(Serge Lang) MATH Encounters With High School Students

Serge Lang
MATH!
Encounters with High School Students
With 103 Illustrations
Springer-Verlag
NewYork Berlin Heidelberg Tokyo
Serge Lang
Department of Mathematics
Yale University
New Haven, CT 06520
U.S.A.
AMS Classifications: OO-A05, OO-A06

Cover photo courtesy of Carol MacPherson.
Library of Congress Cataloging in Publication Data

Lang, Serge
Math! : encounters with high school students.
1. Mathematics-Addresses, essays, lectures.
1. Title.
QA7.L284 1985
510
85-13837
A French version was published as: Serge Lang, des Jeunes et des Maths, Belin, 1984.
CD 1985 by Springer- Verlag New York Inc.
Ail rights reserved. No part of this book may be translated or reproduced in
any form without written permission from Springer-Verlag, 175 Fifth Avenue,
New York, New York 10010, U.S.A.
Typeset by House of Equations, Inc., Newton, New Jersey.

Printed and bound by Halliday Lithograph, West Hanover, Massachusetts.
Printed in the United States of America.
9 8 7 6 5 4 3 2 1
ISBN 0-387-96129-1 Springer-Verlag New York Berlin Heidelberg Tokyo
ISBN 3-540-96129-1 Springer-Verlag Berlin Heidelberg New York Tokyo
Acknowledgemen t
1 want to thank the teachers who made it possible for me to meet their
classes: William Bisset, Patricia Chwat, Peter Edwards, Marie-Therese
Giacomo, Michel Ricart. 1 know that there are sorne bad teachers, but
there are also sorne good ones, and the students with whom 1 did
mathematics would certainly not have reacted as they did if their teachers
had been bad. 1 thank Abe Shenitzer, who induced me to give the talks in
Canada, and organized them. 1 also thank Jean Brette, who directs the
Mathematics Section of the Palais de la Dcouverte in Paris, for the contacts which arose through him with high school teachers in Paris, and also
for his interest in the whole pedagogical enterprise. 1 also appreciated
Stephane Brette's interest, and his willingness to take part in a mathematical dialogue with me, after one of the talks. 1 thank Patrick Huet who
video-taped for the Paris IREM two of the talks, and the discussion reproduced at the end of this book. Finally 1 thank Carol MacPherson for the
photograph on the cover.
SERGE LANG
Contents
Who is Serge Lang?
lX
Dear Christopher, Rachel, Sylvain, Yaelie,

and ali the others
Xl
What is pi?
Volumes in higher dimension
28
The volume of the bali
52
The length of the circle
77
The area of the sphere
91
Pythagorean triples
95
Infinities
Postscript
Il 0
124
Who is Serge Lang?
Serge Lang was born in Paris in 1927. He went to school until the lOth
grade in the suburbs of Paris, where he lived. Then he moved to the
United States. He did two years of high school in California, then entered
the California Institute of Technology (Caltech), from which he graduated
in 1946. After a year and a half in the American army, he went to Princeton in the Philosophy Department where he spent a year. He then
switched to mathematics, also at Princeton, and received his PhD in 1951.
He taught at the university and spent a year at the Institute for Advanced
Study, which is also in Princeton.
Then he got into more regular positions: Instructor at the University of
Chicago, 1953-1955; Professor at Columbia University, 1955-1970. In
between, he spent a year as a Fulbright scholar in Paris in 1958.
He left Columbia in 1970. He was Visiting Professor at Princeton in
1970-1971, and Harvard in 1971-1972. Sin ce 1972 he bas been a professorat Yale.
Besides math, he mostly likes music. During different periods of his
life, he played the piano and the lute.
From 1966 to 1969, Serge Lang was politically and socially active, during a period when the United States faced numerous problems which
affected the universities very deeply.
He has also been concerned with the problems of financing the universities, and of their intellectual freedom, threatened by political and
bureaucratie interference. As he says, such problems are invariant under
ism transformations: socialism, communism, capitalism, or any other ism
in the ology.
However, his principal interest has always been for mathematics. He
has published 28 books and more than 60 research articles. He received
the Cole Prize in the U.S. and Prix Carriere in France.
Dear Christopher, Rachel, Sylvain,

Yaelle, and all the others
I am writing this to you because I came to schools like yours, in France as

weil as in Canada, to talk mathematics with students who could have
been your own friends, and who th us contributed to a joint enterprise.
I wanted to show them beautiful mathematics, at the level of your
class, but conceived the way a mathematician does it. In most school
books, the tapies are usually treated in a way which I find incoherent.
They pile up one little thing on another, without rhyme or reason. They
accumulate technical details endlessly, without showing the great tines of
thought in which technique can be inserted, so that it becomes both
appealing and meaningful. They don't show the great mathematical lines,
similar to musical lines in a great piece of music. And it's a great pity,
because to do mathematics is a lively and beautiful activity.
This book is made up of severa! lectures, or rather dialogues, which
have been transcribed as faithfully as possible from the tapes, to preserve
their li vely style. I t gave me great pleasure to have this ki nd of exchange
with ali the students, in different classes. The subjects concern geometrie
and algebraic tapies, understandable at the ninth and tenth grade level. I
even gave one of the talks to an eighth grade class! If students at those
levels could understand and enjoy the mathematics involved, so can you.
Each dialogue is self-contained, so you don't have to read this book
continuously from beginning to end. Each topic forms a single unit, which
you can enjoy independently of the others. If, while reading any one of
them, you find the going too rough, don't let that put you off. Keep on
reading, skip those passages that don't sink in right away, and you will
probably find something later in the lecture which is easier and more
accessible to you. If you are still interested, come back to those passages
which gave you trouble. You will be surprised how often, after sleeping
on it, something which appeared hard suddenly becomes easy. Just
browse through the book, pick and choose, and mostly get your mind to
function.
A lot of the curriculum of elementary and high schools is very dry.
You may never have had the chance to see what beautiful mathematics
is like. 1 hope that if you are a high school student, you will be able to
complement whatever math course you are taking by reading through this
x11
DEAR CHRISTOPHER, RACHEL, SYLVAIN, YAELLE, AND ALL THE OTHERS
book.* 1 have many objections to the high school curriculum. Perhaps the
main one is the incoherence of what is done there, the lack of sweep, the
little exercises that don't mean anything. You will find something quite
different here, which 1 hope will inspire you. And by doing mathematics,
you might end up by liking math as you like music, or as 1 like it.
SERGE LANG
* The
first five talks would fit weil in a geometry course, and the la~t two in an algebra
course. See also the book which 1 wrote in collaboration with Gene Murrow: Geometry, published by Springer-Verlag.
What is pi?
The following talk was given at a high school in the suburbs of

Toronto, April 1982, to a class of students about 15 years old. The ta/k
lasted about 1 hour and 15 minutes.
SERGE LANG. My name is Serge Lang, 1 usually teach at Yale, but
today 1 came here to do mathematics with you.

We are going to study the area of sorne simple geometrie abjects, like
rectangles, triangles, and circles which you must have heard about in this
course. Let's start with rectangles. We assume that its area is the product
of the base times the height, so if the sides have lengths a and b, then the
area of the rectangle is ab. For instance, if a rectangle has sides of lengths
3 cm and 4 cm, then its area is 12 cm 2. You can check this on the figure.
1
1
1
1
1
1
r - - - - - L - - - - _l_ - - - - -1- - - - -
3cm
1
1
1
1
---- -j----
1
1
1
-~---- -~-----
1
1
1
1
1
4cm
Area = 12 cm2
Area =ba
If y ou eut a rectangle in half, like this:
WHAT IS PI?
then you get a right triangle, so the area of the right triangle is one half
the product of the base times the height. We can write
a rea of righ t triangle
J_bh
2
where b is the base and h is the height.
1
A rea=- bh
2
You should also know that this formula is true for any triangle, if h is the
perpendicular height. 1 can show you this on two possible figures:
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
:h
1
1
1
1
~----------'--- ------ _d
b
1
Area = -bh
Area=
21 bh
Try to prove the formula yourself, because 1 want to have time to discuss
something more interesting, the circle. 1 So we assume you know about the
area of a triangle. Now you have the circle radius r.
1 1 give the proof for the first figure. Drop the perpendicular height from one vertex to the
opposite side as shown on the next figure.
WHAT IS PI?
Do you know what is the formula for its area?

A STUDENT. lt's pi
r squared.
SERGE LANG. That's right, it's wr 2 . Weil, what is pi?

A STUDENT. What is pi?
SERGE LANG. Yes.
THE STUDENT. 3.14
SERGE LANG. You claim that w is 3.14. ls that an exact expression?
THE STUDENT. No, 1 don't think so.
SERGE LANG. Then why did you say 3.14?
STUDENT. Weil, it goes on and on.
SERGE LANG. OK. So we put dot dot dot like this, 3.14 ... to signify
that it goes on and on and on. How do you know how it goes on?
[The students rea ct variously.]

SERGE LANG. lt's not so clear! So it's a problem how it goes on. 1
mean, how are you going to compute it?

A STUDENT. You measure the perimeter of the circle and divide by
twice the radius.

SERGE LANG. Ah, now you're telling me something else. lnstead of
the perimeter, let me cali it the circumference. Do you mind if 1 cali it the
circumference?
STUDENT. No.
Then the triangle i~ decomposed into two right triangles, whose base~ are b and b 2 such that
h + b2 = h. The two right triangle~ have the same height h. Then using the formula for the
area of a right triangle, we now get:
1
This proves the formula for the first figure. Treat the second figure similarly. You will need
a subtraction instead of an addition.
WHAT IS PI?
SERGE LANG. OK. So first you told me the area is '!l'r 2 , and now you
mention the circumference. We call the circumference c. And what did

you just tell me? You made the assertion that
c == 2'll'r.
That's what you said: 2r is the diameter, so the circumference is pi times

the diameter, so we can also write
c == '!l'd,
where d is the diameter, d == 2r. But now look. You have two formulas,
for the area and for the circumference:
and
2'll'r.
By the way, what's your name?

STUDENT. Serge.
SERGE LANG. Oh Serge, just like me! [Laughter.] Serge said, to com-
pute 'Il', you look at the circumference and divide by the diameter. The circumference is something you can measure. You can get a soft tape at
home, you put it around a frying pan, and you measure the circumference. Then you measure the diameter with a ruler, and divide. Actually
you can get one or two decimais accuracy out of that, probably you can
get two decimais if you are careful. You get sorne sort of value, which is
an approximation for 'Il'.
You would have a much harder time trying to measure the area to get
an approximate value for '!1'.
Now the question is: you've got these two formulas, one for the area,
one for the circumference. How do you know these formulas are true?
STUDENTS. [Silence, questioning looks.]
SERGE LANG. How do you prove them? Has anybody ever broached
the problem of proving these formulas? At any time? You were just given
the formulas.
STUDENTS. [Negative looks on most faces, one or two raise their hands.]
SERGE. You can just say that
'Il'
equals the circumference divided by
the diameter, and work it out.

SERGE LANG. Work what out? You just repeated one of the two formulas. You have two formulas. Suppose 1 want to prove them. To prove
them 1 have to start from something and then 1 have to get to the formulas by logic. So 1 start from what?
SERGE. You start from where you divide the circumference by the
diameter, that equals
'Il'.
WHAT IS PI?
SERGE LANG. And then what? Now you have to reach the area. What
is the definition of 'Il'? Before you can prove something, y ou must have a
definition.
SERGE. It's what 1 said, the circumference divided by the diameter.
SERGE LANG. But then, you have to show that it's the same
in the
formula for the area. If you tell me that 'Il' is the circumference divided by
the diameter, which is twice the radius, you can start with that as a
definition, but then you have to prove something, which is the other formula?
'Il'
So we have to start from something, with a definition, otherwise 1 can't

prove anything. And then logically, derive the formulas. So the question
is, where do we start from? That's what 1 am after. 1 want to start from
somewhere, and get to these two formulas.
1 will have to explain two things about these formulas. One is where
the r 2 cornes from; and second, where the 'Il' cornes from. They come from
two different aspects of the problem. One of the aspects has to do with the
r 2. Why is there an r 2 in the formulas for the area? And why is there an r
(but not r 2 ) for the circumference? The presence of the r and r 2 has to be
discussed. And the other thing 1 have to dicuss is the 'Il'.
So we start ali over. l'Il first explain the r 2, and after that l'Il explain
the 'Il'.
Let's go back to the even simpler case of the rectangle. Suppose 1 have
a rectangle of sides a, b. Then the area of the rectangle is just the product,
ab. Now suppose 1 take a rectangle whose sides are twice a and twice b, so
1 blow up the rectangle by a factor of 2. How does the area change?
A STUDENT. lt doubles.
SERGE LANG. What's your name?
THE STUDENT. Adolph.
SERGE LANG. The area doubles? What is the area of the new rectan-
gle?
[Another student starts ta/king.]

SERGE LANG. No. Adolph, 1 am asking Adolph. The area of a rectan-
gle is the product of the sides. Right?

ADOLPH. Yes.
SERGE LANG. So one side of the new rectangle is 2a, and what is the
other side?
ADOLPH. lt's 2b.
You also have to prove that no matter what circle you take, the ratio of the circumference
by the diameter gives the same number. This is precisely one of the things we are trying to
prove. 1 was not on the bali when 1 did not raise this objection explicitly that way.
WHAT IS PI?
SERGE LANG. That's right, so the total area is 2a times 2b, which is
4ab.
1
1
2a~----+-----
l
1
2b
Area =ab
Area
= 2a2b
= 4ab
N ow suppose 1 take a rectangle with three times the si des. So he re 1 have

si des of lengths 3a and 3b.
1
1
1
1
1
1
1
1
1
1
~---------,----------t---------
3a
----------~----------~--------1
3b
What is the area of the rectangle with three times the sides? Adolph.
ADOLPH. 9ab.
SERGE LANG. That's right, 9ab, it's 3a times 3b, which is 9ab. Suppose
1 now take a rectangle with one half the sides, so 1 have here
~b.
~ a,
and
What is the area of this rectangle, with one half the sides?
~bCJ
1
-a
2
ADOLPH. lt's ab over 4.

SERGE LANG. That's right. One fourth ab. Now suppose in general 1
take a rectangle wi th the si des ra and rb. Like this.
ra
rb
WHAT IS PI?
Adolph, what's the area of this rectangle?

ADOLPH. ra times rb.
SERGE LANG. That's right, ra times rb, which is what?
ADOLPH. r squared times ab.
SERGE LANG. Yes, r 2ab. So if 1 change the rectangle by a blow up by
a factor of r, how does the area change? Adolph.

ADOLPH. Can you repeat that, please?
SERGE LANG. Yes. 1 have my old rectangle. 1 blow it up by a factor of
r. In both directions. You see, the new sides are ra, rb? How does the
area change? The old area was ab. What is the new area?
ADOLPH. r squared ab.
SERGE LANG. Yes, r 2ab, not rab. The area changes, by what factor?
ADOLPH. By r 2.
SERGE LANG. You said r a minute ago. Weil it's not r. It's r 2. You see
it? Does everybody see it?

[Students agree th at they see it.]
SERGE LANG. So if 1 make a blow up by a factor of r, the area of a
rectangle changes by a factor of r 2. And of course, r can be bigger than 1,

or r can be smaller th an 1, like r == 1/2 or r == 1/3. So now you see how
area changes for rectangles. Any questions? Everybody got that?
[No questions.]
Ali right, now instead of a rectangle, suppose 1 have another figure.

Suppose 1 have a curved figure, like this.
Just like a kidney. Suppose 1 have a kidney, and it has a certain a rea A.
Now 1 blow up the kidney by a factor of 2, for example. What will be the
area of the blown up kidney?
A STUDENT. A 2?
SERGE LANG. No, let's go back. 1 have a rectangle with area A. 1 blow
this rectangle up by a factor of 2. What is the new area?
A STUDENT. lt's 4A.
WHAT IS PI?
SERGE LANG. Yes, the area changes by 2 squared. If I blow up the
rectangle by a factor of r, the area changes by a factor of r 2. Suppose I

blow up a curved figure. 1 blow up by a factor of 2. The area will change
by what factor?
A STUDENT. 4.
SERGE LANG. OK, the area changes by this same factor of 4. Why?
What is the proof? 1 have a curved figure, a kidney, not a rectangle. How
do 1 prove it? Anybody have any ideas? Ali right, Adolph.
ADOLPI-1. You measure around.
SERGE LANG. No, you don't determine area by measuring around. If I
measure around, 1 get the perimeter, the circumference. 1 am now dealing

with the area. The whole thing inside.
[A student raises her hand. There is a certain amount offumbling around

by severa! students. Afier a wh ile, Serge Lang picks up again.]
SERGE LANG. 1 try to reduce the question to rectangles. I make a grid
like this.
You see the grid? Then the area of the kidney is approximated by the area
of the rectangles which are inside the kidney. I look at ali the rectangles
here which lie completely inside the kidney. [Serge Lang draws the thick
li ne in the next figure.] They go like that, ali the way down there, up here,
there, and here.
WHAT IS Pl?
So if 1 take the rectangles here, inside the thick line, I get an approximation of the area of the kidney by rectangles. If 1 make the grid very fine,
as in the next picture, 1 get an even better approximation.
-r--..
~l"'"'"
/
Il
""'"
"'
'
1
1
l
~
'
"
\
1
l"'"'"'
7
~"""-- ~
..-
.;~il
You see ali these little rectangles there inside? If 1 take the sum of these
little rectangles, the sum of the area of these little rectangles, then I get a
good approximation of the area of the kidney.
Now I blow up the whole picture by a certain factor r. We make a dilation, that is an expansion, or a contraction, by a certain factor r. Then the
rectangles also dilate by a factor of r. For instance, if r is grea ter th an 1,
then a rectangle in the first drawing gets blown up to a bigger rectangle.
A STUDENT. Yes, if you blow up these rectangles inside the kidney by
a factor of 2, then the area of the blown up rectangles will be four times
the old area.
10
WHAT IS PI?
SERGE LANG. That's right. And if 1 blow up by a factor of r, the area
of the new rectangles will be r 2 times the area of the old rectangle. Everybody sees that? Look at ali the rectangles here, like this, there is a first
one, second one, third one, and so on. With areas A 1, A 2 ,
A3, A 4, As, A 6 , A 7 , A 8, A 9 , A 10 . So 1 number the areas of the little rectangles.
You see these rectangles there inside? Then 1 take the sum of the areas of
these rectangles, like the sum
A1
A2
A3
A4
As
A6
ali the way up to A 10.
1 have ten of them. This sum approxima tes the area of the kidney. Right?
Now 1 make a blow up of the whole situation, like in photography by a
factor-of whatever you want. What do you want: 2 or do you want r? Do
you want a specifie number, or can 1 user?
SEVERAL STUDENTS. r. You canuse r.
SERGE LANG. 1 can use r. OK. So 1 make a blow up by a factor of r.
Now if one of these rectangles has surface area A, as she said-what is

your name?
STUDENT. Rachel.
SERGE LANG. Rachel. As Rachel said, the area of the dilated rectangle
will be r 2A. So if 1 have here a rectangle with a rea A, over the re 1 will
WHAT IS PI?
Il
have a rectangle with area r 2A. So what will be the sum of ali the blown
u p rectangles? I t will be
r 2A 1
r 2A2
r 2A3
up to r 2A 10
RACHEL. Yes.
SERGE LANG. If I factor out r 2, 1 get r 2 in front, so I get
r 2(A 1
But A 1 + A 2
ney, and
+ +
10
A2
up to Ato).
is an approximation of the area of the kid-
is an approximation of the area of the kidney dilated by r, so it is an

approximation of the new area, that is r 2 times the old area.
N ow if 1 make a very fine grid, the a rea of the rectangles inside the
kidney gets closer and closer to the area of the kidney. And that's the reason why the area of the kidney also changes by a factor of r 2. Do you
understand the argument? Not quite?
A STUDENT. Kind of.
SERGE LANG. Kind of ... Did you understand the argument about
the rectangles? If I blow up each rectangle by a factor of r, its area

changes by a factor of r 2. Do you see that I can approximate the curved
figure by rectangles? How do I approxima te the curved figure by rectangles?
RACHEL. By adding up the area of the little rectangles.
SERGE LANG. That's right. I first make a grid and then I add up ali
the areas of the little rectangles. And that gives me a good approximation
of the area of the kidney. Then 1 blow up the whole picture by a factor of
r. Each little rectangle gets blown up by a factor of r, and the area of each
rectangles changes by a factor of r 2 , so the ir sum also changes by a factor
of r 2. And so I claim that is an argument why the area of the curved
figure will also change by a factor of r 2 . Do you accept that? Any comtnents?
[ Various students discuss the question.]
ONE STUDENT. lt gets bigger?
SERGE LANG. Yes, if r is bigger than l, it gets bigger. If r is l /2 it will
shrink by a factor of l /4th. That's right. Do you accept it? Anybody else?
ANOTHER STUDENT. Yeah, 1 accept it.
[Laughter, the statement
sounded funny.]
SERGE LANG. Do you have any objections to the argument?
12
WHAT IS Pl?
A STUDENT. I have a question. If r was less than 1, would it reduce by
a factor r to the negative 2?

SERGE LANG. No. If r is less than 1 then you would get the blow up
by a factor of r 2 because r 2 would be 1ess than 1. You can write r as sorne

other number s to the negative one, that is r == s-I, so r 2 is s to the negative 2, that is r 2 == s- 2 . Multiplying by r 2 is then really a shrink down, by
a factor less than 1.
A STUDENT. So it's not really a blow up.
SERGE LANG. Ali right, let's use a neutra! name, let's cali it a dilation,
that's a sort of neutra! name. If r is like 2 or 3, the dilation increases the

area. If r is 1 /2, then the area changes by a factor of 1 /4th, which is a
shrink down. So we use the neutra! name, dilation.
Our argument was by approximating the curved figure by rectangles.
We can say that under a dilation by a factor of r, area changes by a factor
of r 2. Can we go on from that? Can I erase? [Students nod.]
We go on. Now we get back to the circle, which is certainly much
better than the kidney. I mean, it's curved too, but it's better than the kidney because the curve is more regular.
Take a circle of radius 1. How do you get a circle of radius 2 from the
circle of radius 1?
A STUDENT. You increase the radius by 1.

SERGE LANG. By a factor of 1 or a factor of 2? Take another radius.
How do you get a circle of radius 10 from the circle of radius 1?
STUDENT. You've increased it by a factor of 10.
SERGE LANG. I've increased it by a factor of 10. And in general, if 1
have a circle of radius r, how do I get the circle of radius r from the circle
of radius 1?
STUDENT. You increase it by a factor of radius r.
SERGE LANG. By a factor of r. Yeh? So 1 can say that a circle of
radius r is the dilation of a circle of radius 1 by a factor of r. Do you
accept that?
STUDENT. Yeh.
WHAT IS PI?
13
SERGE LANG. And the circle of radius l /2 is the dilation of the circle
of radius l by a factor l /2. Now suppose you know the area of a circle of
radius l? Wh at will be the are a of a circle of radius 2? Let A be the are a
of the circle of radius l. What is the area of the circle of radius 2?
STUDENT. 4A.
SERGE LANG. Correct! What's you name?
STUDENT. Ho.
SERGE LANG. Ho. Good for you. 4A. And what is the area of the cir-
cle of radius 10?

ANOTHER STUDENT. 100A.
SERGE LANG. And what is the area of a circle of radius l /2? Rachel.
RACHEL. One quarter A.
SERGE LANG. Right. And what is the area of the circ1e of radius r?
RACHEL. r 2 ?
SERGE LANG. r 2 ? [Questioning]
RACHEL. r 2A.
SERGE LANG. r 2A. That's it. Now do you see the r 2 coming in? That's
the r 2.
By the way, there is a confusion of course about what we mean by a
circle. The periphery of the circle, that's what is called the circle. If you
want the inside of the circle, we have to make a distinction, we cali it a
dise. Like a frisbee. We cali it a dise because the Greeks called it a dise. If
the Greeks had called it a frisbee we would have called it a frisbee.
So we can say that the area of a dise of radius r is r 2A, where A is the
area of a dise of radius 1. Weil, you want to give a name for the area of
the dise of radius 1, and that's what we are going to cali pi. So I defi ne 'Il'
to be the area of a dise of radius 1. For our purposes today, we are going
to assume 'Il' as being that number. Of course we assume that we have
given units of measurement, whichever way you want. And that is what
we will take as the definition of 'Il'. The area of the dise of radius 1.
14
WHAT IS PI?
Therefore, if we do that, then by what we have seen before, the area of

the dise of radius r is 'll'r 2. Before we called A the area of the dise of radius
l. Now we cali it 'Il'. That's the constant. And we have explained the r 2 ,
starting with this definition of 'Il'. This takes care of one of our formulas.
OK? Any questions? Can 1 go on? [No questions.]
So let's go back to our original problem, when two questions Wl..re
raised. The area of a dise, and also the circumference of the circle. We
have taken care of the area. There remains the circumference. And the
theorem is that:
Theorem. The circumference of a circ le of radius r is 2'!1'r.
That is now the theorem. Let's cali the circumference c. We started with a
definition of 'Il', we got the 'll'r 2, and we now want to prove that
c == 2'!1'r.
That's what 1 want to do: give a systematic treatment of the formulas 'll'r 2
and 2'll'r. Right? Any complaints?
[Laughter.]
STUDENT. No complaints.
ANOTHER STUDENT. Not yet!
SERGE. Why do you have the radius, not the diameter?
SERGE LANG. 1 could have written the formula as
c == 'll'd, where d is
the diameter.
STUDENT. But wh y the diameter? How do you know it is
2r?
SERGE LANG. You don't. That's exactly what 1 am going to prove!
am making an assertion; 1 haven't proved it yet. That is the whole point

of the operation. To prove that this is the correct formula. 1 haven't
proved it yet. The proof is going to take place. 1 don't know, it's quite a
remarkable fact, that the same constant that appeared in the formula for
the area is going to come in the formula for the circumference, exactly in
that pattern. That is precisely the content of the theorem. That it cornes in
the formula in exactly this manner: the circumference is 'll'd, pi times the
diameter.
At first there is no reason why it should come out like that. 1 mean, it
could have come out any other way. If it cornes out like that, don't biarne
me. 1 mean, God made it that way! [Laughter.] 1 didn't fix it up that way.
It was fixed up by somebody else. But you ask me why, and l'rn going to
answer why. That's what the word "proof' means. [Laughter.] So 1 write
the word:
Pro of
And l'Il give the proof. What's your name?
WHAT IS PI?
15
STUDENT. Sheryl.
SERGE LANG. So Sheryl asks why. [Strong laughter.] Serge also asks
why. [Laughter.] And 1 am going to answer why. So what method am 1

going to use? 1'11 use the approximation method again. The only thing we
are sure of about areas is the area of rectangles and triangles. That's the
thing we are going to work with. So 1 look at my circle of radius r, and 1
approximate it with triangles. At first, let's use four triangles, like this.
You see four triangles. lt's a very rough approximation. lt's not especially
good. Then 1 can take my circle of radius r and approximate it with more
triangles, like for instance six triangles.
1 can take an approxima ti on wi th seven, or eigh t
or more like these, which get better and better.
16
WHAT IS Pl?
Have you seen regular polygons before? Yes? Good. So 1 approximate

the circle with polygons. The first polygon 1 used was a square, with four
sides. If the regular polygon has six sides, what do you call it? Do you
know what that is called, a regular polygon with six sides?
A STUDENT. Hexagon.
SERGE LANG. That's right. A hexagon. l'rn just checking what you
know. So here are the four triangles, next are the six triangles, seven triangles, and of course 1 can take more. As 1 take polygons with more and more
sides, 1 get a better and better approximation to the circle. Right? [During
al/ this, Serge Lang draws illustrations on the blackboard.] Here 1 draw one
wi th n, wi th ar bi trary n.
You know about n?

A STUDENT. lt's any number.
SERGE LANG. Yes. 1 don't want to treat just specifie cases, 1 want to be
able to take account of the general situation. So 1 will use n. Does it

bother you?
A STUDENT. No.
SERGE LANG. OK. So that's the method. 1 approximate the circle by
regular polygons. And so the area gets approximated by triangles, and the
circumference of the circle gets approxima ted by the perimeter of the
polygon. Here with the square it's not very good, we have a certain triangle which is repeated four times. In the next picture, we have a certain triangle which is repeated six times. Over there the triangle is repeated seven
times. Right? 1 give a name to the triangle each time. For example, 1 call
the first one T 4 The number 4 indicates that 1 have four triangles. 1 call
its base b4 and its height h 4 . The triangle T 4 is repeated four times.
WHAT IS PI?
17
Then 1 have the triangle T 6 , with base b 6 and height h 6 . The triangle T 6
is repeated six times.
Over there 1 have the triangle T 7 which is repeated seven times. In general 1 have the triangle Tn, which is repeated n times. 1 will cali its base bn
and its height hn.
What is the area of T 4 ? What's the area of a triangle?

SHERYL. One half bn times hn.
SERGE LANG. You are way ahead of me. Let's go slowly for every-
body. The area of T 4 is _}_b 4 h4 She gave me the general answer. Let's go
2
step by step. What's the area of T 6 ?
A STUDENT. One half b6 times h6 .
SERGE LANG. Yes, and what's the area of T 7 ? It's
~ b7 h7 .
And finally
what's the area of Tn ? Sheryl already gave it to me.

A STUDENT. One half hn times bn.
SERGE LANG. That's right. Now can 1 leave the picture just with n,
and erase the others, because 1 need the blackboard space? 1 picked the
pictures with 4, 6, 7 just to lead to the n, but 1 only need the n. Right?
Now what is the area inside the polygon?
STUDENT. The area of Tn times n.
SERGE LANG. That's right. What's your name?
18
WHAT IS PI?
STUDENT. Charlie.
SERGE LANG. So Charlie said, rightly, that the area inside the polygon
is n times the area of Tn. Which is n times
~ bnhn,
and we can write the
formula
And what is the length of the n-th polygon?

SHERYL. n times bn.
SERGE LANG. That's right. l'Il write the formula over there:
Do you follow? Adolph, do you follow?

ADOLPH. Yes.
SERGE LANG. Since the area inside the n-th polygon is An
1 can wri te this as
where Ln is the length of the n-th polygon.

Suppose 1 make n very large. So 1 increase the number of sides of the
polygon, and 1 get a better approximation of the circle.
After a while, 1 can't draw it any more. When n becomes very large, what
does this quantity on the right,
~ Lnhn
approach? The length Ln of the
polygon approaches what? Charlie.

CHARLIE. The circumference of the circle.
SERGE LANG. Yes. And what does the height hn approach?
STUDENT. The radius.
WHAT IS PI?
19
SERGE LANG. Very good. What's your name?

STUDENT. Joe.
SERGE LANG. So Joe says that hn approaches the radius of the circle
when n becomes very large. You've been doing very weil-Charlie, Joe,
Sheryl, Serge ... [Laughter.], Rachel ... It's hard to get ali the names
straight, for the first time. So hn approaches the radius [Serge Lang writes
on the blackboard and runs out of cha/k.]
A STUDENT. There's more chalk in the drawer. [Laughter.]
SERGE LANG. Good. So on the right hand side of the equality
the length Ln approaches the circumference c, and hn approaches the

radius. On the left hand side the area An inside the polygon approacheswhat? Let's pick on somebody new. [Serge Lang points to another student.]
THE STUDENT. Pi? [Serge Lang raises his eyebrows.] No ...
SERGE LANG. 1 have a circle of radius r. We've decided that the area
of the circle of radius r is?
THE STUDENT. wr 2 .
SERGE LANG. Good. So the left hand side approaches wr 2, and the
right band side approaches
_!_cr. Therefore 1 get

2
wr 2
== -cr
2 .
Now what do you do by algebra? Charlie.

CHARLIE. You cancel the r's.
SERGE LANG. Yes, you can cancel one r on each side, on r on the left
and the r on the righ t, so y ou get
wr
l
2
== -c.
Then what do you do?

CHARLIE. Multiply by 2.
SERGE LANG. Yes, you multiply by 2, and what do you get?
CHARLIE. 2wr equals c.
SERGE LANG. Yes, 2wr == c, and that's the formula you wanted to
prove. [Laughter throughout the class.] You see it?
We started with wr 2 as the area of the dise of radius r. Then we used an

approximation, by computing the perimeter of the polygon, the area of
the polygon. Th en we let n tend to infini ty and get the approxima ti on of
20
WHAT IS PI?
the length of the circumference and the area inside the circle. The right
band side approaches
~ cr;
the left band side approaches the area of the
dise which is 'll'r 2. We have already proved that before. And then you use
a little algebra, you cancel one r and multiply by 2, and you get the formula
2'll'r
== c.
There is your formula. Do you agree that's a proof? [Serge Lang points to
Rachel.]
RACHEL. Yes. [Her tone is uncertain.]
SERGE LANG. You do?
RACHEL. Yes. [Laughing a little.]
SERGE LANG. What do you mean "yes"? ls it a yes by intimidation or
ayes by conviction? Or a little bit of both?

RACHEL. A little bit of both. [Laughter.]
SERGE LANG. Weil, where is the intimidation?
RACHEL. 1 don't know.
SERGE LANG. You don't know? [Laughter.] Ali right, let's make it ali
conviction. Look, where did 1 start from? You granted me that '!l'r 2 is the
area of the circle of radius r. Now after that what did 1 do? Heh?
RACHEL. You drew a circle. [Laughter.]
SERGE LANG. Yes, 1 drew a circle. And after that what did 1 do?
RACHEL. You divided it into triangles.
SERGE LANG. That's right, 1 divided the inside of the circle into triangles. 1 drew a regular polygon inside the circle, which 1 divided into triangles.
RACHEL. Yeah.
SERGE LANG. Then, 1 used logic. 1 mean, you granted me the area of
triangles. So 1 took a regular polygon with n sides, 4, 5, 6, 7, 8, 9,

10 ... and then in general, 1 do it with n. "n" doesn't bother you?
RACHEL. No.
SERGE LANG. Not at ali? OK. Then 1 give names to the height and
the base of the triangles. 1 cali the base bn and 1 cali the height h 11 Does
that bother you?
RACHEL. No.
SERGE LANG. ls that intimidation or conviction?
RACHEL. That's conviction.
SERGE LANG. So far no intimidation?
WHAT IS PI?
21
RACHEL. So far so good.

SERGE LANG. So far so good. [Laughter.] Now what's the area of each
triangle?
RACHEL. Half the base times the height.
SERGE LANG. Half the base times the height, you said it yourself. So
that's not intimidation.

RACHEL. No, that's conviction.
SERGE LANG. l'rn glad that's conviction. [Laughter.] Ali right, what's
the a rea inside the polygon? How many triangles do 1 have inside? [Serge
Lang draws a new pic ture.]
RACHEL. [She starts counting:] One, two, three, ... ,seven-1 mean n.
SERGE LANG. n triangles, right. And if 1 have the area of each trian-
gle, and fi nd n triangles wi th this sa me a rea, th en wh at is the total a rea

inside the polygon?
RACHEL. n times the area of each triangle.
SERGE LANG. So the area inside the polygon is n times
~ bnhn-
So far
so good?
RACHEL. Yes.
SERGE LANG. Conviction?
RACHEL. Yeah [smiling].
SERGE LANG. Ali right. Weil, what was my next step? The 1 /2 is 1 /2,
and then 1 took n times bn, n times that base. What is the length of the
perimeter of the polygon? Do you see the polygon? How many sides does
it have?
RACHEL. Seven-oh, l'rn sorry, 1 mean n.
SERGE LANG. That's ali right, you start with the special number 7, but
then your mind works it out in general and you say n, which shows you
understand. You end up saying there are n sides ...
RACHEL. Yes.
22
WHAT IS Pl?
SERGE LANG. Each side has Iength bn. So what is the Iength of the
perimeter of the poiygon?

RACHEL. n times bn.
SERGE LANG. Yes, do you see the n times bn here?
RACHEL. Yes.
SERGE LANG. So 1 get nbn, which is Ln, the Iength of the n-th
poiygon, that's the perimeter of the n-th poiygon. Then the area inside the
n-th poiygon is
ls that conviction or intimidation?

RACHEL. lt's a conviction.
SERGE LANG. Ali right. Now suppose you make n Iarger and Iarger to
get a better and better approximation of the circie. What do these quanti-
ties approach?
approaches
~.
The length Ln of the polygon
approaches what?
RACHEL. The circumference of the circie.
SERGE LANG. [To the class:] That's what she said, the circumference of
the circie. [To Rachel:] ls that conviction or intimidation?

RACHEL. Conviction.
SERGE LANG. The hn approaches what?
RACHEL. The radius.
SERGE LANG. Definiteiy.
So the product of these three things

approaches one haif times the circumference times the radius. Do you see
it? Any questions so far?
RACHEL. No.
SERGE LANG. ls that conviction or intimidation?
RACHEL. Conviction. [Smiling.]
SERGE LANG. Ali right. Now 1 have the area An inside the n-th
poiygon on the Ieft hand side, which approaches what?

RACHEL. The area of the circie.
SERGE LANG. Which is what?
RACHEL.
'll'r .
SERGE LANG. OK. So 1 get that '1Tr 2 is equal to

RACHEL. So far so good!
~cr.
WHAT IS PI?
23
SERGE LANG. So far so good. Now if 1 have

'Tfr
==
-cr
'
you should have an irresistible impulse to do something to this equation.

[Laughter.] What's your irresistible impulse?
RACHEL. To change this to 2'Tfr
==
c.
SERGE LANG. Yes, and how does this change come about?
RACHEL. You cancel the r, and then you multiply by 2 on one side
and by 2 on the other si de, and th en 2'Tfr == c.

SERGE LANG. That's the formula 1 wanted to prove!
RACHEL. You proved it. [Laughter.]
SERGE LANG. 1 proved it. Now where is the intimidation?
RACHEL. 1t's go ne. [Sa id with definiteness.]
SERGE LANG. It's gone!? Ah ... [Laughter.] So we won, the mathematics won! Right?
RACHEL. Right.
[Serge rai ses his ha nd.]

SERGE LANG. Yes?
SERGE. That's only approximate.
SERGE LANG. No. At the n-th level it's only approximate, but if 1 take
the limit, what the quantities approach, it is not approximate.

SERGE. But still, my point is that it will never quite reach the circle. If
you divide the circle by a certain number ... even an infinite number ...
SERGE LANG. 1 don't divide by an infinite number. 1 use the word
~'approach". Do you grant me that the area inside the polygon approaches
the area of the dise?
SERGE. Yes, but it never quite reaches it.

SERGE LANG. No it doesn't, but it approaches it. Whatever the area of
the dise is, it is approached by the area inside the n-th polygon.
SERGE. Yes, but it never quite reaches, it's not exact.
SERGE LANG. What is not exact? [Serge Lang and Serge talk simultaneous~y.]
l'rn going to write something down on the black board. 1 have

the area of the circle, which 1 know is 'Tfr 2. 1 have the area An of the n-th
polygon. The area An approaches the area of the circle. 1 did not say it is
exactly eq ual, i t approaches.
SERGE. Since it never quite reaches, how can you say ...
24
WHAT IS PI?
SERGE LANG. 1 don't understand that sentence. What do you mean
"never qui te reaches"? [Serge Lang and Serge talk at the same time.]
SERGE. The area of the polygon never quite becomes the same as the
area of the circle.

SERGE LANG. No, it does not. That's right.
SERGE. OK, then how can you say that your formula is exact?
SERGE LANG. Ah, the area of the n-th polygon approaches the area of
the circle. The area of the n-th polygon is
~ Lnhn.
What does
~ Lnhn
approach?
SERGE. Weil, the, euh ...
SERGE LANG. What number does
SERGE.
Lnhn approach?
cr.
SERGE LANG. That's right. On the one hand the area of the n-th
polygon approaches
'll'r ;
and on the other hand this other number
_]_Lnhn, which is equal to that area, approaches _]_cr. If 1 have an expres2

2
sion which approaches two numbers, but the same expression, then the
two numbers have to be equal.
SERGE. Oh, 1 see what you mean.
SERGE LANG. When 1 write
1 have an equality sign here. This equality sign is not approximate. This
is a precise equality. Yeah? And on the one hand, this number An

1
approaches '!Tr 2 ; and on the other band that same number
Lnhn
approaches
_]_cr.
So the same number approaches two possibilities. Then

2
the two possibilities have to be equal.
SERGE. Ah, yes.
SERGE LANG. Do you see that? Does that answer your question?
SERGE. Yes, it does.
SERGE LANG. Anybody else have a similar question? Do you under-
stand the argument? [Serge Lang points to a student.]

THE STUDENT. Yes.
STUDENT. Mike.
WHAT IS PI?
25
SERGE LANG. Will y ou repeat the argument? Mike! [Laughter.]

MIKE. OK. You've got a circle, and [Laughter] and you divide it, OK,
and the radius is r. And you divide it up n times. Here n is any number.
So, then with those little ... when you divide it you have little triangles,
and the area of the triangle is half base times height.
SERGE LANG. Yes.
MIKE. And the base of that triangle would be bn, the height would be
hn, so the area is half bnhn
SERGE LANG. Right.
MIKE. Then you times it by n, which is the number that you got.
[Laughter.] And then n times bn is the length, so that becomes Ln. Then
half Lnhn is half times c times r . ..
SERGE LANG. Approaches, not "is".
MIKE. Oh, approaches. Half times c times r. And the, OK, '!l'r 2 equals
_!_cr. You cancel one r from each side; then you times both sides by 2, so
2
2'!1'r
equals c.
SERGE LANG. Bravo! You re ally got it! [To the class:] Do y ou see what
Mike did? He was able to repeat the whole proof, ali at once, the whole
sequence of ideas. That's wonderful. Good for you.
Comments
The response to the lecture was especially gratifying. A teacher told me
afterward that she had been apprehensive about Rachel, when 1 started
the "conviction-intimidation" sequence, whether Rachel would bear up
under the strain. Not only did Rachel bear up (1 don't even know if it was
a strain), but she progressively acquired confidence, and the tone of her
answers changed from uncertain to being quite assertive. Regretfully, the
tone cannat be reproduced on the printed page.
1 would not engage in this kind of sequence with any student. There
has to be a quick judgment on my part whether the student will indeed
find such an exchange too much of a strain, or on the contrary, will reach
a new level of understanding because of it.
The final performance by Mike, who is able to reproduce the whole
proof, is a great success. One of the objections which 1 have to elementary
and secondary school curriculum and teaching is the overemphasis on
technical matters, the incoherence, and the lack of sweep (analogous to
musical phrases) in the questions considered. The topic 1 selected has a
certain sweep, and it was fascinating to see the mental wheels turning,
reftected in Mike's facial expression, while he was reproducing the proof.
He had to do it ali verbally, put the sentences together, and solve a
26
WHAT IS PI?
number of problems of organization, choice of words, sequence of ideas,

which gave pleasure to him, to the rest of the class, and to me.
On a more technical level, the approximation discussed in the first part
of the proof (concerning the change of a rea und er di la ti on) should be
pushed a little beyond the point where I did, but 1 did not have the time
to do so since 1 wanted to deal completely with the second part of the
proof. If I had another hour at my disposai, I would expand on the reasons why the area so changes along the foliowing lines. Consider a grid
consisting of vertical and horizontal lines:
........ ~
~,
~~
1
\
' ,,
'
.,.,~
To determine the area of the dise approximately, we could count ali the
squares that lie inside the circle, measure their sides, add up their areas,
and get the desired approximation. But we want to estimate how good is
the approximation. The difference between the sum of the areas of ali the
little squares contained in the dise and the area of the dise itself is determined by ali the smali portions of squares which touch the boundary of
the dise, i.e. which touch the circle. We have a very strong intuition that
the sum of such little squares is quite smali if the grid is fine enough. We
can give an estimate for this smaliness. Suppose that we make the grid so
that the squares have sides of length a. Then the diagonal of such a
square has length a V'i. If a square intersects the circle, then any point on
the square is at distance at most a V2 from the circle. Look at figure (a) .
.....__
~
'"
(a)
'
(b)
WHAT IS PI?
27
This is because the distance between any two points of the square is at
most a V2. Let us draw a band of width a V2 on each si de of the circle, as
shown in figure (b ). Then ali the squares which intersect the circle must
lie within that band. It is very plausible that the area of the band is at
most equal to
2a V2 times the length of the circle.
Th us if we take a to be very small, i.e. if we take the grid to be a very fine

grid, then the area of the band is small, namely at most 2a V2L where L
is the circumference of the circle.
Therefore we have estimated the error in approximating the area of the
circle (dise) by the sum of the are as of the squares of the grid which intersect the circle. This error approaches 0 as the size of the grid approaches
O. Therefore, as the size of the grid approaches 0, the sum of the areas of
the squares of the grid which lie entirely inside the circle approaches the
area of the dise.
A similar argument applies to any curved figure.
Finally, in a later talk, you will see another way of deducing the length
of the circle c == 2'll'r from the a rea of the dise A == 'll'r 2, which in sorne
ways is simpler than the one chosen here, and which generalizes to finding
the area of the sphere.
Volumes in higher dimension
This is the second talk, following "What is pi?" and dealing with
volumes of standard figures like pyramids and cones. The talk was
given to the same class, with lOth grade students, approximate!J' 15
years old, in a high school in the suburbs of Toronto.
SERGE LANG. Today, we are going to do similar things, but in higher
dimension. But first, a remark in 2 dimensions. Let's start again with our
rectangle, with sides a, b.
Area =A= ab
Yesterday, 1 made dilations by a factor of r in ali directions. It could be a

blow up or a shrink down-1 called it a dilation. Now you should note
that 1 can dilate one direction by a certain factor, and another direction
by another factor. For instance, 1 can keep one direction fixed, and blow
up in the other direction by a factor of 2. Then this side has length 2a, but
the other si de still has length b.
If the area of the first rectangle is A, wh at is the new are a? Sheryl.
SHERYL. 2A.
SERGE LANG. Yes, 2A. In general, if 1 make a blow up by a factor of r
in one direction, and s in the other direction, the area will change by what
factor?
SHERYL. rsA?
VOLUMES IN HIGHER DIMENSION
2a
~---------------------------
29
Area = 2ab = 2A
SERGE LANG. That's right. The area of the new rectangle will be rsA.
ra
Sb
A rea
= ra sb = rsab = rsA
So let's write this down. If we dilate one direction by a factor of r, and if

we dilate the other direction by a factor of s, then area changes by a factor of rs.
See, yesterday, we dilated both directions by a factor of r, and we saw
that area changes by a factor of r 2. Now today we allow dilations in one
direction by a factor of r, and in another direction by a factor of s. Is that
clear? By directions, 1 mean that we have picked two directions in the
plane, which are perpendicular to each other.
Like yesterday, we started with rectangles. What happens for general
areas? Let's go back to the kidney, or even more simply, suppose 1 start
with a circle of radius 1, and 1 dilate the circle by a factor of a tn one
direction, and a factor of b in another direction, like this:
30
We know from yesterday that the area of the circle of radius 1 is 'Il'.
What is the area of the figure obtained by dilating the circle by a factor of
a in one direction, and a factor of b in the other direction? Let's ask sornebody. Joe?
JOE. '!l'ab.
SERGE LANG. '!l'ab, that's right! By the way, do you know what it's
called, a dilated circle, by a factor of a in one direction and a factor of b

in the other direction? You know what it's called?
A STUDENT. An ellipse.
SERGE LANG. That's right. That's the definition of an ellipse. That's
exactly what it's called. Good for you. And we just saw, right from the
definition, that the area of an ellipse is '!l'ab.
If you have a curved figure like a kidney, you can again approximate
its area by little rectangles. If you dilate by a factor of a in one direction,
and a factor of b in the other direction, then the area of each rectangle
changes by a factor of ab, so by approximation, the area of the curved
figured changes also by a factor of ab.
Now let's go to 3 dimensions. We have worked in the plane.

In 3 dimensions, what is the analogous thing to a rectangle? What is
the 3-dimensional analogue of a rectangle?
SHERYL. Oh-length times width times depth.
SERGE LANG. Yeah, 1 have length, width, and depth. So 1 have 3
directions, and what's the analogue of a rectangle?
31
[Sorne fumbling by severa! students.]

SERGE LANG. Weil, it's called a rectangular box, OK? Do you have
any other name for it?

SHERYL. A cube?
SERGE LANG. Weil, a cube is a special case of a box, when ali the
sides are equal. A cube would be the analogous thing to a square. The
analogous thing to a rectangle, l'Il just cali it a rectangular box. Let's draw
one. See the box?
., /- --- ----- ---- - -- - ----- -- - --
.)-----
/
/
-------
//
/
/
/
/
a
cube
b
rectangu lar box
Suppose its sides are a,b,c. Then what is the volume?

SHERYL. abc.
SERGE LANG. That's right, the volume is the product of the sides,
V == abc.
Now suppose 1 make a dilation by a factor of r in ali directions in three

dimensions. So 1 blow up the whole 3-space by a factor of r.
[A student starts interrupting getting into volumes . .. ]
SERGE LANG. Wait, wait, wait! Let me draw it. You guys are too fast
for me. [Laughter.] Then the sides of the dilated box are ra, rb, re.
.,J--- ----------- ----------------//

/
ra
/
//
///
re
/
///
rb
32
So what is the volume of the box dilated by a factor of r in ali directions?

Let's pick on somebody. [Serge Lang points to a student.]
STUDENT. r 3abc.
SERGE LANG. That's right, r 3abc. Good for you. So the dilated volume
is r 3 V, if V was the volume of the original box. What's y our name?

STUDENT. Gary.
SERGE LANG. Yes, 1 remember from yesterday.
Let's see-Sheryl,
Serge, where is Rachel? Oh, over there, you moved! You confused me.
[Laughter.] Ail right. So the volume changes by a factor of r 3 .
Weil, let's look at three dimension al kidneys. Yesterday, wh en 1 had a
2-dimensional curved figure, 1 made a blow up by a factor of r, or a dilation by a factor of r, the area changed by a factor of r 2 . Now suppose 1
have a 3-dimensional kidney.
And it has a certain volume V. 1 make a dilation by a factor of r. What

will be the new volume?
SERGE.
r 3 V.
r 3 V, again. And how would you prove it? Weil how did
1 prove it yesterday?
SERGE LANG.
STUDENT. You drew a grid.

SERGE LANG. Yes, 1 drew a grid. Yesterday, 1 drew a 2-dimensional
grid, and 1 approximated the a rea by a lot of rectangles. N ow 1 draw a 3dimensional grid. What's your name?
STUDENT. Sandra.
SERGE LANG. Ali right. So to prove it, you make a 3-dimensional grid.
But it's hard to draw!

SERGE LANG. [Tries to make a 3-dimensiona/ grid.] If 1 make a 3-
dimensional grid, then 1 can approximate the volume of the kidney by a

whole lot of rectangular boxes inside the kidney, and it gives me a good
approximation. And when 1 dilate by a factor of r, each rectangular box is
dilated, and its volume changes by a factor of r 3 .
33
rb
This will be true for each one of the little boxes. And since ali the little
boxes approximate the volume of the kidney, that means that the actual
volume of the kidney changes also by a factor of r 3 . So do you accept
that? But 1 have difficulties drawing it.
By the way, just like in two dimensions, 1 could make a dilation by a
factor in one direction, and another factor in the other direction. So let's
draw it here.
1 start wi th a rectangular box, and 1 make a dilation by a factor of r in
one direction, s in another, and t in the third direction. If the volume of
the original box is V, what is the volume after 1 make these dilations?
STUDENT. rst V.
SERGE LANG. Yes, so the volume changes by the product of the three
factors.
ANOTHER STUDENT. Yes, the volume is rasbtc, so rstabc and the
volume changes by the product rst.
34
SERGE LANG. And similarly, if 1 make this dilation for a curved
figure, the same result will be true; by making a 3-dimensional grid, and
approximating the curved figure by rectangular boxes. Do you accept ali
that? [Students approve.] So we can state the general result:
Und er dilation by factors of r, s, t, in the three dimensions, the volume of
a sol id changes by a factor of the product, rst.
Just like yesterday: area changes by a factor of r 2 if we dilate by r in each
direction; a factor of rs if we dilate by a factor of r in one dimension and
s in the other dimension; and now volume changes by a factor of rst if
you dilate by a factor of r in one dimension, s in another and t in the
third. And the three dimensions are in perpendicular directions.
Now l'li deal mostly in three dimensions, but what would be a natural
generalization of this? Serge.
SERGE. 1 don't know.
SERGE LANG. What's a generalization of what 1 have just done there?
1 started in 2 dimensions, th en 1 wen t to 3 dimensions ...
SERGE. [Interrupts.] Four dimensions. OK. lt's the next product. 1 see.
lt's rst whatever.
SERGE LANG. Ah, rst whatever. That's right. So suppose 1 have a solid
in four dimensions. You see the four dimensions? Now 1 can't draw it.
STUDENT. Weil, you could not draw it either in three dimensions!
SERGE LANG. That's a very good remark. You are absolutely right. So
the truth of what 1 am saying does not depend on my ability to draw the
picture! So suppose 1 have a solid in four dimensions, and 1 make a dilation by a factor of r in ali four dimensions. How does the volume of the
solid change? Sheryl.
35
SHERYL. r to the fourth, times V.

SERGE LANG. r 4 V, where Vis the 4-dimensional volume. And suppose
1 have a solid in n dimensions, and 1 make a dilation by a factor of r in
ali directions, in ali n dimensions. How does the volume change?

SERGE. r to the power n.
SERGE LANG. rn, and that's how it is in n dimensions. OK? Any prob-
lems? Sandra.
SANDRA. No. [The other students nod, and seem perfectly at ease.]
SERGE LANG. Ali right. So we see how volume changes in n dimen-
sions.
Let's go back to three dimensions, and see how we can find the volume
of sorne standard figures. But 1 think it's remarkable how you react to the
possibility of n dimensions. [Laughter.] 1 am slightly taken aback at the
way you just went along with it.
Ali right, let's try to determine volumes for figures like the pyramid
and the cone. Let's take the pyramid. First, take the most simple-minded
pyramid, when the base is a square, and the pyramid is straight up, like
the Egyptian pyramids. See the pyramid there?
square base
rectangu lar base
The top of the pyramid is calied the vertex. The pyramid has a square
base, with sides a, a; and h is the perpendicular height of the pyramid.
What do you think is the volume? What's the formula for the volume?
Also the pyramid with rectangular base. Ken: what do you guess the
volume is?
KEN. One half abh?
SERGE LANG. That's your guess?
KEN. Yes.
SERGE LANG. Sheryl, what do you say?
36
SHERYL. The same.

SERGE LANG. You say 1/2? What would you say? [Pointing to a stu-
dent.]
CHARLIE. One third.
SERGE LANG. Ah, Adolph says one third. [Laughter.
ft was not
Adolph . .. ] Euh, sorry, what was your name?

CHARLIE. Charlie.
SERGE LANG. Sorry, 1 got mixed up. Charlie says 1 /3. Any other
gu esses?
A STUDENT. One quarter abh.
SERGE LANG. _}_abh?
ANOTHER STUDENT. One fifth abh? [Laughter.]

SERGE LANG. Weil, you could go on quite a while like that. [Severa/
students discuss the matter among themselves.] Ali right, l'Il tell y ou. The
answer that is true is the one third. That's what's true:
The volume of the pyramid with rectangular base is
abh.
Where did you pick that up, by the way?

CHARLIE. 1 thought that you had three different dimensions, so it
should be one third, because you had one half for a triangle. From looking at it, it should be like that.
SERGE LANG. That's pretty good intuition. Of course, it does not
prove it, but it allows you to guess it. So now we have the problem of
proving it. Charlie, how do you prove it? [Silence.] How do you obtain a
pyramid? There are three numbers which come into it. The a for one side,
b on the other side, and the h for the third direction, the height. And you
know how volumes change under dilations by certain factors. So take the
simplest case. What appears to be the simplest case, when 1 have a square
of side 1 for the base-and what do you think will be the simplest case for
the height? Charlie.
CHARLIE. 3
SERGE LANG. No. Why should the height be 3?
CHARLIE. Because you divide by 3.
SERGE LANG. Ali right, let's consider that. He wants to make a
pyramid of height 3. ls that the simplest thing you can think of? How are
we going to prove what the volume is, of this abject? See, if you knew the
volume of this abject, you could get the volume of that one (a dilated
one) by dilations.
37
3
rh
l_
ln your pyramid of height 3, there is not enough symmetry. The height is
still too arbitrary. You have to try to get as much symmetry as you possibly can find.
SHERYL. The height is 2?
SERGE, OR KEN. One?
SERGE LANG. One looks already better. That's right, that's more sym-
metric.
But even with height l, how are you going to prove it? You still have a
problem. So you stare at the problem for a while, and 1 think eventually
you would find how to do it. But 1 have only forty minutes more, so in
fact l'rn going to pull a rabbit out of a hat. 1 mean, 1 have no choice. If 1
could meet you once more, 1 would leave it like this today, tell you to go
home and think about it before you go to bed; and if you solve it yourselves, you'd get a kick out of it. Now, 1 have to interfere with that kick,
because 1 have only forty more minutes.
So the proof is as follows. 1 take the sim pl est case [Serge raises his
hand]- Yes?
SERGE. 1 think that you could, by taking three pyramids, if you put
them together with their corners straight, you'd get a box ...
SERGE LANG. Eh, eh, eh, wow, you see, he's getting there! What do
you want to do? What does Serge want to do?
38
SHERYL. He wants to take two pyramids, with the same direction, that
should fit in a box.

SERGE LANG. That's right. But do you think he can do it with two?
SERGE. No, I think three will fit.
SERGE LANG. You think three pyramids will fit in a rectangular box?
SERGE. If you make the corners of the pyramid fit the corners of the
box in the straight angles. You can make two of the angles straight.
SERGE LANG. Ah, that's a possibility. But can I really do it, and how?
If I put one pyramid like this, in one corner, then the others will have to
slant. They won't be straight pyramids, with the top vertex lying above the
middle of the base.
\
\
'J--
I
1
1
1
The moment the pyramid slants, l'rn facing a new phenomenon, which I
will discuss immediately afterward, because it's a nice phenomenon. But
the slanting creates a new problem.
[There follows a rapid discussion between Serge Lang and Serge about
how to place the pyramids inside the rectangular box.]
SERGE. Can 1 come to the blackboard?
SERGE LANG. Yes, if you want.
[Serge cornes to the blackboard and starts drawing various possibilities.

Laughter by the rest of the class at the vivacious exchange. There is an
active discussion between Serge, members of the class and Serge Lang about
the various possibilities.]
SERGE LANG. Ali right, let's come back to what Serge wants to do
later. Suppose now we consider this picture.
39
SERGE LANG. Suppose you decompose the box like this, putting the
vertex of the pyramid in the middle. How many pieces inside the box will
you get?
A STUDENT. You want to know how many parts there are? There are
SIX.
SERGE LANG. Yes. Six together make up the box. What's your name?
STUDENT. Lisa.
SERGE LANG. She's absolutely right. She's going to get it. You have to
put six pyramids together to make up a box. Now then, why don't you
start in the opposite way, with a box that you know about. What is the
simplest box?
LISA. One by one by one.
SERGE LANG. Perfect. So take the simplest box, a cube which is one
by one by one.
1
1
1
1
1
1
'
~------
That's what Lisa said. Where will 1 put the vertex of the pyramid?
LISA. ln the middle.
SERGE LANG. Right here, in the middle. Lisa said, put the six pyram-
ids together. l'Il do what she said. Like that.
You see, there are six sides to the box, which is a cube. Then there are six
pyramids, with vertex at the center of the cube, and whose bases are the
40
six sides of the cube. These six pyramids together fill out the total cube.
So what is the volume of each pyramid? Lisa.
LISA. One sixth.
SERGE LANG. l /6. And the total volume of the whole cube is l, so
each pyramid is one sixth of the total volume, which is l /6. So now 1
have the volume of one pyramid, whose base has si des l, 1- and what is
the height of that pyramid?
A STUDENT. One half the side of the cube.
SERGE LANG. Yes, l /2. The height is l /2. Now suppose 1 have a
pyramid whose base has sides a, b and the height is still l /2.
h = 1/2
!
b
How is this pyramid obtained from the previous one? Mike.

MIKE. By dilating one side by a, one side by b, and the height by noth-
tng.
SERGE LANG. That's right. So the pyramid whose base has sides a, b
and with height l /2 is obtained by making a dilation by a factor of a, b in
the two base directions, and a factor of l in the third direction. As Mike
said. So what is the volume of this pyramid?
MIKE. Half ab?
SERGE LANG. No. The other pyramid, with sides l, l and height l /2
had volume l /6. If 1 make a dilation by a factor of a on one side, b on

the other, and l on the third, what is the new volume?
MIKE. One sixth ab.
SERGE LANG. Yes,
ab. This thing bas volume
ab.
1
2
!
b
Volume=
6 ab
41
Now suppose 1 change the height, and 1 look at the following pyramid,
with base ab and the heigh t is 1?
Volume= 2-ab =-ab

6
3
1
1
a
'/
)----t--
1 make a dilation in the third direction, the vertical direction, by what factor?
MIKE. 1.
SERGE LANG. No. 1 start with the pyramid with height 1 /2. The
height of the new pyramid is 1. By how much have 1 dilated in this direction?
MIKE. lt's doubled.
SERGE LANG. Certainly, so 1 haye dilated by what factor?
MIKE. 2.
SERGE LANG. That's right. So 1 have three pyramids. The first with
sides 1, 1 and height 1/2. The second with sides a, b and height 1/2. The
third with si des a, b and height 1, which is obtained from the second by
dilation by a factor of 2 in one direction. The volume of the second
pyramid is
ab. What is the volume of the third pyramid?
MIKE. Two one sixth ab.

SERGE LANG. Yes, 2 X
ab, which is what?
MIKE. One third ab.

SERGE LANG.
ab. And you see that one third coming out? We have
caught the one third.

Now 1 draw the final pyramid, with arbitrary height h. How is the
pyramid with sides a, b on the base and height h obtained from the
pyramid with si des a, b on the base, and height 1? Who knows?
[Most hands in the class go up.]

MIKE. Times h.
42
Volume= 3abh
SERGE LANG. Yes. The pyramid with sides a, b on the base and height
h is obtained as a dilation by a factor of h in the vertical direction, of the

pyramid with sides a, b on the base and height 1. So then, what is the
volume of this new pyramid of height h and sides a, b on the base?
SANDRA. One third abh.
SERGE LANG. Yes, so we write
Volume of pyramid
But th at is the most general theorem!
With this method, 1 take the pyramids with the top at the center of the
cube, and 1 get six pyramids. In this situation, the simplest case was of a
pyramid with a base of sides l but height l /2. Then 1 could get the
volume directly, for a pyramid like the Egyptian pyramids, when the top
lies directly above the center of the base.
SERGE. 1 think you could get it a different way. There are three
pyramids which make up the cube. The volume of the cube is abh.
SERGE LANG. It's not a cube, it's a rectangular box.
SERGE. Ail right. But you have three pyramids making up the box.
43
[Serge draws picture on the blackboard.] You take the three pyramids
with bases which are the sides opposite to one corner of the box.
SERGE LANG. No, not quite, because if you take a, b, and h different,
then the pyramids are not ali the same. If I take a cube and ali its sides
are equal to l, then you do end up with three pyramids which are the
same, and your idea is a good one.
Then by symmetry ali three pyramids have the same volume. If you have
a rectangular box, not a cube, then you don't have the symmetry. That's
why we have to deal with a cube, rather than a rectangular box. To have
more symmetry. With your method, the pyramids which you get are
slanted, like this:
Volume= 1/3
The top of each one of your pyramids does not lie exactly above the
center of the base. So you have a different kind of pyramid. But you are
right that the volume of each one is one third the volume of the cube, so
you get the right answer for pyramids whose top lies verticaliy above one
corner of the base. In this case the pyramid is one third of a cube of side
1.
Then we can make a dilation by a factor of a on one side of the base, a
factor of b on another side of the base, and a factor of h verticaliy, and
you get the formula for your type of pyramid, when the vertex lies
directly above one corner of the base:
44
Volume=.! abh =.! Bh

3
\
\
\
\
\-
/
/
/
a
But then you still have the problem of slanting to deal with.
So let's do what Serge wants to do. Suppose 1 have a rectangle for the
base, and suppose 1 have a slanted pyramid, as on the picture.
\
\
\
\
\
\
r--1
-
-/-------
What is the height of the slanted pyramid? Charlie.

CHARLIE. Wouldn't it be the distance from the point on top to the
base?
SERGE LANG. That's right. So what will be the volume of the slanted
pyramid?
CHARLIE. abh one third?
SERGE LANG. Yes, that's the guess, as Charlie said. But we have not
proved it. So we have to know about the slanting process, so let's study
slanting, which is wh at Serge wanted to do. 1t is also called shearing.
Let's go back to two dimensions. 1 take a rectangle.
STUDENT. Not a square?
SERGE LANG. No, not necessarily. You can think of a rectangle as the
side of a deck of cards.
45
And 1 slant it. This way.
1/
b
Z __ _
So 1 geta parallelogram. Then the height is still the same, the perpendicular height. So if A is the area of the rectangle, what's the area of the
parallelogram?
SANDRA. A.
SERGE LANG. It's still A. The are a does not change. Have y ou seen
that before? You have done the area of a parallelogram, last year?
CHARLIE. Yeah.
SERGE LANG. And how did you prove it? [Students simultaneously
recall the proof with triangles.] That's right, you use this picture:
'--/J~i--------L.Zl_
With the two triangles. So we assume that you know that the area of a
parallelogram is the base times the height. Th en we can state the theo rem:
Under slanting, or onder shearing, the area does not change.

Now in three dimensions, take a deck of cards, so the rectangle is the
side of a deck of cards. And th en slant it.
deck of cards
slanted deck
46
So you see the deck of cards, and over there is the sheared deck. The top
is moved over like that. So if you shear a deck of cards, the volume of the
deck of cards won't change. We proved for rectangles that if you shear in
one direction, then the area does not change. Then in three dimensions
we get the same type of theorem:
In three dimensions the volume does not change onder shearing.
We have done this for rectangular boxes, shearing in one dimension.
But now we can do it to pyramids, with a square base, or a rectangular
base. And it's a straight pyramid. And 1 shear it like that, so the height
still re mains the sa me. The perpendicular heigh t.
/----
'/
straight pyramid
sheared pyramid
Then the volume will not change under shearing. And this holds for any
3-dimensional body. How do 1 prove it? Suppose 1 take any 3dimensional body. 1 claim that its volume does not change under shearing. How do 1 prove it?
SHERYL. You approximate it with regular figures. Like a cube.
SERGE LANG. That's right! Perfect. You guys are very good, you
know. 1 make a grid inside my 3-dimensional figure, a grid of rectangular
boxes. Then 1 make a shearing. Then each rectangle goes into-what? A
3-dimensional parallelogram. It's called a parallelepiped. You know the
word? If 1 shear a rectangular box, then this thing is called a parallelepiped. So if the volume of a rectangular box does not change under
shearing, then the volume of an arbitrary solid will not change under
shearing. And the proof is as Sheryl said. You approxima te the volume by
rectangular boxes. The theorem is true for boxes, that the volume does
not change under shearing, so it's true by approximation for any solid.
So now we have the theorem: volume does not change onder shearing.
And we can do what Serge wanted to do at the beginning, to deal with
slanted pyramids. We have now proved that the volume of a slanted
pyramid is the same as the volume of the straight pyramid, one third the
47
base times the height. If I have a pyramid with square base, or rectangular base, and the base is B, then
Volume of slanted pyramid
1
-abh
3
JBh.
T
h
_j
b
SHERYL. What about a triangular base pyramid?

SERGE LANG. Very good! But why stop at triangles? In fact why not a
curved base pyramid, a pyramid with any base? The pyramid is not the
only thing I want. You see the cone here, with a circle base?
It has a base B and a certain height h. What's the volume of the cone?
Who can guess? [Severa! hands go up.]
CHARLIE. It will be '1Tr 2, times the height, times [pause] a third?
SERGE LANG. Yes! You got it just right. Perfect.
The volume of the cone is
[To another student.] What would you have said?

ANOTHER STUDENT. Same thing.
SERGE LANG. And Mike back there?
MIKE. Same thing.
7Tr 2h.
48
SERGE LANG. That's right. That's the theorem. And suppose you had
an arbitrary base, with a kidney, and 1 form the cone, like this.
Vertex
Volume=- Bh
3
1 can make the base as cockeyed as you want. 1 take an arbitrary base.
Sometimes it might be square, sometimes it might be curved, or whatever.
Then 1 can form the cone over this base. 1 pick one point somewhere in
space, and 1 join the point with ali points of the base. This gives me the
cone over the arbitrary base. Cali the area of the base B. Let h be the perpendicular height. What is the volume of this cone? What's the volume?
SHERYL. One third Bh.
SERGE LANG. Yes, that's the theorem:
For any cone such that area of the base is B and the perpendicular
height is h, the volume is
Bh .
How do we prove it?

SERGE. Start with the simplest case?
SERGE LANG. The simplest case was that of a pyramid with square
base, or rectangular base, and a given height. Now suppose you have an
arbitrary curved base. How can you deal with an arbitrary curved base?
What do 1 do?
SERGE. Oh, 1 see. You try to fit other figures by means of a grid.
SHERYL. You approximate the base with a grid.
SERGE LANG. 1 put a rectangular grid on the base, that's right. Then 1
start drawing ali these lines
49
Vertex
RACHEL. You approximate the volume by approximating the base
with rectangles. Then you approximate the cone by pyramids. Then you
can put it together to get the volume.
SERGE LANG. That's right. And 1 guess Sheryl would have said the
sa me thing. [Sheryl nods.] And 1 guess Michael back the re? [Laughter.]
You see, for each rectangle of the grid, you can form the pyramid with
this rectangle as the base, with the same vertex as the cone. You do this
by drawing ali the lines from the vertex to the points of the grid. Then
you form ali slanted pyramids with the vertex. Then you know that for
each of these slanted pyramids, the volume is one third the base times the
perpendicular height. So if you take their sum, and use the theorem for
each of the slanted pyramids, you get the theorem for the whole cone,
with an arbitrary base. So that's how we get
Bh.
Now if the base is a circle of radius r, then the area of the base is '!Tr 2.
That's how we get
'1Tr 2h for the volume of the ordinary co ne. 1t's just a
very special case of the general theorem with an arbitrary base.
Comments
The ready acceptance of n dimensions by students at an early age is a
common experience for me. Given the history of "dimensions", and ali
the fuss made around Einstein's "fourth dimension" as time, this might be
surprising to sorne people. But on the basis of experience, 1 find that
young people will find problems about higher dimensions only if grown
ups suggest the existence of problems.
50
The question: "Is time the fourth dimension?" is a very bad question,
because it already prejudices what one might mean by "dimension", and
also the use of the definite article "the" implies that if there is a "fourth
dimension" there is only one. But today it is generally accepted that
whenever you can associate a number with a notion, then you have a
dimension. This idea was already seen clearly by d'Alembert, when he
wrote the article on "dimension" for Diderot's enclopoedia:
Cette manire de considrer les quantits de plus de trois dimensions est

aussi exacte que l'autre; car les lettres algbriques peuvent toujours tre
regardes comme reprsentant des nombres, rationnels ou non. J'ai dit
plus haut qu'il n'tait pas possible de concevoir plus de trois dimensions.
Un homme d'esprit de ma connaissance croit qu'on pourrait cependant
regarder la dure comme une quatrime dimension et que le produit du
temps par la solidit serait en quelque manire un produit de quatre
dimensions. Cette ide peut tre conteste mais elle a, ce me semble,
quelque mrite, quand ce ne serait que celui de la nouveaut.
Which, translated in English, reads:
This way of considering quantifies in more than three dimensions is just

as exact as the other one; because algebraic letters can a/ways be
regarded as representing numbers, rational or not. 1 said above that it was
not possible to conceive more than three dimensions. A elever gentleman
with whom 1 am acquainted believes that nevertheless, one could regard
duration as a fourth dimension, and that the product of time by solidity
would be in sorne way a product of four dimensions. This idea can
perhaps be challenged, but it has, it seems to me, sorne merit, were it on/y
that of being new.
Of course, one of his friends is presumably himself, but he is prudent,
considering what must have been at the time a rather far out idea. Today,
the idea is current. In physics, you have dimensions besides the three spatial dimensions and time, for instance speed, acceleration, curvature, etc.
In economies, you can associate a number, say the total gross profits of
sorne company in any given year, so you have a dimension for, say, the
steel industry, oil industry, fisheries, agriculture, automobile industry, and
whatnot. In that way you can have as many dimensions as you can think
of.
When 1 brought up higher dimensions with the students during my
talk, there was absolutely no resistance from anyone. The students saw
right away that the important thing is how we operate with numbers in
the context of several dimensions. Then they have the correct idea concerning this abstraction, and give the righ t answers. Sometimes in similar
circumstances, 1 do get the question whether a fourth dimension exists. 1
answer it as above, painting out that it ali depends on how you wish to
use the word. If by "dimension" you want to mean only the spatial
51
dimensions of ordinary space around you, then there are only three dimensions. If you accept to give the word a wider meaning, th en you have
arbitrarily many dimensions, with which you can work just as easily. The
automatic response that in n dimensions under a dilation by a factor of r
volume changes by a factor rn is then both immediate and correct. Of
course, when 1 ask how volume changes under dilations in four dimensions, Serge answers: "lt's rst whatever." He shows that he has weil understood what's going on. The only thing that would remain to be done is to
point out that his "whatever" corresponds to a choice of letters, and that
if we continue to use letters for dimensions, we shall run out of choices
after the 26 letters of the alphabet have been exhausted. Therefore, one
may denote the numbers associated with three dimensions by something
like x 1, x 2, x 3 ; and th en there is no difficulty to denote the numbers associated with n dimensions by x 1 , x 2 , ,xn. Then if we dilate by factors
of r 1, r 2, . , rn one sees at once th at volume of boxes changes by a factor of r 1r2 rn (the product). Whatever difficulty exists here (and it is
very slight) lies only in an appropriate choice of letters and indices; that
is, in the choice of notation to transcribe in symbols what the mind has
already grasped.
Concerning the definition of an ellipse, 1 asked the class if they knew
about coordinates, and they did not. To a slightly older class, 1 would
expand a little as follows. Traditionally, ellipses, parabolas, and other
standard curves are defined by what 1 regard as complicated properties,
for instance: the ellipse is the locus of points such that the sum of the distances from two given points is constant. 1 object very much to this traditional approach, which makes it difficult first to see the standard equation
exhibiting the fact that the ellipse is a dilated circle; and second, makes it
difficult to see that the area is '!Tab, which is immediate from the definition
by means of the notion of dilation, and the technique of approximation
by rectangles.
The volume of the bali
This lecture was given in two hours, one before lunch and one after
lunch, to a class of students in Paris at about the 9th grade leve!, so 14
years old. They had not yet covered Pythagoras' theorem in class, for
instance.
SERGE LANG. What do you do in this class? Geometry? Algebra?

THE CLASS. Yes, algebra.
SERGE LANG. And geometry?
THE CLASS. Yes, a little.
SERGE LANG. So you know sorne geometry?
THE CLASS. We are just beginning.
SERGE LANG. [Pointing to a student.] You know what
'lT
is?
STUDENT. Yes.
SERGE LANG. So, what is it?
STUDENT. It's a number ... geometrie.
SERGE LANG. Oh yes? Which one?
STUDENT. 3.14 and so on.
SERGE LANG. 3.14 and so on. And what does
'lT
represent?
STUDENT. ???
SERGE LANG. OK, who knows what
'lT
represents? You, what's your
na me?
STUDENT. Christopher.
SERGE LANG. Ali right, Christopher. So?
NATHALIE. The circumference of a circle.
SERGE LANG. So what is the circumference of a circle?
NATHALIE. What do you mean, what is it?
SERGE LANG. 1 mean, as a function of 'lT.
NATHALIE. ???
THE VOLUME OF THE BALL
53
SERGE LANG. Weil, you have a circle of radius r ...

A STUDENT. Ah! lt's 2'1Tr.
SERGE LANG. Very good. 2'1Tr, where r is the radius. And the surface,
the area, what is it?

THE STUDENT. lt's
'lTr
squared.
SERGE LANG. That's right. So we have two formulas, and as Christopher said, 'lT is equal to 3.14 .... Good, 1 am not going to do this, but 1
want to do the same thing in one higher dimension. If we go to a higher

dimension ... you see, before we had the circumference and the area, but
if we have one more dimension, what do we get?
THE STUDENT. A bali.
SERGE LANG. Right, a bali. 1 draw it.
And what do we want to find about the bali?

THE STUDENT. The volume.
SERGE LANG. That's right, the volume, What's your name?
STUDENT. Ann.
SERGE LANG. You know the formula for the volume of the bali of
radius r?
ANN. ???
SERGE LANG. Who knows the formula?
THE CLASS. ???
SERGE LANG. Ali right. The length of the circle is 2'1Tr and the area is
'1Tr
But for the bali, what is it going to be?
A STUDENT.
'lT
squared times
to the power 4.
SERGE LANG. No. So let's go back to boxes. Suppose 1 have a square
with sides equal to r. Then its area is equal to what?

THE CLASS. r squared.
SERGE LANG. Now, suppose 1 have a cube. Its volume is what?
54
THE CLASS. r cube.

SERGE LANG. Have you done sorne dilations?
THE CLASS. A little.
SERGE LANG. Good. Let's take a cube with side a. Then the volume is
a . If I make a dilation by a factor of r in ali directions, then the volume
3
becomes what? Ann.

ANN. ra quantity cubed.
SERGE LANG. That's right, (ra) 3. In other words, r 3a 3. OK? Everybody
understood? So we can say that if I make dilation by a factor of r, then

the volume changes, how? By a factor of r 3 OK? Good. Now let's go on
to the sphere. The volume is going to be what?
SOPHIE.
'lT
r cube.
SERGE LANG. '!Tr 3-it's not at ali stupid to say th at. And you, what do
you say?
ANN.
'lT
cube r cube.
SERGE LANG. Ah! It's not so clear, heh? But there must be the factor
of r 3 no matter what, because of the dilation. So the answer is that there is
a '!Tr 3, but with a constant in front. And the constant is equal to 4/3.
That's the volume of the bali:
==
4
3
-'!Tr .
Do you know the volume of other geometrie figures? Or do you just know
surfaces? Is this the first volume that you have ever seen?
THE CLASS. No, there is the cube.
SERGE LANG. Ail right. So what I want to do today is to derive the
volume of the bali: i'1Tr 3 1 want to prove it. How are we going to prove
3
it? Nobody knows? First we are going to find the volume of the bali of
radius l. It's simpler. What is the volume of the bali of radius l? Ann.
ANN. ???
SERGE LANG. If I let r
ANN.
== l in the formula, what do I get?
'1T.
SERGE LANG. That's right, : '1T. We want to prove that the volume of
the bali of radius l is
i'lT.
And how do we do it?
THE CLASS. ???

SERGE LANG. Weil, we have to start from things we know, for instance
cylinders, or boxes.
55
h
1
'----------- ----
//
/
/
The volume of a rectangular box of base B and height h is equal to what?

SOPHIE. The base times the height.
SERGE LANG. That's right, B times h. Here 1 took a base which was a
rectangle. But suppose 1 took a curved base? Any base, like this for
instance.
-...
THE CLASS. lt's the same thing.

SERGE LANG. Yes, it's the same thing, it's always the base times the
height. And if 1 have a real cylinder, whose base is a dise of radius r, and
height h? You. [Pointing to a student.]
VERONIQUE. ???
SERGE LANG. The area of the base is what?
VERONIQUE. ???
SERGE LANG. It's a dise, the base is a .dise. The area of a dise is what?
RIVKA. wr 2.
SERGE LANG. And the volume of the cylinder, it's equal to what?
THE CLASS. wr 2h.
SERGE LANG. That's right:
Volume of cylinder of radius r and height h == wr 2h.

OK? If you know the area of the dise, wr 2 , then the volume of the
cylinder is wr 2h. Everybody agrees? Good. Now we come to the bali.
What do we do? We have to eut up the bali into pieces that we know, and
for the moment, what we know are cylinders.
56
SOPHIE. Cut it in half.

SERGE LANG. If we eut the bali in half, we get something like this .
., -- ...... --------------- - .....
SOPHIE. We could eut it up in sliees, the base would be a dise.

SERGE LANG. The base would be a dise, and you know the area of the
base, wr 2. That's right. But there still remains the thing on top, whieh is
not quite a eylinder. So what do we do?
ANN. Cut it up into dises.
SERGE LANG. You are really very good. [Laughter.] That's right, we
eut it up into sliees, like this:
Good. Do you know Pythagoras' theorem?

THE CLASS. No.
SERGE LANG. 1 have a right triangle, like this:
57
One side is a, another side is b, and the third side, the hypotenuse, is c.
You don't know the relation between a,b, and c? You have never seen it?
We have
You've never seen this?

THE TEACHER. Next year.
A STUDENT (VANYA). Why the squares?
SERGE LANG. It's not at ali obvious why you need the squares. One
has to prove it, but it's not what 1 wanted to do today. So 1 am asking you
to accept i t. Can you a cee pt it?
THE CLASS. Yes, OK.
SERGE LANG. Good. Now 1 eut up the bali and 1 make cylindrical
slices, like this.
Actually, 1 took a half bali, and 1 eut it up. Does everybody see the slices?
If we look at them sideways, they look like rectangles. But if 1 turn them
around the vertical axis, then 1 get cylinders. And 1 know the volumes of
these cylinders once 1 know their radius and their height. 1t's what?
STUDENTS. wr 2h.
SERGE LANG. OK. Now 1 need to know the radius and the heights.
Weil, the heights will depend on the number of cylinders. 1 can have 5, or
6, or more. Let's draw the pi ct ure wi th six.
4/6
3/6
1--------2/6
~--------------
1/6
+---------------\
1
~----------------~
58
If I call the origin 0, then the first point is at what height?

THE CLASS. 116.
SERGE LANG. And the next?
THE CLASS. [Ali together.] 2/6.
THE CLASS. 316.
SERGE LANG. That's right. And the next ones have height 416, 516
and 616.
THE CLASS. Yes.
SERGE LANG. And if I had seven slices? What would be the height of
the first point?

THE CLASS. 117.
SERGE LANG. And the others?
THE CLASS. 217, 317, up to 717.
SERGE LANG. Yes. And if I want to do this in general? What do I call
it?
NATHALIE. 1ln.
SERGE LANG. 1ln? You got it! In the general case, we take 1ln. The
first point will be where?

THE CLASS. At 1ln.
THE CLASS. At 21n.
THE CLASS. At 31n.
SERGE LANG. And the last one, all the way at the end?
THE CLASS. nin.
SERGE LANG. That's right, that's what happens in general. And what
is the volume of each cylinder going to be? The height is equal to what?
A STUDENT. It's h.
SERGE LANG. Right, it's h, but h is equal to what?
THECLASS....
SERGE LANG. When I eut up the ball into six slices, what is h?
VANYA. It depends how you eut it up.
SERGE LANG. Yes, but the cylinders all have the same height. Do you
see that?
VANYA. Yes.
59
SERGE LANG. What is the height, if 1 have six slices?

VANYA. 1/6.
SERGE LANG. Yes, when 1 have six slices, then the first height is 1/6;
and the second height is also 1/6. And the others also. OK?
CLASS. Yes.
SERGE LANG. Ail right, then in the case of six slices, ali the heights
are equal to 1/6. In the case of n slices, these heights are equal to what?
THE CLASS. [Ali together.] 1 over n.
SERGE LANG. Very good, 1ln. Now the radius, what is the radius
equal to? The radius of the first slice?

THE CLASS. r.
SERGE LANG. Yes, it's r, but ris equal to what? OK, 1 draw it with six.
There is a first r, then a second r, and so forth until the last one.
The radii change, don't they? Then the first radius is equal to what? lt's
the radius of the big circle, the radius of the bali, which 1 take to be equal
to 1. Remember, we are trying to figure out the volume of the bali of
radius 1.
And the radius of the first cylinder, it's OA 1 It's equal to what?
THE CLASS. ???
SERGE LANG. We don't quite know, it's smaller than 1. Agreed?
THE CLASS. Yes.
SERGE LANG. Weil, here is where we need Pythagoras' theorem. Let's
draw the first cylinder.
~b
1/6
0
~---r 1 ---~A 1
60
We have a right triangle OA I BI in the picture. And in a right triangle, we

have agreed before that
In the right triangle OA 1BI, what is a?

THE CLASS. It's the radius of the cylinder.
SERGE LANG. That's right, and the radius of the cylinder is r 1 And
what is b?
A STUDENT (PHILIP). lt's the height.
SERGE LANG. And the height is equal to what?
PHILIP. 1/6.
SERGE LANG. Right. And c is equal to what?
PHILIP. 1t's the radius of the bali.
SERGE LANG. Which is equal to what?
PHILIP. Euh ... 1.
SERGE LANG. Exactly. Then we have
ri+
(1/6) == 1
Do you agree with this?

PHILIP. Yes.
SERGE LANG. Then
ri is equal to what?
PHILIP. 1 - ( __!_ ) 2 ,
6
SERGE LANG. That's right, very good. Now let's look at the second
cylinder. Let me draw it.
b
a
1 have a new triangle OA 2 B 2 What is a in this triangle?
61
PHILIP. 1t's the radius of the cylinder.

SERGE LANG. Yes, the radius of the second cylinder, r2 . And b?
PHILIP. lt's A 2 B 2 .
SERGE LANG. Which is equal to ... ?
PHILIP. 2/6.
SERGE LANG. Very good. And
ANOTHER STUDENT. Th en
cis still equal to 1. Th en wh at is ri?
ri will be equal to 1
(2 /6) 2 ?
SERGE LANG. There you are. And the next one will be r3 , and we
have
rj ==
THE STUDENT.
rj =
1- ( ~
SERGE LANG. That's right. OK? We can go on? We go on like this

with r4 , r5, r6 . 1 write it down [and the students repeat out loud]:
and r 6 == 1 - (
)2 .
Now we have sorne idea about the radii of the cylinders. And the
height of these cylinders, what is it?
A STUDENT (KARIN). lt depends. 1/6, 2/6 ...
SERGE LANG. No, that was for the triangles, when we computed the
radii. But the cylinders ali have the same height. And it's equal to what?
THE STUDENT. 1/6.
62
SERGE LANG. That's right, very good, 1 /6. And the radii of the
cylinders depend on the cylinders. The rad ii are going to be different. The
first radius is r 1 The second radius is r 2 . The third is r3 . They become
smaller and smaller. OK? Now, the volume of the cylinder is what?
A STUDENT (RAJNA). The base times the height.
SERGE LANG. Yes, it's wr 2h. The radius of the first cylinder is r 1 And
the volume of the first cylinder is ...

SOME STUDENTS. [Ali together]. wr~ times 1/6.
SERGE LANG. And the volume of the second cylinder is what?
THE STUDENTS. wri times 1/6.
OTHERS. wri times 2/6.
SERGE LANG. Who said 1 /6?
[Sorne hands go up.]

SERGE LANG. Who said 2/6?
[Other hands go up.]

SERGE LANG. No, no! The height of the second cylinder is still 1 /6.
OK? Then the volume of the second cylinder is what? lt is
==
1
6
'!T'f2 - .
And the volume of the third cylinder?

RAJNA. It's
'lTri
L6
SERGE LANG. That's right. And the next, Ann?
2 .
A NN. wr 4 times -.
SERGE LANG. Yes. But the squares of the radii r 1, r2 , we have already
found them a minute ago. What are these squares equal to?
NORA. 1 -
( 1 )2, 1 - ( 2 )2, ...
SERGE LANG. So what do 1 get when 1 take the sum of the volumes of
ali these cylinders? You, what's your name?

A STUDENT. Siem Lang.
SERGE LANG. Gee, like me! [Laughter.] OK, so 1 take the sum [and
Serge Lang writes as he questions the students ]. What's the first one, Siem
Lang?
2
SIEM LANG. wr 1 -
6"
SERGE LANG. And r~ is equal to what?
63
2
( __!_
6) .
SIEM LANG. l -
SERGE LANG. OK. Therefore the volume of the first cylinder is
What about the second?

SIEM LANG.
'TT
(1 - (
~ )2)
SERGE LANG. And the third?
( 2_ )2) X __!_.
SIEM LANG. 'lT ( l -
SERGE LANG. Good. Then we go on, and we find that the sum of the
volumes is equal to:

sum of the volumes of the cylinders ==
l 2
l
'lT( l - (-) ) . -
2 2
l
'lT( l - (-) ) . -
6 2
l
'lT( l - (-) ) . - .
6
6
Now, suppose 1 took seven cylinders instead of six. Then what would be
the sum? You, what's your name?
A STUDENT. Cyrille.
SERGE LANG. OK, Cyrille, what's the sum?
CYRILLE. You must replace the six by seven. lt's
SERGE LANG. Very good. And it will continue up to what?

CYRILLE. 'TT( l -
~ )2) ~.
SERGE LANG. Now, we could take 6, 7, 8, 9, 10, ... What if 1 just take
n, wh at do 1 get for the volume of the first cylinder?
( __!_ ) 2) __!_.
n
n
SERGE LANG. Fantastic! And the next? Someone else.
CYRILLE. 'lT( l -
(2) 2) __!__
n
n
SERGE LANG. Yes. it goes on like this. Wh at about the last? [Pointing.]
What's your name?
A STUDENT. 'lT(l -
64
(-) 2) -.
n
n
SERGE LANG. That's right. So the sum of the volumes of ali the
cylinders is:
THE STUDENT. Kalasa. '1T(l -
sum of volumes
Now, when n gets bigger and bigger, what happens to this sum? lt gets
close to wh at?
THE CLASS. ???
SERGE LANG. 1 take more and more cylinders, like this.
What does the sum of the volumes of the cylinders approach?

PHILIP. The volume of the bali.
SERGE LANG. [ Writing:]
When n becomes larger and larger, the sum of the volumes of the
cylinders approaches the volume of the half bali of radius 1.
So now, the problem becomes: what happens to the expression on the
right when n becomes bigger and bigger?
ANN. You can compute it.
SERGE LANG. You want to compute it directly? OK, but how?
ANN. Compute the volumes one after the other.
SERGE LANG. Yes, it works if you have any specifie number, like six,
or ten. But what about n? When n becomes bigger and bigger? You don't
know any specifie value for n. So what we now have to do is to figure out
what happens to the sum when n gets bigger and bigger. But we'li do this
after lunch. See you this afternoon.
65
The second hour

[The class returns after lunch.]
SERGE LANG. So we're back, and the problem was to find the volume
of the ball of radius l. We had derived the expression for the volume of
the half baH, which is approached by
when n becomes bigger and bigger. We can already factor out

l so we can write
. t h e expression
.
r.actor out -,
as
n
'Il',
and also
Now let's look at the big parenthesis. We see sorne one's, l, then l, then
1, ... How many one's are the re?
THE CLASS. n.
SERGE LANG. Yes. So 1 get
'Il'
- (n - something to be subtracted).
n
What do we subtract?
RAJNA. (l !n) 2.
SERGE LANG. That's right. And after that?
RAJNA. (2/n) 2.
SERGE LANG. Very good. And the next, it's what? Christopher?
CHRISTOPHER. (3 ln )2.
SERGE LANG. Right, up to (n!n) 2. So we subtract
And if we write it all down together, then what's the full expression,
Christopher?
CHRISTOPHER.
n
n
'!1'-
l 2 + (-)
2 2 + ... + (-)
n 2] .
- -'Il' [ (-)
n
66
SERGE LANG. But w n ln is equal to wh at?

CHRISTOPHER. w.
SERGE LANG. Yes. Now subtract ...
'!1'
NATHALIE. - ... euh ...

n
SERGE LANG. Yes, go on.
NATHALIE. Times (l /n) 2
+ (2/n) 2 + (3/n) 2 +
SERGE LANG. Very good, up to (n!n) 2. So the formula now becomes
l2
'!1'- '!1' ( - )
22 . -l + (-)
32 . -l + ... + (-)
n2
+ (-)
. -l] '
and this expression approaches the volume of the half bali when n gets
bigger and bigger. Now, if the formula 1 gave at the beginning is correct,
then what does the expression in parentheses approach?
THE CLASS. ???
SERGE LANG. The volume of the half bali of radius 1 is supposed to
be equal to }:_'TT. Here we have the expression

3
[
l2l
n n
'!1'- '!1' ( - )
n2l]
+ ... + (-)
- .
If this expression approaches two thirds of w, then what does the sum in
the parentheses approach? Do you see it? 1 have a w, and 1 subtract a certain fraction of w. 1 want what remains to be two thirds of w. Then the
fraction is what?
A STUDENT. l /3.
SERGE LANG. That's it! One third. What's your name? Elizabeth? She
got it, Elizabeth got it. The expression in parentheses approaches l /3. It's
very good. And that's what we have to show. And if we succeed, then we
win. OK, let's write this down. Up to now, we have shown that
the sum of the volumes of the cylinders is equal to
and now, we want to prove that the sum
12 1
[(-)
n
22 1
n2
1]
+ (-)
+
(-)
approaches
n
n
n
n
- .
3
67
At first sight, it looks pretty tough. There are sorne squares, lots of n, and
it is not at ali clear how the sum behaves. So what do we do?
THE CLASS. ???
SERGE LANG. Suppose there were no squares, do you think you might
be able to figure it out? OK, we found an expression with squares, and we

don't know what to do. The problem looks difficult, and we try to consider a similar, but simpler problem. So let's try it without the squares.
Suppose 1 write the sum without the squares. Then it's like what? Christopher?
CHRISTOPHER.
--+--+--+
... +-n n
n n
n n
n n
SERGE LANG. That's right. Now we have a sum of fractions, whose
denominator is ... Elizabeth?
ELIZABETH. n n.
SERGE LANG. Yes, you can also say n square, but you are right. We
can leave it as n times n. And the numerators?
ELIZABETH. l, 2, 3, 4, ... ,n.
SERGE LANG. Yes. And how do 1 figure the sum out? Let's draw a pic-
ture. l'Il do it like this. 1 take a square of si des l, and 1 eut it up in pieces,
1 eut it up into little squares whose sides are l ln.
~4-~~--+-~-+--~~~~--+-~
1
n~4-~~--+--~-+--r-~-r~--+-,
1
n~4-~~--+-~-+--~+-~~--~~
n~~~~--._~_.--~~~~--~~
0n
- ..!.. ........................ n n
n
Take the point 0 in the lower left hand corner. Then if 1 go to the right,
get points l ln, 2 ln, 3 ln, and the last one is wh at ... Christopher?
68
CHRISTOPHER. n ln.
SERGE LANG. Good. And the next to the last?
NATHALIE. (n-l)ln.
SERGE LANG. Right, and the one before that?
ALL TOGETHER. (n-2)1n.
0
n
2
n
n- 1
n
n
n
SERGE LANG. Yes, and 1 do the same thing vertically, up and down.
Now, 1 want to find
l l
2
+ ... +
--+-n n
n n
Take
n-I
n
n
n
+ -n
_!_ _!_, What does it look like, on the picture?

n
THE CLASS. ???

SERGE LANG. lt's l ln times l ln, which is what? It is the area of a
square whose sides are l ln. So 1 draw this little square inside the big
square, let's draw it in the upper left band corner.
1/n
1/n
What about the next one? lt's 21n times l ln. This represents what?
69
A STUDENT. It's two little squares.

SERGE LANG. That's right, it's a rectangle of length 21n and of width
l ln. 1 draw it just below the first square.

1ln
1ln
11n
11n
What about 31n times l ln?

THE CLASS. It's a rectangle of length 3 ln.
SERGE LANG. And the term with (n- l) ln?
THE CLASS. It's a rectangle of length (n- l)ln.
SERGE LANG. And the last one?
THE CLASS. 1t's a rectangle of length n 1n.
SERGE LANG. Right, of length l, like this, ali the way at the bottom.
So the pi ct ure is like a staircase:
........
..........
.. .. .. .. .. .........
.....
.. ..
1
................................
. . . . . . . . . . . . . . . . . . . . ... . . . . .. . .. . . .
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
And the sum is just the area of the staircase. Now if 1 increase n, if n gets
bigger and bigger, what does the staircase look like?
VANYA. It will have more steps.
SERGE LANG. Precisely, it will have more steps. And the area of the
staircase approaches what? When n gets bigger and bigger?

THE STUDENT. It approaches half the square.
70
THE STUDENT. Rivka.

SERGE LANG. You got it, Rivka got it! Let me write it down:
The area of the staircase approaches 1/2 when n bec ornes large.
Do you ali agree with Rivka? Everything OK? Good. So we started with a
sum having squares, and then we considered a simpler sum without the
squares. And now we find that this simpler sum is a staircase, and we
showed that when n becomes bigger and bigger then the sum approaches
l 12. That's terrifie. You are really very good. [Laughter.]
Now that we know how to do this, we can go back to the sum at the
beginning, with the squares. OK? [The class approves.] Here is the sum
a gain.
Wh at do we do? [Questioning looks.]

Ali right, take one term in the sum, for instance (
2n )2 _!_.
n
Wh at is it?
What does it represent? It's a volume, isn't it? If 1 multiply the square of
something by something else, 1 get the volume of a box with one side
equal to l ln and the other two sides equal to 21n. Do you agree? OK.
Now how many boxes are there?
A STUDENT (STEPHANIE). n.
SERGE LANG. Right, n. Before we had n rectangles, and we drew
them inside a big square. Now we have n boxes. We draw them
in ... what?
STEPHANIE. A big cube with little cubes.
SERGE LANG. [Appreciatingly] Ah! You got the idea. That's right, so 1
draw a big cube with si des l, and 1 slice it up to get little cubes of sides
l ln.
j
1/
1/
//
/ /
/ /
1.,.
,'
-~-
-- --
v
vv
11
1
~
~
~ ~ Il
/
/
// //
/ /
/
L I
/
f
1
J
v
~
~ v ~1
v J
~
v
~
v ~ Il
~ v ~
~
v )
v
v~
v
,.,
-v
~
llv ~
1
Il
1
Il
1
71
So here is the big cube. Now the first term in my sum is ( _!_ ) 2 _!_, and it
represents what? lt's the volume of a cube ali of whose sides are equal to
1ln. 1 draw it just like the square before.
1/n
,L=7J
U!n
1/n
And the next one, ( 3_ )2 _!_, it's what? lt's not a cube.
PHILIP. lt's a flat box.

SERGE LANG. That's right, it's a flat box, with two sides equal to 21n,
and with height 1ln. So 1 draw this flat box just below the first one.
1/n
2/n
And the third, ( ]__ ) 2 _!_? Elisa beth?
ELISABETH. A flat box of sides 31n.

SERGE LANG. That's right. The height is still 1ln, and the length of
the sides is 3 ln. And we go on like this up ton ln.
72
So in the sum with the squares, we also get a staircase, but these are real
stairs, with a volume; they are three dimensional stairs, like those in real
life. Now, when n becomes bigger and bigger, the stairs approach what?
A STUDENT. 1/3.
ANOTHER. 1/2.
SERGE LANG. Any other suggestions?
SOPHIE. A pyramid.
SERGE LANG. Ah, there you are! That's what 1 wanted to hear. When
n becomes large, the stairs approach a pyramid. You agree with this? And
how do we get this pyramid? By joining one corner, the vertex, ali the
way up on the left, with the opposite face on the bottom. How many faces
are there which are opposite to the upper left hand corner?
RIVKA. Three?
SERGE LANG. Yes, there are three opposite faces. This means that 1
could have three pyramids like the one we just drew for the staircase. And
these three pyramids fill out the cube. Therefore, the volume of one of the
pyramids is equal to what?
VANYA. 2/3.
SERGE LANG. No. How many pyramids are then in the cube?
VANYA. Three.
SERGE LANG. And the volume of the cube is ... ?
VANYA. 1.
SERGE LANG. Therefore the volume of the pyramid is ... ?
VANYA. 1/3.
SERGE LANG. That's right, 1 /3. Let's go back to the stairs with steps
of height 1ln. When n becomes large, then the stairs approach the
pyramid, OK? So 1 write it down:
When n becomes large, the volume of the stairs approaches the volume
of the pyramid, which is equal to 1/3.
But we have won! That's exactly what we wanted to prove. Do you
remember what we started from? We started with the sum of the volumes
of the cylinders, and we found the expression:
'TT -
'TT [ ( - )
2 2 1
3 2 1
n-1 2 . -1 + (-)
n 2 . -1] '
+ (-)
. - + (-)
. - + . . . + (--)
n
with a sum containing lots of squares. To figure out what this sum is like,
we first tried a similar sum without the squares, and we found a fiat
73
staircase, with an area which approaches 1/2 when n becomes large. Now,
for the sum with the squares, we find that it approaches three-dimensional
stairs, and that it approaches 1/3. The sum is the volume of a staircase,
which approaches 1/3 when n becomes large. Therefore the sum of the
volumes of cylinders approaches
'!1'-
And this is equal to what?

ELISABETH. 2!!_.
3
SERGE LANG. So what we have just proved is that the volume of the
half bali is equal to what? Elisabeth?
ELISABETH.
2'!1'
/3.
SERGE LANG. And therefore the volume of the whole bali is ... ?
ELISABETH. 4'!1' /3.
SERGE LANG. And that's the theorem we wanted to prove. We won!
Theorem.
The volume of the bali of radius 1 is equal to 4'17' /3.
By the way, we also found out something about those sums and what they
approach when n becomes large.
The first one, wi tho ut the squares, is
2
+ n
n
+ n
n
+ ... + n-1
n
+ n
and it approaches 1/2.

The second one, with the squares, is
12
22
3 2
(-) - + ( - ) - + ( - ) - +
n
n
n
n
n
n
n-1
n
n
n
+ ( - ) 2 . - + (-) 2 . n
and it approaches 1/3.

What's the next sum after that?
CHRISTOPHER. With cubes,
13
23
33
(-) - + ( - ) - + ( - ) - +
n
n
n
n
n
n
n-1 3 1
n 3 1
+(-) -+(-) n
n
n
n
approaches 114.
SERGE LANG. Who says 1/4? Ann, what do you think?
74
ANN. Nothing, 1 don't know.

SERGE LANG. Elisabeth? Siem Lang?
SIEM LANG. 1/4.
SERGE LANG. We proved it for the first two sums. But this third one,
we have not proved it. And the next one, what is it going to be?
ELISABETH.
14 1
24 1
34 1
(-) - + ( - ) - + ( - ) - +
n
n
n
n
n
n
n-1
n
1
n
n
n
1
n
+ ( - )4 . - + (-)4 . -
approaches 1/5.
SERGE LANG. Terrifie. That's right! lt's a theorem. But how are we
going to prove i t?
PHILIP. There are no dimensions which allow us to expand ( _!_ )3 _!_
to fi nd l 14.
n
n
SERGE LANG. That's very good. You have understood what 1 did very
weil. So we may be stuck. Up to three dimensions, 1 managed by drawing

pictures and stairs, but if 1 want to handle the next case ...
A STUDENT. Weil, you could imagine ...
SERGE LANG. That's right, you can imagine it! And you shou1d. But
it's not quite as safe, one feels less secure. lt raises sorne problems, and
actually one has to use another method because our intuition in higher
dimensions ... is more delicate. For the sum with cubes, we would have
to work in how many dimensions?
PHILIP. Four.
SERGE LANG. Right. And five for the next. But now, it's not so sure
any more that we are doing things right, the intuition may break down.
So one has to give algebraic arguments to complement the geometrie
intuition of higher dimensions. One can do it, but 1 won't do it today.
Anyhow, it's true.
The method which 1 used today, which is to eut up the bali into
cylinders and to take the sum, do you know who used it for the first
time? lt was Archimedes. Have you beard of Archimedes?
Christopher ... Right, that's the guy who shouts "Eureka" when he
jumps naked out of his bathtub. You beard about this? He was very
elever. And it was he who invented this kind of method. And 1 copied
Archimedes, approximately. So let me summarize:
Volume of bali of radius l is equal to
~ 'TT.
And the volume of the bali of radius r? Christopher? With dilations?
75
PHILIP.
The volume of the bali of radius ris equal to
~ '1Tr 3
SERGE LANG. Yes, you catch the r 3 by dilations, that's right. You are
really very good in this class; you are very much on the bali. 1 have rarely
seen a class which does as weil. You are ali quite smart.
Do you have any questions? On mathematics, or anything else? [A
hand goes up.] Yes?
A STUDENT (CAROLINE). To compute
the surface of a sphere,
couldn't you turn the dise around?

SERGE LANG. Yes, you can. [Laughing.] Oh! Oh! Yes, yes, you can
but ... OK, let's try. To find the surface of the sphere by turning the dise
around the vertical axis.
gives
OK, take the circumference, or rather half the circumference, which is

'!l'r. Turn it around. Then what happens? You have to multiply '!l'r by
something?
THE STUDENT. By the length of the circle in the middle.
SERGE LANG. Yes, but then you will find 2'11' 2 r 2 .
76
THE STUDENTS. Oh!

SERGE LANG. And that would be true if ali the smali segments of the
half circumference turned around a circle of radius r. But you see, those
on top and on the bottom turn around circles which are smalier than that.
So the formula 2'11'2r 2 cannat be the right one. 1 agree that you get the surface of the sphere by turning the half circumference around, but then how
do get the right formula? Do you see the problem? There is a difficulty.
You have to take account of the fact that certain pieces turn around smali
circles.
You had a good idea, but you would have to use an "average" circle.
There is also a theorem for that, but it's too complicated to do it now. It's
not at ali clear what happens. Do you know the formula for the area of a
sphere? 1t is that:
Area of the sphere == 4'!1'r 2 .
So the formula says that it's the half circumference times 4r. There is no
other factor of 'Il'. And one can prove it by your method, but it's much
more difficult. That's why 1 prefer to do it another way, but this will have
to be done sorne other time.
Comments
1 was very pleased with the class, which was quite young for this topic. In
general, it is probably better to wait till the 1Oth or Il th grade to do the
proof, although a teacher can always try it out. Again as in other cases,
the geometry raises interesting problems, which lead into algebra. 1 find
these tapies the perfect setting to practice algebra, and to get students
acq uain ted wi th n.
The area of the sphere is the subject of the next lectures.
The length of the circle
The next lecture was given to an 8th grade class in Paris in Spring
1983. We did the length of the circle, starting with the formula for the
area. This method can be used instead of the method followed in
"What is pi?'' and has the advantage that it adapts easily to jind the
area of the sphere from the volume of the bali.
SERGE LANG. So, what do you do in this class?

STUDENTS. [Laughing.] Nothing!
SERGE LANG. Oh come on! You must do something! How far have
you got?
SOME STUDENTS. Factorizations, identities.
SERGE LANG. And in geometry? We are going to do sorne geometry
today. You know the circle, the dise?

A STUDENT. Yep.
SERGE LANG. You know the surface of a dise?
THE STUDENT. [Hesitating.] Yes.
SERGE LANG. What is it?
THE STUDENT. r squared times
'!1'.
SERGE LANG. That's right. The area of a dise of radius r is equal to

'!l'r
And the circumference of the circle?
THE STUDENT. lt's
'Il'
times the diameter.
SERGE LANG. And as a function of the radius?

THE STUDENT. 2'll'r.
SERGE LANG. Yes. And did you do the proof? How do you prove
these two formulas? [Questioning silence.]

STUDENTS. Euh ...
SERGE LANG. Weil, that's what 1 want to do, show you how you can
prove the second formula, starting from the first. In other words, let's suppose that we know the first formula, and we want to prove the second. Let
me write this down.
78
THE LENGTH OF THE CIRCLE
We suppose known that the area of the dise is 'll'r 2, and we want to prove
the second formula, that
c
2'll'r,
where c is the circumference of the circle.

So 1 have a eirele of radius r, and 1 also draw a eirele with a slightly
bigger radius, r plus something.
Let's eall this something h. Then the radius of the bigger eirele is r+ h.
In between the two eireles, 1 have a band. You see the band, 1 draw a
shading of the band like this.
What's the area of the band?

THE CLASS. lt's the area of the big dise minus the area of the small
dise.
SERGE LANG. Preeisely, the area of the big dise minus the area of the
small dise. But we know these areas, and 1 ean write down the formulas
for them. What's the area of the small dise?
A STUDENT. lt's '!l'r 2.
79
SERGE LANG. Right. And the area of the big dise? It's equal to what?
[Questioning looks.]
We know the formula for a dise of radius r, it's '!l'r 2 . If the dise has
radius 2, what is its area? It's '11'2 2, so 4'11'. OK? Now the area of a dise of
radius 3 is equal to what? You, what's your name?
THE STUDENT. Nancy. It's
'Il'
times 3 squared, so it's 9'!1'.
SERGE LANG. Right. And the area of a dise of radius 4, what is it?
Nancy.
NANCY. '!1'4 2, so 16'!1'.
SERGE LANG. OK, and the area of a dise of radius 10?
NANCY. '11'10 2.
SERGE LANG. And the area of a dise of radius r?
NANCY. 1rr 2.
SERGE LANG. And the area of a dise of radius r+h?
NANCY. [Hesitating] '11'(r+h) 2 ?
SERGE LANG. Yes, you got it, '11'(r+h) 2. Now we can write the area of
the band. It's equal to
Do you ali agree? [The students approve.] Now what do we do to this formula?
A STUDENT. We can work it out.
SERGE LANG. Yes, and then (r+h) 2, what do we do with it? [Turning
toward the teacher:] You have do ne this?

TEACHER. 1 should say so!
SERGE LANG. (a+b) 2 is equal to what?
STUDENTS, ALL TOGETHER: a 2
+ 2ab +
SERGE LANG. Wait, not ali together. You, what's your name?
STUDENT. Rita.
SERGE LANG. OK, Rita, (a+b) 2 is equal to what?
A STUDENT. It's an algebraic identity.
SERGE LANG. Who knows?
A BOY. lt's a 2
+ 2ab + b 2.
SERGE LANG. That's right, (a+b) 2
done this before, no?

STUDENTS. We did it with v and w.
== a 2 + 2ab + b 2. You must have
80
SERGE LANG. When 1 was a kid, we did it with a and b, but it's the
same thing. And it's again the same thing if we do it with r and h. So
(r + h )2 is equal to what?
STUDENTS. [Ali together]: r 2
SERGE LANG. Very good, r 2
+ 2rh + h 2 .
+ 2rh + h 2 And how do you prove it?
STUDENTS. Take the product.

SERGE LANG. Yes, let's do it in detail. (r+h) 2 it's just (r+h) times
(r+h). Now what do 1 do? [Ali students talk at once. Serge Lang writes:]
== r(r+ h) + h (r+ h).
(r+ h) (r+ h)
And after that, what do we do? You, what's your name?

A BOY. Christopher. We get r 2
rh
hr
h 2.
SERGE LANG. So ...

CHRISTOPHER. r 2
2rh
h 2.
SERGE LANG. You got it. So we have proved the formula for (r+h) 2.
Now to learn it by heart, we do it with a and b, or v and w, or what? Let's
take a and b. Then a plus b squared is equal to a squared plus two ab plus
b squared. Let's repeat this ali together.
SERGE LANG AND THE CLASS. a plus b squared equals a squared
plus two ab plus b squared. a plus b squared equals a squared plus two ab
plus b squared.
SERGE LANG. OK? You got it in your ears? You'll never forget it?
That's the way one memorizes something, you learn it by heart. You
repeat it three or four times at night, before going to bed, and the next
day, you know it. That's the way 1 learned it.
So with (r+h) 2?
STUDENTS. r+h squared equals r squared plus two rh plus h 2.
SERGE LANG. Good. Now we can do the subtraction:
lt's equal to what? lt's equal to

w(r 2
2rh
h 2)
wr 2 .
So what do we get. You [pointing to a student], what's your name?

A STUDENT. Jerry.
SERGE LANG. So what do you get?
JERRY. 2rh
h 2.
SERGE LANG. ls that true? [Sorne students indicate something is missing.] Watch out! We had a w in front of the parenthesis. Then we get wr 2
81
and - wr 2, which cancel, and we still have le ft w times 2rh + h 2 Right?

Then we get a formula which gives us the area of the band. 1 write it
down:
The area of the band is equal to

area of the dise of radius (r + h) - a rea of dise of radius r
in other words
area of the band = "'(2rh
h 2 ).
OK? No protest?
A STUDENT. No, we trust you. [Laughter.)
SERGE LANG. No, no! You should not trust me! The question ts
whether you have understood or not.

A STUDENT. So so.
SERGE LANG. Did you understand how we got the area of the band?
THE STUDENT. Heu ... [Looking at the picture.] We took the big circle minus the small one.
SERGE LANG. That's almost it. We took the area of the big circle
minus the area of the small circle. And the area of a dise of radius r is
wr 2 Then the radius of the big dise is equal to what?
CHRISTOPHER. r+h.
SERGE LANG. And the area? lt's w(r+h) 2. After that, 1 subtract the
area of the dise of radius r, which is r 2. Then we do a little algebra to

compute (r+h) 2, and we find the expression
are a of band == , (2rh + h 2).
Are you convinced?
CHRISTOPHER. Yep.
SERGE LANG. By conviction or intimidation?
CHRISTOPHER. By conviction.
SERGE LANG. Good, so now we have found the area of the band.
Next suppose 1 have a rectangle, like this:
h~l------------------~
with base of length /, and height h. Then the area of the rectangle ts
what? [Pointing to a student.]
THE STUDENT. . ..
SERGE LANG. The area of a rectangle is equal to what?
82
THE LENGTH oF THE CIRCLE
THE STUDENT. lt's the base times the height.

SERGE LANG. That's right, it's lh. You do know this or not?
CHRISTOPHER. Yes, we know it.
SERGE LANG. Good. Then what 1 want to do, is to find another
expression for the area of the band. 1 want to find inequalities for the area
of the band. A while ago, 1 found an exact expression. Now 1 want to find
another way of expressing the area of the band. Approximately, it will be
the length times the height.
h~l--------------~
But 1 have two possible lengths, with the two circles. 1 have the length of
the small circle, and 1 have the length of the big circle. There is a small
circumference and a big one. And the area of the band will satisfy certain
inequalities. 1t is bigger then the small circumference times the width.
Who sees this? Who sees th at the a rea of the band is bigger th en the
length of the small circle times the width of the band? [Serge Lang points
to severa! students.] You see this.
THE STUDENT. Yes, it's OK.
SERGE LANG. lt's like on this picture. 1 fold the band over the circle.
So 1 find the inequality:
(small circumference) X height
area of the band.
A STUDENT. 1 have not quite understood why.
83
SERGE LANG. Suppose that the length of the rectangle is equal to the
small circumference. If I curve the rectangle and wind it around the small
circle, then the rectangle sticks exactly around the small circle, but 1 have
to stretch it so it sticks around the bigger one. And if I have to stretch it,
this means that the area of the rectangle is smaller than the area of the
band. You see this?
THE STUDENT. Yes, I see it now.
SERGE LANG. Good. But on the other side, what kind of an inequality
do I get?
A STUDENT. [Hesitating.] It's going to be bigger than ... heu ...
smaller ...
SERGE LANG. Yes, you are on the right track. 1 have the band, I have
the big circle, and I have the distance between the two circles, which I call
the height h. I take a rectangle whose base is equal to the big circumference, the circumference of the big circle, and whose height is h. Now I
curve this rectangle, to get a curved band. The area of the rectangle is
simply the product of the big circumference times h. And the area of the
rectangle is bigger th en the a rea of the curved band, be ca use if 1 curve the
rectangle so it sticks exactly along the big circle, then it's going to fold up
along the small circle, it's going to make pleats.
h~L---------------c------------~.1
Do you see it? Th en I can wri te the two ineq ua li ti es:
(small circumference) h
area of the band
(big circumference) h.
Now, let's di vide everything by h. I di vide by h on the le ft, on the righ t,

and in the middle. What do 1 get?
[Questioning silence in the class.]
SERGE LANG. If you know that a
ah
== b, then do you know that
== bh?
THE CLASS. Yes.
SERGE LANG. And conversely, if ah
what?
THE CLASS. Then a
== b.
== bh and h ts not zero, then
84
SERGE LANG. Now, if a ~ b then ah ~ bh and if ah ~ bh then ... ?
If h is positive?
A STUDENT. a
b.
SERGE LANG. Right. If h is positive. So if 1 have something which is
smaller than another thing, and if 1 divide both things by h, then 1 still
have this same inequality. So what do we get? When 1 divide on the left
by h, then 1 get the circumference. When 1 divide the middle by h, 1 get
the area of the band divided by h. Which gives what? [Pointing to a
student.]
[The student keeps quiet.]
SERGE LANG. What's the area of the band? We found the formula a
minute ago.
THE STUDENT. w (2rh
h 2).
SERGE LANG. And if 1 divide by h?

STUDENT. 2wr.
SERGE LANG. Yes, that's the first term. And the second term?
SEVERAL STUDENTS TOGETHER. wh.
SERGE LANG. Very good. So we find 2wr
wh. Now, on the right,
what do we get?
STUDENTS. The big circumference.
SERGE LANG. Right. So we can write:
small circumference
2wr
wh
big circumference.
But our goal is to find a formula for the circumference. Suppose 1 take
h smaller and smaller. What happens?
A STUDENT. It gives the same thing.
SERGE LANG. It gives the same thing, yes. In other words, the big cir-
cumference approaches the small circumference. So if h approaches zero,

what happens? The small circumference stays whatever it is, it does not
change. The middle term is 2wr + wh. Then 2wr does not change, it does
not depend on h. But wh, what happens to wh when h becomes smaller
and smaller?
A STUDENT. It becomes different.
SERGE LANG. It becomes different, OK. But how? Weil, w is approximately equal to 3.14. If 1 take h == 1/10 then wh is approximately equal
to what?
CHRISTOPHER. 0.314.
SERGE LANG. And if h
== 1 /100?
SEVERAL STUDENTS. 0.0314.
SERGE LANG. And if h
85
== 1 /1000?
THE CLASS. 0.00314.

SERGE LANG. Now if h becomes smaller and smaller, then what hap-
pens to 'll'h?
THE CLASS. 1t becomes smaller and smaller.
SERGE LANG. That's right, and therefore it approaches zero. And
2TTr
+ TTh approaches what?
THE CLASS. The small circumference.

SERGE LANG. You are right, 2'!1'r
+ 'll'h approaches the small cir-
cumference, but when h approaches zero, then we have 2'll'r plus a

number which becomes smaller and smaller:
if h == 1110 then we have 2'll'r + 0.314;
if h == 11100 th en we have 2'll'r + 0.0314;
after that, we get 2'll'r + 0.00314; then 2'll'r + 0.000314. And if 1 continue
like this, when h becomes smaller and smaller, the sum approaches what?
A STUDENT. lt approaches 2'll'r.
SERGE LANG. Yes. Now on the right band side, 1 have the big cir-
cumference. When h approaches zero, what does the big circumference

approach? You already said it.
THE CLASS. [Ali together] The small circumference.
SERGE LANG. Very good. So when h approaches zero, then 1 get the
small circumference on each side, on the left and on the right, and 1 have
2'll'r in the middle. This means that 2'll'r is squeezed between these two
numbers: one is the small circumference c, and the other is the big circumference C. When the width h approaches zero, then C approaches c,
and since we squeezed 2'll'r between them, we get
== 2'll'r.
That's just what we wanted to prove!

Remember, we started from the area of a dise, and we found the formula for the circumference.
Now if we go to dimension 3, it won't be a dise any more, it will be
what?
A STUDENT. A sphere.
SERGE LANG. Yes, you are right. We can draw a sphere, and another
slightly bigger. 1 don't have the time to do this now, but suppose 1 know
the volume of the bali. What do 1 do to find the surface of the sphere.
Christopher, do you know?
CHRISTOPHER. . ..
SERGE LANG. What did we do for the circle. Do you remember?
86
CHRISTOPHER. No.
SERGE LANG. We just did it. Let's try to go through the proof quickly.
We started with the area of the dise, and we computed the area of the
band. How did we doit? [Pointing] You, what's your name?
STUDENT. Roger.
SERGE LANG. OK, Roger, how did we compute the area of the band?
ROGER. We took the difference between the two dises.
SERGE LANG. OK. And after that, we wrote inequalities for the area
of the band. You, what's your name?

STUDENT. Yaelle.
SERGE LANG. So then, Yaelle?
YAELLE. We said that the area of the band is bigger th en the product
of the small circumference times h.
SERGE LANG. And after that, what did we do?
YAELLE. You did the same thing with the big circumference.
SERGE LANG. Yes. And then?
YAELLE. You eance lied h and divided by h.
SERGE LANG. Right. Now when h becomes smaller and smaller, the
small circumference remains whatever it is, and the big one?

YAELLE. 1t approaches the small one?
SERGE LANG. Now you have to know what happens to the expression
in the middle, which is the area of the band. Before that, we still had one
thing to do. Roger?
ROGER. You took the area of the big dise minus the area of the small
dise.
SERGE LANG. Right, and that was 7r(r+h) -
what did we do?

ROGER. You used an identity.
SERGE LANG. Yes, and what did we find?
'!Tr 2. And after that,
ROGER. 2wrh
87
wh 2.
SERGE LANG. Very good. And Yaelle said that we divided everything
by h. Then what do we get?

ROGER. 2wr
wh.
SERGE LANG. Now, suppose that h becomes smaller and smaller.

What happens to wh? Let's pick on somebody else. You! [Pointing.]
THE STUDENT. It becomes smaller and smaller.
SERGE LANG. That's right.
When h approaches 0, theo wh also approaches O.

And the big circumference ... ?
THE STUDENT. It approaches the small one.
SERGE LANG. Right, the big circumference approaches c, which is the
small circumference. Therefore, we find:
2wr
c,
and this means that c == 2wr. And that's the proof.

OK, who can repeat the entire proof?
[Here, Serge Lang asks another student to repeat the proof, and the student still needs to be guided through it.]
OK, let's stop this for today, and let's talk about something else.
A STUDENT. lt's complicated.
SERGE LANG. It was too hard? We should go through it one more
ti me.
THE CLASS. [Ali together.] No.
SERGE LANG. No, not today, not today.
A STUDENT. But we are not used to this.
SERGE LANG. Oh, it's not that bad. What did we need? Just (r+h) 2.
lt's not that complicated.

THE STUDENT. No.
[General discussion among the students.]

SERGE LANG. Do you do proofs like this, in class?
THE STUDENT. Not like this one.
SERGE LANG. You want to talk about something else?
A STUDENT. What do you do? What's your work?
SERGE LANG. 1 am a professor of mathematics.
STUDENT. And physics?
88
SERGE LANG. No, just mathematics, at the university. I don't know
much physics. And I do research.

A STUDENT. [Disbelieving.] In mathematics?
SERGE LANG. The theorem that we just proved, somebody had to
prove it for the first time once.

A STUDENT. Oh yes.
SERGE LANG. It was done long ago, but there are other things which
have not yet been discovered.

YAELLE. And you look for them?
SERGE LANG. Yes.
ANOTHER STUDENT. What if you don't find anything? [Laughter.]
SOME STUDENTS. Resign! Resign! [Laughter.]
SERGE LANG. If I don't find anything? Yes, it can happen, but if I did
not find anything, ever, then I would not be a research mathematician. 1

would not have taken this direction, and I would have done something
el se.
A STUDENT. You find theorems?
SERGE LANG. Yes, sorne.
STUDENT. About what?
SERGE LANG. Oh you know, it's at a level which I can't explain here,
it would take too much preparation.

THE STUDENT. People willlearn them in 200 years?
SERGE LANG. In high school? Maybe, but there are people who learn
them now.
[The discussion continued on a variety of topics, including "Dallas",

which the French kids watch just like the American ones. One kid said that
her French teacher ca/led "Dallas" ''feeble minded". There was no disagreement, but they watch it. The students also brought up American culture and ils
influence on French culture, and economie problems faced by their country.
1t was remarkable on the whole to see how television influences their thinking and makes them aware of contemporary problems. After nearly an hour,
1 felt we should not end that way, but that we should go back to mathematics. So 1 picked up the reins once more amidst the general brouhaha.]
SERGE LANG. Look, the time is almost over, we can't end like this.
Let's finish with sorne mathematics. Maybe we can go through the proof
once more.
A FEW STUDENTS. Not again!
SERGE LANG. Let's come back to the proof. [Pointing to Yaelle:] You!
I started from the area of the dise '!l'r 2, and wanted to find 2'll'r for the circumference. So how did we do it?
89
[The brouhaha stops as soon as Yaelle starts to speak. A Il the students

remain quiet and listen intently as she goes through the proof]
YAELLE. Weil, you took the area of the big dise which is of radius
r+h, and you subtracted the area of the small dise which has radius r ...
SERGE LANG. [Encouragingry] Yes.
YAELLE. So this gives, after you work out the product: 2'll'rh
+ 'll'h 2.
Then you had sorne inequalities, and you divided by h ...

SERGE LANG. Hold it! Which inequalities?
YAELLE. The area of the band is bigger th an the circumference of the
small circle times h, and it is smaller than the circumference of the big circle times h.
SERGE LANG. Right.
YAELLE. Then, you cancelled the two h's dividing the area of the band
by h. This gave that when it approached ... euh ... wait .

2'll'r.
It gives
[She hesitates.]
SERGE LANG. Yes, you are doing OK. You get 2'll'r
+ 'll'h 2 in the mid-
dle; and on the two si des?

YAELLE. On the le ft, y ou have the small circumference, and on the
right you have the big circumference, and we saw that when h becomes
smaller and smaller, it approaches c, and we concluded that c was smaller
than 2'll'r and also that 2'll'r was smaller than c, so it gives 2'll'r equals c.
SERGE LANG. [Very happy with this performance-the students a Iso]
That's right! Congratulations. You have learned something.

[The en tire class applauds Yaelle warmly.]
SERGE LANG. lt's really good. 1 am very happy to have met ali of you.
It went very weil ... Yes, it was really good.
Comments
1 was talking to qui te a young class (8th grade). Shortly after, 1 gave a
similar talk to a lOth grade class (unfortunately not taped), and certain
difficulties which 1 encountered with the younger students had disappeared for this more advanced group. In both cases, 1 asked a student at
the end to repeat the proof, and in both cases, it was a success, giving rise
to the spontaneous applause. This shows how much students like to learn
and understand mathematics.
More technically, we see here how we can insert technical considerations (like the identity for (a+ b ) 2) in a conceptual context. Starting with a
problem which students grant is interesting (finding the length of the cir-
90
cie), we meet a technical question, which is solved as we go along. We mix

the general proof, based on general ideas, with such technical material,
and with rote memorization achieved by repeating the formula
(a+b) 2 == a 2 +2ab+b 2 out loud. The formula must be learned as a conditioned reflex, and students should be able to reproduce it instantaneously. This is complementary to being able to reproduce the proof for the
main theorem.
The area of the sphere
1 had the previous class on/y for an hour, and so was able to do on/y
the length of the circ le with them. But Stephane Brette ( 14 years old)
was visiting from another school, and so we were able to do the area of
the sphere together right after the class, by the same method.
SERGE LANG. [Addressing Stephane.] So, you came from another high
school to listen to today's class. Did you understand what 1 did?

STEPHANE. Yes.
SERGE LANG. Could you do the same thing to find the area of a
sphere? Suppose 1 give you the volume of the bali,

V == i'1Tr3.
3
What's the a rea of a sphere of radius r?

STEPHANE. 1 forgot the formula.
SERGE LANG. The formula is
A == 4'1Tr 2 .
OK, so you don't forget it again, we are going to repeat it together. [And
we repeat to~ether out loud: the area of a sphere is 4'1Tr 2, the area of a
sphere is 4'1Tr ... ].
You won't forget it any more?
STEPHANE. No.
SERGE LANG. Good. Then go to the blackboard, and prove the for-
mula.
[Stephane goes to the blackboard and draws the following picture.]
92
THE AREA OF THE SPHERE
SERGE LANG. That's right, you take two halls, one of radius r and the
other of slightly bigger radius r+h. And then?
STEPHANE. Then you have inequalities. [And Stephane writes:]
(area of the small sphere) h
~volume
of the band
~ (area of the big sphere) h.
SERGE LANG. Yes, you have a band between the two spheres, and
these are the inequalities which correspond to the band between two circles, but now we are in dimension 3. The volume of the band is equal to
wh at?
STEPHANE. It's the volume of the big bali minus the volume of the
small bali.
SERGE LANG. OK, so write them down.
[Stephane writes:]
Volume of the band

== volume of the big bali - volume of the small bali.
== i'lT(r+ h) 3
3
i'lTr 2 .
3
SERGE LANG. There you are. But you see, instead of (r+h) 2 and r 2 for
the circ le, we now have (r + h )3 and r 3 for the bali. We need sorne identity
for the expansion of (r+h) 3 . Do you know it?
STEPHANE. No, we haven't done that in class.
SERGE LANG. It doesn't matter, l'rn going to show you the answer
now, you can work with it, and then we'll prove it afterwards. The identity states that
Now substitute. What do you find?

STEPHANE. [ Writes on the blackboard:]
-'1T(r 3
3
i'lTr 3
3
3r 2h
2rh 2
i'lT3r 2h
3
i'lT3rh 2
3
h 3)
-'1Tr 3
3
+ ~'7Th 3
3
i'lTr 3 .
3
SERGE LANG. That's very good. And then, what happens?

STEPHANE. You can simplify.
SERGE LANG. OK, do it.
93
STEPHANE. One can simplify i'lTr 3 . Then we get
SERGE LANG. And then?

STEPHANE. Weil, after, you divide everywhere by h.
SERGE LANG. So go ahead, divide by h. What do you find?
STEPHANE. [ Writes:]
area of small sphere
4'1Tr 2
4'1Trh
i'lTh 2
3
are a of big sphere.
SERGE LANG. Now, what do we do?

STEPHANE. The big sphere approaches the little sphere, and the area
of the big sphere approaches the area of the little sphere.
SERGE LANG. Precisely. And the expression in the middle is squeezed
between the two. What happens to the expression in the middle? For
instance, the term 4'll'rh approaches what when h becomes smaller and
smaller?
STEPHANE. It approaches zero.
SERGE LANG. Yes, and the other term also approaches zero, the
~ '1Th 2.
So the middle expression approaches what when h approaches
zero?
STEPHANE. lt approaches 4'!1'r 2.
SERGE LANG. There you are! So if A ts the area of the sphere of
radius r, then we find
and therefore we get the equality

A
== 4'll'r 2 .
STEPHANE. That's what we were supposed to prove.

SERGE LANG. You said it, that's what we were supposed to prove. You
see that the proof went just as easily as the proof for the length of the circle.
Ali that remains to dois to deal with the little identity for (r+h) 3 . How
do we take care of that?
STEPHANE. You multiply.
94
SERGE LANG. Yes, we multiply like this [and Serge Lang writes on the
blackboard:]
(r+h) 3 == (r+h)(r+h) 2 == (r+h)(r 2
== r(r 2
+ 2rh + h 2 )
+ 2rh + h 2 ) + h(r 2 + 2rh + h 2 )
and we have proved the identity.

Now that we have given the proof, it's worth while to learn the formula
by heart, just like (r+h) 2. There are similar formulas for (r+h) 4, (r+h) 5
and so on, but they are a little more complicated. You can experiment
with these higher expressions, one after the other, to see what goes on,
and to see if you can find the general rule. But that's enough for today.
You can do this sorne other time.
Pythagorean triples
The following talk is the jirst of two given to an 11 th grade class in a

high school in the suburbs of Toronto in Spring 1982. The students
were about 16 years old. The talk lasted about 50 minutes.
SERGE LANG. What is this class? Eleventh grade? [Students nod.] And
what do you guys do? Algebra?

A STUDENT. Nothing. [Laughter. Many students talk at once.]
SERGE LANG. Weil, you must do something. [Laughter.] l'rn going to
do mostly algebra, but 1 start from a problem with geometrie overtones.

This problem goes back to Euclid. Suppose you have a right triangle, with
si des a, b, c.
c
b
You guys know the Pythagoras theorem?

A STUDENT. Yeah.
SERGE LANG. What's Pythagoras' theorem?
A STUDENT. a 2
b 2 == c 2
SERGE LANG. That's right. Now can you give me an example of a
right triangle where a, b, c, are integers? You know what an integer is?
[Students give answers. One of them answers the question about a right
triangle:]
A STUOENT. 3, 4, 5.
SERGE LANG. Yes, because 3 2
+ 42
== 52 ; 9
16 == 25.
96
PYTHAGOREAN TRIPLES
Can you give me another one?

A STUDENT. 6, 8, 10. [Laugher.]
SERGE LANG. That's not so stupid. 6, 8, 10 is another one. How did
y ou get 6, 8, l 0?
THE STUDENT. 1 multiplied 3, 4, 5 by 2.
SERGE LANG. That's right. But you could also multiply them by 3,
right? You can multiply them by anything, by any integer (a whole

number). For example, if dis any integer, then
Now can you give me an example which is not obtained by multiplying 3,

4, 5 by an integer? You, what's your name?
STUDENT. Charles.
SERGE LANG. OK, Charles, give me an example with integers a, b, c
which are not multiples of 3, 4, 5. Is there an example like that?
[Long silence. One student says: You di vide.]

SERGE LANG. If you divide, do you still get an integer?
STUDENT. No.
SERGE LANG. But 1 want integers.
ANOTHER STUDENT. 5, 12, 13.
SERGE LANG. Let's see, 52
which is
13 2
== 25, 12 2 == 144 and 25 + 144
169,
lt works. Ali right. Can you give me another one?
[Silence.]
Think hard. Is there another one?
A STUDENT. 10, 24, 26.
SERGE LANG. Yes, but that's a multiple of 5, 12, 13. You multiplied
them by 2. You see, once you have a solution, you can take any multiple
to get another one. Try to give me another one which is not a multiple of
the solutions we already have. 1 don't want multiples.
PYTHAGOREAN TRIPLES
97
A STUDENT. 7, 24, 25.

SERGE LANG. Let's see. lt seems to work. How did you get it, by the
way?
STUDENT. [The answer cannot be understood on the tape.]
SERGE LANG. Weil, obviously you compute pretty fast. That's very
good. Let's check it out. Somebody check it out quick.

SOME STUDENTS. lt's OK.
SERGE LANG. Can you give me another one? [Silence.] 1 mean, look.
If you try to compute one by one like that ...

A STUDENT. 1 can find another one.
SERGE LANG. Wait a minute. Suppose you try to compute like that.
Sure, maybe you can find another one. The next question is, will you be
able to find infinitely many? [Silence.] Do you understand the question?
STUDENTS. Yes.
SERGE LANG. So how many say yes, there are infinitely many? Will
you be able to find infinitely many which are not obtained by multiplying
one of them by an integer, which are not multiples of a solution? How
many say you can?
[Sorne hands rise.]

How many say you cannat?
[Many hesitate, sorne hands rise.]

How many don't know? How many keep a prudent silence? [Laughter.]
SERGE LANG. [Pointing to a student.] What do you think?
THE STUDENT. lnfinitely many.
SERGE LANG. Why do you think that?
THE STUDENT. Right now you can tell.
SERGE LANG. Oh no, you can't tell. Right now you have gtven me
three.
STUDENT. 1 don't know.
ANOTHER STUDENT. There are infinitely many squares. They go on
forever.
SERGE LANG. Yes, the squares go on forever. There are infinitely
many squares. But 1 want to find squares which are related by this equation. That's the question. Do they go on forever?
STUDENT. Yeah, because eh ... you had a first one, 3, 4, 5. Then you
have 5, 12, 13. Then 7, 24, 25. So the first numbers in each thing are 3, 5,
7. Then the next one would be 9, but that might be ... 1 would geta multiple of 3.
98
PYTHAGOREAN TRIPLES
SERGE LANG. Yes, this might give a multiple of 3. And then what? Do
you think you might find one with Il?

STUDENT. 1 am not sure if you can.
SERGE LANG. lt's very nice what he is doing. He is really thinking
very well. 1 And that, indeed, is how you might go about trying to determine whether there are infinitely many or not. Now what 1 am going to
do, is first to show you that there are infinitely many, and then to describe
them ali. Absolutely ali. That's what we are going to do. How many think
it's an interesting question?
[Laughter.]
SERGE LANG. Why did you laugh? [More laughter.] [Severa/ students
say they Ji nd it interesting.]

SERGE LANG. And are there people who don't find it interesting?
[Laughter.] Ah, there are a number of hands up, sorne fi nd it interesting,

and others don't find it interesting. Weil, OK, so 1 do it for those who find
it interesting. OK?
So 1 want to find ali solutions in integers a, b, e for the equation
(*)
Suppose 1 divide the equation (*) by e 2 1 get
OK? Now if a and e are integers, and b, e are integers, what do we cali
ale and ble? Do you know these numbers are called?
A STUDENT. Rational numbers.
SERGE LANG. Yes. So 1 let
x == ale
and
y== ble.
Then x, y are rational numbers. So what 1 have obtained is the equation

(**)
The equation x 2
+y2
l is the equation for what?
A STUDENT. A circle.
ldeally, 1 should now continue this exchange and let the student go on with his correct line
of discovery. 1 should involve others in a similar activity, and ask them to think about it un til
the next class. Then 1 would pursue the topic in the next class after they have had a chance
to show their own originality. But 1 am pressed for time, in this isolated lecture. so 1 choose
togo on.
PYTHAGOREAN TRIPLES
99
SERGE LANG. Yes, it's the equation of a circle.
What's the radius of the circle?

STUDENT. One.
SERGE LANG. So 1 have a circle of radius 1, centered at the ongtn.
Here 1 have a point (x, y), and wh at 1 want to fi nd are ali the rational
points on the circle. Th at means ali the points with coordinates x, y which
are rational numbers, and which satisfy the equation x 2 + y 2 == 1. If 1
have any such rational point, if 1 have rational numbers x and y such that
x 2 + y 2 == l, how do 1 geta solution of a 2 + b 2 == c 2, in integers?
STUDENT. 1 don't know.
SERGE LANG. What's a rational number?
[Silence.]
SERGE LANG. It's a fraction, it's a quotient of two integers, right? Suppose 1 have two fractions x and y such that x 2 + y 2 == 1. How do 1 find
integers a, b, c such that a 2 + b 2 == c 2 ? You can write x and y as fractions
with a certain denominator. Any fraction can be written as a numerator

over a denominator. Suppose 1 have two fractions x and y. You can write
them over a common denominator. You know that? So let c be a cornmon denominator, so you can write x as a numerator over c; and y as a
numerator over c also. So you can write
a
x==-
and
y==
with a and b integers. OK? If 1 have such fractions, then what will be the
relation between a, b and c? If x 2 + y 2 == 1 then 1 have
100
PYTHAGOREAN TRIPLES
which 1 can also write as

1.
Now what can 1 do to get a 2

What's your name?
+ b 2 == c 2 ? [Serge Lang points to a student.]
STUDENT. Laura.
SERGE LANG. Laura. Suppose you clear denominators in this equation
1.
What do you get?

LAURA. l'rn not sure.
SERGE LANG. What does it mean to clear denominators?
ANOTHER STUDENT. [He starts giving the right answer.]
SERGE LANG. You shut up! [Laughter.] 1 know you can answer. The
point is not with you, the point is with her. Ali right?
LAURA. You multiply by c
SERGE LANG. That's right. You multiply by c 2. And what do you get
when you multiply by c 2 ?

LAURA. You get a 2
+ b 2 == c 2.
SERGE LANG. That's ri~ht. You get a

2
+ b2
= c 2.
If 1 started with a + b == c then 1 ended up with x 2 + y 2 == 1. If

a, b, c were integers, x and y are rational numbers. If 1 start with rational
numbers, with fractions x and y, and put them over a common denominator, and then clear the denominators, 1 end up with a 2 + b 2 == c 2. So
whether 1 work with the equation x 2 + y 2 == 1 with rational numbers, or
with the equation a 2 + b 2 == c 2 with integers, 1 have equivalent problems.
Right?
OK, so let's look at x 2 + y 2 == l. We have the problem of getting solutions to the equation. To find ali rational numbers x and y such that
x 2 + y 2 == 1. Now we'li follow this idea how you can get them. l'rn going
to write down sorne formulas. Suppose you have played with this idea a
lot, how you can get them. Then you would end up with the foliowing
formulas. Let
x ==
y==
2t
1+t 2
1 let x and y be those expressions. OK? Then what is x 2

al ge bra? Charles.
CHARLES. One.
+ y 2 ? You know
PYTHAGOREAN TRIPLES
101
SERGE LANG. Yes. Why? Square x. What do you get when you square
x? Figure it out. Dicta te. What's the square of the numerator?

CHARLES. 1 -
2t 2
t4 .
SERGE LANG. That's right. And now suppose you add the y squared.
CHARLES. You get
SERGE LANG. That's it. That's good. That's exactly it. Both x and y
are over the same denominator. So what happens if you combine them?
Combine the numerators. What do you get? Charles.
CHARLES. 1 -
2t 2
t4
4t 2, which is 1
2t 2
t 4.
SERGE LANG. That's right, Charles is right, because - 2t 2
4t 2 ts
+ 2t 2. So you get
where the numerator is the same as the denominator. So what do you get
finally, Charles?
CHARLES. You get 1.
SERGE LANG. That's right. You get x 2
+y2
== 1. Now suppose you let
have special values. Try out sorne special values. Let's take sorne examples. Somebody give me a value fort.
A STUDENT. Two.
SERGE LANG. Ali right, take t == 2. What do you get?
x ==
A STUDENT. And y
==
{2
{2
22
+ 22
-3
5
2t
2
5
1+t
SERGE LANG. Yes, so that's my ale and ble. 1 can take
a == 3, b == 4, e == 5. We get the solution 3, 4, 5 that we had before.
Give me another value fort.
A STUDENT. t == 3.
SERGE LANG. Ali right, that's as good as any. Now plug it in. What
do you get?
[Student dietates:]
102
PYTHAGOREAN TRIPLES
SERGE LANG. And what is y?

STUDENT. y == 6/10.
SERGE LANG. OK. That's not so good, that's not interesting because
you are getting a multiple of 3, 4, 5. Let's do another example. Take

another value of t.
A STUDENT. Seven.
SERGE LANG. Ail right, when t == 7, what do you get for x?
STUDENT. 1 -
49 == -48, and 1 + 49 == 50 so x == 48/50.
SERGE LANG. And what do you get for y?

STUDENT. y == 14/50.
SERGE LANG. Yes, so if you reduce it to lowest form, you get?
STUDENT. 24/25 and 7/25.
SERGE LANG. Yes, that's another you picked out earlier, 7, 24, 25. Try
another example.
STUDENT. t
== Il.
SERGE LANG. Ail right. Then you get?

STUDENT. x
== l - 121 == -l 20
22
y == 122.
and
1 + 121
122
SERGE LANG. OK. And now if 1 reduce the fractions, what do you
get?
STUDENT. -60/61 and 11/61.
SERGE LANG. Ah, Ah! That's a new one! 11, 60, 61. Do you want to
give me one more value?

STUDENT. 13.
SERGE LANG. OK. Let's do the computation. [Serge Lang and students
do the computation.] If t == 13, we get
x ==
-168
170
-84
85
and
26
y
== 170
13
ss
So this is another one, 13, 84, 85. Now it's clear it will go on. Also you
gave me values of t which are whole numbers, which are integers. You
could have given me a fraction. Want to try a couple of fractions? [Students nod.] Ail right try a couple of fractions. Give me a value of t which
is a fraction.
A STUDENT. 1/2.
SERGE LANG. Then x is what?
[Student dictates:]
1/4
x==--+ 1/4
3/4
5/4
and
y==
2 l/2
5/4
PYTHAGOREAN TRIPLES
103
Everybody sees this? [Students agree.] So 1 am getting back 3, 4, 5.

[Serge Lang and the class go through a couple more examples with fractions, like 1/7.] So it's clear 1 can substitute any fraction for t, and 1 get
a, b, c; so 1 get a whole slew of solutions. Obviously 1 get infinitely many,
just by substituting fractions fort, by substituting rational numbers fort.
Now we have made a lot of progress from a few minutes ago, because
a few minutes ago, even using ali your brainpower, you were still getting
them one at a time and experimenting. And it was not clear whether
you'd get infinitely many. Now we've made the progress that we can actually write down a whole slew of them. Ali right, let's cali it a parametrization. Have you heard the word "parametrization"? t is called the parameter. We substitute values for t in perfectly weil defined expressions. We've
written down expressions, we substitute values for t, and we get solutions
to the problem of solving x 2 + y 2 == 1 in rational numbers.
The next question is, have we got ali the solutions or not? Do you think
there are any other solutions besides those which you write down like
this? This is quite a different type of question. lt looks much more complicated. Do you see that we are able to write down a bunch of solutions.
Are there any others? Have we missed any? Do you have any opinion?
A STUDENT. 1 don't know. Maybe we have missed sorne.
THE TEACHER. ls it possible that any of these could duplicate?
SERGE LANG. Yes, sorne duplicate. We have already seen that we get
3, 4, 5 from two of them. But now l'rn dealing with the problem of ali the
solutions of x 2 + y 2 == 1. The problem of duplication can be discussed
afterward. At the moment, 1 want to know: have we got them ali? What
do you think?
A STUDENT. 1 don't think so.
SERGE LANG. You don't think I've got them ali?
STUDENT. You're using rational numbers?
SERGE LANG. Sure, l'rn using rational numbers. Using the formulas
and
2t
and substituting rational numbers for t, do 1 get ali the rational numbers
(x,y) which satisfy the equation x 2 + y 2 == 1? How do you know that
there isn't another formula which gives me numbers x and y which don't
come from this particular way of setting them up?
STUDENT. 1 don't.
SERGE LANG. You don't. What do you think?
ANOTHER STUDENT. 1 don't know!
SERGE LANG. But you understand the question?
STUDENT. Yes.
104
PYTHAGOREAN TRIPLES
SERGE LANG. [Pointing to another student.] What do you think?

STUDENT. I think there may be another formula.
SERGE LANG. You think there may be another formula? Weil, I want
to show you one solution that we didn't get,
x == -1,
== O.
This is certainly a solution. If x == - 1, wh at is x 2 ?

A STUDENT. 1.
== 0, what is y 2 ?
SERGE LANG. If y
STUDENT. 0.
SERGE LANG. Add 0 to 1 and you get 1. Now I claim I can never get
x == - 1 from my formula, or y == O. Because if y == 0 what is t?

STUDENT. Zero.
SERGE LANG. Yes, t
O. And if t
0 what is x?
STUDENT. l.
SERGE LANG. So x is not equa1 to - 1. Therefore I cannat possibly get
the solution x == - 1 and y == O. So I miss at least one solution. Have I
missed an y other?
== O.
If x == 0, what is t?
STUDENT. When x
SERGE LANG.
STUDENT. Either 1 or - 1.
SERGE LANG. Weil, that's still OK. Suppose t
== 1 or
== - 1. I get a
solution with x == O. So that's now the question: are there any other solutions other than this one, x == - '1, y == 0, and the on es given by the formulas? How many think it's a cute question? [Laughter.] How many don't
give a damn? [Strong laughter.]
A STUDENT. Do you think x and y can both be zero?
SERGE LANG. No, if x and y are both zero, they won't add up to 1.
STUDENT. Oh, OK. Yeah.
SERGE LANG. x 2
y 2 has to be 1. I repeat the question, have I got

ali the solutions? [The students give signs of thin king.]
Ali right, now l'Il state the theorem.
Theorem. Ali solutions of x 2

obtained from the formulas
+ y2 =
1 in rational numbers x,y are
2t
x
by
gtvtng special
x = -l,y =O.
rational
values
of
t,
except
the
solution
PYTHAGOREAN TRIPLES
105
That is the only one 1 have missed. Ali the others have been obtained. Do
you know how to prove it? If 1 can prove it, then 1 have completely solved
the problem of solving x 2 + y 2 == 1 when x and y are rational numbers.
This is a complete solution of the problem 1 posed at the beginning.
1 am now going to give the proof. Suppose 1 have a solution. 1 want to
show that there exists a value of t, which is a rational number, such that
and
2t
y ==
1+ {2
If y ou try to discover how to get the t, y ou will fi nd yourself the patterns

that-what's your name?
STUDENT. Selim.
SERGE LANG. The patterns that Selim gave at the beginning. But of
course 1 gave you solutions so that 1 can give the complete proof before
the end of the hour. We didn't have time to experiment. You would have
to experiment, maybe for two days, before you would find how to do it.
One would have to systematize the patterns Selim was trying to find at the
beginning, experiment, and maybe after sorne time you would get the
proof. Now l'Il show you the proof by writing it down. It's a "let". You
know what a "let" means? [Laughter.] "Let" means you had a brainstorm
and you write down the solution, and you prove afterward that it's the
solution. So it's a "let". 1 let
t ==
y
x+l
It's a "let". Now either this will work or it won't. So what 1 have to show,
is that if 1 let t equal y over x+ 1, as in the picture:
then x and y are given by the formulas. From here on, it's just a question
of algebraic manipulation. If 1 let t == y /(x+ 1) and x, y are rational
numbers, then t is a rational number, right? If you add fractions, you get
a fraction. If you multiply fractions, you get a fraction. If you take the
106
PYTHAGOREAN TRIPLES
quotient of two rational numbers, you get a rational number. So 1 is a

rational number. Now I want to show that
and
Observe th at in the formula 1 == y !(x+ l ), the denominator x+ l does not

make sense if x == - 1. So I have to omit this case. But otherwise, it does
make sense. Now it's just a question-you know-like in the kitchen. You
cook a little, and it cornes out. We have
(x+ 1) 1 == y.
If you square this you get
(x+ 1)212 == y2.

And y 2 is wh at in terms of x? If x 2
+ y 2 == 1, th en y 2 equals wh at?
SELIM. 1 - x 2.
SERGE LANG. Yes, and l - x 2 can be factored. How?
SELIM. l +x times 1 -x.
SERGE LANG. Yes, you are doing very weil. You guys are very good,
very much at ease with algebra. So I have

(x
+ l ) 212 == ( l +
x)( l - x) .
Now you should have an irresistible impulse to do something, which ts

what? [Poinling lo a sludenl.] Wh at is your name?
STUDENT. Uno.
SERGE LANG. Ail right, Uno, what do you want to do now?
UNO. Divide by (x
+ 1) 2.
+ 1) 2 on the left but only
SERGE LANG. No, you have (x
+ 1 on the
right. So you have to divide by ... ?

UNO. x+ l.
+ l. That's right. Uno says divide by x + l.

[Laughler.] I always do wh at l'rn told, so I di vide by x + 1. What do I
SERGE
LANG. x
get?
UNO. (x+ 1)1 2
==
1- x.
SERGE LANG. Yes, and the left hand side is x1 2
x1 2 + 12 == l - x.
Now you solve for x. What do I get?
x( l
1 )
== l -
1 .
+ 12, so 1 have
PYTHAGOREAN TRIPLES
107
And then 1 can get
x
Th en
y == (x
1) t,
and you can use algebra again to see that y

what we wanted to prove.
2t /( 1 + t 2). This is just
STUDENT. How do you find this proof?

SERGE LANG. A good mathematician, fooling around for two days,
would be able to get the "let", write down the "let", and then do sorne
algebraic twiddling to prove that the "let" works. That's the difference
between a good mathematician, and someone who would not find the
"let", and would not find the proof. Sorne of you can, and sorne of you
cannat, find the solution. That's the difference between doing research
mathematics and not doing it. But that's what research mathematicians
do. They start the way Selim did. They try to experiment, and after a
while, one sees what works, and one finds the proof that it is a solution.
That's the answer to your question.
Comments
The period ended before there was time to expand on this. Pressed on
several occasions concerning the history of the formulas and the "let"
which 1 introduced somewhat abruptly, 1 finally inquired into the history
of the formulas. G. Lachaud informed me that it took centuries before
the formulas took the final form which 1 have given here, with contributions by several mathematicians. Euclid, or mathematicians of his time,
knew three centuries BC that you could find Pythagorean triples a, b, c by
using the formulas
b == 2mn,
where m, n are integers. Check using algebra that you indeed get
a 2 +b 2 == c 2 from these formulas. Then you can give m, n arbitrary values
in integers to get solutions. Mathematicians at the time of Euclid did not
like fractions, and preferred to deal entirely with integers. Only much
later, Diophantus (three centuries AD) worked with fractions, and knew
that ifyou divide the above formulas by m 2 + n 2, and then putt== n/m,
then you find the formulas
108
PYTHAGOREAN TRIPLES
and
2t
1+t 2
So it took six centuries between Euclid and Diophantus (who worked at

Alexandria in Egypt, but was a Greek mathematician) to find the formulas as I wrote them down at the beginning, to get infinitely many solutions
of the equation x 2 + y 2 == l in rational numbers.
Diophantus himself did not raise the question whether these formulas
give ali solutions in rational numbers. For this one had to wait another six
centuries. In the lOth and Il th cent ury, Arab matherr1aticians like Al
Khazin raised this further question, and solved it by algebra, essentially as
we did it today. Since the algebra is about at the same level as that known
to Diophantus, we now see after the fact that if Diophantus had raised the
question, he would have been able to solve it.
This history shows that a complete solution resulted from the work of
mathematicians ranging over thirteen centuries. Since then, mathematicians (including me) have copied the solution, as I did today.
It may interest you to pursue the idea of Selim, who noticed the pattern that the odd integers 3, 5, 7, 9, Il seem to occur systematically at the
beginning of Pythagorean triples of integers, which are solutions of
a 2 + b 2 == c 2 Is it true that every odd integer will be the first integer in
such a triple? If it is true, prove it. If it is false, give an example of an odd
integer which does not occur.
There was also no time to deal with the question raised by one of the
teachers (and sometimes raised by students) whether and how there are
duplications in the solutions coming from the formulas. This is in fact
quite easy to do.
For example, suppose t ranges from 0 to 1, so t increases from 0 to 1.
Then 1 - t 2 decreases from l to 0; and 1 + t 2 increases from 1 to 2.
Hence the quotient
x ==
decreases from 1 to 0, and there are no duplications of x for this range of
values of t.
What happens wh en t ranges over ali numbers grea ter than 1? Write
t == 1/s and figure it out for yourself. Also figure out what happens when
t is a negative number. And finally, figure out what happens when you
write t == min, whether in lowest form or not to answer completely the
question about duplication of solutions, or the possibility of getting
integer multiples of solutions a, b, c for a 2 + b 2 == c 2
Today, students at the age of 16 know something of coordinates and
the representation of the circle by the equation x 2 + y 2 == 1. Neither
Euclid, Diophantus, nor Al Khazin knew about that. The idea of using
coordinates arose only in the 16th and l7th century. You will notice that
although I drew a circle and interpreted the problem of finding solutions
PYTHAGOREAN TRIPLES
109
of the equation in terms of finding "rational points on a eirele", the figure

was not used in the arguments. I argued eompletely algebraieally, mueh as
the aneients would have done. However, today it is also useful to give a
geometrie interpretation of the algebraie steps. In partieular, when I let
t ==
x+ l '
what is the geometrie interpretation of t? If I had another hour, I would

lead students to realize that the equation
y == t(x
+ l)
is the equation of a straight line, with slope t.
(x= -1, y= 0)
the 1ine y= t(x + 1)
This line passes through the point x == - l, y == O. The point where the
line interseets the eirele is preeisely the point given by
2t
y ==
{2 .
This is the geometrie interpretation of the part of the proof showing how
to get back these formulas for x and y in terms of t == y !(x + l ).
Infini ti es
The following talk is the second given to an 11 th grade class in a high

school in the suburbs of Toronto in Spring 1982. The talk lasted about
50 minutes. Although the first few minutes are spent discussing a topic
extending that of the first talk, 1 have preserved the timing and balance of the talk intact.
SERGE LANG. Any questions on what 1 did yesterday? No? l'Il deal
with a different topic today, but just to connect with yesterday, let me
mention one more thing. In the lecture yesterday, we looked at
And we wanted solutions when x and y are rational numbers, that is fractions. Now 1 do want to mention-suppose you ask for more complicated
equations. What's the next more complicated equation that you can think
of?
SELIM. x 3
y 3 == 1.
SERGE LANG. That's right, you c.an either have this, or something like
x3
+ y2
== l.
Either one or the other. Now do you think there are infinitely many solutions of this equation in rational numbers?
SELIM. 1 don't know.
SERGE LANG. You don't know?
SELIM. There should be.
SERGE LANG. Oh you think so?
SELIM. Yeah. There were infinitely many yesterday.
SERGE LANG. There were infinitely many for x 2
about for x 3 + y 2 == l?
SELIM. 1 can see at least four. Right away.
SERGE LANG. Are you sure? Which four?
+y2
1. What
INFINITIES
SELIM. x
Ill
== -1, y == l.
SERGE LANG. [lnterrupting.] What?

SELIM. Eh, no [he computes in his head and part/y out loud].
SERGE LANG. It's not so clear, is it?
SELIM. l'rn not sure exactly.
SERGE LANG. Ah, now you are not sure. That's right. Weil, the situa-
tion is infinitely more difficult for equations with a cube than only with
squares. And I think the only solutions are essentially x == 0, y == 1, or
x == l, y == 0, or x == - 2, y == 3. The re is no other, unless y ou allow
y == - l or y == -3, because y 2 gets rid of the minus sign. That's qui te
hard to prove. Qui te hard. And similarly for higher powers. I t is exceedingly difficult to handle such equations the moment the degree is bigger
than 2. It's a very famous unsolved problem, that nobody knows how to
solve. If you take an arbitrary n, and the equation
xn +yn == 1,
show that only solutions are with x == 0, y == 1 or x == l, y == 0, except
for putting a possible minus sign. That's a very difficult problem, which is
called the Fermat problem, and nobody knows how to do it.
One knows how to do it for low values of n, like n == 3, 4, 5, 6, 7. For
a while, for low values of n, people can hack it out, by brute force, using
ad hoc methods. But to give a proof in general, nobody knows how to do
it. It's what's called an unsolved problem in mathematics. That's what
mathematicians do, by the way. They pick nice unsolved problems, that
they get a high on, th en they work on them. I t gives them a high to work on
unsolved problems. Do you think it would give you a high to work on an
unsolved problem of mathematics?
A STUDENT. Sure. [Laughter.]
ANOTHER STUDENT. We'll have it ready for you next time.
SERGE LANG. You'll have it ready for me next time? If you do, you'll
make it into the history books. [Laughter.] Want to make it into the his-
tory books?
Actually, I was just mentioning this, now let me get on to the topic I
want to discuss today. I want to discuss questions of infinity. For example,
take the integers,
l, 2, 3, 4, 5 and so on.
They go on indefinitely. And there are infinitely many of them. Now suppose I take the odd numbers. You know what the odd numbers are?
A STUDENT. Those that are not even. [Laughter.]
SERGE LANG. Ail right, what are the even numbers?
THE STUDENT. 2, 4, 6, 8, ...
IN FINITIES
112
SERGE LANG. Ail right, 1 write them down,
positive integers:
even numbers:
7 .. .
10
12
14 .. .
Observe that you can make them correspond with the positive integers. So
there are no even numbers missing in the correspondence. 1 can line up
the even numbers, a first one, second one, third one, and so on, so 1 won't
miss any of them. When that happens, 1 say that the even numbers are
denumerable. "Denumerable" means that 1 can enumerate them. A first
one, second one, third one, and so on, to take them ali into account. What
about the odd numbers? Are they denumerable? [Serge Lang points to a
student.] What's your name?
STUDENT. Ted.
SERGE LANG. OK, Ted, are the odd numbers denumerable?
TED. l'rn not sure what you mean.
SERGE LANG. "Denumerable" means that 1 can list them, a first,
second, third, etc. so that 1 don't leave out any of them. Here are the odd
numbers:
oddnumbers: 1 3
Il
13
15 ...
For the even numbers, 2 is the first, 4 is the second, 6 is the third, 8 is the
fourth, 10 is the fifth, and so on. Which is the tenth even number?
TED. 20.
SERGE LANG. That's right. Which is the thirtieth?
TED. 60.
SERGE LANG. Yeah. Have 1 missed any?
TED. No.
SERGE LANG. So 1 have enumerated the even numbers. Which is the
first odd number?

TED. One.
SERGE LANG. Which is the second odd number?
TED. 3.
SERGE LANG. Which is the third odd number?
TED. 5.
SERGE LANG. And then?
TED. The fourth is 7, then the fifth is 9, and so on.
SERGE LANG. Right. So 1 have a one to one correspondence between
the odd numbers and the positive integers. They correspond, it's one-one.
INFINITIES
113
One in te ger corresponds to one odd number. l corresponds to 1, 2

corresponds to 3, 3 corresponds to 5, 4 corresponds to 7, 5 corresponds to
9, and so forth. 1 have enumerated the odd numbers.
23456
3 5 7 9 Il
7 8
13 15
Have 1 missed any odd numbers in this way?

TED. No.
SERGE LANG. So then 1 would say that the odd numbers are denu-
merable. By the way, could you describe this correspondence by a formula? If 1 cali n the n-th integer, like the first one, second one, third one,
n-th one, what will be the n-th even number?
TED. 2n.
SERGE LANG. That's right, the n-th even number is 2n. The n-th odd
number would be what?

TED. 2n- l.
SERGE LANG. Yes. That's right, good for you. You said it, the n-th odd
number is 2n - 1. So that's how 1 enumerate the odd numbers and the
even numbers. Even though the odd numbers and the even numbers form
part of ali the positive integers, 1 cali still make this one-one correspondence between them, and enumerate them.
Now 1 have a question. What's the next kind of number you know of?
The rational numbers, yes? That's ordinary fractions.
Question. Are the rational numbers denumerable?
Do you understand what 1 mean? Take the rationals, say the positive ones.
Forget about the negative ones. So it's numbers like 2/3, 7/5, 13/37, fractions. Can 1 enumerate them? Any integer is of course a fraction. 1 can
write l == l /1, 2 == 2/l, 3 == 3/l. So certainly the positive integers constitute part of ali the positive rational numbers. N ow can 1 take ali the
rational numbers, and line them up, so there is a first one, a second, a
third, and so on, and so that 1 don't miss any of them?
How many say yes? [A few hands go up.]
How many say no? [Sorne hands go up. Most/y nobody says anything.]
How many keep a prudent silence? [Laughter.]
Ali right, let's see those who say no. Who said no? Did you say no?
STUDENT. 1 said neither.
SERGE LANG. You said neither? [Severa/ students talk simultaneously.]
Ali right, you [Serge Lang points to a student]. [Laughter.] What did you
say?
STUDENT. No 1 don't think so, but 1 don't know.
SERGE LANG. Why? Why not? Can you give a reason for it?
STUDENT. No.
114
INFINITIES
ANOTHER STUDENT. Weil, if anybody says you can enumerate them,
you give me two, and you can find another one between them.
SERGE LANG. Ah! that is in the usual ordering. Sure, if you take ali
the rational numbers on the number line, and if you take two rational
numbers, you can find another one in between.
1 -3
2 4
0
3
2
2
2
But 1 didn't say to take the usual ordering. Suppose you change the ordering?
STUDENT. 1 think in any ordering you could do the same.
SERGE LANG. Any ordering?
STUDENTS. The fractions are infinite.
SERGE LANG. Oh, but the even numbers were infinite too!
STUDENT. No, but what 1 am saying, is the fractions between two
integers are infinite.

SERGE LANG. Do you think that's an argument? lt's true, if 1 take two
fractions in the usual ordering, there will be another fraction in between.

Now can 1 line up the fractions in a different ordering, so there is a first, a
second, a third, and 1 don't miss any?
ANOTHER. You have an infinite number of denominators, right?
SERGE LANG. Yes 1 do.
STUDENT. And each denominator has infinitely many numerators. So
how can you line it up when you don't know how high this is going to go?
SERGE LANG. My answer to you is: you ask me how can 1 line them
up. Would my failure to line them up be due to my stupidity ...

STUDENT. No.
SERGE LANG .... or because of the mathematics?
STUDENT. Because there is an infinite number of denotninators. Like
the numerator of the fractions can keep changing and changing if 1 made
the denominators bigger.
SERGE LANG. That's true. But that's still not an argument. How do
you know ...

STUDENT. How can you line them up if you can keep changing the
denominator? Like for those ...

SERGE LANG. Weil, suppose 1 was elever enough! You haven't given
an argument. 1 mean, 1 don't know at the moment. 1 haven't told you
INFINITIES
115
what's true or not. l'rn simply questioning you. How do you know if it's
built in the structure of the mathematics, or if it's built in to the accidentai
thing that 1 am not elever enough to do it?
[Long silence.]
You have to give a proof. If you assert you cannat do it, you will have
to prove it. [A hand goes up.] Yes?
STUDENT. [The tape cannot be heard clearly. The student appears to
make a reference to the step by step ordering of the even and the odd
numbers. One can hear faintly: the smallest step could be one.]
SERGE LANG. It's true that the even and odd numbers are arranged
by steps. Maybe in dealing with the rational numbers 1 cannat deal with
something that you cali the steps. 1 have to deal with something else. Ali
you have said is that the method which 1 used for the even and odd
numbers does not apply to the rational numbers. That's ali you have said
so far. How do you know that if l'rn elever enough, 1 can't find another
method? [A hand goes up.] Yes?
STUDENT. Weil you take l, 2, 3, 4, 5, 6 and so on. You put them ali,
the whole thing, you put them over 1. And then you do it-you put them
over 2, and you do it over 3. Then the denominators, they go up ...
SERGE LANG. Yes, they go up. That's very good. What's your name?
STUDENT. Gary.
SERGE LANG. So what Gary is proposing is this. Let's line up ali the
fractions according to their denominators. By the way, what is your name?
[Pointing to the student who had raised the question about denominators.]
What's your name?
STUDENT. Ken.
SERGE LANG. Ken also said what happens when the denominators
grow bigger. So both Ken and Gary said something similar. We have the
fractions, say with denominator 1:
1 2 3 4 5 6 7 8
- - - - - - - - and so on.
1 1 1 l l 1 1 1
Now they were mentioning the denominators which grow bigger.
GARY. With 2.
SERGE LANG. Yes, so you have
__!__
3_ 2_
2_
i 2
2 2 2 2 2 2 2 2
and so on.
Then you have

1 2 3 4 5 6 7 8
- - - - - - - - and so on.
3 3 3 3 3 3 3 3
116
INFINITIES
Then, what would be the next one? Gary.

l 2 3 4 5 6 7 8
GARY. - - - - - - - and so on.
4 4 4 4 4 4 4 4
SERGE LANG. So 1 have lined up the fractions, horizontally for those
with the same denominator, and vertically with the denominator getting
bigger.
1 2 3 4 5 6 7 8
TTTTTTTT
l
2 3 4 5 6 7 8
2 2 2 2 2 2 2 2
2 3 4 5 6 7 8
3 3 3 3 3 3 3 3
--------
2 3 4 5 6 7 8
--------
4 4 4 4 4 4 4 4
Now can 1 enumerate them? 1 have lined them up the way you wanted
me to li ne them up. N ow can 1 make a list? A first, second, third, fourth,
so that 1 don't miss any of them?
[Silence. A hand goes up.] Yes?
A STUDENT. No, because the denominators and numerators are get-
ting bigger.
SERGE LANG. ls that a reason? You can make the list any way you
want.
THE STUDENT. You have to use [tape is not clear].
SERGE LANG. 1 can use anything you want. You can make the list any
way you want. As long as there is a first, a second, a third and so on. 1
can use any method you want, as long as they are lined up and 1 get ali of
them. Ail you have to do is to be elever. Do you think it's a nice question?
[Smiles, sorne students say yes.] Do you think it's a dumb question?
[Silence.] Weil, you are paying attention.
A STUDENT. Weil you take 1 /4; and then you take diagonally across,
it would be l /3 and 2/4; then you have 2/3 ...

SERGE LANG. Ah, Ah! You got it! Heh!? He got it! What's your
name?
STUDENT. Sunil.
SERGE LANG. OK, l'Il do exactly what Sunil says. Let's start with this
corner [points to the upper left hand corner, to the fraction l /1]. This is my
first, l 1 l. This is the second, 21 l. This is the third, l /2. This is the fourth,
l /3. This is the fifth, 2/2. Six th, 31 l. Seventh, eighth, nin th, tenth, ...
INFINITIES
117
-2
-3
-3
;;4/4/4/4
~/2/2/2/2/2/2
3/3 ,.,3/3/3/3
1
2/ 3
eleven, twelve, thirteen, fourteen, fifteen, sixteen, seventeen, eighteen, ...

Now do you see it? Have 1 enumerated them? Am 1 going to miss any
of them by this method?
STUDENT. No.
SERGE LANG. Eventually, 1 will encounter every one of them. 1 will
meet ali of them, and there will be a first, a second, a third, a fourth, and
so on. So in fact they are denumerable. You understand the proof? That's
the theorem. The answer is yes.
Theorem. The positive rational numbers are denumerable.
And the proof is right there. 1 have given you the enumeration. You agree
that 1 have enumerated them?
STUDENT. Yeah.
SERGE LANG. So what about the people who said 1 couldn't do it?
What happened, heh? 1 mean, it's tricky, the mind. It's tricky.
A STUDENT. Is there an equation for that? Can you give an equation?
SERGE LANG. Yes, you can, but it's more complicated. 1 would have to
give the exact recipe how to do it, by an equation. In fact, 1 don't know
ofthand, how to write the equation. Why don't you do that as homework?
[Strong laughter.] Try to think through ... [Very strong laughter.] No, it's
not so bad, it's not so bad. Weil, it will take you ... depending how fast
your mind works, it might take you from five minutes to an hour. To give
the equation, to give the relationship, as we did for the even and odd
numbers: 2n and 2n- 1. The se were the formulas. N ow how do you give
the formulas in this case, for the n-th rational number, it's an exercise. 1
But it's a different question, and 1 want to do something else, and if 1 start
working on the formulas for this, 1 won't have time to do something else.
Thinking about this later, 1 don't see how to give a single formula. 1 would do it only with
severa) formulas, to fit the exact enumeration as on the diagram.
118
INFINITIES
I'd rather do something else rather than give the formulas for this one. So
l'Il let you work out the formulas.
But you see how tricky the mind is?
STUDENT. Yeah.
SERGE LANG. You should learn both, simultaneously: to trust your
intuition, to distrust it; to use your intuition to make guesses, to use your
analytical powers to test the guesses, whether they are true or false. Ali of
this simultaneously. You have to learn how to control your brain.
[Simultaneous talk by students, Serge Lang, sorne teachers in the back
who say "it's a big, tai/ order".]
You think that's a big, tall order? lt's not that bad.
So now we go to the next question. We decided that the rational
numbers are denumerable. Although it looks as if there are a lot more
rational numbers than integers, we have just enumerated them. Now do
you know what real numbers are? The number line. You know the
number line? Ail the numbers on the line.
-v/2
-tr
1
-3
-2
-1
1
2
V2
1r
1
1
3
STUDENTS. Yes we know. [Severa/ talk simultaneously.]

SERGE LANG. Do you think those numbers are denumerable? Can
you line them up so there is a first, a second, a third, a fourth, and so on?
How many of you say yes? That's my question. Let me write down the
question.
Are ali the numbers on the number line denumerable?
By the way, we might as weil limit ourselves to the numbers between 0
and l. For simplicity. So how many say yes? Raise your hands. How
many say no? [Laughter.] How many keep a prudent silence? [A/most
no body raised the ir hands either time. Laughter wh en no hands are raised.]
SERGE LANG. Oh, you guys are getting careful! What do you think?
Not so clear, heh. After you got caught a few minutes ago, with the
rationals, one has to be careful. What do you think? [Serge Lang points to
a student.]
A STUDENT. 1 think you can show they are denumerable.
SERGE LANG. They are denumerable? Why?
STUDENT. You could use the same reason as the other one.
ANOTHER STUDENT. No, 1 don't think so.
SERGE LANG. Weil, let's see. You look at ali the numbers between 0
and l.
INFINITIES
119
0
KEN. Between 0 and 1 you have infinitely many of them.
SERGE LANG. But there is also an infinite number of fractions,
between 0 and 1, and you can enumerate the fractions. [A hand is raised.]
Yes?
STUDENT. 1 wonder if you could use denominators again.
SERGE LANG. No there are no denominators. Ail the numbers x
between 0 and l ... What is by the way a "number" on the number line?
What's the easiest way of representing a number?
KEN. Decimais.
SERGE LANG. Decimais, OK. So 1 take ali the decimais. For example.
0.13251986127824 ... and so on.

That's a typical infinite decimal. So that's what my numbers are. Decimais
like that. 1 take the decimais as the definition of numbers. A number is an
infinite decimal. Now, can 1 line up the decimais so there is a first
decimal, a second decimal, a third decimal, and so on, so that 1 don't miss
any? Can 1 do that? That's the question.
A STUDENT. Each number is an infinite decimal.
SERGE LANG. Yes, that's right. Each number has an infinite decimal.
That's why 1 put the dot, dot, dot ...

THE STUDENT. Wh at about
SERGE LANG. Sure,
'lT
'lT? 'lT
does not have a decimal value.
== 3.14159. . . You know there is an infinite
decimal which gives the value for 'TT. The value is itself the infinite
decimal. That's the value of the number. You know 'lT does not stop after
3.14. Don't you know that? 1 mean, there is a 159 [Laughter]. It goes on
like this:
3.141592653589 ...
[Serge Lang writes many of the decimais. Laughter.]
TEACHER. You're using a ...
SERGE LANG. Shhhh!
TEACHER. OK, 1 won't say.
SERGE LANG. Don't give away my secret! 2
2
The secret is a poem in French: Que j'aime faire apprendre un nombre utile aux sages!
Immortel Archimede, arti~te, ingenieur, qui de ton jugement peut priser la valeur? Pour moi
ton problme eut de pareils avantages.
120
INFINITIES
TEACHER. He is working it out in his head, that's right.
[Laughter. Talk by students. He's doing it in his head! How are you doing
it? What are you thinking?]
SERGE LANG. So a number is an infinite decimal. Can 1 line up the
numbers so the re is a first one, second one, third one, and so forth? Wh at
do you say?
STUDENT. No.
SERGE LANG. He says no.
STUDENT. With the rationals, you had a place to start. [Tape cannot
be made out.] Now you don't have a place to start.

SERGE LANG. Yes, but again, ali that you are saying is that the
method which applied to the rationals will not apply to the real numbers.
The people who started to give reasons why 1 could not do it for the
rationals: ali they were saying is that the method which applied to the
even numbers did not apply to the rationals.
STUDENT. Probably there is.
SERGE LANG. Oh yeah, probably there is?
STUDENT. So far you used different methods; you first used one
method, then you changed the method; probably a different method

would do it for ali the numbers.
SERGE LANG. That's a very weak argument. [Laughter.] Because the
argument is based on psychology, and 1 am asking you to deal with
mathematical problems. Not psychological one s. [Laughter.] So if you start
basing your mathematical intuition on my psychology, [Laughter.] you're
going to have a hard time with it. That's dangerous. Think again.
[Students talk among each other.]

SERGE LANG. Ail right, l'Il tell you the theorem.
Theorem. The infinite decimais are NOT denumerable.

That's the theorem. You cannat do it. And l'Il prove it. Weil, suppose you
could enumerate the infinite decimais. Suppose there is a first one. Suppose you make a list of them. So there is a first one, a second, third,
fourth ... 1 have to write them as infinite decimais. 1 can't write an
infinite decimal completely. 1 have to use sorne means of expression for
the decimais. Suppose we have the first decimal
So each a represents an integer between 0 and 9, each a 11. a 12 , a 13 , and so

on, is an integer between 0 and 9. Suppose there is a second one, which 1
write
INFINITIES
second:
. a2 1a22 a23 a24 a25 a26
. . .
and so on.
third:
. a3 1a32 a33 a3 4 a35 a 36
, , ,
and so on.
121
Now the fourth one, how would 1 denote it?

GARY . . a 41 a 42a 43a 44 a 45 a 46 and so on.
SERGE LANG. Yes. And what would be the n-th one? How would 1
denote the decimal for the n-th one? Gary.

GARY . . anlan2an3an4
SERGE LANG. So suppose 1 have such an enumeration of my infinite
decimais:
first one:
.a 11 a 12a13at4ats .. .
second one:
.a21 a22a23a24a2s .. .
third one:
.a3 1a32a33a34a35 .. .
1 have to show that there is one number, one infinite decimal, that is not
in the list. If 1 show that, 1 have proved the theorem. Because 1 have
shown that if you make any list, 1 can always find one number that is not
in the list. So 1 have to give you a method for finding one which is not in
the list. How do 1 find one which is not in this list?
[Silence.]
1 have to give you a recipe, how to find one which is not in the list.
Any one which is not in the list is going to be different from those in the
list. 1 do it this way. 1 pick a number b 1 , an integer between 0 and 8 such
that b 1 is not equal to a 11 . 1 mean, a 11 is whatever it is. Suppose a 11 is 2.
Pick the first number b 1 to be 7. Whatever you want, except 2. Now pick
the second decimal. Pick a number b 2 , any integer between 0 and 8, such
that b 2 is not equal to a 22 . Next pick an integer b 3 such that b 3 is not
equal to a 33 . What's my next step?
STUDENT. Pick an integer b 4 such that b 4 is not equal to a 44 .
SERGE LANG. That's right. And in general, pick bn such that bn is not
equal to-what?
STUDENT. ann
SERGE LANG. Perfect. And now 1 form the following decimal:
Can this decimal be equal to any of the ones which 1 have listed here?
122
INFINITIES
STUDENTS. No.
SERGE LANG. Why not?
STUDENT. Because the numbers are different. [The tape is not clear,
but the student has the right idea.]

SERGE LANG. That's right. Let's say it a little more precisely. Suppose
this decimal
is equal to one of those which I listed. Then it would be in this list, somewhere. Which one? Say the n-th one. But it can't be the n-th one, because
the n-th one is
and it is different from the n-th one because we have made it different, by
picking bn different from ann. So the decimal . b 1b2 b3 b 4 . . . cannat be
equal to any decimal in the list.
So what I have shown is that if you have a list to begin with, I can
always find a decimal which is NOT in the list. Therefore there could not
be a list to begin with, of ali the decimais. Do you understand the argument? Selim, what do you say?
SELIM. Yes.
A TEACHER. But this rests on the assumption that you are able to find
a b 1 which is not equal to a 11 .
SERGE LANG. a 11 is only one number. Suppose a 11 is 2. Let b 1 be
equal to 3. Suppose a 22 is 5. Let b 2 be equal to 7. Suppose a 33 is l. Let b3

be equal to 8. Suppose a 44 is 4. Let b 4 be equal to 5. Ail I have to do is to
make the b's different one at a time.
A STUDENT. Why did you take the b's between 0 and 8?
SERGE LANG. I took them between 0 and 8 for a technical reason. If
you look at a number like this:

0.1399999999999 ...
which goes on like this, with ali 9's at the end; and you take
0. 1400000000000 . . .
like that; then in fact, these two numbers are equal, and I want to avoid
this kind of ambiguity which occurs in decimal expansions. So it's just a
very minor technical point. If I hadn't mentioned it, you would not even
have noticed it. I mentioned it for my own peace of mind. But if I had not
mentioned it, I don't think you would have picked it up. That's the only
reason. It's the only kind of ambiguity that can occur. If the same number
INFINITIES
123
has two different decimal expansions, the only difference can be that you
have ali 9's at the end of one of them, and ali O's at the end of the other.
Otherwise, that's the only way you can have two different expansions for
the same number. And that's the only reason 1 picked them between 0
and 8. Otherwise, forget it.
So you agree that the set of ali infinite decimais is not denumerable?
So it was pretty, yeah? The rational numbers look as if there are a lot
more than the integers. Your first thinking was that-for most of you-the
rational numbers are not denumerable. But they are, even though the
denominators get bigger. Then 1 caught you psychologically with the real
numbers-caught sorne of you-thinking, weil, since it's going like that,
they will also be denumerable. And then finally 1 pulled the rug from
under you [Laughter] and showed that the real numbers are not denumerable.
So, this is what 1 wanted you to see. And you see, we are using very little here, about mathematics. We are just using the notion of correspondence, enumeration. We are not using arithmetic, addition, or multiplication, or anything like that. lt's ali just making things correspond to each
other.
Postscript
What follows are lengthy excerpts from a discussion between Patricia

Chwat, a teacher in a Paris High School; Jean Brette, director of the
Mathematics Section of the Palais de la Dcouverte (the main science
museum in Paris); Stephane Brette (14 years old), his son; and of
course, Serge Lang.
The discussion brings up various points concerning the teaching of
mathematics, and was transcribed from a video tape by Patrick Huet, of
the Paris Research Institute for the Teaching of Mathematics (in
French: /REM, Institut de Recherche pour l'Enseignement des
Mathmatiques). As far as 1 can tell, the points raised here are va/id
for ali the countries that 1 know of, and provide an appropriate conclusion for this book.
Serge Lang
JEAN BRETTE. We just saw a math course you gave to 9th grade students in a Paris high school. Maybe we can recall first how it ali started.
SERGE LANG. Where, at the Palais? 1
JEAN BRETTE. No, in class. Weil, of course, it ali started at the Palais
de la Dcouverte. Two years ago you gave us a talk there, and 1 asked a
friend of mine if we could go and do sorne mathematics with the kids in
his high school class. Last year, you came back for another talk, again at
the Palais, and there was a polemic about teaching in high schools, in
France and elsewhere, what you thought about the whole situation, and
you recalled your experience about sorne talks you gave in a school in
Canada.
SERGE LANG. Yes, that's right.
JEAN BRETTE. And after this discussion, Miss Chwat asked you to ...
Serge Lang gave three talks at the Palais de la Dcouverte, the main science museum in
Paris, for three consecutive years from 1981 to 1983. These talks were first published in the
Journal of the Palais, and then were put together in a book published in France by Belin in
1984, and in English: The Beauty of Doing Mathematics, Springer-Verlag, 1985.
POSTSCRIPT
125
PATRICIA CHWAT. 1 asked, very naively, if Mr. Lang could come to
my school to talk to my students, and he agreed. He came last year, and

this year he must have liked it, because he came back.
SERGE LANG. That's right, last year you had a really good class, an
eighth grade class, and they ali appeared very smart. lt worked out so weil
that 1 thought 1 would come back this year, and 1 met an eighth and ninth
grade class. It also went very weil.
JEAN BRETTE. But the subjects you talked about were a little difficult.
Usually, one does not give many proofs at the eighth or ninth grade level.
PATRICIA CHWAT. But one can, one can.
SERGE LANG. One can?
PATRICIA CHWAT. Not necessarily in the eighth grade, but in the
ninth or tenth grade one can prove a lot in arithmetic. 1 believe this is
the context which lends itself best to proofs. In geometry, they "see"
things, and so they don't feel the need for proofs so much; also because
previously they have made experiments, constructions, and it ali seems
clear to them. But when they see multiples, divisors, that's something new,
and they feel that to be sure a result is true, for any number, because
there are infinitely many of them, a proof is necessary and contributes to
their understanding.
SERGE LANG. Do you do the proofs? Are they part of the program?
PATRICIA CHWAT. 1 don't know if they are officially on the program,
but 1 certainly do them. The theorems, more or less, are on the program,
and 1 do the proofs.
SERGE LANG. Yes, the theorems are on the program. 1 have seen
books which follow the program, which have the theorems, but which
don't have the proofs, whether in algebra or geometry. For instance, for
the are a of the dise or the vol ume of the bali, or the vol ume of a pyramid,
they've got the formulas, but no proofs. That's what they cali being "on
the pro gram".
PATRICIA CHWAT. But whatever the level, even at the university, you
must accept sorne things without proof, things which you agree will be
proved in later years. You need something to work with, right away.
SERGE LANG. Of course. 1 am not against granting sorne statements
without proof, and sorne proofs can be given later, but the question is:
"What is fundamental, and which proofs must be given, when?"
PATRICIA CHWAT. Yes, but if you manage to give the proof for the
formula giving the volume of the bali in the eighth grade, 1 invite you to
do so.
SERGE LANG. OK, l'Il come.
126
POSTSCRIPT
PATRICIA CHWAT. In the eighth grade? Because that's when they see
the formula for the first time.

SERGE LANG. Weil, no. OK, I agree, you can give them the formula
without proof in the eighth grade, and prove it in the ninth or tenth
grade. But then it's not part of the program any more.
PATRICIA CHWAT. Rather cleverly, spatial geometry is part of the pro-
gram in the eighth grade in physics, and in the ninth grade in math.
Mathematicians think that children are not sufficiently mature to understand these notions in the eighth grade. The people responsible for making up the physics program decided otherwise. Then spatial geometry
disappears, until the children are 16 years old. Hence they don't see
proofs of these formulas at ali.
SERGE LANG. But even in physics, they don't see the proofs.
PATRICIA CHWAT. That's true. I myself saw them only at the univer-
sity.
SERGE LANG. There is the scandai! Those proofs are very beautiful,
it's real mathematics. They allow you to appreciate mathematics, to see

why something is true by using arguments which are quite understandable. I did it to students in the ninth, tenth and eleventh grade, and they
understood, they got a lot of pleasure out of it. But it's not officially on
the program, and it's not usually done at ali.
PATRICIA CHWAT. I don't want to defend the programs systematically,
but still you need to teach a large number of notions. What you did is not
in the program, it's like folklore, it's very exceptional.
SERGE LANG. It's exceptional because it's not in the program, but it
would be natural if it were on the program. Ali you have to do is make
up programs where it would be natural to give such proofs.
JEAN BRETTE. Do you want to make one up?
SERGE LANG. No, I don't want to make up a program, but I wrote a
geometry book. 2 I know what should go into a program. If it isn't there,

that's not a law of nature, it's because the people who make up the programs have decided one way rather then another.
PATRICIA CHWAT. Of course, of course. One of the problems is that
those who make up the frograms are not actually teaching. As long as
there will be "inspectors" who don't teach, there won't be coherent programs.
Geometry, a High School Course, in collaboration with Gene Murrow, Springer- Verlag,
1983.
3
The French system of education is very centralized, and based in part on a system of
"Inspectors" who make up programs and see to it that they are followed.
POSTSCRIPT
127
SERGE LANG. [Laughing] So you have to fight a bureaucracy, you
have to fight a system, here as weil as in the United States. But that's
what one has to challenge, one has to try to do something else.
JEAN BRETTE. As far as you know, the situation is the same every-
where?
SERGE LANG. Yes, it's the same everywhere.
JEAN BRETTE. You mentioned one possible exception, 1 think? Where
it was perhaps somewhat better?
SERGE LANG. 1 don't know exactly what happens in the Soviet Union.
1 have seen sorne books, elementary books, which seemed to me better

than those which 1 saw here.
JEAN BRETTE. What do you mean by "elementary"? Eighth grade?
Ninth grade?
SERGE LANG. Yes, and tenth grade. They seemed better, but 1 haven't
seen enough of them to have a weil documented judgment.

PATRICIA CHWAT. Anyway, books about mathematics right now are
changing. Sorne of them are old-fashioned, and are similar to those which
my teachers had when they were students, but others reflect a new spirit
and force the student to think more, to raise questions. But when you can,
there is nothing that prevents you from doing things which are not
officially in the program. ln my courses, 1 don't say that 1 spend most of
my ti me doing things which are not in the pro gram, but still, 1 do things
at the edge, and try to do extra material, while trying to teach them whatever they have to know officially.
SERGE LANG. Not in the program ... But you can do everything
that's in the program in a coherent framework. That's what 1 abject to,

that's what drove me out of my mind last year, just before the conference
at the Palais, to see a text for the Il th grade which was incoherent from
beginning to end, with an accumulation of meaningless exercises. For
instance, in geometry, you try to find the length of the circle, in the ninth
grade, starting with the area of the dise. You need a formula which gives
the expansion of (r + h) 2, what they cali an identity. OK, you need it, and
you have to do it anyhow, whether it's officially on the program or not.
You need that (r + h) 2 is equal to r 2 + 2rh + h 2 .
JEAN BRETTE. lt's on the program for the lOth grade.
SERGE LANG. But you can do it before, what does it matter, do it
whenever you need it. ln the long run, you end up doing what is in the
program, but you do it in a coherent framework, following a beautiful
phrase, like a musical phrase, which hangs together. lnstead of little
things, one after the other, piled up without rhyme or reason. That's what
1 protest against, this accumulation of little things, which don't fit into a
larger pattern which would help to remember them. [Turning to Stephane,
14 _years old.] Wh at do you say?
128
POSTSCRIPT
STEPHANE. 1 agree completely, because they give us exercises with
numbers, but they don't mean anything. They give us numbers, and we
have to add, subtract, that's ali we have to do.
SERGE LANG. So you mean it's artificial?
STEPHANE. Yes. Besides, it's been two years that we have done almost
nothing else, except a little geometry. Just checking that we know how to
cou nt.
PATRICIA CHWAT. But you must work out basic exercises to get
acquainted with mathematical notions.
SERGE LANG. 1 never said you should eliminate basic exercises. The
negation of one extreme is not the extreme of opposite type! Don't claim
that 1 said what 1 did not. 1 don't want to eliminate ali systematic exercises. You need those too, but the programs have eliminated the coherent
stuff, the beautiful musical phrases in mathematics.
PATRICIA CHWAT. You can always put them back in.
SERGE LANG. Yes, you can put them in a book, but then sorne teachers will tell you that it isn't officially on the program, so they can't do
them.
JEAN BRETTE. Or that they don't have the time to deal with them.
SERGE LANG. That's it, "1 don't have the time to do it''. But that's
totally false, they could do it in addition to what's on the program.

JEAN BRETTE. [To Patricia Chwat.] And it isn't true, because you
manage to do both.
PATRICIA CHWAT. Yes, because one needs to establish a certain con-
tinuity, and it bothers me that the students haven't seen certain things.
But they need sorne continuity in what they learn, and to go from one
grade to the next, they need to have acquired sorne basic knowledge
which will be expected of them when they more from the eighth to the
ninth grade.
SERGE LANG. But you can learn this basic knowledge in a coherent
framework. 1 claim that to do basic material in a coherent framework is
not incompatible. Whereas the official programs are designed as if there
were sorne incompatibility. lt's just not true. lt reminds me of Stephane's
notebook . which he showed me earlier. What did your teacher write?
STEPHANE. "Not in the program." 1 was not allowed to use a square
root because it had not been officially discussed in class.

SERGE LANG. And you got a bad grade because of this?
STEPHANE. Yes.
SERGE LANG. Weil, it's ridiculous, that's ali.
PATRICIA CHWAT. He is not in my school. [Laughter.]
POSTSCRIPT
129
JEAN BRETTE. I t's another school in the same district, but we won't
say which one out of charity.
SERGE LANG. But it's scandalous! To give a bad grade to sorne kid
because he did something very coherent, very intelligent, which showed

he understood what was going on very weil, but it just happened not to be
on the program. It's insane to grade that way, but it happens. That's what
disgusts kids from doing mathematics.
JEAN BRETTE. It depends more on the teacher than on the program,
in this instance.
SERGE LANG. It depends on both. They are not mutually exclusive. It
isn't because one is lousy that the other is good. The other could be good
if they wanted, but they reinforce each other. Naturally, there are sorne
teachers who don't know what they are doing and can't do better than
what they do. There are many of those. There are also good ones, very
good ones.
PATRICIA CHWAT. Now you raise the more general problem of train-
ing math teachers in France.

SERGE LANG. Not only in France. In the United States too, I know
the situation better over there. In the teachers' colleges, for the most part
in mathematics, it's a disaster.
PATRICIA CHWAT. In France, we have the IREM.
SERGE LANG. What's that?
JEAN BRETTE. Research Institutes for the Teaching of Mathematics,
which were spread ali over, and which are now dying because they are
not properly funded any more. There aren't enough subsidies to allow
teachers to attend the special courses, to allow them to teach less so they
have time to learn new things.
PATRICIA CHWAT. They started when there was a big reform in high
school teaching, and when the "new math" started. Now teachers are supposed to have been ali recycled.
JEAN BRETTE. . . . and so the government thinks that the IREM don't
serve any purpose any more.

another stupidity, "new math", so called
"modern" math, or "old" math. Those categories don't exist. There is
good math, and lousy math, which are not interesting. OK, in what they
used to do before the "new math", there was also sorne incoherence, little
things which did not mean anything. But they replaced old fashioned
meaningless little things by new-fangled little things which still don't
mean anything. So what's the improvement?
SERGE
LANG. That's
JEAN BRETTE. "Modern math" is not so new as you think. Here I

have a book, an old book, dating back to 1746, where they refer to
130
POSTSCRIPT
Modern Mathematics, with a capital M for "Math" and "Modern".

Polemics on pedagogy don't seem to have changed much.
SERGE LANG. But there should not be polemics on words like that. It
can't lead to anything constructive. You should not compare modern

mathematics with ancient mathematics. That's not the proper approach.
They are not opposed to each other. What are in opposition are coherent,
beautiful mathematics, instructive and useful mathematics, with
mathematics that aren't good for anything, or rather which are good only
to turn off the kids. That's the opposition. Just using words like modern
math or old math already prejudices the discussion along the wrong channels. 1 am not for Euclid or against Euclid. Take in Euclid whatever is still
useful, and forget about a lot of it which isn't. Of course, Euclid piles up
hundreds of pages, but today we can manage to cover what is important
in thirty or forty pages, and that's it. You go on to something else which
was discovered later in the history of mathematics. We have another perspective. What do you say, Stephane?
STEPHANE. 1 agree. Perhaps the new math is useful, but what's the
point of doing it for a year or a year and a half!

SERGE LANG. What do you mean by the "new math"?
STEPHANE. Weil, union, intersection, ali of that. What you draw with
ci rel es.
SERGE LANG. There you are, sorne success! lt shouldn't be done that
way. It should just be an easy language, which is used and developed as

you need it. There is no point making a big fuss about it.
PATRICIA CHWAT. There is sorne movement, today, to reduce the
amount of vocabulary that students have to swallow.

SERGE LANG. 1 repeat: The vocabulary can be developed as you need
it. If you spend weeks or months building up a vocabulary which is not

used right away, but is an end in itself, then again it's a little thing. It's a
"new" thing, but it's just as little, eut up in little slices, without nice
"phrases", without beauty. There is nothing except an accumulation of little things. That is what should be eliminated. And it's not incompatible
with learning basic notions and techniques. Your guy there, in the book
from 1700, he saw it very weil. What did he write?
JEAN BRETTE. Yes, for instance, he has a footnote here, which said
just that:
General/y one should never give a definition to children without previously
having shown them the thing that was being defined. The name should
Institutions de Geometrie, by M. De La Chapelle. Vol. 1, 1746.
POSTSCRIPT
131
come on/y after the idea, because the name was used only to evoke the
idea.
SERGE LANG. That's it.
JEAN BRETTE. Yes, and there are lots of other instructive comments in
this book. The author wanted to persuade his contemporaries that one
could teach mathematics to very young children, and he had given a great
deal of thought on how to do this.
SERGE LANG. What's the guy's name?
JEAN BRETTE. M. de La Chapelle. 1 don't know much about him, but
think he was a friend of d'Alembert, and his book had the official
approval of the Academy of Sciences. We don't have the time to read it
now, but it might be worth while to publish sorne parts of it, in contemporary french.
SERGE LANG. Let me come back to a concrete case, because to be
useful, a discussion must be based on concrete cases. 1 am not proposing a

new ideology. If you put a label on something, you prejudice the discussion. Let's come back to the con crete case of (r + h ) 2. Not ali the kids
know the formula for this, and it's like many things in mathematics. On
one hand you need to understand theoretically what happens, and you
must see the proof. On the other hand, you must also know how to use
the formula automatically when you need it. There is no way to avoid
this, and so you must ask the kids to repeat the formula ten times:
JEAN BRETTE. Ali together?

SERGE LANG. Definitely, ali together, like a chorus. 1 make them do
this. Do you?
PATRICIA CHWAT. Sometimes 1 make them recite something like that,
but rarely. 1 don't make a habit of it.

SERGE LANG. When 1 was a kid, that's how 1 learned (a
b) 2, and 1
never forgot it. That's how my teacher taught it to me, and that's how it
must be done. There is no other way. lt must be driven into their ears like
music. You shouldn't ask every time why the formula is true! You must be
able to use it when you need it. And the only way-or rather: 1 never
heard of another way-is to repeat the formula ten times, and again ten
times before going to bed. 1 tell them: repeat the formula. At first they
don't take it seriously, but they repeat it with me, like a chorus, and it
amuses them. Stephane, that's the way 1 did it with you. What did 1 ask
you to repeat?
STEPHANE. The volume of the bali.
132
POSTSCRIPT
SERGE LANG. Right, we had proved the formula, but you did not
remember it at first. Then what did 1 make you do?

STEPHANE. Weil, repeat 4/3 of 7Tr 3 .
SERGE LANG. Exactly. Two weeks ago. And now you will remember it
till death do us part?
STEPHANE. Yes, 1 think so.
SERGE LANG. So, you must learn the formula by heart by repeating it
ten times, but you must also see the proof. Because there are two different
ways the brain works, but they are definitely not incompatible, they are
complementary to each other. You must know both. And it's the same
thing at Yale, with university students who are 17 and 18 years old. Sorne
of them, too many, are not able to repeat this formula automatically when
they come from high school, and 1 ask them. Maybe a third, or half the
class. 1t's a scandai!
JEAN BRETTE. lt isn't worse than here.
SERGE LANG. lt's the same, no doubt about it. And even though they
are older, like 17 or 18, 1 still make them repeat the formula ten times.
[Laughter.] First they think 1 have it in for them, but after a few da ys,
they end up by understanding that 1 am right~ that it's an effective way to
learn, and that they have to use it on certain occasions. But they must also
see the proof, and understand how to formulate it. They need both. Afterward, they make jokes about the fact that 1 make them repeat ten times
the same thing.
JEAN BRETTE. You have gone through several experiences of high
school teaching, around Paris as weil as in Canada, at Toronto. How do

you appreciate the students' reaction to what you do, which is usually not
in their standard program, or what they usually do?
SERGE LANG. You have seen how the students react.
JEAN BRETTE. Yes, we have seen sorne examples, but you, how do
you appreciate it?

SERGE LANG. lt's a pleasure for me! They open up, they react positively, they are immediately interested, and they start taking notes. After
that, 1 think you asked them ...
PATRICIA CHWAT. Yes. They have a special homework every two
weeks, and the other day, 1 asked them to write up what you did on the
length of the circumference of the circle.
SERGE LANG. 1 saw their homework, which you brought today, and
many of them were able to reproduce everything. You did not help them?
PATRICIA CHWAT. No. Of course, not ali of them remembered every-
thing, not ali the details, but almost ali of them remembered the general
idea for the proof.
POSTSCRIPT
133
SERGE LANG. 1 saw how they wrote it up. It was wonderful. It was
neat, coherent, just great.

PATRICIA CHWAT. They certainly were neat.
SERGE LANG. lt was not only neat, it was also coherent. There were
sentences, explanations, weil thought out expositions. They learned certain

formulas, so they learned sorne algebra, they learned a proof, and they
learned how to use language to express their mathematical thoughts. That
must come at the end. When we are doing math together, at first 1 start a
sentence, and 1 ask them to complete it with one or two words, or to finish
a sentence which 1 myself began. But as we go further along, 1 ask them
to create their own sentences, to explain their own mathematics. Then at
sorne point, 1 reverse the process, and 1 ask the students to express themselves what's in their heads, and to extract by themselves their own
mathematical thoughts from their own head. 1t's for the student to start
his own sentence. You saw it with Yaelle, at the end, when 1 asked her to
repeat the proof which gives the length of the circle, c == 2'1T r. That's
exactly what she did. With her hands, with her eyes, with her brain, she
did it.
JEAN BRETTE. We almost heard her think.
SERGE LANG. We saw her think. The expression on her face, her con-
centration ...
PATRICIA CHWAT. It's true, 1 have gotten them to talk. We haven't
mentioned this yet, but in my classes, they are used to speaking up, finishing sentences, and then making up their own sentences. In your case, it
wasn't the same persan in front of them, but it was not a technique fundamentally different from what they often see in class.
SERGE LANG. 1 know university students who are not able to do that
in mathematics. Maybe half, in an average class, can't do it. Of course,
there are sorne honour classes, brilliant classes, but in an average class in
the United States ... ali they are required to do is to fill out a little
square. They are given a problem, with a numerical answer, and ali they
are asked to do is to put a number in a little box. That they usually can
do weil. But to create their own sentences, to explain by themselves what's
going on, to write up a proof, they can't do it. It's a scandai! And there
are many classes, here in France-1 don't mean yours, because you do
better than that [Laughter]-what about in your own class? Stephane?
STEPHANE. lt's the same thing. Usually they just ask for a result. Weil,
sometimes they ask us to write something up, but not often.
JEAN BRETTE. One shouldn't wonder that students can't write things
up if their teachers never ask them to learn proofs.

SERGE LANG. Of course, but it helps to clarify ideas, to write them up,
and the subject becomes much more interesting for students if one asks
134
POSTSCRIPT
them to furnish this effort, which anyway is not so great. It gives them
pleasure. Listen, when Yaelle repeated the whole proof, ali at once, you
saw what happened, the whole class applauded spontaneously. [p. 89, The
length of the circ le.]
PATRICIA CHWAT. But you happened to pick on Yaelle, it would not
have gone exactly the same way for every student in the class. It was clear
it would give her pleasure. Others would not have taken it so weil.
JEAN BRETTE. What do you mean, being picked on?
PATRICIA CHWAT. Yes, to redo the whole proof.
JEAN BRETTE. But there is also the example of Rachel, in Toronto.
SERGE LANG. That was another case. In Toronto, there was one stu-
dent who would not have been able to do what Yaelle did, reproduce the
whole proof, and I had to lead her through it step by step. That's true,
mathematical talents differ, and ali students are not able to do it, the first
time, but if I met them again for another hour the following week, many
would become able to do it.
PATRICIA CHWAT. But it's a question of time, and if we had enough
time, we could retrieve practically everybody, except in pathological cases.

But we have 25 students, sometimes more; sorne students are already very
mu ch be hi nd, and unfortunately, we have to leave them like that. I t's a
dramatic situation, but ...
JEAN BRETTE. I agree, but on the other hand, an exchange like the
one with Yaelle was valuable for the whole class, because there was a certain psychological tension, there was suspense, the others were waiting to
see if she would succeed in reconstructing everything, and they were following along at about the same speed. It benefited everybody.
PATRICIA CHWAT. But sorne of them were quite glad that they were
not picked on, and probably were thinking about something else, you
have to be honest.
JEAN BRETTE. Possibly, I am not saying that everybody was follow-
ing, but an exchange with so much tension, so much concentration ...

PATRICIA CHWAT. For sure, Yaelle was carried by the class, she was
encouraged, but sorne of the others were very glad to be doing only that.
SERGE LANG. I would have handled another student differently, I
would adjust to the ability of the student, if I picked someone else. I don't
do exactly the same thing with different students. If I realized that a student was less able then Yaelle to repeat the proof, 1 would have handled
the discussion differently, just as I did it with Rachel in Toronto: first I
had to lead her step by step to repeat the proof, but she ended up by saying more than she knew in the beginning. I pick on them at random. I
don't take those who raise their hand, 1 pick on those who don't.
POSTSCRIPT
135
PATRICIA CHWAT. There is a problem with that. One of my students
was very disappointed because she had made an effort to raise her hand,
she is rather shy, and when she raised her band, you did not ask her.
JEAN BRETTE. She was in the first row?
PATRICIA CHWAT. She sat next to Yaelle. She is not very gifted in
mathematics, but in this case she wanted to show you that she could do
something, and you did not ask her just because of that. [Laughter.]
SERGE LANG. Yes, OK, there are risks, I don't know the students, and
I see them only once ... What I wanted to avoid was that only those who
usually know what's going on should answer. It's very easy to fall into this
trap. Ail situations can arise. There are those who are not able to repeat
the proof by themselves, and there are those who are able, even if they
don't think so. Ail cases are possible. What is needed is to bring them ali
to the maximal level of understanding.
PATRICIA CHWAT. And that requires time.
SERGE LANG. Yes, it requires time, but it's a long term investment.
Once they have learned how to do it in one case, they will do it much
more easily in the next case. The time is not lost.
JEAN BRETTE. This goes back to what you were saying a while ago. If
you eliminate from the programs what you think is unnecessary, then it
gives time to do the rest.
SERGE LANG. Exactly, that's right. If you clear the deadwood from the
programs, and if you construct them coherently, then it leaves plenty of

time to do beautiful mathematics, which will be useful, with ali the necessary proofs, and whatever formulas you have to know automatically, just
like that.
JEAN BRETTE. Stephane, you were sitting in the back of the class, as a
guest. What do you think?

STEPHANE. It was OK, but perhaps if it happened ali the time, one
might get tired of it, the students might be less awake, in the long run.
SERGE LANG. Oh you know, at the university, even with 17 year olds,
I know how to wake them up. It suffices to do this [pointing the finger] and
I guarantee you that it wakes them up. Besides, they never know when it's
going to hit them. [Laughter.] I t's not a big problem to keep them awake.
Besides, I can keep them awake by showing them real mathematical
problems.
PATRICIA CHWAT. But in the 8th or 9th grade, they're too young. Stu-
dents in the 8th or 9th grade know almost no mathematics.

SERGE LANG. Yes, but still there are certain tapies ... when do you
do prime numbers, in what class?

PATRICIA CHWAT. In the 8th grade.
POSTSCRIPT
136
SERGE LANG. Then in the 9th grade, they are a little more mature.
PATRICIA CHWAT. Yes, but then we don't talk about this any more.
SERGE LANG. You don't talk about it any more? Here we come back
to the same problem we met before. You state a result in the 8th grade,
and after that, it's over, you don't follow it up.
PATRICIA CHWAT. You use prime numbers just to write rational
numbers in the 9th grade. You have to know simple fractions, but conceptually, there is no progress. You use notions which were introduced in the
8th grade.
SERGE LANG. Ail right, but 1 can tell you a problem about prime
numbers which could be discussed in the 8th grade, just when you define
prime numbers. First you can ask them if there is a infinity of them, and
you can prove it. At this point, you tell them: look at the twin primes,
3, 5; 5, 7; 11, 13; and the next one, Stephane?
STEPHANE. 17, 19.
SERGE LANG. Right. And you can raise the problem of the twin
primes, are there infinitely many? And you see how they react. They can
have fun with this, and find a number of them, maybe up to 100. It will
make them work with integers, addition, subtractions, divisions. But they
will do so in a meaningful context, to answer a question. After they have
computed like that, up to 100, they will observe that the list of twin
primes seems to go on, and you ask them: "Is there an infinite number of
them?" Then you'll see what they answer. Sorne will say yes, sorne will
say no, then you ask them why, for what reasons, what is their intuition,
and after a while, the same day maybe, you let them go home and compute more of them, so that they arrive at a more sophisticated conclusion.
And a week later, after sorne suspense-and there is suspense: Is there or
is there not an infinity of twin primes-you tell them: "Nobody knows the
answer! And if you succeed to prove it, you make it in the history of
mathematics."
PATRICIA CHWAT. Yes, but to raise this kind of problem, there is one
necessary condition, you have to know them. You need a certain
mathematical culture.
SERGE LANG. But for god's sake, what's the use of teacher's colleges,
which are supposed to train the teachers?

PATRICIA CHWAT. But there aren't any!
SERGE LANG. So it's for society to figure out how to give this training.
Instead of putting ali sorts of stupid things in their books, they should put
this in. lt will make the students work and think. They will learn to divide
just as weil, they will get trained in the basics just as weil, but starting
from problems which have sorne meaning, which are coherent, and quite
beautiful! People re act very positively to su ch a problem. And it's the
POSTSCRIPT
137
same thing for sorne equations. When you do x and y, you can do what 1
did in my lecture at the Palais de la Dcouverte 5 or in Toronto.
PATRICIA CHWAT. By the way, sorne of my students got a reprint of
the talk which you gave last year, and on the front page, in big type you
have the equation y 2 == x 3 + l. During my next math class, Yaelle asked
me what is an equation. 1 had not yet done this (even now, 1 have not
done it in class ), but 1 explained this mathematical ter rn to her, th at is the
notion of an equation; and Yaelle told me: "Ah, that's an equation, so it's
very easy, and 1 thought it was something very complicated." And
Nathalie, the one you did not want to question, tried to find solutions,
and she said: "Weil, let y == 3 and x == 2."
SERGE LANG. There you are, 1 have nothing to add. [Laughter.]
JEAN BRETTE. And then, did she read the whole conference?
PATRICIA CHWAT. 1 don't know, 1 did not ask her, but both were very
disappointed to see that in the next homework, there was no equation to
solve. 1 told them that the problem did not consist only in finding solutions, but also in showing that there were no others.
JEAN BRETTE. But you have to start by finding sorne, there are two
problems.
PATRICIA CHWAT. What 1 mean is that for them, it was agame.
SERGE LANG. Absolutely. That's a marvelous reaction, because if a
child reacts that way, then she is caught by the problem, and then you can
develop it and learn even what's officially on the program, but in the context of this problem.
PATRICIA CHWAT. Yes, but one has to find contexts which are
sufficiently rich to allow y ou to encom pass ...

SERGE LANG. There will be algebra, there will be sorne geometry,
there will be coordinates, the whole thing will come together. But you are
then motivated by a larger structure, a musical phrase, a mathematical
phrase, which is a beautiful phrase ...
JEAN BRETTE. You really like music!
SERGE LANG. Definitely, 1 like music. Yes, because 1 have to com-
municate, and if 1 say "music", it stimulates people. They are not used to
think in similar terms for music and mathematics. But not in a trivial way,
not by counting seconds, or the number of vibrations. That's not what 1
mean by a musical phrase and its mathematical analogue, the mathematical theme which can be developed.
Mathematics is not just "numbers", any more than music is notes.
Music, that's what happens when you enjoy it, when you get a high out of
The Beau~ v (?l DoinR Mat hematies, Springer- Ver1ag, 1985.
138
POSTSCRIPT
it. You can get a high with mathematics. You just said so, when Nathalie
found a solution for her equation, she was very happy.
PATRICIA CHWAT. Me too, naturally.
JEAN BRETTE. Yes, and what's nice is that this conference, you can
read a good half of it before you reach a point which you don't understand, which will give you trouble.
PATRICIA CHWAT. Last year, 1 had used your reprint on prime
numbers 5 and we had looked at it during the last class, as a reward, and
we had gone quite deeply into it. Of course, we had sorne trouble with
logarithms, because in the 8th grade, you know, it's tough.
SERGE LANG. Weil, naturally.
PATRICIA CHWAT. In any case, this talk was not addressed to 8th
grade students.
JEAN BRETTE. It wasn't addressed to anyone in particular. lt was
addressed to those who were there, and if at sorne point they stop understanding, they have already gone that far, and what's more, they know
that it's not the end of it. It's true that rarely if ever in elementary teaching does one tell the students that mathematics is not something which is
finite, which was invented two thousand years ago, or two centuries ago,
or fifty years ago. It makes people feel good to know that great mathematicians also don't know certain things and that there are problems which
nobody knows how to solve.
PATRICIA CHWAT. lt's good for other people's morale!
SERGE LANG. For one thing, but for another, 1 said it explicitly in one
of the classes, when 1 told them that ali those theorems, someone had to
discover them, maybe a thousand years ago, or two hundred years ago,
and then one of the kids said: "Weil, then, what you discover today, it will
be taught to us in a class in two hundred years." Do you remember? So
they understood the human side of this thing, because one of them asked
what 1 did in life, and 1 said that 1 do research, 1 discover theorems. And
she said: "And what happens if you can't discover theorems any more?"
And one of the kids behind her shouted: "Resign, Resign!" [Laughter.]
They sure got the idea, they understood, and mathematics became alive
for them.
PATRICIA CHWAT. 1 tell them that mathematics exists, that mathemat-
ics is alive, and not stereotyped, old and musty, but it's not always easy
with very young children.
SERGE LANG. You are right. Each teacher must do according to his
own way, his own taste. Each one must use their own means to excite the
studen ts. One needs everything, wi thou t excl usi vi ty.
JEAN BRETTE. Here is a perfect conclusion: One needs everything,
without exclusivity.

(Serge Lang) MATH Encounters With High School Students

Uploaded by

Copyright:

Available Formats

(Serge Lang) MATH Encounters With High School Students

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Serge Lang) MATH Encounters With High School Students

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What did the students find rewarding about solving equations?

What did the students find rewarding about solving equations?

Serge Lang

With 103 Illustrations

AMS Classifications: OO-A05, OO-A06

Library of Congress Cataloging in Publication Data

Typeset by House of Equations, Inc., Newton, New Jersey.

Who is Serge Lang?

Dear Christopher, Rachel, Sylvain, Yaelie,

Volumes in higher dimension

The volume of the bali

The length of the circle

The area of the sphere

Who is Serge Lang?

Dear Christopher, Rachel, Sylvain,

I am writing this to you because I came to schools like yours, in France as

DEAR CHRISTOPHER, RACHEL, SYLVAIN, YAELLE, AND ALL THE OTHERS

The following talk was given at a high school in the suburbs of

SERGE LANG. My name is Serge Lang, 1 usually teach at Yale, but

today 1 came here to do mathematics with you.

If y ou eut a rectangle in half, like this:

where b is the base and h is the height.

Do you know what is the formula for its area?

SERGE LANG. That's right, it's wr 2 . Weil, what is pi?

[The students rea ct variously.]

mean, how are you going to compute it?

twice the radius.

mention the circumference. We call the circumference c. And what did

That's what you said: 2r is the diameter, so the circumference is pi times

By the way, what's your name?

equals the circumference divided by

the diameter, and work it out.

diameter, that equals

So we have to start from something, with a definition, otherwise 1 can't

[Another student starts ta/king.]

gle is the product of the sides. Right?

N ow suppose 1 take a rectangle with three times the si des. So he re 1 have

ADOLPH. lt's ab over 4.

Adolph, what's the area of this rectangle?

a factor of r, how does the area change? Adolph.

it? Does everybody see it?

rectangle changes by a factor of r 2. And of course, r can be bigger than 1,

Ali right, now instead of a rectangle, suppose 1 have another figure.

SERGE LANG. Yes, the area changes by 2 squared. If I blow up the

rectangle by a factor of r, the area changes by a factor of r 2. Suppose I

measure around, 1 get the perimeter, the circumference. 1 am now dealing

[A student raises her hand. There is a certain amount offumbling around

A STUDENT. Yes, if you blow up these rectangles inside the kidney by

SERGE LANG. That's right. And if 1 blow up by a factor of r, the area

ali the way up to A 10.

Now if one of these rectangles has surface area A, as she said-what is

is an approximation of the area of the kid-

is an approximation of the area of the kidney dilated by r, so it is an

the rectangles? If I blow up each rectangle by a factor of r, its area

[Laughter, the statement

A STUDENT. I have a question. If r was less than 1, would it reduce by

a factor r to the negative 2?