1
m
0 0 0 0
where b
i
is the linking number between the closed curve ai on F
and a
'
#, is the linking number between a' and af, and finally
b =lk(a', a'#).
If we apply A
l
to M' we shall obtain,
b
1 0
M
MI/=
b
m
0
0 0 0 1
0 0 0 0
What we have shown is that the Seifert matrix M', constructed
from F', can be obtained by first performing A
2
to M, the Seifert ma-
trix of F, and then performing Al several times. Therefore, these Seifert
matrices obtained from F and F' are S-equivalent. The other variations
on this Reidemeister move can also be shown to give S-equivalence of
the relevant Seifert matrices.
The single Reidemeister move applied is {la.
We shall consider only the typical case shown in Figure 5.4.8(a).
In Figure 5.4.8, the numbers 1, 2, ... ,6 and the letters a, b, c indicate
crossing points, and the Seifert graphs r(D) and r(D') are identical,
except inside the broken circles.
Now, from the construction of F' we may assume that F and F'
are identical, except at a few places. The exact nature of these places
is best visualized from their respective Seifert graphs, since then they
correspond precisely to the interiors of the broken circles, similar to the
type shown in Figure 5.4.8(c). It is then easy, from Figure 5.4.8(c), to
see that F and F' have the same number of bands, but F has 2 more
disks than F'. [Note that a disk corresponds to a vertex and a band
corresponds to an edge in T(D) or r(D').]
97 Seifert Matrices
D
(a)
D'
--+
(b)
(i) (ii)
(c)
c
p
Figure 5.4.8
If we regard the shaded area as corresponding to the unbounded
region, then when a disk F
2
lies over F
1
, see Figure 5.4.9, the Seifert
matrices may be seen to be of the forms,
N' ]
p 0
-1 q
Chapter 5 98
and
N N' 0
Nil
P
0
0
M
F
,= 0 q
0
1 0 0 0
0 1 1 0
where the last two rows (and columns) of MF correspond to the simple .,
closed curves a and (3 that bound the domains A,B in r(D) [see
Figures 5.4.8 (c) and 5.4.9(a)], while the last four rows (and columns) of
MF' correspond to the four simple closed curves a, {3, 1,8 that bound
the domains A,B,C, C' in r(D') [see Figures 5.4.8(c) and 5.4.9(b)].
(a) (b)
Figure 5.4.9
If we now subtract the second last row (which corresponds to 1)
from the third last row (which corresponds to {3) of M
F
" and then do
the same to the respective columns, we obtain
N N' 0
Nil
P
0
0
MF'=
-1 q
0
1 0 0 0
0 0 1 0
This matriX, clearly, may be reduced to MF by applying A2"l, and '
s
therefore, MFrvMF/.
We leave it as an exercise for the reader to show the same argument
works when F
2
lies underJ Fl.
99 Seifert Matrices
Exercise 5.4.3. Prove the remaining case, see Figure 5.4.10, required
in the proof of Proposition 5.4.3.
--+
Figure 5.4.10
Proof of Proposition 5.4.4.
We do not wish to burden the reader with yet another turgid proof,
and so we only sketch a proof by means of diagrams, leaving the reader
to flesh out the details at leisure.
Let us work with the surface shown in Figure 5.4.11.
Figure 5.4.11
The ambiguity, and hence the problem arises, when "within" a
disk there are other disks. In our example, Figure 5.4.11, there are
two such disks, F
12
and F
31
The strategy involved in order to prove
this proposition is to replace such a disk by a narrow disk, obtained by
applying Reidemeister moves. This will be done in such a way that if F2
lies over F1; then the resultant disk lies under all the bands connecting
Chapter 5 100
F
1
and F2' Figure 5.4.12{a). However, if F
2
lies under F
1
then the
narrow disk lies over the bands, Figure 5.4.12{b).
(a)
--+
(b)
Figure 5.4.12
Repeating this process, we will eventually obtain a flat surface.
For example, in Figure 5.4.11 suppose that F
21
lies over F
12
, F
31
lies under F
12
, and F
41
, F
42
, F
43
all lie over F
3t
. Then the suitably
transformed (we leave it as an exercise to show this can be done) fiat
surface, F
o
, is shown in Figure 5.4.13.
. Figure 5.4.13
101 Seifert Matrices
It is easy to see that Fa has the same Seifert matrix as F (up to
AI-equivalence), see Figure 5.4.14.
lk(p,a#) =-1
(a)
Ik( a,p*) = 1
(b)
Figure 5.4.14
Exercise 5.4.4. Show that the Seifert matrices M, M' in Exam-
ple 5.3.4 are S-equivalent. (In fact, they are Al-equivalent.)
From a regular diagram of a knot (or link) K we can create a Seifert
surface, and then from this a Seifert matrix. However, to calculate the
Seifert matrix it may be possible to avoid this process, for on occasion
we may construct a Seifert surface that consists of disks and bands, and
on which we place m (= 2g(F) +JL(K) - 1) closed curves. However, the
Seifert graph r obtained from F may not be a plane graph. Therefore,
it is possible that what has gone before does not hold in this case.
However, we bring to our aid the next theorem.
Theorem 5.4.5 [Tr].
Suppose that M
1
and M
2
are two Seifert matrices obtained from
Seifert surfaces F
1
and F
2
of K. Then M
1
and M
2
are S-equivalent.
Chapter 5 102
We now give an example of the type of Seifert surface alluded to in
above discussion.
Example 5.4.1. We may think of the surface F in Figure 5.4.15(b),
which has been constructed from three bands (note that one of these
is knotted) and two disks, to be a Seifert surface for the knot K, Fig-
ure 5.4.15{a). The Seifert graph r of F, however, is not a plane graph,
Figure 5.4.15{c). In spite of this, we may still place two closed curves
Q;1 and a2, Figure 5.4.15{d). It is an easy exercise to calculate the
appropriate linking numbers, and hence the Seifert matrix,
[
-1 0]
M = 1 -1 .
It can be shown that this matrix is S-equivalent to the Seifert ma-
trix obtained from the regular diagram, Figure 5.4.15(a), of K by the
methods described in Sections 1 rv 3.
(a)
(c)
(b)
(d)
. Figure 5.4.15
103 Seifert Matrices
So, to be exact, the Seifert matrix should be said to be constructed
from the Seifert surface, F, of a knot K. In general, however, we shall
simply say that is the Seifert matrix of K. In this vein, we may rewrite
Theorem 5.4.5 as folows: Two Seifert matrices of K are S-equivalent.
However, we should make it quite clear that when we say a Seifert
matrix obtained from a regular diagram D of K, we have in mind that
this Seifert matrix has been constructed from D using the methods
outlined in Sections 1 rv 3.
We shall round of this section by proving two properties of Seifert
matrices. We shall denote by MK the Seifert matrix of a knot (or
link) K.
Proposition 5.4.6.
Suppose that K is an oriented knot (or link) and -K is the knot
with the reverse orientation to K (in the case, of a link, the orientation
is reversed on all the components). Then where is the
transpose matrix of M
K

Proof
If we suppose that D is a regular diagram of K, we may take as a
regular diagram D' for -K, the regular diagram D with all the orien-
tations reversed. Therefore, the orientations of the subsequent Seifert
surfaces are completely opposite. Hence, the under and over relations
for Q;i and at are also completely reversed. The Seifert matrix ob-
tained from D' is therefore just the transpose of that from D. It follows
from Theorem 5.4.1 (or Theorem 5.4.5) that

Proposition 5.4. 7.
Suppose that K* is the mirror image of a knot (or link) K, then

Proof
We can obtain a regular diagram D* of K* from K by changing
the under- and over-crossing segments at each of the crossing points.
Therefore, since the under and over relations for the closed curves that
, S
follow from D and D* are completely reversed, (Compare
this with Examples 5.3.2 and 5.3.3.)

In order to find a knot (or link) invariant from a Seifert matrix, we

need to look for something that will not change under the operations
Al and ~ l defined in Theorem 5.4.1. We will see in this chapter
that the Alexander polynomial is such an invariant. The Alexander
polynomial is not the only important invariant that we can extricate
from the Seifert matrix, the signature of a link can also be defined from
it. In addition to defining these two invariants we shall, in this chapter,
prove some of their basic characteristics. Nota bene, throughout this
chapter we shall asslIme all the knots and links are oriented.
105
1 The Alexander polynomial
Invariants from the Seifert Matrix
For a mathematician it is natural to ask, since we have such a nice
tool as the Seifert matrix, what matrix properties do we know that, via
Al and A
2
, might yield a knot (or link) invariant.
Exercise 6.1.1. Find an example that shows that the determinant,
det M, of the Seifert matrix M of a knot K is not a knot invariant.
However, we should not discard the idea of using the determinant.
Let us, first, symmetrize the matrix M to form the matrix sum M+MT.
If we now look at the absolute value of the determinant of M.+ M
T
,
this does lead to a link invariant.
Proposition 6.1.1.
If M is the Seifert matrix of knot (or link) K, then Idet(M +MT)I
is an invariant of the knot K. This invariant is called the determinant
afK.
Exercise 6.1.2. Find a proof of Proposition 6.1.1. (Hint: Show that
the determinant does not change its value if we apply the operations
Al and A ~ l
We will prove later, in Chapter 11, Section 2, that the determi-
nant of a knot (or link) K is completely independent of the orientation
assigned to K.
This invariant, the determinant of a knot, is quite an old invariant.
One of its useful properties is that since the determinant of the trivial
knot is 1 (we define the determinant of the empty matrix to be 1), it
can and has, over the years, been used to prove that certain knots are
not the trivial knot.
(a) (b)
Figure 6.1.1
Chapter 6 106
(6.1.1)
Example 6.1.1.
Since the determinant of the trefoil knot is 3, it is not equivalent
to the trivial knot. There are knots, however, that have determinant 1
but are not equivalent to the trivial knot; Figures 6.1.1{a) and (b) are
such examples.
The proof of the following proposition we shall postpone until later,
since the proof is an easy consequence of Proposition 6.3.1, which is
proven a bit later.
Proposition 6.1.2.
Suppose that M is the Seifert matrix of a knot (but not a link) K,
then
At this stage, we would like to ask the reader's indulgence as we
jump directly from the above determinant to consider the polynomial,
det(M - tM
T
),
which resembles the characteristic polynomial of M. The determinant
is now a polynomial with indeterminate t. The next logical step is to
examine how this polynomial changes when we apply Al and ~ l
Firstly, since det P = det pT = 1,
det(A
1
(M - tM
T
)) = det[P{M - tMT)pT]
=det(M - tM
T
).
Therefore, it is not affected by the operation At. However, if we
apply A
2
,
det{A
2
(M - tM
T
)) = det b
m
0
-bIt -bmt 0 1
0 0 -t 0
b
1
0
M-tM
T
=det b
m
0
0 0 0 1
0 0 -t 0
107 Invariants from the Seifert Matrix
= t det(M - tM
T
). (6.1.2)
Similarly, we can obtain det(A
2
1
(M
2
- tMf)) = t-
1
det(M
l
-
tMI)
These three formulae lead us to the following theorem.
Theorem 6.1.3.
Suppose that M
1
and M
2
are the Seifert matrices for a knot (or
link) K. Further, if rand s are, respectively, the orders of M
1
and
M
2
, then the following equality holds:
det(Mt - tMI) = t-! det(M
2
- tMi).
Therefore, if M is a Seifert matrix of K and its order is k, then
det(M - tM
T
)
is an invariant of K. This invariant is known as the Alexander polynomial
of K and is denoted by L\K(t).
It follows directly from our previous discussions that k = 2g(F) +
J.t(K) - 1, where as before F is the Seifert surface from which we have
constructed M, and JL(K) is the number of components of the link
K. In most cases, has some terms with a negative exponent;
however, if we multiply by a suitable factor, then we can obtain
a polynomial with only positive exponents. Sometimes it is preferable to
work with such an interpretation of If K is a link with an even
number of components, then k is odd. Therefore, for such links
is a polynomial with terms as powers of ti(= y't) or t-!(= Jt). In
these cases we define (t!)2 = t. [In Appendix (II) we tabulate the
Alexander polynomial of all prime knots with up to 8 crossings.]
We shall next prove an important property of the Alexander poly-
nomial.
Theorem 6.1.4.
Suppose K is a knot; then (t) is a symmetric Laurent polyno-
mial, i.e.,
= +a_(n_l)t-(n-l) +... +an_lt
n
-
1
+ant
n
and
(6.1.3)
[The more general link case is considered in Exercise 6.2.4(2).}
Chapter 6 108
Proof
Suppose that M is a Seifert matrix of K and k is the order of M.
Since K is a knot, k is necessarily even. Therefore,
aK(t-
1
) =t
f
det(M - t-1M
T
) = t-t det(tM - M
T
)
= (-l)kt-t det(M
T
- tM) =t-t det(M - tMT)T
= t-! det(M - tM
T
) = L\K(t).
It is now easy to see that (6.1.3) follows directly from this.

Proposition 6.1.5.
I ~ K -1)1 is equal to the determinant of a knot K.
Proof
IL\K(-l)1 = I(-l)-t det(M + MT)I
= Idet(M + MT)I

Example 6.1.2. If K is a trivial knot, then ~ K t = 1.

Example 6.1.3. IfK is the right-hand trefoil knot (cf. Example 5.3.2),
then
~ K t = t-1(M - tM
T
) = t-
1
det [-(1
1
- t) -t]
-(1 - t)
= t-
1
-1 +t.
Exercise 6.1.3. Evaluate the Alexander polynomial of the knots in
Exercise 5.3.2 and Example 5.3.4.
2 The Alexander-Conway polynomial
The reader will soon find, by experimenting with the above proce-
dure, that. if we wish to use the Alexander polynomial to obtain at least
a partial knot table, the above procedure is quite cumbersome. How-
ever, due to the constant state of flux in knot theory and its interaction
with other disciplines, the S),bove problem can be obviated.
109 Invariants from the Seifert Matrix
In the late 1950s and the 1960s, computers were transformed from
a research project into a research tool. Although the number-crunching
abilities of computers were of tremendous advantage, an extra impetus
was still required to make the Alexander polynomial more computer
friendly. This spark of ingenuity was provided by J.H. Conway in the
late 1960s, when he devised an extremely efficient mechanical procedure
to compute the Alexander polynomial. (With hindsight, if we carefully
reread Alexander's original paper, it is possible to glean from it Con-
way's method. So perhaps, rather like in the case of fractals, this is a
case of technology catching up with mathematical theory.)
Definition 6.2.1. Given an oriented knot (or link) K, then we may
assign to it a Laurent polynomial, VK(Z), with a fixed indeterminate
z, by means of the following two axioms:
Axiom 1 If K is the trivial knot, then we assign VK(Z) =1.
Axiom 2 Suppose that D+, D_, Do are the regular diagrams, respec-
tively, of the three knots (or links), K+, K_, K
o
. These reg-
ular diagrams are exactly the same except at a neighbour-
hood of one crossing point. In this neighbourhood, the reg-
ular diagrams differ in the manner shown in Figure 6.2.1.
(Note: In the case of D+ (D_) within this neighbourhood,
there exists only a positive (negative) crossing.)
,:g\
lX"
(j---r)
: ~
: j/ : , ,
, .
\ #
, #
~ ..-.,., ~ ~ ..... -."
~ ~ ..-."
D+
D_
Do
Figure 6.2.1
Then the Laurent polynomials of the three knots (or links) are
related as follows:
(6.2.1)
The three regular diagrams D+, D_, Do formed as above are called
skein diagrams, and the relation, (6.2.1), between the Laurent polyno-
mials of K+, K_, K
o
(whose regular diagrams these are) is called the
skein relation. Also, an operation that replaces one of D+, D_, Do by
the other two is called a skein operation.
Chapter 6 110
We shall write VD+ (z) instead of VK+ (z), et cetera, since there
is no need to distinguish between the knots K+, K_, K
o
and their
respective regular diagrams D+, D_, Do.
Exercise 6.2.1. Show that if K+ is a j.t-component link, then K_
is also a j.t-component link, but K
o
is either a (J..L - 1)-component or
a (/1 +1)-component link.
The polynomial VK(Z) , defined as above, is called the Conway
polynomial. To actually show that the Laurent polynomial VK(Z) ,
obtained from Axioms 1 and 2, is well-defined and unique is quite trou-
blesome (a complete proof can be found in [LM]). However, if we assume
the well-definedness and uniqueness of VK(Z) , then by proving the fol-
lowing theorem, we can show that VK (z) and the Alexander polynomial
are essentially the same.
Theorem 6.2.1.
In other words, if we replace z by .Jt - Jt in the Conway poly-
nomial, the resultant transformation yields the Alexander polynomial.
Due to this relationship, VK(Z) is often called the Alexander-Conway
polynomial.
Proof
We have already inadvertently shown in Example 6.1.2 that the
Alexander polynomial satisfies Axiom 1; therefore, we need only prove
that it also satisfies Axiom 2, taking into account the substitution z =
.Jt - Jt.
Figure 6.2.2
Let us first consider the skein diagrams, Figure 6.2.2. If F+, F_,
F
o
are the Seifert surfaces that correspond to D+, D_, Do, and
r +, r _, r 0 are the corresppnding graphs, then we may, by using the
111 Invariants from the Seifert Matrix
methods of the previous chapter, determine the respective Seifert ma-
trices, M+, M_, Mo.
The crossing point of D+ (respectively D_), see Figure 6.2.2,
corresponds to a positive (negative) band in F+ (respectively F_ ),
while in the graph r + (r_) the crossing point corresponds to a positive
(negative) edge e+ (_), see Figure 6.2.3.
Figure 6.2.3
Let f + (f- ) be the domain of r+ ( r _) that contains the edge
e+ (e_). Suppose e+ (e_) is one of the common boundary edges of
1+ (/-) and Figure 6.2.3. It is possible that 1+ and are
the same face (and so are 1- and 1'- ). In this case r0 is disconnected
and Do is not connected. Definition 5.4.2 and Proposition 5.4.2 then
imply that = 0 and VKo(Z) = O. On the other hand, in this
case K+ and K_ are equivalent, since they are the connected sum of
two knots (or links). Therefore, aK+(t) = aK_(t), and hence aK(t)
satisfies Axiom 2. Therefore, we suppose that 1+ (I_) and
are different. Now let the order of M+ and M_ be k. We may then
assume that 1+ (/-) and correspond to the (k - l)th row
(and column) and the last k
th
row (and column), respectively. Since
1+ (I-) and (/'-) in r + (r-) are "amalgamated" to form r 0, M
o
is of order k -1. Now, let ai,j(+), ai,j(-), ai,j(O) denote, respectively,
the entries of M+, M_, Mo. Then these entries are related as follows,
Figure 6.2.4:
Figure 6.2.4
(A) If i,j i= k - 1, k, then
(1) ai,j(+) = ai,j(-) = ai,j(O).
Chapter 6
(B) If i # k - 1,
(1) ai,k-l (0) =ai,k-l (+) + ai,k-l (- )
(2) ak-l,i(O) = ak-l,i (+) + ak-l,i(-).
(C) If i = k - 1, or k,
(1) ai,i(+) = ai,i(-)-1
(2) ak-l,k(+) = ak-l,k(-) + 1
ak,k-l(+) = ak,k-l(-) + 1
(3) ak-l,k-l(O) = L {ai,j(+) +ai,j(-)}.
i,j=k-l,k
Exercise 6.2.2. Show that (6.2.2) actually holds.
Using (6.2.2), a further simple calculation shows that
112
(6.2.2)
det (M+ - tMJ) - det (M_ - = (-1)(1 - t) det (M
o
- tMl).
So if we use this to calculate we obtain
(t) = det (M+ - tMJ)
= det (M_ - + (t - 1) det (M
o
- tMJ-')
= (t) + (t! -
This now completes the proof of Theorem 6.2.1.

Before using the skein relation to calculate the Alexander polyno-

mial, we shall prove the following proposition:
Proposition 6.2.2.
The Alexander polynomial of the trivial link with p,-components
( J.L 2:: 2) is O. (Henceforth, we shall denote the trivial link by 0 Jl.' )
Proof
The skein formula corresponding to the skein diagrams in Fig-
ure 6.2.5 is
V'o+ (z) - V'o_ (z) = zV'oo(z).
Since both D+ and D_ are trivial (J.L - I)-component links,
VD+ (z) = Vo_ (z). 0 =ZV'Do (z), i.e., VO
o
(z) = O.
113 Invariants from the Seifert Matrix
y
Jl-2 copies
y
Jl-2 copies
y
J.1-2 copies
Figure 6.2.5

Proposition 6.2.2 is also an immediate consequence of Defini-

tion 5.4.2, since the Seifert matrix of 011- contains the zero matrix of
order J.L - 1.
Usually, the most effective way to calculate the Conway polynomial
is to make use of the skein tree diagram. Since it is a calculating aid, it
is best illustrated/defined by means of an example.
To facilitate our next set of calculations we shall rewrite (6.2.1) as
follows:
V'o+ (z) = V'o_ (z) + zV'oo (z) }
V'o_ (z) = V'o+ (z) - zV'oo (z)
(6.2.3)
~ x n n p l 6.2.1.
Suppose that K is the right-hand trefoil knot and D is a regular
diagram of K, which in Figure 6.2.6 is the topmost diagram.
Within the dotted circle on D, we will perform a skein operation.
Since within this circle the crossing point is positive, it is better to re-
name this regular diagram D+. By performing a skein operation, D+
is transformed into two other regular diagrams: one, D_, is the regu-
lax diagram obtained by changing the original positive crossing point to
a negative crossing point; the other, Do, is obtained by removing the
positive crossing point (by splicing the regular diagram at this cross-
ing point), see Figure 6.2.6. Now connect D+ and D_ (and, similarly,
Chapter 6 114
connect D+ and Do) by drawing a line segment and assign +1 (respec-
tively z) to the line segment, see Figure 6.2.6. The appropriate assign-
ment follows directly from (6.2.3), namely, the coefficients of VD_ (z)
and VDo(Z). So for our skein tree diagram we have our first pair of
branches, and they correspond to 1VD_ (z) +ZVDo (z) in the evaluation
of VD+(Z).
Figure 6.2.6
It is straightforward to see that D_ is equivalent to the trivial knot,
and hence VD_ (z) = 1. Therefore D_ will not produce' any further
branches, i.e., we cannot perform a subsequent skein operation here.
However, Do is not equivalent to the trivial knot or links, and so we
can again perform a skein operation within another dotted circle. Since
within this circle, see Figure 6.2.6, the crossing point is also positive, let
us rename Do as D+. Then, as before, let us denote the subsequent
left-hand diagram by D_ and the right-hand diagram Do. Again, we
draw line segments and assign to them coefficients by means of (6.2.3),
see Figure 6.2.6. It is easy to see that D_ and Do are, respectively, the
trivial 2-component link and the trivial knot. Hence no further branches
may be formed and our skein tree diagram for K is complete.
VK (z) can now be calculated as the sum of the Conway polynomial
of the each terminating trivial knot (or link) multiplied by the product of
the coefficients (on the line segments) along the (uniquely determined)
branch path that begjns with our original regular diagram, D, of K and
terminates, by construction, with the regular diagram of this trivial
115 Invariants from the Seifert Matrix
knot (or link).
Therefore, in the above example we obtain the following sum:
\7K(Z) = 1\7o(z) + z\70 o(z) + z2\7o(z).
Since \7o(z) =1 and \70 o(z) =0, the calculation collapses down
to \7K(Z) = 1 +z2. Therefore, by applying Theorem 6.2.1 we obtain
. 1
~ K t = 1 + (Jt - Jt)2 =t-
1
-1 +t.
Example 6.2.2.
The skein tree diagram for the Conway polynomial for the figure 8
knot is given in Figure 6.2.7.
Figure 6.2. 7
The following calculation is a direct result of this skein tree diagram:
Therefore,
( )
. h 1 2 -1
~ K t = 1 - (vt - Vi") = -t + 3 - t.
Chapter 6 116
O ~
From the above two examples it is possible to surmise how in gen-
eral we can calculate the Conway polynomiaL We start with the regular
diagram D of a knot (or link) and then perform on it an unknotting oper-
ation at one of the crossing points. This produces a new regular diagram
D' of a "simpler" knot. Continuing this process, we finally arrival at a
set of regular diagrams of trivial knots and links. (The reader should
consult Proposition 4.4.1 to see why, in fact, this is possible; however,
a small caveat, in this case the number of crossing points of Do is one ,1
less than for D+ and D_.)
Exercise 6.2.3. Determine the Alexander and the Conway polyno- '
mials for the Borromean rings (Figure 1.4.5), and the Whitehead link
(Figure 1.4.4) and the knot in Figure 5.1.5(a).
Exercise 6.2.4. (1) Show that if K is a knot, then VK(Z) is an
integer polynomial in z2.
(2) Show that if K is a jL-component link, then we '
may write VK(Z) = zp-lg(z), where g(z) is an integer polynomial in '
z2.
In this section we have described an extremely efficacious method .
to calculate the Alexander polynomial for an arbitrary knot (or link). In
fact, this method has been taken up with great gusto since it was first
introduced by Conway, and innumerable Alexander polynomials have
been calculated. This method is still very powerful if in our research
we want to calculate the Alexander polynomial of a specific knot (or
link). By the 1960s, however, the Alexander polynomial had already
been used to the point of exhaustion to detect Global properties of
knots (or links). That is to say, it would seem to be a futile exercise to 0
just carryon calculating the Alexander polynomial for knots (or links)
with arbitrary large number of crossing points, since it is very possible
that no further insight into the Global properties of knot (or link) will
be garnered by so doing. Around 15 years after Conway introduced his
method, it was shown that in addition to the Alexander polynomial, his
approach is very useful in the calculation of the "new" knot invariants.
3 Basic properties of the Alexander polynomial
In this section we shall prove several important properties of the
Alexander polynomial. T h ~ first of these properties is related to Propo-
117
sition 6.1.2.
Invariants from the Seifert Matrix
Proposition 6.3.1.
If K is a knot, then a
K
(l) = 1.
Proof
We know, due to Theorem 6.2.1, that if we set t = 1 then
~ K 1 = VK(O). On the other hand, by (6.2.1), if we set z = 0 then we
obtain VD+ (0) =VD_ (0). Therefore, even if we perform an unknotting
operation to K, the value of VK(O) remains unchanged. However, by
performing the unknotting operation several times on K, eventually K
is transformed into a trivial knot (Proposition 4.4.1). From the above
observations we may deduce, VK(O) = Vo(O) = 1.

Exercise 6.3.1. Show that if L is a p,-component (J..L 2:: 2) link, then

~ L 1 = o.
The reader may have noticed that if we juxtapose Proposition 6.3.1
and Theorem 6.1.4, then these actually seem to characterize the Alexan-
der polynomial, as seen in Theorem 6.3.2.
Theorem 6.3.2.
Suppose that f(t) is a Laurent polynomial that satisfies the follow-
ing two conditions:
(1) f(l) = 1
(2) f(t) = f(t-
1
).
(6.3.1)
Then there exists a knot that has as its Alexander polynomial
f(t). Equivalently, if g(z) is an integer polynomial in Z2 with g(O) = 1,
then there exists a knot K that has as its Alexander-Conway polynomial
g(z).
The proof requires finding an appropriate orientable surface, F,
with its Seifert matrix M of order k satisfying t -! det (M- tMT) = f(t ).
The reader may wish to try and develop a proof or consult [Sei] for more
details.
Example 6.3.1. In Figure 6.3.1 we have drawn one of the knots,
which has as its Alexander polynomial f(t) = 2t-
2
-IOt-
1
+ 17- lOt +
2t
2
(= 1 - 2z
2
+2z
4
).
The Alexander polynomial is particularly useful with regard to the
Global problem of the classification of knots (or links). However, it
Chapter 6
118
is powerless in deciding the Local problems of invertibilty and am-
phicheirality.
Figure 6.3.1
Theorem 6.3.3.
Suppose that K is a knot.
(1) H -K is the knot obtained from K by reversing the orientation
on K, then
=
(2) If K* is the mirror image ofK, then
=
Proof
(1) From the proof of Proposition 5.4.6, we may assume that
M-K = MR If MK is of order 2g, say, then M-K is also of order
2g. So,
=t-
g
det(MK -
=t-
g
- tM-K)
=t-
g
det(M-
K
- tM':K)T =
(2) Since - MR and utilizing the proof of Proposition 5.4.7,
we may assume that MK- = -MR. This calculation now follows along
similar lines to (1),
(t) =t-
g
det(MK* - tMJ'-)
=t-
g
+tMK)
= (-1)2
g
t -
g
det(MK - tMJ)T =

119 Invariants from the Seifert Matrix

We know that the Alexander polynomial of the trivial knot is 1; the
converse, however, does not hold, i.e., there exist non-trivial knots that
have Alexander polynomial 1. An example of one such knot is given in
Figure 6.3.2; this knot is known as the Kinoshita- Terasaka knot.
KT knot
Figure 6.3.2
The Alexander polynomial of the trivial Jt-component (Jt ~ 2)
link has been shown to be o. We may also generalize this result.
Definition 6.3.1. The regular diagram D of a link L that is a compo-
sition of the regular diagrams of two links with no points in common is
said to be split, see Figure 6.3.3. A link L that has a split regular dia-
gram is said to be a split link. (The link in Figure 6.3.3 is a 3-component
split link.)
L
Figure 6.3.3
Proposition 6.3.4.
If L is a split link, then ~ L t = o.
Proof
Suppose that D is a split regular diagram of L. Then the terminal
point of the skein tree diagram of D is always a trivial link; it can never
terminate in a trivial knot. (Why is this the case?) Therefore, it follows
from Proposition 6.2.2 that VL(Z) =0, i.e., ~ L t = o.

Chapter 6 120
Exercise 6.3.2. If ~ L t = 0 it does not necessarily follow that L
is a split link. Show that the Alexander polynomial of the link in Fig-
ure 4.5.5(b) is o.
Theorem 6.3.5.
Suppose K
1
#K
2
is the connected sum of two knots (or links) K
1
and K
2
, then
Proof
Firstly, create in the prescribed way the Seifert surfaces F1 and
F
2
of, respectively, K
1
and K
2
Then the orientable surface formed by ~
joining these surfaces by a band becomes a Seifert surface for K
1
#K
2
, j
see Figure 2.1.7. If we suppose M1 and M
2
are the Seifert matrices of J
K
1
and K
2
obtained from F
1
and F
2
, then M the Seifert matrix of ,:.
K
1
#K
2
has the following form:
Therefore,
Theorem 6.3.5 follows immediately from this equality.

If L is a It-component link, then we may write VL(Z) =zJL-lg(z),

where g(z) is an integer polynomial in z2; cf. Exercise 6.2.4(2).
So, if we let
,.., .1: 1
ddt) =g(vt - yt),
then
and thus ~ L t is a symmetric integer polynomial. This polynomial is
called the Hosokawa polynomial. With regard to the Hosokawa polyno-
mial for links, the followblg theorem, similar to Theorem 6.3.2, holds:
121 Invariants from the Seifert Matrix
Theorem 6.3.6 [H].
Suppose f(t), a Laurent polynomial, is symmetric, i.e., f(t) =
f(t-
1
). Then there exists a link, with an arbitrary number of compo-
nents, that has as its Hosokawa polynomial f(t). [Note that f(l) can be
an arbitrary integer.]
Exercise 6.3.3. Calculate and verify that the Hosokawa polynomial
of the 2-component link L in Figure 6.3.4 is ~ L t = t-
1
+3+t (= 5+z
2
).
Find a 3-component link L' with the same Hosokawa polynomial as L.
(Hint: The Alexander-Conway polynomial of the Hopf link is z, and use
Theorem 6.3.5.)
Figure 6.3.4
In general, the problem of determining the genus of a knot is quite
difficult. However, it is possible to estimate, and for some cases to
determine, the genus of a knot using the Alexander polynomial.
Suppose for a knot, K, of genus g(K), F is a Seifert surface for
K with this particular genus. Then the order of the Seifert matrix M
calculated from F is 2g(K). Therefore, the maximum degree of t in
~ K t is at most g(K). We can put this result on a more formal footing
in the following theorem:
Theorem 6.3.7.
Suppose that the genus of a knot K is g(K), then the maximum
degree of t in the Alexander polynomial cannot exceed g(K).
In particular, if det M =1= 0 then since det(M - tM
T
) is a Laureq,t
polynomial of exactly degree 2g, we obtain the next proposition.
Proposition 6.3.8.
A necessary and sufficient condition for the maximum degree oft in
~ K (t), the Alexander polynomial ofK, to be exactly g(K) is det M =1= o.
Chapter 6 122
Using Proposition 6.3.8 we can prove that the genus of an alternat-
ing knot is equal to the maximal degree of its Alexander polynomial; cf.
Chapter 11, Section 5.
4 The signature of a knot
In this section, we shall define still one more important invariant
that depends on the Seifert matrix M, the invariant in question is the '
signature of a knot. As was noted in Proposition 6.1.1, the absolute
value of the determinant of the symmetric matrix M +M
T
is a knot
invariant. However, this is not the only knot invariant that can be
formed with this symmetric matrix as its core element.
A well-known result in linear algebra allows us to diagonalize any
symmetric matrix that has in all its entries real numbers. More precisely,
Theorem 6.4.1.
Suppose A is a n x n symmetric matrix with its entries real num-
bers. Then it is possible to find a real (with its entries real numbers)
invertible matrix P such that PApT = B is a diagonal matrix. In .,
addition, we may assume that detP = 1.
We may rephrase the essence of this theorem in the terminology
of Chapter 5 as follows: A symmetric matrix is Al-equivalent to a ,
diagonal matrix.
Instead of giving a proof of Theorem 6.4.1, which is a bit tedious
and will not shed any insight in what follows, we propose to illustrate
the method of diagonalizing a matrix by means of a couple of exam-
ples. These examples will, hopefully, indicate to the reader the idea
of the proof, and thus the proof will become only an exercise in (the
manipulations of) linear algebra.
Example 6.4.1.
(a) Let us diagonalize the matrix,
A = [ ~ ; ~ ]
-1 1 3
First interchange the first row and the second row, and then also
interchange the first and second columns. By this two manipulations,
123 Invariants from the Seifert Matrix
the new matrix Al will not have 0 in its (1, I)-entry:
Next, we will apply several AI-operations to eliminate all the en-
tries in the first row and column except the {l,l)-entry.
[
20
Al -+ ~
o ~
2
Next, consider the 2 x 2 matrix, with non-zero elements, which
forms the lower right-hand block of A
2
On this 2 x 2 matrix apply,
in a similar way to our previous step, several AI-operations, so that
within this 2 x 2 matrix, also, all the entries in the first row and column
except the {l,l)-entry are eliminated (in fact, we need only eliminate
two entries):
[ 2
0
~ ] =B. A2- ~
1
-2'
0
(b) Let us next diagonalize:
A= [ ~
~ ] .
Since all the diagonal entries are 0, we cannot apply the previous
method. However, by adding the second row to the first row, and then
adding the second column to the first column, we can get a new matrix,
AI, with non-zero (1, I)-entry:
We may now diagonalize Al as in (a), and so
Chapter 6 124
[
21
A= 1 0
2 3
o -1
6.4.1.
(a) Show that if (at least) one of the diagonal entrief? of a n x n
symmetric matrix is positive, then O'(A) > -no
(b) Diagonalize A, given below, and thereupon calculate IT(A) :
-
1 -2
Returning, now, to Theorem 6.4.1, suppose that the entries in the
diagonal of B are aI, a2, ... , an' We may lift the following two, seem-
ingly innocuous, definitions directly from their linear algebra roots.
(1) The number of at, a2, ... , an that are zero
is called the nullity of A, and is denoted by n(A). (6.4.1)
(2) (The number of positive ai) - (the number of negative ai)
is called the signature of A, and is denoted by O'(A).
The signature, u(A), and the nullity, n(A), depend, as the notation
implies, only on A, completely irrespective of the P chosen. The sig-
nature may also, although this does not follow directly from the above .,1
definition, be obtained from the following:
u(A) = (the number of positive eigenvalues of A) - \;

(the number of negative eigenvalues of A).


Our intention is to prove that the signature of M +MT is, in fact, I
a knot invariant, but first we need to prove a general theorem.
Theorem 6.4.2.
If N1 and N2 are two integer square matrices that are S-equivalent
then
n(N
1
+ N;) =n(N
2
+
u(Nl +Nr) =u(N2 +Nf)
Proof
To avoid getting tangled up in notation, let N = N1 + Nf and
N' = N
2
+ Nr. In order to prove the theorem, we need to show the
following two properties:
(1) If N
2
=A
1
(N
t
), then n(N) =n(N') and O'(N) = u(N')
(2) If N
2
=A
2
(N
1
), then n(N) =n(N') and u(N) =O'(N').
(6.4.2)
125 Invariants from the Seifert Matrix
Proof of (1)
If N
2
= A1(N
1
), then we can write N
2
= PN1PT. Therefore,
In addition to this, we know that P is an invertible matrix. Using
these facts, the required results, namely, n(N') = n(N) and u(N') =
O'(N), follow directly from well-known results in linear algebra.
Proof of (2)
If N
2
=A
2
(N
1
), then we can write
b
l
0 0 0
N
1
N
1
N2 =
b
m
0
or
0 0
0 0 0 1
b
l
b
m
0 0
0 0 0 0
0 0 1 0
So,
b
l
0
N
N' =N
2
+Ni =
b
m
0
b
l
...
b
m
0 1
0
...
0 1 0
Now, if we choose our invertible matrix P judiciously (we leave this
as a straightforward exercise for the reader), we can obtain
PN'p
T
= [ ::
Therefore,
n(N') = n(N) + n [ ~ ~ ] u(N') =u(N) + u [ ~ ~ ]
However,
det [ ~ ~ ] =-1 # 0, so n [ ~ ~ ] =0,
Chapter 6 126 ~
~
and hence n(N') = n(N). !J',
In the case of the signature, we have already shown in Exam-
ple 6.4.1(b) that we may diagonalize [ ~ ~ ] as [ ~ _ ~ ] .
It follows immediately from this that 0- [ ~ ~ ] = o. Therefore,;j
also, u(N') =u(N).

Theorem 6.4.3.
Suppose M is the Seifert matrix of a knot (or link) K. If we set
A = M + M
T
, then n(A), the nullity of A, and u(A), the signature
of A, are invariants of the knot (or link) K. Hence, we can write u(K)
instead of u(A), and this is called the signature of the knot (or link)
K. Similarly, we can write n(K) instead of n(A), and this is called the
nullity of K.
Proof
Since the two matrices M
1
and M
2
ofK are S-equivalent, it follows
from Theorem 6.4.2 that the signature and nullity of M
1
+ M'f and
M
2
+MI are equal. The theorem follows directly from this result.

Theorem 6.4.4.
If K is a knot, then n(K) = 0 and u(K) is always even.
Proof
The Seifert matrix, M, for K is a square matrix of even order.
Further, due to Proposition 6.3.1, since det(M - M
T
) = ~ K l ) = 1,
det(M + MT) is an odd integer and so non-zero. Consequently, n(M +
M
T
) = 0, and n(K) = o. Therefore, the number of eigenvalues of
M +MT that are not zero is even; hence, u(M +M
T
) is also even.

Example 6.4.2. The Seifert matrix, M, of the right-hand trefoil knot,

calculated in Example 5.3.2, is
[
-1 0]
1 -1
Therefore,
127 Invariants from the Seifert Matrix
We may diagonalize this in the same way as our previous examples
to obtain
-!];
so u(K) = -2. Similarly, the signature of the left-hand trefoil knot,
K', may be shown to be 2.
Calculating the signature is a very effective way of rooting out the
local properties of a knot, and hence it has had a profound influence in
the solution of Local problems. As probably the most obvious example
of this, we shall look at the problem of amphicheirality. To distinguish
a knot K from its mirror image, K*, is, in general, a very difficult
problem. To show that the trefoil knot and its mirror image are not
equivalent originally required quite an elaborate proof by Dehn. How-
ever, this follows immediately from the next theorem.
Theorem 6.4.5.
The signature of a knot (or link) K has the following properties:
(1) u(K
1
#K
2
) =u(K
1
) +u(K
2
).
(2) If K* is the mirror image ofK, then u(K*) = -u(K). (6.4.3)
(3) If -K is obtained from K' by reversing the orientation on K,
then u(-K) = u(K).
Proof
(1) It was shown in the proof of Theorem 6.3.5 that MKl#K2 has
the following form:
So,
u(K1#K2 ) = U(MK
1
#K2 +
= U(M
K1
+ + U(MK2 +
=u(K
1
) + u(K
2
).
(2) From Proposition 5.4.7 we know that MK* and -MR are S-
equivalent, so by Theorem 6.4.2 we have
u(K*) =U(MK* +
= u(-MR - MK)
= (-l)u(MK +
=-u(K).
Chapter 6 128
(3) We leave the proof of this part as an exercise for the reader.
,I'
"
'"
Proposition 6.4.6.
If a knot (or link) K is amphicheiral then u(K) =o.
",
.J:
. ',:'
Proof
If K is amphicheiral then K and K* are equivalent. Consequently,
u(K) =u(K*). On the other hand) from (2) of (6.4.3) u(K*) = -u(K).
So, u(K) =u(K*) =-u(K). Hence, 17(K) =o.
Example 6.4.3. If K is the right-hand trefoil knot, we have shown
u(K) = -2. Consequently, K cannot be amphicheiral.
Exercise 6.4.2. Let AK(t) be the Alexander polynomial of a knot
(but not a link) K. Show
(6.4.4)
a(K)
-1 = (-1) 2 ,
Example 6.4.4. The signature of the square knot [Figure 1.5.6(a)] is
zero, while the signature of the granny knot [Figure 1.5.6(b)] is -4. So, .,
these two knots are not equivalent (cf. Exercise 1.5.2). It can be shown '::;;
that the square knot is, in fact, amphicheiral.

if!


,'\
where -1) is just the sign of -1), Le., either +1 or
-1.
Using (6.4.4), show
a{K)
== (-1) 2 (mod 4). (6.4.5)
Show the above formulae hold in practice in the case of the trefoil I
knot and the square knot.
It follows from (6.4.5) that a knot with == -1 (mod 4)
cannot be amphicheiral. A more detailed discussion of the relationship
between and q(K) can be found in a paper by J. Milnor [Mil.
In particular, he proves that if =1 then u(K) =O.
To calculate u(K) using the methods so far described requires the
calculation of the Seifert matrix. However, we may avoid this by using
the skein formula.
129 Invariants from the Seifert Matrix
Theorem 6.4.7 [G].
Suppose K is a knot (but not a link) and D is a regular diagram
for K. Then u(K) can be determined by means of the following three
axioms.
(1) If K is the trivial knot, then u(K) = O.
(2) If D+, D_, Do are the skein diagrams, then
u(D_) - 2 ~ u(D+) ~ u(D_)
(3) If ~ K t is the Alexander polynomial ofK, then
sign(Ll
K
( -1)) = (-1) < 7 ~ K }
(6.4.6)
Since we will not give a proof of the above theorem, we would like
to make a few short comments on (2) and (3) of (6.4.6). Firstly, because
K is a knot (but not a link), D+ and D_ are regular diagrams of knots.
Since Do is a link, u(Do) cannot be calculated from these two regular
diagrams. Moreover, from Theorem 6.4.4, we know that O'(K) is always
even, which means that u(D+) cannot be O'(D_) - 1. Therefore, we
may rewrite (2) of (6.4.6) as
(2)' O'(D+) =O'(D_) or O'(D+) = u(D_) - 2.
Further, (6.4.6) part (3) is exactly (6.4.4). Inconveniently, there are
no known similar skein formulae that allow us to evaluate the signature
of a link.
Let us illustrate how Theorem 6.4. 7 may be used to calculate the
signature of the figure 8 knot.
o
Figure 6.4.1
Example 6.4.5. If we consider the skein diagrams in Figure 6.4.1, it
is easy to see that D_ is the trivial knot, so from (2)' of (6.4.6),
Chapter 6 130
On the other hand, since dK(t) = -t-
1
+3-t (cf. Example 6.2.2),
We may now substitute this positive value into (3) of (6.4.4), and
so obtain the equality 1 == (-1) < 7 ~ K It then follows immediately that
a(K) must be zero, rather than -2. Since it is known that the figure 8
knot is amphicheiral, this is the expected result.
Exercise 6.4.3. Calculate the signature of the right-hand and left-
hand trefoil knots using Theorem 6.4.7.
Exercise 6.4.4. Confirm that the signature in Figure 6.4.2 is zero.
This knot, however, is known not to be amphicheiral.
Figure 6.4.2
If we perform a single unknotting operation on a knot K, then we
change D+ to D_ or D_ to D+. It follows from (2) of..(6.4.6) that the
signature is unchanged or changes only by 2. If u(K) is the unknotting
number of K, then by performing the unknotting operations u(K) times
on K, we will transform K to the trivial knot. Since the signature of the
trivial knot is zero, the following theorem holds:
Theorem 6.4.8.
If u(K) is the unknotting number of the knot K, then
la(K)1 ~ 2u(K).
(This theorem also holds in the case of links; however, an alternative
proof to the knot case is required.)
Example 6.4.6. The signature of the knot, K, in Figure 6.4.3 is -4.
So, lu(K)1 = 4 ~ 2u(K), and hence u(K) ~ 2. However, only two
knotting operations are required to transform K to the trivial knot;
hence, u(K) :c:; 2. From these two facts it follows that u(K) = 2.
131 Invariants from the Seifert Matrix
Figure 6.4.3
Sadly, Theorem 6.4.8 is not sufficient if we wish to determine the
unknotting number of an arbitrary knot. Except for particular classes
of knots, see Chapter 7, a general method for determining u(K) is not
known.
However, u(K) is one of the few knot invariants that gives a nec-
essary condition for a knot to be a slice knot.
Theorem 6.4.9 [Mus3].
If K is a slice knot, then u(K) = o.
It is also possible to use the Alexander polynomial to obtain con-
ditions for a knot to be a slice knot [FM].
Exercise 6.4.5. Calculate the signature of the knot in Figure 6.1.1(a)
and determine u(K) for it. Do the same for the knot in Figure 6.1.1(b).
Exercise 6.4.6. If L is a J.t-component link show that n(L) :5 J.t - 1.
Exercise 6.4.7. Let L be an oriented J.t-component (JJ ~ 2) link.
Let L' be the link obtained from L by reversing the orientation of one,
say, tL
th
component. Then it is known [Mus4] ~
17(L) +lk(L) =17(L') + lk(L'), (6.4.7)
where lk(L) is the total linking number afL (for the definition see Chap-
ter 4, Section 5). Therefore, e(L) = a(L) +Ik(L) does not depend on
the orientation of L, Le., e(L) is an invariant of an unoriented link L.
Nota bene, 17(L) may change considerably when the orientation of some
of the components are reversed, since a(L) depends on the Seifert sur-
face. Finally, the problem is as follows: Confirm that (6.4. 7) holds for
the link in Figure 4.5.5(a).
,"". /l"".
If we take two knots (or links) at random, what we would like to
have is an efficient method that will determine for us whether or not
they are equivalent knots (or links). In general, sadly, such an efficient
method has yet to be discovered. So, at present a concise classification
of knots is not possible. The next most obvious step is to try to group
together knots (or links) with a particular property or properties in
common, and then try to classify them. In fact, the techniques we have
already discussed are sufficient for us to extract the characteristics of
certain particular types of knots.
133 Torus Knots
In this chapter, using our aforementioned techniques, we shall in-
vestigate torus knots, which form one such set of knots (and links) with
certain properties in common. The torus knots are not only interesting
in themselves, but have a further importance in that it is often pos-
sible to gain insight into the general properties of knots (or links) by
extrapolating the characteristics of the torus knots.
Although torus knots can be completely classified by our already
established methods, we should add a caveat that these methods are
not sufficient to totally determine all the properties of torus knots. For
example, it is only quite recently that the general formula for the un-
knotting nwnber of torus knots has been established. This case shows
what we have already mentioned that the solution of Local problems
does not follow automatically from the solution of the Global problem.
1 Torus knots
A knot (or link) is a torus knot if it is equivalent to a knot (or
link) that can be drawn without any points of intersection on the trivial
torus. The trivial torus is a solid T obtained by rotating around the
y-axis the circle m: (x - 2)2 +y2 = 1, on the xy-plane, which has as
its centre the point (2,0), radius 1 unit, Figure 7.1.1.
y
o
Figure 7.1.1
An alternative way to construct the trivial torus is to take a cylinder
in R3 with the unit circle C
1
as its base and the unit circle C2 as
its top, Figure 7.1.2(a). We now glue together C
t
and C
2
in R3
so that C, the central axis of the cylinder, becomes the trivial knot,
Figure 7.1.2(b). [Nota bene, if we glue C
1
and C
2
in the way shown
in Figure 7.1.2(c), then the subsequent C is not equivalent to the trivial
knot and hence the torus is not the trivial torus.]
A knot (or link) that lies on the trivial torus' is said to be a torus
knot, and we can express it in terms of a certain standard form.
Chapter 7 134
c
c
(a) (b) (c)
Figure 7.1.2
Let us use more concretely the above cylinder with height 1 unit and
as its base a unit circle in xy-plane. We may assign to the base C
1
and
to the top C
2
the r points A
o
, AI, ... , A
r
-
I
and B
o
, B
I
, ... , B
r
-
I
,
respectively. The co-ordinates are as follows, see also Figure 7.1.3(a):
A (1 0 0) A
- ( 21r 21r 0)
0= " , 1 - cosr,sln
r
, , ... ,
A
- ( 2(r-I)1r. 2(r-l)1r 0)
r-l - cos r ,Sin r '
B (1 0 1) B
- ( 211' 21r 1)
o = " , I - cosr,Sln
r
, ,... ,
B
= (c S 2(r-l)1r sn 2(r-l)1r 1)
r-l 0 r ,I r '
(c) (a) (b)
Figure 7.1.3
Let us now connect the point Ak and Bk (k = 0,1, ... , r -1) on
the cylinder by the segments ak. Next, keeping the base, C
1
, fixed, let
us give the whole cylinder a twist by rotating the top about the z-axis
by an angle of (In this case, q is either a positive or negative
integer.)
135 Torus Knots
In Figure 7.1.3(b) the case q = 2 and r = 3 is shown, while in
Figure 7.1.3(c) the case q = -2 and r =3 is shown.
Finally, let us identify (Le., glue in a very natural way) the point
(x,y,O) of 0
1
to the point (x,y,l) of O
2
(as before, the centre
C becomes the trivial knot). This creates a single trivial torus T,
with the r segments 00, ai, ... , a
r
-1 that have been transformed
into a knot (or link) on its surface. This knot (or link) is called a
(q, r)-torus knot (or link) and is denoted by Kq,r, Figure 7.1.4.
Figure 7.1.4
Let us once again consider the circle m that we used originally to
form the trivial torus T. FUrther, suppose K is the boundary of a disk
that lies on T, Figure 7.1.5.
Figure 7.1.5
These trivial knots m and K, or a trivial link consisting of some
of these trivial knots on T, is not a torus knot as outlined above. In
fact, we shall consider these type of knots (or links) to correspond to
the case r = 0, and we say that m is a (l,O)-torus knot, K
1
,o. The
other knot, K, we shall call a (O,O)-torus knot, Ko,o.
Exercise 7.1.1. Suppose that n is the greatest common divisor
(g.c.d) of q and r. Show that Kq,r is a n-component link.
Exercise 7.1.2. Given a torus knot (or link), that does not contain
within its components the (1,0)-torus knot and/or (O,O)-torus knot
Chapter 7 136
(also excluding orientation), show that it is equivalent to some
(q, r)-torus knot (or link).
In the above exercise we excluded orientation. However, it is a
straightforward matter to assign an orientation to a torus knot Kq,r;
just assign an orientation to each segment ai, the orientation should
flow from Ai to B
i
. We shall denote this oriented torus knot K(q, r).
In addition, if we now reverse the orientation of each ai, then we shall
denote the subsequent oriented torus knot K( -q, -r).
Example 7.1.1. K(3,2) and K(-3, -2), Figure 7.1.6, are equivalent
to the versatile right-hand trefoil.
K(3,2)
Figure 7.1.6
K(-3,-2)
Exercise 7.1.3. Show that K(q, r), q, r > 0, has a regular diagram
of the form shown in Figure 7.1.7. For example, Figure 5.3.6(a) is a
regular diagram for K(3,4).
Figure 7.1.7
Exercise 7.1.4. Draw the oriented regular diagrams, of the form in
Figure 7.1.7, for the torus knot K(q, r) in the following three cases:
(i) q > 0, r < 0; (ii) q < 0, r > 0; (iii) q < 0, r < O.
137 Torus Knots
We know that from a torus knot Kq,T we can obtain either the
(oriented) torus knot K(q, r) or K( -q, -r). However, it is also pos-
sible to obtain from the torus link Kq,r an oriented torus link that is
neither of these. For example, suppose q = 4 and r = 2, and ao has
an upward orientation, while assigning to at a downward orientation,
Figure 7.1.8(a).

<Xt:
a
o
:

"--1'
*" .At.. ,
Ao
(a) (b)
Figure 7.1.8
(c)
Then, by identifying the ends as described above, we shall obtain
an oriented torus link, Figure 7.1.8(b) and (c). However, this torus link
is neither K(4,2) nor K( -4, -2). Further, there do not exist a q and
r such that it will become an oriented torus link of the type K(q, r),
cf. Exercise 7.3.3. Therefore to avoid such confusion occurring, we shall
consider, from this point onwards, only oriented torus knots or links
K(q, r), with the orientation assigned as described above.
Exercise 7.1.5. Show that the knot in Figure 6.1.1(a) is a torus knot
and determine its type, Le., find q and r.
Exercise 7.1.6. Show that each component of a torus link K(q, r) is
a torus knot, and determine its type.
2 The classification of torus knots (I)
In this and the following sections, to avoid convoluted notation,
we shall mostly consider torus knots rather than torus links. Most of
the results, however, can be shown to also hold for torus links, see
Exercise 7.4.1.
If we draw the diagrams that correspond to the torus knots in the
Chapter 7 138
following proposition, then the results given in the proposition come out
almost immediately.
Proposition 7.2.1.
Suppose g.c.d(q, r) =1 and r # O.
(1) If q = 0, 1 or r = 1, then K(q, r) is the trivial knot.
(2) If q, r are integers that are not equal to 0, 1, then
(i) K(-q, r) is the mirror image of K(q, r);
(ii) K(-q, -r) is the torus knot with reverse orientation to
that on K(q, r). Further, K(q, r) K(-q, -r), i.e., K(q, r)
is invertible.
A natural step, from the point of view of topology, is- to "fill" the
inside of the trivial torus, the resulting solid is the trivial solid torus
V. Strictly speaking, the trivial solid torus is obtained by rotating once
round the y-axis the disk D
2
: {(x, y) I x
2
+y2 1}; the boundary of
this disk is the circle m in Figure 7.1.1. Therefore, V is homeomorphic
to D
2
x S1.
Proposition 7.2.2.
Suppose that V is a trivial solid torus in 8
3
. If VO is the set of
all the internal points of V, then S3 - VO is also a trivial solid torus
in S3.
Proof
In the same way as explained in Chapter 1, Section 5, we shall
consider S3 to be constructed from two 3-balls that have been glued
along their respective boundaries, namely, 2-spheres. J
(a) (b)
Figure 7.2.1
139 Torus Knots
We may alter these two 3-balls so that they become two solid cylin-
ders, WI and W2. Let us now glue these cylinders together along their
surfaces. First, we glue the top, base, and side surface of WI to the
respective top, base, and side surface of W
2
, Figure 7.2.1{a).
Next, consider relatively small, in terms of diameter, cylinders E
I
and E
2
from W
l
and W
2
, respectively, Figure 7.2.1{b). These two
cylinders may be thought of as the "fattened" centres of WI and W2. If
we now glue W1 and W2 along their boundaries in the above manner,
then E
I
and E
2
form a trivial solid torus, V, in S3 (= WI UW
2
).
Therefore, to prove Proposition 7.2.2, it is sufficient to show that the
solid obtained by extracting the internal points of this V from S3 is
also a solid torus.
(a)
a

p:
d
II,
d.
(b)
.P'
Figure 7.2.2
As in Figure 7.2.2, in each of WI and W
2
create a rectangle.
These two rectangles form a disk fi in S3 when we glue together the
surfaces of WI and W
2
To be precise, we glue together the edges ab
and a'b', bc and b'c', and cd and c'd', Figure 7.2.2{b). If we now
rotate the disk once round E
I
and E
2
, which have also been glued
together in the prescribed manner, then what we obtain is S3 - vo,
and by construction this is also a (trivial) solid torus.

Contained in the above proof is something that in itself is of impor-

tance. In order to explain it clearly, let us first consider the two special
circles on the trivial torus T that were discussed briefly in the previous
section. As before, suppose that m is the circle from which we obtain
T. By rotating the point P(3,O) on m once around the y-axis, we shall
form another circle l. We say that m is the meridian of T and l is
Chapter 7 140
the longitude of T. (In general, the meridian and longitude are circles
on T that are obtained by performing elementary knot moves on T to
m and 1, respectively.) Moreover, we say that the circle C, obtained
by rotating the centre, (2,0), of the circle m is the centre of T. We may
think of C and l as being "parallel." Since C and m form a link in R
3
,
we may assign an orientation to C and m such that the linking number
lk(m, C) = 1, Figure 7.2.3. In addition, we assign the same orientation
to l as on C. (We shall also say that m and 1 are the meridian and
longitude of the trivial solid torus V.)
Figure 7.2.3
Let us return to the details of our previous proof. The circle m '
surrounding the core of E
1
, Figure 7.2.1(b), is the meridian ofV. How- I
ever, for the trivial solid torus S3 - VO it is the longitude. Similarly, the
meridian of S3- V
O
, namely, apdp'a, Figure 7.2.2(b), is a longitude for '
V. This, in a sense, gives a refinement of the gluing process, the exact
nature of which is the substance of the next proposition.
Proposition 7.2.3.
8
3
can be obtained by identifying (gluing) the s_urfaces (i.e., tori)
of two trivial solid tori V1 and V2 in such a way that the meridian
and longitude of V1 glue to the longitude and meridian, respectively,
of V2.
This proposition can be put to immediate use in proving the fol-
lowing theorem:
Theorem 7.2.4.
K(q, r) K(r, q).
Proof
By Proposition 7.2.3, we may think of S3 as two trivial solid tori
that have been glued together along their surfaces, the (trivial) tori
T
1
and T
2
, respectively. FUrther, the meridian and longitude of Tl
141 Torus Knots
correspond to the longitude and meridian of T
2
Finally, in our as-
sumptions, we may think of K(q, r) as a (q, r)-torus knot that lies on
T1 (cf. Exercise 7.2.1 below). However, if instead we think of this knot
to be on T
2
, then due to the interchange of meridians and longitudes
it is a (r, q)-torus knot. But the knot itself has not been altered in any
way; hence, K(q, r) K(r, q).

Exercise 7.2.1. Show that a torus knot (or link) K(q, r) is equivalent
to a knot (or link) K on a trivial torus T such that K intersects a
meridian, m, of T at exactly r points and a longitude, l, of T at Iql
points.
The above proof is a mathematical sleight of hand, but the above
theorem will become clearer if instead we use the Reidemeister moves
to transform the regular diagram of K(q, r) into the regular diagram of
K(r,q).
Exercise 7.2.2. Show using the Reidemeister moves that the regular
diagrams, with orientation assigned, ofK(5,3) and K(3,5) are equivalent.
Exercise 7.2.3. Find a formula that determines the total linking num-
ber of the torus link K(q, r).
Exercise 7.2.4. Show that the Whitehead link and the Borromean
rings are not torus links.
3 The Seifert matrix of a torus knot
Before we classify the torus knots K(q,r), we would like to look at
their Seifert matrices. Suppose, firstly, that both q and r are positive
integers. The knot in Figure 5.3.6(a), whose Seifert matrix we have
already investigated in detail in Example 5.3.4, is in faCt the torus knot
K(3,4). It is reasonable to infer from this example that the Seifert
matrix of K(q, r) has, the f?llowing form.
Proposition 7.3.1.
Let q and r be positive integers. The Seifert matrix, M, of K(q, r)
is (S-equivalent to) a square matrix of order (r - l)(q - 1), which can
Chapter 7
be divided into the following (r - 1)2 blocks:
142
B
1I
B
21
B
22
B3
2
B
33
M=
0
o
Br - 2,r-2
Br - 1, r-2 Br - 1, r-l
such that
(1) Each block is a square matrix of order (q - 1);
(2) The diagonal blocks B
i
, i (i = 1, 2, ... , r - 1) and the off diag-
onal blocks B
i
+1, i (i = 1, 2, ... , r - 2) are the only non-zero
matrices;
(3) Bi,i (i=1,2, ... ,r-l) isa (q-1)x(q-1) matrixand
-1
1 -1
Bi,i = 1
-1
1 -1
(4) Bi+1,i=-B1,1 (i=1,2, ... ,r-2).
We can use the above matrix to calculate the Alexander polynomial
of K(q, r).
Theorem 7.3.2.
Let be the Alexander polynomial of K(q, r), q, r =F o.
If g.c.d(q, r) = d 1, i.e., K(q, r) is a d-component torus
link, then
)
.i!: d
(q-l)(r-l) d-l (1 - t (1 - t d )
t 2 (t) = (-1)
q,r (1 - t
q
)(l - tr)
Since the calculation of det(M - tM
T
) is a tad cumbersome, we
shall not give a proof here but instead consider an example.
Example 7.3.1. We shall call the torus knot (or link) K(n,2), the
elementary torus knot (or link). Suppose n > 0 and let us calculate

143
Let
(n-l)
t 2 dK(n,2)(t) = det
-l+t
1
-t
-1 +t -t
1
-l+t
1
Torus Knots
=d
n
.
-t
-l+t
A direct calculation shows that d
2
= -1 +t and dg = 1 - t + t
2
.
We shall calculate d
n
+
1
by induction on n. Suppose
In order to determine d
n
+
1
, first add all the rows to the first row,
and then expand it along the first row so that we obtain
d
n
+
1
= td
n
+ (-l)n
= (-1)n-
1
t(1- t + ... + (_1)n-
1
t
n
-
1
) + (-l)n.
= (-l)n(l - t + .. + (-l)nt
n
)
Finally, it is easy to check that for d = gcd(n,2) (= 1 or 2),
d = (_l)d-l (1 - t)(1 - t ~ d .
n (1 - tn)(l - t2)
Exercise 7.3.1. Without using Theorem 7.3.2, calculate the Alexan-
der polynomial of K(3,4) and K(3,3). Confirm your answers by using
Theorem 7.3.2.
Exercise 7.3.2. Compute ~ q r and d_q,r (q,r =1= 0) using the for-
mula given in Theorem 7.3.2 and confirm that dq,r = d_q,ro
Exercise 7.3.3. Show that there do not exist a q and r that will
make the oriented torus link in Figure 7.1.8(c) equivalent to K(q, r).
4 The classification of torus knots (II)
As mentioned in the introduction to this chapter, to classify torus
knots K(q, r) (q, r > 0) we already have the necessary techniques. In
fact, all that is required is the Alexander polynomial.
Chapter 7 144
Theorem 7.4.1.
Suppose that K(q, r) and K(p, s) are two torus knots, and that
q,r,p,s 2. Jrhen
K(q,r) {q,r} = {p,8}.
Proof
It follows from Theorem 7.2.4 that we may assume q rand
p s. Since the proof in the direction = is fairly immediate, we
shall only show the proof in the direction ==}, i.e., q = p and r = s.
We shall only prove the theorem for torus knots, so q > rand
p > s. The torus link case is left 88 an exercise for the reader.
Now, since K(q, r) K(p, s), LlK(q,r)(t) = The maxi-
mum degree of is
1 ( ) (q-l)(r-l) (q-l)(r-1)
qr+ - q+r - 2 = 2 '
while the maximum degree of LlK(p,s)(t), similarly, is
(p - 1)(8 - 1)
2
So, since dK(q,r)(t) = dK(p,s) (t),
(q - 1)(r - 1) = (p - 1)(s - 1).
Therefore, due to Theorem 7.3.2, the following formula holds:
(1 - t)(l - t
qr
) _ (1 - t)(1 - t
PS
)
(1 - tq)(l - t
r
) - (1 - t
p
)(1 - t
s
) .
(7.4.1)
(7.4.2)
If we clear the denominators in (7.4.2) and divide out by (1 - t), '
we obtain that
1 - t
P
- t
S
+t
P
+
s
- t
qr
+t
p
+
qr
+t
s
+
qr
_ t
p
+
s
+
qr
=1 - t
q
- t
r
+t
r
+
q
- t
PS
+ t
q
+
ps
+ t
r
+
ps
_ t
q
+
r
+
ps

(7.4.3)
Let us, in both sides of (7.4.3) compare the minimum degree of the
non-constant negative terms. Since p > s, the minimum degree of the
negative terms on the left-hand side is either s or qr, while on the
145 Torus Knots
right-hand side the minimum degree is either r or pSt So it follows
that there are four possible cases:
(i) s =r, (ii) s =ps, (iii) qr =r, (iv) qr =pSt
However, since p, q > 1, the cases (ii) and (iii) cannot occur. So,
suppose case (iv) holds, then since qr is the minimum degree we must
have qr :::; s < pSt This now contradicts our assumption that qr =pSt
Hence, we are left with only case (i), Le., s = T. Then from (7.4.1) we
also obtain the other required result, namely, q = p.

In order to obtain a complete classification, we need to consider the

cases when either q or T are negative. As a first step towards this, we
shall prove that K(q, r) is not amphicheiral.
Theorem 7.4.2.
The signature of K(q, r) (q, r > 1) is not zero. Therefore, K(q, r)
is not amphicheiral.
N+NT
-N
o
p=[
Proof
To actually determine the signature of K(q, r) is quite a compli-
cated procedure, cf. Theorem 7.5.1. However, to show that it is not
zero is a slightly easier task. In this regard, let us look at the Seifert
matrix M. We shall show that the signature of P = -(M + M
T
) is
positive, i.e., the signature of M+M
T
is a negative integer. The proof
is divided into two cases: r is an even integer, and r is an odd integer.
We shall confine ourselves to showing the theorem is true in the cases
of K(4,4) and K(4,5). The method to generalize and hence prove the
theorem can easily be deduced from these two examples, and so we leave
the generalization as an exercise for the reader.
(i) The case of the torus link K(4,4).
For this case we can divide P = - (M +M
T
) into 9 blocks.
~ ]
N+NT '
where N = [ - ~ _~ ~ .
Each block is a matrix of order 3. For the sake of convenience, these
blocks will be labeled the (i,j)-block ofP [similar to the (i,j)-entry of
Chapter 7 146
a matrix] and denoted by Pi,j. For example, PI2 = -NT and P
2I
=
-No
Next, we move P
2i
+
I
,2j+1 (i,j =0,1) to the top left-hand corner
and P2k,2l (k, l = 1) to the bottom right-hand corner. (This can be
done by applying a AI-operation on P.) The resultant matrix pI is
therefore A
1
-equivalent to P. To be more precise,
where
[
P11 PI3 P12] [A
pI = P3I P33 P
32
= 0
P
21
P
23
P22 -N
-NT]
-N ,
A
A=N+N
T
= [ ~ ~ -[].
Now, let B = NA-1 and C =NTA-1 0 To clarify matters further,
let us define a matrix, Q, of order 9,
[
I 00]
Q= 0 I 0 ,
Bel
where I denotes the third-order identity matrix, and use it to compute
QP/QT,
[
A 0 0 ]
QP'QT = 0 A 0 .
o 0 A-BN
T
-CN
Then (Y(P) = cr(QP/QT) = 2cr(A) + u(A - BNT - eN), where
A= A - BN
T
- CN is a matrix of order 3. Hence -3 ~ u(A.) ~ 3, and
since we may diagonalize A to
[
2 0 0]
o ~ 0 ,
o 0 i
3
it also follows that u(A) =3. Therefore, u(P) ~ 6 - 3 = 3 > 00
(ii) The case of the torus knot K(4,5).
If we apply a similar method to (i), we can only show that u(P) ~ 0,
so we need to slightly modify the proof of the previous case.
147 Torus Knots
As in case (i), we may divide P into blocks, in this case into 16
blocks Pi,j (i,j = 1,2,3,4). Following along exactly the same lines as
in case (i), we can move P2i+l,2j+l (i,j = 0,1) to the top left-hand
corner and P2k,2l (k, l = 1,2) to the bottom right-hand corner. For
the same reasons as before, the resultant matrix P' is AI-equivalent
to P:
P
II
P
I3 PI2
P
I4
A 0
-NT
0
P
31 P33
P
32
P
34
0 A -N
-NT
p'==
P21 P23
P
22 P24 -N
-NT
A 0
P
41 P43
P
42 P44
0 -N 0 A
where A and N are as in (i).
As in (i), we may define a matrix Q using the same matrices Band
C, but in this case the matrix is of order 12,
Q=
I 0
o I
B C
o B
o
I 0
o I
Also as in (i) we need to calculate QP'QT,
QP'QT ==
A 0
o A
o
o
A-BNT -ON
-BN
Let us denote by R the right lower matrix (of order 6) of QP'QT.
Then u(P) = u(QP/QT) == 2u(A) + u(R).
Now, since
[
321]
A-I =! 2 4 2 I
4 1 2 3
a simple calculation shows that BN
T
= NA-IN
T
== NTA-IN = eN ==
! [ ~ -; = ~ ]
4 -1 -1 3
~
I,
'.1:
":
",I
: ~
:1
Chapter 7 148 ; ~
: ~
Since all the diagonal entries of R are positive, it follows from Ex- f
ercise 6.4.1(a) that -5 ~ u(R) ~ 6. Further, since u(A) =3, we have
that u(P) = 6 + u(R) ~ 6 - 5 = 1 > o.

If we gather the above results together, then this leads to a classi-

fication of the torus knots.
Theorem 7.4.3 (Classification of torus knots).
(1) If q or r is 0 or 1 then K(q, r) is the trivial knot.
(2) Suppose that q, r, p, s are not 0 or 1. Then
K(q, r) ~ K(p, s) => {q, r} = {p, s} or {q, r} = {-p, -s}.
. ~
Exercise 7.4.1. Generalize the above classification to include torus : : ~
links K(q, r). (Hint: Prove Proposition 7.2.1, Theorem 7.2.4, and The- : ~
orem 7.4.1 for torus links.)
5 Invariants of torus knots
The proof of Theorem 7.4.2 seems to affirm that, in comparison
with the Alexander polynomial, to actually calculate the signature of
a torus knot (or link) is quite a complicated process. However, the
following recurrence formula allows to calculate the signature in a more
accessible way.
Theorem 7.5.1 [GLM]. ,;,
Suppose that q, r > 0 and denote -u(K(q, r by u(q, r), then
the following recurrence formula holds:
(1) Let us assume 2r < q, then
(i) if r is an odd integer, u(q, r) =u(q - 2r, r) +r
2
- 1;
(ii) if r is an even integer, O'(q, r) = u(q - 2r, r) +r
2

(II) u(2r,r) = r
2
-1.
(III) Let us assume r ~ q < 2r, then
(i) if r is an odd integer, then a(q,r) +u(2r - q,r) =r
2
-1;
(ii) if r is an even integer, then u(q, r) +a(2r - q, r) = r
2
- 2;
(IV) u(q,r) = u(r,q), a(q, 1) =0, u(q,2) = q -1.
149 Torus Knots
Example 7.5.1. Let us calculate 0"(14,5). First from (I)
u(14, 5) =0"(4,5) + 24 =0"(5, 4) + 24.
Next, by (III)
u(5, 4) +0"(3,4) = 14.
However,
0"(4,3) +0"(2, 3) =8 and u(4,3) = 0"(3,4) =6.
Substituting these into the original equation gives u(14,5) =32.
Exercise 7.5.1. Calculate 0"(16,5) and 0"(16,6).
Exercise 7.5.2. Determine the general form for u(q, 3). (Hint: Clas-
sify q into 6 residue classes.)
Exercise 7.5.3. Determine the signature of the oriented torus link
of type (8,6) and compute e(L) [for a definition of e(L), see Exer-
cise 6.4.7]. Using e(L), determine the signature of the torus link L'
obtained from L by reversing the orientation of only one component
ofL.
Exercise 7.5.4. Show that if q and r are both odd and gcd(q, r) =
1, then u(q, r) == 0 (mod 8).
The signature of the torus knot of the type in Exercise 7.5.4 has
been determined directly without recourse to the above recurrence for-
mula in [HZ*]. In general, it is known that if ( -1) = 1, then
u(K) == 0 (mod 8).
We shall now look at several classical invariants for torus knots.
Theorem 7.5.2.
Let gcd(q,r) = 1. The genus, g(K), of the torus knot K(q,r)
( >0)
(q-l)(r-l)
q, r 18 2
Proof
Suppose M is the Seifert matrix of K(q, r) obtained from the Seifert
surface constructed from the regular diagram in Figure 7.1.7. By Propo-
sition 7.3.1,
detM = (detBll)(detB22) ... (detBr-1,r-l)
= [(_l)q-l]r-l : o.
Chapter 7 150
We know from Proposition 6.3.8 that g(K) is the maximum degree
of the Alexander polynomial, which is equal to

As an aside, the Seifert surface constructed from the standard reg-

ular diagram, see Figure 7.1.7, is, in fact, a spanning surface of K with
this (minimal) genus.
It follows easily from the regular diagram of K(q, r) that K(q, r)
has a Iql-bridge presentation. However, since K(q,r) K(r,q) (cf.
Theorem 7.2.4), K(q, r) has, at worst, a min {Iql, Irll-bridge presenta-
tion. In fact, this is its bridge number.
Theorem 7.5.3 [Sc2].
If q or r # 0, then
br(K(q, r)) = min {Iql, Irl}
The regular diagram of K(q, r) has exactly Iql(lrl - 1) crossing
points. On the other hand, the regular diagram of K(r, q) has exactly
Irl(lql-l) crossing points. In fact, it has recently been proven that the
minimum number of crossing points of K(q, r) depends on these two
numbers.
Theorem 7.5.4 [Mus6].
The minimum number of crossing points of the torus link K(q, r)
is
c(K(q, r)) = min {lql(lrl- 1), Irl(lql- 1)}.
Therefore, if Iql Irl the regular diagram of K(q,r) in Figure 7.1.7
is the minimum regular diagram. To determine, on the other hand, the
unknotting number of a torus knot is an extremely difficult 'problem
Exercise 7.5.5. Show that u(K(q,2)) = Iq'2-
1
(Hint: Use Theo-
rem 6.4.8.)
Exercise 7.5.6. For the torus knot K(q, r) (q, r > 0), show that
(K( ))
(q - l)(r - 1)
u q, r 2
(7.5.1)
For a long time, it was believed that (7.5.1) could be improved if
the stronger supposition is replaced by equality. Quite recently,
this supposition has been shown to be true.
151
Theorem 7.5.5 [KM].
For q,r > 0 and gcd(q,r) =1,
(K( ))
_ (q - 1)(1- - 1)
u q,r - 2
Torus Knots
Exercise 7.5.7. Show that K(q,r) has period 'ql and Irl. (For a
definition of period, refer to Chapter 3, Section 2.)
So far in this book we have concerned ourselves with the problem
of classifying knots (and, of course, links). Intrinsically, this is a knot
theoretical problem. This book, however, is twofold in nature and we
wish to balance the purely theoretical with some practical applications
of knot theory. The various applications of knot theory are discussed
in detail in the latter chapters of this book; we would, however, in this
chapter like to consider what might be called the classic application of
knot theory. One of the most important, even fundamental, problems
in algebraic topology is the general classification of manifolds (see Def-
153 Creating Manifolds from Knots
inition 8.0.1 below). In this chapter we will show that it is possible to
create from an arbitrary knot (or link) a 3-dimensional manifold (usu-
ally shortened to 3-manifold). Hence by studying the properties of knots
we can gain insight into the properties of 3-manifolds.
A 3-manifold may be thought of as having one more dimension
than a 2-manifold, which are more commonly known as surfaces. Sur-
faces, in general, are quite easy to visualize; however, once we add an
extra dimension the actual shape can become ambiguous, and so vi-
sualizing a 3-manifold is not really possible, especially in the printed
form. Hence, we shall in this chapter need to rely quite heavily on our
imagination/geometrical awareness.
Definition 8.0.1. A 3-manifold is a topological space in which every
point has a neighbourhood that is homeomorphic to the 3-ball.
For example, the 3-sphere 8
3
and the ball B3 are typical 3-
manifolds that have already been encountered. To be precise, 8
3
is a
3-manifold without boundary, while B3 has a boundary that is homeo-
morphic to 8
2
Also, the idea of a torus can be extended (exactly how
we shall explain below) to a 3-dimensional torus - a 3-manifold without
boundary.
One method to actually create a 3-manifold is to consider a solid
polyhedron, Le., like a ball its inside has been filled. The 3-manifold
is made by suitably gluing to one another the polygons that make up
the boundary of this polyhedron. This is essentially the analogue of the
gluing process in 2 dimensions. The standard example in 2 dimensions
is the torus, which is usually made by gluing together the two opposite
sides of a square, see Figure 8.0.1.
aDd ----+
b c
----+
Figure 8.0.1
Carrying through this example into 3 d,imensions, if we take a cube
and in the "natural" way glue together the top and bottom faces, the
front and back faces, and the left and right faces, then the resultant
object is a 3-manifold called a 3-dimensional torus, see Figure 8.0.2.
Chapter 8
--+ --+
154
--+ 3-dimeDSioDal
torus
Figure 8.0.2
In this chapter, we shall explain two fundamental ways, both of
which are closely related to knots, of constructing 3-manifolds. One ap-
proach is via Dehn surgery, while the other is via covering spaces. The
fact that is of overwhelming importance concerning these two methods
is that every closed connected orientable 3-manifold can be constructed
by either of these methods. Therefore, in theory, we should be able to
gain insight into the classification of 3-manifolds by using knot theory.
However, at present the relationship is weighted in the opposite direc-
tion, more has been gleaned about knots by understanding their related
3-manifolds.
Throughout this chapter all the knots and links shall be considered
to lie in S3.
1 Dehn surgery
Suppose that K is an (oriented) knot in 8
3
. Now, "slightly" thicken
K, so that we form a 3-dimensional manifold called the tubular neigh-
bourhood of K, and denoted by V(K), Figure 8.1.1. '
(a) (b)
Figure 8.1.1
(c)
I
155 Creating Manifolds from Knots
Since K is a knot, V(K) is a solid torus, and its boundary T
[= 8V(K)] is a torus. Therefore, we may define two characteristic
simple closed curves, a meridian and longitude of T, as we discussed
in Chapter 7, Section 2. However, due to the fact that we do not re-
strict ourselves solely to the trivial knot,. T is not necessarily the trivial
torus. Therefore, we have to slightly amend the definitions of our pre-
vious chapter. The definition of the meridian, m, can be naturally
extended to the present case. The longitude, l, however, we shall fix as
the intersection between the Seifert surface of K and T. [If the tubular
neighbourhood V(K) is made sufficiently thin, then this intersection is a
simple closed curve on T. Further, l is independent of the choice of the
Seifert surface.] We may view l as being "parallel" to the knot K, which
in this case is represented by the centre of V(K), Figure 8.1.1(c). We
shall assign orientations to m and K in such a way that lk(m, K) = 1.
The orientation of l is the same as the orientation of K, and, as noted
above, we have assumed they are parallel.
Exercise 8.1.1. Show that the definition of the meridian m and the
longitude l, given above, agree with our previous definition when T is
the trivial torus. Further, show that lk(l, K) = O.
By removing the interior points of V(K) from 8
3
, another 3-
manifold Y is formed. Y and V(K) have a common boundary in
T. For convenience, we shall denote the boundary of Y by T. The
meridian and longitude that were defined on T may also be thought to
lie on T (and shall also be called a meridian and longitude of T).
K(1,2)
"
K
Figure 8.1.2
Now, let us take a trivial torus To and let us place on To a
K(q,r)-torus knot. Let h: To ---+ T(= BY) beahomeomorphism that
sends the meridian and longitude of To to the meridian and longitude of
T. This homeomorphism also sends the torus knot on To onto T. We
Chapter 8 156
-.. -..
may consider it to be a {q,r)-"torus" knot, K on T. (In Figure 8.1.2,
the case q = 1, r =2 is shown.)
Next, let us consider V(K) and V. Let h: oV(K) =T ----t BV = if
be the auto-homeomorphism that sends the meridian, m, of aV(K) to
a torus knot, K on av. We now glue together, using this h, the 4
solid torus V(K) to V along their boundaries T and T. The resulting
object is a (closed, orientable) 3-manifold M, which is said to have been
constructed from S3 by means of Dehn surgery along K, and s =
is said to be its surgery coefficient. If instead of a knot we use a link
L = {K
1
,K
2
, ,Kp,}, then again we can construct a 3-manifold, M.
However, in this case M is obtained by performing Dehn surgery along
each component. There exists a graphical notation for these 3-manifolds "
in which the surgery coefficient is placed next to the knot, Figure 8.1.3. .

(a) (b) (c)

(d)
Figure 8.1.3
(e)
Figure 8.1.3(a) denotes the 3-manifold obtained by performing
Dehn surgery on Figure 8.1.2, while Figure 8.1.3{b) denotes the 3-
manifold that can be obtained by a Dehn surgery with coefficient 4
3
along the trivial knot. [To be a bit more precise, in this case m is sent .
to the torus knot K(-3,4).] We shall now explain the Dehn surgeries
denoted in Figure 8.1.3(d) and (e).
Example 8.1.1. The Dehn surgery we perform has coefficient 00.
Since gcd(q, r) = 1, if r = 0 the only possibility for q is 1. Then,
we may fix s = = 00. Hence, in this case, the auto-homeomorphism
h sends the meridian m of aV(K) to the meridian m or -m of av.
So this Dehn surgery is nothing but the operation that puts back the '
157 Creating Manifolds from Knots
solid torus that we had previously removed. Therefore, the operation
does not alter S3, i.e., what is obtained after we have performed this
Dehn surgery is the original S3. Since by performing a Dehn surgery
with coefficient 00 along a knot (irrespective of the knot considered), we
will always obtain S3, in essence a trivial surgery, we shall not include
such surgeries in our further discussions.
Example 8.1.2. Let us consider the 3-manifold M
o
denoted by Fig-
ure 8.1.3(e). Since the coefficient is zero, by definition we must have
r = 1, q = O. Therefore, in this case the meridian m on BV(K) is
sent to the longitude l of BV. However, since K is the trivial knot,
V(K) is the trivial solid torus. By Proposition 7.2.2, Vis also a trivial
solid torus. Furthermore, since the meridian and longitude of the solid
torus Vbecome the longitude and meridian, respectively, of BV, we
may think of M
o
as t ~ 3-manifold obtained by l u i n ~ together, via the
auto-homeomorphism h, the two trivial solid tori V and V(K) such
that the meridian of V(K) is sent to the meridian of V, Figure 8.1.4.
V(K)
Figure 8.1.4
These meridians are each the boundaries of disks D
1
and D
2
in
V and V(K), respectively. If we glue these boundaries together, then
the result is a sphere. Since this sphere lies along the longitude, M
o
is homeomorphic to S2 x S1. Succinctly, Figure 8.1.3(e) denotes the
3-manifold S2 x S1.
Figure 8.1.5
Chapter 8 158
Example 8.1.3. The 3-manifold denoted by Figure 8.1.5 is called a
(q, r)-Lens space and is denoted by L(q, r). It is known that for any
integers p and q, L(p, q) and L(-p, -q) are homeomorphic (with
orientation preserved), and for any integer n, L(p, q) and L(p, q +np)
are also homeomorphic (with orientation preserved). Therefore, we may
assume that both p and q are non-negative integers. The study of Lens
spaces has a long history, included in which is the complete classification .
of these spaces.
Theorem 8.1.1.
The necessary and sufficient conditions for two Lens spaces L(q, r) .
(q, r > 0) and L(p, s) (p, s > 0), to be homeomorphic (with orientation
preserved) are
(1) q =p
and
(2) r == s (mod q) or rs == 1 (mod q).
For an interesting proof of this classical theorem using knot theory, '\)
we refer the reader to Brody [Br].
Example 8.1.4. The Lens spaces L(3,4) and L(3,1) are homeomor-
phic, but L(7,3) and L(7,4) are not homeomorphic (with orientation
preserved). Moreover, if, in particular, k is an arbitrary integer, then
L(l, k) is homeomorphic to S3, while L(O, 1) is homeomorphic to
S2 x SI. (Note: k may also take the value zero.)
Finally in this section, let us state the fundamental theorem on
Dehn surgery.
Theorem 8.1.2.
Any closed orientable connected 3-manifold can be obtained by
performing a Dehn surgery to some knot (or link) in 8
3
. Further, we
can take the surgery coefficient to be an integer.
Exercise 8.1.2. Show that Figure 8.1.3{c) denotes S2 x 8
1
.
From the above theorem it follows that every closed orientable con-
nected 3-manifold corresponds to some link with an (integer) coefficient I
assigned to it. Since it is possible for two inequivalent links, with co-
efficients assigned, to represent the same 3-manifold [see, for example,
Figures 8.1.3(c) and (e)], a classification of links will not induce a sub-
sequent classification of 3-manifolds. However, in 1976 R. Kirby [Ki]
introduced a concept of equivalence links with (Dehn surgery) integer
159 Creating Manifolds from Knots
coefficients. In that paper it was shown that there is a 1-1 correspon-
dence between the equivalence type of links with integer coefficients
and closed orientable connected 3-manifolds. Therefore, an invariant
for links with coefficients assigned will pass through to become an in-
variant of 3-manifolds. In fact, there are numerous 3-manifold invariants
that have been defined in this way.
2 Covering spaces
In topology one of the most important concepts is a (topological)
space called the covering space. Intuitively speaking, the covering space
of a space M has locally the same structure as M. Globally, however,
M can be said to be a "large" space that covers M equally several times.
Example 8.2.1. Let us suppose M is the graph of the figure 8, Fig-
ure 8.2.1(a). Locally, this graph at its only vertex can be divided into
four segments. The question now is, Which "large" graph possesses the
very same characteristic? For example, Figure 8.2.1(b), (c), and (d)
all have this characteristic but Figure 8.2.1(e) does not. Therefore (e)
cannot be a covering space of (a). In fact, (b) rv (d) are all covering
spaces of (a).
(a) (b) (c)
(d) (e)
Figure 8.2.1
Let us now formally define a covering space.
Definition 8.2.1. Suppose X is an arcwise-connected topological
space, Le., any two points in X are joined by a simple arc in X. The
Chapter 8 160
arcwise-connected topological space X is a covering space of X if
(1) there exists a continuous map p: X~ X;
(2) for every point x of X there exists a (open) neighbourhood U
x
of
x such that
(i) E-
1
(U
x
) is the union of a family F = {W
a
} of open sets of
X that mutually do not intersect, i.e.,
(8.2.1)
p-l(U
x
) = U Wae
F3W a:
(ii) the restriction p IW
a
for each W
a
is a homeomorphism.
If the above conditions hold, then p is said to be a covering map.
From (8.2.1) part (2)(i) the pre-image p-l(x) of a point x of X is a
collection of isolated points; the number of points may be infinite. Also,
the collection p-l(x) never includes segments. The order of this set of
points is calle2 the degree of the covering. Therefore, if p-l(x) has n
points, then X is said to be the n-fold covering space of X.
Exercise 8.2.1. Confirm that Figures 8.2.1(b) r.J (d) are covering
spaces of Figure 8.2.1{a). Also, determine their degrees.
Exercise 8.2.2. Show that a torus T is not a covering space of 8
2
,
and conversely show that S2 is not a covering space of T.
Exercise 8.2.3. Show that a covering space of finite degree for a torus
is a torus.
Exercise 8.2.4. Show that the only covering space for 8
2
is of degree
1, Le., itself. (Can the same be said for S3?)
An important role in knot theory is played by a space called a
branched covering space, which is a generalization of the concept of a
covering space. We shall explain the generalization by means of an
example.
Example 8.2.2. Suppose F is the sphere S2, but with two small
holes on the surface. Our intention is to make a 2-fold covering space
for F. F is homeomorphic to a cylinder without a top and base (see
Figure 8.2.2{a); the top and ~ correspond to the two holes a and (3).
We can make a square F by cutting open F along a segment I
that connects two points A and B on F, where A and B are points on
161 Creating Manifolds from Knots
a and (3, respectively. Suppose F' is a copy_of F, Figure 8.2.2(b).
Now, glue together the right-hand edge, of F to the left-hand edge
" of F', and the right-hand edge " of F' to the left-hand edge ,
of F. The result of this process is a cylinder M = FU F', which also
has no top and base, Figure 8.2.2{c). This is the 2-fold covering space
(covering surface) of F. To show this is in fact the case, we must find a
continuous map p : M ~ F that satisfies (8.2.1). Firstly, from a point
P of F we can take the "same P2int" P of F and pI of F', which by
construction is just a copy of F. Therefore, we can take p to be the
map that sends the two points P and P' on M to the point P on F.
F'
tt1
u
B P B
A=.K.
.
:1
~ :...1!l=B'
,.r.. ~ ~
B=B'
(c)
j ~ A.
p
l' l'
B' p' B'
(b)
Figure 8.2.2
Exercise 8.2.5. Show that the above map p : M ~ F satis-
fies (8.2.1).
If we close F by attaching two disks D
1
and D
2
to the top and
base, then we create a sphere F, Figure 8.2.3{a).
(a)
Figure 8.2.3
Similarly, we may create a sphere 1M by attaching two disks D
1
Chapter 8 162
and 52 to the top and base of M, Figure 8.2.3(b). We may, as shown be-
low, extend the previous continuous map p to a map p from this sphere
M to the sphere F. For the sake of clarity, suppose that D
l
, D
2
, 51,
and 52 are all disks of the same size. Since P and P' are symmetric
with regard to the axis of the cylinder, using polar co-ordinates we may
define a map PI : 51 ----+ D
1
, given by
PI (r, 8) = (r,28),
see also Figure 8.2.4.
p'
p
Q(r,9)
Figure 8.2.4
Q(r,29)
Similarly, we can define a map P2 between the two base disks D
2
and D
2
These two maps, PI and P2, enable us to extend p to the
continuous map p : if F. This map p, excluding the centres of J
the disks :0
1
and 52, sends two points of Mto one point in F. The
centres, 0
1
and O
2
, of D
l
and D
2
are in 1-1 correspondence with the
centres, 0
1
and O
2
, of 51 and :0
2
Therefore, M to be exact is not
the 2-fold covering space of F, but M- {0
1
, 0
2
} is the 2-fold covering
space of F- {0
1
, 0
2
}. So we say that Mis the 2-fold covering space of
F branched along {0
1
' 02}, or simply the 2-fold branch covering space
of F.
Roughly speaking, a topological space M is said to be a covering
space of a topological space N branched along a subspace B if there exists
a subspace A of M such that M - A is a covering space for N - B, and
A is a covering space for B. Nota bene, the respective degrees of the two
covering spaces do not necessarily need to agree. (We should emphasize
that this is only a rough interpretation, because if A is not connected
then A may not be a covering space, an example of such a case is given in
Exercise 8.4.1. For a precise definition we refer the reader to Fox [Fl].)
163 Creating Manifolds from Knots
3 The cyclic covering space of a knot
The cyclic covering space, Le., one of the covering spaces of S3
branched along a knot, has been extensively studied over quite a number
of years. Therefore, it has played a very important role in what can be
said to be the first phase, i.e., the topological phase, of research into
knot theory.
Suppose F is the Seifert surface of a knot K. Then K is the boundary
of F. Let us now open up S3 by cutting it open along F. The result is a
3-manifold, N, obtained from S3, with a boundary. The boundary of N
consists of two copies, F' and F" of F, which have been glued together
along K, Figure 8.3.1(a) and (b).
p.
(a)
--+

p
(b) (c)
Figure 8.3.1
The correspondence between S3 and N is as follows:
Let us suppose that P is a point of S3.
__ (1) If P does not lie on F then P corresponds to the same point,
P,onN;
(2) If P is a point of K, then it corresponds to the same point
po on the knot K on N;
(3) If P lies on F and is not a point of K, then it corresponds
to exactly two points pi and P" on the boundary of N. For the sake
of clarity, let us denote the boundary of N with pi by F
'
, and the
boundary of N with pI! by F", Figure 8.3.1(b). Now make n copies
of N and denote them by N1, N2, ... , Nn. Let us suppose that P ~ and
P ~ are copies of pI and P", which lie on F ~ and F ~ respectively. F ~
and F ~ are copies on N
i
of F' and F". We now glue together F1 (the
boundary of N
l
) and F ~ (the boundary of N2 ), so that the points P1
and P ~ are in agreement. In a similar way, we glue together F ~ and
Chapter 8 164
and F4' ... , and and finally and Fi in such
a way that the points and (i = 2, ... ,n) are identified.
By carrying out this gluing and identifying process, what we should
have constructed, finally, is a closed, connected, orientable 3-manifold,
Figure 8.3.1(c). This 3-manifold is called the n-fold cyclic covering of
S3 branched along K and is denoted by Mn{K). If K is the trivial knot,
since F is a disk, N becomes the 3-ba11. Mn(K) is constructed by gluing
the surfaces of these balls, i.e., it is homeomorphic to the 3-sphere S3.
The reason why we say it is "cyclic" is because the points
...
form a cycle. Nota bene, Mn(K) depends only on the knot K and is
independent of the Seifert surface of K that is used to construct it.
Example 8.3.1. Consider the graph constructed by placing two semi-
circles a and (3 at four points of K, where K is the trivial knot in S3,
Figure 8.3.2(a).

d K
(a)
(c)
(b)
(d)
Figure 8.3.2
Suppose M
2
(K) is the 2-fold cyclic covering space of S3 branched
along K. It follows from our discussions above that M
2
{K) is also S3,
165 Creating Manifolds from Knots
so let us see what happens to 0 and (3 in M
2
(K). Suppose the disk F is
the Seifert surface of K. Then in the 3-ball N, obtained by cutting open
S3 along F, a and (3 are as shown in Figure 8.3.2(b). Therefore, the
images 01, /31 and 02, /32 of 0 and /3 formed in the 3-balls N1 and
N
2
, Figure 8.3.2, become, respectively, the closed curves a = 01 U02
and fj = {31 U (32 in M
2
(K) ~ sg), which has been obtained by the
above method of gluing FI" to F
2
', and F
2
" to F
1
/
So in N
2
they
form a link, Figure 8.3.2(d).
Exercise 8.3.1. Using a similar method to Example 8.3.1, investigate
what happens to the three arcs 0, (3, 1, shown in Figure 8.3.3, in the
2-fold cyclic covering of 8
3
branched along K.
Figure 8.3.3
Exercise 8.3.2. (i) Show that the knot a in Figure 8.3.4 becomes
the 3-component link a, Figure 8.3.4(b), in M
3
(K) ~ S 3 , the 3-fold
cyclic covering space of S3 branched along K.
(a) (b)
Figure 8.3.4
(ii) Similarly, show that the knot 0 in Fig-
ure 8.3.5(a) becomes the knot a, Figure 8.3.5(b), in Mg(K).
It is possible to deduce from Exercise 8.3.2 that the knot (or link)
a created from a knot that does not intersect the trivial knot K in
the 3-fold cyclic covering space of S3 branched along K has period 3. In
general, it is possible to show that if K is a knot (or link) with period
n (2: 3), then K is equivalent to a knot (or link) in the n-fold cyclic
covering space of S3 branched along the trivial knot K
o
that has been
Chapter 8 166
transformed, by means of the above process from some other knot K1
that does not intersect K
o
. It is often the case that due to the above,
the study of the period of a knot involves studying the links {K
o
, K
1
}
and the associated branched cyclic covering space.
(a) (b)
Figure 8.3.5
Exercise 8.3.3. Show that the 2-fold cyclic covering space of S3
branched along the 2-component trivial link is S2 x 8
1
[In general, it
is known that the n-fold (n 2) cyclic covering space of S3 branched
along a I--component (1- 2: 2) link is never homeomorphic to S3.]
4 A theorem of Alexander
By using the cyclic covering space method, we can from a sin-
gle knot (or link) construct countless closed orientable connected 3-
manifolds. Sadly, however, it is not possible to construct every 3-
manifold by this method; an example is the 3-dimensional torus. There-
fore, in order to construct an arbitrary 3-manifold, it would seem that
we will need to consider even more complicated covering spaces than the
cyclic covering space. However, before we spiral into a shroud of doom,
actually we will find that the required covering space is surprisingly
(relatively speaking) more simple than might be expected.
Theorem 8.4.1 (Alexander's theorem).
An arbitrary closed orientable connected 3-manifold can be con-
structed by means of a 3-fold (in general non-cyclic) covering space of
S3 branched along some knot.
The proof of this theorem is given in Burde and Zieschang [BZ*],
so here we shall explain how to construct such manifolds.
In the way described at the beginning of this chapter, we can con-
167 Creating Manifolds from Knots
struct a 3-manifold by gluing together the faces of a polyhedron. A
covering space (of finite degree) can be constructed by generalizing the
cyclic covering space method, i.e., we first make a (finite) number of
copies of the solid polyhedron and then glue together, in a suitable
manner, their faces. A requirement on the knot, used in Theorem 8.4.1
to construct a 3-fold branched covering space, is that its determinant is
divisible by 3. In fact, we have already encountered such knots, namely,
the knots that can be 3-coloured, cf. Chapter 4, Section 6. A typical
example of such a knot is the trefoil knot; the figure 8 knot, however,
is not suitable. Therefore, using the trefoil knot let us explain how to
construct a covering space of the type in Theorem 8.4.1.
Firstly, let us take a point P that does not lie on K. We can now,
in the obvious fashion by connecting (in 8
3
) P to each point of K, form
a cone C, Figure 8.4.1{a).
(a) (b)
Figure 8.4.1
(c)
Cwill intersect itself at the crossing points of the regular diagram
of K. The intersection takes the form of a straight line, Figure 8.4.1{b).
Next, open S3 by cutting along C. Since aintersects itself, at these
points of intersection (straight lines) the subsequent shape is quite com-
plicated; Figure 8.4.1{c) shows a section that has been opened in this
way. As the result of cutting along the surface 00, the point C has now
been divided into two points, C' and C". Since the point F lies on the
surfaces 00 and (2), when we cut along the surfaces 00 and (2) it be-
comes divided into four points, F(l), F(2), F(3), and F(4). The points
A, B, D, E, on the other hand, since they lie on the knot, will remain as
single points after a has been cut open. (Jj and 00" are the result of
cutting along 00, i.e., the cutting process opens up 00 into two parts.
Chapter 8 168
Similarly, we obtain sections (2f, (j)H , (3f , and (3)". Further, the seg-
ment that connects the vertex P of the cone to the point C divides into
fOUf parts. Suppose these four segments are a, a', a", and a
'
'', see Fig-
ure 8.4.1(c). Moreover, we may assign a front and back to the surfaces
that have been cut open. (There is no specific rule to designate which
is which. It is perfectly feasible to call the surface (Jj the front and
the surface (J)H the back.) So we obtain from 8
3
a ball that has on its
surface a specific design. In Figure 8.4.2 we have shown the surface of
the ball constructed for the case of the trefoil knot.
(a) (b) ~
Figure 8.4.2 ~
Now, make three copies of this ball. We may index the three balls
by colouring them, namely a red ball, B
r
; a blue ball, Bh; and a yellow
ball, By. The polygons on each of these surfaces are all pentagons, and
i;:
they are glued together by means of the following rule. As we explained ~ }
in Chapter 4, Section 6, we can colour K using 3 colours. ~
Suppose AB is part of a segment of the knot that is coloured red. 0,.,
Then,
(1) We may glue together, without alteration, the pentagons '
that are on either side of AB if these surfaces are part of the red ball
B
r
, Figure 8.4.3(a) (in other words, we restore the surface to its original
form. If AB is blue (yellow), then do the same thing, but on this occasion
the surfaces are part of the blue (yellow) ball Bb (By).
(2) If the pentagons that are on either side of AB are surfaces
of the blue ball Bb and the yellow ball By, glue the pentagon that is
on the front of Bb to the pentagon that is on the back of By in such a
way that AB agrees [Figure 8.4.3(b)]. [If AB is blue (yellow), perform
169 Creating Manifolds from Knots
a similar process, but now with B
r
and By (B
r
and Bb )].
(a) (b)
Figure 8.4.3
So, if we glue together the pentagons of all the surfaces of the three
balls in the above prescribed manner, we shall obtain from B
r
, Bb' and
By a single 3-manifold. This 3-manifold is the 3-fold covering space of
8
3
branched along K (with regard to the colouring on K). In fact, the
3-manifold constructed by this process from the trefoil knot is S3.
Exercise 8.4.1. Show that the trefoil knot via the above process is
transformed, in 8
3
, into a 2-component link.
Example 8.4.1. The 3-fold branched covering space obtained with
regard to the 3-colouring of the knot in Figure 8.4.4 is a Lens space.
Figure 8.4.4
In Figure 8.4.4 the thick line denotes say the colour red, the wavy line
the colour yellow, and the other line the colour blue.
The above method of construction need not be limited to 3-fold
branched covering spaces, it is possible to generalize it to construct a
covering space of finite degree branched along some knot.
If K and K' are equivalent knots, then M and M', the branched
covering spaces (of the same type) constructed from K and K', respec-
tively, are homeomorphic 3-manifolds. In M and M', the new knots (or
Chapter 8 170
links) K and K' that arise from K and K', respectively, are equivalent.
In other words, there exists an orientation-preserving homeomorphism
from M to M' that sends K to K'. (In the above case of a 3-jold
branched covering space, it is known that if K is a knot, then K is
always a 2-component link, cf. Exercise 8.4.1.) Therefore, an invariant
of K (as a knot or link in M) is also an invariant of the original K
(as a knot in 8
3
) . For example, suppose that M is 8
3
and so K is
a knot or link in 8
3
; as before, it is possible to define the Alexander
polynomial Lli( (t). This Alexander polynomial is usually different from
the Alexander polynomial of the original knot K, LlK(t). In addition,
since Lli{(t) is also an invariant of the original knot K, we can use this
second Alexander polynomial of K to help us classify knots.
In general, it is not easily possible to extend knot invariants in 8
3
;
for example, the Alexander polynomial to knot invariants in an arbi-
trary 3-manifold M. However, for very simple invariants, for example,
the linking number, it is possible to extend them to certain types of
3-manifolds.
13
These simple invariants are, in fact, powerful invariants
of the original knot. These invariants can even, on occasion, distinguish .
two knots that have the same Alexander polynomial [BS].
During the period from the end of the 1960s through to the begin-
ning of the 1970s, Conway pursued the objective of forming a complete
table of knots. As we have seen in our discussions thus far, the knot
invariants that had been discovered up to that point in time were not
sufficient to accomplish this aim. Therefore, Conway pulled another
jewel from his bag of cornucopia and introduced the concept of a tan-
gle. Using this variation on a knot, a new class of knots could be defined:
algebraic knots. By studying this class of knots, various Local problems
were able to be solved, which led to a further jump in the level of un-
Chapter 9 172
derstanding of knot theory. However, since there are knots that are
not algebraic, the complete classification of knots could not be realized.
Nevertheless, the introduction of this new research approach has had
a significant impact on knot theory. In this chapter we shall investi-
gate 2-bridge knots (or links), which are a special kind of algebraic knot
obtained from trivial tangles.
1 Tangles
On the sphere 8
2
- the surface (boundary) of the 3-ball B
3
- place
2n points. A (n, n)-tangle T is formed by attaching, within B
3
, to
these points n curves, none of which should i n t ~ s t each other, Fig-
ure 9.1.1. (Strictly speaking, from the point of view of our original
assumption, the curves should be polygonal.)
(a) (b) (c)
(d) (e)
Figure 9.1.1
To be precise, we should say that a tangle is the set (B3, T). How-
ever, since all our tangles will be within the ball B
3
, we shall abbreviate
the notation for a tangle to simply T. The astute reader may recall our
discussions in Chapter 1, Section 5; the things discussed there should in
keeping with the above definition be called (1, I)-tangles. The case in
which, in addition to our original n curves, there exists in B
3
a closed
curve, Figure 9.I.I(d), shall not be considered in this book.
173 Tangles and 2-Bridge Knots
From the definition above, it follows immediately that Fig-
ure 9.1.1{a) is a (1, I)-tangle; Figure 9.1.1{b) rv (d) are (2,2)-tangles;
and Figure 9.1.1{e) is a {3,3)-tangle. In what follows we will work vir-
tually exclusively with {2,2)-tangles, and so for simplicity we shall refer
to then as just tangles.
Suppose that we fix four points on the sphere 8
2
, namely,
NE, NW, SE, SW (where the abbreviations are the obvious ones, north
east, et cetera.), Figure 9.1.2.
11
Figure 9.1.2
These points can be precisely described in R3 in terms of the
following co-ordinates:
1 1 1 1
NE = (0, J2' J2)' NW = (0,- J2' J2)'
1 1 1 1
SE = (0, J2' - J2)' SW = (0, - J2' - J2).
A cursory glance at these co-ordinates tells us ~ t four points
all lie in the yz-plane. By attaching the end points of)two polygonal
curves in B3 to these four points, we can form a tangle.
So, if we project this tangle onto the yz-plane, as in the case of a
knot, we have what may be called a regular diagram of the tangle, Fig-
ure 9.1.3. (So as not to overcomplicate matters, we shall often consider
a tangle to be simply this regular diagram.)
Figure 9.1.3
Chapter 9 174
The knot (or link) obtained by connecting the points NW and NE,
SW and SE by simple curves outside B3, as is shown in Figure 9.1.4(a),
is called the numerator and is denoted by N(T). Similarly, we may
connect the points NW and SW, NE and SE by simple curves outside
B3, as is shown in Figure 9.1.4(b), and the subsequent knot (or link) is
called the denominator and is denoted by D(T).
N(T) D(T)
(a) (b)
Figure 9.1.4



:1
.


Definition 9.1.1. Suppose T
1
and T
2
are two tangles in B3. If ":-
we can change T1 into T2 by repeatedly performing elementary knot "
moves in B
3
, keeping the four points (NE, NW, SE, SW) fixed, then Tl
and T
2
are said to be equivalent (or equal).
So, the above process allows us in a natural way to construct from
a tangle two different knots (or links). The reader might be slightly con-
fused with our new terms, since the numerator and denominator do not
.,;,1
seem apt for the diagrams they represent. The reason for this somewhati
strange terminology will be revealed slightly later (cf. Exercise 9.3.8). ,1;

Since a tangle is a "part" of a knot, we can extend the various
definitions we have so far encountered to tangles, i.e., equivalence, the
connected sum, et cetera.
Intuitively, if we can continuously move in B3 T
1
to T
2
without
causing any self-intersections of the tangles and keeping the endpoints
fixed, then T
1
may be said to be equivalent to T
2
Since we may
think of knot equivalence in this way, the following definition is just a
extension of the knot case:
Definition 9.1.2. If an orientation-preserving auto-homeomorphism
175 Tangles and 2-Bridge Knots
<p : B3 ---+ B3 satisfies the following conditions, then T
1
and T2 are
said to be equivalent:
(1) cp is an identity map for 8
2
, i.e., the map keeps 8
2
fixed.
(9.1.1)
(2) <p(T
1
) = T
2

It would seem obvious that there should be some way to connect two
endpoints of one tangle to two endpoints of another tangle, in essence
the sum of two tangles. To accomplish this, firstly place T1 in a ball
Bi and T
2
in a ball Secondly, position T
1
and T
2
so that they
become parallel, and form a "large" that contains of and
Then we connect by parallel segments NE and SE of T1 to NW and
8W, respectively, of T2. In this new configuration, the NW and SW of
T1 and NE and SE of T2 become the NE, NW, SE, SW of the new ball
and the "summed" tangle is denoted by T
1
+T2' Figure 9.1.5.
+
=
Figure 9.1.5
We may think of this process in a slightly more precise fashion.
Regard the ball as a globe, then after gluing the east hemisphere of
to the west hemisphere of if we remove the parts of the hemispheres
that have been glued together we shall form a ball. The "knotted" string
in this ball is again the tangle sum.
It might seem at first sight that we may treat the sum of tangles
in a similar way to the sum of knots (or links). However, there are
considerably differences, as the next example and exercise show.
Example 9.1.1. Even if N(T
1
) and N(T2) non-trivial
knots, it is quite possible that N(T
1
+ T
2
) is a triVIal knot.
This always occurs in the case of rational tangles, which we will
define a bit later (see also Exercise 9.3.9).
Chapter 9 176
Exercise 9.1.1. Show that the numerator of the tangles T
1
and
T
2
in Figure 9.1.5 is not a trivial knot or link, but the numerator of
T
1
+T2 is a trivial knot.
When the surface of a 3-ball Bthat lies in B3 intersects the (2,2)-
tangle T in only two points, then (8,8 n T) is a (1, I)-tangle. If this
(1, I)-tangle is always the trivial tangle, then T is said to be locally triv-
ial [for a definition of a trivial (1, I)-tangle, see Chapter 1, Section 5].
The tangles in Figure 9.1.1(b) and (d) are locally trivial tangles, while
the one in (c) is not locally trivial.
2 Trivial tangles (rational tangles)
Let us consider the simple tangles shown in Figure 9.2.1. Start-
ing from the left-hand tangle, we shall call these tangles of (O)-type, "
(0, a)-type, (-I)-type, and (I)-type. The (a, a)-type tangle is some-
times also called the (00)-type tangle. Collectively, we shall call these
tangles the exceptional tangles.
(0) type (0,0) type (-1) type (1) type
Figure 9.2.1
Exercise 9.2.1. Determine the knot (or links) that are the numerators
and denominators of the exceptional tangles.
If we limit ourselves to (1, I)-tangles, it is obvious what should
be the trivial (1, I)-tangle. However, it is not so straightforward to '
actually say without much thought what a trivial (2,2)-tangle is ex-
actly. For example, we may say that the exceptional tangles are all
"trivial" tangles. This, unfortunately, is not the limit of all possible ..
trivial (2,2)-tangles; there are numerous other possibilities. Therefore,
we need to give a formal definition of a trivial tangle.
Definition 9.2.1. Suppose f is a homeomorphism that maps the
ball B3 to itself and maps the set of four points {NW, NE, SW, SE} to
itself, but not necessarily as the identity map (Le., f need not map NW
177 Tangles and 2-Bridge Knots
to NW, et cetera). Then a trivial tangle (or rational tangle) is a tangle
that is the image of the (O,O)-type tangle under this homeomorphism.
(Hence, there are countless examples of such tangles.)
Example 9.2.1. If we rotate R3 about the x-axis through an angle
of then the (O,O)-type tangle, T(O,O), is sent to the (O)-type tangle,
T(O). Therefore, T(O) is a trivial tangle, Figure 9.2.2.
(a)
Figure 9.2.2
(b)
However, such "trivial" ones as the above are not the only triv-
ial tangles. In fact, such homeomorphisms are numerous; as typical
examples consider the following two examples.
Firstly, let us rotate the sphere about the z-axis, but keeping the
northern hemisphere and the south pole fixed. Then the southern hemi-
sphere is given a twist such that SE and SW exchange positions, Fig-
ure 9.2.3.
o
y
Figure 9.2.3
Secondly, let us consider the rotation of the sphere about the y-axis
but on this occasion keeping the western hemisphere and the point
E(O,l,O) on the equator fixed. In this case, the eastern hemisphere is
Chapter 9 178
given a twist and the points NE and SE exchange positions, Figure 9.2.3. '
The former rotation/twist we shall call a vertical twist, and the latter
a horizontal twist. Further, we may assign an orientation to the twists
as described below. In the case of a vertical twist, a positive twist is a
right twist [Figure 9.2.4(a)], while for a horizontal twist a positive twist
is a left twist [Figure 9.2.4(b)]. The respective inverse twists are the
negative twists.
xx
right twist
(a)
left twist
(b)
Figure 9.2.4
Example 9.2.2. If we give T(O,O) a 3-fold (i.e., rotate it three times) \
vertical (hence positive) twist and then in addition apply a (-4)-fold
horizontal twist, then we obtain the tangle in Figure 9.1.3. Therefore, f
this is also a trivial tangle. In particular, T(l) and T (-1) are triv-
ial tangles.
Example 9.2.3. We can obtain the tangle T
1
of Figure 9.1.5 by
first sending T(O,O) to T(O) and then performing 2 horizontal twists, 3
vertical twists, and finally 1 horizontal twist in that order. Therefore,
T1 is also a trivial tangle.
The trivial tangles that we have obtained depend on performing al- I
ternatively vertical and horizontal twists, Le., it is possible to formulate
the following proposition. (We omit the proof since it depends on the
theory of surface mappings.)
Proposition 9.2.1. !.'
A trivial tangle can be obtained by performing a finite
sequence of vertical and horizontal twists to T(O) or T(O,O). '::

It follows from this proposition that the trivial tangles can be com- ,:-
pletely determined by how we perform, alternatively, the several vertical
and horizontal twists. Let us express this sequence by T(al' a2, ... ,an) :.1
and explain in a little more detail how this sequence occurs.
If n is odd, we can obtain the relevant tangle by first performing
at horizontal twists on T(O), then a2 vertical twists, a3 horizontal
twists, and, continuing in this vein, repeating alternatively the twists
until finally we perform a horizontal twist an times; an example is given
179
in Figure 9.2.5{b).
T(2,S,-4,2)
(a)
Tangles and 2-Bridge Knots
T(-S,2,4)
(b)
Figure 9.2.5
If n is even, we can obtain the relevant tangle by first performing
al vertical twists on T(O,O), then a; horizontal twists, and then alter-
nating the twists until finally we perform a horizontal twist an times;
see also Figure 9.2.5{a).
If all the ai are of the same sign, then the regular diagram that is
obtained is an alternating diagram. The regular diagram is no longer
alternating if the signs of some ai and ai+l are different. We also
allow the case of ai equal to zero. However, since if ai is zero we
can "shorten" the tangle, we shall, to avoid unnecessary complications,
assume that ai 1= 0 (i 1= n).
Exercise 9.2.2. Show that T (2, -3,0,2,1, -2) and T (2, -1,1, -2)
are equivalent.
A nice piece of number theory states that a real number may be
expressed as a continued fraction. An example of such a continued
fraction is
3
2+-
2
r
4+ 5 )
From the "fraction" we can calculate the real number It expresses,
starting from the last fraction in the sequence,
3 3 15 59
2 + 4 +a = 2 + 22 = 2 + 22 = 22'
5 5
Hence, the above continued fraction is equal to
Chapter 9 180
However, it will help us in our further discussions if we express the
continued fraction in the following way:
3 2
2+- -
4+5
and call it the continued fraction of ~ ~ So far we have considered only
rational numbers; however, an irrational number may also be expressed
as a continued fraction, but in this case the expression is infinite. We .'
shall in this book restrict ourselves to rational numbers, and hence all
our continued fraction, expressions will be finite. Unfortunately, there
is more than one way of expressing a rational number as a continued
fraction, our above example can be expressed in the following further
two ways.
59 1 1 1 1 1
22 = 2 +1+2+7 =3 + (-3) + (-7)"
In particular, if except the initial integer, the numerator of all the
fractions is 1; then we shall express it as [2,1,2,7] (= [3, -3, -7)). .f
. ~
:.\1
Exercise 9.2.3. Find at least 3 continued fraction expansions for 2
3
5
2
1 ~
and ~ l
The trivial tangle T(al' a2, ... ,an), where al =F 0, corresponds
to the fraction ~ that has a continued fraction expansion given by
This number is called the fraction of the tangle. In particular, the I
fraction of the tangle T(O) is 0, and the fraction of the tangle T(O,O) is
0+ t, which we will denote by 00. Conversely, given a rational number
piq by expressing it in terms of a continued fraction,
we can associate it with the trivial tangle "T(al, a2, ... , an). (Care r
needs to be taken with the order of aI, ... , an.) The next theorem
shows that there exists a very nice correspondence between the rational
numbers (including 00) and the trivial tangles.
181 Tangles and 2-Bridge Knots
Theorem 9.2.2 [C).
There exists a 1-1 correspondence between the set of all ratio-
nal numbers (including (0) and the equivalence classes of the triv-
ial tangles. In other words, if the trivial tangles T(al' a2, ... , an)
and T(b
l
, b
2
, , b
n
) are equivalent, then their respective fractions
that are expressed by the corresponding continued fraction expressions
[an, an-I, ... , a2, al] and [b
m
, b
m
-
l
, ... , b
2
, b
I
] are equal. The con-
verse also holds.
Since an arbitrary rational number (including 00) corresponds to
a trivial tangle, the trivial tangles are sometimes referred to as rational
tangles.
Exercise 9.2.4. Show the validity of Theorem 9.2.2 by showing
T(2,1,1,0) and T( - 2, - 2,1) are equivalent; this should follow since
3/5 has the the continued fraction expansions [0,1,1,2] and [1, - 2, - 2].
If in the process of determining the rational number from a given
continued fraction, we set t = 00, = 0, and k +00 = 00, then this
will obviate any problems that might occur if any of the ai are zero.
Exercise 9.2.5. Show using regular diagrams that T(O, 3, 0) = T(O)
and T(1, -1,2) = T(O, 0).
Exercise 9.2.6. Show that if the trivial tangle T(al' a2, ... , an) has
as its fraction a rational number other than 0 or 00, then we can always
find ai such that the signs of all the ai are the same.
The sum of two rational tangles is not necessarily a rational tangle.
Figure 9.2.6(c) is the sum of two rational tangles but itself is not a
rational tangle.
(a) (b)
Figure 9.2.6
Chapter 9 182
In fact, from Theorem 9.3.1, which we will prove in the next section,
the denominator of a rational tangle is always a prime knot (or link),
but the denominator of T
1
+T
2
of Figure 9.2.6(c) is not prime. We
say a tangle is an algebraic tangle if it can be expressed as the sum of
a finite number of tangles, each of which is a rational tangle and/or
its homeomorphic image (mirror image, rotation, et cetera). Although
we will give a few comments concerning algebraic tangles at the end of
this chapter, we shall not delve too deeply into the concept of algebraic
tangles. A simple example of an algebraic tangle is given in the final
example of this section.
Example 9.2.4. From Figure 9.2.7, since T
1
'
is T
1
rotated by I
the sum of T
1
' and T
2
, Tl
'
+T
2
, is an algebraic tangle.
T'
1
Figure 9.2.7
3 2-bridge knots (rational knots)
2-bridge knots are a family of knots that are very closely related
to rational tangles. These knots (or links} as the name implies, have I
bridge number 2 (cf. Chapter 4, Section 3). We know that the trivial.
knot is a I-bridge knot, so 2-bridge knots may be said to be the next I
most simplest set of knots (and links) to investigate. 2-bridge knots
183 Tangles and 2-Bridge Knots
(or links) are always prime and have at most 2 components (why?).
Further, this set of knots and links has been completely classified, but
the local characteristics of these knots (or links) are still an important
area of research. Since a non-trivial (q, r)-torus knot has bridge number
min {Iq), Irl}, if q or r is not 2, they do not belong to this class of
knots.
Our first task in this section is to show the relation between 2-
bridge knots and rational tangles. To this end we shall prove the next
theorem. Caveat lector, in this section we will only consider knots and
links that are not oriented.
Theorem 9.3.1.
(1) A 2-bridge knot (or link) is the denominator of some rational
tangle.
(2) Conversely, the denominator of a rational tangle is a 2-bridge
knot (or link).
Due to the above theorem, 2-bridge knots are often called ratio-
nal knots.
Proof
Let us consider a regular diagram, D, of a 2-bridge knot (or link).
Since D has only two local maxima, the regular diagram may be drawn
as the reduced regular diagram shown in Figure 9.3.1(a); refer to The-
orem 4.3.3. (In the diagram, the local maxima occur, sideways, at the
left-hand edge.)


(a)
(b)
Figure 9.3.1
Chapter 9
"
184 ;,
,I
Our aim is to move the crossing points on the bottom so that in the :
diagram all the crossing points are in the centre or on top. (Since the "
proof does not rely on the crossing information at the crossing points, ;'
we shall not distinguish between over- and under-crossings. So, let us ::
denote the number of crossing points by ai, b
j
, Ck. We allow the possi-
bility that some of ai, b
j
, Ck are zero.)
Now, we deform the diagram of Figure 9.3.1(a) into the diagram
of Figure 9.3.1(b). For the sake of convenience, the diagram D in Fig-
ure 9.3.1(b) will be denoted by the following notation:
Explicitly, we want to move all the ai, a2, ... ,an crossing points
on the left to the right. In other words, we want to show that D can be .
deformed into a new diagram
For simplicity, let d
i
=ai + Ci (i = 1,2, ... , n), and call the half-
circular band with b
i
crossing points the i
th
arm of D. ,!;:
First, move the an, the left-most crossing point, to the right by'}ll
twisting the (n + l)th arm either right or left an appropriate number '}
of times. '::,i
Thus, D is deformed into
"'\
Secondly, rotate, an-l times (in either direction), the interior part
of the dotted line around the horizontal axis A, keeping the exterior
part of the dotted line fixed, Figure 9.3.2. ,:J
.-.__
Figure 9.3.2
A
185 Tangles and 2-Bridge Knots
Thus, an-l crossing points have been moved to the right. Such a
rotation will hereafter be called a horizontal rotation.
There are two cases to consider.
(I) If an-l is even, then the resulting diagram is
(II) If an-I is odd, then
D2 = (aI, a2, ., a
n
-2, 0,0 I b
l
, b
2
, ... ,
b
n
-
1
, b
n
, b
n
+
1
I Cl, C2 , C
n
-2, d
n
-
I
, lin),
where b
i
means that the i
th
arm with bi crossing points is above the
horizontal axis, as shown in Figure 9.3.3.
Figure 9.3.3
Now to lower the (n - l)th arm, we shall form a new horizontal
(n _l)th arm (without crossing points) under A, Figure 9.3.4.
B
Figure 9.3.4
Chapter 9
.;:!
Next, rotate, b
n
-
1
times the interior of the dotted line, in Fig-
ure 9.3.4, around the vertical axis B so that the b
n
-
1
crossing points
that lie above A disappear, but bn-l crossing points are created in the
lower arm. Such a rotation will be called a vertical rotation.
The resulting diagram, Dg is of the form:
(a) if b
n
-
1
is even, then
D
3
= (aI, a2, . , a
n
-2,0, I bl , b2 , ,
',f:
b
n
-2' b
n
-
I
, b
n
, b
n
+
1
I Cl, ... , Cn-2, d
n
-
l
, d
n
). :;,
(b) if b
n
-
l
is odd, then
J
By repeated application of horizontal rotations and vertical rota- ~
tions an appropriate number of times, we shall eventually obtain a new } ~
diagram, Figure 9.3.5,
or
D
n
= (Cl' 0, ... , I b
l
, b
2
, , bn+11 aI, d2 , , d
n
).
Figure 9.3.5 .
It is easy to move the final al or CI crossing points on the left
to the right by applying a horizontal rotation on the interior of the
dotted square.
187 Tangles and 2-Bridge Knots
Exercise 9.3.1. Confirm that the procedures outlined in the above
part of the proof work by performing them on the regular diagram in
Figure 9.3.6.
Figure 9.3.6
Therefore, we may think of a standard regular diagram for a 2-
bridge knot (or link) as the one shown in Figure 9.3.7.
...C
"'-----_... ~ ...
Figure 9.3.7
The number of twists is denoted by the integer ai, and we can define
the sign of ai as follows:
If i is odd then the right twist is positive, if i is even then the left
twist is positive (cf. Figure 9.2.4).
The regular diagram (Figure 9.3.7) may also be considered to be the
denominator of the rational tangle T(al' a2, ... , a2k+l, 0). So Theo-
rem 9.3.1(1) has been proven. Conversely, it can be easily seen that the
denominator of a rational tangle has a regular diagram that is the stan-
dard regular diagram of a 2-bridge knot (or link), such as for example,
in Figure 9.3.7.

We say that a 2-bridge knot (or link) that has a standard regu-
lar diagram of the form in Figure 9.3.7 is a 2-bridge knot (or link) of
(ai, ... , a2k+l) type, and we shall denote it by C (ai, a2, ... , a2k+I).
Exercise 9.3.2. Show that the 2-bridge knot C (4, 2 ~ 1 ,-2) is
D(T (4, -2,5,1, -2, b, where b is an arbitrary integer. ~ ~ h e r by
changing the regular diagram, show that it is equivalent to C(3',2,6), see
Figure 9.3.8.
Chapter 9
(a)
Figure 9.3.8
(b)
188
Proposition 9.3.2.
2-bridge knots (or links) are alternating.
Exercise 9.3.3. (1) Show that C( -aI, -a2, ... , -a2k+1 ) is the mir- ,':;;
ror image of C( ai, a2, . , a2k+1 ). tJ
(2) Show that
C(al, a2, ... , a2k+l) C(a2k+l, ... , a2, al).
J
We may assume, due to Exercise 9.2.6, that the signs of all the ai of
a rational tangle, T( al, a2, ... , a2k+1 ), excluding exceptional
are the same. Therefore, this regular diagram is an alternating diagram.
';,J
','

:;;
i,;
:t
Exercise 9.3.4. Prove Proposition 9.3.2 without using Theorem 9.3.1
and Exercise 9.2.6. What needs to be shown is that the standard dia-
gram of a 2-bridge knot (or link) can be deformed into an alternating
diagram. In particular, find an alternating diagram of the knot in Fig-
ure 9.3.6.
In what follows, we shall consider only 2-bridge knots (or links)
that are obtained from non-exceptional rational tangles.
In a similar way to a rational tangle, we can associate a 2-bridge
knot (or link) C(al, a2, ... , a2k+I), with ai =I 0, with a rational num-
ber
189 Tangles and 2-Bridge Knots
(Nota bene, care must be taken when dealing with the continued
fraction expansion.)
If all the ai are positive integers, then 0 < {3 < Ct. However, all
the ai are negative, since ~ < -1, we may assume that Ct > 0 and
f3 < o. We then say that (a, (3) is the type of C(al, a2, . , a2k+l).
Hence, any 2-bridge knot (or link) corresponds to some (a, (3) , where
-a < (3 < a and Ct > O.
Example 9.3.1. The type of the 2-bridge knots in Figures 9.3.8(a)
and (b) is (45,13). In this case, (a) and (b) correspond to the continued
fractional expansions [4, - 2,5,1, - 2] and [3,2,6], respectively, of ~ ~ .
Conversely, from a rational number ~ I ~ I > 1), we can create a
2-bridge knot (or link) in the way described below.
Let us assume that 0 < (3 < Ct. Firstly, let us look for a continued
fraction expansion [aI, a2, ... , am] for ~ such that ai > O.
If m is even and am > 1, then we write
In the case am = 1, we may rewrite it as
From the above continued fraction expansion of ~ we may always
assume that m is odd. If, in fact, we assume this, then we shall obtain
a correspondence between the 2-bridge knot C(al, a2, ... , am) and
~ . Therefore, the type of this 2-bridge knot (or link) is (a, (3). (In
the case when (3 < 0, the subsequent correspondence for the 2-bridge
knot has all the ai negative). If, from two different continued fraction
expansions of " we obtain regular diagrams of two 2-bridge knots,
then they are regular diagrams for the same knot. In short, 2-bridge
knots (or links) are completely classified by their type (a, {3}, as the
following theorem clarifies:
Theorem 9.3.3.
Suppose that K and K' are 2-bridge knots (or links) of type (a, {3}
and (a', (3'), respectively. Then K and K' are equivalent (excluding
orientation in our considerations) if and only if the following holds:
(1) a = a', {3 == {3' (mod a)
or
(2) a = a', {3{3' == 1 (mod a)
Chapter 9

l'f-'
"h
190


Further, the mirror image K* of K is a, 2-bridge knot of type '>
(0, -(3). Therefore, a necessary and sufficient condition for K to be
amphicheiral is that
(3) (32 == -1 (mod a).
The essence of the proof of the above theorem relies on the
that the 2-fold cyclic covering space branched along K is a Lens space .
of type k, 13). Then Theorem 8.1.1 implies Theorem 9.3.3. Besides this
observation, we will not plow into the details of the proof, but direct iii
the reader to Schubert [Sc3] for a more detailed discussion. '1:
It is also known but again we will spare the reader at this juncture 1
the exact details of another proof that the determinant of a 2-bridge knot
(or link) K is a. Therefore, 2-bridge knots cannot solely be classified
by means of their determinants. (Note: The determinant of K does not
depend on the orientation.)
Exercise 9.3.5. Show by determining their types that the two knots
C(1,1,1,2,1) and C(2,4,1) are not equal. (These two knots have the same
knot determinant.)
"
",
Exercise 9.3.6. (1) Find the 2-bridge knots that correspond to 2;
and 2
5
3, and determine whether they are equivalent. ;,t,;
(2) As in (1) but with regard to and 2:.
Exercise 9.3.7. Let K be a 2-bridge knot (or link) of type (0, (3).
Show that K is a knot if and only if a is odd.
So far we have shown that the denominator of a rational tangle
T(al, a2, ... , a2k+l) is a 2-bridge knot (or link). The natural question
is, What can we say about the numerator? The next proposition shows,
by comparing various regular diagrams, that the numerator is also a I
2-bridge knot (or link).
Proposition 9.3.4.
For any integer b, the following hold:
(1) N(T(al, a2, ... , a2k+I)) N(T(al' a2, ... , a2k+l, b, 0)
P(T(-aI, -a2, ... , -a2k+l, b
C(-aI, -a2, ... , -a2k+l).
(2) N(T(al, a2, ... ,a2k D(T(-aI, -a2, ... , -a2k, b)
C(al, a2, ... , a2k - 1, 1).
191 Tangles and 2-Bridge Knots
(3) D(T(al' a2, ... , a2k)) D(T(al, a2, ... , a2k-l, 0))
C(al' a2, ... , a2k-l).
(4) D(T(at, a2, ... , a2k+l)) D(T(al' a2, ... , a2k, 0))
C(l, at - 1, a2, ... , a2k).
Exercise 9.3.8. Calculate the determinant of both D(T(2,3,4)) and
N(T(2,3,4)), and compare them with the continued fraction for [4,3,2].
(This problem explains why we use the terms numerator and denomi-
nator of a rational tangle.)
We know that the sum A +B, of rational tangles A and B, is not
always a rational tangle. In fact, D(A +B) need not necessarily be a
2-bridge knot (or link). However, the numerator N(A+B) is a 2-bridge
knot, or link (why is this?). Further, the following theorem explicitly
calculates the determinant of N(A +B).
Theorem 9.3.5.
Suppose that A and B are rational tangles with fractions and
;, respectively. If we further suppose that the 2-bridge knot (or link)
N(A +B) is of type (a, /3), then a = Ips +qrl. [In this case a is the
determinant of N(A + B).]
Since a detailed proof of the above theorem is given in Ernst and
Sumners [ES2], we refer the reader to that source. It is also possible to
determine /3; however, since its form is not as neat and it will not shed
any substantial insight in what follows, we shall not discuss it further.
The value of a, on the other hand, is expressed by a fraction + ;
that is not reduced is its numerator.
We shall verify that the above theorem is plausible by working
through the next example.
Example 9.3.2. The fraction of T(2,3,4) is [4, 3, 2] = 37, while the
fraction of T (-3, -2) is [-2, -3] = 3
7
Therefore, the determinant of
the numerator of T(2, 3, 4) +T(-3, -2) is 41, since
30 7 90 - 49 41
7 - 3 = 21 = 21-
On the other hand, if we look at the regular diagrams, see Fig-
ure 9.3.9, we have that
N(T(2, 3,4) +T( -3, -2)) = C(-2, -3, -1, -1, -2)_
This knot is a 2-bridge knot of type (41, -18).
Chapter 9 192
Figure 9.3.9
Exercise 9.3.9. Prove that if A is a rational tangle, then there always
exists a rational tangle B such that N(A+B) is a trivial knot. Further,
find B in the case when A = T(2, 3, 4).
2-bridge knots (or links) cannot be distinguished by means of \:,
their Alexander polynomials. For example, the 2-bridge knots of type
(15,4) and (15,7) have both the same Alexander polynomial, namely,
4t-
1
- 7 + 4t, but they are not equivalent. Since 2-bridge knots (or
links) are completely determined by their type (a,{3), one would ex-
,.,;1
pect in theory all their invariants can be determined by a and (3. In '"
fact, the Alexander polynomial can be expressed via the continued frac- )':
tion expansion lab a2, ... , an] of However, the actual method is;t
quite complicated, cf. [BZ*]; so to avoid a turgid explanation, we shall

restrict ourselves to showing how we can express the signature via the
expansion.
Suppose K is a 2-bridge knot of type (0, (3). It is sufficient (why?) "f
to consider only the case < (3 < o. We can also assume that (3 is odd
[cf. Exercise 9.3.11(1)]. So, consider the sequence {a, (3, 2{3, ... , (a -
1)(3}, and divide each element by 20 to obtain a remainder r, -0 <
r < 0:, and these we may write down as the following sequence:
{a, rl, r2, ... , ra-l} (9.3.1)
where ri =1= 0, i =1, 2, ... ,0: - 1. Then the following theorem holds:
Theorem 9.3.6.
The signature, u(K), ofK is equal to the number ofpositive entries
minus the number ofnegative entries in the sequence ofremainders given
in (9.3.1). .
9.3.3. Suppose (0:, (3) =(7,3). So, since
{k(3} ={O, 3,6, 9, 12, 15, IS},
193
we have that
so a(K) = 4 - 2 = 2.
Tangles and 2-Bridge Knots
{rk} = {O,3,6,-5,-2,1,4};
Of the classical invariants, the unknotting number of a 2-bridge
knot is the only one that has yet to be completely decided.
Exercise 9.3.10. Determine the signature of the 2-bridge knots of
type (15,4) and (25,7).
Exercise 9.3.11. (1) Show that for a 2-bridge knot (or link) of type
(0., (3), we can always take {3 to be an odd integer.
(2) Suppose L is a 2-bridge link of type (20, (3).
(i) Show that each component K
1
and K
2
of L is a trivial knot.
(ii) Show that the knot obtained from K
1
in the 2-fold cyclic cov-
ering space M ~ S 3 of S3 branched along K
2
is a 2-bridge knot of
type (0, (3).
(3) Show that a 2-bridge knot (or link) always has
period 2. [Hint: Use (2)(ii).]
(Unoriented) 2-bridge knots (or links) can be characterized by their
graphs (cf. Chapter 2, Section 3). We know that a 2-bridge knot (or
link) has a standard diagram D given in Figure 9.3.7. By colouring the
unbounded domain black, we obtain a signed graph G(D) of the form
shown in Figure 9.3.10(a).
G(D)
(a)
G(D)- {v}
--l......--.04...- - - _. - - t . ~ __.l---.......
(b)
Figure 9.3.10
It is possible to find in G(D) a vertex v such that the subgraph
G= G(D) - {v} that is obtained from G(D) by deleting v and all
the edges incident to v, is a simple line segment, Figure 9.3.10(b). The
converse is also true, and this is more formally expressed by the following
proposition.
Chapter 9 194
Proposition 9.3.7.
An (unoriented) knot (or link) K is a 2-bridge knot (or link) if and
only if K has a regular diagram D with the following property:
The link graph G(D) (defined in Chapter 2, Section 3) has
a vertex v such that G{D) - {v} is a simple line segment (9.3.2)
(with vertices on it).
If condition (9.3.2) is slightly generalized to
G{D) has a vertex v such that G(D) - {v} is a tree, (9.3.3)
then we can define a new class of knots (or links). A knot (or link) in
this class is a called an algebraic link.
Exercise 9.3.12. Consider the three positive graphs in Figure 9.3.11.
(a) (b) (c)
Figure 9.3.11
Let K
a
, Kb' K
c
, respectively, be knots or links whose graphs are
(a), (b), (c) in Figure 9.3.11.
(I) Show that K
a
is a 2-bridge knot and determine its type.
(2) Show that K
b
is an algebraic knot but is not a 2-bridge knot.
Express Kb as the denominator of an algebraic tangle.
(3) What can be said about K
c
?
4 Oriented 2-bridge knots
The bridge number of a knot or link, we know, does not depend
on the orientation. Since we did not assign an orientation to the trivial
tangle that we used originally to create the 2-bridge knots, these 2-
bridge knots and links have no original orientation. Therefore, in order
195 Tangles and 2-Bridge Knots
to investigate further the problem of equivalence with orientation of
oriented knots, Theorem 9.3.3 is not sufficient. So in this final section
we would like to look at the classification of oriented 2-bridge knots
and links.
In the case of 2-bridge knots, there is, in fact, no problem to con-
sider. The reason is that a 2-bridge knot is equivalent with orientation
to the original knot, even if we reverse the orientation.
Exercise 9.4.1. Show that a 2-bridge knot is invertible.
Suppose L is a 2-bridge link of type (a, (3), with a even and
(3 odd. Firstly, as before, let us determine the continued fraction
expansion for and hence construct the unoriented 2-bridge link
C(al, a2, ... , a2k+l). Then to this link we may assign an orientation in
the manner of Figure 9.4.1, making it an (standard) oriented 2-bridge
link of type (a,{3).
Figure 9.4.1
Exercise 9.4.2. (1) Show that we can always assign an orientation to
a 2-bridge link in the manner of Figure 9.4.1.
(2) Show that the 2-bridge link obtained by reversing
the orientation on each of the two components of the (standard) ori-
ented 2-bridge link is equivalent with orientation to the original link.
Therefore, 2-bridge links are invertible.
If we reverse the orientation of only one component of a (standard)
oriented 2-bridge link, L, of type (a, (3), then, in general, it is not
equivalent with orientation to L.
Exercise 9.4.3. Show that the link obtained by reversing the orien-
tation of one component of the (standard) oriented 2-bridge link of type
(4,1) is equivalent with orientation to the (standard) oriented 2-bridge
link of type (4, -3). Consider a similar question in conjunction with
the 2-bridge links of type (10,3) and (10, -7).
The above problem suggests there might exist some sort of equiv-
alence relations, with respect to orientation, for oriented 2-bridge links.
The next theorem actually clarifies the matter.
Chapter 9
196 r
Theorem 9.4.1 [Sc3].
(1) Suppose we reverse the orientation of one component of an (stan-
dard) oriented 2-bridge link of type (a, (3), then the (standard) oriented
2-bridge link obtained is
(i) of type (a, (3 - a) if {3 > 0;
(ii) of type (a,,B +a) if /3 < o.
(2) Two (standard) oriented 2-bridge links of type (a, (3) and (a', (3')
are equivalent with orientation ifone of the following cases hold; further,
these are sufficient conditions for them to be equivalent.
(i) a = a', f3 ={3' (mod 20,)
or
(ii) a =a', {3{3' =1 (mod 20,).
Exercise 9.4.4. Show Theorem 9.4.1(2) is in the case of a 2-bridge
knot equivalent to Theorem 9.3.3.
Exercise 9.4.5. Determine the type (a, (3) of the (standard) oriented
2-bridge link that is equivalent with orientation to the original link but
that has the orientation of one component reversed. Further, find some
examples of such a link. [Hint: The 2-bridge link of type (8,3) is one
such example.]
Exercise 9.4.6. Determine the formula that expresses the linking
number of the (standard) oriented 2-bridge link of type (a, (3). Using
this formula, calculate the linking number of the (standard) oriented
2-bridge links of type (8,3) and (16,5).
Exercise 9.4.7. We can, by means of Theorem 9.3.6, calculate the
signature of a (standard) oriented 2-bridge link. Using this theorem,
determine the signature, u(L), of the (standard) oriented 2-bridge link
of type (16,5), and the signature, u(L'), of the 2-bridge link, L', ob-
tained by reversing the orientation of one component. Finally, confirm
the validity of (6.4.7) for this link.
111, 111"'1J '1 8,"ill,
At the beginning of the 1930s, as a means of studying knots,
E. Artin introduced a concept of a (mathematical) braid(s). This re-
markable insight itself was not sufficient to sustain research in this area,
and so it slowly began to wither. However, in the 1950s this concept
of braids was found to have applications in other fields, and this gave
fresh impetus to the study of braids, rekindling research in this area.
The iridescent hue of this concept flowering into full bloom and activity
occurred in 1984, when V. Jones put into action with inordinate success
the original aim of Artin, i.e., the application of braids to knot theory.
Chapter 10 198
In this chapter our intention is to introduce certain necessary aspects
of the theory of braids that will prove useful when we explain recent
developments in knot theory in the subsequent chapters.
1 Braids
An n-braid is a very particular example of an (n, n)-tangle. On
the top and base of a cube, B, mark out n points, AI, A
2
, .. , An
and AI', A
2
', , An', respectively. These points may be arbitrarily
placed, however, for the sake of neatness, we shall express them in terms
of specific co-ordinates.
Firstly, the co-ordinates for B in R
3
are
B = {(x,y,z) 10:5 x,y,z I}.
Let us choose AI, ... , An, AI" ... , An' as follows
By construction each is directly below the corresponding Ai,
Figure 10.1.1.
y
Figure 10.1.1
Now, join the AI, A
2
, . , An' to A
l
', A
2
', ... , An' by means of
n curves (again, to be precise they should be polygonal arcs) in B. As
usual, they are joined in such a way that these curves (including the
endpoints) do not mutually intersect each other. It is not necessary to
199 The Theory of Braids
join Ai to ~ but we cannot join Ai to some A
j
. We will call these
polygonal arcs strings.
Suppose, now, that we divide the cube into two parts by an ar-
bitrary plane E that is parallel to the base of the cube B. Then, if E
intersects each string (polygonal arc) at one and only one point, we say
that these n strings in B are an n-braid.
(a) (b) (c)
(d) (e)
Figure 10.1.2
Example 10.1.1. Figure lO.1.2(a) and (b) are both examples of
I-braids. Figure 10.1.2(c), however, is not a (l-)braid. Figure IO.l.2(d)
and (e) are typical examples of 2-braids.
If, given two n-braids in a cube, we can, by performing the ele-
mentary knot moves on these strings, transform one to the other, then
we say that these two n-braids are equivalent (or equal).
(a) (b)
Figure 10.1.3
(c)
Chapter 10 200
In the process of applying the elementary knot moves, it is per-
fectly possible that at some stage within the cube we obtain something
that does not conform to our definition of a braid, Figure 10.1.3. Any
I-braid, therefore, is equivalent to the one in Figure 10.1.2(a).
Intuitively, two braids (in a cube), whose endpoints we keep fixed,
can be said to be equivalent if we can continuously deform one to the'
other without causing any of the strings to intersect each other.
Exercise 10.1.1. Show the two braids in Figure 10.1.4 are equivalent.
Figure 10.1.4
In a similar manner as in the case of knots, we can obtain the
regular diagram of a braid by projecting the braid onto the yz-plane.
Figure 10.1.5 is the regular diagram of the braid in Figure 10.1.4.
Figure 10.1.5
By connecting At to AI', A
2
to A
2
', , An to An' by n line
segments, Figure 10.1.6, we can form a special type of braid. In keeping
with our previous nomenclature, we shall call this the trivial n-braid.
At A
2
An
X X X_
l 2
Figure 10.1.6
201 The Theory of Braids
Suppose that a n-braid a has its strings connected as follows:
Al to A ~ l A
2
to A ~ 2 ... , An to A ~ n . Then we can assign to a a
permutation,
(
1 2 n)
il i2 in
We call this permutation the braid permutation. The trivial braid
corresponds to the identity permutation,
(
1 2 n)
1 2 n
Example 10.1.2. The braid permutation for Figure 10.1.2(d) is
~ i) =(1 2),
while the braid permutation for Figure 10.1.4 is
(
123)
3 2 1 =(1 3).
Since if two braids are equivalent, it follows that their braid per-
mutations are equal; so the braid permutation is a braid invariant. This
invariant is not a number or polynomial, as we have used before, but
still a mathematical concept, namely, a permutation. In fact, it is by
far the simplest braid invariant.
2 The braid group
Suppose that B
n
is the set of all n-braids (to be more precise,
all the equivalence classes of these braids). For two elements in B
n
,
i.e., for two n-braids a and {3, it is possible to define a product for
two n-braids a and {3. Firstly, glue the base of the cube that contains
a to the top face of the cube that contains {3. The gluing together
of the two cubes produces a rectangular solid in which there exists a
braid that has been created from the vertical juxtaposition of a and
{3, Figure 10.2.1. (Obviously, we can recover a cube by shrinking the
rectangular solid in half.)
(b)
=
Figure 10.2.2
(a)
1 1
8
Although not necessarily commutative, braids are associative, Le.,
(a(3)'Y =a({3'Y).
'1
Exercise 10.2.1. Show the two products a{3 and {3a, of the a, (3
in Figure 10.2.2(a) and (b), are not equivalent. (Hint: Consider the ; ~
permutations of a{3 and (3a.) I
a Ji : ~
I.
I
~
)l
; : ~
:f
~
~
'I
:J
; ~
t1.
: ~
~
.. ~
~
~ .
.
\ ; ~
~ ~ ;
:g
~
i
: ; ~
.1
'j
; ~
~ ~
Figure 10.2.1 i l ~
This braid is called the product of a and {3, and is denoted by a(3. ~
Similarly, we may define the product, (3a, of {3 and a. In general, it is ~ ;
not true that a(3 = (3a, i.e., a{3 and (3a need not be equivalent braids. ,}
~
: ; l ~
Chapter 10
Figure 10.2.3
203 The Theory of Braids
So far we have described a set B
n
, a product in this set, and also
that associativity holds in the set. Therefore, the natural question to
consider is, Can we make B
n
a group under the action of the product?
In order to show this, we must find a unit element and an inverse ele-
ment. The unit e is simply the trivial braid, Figure 10.1.6. It follows
readily from Figure 10.2.4 that, irrespective of the braid, a, ae = a,
and similarly, eo = o.
a
e
=
Figure 10.2.4
a
In order to find an inverse for an arbitrary 0, let us consider the
mirror image, a*, of a. If we consider the base of the cube to be a
mirror, then the mirror image is the image of a reflected in this mirror.
Exercise 10.2.2. Show that a*a and aa* are equivalent to the
trivial braid e; see also Figure 10.2.5.
Figure 10.2.5
Since this exercise can be solved, we may write that aa* = e and
a* Q = e. Therefore, we now have all the essentials for B
n
to be a
group. This group is called the n-braid group. The inverse element,
a*, of a we shall denote by a-I.
Let us delve a bit further into the structure of these groups. Firstly,
Chapter 10 204
since the I-braid group, B
1
, contains only one element, namely, the
trivial braid, B
1
={e}. The elements of B
2
are equivalent to the two
types of braids drawn in Figure 10.2.6. We have, in fact,
.
I
.
m left twists
Figure 10.2.6
.
.
I
m right twists
Proposition 10.2.1.
Two 2-braids are equivalent if and only if they have been twisted
in the same direction the same number of times.
(For a proof see Exercise 10.3.4.) Therefore, B
2
has an infinite
number of non-equivalent elements; Le., it is a group of infinite order.
For n ~ 2, every B
n
is a group of infinite order; however, there exists
a very easy way of actually writing a general element in one of these
groups.
Among the n-braids, we can create certain specific n-braids by
connecting Ai to A ~ 1 and A
i
+1 to A ~ and then connecting the re-
maining A
j
and Aj (j #i,i+l) by line segments, see Figure 10.2.7(a).
An
An
(J,
0::-
1
,
x. x.
' ,
X- x. x.
11; 11;+1
X-
l 2
X; .l\:i+1
n I 2
n
(a) (b)
. Figure 10.2.7
We shall denote these types of n-braids by Ui. In this way we can
create n-l special n-braids 0'1,0'2, .. , l1
n
-l- In Figure 10.2.7(b) we
have also drawn the inverse element of O'i, the n-braids 0';1. We may
205 The Theory of Braids
now use these elements to express any element in the braid group. For
example, in Figure 10.2.8 we have drawn the braid a =121112"10"30"1.
a
Figure 10.2.8
Conversely, to express any element of B
n
in terms of these O"i and
u:;
1
, first we divide the regular diagram of a braid by lines parallel to
the bottom edge, so that in each rectangle we have only one crossing
point. (If two crossing points are at the same level, then 'by shifting one
slightly upwards and the other slightly downwards, we can eliminate the
problem of having two crossing points at the same level.)
In each of these rectangles, by construction, we have a braid that
is of the form 1i or 1:;1. By definition of the product of braids, we
can decompose (3 into the product of these 1i and u:;l. For example,
the braid f3 in Figure 10.2.9 is u3111u21g121.
=
Figure 10.2.9
Therefore, given any braid, we can express it 88 the finite product
of the O"i and lTil. For this reason, the braids UI, lT2, , 1
n
-1 are
said to generate the braid group B
n
, and so we call Ul, U2, -, Un-l
the generators of B
n
- For example, since any 2-braid may be written
as 1r or ul
m
, where of course m 2:: 0 and
11 = UI ... Ul
~
m times
and
-m -1 -1
0'1 = 1
1
11
~ V
m times
Chapter 10
.\
B
2
is generated by a single element, 0'1. '\:
'....
From the above, we have a way of describing algebraically a braid ;)1
as a product of O'i and u;1. However, these algebraic representations
are not unique. For example, the two braids 0'10'3 and 0"30'1, in Fig-
ure 10.2.10, are equivalent 4-braids.
=
'1
This equality holds even if this braid is considered as a general
n-braid (n 2:: 3), Le., the regular diagram has a few extra non-
intersecting lines added, see Figure 10.2.11(a) and (b).
=
Figure 10.2.10

Therefore, in the 4-braid group, B
4
, the equation 0'10'3 = 0'3U l

holds. FUrther, since 0'10'20'1 and 0"20'10'2 are equivalent 3-braids (cf.
Figure 10.1.4 10.1.5), the following relation holds: I
0'10'20'1 = 0"20'1 0'2
::.t
I

;1
.)
1:
f.
:j

OJ,''
(a)
Figure 10.2.11
(b)
.J
'"
These equalities are called (braid) relations of the braid group. In
fact, if two n-braids are equivalent, then we can change one to other
by using these equalities several times; an example is given a bit later. r
in Example 10.2.1.
A fundamental result on the braid group B
n
is that it only has the
207 The Theory of Braids
following two type of relations called the fundamental relations:
(1) aiaj = ajai
(2) aiai+lO'i = O'i+10'iO'i+1
(Ii - jl 2);
(i = 1,2, ... , n - 2).
(10.2.1)
(Of course, there exist trivial relations, namely, O'ia;l = e and
O"iO"j = (7iO'j; however, we will not consider these as bona fide relations,
and so we will ignore them.)
Collecting together the various relations we have discussed this far,
we may write B
n
in terms of its generators 0"1,0'2, . , O'n-1 and the
these fundamental relations,
(Ii - jl 2) )
(i =1,2, ... ,n - 2).
where the right-hand side is said to be a presentation of B
n
.
For example,
B
1
= (0'1 1--),
B
2
= (0'1, 0'2 10'10"2(71 = 0'20'10'2 ),
B
3
= (0'1, 0'2, 0'3 I 0'10'3 = 0"30'1, 0"10"20'1 = 0'20"10'2, 0'20'30'2 = 0"30'20'3 ).
Since, except for the trivial relation 0"1al1 = e, B
1
does not have any
relations, we denote this lack of relations by --.
Exercise 10.2.3. Determine presentations for the braid groups B
s
and B
6
. (Hint: B
s
has 6 fundamental relations, and B
6
has 10 fun-
damental relations.)
Exercise 10.2.4. Show that the following equalities hold for any B
n
(n 3):
(1) 0"10"20"1
1
= 0'2
1
0'10'2
(2) 0'2a l
1
0'2
1
= (711a21a1.
Example 10.2.1. It follows from Figure 10.2.12 that the two ele-
ments, a and {3, of B
3
given by
are equal.
Chapter 10 208
a
In Exercise 10.2.4 (1) we showed that 0"10'20'1
1
= 0"2
1
0"1(72. Using
this equality we can change the portion inside [=:=J to 0"2
1
(110'2, thus
obtaining
Figure 10.2.12
"
1
\
Let us show this, however, by transforming a into {3 by applying
various braid relations. ,I
We intend to use the relations in Exercise 10.2.4, rather than the 1
'l
fundamental ones, since this will make the calculations much more trans-
parent. Hence, we may change the part 0"20"1
1
0"2
1
in a to 0"1
1
0"2
1
0"1, j
1






'A
]

;i
.i')

't!

:t



Figure 10.2.13
209 The Theory of Braids
Exercise 10.2.5. Show by means of the regular diagrams that the
two elements {31 and (32 in Figure 10.2.13, of B4' are equivalent. By
expressing (31 and {32, respectively, in terms of ai and ail, change
f31 to f32 by means of the braid relations of B
4
(and the equalities that
are derived from these braid relations).
3 Knots and braids
Let us connect, by a set of parallel arcs that lie outside the square,
the points AI, A
2
, . , An on the top of a rectangular diagram of a
braid Ct to the points AI', A
2
', , An', respectively, on the bottom
of the same diagram; see, for example, Figure 10.3.1.
Figure 10.3.1
Then in a natural way we form a regular diagram of knot or link
from a braid. A knot that has been created in this way is said to be
a knot (or link), K, created (or formed) from the braid Ct. Conversely,
we can say that K is the closed bmid (or the closure of Ct). Usually
we assign to each string an orientation that starts from Ai and then
moves downwards along the corresponding string in the cube. Hence,
from a braid a we can form an oriented knot (or link) K. Conversely,
given an oriented knot we can change it suitably so that it becomes an
oriented closed braid. This is encapsulated in the next theorem.
Theorem 10.3.1 (Alexander's theorem).
Given an arbitrary (oriented) knot (or link), then it is equivalent
(with orientation) to a knot (or link) that has been formed from a braid.
Chapter 10 210
Proof
(In Example 10.3.1 and its accompanying diagrams, we show ex- '
plicitly how to form a braid from a knot, so we refer the reader to this
example as template for the proof of this theorem.)
Suppose D is an oriented regular diagram of a knot K. Firstly, cut
D at a point (but not a crossing point) Po, and then pull the loose ,
ends apart so that we now have a (regular diagram of a) (1, I)-tangle
T, Figure 10.3.2. We shall show that we can change this tangle into a
braid Q. The knot, in a sense induced as described previously from this
braid, is equivalent to K.
Figure 10.3.2
(a)

;1,
';i


If the tangle T has m local maxima, then it also has m local i
minima. In the case m = 0, as previously noted, T is a I-braid and
so no proof is required.


So suppose m > 0, then there exists an arc ab in T, which we ;if.
may say is rising upwards, connecting a local minimum a to a local
maximum b, Figure 1O.3.3{a). __ G
Further, we may assume that ab intersects with the other parts

of the tangle at n places. Let us now mark n + 1 points on ab,

i.e., a = ao, aI, ... , an = b, such that the arc 'aiai+l' intersects only
one other part of the tangle, see Figure 10.3.3(a). Next, replace the ::1
,. ,
arc by the much larger arc 8{)p{ PIal. The (large) arc PIP{ ,

lies outside the tangle T, and the arcs aoP{ and alP
I
are selected
in such a way that if passes over (or under) the other segment,
211 The Theory of Braids
~ ~
then they (aoP{ and alP
l
) also pass over (or under) all the other
segments. The result of the above manipulations is a (2,2)-tangle, see
Figure IO.3.3(b).
(a)
Figure 10.3.3
.
.
,
,
,
,
I
,
#
~
"'-- * * ~
...... -.-
(b)
It follows immediately that the oriented knot obtained by joining
(outside the square) the four endpoints of this (2,2)-tangle by curves
is equivalent to'the original knot. [In this tangle, a (= ao) is no longer
a local minimum, and al is a new local minimum.] By using the same
methods as above, with regard to the arcs ~ , ~ , ... , 'an-Ian"(=
'an-Ib;), we will eventually form a (n + 1, n + I)-tangle T that does
not have a local minimum and maximum at a and b respectively, and
further has at most m - 1 local maxima and minima. Continuing this
procedure, we shall finally create a tangle that has no local maxima and
minima. This tangle is our required braid. The same proof will also
work for a link.

Example 10.3.1. Figure 10.3.4(a) rv (d) shows how the procedure of

the above proof works in practice on a regular diagram of a knot. Con-
firm that the knot formed from this braid is equivalent to the origi-
nal knot.
Chapter 10

I:.,/

o\
\
(a) (b) (c)
212 0
(d)
Figure 10.3.4
Exercise 10.3.1. Find a braid (with the fewest possible strings)
.
whose closure (i.e., a closed braid) is equivalent to the knot in Fig- .)
ure 10.3.5.

i;;;

Figure 10.3.5
Now, if two braids are equivalent, their knots (obtained by closing
the braid), it goes without saying, are also equivalent. Caveat lector,
"I!
it is also possible to obtain equivalent knots from the closure of non- ';.1
equivalent braids. For example, the braids in Figure 10.3.6 are non-
equivalent, but their closures are equivalent, to the trivial knot.
e
Figure 10.3.6
213 The Theory of Braids
Exercise 10.3.2. Show that the closure of the 2-braid 0'1
1
,
the 3-braids 0'1
1
0'2 and 0'10'2
1
, and the 4-braids 0'10'20'3
1
and
0'10'"2
1
0'3
1
are all equivalent to the trivial knot.
Therefore, if we wish to apply braid theory to knot theory, we must
first of all explain clearly "how to restrict the braids from which we can
form equivalent knots." To this end, we shall introduce between two
braids the concept of M-equivalence.
Definition 10.3.1. Suppose that Boo is the union of the groups
B
1
, B
2
, , B
n
, ... ,Le., Boo = UBk. We may perform the fol-
~
lowing two operations in Boo; the operations are called Markov moves:
(1) IT f3 is an element of the braid group B
n
(i.e., (3 is an
n-braid), then M
1
is the operation that transforms {3 into
the n-braid ,(3,-1, where , is some element of B
n
, see
Figure IO.3.7(a). The element ,{3i-
1
is the conjugate of {3.
(2) M
2
is the operation that transforms an-braid, {3, into either
of the two (n +I)-braids {3O'n or {30';1, where O'n is a gen-
erator of B
n
+
1
, the (n+ I)-braid group, see Figure 10.3.7(b).
~
M
1
M
2
---+
-(- -)-
---+
r
1
(a)
Figure 10.3.7
(b)
Example 10.3.2. Figure 10.3.7(a) shows how {3 = 0'20'1
1
0'2, an
element of B
3
, changes when M
1
is applied; namely, (3 becomes
i{3i-
1
, where in this case "y = 0'20'1
1
Figure IO.3.7(b) shows how
{3 = 0'20'1
1
0'20'1
1
, an element of B
3
, changes when we apply M
2
; Le.,
we obtain the element {30'3 or {3O'i
l
of B4.
Definition 10.3.2. Suppose that a and {3 are elements of Boo. If
we can transform a into f3 by performing the Markov moves M
1
, M
2
,
Chapter 10 214
and their inverses MIl, M2"l a finite number of times, then a is said to
be Markov equivalent (M-equivalent) to {3 and is denoted by a f"V {3. If
M
a f"V {3, then since {3 rv a, a and {3 are said to be Markov equivalent.
M M
The following theorem shows that Markov equivalence is the fun-
damental concept that connects a knot to a braid.
Theorem 10.3.2 (Markov's theorem).
Suppose that K
I
and K
2
are two oriented knots (or links), which
can be formed from the braids {31 and {32, respectively. Then
Exercise 10.3.3. Confirm that if {31 (32 then K
I
K
2
for the
M
cases in Figure 10.3.7(a) and (b).
The above theorem was first announced by Markov in 1936, however
an exemplary, complete proof appears in Birman [B*].
Exercise 10.3.4. Suppose (11 is a generator of B
2
Prove that
ur = uf <===> m= n.
[Hint: Consider the knot (or link) formed from u1.]
4 The braid index
It follows from Exercise 10.3.2 that a knot (or link) K can be formed
from a (infinite number of) braid(s). So, within this set of braids (from .
which K is formed) there exists a braid, (}, that has the fewest number .::
of strings. The braid a is called the minimum. braid (presentation) of
K, and the number of strings is said to be the braid index of K, and is
denoted by b(K). (Caveat lector, the minimum braid presentation of K
is not unique.) For example, the trivial knot has braid index 1, while,
conversely, the knot(s) of braid index 1 is only the trivial knot.
Exercise 10.4.1. Show that b(K), the braid index ofK, is an invariant I
ofK. .
The knots (or links) with braid index 2 are the elementary torus
knots, Le., only the torus knots of type (n,2), where n # 0, 1.
215 The Theory of Braids
The knots (or links) with braid index 3, in general, are difficult
to list. In fact, it was only relatively recently that it was shown that
certain types of oriented 2-bridge knots and links have braid index 3,
cf. Chapter 11, Section 5. We should now add our customary remark:
As yet no general algorithm has been found to calculate b(K).
Exercise 10.4.2. Determine the minimum braid presentation of the
knot in Figure 10.3.5. (Hint: The braid index is 3.)
Exercise 10.4.3. Show the braid index of the 2-bridge knot of type
(45,7) is 3.
If L is a link, then b(L) is related to the orientation of L. It is often
the case that if we reverse the orientation of a single component, then
the braid index will also change.
Exercise 10.4.4. Show that if we give the torus link K4,2 two differ-
ent orientations, then the respective braid indices are different.
In specific cases, for example, torus knots, as shown below, we can
completely determine their braid index.
Suppose K is a knot and 0 is the minimum braid presentation of K.
Then b(K) is the number of strings of o. In this case, since the regular
diagram of K is obtained by joining b(K) semicircles (the closure strings)
in parallel with the outside of the regular diagram of a (Figure 10.3.1),
the regular diagram of K has exactly b(K) local maxima. Since the
bridge number of K, br(K), is the minimum number of these maxima
(Theorem 4.3.3), it follows that br(K) b(K).
Proposition 10.4.1.
Suppose br(K) is the bridge number and b(K) is the braid index of
a (oriented) knot (or link) K, then
br(K) b(K).
We know (Theorem 7.5.3) that the bridge number of the torus knot
of type (q, r), where q, r > 0, K(q, r), is
br(K(q, r)) = min (q, r).
But, since K(q, r) = K(r, q), we may assume that r < q. Further,
K(q, r) can be formed from the r-braid (cf. Figure 7.1. 7):
Chapter 10
Therefore,
b(K(q, r)) r = br(K(q, r)).
So, combining this with Proposition 10.4.1, we obtain
b(K(q, r)) = br(K(q, r)) = r (= min {q, r}).
Proposition 10.4.2.
If K is a torus knot of type (q, r), q # 0 # r, then
b(K) = min {Iql, Irl}
216
In general, determining the braid index is a difficult problem. How- \
ever, recently, as will be explained in the next chapter, it has been shown
that b(K) is related to the degree of the skein polynomial. So, we can
end this chapter optimistically and without our usual statement: "As
yet no algorithm ... ." Also, due to the above-mentioned result, we
can completely determine the braid index of 2-bridge knots or links; the
exact value will be discussed in Chapter 11, Section 5.
7", """R,,,,t,,,;,,,
In 1984, after nearly half a century in which the main focus in knot
theory was the knot invariants derived from the Seifert matrix, for ex-
ample, the Alexander polynomial, the signatme of a knot, et cetera,
V. Jones announced the discovery of a new invariant. Instead of further
propagating pure theory in knot theory, this new invariant and its sub-
sequent offshoots unlocked connections to various applicable disciplines,
some of which we will discuss in the subsequent chapters.
In the previous chapter we showed how to transform a braid into
a. knot and how to create a group fronl the braids. Therefore, we have
Chapter 11
the following correspondence:
218
knot braid braid group, B
n
.
Suppose we can map the braid group B
n
into some sort of algebraic
system, say, A, whose structure we understand, for example, the group
of invertible matrices, or more generally, an algebra such as a group
ring in which the sum and product have been defined. The aim is to be
able to represent an arbitrary knot by an element of A. (Of course, we
must take care in our selection of A and the correspondence between t;he
knots and A, since a situation may arise in which the correspondence
assigns the same element to each knot.) It should be noted that such an
approach is not exactly new; the Alexander polynomial can be obtained
as an expression in a matrix ring with elements Laurent polynomials in '
a single indeterminate. However, an initial stumbling block is that the
correspondence
knot braid
is not in 1-1 correspondence. To be more precise, to a single knot we
may assign an infinite number of braids. It is our good fortune, due I
to Markov's theorem (Theorem 10.3.2), that each knot corresponds to '
only one M-equivalence class. Therefore, when a braid a corresponds
to a certain value, say, 4>(a:), then if this value 4>(a:) is the same for any
other M-equivalent braid (3, it follows that this 4>(a) is an invariant
of the knot K
a
formed from the braid o. So, from the first condition
of M-equivalence 4> must have the same value for a and ,0,-1. A
typical example of such a function is the trace of a square matrix (Le.,
the sum of its diagonal elements). If we associate 0 with some square
matrix, then if 4>(0) is the trace of this matrix, it follows immediately
that </>(0:) is invariant under the Markov move MI.
Now, if we want to represent the braid group by some algebraic sys-
tem, A, it must have a similar structure to B
n
Further, A should have
a simpler or more restricted algebraic structure than B
n
; otherwise, we ,.
will probably not gain any further insight into B
n
..
Jones, serendipitously, found that one of the (operator) algebras,
which he was studying for other purposes, had a structure that resem-
bled that of the braid group. By means of this insight, Jones was able to
define a function that was invariant under both the Markov moves, M
1
I
and M
2
This function could, in fact, be written in terms of a complex
number q. Following from this, it was possible to assign to each knot a
complex number. To be more exact, to every knot it became possible to
associate a Laurent polynomial in this complex number q (if we replace
219 The Jones Revolution
q by an indeterminate t, then we shall :recover the usually defined Lau-
rent polynomial). This (Laurent) polynomial, one of the new invariants,
is now called the Jones polynomial.
Soon after Jones' announcement of the discovery of his polynomial,
it became clear that this polynomial could be constructed using meth-
ods from other disciplines, for example, statistical mechanics, quantum
groups, et cetera. Hence, knot theory was once again entwined with
fields outside pure mathematics, generating a tremendous amount of
interdisciplinary research and virtually spawning a whole new area of
research.
In this chapter, our intention is to study the new invariants from
the point of view of knot theory, explaining several of their fundamental
properties. Also, in the final section we shall show an application of the
Jones polynomial to knot theory itself, namely, solving a couple of the
Tait conjectures, the original knot theory conjectures.
1 The Jones polynomial
Let us begin by defining the Jones polynomial from the perspective
of knot theory, rather than, say, operator algebras or quantum groups,
which would require the introduction of a great deal of new notation and
definitions, without significantly illuminating our further discussions.
Definition 11.1.1. Suppose K is an oriented knot (or link) and D is
a (oriented) regular diagram for K. Then the Jones polynomial of K,
VK(t) , can be defined (uniquely) from the following two axioms. The
polynomial itself is a Laurent polynomial in 0, i.e., it may have terms
in which 0 has a negative exponent. [We assume (0)2 = t.] The
polynomial VK (t) is an invariant of K.
Axiom 1: If K is the trivial knot, then VK(t) = 1.
Axiom 2: Suppose that D+, D_, Do are skein diagrams (cf. Fig-
ure 6.2.1), then the following skein relation holds.
(11.1.1)
The reader may at first sight mistake these axioms for those that
define the Alexander polynomial, in particular, equations (11.1.1) and
(6.2.1). In fact, it was only a number of years after the discovery of the
Chapter 11 220
Jones polynomial that an extremely unexpected relationship was found
between the Alexander polynomial and in a sense a "truncated" form of .
the Jones polynomial [MeIM]; see also Chapter 15, Section 3. When the
Jones polynomial was first announced, however, many people, not just
mathematicians, were taken aback by the fact that such a seemingly
incongruous change in the second axiom of the Alexander polynomial
should have such a profound significance, and indeed it has been pro-
found. For the proof that the Jones polynomial is defined uniquely by
the above axioms, we refer the reader to Lickorish and Millett [LM3]
and Jones [J1].

The algorithm to calculate the Jones polynomial is completely anal-

ogous to the one for the Alexander polynomial, which we explained in
Chapter 6, Section 2; i.e., it is necessary to form a skein tree diagram.
However, since the coefficients of the two skein relations are different,
the calculation in the Jones case becomes a trifle difficult. The diffi-::
f
:
culty (or perhaps its strength) may immediately be seen if we write out
the Jones polynomial as the sum of the Jones polynomials of the trivial 'J;:!
It-component links, Op, the result of using the skein tree diagram,
.::
.\(
In the Alexander case, a similar expression to that above is super-
fluous since we have already shown that the Alexander polynomial of
Op, for J.L 2 is zero in Proposition 6.2.2, while in the Jones case and
herein lies the fundamental difference, they are not zero. Therefore, our ':.",:1
first step is to calculate the Jones polynomial of 0 p, V0 p. (t).

Proposition
For the trivial It-component link 0""
(11.1.2)
Proof
The proof will be by induction on J.L. If J.L =1, then this is just Ax-
iom 1. So, let us assume for our induction hypothesis that the following
holds:
If we now consider the skein diagram in Figure 6.2.5, then since
D+ D_ 0J1,-1 and Do OJ1,' by the above induction hypothesis
221
and the skein relation (11.1.1),
The Jones Revolution
1 ( 1 )1t-2 (1)1t-2
-(-1)#-2 v't + - - t( _1)1t-2 v't + -
t v't v't
= (v't- ~ V O , . t .
Since the left-hand side of the above formula is
2( 1 )1t-2(1 )
(-1)14- v't + v't t - t
1 ( 1 )1t-2 ( 1 ) ( 1 )
= (-1)Jt- vt + - vt + - vt - -
v't v't v't '
the required result (11.1.2) follows immediately.

We now have the essentials to give an example of a calculation of

the Jones polynomial. As in the Alexander polynomial case, let us write
down the following equalities:
Vo+ (t) =t2Vo_ (t) +tzVoo(t)
VD_ (t) = t-
2
VD+ (t) - t-1zVDo (t),
where for clarity we have set z = (v't - Jt).
Figure 11.1.1
Chapter 11 222
Example 11.1.1. In Figure 11.1.1 we have drawn the skein tree dia-
gram for the calculation of the Jones polynomial of the right-hand trefoil I
knot (it is an interesting exercise for the reader to compare it with the
one in Figure 6.2.6).
It follows from the skein tree diagram that
YK(t) = t2YO(t) +t3zYOO(t) +t2z2YO(t)
= t +t
3
- t
4

Exercise 11.1.1. Calculate the Jones polynomial of the figure 8 knot,

the knot in Figure 5.1.5(a), and the Whitehead link.
In Appendix (II) we list the Jones polynomial of each prime knot
with up to 8 crossing points.
2 The basic characteristics of the Jones polynomial
Let us denote by X II Y the union of two sets X and Y that have
no points in common. For example, we may write the regular diagram
of the 2-component trivial link as 0 II o. Before discussing some of
the characteristics of the Jones polynomial, let us first extend Proposi-
tion 11.1.1.
Proposition 11.2.1.
D_
Figure 11.2.1
Proof
Consider the skein diagrams in Figure 11.2.1. Since D+ and D_
are obtained by just using a Reidemeister move, f2
1
, to add one ex-
tra loop to D, D+ D_ D. Thus substituting this information in
223 The Jones Revolution
(11.1.1), namely, the skein relation is
- tVo_ (t) = (Vi -
then
1 1 (I: 1)
Voo(t) = (Jt _ (t - t)Vo(t) =- vt + Jt Vo(t).
Hence, the case J.t = 1 has been proven. The general case follows
by using mathematical induction, and since this is a straightforward
exercise, we leave it to the reader.

Theorem 11.2.2.
Suppose that K
1
#K
2
is the (connected) sum of two knots (or
links); then
Proof
Suppose D
1
and D2 are, respectively, the regular diagrams of Kl
and K
2
Let us suppose, tempomrily, that D
2
is invisible (for example,
associate D
2
with a single black spot). So, ignoring D
2
let us create
the skein tree diagram for D
1
. Hence, from this we may write down the
Jones polynomial of K
1
,
To continue the calculation, we must now make D
2
visible again.
By construction one component of each trivial link that makes up the
endpoints of the skein tree diagram of Dl has a black dot on it. This
component, when we make D
2
visible, represents D2. We may thus
modify (11.2.1) to
VK
1
#K2(t) = f
1
(t)VD2(t) +f
2
(t)VD21IO(t)
(11.2.2)
+f
3
(t)VD2II02 (t) + ... +fm(t)VD2IIOm_1 (t).
However, from Proposition 11.2.1 it follows that
V0
2
IIO/c(t) = (_l)k(Vi +
=VO
k
+
1
(t)V
D2
(t).
(11.2.3)
Chapter 11
Combining (11.2.3) and (11.2.2) gives us
224
VK1#K2(t) = f
1
(t)VO(t)V0
2
(t) +f2(t)Vo
2
(t)VD
2
(t) +
+fm(t)Vo
m
(t)V0
2
(t)
= {f
1
(t)Vo(t) +f
2
(t)Vo
2
(t) + ... +f
m
(t)VO
m
(t)}V0
2
(t)
= VK
1
(t)VK2 (t).

The following theorem follows immediately from Theorem 11.2.2, ';

so we leave the proof as an exercise for the reader:
Theorem 11.2.3.
Theorem 11.2.4.
Suppose - K is the knot with the reverse orientation to that on K,
then

C
'.$
;)
Proof
In order to calculate V-K(t), we may use the same skein dia-
gram as for the calculation of VK(t) (why?). Hence the result follows..
As a consequence of the above theorem, the Jones polynomial is not
..J
a useful tool in the study of whether or not a knot is invertible. However,
the Jones polynomial is a powerful tool in the study of amphicheirality
of a knot.
Theorem 11.2.5.
Suppose K* is the mirror image of a knot (or link) K, then
"Therefore, if a knot K is amphicheiral, then VK(t) =VK (t -1), Le., '
VK (t) is symmetric.
225 The Jones Revolution
Proof
Suppose D is a regular diagram of K and D* is its mirror image.
If the skein tree diagram of D is, say, R, then we can form the skein tree
diagram of D*, R*, as follows: When we perform a skein operation at
a crossing point, c, of D to make R, at the equivalent crossing point of
D* also perform a skein operation, so forming R*. Since the signs of
the crossing point c of D and the equivalent crossing point in D* are
opposite, the coefficients assigned at this juncture to R and R* differ.
If at a certain segment a of R we have assigned t
2
(or t-
2
), then
at the equivalent segment, a*, of R* we assign t-
2
(or t
2
). On the
other hand, if have assigned tz (or -t-1z), then equivalently at a* we
assign -t-1z (or tz).
Since this change
[note:
is nothing but the replacement of t by t -1, it follows that VK* (t) =
VK(t-
1
).

Example 11.2.1. The Jones polynomial of the right-hand trefoil knot

K is VK (t) =t +t
3
- t
4
, and since this is not symmetric by the above
theorem, K is not amphicheiral.
Example 11.2.2. The Jones polynomial of the figure 8 knot is sym-
metric (cf. Exercise 11.1.1).
Exercise 11.2.1. Calculate the Jones polynomial of the knot K in
Figure 6.4.2 and show it is not symmetric. Therefore, this knot is not
amphicheiral; however, u(K) = 0 (cf. Exercise 6.4.4).
Caveat lector, a symmetric Jones polynomial does not imply that
the knot is amphicheiral.
Exercise 11.2.2. Show that the Jones polynomial of the knot, K, in
Figure 11.2.2 is
Chapter 11 226
However, prove, by calculating its signature, that this knot is not am-
phicheiral.
Figure 11.2.2
Example 11.2.3. The Alexander polynomials of the 2-bridge knots
K
1
and K
2
of type (15,4) and (15,7), respectively, are the same, namely,
4t-
1
-7+4t. But the Jones polynomials of these two knots are different:
V
K1
(t) =t - 2t
2
+ 3t
3
- 2t
4
+ 3t
5
- 2t
6
+t
7
- t
8
V
K2
(t) =t - t
2
+ 2t
3
- 2t
4
+2t
5
- 2t
6
+ 2t
7
- t
8
+t
9
- t
10
,
and so they are not equivalent. (In fact, this also follows from Theo-
rem 9.3.3.)
Example 11.2.4. Calculations of the Jones polynomial have shown
that it is non-trivial in the cases when the Alexander polynomial is
trivial, i.e., equal to 1. However, whether this is true in general is still a I
conjecture. An example of this phenomenon is the Kinoshita-Terasaka
knot (Figure 6.3.2); its Jones polynomial is
VK(t) = -t-
4
+2t-
3
- 2t-
2
+ 2t-
1
+t
2
- 2t
3
+2t
4
- 2t
5
+t
6
.
More generally, at present there are no non-trivial examples of knots
that have the Jones polynomial equal to 1. If, in fact, we can prove that
such an example does not exist, then the Jones polynomial, at the very
least and not insignificantly, will enable us to determine whether or not
a knot is trivial. In this regard, we offer the following exercise for the
ambitious reader:
Exercise (no known solution). Does there exist a non-trivial knot
K whose Jones polynomial is equal to I? Further, is there a non-trivial
j.L-component link, L, u ~ that
(
1 )J.I.-l
VL(t)= (_1)J'-1 vt+ vt ?
227 The Jones Revolution
Although the Jones polynomial is a strong invariant, it is not a
complete invariant. That is to say, there exist an infinite number of
non-equivalent knots that have the same Jones polynomial.
Example 11.2.5 [Kan]. The Jones polynomial of the two knots K
1
and K
2
shown in Figure 11.2.3(a) and (b), respectively, are the same,
namely, are equal to (t-
2
-t-
1
+ I-t+t
2
)2. However, their Alexander
polynomials are
~ K l (t) = (t-
1
- 3 +t)2
~ K 2 t = -t-
3
+ 3t-
2
- 5t-
1
+ 7 - 5t +3t
2
- t
3

Therefore, K
1
and K
2
are not equivalent.
(a) (b)
Figure 11.2.3
This example shows that even though there are cases when knots
can be distinguished by the Jones polynomial and not by the Alexan-
der polynomial, the reverse is also true. So asking which is the more
powerful invariant is to a certain degree a meaningless question.
Example 11.2.6 [LM3]. The two knots K
1
and K
2
in Fig-
ure 11.2.4(a) and (b) are not equivalent; however, their Jones polynomial
are the same,
and their Alexander polynomials are also equal,
Chapter 11
(a) (b)
228
Figure 11.2.4
Showing they are not equivalent can be done by calculating their
respective unknotting numbers, u(K
1
) = 2 and u(K
2
) = 1. However,
it is not an easy matter to show that u(K
1
) = 2.
Exercise 11.2.3. Show that the unknotting number of the knot K
2
in Figure 11.2.4(b) is 1.
If we substitute for t some numerical value, for example, w, the
pth root of unity, then the Jones polynomial is in fact related to other
known knot invariants, at least for relatively small values of p. We shall
discuss some of the known cases.
Proposition 11.2.6.
Suppose that L is a It-component (oriented) link, then
A consequence of this proposition is that the Jones polynomial can
never be zero.
Proof
Since if we substitute t = 1 into the skein formula (11.1.1) we
obtain VL+ (1) - VL_ (1) =0, it follows that
VL+ (1) = VL_ (1) = ~ (1)
(cf. the proof of Proposition 6.3.1). However, due to Proposition 11.1.1,

229
Proposition 11.2.7.
If K is a knot or link, then
The Jones Revolution
where It(K) denotes the number of components of K.
Proof
If we let t = -1 in (11.1.1), then
(11.2.4)
On the other hand, using Theorem 6.2.1, we obtain that
(11.2.5)
Substituting t = -1 in (11.2.5), we obtain
(11.2.6)
Next, we multiply (11.2.6) by the factor -(_l)p(K+)-l, and hence
+
=_(_l)#(K+>-l(H -
Since JL(K_) = JL(K+) and JL(K
o
) = JL(K+) + 1 or JL(K+) - 1,
-(_l)p(K+)-l AKa (-1) = (_l)p(Ko)-l AKa (-1)
and
-(_I)JL(K+)-l (-1) + (_I)I-(K_)-l (-1)
=(_l)#(K
o
>-l(H -
(11.2.7)
Since (11.2.4) and (11.2.7) are exactly the same skein relation, the
required result follows.

Chapter 11 230
A few more cases are given in the next proposition, which we will
not prove.
Proposition 11.2.8.
Suppose K is a knot. Then
(1) [Jl] If w is a primitive 3
th
root of unity, then VK(W) = 1;
(2) [Muk] Let i = A (i.e., a primitive 4
th
root of unity),
(a) if ~ K - l is of the form 8k 1, VK(i) =1;
(b) if ~ K -1) is of the form 8k 3, VK(i) = -1.
Note in the above ~ K -1) is always an odd integer. If K is a link,
then we can obtain similar formulae.
Also, if e is a primitive 6
th
root of unity, then we may calculate
VK(e) [LM1]. A general formula for an arbitrary pth root of unity has
yet to be discovered, this task seems to be by no means an easy one.
The Alexander polynomial of a link L is an invariant of an oriented
link. If, however, we change the orientation of a single component of L,
then the polynomial is completely altered. This change is quite difficult
to categorize. On the other hand, the Jones polynomial of L is an
invariant that is almost unrelated to the assigned orientation, because
if we change the orientation of a component, we can exactly describe
the effect on the Jones polynomial.
Theorem 11.2.9 [LM2].
Suppose L = {Kl' K
2
, , KJ.} is a p,-component (oriented) link.
Further, suppose L = {K
1
, K
2
, , Kp.-l, -Kp.} is the link with the
same orientation as L, except the orientation on the component KIJ. has
been reversed. Then
p.-l
where l = E lk(Kp., K
i
).
i=l
Exercise 11.2.4. Suppose Land L' are two oriented links obtained
from two different orientations on the torus link K
4
,2. Calculate the
Jones polynomial of each, and so check the result of Theorem 11.2.9.
Further, calculate their Alexander polynomials and then compare the
differences between the two cases.
Exercise 11.2.5. Reread the proofs of Theorems 11.2.2, 11.2.4, and
11.2.5, and along the same lines reprove Theorems 6.3.3 and 6.3.5.
231
3 The skein invariants
The Jones Revolution
(11.3.3)
If we look carefully at the skein relation (11.1.1) that defines the
Jones polynomial and compare it in particular with (6.2.1), then it is
possible to perceive that the coefficient of VK+ (t) and VK_ (t) need not
necessarily be limited to t and t. In general, it seems that we may take
an arbitrary function of t and allow it to vary. In other words, it should
be possible to choose a coefficient with more than one indeterminate.
Actually, in this way it is possible to define the most general polynomial.
Definition 11.3.1. Suppose K is a (oriented) knot (or link) and D is
a regular diagram of K. Then we may, by means of the following axioms,
define a polynomial of K, SK(X, y, w), in the three indeterminates x,y,w
that may have negative exponents.
(1) If K is the trivial knot 0, then So(x,y, w) = 1.
(2) With regard to the skein diagrams D+, D_, Do,
the following equality holds: (11.3.1)
XSD+(X,y, w) - ySD_ (x,y, w) = WSDo(X,y, w).
The polynomial SK(X, y, w) is an invariant of K.
Before we calculate SK(X, y, w) for several knots, let us consider a
slight simplification of SK. From the above definition it seems that SK
is a polynomial in 3 indeterminates; however, essentially it is a polyno-
mial in only 2 indeterminates. In order to show this, let us consider the
2-variable polynomial PK(v, z) defined by (11.3.2).
(1) For the trivial knot 0, Po(v, z) = 1
(2) With regard to the skein diagrams D+, D_, Do,
the following formula holds: (11.3.2)
~ P D + (v, z) - vP
D
_(v, z) = ZPDo(V, z).
That ~ 1 (v, z) and SK(x, y, w) are intrinsically the same follows
from the two equalities in (11.3.3).
Exercise 11.3.1. Show that
(1) SK(';' v,z) = PK(V,Z)
(2) P K ~ ;..;y) =SK(X,y,W).
Chapter 11 232
In general, a polynomial defined by skein relations is called a skein
polynomial. PK(v, z) is the most generalized form of a skein polynomial
and is usually called the HOMFLY polynomial; the initials stand for
the surnames of the mathematicians who, at roughly the same time,
discovered this polynomial.
If we compare, the skein relation definitions of the Alexander and
the Jones polynomial [Definitions (6.2.1) and (11.1.1)] with the skein
relation of PK(V, z), (11.3.2), then the following is an easy consequence:
Proposition 11.3.1.
Suppose that K is a (oriented) knot (or link), then
(1) VK(t) = PK(t, v't -
(2) .6.
K
(t) = PK(l, v't -
So it may be said that VK (t) and a
K
( t ) are special cases of
PK(v, z). However, is not a special case of VK(t). As we have
already mentioned, they are essentially different polynomials.
To calculate SK(X,y,W) and PK(v,z) for an arbitrary knot (or
link) K, it is better to use the skein tree diagram.
Exercise 11.3.2. Show
1-v
2
Poo(v,z) = --.
vz
Exercise 11.3.3. Show, using a skein tree diagram, that the skein
polynomial of the right-hand trefoil knot K is
4 The Kauffman polynomial
We know that if we can transform one knot diagram to another knot
diagram via the moves n
1
, O
2
, 0
3
or their inverses, then
they are equivalent. So a possible approach to show that the function
we have developed is a knot invariant is to show it remains unchanged
233 The Jones Revolution
under these Reidemeister moves. Hence, we must investigate in what
way all three Reidemeister moves affect the function, especially since
their characteristics differ completely in several aspects. For example,
we know that if we apply 0
3
or (li
l
, then the number of crossing
points of the regular diagram D remains unchanged. A further impor-
tant consideration is that even if we give D an orientation, the Tait
number (Definition 4.5.2), w(D), remains unchanged when we apply ei-
ther (l2 or 0
3
, or its inverse. However, the Tait number itself is not
a knot invariant; therefore, it would seem it is not sufficient to show
two regular diagrams are invariant under just (l2 and 0
3
and their
inverses. The question is, How "far" are we from a knot invariant if
we restrict ourselves to just to O
2
or (l3 or its inverse? In this re-
gard, L. Kauffman, with extreme perspicacity, arrived at the following
important observation:
Definition 11.4.1. Let us call the Reidemeister moves O
2
or 0
3
and their inverses regular moves. Then, if we can obtain a regular
diagram D' by applying these regular moves a finite number of times
to a regular diagram D of some knot (or link), we say D and D' are
regular equivalent.
Kauffman's principle.
Suppose a function, f, with indeterminate t (for a multivariable
[unction, see Definition 11.4.2) is invariant under the regular moves. If
we choose m suitably (it will depend on the regular diagram), then
tmf is an invariant of knots (and links).
Exercise 11.4.1. Show that the Tait number (or the writhe) of an
oriented regular diagram is invariant under regular equivalence.
Let us explain the above principle by considering a couple of exam-
ples. As in the first example, we shall consider the Kauffman bracket
polynomial defined below. It is essentially the same as the Jones poly-
nomial, however, this polynomial has certain special properties for some
particular types of knots, such as alternating knots and links. Conse-
quently, it has had a significant impact on the study of alternating knots
(and links); we shall discuss this in more detail in the next section.
Suppose K is an unoriented knot (or link) and D is a regular dia-
gram of K. Cut (splice) each crossing point of D in the two ways shown
in Figure 11.4.1 (nota bene, this splicing process is independent of the
sign of the crossing point).
Chapter 11
x
)(
234
Figure 11.4.1
The reason we need to make sure that our knot (or link) is un-
oriented is because if we assign an orientation to D and then cut, the
orientation on the new regular diagram will not longer be compatible,
see Figure 11.4.2.
x
-+
)(
Figure 11.4.2
We shall now use the above process of splicing a crossing point to
define the Kauffman bracket polynomial.
Theorem 11.4.1 [Kaul].
Let D be an unoriented regular diagram of a knot or link K. Then
there exists a unique one-variable integer polynomial PD(A) (with pos-
sibly negative exponents) that satisfies the following four conditions:
(1) PD(A) is invariant under regular equivalence.
(2) If D is the trivial diagram 0 of a trivial knot, then
Po(A) = 1.
(3) H D consists of two split regular diagrams D
1
, D
2
, i.e., D =
D
1
II D
2
, then
(11.4.2)
(4) Let D, D, fil be the skein diagrams given in Figure 11.4.3.
Then the following equality holds:
(11.4.3)
235
~ - _ .....~
/' /\
\./ ../
~ _#
-.......
~ _ ......~
- - - / ~ ~ ,
, ,
, ,
, ,
:/\,:
, ,
, ,
" I
~ ~ - ~ ~
......-
The Jones Revolution
D fi D
Figure 11.4.3
Po(A) is called Kauffman's bracket polynomial, as noted from the
theorem defined on the regular diagram D of a knot or link. For example,
(11.4.1) does not mean that PK(A) = 1 for the trivial knot K. In fact, to
evaluate Po(A) for D = 00, we must use (11.4.3) to eventually obtain
P
co
(A) = -A-3. Therefore, Po(A) is not invariant under the first
Reidemeister move, 0
1
. However, it is possible to define an invariant
from PD(A) that is also invariant under 0
1
; this is an implication of
Kauffman's principle.
Theorem 11.4.2.
Suppose D is an oriented regular diagram of an oriented knot (or
link) K. If Po(A) is the Kauffman bracket polynomial of the "unori-
ented" diagram D, and w(D) is the Tait number (writhe) of D, then
define
(11.4.4)
Then Po(A) is an invariant of an oriented knot (or link), denoted
by PK(A).
If we substitute A = t-!, then PK(A) coincides with the Jones
polynomial VK(t) of K. Namely,
(11.4.5)
Therefore, PK(A) is essentially the same as the Jones polynomial.
[We should note that Po(A) is multiplied by (-A-3)w(D) to eliminate
the effect on Po(A) of the kink, since PCO (A) = -A-3.]
Proof of Theorem 11.4.2.
Suppose that D' is a regular diagram of. K that has been obtained
by performing a single Reidemeister move on D. Then it is sufficient to
show
Chapter 11 236
Firstly, let us suppose that D' has been obtained from D by per-
forming n
2
, n
3
or their inverses. By Definition 11.4.1, D and D' are
regular equivalent, and so by Theorem 11.4.1,
Po(A) = PD/(A).
Further, since w(D) = w(D') (cf. Exercise 11.4.1), it follows that
This leaves the case of D' obtained by applying 0
1
(or nIl)
to D.
D
.. -.- ......
,
# ,
# ,
, \
, ,
I I
. .
. .
, ,
, ,
, #
,
.. ..'
... _--
Figure 11.4.4
D
Since D' has an extra crossing point, to evaluate po/ (A) we need
to use the following skein tree diagram.
, ........
,
,
,
,
.
.
,
,
,
'"
' .. ......
DUO
Then, by (11.4.2) and, (11.4.3), we have
237 The Jones Revolution
Irrespective of how we assign the orientation to D, the sign of the
new crossing point is -1. Therefore, w(D') = w(D) - 1. This fact, in
conjunction with (11.4.4), allows us to write the following:
PO' (A) =(-A-3)w(O')PD,(A) = (-A-3)w(D)-1(-A-3)Pn(A)
= (-A-3)w(D)po(A) = Pn(A).

.-. 1 3 () 1
Exercise 11.4.2. Check that PK(t-
i
) = (-t:r)W D Po(t-:r) satisfies
Axiom 2 in Definition 11.1.1 and show that Pn(t-!) =VK(t).
Example 11.4.1. To evaluate Po(A), Pn(A) for a regular diagram
of the positive Hopf link, L, we shall use the skein tree diagram, but
first we should note that w(D) = 2.


GD
y

y
Therefore, by reading off the coefficients from the skein tree dia-
gram,
Po(A) = A
2
(_(A
2
+A-
2
+ 1 + 1 +A-
2
(_(A
2
+A-
2

= _A
4
-A-
4

Since w(D) = 2, Pn(A) = (_A-3)2(_A4_A-4) =_A-

2
_A-IO,
and so PD(t-
i
) = -t! - ti, which not surprisingly is the same as the
Jones polynomial of L.
Chapter 11 238
Exercise 11.4.3. Evaluate Po(A) for the regular diagram of the
right-hand trefoil knot K, Figure O.l(c). Subsequently evaluate FD(A)
to check that Po(t-
i
) = VK(t).
As an extension of the Kauffman bracket polynomial (and the Jones
polynomial) of a knot (or link) K, it is possible to define a two-variable
integer polynomial (with possible negative exponents), FK(a, x) , using
similar arguments as the above. This will form our second example of
the efficacy of Kauffman's principle.
Theorem 11.4.3 [Kau2].
Suppose D is a regular diagram of a knot (or link) on which no
,orientation has been assigned. Then there exists the 2-variable polyno- I
mial AD(a,x) that is invariant under the regular moves, provided the
following 3 conditions are satisfied.
(1) For the trivial diagram 0,
Ao(a,x) = 1. (11.4.6)
(2) Suppose D, D', 0, 0
'
are the same except in the neighbour-
hood ofa single crossing point. Within this neighbourhood, the ,
regular diagrams of the knot (or link) are as in Figure 11.4.5. I
Then the following relation holds:
AD(a,x) + Ao,(a, x) =x{Aj)(a,x) + Ao,(a, x)}. (11.4.7)
D
:/x\
. ""
, ,
, ,
, #
... _.-,'
n' fi fi'
0,'\
Figure 11.4.5
(3) Suppose D, D, Do are the same except within the neighbour-
hood ofa single crossing point. Within this neighbourhood the )!
regular diagrams of the knot (or link) are as in Figure 11.4.6.
Then the following formulae hold:
(i) AD(a,x) = aADo(a,x)
(ii) An(a,x) = a-1ADo(a,x).
(11.4.8) I
239 The Jones Revolution
'(5-- .......... ,
, \
, ,
, ,
,
.
, , ,
\ I
\ ,
, ,
....... -..-,-'
D
... ---
{ ~ ... ~
I ~ \
. ,
I ,
. .
. ,
, ,
\ I
, ,
"', ~
-- ....-..,'-
D
'0;;-"
, ,
, \
, ,
, .
. .
. .
, I
\ I
, ,
" ,
....... -.- .. -'"
Do
Figure 11.4.6
This 2-variable polynomial AD(a, x) is also unique. As in the case
of the Kauffman bracket polynomial, this invariant is defined from the
regular diagram. So, as before we do not define AK(a,x) = 1 if K is
the trivial knot. In fact, AD(a, x) is not invariant under the first Reide-
meister move, 0
1
Therefore, in order for AD(a,x) to be also invariant
under 0
1
, we need to define a new polynomial.
Definition 11.4.2. Suppose K is an oriented knot (or link) and D is
a (oriented) regular diagram for K. Further suppose that w(D) is the
Tait number of D. Then, let
FD(a, x) = aw(D)AD(a,x). (11.4.9)
Theorem 11.4.4.
Fn(a, x) is an invariant of the oriented knot (or link) K, indepen-
dent of its regular diagram D. This invariant is called the (2-variable)
Kauffman polynomial.
The proof of this theorem runs along the same lines as the proof of
Theorem 11.4.2, so it is left as an exercise for the reader.
Let us now give an example of a calculation of the Kauffman poly-
nomial. For the sake of clarity, we shall write AD(a,x) as A(D).
Example 11.4.2. Suppose L = {00}. In order to calculate A(D)
we need to uSe (11.4.7),
A(CO) + A(OQ) =x{A(CJ) + A(OO)},
(11.4.10)
where CO is a regular diagram of the trivial knot, but it is not regular
equivalent to 0, so A(CO) # A(O).
By (11.4.8),
A(OQ) = aA(O) = a
and
Chapter 11
Substituting these into (11.4.10) gives us that
a-I + a =x{l + A(OO)}.
Therefore,
240
A(OO) = a+ a-I -1.
x
On the other hand, weD) = 0, so the Kauffman polynomial of L is
FL(a, x) = AL(a,x) =x-lea + a-
l
) -1.
Exercise 11.4.4. Show that the Kauffman polynomial of the right-
hand trefoil knot K is
FK(a, x) =a-
3
{x
2
(a+a-
1
) +x(1 +a-
2
) - (2a+a-
l
)}
[Hint: Since CD) and 00 are regular equivalent, Aj)) = A(OO).]
As in the case of the bracket polynomial, the Kauffman polynomial
and Jones polynomial are related.
Theorem 11.4.5 [Kau2].
Suppose K is a (oriented) knot (or link), then
FK(-t-
i
,t-l + t--l) =VK(t).
Exercise 11.4.5. Check the validity of Theorem 11.4.5 by substitut-
ing the above values into the Kauffman polynomial calculated in Exer-
cise 11.4.4.
Hence, by the above theorem the Kauffman polynomial, as well as
the Kauffman bracket polynomial, may be thought of as extensions of
the Jones polynomial. As in the case of the Jones polynomial, the two
Kauffman polynomials are not complete invariants, because there exist
many non-equivalent knots with the same Kauffman polynomials and
Kauffman bracket polynomials.
In general, no matter how we construct the skein polynomials, there
will always exist a knot (or link) that cannot be distinguished by the
given polynomial. Therefore, it is impossible to find a skein polynomial
that will be a complete knot invariant.
241 The Jones Revolution
The Kauffman principle, as we shall see in the next chapter, plays
an important role when a knot invariant is defined from an exactly
solvable model of the type found in statistical mechanics.
5 The skein polynomials and classical knot invariants.
(Alternating knots and the Tait conjectures)
Although in definition an alternating knot is a simple matter, from
the inchoate stages of knot theory to the present, this knot has been
thought to be non-trivial. Furthermore, a reduced alternating diagram
can readily be understood intuitively, Le., it is a regular diagram that
we cannot change into one with a fewer number of crossings. However,
it is only quite recently that this has been proven to the standards of
mathematical rigour. The study of alternating knots has a long history,
and over this period a panoply of its characteristics had been proven,
but the Tait conjectures had haunted and proved unyielding.
The new invariants, the Jones polynomial and the skein polyno-
mials, which have been the focus of our attention, in this chapter, in
actual fact have played the crucial role in solving the primary Local
problems. These were the problems/conjectures that concerned Tait in
the 19
th
century and from which germinated modern knot theory. The
conjectures deal with a specific type of knot, namely, the alternating
knots. In order to develop these conjectures, we shall in this chapter
take a more detailed look at Local problems for alternating knots and
links.
The alternating in alternating knot signifies a geometric character-
istic, i.e., the alternating nature of the crossings. It is the case that on
occasions geometric invariants of knots can be determined by algebraic
invariants. The next theorem is a classic example of such a case.
Theorem 11.5.1 [Mus1].
Suppose K is an alternating knot. The genus ofK, g(K), is equal to
the degree of the Alexander polynomial ofK. Further, the Seifert surface
constructed from an alternating diagram has the minimal genus g(K).
This theorem shows that the genus of K, a geometric invariant,
is completely determined by the (maximum) degree of the Alexander
polynomial, an algebraic invariant. In the case of torus knots, we know
(Theorem 7.5.2) that a similar result holds. This gives a further dis-
tinct example of a geometric characteristic, namely, a torus knot, being
Chapter 11 242
reflected in its algebraic nature.
In addition, the Alexander polynomial for an alternating knot has
a special form.
Theorem 11.5.2 [Mus2].
Suppose K is an alternating knot and
~ K t = a_mt-
m
+a_m+lt-m+1 +... +amt
m
is its Alexander polynomial (am =1= 0 =1= a_
m
). Then
(1) a-
m
, a-
m
+l, ... , am are never equal to zero;
(2) the sign of two consecutive coeflicients alternates, i.e.,
aiai+1 < 0 (i = -m, -m + 1, ... , m - 1).
Exercise 11.5.1. By looking again at the calculations of the trefoil
knot and the figure 8 knot, show the veracity of the above theorem.
Exercise 11.5.2. If K(q, r) is a torus knot with min {Iql, Irl} ~ 3,
show that it cannot be an alternating knot.
From condition (2) in the above, the Alexander polynomial of an
alternating knot is an alternating polynomial. 14
The signature of an alternating knot is also comparatively easy
to calculate. However, since a few preliminary preparations are re-
quired, we shall not discuss the details here but refer the reader to
Murasugi [Mus3].
Also, by looking at the reduced regular diagram, it is relatively easy
to determine whether or not a given alternating knot (or link) is prime.
Theorem 11.5.3 [Men].
Suppose D is a reduced alternating diagram of an alternating knot
(or link) K. If K is not prime, then there exists a circle C on the plane
that intersects D in exactly two points, and this circle C dividesD into
two non-trivial (1, I)-tangles (one of the tangles lies within the circle,
while the lies without).
Caveat lector, Theorem 11.5.3 does not hold if D is not an alter-
nating diagram.
Exercise 11.5.3. Show that 2-bridge knots (or links) are prime knots.
Exercise 11.5.4. Determine whether the two knots in Figure 11.5.1
are prime knots.
243
(a)
The Jones Revolution
(b)
Figure 11.5.1
If we take a second look at Theorem 11.5.1, this theorem seems
to suggest that perhaps the degree of a polynomial is the most impor-
tant aspect of the polynomial as an invariant. Previously we showed
that the degree of the Alexander polynomial of a knot K is not greater
than its genus (Theorem 6.3.7). The natural question is, In what way
(if any) is the degree of the Jones polynomial of K an invariant of K
or on what knot theoretical property does it depend? Since the Jones
polynomial may also have negative exponents in addition to defining
its maximum degree, maxdeg VK(t), we also define the minimum de-
gree, mindeg VK(t). We can prove that these can be estimated using
other invariants.
So, suppose D is a regular diagram of an oriented knot (or link) K.
Further, let c+(D) and c_ (D) be the number, respectively, of positive
and negative crossing points of D. Finally, let O'(K) be the signature of
K.
Theorem 11.5.4 [MusS].
For an arbitrary connected regular diagram D of K, the following
inequality holds:
mindeg VK(t) -c_(D) - iO"(K)
1
maxdeg VK(t) c+(D) - 20"(K).
(11.5.1)
(11.5.2)
Therefore, if we set span VK(t) = maxdeg VK(t) -min deg VK(t),
then the following holds:
(11.5.3)
If K is a (non-split) alternating knot (or link) and D is a reduced
alternating diagram for K, then in (11.5.1) and (11.5.2) the inequalities
Chapter 11
become equalities; therefore,
244 I,
span VK(t) =c+(D) + c_(D) = c(D) (11.5.4)
maxdeg VK(t) + min deg VK(t) = c+(D) - c_(D) - u(K)
=w(D) - u(K). (11.5.5)
In the case when K is a prime knot and D is either a non-alternating .
diagram or is a regular diagram that is not reduced, then (11.5.4) and
(11.5.5) do not hold.
Exercise 11.5.5. Confirm (11.5.3) and (11.5.4) hold for all the knots .
listed in Appendix (II).
Exercise 11.5.6. Let L be the oriented 2-bridge link of type (16,5) ~
and L' the oriented 2-bridge link obtained from L by reversing the ori-
entation of one of the components. [L' is of type (16, -11).] Determine
the maximal and minimal degrees of the Jones polynomials of L and L' ,f,
-J.
and confirm these values are consistent with Theorem 11.2.9. } ~
; ; ~ ~
An application of Theorem 11.5.4 is in the proof of the next two ; ~
"primary" theorems, called Tait's first and second conjectures, respec- ::;,
tively.
Theorem 11.5.5 (Tait's First Conjecture).
A reduced alternating diagram is the minimum diagram of its al-
ternating knot (or link). Moreover, the minimum diagram of a prime
alternating knot (or link) can only be an alternating diagram. In other
words, a, non-alternating diagram can never be the minimum diagram
of a prime alternating knot (or link).
Proof
If we restrict ourselves to reduced (connected) alternating diagrams,
then since (11.5.4) and (11.5.5) always hold, the number of crossing
points of D, c(D) = c+(D) + c_(D), is fixed. Also span VK(t) is an
invariant of K that is independent of D. Therefore, c(D) in this case
is also an invariant of K. Further, by (11.5.3) the number of crossing .
points cannot decrease below span VK(t). Hence, D must be a minimum
diagram. Now, (11.5.4) can never hold for a non-alternating diagram of <
a prime alternating knot .. So, such a non-alternating diagram can never
be a minimum diagram.

245 The Jones Revolution

In Chapter 4, Section 5 we defined (cf. Definition 4.5.2) the Tait
number (or writhe), w(D), with respect to a regular diagram D of a knot
(or link) K. This number in general is not an invariant of K. However,
if D is a reduced alternating diagram (and hence a minimum diagram),
then it becomes an invariant of K. This is the essence of Tait's second
conjecture.
Theorem 11.5.6 (Tait's Second Conjecture).
Suppose that D
1
and D
2
are two reduced alternating diagrams of
an alternating knot (or link) K, then w(D
1
) =w(D
2
).
Proof
Let us consider (11.5.5). Since we restrict ourselves to diagrams D
that are reduced (connected) alternating diagrams, (11.5.5) holds. Fur-
ther, the maximum and minimum degrees of VK (t) and the signature
O'(K) are invariants of K that are independent of D. Therefore, w(D) is
an invariant of K.

Let us now give a straightforward application of this theorem. Sup-

pose D is a reduced alternating diagram of an alternating knot K. Then
we can form D*, the reduced alternating diagram of the mirror image
K* of K, by switching the over and under segments at each crossing
point of D. Due to this switch, it is easy to see that the sign of each
crossing point of D* is exactly opposite to the corresponding one on
D. Therefore, w(D*) = -w(D). If, now, we suppose that K is am-
phicheiral, then since K ~ K*, it is also true that D ~ D*. Hence,
w(D) = w(D*) = -w(D), which implies w(D) = O. Hence, the number
of crossing points of D must be even. The thrust of the discussion is
encapsulated in the next theorem.
Theorem 11.5.7.
An t e ~ n t i n g knot whose minimum number of crossing points is
odd can never be amphicheiraJ.
As yet, an amphicheiral non-alternating knot whose minimum num-
ber of crossing points is odd has not been found. So, we cannot say
whether or not Theorem 11.5.7 holds for a general non-alternating knot.
The Tait conjectures were initially bruited about by Tait at the
beginning of the 20
th
century after studying copious alternating knots.
Tait himself tried to prove them, but without much success.
Tait also put forward another conjecture, which we may call the
Chapter 11 246
Third Tait Conjecture.
Suppose an alternating diagram includes a (2,2)-tangle, as shown
in Figure 11.5.2(a).
(a)
--+
(b)
Figure 11.5.2
Let us fix the four points a,b,c,d, and then rotate this tangle by
a half-revolution, Figure 11.5.2(b). Hence, a twist on the right in Fig-
ure 11.5.2(a) has moved to the left. Such an operation is called a (Con-
way) ftype.
Tait's Third Conjecture.
Suppose 0
1
and O
2
are two reduced alternating diagrams of an
alternating knot K. Then we can change 0
1
into D
2
by performing a
finite number of flypes.
This conjecture has very recently shown to be true (MenT].
Exercise 11.5.7. Show that this conjecture implies the second con-
jecture.
(a)
--+
(b)
Figure 11.5.3
Example 11.5.1. Figures 11.5.3(a) and (b) are equivalent regular di-
agrams of an alternating knot. By performing a Hype within the dotted
circle in Figure 11.5.3(a), we may transform it into Figure 11.5.3(b).
247 The Jones Revolution
As was shown above, the degree of the Jones polynomial is clearly
related to the number of crossing points of a regular diagram D of a
knot (or link) K. A natural question to ask is, What types of invariants
of K are the degree of the skein polynomial PK(V, z) related to? Since
PK(V, z) has two variables, we can define two degrees, one that depends
on v, the v-deg PK(V, z), and the other that depends on z, the z-deg
PK(V, z). In general,
z-deg PK(1, z) z-deg PK(V, z). (11.5.6)
Since PK(l, z) is the Alexander-Conway polynomial [Proposi-
tion 11.3.1(2)], its degree is equal to twice the degree of the Alexan-
der polynomial aK(t). Even if the equality does not hold in (11.5.6),
z-deg PK(V,Z) is approximately equal to twice the degree of aK(t).
Therefore, it is v-deg PK(V, z) that piques our interest.
Proposition 11.5.8 [Mol.
Suppose K is a (oriented) knot (or link) that has been formed from
an n-braid (3. Then
v-span PK(V, z) = max v-deg PK(v, z) - min v-deg PK(V, z)
$ 2(n - 1). (11.5.7)
In general, the equality does not hold in (11.5.7). However, there
are quite a few knots for which the equality does hold. For such a knot
K, since K can never be represented by a braid that has fewer braid
strings than n, the braid index of K is exactly n. For example, for
2-bridge knots their braid index is completely determined by the degree
of the skein polynomial.
Theorem 11.5.9 [Mus6].
The braid index, b(K), of an oriented 2-bridge knot (or link) K is
1
i{v-span PK(V,Z)} + 1.
However, excluding these 2-bridge knots, it is not known for which
other knots (if any) their braid index is completely determined by
PK(v, z).
Exercise 11.5.8. Evaluate PK(V, z) for the figure 8 knot K and con-
firm Theorem 11.5.9 for K. (Hint: K is a 2-bridge knot with braid
index 3.)
The motivation behind statistical mechanics is to try to understand,
by using statistical methods, macroscopic properties - the easiest ex-
ample being to determine what happens to water in a kettle when we
boil it - by looking at the microscopic properties, Le., how the various
molecules interact. Statistical mechanics together with quantum me-
chanics have formed a basis for studying the physics of matter, i.e., the
study from the atomic p ~ t of view of the various properties of matter.
In general, the constituent molecules, even if we assume they obey the
principles of dynamics, have extremely complicated means of motion.
249 Knots via Statistical Mechanics
At present, mathematically these motions are virtually impossible to
categorize. So, one reasonably successful method around this problem
has been to form an ideal realization of matter. This realization takes
the form of a statistical mechanical model that is a simplified copy of
matter. The pivot that is essential for the model to at least have math-
ematical meaning is a function Z called the partition junction,
" (-E(O"))
Z = L.Jexp kT '
q
in which we define 0" to be a state of the particular model, E(0") to be
the total energy of this state, T to be the absolute temperature, and k
to be Boltzmann's constant. The sum itself is taken over all the states
of the particular model.
If the partition function of a model can be derived exactly, then
this model is said to be exactly solvable. Numerous models have been
shown to be exactly solvable, especially since the advent of Drinfel'd's
quantum group. Due to this idea of a quantum group, and also by
independent work in statistical mechanics, the partition function has
shown to be closely related to invariants of knots (and links).
In classical two-dimensional statistical mechanics (Le., statistical
mechanics that applies classical dynamics to the dynamics of the micro-
scopic world), two types of models have shown to be the most effective,
the vertex model and the IRF (interaction round a face) model. These
models not only allow us, in a very straightforward manner, to recover
from the partition function the Jones polynomial, but in fact lead to a
whole new series of Jones-type knot invariants [WAD].
In this chapter, we shall describe the exactly solvable 6-vertex
model, which is the first in the above series of knot invariants. This
model, or to be precise, the actual partition function, allows us to in-
troduce a method that culminates in a different approach to the Jones
polynomial. In the third, final section, we explain how to place statis-
tical mechanical concepts and properties in a knot theoretical setting,
which in turn allows a general method of constructing skein invariants
to be introduced.
1 The 6-vertex model
Let lIS consider a 2-dimensionallattice, Figure 12.1.1.
Chapter 12
250 I
Figure 12.1.1
On the lattice we shall define a model called the vertex model. This
model is one of the classical models of 2-dimensional statistical mechan-
ics. As the name suggests, on the four edges that emanate from a vertex,
of the vertex model, we assign state variables, i, j, k, l, Figure 12.1.2.

, 3
.,.1)

t
oA
,
(12 11) to1I
" i,i
Figure 12.1.2
L w(b, c, q, r)(u)w(a, k,p, c)(u +v)w(i,j, a, b) (v)
a,b,c
= Lw(a, b,p, q)(v)w(i, c, a, r)(u +v)w(j, k, b, c)(u).
a,b,c


For this model it is possible to determine the realization probability,
'1
w, at a vertex. This probability is denoted by w(i,j, k, l)(u) and is ..;,
usually called a Boltzmann weight. The variable u is called the spectral
it;
parameter, and it indicates the mutual interactions of the system. .:i
It is well known that for a statistical mechanical model to be solv-
:j;
able, the sufficient condition is the Yang-Baxter equation. In the case of :'ii
the vertex model, this equation has the following form, and a diagram-
matic interpretation is given in Figure 12.1.3:


r
i
p
k
=
r
i
p
k
Figure 12.1.3
251 Knots via Statistical Mechanics
(12.1.2)
Therefore, if we can find a Boltzmann weight, w, that satisfies (12.1.1),
the model is solvable, Le., it is an exactly solvable model.
The 6-vertex model is an example of such an exactly-solvable model.
In this case, the state variables can only take the value or -!, and at
a vertex we assume conservation of the state variables, Le., i +j = k+1,
Figure 12.1.2. If the above conservation condition is not satisfied the
Boltzmann weight at that vertex is zero.
Proposition 12.1.1.
The 6-vertex model has a set of Boltzmann weights w(i,j,k,l)(u),
defined below, that satisfy the Yang-Baxter equation (12.1.1) :
1111 1 1 1 1
w(2' 2' 2' 2)(u) = w(-2' -2' -2' -2)(u) = 1
1 1 1 1 1 1 1 1 sinh u
w(2' -2' 2' -2)(u) = w(-2' 2' -2' 2)(u) = sinh('x - u)
1 1 1 1 1 1 1 1 sinh ,\
w(2' -2' -2' 2)(u) =w(-2' 2' 2' -2)(u) = sinh('x - u)
in all other cases w(i,j,k,l)(u) = o.
In the above, sinh u is the hyperbolic sine function, namely,
sinhu = !(e
U
- e-
U
), and A is another parameter.
In order for the Boltzmann weights to be more easily digested we
shall rewrite them in the form of a 4 x 4 matrix, which we will denote
by R. (Often in statistical mechanics, this matrix is called an S-matrix;
however, it is more common, in the present context, due to another
definition/extension of this same concept by Drinfel'd, to call it an R-
matrix.) The columns of the R-matrix are indexed by the sets (i,j)
and the rows by the sets (k,l), and the order is the reverse dictionary
order. Excoriating the mathematical argot, the first column (and also
the first row) is indexed by (!' !), the next column (row) by
followed by and the final column (row) by -!).
Therefore,
1 0 0 0
0
sinhu sinhA
0
R = IIw(i,j, k, l)(u) II =
sinh(A-u) sinh(A-u)
(12.1.3)
0
sinhA sinhu
0
sinh(A-u)
0 0 0 1
It is easy to see from this representation that if i + j # k + l, then
w(i,j,k,l)(u) =0.
Chapter 12 252
In fact, these Boltzmann weights satisfy other important conditions
in statistical mechanics besides the Yang-Baxter equation, for example,
w(i,j, k, l)(O) == bilbjk, (12.1.4)
where 6
pq
is the Kronecker delta symbol, which is 1 if P == q and 0
otherwise. Also,
wei, j, k, l)(u) == w(-i, -j, -k, -l)(u). (12.1.5)
From the point of view of knot theory, these Boltzmann weights also
satisfy the following important condition, called the unitary condition.
L w(i,j,p, q)(u)w(q,p, l, k)( -u) == bik6jl.
p,q=!,-!
(12.1.6)
If we depict this unitary condition (12.1.6), as shown in Fig- '
ure 12.1.4, then it should immediately bring to mind the Reidemeis-
ter move, {l2 (if we ignore the under- and over-crossing information at
the crossing points). Moreover, the Yang-Baxter equation (12.1.1) is
essentially nothing but the Reidemeister move 0
3
. In other words, the
partition function is unchanged by the regular moves 11
2
and 11
3
So, ..
in essence, it should provide us with a regular knot invariant. Therefore,
by Kauffman's principle (Chapter 11, Section 4) it should yield a knot

invariant. So finding solutions to the Yang-Baxter equation seems to
imply that we then may transform these solutions into knot invariants.
In fact, this is not only feasible but in the late 1980s and early 1990s
was undertaken with great gusto, mainly through the use of Drinfel'd'sif
quantum group, [D]. However, to explain the quantum group approach
would require a substantial amount of new notation and definitions with- ':t
out a significant increase of insight into knot theory. Therefore, we shall
work within the more accessible framework of statistical mechanics. The
interested reader might wish to consult [Tn].
Ie
q
i
Figure 12.1.4
Ie
i
I
j
, .
..Il
253 Knots via Statistical Mechanics
Fromthe point of view of statistical mechanics, in order to fulfill the
objective of deriving a knot invariant, we need to remove the parameters
u and v that occur in (12.1.1) and (12.1.6), i.e., we need a set of
Boltzmann weights that are "independent" of u and v. Hence, the
necessary requirement is to find u and v in (12.1.1) such that u =
u +v =v. The most obvious solution is to set u = v = O. However, if
we do this, then by (12.1.4) it follows that the only non-zero Boltzmann
weights are w(i,j,j, i), and these are equal to 1. So the R-matrix is
[
1000]
o 0 1 0
o 1 0 0 '
000 1
and this will not lead to a new knot invariant. Another solution is
to let u, v ~ 00. In this case what does the matrix in (12.1.3) of
the subsequent Boltzmann weights look like? Let us do the necessary
calculations (we assume A is fixed):
1 11 1 11 11
w(2' -2' 2' -2)(00) = w(-2' 2' -2' 2)(00)
e
U
- e-
U
= lim ------:---
u ~ o o e
A
-
u
- e-(A-u)
1 - e-
2u
= lim -----
u ~ o o e
A
-
2u
- e-
A
= -e
A
,
and
1 1 11 111 1
w(2' -2' -2' 2)(00) =w(-2' 2' 2' -2)(00)
I
. sinhA
= 1m ----
u ~ o o sinh(A - u)
= O.
Therefore, the matrix in (12.1.3) now has the following form
[
1 0
/Iw(i,j,k,l)(oo)/1 = ~ ~ >
o 0
o 0]
o 0
-e
A
0
o 1
(12.1.7)
However, this still does not lead to a very interesting invariant.
So, in order to construct "non-trivial," from the point of view of knot
e! (k-i-l+j)u
,

254 :1
theory, Boltzmann weights, we need to multiply w(i,j, k, l)(u) by a .,1
crossing multiplier and let u --+ 00. The crossing multiplier we need is ::':


Chapter 12
and the subsequent Boltzmann weights are
w(l,k,i,j)(u) = e!(k-i-l+i)uw(i,j,k,l)(u).
4
1
'1

(Nota bene, particular care needs to be taken with the order of i, j, k, 1,
see Figure 12.1.2.) };
',:;!
i

w(i,j, k, 1) = lim w(i,j, k, l)(u).

For example,
- ( 1 1 1 1) 1. -u (1 1 1 1)( )
w 2' - 2' 2' - 2 = e w 2' - 2' -2' 2 u
1
. -u sinh'\
= 1m e
sinh(,\ - u)
=0;
w(-! ! = lim eUw(-! ! ! _!)(u)
2' 2' 2' 2 u-+oo 2' 2' 2' 2
1
. eU sinh'\
= 1m ----
u-+oo sinh('\ - u)
= 1 -
If we set e
2A
= t, then we can write the R-matrix of the (new) 1

Boltzmann weights, w(i,j, k, 1) :
:jf.

'I
::1
(12.1.8)
Exercise 12.1.1. Confirm the matrix in (12.1.8) by calculating the l
remaining w(i, j, k, 1).
255 Knots via Statistical Mechanics
Since R is a square matrix, with a non-zero determinant, we can
calculate its inverse matrix R-1 :
1 0 0 0
0 1-
1
1
0
R-
1
=
t
-Vi
0
1
0 0
-Yt
0 0 0 1
Let us denote the element i,j), (k,l)) of R-
1
by w_(i,j,k,l);
similarly, let us write as w+(i,j, k, 1) the element i,j), (k,l of R.
Although the w(i,j, k, l)(u) differ from the original w(i,j, k, l)(u), the
sina qua non for a knot invariant, the Yang-Baxter equation, (12.1.1),
and (12.1.6) still hold. (The other conditions mentioned above may not
hold.)
Exercise 12.1.2. Show that the Yang-Baxter equation holds for the
Boltzmann weights w(i,j, k, l)(u).
Exercise 12.1.3. Show that the following formula holds:
L w(i,j,p,q)(u)w(p,q, k, l)(-u) = DikDjl.
p,q=!,-!
[Hint: Note that w(i,j,k,l)(u) isOif i+j#k+l.]
2 The partition function for braids
In the previous section we took a lattice to be our model of "mat-
ter." A lattice without much scrutiny may be thought to be a braid.
Hence, our objective in this section is to define the partition function of
this model, Le., a braid, using the R-matrix.
So, suppose {3 is a (oriented) n-braid and D is a regular diagram
of {3. At each crossing point of D, let us look at the four segments that
make up a neighbourhood of that crossing point. We may assign a state
s on the braid by placing a state variable ! or - on each of these four
segments (see Figures 12.2.3 and 12.2.4) at each crossing point. For this
given state, we may assign a Boltzmann weight at each crossing point
of D, as described below.
On the four segments close to a crossing point, c, suppose the state
variables are assigned as shown in Figure 12.2.1.
Chapter 12
'X
k
i
(a)
Figure 12.2.1
'X
k
i/ j
(b)
256
Then,
(i) if the crossing point, c, is positive, Figure 12.2.1(a), then assign
w+(l,k,i,j) toe;
(ii) if the crossing point, c, is negative, Figure 12.2.1(b), then as-
sign w_(l,k,i,j) to c.
Finally, for a fixed state, s, we take the product of all the Boltzmann
weights, namely,
ITw:l:(l, k, i,j).
c
{12.2.1}
We form a knot (or link) from a braid by adding closure strings, I
Figure 12.2.2.
Figure 12.2.2
These closure strings will also have a contribution to a subsequent
knot invariant. Hence, we need also to assign state variables to these
closure strings. But if, for a given state, s, the state variable ak is
assigned to the top half of the k
th
closure string, and the state variable
bk to the bottom half of the k
th
closure string, then we shall assume
257 Knots via Statistical Mechanics
they are equal, see Figure 12.2.2. By adding these closure strings, we
no longer have a lattice model in the original sense, but a model with
certain boundary conditions. Therefore, we need to perform "some sort
of modification" to the product in (12.2.1). In fact, it is known that
even for a statistical mechanical model with boundary conditions, a
modification is required. The result is that for a knot (or link) K with a
regular diagram D formed from a braid (3 and with an assigned state,
s, we have the following modified function:
II
W (l k i JO)t-(a
l
+a2+... +
a
n)
, , , ,
c
(12.2.2)
where (al' ... ,an) are the state variables that have been assigned to
the top half of the closure strings of the braid (3.
The factor t-(al+a2+...+
a
n) that has been added is the "some sort
of modification" that was alluded to previously. The product given in
(12.2.2) is calculated separately for each state, s, on D. The sum (over all
states) of these products is the partition junction Z/3 for this "matter"
(Le., the closed braid),
Zf3 = L:IIw(l,k,i,j)t-(a
1
+a
2
+...+a
n
).
s c
(12.2.3)
In order to calculate Z/3' usually it is not necessary to consider all
the states, s, but rather only those for which the product in (12.2.2) is
non-zero. Such states are called contributing states.
Let us now use the above partition function to calculate several
examples with the Boltzmann weights of the R-matrix in (12.1.8).
Example 12.2.1. For the case {3 = 0'1, there are only three con-
tributing states, as shown in Figure 12.2.3.
\

-t\
1
1\

2 2
-I \-1 -t \1
.1..
\1 2
(a) (b) (c)
Figure 12.2.3
Chapter 12 258
We have as the product of (12.2.2), respectively,
(
....., (1 1 1 11
a) w+ -2' -2' -2' -2)t = t;
....., 1 1 1 1 0
(b) w+(-2' 2' -2' 2)t = 1- t;
(
....., (1 1 1 1 -1 -1
c) W+ 2' 2' 2' 2)t =t
Hence, by means of (12.2.3), the partition function is
Example 12.2.2. If {3 = O ~ then the number of contributing states
is 5, as shown in Figure 12.2.4.
~
~
-t
~
- ~
2 2
~ .1.
~
1
-t -t -t
~
- ~
~ 2 2
-2
2 2
.l..
- ~
1
- ~ ~ 2
-T
Figure 12.2.4
In a similar way as in the previous example, the partition function is
Z{3 =
Exercise 12.2.1. Calculate the partition function for {31 = O ~ and
{3
-1-1
2 = 0'10'2 0'10'2
To find a "new" knot invariant, the first stage has been achieved,
and we have a viable candidate in the partition function. However, to
259 Knots via Statistical Mechanics
show that this leads to a knot invariant, we need
"the partition function to be equal for M-equivalent braids."
Without involving ourselves in unnecessarily messy definitions, the
easiest approach is to associate a braid , via the Boltzmann weights
(12.1.8), with some matrix. In the next section we shall show that then
Zf3 may be thought of as the trace of the matrix, and this trace is
invariant under the Markov move MI. Then by Kauffman's principle,
if we multiply Zf3 by a suitable factor, we shall have the "new" knot
invariant. At this juncture we shall just introduce this knot invariant.
Theorem 12.2.1.
Suppose K is an (oriented) knot (or link) formed from a braid {3,
and that Zf3 is the partition function for (3. Then,
(12.2.4)
is an invariant of K, where I w(f3) is the Tait number of the regular
diagram D of the closure of {3.
For a closed braid the Tait number is very easy to calculate. Sup-
pose f3 = (ej = 1), then its Tait number is just the sum
of its exponents, Le., w{(3) = el +C2 +... +em.
If we set
15 (t) = PK(t)
K 1+t'
then this is equivalent to the Jones polynomial of K. (When K is the
trivial knot, then PK(t) =1 +t [cf. Example 12.2.1 and (12.2.4)]; for
this reason, we normalize PK(t) by the factor 1 +t. In essence, there
is no difference between 15
K
(t) and PK(t). )
Exercise 12.2.2. Calculate the partition function Z{3 for the cases of
(3I = 0"10'2 and /32 = 0"10";1. Compare these to the partition function
in Example 12.2.1. Also, determine PK(t) for these two braids.
Exercise 12.2.3. Prove that PK(t) with regard to the skein diagrams
D+, D- and Do satisfies
- tPD_(t) = - v't) PDo(t),
and further show that if K is a p,-component link, then
PK(t) = (-1)JJ-
1
VK(t).
(12.2.5)
Chapter 12
3 An invariant of knots
260
So far in this chapter we have concerned ourselves with the 6-vertex
model, but there are infinitely many exactly solvable models. Using the
Boltzmann weights from the various exactly solvable models, Wadati
and his co-workers were able to discover an (infinite) series of invari-
ants, which may be said to be a hierarchical extension of the Jones
polynomial. In this section, we shall show in a systematic fashion how
these skein invariants may be constructed.
Let us suppose, in what follows, that N ;::: 2 is a positive inte-
ger and R is an N
2
x N
2
invertible matrix. We may denote R as
R = IIR(i,j I k, l)lI, where (i,j), (k, l) are chosen from the N
2
sets
of pairs (1,1), (1,2), ... , (1, N), (2,1), ... , (N, N) and (i,j) signi-
fies the appropriate row of Rand (k, l) the appropriate column of R.
Suppose also the element R(i, j I k, l) of R is an element of some ring Q.
The ring Qcan be arbitrary, with the sole proviso that it is a ring that
contains t and t! (= 0). The set of Laurent polynomials in 0 with
rational coefficients is a typical example of Q. This matrix R may now
be used to form (r -1) N
2
x N
2
matrices Rt(r) (i =1,2, ... , r -1).
This is done as follows. Suppose r ~ 2 ) is a positive integer, then
with respect to i = 1,2, ... , r - 1, let
R
i
= II R II
~ ~
(i-I) terms (r-i-l) terms
(12.3.1)
where I is the N x N identity matrix. The tensor product A B of
two matrices A and B is defined as follows. Suppose Ais a p x p matrix
of the form
A=
and similarly let B be a q x q matrix. Then AB is a pq x pq matrix
of the form
a11B a12
B
a21
B
a22
B
AB="
261
Example 12.3.1 Suppose
then
Knots via Statistical Mechanics
We shall index the rows (and columns) of R by the r-tuple pos-
itive integers (aI, a2, ... , ar), 1 aI, a2, ... , a
r
N, arranged in
dictionary order. So the index ordering is as follows: (1, 1, ... , 1),
(1,1, ... ,2), ... , (1, 1, ... , N), (1, 1,'... ,2, 1), ... , (1, 1, ... ,2, N), with
the sequence continuing in this manner until the final indexing term
(N, N, ... ,N) is reached.
Definition 12.3.1. For every i = 1,2, ... , r - 2, if Ri(r) satisfies
the following condition
(12.3.2)
then {RI(r), R
2
(r), ... , Rr-I(r)} are called Yang-Baxter opemtors.
These matrices Ri(r), by definition, satisfy the following condition:
=Rj(r)Ri(r) if Ii - jl 2. (12.3.3)
Therefore, due to (12.3.2) and (12.3.3), we have the correspondence
which associates an element of the braid group, B
r
, to some matrix.
Let us now give some examples of such matrices R.
Example 12.3.2. Suppose N = 2, then
R= IIR(i,j I k,l)1I =
[Note: This matrix is the same as the one in (12.1.8).]
Chapter 12 262
Example 12.3.3. It is possible to generalize the R-matrix of the pre-
vious example. Suppose 1 a, b, c, d N, where N ;::: 2, then
(1) if a +b# c +d, R(a, b Ic, d) = 0;
(2) Suppose m = a +b = c +d,
(i) if a - d = c - b < 0, then R(a, b I c, d) = 0;
(ii) if a - d = c - b ;::: 0 then
R(a,b I c,d) =(_I)a+ct-!(ab+cd+N(k-m+2)-(k+m
.1
[
(t:a-l)(t:N-d) (t:C-1)(t:N-b)]2
x (t: k)(t : d - l)(t : N - a) (t : k)(t : b- l)(t : N - c) ,
where k =a-d = c-b, and the term (t : n) = (l-t)(I-t
2
) . (l-t
n
),
if n is a positive integer, and equal to 1 if n =o.
The above Boltzmann weights are basically the same as those ob-
tained by Wadati and his co-workers from the Boltzmann weights of the
N-vertex model.
Example 12.3.4. Suppose 1 i,j, k, 1 N, then
(1) if i = j = k = l,
(2) if i =1=I k =j,
(3) if i = k < j = l,
(4) in all other cases,
R(i,j I k,l) = -t;
R(i,j I k, l) = 1;
R(i,j I k, l) =t-
1
- t;
R(i,j I k,l) = O.
Therefore, R(i, j I k, l) # 0 only if the condition {i, j} = {k, l}
holds.
Example 12.3.5. Suppose N = 2, then we may set
[
1 0 0 0 ]
R= IIR(i,j I k,l)1I = -Jt
o 0 0 -t
All four of the above examples are Yang-Baxter operators, and
hence from them we may define invariants of knots (and links). In
particular, in the final example (Example 12.3.5) the question of how we
define a non-trivial invariant is of immediate interest since this invariant
is zero for all knots and links (see Exercise 12.3.6).
So, how exactly do we define an invariant of knots (or links) from
a given set of Boltzmann weights? First of all, we must describe the
263 Knots via Statistical Mechanics
partition function in terms of the Yang-Baxter operators, and then, as
in the previous section, we shall need "some sort of modification" of
the partition function. The necessary "modification" corresponds to a
N x N-diagonal matrix J.t:
J.tl
o J.LN-l
IJN
where J.Li is a non-zero element of Q.
Definition 12.3.2. Suppose a and b are non-zero invertible elements
of Q. Then if the set {R,J.L, a, b} satisfies the conditions in (12.3.4), it
is called an enhanced Yang-Baxter operator (or matrix).
(1) For 1 ~ i, j, k, l ~ N, (J.LiJ.Lj - J.LkJ.Ll)R(i, j I k, l) = O.
N
(2) (i) ,ER(i,j I k,j)J.tj =aboiki
j=1 (12.3.4)
N
(ii) ,ER-1(i,j I k,j)J.tj = a-1boiko
j=1 I
The (homomorphic) map CPR, which, if we recall, sends a gener-
ator Ui of the r-braid group B
r
to ~ r , allows us to represent an
arbitrary element, f3 =uj:uj: ... uj: of B
r
, by an Nr x Nr matrix,
Le.,
epr({3) =Rj: (r)Rj: (r) ... Rj::: (r).
This product is "modified" (multiplied) by the Nr x N
r
matrix
given by
J.L(r) = J.L J.L &J J.L .
, ~
".
r times
The final operation required to define a knot (or link) invariant is
to take the trace, Le., tr(CPr ({3)J.L(r).
Theorem 12.3.1.
Suppose K is a (oriented) knot (or link) that is represented by the
r-braid {3, i.e., K is the closure of {3. Then if a (# 0) and b (# 0) are
Chapter 12 264
the elements of Q defined above (Definition 12.3.2), then the following
is an invariant of K:
where w({3) is the Tait number of (3 .
Let us denote J{3 by JK. Also, if Jo = b-1tr(J.t) is not zero,
i.e., JK of the trivial knot 0 is not zero, we can normalize JK in the
following way:
Exercise 12.3.1. Show that if a rv {3 (Definition 10.3.2), then J
a
=
M
J/3' and hence prove JK is an invariant of K.
Therefore, to find a knot invariant by the above method, the im-
portant fact is to find an N2 x N
2
matrix R that is a Yang-Baxter
operator. We have already found such Yang-Baxter operators in Exam-
ples 12.3.2 rv 12.3.5. The question now is, To what type, if any, of the
previous knot (skein) invariants are they related to?
Example 12.3.2 (continued). If we set
then {R, J.L, a, b} is an enhanced Yang-Baxter operator. It easily follows
thatifi+j#k+l, then R(i,jlk,l)=O; andifi+j=k+l, then
J.LiJ.tj - J.tkJ-Ll = o. So condition (1) of (12.3.4) is satisfied. We can
calculate directly the appropriate equations of condition (2) in (12.3.4),
R(I,l 11, 1)J.tl + R(l, 2 11, 2)J.t2 = 1 + 0 = 1 = ab
R(2, 1 I 2, l)J.Ll + R(2, 2 I 2, 2)J.L2 = (1 - t) + t = 1 = ab.
The calculation for (ii) of this condition is completely analogous to
the above. In fact, the invariant JK that is derived from this enhanced
Yang-Baxter operator is nothing other than the Jones polynomial.
Exercise 12.3.2. Show J
K
= PK(t), where JK is as in Exam-
ple 12.3.2 and PK(t) is as in Exercise 12.2.3.
265 Knots via Statistical Mechanics
Exercise 12.3.3. Use the Yang-Baxter operator in Example 12.3.2 to
calculate J{3, by means of the trace, for f3 =
Example 12.3.3 (continued). If we set J.Li = t
i
-
1
(i = 1,2, ... , N)
N-l N-l
and a = t--
2
-, b = t-
2
-, then {R,JI"a,b} is an enhanced Yang-
Baxter operator, and for each N = 2,3, ... , we obtain a knot invariant
of the form,
"(N) _ (t N;l )w({3)-r+ltr(<prCB)JI,(r
J
K
- 1+t++t
N
-
1
'
where K is a knot (or link) that has been represented by the r-braid
(3. Moreover, = t_
N
;l (1 + t + ... + tN-I). We leave it as a
straightforward exercise for the reader to show that if N = 2, then this
knot invariant is the same as the Jones polynomial, so it is appropriate
to call jW) the Nth degree Jones polynomial.
As an example of one of the polynomials that can be calculated,
let N =3 and K be the right-hand trefoil knot, then
= t
2
+t
5
_ t
7
+t
8
_ t
9
_ t
lO
+tIl.
Exercise 12.3.4. For the Boltzmann weights in Example 12.3.3, with
N = 3, determine the 9x9 "matrix R. Using this R-matrix calculate the
3
rd
degree Jones polynomial of the (oriented) Hopflink, Figure 4.3.2{c).
Example 12.3.4 (continued). If we set J.Li = t
2i
-
N
-
I
and a =
-tN, b = 1, then {R, J.t, a, b} is an enhanced Yang-Baxter operator.
In this case, JK of the trivial knot is
t
N
- t-
N
Jo = t _ t-1 '
and
J{3 = (-t
N
)-w(J3)tr(cpr(,B)p,(r)
is an invariant of a knot (or link) K that has been represented by the
r-braid {3.
By considering the R-matrices, it may be shown that JK = t
satisfies the following skein relation:

t JD+ - t- JD_ = (t - t- )JDo'
Since t and N are independent of each other, we may think of t
N
as a distinct variable, then from the infinite series J
2
, J3' ... , we can
recover the skein polynomial PL(V, z).
Chapter 12 266
Exercise 12.3.5. Using the enhanced Yang-Baxter operator of Ex- I
ample 12.3.4, calculate J(3 for N = 3 and f3 =
Example 12.3.5 (continued). If we set
1 1 [1 0]
a =t 2", b = t - '2 and p, = 0 -1 '
then {R, p" a, b} is an enhanced Yang-Baxter operator. However, for
this enhanced Yang-Baxter operator, irrespective of the braid (3 cho-
sen, J{3 = o. Also, since Jo = b-1tr(J.t) = 0, J{3 = g, i.e., the "knot I
invariant" cannot be determined from the methods described above.
However, by delving a bit deeper (but not much deeper) into the theory .
this chapter is based on, it is possible to define a non-zero knot invari- 'I;
ant from this enhanced Yang-Baxter operator and then show that it is ':::1
equivalent to the Alexander polynomial.
Exercise 12.3.6. Show, using the enhanced Yang-Baxter operator of
Example 12.3.5, that for (3 = JI3 = o. \:
'I;
These knot invariants, which depend on the R-matrices that satisfy ,'::;

the Yang-Baxter equation, gave rise to a deluge of research into this area.
In this chapter we have only been able to provide an introduction into
exactly-solvable models and only the basics of the subsequent Jones-type

invariants. The interested reader may wish to refer, amongst others, to ,,'

Jones [J2] and Turaev [Tu]. ::;
It should be noted that there exist methods that allow us to cal-

culate the above invariants directly from an arbitrary regular diagram
of a knot (or link), rather than as we have done in this chapter from

a braid. Finally, it should be underlined that these invariants are very
closely related to the new invariants for closed orientable 3-manifolds

F.H.C. Crick and J.D. Watson, in one of the most remarkable in-
sights of the 20
th
century, unraveled the basic structure of DNA. For
this profundity into the substance of living matter, they were jointly
awarded the Nobel Prize for Medicine in 1962. Essentially, a molecule
of DNA may be thought of as two linear strands intertwined in the form
of a double helix with a linear axis. A molecule of DNA may also take
the form of a ring, and so it can become tangled or knotted. Further,
a piece of DNA can break temporarily. While in this broken state the
structure of tIle DNA may undergo a physical change, and finally the
Chapter 13 268
DNA will recombine. In fact, in the early 1970s it was discovered that a
single enzyme called a (DNA) topoisomerase can facilitate this complete
process, from the initial break to the recombination. The reader who
might have picked up this book, looked at the title, and then randomly
opened the book at this page may think that the publisher has some-
how inserted some pages ~ an elementary textbook on biology here by
mistake. But, let us reconsider the above. The double-helix structure of
DNA - on some occasions DNA may even have only a single strand - is
a geometrical entity, or more precisely, a topological configuration. This
topological configuration is itself a manifestation of linking or knotting.
Further, it has been shown when a topoisomerase causes DNA to change
its form that the procesS is very similar to what happens locally in the
skein diagrams.
Therefore, for the geometrical entity - knotted or linked - the link-
ing number is an important concept, while the action of the topoiso-
merase is related to the new skein invariants. In this chapter we shall
give an outline of exactly how knot theory is interpreted and used in
trying to understand the changes DNA undergoes; this is sometimes
called the topological approach to enzymology.
1 DNA and knots
Living matter, be it a person, an animal, or some type of plant,
et cetera, is composed of countless molecules. Within these countless
molecules, we find the DNA of the living matter. The nature of a living
thing and how it develops depends largely on the information it inherits
from its DNA. Technically, it is t ~ e genes that carry the information .of ,
the DNA and are passed on 'from the progenitor to its offspring.
In general, as we have already mentioned, DNA has the structure
of two linear strands intertwined along a linear axis, forming a double
helix. This, however, is the not only possible structure for DNA, and in
what follows the next description is probably more easy to comprehend
in the context of knot theory. On some occasions, it has been found
that DNA has the form of a ring consisting of either a single strand
or two strands coiled in a double helix. This single-strand DNA can
literally be knotted, i.e., the objects (knots and links) we have so far
discussed, to a degree in a staid abstract way, can actually be seen under
an electron microscope. The information the DNA molecule carries, i.e.,
the arrangement of its nucleotide base pairs, is unrelated to how it is
269 Knot Theory in Molecular Biology
knotted (or tangled). So, maybe we should dismiss the knot (or link) as
a useful tool in molecular biology, but without much significance. How-
ever, recent research has shown that the knot type (of a DNA molecule)
has an important effect on the actual function of the DNA molecule in
the cell. Therefore, using knot theory techniques, it may be possible to
bring further insight into the structure of a DNA molecule.
Let us now be bit more precise and describe a DNA molecule purely
mathematically. A mathematical model for a DNA molecule is usually
a thin, long, narrow (oriented) twisted ribbon, Figure 13.1.1. (In this
figure, the ribbon is homeomorphic to 8
1
X [-1,1], but not to the
Mobius band.)
C2
Ik(C
1
, C
2
) = 1
Wr(B) =0, Tw(B) =1
(a)
Ik(Cl'C
2
) =-1
Wr(B) =-1, Tw(B) = 0
(b)
Figure 13.1.1
c
The two curves C
l
and C
2
that form the boundaries of the rib-
bon B represent the closed DNA strands. We may fix an orientation on
the curve C that forms an axis for B (Le., the central curve 8
1
x {O} ).
This orientation on C induces similar orientations on C
1
and C
2
, on
the boundary of B. In fact, the linking number between C
1
and C
2
[lk(C
1
, C
2
)] is an invariant, and its change has a very important effect
on the structure of the DNA molecule. For example, it is known that if
we reduce the linking number of a double-strand DNA molecule, then
the effect is to cause the DNA molecule to twist and coil, Le., what is
known as 8upercoiling. The actual reduction of the linking number of
DNA molecule can be caused by a topoisomerase acting on the DNA
molecule. (In fact, the orientation on an actual DNA molecule is not
precisely as above. The orientations, in reality, on the two DNA strands
are mutually opposite. Therefore, maybe we should assign mutually op-
posite orientations to Cl and C
2
However, in the case of a link formed
from C
l
and C
2
, the linking number defined on the DNA molecule
Chapter 13 270
in biology, and the linking number calculated from the mathematical
model, as above, are equal. So, from a numerical point of view, there is
no incongruity.)
The number of twists the ribbon B has along the axis C is called the
twisting number, and is denoted by Tw(B). The writhe, Wr(B), in the
case of mathematical biology differs slightly from our previous definition,
Definition 4.5.2. For the purposes of this chapter, we shall define the
writhe as the average value of the sum of the signs of the crossing points,
averaged over all the projections. Succinctly, the writhe is determined
from the axis C by considering it as a spatial curve. These numbers,
Wr(B) and Tw(B), are invariants. They are not, however, invariants
of the knot (or link) obtained from the DNA molecule, but differential '
geometry invariants of the ribbon B as a surface in space. [If C is a
plane curve, then Wr(B) = o. So, Wr(B) may be said to calculate the
non-planarity of B. We should also note that Tw(B) and Wr(B) are not .
necessarily integers.]
The three "invariants" mentioned above are related by the following
basic formula.
lk(C
1
, C
2
) = Tw(B) +Wr(B). (13.1.1)
On occasions - when the double helix is unwound a few turns due
to a cut in one of the strands - the axis of the double helix of a DNA
molecule twists into a helix. As mentioned above, this causes the super-
coiling of the DNA, and (13.1.1) is very useful in picking up this quality. '
Although supercoiling is interesting in its own right, we shall, to avoid
having to swathe the reader in concepts from molecular biology, not
delve any deeper into this concept. If the reader would like to pursue or
become acquainted with these concepts, a good reference is Wang [Wale
Thankfully, DNA is very malleable, being able to recombine
through a series of phases; otherwise the world would be populated .
by clones. In the phases of this process, the knot type of the DNA
molecule is actually changed. At first, it might seem that to understand
this process from the point of view of molecular biology will be compli- ,
cated. However, in the early 1970s it was found the whole process, from
the original splicing to the recombination, was the result of the effect
of a single enzyme/catalyst called a topoisomerase. The term topoiso-
merase may seem rather strange, but it is relatively easy to explain.
Chemically, two molecules with the same chemical composition but dif-
ferent structure are called isomers. It follows that two DNA molecules
with the same sequence of base pairs but different linking numbers are
also isomers. Due to the difference in linking numbers, "topologically"
271 Knot Theory in Molecular Biology
they are inequivalent. So, these DNA molecules are called topoisomers.
Hence, the enzyme that causes the linking number to change is termed
an topoisomerase. The process of mutation due to a topoisomerase can
be in simple terms be described as follows: First a strand of the DNA is
cut at one place, then a segment of DNA passes through this cut, and
finally the DNA reconnects itself.
(a)

(b)
Figure 13.1.2
In Figure 13.1.2, we give two examples, of the action of a topoiso-
merase on a DNA molecule (for clarity, we have not drawn the helical
twist). The place where the strand is cut is denoted by "0." The two
figures [Figures 13.1.2{a) and (b)] are relatively self-explanatory. The
single strand, in Figure 13.1.2{a), has a single cut due to a topoiso-
merase and the DNA passes through it and recombines; this is called
a Type I topoisomerase. While, in Figure 13.1.2{b), a cut in a double-
strand DNA., due again to a topoisomerase, allows a double-strand DNA
to pass through it and recombine, this is as expected called a Type II
topoisomerase. Finding such a topoisomerase is relatively straightfor-
ward, since they occur in organisms small and large, from bacteria to
within the reader of this book.
In the next few sections we shall discuss in slightly more detail
the effect of a certain topoisomerase (to be precise, it should really
be called a recombinase). This effect is usually called a site-specific
recombination.
2 Site-specific recombination
As the name suggests, a site-specific recombination is a local op-
eration. The effect of the recombinase on a DNA molecule is to either
move a piece of this DNA molecule to another position within itself or
to import a foreign piece of a DNA molecule into it. The result is that
the gene transmutes itself. It is known, in fairly advanced organisms,
of which we are an example, that various antibodies form through such
Chapter 13 272
site-specific recombination of a DNA molecule.
The exact process of a site-specific recombination is fairly easy to
understand. Firstly, two points of the same or different DNA molecules
are drawn together, either by a recombinase or by random (thermal)
motion (or even possibly both). The recombinase then sets to work,
causing the DNA molecule to be cut open at two points on the parts
that have been drawn together. The loose ends are then recombined
by the recombinase in a different combination than the original DNA
molecule. In Figure 13.2.1(a) (c), we have shown a simple site-specific
recombination that has been carried out in the manner described above.
writhing
--+
(a) (b)
Figure 13.2.1
(c)
The above description is loosely what occurs in a site-specific re-
combination. For the reader who might want to read further and more
precisely, we shall define the relevant terms involved in this process
found in literature on this subject. The DNA molecule before the ac-
tion of the recombinase is called a substrate; after the recombination it
is called a product. The process of going from the DNA molecule to a
state in which two parts of the DNA molecule have been drawn together,
Figures 13.2.1(a) up to just before (b), is said to be the writhing process.
When at this stage the recombinase combines with the substrate, the re-
sultant combined complex is called a synaptic complex, Figure 13.2.1(b).
Within the synaptic complex, we can assign local orientations to the re-
spective, relatively small parts of the DNA molecule (or molecules) on
which the recombinase acts [within the circle in Figure 13.2.2(a), (b),
and (c)].
If the orientations on the DNA molecule and the orientation in-
duced by these local orientations agree, then this arrangement is called
a direct repeat, Figure 13.2.2{a). On the other hand, if they do not agree,
then the arrangement is said to be an inverted repeat, Figure 13.2.2(c).
273
(a)
Knot Theory in Molecular Biology
(b)
(c)
Figure 13.2.2
Exercise 13.2.1. For a site-specific recombination show the following:
(1) If the substrate is a DNA knot and the arrangement is a direct
repeat, then the product is a 2-component DNA link, Figure 13.2.2{a).
If, however, the arrangement is an inverted repeat, then the product is
a DNA knot, Figure 13.2.2{c);
(2) If the substrate is a DNA link (Le., two DNA molecules entwined),
then after recombination the product is a DNA knot, Figure 13.2.2(b).
In the next section, we shall describe a mathematical model, with
empirical constraints, for the site-specific recombination due to the ac-
tion of a recombinase.
3 A model for site-specific recombination
The following proposition follows from empirical evidence:
Proposition 13.3.1.
(1) Almost all the products obtained by the site-specific recombi-
nation of trivial knot substrates are rational knots (or links),
i.e., 2-bridge knots (or links).
Chapter 13 274
(2) The part of the synaptic complex acted on by an enzyme (re-
combinase), mathematically within the 3-ba1l, is a (2,2}-tangle, Fig-
ure 13.3.1.
+
R
Figure 13.3.1
Therefore, the product is just the replacement of one (2,2)-tangle
by another (2,2}-tangle. For example, the (2,2}-tangle within the cir-
cle T in Figure 13.3.1 is replaced by the tangle R to form the product
shown. This process may be expressed by means of our definition of
the sum of tangles (cf. Chapter 9, Section 1). The good thing about ::;
mathematics is that inside may be outside, and outside may be in-
side. Mathematically, it is perfectly reasonable to consider S to be a 1
(2,2)-tangle in T. The numerator of the sum of 8 and R is then the '
product, Figure 13.3.2.
s
+
R
Figure 13.3.2
80 the following "equation" holds.
N(S +R) =the product. (13.3.1)
Further, we may divide the substrate into the external tangle 8 and
the internal tangle E, since the substrate is then the numerator of the
sum of S and E, Figure 13.3.3.
275 Knot Theory in Molecular Biology
N
s
+
E
Figure 13.3.3
Again we have a quasi-equation holding,
N(S + E) =the substrate. (13.3.2)
If it is possible to observe the substrate and the product, then
the ideal situation would be to determine S, E, R from the two quasi-
equations (13.3.1) and (13.3.2). Mathematically, however, since there
are only two equations but three "unknowns," it is not possible without
further assumptions to determine these unknowns:
So, we need to fall back on experimental data to make some further
progress. Recently, the following has been observed:
Supposition 13.3.2.
The effect of the enzyme - the change due to this enzyme from
the tangle E to the tangle R - depends only on the original enzyme, so
this process of change is independent of the shape, position, and size of
the substrate.
For each recombination we shall obtain a quasi-equation as above.
However, by Supposition 13.3.2 no new'indeterminates are added, and
the indeterminate is always R. Therefore, even although the number
of equations increases, the number of indeterminates remains constant.
This means, mathematically, that there is a possibility that we can
solve the collective equations. Another assumption from experimental
observation is that the repetition of site-specific recombinations can be
expressed as the sum of tangles.
Supposition 13.3.3.
The product of a series of site-specific- recombinations can be ex-
pressed as the numerator of the sum of tangles, namely, it is of the form
N(S + R+ R+ +R).
Chapter 13
276 .
Exercise 13.3.1. Suppose A is a type (2,3) rational tangle, T(2,3),
and K is a 2-bridge knot of type (19,5). Determine the rational tangle
X for which the equation N(A + X) = K holds.
In the next section, under the above assumptions, we shall show
it is possible to construct a virtually solvable model for a site-specific
recombination due to the recombinase Tn3 Resolvase.
4 Recombination due to the recombinase Tn3 Resolvase
As already mentioned, Tn3 Resolvase is an enzyme (recombinase)
that is a catalyst for a site-specific recombination on a circular DNA sub-
strate with directly repeated recombination sites. When this resolvase
acts on a circular DNA substrate that is supercoiled and unknotted,
then the product is a link. In most cases, the product is the Hopf link,
Figure 13.4.1(a). (In addition, if the orientation of the DNA molecule
is taken into account, then the linking number of the recombined DNA
molecule may be considered to be -1. In the sequel, this fact will not
be of relevance, but see Theorem 13.4.2.)
(a) (b) (c) (d)
Figure 13.4.1
If the resolvase causes a further recombination, then the subse- 1
quent product is the figure 8 knot, Figure 13.4.1(b). Continuing, a fur-
ther recombination (so three recombinations have occurred) produces
the Whitehead link as the product, Figure 13.4.1(c). Up to three re-
combinations due to Tn3 resolvase have been shown experimentally to
agree with the above. By experimental observation it has also been
shown that the product of the fourth recombination is the knot in Fig-
ure 13.4.1(d). However, to find the original S, E, and R, we shall show
that this fourth recombination is not required and will only be used as
277 Knot Theory in Molecular Biology
a check for the model we shall put forward.
Assuming Suppositions 13.3.2 and 13.3.3 hold, we can draw the
series of diagrams in Figure 13.4.2.
(1)
=0
(2)
=
8
(3)
=

{(j
(4)
=
(5)
Figure 13.4.2
Since the knots and links on the right-hand side of the "equations"
in Figure 13.4.2 are 2-bridge knots or links, we may rewrite them as
mathematical formulae using the notation created in Chapter 9, Sec-
tion 3:
(1) N(S +E) = C(I)
(2) N(S +R) = C(2)
(3) N(S +R +R) = C(2, 1,1)
(4) N(S + R +R +R) = C(I, 1,1,1,1)
(5) N(S + R +R + R + R) = C(I, 1, 1,2, 1).
(13.4.1)
Chapter 13 278
We can, by looking carefully at these "equations," determine the
tangles Rand S. (It is not, however, possible to determine E from these
equations; see Theorem 13.4.2). 80, finally in this chapter we are ready
to apply some mathematics.
Theorem 13.4.1.
(1) The possibilities for the tangles 8 and R that satisfy
(13.4.1)(1) rv (3) are limited to the four tangles in Fig-
ure 13.4.3.
(2) In addition, if (13.4.1)(4) holds, then the only possibility for
Rand S is as in Figure 13.4.3(a).
(a) (b)
,....
.\a

(c)
Figure 13.4.31

Therefore, the effect of the first recombination due to the recombi- .j
nase Tn3 may be thought of as that in Figure 13.4.4.'

Figure 13.4.4
Since we cannot as yet determine E, the above recombination is to
a certain degree not precise; however, if we assume E is the (0) tangle,
279 Knot Theory in Molecular Biology
the above results are known to hold.
Proof
The first step is for the reader to make sure, by drawing the
relevant diagrams, the above tangles (Figure 13.4.3) are solutions to
(13.4.1)(1) rv (3). We shall give an outline of a proof showing these are
the only possible solutions. (For a detailed proof refer to Ernst and
Sumners [ES2].)
We can prove both Rand S are rational tangles by calling on The-
orem 9.3.1 and Proposition 9.3.4, in which we proved that a 2-bridge
knot (or link) can be represented as a denominator or numerator of a
rational tangle. (The proof itself is not very straightforward; so in order
not to make this proof too dense, we shall omit the details of this part.
In molecular biology, as a first assumption R and S are taken to be
rational tangles.)
Next. let us determine from equations (13.4.1)(2) and (3) the ra-
tional tangles Sand R. By Theorem 9.2.2, we know Sand R may be
represented, respectively, by the fractions t and ~ (recall, 00 = t).
Exercise 13.4.1. It is known that if both i and ~ are not integers,
then N(S +R +R) is not a rational knot (or link). Confirm this is the
case for ~ = i and ~ = - ~
So, if both R and S correspond, respectively, to integers [i.e.,
b = d = 1), N(S + R + R) = N(T(a + 2c)], which is a torus knot
(or link) of type (a +2c, 2). This implies that the resultant knot cannot
be the figure 8 knot. (This may be shown by comparing the Alexan-
der polynomials.) Therefore, only one of Sand R may correspond to
an integer.
So, now, suppose S is a (0, D)-tangle i . ~ e . b = 0), then (see also
Figure 13.4.5)
N(S +R +R) = N(T(O, 0) +R +R) = D(R +R).
Figure 13.4.5
Chapter 13 280
This also cannot be the figure 8 knot (why?). Moreover, if R is a
(O,O)-tangle, then N(S + R + R) is at the very least a 2-component
link, and thus obviously not a knot, see Figure 13.4.6.
Figure 13.4.6
Therefore, neither R nor S may be a (0, D)-tangle.
So, let us assume R is an integer tangle, T( r). If r = 0, then
N(S + R) = N(S), and similarly N(S + R + R) = N(S). Therefore,
N(S + R) = N(S + R + R), which directly contradicts (13.4.1)(2) and
(3). The consequence of this is that r cannot be equal to zero. Hence,
if we suppose R = T(r) (r '# 0), then S must correspond to a rational '
tangle ~ It is now possible to determine r, u, v by making use of
Theorem 9.3.5.
Firstly, from N(S+R) = C(2), the absolute value of the numerator
of
u rv+u
r+-=--
v v
is equal to the determinant of C(2), Le.,
lrv +ul = 2.
(13.4.2) ~
'1
Similarly, from N(S +R +R) = C(2, 1, 1), the absolute value of ~
the numerator of
u 2 u + 2rv
-+ r=---
v v
is equal to the determinant of C(2,1,1). So, in this case it follows that
lu+ 2rvl = 5.
(13.4.3)
Exercise 13.4.2. Show the possible solutions for r, u, v from the
system of equations (13.4.2) and (13.4.3) are
{(u,rv)} = {(-1,3), (1,-3), (9,-7), (-9,7)}.
281 Knot Theory in Molecular Biology
Let us look at the first set in these solutions, u = -1 and TV = 3.
Since T and V are integers, we have the following possible solutions:
= {( -1), (-1,3), (1,-3)}.
Since ; is not an integer, we may remove from our considerations
the final two solutions. Therefore, in the case ; = -1, the corre-
sponding tangle is T(-3,0), and this is the tangle in Figure 13.4.3(a)
(cf. Chapter 9, Section 2), while in the case; = 1, the corresponding
tangle is T(3,0), and this is the tangle in Figure 13.4.3(b).
In a similar way, we may investigate the three other possibilities.
For the case (u, TV) = (9, -7),
= {(-;,1), (;, -1), (-9,7), (9,-7)}.
As above, we may throwaway the final two solutions and concentrate
our attention on the first two possibilities. Since
9 1
7=1+
3
+
1
,
2
the corresponding tangle is T(2,3,1), and with due consideration of the
minus signs, these are the two tangles in Figure 13.4.3(c) and (d).
Finally, if S is an integer tangle, then we may show S does not
satisfy (13.4.1)(2) and (3). The process is almost the same as above,
but a touch more complicated. For example, suppose S = T(s), where
s is an integer.
We may now suppose the rational number corresponding to R is ;
(v > 1). As above, if we again use Theorem 9.3.5, we shall obtain the
following formula:
Jvs+ul = 2.
Exercise 13.4.3. Show that if S is an integer tangle, then
N(S +R +R) = N(R + (R +S)).
(13.4.4)
Since R +S is a rational tangle, from Theorem 9.3.5 the determi-
nant of N(R + S) = N(S + R) is lu + vsl. Similarly, the determinant
of N(S + (R + R)) is the absolute value of the numerator of
u u 2uv + sv
2
-+-+s= 2 '
V V V
Chapter 13
which can be shown to be 12uv +sv
2
1 =5. Hence, v = 5 and
12u+svl = 1.
282
(13.4.5)
Exercise 13.4.4. Confirm the following are solutions to (13.4.4) and
(13.4.5): v =5 and
Since the first two do not give integer solutions for s, we may
ignore them. The final two, on the other hand, do not satisfy (13.4.1)(3)
(show this by drawing the diagrams). Hence, we have proven the first
of Theorem 13.4.1.
Exercise 13.4.5. Show, by drawing the relevant diagrams, the second
part of Theorem 13.4.1.
To show, with the above tangles, that we are along the right lines
for a correct model of the effect of the recombinase Tn3, we can confirm
by drawing the relevant diagram that [13.4.1(5)] also holds.
Hence, from the above it should be possible to predict the result of
further recombinations (in reality, this is possible). o ~
By means of Theorem 13.4.1, we can determine {S, R}; however, ..{
the theorem does not shed any light on what E may be. It is not ~ ~
possible with certainty to be able to understand the structure of E. ~
However, to a certain degree some of the structure can be seen from the ~
.:!,,'
following theorem:
Theorem 13.4.2. ~
Suppose there is a tangle E that satisfies the following two condi- Of:
tions:
(1) N(T(-3,O) +E) =C(I)
(2) N(T(-3,0) +T(l = 0(2)
(13.4.6)
with linking number -1.
If, in addition, E is a rational tangle, then E is a tangle of the form
T(2x, 3,0), where x is an arbitrary integer.
Exercise 13.4.6. Confirm, by drawing the diagrams that for arbitrary
n, E = T(n, 3, 0) satisfies (13.4.6)(1).
283 Knot Theory in Molecular Biology
If we add the second condition (13.4.6)(2), then T(-3,0) has an
orientation induced on it. With regard to this orientation, if the site-
specific recombination by the Tn3 resolvase is a direct repeat, then E
must be of the form T(2x, 3, 0).
In this chapter, we have seen via knot theoretical techniques that
we may shed some light and elucidate the models for the recombination
of a DNA molecule. At present, the extent knot theory may further help
in the understanding of the mechanism of recombination of the DNA
molecule is not clear. This area of research is still in its inchoate stages.
Hopefully in the future this interaction will lead to some interesting,
maybe coruscating, results.
In our discussions thus far we have considered a graph to be a figure,
to put it naively, composed of dots and line segments (topologically this
is called a I-complex). To be more exact, less intuitive, and more math-
ematical, a graph is usually thought of in an abstract sense. Therefore,
strictly speaking, a (finite) graph G is a pair of (finite) sets {VG, EG}
that fulfills an incidence relation. An element of VG is then said to
be a vertex of G, while an element of EG is said to be an edge of G.
The relation/condition mentioned above stipulates that an element, e,
of EG is incident to elements, say, a and b, of VG (nota bene, tIle
285 Graph Theory Applied to Chemistry
condition does not require a and b to be distinct.) The two vertices a
and b are said to be endpoints of e. If it is the case that a = b, then e
is said to be a loop.
If there exist between two graphs, G = {V0, Eo} and G' =
{Va, Eo}, 1-1 correspondences fv : VG Va and fE : EG Eo
that satisfy condition (14.0.1) given below, then G and G' are said to
be isomorphic.
If a vertex, a, of G is an endpoint of an edge, e, of G,
then fv(a) is also an endpoint of fE(e). Conversely,
if a vertex a' of G' is an endpoint of an edge, e', (14.0.1)
of G', then fy
1
(a') is also an endpoint of f
E
1
(e
/
).
Graph theory often relies on - as a way of illustrating its con-
cepts and research results - models of graphs in space, Le., the before-
mentioned I-complexes. Frequently, the model of a graph and the ab-
stract graph itself are perceived to be one and the same. However,
the model is a mere tool for presentation and expository purposes, and
should not really be confused with the abstract graph.
In a sense it is quite easy to come to the conclusion that an abstract
graph is some sort of generalization of a knot (or link); however, if we
are to be precise in our definition of a graph, this strictly is not the case.
For if we consider problems concerning graphs, the obvious problem, the
shape of the graph in space, is not a problem in the theory of (abstract)
graphs. This problem is dealt with separately in the theory of spatial
graphs, which may be thought of as a generalization of knots (or links);
we shall consider this in more detail in Section 2. For example, in
Figure 14.0.1 we have shown two models of the same graph; however,
as spatial graphs they are distinct.
(a) (b)
Figure 14.0.1
In this chapter we will firstly define an abstract graph invariant,
from whicll we may derive a spatial graph invariant. Since a spatial
graph may be thought of as a generalization of a knot (or link), then
Chapter 14 286
this invariant is a generalization of a knot invariant. Having established
this graph invariant, in the final section we shall look at the chiral
properties of spatial graphs, which are of interest to chemists since it is
possible to relate spatial graphs to the structure of molecules.
1 An invariant of graphs: the chromatic polynomial
In the same way as for invariants of knots, a quantity that has
the same value for two isomorphic graphs is said to be an invariant of
graphs. Hence, no matter with what type of models we represent this
graph, the value on the model will be invariant.
For example, the number of vertices, Le., the number of edges of
a graph, are the most obvious invariants. Besides such numerical in- ,
variants, it is possible also to construct polynomial invariants. As a
typical example we shall instead take something called the chromatic
polynomial.
Let G be a graph. We shall colour the vertices of the graph using
a palette of n colours. The way to apply these colours is to paint
two adjacent vertices (the vertices are the endpoints of an edge) with
distinct colours. The above process is said to be a (vertex) colouring of
G. Although we have on our palette n colours, it is not necessary to
use all of them in the colouring process, Figure 14.1.1(a).
yellow.... -. blue
(a) (b)
Figure 14.1.1
Let us denote by Pn(G) the number of possible colourings of G with
(at most) n colours, and let Pn(G) = ~ P n G . If we fix the colour
of a vertex, Pn (G) is the number of ways we can colour the other
vertices using n colours. Hence, this number can never be negative.
From Pn (G) for various n we can define, in the following manner, a
polynomial (to be precise a power series) PG(t) :
PG(t) = Pl(G)t + P2(G)t
2
+... +Pn{G)t
n
+... (14.1.1)
287 Graph Theory Applied to Chemistry
The power series PG(t) is called the chromatic polynomial of G.
If G has a loop e [Figure 14.1.1(b)], then Pn(G) becomes zero. In
this case, since the end point P of the loop is self-adjacent, we cannot
assign a colour to P. Let us calculate Pn(G) for several examples.
Example 14.1.1. Suppose G is a graph consisting of only m vertices,
Le., G has no edges, Figure 14.1.2(a). Then each vertex can be coloured
totally independently of the rest; hence, Pn(G) =nm. Therefore,
00
PG(t) = L: nm-1t
n
.
n=l

m
....

<a>
'Vi '2 Vm
~ t ~ - - - I ~
-. ~ em-l
(b)
Figure 14.1.2
Example 14.1.2. Suppose G is a graph consisting of m vertices,
{VI, V2, ... , v
m
}, and m-l edges, {el' e2, ... , em-I}, such that the
endpoints of ei are Vi and Vi+l (i = 1,2, ... ,m-l), Figure 14.1.2{b).
We may colour the vertex Vt with n colours. The next vertex, V2,
since it cannot be coloured with the same colour as VI, can be coloured
with n - 1 possible colours. Similarly, it is possible to colour Va with
n - 1 possible colours, and so on. This leads to
Pn(G) =n(n - l)m-l.
Therefore,
00
PG(t) = L:(n - l)m-
1
t
n
.
n=l
Exercise 14.1.1. Calculate PG(t) when G is a polygon with m
edges.
Let us denote by G
e
the graph obtained from G by removing a
single edge e, Figure 14.1.3(a). Similarly, let us denote by G/e the
graph obtained from G by contracting e so that its two endpoints amal-
gamate, Figure 14.1.3(b). This latter process is called the contraction
(with respect to e) of G. If e is a loop then G
e
= G/e.
Chapter 14
I
I

'. '
'. ' , '
(a)
G
,
"
Gle
(b)
288
Figure 14.1.3
From our definition of a graph, we do not allow a graph with mul-
tiple edges, for example, as in Figure 14.1.4(a).
,
, ,
, ,
, ,
,
, , ,
>-<
,
--+
, , , ,
,
, ,
, ,
, ,
,
Figure 14.1.4
However, the contraction may produce multiple edges in G/e.
Then by removing all but one of the edges, we can make GI e con-
form to our definition (see also Example 14.1.3).
Exercise 14.1.2. Show that the three graphs G, G
e
, and G/e are
related by the following equation:
(14.1.2)
Since Pn(G
e
) and Pn(G/e) have at least one edge less than G, we
can determine Pn(G) by using (14.1.2) and mathematical induction.
If G is a graph without any edges, then this is the graph described
in Example 14.1.1, and so Pn(G) = n
m
-
1
This leads us to the next
theorem.
Theorem 14.1.1.
The colouring number, Pn(G), of a graph G coloured with (at
most) n colours can be calculated by means of the following two for-
mulae.
(1) lfG consists of m vertices, then
(14.1.3)
289 Graph Theory Applied to Chemistry
(2) If e is an edge of G, then
Pn(G) = Pn(G
e
) - Pn(G/e).
Example 14.1.3.
Pn (A) = Pn (L) - Pn (0) = Pn (.-.-.) - Pn ( .....)
== (n - 1)2 - (n - 1) = (n - l){n - 2).
A graph is said to be a complete graph if it has no loops and two dis-
tinct vertices are always the endpoints of only one edge. If K
m
denotes
a complete graph with m vertices, then K
m
has m ~ - l edges. In
Figure 14.1.5 we have drawn models for complete graphs with m ~ 6.
(a) (b) (c)
(d) (e)
Figure 14.1.5
Exercise 14.1.3. Find a formula to calculate Pn{K
m
).
2 Bing's conjecture and spatial graphs
We would like once again to underline the fact, mentioned in the
previous section, that there is a clear distinction between a spatial graph
Chapter 14 290
as a model of an abstract graph and the abstract graph itself. Again, as
noted previously and importantly, a spatial graph may be considered to
be a generalization of a knot (or link). Along this vein, it is possible to
open a seam for a theory of spatial graphs and then mine this seam. As
a first step in an attempt to extract some information from this seam,
we shall consider a conjecture due to R.H. Bing, whose main interest
during his lifetime, however, was to find the right seam to quarry for a
solution for the Poincare conjecture.
Bing's conjecture.
Suppose that K
m
is a complete graph with m vertices. If m 7,
then regardless of the spatial graph we use as a model for K
m
, we
can find within these spatial graphs a partial graph that represents a
non-trivial knot.
This conjecture was shown to be true, using knot theoretical tech-
niques, by J. Conway and C.McA Gordon [CG].
Exercise 14.2.1. Show that this conjecture does not hold for the com-
plete graphs in Figure 14.1.5(a) rv (d). But find a non-trivial knot in
Figure 14.2.1(e).
For m = 7, Figure 14.2.1{a) is a model for K
7
The cycle 13642571
forms a partial graph, which can easily be seen to be the trefoil knot,
Figure 14.2.1{b).
(a) (b)
Figure 14.2.1
Exercise 14.2.2. By drawing other spatial graph models for K7, find
other partial graphs that are not trivial knots.
291 Graph Theory Applied to Chemistry
From the above, spatial graph theory may be thought of as an
extension of knot theory, rather than being a part of graph theory. So
the research into spatial graphs may be tailored accordingly, Le., with
recourse to knot theory. However, if we are to consider spatial graphs as
generalizations of knots (and links), we must include in our ruminations
graphs of the type shown in Figure 14.2.2, Le., ones that have closed
curves with no vertices.
Figure 14.2.2
However, since a graph without vertices is an oxymoron, this type
of spatial graph is not a model (in space) of a graph. Rather, it is better
to think of a spatial graph as the underlying space of a I-dimensional
finite complex in R
3
. A vertex in this context is a point at which at
least three arcs emerge. Also, it is possible that a spatial graph may
have multiple edges. In this respect, we shall define equivalence of two
spatial graphs.
Definition 14.2.1. If for two spatial graphs G
1
and G
2
there exists
an auto-homeomorphism, C{J, which preserves the orientation of R3
such that cp(Gl) = G
2
, then G
1
and G
2
are said to be equivalent (or
equal).
In the past, research into spatial graphs has been closely connected
with the theory of surfaces embedded in R
3
One such typical example
is the study of the Kinoshita 9-curve. This graph is a spatial graph
with two vertices, A and B, and three curves connected to them, and
these curves do not mutually intersect each other..
Significantly, the two spatial graphs in Figure 14.2.3(a) and (b),
with the vertices fixed in space, cannot be continuously deformed into
each other. Therefore, as spatial gmphs they are not "equivalent" (see
also Definition 14.2.1).
Chapter 14
Ae
B
(a) (b)
292
Figure 14.2.3
Graphs like the 8-curve, which have at each vertex exactly three
incident edges, are called 3-regular graphs. The Jones polynomial, stud-
ied in Chapter 11, may be generalized so that it becomes an invariant of
these graphs. However, we shall first take a quick look at a special set of
spatial graphs, namely, plane graphs, which we previously encountered
in Chapter 2, Section 3.
If we can place a spatial graph G on 8
2
, then G is said to be a
planar graph. The graph that lies on 8
2
is said to be a ]!lane graph.
With regard to a plane graph H, we can construct a graph H, called the
dual graph, by means of the following procedure: Firstly, let us divide,
by means of H, 8
2
into a finite number of regions, R
1
, R
2
, ... , R
m
,
and from (within) each region R
i
select a point Vi. These Vi will now
become the vertices of H. The of if are polygonal arcs e that
are connected to the vertices of H as described below.
If (and only if) two regions R
i
and R
j
have an edge,
e, of H in common, then join Vi and Vj on 8
2
by
a simple polygonal line that intersects e at only one
point. [In Figure 14.2.4{a) we have depicted these
polygonal lines by dotted lines.]
(14.2.1)
.,..,. ...... - R
, #'OJ 'II
, J
: e .,..,."'.
.,..,.", :
__.... :--
R.
. .

\ R
k
(a)
H
Figure 14.2.4
(b)
293 Graph Theory Applied to Chemistry
- -
In Figure 14.2.4(b), we have drawn the dual graph H of H. H is
uniquely determined from the plane graph H, which itself is a represen-
tation of the planar graph G; however, it is not uniquely determined
from G.
Exercise 14.2.3. The two plane graphs HI and H
2
in Figure 14.2.5
of a planar graph G. Draw the respective dual graphs
HI and H
2

Figure 14.2.5
Exercise 14.2.4. Suppose G is a planar graph and HI and H
2
are
plane graph representations of G. Show that if HI and H
2
are their
respective dual graphs, then for all n
Further, show that this equality holds for the respective HI and H
2
in
Exercise 14.2.3.
Now, as in the case of knots (or links), if we project the spatial
graph G onto the plane, we can obtain a regular diagram, D, of G,
Figure 14.2.6.
(a)
Figure 14.2.6
D
(b)
Chapter 14 294
However, a vertex of G should not translate via the projection to
a crossing point of D.
Suppose m is the number of crossing points ofD. We obtain a three
plane graph by replacing a crossing point (not a vertex) of the regular
diagram D by one of the three local diagrams shown in Figure 14.2.7.
x
) (
+
x
o
Figure 14.2.7
Running through all the possible choices at each crossing point, we
can draw 3
m
plane graphs. At the places where originally there was
a crossing point assign, one of +, -, or 0 according to Figure 14.2.7.
Each plane graph with +, -, or 0 assigned as above is called a state
of D. [In Figure 14.2.8 we have shown the three possible states from the
above operation on the regular diagram in Figure 14.2.6(b).]
8
2
Figure 14.2.8
To each state, s, we can assign a monomial denoted by < s >,
< s >= (-l)lt
k
,
where l is the number of 0 assigned to sand k is
(the number of + in s) (the number of - in s).
Example 14.2.1. With regard to the three states in Figure 14.2.8,
1 t t
-l
< SI >= -, < S2 >= < Sa >=
For a plane graph with a state s, we can construct, as described
above, its dual graph, which we shall denote by s. In fact, s is a
295 Graph Theory Applied to Chemistry
genuine plane graph (without multiple edges). Then in conjunction with
the colouring number, Pn(s), with n colours, we may define another
polynomial, denoted by < D >, for the regular diagram D of G:
< D >= L < S > Pn (8) I
s
where the sum on the right-hand side is taken over all the states s
(3
m
in total) on D. Performing the calculation, it is easy to see that
< D > is a polynomial in t and n; so if we replace n by t + 2 +t-
1
,
< D > is transformed into a polynomial only in t, which we shall call
the polynomial for D.
Theorem 14.2.1 [Y].
The polynomial for D, up to multiplication by a sign and t to
some exponent, is an invariant of the 3-regular spatial graphs G. In
other words, if D
1
and D2 are, respectively, the regular diagrams of
two equivalent 3-regular spatial graphs G
1
and G
2
, tben for some
integer p,
Therefore, the common value of the polynomial for G, ignoring the
exponent of t and its sign, is an invariant of G and denoted by AG(t).
[This invariant is sometimes or also called the Yamada polynomial.]
Figure 14.2.9
Example 14.2.1 (continued).
AG(t) =< 81 > Pn(Sl)+ < S2 > Pn(S2)+ < sa > Pn(Sa)
= (-l)Pn(Sl) + tpn(S2) +t-
1
Pn(S3),
where 81 is the dual graph for the state 81, Figure 14.2.9.
Chapter 14 296
Exercise 14.2.5. Using Example 14.2.1, show the following formulae
for Pn(Si) hold, and hence determine Aa(t) :
Pn(Sl) = (n - 1)(n - 2)(n
3
- 8n
2
+23n - 23)
Pn(S2) = (n - l)(n - 2)(n - 3)2
Pn(sa) = (n - 1)(n - 2)3.
In the special case when G is a knot (or link), the invariant of G,
Aa(t), agrees, up to a factor t
P
, with the 3
rd
degree Jones polyno-
mial, J ~ (cf. Example 12.3.3). Therefore, AG(t) may be considered
as a generalization of the Jones polynomial.
Exercise 14.2.6. Show that if G* is the mirror image of the spatial
graph G, then
Exercise 14.2.7. Calculate the polynomial of the two 8-curves in
Figure 14.2.3, and show they are not equivalent.
Exercise 14.2.8. Calculate AK(t) for the right-hand trefoil knot and
compare it with its 3
rd
degree Jones polynomial, J ~
3 The chirality of spatial graphs
As for knots, we may define a concept of amphicheirality for spatial
graphs. However, in the case of spatial graphs, since we shall discuss in
this section their connection to concepts in chemistry, we shall use the
terms more commonly found in chemistry, namely, chiral, achiral.
Definition 14.3.1. If there exists an orientation preserving auto-
homeomorphism of R
3
that transforms the spatial graph G in R3 to its
mirror image G*, then G is said to be topologically achiral. Otherwise,
G is chiral. We can make this condition stronger, and say G is rigidly
achiral, if we can rotate G, about some axis, to its mirror image G*.
It is an immediate consequence that if G is rigidly achiral then it
is also topologically achiral. However, the converse is not true.
Exercise 14.3.1. Show that a plane graph is rigidly achiral.
297 Graph Theory Applied to Chemistry
Exercise 14.3.2. Show that the spatial graph in Figure 14.3.1 is topo-
logically achiral but not rigidly achiral.
Figure 14.3.1
The question whether or not a spatial graph is achiral is of particu-
lar interest to chemists. For we may think of a vertex as an atom of some
molecule. If in this molecule two atoms have a common bond, we may
represent this by an edge connecting the two vertices that correspond
to the atoms. In this way, the spatial graph we construct represents
the structure (model) of a molecule. An attempt to create a molecule
that has in its molecular structure a spatial graph was undertaken at
the beginning of the 20
th
century. However, it was only in 1981 that
the chemists H. Simmons and A. Paquette managed to synthesize a
molecule with the graph in Figure 14.3.2 in its molecular structure.
Figure 14.3.2
In Figure 14.3.2, the symbol "0" denotes an oxygen atom, and the
other atoms are carbon atoms, but the hydrogen atoms have been omit-
ted. Before this molecule was found, D.M. Walba had been successful in
synthesizing a molecule whose molecular structure was a Mobius band
M
3
, Figure 14.3.3(a).
M
3
, more generally M
n
[Figure 14.3.3(b)], and its mirror image
, from the point of view of chemistry, do not mutually change, so
they should not be equivalent, Le., it is conjectured that M
n
is chiral.
In this section, we shall show that this conjecture can be solved by
applying knot theoretical techniques.
Chapter 14
Figure 14.3.3
298
(b)
Theorem 14.3.1.
The Mobius band M
n
(n 4) is topologically chiral. Therefore,
it cannot be rigidly achiral.
Theorem 14.3.2.
If n = 3 there does not exist an (orientation preserving) auto-
homeomorphism of R
3
that transforms Mg to Mg and
to its mirror image for each i = 1,2,3. (14.3.1)
Exercise 14.3.3. Show that if we remove condition (14.3.1), we can
transform Mg to Mg.
Exercise 14.3.4. Suppose n 4, show that if there is an orientation
preserving auto-homeomorphism, f, of R3, which transforms M
n
to
then there exists an orientation preserving auto-homeomorphism,
g, of R3, which transforms M
n
to M: and transforms to
for each i = 1,2, ... , n.
Since Theorem 14.3.1 can be proven using Theorem 14.3.2 and
Exercise 14.3.4, we shall only prove Theorem 14.3.2 (cf. Exercise 14.3.5).
Proof of Theorem 14.3.2.
Suppose that the auto-homeomorphism of Theorem 14.3.2 exists
and denote it by h. Firstly, let us change M
3
to the form in Fig-
ure 14.3.4(a).
Since the cycle C = is a trivial knot, the 2-fold
cyclic covering space 1M of S3 branched along C is S3 (cf. Chapter 8,
Section 3, here we assume the graph is in S3). In 1M (= 8
3
), the three
semicircles Gi = are extended to three circles 0i = Cti U as
shown in Figure 14.3.4(b) (cf. Exercise 8.3.1). These three circles form
the same link L as in Figure 4.5.6.
299
(a)
Graph Theory Applied to Chemistry
(b)
Figure 14.3.4
Similarly, we can consider the 2-fold cyclic covering space M* of
8
3
branched along the trivial knot C* = b l b b b i b ~ b ~ b l in Mg, the
mirror image of Mg. As above, the semicircles {3i = bib ~ form circles
~ i = Pi U { ~ in M* (= 8
3
), Figure 14.3.5(b). These three circles form a
link L* that is the mirror image of L. Since L is chiral (Exercise 4.5.7),
there does not exist an orientation preserving homeomorphism of Rg
(and S3) onto itself that maps L to L*. It contradicts the existence of h,
since h can be extended to an (orientation-preserving) homeomorphism
from M to M* that maps L to L*.
(a)
(b)
Figure 14.3.5

Exercise 14.3.5. Prove that for n ~ 4, M

n
is topologically chiral.
Exercise 14.3.6. Calculate the graph invariant, "\(t) , for M
4
, and
hence show M
4
and M
4
are not equivalent (cf. Exercise 14.2.6).
Towards the end of the 1980s in the midst of the Jones revolution,
V.A. Vassiliev introduced a new concept that has had profound signifi-
cance in the immediate aftermath of the Jones revolution in knot theory
[V]. The importance of these so-called Vassiliev invariants lies in that
they may be used to study Jones-type invariants more systematically.
The other significance of the Vassiliev invariants to knot theorists
is in the fact that they give a certain topological interpretation to the
Jones-type invariants. But, to some dismay, this topological interpre-
tation is not along classic lines. So, although the Vassiliev invariants
301 Vassiliev Invariants
add further insight to the Jones-type invariants, the question of how the
Jones-type invariants relate to classic knot theory, for example, covering
spaces, remains open.
We shall see in this chapter that the exact form of the Vassiliev
invariants differs from anything we have discussed thus far in this book.
The knot invariants, like the Alexander polynomial, associate a knot
with some sort of mathematical quantity. A Vassiliev knot invariant,
on the other hand, is an invariant that satisfies a set of conditions. In
this sense, all the invariants introduced or redefined in Chapter 11 - the
Jones polynomial, skein polynomial, and the Alexander polynomial -
can all be shown to be Vassiliev invariants. However, not all the knot
invariants we have defined in this book are Vassiliev invariants; we shall
see later that the signature of a knot is not a Vassiliev invariant.
At the time of writing, the concept of Vassiliev invariants is still in a
state of flux, with much of the research scattered in various journals and
languages. We aim in this chapter to give an elementary introduction
to this new concept, which will hopefully allow the reader to follow and
maybe develop this concept.
1 Singular knots
In the previous chapter we discussed spatial graphs, and in par-
ticular we defined an invariant of 3-regular graphs called the Yamada
polynomial. The Vassiliev invariant is also connected to spatial graphs,
and it can best be described as an invariant of certain 4-regular spatial
graphs called singular knots.
In its most basic understanding, a singular knot is a "knot" with
self-intersections. More precisely, a singular knot, K, is (the image of)
a (PL-)mapping f: 8
1
---+ R3, which is one-to-one except for a finite
number points on S1. FUrther, (1) if two points on S1 have the same
image, then K =f(Sl) intersects itself at right angles at this common
point, and (2) no three points on 8
1
can have the same image.
Throughout this chapter, we shall call a point of self-intersection
of a singular knot, K, a vertex of K. Some examples of singular knots
are given in Figure 15.1.1 (a) rv (f).
It is possible to extend the definition of a singular knot to the case
of links; however, in the discussion that follows we will restrict ourselves
to singular knots. Further, for convenience we shall not explicitly dis-
tinguish between singular knots and the usual kind of knots we have
Chapter 15 302
discussed in this book; however, if we refer just to a knot, then this will
never mean a singular knot.
(a)
(b)
(c)
(d)
Figure 15.1.1
(e)
(f)
As mentioned above and as is now clear from our definition of a
singular knot, we may consider a singular knot to be a 4-regular spatial
graph. However, we need to refine what we mean by spatial graph the-
ory, since a singular knot does not fall exactly into this category. Firstly,
we may assign to a singular knot an orientation that is induced from
the orientation on S1; for spatial graphs, in general, there is no obvious
way of assigning an orientation. Secondly, the notion of equivalence of
singular knots is stronger than that for spatial graphs in general.
To be precise, let us associate with each vertex A of a singular knot
K a small (closed) neighbourhood BA and a plane EA that contains
BAn K - that part of K that is near A. Then a (plane) disk FA =
EA n BA is called a fiat disk associated with A. A collection of these
flat disks, one for each vertex A, will be denoted by F(K).
Definition 15.1.1. Two singular knots K, K' are said to be equiv-
alent, denoted by K K', if there is an orientation-preserving auto-
homeomorphism <p: R
3
R
3
that satisfies the following conditions:
(1) <p(K) =K';
(2) There exist collections of flat disks :F(K) and :F(K') for K
and K', respectively, such that cp maps F(K) to F(K').
(Such a transformation is called fiat.)
Intuitively, K is equivalent to K' if we can continuously deform K
to K' without causing any further self-intersections, and during the de-
formation the segments at the intersection points remain at right angles.
303 Vassiliev Invariants
The second part of the above definition is important since therein
lies the criterion that makes the equivalence of singular knots stronger
than for spatial graphs in general.
15
Example 15.1.1.
(a) (b)
Figure 15.1.2
In each of Figure 15.1.2(a) and (b) we performed a vertical twist
inside the ball W (keeping the outside fixed). In case (a) this produces
two equivalent singular knots, but in case (b) they are not equivalent.
However, cases (a) and (b) are equivalent as spatial graphs.
Example 15.1.2. I The two singular knots (a) and (b) in Figure 15.1.1
are not equivalent, but again they are equivalent as spatial graphs. The
same holds in the case of the two singular knots (c) and (d) of the same
figure. (See also Exercise 15.3.3.)
In line with our previous discussion, we can, using singular dia-
grams, redefine the equivalence of singular knots.
Proposition 15.1.1.
Two singular knots, K and K', are equivalent if and only if there
exist singular diagrams D and D', respectivley, which can be deformed
into each other by applying a finite number of times (plane isotopy and)
(1) The Reidemeister moves or their inverses except within a small
neighbourhood of each vertex;
(2) The following operation, 0, near the vertices:
or
Figure 15.1.3
Chapter 15 304
Exercise 15.1.1. (1) Show that the two singular knots (e), (f) in
Figure 15.1.1 are equivalent.
(2) Show that the two singular knots in Figure 15.1.2(b) are equiv-
alent to (c) and (d) in Figure 15.1.1.
2 Vassiliev invariants
Let us assume Vo is some numerical knot invariant; i.e., to each
knot, K, Vo assigns a rational number.
16
Then we may extend Vo to
an invariant v for singular knots as follows.
Suppose that v has already been defined for singular knots with at
most n-1 vertices. Now, let K be a singular knot with n vertices, and
further K, K+ and K_ are (regular diagrams of) singular knots that
are the same everywhere except at a neighbourhood of one vertex. In
this neighbourhood, they differ only in the way shown in Figure 15.2.1.
K
Then we shall define
K+
Figure 15.2.1
v(K) = v(K+) - v(K_) (15.2.1)
Since K has n vertices and both K+ and K_ have n - 1 vertices, we
may, by hypothesis, evaluate v(K).
Exercise 15.2.1. Show that v defined by means of (15.2.1) is in fact
an invariant for singular knots. In other words, if K and K' are equiv-
alent singular knots, then v(K) = v(K').
The above singular knot invariant, v, depends implicitly on the knot
invariant Vo from which we start the inductive process. Therefore, this
invariant v will be called the singular knot invariant induced by the knot
invariant vo. The relation (15.2.1), which we may call a singular skein
relation, can be rewritten diagrammatically as
v(X) = v(X) - v(X)
(15.2.1a)
305 Vassiliev Invariants
As might be expected from our previous discussions, we shall call
the diagrams in Figure 15.2.1 singular skein diagrams. Further, we
shall call the operation that replaces one diagram in Figure 15.2.1 by
the other two skein diagrams a singular skein operation. When we apply
the operation that replaces K+ (or K_) by K_ (or K+) and K, it
can be said that we are applying an unknotting operation that results
in a new vertex.
Definition 15.2.1. A singular knot invariant v [satisfying (15.2.1)] is
called a Vassiliev invariant of order (at most) m (or can also be said
to be of finite type) if for any singular knot with m + 1 vertices,
v(K) = O. (15.2.2)
In particular, if v is of order at most m but not of order m - 1,
Le., there exists a singular knot with exactly m vertices for which v is
non-zero, then v is called a Vassiliev invariant of order (exactly) m.
Exercise 15.2.2. Show that if v is a Vassiliev invariant of order at
most m, then v vanishes for any singular knot with at least m + 1
vertices.
As can be seen from the above definitions, a Vassiliev invariant
is essentially different from our previous knot invariants, for example,
the Jones polynomial, the Alexander polynomial, and the bridge num-
ber. These former knot invariants associate some sort of mathematical
quantity to each knot. Rather, a Vassiliev invariant is a singular knot
invariant that satisfies (15.2.1) and (15.2.2). Therefore, there are (in-
finitely) many Vassiliev invariants, and it seems more natural to call
them invariants of Vassiliev type.
Obviously there are many knot invariants that induce Vassiliev in-
variants; however, not all knot invariants induce Vassiliev invariants, for
a more detailed discussion, see Section 3. Before we give some examples
of the knot invariants that induce Vassiliev invariants, we will prove
some propositions that follow easily from the definition of a Vassiliev
invariant.
Proposition 15.2.1.
Let v be a Vassiliev invariant. If a singular knot K has a "loop"
X), then
v(K) = o.
Chapter 15 306
Proof
If we apply (15.2.1) to the vertex that forms part of the loop, then
we immediately write down the following equation:
V(K)=V([)0) =V([)0)-V([)0) =0.

Exercise 15.2.3. Show that if a singular knot K has a vertex as in

Figure 15.2.2, then v(K) = O.
Figure 15.2.2
Proposition 15.2.2.
Let v be a Vassiliev invariant of order o. Then for any non-singular
knot K,
v(K) =v(O).
Therefore, there is essentially only one Vassiliev invariant of order
zero.
Proof
Since v is of order 0, it follows that v ( X) = 0, and hence from
the singular skein relation v ( X) = v ( X). This implies that if we
apply an unknotting operation at any crossing point of K, then the value
of v remains constant. Since a knot can be deformed to a trivial knot by
applying several unknotting operations (cf. Proposition 4.4.1), it follows
that v(K) =v(0). Therefore, vis a constant for any non-singular
knot.
Proposition 15.2.3.
There is no Vassiliev invariant of order (exactly) one.
Proof
Let us assume that there exists a Vassiliev invariant of order one,
say, v. Then, by assumption, v is zero for any singular knot with two
307 Vassiliev Invariants
or more vertices. But there does exist a singular knot K
1
with a single
vertex for which v(K
1
) # o. We can apply unknotting operations to
K
1
which in turn increases the number of vertices, and thus K
1
can be
deformed to the singular knot K in Figure 15.1.1(a).
Since v vanishes for any singular knot with two or more vertices,
it follows from (15.2.1) and Proposition 15.2.1 that v(K
1
) == v(K) == 0,
see Figure 15.1.3. However, this contradicts our original hypothesis.

Let K be a singular knot with m +1 vertices, and number these

vertices 1,2, ... m +1. Now let us apply a singular skein operation at
each vertex to eliminate these vertices. This process will create 2
m
+
1
knots {K1'2' 'm+l} fromK,where 1'2' ... , f
m
+
1
, are either +
or -, and K1, , m+1 denotes the knot obtained from K by replacing
the vertex k (k == 1,2, ... , m+1) by a positive crossing point if k == +,
or by a negative crossing point if k = -. By repeatedly applying the
singular skein relation, we obtain the following formula:
v(K) = L (-1)lv(K1'2''m+l)'
1'2''m+l
(15.2.3)
where the summation is taken over all the set of 2
m
+
1
elements
1,f
2
,, m+l' and l is the number of - in the sequence 1' ... ,
m+l. Using (15.2.3) we may replace (15.2.2) by a more convenient form
(15.2.4) of the following proposition.
Proposition 15.2.4.
A singular knot invariant (that satisfies (15.2.1)) is a Vassiliev in-
variant of order at most m if and only if for any singular knot with
m +1 vertices,
(15.2.4)
where the summation is taken over the same set as in (15.2.3).
To evaluate a Vassiliev invariant, it is as in the case of the skein
polynomials more convenient to draw singular skein tree diagrams (cf.
Example 6.2.1).
Example 15.2.1. In Figure 15.2.3 we have drawn the singular tree
diagram to evaluate a Vassiliev invariant for a singular knot with two
vertices.
Chapter 15
v=o

Figure 15.2.3
Therefore,
v(K) = -v(K_+) +v(K__) = -v(O) +v(K__).
308
Exercise 15.2.4. Draw singular skein diagrams for the singular knots
in Figure 15.1.1.
3 Some examples of Vassiliev invariants
In this section we shall give several concrete examples of Vassiliev
invariants. We shall also show that there exist some knot invariants
that are not of finite type, and hence are not Vassiliev invariants.
Proposition 15.3.1.
Let K be a knot and
V'K(t) = 1 +a2z2 + a4z4 + ... + a2m
z2m
+ ...
be the Alexander-Conway polynomial of K. Then V'2m(K) a2m
(m = 1,2, ... ) induces a Vassiliev invariant of type exactly 2m.
309 Vassiliev Invariants
Proof
It is sufficient to show that the following two conditions hold:
(1) For a singular knot with 2m +1 vertices, K, V
2m
(K) = 0;
(2) There is a singular knot with 2m vertices, K
o
, such that
V'2m(K
o
) =p o.
By our work in the previous section, we may rewrite (1) as
(1)' E(-1)lV2m(K
f1
, 2' ... , 2m+l ) =o.
Our aim will be to prove (I)' rather than (1). In this regard, let us
denote by (Kl' 2' ... , i' 0, ... , 0) the knot obtained from Kl' 2' ... , 2m+l
by replacing the last 2m+1-i crossing points by 0. We are now
in a position to apply the skein relation (6.2.1) that defines the Conway
polynomial, and so obtain the following formulae:
For any 1 ' 2' , 2m+1 '
If we apply the skein relation once again, but this time to
K1'2" .. '2m'0' we obtain
V'K (z) - VK (z) = ZVK (z).
l' ... E:
2m
-
1
, +, 1' ... , 2m-l' -,0 E
1
, ... 2m-l' 0, 0
We keep applying the skein relation until we reach the following
equation:
VK (z) - VK (z) = ZVK (z).
+,0, ... 0 -,0, .... 0 0,0, ... ,0
Collecting these terms together, we finally obtain
(15.3.1)
Since the right-hand side of (15.3.1) does not contain the term z2m,
V
2m
(K) = o.
Now turning our attention to the proof of (2), let us consider the
singular knot K[p, q], depicted in Figure 15.3.1, where p is the number
of vertices and Iql is the number of crossing points.
Chapter 15
K[5,S]
K[5,-3]
310
(a) (b)
Figure 15.3.1
Let us look at the specific singular knot K[2m,1] and compute
V
2m
(K[2m,1]), the required formula is given in (15.3.2):
(15.3.2)
If we refer back to our discussion on torus knots and their generic
form (Chapter 7), then it is easy to see that K[2m,1]1'2' ... '2 is a
torus knot of type (2m-21+1,2), where l is the number of" - ,Fin the
sequence 1' 2' , 2m Therefore, ~ t = 1 if 1=m or m +1 and
~ t =t-(m-l) + +t
m
-
l
=1 +... +z2{m-l) if 0 ~ 1< m,
~ t =t-(l-m-l) + +t
'
-
m
-
1
= 1 +... +z2(l-m-l) if m +1 < 1.
Therefore, V2m(K[2m,l]1'2'.'.'2m) = 1 if each i is + and
zero otherwise. Hence, we may deduce that V2m(K[2m, 1]) =1.

The above proposition shows that there exists a Vassiliev invariant

induced from the Alexander-Conway polynomial, V2m- The reader may
ask, Can the same be said of the Jones polynomial? The next theorem
answers this question in the affirmative.
Theorem 15.3.2 (BL].
Let VK(t) be the Jones polynomial of a knot K. Let VK(q) be
the infinite series obtained from VK (t) by substituting e
q
(= 1 + q +
~ ~ ~ .
2! + - - = L-J n! for t. So we may wrlte
n=O
VK(q) = b
o
+b1q +b2q2 +... .
Then Jm(K) = b
m
is a Vassiliev invariant induced by the Jones
polynomial of order (at most) m.
311 Vassiliev Invariants
As might be expected, it is also possible to define a Vassiliev invari-
ant, induced from the Nth (N 2) degree Jones polynomial,
using the same substitution (see Example 12.3.3). However, we
shall here discuss another way of obtaining a Vassiliev invariant from
the Jones polynomial.
Proposition 15.3.3.
Let <PK (t) be the Taylor expansion of the Jones polynomial of a
knot K at t = 1, which we may write as
where if we recall some first-year calculus, em = t=l
Then <PK(t) = C
m
is a Vassiliev invariant of order (at most) m.
Exercise 15.3.1. (1) Prove Proposition 15.3.3.
(2) If K is the right-hand trefoil knot, show that
(i) V2(K) = 1 and Vi(K) = 0 for i 3;
(ii) J
2
(K) = -3 and J3(K) = -6;
(iii) Evaluate (see Example 12.3.3), and <Pi(K)
for i 2.
(3) For the figure 8 knot, evaluate J
i
for i = 2,3,4, Vi, and <Pi
for i 2.
(4) For K =K[4, 1], show that J
2
(K) = J
3
(K) =0, and evaluate
J
4
(K).
Exercise 15.3.2. Show that if K* is the mirror image of a knot K,
then Jm(K*) = (-I)mJm(K) for all m 2.
Exercise 15.3.3. (1) Show, by evaluating J
2
or V
2
, that the two
singular knots in Figure 15.1.1(a), (b) are not equivalent.
(2) Show, by evaluating J
a
, that the two singular knots in Fig-
ure 15.1.1(c), (d) are not equivalent, and hence the two singular knots
in Figure 15.1.2(b) are not equivalent (see Exercise 15.1.2).
Remark 15.3.1. It is possible, via the substitution t = e
q
, to in-
duce Vassiliev invariants from the skein polynomials and the Kauff-
man polynomial in the same way as in Theorem 15.3.2. However, since
these polynomials have two variables, they need to reduced to I-variable
polynomials. In the case of the skein polynomial, P
K
(v, z), the polyno-
mial JK(t) given in Example 12.3.4 is such a I-variable specialization
Chapter 15 312
of PK(V, z). Not surprisingly, JK(t) induces a Vassiliev invariant, for
more details see Birman and Lin [BL].
We have seen that it is relatively easy to build up a variety of Vas-
siliev invariants induced from the knot invariants in Chapter 11. How-
ever, when we start to look at what we may call the "classic" geometric
invariants rather than the "algebraic," then the stream of Vassiliev in-
variants dries up. Let us look a bit more closely at why this is the
case.
Proposition 15.3.4.
The singular knot invariant, v, defined by the signature of a knot
K is not of finite type; hence it is not a Vassiliev invariant.
Proof
Let us assume that v is of order at most n ~ 2).
We intend to show that v(K[n + 1, n]) =1= 0, or equivalently,
(15.3.3)
Now, since K[n + 1, n]1'2' ."'n+l is a torus knot of type (2n-
2l + 1,2), its signature, see Theorem 7.5.1(IV), is
(i) If If:.n,n+1, u(K[n+l,n]1'2, ,f
n
+
1
)=-(2n-2l);
(ii) If l = n or n + 1, u(K[n + 1, n]l' 2' , n+l ) = o.
It is quite easy to see that the number of knots with l negative
signs in the sequence f 1 ' f 2' . , f n+l' is (nt
1
). So the left-hand side
of (15.3.3) is
2:(-l)lu(K[n + 1, n]1,f
2
, ...'n+1 ) =~ - I / + 1 (n? 1) (2n - 2l),
which can be shown to be non-zero, see Exercise 15.3.4. This now
contradicts the assumption that v is of order at most n.

Exercise 15.3.4. Show that if n ~ 1,

If we look closely at the proof of the previous proposition, we can
easily prove the following proposition:
313 Vassiliev Invariants
Proposition 15.3.5.
None of the following classical geometric invariants: the minimal
crossing number c(K), the unknotting number u(K), the bridge number
br(K), the braid index b(K) and the genus g(K) of a knot, K is of finite
type; hence, they cannot be Vassiliev invariants.
Exercise 15.3.5. By considering the singular knot K[n +1, n]' prove
Proposition 15.3.5.
4 Chord diagrams
Besides the equality in Proposition 15.2.1 (and Exercise 15.2.3),
any Vassiliev invariant, v, satisfies a further important "linear" equation
called the 4-term formula.
Theorem 15.4.1 (4-term formula).
The following equality holds for any Vassiliev invariant, v,
v - v
= v - v
(15.4.1)
Figure 15.4.1
where the 4 singular knots are identical outside the circles.
The above "graphical" formula may be interpreted as the equivalent
of a "Reidemeister move 0
3
" for singular knots. However, the keen
reader may notice that there is in fact a slight difference between the
over-crossing and under-crossing at each crossing point.
Proof
Let us consider the two equivalent singular knots K and K'. K and
K' are identical, except inside the circles shown in Figure 15.4.2.
(a)
Figure 15.4.2
(b)
Chapter 15 314
If we apply singulax skein operations at the crossing points a and
b in the circles, Figure 15.4.3, then we obtain the following equalities:
v(K) = v
= v
v(K') = v
= v
v
+ v
v
+ v
+ v
+ v
+ v
+ v
Figure 15.4.3
The first terms of the final expansions of v(K) and v(K') axe equal;
this follows because they are the v-value of equivalent singular knots.
So, if we rearrange the remaining terms we obtain the desired formula
(15.4.1). '

Exercise 15.4.1. Show that for any Vassiliev invariant, v, the follow-
ing formula holds:
v
- v
- v - v
Figure 15.4.4
315 Vassiliev Invariants
Due to Proposition 15.2.2, the Vassiliev invariant, vo, of order 0 is
essentially "unique." Further, we have shown in Proposition 15.2.3 that
there is no Vassiliev invariant of order 1. Using the same ideas, we may
determine the Vassiliev invariant of order (exactly) 2.
To this end, let us apply unknotting operations to a singular knot
K, which will cause vertices to be added to the subsequent singular knots
and also a trivial knot to be formed. Since V2 vanishes for any singular
knot with more than two vertices, we can write down the following
equality:
v2(K) =aV2 ( 0) +b
V
2 (00) +CV2 ((XX)) +dV2 ( ,
(15.4.2)
where a, b, c and d are integers. (See also Example 15.4.1 below.)
Example 15.4.1. Suppose K is the right-hand trefoil knot. To derive
an expression for v2(K) in the form of (15.4.2), we make use of the
singular skein diagram, see Figure 15.4.5.
Figure 15.4.5
Chapter 15
Since V2 ( ~ = 0, we can write
v2(K) =V2(O) +V2(OO) + V 2 ~ .
316
Now, since V2(OO) = 0 and V 2 ~ = 0, V2 is es-
sentially determined by V2 (0) and V2 ( ~ . However, anal-
ogously to V2(0) and J 2(0), it is natural to assign for any
Vassiliev invariant of order m (m ~ 2),
Vm(O) =0.
(15.4.3)
Therefore, it follows that v2(K) is completely determined by the
value of V2 ( ~ . Hence, V2 is essentially unique.
17
So let us
assign
(15.4.4)
(We may assign any non-zero number to it.)
Then the following equality
shows that the v2-value of a singular knot K with 2 vertices does not
depend on the sign of the crossing points in K. In other words, we can
ignore the difference between an over- and under-crossing point for a
singular knot with 2 vertices. Therefore, if we consider a singular knot
K as a mapping f: 8
1
~ R3, in the case of the v2-value of K all
that we need to concern ourselves with is, Which two points of the four
points a, b, c, and d on 8
1
have the same image?
Example 15.4.2. Consider a singular knot, K, as the mapping f :
8
1
~ R3.
317 Vassiliev Invariants
Case 1 f(a) = f(b) and f(c) = f(d).
Although K and the singular knot in Figure 15.4.6(a) may not be
equivalent, they have the same v2-value.
feb) =fed)
(b)
Figure 15.4.6
Case 2 f(a) = f(c) and f(b) = f(d).
In this case, K and the singular knot in Figure 15.4.6(b) have the
same v2-value, even though they may not be equivalent, this follows
from (15.4.5).
The remaining case, Le., f(a) = f(d) and f(b) = f(c), can be shown
to reduce to Case 1.
To emphasize the above connection between the four points, we
shall assign diagram (a) in Figure 15.4.7 to Case 1 and diagram (b)
of the same figure to Case 2. [As noted above, we may ignore the
under- and over-crossing information of the intersection of the arcs in
Figure 15.4.7(b).]
(a)
Figure 15.4.7
(b)
The diagrams in Figure 15.4.7 are usually called chord diagrams.
Once we assign v2-values to the two chord diagrams in Figure 15.4.7,
it is possible to evaluate V2 for any singular knot. The table of chord
diagrams with their v2-values is called the Actuality Table.
Chapter 15
Actuality Table for V2
~ = o ~ =
318
Figure 15.4.8
Since V2 (((j)) must be zero by Proposition 15.2.1, this chord
diagram is usually omitted from the Actuality Table. In essence, the
choice of value for V2 ( $) is arbitrary as long as we do not assign
zero to it.
It is possible to keep on building Actuality Tables ad infinitum;
however, we shall only consider one more case - Vassiliev invariants of
order 3, Va.
In a similar way to (15.4.2), we can write for any singular knot, K,
(15.4.6)
where a, b, and c are integers. In the above formula we have omitted
those singular knots, for example, 0 and 00, whose va-value,
by (15.4.3) or Proposition 15.2.1, must be zero.
Exercise 15.4.2. Determine the values of a, b, and c in (15.4.6) for
the right-hand trefoil knot and the figure 8 knot.
Therefore, to determine vg(K) it is sufficient to assign values to
the singular knots in (15.4.6). .
Caveat lector, the value of va ( ~ in the Actuality Table need
not necessarily be the same value as V2 ( ~ = V2 ($). The
reason is that the vg-value of a singular knot, K, with 2 vertices may
depend on the sign of the crossing points of K. In fact, from (15.2.1),
319 Vassiliev Invariants
In comparison to V2, there are 3 chord diagrams, shown below,
for Vs.
EB
To construct the Actuality Table for V3, we must assign a vs-value
to each of these three chord diagrams.
However, in contrast to the previous case, this is not so simple a
matter. The sticking point is that we may no longer assign arbitrary
values to the chord diagrams. For example, V3 (@) and V3 ( <!9)
are not independent. There is, in fact, an equation, see Figure 15.4.9,
that involves the both of them.
Figure 15.4.9
On close inspection, the four configurations inside the circles of the
above figure can be seen to satisfy the 4-term formula of Theorem 15.4.1.
It follows immediately from Proposition 15.2.1 that v3(K
2
) = o.
Exercise 15.4.3 Show that v3(K
1
) = v3(K2'). (Caveat lector, Kl
and K
2
' are not necessarily equivalent.)
The above exercise allows us to write down the following equality:
In this way, we must find all the equations involving these chord
diagrams and then assign v3-values to these chord diagrams so that
these equations are satisfied. (The situation is not as bad as it seems.
Chapter 15 320
It is known that any equality that holds for these chord diagrams of any
order is obtained "only" from the 4-term formula and variations on this
formula [BN].) This will lead to the Actuality Table for Va.
Actuality Table for Va
v
a
=1
va=o
Figure 15.4.10
Since there is only one arbitrary choice for either V3 (@) or
V3 ( (3), a Vassiliev invariant of order 3 is essentially unique.
Following the above procedure, Le., by means of (15.2.1), (15.2.2),
and the Actuality Table, we may completely determine the Vassiliev
invariants, v
m
, of order m. Therefore, it is possible to axiomize V
m
by taking as Axioms I, II, and III the equations (15.2.1), (15.2.2), and
(15.4.3) and the Actuality Table as the initial data. More precise details
are given in Birman and Lin [BL).
Even in the case m = 3, determining the Actuality Table was not
an easy matter. In fact, as the value of m increases, determining the
Actuality Table requires tremendous computing power. For example,
for m = 8 it is necessary to solve more than 300,000 linear equations
with more than 40,000 unknowns! For the adventurous reader, further
details may be found in Bar-Natan [BN].
Exercise 15.4.4. (1) Show that for the right-hand trefoil knot, K,
and its mirror image, K*,
v2(K) =1 and v3(K) = 2; v2(K*) = 1 and v3(K*) = 0.
(2) Show v2(K) = -1 and v3(K) = -1 for the figure 8 knot, K.
Exercise 15.4.5. Show that for any knot K, V'2(K) = v2(K). Is it
possible to express the Vassiliev invariants J2 and J3 in terms of V2
or V3?
321 Vassiliev Invariants
Exercise 15.4.6. (1) Suppose that a chord diagram D has a simple
arc that is disjoint from the other arcs in D. Show that for any Vassiliev
invariant, v, v(D) = O.
(2) Find the chord diagrams for a Vassiliev invariant of order 4
whose V4-value is not automatically zero by Proposition 15.2.1. (Hint:
It is known there are 7 such diagrams, and there are essentially 3 dif-
ferent Vassiliev invariants of order 4.)
5 Final Remarks
As we have seen in this chapter, the Alexander-Conway polynomial,
the Jones polynomial, and the skein polynomial induce Vassiliev invari-
ants. So, in a sense we may say that Vassiliev invariants are "stronger"
than the polynomial invariants. This slightly ambiguous last statement
may allow some optimism that the Vassiliev invariants may distinguish
two knots. However, at the time writing, this conjecture by Vassiliev
remains open. We should emphasize that it is not sufficient just to
consider specific Vassiliev invariants. For this conjecture to hold "all"
Vassiliev invariants must be taken into account, because we have exam-
ples that show there are infinitely many distinct knots that cannot be
distinguished by finitely many Vassiliev invariants.
In this regard, let K(n; l), n ~ 2, l ~ 0, be the knot depicted in
Figure 15.5.1, where 2l is the number of positive crossing points on the
far right band.
K(5;2)
~ ~ ~ ~ ~ J
8
Figure 15.5.1
K(n; 0), n 2:: 2, is an alternating knot. Since it is possible to find
for it an alternating diagram with 3n crossing points, we shall leave this
as an exercise for the reader. Therefore.. K 1 1 . ~ 0) is not a trivial knot.
Chapter 15 322
The Alexander-Conway polynomial and the Jones polynomial of
K(n; I) can be found in the following proposition.
Proposition 15.5.1.
For any integer I ~ 0,
(1) V'K(n;O)(z) = V'K(n;l)(Z)
(2) V'K(n;O)(z) = 1 - 2z
n
+ ... if n is even
V'K(n;O)(z) = 1 - 2z
n
+
1
+... if n is odd
(3) VK(n;l)(t) = t
2l
(V
K
(n;O)(t) - 1) + 1.
(15.5.1)
Exercise 15.5.1. (1) Prove (15.5.1)(1) and (3).
(2) Confirm that (15.5.1)(2) holds for K(5; 2).
(3) Show that K(n; I) ~ K(n; I') if and only if l = l'e
It follows from (15.5.1)(2) for the Vassiliev invariant V'm of order
ill, ~ <n,
V'm(K(n; l) = 0. (15.5.2)
The result of (15.5.2) may be extended to any Vassiliev invariant,
v
m
. In fact, it has been proven [0] that for an arbitrary Vassiliev in-
variant, v
m
, of order m, with 2 ~ m < n,
vm(K(n; l)) = o. (15.5.3)
Therefore, it is an immediate consequence of (15.5.3) that there
are infinitely many knots, K(n; l) that are indistinguishable from the
trivial knot by any Vassiliev invariant of order m n). Since we may
take n to be arbitrarily large, K(n; l) and the trivial knot cannot be
distinguished by "finitely" many Vassiliev invariants.
Using K(n; l), we can prove the next theorem:
Theorem 15.5.2 [0].
For a knot K and n ~ 2, let K
'
= K#K(n; l), l ~ 0, be the
connected sum of K and K(n; l). Then
(1) K
'
is not equivalent to K;
(2) For any Vassiliev invariant V
m
of order m, if m < n then
vm(K') = vm(K).
323 Vassiliev Invariants
Therefore, there are infinitely many distinct knots that cannot be
distinguished by finitely many Vassiliev invariants.
In the above example K' is not a prime knot. However, there does
exist an example of the same property as above in which both K and
K' are prime knots [St].
If, on the other hand, we consider "all" Vassiliev invariants, then
the situation is quite different. One of the strengths of the theory that
surrounds the Vassiliev invariants is that it allows us to treat the poly-
nomial invariants in a systematic way. Hence, the Vassiliev invariants
may reveal relationships between the polynomial invariants. In one par-
ticular case, a surprising result of this kind has been found [MelM].
Theorem 15.5.3 (BN-G].
Let K and K' be two knots. If for all N 2: 2 the Nth degree Jones
polynomials are equal, then their Alexander polynomials are also equal.
The above theorem may be paraphrased as follows: The set of all
Nth degree Jones polynomials {JW)(t), N 2: 2} determines the Alexan-
der polynomial. This result in itself is quite surprising, since from Chap-
ter 11 we know that the Jones polynomial, VK(t), does not determine
the Alexander polynomial, ~ K t ) , and the converse is also true.
The importance of Theorem 15.5.3 is in the fact that it may be a
crucial part in the quest to find a non-trivial knot with the trivial Jones
polynomial (cf. Chapter 11, Section 2). The equation "VK ( t) = 1 " may
not imply ~ K t ) = 1," but as a consequence
18
of Theorem 15.5.3,
we now know that
~ N )
J
K
(t) = 1, for all N 2: 2, ===> AK(t) = 1. (15.5.4)
This may be used as a basis for the following conjecture.
Conjecture. If for all N 2: 2, iW)(t) = 1, then K is the trivial knot.
So, even if Vassiliev invariants are not strong enough to distinguish
two knots, they may be able to distinguish knots from the trivial knot.
As we mentioned in the introduction to this chapter, the study of
Vassiliev invariants is still in its nascent stages. But on the basis of
present research, we may predict that the Vassiliev invariants will play
a major role in the future development of knot theory.
For an excellent more detailed introduction to the theory of Vas-
siliev invariants, we refer the reader to [Bi]. The ambitious reader may
care to peruse the more advanced references that may be found in [BN].
A table of knots and their knot invariants
Appendix (I) is a complete list of prime knots with up to 8 crossing
points. In this set of knots,
(1) The amphicheiral knots are 4
1
, 6
3
, 8
3
, 8g , 812, 817, 8
18
;
(2) There is only one non-invertible knot, 8
17
;
(3) The only non-alternating knots are 819, 820, 821;
(4) All the knots are 2-bridge knots, except the following nine
knots 8
5
, 810, and 815 -8
21
, which are 3-bridge knots.
Appendix (II) is a table of the Alexander polynomials and the Jones
polynomials of the knots in Appendix (I).
The notation
denotes the polynomial
tn(ao +al t +a2
t2
+ ... + amt
m
).
For example,
(-2)[-1 +2 - 3 +0+4 +1] = t-
2
(-1 +2t - 3t
2
+Ot
3
+4t
4
+t
5
)
= _t-
2
+2t-
1
- 3 +4t
2
+t
3

If the Jones polynomial of a knot is given by

then the Jones polynomial of the mirror image, K*, of K is given by
In the case of the Alexander polynomials, the polynomials for K
and K* are the same.
Appendix
Appendix (I): A table of knots
326
327
Appendix
Appendix (II): Alexander and Jones polynomials
Knot Alexander polynomial
31 (-1)[1-1+1]
41 (-1)[-1+3-1]
51 (-2}[1-1 +1-1+ IJ
52 (-1)[2 - 3 +2]
61 (-1}[-2+5-2]
62 (-2}[-1 +3 - 3+3 -1]
63 (-2)(1 - 3 +5 - 3 +1)
71 (-3)[1-1+1-1+1-1+1)
72 (-1)[3-5+3]
73 (-2)[2-3+3-3+2]
74, (-1)(4-7+4]
15 (-2)[2-4+5-4+21
78 (-2)[-1+5-7+5-1]
71 (-2)[1- 5 +9 - 5 +1]
81 (-1)(-3+7-3]

83 (-1)[-4+9-4]
84 (-2)[-2 +5 - 5 +5 - 2]

86 (-2)[-2+6-7+6-2]

88 ( -2)[2 - 6 +9 - 6 +2J

810 (-3)[1 - 3 +6 - 7 +6 - 3 +1]
811 (-2}[-2 +7 - 9 +7 - 2]
812 (-2)[1-7+ 13 -7+1]
813 ( -2)[2 - 7+11 - 7 +2]
814, (-2)[-2 +8 -11 +8 - 2]
8Ui (-2)[3 - 8 +11 - 8 +3]
816 (-3)[1- 4 +8 - 9 +8 - 4 +1)
8
1
7 (-3)[-1+4-8+11-8+4-1]
818 (-3}[-1 +5 -10+ 13 -10 +5 -I)
8
1
9 (-3)[1 - 1+0 +1 +0 - 1 +1]
820 (-2)[1 - 2 +3 - 2 +1]
821 (-2)[-1+4-5+4-1]
Jones Iynomial
(1)[1 +0 + 1 - 1]
(-2}[1 - 1 + 1 - 1 +1]
(2}[1 +0 + 1 - 1+1 - 1]
, (1)[1 - 1+2 - 1 +1 - 1)
(-2)[1 - 1 +2 - 2 +1 - 1 +1]
(-1)[1-1+ 2 - 2+ 2 - 2+ 1]
(-3)[-1 +2 - 2 +3- 2+2 -1]
(3)[1 +0 + 1 - 1 + 1 - 1+1 - 1]
(1)[1 - 1+ 2 - 2 + 2 - 1 + 1 - 1]
(2)[1 - 1 +2 - 2 +3 - 2 +1 - 1]
(1)[1 - 2 +3 - 2 +3 - 2 +1 - 1]
(2)[1 - 1 +3 - 3 +3 - 3 +2 - 1]
(-1)(1- 2 +3 - 3+4- 3 +2 -1]
(-3)[-1 +3 - 3 +4 - 4 +3 - 2 +1]
(-2)[1-1 +2 - 2 +2 - 2 + 1-1+1]
(0)[1-1 +2 - 2 +3 - 3 +2 - 2 +1]
(-4)(1 - 1 +2 - 3 +3 - 3 +2 - 1+1]
(-3)[1 - 1 +2 - 3 +3 - 3 +3 - 2 +1]
(0)[1-1 +3 - 3 +3 - 4 +3 - 2 +1]
(-1)[1-1 +3 - 4+4- 4+3 - 2 +1]
(-2}[-1 +2 - 2 +4 - 4 +4 - 3 +2 - 1]
(-3)[-1 +2 - 3+5 - 4+4 - 3+2 -1)
(-4)[1- 2 +3 - 4 +5 - 4 +3 - 2 +1]
(-2)[-1 +2 - 3+ 5 - 4+5 -4+2 -1]
(-1)[1- 2+4-4+5 - 5+3 - 2 +IJ
(-4)[1 - 2 +4 - 5 +5 - 5 +4 - 2 +1]
(-3}[-1 +3 - 4 +5 - 5 +5 - 3 +2 - 1]
(-1}ll - 2 +4 - 5 +6 - 5 +4 - 3 +1]
(2)(1 - 2 +5 - 5 +6 - 6 +4 - 3 +1]
(-2)(-1 +3 - 4 +6 - 6+6 - 5 +3 -IJ
(-4)(1 - 3 +5 - 6 +7 - 6 +5 - 3 +1]
(-4)(1 -4+6 -7+9 -7+6 - 4+ 1]
(3)[1+0+1+0+0+1]
(-1)[-1 +2 - 1 +2 - 1+1 - IJ
(1)[2 - 2 +3 - 3 +2 - 2 +1]
(1) Combinatorial topology (or PL-topology) is a branch of topology
that concerns itself with the study of complexes in Euclidean space.
Such complexes consist of points, segments, triangles, tetrahedra,
et cetem, Figure A.I. These constituent parts are called simplexes.
The union of all simplexes contained in a complex C is called a
polyhedron and is denoted by ICI. The dimension of a complex
is the highest dimension of the simplexes in C. In this book, the
simplexes that we shall encounter will not have dimension greater
than 3.
A complex C is locally finite, i.e., for each point P there exists
a neighbourhood of P that intersects only finitely many simplexes
in C.
A knot is a I-complex (or equivalently, a polygon constructed
from I-complexes) and so by the above definition of a complex a
point P of the type in Figure 1.0.1(b) cannot exist on a knot.
For a more detailed discussion of combinatorial topology we refer
the reader to, for example Glaser [G*] or Massey [M*] .
I-simplex

O-simplex

2-simplex
3-simplex
Figure A.I
(2) The set of all knots (or links) that are equivalent to a knot (or link)
K is called the type of K. Therefore, when we say that the knots K
and K' are equivalent, we mean that the type of K and the type of
K' are equal. Therefore, to be precise, knot theory concerns itself
with the type of knots (or links) rather than the knot itself.
Notes 330
(3) For the purposes of this book we shall deem all maps to be PL-
maps, namely, these are continuous maps from a polyhedron IXI to
a polyhedron IYI, which are also linear maps with regard to some
division of X. If such a map f is also a homeomorphic map, then' f
is said to be a PL-homeomorphism. An-ball Bn is an example of
a topological space that is PL-homeomorphic to an n-sirnplex. In
particular, a 2-ball is called a disk. While an (n -I)-sphere sn-l
is PL-homeomorphic to the boundary of an n-ball.
(4) This is called a one-point compactification of R
2
In general, a
one-point compactification of Rn is homeomorphic to sn, and so
it will be more convenient for us to think of sn as a compact space
obtained from Rn by adding a point to Rn.
(5) Let G be a (non-empty) set. G is called a group if for two arbitrary
elements a, b in G, we may uniquely define a product ab such that
(i) (ab)c = a(bc) (associative law).
(ii) There exists a (unique) element e in G such that for each a in
G, ea = a and ae = a. We shall call e the identity element.
(iii) For each element a in G, there exists a (unique) element a*
such that aa* = e and a*a = e. We shall call a* the inverse
element of a and denote it by a-I.
In addition, if ab = ba for any two elements a, b in G, then the
product is called commutative and G is called a commutative group.
If a set G satisfies (i) [but not necessarily (ii) or (iii)] then G
is called a semi-group. For example, the set of all non-zero ratio-
nal numbers is a group under the usual product, while the set of
all integers is not a group, but a semi-group, under the product.
Occasionally we may also mention some other algebraic structures,
for example, rings and fields; however, an understanding of such
(algebraic) structures is not essential for the reader of this book.
The interested reader should refer to any book on abstract algebra.
(6) It would seem that this is just a guess based on the 2-dimensional
case, but in fact it has a stronger foundation, namely, it is the
Schonflies Theorem.
Schonfties Theorem
A 2-dimensional PL-sphere S2 in a 3-sphere S3 divides S3
into 2 parts called the interior and the exterior, and each part is
PL-homeomorphic to a 3-ba11.
This is the generalization of the famous Jordan Curve Theorem:
331 Notes
Jordan Curve Theorem
A simple closed polygon C in R
2
divides R
2
into two domains
G
1
and G
2
, namely, G
1
UG
2
= R
2
and G
1
n G
2
= C.
We should note that there are examples of 2-spheres that are not
PL-2-spheres for which the SchonHies Theorem does not hold.
(7) L
1
consists of two trivial knots but one component of L
2
is a
(right-hand) trefoil knot, which we know is not equivalent to a triv-
ial knot. Therefore, L
1
and L
2
cannot be equivalent. To show
that their complements are homeomorphic requires a straightfor-
ward application of a Dehn surgery, a concept that we shall discuss
in Chapter 8.
(8) The root of the difficulty of this problem is that almost all alge-
braic invariants of knots are powerless when confronted with this
problem. H. 'Trotter had to apply techniques from 2-dimensional
hyperbolic geometry to show that the knot in Figure 3.2.2{a) is not
invertible. It is not particularly surprising that 2-dimensional hy-
perbolic geometry is useful in the above case, since 3-dimensional
hyperbolic geometry plays a fundamental role in the study of 3-
dimensional topology.
(9) For a link there may exist a closed polygon (3 in the interior of
Therefore, an addition to the definition of upper type and lower
type may be needed for these (3, but this is fairly straightforward
from the proof of Lemma 4.1.3.
(10) Let n be a positive integer. We write a == b (mod n) or a - b ==
o(mod n) if and only if a - b is divisible by n.
(11) Classification of closed orientable (connected) surfaces
Let F
m
be a 2-sphere with m (;::: 0) handles attached, see
Figure 5.2.1(a) or (b). Then
(i) A closed orientable connected surface is homeomorphic to Fm
for some m ;::: 0;
(ii) F
m
and F
n
are homeomorphic if and only if m = n.
For the classification of non-orientable closed surfaces refer to
Massey [M*].
(12) To be exact, these m (= 2g(F) + J-t{K) - 1) closed curves
at, a2 , am are chosen so that their homology classes [at],
[a2], , [am] form a basis for the I-dimensional homology group
(with coefficients in Z) H1(F; Z) of a Seifert surface F.
Notes 332
(13) For two simple closed curves A, B in a Lens space L(q, r), (q =I 0),
one can define the linking number. In this case the linking number
Ik(A, B) is not necessarily an integer, it could be a rational number
: (0 ~ ~ < 1). However, for two simple closed curves in a general
3-manifold, it may not be possible to define a linking number. For
example, the linking number between two simple closed curves in
8
2
x 8
1
cannot be defined.
(14) The Jones polynomial of an alternating knot (or link) K is also
alternating, but Theorem 11.5.2(1) may not hold, and so (2) should
be written as aiai+1 ~ o.
(15) Since a flat disk is irrelevant to the orientation, a system of flat
disks can be defined for any spatial graph, and flat equivalence
between two spatial graphs may be defined along similar lines to
Definition 15.1.1. In fact, for a 3-regular spatial graph there is no
distinction between (ordinary) equivalence and flat equivalence.
(16) Instead of the set of rational numbers, we may use the set of
integers or the set of real numbers.
(17) The set of all Vassiliev invariants of order (at most) m, V
m
(m ~ 2), forms a vector space (over the rational numbers). The
dimension of Vm/Vm-l [the vector space of all Vassiliev invariants
of order (exactly) m] is considered as the number of essentially
different Vassiliev invariants of order m.
(18) Equation (15.5.4) is not an immediate consequence of Theo-
rem 15.5.3. A precise relationship between {lW>, N ~ 2} and
~ K t has been established. Theorem 15.5.3 and (15.5.4) are con-
sequences of this relationship.
Books on Knots and Related Topics
[B*] J.S. Birman, Braids, Links and Mapping Class Groups,
Ann. of Math. Studies 82, Princeton Univ. Press (1974).
[BZ*] G. Burde and H. Zieschang, Knots, Studies in Math. 5, Walter
de Gruyter (1985).
[CF*] R.H. Crowell and R.H. Fox, Introduction to Knot Theory,
Graduate Texts in Math. 57, Springer-Verlag (1977).
[G*] L.C. Glaser, Geometrical Combinatorial Topology vol.I,
Van Nostrand Reinhold (1970).
[HZ*] F. Hirzebruch and D. Zagier, The Atiyah-Singer Theorem
and Elementary Number Theory, Math. Lecture Series 3,
Publish or Perish (1974).
[K*] A. Kawauchi, editor, A Survey of Knot Theory, Birkhauser
(1996).
[M*] W.S. Massey, Algebraic Topology: an Introduction, Grad-
uate Texts in Math. 56, Springer-Verlag (1977).
[R*] D. Rolfsen, Knots and Links, Math. Lecture Series 7, Publish
or Perish (1976).
Papers on Knots and Related Topics
[Bi] J.S. Birman, New points of view in knot theory, Bull. Amer.
Math. Soc. 28 (1993) 253-287.
[Br] E.J. Brody, The topological classification of the lens spaces,
Ann. of Math. 71 (1960) 163-184.
[BLl J.S. Birman and X-So Lin, Knot polynomials and Vassiliev's
invariants, Invent. Math. 111 (1993) 225-270.
[BN] D. Bar-Natan, On the Vassiliev knot invariants, Topology 34
(1995) 423-472.
[BN-G] D. Bar-Natan and S Garoufalidis, On the Melvin-Morton-
Rozansky conjecture, preprint, Harvard Univ. (1994).
Bibliography 334
[BS] C. Bankwitz and H.G. Schumann, Uber Viergeflechte, Abh.
Math. Sem. Univ. Hamburg 10 (1934) 263-284.
[C] J.H. Conway, An enumeration of knots and links, and some
of their algebraic properties, Computational Problems in
Abstract Algebra, Pergamon Press (1970) 329-358.
[CG] J.H. Conway and C.McA. Gordon, Knots and links in spatial
graphs, J. Graph Theory 7 (1983) 445-453.
[D] V.G. Drinfel'd, Quantum Groups, Proceedings of the ICM,
Berkeley (1986) 798-820.
[DT] C.H. Dowker and M.B. Thistlethwaite, Classification of knot
projections, Topology and its Applications 16 (1983) 19-31.
[ES1] C. Ernst and D.W. Sumners, The growth of the number of
prime knots, Math. Proc. Cambridge Phil. Soc. 102 (1987)
303-315.
[ES2] C. Ernst and D.W. Sumners, A calculus for rational tangles;
applications to DNA recombination, Math. Proc. Cambridge
Phil. Soc 108 (1990) 489-515.
[F1] R.H. Fox, A quick trip through knot theory, Topology of 3-
manifolds and Related Topics, Prentice-Hall (1962) 120-
167.
[F2] R.H. Fox, Metacyclic invariants of knots and links, Canad. J.
Math. 22 (1970) 193-201.
[FM] R.H. Fox and J.W. Milnor, Singularities of 2-spheres in 4-space
and cobordism of knots, Osaka J. Math. 3 (1966) 257-267.
[G] C.A. Giller, A family of links and the Conway calculus, Trans.
Amer. Math. Soc. 270 (1982) 75-109.
[GL] C.MeA. Gordon and J. Luecke, Knots are determined by their
complements, J. Amer. Math. Soc. 2 (1989) 371-415.
[GLM] C.MeA. Gordon, R.A. Litherland and K. Murasugi, Signatures
of covering links, Canad. J. Math. 33 (1981) 381-394.
[H] F. Hosokawa, On V -polynomials of links, Osaka Math. J. 10
(1958) 273-282.
[Jl] V.F.R. Jones, Hecke algebra representations of braid groups
and link polynomials, Ann. of Math. 126 (1987) 335-388.
[J2] V.F.R. Jones, On knot invariants related to some statistical
mechanical models, Pacific J. Math. 137 (1989) 311-334.
[Kan] T. Kanenobu, Examples on polynomial invariants of knots and
links, Math. Ann. 275 (1986) 555-572.
335 Bibliography
[Kau1] L.H. Kauffman, State models and the Jones polynomial, Topol-
ogy 26 (1987) 395-407.
[Kau2] L.H. Kauffman, An invariant of regular isotopy, Trans. Amer.
Math. Soc. 318 (1990) 417-471.
[Ki] R. Kirby, A calculus for framed links in S3, Invent. Math. 45
(1978) 35-56.
(KM] P. Kronheimer and T. Mrowka, Gauge theory for embedded
surfaces I, Topology 32 (1993) 773-826, II (ibid) 34 (1995)
37-97.
[L] W.B.R. Lickorish, Three-manifolds and the Temperley-Lieb al-
gebra, Math. Ann. 290 (1991) 657-670.
[LM1] W.B.R. Lickorish and K.C. Millett, Some evaluations of link
polynomials, Comment. Math. Helv. 61 (1986) 349-359.
[LM2] W.B.R. Lickorish and K.C. Millett, The reversing formula for
the Jones polynomial, Pacific J. Math. 124 (1986) 173-176.
[LM3] W.B.R. Lickorish and K.C. Millett, A polynomial invariant of
oriented links, Topology 26 (1987) 107-141.
[MelM] P.M. Melvin and H.R. Morton, The coloured Jones function,
Comm. Math. Physics 169 (1995) 501-520.
[Men] W. Menasco, Closed incompressible surfaces in alternating knot
and link complements, Topology 23 (1984) 37-44.
[MenT] W. Menasco and M. Thistlethwaite, The classification of alter-
nating links, Ann. of Math. 138 (1993) 113-171.
[Mi] J.W. Milnor, Infinite cyclic covers, Conference on the Topol-
ogy of Manifolds, Prindle, Weber and Schmidt (1968) 115-
133.
[Mo] H.R. Morton, Seifert circles and knot polynomials, Math. Proc.
Cambridge Phil. Soc. 99 (1986) 107-109.
[Muk] H. Murakami, A recursive calculation of the Arf invariant of a
link, J. Math. Soc. Japan 38 (1986) 335-338.
[Mus1] K. Murasugi, On the genus of the alternating knots I, J. Math.
Soc. Japan 10 (1958) 94-105, II (ibid) 10 (1958) 235-248.
[Mus2] K. Murasugi, On the Alexander polynomial of the alternating
knot, Osaka Math. J. 10 (1958) 181-189.
[Mus3] K. Murasugi, On a certain numerical invariant of link type,
Trans. Amer. Math. Soc. 117 (1965) 387-422.
[Mus4] K. Murasugi, On the signature oflinks, Topology 9 (1970) 283-
298.
Bibliography 336
[Mus5] K. Murasugi, On invariants of graphs with application to knot
theory, Thans. Amer. Math. Soc. 314 (1989) 1-49.
[Mus6] K. Murasugi, On the braid index of alternating links, Thans.
Amer. Math. Soc. 326 (1991) 237-260.
[0] Y. Ohyama, Vassiliev invariants and similarity of knots, Proc.
Amer. Math. Soc. 123 (1995) 287-291.
[ScI] H. Schubert, Die eindeutige Zerlegbarkeit eines Knotens in
Primknoten, Sitzungsber Heidelberger Akad. Wiss. Math.
Nat. KI 3 (1949) 57-104.
[Sc2] H. Schubert, Uber eine numerische Knoteninvariante, Math.
Zeit. 61 (1954) 245-288.
[Sc3] H. Schubert, Knoten mit zwei Briicken, Math. Zeit. 65 (1956)
133-170.
[Stl T. Stanford, Braid commutators and Vassiliev invariants, to
appear in Pacific J. Math.
[Th] H.F. Trotter, Homology of group systems with applications to
knot theory, Ann. of Math. 76 (1962) 464-498.
[Tu] V.G. Turaev, The Yang-Baxter equation and invariants oflinks,
Invent. Math. 92 (1988) 527-553.
[V] V.A. Vassiliev, Cohomolgy of knot spaces, Theory of Singu-
larities and its Applications, Amer. Math. Soc. (1990)
23-69.
(Wa] J.e. Wang, DNA topoisomerases, Scientific American 247
(1982) 94-109.
[Wi] E. Witten, Quantum field theory and Jones polynomial, Comm.
Math. Physics 121 (1989) 351-399.
[WAD] M. Wadati, Y. Akutsu and T. Deguchi, Knot theory based on
solvable models at criticality, Integrable systems in quan-
tum field theory and statistical mechanics, Academic
Press (1989) 193-285.
(Y] S. Yamada, An invariant of spatial graphs, J. Graph Theory 13
(1989) 537-551.
2-bridge knot, 182, 183, 187, 188-190,
191,195,242,247,273,277
linking number of --, 196
oriented ,194,196
regular diagram of --, 187, 191
signature of , 196
3-.<olourable, 70
3-rrwnifold, 153, 154,156,158,266
4-term formula, 313
(4
achiral
rigidly , 296, 298
topologically --, 296, 298
Actuality Table, 317, 318, 320
Alexander, 4, 75
--'8 theorem, 166, 209
--- Conway polynomial, 110,
117,247,308,310
--polynomial, 76, 107-108,
112, 117-122, 142-143, 170,
220,241,242,266,323
skein relation of-- , 232
algebraic
--geometry, 4
--link, 194
--number theory, 4
--topology,9, 152
alternating
--diagram, 29, 38, 44
--knot, 29, 188,241-242,245
--link, 29
amphicheiral, 43, 128, 145, 224, 226, 245
arm, 184
Artin,197
B
Bing's conjecture, 290
bipartite, 79
Boltzmann weights, 250-251, 262
Borromean rings, 17,33, 141
braid(s), 198, 214, 218, 247, 257
closed ,209
partition function of--, 255
product of , 201-202
--group, 203, 207, 218, 261
-generators, 205, 214
-presentation, 207, 214
-----index, 214, 215,247, 313
--permutation, 201
--relations, 206-207
bridge number, 58-59, 150, 183,215,313
e
chiral, 296
chord diagram, 317
chromatic polynomial, 286-287
circle, 5
clover-leaf knot, see trefoil knot
cobordant, 23
code, see knot
colouring number set, 73
complement, see knot
complex, 329
continued fraction, 179
contributing state(s), 257
Conway polynomial, 110, 113
covering map, 160
covering space, 159, 160
cyclic --, 163, 164, 166
branch --, 162
n-fold --, 166
crossing multiplier, 254
crossing points,
minimum number of --, 56-57,
150,313
Index
"
degree of a polynomial, 243
Dehn surgery, 156, 158
denominator, 174,183,191
detenrrrinant, 105, 191
--of a knot, 105
diagram
regular , 27, 28, 37,47,
62,173,187,235
equivalence of --, 50
reduced ,28
direct repeat, 272
DNA
--knot, 273
--link, 273
--molecule(s), 4,267,269
t
elementary knot moves(s), 6, 7
oriented ,9
elementary symmetric matrix operator, 90
enzymology
topological approach to--, 268
equivalence of knots see knot equivalence
Euler characteristic, 81
exactly solvable model, 249
,
figure 8 knot, 4, 24,43, 115, 225, 276
flat
--disk, 302
--transfonnation, 302
flype,246
f
Gauss, 3, 30
genus, 80
-- of a knot, 81, 121, 149, 241
-- of a Seifert surface, 81
global move/transfonnation, 9
GIQbal problem, 40-43, 116, 117
granny knot, 22
338
graph(s), 35, 36, 193, 194, 284-285
4-regular , 301
abstract , 285
complete , 289
colouring of , 286,
colouring number of --, 288
dual ,292
equivalence of , 291
isomorphic , 285
knot see knot
planar ,292
plane see plane
spatial , 285,
289,291,302
1#
homeomorphism, 10
HOMFLYpolynomial, 232, 247,311
Hopf link, 59
Hosokawa polynomial, 120-121
J
incidence relation, 284
inverted repeat, 272
isotopy,13
I
Jordan curve theorem, 32, 331
Jones, 4
--polynomial, 219, 223, 240
264,296,310,311,332
Nth degree , 265,
311,323
skein relation of -, 232
---type invariant, 261, 300
Il
Kauffman
--bracket polynomial, 234,311
--(2-variable) polynomial, 238
--principle, 233, 259
Kinoshita a-curve, 291
339
Kinoshita-Terasaka knot, 119, 226
knot(s), 1, 5
cobordism group of - , 23
code ofa ,32
connected sum---, see sum of
determinant of a-- , 105
genus of a ,81, 121, 149
graph of a , 36
invertible , 44, 195
nullity of a , 126
oriented , 194, 196
period of a , 45
rational , 183
signature of a , 126, 129,
131,145,148,192,196,312
sum of , 17, 20, 22,
24,29,57,60,120,127,130,223,302
--complement, 14, 41
--decomposition, 17, 19, 20
--equivalence, 3, 8, 11, 50, 329
- with orientation, 9
--invariant, 42, 50
--projection, 26, 27
--semi-group, 22
--tables, 25, 30
--theory, 3
1
L-polynomial, 76
Lens space, 158, 169
linking number in -,332
link(s), 15
code ofa ,33
equivalence of a --, 15
linking number, 64--66, 196,268,269,332
totw ,68,131
Listing,3
--knot, 4
Little, 25, 32, 37
Local problems, 40, 43-45
longitude, 140, 155, 157
loop, 1,285
Index
II
M-equivalence, 213-214
manifold, 153, 154, 156
Markov moves, 213, 214
mathematical physics, 4
meridian, 139, 155, 157
mirror image, 11,24,43, 72, 103, 127,224
model
exactly solvable --, 249, 251
IRF ,249
vertex ,249
"nullity, 124, see also knot
numerator, 174, 190, 191, 274-275
,
p-colourable, 73
partition function, 249, 253, 257, 259
period, see knot
Perko's pair, 8
permutation, 31
braid------ , 201
plane graph, 35
connected---- , 38
signed , 36, 37
Poincare's conjecture, 290
prime knot, 19,44,242
projection, see also knot
regular -----, 27
Q
quantum group, 249, 252
R
R-matrix, 251
R-move(s),51
regular diagram, see diagram
regular equivalence, 233
regular moves, 233
regular projection, see projection
Reidcmcister moves, 48, 50, 90
Index
Il continued
repeat
direct -----, 272
inverted , 272
S
S-equivalent, 91, 93, 100
S-matrix, see R-matrix
Schonflies theorem, 330
Seifert, 4
--curves, 77
--graph, 79, 83, 101
--matrix, 76, 85-89, 91, 101,
103, 141-142
determinant of a - , 105
--surface, 79
genus of a-- ,81
signature, 124, see also knot
simple closed curve, 8
simplex, 329
singular knot(s), 301, 307
equivalence of , 303
site-specific recombination, 271
skein
--diagram, 109
--invariants, 231
--operation, 109
--polynomial, 232, 247,265,311
--relation, 109,220,232
--tree diagram, 113-115, 307
slice knot, 23, 24, 46, 131
Smith's conjecture, 4
splicing, 77
split link, 119
square knot, 22, 24
state variable, 250, 256
substrate, 272
synaptic complex, 272
1
Tait, 25, 34, 37
--conjectures, 241, 244, 245, 246
340
Tait number, 68, 233, 245, 264
tangle(s), 18, 171, 173
(1,1)- , 18, 172
trivial (1,1)-- , 18
(n,n)- ,18,172, 198
wgebrmc ,182
equivalence --, 174
exceptional , 176
fraction of , 180
rational ,176
regulardiagramof - ,173
trivial , 176, 178,
180, see also triviw (1, I)-tangle
Tn3 Resolvase, 276, 282
topoisomers, 271
topoisomerase, 268, 270, 271
topological space, 9
topology,
3-dimensional -- ,4
algebraic , 9, 152
combinatoriw-- , 5, 329
torus, 133
solid , 138-139
trivial , 133
--knot, 133, 135, 138, 148,
183,214,216,242
oriented , 137
regular diagram of - ,136
trefoil knot, 2, 24, 43, 57, 59, 115, 225
trivial colouring, 73
trivial knot, 2, 29, 57, 59, 192,321
trivial link, 16, 112, 220
tubular neigbourhood, 154
twist, 13
twisting number, 270
U
unitary condition, 252
unknotting number, 63-64, 130-131,
150-151,193,313
unknotting operation, 62
unknotted knot, see trivial knot
341
,
Vassiliev invariant, 304, 307
--of finite type, 305
--of order 0, 306
--of order 1, 307
--of order (at most) m, 305
--of order (exactly) m, 305
Vassiliev type invariant, 305
ft1
Whitehead link, 17, 141
wild knot(s), 5, 6
writhe, 68, 233, 245, 264
--in mathematical biology, 270
writhing process, 272
V
Yamada polynomial, 295
Yang-Baxter
--equation, 250, 251,252,255
--operator, 261
enhanced--, 263
Index