Taming

the Infinite
The Story of Mathematics
 Contents
Preface
1. Tokens, Tallies and Tablets
2. The Logic of Shape
3. Notations and Numbers
4. Lure of the Unknown
5. Eternal Triangles
6. Curves and Coordinates
7. Patterns in Numbers
8. The System of the World
9. Patterns in Nature
10. Impossible Quantities
11. Firm Foundations
12. Impossible Triangles
13. The Rise of Symmetry
14. Algebra Comes of Age
15. Rubber Sheet Geometry
16. The Fourth Dimension
17. The Shape of Logic
18. How Likely is That?
19. Number Crunching
20. Chaos and Complexity

Further Reading

Index

Acknowledgements
 Preface

Mathematics did not spring into being fully formed. It grew from the
cumulative efforts of many people, from many cultures, who spoke many
languages. Mathematical ideas that are still in use today go back more than 4000
years.
Many human discoveries are ephemeral – the design of chariot wheels was
very important in New Kingdom Egypt, but it’s not exactly cutting-edge
technology today. Mathematics, in contrast, is often permanent. Once a
mathematical discovery has been made, it becomes available for anyone to use,
and thereby acquires a life of its own. Good mathematical ideas seldom go out of
fashion, though their implementation can change dramatically. Methods for
solving equations, discovered by the ancient Babylonians, are still in use today.
We don’t use their notation, but the historical link is undeniable. In fact, most of
the mathematics taught in schools is at least 200 years old. The advent of
‘modern’ mathematics curricula in the 1960s brought the subject into the 19th
century. But contrary to appearances, mathematics has not stood still. Today,
more new mathematics is created every week than the Babylonians managed in
two thousand years.
The rise of human civilization and the rise of mathematics have gone hand in
hand. Without Greek, Arab and Hindu discoveries in trigonometry, navigation
across the open ocean would have been an even more hazardous task than it was
when the great mariners opened up all six continents. Trade routes from China to
Europe, or from Indonesia to the Americas, were held together by an invisible
mathematical thread.
Today’s society could not function without mathematics. Virtually everything
that we now take for granted, from television to mobile phones, from wide-
bodied passenger jets to satellite navigation systems in cars, from train schedules
to medical scanners, relies on mathematical ideas and methods. Sometimes the
mathematics is thousands of years old; sometimes it was discovered last week.
Most of us never realize that it is present, working away behind the scenes to
empower these miracles of modern technology.
 This is unfortunate: it makes us think that technology works by magic, and
leads us to expect new miracles on a daily basis. On the other hand, it is also
entirely natural: we want to use these miracles as easily and with as little thought
as possible. The user should not be burdened by unnecessary information about
the underlying trickery that makes the miracles possible. If every airline
passenger were made to pass an examination in trigonometry before boarding
the aircraft, few of us would ever leave the ground. And while that might reduce
our carbon footprint, it would also make our world very small and parochial.
To write a truly comprehensive history of mathematics is virtually impossible.
The subject is now so broad, so complicated and so technical, that even an expert
would find such a book unreadable – leaving aside the impossibility of anyone
writing it. Morris Kline came close with his epic Mathematical Thought from
Ancient to Modern Times. It is more than 1200 pages long, in small print, and it
misses out almost everything that has happened in the last 100 years.
This book is much shorter, which means that I have had to be selective,
especially when it comes to mathematics of the 20th and 21st centuries. I am
acutely conscious of all the important topics that I have had to omit. There is no
algebraic geometry, no cohomology theory, no finite element analysis, no
wavelets. This list of what’s missing is far longer than the list of what is
included. My choices have been guided by what background knowledge readers
are likely to possess, and what new ideas can be explained succinctly.
The story is roughly chronological within each chapter, but the chapters are
organized by topic. This is necessary to have any kind of coherent narrative; if I
put everything into chronological order the discussion would flit randomly from
one topic to another, without any sense of direction. This might be closer to the
actual history, but it would make the book unreadable. So each new chapter
begins by stepping back into the past, and then touching upon some of the
historical milestones that were passed as the subject developed. Early chapters
stop a long way in the past; later chapters sometimes make it all the way to the
present.
I have tried to give a flavour of modern mathematics, by which I mean
anything done in the last 100 years or so, by selecting topics that readers may
have heard of, and relating them to the overall historical trends. The omission of
a topic does not imply that it lacks importance, but I think it makes more sense
to spend a few pages talking about Andrew Wiles’s proof of Fermat’s Last
Theorem – which most readers will have heard of – than, say, non-commutative
geometry, where the background alone would occupy several chapters.
In short, this is a history, not the history. And it is history in the sense that it
tells a story about the past. It is not aimed at professional historians, it does not
make the fine distinctions that they find necessary, and it often describes the
ideas of the past through the eyes of the present. The latter is a cardinal sin for a
historian, because it makes it look as though the ancients were somehow striving
towards our current way of thinking. But I think it is both defensible and
essential if the main objective is to start from what we now know and ask where
those ideas came from. The Greeks did not study the ellipse in order to make
possible Kepler’s theory of planetary orbits, and Kepler did not formulate his
three laws of planetary motion in order for Newton to turn them into his law of
gravity. However, the story of Newton’s law rests heavily on Greek work on the
ellipse and Kepler’s analysis of observational data.
A sub-theme of the book is the practical uses of mathematics. Here I have
provided a very eclectic sample of applications, both past and present. Again, the
omission of any topic does not indicate that it lacks importance.
Mathematics has a long, glorious, but somewhat neglected history, and the
subject’s influence on the development of human culture has been immense. If
this book conveys a tiny part of that story, it will have achieved what I set out to
do.

COVENTRY, MAY 2007

 1
Tokens, Tallies and Tablets

The birth of numbers

Mathematics began with numbers, still fundamental, even though the subject
is no longer limited to numerical calculations. By building more sophisticated
concepts on the basis of numbers, mathematics has developed into a broad and
varied area of human thought, going far beyond anything that we encounter in a
typical school syllabus. Today’s mathematics is more about structure, pattern
and form than it is about numbers as such. Its methods are very general, and
often abstract. Its applications encompass science, industry, commerce – even
the arts. Mathematics is universal and ubiquitous.

It started with numbers

Over many thousands of years, mathematicians from many different cultures

have created a vast superstructure on the foundations of number: geometry,
calculus, dynamics, probability, topology, chaos, complexity and so on. The
journal Mathematical Reviews, which keeps track of every new mathematical
publication, classifies the subject into nearly a hundred major areas, subdivided
into several thousand specialities. There are more than 50,000 research
mathematicians in the world, and they publish more than a million pages of new
mathematics every year. Genuinely new mathematics, that is, not just small
variations on existing results.
Mathematicians have also burrowed into the logical foundations of their
subject, discovering concepts even more fundamental than numbers –
mathematical logic, set theory. But again the main motivation, the starting point
from which all else flows, is the concept of number.

‘Numbers seem very simple and straightforward, but

appearances are deceptive.’

Numbers seem very simple and straightforward, but appearances are

deceptive. Calculations with numbers can be hard; getting the right number can
be difficult. And even then, it is much easier to use numbers than to specify what
they really are. Numbers count things, but they are not things, because you can
pick up two cups, but you can’t pick up the number ‘two’. Numbers are denoted
by symbols, but different cultures use different symbols for the same number.
Numbers are abstract, yet we base our society on them and it would not function
without them. Numbers are some kind of mental construct, yet we feel that they
would continue to have meaning even if humanity were wiped out by a global
catastrophe and there were no minds left to contemplate them.

Writing numbers

The history of mathematics begins with the invention of written symbols to

denote numbers. Our familiar system of digits 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 to
represent all conceivable numbers, however large, is a relatively new invention;
it came into being about 1500 years ago, and its extension to decimals, which
lets us represent numbers to high precision, is no more than 450 years old.
Computers, which have embedded mathematical calculations so deeply into our
culture that we no longer notice their presence, have been with us for a mere 50
years; computers powerful enough and fast enough to be useful in our homes
and offices first became widespread about 20 years ago.
Without numbers, civilization as we now know it could not exist. Numbers are
everywhere, hidden servants that scurry around behind the scenes – carrying
messages, correcting our spelling when we type, scheduling our holiday flights
to the Caribbean, keeping track of our goods, ensuring that our medicines are
safe and effective. And, for balance, making nuclear weapons possible, and
guiding bombs and missiles to their targets. Not every application of
mathematics has improved the human condition.

‘It all began with little clay tokens, 10,000 years ago
in the Near East.’

How did this truly enormous numerical industry arise? It all began with little
clay tokens, 10,000 years ago in the Near East.
Even in those days, accountants were keeping track of who owned what, and
how much – even though writing had not then been invented, and there were no
symbols for numbers. In place of number symbols, those ancient accountants
used small clay tokens. Some were cones, some were spheres and some were
shaped like eggs. There were cylinders, discs and pyramids. The archaeologist
Denise Schmandt-Besserat deduced that these tokens represented basic staples of
the time. Clay spheres represented bushels of grain, cylinders stood for animals,
eggs for jars of oil. The earliest tokens date back to 8000 BC, and they were in
common use for 5000 years.
As time passed, the tokens became more elaborate and more specialized.
There were decorated cones to represent loaves of bread, and diamond-shaped
slabs to represent beer. Schmandt-Besserat realized that these tokens were much
more than an accountancy device. They were a vital first step along the path to
number symbols, arithmetic and mathematics. But that initial step was rather
strange, and it seems to have happened by accident.
It came about because the tokens were used to keep records, perhaps for tax
purposes or financial ones, or as legal proof of ownership. The advantage of
tokens was that the accountants could quickly arrange them in patterns, to work
out how many animals or how much grain someone owned or owed. The
disadvantage was that tokens could be counterfeited. So to make sure that no one
interfered with the accounts, the accountants wrapped the tokens in clay
envelopes – in effect, a kind of seal. They could quickly find out how many
tokens were inside any given envelope, and of what kind, by breaking the
envelope open. They could always make a new envelope for later storage.
However, repeatedly breaking open an envelope and renewing it was a rather
inefficient way to find out what was inside, and the bureaucrats of ancient
Mesopotamia thought of something better. They inscribed symbols on the
envelope, listing the tokens that it contained. If there were seven spheres inside,
the accountants would draw seven pictures of spheres in the wet clay of the
envelope.
At some point the Mesopotamian bureaucrats realized that once they had
drawn the symbols on the outside of the envelope, they didn’t actually need the
contents, so they didn’t need to break open the envelope to find out which tokens
were inside it. This obvious but crucial step effectively created a set of written
number symbols, with different shapes for different classes of goods. All other
number symbols, including those we use today, are the intellectual descendants
of this ancient bureaucratic device. In fact, the replacement of tokens by symbols
may also have constituted the birth of writing itself.

Tally marks

These clay marks were by no means the earliest examples of number-writing,

but all earlier instances are little more than scratches, tally marks, recording
numbers as a series of strokes – such as ||||||||||||| to represent the number 13. The
oldest known marks of this kind – 29 notches carved on a baboon’s leg bone –
are about 37,000 years old. The bone was found in a cave in the Lebombo
mountains, on the border between Swaziland and South Africa, so the cave is
known as the Border Cave, and the bone is the Lebombo bone. In the absence of
a time machine, there is no way to be certain what the marks represented, but we
can make informed guesses. A lunar month contains 28 days, so the notches may
have related to the phases of the Moon.
The Ishango bone showing the patterns of marks and the numbers they may represent.
 Tally marks have the advantage that they can be built up one at a time, over long periods, without altering
or erasing previous marks. They are still in use today, often in groups of five, with the fifth stroke cutting
diagonally across the previous four.

The presence of tally marks can still be seen in modern numerals. Our symbols 1, 2, 3 are derived from a
single stroke, two horizontal strokes linked by a sloping line, and three horizontal strokes linked by a
sloping line.

There are similar relics from ancient Europe. A wolf bone found in former
Czechoslovakia has 57 marks arranged in eleven groups of five with two left
over, and is about 30,000 years old. Twice 28 is 56, so this might perhaps be a
two-month long lunar record. Again, there seems to be no way to test this
suggestion. But the marks look deliberate, and they must have been made for
some reason.
Another ancient mathematical inscription, the Ishango bone from Zaire, is
25,000 years old (previous estimates of 6000–9000 years were revised in 1995).
At first sight the marks along the edge of the bone seem almost random, but
there may be hidden patterns. One row contains the prime numbers between 10
and 20, namely 11, 13, 17 and 19, whose sum is 60. Another row contains 9, 11,
19 and 21, which also sum to 60. The third row resembles a method sometimes
used to multiply two numbers together by repeated doubling and halving.
However, the apparent patterns may just be coincidental, and it has also been
suggested that the Ishango bone is a lunar calendar.

The first numerals

The historical path from accountants’ tokens to modern numerals is long and
indirect. As the millennia passed, the people of Mesopotamia developed
agriculture, and their nomadic way of life gave way to permanent settlements, in
a series of city states – Babylon, Eridu, Lagash, Sumer, Ur. The early symbols
inscribed on tablets of wet clay mutated into pictographs – symbols that
represent words by simplified pictures of what the words mean – and the
pictographs were further simplified by assembling them from a small number of
wedge-shaped marks, impressed into the clay using a dried reed with a flat,
sharpened end. Different types of wedge could be made by holding the reed in
different ways. By 3000 BC, the Sumerians had developed an elaborate form of
writing, now called cuneiform or wedge-shaped.
The history of this period is complicated, with different cities becoming
dominant at different times. In particular, the city of Babylon came to
prominence, and about a million Babylonian clay tablets have been dug from the
Mesopotamian sands. A few hundred of these deal with mathematics and
astronomy, and they show that the Babylonian knowledge of both subjects was
extensive. In particular, the Babylonians were accomplished astronomers, and
they developed a systematic and sophisticated symbolism for numbers, capable
of representing astronomical data to high precision.
The Babylonian number symbols go well beyond a simple tally system, and
are the earliest known symbols to do so. Two different types of wedge are used:
a thin vertical wedge to represent the number 1, and a fat horizontal one for the
number 10. These wedges are arranged in groups to indicate the numbers 2–9
and 20–50. However, this pattern stops at 59, and the thin wedge then takes on a
second meaning, the number 60.
The Babylonian number system is therefore said to be ‘base 60’, or
sexagesimal. That is, the value of a symbol may be some number, or 60 times
such a number, or 60 times 60 times such a number, depending on the symbol’s
position. This is similar to our familiar decimal system, in which the value of a
symbol is multiplied by 10, or by 100, or by 1000, depending on its position. In
the number 777, for instance, the first 7 means ‘seven hundred’, the second
means ‘seventy’ and the third means ‘seven’. To a Babylonian, a series of three
repetitions of the symbol for ‘7’ would have a different meaning, though
based on a similar principle. The first symbol would mean 7 × 60 × 60, or
25,200; the second would mean 7 × 60 = 420; the third would mean 7. So the
group of three symbols would mean 25,200 + 420 + 7, which is 25,627 in our
notation. Relics of Babylonian base-60 numbers can still be found today. The 60
seconds in a minute, 60 minutes in an hour and 360 degrees in a full circle, all
trace back to ancient Babylon.

Babylonian symbols for the numbers 1–59

Because it is awkward to typeset cuneiform, scholars write Babylonian

numerals using a mixture of our base-10 notation and their base-60 notation. So
the three repetitions of the cuneiform symbol for 7 would be written as 7,7,7.
And something like 23,11,14 would indicate the Babylonian symbols for 23, 11
and 14 written in order, with the numerical value (23 × 60 × 60) + (11 × 60) +
14, which comes to 83,474 in our notation.

What numbers did for them

The Babylonian Jupiter Table. The Babylonians used their number system for day-to-day
commerce and accountancy, but they also used it for a more sophisticated purpose:
astronomy. Here the ability of their system to represent fractional numbers with high
accuracy was essential. Several hundred tablets record planetary data. Among them is a
single, rather badly damaged, tablet which details the daily motion of the planet Jupiter
over a period of about 400 days. It was written in Babylon itself, around 163 BC. A typical
entry from the tablet lists the numbers 126 8 16;6,46,58 −0;0,45,18 −0;0,11,42
+0;0,0,10, which correspond to various quantities employed to calculate the planet’s
position in the sky. Note that the numbers are stated to three sexagesimal places –
slightly better than to five decimal places.

Symbols for small numbers

Not only do we use ten symbols to represent arbitrarily large numbers: we also
use the same symbols to represent arbitrarily small ones. To do this, we employ
the decimal point, a small dot. Digits to the left of the dot represent whole
numbers; those to the right of the dot represent fractions. Decimal fractions are
multiples of one tenth, one hundredth, and so on. So 25.47, say, means 2 tens
plus 5 units plus 4 tenths plus 7 hundredths.
The Babylonians knew this trick, and they used it to good effect in their
astronomical observations. Scholars denote the Babylonian equivalent of the
decimal point by a semicolon (;), but this is a sexagesimal point and the numbers
to its right are multiples of 1/60, (1/60 × 1/60) = 1/3600, and so on. As an example, the
list of numbers 12,59;57,17 means

12 × 60 + 59 + 57/60 + 17/3600

which is roughly 779.955.

Nearly 2000 Babylonian tablets with astronomical information are known,
though many of these are fairly routine, consisting of descriptions of ways to
predict eclipses, tables of regular astronomical events and shorter extracts. About
300 tablets are more ambitious and more exciting; they tabulate observations of
the motion of Mercury, Mars, Jupiter and Saturn, for example.
Fascinating as it may be, Babylonian astronomy is somewhat tangential to our
main story, which is Babylonian pure mathematics. But it seems likely that the
application to astronomy was a spur to the pursuit of the more cerebral areas of
that subject. So it is a good idea to recognize just how accurate the Babylonian
astronomers were when it came to observing heavenly events. For example, they
found the orbital period of Mars (strictly, the time between successive
appearances in the same position in the sky) to be 12,59;57,17 days in their
notation – roughly 779.955 days, as noted above. The modern figure is 779.936
days.

The Ancient Egyptians

Perhaps the greatest of the ancient civilizations was that of Egypt, which
flourished on the banks of the Nile and in the Nile Delta between 3150 BC and 31
BC, with an extensive earlier ‘pre-dynastic’ period stretching back to 6000 BC,
and a gradual tailing off under the Romans from 31 BC onwards. The Egyptians
were accomplished builders, with a highly developed system of religious beliefs
and ceremonies, and they were obsessive record-keepers. But their mathematical
achievements were modest compared to the heights scaled by the Babylonians.
The ancient Egyptian system for writing whole numbers is simple and
straightforward. There are symbols for the numbers 1, 10, 100, 1000 and so on.
Repeating these symbols up to nine times, and then combining the results, can
represent any whole number. For example, to write the number 5724, the
Egyptians would group together five of their symbols for 1000, seven symbols
for 100, two symbols for 10 and four symbols for 1.

Egyptian number symbols

Fractions caused the Egyptians severe headaches. At various periods, they

used several different notations for fractions. In the Old Kingdom (2700–2200
BC), a special notation for our fractions, 1/2, 1/4, 1/8, 1/16, 1/32 and 1/64, was obtained
by repeated halving. These symbols used parts of the ‘eye of Horus’ or ‘wadjet-
eye’ hieroglyph.
The best known Egyptian system for fractions was devised during the Middle
Kingdom (200–1700 BC). It starts with a notation for any fraction of the form
1/n, where n is a positive integer. The symbol (the hieroglyph for the letter
R) is written over the top of the standard Egyptian symbols for n. So, for
example, 1/11 is written . Other fractions are then expressed by adding together
several of these ‘unit fractions’. For instance, 5/6 = 1/2 + 1/3.
 The number 5724 in Egyptian hieroglyphs

Special fractions formed from parts of the wadjet eye

Special symbols for special fractions

 Special symbols for special fractions

Interestingly, the Egyptians did not write 2/5 as 1/5 + 1/5. Their rule seems to
have been: use different unit fractions. There were also different notations for
some of the simpler fractions, such as ½, ⅔ and ¾.
The Egyptian notation for fractions was cumbersome and poorly adapted to
calculation. It served them well enough in official records, but it was pretty
much ignored by subsequent cultures.

Numbers and people

Whether you like arithmetic or not, it is hard to deny the profound effects that
numbers have had on the development of human civilization. The evolution of
culture, and that of mathematics, has gone hand in hand for the last four
millennia. It would be difficult to disentangle cause and effect – I would hesitate
to argue that mathematical innovation drives cultural change, or that cultural
needs determine the direction of mathematical progress. But both of those
statements contain a grain of truth, because mathematics and culture co-evolve.
There is a significant difference, though. Many cultural changes are clearly
apparent. New kinds of housing, new forms of transport, even new ways to
organize government bureaucracies, are relatively obvious to every citizen.
Mathematics, however, mostly takes place behind the scenes. When the
Babylonians used their astronomical observations to predict solar eclipses, for
instance, the average citizen was impressed by how accurately the priests
forecast this astonishing event, but even the majority of the priesthood had little
or no idea of the methods employed. They knew how to read tablets listing
eclipse data, but what mattered was how to use them. How they had been
constructed was an arcane art, best left to specialists.

What numbers do for us

Most upmarket modern cars now come equipped with satnav – satellite navigation.
Stand-alone satnav systems can be purchased relatively cheaply. A small device, affixed
to your car, then tells you exactly where you are at any moment and displays a map –
often in fancy colour graphics and perspective – showing the neighbouring roads. A
voice system can even tell you where to go to reach a specified destination. If this
 sounds like something out of science fiction, it is. An essential component, not part of the
small box attached to the car, is the Global Positioning System (GPS), which comprises
24 satellites orbiting the Earth, sometimes more as replacements are launched. These
satellites send out signals, and these signals can be used to deduce the location of the
car to within a few metres.
Mathematics comes into play in many aspects of the GPS network, but here we
mention just one: how the signals are used to work out the location of the car.
Radio signals travel at the speed of light, which is roughly 300,000 kilometres per
second. A computer on board the car – a chip in the box you buy – can work out the
distance from your car to any given satellite if it knows how long the signal has taken to
travel from the satellite to your car. This is typically about one tenth of a second, but
precise time measurement is now easy. The trick is to structure the signal so that it
contains information about timing.
In effect, the satellite and the receiver in the car both play the same tune, and
compare its timing. The ‘notes’ coming from the satellite will lag slightly behind those
produced in the car. In this analogy, the tunes might go like this:

CAR . . . feet, in ancient times, walk upon England’s . . .

SATELLITE . . . And did those feet, in ancient times, walk . . .

Here the satellite’s song is lagging some three words behind the same song in the car.
Both satellite and receiver must generate the same ‘song’, and successive ‘notes’ must
be distinctive, so that the timing difference is easy to observe.
Of course, the satnav system doesn’t actually use a song. The signal is a series of
brief pulses whose duration is determined by a ‘pseudorandom code’. This is a series of
numbers, which looks random but is actually based on some mathematical rule. Both the
satellite and the receiver know the rule, so they can generate the same sequence of
pulses.

Some priests may have had good mathematical educations – all trained scribes
did, and trainee priests took much the same lessons as scribes, in their early
years – but an appreciation of mathematics wasn’t really necessary to enjoy the
benefits that flowed from new discoveries in that subject. So it has ever been,
and no doubt always will be. Mathematicians seldom get credit for changing our
world. How many times do you see all kinds of modern miracles credited to
computers, without the slightest appreciation that computers only work
effectively if they are programmed to use sophisticated algorithms, that is
procedures to solve problems, and that the basis of almost all algorithms is
mathematics?

‘The evolution of culture, and that of mathematics,

has gone hand in hand for the last four millennia.’

The main mathematics that does lie on the surface is arithmetic. And the
invention of pocket calculators, tills that tot up how much you have to pay and
tax accountants who do the sums for you, for a fee, are pushing even arithmetic
further behind the scenes. But at least most of us are aware that the arithmetic is
there. We are wholly dependent on numbers, be it for keeping track of legal
obligations, levying taxes, communicating instantly with the far side of the
planet, exploring the surface of Mars or assessing the latest wonder drug. All of
these things trace back to ancient Babylon, and to the scribes and teachers who
discovered effective ways to record numbers and calculate with them. They used
their arithmetical skills for two main purposes: down-to-earth everyday affairs of
ordinary human beings, such as land-measurement and accountancy, and
highbrow activities like predicting eclipses or recording the movements of
planets across the night-time sky.
We do the same today. We use simple mathematics, little more than
arithmetic, for hundreds of tiny tasks – how much anti-parasite treatment to put
into a garden fishpond, how many rolls of wallpaper to buy to paper the
bedroom, whether it will save money to travel further for cheaper petrol. And
our culture uses sophisticated mathematics for science, technology and
increasingly for commerce too. The inventions of number notation and
arithmetic rank alongside those of language and writing as some of the
innovations that differentiate us from trainable apes.
 2
The Logic of Shape

First steps in geometry

There are two main types of reasoning in mathematics: symbolic and visual.
Symbolic reasoning originated in number notation, and we will shortly see how
it led to the invention of algebra, in which symbols can represent general
numbers (‘the unknown’) rather than specific ones (‘7’). From the Middle Ages
onwards, mathematics came to rely increasingly heavily on the use of symbols,
as a glance at any modern mathematics text will confirm.

The beginnings of geometry

As well as symbols, mathematicians use diagrams, opening up various types of

visual reasoning. Pictures are less formal than symbols, and their use has
sometimes been frowned upon for that reason. There is a widespread feeling that
a picture is somehow less rigorous, logically speaking, than a symbolic
calculation. It is true that pictures leave more room for differences of
interpretation than symbols. Additionally, pictures can contain hidden
assumptions – we cannot draw a ‘general’ triangle; any triangle we draw has a
particular size and shape, which may not be representative of an arbitrary
triangle. Nonetheless, visual intuition is such a powerful feature of the human
brain that pictures play a prominent role in mathematics. In fact, they introduce a
second major concept into the subject, after number. Namely, shape.
Mathematicians’ fascination with shapes goes back a long way. There are
diagrams on Babylonian tablets. For example, the tablet catalogued as YBC
7289 shows a square and two diagonals. The sides of the square are marked with
cuneiform numerals for 30. Above one diagonal is marked 1;24,51,10, and
below it 42;25,35, which is its product by 30 and therefore the length of that
diagonal. So 1;24,51,10 is the length of the diagonal of a smaller square, with
sides 1 unit. Pythagoras’s Theorem tells us that this diagonal is the square root of
two, which we write as . The approximation 1;24,51,10 to is very good,
correct to 6 decimal places.

Tablet YBC 7289 and its cuneiform numerals

 The first systematic use of diagrams, together with a limited use of symbols
and a heavy dose of logic, occurs in the geometric writings of Euclid of
Alexandria. Euclid’s work followed a tradition that went back at least to the
Pythagorean cult, which flourished around 500 BC, but Euclid insisted that any
mathematical statement must be given a logical proof before it could be assumed
to be true. So Euclid’s writings combine two distinct innovations: the use of
pictures and the logical structure of proofs. For centuries, the word ‘geometry’
was closely associated with both.
In this chapter we follow the story of geometry from Pythagoras, through
Euclid and his forerunner Eudoxus, to the late period of classical Greece and
Euclid’s successors Archimedes and Apollonius. These early geometers paved
the way for all later work on visual thinking in mathematics. They also set
standards of logical proof that were not surpassed for millennia.

Pythagoras

Today we almost take it for granted that mathematics provides a key to the
underlying laws of nature. The first recorded systematic thinking along those
lines comes from the Pythagoreans, a rather mystical cult dating from roughly
600 BC to 400 BC. Its founder, Pythagoras, was born on Samos around 569 BC.
When and where he died is a mystery, but in 460 BC the cult that he founded was
attacked and destroyed, its meeting places wrecked and burned. In one, the
house of Milo in Croton, more than 50 Pythagoreans were slaughtered. Many
survivors fled to Thebes in Upper Egypt. Possibly Pythagoras was one of them,
but even this is conjectural, for legends aside, we know virtually nothing about
Pythagoras. His name is well known, mainly because of his celebrated theorem
about right-angled triangles, but we don’t even know whether Pythagoras proved
it.
We know much more about the Pythagoreans’ philosophy and beliefs. They
understood that mathematics is about abstract concepts, not reality. However,
they also believed that these abstractions were somehow embodied in ‘ideal’
concepts, existing in some strange realm of the imagination, so that, for instance,
a circle drawn in sand with a stick is a flawed attempt to be an ideal circle,
perfectly round and infinitely thin.
 The number ten forms a triangle

The most influential aspect of the Pythagorean cult’s philosophy is the belief
that the universe is founded on numbers. They expressed this belief in
mythological symbolism, and supported it with empirical observations. On the
mystic side, they considered the number 1 to be the prime source of everything
in the universe. The numbers 2 and 3 symbolized the female and male principles.
The number 4 symbolized harmony, and also the four elements (earth, air, fire,
water) out of which everything is made. The Pythagoreans believed that the
number 10 had deep mystical significance, because 10 = 1 + 2 + 3 + 4,
combining prime unity, the female principle, the male principle and the four
elements. Moreover, these numbers formed a triangle, and the whole of Greek
geometry hinged upon properties of triangles.

Harmony of the World

The main empirical support for the Pythagorean concept of a numerical universe came from
music, where they had noticed some remarkable connections between harmonious sounds and
simple numerical ratios. Using simple experiments, they discovered that if a plucked string
produces a note with a particular pitch, then a string half as long produces an extremely
harmonious note, now called the octave. A string two-thirds as long produces the next most
harmonious note, and one three-quarters as long also produces a harmonious note.
Today these numerical aspects of music are traced to the physics of vibrating strings, which
move in patterns of waves. The number of waves that can fit into a given length of string is a
 whole number, and these whole numbers determine the simple numerical ratios. If the numbers do
not form a simple ratio then the corresponding notes interfere with each other, forming discordant
‘beats’ which are unpleasant to the ear. The full story is more complex, involving what the brain
becomes accustomed to, but there is a definite physical rationale behind the Pythagorean
discovery.

The Pythagoreans recognized the existence of nine heavenly bodies, Sun,

Moon, Mercury, Venus, Earth, Mars, Jupiter and Saturn, plus the Central Fire,
which differed from the Sun. So important was the number 10 in their view of
cosmology that they believed there was a tenth body, Counter-Earth, perpetually
hidden from us by the Sun.
As we have seen, the whole numbers 1, 2, 3, . . ., naturally lead to a second
type of number, fractions, which mathematicians call rational numbers. A
rational number is a fraction a/b where a, b are whole numbers (and b is non-
zero, otherwise the fraction makes no sense). Fractions subdivide whole numbers
into arbitrarily fine parts, so that in particular the length of a line in a geometric
figure can be approximated as closely as we wish by a rational number. It seems
natural to imagine that enough subdivision would hit the number exactly; if so,
all lengths would be rational.

These two shapes are similar

If this were true, it would make geometry much simpler, because any two
lengths would be whole number multiples of a common (perhaps small) length,
and so could be obtained by fitting lots of copies of this common length
together. This may not sound very important, but it would make the whole
theory of lengths, areas and especially similar figures – figures with the same
shape but different sizes – much simpler. Everything could be proved using
diagrams formed from lots and lots of copies of one basic shape.
Unfortunately, this dream cannot be realized. According to legend, one of the
followers of Pythagoras, Hippasus of Metapontum, discovered that this
statement was false. Specifically, he proved that the diagonal of a unit square (a
square with sides one unit long) is irrational: not an exact fraction. It is said (on
dubious grounds, but it’s a good story) that he made the mistake of announcing
this fact when the Pythagoreans were crossing the Mediterranean by boat, and
his fellow cult-members were so incensed that they threw him overboard and he
drowned. More likely he was just expelled from the cult. Whatever his
punishment, it seems that the Pythagoreans were not pleased by his discovery.
The modern interpretation of Hippasus’s observation is that is irrational. To
the Pythagoreans, this brutal fact was a body-blow to their almost religious
belief that the universe was rooted in numbers – by which they meant whole
numbers. Fractions – ratios of whole numbers – fitted neatly enough into this
world-view, but numbers that were provably not fractions did not. And so,
whether drowned or expelled, poor Hippasus became one of the early victims of
the irrationality, so to speak, of religious belief.

Taming irrationals

Eventually, the Greeks found a way to handle irrationals. It works because any
irrational number can be approximated by a rational number. The better the
approximation, the more complicated that rational becomes, and there is always
some error. But by making the error smaller and smaller, there is a prospect of
approaching the properties of irrationals by exploiting analogous properties of
approximating rational numbers. The problem is to set this idea up in a way that
is compatible with the Greek approach to geometry and proof. This turns out to
be feasible, but complicated.
 Is the ratio a:b the same as ratio c:d?

The Greek theory of irrationals was invented by Eudoxus around 370 BC. His
idea is to represent any magnitude, rational or irrational, as the ratio of two
lengths – that is, in terms of a pair of lengths. Thus two-thirds is represented by
two lines, one of length two and one of length three (a ratio 2:3). Similarly, is
represented by the pair formed by the diagonal of a unit square, and its side (a
ratio :1). Note that both pairs of lines can be constructed geometrically.
The key point is to define when two such ratios are equal. When is a:b = c:d?
Lacking a suitable number system, the Greeks could not do this by dividing one
length by the other and comparing a ÷ b with c ÷ d. Instead, Eudoxus found a
cumbersome but precise method of comparison that could be performed within
the conventions of Greek geometry. The idea is to try to compare a and c by
forming integer multiples ma and nc. This can be done by joining m copies of a
end to end, and similarly n copies of b. Use the same two multiples m and n to
compare mb and nd. If the ratios a:b and c:d are not equal, says Eudoxus, then
we can find m and n to exaggerate the difference, to such an extent that ma > nc
but mb < nd. Indeed, we can define equality of ratios that way.
This definition takes some getting used to. It is tailored very carefully to the
limited operations permitted in Greek geometry. Nonetheless, it works; it let the
Greek geometers take theorems that could easily be proved for rational ratios,
and extend them to irrational ratios.

‘The Greek theory of irrationals was invented by

 Eudoxus around 370 BC.’

Often they used a method called ‘exhaustion’, which let them prove theorems
that we would nowadays prove using the idea of a limit and calculus. In this
manner they proved that the area of a circle is proportional to the square of its
radius. The proof starts from a simpler fact, found in Euclid: the areas of two
similar polygons are in the same proportion as the squares of corresponding
sides. The circle poses new problems because it is not a polygon. The Greeks
therefore considered two sequences of polygons: one inside the circle, the other
outside. Both sequences get closer and closer to the circle, and Eudoxus’s
definition implies that the ratio of the areas of the approximating polygons is the
same as the ratio of the areas of the circles.

Euclid

The best-known Greek geometer, though probably not the most original
mathematician, is Euclid of Alexandria. Euclid was a great synthesizer, and his
geometry text, the Elements, became an all-time bestseller. Euclid wrote at least
ten texts on mathematics, but only five of them survive – all through later copies,
and then only in part. We have no original documents from ancient Greece. The
five Euclidean survivors are the Elements, the Division of Figures, the Data, the
Phaenomena and the Optics.
The Elements is Euclid’s geometrical masterpiece, and it provides a definitive
treatment of the geometry of two dimensions (the plane) and three dimensions
(space). The Division of Figures and the Data contain various supplements and
comments on geometry. The Phaenomena is aimed at astronomers, and deals
with spherical geometry, the geometry of figures drawn on the surface of a
sphere. The Optics is also geometric, and might best be thought of as an early
investigation of the geometry of perspective – how the human eye transforms a
three-dimensional scene into a two-dimensional image.
Perhaps the best way to think of Euclid’s work is as an examination of the
logic of spatial relationships. If a shape has certain properties, these may
logically imply other properties. For example, if a triangle has all three sides
equal – an equilateral triangle – then all three angles must be equal. This type of
statement, listing some assumptions and then stating their logical consequences,
is called a theorem. This particular theorem relates a property of the sides of a
triangle to a property of its angles. A less intuitive and more famous example is
Pythagoras’s Theorem.
The Elements breaks up into 13 separate books, which follow each other in
logical sequence. They discuss the geometry of the plane, and some aspects of
the geometry of space. The climax is the proof that there are precisely five
regular solids: the tetrahedron, cube, octahedron, dodecahedron and icosahedron.
The basic shapes permitted in plane geometry are straight lines and circles, often
in combination – for instance, a triangle is formed from three straight lines. In
spatial geometry we also find planes, cylinders and spheres.
To modern mathematicians, what is most interesting about Euclid’s geometry
is not its content, but its logical structure. Unlike his predecessors, Euclid does
not merely assert that some theorem is true. He provides a proof.

Pythagoras’s Theorem: if the triangle has a right angle, then the largest square, A, has the same area as the
other two, B and C, combined

What is a proof? It is a kind of mathematical story, in which each step is a

logical consequence of some of the previous steps. Every statement that is
asserted has to be justified by referring it back to previous statements and
showing that it is a logical consequence of them. Euclid realized that this process
cannot go back indefinitely: it has to start somewhere, and those initial
statements cannot themselves be proved – or else the process of proof actually
starts somewhere different.
 To start the ball rolling, Euclid started by listing a number of definitions:
clear, precise statements of what certain technical terms, such as line or circle,
mean. A typical definition is ‘an obtuse angle is an angle greater than a right
angle’. The definitions gave him the terminology that he needed to state his
unproved assumptions, which he classified into two types: common notions and
postulates. A typical common notion is ‘things which are equal to the same thing
are equal to one another’. A typical postulate is ‘all right angles are equal to one
another’.
Nowadays we would lump both types together and call them axioms. The
axioms of a mathematical system are the underlying assumptions that we make
about it. We think of the axioms as the rules of the game, and insist that the
game is played according to the rules. We no longer ask whether the rules are
true – we no longer think that only one game can be played. Anyone who wants
to play that particular game must accept the rules; if they don’t, they are free to
play a different game, but it won’t be the one determined by those particular
rules.

Regular Polyhedra
A solid is regular (or Platonic) if it is formed from identical faces, arranged in the same way at
each vertex, with each face a regular polyhedron. The Pythagoreans knew of five such solids:

The five Platonic solids

• The tetrahedron, formed from four equilateral triangles.

• The cube (or hexahedron), formed from six squares.
• The octahedron, formed from eight equilateral triangles.
• The dodecahedron, formed from 12 regular pentagons.
• The icosahedron, formed from 20 equilateral triangles.
 They associated them with the four elements of antiquity – earth, air, fire and water – and with a
fifth element, quintessence, which means fifth element.

In Euclid’s day, and for nearly 2000 years afterwards, mathematicians didn’t
think that way at all. They generally viewed the axioms as self-evident truths, so
obvious that no one could seriously question them. So Euclid did his best to
make all of his axioms obvious – and very nearly succeeded. But one axiom, the
‘parallel axiom’, is unusually complicated and unintuitive, and many people
tried to deduce it from simpler assumptions. Later, we’ll see the remarkable
discoveries that this led to.
From these simple beginnings, the Elements proceeded, step by step, to
provide proofs of increasingly sophisticated geometrical theorems. For example,
Book I Proposition 5 proves that the angles at the base of an isosceles triangle
(one with two equal sides) are equal. This theorem was known to generations of
Victorian schoolboys as the pons asinorum or bridge of asses: the diagram looks
like a bridge, and it was the first serious stumbling block for students who tried
to learn the subject by rote instead of understanding it. Book I Proposition 32
proves that the angles of a triangle add up to 180°. Book I Proposition 47 is
Pythagoras’s Theorem.
Euclid deduced each theorem from previous theorems and various axioms. He
built a logical tower, which climbed higher and higher towards the sky, with the
axioms as its foundations and logical deduction as the mortar that bound the
bricks together.

Euclid of Alexandria
325–265 BC
 Euclid is famous for his geometry book The Elements, which was a prominent – indeed, the
leading – text in mathematical teaching for two millennia.
We know very little about Euclid’s life. He taught at Alexandria. Around 45 BC the Greek
philosopher Proclus wrote:
‘Euclid . . . lived in the time of the first Ptolemy, for Archimedes, who followed closely upon
the first Ptolemy, makes mention of Euclid . . . Ptolemy once asked [Euclid] if there were a
shorter way to study geometry than the Elements, to which he replied that there was no royal
road to geometry. He is therefore younger than Plato’s circle, but older than Eratosthenes and
Archimedes . . . he was a Platonist, being in sympathy with this philosophy, whence he made the
end of the whole Elements the construction of the so-called Platonic figures [regular solids].’

Today we are less satisfied with Euclid’s logic, because it has many gaps.
Euclid takes a lot of things for granted; his list of axioms is nowhere near
complete. For example, it may seem obvious that if a line passes through a point
inside a circle then it must cut the circle somewhere – at least if it is extended far
enough. It certainly looks obvious if you draw a picture, but there are examples
showing that it does not follow from Euclid’s axioms. Euclid did pretty well, but
he assumed that apparently obvious features of diagrams needed neither proof
nor an axiomatic basis.
This omission is more serious than it might seem. There are some famous
examples of fallacious reasoning arising from subtle errors in pictures. One of
them ‘proves’ that every triangle has two equal sides.
The golden mean

Book V of the Elements heads off in a very different, and rather obscure,
direction from Books I–IV. It doesn’t look like conventional geometry. In fact,
at first sight it mostly reads like gobbledegook. What, for instance, are we to
make of Book V Proposition 1? It reads: If certain magnitudes are equimultiples
of other magnitudes, then whatever multiple one of the magnitudes is of one of
the others, that multiple also will be of all.
The language (which I have simplified a little) doesn’t help, but the proof
makes it clear what Euclid intended. The 19th-century English mathematician
Augustus De Morgan explained the idea in simple language in his geometry
textbook: ‘Ten feet ten inches makes ten times as much as one foot one inch.’
What is Euclid up to here? Is it trivialities dressed up as theorems? Mystical
nonsense? Not at all. This material may seem obscure, but it leads up to the most
profound part of the Elements: Eudoxus’s techniques for dealing with irrational
ratios. Nowadays mathematicians prefer to work with numbers, and because
these are more familiar I will often interpret the Greek ideas in that language.
Euclid could not avoid facing up to the difficulties of irrational numbers,
because the climax to the Elements – and, many believe, its main objective – was
the proof that there exist precisely five regular solids: the tetrahedron, cube (or
hexahedron), octahedron, dodecahedron and icosahedron. Euclid proved two
things: there are no other regular solids, and these five actually exist – they can
be constructed geometrically, and their faces fit together perfectly, with no tiny
errors.
Two of the regular solids, the dodecahedron and the icosahedron, involve the
regular pentagon: the dodecahedron has pentagonal faces, and the five faces of
the icosahedron surrounding any vertex determine a pentagon. Regular
pentagons are directly connected with what Euclid called ‘extreme and mean
ratio’. On a line AB, construct a point C so that the ratio AB:AC is equal to
AC:BC. That is, the whole line bears the same proportion to the larger segment
as the larger segment does to the smaller.

‘Is it trivialities dressed up as theorems? Not at all.’

If you draw a pentagon and inscribe a five-pointed star, the edges of the star
are related to the edges of the pentagon by this particular ratio.

Nowadays we call this ratio the golden mean. It is equal to , and this
number is irrational. Its numerical value is roughly 1.618. The Greeks could
prove it was irrational by exploiting the geometry of the pentagon. So Euclid and
his predecessors were aware that, for a proper understanding of the
dodecahedron and icosahedron, they must come to grips with irrationals.
This, at least, is the conventional view of the Elements. David Fowler argues
in his book The Mathematics of Plato’s Academy that there is an alternative view
– essentially, the other way round. Perhaps Euclid’s main objective was the
theory of irrationals, and the regular solids were just a neat application. The
evidence can be interpreted either way, but one feature of the Elements fits this
alternative theory more tidily. Much of the material on number theory is not
needed for the classification of the regular solids – so why did Euclid include
this material? However, the same material is closely related to irrational
numbers, which could explain why it was included.

The ratio of the diagonals to the sides is golden

Extreme and mean ratio (now called the golden mean). The ratio of the top line to the middle one is equal to
Extreme and mean ratio (now called the golden mean). The ratio of the top line to the middle one is equal to
that of the middle one to the bottom one

Archimedes

The greatest of the ancient mathematicians was Archimedes. He made important

contributions to geometry, he was at the forefront of applications of mathematics
to the natural world, and he was an accomplished engineer. But to
mathematicians, Archimedes will always be remembered for his work on circles,
spheres and cylinders, which we now associate with the number π (‘pi’), which
is roughly 3.14159. Of course the Greeks did not work with π directly: they
viewed it geometrically as the ratio of the circumference of a circle to its
diameter.
Earlier cultures had realized that the circumference of a circle is always the
same multiple of its diameter, and knew that this multiple was roughly 3, maybe
a bit bigger. The Babylonians used 3⅛. But Archimedes went much further; his
results were accompanied by rigorous proofs, in the spirit of Eudoxus. As far as
the Greeks knew, the ratio of circumference of a circle to its diameter might be
irrational. We now know that this is indeed the case, but the proof had to wait
until 1770, when one was devised by Johann Heinrich Lambert. (The school
value of 3 1/7 is convenient, but approximate.) Be that as it may, since
Archimedes could not prove π to be rational, he had to assume that it might not
be.
Greek geometry worked best with polygons – shapes formed by straight lines.
But a circle is curved, so Archimedes sneaked up on it by way of approximating
polygons. To estimate π, he compared the circumference of a circle with the
perimeters of two series of polygons: one series situated inside the circle, the
other surrounding it. The perimeters of polygons inside the circle must be shorter
than the circle, whereas those outside the circle must be longer than the circle.
To make the calculations easier, Archimedes constructed his polygons by
repeatedly bisecting the sides of a regular hexagon (six-sided polygon) getting
regular polygons with 12 sides, 24, 48 and so on. He stopped at 96. His
calculations proved that 3 10/71 < π < 3 1/7; that is, π lies somewhere between
3.1408 and 3.1429 in today’s decimal notation.
Archimedes’s work on the sphere is of special interest, because we know not
just his rigorous proof, but how he found it – which was decidedly non-rigorous.
The proof is given in his book On the Sphere and Cylinder. He shows that the
volume of a sphere is two-thirds that of a circumscribed cylinder, and that the
surface areas of those parts of the sphere and the cylinder that lie between any
two parallel planes are equal. In modern language, Archimedes proved that the
volume of a sphere is 4/3 πr3, where r is the radius, and its surface area is 4πr2.
These basic facts are still in use today.
The proof is an accomplished use of exhaustion. This method has an
important limitation: you have to know what the answer is before you have
much chance of proving it. For centuries, scholars had no idea how Archimedes
guessed the answer. But in 1906 the Danish scholar Heiberg was studying a
13th-century parchment, with prayers written on it. He noticed faint lines from
an earlier inscription, which had been erased to make room for the prayers. He
discovered that the original document was a copy of several works by
Archimedes, some of them previously unknown. Such a document is called a
palimpsest – a piece of parchment which has later writing superimposed on
erased earlier writing. (Astonishingly, the same manuscript is now known to
contain pieces of lost works by two other ancient authors.) One work by
Archimedes, the Method of Mechanical Theorems, explains how to guess the
volume of a sphere. The idea is to slice the sphere infinitely thinly, and place the
slices at one end of a balance; at the other end, similar slices of a cylinder and a
cone – whose volumes Archimedes already knew – are hung. The law of the
lever produces the required value for the volume. The parchment sold for two
million dollars in 1998 to a private buyer.

Archimedes of Syracuse
287–212 BC
Archimedes was born in Syracuse, Greece, son of the astronomer Phidias. He visited Egypt,
where supposedly he invented the Archimedean screw, which until very recently was still widely
used to raise Nile water for irrigation. He probably visited Euclid in Alexandria; he definitely
corresponded with Alexandrian mathematicians.

His mathematical skills were unsurpassed and wide-ranging. He turned them to practical use,
and constructed huge war machines based on his ‘law of the lever’, able to hurl huge rocks at the
enemy. His machines were used to good effect in the Roman siege of Alexandria in 212BC. He
even used the geometry of optical reflection to focus the Sun’s rays on to an invading Roman
fleet, burning the ships.
His surviving books (in later copies only) are On Plane Equilibria, Quadrature of the
Parabola, On the Sphere and Cylinder, On Spirals, On Conoids and Spheroids, On Floating
Bodies, Measurement of a Circle and The Sandreckoner, together with The Method, found in
1906 by Johan Heiberg.
 Archimedean screw

Pi to Vast Accuracy
The value of π has now been calculated to several billion digits, using more sophisticated
methods. Such computations are of interest for their methods, to test computer systems, and for
pure curiosity, but the result itself has little significance. Practical applications of π generally
require no more than five or six digits. The current record is 1.24 trillion decimal digits, computed
by Yasumasa Kanada and a team of nine other people in December 2002. The computation took
600 hours on a Hitachi SR8000 supercomputer.

Problems for the Greeks

Greek geometry had limitations, some of which it overcame by introducing new

methods and concepts. Euclid in effect restricted the permitted geometrical
constructions to those that could be performed using an unmarked straight edge
(ruler) and a pair of compasses (henceforth ‘compass’ – the word ‘pair’ is
technically needed, for the same reason that we cut paper with a pair of scissors,
but let’s not be pedantic.) It is sometimes said that he made this a requirement,
but it is implicit in his constructions, not an explicit rule. With extra instruments
– idealized in the same way that the curve drawn by a compass is idealized to a
perfect circle – new constructions are possible.

A sphere and its circumscribed cylinder

For example, Archimedes knew that you can trisect an angle using a straight
edge with two fixed marks on it. The Greeks called such procedures ‘neusis
constructions’. We now know (as the Greeks must have suspected) that an exact
trisection of the angle with ruler and compass is impossible, so Archimedes’s
contribution genuinely extends what is possible. Two other famous problems
from the period are duplicating the cube (constructing a cube whose volume is
twice that of a given cube) and squaring the circle (constructing a square with
the same area as a given circle). These are also known to be impossible using
ruler and compass.
 The Archimedes palimpsest

A far-reaching extension of the allowed operations in geometry, which bore

fruit in Arab work on the cubic equation around 800 and had major applications
to mechanics and astronomy, was the introduction of a new class of curves,
conic sections. These curves, which are extraordinarily important in the history
of mathematics, are obtained by slicing a double-cone with a plane. Today we
shorten the name to conics. They come in three main types:

• The ellipse, a closed oval curve obtained when the plane meets only one half
of the cone. Circles are special ellipses.
• The hyperbola, a curve with two infinite branches, obtained when the plane
 meets both halves of the cone.
• The parabola, a transitional curve lying between ellipses and hyperbolas, in
the sense that it is parallel to some line passing through the vertex of the
cone and lying on the cone. A parabola has only one branch, but extends to
infinity.

Conic sections

Conic sections were studied in detail by Apollonius of Perga, who travelled

from Perga in Asia Minor to Alexandria to study under Euclid. His masterwork,
the Conic Sections of about 230 BC, contains 487 theorems. Euclid and
Archimedes had studied some properties of cones, but it would take an entire
book to summarize Apollonius’s theorems. One important idea deserves mention
here. This is the notion of the foci (plural of focus) of an ellipse (or hyperbola).
The foci are two special points associated with these two types of conic. Among
their many properties, we single out just one: the distance from one focus of an
ellipse, to any point, and back to the other focus is constant (equal to the long
diameter of the ellipse). The foci of a hyperbola have a similar property, but now
we take the difference of the two lengths.

What geometry did for them

Around 250 BC Eratosthenes of Cyrene used geometry to estimate the size of the Earth.
He noticed that at midday on the summer solstice, the Sun was almost exactly overhead
at Syene (present-day Aswan), because it shone straight down a vertical well. On the
same day of the year, the shadow of a tall column indicated that the Sun’s position at
Alexandria was one fiftieth of a full circle (about 7.2°) away from the vertical. The Greeks
knew that the Earth is spherical, and Alexandria was almost due north of Syene, so the
geometry of a circular section of the sphere implied that the distance from Alexandria to
Syene is one fiftieth of the circumference of the Earth.
Eratosthenes knew that camel trains took 50 days to get from Alexandria to Syene,
and they travelled a distance of 100 stadia each day. So the distance from Alexandria to
Syene is 5000 stadia, making the circumference of the Earth 250,000 stadia.
Unfortunately we don’t know for sure how long a stadium was, but one estimate is 157
metres, leading to a circumference of 39,250 km (24,500 miles). The modern figure is
39,840 km (24,900 miles).
 How Eratosthenes measured the size of the Earth

‘The Greeks knew how to trisect angles and

duplicate the cube using conics. . . they could also
square the circle.’

The Greeks knew how to trisect angles and duplicate the cube using conics.
With the aid of other special curves, notably the quadratrix, they could also
square the circle.
Greek mathematics contributed two crucial ideas to human development. The
more obvious was a systematic understanding of geometry. Using geometry as a
tool, the Greeks understood the size and shape of our planet, its relation to the
Sun and Moon, even the complex motions of the remainder of the solar system.
They used geometry to dig long tunnels from both ends, meeting in the middle,
which cut construction time in half. They built gigantic and powerful machines,
based on simple principles like the law of the lever, for purposes both peaceful
and warlike. They exploited geometry in shipbuilding and in architecture, where
buildings like the Parthenon prove to us that mathematics and beauty are not so
far apart. The Parthenon’s visual elegance derives from a host of clever
mathematical tricks, used by the architect to overcome limitations of the human
visual system and irregularities in the very ground on which the building rested.

The new Wembley stadium. Built using basic principles discovered in Ancient Greece and developed over
succeeding centuries by many cultures

The second Greek contribution was the systematic use of logical deduction to
make sure that what was being asserted could also be justified. Logical argument
emerged from their philosophy, but it found its most developed and explicit form
in the geometry of Euclid and his successors. Without solid logical foundations,
later mathematics could not have arisen.
 Hypatia of Alexandria
AD 370–415

Hypatia is the first woman mathematician in the historical record. She was the daughter of Theon
of Alexandria, himself a mathematician, and it is probable that she learned her mathematics from
him. By 400 she had become the head of the Platonist school in Alexandria, lecturing on
philosophy and mathematics. Several historical sources state that she was a brilliant teacher.
We do not know whether Hypatia made any original contributions to mathematics, but she
helped Theon to write a commentary on Ptolemy’s Almagest, and may also have helped him to
prepare a new edition of the Elements, upon which all later editions were based. She wrote
commentaries on the Arithmetica of Diophantus and the Conics of Apollonius.

Among Hypatia’s students were several leading figures in the growing religion of Christianity,
among them Synesius of Cyrene. Some of his letters to her are recorded, and these praise her
abilities. Unfortunately, many early Christians considered Hypatia’s philosophy and science to be
rooted in paganism, leading some to resent her influence. In 412 the new patriarch of Alexandria,
Cyril, engaged in political rivalry with the Roman prefect Orestes. Hypatia was a good friend of
Orestes, and her abilities as a teacher and orator were seen as a threat by the Christians. She
became a focus for political unrest, and was dismembered by a mob. One source blames a
fundamentalist sect, the Nitrian monks, who supported Cyril. Another blames an Alexandrian
mob. A third source claims that she was part of a political rebellion, and her death was
unavoidable.

Her death was brutal, hacked to pieces by a mob wielding sharp tiles (some say oyster-shells).
 Her mangled body was then burned. This punishment may be evidence that Hypatia was
condemned for witchcraft – indeed, the first prominent witch to be killed by the early Christians –
because the penalty for witchcraft prescribed by Constantius II was for their flesh to be ‘torn off
their bones with iron hooks’.

Both influences remain vital today. Modern engineering – computer-based

design and manufacture, for example – rests heavily on the geometric principles
discovered by the Greeks. Every building is designed so that it doesn’t fall down
of its own accord; many are designed to resist earthquakes. Every tower block,
every suspension bridge, every football stadium is a tribute to the geometers of
ancient Greece.
Rational thinking, logical argument, is equally vital. Our world is too
complex, potentially too dangerous, for us to base decisions on what we want to
believe, rather than on what is actually the case. The scientific method is
deliberately constructed to overcome a deep-seated human wish to assume that
what we want to be true – what we claim to ‘know’ – really is true. In science,
emphasis is placed on trying to prove that what you deeply believe to be the case
is wrong. Ideas that survive stringent attempts to disprove them are more likely
to be correct.

‘. . . every football stadium is a tribute to the

geometers of ancient Greece.’

What geometry does for us

Archimedes’s expression for the volume of a sphere is still useful today. One application, which
requires knowing to high accuracy, is the standardized unit of mass for the whole of science. For
many years, for example, a metre was defined to be the length of a particular metal bar when
measured at a particular temperature.
Many basic units of measurement are now defined in terms of such things as how long it takes
an atom of some specific element to vibrate some huge number of times. But some are still based
on physical objects, and mass is a case in point. The standard unit of mass is the kilogram. One
kilogram is currently defined to be the mass of a particular sphere, made of pure silicon and kept
in Paris. The sphere has been machined to extremely high accuracy. The density of silicon has
also been measured very precisely. Archimedes’s formula is needed to compute the volume of the
sphere, which relates density to mass.

Principle of ray-tracing and a sample image

Another modern use of geometry occurs in computer graphics. Movies make extensive use of
computer-generated images (CGI), and it is often necessary to generate images that include
reflections – in a mirror, a wineglass, anything that catches the light. Without such reflections the
image will not appear realistic. An efficient way to do this is by ray-tracing. When we look at a
scene from some particular direction, our eye detects a ray of light that has bounded around the
objects in the scene and happens to enter the eye from that direction. We can follow the path of
this ray by working backwards. At any reflecting surface, the ray bounces so that the original ray
and the reflected ray make equal angles at the surface. Translating this geometric fact into
numerical calculations allows the computer to trace the ray backwards through however many
bounces might be needed before it meets something opaque. (Several bounces may be necessary –
if, for example, the wine glass is sitting in front of a mirror.)
 3
Notations and Numbers

Where our number symbols came from

We are so accustomed to today’s number system, with its use of the ten
decimal digits 0, 1, 2, 3, 4, 5, 6, 7, 8 and 9 (in Western countries), that it can
come as a shock to realize that there are entirely different ways to write
numbers. Even today, many cultures – Arabic, Chinese, Korean – use different
symbols for the ten digits, although they all combine these symbols to form
larger numbers using the same ‘positional’ method (hundreds, tens, units). But
differences in notation can be more radical than that. There is nothing special
about the number 10. It happens to be the number of human fingers and thumbs,
which are ideal for counting, but if we had evolved seven fingers, or twelve,
very similar systems would have worked equally well, perhaps better in some
cases.

Roman numerals

Most Westerners know of at least one alternative system, Roman numerals, in

which – for example – the year 2007 is written MMVII. Most of us are also
aware, at least if reminded, that we employ two distinct methods for writing
numbers that are not whole numbers – fractions like ¾ and decimals such as
0.75. Yet another number notation, found on calculators, is the scientific
notation for very large or very small numbers – such as 5 × 109 for five billion
(often seen as 5E9 on calculator displays) or 5 × 10−6 for five millionths.
These symbolic systems developed over thousands of years, and many
alternatives flourished in various cultures. We have already encountered the
Babylonian sexagesimal system (which would come naturally to any creature
that had 60 fingers), and the simpler and more limited Egyptian number
symbols, with their strange treatment of fractions. Later, base-20 numbers were
used in Central America by the Mayan civilization. Only recently did humanity
settle on the current methods for writing numbers, and their use became
established through a mixture of tradition and convenience. Mathematics is
about concepts, not symbols – but a good choice of symbol can be very helpful.

Greek numerals

We pick up the story of number symbols with the Greeks. Greek geometry was a
big improvement over Babylonian geometry, but Greek arithmetic – as far as we
can tell from surviving sources – was not. The Greeks took a big step
backwards; they did not use positional notation. Instead, they used specific
symbols for multiples of 10 or 100, so that, for instance, the symbol for 50 bore
no particular relationship to that for 5 or 500.
The earliest evidence of Greek numerals is from about 1100 BC. By 600 BC the
symbols had changed, and by 450 BC they had changed again, with the adoption
of the Attic system, which resembles Roman numerals. The Attic system used |,
||, ||| and |||| for the numbers 1, 2, 3 and 4. For 5 the Greek capital letter pi (Π)
was employed, probably because it is the first letter of penta. Similarly, 10 was
written Δ, the first letter of deka; 100 was written H, the first letter of hekaton;
1000 was written , the first letter of chilioi; and 10,000 was written M, the first
letter of myrioi. Later Π was changed to Γ. So the number 2178, for example,
was written as

Although the Pythagoreans made numbers the basis of their philosophy, it is

not known how they wrote them. Their interest in square and triangular numbers
suggests that they may have represented numbers by patterns of dots. By the
classical period, 600–300 BC, the Greek system had changed again, and the 27
different letters of their alphabet were used to denote numbers from 1 to 900,
like this:

These are the lower-case Greek letters, augmented by three extra letters derived
from the Phoenician alphabet: (stigma), (koppa), and (sampi).
Using letters to stand for numbers might have caused ambiguity, so a
horizontal line was placed over the top of the number symbols. For numbers
bigger than 999, the value of a symbol could be multiplied by 1000 by placing a
stroke in front of it.
The various Greek systems were reasonable as a method for recording the
results of calculations, but not for performing the calculations themselves.
(Imagine trying to multiply σ μ γ by ω λ δ, for instance.) The calculations
themselves were probably carried out using an abacus, perhaps represented by
pebbles in the sand, especially early on.
The Greeks wrote fractions in several ways. One was to write the numerator,
followed by a prime (′), and then the denominator, followed by a double prime
(″). Often the denominator was written twice. So 21/47 would be written as

κ α‫ ׳‬μ ζ‫ ״‬μ ζ‫״‬,

where κ α is 21 and μ ζ is 47. They also used Egyptian-style fractions, and there
was a special symbol for ½. Some Greek astronomers, notably Ptolemy,
employed the Babylonian sexagesimal system for precision, but using Greek
symbols for the component digits. It was all very different from what we use
today. In fact, it was a mess.

Indian number symbols

The ten symbols currently used to denote decimal digits are often referred to as
Hindu–Arabic numerals, because they originated in India and were taken up and
developed by the Arabs.
The earliest Indian numerals were more like the Egyptian system. For
example, Khasrosthi numerals, used from 400 BC to AD 100, represented the
numbers from 1 to 8 as

with a special symbol for 10. The first traces of what eventually became the
modern symbolic system appeared around 300 BC in the Brahmi numerals.
Buddhist inscriptions from the time include precursors of the later Hindu
symbols for 1, 4 and 6. However, the Brahmi system used different symbols for
multiples of ten or multiples of 100, so it was similar to the Greek number
symbolism, except that it used special symbols rather than letters of the alphabet.
The Brahmi system was not a positional system. By AD 100 there are records of
the full Brahmi system. Inscriptions in caves and on coins show that it continued
in use until the fourth century.
Between the fourth and sixth centuries, the Gupta Empire gained control of a
large part of India, and the Brahmi numerals developed into Gupta numerals.
From there they developed into Nagari numerals. The idea was the same, but the
symbols differed.
The Indians may have developed positional notation by the first century, but
the earliest datable documentary evidence for positional notation places it in 594.
The evidence is a legal document which bears the date 346 in the Chedii
calendar, but some scholars believe this date may be a forgery. Nevertheless, it is
generally agreed that positional notation was in use in India from about 400
onwards.
There is a problem with the use of only the symbols 1–9: the notation is
ambiguous. What does 25 mean, for instance? It might (in our notation) mean
25, or 205, or 2005 or 250, etc. In positional notation, where the meaning of a
symbol depends on its location, it is important to specify that location without
ambiguity. Today we do that by using a tenth symbol, zero (0). But it took early
civilizations a long time to recognize the problem and solve it in that manner.
One reason was philosophical: how can 0 be a number when a number is a
quantity of things? Is nothing a quantity? Another was practical: usually it was
clear from the context whether 25 meant 25 or 250 or whatever.

Brahmi numerals 1–9

Some time before 400 BC – the exact date is unknown – the Babylonians
introduced a special symbol to show a missing position in their number notation.
This saved the scribes the effort of leaving a carefully judged space, and made it
possible to work out what a number meant even if it was written sloppily. This
invention was forgotten, or not transmitted to other cultures, and eventually
rediscovered by the Hindus. The Bhakshali manuscript, the date of which is
disputed but lies somewhere between AD 200 and 1100, uses a heavy dot •. The
Jain text Lokavibhaaga of AD 458 uses the concept of zero, but not a symbol. A
positional system that lacked the numeral zero was introduced by Aryabhata
around AD 500. Later Indian mathematicians had names for zero, but did not use
a symbol. The first undisputed use of zero in positional notation occurs on a
stone tablet in Gwalior dated to AD 876.

Brahmagupta, Mahavira and Bhaskara

The key Indian mathematicians were Aryabhata (born AD 476), Brahmagupta

(born AD 598), Mahavira (9th century) and Bhaskara (born 1114). Actually they
should be described as astronomers, because mathematics was then considered
to be an astronomical technique. What mathematics existed was written down as
chapters in astronomy texts; it was not viewed as a subject in its own right.
Aryabhata tells us that his book Aryabhatiya was written when he was 23
years old. Brief though the mathematical section of his book is, it contains a
wealth of material: an alphabetic system of numerals, arithmetical rules, solution
methods for linear and quadratic equations, trigonometry (including the sine
function and the ‘versed sine’ 1 – cos θ). There is also an excellent
approximation, 3.1416, to π.

‘. . . Lilavati could never get married. To cheer her

up, Bhaskara wrote a mathematics textbook for her.’

Brahmagupta was the author of two books: Brahma Sputa Siddhanta and
Khanda Khadyaka. The first is the most important; it is an astronomy text with
several sections on mathematics, with arithmetic and the verbal equivalent of
simple algebra. The second book includes a remarkable method for interpolating
sine tables – that is, finding the sine of an angle from the sines of a larger angle
and a smaller one.
Mahavira was a Jain, and he included a lot of Jain mathematics in his Ganita
Sara Samgraha. This book included most of the contents of those of Aryabhata
and Brahmagupta, but went a great deal further and was generally more
sophisticated. It included fractions, permutations and combinations, the solution
of quadratic equations, Pythagorean triangles and an attempt to find the area and
perimeter of an ellipse.
 The ancient Jantar Mantar observatory near Jaipur. It is apparent today that the designer was an
accomplished mathematician.

Bhaskara (known as ‘the teacher’) wrote three important works: Lilavati,

Bijaganita and Siddhanta Siromani. According to Fyzi, court poet of the Mogul
emperor Akbar, Lilavati was the name of Bhaskara’s daughter. Her father cast
his daughter’s horoscope, and determined the most auspicious time for her
wedding. To dramatize his forecast, he put a cup with a hole in it inside a bowl
of water, constructed so that it would sink when the propitious moment arrived.
But Lilavati leaned over the bowl and a pearl from her clothing fell into the cup
and blocked the hole. The cup did not sink, which meant that Lilavati could
never get married. To cheer her up, Bhaskara wrote a mathematics textbook for
her. The legend does not record what she thought of this.
Lilavati contains sophisticated ideas in arithmetic, including the method of
casting out the nines, in which numbers are replaced by the sum of their digits to
check calculations. It contains similar rules for divisibility by 3, 5, 7 and 11. The
role of zero as a number in its own right is made clear. Bijaganita is about the
solution of equations. Siddhanta Siromani deals with trigonometry: sine tables
and various trigonometric relations. So great was Bhaskara’s reputation that his
works were still being copied around 1800.

What arithmetic did for them

The oldest surviving Chinese mathematics text is the Chiu Chang, which dates from
about AD 100. A typical problem is: Two and a half piculs of rice are bought for 3/7 of a
tael of silver. How many piculs can be bought for 9 taels? The proposed solution uses
what medieval mathematicians called the ‘rule of three’. In modern notation, let x be the
required quantity.
Then

so x = 52½ piculs. A picul is about 65 kilograms.

The Hindu System

The Hindu system started to spread into the Arabic world, before it was fully
developed in its country of origin. The scholar Severus Sebokht writes of its use
in Syria in 662: ‘I will omit all discussion of the science of the Indians . . . of
their subtle discoveries in astronomy . . . and of their valuable methods of
calculation . . . I wish only to say that this computation is done by means of nine
signs.’
In 776 a traveller from India appeared at the court of the Caliph and
demonstrated his prowess in the ‘siddhanta’ method of calculation, along with
trigonometry and astronomy. The basis for the computational methods seems to
have been the Brahmasphutasiddhanta of Brahmagupta, written in 628, but
whichever book it was, it was promptly translated into Arabic.
 Initially the Hindu numerals were mainly used by scholars; older methods
remained in widespread use among the Arabic business community and in daily
life, until about 1000. But Al-Khwarizmi’s On Calculation with Hindu Numerals
of 825 made the Hindu system widely known in the Arab world. The
mathematician Al-Kindi’s four-volume treatise On the Use of the Indian
Numerals (Ketab fi Isti’mal al-’Adad al-Hindi) of 830 increased awareness of
the possibility of performing all numerical calculations using only the ten digits.

The Dark Ages?

While Arabia and India were making significant advances in mathematics and
science, Europe was comparatively stagnant, although the medieval period was
not quite the ‘Dark Ages’ of popular conception. Some advances were made, but
these were slow and not particularly radical. The pace of change began to
accelerate when word of the Eastern discoveries came to Europe. Italy lies closer
to the Arabian world than most parts of Europe, so it was probably inevitable
that Arab advances in mathematics made their way to Europe through Italy.
Venice, Genoa and Pisa were significant trading centres, and merchants sailed
from these ports to North Africa and the eastern end of the Mediterranean. They
exchanged wool and European wood for silks and spices.
There was metaphorical trade in ideas as well as literal trade in goods.
Arabian discoveries in science and mathematics made their way along the trade
routes, often by word of mouth. As trade made Europe more prosperous, barter
gave way to money, and keeping accounts and paying taxes became more
complex. The period’s equivalent of a pocket calculator was the abacus, a device
in which beads moving on wires represented numbers. However, those numbers
also had to be written down on paper, for legal purposes and for general record-
keeping. So the merchants needed a good number notation as well as methods
for doing calculations quickly and accurately.
An influential figure was Leonardo of Pisa, also known as Fibonacci, whose
book Liber Abbaci was published in 1202. (The Italian word ‘abbaco’ usually
means ‘calculation’, and need not imply the use of the abacus, a Latin term.) In
this book, Leonardo introduced Hindu–Arabic number symbols to Europe.
The Liber Abbaci includes, and promoted, one further notational device that
remains in use today: the horizontal bar in a fraction, such as for ‘three-
quarters’. The Hindus employed a similar notation, but without the bar; the bar
seems to have been introduced by the Arabs. Fibonacci employed it widely, but
his usage differed from what we do today in some respects. For instance, he
would use the same bar as part of several different fractions.
Because fractions are very important in our story, it may be worth adding a
few comments on the notation. In a fraction like , the 4 on the bottom tells us to
divide the unit into four equal parts, and the 3 on top then tells us to select three
of those pieces. More formally, 4 is the denominator and 3 is the numerator. For
typographical convenience, fractions are often written on a single line in the
form 3/4, or sometimes in the compromise form ¾. The horizontal bar then
mutates into a diagonal slash.

Evolution of western number symbols

Leonardo of Pisa (Fibonacci)

1170–1250
 Leonardo, Italian born, grew up in North Africa, where his father Guilielmo was working as a
diplomat on behalf of merchants trading at Bugia (in modern Algeria). He accompanied his father
on his numerous travels, encountered the Arabic system for writing numbers and understood its
importance. In his Liber Abbaci of 1202 he writes: ‘When my father, who had been appointed by
his country as public notary in the customs at Bugia acting for the Pisan merchants going there,
was in charge, he summoned me to him while I was still a child, and having an eye to usefulness
and future convenience, desired me to stay there and receive instruction in the school of
accounting. There, when I had been introduced to the art of the Indians’ nine symbols through
remarkable teaching, knowledge of the art very soon pleased me above all else.’

The book introduced the Hindu–Arabic notation to Europe, and formed a comprehensive
arithmetic text, containing a wealth of material related to trade and currency conversion.
Although it took several centuries for Hindu–Arabic notation to displace the traditional abacus,
the advantages of a purely written system of calculation soon became apparent.

Leonardo is often known by his nickname ‘Fibonacci’, which means ‘son of Bonaccio’, but this
name is not recorded before the 18th century and was probably invented then by Guillaume Libri.

On the whole, however, we seldom use fractional notation in practical work.

Mostly we use decimals – writing π as 3.14159, say, which is not exact, but
close enough for most calculations. Historically, we have to make a bit of a leap
to get to decimals, but we are following chains of ideas, not chronology, and it
will be much simpler to make the leap anyway. We therefore jump forward to
1585, when William the Silent chose the Dutchman Simon Stevin as private
tutor to his son Maurice of Nassau.
Building on this recognition, Stevin made quite a career for himself, becoming
Inspector of Dykes, Quartermaster-General of the Army and eventually the
Minister of Finance. He quickly realized the need for accurate accounting
procedures, and he looked to the Italian arithmeticians of the Renaissance
period, and the Hindu–Arabic notation transmitted to Europe by Leonardo of
Pisa. He found fractional calculations cumbersome, and would have preferred
the precision and tidiness of Babylonian sexagesimals, were it not for the use of
base-60. He tried to find a system that combined the best of both, and invented a
base-10 analogue of the Babylonian system: decimals.
He published his new notational system, making it clear that it had been tried,
tested and found to be entirely practical by entirely practical men. In addition, he
pointed out its efficacy as a business tool: ‘all computations that are met in
business may be performed by integers alone without the aid of fractions’.
His notation did not include the familiar decimal point, but it led very quickly
to today’s decimal notation. Where we would write 5.7731, say, Stevin wrote
. The symbol indicated a whole number, indicated one
tenth, one hundredth and so on. As people got used to the system they
dispensed with , and so on, and retained only – which shrunk, and
simplified, until it became the usual decimal point.

Negative numbers

Mathematicians call the system of whole numbers the natural numbers.

Including negative numbers as well, we obtain the integers. The rational
numbers (or merely ‘rationals’) are the positive and negative fractions, the real
numbers (or merely ‘reals’) are the positive and negative decimals, going on
forever if necessary.
How did negative numbers come into the story?
Early in the first millennium, the Chinese employed a system of ‘counting
rods’ instead of an abacus. They laid the rods out in patterns to represent
numbers.
The top row of the picture shows heng rods, which represented units,
hundreds, tens of thousands and so on, according to their position in a row of
such symbols. The bottom row shows tsung rods, which represented tens,
thousands and so on. So the two types alternated. Calculations were performed
by manipulating the rods in systematic ways.

‘Early in the first millennium, the Chinese employed

a system of ‘counting rods’ instead of an abacus.’

When solving a system of linear equations, the Chinese calculators would

arrange the rods in a table. They used red rods for terms that were supposed to
be added and black rods for terms that were supposed to be subtracted. So to
solve equations that we would write as they would set out the two equations as
two columns of a table: one with the numbers 3 (red), 2 (black), 4 (red), and the
other 1 (red), 5 (red), 7 (red).

3x – 2y = 4
x + 5y = 7

Ancient Chinese counting rods

 Laying out equations, Chinese style. Shaded rods are red

The red/black notation was not really about negative numbers, but the
operation of subtraction. However, it set the stage for a concept of negative
numbers, cheng fu shu. Now a negative number was represented by using the
same arrangement of rods as for the corresponding positive number, by placing
another rod diagonally over the top.
To Diophantus, all numbers had to be positive, and he rejected negative
solutions to equations. Hindu mathematicians found negative numbers useful to
represent debts in financial calculations – owing someone a sum of money was
worse, financially, than having no money, so a debt clearly should be less than
zero. If you have 3 pounds and pay out 2, then you are left with 3 – 2 = 1. By the
same token, if you owe a debt of 2 pounds and acquire 3, your net worth is −2 +
3 = 1. Bhaskara remarks that a particular problem had two solutions, 50 and –5,
but he remained nervous about the second solution, saying that it was ‘not to be
taken; people do not approve of negative solutions’.
Despite these misgivings, negative numbers gradually became accepted. Their
interpretation, in a real calculation, needed care. Sometimes they made no sense,
sometimes they might be debts, sometimes they might mean a downwards
motion instead of an upwards one. But interpretation aside, their arithmetic
worked perfectly well, and they were so useful as a computational aid that it
would have been silly not to use them.

Arithmetic lives on

Our number system is so familiar that we tend to assume that it is the only
possible one, or at least the only sensible one. In fact, it evolved, laboriously and
with lots of dead ends, over thousands of years. There are many alternatives;
some were used by earlier cultures, like the Mayans. Different notations for the
numerals 0 − 9 are in use today in some countries. And our computers represent
numbers internally in binary, not decimal: their programmers ensure that the
numbers are turned back into decimal before they appear on the screen or in a
print-out.
Since computers are now ubiquitous, is there any point in teaching arithmetic
any more? Yes, for several reasons. Someone has to be able to design and build
calculators and computers, and make them do the right job; this requires
understanding arithmetic – how and why it works, not just how to do it. And if
your only arithmetical ability is reading what’s on a calculator, you probably
won’t notice if the supermarket gets your bill wrong. Without internalizing the
basic operations of arithmetic, the whole of mathematics will be inaccessible to
you. You might not worry about that, but modern civilization would quickly
break down if we stopped teaching arithmetic, because you can’t spot the future
engineers and scientists at the age of five. Or even the future bank managers and
accountants.

Mayan Numerals
A remarkable number system, which used base-20 notation instead of base-10, was developed by
the Mayans, who lived in South America around 1000. In the base-20 system, the symbols
equivalent to our 347 would mean

3 × 400 + 4 × 20 + 7 × 1

(Since 20 × 20 = 400) which is 1287 in our notation. The actual symbols are shown here.
Early civilizations that use base-10 probably did so because humans have ten fingers (including
thumbs). It has been suggested that the Mayans counted on their toes as well, which is why they
used base-20.

What arithmetic does for us

We use arithmetic throughout our daily lives, in commerce, and in science. Until the
development of electronic calculators and computers, we either did the calculations by
 hand, with pen and paper, or we used aids such as the abacus or a ready reckoner (a
printed book of tables of multiples of amounts of money). Today most arithmetic goes on
electronically behind the scenes – supermarket checkout tills now tell the operator how
much change to give back, for instance, and banks total up what is in your account
automatically, rather than getting their accountant to do it. The quantity of arithmetic
‘consumed’ by a typical person during the course of a single day is substantial.
Computer arithmetic is not actually carried out in decimal format. Computers use
base-2, or binary, rather than base-10. In place of units, tens, hundreds, thousands and
so on, computers use 1, 2, 4, 8, 16, 32, 64, 128, 256, and so on – the powers of two,
each twice its predecessor. (This is why the memory card for your digital camera comes
in funny sizes like 256 megabytes.) In a computer, the number 100 would be broken
down as 64 + 32 +4 and stored in the form 1100100.

Of course, once you have a basic grasp of arithmetic by hand, using a

calculator is a good way to save time and effort. But, just as you won’t learn to
walk by always using a crutch, you won’t learn to think sensibly about numbers
by relying solely on a calculator.

‘. . . modern civilization would quickly break down

if we stopped teaching arithmetic. . .’
 4
Lure of the Unknown

X marks the spot

The use of symbols in mathematics goes well beyond their appearance in

notation for numbers, as a casual glance at any mathematics text will make clear.
The first important step towards symbolic reasoning – as opposed to mere
symbolic representation – occurred in the context of problem-solving. Numerous
ancient texts, right back to the Old Babylonian period, present their readers with
information about some unknown quantity, and then ask for its value. A standard
formula (in the literary sense) in Babylonian tablets goes ‘I found a stone but did
not weigh it’. After some additional information – ‘when I had added a second
stone of half the weight, the total weight was 15 gin’ – the student is required to
calculate the weight of the original stone.

Algebra

Problems of this kind eventually gave rise to what we now call algebra, in which
numbers are represented by letters. The unknown quantity is traditionally
denoted by the letter x, the conditions that apply to x are stated as various
mathematical formulas, and the student is taught standard methods for extracting
the value of x from those formulas. For instance, the Babylonian problem above
would be written as x + ½x = 15, and we would learn how to deduce that x = 10.
At school level, algebra is a branch of mathematics in which unknown
numbers are represented by letters, the operations of arithmetic are represented
by symbols and the main task is to deduce the values of unknown quantities
from equations. A typical problem in school algebra is to find an unknown
number x, given the equation x2 + 2x = 120. This quadratic equation has one
positive solution, x = 10. Here x2 + 2x = 102 + 2 × 10 = 100 + 20 = 120. It also
has one negative solution, x = −12. Now x2 + 2x = (−12)2 + 2×(−12) = 144 − 24
= 120. The ancients would have accepted the positive solution, but not the
negative one. Today we admit both, because in many problems negative
numbers have a sensible meaning and correspond to physically feasible answers,
and because the mathematics actually becomes simpler if negative numbers are
permitted.

‘How did algebra arise?’

In advanced mathematics, the use of symbols to represent numbers is only one

tiny aspect of the subject, the context in which it got started. Algebra is about the
properties of symbolic expressions in their own right; it is about structure and
form, not just number. This more general view of algebra developed when
mathematicians started asking general questions about school-level algebra.
Instead of trying to solve specific equations, they looked at the deeper structure
of the solution process itself.
How did algebra arise? What came first were the problems and methods. Only
later was the symbolic notation – which we now consider to be the essence of
the topic – invented. There were many notational systems, but eventually one
eliminated all of its competitors. The name ‘algebra’ appeared in the middle of
this process, and it is of Arabic origin. (The initial ‘al’, Arabic for ‘the’, indicates
its origin.)
 An Old Babylonion cuneiform tablet showing an algebraic geometric problem

Equations

What we now call the solution of equations, in which an unknown quantity must
be found from suitable information, is almost as old as arithmetic. There is
indirect evidence that the Babylonians were solving quite complicated equations
as early as 2000 BC, and direct evidence for solutions of simpler problems, in the
form of cuneiform tablets, dating from around 1700 BC.
The surviving portion of Tablet YBC 4652, from the Old Babylonian period
(1800–1600 BC), contains eleven problems for solution; the text on the tablet
indicates that originally there were 22 of them. A typical question is:
‘I found a stone, but did not weigh it. After I weighed out six times its weight,
added 2 gin and added one third of one seventh multiplied by 24, I weighed it.
The result was 1 ma-na. What was the original weight of the stone?’
A weight of 1 ma-na is 60 gin.
In modern notation, we would let x be the required weight in gin. Then the
question tells us that

and standard algebraic methods lead to the answer x = 4⅓ gin. The tablet states
this answer but gives no clear indication of how it is obtained. We can be
confident that it would not have been found using symbolic methods like the
ones we now use, because later tablets prescribe solution methods in terms of
typical examples – ‘halve this number, add the product of these two, take the
square root . . .’ and so on.
This problem, along with the others on YBC 4652, is what we now call a
linear equation, which indicates that the unknown x enters only to the first
power. All such equations can be rewritten in the form

ax + b = 0

with solution x = −b/a. But in ancient times, with no concept of negative

numbers and no symbolic manipulation, finding a solution was not so
straightforward. Even today, many students would struggle with the problem
from YBC 4652.
More interesting are quadratic equations, in which the unknown can also
appear raised to the second power – squared. The modern formulation takes the
form

ax2 + bx + c = 0

and there is a standard formula to find x. The Babylonian approach is

exemplified by a problem on Tablet BM 13901:
‘I have added up seven times the side of my square and eleven times the area,
[getting] 6;15.’
(Here 6;15 is the simplified form of Babylonian sexagesimal notation, and
means 6 plus 15/60, or 6¼ in modern notation.) The stated solution runs:
 ‘. . . the tablet tells its reader what to do, but not
why.’

‘You write down 7 and 11. You multiply 6;15 by 11, [getting] 1,8;45. You
break off half of 7, [getting] 3;30 and 3;30. You multiply, [getting] 12;15. You
add [this] to 1,8;45 [getting] result 1,21. This is the square of 9. You subtract
3;30, which you multiplied, from 9. Result 5;30. The reciprocal of 11 cannot be
found. By what must I multiply 11 to obtain 5;30? [The answer is] 0;30, the side
of the square is 0;30.’
Notice that the tablet tells its reader what to do, but not why. It is a recipe.
Someone must have understood why it worked, in order to write it down in the
first place, but once discovered it could then be used by anyone who was
appropriately trained. We don’t know whether Babylonian schools merely taught
the recipe, or explained why it worked.
The recipe as stated looks very obscure, but it is easier to interpret the recipe
than we might expect. The complicated numbers actually help: they make it
clearer which rules are being used. To find them, we just have to be systematic.
In modern notation, write

a = 11, b = 7, c = 6; 15 = 6¼

Then the equation takes the form

ax2 + bx = c

with those particular values for a, b, c. We have to deduce x. The Babylonian

solution tells us to:

(1) Multiply c by a, which gives ac.

(2) Divide b by 2, which is b/2.
(3) Square b/2 to get b2/4.
(4) Add this to ac, which is ac + b2/4.
(5) Take its square root .
(6) Subtract b/2, which makes .
(7) Divide this by a, and the answer is .
This is equivalent to the formula

that is taught today because we put the term c on the left hand side, where it
becomes −c.
It is quite clear that the Babylonians knew that their procedure was a general
one. The quoted example is too complex for the solution to be a special one,
designed to suit that problem alone.
How did the Babylonians think of their method, and how did they think about
it? There had to be some relatively simple idea lying behind such a complicated
process. It seems plausible, though there is no direct evidence, that they had a
geometric idea, completing the square. An algebraic version of this is taught
today, too. We can represent the question, which for clarity we choose to write
in the form x2 + ax = b, as a picture:

Here the square and the first rectangle have height x; their widths are,
respectively, x and a. The smaller rectangle has area b. The Babylonian recipe
effectively splits the first rectangle into two pieces,
 We can then rearrange the two new pieces and stick them on the edge of the
square:

The left-hand diagram now cries out to be completed to a larger square, by

adding the shaded square:

To keep the equation valid, the same extra shaded square is added to the other
diagram too. But now we recognize the left-hand diagram as the square of side
(x + a/2), and the geometric picture is equivalent to the algebraic statement
Since the left-hand side is a square, we can rewrite this as

and then it is natural to take a square root

and finally rearrange to deduce that

which is exactly how the Babylonian recipe proceeds.

There is no evidence on any tablet to support the view that this geometric
picture led the Babylonians to their recipe. However, this suggestion is plausible,
and is supported indirectly by various diagrams that do appear on clay tablets.

Al-jabr

The word algebra comes from the Arabic al-jabr, a term employed by
Muhammad ibn Musa al-Khwarizmi, who flourished around 820. His work The
Compendious Book on Calculation by al-jabr w’al-muqabala explained general
methods for solving equations by manipulating unknown quantities.
Al-Khwarizmi used words, not symbols, but his methods are recognizably
similar to those taught today. Al-jabr means ‘adding equal amounts to both sides
of an equation’, which is what we do when we start from

x − 3 = 5

and deduce that

x = 8

In effect, we make this deduction by adding 3 to both sides. Al-muqabala has

two meanings. There is a special meaning: ‘subtracting equal amounts from both
sides of an equation’, which we do to pass from

x + 3 = 5
to the answer

x = 2

But it also has a general meaning: ‘comparison’.

Al-Khwarizmi gives general rules for solving six kinds of equation, which
between them can be used to solve all linear and quadratic equations. In his
work, then, we find the ideas of elementary algebra, but not the use of symbols.

‘The word ‘algebra’ comes from the Arabic al-

jabr. . .’

Cubic equations

The Babylonians could solve quadratic equations, and their method was
essentially the same one that is taught today. Algebraically, it involves nothing
more complicated than a square root, beyond the standard operations of
arithmetic (add, subtract, multiply, divide). The obvious next step is cubic
equations, involving the cube of the unknown. We write such equations as

ax3 + bx2 + cx + d = 0

where x is the unknown and the coefficients a, b, c, d are known numbers. But
until the development of negative numbers, mathematicians classified cubic
equations into many distinct types – so that, for example, x3 + 3x = 7 and x3 – 3x
= 7 were considered to be completely different, and required different methods
for their solution.

Fibonacci Sequence
The third section of the Liber Abbaci contains a problem that seems to have originated with
Leonardo: ‘A certain man put a pair of rabbits in a place surrounded on all sides by a wall. How
many pairs of rabbits can be produced from that pair in a year if in every month, each pair begets
a new pair, which from the second month onwards becomes productive?’
This rather quirky problem leads to a curious, and famous, sequence of numbers:

1 2 3 5 8 13 21 34 55
 1 2 3 5 8 13 21 34 55

and so on. Each number is the sum of the two preceding numbers. This is known as the Fibonacci
Sequence, and it turns up repeatedly in mathematics and in the natural world. In particular, many
flowers have a Fibonacci number of petals. This is not coincidence, but a consequence of the
growth pattern of the plant and the geometry of the ‘primordia’ – tiny clumps of cells at the tip of
the growing shoot that give rise to important structures, including petals.
Although Fibonacci’s growth rule for rabbit populations is unrealistic, more general rules of a
similar kind (called Leslie models) are used today for certain problems in population dynamics,
the study of how animal populations change in size as the animals breed and die.

What algebra did for them

Several chapters of the Liber Abbaci contain algebraic problems relevant to the needs of
merchants. One, not terribly practical, goes like this: ‘A man buys 30 birds – partridges,
doves and sparrows. A partridge costs 3 silver coins, a dove 2, and a sparrow ½. He
pays 30 silver coins. How many birds of each type does he buy?’ In modern notation, if
we let x be the number of partridges, y the number of doves, and z the number of
sparrows, then we must solve two equations

x + y + z = 30
3x + 2y + ½z = 30

In real or rational numbers, these equations would have infinitely many solutions, but
there is an extra condition implied by the question: the numbers x, y, z are integers. It
turns out that only one solution exists: 3 partridges, 5 doves and 22 sparrows.
Leonardo also mentions a series of problems about buying a horse. One man says to
another, ‘If you give me one-third of your money, I can buy the horse’. The other says, ‘if
you give me one-quarter of your money, I can buy the horse’. What is the price of the
horse? This time there are many solutions; the smallest one in whole numbers sets the
price of the horse at 11 silver coins.

The Greeks discovered how to use conic sections to solve some cubic
equations. Modern algebra shows that if a conic intersects another conic, the
points of intersection are determined by an equation of third or fourth degree
(depending on the conics). The Greeks did not know this as a general fact, but
they exploited its consequences in specific instances, using the conics as a new
kind of geometrical instrument.
This line of attack was completed and codified by the Persian Omar
Khayyam, best known for his poem the Rubaiyat. Around 1075 he classified
cubic equations into 14 kinds, and showed how to solve each kind using conics,
in his work On the Proofs of the Problems of Algebra and Muqabala. The
treatise was a geometric tour de force, and it polished off the geometric problem
almost completely. A modern mathematician would raise a few quibbles – some
of Omar’s cases are not completely solved because he assumes that certain
geometrically constructed points exist when sometimes they do not. That is, he
assumes his conics meet when they may fail to do so. But these are minor
blemishes.
Geometric solutions of cubic equations were all very well, but could there
exist algebraic solutions, involving such things as cube roots, but nothing more
complicated? The mathematicians of Renaissance Italy made one of the biggest
breakthroughs in algebra when they discovered that the answer is ‘yes’.
In those days, mathematicians made their reputation by taking part in public
contests. Each contestant would set his opponent problems, and whoever solved
the most was adjudged the winner. Members of the audience could place bets on
who would win. The contestants often wagered large sums of money – in one
recorded instance, the loser had to buy the winner (and his friends) thirty
banquets. Additionally, the winner’s ability to attract paying students, mostly
from the nobility, was likely to be enhanced. So, public mathematical combat
was serious stuff.

‘So, public mathematical combat was serious stuff.’

In 1535 there was just such a contest, between Antonio Fior and Niccolo
Fontana, nicknamed Tartaglia, ‘the stammerer’. Tartaglia wiped the floor with
Fior, and word of his success spread, coming to the ears of Girolamo Cardano.
And Cardano’s ears pricked up. He was in the middle of writing a
comprehensive algebra text, and the questions that Fior and Tartaglia had posed
each other were – cubic equations. At that time, cubic equations were classified
into three distinct types, again because negative numbers were not recognized.
Fior knew how to solve just one type. Initially, Tartaglia knew how to solve just
one different type. In modern symbols, his solution of a cubic equation of the
type x3 + ax = b is
In a burst of inspired desperation, a week or so before the contest, Tartaglia
figured out how to solve the other types too. He then set Fior only the types that
he knew Fior could not solve.

Omar Khayyam was better known for his poetry but he was also a considerable mathematician

Cardano, hearing of the contest, realized that the two combatants had devised
methods for solving cubic equations. Wanting to add them to his book, he
buttonholed Tartaglia and asked him to reveal his methods. Tartaglia was
naturally reluctant, because his livelihood depended on them, but eventually he
was persuaded to divulge the secret. According to Tartaglia, Cardano promised
never to make the method public. So Tartaglia was understandably peeved when
his method appeared in Cardano’s Ars Magna – the Great Art of Algebra. He
complained bitterly and accused Cardano of plagiarism.
Now, Cardano was far from lilywhite. He was an inveterate gambler, who had
made and lost considerable sums of money at cards, dice and even chess. He lost
the entire family fortune in this manner, and was reduced to penury. He was also
a genius, a competent doctor, a brilliant mathematician and an accomplished
self-publicist – though his positive attributes were mitigated by frankness that
often became offensively blunt and insulting. So Tartaglia can be forgiven for
assuming that Cardano had lied to him and stolen his discovery. That Cardano
had given full credit to Tartaglia in his book only made things worse; Tartaglia
knew that it was the book’s author who would be remembered, not some obscure
figure given a sentence or so of mention.
However, Cardano had an excuse, quite a good one. And he also had a strong
reason to bend his promise to Tartaglia. The reason was that Cardano’s student
Lodovico Ferrari had found a method for solving quartic equations, those
involving the fourth power of the unknown. This was completely new, and of
huge importance. So of course Cardano wanted quartic equations in his book,
too. Since it was his student who had made the discovery, this would have been
legitimate. But Ferrari’s method reduced the solution of any quartic to that of an
associated cubic, so it relied on Tartaglia’s solution of cubic equations. Cardano
could not publish Ferrari’s work without also publishing Tartaglia’s.
Then news reached him that offered a way out. Fior, who had lost to Tartaglia
in public combat, was a student of Scipio del Ferro. Cardano heard that del Ferro
had solved all three types of cubic, not just the one that he had passed on to Fior.
And a certain Annibale del Nave was rumoured to possess del Ferro’s
unpublished papers. So Cardano and Ferrari went to Bologna in 1543 to consult
del Nave, viewed the papers – and there, as plain as the nose on your face, were
solutions of all three types of cubic. So Cardano could honestly say that he was
not publishing Tartaglia’s method, but del Ferro’s.
Tartaglia didn’t see things that way. But he had no real answer to Cardano’s
point that the solution was not Tartaglia’s discovery at all, but del Ferro’s.
Tartaglia published a long, bitter diatribe about the affair, and was challenged to
a public debate by Ferrari, defending his master. Ferrari won hands down, and
Tartaglia never really recovered from the setback.

Algebraic symbolism
The mathematicians of Renaissance Italy had developed many algebraic
methods, but their notation was still rudimentary. It took hundreds of years for
today’s algebraic symbolism to develop.
One of the first to use symbols in place of unknown numbers was Diophantus
of Alexandria. His Arithmetica, written around 250, originally consisted of 13
books, of which six have survived as later copies. Its focus is the solution of
algebraic equations, either in whole numbers or in rational numbers – fractions
p/q where p and q are whole numbers. Diophantus’s notation differs
considerably from what we use today. Although the Arithmetica is the only
surviving document on this topic, there is fragmentary evidence that Diophantus
was part of a wider tradition, and not just an isolated figure. Diophantus’s
notation is not very well suited to calculations, but it does summarize them in a
compact form.

‘It took hundreds of years for today’s algebraic

symbolism to develop.’

The Arabic mathematicians of the Medieval period developed sophisticated

methods for solving equations, but expressed them in words, not symbols.
The move to symbolic notation gained momentum in the Renaissance period.
The first of the great algebraists to start using symbols was François Vieta, who
stated many of his results in symbolic form, but his notation differed
considerably from the modern one. He did, however, use letters of the alphabet
to represent known quantities, as well as unknown ones. To distinguish these, he
adopted the convention that consonants B, C, D, F, G. . . represented known
quantities, whereas vowels A, E, I,. . . represented unknowns.

Girolamo Cardano
(aka Hieronymus Cardanus, Jerome Cardan)
1501–1576
Girolamo Cardano was the illegitimate son of the Milanese lawyer Fazio Cardano and a young
widow named Chiara Micheria who was trying to bring up three children. The children died of
the plague in Milan while Chiara was giving birth to Girolamo in nearby Pavia. Fazio was an able
mathematician, and he passed on his passion for the subject to Girolamo. Against his father’s
wishes, Girolamo studied medicine at Pavia University; Fazio had wanted him to study law.

While still a student, Cardano was elected rector of the University of Padua, to which he had
moved, by a single vote. Having spent a small legacy from his recently deceased father, Cardano
turned to gambling to augment his finances: cards, dice and chess. He always carried a knife and
once slashed the face of an opponent whom he believed he had caught cheating.
In 1525 Cardano gained his medical degree, but his application to join the College of
Physicians in Milan was rejected, probably because he had a reputation for being difficult. He
practised medicine in the village of Sacca, and married Lucia Bandarini, a militia captain’s
daughter. The practice did not prosper, and in 1533 Girolamo again turned to gambling, but now
he lost heavily, and had to pawn his wife’s jewellery and some of the family furniture.
Cardano struck lucky, and was offered his father’s old position as lecturer in mathematics at
the Piatti Foundation. He continued practising medicine on the side, and some miraculous cures
enhanced his reputation as a doctor. By 1539, after several attempts, he was finally admitted to
the College of Physicians. He began to publish scholarly texts on a variety of topics, including
mathematics. Cardano wrote a remarkable autobiography, The Book of My Life, a miscellany of
chapters on numerous topics. His fame was at its peak, and he visited Edinburgh to treat the
Archbishop of St Andrews, John Hamilton. Hamilton suffered from severe asthma. Under
Cardano’s care, his health improved dramatically, and Cardano left Scotland 2000 gold crowns
the richer.
 He became professor at Pavia University, and things were going swimmingly until his eldest son
Giambatista secretly married Brandonia di Seroni, ‘a worthless, shameless woman’ in Cardano’s
estimation. She and her family publicly humiliated and taunted Giambatista, who poisoned her.
Despite Cardano’s best efforts, Giambatista was executed. In 1570 Cardano was tried for heresy,
having cast the horoscope of Jesus. He was imprisoned, then released, but banned from university
employment. He went to Rome, where the Pope unexpectedly gave him a pension and he was
admitted to the College of Physicians.
He forecast the date of his own death, and allegedly made sure he was right by committing
suicide. Despite many tribulations, he remained an optimist to the end.

In the 15th century, a few rudimentary symbols made their appearance,

notably the letters p and m for addition and subtraction: plus and minus. These
were abbreviations rather than true symbols. The symbols + and – also appeared
around this time. They arose in commerce, where they were used by German
merchants to distinguish overweight and underweight items. Mathematicians
quickly began to employ them too, the first written examples appearing in 1481.
William Oughtred introduced the symbol × for multiplication, and was roundly
(and rightly) criticized by Leibniz on the grounds that this was too easily
confused with the letter x.

Diophantus’s Notation and Ours

Meaning Modern symbol Diophantus’s symbol

The unknown x γ

Its square x2 Δγ

Its cube x3 Kγ

Its fourth power x4 ΔγΔ

Its fifth power x5 ΔKγ

Its sixth power x6 KγK

Addition + Juxtapose terms (use AB for A+B)

 Subtraction –

Equality = ɩσ

In 1557, in his The Whetstone of Witte, the English mathematician Robert

Recorde invented the symbol = for equality, in use ever since. He wrote that he
could think of no two things that were more alike than two parallel lines of the
same length. However, he used much longer lines than we do today, more like
=========. Vieta initially wrote the word ‘aequalis’ for equality, but later
replaced it by the symbol ~. René Descartes used a different symbol, .
The current symbols > and < for ‘greater than’ and ‘less than’ are due to
Thomas Harriot. Round brackets ( ) show up in 1544, and square [ ] and curly
{ } brackets were used by Vieta around 1593. Descartes used the square root
symbol , which is an elaboration on the letter r for radix, or root; but he wrote
for the cube root.
To see how different Renaissance algebraic notation was from ours, here is a
short extract from Cardano’s Ars Magna:

5p: m:15
5m: m:15
25m:m:15 qd. est 40

In modern notation this would read:

So here we see p: and m: for plus and minus, for ‘square root’, and ‘qd. est’
abbreviating the Latin phrase ‘which is’. He wrote

qdratu aeqtur 4 rebus p:32

where we would write

 x2 = 4x + 32

and therefore used separate abbreviations ‘rebus’ and ‘qdratu’ for the unknown
(thing) and its square. Elsewhere he used R for the unknown, Z for its square and
C for its cube.
An influential but little-known figure was the Frenchman Nicolas Chuquet,
whose book Triparty en la Science de Nombres of 1484 discussed three main
mathematical topics: arithmetic, roots and unknowns. His notation for roots was
much like Cardano’s, but he started to systematize the treatment of powers of the
unknown, by using superscripts. He referred to the first four powers of the
unknown as premier, champs, cubiez and champs de champs. For what we
would now write as 6x, 4x2 and 5x3 he used .6.1, .4.2 and .5.3. He also used
zeroth and negative powers, writing .2.0 and .3.1.m. where we would write 2 and
3x−1. In short: he used exponential notation (superscripts) for powers of the
unknown, but had no explicit symbol for the unknown itself.
That omission was supplied by Descartes. His notation was very similar to
what we use nowadays, with one exception. Where we would write

5 + 4x + 6x2 + 11x3 + 3x4

say, Descartes wrote

5 + 4x + 6xx + 11x3 + 3x4

That is, he used xx for the square. Occasionally, though he used x2. Newton
wrote powers of the unknown exactly as we do now, including fractional and
negative exponents, such as x3/2 for the square root of x3. It was Gauss who
finally abolished xx in favour of x2; once the Grand Master had done this,
everyone else followed suit.

The logic of species

Algebra began as a way to systematize problems in arithmetic, but by the time of

Vieta it had acquired a life of its own. Before Vieta, algebraic symbolism and
manipulation were viewed as ways to state and carry out arithmetical
procedures, but numbers were still the main point. Vieta made a crucial
distinction between what he called the logic of species and the logic of numbers.
In his view, an algebraic expression represented an entire class (species) of
arithmetical expressions. It was a different concept. In his 1591 In Artem
Analyticam Isagoge (Introduction to the Analytic Art) he explained that algebra
is a method for operating on general forms, whereas arithmetic is a method for
operating on specific numbers.
This may sound like logical hair-splitting, but the difference in the point of
view was significant. To Vieta, an algebraic calculation such as (in our notation)

(2x + 3y) − (x + y) = x + 2y

expresses a way to manipulate symbolic expressions. The individual terms 2x +

3y and so on are themselves mathematical objects. They can be added,
subtracted, multiplied and divided without ever considering them as
representations of specific numbers. To Vieta’s predecessors, however, the same
equation was simply a numerical relationship that was valid whenever specific
numbers were substituted for the symbols x and y. So algebra took on a life of its
own, as the mathematics of symbolic expressions. It was the first step towards
freeing algebra from the shackles of arithmetical interpretation.

What algebra does for us

The leading consumers of algebra in the modern world are scientists, who represent
nature’s regularities in terms of algebraic equations. These equations can be solved to
represent unknown quantities in terms of known ones. The technique has become so
routine that no one notices they’re using algebra.
Algebra was very nearly applied to archaeology in one episode of Time Team, when
the intrepid TV archaeologists wanted to work out how deep a mediaeval well was. The
first idea was to drop something down it, and time how long it took to reach the bottom. It
took six seconds. The relevant algebraic formula here is

s = ½gt2

where s is the depth, t is the time taken to hit the bottom, and g is the acceleration due to
gravity, roughly 10 metres per second2. Taking t = 6, the formula tells us that the well is
roughly 180 metres deep.
Because of some uncertainty about the formula – which in fact they had remembered
correctly – the Time Team used three long tape-measures tied together.
The measured depth was in fact very close to 180 metres.
Algebra enters more obviously if we know the depth and want to calculate the time.
Now we have to solve the equation for t in terms of s, leading to the answer

Knowing that s = 180 metres for instance, lets us predict that t is the square root of
360/10, that is, the square root of 36 – which is 6 seconds.
 5
Eternal Triangles

Trigonometry and logarithms

Euclidean geometry is based on triangles, mainly because every polygon can be

built from triangles, and most other interesting shapes, such as circles and
ellipses, can be approximated by polygons. The metric properties of triangles –
those that can be measured, such as the lengths of sides, the sizes of angles or the
total area – are related by a variety of formulas, many of them elegant. The
practical use of these formulas, which are extremely useful in navigation and
surveying, required the development of trigonometry, which basically means
‘measuring triangles’.

Trigonometry
Trigonometry spawned a number of special function – mathematical rules for
calculating one quantity from another. These functions go by names like sine,
cosine and tangent. The trigonometric functions turned out to be of vital
importance for the whole of mathematics, not just for measuring triangles.
Trigonometry is one of the most widely used mathematical techniques,
involved in everything from surveying to navigation to GPS satellite systems in
cars. Its use in science and technology is so common that it usually goes
unnoticed, as befits any universal tool. Historically, it was closely associated
with logarithms, a clever method for converting multiplications (which are hard)
into additions (which are easier). The main ideas appeared between about 1400
and 1600, though with a lengthy prehistory and plenty of later embellishments,
and the notation is still evolving.

‘Humanity owes a great deal to these dedicated and

dogged pioneers.’

In this chapter we’ll take a look at the basic topics: trigonometric functions,
the exponential function and the logarithm. We also consider a few applications,
old and new. Many of the older applications are computational techniques,
which have mostly become obsolete now that computers are widespread. Hardly
anyone now uses logarithms to do multiplication, for instance. No one uses
tables at all, now that computers can rapidly calculate the values of functions to
high precision. But when logarithms were first invented, it was the numerical
tables of them that made them useful, especially in areas like astronomy, where
long and complicated numerical calculations were necessary. And the compilers
of the tables had to spend years – decades – of their lives doing the sums.
Humanity owes a great deal to these dedicated and dogged pioneers.

The origins of trigonometry

The basic problem addressed by trigonometry is the calculation of properties of a

triangle – lengths of sides, sizes of angles – from other such properties. It is
much easier to describe the early history of trigonometry if we first summarize
the main features of modern trigonometry, which is mostly a reworking in 18th
century notation of topics that go right back to the Greeks, if not earlier. This
summary provides a framework within which we can describe the ideas of the
ancients, without getting tangled up in obscure and eventually obsolete concepts.

Trigonometry – a Primer
Trigonometry relies on a number of special functions, of which the most basic are the sine, cosine
and tangent. These functions apply to an angle, traditionally represented by the Greek letter θ
(theta). They can be defined in terms of a right triangle, whose three sides a, b, c are called the
adjacent side, the opposite side and the hypotenuse.

Then:

The sine of theta is sin θ = b/c

The cosine of theta is cos θ = a/c
The tangent of theta is tan θ = b/a

As it stands, the values of these three functions, for any given angle θ, are determined by the
geometry of the triangle. (The same angle may occur in triangles of different sizes, but the
geometry of similar triangles implies that the stated ratios are independent of size.) However,
once these functions have been calculated and tabulated, they can be used to solve (calculate all
the sides and angles of) the triangle from the value of.
The three functions are related by a number of beautiful formulas. In particular, Pythagoras’s
Theorem implies that
 sin2θ + cos2θ = 1

Trigonometry seems to have originated in astronomy, where it is relatively

easy to measure angles, but difficult to measure the vast distances. The Greek
astronomer Aristarchus, in a work of around 260 BC, On the Sizes and Distances
of the Sun and Moon, deduced that the Sun lies between 18 and 20 times as far
from the Earth as the Moon does. (The correct figure is closer to 400, but
Eudoxus and Phidias had argued for 10.) His reasoning was that when the Moon
is half full, the angle between the directions from the observer to the Sun and the
Moon is about 87° (in modern units). Using properties of triangles that amount
to trigonometric estimates, he deduced (in modern notation) that sin 3° lies
between 1/18 and 1/20, leading to his estimate of the ratio of the distances to the
Sun and the Moon. The method was right, but the observation was inaccurate;
the correct angle is 89.8°.
The first trigonometric tables were derived by Hipparchus around 150 BC.
Instead of the modern sine function, he used a closely related quantity, which
from the geometric point of view was equally natural. Imagine a circle, with two
radial lines meeting at an angle θ. The points where these lines cut the circle can
be joined by a straight line, called a chord. They can also be thought of as the
end points of a curved arc of the circle. Hipparchus drew up a table relating arc
and chord length for a range of angles. If the circle has radius 1, then the arc
length is equal to θ when this angle is measured in units known as radians. Some
easy geometry shows that the chord length in modern notation is 2sin . So
Hipparchus’s calculation is very closely related to a table of sines, even though it
was not presented in that way.

Relation between the Sun, Moon, and Earth when the Moon is half full
 Arc and chord corresponding to an angle θ

Astronomy

Remarkably, early work in trigonometry was more complicated than most of

what is taught in schools today, again because of the needs of astronomy (and,
later, navigation). The natural space to work in was not the plane, but the sphere.
Heavenly objects can be thought of as lying on an imaginary sphere, the celestial
sphere. Effectively, the sky looks like the inside of a gigantic sphere surrounding
the observer, and the heavenly bodies are so distant that they appear to lie on this
sphere.
Astronomical calculations, in consequence, refer to the geometry of a sphere,
not that of a plane. The requirements are therefore not plane geometry and
trigonometry, but spherical geometry and trigonometry. One of the earliest
works in this area is Menelaus’s Sphaerica of about AD100. A sample theorem,
one that has no analogue in Euclidean geometry, is this: if two triangles have the
same angles as each other, then they are congruent – they have the same size and
shape. (In the Euclidean case, they are similar – same shape but possibly
different sizes.) In spherical geometry, the angles of a triangle do not add up to
180°, as they do in the plane. For example, a triangle whose vertices lie at the
North Pole and at two points on the equator separated by 90° clearly has all three
angles equal to a right angle, so the sum is 270°. Roughly speaking, the bigger
the triangle becomes, the bigger its angle-sum becomes. In fact, this sum, minus
180°, is proportional to the triangle’s total area.
 North Pole

These examples make it clear that spherical geometry has its own
characteristic and novel features. The same goes for spherical trigonometry, but
the basic quantities are still the standard trigonometric functions. Only the
formulas change.

Ptolemy

By far and above the most important trigonometry text of antiquity was the
Mathematical Syntaxis of Ptolemy of Alexandria, which dates to about AD150. It
is better known as the Almagest, an Arabic term meaning ‘the greatest’. It
included trigonometric tables, again stated in terms of chords, together with the
methods used to calculate them, and a catalogue of star positions on the celestial
sphere. An essential feature of the computational method was Ptolemy’s
Theorem which states that if ABCD is a cyclic quadrilateral (one whose vertices
lie on a circle) then

AB × CD + BC × DA = AC × BD
 Cyclic quadrilateral and its diagonals

(the sum of the products of opposite pairs of sides is equal to the product of the
diagonals).
A modern interpretation of this fact is the remarkable pair of formulas

sin (θ + φ) = sin θ cos φ + cos θ sin φ

cos (θ + φ) = cos θ cos φ – sin θ sin φ

The main point about these formulas is that if you know the sines and cosines of
two angles, then you can easily work the sines and cosines out for the sum of
those angles. So, starting with (say) sin 1° and cos 1°, you can deduce sin 2° and
cos 1° by taking θ = φ = 1°. Then you can deduce sin 3° and cos 3° by taking θ =
1°, φ = 2°, and so on. You had to know how to start, but after that, all you
needed was arithmetic – rather a lot of it, but nothing more complicated.
Getting started was easier than it might seem, requiring arithmetic and square
roots. Using the obvious fact that θ/2 + θ/2 = θ, Ptolemy’s Theorem implies that

Starting from cos 90° = 0, you can repeatedly halve the angle, obtaining sines
and cosines of angles as small as you please. (Ptolemy used ¼°.) Then you can
work back up through all integer multiples of that small angle. In short, starting
with a few general trigonometric formulas, suitably applied, and a few simple
values for specific angles you can work out values for pretty much any angle you
want. It was an extraordinary tour de force, and it put astronomers in business
for well over a thousand years.
A final noteworthy feature of the Almagest s how it handled the orbits of the
planets. Anyone who watches the night sky regularly quickly discovers that the
planets wander against the background of fixed stars, and that the paths they
follow seem rather complicated, sometimes moving backwards, or travelling in
elongated loops.
Eudoxus, responding to a request from Plato, had found a way to represent
these complex motions in terms of revolving spheres mounted on other spheres.
This idea was simplified by Apollonius and Hipparchus, to use epicycles –
circles whose centres move along other circles, and so on. Ptolemy refined the
system of epicycles, so that it provided a very accurate model of the planetary
motions.

Early trigonometry

Early trigonometric concepts appear in the writings of Hindu mathematicians

and astronomers: Varahamihira’s Pancha Siddhanta of 500, Brahmagupta’s
Brahma Sputa Siddhanta of 628 and the more detailed Siddhanta Siromani of
Bhaskaracharya in 1150.
 Motion of Mars as viewed from the Earth

Indian mathematicians generally used the half-chord, or jya-ardha, which is in

effect the modern sine. Varahamihira calculated this function for 24 integer-
multiples of 3°45, up to 90°. Around 600, in the Maha Bhaskariya, Bhaskara
gave a useful approximate formula for the sine of an acute angle, which he
credited to Aryabhata. These authors derived a number of basic trigonometric
formulas.
The Arabian mathematician Nasîr-Eddin’s Treatise on the Quadrilateral
combined plane and spherical geometry into a single unified development, and
gave several basic formulas for spherical triangles. He treated the topic
mathematically, rather than as a part of astronomy. But his work went unnoticed
in the West until about 1450.

‘Early trigonometric concepts appear in the writings

of Hindu mathematicians and astronomers.’
 Because of the link with astronomy, almost all trigonometry was spherical
until 1450. In particular, surveying – today a major user of trigonometry – was
carried out using empirical methods, codified by the Romans. But in the mid-
15th century, plane trigonometry began to come into its own, initially in the
north German Hanseatic League. The League was in control of most trade, and
consequently was rich and influential. And it needed improved navigational
methods, along with improved timekeeping and practical uses of astronomical
observations.
A key figure was Johannes Müller, usually known as Regiomontanus. He was
a pupil of George Peuerbach, who began working on a new corrected version of
the Almagest. In 1471, financed by his patron Bernard Walther, he computed a
new table of sines and a table of tangents.
Other prominent mathematicians of the 15th and 16th centuries computed
their own trigonometric tables, often to extreme accuracy. George Joachim
Rhaeticus calculated sines for a circle of radius 1015 – effectively, tables accurate
to 15 decimal places, but multiplying all numbers by 1015 to get integers – for all
multiples of one second of arc. He stated the law of sines for spherical triangles

and the law of cosines

cos a = cos b cos c + sin b sin c cos A

in his De Triangulis, written in 1462–3 but not published until 1533. Here A, B,
C are the angles of the triangle, and a, b, c are its sides – measured by the angles
they determine at the centre of the sphere.
Vieta wrote widely on trigonometry, with his first book on the topic being the
Canon Mathematicus of 1579. He collected and systematized various methods
for solving triangles, that is, calculating all sides and angles from some subset of
information. He invented new trigonometric identities, among them some
interesting expressions for sines and cosines of integer multiples of θ in terms of
the sine and cosine of θ.

Logarithms
The second theme of this chapter is one of the most important functions in
mathematics: the logarithm, log x. Initially, the logarithm was important because
it satisfies the equation

log xy = log x + log y

and can therefore be used to convert multiplications (which are cumbersome)

into additions (which are simpler and quicker). To multiply two numbers x and
y, first find their logarithms, add them and then find the number whose logarithm
is that result (its antilogarithm). This is the product xy.
Once mathematicians had calculated tables of logarithms, they could be used
by anyone who understood the method. From the 17th century until the mid-
20th, virtually all scientific calculations, especially astronomical ones, employed
logarithms. From the 1960s onwards, however, electronic calculators and
computers rendered logarithms obsolete for purposes of calculation. But the
concept remained vital to mathematics, because logarithms had found
fundamental roles in many parts of mathematics, including calculus and complex
analysis. Also many physical and biological processes involve logarithmic
behaviour.
Nowadays we approach logarithms by thinking of them as the reverse of
exponentials. Using logarithms to base 10, which are a natural choice for
decimal notation, we say that x is the logarithm of y if y = 10x. For example,
since 103 = 1000, the logarithm of 1000 (to base 10) is 3. The basic property of
logarithms follows from the exponential law

10a+b = 10a × 10b

However, in order for the logarithm to be useful, we have to be able to find a

suitable x for any positive real y. Following the lead of Newton and others of his
period, the main idea is that any rational power 10p/q can be defined to be the qth
root of 10p. Since any real number x can be approximated arbitrarily closely by a
rational number p/q, we can approximate 10x by 10p/q. This is not the most
efficient way to calculate the logarithm, but it is the simplest way to prove that it
exists.
Historically, the discovery of logarithms was less direct. It began with John
Napier, Baron of Murchiston in Scotland. He had a lifelong interest in efficient
methods for calculation, and invented Napier’s rods (or Napier’s bones), a set of
marked sticks that could be used to perform multiplication quickly and reliably
by simulating pen-and-paper methods. Around 1594 he started working on a
more theoretical method, and his writings tell us that it took him 20 years to
perfect and publish it. It seems likely that he started with geometric progressions,
sequences of numbers in which each term is obtained from the previous one by
multiplying by a fixed number – such as the powers of 2

Plane Trigonometry
Nowadays trigonometry is first developed in the plane, where the geometry is simpler and the
basic principles are easier to grasp. (It is curious how often new mathematical ideas are first
developed in a complicated context, and the underlying simplicities emerge much later.) There is
a law of sines, and a law of cosines, for plane triangles, and it is worth a quick digression to
explain these. Consider a plane triangle with angles A, B, C and sides a, b, c.
Now the law of sines takes the form

and the law of cosines is

a2 = b2 + c2 – 2bc cos A,

with companion formulas involving the other angles. We can use the law of cosines to find the
angles of a triangle from its sides.
 Sides and angles of a triangle

1 2 4 8 16 32 . . .

or powers of 10

1 10 100 1000 10,000 100,000 . . .

Here it had long been noticed that adding the exponents was equivalent to
multiplying the powers. This was fine if you wanted to multiply two integer
powers of 2, say, or two integer powers of 10. But there were big gaps between
these numbers, and powers of 2 or 10 seemed not to help much when it came to
problems like 57.681 × 29.443, say.

Napierian logarithms

While the good Baron was trying to somehow fill in the gaps in geometric
progressions, the physician to King James VI of Scotland, James Craig, told
Napier about a discovery that was in widespread use in Denmark, with the
ungainly name prosthapheiresis. This referred to any process that converted
products into sums. The main method in practical use was based on a formula
discovered by Vieta:

If you had tables of sines and cosines, you could use this formula to convert a
product into a sum. It was messy, but it was still quicker than multiplying the
numbers directly.
Napier seized on the idea, and found a major improvement. He formed a
geometric series with a common ratio very close to 1. That is, in place of the
powers of 2 or powers of 10, you should use powers of, say, 1.0000000001.
Successive powers of such a number are very closely spaced, which gets rid of
those annoying gaps. For some reason Napier chose a ratio slightly less than 1,
namely 0.9999999. So his geometric sequence ran backwards from a large
number to successively smaller ones. In fact, he started with 10,000,000 and then
multiplied this by successive powers of 0.9999999. If we write Naplog x for
Napier’s logarithm of x, it has the curious feature that

‘Napier seized on the idea, and found a major

improvement.’

Naplog 10,000,000 = 0
Naplog 9,999,999 = 1

and so on. The Napierian logarithm, Naplog x, satisfies the equation

Naplog(107xy) = Naplog(x) + Naplog(y)

You can use this for calculation, because it is easy to multiply or divide by a
power of 10, but it lacks elegance. It is, however, much better than Vieta’s
trigonometric formula.

Base ten logarithms

The next improvement came when Henry Briggs, the first Savilian professor of
geometry at the University of Oxford, visited Napier. Briggs suggested replacing
Napier’s concept by a simpler one: the (base ten) logarithm, L = log10 x, which
satisfies the condition

x = 10L.

Now

log10 xy = log10 x + log10 y

and everything is easy. To find xy, add the logarithms of x and y and then find
the antilogarithm of the result.
Before these ideas could be disseminated, Napier died; the year was 1617, and
his description of his calculating rods, Rhabdologia, had just been published. His
original method for calculating logarithms, the Mirifici Logarithmorum Canonis
Constructio, appeared two years later. Briggs took up the task of computing a
table of Briggsian (base 10, or common) logarithms. He did this by starting from
log1010 = 1 and taking successive square roots. In 1617 he published
Logarithmorum Chilias Prima, the logarithms of the integers from 1 to 1000,
stated to 14 decimal places. His 1624 Arithmetic Logarithmica tabulated
common logarithms of numbers from 1 to 20,000 and from 90,000 to 100,000,
also to 14 places.

What trigonometry did for them

Ptolemy’s Almagest formed the basis of all studies of planetary motion prior to Johannes
Kepler’s discovery that orbits are elliptical. The observed movements of a planet are
complicated by the relative motion of the Earth, which was not recognized in Ptolemy’s
time. Even if planets moved at uniform speed in circles, the Earth’s motion round the
Sun would effectively require a combination of two different circular motions, and an
accurate model has to be distinctly more complicated than Ptolemy’s model. Ptolemy’s
scheme of epicycles combines circular motions by making the centre of one circle
revolve around another circle. This circle can itself revolve round a third circle, and so
on. The geometry of uniform circular motion naturally involves trigonometric functions,
and later astronomers used these for calculations of orbits.

An epicycle. Planet P revolves uniformly around point D, which in turn revolves uniformly
around point C.

The idea snowballed. John Speidell worked out logarithms of trigonometric

functions (such as log sin x) published as New Logarithmes in 1619. The Swiss
clockmaker Jobst Bürgi published his own work on logarithms in 1620, and may
well have possessed the basic idea in 1588, well before Napier. But the historical
development of mathematics depends on what people publish – in the original
sense of make public – and ideas that remain private have no influence on
anyone else. So credit, probably rightly, has to go to those people who put their
ideas into print, or at least into widely circulated letters. (The exception is people
who put the ideas of others into print without due credit. This is generally
beyond the pale.)

The number e

Associated with Napier’s version of logarithms is one of the most important

numbers in mathematics, now denoted by the letter e. Its value is roughly
2.7128. It arises if we try to from logarithms by starting from a geometric series
whose common ratio is very slightly larger than 1. This leads to the expression
(1 + 1/n)n, where n is a very large integer, and the larger n becomes, the closer
this expression is to one special number, which we denote by e.
This formula suggests that there is a natural base for logarithms, and it is
neither 10 nor 2, but e. The natural logarithm of x is whichever number y
satisfies the condition x = ey. In today’s mathematics the natural logarithm is
written y = log x. Sometimes the base e is made explicit, as y = loge x, but this
notation is mainly restricted to school mathematics, because in advanced
mathematics and science the only logarithm of importance is the natural
logarithm. Base ten logarithms are best for calculations in decimal notation, but
natural logarithms are more fundamental mathematically.
The expression ex is called the exponential of x, and it is one of the most
important concepts in the whole of mathematics. The number e is one of those
strange special numbers that appear in mathematics, and have major
significance. Another such number is π. These two numbers are the tip of an
iceberg – there are many others. They are also arguably the most important of
the special numbers because they crop up all over the mathematical landscape.

Where would we be without them?

It would be difficult to underestimate the debt that we owe to those far-seeing

individuals who invented logarithms and trigonometry, and spent years
calculating the first numerical tables. Their efforts paved the way to a
quantitative scientific understanding of the natural world, and enabled
worldwide travel and commerce by improving navigation and map-making. The
basic techniques of surveying rely on trigonometric calculations. Even today,
when surveying equipment uses lasers and the calculations are done on a
custom-built electronic chip, the concepts that the laser and the chip embody are
direct descendants of the trigonometry that intrigued the mathematicians of
ancient India and Arabia.
Logarithms made it possible for scientists to do multiplication quickly and
accurately. Twenty years of effort on a book of tables, by one mathematician,
saved tens of thousands of man-years of work later on. It then became possible
to carry out scientific analyses using pen and paper that would otherwise have
been too time-consuming. Science could never have advanced without some
such method. The benefits of such a simple idea have been incalculable.

What trigonometry does for us

Trigonometry is fundamental to surveying anything from building sites to continents. It is
relatively easy to measure angles to high accuracy, but harder to measure distances,
especially on difficult terrain. Surveyors therefore begin by making a careful
measurement of one length, the baseline, that is, the distance between two specific
locations. They then form a network of triangles, and use the measured angles, plus
trigonometry, to calculate the sides of these triangles. In this manner, an accurate map
of the entire area concerned can be constructed. This process is known as triangulation.
To check its accuracy, a second distance measurement can be made once the
triangulation is complete.
The figure here shows an early example, a famous survey carried out in South Africa
in 1751 by the great astronomer Abbé Nicolas Louis de Lacaille. His main aim was to
catalogue the stars of the southern skies, but to do this accurately he first had to
measure the arc of a suitable line of longitude. To do this, he developed a triangulation
to the north of Cape Town.
 LaCaille’s triangulation of South Africa

His result suggested that the curvature of the earth is less in southern latitudes than in
northern ones, a surprising deduction which was verified by later measurements. The
Earth is slightly pear-shaped. His cataloguing activities were so successful that he
named 15 of the 88 constellations now recognized, having observed more than 10,000
stars using a small refracting telescope.
 6
Curves and Coordinates

Geometry is algebra is geometry

Although it is usual to classify mathematics into separate areas, such as

arithmetic, algebra, geometry and so on, this classification owes more to human
convenience than the subject’s true structure. In mathematics, there are no hard
and fast boundaries between apparently distinct areas, and problems that seem to
belong to one area may be solved using methods from another. In fact, the
greatest breakthroughs often hinge upon making some unexpected connection
between previously distinct topics.

Fermat

Greek mathematics has traces of such connections, with links between

Pythagoras’s Theorem and irrational numbers, and Archimedes’s use of
mechanical analogies to find the volume of the sphere. The true extent and
influence of such cross-fertilization became undeniable in a short period ten
years either side of 1630. During that brief time, two of the world’s greatest
mathematicians discovered a remarkable connection between algebra and
geometry. In fact, they showed that each of these areas can be converted into the
other by using coordinates. Everything in Euclid, and the work of his successors,
can be reduced to algebraic calculations. Conversely, everything in algebra can
be interpreted in terms of the geometry of curves and surfaces.
It might seem that such connections render one or other area superfluous. If all
geometry can be replaced by algebra, why do we need geometry? The answer is
that each area has its own characteristic point of view, which can on occasion be
very penetrating and powerful. Sometimes it is best to think geometrically, and
sometimes algebraic thinking is superior.
The first person to describe coordinates was Pierre de Fermat. Fermat is best
known for his work in number theory, but he also studied many other areas of
mathematics, including probability, geometry and applications to optics. Around
1620, Fermat was trying to understand the geometry of curves, and he started by
reconstructing, from what little information was available, a lost book by
Apollonius called On Plane Loci. Having done this, Fermat embarked upon his
own investigations, writing them up in 1629 but not publishing them until 50
years later, as Introduction to Plane and Solid Loci. By so doing, he discovered
the advantages of rephrasing geometrical concepts in algebraic terms.

Focal property of the ellipse

Locus, plural loci, is an obsolete term today, but it was common even in 1960.
It arises when we seek all points in the plane or space that satisfy particular
geometric conditions. For example, we might ask for the locus of all points
whose distances from two other fixed points always add up to the same total.
This locus turns out to be an ellipse with the two points as its foci. This property
of the ellipse was known to the Greeks.
Fermat noticed a general principle: if the conditions imposed on the point can
be expressed as a single equation involving two unknowns, the corresponding
locus is a curve – or a straight line, which we consider to be a special kind of
curve to avoid making needless distinctions. He illustrated this principle by a
diagram in which the two unknown quantities A and E are represented as
distances in two distinct directions.

Fermat’s approach to coordinates

He then listed some special types of equation connecting A and E, and

explained what curves they represent. For instance, if A2 = 1 + E2 then the locus
concerned is a hyperbola.

‘Fermat introduced oblique axes in the plane.’

In modern terms, Fermat introduced oblique axes in the plane (oblique

meaning that they do not necessarily cross at right angles). The variables A and
E are the two coordinates, which we would call x and y, of any given point with
respect to these axes. So Fermat’s principle effectively states that any equation
in two-coordinate variables defines a curve, and his examples tell us what kind
of equations correspond to what kind of curve, drawing on the standard curves
known to the Greeks.

Descartes

The modern notion of coordinates came to fruition in the work of Descartes. In

everyday life, we are familiar with spaces of two and three dimensions, and it
takes a major effort of imagination for us to contemplate other possibilities. Our
visual system presents the outside world to each eye as a two-dimensional image
– like the picture on a TV screen. Slightly different images from each eye are
combined by the brain to provide a sense of depth, through which we perceive
the surrounding world as having three dimensions.
The key to multidimensional spaces is the idea of a coordinate system, which
was introduced by Descartes in an appendix La Géometrie to his book Discours
de la Méthode. His idea is that the geometry of the plane can be reinterpreted in
algebraic terms. His approach is essentially the same as Fermat’s. Choose some
point in the plane and call it the origin. Draw two axes, lines that pass through
the origin and meet at right angles. Label one axis with the symbol x and the
other with the symbol y. Then any point P in the plane is determined by the pair
of distances (x, y), which tells us how far the point is from the origin when
measured parallel to the x- and y-axes, respectively.
For example, on a map, x might be the distance east of the origin (with
negative numbers representing distances to the west), whereas y might be the
distance north of the origin (with negative numbers representing distances to the
south).

René Descartes
1596–1650

Descartes first began to study mathematics in 1618, as a student of the Dutch scientist Isaac
Beeckman. He left Holland to travel through Europe, and joined the Bavarian army in 1619. He
continued to travel between 1620 and 1628, visiting Bohemia, Hungary, Germany, Holland,
 France and Italy. He met Mersenne in Paris in 1622, and corresponded regularly with him
thereafter, which kept him in touch with most of the leading scholars of the period.
In 1628 Descartes settled in Holland, and started his first book Le Monde, ou Traité de la
Lumière, on the physics of light. Publication was delayed when Descartes heard of Galileo
Galilei’s house arrest, and Descartes got cold feet. The book was published, in an incomplete
form, after his death. Instead, he developed his ideas on logical thinking into a major work
published in 1637: Discours de la Méthode. The book had three appendices: La Dioptrique, Les
Météores and La Géométrie.

His most ambitious book, Principia Philosophiae, was published in 1644. It was divided into
four parts: Principles of Human Knowledge, Principles of Material Things, The Visible World
and The Earth. It was an attempt to provide a unified mathematical foundation for the entire
physical universe, reducing everything in nature to mechanics.

In 1649 Descartes went to Sweden, to act as tutor to Queen Christina. The Queen was an early
riser, whereas Descartes habitually rose at 11 o’clock. Tutoring the Queen in mathematics at 5
o’clock every morning, in a cold climate, imposed a considerable strain on Descartes’s health.
After a few months, he died of pneumonia.

Coordinates work in three-dimensional space too, but now two numbers are
not sufficient to locate a point. However, three numbers are. As well as the
distances east–west and north–south, we need to know how far the point is above
or below the origin. Usually we use a positive number for distances above, and a
negative one for distances below. Coordinates in space take the form (x, y, z).
This is why the plane is said to be two-dimensional, whereas space is three-
dimensional. The number of dimensions is given by how many numbers we need
to specify a point.
In three-dimensional space, a single equation involving x, y and z usually
defines a surface. For example, x2 + y2 + z2 = 1 states that the point (x, y, z) is
always a distance 1 unit from the origin, which implies that it lies on the unit
sphere whose centre is the origin.
Notice that the word ‘dimension’ is not actually defined here in its own right.
We do not find the number of dimensions of a space by finding some things
called dimensions and then counting them. Instead, we work out how many
numbers are needed to specify where a location in the space is, and that is the
number of dimensions.
 Coordinates as Used Today
The early development of coordinate geometry will make more sense if we first explain how the
modern version goes. There are several variants, but the commonest begins by drawing two
mutually perpendicular lines in the plane, called axes. Their common meeting point is the origin.
The axes are conventionally arranged so that one of them is horizontal and the other vertical.
Along both axes we write the whole numbers, with negative numbers going in one direction
and positive numbers in the other. Conventionally, the horizontal axis is called the x-axis, and the
vertical one is the y-axis. The symbols x and y are used to represent points on those respective
axes – distances from the origin. A general point in the plane, distance x along the horizontal axis
and distance y along the vertical one, is labelled with a pair of numbers (x, y). These numbers are
the coordinates of that point.
Any equation relating x and y restricts the possible points. For example, if x2 + y2 = 1 then (x,
y) must lie at distance 1 from the origin, by Pythagoras’s Theorem. Such points form a circle. We
say that x2 + y2 = 1 is the equation for that circle. Every equation corresponds to some curve in
the plane; conversely every curve corresponds to an equation.

Cartesian coordinates

Cartesian coordinate geometry reveals an algebraic unity behind the conic

sections – curves that the Greeks had constructed as sections of a double cone.
Algebraically, it turns out that the conic sections are the next simplest curves
after straight lines. A straight line corresponds to a linear equation

ax + by + c = 0

with constants a, b, c. A conic section corresponds to a quadratic equation

ax2 + bxy + cy2 + dx + ey + f = 0

with constants a, b, c, d, e, f. Descartes stated this fact, but did not provide a
proof. However, he did study a special case, based on a theorem due to Pappus
which characterized conic sections, and he showed that in this case the resulting
equation is quadratic.
He went on to consider equations of higher degree, defining curves more
complex than most of those arising in classical Greek geometry. A typical
example is the folium of Descartes, with equation

x3 + y3 – 3axy = 0

which forms a loop with two ends that tend to infinity.

The folium of Descartes

Perhaps the most important contribution made by the concept of coordinates

occurred here: Descartes moved away from the Greek view of curves as things
that are constructed by specific geometric means, and saw them as the visual
aspect of any algebraic formula. As Isaac Newton remarked in 1707, ‘The
moderns advancing yet much further [than the Greeks] have received into
geometry all lines that can be expressed by equations’.
Later scholars invented numerous variations on the Cartesian coordinate
system. In a letter of 1643 Fermat took up Descartes’s ideas and extended them
to three dimensions. Here he mentioned surfaces such as ellipsoids and
paraboloids, which are determined by quadratic equations in the three variables
x, y, z. An influential contribution was the introduction of polar coordinates by
Jakob Bernoulli in 1691. He used an angle θ and a distance r to determine points
in the plane instead of a pair of axes. Now the coordinates are (r, θ).
Again, equations in these variables specify curves. But now, simple equations
can specify curves that would become very complicated in Cartesian
coordinates. For example the equation r = θ corresponds to a spiral, of the kind
known as an Archimedean spiral.

Polar coordinates

Functions

An important application of coordinates in mathematics is a method to represent

functions graphically.
A function is not a number, but a recipe that starts from some number and
calculates an associated number. The recipe involved is often stated as a
formula, which assigns to each number, x (possibly in some limited range),
another number, f(x).

Archimedean spiral

For example, the square root function is defined by the rule f(x) = , that is,
take the square root of the given number. This recipe requires x to be positive.
Similarly the square function is defined by f(x) = x2, and this time there is no
restriction on x.

Which Bernoulli Did What?

A Bernoulli Checklist
The Swiss Bernoulli family had a huge influence on the development of mathematics. Over some
four generations they produced significant mathematics, both pure and applied. Often described
as a mathematical mafia, the Bernoullis typically started out in careers like law, medicine or the
church, but eventually reverted to type and became mathematicians, either professional or
amateur.
Many different mathematical concepts bear the name Bernoulli. This is not always the same
Bernoulli. Rather than providing biographical details about them, here is a summary of who did
what.

Jacob I (1654–1705)
Polar coordinates, formula for the radius of curvature of a plane curve. Special curves, such as the
catenary and lemniscate. Proved that an isochrone (a curve along which a body will fall with
uniform vertical velocity) is an inverted cycloid. Discussed isoperimetric figures, having the
shortest length under various conditions, a topic that later led to the calculus of variations. Early
student of probability and author of the first book on the topic, Ars Conjectandi. Asked for a
logarithmic spiral to be engraved on his tombstone, along with the inscription Eadem mutata
resurgo (I shall arise the same though changed).

Johann I (1667–1748)
Developed the calculus and promoted it in Europe. The Marquis de L’Hôpital put Johann’s work
into the first calculus textbook. ‘L’Hôpital’s rule’ for evaluating limits reducing to 0/0 is due to
Johann. Wrote on optics (reflection and refraction), orthogonal trajectories of families of curves,
lengths of curves and evaluation of areas by series, analytical trigonometry and the exponential
function. Brachistochrone (curve of quickest descent), length of cycloid.

Nicolaus I (1687–1759)
Held Galileo’s chair of mathematics at Padua. Wrote on geometry and differential equations.
Later, taught logic and law. A gifted but not very productive mathematician. Corresponded with
Leibniz, Euler and others – his main achievements are scattered among 560 items of
correspondence. Formulated the St Petersburg Paradox in probability.
Criticized Euler’s indiscriminate use of divergent series. Assisted in publication of Jacob
Bernoulli’s Ars Conjectandi. Supported Leibniz in his controversy with Newton.

Nicolaus II (1695–1726)
Called to the St. Petersburg Academy, died by drowning eight months later. Discussed the St
 Petersburg Paradox with Daniel.

Daniel (1700–1782)
Most famous of Johann’s three sons. Worked on probability, astronomy, physics and
hydrodynamics. His Hydrodynamica of 1738 contains Bernoulli’s principle about the relation
between pressure and velocity. Wrote on tides, the kinetic theory of gases and vibrating strings.
Pioneer in partial differential equations.

Johann II (1710–1790)
Youngest of Johann’s three sons. Studied law but became professor of mathematics at Basel.
Worked on the mathematical theory of heat and light.

Johann III (1744–1807)

Like his father, studied law but then turned to mathematics. Called to the Berlin Academy at age
19. Wrote on astronomy, chance and recurring decimals.

Jacob II (1759–1789)
Important works on elasticity, hydrostatics, and ballistics.

We can picture a function geometrically by defining the y-coordinate, for a

given value of x, by y = f(x). This equation states a relationship between the two
coordinates, and therefore determines a curve. This curve is called the graph of
the function f.
The graph of the function f(x) = x2 turns out to be a parabola. That of the
square root f(x) = is half a parabola, but lying on its side. More complicated
functions lead to more complicated curves. The graph of the sine function y = sin
x is a wiggly wave.
 Graph of a function f

Coordinate geometry today

Coordinates are one of those simple ideas that has had a marked influence on
everyday life. We use them everywhere, usually without noticing what we’re
doing. Virtually all computer graphics employ an internal coordinate system, and
the geometry that appears on the screen is dealt with algebraically. An operation
as simple as rotating a digital photograph through a few degrees, to get the
horizon horizontal, relies on coordinate geometry.

Graphs of the square and square root

 Graph of the sine function

The deeper message of coordinate geometry is about cross-connections in

mathematics. Concepts whose physical realizations seem totally different may be
different aspects of the same thing. Superficial appearances can be misleading.
Much of the effectiveness of mathematics as a way to understand the universe
stems from its ability to adapt ideas, transferring them from one area of science
to another. Mathematics is the ultimate in technology transfer. And it is those
cross-connections, revealed to us over the past 4000 years, that make
mathematics a single, unified subject.

What coordinates did for them

Coordinate geometry can be employed on surfaces more complicated than the plane,
such as the sphere. The commonest coordinates on the sphere are longitude and
latitude. So map-making, and the use of maps in navigation, can be viewed as an
application of coordinate geometry.
The main navigational problem for a captain was to determine the latitude and
longitude of his ship. Latitude is relatively easy, because the angle of the Sun above the
horizon varies with latitude and can be tabulated. Since 1730, the standard instrument
for finding latitude was the sextant (now made almost obsolete by GPS). This was
invented by Newton, but he did not publish it. It was independently rediscovered by the
English mathematician John Hadley and the American inventor Thomas Godfrey.
Previous navigators had used the astrolabe, which goes back to medieval Arabia.
Longitude is trickier. The problem was eventually solved by constructing a highly
accurate clock, which was set to local time at the start of the voyage. The time of sunrise
and sunset, and the movements of the Moon and stars, depend on longitude, making it
possible to determine longitude by comparing local time with that on the clock. The story
of John Harrison’s invention of the chronometer, which solved the problem, is famously
told in Dava Sobel’s Longitude.

Longitude and latitude as coordinates

What coordinates do for us

We continue to use coordinates for maps, but another common use of coordinate
geometry occurs in the stock market, where the fluctuations of some price are recorded
as a curve. Here the x-coordinate is time, and the y-coordinate is the price. Enormous
quantities of financial and scientific data are recorded in the same way.
Stock market data represented in coordinates
 7
Patterns in Numbers

The origins of number theory

Despite becoming ever more fascinated by geometry, mathematicians did not

lose their interest in numbers. But they did start asking deeper questions, and
answered many of them. A few had to wait for more powerful techniques. Some
remain unanswered to this day.

Number theory

There is something fascinating about numbers. Plain, unadorned whole numbers,

1, 2, 3, 4, 5, . . . What could possibly be simpler? But that simple exterior
conceals hidden depths, and many of the most baffling questions in mathematics
are about apparently straightforward properties of whole numbers. This area is
known as number theory, and it turns out to be difficult precisely because its
ingredients are so basic. The very simplicity of whole numbers leaves little
scope for clever techniques.
The earliest serious contributions to number theory – that is, complete with
proofs, not just assertions – are found in the works of Euclid, where the ideas are
thinly disguised as geometry. The subject was developed into a distinct area of
mathematics by the Greek, Diophantus, some of whose writings survive as later
copies. Number theory was given a big boost in the 1600s by Fermat, and
developed by Leonhard Euler, Joseph-Louis Lagrange and Carl Friedrich Gauss
into a deep and extensive branch of mathematics which touched upon many
other areas, often seemingly unrelated. By the end of the 20th century these
connections had been used to answer some – though not all – of the ancient
puzzles, including a very famous conjecture made by Fermat around 1650,
known as his Last Theorem.
For most of its history, number theory has been about the internal workings of
mathematics itself, with few connections to the real world. If ever there was a
branch of mathematical thought that lived in the heady reaches of the ivory
towers, it was number theory. But the advent of the digital computer has
changed all that. Computers work with electronic representations of whole
numbers, and the problems and opportunities raised by computers frequently
lead to number theory. After 2500 years as a purely intellectual exercise, number
theory has finally made an impact on everyday life.

Primes

Anyone who contemplates the multiplication of whole numbers eventually

notices a fundamental distinction.
Many numbers can be broken up into smaller pieces, in the sense that the
number concerned arises by multiplying those pieces together. For instance, 10
is 2 × 5, and 12 is 3 × 4. Some numbers, however, do not break up in this
manner. There is no way to express 11 as the product of two smaller whole
numbers; the same goes for 2, 3, 5, 7 and many others.
The numbers that can be expressed as the product of two smaller numbers are
said to be composite. Those that cannot be so expressed are prime. According to
this definition, the number 1 should be considered prime, but for good reasons it
is placed in a special class of its own and called a unit. So the first few primes
are the numbers

2 3 5 7 11 13 17 19 23 29 31 37 41

As this list suggests, there is no obvious pattern to the primes (except that all
but the first are odd). In fact, they seem to occur somewhat irregularly, and there
is no simple way to predict the next number on the list. Even so, there is no
question that this number is somehow determined – just test successive numbers
until you find the next prime.
Despite, or perhaps because of, their irregular distribution, primes are of vital
importance in mathematics. They form the basic building blocks for all numbers,
in the sense that larger numbers are created by multiplying smaller ones.
Chemistry tells us that any molecule, however complicated, is built from atoms –
chemically indivisible particles of matter. Analogously, mathematics tells us that
any number, however big it may be, is built from primes – indivisible numbers.
So primes are the atoms of number theory.

‘Primes form the basic ‘building blocks’ for all

numbers.’

This feature of primes is useful because many questions in mathematics can

be solved for all whole numbers provided they can be solved for the primes, and
primes have special properties that sometimes make the solution of the question
easier. This dual aspect of the primes – important but ill-behaved – excites the
mathematician’s curiosity.

Euclid

Euclid introduced primes in Book VII of the Elements, and he gave proofs of
three key properties. In modern terminology, these are:

(i) Every number can be expressed as a product of primes.

(ii) This expression is unique except for the order in which the primes occur.
(iii) There are infinitely many primes.
 What Euclid actually stated and proved is slightly different.
Proposition 31, Book VII tells us that any composite number is measured by
some prime – that is, it can be divided exactly by that prime. For example, 30 is
composite, and it is exactly divisible by several primes, among them 5 – in fact,
30 = 6 × 5. By repeating this process of pulling out a prime divisor, or factor, we
can break any number down into a product of primes. Thus, starting from 30 = 5
× 6, we observe that 6 is also composite, with 6 = 2 × 3. Now 30 = 2 × 3 × 5, and
all three factors are prime.
If instead we had started from 30 = 10 × 3, then we would break down 10
instead, as 30 = 10 × 3, leading to 30 = 2 × 5 × 3. The same three primes occur,
but multiplied in a different order – which of course does not affect the result. It
may seem obvious that however we break a number into primes, we always get
the same result except for order, but this turns out to be tricky to prove. In fact,
similar statements in some related systems of numbers turn out to be false, but
for ordinary whole numbers the statement is true. Prime factorization is unique.
Euclid proves the key fact needed to establish uniqueness in Proposition 30,
Book VII of the Elements: if a prime divides the product of two numbers, then it
must divide at least one of those numbers. Once we know Proposition 30, the
uniqueness of prime factorization is a straightforward consequence.

Why Uniqueness of Prime Factors is not Obvious

Since the primes are the atoms of number theory, it might seem obvious that the same atoms
always turn up when a number is broken into primes. After all, atoms are the indivisible pieces. If
you could break a number up in two distinct ways, wouldn’t that involve splitting an atom? But
here the analogy with chemistry is slightly misleading.
To see that uniqueness of prime factorization is not obvious, we can work with a restricted set
of numbers:

1 5 9 13 17 21 25 29

and so on. These are the numbers that are one greater than a multiple of 4. Products of such
numbers also have the same property, so we can build such numbers up by multiplying smaller
numbers of the same type. Define a ‘quasiprime’ to be any number in this list that is not the
product of two smaller numbers in the list. For instance, 9 is quasiprime: the only smaller
numbers in the list are 1 and 5, and their product is not 9. (It is still true that 9 = 3 × 3, of course,
but the number 3 is not in the list.)
 It is obvious – and true – that every number in the list is a product of quasiprimes. However,
although these quasiprimes are the atoms of the set, something rather strange happens. The
number 693 breaks up in two different ways: 693 = 9 × 77 = 21 × 33, and all four factors, 9, 21,
33 and 77, are quasiprime. So uniqueness of factorization fails for this type of number.

‘In modern terms, the list of primes is infinite.’

Proposition 20, Book IX states that: ‘Prime numbers are more than any
assigned multitude of prime numbers.’ In modern terms, this means that the list
of primes is infinite. The proof is given in a representative case: suppose that
there are only three prime numbers, a, b and c. Multiply them together and add
one, to obtain abc + 1. This number must be divisible by some prime, but that
prime cannot be any of the original three, since these divide abc exactly, so they
cannot also divide abc + 1, since they would then divide the difference, which is
1. We have therefore found a new prime, contradicting the assumption that a, b,
c are all the primes there are.
Although Euclid’s proof employs three primes, the same idea works for a
longer list. Multiply all primes in the list, add one and then take some prime
factor of the result; this always generates a prime that is not on the list. Therefore
no finite list of primes can ever be complete.

The Largest Known Prime

There is no largest prime, but the largest known prime as of September 2006 is 232,582,657 − 1,
which has 9,808,358 decimal digits. Numbers of the form 2p − 1, with p prime, are called
Mersenne primes, because Mersenne conjectured in his Cogitata Physica-Mathematica of 1644
that these numbers are prime for p = 2, 3, 5, 7, 13, 17, 19, 31, 67, 127 and 227, and composite for
all other whole numbers up to 257.
There are special high-speed methods for testing such numbers to see if they are primes, and
we now know that Mersenne made five mistakes. His numbers are composite when p = 67 and
257, and there are three more primes with p = 61, 89, 107. Currently 44 Mersenne primes are
known. Finding new ones is a good way to test new supercomputers, but has no practical
significance.
Diophantus

We have mentioned Diophantus of Alexandria in connection with algebraic

notation, but his greatest influence was in number theory. Diophantus studied
general questions, rather than specific numerical ones, although his answers
were specific numbers. For example: ‘Find three numbers such that their sum,
and the sum of any two, is a perfect square.’ His answer is 41, 80 and 320. To
check: the sum of all three is 441 = 212. The sums of pairs are 41 + 80 = 112, 41
+ 320 = 192 and 80 + 320 = 202.

The 3-4-5 right-angled triangle

One of the best known equations solved by Diophantus is a curious offshoot

of Pythagoras’s Theorem. We can state the theorem algebraically: if a right
triangle has sides a, b, c with c being the longest, then a2 + b2 = c2. There are
some special right triangles for which the sides are whole numbers. The simplest
and best known is when a, b, c are 3, 4, 5, respectively; here 33 + 42 = 9 + 16 =
25 = 52. Another example, the next simplest, is 52 + 122 = 132.
In fact, there are infinitely many of these Pythagorean triples. Diophantus
found all possible whole number solutions of what we now write as the equation
a2 + b2 = c2. His recipe is to take any two whole numbers, and form the
difference between their squares, twice their product and the sum of their
squares. These three numbers always form a Pythagorean triple, and all such
triangles arise in this manner provided we also allow all three numbers to be
multiplied by some constant. If the numbers are 1 and 2, for example, we get the
famous 3-4-5 triangle. In particular, since there are infinitely many ways to
choose the two numbers, there exist infinitely many Pythagorean triples.

Fermat

After Diophantus, number theory stagnated for over a thousand years, until it
was taken up by Fermat, who made many important discoveries. One of his most
elegant theorems tells us exactly when a given integer n is a sum of two perfect
squares: n = a2 + b2. The solution is simplest when n is prime. Fermat observed
that there are three basic types of prime:

What We Don’t Know about Prime Numbers

Even today, the primes still have some secrets. Two famous unsolved problems are the Goldbach
Conjecture and the Twin Primes Conjecture.
Christian Goldbach was an amateur mathematician who corresponded regularly with Euler. In a
letter of 1742, he described evidence that every whole number greater than 2 is the sum of three
primes. Goldbach viewed 1 as a prime, which we no longer do; as a consequence, we would now
exclude the numbers 3 = 1 + 1 + 1 and 4 = 2 + 1 +1. Euler proposed a stronger conjecture: that
every even number greater than 2 is the sum of two primes. For example, 4 = 2 + 2, 6 = 3 +3, 8 =
5 + 3, 10 = 5 + 5 and so on. This conjecture implies Goldbach’s. Euler was confident that his
conjecture was true, but could not find a proof, and the conjecture is still open. Computer
experiments have shown it to be true for every even number up 1018. The best known result was
obtained by Chen Jing-Run in 1973 using complicated techniques from analysis. He proved that
every sufficiently large even number is the sum of two primes, or a prime and an almost-prime
(the product of two primes).
The twin prime conjecture is much older, and goes back to Euclid. It states that there are
infinitely many twin primes p and p + 2. Examples of twin primes are 5, 7 and 11, 13. Again, no
proof or disproof is known. In 1966, Chen proved that there are infinitely many primes p, such
that p + 2 is either prime or almost-prime. Currently, the largest known twin primes are
2,003,663,613 × 2195,000 ± 1, found by Eric Vautier, Patrick McKibbon and Dmitri Gribenko in
2007.

(i) The number 2, the only even prime.

 (ii) Primes that are 1 greater than a multiple of 4, such as 5, 13, 17 and so on –
these primes are all odd.
(iii) Primes that are 1 less than a multiple of 4, such as 3, 7, 11 and so on –
these primes are also odd.

He proved that a prime is a sum of two squares if it belongs to categories (i) or

(ii), and it is not a sum of two squares if it belongs to category (iii).
For instance, 37 is in category (ii), being 4 × 9 + 1, and 37 = 62 + 12, a sum of
two squares. In contrast, 31 = 4 × 8 − 1 is in category (iii), and if you try all
possible ways to write 31 as a sum of two squares, you will find that nothing
works. (For instance, 31 = 25 + 6, where 25 is a square, but 6 is not.)
The upshot is that a number is a sum of two squares if and only if every prime
divisor of the form 4k – 1 occurs to an even power. Using similar methods,
Joseph-Louis Lagrange proved in 1770 that every positive integer is a sum of
four perfect squares (including one or more 0s if necessary). Fermat had
previously stated this result, but no proof is recorded.
One of Fermat’s most influential discoveries is also one of the simplest. It is
known as Fermat’s Little Theorem, to avoid confusion with his Last (sometimes
called Great) Theorem, and it states that if p is any prime and a is any whole
number, then ap – a is a multiple of p. The corresponding property is usually
false when p is composite, but not always.

Pierre de Fermat
1601–1665
 Pierre Fermat was born at Beaumont-de-Lomagne in France in 1601, son of the leather merchant
Dominique Fermat and Claire de Long, the daughter of a family of lawyers. By 1629 he had
made important discoveries in geometry and the forerunner of the calculus, but he chose law as a
career, becoming a Councillor at the parliament of Toulouse in 1631. This entitled him to add the
‘de’ to his name. As an outbreak of plague killed off his superiors, he advanced rapidly. In 1648
he became a King’s Councillor in the local parliament of Toulouse, where he served for the rest
of his life, reaching the most senior level of the Criminal Court in 1652.

He never held an academic position, but mathematics was his passion. In 1653 he contracted
plague, was rumoured to be dead, but survived. He carried out extensive correspondence with
other intellectuals, notably the mathematician Pierre de Carcavi and the monk Marin Mersenne.

He worked in mechanics, optics, probability and geometry, and his method for locating the
maximum and minimum values of a function paved the way for calculus. He became one of the
world’s leading mathematicians, but published little of his work, mainly because he was not
willing to spend the time required to bring it into publishable form.
His most long-lasting influence was in number theory, where he challenged other
mathematicians to prove a series of theorems and solve various problems. Among these was the
(misnamed) ‘Pell equation’ nx2 + 1 = y2, and the statement that the sum of two nonzero perfect
cubes cannot be a perfect cube. This is a special case of a more general conjecture, ‘Fermat’s Last
Theorem’, in which cubes are replaced by nth powers for any n ≥ 3.
He died in 1665, just two days after concluding a legal case.

Fermat’s most celebrated result took 350 years to prove. He stated it around
1640, and he claimed a proof, but all we know of his work is a short note.
Fermat owned a copy of Diophantus’s Arithmetica, which inspired many of his
investigations, and he often wrote down his own ideas in the margin. At some
point he must have been thinking about the Pythagorean equation: add two
squares to get a square. He wondered what would happen if instead of squares
you tried cubes, but found no solutions. The same problem arose for fourth, fifth
or higher powers.
In 1670 Fermat’s son Samuel published an edition of Bachet’s translation of
the Arithmetica, which included Fermat’s marginal notes. One such note became
notorious: the statement that if n ≥ 3, the sum of two nth powers is never an nth
power. The marginal note states ‘To resolve a cube into the sum of two cubes, a
fourth power into two fourth powers or, in general, any power higher than the
second into two of the same kind is impossible; of which fact I have found a
remarkable proof. The margin is too small to contain it.’
It seems unlikely that his proof, if it existed, was correct. The first, and
currently only, proof was derived by Andrew Wiles in 1994; it uses advanced
abstract methods that did not exist until the late 20th century.
After Fermat, several major mathematicians worked in number theory,
notably Euler and Lagrange. Most of the theorems that Fermat had stated but not
proved were polished off during this period.

Gauss

The next big advance in number theory was made by Gauss, who published his
masterpiece, the Disquisitiones Arithmeticae (Investigations in Arithmetic) in
1801. This book propelled the theory of numbers to the centre of the
mathematical stage. From then on, number theory was a core component of the
mathematical mainstream. Gauss mainly focused on his own, new, work, but he
also set up the foundations of number theory and systematized the ideas of his
predecessors.
The most important of these foundational changes was a very simple but
powerful idea: modular arithmetic. Gauss discovered a new type of number
system, analogous to integers but differing in one key respect: some particular
number, known as the modulus, was identified with the number zero. This
curious idea turned out to be fundamental to our understanding of divisibility
properties of ordinary integers.
Here is Gauss’s idea. Given an integer m, say that a and b are congruent to the
modulus m, denoted

a ≡ b (mod m)

if the difference a − b is exactly divisible by m. Then arithmetic to the modulus

m works exactly the same as ordinary arithmetic, except that we may replace m
by 0 anywhere in the calculation. So, any multiple of m can be ignored.
The phrase ‘clock arithmetic’ is often used to capture the spirit of Gauss’s
idea. On a clock, the number 12 is effectively the same as 0 because the hours
repeat after 12 steps (24 in continental Europe and military activities). Seven
hours after 6 o’clock is not 13 o’clock, but 1 o’clock, and in Gauss’s system 13 ≡
1(mod 12). So modular arithmetic is like a clock that takes m hours to go full
circle. Not surprisingly, modular arithmetic crops up whenever mathematicians
look at things that change in repetitive cycles.
The Disquisitiones Arithmeticae used modular arithmetic as the basis for
deeper ideas, and we mention three.
The bulk of the book is a far-reaching extension of Fermat’s observations that
primes of the form 4k + 1 are a sum of two squares, whereas those of the form 4k
− 1 are not. Gauss restated this result as a characterization of integers that can be
written in the form x2 + y2, with x and y integers. Then he asked what happens if
instead of this formula we use a general quadratic form, ax2 + bxy + cy2. His
theorems are too technical to discuss, but he obtained an almost complete
understanding of this question.
Another topic is the law of quadratic reciprocity, which intrigued and
perplexed Gauss for many years. The starting point is a simple question: what do
perfect squares look like to a given modulus? For instance, suppose that the
modulus is 11. Then the possible perfect squares (of the numbers less than 11)
are

0 1 4 9 16 25 36 49 64 81 100

Carl Friedrich Gauss

1777–1855
Gauss was highly precocious, allegedly correcting his father’s arithmetic when aged three. In
1792, with financial assistance from the Duke of Brunswick-Wolfenbüttel, Gauss went to
Brunswick’s Collegium Carolinum. There he made several major mathematical discoveries,
including the law of quadratic reciprocity and the prime number theorem, but did not prove them.
From 1795–98 he studied at Göttingen, where he discovered how to construct a regular 17-sided
polygon with ruler and compass. His Disquisitiones Arithmeticae, the most important work in
number theory to date, was published 1801.

Gauss’s public reputation, however, rested on an astronomical prediction. In 1801 Giuseppe

Piazzi discovered the first asteroid: Ceres. The observations were so sparse that astronomers were
worried that they might not find it again when it reappeared from behind the Sun. Several
astronomers predicted where it would reappear; so did Gauss. Only Gauss was right. In fact,
Gauss had used a method of his own invention, now called the ‘method of least squares’, to
derive accurate results from limited observations. He did not reveal this technique at the time, but
it has since become fundamental in statistics and observational science.
In 1805 Gauss married Johanna Ostoff, whom he loved dearly, and in 1807 he left Brunswick
to become director of the Göttingen observatory. In 1808 his father died, and Johanna died in
1809 after giving birth to their second son. Soon after, the son died too.
Despite these personal tragedies, Gauss continued his research, and in 1809 he published his
Theoria Motus Corporum Coelestium in Sectionibus Conicis Solem Ambientium, a major
contribution to celestial mechanics. He married again, to Minna, a close friend of Johanna’s, but
the marriage was more out of convenience than love.

Around 1816 Gauss wrote a review of deductions of the parallel axiom from the other axioms of
 Euclid, in which he hinted at an opinion he had probably held since 1800, the possibility of a
logically consistent geometry that differed from Euclid’s.
In 1818 he was placed in charge of a geodetic survey of Hanover, making serious contributions
to the methods employed in surveying. In 1831, after the death of Minna, Gauss began working
with the physicist Wilhelm Weber on the Earth’s magnetic field.
They discovered what are now called Kirchhoff’s laws for electrical circuits, and constructed a
crude but effective telegraph. When Weber was forced to leave Göttingen in 1837, Gauss’s
scientific work went into decline, though he remained interested in the work of others, notably
Ferdinand Eisenstein and Georg Bernhard Riemann. He died peacefully in his sleep.

which, when reduced (mod 11), yield

0 1 3 4 5 9

with each nonzero number appearing twice. These numbers are the quadratic
residues, mod 11.
The key to this question is to look at prime numbers. If p and q are primes,
when is q a square (mod p)? Gauss discovered that while there is no simple way
to answer that question directly, it bears a remarkable relation to another
question: when is p a square (mod q)? For example, the list of quadratic residues
above shows that q = 5 is a square modulo p ≡ 11. It is also true that 11 is a
square modulo 5 – because 11 ≡ 1 (mod 5) and 1 ≡ 12. So here both questions
have the same answer.
Gauss proved that this law of reciprocity holds for any pair of odd primes,
except when both primes are of the form 4k – 1, in which case the two

What number theory did for them

One of the earliest practical applications of number theory occurs in gears. If two cogs
are placed together so that their teeth mesh, and one cog has m teeth, the other n teeth,
then the movement of the cogs is related to these numbers. For instance, suppose one
cog has 30 teeth and the other has 7. If I turn the big cog exactly once, what does the
smaller one do? It returns to the initial position after 7, 14, 21 and 28 steps. So, the final
2 steps, to make 30, advance it by just 2 steps. This number turns up because it is the
remainder on dividing 30 by 7. So the motion of cogs is a mechanical representation of
division with remainder, and this is the basis of modular arithmetic.
 Cogwheels were used by ancient Greek craftsmen to design a remarkable device, the
Antikythera mechanism. In 1900 the sponge diver Elias Stadiatis found a formless lump
of corroded rock in a wreck of 65 BC near the island of Antikythera, some 40 metres (140
feet) down. In 1902 the archaeologist Valerios Stais noticed that the rock contained a
gear wheel, and was actually the remains of a complicated bronze mechanism. It was
inscribed with words in the Greek alphabet.
The mechanism’s function has been worked out from its structure and its inscriptions,
and it turns out to be an astronomical calculator. There are more than 30 gear wheels –
the latest reconstruction, in 2006, suggests there were originally 37. The numbers of
cogs correspond to important astronomical ratios. In particular, two cogs have 53 teeth –
a difficult number to manufacture – and this number comes from the rate at which the
Moon’s furthest point from the Earth rotates. All the prime factors of the numbers of teeth
are based on two classical astronomical cycles, the Metonic and Saros cycles. X-ray
analysis has revealed new inscriptions and made them readable, and it is now certain
that the device was used to predict the positions of the Sun, the Moon and probably the
then-known planets. The inscriptions date to around 150–100 BC.
The Antikythera mechanism is of a sophisticated design, and appears to incorporate
Hipparchus’s theory of the motion of the Moon. It may well have been built by one of his
students, or at least with their aid. It was probably an executive toy for a royal
personage, rather than a practical instrument, which may explain its exquisite design
and manufacture.

The Antikythera mechanism and a reconstruction

 The Antikythera mechanism and a reconstruction

questions always have opposite answers. That is: for any odd primes p and q,

q is a square (mod p) if and only if p is a square (mod q),

unless both p and q are of the form 4k – 1, in which case

q is a square (mod p) if and only if p is not a square (mod q).

Initially Gauss was unaware that this was not a new observation: Euler had
noticed the same pattern. But unlike Euler, Gauss managed to prove that it is
always true. The proof was very difficult, and it took Gauss several years to fill
one small but crucial gap.
A third topic in the Disquisitiones is the discovery that had convinced Gauss
to become a mathematician at the age of 19: a geometric construction for the
regular 17-gon (a polygon with 17 sides). Euclid provided constructions, using
ruler and compass, for regular polygons with three, five and 15 sides; he also
knew that those numbers could be repeatedly doubled by bisecting angles,
yielding regular polygons with four, six, eight and 10 sides, and so on. But
Euclid gave no constructions for 7-sided polygons, 9-sided ones, or indeed any
other numbers than the ones just listed. For some two thousand years, the
mathematical world assumed that Euclid had said the last word, and no other
regular polygons were constructable. Gauss proved them wrong.
It is easy to see that the main problem is constructing regular p-gons when p is
prime. Gauss pointed out that such a construction is equivalent to solving the
algebraic equation

xp–1 + xp–2 + xp–3 + . . . + x2 + x + 1 = 0

‘The world assumed that Euclid had said the last

word. Gauss proved them wrong.’

Now, a ruler-and-compass construction can be viewed, thanks to coordinate

geometry, as a sequence of quadratic equations. If a construction of this kind
exists, it follows (not entirely trivially) that p – 1 must be a power of 2.
The Greek cases p = and 5 satisfy this condition: here p – 1 = 2, respectively.
But they are not the only such primes. For instance 17 – 1 = 16 is a power of 2.
This does not yet prove that the 17-gon is constructable, but it provides a strong
hint, and Gauss managed to find an explicit reduction of his 16th degree
equation to a series of quadratics. He stated, but did not prove, that a
construction is possible whenever p – 1 is a power of 2 (still requiring p to be
prime), and it is impossible for all other primes. The proof was soon completed
by others.
These special primes are called Fermat primes, because they were studied by
Fermat. He observed that if p is a prime and p – 1 = 2k, then k must itself be a
power of 2. He noted the first few Fermat primes: 2, 3, 5, 17, 257, 65,537. He
conjectured that numbers of the form 22m + 1 are always prime, but this was
wrong. Euler discovered that when m = 5 there is a factor 641.

Marie-Sophie Germain
1776–1831
Sophie Germain was the daughter of the silk merchant Ambroise-François Germain and Marie-
Madelaine Gruguelin. At age 13 she read of the death of Archimedes, killed by a Roman soldier
while contemplating a geometric diagram in the sand, and was inspired to become a
mathematician. Despite her parents’ well-meaning efforts to deter her – mathematics was not then
considered a suitable vocation for a young lady – she read the works of Newton and Euler,
wrapped in a blanket while her mother and father were sleeping. When her parents became
convinced of her commitment to mathematics, they relented and started to help her, giving her
financial support throughout her life.
She obtained lecture notes from the École Polytechnique, and wrote to Lagrange with some
original work of her own under the name ‘Monsieur LeBlanc’. Lagrange, impressed, eventually
discovered that the writer was a woman, and had the good sense to encourage her and become her
sponsor. The two worked together, and some of her results were included in a later edition of
Legendre’s 1798 Essai sur le Théorie des Nombres.

Her most celebrated correspondent was Gauss. Sophie studied the Disquisitiones Arithmeticae,
and from 1804 to 1809 she wrote a number of letters to its author, again concealing her gender by
using the name LeBlanc. Gauss praised LeBlanc’s work in letters to other mathematicians. In
1806 he discovered that LeBlanc was actually female, when the French occupied Braunschweig.
Concerned that Gauss might suffer the same fate as Archimedes, Sophie contacted a family friend
who was a senior officer in the French Army, General Pernety. Gauss learned of this, and
discovered that LeBlanc was actually Sophie.
Sophie need not have worried. Gauss was even more impressed, and wrote to her: ‘But how to
describe to you my admiration and astonishment at seeing my esteemed correspondent Monsieur
Le Blanc metamorphose himself into this illustrious personage . . . When a person of the sex
 which, according to our customs and prejudices, must encounter infinitely more difficulties than
men to familiarize herself with these thorny researches, succeeds nevertheless in surmounting
these obstacles and penetrating the most obscure parts of them, then without doubt she must have
the noblest courage, quite extraordinary talents and superior genius.’

Sophie obtained some results on Fermat’s Last Theorem, the best available until 1840. Between
1810 and 1820 she worked on the vibrations of surfaces, a problem posed by the Institut de
France. In particular, an explanation was sought for ‘Chladni patterns’ – symmetric patterns that
appear if sand is sprinkled on a metal plate, which is then caused to vibrate using a violin bow.
On her third attempt she was awarded a gold medal, although for unknown reasons, possibly a
protest about the unfair treatment of women scientists, she did not appear at the award ceremony.
In 1829 she developed breast cancer, but continued to work on number theory and the
curvature of surfaces until her death two years later.

It follows that there must also exist ruler-and-compass constructions for the
regular 257-gon and 65,537-gon. F.J. Richelot constructed the regular 257-gon
in 1832, and his work is correct. J. Hermes spent ten years working on the
65,537-gon, and completed his construction in 1894. Recent studies suggest
there are mistakes.
Number theory started to become mathematically interesting with the work of
Fermat, who spotted many of the important patterns concealed in the strange and
puzzling behaviour of whole numbers. His annoying tendency not to supply
proofs was put right by Euler, Lagrange and a few less prominent figures, with
the sole exception of his Last Theorem, but number theory seemed to consist of
isolated theorems – often deep and difficult, but not very closely connected to
each other.

What number theory does for us

Number theory forms the basis of many important security codes used for Internet
commerce. The best known such code is the RSA (Ronald Rivest, Adi Shamir and
Leonard Adleman) cryptosystem, which has the surprising feature that the method for
putting messages into code can be made public without giving away the reverse
procedure of decoding the message.
Suppose Alice wants to send a secret message to Bob. Before doing this, they agree
on two large primes p and q (having at least a hundred digits) and multiply them together
 to get M = pq. They can make this number public if they wish. They also compute K =
(p−1)(q−1), but keep this secret.
Now Alice represents her message as a number x in the range 0 to M (or a series of
such numbers if it’s a long message). To encode the message she chooses some
number a, which has no factors in common with K, and computes y = − xa(mod M). The
number a must be known to Bob, and can also be made public.
To decode messages, Bob has to know a number b such that ab ≡ 1mod K. This
number (which exists and is unique) is kept secret. To decode y, Bob computes

yb (mod M).

Why does this decode? Because

yb ≡ (xa)b ≡ xab ≡ x1 ≡ x (mod M),

using a generalization of Fermat’s Little Theorem due to Euler.

This method is practical because there are efficient tests to find large primes.
However, there are no known methods for finding the prime factors of a large number
efficiently. So telling people the product pq does not help them find p and q, and without
those, they cannot work out the value of b, needed to decode the message.

All that changed when Gauss got in on the act and devised general conceptual
foundations for number theory, such as modular arithmetic. He also related
number theory to geometry with his work on regular polygons. From that
moment, number theory became a major strand in the tapestry of mathematics.
Gauss’s insights led to the recognition of new kinds of structure in
mathematics – new number systems, such as the integers mod n, and new
operations, such as the composition of quadratic forms. With hindsight, the
number theory of the late 18th and early 19th centuries led to the abstract algebra
of the late 19th and 20th centuries. Mathematicians were starting to enlarge the
range of concepts and structures that were acceptable objects of study. Despite
its specialized subject matter, the Disquisitiones Arithmeticae marks a significant
milestone in the development of the modern approach to the whole of
mathematics. This is one of the reasons why Gauss is rated so highly by
mathematicians.

‘It often takes time for a good mathematical idea to

 acquire practical importance. . . sometimes hundreds
of years.’

Until the late 20th century, number theory remained a branch of pure
mathematics – interesting in its own right, and because of its numerous
applications within mathematics itself, but of little real significance to the
outside world. All that changed with the invention of digital communications in
the late 20th century. Since communication then depended on numbers, it is
hardly a surprise that number theory came to the fore in those areas of
application. It often takes time for a good mathematical idea to acquire practical
importance – sometimes hundreds of years – but eventually most topics that
mathematicians find significant for their own sake turn out to be valuable in the
real world too.
 8
The System of the World

The invention of calculus

The most significant single advance in the history of mathematics was calculus,
invented independently around 1680 by Isaac Newton and Gottfried Leibniz.
Leibniz published first, but Newton – egged on by over-patriotic friends –
claimed priority and portrayed Leibniz as a plagiarist. The row soured relations
between English mathematicians and those of continental Europe for a century,
and the English were the main losers.

The system of the world

Even though Leibniz probably deserves priority, Newton turned calculus into a
central technique of the budding subject of mathematical physics, humanity’s
most effective known route to the understanding of the natural world. Newton
called his theory ‘The System of the World’. This may not have been terribly
modest, but it was a pretty fair description. Before Newton, human
understanding of patterns in nature consisted mainly of the ideas of Galileo
about moving bodies, in particular the parabolic trajectory of an object such as a
cannonball, and Kepler’s discovery that Mars follows an ellipse through the
heavens. After Newton, mathematical patterns governed almost everything in the
physical world: the movement of terrestrial and heavenly bodies, the flow of air
and water, the transmission of heat, light and sound, and the force of gravity.
Curiously, though, Newton’s main publication on the mathematical laws of
nature, his Principia Mathematica, does not mention calculus at all; instead, it
relies on the clever application of geometry in the style of the ancient Greeks.
But appearances are deceptive: unpublished documents known as the
Portsmouth Papers show that when he was working on the Principia, Newton
already had the main ideas of calculus. It is likely that Newton used the methods
of calculus to make many of his discoveries, but chose not to present them that
way. His version of calculus was published after his death in the Method of
Fluxions of 1732.

Calculus

What is calculus? The methods of Newton and Leibniz are more easily
understood if we preview the main ideas. Calculus is the mathematics of
instantaneous rates of change – how rapidly is some particular quantity changing
at this very instant? For a physical example: a train is moving along a track: how
fast is it going right now? Calculus has two main branches. Differential calculus
provides methods for calculating rates of change, and it has many geometric
applications, in particular finding tangents to curves. Integral calculus does the
opposite: given the rate of change of some quantity, it specifies the quantity
itself. Geometric applications of integral calculus include the computation of
areas and volumes. Perhaps the most significant discovery is this unexpected
connection between two apparently unrelated classical geometric questions:
finding tangents to a curve and finding areas.
Calculus is about functions: procedures that take some general number and
calculate an associated number. The procedure is usually specified by a formula,
assigning to a given number x (possibly in some specific range) an associated
number f(x). Examples include the square root function f(x) = (which requires
x to be positive) and the square function f(x) = x2 (where there is no restriction on
x).
 The first key idea of calculus is differentiation, which obtains the derivative of
a function. The derivative is the rate at which f(x) is changing, compared to how
x is changing – the rate of change of f(x) with respect to x.
Geometrically, the rate of change is the slope of the tangent to the graph of f at
the value x. It can be approximated by finding the slope of the secant – a line
that cuts the graph of f at two nearby points, corresponding to x and x + h,
respectively, where h is small. The slope of the secant is

Geometry of approximations to the derivative

Now suppose that h becomes very small. Then the secant approaches the
tangent to the graph at x. So in some sense the required slope – the derivative of f
at x – is the limit of this expression as h becomes arbitrarily small.
Let’s try this calculation with a simple example, f(x) = x2. Now

As h becomes very, very small, the slope 2x + h gets closer and closer to 2x.
So the derivative of f is the function g for which g(x) = 2x.
 The main conceptual issue here is to define what we mean by limit. It took
more than a century to find a logical definition.
The other key idea in calculus is that of integration. This is most easily
viewed as the reverse process to differentiation. Thus the integral of g, written

∫g(x)dx

is whichever function f(x) has derivative g(x). For instance, because the
derivative of f(x) = x2 is g(x) = 2x, the integral of g(x) = 2x is f(x) = x2. In
symbols,

∫ 2x dx = x2

The need for Calculus

Inspiration for the invention of calculus came from two directions. Within pure
mathematics, differential calculus evolved from methods for finding tangents to
curves, and integral calculus evolved from methods for calculating the areas of
plane shapes and the volumes of solids. But the main stimulus towards calculus
came from physics – the growing realization that nature has patterns. For reasons
we still do not really understand, many of the fundamental patterns in nature
involve rates of change. So they make sense, and can be discovered, only
through calculus.
Prior to the Renaissance, the most accurate model of the motion of the Sun,
Moon and planets was that of Ptolemy. In his system, the Earth was fixed, and
everything else – in particular, the Sun – revolved around it on a series of (real
or imaginary, depending on taste) circles. The circles originated as spheres in the
work of the Greek astronomer Hipparchus; his spheres spun about gigantic axles,
some of which were attached to other spheres and moved along with them. This
kind of compound motion seemed necessary to model the complex motions of
the planets. Recall that some planets, such as Mercury, Venus and Mars, seemed
to travel along complicated paths that included loops. Others – Jupiter and
Saturn were the only other planets known at that time – behaved more sedately,
but even these bodies exhibited strange irregularities, known since the time of
the Babylonians.
We have already met Ptolemy’s system, known as epicycles, which replaced
the spheres by circles, but retained the compound motion. Hipparchus’s model
was not terribly accurate, compared to observations, but Ptolemy’s model fitted
the observations very accurately indeed, and for over a thousand years it was
seen as the last word on the topic. His writings, translated into Arabic as the
Almagest, were used by astronomers in many cultures.

‘Hipparchus’s model was not terribly accurate. . .’

God v science

Even the Almagest, however, failed to agree with all planetary movements.
Moreover, it was rather complicated. Around the year 1000, a few Arabian and
European thinkers began to wonder whether the daily motion of the Sun might
be explained by a rotating Earth, and some of them also played with the idea that
the Earth revolves round the Sun. But little came of these speculations at the
time.
In Renaissance Europe, however, the scientific attitude began to take root, and
one of the first casualties was religious dogma. At that time, the Roman Catholic
Church exerted substantial control over its adherents’ view of the universe. It
wasn’t just that the existence of the universe, and its daily unfolding, were
credited to the Christian God. The point was that the nature of the universe was
believed to correspond to a very literal reading of the Bible. The Earth was
therefore seen as the centre of all things, the solid ground around which the
heavens revolved. And human beings were the pinnacle of creation, the reason
for the universe’s existence.
No scientific observation can ever disprove the existence of some invisible,
unknowable creator. But observations can – and did – debunk the view of the
Earth as the centre of the universe. And this caused a huge fuss, and got a lot of
innocent people killed, sometimes in hideously cruel ways.

Johannes Kepler
1571–1630
Kepler was the son of a mercenary and an innkeeper’s daughter. As a child, he lived with his
mother in his grandfather’s inn after the death of his father, probably in a war between the
Netherlands and the Holy Roman Empire. He was mathematically precocious, and in 1589 he
studied astronomy under Michael Maestlin at the University of Tübingen. Here he came to grips
with the Ptolemaic system. Most astronomers of the period were more concerned with calculating
orbits than in asking how the planets really moved, but from the beginning Kepler was interested
in the precise paths followed by the planets, rather than the proposed system of epicycles. He
became acquainted with the Copernican system, and was quickly convinced that it was literally
true, and not just a mathematical trick.

Work with Brahe In 1596 he made his first attempt to find patterns in the movements of the
planets, via his Mysterium Cosmographicum (Mystery of the Cosmos), with its strange model
based on regular solids. This model did not agree well with observations, so Kepler wrote to a
leading observational astronomer, Tycho Brahe. Kepler became Brahe’s mathematical assistant,
and was set to work on the orbit of Mars. After Brahe’s death, he continued to work on the
problem. Brahe had left a wealth of data, and Kepler struggled to fit a sensible orbit to them. The
surviving calculations occupy nearly 1000 pages, which Kepler referred to as ‘my war with
Mars’. His final orbit was so precise that the only difference from modern data arises from
minute drifting of the orbit over the intervening centuries.

Difficult times 1611 was a bad year. Kepler’s son died aged seven. Next, his wife died. Then
Emperor Rudolf, who tolerated Protestants, abdicated, and Kepler was forced to leave Prague. In
1613 Kepler remarried, and a problem that occurred to him during their wedding celebrations led
to the writing of his New Stereometry of Wine Barrels of 1615.
 In 1619 he published Harmonices Mundi (The Harmony of the World), a sequel to his Mystery
of the Cosmos. The book contained a wealth of new mathematics, including tiling patterns and
polyhedra. It also formulated the third law of planetary motion. While he was writing the book,
his mother was accused of being a witch. With the aid of the law faculty at Tübingen, she was
eventually freed, partly because the prosecutors had not followed the correct legal procedures for
torture.

Copernicus

The fat hit the fire in 1543, when the Polish scholar Nicholas Copernicus
published an astonishing, original and somewhat heretical book: On the
Revolutions of the Heavenly Spheres. Like Ptolemy, he used epicycles for
accuracy. Unlike Ptolemy, he placed the Sun at the centre, while everything else,
including the Earth, but excluding the Moon, turned around the Sun. The Moon
alone revolved around the Earth.
Copernicus’s main reason for this radical proposal was pragmatic: it replaced
Ptolemy’s 77 epicycles by a mere 34. Among the epicycles envisaged by
Ptolemy, there were many repetitions of a particular circle: circles with that
specific size, and speed of rotation, kept appearing, associated with many
distinct bodies. Copernicus realized that if all these epicycles were transferred to
the Earth, only one of them would be needed. We now interpret this in terms of
the motion of the planets relative to the Earth. If we mistakenly assume the Earth
is fixed, as it seems to be to a naive observer, then the motion of the Earth round
the Sun becomes transferred to all of the planets as an additional epicycle.
Another advantage of Copernicus’s theory was that it treated all the planets in
exactly the same manner. Ptolemy needed different mechanisms to explain the
inner planets and the outer ones. Now, the only difference was that the inner
planets were closer to the Sun than the Earth was, while the outer planets were
further away. It all made excellent sense – but on the whole it was rejected, for a
variety of reasons, not all of them religious.
Copernicus’s theory was complicated, unfamiliar and his book was difficult to
read. Tycho Brahe, one of the best astronomical observers of the period, found
discrepancies between Copernicus’s heliocentric theory and some subtle
observations, which also disagreed with Ptolemy’s theory; he tried to find a
better compromise.

Kepler

When Brahe died, his papers were inherited by Kepler, who spent years
analysing the observations, looking for patterns. Kepler was something of a
mystic, in the Pythagorean tradition, and he tended to impose artificial patterns
on observational data. The most famous of these abortive attempts to find
regularities in the heavens was his beautiful but utterly misguided explanation of
the spacing of the planets in terms of the regular solids. In his day, the known
planets were six in number: Mercury, Venus, Earth, Mars, Jupiter and Saturn.
Kepler wondered whether their distances from the Sun had a geometric pattern.
Moreover, he wondered why there were six planets. He noticed that six planets
leave room for five intervening shapes, and since there were exactly five regular
solids, this would explain the limit of six planets. He came up with a series of six
spheres, each one bearing the orbit of a planet around its equator. Between the
spheres, nestling tightly outside one sphere and inside the next, he placed the
five solids, in the order

‘Kepler wondered why there were six planets.’

Mercury Octahedron
Venus Icosahedron
Earth Dodecahedron
Mars Tetrahedron
Jupiter Cube
Saturn
 Kepler’s theory of the spacing of planetary orbits

The numbers fitted reasonably well, especially given the limited accuracy of
observations at that time. But there are 120 different ways to rearrange the five
solids, which between them give an awful lot of different spacings. It is hardly
surprising that one of these was in reasonably close agreement with reality. The
later discovery of more planets knocked this particular piece of pattern-seeking
firmly on the head, consigning it to the waste bin of history.
Along the way, though, Kepler discovered some patterns that we still
recognize as genuine, now called Kepler’s Laws of Planetary Motion. He
extracted them, over some twenty years of calculation, from Brahe’s
observations of Mars. The laws state:

(i) Planets move round the Sun in elliptical orbits.

(ii) Planets sweep out equal areas in equal times.
(iii) The square of the period of revolution of any planet is proportional to the
 cube of its average distance from the Sun.

The most unorthodox feature of Kepler’s work is that he discarded the

classical circle (allegedly the most perfect shape possible) in favour of the
ellipse. He did so with some reluctance, saying himself that he only settled on
the ellipse when everything else had been eliminated. There is no particular
reason to expect these three laws to bear any closer relation to reality than the
hypothetical arrangement of regular solids, but as it happened, the three laws
were of real scientific significance.

Planet moves over given time interval

Galileo

Another major figure of the period was Galileo Galilei, who discovered
mathematical regularities in the movement of a pendulum and in falling bodies.
In 1589, as professor of mathematics at the University of Pisa, he carried out
experiments on bodies rolling down an inclined plane, but did not publish his
results. It was at this time that he realized the importance of controlled
experiments in the study of natural phenomena, an idea that is now fundamental
to all science. He took up astronomy, making a series of fundamental
discoveries, which eventually led him to espouse the Copernican theory of the
Sun as the centre of the solar system. This set him on a collision course with the
Church, and he was eventually tried for heresy and placed under house arrest.
During the last few years of his life, his health failing, he wrote Discourses
and Mathematical Demonstrations Concerning the Two New Sciences,
explaining his work on the motion of bodies on inclined planes. He stated that
the distance an initially stationary body moves under uniform acceleration is
proportional to the square of the time. This law is the basis of his earlier
discovery that a projectile follows a parabolic path. Together with Kepler’s laws
of planetary motion, it brought into being a new subject: mechanics, the
mathematical study of moving bodies.
That’s the physical astronomical background that led up to calculus. Next,
we’ll look at the mathematical background.

The invention of calculus

The invention of calculus was the outcome of a series of earlier investigations of

what seem to be unrelated problems, but which possess a hidden unity. These
included calculating the instantaneous velocity of a moving object from the
distance it has travelled at any given time, finding the tangent to a curve, finding
the length of a curve, finding the maximum and minimum values of a variable
quantity, finding the area of some shape in the plane and the volume of some
solid in space. Some important ideas and examples were developed by Fermat,
Descartes and the more obscure Englishman, Isaac Barrow, but the methods
remained special to particular problems. A general method was needed.

Galileo Galilei
1564–1642

Galileo was the son of Vincenzo Galilei, a music teacher who had performed experiments with
strings to validate his musical theories. At the age of ten Galileo went to a monastery at
Vallombrosa to be educated, with a view to becoming a doctor. But Galileo wasn’t really
interested in medicine, and spent his time doing mathematics and natural philosophy – what we
now call science.
In 1589 Galileo became professor of mathematics at the University of Pisa. In 1591 he took up
a better-paid position in Padua, where he taught Euclidean geometry and astronomy to medical
students. At that time doctors made use of astrology in their treatment of patients, so these topics
were a necessary part of the curriculum.
Learning about the invention of the telescope, Galileo built one for himself and became so
proficient that he gave his methods to the Venetian Senate, granting them sole rights in their use
in return for a salary increase. In 1609 Galileo observed the heavens, making discovery after
discovery: four of Jupiter’s moons, individual stars within the Milky Way, mountains on the
Moon. He presented a telescope to Cosimo de Medici, Grand Duke of Tuscany, and soon became
the Duke’s chief mathematician.

He discovered the existence of sunspots and published this observation in 1612. By now his
astronomical discoveries had convinced him of the truth of Copernicus’s heliocentric theory, and
in 1616 he made his views explicit in a letter to the Grand Duchess Christina, saying that the
Copernican theory represents physical reality and is not just a convenient way to simplify
calculations.
At this time, Pope Paul V ordered the Inquisition to decide on the truth or falsity of the
heliocentric theory, and they declared it false. Galileo was instructed not to advocate the theory,
but a new pope was elected, Urban VIII, who seemed more relaxed about the issue, so Galileo did
not take the prohibition very seriously. In 1623 he published Il Saggiatore (The Assayer),
dedicating it to Urban. In it, he made a famous statement that the universe ‘is written in the
language of mathematics, and its characters are triangles, circles and other geometric figures,
without which it is humanly impossible to understand a single word of it’.

In 1630 Galileo asked permission to publish another book, Dialogue Concerning the Two Chief
Systems of the World, about the geocentric and heliocentric theories. In 1632, when permission
arrived from Florence (but not Rome), he went ahead. The book claimed to prove that the Earth
moves, the main evidence being the tides. In fact Galileo’s theory of tides was completely wrong,
but the Church authorities saw the book as theological dynamite and the Inquisition banned it,
summoning Galileo to Rome to be tried for heresy. He was found guilty, but escaped with a
sentence of life imprisonment, in the form of house arrest. In this he fared better than many other
heretics, for whom being burnt at the stake was a common punishment. While under house arrest
he wrote the Discourses, explaining his work on moving bodies to the outside world. It was
 smuggled out of Italy and published in Holland.

Leibniz

The first real breakthrough was made by Gottfried Wilhelm Leibniz, a lawyer by
profession, who devoted much of his life to mathematics, logic, philosophy,
history and many branches of science. Around 1673 he began work on the
classical problem of finding the tangent to a curve, and noticed that this was in
effect the inverse problem to that of finding areas and volumes. The latter boiled
down to finding a curve given its tangents; the former problem was exactly the
reverse.
Leibniz used this connection to define what, in effect, were integrals, using
the abbreviation, omn (an abbreviation of omnia, the Latin word for ‘all’). Thus
we find, in his manuscript papers, formulas such as

By 1675 he had replaced omn by the symbol still used today, which is an
old-fashioned elongated letter s, standing for sum. He worked in terms of small
increments dx and dy to the quantities x and y, and used their ratio dy/dx to
determine the rate of change of y as a function of x. Essentially, if f is a function
then Leibniz wrote

dy = f(x + dx) – f(x)

so that

which is the usual secant approximation to the slope of the tangent.

Leibniz recognized that this notation has its problems. If dy and dx are
nonzero then dy/dx is not the instantaneous rate of change of y, but an
approximation. He tried to circumvent this problem by assuming dx and dy to be
infinitesimally small. An infinitesimal is a nonzero number that is smaller than
any other nonzero number. Unfortunately, it is easy to see that no such number
can exist (half an infinitesimal is also nonzero, but smaller) so this approach
does little more than displace the problem elsewhere.
By 1676 Leibniz knew how to integrate and differentiate any power of x,
writing the formula

dxn = nxn−1dx

which we would now write as

In 1677 he derived rules for differentiating the sum, product and quotient of
two functions, and by 1680 he had obtained the formula for the length of an arc
of a curve, and the volume of a solid of revolution, as integrals of various related
quantities.
Although we know these facts, and the associated dates, from his unpublished
notes, he first published his ideas on calculus rather later, in 1684. Jakob and
Johann Bernoulli found this paper rather obscure, describing it as ‘an enigma
rather than an explanation’. With hindsight, we see that by that time Leibniz had
discovered a significant part of basic calculus, with applications to complicated
curves like the cycloid, and a sound grasp of concepts such as curvature.
Unfortunately, his writings were fragmented and virtually unreadable.

Newton

The other creator of calculus was Isaac Newton. Two of his friends, Barrow and
Edmond Halley, came to recognize his remarkable abilities, and encouraged him
to publish his work. Newton disliked being criticized, and when in 1672 he
published his ideas about light, his work provoked a storm of criticism, which
reinforced his reluctance to commit his thoughts to print. Nevertheless, he
continued to publish sporadically, and wrote two books. In private he continued
developing his ideas about gravity, and in 1684 Halley tried to convince Newton
to publish the work. But aside from Newton’s general misgivings about
criticism, there was a technical obstacle. He had been forced to model planets as
point particles, with nonzero mass but zero size, which he felt was unrealistic
and would invite criticism. He wanted to replace these unrealistic points by solid
spheres, but he could not prove that the gravitational attraction of a sphere is the
same as that of a point particle of the same mass.

‘The Principia saw the light of day in 1687.’

In 1686 he succeeded in filling the gap, and the Principia saw the light of day
in 1687. It contained many novel ideas. The most important were mathematical
laws of motion, extending the work of Galileo, and gravity, based on the laws
found by Kepler.
Newton’s main law of motion (there are some subsidiary ones) states that the
acceleration of a moving body, multiplied by its mass, is equal to the force that
acts on the body. Now velocity is the derivative of position, and acceleration is
the derivative of velocity. So even to state Newton’s Law we need the second
derivative of position with respect to time, nowadays written

Newton wrote two dots over the top of x instead (ẍ).

The law of gravity states that all particles of matter attract each other with a
force that is proportional to their masses, and inversely proportional to the
square of the distance between them. So, for example, the force attracting the
Earth to the Moon would become one quarter as great if the Moon were removed
to twice its distance, or one ninth as great if its distance were tripled. Again,
because this law is about forces, it involves the second derivative of position.
Newton deduced this law from Kepler’s three laws of planetary motion. The
published deduction was a masterpiece of classical Euclidean geometry. Newton
chose this style of presentation because it involved familiar mathematics, and so
could not easily be criticized. But many aspects of the Principia owed their
genesis to Newton’s unpublished invention of calculus.
Among his earliest work on the topic was a paper titled On Analysis by Means
of Equations with an Infinite Number of Terms, which he circulated to a few
friends in 1669. In modern terminology, he asked what the equation of a
function f(x) is, if the area under its graph is of the form xm. (Actually he asked
something slightly more general, but let’s keep it simple.) He deduced, to his
own satisfaction, that the answer is f(x) = mxm−1
 Isaac Newton
1642–1727

Newton lived on a farm in the tiny village of Woolsthorpe, in Lincolnshire. His father had died
two months before he was born, and his mother managed the farm. He was educated in very
ordinary local schools, and exhibited no special talent of any kind, except for a facility with
mechanical toys. He once made a hot air balloon and tested it out with the family cat as pilot; the
balloon and the cat were never seen again. He went to Trinity College at Cambridge University,
having done reasonably well in most of his examinations – except geometry. As an
undergraduate, he made no great impact.

The Plague
Then, in 1665, the great plague began to devastate London and the surrounding area, and the
students were sent home before the same thing happened in Cambridge. Back in the family
farmhouse, Newton began thinking much more deeply about scientific and mathematical matters.

Gravity
During 1665–66 he devised his law of gravity to explain planetary motion, developed the laws of
mechanics to explain and analyse any kind of moving body or particle, invented both differential
and integral calculus, and made major advances in optics. Characteristically, he published none of
this work, returning quietly to Trinity to take his master’s degree and being elected a fellow of the
college. Then he secured the position of Lucasian Professor of Mathematics, when the incumbent,
 Barrow, resigned in 1669. He gave very ordinary lectures, rather badly, and very few
undergraduates went to them.

Newton’s approach to calculating derivatives was much like that of Leibniz,

except that he used o in place of dx, so his method suffers from the same logical
problem: it seems to be only approximate. But Newton could show that by
assuming o to be very small, the approximation would become ever better. In the
limit, when o becomes as small as we please, the error vanishes. So, Newton
maintained, his end result was exact. He introduced a new word, fluxion, to
capture the main idea – that of a quantity flowing towards zero but not actually
getting there.
In 1671 he wrote a more extensive treatment, the Method of Fluxions and
Infinite Series. The first book on calculus was not published until 1711; the
second appeared in 1736. It is clear that by 1671 Newton possessed most of the
basic ideas of calculus.
Objectors to this procedure, notably Bishop George Berkeley in his 1734 book
The Analyst, a Discourse Addressed to an Infidel Mathematician, pointed out
that it is illogical to divide numerator and denominator by o when later o is set to
0. In effect, the procedure conceals the fact that the fraction is actually 0/0, which
is well known to be meaningless. Newton responded that he was not actually
setting o equal to 0; he was working out what happened when o became as close
as we wish to 0 without actually getting there. The method was about fluxions,
not numbers.
The mathematicians sought refuge in physical analogies – Leibniz referred to
the ‘spirit of finesse’ as opposed to the ‘spirit of logic’ – but Berkeley was
perfectly correct. It took more than a century to find a good answer to his
objections, by defining the intuitive notion of ‘passing to a limit’ in a rigorous
manner. Calculus then turned into a more subtle subject, analysis. But for a
century after the invention of calculus, nobody except Berkeley worried much
about its logical foundations, and calculus flourished despite this flaw.
It flourished because Newton was right, but it would take nearly 200 years
before his concept of a fluxion was formulated in a logically acceptable way, in
terms of limits. Fortunately for mathematics, progress was not halted until a
decent logical foundation was discovered. Calculus was too useful, and too
important, to be held up over a few logical quibbles. Berkeley was incensed,
maintaining that the method seemed to work only because various errors
cancelled each other out. He was right – but he failed to enquire why they always
cancelled out. Because if that were the case, they weren’t really errors at all!

The definite integral

Associated with differentiation is the reverse process, integration. The integral

of f(x), written, f(x)dx, is whichever function yields f(x) when it is differentiated.
Geometrically it represents the area under the graph of the function f. The
definite integral f(x)dx is the area under this graph between the values x = a and
x = b.
Derivatives and integrals solved problems that had taxed the ingenuity of
previous mathematicians. Velocities, tangents, maxima and minima could all be
found using differentiation. Lengths, areas and volumes could be calculated by
integration. But there was more. Surprisingly, it seemed that the patterns of
nature were written in the language of the calculus.

The English get left behind

As the importance of the calculus became ever clearer, greater prestige became
attached to its creator. But who was the creator?
We have seen that Newton began thinking about calculus in 1665, but did not
publish anything on the topic until 1687. Leibniz, whose ideas ran along roughly
similar lines to Newton’s, had started working on calculus in 1673, and
published his first papers on the topic in 1684. The two worked independently,
but Leibniz might have learned about Newton’s work when he visited Paris in
1672 and London in 1673; Newton had sent a copy of On Analysis to Barrow in
1669, and Leibniz talked to several people who also knew Barrow and so might
have known about this work.
When Leibniz published his work in 1684, some of Newton’s friends took
umbrage – probably because Newton had been pipped to the publication post
and they all belatedly realized what was at stake – and accused Leibniz of
stealing Newton’s ideas. The continental mathematicians, especially the
Bernoullis, leaped to Leibniz’s defence, suggesting that it was Newton, not
Leibniz, who was guilty of plagiarism. In point of fact, both men had made their
discoveries pretty much independently, as their unpublished manuscripts show;
to muddy the waters, both had leaned heavily on previous work of Barrow, who
probably had better grounds for grievance than either of them.

What calculus did for them

An early use of calculus to understand natural phenomena was the question of the
shape of a hanging chain. The question was controversial; some mathematicians
thought the answer was a parabola, others disagreed. In 1691 Leibniz, Christiaan
Huygens and Joahnn Bernoulli all published proposed solutions. The clearest was
Bernoulli’s. He wrote down a differential equation to describe the position of the chain,
based on Newtonian mechanics and Newton’s laws of motion.
The solution, it turned out, was not a parabola, but a curve known as a catenary, with
equation

y = k(ex + e−x)

for constant k.
 Hanging chain forms a catenary

The cables of suspension bridges, however, are parabolic. The difference arises
because these cables carry the weight of the bridge, as well as their own weight. Again,
this can be demonstrated using calculus.

The Clifton suspension bridge – a parabola

The accusations could easily have been withdrawn, but instead the dispute
grew more heated; John Bernoulli extended his distaste from Newton to the
entire English nation. The end result was a disaster for English mathematics,
because the English doggedly stuck with Newton’s geometric style of thinking,
which was difficult to use, whereas the continental analysts employed Leibniz’s
more formal, algebraic methods, and pushed the subject ahead at a rapid pace.
Most of the payoff in mathematical physics therefore went to the French,
Germans, Swiss and Dutch, while English mathematics languished in a
backwater.

The differential equation

The most important single idea to emerge from the flurry of work on calculus
was the existence, and the utility, of a novel kind of equation – the differential
equation. Algebraic equations relate various powers of an unknown number.
Differential equations are much grander: they relate various derivatives of an
unknown function.
Newton’s laws of motion tell us that if y(t) is the height of a particle moving
under gravity near the Earth’s surface, then the second derivative d2y/dt2 is
proportional to the force g that acts; specifically,

where m is the mass. This equation does not specify the function y directly.
Instead, it specifies a property of its second derivative. We must solve the
differential equation to find y itself. Two successive integrations yield the
solution

where b is the initial height of the particle and a is its initial velocity. The
formula tells us that the graph of height y against time t is an upside-down
parabola. This is Galileo’s observation.
 Parabolic trajectory of a projectile

What calculus does for us

Differential equations abound in science: they are by far the commonest way to model
natural systems. To choose one application at random, they are used routinely to
calculate the trajectories of space probes, such as the Mariner mission to Mars, the two
Pioneer craft that explored the solar system and gave us such wonderful images of
Jupiter, Saturn, Uranus and Neptune, and the recent Mars Rovers Spirit and
Opportunity, six-wheeled robot vehicles that explored the Red Planet.
The Cassini mission, currently exploring Saturn and its moons, is another example.
Among its discoveries is the existence of lakes of liquid methane and ethane on Saturn’s
moon Titan. Of course, calculus is not the only technique used by space missions – but
without it, these missions would literally never have got off the ground.
More practically, every aircraft that flies, every car that travels the road and every
suspension bridge and earthquake-proof building owes its design in part to calculus.
Even our understanding of how animal populations change size over time stems from
differential equations. The same goes for the spread of epidemics, where calculus
models are used to plan the most effective way to intervene and prevent disease
spreading. A recent model of the foot and mouth disease epidemic in the UK has shown
that the strategy adopted at the time was not the best available.
 The Mars Rover Spirit (artist’s impression, NASA)

The pioneering efforts of Copernicus, Kepler, Galileo and other Renaissance

scientists led to the discovery of mathematical patterns in the natural world.
Some apparent patterns turned out to be spurious, and were discarded; others
provided very accurate models of nature, and were retained and developed. From
these early beginnings, the notion that we live in a ‘clockwork universe’, running
according to rigid, unbreakable rules, emerged, despite serious religious
opposition, mainly from the Church of Rome.
Newton’s great discovery was that nature’s patterns seem to manifest
themselves not as regularities in certain quantities, but as relations among their
derivatives. The laws of nature are written in the language of calculus; what
matters are not the values of physical variables, but the rates at which they
change. It was a profound insight, and it created a revolution, leading more or
less directly to modern science, and changing our planet forever.
 9
Patterns in Nature

Formulating laws of physics

The main message in Newton’s Principia was not the specific laws of nature
that he discovered and used, but the idea that such laws exist – together with
evidence that the way to model nature’s laws mathematically is with differential
equations. While England’s mathematicians engaged in sterile vituperation over
Leibniz’s alleged (and totally fictitious) theft of Newton’s ideas about calculus,
the continental mathematicians were cashing in on Newton’s great insight,
making important inroads into celestial mechanics, elasticity, fluid dynamics,
heat, light and sound – the core topics of mathematical physics. Many of the
equations that they derived remain in use to this day, despite – or perhaps
because of – the many advances in the physical sciences.

Differential equations

To begin with, mathematicians concentrated on finding explicit formulas for

solutions of particular kinds of ordinary differential equation. In a way this was
unfortunate, because formulas of this type usually fail to exist, so attention
became focused on equations that could be solved by a formula rather than
equations that genuinely described nature. A good example is the differential
equation for a pendulum, which takes the form

‘. . .attention became focused on equations that could

be solved by a formula. . .’

for a suitable constant k, where t is time and θ is the angle at which the
pendulum hangs, with θ = 0 being vertically downwards. There is no solution of
this equation in terms of classical functions (polynomial, exponential,
trigonometric, logarithmic and so on). There does exist a solution using elliptic
functions, invented more than a century later. However, if it is assumed that the
angle is small, so we are considering a pendulum making small oscillations, then
sin θ is approximately equal to θ, and the smaller θ becomes the better this
approximation is. So the differential equation can be replaced by

and now there is a formula for the solution, in general,

θ = A sin kt + B cos kt

for constants A and B, determined by the initial position and angular velocity of
the pendulum.
This approach has some advantages: for instance, we can quickly deduce that
the period of the pendulum – the time taken to complete one swing – is 2π/k. The
main disadvantage is that the solution fails when θ becomes sufficiently large
(and here even 20° is large if we want an accurate answer). There is also a
question of rigour: is it the case that an exact solution to an approximate
equation yields an approximate solution to the exact equation? Here the answer
is yes, but this was not proved until about 1900.
The second equation can be solved explicitly because it is linear – it involves
only the first power of the unknown θ and its derivative, and the coefficients are
constant. The prototype function for all linear differential equations is the
exponential y = ex. This satisfies the equation

That is, ex is its own derivative. This property is one reason why the number e is
natural. A consequence is that the derivative of the natural logarithm log x is 1/x,
so the integral of 1/x is log x. Any linear differential equation with constant
coefficients can be solved using exponentials and trigonometric functions (which
we will see are really exponentials in disguise).

Types of differential equation

There are two types of differential equation. An ordinary differential equation

(ODE) refers to an unknown function y of a single variable x, and relates various
derivatives of y, such as dy/dx and d2y/dx2. The differential equations described
so far have been ordinary ones. Far more difficult, but central to mathematical
physics, is the concept of a partial differential equation (PDE). Such an equation
refers to an unknown function y of two or more variables, such as f(x, y, t) where
x and y are coordinates in the plane and t is time. The PDE relates this function
to expressions in its partial derivatives with respect to each of the variables. A
new notation is used to represent derivatives of some variables with respect to
others, while the remainder are held fixed. Thus, ∂x/∂t indicates the rate of
change of x with respect to time, while y is held constant. This is called a partial
derivative – hence the term partial differential equation.
Euler introduced PDEs in 1734 and d’Alembert did some work on them in
1743, but these early investigations were isolated and special. The first big
breakthrough came in 1746, when d’Alembert returned to an old problem, the
vibrating violin string. John Bernoulli had discussed a finite element version of
this question in 1727, considering the vibrations of a finite number of point
masses spaced equally far apart along a weightless string. D’Alembert treated a
continuous string, of uniform density, by applying Bernoulli’s calculations to n
masses, and then letting n tend to infinity. Thus, a continuous string was in effect
thought of as infinitely many infinitesimal segments of string, connected
together.
Starting from Bernoulli’s results, which were based on Newton’s law of
motion, and making some simplifications (for example, that the size of the
vibration is small), d’Alembert was led to the PDE

where y = y(x,t) is the shape of the string at time t, as a function of the horizontal
coordinate x. Here a is a constant related to the tension in the string and its
density. By an ingenious argument, d’Alembert proved that the general solution
of his PDE has the form

y(x, t) = f(x + at) + f(x − at)

where f is periodic, with period twice the length of the string, and f is an odd
function – that is, f(−z) = −f(z). This form satisfies the natural boundary
condition that the ends of the string do not move.

Wave equation

We now call d’Alembert’s PDE the wave equation, and interpret its solution as a
superposition of symmetrically placed waves, one moving with velocity a and
the other velocity –a (that is, travelling in the opposite direction). It has become
one of the most important equations in mathematical physics, because waves
arise in many different circumstances.
Euler spotted d’Alembert’s paper, and immediately tried to improve on it. In
1753 he showed that without the boundary conditions, the general solution is

y(x, t) = f(x + at) + g(x − at)

where f and g are periodic, but satisfy no other conditions. In particular, these
functions can have different formulas in different ranges of x, a feature that Euler
referred to as discontinuous functions, though in today’s terminology they are
continuous but have discontinuous first derivatives.
 Successive snapshots of a wave travelling from left to right

In an earlier paper published in 1749 he pointed out that (for simplicity we

will take the length of the string to be 1 unit) the simplest odd periodic functions
are trigonometric functions
f(x) = sin x, sin 2x, sin 3x, sin 4x, . . .

Vibrational modes of a string

and so on. These functions represent pure sinusoidal vibrations of frequencies 1,

2, 3, 4 and so on. The general solution, said Euler, is a superposition of such
curves. The basic sine curve sin x is the fundamental mode of vibration, and the
others are higher modes – what we now call harmonics.
Comparison of Euler’s solution of the wave equation with d’Alembert’s led to
a foundational crisis. D’Alembert did not recognize the possibility of
discontinuous functions in Euler’s sense. Moreover, there seemed to be a
fundamental flaw in Euler’s work, because trigonometric functions are
continuous, and so are all (finite) superpositions of them. Euler had not
committed himself on the issue of finite superpositions versus infinite ones – in
those days no one was really very rigorous about such matters, having yet to
learn the hard way that it matters. Now the failure to make such a distinction was
causing serious trouble. The controversy simmered until later work by Fourier
caused it to boil over.

‘The ancient Greeks knew that a vibrating string can

produce many different musical notes. . .’

Music, light, sound and electromagnetism

The ancient Greeks knew that a vibrating string can produce many different
musical notes, depending on the position of the nodes, or rest points. For the
fundamental frequency, only the end points are at rest. If the string has a node at
its centre, then it produces a note one octave higher; and the more nodes there
are, the higher the frequency of the note will be. The higher vibrations are called
overtones.
The vibrations of a violin string are standing waves – the shape of the string at
any instant is the same, except that it is stretched or compressed in the direction
at right angles to its length. The maximum amount of stretching is the amplitude
of the wave, which physically determines how loud the note sounds. The
waveforms shown are sinusoidal in shape; and their amplitudes vary sinusoidally
with time.
In 1759 Euler extended these ideas from strings to drums. Again he derived a
wave equation, describing how the displacement of the drumhead in the vertical
direction varies over time. Its physical interpretation is that the acceleration of a
small piece of the drumhead is proportional to the average tension exerted on it
by all nearby parts of the drumhead. Drums differ from violin strings not only in
their dimensionality – a drum is a flat two-dimensional membrane – but in
having a much more interesting boundary. In this whole subject, boundaries are
absolutely crucial. The boundary of a drum can be any closed curve, and the key
condition is that the boundary of the drum is fixed. The rest of the drumhead can
move, but its rim is firmly strapped down.
The mathematicians of the 18th century were able to solve the equations for
the motion of drums of various shapes. Again they found that all vibrations can
be built up from simpler ones, and that these yield a specific list of frequencies.
The simplest case is the rectangular drum, the simplest vibrations of which are
combinations of sinusoidal ripples in the two perpendicular directions. A more
difficult case is the circular drum, which leads to new functions called Bessel
functions. The amplitudes of these waves still vary sinusoidally with time, but
their spatial structure is more complicated.
The wave equation is exceedingly important. Waves arise not only in musical
instruments, but in the physics of light and sound. Euler found a three-
dimensional version of the wave equation, which he applied to sound waves.
Roughly a century later, James Clerk Maxwell extracted the same mathematical
expression from his equations for electromagnetism, and predicted the existence
of radio waves.
 Vibrations of a circular drumhead, and of a real guitar

Gravitational attraction

Another major application of PDEs arose in the theory of gravitational attraction,

otherwise known as potential theory. The motivating problem was the
gravitational attraction of the Earth, or of any other planet. Newton had modelled
planets as perfect spheres, but their true form is closer to that of an ellipsoid.
And whereas the gravitational attraction of a sphere is the same as that of a point
particle (for distances outside the sphere), the same does not hold for ellipsoids.
Colin Maclaurin made significant headway on these issues in a prizewinning
memoir of 1740 and a subsequent book Treatise of Fluxions published in 1742.
His first step was to prove that if a fluid of uniform density spins at a uniform
speed, under the influence of its own gravity, then the equilibrium shape is an
oblate spheroid – an ellipsoid of revolution. He then studied the attractive forces
generated by such a spheroid, with limited success. His main result was that if
two spheroids have the same foci, and if a particle lies either on the equatorial
plane or the axis of revolution, then the force exerted on it by either spheroid is
proportional to their masses.
In 1743 Clairaut continued working on this problem in his Théorie de la
Figure de la Terre. But the big breakthrough was made by Legendre. He proved
a basic property of not just spheroids, but any solid of revolution: if you know its
gravitational attraction at every point along its axis, then you can deduce the
attraction at any other point. His method was to express the attraction as an
integral in spherical polar coordinates. Manipulating this integral, he expressed
its value as a superposition of spherical harmonics. These are determined by
special functions now called Legendre polynomials. In 1784, he pursued this
topic, proving many basic properties of these polynomials.
 The fundamental PDE for potential theory is Laplace’s equation, which
appears in Laplace’s five-volume Traité de Mécanique Céleste (Treatise on
Celestial Mechanics) published from 1799 onwards. It was known to earlier
researchers, but Laplace’s treatment was definitive. The equation takes the form

An ellipsoid

where V(x, y, z) is the potential at a point (x, y, z) in space. Intuitively, it says that
the value of the potential at any given point is the average of its values over a
tiny sphere surrounding that point. The equation is valid outside the body: inside,
it must be modified to what is now known as Poisson’s equation.

Heat and temperature

Successes with sound and gravitation encouraged mathematicians to turn their

attention to other physical phenomena. One of the most significant was heat. At
the start of the 19th century the science of heat flow was becoming a highly
practical topic, mainly because of the needs of the metalworking industry, but
also because of a growing interest in the structure of the Earth’s interior, and in
particular the temperature inside the planet. There is no direct way to measure
the temperature a thousand miles or more below the Earth’s surface, so the only
available methods were indirect, and an understanding of how heat flowed
through bodies of various compositions was essential.
In 1807 Joseph Fourier submitted a paper on heat flow to the French Academy
of Sciences, but the referees rejected it because it was insufficiently developed.
To encourage Fourier to continue the work, the Academy made heat flow the
subject of its 1812 grand prize. The prize topic was announced well ahead of
time, and by 1811 Fourier had revised his ideas, put them in for the prize, and
won. However, his work was widely criticized for its lack of logical rigour and
the Academy refused to publish it as a memoir. Fourier, irritated by this lack of
appreciation, wrote his own book, Théorie Analytique de la Chaleur (Analytic
Theory of Heat), published in 1822. Much of the 1811 paper was included
unchanged, but there was extra material too. In 1824 Fourier got even: he was
made Secretary of the Academy, and promptly published his 1811 paper as a
memoir.
Fourier’s first step was to derive a PDE for heat flow. With various
simplifying assumptions – the body must be homogeneous (with the same
properties everywhere) and isotropic (no direction within it should behave
differently from any other), and so on. He came up with what we now call the
heat equation, which describes how the temperature at any point in a three-
dimensional body varies with time. The heat equation is very similar in form to
Laplace’s equation and the wave equation, but the partial derivative with respect
to time is of first order, not second. This tiny change makes a huge difference to
the mathematics of the PDE.

‘In 1824 Fourier got even: he was made Secretary of

the Academy.’

There were similar equations for bodies in one and two dimensions (rods and
sheets) obtained by removing the terms in z (for two dimensions) and then y (for
one). Fourier solved the heat equation for a rod (whose length we take to be π),
the ends of which are maintained at a fixed temperature, by assuming that at
time t = 0 (the initial condition) the temperature at a point x on the rod is of the
form

b1 sin x + b2 sin 2x + b3 sin 3x + . . .

(an expression suggested by preliminary calculations) and deduced that the

temperature must then be given by a similar but more complicated expression in
which each term is multiplied by a suitable exponential function. The analogy
with harmonics in the wave equation is striking. But there each mode given by a
pure sine function oscillates indefinitely without losing amplitude, whereas here
each sinusoidal mode of the temperature distribution decays exponentially with
time, and the higher modes decay more rapidly.
The physical reason for the difference is that in the wave equation energy is
conserved, so the vibrations cannot die down. But in the heat equation, the
temperature diffuses throughout the rod, and is lost at the ends because these are
kept cool.
The upshot of Fourier’s work is that whenever we can expand the initial
temperature distribution in a Fourier series – a series of sine and cosine
functions like the one above – then we can immediately read off how the heat
flows through the body as time passes. Fourier considered it obvious that any
initial distribution of temperature could be so expressed, and this is where the
trouble began, because a few of his contemporaries had been worrying about
precisely this issue for some time, in connection with waves, and had convinced
themselves that it was much harder than it seemed.

How Fourier Series Work

Fourier expansion of a square wave: above, the component sine curves; below, their sum

A typical discontinuous function is the square wave S(x), which takes the values 1 when −π < x ≤
0 and –1 when 0 < x ≤ π and has period 2π. Applying Fourier’s formula to the square wave we get
the series
 The terms add up, as shown in the diagram below.
Although the square wave is discontinuous, every approximation is continuous. However, the
wiggles build up as more and more terms are added, making the graph of the Fourier series
become increasingly steep near the discontinuities. This is how an infinite series of continuous
functions can develop a discontinuity.

Fourier’s argument for the existence of an expansion in sines and cosines was
complicated, confused and wildly non-rigorous. He went all round the
mathematical houses to derive, eventually, a simple expression for the
coefficients b1, b2, b3, etc. Writing f(x) for the initial temperature distribution, his
result was

Euler had already written down this formula in 1777, in the context of the
wave equation for sound, and he proved it using the clever observation that
distinct modes, sin mπx and sin mπx, are orthogonal, meaning that

is zero whenever m and n are distinct integers, but non-zero – in fact, equal to
π/2 – when m = n. If we assume that f(x) has a Fourier expansion, multiply both
sides by sin nx and integrate, then every term except one disappears, and the
remaining terms yield Fourier’s formula for bn.

Fluid dynamics

No discussion of the PDEs of mathematical physics would be complete without

mentioning fluid dynamics. Indeed, this is an area of enormous practical
significance, because these equations describe the flow of water past submarines,
of air past aircraft, and even the flow of air past Formula 1 racing cars.
Euler started the subject in 1757 by deriving a PDE for the flow of a fluid of
zero viscosity, that is zero ‘stickiness’. This equation remains realistic for some
fluids, but it is too simple to be of much practical use. Equations for a viscous
fluid were derived by Claude Navier in 1821, and again by Poisson in 1829.
They involve various partial derivatives of the fluid velocity. In 1845 George
Gabriel Stokes deduced the same equations from more basic physical principles,
and they are therefore known as the Navier–Stokes equations.

What differential equations did for them

Kepler’s model of elliptical orbits is not exact. It would be if there were only two bodies in
the solar system, but when a third body is present, it changes (perturbs) the elliptical
orbit. Because the planets are spaced quite a long way apart, this problem affects the
detailed motion, and most orbits remain close to ellipses. However, Jupiter and Saturn
behave strangely, sometimes lagging behind where they ought to be and sometimes
pulling ahead. This effect is caused by their mutual gravitation, together with that of the
Sun.
Newton’s law of gravitation applies to any number of bodies, but the calculations
become very difficult when there are three bodies or more. In 1748, 1750 and 1752 the
French Academy of Sciences offered prizes for accurate calculations of the movements
of Jupiter and Saturn. In 1748 Euler used differential equations to study how Jupiter’s
gravity perturbs the orbit of Saturn, and won the prize. He tried again in 1752 but his
work contained significant mistakes. However, the underlying ideas later turned out to be
useful.

Jupiter and Saturn in a composite picture

 Jupiter and Saturn in a composite picture

Sofia Vasilyevna Kovalevskaya

1850–1891

Sofia Kovalevskaya was the daughter of an artillery general and a member of the Russian
nobility. It so happened that her nursery walls had been papered with pages from lecture notes on
analysis. At the age of 11 she took a close look at her wallpaper and taught herself calculus. She
became attracted to mathematics, preferring it to all other areas of study. Her father tried to stop
her, but she carried on regardless, reading an algebra book when her parents were sleeping.
In order to travel and obtain an education, she was obliged to marry, but the marriage was
never a success. In 1869 she studied mathematics in Heidelberg, but because female students
were not permitted, she had to persuade the university to let her attend lectures on an unofficial
basis. She showed impressive mathematical talent, and in 1871 she went to Berlin, where she
studied under the great analyst Karl Weierstrass. Again, she was not permitted to be an official
student, but Weierstrass gave her private lessons.

She carried out original research, and by 1874 Weierstrass said that her work was suitable for a
doctorate. She had written three papers, on PDEs, elliptic functions and the rings of Saturn. In the
same year Göttingen University awarded her a doctoral degree. The PDE paper was published in
1875.
In 1878 she had a daughter, but returned to mathematics in 1880, working on the refraction of
light. In 1883 her husband, from whom she had separated, committed suicide, and she spent more
 and more time working on her mathematics to assuage her feelings of guilt. She obtained a
university position in Stockholm, giving lectures in 1884. In 1889 she became the third female
professor ever at a European university, after Maria Agnesi (who never took up her post) and the
physicist Laura Bassi. Here she did research on the motion of a rigid body, entered it for a prize
offered by the Academy of Sciences in 1886, and won. The jury found the work so brilliant that
they increased the prize money. Subsequent work on the same topic was awarded a prize by the
Swedish Academy of Sciences, and led to her being elected to the Imperial Academy of Sciences.

Ordinary differential equations

We close this section with two far-reaching contributions to the use of ODEs
(ordinary differential equations) in mechanics. In 1788 Lagrange published his
Mécanique Analytique (Analytical Mechanics), proudly pointing out that

‘One will not find figures in this work. The methods that I expound require neither
constructions, nor geometrical or mechanical arguments, but only algebraic
operations, subject to a regular and uniform course.’

At that period, the pitfalls of pictorial arguments had become apparent, and
Lagrange was determined to avoid them. Pictures are now back in vogue, though
supported by solid logic, but Lagrange’s insistence on a formal treatment of
mechanics inspired a new unification of the subject, in terms of generalized
coordinates. Any system can be described using many different variables. For a
pendulum, for instance, the usual coordinate is the angle at which the pendulum
is hanging, but the horizontal distance between the bob and the vertical would do
equally well.
The equations of motion look very different in different coordinate systems,
and Lagrange felt this was inelegant. He found a way to rewrite the equations of
motion in a form that looks the same in every coordinate system. The first
innovation is to pair off the coordinates: to every position coordinate q (such as
the angle of the pendulum) there is associated the corresponding velocity
coordinate, (the rate of angular motion of the pendulum). If there are k position
coordinates, there are also k velocity coordinates. Instead of a second-order
differential equation in the positions, Lagrange derived a first-order differential
equation in the positions and the velocities. He formulated this in terms of a
quantity now called the Lagrangian.

Global wind speed and temperature computed from an extended version of the Navier– Stokes equations

Hamilton improved Lagrange’s idea, making it even more elegant. Physically,

he used momentum instead of velocity to define the extra coordinates.
Mathematically, he defined a quantity now called the Hamiltonian, which can be
interpreted – for many systems – as energy. Theoretical work in mechanics
generally uses the Hamiltonian formalism, which has been extended to quantum
mechanics as well.

What differential equations do for us

There is a direct link between the wave equation and radio and television.
Around 1830 Michael Faraday carried out experiments on electricity and magnetism,
investigating the creation of a magnetic field by an electric current, and an electric field
 by a moving magnet. Today’s dynamos and electric motors are direct descendants from
his apparatus. In 1864 James Clerk Maxwell reformulated Faraday’s theories as
mathematical equations for electromagnetism: Maxwell’s equations. These are PDEs
involving the magnetic and electric fields.
A simple deduction from Maxwell’s equations leads to the wave equation. This
calculation shows that electricity and magnetism can travel together like a wave, with the
speed of light. What travels at the speed of light? Light. So light is an electromagnetic
wave. The equation placed no limitations on the frequency of the wave, and light waves
occupy a relatively small range of frequencies, so physicists deduced that there ought to
be electromagnetic waves with other frequencies. Heinrich Hertz demonstrated the
physical existence of such waves, and Guglielmo Marconi turned them into a practical
device: radio. The technology snowballed. Television and radar also rely on
electromagnetic waves. So do GPS satellite navigation, mobile phones and wireless
computer communications.

Radio waves

Physics goes mathematical

Newton’s Principia was impressive, with its revelation of deep mathematical

laws underlying natural phenomena. But what happened next was even more
impressive. Mathematicians tackled the entire panoply of physics – sound, light,
heat, fluid flow, gravitation, electricity, magnetism. In every case, they came up
with differential equations that described the physics, often very accurately.
The long-term implications have been remarkable. Many of the most
important technological advances, such as radio, television and commercial jet
aircraft depend, in numerous ways, on the mathematics of differential equations.
The topic is still the subject of intense research activity, with new applications
emerging almost daily. It is fair to say that Newton’s invention of differential
equations, fleshed out by his successors in the 18th and 19th centuries, is in
many ways responsible for the society in which we now live. This only goes to
show what is lurking just behind the scenes, if you care to look.

‘. . .radio, television and commercial jet aircraft

depend on the mathematics of differential
equations.’
 10
Impossible Quantities

Can negative numbers have square roots?

Mathematicians distinguish several different kinds of number, with different

properties. What really matters is not the individual numbers, but the system to
which they belong – the company they keep.
Four of these number systems are familiar: the natural numbers, 1, 2, 3, . . .;
the integers, which also include zero and negative whole numbers; the rational
numbers, composed of fractions p/q where p and q are integers and q is not zero;
and the real numbers, generally introduced as decimals that can go on forever –
whatever that means – and represent both the rational numbers, as repeating
decimals, and irrational numbers like , e and π whose decimal expansions do
not ever repeat the same block of digits.

Integers
The name integer just means whole; the other names give the impression that the
systems concerned are sensible, reasonable things – natural, rational and of
course real. These names reflect, and encourage, a long-prevailing view that
numbers are features of the world around us.
Many people think that the only way you can do mathematical research is to
invent new numbers. This view is almost always wrong; a lot of mathematics is
not about numbers at all, and in any case the usual aim is to invent new
theorems, not new numbers. Occasionally, however, ‘new numbers’ do arise.
And one such invention, a so-called ‘impossible’ or ‘imaginary’ number,
completely changed the face of mathematics, adding immeasurably to its power.
That number was the square root of minus one. To early mathematicians, such a
description seemed ridiculous, because the square of any number is always
positive. So, negative numbers cannot have square roots.
But just suppose they did. What would happen?
It took mathematicians a long time to appreciate that numbers are artificial
inventions made by human beings; very effective inventions for capturing many
aspects of nature, to be sure, but no more a part of nature than one of Euclid’s
triangles or a formula in calculus. Historically, we first see mathematicians
starting to struggle with this philosophical question when they began to realize
that imaginary numbers were inevitable, useful and somehow on a par with the
more familiar real ones.

‘. . .negative numbers cannot have square roots. But

just suppose they did. . .’

Problems with cubics

Revolutionary mathematical ideas are seldom discovered in the simplest and

(with hindsight) most obvious context. They almost always emerge from
something far more complicated. So it was with the square root of minus one.
Nowadays, we normally introduce this number in terms of the quadratic equation
x2 + 1 = 0, the solution of which is the square root of minus one – whatever that
means. Among the first mathematicians to wonder whether it had a sensible
meaning were the Renaissance algebraists, who ran into square roots of negative
numbers in a surprisingly indirect way: the solution of cubic equations.
 Recall that del Ferro and Tartaglia discovered algebraic solutions to cubic
equations, later written up by Cardano in his Ars Magna. In modern symbols, the
solution of a cubic equation x3 + ax = b is

The Renaissance mathematicians expressed this solution in words, but the

procedure was the same.
Sometimes this formula worked beautifully, but sometimes it ran into trouble.
Cardano noticed that when the formula is applied to the equation x3 = 15x + 4,
with the obvious solution x = 4, the result is expressed as

This expression seemed to have no sensible meaning, however, because –121

has no square root. A puzzled Cardano wrote to Tartaglia, asking for
clarification, but Tartaglia missed the point and his reply was distinctly
unhelpful.
An answer of sorts was provided by Rafael Bombelli in his three-volume book
L’Algebra, printed in Venice in 1572 and Bologna in 1579. Bombelli was
worried that Cardano’s Ars Magna was rather obscure, and he set out to write
something clearer. He operated on the troublesome square root as if it were an
ordinary number, noticing that

and deducing the curious formula

Similarly, Bombelli obtained the formula

Now he could rewrite the sum of the two cube roots as

 So this strange method yielded the right answer, a perfectly normal integer,
but it got there by manipulating ‘impossible’ quantities.
This was all very interesting, but why did it work?

Imaginary numbers

To answer this question, mathematicians had to develop good ways to think

about square roots of negative quantities, and do calculations with them. Early
writers, among them Descartes and Newton, interpreted these imaginary
numbers as a sign that a problem has no solutions. If you wanted to find a
number whose square was minus one, the formal solution, square root of minus
one, was imaginary, so no solution existed. But Bombelli’s calculation implied
that there was more to imaginaries than that. They could be used to find
solutions; they could occur when solutions did exist.
In 1673 John Wallis invented a simple way to represent imaginary numbers as
points in a plane. He started from the familiar representation of real numbers as a
line, with the positive numbers on the right and the negative ones on the left.

The real number line

Then he introduced another line, at right angles to the first, and along this new
line he placed the imaginaries.
 Two copies of the real number line, placed at right angles

This is like Descartes’s algebraic approach to plane geometry, using

coordinate axes. Real numbers form one axis in the picture, imaginaries another.
Wallis did not state the idea in quite this form – his version was closer to
Fermat’s approach to coordinates than Descartes’s. But the underlying point is
the same. The remainder of the plane corresponds to complex numbers, which
consist of two parts: one real, one imaginary. In Cartesian coordinates, we
measure the real part along the real line and measure the imaginary part parallel
to the imaginary line. So 3 +2i lies 3 units to the right of the origin and 2 units
up.
Wallis’s idea solved the problem of giving meaning to imaginary numbers,
but no one took the slightest notice. However, his idea slowly gained ground
subconsciously. Most mathematicians stopped worrying that the square root of
minus one could not occupy any position on the real line, and realized that it
could live somewhere in the wider world of the complex plane. Some failed to
appreciate the idea: in 1758 François Daviet de Foncenex, in a paper about
imaginary numbers, stated that it was pointless to think of imaginaries as
forming a line at right angles to the real line. But others took it to heart and
understood its importance.
 The complex plane according to Wessel

The idea that a complex plane could extend the comfortable real line and give
imaginaries a home was implicit in Wallis’s work, but slightly obscured by the
way he presented it. It was made explicit by the Norwegian Caspar Wessel in
1797. Wessel was a surveyor, and his main interest was to represent the
geometry of the plane in terms of numbers. Working backwards, his ideas could
be viewed as a method for representing complex numbers in terms of planar
geometry. But he published in Danish, and his work went unnoticed until a
century later, when it was translated into French. The French mathematician
Jean-Robert Argand independently published the same representation of
complex numbers in 1806, and Gauss discovered it independently of them both
in 1811.

Complex analysis

If complex numbers had been good only for algebra, they might have remained
an intellectual curiosity, of little interest outside pure mathematics. But as
interest in calculus grew, and it took on a more rigorous form as analysis, people
began to notice that a really interesting fusion of real analysis with complex
numbers – complex analysis – was not only possible, but desirable. Indeed, for
many problems, essential.
This discovery stemmed from early attempts to think about complex
functions. The simplest functions, such as the square or the cube, depended only
on algebraic manipulations, so it was easy to define these functions for complex
numbers. To square a complex number, you just multiply it by itself, the same
process that you would apply to a real number. Square roots of complex numbers
are marginally trickier, but there is a pleasant reward for making the effort: every
complex number has a square root. Indeed, every non-zero complex number has
precisely two square roots, one equal to minus the other. So not only did
augmenting the real numbers with a new number, i, provide –1 with a square
root, it provided square roots for everything in the enlarged system of complex
numbers.

‘Square roots of complex numbers are marginally

trickier.’

What about sines, cosines, the exponential function and the logarithm? At this
stage, things started to get very interesting, but also very puzzling, especially
when it came to logarithms.
Like i itself, logarithms of complex numbers turned up in purely real
problems. In 1702 Johann Bernoulli was investigating the process of integration,
applied to reciprocals of quadratics. He knew a clever technique to carry out this
task whenever the quadratic equation concerned has two real solutions, r and s.
Then we can rewrite the expression to be integrated in terms of ‘partial fractions’

which leads to the integral

A log (x − r) + B log (x − s)

But what if the quadratic has no real roots? How can you integrate the
reciprocal of x2 + 1, for instance? Bernoulli realized that once you have defined
complex algebra, the partial fraction trick still works, but now r and s are
complex numbers. So, for example,

and the integral of this function takes the form

 ½ log (x + i) + ½ log (x – i)

This final step was not fully satisfactory, because it demanded a definition of
the logarithm of a complex number. Was it possible to make sense of such a
statement?
Bernoulli thought it was, and proceeded to use his new idea to excellent
effect. Leibniz also exploited this kind of thinking. But the mathematical details
were not straightforward. By 1712 the two of them were arguing about a very
basic feature of this approach. Forget complex numbers – what was the
logarithm of a negative real number? Bernoulli thought that the logarithm of a
negative real number should be real; Leibniz insisted that it was complex.
Bernoulli had a kind of proof for his claim: assuming the usual formalism of
calculus, the equation

can be integrated to yield

log(−x) = log(x)

However, Leibniz was unconvinced, and believed that the integration was
correct only for positive real x.
This particular controversy was sorted out by Euler in 1749, and Leibniz was
right. Bernoulli, said Euler, had forgotten that any integration involves an
arbitrary constant. What Bernoulli should have deduced was that

log(−x) = log(x) + c

for some constant c. What was this constant? If logarithms of negative (and
complex) numbers are to behave like logarithms of real numbers, which is the
point of the whole game, then it should be true that

log(−x) = log(−1 × x) = log (−1) + log x

so that c = log (−1). Euler then embarked on a series of beautiful calculations

that produced a more explicit form for c. First, he found a way to manipulate
various formulas involving complex numbers, assuming they behaved much like
real numbers, and deduced a relation between trigonometric functions and the
exponential:

eiθ = cos θ + i sin θ

a formula that had been anticipated in 1714 by Roger Cotes. Putting θ = π, Euler
obtained the delightful result that

eiπ = −1

relating the two fundamental mathematical constants e and π. It is remarkable

that any such relation should exist, and even more remarkable that it is so
simple. This formula regularly tops league tables for the ‘most beautiful formula
of all time’.
Taking the logarithm, we immediately deduce that

log (−1) = iπ

revealing the secret of that enigmatic constant c above: it is iπ. As such, it is

imaginary, so Leibniz was right and Bernoulli was wrong.
However, there is more, and it opens Pandora’s box. If we put θ = 2π, then

e2iπ = 1

So log (1) = 2iπ. Then the equation x = x × 1 implies that

log x = log x + 2iπ

from which we conclude that if n is any integer whatever,

log x = log x + 2niπ

At first sight, this makes no sense – it seems to imply that 2niπ = 0 for all n.
But there is a way to interpret it that does make sense. Over the complex
numbers, the logarithmic function is many-valued. Indeed, unless the complex
number z is zero, the function log z can take infinitely many distinct values.
(When z = 0., the value log 0 is not defined.)

What complex numbers did for them

 The real and imaginary parts of a complex function satisfy the Cauchy–Riemann
equations, which are closely related to the PDEs for gravitation, electricity, magnetism
and some types of fluid flow in the plane. This connection made it possible to solve
many equations of mathematical physics – but only for two-dimensional systems.

Magnetic field of a bar magnet, revealed by iron filings: complex analysis can be used to
calculate such fields

Mathematicians were used to functions that could take several distinct values,
the square root being the most obvious example: here, even a real number
possessed two distinct square roots, one positive and the other negative. But
infinitely many values? This was very strange.

Cauchy’s theorem

What really put the cat among the pigeons was the discovery that you could do
calculus – analysis – with complex functions, and that the resulting theory was
elegant and useful. So useful, in fact, that the logical basis of the idea ceased to
be an important issue. When something works, and you feel that you need it, you
generally stop asking whether it makes sense.
The introduction of complex analysis seems to have been a conscious decision
by the mathematical community – a generalization so obvious and compelling
that any mathematician with any kind of sensitivity would want to see what
transpired. In 1811 Gauss wrote a letter to a friend, the astronomer Friedrich
Bessel, revealing his representation of complex numbers as points in a plane; he
also mentioned some deeper results. Among them is a basic theorem upon which
the whole of complex analysis hangs. Today we call it Cauchy’s Theorem,
because it was published by Cauchy, but Gauss had the idea much earlier in his
unpublished writings.

Two distinct paths P and Q from –1 to 1 in the complex plane

Augustin-Louis Cauchy
1789–1857

Augustin-Louis Cauchy was born in Paris during a time of political turmoil. Laplace and
Lagrange were family friends, so Cauchy was exposed to higher mathematics at an early age. He
 went to the École Polytechnique, graduating in 1807. In 1810 he carried out engineering work in
Cherbourg, preparing for Napoleon’s planned invasion of England, but he continued thinking
about mathematics, reading Laplace’s Mécanique Céleste (Celestial Mechanics) and Lagrange’s
Théorie des Fonctions (Theory of Functions).
He continually sought academic positions, without success, but kept working on mathematics.
His famous paper on complex integrals, which effectively founded complex analysis, appeared in
1814, and he finally achieved his goal of an academic post, becoming assistant professor of
analysis at the École Polytechnique a year later. His mathematics now flourished, and a paper on
waves won him the 1816 prize of the Academy of Sciences. He continued to develop complex
analysis, and in his 1829 Leçons sur le Calcul Différentiel he gave the first explicit definition of a
complex function.

After the revolution of 1830 Cauchy briefly went to Switzerland, and in 1831 he became
professor of theoretical physics in Turin. His courses were reported as being highly disorganized.
By 1833 he was in Prague tutoring the grandson of Charles X, but the prince disliked both
mathematics and physics, and Cauchy often lost his temper. He returned to Paris in 1838,
regaining his post at the Academy but did not regain his teaching positions until Louis Philippe
was deposed in 1848. Altogether, he published an astonishing 789 research articles on
mathematics.

This theorem concerns definite integrals of complex functions: that is,

expressions

where a and b are complex numbers. In real analysis this expression can be
evaluated by finding an antiderivative F(z) of f(z), that is, a function F(z) such
that its derivative dF(z)/dz = f(z). Then the definite integral is equal to F(b) –
F(a). In particular, its value depends only on the end points a and b, not on how
you move from one to the other.
Complex analysis, said Gauss, is different. Now the value of the integral may
depend on the path that the variable z takes as it moves from a to b. Because the
complex numbers form a plane, their geometry is richer than that of the real line,
and this is where that extra richness matters.
For example, suppose you integrate f(z) = 1/z from a = −1 to b = 1. If the path
concerned is a semicircle P lying above the real axis, then the integral turns out
to be −πi. But if the path is a semicircle Q lying below the real axis, then the
integral turns out to be πi. The two values are different, and the difference is 2πi.
This difference, said Gauss, occurs because the function 1/z is badly behaved.
It becomes infinite inside the region enclosed by the two paths. Namely, at z = 0,
which here is the centre of the circle formed by the two paths. ‘But if this does
not happen . . . I affirm,’ Gauss wrote to Bessel, ‘that the integral has only one
value even if taken over different paths provided [the function] does not become
infinite in the space enclosed by the two paths. This is a very beautiful theorem,
whose proof I shall give on a convenient occasion’. But he never did.
Instead, the theorem was rediscovered, and published by Augustin-Louis
Cauchy, the true founder of complex analysis. Gauss may have had the ideas, but
ideas are useless if no one gets to see them. Cauchy published his work. In fact,
Cauchy seldom stopped publishing. It is said that the rule, still in force today,
that the journal Comptes Rendus de l’Academie Française accepts papers no
more than four pages long, was introduced explicitly to stop Cauchy filling it
with his huge output. But when the rule was introduced, Cauchy just wrote lots
of short papers. From his prolific pen the main outlines of complex analysis
quickly emerged. And it is a simpler, more elegant and in many ways more
complete theory than real analysis, where the whole idea started.
For instance, in real analysis a function may be differentiable, but its
derivative may not be. It may be differentiable 23 times, but not 24. It may be
differentiable as many times as you wish, but not possess a power series
representation. None of these nasty things can happen in complex analysis. If a
function is differentiable, then it can be differentiated as many times as you
wish; moreover, it has a power series representation. The reason – closely related
to Cauchy’s Theorem and probably a fact used by Gauss in his unknown proof –
is that in order to be differentiable, a complex function must satisfy some very
stringent conditions, known as the Cauchy–Riemann equations. These equations
lead directly to Gauss’s result that the integral between two points may depend
on the path chosen. Equivalently, as Cauchy noted, the integral round a closed
path need not be zero. It is zero provided the function concerned is differentiable
(so in particular is not infinite) at all points inside the path.
There was even a theorem – the residue theorem – that told you the value of
an integral round a closed path, and it depended only on the locations of the
points at which the function became infinite, and its behaviour near those points.
In short, the entire structure of a complex function is determined by its
singularities – the points at which it is badly behaved. And the most important
singularities are its poles, the places where it becomes infinite.
The square root of minus one puzzled mathematicians for centuries. Although
there seemed to be no such number, it kept turning up in calculations. And there
were hints that the concept must make some kind of sense, because it could be
used to obtain perfectly valid results which did not themselves involve taking the
square root of a negative number.
As the successful uses of this impossible quantity continued to grow,
mathematicians began to accept it as a useful device. Its status remained
uncertain until it was realized that there is a logically consistent extension of the
traditional system of real numbers, in which the square root of minus one is a
new kind of quantity – but one that obeys all of the standard laws of arithmetic.

What complex numbers do for us

Today, complex numbers are widely used in physics and engineering. A simple example
occurs in the study of oscillations: motions that repeat periodically. Examples include the
shaking of a building in an earthquake, vibrations in cars and the transmission of
alternating electrical current.
The simplest and most fundamental type of oscillation takes the form a cos ωt,, where
t is time, a is the amplitude of the oscillation ω and is its frequency. It turns out to be
convenient to rewrite this formula as the real part of the complex function eiωt. The use
of complex numbers simplifies calculations because the exponential function is simpler
than the cosine. So engineers studying oscillations prefer to work with complex
exponentials, and revert to the real part only at the end of the calculation.
Complex numbers also determine the stabilities of steady states of dynamical
systems, and are widely used in control theory. This subject is about methods for
stabilizing systems that would otherwise be unstable. An example is the use of
computer-controlled moving control surfaces to stabilize the space shuttle in flight.
Without this application of complex analysis, the space shuttle would fly like a brick.

Geometrically, the real numbers form a line and the complex numbers form a
plane; the real line is one of the two axes of this plane. Algebraically, complex
numbers are just pairs of real numbers with particular formulas for adding the
pairs or multiplying them.
Now accepted as sensible quantities, complex numbers quickly spread
throughout mathematics because they simplified calculations by avoiding the
need to consider positive and negative numbers separately. In this respect they
can be considered as analogous to the earlier invention of negative numbers,
which avoided the need to consider addition and subtraction separately. Today,
complex numbers, and the calculus of complex functions, are routinely used as
an indispensable technique in virtually all branches of science, engineering and
mathematics.
 11
Firm Foundations

Making calculus make sense

By 1800 mathematicians and physicists had developed calculus into an

indispensable tool for the study of the natural world, and the problems that arose
from this connection led to a wealth of new concepts and methods – for
example, ways to solve differential equations – that made calculus one of the
richest and hottest research areas in the whole of mathematics. The beauty and
power of calculus had become undeniable. However, Bishop Berkeley’s
criticisms of its logical basis remained unanswered, and as people began to
tackle more sophisticated topics, the whole edifice started to look decidedly
wobbly. The early cavalier use of infinite series, without regard to their meaning,
produced nonsense as well as insights. The foundations of Fourier analysis were
non-existent, and different mathematicians were claiming proofs of
contradictory theorems. Words like ‘infinitesimal’ were bandied about without
being defined; logical paradoxes abounded; even the meaning of the word
‘function’ was in dispute. Clearly these unsatisfactory circumstances could not
go on indefinitely.
Sorting it all out took a clear head, and a willingness to replace intuition by
precision, even if there was a cost in comprehensibility. The main players were
Bernard Bolzano, Cauchy, Niels Abel, Peter Dirichlet and, above all,
Weierstrass. Thanks to their efforts, by 1900 even the most complicated
manipulations of series, limits, derivatives and integrals could be carried out
safely, accurately and without paradoxes. A new subject was created: analysis.
Calculus became one core aspect of analysis, but more subtle and more basic
concepts, such as continuity and limits, took logical precedence, underpinning
the ideas of calculus. Infinitesimals were banned, completely.

Fourier

Before Fourier stuck his oar in, mathematicians were fairly happy that they knew
what a function was. It was some kind of process, f, which took a number, x, and
produced another number, f(x). Which numbers, x, make sense depends on what
f is. If f(x) = 1/x, for instance, then x has to be non-zero. If f(x) = , and we are
working with real numbers, then x must be positive. But when pressed for a
definition, mathematicians tended to be a little vague.
The source of their difficulties, we now realize, was that they were grappling
with several different features of the function concept – not just what a rule
associating a number x with another number f(x) is, but what properties that rule
possesses: continuity, differentiability, capable of being represented by some
type of formula and so on.
In particular, they were uncertain how to handle discontinuous functions, such
as

f(x) = 0 if x ≤ 0, f(x) = 1 if x > 0

This function suddenly jumps from 0 to 1 as x passes through 0. There was a

prevailing feeling that the obvious reason for the jump was the change in the
formula: from f(x) = 0 to f(x) = 1. Alongside that was the feeling that this is the
only way that jumps can appear; that any single formula automatically avoided
such jumps, so that a small change in x always caused a small change in f(x).
 Another source of difficulty was complex functions, where – as we have seen
– natural functions like the square root are two-valued, and logarithms are
infinitely many valued. Clearly the logarithm must be a function – but when
there are infinitely many values, what is the rule for getting f(z) from z? There
seemed to be infinitely many different rules, all equally valid. In order for these
conceptual difficulties to be resolved, mathematicians had to have their noses
firmly rubbed in them to experience just how messy the real situation was. And
it was Fourier who really got up their noses, with his amazing ideas about
writing any function as an infinite series of sines and cosines, developed in his
study of heat flow.

‘And it was Fourier who really got up their noses. . .’

Fourier’s physical intuition told him that his method should be very general
indeed. Experimentally, you can imagine holding the temperature of a metal bar
at 0 degrees along half of its length, but 10 degrees, or 50, or whatever along the
rest of its length. Physics did not seem to be bothered by discontinuous
functions, whose formulas suddenly changed. Physics did not work with
formulas anyway. We use formulas to model physical reality, but that’s just
technique, it’s how we like to think. Of course the temperature will fuzz out a
little at the junction of these two regions, but mathematical models are always
approximations to physical reality. Fourier’s method of trigonometric series,
applied to a discontinuous function of this kind, seemed to give perfectly
sensible results. Steel bars really did smooth out the temperature distribution the
way his heat equation, solved using trigonometric series, specified. In The
Analytical Theory of Heat he made his position plain: ‘In general, the function
f(x) represents a succession of values or ordinates each of which is arbitrary. We
do not suppose these ordinates to be subject to a common law. They succeed
each other in any manner whatever’.
Bold words; unfortunately, his evidence in their support did not amount to a
mathematical proof. It was, if anything, even more sloppy than the reasoning
employed by people like Euler and Bernoulli. Additionally, if Fourier was right,
then his series in effect derived a common law for discontinuous functions. The
function above, with values 0 and 1, has a periodic relative, the square wave.
And the square wave has a single Fourier series, quite a nice one, which works
equally well in those regions where the function is 0 and in those regions where
the function is 1. So a function that appears to be represented by two different
laws can be rewritten in terms of one law.
Slowly the mathematicians of the 19th century started separating the different
conceptual issues in this difficult area. One was the meaning of the term,
function. Another was the various ways of representing a function – by a
formula, a power series, a Fourier series or whatever. A third was what
properties the function possessed. A fourth was which representations
guaranteed which properties. A single polynomial, for example, defines a
continuous function. A single Fourier series, it seemed, might not.

The square wave and some of its Fourier approximations

Fourier analysis rapidly became the test case for ideas about the function
concept. Here the problems came into sharpest relief and esoteric technical
distinctions turned out to be important. And it was in a paper on Fourier series,
in 1837, that Dirichlet introduced the modern definition of a function. In effect,
he agreed with Fourier: a variable y is a function of another variable x if for each
value of x (in some particular range) there is specified a unique value of y. He
explicitly stated that no particular law or formula was required: it is enough for y
to be specified by some well-defined sequence of mathematical operations,
applied to x. What at the time must have seemed an extreme example is ‘one he
made earlier’, in 1829: a function f(x) taking one value when x is rational, and a
different value when x is irrational. This function is discontinuous at every point.
(Nowadays functions like this are viewed as being rather mild; far worse
behaviour is possible.)
For Dirichlet, the square root was not one two-valued function. It was two
one-valued functions. For real x, it is natural – but not essential – to take the
positive square root as one of them, and the negative square root as the other.
For complex numbers, there are no obvious natural choices, although a certain
amount can be done to make life easier.

Continuous functions

By now it was dawning on mathematicians that although they often stated

definitions of the term ‘function’, they had a habit of assuming extra properties
that did not follow from the definition. For example, they assumed that any
sensible formula, such as a polynomial, automatically defined a continuous
function. But they had never proved this. In fact, they couldn’t prove it, because
they hadn’t defined ‘continuous’. The whole area was awash with vague
intuitions, most of which were wrong.
The person who made the first serious start on sorting out this mess was a
Bohemian priest, philosopher and mathematician. His name was Bernhard
Bolzano. He placed most of the basic concepts of calculus on a sound logical
footing; the main exception was that he took the existence of real numbers for
granted. He insisted that infinitesimals and infinitely large numbers do not exist,
and so cannot be used, however evocative they may be. And he gave the first
effective definition of a continuous function. Namely, f is continuous if the
difference f(x + a) − f(x) can be made as small as we please by choosing a
sufficiently small. Previous authors had tended to say things like ‘if a is
infinitesimal then f(x + a) − f(x) is infinitesimal’. But for Bolzano, a was just a
number, like any others. His point was that whenever you specify how small you
want f(x + a) − f(x) to be, you must then specify a suitable value for a. It wasn’t
necessary for the same value to work in every instance.
So, for instance, f(x) = 2x is continuous, because 2(x + a) – 2x = 2a. If you
want 2a to be smaller than some specific number, say 10−10, then you have to
make a smaller than 10−10/2. If you try a more complicated function, like f(x) =
x2, then the exact details are a bit complicated because the right a depends on x
as well as on the chosen size, 10−10, but any competent mathematician can work
this out in a few minutes. Using this definition, Bolzano proved – for the first
time ever – that a polynomial function is continuous. But for 50 years, no one
noticed. Bolzano had published his work in a journal that mathematicians hardly
ever read, or had access to. In these days of the Internet it is difficult to realize
just how poor communications were even 50 years ago, let alone 180.

‘. . .gradually order emerged from the chaos.’

In 1821 Cauchy said much the same thing, but using slightly confusing
terminology. His definition of continuity of a function f was that f(x) and f(x + a)
differ by an infinitesimal amount whenever a is infinitesimal, which at first sight
looks like the old, poorly-defined approach. But infinitesimal to Cauchy did not
refer to a single number that was somehow infinitely small, but to an ever-
decreasing sequence of numbers. For instance, the sequence 0.1, 0.01, 0.001,
0.0001 and so on is infinitesimal in Cauchy’s sense, but each of the individual
numbers, such as 0.0001, is just a conventional real number – small, perhaps, but
not infinitely so. Taking this terminology into account, we see that Cauchy’s
concept of continuity amounts to exactly the same thing as Bolzano’s.
Another critic of sloppy thinking about infinite processes was Abel, who
complained that people were using infinite series without enquiring whether the
sums made any sense. His criticisms struck home, and gradually order emerged
from the chaos.

Limits

Bolzano’s ideas set these improvements in motion. He made it possible to define

the limit of an infinite sequence of numbers, and from that the series, which is
the sum of an infinite sequence. In particular, his formalism implied that

What analysis did for them

The mathematical physics of the 19th century led to the discovery of a number of
important differential equations. In the absence of high-speed computers, able to find
numerical solutions, the mathematicians of the time invented new special functions to
solve these equations. These functions are still in use today. An example is Bessel’s
equation, first derived by Daniel Bernoulli and generalized by Bessel. This takes the form
 and the standard functions, such as exponentials, sines, cosines and logarithms, do not
provide a solution.
However, it is possible to use analysis to find solutions, in the form of power series.
The power series determine new functions, Bessel functions. The simplest types of
Bessel function are denoted Jk(x); there are several others. The power series permits
the calculation of Jk(x) to any desired accuracy.
Bessel functions turn up naturally in many problems about circles and cylinders, such
as the vibration of a circular drum, propagation of electromagnetic waves in a cylindrical
waveguide, the conduction of heat in a cylindrical metal bar, and the physics of lasers.

Laser beam intensity described by the Bessel function J1(x)

1 + 1/2 + 1/4 + 1/8 + 1/16 + . . .

carries on forever, is a meaningful sum, and its value is exactly 2. Not a little bit
less; not an infinitesimal amount less; just plain 2.
To see how this goes, suppose we have a sequence of numbers

a0, a1, a2, a3, . . .

going on forever. We say that an tends to limit a as n tends to infinity if, given
any number ε > 0 there exists a number N such that the difference between an
and a is less than whenever n > N (The symbol ε which is the one traditionally
used, is the Greek epsilon.) Note that all of the numbers in this definition are
finite – there are no infinitesimals or infinities.
To add up the infinite series above, we look at the finite sums

and so on. The difference between an and 2 is 1/2n. To make this less than ε we
take n > N = log2 (1/ε).
A series that has a finite limit is said to be convergent. A finite sum is defined
to be the limit of the sequence of finite sums, obtained by adding more and more
terms. If that limit exists, the series is convergent. Derivatives and integrals are
just limits of various kinds. They exist – that is, they make mathematical sense –
provided those limits converge. Limits, just as Newton maintained, are about
what certain quantities approach as some other number approaches infinity, or
zero. The number does not have to reach infinity or zero.
The whole of calculus now rested on sound foundations. The downside was
that whenever you used a limiting process, you had to make sure that it
converged. The best way to do that was to prove ever more general theorems
about what kinds of functions are continuous, or differentiable or integrable, and
which sequences or series converge. This is what the analysts proceeded to do,
and it is why we no longer worry about the difficulties pointed out by Bishop
Berkeley. It is also why we no longer argue about Fourier series: we have a
sound idea of when they converge, when they do not and indeed in what sense
they converge. There are several variations on the basic theme, and for Fourier
series you have to pick the right ones.

Power series
Weierstrass realized that the same ideas work for complex numbers as well as
real numbers. Every complex number z = x + iy has an absolute value ,
which by Pythagoras’s Theorem is the distance from 0 to z in the complex plane.
If you measure the size of a complex expression using its absolute value, then
the real-number concepts of limits, series and so on, as formulated by Bolzano,
can immediately be transferred to complex analysis.
Weierstrass noticed that one particular type of infinite series seemed to be
especially useful. It is known as a power series, and it looks like a polynomial of
infinite degree:

f(z) = a0 + a1z + a2z2 + a3z3 + . . .

where the coefficients an are specific numbers. Weierstrass embarked on a huge

programme of research, aiming to found the whole of complex analysis on
power series. It worked brilliantly.
For example, you can define the exponential function by

where the numbers 2, 4, 6 and so on are factorials: products of consecutive

integers (for instance 120 = 1×2×3×4×5). Euler had already obtained this
formula heuristically; now Weierstrass could make rigorous sense of it. Again
taking a leaf from Euler’s book, he could then relate the trigonometric functions
to the exponential function, by defining

All of the standard properties of these functions followed from their power
series expression. You could even define π, and prove that eiπ = −1, as Euler had
maintained. And that in turn meant that complex logarithms did what Euler had
claimed. It all made sense. Complex analysis was not just some mystical
extension of real analysis: it was a sensible subject in its own right. In fact, it
was often simpler to work in the complex domain, and read off the real result at
the end.
 To Weierstrass, all this was just a beginning – the first phase of a vast
programme. But what mattered was getting the foundations right. If you did that,
the more sophisticated material would readily follow.
Weierstrass was unusually clear-headed, and he could see his way through
complicated combinations of limits and derivatives and integrals without getting
confused. He could also spot potential difficulties. One of his most surprising
theorems proves that there exists a function f(x) of a real variable x, which is
continuous at every point, but differentiable at no point. The graph of f is a
single, unbroken curve, but it is such a wiggly curve that it has no well-defined
tangent anywhere. His predecessors would not have believed it; his
contemporaries wondered what it was for. His successors developed it into one
of the most exciting new theories of the 20th century, fractals.

The Riemann Hypothesis

The most famous unsolved problem in the whole of mathematics is the Riemann Hypothesis, a
problem in complex analysis which arose in connection with prime numbers but has repercussions
throughout mathematics.
Around 1793 Gauss conjectured that the number of primes less than x is approximately x/log x.
In fact he suggested a more accurate approximation called the logarithmic integral. In 1737 Euler
had noticed an intriguing connection between number theory and analysis: the infinite series

1 + 2−s + 3−s + 4−s + . . .

is equal to the product, over all primes p, of the series

Here we must take s > 1 for the series to converge.

In 1848 Pafnuty Chebyshev made some progress towards a proof of Gauss’s conjecture, using a
complex function related to Euler’s series, later called the zeta function ζ(z). The role of this
function was made clear by Riemann in his 1859 paper On the Number of Primes Less Than a
Given Magnitude. He showed that the statistical properties of primes are closely related to the
zeros of the zeta function, that is, the solutions, z, of the equation ζ(z) = 0.
In 1896 Jacques Hadamard and Charles de la Vallée Poussin used the zeta function to prove the
Prime Number Theorem. The main step is to show that ζ(z) is non-zero for all z of the form 1+ it.
The more control we can gain over the location of the zeros of the zeta function, the more we
 learn about primes. Riemann conjectured that all zeros, other than some obvious ones at negative
even integers, lie on the critical line. z = ½ + it.
In 1914 Hardy proved that an infinite number of zeros lie on this line. Extensive computer
evidence also supports the conjecture. Between 2001 and 2005 Sebastian Wedeniwski’s program
ZetaGrid verified that the first 100 billion zeros lie on the critical line.
The Riemann Hypothesis was part of Problem 8 in Hilbert’s famous list of 23 great unsolved
mathematical problems, and is one of the Millennium prize problems of the Clay Mathematics
Institute.

But more of that story later.

A firm basis

The early inventors of calculus had taken a rather cavalier approach to infinite
operations. Euler had assumed that power series were just like polynomials, and
used this assumption to devastating effect. But in the hands of ordinary mortals,
this kind of assumption can easily lead to nonsense. Even Euler stated some
pretty stupid things. For instance, he started form the power series

1 + x + x2 + x3 + x4 + . . .

which sums to 1/(1 − x), set x = −1 and deduced that

1 − 1 + 1 − 1 + 1 − 1 + . . . = ½

which is nonsense. The power series does not converge unless x is strictly
between –1 and 1, as Weierstrass’s theory makes clear.
Taking criticisms like those made by Bishop Berkeley seriously did, in the
long run, enrich mathematics and place it on a firm footing. The more
complicated the theories became, the more important it was to make sure you
were standing on firm ground.

‘Even Euler stated some pretty stupid things.’

Today, most users of mathematics once more ignore such subtleties, safe in
the knowledge that they have been sorted and that anything that looks sensible
probably has a rigorous justification. They have Bolzano, Cauchy and
Weierstrass to thank for this confidence. Meanwhile, professional
mathematicians continue to develop rigorous concepts about infinite processes.
There is even a movement to revive the concept of an infinitesimal, known as
non-standard analysis, and it is perfectly rigorous and technically useful on some
otherwise intractable problems. It avoids logical contradictions by making
infinitesimals a new kind of number, not a conventional real number. In spirit, it
is close to the way Cauchy thought. It remains a minority speciality – but watch
this space.

Absolute value of the Riemann zeta function

What analysis does for us

Analysis is used in biology to study the growth of populations of organisms. A simple
example is the logistic or Verhulst–Pearl model. Here the change in the population, x, as
a function of time t is modelled by a differential equation

where the constant M is the carrying capacity, the largest population that the
environment can sustain.
Standard methods in analysis produce the explicit solution
which is called the logistic curve. The corresponding pattern of growth begins with rapid
(exponential) growth, but when the population reaches half the carrying capacity it
begins to level off, and eventually levels off at the carrying capacity.
This curve is not totally realistic, although it fits many real populations quite well. More
complicated models of the same kind provide closer fits to real data. Human
consumption of natural resources can also follow a pattern similar to the logistic curve,
making it possible to estimate future demand, and how long the resources are likely to
last.

World consumption of crude oil, 1900–2000:smooth curve, logistic equation; jagged

curve, actual data
 12
Impossible Triangles

Is Euclid’s geometry the only one?

Calculus was based on geometric principles, but the geometry was reduced to
symbolic calculations, which were then formalized as analysis. However, the
role of visual thinking in mathematics was also developing, in a new and initially
rather shocking direction. For more than 2000 years the name Euclid had been
synonymous with geometry. His successors developed his ideas, especially in
their work on conic sections, but they did not make radical changes to the
concept of geometry itself. Essentially, it was assumed that there can be only one
geometry, Euclid’s, and that this is an exact mathematical description of the true
geometry of physical space. People found it difficult even to conceive of
alternatives.
It couldn’t last.
Spherical and projective geometry

The first significant departure from Euclidean geometry arose from the very
practical issue of navigation. Over short distances, the Earth is almost flat, and
its geographical features can be mapped on a plane. But as ships made ever
longer voyages, the true shape of the planet had to be taken into account. Several
ancient civilizations knew that the Earth is round – there is ample evidence, from
the way ships seem to disappear over the horizon to the shadow of the planet on
the Moon during lunar eclipses. It was generally assumed that the Earth is a
perfect sphere.
In reality, the sphere is slightly flattened: the diameter at the equator is 12,756
km, whereas that at the poles it is 12,714 km. The difference is relatively small –
one part in 300. In times when navigators routinely made errors of several
hundred kilometres, a spherical Earth provided an entirely acceptable
mathematical model. At that time, however, the emphasis was on spherical
trigonometry rather than geometry – the nuts and bolts of navigational
calculations, not the logical analysis of the sphere as a kind of space. Because
the sphere sits naturally within three-dimensional Euclidean space, no one
considered spherical geometry to be different from Euclidean. Any differences
were the result of the Earth’s curvature. The geometry of space itself remained
Euclidean.
A more significant departure from Euclid was the introduction, from the early
17th century onwards, of projective geometry. This topic emerged not from
science but from art: theoretical and practical investigations of perspective by
the Renaissance artists of Italy. The aim was to make paintings look realistic; the
outcome was a new way to think about geometry. But, again, this development
could be seen as an innovation within the classical Euclidean frame. It was about
how we view space, not about space.
The discovery that Euclid was not alone, that there can exist logically
consistent types of geometry in which many of Euclid’s theorems fail to hold,
emerged from a renewed interest in the logical foundations of geometry, debated
and developed from the middle of the 18th century to the middle of the 19th. The
big issue was Euclid’s Fifth Postulate, which – clumsily – asserted the existence
of parallel lines. Attempts to deduce the Fifth Postulate from the remainder of
Euclid’s axioms eventually led to the realization that no such deduction is
possible. There are consistent types of geometry other than Euclidean. Today,
these non-Euclidean geometries have become indispensable tools in pure
mathematics and mathematical physics.

Geometry and art

As far as Europe was concerned, geometry was becalmed in the doldrums

between the years 300 and 1600. The revival of geometry as a living subject
came from the question of perspective in art: how to render a three-dimensional
world realistically on two-dimensional canvas.
The artists of the Renaissance did not just create paintings. Many were
employed to carry out engineering works, for peaceful or warlike purposes.
Their art had a practical side, and the geometry of perspective was a practical
quest, applying to architecture as well as to the visual arts. There was also a
growing interest in optics, the mathematics of light, which blossomed once the
telescope and microscope were invented. The first major artist to think about the
mathematics of perspective was Filippo Brunelleschi. In fact, his art was mostly
a vehicle for his mathematics. A seminal book is Leone Battista Alberti’s Della
Pittura, written in 1435 and printed in 1511. Alberti began by making some
important, and relatively harmless, simplifications – the standard reflex of a true
mathematician. Human vision is a complex subject. For example, we use two
slightly separated eyes to generate stereoscopic images, providing a feeling of
depth. Alberti simplified reality by assuming a single eye with a pinprick pupil,
which worked like a pinhole camera. He imagined an artist painting a scene,
setting up his easel and trying to make the image on the canvas match the one
perceived by his (single) eye. Both canvas and reality project their images on to
the retina, at the back of the eye. The simplest (conceptual) way to ensure a
perfect match is to make the canvas transparent, look through it from a fixed
location and draw upon the canvas exactly what the eye sees. So the three-
dimensional scene is projected on to the canvas. Join each feature of the scene to
the eye by a straight line, note where this line meets the plane of the canvas:
that’s where you paint that feature.

‘. . .geometry was in the doldrums between the years

300 and 1600.’
 This idea is not terribly practical if you take it literally, although some artists
did just that, using translucent materials, or glass, in place of a canvas. They
often did this as a preliminary step, transferring the resulting outline to canvas
for proper painting. A more practical approach is to use this conceptual
formulation to relate the geometry of the three-dimensional scene to that of the
two-dimensional image. Ordinary Euclidean geometry is about features that
remain unchanged by rigid motions – lengths, angles. Euclid did not formulate it
that way, but his use of congruent triangles as a basic tool has the same effect.
(These are triangles of the same size and shape, but in different locations.)
Similarly, the geometry of perspective boils down to features that remain
unchanged by projection. It is easy to see that lengths and angles do not behave
like that. You can cover the Moon with your thumb, so lengths can change.
Angles fare no better – when you look at the corner of a building, a right angle,
it only looks like a right angle if you view it square on.

Projecting a scene – Albrecht Dürer

What properties of geometrical figures, then, are preserved by projection? The

most important ones are so simple that it is easy to miss their significance. Points
remain points. Straight lines remain straight. The image of a point lying on a
straight lines lies on the image of that line. Therefore, if two lines meet at a
point, their images meet at the corresponding point. Incidence relations of points
and lines are preserved by projection.
An important feature that is not quite preserved is the relation ‘parallel’.
Imagine standing in the middle of a long, straight road and look ahead. The two
sides of the road, which in three-dimensional reality are parallel – so never meet
– do not appear parallel. Instead, they converge towards a single point on the
distant horizon. They behave like this on an ideal infinite plane, not just on a
slightly rounded Earth. In fact, they only behave exactly like this on a plane. On
a sphere, there would be a tiny gap, too small to see, where the lines cross the
horizon. And the whole issue of parallel lines on a sphere is tricky anyway.
This feature of parallel lines is very useful in perspective drawing. It lies
behind the usual way of drawing right-angled boxes in perspective, using a
horizon line and two vanishing points, which are where parallel edges of the box
cross the horizon in perspective. Piero della Francesca’s De Prospettiva
Pingendi (1482–87) developed Alberti’s methods into practical techniques for
artists, and he used them to great effect in his dramatic and very realistic
paintings.
The writings of the Renaissance painters solved many problems in the
geometry of perspective, but they were semi-empirical, lacking the kind of
logical foundation that Euclid had supplied for ordinary geometry. These
foundational issues were finally resolved by Brook Taylor and Johann Heinrich
Lambert in the 18th century. But by then, more exciting things were going on in
geometry.

Desargues

The first non-trivial theorem in projective geometry was found by the

engineer/architect Girard Desargues and published in 1648 in a book by
Abraham Bosse. Desargues proved the following remarkable theorem. Suppose
that triangles ABC and A′B′C′ are in perspective, which means that the three lines
AA′, BB′ and CC′ all pass through the same point. Then the three points P, Q and
R at which corresponding sides of the two triangles meet all lie on the same line.
This result is called Desargues’s Theorem to this day. It mentions no lengths, no
angles – it is purely about incidence relations among lines and points. So it is a
projective theorem.

Desargues’s Theorem

There is a trick which makes the theorem obvious: imagine it as a drawing of

a three-dimensional figure, in which the two triangles lie in two planes. Then the
line along which those planes intersect is the line containing Desargues’s three
points P, Q and R. With a little care, the theorem can even be proved this way,
by constructing a suitable three-dimensional figure whose projection looks like
the two triangles. So we can use Euclidean methods to prove projective
theorems.

Euclid’s axioms

Projective geometry differs from Euclidean geometry as far as its viewpoint goes
(pun intended), but it is still related to Euclidean geometry. It is the study of new
kinds of transformation, projections, but the underlying model of the space that
is being transformed is Euclidean. Nevertheless, projective geometry made
mathematicians more receptive to the possibility of new kinds of geometric
thinking. And an old question, one that had lain dormant for centuries, once
more came to the fore.
Nearly all of Euclid’s axioms for geometry were so obvious that no sane
person could seriously question them. All right angles are equal, for instance. If
that axiom failed, there had to be something wrong with the definition of a right
angle. But the Fifth Postulate, the one that was really about parallel lines, had a
distinctly different flavour. It was complicated. Euclid states it this way: If a
straight line falling on two straight lines make the interior angles on the same
side less than two right angles, the two straight lines, if produced indefinitely,
meet on that side on which are the angles less than the two right angles.
It sounded more like a theorem than an axiom. Was it a theorem? Might there
be some way to prove it, perhaps starting from something simpler, more
intuitive?
One improvement was introduced by John Playfair in 1795. He substituted the
statement that for any given line, and any point not on that line, there exists one
and only one line through the point that is parallel to the given line. This
statement is logically equivalent to Euclid’s Fifth Postulate – that is, each is a
consequence of the other, given the remaining axioms.

Legendre

In 1794 Adrien-Marie Legendre discovered another equivalent statement, the

existence of similar triangles – triangles having the same angles, but with edges
of different sizes. But he, and most other mathematicians, wanted something
even more intuitive. In fact, there was a feeling that the Fifth Postulate was
simply superfluous – a consequence of the other axioms. All that was missing
was a proof. So Legendre tried all sorts of things. Using only the other axioms,
he proved – to his own satisfaction, at any rate – that the angles of a triangle
either add up to 180° or less. (He must have known that in spherical geometry
the sum is greater, but that is the geometry of the sphere, not the plane.) If the
sum is always 180°, the Fifth Postulate follows. So he assumed that the sum
could be less than 180°, and developed the implications of that assumption.

‘Most mathematicians wanted something more

intuitive.’

A striking consequence was a relation between the triangle’s area and the sum
of its angles. Specifically, the area is proportional to the amount by which the
angle sum falls short of 180°. This seemed promising: if he could construct a
triangle whose sides were twice those of a given triangle, but with the same
angles, then he would obtain a contradiction, because the larger triangle would
not have the same area as the smaller one. But however he tried to construct the
larger triangle, he found himself appealing to the Fifth Postulate.
He did manage to salvage one positive result from the work. Without
assuming the Fifth Postulate, he proved that it is impossible for some triangles to
have angle-sums greater than 180°, while others have angle-sums less than 180°.
If one triangle has angles that summed to more than 180°, so does every triangle;
similarly if the sum is less than 180°. So there are three possible cases:

• The angles of every triangle add up to 180° exactly (Euclidean geometry)

• The angles of every triangle add up to less than 180°
• The angles of every triangle add up to more than 180° (a case that Legendre
thought he had excluded; it later turned out that he had made other unstated
assumptions to do so).

Saccheri

In 1733 Gerolamo Saccheri, a Jesuit priest at Pavia, published a heroic effort,

Euclides ab Omni Naevo Vindicatus (Euclid Vindicated from All Flaws). He also
considered three cases, of which the first was Euclidean geometry, but he used a
quadrilateral to make the distinction. Suppose the quadrilateral is ABCD, with A
and B right angles and AC = BD. Then, said Saccheri, Euclidean geometry
implies that angles C and D are right angles. Less obviously, if C and D are right
angles in any one quadrilateral of this kind, then the Fifth Postulate follows.
Without using the Fifth Postulate, Saccheri proved that angles C and D are
equal. So that left two distinct possibilities:

• Hypothesis of the obtuse angle: both C and D are greater than a right angle.
• Hypothesis of the acute angle: both C and D are less than a right angle.
Saccheri’s quadrilateral: the line CD has been drawn curved to avoid Euclidean assumptions about angles C
and D

Saccheri’s idea was to assume each of these hypotheses in turn, and deduce a
logical contradiction. That would then leave Euclidean geometry as the only
logical possibility.
He began with the hypothesis of the obtuse angle, and in a series of theorems
deduced – so he thought – that angles C and D must in fact be right angles after
all. This was a contradiction, so the hypothesis of the obtuse angle had to be
false. Next, he assumed the hypothesis of the acute angle, which led to another
series of theorems, all correct, and fairly interesting in their own right.
Eventually he proved a rather complicated theorem about a family of lines all
passing through one point, which implied that two of these lines would have a
common perpendicular at infinity. This is not actually a contradiction, but
Saccheri thought it was, and declared the hypothesis of the acute angle to be
disproved as well.

What Non-Euclidean geometry did for them

By 1813 Gauss was becoming ever more convinced that what he first called anti-
Euclidean, then astral and finally non-Euclidean geometry was a logical possibility. He
began to wonder what the true geometry of space was, and measured the angles of a
triangle formed by three mountains – the Brocken, the Hohehagen and the Inselberg. He
used line-of-sight measurement so the curvature of the Earth did not come into play. The
sum of the angles that he measured was 15 seconds of arc greater than 180°. If
anything, this was the obtuse-angle case, but the likelihood of observational errors made
the whole exercise moot. Gauss needed a much bigger triangle, and much more
 accurate instruments to measure its angles.

That left only Euclidean geometry, so Saccheri felt that his programme was
vindicated, along with Euclid. But others noticed that he had not really obtained
a contradiction from the hypothesis of the acute angle; just a rather surprising
theorem. By 1759 d’Alembert declared the status of the Fifth Postulate to be ‘the
scandal of the elements of geometry’.

Lambert
A German mathematician, Georg Klügel, read Saccheri’s book, and offered the
unorthodox and rather shocking opinion that belief in the truth of the Fifth
Postulate was a matter of experience rather than logic. Basically, he was saying
that something in the way we think about space makes us believe in the
existence of parallel lines of the kind envisaged by Euclid.
In 1766 Lambert, following up on Klügel’s suggestion, embarked on an
investigation that was similar to Saccheri’s, but he started from a quadrilateral
with three right angles. The remaining angle must either be a right angle
(Euclidean geometry), acute or obtuse. Like Saccheri, he thought that the obtuse
angle case led to a contradiction. More precisely, he decided that it led to
spherical geometry, where it had long been known that the angles of a
quadrilateral add up to more than 360°, because the angles of a triangle add up to
more than 180°. Since the sphere is not the plane, the obtuse case is ruled out.
However, he did not claim the same for the acute angle case. Instead, he
proved some curious theorems, the most striking being a formula for the area of
a polygon with n sides. Add all the angles, and subtract this from 2n – 4 right
angles: the result is proportional to the polygon’s area. This formula reminded
Lambert of a similar formula for spherical geometry: add all the angles, and
subtract 2n – 4 right angles from this: again the result is proportional to the
polygon’s area. The difference is minor: the subtraction is performed in the
opposite order. He was led to a remarkably prescient but obscure prediction: the
geometry of the acute-angle case is the same as that on a sphere with imaginary
radius.
He then wrote a short article about trigonometric functions of imaginary
angles, obtaining some beautiful and perfectly consistent formulas. We now
recognize these functions as the so-called hyperbolic functions, which can be
defined without using imaginary numbers, and they satisfy all of Lambert’s
formulas. Clearly something interesting must lie behind his curious, enigmatic
suggestion. But what?

Gauss’ dilemma

By now the best-informed geometers were getting a definite feeling that Euclid’s
Fifth Postulate could not be proved from the remaining axioms. The acute-angle
case seemed too self-consistent ever to lead to a contradiction. On the other
hand, a sphere of imaginary radius was not the sort of object that could be
proposed to justify that belief.
One such geometer was Gauss, who convinced himself from an early age that
a logically consistent non-Euclidean geometry was possible, and proved
numerous theorems in such a geometry. But, as he made clear in an 1829 letter
to Bessel, he had no intention of publishing any of this work, because he feared
what he called the ‘clamour of the Boeotians’. Unimaginative people would not
understand, and in their ignorance and hidebound adherence to tradition, they
would ridicule the work. In this he may have been influenced by the overarching
status of Kant’s widely acclaimed work in philosophy; Kant had argued that the
geometry of space must be Euclidean.

‘Kant had argued that the geometry of space must be

Euclidean.’

By 1799 Gauss was writing to the Hungarian Wolfgang Bolyai, telling him
that the research ‘seems rather to compel me to doubt the truth of geometry
itself. It is true that I have come upon much which by most people would be held
to constitute a proof [of the Fifth Postulate from the other axioms]; but in my
eyes it is as good as nothing.’
Other mathematicians were less circumspect. In 1826 Nikolai Ivanovich
Lobachevsky, at the University of Kazan in Russia, gave lectures on non-
Euclidean geometry. He knew nothing of Gauss’s work, but had proved similar
theorems using his own methods. Two papers on the topic appeared in 1829 and
1835. Rather than starting riots, as Gauss had feared, these papers sank pretty
much without trace. By 1840 Lobachevsky was publishing a book on the topic,
in which he complained about the lack of interest. In 1855 he published a further
book on the subject.
Independently, Wolfgang Bolyai’s son János, an officer in the army, came up
with similar ideas around 1825, writing them up in a 26-page paper that was
published as an appendix to his father’s geometry text Tentamen Juventum
Studiosam in Elementa Matheseos (Essay on the Elements of Mathematics for
Studious Youths) of 1832. ‘I have made such wonderful discoveries that I am
myself lost in astonishment,’ he wrote to his father.
Gauss read the work, but explained to Wolfgang that it was impossible for
him to praise the young man’s efforts, because Gauss would in effect be praising
himself. This was perhaps a little unfair, but it was how Gauss tended to operate.

Non-Euclidean geometry

The history of non-Euclidean geometry is too complicated to describe in any

further detail, but we can summarize what followed these pioneering efforts.
There is a deep unity behind the three cases noticed by Saccheri, by Lambert,
and by Gauss, Bolyai and Lobachevsky. What unites them is the concept of
curvature. Non-Euclidean geometry is really the natural geometry of a curved
surface.

Poincaré’s model of hyperbolic geometry makes it clear there are infinitely many parallel lines through a
point that do not meet a given line
 point that do not meet a given line

If the surface is positively curved like a sphere, then we have the case of the
obtuse angle. This was rejected because spherical geometry differs from
Euclidean in obvious ways – for example, any two lines, that is, great circles
(circles whose centres are at the centre of the Earth), meet in two points, not the
one that we expect of Euclidean straight lines.
Actually, we now realize that this objection is unfounded. If we identify
diametrically opposite points of the sphere – that is, pretend they are identical –
then lines (great circles) still make sense, because if a point lies on a great circle,
so does the diametrically opposite point. With this identification, nearly all of the
geometric properties remain unchanged, but now lines meet in one point.
Topologically, the surface that results is the projective plane, although the
geometry concerned is not orthodox projective geometry. We now call it elliptic
geometry, and it is considered just as sensible as Euclidean geometry.
If the surface is negatively curved, shaped like a saddle, then we have the case
of the acute angle. The resulting geometry is called hyperbolic. It has numerous
intriguing features, which distinguish it from Euclidean geometry.
If the surface has zero curvature, like a Euclidean plane, then it is the
Euclidean plane, and we get Euclidean geometry.
All three geometries satisfy all of Euclid’s axioms other than the Fifth
Postulate. Euclid’s decision to include his postulate is vindicated.
These various geometries can be modelled in several different ways.
Hyperbolic geometry is especially versatile in this respect. In one model the
space concerned is the upper half of the complex plane, omitting the real axis
and everything below it. A line is a semicircle that meets the real axis at right
angles. Topologically, this space is the same as a plane and the lines are identical
to ordinary lines. The curvature of the lines reflects the negative curvature of the
underlying space.
In a second model of hyperbolic geometry, introduced by Poincaré, the space
is represented as the interior of a circle, not including its boundary, and the lines
are circles that meet the boundary at right angles. Again, the distorted geometry
reflects the curvature of the underlying space. The artist Maurits Escher
produced many pictures based on this model of hyperbolic geometry, which he
learned from the Canadian geometer Coxeter.
These two models hint at some deep connections between hyperbolic
geometry and complex analysis. These connections relate to certain groups of
transformations of the complex plane; hyperbolic geometry is the geometry of
their invariants, according to Felix Klein’s Erlangen Programme. Another class
of transformations, called Möbius transformations, brings elliptic geometry into
play as well.

The geometry of space

What of the geometry of space? We now agree with Klügel, and disdain Kant.
This is a matter for experience, not something that can be deduced by thought
alone. Einstein’s General Relativity tells us that space (and time) can be curved;
the curvature is the gravitational effect of matter. The curvature can vary from
one location to another, depending on how the matter is distributed. So the
geometry of space is not really the issue. Space can have different geometries in
different places. Euclid’s geometry works well on human scales, in the human
world, because gravitational curvature is so small that we don’t observe it in our
daily lives. But out there in the greater universe, non-Euclidean geometries
prevail.
To the ancients and indeed well into the 19th century, mathematics and the
real world were hopelessly confused. There was a general belief that
mathematics was a representation of basic and inevitable features of the real
world, and that mathematical truth was absolute. Nowhere was this assumption
more deeply rooted than in classical geometry. Space was Euclidean, to virtually
everyone who thought about the question. What else could it be?

What Non-Euclidean geometry does for us

What shape is the universe? The question may seem simple but answering it is difficult –
partly because the universe is so big, but mainly because we are inside it and cannot
stand back and see it as a whole. In an analogy that goes back to Gauss, an ant living
on a surface, and observing it only within that surface, could not easily tell if the surface
was a plane, a sphere, a torus or something more complicated.
General relativity tells us that near a material body, such as a star, space– time is
curved. Einstein’s equations, which relate the curvature to the density of matter, have
many different solutions. In the simplest ones, the universe as a whole has positive
curvature and its topology is that of a sphere. But for all we can tell, the overall curvature
of the real universe might be negative instead. We don’t even know whether the
universe is infinite, like Euclidean space, or is of finite extent, like a sphere. A few
physicists maintain that the universe is infinite, but the experimental basis for this
assertion is highly questionable. Most think it is finite.

Space with positive, negative and zero curvature

A closed universe curves back on itself. Lines that were diverging apart come
back together. Density > critical density.

An open universe curves away from itself. Diverging lines curve at increasing
angles away from each other. Density < critical density.
 A flat universe has no curvature. Diverging lines remain at a constant angle with
respect to each other. Density = critical density.

Surprisingly, a finite universe can exist without having a boundary. The sphere is like
that in two dimensions, and so is a torus. The torus can be given a flat geometry,
inherited from a square by identifying opposite edges. Topologists have also discovered
that space can be finite yet negatively curved: one way to construct such spaces is to
take a finite polyhedron in hyperbolic space and identify various faces, so that a line
passing out of the polyhedron across one face immediately re-enters at another face.
This construction is analogous to the way the edges of the screen wrap round in many
computer games.

To get Poincaré’s dodecahedral space, identify opposite faces

If space is finite, then it should be possible to observe the same star in different
directions, though it might seem much further away in some directions than in others,
and the observable region of the universe might be too small anyway. If a finite space
has hyperbolic geometry, these multiple occurrences of the same stars in different
 directions determine a system of gigantic circles in the heavens, and the geometry of
those circles determines which hyperbolic space is being observed. But the circles could
be anywhere among the billions of stars that can be seen, and so far attempts to
observe them, based on statistical correlations among the apparent positions of stars,
have not produced any result.
In 2003 data from the Wilkinson Microwave Anisotropy Probe led Jean-Pierre Luminet
and his collaborators to propose that space is finite but positively curved. They found
that Poincaré’s dodecahedral space – obtained by identifying opposite faces of a curved
dodecahedron – gives the best agreement with observations. This suggestion received
wide publicity as the assertion that the universe is shaped like a football. This suggestion
has not been confirmed and we currently have no idea of the true shape of space.
However, we do have a much better understanding of what has to be done to find out.

This question ceased to be rhetorical when logically consistent alternatives to

Euclid’s geometry began to appear. It took time to recognize that they were
logically consistent – at least, just as consistent as Euclid’s geometry – and even
longer to realize that our own physical space might not be perfectly Euclidean.
As always, human parochialism was to blame – we were projecting our own
limited experiences in one tiny corner of the universe on to the universe as a
whole. Our imaginations do seem to be biased in favour of a Euclidean model,
probably because, on the small scales of our experience, it is an excellent model
and also the simplest one available.

‘. . . there are many alternatives to Euclid’s

geometry.’

Thanks to some imaginative and unorthodox thinking, often viciously

contested by a less imaginative majority, it is now understood – by
mathematicians and physicists, at least – that there are many alternatives to
Euclid’s geometry, and that the nature of physical space is a question for
observation, not thought alone. Nowadays, we make a clear distinction between
mathematical models of reality, and reality itself. For that matter, much of
mathematics bears no obvious relation to reality at all – but is useful, all the
same.
 13
The Rise of Symmetry

How not to solve an equation

Around 1850 mathematics underwent one of the most significant changes in its
entire history, although this was not apparent at the time. Before 1800, the main
objects of mathematical study were relatively concrete: numbers, triangles,
spheres. Algebra used formulas to represent manipulations with numbers, but the
formulas themselves were viewed as symbolic representations of processes, not
as things in their own right. But by 1900 formulas and transformations were
viewed as things, not processes, and the objects of algebra were much more
abstract and far more general. In fact, pretty much anything went as far as
algebra was concerned. Even the basic laws, such as the commutative law of
multiplication, ab = ba, had been dispensed with in some important areas.

Group Theory

These changes came about largely because mathematicians discovered group

theory, a branch of algebra that emerged from unsuccessful attempts to solve
algebraic equations, especially the quintic, or fifth degree, equation. But within
50 years of its discovery, group theory had been recognized as the correct
framework for studying the concept of symmetry. As the new methods sank into
the collective consciousness, it became clear that symmetry is a deep and central
idea, with innumerable applications to the physical sciences, indeed to the
biological ones as well. Today, group theory has become an indispensable tool in
every area of mathematics and science, and its connections with symmetry are
emphasized in most introductory texts. But this point of view took several
decades to develop. Around 1900 Henri Poincaré said that group theory was
effectively the whole of mathematics reduced to its essentials, which was a bit of
an exaggeration, but a defensible one.

‘Today, group theory has become an indispensable

tool. . .’

The turning-point in the evolution of group theory was the work of a young
Frenchman, Évariste Galois. There was a long and complicated pre-history –
Galois’s ideas did not arise from a vacuum. And there was an equally
complicated and often somewhat muddled post-history, as mathematicians
experimented with the new concept, and tried to work out what was important
and what was not. But it was Galois, more than anyone else, who understood
clearly the need for groups, worked out some of their fundamental features, and
demonstrated their value in core mathematics. Not entirely surprisingly, his work
went almost unnoticed during his lifetime. It was a little too original, perhaps,
but it has to be said that Galois’s personality, and his fierce involvement in
revolutionary politics, did not help. He was a tragic figure living in a time of
many personal tragedies, and his life was one of the more dramatic, and perhaps
romantic, among those of major mathematicians.

Solving equations

The story of group theory goes right back to the ancient Babylonian work on
quadratic equations. As far as the Babylonians were concerned, their method
was intended for practical use; it was a computational technique, and they seem
not to have asked deeper questions about it. If you knew how to find square
roots, and had mastered basic arithmetic, you could solve quadratics.
 There are a few hints, in surviving clay tablets, that the Babylonians also
thought about cubic equations, even some quartic equations. The Greeks, and
after them the Arabs, discovered geometric methods for solving cubic equations
based on conic sections. (We now know that the traditional Euclidean lines and
circles cannot solve such problems exactly. Something more sophisticated was
necessary; as it happens, conics do the job.) One of the prominent figures here
was the Persian, Omar Khayyam. Omar solved all possible types of cubic by
systematic geometric methods. But, as we have seen, an algebraic solution of
cubic and quartic equations had to wait until the Renaissance, with the work of
Del Ferro, Tartaglia, Fior, Cardano and his pupil Ferrari.
The pattern that seemed to be emerging from all this work was
straightforward, even though the details were messy. You can solve any cubic
using arithmetical operations, plus square roots, plus cube roots. You can solve
any quartic using arithmetical operations, plus square roots, plus cube roots, plus
fourth roots – though the latter can be reduced to two square roots taken in
succession. It seemed plausible that this pattern would continue, so that you
could solve any quintic using arithmetical operations plus square roots, cube
roots, fourth roots and fifth roots. And so on, for equations of any degree. No
doubt the formulas would be very complicated, and finding them would be even
more complicated, but few seem to have doubted that they existed.

‘No doubt the formulas would be very

complicated. . .’

As the centuries went by, with no sign of any such formulas being found, a
few of the great mathematicians decided to take a closer look at the whole area,
to work out what was really going on behind the scenes, unify the known
methods and simplify them so that it became obvious why they worked. Then,
they thought, it would just be a matter of applying the same general principles,
and the quintic would give up its secret.
The most successful and most systematic work along these lines was carried
out by Lagrange. He reinterpreted the classical formulas in terms of the solutions
that were being sought. What mattered, he said, was how certain special
algebraic expressions in those solutions behaved when the solutions themselves
were permuted – rearranged. He knew that any fully symmetric expression – one
that remained exactly the same no matter how the solutions were shuffled –
could be expressed in terms of the coefficients of the equation, making it a
known quantity. More interesting were expressions that only took on a few
different values when the solutions were permuted. These seemed to hold the
key to the whole issue of solving the equation.

The symmetries of a quadratic

Consider a quadratic equation, in the slightly simplified form

x2 + px + q = 0

Suppose that the two solutions are x = a and x = b

x2 + px + q = (x – a) (x – b)

Then this tells us that

a + b = −p ab = q

So although we don’t yet know the solutions, we do know their sum and their product –
without doing any serious work.
Why is this? The sum a + b is the same as b + a − it does not change when the solutions are
permuted. The same goes for ab = ba. It turns out that every symmetric function of the solutions
can be expressed in terms of the coefficients p and q. Conversely, any expression in p and q is
always a symmetric function of a and b. Taking a broad view, the connection between the
solutions and the coefficients is determined by a symmetry property.
Asymmetric functions do not behave like this. A good example is the difference a − b. When
we swap a and b, this becomes b − a, which is different. However – the crucial observation – it is
not very different. It is what we get from a − b by changing its sign. So the square (a – b)2 is fully
symmetric. But any fully symmetric function of the solutions must be some expression in the
coefficients. Take the square root, and we have expressed a – b in terms of the coefficients, using
nothing more esoteric than a square root. We already know a + b – it is equal to –p. Since we also
know a – b, the sum of these two numbers is 2a and the difference is 2b. Dividing by 2, we obtain
formulas for a and for b.
What we’ve done is to prove that there must exist a formula for the solutions a and b involving
nothing more esoteric than a square root, based on general features of the symmetries of algebraic
expressions. This is impressive: we have proved that the problem possesses a solution, without
going to the bother of working out all the messy details that tell us what it is. In a sense, we have
tracked down why the Babylonians were able to find a method. This little story puts the word
 understand in a new light. You can understand how the Babylonian method produces a solution,
by working through the steps and checking the logic. But now we have understood why there had
to be some such method – not by exhibiting a solution, but by examining general properties of the
presumed solutions. Here, the key property turned out to be symmetry.
With a bit more work, leading to an explicit expression for (a – b)2, this method yields a
formula for the solutions. It is equivalent to the formula that we learn in school, and to the method
used by the Babylonians.

Lagrange’s well-developed sense of mathematical form and beauty told him

that this was a major idea. If something similar could be developed for cubic and
quartic equations, then he might discover how to solve the quintic.
Using the same basic idea, he found that partly symmetric functions of the
solutions allowed him to reduce a cubic equation to a quadratic. The quadratic
introduced a square root, and the reduction process could be sorted out using a
cube root. Similarly, any quartic equation could be reduced to a cubic, which he
called the resolvent cubic. So you could solve a quartic using square and cube
roots to deal with the resolvent cubic and fourth roots to relate the answer to the
solutions you wanted. In both cases, the answers were identical to the classical
Renaissance formulas. They had to be, really – those were the answers. But now
Lagrange knew why those were the answers, and better still, he knew why
answers existed to be found.

‘But now Lagrange knew why those were the

answers, and better still, he knew why answers
existed to be found.’

He must have got quite excited at this stage of his research. Moving on to the
quintic, and applying the same techniques, you expect to get a resolvent quartic
– job done. But, presumably to his disappointment, he didn’t get a resolvent
quartic. He got a resolvent sextic – an equation of the sixth degree.
Instead of making things simpler, his method made the quintic more
complicated.
Was this a flaw in the method? Might something even cleverer solve the
quintic? Lagrange seems to have thought so. He wrote that he hoped his new
viewpoint would be useful to anyone trying to develop a way to solve quintics. It
does not seem to have occurred to him that there might not be any such method;
that his approach failed because in general, quintics do not have solutions in
‘radicals’ – which are expressions involving arithmetical operations and various
roots, such as fifth roots. To confuse things, some quintics do have such
solutions, for instance, x5 – 2 = 0 has the solution x = . But that is a rather
simple case, and not really typical.
All quintic equations have solutions, by the way; in general these are complex
numbers, and they can be found numerically to any accuracy. The problem was
about algebraic formulas for solutions.

Search for a solution

As Lagrange’s ideas started to sink in, there was a growing feeling that perhaps
the problem could not be solved. Perhaps the general quintic equation cannot be
solved by radicals. Gauss seems to have thought so, privately, but expressed the
view that this was not a problem he thought was worth tackling. It is perhaps one
of the few instances where his intuition about what is important let him down;
another was Fermat’s Last Theorem, but here the necessary methods were
beyond even Gauss, and took a couple of centuries to emerge. But, ironically,
Gauss had already initiated some of the necessary algebra to prove the
insolubility of the quintic. He had introduced it in his work on the construction
of regular polygons by ruler and compass. And he had also, in this work, set a
precedent, by proving (to his own satisfaction, at any rate) that some polygons
could not be constructed in that manner. The regular 9-gon was an example.
Gauss knew this, but never wrote down a proof; one was supplied a little later by
Pierre Wantzel. So Gauss had established a precedent for the proposition that
some problems might not be soluble by particular methods.
The first person to attempt a proof of the impossibility was Paolo Ruffini, who
became a mathematics professor at the University of Modena in 1789. By
pursuing Lagrange’s ideas about symmetric functions, Ruffini became
convinced that there is no formula, involving nothing more esoteric than nth
roots, to solve the quintic. In his General Theory of Equations of 1799 he
claimed a proof that ‘The algebraic solution of general equations of degree
greater than four is always impossible’. But the proof was so long – 500 pages –
that no one was willing to check it, especially since there were rumours of
mistakes. In 1803 Ruffini published a new, simplified proof, but it fared no
better. During his lifetime, Ruffini never managed to secure the credit for
proving that the quintic is insoluble.
Ruffini’s most important contribution was the realization that permutations
can be combined with each other. Until then, a permutation was a rearrangement
of some collection of symbols. For instance, if we number the roots of a quintic
as 12345, then these symbols can be rearranged as 54321, or 42153, or 23154 or
whatever. There are120 possible arrangements. Ruffini realized that such a
rearrangement could be viewed in another way: as a recipe for rearranging any
other set of five symbols. The trick was to compare the standard order 12345
with the rearranged order. As a simple example, suppose the rearranged order
was 54321. Then the rule for getting from the initial standard order to the new
order was simple: reverse the order. But you can reverse the order of any
sequence of five symbols. If the symbols are abcde, the reverse is edcba. If the
symbols start out as 23451, then their reverse is 15432. This new way of viewing
a permutation meant that you could perform two permutations in turn – a kind of
multiplication of permutations. The algebra of permutations, multiplied in this
way, held the key to the secrets of the quintic.

Abel

We now know that there was a technical error in Ruffini’s proof, but the main
ideas are sound and the gap can be filled. He did achieve one thing: his book led
to a vague but widespread feeling that the quintic is not soluble by radicals.
Hardly anyone thought that Ruffini had proved this, but mathematicians began to
doubt that a solution could exist. Unfortunately the main effect of this belief was
to dissuade anyone from working on the problem.
An exception was Abel, a young Norwegian with a precocious talent for
mathematics, who thought that he had solved the quintic while still at school. He
eventually discovered a mistake, but remained intrigued by the problem, and
kept working on it intermittently. In 1823 he found a proof of the impossibility
of solving the quintic, and this proof was completely correct. Abel used a similar
strategy to Ruffini’s but his tactics were better. At first he was unaware of
Ruffini’s research; later he clearly knew of it, but he stated that it was
incomplete. However, he did not refer to any specific problem with Ruffini’s
proof. Ironically, one step in Abel’s proof is exactly what is needed to fill the
gap in Ruffini’s.
 We can get a general idea of Abel’s methods without going into too many
technicalities. He set up the problem by distinguishing two kinds of algebraic
operation. Suppose we start with various quantities – they may be specific
numbers or algebraic expressions in various unknowns. From them we can build
many other quantities. The easy way to do this is to combine the existing
quantities by adding them, subtracting them, multiplying them or dividing them.
So from a simple unknown, x, we can create expressions like x2, 3x + 4 or .
Algebraically, all of these expressions are on much the same footing as x itself.
The second way to get new quantities from existing ones is to use radicals.
Take one of the harmless modifications of existing quantities, and extract some
root. Call such a step adjoining a radical. If it is a square root, say that the degree
of the radical is 2, if a cube root, then the degree is 3, and so on.
In these terms, Cardano’s formula for the cubic can be summarized as the
result of a two-step procedure. Start with the coefficients of the cubic (and any
harmless combination of them). Adjoin a radical of degree 2. Then adjoin a
further radical of degree 3. That’s it. This description tells us what kind of
formula arises, but not exactly what it is. Often the key to answering a
mathematical riddle is not to focus on fine details, but too look at broad features.
Less can mean more. When it works, this trick is spectacular, and here it worked
beautifully. It allowed Abel to reduce any hypothetical formula for solving the
quintic to its essential steps: extract some sequence of radicals, in some order,
with various degrees. It is always possible to arrange for the degrees to be prime
– for instance, a sixth root is the cube root of a square root.
Call such a sequence a radical tower. An equation is soluble by radicals if at
least one of its solutions can be expressed by a radical tower. But instead of
trying to find a radical tower, Abel merely assumed that there was a radical
tower, and asked what the original equation must look like.

‘When it works, this trick is spectacular, and here it

worked beautifully.’

Without realizing it, Abel now filled the gap in Ruffini’s proof. He showed
that whenever an equation can be solved by radicals, there must exist a radical
tower leading to that solution, involving only the coefficients of the original
equation. This is called the Theorem on Natural Irrationalities and it states that
nothing can be gained by including a whole pile of new quantities, unrelated to
the original coefficients. This ought to be obvious, but Abel realized that it is in
many ways the crucial step in the proof.
The key to Abel’s impossibility proof is a clever preliminary result. Suppose
we take some expression in the solutions x1, x2, x3, x4, x5 of the equation, and
extract its pth root for some prime number p. Moreover, assume that the original
expression is unchanged when we apply two special permutations

S: x1, x2, x3, x4, x5 → x2, x3, x1, x4, x5

And

T: x1, x2, x3, x4, x5 → x1, x2, x4, x5, x3.

Then, Abel showed, the pth root of that expression is also unchanged when
we apply S and T. This preliminary result leads directly to the proof of the
impossibility theorem, by ‘climbing the tower’ step by step. Assume that the
quintic can be solved by radicals, so there is a radical tower that starts with the
coefficients and climbs all the way to some solution.
The first floor of the tower – the harmless expressions in the coefficients – is
unchanged when we apply the permutations S and T, because those permute the
solutions, not the coefficients. Therefore, by Abel’s preliminary result, the
second floor of the tower is also unchanged when we apply S and T, because it is
reached by adjoining a pth root of something in the ground floor, for some prime
p. By the same reasoning, the third floor of the tower is unchanged when we
apply S and T. So is the fourth floor, the fifth floor, . . ., all the way to the top
floor.
However, the top floor contains some solution of the equation. Could it be x1?
If so, then x1 must be unchanged when we apply S. But S applied to x1 gives x2,
not x1, so that’s no good. For similar reasons, sometimes using T, the solution
defined by the tower cannot be x2, x3, x4 or x5 either. All five solutions are
excluded from any such tower – so the hypothetical tower cannot, in fact,
contain a solution.
There is no escape from this logical trap. The quintic is unsolvable because
any solution (by radicals) must have self-contradictory properties, and therefore
cannot exist.
Galois

The pursuit not just of the quintic, but of all algebraic equations, was now taken
up by Évariste Galois, one of the most tragic figures in the history of
mathematics. Galois set himself the task of determining which equations could
be solved by radicals, and which could not. Like several of his predecessors, he
realized that the key to the algebraic solution of equations was how the solutions
behaved when permuted. The problem was about symmetry.
Ruffini and Abel had realized that an expression in the solutions did not have
to be either symmetric or not. It could be partially symmetric: unchanged by
some permutations but not by others. Galois noticed that the permutations that
fix some expression in the roots do not form any old collection. They have a
simple, characteristic feature. If you take any two permutations that fix the
expression, and multiply them together, the result also fixes the permutation. He
called such a system of permutations a group. Once you’ve realized that this is
true, it is very easy to prove it. The trick is to notice it, and to recognize its
significance.

Évariste Galois
1811-1832

Évariste Galois was the son of Nicholas Gabriel Galois and Adelaide Marie Demante. He grew
up in revolutionary France, developing distinctly left-wing political views. His great contribution
to mathematics went unrecognized until 14 years after his death.
The French revolution had begun with the storming of the Bastille in 1789 and the execution of
Louis XVI in 1793. By 1804 Napoleon Bonaparte had proclaimed himself Emperor, but after a
series of military defeats he was forced to abdicate, and the monarchy was restored in 1814 under
Louis XVIII. By 1824, Louis had died and the king was now Charles X.

In 1827 Galois began to display an unusual talent for—and an obsession with—mathematics. He

tried to gain entrance to the prestigious École Polytechnique, but failed the examination. In 1829
his father, then town Mayor, hanged himself when his political enemies invented a phoney
scandal. Shortly after, Galois tried once more to enter the École Polytechnique, and failed again.
Instead, he went to the École Normale.
In 1830 Galois submitted his researches on the solution of algebraic equations for a prize
offered by the Academy of Sciences. The referee, Fourier, promptly died, and the paper was lost.
The prize went to Abel (who was by then dead of tuberculosis) and to Carl Jacobi. In the same
year Charles X was deposed and fled for his life. The director of the École Normale locked his
students in to prevent them joining in. Galois, furious, wrote a sarcastic letter attacking the
director for cowardice, and was promptly expelled.
As a compromise, Louis-Phillipe was made king. Galois joined a republican militia, the
Artillery of the National Guard, but the new king abolished it. Nineteen of the Guard’s officers
were arrested and tried for sedition, but they jury threw the charges out, and the Guard held a
dinner to celebrate. Galois proposed an ironic toast to the king, holding a knife in his hand. He
was arrested, but acquitted because (so he claimed) the toast had been ‘To Louis-Phillipe, if he
betrays’, and not a threat to the king’s life. But on Bastille Day Galois was arrested again, for
wearing the now illegal uniform of the Guard.

In prison, he heard what had happened to his paper. Poisson had rejected it for being
insufficiently clear. Galois tried to kill himself, but the other prisoners stopped him. His hatred of
officialdom now became extreme, and he displayed signs of paranoia. But when a cholera
epidemic began, the prisoners were released.
At that point Galois fell in love with a woman whose name was for many years a mystery; she
turned out to have been Stephanie du Motel, the daughter of the doctor in Galois’s lodgings. The
affair did not prosper, and Stephanie ended it. One of Galois’s revolutionary comrades then
challenged him to a duel, apparently over Stephanie. A plausible theory, advanced by Tony
Rothman, is that the opponent was Ernest Duchâtelet, who had been imprisoned along with
Galois. The duel seems to have been a form of Russian roulette, involving a random choice from
two pistols, only one being loaded, and fired at point blank range. Galois chose the wrong pistol,
was shot in the stomach, and died the next day.
The night before the duel he wrote a long summary of his mathematical ideas, including a
description of his proof that all equations of degree 5 or higher cannot be solved by radicals. In
this work he developed the concept of a group of permutations, and took the first important steps
towards group theory. His manuscript was nearly lost, but it made its way to Joseph Liouville, a
member of the Academy. In 1843 Liouville addressed the Academy, saying that in Galois papers
he had found a solution ‘as correct as it is deep of this lovely problem: given an irreducible
equation of prime degree, decide whether or not it is soluble by radicals’. Liouville published
Galois’s papers in 1846, finally making them accessible to the mathematical community.

Part of a manuscript written by Évariste Galois

What group theory did for them

 One of the first serious applications of group theory to science was the classification of
all possible crystal structures. The atoms in a crystal form a regular three-dimensional
lattice, and the main mathematical point is to list all possible symmetry groups of such
lattices, because these effectively form the symmetries of the crystal.
In 1891 Evgraf Fedorov and Arthur Schönflies proved that there are exactly 230
distinct crystallographic space groups. William Barlow obtained a similar but incomplete
list.
Modern techniques for finding the structure of biological molecules, such as proteins,
rely on passing X-rays through a crystal formed by that molecule and observing the
resulting diffraction patterns. The symmetries of the crystal are important in deducing the
shape of the molecule concerned. So is Fourier analysis.

The upshot of Galois’s ideas is that the quintic cannot be solved by radicals
because it has the wrong kind of symmetries. The group of a general quintic
equation consists of all permutations of the five solutions. The algebraic
structure of this group is inconsistent with a solution by radicals.
Galois worked in several other areas of mathematics, making equally
profound discoveries. In particular he generalized modular arithmetic to classify
what we now call Galois fields. These are finite systems in which the
arithmetical operations of addition, subtraction, multiplication and division can
be defined, and all the usual laws apply. The size of a Galois field is always a
power of a prime, and there is exactly one such field for each prime power.

Jordan

The concept of a group first emerged in a clear form in the work of Galois,
though with earlier hints in Ruffini’s epic writings and the elegant researches of
Lagrange. Within a decade of Galois’s ideas becoming widely available, thanks
to Liouville, mathematics was in possession of a well-developed theory of
groups. The main architect of this theory was Camille Jordan, whose 667-page
work Traité de Substitutions et des Équations Algébriques was published in
1870. Jordan developed the entire subject in a systematic and comprehensive
way.
Jordan’s involvement with group theory began in 1867, when he exhibited the
deep link with geometry in a very explicit manner, by classifying the basic types
of motion of a rigid body in Euclidean space. More importantly, he made a very
good attempt to classify how these motions could be combined into groups. His
main motivation was the crystallographic research of Auguste Bravais, who
initiated the mathematical study of crystal symmetries, especially the underlying
atomic lattice. Jordan’s papers generalized the work of Bravais. He announced
his classification in 1867, and published details in 1868–9.
Technically, Jordan dealt only with closed groups, in which the limit of any
sequence of motions in the group is also a motion in the same group. These
include all finite groups, for trivial reasons, and also groups like all rotations of a
circle about its centre. A typical example of a non-closed group, not considered
by Jordan, might be all rotations of a circle about its centre through rational
multiples of 360°. This group exists, but does not satisfy the limit property
(because, for example, it fails to include rotation by 360 × degrees, since is
not rational). The non-closed groups of motions are enormously varied and
almost certainly beyond any sensible classification. The closed ones are
tractable, but difficult.
The main rigid motions in the plane are translations, rotations, reflections and
glide reflections. In three-dimensional space, we also encounter screw motions,
like the movement of a corkscrew: the object translates along a fixed axis and
simultaneously rotates about the same axis.
Jordan began with groups of translations, and listed ten types, all mixtures of
continuous translations (by any distance) in some directions and discrete
translations (by integer multiples of a fixed distance) in other directions. He also
listed the main finite groups of rotations and reflections: cyclic, dihedral,
tetrahedral, octahedral and icosahedral. He distinguished the group O(2) of all
rotations and reflections that leave a line in space, the axis, fixed, and the group
O(3) of all rotations and reflections that leave a point in space, the centre, fixed.

‘But his work was a major step towards the

understanding of Euclidean rigid motions. . .’

Later it became clear that his list was incomplete. For instance, he had missed
out some of the subtler crystallographic groups in three-dimensional space. But
his work was a major step towards the understanding of Euclidean rigid motions,
which are important in mechanics, as well as in the main body of pure
mathematics.
 Jordan’s book is truly vast in scope. It begins with modular arithmetic and
Galois fields, which as well as providing examples of groups also constitute
essential background for everything else in the book. The middle third deals with
groups of permutations, which Jordan calls substitutions. He sets up the basic
ideas of normal subgroups which are what Galois used to show that the
symmetry group of the quintic is inconsistent with a solution by radicals, and
proves that these subgroups can be used to break a general group into simpler
pieces. He proves that the sizes of these pieces do not depend on how the group
is broken up. In 1889 Otto Hölder improved this result, interpreting the pieces as
groups in their own right, and proved that their group structure, not just their
size, is independent of how the group is broken up. Today this result is called the
Jordan–Hölder Theorem.
A group is simple if it does not break up in this way. The Jordan–Hölder
Theorem effectively tells us that the simple groups relate to general groups in the
same way that atoms relate to molecules in chemistry. Simple groups are the
atomic constituents of all groups. Jordan proved that the alternating group An,
comprising all permutations of n symbols that switch an even number of pairs of
symbols, is simple whenever n ≥ 5. This is the main group-theoretic reason why
the quintic is insoluble by radicals.
A major new development was Jordan’s theory of linear substitutions. Here
the transformations that make up the group are not permutations of a finite set,
but linear changes to a finite list of variables. For example, three variables x, y, z
might transform into new variables X, Y, Z by way of linear equations

X = a1x + a2y + a3z

Y = b1x + b2y + b3z
Z = c1x + c2y + c3z

where the a’s, b’s and c’s are constants. To make the group finite, Jordan usually
took these constants to be elements of the integers modulo some prime, or more
generally a Galois field.
Also in 1869, Jordan developed his own version of Galois theory and included
it in the Traité. He proved that an equation is soluble if and only if its group is
soluble, which means that the simple components all have prime order. He
applied Galois’s theory to geometric problems.
Symmetry

The 4000-year-old quest to solve quintic algebraic equations was brought to an

abrupt halt when Ruffini, Abel and Galois proved that no solution by radicals is
possible. Although this was a negative result, it had a huge influence on the
subsequent development of both mathematics and science. This happened
because the method introduced to prove the impossibility turned out to be central
to the mathematical understanding of symmetry, and symmetry turned out to be
vital in both mathematics and science.
The effects were profound. Group theory led to a more abstract view of
algebra, and with it a more abstract view of mathematics. Although many
practical scientists initially opposed the move towards abstraction, it eventually
became clear that abstract methods are often more powerful than concrete ones,
and most opposition has disappeared. Group theory also made it clear that
negative results may still be important, and that an insistence on proof can
sometimes lead to major discoveries. Suppose that mathematicians had simply
assumed without proof that quintics cannot be solved, on the plausible grounds
that no one could find a solution. Then no one would have invented group theory
to explain why they cannot be solved. If mathematicians had taken the easy
route, and assumed the solution to be impossible, mathematics and science
would have been a pale shadow of what they are today.

What group theory does for us

Group theory is now indispensable throughout mathematics, and its use in science is
widespread. In particular, it turns up in theories of pattern formation in many different
scientific contexts.
One example is the theory of reaction–diffusion equations, introduced by Alan Turing
in 1952 as a possible explanation of symmetric patterns in the markings of animals. In
these equations, a system of chemicals can diffuse across a region of space, and the
chemicals can also react to produce new chemicals. Turing suggested that some such
process might set up a pre-pattern in a developing animal embryo, which later on could
be turned into pigments, revealing the pattern in the adult.
Suppose for simplicity that the region is a plane. Then the equations are symmetric
under all rigid motions. The only solution of the equations that is symmetric under all
rigid motions is a uniform state, the same everywhere. This would translate into an
 animal without any specific markings, the same colour all over. However, the uniform
state may be unstable, in which case the actual solution observed will be symmetric
under some rigid motions but not others. This process is called symmetry-breaking.
A typical symmetry-breaking pattern in the plane consists of parallel stripes. Another is
a regular array of spots. More complicated patterns are also possible. Interestingly,
spots and stripes are among the commonest patterns in animal markings, and many of
the more complicated mathematical patterns are also found in animals. The actual
biological process, involving genetic effects, must be more complicated than Turing
assumed, but the underlying mechanism of symmetry-breaking must be mathematically
very similar.

A mathematical model and a fish both showing Turing patterns

That is why mathematicians insist on proofs.

 14
Algebra Comes of Age

Numbers give way to structures

By 1860 the theory of permutation groups was well developed. The theory of
invariants – algebraic expressions that do not change when certain changes of
variable are performed – had drawn attention to various infinite sets of
transformations, such as the projective group of all projections of space. In 1868
Camille Jordan had studied groups of motions in three-dimensional space, and
the two strands began to merge.

Sophisticated concepts

A new kind of algebra began to appear, in which the objects of study were not
unknown numbers, but more sophisticated concepts: permutations,
transformations, matrices. Last year’s processes had become this year’s things.
The long-standing rules of algebra often had to be modified to fit the needs of
these new structures. Alongside groups, mathematicians started to study
structures called rings and fields, and a variety of algebras.
One stimulus to this changing vision of algebras came from partial differential
equations, mechanics and geometry: the development of Lie groups and Lie
algebras. Another source of inspiration was number theory: here algebraic
numbers could be used to solve Diophantine equations, understand reciprocity
laws and even to attack Fermat’s Last Theorem. Indeed, the culmination of such
efforts was the proof of Fermat’s Last Theorem by Andrew Wiles in 1995.

Lie and Klein

In 1869 the Norwegian mathematician Sophus Lie became friendly with the
Prussian mathematician Klein. They had a common interest in line geometry, an
offshoot of projective geometry introduced by Julius Plücker. Lie conceived a
highly original idea: that Galois’s theory of algebraic equations should have an
analogue for differential equations. An algebraic equation can be solved by
radicals only if has the right kind of symmetries – that is, it has a soluble Galois
group. Analogously, Lie suggested, a differential equation can be solved by
classical methods only when the equation remains unchanged by a continuous
family of transformations. Lie and Klein worked on variations of this idea
during1869–1870; it culminated in 1872 in Klein’s characterization of geometry
as the invariants of a group, laid down in his Erlangen programme.
This programme grew out of a new way of thinking about Euclidean
geometry, in terms of its symmetries. Jordan had already pointed out that the
symmetries of the Euclidean plane are rigid motions of several kinds:
translations, which slide the plane in some direction; rotations, which turn it
about some fixed point; reflections, which flip it over about a fixed line; and,
less obviously, glide reflections, which reflect it and then translate it in a
direction perpendicular to the mirror line. These transformations form a group,
the Euclidean group, and they are rigid in the sense that they do not change
distances. Therefore they also do not change angles. Now lengths and angles are
the basic concepts of Euclid’s geometry. So Klein realized that these concepts
are the invariants of the Euclidean group, the quantities that do not change when
a transformation from the group is applied.
 Felix Klein
1849–1925

Klein was born in Düsseldorf to an upper-class family – his father was secretary to the head of
the Prussian government. He went to Bonn University, planning to become a physicist, but he
became laboratory assistant to Julius Plücker. Plücker was supposed to be working in
mathematics and experimental physics, but his interests had focused on geometry, and Klein fell
under his influence. Klein’s 1868 thesis was on line geometry as applied to mechanics.

By 1870 he was working with Lie on group theory and differential geometry. In 1871 he
discovered that non-Euclidean geometry is the geometry of a projective surface with a
distinguished conic section. This fact proved, very directly and obviously, that non-Euclidean
geometry is logically consistent if Euclidean geometry is. This pretty much ended the controversy
over the status of non-Euclidean geometry.
In 1872 Klein became professor at Erlangen, and in his Erlangen programme of 1872 he
unified almost all known types of geometry, and clarified links between them, by considering
geometry as the invariants of a transformation group. Geometry thereby became a branch of
group theory. He wrote this article for his inaugural address, but did not actually present it on that
occasion. Finding Erlangen uncongenial, he moved to Munich in 1875. He married Anne Hegel,
granddaughter of the famous philosopher. Five years later he went to Leipzig, where he
blossomed mathematically.
Klein believed that his best work was in the theory of complex functions, where he made deep
studies of functions invariant under various groups of transformations of the complex plane. In
 particular he developed the theory of the simple group of order 168 in this context. He engaged in
rivalry with Poincaré to solve the uniformization problem for complex functions, but his health
collapsed, possibly because of the strenuous effort involved.

In 1886 Klein was made professor at the University of Göttingen, and concentrated on
administration, building one of the best schools of mathematics in the world. He remained there
until retirement in 1913.

In fact, if you know the Euclidean group, you can deduce its invariants, and
from these you get Euclidean geometry.
The same goes for other kinds of geometry. Elliptic geometry is the study of
the invariants of the group of rigid motions in a positively curved space,
hyperbolic geometry is the study of the invariants of the group of rigid motions
in a negatively curved space, and projective geometry is the study of the
invariants of the group of projections and so on. Just as coordinates relate
algebra to geometry, so invariants relate group theory to geometry. Each
geometry defines a corresponding group, the group of all transformations that
preserve the relevant geometric concepts. Conversely, every group of
transformations defines a corresponding geometry, that of the invariants.
Klein used this correspondence to prove that certain geometries were
essentially the same as others, because their groups were identical except for
interpretation. The deeper message is that any geometry is defined by its
symmetries. There is one exception: Riemann’s geometry of surfaces, the
curvature of which can change from point to point. It does not quite fit into
Klein’s programme.

Lie groups

Lie and Klein’s joint research led Lie to introduce one of the most important
ideas in modern mathematics, that of a continuous transformation group, now
known as a Lie group. It is a concept that has revolutionized both mathematics
and physics, because Lie groups capture many of the most significant
symmetries of the physical universe, and symmetry is a powerful organizing
principle – both for the underlying philosophy of how we represent nature
mathematically, and for technical calculations.
 Sophus Lie created the theory of Lie groups in a flurry of activity beginning in
the autumn of 1873. The concept of a Lie group has evolved considerably since
Lie’s early work. In modern terms, a Lie group is a structure having both
algebraic and topological properties, the two being related. Specifically, it is a
group (a set with an operation of composition that satisfies various algebraic
identities, most notably the associative law) and a topological manifold (a space
that locally resembles Euclidean space of some fixed dimension but which may
be curved or otherwise distorted on the global level), such that the law of
composition is continuous (small changes in the elements being composed
produce small changes in the result). Lie’s concept was more concrete: a group
of continuous transformations in many variables. He was led to study such
transformation groups while seeking a theory of the solubility or insolubility of
differential equations, analogous to that of Évariste Galois for algebraic
equations, but today they arise in an enormous variety of mathematical contexts,
and Lie’s original motivation is not the most important application.

‘Lie’s original motivation is not the most important

application.’

Perhaps the simplest example of a Lie group is the set of all rotations of a
circle. Each rotation is uniquely determined by an angle between 0° and 360°.
The set is a group because the composition of two rotations is a rotation –
through the sum of the corresponding angles. It is a manifold of dimension one,
because angles correspond one-to-one with points on a circle, and small arcs of a
circle are just slightly bent line segments, a line being Euclidean space of
dimension one. Finally, the composition law is continuous because small
changes in the angles being added produce small changes in their sum.
A more challenging example is the group of all rotations of three-dimensional
space that preserve a chosen origin. Each rotation is determined by an axis – a
line through the origin in an arbitrary direction – and an angle of rotation about
that axis. It takes two variables to determine an axis (say the latitude and
longitude of the point in which it meets a reference sphere centred on the origin)
and a third to determine the angle of rotation; therefore this group has dimension
three. Unlike the group of rotations of a circle, it is non-commutative – the result
of combining two transformations depends upon the order in which they are
performed.
 In 1873, after a detour into PDEs, Lie returned to transformation groups,
investigating properties of infinitesimal transformations. He showed that
infinitesimal transformations derived from a continuous group are not closed
under composition, but they are closed under a new operation known as the
bracket, written [x,y]. In matrix notation this is the commutator xy yx of x and y.
The resulting algebraic structure is now known as a Lie algebra. Until about
1930 the terms Lie group and Lie algebra were not used: instead these concepts
were referred to as continuous group and infinitesimal group respectively.
There are strong interconnections between the structure of a Lie group and
that of its Lie algebra, which Lie expounded in a three-volume work Theorie der
Transformationsgruppen (Theory of Transformation Groups) written jointly
with Friedrich Engel. They discussed in detail four classical families of groups,
two of which are the rotation groups in n-dimensional space for odd or even n.
The two cases are rather different, which is why they are distinguished. For
example, in odd dimensions a rotation always possesses a fixed axis; in even
dimensions it does not.

Killing

The next really substantial development was made by Wilhelm Killing. In 1888
Killing laid the foundations of a structure theory for Lie algebras, and in
particular he classified all the simple Lie algebras, the basic building blocks out
of which all other Lie algebras are composed. Killing started from the known
structure of the most straightforward simple Lie algebras, the special linear Lie
algebras sl(n), for n ≥ 2. Start with all n × n matrices with complex entries, and
let the Lie bracket of two matrices A and B be AB – BA. This Lie algebra is not
simple, but the sub-algebra, sl(n), of all matrices whose diagonal terms sum to
zero, is simple. It has dimension n2 − 1.
Lie knew the structure of this algebra, and he showed that any simple Lie
algebra had a similar kind of structure. It is remarkable that he could prove
something so specific, starting only with the knowledge that the Lie algebra is
simple. His method was to associate to each simple Lie algebra a geometric
structure known as a root system. He used methods of linear algebra to study and
classify root systems, and then derived the structure of the corresponding Lie
algebra from that of the root system. So classifying the possible root system
geometries is effectively the same as classifying the simple Lie algebras.
 ‘The upshot of Killing’s work is remarkable.’

The upshot of Killing’s work is remarkable. He proved that the simple Lie
algebras fall into four infinite families, now called An, Bn, Cn and Dn.
Additionally, there were five exceptions: G2, F4, E6, E7 and E8. Killing actually
thought there were six exceptions, but two of them turned out to be the same
algebra in two different guises. The dimensions of the exceptional Lie algebras
are 12, 56, 78, 133 and 248. They remain a little mysterious, although we now
understand fairly clearly why they exist.

Simple Lie groups

Because of the close connections between a Lie group and its Lie algebra, the
classification of simple Lie algebras also led to a classification of the simple Lie
groups. In particular the four families An, Bn, Cn and Dn are the Lie algebras of
the four classical families of transformation groups. These are, respectively, the
group of all linear transformations in (n + 1)-dimensional space, the rotation
group in (2n + 1)-dimensional space, the symplectic group in 2n dimensions,
which is important in classical and quantum mechanics and optics, and the
rotation group in 2n-dimensional space. A few finishing touches to this story
were added later; notably the introduction by Harold Scott MacDonald Coxeter
and Eugene (Evgenii) Dynkin of a graphical approach to the combinatorial
analysis of root systems, now known as Coxeter or Dynkin diagrams.
Lie groups are important in modern mathematics for many reasons. For
example, in mechanics, many systems have symmetries, and those symmetries
make it possible to find solutions of the dynamical equations. The symmetries
generally form a Lie group. In mathematical physics, the study of elementary
particles relies heavily upon the apparatus of Lie groups, again because of
certain symmetry principles. Killing’s exceptional group E8 plays an important
role in superstring theory, an important current approach to the unification of
quantum mechanics and general relativity. Simon Donaldson’s epic discovery of
1983 that four-dimensional Euclidean space possesses non-standard
differentiable structures rests, fundamentally, on an unusual feature of the Lie
group of all rotations in four-dimensional space. The theory of Lie groups is vital
to the whole of modern mathematics.
Abstract groups

In Klein’s Erlangen programme it is essential that the groups concerned consist

of transformations; that is, the elements of the group act on some space. Much of
the early work on groups assumed this structure. But further research required
one extra piece of abstraction: to retain the group property but throw away the
space. A group consisted of mathematical entities that could be combined to
yield similar entities, but those entities did not have to be transformations.
Numbers are an example. Two numbers (integer, rational, real, complex) can
be added, and the result is a number of the same kind. Numbers form a group
under the operation of addition. But numbers are not transformations. So even
though the role of groups as transformations had unified geometry, the
assumption of an underlying space had to be thrown away to unify group theory.
Among the first to come close to taking this step was Arthur Cayley, in three
papers of 1849 and 1854. Here Cayley said that a group comprises a set of
operators 1, a, b, c and so on. The compound ab of any two operators must be
another operator; the special operator 1 satisfies 1a = a and a1 = a for all
operators a; finally, the associative law (ab)c = a(bc) must hold. But his
operators still operated on something (a set of variables). Additionally, he had
omitted a crucial property: that every a must have an inverse a′ such that a′a =.
aa' = 1. So Cayley came close, but missed the prize by a whisker.
In 1858 Richard Dedekind allowed the group elements to be arbitrary entities,
not just transformations or operators, but he included the commutative law ab =
ba in his definition. His idea was fine for its intended purpose, number theory,
but it excluded most of the interesting groups in Galois theory, let alone the
wider mathematical world. The modern concept of an abstract group was
introduced by Walther van Dyck in 1882–3. He included the existence of an
inverse, but rejected the need for the commutative law. Fully-fledged axiomatic
treatments of groups were provided soon after, by Edward Huntington and
Eliakim Moore in 1902, and by Leonard Dickson in 1905.
With the abstract structure of groups now separated from any specific
interpretation, the subject developed rapidly. Early research consisted mostly of
‘butterfly collecting’ – people studied individual examples of groups, or special
types, looking for common patterns. The main concepts and techniques appeared
relatively quickly, and the subject thrived.
Number theory

Another major source of new algebraic concepts was number theory. Gauss
started the process when he introduced what we now call Gaussian integers.
These are complex numbers a + bi, where a and b are integers. Sums and
products of such numbers also have the same form. Gauss discovered that the
concept of a prime number generalizes to Gaussian integers. A Gaussian integer
is prime if it cannot be expressed as a product of other Gaussian integers in a
non-trivial way. Prime factorization for Gaussian integers is unique. Some
ordinary primes, such as 3 and 7, remain prime when considered as Gaussian
integers, but others do not: for example 5 = (2 + i)(2 – i). This fact is closely
connected with Fermat’s theorem about primes and sums of two squares, and
Gaussian integers illuminate that theorem and its relatives.

‘Gauss started the process when he introduced what

we now call Gaussian integers.’

If we divide one Gaussian integer by another, the result need not be a

Gaussian integer, but it comes close: it is of the form a + bi, where a and b are
rational. These are the Gaussian numbers. More generally, number theorists
discovered that something similar holds if we take any polynomial p(x) with
integer coefficients, and then consider all linear combinations a1x1 +. . .+ anxn of
its solutions x1, . . ., xn. Taking a1, . . ., an to be rational, we obtain a system of
complex numbers that is closed under addition, subtraction, multiplication and
division – meaning that when these operations are applied to such a number, the
result is a number of the same kind. This system constitutes an algebraic number
field. If instead we require a1, . . ., an, to be integers, the system is closed under
addition, subtraction and multiplication, but not division: it is an algebraic
number ring.
The most ambitious application of these new number systems was Fermat’s
Last Theorem: the statement that the Fermat equation, xn + yn = zn, has no whole
number solutions when the power n is three or more. Nobody could reconstruct
Fermat’s alleged ‘remarkable proof’ and it seemed increasingly doubtful that he
had ever possessed one. However, some progress was made. Fermat found
proofs for cubes and fourth powers, Peter Lejeune-Dirichlet dealt with fifth
powers in 1828 and Henri Lebesgue found a proof for seventh powers in 1840.
 In 1847 Gabriel Lamé claimed a proof for all powers, but Ernst Eduard
Kummer pointed out a mistake. Lamé had assumed without proof that
uniqueness of prime factorization is valid for algebraic numbers, but this is false
for some (indeed most) algebraic number fields. Kummer showed that
uniqueness fails for the algebraic number field that arises in the study of
Fermat’s Last Theorem for 23rd powers. But Kummer did not give up easily,
and he found a way round this obstacle by inventing some new mathematical
gadgetry, the theory of ideal numbers. By 1847 he had disposed of Fermat’s Last
Theorem for all powers up to 100, except for 37, 59 and 67. By developing extra
machinery, Kummer and Dimitri Mirimanoff disposed of those cases too in
1857. By the 1980s similar methods had proved all cases up to the 150,000th
power, but the method was running out of steam.

Rings, fields and algebras

Kummer’s notion of an ideal number was cumbersome, and Dedekind

reformulated it in terms of ideals, special subsystems of algebraic integers. In the
hands of Hilbert’s school at Göttingen, and in particular Emmy Noether, the
entire area was placed on an axiomatic footing. Alongside groups, three other
types of algebraic system were defined by suitable lists of axioms: rings, fields
and algebras.

‘In 1847 Gabriel Lamé claimed a proof for all

powers, but Ernst Eduard Kummer pointed out a
mistake.’

In a ring, operations of addition, subtraction and multiplication are defined,

and they satisfy all the usual laws of algebra except for the commutative law of
multiplication. If this law also holds, we have a commutative ring.
In a field, operations of addition, subtraction, multiplication and division are
defined, and they satisfy all the usual laws of algebra including the commutative
law of multiplication. If this law fails, we have a division ring.

Emmy Amalie Noether

1882–1935
 1882–1935

Emmy Noether was the daughter of the mathematician Max Noether and Ida Kaufmann, both of
Jewish origin. In 1900 she qualified to teach languages but instead decided her future lay in
mathematics. At that time German universities allowed women to study courses unofficially if the
professor gave permission, and she did this from 1900 to 1902. Then she went to Göttingen,
attending lectures by Hilbert, Klein and Minkowski in 1903 and 1904.
She gained a doctorate in 1907, under the invariant theorist Paul Gordan. Her thesis calculated
a very complicated system of invariants. For men, the next step would be Habilitation, but this
was not permitted for women. She stayed home in Erlangen, helping her disabled father, but she
continued her research and her reputation quickly grew.

In 1915 she was invited back to Göttingen by Klein and Hilbert, who struggled to get the rules
changed to allow her on to the faculty. They finally succeeded in 1919. Soon after her arrival she
proved a fundamental theorem, often called Noether’s Theorem, relating the symmetries of a
physical system to conservation laws. Some of her work was used by Einstein to formulate parts
of general relativity. In 1921 she wrote a paper on ring theory and ideals, taking an abstract
axiomatic view. Her work formed a significant part of Bartel Leendert van der Waerden’s classic
text Moderne Algebra.

When Germany fell under Nazi rule she was dismissed because she was Jewish, and she left
Germany to take up a position in the USA. Van der Waerden said that for her, ‘relationships
among numbers, functions and operations became transparent, amenable to generalization and
productive only after they have been . . . reduced to general conceptual relationships.’
 An algebra is like a ring, but its elements can also be multiplied by various
constants, the real numbers, complex numbers or – in the most general setting –
by a field. The laws of addition are the usual ones, but the multiplication may
satisfy a variety of different axioms. If it is associative, we have an associative
algebra. If it satisfies some laws related to the commutator xy − yx, it is a Lie
algebra.
There are dozens, maybe hundreds of different types of algebraic structure,
each with its own list of axioms. Some have been invented just to explore the
consequences of interesting axioms, but most arose because they were needed in
some specific problem.

Finite simple groups

The high point of 20th century research on finite groups was the successful
classification of all finite simple groups. This achieved for finite groups what
Killing had achieved for Lie groups and their Lie algebras. Namely, it led to a
complete description of all possible basic building blocks for finite groups, the
simple groups. If groups are molecules, the simple groups are their constituent
atoms.
Killing’s classification of simple Lie groups proved that these must belong to
one of four infinite families An, Bn, Cn and Dn, with exactly five exceptions, G2,
F4, E6, E7 and E8. The eventual classification of all finite simple groups was
achieved by too many mathematicians to mention individually, but the overall
programme for solving this problem was due to Daniel Gorenstein. The answer,
published in 1888–90, is strangely similar: a list of infinite families, and a list of
exceptions. But now there are many more families, and the exceptions number
26.

Andrew Wiles
1953 –
 Andrew Wiles was born in 1953 in Cambridge, England. At the age of 10 he read about Fermat’s
Last Theorem and resolved to become a mathematician and prove it. By the time of his PhD he
had pretty much abandoned this idea, because the theorem seemed so intractable, so he worked
on the number theory of ‘elliptic curves’, an apparently different area. He moved to the USA and
became a Professor at Princeton.

By the 1980s it was becoming clear that there might be an unexpected link between Fermat’s
Last Theorem and a deep and difficult question about elliptic curves. Gerhard Frey made this link
explicit, by means of the so-called Taniyama–Shimura conjecture. When Wiles heard of Frey’s
idea he stopped all of his other work to concentrate on Fermat’s Last Theorem, and after seven
years of solitary research he convinced himself that he had found a proof, based on a special case
of the Taniyama–Shimura Conjecture. This proof turned out to have a gap, but Wiles and Richard
Taylor repaired the gap and a complete proof was published in 1995.

Other mathematicians soon extended the ideas to prove the full Taniyama–Shimura Conjecture,
pushing the new methods further. Wiles received many honours for his proof, including the Wolf
Prize. In 1998, being just too old for a Fields Medal, traditionally limited to people under 40, he
was awarded a special silver plaque by the International Mathematical Union. He was made a
Knight Commander of the Order of the British Empire in 2000.

The families comprise the alternating groups (known to Galois) and a host of
groups of Lie type which are like the simple Lie groups but over various finite
fields rather than the complex numbers. There are some curious variations on
this theme, too. The exceptions are 26 individuals, with hints of some common
patterns, but no unified structure. The first proof that the classification is
complete came from the combined work of hundreds of mathematicians, and its
total length was around 10,000 pages. Moreover, some crucial parts of the proof
were not published. Recent work by those remaining in this area of research has
involved reworking the classification in a more streamlined manner, an approach
made possible by already knowing the answer. The results are appearing as a
series of textbooks, totalling around 2000 pages.

‘Andrew Wiles, as a child, had dreamed of proving

Fermat’s Last Theorem.’

The most mysterious of the exceptional simple groups, and the largest, is the
monster. Its order is

246×320×59×76×112×133×17×19×23×29×31×41×47×59×71

which equals

808017424794512875886459904961710757005754368000000000

and is roughly 8 × 1053. Its existence was conjectured in 1973 by Bernd Fischer
and Robert Griess. In 1980 Griess proved that it existed, and gave an algebraic
construction as the symmetry group of a 196,884-dimensional algebra. The
monster seems to have some unexpected links with number theory and complex
analysis, stated by John Conway as the Monstrous Moonshine conjecture. This
conjecture was proved by Richard Borcherds in 1992, and he was awarded a
Fields Medal – the most prestigious award in mathematics – for it.

Fermat’s Last Theorem

The application of algebraic number fields to number theory developed apace in

the second half of the 20th century, and made contact with many other areas of
mathematics, including Galois theory and algebraic topology. The culmination
of this work was a proof of Fermat’s Last Theorem, some 350 years after it was
first stated.
The really decisive idea came from a beautiful area that lies at the heart of
modern work on Diophantine equations: the theory of elliptic curves. These are
equations in which a perfect square is equal to a cubic polynomial, and they
represent the one area of Diophantine equations that mathematicians understand
pretty well. However, the subject has its own big unsolved problems. The
biggest of all is the Taniyama–Weil conjecture, named after Yutaka Taniyama
and André Weil. This says that every elliptic curve can be represented in terms
of modular functions – generalizations of trigonometric functions studied in
particular by Klein.

What abstract algebra did for them

In his 1854 book The Laws of Thought, George Boole showed that algebra can be
applied to logic, inventing what is now known as Boolean algebra.
Here we can do no more than convey a flavour of Boole’s ideas. The most important
logical operators are not, and and or. If a statement S is true, then not-S is false, and
vice versa. S and T are true if, and only if, both S and T are true. S or T is true provided
at least one of S, T is true – possibly both. Boole noticed that if we rewrite T as 1 and S
as 0, then the algebra of these logical operations is very similar to ordinary algebra,
provided we think of 0 and 1 as integers modulo 2, so that 1 + 1 = 0 and –S is the same
as S. So not-S is 1+S, S and T is ST, and S or T is S+T+ST. The sum S+T corresponds
to exclusive or (written xor by computer scientists). S xor T is true provided T is true or S
is true, but not both. Boole discovered that his curious algebra of logic is entirely self-
consistent if you bear its slightly weird rules in mind and use them systematically. This
was one of the first steps towards a formal theory of mathematical logic.

Early in the 1980s Gerhard Frey found a link between Fermat’s Last Theorem
and elliptic curves. Suppose that a solution to Fermat’s equation exists; then you
can construct an elliptic curve with very unusual properties – so unusual that the
curve’s existence seems highly improbable. In 1986 Kenneth Ribet made this
idea precise by proving that if the Taniyama–Weil conjecture is true, then Frey’s
curve cannot exist. Therefore the presumed solution of Fermat’s equation cannot
exist either, which would prove Fermat’s Last Theorem. The approach depended
on the Taniyama–Weil conjecture, but it showed that Fermat’s Last Theorem is
not just an isolated historical curiosity. Instead, it lies at the heart of modern
number theory.
Andrew Wiles, as a child, had dreamed of proving Fermat’s Last Theorem,
but when he became a professional he decided that it was just an isolated
problem – unsolved, but not really important. Ribet’s work changed his mind. In
1993 he announced a proof of the Taniyama–Weil conjecture for a special class
of elliptic curves, general enough to prove Fermat’s Last Theorem. But when the
paper was submitted for publication, a serious gap emerged. Wiles had almost
given up when ‘suddenly, totally unexpectedly, I had this incredible
revelation . . . it was so indescribably beautiful, it was so simple and so elegant,
and I just stared in disbelief.’ With the aid of Richard Taylor, he revised the
proof and repaired the gap. His paper was published in 1995.

‘It was not only algebra that became abstract.’

We can be sure that whatever ideas Fermat had in mind when he claimed to
possess a proof of his Last Theorem, they must have been very different from
the methods used by Wiles. Did Fermat really have a simple, clever proof, or
was he deluding himself? It is a puzzle that, unlike his Last Theorem, may never
be resolved.

Abstract mathematics

The move towards a more abstract view of mathematics was a natural

consequence of the growing variety of its subject matter. When mathematics was
mostly about numbers, the symbols of algebra were simply placeholders for
numbers. But as mathematics grew, the symbols themselves started to take on a
life of their own. The meaning of the symbols became less significant than the
rules by which those symbols could be manipulated. Even the rules were not
sacred: the traditional laws of arithmetic, such as the commutative law, were not
always appropriate in new contexts.
It was not only algebra that became abstract. Analysis and geometry also
focused on more general issues, for similar reasons. The main change in
viewpoint occurred from the middle of the 19th century to the middle of the
20th. After that, a period of consolidation set in, as mathematicians tried to
balance the conflicting needs of abstract formalism and applications to science.
Abstraction and generality go hand in hand, but abstraction can also obscure the
meaning of mathematics. But the issue is no longer whether abstraction is useful
or necessary: abstract methods have proved their worth by making it possible to
solve numerous long-standing problems, such as Fermat’s Last Theorem. And
what seemed little more than formal game-playing yesterday may turn out to be
a vital scientific or commercial tool tomorrow.

What abstract algebra does for us

Galois fields form the basis of a coding system that is widely used in a variety of
commercial applications, especially CDs and DVDs. Every time you play music, or watch
a video, you are using abstract algebra.
These methods are known as Reed–Solomon codes, after Irving Reed and Gustave
Solomon who introduced them in 1960. They are error-correcting codes based on a
polynomial, with coefficients in a finite field, constructed from the data being encoded,
such as the music or video signals. It is known that a polynomial of degree n is uniquely
determined by its values at n distinct points. The idea is to calculate the polynomial at
more than n points. If there are no errors, any subset of n data points will reconstruct the
same polynomial. If not, then provided the number of errors is not too large, it is still
possible to deduce the polynomial.
In practice the data are represented as encoded blocks, with 2m − 1 m-bit symbols per
block, where a bit is a binary digit, 0 or 1. A popular choice is m = 8, because many of
the older computers work in bytes – sequences of eight bits. Here the number of
symbols in a block is 255. One common Reed–Solomon code puts 223 bytes of
encoded data in each 255 byte block, using the remaining 32 bytes for parity symbols
which state whether certain combinations of digits in the data should be odd or even.
This code can correct up to 16 errors per block.
 15
Rubber Sheet Geometry

Qualitative beats quantitative

The main ingredients of Euclid’s geometry – lines, angles, circles, squares and
so on – are all related to measurement. Line segments have lengths, angles are a
definite size with 90° differing in important ways from 91° or 89°, circles are
defined in terms of their radii, squares have sides of a given length. The hidden
ingredient that makes all of Euclid’s geometry work is length, a metric quantity,
one which is unchanged by rigid motions and defines Euclid’s equivalent
concept to motion, congruence.

Topology

When mathematicians first stumbled across other types of geometry, these too
were metric. In non-Euclidean geometry, lengths and angles are defined; they
just have different properties from lengths and angles in the Euclidean plane.
The arrival of projective geometry changed this: projective transformations can
change lengths, and they can change angles. Euclidean geometry and the two
main kinds of non-Euclidean geometry are rigid. Projective geometry is more
flexible, but even here subtler invariants exist, and in Klein’s picture what
defines a geometry is a group of transformations and the corresponding
invariants.
As the 19th century approached its end, mathematicians began to develop an
even more flexible kind of geometry; so flexible, in fact, that it is often
characterized as rubber-sheet geometry. More properly known as topology, this
is the geometry of shapes that can be deformed or distorted in extremely
convoluted ways. Lines can bend, shrink or stretch; circles can be squashed so
that they turn into triangles or squares. All that matters here is continuity. The
transformations allowed in topology are required to be continuous in the sense of
analysis; roughly speaking, this means that if two points start sufficiently close
together, they end up close together – hence the ‘rubber sheet’ image.
There is still a hint of metric thinking here: ‘close together’ is a metric
concept. But by the early 20th century, even this hint had been removed, and
topological transformations took on a life of their own. Topology quickly
increased in status, until it occupied centre stage in mathematics – even though
to begin with it seemed very strange, and virtually content-free. With
transformations that flexible, what could be invariant? The answer, it turned out,
was quite a lot. But the type of invariant that began to be uncovered was like
nothing ever before considered in geometry. Connectivity – how many pieces
does this thing have? Holes – is it all one lump, or are there tunnels through it?
Knots – how is it tangled up, and can you undo the tangles? To a topologist, a
doughnut and a coffee-cup are identical (but a doughnut and a tumbler are not);
however, both are different from a round ball. An overhand knot is different
from a figure-eight knot, but proving this fact required a whole new kind of
machinery, and for a long time no one could prove that any knots existed at all.
It seems remarkable that anything so diffuse and plain weird could have any
importance at all. But appearances are deceptive. Continuity is one of the basic
aspects of the natural world, and any deep study of continuity leads to topology.
Even today we mostly use topology indirectly, as one technique among many.
You don’t find anything topological sitting in your kitchen – not obviously, at
least. (However, you can occasionally find such items as a chaotic dishwasher,
which uses the strange dynamics of two rotating arms to clean dishes more
efficiently. And our understanding of the phenomenon of chaos rests on
topology.) The main practical consumers of topology are quantum field theorists
– a new use of the word ‘practical’ perhaps, but an important area of physics.
Another application of topological ideas occurs in molecular biology, where
describing and analysing the twists and turns of the DNA molecule requires
topological concepts.
Behind the scenes, topology informs the whole of mainstream mathematics,
and enables the development of other techniques with more obvious practical
uses. It is a rigorous study of qualitative geometric features, as opposed to
quantitative ones like lengths. This is why mathematicians consider topology to
be of vast importance, whereas the rest of the world has hardly heard of it.

Polyhedra and the Königsberg Bridges

Although topology really only began to take off around 1900, it made an
occasional appearance in earlier mathematics. Two items in the prehistory of
topology were introduced by Euler: his formula for polyhedra and his solution of
the puzzle of the Königsberg Bridges.
In 1639 Descartes had noticed a curious feature of the numerology of regular
solids. Consider, for instance, a cube. This has six faces, 12 edges and eight
vertices. Add 6 to 8 and you get 14, which is 2 larger than 12. How about a
dodecahedron? Now there are 12 faces, 30 edges and 20 vertices. And 2 + 20 =
30, which exceeds 30 by 2. The same goes for the tetrahedron, octahedron and
icosahedron. In fact, the same relation seemed tomm work for almost any
polyhedron. If a solid has F faces, E edges and V vertices, then F + V = E + 2,
which we can rewrite as

F − E + V = 2

Descartes did not publish his discovery, but he wrote it down and his manuscript
was read by Leibniz in 1675.
Euler was the first to publish this relationship, in 1750. He followed this up
with a proof in 1751. He was interested in the relationship because he had been
trying to classify polyhedra. Any general phenomenon such as this one had to be
taken into account when performing such a classification.
 Polyhedron with a hole

Is the formula valid for all polyhedra? Not quite. A polyhedron in the form of
a picture frame, with square cross-section and mitred corners, has 16 faces, 32
edges and 16 vertices, so that here F + V − E = 0. The reason for the discrepancy
turns out to be the presence of a hole. In fact, if a polyhedron has g holes, then

F + V − E = 2 − 2g

Cauchy’s Proof of the Descartes–Euler Formula

Remove one face and stretch the surface of the solid out on a plane. This reduces F by
1, so we now prove that the resulting plane configuration of faces, lines and points has F
− E + V = 1. To achieve this, first convert all faces to triangles by drawing diagonals.
Each new diagonal leaves V unchanged, but increases both E and F by 1, so F − E + V
remains the same as before. Now start deleting edges, starting from the outside. Each
such deletion reduces both F and E, so F − E + V is once more unchanged. When you
run out of faces to delete, you are left with a tree of edges and vertices, containing no
closed loops. One by one, delete terminal vertices, along with the edge that meets them.
Now E and V both decrease by 1, and again F − E + V is unchanged. Eventually this
process stops with one solitary vertex. Now F = 0, E = 0 and V = 1, so F − E + V = 1, as
required.
 Example of Cauchy’s proof

What, exactly, is a hole? This question is harder than it looks. First, we are
talking about the surface of the polyhedron, not its solid interior. In real life we
make a hole in something by burrowing through its solid interior, but the above
formulas make no reference to the interior of the polyhedron – just to the faces
that constitute its surfaces, along with their edges and vertices. Everything we
count lies on the surface here. Second, the only holes that change the
numerology are those that burrow all the way through the polyhedron – tunnels
with two ends, so to speak, not holes like workmen dig in the road. Third, such
holes are not in the surface at all, though they are somehow delineated by it.
When you buy a doughnut, you also buy its hole, but you can’t buy a hole in its
own right. It exists only by virtue of the doughnut, even though in this case you
also buy the doughnut’s solid interior.
It is easier to define what ‘no holes’ means. A polyhedron has no holes if it
can be continuously deformed, creating curved faces and edges, so that it
becomes (the surface of) a sphere. For these surfaces, F + V − E really is always
2. And the converse holds as well: if F + V − E = 2 then the polyhedron can be
deformed into a sphere.
The picture-frame polyhedron does not look as though it can be deformed into
a sphere – where would the hole go? For a rigorous proof of the impossibility,
we need look no further than the fact that for this polyhedron, F + V − E = 0.
This relationship is impossible for surfaces deformable to spheres. So the
numerology of polyhedra tells us significant features of their geometry, and
those features can be topological invariants – unchanged by deformations.
Euler’s formula is now viewed as a significant hint of a useful link between
combinatorial aspects of polyhedra, such as numbers of faces, and topological
aspects. In fact, it turns out to be easier to work backwards. To find out how
many holes a surface has, work out F + V − E − 2, divide by 2, and change sign:
 g = − (F + V – E − 2)/2

A curious consequence: we can now define how many holes a polyhedron has,
without defining ‘hole’.
An advantage of this procedure is that it is intrinsic to the polyhedron. It does
not involve visualizing the polyhedron in a surrounding three-dimensional space,
which is how our eyes naturally see the hole. A sufficiently intelligent ant that
lived on the surface of the polyhedron could work out that it has one hole, even
if all it could see was the surface. This intrinsic point of view is natural in
topology. It studies the shapes of things in their own right, not as part of
something else.
At first sight, the problem of the Königsberg Bridges bears no relationship to
the combinatorics of polyhedra. The city of Königsberg, then in Prussia, was
situated on both banks of the river Pregelarme, in which there were two islands.
The islands were linked to the banks, and to each other, by seven bridges.
Apparently, the citizens of Königsberg had long wondered whether it was
possible to take a Sunday stroll that would cross each bridge exactly once.

Problem of the Königsberg Bridges C

In 1735 Euler solved the puzzle; rather, he proved that there is no solution,
and explained why. He made two significant contributions: he simplified the
problem and reduced it to its bare essentials, and then he generalized it, to deal
with all puzzles of a similar kind. He pointed out that what matters is not the size
and shape of the islands, but how the islands, the banks and the bridges are
connected. The whole problem can be reduced to a simple diagram of dots
(vertices) joined by lines (edges), here shown superimposed on a map.
To form this diagram, place one vertex on each land-mass – north bank, south
bank, and the two islands. Join two vertices by an edge whenever a bridge exists
that links the corresponding land masses. Here we end up with four vertices A,
B, C, D and seven edges, one for each bridge.
The puzzle is then equivalent to a simpler one about the diagram. Is it possible
to find a path – a connected sequence of edges – that includes each edge exactly
once?
Euler distinguished two types of path: an open tour, which starts and ends at
different vertices, and a closed tour, which starts and ends at the same vertex. He
proved that for this particular diagram neither kind of tour exists.
The key to the puzzle is to consider the valency of each vertex, that is, how
many edges meet at that vertex. First, think of a closed tour. Here, every edge
where the tour enters a vertex is matched by another, the next edge, along which
the tour leaves that vertex. If a closed tour is possible, then the number of edges
at any given vertex must therefore be even. In short, every vertex must have
even valency. But the diagram has three vertices of valency 3 and one of valency
5 – all odd numbers. Therefore no closed tour exists.
A similar criterion applies to open tours, but now there will be exactly two
vertices of odd valency: one at the start of the tour, the other at its end. Since the
Königsberg diagram has four vertices of odd valency, there is no open tour
either.
Euler went one step further: he proved that these necessary conditions for the
existence of a tour are also sufficient, provided the diagram is connected (any
two vertices are linked by some path). This general fact is a little trickier to
prove, and Euler spends some time setting his proof up. Nowadays we can give a
proof in a few lines.

Geometric properties of plane surfaces

Euler’s two discoveries seem to belong to entirely different areas of

mathematics, but on closer inspection they have elements in common. They are
both about the combinatorics of polyhedral diagrams. One counts faces, edges
and vertices and the other counts valencies; one is about a universal relation
between three numbers, the other a relation that must occur if there exists a tour.
But they are visibly similar in spirit. More deeply – and this went unappreciated
for more than a century – both are invariant under continuous transformations.
The positions of the vertices and edges do not matter: what counts is how they
connect to each other. Both problems would look the same if the diagrams were
drawn on a rubber sheet and the sheet were distorted. The only way to create
significant differences would be to cut or tear the sheet, or glue bits together –
but these operations destroy continuity.

The Möbius Band

Topology has some surprises. The best known is the Möbius band (or Möbius strip), which can be
formed by taking a long strip of paper, and joining the ends with a half-twist. Without this twist,
we get a cylinder. The difference between these two surfaces becomes apparent if we try to paint
them. We can paint the outer surface of a cylinder red and the inner surface blue. But if you start
painting a Möbius band red on one side, and keep going until every part of the surface that
connects to the red region has been covered, you end up coating the entire band in red paint. The
inside surface connects to the outside thanks to that half-twist.

Another difference appears if you cut along the centre line of the band. It falls into two pieces
but they remain connected.

The glimmerings of a general theory were apparent to Gauss, who from time
to time made quite a lot of fuss about the need for some theory of the basic
geometric properties of diagrams. He also developed a new topological
invariant, which we now call the linking number, in work on magnetism. This
number determines how one closed curve winds round another. Gauss gave a
formula to calculate the linking number from analytic expressions for the curves.
A similar invariant, the winding number of a closed curve relative to a point, was
implicit in one of his proofs of the Fundamental Theorem of Algebra.
Gauss’s main influence on the development of topology came from one of his
students, Johann Listing, and his assistant Augustus Möbius. Listing studied
under Gauss in 1834, and his work Vorstudien zur Topologie introduced the
word topology. Listing himself would have preferred to call the subject
geometry of position, but this phrase had been pre-empted by Karl von Staudt to
mean projective geometry, so Listing found another word. Among other things,
Listing looked for generalizations of Euler’s formula for polyhedra.
It was Möbius who made the role of continuous transformations explicit.
Möbius was not the most productive of mathematicians, but he tended to think
everything through very carefully and very thoroughly. In particular, he noticed
that surfaces do not always have two distinct sides, giving the celebrated Möbius
band as an example. This surface was discovered independently by Möbius and
Listing in 1858. Listing published it in Der Census Räumlicher Complexe, and
Möbius put it in a paper on surfaces.
For a long time Euler’s ideas on polyhedra were something of a side-issue in
mathematics, but several prominent mathematicians began to glimpse a new
approach to geometry, which they called ‘analysis situs’ – the analysis of
position. What they had in mind was a qualitative theory of shape, in its own
right, to supplement the more traditional quantitative theory of lengths, angles,
areas and volumes. This view began to gain ground when issues of that kind
emerged from traditional investigations in mainstream mathematics. A key step
was the discovery of connections between complex analysis and the geometry of
surfaces, and the innovator was Riemann.

The Riemann sphere

The obvious way to think of a complex function f is to interpret it as a mapping

from one complex plane to another. The basic formula w = f(x) for such a
function tells us to take any complex number z, apply f to it and deduce another
complex number w associated with z. Geometrically, z belongs to the complex
plane, and w belongs to what is in effect a second, independent copy of the
complex plane.
 However, this point of view turns out not to be the most useful one, and the
reason is singularities. Complex functions often have interesting points at which
their normal, comfortable behaviour goes horribly wrong. For example, the
function f(z) = 1/z is well behaved for all z except zero. When z = 0, the value of
the function is 1/0, which makes no sense as an ordinary complex number, but
can with some stretch of the imagination be thought of as infinity (symbol ∞).
Specifically, if z gets very close to 0, then 1/z gets very big. Infinity in this sense
is not a number, but a term that describes a numerical process: become as large
as you wish. Gauss had already noticed that infinities of this type create new
types of behaviour in complex integration. They mattered.
Riemann found it useful to include ∞ among the complex numbers, and he
found a beautiful geometric way to do this. Place a unit sphere so that it sits on
top of the complex plane. Now associate points in the plane with points on the
sphere by stereographic projection. That is, join the point in the plane to the
North Pole of the sphere, and see where that line meets the sphere.

The Riemann sphere and the complex plane

This construction is called the Riemann sphere. The new point at infinity is
the North Pole of the sphere – the only point that does not correspond to a point
in the complex plane. Amazingly, this construction fits beautifully into the
standard calculations in complex analysis, and now equations like 1/0 = ∞ make
perfect sense. Points at which a complex function f takes the value ∞ are called
poles, and it turns out that you can learn a great deal about f if you know where
its poles lie.
The Riemann sphere alone would not have drawn attention to topological
issues in complex analysis, but a second kind of singularity, called a branch
point, made topology essential. The simplest example is the complex square root
function, f(z) = . Most complex numbers have two distinct square roots, just
like real numbers. These square roots differ only in their sign: one is minus the
other. For example, the square roots of 2i turn out to be 1 + i and −1−i, much as
the real square roots of 4 are 2 and –2. However, there is one complex number
with only one square root, namely 0. Why? Because +0 and –0 are equal.
To see why 0 is a branch point of the square root function imagine starting at
the point 1 in the complex plane, and choosing one of the two square roots. The
obvious choice is also 1. Now gradually move the point round the unit circle,
and as you go, choose, for each position of the point, whichever of the two
square roots keeps everything varying continuously. By the time you get half
way round the circle, to –1, the square root has only gone a quarter of the way
round, to +i, since = +i or −i. Continuing all the way round, we get back to
the starting point 1. But the square root, moving at half the speed, ends up at –1.
To return the square root to its initial value, the point has to go round the circle
twice.
Riemann found a way to tame this kind of singularity, by doubling up the
Riemann sphere into two layers. These layers are separate except at the points 0
and ∞, which is a second branch point. At these two points the layers merge – or,
thinking about this the other way round, they branch from the single layers at 0
and –∞. Near these special points, the geometry of the layers is like a spiral
staircase – with the unusual feature that if you climb two full turns up the
staircase, you get back to where you started. The geometry of this surface tells us
a lot about the square root function, and the same idea can be extended to other
complex functions.

Sphere
 Torus

Two-holed torus

The description of the surface is rather indirect, and we can ask what shape is
it? It is here that topology comes into play. We can continuously deform the
spiral staircase description into something easier to visualize. The complex
analysts found that, topologically, every Riemann surface is either a sphere, or a
torus, or a torus with two holes, or a torus with three holes, etc. The number of
holes, g, is known as the genus of the surface, and it is the same g that occurs in
the generalization of Euler’s formula to surfaces.
Orientable surfaces

The genus turned out to be important for various deep issues in complex
analysis, which in turn attracted attention to the topology of surfaces. It then
turned out that there is a second class of surfaces, which differ from g-holed tori,
but are closely related. The difference is that g-holed tori are orientable surfaces,
which intuitively means that they have two distinct sides. They inherit this
property from the complex plane, which has a top side and a bottom side
because the spiral staircases are joined up in a way that preserves this distinction.
If instead you joined two flights of the staircase by twisting one floor upside
down, the apparently separate sides would join together.
The possibility of this type of joining was first emphasized by Möbius, whose
Möbius band has one side and one edge. Klein went one step further, by
conceptually gluing a circular disc along the edge of the Möbius band to
eliminate the edge altogether. The resulting surface, jokingly named the Klein
bottle, has a single side and no edges at all. If we try to draw it sitting inside
normal three-dimensional space, it has to pass through itself. But as an abstract
surface in its own right (or as a surface sitting inside four-dimensional space)
this self-intersection does not occur.

The Klein bottle. The apparent self-intersection is an artefact of its representation in three-dimensional
space

The theorem about g-holed tori can be restated like this: any orientable surface
(of finite extent with no edges) is topologically equivalent to a sphere with g
extra handles (where g might be zero). There is a similar classification of the
non-orientable (one-sided) surfaces: they can be formed from a surface called
the projective plane by adding g handles.
The combination of these two results is called the Classification Theorem for
Surfaces. It tells us, up to topological equivalence, all possible surfaces (of finite
extent with no edges). With the proof of this theorem, the topology of two-
dimensional spaces – surfaces – could be considered known. That did not mean
that every possible question about surfaces could be solved without further
effort, but it did give a reasonable place to start from when considering more
complicated issues. The Classification Theorem for Surfaces is a very powerful
tool in two-dimensional topology.

Jules Henri Poincaré

1854–1912

Henri Poincaré was born in Nancy, France. His father Léon was Professor of Medicine at the
University of Nancy, and his mother was Eugénie (née Launois). His cousin, Raymond Poincaré,
became the French prime minister and was president of the French Republic during the First
World War. Henri came top in every subject at school and was absolutely formidable in
mathematics. He had an excellent memory and could visualize complicated shapes in three
dimensions, which helped compensate for eyesight so poor that he could hardly see the
blackboard, let alone what was written on it.
 His first university job was in Caen in 1879, but by 1881 he had secured a far more prestigious
job at the University of Paris. There he became one of the leading mathematicians of his age. He
worked systematically – four hours per day in two two-hour periods, morning and late afternoon.
But his thought processes were less organized, and he often started writing a research paper
before he knew how it would end or where it would lead. He was highly intuitive, and his best
ideas often arrived when he was thinking about something else.
He ranged over most of the mathematics of his day, including complex function theory,
differential equations, non-Euclidean geometry and topology – which he virtually founded. He
also worked on applications: electricity, elasticity, optics, thermodynamics, relativity, quantum
theory, celestial mechanics and cosmology.

He won a major prize in a competition started in 1887 by King Oscar II of Sweden and Norway.
The topic was the ‘three-body problem’ – the motion of three gravitating bodies. The work he
actually submitted contained a major mistake, which he quickly corrected; as a result he
discovered the possibility of what is now called chaos – irregular, unpredictable motion in a
system governed by deterministic laws. He also wrote several bestselling popular science books:
Science and Hypothesis (1901), The Value of Science (1905) and Science and Method (1908).

When thinking about topology, it is often useful to suppose that the space
concerned is everything that exists. There is no need to embed it in a
surrounding space. So doing focuses attention on the intrinsic properties of the
space. A vivid image is that of a tiny creature living on, say, a topological
surface. How could such a creature, ignorant of any surrounding space, work out
which surface it inhabits? How can we characterize such surfaces intrinsically?
By 1900 it was understood that one way to answer such questions is to
consider closed loops in the surface, and how these loops can be deformed. For
example, on a sphere any closed loop can be continuously deformed to a point –
shrunk. For example, the circle running round the equator can be gradually
moved towards the north pole, becoming smaller and smaller until it coincides
with the north pole itself.
In contrast, every surface that is not equivalent to a sphere contains loops that
cannot be deformed to points. Such loops pass through a hole, and the hole
prevents them being shrunk. So the sphere can be characterized as the only
surface in which any closed loop can be shrunk to a point.

Topology in three dimensions

Topology in three dimensions

The natural next step after surfaces – two-dimensional topological spaces – is

three dimensions. Now the objects of study are manifolds in Riemann’s sense,
except that notions of distance are ignored. In 1904 Henri Poincaré, one of the
all-time greats in mathematics, was trying to understand three-dimensional
manifolds. He introduced a number of techniques for achieving this goal. One of
them, homology, studies the relations between regions in the manifold and their
boundaries. Another, homotopy, looks at what happens to closed loops in the
manifold as the loops are deformed.
Homotopy is closely related to the methods that had served so well for
surfaces, and Poincaré looked for analogous results in three dimensions. Here he
was led to one of the most famous questions in the whole of mathematics.
He knew the characterization of the sphere as the only surface in which any
closed loop can be shrunk. Did a similar characterization work in three
dimensions? For a time he assumed that it did; in fact, this seemed so obvious
that he didn’t even notice he was making an assumption. Later he realized that
one plausible version of this statement is actually wrong, while another closely
related formulation seemed difficult to prove but might well be true. He posed a
question, later reinterpreted as the Poincaré conjecture – if a three-dimensional
manifold (without boundary, of finite extent, and so on) has the property that any
loop in it can be shrunk to a point, then that manifold must be topologically
equivalent to the 3-sphere (a natural three-dimensional analogue of a sphere).
Subsequent attempts to prove the conjecture succeeded for generalizations to
four or more dimensions. Topologists continued to struggle with the original
Poincaré conjecture, in three dimensions, without success.
In the 1980s William Thurston came up with an idea that might make it
possible to sneak up on the Poincaré conjecture by being more ambitious. His
geometrization conjecture goes further, and applies to all three-dimensional
manifolds, not just those in which every loop can be shrunk. Its starting-point is
an interpretation of the classification of surfaces in terms of non-Euclidean
geometry.

‘The sphere has constant positive curvature.’

The torus can be obtained by taking a square in the Euclidean plane, and
identifying opposite edges. As such, it is flat – has zero curvature. The sphere
has constant positive curvature. A torus with two or more holes can be
represented as a surface of constant negative curvature. So the topology of
surfaces can be reinterpreted in terms of three types of geometry, one Euclidean
and two non-Euclidean, namely, Euclidean geometry itself, elliptic geometry
(positive curvature) and hyperbolic geometry (negative curvature).

What topology did for them

One of the simplest topological invariants was invented by Gauss. In a study of electrical
and magnetic fields, he became interested in how two closed loops can link together. He
invented the linking number, which measures how many times one loop winds round the
other one. If the linking number is non-zero, then the loops cannot be separated by a
topological transformation. However, this invariant does not completely solve the
problem of determining when two linked loops cannot be separated, because sometimes
the linking invariant is zero but the links cannot be separated.
He even developed an analytic formula for this number, by integrating a suitable
quantity along the curve concerned. Gauss’s discoveries provided a foretaste of what is
now a huge area of mathematics, algebraic topology.

(Left) Loops with linking number 3; (right) these links cannot be separated topologically,
even though they have linking number 0

Might something similar hold in three dimensions? Thurston pointed out some
complications: there are eight types of geometry to consider, not three. And it is
no longer possible to use a single geometry for a given manifold: instead, the
manifold must be cut into several pieces, using one geometry for each. He
formulated his geometrization conjecture: there is always a systematic way to
cut up a three-dimensional manifold into pieces, each corresponding to one of
the eight geometries.
 The Poincaré conjecture would be an immediate consequence, because the
condition that all loops shrink rules out seven geometries, leaving just the
geometry of constant positive curvature – that of the 3-sphere.
An alternative approach emerged from Riemannian geometry. In 1982
Richard Hamilton introduced a new technique into the area, based on the
mathematical ideas used by Albert Einstein in general relativity. According to
Einstein, space-time can be considered as curved, and the curvature describes the
force of gravity. Curvature is measured by the so-called curvature tensor, and
this has a simpler relative known as the Ricci tensor after its inventor Gregorio
Ricci-Curbastro. Changes in the geometry of the universe over time are
governed by the Einstein equations, which say that the stress tensor is
proportional to the curvature. In effect, the gravitational bending of the universe
tries to smooth itself out as time passes, and the Einstein equations quantify that
idea.
The same game can be played using the Ricci version of curvature, and it
leads to the same kind of behaviour: a surface that obeys the equations for the
Ricci flow will naturally tend to simplify its own geometry by redistributing its
curvature more equitably. Hamilton showed that the two-dimensional Poincaré
conjecture can be proved using the Ricci flow – basically, a surface in which all
loops shrink simplifies itself so much as it follows the Ricci flow that it ends up
as a perfect sphere. Hamilton also suggested generalizing this approach to three
dimensions, and made some progress along those lines, but hit some difficult
obstacles.

Perelman

In 2002 Grigori Perelman caused a sensation by placing several papers on the

arXiv, a website for physics and mathematics research that lets researchers
provide public access to unrefereed, often ongoing, work. The aim of the website
is to avoid long delays that occur while papers are being refereed for official
publication. Previously, this role had been played by informal preprints.
Ostensibly these papers were about the Ricci flow, but it became clear that if the
work was correct, it would imply the geometrization conjecture, hence that of
Poincaré.
The basic idea is the one suggested by Hamilton. Start with an arbitrary three-
dimensional manifold, equip it with a notion of distance so that the Ricci flow
makes sense, and let the manifold follow the flow, simplifying itself. The main
complication is that singularities can develop, where the manifold pinches
together and ceases to be smooth. At singularities, the proposed method breaks
down. The new idea is to cut it apart near such a singularity, cap off the resulting
holes and let the flow continue. If the manifold manages to simplify itself
completely after only finitely many singularities have arisen, each piece will
support just one of the eight geometries, and reversing the cutting operations
(surgery) tells us how to glue those pieces back together to reconstruct the
manifold.

Grigori Perelman
1966 –

Perelman was born in what was then the Soviet Union in 1966. As a student he was a member of
the USSR team competing in the International Mathematical Olympiad, and won a gold medal
with a 100% score. He has worked in the USA and at the Steklov Institute in St Petersburg, but
currently holds no specific academic position. His increasingly reclusive nature has added an
unusual human dimension to the mathematical story. It is perhaps a pity that this story reinforces
the stereotype of the eccentric mathematician.

The Poincaré conjecture is famous for another reason: it is one of the eight
Millennium Mathematics Problems selected by the Clay Institute, and as such its
solution – suitably verified – attracts a million-dollar prize. However, Perelman
had his own reasons for not wanting the prize – indeed, any reward save the
solution itself – and therefore had no strong reason to expand his often-cryptic
papers on the arXiv into something more suitable for publication.
Experts in the area therefore developed their own versions of his ideas, trying
to fill in any apparent gaps in the logic, and generally tidy up the work into
something acceptable as a genuine proof. Several such attempts were published,
and a comprehensive and definitive version of Perelman’s proof has now been
accepted by the topological community. In 2006 he was awarded a Fields Medal
for his work in this area, which he declined. Not everyone craves worldly
success.

Topology and the real world

Topology was invented because mathematics could not function without it,
stimulated by a number of basic questions in areas like complex analysis. It
tackles the question ‘what shape is this thing?’ in a very simple but deep form.
More conventional geometric concepts, such as lengths, can be seen as adding
extra detail to the basic information captured by topology.
A few early forerunners of topology exist, but it did not really become a
branch of mathematics with its own identity and power until the middle of the
19th century, when mathematicians obtained a fairly complete understanding of
the topology of surfaces, two-dimensional shapes. The extension to higher
dimensions was given a huge boost in the late 19th and early 20th centuries,
notably with the investigations of Henri Poincaré. Further advances occurred in
the 1920s; the subject really took off in the 1960s, although, ironically, it pretty
much lost contact with applied science.
Confounding those critical of the abstraction of 20th century pure
mathematics, the resulting theory is now vital to several areas of mathematical
physics. Even its most intractable obstacle, the Poincaré conjecture, has now
been overcome. In retrospect, the main difficulties in developing topology were
internal ones, best solved by abstract means; connections with the real world had
to wait until the techniques were sorted out properly.

What topology does for us

In 1956 James Watson and Francis Crick discovered the secret of life the double-helix
structure of the DNA molecule, the backbone on which genetic information is stored and
manipulated. Today the topology of knots is being used to understand how the two
strands of the helix disentangle as the genetic blueprint controls the development of a
living creature.
The DNA helix is like a two-stranded rope, with each strand twisted repeatedly around
the other. When a cell divides, the genetic information is transferred to the new cells by
splitting the strands apart, copying them and joining the new strands to the old in pairs.
Anyone who has tried to separate the strands of a long piece of rope knows how tricky
this process is: the strands tangle in lumps as you try to pull them apart. In fact DNA is
much worse: the helices themselves are supercoiled as if the rope itself has been wound
into a coil. Imagine several kilometres of fine thread stuffed into a tennis-ball and you
have some idea of how tangled the DNA in a cell must be.
The genetic biochemistry must ravel and unravel this tangled thread, rapidly,
repeatedly and faultlessly; the very chain of life depends upon it. How? Biologists tackle
the problem by using enzymes to break the DNA chain into pieces, small enough to
investigate in detail. A segment of DNA is a complicated molecular knot, and the same
knot can look very different after a few tucks and turns have distorted its appearance.

Knotted strands of DNA

The new techniques for studying knots open up new lines of attack on molecular
genetics. No longer a plaything of the pure mathematicians, the topology of knots is
becoming an important practical issue in biology. A recent discovery is a mathematical
connection between the amount of twisting in the DNA helix and the amount of
supercoiling.
 16
The Fourth Dimension

Geometry out of this world

In his science fiction novel The Time Machine, Herbert George Wells described
the underlying nature of space and time in a way that we now find familiar, but
which must have raised some eyebrows among his Victorian readers: ‘There are
really four dimensions, three which we call the three planes of Space, and a
fourth, Time’. To set up the background for his story, he added: ‘There is,
however, a tendency to draw an unreal distinction between the former three
dimensions and the latter, because it happens that our consciousness moves
intermittently in one direction along the latter from the beginning to the end of
our lives. But some philosophical people have been asking why three dimensions
particularly – why not another direction at right angles to the three? – and have
even tried to construct a four-dimensional geometry’. His protagonist then goes
one better, overcomes the alleged limitations of human consciousness and
travels along the fourth dimension of time, as if it were a normal dimension of
space.

The fourth dimension

The art of the science fiction writer is suspension of disbelief, and Wells
achieved this by informing his readers that ‘Professor Simon Newcomb was
expounding this to the New York Mathematical Society only a month or so ago’.
Here Wells was probably referring to a real event; we know that Newcomb, a
prominent astronomer, gave a lecture on four-dimensional space at roughly the
right time. His lecture reflected a major change in mathematical and scientific
thinking, freeing these subjects from the traditional assumption that space must
always have three dimensions. This does not imply that time travel is possible,
but it gave Wells an excuse to make penetrating observations about present-day
human nature by displacing his time traveller into a disturbing future.
The Time Machine, published in 1895, resonated with a Victorian obsession
with the fourth dimension, in which an additional, unseen dimension of space
was invoked as a place for ghosts, spirits, or even God to reside. The fourth
dimension was championed by charlatans, exploited by novelists, speculated
upon by scientists and formalized by mathematicians. Within a few decades, not
only was four-dimensional space standard in mathematics: so were spaces with
any number of dimensions – five, ten, a billion, even infinity. The techniques
and thought-patterns of multidimensional geometry were being used routinely in
every branch of science – even biology and economics.

‘The fourth dimension was championed by

charlatans, exploited by novelists. . .’

Higher-dimensional spaces remain almost unknown outside the scientific

community, but very few areas of human thought could now function effectively
without these techniques, remote though they may seem from ordinary human
affairs. Scientists trying to unify the two great theories of the physical universe,
relativity and quantum mechanics, are speculating that space may actually have
nine dimensions, or ten, rather than the three that we normally perceive. In a
rerun of the fuss about non-Euclidean geometry, space of three dimensions is
increasingly being viewed as just one possibility out of many, rather than the
only kind of space that is possible.
These changes have come about because terms like space and dimension are
now interpreted in a more general manner, which agrees with the usual
dictionary meanings in the familiar contexts of a TV screen or our normal
surroundings, but opens up new possibilities. To mathematicians, a space is a
collection of objects together with some notion of the distance between any two
of those objects. Taking a hint from Descartes’s idea of coordinates, we can
define the dimension of such a space to be how many numbers are required to
specify an object. With points as the objects, and the usual notion of distance in
the plane or space, we find that the plane has two dimensions and space has
three. However, other collections of objects may have four dimensions, or more,
depending on what the objects are.
For example, suppose that the objects are spheres in three-dimensional space.
It takes four numbers (x, y, z, r) to specify a sphere: three coordinates (x, y, z) for
its centre, plus the radius r. So the space of all spheres in ordinary space has four
dimensions. Examples like this show that natural mathematical questions can
easily lead to higher-dimensional spaces.
Indeed, modern mathematics goes further. Abstractly, space of four
dimensions is defined as the set of all quadruples (x1, x2, x3, x4) of numbers. More
generally, space of n dimensions – for any whole number n – is defined as the
set of all n-tuples (x1, x2, . . ., xn) of numbers. In a sense, that is the whole story;
the intriguing and baffling notion of many dimensions collapses to a triviality:
long lists of numbers.
That viewpoint is now clear, but historically, it took a long time to become
established. Mathematicians argued, often very forcibly, about the meaning and
reality of higher-dimensional spaces. It took about a century for the ideas to
become widely accepted. But the applications of such spaces, and the geometric
imagery that went with them, proved so useful that the underlying mathematical
issues ceased to be controversial.

Three-or four-dimensional space

Ironically, today’s conception of higher-dimensional spaces emerged from

algebra, not geometry, as a consequence of a failed attempt to develop a three-
dimensional number system, analogous to the two-dimensional system of
complex numbers. The distinction between two and three dimensions goes back
to Euclid’s Elements. The first part of the book is about the geometry of the
plane, a space of two dimensions. The second part is about solid geometry – the
geometry of three-dimensional space. Until the 19th century, the word
dimension was limited to these familiar contexts.
Greek geometry was a formalization of the human senses of sight and touch,
which allow our brains to build internal models of positional relationships of the
outside world. It was constrained by the limitations of our own senses, and of the
world in which we live. The Greeks thought that geometry described the real
space in which we live, and they assumed that physical space has to be
Euclidean. The mathematical question ‘can four-dimensional space exist in some
conceptual sense?’ became confused with the physical question ‘can a real space
with four dimensions exist?’ And that question was further confused with ‘can
there be four dimensions within our own familiar space?’ to which the answer is
‘no’. So it was generally believed that four-dimensional space is impossible.
Geometry began to free itself from this restricted viewpoint when the
algebraists of Renaissance Italy unwittingly stumbled upon a profound extension
of the number concept, by accepting the existence of a square root of minus one.
Wallis, Wessel, Argand and Gauss worked out how to interpret the resulting
complex numbers as points in a plane, freeing numbers from the one-
dimensional shackles of the real number line. In 1837, The Irish mathematician
William Rowan Hamilton reduced the whole topic to algebra, by defining a
complex number x + iy to be a pair of real numbers (x, y). He further defined
addition and multiplication of pairs by the rules

(x, y) + (u, v) = (x + u, y + v)
(x, y)(u, v) = (xu – yv, xv + yu)

William Rowan Hamilton

1805–1865
 Hamilton was so precocious mathematically that he was made Professor of Astronomy at Trinity
College Dublin while still an undergraduate, at the age of 21. This appointment made him Royal
Astronomer of Ireland.

He made numerous contributions to mathematics, but the one that he himself believed to be
most significant was the invention of quaternions. He tells us that ‘Quaternions . . . started into
life, fully grown, on the 16th of October, 1843, as I was walking with Lady Hamilton to Dublin,
and came up to Brougham Bridge. That is to say, I then and there felt the galvanic circuit of
thought closed, and the sparks which fell from it were the fundamental equations between i, j, k;
exactly such as I have used them ever since. I pulled out, on the spot, a pocketbook, which still
exists, and made an entry, on which, at the very moment, I felt that it might be worth my while to
expend the labour of at least ten (or it might be fifteen) years to come. I felt a problem to have
been at that moment solved, an intellectual want relieved, which had haunted me for at least
fifteen years before.’
Hamilton immediately carved the equation

i2 = j2 = k2 = ijk − 1

in the stone of the bridge.

In this approach, a pair of the form (x,0) behaves just like the real number x,
and the special pair (0,1) behaves like i. The idea is simple, but appreciating it
requires a sophisticated concept of mathematical existence.
Hamilton then set his sights on something more ambitious. It was well known
that complex numbers make it possible to solve many problems about the
mathematical physics of systems in the plane, using simple and elegant methods.
A similar trick for three-dimensional space would be invaluable. So he tried to
invent a three-dimensional number system, in the hope that the associated
calculus would solve important problems of mathematical physics in three-
dimensional space. He tacitly assumed that this system would satisfy all the
usual laws of algebra. But despite heroic efforts, he could not find such a system.
Eventually, he discovered why. It’s impossible.
Among the usual laws of algebra is the commutative law of multiplication,
which states that ab = ba. Hamilton had been struggling for years to devise an
effective algebra for three dimensions. Eventually he found one, a number
system that he called quaternions. But it was an algebra of four dimensions, not
three, and its multiplication was not commutative.
Quaternions resemble complex numbers, but instead of one new number, i,
there are three: i, j, k. A quaternion is a combination of these, for example 7 + 8i
– 2j + 4k. Just as the complex numbers are two-dimensional, built from two
independent quantities 1 and i, so the quaternions are four-dimensional, built
from four independent quantities 1, i, j and k. They can be formalized
algebraically as quadruples of real numbers, with particular rules for addition
and multiplication.

Higher-dimensional space

When Hamilton made his breakthrough, mathematicians were already aware that
spaces of high dimension arise entirely naturally, and have sensible physical
interpretations, when the basic elements of space are something other than
points. In 1846 Julius Plücker pointed out that it takes four numbers to specify a
line in space. Two of those numbers determine where the line hits some fixed
plane; two more determine its direction relative to that plane. So, considered as a
collection of lines, our familiar space already has four dimensions, not three.
However, there was a vague feeling that this construction was rather artificial,
and that spaces made from four dimensions worth of points were unnatural.
Hamilton’s quaternions had a natural interpretation as rotations, and their
algebra was compelling. They were as natural as complex numbers – so four-
dimensional space was as natural as a plane.
 ‘His presentation was mystical and rather
abstract. . .’

The idea quickly went beyond just four dimensions. While Hamilton was
promoting his beloved quaternions, a mathematics teacher named Hermann
Günther Grassmann was discovering an extension of the number system to
spaces with any number of dimensions. He published his idea in 1844 as
Lectures on Lineal Extension. His presentation was mystical and rather abstract,
so the work attracted little attention. In 1862, to combat the lack of interest, he
issued a revised version, often translated as The Calculus of Extension, which
was intended to be more comprehensible. Unfortunately, it wasn’t.
Despite its cool reception, Grassmann’s work was of fundamental importance.
He realized that it was possible to replace the four units, 1, i, j and k, of
quaternions by any number of units. He called combinations of these units
hypernumbers. He understood that his approach had limitations. You had to be
careful not to expect too much from the arithmetic of hypernumbers; slavishly
following the traditional laws of algebra seldom led anywhere.
Meanwhile, physicists were developing their own notions of higher-
dimensional spaces, motivated not by geometry, but by Maxwell’s equations for
electromagnetism. Here both the electric and magnetic fields are vectors –
having a direction in three-dimensional space as well as a magnitude. Vectors
are arrows, if you wish, aligned with the electric or magnetic field. The length of
the arrow shows how strong the field is, and its direction shows which way the
field is pointing.

‘Physicists were developing their own notions of

higher-dimensional spaces.’

In the notation of the time, Maxwell’s equations were eight in number, but
they included two groups of three equations, one for each component of the
electric or the magnetic field in each of the three directions of space. It would
make life much easier to devise a formalism that collected each such triple into a
single vector equation. Maxwell achieved this using quaternions, but his
approach was clumsy. Independently, the physicist Josiah Willard Gibbs and the
engineer Oliver Heaviside found a simpler way to represent vectors
algebraically. In 1881 Gibbs printed a private pamphlet, Elements of Vector
Analysis, to help his students. He explained that his ideas had been developed for
convenient use rather than mathematical elegance. His notes were written up by
Edwin Wilson, and they published a joint book Vector Analysis in 1901.
Heaviside came up with the same general ideas in the first volume of his
Electromagnetic Theory in 1893 (the other two volumes appeared in 1899 and
1912).
The different systems of Hamilton’s quaternions, Grassmann’s hypercomplex
numbers and Gibbs’s vectors rapidly converged to the same mathematical
description of a vector: it is a triple (x, y, z) of numbers. After 250 years, the
world’s mathematicians and physicists had worked their way right back to
Descartes – but now the coordinate notation was only part of the story. Triples
did not just represent points: they represented directed magnitudes. It made a
huge difference – not to the formalism, but to its interpretation, its physical
meaning.
Mathematicians wondered just how many hypercomplex number systems
there might be. To them, the question was not ‘are they useful?’ but ‘are they
interesting?’ So, mathematicians mainly focused on the algebraic properties of
systems of n-hypercomplex numbers, for any n. These were, in fact, n-
dimensional spaces, plus algebraic operations, but to begin with everyone
thought algebraically and the geometric aspects were played down.

Differential geometry

Geometers responded to the algebraists invasion of their territory by

reinterpreting hypercomplex numbers geometrically. The key figure here was
Riemann. He was working for his ‘Habilitation’ which gave him the right to
charge lecture fees to students. Candidates for Habilitation must give a special
lecture on their own research. Following the usual procedure, Gauss asked
Riemann to propose a number of topics, from which Gauss would make the final
choice. One of Riemann’s proposals was On the Hypotheses Which Lie at the
Foundation of Geometry, and Gauss, who had been thinking about the same
question, chose that topic.

‘Riemann was terrified – he disliked public speaking

 and he hadn’t fully worked out his ideas.’

Riemann was terrified – he disliked public speaking and he hadn’t fully

worked out his ideas. But what he had in mind was explosive: a geometry of n
dimensions, by which he meant a system of n coordinates (x1, x2, . . ., xn),
equipped with a notion of distance between nearby points. He called such a
space a manifold. This proposal was radical enough, but there was another, even
more radical feature: manifolds could be curved. Gauss had been studying the
curvature of surfaces, and had obtained a beautiful formula which represented
curvature intrinsically – that is, in terms of the surface alone, not of the space is
which it was embedded.
Riemann had intended to develop a similar formula for the curvature of a
manifold, generalizing Gauss’s formula to n dimensions. This formula would
also be intrinsic to the manifold – it would not make explicit use of any
containing space. Riemann’s efforts to develop the notion of curvature in a space
of n dimensions led him to the brink of a nervous breakdown. What made
matters worse was that at the same time, he was helping Gauss’s colleague
Weber, who was trying to understand electricity. Riemann battled on, and the
interplay between electrical and magnetic forces led him to a new concept of
force based on geometry. He had the same insight that led Einstein to general
relativity, decades later: forces can be replaced by the curvature of space.
In traditional mechanics, bodies travel along straight lines unless diverted by a
force. In curved geometries, straight lines need not exist and paths are curved. If
space is curved, what you experience when you are obliged to deviate from a
straight line feels like a force. Now Riemann had the insight he needed to
develop his lecture, which he gave in 1854. It was a major triumph. The ideas
quickly spread, to growing excitement. Soon scientists were giving popular
lectures on the new geometry. Among them was Hermann von Helmholtz, who
gave talks about beings that lived on a sphere or some other curved surface.
The technical aspects of Riemann’s geometry of manifolds, now called
differential geometry, were further developed by Eugenio Beltrami, Elwin Bruno
Christoffel and the Italian school under Gregorio Ricci and Tullio Levi-Civita.
Later, their work turned out to be just what Einstein needed for general
relativity.

Matrix algebra
Matrix algebra

Algebraists had also been busy, developing computational techniques for n-

variable algebra – the formal symbolism of n-dimensional space. One of these
techniques was the algebra of matrices, rectangular arrays of numbers,
introduced by Cayley in 1855. This formalism arose naturally from the idea of a
change of coordinates. It had become commonplace to simplify algebraic
formulas by replacing variables such as x and y by linear combinations, for
example

u = ax + by
v = cx + dy

for constants a, b, c and d. Cayley represented the pair (x, y) as a column vector
and the coefficients by 2 × 2 table, or matrix. With a suitable definition of
multiplication, he could rewrite the coordinate change as

The method extended easily to tables with any number of rows and columns,
representing linear changes in any number of coordinates.
Matrix algebra made it possible to calculate in n-dimensional space. As the
new ideas sank in, a geometric language for n-dimensional space came into
being, supported by a formal algebraic computational system. Cayley thought
that his idea was no more than a notational convenience, and predicted that it
would never acquire any applications. Today, it is indispensable throughout
science, especially in areas like statistics. Medical trials are a great consumer of
matrices, which are used to work out which associations between cause and
effect are statistically significant.
The geometric imagery made it easier to prove theorems. Critics countered
that these newfangled geometries referred to spaces that didn’t exist. The
algebraists fought back by pointing out that the algebra of n variables most
certainly did exist, and anything that helped advance many different areas of
mathematics must surely be interesting. George Salmon wrote: ‘I have already
completely discussed this problem [solving a certain system of equations] when
we are given three equations in three variables. The question now before us may
be stated as the corresponding problem in space of p dimensions. But we
consider it as a purely algebraic question, apart from any geometrical
considerations. We shall however retain a little of geometrical language. . .
because we can thus more readily see how to apply to a system of p equations,
processes analogous to those which we have employed in a system of three.’

What high-dimensional geometry did for them

Around 1907 the German mathematician Hermann Minkowski formulated Einstein’s
theory of special relativity in terms of a four-dimensional space-time, combining one-
dimensional time and three-dimensional space into a single mathematical object. This is
known as Minkowski space-time.
The requirements of relativity imply that the natural metric on Minkowski space-time is
not the one determined by Pythagoras’s theorem, in which the square of the distance
from a point (x, t) to the origin is x2 + x2. Instead, this expression should be replaced by
the interval x2 – c2t2, where c is the speed of light. The crucial change here is the minus
sign, which implies that events in space-time are associated with two cones. One cone
(here a triangle because space has been reduced to one dimension) represents the
future of the event, the other its past. This geometric representation is employed almost
universally by modern physicists.
Real space

Do higher dimensions exist? Of course, the answer depends on what we mean by

‘exist’, but people tend not to understand that kind of thing, especially when
their emotions are aroused. The issue came to a head in 1869. In a famous
address to the British Association, later reprinted as A Plea for the
Mathematician, James Joseph Sylvester pointed out that generalization is an
important way to advance mathematics. What matters, said Sylvester, is what is
conceivable, not what corresponds directly to physical experience. He added that
with a little practice it is perfectly possible to visualize four dimensions, so four-
dimensional space is conceivable.

‘The nature of real space is irrelevant to the

mathematical issues.’

This so infuriated the Shakespearean scholar Clement Ingleby that he invoked

the great philosopher Immanuel Kant to prove that three-dimensionality is an
essential feature of space, completely missing Sylvester’s point. The nature of
real space is irrelevant to the mathematical issues. Nevertheless, for a time most
British mathematicians agreed with Ingleby. But some continental
mathematicians did not. Grassmann said: ‘The theorems of the Calculus of
Extension are not merely translations of geometrical results into an abstract
language; they have a much more general significance, for while the ordinary
geometry remains bound to three dimensions of [physical] space, the abstract
science is free of this limitation’.
Sylvester defended his position: ‘There are many who regard the alleged
notion of a generalized space as only a disguised form of algebraic
formulization; but the same might be said with equal truth of our notion of
infinity, or of impossible lines, or lines making a zero angle in geometry, the
utility of dealing with which no one will be found to dispute. Dr Salmon in his
extension of Chasles’s theory of characteristics to surfaces, Mr Clifford in a
question of probability, and myself in the theory of partitions, and also in my
paper on barycentric projection, have all felt and given evidence on the practical
utility of handling space of four dimensions as if it were conceivable space’.

Multidimensional space

In the end, Sylvester won the debate. Nowadays mathematicians consider

something to exist if it is not logically contradictory. It may contradict physical
experience, but that is irrelevant to mathematical existence. In this sense,
multidimensional spaces are just as real as the familiar space of three
dimensions, because it is just as easy to provide a formal definition.
The mathematics of multidimensional spaces, as now conceived, is purely
algebraic, and based on obvious generalizations from low-dimensional spaces.
For example, every point in the plane (a two-dimensional space) can be specified
by its two coordinates, and every point in three-dimensional space can be
specified by its three coordinates. It is a short step to define a point in four-
dimensional space as a set of four coordinates, and more generally to define a
point in n-dimensional space as a list of n coordinates. Then n-dimensional space
itself (or n-space for short) is just the set of all such points.
Similar algebraic machinations let you work out the distance between any two
points in n-space, the angle between any two lines, and so on. From there on out,
it’s a matter of imagination: most sensible geometric shapes in two or three
dimensions have straightforward analogues in n dimensions, and the way to find
them is to describe the familiar shapes using the algebra of coordinates and then
extend that description to n coordinates.

‘Nowadays, this point of view is called linear

algebra.’

For example, a circle in the plane, or a sphere in 3-space, consists of all points
that lie at a fixed distance (the radius) from a chosen point (the centre). The
obvious analogue in n-space is to consider all points that lie at a fixed distance
from a chosen point. Using the formula for distances, this becomes a purely
algebraic condition, and the resulting object is known as an (n − 1)-dimensional
hypersphere, or (n − 1)-sphere for short. The dimension drops from n to n − 1
because, for example, a circle in 2-space is a curve, which is a one-dimensional
object; similarly a sphere in space is a two-dimensional surface. A solid
hypersphere in n dimensions is called an n-ball. So the Earth is a 3-ball and its
surface is a 2-sphere.

A four-dimensional hypercube, projected on to the plane

Nowadays, this point of view is called linear algebra. It is used throughout

mathematics and science, especially in engineering and statistics. It is also a
standard technique in economics. Cayley stated that his matrices were unlikely
ever to have any practical application. He could not have been more wrong.
By 1900 Sylvester’s predictions were coming true, with an explosion of
mathematical and physical areas where the concept of multidimensional space
was having a serious impact. One such area was Einstein’s relativity, best
considered as a special kind of four-dimensional space-time geometry. In 1908
Hermann Minkowski realized that the three coordinates of ordinary space,
together with an extra one for time, form a four-dimensional space-time. Any
point in space-time is called an event: it is like a point particle that winks into
existence at just one moment in time, and then winks out again. Relativity is
really about the physics of events. In traditional mechanics, a particle moving
through space occupies coordinates (x(t), y(t), z(t)) at time t, and this position
changes as time passes. From Minkowski’s space-time viewpoint, the collection
of all such points is a curve in space-time, the world line of the particle, and it is
a single object in its own right, existing for all time. In relativity, the fourth
dimension has a single, fixed interpretation: time.
The subsequent incorporation of gravity, achieved in general relativity, made
heavy use of Riemann’s revolutionary geometries, but modified to suit
Minkowski’s representation of the geometry of flat space-time – that is, what
space and time do when no mass is present to cause gravitational distortions,
which Einstein modelled as curvature.
Mathematicians preferred a more flexible notion of dimensionality and space
and as the late 19th century flowed into the early 20th, mathematics itself
seemed, ever more, to demand acceptance of multidimensional geometry. The
theory of functions of two complex variables, a natural extension of complex
analysis, required thinking about space of two complex dimensions – but each
complex dimension boils down to two real ones, so like it or not you are looking
at a four-dimensional space. Riemann’s manifolds and the algebra of many
variables provided further motivation.

Generalized coordinates

Yet another stimulus towards multidimensional geometry was Hamilton’s 1835

reformulation of mechanics in terms of generalized coordinates, a development
initiated by Lagrange in his Analytical Mechanics of 1788. A mechanical system
has as many of these coordinates as it has degrees of freedom – that is, ways to
change its state. In fact the number of degrees of freedom is just dimension in
disguise.
For example, it takes six generalized coordinates to specify the configuration
of a rudimentary bicycle: one for the angle at which the handlebars sit relative to
the frame, one each for the angular positions of the two wheels, another for the
axle of the pedals, two more for the rotational positions of the pedals themselves.
A bicycle is, of course, a three-dimensional object – but the space of possible
configurations of the bicycle is six-dimensional, which is one of the reasons why
learning to ride a bicycle is hard until you get the knack. Your brain has to
construct an internal representation of how those six variables interact – you
have to learn to navigate in the six-dimensional geometry of bicycle-space. For a
moving bicycle, there are six corresponding velocities to worry about too: the
dynamics is, in essence, twelve-dimensional.
By 1920 this concurrence of physics, mathematics and mechanics had
triumphed, and the use of geometric language for many-variable problems –
multidimensional geometry – had ceased to raise eyebrows, except perhaps
among philosophers. By 1950, the process had gone so far that mathematicians’
natural tendency was to formulate everything in n dimensions from the
beginning. Limiting theories to two or three dimensions seemed old-fashioned
and ridiculously confining.
The language of higher-dimensional space rapidly spread into every area of
science, and even invaded subjects like economics and genetics. Today’s
virologists, for instance, think of viruses as points in a space of DNA sequences
that could easily have several hundred dimensions. By this they mean, at root,
that the genomes of these viruses are several hundred DNA bases long – but the
geometric image goes beyond mere metaphor: it provides an effective way to
think about the problem.

‘Our universe may actually have ten dimensions.’

None of this, however, means that the spirit world exists, that ghosts now have
a credible home, or that one day we might (as in Edwin Abbott’s Flatland)
receive a visit from the Hypersphere, a creature from the Fourth Dimension, who
would manifest himself to us as a sphere whose size kept mysteriously changing,
able to shrink to a point and vanish from our universe. However, physicists
working in the theory of superstrings currently think that our universe may
actually have ten dimensions, not four. Right now they think that we’ve never
noticed the extra six dimensions because they are curled up too tightly for us to
detect them.
Multidimensional geometry is one of the most dramatic areas in which
mathematics appears to lose all touch with reality. Since physical space is three-
dimensional, how can spaces of four or more dimensions exist? And even if they
can be defined mathematically, how can they possibly be useful?
The mistake here is to expect mathematics be an obvious, literal translation of
reality, observed in the most direct manner. We are in fact surrounded by objects
that can best be described by a large number of variables, the ‘degrees of
freedom’ of those objects. To state the position of a human skeleton requires at
least 100 variables, for example. Mathematically, the natural description of such
objects is in terms of high-dimensional spaces, with one dimension for each
variable.
It took mathematicians a long time to formalize such descriptions, and even
longer to convince anyone else that they are useful. Today, they have become so
deeply embedded in scientific thinking that their use has become a reflex action.
They are standard in economics, biology, physics, engineering, astronomy. . . the
list is endless.
The advantage of high-dimensional geometry is that it brings human visual
abilities to bear on problems that are not initially visual at all. Because our brains
are adept at visual thinking, this formulation can often lead to unexpected
insights, not easily obtainable by other methods. Mathematical concepts that
have no direct connection with the real world often have deeper, indirect
connections. It is those hidden links that make mathematics so useful.

What high-dimensional geometry does for us

Your mobile phone makes essential use of multidimensional spaces.
So do your Internet connection, your satellite or cable TV and virtually any other piece
of technology that sends or receives messages. Modern communications are digital. All
messages, even telephone voice messages, are converted into patterns of 0s and 1s –
binary numbers.
Communications are not much use unless they are reliable – the message that is
received should be exactly the same as the one that was sent. Electronic hardware
cannot guarantee this kind of accuracy, because interference, or even a passing cosmic
ray, can cause mistakes. So, electronic engineers use mathematical techniques to put
signals into code, in such a way that errors can be detected, and even corrected. The
basis of these codes is the mathematics of multidimensional spaces.
Such spaces turn up because a string of, say, ten binary digits, or bits, such as
1001011100, can profitably be viewed as a point in a ten-dimensional space with
coordinates restricted either to 0 or to 1. Many important questions about error-detecting
and error-correcting codes are best tackled in terms of the geometry of this space.
 Geometry of pairs of binary digits

For example, we can detect (but not correct) a single error if we code every message
by replacing every 0 by 00 and every 1 by 11. Then, a message such as 110100 codes
as 111100110000. If this is received as 111000110000, with an error in the fourth bit, we
know something has gone wrong because the boldface pair 10 should not occur. But we
don’t know whether it should have been 00 or 11. This can be neatly illustrated in a two-
dimensional figure (corresponding to the length 2 of the code words 00 and 11). By
thinking of the bits in the code words as coordinates relative to two axes (corresponding
to the first and second digits of the code word, respectively) we can draw a picture in
which the valid code words 00 and 11 are diagonally opposite corners of a square.
Any single error changes them to code words at the other two corners – which are not
valid code words. However, because these corners are adjacent to both of the valid
code words, different errors can lead to the same result. To get an error-correcting code,
we can use code words of length three and encode 0 as 000 and 1 as 111. Now the
code words live at the corners of a cube in three-dimensional space. Any single error
results in an adjacent code word; moreover, each such invalid code word is adjacent to
only one of the valid code words 000 or 111.
This approach to coding of digital messages was pioneered by Richard Hamming in
1947. The geometric interpretation came soon after, and it has proved crucial for the
development of more efficient codes.
Error-correcting code using strings of length three
 17
The Shape of Logic

Putting mathematics on fairly firm foundations

As the superstructure of mathematics grew ever larger, a small number of

mathematicians began to wonder whether the foundations could support its
weight. A series of foundational crises – in particular the controversies over the
basic concepts of calculus and the general confusion about Fourier series – had
made it clear that mathematical concepts must be defined very carefully and
precisely to avoid logical pitfalls. Otherwise the subject’s towers of deduction
could easily collapse in logical contradictions, because of some underlying
vagueness or ambiguity.
At first, such worries were focused on complicated, sophisticated ideas, such
as Fourier series. But slowly the mathematical world came to realize that very
basic ideas might also be suspect. Paramount among them was the concept of a
number. The dreadful truth was that mathematicians had devoted so much effort
to the discovery of deep properties of numbers that they had neglected to ask
what numbers were. And when it came to giving a logical definition, they didn’t
know.

Dedekind

In 1858, teaching a calculus course, Dedekind became worried about the basis of
calculus. Not its use of limits, but the system of real numbers. He published his
thoughts in 1872 as Stetigkeit und Irrationale Zahlen, pointing out that
apparently obvious properties of real numbers had never been proved in any
rigorous way. As an example he cited the equation . Obviously this fact
follows by squaring both sides of the equation – except that multiplication of
irrational numbers had never actually been defined. In his 1888 book Was Sind
und was Sollen die Zahlen? (What are Numbers, and What do They Mean?) he
exposed serious gaps in the logical foundations of the system of real numbers.
No one had really proved that the real numbers exist.

‘No one had really proved that the real numbers

exist.’

He also proposed a way to fill these gaps, using what we now call Dedekind
cuts. The idea was to start from an established number system, the rational
numbers, and then to extend this system to obtain the richer system of real
numbers. His approach was to start from the properties required of real numbers,
find some way to rephrase them solely in terms of rational numbers and then
reverse the procedure, interpreting those features of rational numbers as a
definition of the reals. This kind of reverse engineering of new concepts from
old ones has been widely used ever since.
Suppose, for the moment, that real numbers do exist. How do they relate to
rational numbers? Some reals are not rational, an obvious example being .
Now, although is not an exact fraction, it can be approximated as closely as we
wish by rationals. It somehow sits in a specific position, sandwiched among the
dense array of all possible rationals. But how can we specify that position?
Dedekind realized that neatly separates the set of rational numbers into two
pieces: those that are less than , and those that are greater. In a sense, this
separation – or cut – defines the number in terms of rationals. The only snag is
that we make use of in order to define the two pieces of the cut. However,
there is a way out. The rational numbers greater than are precisely those that
are positive, and for which the square is greater than . The rational numbers
less than are all the others. These two sets of rational numbers are now defined
without any explicit use of , but they specify its location on the real-number
line precisely.
Dedekind showed that if, for the sake of argument, we assume that the real
numbers exist, then a cut satisfying these two properties can be associated with
any real number, by forming the set R of all rationals that are larger than that real
number, and the set L of all rationals that are smaller than that real number, or
equal to it. (The final condition is needed to associate a cut with any rational
number. We don’t want to leave them out.) Here L and R can be read as left and
right in the usual picture of the real number line.

These two sets L and R obey some rather stringent conditions. First, every
rational number belongs to precisely one of them. Second, every number in R is
greater than any number in L. Finally, there is a technical condition that takes
care of the rational numbers themselves: L may or may not have a largest
member, but R never has a smallest member. Call any pair of subsets of the
rationals with these properties a cut.
In reverse engineering, we do not need to assume that real numbers exist.
Instead, we can use cuts to define real numbers, so that effectively a real number
is a cut. We don’t usually think of a real number that way, but Dedekind realized
that we can do so if we wish. The main task is to define how to add and multiply
cuts, so that the arithmetic of real numbers makes sense. This turns out to be
easy. To add two cuts (L1, R1) and (L2, R2), define L1 + L2 to be the set of all
numbers obtainable by adding a number in L1 to a number in L2 and similarly
define R1 + R2. Then the sum of the two cuts is the cut (L1 + L2, R1 + R2).
Multiplication is similar, but positive and negative numbers behave slightly
differently.
Finally, we have to verify that the arithmetic of cuts has all the features that
we expect of real numbers. These include the standard laws of algebra, which all
follow form analogous features of rational numbers. The crucial feature, which
distinguishes the reals from the rationals, is that the limit of an infinite sequence
of cuts exists (under certain technical conditions). Equivalently, there is a cut
corresponding to any infinite decimal expansion. This is also fairly
straightforward.
Assuming all this can be done, let’s see how Dedekind can then prove that
. We have seen that corresponds to the cut (L1, R1) where R1 consists
of all positive rationals the square of which is greater than 2. Similarly,
corresponds to the cut (L2, R2) where R2 consists of all positive rationals the
square of which is greater than 3. The product of these cuts is easily proved to be
(L3, R3) where R3 consists of all positive rationals the square of which is greater
than 6. But this is the cut corresponding to . Done!
The beauty of Dedekind’s approach is that it reduces all issues concerning real
numbers to corresponding issues about rational numbers – specifically, about
pairs of sets of rational numbers. It therefore defines real numbers purely in
terms of rational numbers and operations on those numbers. The upshot is that
real numbers exist (in a mathematical sense) provided rational numbers do.
There is a small price to pay: a real number is now defined as a pair of sets of
rationals, which is not how we usually think of a real number. If this sounds
strange, bear in mind that the usual representation of a real number as an infinite
decimal requires an infinite sequence of decimal digits 0–9. This is conceptually
at least as complicated as a Dedekind cut. But it is actually very tricky to define
the sum or product of two infinite decimals, because the usual arithmetical
methods for adding or multiplying decimals start from the right-hand end – and
when a decimal is infinite, it doesn’t have a right-hand end.

Axioms for whole numbers

Dedekind’s book was all very well as a foundational exercise, but as the general
point about defining your terms sank in, it was soon realized that the book had
merely shifted attention from the reals to the rationals. How do we know that
rational numbers exist? Well, if we assume the integers exist, this is easy: define
a rational p/q to be a pair of integers (p, q) and work out the formulas for sums
and products. If integers exist, so do pairs of integers.

‘If integers exist, so do pairs of integers. Yes, but

 how do we know that integers exist?’

Yes, but how do we know that integers exist? Apart from a plus or minus sign,
integers are ordinary whole numbers. Taking care of the signs is easy. So,
integers exist provided whole numbers do.
We still haven’t finished, though. We are so familiar with whole numbers that
it never occurs to us to ask whether the familiar numbers 0, 1, 2, 3 and so on
actually exist. And if they do, what are they?
In 1889, Giuseppe Peano sidestepped the question of existence by taking a
leaf out of Euclid’s book. Instead of discussing the existence of points, lines,
triangles and the like, Euclid simply wrote down a list of axioms – properties
that would be assumed without further question. Never mind whether points and
so on exist – a more interesting question is: if they did, what properties would
follow? So Peano wrote down a list of axioms for whole numbers. The main
features were:

• There exists a number 0.

• Every number n has a successor, s(n), (which we think of as n + 1).
• If P(n) is a property of numbers, such that P(0) is true, and whenever P(n) is
true then P(s(n)) is true, then P(n) is true for every n (Principle of
Mathematical Induction).

He then defined the numbers 1, 2 and so on in terms of those axioms,

essentially by setting

1 = s(0)
2 = s(s(0))

and so on. And he also defined the basic operations of arithmetic and proved that
they obey the usual laws. In his system, 2 + 2 = 4 is a provable theorem, stated
as s(s(0)) + s(s(0)) = s(s(s(s(0)))).
A great advantage of this axiomatic approach is that it pins down exactly what
we have to prove if we want to show, by some means or other, that whole
numbers exist. We just have to construct some system that satisfies all of
Peano’s axioms.
The deep question here is the meaning of ‘exist’ in mathematics. In the real
world, something exists if you can observe it, or, failing that, infer its necessary
presence from things that can be observed. We know that gravity exists because
we can observe its effects, even though no one can see gravity. So, in the real
world, we can sensibly talk about the existence of two cats, two bicycles or two
loaves of bread. However, the number two is not like that. It is not a thing, but a
conceptual construct. We never meet the number two in the real world. The
closest we get is a symbol, 2, written or printed on paper, or displayed on a
computer screen. However, no one imagines that a symbol is the same as the
thing it represents. The word ‘cat’ written in ink is not a cat. Similarly, the
symbol 2 is not the number two.
The meaning of ‘number’ is a surprisingly difficult conceptual and
philosophical problem. It is made all the more frustrating by the fact that we all
know perfectly well how to use numbers. We know how they behave, but not
what they are.

Sets and classes

In the 1880s Gottlob Frege tried to resolve this conceptual issue by constructing
whole numbers out of even simpler objects – namely sets, or classes as he called
them. His starting point was the standard association of numbers with counting.
According to Frege, two is a property of those sets – and only those – that can be
matched one-to-one with a standard set {a, b} having distinct members a and b.
So

{one cat, another cat}

{one bicycle, another bicycle }
{one loaf, another loaf}

can all be matched with {a, b}, so they all determine – whatever that means – the
same number.
Unfortunately, using a list of standard sets as numbers seems to beg the
question – it’s very like confusing a symbol with what it represents. But how can
we characterize ‘a property of those sets that can be matched one-to-one with a
standard set’? What is a property? Frege had a wonderful insight. There is a
well-defined set that is associated with any property; namely, the set consisting
of everything that possesses that property. The property ‘prime’ is associated
with the set of all prime numbers; the property ‘isosceles’ is associated with the
set of all isosceles triangles, and so on.
So Frege proposed that number two is the set comprising all sets that can be
matched one-to-one with the standard set {a, b}. More generally, a number is the
set of all sets that can be matched to any given set. So, for example, the number
3 is the set

{. . .{a, b, c}, {one cat, another cat, yet another cat}, {X, Y, Z},. . .}

although it is probably best to use mathematical objects in place of cats or

letters.
On this basis, Frege discovered that he could put all of the arithmetic of the
whole numbers on a logical basis. It all reduced to obvious properties of sets. He
wrote everything up in his masterwork Die Grundlagen der Arithmetik (The
Foundations of Arithmetic) of 1884, but to his bitter disappointment Georg
Cantor, a leading mathematical logician, dismissed the book as worthless. In
1893 Frege, undaunted, published the first volume of another book, Die
Grundgesetze der Arithmetik (The Basic Laws of Arithmetic), in which he
provided an intuitively plausible system of axioms for arithmetic. Peano
reviewed it, and everyone else ignored it. Ten years later, Frege was finally
ready to publish volume two, but by then he had noticed a basic flaw in his
axioms. Others noticed this as well. While volume two was in press, disaster
struck. Frege received a letter from the philosopher-mathematician Bertrand
Russell, to whom he had sent an advance copy of his book. To paraphrase, the
letter said roughly this: ‘Dear Gottlob, consider the sets of all sets that are not
members of themselves. Yours, Bertrand.’
Frege was a superb logician and he immediately got Russell’s point – indeed,
he was already aware of the potential for trouble. Frege’s whole approach had
assumed, without proof, that any reasonable property defined a meaningful set,
consisting of those objects that possess the property concerned. But here was an
apparently reasonable property, not a member of itself, which manifestly did not
correspond to a set.

Russell’s Paradox
A less formal version of the paradox proposed by Russell is the village barber, who shaves
 everyone who does not shave themselves. Who shaves the barber? If he does shave himself, then
by definition he is shaved by the village barber – himself! If he does not shave himself, then he is
shaved by the barber – which, again, is himself.
Aside from various cooks – the barber is a woman, for instance – the only possible conclusion
is that no such barber exists. Russell reformulated this paradox in terms of sets. Define a set X to
consist of all sets that are not members of themselves. Is X a member of itself, or not? If it is not,
then by definition it belongs to X – itself. If it is, a member of itself, then like all members of X, it
is not a member of itself. This time there is no way out – female sets are not yet part of the
mathematical enterprise.

A glum Frege penned an appendix to his magnum opus, discussing Russell’s

objection. He found a short-term fix: eliminate from the realm of sets any that
are members of themselves. But he was never really happy with this proposal.
Russell, for his part, tried to repair the gap in Frege’s construction of whole
numbers from sets. His idea was to restrict the kind of property that could be
used to define a set. Of course he had to find a proof that this restricted kind of
property never led to a paradox. In collaboration with Alfred North Whitehead,
he came up with a complicated and technical theory of types which achieved that
objective, to their satisfaction at least. They wrote their approach in a massive
three-volume tome Principia Mathematica of 1910–13. The definition of the
number 2 is near the end of volume one, and the theorem 1 + 1 = 2 is proved on
page 86 of volume two. However, Principia Mathematica did not end the
foundational debate. The theory of types was itself contentious. Mathematicians
wanted something simpler and more intuitive.

Cantor

These analyses of the fundamental role of counting as the basis for numbers led
to one of the most audacious discoveries in the whole of mathematics: Cantor’s
theory of transfinite numbers – different sizes of infinity.
Infinity, in various guises, seems unavoidable in mathematics. There is no
largest whole number – because adding one always produces a larger number
still – so there are infinitely many whole numbers. Euclid’s geometry takes place
on an infinite plane, and he proved that there are infinitely many prime numbers.
In the run-up to calculus, several people, among them Archimedes, found it
useful to think of an area or a volume as being the sum of infinitely many
infinitely thin slices. In the aftermath of calculus, the same picture of areas and
volumes was used for heuristic purposes, even if the actual proofs took a
different form.
These occurrences of the infinite could be rephrased in finite terms to avoid
various philosophical difficulties. Instead of saying ‘there are infinitely many
whole numbers’, for instance, we can instead say ‘there is no largest whole
number’. The second statement avoids explicit mention of the infinite, while
being logically equivalent to the first. Essentially, infinity is here being thought
of as a process, which can be continued without any specific limit, but is not
actually completed. Philosophers call this kind of infinity potential infinity. In
contrast, explicit use of infinity as a mathematical object in its own right is
actual infinity.
Mathematicians prior to Cantor had noticed that actual infinities had
paradoxical features. In 1632 Galileo wrote his Dialogue Concerning the Two
Chief World Systems, in which two fictional characters, the sagacious Salviati
and the intelligent layman Sagredo, discuss the causes of tides, from the
geocentric and heliocentric viewpoints. All mention of tides was removed at the
request of church authorities, making the book a hypothetical exercise which
nonetheless makes a powerful case for Copernicus’s heliocentric theory. Along
the way, the two characters discuss some of the paradoxes of infinity. Sagredo
asks ‘are there more numbers than squares?’ and points out that since most
whole numbers are not perfect squares, the answer must be yes. Salviati replies
that every number can be matched uniquely to its square:

Therefore the number of whole numbers must be the same as that of squares,
so the answer is no.
Cantor resolved these difficulties by recognizing that in the dialogue, the
adjective ‘more’ is being used in two different ways. Sagredo is pointing out that
the set of all squares is a proper subset of the set of all whole numbers. Salviati’s
position is subtler: he is arguing that there is a one-to-one correspondence
between the set of squares and the set of all whole numbers. Those statements
are different, and both can be true without leading to any contradiction.
 Pursuing this line of thought, Cantor was led to the invention of an arithmetic
of the infinite, which explained the previous paradoxes while introducing some
new ones. This work was part of a more extensive programme, Mengenlehre, the
mathematics of sets (Menge in German means set or assemblage). Cantor began
thinking about sets because of some difficult questions in Fourier analysis, so the
ideas were rooted in conventional mathematical theories. But the answers he
discovered were so strange that many mathematicians of the period rejected
them out of hand. Others, however, realized their value, notably David Hilbert,
who affirmed ‘No one shall expel us from the paradise that Cantor has created’.

Set size

Cantor’s starting point was the naive concept of a set, which is a collection of
objects, its members. One way to specify a set is to list the members, using curly
brackets. For instance, the set of all whole numbers between 1 and 6 is written

{1, 2, 3, 4, 5, 6}

Alternatively, a set can be specified by stating the rule for membership:

{n : 1 ≤ n ≤ 6 and n is a whole number}

The sets specified above are identical. The first notation is limited to finite
sets, but the second has no such limitations. Thus the sets

{n : n is a whole number}

and

{n : n is a perfect square}

are both specified precisely, and both are infinite.

One of the simplest things you can do with a set is count its members. How
big is it? The set {1, 2, 3, 4, 5, 6} has six members. So does the set {1, 4, 9, 16,
25, 36} consisting of the corresponding squares. We say that the cardinality of
the set is 6, and call 6 a cardinal number. (There is a different concept, ordinal
number, associated with putting numbers in order, which is why the adjective
‘cardinal’ is not superfluous here.) The set of all whole numbers cannot be
counted in this manner, but Cantor noticed that despite that, you can place the set
of all whole numbers and the set of all squares in one-to-one correspondence,
using the same scheme as Galileo. Each whole number n is paired with its square
n2.
Cantor defined two sets to be equinumerous (not his word) if there exists a
one-to-one correspondence between them. If the sets are finite, this property is
equivalent to ‘having the same number of members’. But if the sets are infinite,
it apparently makes no sense to talk of the number of members; nevertheless the
concept of equinumerosity makes perfect sense. But Cantor went further. He
introduced a system of transfinite numbers, or infinite cardinals, which did make
it possible to say how many members an infinite set has. Moreover, two sets
were equinumerous if and only if they had the same number of members – the
same cardinal.
The starting point was a new kind of number, which he denoted by the symbol
0. This is the Hebrew letter aleph with a subscript zero, read as aleph-null in
German, nowadays aleph-zero. This number is defined to be the cardinality of
the set of all whole numbers. By insisting that equinumerous sets have the same
cardinality, Cantor then required any set that can be placed in one-to-one
correspondence with the set of whole numbers also to have cardinality 0‫א‬. For
instance, the set of all squares has cardinality 0‫א‬. So does the set of all even
numbers:

And so does the set of all odd numbers:

One implication of these definitions is that a smaller set can have the same
cardinality as a bigger one. But there is no logical contradiction here to Cantor’s
definitions, so he considered this feature to be a natural consequence of his set-
up, and a price worth paying. You just have to be careful not to assume that
infinite cardinals behave just like finite ones. But why should they? They are not
finite!
Are there more integers (positive and negative) than whole numbers? Aren’t
there twice as many? No, because we can match the two sets like this:

The arithmetic of infinite cardinals is also strange. For instance, we have just
seen that the sets of even and odd whole numbers have cardinal 0‫א‬. Since these
sets have no members in common, the cardinal of their union – the set formed by
combining them – should, by analogy with finite sets, be 0‫ א‬+ 0‫א‬. But we know
what that union is: it is the whole numbers, with cardinal 0‫א‬. So apparently we
are forced to deduce that

0‫ = א‬0‫ א‬+ 0‫א‬

And so we are. But again, there is no contradiction: we cannot divide by 0‫א‬, to

deduce that 1 + 1 = 1, because 0‫ א‬is not a whole number and division has not
been defined, let alone been shown to make sense. Indeed, this equation shows
that division by 0‫ א‬does not always make sense. Again, we accept this as the
price of progress.
This is all very well, but it looks as though 0‫ א‬is just a fancy symbol for good
old ∞, and nothing new is really being said. Is it not the case that all infinite sets
have cardinal 0‫ ?א‬Surely all infinities are equal?
One candidate for an infinite cardinal greater than 0‫ – א‬that is, an infinite set
that cannot be put in one-to-one correspondence with the set of all whole
numbers – is the set of all rational numbers, which is usually denoted by . After
all, there are infinitely many rational numbers in the gap between any two
consecutive integers, and the kind of trick we used for the integers no longer
works.
However, in 1873 Cantor proved that also has cardinal 0‫א‬. The one-to-one
correspondence shuffles the numbers quite thoroughly, but no one said that they
had to remain in numerical order. It was beginning to look remarkably as though
every infinite set had cardinal 0‫א‬.
 In the same year, though, Cantor made a breakthrough. He proved that the set
of all real numbers does not have cardinal 0‫א‬, an astonishing theorem that he
published in 1874. So even in Cantor’s special sense, there are more reals than
integers. One infinity can be bigger than another infinity.
How big is the cardinal of the reals? Cantor hoped that it would be 1‫א‬, the next
largest cardinal after 0‫א‬. But he could not prove this, so he named the new
cardinal c, for continuum. The hoped-for equation c = 1‫ א‬was called the
continuum hypothesis. Only in 1960 did mathematicians sort out the relationship
between c and 0‫א‬, when Paul Cohen proved that the answer depends on which
axioms you choose for set theory. With some sensible axioms, the two cardinals
are the same. But with other, equally reasonable axioms, they are different.
Although the validity of the equation c = 1‫ א‬depends on the chosen axioms, an
associated equality does not. This is c = 20‫א‬. For any cardinal A we can define 2A
as the cardinal of the set of all subsets of A. And we can prove, very easily, that
2A is always bigger than A. This means that not only are some infinities bigger
than others: there is no largest infinite cardinal.

Contradictions

The biggest task of foundational mathematics, however, was not proving that
mathematical concepts exist. It was to prove that mathematics is logically
consistent. For all mathematicians knew – indeed, for all they know today –
there might be some sequence of logical steps, all perfectly correct, leading to an
absurd conclusion. Maybe you could prove that 2 + 2 = 5, or 1 = 0, for instance.
Or that 6 is prime, or π = 3.
Now, it might seem that one tiny contradiction would have limited
consequences. In everyday life, people often operate very happily within a
contradictory framework, asserting at one moment that, say, global warming is
destroying the planet, and a moment later that low-cost airlines are a great
invention. But in mathematics, consequences are not limited, and you can’t
evade logical contradictions by ignoring them. In mathematics, once something
is proved you can use it in other proofs. Having proved 0 = 1, much nastier
things follow. For instance, all numbers are equal. For if x is any number, start
from 0 = 1 and multiply by x. Then 0 = x. Similarly, if y is any other number, 0 =
y. But now x = y.
 Worse, the standard method of proof by contradiction means that anything can
be proved once we have proved that 0 = 1. To prove Fermat’s Last Theorem,
say, we argue like this:

Suppose Fermat’s Last Theorem is false.

Then 0 = 1.
Contradiction.
Therefore Fermat’s Last Theorem is true.

As well as being unsatisfying, this method also proves Fermat’s Last Theorem
is false:

Suppose Fermat’s Last Theorem is true.

Then 0 = 1.
Contradiction.
Therefore Fermat’s Last Theorem is false.

David Hilbert
1862–1943

David Hilbert graduated from the University of Königsberg in 1885 with a thesis on invariant
theory. He was on the university’s staff until taking up a professorship at Göttingen in 1895. He
 continued to work in invariant theory, proving his finite basis theorem in 1888. His methods were
more abstract than the prevailing fashion, and one of the leading figures in the field, Paul Gordan,
found the work unsatisfactory. Hilbert revised his paper for publication in the Annalen, and Klein
called it ‘the most important work on general algebra that [the journal] has ever published.’

In 1893 Hilbert began a comprehensive report on number theory, the Zahlbericht. Although this
was intended to summarize the known state of the theory, Hilbert included a mass of original
material, the basis of what we now call class field theory.
By 1899 he had changed fields again, now studying the axiomatic foundations of Euclidean
geometry. In 1923, at the Second International Congress of Mathematicians in Paris, he presented
a list of 23 major unsolved problems. These Hilbert problems had a tremendous effect on the
subsequent direction of mathematical research.
Around 1909 his work on integral equations led to the formulation of Hilbert spaces, now
basic to quantum mechanics. He also came very close to discovering Einstein’s equations for
general relativity in a 1915 paper. He added a note in proof to the effect that the paper was
consistent with Einstein’s equations, which gave rise to an erroneous belief that Hilbert might
have anticipated Einstein.

In 1930, on his retirement, Hilbert was made an honorary citizen of Königsberg. His acceptance
speech ended with the words ‘Wir müssen wissen, wir werden wissen’ (we must know, we shall
know) which encapsulated his belief in the power of mathematics and his determination to solve
even the most difficult problems.

Once everything is true – and also false – nothing meaningful can be said. The
whole of mathematics would be a silly game, with no content.

Hilbert

The next major foundations step was taken by David Hilbert, probably the
leading mathematician of his day. Hilbert had a habit of working in one area of
mathematics for about ten years, polishing off the main problems, and moving
on to a new area. Hilbert became convinced that it should be possible to prove
that mathematics can never lead to a logical contradiction. He also realized that
physical intuition would not be useful in such a project. If mathematics is
contradictory, it must be possible to prove that 0 = 1, in which case there is a
physical interpretation: 0 cows = 1 cow, so cows can disappear in a puff of
smoke. This seems unlikely. However, there is no guarantee that the
mathematics of whole numbers actually matches the physics of cows, and it is at
least conceivable that a cow might suddenly disappear. (In quantum mechanics,
this could happen, but with a very low probability.) There is a limit to the
number of cows in a finite universe, but there is no limit to the size of
mathematical integers. So physical intuition could be misleading, and should be
ignored.
Hilbert came to this point of view in his work on the axiomatic basis of
Euclid’s geometry. He discovered logical flaws in Euclid’s axiom system, and
realized that these flaws had arisen because Euclid had been misled by his visual
imagery. Because he knew that a line was a long thin object, a circle was round
and a point was a dot, he had inadvertently assumed certain properties of these
objects, without stating them as axioms. After several attempts, Hilbert put
forward a list of 21 axioms and discussed their role in Euclidean geometry in his
1899 Grundlagen der Geometrie (Foundations of Geometry).

‘After his success in geometry, Hilbert now set his

sights on a far more ambitious project.’

Hilbert maintained that a logical deduction must be valid, independently of the

interpretation imposed on it. Anything that relies on some particular
interpretation of the axioms, but fails in other interpretations, involves a logical
error. It is this view of axiomatics, rather than the specific application to
geometry, that is Hilbert’s most important influence on the foundations of
mathematics. In fact, the same point of view also influenced the content of
mathematics, by making it much easier – and more respectable – to invent new
concepts by listing axioms for them. Much of the abstraction of early 20th
century mathematics stemmed from Hilbert’s viewpoint.
It is often said that Hilbert advocated the idea that mathematics is a
meaningless game played with symbols, but this overstates his position. His
point was that in order to place the subject on a firm logical basis, you have to
think about it as if it is a meaningless game played with symbols. All else is
irrelevant to the logical structure. But no one who takes a serious look at
Hilbert’s mathematical discoveries, and his deep commitment to the subject, can
reasonably deduce that he thought he was playing a meaningless game.
 What logic did for them
Charles Lutwidge Dodgson, better known as Lewis Carroll, used his own formulation of a
branch of mathematical logic, now known as propositional calculus, to set and solve
logical puzzles. A typical example from his Symbolic Logic of 1896 is:

• Nobody, who really appreciates Beethoven, fails to keep silence while the Moonlight
Sonata is being played;
• Guinea-pigs are hopelessly ignorant of music;
• No one, who is hopelessly ignorant of music, ever keeps silence while the Moonlight
Sonata is being played.

The deduction is that no guinea-pig really appreciates Beethoven. This form of logical
argument is called a syllogism, and it goes back to classical Greece.

After his success in geometry, Hilbert now set his sights on a far more
ambitious project: to place the whole of mathematics on a sound logical footing.
He followed the work of the leading logicians closely, and developed an explicit
programme to sort out the foundations of mathematics once and for all. As well
as proving that mathematics was free of contradictions, he also believed that in
principle every problem could be solved – every mathematical statement could
either be proved or disproved. A number of early successes convinced him that
he was heading along the correct path, and that success was not far away.

Gödel

There was one logician, however, who remained unconvinced by Hilbert’s

proposal to prove that mathematics is logically consistent. His name was Kurt
Gödel, and his worries about Hilbert’s programme forever changed our view of
mathematical truth.
Before Gödel, mathematics was simply though to be true – and it was the
highest example of truth, because the truth of a statement like 2 + 2 = 4 was
something in the realm of pure thought, independent of out physical world.
Mathematical truths were not things that might be disproved by later
experiments. In this they were superior to physical truths, such as Newton’s
inverse square law of gravity, which was disproved by observations of the
motion of the perihelion of Mercury, supporting the new gravitational theory
suggested by Einstein.
After Gödel, mathematical truth turned out to be an illusion. What existed
were mathematical proofs, the internal logic of which might well be faultless,
but which existed in a wider context – foundational mathematics – where there
could be no guarantee that the entire game had any meaning at all. Gödel did not
just assert this: he proved it. In fact, he did two things, which together left
Hilbert’s careful, optimistic programme in ruins.

‘After Gödel, mathematical truth turned out to be an

illusion.’

Gödel proved that if mathematics is logically consistent, then it is impossible

to prove that. Not just that he could not find a proof, but that no proof exists. So,
remarkably, if you do succeed in proving that mathematics is consistent, it
immediately follows that it’s not. He also proved that some mathematical
statements can neither be proved nor disproved. Again, not just that he
personally could not achieve this, but that it is impossible. Statements of this
type are called undecidable.
He proved these statements initially within a particular logical formulation of
mathematics, that adopted by Russell and Whitehead in their Principia
Mathematica. To begin with, Hilbert thought that there might be a way out: find
a better foundation. But as the logicians studied Gödel’s work, it quickly became
apparent that the same ideas would work in any logical formulation of
mathematics, strong enough to express the basic concepts of arithmetic.

Kurt Gödel
1906–78
In 1923, when Gödel went to the University of Vienna, he was still unsure whether to study
mathematics or physics. His decision was influenced by lectures of a severely disabled
mathematician, Philipp Furtwängler (brother of the famous conductor and composer Wilhelm).
Gödel’s own health was fragile, and Furtwängler’s will to overcome his disabilities made a big
impression. In a seminar given by Moritz Schlick, Gödel began studying Russell’s Introduction
to Mathematical Philosophy, and it became clear that his future lay in mathematical logic.

His doctoral thesis of 1929 proved that one restricted logical system, first-order propositional
calculus, is complete – every true theorem can be proved and every false one can be disproved.
He is best known for his proof of ‘Gödel’s Incompleteness Theorems’. In 1931 Gödel published
his epic paper ‘Über formal unentscheidbare Sätze der Principia Mathematica und verwandter
Systeme’. In it, he proved that no system of axioms rich enough to formalize mathematics can be
logically complete. In 1931 he discussed his work with the logician Ernst Zermelo, but the
meeting fared badly, possibly because Zermelo had already made similar discoveries but had
failed to publish them.
In 1936 Schlick was murdered by a Nazi student, and Gödel had a mental breakdown (his
second). When he had recovered, he visited Princeton. In 1938 he married Adele Porkert, against
his mother’s wishes, and returned to Princeton, shortly after Austria had been incorporated into
Germany. When the Second World War started, he was worried that he might be called up into
the German Army, so he emigrated to the USA, travelling through Russia and Japan. In 1940 he
produced a second seminal work, a proof that Cantor’s continuum hypothesis is consistent with
the usual axioms for mathematics.

He became a US citizen in 1948, and spent the rest of his life in Princeton. Towards the end of
 his life he worried more and more about his health, and eventually became convinced that
someone was trying to poison him. He refused food, and died in hospital. To the end, he liked to
argue philosophy with his visitors.

An intriguing consequence of Gödel’s discoveries is that any axiomatic

system for mathematics must be incomplete: you can never write down a finite
list of axioms that will determine all true and false theorems uniquely. There was
no escape: Hilbert’s programme cannot work. It is said that when Hilbert first
heard of Gödel’s work he was extremely angry. His anger may well have been
directed at himself, because the basic idea in Gödel’s work is straightforward.
(The technical implementation of that idea is distinctly difficult, but Hilbert was
good with technicalities.) Hilbert probably realized that he should have seen
Gödel’s theorems coming.
Russell demolished Frege’s book with a logical paradox, the paradox of the
village barber who shaves everyone who does not shave himself: the set of all
sets that are not members of themselves. Gödel demolished Hilbert’s programme
with another logical paradox, the paradox of someone who says: this statement is
a lie. For in effect Gödel’s undecidable statement – on which all else rests – is a
theorem T that states: ‘this theorem cannot be proved’.

‘Gödel’s theorems changed the way we view the

logical foundations of mathematics.’

If every theorem can either be proved, or disproved, then Gödel’s statement T

is contradictory in both cases. Suppose T can be proved: then T states that T
cannot be proved, a contradiction. On the other hand, if T can be disproved, then
the statement T is false, so it is wrong to state that T cannot be proved. Therefore
T can be proved, another contradiction. So the assumption that every theorem
can either be proved or disproved tells us that T can be proved if and only if it
cannot be proved.

Where we are now?

Gödel’s theorems changed the way we view the logical foundations of

mathematics. They imply that currently unsolved problems may have no solution
at all –neither true nor false, but in the limbo of undecidability. And many
interesting problems have been shown to be undecidable. However, the effect of
Gödel’s work has not, in practice, extended much beyond the foundational area
where it took place. Rightly or wrongly, mathematicians working on the
Poincaré conjecture, or the Riemann hypothesis, spend their time either looking
for proofs or disproofs. They are aware that the problem may be undecidable,
and might even look for a proof of undecidability if they could see how to get
started. But most known undecidable problems have a self-referential feel to
them, and without that, a proof of undecidability seems unattainable anyway.

What logic does for us

A profound variant on Gödel’s incompleteness theorem was discovered by Alan Turing
in an analysis of which computations are feasible, published in 1936 as On Computable
Numbers, with an application to the Entscheidungsproblem (The decision problem).
Turing began by formalizing an algorithmic computation – one that follows a pre-stated
recipe – in terms of a so-called Turing machine. This is a mathematical idealization of a
device which writes symbols 0 and 1 on a moving tape according to specific rules. He
proved that the halting problem for Turing machines – does the computation eventually
stop for a given input? – is undecidable. This means that there is no algorithm that can
predict whether the computation halts or not.
Turing proved his result by assuming the halting problem to be decidable and
constructing a computation that halts if and only if it does not halt, a contradiction. His
result demonstrates that there are limits to computability. Some philosophers have
extended these ideas to determine limits on rational thinking, and it has been suggested
that a conscious mind cannot function algorithmically. However, the arguments here are
currently inconclusive. They do show that it is naïve to think that a brain works much like
a modern computer, but this may not imply that a computer cannot simulate a brain.

As mathematics built ever more complicated theories on top of earlier ones, the
superstructure of mathematics began to come to pieces because of unrecognized
assumptions that turned out to be false. In order to shore up the entire edifice,
serious work was needed on the foundations.
The subsequent investigations delved into the true nature of numbers, working
backwards from complex numbers to reals, to rationals and then to whole
numbers. But the process did not stop there. Instead, the number systems
themselves were reinterpreted in terms of even simpler ingredients, sets.
Set theory led to major advances, including a sensible, though unorthodox,
system of infinite numbers. It also revealed some fundamental paradoxes related
to the notion of a set. The resolution of these paradoxes was not, as Hilbert
hoped, a complete vindication of axiomatic mathematics, and a proof of its
logical consistency. Instead, it was a proof that mathematics has inherent
limitations, and that some problems do not have solutions. The upshot was a
profound change in the way we think about mathematical truth and certainty. It
is better to be aware of our limitations than to live in a fool’s paradise.
 18
How Likely is That?

The rational approach to chance

The growth of mathematics in the 20th and early 21st centuries has been
explosive. More new mathematics has been discovered in the last 100 years than
in the whole of previous human history. Even to sketch these discoveries would
take thousands of pages, so we are forced to look at a few samples from the
enormous amount of material that is available.
An especially novel branch of mathematics is the theory of probability, which
studies the chances associated with random events. It is the mathematics of
uncertainty. Earlier ages scratched the surface, with combinatorial calculations
of odds in gambling games and methods for improving the accuracy of
astronomical observations despite observational errors, but only in the early 20th
century did probability theory emerge as a subject in its own right.

Probability and statistics

Nowadays, probability theory is a major area of mathematics, and its applied

wing, statistics, has a significant effect on our everyday lives – possibly more
significant than any other single major branch of mathematics. Statistics is one
of the main analytic techniques of the medical profession. No drug comes to
market, and no treatment is permitted in a hospital, until clinical trials have
ensured that it is sufficiently safe, and that it is effective. Safety here is a relative
concept: treatments may be used on patients suffering from an otherwise fatal
disease when the chance of success would be far too small to use them in less
serious cases.
Probability theory may also be the most widely misunderstood, and abused,
area of mathematics. But used properly and intelligently, it is a major contributor
to human welfare.

Games of chance

A few probabilistic questions go back to antiquity. In the Middle Ages, we find

discussions of the chances of throwing various numbers with two dice. To see
how this works, let’s start with one die. Assuming that the die is fair – which
turns out to be a difficult concept to pin down – each of the six numbers 1, 2, 3,
4, 5 and 6 should be thrown equally often, in the long run. In the short run,
equality is impossible: the first throw must result in just one of those numbers,
for instance. In fact, after six throws it is actually rather unlikely to have thrown
each number exactly once. But in a long series of throws, or trials, we expect
each number to turn up roughly one time in six; that is, with probability 1/6. If
this didn’t happen, the die would in all likelihood be unfair or biased.
An event of probability 1 is certain, and one of probability 0 is impossible. All
probabilities lie between 0 and 1, and the probability of an event represents the
proportion of trials in which the event concerned happens.
Back to that medieval question. Suppose we throw two dice simultaneously
(as in numerous games from craps to Monopoly™). What is the probability that
their sum is 5? The upshot of numerous arguments, and some experiments, is
that the answer is 1/9. Why? Suppose we distinguish the two dice, colouring one
blue and the other red. Each die can independently yield six distinct numbers,
making a total of 36 possible pairs of numbers, all equally likely. The
combinations (blue + red) that yield 5 are 1 + 4, 2 + 3, 3 + 2, 4 + 1; these are
distinct cases because the blue die produces distinct results in each case, and so
does the red one. So in the long run we expect to find a sum of 5 on four
occasions out of 36, a probability of 4/36 = 1/9.
Another ancient problem, with a clear practical application, is how to divide
the stakes in a game of chance if the game is interrupted for some reason. The
Renaissance algebraists Pacioli, Cardano, and Tartaglia all wrote on this
question. Later, the Chevalier de Meré asked Pascal the same question, and
Pascal and Fermat wrote each other several letters on the topic.
From this early work emerged an implicit understanding of what probabilities
are and how to calculate them. But it was all very ill-defined and fuzzy.

Combinations

A working definition of the probability of some event is the proportion of

occasions on which it will happen. If we roll a die, and the six faces are equally
likely, then the probability of any particular face showing is 1/6. Much early work
on probability relied on calculating how many ways some event would occur,
and dividing that by the total number of possibilities.
A basic problem here is that of combinations. Given, say, a pack of six cards,
how many different subsets of four cards are there? One method is to list these
subsets: if the cards are 1 − 6, then they are

1234 1235 1236 1245 1246

1256 1345 1346 1356 1456
2345 2346 2356 2456 3456

so there are 15 of them. But this method is too cumbersome for larger numbers
of cards, and something more systematic is needed.
Imagine choosing the members of the subset, one at a time. We can choose the
first in six ways, the second in only five (since one has been eliminated), the
third in four ways, the fourth in three ways. The total number of choices, in this
order, is 6 × 5 × 4 × 3 = 360. However, every subset is counted 24 times – as
well as 1234 we find 1243, 2134, and so on, and there are 24 (4 × 3 × 2) ways to
rearrange four objects. So the correct answer is 360/24, which equals 15. This
argument shows that the number of ways to choose m objects from a total of n
objects is

These expressions are called binomial coefficients, because they also arise in
algebra. If we arrange them in a table, so that the nth row contains the binomial
coefficients

then the result looks like this:

In the sixth row we see the numbers 1, 6, 15, 20, 15, 6, 1. Compare with the
formula (x + 1)6 = x6 + 6x5 + 15x4 + 20x3 + 15x2 + 6x + 1 and we see the same
numbers arising as the coefficients. This is not a coincidence.

Pascal’s triangle

The triangle of numbers is called Pascal’s triangle because it was discussed

by Pascal in 1655. However, it was known much earlier; it goes back to about
950 in a commentary on an ancient Indian book called the Chandas Shastra. It
was also known to the Persian mathematicians Al-Karaji and Omar Khayyám,
and is known as the Khayyám triangle in modern Iran.
Probability theory

Binomial coefficients were used to good effect in the first book on probability:
the Ars Conjectandi (Art of Conjecturing) written by Jacob Bernoulli in 1713.
The curious title is explained in the book:

‘We define the art of conjecture, or stochastic art, as the art of evaluating as exactly
as possible the probabilities of things, so that in our judgments and actions we can
always base ourselves on what has been found to be the best, the most appropriate,
the most certain, the best advised; this is the only object of the wisdom of the
philosopher and the prudence of the statesman.’

So a more informative translation might be The Art of Guesswork.

Bernoulli took it for granted that more and more trials lead to better and better
estimates of probability.

‘Suppose that without your knowledge there are concealed in an urn 3000 white
pebbles and 2000 black pebbles, and in trying to determine the numbers of these
pebbles you take out one pebble after another (each time replacing the pebble. . .)
and that you observe how often a white and how often a black pebble is drawn. . .
Can you do this so often that it becomes ten times, one hundred times, one thousand
times, etc, more probable. . . that the numbers of whites and blacks chosen are in the
same 3:2 ratio as the pebbles in the urn, rather than a different ratio?’

Here Bernoulli asked a fundamental question, and also invented a standard

illustrative example, balls in urns. He clearly believed that a ratio of 3:2 was the
sensible outcome, but he also recognized that actual experiments would only
approximate this ratio. But he believed that with enough trials, this
approximation should become better and better.
There is a difficulty here, and it stymied the whole subject for a long time. In
such an experiment, it is certainly possible that by pure chance every pebble
removed could be white. So there is no hard and fast guarantee that the ratio
must always tend to 3:2. The best we can say is that with very high probability,
the numbers should approach that ratio. But now there is a danger of circular
logic: we use ratios observed in trials to infer probabilities, but we also use
probabilities to carry out that inference. How can we observe that the probability
of all pebbles being white is very small? If we do that by lots of trials, we have
to face the possibility that the result there is misleading, for the same reason; and
the only way out seems to be even more trials to show that this event is itself
highly unlikely. We are caught in what looks horribly like an infinite regress.

What probability did for them

In 1710 John Arbuthnot presented a paper to the Royal Society in which he used
probability theory as evidence for the existence of God. He analysed the annual number
christenings for male and female children for the period 1629–1710, and found that there
are slightly more boys than girls. Moreover, the figure was pretty much the same in
every year. This fact was already well known, but Arbuthnot proceeded to calculate the
probability of the proportion being constant. His result was very small, 2-82. He then
pointed out that if the same effect occurs in all countries, and at all times throughout
history, then the chances are even smaller, and concluded that divine providence, not
chance, must be responsible.
On the other hand, in 1872 Francis Galton used probabilities to estimate the efficacy
of prayer by noting that prayers were said every day, by huge numbers of people, for the
health of the royal family. He collected data and tabulated the ‘mean age attained by
males of various classes who had survived their 30th year, from 1758 to 1843,’ adding
that ‘deaths by accident are excluded.’ These classes were eminent men, royalty, clergy,
lawyers, doctors, aristocrats, gentry, tradesmen, naval officers, literature and science,
army officers, and practitioners of the fine arts. He found that ‘The sovereigns are
literally the shortest lived of all who have the advantage of affluence. The prayer has
therefore no efficacy, unless the very questionable hypothesis be raised, that the
conditions of royal life may naturally be yet more fatal, and that their influence is partly,
though incompletely, neutralized by the effects of public prayers.’

Fortunately, the early investigators of probability theory did not allow this
logical difficulty to stop them. As with calculus, they knew what they wanted to
do, and how to do it. Philosophical justification was less interesting than
working out the answers.
Bernoulli’s book contained a wealth of important ideas and results. One, the
Law of Large Numbers, sorts out in exactly what sense long-term ratios of
observations in trials correspond to probabilities. Basically, he proves that the
probability that the ratio does not become very close to the correct probability
tends to zero as the number of trials increases without limit.
 Another basic theorem can be viewed in terms of repeatedly tossing a biased
coin, with a probability p of giving heads, and q = 1 − p of giving tails. If the
coin is tossed twice, what is the probability of exactly 2, 1 or 0 heads?
Bernoulli’s answer was p2, 2pq and q2. These are the terms arising from the
expansion of (p + q)2 as p2 + 2pq + q2. Similarly if the coin is tossed three times,
the probabilities of 3, 2, 1 or 0 heads are the successive terms in (p + q)3 = p3 +
3p2q + 3q2p + q3.
More generally, if the coin is tossed n times, the probability of exactly m
heads is equal to

the corresponding term in the expansion of (p + q)n.

Between 1730 and 1738 Abraham De Moivre extended Bernoulli’s work on
biased coins. When m and n are large, it is difficult to work out the binomial
coefficient exactly, and De Moivre derived an approximate formula, relating
Bernoulli’s binomial distribution to what we now call the error function or
normal distribution

De Moivre was arguably the first person to make this connection explicit. It
was to prove fundamental to the development of both probability theory and
statistics.

Defining probability

A major conceptual problem in probability theory was to define probability.

Even simple examples – where everyone knew the answer – presented logical
difficulties. If we toss a coin, then in the long run, we expect equal numbers of
heads and tails, and the probability of each is ½. More precisely, this is the
probability provided the coin is fair. A biased coin might always turn up heads.
But what does ‘fair’ mean? Presumably, that heads and tails are equally likely.
But the phrase ‘equally likely’ refers to probabilities. The logic seems circular.
In order to define probability, we need to know what probability is.
 The way out of this impasse is one that goes back to Euclid, and was brought
to perfection by the algebraists of the late 19th and early 20th centuries.
Axiomatize. Stop worrying about what probabilities are. Write down the
properties that you want probabilities to possess, and consider those to be
axioms. Deduce everything else from them.
The question was: what are the right axioms? When probabilities refer to
finite sets of events, this question has a relatively easy answer. But applications
of probability theory often involved choices from potentially infinite sets of
possibilities. If you measure the angle between two stars, say, then in principle
that can be any real number between 0° and 180°. There are infinitely many real
numbers. If you throw a dart at a board, in such a way that in the long run it has
the same chance of hitting any point on the board, then the probability of hitting
a given region should be the area of that region, divided by the total area of the
dartboard. But there are infinitely many points on a dartboard, and infinitely
many regions.
These difficulties caused all sorts of problems, and all sorts of paradoxes.
They were finally resolved by a new idea from analysis, the concept of a
measure.
Analysts working on the theory of integration found it necessary to go beyond
Newton, and define increasingly sophisticated notions of what constitutes an
integrable function, and what its integral is. After a series of attempts by various
mathematicians, Henri Lebesgue managed to define a very general type of
integral, now called the Lebesgue integral, with many pleasant and useful
analytic properties.
The key to his definition was Lebesgue measure, which is a way to assign a
concept of length to very complicated subsets of the real line. Suppose that the
set consists of non-overlapping intervals of lengths 1, ½, ¼, ⅛ and so on. These
numbers form a convergent series, with sum 2. So Lebesgue insisted that this set
has measure 2. Lebesgue’s concept has a new feature: it is countably additive. If
you put together an infinite collection of non-overlapping sets, and if this
collection is countable in Cantor’s sense, with cardinal 0‫א‬, then the measure of
the whole set is the sum of the infinite series formed by the measures of the
individual sets.
In many ways the idea of a measure was more important than the integral to
which it led. In particular, probability is a measure. This property was made
explicit in the 1930s by Andrei Kolmogorov, who laid down axioms for
probabilities. More precisely, he defined a probability space. This comprises a
set X, a collection B of subsets of X called events and a measure m on B. The
axioms state that m is a measure, and that m(X) = 1 (that is, the probability that
something happens is always 1.) The collection B is also required to have some
set-theoretic properties that allow it to support a measure.
For a die, the set X consists of the numbers 1–6, and the set B contains every
subset of X. The measure of any set Y in B is the number of members of Y,
divided by 6. This measure is consistent with the intuitive idea that each face of
the die has probability 1/6 of turning up. But the use of a measure requires us to
consider not just faces, but sets of faces. The probability associated with some
such set Y is the probability that some face in Y will turn up. Intuitively, this is
the size of Y divided by 6.
With this simple idea, Kolmogorov resolved several centuries of often heated
controversy, and created a rigorous theory of probability.

Statistical data

The applied arm of probability theory is statistics, which uses probabilities to

analyse real world data. It arose from 18th century astronomy, when
observational errors had to be taken into account. Empirically and theoretically,
such errors are distributed according to the error function or normal distribution,
often called the bell curve because of its shape. Here the error is measured
horizontally, with zero error in the middle, and the height of the curve represents
the probability of an error of given size. Small errors are fairly likely, whereas
large ones are very improbable.

The bell curve

In 1835 Adolphe Quetelet advocated using the bell curve to model social data
– births, deaths, divorces, crime and suicide. He discovered that although such
events are unpredictable for individuals, they have statistical patterns when
observed for an entire population. He personified this idea in terms of the
‘average man’, a fictitious individual who was average in every respect. To
Quetelet, the average man was not just a mathematical concept: it was the goal
of social justice.
From about 1880 the social sciences began to make extensive use of statistical
ideas, especially the bell curve, as a substitute for experiments. In 1865 Francis
Galton made a study of human heredity. How does the height of a child relate to
that of its parents? What about weight, or intellectual ability? He adopted
Quetelet’s bell curve, but saw it as a method for separating distinct populations,
not as a moral imperative. If certain data showed two peaks, rather than the
single peak of the bell curve, then that population must be composed of two
distinct sub-populations, each following its own bell curve. By 1877 Galton’s
investigations were leading him to invent regression analysis, a way of relating
one data set to another to find the most probable relationship.

Quetelet’s graph of how many people have a given height: height runs horizontally, the number of people
vertically

Another key figure was Ysidor Edgeworth. Edgeworth lacked Galton’s vision,
but he was a far better technician, and he put Galton’s ideas on a sound
mathematical basis. A third was Karl Pearson, who developed the mathematics
considerably. But Pearson’s most effective role was that of salesman: he
convinced the outside world that statistics was a useful subject.

What probability does for us

A very important use of probability theory occurs in medical trials of new drugs. These
trials collect data on the effects of the drugs – do they seem to cure some disorder, or do
 they have unwanted adverse effects? Whatever the figures seem to indicate, the big
question here is whether the data are statistically significant. That is, do the data result
from a genuine effect of the drug, or are they the result of pure chance? This problem is
solved using statistical methods known as hypothesis testing. These methods compare
the data with a statistical model, and estimate the probability of the results arising by
chance. If, say, that probability is less than 0.01, then with probability 0.99 the data are
not due to chance. That is, the effect is significant at the 99% level. Such methods make
it possible to determine, with considerable confidence, which treatments are effective, or
which produce adverse effects and should not be used.

Newton and his successors demonstrated that mathematics can be a very

effective way to understand the regularities of nature. The invention of
probability theory, and its applied wing, statistics, made the same point about the
irregularities of nature. Remarkably, there are numerical patterns in chance
events. But these patterns show up only in statistical quantities such as long-term
trends and averages. They make predictions, but these are about the likelihood of
some event happening or not happening. They do not predict when it will
happen. Despite that, probability is now one of the most widely used
mathematical techniques, employed in science and medicine to ensure that any
deductions made from observations are significant, rather than apparent patterns
resulting from chance associations.
 19
Number Crunching

Calculating machines and computational mathematics

Mathematicians have always dreamed of building machines to reduce the

drudgery of routine calculations. The less time you spend calculating, the more
time you can spend thinking. From prehistoric times sticks and pebbles were
used as an aid to counting, and piles of pebbles eventually led to the abacus, in
which beads slide along rods to represent the digits of numbers. Especially as
perfected by the Japanese, the abacus could carry out basic arithmetic rapidly
and accurately in the hands of an expert. Around 1950 a Japanese abacus out-
performed a mechanical hand-calculator.

A dream comes true?

By the 21st century, the advent of electronic computers and the widespread
availability of integrated circuits (chips) had given machines a massive
advantage. They were far faster than the human brain or a mechanical device –
billions or trillions of arithmetical operations every second are now
commonplace. The fastest as I write, IBM’s Blue Gene/L, can perform one
quadrillion calculations (floating-point operations) per second. Today’s
computers also have vast memory, storing the equivalent of hundreds of books
ready for almost instant recall. Colour graphics have reached a pinnacle of
quality.

The rise of the computer

Earlier machines were more modest, but they still saved a lot of time and effort.
The first development after the abacus was probably Napier’s bones, or Napier’s
rods, a system of marked rods that Napier invented before he came up with
logarithms. Essentially they were the universal components of traditional long
multiplication. The rods could be used in place of pencil and paper, saving time
writing numerals, but they mimicked hand calculations.
In 1642 Pascal invented the first genuinely mechanical calculator, the
Arithmetic Machine, to help his father with his accounts. It could perform
addition and subtraction, but not multiplication or division. It had eight rotating
dials, so it effectively worked with eight-digit numbers. In the succeeding
decade, Pascal built fifty similar machines, most of which are preserved in
museums to this day.
In 1671 Leibniz designed a machine for multiplication, and built one in 1694,
remarking that ‘It is unworthy of excellent men to lose hours like slaves in the
labour of calculation, which could be safely relegated to anyone else if machines
were used.’ He named his machine the Staffelwalze (step reckoner). His main
idea was widely used by his successors.
One of the most ambitious proposals for a calculating machine was made by
Charles Babbage. In 1812 he said he ‘was sitting in the rooms of the Analytical
Society, at Cambridge, my head leaning forward on the table in a kind of dreamy
mood, with a table of logarithms lying open before me. Another member,
coming into the room, and seeing me half asleep, called out, “Well, Babbage,
what are you dreaming about?” to which I replied “I am thinking that all these
tables” (pointing to the logarithms) “might be calculated by machinery”.’
Babbage pursued this dream for the rest of his life, constructing a prototype
called the difference engine. He sought government funding for more elaborate
machines. His most ambitious project, the analytical engine, was in effect a
programmable mechanical computer. None of these machines was built,
although various components were made. A modern reconstruction of the
difference engine is in the Science Museum in London – and it works. Augusta
Ada Lovelace contributed to Babbage’s work by developing some of the first
computer programs ever written.
The first mass-produced calculator, the Arithmometer, was manufactured by
Thomas de Colmar in 1820. It employed a stepped drum mechanism and was
still in production in 1920. The next major step was the pin wheel mechanism of
the Swedish inventor Willgodt T. Odhner. His calculator set the pattern for
dozens, if not hundreds of similar machines, made by a variety of manufacturers.
The motive power was supplied by the operator, who turned a handle to revolve
a series of discs on which the digits 0–9 were displayed. With practice,
complicated calculations could be carried out at high speed. The scientific and
engineering calculations for the Manhattan project of the Second World War, to
make the first atomic bomb, were performed using such machines by a squad of
‘calculators’ – mainly young women. The advent of cheap powerful electronic
computers in the 1980s made mechanical calculators obsolete, but their use in
business and scientific computing was widespread until that time.
Calculating machines contribute more than just simple arithmetic, because
many scientific calculations can be implemented numerically as lengthy series of
arithmetical operations. One the earliest numerical methods, which solves
equations to arbitrarily high precision, was invented by Newton, and is
appropriately known as Newton’s method. It solves an equation f(x) = 0 by
calculating a series of successive approximations to a solution, each improving
on the previous one but based on it. From some initial guess, x1, improved
approximations x2, x3, . . ., xn, xn+1, are derived using the formula
 Newton’s method for solving an equation numerically

where fʹ is the derivative of f. The method is based on the geometry of the

curve y = f(x) near the solution. The point xn+1 is where the tangent to the curve at
xn crosses the x-axis. As the diagram shows, this is closer to x than the original
point.

Augusta Ada King, Countess of Lovelace

1815–52
 Augusta Ada was the daughter of the poet Lord Byron and Anne Milbanke. Her parents separated
a month after her birth and she never saw her father again. The child showed mathematical
ability, and unusually Lady Byron thought that mathematics was good training for the mind and
encouraged her daughter to study it. In 1833 Ada met Charles Babbage at a party, and soon after
she was shown his prototype difference engine, finding it fascinating and rapidly understanding
how it worked. She became Countess of Lovelace when her husband William was created an Earl
in 1838.
In her 1843 translation of Luigi Menabrea’s Notions sur la Machine Analytique de Charles
Babbage she added what are in effect sample programs of her own devising. She wrote that ‘The
distinctive characteristic of the Analytical Engine. . . is the introduction into it of the principle
which Jacquard devised for regulating, by means of punched cards, the most complicated patterns
in the fabrication of brocaded stuffs . . . We may say most aptly that the Analytical Engine
weaves algebraical patterns just as the Jacquard loom weaves flowers and leaves.’
At the age of 36 she developed uterine cancer, and died after a lengthy period of severe pain,
bled to death by her physicians.

A second important application of numerical methods is to differential

equations. Suppose we wish to solve the differential equation

given that x = x0 at time t = 0. The simplest method, due to Euler, is to

approximate by , where is very small. Then an approximation to
the differential equation takes the form

x(t + ε) = x(t) + εf(x(t))

Starting with x(0) = x0, we successively deduce the values of f(ε), f(2ε), f(3ε)
and, in general, f(nε) for any integer n > 0. A typical value for might be 10-6, say.
A million iterations of the formula tells us x(1), another million leads to x(2) and
so on. With today’s computers a million calculations are trivial, and this
becomes an entirely practical method.
However, the Euler method is too simple-minded to be fully satisfactory, and
numerous improvements have been devised. The best known are a whole class
of Runge–Kutta methods, named after the German mathematicians Carl Runge
and Martin Kutta, who invented their first such method in 1901. One of these,
the so-called fourth order Runge–Kutta method, is very widely used in
engineering, science and theoretical mathematics.
The needs of modern nonlinear dynamics have spawned several sophisticated
methods that avoid accumulating errors over long periods of time by preserving
certain structure associated with the exact solution. For example, in a mechanical
system without friction, the total energy is conserved. It is possible to set up the
numerical method so that at each step energy is conserved exactly. This
procedure avoids the possibility that the computed solution will slowly drift
away from the exact one, like a pendulum that is slowly coming to rest as it loses
energy.
More sophisticated still are symplectic integrators, which solve mechanical
systems of differential equations by explicitly and exactly preserving the
symplectic structure of Hamilton’s equations, which is a curious but hugely
important kind of geometry tailored to the two types of variable, position and
momentum. Symplectic integrators are especially important in celestial
mechanics, where – for example – astronomers may wish to follow the
movements of the planets in the solar system over billions of years. Using
symplectic integrators, Jack Wisdom, Jacques Laskar and others have shown that
the long-term behaviour of the solar system is chaotic, that Uranus and Neptune
were once much closer to the Sun than they are now, and that eventually
Mercury’s orbit will move towards that of Venus, so that one or other planet
may well be thrown out of the solar system altogether. Only symplectic
integrators offer any confidence that results over such long time periods are
accurate.

What numerical analysis did for them

Newton not only had to sort out patterns in nature: he had to develop effective methods
of calculation. He made extensive use of power series to represent functions, because
he could differentiate or integrate such series term by term. He also used them to
calculate values of functions, an early numerical method still in use today. One page of
his manuscripts, dating to 1665, shows a numerical calculation of the area under a
hyperbola, which we now recognize as the logarithmic function. He added together the
terms of an infinite series, working to an astonishing 55 decimal places.

Newton’s calculation of the area under a hyperbola

Computers need mathematics

As well as using computers to help mathematics, we can use mathematics to help

computers. In fact, mathematical principles were important in all early computer
designs, either as proof of concept or as key aspects of the design.
All digital computers in use today work with binary notation, in which
numbers are represented as strings of just two digits: 0 and 1. The main
advantage of binary is that it corresponds to switching: 0 is off and 1 is on. Or 0
is no voltage and 1 is 5 volts, or whatever standard is employed in circuit design.
The symbols 0 and 1 can also be interpreted within mathematical logic, as truth
values: 0 means false and 1 means true. So computers can perform logical
calculations as well as arithmetical ones. Indeed, the logical operations are more
basic, and the arithmetical operations can be viewed as sequences of logical
operations. Boole’s algebraic approach to the mathematics of 0 and 1, in The
Laws of Thought, provides an effective formalism for the logic of computer
calculations. Internet search engines carry out Boolean searches, that is, they
search for items defined by some combination of logical criteria, such as
‘containing the word “cat” but not containing “dog”’.

Algorithms

Mathematics has aided computer science, but in return computer science has
motivated some fascinating new mathematics. The notion of an algorithm – a
systematic procedure for solving a problem – is one. (The name comes from the
Arabian algebraist al-Khowarizmi.) An especially interesting question is: how
does the running time of an algorithm depends on the size of the input data?

‘. . . we have no idea whether any ‘sensible’ problem

is not-P.’

For example, Euclid’s algorithm for finding the highest common factor of two
whole numbers m and n, with m ≤ n, is as follows:

• Divide n by m to get remainder r.

• If r = 0 then the highest common factor is m: STOP.
• If r > 0 then replace n by m and m by r and go back to the start.

It can be shown that if n has d decimal digits (a measure of the size of the
input data to the algorithm) then the algorithm stops after at most 5d steps. That
means, for instance, that if we are given two 1000-digit numbers, we can
compute their highest common factor in at most 5000 steps – which takes a split
second on a modern computer.
The Euclidean algorithm has linear running time: the length of the
computation is proportional to the size (in digits) of the input data. More
generally, an algorithm has polynomial running time, or is of class P, if its
running time is proportional to some fixed power (such as the square or cube) of
the size of the input data. In contrast, all known algorithms for finding the prime
factors of a number have exponential running time – some fixed constant raised
to the power of the size of the input data. This is what makes the RSA
cryptosystem (conjecturally) secure.
Roughly speaking, algorithms with polynomial running time lead to practical
computations on today’s computers, whereas algorithms with exponential
running time do not – so the corresponding computations cannot be performed in
practice, even for quite small sizes of initial data. This distinction is a rule of
thumb: a polynomial algorithm might involve such a large power that it is
impractical, and some algorithms with running time worse than polynomial still
turn out to be useful.

‘The solution will win a million-dollar prize.’

The main theoretical difficulty now arises. Given a specific algorithm, it is

(fairly) easy to work out how its running time depends on the size of the input
data and to determine whether it is class P or not. However, it is extraordinarily
difficult to decide whether a more efficient algorithm might exist to solve the
same problem more rapidly. So, although we know that many problems can be
solved by an algorithm in class P, we have no idea whether any sensible problem
is not-P.
Sensible here has a technical meaning. Some problems must be not-P, simply
because outputting the answer requires non-P running time. For example list all
possible ways to arrange n symbols in order. To rule out such obviously non-P
problems, another concept is needed: the class NP of non-deterministic
polynomial algorithms. An algorithm is NP if any guess at an answer can be
checked in a time proportional to some fixed power of the size of input data. For
example, given a guess at a prime factor of a large number, it can quickly be
checked by a single division sum.
A problem in class P is automatically NP. Many important problems, for
which P algorithms are not known, are also known to be NP. And now we come
to the deepest and most difficult unsolved problem in this area, the solution of
which will win a million-dollar prize from the Clay Mathematics Institute. Are P
and NP the same? The most plausible answer is no, because P = NP means that a
lot of apparently very hard computations are actually easy – there exists some
short cut which we’ve not yet thought of.

What numerical analysis does for us

Numerical analysis plays a central role in the design of modern aircraft. Not so long ago,
engineers worked out how air would flow past the wings and fuselage of an aircraft using
wind-tunnels. They would place a model of the aircraft in the tunnel, force air past it with
a system of fans and observe the flow patterns. Equations such as those of Navier and
Stokes provided various theoretical insights, but it was impossible to solve them for real
aircraft because of their complicated shape.
Today’s computers are so powerful, and numerical methods for solving PDEs on
computers have become so effective, that in many cases the wind-tunnel approach has
been discarded in favour of a numerical wind-tunnel – that is, a computer model of the
aircraft. The Navier–Stokes equations are so accurate that they can safely be used in
this way. The advantage of the computer approach is that any desired feature of the flow
can be analysed and visualized.
 Numerical computation of flow of air past an aircraft

The P = NP? problem is made more difficult by a fascinating phenomenon,

called NP-completeness. Many NP problems are such that if they are actually
class P, then every NP problem is class P as well. Such a problem is said to be
NP-complete. If any particular NP-complete problem can be proved to be P, then
P = NP. On the other hand, if any particular NP-complete problem can be proved
to be not-P, then P is not the same as NP. One NP-complete problem that
attracted attention recently is associated with the computer game Minesweeper.
A more mathematical one is the Boolean Satisfiability Problem: given a
statement in mathematical logic, can it ever be true for some assignment of truth
values (true or false) for its variables?

Numerical analysis

Mathematics involves much more than calculations, but calculations are an

unavoidable accompaniment to more conceptual investigations. From the earliest
times, mathematicians have sought mechanical aids to free them from the
drudgery of calculation, and to improve the likelihood of accurate results. The
mathematicians of the past would have envied our access to electronic
computers, and marvelled at their speed and accuracy.
Calculating machines have done more for mathematics than just act as
servants. Their design and function have posed new theoretical questions for
mathematicians to answer. These questions range from justifying approximate
numerical methods for solving equations, to deep issues in the foundations of
computation itself.
At the start of the 21st century, mathematicians now have access to powerful
software, making it possible not just to perform numerical calculations on
computers, but to perform algebraic and analytic ones. These tools have opened
up new areas, helped to solve long-standing problems, and freed up time for
conceptual thinking. Mathematics has become much richer as a result, and it has
also become applicable to many more practical problems. Euler had the
conceptual tools to study fluid flow round complicated shapes, and even though
aircraft had not been invented, there are plenty of interesting questions about
ships in water. But he did not have any practical methods to implement those
techniques.
A new development, not touched on above, is the use of computers as an aid
to proof. Several important theorems, proved in recent years, rely on massive but
routine calculations, carried out by computer. It has been argued that computer-
assisted proofs change the fundamental nature of proof, by removing the
requirement that the proof can be verified by a human mind. This assertion is
controversial, but even if it is true, the result of the change is to make
mathematics into an even more powerful aid to human thought.
 20
Chaos and Complexity

Irregularities have patterns too

By the middle of the 20th century, mathematics was undergoing a rapid phase
of growth, stimulated by its widespread applications and powerful new methods.
A comprehensive history of the modern period of mathematics would occupy at
least as much space as a treatment of everything that led up to that period. The
best we can manage is a few representative samples, to demonstrate that
originality and creativity in mathematics are alive and well. One such topic,
which achieved public prominence in the 1970s and 1980s, is chaos theory, the
media’s name for nonlinear dynamics. This topic evolved naturally from
traditional models using calculus. Another is complex systems, which employs
less orthodox ways of thinking, and is stimulating new mathematics as well as
new science.
Chaos

Before the 1960s, the word chaos had only one meaning: formless disorder. But
since that time, fundamental discoveries in science and mathematics have
endowed it with a second, more subtle, meaning – one that combines aspects of
disorder with aspects of form. Newton’s Mathematical Principles of Natural
Philosophy had reduced the system of the world to differential equations, and
these are deterministic. That is, once the initial state of the system is known, its
future is determined uniquely for all time. Newton’s vision is that of a clockwork
universe, set in motion by the hand of the creator but thereafter pursuing a single
inevitable path. It is a vision that seems to leave no scope for free will, and this
may well be one of the early sources of the belief that science is cold and
inhuman. It is also a vision that has served us well, giving us radio, television,
radar, mobile phones, commercial aircraft, communications satellites, man-made
fibres, plastics and computers.
The growth of scientific determinism was also accompanied by a vague but
deep-seated belief in the conservation of complexity. This is the assumption that
simple causes must produce simple effects, implying that complex effects must
have complex causes. This belief causes us to look at a complex object or
system, and wonder where the complexity comes from. Where, for example, did
the complexity of life come from, given that it must have originated on a lifeless
planet? It seldom occurs to us that complexity might appear of its own accord,
but that is what the latest mathematical techniques indicate.

A single solution?

The determinacy of the laws of physics follows from a simple mathematical fact:
there is at most one solution to a differential equation with given initial
conditions. In Douglas Adams’s The Hitchhiker’s Guide to the Galaxy the
supercomputer Deep Thought embarked on a five million year quest for the
answer to the great question of life, the universe and everything, and famously
derived the answer, 42. This incident is a parody of a famous statement in which
Laplace summed up the mathematical view of determinism:

‘An intellect which at any given moment knew all the forces that animate nature and
the mutual positions of the beings that comprise it, if this intellect were vast enough
 to submit its data to analysis, it could condense into a single formula the movement
of the greatest bodies of the universe and that of the lightest atom: for such an
intellect nothing could be uncertain, and the future just like the past would be
present before its eyes.’

He then brought his readers down to earth with a bump, by adding:

‘Human mind offers a feeble sketch of this intelligence in the perfection which it has
been able to give to astronomy’.

Ironically, it was in celestial mechanics, that most evidently deterministic part

of physics, that Laplacian determinism would meet a sticky end. In 1886 King
Oscar II of Sweden (who also ruled Norway) offered a prize for solving the
problem of the stability of the solar system. Would our own little corner of the
clockwork universe keep ticking forever, or might a planet crash into the Sun or
escape into interstellar space? Remarkably, the physical laws of conservation of
energy and momentum do not forbid either eventuality – but could the detailed
dynamics of the solar system shed further light?
Poincaré was determined to win the prize, and he warmed up by taking a close
look at a simpler problem, a system of three celestial bodies. The equations for
three bodies are not much worse than those for two, and have much the same
general form. But Poincaré’s three-body warm-up turned out to be surprisingly
hard, and he discovered something disturbing. The solutions of those equations
were totally different from those of the two-body case. In fact, the solutions were
so complicated that they could not be written down as a mathematical formula.
Worse, he could understand enough of the geometry – more precisely, the
topology – of the solutions to prove, beyond any shadow of doubt, that the
motions represented by those solutions could sometimes be highly disordered
and irregular. ‘One is struck,’ Poincaré wrote, ‘with the complexity of this figure
that I am not even attempting to draw. Nothing can give us a better idea of the
complexity of the three-body problem.’ This complexity is now seen as a classic
example of chaos.

‘This complexity is now seen as a classic example of

chaos.’

His entry won King Oscar II’s prize, even though it did not fully solve the
problem posed. Some 60 years later, it triggered a revolution in how we view the
universe and its relation to mathematics.
In 1926–7 the Dutch engineer Balthazar van der Pol constructed an electronic
circuit to simulate a mathematical model of the heart, and discovered that under
certain conditions the resulting oscillation is not periodic, like a normal
heartbeat, but irregular. His work was given a solid mathematical basis during
the Second World War by John Littlewood and Mary Cartwright, in a study that
originated in the electronics of radar. It took more than 40 years for the wider
significance of their work to become apparent.

Poincaré’s Blunder
June Barrow-Green, delving into the archives of the Mittag-Leffler Institute in Stockholm,
recently discovered an embarrassing tale that had previously been kept quiet. The work that had
won Poincaré the prize contained a serious mistake. Far from discovering chaos, as had been
supposed, he had claimed to prove that it did not occur. The original submission proved that all
motions in the three-body problem are regular and well-behaved.
After the award of the prize, Poincaré belatedly spotted an error, and quickly discovered that it
demolished his proof completely. But the prize-winning memoir had already been published as an
issue of the Institute’s journal. The journal was withdrawn, and Poincaré paid for a complete
reprint, which included his discovery of homoclinic tangles and what we now call chaos. This
cost him significantly more than the money he had won with his flawed memoir. Almost all of the
copies of the incorrect version were successfully reclaimed and destroyed, but one, preserved in
the archives of the Institute, slipped through the net.

Nonlinear dynamics

In the early 1960s the American mathematician Stephen Smale ushered in the
modern era of dynamical systems theory by asking for a complete classification
of the typical types of behaviour of electronic circuits. Originally expecting the
answer to be combinations of periodic motions, he quickly realized that much
more complicated behaviour is possible. In particular he developed Poincaré’s
discovery of complex motion in the restricted three-body problem, simplifying
the geometry to yield a system known as ‘Smale’s horsehoe’. He proved that the
horsehoe system, although deterministic, has some random features. Other
examples of such phenomena were developed by the American and Russian
schools of dynamics, with especially notable contributions by Oleksandr
Sharkovskii and Vladimir Arnold, and a general theory began to emerge. The
term ‘chaos’ was introduced by James Yorke and Tien-Yien Li in 1975, in a
brief paper that simplified one of the results of the Russian school: Sharkovskii’s
Theorem of 1964, which described a curious pattern in the periodic solutions of
a discrete dynamical system – one in which time runs in integer steps instead of
being continuous.
Meanwhile, chaotic systems were appearing sporadically in the applied
literature – again, largely unappreciated by the wider scientific community. The
best known of these was introduced by the meteorologist Edward Lorenz in
1963. Lorenz set out to model atmospheric convection, approximating the very
complex equations for this phenomenon by much simpler equations with three
variables. Solving them numerically on a computer, he discovered that the
solution oscillated in an irregular, almost random manner. He also discovered
that if the same equations are solved using initial values of the variables that
differ only very slightly from each other, then the differences become amplified
until the new solution differs completely from the original one. His description
of this phenomenon in subsequent lectures led to the currently popular term,
butterfly effect, in which the flapping of a butterfly’s wings leads, a month later,
to a hurricane on the far side of the globe.
This weird scenario is a genuine one, but in a rather subtle sense. Suppose you
could run the world’s weather twice: once with the butterfly and once without.
Then you would indeed find major differences, quite possibly including a
hurricane on one run but no hurricane on the other. Exactly this effect arises in
computer simulations of the equations normally used to predict the weather, and
the effect causes big problems in weather forecasting. But it would be a mistake
to conclude that the butterfly caused the hurricane. In the real world, weather is
influenced not by one butterfly but by the statistical features of trillions of
butterflies and other tiny disturbances. Collectively, these have a definite
influence on where and when hurricanes form, and where they subsequently go.
 The Lorenz attractor

Using topological methods, Smale, Arnold and their coworkers proved that
the bizarre solutions observed by Poincaré were the inevitable consequence of
strange attractors in the equations. A strange attractor is a complex motion that
the system inevitably homes in on. It can be visualized as a shape in the state-
space formed by the variables that describe the system. The Lorenz attractor,
which describes Lorenz’s equations in this manner, looks a bit like the Lone
Ranger’s mask, but each apparent surface has infinitely many layers.
The structure of attractors explains a curious feature of chaotic systems: they
can be predicted in the short term (unlike, say, rolls of a die) but not in the long
term. Why cannot several short-term predictions be strung together to create a
long-term prediction? Because the accuracy with which we can describe a
chaotic system deteriorates over time, at an ever-growing rate, so there is a
prediction horizon beyond which we cannot penetrate. Nevertheless, the system
remains on the same strange attractor – but its path over the attractor changes
significantly.
This modifies our view of the butterfly effect. All the butterflies can do is
push the weather around on the same strange attractor – so it always looks like
perfectly plausible weather. It’s just a bit different from it what would have been
without all those butterflies.
David Ruelle and Floris Takens quickly found a potential application of
strange attractors in physics: the baffling problem of turbulent flow in a fluid.
The standard equations for fluid flow, called the Navier–Stokes equations, are
partial differential equations, and as such they are deterministic. A common type
of fluid flow, laminar flow, is smooth and regular, just what you would expect
from a deterministic theory. But another type, turbulent flow, is frothy and
irregular, almost random. Previous theories either claimed that turbulence was an
extremely complicated combination of patterns that individually were very
simple and regular, or that the Navier–Stokes equations broke down in the
turbulent regime. But Ruelle and Takens had a third theory. They suggested that
turbulence is a physical instance of a strange attractor.
Initially this theory was received with some scepticism, but we now know that
it was correct in spirit, even if some of the details were rather questionable.
Other successful applications followed, and the word chaos was enlisted as a
convenient name for all such behaviour.

Mary Lucy Cartwright

1900–98
 Mary Cartwright graduated from Oxford University in 1923, one of only five women studying
mathematics in the university. After a short period as a teacher, she took a doctorate at
Cambridge, nominally under Godfrey Hardy but actually under Edward Titchmarsh because
Hardy was in Princeton. Her thesis topic was complex analysis. In 1934 she was appointed
assistant lecturer at Cambridge, and in 1936 she was made director of studies at Girton College.
In 1938, in collaboration with John Littlewood, she undertook research for the Department of
Scientific and Industrial Research on differential equations related to radar. They discovered that
these equations had highly complicated solutions, an early anticipation of the phenomenon of
chaos. For this work she became the first woman mathematician to be elected a Fellow of the
Royal Society, in 1947. In 1948 she was made Mistress of Girton, and from 1959 to 1968 she was
a reader at the University of Cambridge. She received many honours, and was made Dame
Commander of the British Empire in 1969.

Theoretical monsters

A second theme now enters our tale. Between 1870 and 1930 a diverse
collection of maverick mathematicians invented a series of bizarre shapes the
sole purpose of which was to show up the limitations of classical analysis.
During the early development of calculus, mathematicians had assumed that any
continuously varying quantity must possess a well-defined rate of change almost
everywhere. For example, an object that is moving through space continuously
has a well-defined speed, except at relatively few instants when its speed
changes abruptly. However, in 1872 Weierstrass showed that this long-standing
assumption is wrong. An object can move in a continuous fashion, but in such an
irregular manner that – in effect – its speed is changing abruptly at every
moment of time. This means that it doesn’t actually have a sensible speed at all.

Stages in the construction of Hilbert’s space-filling curve and the Sierpinski gasket

Other contributions to this strange zoo of anomalies included a curve that fills
an entire region of space (one was found by Peano in 1890, another by Hilbert in
1891), a curve that crosses itself at every point (discovered by Waclaw
Sierpinski in 1915) and a curve of infinite length that encloses a finite area. This
last example of geometric weirdness, invented by Helge von Koch in 1906, is
the snowflake curve, and its construction goes like this. Begin with an equilateral
triangle, and add triangular promontories to the middle of each side to create a
six-pointed star. Then add smaller promontories to the middle of the star’s
twelve sides, and keep going forever. Because of its sixfold symmetry, the end
result looks like an intricate snowflake. Real snowflakes grow by other rules, but
that’s a different story.
 The snowflake curve

The mainstream of mathematics promptly denounced these oddities as

‘pathological’ and a ‘gallery of monsters’, but as the years went by several
embarrassing fiascos emphasized the need for care, and the mavericks’
viewpoint gained ground. The logic behind analysis is so subtle that leaping to
plausible conclusions is dangerous: monsters warn us about what can go wrong.
So, by the turn of the century, mathematicians had become comfortable with the
newfangled goods in the mavericks’ curiosity shop – they kept the theory
straight without having any serious impact on applications. Indeed by 1900
Hilbert could refer to the whole area as a paradise without causing ructions.
In the 1960s, against all expectations, the gallery of theoretical monsters was
given an unexpected boost in the direction of applied science. Benoit Mandelbrot
realized that these monstrous curves are clues to a far-reaching theory of
irregularities in nature. He renamed them fractals. Until then, science had been
happy to stick to traditional geometric forms like rectangles and spheres, but
Mandelbrot insisted that this approach was far too restrictive. The natural world
is littered with complex and irregular structures – coastlines, mountains, clouds,
trees, glaciers, river systems, ocean waves, craters, cauliflowers – about which
traditional geometry remains mute. A new geometry of nature is needed.
Today, scientists have absorbed fractals into their normal ways of thinking,
just as their predecessors did at the end of the 19th century with those maverick
mathematical monstrosities. The second half of Lewis Fry Richardson’s 1926
paper ‘Atmospheric diffusion shown on a distance–neighbour graph’ bears the
title ‘Does the wind have a velocity?’. This is now seen as an entirely reasonable
question. Atmospheric flow is turbulent, turbulence is fractal and fractals can
behave like Weierstrass’s monstrous function – moving continuously but having
no well defined speed. Mandelbrot found examples of fractals in many areas
both in and outside science – the shape of a tree, the branching pattern of a river,
the movements of the stock market.

Chaos everywhere!

The mathematicians’ strange attractors, when viewed geometrically, turned out

to be fractals, and the two strands of thought became intertwined in what is now
popularly known as chaos theory.
Chaos can be found in virtually every area of science. Jack Wisdom and
Jacques Laskar have found that the dynamics of the solar system is chaotic. We
know all the equations, masses and speeds that are required to predict the future
motion forever, but there is a prediction horizon of around ten million years
because of dynamical chaos. So if you want to know what side of the Sun Pluto
will be in AD 10,000,000 – forget it. These astronomers have also shown that
the Moon’s tides stabilize the Earth against influences that would otherwise lead
to chaotic motion, causing rapid shifts of climate from warm periods to ice ages
and back again; so chaos theory demonstrates that without the Moon, the Earth
would be a pretty unpleasant place to live.
Chaos arises in nearly all mathematical models of biological populations, and
recent experiments (letting beetles breed under controlled conditions) indicate
that it arises in real biological populations as well. Ecosystems do not normally
settle down to some kind of static balance of nature: instead they wander around
on strange attractors, usually looking fairly similar, but always changing. The
failure to understand the subtle dynamics of ecosystems is one reason why the
world’s fisheries are close to disaster.

Complexity

From chaos, we turn to complexity. Many of the problems facing today’s science
are extremely complicated. To manage a coral reef, a forest or a fishery it is
necessary to understand a highly complex ecosystem, in which apparently
harmless changes can trigger unexpected problems. The real world is so
complicated, and can be so difficult to measure, that conventional modelling
methods are hard to set up and even harder to verify. In response to these
challenges, a growing number of scientists have come to believe that
fundamental changes are needed in the way we model our world.
In the early 1980s George Cowan, formerly head of research at Los Alamos,
decided that one way forward lay in the newly developed theories of nonlinear
dynamics. Here small causes can create huge effects, rigid rules can lead to
anarchy and the whole often has capabilities that do not exist, even in
rudimentary form, in its components. In general terms, these are exactly the
features observed in the real world. But does the similarity run deeper – deep
enough to provide genuine understanding?

What Nonlinear dynamics did for them

Until nonlinear dynamics became a major issue in scientific modelling, its role was
mainly theoretical. The most profound work was that of Poincaré on the three-body
problem in celestial mechanics. This predicted the existence of highly complex orbits,
but gave little idea of what they looked like. The main point of the work was to
demonstrate that simple equations may not have simple solutions – that complexity is
not conserved, but can have simpler origins.

Modern computers can calculate complicated orbits in the three-body problem

Cowan conceived the idea of a new research institute devoted to the

interdisciplinary applications and development of nonlinear dynamics. He was
joined by Murray Gell-Mann, the Nobel-winning particle physicist, and in 1984
they created what was then called the Rio Grande Institute. Today it is the Santa
Fe Institute, an international centre for the study of complex systems.
Complexity theory has contributed novel mathematical methods and viewpoints,
exploiting computers to create digital models of nature. It exploits the power of
the computer to analyse those models and deduce tantalizing features of complex
systems. And it uses nonlinear dynamics and other areas of mathematics to
understand what the computers reveal.

Cellular automaton

In one type of new mathematical model, known as a cellular automaton, such

things as trees, birds and squirrels are incarnated as tiny coloured squares. They
compete with their neighbours in a mathematical computer game. The simplicity
is deceptive – these games lie at the cutting edge of modern science.
Cellular automata came to prominence in the 1950s, when John von Neumann
was trying to understand life’s ability to copy itself. Stanislaw Ulam suggested
using a system introduced by the computer pioneer Konrad Zuse in the 1940s.
Imagine a universe composed of a large grid of squares, called cells, like a giant
chessboard. At any moment, a given square can exist in some state. This
chessboard universe is equipped with its own laws of nature, describing how
each cell’s state must change as time clicks on to the next instant. It is useful to
represent that state by colours. Then the rules would be statements like: ‘If a cell
is red and has two blue cells next to it, it must turn yellow.’ Any such system is
called a cellular automaton – cellular because of the grid, automaton because it
blindly obeys whatever rules are listed.
To model the most fundamental feature of living creatures, von Neumann
created a configuration of cells that could replicate – make copies of itself. It had
200,000 cells and employed 29 different colours to carry around a coded
description of itself. This description could be copied blindly, and used as a
blueprint for building further configurations of the same kind. Von Neumann did
not publish his work until 1966, by which time Crick and Watson had discovered
the structure of DNA and it had become clear how life really does perform its
replication trick. Cellular automata were ignored for another 30 years.
 Cellular automaton

By the 1980s, however, there was a growing interest in systems composed of

large numbers of simple parts, which interact to produce a complicated whole.
Traditionally, the best way to model a system mathematically is to include as
much detail as possible: the closer the model is to the real thing, the better. But
this high-detail approach fails for very complex systems. Suppose, for instance,
that you want to understand the growth of a population of rabbits. You don’t
need to model the length of the rabbits’ fur, how long their ears are or how their
immune systems work. You need only a few basic facts about each rabbit: how
old it is, what sex it is, whether it is pregnant. Then you can focus your computer
resources on what really matters.

‘Complex systems support the view that with

sufficiently complex chemistry, life is likely to
 arise.’

For this kind of system, cellular automata are very effective. They make it
possible to ignore unnecessary detail about the individual components, and
instead to focus on how these components interrelate. This turns out to be an
excellent way to work out which factors are important, and to uncover general
insights into why complex systems do what they do.

Geology and biology

A complex system that defies analysis by traditional modelling techniques is the

formation of river basins and deltas. Peter Burrough has used cellular automata
to explain why these natural features adopt the shapes that they do. The
automaton models the interactions between water, land and sediment. The results
explain how different rates of soil erosion affect the shapes of rivers, and how
rivers carry soil away, important questions for river engineering and
management. The ideas are also of interest to oil companies, because oil and gas
are often found in geological strata that were originally laid down as sediment.
Another beautiful application of cellular automata occurs in biology. Hans
Meinhardt has used cellular automata to model the formation of patterns on
animals, from seashells to zebras. Key factors are concentrations of chemicals.
Interactions are reactions within a given cell and diffusion between neighbouring
cells. The two types of interaction combine to give the actual rules for the next
state. The results provide useful insights into the patterns of activation and
inhibition that switch pigment-making genes on and off dynamically during the
animal’s growth.
Stuart Kauffman has applied a variety of complexity-theoretic techniques to
delve into another major puzzle in biology: the development of organic form.
The growth and development of an organism must involve a great deal of
dynamics, and it cannot be just a matter of translating into organic form the
information held in DNA. A promising way forward is to formulate development
as the dynamics of a complex nonlinear system.
Cellular automata have now come full circle and given us a new perspective
on the origins of life. Von Neumann’s self-replicating automaton is enormously
special, carefully tailored to make copies of one highly complex initial
configuration. Is this typical of self-replicating automata, or can we obtain
replication without starting from a very special configuration? In 1993 Hui-
Hsien Chou and James Reggia developed a cellular automaton with 29 states for
which a randomly chosen initial state, or primordial soup, leads to self-
replicating structures more than 98 per cent of the time. In this automaton, self-
replicating entities are a virtual certainty.
Complex systems support the view that on a lifeless planet with sufficiently
complex chemistry, life is likely to arise spontaneously and to organize itself
into ever more complex and sophisticated forms. What remains to be understood
is what kinds of rule lead to the spontaneous emergence of self-replicating
configurations in our own universe – in short, what kind of physical laws make
this first crucial step towards life not only possible, but inevitable.

How mathematics was created

The story of mathematics is long and convoluted. As well as making remarkable

breakthroughs, the pioneers of mathematics headed off down blind alleys,
sometimes for centuries. But this is the way of the pioneer. If it is obvious where
to go next, anyone can do it. And so, over some four millennia, the elaborate,
elegant structure that we call mathematics came into being. It arose in fits and
starts, with wild bursts of activity followed by periods of stagnation; the centre
of activity moved around the globe following the rise and fall of human culture.
Sometimes it grew according to the practical needs of that culture; sometimes
the subject took off in its own direction, as its practitioners played what seemed
to everyone else to be mere intellectual games. And surprisingly often, those
games eventually paid off in the real world, by stimulating the development of
new techniques, new points of view and new understanding.

What Nonlinear dynamics does for us

It might seem that chaos has no practical applications, being irregular, unpredictable and
highly sensitive to small disturbances. However, because chaos is based on
deterministic laws, it turns out to be useful precisely because of these features.
One of the potentially most important applications is chaotic control. Around 1950 the
mathematician John von Neumann suggested that the instability of weather might one
day be turned to advantage, because it implies that a large desired effect can be
generated by a very small disturbance. In 1979 Edward Belbruno realized that this effect
could be used in astronautics to move spacecraft through large distances with very little
expenditure of fuel. However, the resulting orbits take a long time – two years from the
Earth to the Moon, for instance – and NASA lost interest in the idea.
In 1990 Japan launched a small Moon probe, Hagoromo, which separated from a
larger probe Hiten that stayed in Earth orbit. But the radio on Hagoromo failed, leaving
Hiten with no role to play. Japan wanted to salvage something from the mission, but
Hiten had only 10% of the fuel required to get it to the Moon using a conventional orbit.
An engineer on the project remembered Belbruno’s idea, and asked him to help. Within
ten months Hiten was on its way to the Moon and beyond, seeking trapped particles of
interstellar dust – with half of its remaining fuel unused. The technique has been used
repeatedly since this first success, in particular for the Genesis probe to sample the solar
wind, and ESA’s SMARTONE mission.

The Genesis probe (NASA).

The technique applies on Earth as well as in space. In 1990 Celso Grebogi, Edward
Ott and James Yorke published a general theoretical scheme to exploit the butterfly
effect in the control of chaotic systems. The method has been used to synchronize a
bank of lasers; to control heartbeat irregularities, opening up the possibility of an
intelligent pacemaker; to control electrical waves in the brain, which might help to
suppress epileptic attacks; and to smooth the motion of a turbulent fluid, which could in
 future make aircraft more fuel-efficient.

Mathematics has not stopped. New applications demand new mathematics,

and mathematicians are responding. Biology, especially, poses new challenges to
mathematical modelling and understanding. The internal requirements of
mathematics continue to stimulate new ideas, new theories. Many important
conjectures remain unsolved, but mathematicians are working on them.
Throughout its lengthy history, mathematics has taken its inspiration from
these two sources – the real world and the world of human imagination. Which
is most important? Neither. What matters is the combination. The historical
method makes it plain that mathematics draws its power, and its beauty, from
both. The time of the ancient Greeks is often seen as a historical Golden Age, as
logic, mathematics, and philosophy were brought to bear on the human
condition. But the advances made by the Greeks are just part of an ongoing
story. Mathematics has never been so active, it has never been so diverse, and it
has never been so vital to our society.

Welcome to the Golden Age of mathematics.

 Further Reading

Books and articles

E. Belbruno, Fly Me to the Moon, Princeton University Press, Princeton 2007.

E.T. Bell, Men of Mathematics (2 vols.), Pelican, Harmondsworth 1953.

E.T. Bell, The Development of Mathematics (reprint), Dover, New York 2000.

R. Bourgne and J.-P. Azra, Écrits et Mémoires Mathématiques d’Évariste Galois, Gauthier-Villars, Paris 1962.

C.B. Boyer, A History of Mathematics, Wiley, New York 1968.

W.K. Bühler, Gauss: a Biographical Study, Springer, Berlin 1981.

J. Cardan, The Book of My Life (translated by Jean Stoner), Dent, London 1931.

G. Cardano, The Great Art or the Rules of Algebra (translated T. Richard Witmer), MIT Press, Cambridge, MA 1968.

J. Coolidge, The Mathematics of Great Amateurs, Dover, New York 1963.

T. Dantzig, Number – the Language of Science (ed. J. Mazur), Pi Press, New York 2005.

Euclid, The Thirteen Books of Euclid’s Elements (3 vols, translated by Sir Thomas L. Heath), Dover, New York 1956.

J. Fauvel and J. Gray, The History of Mathematics – a Reader, Macmillan Education, Basingstoke 1987.

D.H. Fowler, The Mathematics of Plato’s Academy, Clarendon Press, Oxford 1987.

C.F. Gauss, Disquisitiones Arithmeticae, Leipzig 1801 (translated by A.A. Clarke), Yale University Press, New Haven, CT 1965.

A. Hyman, Charles Babbage, Oxford University Press, Oxford 1984.

G.G. Joseph, The Crest of the Peacock – non-European Roots of Mathematics, Penguin, Harmondsworth 2000.

V.J. Katz, A History of Mathematics (2nd edn), Addison-Wesley, Reading, MA 1998.

M. Kline, Mathematical Thought from Ancient to Modern Times, Oxford University Press, Oxford 1972.

A.H. Koblitz, A Convergence of Lives – Sofia Kovalevskaia, Birkhäuser, Boston 1983.

N. Koblitz, A Course in Number Theory and Cryptography (2nd edn.), Springer, New York 1994.

M. Livio, The Golden Ratio, Broadway, New York 2002.

M. Livio, The Equation That Couldn’t Be Solved, Simon & Schuster, New York 2005.

E. Maior, e – the Story of a Number, Princeton University Press, Princeton 1994.

E. Maior, Trigonometric Delights, Princeton University Press, Princeton 1998.

D. McHale, George Boole, Boole Press, Dublin 1985.

O. Neugebauer, A History of Ancient Mathematical Astronomy (3 vols.) Springer, New York 1975.
 O. Ore, Niels Hendrik Abel: Mathematician Extraordinary, University of Minnesota Press, Minneapolis 1957.

C. Reid, Hilbert, Springer, New York 1970.

T. Rothman, ‘The short life of Évariste Galois’, Scientific American (April 1982) 112–120. Collected in T. Rothman, A Physicist
on Madison Avenue, Princeton University Press 1991.

D. Sobel, Longitude (10th anniversary edn), HarperPerennial, New York 2005.

I. Stewart, Does God Play Dice? – The New Mathematics of Chaos, (2nd edition) Penguin, Harmondsworth 1997.

I. Stewart, Why Beauty is Truth, Basic Books, New York 2007.

S.M. Stigler, The History of Statistics, Harvard University Press, Cambridge, MA 1986.

B.L. van der Waerden, A History of Algebra, Springer-Verlag, New York 1994.

D. Welsh, Codes and Cryptography, Oxford University Press, Oxford 1988.

Internet
Most topics can be located easily using a search engine. Three very good general sites are: The MacTutor History of

Mathematics archive:

http://www-groups.dcs.st-and.ac.uk/~history/index.html

Wolfram MathWorld, a compendium of information on mathematical topics:

http://mathworld.wolfram.com

Wikipedia, the free online encyclopaedia:

http://en.wikipedia.org/wiki/Main_Page
 Index
Abacus 258

Abbott, Edwin 228

Abel, Niels 149, 152, 179, 180, 184

Abstract algebra 187-99

Abstract groups 192

Adam, Douglas 267

Alberti, Leone Battista 161

Algebra 54-68, 187-99

Algebras 195

Algebraic symbolism 64

Algoriths 262

Al-Khwarizmi 46, 60

Al-Kindi 46

Almagest 73, 75

Analyst, a Discourse Addressed to an Infidel Mathematician, The 117

Analytical Mechanics 227

Analytical Theory of Heat, The 150

A Plea for the Mathematician 225

Appollonius 22, 34, 83

Arabian mathematics 6, 47, 75, 175

Arbuthnot, John 252

Archimedean spiral 87

Archimedes 22, 30-32, 34, 83

Argand, Jean-Robert 141, 219

Aristarchus 71

Arithmetic Logarithmica 79

Arithmetic 40-53

Arithmetica 98

Arnold, Vladimir 269

Ars Conjectandi 251

Ars Magna 63

Art 161

Aryabhata 44

Astronomy 72

Atomic bomb 259

Babbage, Charles 258

Babylonian mathematics 6, 13-15, 21, 30, 175, 176

Barlow, William 182

Barrow, Isaac 114

Base ten logarithms 78

Belbruno, Edward 277

Beltrami, Eugenio 223

Berkeley, Bishop George 117, 149

Bernoulli, Daniel 153

Bernoulli family 88

Bernoulli, Jacob 251-3

Bernoulli, Johann 142, 150

Bessel, Friedrich 144, 167

Bessel functions 153

Bessel’s equation 153

Bhaskara 44, 45, 50

Bhaskaracharya 74

Binary numbers 229

Biology 276

Bolyai, Janos 168

Bolyai, Wolfgang 167

Bolzano, Bernard 149, 152, 157

Bombelli, Rafael 139, 140

Boolean algebra 197

Boole, George 197, 262

Borcherds, Richard 197

Brahe, Tycho 111

Brahmagupta 44

Bravais, Auguste 183

Briggs, Henry 78

Brunelleschi, Filippo 161

Burgi, Jobst 79

Burrough, Peter 276

Byron, Lord 260

Calculus 106-122, 148-158

Calculus of Extension, The 221

Canon Mathematicus 75

Cantor, Georg 236-8, 245, 254

Cardano, Girolamo 62-5, 139, 175, 179, 250

Carroll, Lewis 243

Cartesian coordinates 86

Cartwright, Mary Lucy 269, 271

Cauchy, Augustin-Louis 144, 146, 152, 157, 203

Cauchy-Riemann equations 146

Cauchy’s theorem 144, 146

Cayley, Arthur 192, 193, 223, 236

CDs 199

Cellular automaton 275

Chandas Shastra 251

Chaos theory 266-78

Chinese mathematics 49

Chou, Hui-Hsien 276

Christoffel, Elwin Bruno 223

Chuquet, Nicolas 67

Classes 235

Classification Theorem for Surfaces 209

Clay Institute 214

Closed universe 170

Cogitata Physica-Mathematica 96
Cohen, Paul 241

Colmar, Thomas de 259

Combinations 250

Complex analysis 141

Complex numbers 137-47

Complex systems 273-6

Comptes Rendus de l’Academie Francaise 146

Computational mathematics 257-65

Computers 258

Conic sections 34

Continuous functions 151

Conway, John 197

Coordinates 82-91, 227

Copernicus, Nicholas 110, 111, 122, 238

Cotes, Roger 143

Cowan, George 273

Coxeter, Harold Scott MacDonald 192

Coxeter diagrams 192

Craig, James 77

Crick, Francis 215, 275

Cubic equations 60, 138

Cuneiform numerals 21

D’Alembert 125, 126

Dedekind, Richard 193, 194, 232-4

Del Ferro 175

Della Pittura 161

Der Census Räumlicher Complexe 206

Desargues, Girard 163

Desargues’s Theorem 163

Descartes-Euler Formula 203

Descartes, Rene 66, 67, 84-7, 114, 140, 202, 218, 221

De Triangulis 75
Dialogue Concerning the Two Chief Systems of the World 114, 238

Die Grundgesetze der Arithmetik 236

Die Grundlagen der Arithmetik 236

Differential equations 120, 123-36

Differential geometry 222

Diophantus 50, 64, 93, 96

Dirichlet, Peter 149, 151

Discours de la Méthode 84

Discourses and Mathematical Demonstrations Concerning the Two New Sciences 113

Disquisitions Arithmeticae 99

DNA 202, 215

Dodgson, Charles Lutwidge 243

DVDs 199

Dyck, Walther van 193

Dynkin diagrams 192

Dynkin, Eugene 192

e 79

Eddin, Nasîr 75

Edgeworth, Ysidor 255

Egyptian mathematics 15-17

Einstein, Albert 169, 195, 213, 222, 223, 227, 242, 244

Electromagnetic Theory 221

Electromagnetism 127

Elements of Vector Analysis 221

Elements, The 25-9, 94, 218

Ellipses 33

Engel, Friedrich 191

Equations 56, 175

Eratosthenes 35

Erlangen Programme 169

Escher, Maurits 169

Euclid 22, 25-9, 37, 83, 93-5, 160, 166, 169, 172, 201, 242, 243, 254, 262
Euclidean geometry 21-39, 69-81

Euclides ab Omni Naevo Vindicatus 165

Euclid’s Fifth Postulate 161

Eudoxus 22

Euler, Leonhard 93, 127, 143, 150, 155-7, 202, 204, 205, 207, 260

Faraday, Michael 128, 135

Fedorov, Evgraf 182

Fermat, Pierre de 83, 97, 114, 140, 250

Fermat’s Last Theorem 93, 102, 177, 188, 194, 196-8, 241

Fermat’s Little Theorem 97

Ferrari 175

Fibonacci 47

Fibonacci Sequence 61

Fields 194

Fior 175

Fischer, Bernd 197

Flatland 228

Flat universe 170

Fluid dynamics 131

Four-dimensional space 217

Fourier analysis 148-58

Fourier, Joseph 129, 149-51, 182, 232

Fowler, David 29

Frege, Gottlob 235, 245

Frey, Gerhard 196, 198

Functions 87

Fundamental Theorem of Algebra 206

Furtwängler, Philipp 245

Galileo, Galilei 107, 113, 122, 238-41

Galois, Evaristel, 174, 180-5, 190, 193, 196, 197, 199

Galton, Francis 252, 255

Games of chance 249

Gauss, Carl Friedrich 93, 141, 144-6, 166-8, 170, 177, 178, 193, 206, 212, 219, 222

Gaussian integers 193

Gaussian numbers 193

Gell-Mann, Murray 274

General relativity 169, 170

General Theory of Equations 178

Geology 276

Geometry 21-39, 69-81, 159-72, 200-215, 216-29

Germain, Marie-Sophie 103

Gibbs, Josiah Willard 221

Global positioning system 18

Gödel, Kurt 244-6

Gödel’s incompleteness theorem 244-7

Godfrey, Thomas 90

Goldbach, Christian 97

Golden mean 28

Gordan, Paul 242

Gorenstein, Daniel 196

GPS 18

Grassmann, Hermann Günther 220

Gravity 128

Grebogi, Celso 277

Greek numerals 41

Greek mathematics 22-37, 41, 62, 83, 175

Griess, Robert 197

Group theory 173-86

Grundlagen der Geometrie 243

Hadley, John 90

Hamiltonian 134

Hamilton, Richard 213

Hamilton, William Rowan 134, 219

Hamming, Richard 230

Harrison, John 90

Heat 129

Heaviside, Oliver 221

Heiberg 31

Helmholtz, Hermann von 223

Hermes, J. 103

Hertz, Heinrich 135

High-dimensional geometry 216-29

Hilbert, David 242, 245

Hilbert Problems 242-5

Hindu mathematics 43-6

Hipparchus 72, 109

Hippasus 24

Hitchhiker’s Guide to the Galaxy, The 267

Hölder, Otto 184

Huntingdon, Edward 193

Hypatia 37

Hyperbolas 34

Hypothesis testing 256

Imaginary numbers 137-47

Incompleteness theorem 244-7

Indian numerals 43

Infinity 149-57, 207

Integers 138

Introduction to Mathematical Philosophy 245

Introduction to Plane and Solid Loci 83

Irrational numbers 24

Jordon, Camille 182-4

Jordon-Hölder Theorem184

Jupiter table 14

Kant, Immanuel 167, 169, 225

Karaji, Al 251

Kauffman, Stuart 276

Kepler, Johannes 7, 107, 110, 112, 113, 122, 132

Kepler’s Laws of Planetary Motion 112

Khayyám, Omar 62, 63, 175, 251

Killing, Wilhelm 191, 195

King, Augusta Ada 259, 260

Klein bottle 209

Klein, Felix 169, 188-90, 192, 198, 209

Kline, Morris 6

Klugel, Georg 166, 169

Kolmogorov, Andrei 254

Königsberg Bridges puzzle 202

Kovalevskaya, Sofia Vasilyevna 133

Kummer, Ernst Eduard 194

Kutta, Martin 260

Lacaille, Abbe Nicolas Louis de 81

La Géometrie 84

Lagrange, Joseph-Louis 93, 133, 134, 177, 182, 227

Lagrangian 134

L’Algebra 139

Lambert, Johann Heinrich 30, 166, 167

Lame, Gabriel 194

Laplace’s equation 130

Lasers 153

Laskar, Jaques 262

Law of Large Numbers 253

Laws of Thought, The 197, 262

Lebesgue, Henri 194, 254

Lebesgue measure 254

Lectures on Lineal Extension 221

Legendre, Adrien-Marie 164

Leibniz, Wilhelm Gottfried 107, 118-120, 124, 142, 143, 202, 258

Lejeune-Dirichlet, Peter 194

Leonardo of Pisa 47, 61

Levi-Civita, Tullio 223

Liber Abbaci 61

Lie groups 190-2

Lie, Sophus 188-192, 195

Limits 152

Listing, Johann 206

Li, Tien-Yien 269

Littlewood, John 269

Lobachevsky, Nikolai Ivanovich 167

Logarithmorum Canonis Constructio 78

Logarithmorum Chilias Prima 78

Logarithms 76-80

Logic 231-47

Longitude 90

Lorenz, Edward 269

Luminet, Jean Pierre 171

Maclaurin, Colin 128

Mahavira 44

Mandelbrot, Benoir 272

Marconi, Guglielmo 135

Mathematics of Plato’s Academy, The 29

Mathematical Reviews 9

Mathematical Syntaxis 73

Matrix algebra 223

Maxwell, James Clark 128, 135, 221

Maxwell’s equations 135

Mayan numerals 51

Meinhardt, Hans 276

Mengenlehre 238
Meré, Chevalier de 250

Mersenne primes 96

Mesopotamian symbols 10

Method of Fluxions 107

Method of Fluxions and Infinite Series 117

Method of Mechanical Theorems 32

Millennium Mathematics Problems 214

Minkowski, Hermann 224

Minkowski space-time 224

Mirimanoff, Dimitri 194

Möbius, Augustus 206

Möbius band 206

Möbius transformations 169

Moivre, Abraham de 253

Monstrous Moonshine Conjecture 197

Moore, Eliakim 193

Morgan, Augustus de 28

Muller, Johannes 75

Multidimensional space 225

Music 127

Mysterium Cosmographicum 110

Napier, John 76

Napierian logarithms 77

Napier’s Rods 76

Natural irrationalities, theorem on 179

Navier-Stokes equations 264, 270, 271

Negative numbers 49

Neumann, John von 275

New Logarithmes 79

Newton, Isaac 7, 67, 76, 90, 107, 116-18, 120, 122, 124, 128, 132, 136, 140, 154, 254, 259, 261, 267

Noether, Emmy Amalie 195

Noether’s Theorem 195

Non-Euclidean geometry 159-72, 200-215

Nonlinear dynamics 266-77

Notions sur la Machine Analytique de Charles Babbage 260

Numbers 8-19, 40-53

Number theory 92-105

Numerals 8-19, 40-53

Numerical analysis 257-65

Odhner, Willgodt T. 259

On Plane Loci 83

On the Revolutions of the Heavenly Spheres 110

On the Sizes and Distances of the Sun and the Moon 71

On the Sphere and the Cylinder 30

Open universe 170

Ott, Edward 277

Pacioli 250

Parabolas 34

Pascal 251, 258

Pascal’s triangle 251

Pearson, Karl 256

Peono, Giuseppe 234, 236

Perelman, Grigori 213

Peuerbach, George 75

Phidias 71

pi 30, 32

Plane trigonometry 77

Planetary motion 112

Playfair, John 164

Plea for the Mathematician, A 225

Plücker, Julius 188, 220

Poincaré, Jules Henri 174, 210-14, 246, 269, 270, 274

Pol, Balthazar van der 268

Polyhedrons 202
Portsmouth Papers 107

Power series 155

Primes 93-7

Principia Mathematica 107

Probability 248-56

Probability theory 251

Prosthapheiresis 77

Pythagorean triples 96

Ptolemy 73, 109-111

Ptolemy’s Theorem 73

Pythagoras 22, 24

Pythagoras’s Theorem 22, 26, 83, 96, 155, 224

Quaternions 219

Quetelet, Adolphe 255

Recorde, Robert 68

Reed, Irving 199

Reed-Solomon codes 199

Reggia, James 276

Regiomontanus 75

Regular polyhedra 27

Rhabdologia 78

Rhaeticus, George Joachim 75

Ribet, Kenneth 198

Ricci-Curbastro, Gregorio 213, 223

Ricci tensor 213

Richardson, Lewis Fry 273

Richelot, F.J. 103

Riemann 156, 189, 207, 222, 246

Riemann Hypothesis 156

Rings 194

Roman numerals 41

RSA code 104

Ruelle, David 270

Ruffini, Paolo 178, 180

Runge, Carl 260

Runge-Kutta methods 260-1

Russell, Bertrand 236, 237, 245

Russell’s paradox 237

Saccheri, Gerolamo 165, 166

Salmon, George 223

Satnav 18

Schlick, Moritz 245

Schmandt-Besserat, Denise 10

Schönflies, Arthur 182

Science and Hypothesis 210

Science and Method 210

Sets 235

Sharkovskii, Oleksandr 269

Sharkovskii’s Theorem 269

Sierpinski, Waclaw 272

Smale, Stephen 269

Snoflake curve 272

Solomon, Gustave 199

Speidell, John 79

Statistics 248-56

Staudt, Karl von 206

Stetigkeit und Irrationale Zahlen 232

Stevin, Simon 48, 49

Stokes, George Gabriel 132

Summerian numerals 12

Sylvester, James Joseph 225

Syllogism 243

Symbolic Logic 243

Symmetry 173-86
Symmetry of a quadratic 176

Takens, Floris 270

Tally marks 10, 12

Taniyama-Shimura conjecture 196

Taniyama-Weil conjecture 197

Tartaglia 62, 139, 175, 250

Taylor, Richard 196

Tentamen Juventum Studiosam in Elementa Matheseos 168

Theorie Analytique de la Chaleur 130

Thurston, William 211

Time Machine, The 217

Topology 200-215

Traité de Substitutions et des Équations Algébriques 182

Transformationsgruppen 191

Triangles 69-81

Treatise on the Quadrilateral 75

Trigonometry 69-81

Turing, Alan 185, 246

Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme 245

Universe, shape of 171

Value of Science, The 210

Varahamihira 74

Vieta 66-8

Vorstudien zur Topologie 206

Wallis, John 140, 219

Walther, Bernard 75

Wantzel, Pierre 178

Was Sind und was Sollen die Zahlen? 232

Watson, James 215, 275

Wave equations 126

Weierstrass 149, 155, 157

Weil, André 198

Wells, Herbert George 217

Wessel, Caspar 141

Whetstone of Witte, The 66

Whitehead, Alfred North 237

Wiles, Andrew 7, 196

Wilkinson Microwave Anisotropy Probe 171

Wisdom, Jack 262

Yorke, James 269, 277

Zermelo, Ernst 245

Zuse, Konrad 275

 Acknowledgements

p.2/3 ©Tetra Images/Corbis; p.11 ©Museum of Natural History Belgium; p.14

©Visual Arts Library (London)/Alamy; p.20 ©The Print Collector/Alamy; p.21
©Bill Casselman, courtesy of the Yale Babylonian Collection, holders of the
YBC7289 tablet; p.28 Painting of Euclid Greek mathematician by Justus von
Ghent/©Bettmann/Corbis; p.31 t ©Hulton-Deutsch Collection/Corbis, b ©Time
Life Pictures/ Getty Images; p.33 ©Maiman Rick/Corbis Sygma; p.36 ©Charles
Bowman/Alamy; p.37 ©Bettmann/Corbis; p.39 ©RubberBall/Alamy; p.40
©Tetra Images/Corbis; p.45 ©Hulton-Deutsch/Corbis; p.48 ©Bettmann/Corbis;
p.52/53 ©iStockphoto/Alija p.54 © Bettmann/Corbis; p.56 ©David Lees/Corbis;
p.63 ©Bettmann/Corbis; p.65 ©Science Source/Science Photo Library; p.69
©Sheila Terry/Science Photo Library; p.82 ©Comstock Select/Corbis; p.85 l
©Bettmann/Corbis, r Reproduced with the permission of the Brotherton
Collection, Leeds University Library; p.92 ©National Archaeological Museum,
Athens; p.98 ©Bettmann/Corbis; p.100 ©Bettmann/Corbis; p.101 ©National
Archaeological Museum, Athens; p.103 Credit: Sophie Germain (1776-1831),
illustration from ‘Histoire du Socialisme’, c.1880 by Leray, Auguste; Private
Collection/ Archives Charmet/ The Bridgeman Art Library p.106 Courtesy
NASA/JPL-Caltech; p.110 ©Bettmann/Corbis; p.112 ©Bettmann/Corbis; p.114
©Bettmann/Corbis; p.117 ©Bettmann/Corbis; p.119 t ©Burke/Triolo
Productions/Brand X/Corbis, b ©Martyn Goddard/Corbis; p.121 Courtesy
NASA/JPL-Caltech; p.123 ©nagelestock.com/Alamy; p.132 ©Jack
Newton/Phototake Inc/Alamy; p.133 ©Bettmann/Corbis; p.137 ©Werner H.
Muller/Corbis; p.144 ©Werner H. Muller/Corbis; p.145 ©Bettmann/Corbis;
p.159 Perspective study of a chalice, 1430-40 (pen & ink on paper) by Uccello,
Paolo (1397-1475) Gabinetto dei Disegni e Stampe, Uffizi, Florence, Italy/
Alinari/ The Bridgeman Art Library Nationality; p.162 ©Stapleton
Collection/Corbis; p.173 ©Robert Yin/Corbis; p.181 ©Bettmann/Corbis, b J-L
Charmet/Science Photo Library; p.185 t ©Los Alamos National
Laboratory/Science Photo Library, b ©Robert Yin/Corbis; p.189 ©Science
Photo Library; p.195 ©Science Photo Library; p196 © C. J. Mozzochi, Princeton
N.J.; p.200 M.C. Escher’s “Mobius II” © 2008 The M.C. Escher Company-
Holland. All rights reserved. www.mcescher.com; p.210 ©Hulton-Deutsch
Collection/Corbis; p.214 ©epa/Corbis; p.215 Phototake Inc./Alamy; p.219
©Hulton-Deutsch Collection/Corbis; p.231 ©The Barber by Amman, Jost
(1539-91)Bibliotheque Nationale, Paris, France/ Giraudon/ The Bridgeman Art
Library; p.242 ©Bettman/Corbis; p.245 ©Alfred Eisenstaedt/Time Life
Pictures/Getty Images; p.248 ©ImageBroker/Alamy; p.257 NASA/Science
Photo Library; p.260 ©Science Photo Library; p.261 ©Cambridge University
Library; p.264 NASA/Science Photo Library; p.270 ©Prof. E. Lorenz, Peter
Arnold Inc., Science Photo Library; p.270 Image kindly supplied by Girton
College, Cambridge; p.277 Artist’s conception courtesy of JPL

All drawings by Tim Oliver