Download Full Version ebook - Visit ebooknice.

com

(Ebook) Statistics in Engineering: With Examples

in MATLAB® and R, Second Edition (Chapman &
Hall/CRC Texts in Statistical Science) by Andrew
Metcalfe, David Green, Tony Greenfield,
Mayhayaudin Mansor, Andrew Smith, Jonathan Tuke
https://ebooknice.com/product/statistics-in-engineering-
ISBN 9781439895474, 1439895473
with-examples-in-matlab-and-r-second-edition-chapman-hall-
crc-texts-in-statistical-science-11359990

Click the button below to download

DOWLOAD EBOOK

Discover More Ebook - Explore Now at ebooknice.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Start reading on any device today!

(Ebook) Biota Grow 2C gather 2C cook by Loucas, Jason;

Viles, James ISBN 9781459699816, 9781743365571,
9781925268492, 1459699815, 1743365578, 1925268497
https://ebooknice.com/product/biota-grow-2c-gather-2c-cook-6661374

ebooknice.com

(Ebook) Matematik 5000+ Kurs 2c Lärobok by Lena

Alfredsson, Hans Heikne, Sanna Bodemyr ISBN 9789127456600,
9127456609
https://ebooknice.com/product/matematik-5000-kurs-2c-larobok-23848312

ebooknice.com

(Ebook) SAT II Success MATH 1C and 2C 2002 (Peterson's SAT

II Success) by Peterson's ISBN 9780768906677, 0768906679

https://ebooknice.com/product/sat-ii-success-
math-1c-and-2c-2002-peterson-s-sat-ii-success-1722018

ebooknice.com

(Ebook) Cambridge IGCSE and O Level History Workbook 2C -

Depth Study: the United States, 1919-41 2nd Edition by
Benjamin Harrison ISBN 9781398375147, 9781398375048,
1398375144, 1398375047
https://ebooknice.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-53538044

ebooknice.com
(Ebook) Master SAT II Math 1c and 2c 4th ed (Arco Master
the SAT Subject Test: Math Levels 1 & 2) by Arco ISBN
9780768923049, 0768923042
https://ebooknice.com/product/master-sat-ii-math-1c-and-2c-4th-ed-
arco-master-the-sat-subject-test-math-levels-1-2-2326094

ebooknice.com

(Ebook) Statistical Rethinking: A Bayesian Course with

Examples in R and STAN (Chapman & Hall/CRC Texts in
Statistical Science) by Richard Mcelreath ISBN
9780367139919, 036713991X
https://ebooknice.com/product/statistical-rethinking-a-bayesian-
course-with-examples-in-r-and-stan-chapman-hall-crc-texts-in-
statistical-science-11118670
ebooknice.com

(Ebook) Bayesian Networks: With Examples in R (Chapman &

Hall/CRC Texts in Statistical Science) by Scutari, Marco,
Denis, Jean-Baptiste ISBN 9780367366513, 0367366517
https://ebooknice.com/product/bayesian-networks-with-examples-in-r-
chapman-hall-crc-texts-in-statistical-science-34669960

ebooknice.com

(Ebook) Bayesian Data Analysis, Second Edition (Chapman &

Hall CRC Texts in Statistical Science) by Andrew Gelman,
John B. Carlin, Hal S. Stern, Donald B. Rubin ISBN
9781584883883, 158488388X
https://ebooknice.com/product/bayesian-data-analysis-second-edition-
chapman-hall-crc-texts-in-statistical-science-2450262

ebooknice.com

(Ebook) Foundations of Statistics for Data Scientists:

With R and Python (Chapman & Hall/CRC Texts in Statistical
Science) by Alan Agresti, Maria Kateri ISBN 9780367748456,
0367748452
https://ebooknice.com/product/foundations-of-statistics-for-data-
scientists-with-r-and-python-chapman-hall-crc-texts-in-statistical-
science-37624826
ebooknice.com
Statistics in Engineering
With Examples in MATLAB® and R
Second Edition
CHAPMAN & HALL/CRC
Texts in Statistical Science Series
Joseph K. Blitzstein, Harvard University, USA
Julian J. Faraway, University of Bath, UK
Martin Tanner, Northwestern University, USA
Jim Zidek, University of British Columbia, Canada

Recently Published Titles

Extending the Linear Model with R

Generalized Linear, Mixed Effects and Nonparametric Regression Models, Second Edition
J.J. Faraway

Modeling and Analysis of Stochastic Systems, Third Edition

V.G. Kulkarni

Pragmatics of Uncertainty
J.B. Kadane

Stochastic Processes
From Applications to Theory
P.D Moral and S. Penev

Modern Data Science with R

B.S. Baumer, D.T Kaplan, and N.J. Horton

Generalized Additive Models

An Introduction with R, Second Edition
S. Wood

Design of Experiments
An Introduction Based on Linear Models
Max Morris

Introduction to Statistical Methods for Financial Models

T. A. Severini

Statistical Regression and Classification

From Linear Models to Machine Learning
Norman Matloff

Introduction to Functional Data Analysis

Piotr Kokoszka and Matthew Reimherr

Stochastic Processes
An Introduction, Third Edition
P.W. Jones and P. Smith

Theory of Stochastic Objects

Probability, Stochastic Processes and Inference
Athanasios Christou Micheas

Linear Models and the Relevant Distributions and Matrix Algebra

David A. Harville

An Introduction to Generalized Linear Models, Fourth Edition

Annette J. Dobson and Adrian G. Barnett

Graphics for Statistics and Data Analysis with R

Kevin J. Keen

Statistics in Engineering
With Examples in MATLAB and R, Second Edition
Andrew V. Metcalfe, David A. Green, Tony Greenfield, Mahayaudin M. Mansor, Andrew Smith, and Jonathan Tuke

For more information about this series, please visit: https://www.crcpress.com/go/textsseries

Statistics in Engineering
With Examples in MATLAB® and R
Second Edition

Andrew Metcalfe
David Green
Tony Greenfield
Mahayaudin Mansor
Andrew Smith
Jonathan Tuke
MATLAB ® is a trademark of The MathWorks, Inc. and is used with permission. The MathWorks does not warrant the
accuracy of the text or exercises in this book. This book’s use or discussion of MATLAB ® software or related products
does not constitute endorsement or sponsorship by The MathWorks of a particular pedagogical approach or particular
use of the MATLAB ® software.

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

© 2019 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: 20191213

International Standard Book Number-13: 978-1-4398-9547-4 (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 copy-
right 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 pho-
tocopying, 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
Contents

Preface xvii

1 Why understand statistics? 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Using the book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Probability and making decisions 3

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Random digits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 Concepts and uses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.2 Generating random digits . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.3 Pseudo random digits . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Defining probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.1 Defining probabilities – Equally likely outcomes . . . . . . . . . . . . 8
2.3.2 Defining probabilities – Relative frequencies . . . . . . . . . . . . . . 11
2.3.3 Defining probabilities – Subjective probability and expected monetary
value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Axioms of probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 The addition rule of probability . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5.1 Complement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.6 Conditional probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.6.1 Conditioning on information . . . . . . . . . . . . . . . . . . . . . . 18
2.6.2 Conditional probability and the multiplicative rule . . . . . . . . . . 18
2.6.3 Independence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.6.4 Tree diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.7 Bayes’ theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.7.1 Law of total probability . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.7.2 Bayes’ theorem for two events . . . . . . . . . . . . . . . . . . . . . . 27
2.7.3 Bayes’ theorem for any number of events . . . . . . . . . . . . . . . 28
2.8 Decision trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.9 Permutations and combinations . . . . . . . . . . . . . . . . . . . . . . . . 31
2.10 Simple random sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.11.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.11.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 36
2.11.3 MATLAB R and R commands . . . . . . . . . . . . . . . . . . . . . 36
2.12 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

v
vi Contents

3 Graphical displays of data and descriptive statistics 55

3.1 Types of variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.2 Samples and populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.3 Displaying data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.3.1 Stem-and-leaf plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.3.2 Time series plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.3.3 Pictogram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.3.4 Pie chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.3.5 Bar chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.3.6 Rose plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.3.7 Line chart for discrete variables . . . . . . . . . . . . . . . . . . . . . 70
3.3.8 Histogram and cumulative frequency polygon for continuous variables 73
3.3.9 Pareto chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.4 Numerical summaries of data . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.4.1 Population and sample . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.4.2 Measures of location . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.4.3 Measures of spread . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.5 Box-plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.6 Outlying values and robust statistics . . . . . . . . . . . . . . . . . . . . . 97
3.6.1 Outlying values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.6.2 Robust statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.7 Grouped data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.7.1 Calculation of the mean and standard deviation for discrete data . 99
3.7.2 Grouped continuous data [Mean and standard deviation for grouped
continuous data] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.7.3 Mean as center of gravity . . . . . . . . . . . . . . . . . . . . . . . . 101
3.7.4 Case study of wave stress on offshore structure. . . . . . . . . . . . . 103
3.8 Shape of distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.8.1 Skewness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.8.2 Kurtosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.8.3 Some contrasting histograms . . . . . . . . . . . . . . . . . . . . . . 105
3.9 Multivariate data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.9.1 Scatter plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.9.2 Histogram for bivariate data . . . . . . . . . . . . . . . . . . . . . . 110
3.9.3 Parallel coordinates plot . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.10 Descriptive time series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.10.1 Definition of time series . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.10.2 Missing values in time series . . . . . . . . . . . . . . . . . . . . . . . 114
3.10.3 Decomposition of time series . . . . . . . . . . . . . . . . . . . . . . 114
3.10.3.1 Trend - Centered moving average . . . . . . . . . . . . . . 114
3.10.3.2 Seasonal component - Additive monthly model . . . . . . . 115
3.10.3.3 Seasonal component - Multiplicative monthly model . . . . 115
3.10.3.4 Seasonal adjustment . . . . . . . . . . . . . . . . . . . . . . 116
3.10.3.5 Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
3.10.4 Index numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
3.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.11.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.11.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 121
3.11.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 122
3.12 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Contents vii

4 Discrete probability distributions 137

4.1 Discrete random variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.1.1 Definition of a discrete probability distribution . . . . . . . . . . . . 138
4.1.2 Expected value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
4.2 Bernoulli trial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.2.2 Defining the Bernoulli distribution . . . . . . . . . . . . . . . . . . . 141
4.2.3 Mean and variance of the Bernoulli distribution . . . . . . . . . . . . 141
4.3 Binomial distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.3.2 Defining the Binomial distribution . . . . . . . . . . . . . . . . . . . 142
4.3.3 A model for conductivity . . . . . . . . . . . . . . . . . . . . . . . . 147
4.3.4 Mean and variance of the binomial distribution . . . . . . . . . . . . 148
4.3.5 Random deviates from binomial distribution . . . . . . . . . . . . . 149
4.3.6 Fitting a binomial distribution . . . . . . . . . . . . . . . . . . . . . 149
4.4 Hypergeometric distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.4.1 Defining the hypergeometric distribution . . . . . . . . . . . . . . . . 151
4.4.2 Random deviates from the hypergeometric distribution . . . . . . . 152
4.4.3 Fitting the hypergeometric distribution . . . . . . . . . . . . . . . . 152
4.5 Negative binomial distribution . . . . . . . . . . . . . . . . . . . . . . . . . 153
4.5.1 The geometric distribution . . . . . . . . . . . . . . . . . . . . . . . 153
4.5.2 Defining the negative binomial distribution . . . . . . . . . . . . . . 154
4.5.3 Applications of negative binomial distribution . . . . . . . . . . . . . 155
4.5.4 Fitting a negative binomial distribution . . . . . . . . . . . . . . . . 157
4.5.5 Random numbers from a negative binomial distribution . . . . . . . 157
4.6 Poisson process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
4.6.1 Defining a Poisson process in time . . . . . . . . . . . . . . . . . . . 158
4.6.2 Superimposing Poisson processes . . . . . . . . . . . . . . . . . . . . 158
4.6.3 Spatial Poisson process . . . . . . . . . . . . . . . . . . . . . . . . . 158
4.6.4 Modifications to Poisson processes . . . . . . . . . . . . . . . . . . . 159
4.6.5 Poisson distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
4.6.6 Fitting a Poisson distribution . . . . . . . . . . . . . . . . . . . . . . 160
4.6.7 Times between events . . . . . . . . . . . . . . . . . . . . . . . . . . 161
4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.7.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.7.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 162
4.7.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 163
4.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

5 Continuous probability distributions 175

5.1 Continuous random variables . . . . . . . . . . . . . . . . . . . . . . . . . 175
5.1.1 Definition of a continuous random variable . . . . . . . . . . . . . . 175
5.1.2 Definition of a continuous probability distribution . . . . . . . . . . 176
5.1.3 Moments of a continuous probability distribution . . . . . . . . . . . 177
5.1.4 Median and mode of a continuous probability distribution . . . . . . 181
5.1.5 Parameters of probability distributions . . . . . . . . . . . . . . . . . 181
5.2 Uniform distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
5.2.1 Definition of a uniform distribution . . . . . . . . . . . . . . . . . . . 182
5.2.2 Applications of the uniform distribution . . . . . . . . . . . . . . . . 183
5.2.3 Random deviates from a uniform distribution . . . . . . . . . . . . . 183
5.2.4 Distribution of F (X) is uniform . . . . . . . . . . . . . . . . . . . . 183
viii Contents

5.2.5 Fitting a uniform distribution . . . . . . . . . . . . . . . . . . . . . . 184

5.3 Exponential distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
5.3.1 Definition of an exponential distribution . . . . . . . . . . . . . . . . 184
5.3.2 Markov property . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
5.3.2.1 Poisson process . . . . . . . . . . . . . . . . . . . . . . . . . 186
5.3.2.2 Lifetime distribution . . . . . . . . . . . . . . . . . . . . . . 186
5.3.3 Applications of the exponential distribution . . . . . . . . . . . . . . 187
5.3.4 Random deviates from an exponential distribution . . . . . . . . . . 189
5.3.5 Fitting an exponential distribution . . . . . . . . . . . . . . . . . . . 190
5.4 Normal (Gaussian) distribution . . . . . . . . . . . . . . . . . . . . . . . . 194
5.4.1 Definition of a normal distribution . . . . . . . . . . . . . . . . . . . 194
5.4.2 The standard normal distribution Z ∼ N (0, 1) . . . . . . . . . . . . 195
5.4.3 Applications of the normal distribution . . . . . . . . . . . . . . . . 199
5.4.4 Random numbers from a normal distribution . . . . . . . . . . . . . 203
5.4.5 Fitting a normal distribution . . . . . . . . . . . . . . . . . . . . . . 203
5.5 Probability plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
5.5.1 Quantile-quantile plots . . . . . . . . . . . . . . . . . . . . . . . . . . 204
5.5.2 Probability plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
5.6 Lognormal distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
5.6.1 Definition of a lognormal distribution . . . . . . . . . . . . . . . . . 205
5.6.2 Applications of the lognormal distribution . . . . . . . . . . . . . . . 208
5.6.3 Random numbers from lognormal distribution . . . . . . . . . . . . . 209
5.6.4 Fitting a lognormal distribution . . . . . . . . . . . . . . . . . . . . . 209
5.7 Gamma distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
5.7.1 Definition of a gamma distribution . . . . . . . . . . . . . . . . . . . 210
5.7.2 Applications of the gamma distribution . . . . . . . . . . . . . . . . 212
5.7.3 Random deviates from gamma distribution . . . . . . . . . . . . . . 212
5.7.4 Fitting a gamma distribution . . . . . . . . . . . . . . . . . . . . . . 212
5.8 Gumbel distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
5.8.1 Definition of a Gumbel distribution . . . . . . . . . . . . . . . . . . . 213
5.8.2 Applications of the Gumbel distribution . . . . . . . . . . . . . . . . 215
5.8.3 Random deviates from a Gumbel distribution . . . . . . . . . . . . . 215
5.8.4 Fitting a Gumbel distribution . . . . . . . . . . . . . . . . . . . . . . 216
5.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5.9.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5.9.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 218
5.9.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 219
5.10 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220

6 Correlation and functions of random variables 233

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6.2 Sample covariance and correlation coefficient . . . . . . . . . . . . . . . . . 236
6.2.1 Defining sample covariance . . . . . . . . . . . . . . . . . . . . . . . 236
6.3 Bivariate distributions, population covariance and correlation coefficient . 244
6.3.1 Population covariance and correlation coefficient . . . . . . . . . . . 245
6.3.2 Bivariate distributions - Discrete case . . . . . . . . . . . . . . . . . 246
6.3.3 Bivariate distributions - Continuous case . . . . . . . . . . . . . . . 248
6.3.3.1 Marginal distributions . . . . . . . . . . . . . . . . . . . . . 248
6.3.3.2 Bivariate histogram . . . . . . . . . . . . . . . . . . . . . . 249
6.3.3.3 Covariate and correlation . . . . . . . . . . . . . . . . . . . 250
6.3.3.4 Bivariate probability distributions . . . . . . . . . . . . . . 251
Contents ix

6.3.4 Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256

6.4 Linear combination of random variables (propagation of error) . . . . . . . 256
6.4.1 Mean and variance of a linear combination of random variables . . . 257
6.4.1.1 Bounds for correlation coefficient . . . . . . . . . . . . . . 259
6.4.2 Linear combination of normal random variables . . . . . . . . . . . . 260
6.4.3 Central Limit Theorem and distribution of the sample mean . . . . 262
6.5 Non-linear functions of random variables (propagation of error) . . . . . . 265
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
6.6.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
6.6.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 267
6.6.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 268
6.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268

7 Estimation and inference 279

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
7.2 Statistics as estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
7.2.1 Population parameters . . . . . . . . . . . . . . . . . . . . . . . . . 280
7.2.2 Sample statistics and sampling distributions . . . . . . . . . . . . . . 280
7.2.3 Bias and MSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
7.3 Accuracy and precision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
7.4 Precision of estimate of population mean . . . . . . . . . . . . . . . . . . . 285
7.4.1 Confidence interval for population mean when σ known . . . . . . . 285
7.4.2 Confidence interval for mean when σ unknown . . . . . . . . . . . . 288
7.4.2.1 Construction of confidence interval and rationale for the
t-distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 288
7.4.2.2 The t-distribution . . . . . . . . . . . . . . . . . . . . . . . 289
7.4.3 Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
7.4.4 Bootstrap methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
7.4.4.1 Bootstrap resampling . . . . . . . . . . . . . . . . . . . . . 292
7.4.4.2 Basic bootstrap confidence intervals . . . . . . . . . . . . . 293
7.4.4.3 Percentile bootstrap confidence intervals . . . . . . . . . . 293
7.4.5 Parametric bootstrap . . . . . . . . . . . . . . . . . . . . . . . . . . 296
7.5 Hypothesis testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
7.5.1 Hypothesis test for population mean when σ known . . . . . . . . . 300
7.5.2 Hypothesis test for population mean when σ unknown . . . . . . . . 302
7.5.3 Relation between a hypothesis test and the confidence interval . . . 303
7.5.4 p-value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
7.5.5 One-sided confidence intervals and one-sided tests . . . . . . . . . . 304
7.6 Sample size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
7.7 Confidence interval for a population variance and standard deviation . . . 307
7.8 Comparison of means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
7.8.1 Independent samples . . . . . . . . . . . . . . . . . . . . . . . . . . 309
7.8.1.1 Population standard deviations differ . . . . . . . . . . . . 309
7.8.1.2 Population standard deviations assumed equal . . . . . . . 312
7.8.2 Matched pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
7.9 Comparing variances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
7.10 Inference about proportions . . . . . . . . . . . . . . . . . . . . . . . . . . 318
7.10.1 Single sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
7.10.2 Comparing two proportions . . . . . . . . . . . . . . . . . . . . . . . 320
7.10.3 McNemar’s test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
7.11 Prediction intervals and statistical tolerance intervals . . . . . . . . . . . . 325
x Contents

7.11.1 Prediction interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325

7.11.2 Statistical tolerance interval . . . . . . . . . . . . . . . . . . . . . . . 326
7.12 Goodness of fit tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
7.12.1 Chi-square test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
7.12.2 Empirical distribution function tests . . . . . . . . . . . . . . . . . . 330
7.13 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
7.13.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
7.13.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 333
7.13.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 335
7.14 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335

8 Linear regression and linear relationships 357

8.1 Linear regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
8.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
8.1.2 The model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
8.1.3 Fitting the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
8.1.3.1 Fitting the regression line . . . . . . . . . . . . . . . . . . . 361
8.1.3.2 Identical forms for the least squares estimate of the slope . 363
8.1.3.3 Relation to correlation . . . . . . . . . . . . . . . . . . . . 363
8.1.3.4 Alternative form for the fitted regression line . . . . . . . . 364
8.1.3.5 Residuals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
8.1.3.6 Identities satisfied by the residuals . . . . . . . . . . . . . . 366
8.1.3.7 Estimating the standard deviation of the errors . . . . . . . 367
8.1.3.8 Checking assumptions A3, A4 and A5 . . . . . . . . . . . . 368
8.1.4 Properties of the estimators . . . . . . . . . . . . . . . . . . . . . . . 368
8.1.4.1 Estimator of the slope . . . . . . . . . . . . . . . . . . . . . 369
8.1.4.2 Estimator of the intercept . . . . . . . . . . . . . . . . . . . 371
8.1.5 Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
8.1.5.1 Confidence interval for mean value of Y given x . . . . . . 371
8.1.5.2 Limits of prediction . . . . . . . . . . . . . . . . . . . . . . 373
8.1.5.3 Plotting confidence intervals and prediction limits . . . . . 374
8.1.6 Summarizing the algebra . . . . . . . . . . . . . . . . . . . . . . . . 375
8.1.7 Coefficient of determination R2 . . . . . . . . . . . . . . . . . . . . . 376
8.2 Regression for a bivariate normal distribution . . . . . . . . . . . . . . . . 376
8.2.1 The bivariate normal distribution . . . . . . . . . . . . . . . . . . . . 377
8.3 Regression towards the mean . . . . . . . . . . . . . . . . . . . . . . . . . 378
8.4 Relationship between correlation and regression . . . . . . . . . . . . . . . 380
8.4.1 Values of x are assumed to be measured without error and can be
preselected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
8.4.2 The data pairs are assumed to be a random sample from a bivariate
normal distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
8.5 Fitting a linear relationship when both variables are measured with error . 383
8.6 Calibration lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
8.7 Intrinsically linear models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
8.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
8.8.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
8.8.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 393
8.8.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 394
8.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
Contents xi

9 Multiple regression 403

9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
9.2 Multivariate data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
9.3 Multiple regression model . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
9.3.1 The linear model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
9.3.2 Random vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
9.3.2.1 Linear transformations of a random vector . . . . . . . . . 406
9.3.2.2 Multivariate normal distribution . . . . . . . . . . . . . . . 407
9.3.3 Matrix formulation of the linear model . . . . . . . . . . . . . . . . . 407
9.3.4 Geometrical interpretation . . . . . . . . . . . . . . . . . . . . . . . . 407
9.4 Fitting the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
9.4.1 Principle of least squares . . . . . . . . . . . . . . . . . . . . . . . . 408
9.4.2 Multivariate calculus - Three basic results . . . . . . . . . . . . . . 409
9.4.3 The least squares estimator of the coefficients . . . . . . . . . . . . . 410
9.4.4 Estimating the coefficients . . . . . . . . . . . . . . . . . . . . . . . . 411
9.4.5 Estimating the standard deviation of the errors . . . . . . . . . . . . 416
9.4.6 Standard errors of the estimators of the coefficients . . . . . . . . . . 417
9.5 Assessing the fit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
9.5.1 The residuals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
9.5.2 R-squared . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
9.5.3 F-statistic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
9.5.4 Cross validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
9.6 Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
9.7 Building multiple regression models . . . . . . . . . . . . . . . . . . . . . . 424
9.7.1 Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
9.7.2 Categorical variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
9.7.3 F-test for an added set of variables . . . . . . . . . . . . . . . . . . . 433
9.7.4 Quadratic terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
9.7.5 Guidelines for fitting regression models . . . . . . . . . . . . . . . . . 447
9.8 Time series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
9.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
9.8.2 Aliasing and sampling intervals. . . . . . . . . . . . . . . . . . . . . . 450
9.8.3 Fitting a trend and seasonal variation with regression . . . . . . . . 451
9.8.4 Auto-covariance and auto-correlation . . . . . . . . . . . . . . . . . . 456
9.8.4.1 Defining auto-covariance for a stationary times series model 457
9.8.4.2 Defining sample auto-covariance and the correlogram . . . 458
9.8.5 Auto-regressive models . . . . . . . . . . . . . . . . . . . . . . . . . . 459
9.8.5.1 AR(1) and AR(2) models . . . . . . . . . . . . . . . . . . . 460
9.9 Non-linear least squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
9.10 Generalized linear model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
9.10.1 Logistic regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
9.10.2 Poisson regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
9.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
9.11.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
9.11.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 474
9.11.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 475
9.12 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
xii Contents

10 Statistical quality control 491

10.1 Continuous improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
10.1.1 Defining quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
10.1.2 Taking measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 492
10.1.3 Avoiding rework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
10.1.4 Strategies for quality improvement . . . . . . . . . . . . . . . . . . . 494
10.1.5 Quality management systems . . . . . . . . . . . . . . . . . . . . . . 494
10.1.6 Implementing continuous improvement . . . . . . . . . . . . . . . . . 495
10.2 Process stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
10.2.1 Runs chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
10.2.2 Histograms and box plots . . . . . . . . . . . . . . . . . . . . . . . . 499
10.2.3 Components of variance . . . . . . . . . . . . . . . . . . . . . . . . . 501
10.3 Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
10.3.1 Process capability index . . . . . . . . . . . . . . . . . . . . . . . . . 510
10.3.2 Process performance index . . . . . . . . . . . . . . . . . . . . . . . 511
10.3.3 One-sided process capability indices . . . . . . . . . . . . . . . . . . 512
10.4 Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
10.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514
10.4.1.1 Reliability of components . . . . . . . . . . . . . . . . . . . 514
10.4.1.2 Reliability function and the failure rate . . . . . . . . . . . 515
10.4.2 Weibull analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517
10.4.2.1 Definition of the Weibull distribution . . . . . . . . . . . . 517
10.4.2.2 Weibull quantile plot . . . . . . . . . . . . . . . . . . . . . 518
10.4.2.3 Censored data . . . . . . . . . . . . . . . . . . . . . . . . . 522
10.4.3 Maximum likelihood . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
10.4.4 Kaplan-Meier estimator of reliability . . . . . . . . . . . . . . . . . . 529
10.5 Acceptance sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
10.6 Statistical quality control charts . . . . . . . . . . . . . . . . . . . . . . . . 533
10.6.1 Shewhart mean and range chart for continuous variables . . . . . . . 533
10.6.1.1 Mean chart . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
10.6.1.2 Range chart . . . . . . . . . . . . . . . . . . . . . . . . . . 535
10.6.2 p-charts for proportions . . . . . . . . . . . . . . . . . . . . . . . . . 538
10.6.3 c-charts for counts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
10.6.4 Cumulative sum charts . . . . . . . . . . . . . . . . . . . . . . . . . 542
10.6.5 Multivariate control charts . . . . . . . . . . . . . . . . . . . . . . . 544
10.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
10.7.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
10.7.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 548
10.7.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 550
10.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550

11 Design of experiments with regression analysis 559

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
11.2 Factorial designs with factors at two levels . . . . . . . . . . . . . . . . . . 562
11.2.1 Full factorial designs . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
11.2.1.1 Setting up a 2k design . . . . . . . . . . . . . . . . . . . . . 562
11.2.1.2 Analysis of 2k design . . . . . . . . . . . . . . . . . . . . . 565
11.3 Fractional factorial designs . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
11.4 Central composite designs . . . . . . . . . . . . . . . . . . . . . . . . . . . 585
11.5 Evolutionary operation (EVOP) . . . . . . . . . . . . . . . . . . . . . . . . 593
11.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
Contents xiii

11.6.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597

11.6.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 597
11.6.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 598
11.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598

12 Design of experiments and analysis of variance 605

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
12.2 Comparison of several means with one-way ANOVA . . . . . . . . . . . . 605
12.2.1 Defining the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
12.2.2 Limitation of multiple t-tests . . . . . . . . . . . . . . . . . . . . . . 606
12.2.3 One-way ANOVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607
12.2.4 Testing H0O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
12.2.5 Follow up procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
12.3 Two factors at multiple levels . . . . . . . . . . . . . . . . . . . . . . . . . 613
12.3.1 Two factors without replication (two-way ANOVA) . . . . . . . . . . 614
12.3.2 Two factors with replication (three-way ANOVA) . . . . . . . . . . . 618
12.4 Randomized block design . . . . . . . . . . . . . . . . . . . . . . . . . . . 621
12.5 Split plot design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 626
12.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636
12.6.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636
12.6.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 637
12.6.3 MATLAB and R commands . . . . . . . . . . . . . . . . . . . . . . . 637
12.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638

13 Probability models 649

13.1 System reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
13.1.1 Series system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
13.1.2 Parallel system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
13.1.3 k-out-of-n system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
13.1.4 Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
13.1.5 Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
13.1.6 Paths and cut sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
13.1.7 Reliability function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
13.1.8 Redundancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
13.1.9 Non-repairable systems . . . . . . . . . . . . . . . . . . . . . . . . . 658
13.1.10 Standby systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
13.1.11 Common cause failures . . . . . . . . . . . . . . . . . . . . . . . . . . 661
13.1.12 Reliability bounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
13.2 Markov chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
13.2.1 Discrete Markov chain . . . . . . . . . . . . . . . . . . . . . . . . . . 663
13.2.2 Equilibrium behavior of irreducible Markov chains . . . . . . . . . . 667
13.2.3 Methods for solving equilibrium equations . . . . . . . . . . . . . . . 670
13.2.4 Absorbing Markov chains . . . . . . . . . . . . . . . . . . . . . . . . 675
13.2.5 Markov chains in continuous time . . . . . . . . . . . . . . . . . . . 681
13.3 Simulation of systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 684
13.3.1 The simulation procedure . . . . . . . . . . . . . . . . . . . . . . . . 685
13.3.2 Drawing inference from simulation outputs . . . . . . . . . . . . . . 689
13.3.3 Variance reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 692
13.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694
13.4.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694
13.4.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 694
xiv Contents

13.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696

14 Sampling strategies 699

14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 699
14.2 Simple random sampling from a finite population . . . . . . . . . . . . . . 702
14.2.1 Finite population correction . . . . . . . . . . . . . . . . . . . . . . . 702
14.2.2 Randomization theory . . . . . . . . . . . . . . . . . . . . . . . . . . 703
14.2.2.1 Defining the simple random sample . . . . . . . . . . . . . 703
14.2.2.2 Mean and variance of sample mean . . . . . . . . . . . . . 704
14.2.2.3 Mean and variance of estimator of population total . . . . 705
14.2.3 Model based analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 707
14.2.4 Sample size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 708
14.3 Stratified sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 708
14.3.1 Principle of stratified sampling . . . . . . . . . . . . . . . . . . . . . 709
14.3.2 Estimating the population mean and total . . . . . . . . . . . . . . . 709
14.3.3 Optimal allocation of the sample over strata . . . . . . . . . . . . . . 711
14.4 Multi-stage sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713
14.5 Quota sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
14.6 Ratio estimators and regression estimators . . . . . . . . . . . . . . . . . . 716
14.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
14.6.2 Regression estimators . . . . . . . . . . . . . . . . . . . . . . . . . . 716
14.6.3 Ratio estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
14.7 Calibration of the unit cost data base . . . . . . . . . . . . . . . . . . . . . 718
14.7.1 Sources of error in an AMP . . . . . . . . . . . . . . . . . . . . . . . 718
14.7.2 Calibration factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719
14.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
14.8.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
14.8.2 Summary of main results . . . . . . . . . . . . . . . . . . . . . . . . 721
14.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 722

Appendix A - Notation 727

A.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
A.2 Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
A.3 Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 728
A.4 Probability distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 729

Appendix B - Glossary 731

Appendix C - Getting started in R 745

C.1 Installing R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
C.2 Using R as a calculator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
C.3 Setting the path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747
C.4 R scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747
C.5 Data entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747
C.5.1 From keyboard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 747
C.5.2 From a file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 748
C.5.2.1 Single variable . . . . . . . . . . . . . . . . . . . . . . . . . 748
C.5.2.2 Several variables . . . . . . . . . . . . . . . . . . . . . . . . 748
C.6 R vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
C.7 User defined functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
C.8 Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
Contents xv

C.9 Loops and conditionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751

C.10 Basic plotting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 752
C.11 Installing packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753
C.12 Creating time series objects . . . . . . . . . . . . . . . . . . . . . . . . . . . 753

Appendix D - Getting started in MATLAB 755

D.1 Installing MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 755
D.2 Using MATLAB as a calculator . . . . . . . . . . . . . . . . . . . . . . . . 755
D.3 Setting the path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756
D.4 MATLAB scripts (m-files) . . . . . . . . . . . . . . . . . . . . . . . . . . . 756
D.5 Data entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757
D.5.1 From keyboard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757
D.5.2 From a file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757
D.5.2.1 Single variable . . . . . . . . . . . . . . . . . . . . . . . . . 757
D.5.2.2 Several variables . . . . . . . . . . . . . . . . . . . . . . . . 758
D.6 MATLAB vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 758
D.7 User defined functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
D.8 Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
D.9 Loops and conditionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
D.10 Basic plotting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 763
D.11 Creating time series objects . . . . . . . . . . . . . . . . . . . . . . . . . . . 764

Appendix E - Experiments 765

E.1 How good is your probability assessment? . . . . . . . . . . . . . . . . . . 765
E.1.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
E.1.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
E.1.3 Question sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765
E.1.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
E.1.5 Follow up questions . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
E.2 Buffon’s needle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
E.2.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
E.2.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
E.2.3 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
E.2.4 Computer simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
E.2.5 Historical note . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
E.3 Robot rabbit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
E.3.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 768
E.3.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769
E.3.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 770
E.3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 770
E.3.5 Follow up question . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
E.4 Use your braking brains . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
E.4.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
E.4.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
E.4.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772
E.5 Predicting descent time from payload . . . . . . . . . . . . . . . . . . . . . 773
E.5.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773
E.5.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773
E.5.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774
E.5.4 Follow up question . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774
E.6 Company efficiency, resources and teamwork . . . . . . . . . . . . . . . . . 774
xvi Contents

E.6.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774

E.6.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774
E.6.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
E.7 Factorial experiment – reaction times by distraction, dexterity and
distinctness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
E.7.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
E.7.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
E.7.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
E.7.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 777
E.7.5 Follow up questions . . . . . . . . . . . . . . . . . . . . . . . . . . . 777
E.8 Weibull analysis of cycles to failure . . . . . . . . . . . . . . . . . . . . . . 778
E.8.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 778
E.8.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 778
E.8.3 Weibull plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 778
E.8.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 779
E.9 Control or tamper? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 779
E.10 Where is the summit? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 781

References 783

Index 789
Preface

Engineering is a wide ranging discipline with a common theme of mathematical modeling.

Probability and statistics is the field of mathematics that deals with uncertainty and vari-
ation, features that are part of every engineering project. Engineering applications have
provided inspiration for the development of mathematics, and just a few examples from
the last century alone are Shannon’s theory of communication, Shewhart’s focus on the im-
provement of industrial processes, Wiener’s contribution to signal processing and robotics,
and Gumbel’s research into extreme values in hydrology.
We aim to motivate students of engineering by demonstrating that probability and statis-
tics are an essential part of engineering design, enabling engineers to assess performance
and to quantify risk. Uncertainties include variation in raw materials, variation in manufac-
turing processes, and the volatile natural environment within which engineered structures
operate. Our emphasis is on modeling and simulation. Engineering students generally have
a good mathematical aptitude and we present the mathematical basis for statistical meth-
ods in a succinct manner, that in places assumes a knowledge of elementary calculus. More
mathematical detail is given in the appendices. We rely on a large number of data sets that
have either been provided by companies or are available from data archives maintained
by various organizations. We appreciate the permission to use these data and thank those
involved. All of these data sets are available on the book website.
A feature of the book is the emphasis on stochastic simulation, enabled by the generation
of pseudo-random numbers. The principal reason for this emphasis is that engineers will
generally perform stochastic simulation studies as part of their design work, but it has other
advantages. Stochastic simulation provides enlightening demonstrations of the concept of
sampling distributions, which is central to statistical analysis but unfamiliar to students
beginning the subject. Stochastic simulation is also part of modern statistical analysis,
enabling bootstrap methods and Bayesian analyses.
The first eight chapters are designed to be read in sequence, although chapters two
and three could be covered in reverse order. The following six chapters cover: multiple
regression; the design of experiments; statistical quality control; probability models; and
sampling strategies. We use the multiple regression model to introduce the important topics
of time series analysis and the design of experiments.
The emphasis on stochastic simulation does not diminish the importance of physical
experiments and investigations. We include an appendix of ten experiments that we have
used with classes of students, alternated with computer based practical classes. These sim-
ple experiments, for example descent times of paper helicopters with paperclip payloads
and cycles to failure when bending paperclips, offer scope for discussing principles of good
experimentation and observing how well mathematical models work in practice. They also
provide an opportunity for students to work together in small groups, which have suggested
intriguing designs for paper helicopters.
The choice of the mathematical software MATLAB R and R rather than a statistical
package provides a good combination of an interactive environment for numerical computa-
tion, statistical analysis, tools for visualization, and facility for programming and simulation.
In particular, MATLAB is very widely used in the engineering community for design and

xvii
xviii Preface

simulation work and has a wide range of inbuilt statistical functions. The R software has
similar capabilities, and has the potential advantage of being open source. It too has a
wide range of inbuilt statistical functions augmented with hundreds of specialist packages
that are available on the CRAN website. Many advances in statistical methodology are
accompanied by new packages written in R.
The exercises at the end of each chapter are an essential part of the text, and are orga-
nized by targeting each section within the chapter and followed by more general exercises.
The exercises fall into three categories: routine practice of the ideas presented; additions to
the explanatory material in the chapter including details of derivations, special cases and
counter examples; and extensions of the material in the chapter. Additional exercises, and
solutions including code are given to odd numbered exercises on the website.
We thank John Kimmel for his generous support and encouragement. We are also grate-
ful to several anonymous reviewers for their helpful comments and constructive advice.

Andrew Metcalfe
David Green
Tony Greenfield
Mahayaudin Mansor
Andrew Smith
Jonathan Tuke

MATLAB R is a registered trademark of The MathWorks, Inc. For 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: [email protected]
Web: www.mathworks.com
1
Why understand statistics?

Engineers need to take account of the uncertainty in the environment and to assess how
engineered products will perform under extreme conditions. They have to contend with er-
rors in measurement and signals that are corrupted by noise, and to allow for variation in
raw materials and components from suppliers. Probability and statistics enable engineers to
model and to quantify uncertainty and to make appropriate allowances for it.

1.1 Introduction

The Voyager 1 and Voyager 2 spacecraft were launched from Cape Canaveral in 1977, taking
advantage of a favorable alignment of the outer planets in the solar system. Thirty five years
later Voyager 1 entered interstellar space traveling “further than anyone or anything in
history” [The Times, 2017]. The trajectory of Voyager 2 included flybys of Jupiter, Saturn,
Uranus and Neptune and the spacecraft is now in the heliosheath where the solar wind is
compressed and turbulent. The robotic spacecraft have control systems that keep their high
gain antennas pointing towards the earth. They have the potential to transmit scientific data
until around 2020 when the radioisotope thermoelectric generators will no longer provide
sufficient power.

The work of renowned engineers such as Rudolf Kalman and Norbert Weiner in electrical
engineering, in particular control theory and robotics, Claude Shannon in communication
theory, and Waloddi Weibull in reliability theory is directly applicable to the space program.
Moreover, statistics is an essential part of all engineering disciplines. A glance at the titles
of journals published by American Society of Civil Engineers (ASCE), American Society
of Mechanical Engineers (ASME), and Institute of Electrical and Electronics Engineers
(IEEE) give an indication of the wide range of applications. These applications have also
led to advances in statistical theory, as seen, for example, in the work of: Emil Gumbel
in hydrology and Walter Shewhart [Shewhart, 1939] in manufacturing. In this book we
consider examples from many engineering disciplines including: hydrology; water quality;
strengths of materials; mining engineering; ship building; chemical processes; electrical and
mechanical engineering; and management.

Engineers have always had to deal with uncertainty, but they are now expected to do
so in more accountable ways. Probability theory provides a mathematical description of
random variation and enables us to make realistic risk assessments. Statistics is the analysis
of data and the subsequent fitting of probability models.

1
2 Statistics in Engineering, Second Edition

1.2 Using the book

The first eight chapters are best read in sequence, although Chapter 3 could be read before
most of Chapter 2. The exercises include: routine examples; further detail for some of the
explanations given in chapters; extensions of the theory presented in chapters; and a few
challenges. Numerical answers to odd numbered exercises are given on the book website,
together with sets of additional exercises.
The following chapters cover more advanced topics, although the mathematical detail is
kept at a similar level to that in the preceding chapters. Chapter 9 on multiple regression is
a pre-requisite for the two chapters on the design of experiments. The other three chapters,
on statistical quality control, probability models, and sampling strategies, can be read in any
order. Of these, probability models relies only on the material on probability in Chapter 2,
whereas the other two chapters assume some familiarity with most of the material in the
first eight chapters.
Appendix D.11 contains experiments which have been designed to complement computer
based investigations. The experiments are simple, such as descent time of paper helicopters
with paper-clip payloads and cycles to failure for bending paperclips, but the rationale
is that doing something is more convincing and more memorable than reading about it.
The website includes answers to odd numbered exercises, additional exercises, and informal
derivations of key results. These include the Central Limit Theorem, Gumbel’s extreme
value distribution, and more detail on the multiple regression model. The proofs of these
results rely only on standard calculus and matrix algebra and show something of the diverse
applications of fundamental mathematics. These proofs are intriguing, but are not needed
to follow the main text.

1.3 Software
We have chosen to implement the analyses in two software environments: R and
MATLAB R . The flexibility of command line programming is offset against the convenience
of menu driven statistical packages such as Minitab. Appendices D and G are tutorials
to get you started in R and MATLAB respectively. Both R and MATLAB have built in
functions for a wide range of modern statistical analyses. To see what a particular com-
mand such as plot does in R, type help(plot) and in MATLAB type help plot. R has
the great advantage of being open source, and hence free for anyone to use, and many re-
search statisticians chose to develop the subject by providing new packages for R. However,
MATLAB is very well suited to engineering applications and is widely used in industry and
universities. Moving between the two is partially facilitated by Hiebeler’s MATLAB/R Ref-
erence [Hiebeler, 2010] and complemented where possible in a direct function comparison
at the end of each chapter. Other useful computing resources are Short’s R Reference card
[Short, 2004], and many of the posts on the World Wide Web (www) once one has learnt the
basic principles of a computing language. In general there are several ways of programming
a statistical routine and also packages and toolboxes that automate the entire process. We
have aimed to use R and MATLAB to demonstrate calculations that follow the statistical
development in the book, and to show the use of the standard statistical functions.
2
Probability and making decisions

Three approaches to defining probability are introduced. We explain the fundamental rules of
probability and use these to solve a variety of problems. Expected monetary value is defined
and applied in conjunction with decision trees. Permutations and combinations are defined
and we make a link with the equally likely definitions of probability. We discuss the concept
of random digits and their use for drawing simple random samples from a population. See
relevant examples in Appendix E:

Appendix E.1 How good is your probability assessment?

Appendix E.2 Buffon’s needle

2.1 Introduction
The Australian Bureau of Meteorology Adelaide Forecast gives the chance of any rain
tomorrow, a summer day in South Australia, as 5%. We will see that it is natural to express
the chance of an uncertain event, such as rain tomorrow, occurring as a probability on
a scale from 0 to 1. If an event is as likely to occur as it is not to occur, then it has a
probability of occurrence of 0.5. An impossible event has a probability of 0 and a certain
event has a probability of 1. Formally, the Bureau’s chance of 5% is a probability of 0.05,
and as this is considerably closer to 0 than 1 we think it is unlikely to rain tomorrow. There
are several ways of giving a more precise interpretation. One is to imagine that similar
weather patterns to today’s have been observed in Adelaide on many occasions during the
Australian summer, and that on 5% of these occasions it has rained on the next day. Another
interpretation is based on the notion of a fair bet (formally defined in Section 2.3.3). The
weather forecaster thinks that the possibility of paying out $95 if it rains is offset by the
more likely outcome of receiving $5 if it is dry. Many engineering decisions are based on such
expert assessments of probability. For example, after drilling a well an oil company must
decide either to prepare it for oil production or to plug and abandon it. Before drilling at a
specific location, the company will seek a geologist’s opinion about the probability of finding
sufficient oil to justify production. There are various strategies, including the notion of a fair
bet, for making assessments of probabilities. One, the quadratic rule [Lindley, 1985], which
has been used in the U.S. in the training of weather forecasters, is covered in Experiment
E.1. Others are discussed later in the chapter.
There is, however, a basic approach to defining probability, which is applicable in special
cases when we can define the outcomes of some experiment so that they are equally likely
to occur. In this context, an experiment is any action that has an uncertain outcome.
Typical experiments that are supposed to have equally likely possible outcomes are games of
chance played with carefully constructed apparatus such as dice, cards, and roulette wheels.
The claim that outcomes are equally likely is based on the symmetry of the apparatus. For
example, all the cards in a deck should have the same physical dimensions and all slots and

3
4 Statistics in Engineering, Second Edition

frets on the roulette wheel should have the same physical dimensions1 . The equally likely
definition of probability was developed in the context of gambling games by Gerolamo
Cardano (1501-1576) and other mathematicians including Galileo, Pascal and Fermat in
the sixteenth century [David, 1955]. Cards and dice may seem unconnected to engineering,
but the generation of digits from 0 up to 9, such that each digit is equally likely to appear
as the next in sequence, is the basis of stochastic simulation. Simulation studies have a wide
range of applications including engineering design and analysis.

2.2 Random digits

2.2.1 Concepts and uses
You can use the R function sample() to obtain a sequence in which each digit from 0 up
to 9 appears equally likely to occur at each turn. To obtain such a sequence of length 20:

> sample(0:9, size=20, replace=TRUE)

[1] 6 2 4 2 8 3 0 8 8 1 3 5 1 6 5 2 8 6 7 0

although your sequence will be different. The R function sample(x,n, replace=TRUE)

takes a sample of size n from the object x as follows. Imagine each element of x has an
associated ticket that is put into a hat. Then n tickets are drawn from the hat, sequentially
with the selected ticket returned to the hat before the next ticket is drawn, in an idealized
manner such that at every draw each ticket in the hat is equally likely to be drawn. The
third argument in the function call, replace=TRUE, gives sampling with replacement, which
means that the ticket is returned to the hat after each draw.
If a ticket is not replaced after being drawn the sampling is without replacement and
this, more common application, is the default for the sample() function. So, when sampling
without replacement the function call sample(x,n,replace=FALSE) can be shortened to
sample(x,n). One use of such sequences is the selection of a random sample from a finite
population.

Example 2.1: Island ferry safety [random selection]

An island is served from a city on the mainland by a fleet of ferries, which each carry
12 inflatable life rafts in hard shelled canisters. The port authority requires that the
crew of each ferry demonstrates the launch of one life raft each year. The life raft will
be chosen at random from the 12, in such a way that each life raft has the same chance
of selection, so that all concerned can agree that there is no possibility of performing
the demonstration with a specially prepared life raft.

1 This understanding of probability can be traced back over thousands of years. The oldest known dice

were excavated as part of a 5 000-year-old backgammon set, at Shahr-i Sokhta in Iran. The concept of
equally likely outcomes is the basis for the kleroterion allotment machine, that was used by the Athenians
in the third century BC for selecting jurors and other representatives. Substantial remnants of a kleroterion
can be seen in the Agora Museum, in Athens.
Probability and making decisions 5

A sequence of random digits can be used to make the selection. One way to do this
is as follows: number the life rafts from 01 to 12; pair consecutive random digits; and
take the life raft corresponding to the first pair in the range from 01 up to 12. With the
sequence 6 2 4 2 8 3 0 8 . . ., the pairing gives: 62, 42, 83, 08, . . ., and life raft 8 would be
selected. If we were asked to sample more than one lifeboat, identifying the 12 lifeboats
by the pairs from 01 to 12 only, and ignoring all pairs between 13 and 99, might require
a long sequence of digits. You are asked to devise a more efficient identification in
Exercise 2.5. Direct use of the software functions is more convenient, as in Exercise 2.6.

Apart from sampling, sequences of random digits are used for computer simulations of
random processes. One example is the simulation of random wave forces and the calculation
of their effects on off-shore structures such as oil rigs and wave energy converters. Other
uses include Monte-Carlo strategies for computing probabilities.

2.2.2 Generating random digits

How can we produce sequences of random digits between 0 and 9, without relying on
software? In principle we might: flip a fair coin, taking a head as 0 and a tail as 1, to obtain
a sequence of random binary digits; take consecutive sets of four digits, and convert from
binary to base 10; accept the digit if the base 10 integers is between 0 and 9, and reject base
10 integers between 10 and 15. For most purposes this would be far too slow. A variant on
a roulette wheel, with an equal number of each digit from 0 to 9, would be a rather more
convenient generator of random digits but it would still be far too slow for simulations. There
is also the impracticality of making a roulette wheel, or any other mechanical apparatus,
perfectly fair 2 .
Given the limitations of mechanical devices for producing random digits, it is natural
to consider electronic alternatives. In 1926, John Johnson, a physicist at Bell Labs, noticed
random fluctuations in the voltage across the terminals of a resistor that he attributed to
thermal agitation of electrons. This noise can be amplified, sampled, and digitized into a se-
quence of 0s and 1s by taking sampled voltages below the nominal voltage as 0 and sampled
voltages above the nominal voltage as 1. Such an electronic device needs to be very stable
and, in particular, the nominal voltage has to be maintained so that a sampled value is
equally likely to be a 0 as a 1. Provided the sampling rate is slow relative to typical fluctua-
tions in voltage, the next binary digit is equally likely to be a 0 as a 1, regardless of previous
values. Similar principles are exploited in solid state devices called hardware random number
generators (HRNG), such as ComScire’s R2000KU, which produces 2 megabits per second
and is guaranteed, by the manufacturer, to pass any test for randomness. A renowned HRNG
is ERNIE (Electronic Random Number Indicator Equipment), first used in 1957 to draw
winning numbers in the Premium Bond investment scheme run by the Office of National
Savings in the UK. ERNIE has been through several incarnations since 1957, but continues
to be used for the draw.

2 In 1875, an engineer Joseph Jagger took advantage of a bias that he detected in one of six roulette

wheels at the Beaux-Arts Casino in Monte-Carlo to win over two million francs. A rather riskier strategy
employed by the adventurer Charles Wells, who is the most likely inspiration for the well known song “The
Man Who Broke the Bank at Monte-Carlo”, is the Martingale betting system (Exercise 2.59).
6 Statistics in Engineering, Second Edition

2.2.3 Pseudo random digits

Despite the availability of HRNG hardware, the sequence of 20 random digits produced by
sample() in R, and most other software, is not random but generated according to some
algorithm. To obtain the same sequence shown in Section 2.2.1, type

> set.seed(16)
> sample(0:9,20,replace=T)

The seed, here 16, is the initial value for the algorithm. If no seed is specified, some coding of
the computer clock time will be used. Once begun, the algorithm is completely deterministic
but the resulting sequence appears to be practically equivalent to a record of an experiment
in which each of the digits from 0 up to 9 was equally likely to occur at the next turn.
Such algorithms are called pseudo random number generators (PRNGs). A relatively
simple example of a PRNG is given in Exercise 2.7, and [Kroese et al., 2011] is a detailed
reference. John von Neumann, a renowned mathematician, quipped that

“Any one who considers arithmetical methods of producing random digits is, of
course in a state of sin”.

So, why do we use them? Reasons for using a PRNG are: they do not involve additional
hardware; hardware does not need to be checked for stability; and if a seed is specified, the
sequences are reproducible. Reproducible sequences can be used to document how a random
selection was made. Also, using a reproducible sequence in a computer simulation allows us
to investigate any anomalous results.
Donald E Knuth discusses why PRNGs work as well as they do in his multi-volume
monograph The Art of Computer Programming [Knuth, 1968]. As a consequence of its
construction, a PRNG must repeat after a given number of digits, known as its cycle length
(or period), but this number can be very large3 .

Example 2.2: Linear Congruential Generator [simple PRNG]

An early algorithm used to generate random numbers was the Linear Congruential
Generator (LCG), specified by the recursive formula using modular integer arithmetic4

Zi = a Zi−1 + c (mod m),

where m is called the modulus

a is called the multiplier
c is called the increment and
Z0 is the seed or starting value.
The variable Zi will be an integer between 0 and m − 1. To get a number U in the
range [0, 1), set Ui = Zi /m. This can be transformed to an integer between 0 and 9
(inclusive) by multiplying by 10 and removing the decimal fraction (using the floor
command).
Choosing the 4 parameters m, a, c and Z0 in a good way, provides sequences of numbers
Zi and Ui that appear for most purposes to be random.

3 The default in R and MATLAB is the Mersenne Twister [Matsumoto and Nishimura, 1998] with a

period length of 219 937 − 1.

4 The expression x = y (mod m) for non-negative integers x and y and positive integer m means that x

is the remainder of the integer division y/m, so that (y − x) = km for some non-negative integer k.
Probability and making decisions 7

We will assume we have some long sequence that is less than the cycle length. We consider
briefly how we might check that sequences from PRNGs appear random. The most basic
requirement is that the digits 0, . . . , 9 appear in the same proportions. Apart from this
requirement, there should be no easily recognizable patterns in the sequence of generated
numbers, such as there is in the following example.

Example 2.3: RANDU [limitations of PRNG]

An example of an LCG with poorly chosen parameters was RANDU, specified by

m = 231 , a = 216 + 3 = 65 539, c = 0, with Z0 odd, which was distributed in the 1960s,
and has bad statistical properties and a period of only 229 .

“... its very name RANDU is enough to bring dismay into the eyes and
stomachs of many computer scientists!”
–Donald Knuth

In fact, RANDU can be re-written as

Zi+2 = 6Zi+1 − 9Zi (mod 231 ),

which implies that its resultant sequences have a marked pattern that can be readily
visualized (see Exercise 2.9).

More demanding tests require the correct proportions of double digits, the correct propor-
tions of digits following a specific digit, and so on. A set of routines to evaluate randomness
called the Diehard tests5 were developed by George Marsaglia, an American mathemati-
cian and computer scientist, who established the lattice structure of linear congruential
generators in his paper memorably titled Random numbers fall mainly in the planes.
A good PRNG is far better than some ad hoc strategy based on long lists of num-
bers. A strategy such as using the first digits from long lists of numbers that range
over orders of magnitude, such as the magnitudes of 248 915 globally distributed earth-
quakes, photon fluxes for 1 452 bright objects identified by the Fermi space telescope
[Sambridge et al., 2011], and populations of cities, is not suitable. The reason is that, apart
from the lack of 0s, measurements with a lower first digit (1, 2, . . .) occur more frequently
than those with a higher first digit (. . .,8,9). This latter result is commonly known as Ben-
ford’s Law (see Exercise 2.61).

2.3 Defining probabilities

We discuss three approaches to defining probabilities: equally likely outcomes; relative fre-
quencies; and subjective assessments. But, no matter which approach is suitable for a par-
ticular application, the rules for manipulating probabilities are the same.

Definition 2.1: Sample space

A set of possible outcomes of an experiment, defined so that exactly one of them must
occur, is known as a sample space.

5 The Diehard tests and the TestU01 library provide standards for checking the output from PRNGs.
8 Statistics in Engineering, Second Edition

Definition 2.2: Mutually exclusive (disjoint)

Outcomes that comprise a sample space are said to be mutually exclusive because
no two of these outcomes can occur simultaneously. A commonly used synonym for
mutually exclusive is disjoint.

Definition 2.3: An event occurs

An event A is some subset of the sample space, and is said to occur if one of its
elements is the outcome of the experiment.

Definition 2.4: Collectively exhaustive

A set of events is collectively exhaustive if it covers the sample space6 . In particular,

the set of all possible outcomes is collectively exhaustive.

The sample space is not necessarily unique and we set up a sample space that will enable
us to answer the questions that we pose in a convenient manner.

2.3.1 Defining probabilities – Equally likely outcomes

In some experiments we can claim that all outcomes are equally likely (EL) because of
symmetry, such as that which a gambling apparatus is designed to possess. How can we
deduce the proportion of occasions on which a particular event A will occur in the long
run? In the case of EL outcomes, the deduced proportion is referred to as a probability and
is given by the following definition.

Definition 2.5: Probability for equally likely events

If all outcomes of an experiment are equally likely, then for any event A, the probability
of the event A occurring is defined as
number of EL outcomes in A
P(A) = .
total number of EL outcomes

This probability measure ranges between 0, when an event is impossible, and 1 when an
event is certain. It also follows that
total number of EL outcomes - number of EL outcomes in A
P(not A) =
total number of EL outcomes
= 1 − P(A) .

Definition 2.6: Complement

The event “not A” is called the complement of A, denoted by A.

6A more precise statement is that the union equals the sample space. See Section 2.3.
Probability and making decisions 9

Example 2.4: Decagon spinner [equally likely outcomes]

A regular decagon laminar with a pin through its center (Figure 2.1) is spun on
the pin. The laminar comes to rest with one edge resting on the table and the cor-
responding number is the outcome of the experiment. The sample space is the set
{0, 1, 2, 3, 4, 5, 6, 7, 8, 9}. If the device is carefully constructed so as to preserve the sym-
metry, and the initial torques differ, the outcomes can be considered as equally likely.
The event “spinning a 7” is the set {7} and so the probability of this event, P(7), is
1/10. The event “spinning an odd number” is the set {1, 3, 5, 7, 9}.
It follows that the probability of obtaining an odd number is

number of elements in {1, 3, 5, 7, 9} 5

P(odd number) = = .
number of elements in {1, . . . , 10} 10

8
9 7

0 6

1 5

2 4
3

(a) (b)

FIGURE 2.1: Decagon spinner (a) at rest (b) spinning.

Example 2.5: Two decagon spinners [equally likely outcomes]

Two decagons are spun, or equivalently the same decagon is spun twice. Either way, a
sample space of 100 equally likely outcomes is shown as a set of points in Figure 2.2.

(a) Find P(total of 3). There are four points that give a total of 3: (3, 0), (2, 1), (1, 2)
and (0, 3). Hence, P(total of 3) = 4/100.

Similarly we can count points in the sample space to obtain:

(b) P(total of 4) = 5/100.

(c) P(doubles) = 10/100.
10 Statistics in Engineering, Second Edition

Doubles

Total of 3

Total of 4

8
Second spin

0 1 2 3 4 5 6 7 8 9
First spin

FIGURE 2.2: Sample space for two spins of a decagon spinner.

Now consider P(total of 3 or doubles)7 .

(d) Calculate P(total of 3 or doubles). We can count points in the sample space to
obtain 14/100.

However, we can express it in terms of the probabilities of the constituent events

“total of 3” and “doubles”. If we look at the sample points corresponding to a
“total of 3 or doubles” we see that they are those corresponding to “total of 3”
together with those corresponding to “doubles”. Hence

P (total of 3) or (doubles) = P(total of 3) + P(doubles)
= 4/100 + 10/100 = 14/100.

7 Doubles is the same digit on both spins.

Exploring the Variety of Random
Documents with Different Content
motor-minds, our movie-nerves, our corner-light religion; where
“Safety First” shall give place to “Derring-Do” as a national motto;
where we shall ascend the empty peaks, and out of the thunder and
smoke of shaking Sinai bring down some daring commandment,
done by the finger of God on new tables of stone.
We are not lacking courage. It is imagination that we lack. We dare.
But we do not think it worth while. We are shallow, skeptical,
conventional, out of tune with the Infinite, and out of touch with
spiritual things. If we do not try the unattempted, it is because we
believe it has already been tried. It is because Homer has preëmpted
Helicon that we tunnel it. Only Milton, among us moderns (and how
ancient Milton seems!), only Milton in his blindness has seen that
there is room and verge enough on Helicon, and deeps within the
abyss of Hades where Dante would be lost. No, Milton is not the only
modern to leave the plains, and, like a star, to dwell apart. Thoreau
did it at Walden; Lanier did it on the marshes of Glynn; Burroughs
did it at Woodchuck Lodge; and Hudson did it on the plains of
Patagonia—proof enough that ponds and plains and the low-lying
marsh may be as high as Helicon for poetry, if only the poet have
the vision to see that
“Like to the greatness of God is the greatness within
The range of the marshes, the liberal marshes of Glynn.”
But here we fail. We no longer see the greatness of God in things.
We have covered God with an atom. We ask for bread, and Science
gives us a stone; for God, and Science gives us an electron. It was a
super-electron that created the heavens and the earth when it saw
that all of the other electrons were without form and void. Atomism
has taken the place of theism in our religion, if it is religion. Man is
only a bunch of willful atoms, or parts of atoms, not any longer the
crowning work of Creation, its center and circumference, its
dominion and destiny and glory, its divine expression, interpretation,
and immortal soul. Are we to be robbed of God? Inhibited forever
from faith by the lensed eyes of Science?
“What is man?” I ask, and Science laughs and answers, “Electrons.”
That is its latest guess. But does man look like them? Does he feel
like them? Does he behave like them? Does he believe like them? In
the laboratory he may. But out here in the hills of Hingham, where I
am returned to the earth and the sky and to my own soul, I know
that I AM, and that I still hold to all of those first things which
Science would shame me out of, offering me electrons instead!
I accept the electrons. Capering little deities, they are the sons of
God. But so are you and I the sons of God—and we are electrons,
trillions of electrons, if you like.
Gods and atoms, we can dwell and think and feel as either, the two
realms distinct and far apart, the roads between in a continual state
of construction, dangerous but passable. The anatomist, laying down
his scalpel, cries, “Such knowledge is too wonderful for me. I am
fearfully and wonderfully made!”—his science passing into poetry,
and from poetry to religion, but not easily in our present frame and
mood.
Science clears the sight and widens its range; but Science can never
clear up the shadows at the bottom of a woodchuck. Only vision can
do that, and Science lacks vision, using a microtome instead, paring
its woodchuck till he is thinner than sliced sunlight before it can see
through so much as a single stained cell of him. Science turns aside
from shadows, walking by sight or else standing still. It deals with
the flesh, not the spirit; and is as impotent in literature and art as in
life and society. The potent thing among men and nations is love.
Love never faileth. Yet never were we so afraid of love as we are to-
day; and never did art and literature seem so fearful of the
imagination, of vision, of the eternal, the divine.
“Go get me a bird,” the old scientist said to me; “I will give you a
lesson in skinning and mounting.” I was a young boy. Hurrying out to
the woods, I was soon back with a cuckoo. The face of the old
scientist darkened. “You should not have killed this bird, it is the
friend of man. See when I open this gizzard.” And with a dexterous
twist of his fingers turned inside out the gizzard, and showed it, like
a piece of plush, its fleshy walls penetrated with millions of
caterpillar hairs.
To this day I feel the wonder of that knowledge, and I thrill at the
meaning of that bird’s gizzard. Here was science and charity and
poetry and religion. What untold good to man! What greater
possible good to man? That was before I knew or understood the
cuckoo’s song. And neither the old scientist, nor yet his book,
“Sixteen Weeks in Zoölogy,” dealt with the song. Science is sure and
beautiful with a gizzard. Poetry is sure and beautiful with both
gizzard and song. And I wonder if the grinding gizzard or the singing
throat is the better part of the cuckoo, even in this world of worms?

“Though babbling only to the vale,

Of sunshine and of flowers,
Thou bringest unto me a tale
Of visionary hours.

“Thrice welcome, darling of the Spring!

Even yet thou art to me
No bird, but an invisible thing,
A voice, a mystery;

“And I can listen to thee yet;

Can lie upon the plain
And listen, till I do beget
That golden time again.

“O blessed Bird! the earth we pace

Again appears to be
An unsubstantial faëry place,
That is fit home for thee!”

I have a great book, published by the Government, devoted entirely

to birds’ gizzards, mills of the gods, and their grindings. It is not a
dull book, though the mills grind slowly and grind exceeding small. It
is a book of bones, of broken beetles, seeds, hairs, feathers, and
fragments. It is a great work of science. One might not like to lay it
down unfinished; but, having finished it, one could hardly say:

“And I can listen to thee yet;

Can lie upon the plain
And listen, till I do beget
That golden time again.

“O blessed Bird! the earth we Pace

Again appears to be
An unsubstantial faëry place,
That is fit home for thee!”

Nature will not do, nor all the truth of nature, for stuff of song and
story. As life is more than meat, so is literature more than life.
Nature conforms to art; and in fiction “the only real people are those
who never existed.”
At Good-Will Farm, Maine, there is a rock marked with a copper
plate. It had been marked for drill and dynamite, until, one day, my
car swung up and over a sharp turn in the road before the
schoolhouse, skidding rather horribly on the smooth outcropping
ledge which had been uncovered and left as part of the roadbed.
“You ought to blast that thing out,” I said, somewhat testily, to the
supervisor who came out to greet me, my nerves, strung a bit too
tight on the long day’s drive, snapping with the skid here at the very
end of the trip.
“I’ll do it,” he replied, apologetically. “I had intended to do it from
the first.”
The next day we were climbing this road on foot, and, standing on
the ledge to take in the wide landscape of the Kennebec below us, I
chanced to look down at my feet and saw, cut deep in the smooth
surface of the stone, several parallel lines.
“Don’t blast out this rock!” I exclaimed. “Tear down your
schoolhouse rather. Build a new road through the grounds, but leave
this stone. This is part of a great book.”
“I don’t understand,” said the supervisor.
“Here is written a page of the greatest story ever penned. These
lines were done by the hand of the glacier who came this way in the
Ice Age. Don’t blot it out. Put a fence about it, and a copper plate
upon it, translating the story so that your students can read it and
understand.”
He did. There was no need of the fence; but he set the plate into
the rock, telling of the Ice Age, how the glacier came down,
ploughing out the valley of the Kennebec, rounding and smoothing
this ledge, and inditing this manuscript for Good-Will Farm School
ages later.
So much does the mere scratch of science enhance the virtue of a
stone! Now add to your science history. Instead of the scratch of a
glacier, let it be a chisel and a human hand, and let the marks be
—“1620.” Now read—if you can read and understand.
I copy it verbatim from a Freshman college theme:

Plymouth Rock
Plymouth Rock is situated in Plymouth Mass. It is the rock upon
which the Mayflower landed in 1620. But it is not now where it
was then. It was moved many years ago up to the street. And
when they moved it it broke. But they cemented it together. It is
four or five feet long; and three or four feet wide; and it is
inscribed with the famous figures 1620, to celebrate the landing
of the Puritans at that time. It is enclosed within a canopy of
stone and an iron fence; but the gate is hardly ever closed.
There are a great many famous stones in the world but this is
as famous as any.
My mother was visiting me. She is a self-contained old Quaker, and
this was the second time in all of her eighty years that she had even
seen New England! What should we do first? What did she most
desire to see? “Take me to see Plymouth Rock first,” she said; and
we were off, mother as excited and as lively as a girl. As we entered
Plymouth, however, I noticed that mother had grown silent, and that
her doctor-daughter, beside her on the back seat, always sensitive to
her moods, was also silent. We descended the hill to the harbor,
came on in sight of the canopy over the Rock, and slowed down to
stop. But the car had not stopped, when mother, the back door
open, her foot on the running-board, was stepping off and through
the open gate, where, falling on her knees, with tears running down
her face, she kissed the blessed stone, her daughter calling, “O
Mother, the germs! the germs!”
When Science and Religion thus clash, Science must give way.
Mother knew as much about germs as her doctor-daughter. She had
lived longer than her daughter; she had lost more, and had loved
more—some things more than life itself.
Science has marked every rock; but only those that are wet with
such tears and kissed with such lips are ripe for sermon and song.
These are the eyes and these the lips of those, who, passing
through Bacca, make it a well. Knowledge alone, though it course
the very heavens, will come back to earth without so much as one
shining fleck of stardust in its hair.
The other day a great astronomer was delivering a lecture in Boston
on the stars. Wonder and awe held the audience as it traveled the
stellar spaces with the help of the astounding pictures on the screen.
The emotion was deep; the tension almost painful as the lecturer
swept on and on through the unthinkable vast, when, coming to his
close, he turned and asked lightly, “Now, what do you think of
immortality? Is it anything more than the neurotic hope of a very
insignificant mote in this immensity?”
The effect was terrific. The scientific smiled. The simple left the hall
dazed and stunned. They lost all sense of time and space, they lost
sight of the very stars in this swift, far fall. They had been carried up
through the seven spheres to the very gate of heaven, then hurled
to earth. The lecture failed—not of instruction, not of emotion, but
of will, leaving the listeners powerless and undone. The lecturer may
be right—for astronomy; and yet be quite wrong, for poetry. He may
have uttered the last word—for science; but this end is only the
beginning for religion.
How much greater an astronomer this college professor than that
shepherd psalmist on the far-off Syrian hills! Ranging the same astral
field as our scientist, sweeping the same stellar spaces, with only a
shepherd’s knowledge, the psalmist’s thought takes the same turn as
the scientist’s, down to man, but on different wings,—the wings of
poetry:
“When I consider thy heavens,
The work of thy fingers,
The moon and the stars
Which thou hast ordained;
What is man, that thou art mindful of him?
And the son of man, that thou visitest him?”
Then, swinging upward on those mighty wings, past the reach of
science, out of the range of knowledge, up, up to the divinest height
ever touched by human thought, the psalmist-astronomer cries
impiously, exultantly,
“For thou hast made him but little lower than God,
And crownest him with glory and honor!”
This starts where the astronomer stopped. This is religion and
literature. And I have these very stars over my hilltop here in
Hingham!
 CHAPTER IV
THE DUTY TO DIG
 CHAPTER IV
THE DUTY TO DIG

I
A young man sat by the roadside, milking. And as he milked, one
drove up in her limousine and stopped and said unto him:
“Young man, why are you not at the front?”
The young man milked on, for that was the thing to do. Then, with
still more slackers in her voice, the woman said a second time unto
him:
“Young man, why are you not at the front?”
“Because, ma’am, the milk is at this end,” he answered.
And the chauffeur, throwing the clutch of the limousine into third
speed ahead, drove off, thinking.
But the young man milking had already thought. To milk is to think.
If “darning is premeditated poverty,” then there is no saner
occupation for human hands, none more thought-inducing, unless it
be milking. Anyhow, when the Great War came on, I went over to a
neighbor’s and bought a cow; I made me a new milking-stool with
spread sturdy legs; and I sat down to face the situation calmly,
where I might see it steadily and whole. I had tried the professorial
chair; I had tried the editorial chair; I had even tried that Siege
Perilous, the high-backed, soft-seated chair of plush behind the
pulpit. I may never preach again; but if I do, it will be on condition
that I sit on a three-legged milking stool instead of on that
upholstered pillowy throne of plush.
Whence cometh wisdom? and where is the place of understanding?
The flaming flambeaux on the Public Library say, “The light is in
here”; the Φ B K key in the middle of the professorial waistcoat says,
“It is in here.” But I say, let the flambeaux be replaced by round-
headed stocking-darners, as the sign of premeditated poverty; and
the dangling Key by a miniature milking-stool, as the symbol of the
wisdom that knows which end of a cow to milk.
Not one of those students in the University who earned Φ B K last
year knew how to milk, and only a few, I believe, of their professors.
One of these, with a Ph.D. from Germany, whose key had charmed
his students across their whole college course, asked me what breed
of cattle heifers were. Might not his teaching have been quite as
practical, had there dangled from his watch-chain those four years,
not this key to the catacombs of knowledge, but a little jeweled
milking-stool?
I too might wear a key, especially as I came innocently by mine,
having had one thrust upon me; still, as I was born on a farm, and
grew up in the fields, and am likely to end my days as I have lived
them, here in the woods, this Φ B K key does not fit the lock to the
door of knowledge that opens widest to me.
I have read a little on the aorist tense, and on the Ygdrasyl tree; a
little, I say, on many things, from the animal aardvark, here and
there, to zythum, a soft drink of the ancient Egyptians, picking a few
rusty locks with this skeleton key; but the doors that open wide at
my approach are those to my house, my barn, and the
unwithholding fields. I know the road home, clear to the end; I know
profoundly to come in when it rains; and I move with absolute
certainty to the right end of the cow when it is time to milk.
I am aware of a certain arrogance in this, a show of pride, and that
unbottomed pomp of those who wear the Φ B K key dangling at
their vests,—as if I could milk any cow! or might have in my barn
the world’s champion cow! I have only a grade Jersey in my barn;
and as for milking heifers with their first calves—I have milked them.
But breaking in a heifer is really a young man’s job.
So I find myself at the middle of my years, stripped of outward
signs, as I hope I am inwardly purged, of all vain shows of wisdom
(quite too humble, truly!), falling in as unnaturally as the birds with
the fool daylight-saving plan, the ways of the sun, who knoweth his
going down, being quite good enough for me.

II
But how far run the ways of nature from the devious ways of men!
The ways of Mullein Hill from the ways of a military camp! The Great
War came and passed and left the earth a vast human grave. But
through it all seedtime and harvest came to Mullein Hill, leaving only
more and more abundant life. The Great War is an illustration on the
grandest scale of what man, departing from the simple ways of
nature, will do to man. War is the logic of our present way of living.
I am not concerned with war in this book, but with the sources of
life and literature. I have a cure for war, however, here on Mullein
Hill; and this cure is the very elixir of life and literature.
War could have destroyed, but it did not change, my going or
coming here in the hills. My garden went on as it had for years gone
on. There was a little more of it, for there was more need in the
village; there was a little larger yield of its reasonableness and joy
and beans. But I did not plough up my front lawn for potatoes. Years
before I had provided myself with a back yard, and got it into tilth
for potatoes, keeping the front lawn green for the cow.
Though she is only a grade Jersey, the cow is a pretty creature, and
gives a rural, ruminant touch to our approach, along with the lilacs
and the hens. Hitched here on the front lawn the cow suggests
economy, too. She is more than a wagon hitched to a star. She is a
mowing-machine, and a rake and tedder, and a churn. The gods
with her do my mowing, gather up and cure my hay, and turn it into
cream.
Every cow gives some skim milk—which we need for the chickens,
for cooking, and cottage cheese. Life is not all cream. If I speak of
gods doing my chores, I will say they do not milk for me in the
mornings, and that it is one of the boys who milks at night. A cow
clipping your lawn is poetry and cream too, but caring for the
creature is often skim milk and prose. Milking ought to be done
regularly. Get a cow and you find her cud a kind of pendulum to all
creation, the time to milk being synchronized twice daily to the stars.
I did not plant war-potatoes on my front lawn, partly because they
would not grow there, and partly because, in times of peace, I had
prepared for war-potatoes; and partly because I think a front lawn
looks better in cows than in potatoes. If thou shouldst
“So live, that when thy summons comes to join
The innumerable caravan ...
Thou go not, like the quarry-slave at night,
Scourged to his dungeon,”
—should we not all live so that, when war comes, we need not
plough up the beds for potatoes where the portulaca and poppy
ought to blow?
But what a confession here! When war comes! As if I expected war
to come again! Many of us are fighting feebly against both the
thought and the hideous thing. But many more are preparing for it.
Our generals of the late war are going up and down the land
preaching preparedness, as they always have. We learn nothing.
They know everything. Their profession is war. Can a man lay down
his life for a profession he does not believe in? The military men
believe in war.
But so do we all as a people. War is the oldest, most honored
profession in the world. In all of the fifty years of its history, the
great University, of which, for almost half of that time, I have been a
teacher, had never conferred an honorary degree. For fifty years it
had carried a single laurel wreath in its hand to crown some honored
head. Prophets came and went; poets came and went; scientists
came and went; scholars came and went; but still the University,
dedicated to life and learning, waited with its single wreath a yet
more honorable brow.
Then came Foch, the professional soldier. Study for nearly ten
thousand students was suspended; a holiday was proclaimed; a
great assembly was called; and here, with speech and song and
academic garb, with national colors mingling and waving palms, the
laurel wreath was placed upon the soldier’s brow. And this
University, founded in the name of the Prince of Peace, dedicated to
Christian life and learning, crowned the profession of arms as it can
crown no other profession, and gave its highest sanction to bloody
war.
I do not fail of gratitude to Foch. I only stand in horror that he is,
and had to be. I would have had the Nation at the pier to meet him,
but clothed in sackcloth, and every citizen with ashes on his head. I,
too, would have suspended study and work everywhere for an hour;
and, stretching crape across the Arctic Circle till the sign of mourning
hid Matamoras and Cape Sable, I would have called an assembly of
the continent, and begged this half of the hemisphere to cry, “O
God, the foolishness and futility of war!”
The Duty to Dig is older than the practice of war. It was designed to
prevent war; and to-day it is the only biological remedy certain to
cure war. I must not stop here to explain its therapeutics as applied
to war, for I am dealing with another theme. But just before the war
broke upon the world, I wrote an editorial for a paper I was serving,
advocating that a hive of my bees be sent to the German Emperor
and one also to the war lord of Austria. I had extra hives for the
British Prime Minister, the Tiger of France, Mr. William Randolph
Hearst, and Mr. Theodore Roosevelt. If I could have interested these
gentlemen, and a few others, in bee-keeping, I could save the world
from war.
The editorial and the offer of bees were both rejected by the careful
editor. “I must stay strictly neutral,” was his timid excuse. At the very
time I was writing, the Reverend Price Collier, a former Hingham
preacher, was publishing a book called “Germany and the Germans,”
in which he set forth the old theory of military preparedness as a
preventive of war. Speaking of the German army (this was in 1913),
he says:

It is the best all-round democratic university in the world; it is a

necessary antidote for the physical lethargy of the German race;
it is essential to discipline; it is a cement for holding Germany
together; it gives a much-worried and many-times-beaten
people confidence; the poverty of the great bulk of its officers
keeps the level of social expenditure on a sensible scale; it
offers a brilliant example, in a material age, of men scorning
ease for the service of their country; it keeps the peace in
Europe; and until there is a second coming of a Christ of pity
and patience and peace, it is as good a substitute for that far-
off divine event as puzzled man has to offer.

It is a minister of the Gospel who makes this profound observation.

But it sounds like our present Secretary of War, a banker, and like
our present Commanding General, a professional soldier. Here is sure
proof that the human race cannot learn the essential things, and so,
is doomed.
But I wish I might try my bees. This old hoax of preventing war by
preparing men to fight has been so often tried! And we are at it
again. But no nation has tried my simple and inexpensive substitute
of bees and plough-shares, and pruning hooks. They have tried,
from time out of memory, to beat their swords and spears into
garden tools, but a sword makes a mighty poor ploughshare. It has
never been successfully done. The manufacturing process is wrong.
It takes the temper out of good garden steel first to heat it in the
fires of war. We must reverse the process: turn the virgin metal into
garden steel first, and give every man a hoe, and a ploughshare,
and a pruning hook; then he will never have need for sword and
spear.

III
Instead of universal military training, I would advocate a hive of bees
for everybody, or a backyard garden. A house should have both lawn
and garden, even though the gooseberries crowd the house out to
the roadside, where the human house instinctively edges to see the
neighbors in their new pony-coats go by. Let my front door stand
open; while over the back stoop the old-fashioned roses and the
grape-vines draw a screen.
But give me a house with a yard rather than a hole in the wall, a city
tenement, or a flat. The whole trend of society is toward the city, or
camp, military, rather than agricultural. The modern city is a social
camp. Life is becoming a series of mass movements, military
maneuvers, at the command of social leaders. Industry has long
been militarized both in form and spirit, and is rapidly perfecting its
organization. Today, 1923, there are more than a hundred different
automobile manufacturing concerns in this country. Only those
capable of “quantity production” will survive. Already the
manufacturers see the entire industry reduced to five different
concerns. This is strictly military, the making of society into a vast,
and vaster machine, which, too great at last for control, will turn
upon and crush its makers.
We are all conscripted. The Draft Board of industrialism allows no
exemptions. The only way I see is to desert, to take to the woods,
as I have done, to return individually to a simple, elemental manner
of life out of the soil. But who can pay the social cost? Our social or
camp psychology is better understood and more easily handled than
the mind of the lone scout within us. We are gregarious by nature;
we hunt in packs, we collect into crowds. Yet we are selves,
separate, single, each of us a cave-man as well as cliff-dweller, a
Remus as well as a Romulus. The city-building brother killed his
country brother. And the murder still goes on.
Out of a commentary on the Bible I take the following observation,
partly for its charm, but also because it holds a profound truth:
 God’s people from the earliest time had never been builders of
cities. The earliest account of city-building is that of the city of
Enoch by Cain, and all the subsequent mention of city-building
is in connection with the apostate families of the earth, such as
Nimrod and his descendants, the Canaanites and the Egyptians.
Sodom is one of the earliest mentioned cities properly so called,
and the story of it is not encouraging for the people of God.

But which is the city whose story is encouraging to the people of

God? Not Boston’s, nor New York’s, nor London’s, nor Vienna’s.
Vienna is starving; the country is bankrupt Austria’s governmental
machine is a total wreck; but the peasant goes his way, suffering
little inconvenience, though the crown is not to-day worth the paper
it is printed on. The peasant lives on the land, not on the bank; he
gets his simple life directly out of the soil instead of a pay envelope;
he has no New York, New Haven and Hartford stock, worth one
hundred and eighty-six dollars yesterday, and ten dollars to-day, to-
morrow, and until he starves. He has a piece of land and, impossible
as it sounds on paper, lives on it, and out of it, and in it, an almost
independent life, as the wage-slave and the coupon-victim cannot
live.
We shall face a famine, so long as our door-yards are all lawn in
front and all garbage-can behind. We have farmers enough—one to
every eight of our population, I believe—who might produce
sufficient raw potatoes; but Aroostook County is barely contiguous to
the United States, and such a barrage of frost was laid down across
its borders this last winter that, if one brought potatoes out of
Aroostook between December and March, he had to bear them in his
bosom.
Aroostook County is the greatest potato-patch in the world; the
American imagination loves to hover over the tubered tracts of
Aroostook, the richest county in the world; loves to feel that the
world could be fed from Aroostook, were it not for the triple alliance
of the cold and the contiguity and a railroad that runs, if not like a
broken tooth, then like a foot out of joint, into these remote
dreamlands of Maine.
Woe to them that go down to the railroads for help; and stay on
engines and trust in empties, because they are many; and in
officials, because they are very strong. Now the officials are men
and not God, and their engines steel and not spirit. Why should a
rational, spiritual human society trust its well-being to such paltry
powers, when all the forces of nature are at its command?
I will put more trust in an acre of land than in a Continental
Congress. I had rather have a hoe at my right hand than an army of
bank presidents. Give me the rising and the setting sun, the four
seasons, and the peasant’s portion; and you may have the portion of
the president.
I said we have farmers enough to raise all we need. We have more
than enough. We have more than enough bankers; more than
enough automobile-makers; more than enough store-keepers; more
than enough coal-miners; more than enough cooks and janitors. But
we have nowhere near enough landowners and peasants. Nothing in
the world would so straighten out society as to declare next year a
Year of Jubilee, and give every man, not a job, but his birthright, a
piece of land.
We are over-organized and almost de-individualized. But the time
must again come when every man shall dig and every woman spin,
and every family build its own automobile, distill its own petrol, and
work out its taxes on the road. We shall always hold to the social
principle of the division of labor—I plough for you; and you shoe my
horse for me. But we have carried the principle, in our over-
organization, to the point where a man’s whole part in the world’s
work consists in putting on the left hind wheel of endless
automobiles.
An eight-hour day will not save that man. And he is typical of all
men to-day. Only by his acceptance of the duty to dig can he be
saved, and society with him. The principle of the division of labor
has been misapplied: instead of specialization and the narrowing of
each man’s portion, it should be applied broadly, multiplying his
labors. Work is creative; it is self-expression; and I should let no
man do for me what I can do. He robs me of living who robs me of
doing.
The theory of present-day society—specialization, organization,
combination, quantity production—is a fatal application of a perfectly
sound principle. Six automobile combinations which in a year can
destroy a hundred lesser combinations, can in another year destroy
all but one of each other; and that remaining one, having nothing
now to destroy, must turn and destroy itself.

IV
But I am concerned with life and literature. How does the
organization of society affect books, admitting that it affects life?
What is a book but a life?—and a more abundant life? Everybody
who has lived has a book to write. But only those who have lived
abundantly should write their books. Starve a nation spiritually, as
ours is being starved; reduce its life to a mechanical routine; rob its
labor of all creative quality, and how shall it write?
The duty to dig applies first to the spirit. There is sound economy in
it if one’s political economy is sound. I do not say money. Economy
and money are not equivalent terms. I digged as a duty last year;
and as a result I did not buy a potato from the “combine” in Maine; I
nearly got on without a pound of beef from Chicago; and could have
made my honey serve for Cuban sugar. The same amount of time
spent putting left hind wheels on automobiles might have brought
me more money, and so, more beef and potatoes and sugar, and so,
more gout and rheumatism. The duty to dig comprehends a great
deal more than ordinary economy.
I would not imply that I can handle the Beef Trust and the potato
pirates and the sugar barons with my humble hoe; or snap my
fingers in the face of Standard Oil, and say, “Go to, I’ll have none of
your twenty-eight-cent gas!” I do say that several million bee-
keepers and potato-patchers, and hen-coopers, keeping busy in their
back yards, as I keep busy in mine, could mightily relieve the
railroad congestion, and save gasoline, and cut in on the demand for
Chicago beef and cold-storage eggs, and generally lower the high
cost of living.
It is not because there are “millions in it” that I would have the
banker plant his back yard to beans. Thoreau planted two acres and
a half to beans and potatoes (on a weak market, however), with a
“pecuniary profit of $8.71½.” Here is no very great financial
inducement to a busy banker, or to a ward boss. Still, who better
than the ward boss or the banker could afford a private beanfield?
I say it is not for the sake of this $8.71½ profit of Thoreau’s that we
must dig; but rather for that chapter on the bean field in Walden
Pond, which proves the real worth of digging.
There can hardly be a form of labor so elemental, so all-demanding,
so abundantly yielding of the fruits of life, as digging. Yet there are
those who doubt the wisdom of digging because things can be
bought cheaper at the store; and those who question their right to
dig when they can hire a man to dig for them; and there are those
who hate to dig, who contemn duty, who, if they plant, will plant a
piece of fallow land with golf-balls only, and hoe it with brassies,
niblicks, cleeks, and spooners, saying with Chaucer’s Monk:
“... how shall the world be servèd?
Let Austyn have his swynk to his reservèd.”
Golf is an ancient game, no doubt, but not so old as gardening,
though golf’s primordial club and vocabulary seem like things long
left over, bits of that Missing-Link Period between our arboreal and
cave-day past. Except for calling the cows from the meadow, or
fighting in war, there is nothing we do that requires words and
weapons, tools, instruments, implements, utensils, apparatus,
machinery, or mechanisms so lacking in character and comeliness as
the words and clubs of golf. The gurglings of infants seem articulate,
even to unparental ears, compared with the jargon of golf; and as
for billiard-cues, baseball-bats, pikes, spades, shillalahs, and
teething-rings, they have the touch of poetry on them; whereas the
golf-club was conceived and shaped in utter unimaginativeness.
Golf is not an ancient game: it has the mark of the Machine upon it;
the Preadamites could not have figured the game out. Gardening, on
the other hand, if we trust Holy Writ, was an institution founded
before the Fall, incorporated with the social order from the start—an
inherent, essential element in the constitution of human things:
“Great nature’s primal course,
Chief nourisher in life’s feast,”
—which civilization doth murder as Macbeth murdered sleep.
Golf belongs to civilization strictly, not to the human race, being one
of life’s post-Edenic precautions, like psychopathic hospitals, jails,
and homes for the feeble-minded. A golf course is a little-wanderers’
home; and if we must have golf courses, let their hazards be
carefully constructed on worthless land, and let the Civil Service
Board examine the caddies, whether they be fit guards for the
golfers, lest some small boy be wasted who might have tended real
sheep on Norfolk Downs or have weeded in a garden.
It is a duty to dig, to nail the Stars and Stripes to a lima-bean pole,
and plant the banner square in the middle of the garden. Profits?
pleasures? Both sorts will grow, especially the pleasures, which really
are part of the profits, till they fairly smother the weeds; not the
least of these being your sense of living and your right to live, which
comes out of actually hoeing your own row—a literal row of beans or
corn or tomatoes.
Somebody must feed the soldiers; but nobody must needs feed me.
It is not necessary that I live, however necessary I find it to eat;
eating, like sleeping and breathing and keeping warm, being strictly
a private enterprise that nobody but I need see as necessary or be
responsible for.
The soldier must carry a shovel nowadays, but he will require a hoe,
too, and a pruning hook, and a ploughshare, before he will be self-
supporting. With such a kit war could support war forever, which is
the Rathenau plan of war, with everything German left out,
consequently everything of war left out. The soldier cannot feed
himself. The crew of a battleship cannot be expected to catch their
own cod and flounders. They must leave that to the trawlers, those
human boats, with human crews who fish for a living. Men of the
navy must die for a living. The captain of a United States destroyer,
writing to his wife, says, “I think that the only real anxiety is lest we
may not get into the big game at all. I do not think that any of us
are bloodthirsty or desirous of glory or advancement, but we have to
justify our existence.”
So does every human being; yet an existence that can be justified
only by fighting and dying is too unproductive, too far from self-
supporting, to warrant the sure calling and election of many of us.
No Grand-Banker ever wrote so to his wife, though he might be
returning with all his salt unwet; no college professor ever wrote so
—not if he could get into his garden—in spite of his pupils, his
college president, the trustees, and Mr. Carnegie’s Efficiency Board.
Teaching may not justify a professor’s existence, though it ought to
justify his salary; so, every time I start for the University, I put a
dozen or two of eggs into my book-bag, that I may have a right to
the tree of life, and may enter in through the gates into the city.
I am not independent of society. I do not wish to be independent. I
wish to be debtor to all and have all debtors to me. But we buy too
much and sell too much of life, and raise too little. We pay for all we
get. Sometimes we get all we pay for, but not often; and if we never
did, still life has so thoroughly adopted the business standard, that
we had rather keep on paying than trying to grow our way.
Business is a way of living by proxy; money is society’s proxy for
every sort of implement and tool. To produce something, however—
some actual wealth, a pennyweight of gold-dust, a pound of honey,
a dozen eggs, a book, a boy, a bunch of beets; some real wealth out
of the soil, out of my loins, out of my brains, out of my muscles and
the sap of the maple, the rains and sunshine and the soil, out of the
rich veins of the earth or the swarming waters of the sea—this is to
be; and to be myself, and not a proxy, is to lose my life and save it,
and to justify my existence.
I have to buy a multitude of things—transportation, coal, dentistry,
news, flour, and clothes. I have paid in money for them. I have also
paid in real wealth, having given, to balance my charge on society,
an equivalent in raw cabbage, pure honey, fresh eggs, and the like,
from my own created store. I am doubtless in debt to society, but I
have tried to give wealth for wealth, not the symbol of it merely; and
last year, as I balanced my books, I think the world was in debt to
me by several bunches of beets. I do not boast of the beets, though
they take me out of the debtor’s prison where most of us live. I can
face the world, however, with those beets; I have gone over the top,
have done my bit, with beets.
The oldest duty on the human conscience is the duty to dig. I am a
college teacher, and that is an honorable, if futile, profession. The
Scriptures say: “He gave some, apostles; and some, prophets; and
some, evangelists; and some, pastors and teachers”; but, before
there were any such multifarious and highly specialized needs, it was
said unto our first father: “Replenish the earth and subdue it”—a
universal human need, a call to duty, from which no Draft Board of
civilization can rightly exempt us.
Wealth is not created, not even increased, in trade. When was one
pennyweight of gold on ’change by any magic metallurgy of trade
made two pennyweight? The magic of the second pennyweight is
the metallurgy of the pick and shovel and cradle rocking the shining
sands of the Yukon. Real wealth is only circulated in trade. It comes
from primal sources—from the gold-fields, the cotton-fields, the
corn-fields, the fir-clad sides of Katahdin, the wide gray waters of
the Grand Banks, the high valleys of the sheeped Sierras, and from
back yards, like mine, that bring forth thirty- and sixty- and an
hundred-fold.
And this is as profoundly true of life and literature as it is of cotton
and lumber and gold.
Give me a garden and the wages of hoeing my row. And if not a
garden, then a little house of hens, a coop of pigeons, a colony of
bees—even in the city I should keep bees, if I had to keep them in
the attic or on the roof. Not every one can have a garden, but every
one can either plant a tree, or raise one pig, or keep a cow or goat,
or feed a few hens, or raise a flock of pigeons, or do something that
will bring him personally into contact with real things, and make it
possible for him to help pay his way with real wealth, and in part, at
least, to justify his existence, and his book.
 CHAPTER V
THE MAN AND THE BOOK
 CHAPTER V
THE MAN AND THE BOOK

Here on my desk lies a new book entitled, “For the Benefit of My

Creditors,” the autobiography of Hinckley Gilbert Mitchell, a scholar, a
teacher in a school of theology—and now this book, a simple, sad
book of human struggle and defeat, of spiritual and scientific
adventure and triumph and romance.
The scholar is not the accepted stuff of literature. What of human
interest can come out of a classroom? Yet I have seen this scholar’s
classroom when it was wilder than ten nights in a barroom crowded
into one. I have seen some lively and human times in my own
classroom; and I know that there is as real a chance, and as magical
a chance, there as Dana found on the high seas. There are frontiers
for the scholar, especially in theology, as dangerous in their crossing
as any to be met with by the overland pioneer.
Dana escaped from the decorous and the conventional life of social
Boston by way of the deep sea; Mitchell escaped from the decorous
and conventional dogma of his church by way of honest study; and
his Church tried him for heresy, and found him guilty, and would
have burned him at the stake had that been the decorous and
conventional manner of dealing with heretics at the moment. As it
was, they only branded him, and cast him out as a thing unclean.
Perhaps that is not a life human enough, and abundant enough, for
a book. It is the simple story of a poor boy picking stones and
building walls on his father’s farm in New York State; then, as
Director of the American School of Oriental Research in Jerusalem,
rebuilding “The Wall of Nehemiah”; then, as scholar and professor,
re-creating “The World before Abraham”; and finally, as the storm
center of one of the bitterest theological controversies of recent
years, dismissed, dishonored, betrayed for less than thirty pieces of
silver, a silent, brokenhearted man. It is only another version of an
old and very common story. Prophets and pioneers are all alike; and
their stories are much alike, whether the pages turn westward,
where new empires take their way, or eastward, back along the
scholar’s crossed and tangled trails to a world before Abraham.
As the manuscript of the book lay upon my table, I wondered if any
publisher would feel the human pathos of the struggle, and the
mighty meaning of it all for truth. Who would publish it? But here it
is, printed and bound, a book—“For the Benefit of My Creditors,” as
if he were debtor to all, his enemies included, and owed them only
love.
This is as modest and self-withholding a story as a man ever told of
himself. There are all too few of such human stories. This one would
never have been told had the author not hated intellectual cowardice
as he hated moral cowardice, with a perfect hatred. He sought the
truth—in the Bible, and in his own mind. The geologist seeks some
of the same truth in the rocks; the astronomer in the stars. The Old
Testament was this scholar’s field. And, laying aside tradition and the
spirit of dogma, he sought as a scientist seeks, patiently, fearlessly,
reverently, for what his long and thorough preparation made him
eminently able to find.
This is the highest type of courage and daring. Who finds truth finds
trial and adventure. In his condemnation by the bishops of his
Church, he felt that truth had been assailed and the scientific
method. He did not write this book to defend the truth, nor to
defend himself; but to examine himself, as he would examine a
difficult fragment of Hebrew manuscript, and make himself easy for
other men to read.
His trial was long past, and most of his life had been lived, before a
page of his book was written. He came at it reluctantly: he might
seem personal—petty or selfish or egotistical; or he might say
something bitter and vindictive and do harm to the Church. But
neither himself nor his Church must stand in the way of truth; and in
his trial, truth had been tried, and the only way of knowing truth had
been condemned. So he sits down to write this story of his life
exactly as he sat down to write a commentary on the Book of
Genesis—to account for his being as a man and a scholar, his
preparation, his methods of study, his attitude, and approach.
How much truth has he discovered? He makes no claims. Darwin
may or may not have the truth about Evolution; but we have a
certain and a great truth in Darwin—in his mind and method. It was
how Darwin tried to solve the problem of life and its forms, rather
than the solution, that has changed the thinking of the world.
For three years I was a student of Hebrew and Old Testament
Exegesis under this scholar. I have forgotten all he taught me, and
more. But the way he taught me has changed forever my outlook
upon life. His attitude was truth, and it flooded not only the whole
mind, but one’s whole being, with light. Many a time I have sat in his
classroom during the discussion of some highly difficult and
dangerous question of doctrine, and said to myself, amid the drawn
daggers of those who had come to trap him, “Right or wrong his
findings, he is himself truth, its life and way.”
Life enough for a book? He could have written a book on teaching.
For he loved to teach! He loved to teach young preachers. He could
not preach; but he was the teacher born. The classroom was his
from the foundation of the world. Here he was preaching truly—from
a thousand future pulpits at the very ends of the earth. He saw his
students scattered over the whole world preaching to the
intelligences of men as well as to their hearts; revealing the wisdom
as well as the love of God; and expounding a diviner Bible because it
was a wholly human Bible. In all of these pulpits he heard himself
speaking with tongues not his own, but the message was his own,
the simple sincere faith of his classroom.
The thought of it thrilled him. It lifted him up. He dwelt in the
presence of the opportunity as in the very presence of the Most
High. As humble a man as ever lived, doubting his every power and
gift, and relying only on the truth to make him free, he would come
into the classroom and take his chair on the six-inch platform, which
raised him by so much above his students, as if that platform were
the Mount of Transfiguration. His face would shine; his voice, his
gestures, his attitude working with his careful words, made his
whole being radiant with zeal for the truth and love for us, his
students, so mysteriously given to his care.
Then suddenly, after more than twenty years of this, he was
expelled—driven from this sacred classroom and branded as
unsound, unsafe, unfit!
No, not suddenly. It was only the verdict of his judges that came
suddenly. No one nowadays could prepare his mind for a judgment
like that. For five or six of the years, during which the trouble-
makers, under pretense of study, had elected his courses at the
Theological School, I had either been a student under him or his
close and sympathetic friend. I knew, as he knew, that his enemies
would stop at nothing in their bitter zeal; still, I remember vividly the
utter shock and astonishment of the bishops’ decision. And I
remember—for I cannot forget—its strange numbing effect upon
him. It came over him slowly, else I think he might have died. It
crept upon him like a dreadful palsy, leaving him dazed and dumb.
He was too simple a man to realize it quickly, too entirely single in
mind and heart to realize it wholly. It slowly crushed him to the
earth. And never in all the after years was he whole again. His heart
was broken. He rose up and taught, until the very hour of his death,
but never again in his old classroom nor with his old spirit. Day after
day he would pass by the Theological School with its hundreds of
eager students; he would see them gathering at the hour of his
lecture; but another teacher (one whom he had trained) would come
in and take his place, while he plodded down the street and on, a
shepherd without his sheep.
Meantime he was called to teach in another graduate school. He
welcomed this new work. He found honor, and love, and fellowship
among his new colleagues. They gave him freedom. They created a
place for him that had not been before. He could teach what he
wished and as he wished. It was enough for them to have him
among them, and many a time he told me of how unworthy he felt
of all this love and honor in his declining years, and how it had
stayed and steadied him in his deep defeat. But they did not need
him here—so he felt. It was more for the honor of scholarship than
for the good he would do them. But he felt that they did need him at
his own beloved school, whose policies he had helped to shape,
whose spirit he had helped to create, whose name and fame he had
so largely helped to establish, and whose students, crowding in from
the east and from the great west, he longed to take into his heart
and his home, as for so many happy years he had been in the habit
of doing.
“O Jerusalem, Jerusalem,” he would cry as he passed by on the
street, a stranger, and saw the students going in and out, “O
Jerusalem, Jerusalem, which killeth the prophets, ... how often
would I have gathered thy children together, even as a hen
gathereth her chickens under her wings, and ye would not!”
This, however, was not the doing of the school. Faculty and
students, with the exception of those few who came for the express
purpose of accusing him, were loyal. The president of the University,
his close friend, was loyal, and did all that lay in his power to
prevent the iniquity of the trial and the decision. This only added to
the tragedy. To have been tried by his peers and co-laborers, by
those who knew him and the field of his labors, would have been
perfectly fair, but to be accused by three or four narrow-minded
students (one of whom recanted later and all of whom deserve
oblivion), who had come with malice aforethought, whose very
presence in the school was a lie, to be accused by such as these, I
say, and then tried by a board of judges, to whom he was largely a
stranger, not one of whom probably was his equal as a scholar in the
field involved—this made the shame to the school, to himself, and to
truth, doubly deep and sore.
There remained one thing more for him to do; and as soon as he
could do it kindly, as a Christian, and dispassionately, as a scholar,
without bias or prejudice or any personal ends except the ends of
gratitude and truth, he set about his autobiography. And I wonder if,
among autobiographies, there is another that approaches his for
detachment, restraint, and self-negation; for absolute adherence to
the facts for the sake of the truth involved, a truth not of self at all,
but wholly of scholarship? This is more of a thesis than an
autobiography—as if the author were writing of another “Wall of
Nehemiah,” and no more involved in it, personally, than he was
present in “The World before Abraham”!
This is one of the most remarkable evidences of severe and scientific
scholarship that I have ever seen; and it is equal evidence of the
inherent literary value of human life. No accusing word is here,
nothing bitter and unchristian. But just the opposite: “For the Benefit
of My Creditors” is a work of love. His very character had been
assailed by his enemies, but this, while it hurt, could not harm him.
He stood upon his conscious integrity, calm and silent. It was not the
attack upon himself that concerned him. It was that Truth had been
attacked. It was an attempt to make the Bible a denominational
book; to confound truth with tradition and give it a doctrinal color or
a denominational slant. The Church may compel its theologians to
do that if it has to, but its scholars, those who discover truth, it
should leave free. God and truth are not denominational, nor
Protestant nor Catholic nor Hebrew. God is truth, and single or
separate, God and Truth belong to the fearless, the frank, and the
pure—in science not more than in religion. For “are ye not as the
children of the Ethiopians unto me, O children of Israel?... Have not
I brought up Israel out of the land of Egypt, and the Philistines from
Caphtor, and the Syrians from Kir?”
I recall the day we came upon that wonderful passage in Amos in
our study of this prophet; and how for the first time in my life the
universality of truth dawned upon me out of that passage. I had had
a tribal, denominational God, up to that time. I had been seeing
different kinds of truth—like the different tribes of old in Palestine—
warring truths, each with its own territory, its own grip upon me,
when suddenly, as the “Rabbi” opened up this mighty saying of
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!

ebooknice.com