The

Mathematics
Curriculum HILARY SHUARD
HUGHNEILL

From Graphs
to Calculus
Blackie
The Mathematics Curriculum
FROM GRAPHS
TO CALCULUS

£V- :.;
V

National STEM Cenln

N22321
 The Mathematics Curriculum: A Critical Review
was a project set up by the Schools Council at the University of Nottingham 1973-77

Members of the central project team were

Professor J. V. Armitage (Principal, College of St. Hild and

St. Bede, Durham), Director
Professor H. Halberstam (Department of Mathematics,
University of Nottingham), Co-Director (1975-77)
Mr. G. R. H. Boys (1973-76)
Mrs. J. A. Gadsden
''.?.. Dr. R. B. Coates (1973-74)

The books in this series are

Geometry
From Graphs to Calculus
Mathematics across the Curriculum
Number
Algebra
Counting and Configurations
Mathematics in the World
Eleven to Thirteen
The Mathematics Curriculum
FROM GRAPHS
TO CALCULUS
Written for the Project by
HILARY SHUARD and HUGH NEILL

Published for the Schools Council by BLACKIE

 ISBN 0216 90341 6

First published 1977

Copyright © 1977 Schools Council Publications

All rights reserved. No part of this publication may be reproduced,

stored in a retrieval system or transmitted in any form or by any
means, electronic, mechanical, by photocopying or otherwise without
permission in writing from the publisher.

Illustrator: Julie Gadsden

Published by Blackie and Son Limited, Bishopbriggs, Glasgow

and 450/452 Edgware Road, London W2 1 EG

Printed in Great Britain by Thomson Litho Ltd., East Kilbride, Scotland

 Preface
This book is one of a series produced by the Schools Council Although the Project was not intended to be an exercise in
Project: 77ie Maf/iew«f:c.s- CmTi'm/:mi /4 Cn'n'm/ ^t^v'fn'. This curriculum development, it was almost inevitable and certainly
Project was initiated by the Mathematics Committee of the desirable that a review of existing syllabuses should lead to a
Schools Council as a result of letters received from teachers asking consideration of the possibility of a synthesis of "modern" and
for guidance on the vast amount of new mathematical literature "traditional". I believe that such a synthesis is possible and,
which had been produced for schools during the 1960s. The indeed, sorely needed. So, although we have not attempted to
Project was set up in 1973 and was based at the Shell Centre spell out an optimum syllabus, we have tried to identify the
for Mathematical Education at Nottingham University. important ideas and skills which should be represented at school,
It was felt that teachers, who faced a daunting array of math- and to show how so-called modern and traditional topics are
ematical literature and novel classroom material, would welcome related. We hope that one of the lessons which emerges from
a basis for constructive and critical discussion of the content of these books is that the two can be integrated in a unified presen-
the school mathematics curriculum. Moreover, whilst the choice tation of mathematics and its applications. Perhaps the current
of syllabus, books, materials, methods and presentation belonged numeracy debate will lead to syllabus revision. If so. it must be
properly to the teacher, the range of choices was so vast as to informed by the sound mathematical and pedagogical consider-
make well-informed decisions consistent with professional integ- ations to which end these books are devoted.
rity well nigh impossible: so that any advice implicit in these In order to focus as wide a range of experience as possible on
books, far from detracting from the teachers' role, would rather the task, planning teams were established, each under the chair-
establish it. manship of a potential author (an Editorial Fellow) and com-
The fundamental aim of the Project, therefore, is to help prising representatives from Universities, Polytechnics. Colleges
teachers to perform their own critical appraisal of existing math- of Education. Schools, the Inspectorate and the Advisory Services.
ematics syllabuses and teaching apparatus for secondary school The material produced was referred to working groups of
pupils in the 11 to 16 age range, with the objective of making, teachers across the country- from Cornwall and the Channel
for them, optimal choices. Such an aim, however fundamental it Islands to Northumberland. Moreover, the groups were in-
may be, is still inadequate. It was never the intention of those vited to make original contributions as well as to comment
responsible for the Project (hat they should provide only a review on planning material. All the material thus made available
of mathematical literature and apparatus, for an exercise of that was then referred once more to the original planning team,
kind would be obsolescent before the material could be pub- who now assume responsibility for advising the Editorial
lished. Instead, the Project was conceived as a contribution both Fellow on the "final" write-up, which in some cases received
to initial and to post-experience in-service training, as well as further redaction by other writers. The result is now before
providing helpful private reading. the reader.
VI PREFACE

It may seem strange that a book which includes the word Further, every school leaver should be able to extract and
"calculus" in its title should appear in a series written about the understand the information contained in the simpler forms of
mathematics curriculum for 11 to 16 year olds. However, although graph. More and more facts are being presented in this way by
the beginnings of formal calculus are the end-point of this book, industry and government, and it is vital that people should have
the authors believe that ideas such as measurement of area and the formed the habit of looking critically at suggested interpretations
concept of a limit should be encountered by every pupil, whether or of the data. This is an important prerequisite for a responsible
not he goes on to use these in differentiation and integration. society.
J. V. Armitage
 Acknowledgements
The authors wish to thank the many people who have in various Lastly, the authors would like to acknowledge with gratitude
ways contributed to the production of this book. Its design and permission to reproduce material from:
outlook were greatly influenced by the planning team, whose The School Mathematics Project, Books A to H, and X to Z,
meetings were both thought-provoking and fun. The authors hope Cambridge University Press, 1968 to 1974
that the team's fresh approach to what at first seemed well-worn The School Mathematics Project, Books 1 to 5. Cambridge
mathematical paths is reflected in the book. The members of the University Press, 1965 to 1969, and Metric Edition, 1969 to
team were: 1971
Dr. J. A. Anderson, Department of Mathematics, University of The Scottish Mathematics Group, Modern Mathematics for
Nottingham Schools, Books 1 to 9, Second Edition, Blackie and Chambers,
Mr. T. Farrell, Sarah Metcalfe Secondary School, Mid- 1971 to 1975
dlesbrough C. V. Durell, Certificate Mathematics, Volumes 1 to 4, Bell, 1974
Mr. A. Gorringe, Chosen Hill School, Churchdown, Nr. Glou- to 1975
cester Nuffield Mathematics Project: Pictorial Representation, Graphs
Mr. D. S. Hale, H.M.I. Leading to Algebra, Chambers and Murray, 1967
Mr. M. W. Pedelty, South Hackney Secondary School, London
Science 5/13: Structures anil Forces, Macdonald Educational,
Professor R. L. E. Schwarzenberger, Department of Mathe-
1972-73
matics, University of Warwick
Mr. E. B. C. Thornton, Claremont Teachers' College, Perth, SMP, Revised Advanced Mathematics, Book Cambridge
Australia University Press, 1973
They received perceptive comments on, and constructive criti- Reference has also been made to:
cism of, the planning material from Mr. J. K. Backhouse, Dr. T. J. J. B. Channon, A. McLeish Smith, H. C. Head, New General
Fletcher and from working groups of teachers in North and South Mathematics. Books 1 to 4. Longman, 1970 to 1971
London. D. E. Mansfield, D. Thompson, M. Bruckheimer, Mathematics: A
They would also like to thank Mrs. J. A. Gadsden, who has New Approach, Chatto & Windus, 1965
prepared an endless succession of typescripts, together with all the Nuffield Physics Teachers' Guide I, Longmans/Penguin, 1966
diagrams, and has helped the authors in countless other ways. B. Holland, P. Rees, Maths Today, Books I to 4, Harrap, 1975

vn
 Contents
Introduction CHAPTER 5 Further Developments of the Function Idea 32
5.1 Introduction 32
5.2 Doing without the "rule" 32
CHAPTER The Early Stages of Graphical Work 1
5.3 Review of texts 33
1.1 Graphical literacy 1
5.4 Recommendations 36
1.2 Beginnings 1
1.3 The introduction of Cartesian graphs 6
1.4 The expectation of science teachers 8 CHAPTER 6 Some Special Graphs 39
6.1 Straight-line graphs and proportionality 39
6.2 Other examples of proportionality 40
CHAPTER 2 Graphs as Mathematical Models 10 6.3 The reciprocal function 41
2.1 Graphical models 10 6.4 Transformations of some common graphs 42
2.2 Discontinuous graphs as models 12 6.5 Graphs of the sine and cosine functions 43
2.3 Aim of graphical work 13 6.6 The exponential function 46
2.4 An example 13 6.7 Logarithms 49
2.5 Looking forward 15
2.6 A recommendation 15
CHAPTER 7 What Is Area? 50
7.1 An approach 50
CHAPTER 3 The Idea of a Function 17 7.2 Sandwich arguments 52
3.1 Dependence 17 7.3 Two remarks 52
3.2 Historical development of the idea of a 7.4 An example of a sandwich inequality 54
function 17 7.5 Volume 54
3.3 Elementary notation and wording 19
3.4 Use of the word /'unction in "traditional"
CHAPTER 8 First Ideas of Speed and Gradient 55
texts ' 19
8.1 First experiences 55
8.2 Average speed 55
CHAPTER 4 Functions in the Early Secondary Years 21 8.3 Gradient of a straight-line distance-time
4.1 The introduction of functions 21 graph 56
4.2 Graphical representation of functions 21 8.4 The gradient of a straight-line graph as a
4.3 The labelling of Cartesian graphs 27 rate 57
IX
 CONTENTS

8.5 Gradient of a road: a red herring 59 10.10 Fibonacci sequences 85

8.6 Gradients of straight-line graphs 59 10.11 Conclusion 86
8.7 A change of wording 59
CHAPTER 11 Area under a Graph: Velocity and Displacement 87
CHAPTER 9 Numbers 61 11.1 Introduction 87
9.1 The need for real numbers in graphical 11.2 The SMP treatment of area under the
work 61 velocity-time graph 87
9.2 Early stages 62 11.3 An incorrect argument 90
9.3 Fractions and decimals on the number-line 62 11.4 A sandwiching argument for displacement
9.4 Recurring decimals on the number-line 63 and area 91
9.5 Calculating with recurring decimals 66 11.5 Negative velocity and negative area 94
9.6 The sandwiching process and measurement 66 11.6 The trapezium rule 95
9.7 Describing points on the number-line 67
9.8 Changing recurring decimals to fractions 68 CHAPTER 12 Velocity and Gradient 96
12.1 The tangent to a curve 96
9.9 Calculation with real numbers 69
12.2 The need for the gradient of a curve 97
9.10 The introduction of irrational numbers in
12.3 Gradients of tangents by drawing 99
texts 70
9.11 Proofs of irrationality 71 12.4 Remarks 100
9.12 Recommendations 71 12.5 Towards definitions of velocity and gra-
dient 102

CHAPTER 10 The Idea of a Limit 72 CHAPTER 13 The Definite Integral 103

10.1 Strategy 72 13.1 Introduction 1Q3
10.2 First ideas 72 13.2 Areas and sandwiches 103
10.3 A geometrical introduction to convergent 13.3 Definition of the definite integral 105
sequences 73 13.4 Integrals of increasing functions 107
10.4 Further simple convergent sequences of 13.5 Extending the result 107
numbers 77 13.6 Integration of polynomials 108
10.5 Sequences given by recurrence relations 78 13.7 Comments HO
10.6 Some mathematical discussion 81
10.7 Non-terminating decimals 83 CHAPTER 14 Differentiation 111
10.8 The key idea 85 14.1 Introduction 111
10.9 A sandwich inequality argument for con- 14.2 Chords: a numerical approach 111
vergence 85 14.3 Differentiation 113
CONTENTS XI

14.4 The scale-factor approach 116 16.2 Graph sketching 130

14.5 The best linear approximation 117 16.3 Design problems 135
14.6 Calculus Made Easy 120 16.4 Step-by-step integration: orienteering 137
14.7 The black-box technique 121 16.5 Volumes of revolution 141
14.8 The derivative of xi > x" 121
14.9 Recommendations 122 CHAPTER 17 The Early Historical Development of Calculus
by R. L. E. Schwarzenberger 145
CHAPTER 15 Linking the Definite Integral with Differentia­ 17.1 The origins of graphical methods 145
tion 124 17.2 Pythagoras 146
15.1 Which comes first? 124 17.3 Eudoxus 148
15.2 The link between integration and differen- 17.4 Archimedes 149
tiation 125 17.5 The early seventeenth century 151
15.3 Further discussion of the link between 17.6 Newton and Leibniz 153
integration and differentiation 126 17.7 Conclusions 154

CHAPTER 16 Some Applications of Integration and Differen­ Suggestions for Further Reading 156
tiation 129
16.1 Displacement, velocity and acceleration 129 Index 157
 Introduction
This book is about graphs, their drawing, their interpretation, "O" Level syllabuses which do not include area under a graph or
their development and their use. It discusses the teaching of graphs gradient of a curve. CSE syllabuses contain just the same range of
from their early introduction in secondary schools as far as the views about the extent of graphical work as GCE syllabuses do.
beginnings of integral and differential calculus. The introduction of a common system of examinations at sixteen
Graphs are used in order to convey in a simple pictorial and would, no doubt, produce syllabuses which reflect a similar range
immediate way ideas which otherwise would require many words, of opinion and practice.
figures or symbols to portray. "Every picture tells a story" is a Some teachers who argue against teaching integration and
particularly apt saying in this context; in fact, to those who differentiation to the more-able pupils before sixteen, do so on the
understand the message, the graphical picture avoids the need for grounds that an introduction to an important topic shortly before
the story by saying it all. an examination may lead to a hurried approach in which a proper
In order to extract as much as possible from graphical messages, understanding takes second place to the learning of rules to answer
pupils must learn to read the messages and to become fluent in the examination questions. It is unfortunate that the types of question
vocabulary of graphical language. Some pupils will learn to read set in a number of GCE and CSE examinations indicate clearly
these messages more quickly than others, and these more-able that an expectation of rote learning is usual in this area of work.
students will progress to the stage where they appreciate the Other teachers argue that it is very important that able pupils
significance of area and gradient; they can then go on to study whose major interests lie outside mathematics, and whose formal
integration and differentiation. Other pupils who are not so adept mathematical education will end at age sixteen, should see
at the language will not progress much further than reading and something of the power of calculus. Some of these pupils will, in the
understanding block graphs. All pupils, whatever their ability, sixth form, study quantitative subjects such as economics or
should learn to read as many of the messages contained in graphs biology or geography, where some knowledge of calculus is very
as they can, for otherwise they will be deprived of a means of useful. Others will eventually take up positions of responsibility in
communication of proven use. government, industry or commerce, and will need to make
It is interesting to observe the differences of opinion between decisions on the basis of numerical and graphical information.
teachers of mathematics about the depth of study of graphs which The authors of this book take the second view. They believe that
is appropriate for pupils of various abilities. Some GCE "O" Level a well-developed understanding of mathematics is a national
syllabuses recognize that the concepts of area and gradient are resource. The more numerate the decision makers, the more likely
closely associated with graphs, and require that integration and the decisions are to be well informed. It is because government,
differentiation should be taken as far as obtaining algebraic industry and business often communicate quantifiable ideas by
formulae and their application. Other syllabuses require only graphs, and because a knowledge of calculus helps citizens to read
approximate calculation of area and gradient. There are still other and understand these graphs, that calculus assumes its position of
xn
INTRODUCTION Xlll

importance. We cannot, as a nation, afford not to develop this reference has been made to computers, as the principles underlying
mathematical skill in our pupils. the use of calculators and computers are the same.
However, while holding the second view, the authors try to meet The discussion of the teaching of graphs, area and gradient form
the criticisms expressed by those who hold the first view by keeping the major part of this book. There follows an essay by Professor
integration and differentiation firmly embedded in the context of R. L. E. Schwarzenberger of the Mathematics Department,
graphs, and by deploring the mere mechanical "learning of rules" University of Warwick. This essay traces the historical develop-
which is not based on understanding of the central concepts of area ment of integration and differentiation, and relates history to
and gradient. The authors hope that those teachers who disagree teaching. In addition to contributing to the book. Professor
with their view, and first teach calculus as part of a sixth-form Schwarzenberger has enthusiastically given the authors most
course, will nevertheless find material which is of interest to them. valuable help and guidance throughout its preparation.
New developments in the mathematics curriculum have affected The authors believe that pupils should build up their knowledge
the teaching of graphs in various ways. First, there have been of important ideas over a number of years, treating the ideas in
considerable changes in the teaching of functions. In many greater depth and using more powerful techniques as their own
syllabuses, functions are taught much earlier than before, some- thinking develops. The two key ideas of the calculus area and
times appearing as special kinds of relations, and often gradient are treated in this way in this book, and the order of the
illustrated by types of diagrams other than the traditional chapters reflects this development. In the early secondary years the
Cartesian graph. Another development has been the increased use treatment is graphical and pictorial. Pupils' increasing arithmetical
of approximate methods to calculate areas under graphs and skill enables many areas under curves to be evaluated numericallv
gradients of graphs. This development is a most valuable stepping and, as the concept of ratio develops, pupils are able to find
stone before the introduction of integration and differentiation, gradients numerically. The final stage of abstraction is the
thereby reducing the number of new ideas with which the beginner generalization to algebraic formulae for integrals and derivatives,
in calculus has to cope at the same time. and the link between integrals and derivatives. By the age of
Both these developments have meant that textbook writers have sixteen, the most-able pupils are reaching a level of thinking when
incorporated new ideas into their books. The new material on they can handle these abstract ideas. For the rather less able,
relations and functions has demanded decisions about their concentration on the details of algebraic manipulation often
relative importance, while the material on area and gradient has prevents appreciation of the underlying ideas, and such pupils
required the writers to handle analytical ideas of some delicacy. should postpone algebraic work on area and gradient until later.
A third development is very new indeed. Modern electronic However, there is a powerful argument in favour of their meeting
calculators are now so cheap and so easily available that they have the ideas of the calculus in graphical and numerical settings which
opened up the possibility of a greatly increased emphasis on enable them to have a conceptual background for the later
numerical work, particularly in the teaching of limits and in- introduction of formal calculus, which is now needed in so many
tegration. These ideas are discussed in some detail in this book. sciences and other subjects. Thus, many teachers will not expect
The greater availability of computing facilities also increases the their pupils to cover the content of this book by the age of sixteen,
possibility of numerical exploration of these ideas, but no specific but the book goes as far as the fundamental theorem in calculus, a
XIV INTRODUCTION

simple version of which provides a suitable staging post for the

most-able pupils.
The first five chapters set the scene by discussing graphical work
and functions. Two separate themes develop from this: area
leading to integration, and gradient leading to differentiation.
These themes are developed throughout the book in parallel with
developments based on pupils' experiences of distance, time and
speed. The chapters on area and integration always precede the
corresponding chapters on gradient and differentiation, because
the authors believe that area is conceptually simpler than gradient.
Chapters 9 and 10 on numbers and limits take their place because
they provide essential background to subsequent work. Chapter 6,
Some Special Graphs, is at what seems to be the logical place in the
development of the book. The actual order and development of
chapters 1 to 15 is shown schematically in the diagram in which
abbreviated chapter headings are followed by the chapter number.

LINKING THE
DEFINITE INTEGRAL
WITH DIFFERENTIATION
15
 1 The Early Stages of Graphical Work

1.1. Graphical Literacy some literacy in graphical work, for the citizen who cannot read a
graph is a handicapped member of society.
Few pupils nowadays reach the age of 11 without having some
experience of drawing block graphs; at this stage, however, most
1.2. Beginnings
children's appreciation of graphical representation is vivid but
imprecise. During the years from 11 to 16 they should develop a We start by looking at some typical examples of graphs drawn
much greater understanding of how information can be conveyed by children before the age of 11. These examples draw our
by Cartesian graphs, and a knowledge of the fact that some attention to several conceptual developments which children need
information cannot be represented in this way. Their early to make in their graphical work. The examples are drawn from the
experience should also help pupils to acquire, through informal Teacher's Guides (Pictorial Representation and Graphs Leading to
examples, an idea of the essential characteristics of functions. Algebra) to the Nuffield Mathematics Project', among the children
Pupils should also come to link together the Cartesian graph and who enter secondary schools will be found some who are capable
other pictures of a function with algebraic representations such as of drawing each type of graph illustrated. However, the repre-
formulae. Graphs are often used to convey information about sentation most commonly used in primary schools is that of block
functions in a vigorous immediate way, but a graph will only get its graphs, and a number of children will have had no other experience
message across to a person who has learnt to read that message. of drawing graphs.
The growth of graphical literacy, both in the reading of graphs and Seven-year-old Richard has drawn his graph (figure 1.1) so that
in the interpretation of their messages, should be one of the main each square of the graph represents a child. Whole squares are
aims of graphical work for all pupils. coloured, and it is natural for Richard to write the numbers of
The more-able pupils will be able to draw graphs and to make children represented in the centres of the intervals on the axes. The
deductions from them, and also to interpret graphs and to make idea that a square of the graph paper represents a unit of a quantity
conjectures about the formulae associated with them. Graphs are leads naturally to the drawing of histograms. We shall also want a
used to convey information and to assist in the analysing of square of graph paper, which is a unit of area, to represent a unit of
information. Graphical representation is one of the most valuable some quantity such as distance at a later stage (see chapter 11).
mathematical tools used by scientists, economists, government, However, in order to draw a Cartesian graph, Richard needs to
industry and commerce. Indeed, many pupils will find a variety of supplement his idea of representing numbers by squares with the
uses for graphs in their activities outside mathematics. For less- idea that an axis of a graph is a number-line, so that a point on the
able pupils the aim of graphical literacy will inevitably be restricted axis represents a number, instead of a square of the paper'
to more limited circumstances, but all pupils can and should have representing a number. Carol, who drew the graph in figure 1.2,
 FROM GRAPHS TO CALCULUS

Tw,e g r^pk he. us ^5 H^»H moSh peoplt

Wolh Uor-oe -Fr 0 ^ School I h V-fl 115 uou t-Uo >-

eleven people c = m«. by b^S. Iv ^e-us uo'j

^oh t, p«o pU co^.e bj car l(- ! « 11 s ^«^ hKah

. t-l- c.rt d rt 53 p««p)t in )-hc class. IK h* i/i

MO^ l~Kal- 10 rv-.»re p»opl«. \A/dl^ V^o^e.

1-^arN 2= kj ccif», /nd S rvio^e ptfop/& co"-ie ^j

kvS hho n bj cur. Arsd ih SK«vv9 uov H^ol- s

^o<~a p opic walk- ^=>^vi c 4Tr.rv\ 5c^W oil HWo

J=> b w b w S

Figure I.I from Pictorial Representation

THE EARLY STAGES OE GRAPHICAL WORK

nc.lui'* graph skewing lit t nvuwVjO- of cnlaren in

O/ all The boyj ai-,d QIC.J m iU« clojii

i-iyk £ hows TKd b ChiU^en in <he c

One ChJJ Kaj A bro!Vi«i'< arJ itileci, and a

iar-j and i.itcvn

re are T-E cU.dre^v ii^ tUe clAi*>

If our clost UaJ a party a*J nv.TeJ all

iitei'^ |n«»"^ would be 65 cLi|cjy

0 i 2 3 456

1-iyiirc 1.2 ITDIII I'iclnruil Ri'i'i\"*

 FROM GRAPHS TO CALCULUS

VfMO te

u
UJ
cr
uJ
Of

Jl J5 J.3 J<r
"ttvo-t~<4i .cawi a*
-
artip acr ucn*u4 citnwt u-iTKe.sa>*w-crvcAtr'lwiudo' wcl^

Figure 1.3 From Pictorial Representation.

was in the process of learning this idea, and used points to of using an ordered pair such as (1, 3^} to satisfy D +2A = 8. Do
represent numbers on the vertical axis, but not on the horizontal the points intermediate to the grid points yet represent number-
axis. Julia (figure 1.3) was able to draw a line graph instead of a pairs to her? What meaning (if any) does she attach to the lines she
block graph. This stage leads naturally to the Cartesian graphs has drawn to join up the points on her graphs? We cannot tell from
drawn by Sandra (figure 1.4). She can represent an ordered pair of these graphs, but these questions are most important in under-
numbers such as (2,3) by a point. Many children of 11 are not as standing children's interpretation of graphical representations.
advanced as this. We notice, however, that Sandra has not thought They commonly join up points they have drawn on the graph
THE EARLY STAGES OF GRAPHICAL WORK

SontirCL ft.

n +2 A *$ » D + 2A
o, v
£ , 3
U , 2
b ( (
\, % , O
^
^ '^
-— ^ "-i
i 3 X 1. V » » ^ Ci 1

A na ^< 1 A, D -t 2 A = 5

5? , 0

^^
^ ^ ^^ 1
—1 ——«! r "<
D
u 1 ^ Jl t

IKooc OK boiK o
S-frmohl
Cjmnruo and aK bo^K at the. Sarrtt
I'igurc 1.4 Front </></;>/is l^'tuliiiti In Ahjchru.
 FROM GRAPHS TO CALCULUS

merely because they have seen graphs which look like this, How la come to school.
although they do not yet understand that all the points of the
graph satisfy its equation. We notice, too, that Sandra is develop-
ing an intuitive idea of gradient, as she says that both graphs are at
the same angle. It is not clear whether she is judging entirely by eye,
or whether she knows that a right-angled triangle with sides in the Q.
Q.
ratio 8:4 is of exactly the same shape as one with sides in the ratio "S
5:2^. However, spontaneous expressions such as her statement <D
that the lines are at the same angle are to be encouraged among .0
E
secondary-school pupils as much as they are amongjunior-school
pupils.

1.3. The Introduction of Cartesian Graphs

A graph is a means of recording and conveying information. Walk Cycle Bus Train
While pupils are learning the techniques of graph drawing, it helps The natural representation of this information, the frequency of
their understanding considerably if the information is vivid and a particular occurrence, is by a block graph. This starts a line of
full of meaning to them. Often, this implies that information should development towards the drawing of a histogram, in which the
be collected or measured by the pupils themselves. frequency of a property is represented by an area.
Much of the information easily gathered by pupils is of the form In order for pupils to progress towards drawing Cartesian
shown in the quotations below from SMP Book B (p. 107). graphs of functions, we need to look for situations in which
numbers in one set depend on numbers in another set. The following
are suitable situations in which pupils can collect information and
record it by drawing graphs:
Here is the result of a survey made on a class of first-form boys and girls to (i) The height of a growing plant or animal depends on the
find out how many are using each method:
time after planting or birth.
walk, 13; cycle, 8; bus, 4; train, 9; car, 2.
(ii) The length of a spring depends on the mass hung from it.
We can just write the results out like this, or we could try to find a way to (iii) The time of a railway journey depends on the length of the
display them so that they were easier to follow, or even to see at a glance.
The first method suggested is a simple table: journey.
Walk Cycle Bus Train Car
(iv) People's European clothing size depends on their British
clothing size.
13 8 4 9 2 Total 36 (v) The charge for a trunk telephone call depends on the
The total has been included and this acts as a useful check. distance.
THE EARLY STAGES OF GRAPHICAL WORK 7

(vi) The charge for a trunk telephone call depends on the

duration of the call.
If pupils are to plot with understanding a point on a graph such
as that of the growth of a broad bean plant (figure 1.5), two pieces
of knowledge need to be linked together in the representation of
the fact that "after 5 weeks, the plant is 6 centimetres high". These
are:
(i) how to plot the point with coordinates (5,6);
(ii) the fact that after 5 weeks of growth the plant is 6 cm high is
represented by the point (5,6).
The progression of thought from 5 to 6 is caught by the wording
5 maps to 6 and is indicated in some representations by a mapping
arrow, although in the graph of figure 1.5 the mapping arrow is not
seen.
Figure 1.6 shows a possible teaching method. It is important
that time is spent on the stages indicated by the first two diagrams.
It may be helpful at this point for the reader to look forward to
section 4.2 (page 21).
It is often easier for pupils to read what others have written than
to write their own stories; so too they may find it easier to read
graphs drawn by other people than to draw their own graphs. They
can be helped to imagine the invisible mapping arrows which a
graph contains if the teacher asks for sets of ordered pairs to be
obtained from a given graph, such as that in figure 1.5:
Fill in the blank in the sentence "The point (7, ) lies on the graph".
Where does 7 map to?
Where has 10 come from?
What number maps to 6?
567! 10 11 12
age in weeks

Figure 1.5 The rate of growth of a broad bean shoot.

 FROM GRAPHS TO CALCULUS

age in height in height in

weeks centimetres centimetres

age in
weeks
Figure 1.6 (i) The axes of the graph are parallel: the arrow shows that 5 maps to 6.

(iii) The mapping arrow is invisible, but the dot marks where it "turns the corner".
height in
centimetres

1.4. The Expectation of Science Teachers

The quotation opposite from Science 5/13: Structures and
Forces, Stages 1 & 2 (p. 34) is typical of the intuitive understanding
of graphs that is often expected of children of age 11 and 12 in
science. The topic under discussion is the rate of cooling of a can of
hot water when insulated by various materials used in house
building.

age in
weeks

(ii) The axes are perpendicular: the mapping arrow is shown with a right-angled
bend in it.
THE EARLY STAGES OF GRAPHICAL WORK

Also typical of the attitudes of science teachers is the quotation

Temperature
below from the Teachers' Guide I to Nuffield Physics (p. 198). The
experiment is that of stretching a spring by hanging weights from
it.
Graph. This is a case for plotting a simple graph, if children are ready for it.
Children take kindly to graphs provided they are introduced casually, not as
something they "should have done in maths", and unfortunately have not.

The authors of this quotation have a very valid point of view. We

have seen that most children have drawn many graphs before the
Time age of 11, but in a number of mathematics courses designed for
secondary schools this fact is ignored.
Children may see that 'the steeper the graph, the
quicker it is cooling'.
Does the cavity in the walls help to keep the
house warm (ie is air a good heat insulator) ? .
An experiment like the one shown can be tried.

Thermometer n-
Plastic bag
enclosing
plenty of air

Tin of hot water

 2 Graphs as Mathematical Models
2.1. Graphical Models height in
centimetres
We have seen that children of 11 and 12 can use graphs in a very
natural way to make a vivid pictorial record of information about
numbers and quantities which they have found in the world 18 -
around them. Graphical representation is, however, more than a
16 .
descriptive tool; its major importance in mathematics is that it is
predictive. The representation of information in a graph has an 14 -
immediate visual impact which encourages us to think about the •

12
information, to ask ourselves questions, and to draw conclusions.
For instance, we can measure the height of the classroom broad 1.0 -
bean plant each Friday afternoon, and draw a dot graph such as
that of figure 2.1. This draws attention to one aspect only of the 8 '
plant's growth: its height. We have ignored its girth, its mass,
6 -
whether it has grown a new leaf this week, and any other aspects of
•
its behaviour apart from its height. 4
We have made a mathematical model which simulates the
2
particular aspect of the real world in which we are interested; it
helps us to focus our attention on the height of the plant to the I i < I i I I I y

exclusion of all else, in a way similar to that in which a small boy 0 12345678 time m
weeks
can concentrate completely on a model car without the noise and
Figure 2.1 First model.
danger and uncontrollability of the street scene in real life. He can
act out his thoughts in a model world which is completely under (iv) Do other plants grow faster or slower?
his control. Similarly, a mathematical model enables us to concen- (v) Do side shoots grow at the same rate as the main stem?
trate on particular features of the situation, and to act out changes These questions are questions about plants, but they can be
and study their consequences in a way which would be impossible handled within a model, and the answers and predictions made
in the real world. The graphical model at this stage is purely de- within the model can then be applied back to the real world.
scriptive. But the graphical description also suggests questions: A next step in our thinking may be to improve the model. A
(i) Does the bean grow steadily? straight-line graph is both a simple model and one which displays
(ii) How fast is it growing? the continuity of plant growth (which the dot-graph of figure 2.1
(iii) How tall should we expect it to be next Friday? does not). It also contains answers to some of our questions.
10
GRAPHS AS MATHEMATICAL MODELS 11

height in
centimetres

123456

Figure 2.2 Improved model.

The straight-line graph represents a rate of growth of about 2

centimetres per week. According to the model, the plant will be
approximately 18cm tall next week. The model has enabled us to
predict. Of course, the prediction may not fit real life very well the
plant may die, or reach maturity and stop growing.
It is a feature of our world that a small number of graphical
models can be used to fit a great variety of situations fairly well. A
straight-line model is a fairly good fit for the velocities at various
times of a stone dropped from the hand, for the extensions of an
elastic band when pulled with different forces, for the quantities of l-igure 2.3 "lypical p.intboiic graphs: (i) ihe areas of squares: (ii) the path of a
potatoes needed for different numbers of school dinners, and lor projectile: liii) the low arch ol the Ro\;il Tweed Bridge. Berwick (after \V. \V.
many more situations. Saw\er. Mailicmuliciiin'a Dcliyht. Penguin I.
12 FROM GRAPHS TO CALCULUS

For pupils aged eleven, parabolic graphs can model the areas of
different-sized squares. Older pupils will find that parabolas are
also reasonably good models of the paths of balls when thrown,
and of the shapes of arches of such bridges as the Royal Tweed
Bridge at Berwick (figure 2.3).
Thus, a number of simple graphical models are worth studying
in their own right because they fit so many different situations. The
process of moving from the situation to the graphical model and
back is also an important subject for study in the secondary
school we need to discuss with pupils the limitations of the
models used. The parabolic model of the path of a ball ignores air
resistance, and is also a poor model of the path of the tip of a javelin
in flight.

40 50 mass in grams
2.2. Discontinuous Graphs as Models
cost in
pence
The graphical models which are used to represent real-life
situations are often, but not always, continuous graphs. For
example, consider the problem of devising a scale of postal charges
for letters of different masses. The principle of fair play might
40
suggest that the postage should be proportional to the mass of the
letter. But this model, the straight-line graph (figure 2.4(i)), is not a
suitable model for practical use for, although mass is a continuous 30
quantity, money is not and it is not much use choosing a model
which would direct the Post Office to charge 10-23 pence for a 20 -
letter. The only suitable model is that of a step-function such as
that shown in figure 2.4(ii). Step-functions turn out to be
fundamental in later work on areas, and so it is helpful to introduce 10
them early in concrete situations such as this.
A situation in which neither variable is continuous is shown in
10 20 30 40 50 mass in grams
figure 2.5. The straight-line model shown by a dotted line is
illuminating, in spite of the fact that interpolation is useless, there Figure 2.4 Scales of postal charges: (i) an impractical straight-line model;
being no shoe sizes between 4^ and 5, for instance. (ii) the step-function used for air-mail letters to America during 1976.
 GRAPHS AS MATHEMATICAL MODELS 13

give rise to straight lines and curves, and investigating the

x'
X
properties of these curves, they should not forget that behind their
42 - model lies the real world, and that the purpose of what they are
doing is to be able to transfer predictions from the models back to
s
s • the situations which gave rise to them.
40 xX ' 2i —— 34
CD
N
—— 35
'</>
X-" 3* — — 35g
8 42 —— 36
2.4. An Example
.C —— 361
en
c. 38 X 41 —— 37i
ro
<D ? —— 38
An example may help to make the development of the modelling
Q.
9
3
—— 39| process clearer. SMP Book A (p. 11) contains the following
LJJ | —— 40
passage which it uses as a starting point.
36 —
x'
*^ 7 2 ——
'
401
f- x 82 ——
7— 41
—— 42
^
34 'ill II I y

2 34567 8

British shoe size

Figure 2.5 Conversion graph for women's shoe sizes.

2.3. Aim of Graphical Work l2 32 42 62

i 9 16 25 36
At the beginning of the secondary school, children think of the What are the next two numbers of this kind?
graphs they draw as direct records of real situations. A major aim
of the work of the years up to 16 is to help pupils to see the Pupils might draw a graph of these square numbers, although
graphical model as an entity in its own right, a piece of the text does not ask them to do so at this stage. Figure 2.6(i) shows
mathematics which can be studied apart from the situation which a line graph, which many children might draw without much help
gave rise to it. However, this process of abstracting a mathematical on entry to the secondary school. During their first year, children
model from a real situation can be treated in such a way that pupils should be encouraged to replace the lines of this graph by points, to
do not always realize that it has taken place, so that the model, for give the result shown in figure 2.6(ii).
instance the straight line or parabola, seems to them to be divorced Should the points of this graph by joined up? If so, how?
from reality, and to have no connection with the real situation If the graph is regarded merely as a record of spots arranged in
which it modelled. The teacher needs to be sure that, while pupils squares, the points of the graph clearly should not be joined up, for
are working within the model, that is, studying equations which we cannot have a square of side 1\ spots.
14 FROM GRAPHS TO CALCULUS

16 16

3
3
cr cr

1234 1234
number of spots on side number of spots on side

Figure 2.6 The squares of positive whole numbers: (i) line graph; (ii) point graph.

However, squares turn up in other contexts as well as in ing chapters. However, we should note here that, when directed
arrangements of spots. We can draw a square of side 2-5 numbers and their multiplication make an appearance, we shall
centimetres, and we can also use the arithmetic process of squaring extend the quadratic or parabolic model which maps the real
the number 2-5. number x to x2, and which is usually written either y = x 2 or
A graph showing the squares of all the real numbers models the x i >x2, to allow x to be a directed number (figure 2.8). Pupils will
squaring of fractional, as well as of whole, numbers. Hence it is also get used to seeing the parabola in a number of different
more useful, and leads pupils towards a greater knowledge of the positions, and will eventually see that this model applies to a
real number system. Thus teachers may think it sensible to problem such as:
encourage pupils to join the points in figure 2.6(ii) by drawing a 100 metres of fencing is used to make three sides of a rectangular enclosure
smooth curve through them (figure 2.7), thus setting the graph against a wall. What is the greatest area which can be enclosed?
which was originally concerned with spots within a wider math-
ematical context. Thus we see that a major aim of the work is to give the parabolic
Later developments of graphical work are discussed in succeed- model, and the formula x i >x 2 for the function, a life of their own
GRAPHS AS MATHEMATICAL MODELS 15
square *
y

16

-4 -3
number

Figure 2.7 The squares of positive real numbers. Figure 2.8 Graph of .vi ».v 2 .

within the pupil's mind. By the age of 14 or 15, the more-able pupils development towards, and study of, these two general categories of
will be considering problems about the mathematical model and, graphical model. The unifying concept of a function is of particular
for instance, the gradient of the graph of XH->x2 can be studied importance in this study, and the next chapter discusses the
without immediate reference back to the real-life situations being concept.
modelled.

2.5. Looking Forward 2.6. A Recommendation

Among the many graphical models available, there are two The early treatment of graphical representation in secondary
types which are particularly useful. These are loosely described as schools is often disappointing. The location and the plotting of
continuous graphs and graphs of step functions (figure 2.9). In this points are usually introduced in the first year, but the subsequent
book, therefore, we shall concentrate particularly on pupils' development of graphical work seems to take two forms:
16 FROM GRAPHS TO CALCULUS

xi—» y

Figure 2.9 (i) A continuous graph; (ii) the graph of a step-function.

(i) Real situations are dealt with in chapters entitled statistics. graphs, their equations and the idea of functional dependence
(ii) The work which leads towards functions and the idea of could grow from children's primary-school experience of collect-
dependence is conducted in an abstract way, not relating the ing, recording and interpreting information in a systematic way,
ideas to graphs which may occur outside mathematics. so as to incorporate the graphs of measured data which pupils
If these two lines of thought could be fused so that Cartesian draw in other subjects, this would be a great improvement.
 3 The Idea of a Function
3.1. Dependence should meet activities which will enable them to develop the idea of
functional dependence. From these examples will emerge the
The idea of a function is one of the basic ideas of mathematics generality and power of functional dependence, and of the uses
and is particularly important in the uses of mathematics in the real which are made of functions in mathematics and other subjects.
world. It is derived in the first place from the fact that a quantity Like the man who discovered that he had spoken prose all his life,
may depend on one or more other quantities. As such, it is a very the pupil needs to discover that he is surrounded by functions,
familiar aspect of everyday life, which does not have to be whose behaviour he can begin to analyse mathematically.
artificially introduced into children's thinking. They know that the For some pupils, their ideas of functional dependence will
price they pay for sweets depends on how many packets they buy, remain simple, pictorial, and mostly qualitative:
that the BBC 1 television programme which is showing depends on The bigger the block of chocolate, the more it costs.
the time of day, that whether it is school or holiday time depends The time it gets dark in the evening depends on the time of the year.
on the time of year, and so on. In the secondary school, teachers
introduce their pupils to many functions: the science teacher shows The majority of pupils will be able to understand that different
them that the length of a spring depends on the mass hung from it, types of functional dependence are expressed by graphs of different
and that the pressure exerted by a given volume of gas depends on shapes, and to realize the significance of a linear function:
its temperature; the geography teacher discusses the fact that the Doubling the weight of chocolate doubles the price.
local time depends on the time zone in which a place is situated; the
physical education teacher demonstrates how the distance travel- They will by age sixteen be able to attach a good deal of
led by a cricket ball depends on a number of aspects of the significance to the steepness of straight-line and other graphs. For
thrower's action, such as the angle at which the ball is thrown. In some students, the building-up of a vocabulary of different types of
mathematics we expect our pupils to notice that the area of a functional dependence will be a necessary preparation for future
rectangle depends on its length and its breadth, that the distance a work. They will need to recognize quadratic as well as linear
ladder reaches up a wall depends on its angle to the floor, that the functions, and to realize the significance of inverse as well as direct
area of a circle depends on its radius; and we find many other proportion.
examples of dependence throughout mathematics.
The concept of function grew in order to express mathematically
this notion of dependence. For users of mathematics, as well as for 3.2. Historical Development of the Idea of a Function
mathematicians, it is one of the key ideas, for so much of both
mathematics and science is concerned with studying how one thing Like many other important ideas, the idea of a function has
depends on another. Hence all pupils in the 11-16 age group gradually been refined and clarified over the years, and has only
17
18 FROM GRAPHS TO CALCULUS

fairly recently been seen in its full generality. It is important that later we read in Hardy's great textbook Pur? M<if/icm«n'cs (1908):
teachers should be aware of this change, for even some recent texts All that is essential (to a function) is that there should be some relation between
which they may use or consult may be based on outmoded x and y such that to some values of x at any rate correspond values of y.
definitions of a function. Moreover, in the last few years some
school texts have adopted a point of view about functions which Hardy had abandoned the idea that a unique y should depend
does not seem to help pupils to realize their importance. We on each x. This admitted "many-valued functions" to mathe­
therefore now discuss the historical development of the function matics, so that to Hardy y^ = 1 — x^ was a function, but had the
idea, and how this is reflected in school texts. property that if x = 0, then y = + 1 or — 1. This generalization has
The following early definitions show how mathematicians been rejected by later writers, who have returned to Dirichlet's
groped to make their ideas of a function more precise. statement that a function is obtained by giving a #•»&' whereby
exact/y o;?e y is obtained from ear/? admissible x.
A quantity composed in any manner of a variable and any constants. (Jean Modern writers have also realized the importance of the scf from
Bernoulli/I 71X)
Any analytic expression whatsoever made up from that variable quantity and which the admissible values of x are drawn and the stV from which
from numbers or constant quantities. (Euler, I74KJ the values of)' are drawn, so that in elementary modern treatments
Quantities dependent on others, such that as the second change, so do the first. the concept of a function is developed so as to have three
are said to be functions. (Euler)
If a variable y is related to a variable .\, so that whenever a numerical value is constituent parts:
assigned to .\- there is a rule according to which a miiVytic value of y is determined. (i) a starting set, called the (/«mm;i of the function, whose
then v is said to be a function of the independent variable x. (Dirichlet, 1837) members are the admissible values of x;
These early definitions reached out to express in words the idea (ii) a target set. called the co&jmam of the function, a single
of the dependence of one quantity on another, together with the member of which is attached to each x;
idea that the second quantity is imi'que/y determined from the first (iii) a set of arrows or a ni/c, to show which member of the
by some ni/e. codomain depends on each member of the domain.
At first, this rule had to be expressed in a single formula, but by This idea of a function is shown diagrammatically in figure 3.1,
the time of Dirichlet the single algebraic formula had been found to and is very simple and suitable for development with pupils
be too restrictive. For example, if x is a counting number greater rule
than I, the rule
v is the highest prime factor of x

gives rise to a perfectly good function, according to Dirichlet's

definition, although the rule is expressed in words rather than
symbols.
Unfortunately, as time went on. Dirichlet's definition, which is
so suitable as a basis for the classroom introduction of functions,
was modified in the search lor further generality. Nearly a century Fiuurell
THE IDEA OF A FUNCTION 19
between 11 and 16. The rule expresses the functional dependence. A more general function is shown in figure 3.3. Here we write
For example, the area of a circle of radius r is nr2 ; the rule is x i >/(x). The use of/'(x) to denote the member of the codomain to
which x maps is becoming more general in texts intended for pupils
i- maps to nr2 .
up to the age of 16, as is the use of a single letter such as /'to
This rule is easily given by a formula, but many rules are not. For represent a function.
the height of a growing plant (section 2.1), the rule is that the f
number of weeks since planting maps to the height of the plant.

3.3. Elementary Notation and Wording

It is convenient to have a standard vocabulary for describing

functions, which can gradually be built up with pupils. Most
Figure 3.3
people talk about the arrow in figure 3.2 as "3 maps to 9" or "3 goes
3.4. Use of the \\'ord function in "Traditional" Texts

Many books still in use in schools show considerable traces of

older views of function; so does the vocabulary of many teachers.
Pupils may be seriously confused when the teacher's colloquial
expressions are not in accordance with the "modern" definitions
which he uses.
Durell is a typical author of the "traditional" school, and his
Figure 3.2
usage is well represented by the following quotation from A New
to 9". If the rule is that each member of the domain maps to its Algebra for Schools (p. 157), written in 1930:
square, we can say more generally that, for a given member .v of the
Since this graph represents the relation between a number x and its square .v 2 ,
domain, it is failed the yraph of the function jc 2 .
.v maps to \ 2 . Any expression, containing x. w hose value can be lound when the \ alue of .v is
2.v - 1
given, is called a function of v. Thus 7,v. ^.v - 5. - . .v( 5.v. etc.. are all
This is often written .v + 3
1'unclions of .v. The letter v is generally used to represent the function of .v.

or For Durell, as for Hardy and all his predecessors, the formula
was the function. They talked about "the function 7.v". So do many
In this book the second of these notations is used. teachers at the present time. However, most modern authors
20 FROM GRAPHS TO CALCULUS

regard 7.v as the member of the codomain of the function which Thus, traditional authors, by emphasizing the formula for a
corresponds to the member .v of the domain (figure 3.4). Thus, in function, emphasized the idea of functional dependence. Unfor-
modern usage, the function which Durell spoke about as "the tunately, the words "the function 7.x" lose the idea of mapping. It is
function 7.x" consists of: .x which maps to 7.x, and this vitally important link between .x and
(i) a domain, which is usually assumed to be the set of real its image 7.x needs emphasis for young pupils. We need to say
numbers, if there is no statement to the contrary;
the function: .x maps to 7.x
(ii) a codomain, which again is usually assumed to be the set of
and to write
real numbers; A'i > 7.x.
(iii) the rule: x\—>7.x.
»7x Moreover, the suppression in traditional texts of the domain or
the codomain of the function sometimes leads to difficulties over
the use of functions as models of real situations.
For example, a function with rule .v i > .v 2 models the areas of
squares which can be built with the rods of structural apparatus.
But this model only applies to structural apparatus when .x is a
natural number. The domain of the function is the set N of natural
numbers.
Similarly, a function with formula s = 5t 2 models the distance s
Figure 3.4 metres fallen in time t seconds by a stone dropped from the hand,
but the model only applies to the situation when t is positive. The
Many teachers have been brought up to speak of "the function domain of the function is the set of positive real numbers.
7.x" and it requires considerable concentration in the classroom to Thus, the modern view of a function as having three parts, a
replace this in everyday speech by domain, a codomain and a rule, is most helpful to pupils in
the function: .v maps to 7.v. understanding the power of the function concept, and how
functions are used in modelling real situations. This development
The view that the formula is the function persisted in school
of the meaning of the word function in school mathematics started
textbooks for many years, so that in 1959 Channon and McLeish
with the introduction of modern mathematics texts in the early
Smith wrote, in General Mathematics Book 4, p. 152:
1960s.
Any algebraic expression which involves the variable .v (and no other However, the treatment of function in most modern texts has
variable) is called a function of x, and its value depends on the value of x. The been greatly influenced by another mathematical point of view
symbol used is/(.v), which is re'ad as "function of .\-";/{2) means "the same
expression with 2 written instead of .x", / ( 1) means "the same expression with which seems to be unhelpful for young pupils, and so to be
- 1 written instead of x", and so on. regrettable. This is described in chapter 5.
 4 Functions in the Early Secondary Years
4.1. The Introduction of Functions cost in
pence

The idea of'functional dependence will have been taught by using

examples in which the members of one set depend on the members
of another set (sections 1.3 and 3.1).
40
More examples can be found within the mathematics in the
classroom. Some are:
30
(i) The image of a point mapped by a reflection depends on
which point is mapped. 20
(ii) When a square is enlarged, the perimeter of the enlarge-
ment depends on the scale-factor. 10 -
(iii) The volume of a cube depends on the length of the edge.
(iv) Each arithmetic operation produces examples of functional
dependence. In the "seven times table", the answer depends
10 20 30 40 50 mass in grams
on the starting number the answer is always 7 times the
Fimire 4.1
starting number.
On what does the postage payable on a letter depend: its length
Straight-line models are appropriate in (ii) and (iv). and width, its mass, whether it is sealed?
In science, and in other subjects, further examples of functional What is the postage on a 35-gram letter?
dependence are studied. For instance, in Nuffiehl Physics Book !, What point does 20 map to?
different masses are hung from the ends of springs and elastic What can we say about a letter which costs 30p?
bands, and measurements are made (including testing to de- What numbers can be mapped by the function?
struction). The model of a straight-line graph is not a good fit near What is the domain of the function ?
the elastic limit.
In all these examples, ideas about functions can be brought out
in discussion or exercises. 4.2. Graphical Representation of Functions
For example, figure 4.1 shows how the airmail postage for a
lelter lo America depends on its mass in grams in 1976. Traditionally, the graphical representation of functions used in
The following questions apply to this graph : schools has been a Cartesian graph. Today several other methods
21
 FROM GRAPHS TO CALCULUS

of showing functions visually are available. Each of these pictures accommodates domains and codomains which have no particular
brings out particular features of the concept of function, and has its structure, and so which need not be sets of numbers. Many pupils
own strengths and weaknesses. Pupils should be able to choose will have used it in the primary school.
from a variety of different representations, according to the
particular function or aspect of a function in which they are
interested at that moment. They therefore need to be introduced to
these representations at appropriate times in the early secondary
years. It is also illuminating for pupils to use more than one
representation of the same function.
Some of the strengths and weaknesses of the various forms of
diagrammatic representation are summarized below, so that the
teacher can help pupils to choose the most appropriate one for a
particular situation. The teacher will realize that some of the points
listed do not apply to 11 and 12-year-old pupils, but he needs to
help pupils to build up a vocabulary which they will use later.

(i) The basic arrow diagram

Figure 4.2 emphasizes the three parts of a function: the domain,
codomain and rule. Pupils can write the names of the sets used
against them, and can express the rule in words on an arrow. This is
the most general way of representing a function pictorially, for it

rule

Figure 4.3

The main strengths of the basic arrow diagram are:

(i) It is very simple indeed, so that it can be used with
confidence by the least-able pupils.
(ii) The domain and codomain need not be sets of numbers;
this diagram can illustrate some examples for which other
graphs fail, ranging from the names of pupils' pets to linear
Fiaurc 4.2 maps of vector spaces.
FUNCTIONS IN THE EARLY SECONDARY YEARS 23
XI—>X+ 1

X \—» X + 1

/-1 0

Figure 4.4 Arrow diagrams and Cartesian graphs for some important functions.

(iii) It brings out properties of the function, such as one-one, Its main weaknesses are:
onto and the existence of an inverse function. (i) If the domain has a large number of members, the diagram
(iv) It shows composite functions very easily. is messy,
(v) It gives a diagrammatic representation of the most general (ii) If the domain has an infinity of members, the diagram can
function without any special features. only be symbolic of the function.
 FROM GRAPHS TO CALCULUS

(v) For calculation and other numerical purposes, a more

sophisticated representation is needed.

(ii) The arrow diagram for functions with numerical domain and
codomain
This is an adaptation of (i) to the case when the domain and
codomain are sets of real numbers. It shows the order of the
numbers and contains the important idea of scale. It is also the
most basic representation in which different functions produce
different-shaped diagrams. Arrow diagrams for a few of the most
important functions are shown in figure 4.4, where their Cartesian
graphs are also shown for comparison. The teacher will realize that
for each function, the two representations stress very different
aspects of the function. In all cases, the domain and codomain of
the function drawn is the set of real numbers.
More experience is needed to associate these simple functions as
easily with their Cartesian graphs as with their arrow diagrams.
xi->2x+1
Able and average pupils should certainly be expected to recognize
the shapes of the Cartesian graphs of the functions shown, and
others, but the arrow diagram, which is often more meaningful to
beginners, gives a useful lead towards realizing that particular
types of functions produce particular patterns of diagram.
Some strengths of the arrow diagram are:
(i) It is very simple and is comprehensible to almost all pupils.
(ii) Good pictures for simple functions. Linear functions
connect easily with transformation geometry.
(iii) The order of the numbers is emphasized. Functions which
(iii) If the domain and codomain are subsets of the real preserve order are distinguished from those which do not.
numbers, the diagram does not show the order of the (iv) The diagram for a composite function is easily constructed,
numbers. (v) The diagram for the inverse function (if any) is easily
(iv) It does not attach a shape of picture to any particular constructed.
function. (vi) It brings out whether the function is one-one or many-
FUNCTIONS IN THE EARLY SECONDARY YEARS 25

one. fresh eyes, and to realize how much building-up of concepts he

(vii) The derivative of the function is meaningful as the local brings to his own interpretation of a Cartesian graph. The reader
scale-factor (but see chapter 14). should notice how automatically he himself draws a smooth curve
Its weaknesses can be summarized as: through a few plotted points, the meaning he attaches to positive
and negative gradients as indicating increasing and decreasing
(i) It can only be symbolic of the function unless the domain functions, the particular significance he gives to points of zero
has a finite number of members, gradient, the range of knowledge he brings to the interpretation of
(ii) Unless the function is of the form xt-^m.\ + c, the diagram straight-line graphs. At the beginning of their secondary-school
may become messy, education, pupils have none of these ideas available to them, and
(iii) There is no representation of the integral of the function. one of the most important tasks is the development of their grasp
of the significance of different features of a Cartesian graph.
Some of the strengths of Cartesian graphs are:
(iii) The Cartesian graph
(i) Familiarity. Cartesian graphs are a part of our culture, and
The Cartesian graph is the most familiar of all the illustrations of pupils must learn to use and interpret them,
a function. Hence, it is very difficult for a teacher to look at it with
(ii) The order of the real numbers is clearly shown in both
domain and codomain.
(iii) A knowledge of the shapes of Cartesian graphs of the
commoner functions is an aspect of mathematical literacy,
(iv) The Cartesian graph is the only representation in which the
concept of a continuous function is visually meaningful,
(v) Good for sum and difference functions, which are easily
constructed.
(vi) The derivative is meaningful as the gradient of the graph,
(vii) The integral is meaningful as the area under the graph.
Its main weaknesses are:
(i) It only shows the mapping arrows implicitly, so it needs
careful teaching in the early stages,
(ii) It can only be used when both the domain and codomain
are sets of real numbers,
(iii) Pupils may think that the complete set of real numbers is
the domain of every function,
(iv) It does not emphasize whether the function is one-one or
Fis>ure4.5 many-one.
26 FROM GRAPHS TO CALCULUS

(v) The graph of the inverse function (if any) is not immediately We often speak about the processes of arithmetic. Addition is a
obvious, although it can easily be found, function which needs two inputs. If we feed in 2 and 3, the addition
(vi) It is not good for composite functions, for which new function produces 5. The function machine illustration is the only
graphs always hnvp
aranlr: alwnvc have tn
to hp
be Hrau/n
drawn. elementary diagram for this situation. It can also be used to focus

25
(iv) The "function machine"

Figure 4.8

input output attention on the algebraic formulation of a functional rule (figure

4.8), and it shows composite and inverse functions very easily
Figure 4.6 (figure 4.9).
5 25
(a, b) i-> a + b XI—>X2
Many mathematical concepts are important because they have a
number of different aspects which make them appropriate for use 3
in different contexts. The idea of a function is one of these.
Functions with particular formulae are used in modelling because
rules found in real-life situations are usually approximately 10
regular. The rule does the same to each element of the domain. For function
example, the function xt >7.x multiplies every number by 7. We
can input any real number x, and the function outputs 7x. The
10
"function machine" diagram is important because it enables this inverse subtract 3
processing aspect of a function to be made vivid to pupils. The
pupil can imagine himself feeding the input of 5 into the "black Figure 4.9
box", which outputs 35. Any appropriate input of a member of the
domain is transformed or processed by the function. The conceptual link with computing is also important. A
computer is a black box into which certain numbers or other
information are fed. The computer prints out further information,
which is a function of the input. The program consists of the
add instructions for the required function. Pupils need to think of a
computer as a processing machine, and so as a machine which is
programmed to generate the result of applying a function to an
Figure 4.7 input.
FUNCTIONS IN THE EARLY SECONDARY YEARS 27
4.3. The Labelling of Cartesian Graphs f(x)
In this book we have suggested that graphs be approached by
keeping records of information obtained from real situations. A
function may then be constructed to model the situation, and a
Cartesian graph of the function may be drawn and studied. For
instance, if a stone falls from the top of a cliff, the number of metres
fallen in t seconds is fairly well modelled by the function with
formula ti >5f2, whose graph is shown in figure 4.10; a problem
then arises about the labelling of the vertical axis.

60

40

20

0123 t

Figure 4.10 A function which models the vertical fall of a stone.

Some texts follow one of the policies illustrated in a general

situation in figure 4.11, but these lead to difficulties if two functions
are to be drawn on the same axes, and the second is particularly
inappropriate in the case of a function such as .vi > 1. Figure 4.11 Labelling of the vertical axis: two recent usages.
28 FROM GRAPHS TO CALCULUS

The traditional method of labelling the axes of graphs, using x

and y or t and s as labels, has considerable advantages, and arises
naturally in the context of plotting points given by ordered pairs of
numbers. Teachers will always give pupils sets of points to plot and
eventually, whether or not set language is actually used, sets such
as {(1,2), (2,4), (3, 6)} and then {(x,y):y = 2x}, will make their
appearance. This leads naturally to the traditional labelling of
graphs shown in figure 4.12, where y = 2.x is called the equation of
the set or graph. When pupils realize that the point (3, 6) of the line
y = 2x indicates that 3 maps to 6 under the function x i > 2x, a
linking of the two points of view becomes apparent.

Note on the equation of a graph

= 2x The coordinate plane is the set of points defined by {(\, y):x e R, v e R}. Since
a function is defined by the set of ordered pairs {(x,f(x)):xeD}, where D is a
(3, 6), subset of R, we consider the set of points given by {(x,y):y = J'(x),xeD} to be
the graph of the function/ y =/(.v) is called the equation of the graph of/
In the above example, y — x 2 is the equation of the graph of the function /for
which/(.x) = x2 , as shown in figure 9.
(2, 4); Example 1. Figure 10 shows the graph of the quadratic function/given by
/(x) = x 2 4.x, with domain {.x: 2 < jc < 6, xeR},i.e. these! of real numbers
from 2 to 6 inclusive.

'(1. 2)

Figure 4.12 Labelling of the vertical axis: traditional usage.

This linking is ably, if rather formally, expounded in the books of

the Scottish Mathematics Group, Modern Mathematics for
Schools, Book 5 (second edition), p. 58: 10
FUNCTIONS IN THE EARLY SECONDARY YEARS 29

Thus both methods of labelling shown in figure 4.13 are

acceptable practice and are convenient in different circumstances.
In (i), the graph is labelled to show the function x i-»-.\-2, and in (ii)
the equation y = .x 2 of the graph of the function is emphasized.

0 1

In this book both notations will be used ; whichever one is more

convenient will be chosen for each particular example.
The writers of the SMP texts use a different style of labelling on
the axes, as the illustration above shows (Bonk A. p. 21).
Unfortunately, this style leads to considerable confusion in pupils'
minds for, although the equation of the horizontal axis certainly is
y = 0, the points 0, 1,2,... marked on it are values of x. Moreover, Figure 4.13 Labelling of graphs: two acceptable practices.
30 FROM GRAPHS TO CALCULUS

.a

Figure4.14 Labelling of an arrow graph.

the practice may seem to pupils who do not appreciate the finer
points of logic to be at variance with the usual style of labelling
when physical quantities are graphed, as in the following extract,
which is taken from SMP Book C, Teachers'" Guide (p. T97).
If we let x stand for the number of hours and y for the height of the candle in
cm, then the relation between x and y is
x+y = 8 or .x = 8 y or y= 8 x
_c
0>
If the height of the candle after n hours is h cm, the axes of the
Cartesian graph could certainly be labelled h = 0 and ;i = 0, but
no such labelling is possible for the arrow graph, although the
labelling shown in figure 4.14 would still be possible.
1 OT >1 C C -7 Q
An additional point which arises from the extract is the conflict
number of hours of conventions between teachers of science and mathematics over
(b) the labelling of graphs. This controversy will no doubt continue to
FUNCTIONS IN THE EARLY SECONDARY YEARS 31

be bitterly fought out between science teachers, who teach that /? when elsewhere in mathematics this use of a bracket denotes
stands for a height, and that the associated number is /i/cm, and multiplication. Our more logical science colleagues write either
mathematics teachers, who usually insist that a letter stands for a
number, and so write that the height of the candle is h cm. height of candle in cm
However, from either point of view, there seems little merit in or
height of candle (cm), height of candle/cm.
 5 Further Developments of the Function Idea
5.1. Introduction
or
We have seen that in "traditional" texts, a formula was regarded (a2,b2 ), (a3 ,b2 )
as a function. More "modern" texts, published since about 1960,
is quite sufficient to specify the function shown, without a verbal
regard a function as having three parts:
description of the rule. Indeed, no convenient verbal description of
domain, codomain and rule. this rule exists. Hence, all that is needed to describe the function is a
In previous chapters, we have emphasized the importance of the
idea offunctional dependence. However, a number of modern texts
treat functions very differently, regarding them as particular
examples of relations.
This approach is unhelpful to pupils' understanding of
functions, and should not be used for the reasons given in the next
section.

5.2. Doing without the "Rule"

The treatment of functions as special cases of relations has its

roots, as have all the previous developments of the idea of a
function, in the advanced work of mathematicians. One of the Figure 5.1
three essential components of a function is the rule, which describes
how each member of the domain maps to a member of the list oRordered pairs. When mathematicians define a function, they
codomain. can do without the word "rule" and merely list the linked elements
In recent years some mathematicians, concerned with the logical in ordered pairs.
and conceptual foundations of their subject, have taken the view This idea has given rise, in advanced work, to a treatment of the
that the use of the word rule begs questions. They have avoided the idea of a function which is more general and so might be
use of the idea of a rule and realized that all that is needed is a list of considered to be more logically satisfactory. The order of pre-
pairings, saying which member of the codomain is to be attached sentation of ideas in this development (which avoids the idea of
to each member of the domain. In figure 5.1, the list of pairings "rule") is as follows.
32
FURTHER DEVELOPMENTS OF THE FUNCTION IDEA 33

(i) The idea of a set is first studied. development that working mathematicians discard the idea of a rule
(ii) The idea that two elements a and b can be put together to in favour of a subset of A x B.
form an ordered pair (a, b) is introduced. Children of 12 and 13 years of age are not yet concerned with the
(iii) This enables the Cartesian product A x B of two sets A and logic of axiomatic development. They are still learning the tools
B to be defined as the set of all ordered pairs (a, b) such that of the mathematician's trade. When teaching beginners about
aeA,beB. functions, it is very unwise to suppress the idea that functional
(iv) A relation from A to B is defined to be a subset of A x B; dependence is expressed by a rule for mapping. Functions are so
that is, a relation is any set of ordered pairs (a,b) whose important in their own right that they should not be taught
members are drawn from A and B respectively; all mention as particular (and perhaps unimportant) types of relations.
of the word "rule" is avoided by simply giving a list of pairs Functions are used as mathematical models in situations where
which would have been linked together by the rule, dependence needs to be expressed. In many "modern" texts this
(v) A function f with domain A and codomain B is now a idea is not clearly brought out.
relation from A to B with the additional property that each The introduction of "modern" work, as at present seen, has not
«eA is the first member of exactly one ordered pair improved pupils' understanding of the key concept of a function.
belonging to the function. In traditional texts, functional ideas were introduced late, and the
Thus a function is a set of ordered pairs (a, b) such that point of view used was not in accordance with the present
(a) the first member a of each pair belongs to A, mainstream of mathematical thinking. By contrast, the "modern"
(b) the second member b of each pair belongs to B, orthodoxy is to introduce functions early, but to minimize their
(c) each member of A belongs to exactly one ordered pair of the importance. This seems very odd.
function and so is uniquely attached to a single member of B.
In other words a function can be considered as a particular type of
relation. 5.3. Review of Texts
This definition of a function is logically satisfactory in an
axiomatic development of mathematics in which the aim is to
present each concept in terms of those ideas which have been The changes described above have only become part of the
previously defined. common currency of mathematics within the last twenty years, so
Unfortunately, the idea of functional dependence has been that different texts in use in schools today show different stages of
totally eliminated from this formal definition of a function. In the historical growth. Moreover, different authors have adopted
process of generalization, the rule which was the essential idea of different definitions for the same words, leading to a most
the function has vanished. unfortunate state of confusion.
Inquiries among mathematicians who use functions as basic The date of publication is no guide to the view of functions taken
tools of their thinking show that working mathematicians usually by an author. The latest editions of Durel! still firmly hold that the
think of functions as rules for mapping, rather than as sets of formula is the function (C. V. Durell. Certificate Mathematics,
ordered pairs. It is onlv when they need the logic of an axiomatic Volume 2 (p. 150), second edition, 1971).
34 FROM GRAPHS TO CALCULUS

Graph of y = x2 SMP Books I to 5 were among the first "modern" texts. For their
25 1 1 | 1 1 1 1 1 ! n | ! 1 1 1| | | .1 1 1 1 1 1 1 1
y ::—:::_. .. :::„!. :...—
authors, a function was a particular relation (SMP Book 2 (1966)
EEEEEEEEEEEEEEEEEEE|EE|EEEEEEEj p. 158).
.__ _U__- __________ -^--- --J----L--

20 ... — —.- — — ..-- — A relation is a connection between members of two sets or members of the
same set. Figure 8 represents the relation "has the prime factor" between the sets
{4,6,7,15,16,20} and {2,3,5,7}.
15 ...................... ....|— -——-/— -^-—-ii-i--— We can call this relation a mapping, saying that each number is mapped onto
7 its prime factors. The set on the left in figure 8 is called the domain.
::::::TZ:::::::::::::::::::::::: A mapping is a function if each member of the domain has only one image, for
. ^ -. . . J example, the mapping which maps each number onto its smallest prime factor is
:-:p:-i:"::ii:----:|^-j a function.
1o;EEEEEEEE!:EElp::i:::" Figure 9 shows this mapping for the two sets above. Only one arrow starts
' ^4r-
' i i x— from each member of the domain. The image set [2,3, 7} is called the range.
::::::::::::::::::::::::::- ^- — ----Lr— — ----- — The mapping represented in Figure 8 is not a function. The ranee in this case
i 1 is [2,3, 5, 7}. "
5 ::::i::::::::rr:::::^::::
---L— — - -4— -j — -44+ — ~ Another word appeared in these texts: SMP, at that time, used
a**
( "'
mapping as another word for relation. They said that a mapping
Q ----.=•"---.-——.--—- ::::±:::::::±:::::::::::::::::: was a function if each member of the domain had a unique image.
0 1 2 3 4 5 x
Writing at about the same time, Mansfield and Bruckheimer in
> Fig. 181 Book4 of Mathematics: A New Approach (1965), which is also in
the "particular relation" school, avoided both words relation
The graph in figure 181 represents the relation between a number .v and its and function, using the word mapping for what is usually called a
square x 2 : it is called the graph of the function x2 . Figure 181 also represents the
graph of y, where y = x 2 , for values of x from 0 to 5. relation, and described functions as one-one or manv-one map­
pings (Mathematics: A New Approach, Pupil's Book 4 (1965), pp.
Channon, McLeish Smith and Head (New General Mathe­ 10-11).
matics 4 (1971) p. 102) use both the "particular relation" and the
"function is the formula" views. Any system by which every member of one set called the domain sat (in the
above example, the set of numbers of therms) has associated with it a member of
Functional notation another set called the range set (in the above example, the set of gas bills) is
A many-to-one relation is called a function. called a mapping.
As a one-to-one relation, often called a one-to-one correspondence, is a special The only necessity in a mapping is that every member of the first set should
have an image and that the mapping should provide a means of finding that
case of a many-to-one relation, it is also a function.
Any algebraic expression which involves the variable x (and no other image. We can, indeed, have mappings where a member of the domain has
variable) is a function of x, and its value depends on the value of x. The symbol several different images. These mappings are called one-many.
used is/(x), which is read as "function of x" ;/(2) means "the same expression
with 2 written instead of x",/( 1) means "the same expression with 1 written In the Teacher's Book 4, they explained their policy. They were
instead of x", and so on. aware of the changing use of the word function, and attempted to
FURTHER DEVELOPMENTS OF THE FUNCTION IDEA 35
were appearing, the following discussion of relations, mappings
and functions appeared in SMP Book D. Teachers Guide (1970),
p.T231:
The following definitions of the different types of mappings are included for
the benefit of the teacher.
A relation is many to one if the image of each element of the domain is a single
element of the range. A relation is one to many if the inverse image of each
element of its range is a single element of the domain. A relation is one to one if it
is both many to one and one to many. A relation is many to many if it is neither
one to many nor many to one.
(iv) The relation 'is the brother of defined on the set of brothers and sisters
[Jane, Janet, June, Jack] is an example of a one to many correspondence.
Many to one and one to one relations are mappings.
It might be worth noting here that, in this course, unlike the SMP O-level
course, we have chosen to use the more commonly recognized definition of a
mapping: that is, a relation in which each member of the domain is related to
one and only one member of the codomain. This statement corresponds to the
definition of a function given in Book 2, p. 158, and the authors have decided
that in Books A-U, in order to keep the mathematical language as simple as
possible, the use of the word 'function' is therefore unnecessary.

Thus, in this series of texts, a mapping is a synonym for a

function, rather than for a relation. This usage is in accordance
with the usual meaning attached to the words map and mapping in
advanced work, when a mapping is always a function (except in the
texts of the Open University). However, functions have now been
downgraded almost out of existence in the SMP lettered books.
The word range has also caused problems. Mansfield and
Domain. Fig. 9
Bruckheimer use range for the set which we have described as the
resolve the dilemma by avoiding the word (Mathematics: A New codomain (Mathematics: A New Approach, Pupil's Book 4, p. 10).
Approach, Teacher's Book 4 (1965), p. 16). For example, consider the set of all the people in a town as domain. Map each
person onto the number of hairs on his head. The range is the set of natural
It is becoming common practice to use the word 'function' for one-one and numbers and zero.
many-one mappings only. Since this usage is not universal and some authors
still use such phrases as 'many-valued functions' (which, in the new use of the It is more usual to adopt the policy of SMP Book D, and to
word 'function', is a contradiction in terms), the word is not used in this book. reserve the word range for the set of actual images, which is a
It is not necessary, of course: the more general 'mapping' is entirely adequate.
subset of the codomain.
Unfortunately, the word mapping has also been used with a The definitions and notation used by the Scottish Mathematics
variety of meanings, so that in 1970, when the SMP "letter" books Group are in accordance with those usual in advanced work. The
36 FROM GRAPHS TO CALCULUS

first section of Modern Mathematics for Schools, Book 5, chapter 3 Notation: If a function/maps an element x of set A to an element y of set B, we
(second edition, 1973) is headed "Mappings, or functions". The write/ :.x -»3', which may be read '/maps .x to /'.
v is called the image of .x under/ and the set of images form the range of the
authors are still members of the "particular relation" school, function.
although they emphasize the importance of functions. This Thus for a function we require:
approach to functions also necessitates their very late (i) a set A, called the domain of the function;
(ii) a relation which assigns each element of A to exactly one element of B.
introduction in this case not until Book 5 . The following extract The set of images in B is called the range of the function.
is the key to what follows (MMS Teacher's Book 5 (1973), pp.
50-51).
1 Mappings, or functions
5.4. Recommendations

In Book 3, we looked at some relations and mappings from one set to another The nomenclature used about functions in school texts has
set. In the present chapter, we develop these ideas further and study the
important concept ofjunction.
become very varied since about 1960, so that a word which is used
with one meaning in one set of texts may be used with another
Example 1. Let A = {1,4,9} and B = {1,2,3,4}. Show in an arrow diagram
the following relations from A to B: (different) meaning in another set of texts. This causes considerable
(i) is greater than (ii) is the square of
confusion for pupils and teachers alike. Pupils frequently change
schools and texts, and good teachers consult more than one text,
and often draw exercises from a number of places. Moreover, many
A B students who proceed into higher education may have to modify
sthesquareof/ their vocabulary at the beginning of their course. There seems no
good reason for the present confusion of vocabulary to be
perpetuated. When choosing a vocabulary for school use, the
vividness with which it conveys the ideas to beginners, its
memorability and its clarity are important considerations. There
seems very little to choose between different versions on these
grounds, so that this criterion gives little guidance, except that
phrases such as
3 maps to 21
In (ii), the relation is a mapping since each element of A is related to exactly one
element of B as is shown in figure 1 (ii) where one arrow leaves each element of A, give a vivid dynamic picture of what is happening, and should be
Relations which are mappings are of prime importance in mathematics and
are often referred to as functional relations, or simply functions. A function is preserved.
therefore another name for a mapping. Both terms are useful; the idea of a In higher mathematics there is a considerable measure of
mapping as a kind of operation helps to give a picture of a function. agreement about the vocabulary of functions, and confusion
A function, or mapping, from a set A to a set B is a relation in which each
element of A is related to exactly one element of B. We write A -> B (A maps to would be diminished if this vocabulary were universally adopted.
It is illustrated diagrammatically in figure 5.2.
FURTHER DEVELOPMENTS OF THE FUNCTION IDEA 37

A relation associates some elements of a first set the domain with some
elements of a second (or the same) set the coilomuin.
/^function associates with each element of the domain one and only one
element of the codomain. The elements of the codomain which are "images" of
elements of the domain form a subset of the codomain the range of the
function. Thus we may have a "many-valued relation" but not a "many-valued
function".
We recommend that mapping be synonymous with function.

Another major problem of notation for functions has no clear

solution. For us, a function has three parts: the domain, the
Figure 5.2 The vocabulary of functions: every member of the domain has exactly codomain and the rule. We have abandoned Durell's view that the
one arrow starting from it; the set of images is called the range of the function.

The variety of uses of the noun mapping and the verb to map
in school mathematics presents particular difficulty as they have
both been used in connection with both functions and relations.
In advanced work, the noun mapping (sometimes abbreviated
to map) is only used to refer to functions (except in texts of the
Open University) and its use in schools should also be confined to
Figure 5.3
functions. Similarly the verb to map should be confined to
functions or mappings, and used as a vivid way of talking about the formula is the function. We should no longer speak of "the function
mapping arrow in sentences such as "Each person maps to the 7.x". In advanced work, a new notation has grown up to describe
number of hairs on his head". such a function. Many mathematicians write
The use of the verb to map in connection with relations should /:R->R given by /(.x) = Ix
be avoided. The words is related to can always replace the words
maps to in relations, and so reduce the danger of ambiguity. For to express the function with domain the set of real numbers R,
example, consider the relation "is joined to" illustrated in figure codomain R and rule x i > 7.x. This is unwieldy for everyday use by
5.3. It is better to say "B is related to D" than "B maps to D". beginners, who are often not interested in the domain and
Thus we concur with the following recommendations of the codomain as much as they are in the rule. In this case it is better to
Teaching Committee of the Mathematical Association (Mathe­ use the abbreviated form of words
matical Gazette 4\8, December 1977): the function ,xi > Ix (the function x maps to 7.x)
There are considerable difficulties in the definition of some of the common
terms; our aim is to suggest a scheme which is usable in the sixth form, with or if the function needs a name:
obvious consequences at earlier stages.
The idea of a relation is a very general one in mathematics. For school we the function/:xH->7.x
make the following recommendations. (the function/under which .x maps to 7.x).
38 FROM GRAPHS TO CALCULUS

This notation has already been adopted in the majority of modern cube plus one. This can be described as x i-+ x3 +1, but the
school texts, and presents no difficulty to the learner. temptation to say
However, the Mathematical Association recommendations
continue: the function x 3 + 1
A function may be denoted by a single letter/ Generally by/(x) we mean that is strong, and certainly this "abuse of notation" is convenient.
clement of the range which is the image of the element x of the domain under the However, if notation is abused before the learner grasps the
function/ but it is sometimes convenient to use/(.v) in an "old-fashioned" way
to describe the function itself (usually where a particular function is involved, concept he is being taught, it may prevent him from fully
e.g. "the function ,\ 2 3.x+ 2"). appreciating the concept. It is better to talk about
The arrow »is used to mean both "is mapped onto" and "tends to". Usually
there is little chance of confusion, but if distinction is necessary the "barred the function x maps to x 3 + 1
arrow" t-> may be used to indicate that an element of the domain is mapped to
an element of the codomain.
until the learner is thoroughly familiar with the constituent parts of
The problem referred to in the second sentence is that a general a function: domain, codomain, and rule. Thus, the "abuse of
function can be described by a single letter/, but there is no equally notation" which talks about "the function 7x" should be avoided
brief way of describing the function which maps each number to its before the age of 16. '
 6 Some Special Graphs
6.1. Straight-Line Graphs and Proportionality example, x might represent the number of kilometres travelled by a
car on a motorway and y the number of litres of petrol used; or x
Functions whose graphs are straight lines through the origin might be the number of people in a household and £y the cost of
turn up so often as models of situations in the real world that a their food per day.
particular jargon has developed to describe them, and is used as In the first case the car may travel about 10 kilometres for each
everyday language by scientists and mathematicians. This jargon litre of petrol, and y = -^x is then the equation of the graph used as
a model. However, in this situation x can only take positive values,
yi\ and so we are concerned with the function .XH-^.X with domain
the non-negative numbers.
In the second case, if it costs about 50p per day to feed a person,
then y = jx. It is obvious that this statement is only approximately
true, for not everyone eats the same amount of food, and it is also
clear that .x must be a positive integer or zero. However, y = %x,
although only an approximation, acts as a mathematical model
which says something useful about the cost of running a house-
hold.
These graphs and functions are often described in the language
of proportionality
y is directly proportional to .x
or
y is proportional to .x.

Figure 6.1
The two examples above are often described as "the petrol used
by a car on the motorway is proportional to the distance travelled"
of proportionality needs to be known by pupils, who often do not and "the cost of the food used in a household is proportional to the
realize that they are dealing with linear functions under a different number of people in the household".
guise. More precisely, the statement that "y is directly proportional to
A straight line through the origin is a graphical model which .x" means that y = kx for some fixed value of k. In the cases in
describes either exactly or approximately many everyday pheno- which the statement "y is directly proportional to .x" is used, the
mena, and can be easily used to make predictions about them. For domain is the set of values of x which are appropriate to the
39
40 FROM GRAPHS TO CALCULUS

situation being described. Moreover, k is positive in cases in which

the language of proportion is used.
Pupils may mistakenly get the impression that all straight-line
graphs are described by "y is proportional to x", and teachers may
need to emphasize that it is only straight lines through the origin
which correspond to direct proportion.
Unfortunately, direct proportion is often taught as a purely
arithmetical idea, divorced from its graphical illustration.

6.2. Other Examples of Proportionality

The idea of proportionality is also used to help pupils to think of

non-linear functions in a way which connects them with linear
functions. Examples can be taken from graphs such as y = k-^fx
and y = kx 2. In the first case, we may say that "y is (directly)
proportional to the square root of x" or "y varies as the square root
of x", and in the second case "y is proportional to the square of x"
or "y varies as the square of x".
Examples of the use of the language of proportionality are "the
distance of the horizon at sea is proportional to the square root of
the height of the observer above sea-level" and "the distance fallen
by a stone is proportional to the square of the time for which it has
fallen". Here too, the domains of the functions may need con-
sideration, and pupils should consider whether a good model of
the situation has been made.
Appropriate language to describe situations modelled by the Figure 6.2
graphs of y = k/x and y = k/x2 is "y is inversely proportional to
x" or "y varies inversely as x" and "y is inversely proportional to (ii) the force of attraction between the sun and a planet is
x 2 " or "y varies inversely as x2 " respectively. Examples of these inversely proportional to the square of the distance between
situations are: them.
(i) for rectangles of given area, the length is inversely pro- In the examples we have discussed, able students should
portional to the width; eventually know what functions are used, should know the shapes
SOME SPECIAL GRAPHS 41

Figure 6.3

of the graphs of these functions, and should be aware that common responding to the function ,xt > 1/x, should be known and
sense should be used in the interpretation of the graph, both recognized by pupils.
regarding the domain of the function and the suitability of the The calculator makes it easy to explore reciprocals numerically,
function as a model of the situation. so that x ^ \jx gives, for example,
2i->0-5
20h->0-05
6.3. The Reciprocal Function 200 1-> 0-005
0-2 )-> 5 ., . .
It is particularly important that the graph of y= l./x, cor- 0-02^50 . . .. .
42 FROM GRAPHS TO CALCULUS

(i)

y = mx

O/

Figure 6.4 Translations of the graph of y = mx.

and so on. Discussion of the idea of an asymptote to a graph may y = mx + c, and of translating it a units in the x-direction, to give
well arise from this. the graph whose equation is y = m(x — a). While the first of these
The calculator's error message also reinforces pupils' knowledge translations is usually taught, the second is equally important
that 0 does not have a reciprocal, so that the domain of xt-> 1/x for a full understanding of the relation between the graph and its
cannot include 0. equation. Of course, most pupils below the age of sixteen will deal
with these transformations largely in numerical terms.
Similarly, the general shape of the graph of the parabola of
6.4. Transformations of Some Common Graphs y = x2 should be known, together with its translations y = x2 + c
and y = (x — a)2. These are shown in figure 6.5.
Pupils who recognize the association between a straight-line An enlargement in the y-direction is also an important transfor-
graph through the origin, its gradient m, and its equation y = mx, mation. If the scale-factor of the enlargement is k, then the
should also see the effect on the equation of translating the graph c parabola y = x2 becomes y = /ex2, and the straight line y = x
units in the y-direction to give the graph whose equation is becomes y = kx (figure 6.6). Knowledge of these transformations
SOME SPECIAL GRAPHS 43

6.5. Graphs of the Sine and Cosine Functions

Some teachers prefer to introduce the sine and cosine functions

using definitions for angles between 0° and 90° based on right-
angled triangles. They then extend the definitions of sine and
cosine to angles outside the range 0° to 90° in such a way that the
new definition (i) agrees with the old one within 0° to 90°, and (ii) is
seen by pupils to be a natural extension outside 0° to 90°. This is
usually done by using a definition based on a circle of unit radius,
centre (0,0), in which a radius OP is drawn at an angle 9 to the
x-axis, and the sine and cosine functions are defined by the
equations (figure 6.8)
sin 6 = y
cosO = x.
Figure 6.5 Translations of the graph of y = .x
Other teachers prefer to begin by defining sin 0 and cos 9 for all
angles, using the circle definition given above, and then to consider
should together enable the pupil to make a reasonable guess of the trigonometry of the right-angled triangle as being concerned
y = ax 2 + bx + c for the equation of a graph such as that of figure particularly with the values of 9 between 0° and 90°.
6.7. Of course, these transformations can be used on all graphs, Whichever order is chosen, a complete graph oi the sine function
not merely straight lines and parabolas. cannot be drawn until it is defined for angles of all magnitudes.
44 FROM GRAPHS TO CALCULUS

f>5

Figure 6.6 Enlargements in the y-direction.

Then its periodic nature will be appreciated. It is this property of cosine graphs in figure 6.9 and should recognize the type of
periodicity which makes the sine function such a useful model for physical situation involving periodic motion for which they are
so many regularly recurring physical situations. good models.
Pupils should know and be able quickly to sketch the sine and Translations and enlargements along the x and y-axis are often
SOME SPECIAL GRAPHS 45

(x,y)P

Figure 6.7 Figure 6.8

needed. Pupils will notice that the graph of x H-> sin x° is the graph Figure 6.10(iii) shows the graph ofy = sin (x — 30)° in which the
of x i—> cos x° translated through 90° along the x-axis, so that whole wave has been shifted in the x-direction, or undergone a
phase change, while figure 6JO(iv) shows y = 2 + sinx° in which
sinx° = cos(x-90)°.
the whole wave has been translated by two units in the y-direction.
Enlargements are particularly important, for enlargement in the Some examples of the uses of sine and cosine graphs in
direction of the x-axis changes the wave-length, and enlargement modelling are shown below.
in the direction of the y-axis changes the amplitude of the wave. The height of a bicycle pedal above the ground as the cycle is
For in stance, the graph of figure 6.10(i) is obtained from y = sinx° being pedalled can be modelled by a function with formula of the
by an enlargement scale-factor j in the x-direction. Its equation is type
y = sin 2x°.
Figure 6.10(ii) is obtained from y = sinx° by an enlargement 15sin(360rf
scale-factor 2 in the y-direction. The equation is y = 2 sin x°, and
the waves are twice as high, or have an amplitude twice that of where r is the time in seconds from the time of starting, and h is the
y = sin x°. height in centimetres. The graph of this is shown in figure 6.1 1.
46 FROM GRAPHS TO CALCULUS

y*.

y = sin 2x°

90\ 180 /270 360 460^ 90 \ /180270X /360 450

y= cos x

180 360 540 N

y = sin x°

Figure 6.10 Transformations of the graph of r = sin .v . (i) v = sin 2.\';

Figure 6.9 Graphs of y = cos.x and y = sin.x'. (ii) y = 2 sin .v ; (iii) y = sin(.\- — 30) : (iv) y = 2 + sin .v .

The distance of the bob of a pendulum from the vertical position and the slower movement up and down of tides. All these are
after time t seconds may be given by a formula such as modelled by sine or cosine graphs which have been transformed
from the basic graphs of figure 6.9.
x = 5 cos (2nt)°
and the graph is that of figure 6.12. 6.6. The Exponential Function
Other examples are the variation in the length of daylight
through the year, the bobbing up and down of a cork on a wave, The exponential function arises from the attempt by mathema-
SOME SPECIAL GRAPHS 47
y., ,, height of pedal
(iii) above ground in
centimetres
1-
40 —
= sin (x-30)°

/30 120 210 360

\ 25--

10--
time in seconds

Figure 6.11 The height of a bicycle pedal.

displacement in
y = 2 + sinx° centimetres

0 360

ticians to find a function which models a commonly-occurring

type of growth.
Consider the mythical pond-weed which doubles its area every
day. On the first day its area is 1 cm2 . What is its area after .v days?
A mathematical model gives its area v cm 2 after .v complete days by
the formula
= 7*

This is shown in figure 6.13(i). Figure 6.12 The displacement of a pendulum hob.
48 FROM GRAPHS TO CALCULUS

(i)

= 2x :x is a non-negative integer

4 x

Figure 6.13 Exponential growth.

In this case, because the weed is growing continuously, it is negative numbers; but this is not a good model. The step-function
appropriate to extend the model to values of x other than integers; shows the actual amount of money accumulated after x years,
this amounts to joining the points by a smooth curve. where x need not be an integer.
Another example of growth, but one in which it is not In general, a function of the form x i—>• a* or x H-> a~ x (a > 1) is
appropriate to join points by a smooth curve, is compound called an exponential function, the first modelling growth and the
interest. Suppose £100 is invested at 8% interest, the interest to be second modelling decay. The study of exponential functions is one
paid at the end of each year and re-invested. Then after n of the major themes in calculus at the stage beyond sixteen. In the
completed years the amount £A of money invested would be given meantime, the calculator makes it easy for pupils to investigate
by the formula A = 100 x 1-08". The dotted line in figure 6.14 is the numerically exponential functions such as those associated with
graph of the function xi—> 100 x 1-08*, with domain the non- continued doubling and with compound interest, which occur in a
SOME SPECIAL GRAPHS 49

A i <• number of examples suitable at this stage. For example, the effect of
the compounding of interest at more frequent intervals than once
each year can be investigated.
220

200 6.7. Logarithms

180 The advent of the calculator has meant that the place of
logarithms in the secondary mathematics curriculum is changing
160 rapidly. Teachers who argue that their pupils will need logarithms
for calculation in real life must think again; no one who has used a
140 calculator will return willingly to logarithms. Some argue that
calculators are too expensive for all pupils to have access to them,
120 and that there are economic if not educational reasons for
continuing to use logarithms; the authors believe that this
100 argument will be short-lived, and that logarithms and slide rules
will take their place alongside Napier's bones as interesting
80 historical curiosities which aided calculation before better me­
thods were developed.
60 In the sixth-form mathematics course, the natural logarithm will
be needed and will be introduced either as the inverse of the
40 exponential function or in terms of the area under the graph of
v = 1/x. Teachers of pupils between eleven and sixteen need to
20 decide whether there are reasons other than computational ones
for teaching their pupils logarithms below the age of sixteen. The
authors think the arguments for the retention of logarithms are
0 10
weak, and that as sufficient calculators become available in schools
for those pupils who do not have their own to have access to them,
'. ' the teaching of logarithms before age sixteen should be discon-
Fiaure6.14 Investment growth with compound interest. •• tinued.
 7 What is Area?
7.1. An Approach we attempt to ask what the other words in these definitions mean,
we see that they are not definitions at all.
What is area? This is one of those deceptively simple questions A more helpful approach than attempting to define area, which
which is much harder to answer than it looks. Some textbooks give is handled very well in SMP Book B, chapter 4, p. 36, is to consider
a definition of the form "Area is the amount of surface covered" or the problem of how to compare two shapes in order to say which is
"Area is the amount of space covered by a surface" but, as soon as the larger. It is necessary to remind pupils (because, of course, they
will previously have met area in primary schools) precisely what it
is that is supposed to be larger, for pupils are often unclear
whether the perimeter or surface is meant. It may be useful to think
in terms of paint required to cover the surface in conversations
with pupils. Once it is established what is being investigated, it is
clear that if the shapes are like A and B in figure 7.1 there is no
problem, because one will fit inside the other; but if the shapes are
like X and Y, a third shape Z is needed so that X and Z can first be
compared, then Y and Z, and so X and Y can be compared with
each other.
A convenient shape is then picked for Z, the only proviso being
that copies of the shape Z must fit together without leaving holes
(see figure 7.2).
In general the most convenient shape for Z will be the square. A
grid of squares is then used to make an estimate of the area. This is
illustrated in the following extract which is taken from SMP
Book B, p. 43.
Measurement of area

To find the area of an irregular figure, it is convenient to use tracing paper.

Either the outline of the figure can be traced and the tracing held over the top of
the grid so that the squares can be counted, or the grid can be placed on tracing
paper and held over the figure.
Figure 8 shows a shape drawn over the square grid. It is possible to count and
Figure 7.1 estimate its area from the figure.

50
WHAT IS AREA / 51

How many complete centimetre squares are there?

The complete squares have area
Two parts marked a
Three parts marked />
Two parts marked c
Two parts marked (/
Two parts marked e
Two parts marked /
together with the small unlettered parts

13 squares

Other books make other suggestions about how to make the

approximation. For example, some suggest first counting the
number of complete squares inside the shape and then counting
the number of part squares inside the shape, allowing each part
square to count as one half of a complete square. In the extract
shown above, there are 7 complete squares and 17 part squares,
making a total of
Figure 7.2 (i) A convenient shape for Z: (ii) a bad choice Tor Z. 7 + 17 x | = 15| squares.
52 FROM GRAPHS TO CALCULUS

No comment need be made on the relative accuracy of these information about A which can be used in further calculations, and
methods, but there is one very important idea which can be put exact limits of accuracy for those calculations are then known.
forward at this stage. Suppose the accuracy of a statement such as A sr 13 is known and
that, for instance, the estimate of 13 units is known to be in error by
less than 3 units. This is precisely equivalent to saying that
10<A< 16
7.2. Sandwich Arguments
that is, it is equivalent to a sandwich inequality.
While the approximations of the previous section Another advantage of the original sandwich inequality
7 < A < 24 is that it can be refined. If a grid with smaller squares
area » 13 squares
and were to be used, we would expect to enclose A within closer
bounds. If each square is \ of the area of the original one, then the
area s; 15-j squares
result
may be reasonably accurate statements, it is impossible to say how
9j < A < 16f
accurate they are. Neither is there any way of knowing whether or
not 13 is a better approximation than 15|. However, there is is obtained, a considerable improvement.
another statement which can be guaranteed to be correct. This is This idea should not be developed very far at this stage, but the
the statement that the area of the figure is sandwiched between an sandwich notion will become important later, especially in in­
area of 7 whole squares and 24( = 7 + 17) whole squares— tegration. It therefore seems a pity not to plant the seed of the idea
in symbols early. The sandwich idea is taken up again in chapter 9.

7 < A < 24
7.3. Two Remarks
where A is the number of units of area covered by the figure. The
essential difference between each of the statements (I) In the example of the previous section, the approximation
A ^ 13 and A
and the statement was obtained by counting the squares inside the figure and by
allowing each part square inside the figure to count as one half-
7 < A < 24
square. The inequality
is that while the first statements are vague, the second is precise.
7 < A < 24
Without additional information there is no way of knowing how
near A is to either 13 or to 15^. On the other hand the second was also obtained, and it will be noted that 1 5^ is the average of 7
statement, 7 < A < 24, is not an approximation at all. It gives and 24, that is 1 5^ = |(7 + 24 ).
WHAT IS AREA? 53
That this is no coincidence may be seen by the following
argument.
Let s be the number of complete squares inside the shape (see
figure 7.3) and 5 the number of squares required to enclose the
shape. Then the sandwich inequality states that

s<A<S
But the number of squares cut by the perimeter of the shape is S —s
and part of each of these squares lies inside the shape. Each of these
S — s cut-squares is taken to count one half-square towards the
area.
Hence

which simplifies to
s+S

It would be good practice to write down the sandwich inequality Figure 7.3 The sandwich inequality.
first and then to use it to write down an approximation for A. This
would prepare the way for the development of the sandwich idea at
a later stage.

(II) A different orientation of the figure, or a different position of

the grid may very well give a different sandwich inequality, and so a
different approximation for the area.
In figure 7.4 different positions of the square grid give different
inequalities for the area A of the shaded square, namely,
1 <A <9 and 0<A <4
and approximations
A* 5 and A ^ 2
Figure 7.4 The effect of grid position on the approximation obtained from a
then follow from these. sandwich inequality.
54 FROM GRAPHS TO CALCULUS

7.4. An Example of a Sandwich Inequality

When pupils have studied a little trigonometry, they can apply it
to a revision of areas of circles and so use a particularly important
type of sandwich inequality which arises.
A circle of radius 1 unit is drawn and its area is sandwiched by
inner and outer regular polygons. Figure 7.5 shows inner and outer
regular polygons with four and seven sides.
When the polygons have n sides, the sandwich formula
area of inner polygon < area of circle < area of outer polygon
becomes
. 360C 360° . 360°
n sin——cos-—— < area ol circle < n tan——
In In 2n
and the results, evaluated on a calculator, for n — 4, 7, 360 and
Figure 7.5 Approximations to the area of a circle.
1800 respectively are
2 < area of circle < 4
2-736 < area of circle < 3-371
were used before. However, it is interesting to reflect on two
3-1414 < area of circle < 3-1417
practical differences.
3-141586 < area of circle < 3-141596
First, it is not possible in reality to put a network of cubes inside
It should be noted that for a calculator which gives nine decimal and outside a given container or shape whose volume is required.
places, the sixth figure after the decimal point in the last line may be The exercise is entirely a mental one; that is, a mathematical model
suspect. for volume has been made.
Secondly, volume is measured in the classroom by using a linear
scale on a measuring cylinder of some kind. This draws attention
7.5. Volume to the fact that almost all physical quantities, for example, voltage,
A theory of volume may also be built up in a similar way to this current, velocity and angle, are measured on a meter or some other
theory of area, except that cubes correspond to the squares which linear scale. Area is a notable exception.
 8 First Ideas of Speed and Gradient
8.1. First Experiences statement that the average is measured over the 3| hours of the
journey. Over a particularly fast section of the journey, the average
Children's first experiences of speeds of movement are concerned
speed may be as great as 150 kilometres per hour. During the time
with comparison of speeds. One running child overtakes another,
of the stop at Darlington the average speed is zero.
one rhythm feels faster than another, one car gets away from traffic
There are many data available from which pupils can work out
lights faster than another. No measurement is involved at this
average speeds of trains and buses, and runners and swimmers.
stage, and understanding of speed may not for some children be far
The calculation of average speeds from data within their ex­
enough advanced for its measurement to be meaningful until very
perience helps pupils to grasp some important ideas about the
late in the primary years. For the vast majority of children, their
measurement of speed:
first acquaintance with the measurement of speed is concerned
(i) To measure speed, there has to be an interval of time over
with the speed of cars. Both the speedometer of a car and the
which the distance travelled is measured,
ubiquitous speed-limit signs serve to connect numbers with visual
(ii) This process always leads to an average speed,
and bodily impressions of speed, and so to introduce the idea that
(iii) The units of average speed are units of distance divided by
speed is measurable. units of time (for instance, kilometres per hour or metres
Unfortunately, the speedometer of a car is almost the only
per second),
device in everyday experience which measures a speed directly. The
(iv) The average speed over an interval conceals variation in
realization does not come easily, that the 30 which a speedometer
actual speed during that time.
registers at an instant means that if the car went on travelling at the
(v) Motion at constant speed may provide a useful mathemati­
same speed for an hour it would cover 30 miles. Nor is it
cal model for the real situation.
immediately clear to pupils that, unless a speedometer is available,
Figure H.l(i) shows a distance-time graph which is a greatly
the only measure of how fast one is travelling is found by
simplified mathematical model of the journey of a
measuring the distance travelled in a particular time, and dividing
London-Durham train.
the distance travelled by the time taken to give an average speed The simple model of the straight-line graph needs careful
over that time. discussion, so that pupils realize that according to this model, the
train has travelled 120 km by the end of 1 hour. 240 km by the end
of 2 hours, and so on. Hence they associate a straight-line distance-
8.2. Average Speed
time graph with constant speed. Pupils may be encouraged to
An average speed must be measured over a particular period of provide more-probable distance-time graphs than that of the
time. When we say that the average speed of the London to straight-line model shown in figure 8.1, for instance by taking
Durham train is 120 kilometres per hour, it is implicit in the account of the stop at Darlington.
55
56 FROM GRAPHS TO CALCULUS

The discussion of such graphs in many texts does not bring out
this modelling aspect of the graph. These books tend to give the
impression that if the stop at Darlington on the Durham train is 500
taken into account as in figure 8.1 (ii), the train actually travels like
that. Although the straight-line model with the stops shown is a
400
good one, and is actually used by British Rail in their day-to-day distance in
timetabling, teachers should emphasize to pupils that the straight- kilometres
line graph is only a simple approximate model which shows 300
average speed very clearly.
Detailed discussion of such straight-line graphical models helps 200
to bring out this fact that the interval of time over which an average
speed is found may be very short. It is quite reasonable to say that
100
the average speed of a train is 150 km/h for a period of 5 minutes; it
means that in the 5 minutes in question the train travelled 12^ km.
To have an average speed of 150 km/h a train does not have to
0 1234
travel for 1 hour.
time in hours

8.3. Gradient of a Straight-Line Distance-Time Graph

It is not a difficult idea that, for straight-line graphs of distance 400

against time, the steeper graph represents the greater speed. It is
important that the idea should be firmly linked with the gradient of distance in
300
kilometres
the straight line, so that pupils implicitly realize that the gradient of
the straight-line graph, which is measured by
200
increase in y
corresponding increase in x '
100
numerically represents the constant speed which is shown by the
graph. Of course, in measuring gradients from graphs, care must be
taken to measure in the units which are used on the axes, as shown time in hours
in figure 8.2(ii). This point is particularly important for distance-
time graphs, where the scale for units of distance is unlikely to be Figure 8.1 Distance-time graphs: (i) simplest model; (ii) model showing a
related to the scale for units of time. stationary period.
 FIRST IDEAS OF SPEED AND GRADIENT
57

8.4. The Gradient of a Straight-Line Graph as a Rate

0)

o
There are many ether examples, as well as distance and time,
where a real-life situation is usefully modelled by a straight-line
graph, in such a way that the gradient of the graph is meaningful. In
all these situations, the gradient of the straight-line graph repre­
300 sents a rate. Some examples are given below.
(i) Figure 8.3(i) shows the petrol consumption of a car. A
straight line has been used as a model, and its gradient represents
200
the rate of consumption of petrol, which is measured in miles per
gallon.
100 (ii) Cloth is sold by the metre; a graph can be drawn of cost
against length (figure 8.3(ii)). The marked price represents a rate
of, for instance, £1-25 per metre. It is the gradient of this straight-
line graph which represents the price rate of £1-25 per metre.
time in hours (iii) The exchange rate of francs against pounds is the gradient
of the graph (figure 8.3(iii)) which shows how many francs the
tourist will obtain for his pounds.
(iv) When cricketers measure the rate of scoring needed to win,
400 they give it as, for instance, 7 runs per over. This is very clearly an
average rate required over the remaining overs of the match, and
fluctuates from over to over. In examples (ii) and (iii), however,
300
although an average rate is used, it is a constant average rate.
These four examples call attention to the fact that in everyday
200 life, a rate or (more properly) an average rate is not always
measured with respect to time. Of course there are many examples
gradient = 80
100
of rates with respect to time, such as the rate of growth of a plant, or
the rate of inflation, as well as our first example of speed.
Speed is not always clearly seen as a rate, but it is in fact the rate
0 123456* at which a moving body changes its distance from a fixed point
with respect to time, and so is a rate of change. The units used for
speed, such as kilometres per hour, are entirely comparable with
Figure 8.2 The gradient of a distance-time graph: (i) the steeper graph represents other units for rates, such as runs per over, or miles per gallon.
the greater speed; (ii) calculation of speed from gradient. The links between speed and other rates, and between rate and
58 FROM GRAPHS TO CALCULUS

petrol in gallons length in metres pounds

Figure 8.3 The gradient of a straight-line graph: (i) rate of consumption of petrol; (ii) price per unit of length; (iii) rate of exchange for foreign currency.

the gradient of a graph, need to be carefully made. The following Rale. The rate giving the number of pence per book is the same for any pair of
numbers in a row of the table.
extracts from Modern Mat hematics for Schools (second edition) From the first row we see that the rate is 30 pence per book.
form a train of thought which needs careful development by the The rate is also given by -'f, or ™, or -Lf2, or j%, or J-.
teacher if the pupils are to make the connections expected. The (MMS Book 3, p. 167)
development from rate oj change is particularly important
Exercise 2
The costs of different numbers of a certain book are shown below. 1 Find the rate in each of the following, in the units stated:
Number of pence a 8 oranges cost 24p: pence per orange
Number of hooks in the coat b 20 bars of chocolate cost 90p; pence per bar
1 30 c 273 units of electricity per week ; units per day
2 60 d 20 km in 4 hours; km per hour
3 90 e 20 apples weigh 2 kg; apples per kg
4 120 / 140 km take 2\-hours; km per hour
10 x (MMS Book 3. p. 168)
v
One-to-one correspondence. There is one. and only one, cost corresponding to
a given number of books: and one, and only one, number of books Speed is rate of change of distance. When the speed is constant, the distance is
corresponding to a given cost. So we say that there is a one-to-one cor­ proportional to the time and. as we saw in Chapter 2, the graph of a direct
respondence between the number of books and the cost, as indicated by the proportion relationship is a straight line.
arrows. {MMS Book 3. p. 206)
FIRST IDEAS OF SPEED AND GRADIENT 59
8.5. Gradient of a Road: a Red Herring

The usual first example of the gradient of a road or railway has

been left until last, for the word gradient is not used in the same
sense as it has been used above. The steep hill whose gradient is 1 in
6 is shown in figure 8.4. It is natural to measure the 6 units along the

Figure 8.5 Gradient of a railway line.

Pupils will also have drawn graphs such as that of the function
x i—> — 2x (figure 8.6(ii)). They will see that to an increase in x there
Figure 8.4 Gradient of a road. corresponds a decrease, or a negative increase, in y. Thus the
definition of the gradient of a straight line,
road, rather than horizontally, so the gradient described by the increase in y
road sign is sin 0, not tan 0. Moreover, a railway gradient post corresponding increase in x
shows
inevitably attaches a negative gradient to such a straight line.
-=—ry or cosec 0. \ Similarly, the graph of the function x i—> mx + c has a gradient of
sin u
m. However, this last statement represents a considerable step in
abstraction from the previous one, for now a general statement
8.6. Gradients of Straight-Line Graphs about the whole class of straight-line functions is made, and the
teacher should not take this last step until he is sure his pupils are
It is now an easy step to discuss the gradients of the graphs of ready for it.
straight-line functions, such as xt->2.\. Pupils who have used
straight-line graphs to model the situations above will realize that
the gradient of the graph shows the rate of change, and so will 8.7. A Change of Wording
realize that the study of the gradient of a straight-line graph such as
that of the function xi->2x gives important information about One further point needs to be made; it concerns the nature of the
rates which can be applied in a variety of contexts. Pupils easily words distance and speed. It is unfortunate that these words, which
verify that the graph of the function ,x H-» 2x has a gradient of 2. are used so much in everyday speech and thought, are not
60 FROM GRAPHS TO CALCULUS

y' •

-2x

^corresponding
^increase in )'

Figure 8.6 (i) A positive gradient; (ii) a negative gradient.

considered by mathematicians to be sufficiently precise. Gradients way the distance between two points is always positive, while
of straight-line graphs may be either positive or negative, but when mathematicians wish to allow for the possibility of negative
average speeds are always thought of as positive. The word velocity distance they use the word displacement.
is used when the direction of a speed is taken into account, so These "official" words will be used in those remaining parts of
velocity is allowed to be negative. Hence it is average velocity which the book which use the ideas of motion.
corresponds to the gradient of the straight-line graph. In a similar The ideas of this chapter are developed in chapter 12.
 9 Numbers
9.1. The Need for Real Numbers in Graphical Work and the pupil is usually (and rightly) expected to accept this
assumption in good faith without even realizing that it is an
When a pupil is asked to draw the graph of a function such as assumption.
xi—>x2 and to join the small number of points he has plotted to If, instead, a pupil attempts to find ^/2Q using a calculator, the
form a curve, there is an implicit assumption that each point of the calculator may show that
x-axis represents a number and that each point of the y-axis also
represents a number. v/20 = 4-47213595
Pupils might then be asked to read off from the graph of x t->x2 but when the number on the right-hand side is squared it is found
an approximation to the positive number whose square is 20. This not to be exactly equal to 20.* Indeed, none of the successive
task carries the implicit assumption that there is such a number, approximations which a calculator might produce,
44, 4.47, 4-472, 4-4721 and so on,
has a square which is exactly 20, although the sequence has a
limit whose square is exactly 20. This concept of a limit, which will
be necessary in the later study of integration and differentiation,
will involve pupils in an intuitive appreciation of the complete
system of real numbers.
This chapter is devoted to a discussion of a way in which this
intuitive appreciation of real numbers may develop in children
before the age of sixteen by using the interplay between numbers
and their representation on the real number-line; that is, by using
the interplay between arithmetical and geometrical thinking about
numbers.
Many of the ideas are subtle and delicate. The teacher will be
introducing them implicitly in discussion and by assumption,
rather than by formal exposition, over a period of some years. The

* Care is needed here for calculator may appear to show that

(4-47213595)•*
(4-47213595) 2 = 20. However, it is clearr that in an exact calculation
calcu there will be a 5
Figure 9.1 in the sixteenth place after the decima' point of the product.
••••••
61
62 FROM GRAPHS TO CALCULUS

development of pupils' concepts of number is one of the most The interaction of three ideas, the number itself, its repre­
important tasks in mathematics, and this chapter is therefore sentation on the number-line, and its written representation in the
addressed to the teacher, so that he may examine the development denary system for the purposes of calculation and comparison,
of the number concept as it is needed for later work in calculus. The forms the theme of this chapter.
level of exposition is often more sophisticated than that which
would be suitable for pupils, who may take for granted much of
what is explicitly stated here. 9.3. Fractions and Decimals on the Number-Line

Most pupils know before the age of eleven how some simple
9.2. Early Stages fractions such as f or 1^ and some decimals can be represented on
the number-line, using different scales as convenient (see figure
By the age of eleven, most children have absorbed the whole 9.3).
numbers into their thinking. They have met the number-line and
H 11
they see the whole numbers as attached to equally spaced points on 1——————————————1
0 1 2 3

0-1 0-2 0-5

012345 I———'———'—————— ———i———————————————|————^
0 1
Figure 9.2
0-01 0-1 0-2
r-1—I—————
the number-line with zero as the starting point on the line. They 0 0-02
also know that the bigger the number, the further to the right it is
on the number-line.
They are also familiar with the denary way of writing whole Figure 9.3
numbers, using hundreds, tens and units, and can carry out, to
varying extents, procedures for adding, subtracting, multiplying
and dividing whole numbers. Pupils may not be aware how The ruler-and-compass construction which enables a line-
efficient the denary system of writing numbers is compared with, segment to be divided into a number of equal parts is a most useful
say, the Roman system. Not only are the denary procedures for the aid in enabling pupils to visualize the position of fractional points
four rules much simpler than corresponding procedures would be on the number-line. To find points on the number-line correspond­
in the Roman system, but the denary system gives such a simple ing loj,j,...,j, for instance (figure 9.4), a line is drawn through O
method for deciding which is the larger of two numbers that we at an angle to the number-line, and seven equal lengths are stepped
take the method for granted, automatically comparing digits of the off along it. Parallels are then drawn to the line joining P to the
two numbers. point 1 of the number-line.
NUMBERS 63
ment from a number-line with equally spaced points marked to a
number-line with fractional points marked.

Whole numbers Fractions

Line with equa Line with

Geometrical ideas spaced points fractional points
marked marked

As well as knowing that fractions may be marked on the

Figure 9.4 Geometrical construction for positioning fraclional points on the
number-line. number-line and may be written using a pair of whole numbers
such as |, many primary-school children know that f may also be
Thus, fractions are easily made to correspond to points of the written as the decimal 0-75. Indeed they may have met the decimal
number-line. Fractions are also comparatively easily added, form first. The decimal fractions which a pupil knows at this stage
subtracted, multiplied and divided by well-known procedures. are very easily ordered on the number-line, although some pupils
However, it is not easy to tell at a glance whether, for instance, jf is may, for a time, be deceived by the digits into thinking that, for
greater than f. When two fractions such as these are to be instance, 0-075 is a larger number than 0-13. It is essential for the
compared, it is necessary to replace them both by equivalent eventual understanding of the real-number system that pupils
fractions with a common denominator, so that for example f > yf, should understand that the decimal notation to which they are
since f = {ff and 4j = T§t- The pupil will not be able actually to accustomed merely gives another way of writing some fractions;
mark these points on a number-line, but we expect that he 0-13 isonly another way of writing YQ- +-TOO or-foij. In the same way,
probably will have some picture such as that of figure 9.5 in his 0-25 is another way of writing f^, which is that very familiar
imagination. fraction ^. Unfortunately, although to every terminating decimal
there corresponds a fraction, it is not true that to every fraction
there corresponds a terminating decimal. The attempt to convert
fractions to decimals leads to the exploration of recurring de­
cimals.

Finure9.5
9.4. Recurring Decimals on the Number-Line
A correspondence has now been set up between some arithmeti­
cal ideas and some geometrical ideas. These ideas are the When 3 is converted by division into a decimal, there is at every
development from whole numbers to fractions, and the develop- stage a remainder. It is usual practice, as in the quotation below
64 FROM GRAPHS TO CALCULUS

from Maths Today, to introduce recurring decimals at this point

(Maths Today, Book /, pp. 159-60).
What is 3 as a decimal fraction? What do you discover when dividing? Why
does the decimal fraction fail to work out exactly? Such a decimal is called a
RECURRING DECIMAL and we frequently put a full stop or point above the
recurring figure to show that it is repeated—e.g.,
^ = 0-3 =0-33333...

If the work is just left here, important opportunities are missed.

Pupils can mark a point which represents 5 on the number-line by
a ruler-and-compass construction. They can also mark on the
number-line the first few decimal approximations
0-3,0-33,0-333,
to the fraction -k

Figure 9.6

In practice, the point on the number-line representing 0-333 is

indistinguishable from that representing 3-; however, because
3 x 0-333 = 0-999, so that 0-333 is not equal to i, and similarly for
each of the approximations. Thus, none of the points which
represent 0-3,0-33,0-333,... should fall exactly on the point which
represents 5. Figure 9.7 Successive decimal approximations to one third on the number-line.
NUMBERS 65
It is well worth discussing the sequence of sandwiching in­ hundredths, etc., then, working in base three instead of base ten, it
equalities which occur at the successive stages of the division which is found that
converts } to a decimal. First
^ = 0-1 (base three).
0-3 < | < 0-4 Similarly the proper fraction p/q is written in base q as 0-p. Thus a
fractional number always has a terminating representation in
since, multiplying by three, 3 x 0-3 < 1, but 1 < 3 x 0-4. Hence
some number-base, even though its decimal (base ten) repre­
sentation may not terminate. Whichever base is chosen, however,
0-3 < { < 0-4 the majority of fractional numbers have recurring rather than
0-33 < ^ < 0-34 terminating representations.
0-333 <i< 0-334 Clearly, each terminating decimal is equivalent to a fraction; for
instance, 0-23 = j$j. It will be shown in section 9.8 that each
0-3333 < i < 0-3334 recurring decimal is equivalent to a fraction; for instance
and so on. Figure 9.7 illustrates these inequalities in terms of 0-15 = 3^. It will be assumed that, conversely, each fraction may be
successive magnifications of a ruler graduated in tenths, hun- written in base ten as either a terminating or a recurring decimal.
Hence the previous correspondence between arithmetical and
dredths, thousandths, etc., of a unit.
The width of the sandwich inequality decreases at each step by a geometrical ideas may be extended to the following:
factor of ten:
Fractions (or
Arithmetical ideas Whole numbers > terminating and
0-3 < | < 0-4 recurring decimals)

0-1

Line with equally Line with

Geometrical ideas spaced points — > fractional points
marked marked
0-01

so that although the decimal graduation poihts on the number-line When directed numbers are introduced, the number-line needs
become indefinitely close together, none of them ever falls exactly to be extended indefinitely in the negative as well as the positive
on the point which corresponds to f The explicit discussion of this direction, so that it can represent positive and negative whole
point may help pupils to extend their ideas about number. numbers and fractions. At this time, official mathematical voca­
But if a number-line which is graduated in thirds, ninths, etc., of bulary can be introduced: the positive and negative whole
a unit is taken instead of one which is graduated in tenths, numbers are the integers, and the positive and negative whole
66 FROM GRAPHS TO CALCULUS

numbers and fractions together make up the rational numbers. The procedures for exact calculation. This might be considered as the
correspondence now contains three sets of ideas: price paid for the beauty of the decimal system in which the relative
sizes of any two numbers may be distinguished by inspection.
Positive and
Positive and negative Pupils who have begun to explore fractions on the number-line
fractions, or positive and probably think that to every point of the number-line there
Arithmetical ideas negative —
negative terminating and
whole numbers
recurring decimals corresponds a number of a type they know: in other words, a
f rational number. At this stage, it is difficult not to reinforce the
belief that every point on the number-line corresponds to a
Line with equally Line with
Geometrical ideas spaced points —— • fractional points rational number, because of the very necessary emphasis which
marked marked must be placed on interpolation from graphs. From the graph of
x i—>x2, a pupil finds that (as nearly as he can tell) ^20 = 4-5. He
should also be encouraged to verify that (4-5)2 = 20-25, so that his
Rational result, although near, is too big: -y/20 < 4-5. With a sharper
Integers
numbers
pencil, a larger scale and a good deal of luck he may
obtain the approximation 4-47 for ^20. But (4-47) 2 = 19-9809.
9.5. Calculating with Recurring Decimals so 4-47 < ^/20 < 4-5. It is likely that pupils will draw from this the
The procedures which children have learned for calculating with belief that if only they could be accurate enough, or if they had a
terminating decimals break down for decimals which do not calculator with enough decimal places, they could find a decimal
terminate. For example, it is not possible to calculate directly whose square is exactly 20.
It is one of the most remarkable discoveries of Greek mathe­
0-33333 ...... x 0-142857142857...... matics, and still comes as a great shock to many pupils today, that
because it is not possible to find "the right-hand end of the there is no rational number whose square is exactly 20. But much
recurring decimal". All that can be done is to cut short the further preparation is needed before pupils will appreciate the
recurring decimals, to calculate with the resulting terminating significance of this fact.
decimal approximations, and hope that the result of the calcu­
lation is a reasonable approximation to the answer.
Exact calculation of the product of the two recurring decimals 9.6. The Sandwiching Process and Measurement
above is possible by converting them to fractions, carrying out the
multiplication and then reconverting to a decimal. This gives A further step in the argument comes from the combination of
sandwich inequalities with ideas derived from practical measure­
i x i = ^L = 0-047619047619047619......
ment. No measurement made with a ruler or other measuring
The change from a fractional representation of a rational instrument is ever exact; measurements are made "to the near­
number to a decimal representation has resulted in the loss of est ...". The statement that the lenath of the room is 4-27 m must
NUMBERS 67

be taken to mean that it is in the range from 4-265 m to 4-275 m. It of enclosing a number in a sequence of sandwich inequalities
has been (presumably) measured to the nearest centimetre, so that whose width can be indefinitely decreased (figure 9.8) is of utmost
the actual measurement is sandwiched in an interval of length 1 cm importance when moving from the rational numbers to the
(or 0-01 m). irrational numbers.

4-265 ^ / ^ 4-275
0-01

Figure 9.8 Successive approximations to a value for a point .v on a number-line.

Similarly, when a pupil says that, from his graph, N/20 is

approximately 4-5, he probably means that the point is nearer to
9.7. Describing Points on the Number-Line
4-5 than to 4-4 or 4-6; for him
The decimal number system provides a way of labelling every
point on the number-line, not only those which correspond to
4o5
terminating decimals. A point marked on the number-line either
(M falls on an integer point, or it falls between two integers. The
number corresponding to the point labelled x in figure 9.9 lies
between 2 and 3, so that 2 ^ A ^ 3, and the decimal representation
It is important for the teacher to realize that the graduation marks of the number either starts 2-, or is exactly 3. Next, tenths of a unit
on measuring instruments are placed at points representing are used ; so that .v satisfies
rational numbers, so that every measurement is always sand­
. 2-7 s; ,x <: 2-8
wiched by a pair of rational numbers. In the example of the length
of the room, the precise length is not known: it is only known that Using hundredths of a unit
the length is sandwiched between the rational numbers 4-265 and 2-74 2-75
4-275.
The previous sandwiching example of decimal approximations It may happen that this process terminates at some stage, so that.
to j contains an additional feature: each inequality sandwiches } for instance, when thousandths of a unit are used, x = 2-743.
between rational numbers such as 0-33 and 0-34, or 0-3333 and However, it is much more likely that the process never terminates.
0-3334, but the sandwich inequalities also lead to intervals of so that the number .v can only be described as a decimal by the
decreasing length, each interval nested within the previous one; sequence of nested intervals within which it is sandwiched, as
thus j can be enclosed in as short an interval as we please. This idea shown on page 68.
 FROM GRAPHS TO CALCULUS

bers, decimals which neither terminate nor recur, are irrational

numbers. The rational and irrational numbers together make up
the real numbers.
Two assumptions are implicit in the above description, and the
intention is that pupils should informally take them for granted in
their own thinking. If they do so, they will have nothing to unlearn
later, and they will be well prepared for a deeper understanding of
the real numbers when they are older. The assumptions are:
(i) Every point of the number-line corresponds to a real
number,
(ii) The position of a given point of the number-line can be
described by a sandwiching process, and the corresponding
real number represented by a decimal.
In this book we have called this second assumption the sandwich
assumption.
At a later stage in their mathematical development, the most-
able students will realize that the second assumption is the
Figure 9.9 Successive approximations to a value for a point \ on a number-line. "Chinese Box Theorem", which states that if we have a sequence of
nested closed intervals on the real number-line whose lengths tend
to zero, then there is exactly one real number which is common to
Sandwich inequality Length of interval
2 s; x < 3 1 all the intervals; that is, every sequence of nested closed intervals
2-7 s; .x =£ 2-8 0-1 converges to a number and not to a gap between numbers. Of
2-74 <: x s; 2-75 0-01 course the real number to which the sequence of nested intervals
2-742 «: A- <: 2-743 0-001
2-7429 s; x < 2-7430 0-0001 converges may be either a rational number or an irrational
number.

Pupils can be shown how this point is represented by

9.8. Changing Recurring Decimals to Fractions
2-7429......
where the dots indicate that further steps in the sandwiching The links between rational numbers and terminating or re­
process have not yet been described. If the decimal corresponding curring decimals are not complete until pupils realize that every
to the point on the number-line neither terminates nor recurs, then recurring decimal, as well as every terminating decimal, cor­
the number is not a rational number, since all rational numbers responds to a rational number.
have terminating or recurring decimal expressions. These num­ The following method of showing how a recurring decimal can
NUMBERS 69
be converted to a fraction is suitable for discussion with abler 9.9. Calculation with Real Numbers
pupils, but it depends on assumptions about the convergence of
infinite series which will not be made explicit at this stage. The Although pairs of irrational numbers are easily ordered, ir­
method is illustrated for 0-J23. rational numbers, like recurring decimals, are extremely awkward
x= 0-123 for purposes of calculation. Whenever irrational numbers occur
they can always be sandwiched between rational numbers. Calcu­
=> 1 000.x = 123-123 lations are always carried out using the rational approximations.
=> 999.x = 123 For instance, if two positive real numbers r and s are sandwiched
between positive terminating decimals a, b and c, d respectively, so
that
— 41
— 333
a^ r^ b
The correspondence which was introduced earlier, between and
arithmetical and geometrical ideas and the official mathematical c< s^ d
language, can now be completed : then
a + c ^ /• + .v ^ h + d
Fractions (or Decimals (including
terminating non-terminating
and
Ariihmcticiil Whole numbers -
ideas and recurring non-recurring ac ^ rs <. bd
decimals) decimals)
Only minor adjustments are needed when the real numbers
concerned are not positive. Moreover, these sandwich inequalities
Line with equally Line with
can always be made as thin as we please by taking decimal
Geometrical spaced points ———> fractional points—^Complete line approximations which are close enough to r and .v.
ideas marked marked This is the basis of all numerical calculation with irrational
numbers, whether done on paper or using a calculator, a computer
or logarithm tables. Mathematical tables are simply dictionaries
which give terminating decimal approximations for numbers such
Official Rational Real as log2, sin 35' and so on. Calculators give similar rational
nunheiutilictil Integers —————— > numhers
* numbers" approximations to real numbers. Once these rational approxi­
mations have been found, then the calculation is carried out
entirely in rational numbers according to the usual procedures, the
sandwich inequalities leading to a guarantee that the resulting
The third line of the (able, which gives the official mathematical rational number will be fairly close to the required real number.
names for the different set of numbers, can be introduced as the Thus, for everyday purposes of calculation, rational numbers
ideas are understood. rather than real numbers arc used. Mathematically, however, it is
70 FROM GRAPHS TO CALCULUS

important that the most-able pupils should realize that irrational Sandwich inequality Length oj interval
numbers are part of the theoretical construction of the number 2 /5 3 1
2-2 5 ^ 2-3 0-1
system.
2-23 = 5 ^ 2-24 0-01
2-236 = S =5 2'237 0-001
9.10. The Introduction of Irrational Numbers in Texts

In most texts, irrational numbers are first encountered in At this stage, pupils may begin to realize intuitively why the
connection with Pythagoras' Theorem and the need for square sandwiching process can never end, for to obtain a terminating
roots. A typical treatment is that of the Scottish Mathematics decimal whose square is exactly 5-000....0, the last digit would
Group shown below (Modern Mathematics for Schools, Book 4, have to be 0, and so would the digit before, and so on. Thus,
p. 19): discussion of the sandwiching process helps to build up the

u
realization that non-terminating non-recurring decimals, or ir­
rational numbers, do actually turn up in mathematics. Apart from
\A the square roots, other irrational numbers which pupils are likely
to meet before the age of sixteen are n and the trigonometric
functions.
-2-101;l*k 3 However, the rational numbers are very few among the totality
24
of the real numbers, and the vast majority of points of the number-
line do in fact correspond to irrational numbers. This fact is on the
From figure 24, using Pythagoras' theorem.
face of it most surprising, when the rational numbers are so
,- 2 = 2 2 + l 2 = 5
familiar. An informal explanation may be given as follows. A real
number between 0 and 1 could be generated in a random way by
Hence r = ^/5 = 2-236 ..., a number which cannot be expressed as a ratio of using a ten-sided die labelled 0,1,... 9, and taking the digit
two integers or as a decimal. Therefore vo does not belong to Q.
obtained as the first decimal place of a number. The die could
^/5 is called an irrational number. We can find its approximate position on
the number line by drawing an arc of a circle with centre O and radius OA as similarly be used to give a digit for the second decimal place, and
shown. any number of further decimal places could be similarly generated.
It is very highly unlikely that such a random process would
This is a good opportunity for raising several points in class generate one of the two types of non-terminating decimals which
discussion. First, on the real number-line which the pupils correspond to rational numbers: those such as
visualize, the exact position of the point representing ^/5 can be 0-342134213421... in which there is always a cyclic repetition, and
constructed by the method described. Secondly, the statement those such as 0-250000000 ... which correspond to terminating
r = v/5 = 2-236 ... can usefully be examined by the sandwich decimals. Thus, it appears probable that comparatively very few
method described above real numbers are rational.
NUMBERS 71

9.11. Proofs of Irrationality that an irrational number cannot be written as a fraction in any
number base, a point about which pupils sometimes have doubts.
Able pupils will be interested to see how the fact that a number By contrast, the expansion of a rational number always terminates
such as ^/5 is irrational can be established. All such arguments are in some base, as was shown in section 9.4.
proofs by contradiction, a method which able pupils should meet
before they are sixteen.
The proof that ^/5 is irrational begins by supposing, on the 9.12. Recommendations
contrary, that ^/5 is a rational number. Hence there are integers p
It is suggested that the following ideas should form the goal of
and q such that
work on the real numbers with able pupils.
1. Every point on the number-line corresponds to a real number
and, conversely, every real number corresponds to a point on
and so the number-line.
P2 = V- 2. To every fraction there corresponds a terminating or re­
curring decimal, and conversely, to every terminating or
Because q2 is a perfect square, it has an even number of prime recurring decimal there corresponds a fraction.
factors. Thus 5q2 has an odd number of prime factors. But 3. Every real number, whether rational or irrational, can be
5q2 = p2 and therefore p2 must have an odd number of prime approximated to within any required degree of accuracy
factors. But this contradicts the fact that p2 is a perfect square. using terminating decimals.
Hence ^/5 is not a rational number. 4. Calculations can only be carried out by the procedures for
This proof clearly extends to a proof that the square root of terminating decimals.
every prime number is irrational. Finally, the concept of real numbers is very closely bound up
It is worth noticing that the above proof is independent of the with the related idea of a limit, which is discussed in the next
number base chosen to represent p and q. This emphasizes the fact chapter.
 10 The Idea of a Limit
10.1. Strategy worksheets. Discussion between pupils will make an important
contribution to the work.
Although the idea of a limit, and the notation for limits, lie at the
heart of the differential calculus, they receive little attention in the
introduction to differentiation in any of the texts reviewed. This is 10.2. First Ideas
not because ideas about limits have been growing steadily
throughout the secondary course—far from it. In fact, it is usual for In the primary school, children meet many situations in which a
the idea of a limit to be met for the first time in the most difficult of process is repeated again and again, and the same process could be
all possible circumstances: the limits commonly used in differen­ repeated indefinitely. Many of these repetitive processes are
tiation, which are limits of fractions where both the numerator and concerned with sequences.
the denominator tend to zero. This is conceptually the most (i) Peter (aged 9) said "Tie a knot in your handkerchief to
baffling situation in which to grasp the idea of a limit. It is remind you. And another to remind you what that's for. And
suggested in this chapter that there are many fairly simple another. And another. And...."
mathematical situations in which pupils can gain informal ex­ (ii) A familiar exercise is "Give the next three terms of the
perience of limits from very early in the secondary school. If they sequences
do this, they will find the limits used in differentiation easier to 1,3,7,13,....
understand, and progress will be correspondingly greater.
It is not suggested that a formal treatment of limits should form 2,3,5,8,12,...."
part of the work done by secondary pupils; that must properly be (iii) Squares are often made with Cuisenaire rods. Children
deferred for higher education. However, some reference to limits is explore how many units are needed to turn each square into the
essential to any work on differentiation and is also needed for the next one (fiaure 10.1).
study of infinite geometric and other series at the next stage of
education.
It is extremely difficult to convey in print the delicacy of touch
which is necessary. The work described subsequently in this
chapter is not suitable for conventional teaching and testing, and
there may be little or no written output by the pupil. The aim is to
encourage ideas to grow gradually, and there need be no attempt in
the early years to draw things together. Many of the ideas
suggested are suitable either for group working or for individual Figure 10.1 Construction of squares with Cuisenaire rods.
72
THE IDEA OF A LIMIT 73

(iv) Repeated doubling produces a sequence whose terms grow We return to this example to discuss its mathematical impli­
very rapidly: 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024,. .. cations in section 10.5.
From these experiences, children gain the idea that many (ii) A variation on this situation is obtained by rearranging the
mathematical processes could be repeated indefinitely, or as they squares and rectangles into a spiral, as shown in figure 10.3. Pupils
say, "for ever". Peter's enjoyment came from his vision of knotted can investigate to which point the spiral converges.
handkerchiefs receding into the dim distance. But, at this stage,
most of the work is done with whole numbers, so that the terms of
the sequences are likely to grow beyond all bounds. Ideas about
limits begin to develop when situations can be explored in which a
sequence "settles down" or converges. This may initially be more
vivid in geometrical rather than arithmetical contexts.

10.3. A Geometrical Introduction to Convergent Sequences

Figure 10.3
(i) A square of side 1 dm is drawn, and rectangles and squares
are alternately built on to it, as shown in figure 10.2. Each shape
added has an area which is one half of that added previously. What (iii) Curve stitching is an activity which many pupils will have
happens to the total area? met in the primary school. Two intersecting lines are taken and
equal intervals are marked off from the point of intersection; the
i dm 2

1 dm 2 1 dm 2

area 1 dm 2 area 1t dm 2

i dm 2 i dm 2

1 dm 2 1 dm 2
J- dm 2
idm 2
\ dm 2
2 3 4 5 Gal 10

Figure 10.2 Geometrical introduction to convergent sequences. Figure 10.4 Curve stitching.
74 FROM GRAPHS TO CALCULUS

points on each line are numbered as shown in figure 10.4, the point Positions of the rabbit are drawn at R0, R 1; ..., R 10, each 1 unit
1 is joined to 10,2 to 9 and so on. The lines appear to form a curve. apart. Each part of the dog's path is 1-5 units long and D0 D, is
This is illusory, but better and better approximations may be directed towards R0, D,D 2 towards R l5 D2 D3 towards R 2 and so
obtained by inserting more and more lines, using the rule on. The dog catches the rabbit after about 12 seconds. By taking
a + b = 11 for deciding which points should be joined; a and b short enough time intervals, the path drawn can be made as close
may be fractional or negative. to the actual path of the dog as we please.
(iv) Pursuit curves give another method of approximating to (v) Another example of a similar type concerns four can­
curves by using small line segments. A dog D is chasing a rabbit R. nibalistic beetles. These beetles, A, B, C and D start at the corners
The rabbit runs along a straight line at 1 m/s. The dog can run at of a square. A wants to eat B, B wants to eat C, C wants to eat D,
1-5 m/s and always aims directly towards the rabbit. and D wants to eat A. All the beetles walk at the same speed, and
R0 R, R2 R3 R4 R5 R6 R7 R8 R9 R 10
the path of each beetle is aimed directly towards the beetle it is
chasing.
path of rabbit

path of dog
Figure 10.6 Pursuit curve: four cannibalistic beetles.

Step-by-step approximations to their paths are shown in figure

Fisure 10.5 Pursuit curve: a doe chasing a rabbit. 10.6. The example breaks down near the point at the centre of the
 THE IDEA OF A LIMIT
75

Figure 10.7 "Snowflake" curve.

spirals, but once again intuition suggests that the use of shorter
steps would have given a closer approximation to the actual paths.
It is interesting that in the last two examples we are, in effect,
solving differential equations by step-by-step methods. This idea is
introduced in section 16.4
(vi) The last geometrical example is of the snowllake curve, and
returns to the sandwiching theme. To construct the curve, an
equilateral triangle is drawn and each side is divided into three
equal parts. An equilateral triangle is constructed on the middle
third, as shown in figure 10.7, and the process of constructing
equilateral triangles on the middle third of each line is repeated
indefinitely.
It is clear that both the area and the perimeter of the snowllake
curve increase at each step. Construction of the first few approxi-
76 FROM GRAPHS TO CALCULUS

Ficure 10.8 "Anti-snowflake" curve. A/yyyvy

AAAAA/y
V
THE IDEA OF A LIMIT
77
3

mations both to the snowflake curve, and to the "anti-snowflake" 10.4. Further Simple Convergent Sequences of Numbers
curve which is obtained by inscribing the original equilateral
triangle in a larger one, and then at each step drawing inward- The advent of cheap electronic calculators makes possible an
facing equilateral triangles on the two outer thirds of each line investigation of convergent sequences in a way which was
(figure 10.8) will convince pupils that the area and perimeter of the previously not possible.
snowflake curve behave differently. The area is sandwiched (i ) Suppose we wish to find ,/6 by successive approximation or
between the areas of the snowflake and anti-snowflake curves by trial and error.
(figure 10.8). The perimeter of the snowflake, however, is multi­ We know that
plied by j at each step, as the effect of the construction at each step
is to increase the length of each line by j- of itself. Thus the sequence 2 < v'6 < 3.
of perimeters is
By squaring 2-4 and 2-5 we find that
3, 3x(f), 3x(f) 2, 3x(-j)3 ,..., 3 x &"-',...
and this increases bevond all bounds. 2-4 < v/6< 2-5.
78 FROM GRAPHS TO CALCULUS

The next step is Teachers should be aware that pupils may think that the
2-44 < 76 < 2-45 calculator has fixed on 10 as the exact value but this is not so. A
and the next is calculator which displays ten digits may store three more digits in
2-449 < v/6 < 2-450. the calculator and the number displayed will be a thirteen-digit
number corrected to ten significant figures. When this corrected
Pupils find it easy to believe that they can get as close as they version is 10-00000000, the calculator simply shows 10.
please to ,/6 by going on long enough, although they do not know
and cannot find the exact value of %/6.
(ii) The classical Newton method for square root converges 10.5. Sequences given by Recurrence Relations
much more rapidly than that discussed above. It uses the idea that
if x is an approximation to ^/N, then When a young child is asked how to get the next term of the
sequence
\( NN
2, 5, 10, 17, 26, 37,...

which is the average of x and N/x, will be a better approximation. he is much more likely to say
Unlike the previous method, Newton's method improves more "you add 3, then 5, then 7, then 9, . . ."
than one decimal place at each step. In fact it roughly doubles the
number of decimal places which are correct at each step, and so is a than he is to say
very rewarding exercise, even if a poor first approximation is taken. "it is the next perfect square, plus 1".
For example, for N = 2, using 1 as a first approximation,
He prefers the recurrence relation, which tells how to calculate the
successive approximations are
next term of the sequence from the previous ones, to the formula
1, 1-5, 1-42, 1-4142, 1-414213562 for the nth term. In official mathematical language, he prefers the
additive approach of the recurrence relation
and although the calculator then shows no further change, pupils
should realize that (1-414213562)2 is not exactly equal to 2. xi=2, xn+l =xn
With N = 100 and 1 as the first approximation, the figures
shown by the calculator are to the formula for the nth term,

1, 50-5, 26-2, 15-0, 10-84, 10-03, 10-0000529, 10

The formula is algebraically convenient, but makes a fresh start to
This illustrates an interesting and sometimes irritating feature of the calculation of each term. For numerical calculations, a
some calculators; when all the figures in the range of the machine recurrence relation is often very useful. Electronic calculators
after the decimal point are zero, the calculator suppresses them. make it possible for teachers to build on pupils' intuitive grasp of
This has happened when the result given is 10. recurrence relations.
THE IDEA OF A LIMIT 79
Several examples for use with calculators for pupils of different x2 = N
ages and abilities are given below. Official mathematical language
jV, since x is positive.
is used; this language will need to be adapted according to
classroom circumstances. For example, the Newton method for (ii) If a calculator with a square root is available, a very
square root can be described by saying attractive method for N l/3 is
"you take the average of the approximations

x andj —
N to get the
, next approximation
• • „ This is extremely quick and easy, giving for N = 10 and x i = 2,
the sequence
rather than 2, 2-1147, 2-1444, 2-1519, 2-1538,....
N
= i A',, + Again, if the (assumed) limit is x, (x ^ 0),
(i) If a reciprocal key is available on the calculator, the routine
=> x4 = NX
. N-l
=>x 3 = N.
(in) By a natural method this can be extended to other integral
is easy to use for
roots. For example
This method does not need any intermediate writing down, but
it does not converge so quickly as Newton's method. Pupils will
easily see that the method does converge, but may doubt whether it
converges to ^/N. If, for example, they take N = 10, and start with
x 1 = 3, they obtain successively (to 4 decimal places) and

3, 3-25, 3-1176, 3-1857, 3-1502, 3-1686,... all give sequences of approximations to N l!S .
so that successive values oscillate and become nearer together, (iv) An interesting method of finding the reciprocal of N
sandwiching a number which lies between 3-1502 and 3-1686. An without using division is the relation
algebraic calculation then gives, if the (assumed) limit is denoted
by .x,
Pupils might be given this relation and asked to explore what it
x = 1, + -——
N-\ does, for different values of ,V, and for different starting values. This
1 +x sequence only converges to I/TV from starting values fairly near
OX(1+X) = 1 + X + /V-1 l/N, and serves as a useful antidote for pupils who are beginning to
o x + x2 = x + N think that all sequences converge.
80 FROM GRAPHS TO CALCULUS

(v) The linear recurrence relation For pupils who are familiar with matrix notation it may be
convenient to use the notation
ax,, + b
Pn+l N\fpn
is well worth exploring for various values of a, b and c. \<J»+\
An interesting exercise is to find the temperature which Then pn+ l /q,,+ j is a better approximation to ^/N than pjqn.
is the same in °C as in °F. This is given by x = |(x — 32) and can be We give an example for N = 3 taking p l = q l = 1. The calcu­
solved iteratively using the recurrence relation lator is used only to verify how the sequence is converging.
Successive terms are
x,, + 1 =f(x,,-32).
4 10 76 208 1151
6 i
Conversely, the temperature which is the same as °F in °C, has the 2> 44' 120' 896 '

recurrence relation We note that (^^f)2 = 3-0003 ... The reader should consider why
this method works.
(viii) In two articles in Mathematics Teaching*, Dr. T. J.
These two recurrence relations have different convergence proper­ Fletcher shows an ingenious and effective method of approximat­
ties. ing using vectors. The article would give excellent material for
(vi) Another interesting recurrence relation is workcards, and once again shows sandwich methods at work.
(ix) The calculator is particularly useful for solving equations of
S l = a, Sn + i = rSn + a, where r < 1. the type x =/(x) by using the recurrence relation x,, + , =/(x,,)
with a suitable starting point. This does not always work, as some
This gives the sum to (n + 1 ) terms of the geometric series of the above examples illustrate, but it works often enough to be
a + ar+ ... + ar" worth trying. A good elementary discussion, though not suitable
for pupils before the sixth form, is found in Elementary Calculus
and can be used for series such as and Co-ordinate Geometry II by C. G. Nobbs, OUP, p. 339.
(x) The matrix
0-5 0-2
0-5 0-8
(vii) The following method of obtaining a sequence of rational
number approximations to ^/N gives the approximations in
"fractional" form. maps the vectors
Suppose pn and qn are positive integers such that pjq,, is 0-5 '0-2
an approximation to JN. Then let Pn +i=pn + Nqn and and to and
0-5 0-8
THE IDEA OF A LIMIT 81
respectively. Then the matrix maps
0-2
0-8

35)
to
0-5 0-2 0-5 '0-5 0-2 '0-2 '0-26
0-5 and 0-8 0-74
0-5 0-8 0-65 / 0-5 0-8
Writing

and

as OP, and OQ, respectively, their images as OP, and OQ 2 , and

so on, this situation is illustrated by figure 10.9. What happens
subsequently?

10.6. Some Mathematical Discussion

We return to the geometrical example in section 10.3 of a Fimire 10.9

sequence of squares and rectangles.
The following conversation was heard between two pupils aged
14 of average ability in a classroom in the North of England: some important points about a sequence which converges to a
limit. These are:
Anyela: You're wasting your time. You'll never fill that corner in.
Gary: 1 could if my pencil was sharper. (i) "You'll never fill that corner in". Angela realizes that it is impossible to
Angela: No, you couldn't. do more than a finite sequence of operations. At the nth step of the
Gary: I could draw more squares than you. construction, n may be very large, but it is always finite. We can never
Angela: You still couldn't fill the corner in. take an infinite number of steps.
(ii) In Gary's mind, the distinction between a particular term of the sequence
Angela and Gary's conversation shows a clear intuitive grasp of and the limit of the sequence is not clear. He realizes that by taking
enough steps, he can get as close as he pleases to an area of 2dm 2 .
However, he seems to think that, if he drew enough squares and
rectangles, he would actually reach 2dm 2 . In fact, if A,, is the area after n
steps, we have (see figure 10.3)
* Mathematics Teacliint/. June 1973 (No. 63) pp. 4 9: September 1973 (No. 64)
pp. 42-44, "Approximating by Vectors I, II". An < 2 for every whole number n.
82 FROM GRAPHS TO CALCULUS

(iii) Gary is clear, however, in his assertion of superiority "I can draw more More-able pupils will be able to see that at the nth step, the
squares than you", that however many terms of the sequence Angela has number of dm 2 filled is
drawn, more terms are always possible, and that the area always
increases. 1 1
Between them they realize that 1 +- + - + ...+
24 /•• - r '
A! < A 2 < A 3 < ... < An < A,, + 1 < ... < 2
and that the difference between An and 2 can be made as small as we The unfilled area of 1/2" ~ l dm2 can be made as small as we please
please by taking n large enough. by taking enough steps, but for every n there always is an unfilled
area. At this stage, the following wording should be used, without
In the class discussion which followed this experiment, the mention of limits:
pupils were equally divided between two points of view. The area
had now been represented arithmetically on the blackboard as We can make the filled area An as close as we please to 2 dm 2 by taking a large
enough number of steps. The filled area converges to 2 dm 2 .

Some pupils thought that the sum must always be less than 2 ('"cos The pupils' attention can also be drawn to the fact that we have a
you can't fill the corner in."). Others maintained that eventually sequence of numbers
the sum would exceed 2 ("You keep on adding more—it must get
bigger".)
In order that pupils gain the maximum of understanding about in which the terms are always increasing, but all the terms remain
limits from this and similar examples involving summation, it is less than 2. This sequence, which is bounded above, is of a
important to use a notation which hides nothing; the usual fundamentally different nature from the many sequences already
notation familiar to the pupils, such as
2, 3, 5, 12,
hides a great deal, and should not be used at this stage. which increase beyond all bounds.
The pupils' attention should be focused on two points: clarity "Achilles and the tortoise" provides another situation which can
about successive steps of the process, and the size of the difference usefully be linked to the last one. It also has the advantage that
between each member of the sequence (AJ of areas and 2. pupils are likely to start off not knowing exactly where or when
Area filled/dm 2 Area unfilled/din 2 Achilles catches the tortoise. A step-by-step analysis with a
1st step 1 distance-time graph is helpful.
2nd step i To simplify the numbers, we use a rather speedy tortoise.
3rd step "Achilles can run twice as fast as the tortoise, so he gives the
4th step
5th step tortoise a kilometre start. While Achilles runs the kilometre, the
tortoise runs \ km; then while Achilles runs that ^-krn, the tortoise
nth step
runs 5 km; and so on."
THE IDEA OF A LIMIT 83
————————————————————————»*a' S

W ' 1 1
."• 4. ' A km to go

X 2 > 2 km to go

km to go

1 1

0 1 km 2 km

Figure 1 0.1 1

If we take another form of the problem, in which Achilles is ten

times as fast as the tortoise, a decimal representation is useful:
Distances in km from Achilles' starting-point
Achilles Tortoise
To start with 0 1
After one stage 1 1+A = H
After two stages 1 +TO+TOO = H 1
After three stages '+To + TO + TBTre=

The catching-up point is not so obvious here, although pupils who

are familiar with recurring decimals will realize that the sequence
first stage second stage third time
stage 1, 1-1, 1-11, 1-111,....
Figure 10.10 Distance-time graph for "Achilles and the tortoise". converges to l£.
This leads us to non-terminating decimals.

Distance's in km from Achilles' starting point 10.7. Non-Terminating Decimals

Achilles Tortoise
To start with 0 1 In the last chapter, we used sandwiching methods to discuss
After one stage 1 non-terminating decimals. For example,
After two stages 1+4
After three stages 1 +3 + 1 0-3 } s; 0-4
0-33 i sc 0-34
After a few stages, pupils will be certain that Achilles catches the
tortoise at the 2 km mark, and should then record the distances of
the two participants from this point at each stage. and so on.
84 FROM GRAPHS TO CALCULUS

We can think of this in several ways. A number which often baffles pupils is 0-999 ... or 0-9. They may
First, the sequence of rational numbers 0-3, 0-33, 0-333..... is say that the difference between 0-9 and 1 is "point nought recurring,
increasing, and it is also bounded above by j. We can, however, get one", indicating that they do not yet understand the significance of
as close as we please to 5 by going on far enough in this sequence. It the recurring figure. The sequence of rational approximations to
is instructive for pupils to calculate the difference between the 0-9 is
successive terms of the sequence and ^:
0-9, 0-99, 0-999,...
•TO = 301
33 1 and the sequence of differences between these numbers and 1 is
100 300
333 1 0-1, 0-01, 0-001,....
' 1000 3000

and so on. Thus we can make the rational number 0-999... 9 as close as we
We say that the sequence please to 1 by going on far enough in the sequence. It is perhaps
unfortunate for young pupils that this fact is written
0-3, 0-33, 0-333,..., converges to i
0-9 = 1
Similarly, the second sequence
in the same way that we write 0-3 = ^. What is meant is that the
0-4, 0-34, 0-334,....
sequence 0-9, 0-99, 0-999,... converges to 1.
is decreasing, all its terms are greater than ^, and it also converges Appropriate forms of words to discuss these phenomena with
to | because the terms can be made as close as we please to 3 by 'pupils are:
going on far enough in the sequence. (i) 0-3 is the number whose approximations are
Alternatively, we may look at the successive sandwich in­
equalities, note that the thickness of the sandwiches can be made as 0-3, 0-33, 0-333,....
Sandwich Thickness of sandwich
0-3 =? i sc 0-4 0-1 The approximation can be made as close as we please to 5, so
0-33 s: 1 s: 0-34 0-01 we say
0-333 < 1 ^ 0-334 0-001
0-3 =i
small as we please by going on far enough and that there is one
number, y, which is common to all the sandwich inequalities. This (ii) 0-9 is the number whose approximations are
is sufficient to show that both the sequences
0-9, 0-99, 0-999,....
0-3, 0-33, 0-333...
and The approximation can be made as close as we please to 1,
0-4, 0-34, 0-334 ... so we say
converge to j. 0-9 = 1.
THE IDEA OF A LIMIT 85
10.8. The Key Idea approximations to ,y6 must also decrease, so an approximation
can be found which is as close as we please to v/6.
Suppose we have a sequence of numbers. We shall say that this
Lower Upper Difference between
sequence converges to a number or approaches a limit if we can approximation approximation v'6 and each approximation
convince ourselves that we can get as close as we please to this 2 3 Less than 1
number or limit by going on far enough in the sequence. 2-4 2-5 Less than 0-1
2-44 2-45 Less than 0-01
We shall use this key idea
2-449 2-450 Less than 0-001
We can get as close as we please to...
Hence we say that the sequence of rational approximations 2,
again and again, for it is the essential first step in developing the 24, 244, 2449,... converges to ^6.
limit concept.
What we cannot do, and it would not be appropriate to try in
any systematic way with sixteen-year-old pupils, is to prove in all
the examples we have given that the sequences have limits. At this
age, an intuitive grasp of the idea of a limit should be the objective. 10.10. Fibonacci Sequences

Sometimes the sandwich argument needs to be used rather

differently. Fibonacci sequences provide an example.
10.9. A Sandwich Inequality Argument for Convergence The standard Fibonacci sequence has 0. 1 as its first two terms.
Each successive term is then the sum of the two previous terms, so
The search for x/6 discussed in section 10.4 provides an that the sequence is
interesting example of a sequence for which the convergence can be 0, 1, 1, 2, 3, 5, 8, 13, 21, 34,....
proved by using a sandwich inequality.
In this work we established that This is, however, only the simplest example of a whole class of
Fibonacci sequences, as a Fibonacci sequence can be constructed
24 < v/6 < 2-5 for any choice of first two terms.
244 < < 2-45 The ratios of successive terms of a Fibonacci sequence can be
calculated to give a new sequence. The standard sequence has as its
and so on. sequence of ratios
It is not possible here to find how close 2-4, 2-44, etc., are to ^/6. 1-5.
However, we can guarantee that at each stage the differences •6. 1.625, 1-615384. 1-619047.
between the upper and lower approximations get smaller, and can This sequence has two interesting properties.
be made as close to zero as we please, by going on far enough in the (i) It does its own sandwiching, successive terms being alter­
sequence. The differences between ^/6 and the rational number nately greater and less than the preceding term. In fact, we can split
86 FROM GRAPHS TO CALCULUS

the sequence into two sequences, one of which is increasing and the useful preparation for the limits "at a point" used in differentiation,
other decreasing: and this preliminary work is important not only in enabling pupils
1-5, to tackle differentiation with understanding, but also in keeping
1, 1-6, 1-615384,...
1-6, mathematics alive as an investigating exploratory subject in the
1-625, 1-619047,...
years preceding the 16+ examinations. Nearly all this work can
(ii) It can be seen that if this pattern continues, and if the terms and should be done in play-like activities in which the limit idea
of the upper sequence can be made as close as we please to those of can develop over several years before being applied in the context
the lower sequence, there will be a number sandwiched in between of differentiation. The key concept in all work on limits is that we
them. It is intuitively clear that the sequence of ratios converges to can get as close as we please by... In different settings, we get "as
a number whose value is 1-61..., although we cannot say precisely close as we please" in different ways. The first two examples are
what number this is. In the case of this sequence, we are sure that it drawn from this chapter, while the third and fourth look forward
converges, although we cannot write down an expression for its to chapter 12.
limit. This is in contrast to the example of section 10.9, when we (i) We get as close as we please to the limit of a sequence by
knew the limit was 6. going far enough in the sequence.
(ii) We get as close as we please in drawing a curve of pursuit by
taking short enough steps.
(iii) We get as close as we please to the tangent to a curve at P by
10.11. Conclusion taking a chord through points P and Q which are close enough
together.
In each activity in this chapter, pupils will have the specific (iv) We get as close as we please to the actual velocity of a
objective of investigating a particular situation, but the teacher will particle by taking the average velocity over a short enough
also have in mind the growth of the concept of a limit. It is helpful if interval.
he uses forms of words such as "Can you get as close as you please Formal work on differentiation and other limit ideas at the next
to the number? (or the curve, or the point?)" stage will be much more intelligible to pupils if the limit idea is
The limits discussed in this chapter are largely limits "at allowed to grow slowly and informally before the official notation
infinity", although we have not used the phrase. They form a very for limits is introduced.
 11 Area under a Graph: Velocity and Displacement
11.1. Introduction made of this aspect of area in the early years in the secondary
school. For instance, in the block graph in figure 11.1 the shaded
The idea that area under a graph can convey useful information area under the graph tells us the total number of pupils in the class.
should not be new to pupils, because it can be brought out in If a new pupil were to arrive, and information about him were
graphical work in the late primary school or early secondary added to the graph, the height of one of the columns would be
school in the context of block graphs or histograms. More could be increased by one unit and hence an additional area of one unit
square would be shaded. Areas of histograms will become more
significant and important in later studies of statistics, but an early
introduction to the interpretation of areas under graphs would be
most profitable.
frequency
In this chapter we concentrate on velocity-time graphs and on
the problem of convincing pupils that the area under a velocity-
time graph gives a measure of displacement. Many traditional
books and courses do not include this work, but a number of
modern courses for average and more-able pupils do.

4- 11.2. The SMP Treatment of Area under the Velocity-Time

Graph
3-
The treatment of area under a velocity-time graph in SMP Book
2-
X, p. 174, is quoted below.
3 Areas under graphs
1 - (a) A lorry carrying a wide, heavy load travels at a speed of 10 km 'h for 3
hours and at 7 km/h for the next 2 hours. How far has the lorry travelled?
(b) A speed-time graph for the lorry's journey is shown in Figure 13.
How many of the red squares can be fitted into the shaded area? We can see
123456
that the area of the red square represents a distance of 1 km. Hence the total area
number of children in family under the graph represents a distance of 44 km. i.e. the area under the graph
represents the total distance travelled by the lorry. [Red is shown here shaded.]
Figure I I.I Block graph of the number of children in each family for a class. (r) Figure 14 shows a speed-time graph for a car travelling along a motorway
87
 FROM GRAPHS TO CALCULUS

E
_^;

T3
CD
CD
Q.
co

123456 12345
Time (h)

Fia. 13. Fig. 14.

at a constant speed of 100 km 'h for 1 hour and then at a constant speed of 120 maximum. The waste is then produced at this constant rate for 4 hours. After
km.'h for 2 hours. this it decreases. The graph in Figure 16 shows the amount of waste per hour
Do you agree that the red square represents a distance of 20 km? How many flowing into the river at various times during the day.
of the red squares can be fitted into the shaded area? What is the total distance The shaded rectangle represents a quantity of 5000 litres of industrial waste.
travelled by the car? The area of triangle .4 is i x2 x 15 = 15. Thus the area of region ,4 represents a
{ti) The graphs in sections (h) and k'| represent idealized situations. Lorries quantity of 15 000 litres of waste.
and cars cannot, in fact, change speeds instantaneously as the graphs suggest. A Calculate (i) the area of region B: (ii) the area of region C. Hence write down
more realistic description of the lorry's journey is represented by Figure 15. the total amount of waste flowing from the factory per day.
Use the graph to describe the lorry's journey in words. Calculate the area
under the graph and hence write down the distance the lorry travelled.
Now sketch a more realistic version of the car's journey than that represented The technique of leading the pupils gently into a new situation,
by Fiaure 14. Do not forget that the car has to travel a total distance of 340 km.
(t-I The rate at which industrial waste flows into a river from a factory
which is so ably used in this exercise, is one which is familiar to and
increases during the first two hours of a working day until it reaches a used by all teachers. The pupils are first convinced that the area
AREA UNDER A GRAPH: VELOCITY AND DISPLACEMENT 89

t 15
<u velocity
0) Q-
w w in km/h
CD g

.I'S 10
If
II
1 5

6 a.m. 7 1112 1
time (h)

Fig. 16.

under a velocity-time graph represents the displacement in the

simple case in which the velocity-time graph is a step-function. Care
needs to be taken over this stage, for pupils are accustomed to time in hours
representing displacement by a length on a graph, and it is a further
abstraction to use a number of units of area to represent a number Figure 11.2
of units of displacement. Fortunately, it is easy for the pupil to
relate ,
displacement = velocity x time
to the area of a rectangle whose length represents velocity and
whose breadth represents time. The units are a great help, and for Similarly, in the industrial waste example in the extract, one axis
instance in figure 11.2, the shaded area represents a displacement represents litres per hour, and the other represents hours, so the
of 8 km, and is a quantity whose units are area represents
kilometres litres
x hours -- x hours,
hours hours

or kilometres. or litres. This is a very effective way of remembering or working out

90 FROM GRAPHS TO CALCULUS

what information the area under the graph conveys, provided that
velocity
the graph is a step-function, so that the area under it is a rectangle in m/s
or a sum of rectangles.
The next stage in the teaching strategy in the extract is shown in
(d), where it is suggested that although the velocity-time graph is
no longer a step-function, the area under it still represents the
displacement. By (e) the use of the word "Thus" in the sentence
"Thus the area of region A represents a quantity of 15 000 litres of
waste." makes the pupil believe that the fact that the area under the
graph gives useful information has been thoroughly established for
all graphs.
This technique, which is extremely useful in many places in the
teaching of mathematics, depends on the fact that it is common for
mathematical ideas to generalize from simple cases into a more
widely applicable setting. However, able pupils may recognize and
be worried when jumps are made without explanation, and the
teacher needs to make conscious decisions whether to discuss these time in seconds
jumps with particular pupils or groups of pupils. The decisions
depend on balancing the development of pupils' critical awareness Figure 11.3
against the degree of difficulty that a fuller treatment may involve.
Unfortunately, when an attempt is made to justify the repre­
shown in figure 11.3. The argument runs as follows. As the particle
sentation of displacement by area under a velocity-time graph, a
starts with velocity 2 m/s and finishes with velocity 4 m/s, the
fallacious argument is sometimes used, and this is described in
average of its initial and final velocities is 3 m/s. Thus it might be
section 11.3.
regarded as travelling for 2 seconds at an average velocity of 3 m/s,
There is available, however, a straightforward argument which
so that it covers a distance of 6 metres. Also, the number of units of
forms an important step towards the understanding of integration,
area below the graph is . .
and which is intelligible and convincing to pupils. This argument is
discussed in section 11.4. 2+4
x 2 = 6.

11.3. An Incorrect Argument Hence this example might be thought to illustrate the result that
the area under a velocity-time graph represents the displacement,
Consider a particle which accelerates uniformly from a velocity even when the velocity is not constant.
of 2 m/s to a velocity of 4 m/s in 2 seconds. Its velocity-time graph is However, in general, the average of the initial and final velocities
AREA UNDER A GRAPH: VELOCITY AND DISPLACEMENT 91
is not equal to the average velocity. It will be recalled that
velocity
total displacement in m/s
average velocity = -
time taken

This is not in general equal to

} (initial velocity + final velocity).

The fallacy can be seen clearly by considering a particle which

travels at 2 m/s for 49 s, and then at 1 2 m/s for 1 s. In this time it has
a displacement of (98 + 12) m = 110 m, and so has an average
velocity of ^5^ m/s, or 2-2 m/s, but the average of its initial and final
velocities is j(2 + 12) m/s or 7 m/s. 6km 4km
It is only when a particle moves with constant acceleration that
its average velocity is equal to the average of its initial and final
velocities. In this case the velocity-time graph is a trapezium,
whose area is found by multiplying the average of its two parallel
sides by its width. To establish correctly in the previous example
that the average velocity is 3 m/s, it is necessary first to calculate time in seconds
that the displacement in the 2 seconds of the journey is 6 metres. Figure 11.4

We now put alongside X a slow man S and a fast man F who

11.4. A Sandwiching Argument for Displacement and Area
walk with the constant velocities shown in figure 11.5(ii).
Two inequalities can now be written down from the pupils'
A straightforward and convincing sandwiching argument can be
intuitive insight. First, it is clear to them that between the times
used to demonstrate that the area under a velocity-time graph still t = } and t = 9 ......
represents the displacement, even when the velocity is not
constant. urea under graph of S ^ area under graph of X ^ area under graph of F.
It is clear that the fact that, for constant velocity, the area under a
Secondly, they know that
velocity-time graph represents the displacement, extends im­
mediately to the situation where the velocity-time graph is a step- the faster you go, the further you get in the time
function (figure 1 1.4). The extension to other velocity-time graphs
so because S is always walking slower than X, and F is always
can then be developed as follows.
walking faster than X,
A man X walks with a variable velocity illustrated in the
velocity-time graph of figure 11.5(1). displacement of S 5C displacement of X «; displacement of F.
 92 FROM GRAPHS TO CALCULUS

velocity velocity
in m/s in m/s

1 -

time in seconds time in seconds

Figure 11.5 A sandwiching representation connecting displacement with the area velocity
under a velocity-time graph. i n m/s

Clearly a sandwich argument is going to develop. L

It has already been established that, for step-function velocity-
time graphs such as those of S and F, the displacement is ('")
represented by the area under the velocity-time graph. Thus the ;
two sets of inequalities can be put together
area under graph of S area under graph of X area under graph of F
represents represents
displacement of S displacement of X displacement of F

The two quantities in each box are equal in magnitude, but it is

important to realize that the pupils have yet to be convinced that
the area under the graph of X represents the displacement of X.
Now for most of the time S is walking considerably slower than X.
time in seconds
 AREA UNDER A GRAPH: VELOCITY AND DISPLACEMENT
93
velocity He could be allowed to speed up for some of the time and still
in m/s
always be walking slower than X. Hence at half-time (or some
other convenient time) his velocity is instantaneously adjusted as
shown in figure 11.5(iii). The velocity of F is similarly adjusted.
The same two inequalities still hold:
area under graph of S area under graph of X area under graph of F
represents represents
displacement of S displacement of X displacement of F

Now another two instantaneous changes of the velocities of S

and F are put in, as shown in figure 11.5(iv).

It is still true that

area under graph of S area under graph of X area under graph of F
represents represents
displacement of S displacement of X displacement of F
9 time in seconds

velocity More instantaneous changes of velocity for S and F can be

in m/s introduced (figure 11.5(v)), and it still remains true that
area under graph of S $ area under graph of X ^ area under graph of F
represents • represents
displacement of S displacement of X displacement of F
(v)

At this stage pupils should be convinced that the inequalities

hold, however many times the velocities of S and F are adjusted.
The areas under the graphs of S and F get closer together, and the
2 - displacements of S and F also get closer together. Moreover, and
this is the crucial point, the displacements of S and F can be made
as close together as we please by the same process.
It follows from the sandwiching assumption of section 9.7, and it
will seem evident to pupils, that only one number is sandwiched
between the areas of all possible graphs S and the areas of all
possible graphs F. This number is the area under the graph of X,
0 1234567 9 time in seconds and represents the displacement of X.
 94 FROM GRAPHS TO CALCULUS

It should be made clear by examples that these ideas do not only regarded as positive or negative according as the values of the
apply to velocity-time graphs, but also to other graphs such as that function are positive or negative. The reason for this is shown by
given as example (e) in the SMP extract. The reason that velocity- the following example. The parallel between displacement and
time graphs are dealt with first is that pupils are more familiar with area having been established, it needs to be maintained in all
velocity than they are with other rates. situations.
Consider the velocity-time graph of the particle X shown in
figure 11.6. Between the times t = 2 and t = 4, X is travelling with a
11.5. Negative Velocity and Negative Area positive but decreasing velocity. At 4 seconds it instantaneously
stops and starts off again in the opposite direction. Hence at the
A distinction needs to be made between the use of the word area end of 10 seconds it has travelled a distance represented by the area
in chapter 7, and some implications of its use in this chapter. The A in figure 11.6 forwards, and a distance represented by the area B
area of a plane figure, discussed in chapter 7, is always regarded as backwards. In order that the total area under the graph should
positive, whereas the area under the graph of a function needs to be represent the displacement at the end of the journey, area B must

velocity '
in m/s
velocity
in m/s

10 time in seconds

-1

-2

1 9 time in seconds

Figure 11.6 Negative velocities and areas. Figure 11.7 The trapezium rule.
AREA UNDER A GRAPH I VELOCITY AND DISPLACEMENT 95

be regarded as negative. Pupils should be aware that it is not an It is not clear in speech whether the trapezium rule refers to the
arbitrary decision that areas below the x-axis are regarded as idea of dividing the region under the graph into trapezia and then
negative. saying that the area under the graph is approximately equal to the
sum of the areas of the trapezia, or whether the trapezium rule
refers to the formula to which the idea leads.
11.6. The Trapezium Rule The idea is much more important than the formula. Formulae
are never easy to remember, and the trapezium-rule formula is no
In numerical work, pupils will often obtain sandwich in­ exception. In more advanced work other formulae are nearly
equalities for the area under a graph, as suggested above. If an always used when approximating to areas under graphs, and in
approximate numerical value is needed for an area under a graph, elementary work it always suffices to calculate the area of each
instead of a pair of inequalities, the trapezium rule is the best trapezium in turn. The calculation of areas under graphs is
method for pupils at this level to use. discussed further in chapter 13.
 12 Velocity and Gradient
12.1. The Tangent to a Curve

This chapter is concerned with the idea of a tangent to a graph

and with developing the relationship between the gradient of a
tangent and velocity. This is generalized so that other rates of
change are also linked with gradients.
A number of diverse ideas from various realms of experience
need to have come together if pupils are to understand the idea of a
tangent to a curve.
Curve stitching (figure 10.4) shows how a line can make a very
close approximation to a curve at a particular point. Each stitch Figure 12.1
forms a tangent to the curve, and the curve is enveloped by all the
tangents. line which shows the direction of the curve at that point. The
The actual direction of motion of a point travelling along a curve tangent may also be considered as the straight line which fits the
is also an important part of the tangent concept. Young pupils curve most closely in the neighbourhood of that point (figure 12.1).
often have an actively wrong idea about this. They think that when Later in the chapter it will be seen how much information is
the string of a conker breaks, the conker flies out along the conveyed by the gradient of a curve. So far gradients have only
direction of the string; they think that when two children whirl one been mentioned in the context of straight lines, so it is natural to try
another around and lose their grasp, they will fall directly to define the gradient of a curve at a point as the gradient of the
outwards. This idea needs to be replaced by the explicit idea that tangent line to the curve at that point. This is why, before any of the
the direction of motion in a curved path is along the tangent. subsequent work in this chapter is attempted, it is important for
Examples from athletics are helpful here, and can often be pupils to have well-developed intuitive ideas about tangents to
clearly seen on television in slow motion. When a discus or curves.
hammer is released it starts to move in the direction along the These points are well brought out in the following extract taken
curve which the thrower described with it, that is, along the tangent from Modern Mathematics for Schools, Book 7, p. 42.
to the curve. A runner rounding a bend is at any instant running in
the direction of the tangent to the bend. A fielder catching a cricket
ball recoils in the direction of the tangent to the ball's path. 2 The gradient of a curve
Discussion of such examples as these will help pupils to become Figure 5 shows an aircraft coming in to land. The direction of travel of the
explicitly aware that the tangent to a curve at a point is the straight aircraft at a point P on the flight path is the same as the direction of the tangent
96
VELOCITY AND GRADIENT 97

at P. The gradient of the tangent at P gives a measure of the steepness of the

curve at P. displacement
The gradient of a curve changes from point to point, and the gradient of the in metres
curve at P is defined to be the gradient of the taiiyent to the curve at P.

12.2. The Need for the Gradient of a Curve

M
In chapter 8, straight-line graphs were used as models of the way
in which the displacement of an object moving at constant velocity
depends on the time taken. In the real world, cars, trains, runners
and swimmers do not move with constant velocity. In these cases
the straight line may not be a good model, and pupils may
recognize this. The question which then arises is how to calculate
velocity at an instant from a curved displacement-time graph in the
same way that the speedometer of a car indicates the velocity of the
car at an instant.
Pupils should first recognize from a number of numerical
examples that if a graph is drawn showing the position of a runner a+h time in seconds
at each instant, then his average velocity over any period of time
corresponds to the gradient of a chord of the graph.
Referring to figure 12.2, the average velocity between the times a Figure 12.2
 FROM GRAPHS TO CALCULUS
displacement
and a + h is found by dividing the displacement LM by the time in metres
interval h. This average velocity must be recognized and identified
as the gradient of the chord PQ.
If pupils feel, in figure 12.2, that at P the runner is actually going
faster than the average velocity calculated, then they are beginning (b)
to see that the gradient of the curve itself will correspond to the
actual velocity at P.
The way forward to establishing the link between actual velocity
and gradient of a curve more firmly is to shorten the time interval
over which the average velocity is calculated. In figure 12.3,
average velocities over the time intervals h : , h2 and h3 correspond
to the gradients of the chords PQ l5 PQ 2 and PQ 3 respectively. On
any graph which a pupil draws, the chords PQ for which Q is very

displacement
in metres time in seconds

displacement
in metres

(a)

(c)

time in seconds
Figure 12.3 Approximations to instantaneous velocity through the link with the
gradient of a displacement-time curve. time in seconds
 VELOCITY AND GRADIENT 99

close to P are very difficult to distinguish from the curve itself in the displacement-time graph using a ruler and pencil, and by calculat­
neighbourhood of P. This leads easily to the conviction that the ing the gradient of their tangent line from measurements made
"actual velocity" at P is given by the gradient of the tangent to the from their graph. They can also learn through examples that the
curve drawn at P. gradient of a curve is useful in other circumstances.
It is true that mathematical difficulties can arise in trying to be For instance, a domestic gas meter measures the volume of gas
precise in explaining what is meant by "velocity at an instant" or used by the householder, and it is possible to draw a graph of the
"gradient of a curve", and mathematical purists may say that volume of gas used against time. This might have the general
neither concept is yet properly defined. At this stage this is characteristics of figure 12.4.
unimportant, for the concepts of "velocity at an instant" and In this example, the gradients of chords and tangents give
"gradient of a curve" are intuitively clear to pupils. It is much more information about the average rate of flow of gas and the actual
important for these concepts to be linked in the minds of the pupils rate of flow of gas. The gradient of the tangent at P corresponds to
than it is for pupils to be able to articulate clearly and concisely the rate of change of the volume of gas at P.
what the concepts mean. In fact, the concepts of velocity and Another example in which gradient is useful is provided by
gradient are strengthened by the link between them, and precise acceleration. In the same way that the average velocity of a moving
definitions are inappropriate until after that link has been made. body is calculated by measuring changes of displacement in a given
time, so also its acceleration can be calculated by measuring

12.3. Gradients of Tangents by Drawing

velocity in
metres per
It is now that pupils can usefully do examples in which they second
calculate velocity at a point by drawing a tangent to a
10
volume of
gas used in
cubic metres

—*
lime in seconds
time in seconds
Fieure 12.4 I-ieurc 12.5
100 FROM GRAPHS TO CALCULUS

changes of velocity in a given time. A sprinter who starts from rest 12.4. Remarks
and reaches a velocity of 10 metres per second in half a second has
an average acceleration over that time of 20 metres per second in This is a convenient stopping point for a school course and the
each second, or 20 m/s 2 . The average acceleration corresponds to SMP course and some other courses go no further than this
the gradient of a chord of the velocity-time graph. The acceleration towards calculating either areas or gradients without approxi­
of a moving body at a particular instant corresponds to the mations being made en route. Pupils who have reached this stage
gradient of the tangent to the velocity-time graph at that instant. are well prepared for integration and differentiation, and in
addition they will understand clearly why integration and differen­
tiation are important processes.
y /
While pupils are drawing tangents by eye, brief informal
discussion of drawing methods can help to form ideas which will be
useful in calculating gradients of curves without drawing. Pupils
use two methods.
(i) The ruler is slid up to the curve, its direction being adjusted
by eye until it approximates to the direction of the curve at
the required point P.
(ii) The pencil is placed at P and the ruler slid round from a
position such as PQ to a position where the second

Figure 12.6

Using the language of rates, velocity is the rate of change of

displacement, and acceleration is the rate of change of velocity. In
general if a graph of a function/: x i-> y is drawn, the gradient of the
tangent at P corresponds to the rate of change of y with respect to x
at P. Very often when we talk of a rate of change it is a rate of
change with respect to time which is meant, and in this case the
phrase "with respect to time" is often omitted. Figure 12.7 Methods of drawing tangents by eye.
VELOCITY AND GRADIENT 101

displacement displacement displacement

in metres in metres in metres

time in seconds time in seconds time in seconds

Figure 12.8 Averaac velocity over a short time interval.

y/
xi—> y xi—» y

Figure 12.9 Approximations to the gradient of a tangent.

102 FROM GRAPHS TO CALCULUS

intersection of ruler and curve, Q, appears to the eye to In order to calculate velocity from a displacement-time graph
move from one side of P to the other. without using drawing and measurement, pupils must return to the
The first of these drawing methods leads ultimately in the study idea that chords PQ for which Q is very close to P are difficult to
of differentiation to the method of defining the derivative of a distinguish from the tangent to the curve at P. When this statement
function at a point as the gradient of the best linear approximation is put into the context of displacement-time graphs, it becomes
to its graph (section 14.5). The second method leads to the idea very difficult to distinguish "the average velocity over a short time
that the derivative of a function at a point is the limit of the interval a to a + /i" from the actual velocity at P (figure 12.8).
gradient of a chord (section 14.3). Both ideas are valuable, and This suggests a method of calculation of velocity at the instant a
both can be built up from pupils" actions in drawing. corresponding to point P on the curve, for it is now a small step to
the conviction that we can get as close as we please to the velocity at
time a by taking a short enough time interval from a to a + h anil
12.5. Towards Definitions of Velocity and Gradient calculating the average velocity over this interval.
In the corresponding language about gradients of tangents to
In section 1 2.2 the intuitive concepts of velocity at an instant and graphs, the words in italics above become (figure 12.9).
gradient of a curve were linked, and in section 12.3 it was suggested we can get as close as we please to the gradient of the tangent at P by
that pupils should do plenty of numerical examples in which they taking
PQ.' 0 ~ close enough to P and calculating the gradient oj the chord
draw a tangent line to a curve and then calculate the gradient of the
tangent line using measurements made from the graph. This idea, differentiation, is developed in chapter 14.
 13 The Definite Integral

13.1. Introduction centres of gravity, moments of inertia, means and standard

deviations.
In chapter 11, the idea that the area under a velocity-time graph In the sections which follow elementary and pictorial arguments
gives a measure of displacement was discussed; it was also are used to convince pupils that
indicated that the areas under other graphs can give useful tp+i aP+ i
information. This chapter discusses methods of calculating areas x"dx =-
p+l p+l
under the graphs of functions, and so introduces the definite
integral. provided, of course, that p ^ — 1. A pictorial demonstration of
Textbooks for pupils under sixteen vary considerably in their
treatment of integration. Some do not consider it at all, others start
with anli-differentialion while yet others emphasize area. None of
these books uses the calculator, which will greatly influence the where /. is a fixed real number is given, and a similar demonstration
teaching of integration in the next few years. of
Most pupils at present first meet integration as the reverse of P f
(f(x) + g(x))dx= f(x)dx+ g(x)dx
differentiation, and only later do they use integrals to calculate
areas. The authors believe that this approach gives pupils a very
may be given, so that a variety of elementary functions can be
wrong impression conceptually, and is at variance with the way in
integrated.
which integration developed historically.
In the next sections we present a possible way of teaching
integration to beginners, based on the use of calculators, area and 13.2. Areas and Sandwiches
summation. The treatment is independent of ideas about gradient
and differentiation. Pupils may or may not be developing ideas The problem is to calculate the area under the graph shown in
which lead to differentiation in parallel with those which lead to figure 13.1 (i), and the only knowledge required is that of the area of
the definite integral, but they will certainly not suspect at this stage a rectangle. The procedure is the same as that described in chapter
that the calculation of areas and the calculation of gradients are 1 1 , when a "fast man" and "slow man" were introduced. In this case
related problems. In later work, in the sixth form, the fact that a the interval on the x-axis fromu to f is divided into« parts, and two
definite integral is calculated by summation becomes increasingly- step-functions, the lower step-function shown in figure 13.1 (iij and
important, for it enables applied mathematicians, statisticians and the upper step-function of figure 13.1 (in). are constructed. A
others to use integration in order to calculate such quantities as sandwich argument can then be used.
103
104 FROM GRAPHS TO CALCULUS

Y = f(x)
y = f(x)

upper step-function

Figure 13.1 (i) Area under the graph of y=f(x); (ii) lower step-function;
(iii) upper step-function. For simple functions, numerical calculations are very quickly
made, as the following examples show.
We first find a sandwich inequality for the area under the graph
of y = x2 between x = 0 and x = 1, using lower and upper step-
functions with ten steps each of width 0-1. Figure 13.2 shows the
lower step-function.
The sandwich inequality is
0-285 sc area ^ 0-385
and is obtained very quickly indeed. (A calculator is more
hindrance than help!)
A calculator is more help if the inequality is refined by taking a
step-function with twenty steps each of width 0-05. This time the
sandwich inequality is
0-30875 sc area s£ 0-35875
If an approximation is made to the area under the graph by
averaging the upper and lower estimates (see section 7.3) the result
lower step-function is 0-33375. Hence, to 2 decimal places, an estimate of the area under
THE DEFINITE INTEGRAL 105
y = x 2 between x = 0 and x = 1 is 0-33, but a clear distinction
must be drawn between the inequality which is precise and the
average result which is approximate. 1 -
This work can be followed up by using similar approximate
methods to calculate areas under the graphs of y = x 3 and y = x4
between x = 0 and x = 1.
The table below gives approximate results, using ten-step step-
functions, for the areas under the graphs of y = xp for
p = 0,1, 2,..., 9. The step-function areas are omitted for p = 0 and
p = 1 because the required areas are those of a rectangle and a
triangle and can be calculated directly. In these cases the number in
the right-hand column is the actual area under the graph.
Average of
the lower
step-function
area and
upper
step-function
Area under Area under area (to
Value Equation lower upper two decimal
of;> of graph step-function step-function places)
0 0-5 1-0 x
0 v= 1 _ _ 1 Figure 13.2 A lower step-function for the graph of y = .v 2 .
1 V =x — — 0-5
2 V = X2 0-285 0-385 0-33
3 V = X3 0-2025 0-3025 0-25 This is not a proof, but it is a very convincing demonstration to
4 y = x4 0-1533 0-2533 0-20 pupils at this stage that the area under r = x'' between x = 0 and
5 y = x' 0-1208 0-2208 0-17
y = x (> 0-15 x = 1 is l/(p+l).
6 0-0978 0-1978
7 y = A" 7 0-0808 0-1808 0-13 It is suggested that the numerical verification of this result, given
8 v = x8 0-0677 0-1677 0-12 above, is sufficient for almost all pupils, and a formal theoretical
9 y = x" 0-0574 0-1574 0-1 1 proof should be postponed to the next stage of work.
The sequence of approximations in the right-hand column is
1,0-5,0-33, 0-25, 0-20, 0-17....
13.3. Definition of the Definite Integral
and suggests very strongly that the actual areas are
The definite integral of a function /between a and / can now be
J 1 1 i L
i it 3' 4> 5' 6> introduced as the area under the graph of y = f(x) between x = a
106 FROM GRAPHS TO CALCULUS

and x = t, where t ^ a. It is written as

r
f(x)dx
M

Thus, the definite integral is merely a new notation for a familiar

idea, and the result obtained at the end of section 13.2 can now be
written

x"dx = ——-

where p is a positive integer or zero.

The same formula can be verified quite easily when p = T and
p = i, by the same method.

Area under Area under

Value Integral lower upper
ofp required step-function step-function Average

x"2 <;.\- 0-6105 0-7105 0-66 0-67

0-6874 0-7874 0-74 0-75

Hence
Hmdx ^ area under graph ^ DM (5.x
It will now seem very justifiable to pupils that
where Zm<5.x may be read as "the sum of all the terms like
xpdx = ——- Moreover, the values of/(.x) on the strip lie between m and M, so
o P+l that the inequality now reads
for all rational numbers p ^ 0.
I.mdx ^ I /(.x) dx ^ I.M&X
The curious notation \f(x)dx often gives rise to comments and
queries. Pupils may be told that 6x is a common notation for the
and I" can be regarded as another form of the summation sign.
width of each strip, so that if HI is the height of the lower step-
The notation for integrals certainly needs further discussion at
function on that strip and M the height of the upper step-function
the next stage, when use will be made of definite integrals to handle
on the same strip, we have
limits of sums, for instance when studying the mean and standard
mdx ^ area of strip ^ Mdx. deviation of a continuous distribution.
THE DEFINITE INTEGRAL 107

13.4. Integrals of Increasing Functions From the sandwich assumption of section 9.7, there is just one
number which is sandwiched between every possible lower step-
By this stage, pupils should have realized intuitively that if they function area and every possible upper step-function area, and we
make the steps of the step-functions narrower, they make the can, by making the steps of our step-function sufficiently narrow,
sandwich inequality thinner. Indeed for all the graphs so far identify this number.
considered the thickness of the sandwich inequality, or the This argument applies not only to the integral of y = xp between
difference between the upper step-function area and the lower step- 0 and 1, but to every increasing function because the shaded areas
function area, is of the form shown by the shaded area in figure can always be stacked. Hence every increasing function has a
13.4. All these areas can be slid over to the left and stacked into the definite integral which can be evaluated by a sandwiching process;
column shown against the y-axis. If the steps all have equal width so also do decreasing functions.
vv, the shaded area is w x 1 = w. Hence the sandwich can be made Pupils under sixteen will not concern themselves with the details
as thin as we please by choosing w small enough. of this argument, but the intuition that smaller step-lengths give a
thinner sandwich should be encouraged. Later, more-advanced
students will prove that not only every increasing function, but
also every continuous function (and some other functions as well)
has a definite integral which can be evaluated by a sandwiching
process.

13.5. Extending the Result

Having assumed that

'i
x" dx = for p ^ 0

we can now use the sandwich method of the previous section to

show that
P+I

Pupils should, however, have the opportunity to explore the

result for themselves by some more summation exercises before
meeting the following demonstration. We consider a lower step-
Figure 13.4 Difference between the upper and lower step-functions for y = x'. function which has been used to arrive at the result
108 FROM GRAPHS TO CALCULUS

P+ 1

A typical step is shown in figure 13.5. The step is of width w and

starts at a, so its height is ap. To obtain a step-function which is
helpful for the integral from 0 to t, we stretch the set of steps in the
x-direction with scale-factor t, so that they occupy the space from
x = 0 to x = t. The steps are also stretched in the y-direction so
that they still form a lower step-function for y = xp. This is shown
in figure 13.5(ii); the height of the step shown is (at)p. Thus the
width of the typical step is multiplied by t, and its height by tp, so its
area is multiplied by tp+l .
Each step-function used in obtaining a sandwich inequality for
1^1 i^x wt—> t x
xp dx
Figure 13.5
could be transformed in the same way, so
f1 tp+1
X"dx = tp+1 X"(lx = ———r . 13.6. Integration of Polynomials
o Jo P+ 1
By a simple subtraction argument we find that An integral such as
fP +i Mp+i
p+1 p+1
has to be written as .
The notation [F(.x)]J, can now be introduced as an abbreviation
forF(t)-F(u).
(3x2 + 2x) dx = 3 I x 2 dx + 2 I x dx
Hence Jo Jo Jo
before it can be evaluated. This relies on two formulae which may
x" dx = be assumed for the present. These formulae, which look very likely
P+l p+1 p+1
to be true, are
whenever p ^ 0 and p is rational.
This formula still holds when p is negative (p ^ — 1) and pupils l/(x) dx = ). /(x) dx, where A is a constant,
can verify it numerically if they wish.
THE DEFINITE INTEGRAL 109

and

) + g(x))dx = f(x)dx+ g(x)dx. (i) Y

Using these formulae, all the usual exercises involving the

calculation of areas under graphs of polynomials can be tackled.
Numerical experiment with lower and upper step-functions for
y = f(x)
integrals such as

3.x 2 dx and x)dx

will rapidly convince pupils of the truth of the formulae. If desired,

the exploration of step-functions can easily be refined to
demonstrate the two formulae, in the way shown below.
In figure 13.6, the graphs of y =/(x) and y = /.f(x) are shown 0 a
for a value of A which is about lj. In the two graphs, the
corresponding steps of the step-functions have equal widths, and
the height of each step in diagram (i) is multiplied by 1 to give the
height of the corresponding step in diagram (ii). From (i) we know
that
lower step-function area under graph upper step-function
area for/ ^ ofy=/(.x) area for/

and hence, multiplying by /*, we obtain

/lower step-functiorA /area under graphN /upper step-functiorA
?,\ ^ At ^ / (1)
\area for/ / \of.y=/(.x) / \area for / /

From (ii) we see that

lower step-function area under graph upper step-function
area for ).f ~~ of y = /./'(.x) "~ area for //

* /. is assumed to be positive. There are minor adjustments to be made if /. is

negative or zero. Figure 13.6 Step-functions fory =/(x) and y — /./(.x).
110 FROM GRAPHS TO CALCULUS

or The second formula

/lower step-function\ area under graph /upper step-function\
"\area for/ j ~~ of y = A/(.x) "" \area for/ (f(x) + g(x))dx = P f(x)dx + \ g(x)dx
Ja Ja

is demonstrated similarly. , ,
Combining inequalities (1) and (2) gives
area under graph
/lower step-function of.V=/(x) /upper step-function
\area for/ area under graph \area for f 13.7. Comments
of >• = //(*)
In the above treatment, no reference is made to differentiation,
Now for y = /"(.v), the sandwich inequality can be made as thin so that the definite integral can be studied as an extension of the
as we please by taking a step-function with thin enough steps, so it pupil's knowledge of area and sandwiching, independently of work
follows that the quantities outlined in the blocks can be made as on gradient and differentiation. In chapter 15 the link between the
close together as we please. We conclude that the middle terms of definite integral and differentiation is discussed.
the two inequalities are equal. It is suggested that this treatment establishes patterns of thought
Hence which are helpful to the future user of mathematics, whether he is a
pure mathematician, applied mathematician, scientist, engineer, or
f(x)dx= lf(x)dx. economist.
 14 Differentiation
L'M'
14.1. Introduction
14.2(ii), the same ratio measures the scale-fact or of the
TjvT
The gradient of the graph of a function at a point is the gradient enlargement from LM to L'M'. The calculations are the same as in
of the tangent to the graph at that point. Traditionally, the method the chord approach, but because the geometrical interpretation is
of calculating this has been to find the gradient of a chord PQ essentially different, we shall call this approach to differentiation
through point P, and then to find the limit of the gradient of PQ as the scale-factor approach.
Q approaches P (figure 14.1). We shall call this the chord approach In 1970, Montgomery and Jones, in Calculus and Elementary
Functions, gave the first elementary treatment of a third approach
to differentiation, that of the best linear approximation. This has, so
far, received little attention in schools, although the treatment was
intended for sixteen-year olds. The picture is again that of the
Cartesian graph, but the idea used is that among all the possible
straight lines drawn through the point P of the graph, there is one
which approximates to the curve more closely than any other. This
best straight-line approximation is the tangent at P.
In the next sections we discuss these three approaches and also
mention briefly one or two other possibilities. Whatever approach
is used, there are considerable notational difficulties, and these are
also considered.

Figure 14.1
14.2. Chords: a Numerical Approach
to differentiation. It was Newton's original method based on
velocity and was used in almost all school textbooks published It is suggested that, before the age of sixteen, the major emphasis
before 1960 which included any calculus. should be on the idea of the gradient of a graph, with careful
The SMP "A" Level course pioneered another approach, using attention to clarity of concept and wording, and many simple
the geometrical picture of a function given by the arrow diagram. numerical examples. Calculators make it possible for pupils easily
L'M' and rapidly to calculate gradients of many chords PQ (figure 14.3)
The ratio ——, in the Cartesian araph of figure 14.2(i), measures through a given point P, and to acquire a feeling for the
LM
the gradient of the chord PQ, but in the arrow diagram of figure relationship between the position of the second point Q (for fixed
112 FROM GRAPHS TO CALCULUS

Figure 14.3

gradients of chords through this point. When pupils perform these

calculations, they do not need to record as much as has been done
below. They can also easily choose their own sequences of
positions of Q, so that many numerical examples together give an
the best linear understanding of the limit process.
approximation
a bad linear x-coordinate v-coordinate Gradient of
approximation ofQ ofQ QR PR QR
~ PR

4 16 7 1 7
3-5 12-25 3-25 0-5 6-5
Figure 14.2 (i) Chord approach to differentiation; (ii) scale-factor approach, to 3-25 10-5625 1-5625 0-25 6-25
differentiation; (iii) best-linear-approximation approach to differentiation.
3-125 9-765625 0-765625 0-125 6-125
3-0625 9-37890625 0-37890625 0-0625 6-0625
P) and the gradient of PQ. When the time comes for a more formal
treatment, this numerical feeling for the gradient will help the pupil The chords in this particular sequence have been chosen so that the
to understand what is going on behind the notation of limits. x-distance of Q from P is continually halved. The sequence of
For example, the gradient of the tangent to the graph of the gradients is
function x i—> x2 at the point (3,9) can be approximated by finding 7, 6-5, 6-25, 6-125, 6-0625,...
DIFFERENTIATION 113
and it would seem that the gradient of the chord can be made as beyond P and Q, so that when P and Q are close together the chord
close as we please to 6 by taking Q close enough to P. is visible on the diagram, and can easily be related to the tangent
Another example of a sequence of chords PQ is the following, in at P.
which Q is on the left of P. Pupils should be encouraged to
experiment with a variety of different positions of Q.
14.3. Differentiation
x-coordinate v-coordinate Gradient of
ofQ ofQ QR PR PQ
2-9
In the numerical work, the same type of calculation has been
841 -0-59 5-9
2-99 8-9401 -0-0599 5-99 repeated many times. The time has now come to generalize, and it
2-999 8-994001 -0-005999 5-999 will seem natural to find the gradient of the chord of xi—>.x 2 joining
P(3,9) and Q(3 + /1, (3 +h) 2 ). Because P and Q are distinct points, h
Here again, it can be seen that we can make the gradient of the cannot be zero (see diagram opposite).
chord as close as we please to 6 by going far enough along the Now
sequence of gradients QR = (3
and
5-9, 5-99, 5-999,..., PR = h,
or by taking Q near enough to P. so the gradient of chord PQ is
It is the calculator which makes it easy for pupils to have
personal experience of the process of finding gradients of sequences QR _ (3 -h)2 -9
~l~'""~
l^R""
of chords through P. They can thus approximate to the gradient of
the tangent through P by using chords whose intersections P and
Q with the graph are very close together. Time spent in this
numerical experimentation is well spent, for soon pupils will try to
find positions of Q such that the .gradient of PQ is within for = 6 + /;.
instance 0-000001 of 6, and will realize that they cannot ever make
the gradient of the chord PQ e.\actly,6. At this stage, the wording As P and Q must be different points, we can never have h = 0, but
the gradient ofthe chord PQ can be made as close as we please to 6
the limit of the gradient of the chord as Q tends to P is 6 by taking /? close enough to 0. It is important that pupils realize
may be introduced to express the idea that although there is no chord whose gradient is exactly 6, there is
a line through P whose gradient is exactly 6, and it is this line which
the gradient ofthe chord can he made us <•/<>.«' us we pli'iise to 6 by taking 0 close is the tangent at P.
enough to P.
After this method of finding the gradient ofthe tangent at a point
When drawing diagrams such as that in figure 14.3, it is helpful with given numerical coordinates, the next step is to generalize to
to pupils' imagination to be sure that the chord is produced find a rule for the gradient ofthe tangent to a graph of a function at
114 FROM GRAPHS TO CALCULUS

X/v an arbitrary point P. The discussion given below is sufficiently

simple to be within the grasp of pupils who have done the same
thing numerically.
We shall find the gradient of the tangent to the graph of the
function xi—>2x2 + x at the point P, whose coordinates are
(a, 2a 2 + a). Since P is a given point, a is a fixed number throughout
this calculation. We take a point Q on the graph other than P. The
point Q is (a + h,2(a + h) 2 + (a + h)) where h may be positive or
negative, but is not zero.
OR
The gradient of PQ is —— , and
PR
QR
PR~
(a + h)-2a2 -a
h
2h 2 +h
a+h

The gradient of the chord can be made as close as we please to

Figure 14.4
4a + 1 by taking h close enough to 0. Hence the gradient of
the tangent at P is 4a+ 1.
When this is thoroughly grasped, the teacher may wish to
introduce the notation
lim(4a + 2/z + l)
/i-O

but this is unnecessary and can be harmful to beginners, for it

suggests to the pupil that he is putting h = 0. It seems better to
expect pupils to say to themselves
the gradient of the chord can be made as close as we please to 4o + 1 by taking h
close enough to 0, so the gradient of the tangent is 4a+ 1.
The next notational step is
Figure 14.5 If/(x) = 2x 2 + .x, then/'(o) = 4u+l.
DIFFERENTIATION 115
using f'(a) to symbolize the gradient or derivative of the function /
at the point a in just the same way that /(a) is used to symbolize the
value of the function at the point a (figure 14.5).
However, notation should not be forced upon pupils who have
not absorbed the ideas. It is quite satisfactory at this stage if they
express the gradient of a graph in words such as

the gradient of ,\ .\ at a is 4<i+ 1.

dv
The notations dx, <5v and -f- are better avoided until later, for a
ax
number of reasons:
(i) The symbols d.\ and 6y are the pupils' first experience of
using two letters to stand for a single number.
The temptation is strong to think of 8x as a multiple of x,
and 5y as the same multiple of y, so that pupils may want to Figure 14.6
(5y y fty y
write —^ = — , in the same way that — = — . This notational
dx x nx x
problem has to be faced at some time, but a time when function /', which is given by /'(x) = 2x for all x. The traditional
pupils are meeting several other new and subtle ideas is not notation
the best time for extra notational difficulties.
(ii) In the traditional notation shown in figure 14.6, the symbol for all
x is used with two quite distinct meanings, as can be seen dx
from the two occurrences of x on the horizontal axis. It is can then be introduced, but this notation has no advantages for
first the x-coordinate of an arbitrary point on the graph pupils before sixteen.
y = x 2 of the function /:xi->x 2 . Secondly, it is the In many "O" Level texts in which differentiation is treated, the
x-coordinate of a particular point P on the graph, and so aim seems to be the achievement of rote learning of the formulae
is a fixed number. It can only be used for both by an abuse of
notation which is very unsuitable until a later stage. which enable pupils to differentiate expressions such as 3x 2 — —.
This is not a suitable aim if it is done at the expense of the
This confusion is easily avoided by using a letter other than x for understanding of how gradients of graphs are found. In the first
the x-coordinate of P. Then we may say that, since/'(«) = 2a for treatment of differentiation, attention should be focused on
each point a, we have established a new function, the derived methods rather than on formulae.
116 FROM GRAPHS TO CALCULUS

14.4. The Scale-Factor Approach scale-factor over the interval PQ. The idea of local scale-factor at P
is introduced, so that
This approach is developed in SMP Revised Advanced Mathe­
matics, Book 1. The calculations are exactly as in the chord lim'- (a + h)-f(a)
approach, but the arrow diagram gives a different geometric
motivation from the gradient of a Cartesian graph. The method is is interpreted as the local scale-factor of the function at P.
now discussed briefly in language more sophisticated than would One or two difficulties occur in this interpretation. The first
be appropriate for most 16-year-old pupils. arises when a point between P and Q maps to a point which is not

4 Q'
f(a + h) Q
3
0 a+h
Q 2 2

1 1 P'
P a f(a) P
0 0
P -1

Figure 14.7 Figure 14.8

The closed interval PQ[ti,a + /i] of the domain is considered. between P' and Q'. This happens in the case of the function x \-> x..2
This interval is mapped so that P maps to P' and Q maps to Q'. The when we consider the interval [—1,2]. Because 0 maps to 0, the
signed lengths P'Q' and PQ are then calculated and found to be whole interval [—1,2] does not map to the interval [ 1,4]; it is P
f(ii + h)—f(a) and h respectively. The quotient which maps to P' and Q which maps to Q', not PQ which maps to
P'Q'. The idea of scale-factor is geometrically more obscure in this
P'Q'
« + /« )-/(«) example.
1S
The other pictorial difficulty is that the local scale-factor can
never actually be seen because it cannot be drawn. It literally
with attention paid to signs, is calculated and called the average almost vanishes as the interval PQ becomes small.
DIFFERENTIATION 117
These difficulties are greater than those which occur in the
h)-f(a)
gradient approach, in which the expression - always
represents the gradient of a chord. Even when h is very close to
zero, the gradient of the extended chord is visible in the diagram,
and the extended chord is almost indistinguishable from the
tangent.
The authors of SMP Revised Advanced Mathematics must have
been aware of these difficulties, for they introduce the gradient
picture very soon after differentiation has been defined using scale-
Fie. 7
factor ideas, and with one exception subsequent work in Revised
Advanced Mathematics I is all based on the gradient-of-chord
picture rather than the scale-factor picture. It is interesting to note that the authors do not attempt to prove
This exception is the rule for differentiating composite functions the result but make it clear that assumptions are made. This
(or the chain rule), which has a very natural interpretation using contrasts favourably with most texts, which claim to "prove" the
scale-factors but no equivalent easy picture using gradients. The chain rule, but take no account of the difficulties mentioned at the
quotation below is taken from Revised Advanced Mathematics I, end of the extract.
and is given to show the use of the scale-factor approach at the next It will be seen that the scale-factor picture forms a useful
stage of work. By this time, this text has introduced the traditional supplement to the gradient-of-chord picture, but cannot replace it.
ox and ov notation (SMP Revised Advanced Mathematics, Book 1,
p. 305). '
14.5. The Best Linear Approximation
2.3 Alternative statement of the chain rule. Using the notation shown in Figure
7, the average scale factor of the function f:x -> rover the interval of length fix
is This approach is not well known, and so is described in some
Sy fiy dti
detail. The interested reader should consult Montgomery and
Ax fin fix
Jones, Calculus and Elementary Functions (CUP) for a teaching
method, and Fraleigh, Calculus. A Linear Approach, Vol. I
To find the value of the derived function, ily ilx. we have to consider the limit as
r5x-»0; 6y/Sx-*d\-/dx, Sv/fin ~> dv du and fiu'fix -» du.dx so it would be
(Addison Wesley) for a more advanced treatment.
reasonable to conclude thai The essential idea is that of trying to find a straight line which fits
the graph of a function at a point better than any other straight
line. This line (if it exists) is defined to be the tangent to the graph at
the point (figure 14.9).
To prove the result formally, which we shall not attempt here, we have to cope
with the limit of a product and also the possibility that fiu might be zero even il
The equation of the tangent is also the equation of the best linear
fix ^ 0. (What difficulty would this create?) approximation to the function at that point. For example, when
118 FROM GRAPHS TO CALCULUS

not a tangent

tangent

Figure 14.9
the values of the function .xi—>2 + 3.x + x 2 are calculated near
x = 0, it is found that the x 2 -term makes little difference, com­
pared with 2 + 3.x.

X 2 + 3.T 2 + 3.X + .V
1 -1 0
0-1 1-7 1-71
0-001 1-997 1-997001
0 2 2
0-001 2-003 2-003001 Figure 14.10
0-1 2-3 2-31
1 5 6
It is essential that pupils should understand from numerical
Thus, for .x near 0, the error in replacing 2 + 3.x + ,x2 by 2 + 3x is
experience that near 0, x2 is small in comparison with x, so that the
small. Discussion of the interval of values of x for which the error is
terms 2 + 3x dominate. Figure 14.10 shows the geometrical
"acceptably" small is helpful. Moreover, the graph of x i-> 2 + 3.x is
significance of the three terms at the point x = /?.
a straight line, where that of x i—> 2 + 3x + x 2 is a curve. The x 2-term
The derivative of the function when x = 0 can now be calculated
can helpfully be regarded as the error term in replacing 2 + 3.x + x2
as the gradient of the best linear approximation near x = 0. Thus,
by 2 + 3x. When finding a best linear approximation near a
the derivative of x i—> 2 + 3x + x2 at x = 0 is 3.
particular point we always try to replace a non-linear function by a This approach is easily adapted when the derivative at a point
linear function such that the error is small in comparison with the other than x = 0 is required, as the following numerical example
linear function near the point. Graph drawing should convince shows. We try to find the derivative of the function
pupils that the graph of x i—> 2 + 3.x is the tangent to x i—> 2 + 3x + x 2
at (0, 2). x H-> 3 - 3.x + x 2 - x 3 at x = 2.
DIFFERENTIATION 119

Again, numerical experiment and graph drawing will be needed to

convince pupils that
(i) —5h 2 — h 3 can be made as small as we please in comparison
with h by taking h near enough to 0,
(ii) the plotted values of —7 — 1 Ih give a straight line of gradient
— 11. This straight line fits the curve extremely closely at
x = 2, and — 11 is the derivative of the function at x = 2.
2 2+h The teacher will realize that the substitution x = 2 + h has
merely moved the origin to x = 2, but pupils will not be
accustomed to seeing the equation of a straight line in the form
_v = —1—llh, where x = 2 + h.
Numerical comparisons between —1—llh and 3 — 3x + x2 — x 3
in the neighbourhood of x = 2 (or /; = 0), will be helpful in
-7 -7-11/7)
enabling pupils to realize the accuracy of the linear approximation.

a 2 + 2ah
Figure 14.11

We consider the point x = 2 + h, at which

3 - 3.x + x2 - x 3 = 3 - 3(2 + /i) + (2 + h)2 - (2 + /;) 3 a a+ h
= 3-6-3/i + 4 + 4/j + /! 2 -8-12/!-6/r-/r3
= -7-11/J-5/2 2 -/! 3 . Figure 14.12
120 FROM GRAPHS TO CALCULUS

-7-11/1 work to polynomial functions is recommended whatever approach

2 0 -7 -7 is used.
2-1 0-1 -8-1 -8-151
2-01 0-01 -7-11 -7-110501
2-001 0-001 -7-011 -7-011005
14.6. Calculus Made Easy
More generally, for the function /:.xi—>.x 2. at the point x = a,
we have (figure 14.12) In Calculus Made Easy (1910), Sylvanus P. Thompson talked
without rigour about 'differentials' and maintained that if dx is a
small enough increase, then (dx)2 is negligible in comparison with
and since h 2 can be made as small as we please in comparison with dx. He then wrote
h by taking h close enough to 0. the best linear approximation is
h H-> a2 + 2ah, which has gradient 2a. Hence /'(a) = 2a.
This way of introducing the derivative of a function at a point is
easy and intuitive for polynomial functions. Its theoretical pro­
= x2 + 2x.dx
blems lie at the next stage, when as small as we please in
comparison with h needs to be made precise. This is, however, no so
more difficult than in the traditional method, for the function / is v = 2x.dx
defined to be differentiate at a and to have derivative A at that and
point if f(a + h) can be written in the form
f(a + h) = f(a) + Ah + nh
This is not a valid argument but methods similar to these have
where the error term is written as a multiple r\h of h, and can be always been used by mathematicians and scientists as a way to find
made as small as we please in comparison with h by taking h small results which they subsequently prove by other means.
enough. For this to happen, we must be able to make j; as close to 0 In 1961, however, Abraham Robinson was able to make
as we please by taking h near enough to 0. rigorous an approach to calculus through infinitesimals.
The best-linear-approximation method is the only one which Robinson's work is known as non-standard analysis, and the first
generalizes to higher dimensions, and so is the most useful in introductory calculus book making use of the ideas of non-
advanced work. There is at present insufficient experience of using standard analysis has recently appeared: H. J. Keisler, Elementary
it at the pre-"O" Level stage to know how effective it is as an Calculus (Prindle, Weber & Schmidt, 1976). While Keisler's text is
introduction to differentiation, but certainly the calculations are clearly unsuitable for pupils under sixteen, it may be that further
extremely easy, so that they do not interfere with the compre­ work will produce greater simplification. Further development in
hension of the ideas if the work is confined to polynomial the introduction of differentiation to beginners along these lines is
functions. However, for pupils under the age of 16. confining the possible in the next few years.
DIFFERENTIATION 121
14.7. The Black-Box Technique 14.8. The Derivative of.x I-H- x"

This technique for teaching differentiation admits that differen­ For some pupils who have learnt to differentiate combinations
tiation is difficult; therefore the teacher explains to the pupils that of functions such as xi—>x2 and xi—»x3, the teacher may wish to
in the seventeenth century, Newton and Leibniz discovered that proceed as far as finding the derivative of the function with formula
for a graph of the form y = x", the gradient is given by the formula f(x) = x", where n is a positive integer.
MX"' 1 .
Care needs to be taken in deciding whether to make this
While no self-respecting teacher admits to using this method, the extension, for it is easy for pupils to lose the thread of the argument
effects of other methods presented with insufficient time and in a maze of algebraic symbols. Most texts find the derivatives of
preparation are equivalent to a black box plus hang-ups on the x i-> x 2, x i—> x3 and x i-> x4, and then draw out a general rule from
part of the student, while the black-box method as described leaves the pattern of the results. At this level this is a satisfactory
a mystery but no hang-up about it! procedure which many teachers will wish to follow.

(a + h) n
(a + hy

/ a a+h a+h

Figure 14.13 The derivative of AH-.X": (i) using Pascal's triangle: (ii) using the
best-linear-approximation method.
122 FROM GRAPHS TO CALCULUS

Although it is not possible to give a rigorous proof that the This can be made as close as we please to na" ' by taking h close
gradient of x\—>x" at x = a is na"" 1 without using proof by enough to 0. The values of A, B,... do not matter.
induction somewhere in the chain of results, it is possible to use It follows that the gradient of x i-> x" at x = a is na" ~ 1 .ln other
Pascal's Triangle to show a reason for the form of the result. words,
Pupils with skill at algebra can calculate if /(x) = x", then f'(a) = ncf~ i .
(a + h)2 = a 2 + 2 If the best-linear-approximation method is used, we consider
(a + h) 2 = (a the point (a + h), at which (a + h)" = a" + na"~ 1 h + Aa"~ 2 lr+ ....
+ ah 2 The best linear approximation is a" + na"~ l h,and this has gradient

No attempt should be made at this stage to show that the

and most of them will recognize that the coefficients form Pascal's formula still holds when n is not a positive integer. Even in the case
Triangle of a positive integer, a proof by induction will be given in the sixth
a+h 1 1 form, but the search for a pattern given above forms a suitable
(a + h)2 1 2 1 introduction.
1331
(a + h)4 14641 14.9. Recommendations
They can see from the pattern that (a + h)" starts
In considering how best to treat the introduction to differen­
a" + na"~ l h+ ..., tiation, it is necessary to consider the backgrounds of both the
but they find the formulae for later terms more difficult. However, pupils and the teacher, the time available and the directions in
formulae for these later terms are not necessary, and the calcu­ which the pupils will pursue their mathematics after the age of
lation proceeds as follows (see figure 14.13): sixteen. Among the three major approaches, through gradient of
The gradient of PQ chord, scale-factor and linear approximation, there is a good case
for able pupils to meet all three at different times, for each
(a + h)" -a" contributes to the understanding of the concept of a derivative.
However, the teacher of 16-year-old pupils has to decide which
single method he will use to introduce differentiation, in the
knowledge that other methods may be used later to deepen his
pupils' understanding.
where A, B.... are later coefficients in the expansion of (a + h)". Notational problems also loom large. It has been suggested
A and B have been used in order to avoid having to work out
formulae. Thus, the gradient of PQ is above that <5x, 6y and — should not be used at this stage, because
ax
na'"' l + Aa"~ 2 h + . of the confusion between ideas inherent in them. However, if
DIFFERENTIATION 123
students' next encounter with calculus is to be in science or
dy It seems clear that for most teachers, with most pupils, the
economics, they are very likely to be expected to use —, and
ax gradient-of-chord approach will continue to be used (the scale-
should be aware of this notation and how it arises. Moreover, the factor picture being helpful for composite functions) but there is a
strong case for experiment with the best-linear-approximation
6x, dy, — notation has stood the test of 300 years' use, and many
ax approach. Innovative teachers may also like to consider non-
mathematicians think intuitively in terms of it; for instance the standard analysis which, it is claimed by its American proponents,
has improved understanding and performance in calculus at both
arc-length formula —= /l+l—-I is remembered from school and college level.
dx \ \dx
 15 Linking the Definite Integral with Differentiation
We have defined the definite integral as an area under a curve. 15.1. Which Comes First?
f . .
Thus /(x) dx is the shaded area shown in figure 15.1. In this book, chapter 13 on the definite integral comes before the
chapter on differentiation (chapter 14). In most school courses
differentiation is tackled first and integration is subsequently
introduced as the process of anti-differentiation. The treatment of
integration given in chapter 13 does not depend on differentiation
so, in order to give emphasis to this, it was put before differen­
tiation. The fact that integration and differentiation are inde­
pendent in the initial stages is crucial, and the authors believe that
it is of great importance that they are taught independently and
y = f(x)
subsequently brought together by the fundamental theorem of the
calculus which is discussed in this chapter.
This leaves open the question of whether integration should be
taught before differentiation or not. The usual order in school is to
teach differentiation first, but there are powerful arguments in
favour of integration first. Some of these arguments for teaching
the definite integral first are now discussed.
(i) As a general principle it is likely that when one thing is
discovered before another it is simpler; more than a thousand
years before anyone knew the methods of differentiation,
Figure 15.1 The definite inteeral as an area under a curve. Archimedes was finding definite integrals, as is shown in
chapter 17.
(ii) The integral is a development of the idea of area, which has
been well known to pupils since their primary-school days. Their
On the other hand, differentiation is linked to the gradient of a first statistical graphs were made by using small squares to make a
curve, a local scale-factor, or a best linear approximation. block graph or a histogram. Thus most pupils have been familiar
The purpose of this chapter is to provide a link between the two with the concept of area for many years. On the other hand, at the
concepts of the definite integral and the derivative which appear on age of sixteen, the idea of a rate is very recent to most pupils.
the surface to be completely unrelated. Although a qualitative impression of steepness is plain in a graph,
124
LINKING THE DEFINITE INTEGRAL WITH DIFFERENTIATION 125

the measurement of gradient is by means of a ratio, and the

understanding of ratio and proportion is found difficult by many
pupils. Moreover, although pupils have drawn tangents to a graph
y = x2 -3x+2
by eye, and have calculated approximations to the gradient, formal
differentiation is a much more complicated process.
(iii) An important mathematical argument for teaching the
definite integral before differentiation is that it prevents integration
being regarded first and foremost as the reverse of differentiation.
Students can then see the fundamental theorem of calculus which
links integration and differentiation for what it really is, a massive
and surprising bridge linking the apparently independent
mathematical structures of integration and differentiation.
(iv) Another advantage of teaching integration first is that it
capitalizes on the certainty in the mind of the pupil that there really
is a number which measures area. This is guaranteed by the Figure 15.2
sandwich theorem for the real numbers, which we have assumed,
and which is intuitively obvious to pupils. The limit idea needed functions, whereas the definite integral is a number. We look for a
for finding the gradient of a tangent from the gradient of a chord function associated with an integral.
is conceptually much harder.
We consider integrals such as~ f 3.x 2 c/.xand f (x 2 — 3x + 2)dx
(v) It is interesting that some attempts have been made to write
books which present integration before differentiation, the most in which the lower limit is fixed and the upper limit may take any
notable example being the first edition of Modern Mathematics for value. These integrals can be called integral functions, and it is
Schools. It is unfortunate that, in the second edition, the authors necessary to ensure that pupils realize that these integrals depend
have reverted to putting differentiation first just at the time that the on the value of I, and that .x is a "dummy variable", which is only
calculator eases the computation problems associated with the needed to tell us what graph we are finding the area under. Thus
definite integral.
However, in the second edition of Modern Mathematics for 3x 2 dx is the shaded area under y = 3x 2 in figure 15.2(i), and
Schools, the authors do carefully re-emphasize the independence
of integration and differentiation. (.x 2 — 3x + 2)dx is the shaded area under v' = .x 2 —3.x + 2 in

15.2. The Link between Integration and Differentiation fiszure 15.2(ii) with sign taken into account. It is clear that 3.x 2 d\

To develop the link between integration and differentiation we and (x 2 —3x + 2}dx depend on the position of the right-hand
need to appreciate that differentiation is a process applied to
126 FROM GRAPHS TO CALCULUS

boundary and hence on t, not x. This can be reinforced by working 15.3. Further Discussion of the Link between Integration and
out the integrals Differentiation
The reader may be worried by the fact that the work of the
previous section relies on a statement which was assumed earlier
and on the basis of numerical evidence only. This statement is

3x.2 (P/-1)

However, a link between derivatives and definite integrals may

It may strike pupils as a curious coincidence that the derivatives
be made in another way, using three results from our knowledge of
of the functions A and B given by
displacement-time and velocity-time graphs and the definition of
an integral.
A(t) = 3.x 2 dx = f3 When a displacement-time graph is drawn for the motion of a
Jo
and particle, the velocity at each instant is the gradient of the graph at
thatinstant (section 12.2). This is illustrated for time t = Tin figure
B(t)=
are
and B'(t) = t2 -3t + 2.
By drawing pictures of some integral functions such as A and B, gradient
finding their formulae and differentiating, pupils can be led to
suggest the result that if a is a fixed number and

A(t)= f(x)dx
Ja
then
A'(t) =f(t).

Pupils will not realize that this result needs/to be continuous at

f, but the time to discuss this point is in the sixth form, when a more
formal proof can be given. However, even at this stage the work Figure 15.3
can go slightly beyond the feeling of surprise that integration and
differentiation are linked, which the above algebraic work sug­ Pupils can plot the velocity-time graph for the motion of the
gests. same particle, by working out the gradient of the displacement-
LINKING THE DEFINITE INTEGRAL WITH DIFFERENTIATION 127
time graph at a number of points. In fact, at each time T,
v(T) = s'(T) y

where s'(T) is the derivative of s at time T. Hence v and s' are the
same function of time.
Now it is also known (section 11.4) that the area under the
velocity-time graph between f = a and f = b represents the
displacement during that time; in symbols gradient f (X)

area under velocity-time graph = s(b) — s(a).

The definite integral of a function is the area under its graph, so

area under velocity-time graph = P v(t)dt

Ji
or
s(b)-s(a)= v(t)dt
Ja

= p S'(t)dt
Ja

from the previous paragraph, using t as a place-holder instead

ofr.
Thus, for displacement and time, the formula

s(h)-s(a)= s'(t)dt
Ja

indicates that the definite integral between u and b of velocity, or

the derivative of displacement, is the change in the displacement
between a and b.
In order to get a more general result, we only have to move from
s and r to y and .v. This is easily done for a differentiable function in
the following way.
Suppose the function /:.YH->/'(.Y) (figure 15.4) has a derivative at
each point. Then certainly we can imagine a particle moving Figure 15.4
128 FROM GRAPHS TO CALCULUS

along a straight line in such a way that its displacement from the This argument is suitable for able pupils who are sufficiently
origin at time x is given by f(x). Its velocity at time x will be/'(.x), familiar with the relation between displacement-time and velocity-
and by an exact repetition of the previous argument we obtain time graphs to argue from them. Proofs which do not depend on
this fact are suitably left for the sixth form.
f(b)-f(a) = \"f'(x)dx. For the majority of pupils who meet the definite integral before
Ja the age of 16, the numerical verification of
The final step in the argument is made by considering a function
/such that x"dx =
f'(x) = x", where p ^ -I.
Experiment easily convinces pupils that the formula for f(x) must carries complete conviction. If they regard the two facts
be of the form
x"dx=\ '-
/(-x) = +c
(i)
p+l
P+l
Axp+1 \. r
where c is any real number. This immediately yields (ii) the derivative of —r ls x (P^-ll
VP+V
1 «" +1
x"clx =
p+1 p+1 as a curious coincidence, no harm will be done, for the first was
b studied for more than a thousand years before its connection with
the second was discovered, as is shown in the final historical
chapter of the book. More careful discussion of the connection
Thus the link between the definite integral, defined as an area between the two can appropriately wait for the next stage of the
under a graph, and antidifferentiation is established. student's education.
 16 Some Applications of Integration and Differentiation
The preceding two chapters were concerned with the calculation
of areas and gradients, and neglected the applications which arise
from integration and differentiation. In this chapter we discuss
some examples of applications, while recognizing that teachers
would want to introduce some of them during the teaching of
velocity in
integration and differentiation. metres/second

16.1. Displacement, Velocity and Acceleration

A particularly important example of the use of integration and

differentiation occurs in the study of motion under gravity. It is time in seconds
found experimentally that when a heavy body falls in a vacuum Figure 16.1 Velocity-time graph for a falling stone.
under gravity, it falls with a constant acceleration. In real life, air
resistance retards the fall, but the constant-acceleration model is seconds is given by
very simple, and is often used. On the earth's surface the constant
\0tdt
acceleration due to gravity is nearly 10m/s 2 ; that is, velocity
increases by nearly 10 metres per second in each second.
It was seen in section 12.3 that the gradient of a velocity-time = 5t
graph represents the rate of change of velocity with respect to time; = 5T 2 .
this is called the acceleration. Thus, if a stone is allowed to fall from
the top of a cliff, and the constant-acceleration model is used, the Thus, after T seconds, the stone has fallen 5T 2 metres, and has a
velocity-time graph is extremely simple, and has a constant velocity of lOTm/s.
gradient of 10. Thus, the equation of the velocity-time graph is In general, the velocity-time graph contains complete infor­
mation about the motion of a body, for its gradient gives the
v = lOf acceleration at an instant, and the area under it represents the
if we assume that the stone falls with zero initial velocity. displacement up to that instant (figure 16.2).
Moreover, we also know that the area under the velocity-time A second example is that in which a ball is thrown upwards with
graph represents the displacement, so that the displacement in T initial velocity 5 m/'s. Here the acceleration is downwards, so the
129
130 FROM GRAPHS TO CALCULUS

r
Figure 16.2

gradient of the velocity-time graph is— 10m/s 2 (figure 16.3(i)).

After \ second, the ball changes from positive upward velocity
to negative upward velocity, and so starts to move down again.
The equation of the velocity-time graph is
v = 5-\0t.
i\
The upward displacement after T seconds is
fr
Figure 16.3 (i) Velocity-time graph; (ii) displacement-time graph, for a ball
s= (5-\Ot)dt thrown upwards.
Jo
= 5T-5T 2
16.2. Graph Sketching
= 5T(1-T)
and the ball returns to the thrower's hand after 1 second. The Pupils first become familiar with the shapes of the graphs of
displacement-time graph is shown in figure 16.3(ii). Its point of well-known functions by numerical plotting of points. At that
zero gradient occurs when T = \, and occurs at the instant at stage every graph is a new experience, its shape unknown until a
which the velocity is zero. number of points have been plotted.
SOME APPLICATIONS OF INTEGRATION AND DIFFERENTIATION 131

It is an important stage in the development of the vocabulary of

mathematics when pupils become explicitly aware of and can
anticipate the shapes of the graphs of some basic functions.
It was suggested in chapter 6 that pupils by the age of sixteen
should know (in numerical examples) that
(i) the graph of v = mx + c is a straight line with gradient m,
(ii) the graph of _v = ax 2 + bx + c is a parabola.
They should also have plotted v = —, and have had some
.\
(0,4)
experience of plotting cubics.
For many purposes, however, it is not necessary to plot an
accurate graph; skill at drawing a quick sketch which shows the
major features of the graph of a function is useful when a graphical
model is used, and should be beginning to develop by age sixteen.
Knowledge of how to calculate the gradient of a function from
its equation is a considerable help in graph-sketching, and this
knowledge should be used as soon as it is available.
Some examples will help to make the situation clear.

1. Sketch the graph of the function f: x i—> x2 — 6.x+ 4. Figure 16.4 Sketch of the graph of/: xi-> x 2 -6x +4.

(i) Pupils will expect the graph of this function to be a

parabola whose vertex points either upwards or down­
wards. Large positive or negative values of x give rise to Thus, the vertex of the parabola is at (3, —5).
large positive values of/(x), so the vertex of the parabola (ii) Sometimes, the points at which j (x) = 0 are easily plotted.
points down. It remains to locate the vertex. However, in this case,/'(x) = 0 when
The point x = 0, y = 4 is easily plotted as a control
point. When x = a, the gradient of the graph is 2a - 6. This x 2 -6x+4 = 0
indicates that the only point where the graph has zero and the roots of the equation are irrational,
gradient is given by (iii) The sketch of the graph shown in figure 16.4 contains
2a-6 = enough information for most purposes. Further con­
or firmation can be obtained by noticing that the gradient,
a = 3. 2a - 6, is positive when a > 3 and negative when a < 3.
132 FROM GRAPHS TO CALCULUS

(0,0)
(3,0)

(2-4)

Figure 16.6 Sketch of the graph of/: .v t-> x3 - 3.x 2 .

(v) (vi)
(iii) or (v); numerically large negative values of x confirm this.
Substitution also shows that when x = 0,/(x) = 0, so (0,0) lies on
the graph. The gradient at x = a is/'(a) = 3a2 —6a and it can be
Figure 16.5 Shapes of the graphs of cubic functions. seen that this is zero when a = 0 or 2; thus, the graph has two
points of zero gradient, at (0,0) and (2, —4), and so is of type (i). In
2. Sketch the graph of the function f: x i—> x 3 — 3x 2
this case, it is easy to see what happens when /(x) = 0, and so
where the graph meets the x-axis. At these points,
Previous experience of plotting graphs of cubic functions should x 3 -3x2 =0
lead pupils to expect a graph of one of the shapes shown in figure
16.5. The only problem is to decide which of these it is and to place and so x = 0 or x = 3. Thus the graph meets the x-axis at (0,0) and
it correctly with respect to the coordinate axes. Numerical (3,0). The sketch in figure 16.6 shows all the main features of the
experiment quickly satisfies pupils that when x is large (positive or graph.
negative), the term x3 is more important than the term —3.x2 , so The teacher may like to introduce vocabulary to describe the
that the sign of/(x) is the same as the sign of x3 . When x is large behaviour of the graph. The function is increasing when x < 0 and
and positive x 3 is large and positive. This restricts the shape to (i), when x > 2; it is decreasing when 0 < x < 2. Pupils should
SOME APPLICATIONS OF INTEGRATION AND DIFFERENTIATION 133

recognize that increasing functions have positive gradients, and At this stage, it is quite inappropriate for pupils to know rules,
decreasing functions have negative gradients. They know that the involving the second derivative, to distinguish between maxima,
gradient when x = a is minima and points of inflexion. A sketch will nearly always suffice,
and if any doubt remains, working out the gradient when x is just
to the right or just to the left of the point of zero gradient will dispel
It is easily verified that, for instance, all doubts.
when a = — \,f'(a) = 9;
when a = 1, f'(a) = -3;
when a = 3, f'(a) = 9. 3. Sketch the graph of the function f: x \—> 3.x 5 — 20x3 .
The point of zero gradient at (0, 0) is a local maximum ; the point of This example is probably too advanced for all but the most-able
zero gradient at (2, —4) is a local minimum. pupils, but illustrates how much information about a graph points
Points of zero gradient are always either local maxima or of zero gradient can give.
minima, or horizontal points of inflexion, as shown in figure 16.7.
(i) 3x 5 is the more important term, so when x is large and
positive, /(x) is large and positive; when x is large and
negative,/(x) is large and negative.
(ii) The gradient at x = a is given by
/»= 15a4 -60a2.
Thus the points of zero gradient are at

15a4 -60a 2 = 0
or
15a 2 (a 2 -4) =

Hence a = 0, a = 2 and a = ~ 2 give points of zero gradient.

These points are (0,0), (2, -64) and (-2,64). The infor­
mation now known is shown in figure 16.8(i), and it is easy to
complete the sketch, as in figure 16.8(ii).
If doubt remains, it. is easy to check on the calculator that, for
Figure 16.7 Points of zero gradient: (i) local maximum; (ii) local minimum; (iii)
instance, the gradient at a = 0-5 is about — 14, as is the gradient at
horizontal points of inflexion.
a= -0-5.
134 FROM GRAPHS TO CALCULUS

(-2,64)

(0,0)

(2,-64)

Figure 16.8 Sketch of the graph of/:X t-> 3x 5 -20.v3 .

in the domain of the function, but when x is close to 1, y is

4. Sketch the graph ofx i numerically large. Its sign can be found by saying that if x is
x-1
greater than 1, x — 1 is positive, and if x is less than 1, x — 1 is
Although pupils will not be able to differentiate the function
negative.
• l/(x — 1), they can draw a sketch showing the salient features of (iii) The information gained so far is shown in figure 16.9(i), and
of the graph. it is easy to complete the sketch as shown in figure 16.9(ii).
Confirmation is gained by noting that the graph does not
(i) For large positive values ofx, the value of——- is very small
x— 1 cross the x-axis, since —— = 0 does not have solutions,
but positive. For large negative values of x, the value of x— 1
—— is very small and negative. The point (0, — 1) lies on The teacher will notice that graph-sketching from clues such as
x— 1 those obtained here also relies implicitly on knowledge of the
the curve. This information is plotted immediately. continuity of the functions sketched. Many pupils are not bold
enough with sketching graphs at this age. They tend to plot more
(ii) Since ——
x— 1 is not denned when x = 1, the point x= lisnot points than they need and to be frightened to look for clues in the
SOME APPLICATIONS OF INTEGRATION AND DIFFERENTIATION 135

(0.-1)

Figure 16.9 Sketch of the graph of/:.\i-> l/(.v-I).

form of the equation and in the form of the derivative (if they can these problems by careful graph-drawing earlier. Now. instead,
calculate it). they can sketch the graph and find the important points, which are
usually points of zero gradient, by exact calculation.
1. How should a loop of string of length 24cm be arranged on a
16.3. Design Problems peg-board to enclose a rectangle of maximum area? This problem
will have been familiar to many pupils since their primary-school
Some of the most compelling and dramatic glimpses of the days. They can now solve it by calculation, by finding the point of
power of calculus come, for 16-year-old pupils, when they see how zero gradient on the graph of
differentiation can solve problems of design. A=x(\2-x)
There are many simple beautiful examples, some of which are
discussed below. Pupils may well have met and solved some of = 12.x-A- 2.
136 FROM GRAPHS TO CALCULUS

A/1 an existing stone wall for the fourth side. How does he arrange the
pen to get the maximum area?
4. A famous problem is that of the open cardboard box which is
made from a square sheet of cardboard by cutting out the shaded
squares at the corners, bending upwards along the dotted lines and
then gluing at the edges. How should the squares be cut to obtain a
t box of maximum volume?
A cm 2 (12-x) cm
This problem brings out further teaching points. Suppose the
i cardboard sheet is a square of side 100 cm and that a square of side
x cm is cut from each corner. The box is shown in figure 16.12 and it
Figure 16.10 xcm xcm
Kcm
They will now expect the graph of this function to be a parabola,
with its vertex upwards, and should note that the function
A = 12.x-x2 (100-2x)cm

only models the situation for 0 < x ^ 12.

2. The long rectangular strip in figure 16.11 has its sides bent up xcm
to form a gutter; where should the bends be made so that the gutter
channels the maximum possible water? xcm
Figure 16.12

is clear that its volume, y cm3 , is given by the function

y = .x(100-2x)2 = 10000x-400x 2 + 4x3 .
The equation models the situation only for 0 ^ x < 50, and pupils
should be explicitly aware of this, for it helps them to visualize the
graph of the function. Clearly, the volume of the box is zero if x = 0
or x = 50. It seems as if the graph will be rather like that of figure
Figure 16.11 16.13(i). However, the function is a cubic, and figure 16.13(i) gives a
poor idea of what actually happens. Moreover, pupils may expect
3. The farmer has a fixed length of fencing and he wishes to the maximum to occur when x = 25.
make a rectangular sheep pen using his fencing for three sides and The gradient at x = ais/'(a) = 10000-800a+ 12a2 , and this is
SOME APPLICATIONS OF INTEGRATION AND DIFFERENTIATION 137

50 50

Figure 16.13

zero when a = " and a = 50. Thus the graph is that shown in and it is required to find the general form of the graph of the
figure 16.13(ii), where the parts of the graph which do not model function. One example of this, which has already been examined
the physical situation are shown by dotted lines. from a different viewpoint in section 6.6, is growth; a situation in
The box has a maximum volume of about 0-07 m 3 ; the base of which at each instant the amount of material present increases at a
this box is a square of side about 67cm, and its height is about rate proportional to the total amount of material there at that
17 cm: a surprising shape to give the maximum volume. instant, that is, the rate of change of material is proportional to the
This kind of argument based on curve sketching and common total amount of material. This gives rise to the equation
sense is all that is necessary for design problems at this level. Rules f'(t) = //(f).
to distinguish maxima from minima are not yet needed, and should
not yet be introduced. For instance, in the problems just given, it is For example, the birthrate is given as 13 live births per 1000 of
surely unnecessary to verify that x = 4p gives a local maximum and population, and the interest rate for investment of money is given
x = 50 gives a local minimum for it is quite clear that when x = 50, as a percentage of the amount invested.
the volume of the box is zero. It is useful and constructive, both for future work and for a
proper understanding of gradient, for pupils to study a step-by-
step approach for finding such approximations to graphs whose
16.4. Step-by-Step Integration; Orienteering gradients are known at each point.
The method is illustrated by the following example which is not
Many situations exist in real life which lead to mathematical related to growth. A graph passes through the point (0,0) and has
models in which the gradient of a function is known at each point, the property that its gradient at each point is equal to the x-
138 FROM GRAPHS TO CALCULUS

gradient 0-1

C(0-3,0-03)
A^jy___^6(0-2,0-01)
0-1 0-1 (0,0) 0-1 A 0-1 0-1

0-5

0-4

0-3

02

0-1

0-1 0-2 0-3 0-4 0-5 0-6 0-7 0-8 0-9 1-0

Figure 16.14 Step-by-step solution of the equation/'^) = x.

SOME APPLICATIONS, OF INTEGRATION AND DIFFERENTIATION 139

0-5

0-4

03

02

0-1

0-1 02 0-3 0-4 0-5 0-6 0-7 0-8 0-9 1-0

Figure 16.15 Graph of v = j.x 2 .

coordinate at that point. An approximation to the shape of this reached. What instruction should it be? Since the gradient at that
graph can be stepped out by using an idea from orienteering. point is known to be the same as its x-coordinate, the instruction is
Imagine you are orienteering, that you are at (0,0), and that you to proceed with a gradient of 0-1 until further orders. These will be
must walk in the direction given by a first set of instructions, and given on reaching the point with x-coordinate 0-2; that is. at
keep walking in that direction until you receive another in­ B. The step in the y-direction from A to B is 0-01, so B has
struction. At (0,0) the x-coordinate is 0, so the gradient is 0, and the coordinate (0-2,0-01).
instructions are to walk along a line with gradient 0 in the direction The new instruction will be to travel with gradient 0-2 until
of the positive x-axis. reaching a point C with x-coordinate 0-3. The step in the y-
The next instruction may come after walking a distance 0-1 km direction from B to C is 0-02, so C has coordinates (0-3,0-03).
or on reaching the line x — 0-1 or according to some other rule. Pupils can proceed in this way using a mixture of drawing on graph
Suppose the next instruction is given when the point (0-1,0) is paper with a large scale and calculation until they see that they
140 / FROM GRAPHS TO CALCULUS

eventually reach the point (1,0-00 + 0-01+0-02+ ...+0-09) or

2-6
(1,0-45).
Clearly this method is approximate and the errors will "grow" in
the sense that the further you go the further away from the true
2-4 solution you will be. Nevertheless, the series of segments of straight
lines have a fair resemblance to the curve y — \x2 (see figure 16.15).
The approximation can be improved by taking shorter steps,
2-2 e.g. by receiving instructions at x-intervals of 0-01 instead of 0-1.
If this is done, then the point reached when x = 1 is the point
(1, 0-0001+0-0002 + ... +0-0099), or the point (1, 0-495).
2-0 In the sixth form, students will be able to investigate what
happens if they receive fresh instructions n times between x = 0
and x = 1, and allow n to become indefinitely large.
We turn now to the growth function, and consider the parti­
cularly simple case where the graph starts at t = 0, y = 1, and the
constant of proportionality X takes the value 1, so that
16 f'(t) =f(t).
Thus, the gradient or rate of growth is always equal to the
y-coordinate.
1-4 We continue the orienteering analogy, receiving instructions
whenever we have moved a distance of 0-1 in the t-direction. The
first point reached is A(0-l, 1-1) and the fresh instruction is to
1-2 proceed with gradient 1-1. This leads to 8(0-2,1-21) and the
instruction to proceed with gradient 1-21. Working to two decimal
places, which is all that can be expected using a graphical method,
1-0 we eventually reach the point (1-0, 2-59) and the y-coordinate is
beginning to increase in value for each step quite quickly.
Pupils will see that the values they obtain at each point are (0,1),
(0-1,1-1), (0-2,1-1 2 ),..., (0-9,1 -I 9 ) (1 -0,1-1 10 ). The calculator shows
that l-l 10 is approximately 2-59. They might guess that with steps
of 0-01, the value at x = 1 would be (l-Ol) 100 ^ 2-70.
0 0-2 04 0-6 08 10 / This method of step-by-step solution of differential equations is
Figure 16.16 Step-by-step solution of the equation/'(0 =/(')• a very general one, and adaptations of it are used when computers
SOME APPLICATIONS OF INTEGRATION AND DIFFERENTIATION 141

are available and the equation of the curve cannot be evaluated

directly. For the pupils we are considering, the beauty of the
method lies first in its simplicity, and secondly in that it con­
centrates the pupils' attention on the meaning of gradient.

16.5. Volumes of Revolution

Many pupils are told, on the basis of very little evidence, that the
volume of a circular cone is jnr 2h, or one-third of the volume of a
cylinder of the same height and radius.
For pupils who have studied the definite integral, the volume of
a cone may be taken as the first example of the volume of a solid of
revolution. The line through the origin with gradient - is rotated
round this x-axis, and sweeps out a circular cone of radius r and
height/; (figure 16.17(ii)).
This section concentrates on establishing that when the graph of
y =f(x) shown in figure 16.18 is rotated around the x-axis the
volume of revolution generated is given by

ny2 dx or n{f(x)} 2 dx.

The method used is to recognize that the volume of revolution

generated between x = a and x = b is equal to the area under the
graph of y = n{f(x)} 2 between x = a and x = b.
The x-axis between x = a and x = ft is divided into n equal parts
each of width w. Each part gives rise to a disc with a curved surface,
and a set of inner and outer cylindrical discs is constructed so that a
sandwich argument can be used (figure 16.19). We see that the
rotation of each strip in figure 16.19 gives
volume of inner disc ^ volume of revolution < volume of outer disc.
An enlarged view of one of these elements with its inner and Figure 16.17 The volume of a circular cone.
142 FROM GRAPHS TO CALCULUS

Y = f (x)

Figure 16.18

N y = f(x) y = f(x)

/ \

volume of revolution inner cylindrical discs outer cylindrical discs

Figure 16.19 Volume of revolution by a sandwiching method.

SOME APPLICATIONS OF INTEGRATION AND DIFFERENTIATION 143

M
• y = f(x) H
m

= F(x)=n{f(x)} 2

\
view along the x-axis

Figure 16.20 Figure 16.21

outer cylindrical discs is shown in figure 16.20. The thickness of But h = Tim 2 and H = nM 2, since F(x) = 7t(/(x)} 2 . Thus
each disc is w, while the radii of the inner and outer cylindrical discs
hw ^ volume of revolution of y =/(x) ^ Hw.
are m and M respectively. Hence
nnrw ^ volume of revolution of y = f(x) ^ nM 2 \v. Now we can compare two sandwich inequalities
But this can be interpreted differently. We consider the curve hw = f- \olume of revolution of Hw =
lower step-function y =/l.v) upper step-function
y = F(x) where F(x) = n[f(x)} 2 and calculate the area under area for F(x) < ;irea under y = /-"(.v| area for F(x)
V = F(x) between x = a and x = b. Dividing the interval between
x = a and x = b into n equal parts as before, we can set up a and add the similar inequalities for all the discs between x = a and
correspondence between a strip of the area under v = F(x) (figure x = b. When this is done we reach the inequality
16.21) and the disc of the volume of revolution of v = /(x) shown in
figure 16.20. If the greatest and the least heights are H and h \olumc of solid of revolution of
lower step-function v = /(.v) between x = a and .v = /) upper step-function
respectively, the sandwich inequality for area is area for F(x) • <: area under y = /-'(.v) between < | area for F(x)
hw < area under v = F(.v) ^ Hw. __ __ __I x — a and x = /'
144 FROM GRAPHS TO CALCULUS

This is a familiar situation. We know that the sandwich inequality We can now return to finding the volume of a circular cone. The
can be made as thin as we please by choosing w small enough. . <-,,.• , , • , • r , <. r
_ .. . . , -11, • • i equation ol the function whose araph is rotated is v = /(x) = -x.
From this we deduce that the two middle expressions in the h
sandwich are equal. Thus Hence the volume of the cone is
volume of solid of revolution of area under v = F(.v) between _ I 1' /"'' fr } 2 jj,- 2 P/i
Y = /'(x) between .x = a and .x = h ~ x = a and .x = b ~ X' 7I<-X> C/X = ~rr \ -x' 2 dx
Jo (h j /1 2 J 0
ButF(x) = 7t{/(x)} 2 , so = !^!r^

volume of solid of revolution generated

by v = /'(.x) between x = « and .v = ft = \nr2 h.
 17 The Early Historical Development of Calculus R.L.E.Schwarzenberger
The teaching of differentiation and integration in Britain has helpful mathematical context is that of graphs and functions
suffered through an unfortunate historical accident. Patriotic pride adopted in this book. The historical context is closely related, but
in the achievements of Newton, ignorance of continental develop­ leads to slightly different insights. It clarifies the physical problems
ments, and distrust of pure mathematics all contributed to a view from which the ideas of differentiation and integration arose, and
of "the Calculus" as an isolated, awe-inspiring, impossibly difficult so may give a guide to possible topics for use as examples or as
and almost magical event. motivation in the classroom today. It makes clear that integration
The material in this book makes it clear that, on the contrary, arose from much simpler problems than differentiation, and at a
differentiation and integration arose in a context of other work and very much earlier date, and so raises the question whether
to serve particular purposes. This is also apparent from the very integration is not more "natural" and more suitable for teaching at
explicit titles of the monographs of Newton himself: On Analysis an early stage. It shows the enormous conceptual simplification
by means of Equations with an Infinite Number of Terms 1 , Methods and unification which came about through the systematic use of
of Fluxions and Infinite Series2 , On Quadrature of Curves3 . On the coordinates and graphs: a good example of the way in which
Continent the successors of Leibniz improved and simplified the mathematics progresses by taking advantage of simpler notation,
subject: it became more common to use ordinary words (for as well as by creating new theories.
example Differentialrechnung and calcul differentiel meaning differ­
entiation, Integralrechnung and calcul integral meaning in­
17.1. The Origins of Graphical Methods
tegration) of the same kind as the words used to describe other
methods of calculation, which had the further advantage of The modern habit of plotting graphs is so familiar that it is now
making clear that differentiation and integration are different hard to realize that it is relatively recent. The plotting of data by
methods which came about in different ways and for different chemists and physicists, whether as a smooth curve or by isolated
purposes. values or by block graphs, is difficult to trace back beyond the
It is therefore particularly necessary in Britain to teach differen­ eighteenth century.
tiation and integration within a context in which they arise The solution of geometrical problems by the use of algebraic
naturally, and not as an isolated abstruse phenomenon with a equations involving coordinates dates from Descartes4 and Fer-
mystical Latin title. The range of current applications of differen­ mat 5 in the early seventeenth century: to the extent that the
tiation and integration gives a clear indication that the most systematic use of coordinates is the essential feature of modern
graphical work, they must be credited with its discovery.
1 De Analvsi per Aeiiuatioiws Numcro Teniuiionuii Iiijiiiitim. 1711 (\vrillen c.
1669).
2 Methotliix Fliixionwn el Sericruni In/inilanim. 1736 (written r. 1671). * La Geometric. 1637.
3 De Quadratura Ciin-cirum, 1704 (written c. 1676). 5 Ail Locus Pianos el Solitlos Isuyoye. 1679 (written c. 1629).

145
146 FROM GRAPHS TO CALCULUS

Figure 17.1 Graphs in a manuscript of Orcsme of 1428 (British Library, Sloane,

MS 2156). The graphs represent a "quality" (given by height above the
"subject line" (//>) which is decreasing steadily. The "uniform quality", or constant.
which has the same area under the graph, is given by the horizontal line fy.

Vwi? Jua. feptSplt- OSfa-

But the representation of quantities by lengths and areas on
geometrical diagrams has a longer history. In the fourteenth
- y u4&xst**t£ Jiffia century, Nicole Oresme6 represented time and velocity by lengths,
and distance travelled by area, on diagrams not dissimilar from
those in chapter 11. In the eleventh century, Omar Khayyam 7 —
better known today for his collection of light verse—used in­
tersecting conies to construct a length representing the root of a
rtt&SiA. •&< -t& fiuf: cubic equation. Both Oresme and Khayyam were not merely
mathematicians: they made contributions to economics and to
-»i. Knoorifftd astronomy respectively. Similarly the writings of the Greek
y •—1 geometers on astronomy, geography, mechanics and optics show
S
«t» clearly that for them geometry was closely linked to physical
.S- -Xtfit •*••
applications: the length and area of a geometrical diagram could
-f A _- - £ be used to represent physical quantities.
Here, even if coordinates are not used systematically, is the
origin of modern graphical work and also the origin of calculus. Its
•8- ut p? development can be seen most clearly in the work of the Greek
geometers on area, and in the problems studied by the Pytha­
gorean school 2500 years ago.

17.2. Pythagoras
•tof

T- <&>if-m9-fSt ttzfft The study of area under the Pythagorean school followed the
<**&, vfixzt-' StffSxii t&ttt utn
ufig -ft» etu'j&iZ'- e^.
method familiar in primary schools of dissecting a region into parts
and reassembling them to form a known area. The rules for doing

6 De Uniformitate et Difformitate Intensionum, c. 1350.

1 Algebra, c. 1080.
THE EARLY HISTORICAL DEVELOPMENT OF CALCULUS 147

Figure 17.2

so were expressed in "axioms" such as "if equals be added to equals, The procedure, of deducing non-trivial results about area from
the wholes are equal" and "the whole is greater than the parts". very elementary assumptions, is precisely that used in "measure
Once the area of rectangles is taken for granted, other areas can be theory" today. If these methods are combined with the notion of
determined without further assumptions. For example, the dissec­ similar triangles, there are further consequences. For example, if r
tion shown in figure 17.2 gives the area of a triangle as ^base is the radius (distance from centre to any vertex) of a regular
x height, and in fact yields three equivalent expressions for this polygon with N sides then the length a of each edge, and the
area (taking different sides as base). The consequences of this distance f from the centre to the mid-point of each edge, are each
equivalence are non-trivial. For example, it is a short step to the proportional to r by the properties of similar triangles. It follows
fact (equivalent to Pythagoras' theorem) that, if ABC is a right- that the area of a regular polygon with N sides is proportional to
angled triangle, then the two areas shaded in figure 17.3 (where the square of its radius.
AB = AD and AC = AE) are equal.

Figure 17.4

The most significant steps towards the calculus, from such

elementary calculations, were those attributed by later Greek
mathematicians to Eudoxus. It is tragic that no edition of his work
survives, but it is possible to guess much of the content from the
Figure 17.3 description given by Euclid.
148 FROM GRAPHS TO CALCULUS

17.3. Eudoxus

It is sufficient for the present purpose to illustrate the power and

precision of Eudoxus by explaining in modern notation one
particular theorem: the fact that a circle of radius r has area
proportional to r2 . This theorem also illustrates the way in which
the successive attempts by mathematicians from Euclid onwards
to make the results of Eudoxus understandable to the reader have
done so at the price of introducing vague concepts which have
harmed mathematical education ever since, and which are still
used by many teachers at all levels. It would, for example, be
asserted that "Polygons inscrib'd in a Circle infinitely at last end in
the Circle", and hence that the area of a circle is proportional to the
square of the radius by an immediate deduction from the Figure 17.5
corresponding fact already established for polygons. The assertion
and the deduction are a common feature of editions of Euclid, but
neither makes much sense to the thoughtful pupil: both fully polygon for a given value of N. Thus A 2N — A N is the area of the
deserve the denunciation made by Berkeley 8 when he asked additional triangles needed to form a regular polygon with 2JV
Whether men may properly be said to proceed in a scientific method, without sides from a regular polygon with N sides, while A - A2N is the area
clearly conceiving the object they are conversant about, the end proposed, and of the segments of the circle which remain outside the larger
the method by which it is pursued? ... Whether the same things which are now polygon. A dissection (figure 17.6) shows that
done by infinities may not be done by finite quantities? And whether this would
not be a great relief to the imaginations and understanding of mathematical

since each side of the inequality can be represented by the sum of N

None of these criticisms apply to Eudoxus who was, as far as one regions (one for each side of the polygon) of the kind shaded in
can tell, careful to avoid the notions of "infinite", "approximation" figure 17.6. Thus the area left outside the polygon is more than
and "exhaustion" which later less-clear-thinking generations halved each time the number of sides is doubled (to see this in
adopted. His work depends on the relationship between the areas practice work out the areas of polygons with 4, 8, ] 6,... etc., sides
A, A N and A 2N of the circle, /V-sided polygon and 2/V-sided using a calculator with a square root—no trigonometry is needed).
But the heart of Eudoxus' method is not in proceeding to alleged
8 The Analyst, or A Discourse Addressed lo an Infidel Mathematician. Wherein it limits; it is in the more straightforward theorem which states that
is examined whether the Object, Principles, and Inferences of the Modern Analysis
are more distinctly conceived, or more evidently deduced, than Religious Mysteries If B < A there is an integer N such that B < A N .
and Points of Faith. "First cast out the beam out of thine own Eye: and then shall
thou see clearly to cast out the mote out of they brother's Eye". 1734. In more colloquial language, given any area B less than that of the
THE EARLY HISTORICAL DEVELOPMENT OF CALCULUS 149

The same combination of rigour and straightforwardness can be

seen even more clearly in the subsequent work of Archimedes,
since his own commentary on the method has survived.

17.4. Archimedes

A good example of the work of Archimedes is his determination

of the area of the segment of a parabola. In the diagram, the
segment in question is bounded by the chord RS and T is the point
A-A2N

Figure 17.6

circle, there are inscribed regular polygons with area larger than B.
It is this theorem which can now be used to prove that the area of a
circle is proportional to the square of its radius.
The argument is as follows. Recall that we already know the
result for a regular polygon with N sides, so that if two circles have
areas A, A' and radii r, ?•', then the areas of the corresponding
polygons satisfy

where k is the ratio r/r'. Now compare k 2 A' with A. If k z A' < A,
then by the theorem established above there is an integer N such
that k2 A' < A N ; but this implies k 2 A' < k 2 A'N in contradiction to
the fact that A'N < A'. Again, if A < k 2 A' there is an integer N such
that A < k 2 A'N because the same theorem can be applied to the Figure 17.7
circle of radius ?•'; but this implies A < A N in contradiction to the
fact that A N < A. Since both k 2 A' < A and A < k 2 A' are false, the with tangent parallel to RS. The problem is to relate the area A of
only possibility remainingis theequality k2 A' = A which was to be the segment to the area A 0 of the triangle RST. Archimedes proves
proved. that A = |/4 0 in the following four stages.
Notice the completely rigorous nature of this proof, and also the (i) A centre-of-gravity argument applied to strips parallel to the
absence of spurious appeals to "infinites" or "approximations". axis of the parabola suggests f A 0 as a likely guess for the value of A.
150 FROM GRAPHS TO CALCULUS

Figure 17.8 Figure 17.9

Archimedes' comment is "the fact is not demonstrated by the (iii) Comparing the area A with that of the parallelogram with
argument used ; but that argument has given a sort of indication side RS we see that A 0 < A < 2A 0. The same argument applied to
that the conclusion is true. Seeing then that the theorem is not each stage of the preceding construction shows that
demonstrated, but at the same time suspecting that the conclusion
is true, we shall have recourse to geometrical demonstration . . .".
(ii) A geometrical argument shows that the two triangles
(iv) Since An = —A 0 is obtained by successive quartering, it is
formed by adding points T, T" with tangents parallel to RT, ST
respectively have a combined area A l = %A 0 . The same con­ true that if fy4 0 < A there is a value of n for which f/! 0 + f/l n < A,
struction performed n times yields a polygon with area while if A < %A 0 there is a value of n for which A < f/4 0 — p4,,-
Either case contradicts the inequality obtained above; the only
— possibility remaining is the equality f/4 0 = A which was to be
proved.
This argument is that of Archimedes subject only to the use of
But I 1 +- + ...+— 1(1-71=1-7X7- and therefore the
\ 4 4"/\ 4/ 44" modern notation. The four stages are typical of the method used in
polygon has area fA 0 — -j/4n. • •.. . • many other problems:
THE EARLY HISTORICAL DEVELOPMENT OF CALCULUS 151

(i) to guess the answer, often by the use of physical argument; particular definitions, and moreover the methods of Eudoxus and
(ii) to establish a geometrical result which ensures that a Archimedes were available and could easily be adapted to the new
known area or volume can be built up from blocks of coordinate geometry.
diminishing size; Three examples must suffice; all are not merely of historical
(iii) to prove that the required area or volume is bounded above interest but could also provide teaching material for pupils at the
and below by known areas or volume; corresponding stage of development. The first is the work of
(iv) to show that the area or volume is neither more nor less Cavalieri" on the area under the curve v = xp. In modern notation
than that guessed initially. his result would be written
The great merit of the method is the absence of "infinites" and
"approximations". Its great defect is that each problem must be x"dx =
attacked separately and may require ingenious geometrical argu­
ments. When Greek mathematics began once again to be studied but in fact he appears first to have noted that the integral is
and understood in the fourteenth, fifteenth and sixteenth centuries, proportional to ap+1 . Today we would prove this fact using the
the stumbling block to further progress was the lack of any general method of substitution: if a' = ta then
method which would be applicable to a wide range of problems
and which would contain as special cases the results of Eudoxus x"dx = x"dx = (ty)"tdy
o Jo Jo
and Archimedes. The discovery of such a method had to await the
systematic use of coordinates in graphical work due to Descartes where x = ty. Thus
and, when it came, the method lacked the straightforward
y>dy
precision characteristic of Eudoxus and Archimedes.

and therefore, writing I (a) for xp dx, we obtain the required

17.5. The Early Seventeenth Century Jo
"homogeneity property"
The turmoil of the first half of the seventeenth century was /(«') = tp+ l l(a).
characterized in mathematics by much speculation on problems
arising from graphical work and from physical applications: the However, this fact can be understood in a more elementary
correct definition of the tangent to a curve at a given point, of fashion. If the region between x = 0 and x = a is cut into strips of
maxima and minima of given functions, of the length of a curve, of width w, then the region between x = 0 and x = ta can be cut into
the relationship between velocity and distance, and so on. But the same number of strips of width iw (figure 17.10). The height
initially the most substantial progress was made on questions ofeach strip is now multiplied by a factor t r. The required result is
concerning area and volumes. There are two natural reasons for now immediately plausible. The method of Eudoxus would then
this: there was no problem of finding correct definitions, since
results on area depend only on very simple axioms and not on the 9 Cenluria de Varii ProMemii. 1639.
152 FROM GRAPHS TO CALCULUS

and confusing notion of "indivisibles". Once it is known that the

integral has the form cp ap +1 , where cp is a constant depending on p,
then it is easy to form recurrence relations which determine cp in
terms of c l5 c2,..., cp _ l . The pupil who is asked to do this will learn
a lot in the process, but should be told that Cavalieri himself only
'y= got up to p = 9.
Example: To find c 2 knowing c, = j. The homogeneity argument implies

8c, = ! x 2 dx = x 2 dx = x 2 dx + X x = C^ (x + 1 )2 dx

0 ta But l) 2 dx = c2 + 2c, + 1 = c 2 + 2. Therefore 8c2 = c2 + c 2 +2. We con­

Figure 17.10 clude that c 2 = j (see figure 17.11).
The second example is the work of Fermat 10 on the same
problem. Essentially he used steps of width w where a = Nw
obtaining the inequality

NP+l v- ' ••• ' v" -/ /

This is an efficient "sandwich" if, but only if, there is available an

= 2c,
algebraic formula for the sum l p + 2p + ... +NP in terms of N.
1 -
Example: Thecasep = 2. The formula ^(WH- l)(2N + l)for the sum of the first
N squares implies

6N 3 (N-\)N(2N-l)< o x 2 dx< 6/V

"
3
Figure 17.11
a a a a3 a3 a3
x dx < — + — + ——
o 3 2/V 6/V 2
serve to establish it rigorously, as would the "sandwich" method and it is easy to see that the integral can be neither more or less than %a3 .
which can be found in chapter 13, but Cavalieri appears not to
have argued in this way: instead he introduced the unfortunate 10 Varia Opera Mathematica, 1679 (written c. 1630).
THE EARLY HISTORICAL DEVELOPMENT OF CALCULUS 153

It is significant that in all these examples it is not necessary to

y= know anything about area beyond the axioms used by the Greeks.
The results can all be established by the method of Eudoxus, that is
to say without introducing "infinites", "approximations" or (a
term which made its appearance in the work of Gregory of St.
Vincent) "exhaustion". However, it was by the use of such
language that mathematicians searched for a general method
(5w)i
which would reduce the calculation of areas to a more mechanical
and less ingenious process.

(4w)f>

17.6. Newton and Leibniz

The long-sought general method came from an unexpected

quarter. It was not an algorithm for integration, but a highly
wp efficient machine for doing the opposite. It was undoubtedly the
study of velocity-time graphs which led Newton to the realization
0 w 2w 3w 4iv 5w 6w that the ordinates of a graph give a direct measure of the rate of
Figure 17.12 change, or gradient, of the area under the graph. His results on this
question yielded a general method for determining gradients. Once
the basic properties of differentiation had been established, it
became evident that Newton and Leibniz had created a tool which
The third example concerns the area under the graph v = A"
and is attributed to was astonishingly easy to use. The derivatives of all kinds of
Gregory of St. Vincent 1 '. The same homogeneity argument, used by Cavalier! for functions could be obtained by a unified method; whenever, by
other powers of x, implies chance, there appeared among the list of derivatives a known
function, then areas under the graph of the latter function could be
determined.
With the use of coordinates it became evident that much of the
As Gregory's pupil Sarasa 12 observed in his own work, the areas behave like
logarithms. Even today, this is probably the most satisfactory way to introduce
work of Eudoxus and Archimedes consisted of different instances
natural logarithms. of the integration of functions like .v —> \ and x H-».v 2 , and that this
itself was merely the consequence of the differentiation of functions
1 ' Opus Geometrician Quadratuni Circuit el Sectiimiim Coni. 1647. like .\-H-».v 2 and .vt—>.v\ The way was open to the calculation of
12 Solutio Problematis a Mersenno Prupoxiii. 1649. more complicated areas and volumes, to an understanding of the
154 FROM GRAPHS TO CALCULUS

length of curves and the calculus of variations and to the solution definite. Inevitably "the Calculus" appeared impossibly abstruse
of differential equations. and the harm described at the beginning of this essay was done. It
With hindsight we can see that the achievement of Newton and was only by a gradual return to more geometrical methods, and by
Leibniz was built upon a vast body of results established by the a rejection of infinitesimals in favour of a more sensible under­
mathematicians of the early seventeenth century: not merely those standing of real numbers and of limits, that the balance was
mentioned above by way of example but also such men as Kepler, gradually restored and the methods of differentiation and in­
Roberval, Wallis, Barrow and Gregory. On the other hand, the tegration could take their place among the ordinary concepts of
dramatic power of the technique of differentiation, and the extreme mathematics taught at an elementary level.
difficulty which was experienced in justifying and in explaining the At a more advanced level the return to more geometric methods
concepts and techniques involved in its creation, combined to set has provided a proper basis for the calculus of functions of several
differentiation (and hence also its opposite, integration) apart as variables. Integration, partial differentiation and calculus of
"the Calculus". The result was a tension between two extremes. On variations all now arise from a fusion of the analytic methods of
the one hand were the geometric problems from which calculus calculus and the geometric methods of linear algebra. The right
arose, especially those concerning area and volume; the concepts balance must be achieved between algebraic manipulations and
were clear, and could be applied to prove satisfying general results geometric concepts, for both are necessary if the subject is to
or to justify results guessed by other methods, but any particular progress healthily.
numerical determination of area or volume required either an
ingenious special argument or the restatement of the problem in 17.7. Conclusions
more algebraic terms followed by recourse to results from
differentiation. What conclusions does the historical development of calculus
On the other hand, there was the supremely efficient algebraic suggest for the teacher of graphical work today? Those who
tool of differentiation which was very difficult to understand and believe that historical development is a good guide to psychologi­
which appeared mystical or even magical when attempts were cal development in individuals might find it useful to chart the
made to explain how or why it worked. former in terms of the dichotomy between "relational" and
So powerful was the spell cast by this algebraic tool that the "instrumental" understanding popularized by Skemp 13 . The sum
origins of calculus in the calculation of areas were almost total of man's knowledge of calculus might be displayed as a trail
forgotten. Differentiation took over from integration, and mathe­ traced on a diagram, in which fast progress was made in the three
maticians, instead of viewing differentiation as a method for periods 400-200 BC, AD 1600-1700, AD 1850-1950 and very slow
finding gradients, began to view it as the subject matter of calculus progress in the periods between. Perhaps there is a case for
par excellence. building up the knowledge of the individual pupil by much the
They even began to introduce a so-called "indefinite in­ same route? If so, three conclusions suggest themselves.
tegration" as the opposite of differentiation, so that the pupil was Firstly, calculus has its roots in geometrical problems so that it is
faced first with the concept of differentiation whose definition was
most obscure, and then with two sorts of integration, indefinite and 13 The Psycholayv of Learning Mathematics, Pelican 1971.
THE EARLY HISTORICAL DEVELOPMENT OF CALCULUS 155
relational Secondly, the concepts associated with calculus are an essential
understanding
part of graphical work and may be expected to enter at an early
-500 stage. The first such concepts are likely to arise from the estimation
\ of area and volume, and these will be accepted naturally by quite
\
-400 + 1950
young pupils. Area is important both for its own sake and to
/
geometric calculate other quantities.
/ \
/
Thirdly, understanding of gradients and derivatives is likely to
area, / 1 slope, rate
be achieved only by pupils at age fifteen or sixteen. Perhaps this is
volume I i of change because ratio is a much more sophisticated concept than area. It
-200 + 1850 suggests that differentiation should be deferred until after in­
\ tegration. It should arise in the first instance as a set of rules for
\
algebraic calculation of gradients which turns out, surprisingly, to be a
\
\ successful aid for evaluating definite integrals. One advantage of
+ 1600 + 1700 this approach is that it stresses that differentiation is an exact
\ process rather than some kind of approximation.
- +1650- - If these conclusions suggested by the history of the development
instrumental of calculus were to be adopted, pupils would see calculus as a
understanding natural development of the ideas of area and gradient. They would
Figure 17.13 Historical development of man's knowledge of calculus. have seen successive stages of the function of the concepts of area
and gradient, through quantitative pictorial ideas to exact calcu­
essential for good understanding to use geometric diagrams ana lations, and would be led by this route to a greater understanding
not merely to rely on algebraic symbols. of calculus.
 SUGGESTIONS FOR FURTHER READING

1. Apostal, T. M., Calculus I and II, Wiley, 1976.

2. Flegg, G. and Meetham, R. (editors), An Introduction to
Calculus and Algebra—Volume I: Background to Calculus,
Volume II: Calculus Applied, Open University Press, 1971.
3. Maxwell, E. A., Analvtical Calculus, Volumes /, //, /// and
IV, CUP.
4. Montgomery, R. M. N. and Jones, T. A., Calculus and
Elementary Functions, CUP, 1970.
5. Spivak, Michael, Calculus, Benjamin, 1967.
6. Wheeler, David, R is for Real, Open University Press, 1974.
7. Williams, E. M. and Shuard, H. B., Primary Mathematics
Today, second edition, Longman, 1976.

156
 Index
'Titles of books and journals are given in italics
abuse of notation, 38 chord approach to differentiation, 111-5 direct proportion, 39
acceleration, 91, 99, 129 chord, gradient of, 97, 111-2 directed numbers, 65
Achilles and the tortoise, 82-3 circular cone, volume of, 144 displacement, 60, 87-95, 129
analysis, non-standard, 120 codomain of a function, 18 represented by an area, 146
area, 50-4, 87-95 composite function, differentiation of, 117 displacement-time graph, 55-6, 126
negative, 94 compound interest, 48 as a model, 56
of circle, 54 cone, circular, volume of, 144 distance, 59. Also see displacement
of irregular figure, 51-4 continuous graph, 15, 16 domain of a function, 18
of regular polygon, 147 convergence of a sequence, 73, 84 Durell, C. V., 19, 33-4
under a graph, 87-95, 103 coordinates, 7
under velocity-time graph, 87-90 cosine function, 43
under v = x", 105, 151 cubic function, 132 equation of a graph, 28
Archimedes, 124, 149-51 curve, gradient of, 98-9 Eudoxus, 147-9
arrow diagram, 7, 8, 22, 24 stitching, 73 exponential functions, 46-8
average rate, 57 tangent to, 96
average speed, 55
average velocity, 60, 90-1, 97 Fermat, 145, 152
axes of graph, labelling of, 27, 29 decimal, 62-8, 83-4 Fibonacci sequence, 85
non-terminating, 83 fractions, 62
recurring, 63-6, 68, 84 on the number-line, 63
ball thrown upwards, 130 terminating, 63, 65 function, 1, 15, 17-38
Berkeley. Bishop, 148 decimal number, 62-6, 67-8 as a relation, 32
best-fitting straight line, tangent as, 96 decreasing function, 132 codomain of, 18
best linear approximation to a graph, 117-8, 120 decreasing sequence, 84 composite, 117
black-box technique, 121 definite integral, 103-10 cosine, 43
block graph, 1, 6 denary system, 62 cubic, 132
dependence, functional, 17, 21, 33 decreasing, 132
derivative, 115, 121 derived, 115
calculator, use of, 48, 49, 54,61,69, 78-80, 104-5, 112-3 derived function, 115 domain of. 18
rounding by, 78 Descartes, R., 145 exponential, 46
Calculus and Elementary Functions, 111, 117 differential equation, step-by-step solution of. 75, 140 formula for, 19
calculus, fundamental theorem of, 125-8, 154 differentiation, 111-23, 124-8, 129-36, 145, 153 4 "function machine", 26
Calculus Made Easy, 120 chord approach to, 111-5 growth function, 140
Cartesian graph, 1, 4, 6, 23-7 limits and. 72 increasing, 132
Cavalieri, 151-3 linear approximation approach to. 117 20, integral, 125
change, rate of, 58, 99, 100 link with integration. 124 8 linear, 10, 17,39. 131
Channon. McLeish Smith and Head. See New General problems of notation in. 122 notation for, 37
Mathematics scale-factor approach to. 116-7 quadratic, 42-3

157
158 INDEX

range of, 35 integral, definite, 103-10 negative velocity, 94

rule for, 18, 19,32 indefinite, 154 nested intervals, 67-8
sine, 43 of increasing function, 107 New General Mathematics, 20, 34
step-function. 12, 15. 16, 104 of polynomial, 108-10. See also integration Newton, 111, 145. 153
functional dependence. 17. 21, 33 integral function, 125 Newton's method for square root, 78
fundamental theorem of calculus, 125-8, 154 integration, 129-44 non-standard analysis, 120
step-by-step, 137 non-terminating decimals, 83
interpolation from a graph, 66 notation, abuse of, 38
geometric series, sum of, 80. 81-2 intervals, nested. 67-68 for functions, 37
gradient, 6. 55-60, 96-102 irrational number, 68-70 problems of in differentiation, 122
as a rate, 57-8 Nuffield Mathematics Project, 1-5
of chord, 97, 111-2 Nuffield Physics, 9, 21
of curve, 98-9 labelling of axes of graph. 27. 29 number, 61-71
of graph of a function, 111 Leibniz 145, 153 decimal, 62-6, 67-8, 83
of road, 59 limit, 72-86. 113, 114 directed, 65
of straight line; 56-9 integer, 65
and differentiation, 72 :" •
of tangent. 97, 99, 102, 113 irrational, 67-8
at a point, 86
graph. 1-16. 39-49 rational, 66, 68
at infinity, 86
area under, 87-95. 103 of a sequence, 81, 82 real, 61, 68, 69
Cartesian, 1,4,6,23-27 number-line, 1, 61, 62
linear approximation to a graph, 96, 117-8, 120
continuous, 15, 16 approach to differentiation through, 117-20
displacement-time, 55-6, 126 linear function, 10, 17, 39, 131
equation of, 28 Omar Khayyam, 146
link between differentiation and integration, 124-8
gradient of. 111 local maximum or minimum, 133 ordered pair, 4, 7
interpolation from, 66 local scale-factor, 116 Oresme, No., 146
labelling axes of. 27-31 logarithm, 49
of squares, 14
straight line, 10, 39, 131 parabola, 11,131
graph sketching, 130-1, 133-5, 145 point, limit at, 86
Gregory of St. Vincent, 153 machine, function, 26 polynomial, integral of, 108-10
growth function, 140 Mansfield, Thompson and Bruckheimer. See proportion, direct, 39^10
Mathematics: a New Approach inverse, 40
mapping, 7, 20, 34-7 pursuit curves, 74
histogram, 1. 6, 87 confusion over, 34—5, 37 Pythagoras, 146
Holland and Rees. See Maths Today Mathematical Association, 37, 38
horizontal point of inflexion, 133 mathematical model, 10-16
Mathematics: a New Approach, 34-5 quadratic function, 42-3
Maths Today, 64
increasing function. 132 maximization, problems of, 136
integral of, 107 maximum or minimum, local, 133 range of a function, 35
increasing sequence, 84 model, mathematical, 10-16 rate, 57-8, 99-100, 124
indefinite integration, 154 Modern Mathematics for Schools, 28, 35-6, 58, 70. and gradient, 57-8
infinity, limit at, 86 96-7, 125 average, 57
inflexion, horizontal point of, 133 Montgomery and Jones. See Calculus and Elementary of change, 58, 99, 100
integer. 65 Functions rational number. 66. 68
INDEX 159

real number, 61, 68, 69 School Mathematics Project, Revised Advanced straight-line graph, 10. 39, 131
reciprocal, 79 Mathematics, 116-7 gradient of. 56-9
reciprocal function, 41 Scottish Mathematics Group. See Modern
recurrence relation, 78-80 Mathematics for Schools tangent to a curve, 96-99, 117-8, 120
recurring decimal. 63-6, 68, 84 sequence, 72, 77 as best linear approximation. 117-8, 120
relation, 32-4 convergence of, 73, 84 gradient of, 97,99, 102, 113
function as 32 decreasing. 84 terminating decimal, 63, 65
recurrence, 78-80 Fibonacci, 85 Thompson, Sylvanus, P., 120
root of a number, 79 given by recurrence relation, 78 transformations of graphs, 42, 45
rule for a function, 18, 19, 32 increasing, 84 trapezium rule, 95
limit of. 81-2
series, geometric, sum of, 80
sine function, 43 units, 89
sandwich inequality, 51-2, 54, 65, 66-8, 70, 77-8, 85, Skemp. R. R., 154
91-2, 103, 106 snowflake curve, 75 velocity, 60. 87-102. 129. 146
scale-factor. 111, 116-7 speed, 55-60, 91 actual, 98-9
approach to differentiation. 111, 116-7 average, 55. See also velocity and speed, 60, 91
average, 116 square root. 61, 80 at an instant, 97, 99
local, 116 Newton's method for, 78 average. 60, 90-1
Science, 5/13, 8-9 squares, graph of, 14 negative, 94
School Mathematics Project, Books A-H and A'-Z. step-by-step method, of integration, 137 velocity-time graph, area under, 87-90
6,13,29,30,35,50-1,87-8 of solution of differential equation, 75, 140 volume, 54, 141-4
School Mathematics Project, Books 1-5, 34 step-function, 12. 15, 16, 104 of revolution, 141-4
•<*:
The Mathematics Curriculum
A Schools Council Project

The mathematics teacher today faces a daunting

array of literature and apparatus which makes well-
informed decisions about syllabuses, methods of
organisation and presentation, books and other
learning materials very difficult. This series provides
a range of surveys of the mathematical topics central
to the curriculum generally taught to 11 to 16 year-
olds, thereby helping teachers to make critical
appraisals in the context of their own needs and
situations. School mathematics departments will find
the series invaluable, as will those involved in the
initial and in-service training of teachers.

The presentation and critical interpretation of data in

the form of graphs are skills of considerable importance
to every school leaver. Starting from the ideas of
graphical representation taught in the primary school,
this book illuminates the stages in the development of
the subject, culminating in possible ways of intro­
ducing integration and differentiation. ISBN 021690341 6

BLACKIE Glasgow and London