Explore the full ebook collection and download it now at textbookfull.

com

Correspondence Analysis in Practice, Third Edition

Greenacre

https://textbookfull.com/product/correspondence-analysis-in-
practice-third-edition-greenacre/

OR CLICK HERE

DOWLOAD EBOOK

Browse and Get More Ebook Downloads Instantly at https://textbookfull.com

Click here to visit textbookfull.com and download textbook now
 Your digital treasures (PDF, ePub, MOBI) await
Download instantly and pick your perfect format...

Read anywhere, anytime, on any device!

Compositional data analysis in practice First Edition

Greenacre

https://textbookfull.com/product/compositional-data-analysis-in-
practice-first-edition-greenacre/

textbookfull.com

Biota Grow 2C gather 2C cook Loucas

https://textbookfull.com/product/biota-grow-2c-gather-2c-cook-loucas/

textbookfull.com

The George Bell Gerhard Leibholz Correspondence In the

Long Shadow of the Third Reich 1938 1958 Andrew Chandler

https://textbookfull.com/product/the-george-bell-gerhard-leibholz-
correspondence-in-the-long-shadow-of-the-third-reich-1938-1958-andrew-
chandler/
textbookfull.com

Structural Analysis Third Edition Coates

https://textbookfull.com/product/structural-analysis-third-edition-
coates/

textbookfull.com
The Practice Manager Third Edition Drury

https://textbookfull.com/product/the-practice-manager-third-edition-
drury/

textbookfull.com

The Strauss-Krüger Correspondence Susan Meld Shell

https://textbookfull.com/product/the-strauss-kruger-correspondence-
susan-meld-shell/

textbookfull.com

Applied Functional Analysis, Third Edition John Tinsley

Oden

https://textbookfull.com/product/applied-functional-analysis-third-
edition-john-tinsley-oden/

textbookfull.com

Modeling and Analysis of Dynamic Systems, Third Edition

Esfandiari

https://textbookfull.com/product/modeling-and-analysis-of-dynamic-
systems-third-edition-esfandiari/

textbookfull.com

Pavement Engineering: Principles and Practice, Third

Edition El-Korchi

https://textbookfull.com/product/pavement-engineering-principles-and-
practice-third-edition-el-korchi/

textbookfull.com
 Correspondence
Analysis in Practice
Third Edition
CHAPMAN & HALL/CRC
Interdisciplinar y Statistics Series
Series editors: N. Keiding, B.J.T. Morgan, C.K. Wikle, P. van der Heijden
Published titles
AGE-PERIOD-COHORT ANALYSIS: NEW MODELS, METHODS, AND
EMPIRICAL APPLICATIONS Y. Yang and K. C. Land
ANALYSIS OF CAPTURE-RECAPTURE DATA R. S. McCrea and B. J.T. Morgan
AN INVARIANT APPROACH TO STATISTICAL ANALYSIS OF SHAPES
S. Lele and J. Richtsmeier
ASTROSTATISTICS G. Babu and E. Feigelson
BAYESIAN ANALYSIS FOR POPULATION ECOLOGY R. King, B. J.T. Morgan,
O. Gimenez, and S. P. Brooks
BAYESIAN DISEASE MAPPING: HIERARCHICAL MODELING IN SPATIAL
EPIDEMIOLOGY, SECOND EDITION A. B. Lawson
BIOEQUIVALENCE AND STATISTICS IN CLINICAL PHARMACOLOGY
S. Patterson and B. Jones
CLINICAL TRIALS IN ONCOLOGY,THIRD EDITION S. Green, J. Benedetti, A. Smith,
and J. Crowley
CLUSTER RANDOMISED TRIALS R.J. Hayes and L.H. Moulton
CORRESPONDENCE ANALYSIS IN PRACTICE,THIRD EDITION M. Greenacre
DESIGN AND ANALYSIS OF QUALITY OF LIFE STUDIES IN CLINICAL TRIALS,
SECOND EDITION D.L. Fairclough
DYNAMICAL SEARCH L. Pronzato, H. Wynn, and A. Zhigljavsky
FLEXIBLE IMPUTATION OF MISSING DATA S. van Buuren
GENERALIZED LATENT VARIABLE MODELING: MULTILEVEL, LONGITUDI-
NAL, AND STRUCTURAL EQUATION MODELS A. Skrondal and S. Rabe-Hesketh
GRAPHICAL ANALYSIS OF MULTI-RESPONSE DATA K. Basford and J. Tukey
INTRODUCTION TO COMPUTATIONAL BIOLOGY: MAPS, SEQUENCES, AND
GENOMES M. Waterman
MARKOV CHAIN MONTE CARLO IN PRACTICE W. Gilks, S. Richardson, and
D. Spiegelhalter
MEASUREMENT ERROR ANDMISCLASSIFICATION IN STATISTICS AND EPIDE-
MIOLOGY: IMPACTS AND BAYESIAN ADJUSTMENTS P. Gustafson
MEASUREMENT ERROR: MODELS, METHODS, AND APPLICATIONS
J. P. Buonaccorsi
Published titles
MEASUREMENT ERROR: MODELS, METHODS, AND APPLICATIONS
J. P. Buonaccorsi
MENDELIAN RANDOMIZATION: METHODS FOR USING GENETIC VARIANTS
IN CAUSAL ESTIMATION S.Burgess and S.G. Thompson
META-ANALYSIS OF BINARY DATA USINGPROFILE LIKELIHOOD D. Böhning,
R. Kuhnert, and S. Rattanasiri
MISSING DATA ANALYSIS IN PRACTICE T. Raghunathan
POWER ANALYSIS OF TRIALS WITH MULTILEVEL DATA M. Moerbeek and
S. Teerenstra
SPATIAL POINT PATTERNS: METHODOLOGY AND APPLICATIONS WITH R
A. Baddeley, E Rubak, and R. Turner
STATISTICAL ANALYSIS OF GENE EXPRESSION MICROARRAY DATA T. Speed
STATISTICAL ANALYSIS OF QUESTIONNAIRES: A UNIFIED APPROACH
BASED ON R AND STATA F. Bartolucci, S. Bacci, and M. Gnaldi
STATISTICAL AND COMPUTATIONAL PHARMACOGENOMICS R. Wu and M. Lin
STATISTICS IN MUSICOLOGY J. Beran
STATISTICS OF MEDICAL IMAGING T. Lei
STATISTICAL CONCEPTS AND APPLICATIONS IN CLINICAL MEDICINE
J. Aitchison, J.W. Kay, and I.J. Lauder
STATISTICAL AND PROBABILISTIC METHODS IN ACTUARIAL SCIENCE
P.J. Boland
STATISTICAL DETECTION AND SURVEILLANCE OF GEOGRAPHIC CLUSTERS
P. Rogerson and I.Yamada
STATISTICS FOR ENVIRONMENTAL BIOLOGY AND TOXICOLOGY A. Bailer
and W. Piegorsch
STATISTICS FOR FISSION TRACK ANALYSIS R.F. Galbraith
VISUALIZING DATA PATTERNS WITH MICROMAPS D.B. Carr and L.W. Pickle
 Ch ap ma n & Hall/CRC
I n t e rd is ci pl in ar y Statistics Series

Correspondence
Analysis in Practice
Third Edition

Michael Greenacre
Universitat Pompeu Fabra
Barcelona, Spain
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2017 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed on acid-free paper

Version Date: 20161031

International Standard Book Number-13: 978-1-4987-3177-5 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and
information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission
to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic,
mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or
retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact
the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides
licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment
has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without
intent to infringe.

Library of Congress Cataloging‑in‑Publication Data

Names: Greenacre, Michael J.

Title: Correspondence analysis in practice / Michael Greenacre.
Description: Third edition. | Boca Raton, Florida : CRC Press, [2017] |
Series: Interdisciplinary statistics | Includes bibliographical references
and index.
Identifiers: LCCN 2016036265| ISBN 9781498731775 (hardback : alk. paper) |
ISBN 9781498731782 (e‑book)
Subjects: LCSH: Correspondence analysis (Statistics)
Classification: LCC QA278.5 .G74 2017 | DDC 519.5/37‑‑dc23
LC record available at https://lccn.loc.gov/2016036265

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com
To Françoise, Karolien and Gloudina
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
1 Scatterplots and Maps . . . . . . . . . . . . . . . . . . . . . . 1
2 Profiles and the Profile Space . . . . . . . . . . . . . . . . . 9
3 Masses and Centroids . . . . . . . . . . . . . . . . . . . . . . 17
4 Chi-Square Distance and Inertia . . . . . . . . . . . . . . . 25
5 Plotting Chi-Square Distances . . . . . . . . . . . . . . . . . 33
6 Reduction of Dimensionality . . . . . . . . . . . . . . . . . . 41
7 Optimal Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . 49
8 Symmetry of Row and Column Analyses . . . . . . . . . . 57
9 Two-Dimensional Displays . . . . . . . . . . . . . . . . . . . 65
10 Three More Examples . . . . . . . . . . . . . . . . . . . . . . 73
11 Contributions to Inertia . . . . . . . . . . . . . . . . . . . . . 81
12 Supplementary Points . . . . . . . . . . . . . . . . . . . . . . 89
13 Correspondence Analysis Biplots . . . . . . . . . . . . . . . 97
14 Transition and Regression Relationships . . . . . . . . . . 105
15 Clustering Rows and Columns . . . . . . . . . . . . . . . . . 113
16 Multiway Tables . . . . . . . . . . . . . . . . . . . . . . . . . . 121
17 Stacked Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
18 Multiple Correspondence Analysis . . . . . . . . . . . . . . 137
19 Joint Correspondence Analysis . . . . . . . . . . . . . . . . 145
20 Scaling Properties of MCA . . . . . . . . . . . . . . . . . . . 153
21 Subset Correspondence Analysis . . . . . . . . . . . . . . . 161
22 Compositional Data Analysis . . . . . . . . . . . . . . . . . . 169
23 Analysis of Matched Matrices . . . . . . . . . . . . . . . . . 177
24 Analysis of Square Tables . . . . . . . . . . . . . . . . . . . . 185
25 Correspondence Analysis of Networks . . . . . . . . . . . . 193
26 Data Recoding . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
27 Canonical Correspondence Analysis . . . . . . . . . . . . . 209
28 Co-Inertia and Co-Correspondence Analysis . . . . . . . . 217
29 Aspects of Stability and Inference . . . . . . . . . . . . . . 225
30 Permutation Tests . . . . . . . . . . . . . . . . . . . . . . . . 233
Appendix A: Theory of Correspondence Analysis . . . . . . . 241
Appendix B: Computation of Correspondence Analysis . . . 255
Appendix C: Glossary of Terms . . . . . . . . . . . . . . . . . . 285
Appendix D: Bibliography of Correspondence Analysis . . . 291
Appendix E: Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . 295
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
ix
Preface

This book is a revised and extended third edition of the second edition of
Correspondence Analysis in Practice (Chapman & Hall/CRC, 2007), first pub-
lished in 1993. In the original first edition I wrote the following in the Preface,
which is still relevant today:
“Correspondence analysis is a statistical technique that is useful to all students, Extract from
researchers and professionals who collect categorical data, for example data col- preface of first
lected in social surveys. The method is particularly helpful in analysing cross- edition of
tabular data in the form of numerical frequencies, and results in an elegant but Correspondence
simple graphical display which permits more rapid interpretation and under- Analysis in Practice
standing of the data. Although the theoretical origins of the technique can be
(1993)
traced back over 50 years, the real impetus to the modern application of corre-
spondence analysis was given by the French linguist and data analyst Jean-Paul
Benzécri and his colleagues and students, working initially at the University of
Rennes in the early 1960s and subsequently at the Jussieu campus of the Univer-
sity of Paris. Parallel developments of correspondence analysis have taken place
in the Netherlands and Japan, centred around such pioneering researchers as
Jan de Leeuw and Chikio Hayashi. My own involvement with correspondence
analysis commenced in 1973 when I started my doctoral studies in Benzécri’s
Data Analysis Laboratory in Paris. The publication of my first book Theory and
Applications of Correspondence Analysis in 1984 coincided with the beginning
of a wider dissemination of correspondence analysis outside of France. At that
time I expressed the hope that my book would serve as a springboard for a much
wider and more routine application of correspondence analysis in the future. The
subsequent evolution and growing popularity of the method could not have been
more gratifying, as hundreds of researchers were introduced to the method and
became familiar with its ability to communicate complex tables of numerical
data to non-specialists through the medium of graphics. Researchers with whom
I have collaborated come from such varying backgrounds as sociology, ecology,
palaeontology, archaeology, geology, education, medicine, biochemistry, microbi-
ology, linguistics, marketing research, advertising, religious studies, philosophy,
art and music. In 1989 I was invited by Jay Magidson of Statistical Innovations
Inc. to collaborate with Leo Goodman and Clifford Clogg in the presentation of a
two-day short course in New York, entitled “Correspondence Analysis and Asso-
ciation Models: Geometric Representation and Beyond”. The participants were
mostly marketing professionals from major American companies. For this course
I prepared a set of notes which reinforced the practical, user-oriented approach to
correspondence analysis. ... The positive reaction of the audience was infectious
and inspired me subsequently to present short courses on correspondence analysis
in South Africa, England and Germany. It is from the notes prepared for these
courses that this book has grown.”

In 1991 Prof. Walter Kristof of Hamburg University proposed that we orga- The CARME
nize a conference on correspondence analysis, with the assistance of Dr. Jörg conferences
Blasius of the Zentralarchiv für Empirische Sozialforschung (Central Archive
xi
xii Preface

for Empirical Social Research) at the University of Cologne. This confer-

ence was the first international one of its kind and drew a large audience
to Cologne from Germany and neighbouring European countries. This ini-
tial meeting developed into a series of quadrennial conferences, repeated in
1995 and 1999 in Cologne, at the Pompeu Fabra University in Barcelona
(2003), the Erasmus University in Rotterdam (2007), Agrocampus Rennes
(2011) and the University of Naples Federico II (2015). The 1991 conference
led to the publication of the book Correspondence Analysis in the Social Sci-
ences, while the 1995 conference gave birth to another book, Visualization
of Categorical Data, both of which received excellent reviews. For the 1999
conference on Large Scale Data Analysis, participants had to present analyses
of data from the multinational International Social Survey Programme (ISSP
— see www.issp.org). This interdisciplinary meeting included presentations
not only on the latest methodological developments in survey data analysis
but also topics as diverse as religion, the environment and social inequality.
In 2003 we returned to the original theme for the Barcelona conference, which
was baptized with the Catalan girl’s name CARME, standing for Correspon-
dence Analysis and Related MEthods; hence the formation of the CARME
network (www.carme-n.org). This led once more to Jörg Blasius and myself
editing a third book, Multiple Correspondence Analysis and Related Methods,
which was published in 2006. As with the two previous volumes, our idea was
to produce a multi-authored book, inviting experts in the field to contribute.
Our editing task was to write the introductory and linking material, unifying
the notation and compiling a common reference list and index. As a result of
the Rennes conference in 2011, which celebrated 50 years of correspondence
analysis, the book Visualization and Verbalization of Data was published, half
of which is devoted to the history of multivariate analysis. These books mark
the pace of development of correspondence analysis, at least in the social sci-
ences, and are highly recommended to anyone interested in deepening their
knowledge of this versatile statistical method as well as methods related to it.

New material in I have been very gratified to be invited to prepare a new edition of Correspon-
third edition dence Analysis in Practice, having accumulated considerably more experience
in social and environmental research in the nine years since the publication of
the second edition. Apart from revising the existing chapters, five new chapters
have been added, on “Compositional Data Analysis” (an area highly related
to correspondence analysis), “Analysis of Matched Matrices” (joint analysis
of data tables with the same rows and columns), “Correspondence Analysis of
Networks” (applying correspondence analysis to graphs), “Co-Inertia and Co-
Correspondence Analysis” (analysis of relationships between two tables with
common rows), and “Permutation Tests” (performing statistical inference in
the context of correspondence analysis and related methods). All in all, I can
say that this third edition contains almost all my practical knowledge of the
subject, after more than 40 years working in this area.
Preface xiii

At a conference I attended in the 1980s, I was given this lapel button with its “Statisticians
nicely ambiguous maxim, which could well be the motto of correspondence count!”
analysts all over the world:

To illustrate the more technical meaning of this motto, and to give an initial Textual analysis of
example of correspondence analysis, I made a count of the most frequent words third edition
in each of the 30 chapters of this new edition. I had to aggregate variations of
the same word, e.g. “coordinate” and “coordinates”, “plot” and “plotting”,
a process called lemmatization in textual data analysis. The top 10 most
frequent words were, in descending order of frequency: “row/s”, “profile/s”,
“inertia” (which is the way correspondence analysis measures variance in a
table), “point/s”, “column/s”, “data”, “CA” (abbreviation for correspondence
analysis), “variable/s”, “value/s” and “average”. I omitted words that occur
in one chapter only, such as “fuzzy” and “degree”, which are specific to a
single chapter, and removed words that described particular applications. This
left an eventual total of 167 words, which can be regarded as reflecting the
methodological content of the book.

Exhibit 0.1:
analyse/sis association/s asymmetric average axis/es ... First few rows and
columns of the table
Chap 1 10 0 0 0 15 ... of counts of the 167
Chap 2 0 0 0 29 22 ... most frequent words
Chap 3 0 0 0 55 0 ... in the 30 chapters of
Chap 4 0 6 0 22 0 ... Correspondence
Chap 5 0 0 0 22 13 ... Analysis in Practice,
Chap 6 0 0 0 8 0 ... Third Edition,
Chap 7 0 0 0 29 0 ... visualized in Exhibit
Chap 8 47 0 0 14 20 ... 0.2 using
correspondence
Chap 9 0 0 14 6 32 ...
analysis.
Chap 10 0 0 17 0 14 ...
.. .. .. .. .. ..
. . . . . . ...

Total 369 12 39 370 277 ...

xiv Preface

Exhibit 0.2: permutation

test
Correspondence
profile/s
analysis display of
30 chapters of the

0.2
present book in
distribution
terms of the most
p-value 30
frequent words in significant/ce
distance/s statistic bootstrap sample/ing
each chapter. row/s hypothesis/es
0.1
frequencies
weighted 29
Numbers in boldface point/s
column/s 5
data pairs
CA dimension 2

indicate the project/ion 23 4

axis/es categorical
vertex/ices 69 15 original
positions of the 13 12 110 interaction/ive
chapters, and their 1411
0.0

8 7 22 16
proximity signifies sets
2627
relative similarity of 21 1723
24
word distribution. 20 28
Directions of the 2518 response/svariable/s
−0.1

19 CA
respondent/s dimension/s
words give the weights
question/s
interpretation for subtables
analysis/analysed
the positioning of indicator
Burt
the chapters.
−0.2

MCA
Technically, this is a diagonal
so-called “contribu-
tion biplot” (see
Chapter 13).
−0.3

matrix/ices

−0.2 −0.1 0.0 0.1 0.2 0.3 0.4

CA dimension 1

The final table of word counts was composed of the 30 chapters as rows,
and the 167 words as columns (see Exhibit 0.1). This table is very sparse,
i.e. it has many zeros. In fact, 80% of the cells of the table have no counts.
Correspondence analysis copes quite well with such data, which has made it a
popular method in research areas such as linguistics, archaeology and ecology,
where data sets of frequency counts occur that are very sparse.
Exhibit 0.2 shows the “map” of the table, resulting from applying correspon-
dence analysis. The first thing to notice is that the rows (chapters, indicated
by their numbers) and the columns (words, connected to the centre by lines)
are displayed with respect to two “dimensions”. These dimensions are de-
termined by the analysis with the objective of exposing the most important
features of the associations between chapters and words. An alternative way
of thinking about this is that the chapters are mapped according to the simi-
larity in their distributions of words, with closer chapters being more similar
and distant chapters more different. Then the directions of the words explain
the differences between the chapters. Not all the words are shown, because
about two-thirds of the words turn out to be not so important for the interpre-
tation of the result, so only those words are shown that contribute highly to
the positioning of the chapters. Without further explanation of the concepts
Preface xv

underlying correspondence analysis (after all, this is the aim of the book that
follows!) the map clearly shows three sets of words emanating from the centre.
The words out to the top right clearly distinguish Chapters 29 and 30 from all
the others — these are the chapters that concentrate on the sampling, distri-
butional and inferential properties of correspondence analysis, with main key-
words “permutation” and “test”. Chapter 15 on clustering also tends in that
direction because it contains some hypothesis testing. Out in the upper left
direction are all the words describing basic concepts and terminology of corre-
spondence analysis associated with Chapters 1–14 that introduce the method
and develop it, exemplified by the most prominent keyword “profile/s”. To-
wards the bottom of the map are the words associated with a generalization of
correspondence analysis, called multiple correspondence analysis, usually ap-
plied to questionnaire data, described in later chapters. This method involves
various coding schemes in different types of matrices, hence the important
keyword “matrix/ices” down below.
Like the second edition, the book maintains its didactic format, with exactly Format of third
eight pages per chapter to provide a constant amount of material in each edition
chapter for self-learning or teaching (a feature that has been commented on
favourably in book reviews of the second edition). One of my colleagues re-
marked that it was like writing 14-line sonnets with strict rules for metre
and rhyming, which was certainly true in this case: the format definitely con-
tributed to the creative process. The margins are reserved for section headings
as well as captions of the tables and figures — these captions tend to be more
informative than conventional one-liner ones. Each chapter has a short intro-
duction and its own “Contents” list on the first page, and the chapter always
ends with a summary in the form of a bulleted list.
As in the first and second editions, the book’s main thrust is towards the Appendices
practice of correspondence analysis, so most technical issues and mathemati-
cal aspects are gathered in a theoretical appendix at the end of the book. It is
followed by a computational appendix, which describes some features of the R
language relevant to the methods in the book, including the ca package for cor-
respondence analysis. R scripts are placed on the website www.carme-n.org,
along with several of the data sets. No references at all are given in the 30
chapters — instead, a brief bibliographical appendix is given to point readers
towards further readings and more complete literature sources. A glossary of
the most important terms in the book is also provided and the book concludes
with some personal thoughts in the form of an epilogue.
The first edition of this book was written in South Africa, and the second and Acknowledgements
present third editions in Catalonia, Spain. Many people and institutions have
contributed in one way or another to this project. I would like to thank the
BBVA Foundation in Madrid and its director Prof. Rafael Pardo, for sup-
port and encouragement in my work on correspondence analysis. The BBVA
Foundation has published a Spanish translation of the second edition of Cor-
respondence Analysis in Practice, called La Práctica del Análisis de Corre-
spondéncias, available for free download at www.multivariatestatistics.org.
xvi Preface

I owe a similar debt of gratitude to my colleagues and the institution of the

Universitat Pompeu Fabra in Barcelona, where I have been working since
1994, one of the most innovative universities in Europe, recently ranked the
most productive Spanish university after only 25 years of its existence.
I would like to thank all my friends and colleagues in many countries for
moral and intellectual support, especially my wife Zerrin Aşan Greenacre,
Jörg & Beate Blasius, Trevor & Lynda Hastie, Carles Cuadras, John Gower,
Jean-Paul Benzécri, Angelos Markos, Alfonso Iodice d’Enza, Patrick Groe-
nen, Pieter Kroonenberg, Cajo ter Braak, Jan de Leeuw, Ludovic Lebart,
Jean-Marie & Annick Monget, Pierre & Martine Teillard, Michael Meimaris,
Michael Friendly, Antoine de Falguerolles, John Aitchison, Michael Browne,
Cas Crouse, Fred Lombard, June Juritz, Francesca Little, Karl Jöreskog, Les-
ley Andres, Barbara Cottrell, Raul Primicerio, Michaela Aschan, Salve Dahle,
Stig Falk-Petersen, Reinhold Fieler, Sabine Cochrane, Paul Renaud, Haakon
Hop, Tor & Danielle Korneliussen, Yasemin El-Menouar, Ekkehard & Ingvill
Mochmann, Maria Rohlinger, Antonella Curci, Gianna Mastrorilli, Paola Bor-
dandini, Oleg Nenadić, Walter Zucchini, Thierry Fahmy, Kimmo Vehkalahti,
Öztaş Ayhan, Simo Puntanen, George Styan, Juha Alho, François Theron,
Volker Hooyberg, Gurdeep Stephens & Pascal Courty, Antoni Bosch & Helena
Trias, Teresa Garcia-Milà, Tamara Djermanovic, Guillem Lopez, Xavier Cal-
samiglia, Andreu Mas-Colell, Xavier Freixas, Frederic Udina, Albert Satorra,
Jan Graffelman & Nuria Satorra, Rosemarie Nagel, Anna Espinal, Carolina
Chaya, Carlos Pérez, Moya Berry, Alan Griffiths, Dianne Fortescue, Tasos &
Androula Ladikos, Jerry & Mary Ann Reedy, Bodo & Bärbel Bilinski, Andries
& Gaby Claassens, Judy Twycross, Romà Revelles & Carme Clotet of Niu Nou
restaurant, Santi Careta, Marta Andreu, Rita Lugli & Danilo Guaitoli, José
Penalva & Nuria Serrano, Gabor Lugosi and the whole community of Gréixer
in the Pyrenees — you have all played a part in this story!
I also fondly remember dear friends and colleagues who have influenced my ca-
reer, but who have sadly passed away in recent years: Paul Lewi, Cas Troskie,
Dan Bradu, Reg & Kay Griffiths, Leo & Wendy Theron, Tony Brink, Jan
Visser, Victor Thiessen, Al McCutcheon and Ingram Olkin, as well as Ruben
Gabriel, from whom I learnt such a lot about statistics and life.
Particular thanks go to Angelos Markos and Antoine de Falguerolles for valu-
able comments on parts of the manuscript, and to Oleg Nenadić for his con-
tinuing collaboration in the development of our ca package in R.
Like the first and second editions, I have dedicated this book to my three
daughters, Françoise, Karolien and Gloudina, who never cease to amaze me
by their joy, sense of humour and diversity.
Finally, I thank the commissioning editor, Rob Calver, as well as Rebecca
Davies and Karen Simon of Chapman & Hall/CRC Press, for placing their
trust in me and for their constant cooperation in making this third edition of
Correspondence Analysis in Practice become a reality.
Michael Greenacre
Barcelona
Scatterplots and Maps 1
Correspondence analysis is a method of data analysis for representing tabu-
lar data graphically. Correspondence analysis is a generalization of a simple
graphical concept with which we are all familiar, namely the scatterplot. The
scatterplot is the representation of data as a set of points with respect to
two perpendicular coordinate axes: the horizontal axis often referred to as
the x-axis and the vertical one as the y-axis. As a gentle introduction to the
subject of correspondence analysis, it is convenient to reflect for a short time
on our perception of scatterplots and how we interpret them in relation to the
data they represent graphically. Particular emphasis will be placed on how we
interpret distances between points in a scatterplot and when scatterplots can
be seen as a spatial map of the data.

Contents
Data set 1: My travels . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Continuous variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Expressing data in relative amounts . . . . . . . . . . . . . . . . . . . . 2
Categorical variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Ordering of categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Distances between categories . . . . . . . . . . . . . . . . . . . . . . . . 3
Distance interpretation of scatterplots . . . . . . . . . . . . . . . . . . . 3
Scatterplots as maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Calibration of a direction in the map . . . . . . . . . . . . . . . . . . . . 4
Information-transforming nature of the display . . . . . . . . . . . . . . 4
Nominal and ordinal variables . . . . . . . . . . . . . . . . . . . . . . . . 5
Plotting more than one set of data . . . . . . . . . . . . . . . . . . . . . 5
Interpreting absolute or relative frequencies . . . . . . . . . . . . . . . . 6
Describing and interpreting data, vs. modelling and statistical inference 7
Large data sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
SUMMARY: Scatterplots and Maps . . . . . . . . . . . . . . . . . . . . 8

During the original writing of this book, I was reflecting on the journeys I Data set 1: My
had made during the year to Norway, Canada, Greece, France and Germany. travels
According to my diary I spent periods of 18 days in Norway, 15 days in
Canada and 29 days in Greece. Apart from these longer trips I also made
several short trips to France and Germany, totalling 24 days. This numerical
description of my time spent in foreign countries can be visualized in the
graphs of Exhibit 1.1. This seemingly trivial example conceals several issues
that are relevant to our perception of graphs of this type that represent data
with respect to two coordinate axes, and which will eventually help us to
1
2 Scatterplots and Maps

Exhibit 1.1: 40% 40%

Graphs of number of 30 30
days spent in foreign 30% 30%
countries in one
Days

Days
year, in scatterplot 20 20
20% 20%
and bar-chart
formats respectively.
10 10
A percentage scale, 10% 10%

expressing days
relative to the total
of 86 days, is given Norway Canada Greece France/Germany Norway Canada Greece France/Germany
on the right-hand
side of each graph.

understand correspondence analysis. Let me highlight these issues one at a

time.

Continuous The left-hand vertical axis labelled Days represents the scale of a numeric piece
variables of information often referred to as a continuous, or numerical, variable. The
scale on this axis is the number of days spent in some foreign country, and the
ordering from zero days at the bottom end of the scale to 30 days at the top end
is clearly defined. In the bar-chart form of this display, given in the right-hand
graph of Exhibit 1.1, bars are drawn with lengths proportional to the values
of the variable. Of course, the number of days is a rounded approximation of
the time actually spent in each country, but we call this variable continuous
because the underlying time variable is indeed truly continuous.

Expressing The right-hand vertical axis of each plot in Exhibit 1.1 can be used to read the
data in relative corresponding percentage of days relative to the total of 86 days. For example,
amounts the 18 days in Norway account for 21% of the total time. The total of 86 is
often called the base relative to which the data are expressed. In this case
there is only one set of data and therefore just one base, so in these plots the
original absolute scale on the left and the relative scale on the right can be
depicted on the same graph.

Categorical In contrast to the vertical y-axis, the horizontal x-axis is clearly not a numer-
variables ical variable. The four points along this axis are just positions where we have
placed labels denoting the countries visited. The horizontal scale represents a
discrete, or categorical , variable. There are two features of this horizontal axis
that have no substantive meaning in the graph: the ordering of the categories
and the distances between them.
Ordering of categories 3

Firstly, there is no strong reason why Norway has been placed first, Canada Ordering of
second and Greece third, except perhaps that I visited these countries in that categories
order. Because the France/Germany label refers to a collection of shorter trips
scattered throughout the year, it was placed after the others. By the way, in
this type of representation where order is essentially irrelevant, it is usually
a good idea to re-order the categories in a way that has some substantive
meaning, for example in terms of the values of the variable. In this example
we could order the countries in descending order of days, in which case we
would position the countries in the order Greece, France/Germany, Norway
and Canada, from most visited to least. This simple re-arrangement assists
in the interpretation of data, especially if the data set is much larger: for
example, if I had visited 20 different countries, then the order would contain
relevant information that is not quickly deduced from the data in their original
ordering.

Secondly, there is no reason why the four points are at equal intervals apart Distances between
on the axis. There is also no immediate reason to put them at different in- categories
tervals apart, so it is purely for convenience and aesthetics that they have
been equally spaced. Using correspondence analysis we will show that there
are substantively interesting ways to define intervals between the categories
of a variable such as this one, when it is related to other variables. In fact,
correspondence analysis will be shown to yield values for the categories where
both the distances between the categories and their ordering have substantive
meaning.

Since the ordering of the countries is arbitrary on the horizontal axis of Ex- Distance
hibit 1.1, as well as the distances between them, there would be no sense interpretation of
in measuring and interpreting distances between the displayed points in the scatterplots
left-hand graph. The only distance measurement that has meaning is in the
strictly vertical direction, because of the numerical nature of the vertical axis
that indicates frequency (left-hand scale) or relative frequency (right-hand
scale).

In some special cases, the two variables that define the axes of the scatterplot Scatterplots as
are of the same numerical nature and have comparable scales. For example, maps
suppose that 20 students have written a mathematics examination consisting
of two parts, algebra and geometry, each part counting 50% towards the final
grade. The 20 students can be plotted according to their pair of grades, shown
in Exhibit 1.2. It is important that the two axes representing the respective
grades have scales with unit intervals of identical lengths. Because of the simi-
lar nature of the two variables and their scales, it is possible to judge distances
in any direction of the display, not only horizontally or vertically. Two points
that are close to each other will have similar results in the examination, just
like two neighbouring towns having a small geographical distance between
4 Scatterplots and Maps

Exhibit 1.2: 50
Scatterplot of grades
of 20 students in two
sections (algebra
and geometry) of a 40
mathematics
examination. The
points have spatial
properties: for 30
Geometry

example, the total

grade is obtained by
projecting each
point perpendi- 20
cularly onto the 45◦
line, which can be
calibrated from 0
(bottom left corner) 10
to 100 (top right
corner).

0
0 10 20 30 40 50

Algebra

them. Thus, one can comment here on the shape of the scatter of points and
the fact that there is a small cluster of four students with high grades and a
single student with very high grades. Exhibit 1.2 can be regarded as a map,
because the position of each student can be regarded as a two-dimensional
position, similar to a geographical location in a region defined by latitude and
longitude co-ordinates.

Calibration of a Maps have interesting geometric properties. For example, in Exhibit 1.2 the
direction in the 45◦ dashed line actually defines an axis for the final grades of the students,
map combining the algebra and geometry grades. If this line is calibrated from 0
(bottom left) to 100 (top right), then each student’s final grade can be read
from the map by projecting each point perpendicularly onto this line. An
example is shown of a student who received 12 out of 50 and 18 out of 50
for the two sections, respectively, and whose position projects onto the line at
coordinates 15 and 15, corresponding to a total grade of 30.

Information- The scatterplots in Exhibit 1.1 and Exhibit 1.2 are different ways of expressing
transforming in graphical form the numerical information in the two sets of travel and
nature of the examination data respectively. In each case there is no loss of information
display between the data and the graph. Given the graph it is easy to recover the data
Nominal and ordinal variables 5

exactly. We say that the scatterplot or map is an “information-transforming

instrument” — it does not process the data at all; it simply expresses the data
in a visual format that communicates the same information in an alternative
way.

In my travel example, the categorical variable “country” has four categories, Nominal and
and, since there is no inherent ordering of the categories, we refer to this vari- ordinal variables
able more specifically as a nominal variable. If the categories are ordered, the
categorical variable is called an ordinal variable. For example, a day could be
classified into three categories according to how much time is spent working:
(i) less than one hour (which I would call a “holiday”), (ii) more than one but
less than six hours (a “half day”, say) and (iii) more than six hours (a “full
day”). These categories, which are based on the continuous variable “time
spent daily working” divided up into intervals, are ordered and this ordering
is usually taken into account in any graphical display of the categories. In
many social surveys, questions are answered on an ordinal scale of response,
for example, an ordinal scale of importance: “not important”/“somewhat
important”/“very important”. Another typical example is a scale of agree-
ment/disagreement: “strongly agree”/“somewhat agree”/“neither agree nor
disagree”/“somewhat disagree”/“strongly disagree”. Here the ordinal position
of the category “neither agree nor disagree” might not lie between “somewhat
agree” and “somewhat disagree”; for example, it might be a category used
by some respondents instead of a “don’t know” response when they do not
understand the question or when they are confused by it. We shall treat this
topic later in this book (Chapter 21) once we have developed the tools that
allow us to study patterns of responses in multivariate questionnaire data.

Exhibit 1.3:
COUNTRY Holidays Half days Full days TOTAL Frequencies of
different types of
Norway 6 1 11 18 day in four sets of
Canada 1 3 11 15 trips.
Greece 4 25 0 29
France/Germany 2 2 20 24
TOTAL 13 31 42 86

Let us suppose now that the 86 days of my foreign trips were classified into one Plotting more than
of the three categories holidays, half days and full days. The cross-tabulation one set of data
of country by type of day is given in Exhibit 1.3. This table can be considered
in two different ways: as a set of rows or a set of columns. For example, each
column is a set of frequencies characterizing the respective type of day, while
each row characterizes the respective country. Exhibit 1.4(a) shows the latter
way, namely a plot of the frequencies for each country (row), where the hori-
zontal axis now represents the type of day (column). Notice that, because the
categories of the variable “type of day” are ordered, it makes sense to connect
6 Scatterplots and Maps

the categories by lines. Clearly, if we want to make a substantive comparison

between the countries, then we should take into account the fact that different
numbers of days in total were spent in each country. Each country total forms
a different base for the re-expression of the corresponding row in Exhibit 1.3
as a set of percentages (Exhibit 1.5). These percentages are visualized in Ex-
hibit 1.4(b) in a plot that expresses better the different compositions of days
in the respective trips.

Exhibit 1.4: (a) (b) 100

30
Plots of (a)
frequencies in 25

Percentage of days (%)

Exhibit 1.3 and (b) 75
relative frequencies 20

in each row
Days

15 50
expressed as
percentages. 10
25
5

0 0
Holiday Half Day Full Day Holiday Half Day Full Day

Norway Canada Norway Canada

Greece France/Germany Greece France/Germany

Exhibit 1.5:
Percentages of types COUNTRY Holidays Half days Full days
of day in each
country, as well as Norway 33% 6% 61%
the percentages Canada 7% 20% 73%
overall for all Greece 14% 86% 0%
countries combined; France/Germany 8% 8% 83%
rows add up to Overall 15% 36% 49%
100%.

Interpreting There is a lesson to be learnt from these displays that is fundamental to the
absolute or relative analysis of frequency data. Each trip has involved a different number of days
frequencies and so corresponds to a different base as far as the frequencies of the types of
days are concerned. The 6 holidays in Norway, compared to the 4 in Greece,
can be judged only in relation to the total number of days spent in these
respective countries. As percentages they turn out to be quite different: 6 out
of 18 is 33%, while 4 out of 29 is 14%. It is the visualization of the relative
frequencies in Exhibit 1.4(b) that gives a more accurate comparison of how I
spent my time in the different countries. The “marginal” frequencies (18, 15,
29, 24 for the countries, and 13, 31, 42 for the day types) are also interpreted
relative to their respective totals — for example, the last row of Exhibit 1.5
shows the percentages of day types for all countries combined, and could have
been plotted similarly in Exhibit 1.4(b).
Describing and interpreting data, vs. modelling and statistical inference 7

Any conclusion drawn from the points’ positions in Exhibit 1.4(b) is purely Describing and
an interpretation of the data and not a statement of the statistical signifi- interpreting data,
cance of the observed feature. In this book we shall address the statistical vs. modelling and
aspects of graphical displays only towards the end of the book (Chapters 29 statistical inference
and 30); for the most part we shall be concerned only with the question of
data visualization and interpretation. The deduction that I had proportion-
ally more holidays in Norway than in the other countries is certainly true in
the data and can be seen strikingly in Exhibit 1.4(b). It is an entirely dif-
ferent question whether this phenomenon is statistically compatible with a
model or hypothesis of my behaviour that postulates that the proportion of
holidays was generally intended to be the same for all countries, in which case
any observed differences are purely random. Most of statistical methodology
concentrates on problems where data are fitted and compared to a theoretical
model or preconceived hypothesis, with little attention being paid to enlight-
ening ways for describing data, interpreting data and generating hypotheses.
A typical example in the social sciences is the use of the ubiquitous chi-square
statistic to test for association in a cross-tabulation. Often statistically sig-
nificant association is found but there are no simple tools for detecting which
parts of the table are responsible for this association. Correspondence analysis
is one tool that can fill this gap, allowing the data analyst to see the pattern of
association in the data and to generate hypotheses that can be tested in a sub-
sequent stage of research. In most situations data description, interpretation
and modelling can work hand-in-hand with one other. But there are situa-
tions where data description and interpretation assume supreme importance,
for example when the data represent the whole population of interest.

As data tables increase in size, it becomes more difficult to make simple graph- Large data sets
ical displays such as Exhibit 1.4, owing to the overabundance of points. For
example, suppose I had visited 20 countries during the year and had a break-
down of time spent in each one of them, leading to a table with many more
rows. I could also have recorded other data about each day in order to study
possible relationships with the type of day I had; for example, the weather on
each day — “fair weather”, “partly cloudy” or “rainy”. So the table of data
might have many more columns as well as rows. In this case, to draw graphs
such as Exhibit 1.4, involving many more categories and with 20 sets of points
traversing the plot, would result in such a confusion of points and symbols
that it would be difficult to see any patterns at all. It would then become clear
that the descriptive instrument being used, the scatterplot, is inadequate in
bringing out the essential features of the data. This is a convenient point to in-
troduce the basic concepts of correspondence analysis, which is also a method
for visualizing tabular data, but which can easily accommodate larger data
sets in a natural and intuitive way.
8 Scatterplots and Maps

SUMMARY: 1. Scatterplots involve plotting two variables, with respect to a horizontal axis
Scatterplots and and a vertical axis, often called the “x-axis” and “y-axis” respectively.
Maps
2. Usually the x variable is a completely different entity to the y variable. We
can often interpret distances along at least one of the axes in the specific
sense of measuring the distance according to the scale that is calibrated on
the axis. It is usually meaningless to measure or interpret oblique distances
in the plot.
3. In a few cases the x and y variables are similar entities with comparable
scales, in which case interpoint distances can be interpreted as a measure
of difference, or dissimilarity, between the plotted points. In this special
case we call the scatterplot a map. For such maps it is important that the
horizontal and vertical scales have physically equal units, i.e. the aspect
ratio of the axes is equal to 1.
4. When plotting positive quantities (usually frequencies in our context), both
the absolute and relative values of these quantities are of interest.
5. The more complex the data are, the less convenient it is to represent these
data in a scatterplot.
6. This book is concerned with visually describing and interpreting complex
information, rather than modelling it.
Profiles and the Profile Space 2
The concept of a set of relative frequencies, or a profile, is fundamental to
correspondence analysis (referred to from now on by its abbreviation CA).
Such sets, or vectors, of relative frequencies have special geometric features
because the elements of each set add up to 1 (or 100%). In analysing a fre-
quency table, relative frequencies can be computed for rows or for columns —
these are called row or column profiles respectively. In this chapter we shall
show how profiles can be depicted as points in a profile space, illustrating the
concept in the special case when each profile consists of only three elements.

Contents
Profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Average profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Row profiles and column profiles . . . . . . . . . . . . . . . . . . . . . . 10
Symmetric treatment of rows and columns . . . . . . . . . . . . . . . . . 11
Asymmetric consideration of the data table . . . . . . . . . . . . . . . . 11
Plotting the profiles in the profile space . . . . . . . . . . . . . . . . . . 11
Vertex points define the extremes of the profile space . . . . . . . . . . . 12
Triangular (or ternary) coordinate system . . . . . . . . . . . . . . . . . 12
Positioning a point in a triangular coordinate system . . . . . . . . . . . 14
Geometry of profiles with more than three elements . . . . . . . . . . . 14
Data on a ratio scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Data on a common scale . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
SUMMARY: Profiles and the Profile Space . . . . . . . . . . . . . . . . 16

Let us look again at the data in Exhibit 1.3, a table of frequencies with four Profiles
rows (the countries) and three columns (the type of day). The first and most
basic concept in CA is that of a profile, which is a set of frequencies divided
by their total. Exhibit 2.1 shows the row profiles for these data: for example,
the profile of Norway is [0.33 0.06 0.61], where 0.33 = 6/18, 0.06 = 1/18, 0.61
= 11/18. We say that this is the “profile of Norway across the types of day”.
The profile may also be expressed in percentage form, i.e. [33% 6% 61%] in
this case, as in Exhibit 1.5. In a similar way, the profile of Canada across the
day types is [0.07 0.20 0.73], concentrated mostly in the full day category, as
is Norway. In contrast, Greece has a profile of [0.14 0.86 0.00], concentrated
mostly in the half day category, and so on. The percentages are plotted in
Exhibit 1.4(b) on page 6.

9
10 Profiles and the Profile Space

Exhibit 2.1:
Row (country) COUNTRY Holidays Half days Full days
profiles: relative
frequencies of types Norway 0.33 0.06 0.61
of day in each set of Canada 0.07 0.20 0.73
trips, and average Greece 0.14 0.86 0.00
profile showing France/Germany 0.08 0.08 0.83
relative frequencies Average 0.15 0.36 0.49
in all trips.

Average profile In addition to the four country profiles, there is an additional row in Exhibit
2.1 labelled Average. This is the profile of the final row [13 31 42] of Exhibit
1.3, which contains the column sums of the table; in other words this is the
profile of all the trips aggregated together. In Chapter 3 we shall explain more
specifically why this is called the average profile. For the moment, it is only
necessary to realize that, out of the total of 86 days travelled, irrespective
of country visited, 15% were holidays, 36% were half days and 49% were full
days of work. When comparing profiles we can compare one country’s profile
with another, or we can compare a country’s profile with the average profile.
For example, eyeballing the figures in Exhibit 2.1, we can see that of all the
countries, the profiles of Canada and France/Germany are the most similar.
Compared to the average profile, these two profiles have a higher percentage
of full days and are below average on holidays and half days.

Row profiles In the above we looked at the row profiles in order to compare the different
and column countries. We could also consider Exhibit 1.3 as a set of columns and compare
profiles how the different types of days are distributed across the countries. Exhibit 2.2
shows the column profiles as well as the average column profile. For example,
of the 13 holidays 46% were in Norway, 8% in Canada, 31% in Greece and 15%
in France/Germany, and so on for the other columns. Since I spent different
numbers of days in each country, these figures should be checked against those
of the average column profile to see whether they are lower or higher than
the average pattern. For example, 46% of all holidays were spent in Norway,
whereas the number of days spent in Norway was just 21% of the total of 86 —
in this sense there is a high number of holidays there compared to the average.

Exhibit 2.2:
Profiles of types of COUNTRY Holidays Half days Full days Average
day across the
countries, and Norway 0.46 0.03 0.26 0.21
average column Canada 0.08 0.10 0.26 0.17
profile. Greece 0.31 0.81 0.00 0.34
France/Germany 0.15 0.07 0.48 0.28
Symmetric treatment of rows and columns 11

Looking again at the proportion 0.46 (= 6/13) of holidays spent in Norway Symmetric
(Exhibit 2.2) and comparing it to the proportion 0.21 (= 18/86) of all days treatment of rows
spent in that country, we can calculate the ratio 0.46/0.21 = 2.2, and conclude and columns
that holidays in Norway were just over twice the average. Exactly the same
conclusion is reached if a similar calculation is made on the row profiles. In
Exhibit 2.1 the proportion of holidays in Norway was 0.33 (= 6/18) whereas
for all countries the proportion was 0.15 (= 13/86). Thus, there are 0.33/0.15
= 2.2 times as many holidays compared to the average, the same ratio as was
obtained when arguing from the point of view of the column profiles (this
ratio is called the contingency ratio and will re-appear in future chapters).
Whether we argue via the row profiles or column profiles we arrive at the
same conclusion. In Chapter 8 it will be shown that CA treats the rows and
columns of a table in an equivalent fashion or, as we say, in a symmetric way.

Nevertheless, it is true in practice that a table of data is often thought of and Asymmetric
interpreted in a non-symmetric, or asymmetric, fashion, either as a set of rows consideration of
or as a set of columns. For example, since each row of Exhibit 1.3 constitutes a the data table
different country (or pair of countries in the case of France/Germany), it might
be more natural to think of the table row-wise, as in Exhibit 2.1. Deciding
which way is more appropriate depends on the nature of the data and the
researcher’s objective, and the decision is often not a conscious one. One
concrete manifestation of the actual choice is whether the researcher refers to
row or column percentages when interpreting the data. Whatever the decision,
the results of CA will be invariant to this choice, but the interpretation will
adapt to the researcher’s viewpoint.

Let us consider the four row profiles and average profile in Exhibit 2.1 and Plotting the
a completely different way to plot them. Rather than the display of Exhibit profiles in the
1.4(b), where the horizontal axis serves only as labels for the type of day profile space
and the vertical axis represents the percentages, we now propose using three
axes corresponding to the three types of day, which is a scatterplot in three
dimensions. To imagine three perpendicular axes is not difficult: merely look
down into an empty corner of the room you are sitting in and you will see
three axes as shown in Exhibit 2.3. Each of the three edges of the room
serves as an axis for plotting the three elements of the profile. These three
values are now considered to be coordinates of a single point that represents
the whole profile — this is quite different from the graph in Exhibit 1.4(b)
where there is a point for each of the three profile elements. The three axes
are labelled holidays, half days and full days, and are calibrated in fractional
profile units from 0 to 1. To plot the four profiles is now a simple exercise.
Norway’s profile of [0.33 0.06 0.61] (see Exhibit 2.1) is 0.33 of a unit along
axis holidays, 0.06 along axis half days and 0.61 along axis full days. To take
another example, Greece’s profile of [0.14 0.86 0.00] has a zero coordinate
in the full days direction, so its position is on the “wall”, as it were, on the
left-hand side of the display, with coordinates 0.14 and 0.86 on the two axes
12 Profiles and the Profile Space

Exhibit 2.3: holidays

Positions of the four
1.00 o
row profiles (•) of
Exhibit 2.1 as well
as their average 0.75
profile (∗)in
three-dimensional
0.50
space, depicted as
the corner of a room
with “floor tiles”. NORWAY
0.25
For example, FRANCE/GERMANY
Norway is 0.06 along
o full days
the half days axis, average CANADA
0.61 along the full
days axis and 0.33 in GREECE
a vertical direction
along the holidays
axis. The unit o
(vertex) points are
half days
also shown as empty
circles on each axis.

holidays and half days that define the “wall”. All other row profile points in this
example, including the average row profile [0.15 0.36 0.49], can be plotted
in this three-dimensional space.

Vertex points With a bit of imagination it might not be surprising to discover that the
define the profile points in Exhibit 2.3 all lie exactly in the plane defined by the triangle
extremes of the that joins the extreme unit points [1 0 0], [0 1 0] and [0 0 1] on the three
profile space respective axes, as shown in Exhibit 2.4. This triangle is equilateral and its
three corners are called vertex points or vertices. The vertices coincide with
extreme profiles that are totally concentrated into one of the day types. For
example, the vertex point [1 0 0] corresponds to a trip to a country consisting
only of holidays (fictional in my case, unfortunately). Likewise, the vertex point
[0 0 1] corresponds to a trip consisting only of full days of work.

Triangular (or Having realized that all profile points in three-dimensional space actually lie
ternary) exactly on a flat (two-dimensional) triangle, it is possible to lay this triangle
coordinate system flat, as in Exhibit 2.5. Looking at the profile points in a flat space is clearly
better than trying to imagine their three-dimensional positions in the corner of
a room! This particular type of display is often referred to as the triangular (or
ternary) coordinate system and may be used in any situation where we have
sets of data consisting of three elements that add up to 1, as in the case of the
row profiles in this example. Such data are common in geology and chemistry,
for example where samples are decomposed into three constituents, by weight
Exploring the Variety of Random
Documents with Different Content
 [825] Marchal, Ouv. cit., p. 455, XXVII.
[826] Marchal, Ouv. cit., p. 64.
[827] Malebranche, XIe Entretien sur la
métaphysique.
[828] Max. Simon, De la préservation du
choléra épidémique et d’une hygiène spéciale,
applicable au traitement de la maladie réalisée.
Paris, 1865.
[829] Fonssagrives, Bulletin général de
thérapeutique, t. LXX, p. 35. 1866.
[830] Ch. Anglada, Traité de la Contagion, t.
II, ch. 3, p. 99-150.
[831] Voy. A. Fauvel, le Choléra, étiologie et
prophylaxie... exposé des travaux de la
conférence sanitaire internationale de
Constantinople..... Paris, 1868.
[832] Joannis Varandæi Opera omnia,
Patholog. universalis pars II, cap. XIV, p.
CXXXIX. Lugduni MDCLVIII.
[833] Bally, Études sur la choladrée
lymphatique. Paris, 1833.
[834] Jules Guérin, Bull. de l’Académie de
médecine. Paris, 1865-66, tome XXXI, p. 11.
[835] Rapport sur la marche et les effets du
choléra-morbus dans Paris. Année 1832. Paris,
MDCCCXXXIV.
[836] Double, Rapp. sur le choléra-morbus
épid., p. 147. MDCCCXXXI.
[837] Voici, à ce propos, ce que dit
Broussais, donnant les résultats d’un certain
nombre de nécroscopies rédigées sous ses
yeux: «Ayant examiné douze ou quinze fois le
plexus solaire, et ne l’ayant trouvé que deux fois
un peu injecté, et une seule fois un peu ramolli,
sans que ces lésions se rapportassent à des
symptômes spéciaux, j’ai cessé cette recherche,
qui, continuée par d’autres, n’a pas donné des
résultats différents.» (Le choléra-morbus épid.,
etc., p. 97. 1832.)
 [838] Voyez Griesinger, Traité des maladies
infectieuses, trad. par G. Lemattre. Paris, 1868.
[839] Broussais, le Choléra-morbus
épidémique, etc., p. 50 (passim).
[840] Dict. de médecine, 12e édition, par
Littré et Robin, art. Choléra.
[841] Mézeray, Hist. de France, t. II, p. 966.
[842] Voy. le Compendium de méd., au mot
choléra.
[843] Zacutus Lusitanus, Praxeos med.
admir., lib. II, obs. XXIII. Lugduni, 1643.
[844] Zacuti Lusitani Operum tomus
secundus, p. 622, numerus VII, De cholera.
[845] Zacutus, Ibid., Praxeos med. admir.,
lib. II, observ. XVI. Choleræ sævissimæ curatio.
[846] Cœlius Aurelianus, De morbis acutis et
chronicis, lib. III, cap. XIX.
[847] Medicæ artis principes, t. I, p. 236.—
Alexandri Tralliani De arte med., lib. secundus,
cap. XIII, de cholera.
[848] Thomæ Sydenham Opera medica, t. I,
sect. quarta, cap. II.
[849] Sydenham, Op. cit., t. I, sect. IV, cap.
II.
[850] Bontii de medicina Indorum, lib. IV, de
cholera, cap. VI, p. 136. Lugduni Batavorum.
1642.
[851] La préparation officinale que Bontius
prescrit sous le nom d’extrait de safran
(extractum croci) a pour base l’opium (opium
electissimum). L’auteur en donne la formule à la
page 131.
[852] Hippocrate, edit. Foës, de morbis
vulgaribus, lib. VII, sect. VII, p. 1106.—Choleræ
origo et causa.—De affectionibus lib., sect. V, p.
587.—Choleræ curatio. Francofurti. MDXCVI.
[853] Pauli Æginetæ de re medica, cap.
XXXIX, de cholera, in medicæ artis principes (t.
I, p. 458. MDLXVIII).
 [854] Celsi de medicina, lib. IV, cap. XI.
[855] Aretæi Cappadocis medici de causis et
signis acutorum morborum, cap. V, de cholera,
in med. art. principes, t. I, p. 11.
[856] Cœlii Aureliani de morbis acutis et
chronicis, lib. III, cap. XIX, p. 253, de cholericis.
[857] Alexandri Tralliani De arte medica, lib.
secundus, cap. XIV, de cholera, in med. art.
princip., t. I, p. 236.
[858] Ballonii opera omnia med., t. II,
consilium LV, p. 252.
[859] Requin, Pathol. méd., t. II, p. 568.
[860] Moreau, Gaz. méd. de Paris, t. IV, 3e
série, p. 213.
[861] Ordonnance du roi. Arrêtés et
règlement pour l’Acad. roy. de médecine. Art. 2.
1820.
[862] Lucrèce, De la nature des choses.
Trad. par Pongerville, Note 43 du livre sixième.
1845.
[863] Broussais, le Choléra-morbus
épidémique, p. 2. 1832.
[864] Victor Gravier (de Marseille),
Dissertation sur le choléra-morbus de l’Inde, 30e
vol. de la collect. des Thèses de Strasbourg.
Première partie. 1823.
[865] Fodéré, Leç. sur les épidémies, t. II, p.
394. 1823.
[866] Broussais, le Choléra-morbus
épidémique, etc. 1832 (passim).
[867] Gravier, Thèse citée, p. 8.
[868] Gravier, Th. cit., p. 16.
[869] Gravier, Th. cit., p. 17.
[870] Gravier, Th. cit., p. 19.
[871] Fuster, des Maladies de la France, etc.,
Appendice, p. 276. 1840.
[872] Requin, Path. médicale, t. II, p. 565.
 [873] Requin, de la spécificité dans les
maladies, Thèse de concours. Art. V, prop. I.
1851.
[874] Requin, Pathol. méd., t. III, p. 495-
496.
[875] Grisolle, Traité de pathol. interne, t. I,
p. 708. 1852.
[876] Andral, Cours de pathologie interne,
rédigé par Am. Latour, t. I, p. 120. 1836.
[877] Cette observation est extraite d’un
mémoire manuscrit présenté par les auteurs
que j’ai nommés, à la section médicale de
l’Académie des sciences et lettres de
Montpellier.
[878] Aretæi Cappadocis medici de causis et
signis acutorum morborum, cap. V.
[879] Cœlius Aurelianus, De morbis acutis et
chronicis, lib. III, cap. XIX.
[880] Josephi Quarin animadversiones
practicæ in diversos morbos, cap. X, p. 205.
Viennæ, 1786.
[881] P. Frank, Traité de médecine pratique,
t. III, p. 469. 1820. Trad.
[882] Lind, Essai sur les maladies des
Européens dans les pays chauds, trad. par
Thion de la Chaume. T. II, p. 48. Paris, 1785.
Voyez Dutroulau, Traité des maladies des
Européens dans les pays chauds, 2e édition.
Paris, 1868.
[883] Fodéré, Leçons sur les épidémies, t.
II, p. 426.
[884] Thomæ Sydenham Opera med. T. I, p.
184. Epistola Roberto Brady.
[885] Alexandri Tralliani de arte med., lib.
secundus, cap. XIIII, de cholera.
[886] Ballonii Op. omnia medica, t. III,
consilium LXV.
[887] Voy. Fauvel, Ouv. cité, p. 134.
 E R R ATA

Page 88, note 95, au lieu de: Thucydide,

lisez: de Pongerville.
Page 125, ligne 8, au lieu de: manière
irritante, lisez: matière irritante.
Ibid., ligne 19, au lieu de: érésipèle, lisez:
érysipèle.
Page 340, note 455, ligne 1re, au lieu de:
præfocante puero, lisez: pueros.
Page 474, note 637, au lieu de: p. 11, lisez:
p. XI.
Page 503, note 679, ligne 1re, au lieu de: de
abitis, lisez: de abditis.
Page 508, ligne 24, au lieu de: séparaton,
lisez: séparation.
Page 576, ligne 31, au lieu de: hanc amen
ipsam, lisez: hanc tamen ipsam.
 TA B LE D E S
MATIÈRES

Dédicace. V
Préface. VII
Introduction. 1
Chapitre Ier. — De la grande épidémie du
Ve siècle avant l’ère
chrétienne (peste
d’Athènes). 51
—— II. — De la grande épidémie du
IIe siècle de l’ère
chrétienne (peste
Antonine). 113
—— III. — De la grande épidémie du
IIIe siècle après Jésus-
Christ (peste sous
Gallus). 140
—— IV. — De la grande épidémie du
VIe siècle (peste
inguinale). 158
—— V. — Des épidémies de fièvres
éruptives nouvelles
apparues au VIe siècle
de l’ère chrétienne. 215
re.
Section I — De la variole 218
considérée
 comme maladie
nouvelle.
—— II. — De la rougeole
considérée
comme maladie
nouvelle. 278
—— III. — De la scarlatine
considérée
comme maladie
nouvelle. 304
—— VI. — De la grande épidémie
gangréneuse du
moyen âge (mal des
ardents, feu Saint-
Antoine). 351
—— VII. — De la grande épidémie du
XIVe siècle (peste
noire). 395
—— VIII. — De la grande épidémie du
XVe siècle (suette
anglaise). 449
—— IX. — De la grande épidémie
syphilitique du XVe
siècle. 535
—— X. — De la grande épidémie
cholérique du XIXe
siècle. 591
Notes. NOTES
Errata. 647

F I N D E LA TA B L E D E S M AT I È R E S

IMPRIMERIE L. TOINON ET Ce, A SAINT-GERMAIN.

 J.-B. BAILLIÈRE et FILS,
L I B R A I R E S D E L’A C A D É M I E I M P É R I A L E D E
MÉDECINE,
Rue Hautefeuille, 19, à Paris.

LONDRES, MADRID, NEW-YORK

Hippolyte Baillière. C. Bailly-Baillière. Baillière Brothers.

Leipzig, E. Jung-Treuttel, Querstrasse, 10.

Janvier 1869.

ŒUVRES COMPLÈTES

D’ H IPPOC R ATE
TRADUCTION NOUVELLE AVEC LE
TEXTE EN REGARD
COLLATIONNÉ SUR LES MANUSCRITS ET
TOUTES LES ÉDITIONS
ACCOMPAGNÉE D’UNE
INTRODUCTION
de Commentaires médicaux, de Variantes et de Notes
philologiques
SUIVIE

D’UNE TABLE GÉNÉRALE DES MATIÈRES

 Par É. LITTRÉ
Membre de l’Institut (Académie des inscriptions et belles-
lettres),
de l’Académie impériale de médecine, de la Société de biologie
de Paris,
et de la Société d’histoire naturelle de Halle.

OUVRAGE COMPLET, 10 forts volumes in-

8o.—100 fr.

En publiant cet ouvrage, M. Littré a eu pour

but de mettre les œuvres hippocratiques
complétement à la portée des médecins de
notre temps; il a voulu qu’elles pussent être
lues et comprises comme un livre
contemporain. Deux difficultés principales s’y
opposaient: la première gisait dans des théories
antiques qui, depuis longtemps, ont cessé d’être
familières aux esprits, et dont l’intelligence est
nécessaire pour l’interprétation d’une foule de
passages; la seconde était dans l’emploi d’une
ancienne langue médicale où les mots ont
quelquefois une acception mal déterminée, et
quelquefois aussi une acception trompeuse,
attendu qu’ils ont changé de signification en
passant dans le langage moderne. Pour
remédier à ces difficultés, en tête de chaque
traité, M. Littré a exposé dans un Argument ce
qui est nécessaire à l’intelligence de ce traité;
puis il a précisé, autant que la nature des
choses le permettait, le langage antique, et, à
cet effet, il a souvent essayé un diagnostic
rétrospectif qui n’est pas entouré de moindres
obscurités que le diagnostic au lit du malade.
Ceci dit, nous croyons que la meilleure et la plus
simple manière de faire connaître un ouvrage
est d’exposer ce qu’il contient.
Le tome Ier est consacré presque
entièrement à une Introduction (pages 1-478).
Là sont traitées les questions préliminaires dont
la solution importe à l’intelligence des livres
hippocratiques. Le volume est terminé par le
traité De l’Ancienne médecine, ouvrage
important de philosophie scientifique, et où la
collation des manuscrits a permis de combler
une lacune considérable et de rétablir une
mention d’Empédocle.
Le tome II renferme le traité Des Airs, des
Eaux et des Lieux, le Pronostic, le livre Du
Régime des maladies aiguës, et le premier livre
Des Épidémies. La polémique d’Hippocrate
contre les médecins de Cnide est trop
importante pour n’être pas appréciée: l’école de
Cos note surtout les symptômes généraux,
l’école de Cnide surtout les symptômes
particuliers; la première a pour doctrine une
sorte de physiologie pathologique, la seconde
est essentiellement descriptive. Voilà pour le
système d’Hippocrate; voici pour un point
considérable de sa médecine, la Pyrétologie. Les
pays chauds sont affectés endémiquement de
fièvres intermittentes, rémittentes et continues,
marquées d’un caractère à peu près étranger
aux régions tempérées non marécageuses. M.
Littré a montré, dans un Argument, que les
fièvres décrites par Hippocrate y devaient être
rapportées. Cette remarque a jeté un jour tout
nouveau sur la pyrétologie du vieux médecin
grec, et donne un élément de plus à l’étude des
maladies suivant leur distribution géographique.
Dans le tome III sont le troisième livre des
Épidémies, le traité Des Plaies de tête, le livre
De l’Officine du médecin et celui Des Fractures,
un des traités les plus importants, le moins
connu, qui par une révision des textes et une
savante interprétation, en fait un livre que tous
les chirurgiens doivent consulter. M. Littré,
recherchant si Hippocrate avait connu la peste à
bubons, a établi, à l’aide de textes irréfragables,
que cette maladie, regardée jusqu’à présent
comme récente, comparativement et comme
datant du VIe siècle de l’ère chrétienne, devait
être reportée plus haut et qu’elle avait sévi
d’une manière épidémique dans le premier
siècle au moins de cette ère et sans doute
beaucoup plus tôt.
Le tome IV comprend le traité Des
Articulations, le Mochlique, les Aphorismes, le
Serment et la Loi. De nombreuses corrections
dans le texte ont rendu très-facile à lire le grand
et important traité Des Articulations. M. Littré a
terminé le volume par des Remarques
rétrospectives; là, il classe les livres qu’il
regarde comme étant d’Hippocrate lui-même,
suivant les objets qui y sont traités; là, enfin,
sont appréciées les connaissances
physiologiques d’Hippocrate, sa doctrine de la
crase et la tentative de physiologie pathologique
qu’il a faite dans le Pronostic.
Dans le tome V se trouvent les IIe, IVe, Ve,
VIe et VIIe livres Des Épidémies, le traité Des
Humeurs, le premier livre du Prorrhétique et les
Prénotions de Cos. Ces cinq livres Des
Épidémies donnent lieu à des études sur la
pratique, la clientèle et le mode de travailler des
médecins hippocratiques: ils donnent lieu aussi
à un essai sur le caractère de plusieurs des
grandes épidémies qui ont affligé l’antiquité.
Le tome VI renferme un grand nombre de
traités relatifs à des objets différents: le traité
De l’Art, destiné à combattre ceux qui
prétendent que la médecine n’existe pas; De la
Nature de l’homme; Du Régime salutaire, qui
donne des préceptes hygiéniques; Des Vents,
qui attribue toutes les maladies à une cause
unique (le vent ou pneuma); De l’Usage des
liquides; le livre premier Des Maladies, ouvrage
dont le but est de donner au médecin des idées
générales sur les nécessités pathologiques qui
font qu’une maladie a telle ou telle issue, et sur
les conditions que le médecin doit remplir pour
exercer habilement; les livres Des Affections;
Des Lieux dans l’homme, qui renferme une
proposition dont l’homœopathie s’est emparée;
De la Maladie sacrée, remarquable surtout par
deux points de doctrine: le premier, c’est que
toutes les maladies sont de cause naturelle; le
second, c’est que toute fonction intellectuelle et
morale appartient au cerveau; Des Plaies; Des
Hémorrhoïdes et des Fistules, traités où il est
parlé du spéculum de l’anus et de la membrane
tapissant les trajets fistuleux; enfin le grand
traité Du Régime et des Songes.
Tome VII. Des Maladies, livres II, III (162
pages). — Des Affections internes (140 pages).
— De la nature de la Femme (50 pages). — Du
Fœtus à sept, huit et neuf mois, de la
Génération, de la nature de l’Enfant (80 pages).
— Des Maladies, livre IV (70 pages), etc.
Tome VIII. Maladies des Femmes, des
Femmes stériles, des Jeunes Filles, de la
Superfétation, de l’Anatomie, de la Dentition,
des Glandes, des Chairs, des Semaines, etc.
Tome IX. Prorrhétique. — Du Cœur. — De
l’Aliment. — De la Vision. — De la nature des
Os. — Du Médecin. — De la Bienséance. —
Préceptes. — Des Crises, Jours critiques,
Décrets, Harangues, Lettres et Discours. —
Appendice.
Tome X et dernier. Dernières remarques. —
Table générale alphabétique, travail
considérable de 400 pages, complément
indispensable dans une collection qui comprend,
comme les Œuvres d’Hippocrate, 70 traités sur
des sujets variés.

Les Œuvres complètes d’Hippocrate

forment dix forts volumes in-8 de 700
pages chacun. Prix de chaque volume 10 fr.
Il a été tiré quelques exemplaires sur
grand papier jésus vélin. Prix de chaque
volume 20 fr.
On peut encore souscrire en retirant un
volume à la fois.

HISTOIRE DE LA MÉDECINE 3 fr.

GRECQUE DEPUIS ESCULAPE,
jusqu’à Hippocrate exclusivement, par
le docteur M. S. Houdart. Paris, 1856,
in-8 de 320 pages (6 fr.).
ÉTUDES HISTORIQUES ET
CRITIQUES sur la vie et la doctrine
d’Hippocrate et sur l’état de la
médecine avant lui, par le docteur
Houdart, membre correspondant de
l’Académie de médecine; 2e édition
augmentée. Paris, 1840, in-8. 4 fr. 50
MOSCHIONIS. DE MULIERUM
PASSIONIBUS. Libri græce et latine
edente, F. Dewetz. Viennæ, 1793, in-8. 3 fr.
ALBUCASIS. DE CHIRURGIA,
arabice et latine cura J. Channing,
Oxonii. 1778, 2 vol. in-4 avec figures. 50 fr.
LA CHIRURGIE D’ALBUCASIS,
traduite par le docteur Lucien Leclerc,
médecin major, précédée d’une
introduction. Paris, 1861, in-8, 342
pages avec pl. 6 fr.
LA MÉDECINE DU PROPHÈTE,
traduit de l’arabe par Perron, ancien
directeur de l’école de médecine du
Caire, directeur du collége français
arabe à Alger, etc. 1860. In-8 de 228
pages. 4 fr.
LA MÉDECINE A TRAVERS LES
SIÈCLES. Histoire et philosophie, par
J. M. Guardia, docteur en médecine et
docteur ès lettres, bibliothécaire
adjoint de l’Académie de médecine. 1
vol. in-8 de 800 pages. 10 fr.
 Table des matières: Histoire. La tradition médicale. La
médecine grecque avant Hippocrate. La légende
hippocratique. Classification des écrits hippocratiques.
Documents pour servir à l’histoire de l’art. — Philosophie.
Questions de philosophie médicale. Nos philosophes
naturalistes. Sciences anthropologiques. Buffon. La
philosophie positive et ses représentants. La
métaphysique médicale. Asclépiade, fondateur du
méthodisme. Esquisse des progrès de la physiologie
cérébrale. De l’enseignement de l’anatomie générale. La
méthode expérimentale et la physiologie. Les vivisections
à l’Académie de médecine. Les misères des animaux.
Abus de la méthode expérimentale. Philosophie sociale.

ŒUVRES
ANATOMIQUES, PHYSIOLOGIQUES

ET MÉDICALES

DE GALIEN
Traduites sur les textes imprimés et
manuscrits
ACCOMPAGNÉES DE SOMMAIRES, DE NOTES,
DE PLANCHES, ETC.
 Par le Dr Ch. DAREMBERG
Bibliothécaire de la bibliothèque Mazarine, Bibliothécaire
honoraire de l’Académie de médecine, etc.

Déjà M. Littré a fait revivre Hippocrate: le

prenant pour guide, M. Daremberg a fait revivre
Galien, le plus illustre médecin de l’antiquité
après Hippocrate.
Galien était un grand anatomiste; il suffit,
pour s’en convaincre, de suivre ses descriptions
sur la nature dans le livre De l’Utilité des
parties; — c’était un habile physiologiste, ses
ingénieuses expériences sur les systèmes
nerveux et sanguins en sont un irrécusable
témoignage; — c’était un pathologiste éminent,
son beau traité Des Lieux affectés ne laisse
aucun doute à cet égard.
Le traité de l’Utilité des parties du corps,
dont on ne paraît pas avoir compris le vrai
caractère, se résume dans cette sentence
d’Aristote: Que la nature ne fait rien en vain.
Aussi Galien, loin d’y traiter les questions de
physiologie proprement dite, ne s’y occupe qu’à
découvrir et à démontrer que les parties ne
pouvaient être mieux disposées qu’elles ne le
sont, et qu’elles sont parfaitement adaptées aux
fonctions qu’elles ont à remplir. — Une
conception hardie, et jusqu’à un certain point
nouvelle, de la parfaite harmonie entre les
diverses parties du corps, est une des qualités
qui distinguent cet ouvrage.
 Dans le Traité des Lieux affectés, Galien a
devancé l’école moderne, en démontrant, par la
théorie et par les faits, combien il importe
d’abord à la connaissance des maladies, puis à
la thérapeutique, de savoir exactement le siége
du mal, en d’autres termes, d’arriver au
diagnostic local. Cet admirable ouvrage, l’un des
plus beaux titres de gloire de Galien, est pour la
première fois traduit en français, il figure tout
entier dans le second volume.
Les traités Des Facultés naturelles, Du
Mouvement des muscles, Des Sectes, aux
étudiants, De la meilleure Secte à Thrasybule,
nous présentent une idée à peu près complète
de la physiologie théorique et expérimentale de
Galien.
Le traité De la Méthode Thérapeutique à
Glaucon donnera une idée de la manière dont il
concevait et exposait les généralités sur la
médecine.
Les Œuvres de Galien forment 2 forts
volumes grand in-8 de 700 pages. Prix de
chaque, 10 francs.

Envoi FRANCO, par la poste, contre un

Mandat.
 ŒUVRES
D’ORIBASE
TEXTE GREC, EN GRANDE PARTIE
INÉDIT
COLLATIONNÉ SUR LES MANUSCRITS

Traduit pour la première fois en

français, avec une Introduction, des
Notes, des Tables et des Planches,
PAR LES DOCTEURS

BUSSEMAKER et DAREMBERG.
6 forts vol. in-8, gr. papier, imprimé à
l’Imprimerie impériale.
Les tomes I à IV, chacun de 750 pages, sont en
vente. — Prix du vol.: 12 fr.

Les amis des lettres et de la médecine

ancienne applaudiront à la publication des
Œuvres complètes d’Oribase; c’est pour la
première fois qu’elles ont été réunies avec de
notables augmentations. — Une partie
seulement de la Collection médicale, véritable
encyclopédie de la médecine ancienne, avait été
publiée en grec. Le Synopsis en neuf livres, et le
traité Ad Eunapium en quatre livres, n’ont
jamais été publiés qu’en latin.
Pour entreprendre un travail de cette
importance, il fallait les longues études, les
laborieuses recherches et le dévouement de
MM. Daremberg et Bussemaker; il fallait, de
plus, les heureuses circonstances où s’est
trouvé M. Daremberg, qui a été chargé par le
ministre de l’instruction publique de quatre
missions littéraires dans les principales
bibliothèques d’Allemagne, de Belgique,
d’Angleterre et d’Italie, pour y recueillir de
nombreux matériaux.
On sait qu’un des grands mérites des
Œuvres d’Oribase est d’être formées d’extraits
textuels de médecins et de chirurgiens anciens,
dont plusieurs nous seraient à peu près
inconnus, si ces précieux fragments n’avaient
été sauvés par le médecin et l’ami de
l’empereur Julien.
Les livres Chirurgicaux d’Oribase ont un
intérêt tout particulier: cependant ils sont peu
connus: cela tient à ce que les uns ne sont
publiés qu’en grec, et que pour les autres la
traduction latine est souvent aussi difficile à
entendre que le texte.
Les livres publiés par Cocchi et par le
cardinal Ang. Mai ont été revus sur les
manuscrits de Florence et du Vatican par MM.
Bussemaker et Daremberg.
Les fragments retrouvés par M. Bussemaker,
trois livres de la Collection médicale entièrement
inconnus et découverts par Dietz et par M.
Daremberg, ajoutent un nouveau prix aux
œuvres d’Oribase.
Les quatre volumes publiés comprennent:
Tome Ier. Plan de la collection des médecins
grecs. — Les rapports de l’Académie des
inscriptions et belles-lettres et de l’Académie
impériale de médecine. — Collection médicale,
livres comprenant les aliments, les boissons, les
exercices (avec des notes sur la Gymnastique
chez les anciens).
Tome IIe. Collection médicale, livres
comprenant les émissions sanguines, les
évacuations; de l’air et des localités; des
médicaments externes; des bains; médication
topique; médicaments simples, médicaments
composés (avec des notes importantes).
Tome IIIe. Physiologie et Pathologie
générales, physiologie de la Génération;
hygiène, pathologie et symptomatologie
générales; Splanchnologie; Nomenclature, os,
muscles, nerfs, vaisseaux; Tumeurs contre
nature (abcès, sinus, vésicules, fistules,
gangrène, etc., du sphacèle, de l’érysipèle, des
squirrhes, de l’herpès, de la phagédéenne, de
l’œdème, des furoncles, etc.).
Tome IVe. Comprenant: des Tumeurs
enkystées; des contractures de la langue; des
varices, des scrofules, de l’emphysème, de
l’anévrysme, du traitement de l’éléphantiasis,
des dépôts, des fractures, des luxations, du
déplacement des os du pied, des lacs, des
bandages et des machines, du plinthium de
Nilée, du glossocome de Nymphadore, machine
de l’artisan, du ban d’Hippocrate, de
l’hypospadias, de la hernie, des ulcères, etc.

PUBLICATIONS DE J.-B.
BAILLIÈRE et FILS.
HISTOIRE GÉNÉRALE DES SCIENCES
MÉDICALES. Résumé du cours fait au Collége
de France par M. Ch. Daremberg, chargé du
cours d’histoire de la médecine au Collége de
France, membre de l’Académie de médecine.
Paris, 1869, 1 vol. in-8 de 750 pages.
RECHERCHES SUR L’ÉTAT DE
LA MÉDECINE DURANT LA
PÉRIODE PRIMITIVE de l’histoire
des Indous, par le docteur Ch.
Daremberg, professeur chargé du
cours d’histoire de la médecine au
Collége de France, bibliothécaire de la
Bibliothèque Mazarine. Paris, 1867,
in-8 de 24 pages. 1 fr. 25
NOTICES ET EXTRAITS DES
MANUSCRITS MÉDICAUX GRECS,
LATINS ET FRANÇAIS, des
principales bibliothèques d’Europe,
1re partie, Bibliothèques d’Angleterre,
par le docteur Ch. Daremberg. Paris,
1853, in-8. 7 fr.
GLOSULÆ QUATUOR 4 fr. 50
MAGISTRORUM SUPER
CHIRURGIAM ROGERII ET
ROLANDI, publiées pour la première
fois par le docteur Ch. Daremberg.
Napoli, 1854, in-8 de LXIV — 228
pages.
DE SECRETIS MULIERUM, De
chirurgia, de modo medendi, libri
septem. Poema medicum; nunc
primum ad fidem codicis Mazarinæi,
edidit C. Daremberg. Napoli, 1854, in-8
de 178 pages. 3 fr. 50
HISTOIRE DE LA MÉDECINE
depuis son origine jusqu’au XIXe
siècle, Paris, 1846, 2 vol. in-8. —
LETTRES PHILOSOPHIQUES ET
HISTORIQUES SUR LA
MÉDECINE AU XIXe SIÈCLE, par le
docteur P.-V. Renouard. Troisième
édition, corrigée et considérablement
augmentée. Paris, 1861. In-8 de 240
pages. 15 fr. 50
— Séparément, Lettres
philosophiques et historiques. 3e
édition. 1861, in-8. 3 fr. 50
Ces lettres traitent: I. La médecine jugée par les
médecins. — II. Est-il, en médecine, un moyen de
discerner le vrai du faux, le certain de l’hypothèse? — III.
Des causes qui engagèrent les médecins à quitter la voie
primitive de l’observation pure. — IV. La physiologie
pathologique peut-elle être, oui ou non, en totalité ou en
partie, le fondement direct et immédiat de la
thérapeutique? — V. De l’éclectisme en médecine. — VI.
De l’homœopathie. — VII. Des méthodes thérapeutiques.
— VIII. Réponse à quelques objections concernant la
doctrine empiri-methodique. — IX. Du rang que la
médecine doit occuper dans un système général des
connaissances humaines, et du degré de certitude qu’elle
peut atteindre. — X. Les doctrines médicales devant
l’Académie impériale de médecine. — XI. Les doctrines
médicales devant les Facultés de médecine de France.
 HISTOIRE DES SCIENCES
NATURELLES AU MOYEN AGE, ou
Albert le Grand et son époque considérés
comme point de départ de l’école
expérimentale, par F.-A. Pouchet, directeur
du Muséum d’histoire naturelle de Rouen.
Paris, 1853. 1 beau vol. in-8. 9 fr.
Table des matières: Introduction. — École scandinave.
— École franco-gothique. — École bizantine. — École
arabe. — École EXPÉRIMENTALE: Albert le Grand, St-
Thomas d’Aquin, Roger Bacon, Alfred le Philosophe,
Raymond de Lulle, Duns Scott, Trithème, Bazile Valentin,
Nicolas Flamel, Vincent de Beauvais, Abélard, Barthélemy,
Bruneto Latini, Richard de Furnival, Agricola, Platearius,
Simon de Cordo, Leoniceno, J. de Dondis, P. Sanctinus,
Leonard de Vinci, Arnaud de Villeneuve, P. d’Abano,
Lanfranc, Guy de Chauliac, J. de Vigo, Mundinus,
Béranger de Carpi, Achillini, Marco Polo, etc.

L’ÉCOLE DE SALERNE.
Traduction en vers français, par Ch.
Meaux Saint-Marc, avec le texte latin en
regard (1870 vers), précédée d’une
introduction par M. le docteur Ch.
Daremberg. — DE LA SOBRIÉTÉ,
conseils pour vivre longtemps, par L.
Cornaro, traduction nouvelle. Paris,
1861, 1 joli vol. in-18 jésus de LXXII-
344, avec 5 vignettes. 3 fr. 50
LETTRES DE GUI PATIN. 21 fr.
Nouvelle édition augmentée de lettres
inédites, précédée d’une notice
biographique, accompagnée de
remarques scientifiques, historiques,
philosophiques et littéraires, par
Réveillé-Parise, membre de l’Académie
impériale de médecine. Paris, 1846, 3
vol. in-8, avec le portrait et le fac-
simile de Gui Patin.
Les lettres de Gui Patin sont de ces livres qui ne
vieillissent jamais, et quand on les a lues, on en conçoit
aussitôt la raison. Ces lettres sont en effet l’expression la
plus pittoresque, la plus vraie, la plus énergique, non-
seulement de l’époque où elles ont été écrites, mais du
cœur humain, des sentiments et des passions qui
l’agitent. — Tout à la fois savantes, érudites, spirituelles,
profondes, enjouées, elles parlent de tout: mouvements
des sciences, hommes et choses, passions sociales et
individuelles, révolutions politiques, etc. C’est donc un
livre qui s’adresse aux savants, aux médecins, aux
érudits, aux gens de lettres, aux moralistes, etc.

CELSI (A.-C.) DE RE MEDICA

LIBRI OCTO, editio nova, curantibus P.
Fouquier, in Facultate Parisiensi
professore, et F.-S. Ratier, D. M. Parisiis,
1823, in-18. 2 fr.
LA MÉDECINE ET LES MÉDECINS,
philosophie, doctrines, institutions,
critiques, mœurs et biographies
médicales, par Louis Peisse. Paris, 1857. 2
vol. in-18 jésus. 7 fr.
Cet ouvrage comprend: Esprit, marche et
développement des sciences médicales. — Découvertes
et découvreurs. — Sciences exactes et sciences non
exactes. — Vulgarisation de la médecine. — La méthode
numérique. — Le microscope et les microscopistes. —
Méthodologie et doctrines. — Comme on pense et ce
qu’on fait en médecine à Montpellier. — L’encyclopédisme
et le spécialisme en médecine. — Mission sociale de la
médecine et du médecin. — Philosophie des sciences
naturelles. — La philosophie et les philosophes par-
devant les médecins. — L’aliénation mentale et les
aliénistes. — Phrénologie: bonnes et mauvaises têtes,
grands hommes et grands scélérats. — De l’esprit des
bêtes. — Le feuilleton. — L’Académie de médecine. —
L’éloquence et l’art à l’Académie de médecine. —
Charlatanisme et charlatans. — Influence du théâtre sur
la santé. — Médecins poëtes. — Biographie.

ÉTUDES SUR LE TRAITÉ DE

MÉDECINE D’ABOUJAFAR
AH’MAD, intitulé: Zad Al-Mocafir, «La
provision du voyageur», par G. Dugat,
membre de la Société asiatique. Paris,
1853, in-8 de 64 pages (2 fr. 50). 1 fr.
STORIA DELLA MEDICINA IN
ITALIA, dell dott. Salvator Renzi,
medico napolitano. Napoli, 1845 à
1848. 5 forts vol. in-8. 40 fr.
FLOS MEDICINÆ, scholæ
salertina, seconde édition entièrement
refondue, comprenant les travaux
inédits de Baudry de Balzac, et les vers
nouvellement recueillis par Ch.
Daremberg, et S. de Renzi, publié par
les soins du docteur S. de Renzi.
Naples, 1859, in-8 de LXVIII-128 pages. 4 fr.
STORIA DOCUMENTA DELLA
SCUOLA MEDICA DI SALERNO,
seconda edizione. Napoli, 1857, in-8
de 608-CLXXXIV pages. 12 fr.
MAGISTRI SALERNI. Tabulæ et
Compendium, extraits des manuscrits
de la Bibliothèque impériale de Paris,
enrichis de notes et de notices
bibliographiques et historiques de
Baudry de Balzac. Naples, 1859, in-8
de 68 pages. 2 fr. 50
HISTOIRE DES FEMMES- 75 c.
MÉDECINS depuis l’antiquité jusqu’à
nos jours, par H. Scoutetten, docteur et
professeur en médecine, officier de la
Légion d’honneur, etc. Paris, 1868, in-
18 de 24 pages.
LE CHARLATANISME ET LES
CHARLATANS EN MÉDECINE, étude
psychologique, par le docteur Verdo.
Paris, 1867, 1 vol. in-12 de 48 pages. 1 fr.
ŒUVRES COMPLÈTES
D’AMBROISE PARÉ, revues et
collationnées sur toutes les éditions,
avec les variantes; ornées de 217
figures et du portrait de l’auteur,
accompagnées de notes historiques et
critiques, et précédées d’une
introduction sur l’origine et les progrès
de la chirurgie en Occident du VIe au
XVIe siècle, et sur la vie et les ouvrages
d’Ambroise Paré, par J.-F. Malgaigne,
chirurgien de l’hôpital de la Charité,
professeur à la Faculté de médecine de
Paris. Paris, 1840, 3 vol. grand in-8 à
deux colonnes. 36 fr.
ÉLOGES LUS DANS LES 7 fr. 50
SÉANCES PUBLIQUES DE
L’ACADÉMIE ROYALE DE
CHIRURGIE DE 1750 A 1792, par
A. Louis, recueillis et publiés pour la
première fois, au nom de l’Académie
impériale de médecine, et d’après les
manuscrits originaux, avec une
Introduction, des notes et des
éclaircissements, par Fréd. Dubois
(d’Amiens), secrétaire perpétuel de
l’Académie impériale de médecine.
Paris, 1859, 1 vol. in-8 de 548 pages.
Cet ouvrage contient: Introduction historique par M.
Dubois, 76 pages; Éloges de J.-L. Petit, Bassuel, Malaval,
Verdier, Rœderer, Molinelli, Bertrandi, Foubert, Lecat,
Ledran, Pibrac, Benomont, Morand, Van Swieten,
Quesnay, Haller, Flurent, Willius, Lamartinière, Houstet,
de la Faye, Bordenave, David, Faure, Caqué, Faguer,
Camper, Hevin, Pipelet, et l’éloge de Louis, par Sue.
Embrassant tout un demi-siècle et renfermant, outre
les détails historiques et biographiques, des appréciations
et des jugements sur les faits, cette collection forme une
véritable histoire de la chirurgie française au XVIIIe siècle.

HISTOIRE DES MEMBRES DE

L’ACADÉMIE ROYALE DE MÉDECINE,
ou Recueil des éloges lus dans les
séances publiques, par E. Pariset,
secrétaire perpétuel de l’Académie de
médecine, etc., édition complète, publiée
sous les auspices de l’Académie, précédée
de l’éloge de Pariset, par F. Dubois
(d’Amiens), secrétaire perpétuel de
l’Académie de médecine. Paris, 1850, 2
beaux vol. in-12. 7 fr.
Cet ouvrage comprend: Discours d’ouverture de
l’Académie royale de médecine. — Éloge de Corvisart, —
Cadet de Gassicourt, — Berthollet, — Pinel, —
Beauchêne, — Bourru, — Percy, — Vauquelin, — G.
Cuvier, — Portal, — Chaussier, — Dupuytren, — Scarpa,
— Desgenettes, — Laennec, — Tessier, — Husard, —
Marc, — Lodibert, — Bourdois de la Motte, — Esquirol, —
Chevreul, — Larrey, — Lerminier, — A. Dubois, — Alibert,
— Geoffroy Saint-Hilaire, — A. Paré, — Broussais, —
Bichat, etc.

CODE MÉDICAL, ou Recueil des Lois, 4 fr.

Décrets et Règlements sur l’étude,
l’enseignement et l’exercice de la
médecine civile et militaire en France, par
Amédée Amette, secrétaire de la Faculté de
médecine de Paris. Troisième édition,
revue et augmentée. Paris, 1859, 1 vol.
in-12 de 560 pages.
Ouvrage traitant des droits et des devoirs des
médecins. Il s’adresse à tous ceux qui étudient,
enseignent ou exercent la médecine, et renferme dans un
ordre méthodique toutes les dispositions législatives et
réglementaires qui les concernent.

TRAITÉ DE LA CONTAGION,
pour servir à l’histoire des maladies
contagieuses et des épidémies, par Ch.
Anglada, professeur à la Faculté de
médecine de Montpellier. Paris, 1853, 2
vol. in-8. 12 fr.
DOCUMENTS INÉDITS SUR LA
GRANDE PESTE DE 1348.
(Consultation de la Faculté de Paris. —
Consultation d’un praticien de
Montpellier. — Description de G. de
Machault), par le docteur Joseph
Michon. Paris, 1860, in-8. 2 fr. 50
ÉTUDE PRATIQUE,
RÉTROSPECTIVE ET COMPARÉE,
SUR LE TRAITEMENT DES
ÉPIDÉMIES au XVIIIe siècle.
Appréciation des travaux et éloge de
Lepecq de la Cloture, médecin
épidémiographe de la Normandie, par
le docteur Max Simon. Paris, 1853, in-8
de 348 pages. 5 fr.
 ENVOI FRANCO CONTRE MANDAT SUR LA
POSTE.

Paris. — Imp. de E. Martinet, rue Mignon, 2.

Au lecteur.
Ce livre reproduit intégralement le texte original,
et l’orthographe d’origine a été conservée.
Cependant quelques erreurs typographiques ont été
corrigées. Les corrections sont soulignées en
pointillés gris. En positionnant le curseur sur ces
mots vous pouvez faire apparaître l'orthographe
originale. Les corrections indiquées dans l'Errata ont
également été apportées, et la ponctuation a fait
l'objet de quelques corrections mineures.
Les notes de bas de page ont été renumérotées
consécutivement et placées à la fin du livre.
*** END OF THE PROJECT GUTENBERG EBOOK ÉTUDE SUR LES
MALADIES ÉTEINTES ET LES MALADIES NOUVELLES ***

Updated editions will replace the previous one—the old editions

will be renamed.

Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States
copyright in these works, so the Foundation (and you!) can copy
and distribute it in the United States without permission and
without paying copyright royalties. Special rules, set forth in the
General Terms of Use part of this license, apply to copying and
distributing Project Gutenberg™ electronic works to protect the
PROJECT GUTENBERG™ concept and trademark. Project
Gutenberg is a registered trademark, and may not be used if
you charge for an eBook, except by following the terms of the
trademark license, including paying royalties for use of the
Project Gutenberg trademark. If you do not charge anything for
copies of this eBook, complying with the trademark license is
very easy. You may use this eBook for nearly any purpose such
as creation of derivative works, reports, performances and
research. Project Gutenberg eBooks may be modified and
printed and given away—you may do practically ANYTHING in
the United States with eBooks not protected by U.S. copyright
law. Redistribution is subject to the trademark license, especially
commercial redistribution.

START: FULL LICENSE

Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com