Egyptian Mathematics: The Sources
Egyptian Mathematics: The Sources
Egyptian Mathematics: The Sources
Contribution to
The Cambridge History of Science
The primary references of the term Egyptian mathematics are the computational
techniques and the underlying mathematical knowledge attested in Pharaonic written
sources. Secondary references are, on one hand, the corresponding techniques etc. as
known from Demotic sources; on the other, the geometrical procedures used in
Pharaonic and subsequent architecture and visual arts. Greek mathematics produced
in Hellenistic Egypt is thus not included. Accordingly, all dates in the following are
BC unless AD is indicated explicitly.
The sources
The most important written sources for Pharaonic mathematics are the Rhind
Mathematical Papyrus (henceforth RMP)1 and the Moscow Mathematical Papyrus (MMP).2
1
of Chaces volume II and includes a translation kept close to that of the same volume
as well as an extensive analysis.
2
Marshall Clagett gives the following approximate dates (Ancient Egyptian Science. A
Source Book. Volume I: Knowledge and Order, pp. 629635. Memoirs of the American
Philosophical Society, 184 A+B; Philadelphia: American Philosophical Society, 1989):
Early dynastic period (dynasties 12): 31102665.
Old Kingdom (dynasties 38): 26642155.
First intermediate period (dynasties 910): 21542052.
Middle Kingdom (dynasties 1113): 20401640.
Second intermediate period (Hyksos dynasties 1516, Theban dynasty 17): 16401532.
New Kingdom (dynasties 1820): 15501070.
Third intermediate period (dynasties 21(initial) 25): 1070712.
Late period (dynasties (final) 2531, including the Assyrian hegemony during dynasty
26 and the Persian dynasties 27 and 31): 712332.
Greco-Roman period: 332 BC to 395 AD.
4
Full edition and translation in Alan H. Gardiner, Egyptian Hieratic Texts. Series I:
Literary Texts from the New Kingdom. Part I: The Papyrus Anastasi I and the Papyrus Koller,
together with Parallel Texts (Leipzig: J. C. Hinrichssche Buchhandlung, 1911).
-2-
with numbers to the early dynastic period); even in this respect, however, the Middle
through New Kingdom is much richer.
The mathematical sources of the Pharaonic period are written in hieratic script
(evidently, some hieroglyphic documents contain numbers); part of the metrological
terminology seems to have been created in hieratic and to have acquired hieroglyphic
equivalents only at a later moment (when at all).
From the Demotic phase (late and Greco-Roman periods), several papyri containing
mathematical problems and tables survive5 with the possible exception of one
uncertain fragment all of Greco-Roman date.
Edition with translation and extensive commentary in Richard A. Parker (ed.), Demotic
Mathematical Papyri (Providence & London: Brown University Press, 1972).
6
The standard reference for Egyptian numerals and number words remains Kurt Sethe,
Von Zahlen und Zahlworten bei den Alten gyptern, und was fr andere Vlker und Sprachen
daraus zu lernen ist. (Schriften der Wissenschaftlichen Gesellschaft in Straburg, 25;
Straburg: Karl J. Trbner, 1916). In the second millennium, the sign for 1000000, and
afterwards that for 100000 went out of use; instead, multiplicative notations were used.
-3-
henu or 320 ro (the part again, but in a different use), according to another by
successive halvings (down to 1/64 ).7 Also multiples of the hekat were expressed by special
signs or by non-standard use of the standard numerals. A special unit for bulky
substances is the khar, equal to 20 hekat and to 2/3 of a royal cubit (probably a secondary
normalization of originally distinct units).
75
150
750
1500
3000
3975
Some texts reveal the underlying thought: If 1 (of the entity we count) is 75, then 2
(of it) is 150, etc.. The multiplier 53, as we see, is split into components that can be
obtained by successive doublings and decuplings (mostly, only doublings would be
employed). Strokes mark addends that are actually used (53 = 1+2+10+40).
The corresponding division of 3975 by 75 would go by the same procedure,
emptying 3975 by multiples of 75:
/1
/10
20
/40
/2
total
75
750
1500
3000
150
3975
A separate phrase would state the result as 53 (=1+10+40+2); strokes will of course
have been inserted a posteriori in the scheme.
This only seems simple until fractions are introduced. An actual multiplication
(of 8 3" 6 18 = 8 8/9 by itself) would run as follows (RMP 42):
The hieroglyphic writings for the successive halves of the hekat can be put together
to the standard representation of the healing sacred eye of Horus; as pointed out by
Peet (op. cit., p. 26), however, the hieroglyphic writings do not antedate the 18th dynasty,
whereas the hieratic forms go back to the third millennium; the mythological
connotations of the system are thus a late construction, notwithstanding their popularity
in standard histories.
-4-
3" 6 18
3" 9
2 18
9
3" 6 18 27
3" 6 12 36 54
3 12 24 72 108
3 9 27 108 324
79 108 324
1
2
4
/8
/3"
3
/6
/18
8
17
35
71
5
2
1
Total
It is no accident that 3" of 8 3" 6 18 is found before 3. Even when only 3 of the
multiplicand is needed, 3" is found first and 3 then by halving. 3" and 2 were the basic
fractions of the Middle Kingdom calculators; if at all possible, further divisions would
be produced from these by successive halvings (the presence of 18 shows that it was
not always possible).
Beyond this, the calculation displays the main difficulties to which multiplication
of fractions gives rise. The first doubling is obvious, since 3" doubled is 1 3; in the
next, however, 9 has to be doubled, and the scribe has to know that this yields 6 18
(after which 3 6 is contracted to 2). Finally, 9 3" 6 18 27 3 12 24 72 108 3 9 27 108
3"
72
6
18
18
6
27
4
3
36
12
9
24
4 2
72
1 2
108
1
3
36
9
12
27
4
108
1
324
3
Since the sum of the red (i.e., italicized) numbers is 217 3 = 2 108+1+3, the sum of
This was first pointed out by Kurt Vogel in Die Grundlagen der gyptischen Arithmetik,
p. 43 (Mnchen 1929, reprint Wiesbaden: Martin Sndig, 1970).
9
of the Rule of Three. The Egyptian conceptualization may be illustrated by RMP 24,
one of the problems treating of an abstract quantity or heap ( h ) the problem
type by which the technique was trained: A quantity, 7 of it added to it, becomes it:
19.
The computation looks as follows
/1
/7
7
1
1
/2
2
/4
/8
8
16
4
2
1
/1
/2
/4
2
4
9
16 2 8
2 4 8
19
4 8
2 4
2
This may be explained as a single false position: As a preliminary value for the heap
we take 7, then the quantity together with its seventh part become 8. This is seen to
be contained 2 4 8 times in 19 (an ordinary division); therefore the true value of the
quantity is 2 4 8 times 7 or, which is more convenient for the final proof, 7 times
The Egyptians, indeed, made ample use of the commutativity of
2 4 8 = 16 2 8.
multiplication, despite the obvious asymmetry of their algorithm; the frequent claim
that their mathematical thought was purely additive is thus blatantly mistaken.
The principles of this computation were applied with flexibility: at times the
preliminary value might be set to 1 (e.g., RMP 32); in combination with the commutativity of operations this might lead to something very close to the Babylonian division
through multiplication by the reciprocal (e.g., RMP 63). The formulations, however,
show that the Egyptian method is based on the usual principles and no borrowing from
abroad.
The 2n table of RMP is the largest extant piece of systematic Egyptian mathematics and may be considered its theoretical high point. Many efforts have hence been
dedicated to finding the principle(s) which underlie its construction the same fraction
9 45,
10 30,
may indeed be split in many ways into aliquot parts ( 2/15 thus into 8 120,
11 30 110,
13 20 156,
14 30 35,
etc.). So much is certain that a standard existed
12 20,
in the later Middle Kingdom the deviations from the RMP-norm are rare enough
to count as aberrations. Kurt Vogel points to three principles (at times in mutual conflict)
that seem to intervene:10
(i)
The members of the sum should be few.
(ii)
The first member should be as large as possible.
10
(iii)
If more than two members are present, the largest denominator should be
kept small.
(ii) might seem to suggest a search for a good first approximation but (iii) shows that
a good second approximation was not aimed at. The principles seem rather to have
been of an aesthetic kind.
The technique that is used consists in dividing 2 into two parts p+r, where p is
an aliquot part of n ( p/n = 1/m ) and the remainder r is an aliquot part of 1 or the sum
of such parts, r = 1/s + 1/t +..., whence r/n = 1/(sn) + 1/(tn) +.... This much is shown explicitly
in the text, which lists p and the constituents of r and tells which part ( 1/m , 1/(sn) , etc.)
each one is of n. The essential trick, however, is of course to find an adequate splitting
of 2. Here several ways were followed, probably reflecting the steps of the historical
process that had engendered the table. If n is a multiple of 3 (n = 3m), the division is
into 1 2 and 2, whence 2n = 1/2m + 1/6m . In many other cases, an adequate p was
probably found by subdivision of 3" of n or 2 of n, as illustrated by the way the text
explains 213:
1
2
4
/8
52
108
/4
/8
13
6
3
1
8 1 2 8
52 4
104 8
2
4
2 8
4
8
73
Find
\3
\4
\5
11
\60
219
292
365
219 3
292 4
365 5
1 6 20
3
4
5
This terminology is used in RMP 61B, cf. Peet, op. cit. (n. 1), p. 104.
-7-
Applied arithmetic
Beyond the abstract h -problems, both RMP and MMP contain many arithmetical
problems of practical or sham-practical character. Most important are distribution
problems and the so-called pesu-problems.
Many of the distribution problems deal with equal partition e.g., the distribution
of n loaves among 10 persons, n = 1, 2, 6, 7, 8, 9 (RMP, 16); they illustrate why Plutarch
and other Greek authors would link social equality to arithmetical justice (and hence
reject the latter as morally unsound). Others follow the principle that the foreman and
other officials get double share (RMP 65), or that the ratio between shares is given (RMP
63). Such problems are true to real life as revealed in administrative texts. RMP 40,
on the contrary, is wholly artificial: loaves are distributed in five shares (say, a, b, c,
d, and e) in arithmetical proportion in such a way that
1
(i)
/7 of the sum of the three major shares equals the sum of the two minor
ones;
(ii) a+b+c+d+e = 100.
The solution makes use of a simple false position: at first an arithmetical progression
, , , , is constructed, starting with = 1 and fulfilling (i); its sum is found to be
60, whence all members are multiplied by 1 3". The first step is not explained, but since
RMP 64 refers explicitly to and makes adequate use of the average share and the excess
of one share over the other when determining the single members of an arithmetical
progression from the sum and the difference, a simple algebraic solution (whether
represented by words or by pebbles or other material tokens) will not have exceeded
the conceptual capabilities of the Egyptian calculator though apparently his standard
discourse: If is 1 and the difference, = 1+, = 1+2, etc; the sum of the three major
shares is thus 3+(2+3+4), which is 7 times + = 2+; thus 3+9 = 14+7, 2 = 11, =
5 2.
Endowed with particular status obviously because of the importance of bread
and beer as staple food are the pesu problems. pfsw is derived from psj, cooking,
and may be understood as baking ratio. The pesu of a loaf is the number of similar
loaves that may be made from one hekat of grain; similarly, the pesu of beer counts
the number of jugs that are produced from one hekat of grain. In both cases, the baking
-8-
ratio thus indicates the reciprocal grain content of the unit of consumption. Pesu
problems may ask for the pesu, given the number of units produced and the total
amount of grain; for the exchange of loaves with different pesu or of bread with beer;
more complex problems deal with the dilution of beer, or with special brews made
from several grain sorts or fruits.12
Together, these and other problems of applied arithmetic cover more or less the
standard types of late medieval and early modern commercial arithmetic proportional
partition, exchange, alloying (only composite interest has no Egyptian counterpart);
often the methods are familiar, although no technique similar to the double false
position is ever applied; at times, however, unexpected steps demonstrate that ad hoc
reasoning was no less important than automatic routines.
Among higher arithmetical problems, one recreational problem in RMP deals with
the geometrical progression 7, 72, 73, 74, 75, and finds the sum as 7 2801; nothing in
the text tells whether the underlying reasoning is simply that 7+...+75 = 7 (1+...+74) =
7 2801, or a formula for the sum of a geometrical progression was known.13
Geometrical computation
Geometrical problems deal with slopes, areas and volumes. The batter (skd) of
pyramids is expressed as the retrocession in palms per cubit height, whereas that of
a different (unidentified) structure is given as a pure-number ratio in RMP 60.
Already the metrology (cf. above) shows that rectangular areas were found as
length times breadth. The area of a triangle was determined as half the base multiplied
by the edge, whose identity has been discussed; however, since RMP 51 takes the
half of the base for the giving of the rectangle of it, there can be little doubt that the
edge between the two parts into which an isosceles triangle is cut is meant that is,
the height.14 The area of the trapezium was found correspondingly.
Area computation serves in a few cases as the basis for homogeneous seconddegree problems. Thus in MMP 6, 7 and 17, the area of a right triangle and the ratio
between the sides is given; doubling of the area and multiplication by the ratio yield
the area of a corresponding square, whose square root (corner) is then one side of
the triangle; similar considerations are used to solve problems about two squares, whose
sides have a given ratio.
The volume of right parallelepipeds was found by multiplication of the three
dimensions measured in cubits, followed by a multiplication by 1 2 in order to express
12
See Peet, op. cit. (n. 1), pp. 112121 and Struve, op. cit. (n. 2), pp. 44101.
13
A third, somehow intermediate possibility is suggested by Robins & Shute, op. cit.
(n. 1), pp. 56f.
14
This was already argued by Peet, op. cit. (n. 1), pp. 9193.
-9-
it in khar. MMP 14 finds the volume of a truncated square pyramid (with height h and
sides a and b of base and top, respectively) correctly as h/3 (a2+ab+b2). No cues are given
as to how the formula was derived. It cannot be excluded that it is the result of a lucky
generalization of the formula for the area of a triangle; nor is a heuristic argument based
on dissection into simpler volumes to be excluded, however.15
The area of the circle was found as that of the square on
8
/9 of the diameter 1.006... times the true value. A diagram
in RMP 48 suggests that this may be a computational approximation to the area of a geometrically approximating octagon,
whose area is 63/81 of the square in question (see diagram).
Volumes of circular cylinders were determined accordingly.
MMP 10 calculates the surface of a basket with mouth
4 2 as 4 1/2 ( 8/9 [ 8/9 9]), with the argument that the basket
is the half of an egg (Struves reading of a damaged word).
The double factor 8/9 leaves no doubt that explicit use is made of the formula for the
circular area no empirical measurement would be able to distinguish ( 8/9 [ 8/9 9]) =
7 9 from 7 and the conjectured egg seems to suggest that a hemisphere with
diameter 4 2 is intended, whose surface (in modern terms) is then found correctly as
2r2. This formula seems much more sophisticated that anything else found in the
sources, for which reason the alternative interpretation of the basket as the curved
surface of a semicylinder (with height = diameter = 4 2) has been suggested. This does
not fit an egg too well, but has the advantage to presuppose only that the Egyptians
knew the relation between circular area and circumference which agrees well with
their explicit transformation of a triangle into a corresponding rectangle.16
Geometrical techniques
Rules for geometrical computation evidently depend on techniques for mensuration. These will have been the responsibility of those rope stretchers (harpedonaptai)
which the Greeks refer to.17 Rope constructions were also used when the ground plans
of prestige buildings were laid out. Architectural designs as well as pictorial art were
constructed within square grids, following a strict canon (the canonical system,
coupled to metrology and already used in Early Dynastic iconography) for how many
15
See, e.g., Robins & Shute, op. cit. (n. 1), p. 49.
16
The two interpretations (due, respectively, to Struve and Peet) are confronted in
grammatical detail in O. Neugebauer, Vorgriechische Mathematik, pp. 129137 (Berlin:
Julius Springer, 1934).
17
See Peet, op. cit. (n. 1), p. 32, and Vogel, Vorgriechische Mathematik. vol. I (n. 1), p. 59f.
- 10 -
Fables
Two remarks should be added concerning what we have no reason to ascribe to
Pharaonic Egyptian geometry.
Firstly, ever since Moritz Cantor proposed that the rope stretchers might have used
the 3-4-5-triangle to construct right triangles it has been a recurrent claim that they
actually did so.19 It has also been presumed that since several pyramids have the batter
3:4, the Egyptians will have known the properties of this triangle.20 It must be
emphasized that the sources do not contain the slightest hint pointing in this direction,
and that the batter in question would be expressed as 5 4 palms [per cubit height],
which is not liable to have furthered any Pythagorean speculations.
Similarly, the attempts to find (or the Golden Section) in the great pyramids
founder on the observation that the approximate occurrence of such ratios in the
construction are automatic consequences of the simple value for the batter; we should
also remember that the Egyptians did not make use of , that is, of the ratio between
4
circular circumference and diameter, but of
(the ratio between the side of the
squared circle and the diameter), which they approximated as 8/9 .21
18
See Erik Iversen, Canon and Proportion in Egyptian Art (2Warminster, England: Aris
& Phillips, 1975; 11955), and (on the influence of the system in later art) idem, The
Canonical Tradition, pp. 5582 in J. R. Harris (ed.), The Legacy of Egypt, second edition
(Oxford: Oxford University Press, 1971).
19
Thus, e.g., Alexander Badawy, Ancient Egyptian Architectural Design. A Study of the
Harmonic System, pp. 3f and passim (Berkeley & Los Angeles: University of California
Press, 1965).
20
See Gay Robins & Charles C. D. Shute, Mathematical Bases of Ancient Egyptian
Architecture and Graphic Art, Historia Mathematica 12 (1985), 107122.
21
For further references regarding the fables and their lack of foundation, see Gillings,
op. cit. (n. 1), pp. 23739. A recent very careful treatment is Roger Herz-Fischler, The
Shape of the Great Pyramid (Waterloo, Ontario: Wilfrid Laurier University Press, 2000).
- 11 -
is recorded during Dynasty 1.22 Though nothing comparable to the bureaucratic precision
of early Mesopotamian state formation was aimed at the legitimization of the
Pharaonic state rested on conquest and perhaps on the affirmation of cosmological
stability, not on redistribution practical mathematics proper was thus certainly present
throughout the third millennium.
This, however, is not yet the mathematics of RMP and MMP. Until Dynasty 5
only metrological sub-units and the fractions 3", 2 and 3 (and a particular sign for
1
/4
Links?
These similarities are evidently to be explained as parallel developments due to similar
conditions, not as borrowings. On the general level, second millennium Egyptian and
Babylonian mathematics are wholly independent from each other. On the level of
particulars, the occasional multiplication with a reciprocal in RMP has sometimes been
seen as a borrowing, but the context where it occurs speaks against that assumption
(cf. above). Only a single problem in RMP (viz no 37) is certainly related to a Babylonian
text:
Go down I [a jug of unknown capacity] times 3 into the hekat-measure, 3 of me is added
to me, 3 of 3 of me is added to me, 9 of me is added to me; return I, filled am I.
The Egyptian solution is quite regular, fully based on aliquot parts and grain metrology;
the Babylonian solution is no solution at all but a trick which presupposes that the
solution be already known. The problem is obviously one of those riddles which the
early Akkadian scribe school borrowed around 1800 BC (see MESOPOTAMIAN
MATHEMATICS). On the other hand, the idiom of ascending continued fractions
26
IM 53 957, ed. p. 37 in Taha Baqir, Some More Mathematical Texts from Tell Harmal,
Sumer 7 (1951), 2845, corrections and interpretation p. 52 in W. von Soden, Zu den
mathematischen Aufgabentexten vom Tell Harmal. Sumer 8 (1952), 4956.
- 13 -
(a, and b of a, where a and b are simple fractions) is typically Semitic, and alien to
the Egyptian context.27 Both a Babylonian borrowing from Egypt and an Egyptian
adoption of a Babylonian school problem are thus excluded; both must build on a
common source, probably a traders environment in contact with both regions. In
contrast to what happened in Babylonia, however, such borrowings are not likely to
have had any deeper influence on Egyptian Middle Kingdom mathematics, which
instead develops material and ideas already present in third-millennium scribal
computation.
27
See in general Hyrup, op. cit. (n. 23); since I had not noticed the Enunna parallel
at the time, this publication contains some speculations about a possible common
Hamito-Semitic language structure; they may now be happily dismissed.
28
P. British Museum 10794, ed. Parker, op. cit. (n. 5), pp. 72f.
- 14 -
a truncated cone as the height times the mid-cross-section;30 and the use of the
surveyors formula (average length times average width) to calculate the areas of
approximately rectangular quadrangles. They are certainly the reason that 8 of 40
problems in P. Cairo J. E. 8912730, 8913743 belong to a characteristic Babylonian type
(a reed first standing vertically against a wall and then moved to a slanted position),
which always involves the Pythagorean theorem, often in a sophisticated way (asking,
e.g., for the legs of a right triangle when the hypotenuse and the difference between
the hypotenuse and one leg are given).
30
The details are of some interest: the surface of the mid-cross-section is not found as
in the corresponding Old Babylonian text, but as 1/4 of the product of diameter and
arc, the arc being 3 diameters see Clagett, Ancient Egyptian Science. Volume II:
Calendars, Clocks and Astronomy, p. 75 (Memoirs of the American Philosophical Society,
214; Philadelphia: American Philosophical Society, 1989). The latter formula belongs
to the lay tradition, is found in one Old Babylonian school tablet (dealing with a
semicircle), and recurs in the pseudo-Heronic material.
- 15 -