Download Full ebookname - Read Now at ebookname.

com

Exploratory Data Analysis with MATLAB Second

Edition Chapman Hall CRC Computer Science Data
Analysis Wendy L. Martinez

https://ebookname.com/product/exploratory-data-analysis-
with-matlab-second-edition-chapman-hall-crc-computer-
science-data-analysis-wendy-l-martinez/

OR CLICK BUTTON

DOWLOAD EBOOK

Discover More Ebook - Explore Now at ebookname.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Think Stats Exploratory Data Analysis Second Edition Allen

B. Downey

https://ebookname.com/product/think-stats-exploratory-data-analysis-
second-edition-allen-b-downey/

ebookname.com

Computational Statistics Handbook with MATLAB 3rd Edition

Wendy L. Martinez

https://ebookname.com/product/computational-statistics-handbook-with-
matlab-3rd-edition-wendy-l-martinez/

ebookname.com

Environmental Data Analysis with MatLab 1st Edition

William Menke

https://ebookname.com/product/environmental-data-analysis-with-
matlab-1st-edition-william-menke/

ebookname.com

Politics and Parentela in Paraiba A Case Study of Family

Based Oligarchy in Brazil Linda Lewin

https://ebookname.com/product/politics-and-parentela-in-paraiba-a-
case-study-of-family-based-oligarchy-in-brazil-linda-lewin/

ebookname.com
The Routledge Handbook of Multilingualism 1st Edition
Marilyn Martin-Jones

https://ebookname.com/product/the-routledge-handbook-of-
multilingualism-1st-edition-marilyn-martin-jones/

ebookname.com

Ecological Consequences of Climate Change Mechanisms

Conservation and Management 1st Edition Erik A. Beever

https://ebookname.com/product/ecological-consequences-of-climate-
change-mechanisms-conservation-and-management-1st-edition-erik-a-
beever/
ebookname.com

Setting Priorities for HIV AIDS Interventions A Cost

Benefit Approach 1st Edition Robert J. Brent

https://ebookname.com/product/setting-priorities-for-hiv-aids-
interventions-a-cost-benefit-approach-1st-edition-robert-j-brent/

ebookname.com

Smith s General Urology 17th Edition Emil Tanagho

https://ebookname.com/product/smith-s-general-urology-17th-edition-
emil-tanagho/

ebookname.com

Advances in Astronomy From the Big Bang to the Solar

System 2005 en 417s J. M. T. Thompson

https://ebookname.com/product/advances-in-astronomy-from-the-big-bang-
to-the-solar-system-2005-en-417s-j-m-t-thompson/

ebookname.com
The Arizona state constitution First Edition Leshy

https://ebookname.com/product/the-arizona-state-constitution-first-
edition-leshy/

ebookname.com
 Exploratory Data Analysis
with MATLAB®
Second Edition

K10616_FM.indd 1 11/15/10 11:50 AM

 Chapman & Hall/CRC
Computer Science and Data Analysis Series

The interface between the computer and statistical sciences is increasing, as each discipline
seeks to harness the power and resources of the other. This series aims to foster the integration
between the computer sciences and statistical, numerical, and probabilistic methods by
publishing a broad range of reference works, textbooks, and handbooks.

SERIES EDITORS
David Blei, Princeton University
David Madigan, Rutgers University
Marina Meila, University of Washington
Fionn Murtagh, Royal Holloway, University of London

Proposals for the series should be sent directly to one of the series editors above, or submitted to:

Chapman & Hall/CRC

4th Floor, Albert House
1-4 Singer Street
London EC2A 4BQ
UK

Published Titles

Bayesian Artificial Intelligence, Second Edition Introduction to Data Technologies

Kevin B. Korb and Ann E. Nicholson Paul Murrell
Clustering for Data Mining: A Data Recovery Introduction to Machine Learning
Approach and Bioinformatics
Boris Mirkin Sushmita Mitra, Sujay Datta,
Computational Statistics Handbook with Theodore Perkins, and George Michailidis
®
MATLAB , Second Edition Microarray Image Analysis:
Wendy L. Martinez and Angel R. Martinez An Algorithmic Approach
Correspondence Analysis and Data Karl Fraser, Zidong Wang, and Xiaohui Liu
Coding with Java and R Pattern Recognition Algorithms for
Fionn Murtagh Data Mining
Design and Modeling for Computer Sankar K. Pal and Pabitra Mitra
Experiments R Graphics
Kai-Tai Fang, Runze Li, and Agus Sudjianto Paul Murrell
Exploratory Data Analysis with MATLAB ,
®
R Programming for Bioinformatics
Second Edition Robert Gentleman
Wendy L. Martinez, Angel R. Martinez, Semisupervised Learning for Computational
and Jeffrey L. Solka Linguistics
Exploratory Multivariate Analysis by Steven Abney
Example Using R Statistical Computing with R
François Husson, Sébastien Lê, and Maria L. Rizzo
Jérôme Pagès

K10616_FM.indd 2 11/15/10 11:50 AM

 Exploratory Data Analysis
with MATLAB®
Second Edition

Wendy L. Martinez
Angel R. Martinez
Jeffrey L. Solka

K10616_FM.indd 3 11/15/10 11:50 AM

 MATLAB® and Simulink® are trademarks of The MathWorks, Inc. and are used with permission. The Math-
Works does not warrant the accuracy of the text or exercises in this book. This book’s use or discussion
of MATLAB® and Simulink® software or related products does not constitute endorsement or sponsorship
by The MathWorks of a particular pedagogical approach or particular use of the MATLAB® and Simulink®
software.

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2011 by Taylor and Francis Group, LLC

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

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1

International Standard Book Number: 978-1-4398-1220-4 (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, transmit-
ted, 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

Martinez, Wendy L.
Exploratory data analysis with MATLAB / Wendy L. Martinez, Angel Martinez,
Jeffrey Solka. -- 2nd ed.
p. cm. -- (Chapman & Hall/CRC computer science and data analysis series)
Includes bibliographical references and index.
ISBN 978-1-4398-1220-4 (hardback)
1. Multivariate analysis. 2. MATLAB. 3. Mathematical statistics. I. Martinez, Angel
R. II. Solka, Jeffrey L., 1955- III. Title.

QA278.M3735 2010
519.5’35--dc22 2010044042

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com

K10616_FM.indd 4 11/15/10 11:50 AM

EDA2ed.book Page i Thursday, November 11, 2010 8:51 AM

Angel and Wendy dedicate this book to their children:

Deborah,
Jeffrey,
Robbi (the middle child),
and Lisa (Principessa)

Jeffrey dedicates this book to his wife Beth,

sons Stephen and Rob, and
future daughter-in-law Rebecca
EDA2ed.book Page vii Thursday, November 11, 2010 8:51 AM

Table of Contents

Preface to the Second Edition...........................................................................xiii

Preface to the First Edition...............................................................................xvii

Part I
Introduction to Exploratory Data Analysis

Chapter 1
Introduction to Exploratory Data Analysis
1.1 What is Exploratory Data Analysis ............................................................. 3
1.2 Overview of the Text ..................................................................................... 6
1.3 A Few Words about Notation ...................................................................... 8
1.4 Data Sets Used in the Book ........................................................................... 9
1.4.1 Unstructured Text Documents ........................................................ 9
1.4.2 Gene Expression Data ..................................................................... 12
1.4.3 Oronsay Data Set ............................................................................. 18
1.4.4 Software Inspection ......................................................................... 19
1.5 Transforming Data ....................................................................................... 20
1.5.1 Power Transformations .................................................................. 21
1.5.2 Standardization ................................................................................ 22
1.5.3 Sphering the Data ............................................................................ 24
1.6 Further Reading ........................................................................................... 25
Exercises .............................................................................................................. 27

Part II
EDA as Pattern Discovery

Chapter 2
Dimensionality Reduction — Linear Methods
2.1 Introduction .................................................................................................. 31
2.2 Principal Component Analysis — PCA .................................................... 33
2.2.1 PCA Using the Sample Covariance Matrix ................................. 34
2.2.2 PCA Using the Sample Correlation Matrix ................................. 37
2.2.3 How Many Dimensions Should We Keep? ................................. 38
2.3 Singular Value Decomposition — SVD .................................................... 42
2.4 Nonnegative Matrix Factorization ............................................................ 47

vii
EDA2ed.book Page viii Thursday, November 11, 2010 8:51 AM

viii Exploratory Data Analysis with MATLAB® , 2nd Edition

2.5 Factor Analysis ............................................................................................. 51

2.6 Fisher’s Linear Discriminant ...................................................................... 56
2.7 Intrinsic Dimensionality .............................................................................. 61
2.7.1 Nearest Neighbor Approach ......................................................... 63
2.7.2 Correlation Dimension ................................................................... 67
2.7.3 Maximum Likelihood Approach .................................................. 68
2.7.4 Estimation Using Packing Numbers ............................................ 70
2.8 Summary and Further Reading ................................................................. 72
Exercises .............................................................................................................. 73

Chapter 3
Dimensionality Reduction — Nonlinear Methods
3.1 Multidimensional Scaling — MDS ............................................................ 79
3.1.1 Metric MDS ...................................................................................... 81
3.1.2 Nonmetric MDS ............................................................................... 91
3.2 Manifold Learning ....................................................................................... 99
3.2.1 Locally Linear Embedding ............................................................. 99
3.2.2 Isometric Feature Mapping — ISOMAP .................................... 101
3.2.3 Hessian Eigenmaps ....................................................................... 103
3.3 Artificial Neural Network Approaches .................................................. 108
3.3.1 Self-Organizing Maps ................................................................... 108
3.3.2 Generative Topographic Maps .................................................... 111
3.3.3 Curvilinear Component Analysis ............................................... 116
3.4 Summary and Further Reading ............................................................... 121
Exercises ............................................................................................................ 122

Chapter 4
Data Tours
4.1 Grand Tour ................................................................................................. 126
4.1.1 Torus Winding Method ................................................................ 127
4.1.2 Pseudo Grand Tour ....................................................................... 129
4.2 Interpolation Tours .................................................................................... 132
4.3 Projection Pursuit ....................................................................................... 134
4.4 Projection Pursuit Indexes ........................................................................ 142
4.4.1 Posse Chi-Square Index ................................................................ 142
4.4.2 Moment Index ................................................................................ 145
4.5 Independent Component Analysis ......................................................... 147
4.6 Summary and Further Reading ............................................................... 151
Exercises ............................................................................................................ 152

Chapter 5
Finding Clusters
5.1 Introduction ................................................................................................ 155
5.2 Hierarchical Methods ................................................................................ 157
EDA2ed.book Page ix Thursday, November 11, 2010 8:51 AM

Table of Contents ix

5.3 Optimization Methods — k-Means ......................................................... 163

5.4 Spectral Clustering ..................................................................................... 167
5.5 Document Clustering ................................................................................ 171
5.5.1 Nonnegative Matrix Factorization — Revisited ....................... 173
5.5.2 Probabilistic Latent Semantic Analysis ...................................... 177
5.6 Evaluating the Clusters ............................................................................. 182
5.6.1 Rand Index ..................................................................................... 182
5.6.2 Cophenetic Correlation ................................................................ 185
5.6.3 Upper Tail Rule .............................................................................. 186
5.6.4 Silhouette Plot ................................................................................ 189
5.6.5 Gap Statistic .................................................................................... 191
5.7 Summary and Further Reading ............................................................... 197
Exercises ............................................................................................................ 200

Chapter 6
Model-Based Clustering
6.1 Overview of Model-Based Clustering .................................................... 205
6.2 Finite Mixtures ........................................................................................... 207
6.2.1 Multivariate Finite Mixtures ........................................................ 210
6.2.2 Component Models — Constraining the Covariances ............ 211
6.3 Expectation-Maximization Algorithm .................................................... 217
6.4 Hierarchical Agglomerative Model-Based Clustering ......................... 222
6.5 Model-Based Clustering ............................................................................ 224
6.6 MBC for Density Estimation and Discriminant Analysis .................... 231
6.6.1 Introduction to Pattern Recognition ........................................... 231
6.6.2 Bayes Decision Theory .................................................................. 232
6.6.3 Estimating Probability Densities with MBC .............................. 235
6.7 Generating Random Variables from a Mixture Model ......................... 239
6.8 Summary and Further Reading ............................................................... 241
Exercises ............................................................................................................ 244

Chapter 7
Smoothing Scatterplots
7.1 Introduction ................................................................................................ 247
7.2 Loess ............................................................................................................. 248
7.3 Robust Loess ............................................................................................... 259
7.4 Residuals and Diagnostics with Loess .................................................... 261
7.4.1 Residual Plots ................................................................................. 261
7.4.2 Spread Smooth ............................................................................... 265
7.4.3 Loess Envelopes — Upper and Lower Smooths ....................... 268
7.5 Smoothing Splines ..................................................................................... 269
7.5.1 Regression with Splines ................................................................ 270
7.5.2 Smoothing Splines ......................................................................... 272
7.5.3 Smoothing Splines for Uniformly Spaced Data ........................ 278
7.6 Choosing the Smoothing Parameter ....................................................... 281
EDA2ed.book Page x Thursday, November 11, 2010 8:51 AM

x Exploratory Data Analysis with MATLAB® , 2nd Edition

7.7 Bivariate Distribution Smooths ................................................................ 285

7.7.1 Pairs of Middle Smoothings ......................................................... 285
7.7.2 Polar Smoothing ............................................................................ 287
7.8 Curve Fitting Toolbox ............................................................................... 291
7.9 Summary and Further Reading ............................................................... 293
Exercises ............................................................................................................ 294

Part III
Graphical Methods for EDA

Chapter 8
Visualizing Clusters
8.1 Dendrogram ................................................................................................ 301
8.2 Treemaps ..................................................................................................... 303
8.3 Rectangle Plots ........................................................................................... 306
8.4 ReClus Plots ................................................................................................ 312
8.5 Data Image .................................................................................................. 317
8.6 Summary and Further Reading ............................................................... 323
Exercises ............................................................................................................ 324

Chapter 9
Distribution Shapes
9.1 Histograms .................................................................................................. 327
9.1.1 Univariate Histograms ................................................................. 327
9.1.2 Bivariate Histograms .................................................................... 334
9.2 Boxplots ....................................................................................................... 336
9.2.1 The Basic Boxplot .......................................................................... 337
9.2.2 Variations of the Basic Boxplot .................................................... 342
9.3 Quantile Plots ............................................................................................. 347
9.3.1 Probability Plots ............................................................................ 347
9.3.2 Quantile-Quantile Plot .................................................................. 349
9.3.3 Quantile Plot .................................................................................. 352
9.4 Bagplots ....................................................................................................... 354
9.5 Rangefinder Boxplot .................................................................................. 356
9.6 Summary and Further Reading ............................................................... 359
Exercises ............................................................................................................ 361

Chapter 10
Multivariate Visualization
10.1 Glyph Plots ................................................................................................ 365
10.2 Scatterplots ................................................................................................ 366
10.2.1 2-D and 3-D Scatterplots ............................................................. 368
10.2.2 Scatterplot Matrices ..................................................................... 371
10.2.3 Scatterplots with Hexagonal Binning ....................................... 372
EDA2ed.book Page xi Thursday, November 11, 2010 8:51 AM

Table of Contents xi

10.3 Dynamic Graphics ................................................................................... 374

10.3.1 Identification of Data .................................................................. 376
10.3.2 Linking ......................................................................................... 378
10.3.3 Brushing ........................................................................................ 381
10.4 Coplots ....................................................................................................... 384
10.5 Dot Charts ................................................................................................. 387
10.5.1 Basic Dot Chart ............................................................................ 387
10.5.2 Multiway Dot Chart .................................................................... 388
10.6 Plotting Points as Curves ........................................................................ 392
10.6.1 Parallel Coordinate Plots ............................................................ 393
10.6.2 Andrews’ Curves ......................................................................... 395
10.6.3 Andrews’ Images ......................................................................... 399
10.6.4 More Plot Matrices ...................................................................... 400
10.7 Data Tours Revisited ............................................................................... 403
10.7.1 Grand Tour ................................................................................... 404
10.7.2 Permutation Tour ........................................................................ 405
10.8 Biplots ........................................................................................................ 408
10.9 Summary and Further Reading ............................................................. 411
Exercises ............................................................................................................ 413

Appendix A
Proximity Measures
A.1 Definitions .................................................................................................. 417
A.1.1 Dissimilarities ............................................................................... 418
A.1.2 Similarity Measures ..................................................................... 420
A.1.3 Similarity Measures for Binary Data ......................................... 420
A.1.4 Dissimilarities for Probability Density Functions ................... 421
A.2 Transformations ........................................................................................ 422
A.3 Further Reading ........................................................................................ 423

Appendix B
Software Resources for EDA
B.1 MATLAB Programs .................................................................................. 425
B.2 Other Programs for EDA .......................................................................... 429
B.3 EDA Toolbox .............................................................................................. 431

Appendix C
Description of Data Sets ................................................................................... 433

Appendix D
Introduction to MATLAB
D.1 What Is MATLAB? .................................................................................... 439
D.2 Getting Help in MATLAB ....................................................................... 440
D.3 File and Workspace Management .......................................................... 440
EDA2ed.book Page xii Thursday, November 11, 2010 8:51 AM

xii Exploratory Data Analysis with MATLAB® , 2nd Edition

D.4 Punctuation in MATLAB ......................................................................... 443

D.5 Arithmetic Operators ............................................................................... 444
D.6 Data Constructs in MATLAB .................................................................. 444
Basic Data Constructs ............................................................................ 444
Building Arrays ...................................................................................... 445
Cell Arrays ............................................................................................... 445
Structures ................................................................................................. 447
D.7 Script Files and Functions ........................................................................ 447
D.8 Control Flow .............................................................................................. 448
for Loop ................................................................................................. 448
while Loop ............................................................................................. 449
if-else Statements ............................................................................... 449
switch Statement .................................................................................. 449
D.9 Simple Plotting .......................................................................................... 450
D.10 Where to get MATLAB Information .................................................... 452

Appendix E
MATLAB Functions
E.1 MATLAB ..................................................................................................... 455
E.2 Statistics Toolbox ....................................................................................... 457
E.3 Exploratory Data Analysis Toolbox ........................................................ 458
E.4 EDA GUI Toolbox ..................................................................................... 459

References ......................................................................................................... 475

Author Index .................................................................................................... 497
Subject Index..................................................................................................... 503
EDA2ed.book Page xiii Thursday, November 11, 2010 8:51 AM

Preface to the Second Edition

In the past several years, many advancements have been made in the area of
exploratory data analysis, and it soon became apparent that it was time to
update this text. In particular, many innovative approaches have been
developed for dimensionality reduction, clustering, and visualization.
We list below some of the major changes and additions in the second
edition.

• We added significant content to the chapter on linear dimension-

ality reduction. The new methods we discuss are nonnegative
matrix factorization and linear discriminant analysis. We also
expanded the set of methods that are available for estimating the
intrinsic dimensionality of a data set.
• Curvilinear component analysis is a nonlinear dimensionality
reduction method that is now described in Chapter 3. Curvilinear
component analysis was developed as an improvement to self-
organizing maps.
• A description of independent component analysis has been added
to the chapter on data tours.
• Several new clustering methods are not included in the text. These
include nonnegative matrix factorization, probabilistic latent
semantic analysis, and spectral-based clustering.
• We included a discussion of smoothing splines, along with a fast
spline method that works with uniformly spaced data.
• Several visualization methods have been added to the text. These
are a rangefinder boxplot for bivariate data, scatterplots with mar-
ginal histograms, biplots, and a new method called Andrews’
images.
• Many of the methods in the text are available via a GUI interface.
This free EDA GUI Toolbox is described in Appendix E.

In a spirit similar to the first edition, this text is not focused on the
theoretical aspects of the methods. Rather, the main focus of this book is on
the use of the EDA methods. So, we do not dwell so much on implementation
and algorithmic details. Instead, we show students and practitioners how the

xiii
EDA2ed.book Page xiv Thursday, November 11, 2010 8:51 AM

xiv Exploratory Data Analysis with MATLAB® , 2nd Edition

methods can be used for exploratory data analysis by providing examples

and applications.
The MATLAB® code for the examples, the toolboxes, the data sets, and
color versions of most figures are available for download. They can be
downloaded from the Carnegie Mellon StatLib site found here:

http://lib.stat.cmu.edu

or from the book’s website:

http://pi-sigma.info

Please review the readme file for installation instructions and information
on any changes.
For MATLAB product information, please contact:

The MathWorks, Inc.

3 Apple Hill Drive
Natick, MA, 01760-2098 USA
Tel: 508-647-7000
Fax: 508-647-7001
E-mail: info@mathworks.com
Web: www.mathworks.com

We would like to acknowledge the invaluable help of those researchers

who wrote MATLAB code for methods described in this book and also made
it available for free. In particular, the authors would like to thank Michael
Berry for helpful discussions regarding nonnegative matrix factorization and
Ata Kaban for allowing us to use her PLSI code. We are also very grateful to
Mia Hubert and Sabine Verboven for granting us permission to use their
bagplot function and for their patience with our emails.
We thank the editors of the book series in Computer Science and Data
Analysis for including this text. We greatly appreciate the help and patience
of those at CRC press: David Grubbs, Bob Stern, and Michele Dimont. As
always, we are indebted to Naomi Fernandes and Tom Lane at The
MathWorks, Inc. for their special assistance with MATLAB.

Disclaimers

1. Any MATLAB programs and data sets that are included with the
book are provided in good faith. The authors, publishers, or dis-
tributors do not guarantee their accuracy and are not responsible
for the consequences of their use.
2. Some of the MATLAB functions provided with the EDA Toolboxes
were written by other researchers, and they retain the copyright.
EDA2ed.book Page xv Thursday, November 11, 2010 8:51 AM

Preface to the Second Edition xv

References are given in Appendix B and in the help section of

each function. Unless otherwise specified, the EDA Toolboxes are
provided under the GNU license specifications:
http://www.gnu.org/copyleft/gpl.html
3. The views expressed in this book are those of the authors and do
not necessarily represent the views of the United States Department
of Defense or its components.

Wendy L. Martinez, Angel R. Martinez, and Jeffrey L. Solka

October 2010
EDA2ed.book Page xvii Thursday, November 11, 2010 8:51 AM

Preface to the First Edition

One of the goals of our first book, Computational Statistics Handbook with
MATLAB® [2002], was to show some of the key concepts and methods of
computational statistics and how they can be implemented in MATLAB.1 A
core component of computational statistics is the discipline known as
exploratory data analysis or EDA. Thus, we see this book as a complement to
the first one with similar goals: to make exploratory data analysis techniques
available to a wide range of users.
Exploratory data analysis is an area of statistics and data analysis, where
the idea is to first explore the data set, often using methods from descriptive
statistics, scientific visualization, data tours, dimensionality reduction, and
others. This exploration is done without any (hopefully!) pre-conceived
notions or hypotheses. Indeed, the idea is to use the results of the exploration
to guide and to develop the subsequent hypothesis tests, models, etc. It is
closely related to the field of data mining, and many of the EDA tools
discussed in this book are part of the toolkit for knowledge discovery and
data mining.
This book is intended for a wide audience that includes scientists,
statisticians, data miners, engineers, computer scientists, biostatisticians,
social scientists, and any other discipline that must deal with the analysis of
raw data. We also hope this book can be useful in a classroom setting at the
senior undergraduate or graduate level. Exercises are included with each
chapter, making it suitable as a textbook or supplemental text for a course in
exploratory data analysis, data mining, computational statistics, machine
learning, and others. Readers are encouraged to look over the exercises
because new concepts are sometimes introduced in them. Exercises are
computational and exploratory in nature, so there is often no unique answer!
As for the background required for this book, we assume that the reader
has an understanding of basic linear algebra. For example, one should have
a familiarity with the notation of linear algebra, array multiplication, a matrix
inverse, determinants, an array transpose, etc. We also assume that the
reader has had introductory probability and statistics courses. Here one
should know about random variables, probability distributions and density
functions, basic descriptive measures, regression, etc.
In a spirit similar to the first book, this text is not focused on the theoretical
aspects of the methods. Rather, the main focus of this book is on the use of the

1 MATLAB® and Handle Graphics ® are registered trademarks of The MathWorks, Inc.

xvii
EDA2ed.book Page xviii Thursday, November 11, 2010 8:51 AM

xviii Exploratory Data Analysis with MATLAB® , 2nd Edition

EDA methods. Implementation of the methods is secondary, but where

feasible, we show students and practitioners the implementation through
algorithms, procedures, and MATLAB code. Many of the methods are
complicated, and the details of the MATLAB implementation are not
important. In these instances, we show how to use the functions and
techniques. The interested reader (or programmer) can consult the M-files for
more information. Thus, readers who prefer to use some other programming
language should be able to implement the algorithms on their own.
While we do not delve into the theory, we would like to emphasize that the
methods described in the book have a theoretical basis. Therefore, at the end
of each chapter, we provide additional references and resources; so those
readers who would like to know more about the underlying theory will
know where to find the information.
MATLAB code in the form of an Exploratory Data Analysis Toolbox is
provided with the text. This includes the functions, GUIs, and data sets that
are described in the book. This is available for download at

http://lib.stat.cmu.edu

Please review the readme file for installation instructions and information
on any changes. M-files that contain the MATLAB commands for the
exercises are also available for download.
We also make the disclaimer that our MATLAB code is not necessarily the
most efficient way to accomplish the task. In many cases, we sacrificed
efficiency for clarity. Please refer to the example M-files for alternative
MATLAB code, courtesy of Tom Lane of The MathWorks, Inc.
We describe the EDA Toolbox in greater detail in Appendix B. We also
provide website information for other tools that are available for download
(at no cost). Some of these toolboxes and functions are used in the book and
others are provided for informational purposes. Where possible and
appropriate, we include some of this free MATLAB code with the EDA
Toolbox to make it easier for the reader to follow along with the examples
and exercises.
We assume that the reader has the Statistics Toolbox (Version 4 or higher)
from The MathWorks, Inc. Where appropriate, we specify whether the
function we are using is in the main MATLAB software package, Statistics
Toolbox, or the EDA Toolbox. The development of the EDA Toolbox was
mostly accomplished with MATLAB Version 6.5 (Statistics Toolbox, Version
4); so the code should work if this is what you have. However, a new release
of MATLAB and the Statistics Toolbox was introduced in the middle of
writing th is book; so we also in corp orate information abou t n ew
functionality provided in these versions.
We would like to acknowledge the invaluable help of the reviewers: Chris
Fraley, David Johannsen, Catherine Loader, Tom Lane, David Marchette,
and Jeffrey Solka. Their many helpful comments and suggestions resulted in
a better book. Any shortcomings are the sole responsibility of the authors. We
EDA2ed.book Page xix Thursday, November 11, 2010 8:51 AM

Preface to the First Edition xix

owe a special thanks to Jeffrey Solka for programming assistance with finite
mixtures and to Rich ard Johnson for allowin g us to use h is Data
Visualization Toolbox and updating his functions. We would also like to
acknowledge all of those researchers who wrote MATLAB code for methods
described in this book and also made it available for free. We thank the
editors of the book series in Computer Science and Data Analysis for
including this text. We greatly appreciate the help and patience of those at
CRC Press: Bob Stern, Rob Calver, Jessica Vakili, and Andrea Demby. Finally,
we are indebted to Naomi Fernandes and Tom Lane at The MathWorks, Inc.
for their special assistance with MATLAB.

Disclaimers

1. Any MATLAB programs and data sets that are included with the
book are provided in good faith. The authors, publishers, or dis-
tributors do not guarantee their accuracy and are not responsible
for the consequences of their use.
2. Some of the MATLAB functions provided with the EDA Toolbox
were written by other researchers, and they retain the copyright.
References are given in Appendix B and in the help section of
each function. Unless otherwise specified, the EDA Toolbox is pro-
vided under the GNU license specifications:
http://www.gnu.org/copyleft/gpl.html
3. The views expressed in this book are those of the authors and do
not necessarily represent the views of the United States Department
of Defense or its components.

Wendy L. and Angel R. Martinez

October 2004
EDA2ed.book Page 1 Thursday, November 11, 2010 8:51 AM

Part I
Introduction to Exploratory Data Analysis
EDA2ed.book Page 3 Thursday, November 11, 2010 8:51 AM

Chapter 1
Introduction to Exploratory Data Analysis

We shall not cease from exploration

And the end of all our exploring
Will be to arrive where we started
And know the place for the first time.

T. S. Eliot, “Little Gidding” (the last of his Four Quartets)

The purpose of this chapter is to provide some introductory and background

information. First, we cover the philosophy of exploratory data analysis and
discuss how this fits in with other data analysis techniques and objectives.
This is followed by an overview of the text, which includes the software that
will be used and the background necessary to understand the methods. We
then present several data sets that will be employed throughout the book to
illustrate the concepts and ideas. Finally, we conclude the chapter with some
information on data transforms, which will be important in some of the
methods presented in the text.

1.1 What is Exploratory Data Analysis

John W. Tukey [1977] was one of the first statisticians to provide a detailed
description of exploratory data analysis (EDA). He defined it as “detective
work - numerical detective work - or counting detective work - or graphical
detective work” [Tukey, 1977, page 1]. It is mostly a philosophy of data
analysis where the researcher examines the data without any pre-conceived
ideas in order to discover what the data can tell him or her about the
phenomena being studied. Tukey contrasts this with confirmatory data
analysis (CDA), an area of data analysis that is mostly concerned with
statistical hypothesis testing, confidence intervals, estimation, etc. Tukey
[1977] states that “Confirmatory data analysis is judicial or quasi-judicial in
character.” CDA methods typically involve the process of making inferences
about or estimates of some population characteristic and then trying to

3
EDA2ed.book Page 4 Thursday, November 11, 2010 8:51 AM

4 Exploratory Data Analysis with MATLAB® , 2nd Edition

evaluate the precision associated with the results. EDA and CDA should not
be used separately from each other, but rather they should be used in a
complementary way. The analyst explores the data looking for patterns and
structure that leads to hypotheses and models.
Tukey’s book on EDA was written at a time when computers were not
widely available and the data sets tended to be somewhat small, especially
by today’s standards. So, Tukey developed methods that could be
accomplished using pencil and paper, such as the familiar box-and-whisker
plots (also known as boxplots) and the stem-and-leaf. He also included
discussions of data transformation, smoothing, slicing, and others. Since our
book is written at a time when computers are widely available, we go beyond
what Tukey used in EDA and present computationally intensive methods for
pattern discovery and statistical visualization. However, our philosophy of
EDA is the same - that those engaged in it are data detectives.
Tukey [1980], expanding on his ideas of how exploratory and confirmatory
data analysis fit together, presents a typical straight-line methodology for
CDA; its steps follow:

1. State the question(s) to be investigated.

2. Design an experiment to address the questions.
3. Collect data according to the designed experiment.
4. Perform a statistical analysis of the data.
5. Produce an answer.

This procedure is the heart of the usual confirmatory process. To incorporate

EDA, Tukey revises the first two steps as follows:

1. Start with some idea.

2. Iterate between asking a question and creating a design.

Forming the question involves issues such as: What can or should be asked?
What designs are possible? How likely is it that a design will give a useful
answer? The ideas and methods of EDA play a role in this process. In
conclusion, Tukey states that EDA is an attitude, a flexibility, and some graph
paper.
A small, easily read book on EDA written from a social science perspective
is the one by Hartwig and Dearing [1979]. They describe the CDA mode as
one that answers questions such as “Do the data confirm hypothesis XYZ?”
Whereas, EDA tends to ask “What can the data tell me about relationship
XYZ?” Hartwig and Dearing specify two principles for EDA: skepticism and
openness. This might involve visualization of the data to look for anomalies
or patterns, the use of resistant (or robust) statistics to summarize the data,
openness to the transformation of the data to gain better insights, and the
generation of models.
EDA2ed.book Page 5 Thursday, November 11, 2010 8:51 AM

Introduction to Exploratory Data Analysis 5

Some of the ideas of EDA and their importance to teaching statistics were
discussed by Chatfield [1985]. He called the topic initial data analysis or
IDA. While Chatfield agrees with the EDA emphasis on starting with the
noninferential approach in data analysis, he also stresses the need for looking
at how the data were collected, what are the objectives of the analysis, and
the use of EDA/IDA as part of an integrated approach to statistical inference.
Hoaglin [1982] provides a summary of EDA in the Encyclopedia of Statistical
Sciences. He describes EDA as the “flexible searching for clues and evidence”
and confirmatory data analysis as “evaluating the available evidence.” In his
summary, he states that EDA encompasses four themes: resistance, residuals,
re-expression, and display.
Resistant data analysis pertains to those methods where an arbitrary
change in a data point or small subset of the data yields a small change in the
result. A related idea is robustness, which has to do with how sensitive an
analysis is to departures from the assumptions of an underlying probabilistic
model.
Residuals are what we have left over after a summary or fitted model has
been subtracted out. We can write this as

residual = data – fit.

The idea of examining residuals is common practice today. Residuals should

be looked at carefully for lack of fit, heteroscedasticity (nonconstant
variance), nonadditivity, and other interesting characteristics of the data.
Re-expression has to do with the transformation of the data to some other
scale that might make the variance constant, might yield symmetric
residuals, could linearize the data, or add some other effect. The goal of re-
expression for EDA is to facilitate the search for structure, patterns, or other
information.
Finally, we have the importance of displays or visualization techniques for
EDA. As we described previously, the displays used most often by early
practitioners of EDA included the stem-and-leaf plots and boxplots. The use
of scientific and statistical visualization is fundamental to EDA, because
often the only way to discover patterns, structure, or to generate hypotheses
is by visual transformations of the data.
Given the increased capabilities of computing and data storage, where
massive amounts of data are collected and stored simply because we can do
so and not because of some designed experiment, questions are often
generated after the data have been collected [Hand, Mannila, and Smyth,
2001; Wegman, 1988]. Perhaps there is an evolution of the concept of EDA in
the making and the need for a new philosophy of data analysis.
EDA2ed.book Page 6 Thursday, November 11, 2010 8:51 AM

6 Exploratory Data Analysis with MATLAB® , 2nd Edition

1.2 Overview of the Text

This book is divided into two main sections: pattern discovery and graphical
EDA. We first cover linear and nonlinear dimensionality reduction because
sometimes structure is discovered or can only be discovered with fewer
dimensions or features. We include some classical techniques such as
principal component analysis, factor analysis, and multidimensional scaling,
as well as some computationally intensive methods. For example, we discuss
self-organizing maps, locally linear embedding, isometric feature mapping,
generative topographic maps, curvilinear component analysis, and more.
Searching the data for insights and information is fundamental to EDA. So,
we describe several methods that ‘tour’ the data looking for interesting
structure (holes, outliers, clusters, etc.). These are variants of the grand tour
and projection pursuit that try to look at the data set in many 2-D or 3-D
views in the hope of discovering something interesting and informative.
Clustering or unsupervised learning is a standard tool in EDA and data
mining. These methods look for groups or clusters, and some of the issues
that must be addressed involve determining the number of clusters and the
validity or strength of the clusters. Here we cover some of the classical
methods such as hierarchical clustering and k-means, as well as an innovative
approach called nonnegative matrix factorization. We also devote an entire
chapter to a lesser-known technique called model-based clustering that
includes a way to determine the number of clusters and to assess the
resulting clusters.
Evaluating the relationship between variables is an important subject in
data analysis. We do not cover the standard regression methodology; it is
assumed that the reader already understands that subject. Instead, we
include a chapter on scatterplot smoothing techniques such as loess and
smoothing splines.
The second section of the book discusses many of the standard techniques
of visualization for EDA. The reader will note, however, that graphical
techniques, by necessity, are used throughout the book to illustrate ideas and
concepts.
In this section, we provide some classic, as well as some novel ways of
visualizing the results of the cluster process, such as dendrograms, silhouette
plots, treemaps, rectangle plots, and ReClus. These visualization techniques
can be used to assess the output from the various clustering algorithms that
were covered in the first section of the book. Distribution shapes can tell us
important things about the underlying phenomena that produced the data.
We will look at ways to determine the shape of the distribution by using
boxplots, bagplots, q-q plots, histograms, and others.
Finally, we present ways to visualize multivariate data. These include
parallel coordinate plots, scatterplot matrices, glyph plots, coplots, dot
EDA2ed.book Page 7 Thursday, November 11, 2010 8:51 AM

Introduction to Exploratory Data Analysis 7

charts, Andrews’ curves, and Andrews’ images. The ability to interact with
the plot to uncover structure or patterns is important, and we present some
of the standard methods such as linking and brushing. We also connect both
sections by revisiting the idea of the grand tour and show how that can be
implemented with Andrews’ curves and parallel coordinate plots.
We realize that other topics can be considered part of EDA, such as
descriptive statistics, outlier detection, robust data analysis, probability
density estimation, and residual analysis. However, these topics are beyond
the scope of this book. Descriptive statistics are covered in introductory
statistics texts, and since we assume that readers are familiar with this subject
matter, there is no need to provide explanations here. Similarly, we do not
emphasize residual analysis as a stand-alone subject, mostly because this is
widely discussed in other books on regression and multivariate analysis.
We do cover some density estimation, such as model-based clustering
(Chapter 6) and histograms (Chapter 9). The reader is referred to Scott [1992]
for an excellent treatment of the theory and methods of multivariate density
estimation in general or Silverman [1986] for kernel density estimation. For
more information on MATLAB implementations of density estimation the
reader can refer to Martinez and Martinez [2007]. Finally, we will likely
encounter outlier detection as we go along in the text, but this topic, along
with robust statistics, will not be covered as a stand-alone subject. There are
several books on outlier detection and robust statistics. These include
Hoaglin, Mosteller, and Tukey [1983], Huber [1981], and Rousseeuw and
Leroy [1987]. A rather dated paper on the topic is Hogg [1974].
We use MATLAB® throughout the book to illustrate the ideas and to show
how they can be implemented in software. Much of the code used in the
examples and to create the figures is freely available, either as part of the
downloadable toolbox included with the book or on other internet sites. This
information will be discussed in more detail in Appendix B. For MATLAB
product information, please contact:

The MathWorks, Inc.

3 Apple Hill Drive
Natick, MA, 01760-2098 USA
Tel: 508-647-7000
Fax: 508-647-7001
E-mail: info@mathworks.com
Web: www.mathworks.com

It is important for the reader to understand what versions of the software or

what toolboxes are used with this text. The book was written using MATLAB
Version 7.10 (R2010a), and we made some use of the MATLAB Statistics
Toolbox. We will refer to the Curve Fitting Toolbox in Chapter 7, where we
discuss smoothing. However, this particular toolbox is not needed to use the
examples in the book.
EDA2ed.book Page 8 Thursday, November 11, 2010 8:51 AM

8 Exploratory Data Analysis with MATLAB® , 2nd Edition

To get the most out of this book, readers should have a basic understanding
of matrix algebra. For example, one should be familiar with determinants, a
matrix transpose, the trace of a matrix, etc. We recommend Strang [1988,
1993] for those who need to refresh their memories on the topic. We do not
use any calculus in this book, but a solid understanding of algebra is always
useful in any situation. We expect readers to have knowledge of the basic
concepts in probability and statistics, such as random samples, probability
distributions, hypothesis testing, and regression.

1.3 A Few Words about Notation

In this section, we explain our notation and font conventions. MATLAB code
will be in Courier New bold font such as this: function. To make the book
more readable, we will indent MATLAB code when we have several lines of
code, and this can always be typed in as you see it in the book.
For the most part, we follow the convention that a vector is arranged as a
column, so it has dimensions p × 1. 1 In most situations, our data sets will be
arranged in a matrix of dimension n × p , which is denoted as X. Here n
represents the number of observations we have in our sample, and p is the
number of variables or dimensions. Thus, each row corresponds to a p-
dimensional observation or data point. The ij-th element of X will be
represented by xij. For the most part, the subscript i refers to a row in a matrix
or an observation, and a subscript j references a column in a matrix or a
variable. What is meant by this will be clear from the text.
In many cases, we might need to center our observations before we analyze
them. To make the notation somewhat simpler later on, we will use the
matrix X c to represent our centered data matrix, where each row is now
centered at the origin. We calculate this matrix by first finding the mean of
each column of X and then subtracting it from each row. The following code
will calculate this in MATLAB:
% Find the mean of each column.
[n,p] = size(X);
xbar = mean(X);
% Create a matrix where each row is the mean
% and subtract from X to center at origin.
Xc = X - repmat(xbar,n,1);

1The notation m × n is read “m by n,” and it means that we have m rows and n columns in an
array. It will be clear from the context whether this indicates matrix dimensions or
multiplication.
EDA2ed.book Page 9 Thursday, November 11, 2010 8:51 AM

Introduction to Exploratory Data Analysis 9

1.4 Data Sets Used in the Book

In this section, we describe the main data sets that will be used throughout
the text. Other data sets will be used in the exercises and in some of the
examples. This section can be set aside and read as needed without any loss
of continuity. Please see Appendix C for detailed information on all data sets
included with the text.

1.4.1 Unstructured Text Documents

The ability to analyze free-form text documents (e.g., Internet documents,
intelligence reports, news stories, etc.) is an important application in
computational statistics. We must first encode the documents in some
numeric form in order to apply computational methods. The usual way this
is accomplished is via a term-document matrix, where each row of the matrix
corresponds to a word in the lexicon, and each column represents a
document. The elements of the term-document matrix contain the number of
times the i-th word appears in j-th document [Manning and Schütze, 2000;
Charniak, 1996]. One of the drawbacks to this type of encoding is that the
order of the words is lost, resulting in a loss of information [Hand, Mannila,
and Smyth, 2001].
We now present a novel method for encoding unstructured text documents
where the order of the words is accounted for. The resulting structure is
called the bigram proximity matrix (BPM).

Bigram Proximity Matrices

The bigram proximity matrix (BPM) is a nonsymmetric matrix that captures
the number of times word pairs occur in a section of text [Martinez and
Wegman, 2002a; 2002b]. The BPM is a square matrix whose column and row
headings are the alphabetically ordered entries of the lexicon. Each element
of the BPM is the number of times word i appears immediately before word
j in the unit of text. The size of the BPM is determined by the size of the
lexicon created by alphabetically listing the unique occurrences of the words
in the corpus. In order to assess the usefulness of the BPM encoding we had
to determine whether or not the representation preserves enough of the
semantic content to make them separable from BPMs of other thematically
unrelated collections of documents.
We must make some comments about the lexicon and the pre-processing
of the documents before proceeding with more information on the BPM and
the data provided with this book. All punctuation within a sentence, such as
commas, semi-colons, colons, etc., were removed. All end-of-sentence
Other documents randomly have
different content
One or two grey cats came in, hopefully, and sat on the ladder-
stairs, purring, observant, receptive.
The cows on test were in the western extension, all becoming a trifle
restless now that their hour was again approaching. And presently
two of Odell’s sons, Si and Willis, came in, scrubbed and clothed in
white, prepared to continue the exhaustive record already well
initiated.
“Eris home yet?” asked Odell over his shoulder.
Si shook his head and picked up a pail.
“Well, where’n the dang-dinged town is she?” growled Odell. “If
she’s staying som’mers to supper, why can’t she send word?”
Willis said: “Buddy went down street to look for her. Mommy sent
him.”
The boys passed on into the extension where the comely cattle on
test stood impatient.
Odell remarked to Lister: “Ever since Eris drove over to Summit to
see them pitcher people makin’ movies she’s acted sulky and
contrary like. Now look at her stayin’ away all day—’n’ out to supper,
too, som’mers.”
“She acts like she’s sot on sunthin’,” suggested Lister, adjusting his
milking stool and clasping the pail between his knees.
“She’s sot on j’ining some danged moving pitcher comp’ny,” grunted
Odell. “That’s what’s in her head all the time these days.”
Lister’s pail hummed with alternate streams of milk drumming on the
tin. For a while he milked in silence save for a low-voiced
remonstrance to the young and temperamental Guernsey whose
near hind leg threatened trouble.
As he rose with the brimming pail he said: “I guess Eris is a good
girl. I guess she wouldn’t go so far as to do nothin’ rash, Elmer.”
“I dunno. You couldn’t never tell what Fanny had in her head. Fanny
allus had her secret thoughts. I never knowed what she was figurin’
out. Eris acts that way; she does what she’s told but she thinks as
she’s a mind to. Too much brain ain’t healthy for no woman.”
Lister weighed his pail, scratched down the record opposite the
cow’s name, turned and looked back at Odell.
“Women oughta think the way their men-folks tell ’em,” he said.
“That’s my idee. But the way they vote and carry on these days is a-
sp’ilin’ on ’em, accordin’ to my way of figurin’.”
Odell said nothing. As he stood weighing his pail of milk, Buddy
came into the barn, eating a stick of shop candy.
“Say, pa,” he called out, “mommy wants you up to the house!”
“When? Now?” demanded his father in dull surprise.
“I guess so. She said you was to come right up.”
Odell placed the empty milk pail on the floor: “Eris home yet?”
“I dunno. I guess not. Will you let me milk Snow-bird, pa?”
“No. Look at your hands! You go up and shake down some hay....
Where’s your ma?”
“She’s up in Eris’ room. She says for you to come. Can’t I wash my
hands and——”
“No. G’wan up to the loft. And don’t step on the pitchfork, neither.”
He turned uncertainly toward Lister and found his father-in-law
looking at him.
“Kinda queer,” he muttered, “Mazie sending for me when she knows
I’m milking....”
Lister made no comment. Odell went out heavily, crossed the farm
yard in the pleasant sunset glow, walked on toward the house with
lagging stride.
As he set foot on the porch he became conscious of his irritation, felt
the heat of it in his cheeks—the same old familiar resentment which
had smouldered through the dingy, discordant years of his first
marriage.
Here it was again, creeping through him after all these placid years
with Mazie—the same sullen apprehension, dull unease verging on
anger, invading his peace of mind, stirred this time by Fanny’s child—
Eris, daughter of Discord.
“Dang Fanny’s breed,” he muttered, entering the house, “—we allus
was enemies deep down, ... deep down in the flesh....”
All at once he understood his real mind. Eris had always been
Fanny’s child. Never his. He remembered what Fanny had said to
him at the approach of death—how, in that last desperate moment
the battered mask of years had slipped from her bony visage and he
had gazed into the stark face of immemorial antipathy, ... the
measureless resentment of a sex.
Fanny was dead. May God find out what she wants and give it to
her. But Fanny’s race persisted. She lived again in Eris. He was face
to face with it again.... After twenty years of peace!...
He went to the foot of the stairs and called to his wife. Her voice
answered from the floor above. He plodded on upstairs.
Mazie was standing in Eris’ room, a pile of clothing on the bed, a
suitcase and a small, flat trunk open on the floor.
She turned to Odell, her handsome features flushed, and the sparkle
of tears in her slanting, black eyes.
“What’s the trouble now?” he demanded, already divining it.
“She’s gone, Elmer. She called me up on the telephone from Albany
to tell me. The Crystal Fillum Company offers her a contract. She
wants her clothes and her money.”
A heavy colour surged through the man’s face.
“That’s the danged secret blood in her,” he said. “I knowed it.
There’s allus sunthin’ hatchin’ deep down in women of her blood....
She’s allus had it in her mind to quit us.... She never was one of
us.... All right, let her go. I’m done with her.”
Mazie began unsteadily: “So many children of—of our day seem to
feel like our Eris——”
“Mine don’t! My boys ain’t got nothin’ secret into them! They ain’t
crazy in the head ’n’ they ain’t full o’ fool notions.”
Mazie remained silent. Her sons were fuller of “notions” than their
father knew. It had required all the magnetism of her affection and
authority to keep them headed toward a future on Whitewater
Farms. For the nearest town was already calling them; they sniffed
the soft-coal smoke from afar and were restless for the iron
dissonance and human bustle of paved and narrow ways.
Theirs was the gregarious excitement instinct in human animals.
Beyond the dingy monochrome of life they caught a glimmer of
distant brightness. The vague summons of unknown but suspected
pleasures stirred them as they travelled the sodden furrow.
Youth’s physical instinct is to gather at the water-hole of this vast
veldt we call the world, and wallow in the inviting mire of a thousand
hoofs, and feel and hear and see the perpetual milling of the human
herds that gather there.
Only in quality did Eris differ from her brothers. It was her mind—
and the untasted pleasures of the mind—that drove her to the
common fount.
There is a picture by Fragonard called “The Fountain of Love.” And,
as eagerly as the blond and glowing girl speeds to the brimming
basin where mischievous little winged Loves pour out for her the
magic waters, so impetuously had Eris sped toward the fount of
knowledge, hot, parched with desire to set her lips to immortal
springs.
Odell’s heavy eyes, brooding anger, followed Mazie’s movements as
she smoothed out the clothing and laid each garment in the trunk.
“You don’t have to do that,” he growled. “Let her come and get ’em
if she wants ’em.”
“But she needs——”
“Dang it, let ’em lay. Like’s not she’ll sicken o’ them pitcher people
before the week’s out. She’ll get her belly full o’ notions. Let her
caper till she runs into barbed wire. That’ll sting some sense into her
hide.”
“She only took her little leather bag, Elmer——”
“She’ll sicken sooner. I ain’t worryin’ none. She ain’t a loose girl;
she’s just a fool heifer that goes bucketin’ over a snake-fence where
it’s half down. Let her kick up and skylark. You bet she’ll hear the
farm bell when it comes supper time——”
He turned away exasperated, but Mazie took him by the sleeve of
his milking jacket:
“She’s got to have money, Elmer——”
“No, she hain’t! She’ll sicken the quicker——”
“Elmer, it’s her money.”
“’Tain’t. It’s mine.”
“It’s her heifer-money——”
“She shan’t have it! Not till she’s twenty-one. And that’s that!”
Mazie looked at her husband in a distressed way, her black eyes full
of tears:
“Elmer, you can’t use a girl like a boy. A girl’s a tender thing. And I
was afraid of this—something like this.... Because Eris is a mite
different. She likes to read and study. She likes to figure out what
she reads about. She likes music and statues and art-things like the
hand-painted pictures we saw in Utica. There’s no harm in art, I
guess.... And you know how she always did love to dress up for
church plays—and how nicely she sang and danced and acted in
school——”
“Dang it all!” shouted Odell, beating one tanned fist within the other
palm, “let her come home and cut her capers! She can do them
things when there’s a entertainment down to the church, can’t she?
“That’s enough for any girl, ain’t it? And she can go to Utica and look
at them hand-painted pitchers in the store windows. And she can
dance to socials and showers like sensible girls and she can sing her
head off Sundays in church when she’s a mind to!
“All she’s gotta do is come home and git the best of everything. But
as long as she acts crazy and stays away, I’m done with her. And
that’s that!”
 CHAPTER VII
SPRING had begun more than a month early. The young year
promised agricultural miracles. All omens were favourable. Ed Lister
predicted it would be a “hog-killin’.”
June’s magic turned Whitewater to a paradise. Crystal mornings
gradually warming until sundown; gentle showers at night to freshen
herbage and start a million planted seeds; blossoms, bees, buds,
blue skies—all exquisitely balanced designs in June’s enchanted
tapestry—and nothing so far to mar the fabric—no late and
malignant frost, no early drouth, broken violently by thunderbolt and
deluge; no hail; no heavy winds to dry and sear; nothing untoward
in the herd,—no milk-fever, no abortion, no terrifying emergency at
night.
The only things to irritate Odell were the letters from Eris. They
aroused in him the dumb, familiar anger of Fanny’s time.

But after the first week in July there were no longer any letters from
Eris. The girl had written two or three times during June, striving to
explain herself, to make him understand her need of doing as she
was doing, the necessity that some of her own money be sent her.
Her last letter arrived about the beginning of that dreadful era of
unprecedented heat and drouth which ushered in July and which
caused that summer to be long remembered in the Old World as well
as in the New.
Odell’s refusal to send her a single penny, and his repeated
summons for her return had finally silenced Eris. No more letters
came. Odell’s attitude silenced Mazie, too, whose primitive sense of
duty was to her man first of all.
Sometimes she ventured to hope that Eris might, somehow, be
successful. Oftener a comforting belief reassured her that the girl
would soon return to material comforts and female duties, which
were all Mazie comprehended of earthly happiness.
Odell’s refusal to send Eris her money and her clothes worried Mazie
when she had time to think. But what could she do? Man ruled
Mazie’s universe. It was proper that he should. All her life she had
had to submit to him,—she had to cook for him, wash, sew, mend,
care for his habitation, bear his children, fed them, wean them, and,
in the endless sequence again, cook, wash, iron, sew, mend for
these men-children which she had borne her man. And it was
proper. It was the way of the world. Of heaven, too, perhaps. God
himself was masculine.... She sometimes wondered whether there
really was any rest there for female angels....
Of what other women desired and did,—of aspiration, spiritual and
intellectual discontent, Mazie knew nothing. For her nothing
desirable existed beyond the barbed wire. And yet, without at all
understanding Eris, always she had felt an odd sympathy for the
girl’s irregularities—had recognized that Fanny’s child was different
from herself, from her offspring—from other women’s children. But
the underlying motive that had sent Eris forth was quite beyond
Mazie’s ken. The resurrection of her sex came too early for her who
had not yet died.
The farm year had begun prosperously. Until July there had been no
cloud on the horizon. In imagination Odell gazed across acres and
acres of golden harvest; saw a beneficent and paternal Government
coming to the relief of all farmers; saw every silo packed, every barn
bursting; saw the steady increase of the herd balanced by profitable
sales; saw ribbons and prizes awaiting his exhibits at County and
State Fairs.
Yet, very often after supper, when standing on the porch chewing his
quid as stolidly as his cows chewed their cuds, he was aware of a
vague unease—as in Fanny’s day.
He could not comprehend the transmission of resentment from
Fanny to Fanny’s child. He could much less understand the inherited
resentment of a sex, now for the first time since creation making its
defiance subtly felt the whole world through. Sub jugum ad astra!
And now the Yoke had fallen; stars blazed beyond. Restless-winged,
a Sex stood poised for flight, turning deaf ears to earthbound voices
calling them back to hoods and bells and jesses.

One stifling hot night in July, after two weeks’ enervating drouth,
Odell’s impotent wrath burst from the depths of bitterness long pent:
“That ding-danged slut will shame us yet if she don’t come back! I’m
done with her if she ain’t in her own bed by Monday night. You write
and tell her, Mazie. Tell her I’m through. Tell her I say so. And that’s
that!”

The “ding-danged slut” at that moment lay asleep on the grass in a

New York public park. And all around her, on the hot and trampled
grass, lay half-naked, beastly, breathing human heaps—the heat-
tortured hordes of the unwashed.
 CHAPTER VIII
JULY began badly in New York. Ambulances became busy, hospitals
overcrowded, seaside resorts thronged. Day after day a heavy
atmosphere hung like a saturated and steaming blanket over the
city. The daily papers recorded deaths from heat. Fountains were full
of naked urchins unmolested by police. Firemen drenched the little
children of the poor with heavy showers from hose and stand-pipe.
Toward midnight, on the tenth day of the heat, a slight freshness
tempered the infernal atmosphere of the streets. It was almost a
breeze. In the Park dry leaves rustled slightly. Sleepers on bench and
withered sward stirred, sighed, relaxed again into semi-stupor.
Two men in light clothes and straw hats, crossing the Park from
West to East, paused on the asphalt path to gaze upon the
thousands of prostrate figures.
“Yonder’s a sob-stuff story for you, Barry,” remarked the shorter
man.
“There’s more than one story there,” said the other.
“No, only one. I’ll tell you that story: these people had rather work
and die in their putrid tenements than work and live in the
wholesome countryside. You can’t kick these town rats out of their
rat-ridden city. They like to fester and swarm. And when any species
swarms, Barry, Nature presently decimates it.”
They moved along slowly, looking out over the dim meadows heaped
with unstirring forms.
“Perhaps,” admitted Annan, who had been addressed as Barry, “the
mass story is about what you outlined, Mike; but there are other
stories there——” He made a slight gesture toward the meadow,
“The whole gamut from farce to tragedy....”
“The only drama in that mess is rooted in stupidity.”
“That’s where all tragedy is rooted.... I could step in among those
people and in ten minutes I could bring back material for a Hugo, a
Balzac, a Maupassant, a Dumas——”
“Why don’t you? It’s your job to look for literary loot in human scrap
heaps. Here’s life’s dumping ground. You’re the chiffonier. Why not
start business?”
“I’m considering it.”
“Go to it,” laughed the other, lighting a cigarette and leaning
gracefully on his walking stick. “Yonder’s the sewer; dig out your
diamond. Uproot your lily!”
Annan said: “Do you want to bet I can’t go in there, wake up one of
those unwashed, and, in ten minutes, get the roots of a story as
good as any ever written?”
“If you weren’t in a class by yourself,” said the other, “I’d bet with
you. Any ordinary newspaper man could go in there and dig up a
dozen obvious news items. But you’ll dig up a commonplace item
and turn it into an epic. Or you’ll dig up none at all, and come back
with a corker——”
“I’ll play square——”
“I know you! The biggest story in the world, Barry, was born a punk
little news item; and it would have died an item except for the
genius who covered it. You’re one of those damned geniuses——”
“Don’t try to hedge!——”
“Don’t tell me! Nothing ever really happens except in clever brains. I
can condense Hamlet’s story into a paragraph. But I’m glad
Shakespeare didn’t. I’m glad the Apostles were——”
“You’re a crazy Irishman, Coltfoot,” remarked Annan, looking about
him at the thousands of spectral sleepers. “Shut up. I need a story
and I’m going to get one.... You don’t want to take my bet, do you?”
“All right. Ten dollars that you don’t get the honest makings of a real
story in ten minutes. No faking! No creative genius stuff. Just bald
facts.” He looked at his wrist-watch, then at his companion. “Ready?”
Annan nodded, glanced out over the waste of withered grass. As he
stepped from the asphalt to the meadow a tepid breeze began to
blow, cooling his perspiring cheeks.
A few sleepers stirred feverishly. Under a wilted shrub a girl lifted her
heavy head from the satchel that had pillowed it. Then, slowly, she
sat upright to face the faint stir of air.
Her hat fell off. She passed slim fingers through her bobbed hair,
ruffling it to the cool wind blowing.
Annan walked directly toward her, picking his way across the grass
among the sleeping heaps of people.
As he stopped beside her, Eris looked up at him out of tired eyes
which seemed like wells of shadow, giving her pinched face an
appearance almost skull-like.
Annan mistook her age, as did everybody; and he calmly squatted
down on his haunches as though condescending to a child.
“Don’t be afraid to talk to me,” he said in his easy, persuasive way. “I
write stories for newspapers. I’m looking for a story now. If you’ll tell
me your story I’ll give you ten dollars.”
Eris stared at him without comprehension. The increasing breeze
blew her mop of chestnut curls upward from a brow as white as
milk.
“Come,” he said in his pleasant voice, “there are ten perfectly good
dollars in it for you. All I want of you is your story—not your real
name, of course,—just a few plain facts explaining how you happen
to be sleeping here in Central Park with your little satchel for your
pillow and the sky for your bed-clothes.”
Eris remained motionless, one slender hand buried in the grass, the
other resting against her temples. The blessed breeze began to
winnow her hair again.
“Won’t you talk to me?” urged Annan. “You’re not afraid, are you?”
“I don’t know what to say to you?”
“Just tell me how you happen to be sleeping here in the Park to-
night.”
“I have to save my money—” She yawned and concealed her lips
with one hand.
“Please excuse me,” she murmured, “I haven’t slept very well.”
“Then you have some money?” he inquired.
“I have twenty dollars.... Money doesn’t last long in New York.”
“No, it doesn’t,” agreed Annan gravely. “Did you work in a shop?”
“In pictures.”
“Moving pictures?”
“Yes. I have a contract with the Crystal Films.”
“Oh, yes. I heard about that outfit. It blew up. Did they ever pay you
any salary?”
“No.”
“How did you happen to hook up with that bunch of crooks?” he
asked.
“I don’t think they are crooks. Mr. Quiss isn’t.”
“Who’s he?”
“Well—I think he looks up places to photograph—and he supplies
extras——”
“A scout. Where did you run into him?”
“Near my home.”
“Did your parents permit you to join that flossy outfit?”
“No.”
“I see. You ran away.”
“I—went away.”
“Could you go home now if you wished to?”
“I don’t wish to.”
“Then you must believe that you really possess dramatic talent.”
Eris passed her fingers wearily through her hair: “I am trying to
learn something,” she said, as though to herself. “I think I have
talents.”
“What is it you most desire to be?”
“I like to act ... and dance.... I’d like to write a play ... or a book ...
or something....”
“Like other people, you’re after fame and fortune. I’m chasing them,
too. Everybody is. But the world’s goal remains the same, no matter
what you are hunting. That goal is Happiness.”
She looked at him, heavy-eyed, silent. She yawned slightly,
murmured an excuse, rubbed her eyes with her forefinger.
“Which is your principal object in life, fame or fortune?” he inquired,
smiling.
“Are those the principal objects in life?” she asked, so naïvely that he
suspected her.
“Some believe that love is more important,” he said. “Do you?”
She rested her pale cheek on her hand: “No,” she said.
“Then what is your principal object in life?” he asked, watching her
intently.
“I think, more than anything, I desire education.”
His surprise was followed by further suspicion. Her reply sounded
too naïve, too moral. He became wary of the latent actress in her.
She sat there huddled up, brooding, gazing into the darkness out of
haunted eyes.
“Do you think an education is really worth this sort of hardship?” he
asked.
That seemed to interest her. She replied:
“I think so.... I don’t know.”
“What are you trying to learn?”
“The truth ... about things.”
“Why don’t you go to school?”
“I’ve been through high-school.”
“Didn’t you learn the truth about things in high-school?”
“I don’t think so.”
“Where are you going to learn it then?”
She was plainly interested now:
“I think the only way is to find out for myself.... I don’t know
anybody who can tell me reasons. I like to be told why. If I don’t
know the facts about life how can I write plays and act them? I must
find out. I’m twenty, and I know scarcely anything worth knowing. It
is awful. It frightens me. I’m crazy to be somebody. I can’t be unless
I learn the truth about things.
“There is nobody at home to tell me.... I couldn’t stand it any
longer.... I had to find out for myself. Books don’t help. They excite.”
She looked at him feverishly: “It is a terrible thing to want only
facts,” she said. “Because nothing else satisfies.”
He thought, incredulously, “Where did she get that line?” He said: “A
taste for Truth spoils one’s appetite for anything else.... So that’s
what you’re after, is it? You’re after the truth about things.”
She did not reply.
He said, always watching her: “When you know the truth what are
you going to do with it?”
“Act it. Write it.”
“Live it, too?” he inquired gravely.
She turned to look at him, not comprehending.
“Where are you going to get the money to do all this?” he asked
lightly.
“It is going to be difficult—without money,” she admitted.
Something in the situation stirred a perverse sort of humour in him.
He didn’t quite believe in her, as she revealed her complexities and
her simplicities out of her own mouth.
“The love of money is the root of all good,” he remarked.
After a silence: “I wonder,” she said thoughtfully. “One needs it to do
good ... perhaps to be good.... Nobody can tell, I suppose, what
starvation might do to them.... Money is good.”
“All things are difficult without money,” he said, pursuing his
perverse thesis. “The love of it is not the root of all evil. Money is
often salvation. Lack of it fetters effort. Want of it retards fulfilment.
Without it ambition is crippled. Aspiration remains a dream. Lacking
a penny-worth of bread, Hamlet had never been written.... I think I’ll
say as much in my next story.”
His was an easy and humorous tongue, facile and creative, too—it
being his business to juggle nimbly with ideas and amuse an
audience at so much a column.
Eris listened, unaware that he was poking fun at himself. Her
shadowy eyes were intent on his in the starlight. The white, sharp
contours of her face interested him. He was alert for any word or
tone or gesture done for dramatic effect.
“So that’s your story, then,” he said in his gay, agreeable voice. “You
are a little pilgrim of Minerva in quest of Wisdom, travelling afoot
through the world with an empty wallet and no staff to comfort you.”
“I understand what you mean,” she said. “Minerva was goddess of
Wisdom. We had mythology in high-school.”
He thought: “She’s a clever comedienne or an utter baby.” He said:
“Is that really all there is to your story?”
“I have no story.”
“No ill-treatment at home to warrant your running away?”
“Oh, no.”
“Not even an unhappy love affair?”
She shook her head slightly as though embarrassed.
“How old are you?”
“Twenty in April.”
Annan was silent. He had not supposed her to be over seventeen.
She had seemed little more than a child in the starlight when she sat
up ruffling her bobbed hair in the first tepid breeze.
She said seriously: “I am growing old. And if I have talent I have no
time to waste. That is why I went away at the first opportunity.”
“What are your talents?”
“I dance. I have acted in school plays. Once I wrote a one-act piece
for myself. They liked it.”
“Go ahead and tell me about it.”
She told him how she had written the act and how she had sung and
danced. Stimulated by the memory of her little success, she
ventured to speak of her connection with the Crystal Films. Then,
suddenly, the long-pent flood of trouble poured out of her lonely
heart.
“I drove over to Summit,” she said, “where they had been shooting
an exterior. Mr. Quiss introduced me to Mr. Donnell, the director. Mr.
Donnell said that they were just leaving for Albany on location, and
he couldn’t give me a test. So I went to Albany the next morning—I
just packed my night-clothes and walked all the way to Gayfield to
catch the six o’clock morning train. It was my first chance. I seemed
to realise that. I took fifty dollars I had saved. I have spent thirty of
it already.
“At Albany Mr. Donnell had a test made of me. It turned out well. He
offered me a contract. I telephoned to my stepmother and told her
what I had done. I explained that I needed money.... I have some
money of my own. But my father wouldn’t let me have it. I wrote
several times, but they only told me to come home. They wouldn’t
let me have any money.
“Then, when the company arrived at the New York studio, Mr.
Donnell seemed to be in trouble. We were not paid. I heard Mr.
Quiss say that the principals had received no salary for a month. He
said that Mr. Donnell had not been paid, either. The carpenters who
were building sets refused to go on until they had their wages.
Somebody cut off the electric current. Our dynamo stopped. We
stood around all day. Somebody said that the bankers who owned
the Crystal Films were in financial difficulties.
“Then, the next morning, when we reported for work at the studio,
we found it locked. I was sorry for our company. Even the principals
seemed to be in need of money. Mr. Quiss was very kind to me. He
offered to pay my fare back home. But I wouldn’t go. Mr. Donnell
offered to lend me ten dollars, but I told him I had twenty. He gave
me a nice letter to the Elite Agency. Mr. Quiss promised to keep me
in mind. But the agencies tell me that all the film companies are
letting their people go this summer. I can’t seem to find any work.
They tell me there won’t be any work until October.... I’m saving my
twenty dollars. And I’m wondering what I shall find to do to keep
busy until October.... Even if I could afford a room, I don’t need it. It
is too hot in New York to sleep indoors.... I can wash my face and
hands in the ladies’ room of any hotel. I give the maid five cents....
But I don’t know what to do for a bath. I must do something.... I
shall hire a room for a day and wash myself and my clothes.... You
see, twenty dollars doesn’t go very far in New York.... I wonder how
far I can go on it.... Do you know what would be the very cheapest
way to live on twenty dollars until October?”
After a silence Annan said: “I owe you ten for your story. That
makes thirty dollars.”
“Oh. But I can’t take money from you!”
“Why?”
“I haven’t earned it. I had no story to tell you. I’ve only talked to
you.”
Annan, sitting cross-legged on the grass, clasped his knees with both
arms. He said, coolly:
“I offered you ten dollars for your story. That was too little to offer
for such a story. It’s worth more.”
“Why, it isn’t worth anything,” she retorted. “I hadn’t any story to tell
you. I shan’t let you give me money just because I’ve talked to you.”
“Can you guess how much I shall be paid by my newspaper for
writing out this story you have told me?” he asked, smiling at her in
the starlight.
She shook her head.
“Well, I won’t bother you with details; but your commission in this
transaction will be considerable. Your commission will amount to a
hundred dollars.”
She sat so rigid and unstirring that he leaned a little toward her to
see her expression. It was flushed and hostile.
“Do you think I am joking?” he asked.
“I don’t know what you are doing.”
He said: “I’m not mean enough to make a joke of your predicament.
I’m telling you very honestly that I can construct a first-rate short
story out of the story you have just told me. I’m workman enough to
do it. That’s my job.
“Every week I write a short story for the Sunday edition of the New
York Planet. My stories have become popular. My name is becoming
rather well known. I am now paid so well for my stories that I can
afford to pay well for the idea you have given me. Your story is full
of ideas, and it’s worth about a hundred dollars to me.”
“It isn’t worth a cent,” she said. “I don’t want you to offer me
money.... Or anything....” She laid both hands against her forehead
as though her head ached, and sat huddled up, elbows resting on
her knees. Presently she yawned.
“Please excuse me,” she murmured, “I seem to be tired.”
There was a long silence. Annan turned his head to see if his friend
Coltfoot still waited. Not discovering him, he inspected his watch.
Surprised, he lit a match to make certain of the time; and discovered
that he had been talking with this girl for more than an hour and a
half.
He said to her in his pleasant, persuasive voice: “You’re not afraid of
me, are you?”
She looked up, white and tired: “I’m not afraid of anybody.”
“Well, you’re not entirely right. However, if you’re not afraid of me,
suppose I help you find a room to-night. You can afford a room
now.”
She shook her head.
“You intend to stay here?”
“Yes, to-night.”
“You’d better not stay here with a hundred and twenty dollars in
your pocket.”
“I shan’t take money from you.”
“Do you want me to lose five hundred dollars?”
“How?” she asked, bewildered by the sudden impatience in his
voice.
“If I write the story I get six hundred. I won’t write it unless you
take your commission.”
She said nothing.
“Come,” he said, almost sharply. “I’m not going to leave you here.
You need a bath, anyway. You can’t get a good rest unless you have
a bath.”
He sprang up from the grass, took her hand before she could
withdraw it, and drew her forcibly to her feet.
“Maybe you’re twenty,” he said, “but some cop is likely to take you to
the Arsenal as a lost child.”
She seemed so startled that he reassured her with a smile,—stooped
to pick up her hat and satchel, still smiling.
“Come on, little pilgrim,” he said, “it’s two o’clock in the morning,
and the Temple of Wisdom is closed. Bath and bed is your best bet.”
She pinned on her hat mechanically, smoothed her wrinkled dress.
Then she looked up at him in a dazed way.
“Ready?” he asked gently.
“Yes. What do you want me to do?”
“Let’s go,” he said lightly, and took her by the hand again.
Slowly through starry darkness he guided her between prone shapes
on the grass, and so along the asphalt, east, until the silvery lamps
of Fifth Avenue stretched away before them in endless, level
constellations.
He was beginning to wonder where to take her at such an hour. But
to the sort of mind that was Annan’s, direct method and simple
solution always appealed. He came to a swift conclusion,—came to it
the more easily because it was an amusing one.
“You’re not afraid of me, you say?” he repeated.
She shook her head. “You seem kind.... Should I be?”
“Well, not in my case,” he said, laughing.... “We’ll take that taxi—”
He hailed it, gave directions, and seated himself beside her, now
keenly amused.
“Little pilgrim,” he said, “you’re going to have a good scrub, a good
sleep in a good bed, and a jolly good breakfast when you wake up.
What do you think of that!”
“I don’t know what to think.... I have found much kindness among
strangers.”
He laughed and lighted a cigarette. The avenue was nearly deserted.
At Forty-second Street the taxi swung west to Seventh Avenue,
south, passing Twenty-third Street, west again through a maze of
crooked old-time streets. It stopped, finally, before a two-story and
basement house of red brick—one of many similar houses that lined
both sides of a dark and very silent block.
Annan got out, paid his fare, took the little satchel, and handed Eris
out.
“Is it a boarding house?” she asked.
“One lodges well here,” he replied carelessly.
They ascended the stoop; Annan used his latch-key, let her in,
switched on the light.
“Come up,” he said briefly.
On the landing at the top of the stairs he switched on another light,
opened a door, lighted a third bracket.
“Come in!”
Eris entered the bed-room. It was large. So was the bed, a four-
poster. So was the furniture.
“Here’s your bath-room,” he remarked, opening a door into a white-
tiled room. He stepped inside to be certain. There were plenty of
towels, soap still in its wrapper, a row of bottles with flowers painted
on them—evidently for masculine use—cologne, bay rum, witch
hazel, hair-tonic.
“Now,” he said, “your worries are over until to-morrow. There’s your
tub, there’s your bed, there’s a key in the door. Lock it when you
turn in. And don’t you stir until they bring your breakfast in the
morning.”
Eris nodded.
“All right. Good-night.”
She turned toward him as though still a little bewildered.
“Are you going?” she asked timidly.
“Yes. Is there anything you need?”
“No.... I would like to thank you—if you are going....”
“Little pilgrim,” he said, “I want to thank you for an interesting
evening.”
He held out his hand; Eris laid hers in it.
“You needn’t tell me your name,” he said smilingly,—“unless you
choose to.”
“Eris Odell.”
“Eris! Well, that’s rather classic, isn’t it? That’s an—unusual—
name.... Eris. Suggests Mount Ida and golden apples, doesn’t it?—Or
is it your stage name?”
Puzzled, smiling, he stood looking at her, still retaining her hand.
“No, it’s my name.”
“Well, then, my name is Barry Annan.... And I think it’s time we both
got a little sleep....” He shook her slender hand formally, released it.
“Good-night, Eris,” he said. “Lock your door and go to sleep.”
“Good-night,” she replied in a tired, unsteady voice.
Annan walked through the corridor into the front bed-room and
turned on his light.
He seemed to be much amused with the situation,—a little worried,
too.
“She’ll get in Dutch if she doesn’t look out,” he thought as he went
about his preparations for the night.... “A funny type.... Rather
convincing.... Or a consummate actress.... But she’s most amusing
anyway. Let’s see how she turns out.... She looks hungry.... What a
little fool!... Now, you couldn’t put this over on the stage or in a
story.... Your public is too wise. They don’t grow that kind of girl
these days.... That’s romantic stuff and it won’t go with the wise
guy.... You can’t pull a character like this girl on any New York
audience. And yet, there she is—in there, scrubbing herself, if I can
judge by the sound of running water.... No, she doesn’t exist.... And
yet, there she is!... Only I’m too clever to believe in her.... There is
no fool like a smart one.... That is why the Great American Ass is the
greatest ass on earth....”
 CHAPTER IX
MRS. SNIFFEN, who had looked after Annan for thirty years, found
him bathed, shaved, and dressed, and busy writing when she
brought him his breakfast tray.
“The gentleman in the other room, Mr. Barry—when is he to ’ave ’is
breakfast?”
“It’s a lady, old dear.”
Mrs. Sniffen’s pointed nose went up with a jerk. He had been
counting on that. He liked to see Mrs. Sniffen’s nose jerk upward.
“A pretty lady,” he added, “with bobbed hair. I met her accidentally
about two o’clock this morning in Central Park.”
When the effect upon Mrs. Sniffen had sufficiently diverted him, he
told her very briefly the story of Eris.
“I’m writing it now,” he added, grinning. “Sob-stuff, Xantippe. I’m
going to make a little gem of it. It’ll be a heart-yanking tragedy—
predestined woe from the beginning. That’s what they want to-day,
—weeps. So I’m going to make ’em snivel.... Moral stuff, old dear.
You’ll like it. Now, be nice to that girl in there when she wakes up
——”
He put his arm around Mrs. Sniffen’s starched and angular shoulders
as she indignantly placed his tray on the desk before him.
“Leave me be, Mr. Barry,” she said sharply.
Some of the parties given by Annan had been attended by what Mrs.
Sniffen considered “hussies.” Annan gave various sorts of parties.
Some were approved by Mrs. Sniffen, some she disapproved. Her
sentiments made a chilling difference in her demeanour, not in her
efficiency. She was a trained servant first of all. She had been in
Annan’s family for forty years.
“Be kind to her,” repeated Annan, giving Mrs. Sniffen a pat and a
hug. “She’s a good little girl.... Too good, perhaps, to survive long.
She’s the sort of girl you read about in romance forty years ago.
She’s a Drury Lane victim. They were all fools, you know. I couldn’t
leave the suffering heroine of a Victorian novel out in the Park all
night, could I, old dear?”
“It’s your ’ouse, Mr. Barry,” said Mrs. Sniffen grimly. “Don’t be trying
to get around me with your imperent, easy ways——”
“I’m not trying to. When you see her and talk to her you’ll agree
with me that she is as virtuous as she is beautiful. Of course,” he
added, “virtue without beauty is unknown in polite fiction, and is to
be severely discouraged.”
“You’re the master,” snapped Mrs. Sniffen. “I know my place. I ’ope
others will know theirs—particularly minxes——”
“Now, Xantippe, don’t freeze the child stiff. I’m very sure she isn’t a
minx——”
Mrs. Sniffen coldly laid down the law of suspects:
“I’ll know what she is when I see her.... There’s minxes and there’s
’ussies; and there’s sluts and scuts. And there’s them that walk in
silk and them that wear h’aprons. And there’s them that would
rather die where they lie than take bed and bread of a strange
young gentleman who follows ’is fancy for a lark on a ’ot night in the
Park. ’Ussies are ’ussies. And I’m not to be deceived at my time o’
life.”
Annan chipped an egg, undisturbed. “I know you, Xantippe,” he
remarked. “You may not like some of the people who come here, but
you’ll be nice to this girl.... Take her breakfast to her at ten-thirty;
look her over; come in and report to me.”
“Very well, sir.”
Annan went on with his breakfast, leisurely. As he ate he read over
his pencilled manuscript and corrected it between bites of muffin
and bacon.
It was laid out on the lines of those modern short stories which had
proven so popular and which had lifted Barry Annan out of the
uniform ranks of the unidentified and given him an individual and
approving audience for whatever he chose to offer them.
Already there had been lively competition among periodical
publishers for the work of this new-comer.
His first volume of short stories was now in preparation. Repetition
had stencilled his name and his photograph upon the public
cerebrum. Success had not yet enraged the less successful in the
literary puddle. The frogs chanted politely in praise of their own
comrade.
The maiden, too, who sips the literary soup that seeps through the
pages of periodical publications, was already requesting his
autograph. Clipping agencies began to pursue him; film companies
wasted his time with glittering offers that never materialised. Annan
was on the way to premature fame and fortune. And to the
aftermath that follows for all who win too easily and too soon.
There is a King Stork for all puddles. His law is the law of
compensations. Dame Nature executes it—alike on species that
swarm and on individuals that ripen too quickly.
Annan wrote very fast. There were about thirty-five hundred words
in the story of Eris. He finished it by half-past ten.
Rereading it, he realised it had all the concentrated brilliancy of an
epigram. Whether or not it would hold water did not bother him. The
story of Eris was Barry Annan at his easiest and most persuasive.
There was the characteristic and ungodly skill in it, the subtle
partnership with a mindless public that seduces to mental
speculation; the reassuring caress as reward for intellectual
penetration; that inborn cleverness that makes the reader see,
applaud, or pity him or herself in the sympathetic rôle of a plaything
of Chance and Fate.
And always Barry Annan left the victim of his tact and technique
agreeably trapped, suffering gratefully, excited by self-approval to
the verge of sentimental tears.
“That’ll make ’em ruffle their plumage and gulp down a sob or two,”
he reflected, his tongue in his cheek, a little intoxicated, as usual, by
his own infernal facility.
He lit a cigarette, shuffled his manuscript, numbered the pages, and
stuffed them into his pocket. The damned thing was done.
Walking to the window he looked out into Governor’s Place—one of
those ancient and forgotten Greenwich streets, and now very still
and deserted in the intense July sunshine.
Already the hazy morning threatened to be hotter than its humid
predecessors. Nothing stirred in the street, not a cat, not an iceman,
not even a sparrow.
Tall old trees, catalpa, maple, ailanthus,—remnants of those old-time
double ranks that once lined both sidewalks,—spread solitary pools
of shade over flagstone and asphalt. All else lay naked in the glare.
Mrs. Sniffen appeared, starched to the throat, crisp, unperspiring in
her calico.
“She’s ’ad her breakfast, sir.”
“Oh! How is she feeling?”
“Could you lend her a bath-robe and slippers, sir?”
He smiled: “Has she concluded to stay here indefinitely?”
“Her clothes are in the tub, Mr. Barry.”
“In the bath-tub?”
“In the laundry tub.”
“Oh. So you’re going to do her laundry for her!”
“It’s no trouble, sir. I can ’ave them for her by early afternoon.”
“You’re a duck, Xantippe. You look after her. I’m going down-town to
the office. Give her some lunch.”
“Very good, sir.”
He followed Mrs. Sniffen to the corridor, where his straw hat and
malacca stick hung on a peg.
“Am I right, or is she a hussie?” he inquired, mischievously.
“She’s an idjit,” snapped Mrs. Sniffen. “Spanking is what she needs.”
“You give her one,” he suggested in guarded tones, glancing
instinctively at the closed door beyond.
“Shall you be back to lunch, sir?”
He was descending the stairs, his story bulging in his coat pocket.
“No; but don’t let her go till I come back. I’m going to try to
persuade her to go home to the pigs and cows.... And, Xantippe,
there’ll be four to dinner. Eight o’clock will be all right.... I’d like a
few flowers.”
“Very well, sir.”
Annan went out. The house had cooled during the night and the
heat in the street struck him in the face.
“Hell,” he muttered, “isn’t there any end to this!”
There is no shabbier, dingier city in the world than New York in
midsummer.
The metropolis seems to be inhabited by a race constitutionally
untidy, indifferent to dirt, ignorant of beauty, of the elements of civic
pride and duty.
For health and comfort alone, tree-shaded streets are a necessity;
but in New York there is a strange hostility to trees. The few that
survive mutilation by vandals,—animal and human,—are species that
ought not to be planted in such a city.
A few miserable elms, distorted poplars, crippled maples, accentuate
barren vistas. Lamp posts and fire boxes fill up the iron void, stark
as the blasted woods of no-man’s land.
Annan found Coltfoot, the Sunday editor, in his undershirt, drops of
sweat spangling the copy he was pencilling.
“You didn’t wait last night,” began Annan.
“What do you think I am!” growled Coltfoot “I need sleep if you
don’t.” He picked up a cold cigar, relighted it.
“Do I get your ten or do you get mine?”
“There’s her story,” said Annan, tossing the manuscript onto the
desk.
“Is it straight?”
“No, of course not. You yourself said that nothing really ever
happens except in the human brain.”
“Then you hand me ten?”
“I found a news item and made a story of it. As the girl is still alive, I
had to end my story by deduction.”
“What do you do, kill her off?”
“I do.”
“You and your morgue,” grunted Coltfoot. “—it’s a wonder your
public stands for all the stiffs you bring in.... But they do.... They
want more, too. It’s a murderous era. Fashion and taste have
become necrological. But mortuary pleasures pass. Happy endings
and bridal bells will come again. Then you tailors of Grubb Street will
have to cut your shrouds according.”
He glanced at the first pencilled page, skimmed it, read the next
sheet more slowly, lingered over the third—suddenly slapped the
manuscript with open palm:
“All right. All right! You get away with murder, as usual.... Your stuff
is dope. Anybody is an ass to try it. It’s habit-forming stuff. I don’t
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!

ebookname.com