Engineering Science at the University of

Oxford, where he researches jet-engine and

rocket technology. His science puzzles have

appeared in the Guardian newspaper and been

featured on BBC Radio 4.

Entry to selective universities will often require students to demonstrate that

they have engaged in super-curricular activities which develop their awareness
and understanding of their subject. Tom Povey has provided in this book
enough super-curricular content to keep an aspiring mathematician, physicist,
engineer or material scientist (and their teachers) happy for months. For a
parent or teacher who wants to pro-actively support a student in preparing for
entry to a competitive science or maths course this is a no-brainer purchase.
Mike Nicholson, Director of Undergraduate Admissions and
Outreach, University of Oxford (2006-14), Director of Student
Recruitment and Admissions, University of Bath
This book is about the excitement of mathematics and physics at the school/
university boundary. It contains a remarkable collection of intriguing intellectual
challenges, all with complete solutions. You can read it to increase your ability
in maths and physics; you can read it to develop your problem-solving skills; but
most of all you can read it for fun. Professor Richard Prager, Head of the
School of Technology, University of Cambridge
This instructive book will allow aspiring young
scientists to test and expand their knowledge of
physics and mathematics, and gain greater
confidence in their abilities. Professor Sir Chris
Llewellyn Smith FRS, Director General
of CERN (1994-98)

ONEWORLD

MATHEMATICS / SCIENCE

jacket illustrations courtesy of the author

www.oneworld-publications.com

Professor Poveys Perplexing Problems T H O M A S P OV E Y

THOMAS POVEY is Professor of

T H O M A S P OV E Y

Professor
Poveys
Perplexing
Problems
Pre-University Physics and Maths
Puzzles with Solutions

In Professor Poveys Perplexing Problems, Thomas

Povey shares 109 of his favourite problems
in physics and maths. A tour de force of
imagination and exposition, he takes us by the
hand and guides us through uncompromisingly
challenging territory that expands our minds and
encourages a playful and exploratory approach
to study. The puzzles, he says, are like toys. We
should pick up the one we most enjoy, and play
with it.
Whether you are an aspiring scientist or an oldhand, pitting yourself against these problems will
test your ability to think, and inspire you with
curiosity and enthusiasm for physics. Presented
with charm and wit, the questions span the
gap between high-school and universityentrance standard material. Detailed answers are
lightened with a fascinating and refreshing blend
of scientific history, application and personal
anecdote.
On this delightful and idiosyncratic romp
through pre-university maths and physics, the
author shows us that behind every single one
of these questions lies a new way of thinking
about subjects we thought we had understood.
He argues that engaging with the unfamiliar is
key to forming deeper insights and developing
intellectual independence. Professor Poveys
Perplexing Problems is a manifesto that science
should be playful, and a celebration of the
curious.

25/$30

Praise for Professor Poveys Perplexing Problems

Entry to selective universities will often require students to demonstrate that
they have engaged in super-curricular activities which develop their awareness
and understanding of their subject. Thomas Povey has provided in this book
enough super-curricular content to keep an aspiring mathematician, physicist,
engineer or material scientist (and their teachers) happy for months. For a
parent or teacher who wants to proactively support a student in preparing for
entry to a competitive science or maths course this is a no-brainer purchase.
Mike Nicholson, Director of Undergraduate Admissions and Outreach,
University of Oxford (2006-14), Director of Student Recruitment and
Admissions, University of Bath

This book is about the excitement of mathematics and physics at the schooluniversity boundary. It contains a remarkable collection of intriguing intellectual challenges, all with complete solutions. You can read it to increase your
ability in maths and physics; you can read it to develop your problem-solving
skills; but most of all you can read it for fun.
Professor Richard Prager, Head of Cambridge University School of
Technology

This instructive book will allow aspiring young scientists to test and expand
their knowledge of physics and mathematics, and gain greater confidence in
their abilities.
Professor Sir Chris Llewellyn Smith FRS, Director General of CERN
(1994-98)

By challenging students with unfamiliar but engaging problems, I hope Professor Povey inspires many more budding physicists and mathematicians to
realise their potential.
Sir Peter Lampl, Founder and Chairman of the Sutton Trust

A real tour de force of scientific problems. Thomas Povey is clearly a master

at this kind of conundrum ... Fascinating.
John Farndon, author of Do You Think Youre Clever?

This book will amuse those who like puzzles. The topics vary from the practical to the quirky. Readers with a basic knowledge of maths and physics will
be entertained and encouraged by the interesting and challenging questions
they are able to solve.
Professor Dame Ann Dowling DBE FRS FREng, President of the Royal
Academy of Engineering

Professor Poveys
Perplexing Problems
Pre-University Physics and Maths Puzzles
with Solutions

Thomas Povey

A Oneworld Book

First published in North America, Great Britain and Australia by

Oneworld Publications, 2015

Copyright Thomas Povey 2015

The moral right of Thomas Povey to be identified as the Author of this work has been
asserted by him in accordance with the Copyright, Designs, and Patents Act 1988

All rights reserved

Copyright under the Berne Convention
A CIP record for this title is available from the British Library

ISBN 978-1-78074-775-0

Printed and bound by CPI Group (UK) Ltd, Croydon CR0 4YY

Oneworld Publications
10 Bloomsbury Street, London WC1B 3SR, England

To Mike Leask and Terry Jonesbrilliant teachers, insatiable adventurers.

With thanks to Tet Amaya, for substantial contributions throughout.

Contents
Introduction

Acknowledgements

An odd journey

Geometry
1.1 Shortest walk F . . . . . . . . . . . . . .
1.2 Intercontinental telephone cable F . . . .
1.3 Chessboard and hoop FF . . . . . . . . .
1.4 Hexagonal tiles and hoop FF . . . . . . .
1.5 Intersecting circles FF . . . . . . . . . . .
1.6 Cube within sphere F . . . . . . . . . . .
1.7 Polygon inscribed within circle F . . . . .
1.8 Circle inscribed within polygon FFF . .
1.9 Triangle inscribed within semicircle F . .
1.10 Big and small tree trunks FF or FFFF
1.11 Professor Fuddlethumbs stamp FF . . .
1.12 Captain Fistfulls treasure F . . . . . . . .
1.13 Captain Fistfulls treasure II FF . . . . .
1.14 Captain Fistfulls treasure III FFF . . . .
1.15 The geometry of Koch Island FFFF . .
1.16 An easyish fencing problem F . . . . . . .
1.17 A hardish fencing problem FFF . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

19
20
21
22
24
26
28
30
31
34
36
39
41
42
45
49
52
54

Mathematics
2.1 Human calculator F . . . . . . . . . . . . .
2.2 Professor Fuddlethumbs reports F . . . . .
2.3 More of Professor Fuddlethumbs reports F
2.4 Ant on a cube I FF . . . . . . . . . . . . .
2.5 Ant on a cube II FFF . . . . . . . . . . .
2.6 Ant on a cube III FFFF . . . . . . . . . .
2.7 A falling raindrop FF . . . . . . . . . . . .
2.8 The Three Door Problem FF . . . . . . . .

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

60
60
63
64
65
66
67
69
72

vii

CONTENTS

viii
2.9
2.10
2.11
2.12
3

Dr Bletchleys PIN FFF . . .

Mr Smiths coins FFF . . . .
The three envelope problem F .
A card game FF . . . . . . . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

74
76
77
79

Statics
3.1 Sewage workers conundrum F . . . . . . . . . .
3.2 Sewage workers escape FF . . . . . . . . . . . .
3.3 Sewage workers resolution FFF . . . . . . . . .
3.4 Aztec stone movers FF . . . . . . . . . . . . . .
3.5 The Wheel Wars I FFF . . . . . . . . . . . . . .
3.6 The Wheel Wars II FF . . . . . . . . . . . . . .
3.7 Obelisk raiser FF . . . . . . . . . . . . . . . . .
3.8 Obelisk razer FFFF . . . . . . . . . . . . . . .
3.9 The Ravine of (Not Quite) Certain Death FFF

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

82
83
85
89
91
96
100
101
108
115

.
.
.
.
.
.
. .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

119
122
124
129
132
136
141
147

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

150
151
153
155
158
162
165
170

Simple harmonic motion

6.1 Oscillating sphere FF . . . . . . . . . . . . . . . .
6.2 Professor Stopclocks time-manipulator FFF . . .
6.3 Dr Springloves Oscillator F . . . . . . . . . . . . .
6.4 Dr Springloves Infernal Oscillator FF . . . . . . .
6.5 Dr Springloves Improved Infernal Oscillator FFF

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

176
180
184
187
190
192

Mad inventions and perpetual motion

7.1 Stevins clootcrans F . . . . . . . .
7.2 Power-producing speed humps FF
7.3 The overbalanced wheel FF . . .
7.4 Professor Sinclairs syphon FF . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

196
197
201
203
209

Dynamics and collisions

4.1 Pulleys F . . . . . . . . . . . . . . . . . . . . . . .
4.2 Dr Lightspeeds elastotennis match FF . . . . . . .
4.3 Accelerating matchbox FFF . . . . . . . . . . . .
4.4 The last flight of Monsieur Canard FF . . . . . .
4.5 Water-powered funicular F . . . . . . . . . . . . .
4.6 Sherlock Holmes and the Bella Fiore emerald FFF
4.7 Equivalent statements for linear collisions FFF . .
Circular motion
5.1 Friction at the superbike races F . . . . . .
5.2 Pole position at the superbike races FF . . .
5.3 Roller coaster FFF . . . . . . . . . . . . .
5.4 Derailed roller coaster FF . . . . . . . . . .
5.5 The last ride of Professor Lazy FF . . . . .
5.6 Wall of Death: car FFF . . . . . . . . . .
5.7 Wall of Death: motorcycle FF or FFFF

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.

CONTENTS
7.5
7.6

ix

Boyles perpetual vase F . . . . . . . . . . . . . . . . . . . .

The curious wheel FFF . . . . . . . . . . . . . . . . . . .

214
217

Kinematics
8.1 Professor Lazy FF . . . . . . . . . . . . . . . . . . . . . . .
8.2 The Unflinching Aviator F . . . . . . . . . . . . . . . . . .
8.3 Target shooting F . . . . . . . . . . . . . . . . . . . . . . .

219
219
222
225

Electricity
9.1 Resistor pyramid FF .
9.2 Resistor tetrahedron F .
9.3 Resistor square FFFF
9.4 Resistor cube FFF . .
9.5 Power transmission F .
9.6 RMS power F . . . . .
9.7 Boiling time F . . . . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

228
230
234
235
237
239
240
240

10 Gravity
10.1 The hollow moon F . . . . . . . . . . . . .
10.2 Lowest-energy circular orbit FF . . . . . .
10.3 Weightless in space FF . . . . . . . . . . .
10.4 Jump into space FFF . . . . . . . . . . . .
10.5 Space graveyard FFF . . . . . . . . . . . .
10.6 Newtons cannonball FF . . . . . . . . . .
10.7 De la Terre la Lune FF . . . . . . . . . .
10.8 Professor Plumbs Astrolabe-Plumb FFF .
10.9 Jet aircraft diet FF . . . . . . . . . . . . . .
10.10 Escape velocity from the Solar System FFF
10.11 Mr Megalopolis expanding Moon FFF . .
10.12 Asteroid games FFFF . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

242
243
245
247
250
254
259
263
265
269
272
276
279

11 Optics
11.1 Mote in a sphere F . . . . . . . . . . . . .
11.2 Diminishing rings of light FFF . . . . . .
11.3 Floating pigs FFF . . . . . . . . . . . . .
11.4 The Martian and the caveman F or FFFF
11.5 Strange fish FFF or FFFF . . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

285
285
286
288
293
298

12 Heat
12.1 The heated plate F . . . . . . .
12.2 The heated cube F . . . . . . .
12.3 Fridge in a room FF . . . . . .
12.4 Ice in the desert FFF . . . . .
12.5 The cold end of the Earth FF .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

305
306
308
310
314
318

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

CONTENTS

x
13 Buoyancy and hydrostatics
13.1 Archimedes crown and Galileos balance F .
13.2 Another Galileos balance puzzle FF . . . .
13.3 Balanced scales F . . . . . . . . . . . . . . .
13.4 The floating ball and the sinking ball FF . .
13.5 Floating cylinders FFF . . . . . . . . . . .
13.6 The hydrostatic paradox FF . . . . . . . .
13.7 A quantitative piston puzzle F . . . . . . .
13.8 The floating bar FFFF . . . . . . . . . .

.
.
.
.
.
.
.
.

14 Estimation
14.1 Mile-high tower F . . . . . . . . . . . . . . .
14.2 How long do we have left? FF . . . . . . . .
14.3 Midas storeroom F . . . . . . . . . . . . . .
14.4 Napoleon Bonaparte and the Great Pyramid F
14.5 Lawnchair Larry FF . . . . . . . . . . . . .
14.6 Do we get lighter by breathing? FF . . . . .

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

322
323
328
331
332
333
336
338
340

. . . .
. . . .
. . . .
. . .
. . . .
. . . .

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

351
352
355
357
358
360
365

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

The Deadly Game of Puzzle Points

367

Endnote

374

Introduction
This is a personal book which celebrates a passion I have always had for playful
problems in physics and maths. Questions that rely on material no harder
than that studied in high school, but which encourage original thinking, and
stretch us in new and interesting directions.
In practically every country that has an established higher education system, the most competitive universities operate some kind of admissions test
or interview. These generally have only one object: to distinguish from many
brilliant applicants those with the greatest potential. To test this potential,
questions are often designed to test an applicants ability to think creatively.
In many countries, these pre-university tests are set by university professors.
Whilst the questions might reference a particular high-school syllabus, they
would not normally be constrained by it. The questions are often deliberately
off-beat. This is to see if applicants can apply what they have already learnt in
new and more challenging situations. The questions test the applicants ability
to pick an unusual problem apart, reduce it to its essential components, and
use standard tools to solve it.
This book is a collection of some of my favourite pre-university problems
in physics and maths. These questions are devised to encourage curiosity and
playfulness, and many are of the standard expected in some university entrance
tests. These questions are perplexing, puzzling, butmost of allfun. You
should regard them like toys. Pick up the one that most appeals, and play
with it. When you have exhausted it, you can entertain a friend with it. It
might seem impossibly nerdy, but for me almost nothing is as enjoyable as
being baffled by an apparently simple problem in an area of classical physics I
thought I understood. This book is a way of sharing the pleasure I have taken
in some of the physics and maths puzzles I have most enjoyed. The questions
will appeal to those high-school students who have mastered the basics, and
feel they have room to play a little with something more unusual. Teachers can
also use the problems to stretch their pupils with something more challenging
or unconventional. Here I am grateful to those teachers who allowed their
classes to pilot these problems, and to the students who bravely attempted
them.
I invented many of the questions myself. Others were suggested or inspired by friends and colleagues with an interest in this book. Still more are
1

INTRODUCTION

well-known classics. Even working outside a syllabus however, it is hard to be

entirely original with questions in physics and maths. Although these questions can be asked in essentially unlimited ways, the number of fundamental
concepts underpinning them is relatively small. Even the ones I am most
proud of will no doubt have been asked in numerous other forms over the
years.
There was one problem I was unable to solve. Many of these questions
especially the harder oneswork best as discussion problems. They are not
meant to be asked and answered straight. Instead they should be discussed
as part of a tutorial-style conversation, with an experienced tutor helping the
student along, giving hints, correcting mistakes, providing encouragement,
and so on. There is no real substitute for this kind of help. A question which
might seem impenetrable can suddenly yield when a tutor provides just the
right hint. I would encourage you to try the problems with friends, or subject
tutors, or anyone with a sound knowledge of physics and maths.
I should say a few words about the level of difficulty of the problems.
Every single one of them is challenging in some way. The publisher asked me
to use stars to indicate their relative difficulty. Although I think this is quite
subjective, I had to concede that it would be a shame for someone to happen
to try all the most difficult problems first and feel defeated. So I have indicated
how hard I found the problems the first time I tried them. If you think I have
rated some of them too generously, it simply means your intuition is better
than mine in that particular area. In general, those questions with one or two
stars should yield to a student who is on top of the pre-university high-school
syllabus1 in physics and maths. The following descriptive gradings should give
you some idea of what to expect.
F Not too difficult. Requires thought, and some insight, but
should be soluble without hints.
FF Challenging. Requires considerable thought, and some insight, and may require simple hints.
FFF Hard. Requires significant thought, and considerable insight, and more substantial hints. In some cases, questions with
three stars are only suitable as discussion problems.
I have also included a handful of exceptionally hard questions, to give the most
confident students an opportunity for a real challenge.
FFFF Exceptionally hard. Suitable for discussion. May be
too difficult for most students to solve without substantial help.
Complex and unusual, these questions allow for different and nonstandard approaches.
1 Here I mean a student who has completed the first year of A-level, International Baccalaureate, Scottish Highers, SAT, etc., and is about a year away from entering higher education.

INTRODUCTION

This book has been what I call my Saturday project, and there are, I increasingly discover, always fewer Saturdays than I hope there will be. I have not
had the luxury of someone to work with, so there will certainly be mistakes,
and errors of logic. They are all my fault. But I hope they are few enough that
this book brings more enlightenment than confusion. The most challenging
aspect of putting it together was striking the right balance with the difficulty
of questions. The questions range from being quite straightforward to really
quite hardthey are as unpredictable as many university entrance tests. Here
I am indebted to the many teachers, students and colleagues who were kind
enough to provide detailed comments on the difficulty of the questions. I
would be delighted to receive further constructive notes and any genuinely
unusual questions which could improve a future edition of this book, and
thank in advance anyone willing to provide this feedback.2 I hope you will
find friends you can share the fun of these questions with, and argue the case
for improved or more elegant solutions (many of which exist). To share your
solutions with a wider audience, and to learn from the ideas of others, you can
contribute to the Puzzle Forum at
www.PerplexingProblems.com
If you are feeling brave, you can also play The Deadly Game of Puzzle Points,
which is explained in the final chapter of this book. Ive provided scoresheets
and ticklists so you can keep track of your progress.
Finally, good luck, and have fun.
T.P., Oxford, 2015
2 You

can contact me by emailing [email protected]

Acknowledgements
I am indebted to all the teachers, headteachers, students, parents of students,
and colleagues who provided many useful comments, particularly on the difficulty and range of questions that were eventually included in this book.
Among those who were kind enough to provide feedback were: Oscar Van
Nooijen, Brian Smith, Terry Jones, Frances Burge, Julie Summers, Robin Parmiter, Elaine Cook, Georgina Allan and Glenn Black. With thanks to Ben
Sumner for copyediting the manuscript.

Chapter 1

Geometry
In this chapter we look at problems in geometry.1 This ancient branch of
mathematics dates back to approximately 600 BC, but advanced most rapidly
around the time of Euclid,2 who developed results into a logical system based
on a set of simple axioms.3 Well over two thousand years later, what is now
known as Euclidean geometry4 is still taught in schools. It is interesting to
think that when we solve problems in this area we often do so with exactly
the same logic, tools and methods as the ancient Greeks used. So we are concerned with developing simple but clever methods for geometric calculations
(areas and volumes, for example) without recourse to integration. That is,
without the need to represent the geometry with explicit functions which are
dealt with using calculus. The latter was developed two thousand years after
Euclidin the seventeenth centuryby Isaac Newton and Gottfried Leibniz.
Some of these puzzles are well-known, others are based on well-known
questions, and still more I invented in the last few years. My personal favourite
is the chess board and hoop question, which I came up with three years ago,
and which students have found quite satisfying and enjoyable to solve.

1 Literally

translated from the ancient Greek as earth measurement.

of Alexandria (c. 300 BC) was a Greek mathematician most famous for Elements, a
collection of works on geometry, number theory and logic, said by some to be the most influential
mathematical work ever written. Euclid is known as the Father of Geometry.
3 An axiom is a premise that is assumed to be so simple and obvious that it is self-evidently
correct. Euclids five axioms were as follows: i) A straight line can be drawn from any point to
any other point; ii) a line segment can be extended continuously in both directions; iii) a circle
can be drawn with any centre and any radius; iv) all right angles are equal; v) if a straight line falls
on two straight lines and makes interior angles on the same side that sum to less than the sum
of two right angles, the two straight lines, if extended indefinitely, intersect on that side (this is
known as the parallel postulate).
4 Non-Euclidean geometries, such as hyperbolic and elliptic geometries, were developed in the
nineteenth century by relaxing the parallel postulate.
2 Euclid

19

CHAPTER 1. GEOMETRY

20

1.1

Shortest walk F

An ant starts at one vertex of a solid cube with side of unity length. Calculate the distance of the shortest route the ant can take to the furthest vertex
from the starting point. There is more than one solution to this problem.

Answer
This is a well-known question, though not one with very much scope for
discussion.
If we consider unwrapping any pair of sides that connect the starting vertex A to the furthest vertex from the starting vertex, B, it is clear that the
unwrapped surface forms a rectangle with
sides of length 1 and 2. The shortest
distance is a straight line with length 5.

This solution is not unique. There are, in fact, six solutions of the same minimum length, which are shown in this diagram of the unwrapped cube.

1.2. INTERCONTINENTAL TELEPHONE CABLE F

1.2

21

Intercontinental telephone cable F

This rather neat little question is simple enough that it does not warrant much
introduction. I have heard of a number of variants over the years.
A telephone company has run a very long telephone cable all the way round
the middle of the Earth. Assuming the Earth to be a sphere, and without
recourse to pen and paper, estimate how much additional cable would be required to raise the telephone cable to the top of 10 m tall telephone poles.

Answer
The circumference of the Earth is given by C = 2RE , where RE is the
radius of the Earth. The new circumference at the top of 10 m tall telephone
poles is given by C 0 = 2 (RE + 10). The difference in the circumferences is
C 0 C = 20. We should know that 3.14, so we should be able to work
out in our heads that the additional length of cable is approximately equal to
62.8 m. The question is designed to be slightly tricky in that we might be
expecting a big number. If we approach it rationally we see that questions
dont come much easier than this!

There are, of course, intercontinental telephone cables, and the history of

submarine cable-laying started not long after the invention of the telegraph.
The first commercial telegraph was developed by William Cooke and Charles
Wheatstone in 1838, and as early as 1840 Samuel Morse was working on the
concept of submerged cables for transatlantic operation. Experiments under

CHAPTER 1. GEOMETRY

22

the waters of New York Harbour in 1842 confirmed the feasibility of such
a project, and in 1850 the first telephone cable was laid under the English
Channel (to France), by the Anglo-French Telegraph Company. Numerous
other submerged telephone cables were laid across the English Channel, the
Irish Sea and the North Sea in the years that followed, and the first attempt
at a transatlantic cable was made in 1858. The first successful transatlantic
project was in 1865, when undersea cables were laid by the SS Great Eastern,
a huge iron steamship designed by the Bristolian Isambard Kingdom Brunel.
Now there are numerous fibre-optic cables snaking across the Atlantic seabed.

1.3

Chessboard and hoop FF

I invented this puzzle a couple of years ago. It is one of my favourites because

students seem to enjoy it, and can generally make good progress without much
help, although sometimes they need a hint to get started.
A thin hoop of diameter d is thrown on to an infinitely large chessboard with
squares of side L. What is the chance of the hoop enclosing two colours?

Answer
This question can be answered by applying very simple geometry, and should
be accessible to anyone with a good grasp of GCSE-level maths. The first
realisation is that if the hoop crosses any tile boundary it will enclose two
colours. It is easier to first pose the reverse of the question askedthat is,
what is the chance of not enclosing two colours? We then subtract this
chance from the chance of any outcome, giving the answer to the question
as posed. We will use notation familiar to students of probability, although

1.3. CHESSBOARD AND HOOP FF

23

only the most elementary understanding of that subject is required here. We

use the notation P (A) to denote the probability of the hoop enclosing two
colours. By definition 0 P (A) 1; that is, the probability lies somewhere
to
between 0 (will not happen) and 1 (will definitely happen). We use P (A)
denote the probability of the hoop not enclosing two colours. There are only
= 1 P (A).
two possible outcomes, so P (A)
We are dealing with two-dimensional space, so we will probably need to
define the landing areas which lead to one outcome or the other. If we were
dealing with three-dimensional space we would need to define volumes. In
one-dimensional space, we would need to define distances, and so on. We can
define the location of the hoop unambiguously by referring to the position of
the centre of the hoop. We could equally use the location of any identifiable
point on the rim of the hoop, but then we would also need to express the
angular orientation of the hoop. This introduces more complexity with no
benefit.
We will now consider three cases: d > L; the special case d = L; and
d < L. Lets consider each in turn.

The Case of d > L. In this case, in which the diameter is larger than
the side of a tile, it is perfectly obvious that the hoop must enclose two
colours. We see that P (A) = 1.

The Special Case of d = L. For the special case that d = L, it is

possible for the hoop to land entirely within a single tile, but the chance
of this happening is 1/. In this limiting case the circle touches the tile
at four points. That is, a single solution out of infinitely many possible
solutions satisfies the condition that only one colour is enclosed. We
= 1, and therefore P (A) = 0.
write P (A)

The Case of d < L. Consider all the circles that can be drawn entirely
within a single tile, and which touch the tile boundary either at one
point (along an edge of the tile) or at two points (in the corner of the
tile). For now we observe that the circles that touch the edges of the tile
at either one or two points have centres which lie on a smaller square
exactly in the middle of the larger tile. The locus of points is shown in
the diagram. For a circle of diameter d, the shortest distance between the
locus of points and the tile is d/2 everywhere. The side of the smaller
square is therefore L d.

CHAPTER 1. GEOMETRY

24

We now observe that any circle with a centre within the smaller square
does not touch the tile boundaries. The locus of all points at which the
hoop centre can be without touching the border of the tile is defined by
the smaller square. The probability of the hoop centre landing in the
smaller square is given by the ratio of the area of the smaller square to
that of the larger square.
=
P (A)

(L d)2
for d L
L2

So
P (A) = 1

1.4

(L d)2
for d L
L2

Hexagonal tiles and hoop FF

An infinitely large floor is tiled with regular hexagonal tiles of side L. Different colours of tiles are used so that no two tiles of the same colour touch.
A hoop of diameter d is thrown onto the tiles. What is the chance of the hoop
enclosing more than one colour?

1.4. HEXAGONAL TILES AND HOOP FF

25

Answer
We can answer this question using a similar method to that used for the chessboard and hoop, but the geometry is a little more complex. We start by defining the locus of hoop centres that allows a hoop entirely enclosed within a tile
to touch the edge of the tile at one position (edges) or two positions (corners).
The locus of such points is defined by the line CDEF GH, which describes a
smaller hexagon centred on the tile centre. The edges of the smaller hexagon
are d/2 away from the edges of the large hexagon. That is, the perpendicular
distance between AB and CD, for example, is d/2. The locus of all hoop
centres for which the hoop does not intersect the edge of the tile is defined by
the area CDEF GH.

To calculate the area of the smaller hexagon we need the length of the side CD,
which is denoted by |CD|. From the geometry of the regular hexagons
we can


see that |CD| = |AB| 2x = L 2x, where x = (d/2) tan 30 = d 3/6 .

We have |CD| = L d 3/3 .

is given by the ratio of

The chance of landing entirely within a tile, P (A),
the areas of the smaller tile and the larger tile. We do not need to calculate

CHAPTER 1. GEOMETRY

26

the areas explicitly, because we know they are of similar shape, so the area is
proportional to the square of any characteristic length. For the characteristic
length we could take the side length. The relevant lengths become |CD| and
|AB|. This gives


=
P (A)

|CD|
|AB|

2
d 3
3

L2

for d

3L

The probability of enclosing more than one colour, P (A), is given as P (A) =
or
1 P (A),

P (A) = 1

1.5

2
d 3
3

L2

for d

3L

Intersecting circles FF

A friend of mine introduced me to this puzzle some years ago. I like it because
it is deceptively simple-looking, but it actually took me some time to solve it.
I think I simply started down the wrong track, and took some time to find my
way back to a sensible method. Hopefully you will find it easier than I did.
Two circles of radius l and 2l intersect as shown. What is the area of the
shaded region?

Answer
Consider first the construction of points A, B, C, D, E and F in the sketch
below. We note first that the distance between A and C is equal to l, because
AC forms a radius of the smaller circle. We write |AC| = |AD| = l. A similar
statement leads to |BC| = |BD| = 2l.

1.5. INTERSECTING CIRCLES FF

27

The area of the shaded element CEDF C is equal to the area of sector ADF CA
plus the area of sector BCEDB minus the area of the quadrilateral BCADB.

We represent these areas by A1 to A4 respectively, and write

A1 = A2 + A3 A4
Expressing this explicitly we have
|CD|
2
To simplify this we require expressions for |AB|, |CD| /2, and . We now
find each of these in turn.
A1 = l2 + 4l2 |AB|

Find |AB|. We see

that AB is the longest diagonal of the square, and
has length |AB| = 2 2l.
Find . Using the cosine rule5 on the triangle ADBA we see that
 2
 
2
(2l) = l2 + 2 2l 2 (l) 2 2l cos
Simplifying, we get
5 Usually written in the form c2 = a2 + b2 2ab cos C, where a, b and c are the lengths of
the sides of a triangle, and C is the angle between the sides of length a and b.

CHAPTER 1. GEOMETRY

28

4 = 1 + 8 4 2 cos
That is

5 2
cos =
8

Find . We find using an identical method to the one we used to find

, and get

11 2
cos =
16
Find |CD| /2. We set |CD| /2 = h, and note h in the construction
ADBA. From the construction we see that

p
l 14
2
h = l sin = l 1 cos () =
8
Now that we have expressions for |AB|, |CD| /2, and , we can evaluate
A1 . We write
"
! #
!
11 2
7
5 2
1
2
1
A1 = l cos
+ 4 cos

8
16
2
This gives the area of the shaded region as
A1 0.108 l2
My original method was considerably longer even than this! I like this problem because it is apparently trivial, but it takes some thought to solve with
relatively compact algebra.

1.6

Cube within sphere F

What is the volume of the largest cube that fits entirely within a sphere of
unity volume?

1.6. CUBE WITHIN SPHERE F

29

Answer
By symmetry, all eight vertices of the cube must touch the sphere. Again,
by symmetry, the length of the longest diagonals of the cube, for example
|ab|, must be equal to the diameter of the sphere D, because the line ab passes
through the centre of the sphere.

From the geometry of the cube, using Pythagoras theorem

2

|ab| =

2
2L + L2

where L is the length of the side of the cube. Thus

|ab| = 3L = D

The volume of the cube is thus given by

D3
33/2
3
All that remains is to substitute for D using the equation for the volume of
the sphere, VS = (4/3) r3 = D3 /6 = 1. Rearranging D3 = 6/. The
volume of the cube within the sphere is given by
VC = L3 =

VC =

2
0.368
3

CHAPTER 1. GEOMETRY

30

1.7

Polygon inscribed within circle F

What is the area of an n-sided regular polygon inscribed within a circle of

radius r?

Answer
This is a straightforward problem really, but one that is a good test of a students ability to derive simple formulae for unfamiliar situations. There is relatively little scope for practising this skill in most high-school courses, so Ive
included a number of problems of this type to develop this aspect of problemsolving.
Consider the constructions for n = 3, 4 and 5-sided polygons.

The polygons are regular, so using their symmetry we can decompose them
into 2n right-angled triangles, each with area (xy) /2. The total area of the
n-sided polygon becomes
xy
= nxy
2
Noting the angle in each of the constructions, we see that sin = x/r and
cos = y/r, where = (2) / (2n) = /n. The area of the polygon now
becomes
An = 2n

cos
n
n
We can simplify this using the trigonometric identity 2 sin A cos B = sin (A + B)+
sin (A B). We get
An = nr2 sin

1.8. CIRCLE INSCRIBED WITHIN POLYGON FFF

An =

31

nr2
2
sin
2
n

As a final note, we see that, for very large n, we can make the approximation
sin (2/n) 2/n. The expression for area becomes
An r2 (for large n)
This is the equation for the area of a circle, as expected.

1.8

Circle inscribed within polygon FFF

For a circle inscribed inside a regular n-sided polygon, what is the minimum
n so that Ashaded /Acircle 1/1000?

This is a variant on the previous question, but has the added problem that the
final expression cannot be solved by simply rearranging the equation. You can
use a number of techniques familiar to high-school students, however. And,
unusually, the suggested methods are now given as hints.
One method is to use a graphical solutionthat is, to find two expressions
so that the intersection defines the solution. A second method is an iterative
numerical techniqueessentially trial and error of a thoughtful sort, in which
you see if you are getting closer or further from the solution and correct your
guess up or down accordingly until you converge on a solution. A third technique, and certainly the most elegant, is to use a Taylor polynomial approximation to the function. Taylor series are no longer taught in all high-school
syllabuses but are certainly taught in some extension papers, and are covered
in the first year of all university courses. Because the Taylor approximation is
such an elegant technique we will say a few words about it here.
The first thing to say is that the concept of a Taylor series is very simple,
so there is no need to be intimidated by it. It says we can represent a function about a particular point as an infinite polynomial series. If, for example,

CHAPTER 1. GEOMETRY

32

we wish to represent sin x around zero, we start with a linear (or first-order)
approximation that only applies at very small angles, namely
sin x x for very small x
This is an approximation that all high-school students should be familiar with,6
and will use routinely. As we move away from zero the approximation starts
to fail because, as we well know, the graph of y = sin x has curvature as we
move away from the origin. To correct for the curvature, we need to add a
higher-order term which itself is approximately zero when very close to the
origin, so as not to harm our first-order approximation, but which gets larger
(either positive or negative in sign) at a greater rate than x as we move away
from the origin. We skip the second-order term because in this series it is zero,
so the third-order approximation7 is
sin x x

x3
3!

where 3! stands for three factorial, or 3 2 1 = 6. The factorial8 of any

integer n is simply the product of all positive integers less than or equal to n.
The third-order approximation extends the region around the origin in which
the approximation is accurate. We can keep adding terms to get what is called
the Taylor series for sin x, which is
sin x = x

x3
x5
x7
x9
+

+
...
3!
5!
7!
9!

and so on with infinite terms. Actually, the ninth-order approximation (the

first five non-zero terms in this case) is very accurate up to x = . Try it.
The general process by which the series approximation terms are derived
for a particular function is due to Brook Taylor, a mathematician who worked
on what later became known as differential calculus. He published this, his
most famous result, in 1712.
6 Heres a simple geometric proof. If we consider a thin sector of a circle of angle and chord
length l, we see that (in radians) we have l = r. The perpendicular distance h is given by
h = r sin . We see from the geometry of the sketch that for very small , h l, and therefore
sin .

7 By

n-th order we mean including terms up to xn .

factorial is equal to unity. The proof is as follows. Noting that (n + 1)! = (n + 1)
n ... 2 1 = (n + 1) n!, and setting n = 0, gives 1! = 1 0!, so 0! = 1! = 1.
8 Zero

1.8. CIRCLE INSCRIBED WITHIN POLYGON FFF

33

The Taylor series for cos x and tan x are as follows

cos x =
tan x =

x2
x4
x6
x8
+

+
...
2!
4!
6!
8!
2
17 7
62 9
1
x +
x ...
x + x3 + x5 +
3
15
315
2835

Answer
Consider the constructions for n = 3 and 4-sided polygons.

The polygons are regular, so using their symmetry we can decompose them
into 2n right-angled triangles, each with area (xr) /2, where r is the radius
of the circle, and x is the half-length of the side of the polygon, given by
x = r tan , where = /n. The total area of the n-sided polygon becomes
An = nxr = nr2 tan

Consider that
Ashaded
An
n

=
1 = tan 1
Acircle
Acircle

n
So we require the minimum n that satisfies
n

1
tan 1

n
1000
Here we must choose either a graphical, numerical (iterative) or series solution
approach. Let us use the series solution, which is the most elegant, although
trying the problem in other ways is also instructive. We need to approximate
tan (/n) using the Taylor series for tan x. Taking a third-order approximation we have
1
tan x x + x3
3

CHAPTER 1. GEOMETRY

34
Our inequality reduces to



n 1  3
1
1
+
n 3 n
1000
That is
1
1  2

3 n
1000
p
This gives n 1000/3 = 57.4. We require n to be an integer, so the
minimum n that satisfies this condition is n = 58, an octapentacontagon, or,
in the language most mathematicians use, a 58-gon.
Because we have used Taylor series approximations it is very important to
check this solution directly, to see if it is correct. We do so by recalling the
condition (n/) tan (/n) 1 1/1000, and evaluating this explicitly for
n = 57 and n = 58. We get

57
1
tan
1 = 1.014... 103 

57
1000

58
1
tan
1 = 9.791... 104

58
1000
We see that this agrees with our result.

1.9

Triangle inscribed within semicircle F

This little problem is an exercise in using vectors.

Give a vector proof that for a triangle inscribed within a semicircle, the
included angle is always /2.

1.9. TRIANGLE INSCRIBED WITHIN SEMICIRCLE F

35

Answer
Consider the semicircle AP B, formed by the intersection of a line AOB with
a circle. From the centre of the circle, if B is represented by the vector9 a,
then A is represented by the vector a. Let point P be represented by the
vector p.

The line AP is represented by the vector p + a, and the line BP is represented

by the vector p a. For two general vectors A and B, the dot product is
defined by A.B = |A| |B| cos , where is the angle between the vectors.
Thus, for = /2, A.B = 0. We recall two properties of the dot product. It
is commutative (that is, A.B = B.A) and distributive (that is, A.(B + C) =
A.B + B.C ).

Consider the dot product of the two vectors defining AP and BP . Expanding
using the distributive property we have
(p + a) . (p a) = p.p p.a + a.p a.a
Simplifying using the definition of the dot product and the commutative property, we have
2

(p + a) . (p a) = |p| |a|

We note that both a and p are radii of the circle, so

9 In this book we will use bold symbols to distinguish a vector a from its scalar magnitude |a|,
or a. In handwriting (the hand-drawn diagrams in this book, for example) vectors are denoted
with underscores (such as a).

CHAPTER 1. GEOMETRY

36

|a| = |p|
We see that the dot product (p + a) . (p a) = 0. However, |p + a| =
6 0 and
|p a| 6= 0. So we conclude from the definition of the dot product that for
(p + a) . (p a) = 0 to be satisfied we must have = /2. Thus, a triangle
inscribed within a semicircle is always a right-angled triangle.
A friend who read this question pointed out, quite rightly, that easier solutions exist that dont require vector notation. He objected to the use of vectors
on the basis that they over-complicate the analysis. He was right, of course,
but I still think its a nice exercise for those not entirely familiar with using
vectors. I did feel obliged to include the alternative solution,10 however.

1.10

Big and small tree trunks FF or FFFF

Quite a number of maths questions are designed to test a students ability

to set up equations for unfamiliar problems and then solve them. This is an
extremely useful skill for a numerate person. The question here is one of several similar problems I came up with on a long-haul flight to the US a couple
of years ago. The geometry is trivial, but I think most high-school students
would struggle with it, simply because the ability to approach completely unfamiliar situations is valued too little in most curricula.
A tree trunk has a circular cross section of radius n, where n is an integer. A
second circular tree trunk has radius m, where m is also an integer. Both tree
trunks rest on the ground, defined by y = 0, with their axes aligned with the
z-axis.
1. Derive the general condition that permits both tree trunk centres to lie
at integer co-ordinates along the x-axis.
2. Give the three smallest solutions for n when m = 1.
10 Noting the construction AP BOA, we see that = 2 and that = 2. Adding
these expressions we get + = 2 2 ( + ). But + = . Combining these expressions
we get = 2 2 ( + ). Simplifying, we get + = /2, as required. This solution is
undeniably simpler.

1.10. BIG AND SMALL TREE TRUNKS FF OR FFFF

37

3. Give a general solution for n and m (extension for FFFF).

Answer
This is a simple geometry problem that I think a lot of students would struggle
to unpack. Lets consider each of the three parts in turn.
1) Derive the general condition that permits both tree trunk centres
to lie at integer co-ordinates along the x-axis
We are told that n and m are integers, so the difference in the y co-ordinates
of their centres is also an integer, n m. We require that the difference in the
x co-ordinate also be an integer. We denote the difference simply as x. We
note that the tree trunks touch at a mutual point of tangency, so the distance
between their centres is n + m.

From Pythagoras theorem we have

2

(n + m) (n m) = x2
Simplifying, we get
4nm = x2
This is the general condition that must be met.
2) Give the three smallest solutions for n when m = 1.
A mathematician would hardly regard it as elegant, but to obtain the first three
solutions for n when m = 1 we can solve for m = 1 simply by substituting
this into the expression and trying values for n.
for m = 1, n =

x2
4

CHAPTER 1. GEOMETRY

38

We can probably see the first three solutions to this. If we cant, we can simply
try the first few integer values for x. We have
x
1
2
3
4
5
6

x2
1
4
9
16
25
36

n = x2 /4
1/4
1
9/4
4
25/4
9

Integer?
N
Y
N
Y
N
Y

The first three solutions are n = 1, 4 and 9.

3) Give a general solution for n and m (extension for FFFF).
Consider
the general condition that we are required to meet, namely 4nm =
x2 , or 2 nm = x. We know that x is an integer. We write this as x Z. Here
means is a member of, and Z represents the
set of all integers. By extension

11
2 nm Z. It is possible to prove that k Z if and only if k is a square
number.12 From this it follows that there are two possibles cases:
n or m is equal to unity and the other is a square number.
m = a2 b and n = bc2 , where b is not a square number and where a, b
and c are integers.13

That
k Z if and Only if k is a Square Number.
Theorem: k is an integer if k is a square number, otherwise it is irrational (it cannot be
expressed as the ratio of two integers). Here k Z+ ; that is, it is a member of the set of all
positive integers.

2
Proof: a) If k is a square number
we write k = l k = l Z; b) We assume that
k is a rational number, that is, k Q, and prove that this is impossible by contradiction.
Consider two sets of numbers {ai } and {bi }
in which members are positive integers: ai Z+
and bi Z+ . We assume that we can write k = ai /bi , where (ai , bi ) is an ordered pair of
positive integers. There must be a unique least element a {ai }. Similarly, unique least element
b {bi } exists. The least element of {ai } must pair up with the least element of {bi }. Hence
k = a/b, so k = a2 /b2 , so b2 k = a2 . Because a, b, k Z+ , and because the left-hand side is
divisible by k, the right-hand side must also be divisible by k, so
kc (provided
we can write a = +
that k is not a square
number,
otherwise
we
could
write
a
=
kc),
where
c

Z
. We write

b2 k = (kc)2 k = b/c. But fork Z+ and k 6= 1, k > 1. Therefore, from k = a/b,

we see that a > b. Similarly, from k = b/c, we see that b > c. So we have found k = b/c
with c < b < a. However, a and b were the least elements of {a
i } and {bi } respectively, which
leads to a contradiction. {ai } and
{b
}
must
both
be
empty,
so
k cannot be written as a rational
i

number. We have proved that k

/ Q.
12 A square number or perfect square is an integer that is the square of an integer.
13 To Show That n and m Satisfy This Condition. If m = a2 b and n = bc2 , then mn =
a2 b2 c2 = (abc)2 . Where a, b, c Z, then abc Z, so mn = (abc)2 is a square number.
11 Proof

Chapter 3

Statics
In this section we look at problems in statics. By definition, statics is the
study of systems that are stationary, and non-accelerating. In these systems
the forces acting on individual masses must sum to zero, and the torques (or
moments) acting on a body must sum to zero. These are requirements for zero
acceleration (both linear and rotational). We generally look at the systems
at a particular instant, and resolve1 the forces so that the conditions for zero
acceleration are satisfied. We should state a few definitions more formally.
Torque or Moment. The magnitude of the torque (or moment) about
a particular point (or pivot) due to a force is the product of the force
and the perpendicular distance between the line of action of the force
and the point of interest. We refer to this distance as the lever arm of
the force. Torque has units of N m.
Equilibrium Requirement for Zero Linear Acceleration. For an
object of finite mass, for the linear acceleration in a particular direction to be zero, the forces in that direction must sum to zero. For the
simplest
two-dimensional
Cartesian system, with axes x and y, we write
P
P
Fx =
Fy = 0. This condition must hold in every direction.
Equilibrium Requirement for Zero Angular Acceleration. Taking an object of finite moment of inertia2 about a particular pivot, A,
for the angular acceleration to be zero, the sum
P of torques, TAi , about
that pivot must be equal to zero. We write i TAi = 0. By the same
logic, given that every real body has a finite moment of inertia, the sum
of torques about any point in a system must be equal to zero, whether
or not that point lies on the body.
1 To

resolve, take the sum of vector components in a particular direction.

the same way as mass is a measure of the resistance of a body to linear acceleration, the
moment of inertia is a measure of the resistance of a body to angular acceleration. The moment
of inertia I of a body is defined about a particular point. For a body composed of mass P
elements
mi at distances ri from a point of interest A, the moment of inertia is defined by IA = mi ri2 .
2 In

82

3.1. SEWAGE WORKERS CONUNDRUM F

83

Centre of Mass (Gravity). The centre of mass (or gravity) of an object

is defined as the point through which any plane drawn would split the
object into two parts of equal mass. It can be thought of as the point
from which the weight of the body acts. It does not necessarily need to
lie on the object (consider a doughnut shape, for example).
Now we are ready to try a few problems in statics. Most of these questions are
my own invention, but a couple are modifications of well-known problems.

3.1

Sewage workers conundrum F

A sewage worker is inside a large underground aqueduct (diameter  a mans

height) of circular cross section, and so smooth that friction is negligibly small.
He has a ladder which is the same length as the diameter of the aqueduct. He
wishes to inspect something in the roof. He mounts the ladder and continues
to climb until he reaches the other end. What happens?

Answer
It hardly matters as far as solving the problem is concerned, but we might
start with the realization that the sewage worker must be at the lowest point
of the aqueduct. Were he anywhere else he would slide to the bottom because
the aqueduct is perfectly smooth. The ladder, interestingly, is equally happy
in any rotational position despite there being no friction. We could argue the
case by drawing forces on the ladder. Or we could simply note that it has the
same potential energy in all rotational positions. This must be so because the
centre of mass of the ladder is always at the centre of the tunnel: the length
of the ladder is equal to the diameter of the tunnel. Energy arguments can be
powerful, and we will use them in this question.
But what does happen next?
Well, as the man takes hold of the ladder and mounts it, the very first thing
that happens is that the ladder tips back slightly. Because the system is free to
move, it moves to the position with the lowest potential energy. Here we are

84

CHAPTER 3. STATICS

using the principle of minimum total potential energy.3 When we talk about the
system, we mean the ladder and the man. The potential energy of the ladder is
the same in every rotational position. The mans potential energy varies with
the angle he makes with the vertical, and is minimised when his centre of mass
is below the centre of the ladder, which can be thought of as a virtual pivot.4
The mans centre of mass hangs slightly away from the plane of the ladder,
and his centre of mass must be under the centre of rotation of the system for
it to be in equilibrium.

As he climbs and gets closer to the halfway point of the ladder, which is the
centre of rotation of the system, the ladder has to tip back more. When the
man approaches the middle of the ladder, the ladder must be almost horizontal, until his weight is directly under the centre of the cylinder and the
ladder is horizontal.

3 This rather impressive sounding principle simply tells us that systems that are free to move do
so until they are at the point of minimum potential energy. The lost energy is dissipated, as heat,
for example. A ball placed on the top of a hill will roll to the bottom of the hill. Water will move
until the free surface is perpendicular to the direction of gravity. The most stable configuration is
the one with minimum potential energy. This is the equilibrium position for the system.
4 If this is not perfectly obvious, consider that when the man is a given distance, say r, from
the midpoint of the ladder, his possible positions (for every angle his centre of mass makes with
respect to the vertical) are defined by a circle of radius r centred on the midpoint of the ladder.
The lowest point on that circle, or the point with the minimum potential energy, is below the
midpoint of the ladder.

3.2. SEWAGE WORKERS ESCAPE FF

85

As the man continues to climb along the ladder from the midpoint, the ladder
rotates past the horizontal position. The man climbs down to the bottom of
the cylinder, this time upside-down.

Eventually he is back to where he started but the wrong way up. It seems his
inspection of the roof will have to wait.

3.2

Sewage workers escape FF

A sewage worker is inside a large and smooth underground aqueduct of circular cross section and diameter d. To escape a tide of effluent, he balances on
the end of a ladder of length d/2 and mass m. If the sewage worker also has
mass m, at what angle does the ladder settle? Assume that the length of the
ladder is very much greater than the height of the man, so he acts as a point
mass on the end of the ladder.

CHAPTER 3. STATICS

86

Answer
We first need to consider the geometry of the problem. Triangle ABC is
equilateral, with side d/2. We might spot at this point, however, that the
answer is unlikely to depend on d/2, as it is determined by the angles of the
system. The angle at which the ladder settles we mark as in the diagram.
There are four external forces acting on the ladder-man system.
The weight due to the ladder, mg, which acts vertically downwards from
the centre of the ladder.
The weight due to the man, mg, which acts vertically downwards from
point B (we treat the man as we would a point mass at B).
The normal reaction force F1 , which acts perpendicular to the smooth
wall at point B.
The normal reaction force F2 , which acts perpendicular to the smooth
wall at point C.
Now we have the four forces we can mark them on the diagram, paying attention to the directions in which they act. We note that forces F1 and F2 both
act through the centre of the cylinder, along lines BA and CA respectively.

For the man-ladder system to be in static equilibrium, the forces in all directions have to balance. If we had a net force in a particular direction we would
have a non-zero acceleration in that direction according to Newtons Second
Law. By similar reasoning we require zero net torque (moment) on the manladder system, so that it is in rotational equilibrium.5 We are interested in the
5 Netwons Second Law tells us that the force in a particular direction F (where we note this
is a vector) is equal to the time rate of change of momentum in that direction, F = dp/dt. For
constant m we have F = m (dv/dt) = ma. To be in static equilibrium we must have zero
acceleration, dv/dt = a = 0. For this we require the sum of forces in every direction to be zero.
To be in static equilibrium we also require the angular acceleration to be zero: = d/dt = 0,
where is the angular velocity vector. For this we require the net torque on the object to be zero,
giving = I = 0.

3.2. SEWAGE WORKERS ESCAPE FF

87

vector forces, and the points of action of the forces, on the man-ladder system.
As long as we preserve the relative magnitudes, angles, and distances we can
consider the diagram in any rotated position. A slightly simpler representation of the man-ladder system is shown below. Note that the weight vectors
make an angle to the ladder, as they did in the original representation.

The standard procedure from here, which should be the mantra of any student
of statics, is to resolve, resolve, take moments. Let us do just that.
Resolve in the horizontal direction: 2mg sin = F1 cos 60 F2 cos 60
Resolve in the vertical direction: F1 sin 60 + F2 sin 60 = 2mg cos 60
Take moments about point B: 2F2 sin 60 = mg cos
Now we just need to solve these equations for , the angle of the ladder to
the horizontal. We should check that we have enough equations: we need as
many independent equations as we have unknowns. The unknowns are F1 ,
F2 and only three. We have three independent equations, so a solution
should be possible. Ive left the algebra as an exercise, as it should be fairly
straightforward from here. If you do it correctly you should get
16.10
The answer feels about right: the ladder is at a shallow angle to the horizontal.
Of course, if we wish, we can then also solve explicitly for F1 and F2 , giving

F1 = 3mg cos 1.66 (mg)

3
F2 =
mg cos 0.555 (mg)
3
Again, the answers feel about right. F1 is approximately three times larger
than F2 , giving greater reaction force at the end where the man is balancing,
as wed expect.

88

CHAPTER 3. STATICS

Further discussion
We did well to solve the last problem. We took a completely new situation,
reduced it to the underlying equations, then solved the equations quite simply
to get an exact result. The resolve, resolve, take moments method worked well.
But there are many other ways to tackle problems of this sort, and here we
demonstrate one that avoids the need to explicitly deal with forces and moments. It is a very common method in statics problems, and a very common
technique in studying systems in general. It is to consider the energy of the
systemspecifically, the point of minimum potential energy for the system
comprising the ladder and the man. Many systems that are free to move to a
lower potential energy point do move to that point. This is one such system.
The lowest potential energy state for this system is the one in which the
centre of mass is as low as possible. The centre of mass of the system is midway between point B and point E, or 1/4 of the way along the ladder from
point B. The lowest potential energy state is when point D is directly below
point A. If this is not obvious, consider varying between 0 and 360 in the
diagram. Point D describes a circle about point A. It should now be obvious
that the lowest point of the circle is directly below A.

It is now a simple exercise in geometry to define the angle that line AE

makes to the vertical. This is the same angle that BC makes to the horizontal.
Consider the geometry of the situation.

3.3. SEWAGE WORKERS RESOLUTION FFF

89

We have
= arctan

(d/8)
1
= arctan 16.10
|AE|
2 3

This is the same answer as before. By using the principle of minimization of

potential energy we have considerably simplified the problem.

Assumptions
We started by saying that the aqueduct was smooth, and we implicitly suggested that the man could get onto the ladder in a position that was not the
point at which the ladder ultimately settles. If we wanted to be pedantic, we
could argue that if there was no friction at all the ladder would simply oscillate indefinitely, and never settle. Likewise, getting onto the ladder could be
problematic. If the expected answer to at what angle does the ladder settle?
was either of these I think we would feel we had been somewhat misled. We
therefore assume that the walls are smooth enough that the reaction forces are
always normal to the wall, but not quite so smooth that the ladder cannot
eventually settle.
An important skill in addressing physics problems is identifying the fundamental subject of interest, the nub of the question, and filtering out everything
that is unimportant. In that latter category are an almost infinite number
of things which we should know we need to discount: the material of the
ladderwe know it is smooth; air resistancewe are considering an equilibrium/stationary solution; the name of the man; the day of the week; etc... If
we feel the need to introduce these latter things, we are probably more interested in riddles than physics problems.

3.3

Sewage workers resolution FFF

A sewage worker needs to use a ladder inside a large and perfectly smooth
underground aqueduct of circular cross section. The ladder is of the same
length as the diameter of the aqueduct. Draw the forces on the ladder when
the ladder is placed: i) vertically; ii) horizontally. Discuss the diagrams.

Answer
We consider the answer in two sections. First the vertical ladder, then the
horizontal ladder.
i) Vertical Ladder
This should seem, and is, very simple. There are only two external forces
acting on the ladder:

CHAPTER 3. STATICS

90

The weight of the ladder, mg, which acts vertically downwards from the
centre of mass of the ladder.
The reaction force, F , which acts normal to the smooth wall of the
cylinder at the point of contact.

If the ladder is in static equilibrium the two forces must be in balance. That
is, F = mg.
ii) Horizontal Ladder
Our first reaction might be that this is equally simple. We can have three
external forces acting on the ladder, one due to the weight of the ladder, and
one at each end of the ladder. The ladder is symmetrically disposed about
the y-axis, so we expect the two forces at the ends of the ladder to be equal
in magnitude and opposite in sign. We know that the wall of the cylinder is
smooth, so the forces at the ends must be normal to the wall of the cylinder.
The arrangement of forces is shown in the diagram below.

A problem is immediately apparent. We can balance forces in the horizontal

direction, but what about the vertical direction? We seem to have a net downward force due to the weight, mg. The ladder cannot be in static equilibrium
with this arrangement of forces. We require upward forces at the walls to balance the weight of the ladder. As the boundary conditions are stated, and,
making assumptions about the ladder having negligible width in the vertical
direction, it is impossible to achieve a balance of forces.

3.4. AZTEC STONE MOVERS FF

91

There are a number of ways we can modify the boundary conditions, or

relax the assumptions, that would allow us to have a more realistic arrangement of forces.
We could relax the smooth wall condition, and have a coefficient of
friction slightly greater than zero.
We could allow the ladder to have finite width in the vertical direction,
so that the points of contact are just above and just below the centre of
the cylinder. This would allow an upward component of reaction force
at the lower points of contact.
We could relax the condition that the ladder is exactly the same length
as the diameter of the cylinder, and allow it to be infinitesimally shorter.
This would allow it to drop a small distance below the centre of the cylinder, allowing an upward component of force at the points of contact.
In practice this would happen due to both bending and compression.
Let us consider the option of the ladder being infinitesimally shorter. This
could be on account of a tiny amount of compression, or due to very slight
bending, or both. Whatever the mechanism, the points of contact with the
cylinder move just below the line through the centre of the cylinder. If we
still make the assumption that the cylinder is perfectly smooth, the symmetrical reaction forces, F , are still normal to the wall, and are therefore directed
through the centre of the cylinder. This arrangement of forces allows us to
have zero net horizontal and vertical force on the ladder. The conundrum is
resolved. It is interesting to note that for a small amount of compression we
require F  mg, so the mechanism for providing measurable compression of
even a fairly rigid ladder is also apparent.

3.4

Aztec stone movers FF

There is some uncertainty about how the Aztecs and Egyptians moved huge
stones great distances to their temples and pyramids, even if we assume that a
practically unlimited supply of slaves must have been available. There is even

92

CHAPTER 3. STATICS

more uncertainty about how our Neolithic ancestors behind such structures
as Stonehenge could have accomplished what even today would have been
quite a feat of engineering. The sarsen stones that form the outer circle of the
monument weigh 50 tonnes each. To lift such a stone would take the strength
of about 1,000 menunfortunately only about 20 men fit round the perimeter
of each stone. The nearest known source of sarsen stone is 25 miles from the
monument. And the stones were transported in about 2000 BC.
We have almost no direct evidence of how such a feat was accomplished,
but most theorists have it that the stones were moved using sleds and ropes.
The sleds were made of tree trunks, and the ropes of leather. The sleds were
loaded onto rollers, also made from tree trunks. It has been estimated that
using this relatively sophisticated system would have required 500 men to pull
the load and a further 100 men to place the rollers.
If a block of stone 2 m long is to be pulled 1 km, how many times do rollers
have to be placed in the path of the stone?

Answer
The obvious answer to this goes as follows. The block of stone is 2 m long and
can only balance on a single roller when that roller is exactly in the middle
of the block. At this point the trailing roller could be removed and a leading
roller could be added. The separation of the rollers is therefore 1 m, and
approximately 1,000 rollers would have to be placed along the 1 km path. As
you have already guessed, this answer is wrong. That would be too easy to be
interesting.
The correct solution is only a little more complicated, however. The key
observation lies in the fact that a circle rolling along the ground is instantaneously stationary at the point of contact. If it were not, slipping would occur.
Consider points A, B and C in the diagram below. Point A is the point of
contact with the ground. It is the point about whichinstantaneouslythe
body appears to be rotating. We conclude that vA = 0. Neither is the ground
moving, so vE = 0. Points B (the centre) and C (the top) lie vertically above
point A. Thus, B and C can have only horizontal components of velocity if
we consider that the body is rotating instantaneously about A. This is an important point, which is worth thinking about. If the wheel is in solid-body
rotation about point A, the velocity of C must be twice the velocity at B, so
vC = 2vB . This is easy to see because B and C are, respectively, one radius

3.4. AZTEC STONE MOVERS FF

93

and two radii from A. The velocity distribution for points on a line between
A and C is linear; that is, a triangle of velocities is formed.6

We can apply a similar argument to the block of stone. If the roller rolls against
the block of stone, rather than slipping, the roller and the stone must have the
same velocity at the point of contact, so vD = vC . The important result is that
the speed of advance of the stone is twice the speed of advance of the roller.
We will return to this in a moment.
For interests sake, lets consider the problem in the frame of reference of
the block, that is, the frame of reference in which the block appears to be
0
= 0. This is the frame we would be observing from if we were
stationary, vD
sitting on top of the block as it was being slid along by the workers. In our
block frame we will mark all velocities v 0 , to distinguish them from velocities
measured in the ground frame, v. In our block frame it looks like the ground
is being slid along on rollers underneath the block. The ground has to go in
the opposite direction to the block when observed from this frame. Because
0
0
= 0.
= vD
there is no slipping at the point of contact with the block, vC
6 A popular, related puzzle is how to derive the equation of a cycloid, or the path traced by
a point on the rim of a wheel as it rolls without slipping along a plane. For a wheel of radius
r, which has rolled about its centre through an angle , it is possible to show using relatively
simple geometry that the parametric equations (that is, the equations for x and y as a function of
) are x = r ( sin ) and y = r (1 cos ). By eliminating it is possible to show that the
Cartesian equation (x = f (y)) is given by


1/2
y
x = r cos 1
2ry y 2
r
This is valid for y = 0 to y = 2r, and gives the first half of the first hump (from the origin to P
in the diagram).

94

CHAPTER 3. STATICS

Once again, the velocity distribution along a vertical line of points through
0
0
the centre of a roller is linear in form, giving vA
= 2vB
. In this frame of
reference, the ground is moving twice as fast as the centre of the roller. Because
0
0
there is no slipping at the ground, vE
= vA
. There are no great surprises here,
but it can sometimes be helpful to look at problems like this from another
perspective to confirm that they still make sense.

The key result of this analysis is that for every 1 m the block is pulled forward,
the roller advances 1/2 m. We decided at the outset that we needed the rollers
to be spaced exactly 1 m apart under the block. But this would be achieved
by having them spaced every 2 m apart on the ground, because as the roller
is pulled forward 2 m, the rollers already underneath the block go back only
1 m relative to the block. The process is illustrated below.

Further discussion
It is interesting to note that the answer is independent of the diameter of the
rollers. Note that we did not specify the diameter in the analysis. In fact,
as we will see in questions relating to imperfect wheels, the force required to

3.4. AZTEC STONE MOVERS FF

95

rotate an imperfect wheel on a smooth surface also depends only on shape,7

and does not depend on diameter. When rolling on a rough surface, however,
the diameter of the wheel is important, and on rough surfaces larger wheels,
or larger rollers, are clearly better.
There is a convincing case that the development of wheels was inspired by
observations about the relative speed of rollers and load. In this version of
history, men started with the observation that pushing or pulling great loads
was made easier by using rollers made from tree trunks.

Around the same period in mans technological development, people started

to use primitive sleds to haul loads, because the smooth runners had lower
friction on most surfaces than the uneven loads did. One day, a particularly
clever person realized that combining the sled and the roller was more efficient
than using either in isolation.

Many many years passed in which the roller-sled system was adopted more
routinely for heavy loads. When objects were hauled very long distances the
sleds cut into the rollers, creating deep grooves in them. This served to guide
the sled, which was helpful, but there was an even more important effect. As
the grooves got deeper, the rollers needed to be placed less frequently. With
very deep grooves you might need to place the rollers, say, ten times less frequently than with un-grooved rollers.8 This made moving loads over long
distances much more efficient.
7 By shape we mean how circular the wheel is. In a question that follows, we model imperfections in the wheel, or facets, by representing the wheel as an n-sided polygon.
8 We require a diagram to explain this interesting effect fully, but Ive left that as an exercise for
the reader. Consider, however, that when the grooves get very deep, the radius r of the remaining
axle of wood (at the very bottom of the groove, that is) is very much less than the radius of the
roller, R. We write r  R. We should be able to show that the forward progression of the load
in the lab frame is proportional to R + r, and the backward progression of the roller with respect
to the load (in the load frame) is proportional to r. The distance we can travel before having to
replace a log is therefore proportional to (R + r) /r, and goes up quickly as r becomes smaller.
Prove it for FFF!

CHAPTER 3. STATICS

96

Neolithic man then started pre-grooving the rollers, creating what we would
now recognize as axles with fixed wheels, which were placed periodically under the load. Later a single axle with fixed wheels was attached through fixed
holes in the sled. Now there was relative motion of the bearing and axle, so
the holes had to be lubricated with animal fat. From here it was only a small
step to design a fixed (non-rotating) axle with removable wheels.
And the rest, as they say, is history.

3.5

The Wheel Wars I FFF

About five thousand years ago, tens of thousands of years of somewhat uncertain prehistory were ending rather abruptly in almost all corners of the world,
with the almost simultaneous beginnings of Copper Ages, Bronze Ages and
Iron Ages. The historical evidence suggests that smelting techniques were first
used about 5,500 years ago in Plocnik, modern day Serbia. And by 2000 BC, a
mere 3,500 years later, metallurgy had spread to South America, where it was
used in Pre-Inca cultures to extract gold and silver. This comparative rush of
technological development marked the end of the Neolithic Era, which was
itself the end of countless millennia we call the Stone Age. It is strange to
think that our human ancestors had developed such a high degree of latent
sophistication, which would go on to suddenly manifest itself in all the great
inventions during the few thousand years between then and now.
One of the most important inventions in human history, along with the
tools to make fire (fire itself had been around for a bit longer), and much later
the printing press and penicillin, was the wheel.
The first wheels seem to have been invented at similar times in several
places: Mesopotamia (a region between the Tigris and Euphrates rivers which
includes a large part of modern-day Iraq), the Indus Valley (modern day Afghanistan and Pakistan), the North Caucasus (home to the Maykop culture) and
Central Europe. The very oldest intact wheel is thought to be the Ljubljana
Marshes Wheel, which is believed to be 5,150 years old. Some people believe
that the oldest known representation of a wheeled vehicle is on the Bronocice
pot, dug up in the 1970s in Poland, and which dates back to 3500 BC.

3.5. THE WHEEL WARS I FFF

97

As with all interesting topics in history, there are naysayers, those who believe the repeated motif is not a cart at all. You can decide for yourself. From
around 3500 BC onwards, however, there is increasing evidence that people
were using the wheel for pottery-making, transportation, and many other applications besides. The fanciful cartoon images of cavemen with stone wheels
are just that. Strangely, however, huge stone wheels were, and to some extent
still are, used on the island of Yap (in Micronesia) as a form of currency. The
wheel truly has wide application.
The wheels ability to roll is a strong function of how smooth it ishow
far it deviates from a perfect circular form. Of course, it also depends on how
flat the road isa problem in ancient times.
During the Wheel Wars, a very brief period of Paleolithic history between
1,000,000 years ago and 1,000,001 years agothe exact dates are disputed by
some scientiststhere was a sudden surge in wheel research, and wheel makers
were selling wheels with three, four, and five sides, or even special made-toorder n-sided wheels. Wheel sellers claimed that it was worth spending more
caveman currency (probably mammoth tusk or some such) on wheels with
more sides, and it certainly seemed to be the case that large-n wheels rolled
very smoothly indeed. But it would be another million years until mathematics was advanced enough to prove or disprove these marketing claims.
How much horizontal force, F , applied at the frictionless axle of a regular
n-sided polygon of mass m, is required to cause it to roll without slipping?9 If
we wish, we can take r to be the distance from the axle to the vertices of the
polygon, defined so that in the limit n , r defines the radius of a circular
wheel.

Answer
It may have been beyond the average Neolithic man, but this is actually relatively easy to answer. Lets start with a polygon with an arbitrary number
of sides: n = 4, say (a square). In a moment we will see that it is a trivial
extension to generalize the answer.
First we consider the point about which the square rotates. If we apply a
force towards the left, and if the square rolls without slipping, the point about
which the square rotates is the bottom left corner. When the object is on the
point of overbalancingthat is, when it is about to rotateall the force from
9 This implies that we must assume that the coefficient of friction is high enough that this
motion can occur.

CHAPTER 3. STATICS

98

the ground is transmitted through the point of rotation. If this is not obvious,
consider the object a moment later.
If we apply a horizontal force F at the centre of the square (the axle), it
must be balanced by a force equal in magnitude and opposite in sign which
acts at the point about which the square rotates. Similarily, the downward
force due to weight, mg, must be balanced by an upward force, mg, acting at
the point of rotation. These two conditions are required for the object to be
in static equilibriumin other words, not accelerating in either the horizontal
or vertical direction. We can now take moments about the point of rotation.
When the applied force, F , is just large enough to cause rotation to occur, the
moments about the pivot due to this force, and the weight mg, must be equal
and opposite.

We have
F r cos = mgr sin
where r is the distance from the axle to the vertices of the polygon. We can
rearrange this to give
F = mg tan
where, for a square, = 2/8. In general for an n-sided polygon, we have
= 2/2n = /n. So the required force becomes

F = mg tan


n

Because the side of a polygon subtends a smaller angle at the axle as n increases,
the required force reduces rapidly as n increases. As n , 0, so
the required force also approaches zero. A perfectly circular wheel (with no
friction at the axle) requires no force to rotate on flat ground. A million years
later we have proved the cavemen rightit was worth spending a bit more
mammoth tusk on a wheel with lots of sides.

3.5. THE WHEEL WARS I FFF

99

If we plot the solution we see that for n = 4 the force required to rotate the
object is equal to mg. For a twenty-sided polygon F = 0.16mg. The solution
asymptotes to zero for n , which represents a perfect circle.

Further discussion
You will probably have noticed that in the analysis we have just performed the
solution for the force required to rotate a wheel does not depend on diameter.
Yet we know from practical experience that larger wheels give a smoother
ride than smaller wheels. Bikes are less shaky than push scooters. There are a
number of things at play here.
Imperfections in the flatness of the ground act in the same way as facets on the wheel, and lead to large starting forces, and, by the same
argument, a bumpy ride.
Manufacturing tolerances become proportionally more important as
wheels become smaller.
Larger wheels often have additional mechanisms to reduce the peak
force. These could include pneumatic tyres, or some form of suspension, or just the compliance10 of the overall structure of the wheel.
10 Compliance is the inverse of stiffness, or the deformation of a structure under a given load.
The compliance of a structure is often responsible for the redistribution of stress, so that peak
stress within the structure is reduced.

Chapter 4

Dynamics and collisions

In this chapter we look at dynamics and collisions. We consider collisions
in terms of the velocities of the interacting bodies, the forces on the bodies
during a collision, and the energy of the bodies before and after a collision.
In the simplest case, we deal with head-on collisionsthat is, one-dimensional
collisions. The mathematics for two- and three-dimensional collisions is not
much more complicated, but we need to consider the vector components (in
the x, y and z directions, or in polar co-ordinates) when solving the equations.
The basic principles remain the same.
To solve problems in dynamics and collisions we need a number of concepts to hand. We wont elaborate on all the equations here, because you
should be able to derive them as you solve these problems. If we have the basic
principles correct, however, the rest should follow naturally. Let us consider
some of the concepts that come up in dynamics problems, and the notation
that is used. If any of these are unfamiliar to you, it may be worth consulting
a textbook before trying the problems.
Velocity. When we consider interacting bodies, we are generally interested in their initial and final velocitiesthat is, the velocities before
and after the collision. We need a notation to distinguish these velocites.
For example, we can use v 1 and v 2 for the initial velocities, and v 01 and
v 02 for the final velocities. The corresponding scalar speeds are v1 , v2 , v10
and v20 . In general we will use the prime symbol (0) to indicate the state
following a collision.
Energy. The kinetic energy of a body is given by KE =
proof1 ).
1 This

1
2
2 mv

(see

proof of the formula for kinetic energy is a popular problem that encourages students
to think more deeply about an equation we normally take for granted. The incremental work
dW performed on a particle is defined as the force F acting on the particle multiplied by the
s2
displacement in the direction of the force ds. In vector notation we write W = s1
F .ds, where
we take the dot product of vectors F and s. If we consider motion in a single directionthe

119

CHAPTER 4. DYNAMICS AND COLLISIONS

120

Linear Momentum. The momentum of a body is given by p = mv.

Because velocity is a vector quantity, momentum must also be a vector quantity. According to the law of conservation of momentum, in
all collisions, whether elastic or inelastic, the linear momentum of the
system is conserved. For two bodies we write
m1 v 1 + m2 v 2
p1 + p2

= m01 v 01 + m02 v 02
= p01 + p02

Elastic and Inelastic Collisions. In an elastic collision the total energy of the bodies is conserved: the sum is the same before and after a
collision. In an inelastic collision energy can be lost. By lost we mean
that it is converted from kinetic energy, which is the useful form that
interests us, to less useful forms of energy like heat.2 For two bodies we
write
2
2
m1 v12 + m2 v22 m1 (v10 ) + m2 (v20 )
where the equality applies for elastic collisions.
Coefficient of Restitution. The coefficient of restitution e is defined
as the ratio of relative velocities after and before a collision. We write
e=

|v 0 v 02 |
speed of separation
= 1
speed of approach
|v 1 v 2 |

where, in general,3 0 e 1. For an elastic collision e = 1, for an

inelastic collision e < 1, and for a perfectly inelastic collision e = 0.
Later in this chapter we prove that, for elastic collisions, saying there is
zero kinetic energy loss is the same as saying that e = 1.
Impulse and Force. Newtons Second Law tells us that force on a body
is equal to the rate of change of momentum of the body. We write
F = dp/dt, where the momentum p = mv. This reduces to
x2
x-direction, saythis reduces to W = x1
Fx dx. The acceleration of a particle of mass m is
x2
related to the force Fx by Newtons Second Law, Fx = max . We write W = x1
max dx. The
acceleration ax is defined in terms of the velocity vx as ax = dvx /dt. We write
x2
x2
x2
dvx
dvx dx
dvx
m
m
m
W =
dx =
dx =
vx dx
dt
dx dt
dx
x1
x1
x1
v2
Simplifying, we write W = v1 mvx dvx , where we change the limits to be in terms of velocity,
the variable in which we are integrating. Integrating we get
1
1
mv22 mv12
2
2
which we define to be the kinetic energies of the initial and final states.
2 One measure of the usefulness of a particular form of energy is the ease with which it can be
converted to another form. In this context kinetic energy can readily be converted to heat, but
heat cannot as readily be converted to kinetic energy.
3 For superelastic collisions, or collisions in which stored energy is released, we can have e > 1.
W =

121

F =m

dv dm
+
v = ma
dt
dt

for

dm
=0
dt

We can integrate this relationship to get the impulse on a body due to a

force, which is simply defined as the integral of the force over a period
of time. So, impulse I is defined by

I=

t0

p0

dp = p0 p

F dt =
p

If the force is constant during the period of time over which we are
integrating, we get
I = F t = p = p0 p
So to achieve a particular change of momentum of a bodyto bring it
to rest, for examplewe require a fixed impulse. The magnitude of the
force required to achieve that impulse depends on the period of time
over which the force acts. If we fall over on a concrete floor, in the
resulting collision we are brought to rest very quickly indeed and the
force is high. It hurts. If we fall over and hit a springy surface we are
brought to rest rather slower and the peak force is lower.4 In each case
the total impulse is the same.
Reference frame. According to the principle of Galilean invariance,
Newtons laws apply in all inertial frames of reference. An inertial frame
of reference is one for which the acceleration is zero. This means that the
laws of conservation of momentum and energy (for elastic collisions)
apply equally in any non-accelerating frame. In 1632 Galileo used the
example of an observer in the belly of a ship sailing over a smooth sea.
According to the invariance principle, when performing experiments
the observer would be unable to detect whether the ship was in constant
motion or stationary. For all practical purposes this statement is true,
but we should note that the Earths surface is not quite an inertial frame.
This is because it rotates about its own axis, and about the Sun, which
itself rotates around the centre of the galaxy, and so on. We will ignore
the secondary non-inertial effects, which are small.
In addition to the laboratory frame of reference, which we take as being the frame in which the observer is stationary, there are two special
frames of reference we might choose for convenience during our calculations. Whichever frame we choose we should get the same result. The
frames are:
4 In collisions between springy/deformable objects, the peak pressure is also lower, because the
force tends to be distributed over a larger surface area.

CHAPTER 4. DYNAMICS AND COLLISIONS

122

A frame in which one body is stationary, allowing us to simplify

the equations.
The centre-of-momentum frame, in which momentum is zero both
before and after the collision (by the law of conservation of momentum). This frame has the special property that in an elastic
collision the bodies rebound with the same speeds as before the
collision (but with the signs of the velocities reversed).

4.1

Pulleys F

Many students find this straightforward question (in the sense that there are
no tricks) harder than they expect.
A light, inextensible rope connecting 5 kg and 10 kg masses is suspended
over a light frictionless pulley. The system is stationary, and has just been
released from rest at the instant shown. Answer the following questions:
1. Are the tensions T1 and T2 the same?
2. What is the magnitude and direction of the force on the pulley?

Answer
Let us consider the two parts of the problem in turn.
1) Are the Tensions T1 and T2 the Same?
The simple answer is yes, they are the same. They have to be the same because
both the rope and the pulley are light, so have no inertia.5 Likewise, the rope
is inextensible so has no ability to store energy. If we pull on one end, the force
must be transmitted to the other end.
5 Here we can think of mass as being the translational inertia. The rotational inertia of the
P
pulley is defined by I =
mr2 , the sum of the mass elements multiplied by their distance from
the centre of rotation.

4.1. PULLEYS F

123

1) What is the Magnitude and Direction of the Force on the Pulley?

For this part we need to solve the forces on all the objects. To do this we need
to draw free-body diagrams of all three components. To simplify the diagram
we let the smaller weight have mass m, and the larger weight have mass M .
We have established that the tensions in the ropes are equal. We represent the
tensions by T .
Lets deal first with the pulley. It should be immediately apparent that to
achieve zero acceleration of the pulley, we require an upward force equal to
2T acting from the axle of the pulley. We now need to establish the value of
T.

Consider the smaller mass m. Resolving forces upwards positive, and allowing
the mass to have a positive upward acceleration a, we have
T mg = ma
Consider the larger mass M . Resolving forces downwards positive, and allowing the mass to have a positive downward acceleration a, we have
Mg T = Ma
Adding these equations we have
(M m) g = (M + m) a
Rearranging for a we have
a=

(M m)
g
(M + m)

Putting in values (m = 5 kg, M = 10 kg) we have a = (1/3) g. We can

now solve either of the equations for the masses to find T . Doing this gives
T = (20/3) g. The upward force on the pulley is equal to 2T , so is therefore
(40/3) g. Taking g 10 m s2 gives a force of approximately 133 N.
We can compare this value to the upward force on the pulley that would
be required if the system were prevented from rotating (and the rope were

CHAPTER 4. DYNAMICS AND COLLISIONS

124

prevented from slipping over the pulley). In this case we would need to support the entire weight of the masses, and the upward force would be 15g, or
approximately 150 N. By allowing the centre of mass of the system (which
comprises both masses and the pulley6 ) to accelerate downwards, we reduce
the force requirement on the pulley axle.

4.2

Dr Lightspeeds elastotennis match FF

Dr Lightspeed and Professor Bumble are lining up for another elastotennis

match. Viewed from a distance it looks an awful lot like ordinary tennis with
very good players. The secret to their apparent prowess is that the elastoball
and elastoracket are both perfectly elastic. It is a futile ceremony in which
Dr Lightspeed wins every point. Fortunately, the ritual usually only lasts a
couple of sets before the cucumber sandwiches arrive and the pair tackle some
problem in mathematics.
It is fascinating, says Professor Bumble, how pre-eminently fast your
returns are. I swear the ball comes off your racket faster than the sum of the
ball speed and the swing speed. That should be impossible. Dr Lightspeed
looks confused for a moment then replies, Why, dear professor, surely the
reverse is true. In elastotennis it would be impossible for it not to come off
my racket faster than the sum of those speeds. They return to the match, in
which Dr Lightspeed wins every point.
Later, over more sandwiches, the professor is rather subdued. They sit
nibbling in silence. Eventually the professor speaks. I still dont understand,
Dr Lightspeed, how you can return so fast. You really must explain.

In elastotennis (tennis in which both the ball and the racket are entirely
elastic), what is the maximum return speed of the ball in terms of the original
ball speed and swing (racket) speed?
6 Though

note that, in this case, the pulley has no moment of inertia.

4.2. DR LIGHTSPEEDS ELASTOTENNIS MATCH FF

125

Answer
First we will take a brute-force approach to the problem, the standard method.
Well solve the momentum and energy equations in the global frame of reference (the frame of reference of the tennis court). Ive also given an alternative
answer, which exploits the centre-of-momentum frame, and which is more
elegant.
In the global frame of reference we define the velocities of the elastoball
and the elastoracket as v1 and v2 respectively.

We arbitrarily define a positive value of velocity as corresponding to motion

to the right, and a negative value of velocity as motion to the left. Letting
the masses of the ball and racket be m1 and m2 , we can write the momentum
equation
m1 v1 + m2 v2

m1 v10 + m2 v20

or
v1 + rv2 = v10 + rv20
where r = m2 /m1 . For a completely elastic collision we have
 0

|v 1 v 02 |
e=
= 1 or v1 v2 = v20 v10
|v 1 v 2 |
where v10 and v20 are the velocities after the collision. Eliminating v20 we have
v10 =

v1 (1 r) + 2rv2
(1 + r)

Taking v1 to be positive (the ball moving to the right) and v2 as negative

(the racket moving to the left), we seek to make v10 as negative as possible.
We therefore look at both limits, lim (v10 ) and lim (v10 ), and investigate all
r0

possible minima. Lets consider each of the three in turn.

CHAPTER 4. DYNAMICS AND COLLISIONS

126

Consider lim (v10 ). We see that lim (v10 ) = v1 . This is because m1 is so

r0

r0

heavy (relative to m2 ) that it continues unimpeded after the collision.

Consider lim (v10 ). We see that lim (v10 ) = 2v2 v1 < 0. In this
r

case m2 is so heavy (relative to m1 ) that it continues unimpeded after

colliding.7 The smaller mass m1 (the ball) bounces off the unaffected
larger mass m2 (the racket).
Consider all possible minima. To do this, we first differentiate to get
dv10 /dr, giving
dv10
2 (v2 v1 )
=
2
dr
(1 + r)
However, v1 > 0 and v2 < 0, so v2 v1 < 0. For finite m1 and m2 ,
r = is not allowed. For finite r, dv10 /dr 6= 0, so there is no minimum
or maximum.
Considering both limits, and having shown that there are no minima or maxima, we see that lim (v10 ) = 2v2 v1 is the solution. Recalling that v1 > 0
r

and v2 < 0 we see that the maximum speed of return |v10 | is equal to the ball
speed plus twice the racket speed. We write
|v10 | = 2 |v2 | + |v1 |
Dr Lightspeed is correct. In elastotennis it is impossible for the elastoball not
to come off the racket faster than the sum of the ball speed and the racket
speed.8 No wonder Professor Bumble is struggling to return his shots.

An elegant approach
We now consider an approach that invokes the centre-of-momentum frame of
reference.
We are interested in determining the maximum return speed. This will
occur when the change in velocity of the ball is greatest, no matter what inertial frame of reference we consider the problem in. Consider the frame of
reference in which the elastoracket is stationary. It should be obvious that in
an elastic collision, the biggest change in velocity of the ball occurs when
the change in velocity of the racket during the collision approaches zero.
This condition occurs when the racket has a much greater mass than the ball
(m2  m1 ). If this doesnt feel right, even in the frame of reference of the
7 To show this, we consider the symmetry of the problem (that is, what would happen if we
interchanged 1 for 2 throughout the analysis). Alternatively we can solve directly for v20 .
8 For an individual shot this is true even for realistic ball and racket weights (see further discussion).

4.2. DR LIGHTSPEEDS ELASTOTENNIS MATCH FF

127

racket, try and invent some examples to convince yourself.9 In the discussion we solve the more general problem (m1 and m2 having the same order of
magnitude10 ).
Because we can perform our calculations in any inertial frame of reference,
and because for m2  m1 the racket will experience no velocity change during
the collision, we consider the problem in the frame of reference of the racket.
This frame is travelling at v2 . The velocities v1 and v2 in the global frame of
reference become v1 v2 and 0, respectively, in the racket frame. We represent
the velocities in both frames below, and show the velocities of the frames with
respect to the global frame.

Considering the collision in the racket frame, for e = 1, and for m2  m1 ,

the velocity of the ball is reversed. A ball velocity before collision, v1 v2 ,
becomes a velocity after collision, v2 v1 . Because m2  m1 , the velocity
of the racket after the collision is still zero. We now step back into the global
frame by adding the racket frame velocity, v2 , to the velocities of the ball and
racket within the racket frame. For m2  m1 , the velocities of the ball and
racket after collision are v10 = 2v2 v1 and v20 = v2 .

9 Imagine floating in a space station with a superball (a highly elastic toy ball) and a number of
hard steel plates of various thicknesses. We throw the superball at a particular floating stationary
plate. If the plate is very heavy, the ball rebounds with approximately the same speed as it had
when it arrived. This is like bouncing the ball off a hard floor. As we make the plate lighter, the
ball rebounds more slowly. When the plate is very much lighter than the ball, it can no longer
impede the balls progress, and the ball does not return.
10 The order of magnitude of a number is its size to the nearest power of 10, and is used to make
approximate comparisons of the sizes of two numbers. The order of maguitude of a number
a is log10 (a) rounded to the nearest integer. If two numbers a and b have the same order of
maguitude we write a b.

CHAPTER 4. DYNAMICS AND COLLISIONS

128

We see that by using this approach we achieve the same result with considerably less complexity. The centre-of-momentum frame can be a powerful tool.

A realistic case FFF

Believe it or not there are quite a number of academic papers on the physics
of tennis. Specifically ball-racket interactions, and ball-surface interactions.11
In both cases the coefficients of restitution are reasonably similar, and both
are functions of the impact speed. The latter defines the deformation of the
tennis ball, which is the main source of energy loss during the collision. The
coefficients of restitution vary between about e = 0.7 at low speed (10 m s1 ),
and about e = 0.4 at high speed (45 m s1 ).
Lets reconsider the problem above, but with realistic values everywhere.
A fast approach speed for a ball after bouncing is v1 = 20 m s1 . A very fast
forehand racket speed would be v2 = 40 m s1 . A typical ball weighs about
m1 = 60 g, and a typical racket weighs about m2 = 300 g. Assume that the
coefficient of restitution is e = 1/2.
We step into the centre-of-momentum frame of reference, defined as moving with velocity V relative to the court frame. The velocities of the ball
and racket relative to the centre-of-momentum frame are u1 = v1 V and
u2 = v2 V . In this frame, by definition, the momentum is zero. That is,
m1 u1 + m2 u2 = 0. Expanding this we find that
m1 (v1 V ) + m2 (v2 V ) = 0
or
V =

m1 v1 + m2 v2
= 30 m s1
m1 + m2

This gives u1 = 50 m s1 and u2 = 10 m s1 . In the centre-of-momentum

frame the relative speeds of the ball and racket are u2 u1 = 60 m s1 . The
coefficient of restitution is e = 1/2. We denote the speeds of the ball and the
racket in the centre-of-momentum frame after collision as u01 and u02 . The relative speeds after collision are therefore given by u02 u01 = 0.5 (u2 u1 ) =
30 m s1 . In the centre-of-momentum frame we must have zero momentum
after the collision. That is, m2 u02 + m1 u01 = 0. Satisfying both of these
conditions yields u01 = 25 m s1 and u02 = 5 m s1 . To step out of the
centre-of-momentum frame and into the court or global frame, we need to
add the frame velocity. The velocities of the ball and racket in the court frame
are therefore v10 = u01 + V = 55 m s1 and v20 = u02 + V = 25 m s1 .
In this more realistic case we achieve a ball velocity of 55 m s1 (123 mph),
for racket and ball speeds of 40 m s1 and 20 m s1 respectively. For a per11 For further discussion see: Miller, S., 2006, Modern tennis rackets, balls, and surfaces, British Journal of Sports Medicine, May 2006, 40(5), pp. 401405, doi: 10.1136/bjsm.2005.023283;
and Brody, H., 1997, The physics of tennisIIIThe ballracket interaction, American Journal
of Physics, 65, pp. 981987.

Chapter 9

Electricity
In this chapter we look at problems involving electricity. Some of the problems test the application of very simple concepts, such as Ohms Law, or the
definition of electrical power. You may feel that some of the questions are
too simple even for high-school students. They are not, however, and it is
surprising how many students struggle to think their way clearly to a cogent
answer. Problems like this can be used to determine whether students have a
sound knowledge of GCSE-level material on these topics. We must not dismiss
them.
I have also included quite a few resistor puzzles, which involve groups of
resistors either in two-dimensional or three-dimensional arrays. The goal of
these is to calculate the overall resistance of the network. Resistor puzzles have
been very popular over the years. The classic question is that of the resistor
cube, where each edge of the cube has a given resistance (R, say) and the aim
is to calculate the overall resistance between two vertices, most commonly the
longest diagonal (this is also the simplest variant). The rules for solving these
problems are simple, so they reduce almost to logic problems.
I have not included any unsteady problemsthose which involve the charging and discharging of capacitors, or the time constant of CR circuits, etc.
even though these are also quite popular pre-university puzzles. They are easy
to find in most textbooks, and are rather formulaic, so (to me anyway) are
therefore less interesting. Nonetheless, this is an important area that students
should be familiar with if theyve studied physics to pre-university level.
Lets review a few simple definitions that will help us with these problems.
Current. The current, I, is related to the charge, Q, according to the
definition I = dQ/dt, where t is time. For steady currents this reduces
to I = Q/t.
Ohms Law. The current, I, through a conductor is proportional to
the potential difference, V , across it, and inversely proportional to the
228

229
resistance1 R. We write I = V /R.
Electrical Power. The electrical power, P , dissipated in a load of
resistance R, is proportional to both the current through the load, I,
and the potential difference across it, V . We write P = IV . Combining
this with Ohms Law we can write instead P = I 2 R or P = V 2 /R.
Resistors in Series. Resistors in series add. In other words, the total
resistance is given by RT = R1 + R2 .

We should prove this relation, because it is used in many of the problems that follow. It is a consequence of Ohms Law: V = IR. The
voltage drops across R1 and R2 happen in series, so the total voltage
drop between A and C is given by VT = V1 + V2 , where V1 and V2
are the individual voltage drops. The current flowing from A to B is
the same as the current flowing from B to C,2 so we can rewrite this
equation as IRT = IR1 + IR2 . Dividing the equation by I, we get
RT = R1 + R2 .
Resistors in Parallel. Resistors in parallel obey the following rule.
1
1
1
=
+
RT
R1
R2

To prove this we note that for a given voltage drop VAB the voltage drop
between C and D must be equal3 to the voltage drop between E and F .
That is, VCD = VEF = VAB , or V1 = V2 = V . The current, on the
1 This is only true when the resistance is not a function of the currentthat is, R 6= f (I). For
many real loads in practical situations this assumption may not be valid.
2 This is because charge is not leaving the circuit, and because there is no place for charge
to accumulate. We can think of this as a one-dimensional continuity equation for a conserved
quantity (charge).
3 Taking the wire to have zero resistance, and noting that voltages add in series, we have V
CD =
VCE + VEF + VF D . The resistances between C and E and between F and D are both zero,
giving VCE = 0 and VF D = 0. So VCD = VEF .

CHAPTER 9. ELECTRICITY

230

other hand, obeys an additive law: IT = I1 + I2 . Combining this with

Ohms Law we get
V
V
V
1
1
1
=
+
or
=
+
RT
R1
R2
RT
R1
R2
Capacitors in Series. Capacitors in series obey 1/CT = 1/C1 + 1/C2 .
Capacitors in Parallel. Capacitors in parallel obey CT = C1 + C2 .
Energy Stored in a Capacitor. The energy stored in a capacitor is
given by E = (1/2) CV 2 .
We are now ready to try a few puzzles. As I remarked, the resistor cube is a
well-established puzzle that has been circulating for several decadesIve also
seen it in at least one physics textbook. The other resistor questions are my
own, the result of a productive hour spent at Detroit airport waiting for a
connecting flight across the US. Every time the waitress topped up my coffee
she asked if I had solved my problem yet. Yes, I said, but I keep finding
new problems. To which she replied, Oh dear, honey, you just keep at it
then. Good advice for anyone attempting these puzzles.

9.1

Resistor pyramid FF

A resistor pyramid is built in a branched structure as shown below. If there

are N levels, constructed from individual resistances of value R, what is the
total resistance RN across the pyramid? What is the resistance of an infinite
pyramid of this type?

9.1. RESISTOR PYRAMID FF

231

Answer
We can think of the structure of the pyramid as a combination of both series
and parallel connections of resistors. From the symmetry of the arrangement
we see that all nodes at a given level n in the pyramid are equipotential points.
We can add dummy wires (shown as dotted lines in the diagram) between these
equipotential points without changing the current flows in the circuit. So we
see that the equivalent circuit has n levels in series, with 2n1 resistors in
parallel at the n-th level.

The resistance of the n-th level is therefore

Rn =

R
2n1

The resistance of a pyramid of size N is the sum of the resistances Rn from

n = 1 to N . That is,
RN



N
X
R
1
1 1
=
= R 1 + + + ... + N 1
2n1
2 4
2
n=1

The right-hand side of this expression is a geometric series4 with the sum5
"
 N #
1
RN = 2R 1
2
4 A geometric series of n terms has the general form S = a(1 + r + r 2 + ... + r n1 ), where
n
a is the value of the first term and the common ratio is r. The sum can be expressed simply as
follows. We multiply by r to give rSn = a(r + r2 + r3 ... + rn ), then subtract this from the
previous expression to give rSn Sn = arn a. Rearranging, Sn = a(rn 1)/ (r 1).
5 Taking a = R and r = 1/2.

CHAPTER 9. ELECTRICITY

232

To calculate the resistance of an infinite pyramid we need the sum to infinity

of the series. The common ratio is r = 1/2 < 1, so the sum to infinity
converges to give
R = 2R
It might have been easier to put two resistors in series.

A current-symmetry argument
Another way of thinking about this problem is in terms of current symmetry
in each tier of the pyramid. At the n-th level of the pyramid there are 2n1
resistors, each with the same current flowing through them (due to the symmetry of the structure). If the current through the entire pyramid is I, then
the current through each resistor in the n-th level is In = I/2n1 . We need
to add the potential (or voltage) drop across each tier in series to find the potential drop across the entire pyramid. The potential drop across a resistor in
the n-th level is Vn = In R = IR/2n1 . So the total potential drop across a
pyramid of N levels is

VN

"


 N #
N
X
IR
1 1
1
1
= IR 1 + + + ... + N 1 = 2IR 1
=
n1
2
2
4
2
2
n=1

The resistance of a pyramid of N levels is given by RN = VN /I. As before,

we get
"
 N #
1
RN = 2R 1
and R = 2R
2

An elegant solution
We now consider an elegant solution that will appeal to the more mathematically minded. A friend came up with the idea, and my brother and I extended
it.
In the diagram we represent an infinite pyramid of resistors as a double
triangle inscribed with R . If we extend the pyramid by one level, its the
same as having two such pyramids in parallel, in series with a single resistor. If
the series converges, the resistance of the extended pyramid must be the same
as that of the original pyramid. We write

R = R +
A very elegant solution.

R
,
2

so

R = 2R

9.1. RESISTOR PYRAMID FF

233

Lets consider how we might extend this to find the resistance of a pyramid
with N levels. In the diagram we represent a pyramid with N levels as a double
triangle inscribed with RN . We see that extending the pyramid by one resistor
gives
RN +1 = R +

RN
2

Here we remind ourselves that RN and RN +1 are terms in a series that we

seek the expression for. Noting that the sum to infinity of the series is given
by R = 2R, we define a new series QN RN 2R. The series QN
converges on zerothat is, Q = 0. We substitute the definition of QN into
the expression for RN +1 , giving
QN +1 + 2R = R +

(QN + 2R)
2

So
QN +1 =

QN
2

Noting that R1 = R, and that therefore Q1 = R, we see that

CHAPTER 9. ELECTRICITY

234

QN =

R
2R
= N
N
1
2
2

Using QN RN 2R, we get the expression for RN .

RN

"
 N #
1
2R
= N + 2R = 2R 1
2
2

If the previous solution was elegant, this is elegance to the power N .

9.2

Resistor tetrahedron F

A resistor tetrahedron has individual resistance values R in each leg. What

is the total resistance RT across any pair of nodes in the pyramid?

Answer
It is easiest to think about this by flattening the structure onto a plane. Doing
this makes it easier to see the essential structure were dealing with. Lets
consider the total resistance between A and B. When its flattenedas shown
in the diagramwe can see that the network reduces to a single resistor (at the
top, A B) in parallel with a group of five resistors (at the bottom, A0
B 0 ). If all the individual resistances take the same value R, the resistance of
the bottom group itself becomes R. We see this by noting from the symmetry
of the arrangement that the potentials at C and D are the same, so no current
flows between those two points. Thus the resistor between C and D can
be removed without affecting the currents in the circuit. The circuit is then
equivalent to having two resistors of value 2R in parallel, giving a resistance
of R for this block. The resistance of the top resistor (between A and B) is
trivially R. The total resistance between A and B is therefore RT = R/2. By
symmetry, this is the resistance between any two nodes.

Chapter 11

Optics
In this section on optics we look at problems built on simple principles of
reflection and refraction. The underlying material is no more than GCSE
level in difficulty, but the questions are designed to be challenging and unusual.
Some are well-known questions, and others are my own invention.
Law of Reflection. The angle between the incident ray and the normal
is the same as the angle between the reflected ray and the normal.
Snells Law of Refraction. A ray of light passing through an interface
between two different media of refractive indices n1 and n2 obeys Snells
Law, n1 sin 1 = n2 sin 2 , where 1 and 2 are the angles to the normal
of the incident and refracted rays.

11.1

Mote in a sphere F

I have only ever heard the word mote used in two places. The first was the
Bible,1 in which one appears in a persons eye. The second was in my Oxford
physics interview, during which I had to answer the following optics puzzle.
Fortunately, at my school a passing knowledge of the Bible was inescapable, so
I knew that a mote was simply a small speck of somethinga dust particle, for
example. However, Id not seen any unusual optics questions before thenbut
I suppose that was the point. The question is as follows.
A mote is at the very centre of a perfect sphere of glass. Where, if anywhere,
do you see the mote?
1 It appears in the Sermon on the Mount, as a moral lesson against hypocrisy and selfrighteousness. The word mote comes from Greek, and means a very small dry body. Today
we use the word speck more commonly.

285

286

CHAPTER 11. OPTICS

Answer
We are, of course, expected to draw a ray diagram to demonstrate the solution.
Snells Law tells us that refraction occurs according to the rule n1 sin 1 =
n2 sin 2 . Because the mote is at the very centre of the sphere, all emitted rays
have an angle of incidence of 1 = 90 with the glass-air interface. So the rays
are not diffracted by passing through the interface, and the mote appears to be
exactly where it is. In the ray diagram we represent the eye with a converging
lens, showing it focussing the mote down to a point on the retina.

11.2

Diminishing rings of light FFF

I first heard this question about 15 years ago, though I think it is much older
than that. My PhD supervisor was also a great fan of physics puzzles, and this
was one of his favourites. I used to think the question was too hard for highschool students, but tried it on a few recently and found that with enough
hints students made surprisingly good progress, and most were able to complete the question with a bit of help. I have also tried it on a number of PhD
students, who did no better. I attributed this to them forgetting most of their
high-school physics: they were all engineers. It is a very enjoyable question
to ask, and best done with an experiment. First I would ask students to describe what they saw before deriving equations from first principles to explain
it. Students really enjoy this, and find it very satisfying when they manage
to derive equations for the interesting optical display. When you attempt the
question, remember that it works best with an experiment and someone to
give you hints.
You are given a long, thin polished metal tube (a 1 m length of the copper

11.2. DIMINISHING RINGS OF LIGHT FFF

287

pipe used in domestic plumbing, for example) and asked to put your eye at
one end and to point the other end at a diffuse light source (a well-illuminated
wall, for example). What you see is a pattern of rings. At the centre there is a
bright circle, which is the diffuse light source seen directly through the hole in
the end of the tube. Surrounding this is a ring of light that is less bright than
the central circle. Around that theres another ring thats less bright still. And
so on. The rings become successively less bright until the pipe appears dark.
The rings appear to have approximately the same width as they get larger. You
are asked to derive an equation for the angle subtended at the eye by the n-th
ring.

Answer
When approaching this question, it may help to first sketch a diagram of the
tube, with the centre marked as a dotted line, in preparation for drawing ray
diagrams. Analyse first the light that is directly incident at the eye from the
diffuse light source without reflection, then consider the first-order reflection,
second-order reflection, and so on.
We start by drawing the ray that subtends the highest angle at the eye
(which is at the centre of the tube). We see that where the diameter of the
tube is D, and where the length of the tube is L, the highest angle that can be
subtended at the eye without reflection is given by tan 1 = (D/2) /L. For a
long tube, taking the small-angle approximation, we have D/2L.

We then consider the first reflected ray. To do this, the only additional thing
we need to know is the law of reflection, which is simply that the incident
and reflected rays make the same angle with the wall. We see that the ray that
subtends the highest angle at the eye, and which is reflected once, divides the
tube into three. We have tan 2 = (D/2) / (L/3). That is, 2 3D/2L.

CHAPTER 11. OPTICS

288

We then consider the second reflected ray. The ray that subtends the highest
angle at the eye, and which is reflected twice, divides the tube into five. We
have tan 3 = (D/2) / (L/5). That is, 3 5D/2L.

We quickly see that the general formula for the angle subtended at the eye by
the n-th ring is
tan n =

(D/2)
(L/ (2n 1))

Taking the small-angle approximation, we see that

(2n 1) D
2L
This is a rather satisfying question, and one that works very well with the
accompanying demonstration. If you havent already done so, track down a
length of copper pipe a meter or so long (from your local plumbers shop if
necessary) and try it. It is very neat indeed.
n

11.3

Floating pigs FFF

If youve read the introduction to this book, youll know that I was asked this
in my university physics interview. Its difficult to recreate here the manner in
which the question was asked, because it was set up as a desktop experiment. I
was asked to explain the strange optical effect that it created. That is certainly
the most logical way to ask the question. I still vividly remember the demonstration. The interviewer took two bowls from a drawer, dusted them off with
a handkerchief, and placed them proudlyand with something of a flourish
on the table in front of me. They were about the size of cereal bowls, and
their insides were perfect mirrors. The top bowl had a hole about two inches
in diameter in its base, and was set upside-down on top of the lower bowl, creating a mirrored cavity which could be viewed through the hole. Before that,
though, the interviewer placed a small, bright pink plastic pigabout the size

Chapter 14

Estimation
I really enjoy questions on estimation, or back-of-the-envelope calculations.
They are popular questions for a whole range of numerate disciplines, and are
certainly very common elements in physics and engineering tests. Strangely,
they are also very common in job interviews, particularly with companies that
consider themselves to be free thinking.
I remember several summer evenings when we were physics undergraduates, sitting in a friends room, staring out over the cricket lawn, drinking coffee, and estimating things simply for the fun of it. The game had two stages.
First, we would guess the answer to the problem at hand, going on instinct
alone, and then compare our guesses. At this stage, to raise the stakes, you
had to pretend to be very confident about your own guess and very sceptical
about everyone elses. It didnt matter that we all made the least accurate guess
about the same number of times. Then we would independently estimate the
answer, usually without recourse to pen and paper. It could take a while, because we often had to devise strategies to do the calculation at hand. This was
just before the internet changed everything. So if we needed some important but esoteric value we couldnt estimate easily, one of us would be sent to
the library to consult a (printed) encyclopaedia, or to look at a set of data
tables, and would reappear with the population of Bolivia, say, or the density
of uranium, scribbled on a scrap of paper. Once wed made our estimates we
would compare them. It was often surprising how close the estimates were,
even when we had taken diverging approaches to the solution.
We worked out that there were fewer blades of grass on the Merton College
cricket ground than there were people in the world,1 and that if we shared
the books in our college library equally across the population of the United
1 Merton

College cricket ground is about 200 yards on a side, and we estimated about 50 blades
of grass per square inch. This gave 50 122 32 2002 = 2.6 109 blades. Back in the
mid-1990s this was roughly half the population of the world, which was 5.7 bn (it is now over
7 bn). I still remember gazing out over the brown field at the end of that summer and thinking
what an awful lot of people that represented, and how insignificant we would each seem if we all
stood there side by side.

351

CHAPTER 14. ESTIMATION

352

Kingdom (something which we felt like doing more than once) we could offer
each citizen a paltry third of a page.2 We worked out how many college rooms
it would take to store all the gold in the worlda question that appears in this
book.
Wherever you look there is something to estimate, and its useful to know
how to set about it. Just recently someone told me that work had started on
the Kingdom Tower in Saudi Arabia. If its completed it will be over a kilometre high. The energy cost of pumping water up the tower, she speculated,
will be terrific. I thought for a moment, and with a little mental arithmetic
worked out the annual cost per person. When I told her the number she
looked at me in surprise. How did you know that? she asked. When I told
her I had just estimated it, but fairly accurately, she looked at me in disbelief.
I had to go through the steps of the calculation to convince her that I really
had just calculated it in my head.
Most everyday things can be estimated easily and relatively accurately.
Its an essential skill for anyone who is serious about being a scientist. I
shouldnt really admit to it here, but Im always a little shocked when someone
who has been through high-school science cant estimateroughlyhow big,
long, heavy etc. some object of interest is. Often people simply shrug their
shoulders and say, I have no idea. But we should always have some idea.
I hope you will enjoy this selection of problems. They are based on the
more unusual questions I have heard over the years, or are favourites of my
own. Some of them require the odd value which you might not be able to
estimate accurately enough to satisfy yourself. In which case you might need
to look it up in a data table, or ask your tutor for it. Not knowing values is
fine. Its the ability to come up with a method for solving a problem thats
important.

14.1

Mile-high tower F

Frank Lloyd Wright, the great American architect, was perhaps not the first
to conceive of the notion of a mile-high building, but he was probably the
first to work seriously on plans for one. His 1956 skyscraper Illinois would
have been 528 stories high, making it four times taller than the Empire State
Building. To get up and down it youd have had to use one of 76 atomicpowered elevators (gulp!). The Illinois was never built, and its unlikely that
the technology to do so even existed at that time.
Almost sixty years later, the tallest structure ever constructed by man is
the Burj Khalifa in Dubai, in the United Arab Emirates. Finished in 2010, it
stands 829.8 m tall. Huge as it is, it is still only half the height of the magic
mile.
2 The library walls totalled a good 300 m, and were stacked six shelves high. A ream of 500
pages of paper is about 5 cm thick. This gives (300 6 500) /0.05 = 18 106 pages, or a
third of a page for each of the then 58 million inhabitants of the UK (the population is now over
63 million).

14.1. MILE-HIGH TOWER F

353

The Kingdom Tower, which is currently being built in Jeddah, Saudi Arabia, was originally designed to be one mile high. Plans were drawn up by the
architect who designed the Burj Khalifa. But the geology of the area proved
unsuitable, and the tower had to be scaled down. The current intended height
is still a closely guarded secret, but it has been announced that it will be over
1 km, setting a new record for super-tall buildings.
In the architectural world there now seems to be little debate over the
technical feasibility of the mile-high buildingthe question is purely one of
finance. Given the sudden explosion in super-tall buildings in the last ten
years, the near future will be very interesting.
I remarked in the introduction that someone I know recently speculated
that the cost of pumping water to the top of a mile-high building would be
fantastic. The purpose of this question is to calculate whether or not that is
so.
Estimate the annual financial cost of the energy required per person to supply water to living accommodation at the very top of a mile-high building.

Answer
Of course, we have to assume that were talking about the current economic
and energy climate. The current cost of electricity in the UK is roughly 15
pence per kWh. Its reasonable to expect students to be able to estimate this
to within, say, a factor of two. Energy is a key household expense, and one
of the most significant drivers of the economy. Now lets calculate the annual
energy required to pump water for a single person.
First we need to calculate the amount of water used by a person in a day.
We can create a rough table to show this, and each of us might fill the table
slightly differently. But we should certainly agree to within a factor of two or
three, say. My rough estimates are as follows.

CHAPTER 14. ESTIMATION

354

Toilet flushes
Showers
Washing clothes
Washing dishes
Sundries

Number per day

5
2
1/3
1
1

Litres
10
25
60
20
10
Total

Total litres
50
50
20
20
10
150

We estimate the total amount of water used at 150 l (or 150 kg) per day.
Now we can estimate the energy required to pump that water to the top of
a mile-high tower. A mile is approximately 1.6 km, or 1, 600 m. The energy
cost is simply the gravitational potential energy gained, mgh, divided by the
total efficiency of the pump. Lets take a very conservative electrical efficiency
value of E = 0.5, and a very conservative mechanical efficiency value of
M = 0.5. The total energy cost per day is
E=

mgh
E M

Taking m = 150 kg, g 10 m s2 and h = 1, 600 m gives E = 9.6 106 J.

This looks like a big number, but lets put it in context by calculating the
number of kWh it represents. One kWh is equal to 3.6 106 J, so we need
2.7 kWh to pump a days water to the top of a mile-high tower. That works
out at a daily cost of approximately 40 pence. The annual cost of the energy
is approximately 146. Its not a trivial sum, but it is far from being a fantastic
amount of money. In fact its only about one-third of the cost of a typical
Oxford water bill. In the scheme of things we can say it is fairly insignificant.
The Burj Khalifa in Dubai has set an unprecedented record for height, and
its likely that it will soon be surpassed by the Kingdom Tower in Jeddah, at
over 1 km high. The smart money says that in the next few years we will build
even higher. Consider the predictions made in a recent paper in the Journal
of Urban Technology:3
Technologically, a one-mile-high (1.6 km) building is possible
todayalthough not without the associated challenges of required
high-strength materials, fire safety provisions, elevator systems,
and, above all, constructability. As the building gets taller, the
slenderness (height-to-width) ratio increases and the tower becomes dynamically extra sensitive to wind so it cannot be inhabited comfortably. To overcome this, the slenderness effect of the
tower is offset by making it wide at the base, or connecting several towers with bridges at different levels so they will enforce
and stiffen each other. Will it be possible someday to build a skyscraper even taller than a mile-high building? Probably yes. Since
3 Al-Kodmany, K., 2011, Tall buildings, design, and technology: visions for the twenty-first
century city, Journal of Urban Technology, Vol. 18, No. 3, pp. 115140.

14.2. HOW LONG DO WE HAVE LEFT? FF

355

1885 when the 10-storey Home Insurance Building was built in

Chicago, we have come a long way to the 2,717ft (828m) tall Burj
Khalifa in Dubai, 2010. Will building height increase at this rate
in the future? We may never know and may not contemplate this
at present but we cannot guarantee that this will not definitely
happen. One thing we know is that an inborn human instinct
is to overcome obstacles through creativity and dreaming. This
raises fundamental philosophical and moral questions about the
so-called cloud-piercing future visionary towers.
So the megastructures of the future may yet dwarf the tallest buildings of
today. People have different views about ultra-tall buildings, but I find them
remarkable and fascinating. I look forward to the mile-high buildings of the
future.

14.2

How long do we have left? FF

A good friend of mine used to test prospective chemistry students with this
estimation question back in the 1970s and 1980s. In those days he was a professor of physical chemistry and was researching the respirative absorption of
oxygen and other gases, as applied to anaesthetics and diving. He must have
been doing something right because one of his students went on to win the
Nobel Prize for chemistry.
The question relies on a basic knowledge of physiology, chemistry and
physics, at lower than pre-university standard. Anyone interested in the world
around us should be able to make at least a good guesstimate4 of most of the
quantities required. The question went as follows.
There are three of us in this room (whatever room you happen to be in).
Assuming it to be perfectly sealed, how long will we live for?

4 Better

than a guess, but worse than an estimate.

The Deadly Game of Puzzle

Points
Rbuffat points must have been responsible for the deaths of quite a number
of young mountaineers. In his 1973 book The Mont Blanc Massif: The Hundred Finest Routes, Gaston RbuffatFrench Alpinist extraordinairemade
the fatal error of organising his climbs roughly in order of difficulty, from
route one to one hundred. Mountaineers are no less competitive creatures
than mathematicians, and the deadly game of collecting Rbuffat points was
quickly established. You got one point for a gentle traverse of the Clocher
Clochetons on the Aiguilles Rouges, the first route in his book, and 100 points
for dicing with death on the unrelentingly vertical walls of the Central Pillar
of Freney, the last route in the book. If you got back alive, you could casually
drop into conversation that you had had a good season. If you were lucky,
someone would ask how many Rbuffat points you had collected, and you
could brag that you had collected x points in that particular year, giving you
y points in total. It was difficult to know whether to go for broke on a few
hard routes or tick off many easy ones.
So it is with Puzzle Points.
The four-star puzzles are just so much harder than the one-star puzzles,
I said to my friend Tet. It would seem unfair to give just four times as many
points.
Then square the number of stars before adding, said Tet.1
And so the Deadly Game of Puzzle Point collecting was born.
The distribution of questions in this book (including optional questions
for which solutions are given) is: 35 F, 40 FF, 28 FFF, and 10 FFFF.
That makes 113 questions in total, and a potential 607 Puzzle Points if we sum
the squares of the number of stars.2
Whats the point of points if you cant brag about them? said Tet.
And so the Puzzle Hall of Fame was born. By keeping track of your
1 In fact, a similar method of scoring was used for many years in Oxford maths exams. The
raw marks for each question were squared before being added. Like any other arbitrary method
of marking exams, it favours a certain kind of approach to the exam. In this case, it favours deep
knowledge of particular topics rather than superficial
knowledge



of many topics.




2 You calculate your points like this: 35 12 + 40 22 + 28 32 + 10 42 = 607.

367

368

THE DEADLY GAME OF PUZZLE POINTS

progress with the ticklists and scoresheet on the pages that follow, you can
calculate your Puzzle Points and your honorary title.
< 35 Puzzle Points: Aspirant Puzzlist
35 Puzzle Points < 195: Novice Puzzlist
195 Puzzle Points < 447: Expert Puzzlist
447 Puzzle Points < 607: Puzzle Master
607 Puzzle Points: Puzzle Overlord
You can enter and update your points in the Hall of Fame by going to the
website created for this book:
www.PerplexingProblems.com
Here you can debate the questions, make the case for alternative or more
elegant solutions, and propose questions of your own. You might think it
strange, but I have avoided clever solutions in favour of more standard methods throughout this book. Of course, I appreciate clever solutions as much as
the next person, but I thought it would be more instructive to focus on the basic techniques, especially as were dealing with such unusual questions. There
is ample scope, therefore, to improve upon many of my solutions in terms of
simplicity and compactness. I look forward to seeing these clever solutions
on the website.
Like all adventures, these puzzles are much more fun when they are shared.
Hopefully you already have someone to share your puzzling with, but the
Puzzle Forum on the website allows you to share your ideas with an even
wider audience, andmore importantlyto learn from others. From time to
time I will also post new questions on the website.
Finally, good luck. And I hope you enjoy the adventure, wherever it takes
you.

Engineering Science at the University of

Oxford, where he researches jet-engine and

rocket technology. His science puzzles have

appeared in the Guardian newspaper and been

featured on BBC Radio 4.

Entry to selective universities will often require students to demonstrate that

they have engaged in super-curricular activities which develop their awareness
and understanding of their subject. Tom Povey has provided in this book
enough super-curricular content to keep an aspiring mathematician, physicist,
engineer or material scientist (and their teachers) happy for months. For a
parent or teacher who wants to pro-actively support a student in preparing for
entry to a competitive science or maths course this is a no-brainer purchase.
Mike Nicholson, Director of Undergraduate Admissions and
Outreach, University of Oxford (2006-14), Director of Student
Recruitment and Admissions, University of Bath
This book is about the excitement of mathematics and physics at the school/
university boundary. It contains a remarkable collection of intriguing intellectual
challenges, all with complete solutions. You can read it to increase your ability
in maths and physics; you can read it to develop your problem-solving skills; but
most of all you can read it for fun. Professor Richard Prager, Head of the
School of Technology, University of Cambridge
This instructive book will allow aspiring young
scientists to test and expand their knowledge of
physics and mathematics, and gain greater
confidence in their abilities. Professor Sir Chris
Llewellyn Smith FRS, Director General
of CERN (1994-98)

ONEWORLD

MATHEMATICS / SCIENCE

jacket illustrations courtesy of the author

www.oneworld-publications.com

Professor Poveys Perplexing Problems T H O M A S P OV E Y

THOMAS POVEY is Professor of

T H O M A S P OV E Y

Professor
Poveys
Perplexing
Problems
Pre-University Physics and Maths
Puzzles with Solutions

In Professor Poveys Perplexing Problems, Thomas

Povey shares 109 of his favourite problems
in physics and maths. A tour de force of
imagination and exposition, he takes us by the
hand and guides us through uncompromisingly
challenging territory that expands our minds and
encourages a playful and exploratory approach
to study. The puzzles, he says, are like toys. We
should pick up the one we most enjoy, and play
with it.
Whether you are an aspiring scientist or an oldhand, pitting yourself against these problems will
test your ability to think, and inspire you with
curiosity and enthusiasm for physics. Presented
with charm and wit, the questions span the
gap between high-school and universityentrance standard material. Detailed answers are
lightened with a fascinating and refreshing blend
of scientific history, application and personal
anecdote.
On this delightful and idiosyncratic romp
through pre-university maths and physics, the
author shows us that behind every single one
of these questions lies a new way of thinking
about subjects we thought we had understood.
He argues that engaging with the unfamiliar is
key to forming deeper insights and developing
intellectual independence. Professor Poveys
Perplexing Problems is a manifesto that science
should be playful, and a celebration of the
curious.

25/$30