Download the full version of the textbook now at textbookfull.

com

Multivariate Kernel Smoothing and Its

Applications 1st Edition José E. Chacón

https://textbookfull.com/product/multivariate-
kernel-smoothing-and-its-applications-1st-edition-
jose-e-chacon/

Explore and download more textbook at https://textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Linux Kernel Programming - Second Edition: A comprehensive

and practical guide to kernel internals, writing modules,
and kernel synchronization Kaiwan N. Billimoria
https://textbookfull.com/product/linux-kernel-programming-second-
edition-a-comprehensive-and-practical-guide-to-kernel-internals-
writing-modules-and-kernel-synchronization-kaiwan-n-billimoria/
textbookfull.com

Linux Kernel source code Heavily Commented Kernel Version

0 12 2019th Edition Zhao Jiong

https://textbookfull.com/product/linux-kernel-source-code-heavily-
commented-kernel-version-0-12-2019th-edition-zhao-jiong/

textbookfull.com

Machine Learning and its Applications 1st Edition Peter

Wlodarczak

https://textbookfull.com/product/machine-learning-and-its-
applications-1st-edition-peter-wlodarczak/

textbookfull.com

The Complete Redux Book Boris Dinkevich

https://textbookfull.com/product/the-complete-redux-book-boris-
dinkevich/

textbookfull.com
Practical Building Energy Modeling with OpenStudio: A
Guide for Students and Professionals Team Educohack

https://textbookfull.com/product/practical-building-energy-modeling-
with-openstudio-a-guide-for-students-and-professionals-team-educohack/

textbookfull.com

You Majored in What Designing Your Path from College to

Career 1st Edition Katharine Brooks

https://textbookfull.com/product/you-majored-in-what-designing-your-
path-from-college-to-career-1st-edition-katharine-brooks/

textbookfull.com

The Crisis of the 14th Century Teleconnections between

Environmental and Societal Change Bauch

https://textbookfull.com/product/the-crisis-of-the-14th-century-
teleconnections-between-environmental-and-societal-change-bauch/

textbookfull.com

SharePoint 2016 User s Guide Learning Microsoft s Business

Collaboration Platform 5th Edition Tony Smith (Auth.)

https://textbookfull.com/product/sharepoint-2016-user-s-guide-
learning-microsoft-s-business-collaboration-platform-5th-edition-tony-
smith-auth/
textbookfull.com

The Apollo Lunar Samples Collection Analysis and Results

1st Edition Anthony Young (Auth.)

https://textbookfull.com/product/the-apollo-lunar-samples-collection-
analysis-and-results-1st-edition-anthony-young-auth/

textbookfull.com
Think Like A Data Scientist Tackle The Data Science
Process Step-by-step 1st Edition Brian Godsey

https://textbookfull.com/product/think-like-a-data-scientist-tackle-
the-data-science-process-step-by-step-1st-edition-brian-godsey-2/

textbookfull.com
Multivariate Kernel
Smoothing and Its
Applications
 MONOGRAPHS ON STATISTICS AND
APPLIED PROBABILITY
Editors: F. Bunea, P. Fryzlewicz, R. Henderson, N. Keiding, T. Louis, R. Smith,
and W. Wong

Semialgebraic Statistics and Latent Tree Models

Piotr Zwiernik 146
Inferential Models
Reasoning with Uncertainty
Ryan Martin and Chuanhai Liu 147
Perfect Simulation
Mark L. Huber 148
State-Space Methods for Time Series Analysis
Theory, Applications and Software
Jose Casals, Alfredo Garcia-Hiernaux, Miguel Jerez, Sonia Sotoca, and
A. Alexandre Trindade 149
Hidden Markov Models for Time Series
An Introduction Using R, Second Edition
Walter Zucchini, Iain L. MacDonald, and Roland Langrock 150
Joint Modeling of Longitudinal and Time-to-Event Data
Robert M. Elashoff, Gang Li, and Ning Li 151
Multi-State Survival Models for Interval-Censored Data
Ardo van den Hout 152
Generalized Linear Models with Random Effects
Unified Analysis via H-likelihood, Second Edition
Youngjo Lee, John A. Nelder, and Yudi Pawitan 153
Absolute Risk
Methods and Applications in Clinical Management and Public Health
Ruth M. Pfeiffer and Mitchell H. Gail 154
Asymptotic Analysis of Mixed Effects Models
Theory, Applications, and Open Problems
Jiming Jiang 155
Missing and Modified Data in Nonparametric Estimation With R
Examples
Sam Efromovich 156
Probabilistic Foundations of Statistical Network Analysis
Harry Crane 157
Multistate Models for the Analysis of Life History Data
Richard J. Cook and Jerald F. Lawless 158
Nonparametric Models for Longitudinal Data
with Implementation in R
Colin O. Wu and Xin Tian 159
Multivariate Kernel Smoothing and Its Applications
José E. Chacón and Tarn Duong 160

For more information about this series please visit:

https://www.crcpress.com/Chapman--HallCRC-Monographs-on-Statistics--
Applied-Probability/book-series/CHMONSTAAPP
Multivariate Kernel
Smoothing and Its
Applications

José E. Chacón
Tarn Duong
The cover figure is the heat map corresponding to the kernel density estimator of the distribution of
lighting locations in the Paris city council.

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

© 2018 by Taylor & Francis Group, LLC

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

No claim to original U.S. Government works

Printed on acid-free paper

Version Date: 20180311

International Standard Book Number-13: 978-1-4987-6301-1 (Hardback)

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

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

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

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Para Anabel, Mario y Silvia, en los que empleé toda la
suerte que me tocaba en esta vida

À mon bien-aimé éternel, Christophe

 Contents

Preface xiii

List of Figures xv

List of Tables xix

List of Algorithms xxi

1 Introduction 1
1.1 Exploratory data analysis with density estimation 1
1.2 Exploratory data analysis with density derivatives estimation 4
1.3 Clustering/unsupervised learning 5
1.4 Classification/supervised learning 6
1.5 Suggestions on how to read this monograph 7

2 Density estimation 11
2.1 Histogram density estimation 11
2.2 Kernel density estimation 14
2.2.1 Probability contours as multivariate quantiles 16
2.2.2 Contour colour scales 19
2.3 Gains from unconstrained bandwidth matrices 19
2.4 Advice for practical bandwidth selection 23
2.5 Squared error analysis 26
2.6 Asymptotic squared error formulas 30
2.7 Optimal bandwidths 35
2.8 Convergence of density estimators 36
2.9 Further mathematical analysis of density estimators 37
2.9.1 Asymptotic expansion of the mean integrated squared
error 37
2.9.2 Asymptotically optimal bandwidth 39
2.9.3 Vector versus vector half parametrisations 40

ix
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
x CONTENTS
3 Bandwidth selectors for density estimation 43
3.1 Normal scale bandwidths 44
3.2 Maximal smoothing bandwidths 45
3.3 Normal mixture bandwidths 46
3.4 Unbiased cross validation bandwidths 46
3.5 Biased cross validation bandwidths 49
3.6 Plug-in bandwidths 49
3.7 Smoothed cross validation bandwidths 52
3.8 Empirical comparison of bandwidth selectors 54
3.9 Theoretical comparison of bandwidth selectors 60
3.10 Further mathematical analysis of bandwidth selectors 61
3.10.1 Relative convergence rates of bandwidth selectors 61
3.10.2 Optimal pilot bandwidth selectors 64
3.10.3 Convergence rates with data-based bandwidths 65

4 Modified density estimation 67

4.1 Variable bandwidth density estimators 67
4.1.1 Balloon density estimators 68
4.1.2 Sample point density estimators 69
4.1.3 Bandwidth selectors for variable kernel estimation 70
4.2 Transformation density estimators 73
4.3 Boundary kernel density estimators 76
4.3.1 Beta boundary kernels 76
4.3.2 Linear boundary kernels 77
4.4 Kernel choice 81
4.5 Higher order kernels 83
4.6 Further mathematical analysis of modified density estimators 84
4.6.1 Asymptotic error for sample point variable band-
width estimators 84
4.6.2 Asymptotic error for linear boundary estimators 86

5 Density derivative estimation 89

5.1 Kernel density derivative estimators 89
5.1.1 Density gradient estimators 90
5.1.2 Density Hessian estimators 92
5.1.3 General density derivative estimators 93
5.2 Gains from unconstrained bandwidth matrices 96
5.3 Advice for practical bandwidth selection 100
5.4 Empirical comparison of bandwidths of different derivative
orders 102
CONTENTS xi
5.5 Squared error analysis 103
5.6 Bandwidth selection for density derivative estimators 108
5.6.1 Normal scale bandwidths 109
5.6.2 Normal mixture bandwidths 110
5.6.3 Unbiased cross validation bandwidths 111
5.6.4 Plug-in bandwidths 112
5.6.5 Smoothed cross validation bandwidths 115
5.7 Relative convergence rates of bandwidth selectors 117
5.8 Case study: The normal density 118
5.8.1 Exact MISE 118
5.8.2 Curvature matrix 119
5.8.3 Asymptotic MISE 120
5.8.4 Normal scale bandwidth 121
5.8.5 Asymptotic MSE for curvature estimation 122
5.9 Further mathematical analysis of density derivative estima-
tors 124
5.9.1 Taylor expansions for vector-valued functions 124
5.9.2 Relationship between multivariate normal moments 124

6 Applications related to density and density derivative estima-

tion 127
6.1 Level set estimation 127
6.1.1 Modal region and bump estimation 129
6.1.2 Density support estimation 132
6.2 Density-based clustering 135
6.2.1 Stable/unstable manifolds 136
6.2.2 Mean shift clustering 137
6.2.3 Choice of the normalising matrix in the mean shift 143
6.3 Density ridge estimation 143
6.4 Feature significance 149

7 Supplementary topics in data analysis 155

7.1 Density difference estimation and significance testing 155
7.2 Classification 159
7.3 Density estimation for data measured with error 163
7.3.1 Classical density deconvolution estimation 164
7.3.2 Weighted density deconvolution estimation 166
7.3.3 Manifold estimation 170
7.4 Nearest neighbour estimation 171
7.5 Further mathematical analysis 178
xii CONTENTS
7.5.1 Squared error analysis for deconvolution kernel
density estimators 178
7.5.2 Optimal selection of the number of nearest neigh-
bours 179

8 Computational algorithms 181

8.1 R implementation 181
8.2 Approximate binned estimation 185
8.2.1 Approximate density estimation 185
8.2.2 Approximate density derivative and functional
estimation 190
8.3 Recursive computation of the normal density derivatives 191
8.4 Recursive computation of the normal functionals 195
8.5 Numerical optimisation over matrix spaces 197

A Notation 199

B Matrix algebra 205

B.1 The Kronecker product 205
B.2 The vec operator 206
B.3 The commutation matrix 206

Bibliography 207

Index 225
 Preface

Kernel smoothers, for their ability to render masses of data into useful and
interpretable summaries, form an integral part of the modern computational
data analysis toolkit. Their success is due in a large part to their intuitive
appeal which makes them accessible to non-specialists in mathematics.
Our goal in writing this monograph is to provide an overview of ker-
nel smoothers for multivariate data analysis. In keeping with their intuitive
appeal, for each different data analytic situation, we present how a kernel
smoother provides a solution to it, illustrating it with statistical graphics cre-
ated from the experimental data, and supporting it with minimal technical
mathematics. These technical details are then gradually filled in to guide the
reader towards a more comprehensive understanding of the underlying sta-
tistical, mathematical and computational concepts of kernel smoothing. It is
our hope that the book can be read at different levels: for data scientists who
wish to apply kernel smoothing techniques to their data analysis problems,
for undergraduate students who aim to understand the basic statistical prop-
erties of these methods, and for post-graduate students/specialist researchers
who require details of their technical mathematical properties. An expanded
explanation of how to read this monograph is included in Section 1.5.
The most well-known kernel smoothers are kernel density estimators,
which convert multivariate point clouds into a smooth graphical visualisa-
tion, and so can be considered as an important improvement over data his-
tograms. Interest in them is not solely restricted to their visual appeal, as their
mathematical form leads to the quantification of key statistical characteristics
which greatly assist in drawing meaningful conclusions from the observed
data. From this base case of density estimation, kernel smoothers can be ex-
tended to a wide range of more complex statistical data analytic situations, in-
cluding the identification of data-rich regions, clustering (unsupervised learn-
ing), classification (supervised learning) and hypothesis testing.
Many of these complex data analysis problems are closely related to
derivatives of the density function, especially the first (gradient) and sec-
ond (Hessian) derivatives. An important contribution of this monograph is
to provide a systematic treatment of multivariate density derivative estima-

xiii
xiv PREFACE
tion. Rather than being an arcane theoretical problem with limited practical
applications, as it has been usually historically viewed, density derivative es-
timation is now recognised as a key component in the toolkit for exploratory
data analysis and statistical inference.
We can cover only a small selection of the possible data analysis situations
in this monograph, and we focus on those with a sufficient level of maturity to
be able to offer meaningful analyses of experimental data which are informed
by solid mathematical justifications. This has lead to the exclusion of im-
portant and interesting problems such as density estimation in non-Euclidean
spaces (e.g., simplices, spheres), mode estimation, receiver operating charac-
teristic (ROC) curve analysis, conditional estimation, censored data analysis,
or regression. Rather than being a pessimistic observation, this augurs well
for the continuing potential of kernel smoothing methods to contribute to the
solution of a wide range of complex multivariate data analysis problems in
the future.
In our voyage through the field of kernel smoothing techniques, we have
been fortunate to benefit from the experience of all of our research collabo-
rators, who are too numerous to mention individually here. Though it would
be remiss of us to omit that we have been greatly influenced by our mentors:
Antonio Cuevas, Agustı́n Garcı́a Nogales, Martin Hazelton, Berwin Turlach
and Matt Wand. A special thank you also goes to Larry Wasserman for his
encouragement at the initial stages of this project, during the stay of J.E.C.
at the Carnegie Mellon University in 2015: probably this book would not
have been written without such initial push. Moreover, regarding the writing
of this monograph itself, we are grateful to all those who have reviewed the
manuscript drafts and whose feedback have much contributed to improving
them. We thank all the host and funding organisations as well, which are also
too numerous to list individually, for supporting our research work over the
past two decades.
And last, but of course not least, J.E.C. would like to thank his family for
their loving and support, and T.D. would like to thank his family and most of
all, Christophe.

José E. Chacón Tarn Duong

Badajoz, Spain Paris, France
 List of Figures

1.1 Scatter plot of the daily temperature data. 2

1.2 Scatter plot matrix of the World Bank development indicators
data. 3
1.3 Scatter plot of the earthquake data. 4
1.4 Scatter plot of the stem cell transplant data. 6
1.5 Scatter plots of the foetal cardiotocographic data. 7

2.1 Anchor point placement for the histograms of the daily

temperature data. 12
2.2 Binwidth on the histograms of the daily temperature data. 13
2.3 Construction of a density estimate on a subset of the daily
temperature data. 16
2.4 Plots of the density estimate of the daily temperature data. 17
2.5 Choice of contour levels for the density estimate of the daily
temperature data. 17
2.6 Probability contours for the density estimate of the daily
temperature data. 18
2.7 Filled contour plots of the density estimate of the daily
temperature data, with alternative colour scales. 20
2.8 Potential gains for an unconstrained over a diagonal band-
width matrix for the Grevillea data. 22
2.9 Potential gains of an unconstrained over a diagonal band-
width matrix for the dumbbell density. 23
2.10 Under- and oversmoothing of the density estimate of the
Grevillea data. 24
2.11 Different bandwidth selectors for the density estimates of the
Grevillea data. 25
2.12 Different bandwidth selectors for the density estimates of the
stem cell data. 27

xv
xvi LIST OF FIGURES
3.1 Contour plot of a moderately difficult-to-estimate normal
mixture density. 55
3.2 Different bandwidth selectors for the density estimates for
the normal mixture data. 57
3.3 Visual performance of the different bandwidth selectors for
the density estimates for the normal mixture data. 59

4.1 Variable bandwidth density estimates for the World Bank

GDP-inflation data. 72
4.2 Transformation density estimate for the World Bank CO2 -
GDP data. 74
4.3 Boundary kernel density estimates for the World Bank
internet-agriculture data. 80

5.1 Density gradient estimates for the daily temperature data. 92

5.2 Density curvature estimates for the daily temperature data. 94
5.3 Potential gains of an unconstrained over a diagonal band-
width matrix for the density derivatives of the Grevillea
data. 97
5.4 Potential gains of an unconstrained over a diagonal band-
width matrix for the derivatives of the dumbbell density. 98
5.5 Asymptotic relative errors for the derivatives of single normal
density. 99
5.6 Different bandwidth selectors for the density gradient esti-
mates of the Grevillea data. 100
5.7 Different bandwidth selectors for the density curvature
estimates of the Grevillea data. 101
5.8 Different bandwidth selectors for the density curvature
estimates of the stem cell data. 102
5.9 Plug-in selectors of different derivative orders for the fountain
density #10 data. 104

6.1 Modal regions of the kernel estimates for the daily tempera-
ture data. 131
6.2 Modal regions of the kernel estimates for the stem cell data. 131
6.3 Density support estimates for the daily temperature data. 133
6.4 Density support estimates for the Grevillea data. 134
6.5 Stable manifolds and population cluster partition for the
trimodal normal mixture density. 137
LIST OF FIGURES xvii
6.6 Mean shift recurrence for the trimodal normal mixture
density. 139
6.7 Mean shift clusters for the daily temperature data. 140
6.8 Mean shift clusters for the stem cell data. 141
6.9 First principal components of the earthquake data. 144
6.10 Density ridge estimate for the earthquake data. 147
6.11 Significant modal region estimates for the daily temperature
data. 151
6.12 Significant modal region estimates for the stem cell data. 152

7.1 Significant density difference regions for the stem cell data. 157
7.2 Target negative and positive significant density difference
regions for the stem cell data. 158
7.3 Classification for the foetal cardiotocographic test data. 161
7.4 Deconvolution density estimate for the air quality data. 168
7.5 Segmentations of the elephants image. 175

8.1 Linear binning counts. 187

8.2 Different grid sizes for the kernel density estimates of the
temperature data. 190
 List of Tables

1.1 Suggestions on how to read this monograph. 9

3.1 Convergence rates for the bandwidth selectors for density

estimation. 61

4.1 Efficiencies for spherically symmetric beta family kernels. 82

5.1 Convergence rates for the bandwidth selectors for density

derivative estimation. 118

7.1 Cross classification table for the foetal cardiotocographic test

data. 162
7.2 Efficiency of nearest neighbour and kernel density estimation. 178

8.1 R functions in the ks package for bandwidth selection. 182

xix
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
 List of Algorithms

1 Normal mixture bandwidth selector for density estimation 47

2 UCV bandwidth selector for density estimation 48
3 BCV bandwidth selector for density estimation 49
4 Two-stage plug-in bandwidth selector for density estimation 52
5 Two-stage SCV bandwidth selector for density estimation 55

6 Normal mixture bandwidth selector for density derivative

estimation 111
7 UCV bandwidth selector for density derivative estimation 112
8 Two-stage plug-in bandwidth selector for density derivative
estimation 114
9 Two-stage SCV bandwidth selector for density derivative
estimation 117

10 Mean shift clustering 142

11 Density ridge estimator 148
12 Significant modal region estimator 152

13 Significant density difference region estimator 158

14 Classification 161
15 Weighted deconvolution density estimator 170
16 Nearest neighbour mean shift clustering for image segmenta-
tion 176

17 Binned density estimator 189

18 Recursive computation of the normal density derivative 192
19 Recursive computation of the minimal Hermite polynomial 193
20 Recursive computation of Qr 197

xxi
 Chapter 1

Introduction

The late twentieth century witnessed the arrival of an omnipresence of data,

ushering in the so-called Information Age (Castells, 2010). Statistics plays a
key role in this age, as it can be broadly characterised as a meta-science which
draws meaningful conclusions from data. This role has become increasingly
critical with the advent of Open Data, which professes the unrestricted access
to data as a source material, as well as similar freedoms to produce and to dis-
seminate any suitable analyses from them (OKI, 2016). To tackle the analysis
of the widely different forms of data, there are correspondingly a wide range
of statistical techniques.

1.1 Exploratory data analysis with density estimation

Exploratory data analysis (EDA), introduced by Tukey (1977), is a fundamen-
tal class of statistical methods for understanding the structure of data sets. As
its name suggests, its relationship with more formal inferential methods is
sometimes cast with the former solely serving as a preliminary role to the ul-
timate goal of inference. This belies its importance in its own right, especially
given that increasingly accessible computing power has rendered it increas-
ingly informative.
Graphical techniques occupy a central place in EDA, as appropriate vi-
sual summaries have the ability to convey concisely the deep structure of
the data which would otherwise be difficult to discern from purely numeri-
cal summaries. The usual starting point is that we suppose that the observed
data set consists of the realisations of a continuous multi-dimensional random
variable X , whose underlying unknown probabilistic mechanism is fully de-
termined by its probability density function f . The density function provides
a wealth of information about the data and its distribution, so considerable ef-
forts, historically and currently, have been expended to develop its statistical
estimation. If each data point is represented by a multi-dimensional vector,

1
2 INTRODUCTION
then the most basic form of a density estimator is the scatter plot which visu-
alises the data as points in this multi-dimensional space.
A motivating data set is the tempb data. In Figure 1.1 is the scatter plot of
the n = 21908 pairs of the daily minimum and maximum temperatures (◦ C)
in the GHCN-D v2 time series (Menne et al., 2012) from 1 Jan 1955 to 31
Dec 2015 recorded at the weather station in Badajoz, Spain. It is located at
(38.88N, 6.83W) and has GHCN-D station code SP000004452. As this is a
dense data set in the central regions, its standard scatter plot results in solid
coloured regions which obscure the underlying data structure. To enhance the
scatter plot we use alpha blending, where the data points are represented by
translucent dots, and where overlapping shapes reinforce rather than obscure
each other as the overlapping regions lead to an increase in the displayed
opacity (Porter & Duffle, 1984). For alpha blended scatter plots, darker re-
gions thus indicate regions of higher data density.

Figure 1.1 Scatter plot of minimum and maximum daily temperatures (◦ C) of the
n = 21908 days between 1 Jan 1955 to 31 Dec 2015, recorded at the weather station
in Badajoz, Spain.

Alpha blended scatter plots can be considered as a low level form of data
smoothing of the density function, though there exist more sophisticated data
smoothing techniques. Data smoothers in general replace each infinitesimal
data point with a smooth function over its local neighbourhood in the data
space. With this smooth representation of the data sample, we are able to pro-
ceed to more suitable visualisations. We will highlight kernel smoothers in
this monograph as they form a class of methods which are highly applicable
for the practical analysis of experimental data, as well as residing within a
solid theoretical framework which assures the statistical rigour of their con-
struction and software implementation. Since the introduction of kernel es-
Random documents with unrelated
content Scribd suggests to you:
The Project Gutenberg eBook of Limehouse
Nights
 This ebook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this ebook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.

Title: Limehouse Nights

Author: Thomas Burke

Release date: August 28, 2020 [eBook #63073]

Language: English

Credits: E-text prepared by Mary Glenn Krause, Charlene Taylor,

and the Online Distributed Proofreading Team
(http://www.pgdp.net) from page images generously
made available by Internet Archive
(https://archive.org)

*** START OF THE PROJECT GUTENBERG EBOOK LIMEHOUSE

NIGHTS ***
E-text prepared by Mary Glenn Krause, Charlene Taylor,
and the Online Distributed Proofreading Team
(http://www.pgdp.net)
from page images generously made available by
Internet Archive
(https://archive.org)

Note: Images of the original pages are available through

Internet Archive. See
https://archive.org/details/limehousenights00burkiala

Limehouse Nights
 Limehouse
Nights

by Thomas Burke

NEW YORK
ROBERT M. McBRIDE & COMPANY
1919

Copyright by
Robert M. McBride & Co.

Second Printing
September, 1917
Third Printing
October, 1917
 Fourth Printing
November, 1917
Fifth Printing
March, 1918
Sixth Printing
September, 1918
Seventh Printing
March, 1919
Eighth Printing
July, 1919
Ninth Printing
September, 1919
Tenth Printing
October, 1919

Published August, 1917

To
Caradoc Evans

Limehouse Nights
 Contents
Page
The Chink and the Child 13
The Father of Yoto 39
Gracie Goodnight 57
The Paw 75
The Cue 97
Beryl, the Croucher and the Rest of England 111
The Sign of the Lamp 133
Tai Fu and Pansy Greers 149
The Bird 169
Gina of the Chinatown 187
The Knight-Errant 231
The Gorilla and the Girl 255
Ding-Dong-Dell 273
Old Joe 291
 The Chink and the Child

It is a tale of love and lovers that they tell in the low-lit Causeway
that slinks from West India Dock Road to the dark waste of waters
beyond. In Pennyfields, too, you may hear it; and I do not doubt
that it is told in far-away Tai-Ping, in Singapore, in Tokio, in
Shanghai, and those other gay-lamped haunts of wonder whither the
wandering people of Limehouse go and whence they return so
casually. It is a tale for tears, and should you hear it in the lilied
tongue of the yellow men, it would awaken in you all your pity. In
our bald speech it must, unhappily, lose its essential fragrance, that
quality that will lift an affair of squalor into the loftier spheres of
passion and imagination, beauty and sorrow. It will sound
unconvincing, a little ... you know ... the kind of thing that is best
forgotten. Perhaps....
But listen.
It is Battling Burrows, the lightning welter-weight of Shadwell, the
box o’ tricks, the Tetrarch of the ring, who enters first. Battling
Burrows, the pride of Ratcliff, Poplar and Limehouse, and the despair
of his manager and backers. For he loved wine, woman and song;
and the boxing world held that he couldn’t last long on that. There
was any amount of money in him for his parasites if only the
damned women could be cut out; but again and again would he
disappear from his training quarters on the eve of a big fight, to
consort with Molly and Dolly, and to drink other things than barley-
water and lemon-juice. Wherefore Chuck Lightfoot, his manager,
forced him to fight on any and every occasion while he was good
and a money-maker; for at any moment the collapse might come,
and Chuck would be called upon by his creditors to strip off that
“shirt” which at every contest he laid upon his man.
Battling was of a type that is too common in the eastern districts
of London; a type that upsets all accepted classifications. He
wouldn’t be classed. He was a curious mixture of athleticism and
degeneracy. He could run like a deer, leap like a greyhound, fight like
a machine, and drink like a suction-hose. He was a bully; he had the
courage of the high hero. He was an open-air sport; he had the
vices of a French decadent.
It was one of his love adventures that properly begins this tale; for
the girl had come to Battling one night with a recital of terrible
happenings, of an angered parent, of a slammed door.... In her arms
was a bundle of white rags. Now Battling, like so many sensualists,
was also a sentimentalist. He took that bundle of white rags; he paid
the girl money to get into the country; and the bundle of white rags
had existed in and about his domicile in Pekin Street, Limehouse, for
some eleven years. Her position was nondescript; to the casual
observer it would seem that she was Battling’s relief punch-ball—an
unpleasant post for any human creature to occupy, especially if you
are a little girl of twelve, and the place be the one-room household
of the lightning welter-weight. When Battling was cross with his
manager ... well, it is indefensible to strike your manager or to throw
chairs at him, if he is a good manager; but to use a dog-whip on a
small child is permissible and quite as satisfying; at least, he found it
so. On these occasions, then, when very cross with his sparring
partners, or over-flushed with victory and juice of the grape, he
would flog Lucy. But he was reputed by the boys to be a good
fellow. He only whipped the child when he was drunk; and he was
only drunk for eight months of the year.
For just over twelve years this bruised little body had crept about
Poplar and Limehouse. Always the white face was scarred with red,
or black-furrowed with tears; always in her steps and in her look was
expectation of dread things. Night after night her sleep was broken
by the cheerful Battling’s brute voice and violent hands; and terrible
were the lessons which life taught her in those few years. Yet, for all
the starved face and the transfixed air, there was a lurking beauty
about her, a something that called you in the soft curve of her cheek
that cried for kisses and was fed with blows, and in the splendid
mournfulness that grew in eyes and lips. The brown hair chimed
against the pale face, like the rounding of a verse. The blue cotton
frock and the broken shoes could not break the loveliness of her
slender figure or the shy grace of her movements as she flitted
about the squalid alleys of the docks; though in all that region of
wasted life and toil and decay, there was not one that noticed her,
until....
Now there lived in Chinatown, in one lousy room over Mr Tai Fu’s
store in Pennyfields, a wandering yellow man, named Cheng Huan.
Cheng Huan was a poet. He did not realise it. He had never been
able to understand why he was unpopular; and he died without
knowing. But a poet he was, tinged with the materialism of his race,
and in his poor listening heart strange echoes would awake of which
he himself was barely conscious. He regarded things differently from
other sailors; he felt things more passionately, and things which they
felt not at all; so he lived alone instead of at one of the lodging-
houses. Every evening he would sit at his window and watch the
street. Then, a little later, he would take a jolt of opium at the place
at the corner of Formosa Street.
He had come to London by devious ways. He had loafed on the
Bund at Shanghai. The fateful intervention of a crimp had landed
him on a boat. He got to Cardiff, and sojourned in its Chinatown;
thence to Liverpool, to Glasgow; thence, by a ticket from the
Asiatics’ Aid Society, to Limehouse, where he remained for two
reasons—because it cost him nothing to live there, and because he
was too lazy to find a boat to take him back to Shanghai.
So he would lounge and smoke cheap cigarettes, and sit at his
window, from which point he had many times observed the lyrical
Lucy. He noticed her casually. Another day, he observed her, not
casually. Later, he looked long at her; later still, he began to watch
for her and for that strangely provocative something about the toss
of the head and the hang of the little blue skirt as it coyly kissed her
knee.
Then that beauty which all Limehouse had missed smote Cheng.
Straight to his heart it went, and cried itself into his very blood.
Thereafter the spirit of poetry broke her blossoms all about his
odorous chamber. Nothing was the same. Pennyfields became a
happy-lanterned street, and the monotonous fiddle in the house
opposite was the music of his fathers. Bits of old song floated
through his mind: little sweet verses of Le Tai-pih, murmuring of
plum blossom, rice-field and stream. Day by day he would moon at
his window, or shuffle about the streets, lighting to a flame when
Lucy would pass and gravely return his quiet regard; and night after
night, too, he would dream of a pale, lily-lovely child.
And now the Fates moved swiftly various pieces on their sinister
board, and all that followed happened with a speed and precision
that showed direction from higher ways.
It was Wednesday night in Limehouse, and for once clear of mist.
Out of the coloured darkness of the Causeway stole the muffled wail
of reed instruments, and, though every window was closely
shuttered, between the joints shot jets of light and stealthy voices,
and you could hear the whisper of slippered feet, and the stuttering
steps of the satyr and the sadist. It was to the café in the middle of
the Causeway, lit by the pallid blue light that is the symbol of China
throughout the world, that Cheng Huan came, to take a dish of
noodle and some tea. Thence he moved to another house whose
stairs ran straight to the street, and above whose doorway a lamp
glowed like an evil eye. At this establishment he mostly took his pipe
of “chandu” and a brief chat with the keeper of the house, for,
although not popular, and very silent, he liked sometimes to be in
the presence of his compatriots. Like a figure of a shadowgraph he
slid through the door and up the stairs.
The chamber he entered was a bit of the Orient squatting at the
portals of the West. It was a well-kept place where one might play a
game of fan-tan, or take a shot or so of li-un, or purchase other
varieties of Oriental delight. It was sunk in a purple dusk, though
here and there a lantern stung the glooms. Low couches lay around
the walls, and strange men decorated them: Chinese, Japs, Malays,
Lascars, with one or two white girls; and sleek, noiseless attendants
swam from couch to couch. Away in the far corner sprawled a lank
figure in brown shirting, its nerveless fingers curled about the stem
of a spent pipe. On one of the lounges a scorbutic nigger sat with a
Jewess from Shadwell. Squatting on a table in the centre, beneath
one of the lanterns, was a musician with a reed, blinking upon the
company like a sly cat, and making his melody of six repeated notes.
The atmosphere churned. The dirt of years, tobacco of many
growings, opium, betel nut, and moist flesh allied themselves in one
grand assault against the nostrils.
As Cheng brooded on his insect-ridden cushion, of a sudden the
lantern above the musician was caught by the ribbon of his reed. It
danced and flung a hazy radiance on a divan in the shadow. He saw
—started—half rose. His heart galloped, and the blood pounded in
his quiet veins. Then he dropped again, crouched, and stared.
O lily-flowers and plum blossoms! O silver streams and dim-
starred skies! O wine and roses, song and laughter! For there,
kneeling on a mass of rugs, mazed and big-eyed, but understanding,
was Lucy ... his Lucy ... his little maid. Through the dusk she must
have felt his intent gaze upon her; for he crouched there, fascinated,
staring into the now obscured corner where she knelt.
But the sickness which momentarily gripped him on finding in this
place his snowy-breasted pearl passed and gave place to great joy.
She was here; he would talk with her. Little English he had, but
simple words, those with few gutturals, he had managed to pick up;
so he rose, the masterful lover, and, with feline movements, crossed
the nightmare chamber to claim his own.
If you wonder how Lucy came to be in this bagnio, the
explanation is simple. Battling was in training. He had flogged her
that day before starting work; he had then had a few brandies—not
many; some eighteen or nineteen—and had locked the door of his
room and taken the key. Lucy was, therefore, homeless, and a girl
somewhat older than Lucy, so old and so wise, as girls are in that
region, saw in her a possible source of revenue. So there they were,
and to them appeared Cheng.
From what horrors he saved her that night cannot be told, for her
ways were too audaciously childish to hold her long from harm in
such a place. What he brought to her was love and death.
For he sat by her. He looked at her—reverently yet passionately.
He touched her—wistfully yet eagerly. He locked a finger in her
wondrous hair. She did not start away; she did not tremble. She
knew well what she had to be afraid of in that place; but she was
not afraid of Cheng. She pierced the mephitic gloom and scanned his
face. No, she was not afraid. His yellow hands, his yellow face, his
smooth black hair ... well, he was the first thing that had ever
spoken soft words to her; the first thing that had ever laid a hand
upon her that was not brutal; the first thing that had deferred in
manner towards her as though she, too, had a right to live. She
knew his words were sweet, though she did not understand them.
Nor can they be set down. Half that he spoke was in village Chinese;
the rest in a mangling of English which no distorted spelling could
possibly reproduce.
But he drew her back against the cushions and asked her name,
and she told him; and he inquired her age, and she told him; and he
had then two beautiful words which came easily to his tongue. He
repeated them again and again:
“Lucia ... li’l Lucia.... Twelve.... Twelve.” Musical phrases they
were, dropping from his lips, and to the child who heard her name
pronounced so lovingly, they were the lost heights of melody. She
clung to him, and he to her. She held his strong arm in both of hers
as they crouched on the divan, and nestled her cheek against his
coat.
 Well ... he took her home to his wretched room.
“Li’l Lucia, come-a-home ... Lucia.”
His heart was on fire. As they slipped out of the noisomeness into
the night air and crossed the West India Dock Road into Pennyfields,
they passed unnoticed. It was late, for one thing, and for another ...
well, nobody cared particularly. His blood rang with soft music and
the solemnity of drums, for surely he had found now what for many
years he had sought—his world’s one flower. Wanderer he was, from
Tuan-tsen to Shanghai, Shanghai to Glasgow ... Cardiff ... Liverpool
... London. He had dreamed often of the women of his native land;
perchance one of them should be his flower. Women, indeed, there
had been. Swatow ... he had recollections of certain rose-winged
hours in coast cities. At many places to which chance had led him a
little bird had perched itself upon his heart, but so lightly and for so
brief a while as hardly to be felt. But now—now he had found her in
this alabaster Cockney child. So that he was glad and had great joy
of himself and the blue and silver night, and the harsh flares of the
Poplar Hippodrome.
You will observe that he had claimed her, but had not asked
himself whether she were of an age for love. The white perfection of
the child had captivated every sense. It may be that he forgot that
he was in London and not in Tuan-tsen. It may be that he did not
care. Of that nothing can be told. All that is known is that his love
was a pure and holy thing. Of that we may be sure, for his worst
enemies have said it.
Slowly, softly they mounted the stairs to his room, and with almost
an obeisance he entered and drew her in. A bank of cloud raced to
the east and a full moon thrust a sharp sword of light upon them.
Silence lay over all Pennyfields. With a bird-like movement, she
looked up at him—her face alight, her tiny hands upon his coat—
clinging, wondering, trusting. He took her hand and kissed it;
repeated the kiss upon her cheek and lip and little bosom, twining
his fingers in her hair. Docilely, and echoing the smile of his lemon
lips in a way that thrilled him almost to laughter, she returned his
kisses impetuously, gladly.
He clasped the nestling to him. Bruised, tearful, with the love of
life almost thrashed out of her, she had fluttered to him out of the
evil night.
“O li’l Lucia!” And he put soft hands upon her, and smoothed her
and crooned over her many gracious things in his flowered speech.
So they stood in the moonlight, while she told him the story of her
father, of her beatings, and starvings, and unhappiness.
“O li’l Lucia.... White Blossom.... Twelve.... Twelve years old!”
As he spoke, the clock above the Millwall Docks shot twelve
crashing notes across the night. When the last echo died, he moved
to a cupboard, and from it he drew strange things ... formless
masses of blue and gold, magical things of silk, and a vessel that
was surely Aladdin’s lamp, and a box of spices. He took these robes,
and, with tender, reverent fingers, removed from his White Blossom
the besmirched rags that covered her, and robed her again, and led
her then to the heap of stuff that was his bed, and bestowed her
safely.
For himself, he squatted on the floor before her, holding one
grubby little hand. There he crouched all night, under the lyric
moon, sleepless, watchful; and sweet content was his. He had fallen
into an uncomfortable posture, and his muscles ached intolerably.
But she slept, and he dared not move nor release her hand lest he
should awaken her. Weary and trustful, she slept, knowing that the
yellow man was kind and that she might sleep with no fear of a steel
hand smashing the delicate structure of her dreams.
In the morning, when she awoke, still wearing her blue and yellow
silk, she gave a cry of amazement. Cheng had been about. Many
times had he glided up and down the two flights of stairs, and now
at last his room was prepared for his princess. It was swept and
garnished, and was an apartment worthy a maid who is loved by a
poet-prince. There was a bead curtain. There were muslins of pink
and white. There were four bowls of flowers, clean, clear flowers to
gladden the White Blossom and set off her sharp beauty. And there
was a bowl of water, and a sweet lotion for the bruise on her cheek.
When she had risen, her prince ministered to her with rice and
egg and tea. Cleansed and robed and calm, she sat before him,
perched on the edge of many cushions as on a throne, with all the
grace of the child princess in the story. She was a poem. The beauty
hidden by neglect and fatigue shone out now more clearly and
vividly, and from the head sunning over with curls to the small white
feet, now bathed and sandalled, she seemed the living interpretation
of a Chinese lyric. And she was his; her sweet self and her prattle,
and her bird-like ways were all his own.
Oh, beautifully they loved. For two days he held her. Soft caresses
from his yellow hands and long, devout kisses were all their
demonstration. Each night he would tend her, as might mother to
child; and each night he watched and sometimes slumbered at the
foot of her couch.
But now there were those that ran to Battling at his training
quarters across the river, with the news that his child had gone with
a Chink—a yellow man. And Battling was angry. He discovered
parental rights. He discovered indignation. A yellow man after his
kid! He’d learn him. Battling did not like men who were not born in
the same great country as himself. Particularly he disliked yellow
men. His birth and education in Shadwell had taught him that of all
creeping things that creep upon the earth the most insidious is the
Oriental in the West. And a yellow man and a child. It was ... as you
might say ... so ... kind of ... well, wasn’t it? He bellowed that it was
“unnacherel.” The yeller man would go through it. Yeller! It was his
supreme condemnation, his final epithet for all conduct of which he
disapproved.
There was no doubt that he was extremely annoyed. He went to
the Blue Lantern, in what was once Ratcliff Highway, and thumped
the bar, and made all his world agree with him. And when they
agreed with him he got angrier still. So that when, a few hours later,
he climbed through the ropes at the Netherlands to meet Bud Tuffit
for ten rounds, it was Bud’s fight all the time, and to that bright
boy’s astonishment he was the victor on points at the end of the ten.
Battling slouched out of the ring, still more determined to let the
Chink have it where the chicken had the axe. He left the house with
two pals and a black man, and a number of really inspired curses
from his manager.
On the evening of the third day, then, Cheng slipped sleepily down
the stairs to procure more flowers and more rice. The genial Ho
Ling, who keeps the Canton store, held him in talk some little while,
and he was gone from his room perhaps half-an-hour. Then he
glided back, and climbed with happy feet the forty stairs to his
temple of wonder.
With a push of a finger he opened the door, and the blood froze
on his cheek, the flowers fell from him. The temple was empty and
desolate; White Blossom was gone. The muslin hangings were torn
down and trampled underfoot. The flowers had been flung from
their bowls about the floor, and the bowls lay in fifty fragments. The
joss was smashed. The cupboard had been opened. Rice was
scattered here and there. The little straight bed had been jumped
upon by brute feet. Everything that could be smashed or violated
had been so treated, and—horror of all—the blue and yellow silk
robe had been rent in pieces, tied in grotesque knots, and slung
derisively about the table legs.
I pray devoutly that you may never suffer what Cheng Huan
suffered in that moment. The pangs of death, with no dying; the
sickness of the soul which longs to escape and cannot; the
imprisoned animal within the breast which struggles madly for a
voice and finds none; all the agonies of all the ages—the agonies of
every abandoned lover and lost woman, past and to come—all these
things were his in that moment.
Then he found voice and gave a great cry, and men from below
came up to him; and they told him how the man who boxed had
been there with a black man; how he had torn the robes from his
child, and dragged her down the stairs by her hair; and how he had
shouted aloud for Cheng and had vowed to return and deal
separately with him.
Now a terrible dignity came to Cheng, and the soul of his great
fathers swept over him. He closed the door against them, and fell
prostrate over what had been the resting-place of White Blossom.
Those without heard strange sounds as of an animal in its last pains;
and it was even so. Cheng was dying. The sacrament of his high and
holy passion had been profaned; the last sanctuary of the Oriental—
his soul dignity—had been assaulted. The love robes had been torn
to ribbons; the veil of his temple cut down. Life was no longer
possible; and life without his little lady, his White Blossom, was no
longer desirable.
Prostrate he lay for the space of some five minutes. Then, in his
face all the pride of accepted destiny, he arose. He drew together
the little bed. With reverent hands he took the pieces of blue and
yellow silk, kissing them and fondling them and placing them about
the pillow. Silently he gathered up the flowers, and the broken
earthenware, and burnt some prayer papers and prepared himself
for death.
Now it is the custom among those of the sect of Cheng that the
dying shall present love-gifts to their enemies; and when he had set
all in order, he gathered his brown canvas coat about him, stole from
the house, and set out to find Battling Burrows, bearing under the
coat his love-gift to Battling. White Blossom he had no hope of
finding. He had heard of Burrows many times; and he judged that,
now that she was taken from him, never again would he hold those
hands or touch that laughing hair. Nor, if he did, could it change
things from what they were. Nothing that was not a dog could live in
the face of this sacrilege.
As he came before the house in Pekin Street, where Battling lived,
he murmured gracious prayers. Fortunately, it was a night of thick
river mist, and through the enveloping velvet none could observe or
challenge him. The main door was open, as are all doors in this
district. He writhed across the step, and through to the back room,
where again the door yielded to a touch.
Darkness. Darkness and silence, and a sense of frightful things.
He peered through it. Then he fumbled under his jacket—found a
match—struck it. An inch of candle stood on the mantelshelf. He lit
it. He looked round. No sign of Burrows, but.... Almost before he
looked he knew what awaited him. But the sense of finality had
kindly stunned him; he could suffer nothing more.
On the table lay a dog-whip. In the corner a belt had been flung.
Half across the greasy couch lay White Blossom. A few rags of
clothing were about her pale, slim body; her hair hung limp as her
limbs; her eyes were closed. As Cheng drew nearer and saw the
savage red rails that ran across and across the beloved body, he
could not scream—he could not think. He dropped beside the couch.
He laid gentle hands upon her, and called soft names. She was warm
to the touch. The pulse was still.
Softly, oh, so softly, he bent over the little frame that had enclosed
his friend-spirit, and his light kisses fell all about her. Then, with the
undirected movements of a sleep-walker, he bestowed the rags
decently about her, clasped her in strong arms, and crept silently
into the night.
From Pekin Street to Pennyfields it is but a turn or two, and again
he passed unobserved as he bore his tired bird back to her nest. He
laid her upon the bed, and covered the lily limbs with the blue and
yellow silks and strewed upon her a few of the trampled flowers.
Then, with more kisses and prayers, he crouched beside her.
So, in the ghastly Limehouse morning, they were found—the dead
child, and the Chink, kneeling beside her, with a sharp knife gripped
in a vice-like hand, its blade far between his ribs.
Meantime, having vented his wrath on his prodigal daughter,
Battling, still cross, had returned to the Blue Lantern, and there he
stayed with a brandy tumbler in his fist, forgetful of an appointment
at Premierland, whereby he should have been in the ring at ten
o’clock sharp. For the space of an hour Chuck Lightfoot was going
blasphemously to and fro in Poplar, seeking Battling and not finding
him, and murmuring, in tearful tones: “Battling—you dammanblasted
Battling—where are yeh?”
His opponent was in his corner sure enough, but there was no
fight. For Battling lurched from the Blue Lantern to Pekin Street. He
lurched into his happy home, and he cursed Lucy, and called for her.
And finding no matches, he lurched to where he knew the couch
should be, and flopped heavily down.
Now it is a peculiarity of the reptile tribe that its members are
impatient of being flopped on without warning. So, when Battling
flopped, eighteen inches of writhing gristle upreared itself on the
couch, and got home on him as Bud Tuffit had done the night before
—one to the ear, one to the throat, and another to the forearm.
Battling went down and out.
And he, too, was found in the morning, with Cheng Huan’s love-
gift coiled about his neck.
 The Father of Yoto

Sweet human hearts—a tale of carnival, moon-haunted nights: a

tale of the spring-tide, of the flower and the leaf ripening to fruit: a
gossamer thing of dreamy-lanterned streets, told by my friend, Tai
Ling, of West India Dock Road. Its scene is not the Hoang Ho or the
sun-loved islands of the East, but Limehouse. Nevertheless it is a
fairy tale, because so human.
Marigold Vassiloff was a glorious girl. The epithet is not mine, but
Tai Ling’s. Marigold lived under the tremendous glooms of the East
and West India Docks; and what she didn’t know about the more
universal aspects of human life, though she was yet short of twenty,
was hardly to be known. You know, perhaps, the East India Dock,
which lies a little north of its big brother, the West India Dock: a
place of savagely masculine character, evoking the brassy mood. By
day-time a cold, nauseous light hangs about it; at night a devilish
darkness settles upon it.
You know, perhaps, the fried-fish shops that punctuate every
corner in the surrounding maze of streets, the “general” shops with
their assorted rags, their broken iron, and their glum-faced basins of
kitchen waste; and the lurid-seeming creatures that glide from
nowhere into nothing—Arab, Lascar, Pacific Islander, Chinky, Hindoo,
and so on, each carrying his own perfume. You know, too, the
streets of plunging hoof and horn that cross and re-cross the
waterways, the gaunt chimneys that stick their derisive tongues to
the skies. You know the cobbly courts, the bestrewn alleys, through
which at night gas-jets asthmatically splutter; and the mephitic
glooms and silences of the dock-side. You know these things, and I
need not attempt to illuminate them for you.
 But you do not know that in this place there are creatures with the
lust for life racing in their veins; creatures hot for the moment and
its carnival; children of delicate graces; young hearts asking only
that they may be happy for their hour. You do not know that there
are girls on these raw edges of London to whom silks and wine and
song are things to be desired but never experienced. Neither do you
know that one of these creatures, my Marigold, was the heroine of
one of the most fantastic adventures of which I have heard.
It may offend your taste, and in that case you may reject it. Yet I
trust you will agree that any young thing, moving in that dank
daylight, that devilish darkness, is fully justified in taking her
moments of gaiety as and when she may. There may be callow
minds that cry No; and for them I have no answer. There are minds
to which the repulsive—such as Poplar High Street—is supremely
beautiful, and to whom anything frankly human is indelicate, if not
ugly. You need, however, to be a futurist to discover ecstatic beauty
in the torn wastes of tiles, the groupings of iron and stone, and the
nightmare of chimney-stacks and gas-works. Barking Road, as it dips
and rises with a sweep as lovely as a flying bird’s, may be a thing to
fire the trained imagination, and so may be the subtle tones of flame
and shade in the byways, and the airy tracery of the Great Eastern
Railway arches. But these crazy things touch only those who do not
live among them: who comfortably wake and sleep and eat in
Hampstead and Streatham. The beauty which neither time nor tears
can fade is hardly to be come by east of Aldgate Pump; if you look
for it there and think that you find it, I may tell you that you are a
poseur; you may take your seat at a St John’s Wood breakfast-table,
and stay there.
Marigold was not a futurist. She was an apple-cheeked girl, lovely
and brave and bright. The Pool at night never shook her to wonder.
Mast-head, smoke-stack, creaking crane, and the perfect chiming of
the overlying purples evoked nothing responsive in her. If she
desired beauty at all, it was the beauty of the chocolate box or the
biscuit tin. Wherefore Poplar and Limehouse were a weariness to
her. She was a malcontent; and one can hardly blame her, for she
was a girl of girls. When she dreamed of happier things, which she
did many times a week, and could not get them, she took the next
best thing. A sound philosophy, you will agree. She flogged a jaded
heart in the loud music hall, the saloons of the dock-side, and found
some minutes’ respite from the eternal grief of things in the arms of
any salt-browned man who caught her fancy.
Tai Ling was right. She was a moon-blossom. Impossible to
imagine what she might have been in gentler surroundings. As it
was she was too cruelly beautiful for human nature’s daily food. Her
face had not the pure and perfect beauty such as you may find in
the well-kept inmates of an Ealing High School. But above that face
was a crown of thunderous hair, shot with an elfish sheen, which
burned the heart out of any man creature who spotted her. She was
small, but ripe-breasted, and moved like a cat. The very lines of her
limbs were an ecstasy, and she had, too, an odd, wide laugh—and
knew how to use it.
Now it happened one night, when her head was tangled in a net
of dreams, that she sought escape in the Causeway, in the little
white café where you may take noodle, chop suey, China tea, and
other exotic foods. She was the only white thing there. Yellow men
and brown were there, and one tan-skinned woman, but Marigold
was the only pure product of these islands. At a far table, behind the
bead curtain in the corner, sat Tai Ling. He saw her, and lit to a
sudden delight of her.
Tai Ling was a queer bird. Not immoral, for, to be immoral, you
must first subscribe to some conventional morality. Tai Ling did not.
You cannot do wrong until you have first done right. Tai Ling had
not. He was just non-moral; and right and wrong were words he did
not understand. He was in love with life, and song, and wine, and
warmth, and the beauty of little girls. The world to him, as to
Marigold, was a pause on a journey, where one might take one’s idle
pleasure, while others strewed the path with mirth and roses. He
knew only two divisions of people—the gay and the stupid. The
problems of this life and the next passed him by. He never turned
aside from pleasure, or resisted an invitation to the feast.
In fact, by our standards, a complete rogue; yet the most joyous I
have known. I never knew a man with so seductive a smile. It has
driven the virtuously indignant heart out of me many a time, and I
never knew a girl, white or coloured, who could withstand it. I
almost believe it would have beaten down the frigid steel ramparts
that begird the English “lady.” It thrilled and tickled you as does the
gayest music of Mozart. It had not the mere lightness of frivolity,
but, like that music, it had the deep-plumbing gaiety of the love of
life, for joy and sorrow.
The moment Tai Ling caught Marigold’s eye, the heart in him
sprang like a bird to song, and he began to smile. I say began, for
an Oriental smile is not an affair of a swift moment. It has a birth
and a beginning. It awakes—hesitates—grows, and at last from the
sad chrysalis emerges the butterfly. A Chinese smile at the full is one
of the subtlest expressions of which the human face is capable.
The mischief was done. Marigold went down before that smile
without even putting up her guard. Swift on the uptake, she tossed
it back to him, and her maddening laugh ran across the room. Tai
Ling waited until she drew out a frowsy packet of cigarettes; then
back to her he carried the laugh, and slipped a lighted match over
her shoulder almost before the cigarette was at her mouth.
It was aptly done. He sat down beside her, and took graceful
charge of her hand, while he encircled her waist. He had been flying
to and fro long enough on P. & O. boats to have picked up, during
his London sojourns, a fair Cockney vocabulary, which he used with
a liquid accent; and he began talk with her, in honey-flavoured
phrases, of Swatow, of Yokohama, Fuji Yama, Sarawak; of flowered
islands, white towns and green bays, and sunlight like wine, and ...
oh, a thousand things that the little cloudy head spun at hearing.
They had more tea and cigarettes, and he bought a scented spice
for her, and they left the café together, at about midnight, very glad.
 When Marigold gave herself to Tai Ling, as I have explained in that
row of dots, she did so because she was happy, and because Tai
Ling had amused her, and was pleased with her. But why she met
him again and yet again, it is difficult to say. It is difficult also to
understand why Tai Ling, who so loved sunshine, and flower and
blue water, should have lingered in fusty Limehouse for the space of
a year. But the two of them seemed to understand their conduct,
and both were happy. For Tai Ling had a little apartment in the
Causeway, and thither Marigold would flit from time to time, until....
One evening, as they loafed together in the hot, lousy dusk, when
the silence was so sharp that a footstep seemed to shatter the night,
he learnt, in a flood of joy and curiosity and apprehension, that he
was about to become papa.
It overwhelmed him. He nearly choked. It was so astounding, so
new, so wonderful, so ... everything that was inexpressible. Such a
thing had not happened before to him. Hitherto, he had but loved
and ridden away, the gay deceiver. But now—— He questioned, and
conjectured what was to be done; and Marigold replied airily that it
didn’t matter much; that if she had a little money she could arrange
things. She spoke of a Poplar hospital ... good treatment ... quite all
right; and thereupon she collapsed at his feet in a tempest of curls
and tears.
With that, his emotions cleared and calmed, and resolved
themselves into one definite quantity—pride. He drew Marigold on to
the cushions, and kissed her, and in his luscious tongue he sang to
her; and this is, roughly, what he sang: an old song known to his
father:
“O girl, the streams and trees glory in the glamour of spring; the
bright sun drops about the green shrubs, and the falling flowers are
scattered and fly away. The lonely cloud moves to the hill, and the
birds find their leafy haunts. All things have a refuge to which they
fly, but I alone have nothing to which to cling. Wherefore, under the
moon I drink and sing to the fragrant blossom, and I hold you fast,
O flower of the waters, O moon-blossom, O perfect light of day!
“Violets shall lie shining about your neck, and roses in your hair.
Your holy hands shall be starred about with gems. Over the green
and golden hills, and through the white streets we will wander while
the dawn is violet-lidded; and I will hide you in your little nest at
night, and love shall be over you for ever!”
That was his song, sung in Chinese. It was old—such songs are
not now written in the country of Tai Ling, except by imitators—and
Tai Ling might well have forgotten it in the hard labours of his
seaman’s life. But he had not, and when it was finished, Marigold
was pleased, and clung to him, and told him that she so loved him
that she must not inflict this trouble upon him. But he would not
hear her.
“Nonono, Malligold,” he murmured, while they raptured, “Malligold
—lou shall not go. Lou shall stay with Tai Ling. Oh, lou’ll have
evelything beautiful, all same English lady. Tai Ling have heap money
—les—and lou shall have a li’l room.... Blimey—les ... clever doctors
... les.”
And he managed it. He arranged that chamber and that landlady,
and that doctor and nurse were duly booked. And he glided in great
joy next evening to the café, to inform his friends that he was about
to have an heir. He talked loudly and volubly in his rich seaman’s
lingo, and suddenly, in the same language, a voice shot through the
clamour:
“Tai Ling, you speak no truth!”
Tai Ling sprang up, and his hand flew to the waist of his cotton
trousers, and flew back, grasping a kreese.
“Tai Ling,” repeated the voice, still in Chinese, “I say you lie. I am
the father of li’l Malligold’s babe!”
At that moment, anything might have happened, had not two
shirt-sleeved waiters slipped dexterously between the claimants, and
grasped their wrists. Tai Ling’s face was aflame with as much
primitive emotion as an Oriental face may show. But his first rage
died, as another voice came from the bead curtain at the rear of the
little cluster.
“Tai Ling, Wing Foo, you both speak no truth. For Malligold has
told me even this evening that the child is mine!” And the third
claimant thrust a vehement face through the curtain, and swam
down among them. “I,” he cried, his hands quarrelling nervously at
his bosom, “I—I am the father of Malligold’s man-child!”
The glances of the three met like velveted blades. For one
moment tragedy was in the air. Knives were still being grasped.
Then Tai Ling began his conquering smile. It was caught by the
crowd and echoed, and in another moment light laughter was
running about, with chattering voices and gesturing hands. The
waiters released their hold on the prospective fathers, and the three
competitors sat down to a table and called for tea and sweet cakes
and cigarettes.
One must admit that Marigold’s conduct was, as the politicians
say, deserving of the highest censure; but, you see, she was young,
and she needed money for this business—her first. Some small
amounts, it appeared, she had managed to collect from Wing Foo
and his friend, but neither of them had done what Tai Ling had done
so magnanimously. You would have thought, perhaps, that by all the
traditions of his race, Ling would have been exceedingly wroth at
this discovery of infidelity on the part of one who had shared his
bed. But he was not. He sat at the table, and smiled that
inscrutable, shattering smile, and in fancy he folded Marigold within
his brown arms. His was an easy-going disposition; human kindliness
counted with him before tradition and national beliefs. A sweet
fellow. A rogue himself, he did not demand perfection in others. No;
the infidelity did not anger him. The only point about the business
that really disturbed him was that there should be others who
aspired to the fatherhood of this, Marigold’s first child, and, he
believed, his.
 So they sat and talked it over, and when they parted, and each
went his way into the night, to tell his tale, Tai Ling went to the
Poplar Hippodrome to drown his perplexity. There he witnessed the
performance of a Chinese juggler, who blasphemed his assistants in
the language of Kennington Gate, and was registered on the voting
list at Camberwell as Rab M’Andrew. After sitting in the hall for some
hour and a half, his ideas were adjusted, and he went to the house
where Marigold was, and gently charged her with what he had
heard. She fell at once to tears and protestations and explanations,
and desired to go away from him for ever. She had not meant
wrong; but ... she did not know ... and she had so wanted the
money ... and....
Well, he would not let her go. He caught her back, and thrust his
forgiveness upon her; and the whole affair ought to have ended in
disaster for both of them. But it did not, as you will see.
The next morning, there was a new development. The story of the
café conversation was racing about Limehouse and Poplar, when it
came to the ears of one, Chuck Lightfoot, a pugilistic promoter. Now
parenthood is not an office which the Englishman lightly assumes,
but Chuck straightway butted in, and demanded to know, with
menaces, what was the matter with his claim. It wasn’t that he was
specially anxious to father the child. Indeed, the success of his
claim, and the resultant financial outlay, would have seriously
disconcerted him. It was just the principle of the thing that riled him.
Damn it, he wasn’t going to stand by and be dished by any lousy
scarleteer of a yellow devil; not much. He asserted further that by
reference to dates he could prove many things which went far to
establish his claim; and, finally, if anyone wanted a fight, they’d only
got to ask for it.
Apparently no one did; for Tai Ling went about with that smile of
his, and shook all seriousness out of them. During the week he
called a convocation at the house where he had installed Marigold,
and where she now lay, and there they gathered—three yellow men,
proud, jealous, reticent, and one vehement white man, hot-eared,