I

Mesopotantia, 1800 BCE

The Pythagorean Theorem We would more properly have to call

''Babylonian" many things which the Greek
tradition had brought down to us as
" Pythagorean . "
-Otto Neugeba uer, quoted in Ba rtel van der Wat'rdt'n,
•:• A 4,000-Year History Science Awakening, p . 77
Maor- The Pythag. Thm.

The vast region stretching from the Euphrates and Tigris Rivers in the east to
the mountains of Lebanon in the west is known as the Fertile Crescent. It was
here, in modem Iraq, that one of the great civilizations of antiquity rose to
ELI MAOR prominence four thousand years ago: Mesopotamia. Hundreds of thousands of
clay tablets, found over the past two centuries, attest to a people who flour­
ished in commerce and architecture, kept accurate records of astronomical
events, excelled in the arts and literature, and, under the rule of Hammurabi,
created the first legal code in history. Only a small fraction of this vast archeo­
logical treasure trove has been studied by scholars; the great majority of tablets
lie in the basements of museums around the world, awaiting their turn to be
deciphered and give us a glimpse into the daily life of ancient Babylon.
Among the tablets that have received special scrutiny is one with the unas­
suming designation "Y BC 7289," meaning that it is tablet number 7289 in the
Babylonian Collection of Yale University (fig. 1 . 1 ). The tablet dates from the
Old Babylonian period of the Hammurabi dynasty, roughly 1 800- 1 600 BCE. It
shows a tilted square and its two diagonals, with some marks engraved along
one side and under the horizontal diagonal. The marks are in cuneiform
(wedge-shaped) characters, carved with a stylus into a piece of soft clay which
was then dried in the sun or baked in an oven. They turn out to be numbers,
written in the peculiar Babylonian numeration system that used the base 60.
In this sexagesimal system, numbers up to 59 were written in essentially our
modem base-ten numeration system, but without a zero. Units were written as
vertical Y-shaped notches, while tens were marked with similar notches written
horizontally. Let us denote these symbols by I and -, respectively. The number
2007

P R I N C E T O N U N I V E R S I T Y P R E S S Princeton and Oxford 23, for example, would be written as - - I I I . When a number exceeded 59,
 Mesopotam·1a, 1 800 BCE ..•.. 5 6 •••• Chapter I

it was arranged in groups of 60 in much the same way as we bunch numbers

into groups of ten in our base-ten system. Thus, 2,4 13 in the sexagesimal system
is 40 x 60 + 13, which was written as - - - - - 1 I I (often a group of
several identical symbols was stacked, evidently to save space).
Because the Babylonians did not have a symbol for the "empty slot"-our
modern zero-there is often an ambiguity as to how the numbers should be
grouped. In the example just given, the numerals - - - - - 1 I I could
also stand for 40 x 602 + 13 x 60 = 144,780; or they could mean 40/60 +
13 = 13. 166, or any other combination of powers of 60 with the coefficients
40 and 13. Moreover, had the scribe made the space between - - - - and
- 1 I I too small, the number might have erroneously been read as - - -
- - 1 I I , that is, 50 x 60 + 3 = 3,003. In such cases the correct interpretation
must be deduced from the context, presenting an additional challenge to schol­
ars trying to decipher these ancient documents.
Luckily, in the case of Y BC 7289 the task was relatively easy. The number
along the upper-left side is easily recognized as 30. The one immediately un­
der the horizontal diagonal is I ;24,5 1, I O ( we are using here the modern nota­
tion for writing Babylonian numbers, in which commas separate the sexagesi­
Maor- The Pythag. Thm.

mal "digits," and a semicolon separates the integral part of a number from its
fractional part). Writing this number in our base- I O system, we get
I + 24/60 + 5 1/602 + I 0/60 3 = 1 .4 142 13, which is none other than the decimal
value of ✓2 , accurate to the nearest one hundred thousandth! And when this
number is multiplied by 30, we get 42.426389, which is the sexagesimal num­
ber 42;25,35-the number on the second line below the diagonal. The conclu­
sion is inescapable: the Babylonians knew the relation between the length of
the diagonal of a square and its side, d = a ✓2. But this in turn means that they
were familiar with the Pythagorean theorem-or at the very least, with its spe­
cial case for the diagonal of a square (d 2 = a2 + a2 = 2a 2 )-more than a thou­
sand years before the great sage for whom it was named.
Two things about this tablet are especially noteworthy. First, it proves that
the Babylonians knew how to compute the square root of a number to a re­
markable accuracy-in fact, an accuracy equal to that of a modem eight-digit
calculator. ' But even more remarkable is the probable purpose of this particu­
lar document: by all likelihood, it was intended as an example of how to find
the diagonal of any square: simply multiply the length of the side by
I ;24,5 1, I0. Most people, when given this task, would fallow the "obvious"
but more tedious route: start with 30, square it, double the result, and take the
square root: d = ✓30 2 + 30 2 = ✓I 800 = 42. 4264, rounded to four places.
But suppose you had to do this over and over for squares of different sizes;
Figure l . l . Y BC 7289

you would have to repeat the process each time with a new number, a rather
tedious task. The anonymous scribe who carved these numbers into a clay
tablet nearly four thousand years ago showed us a simpler way: just multiply
the side of the square by ✓2 (fig. 1.2). Some simplification!
 Mesopotamia, 1 800 BCE •:• 7 8 •••• Chapter I

Figure 1 .2. A square and its diagonal

But there remains one unanswered question: why did the scribe choose a
Maor- The Pythag. Thm.

side of 30 for his example? There are two possible explanations: either this
tablet referred to some particular situation, perhaps a square field of side 30 for
which it was required to find the length of the diagonal; or-and this is more Figure 1 .3 . Piimpton 322
plausible-he chose 30 because it is one-half of 60 and therefore lends itself to
easy multiplication. In our base-ten system, multiplying a number by 5 can
be quickly done by halving the number and moving the decimal point one Plimpton 322 (so named because it is number 322 in the G. A. Plimpton Col­
place to the right. For example, 2.86 x 5 = (2.86/2) x 10 = 1.43 x 10 = 14.3 lection at Columbia University; see fig. 1 .3), it is a table of four columns,
(more generally, a x 5 = � x 10). Similarly, in the sexagesimal system multi­ which might at first glance appear to be a record of some commercial transac­
plying a number by 30 can be done by halving the number and moving the tion. A close scrutiny, however, has disclosed something entirely different: the
"sexagesimal point" one place to the right (a x 30 = � x 60) . tablet is a list of Pythagorean triples, positive integers (a, b, c) such that
a2 + b2 = c2 • Examples of such triples are (3, 4, 5), (5, 1 2, 1 3), and (8, 1 5, 1 7).
Let us see how this works in the case of Y BC 7289. We recall that
1;24,5 1, 10 is short for 1 + 24/60 + 5 1/602 + 10/603 . Dividing this by 2, we get
Because of the Pythagorean theorem, 3 every such triple represents a right tri­

i+ lii + 2560i + 6i3 , which we must rewrite so that each coefficient of a power
.l angle with sides of integer length.
Unfortunately, the left edge of the tablet is partially missing, but traces of
of 60 is an integer. To do so, we replace the 1/2 in the first and third terms by modern glue found on the edges prove that the missing part broke off after

by 30/60, getting + 60 + 60+ 260 + 605 3 - + 60252 + 60353 _

- 0, 42, 25, 35. Finally,
25
30 the tablet was discovered, raising the hope that one day it may show up on
. 30 12 _ 42 .
the antiquities market. Thanks to meticulous scholarly research, the missing
moving the sexagesimal point one place to the right gives us 42;25,35, the
60 60
part has been partially reconstructed, and we can now read the tablet with rel­
length of the diagonal. It thus seems that our scribe chose 30 simply for prag­ ative ease. Table 1 . 1 reproduces the text in modern notation. There are four
matic reasons: it made his calculations that much easier. columns, of which the rightmost, headed by the words "its name" in the orig­

•••• ••• ••••

inal text, merely gives the sequential number of the lines from I to 1 5 . The
• second and third columns (counting from right to left) are headed "solving
number of the diagonal" and Hsolving number of the width," respectively;
If Y BC 7289 is a remarkable example of the Babylonians' mastery of ele­ that is, they give the length of the diagonal and of the short side of a rectan­
mentary geometry, another clay tablet from the same period goes even further: gle, or equivalently, the length of the hypotenuse and the short leg of a right
it shows that they were familiar with algebraic procedures as well. 2 Known as triangle. We will label these columns with the letters c and b, respectively. As
 Mesopotamia, 1 800 BCE •••• 9 10 •••• Chapter I

I.1
TA B L E
Plimpton 322

(cla) 2 b C

[ 1,59,0,] 15 1,59 2,49 1 Newsome Note: Table 1.1

[ 1,56,56,]58, 14,50,6, 15 56,7 3, 12, 1 2 doesn't make any
[ 1,55, 7 ,]4 1, 15,33,45 I , 16,4 1 1,50,49 3
corrections to the errors
a
on Plimpton 322.
[ 1,]5[3, 1 ]0,29,32,52, 16 3,3 1,49 5,9, 1 4
[ l , ]48,54, 1,40 1,5 1,37 5
[ l ,]47,6,4 1,40 5, 19 8, 1 6
[ l ,]43, 1 1,56,28,26,40 38, 1 1 59, 1 7
[ l ,]4 1,33,59,3,45 1 3, 19 20,49 8
[ l ,]38,33,36,36 9, I 12,49 9 Figure 1.4. The cosecant of an angle: csc A = c/a
Maor- The Pythag. Thm.

1,35, 10,2,28,27 ,24,26,40 1,22,4 1 2, 16, I 10

Again, in line 13 we have b = 7, 12, 1 = 7 x 602 + 12 x 60 + 1 = 25 92 1 and c =
4,49 = 4 x 60 + 49 = 289, and these do not form a Pythagorean triple; but we
1 ,33,45 45 I , 15 11

may notice that 25 92 1 is the square of 16 1 , and the numbers 1 6 1 and 289 do
1,29,2 1,54,2, 15 27,59 48,49 12
[ l ,]27,0,3,45 7, 12, 1 4,49 13 form the triple ( 16 1, 240, 289). It seems the scribe simply forgot to take the
square root of 25 92 1. And in row 1 5 we find c = 53, whereas the correct entry
should be twice that number, that is, I06 = 1,46, producing the triple (56, 90,
l ,25,48,5 1,35,6,40 29,3 1 53,49 14
[ l ,]23, 13,46,40 56 53 15 106). 4 These errors leave one with a sense that human nature has not changed
over the past four thousand years; our anonymous scribe was no more guilty
of negligence than a student begging his or her professor to ignore "just a little
Note: The numbers i n brackets are reconstructed.

stupid mistake" on the exam. 5

an example, the first line shows the entries b = l ,59 and c = 2,49, which rep­ The leftmost column is the most intriguing of all. Its heading again men­
resent the numbers 1 x 60 + 59 = 1 1 9 and 2 x 60 + 49 = 169. A quick calcu- tions the word "diagonal," but the exact meaning of the remaining text is not
lation gives us the other side as a = ✓ 169 2 - 1 19 2 = ✓ 14400 = 1 20; hence entirely clear. However, when one examines its entries a startling fact comes
( 1 1 9, 120, 1 69) is a Pythagorean triple. Again, in the third line we read to light: this column gives the square of the ratio c/a, that is, the value of csc2 A ,
b = 1 , 1 6,4 1 = 1 x 602 + 16 x 60 + 4 1 = 460 1 , and c = 1 ,50, 49 = 1 x 60 2 + 50 x where A is the angle opposite side a and csc is the cosecant function studied in
trigonometry (fig. 1.4). Let us verify this for line 1. We have b = 1,59 = 1 19
60 + 49 = 6649; therefore, a = ✓6649 2 - 460 1 2 = ,J23 040 000 = 4800, giv­
and c = 2,49 = 169, from which we find a = 120. Hence (c/a ) 2 = ( 1 69/ 1 20) 2 =
ing us the triple (460 1 , 4800, 6649). 1.983, rounded to three places. And this indeed is the corresponding entry in
The table contains some obvious errors. In line 9 we find b = 9, 1 = column 4: 1;59,0, 15 = l + 59/60 + 0/60 2 + 15/60 3 = 1.983. (We should note
9 x 60 + 1 = 54 1 and c = 1 2, 49 = 1 2 x 60 + 49 = 769, and these do not form again that the Babylonians did not use a symbol for the "empty slot" and
a Pythagorean triple (the third number a not being an integer). But if we therefore a number could be interpreted in many different ways; the correct in­
replace the 9, 1 by 8, 1 = 48 1 , we do indeed get an integer value for a: terpretation must be deduced from the context. In the example just cited, we
a = ✓769 2 - 48 1 2 = ✓360 000 = 600, resulting in the triple (48 1 , 600, 769). assume that the leading l stands for units rather than sixties.) The reader may
It seems that this error was simply a "typo"; the scribe may have been momen­ check other entries in this column and confirm that they are equal to (c/a ) 2 .
tarily distracted and carved nine marks into the soft clay instead of eight; and Several questions immediately arise: Is the order of entries in the table ran­
once the tablet dried in the sun, his oversight became part of recorded history. dom, or does it follow some hidden pattern? How did the Babylonians find
 Mesopotamia, 1 800 BCE •:• 11 12 ..•.. Chapter I

those particular numbers that form Pythagorean triples? And why were they Otto Neugebauer, The Exact Sciences in Antiquity ( 1957 ; rpt. New York: Dover, 1969),
interested in these numbers-and in particular, in the ratio (c/a)2-in the first chap. 2. See also Eves, pp. 44-47 .
place? The first question is relatively easy to answer: if we compare the values 3 . More precisely, its converse: i f the sides of a triangle satisfy the equation
of (c/a)2 line by line, we discover that they decrease steadily from 1. 983 to a 2 + b2 = c2 , the triangle is a right triangle.
1.387, so it seems likely that the order of entries was determined by this se­ 4. This, however, is not a primitive triple, since its members have the common fac­
quence. Moreover, if we compute the square root of each entry in column 4- tor 2; it can be reduced to the simpler triple (28, 45, 53). The two triples represent sim­
that is, the ratio c/a = csc A-and then find the corresponding angle A , we dis­
ilar triangles.
cover that A increases steadily from just above 45 ° to 58 ° . It therefore seems
5. A fourth error occurs in line 2, where the entry 3, 12, 1 should be 1,20,25, produc­

that the author of this text was not only interested in finding Pythagorean
ing the triple (3367, 3456, 4825). This error remains unexplained.

triples, but also in determining the ratio c/a of the corresponding right trian­
gles. This hypothesis may one day be confirmed if the missing part of the tablet
shows up, as it may well contain the missing columns for a and c/a. If so,
Plimpton 322 will go down as history's first trigonometric table.
As to how the Babylonian mathematicians found these triples-including
such enormously large ones as (460 1, 4800, 6649)-there is only one plausi­
ble explanation: they must have known an algorithm which, 1,500 years later,
would be formalized in Euclid's Elements: Let u and v be any two positive in­
tegers, with u > v; then the three numbers
Maor- The Pythag. Thm.

a = 2uv, b = u 2 - v2 , c = u 2 + v2 (1)

form a Pythagorean triple. (If in addition we require that u and v are of oppo­
site parity-one even and the other odd-and that they do not have any com­
mon factor other than 1, then (a, b, c) is a primitive Pythagorean triple, that is,
a, b, and c have no common factor other than 1.) It is easy to confirm that the
numbers a, b, and c as given by equations ( 1) satisfy the equation a2 + b 2 = c2 :
a2 + b 2 = (2uv ) 2 + ( u 2 - v2 ) 2
= 4u 2 v2 + u4 - 2u 2 v 2 + v4
= u4 + 2u 2 v2 + v4
= ( u 2 + v2 ) 2 = c2 .
The converse of this statement-that every Pythagorean triple can be found in
this way-is a bit harder to prove (see Appendix B).
Plimpton 322 thus shows that the Babylonians were not only familiar with
the Pythagorean theorem, but that they knew the rudiments of number theory
and had the computational skills to put the theory into practice-quite remark­
able for a civilization that lived a thousand years before the Greeks produced
their first great mathematician.

Notes and Sources

I . For a discussion of how the Babylonians approximated the value ofJi.,

see Ap­
pendix A.
2. The text that follows is adapted from Trigonometric Delights and is based on