Chi-Squared Data Analysis and Model

Testing for Beginners 1st Edition Carey

Witkov
Visit to download the full and correct content document:
https://ebookmass.com/product/chi-squared-data-analysis-and-model-testing-for-begi
nners-1st-edition-carey-witkov/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Accelerated Life Testing of One-shot Devices: Data

Collection and Analysis Narayanaswamy Balakrishnan

https://ebookmass.com/product/accelerated-life-testing-of-one-
shot-devices-data-collection-and-analysis-narayanaswamy-
balakrishnan/

Data Analysis for the Life Sciences with R 1st Edition

https://ebookmass.com/product/data-analysis-for-the-life-
sciences-with-r-1st-edition/

Data-Driven and Model-Based Methods for Fault Detection

and Diagnosis Majdi Mansouri

https://ebookmass.com/product/data-driven-and-model-based-
methods-for-fault-detection-and-diagnosis-majdi-mansouri/

Data Wrangling on AWS: Clean and organize complex data

for analysis Shukla

https://ebookmass.com/product/data-wrangling-on-aws-clean-and-
organize-complex-data-for-analysis-shukla/
Statistics and Data Analysis for Nursing Research (2nd
Edition ) 2nd Edition

https://ebookmass.com/product/statistics-and-data-analysis-for-
nursing-research-2nd-edition-2nd-edition/

Volume III: Data Storage, Data Processing and Data

Analysis 1st ed. 2021 Edition Volker Liermann (Editor)

https://ebookmass.com/product/volume-iii-data-storage-data-
processing-and-data-analysis-1st-ed-2021-edition-volker-liermann-
editor/

Numerical Methods Using Kotlin: For Data Science,

Analysis, and Engineering 1st Edition Haksun Li

https://ebookmass.com/product/numerical-methods-using-kotlin-for-
data-science-analysis-and-engineering-1st-edition-haksun-li-2/

Numerical Methods Using Kotlin: For Data Science,

Analysis, and Engineering 1st Edition Haksun Li

https://ebookmass.com/product/numerical-methods-using-kotlin-for-
data-science-analysis-and-engineering-1st-edition-haksun-li/

Introduction to Python for Econometrics, Statistics and

Data Analysis Kevin Sheppard

https://ebookmass.com/product/introduction-to-python-for-
econometrics-statistics-and-data-analysis-kevin-sheppard/
Chi-Squared Data Analysis and Model
Testing for Beginners

Carey Witkov
Harvard University

Keith Zengel
Harvard University

OXFORD
UNIVERSITY PRESS
 OXFORD
UNIVERSITY PRESS

Great Clarendon Street, Oxford, OX2 6DP,

United Kingdom

Oxford University Press is a departmentof the University of Oxford.

It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK andin certain other countries
© Carey Witkov and Keith Zengel 2019
The moralrights of the authors have been asserted
First Edition published in 2019
Impression: 1
All rights reserved. No part of this publication may be reproduced,stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must notcirculate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2019941497
ISBN 978-0-19-884714—4 (hbk.)
ISBN 978-0-19-8847 15-1 (pbk.)
DOI: 10.1093/0s0/9780 198847 144.001.0001
Printed and bound by
CPI Group (UK) Ltd, Croydon, CRO 4YY
Linksto third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Preface

Recent ground-breaking discoveries in physics, including the discovery of the

Higgs boson and gravitational waves, relied on methodsthat included chi-squared
analysis and modeltesting. Chi-squared Analysis and Model Testing for Beginners
is the first textbook devoted exclusively to this method at the undergraduate
level. Long considered a “secret weapon”in the particle physics community, chi-
squared analysis and model testing improves upon popular (and centuries old)
curvefitting and model testing methods whenever measurementuncertainties and
a model are available.
The authors teach chi-squared analysis and model testing as the core method-
ology in Harvard University’s innovative introductory physics lab courses. This
textbook is a greatly expanded version of the chi-squared instruction in these
courses, providing students with an authentic scientist’s experience of testing and
revising models.
Over the past several years, we’ve streamlined the pedagogy of teaching
chi-squared and model testing and have packaged it into an integrated,self-
contained methodology that is accessible to undergraduates yet rigorous enough
for researchers. Specifically, this book teaches students and researchers how to
use chi-squared analysis to:

¢ obtain the best fit model parameter values;

* test if the best fit is a goodfit;
¢ obtain uncertainties on best fit parameter values;
¢ use additional modeltesting criteria;
¢ test whether modelrevisions improvethefit.

Why learn chi-squared analysis and model testing? This book provides a
detailed answerto this question, but in a nutshell:
In the “old days” it was not uncommon (though mistaken even then) for
students and teachers to eyeball data points, draw a line of best fit, and form an
opinion about how goodthefit was. With the advent of cheap scientific calculators,
Legendre’s and Gauss’s 200 year old methodofordinary least squares curvefitting
became available to the masses. Graphical calculators and spreadsheets made
curve fitting to a line (linear regression) and other functionsfeasible for anyone
with data.
Althoughthis newly popular form of curve fitting was a dramatic improvement
over eyeballing, there remained a fundamental inadequacy. Thescientific method
is the testing and revising of models. Science students should be taught how to
vi Preface

test models, not just fit points to a best fit line. Since models can’t be tested
without assessing measurement uncertainties and ordinary least squares andlinear
regression don’t accountfor uncertainties, one wondersif maybe they’re the wrong
tools for the job.
The right tool for the job of doing science is chi-squared analysis, an out-
growth of Pearson’s chi-square testing of discrete models. In science, the only
things that are real are the measurements, andit is the job of the experimental
scientist to communicate to others the results of their measurements. Chi-squared
analysis emerges naturally from the principle of maximumlikelihood estimation
as the correct way to use multiple measurements with Gaussian uncertainties
to estimate model parameter values and their uncertainties. As the central limit
theorem ensures Gaussian measurementuncertainties for conditions found in a
wide range of physical measurements, including those commonly encountered
in introductory physics laboratory experiments, beginning physics students and
students in other physical sciences should learn to use chi-squared analysis and
model testing.
Chi-squared analysis is computation-intensive and would not have been
practical until recently. A unique feature of this book is a set of easy-to-use,
lab-tested, computer scripts for performing chi-squared analysis and model
testing. MATLAB® and Python scripts are given in the Appendices. Scripts
in computer languages including MATLAB, Python, Octave, and Perl can be
accessed at HYPERLINK “https://urldefense.proofpoint.com/v2/url?u=http-
3A__www.oup.co.uk_companion_chi-2Dsquared2020&d=DwMFAg&c=WO-
RGvefibhHBZq3fL85hQ&r=T6Yii733Afo_ku3s_Qmf9XTMFGvqD_VBKcN4
fEIAbDL4&m=Q09HDomceY3L7G_2mQXKk3T9K1Tm_232yEpwBzpfoNP4
&s=DFO05tbd6j37MCerXNmVwfVLRSpA7pXaEx29 lhsfBak4&e="www.oup.co.
uk/companion/chi-squared2020.
Another uniquefeature ofthis bookis thatsolutionsto all problemsare included
in an appendix, making the booksuitable for self-study.
We hope that after reading this book you too will use chi-squared analysis to
test your models and communicate the results of your measurements!

MATLAB®js a registered trademark of The MathWorks, Inc. For MATLAB® productinforma-

tion, please contact:
The MathWorks,Inc.
3 Apple Hill Drive Natick, MA, 01760-2098 USA
Tel: 508-647-7000
Fax: 508-647-7001
E-mail: info@mathworks.com
Web: https://www.mathworks.com
Howto buy: https://www.mathworks.com/store
Find yourlocal oce: https://www.mathworks.com/company/worldwide
Acknowledgments

All textbook authors owe debts to many people, from teachers to supporters.
The innovative Harvard introductory physics lab course sequence, Principles
of Scientific Inquiry (PSI lab), that trains undergraduate students to thinklike
scientists using chi-squared model testing, was innovated by physics professors
Amir Yacoby and Melissa Franklin more than a decade ago. Important early
contributions to PSI lab were also made byinstructors and instructional physics
lab staff Robert Hart, Logan McCarty, Joon Pahk, and Joe Peidle. Both authors
have the pleasure each semester of teaching PSI lab courses with Professors
Yacoby and Bob Westervelt, Director of Harvard’s Center for Nanoscale Systems.
We also gratefully acknowledge the helpful feedback provided by our many
students, undergraduate Classroom Assistants and graduate Teaching Fellows
over the years.
We are indebted to colleagues for reviewing this book and making valuable
suggestions, including Bob Westervelt. Special thanks go to Preceptor Anna
Klales, Instructional Physics Lab ManagerJoe Peidle, and Lecturer on Physics and
Associate Director of Undergraduate Studies David Morin, who made detailed
suggestions (nearly all of which were adopted) for every chapter.
The authors, while thanking many people above for their contributions and
feedback, blame no one but themselves for any errors and/or omissions.
Below are someauthor-specific acknowledgements.
CW: I thought I understood chi-squared analysis and model testing until I
met Keith Zengel. Keith’s willingness to share his in-depth understanding of the
technicalities of chi-squared analysis and model testing over the past three years
we’ve taught together is greatly appreciated and gratefully acknowledged. I'd like
to thank my 12-year-old son Benjamin (who decided sometime during the writing
of this book that he doesn’t wantto be a physicist!) for not complaining (too much)
over time lost together while this book was being written.
KZ: I used chi-squared analysis a lot of times as a graduate student at the
ATLASExperimentwithoutreally understanding whatI was doing. Later, I had to
do a lot of secretive independent research to be able to answer the good questions
posed by Carey Witkov whenwestarted teachingit together. I think we managed to
learn a good amountaboutthe subject together, mostly motivated by his curiosity,
and I hopethat others will benefit from Carey’s vision of science education, where
students learn how to speak clearly on behalf of their data.
Contents

1 Introduction

2 Statistical Toolkit
2.1 Averaging
2.2 Standard Deviation (RMSD)
2.3 Standard Error 10
2.4 Uncertainty Propagation 12
2.5 Random versus Systematic Uncertainties and Rounding Rules 14
2.6 Problems for Chapter 2 16

3 One Parameter Chi-squared Analysis 18

3.1 Modeling a System 18
3.2 Uncertainties and the Central Limit Theorem 19
3.3 Using a Fit to Combine Measurements 20
3.4 Chi-squared Fitting 25
3.5 Maximum Likelihood Estimation 27
3.6 Example: Block on a Frictionless Incline 30
3.7 Problems for Chapter 3 31

4 Two-Parameter Chi-squared Analysis 32

4.1 Problems for Chapter 4 36
5 Case Study 1: Falling Chains 37
5.1 Falling Chain Model 37
5.2 Data Collection 39
5.3 Chi-squared Analysis 40
5.4 Problems for Chapter 5 42

6 Case Study 2: Modeling Air Resistance on Falling Coffee Filters 43

6.1 Viscous Drag Model 44
6.2 Data Collection 45
6.3 Chi-squared Analysis 46
6.4 Revising the Model 47
6.5 Testing the Revised Model 49
6.6 Finding the Optimal Modelfor a Dataset 51
6.7 Using Chi-squared to Debug Experimental Issues 54
6.8 Problems for Chapter 6 56
xX Contents

7 Advanced Topics 57
7.1 Probability Density Functions 57
7.2 The Chi-squared Probability Density Function 62
7.3 The Reduced Chi-squared and Degrees of Freedom 67
7.4 The 68% and 95% Contours for Two Parameters 67
7.5 Poisson Distributed Uncertainties 70
7.6 Maximum Likelihood Estimation for Nonlinear Models
and OtherProbability Density Functions 71
7.7 Problems for Chapter 7 72

Appendices 75
Appendix A MATLAB: One-Parameter Linear Chi-squared Script 75
Appendix B MATLAB: Two-Parameter Linear Chi-squared Script 77
Appendix C Python: One-Parameter Linear Chi-squared Script 79
Appendix D Python: Two-Parameter Linear Chi-squared Script 81
Appendix E Solutions 83
E.1 Solutions for Chapter 2 83
E.2 Solutions for Chapter 3 86
E.3 Solutions for Chapter 4 90
E.4 Solutions for Chapter 5 91
E.5 Solutions for Chapter 6 91
E.6 Solutions for Chapter 7 92
Glossary 97
Books for Additional Study 99
Index 101
Introduction

Say you discover a phenomenonthat you thinkis interesting or important or by

some miracle both. You want to understand and hopefully use your discovery to
help the world, so using your knowledge of science and mathematics you derive
a simple mathematical model that explains how the phenomenon works. You
conclude that there is somevariable “y” that depends on some other variable “x”
that you can control. The simplest version of a model is that they are the same,
that y = x. You probablyaren’t this lucky, and other parameters that you don’t
control, like “A” and “B,” show up in your model. A simple example ofthis is the
linear model y = Ax + B. You don’t control A or B, but if you had some way of
estimating their values, you could predict a y for any x. Of course,in real life, your
model may have morevariables or parameters, and you might give them different
names, but for now let’s use the example of y = Ax + B.
What we want is a way to estimate the parameter values (A and B) in your
model. In addition to that, we’d like to know the uncertainties of those estimates.
It takes a little extra work to calculate the uncertainties, but you need them to give
your prediction teeth—imaginehowdifficult it would be to meaningfully interpret
a prediction with 100% uncertainties! But even that isn’t enough. We could make
parameter estimates for any model, but that alone wouldn’t tell us whether we
had a good model. We’d also like to know that the model in which the parameters
appearcan pass some kind of experimental “goodnessoffit” test. A ““go/no go” or
“accept/reject” criterion would be nice. Of course, there’s no test you can perform
that will provide a believable binary result of “congratulations, you gotit right!” or
“wrong model, but keep trying.” We can hopefor relative statements of “better”
or “worse,” some sort of continuous (not binary) metric that allows comparison
between alternatives, but not a statement of absolute truth.
Summarizing so far, we’ve specified four requirements for a method of analyz-
ing data collected on a system for which a modelis available:

Parameter estimation (often called curve fitting).

PWN

Parameter uncertainty estimation.

Modelrejection criterion.
Model testing method with continuous values.

Chi-Squared Data Analysis and Model Testing for Beginners. Carey Witkov and Keith Zengel.
© Carey Witkov and Keith Zengel 2019. Published in 2019 by Oxford University Press.
DOI: 10.1093/0s0/9780 198847 144.001.0001
2 Introduction

Since the general problem we presented (parameter estimation and model

testing) is a universal one andsince the specific requirements we came up with
to solve the problem arefairly straightforward, you might expect there to be a
generally accepted solution to the problem that is used by everyone. The good
newsis that there is a generally accepted methodto solve the problem. The bad
newsis thatit is not used by everyone andthe people whodouseit rarely explain
howto useit (intelligibly, anyway).
That’s why wewrote this book.
This generally accepted solution to the system parameter estimation and model
testing problem is used by physicists to analyze results of some of the most
precise and significant experiments of our time, including those that led to the
discoveries of the Higgs boson and gravitational waves. We teach the method
in our freshman/sophomore introductory physics laboratory course at Harvard
University, called Principles of Scientific Inquiry. Using the software available at
this book’s website to perform the necessary calculations (the method is easy
to understand but computation intensive), the method could be taught in high
schools, used by more than the few researchers who useit now, and used in many
morefields.
The method is called chi-squared analysis (x? analysis, pronounced
“kai-squared”) and will be derived in Chapter 3 using the maximum likelihood
method. If in high schoolor college you tested a hypothesis by checking whether
the difference between expected and observed frequencies wassignificantin a fruit
fly or bean experiment, you likely performed the count data form of chi-squared
analysis introduced bystatistician Karl Pearson. Karl Pearson’s chi-squared test
was included in an American Association for the Advancementof Sciencelist of
the top twenty discoveries of the twentieth century.! The version of chi-squared
analysis that you will learn from this book works with continuous data, the type
of the data you get from reading a meter, and provides best fit model parameter
values and model testing. Parameter estimation and modeltesting,all from one
consistent, easy to apply method!
Butfirst, let’s start with a practical example, looking at real data drawn from an
introductory physics lab experimentrelating the length and period of a pendulum.
The period of a simple pendulum depends onits length and is often given by
T=2n JE. This pendulum modelisn’t linear, but it can be cast in the linear form
y = Ax as follows: L= aT’, where X =T?, Y=L, and A= gt: However,
something seems to be wrong here. In this experiment we change the length
(L) and observe the period (T), yet L in the model is the dependentvariable
y! There’s a reason for this. One of the simplifications we will be using
throughout most of this book is the assumption that all uncertainties
are associated with measurementsof the dependentvariable y. You might
think that measurements of pendulum length should have even less uncertainty

1 Hacking,I. (1984). Science 84, 69-70.

 Introduction 3

than pendulum period, but we assure you that from experience teaching many,
many, physics lab sections, it doesn’t! The accuracy of photogate sensors or
video analysis, coupled with the large number of period measurements that
can be averaged, makes the uncertainty in period data fairly small, while meter
stick measurements, especially when long pendulums(greater than a meter) are
involved, often exhibit surprisingly large uncertainties.
Figure 1 displays a plot of data collected over a wide range of pendulum lengths
and periods. There are two importantfacts to note aboutthe plot. First, each data
point represents a mean value, not an individualtrial value, so each data pointis
the result of repeated measurements. Second,the plot line is not the best fit line.
Instead, the plot line is the model line calculated using known parameter values.
This is a model where we havethe luxury of knowing parametervaluesprecisely
(2, 2, and the acceleration due to gravity, g) before performing any experiments
on our system.
Looking at the data points and modelline in figure 1, what are your thoughts
about how well the data fits the model line? At first glance the fit looks pretty
good, right? A naive approach would be to assume from theclose visual fit that
the model must be good and then proceed to obtain a best fit line using linear
regression (least squares with a linear model).
Least squares goes back more than two centuries to Legendre and Gauss and
involves minimizing the sum of the squared differences between each data point
and the model,
Yor _- Ymodel)”s
i

1.8,
[ fp Data 7
1.6. F) essssne Model Line eo 7

1.4
TIT

1.2
Length [m]

*.
Oo
oo

1
So
pb
mT

ot pe oo
0 2 4 6 8
Period? [s?]

Figure 1 Pendulum data and model line.

4 Introduction

a
t hi Data °
157)... Model Line 7

1.495 F i 4

= 1.49 “ 4
ep L
G
uo
L eS
ja r Poa 4
1.485 7 a 4

1.475E, boa tt 1 4
5.96 5.98 6 6.02 6.04
Period?[s2]

Figure 2 Pendulum data with error bars after zooming.

to arrive at best estimates for parameters A and B in the model ofa straight line
y= Axt B.
By zooming in on a few ofthe data points, as seen in Fig. 2, we see that what
initially looked like a good fit is not. Each error bar correspondsto plus or minus
one standard error (“one sigma” or 1o).? It is now clear that the discrepancy
between data and model is big enoughto drive a (several o) truck through! Note
that some data points are many sigmas (o’s) from the model. A discrepancy of
only two sigma is sometimes considered sufficient to reject a model. In this case
the model was entirely wrong, but we couldn’t tell simply by “eyeballing”’it.
It might seem like any curve fitting technique could constitute a form of model
testing, but, as shownin this example,failure to include measurementuncertainties
can prevent us from seeing that the fit to the data is poor. For example, linear
regression, which is built into even the least expensive scientific calculators and
is one of the most commonuses for spreadsheets in business and science, does
not include measurementuncertainties. Therefore, ordinary least squares is not an
acceptable methodfor parameter estimation becauseit treats all measurements the
same,even if their uncertainties vary greatly. Tacking other methods onto ordinary
least squares (like correlation coefficients or the F-test) in an effort to add some
form of model testing is a fragmented approach that doesn’t addressthe keyissues:
that parameter estimation and model testing require measurement uncertainties
and that parameter estimation and model testing cannot be separated. Parameter

2 The standard erroris the standard deviation of the mean. Standard deviation and standard error
are widely used measures of uncertainty that are formally introducedin the Chapter2.
 Introduction §

estimation and modeltesting is a “chicken or egg” problem: how can we know if

a modelis good withoutfirst estimating its parameters, yet how can weestimate
meaningful model parameters without knowingif the model is good?
In this book we present one consistent methodology, chi-squared analysis, built
on the terra firma of probability theory, that combines parameter estimation and
modeltesting by including measurement uncertainties and answersfive important
questions about the results of an experiment on a system for which a modelis
available:

Whatare the best fit parameter values?

wk WNT

Is the bestfit a goodfit?

Whatare the uncertainties on the bestfit parameters?
Evenif the fit is good, should the modelstill be rejected?
Is the revised model an improvement?
2
Statistical Toolkit

2.1 Averaging
Chi-squared analysis requires familiarity with basic statistical concepts like the
mean, standard deviation, and standard error. Even if you are familiar with these
termsit would still be a good idea not to skip this chapter as we present them in a
way that is most useful for chi-squared analysis and discuss some novelaspects.
Let’s start with a simple reaction time experiment: you hold a vertical meter
stick out in front of you while a friend (or a lab partner if you don’t have any
friends) holds their open hand near the zero mark at the bottom, then you drop
it and they catch it as quickly as they can. You and your friend (or whoever)
can estimate their reaction time from this experiment using the kinematic relation
between displacementand time for constant acceleration with zeroinitial position
and zeroinitial velocity,

1
d= —gi’,
58 1
(1)

which can be rearranged to find!

2d
=,/—. (2)
&

If you perform the experiment twenty times, discard the first trial to avoid
measuring “practice effects,”* starting with your handhalf open,you mightobtain
results like those in Table 1, forfivetrials.

! We’re going to go ahead and throw awaythe negative reaction times that result from the square
roothere, butit’s probably worth noting that at least once in physics history a Nobel Prize was awarded
to someone whorefused to disregard apparently non-physical negative values. In 1929, Paul Dirac
interpreted positive and negative energies resulting from a square root as corresponding to particles
and anti-particles. He was awarded the Nobel Prize in 1933, and Carl Anderson, who discovered the
first anti-particle in 1932, was awarded the NobelPrize in 1936.
2 Del Rossi G., Malaquti A., and Del Rossi S. (2014). Practice effects associated with repeated
assessmentofa clinical test of reaction time. Journal ofAthletic Training 49(3), 356-359.

Chi-Squared Data Analysis and ModelTesting for Beginners. Carey Witkov and Keith Zengel.
© Carey Witkov and Keith Zengel 2019. Published in 2019 by Oxford University Press.
DOI: 10.1093/os0/9780 198847 144.001.0001
 Averaging 7

Table 1 Falling meter stick

dataforthe first five trials.

Displacement (m)

0.27

0.33

0.29

0.31

0.26

How would you use the displacement data to estimate your reaction time? One
possibility is to average the displacements and substitute the mean displacement
into Eq. 2. We could write this in equation form as

= |—, (3)

where the angular brackets ( and ) denote the mean value of the thing enclosed
and an asterisk ¢* denotes the best estimate of t. Another possibility is to add a
second column to Table 1 labeled Time (s), compute a reaction time for each
displacement using Eq. 2, and average the reaction times. We could write this in
equation form as

. 2d
eo ( “4 (4)
&

Would these two algorithms give the sameresults? If not, which one would give
the correct result?
As both algorithmsinvolve averaging it may be helpful to review why we average
data in the first place. We average data because webelieve that there is a “true”
value and that the measurement uncertainty causes small, random deviations from
this true value. These deviations are equally likely to be positive or negative, so
averaging (which involves summing) will tend to cancel these positive and negative
deviations from the true value.
So which algorithm, averaging displacements or averaging reaction times, is the
“correct way” to estimate reaction time from displacement data? Well, averaging
causes deviations from the true value to cancel when each deviation contributes
equally and has an equal chanceof being positive or negative. The equation for
computing the reaction time (Eq. 2) applies a square root operation to the dis-
placement measurements. The square root operation weights some displacements
8 Statistical Toolkit

more than others, so the positive and negative deviations from the true value are
weighted differently and can no longer be expected to cancel by simple averaging.
Figure 3 showsthe subtle shift caused in this case by this square root.
Therefore we should average the displacements in order to cancel these
deviations. This is a rare case where the simpler of two algorithms, merely
substituting once into the equation using mean displacement, is better! This
can bestated in the form of a maxim: Averagethe data, not the calculations!

(a) Meter stick drop displacement data and mean displacementvalue.

VSO ro or
H {__] Data
ot MeanDisp.

100 - 4
a
et
2
S
a
a Fr

sob 4

ghee tith id td
0.1 0.15 0.2 0.25 0.3
d [ml]
(b) Meterstick drop calculated time data, with mean values for the average-then-calculate and
calculate-then-average algorithms shown.
160 -—— r 7
[ | Data |
140 [| ....... Calc. First 7
[| -~-- Avg. First ]
120 I
1 7
°
Oo
—
Trials/0.02 s

TT
oo
So

60F
4of

0.15 0.2 0.25

t (s]

Figure 3 Histograms of the displacement (a) and time (b) for 500 simulated meter stick drops.
 Standard Deviation (RMSD) 9

2.2 Standard Deviation (RMSD)

Let’s say you and yourfriend (or whoever) thoughtthis calculated reaction time
was an important scientific result. You write a paper on your experiment and
results and send it to The fournal of Kinesthetc Experiments (The JoKE), but it
gets bounced back. You left something out. You forgot to include an estimate of
uncertainty in your reaction time. The uncertainty in your reaction time depends
on the spread (synonyms: dispersion, scatter, variation) in the data. How do you
characterize the spread in data? Listing all of your data is usually difficult to read
and impractical. Giving highs and lows is better than nothing (though not by
much)as they disregardall of the other data and onlycharacterize outliers. Taking
the deviation (difference) of each value from the mean seems useful but runs into
the problem that the meandeviationis zero.
‘To get around the problem of zero mean deviation, we have a few options. We
could take the absolute value of the deviations before finding their mean. That
would give a positive and non-zero value, butit’s not fun working with absolute
values in calculus. A second option is to square the deviations. This solves one
problem but creates another. The mean squared deviation is positive but we now
have square units (square meters in this case). To solve the square units problem
wetake the square root of the mean of the square deviations. So the algorithm is:
calculate the deviations, square them, find their mean, and take the square root.
If you read this backwards you have the root-mean-square deviation, or RMS
deviation. In fact, you can re-create the algorithm for N measurementsof a variable
y by reading root-mean-square deviation word by word from righttoleft.
First, write the deviations: y; —(y).
Second, square the deviations: (y;—(y))?.
. 2
Third, take the mean of the square of the deviations: LOE

Fourth, take the square root: \/ Lor |

Ofcourse, this result is better known asthe standard deviation, butit is really
the root-mean-square deviation (RMSD),

6)

3 You may find that in the standard deviation equation people sometimes use an N in the
denominator, while other people use an N — 1. You mayalso find that people have long arguments
about which is better or correct. We'll keep it simple and stick to the RMSD,which by definition has
an N in the denominator.
10 Statistical Toolkit

2.3. Standard Error

You resubmit yourresults to The JoKE, this tme presenting your mean reaction
displacementplus or minusits standard deviation, but your submission is bounced
back again. The editor explains that the standard deviation does not mea-
sure uncertainty in your mean reaction displacement. Then what does standard
deviation measure?
The standard deviation measures the spread of your entire set of measurements.
In this experiment, the standard deviation of the displacements represents the
uncertainty of each individual displacement measurement.
But the uncertainty of an individual measurementis not the uncertainty of the
mean of the measurements, Consider the (mercifully) hypothetical case of taking
a hundred measurements of yourreaction time. If you repeated and replaced one
of your measurements, you wouldn’t be surprisedif it was one standard deviation
from the mean value of your original 100 measurements. But unless you were
using a very long stick and something wentterribly, medically wrong, you wouldn’t
expect the mean value of your new set of a hundred measurements to be very
different from the mean value of your original hundred measurements. See figure
4 for an example of data from one hundred measurements, with and without the
last trial repeated and replaced.
Now consider the hypothetical case of taking only four measurements, finding
the mean and standard deviation, and then repeating and replacing one of those
measurements. See figure 5 for an example of data from four measurements, with
and withoutthe last trial repeated and replaced. You would still find it perfectly
natural if the new measurement were one standard deviation from the mean of
your original four measurements, but you also wouldn’t be so surprised or find
much reason for concern for your health if your new mean value shifted by a
considerable amount. In other words, the uncertainty of your next measurement
(the standard deviation) won’t changeall that much as you take more data, but
the uncertainty on the mean of your data should get smaller as you take more
data. This is the same logic we used previously to explain why we averaged
the displacementdata: every measurement includes some fluctuation around the
true value, which cancels out when we take the average. We don’t expect these
experimental uncertainties to be magically resolved before the nexttrial, but as we
take more data, the mean valueis less susceptible to being significantly shifted by
the fluctuation in any one particular measurement.
The uncertainty of the mean (called the standard error or o) should depend on
the number of measurements (N), as well as on the spread (RMSD)ofthe data.
The equation for standarderroris

RMSD
0 = Ty (6)
For now we'll just state the result, but readers whoare interested in further
details should consult Problem 2.2.
 Standard Error 11

(a) Meter stick drop displacement data and mean displacementvalue.

30 oe
+] L_] Data
os tL MeanDisp.

20F 4

Trials/0.02 m

poo ti
15¢

10 4

0 “ J 1 ati ht J

0.1 0.15 0.2 0.25 0.3

d [m]
(b) Meterstick drop displacement data and mean displacementvalue for one hundred
measurements, with a different measured value forthelasttrial.

30 Cc | TT

[_] Data
sesenne MeanDisp.
254

20 4
Trials/0.02 m

15 :

10F 4

0.1 0.15 0.2 0.25 0.3

d [m]

Figure 4 Histograms of the displacementfor one hundred simulated meter stick drops (a) and
jor the same one hundred simulated meter stick drops with the last trial repeated and replaced
(b). The shift in the mean value is barely noticeable, even though one measurement has moved
from the farleft bin to the far right bin in (b).
12.) Statistical Toolkit

(a) Meter stick drop displacement data and mean displacementvalue.

2.5 | OS |

[|] [_] Data

[were Mean Disp.
2 L 4

E t
nq lb 4
o
Ss f
a L
e& fr 1

0.5 4

0 1 11 1 Ped ns 4 ]

0.1 0.15 0.2 0.25 0.3

d [m]

(b) Meterstick drop displacement data and mean displacementvalue, with a different measured
value for the lasttrial.
2.5 je mT TT TTT

(! [__] Data
[] vseees Mean Disp.
2 L 4

g L
a
atlS 4

g f

Bi
3 L

ost 4

0 L fe eo te pda al | at

0.1 0.15 0.2 0.25 0.3

d [m]
Figure 5 Meter stick drop displacement data and mean displacement valueforfour
measurements, with different measured valuesfor the lasttrial.

2.4 Uncertainty Propagation

You resubmit your mean reaction displacement and its uncertainty, expressed as
its standard error. Again, your manuscript bounces back!
Another random document with
no related content on Scribd:
mutton-pies and “Stuckling.” This latter is a production which, I am
happy to say, is peculiar to Winchester, and fortunately only to be
found there during two days in Election week. In external
appearance and in section it is similar to a mince pie, and in taste it
is something like one that has been soaked in vinegar, and then kept
till it is mouldy. It is made of chopped beef, currants, suet, apples,
and carraway seeds. I have seldom seen anybody taste it a second
time. These rations were, however, scorned by the boys; and as half
the good things from the centre table were handed over to them,
they made up for any past deficiencies in the way of dinner, in a
manner that must have been surprising to the lookers on. Venison
pasty, currant-pies, pea-soup, ham, ducks, cucumber, fish, &c.,
disappeared as if by magic, washed down by libations of “Huff,” (very
strong ale.) The gospel was read by the Bible Clerk for the week
after the first course, and at the conclusion the beautiful grace was
sung by the boys, with the help of the cathedral choristers and
singing men. On Friday the boys had no dinner in hall, where the
Dons had it all to themselves, and afterwards adjourned for wine and
dessert to the Warden’s Gallery, where they complimented each
other in the usual after-dinner style.
All through the week the Fags revelled in cricket on their own
account, and strawberries and cream and ices were supplied in a
continual stream by the Oxford men and old Wykehamists, who
resorted to Meads in considerable numbers. One can get too much
of a good thing; for on one of these days I came up melting with heat
from Meads, and going into chambers washed my hands and face,
and consumed a quantity of ice. The next morning when I awoke I
found my pillow saturated with blood, which had flowed from my
nose during my sleep,—this was the only time in my life that I have
ever had a spontaneous effusion from that organ. The morning
following, when I awoke, I felt a strange irritation over my face and
hands, with some symptoms of eruption. I sent for a doctor, who
informed me I had the smallpox. I denied the soft impeachment, as I
felt perfectly well. I was sent home, and my face and hands soon
became one mass of pimples, and, after about a fortnight, all the
skin peeled off, and I recovered with the loss of my complexion,
which exchanged its former peculiar delicate hue for that of boiled
veal, which it has ever since retained.
When a boy was going to leave the School for good, he had to
send in his resignation to the Warden, written on a slip of paper. It
ran as follows:—

“Ego Robertus Jones omne jus quod habeo aut unquam habui in hoc Collegio
Beatæ Mariæ Winton prope Winton, ætate cogente (or parentibus volentibus)
libenter resigno.”

The last Friday evening of the half year there was a grand
celebration of “Domum” in school, which was adorned with branches
and banners. A military band was engaged, and “Domum” was sung
by all the company till they were hoarse. It was an exceedingly gay
and animated affair. Numbers of ladies graced the school with their
presence, and crowds of old Wykehamists, fancying themselves
boys once more, wandered about among the old familiar scenes.
From school the company adjourned to Meads, and then to the
Chamber quadrangle, at each place singing “Domum” louder than
before, till at last the power failed, and the ladies, visitors, and
Superannuates went to recruit their energies for a brief period before
going to the “Domum ball” at St John’s rooms, of which the
Superannuates were stewards, and which was generally kept up
with tremendous energy far into the small hours. The boys’ last
breakfast at the George or White Hart formed an appropriate supper
for the most enthusiastic dancers.
After chapel on the last morning all the boys, with the second
master at their head, marched round the quadrangle, singing the
beautiful Latin hymn commencing with Jam lucis orto sidere; they
then adjourned to Sixth, where the journey money was distributed,
and then went off rejoicing to the “Last breakfast.”
There was generally an exciting contest between the adherents of
the George and White Hart hotels, as to which should provide this
entertainment. The former generally had the call; but one year, when
some of us had made a particularly good feed at the latter
establishment, we got up a little private canvass before the usual
time, and to the great disgust of the Præfects, who were all
Georgists, we carried the day triumphantly. The expense was
defrayed by the boys subscribing the last three “Batlings,” (i.e., the
weekly shilling allowed each boy.) This was rather an illusory coin,
for we seldom actually fingered it, as some one of the College
servants generally had a kind of prescriptive right to a benefit; and
whenever Saturday arrived, Præfect of Hall’s valet was sure to come
round to ask the boys if they would give their batlings to Rat
Williams, or Dungy, or Purver, or Long John, or some other equally
deserving individual. The last breakfast was a decidedly jolly and a
very good repast,—in fact, it was a regular dinner, with tea and
coffee instead of wine. Then the inn-yard was full of postchaises and
four-wheelers, with a four-horse coach or two drawn up outside; and
the exulting Junior slipping a shilling into the expectant hand of the
obsequious “Yessah,” stepped proudly into his vehicle, and soon
forgot all the troubles of his first half in delightful anticipation of
coming festivities.
 CHAPTER XV.
CONCLUSION.

If I could only flatter myself that any of my readers have felt one
quarter of the pleasure in reading my little book that I have derived
from writing it, I should be more than satisfied. On looking it over,
however, I have some misgivings; I fear that those who know not
Winchester may think the subject too trivial for illustration, and that
Wykehamists may regret that so good a subject has been so
indifferently handled.
It has surprised me to find how many occurrences, names, words,
and faces that for years had lain dormant in my mind, started up
again in my memory, as I sauntered through Chambers, School, and
Meads; how strange was the sensation to be looked at as a stranger,
in a place every stone of which was as familiar to me, nay, I may say
more familiar to me, than any other spot in the world. The boys, their
names, their dress, their games, their quaint language, (of which I
have endeavoured to draw up a glossary,) all seem unchanged; and
even I fancied that I myself was once more a Junior, till mechanically
snatching at a ball rolling past me, my vertebræ most disagreeably
reminded me what o’clock it was.
It has often occurred to me how very remarkably the relative
positions of boys change in after-life; and if all those who were my
school-fellows could be collected together how differently we should
regard each other. Such a gathering would be of course impossible,
even if death had not cut short the career of very many; scattered
through every portion of the globe, working hard in their different
professions, or vegetating in luxurious ease or careless indolence,
how contrary to all our expectations has many a one turned out;
some, neglected and despised while at school, have risen to
eminence, while others, whose early youth seemed to give bright
promise, have never raised their heads above the common throng.
This change is even more remarkable in the physical appearance of
individuals than in their moral character. I am myself, in some
respect, an example of this. I left school when I was seventeen, if
anything rather smaller than other boys of my age; some years after
I was at a ball at Glasgow, when I saw a little officer whose face was
perfectly familiar to me, and I perceived that mine was not strange to
him, so we soon fraternized; I found that my friend was Long Eves,
as we used to call him from his being so tall. I am no giant, and I now
found that I had overtopped him by a head. At Oxford I found men
who, when at Winchester with me, could have thrashed me with one
hand, but who then, if engaged in such a contest, would I think have
preferred the use of three.
But I must not quit my subject altogether without noticing some of
the more important changes that have taken place in the institutions
of the College since the time when I was first admitted. I will mention
them in the same order that I observed when treating of them as
they were in my time.
The Choristers are now no longer mere fags to the boys, but really
do sing, and have had a handsome school built for their
accommodation. The old Commoners’ buildings were removed in my
time, and replaced by a more commodious and hideous structure, in
which for some time the head-master received boarders. These have
all migrated to four masters’ houses in Culver’s Close. They still call
themselves Commoners, so that, for purposes of athletic
competitions, &c., the school is divided into College, Commoners,
and Houses (these latter being the occupants of the other five
masters’ houses not in Culver’s Close). The building which was
vacated by Commoners has been adorned and renovated by Mr.
Butterfield, and now presents quite a respectable appearance. In it
the new Library and various lecture rooms are located. The College
boys are still 70 in number, but the number in Commoners and
Houses has more than doubled since 1836, when there were about
120 (I believe at one time they dwindled down to 65). There are now
about 350, and, if more masters’ houses were built, this number
might be greatly increased. There are now nine masters’ houses, in
each of which from 35 to 40 boys are accommodated, and the
number of applications from anxious parents who wish to get their
boys into the school is so excessive as to cause great
embarrassment to the masters and much disappointment to the
candidates.
In College the office of Præfect of Tub has been abolished, and a
Præfect of Library installed in his place. There are now only two half
remedies (Tuesdays and Thursdays) in each week; but Thursday is
a whole remedy in the Summer term, and there is a vacation of three
weeks at Easter. Going in procession to “Hills” is discontinued, and
the boys may go where they like (except into the town) out of school
hours. Watching-out at cricket is considerably limited, and “kicking-
in” at football entirely abolished. The College Chambers[13] (except
Fourth) are removed to the two storeys above, which were supposed
to be occupied by the Fellows, but were hardly ever used by them. In
the new Chambers there is every necessary appliance for decency
and comfort, so that the Junior in winter has no longer to rub his
miserable little nose with a bit of ice, and think he is washing himself,
but can revel in any amount of fresh water indoors. The old
Chambers are mostly used as studies, book-chambers, and toy-time
as before; but one is used as a laundry and another as a dressing-
room after football, cricket, &c. The boys have tea found for them in
Hall, instead of having to go to sick-house for it. The dinner takes
place at one o’clock. The boys get crockery instead of trenchers to
eat their meals off, and knives and forks to eat them with.[14] Prime
joints are carved by the servants, and a master is present. The
pandemonium which took place when I was a Junior at the six
o’clock dinner has happily entirely departed. At supper gaslight
illumines the Hall, instead of the feeble flickering of a few dips. The
ceremony of the oath is abolished. The Morning Service in chapel is
at 7.50, on Saturday evenings at 5.30, on Sundays at 9, 11, and 5,
and on Saints’ days at 9 a.m. Only the members of the Choir wear
surplices. The boys only go to Cathedral on the second Sunday in
each month.
School is now kept closed, and is only opened for concerts,
lectures, &c. Its appearance has much changed. The seats and the
fixed forms which supported them are swept away, a handsome
organ is erected at the east end, and the “Tabula Legum” has been
moved to another position over the entrance-door. The mouldings of
the cornices are picked out in colour, the wainscoting varnished; and
the magnificent room presents a great contrast to what it was when it
was the only living room for the boys in play as well as in school
hours.
The area of College Meads has been much extended by throwing
down the walls which separated it from “Long” and “sick house”
Meads. A racket and fives courts have been erected by the Ridding
family for the benefit of the School. There is a large gymnasium, and
very extensive buildings erected for additional accommodation for
invalids. A splendid additional cricket-ground has been laid out
adjoining the old Meads, which together make, I fancy, the finest
playground attached to any school in the world.
The glories of the old Election week have altogether vanished. The
election to Winchester is still in the middle of July; and there is a
dinner in Hall on Domum Day, but there are very few old
Wykehamists in attendance. The parents and tutors of the
candidates for election muster strong for the examinations, which
last three days; and the ordeal they have now to pass through is very
different indeed from that which I have described as taking place
when I was a candidate. The distinction and privileges of Founders’
kin are entirely abolished. Commoners are admitted to Election
Chamber, and have an equal chance with College boys of getting
scholarships at New College.[15]
Besides all these alterations in the buildings and regulations,[16]
there are many fresh institutions which are sources of great interest
to the boys. There is a Rifle Corps, a Debating Society, Shakespeare
Society, Glee Club, Chapel Choir, a Boat and a Golf Club. Golf,
however, is not allowed in the Summer term.
On the whole, I do not think it would be a great exaggeration to
say that there have been more changes in the school arrangements
since 1835 than took place in the whole period from the foundation
to that date.
 As regards the changes that have been enforced on the College
by the Government, I feel myself incompetent to form an opinion. I
have heard them both extolled and vituperated by individuals whose
opinions are worthy of respect, and who have the welfare of the
school at heart. At any rate, I earnestly hope that the good effects
expected by the former will be fully realised, and the forebodings of
the latter prove illusory. I have little doubt myself that as long as the
moral and physical wellbeing of the boys is so carefully looked after
as it now is, that Wykeham’s College will continue to maintain its
high reputation; and though but few of her sons may dazzle the
world by their wealth, rank, or power, the majority may yet, at any
rate, prove always by their manly and honourable bearing, that they
know the meaning of the motto of their illustrious founder,

MANNERS MAKYTH MAN

 NOTE.
Mr. Wrench has recently published a “Dictionary of Winchester
Slang,” which has probably had a wide circulation among
Wykehamists. He has given a list of all the words he could collect,
used in days before my time, and those of more recent coinage. In
the following Glossary I have strictly confined myself to those in use
while I was in the school. It interested me much to learn from Mr.
Wrench’s book, not only the number of quite new words (for
instance, the title of his work—“Notions”—was invented subsequent
to my period), but also the change in the meaning of several of the
words which are still used.
 GLOSSARY
OF

WORDS, PHRASES, AND CUSTOMS, PECULIAR

TO WINCHESTER COLLEGE.[17]

⁂ The few woodcuts interspersed through the Glossary are not

such faithful representations as the author could have wished.
However, they serve in some measure to explain the various
colloquialisms.

A.
ABROAD—A boy was said “to come Abroad” when he returned to
his school duties after having been “Continent,” (v. inf.)
ABS.—Was written after a boy’s name on a Roll, (v. inf.,) to show
that he was absent from Chapel or Hills, or from any place,
where names had been called.
AD PORTAS—A Latin oration addressed by the Præfect of Hall to
the Electors, under Middle gate, at the commencement of
Election week.
AMEN CHAPEL—A long service performed on the four days set
apart for commemorating the Founder, and on the anniversary
of his death.
APPLE-PIE DAY—The last Thursday in Short half, on which we had
apple-pies for dinner.

B.
BAKER—Anything (such as a cushion or blotting-book) placed on a
form to sit upon.
BANGIES—Drab trousers; so called from
BANGY—Brown sugar.
BARTER—A half-volley (cricket term), so called from the late
Warden, who was celebrated for the vigour with which he
disposed of a ball of this description.
BATLINGS—The boys’ weekly allowance of one shilling.
BEESWAXERS—Thick laced boots.
BEEVER-TIME—A quarter of an hour’s relaxation allowed to the
boys in the middle of afternoon school in summer, to give them
an opportunity of disposing of
BEEVERS—A portion of bread and allowance of beer laid out in hall
at the time above mentioned; from the French boire, buveur.
BIBLE CLERK—An office held by the Præfects in full power in
rotation; it lasted for a week. The duties were to collect the
Verse and Prose tasks; to keep order in School; to give in the
names of those boys who were condemned to be “Bibled,” (v.
inf.,) and afterwards assist at their execution; to take the place
of the Præfect of Hall when absent, &c., &c.
BIBLER, or BIBLING—A flogging of six cuts, when the Bible Clerk
and Ostiarius laid bare a hand’s-breadth of the small of the
victim’s back.
BILL BRIGHTERS—Small faggots used for lighting kitchen fire.
BLOW—To blush.
BLUCHERS—Præfects not in full power; a metaphor taken from the
half boot.
BOB—A large white jug, holding about a gallon.
BONER—A blow given with the fist on the lowest vertebra.
BOOK—The School was divided into three classes, called Sixth,
Fifth, and Fourth Book. The Præfects formed the former. Fifth
Book was divided into Senior, Middle, and Junior Part. The
remaining lower boys constituted Fourth Book.
BOOKS, THE—The prizes given to the two boys at the head of each
Part at the end of each Half; they used at one time to be given
by the Duke of Buckingham, now by Lord Saye and Sele. Also
anybody who had got the largest score on his side in a cricket
match, was said “to have got Books.”
BOOKS CHAMBERS—Certain hours on a Remedy, (v. inf.,) during
which the boys had to attend in School, no Master being
present.
BOOKS, UP TO—A Part (v. s.) was said to be so, when it was
ranged on the “Rows” (v. inf.) before a Master, for the purpose of
repeating a lesson.
BOOTS AND LEATHERS—A Commoner Peal, (v. inf.)
BREADPICKERS—A nominal office, one being in the gift of each of
the four senior Commoner Præfects, which excused the holder
from fagging.
BROCK—To teaze, chaff, or badger.
BROLLY—An umbrella.
BRUM—Without money.
BULKY—Generous.

C.
CANDLEKEEPERS—The seven Inferiors (v. inf.) who had been in
College the longest time. They were excused from all fagging,
though if there was an absolute dearth of Juniors, Præfects in
full power had (though they rarely exercised it) the right to fag
them. They were allowed a Breakfast fag in Hall, and a Valet (v.
inf.) in Chambers; and the Senior Candlekeeper had the power
of fagging the twenty Juniors on the School side of Seventh
Chamber Passage Gate.
CANDLESTICKS—Candidates for admission into College.
CARGO—A hamper of eatables sent to a boy by his friends.
CAT’S HEAD—A Dispar, (v. inf.,) the end of a shoulder of mutton.
CHAMBER DAY—A day on which access was allowed to Chambers
during the whole day.
CHILD—Each of the Electors appointed one; his principal duty was
to pocket a sovereign.
CHINTZ—A chance.
CLASSICUS—The boy who at the end of the week has the smallest
score in the
CLASSICUS PAPER—The book in which, each day after morning
school, the position of each boy in his Part was marked, the
lowest being marked one, and so on. The Classicus had to get
the lessons set, as well as the Themes for composition.
CLEAN STRAW—Clean sheets.
CLOISTERS—The name given to Middle and Junior Part of Fifth
Book, when combined together in
CLOISTER TIME—Ten or twelve weeks at the latter end of Long
Half, commencing about Whitsunday and ending at Standing-up
week, (v. inf.)
CLOISTER ROUSH—An annual engagement between Cloisters and
Senior Part in School at the beginning of Cloister Time.
CLOW—A box on the ear.
COLLEGE WARE—Crockery that fell without breaking.
COMMON TIME—The Short Half, and beginning of Long, up to
Easter Time.
COMMONERS—The boys not on the foundation. Also the building
they lived in.
COMMONER GRUB—An entertainment given by Commoners to
College after the cricket matches.
COMMONERS SPEAKING—The day on which the speakers
selected from among the Inferiors declaimed.
CON—A blow on the head given with the knuckles or any other hard
substance, derived from κoνδυλον, a knuckle.

CON.

CONDUIT—A water tap.

CONTINENT—A boy was said to go Continent when he left School
for Sick-house.
COURSE, IN—The two Præfects of Chapel are said “to be in
Course” on alternate weeks, when they have to call names at
the end of the service. The Præfects in each Chamber are “in
Course” in rotation, during which time they are especially
 responsible for what goes on within their jurisdiction. A new boy
on his arrival was not liable to be fagged for the first fortnight;
when his fagging commenced he was said to be “in Course.”
COURSEKEEPER—An office in the patronage of the Commoner
Præfects, the duties of which were principally connected with
the organisation of the fagging department. He was required to
have been three years in the school, to be of reasonable bodily
strength, and in Middle Part. His privileges were numerous, the
principal being that he was allowed to fag. When he ascended
into Senior Part his duties ceased, but his privileges remained;
he was then called “Ex-Coursekeeper.”
CROCKETTS—Miniature Cricket, played with a stump and a five’s
ball. Also, when a boy had got no runs in a cricket match, or had
failed in a lesson of Standing Up, (v. inf.,) he was said to have
“got Crocketts.”
CROPPLED—To be floored in any examination.
CUD—Pretty, Nice.
CUSE PAPER—The same as Classicus Paper, (v. sup.)
CUT IN A BOOK—A method of drawing lots. A certain letter was
fixed on, (e.g., the first in the second line on the left page,) each
boy then turned over a leaf, and whoever turned over the leaf in
which the corresponding letter was nearest to A, won.

D.
DEPUTY—The Junior Candlekeeper, who had the organisation of
the Fagging department, and assisted the Senior Candlekeeper
in thrashing the Juniors in Hall.
DISPAR—The rations of meat served out for dinner: derived either
from “Dispertio,” to divide, or “Dispar,” uneven.
DOCK A BOOK—To tear out the leaves.
DOLE—A stratagem; a clever invention, trick, or dodge.
DOMUM—A Latin canticle, supposed to have been written by a boy
who was not allowed to go home for the holidays. The tradition
says that he carved it on a tree, and then committed suicide. On
the last Friday in Long Half, after Election, a kind of festival was
held in the evening, when numbers of people came into College,
and “Domum” was sung over and over again in School, Meads,
and the principal Quadrangle. Ten minutes before the time for
descending Hills, the three Juniors in College had to walk round
and across Hills, calling “Domum,” as a signal to the boys to
prepare to start.

E.
ELECTION—The examination of Præfects and Senior Part for New
College; and of Candidates for admission to Winchester.
ELECTORS—The Warden and two Fellows of New College; and the
Warden, Sub-warden, and Head-master of Winchester, who
conducted the examination above mentioned.
END—At dinner-time the Inferiors (v. inf.) were divided into six
companies, each being presided over by a Candlekeeper.
These companies, and the tables at which they sat, were called
“Ends.”
EX-COURSEKEEPER—(V. sup.)
EX TRUMPS—Extempore.

F.
FARDEL—Sixth Book and Senior Part were divided into three
Fardels (or parts) for the examination in Election week.
FAT FLAB—The name of a Dispar; part of the breast of mutton.
FINJY—When some one of a number of boys had something
unpleasant to do, he who said “Finjy” last had to do it.
FLESHY—The name of a Dispar; a thick cut out of the middle of a
shoulder of mutton.
FORICUS—Latrine.
FOUNDER’S COM—The four days on which there were festivals in
commemoration of the Founder, when there was Amen Chapel,
the Fellows and Masters gave a dinner in Common-room, and
the Founders (v. inf.) received a sovereign each.
FOUNDER’S OB—The Anniversary of the Founder’s death.

FOUNDERS KIN—(see p. 55.)

FOUNDERS—Boys who proved their descent from the Founder, and

were afterwards elected (by vote among the Electors) as such.
Only two were admitted each year; and only two were sent to
New College, but these two were put at the head of the “Roll,”
(v. inf.,) whatever their previous position in Sixth Book might
have been. They were not obliged to leave at the age of
eighteen, as the other boys were, but were allowed to remain till
they were twenty-five. They were supposed to have particularly
thick skulls.
FOUR-HOLED MIDDLINGS—Ordinary walking shoes
FOURTH BOOK—All the boys below Junior Part the Fifth.
FRAGMENT—A private dinner-party given in Hall to a certain
number of boys, by the Warden, or one of the Masters or
Fellows.
FRATER—Brother.
FROUT—Angry.
FUNCTIOR—The rushlight in each chamber.
FURKED—To be expelled.

G.
GAIN A YEAR BY ELECTION—A boy (not being a Founder) was
obliged to leave at the Election immediately succeeding his
eighteenth birthday; he whose birthday came shortly after
Election, was thus enabled to stay till he was nearly nineteen,
and was so said “to gain a year.”
GAGS—Slang name for
GATHERINGS—Criticisms on some Greek or Latin author, written in
Latin by Sixth Book and Senior Part, eight times in the year. In
the other Parts an analysis of some history, in English, was so
called.
GATER—A spring head-foremost into “Pot,” (v. inf.,) over one of the
projecting handles of the canal lock-gate.
GATES—When the boys were assembled together in Seventh
chamber passage, preparatory to going on to Hills or Cathedral,
they were said “to be at Gates.”
GLOPE—To spit.