22 Chapter I

Mesopotamia but nothing inhinsically inherent in the Babylonian

notation. This is different for the second problem, however. The
Babylonian scribe would know (or take this information from a
table of reciprocals) that i corresponds to "12" (0;12 = if in
our notation when we use a zero symbol). Hence y leads again
to finding the value of 12 times 12 or again to 2,24 (we would
write 2;24). In other words the Babylonian process completely
avoids special rules for computing with fractions, whether unit
fractions or not, and requires only that one remember correctly
the place value of each contributing number, exactly as we
must do in placing the final decimal point. The historical con-
sequences of this simplification can scarcely be overestimated.
14 Q. The advantages of the Babylonian place value system over
the Egyptian additive computation with unit fractions are so
obvious that the sexagesimal system was adopted for all astron-
omical computations not only by the Greek astronomers but also
by their followers in India and by the Islamic and European
astronomers. Nevertheless the sexagesimal notation is rarely
applied with the strictness with which it appears in the cuneiform
texts of the Seleucid period in Mesopotamia. Ptolemy, for ex-
ample, uses the sexagesimal place value system exclusively for
fractions but not for integers. Thus he will write 365 as '&' E 8
(300,60, 5) but not as ~ 8(6,5). The same procedure was followed
by the Islamic astronomers and is the reason for our present
astronomical custom to write integers decimally and then use
sexagesimal minutes and seconds.
Extreme consistency in the use of the sexagesimal place value
system is found in the Latin version of the "Alfonsine Tables"
(about 1280). Here we find a date like 1477 Sept. 20 6;1,36h
expressed as 2,29,49,32;15,4,0 days. Indeed 1476 Julian years
(of 3651 days each) contain 24,36 ·6,5;15 days = 2,29,45,9 days.
To this are added the 4,23 days until Sept. 20 and the fraction
O;15,4d = 6;1,36h • This gives the above total of days, counted
from A.D. 1 Jan. O.
Also Copernicus often used consistently written sexagesimal
numbers, pal1icularly in his tables of mean motions. For example,
for the moon he gives the following mean motions in consecutive
Egyptian years (of 365 days each)
 Numbers 23

1 2,9;37,22,36°
2 4,19;14,45,12
3 0,28;52,7,49 etc.
where we (and Ptolemy) would write for the integers 129, 259,
and 28 respectively.
The perfection to which Islamic scholars developed numerical
methods has only recently become clear, especially through the
work of P. Luckey on aI-Kishi, the astronomer royal of Ulugh
Beg in Samarqand. AI-Kishi died in 1429; one of his last works
is a treatise on the circumference of the circle in which he de-
termines (correctly) 2n as 6;16,59,28,1,34,51,46,15,50. And since
he had invented, a few years earlier, the decimal analog of the
sexagesimal fractions, he also converts the above number into
decimal fractions: 6.2831853071795865.

BIBLIOGRAPHY TO CHAPTER I
The best existing book on numbers and number systems is Karl Menninger,
Zahlwort und ZifTer. Aus der Kullurgeschichte unserer Zahlsprache, unserer
Zahlschrift und des Rechenbretts. Breslau, Hirt, 1934. The sections about the
Babylonian number system are not always reliable. In general, however, this
work is far superior to the majority of books on the history of numbers and of
elementary computing.
Kurt Sethe, Von Zahlen und Zahlworten bei den alten Aegyptern und was
fur andere V6lker und Sprachen daraus zu lemen ist. Schriften d. wiss. Ges.
Strassburg 25 (1916). This work is fundamental for the understanding of the
role of fractions within number systems in general.
For calendar computation in general see F. K. Ginzel, Handbuch der
mathematischen und technischen Chronologie, 3 vols., Leipzig, Hinrichs, 1906-
1914. The lunar calendar of Western Europe during the Middle Ages is discussed
in vol. III.
An excellent introduction to Babylonian civilization is given in Edward
Chiera, They Wrote on Clay, Univ. of Chicago Press; several editions since
1988.

NOTES AND REFERENCES TO CHAPTER I

ad 1. The "Book of the Hours" of the Duke of Berry was originally published
by Paul Durrieu, Les Ires riches heures de Jean de France, Duc de Berry,
Paris 1904. The twelve calendar miniatures are reproduced in color in Verve
24 Chapter I

No. 7 (1940), unfortunately excluding the zodiacal figure (Umelothesia") which

followed the calcndar. It was discussed in detail by Harry Bober, The Zodiacal
Miniature of the Tris Riches Heures of the Duke of Berry -Its Sources and
Meaning, Journal of the Warburg and Courtauld Institutes 11 (1948) p. 1-34.
ad 2. D. E. Smith and L. C. Karpinski, The Hindu-Arabic Numerals,
Boston, 1911. Julius Ruska, Zur Iltesten arabisehen Algebra und Reehenkunst,
Sitzungsber. d. Heidelberger Akad. d. Wiss., philos.-hist. Kl. 1917, 2. G. F. Hill,
The Development of Arabic Numerals in Europe exhibited in 64 Tables. Oxford,
Clarendon Press, 1915.
It is a mistake to assume that the Islamic mathematicians and astronomers
consistently used the "Hindu-Arabic" numerals. By and large the Hindu-Arabic
numerals are restricted to mathematical context, whereas astronomical tables
use the alphabetic numerals. In Egypt the Greek or Coptic alphabetic numerals
remained in use for centuries after the Arabic conquest.
A nice detail about the transmission of the numerical notation of the Hindus
to the Islamic world is accidentally preserved in the autobiography of Ibn SInl,
the "Avicenna" of the Middle Ages. He was born in 980 near Bukhara, which
was then undcr the Iranian Dynasty of the Samanids. When he was about ten
years old, missionaries of an Islamic sect, called Ismaelites, came to Bukhara
from Egypt. Through the teaching of these missionaries Ibn Sinllearned about
the Hindu method of ~ompuling. Without this explicit bit of information nobody
would have dreamed that Indian influcnce reached southem Russia via Egyptl
(Cr. Ibn Sini's autobiography, translated in Arthur J. Arberry, Avincenna on
Theology, London 1951.)
ad 3. For the development of the Greek number system and its relation to
Phoenicia cf. Wilhelm Larfeld, Grieehische Epigraphik, 3rd ed., MUnchen,
Beck, 1914 [Handbuch der k1assisehen A1tertumswissenschaft vol. 1]. See esp.
p. 290 ft'. Furthermore M. N. Tod, The alphabetic numeral system in Attica.
The Annual of the British School at Athens, 45 (1950) p. 12&-139.
The acrophonic numerals are often called "Herodianic" because a gramma-
rian Herodianus (second century A. D.) discussed these numbers. The name
seems to have been introduced by Woisin in his thesis, De graecorum notis
numeralibus, Lipsia 1886.
Recent discussion, textual evidence and bibliography in Marcus Niebuhr Tod,
The Greek acrophonic numerals, The Annual of the British School at Athens
No. 37, Sessions 193&-37, p. 23&-258 (London 1940). For examples cf. B. D.
Meritt-H. T. Wade-Gery-M. F. McGregor, The Athenian Tribute Lists (Cam-
bridge, Harvard Univ. Press. 1939) vol. I passim; e. g. the photo p. 74 Fig. 98
and corresponding copy on PI. XXI.
ad 6. W. E. van Wijk, Le nombre d'or. :Etude de chronologie technique
suivie du texte de la Massa Compoti d'A1exandre de Villedieu. La Haye, Nijhoft',
1936. This work contains a valuable introduction to the medieval cyclic calendars
in Europe. Cf. also Nils Lithberg, Computus, Stockholm 1953 (Swedish) =
Nordiska Muscets Handlingar 29; very complete bibliography.
ad 7. Jean-Gabriel Lemoine, Les anciens procedes de calcul sur les doigts
en orient et en occident. Revue des etudes islamiques 6 (1932) p. 1-58 [with
extensive critical bibliography].
Egyptian numbering of fingers: Kurt Sethe, Ein a1tlgyplischer Fingerzlhl-
 Notes and References 25
reim. Zeitsehr. fUr Aegyptische Sprache 54 (1918) p. 16-39. Battiscombe Gunn.
UFinger-Numbering" in the Pyramid Texts. ibid. 57 (1922) p. 71 f.
For the relations between alphabetic numerals, number mysticism, astrology,
etc., see Franz Dornseiff, Das Alphabet in Mystik und Magie, Stoicheia 7,
2nd ed., Leipzig, Teubner, 1925.
ad 8. Examples for Athenian coins of the New Style: Bulletin de Corre-
spondance Hellenique 58 (1934) PI. I. These coins show the Athenian owl
standing on the Panathenaic amphora. The numerals for the months are often
inseribed on the amphora and are therefore called "amphora letters". For
Greek coins in general see. e. g.• Barkley V. Head. Historia Numorum, A
Manual of Greek Numismaties, Oxford, Clarendon 1911.
For the Ptolemaic coins with double letter numerals cf. Reginald Stuart
Poole, Catalogue of Greek Coins. The Ptolemies, Kings of Egypt, p. 44 and
Pl. VIII, 5. These numbers seem to represent the years of an era in honor of
Queen Arsinoti U (270 B.C.): cf. Head, Hist. Num. p. 850.
ad 9. The Greek symbols for 6, 90, and 900 are usually called stigma,
qoppa, and sampi respectively. The fll'St is originally F (or similar) and therefore
also called cCdigamma", that is double-gamma. Later, it assumed forms which
were similar to the ligature of c and T in Byzantine manuscripts and it was
therefore called "stigma" (since the 7th or 8th cent. A.D.). Its original name
is Waw. Qoppa is the Q of the Phoenician alphabet. The sampi is originally
written with only one middle stroke (cr. the form in PI. 5). The name usampi"
has been in use since the 17th century A.D.; the Phoenician original is an
S-sound called ~ade.
The division of the circumference of the circle into 360 parts originated in
Babylonian astronomy of the last centuries B.C. The sexagesimal number
system as such is many centuries older and has nothing to do with astronomical
concepts.
ad 10. For an example of an inscription with large number symbols in the
alphabetic notation see Inscriptiones Graecae vol. 12,1 (Insularum Maris
Aegaei) Berlin 1895, No. 913. This inscription, found in Keskinto (Rhodes) and
dating from the second century B.C., lists the basic numbers of a theory of
planetary motion; the author is unknown. For a discussion cr. P. Tannery,
Memoires scientifiques vol. 2 p. 487 fl.
In the ordinary alphabetic notation the numbers 1000, 2000 etc. are written
by means of (x, P. etc. which precede the symbols of lower order. Often accents
are added in order to avoid confusion with I, 2. etc. Several cases, both from
inscriptions and papyri, are known, where the symbol for 900, the usampi",
with (x, p, ... as superscript was used for 1000,2000. etc. (cf. Larfeld. quoted
above p. 24 in the note to Section 3, p. 294).
In papyri of the early Ptolemaic period ODe finds, in addition to sampi,
also fI'U = 500 ( +) 400 for 900, and besides Q) also fJT = 500 ( +) 300 for 800.
Cf. Mahaffy. Flinders Petrie Papyri vol. 3 p. 98 etc. An example from a school-
book of the third century B.C. is shown on PI. 5 from P. Cairo. Inv.65445
(published by O. Gueraud et P. J ouguet, Publications de la Societe Royale
EuPtienne de Papyrologie, Textes et Documents. Vol. 2, Cairo 1938). The
column on the left and the middle column constitute a table of squares of which
the fonowing part is clearly readable:
26 Chapter I

6 6 36 100 100 1 ·10000

7 7 49 200 200 4 ·10000
8 8 64 300 300 9 ·10000
9 9 81 400 400 16 ·10000
10 10 100 500 500 25 ·10000
20 20 400 600 600 36 ·10000
30 30 900 700 700 49 ·10000
40 40 1600 800 800 64 ·10000

Note in PI. 5 the signs for 6 and 900. The sign for 1000 (in 1600) is an (¥
with an attached loop. The multiples of 10000 are written as a II (first letter of
the Greek word for 10000) with the factor written over it. The last column gives
a list of the fractions of the drachma; preserved are the symbols for the following
unit fractions: 8, 12, 24,3, 6,8, 48. Note that the unit fractions are written with
the ordinary number signs plus an a£cent. The only exception is p' which does
not mean} but f denoted here by 3. Its corresP.9nding drachma sYmbol is a
combination of the symbols for 2 and 6. Indeed, 3" = 2 6. +
Ordinarily the arrangement of the alphabetic numerals is stricOy from higher
to lower numbers. In datings, however, one finds also the inverted order: ct. for
examples from Mesopotamia Yale Classical Studies 3 p. 30 fT. (clay bullae
from Uruk); Excavations in Dura-Europus, Preliminary Report IX, 1 p. 169 fT.;
Klio 9 p. 353. For Macedonian inscriptions (between 131 B.C. and 322 A.D.)
cf., e. g., Tod. The Maeedonian Era; The Annual of the British School at
Athens, No. 23 (1918-1919) p. 206-217 and No. 24 (1919-1921) p. 54-67.
That the Arabic form for the zero sYmbol (a litOe circle with a bar over it
and related forms) is simply taken from Greek astronomical manuscripts was
recognized by F. Woepcke in 1863 (Journal Asiatique, Sere 6 voL 1 p. 466 ft.).
A table showing dift'erent forms in Arabic manuscripts as well as in Greek
papyri is given by Rida A. K. Irani, Arabic numeral forms, Centaurus 4 (1955)
p. 1-12. In a Byzantine manuscript, written about 1300 A.D. a sign like II is
used for zero beside 0 (Vat. Graec. 1058 fol. 261 ft.), apparently under Islamic
influence.
ad 13. The most comprehensive collection of the evidence on early number
signs is found in the fIrSt edition of Anton Deimel, Sumerische Grammatik der
arehaistisehen Texte, Roma, Pontificium Institutum Biblicum, 1924 (Chapter IV).
More recent evidence, especially concerning the decimal system, is given in
A. Falkenstein, Archaisehe Texte aus Uruk, Leipzig 1936, (sign list at the end).
For the picture of a stylus see, e. g., S. Langdon, Excavations at Kish, vol. 1,
Paris 1924, PI. XXIX and Falkenstein, 1. c., p. 6.
The texts from Uruk also revealed the existence of a system of fractions
strictly proceeding on the principle of repeated halving. A very important
feature of cuneiform numerical notation is the existence of special signs for },
-1-, I, and t which are in very common use also in later periods, even occasionally
in mathematical texts. These "natural fractions" undoubtedly play an important
role in the arrangement of metrological units. Obviously one will group higher
units in such a form that they admit directly the forming of these most common
parts. This leads naturally to a grouping in 12 or 30 or 60. All these ratios do
occur in one or another of the parallel systems ofunits in Mesopotamian metrology.
 Notes and References 27

A strictly decimal notation occurs occasionally in mathematical texts. The

following is an example from an Old-Babylonian text (published Neugebauer-
Sachs, Math. Cuneiform Texts, p. 18). The number 1,12 which occurs in the
text is transcribed in the heading as "4 thousand 3 hundred and 20" which is
indeed the equivalent of 1,12,0. This example shows at the same time the lack
of an absolute determination of the place value in this period of number writing.
We may interpret 1,12 as 1,12,0 = 4320 or as 1,12 = 72 or as 1;12 = I} etc.
Only the context permits the determination of the absolute value of a number
written sexagesimally.
The lack of a notation which determines the absolute value of a number
made it possible to misinterpret simple tables of multiplication or reciprocals.
When Hilprecht, in 1906, published a volume of "mathematical, metrological
and chronological tablets from the Temple Library of Nippur" he was convinced
that these texts showed a relation to Plato's number mysticism. In book VIII of
the "Republic" Plato gives some cabballstic rules as to how guardians of his
dictatorially ruled community should arrange for proper marriages. By some

found on the tablets. Thus 1,10 (i. e. 70 or 1.

wild artifices, Plato's cabbala was brought into relationship with the numbers
etc.) was interpreted to mean
195,955,200,000,000 and in this fashion whole tablets were transcribed and
"explained".
As to the origin of the sexagesimal place value notation, it may be noted
that it is quite common that fractions of monetary units came to mean fractions
in general. As an example can be quoted the Roman as, which is It, of the
uncia (ounce). In the measurement of time, however, as is if of one hour
(Jahreshefte d. oesterreichischen archaeol. lost. in 'Vien 37, 1948, p. 111).
Mixed writings are also quite common. An example from an astronomical
procedure text (ACT No. 811 a, obv. 27) is 1 me 1,30 me for "190 days". Here
1 me means "1 hundred" (me being an abbreviated writing of the Babylonian
word for 100), while 1,30 is the sexagesimal writing for 90, and the final me
means "day", probably an abbreviation.
ad 14. No definite answer can be given to the question when the zero sign
was introduced in Babylonian mathematics. \Ve feel sure that it did not exist.
say, before 1500: and we find it in full use from 300 B.C. on. A table of squares,
found at Kish. tentatively dated by the excavator, S. Langdon, to the period
of Darius (500 B.C.), contains four cases of a "zero" written exactly like 30.
It is omitted in one case. Cf. Neugebauer, MKT I p. 73 and II pI. 34. For the
possibility of an earlier date (about 700) see MeT p. 34 note 95.
One might expect that the Babylonian notation should often lead to errors,
e. g., by mistaking 10,2 for 12 and vice versa. Numbers of this type are, however.
ordinarily written with very careful spacing such that <rr is hardly ever to
be taken as -«Tr. The descriptions of the Babylonian number systems in the
current textbooks arc generally quite misleading on this very point. Nevertheless,
there do exist cases where the proper combination of tens and units becomes
very doubtful. We even have examples of large numbers, written in two or more
lines, where, e. g., the 50 of a 56 was written at the end of one line and the 6
at the beginning of the next. For such "split writings" cf. Neugebauer-Sachs,
Math. Cun. Texts p. 13 note 69 and Neugebauer, ACT vol. 2, index.
28 Chapter I

In practical computations, the Babylonian scribes oceasionally committed the

same type of mistakes which arise when we are careless with the decimal point.
As an example may be quoted an astronomical text which concerns the risings
and settings ofMereury dUl'iDg the years 146 to 122 B.C. (BM 34585, Neugebauer
ACT No. 302 obv. IV. 30). The scribe had trouble with interpolations. The
table at his disposal contained the entries
15 42
45 36
The problem consists in interpolating the value of the right-hand column for
the value 31;20 of the left-hand argument. Obviously the answer should be
36 + 13;40 • 0;12 - 36 + 2;44 - 38j44. The utronomical problem required
the addition of this result to a~other number 1;20. Henee the ftnaI result should
be 40;4. In the text. however. we find 37;22,44. Obviously the scribe incorreedy
determined the place value of 13;40 • 0;12 and wrote Oj2.44 instead of 2j44.
This gave him u the result of the interpolation 36;2,44 and therefore as the
final answer Ij20 + 36;2.44 - 37;22.44.
It must be said. however, that the number of errors in the texts is compara-
tively very small. I have had the experience that I committed many more errors
in eheekiDg the ancient computations than there were in the original documents.
Often errors in a text are very helpful because they constitute one of the main
tools for establishing the details of a numerical procedure followed by the
ancient computer.
ad 14. Paul Luckey, Der Lehrbrief iiber den KreisumfaDg (ar-risila aI-
muhltlya) von c}amlld b. MU·Od ai-Kill. Abh. d. Deutsehen Akad. d. Wiss.
zu Berlin. KI. f. Math. u. aJJcem. Naturwiss. Jahrg. 1950 Nr. 6. Akad.
... Verl.•
Berlin 1953. Furthermore: Paul Luckey, Die Reehenkunst bei Gamlld. b.
Mu·Od ai-Kill mit RfickbUcken auf die Jiltere Geschiehte des Rechnens. Abh.
f. d. Kunde d. Morgenlandes 31,1 (1951).
 29

CHAPTER II

Babylonian Mathematics.
15. The following chapter does not attempt to give a history of
Babylonian mathematics or even a complete summary of its
contents. All that it is possible to do here is to mention certain
features which might be considered characteristic of our present
knowledge.
I have remarked previously that the texts on which our study
is based belong to two sharply limited and widely separated
periods. The great majority of mathematical texts are "Old-
Babylonian"; that is to say, they are contemporary with the
Hammurapi dynasty, thus roughly belonging to the period from
1800 to 1600 B.C. The second, and much smaller, group is
"Seleucid", i. e. datable to the last three centuries B.C. These
dates are arrived at on quite reliable palaeographic and linguistic
grounds. The more than one thousand intervening years influenced
the forms of signs and the language to such a degree that one is
safe in assigning a text to either one of the two periods.
So far as the contents are concerned, little change can be
observed from one group to the other. The only essential progress
which was made consists in the use of the "zero" sign in the
Seleucid texts (cr. p. 20). It is further noticeable that numerical
tables, expecially tables of reciprocals, were computed to a much
larger ex~ent than known from the earlier period, though no new
principle is involved which would not have been fully available
to the Old-Babylonian scribes. It seems plausible that the
expansion of numerical procedures is related to the development
of a mathematical astronomy in this latest phase of Mesopotamian
science.
For the Old-Babylonian texts no prehistory can be given.
We know absolutely nothing about an earlier, presumably
30 Chapter II

Sumerian, development. All that will be described in the sub-

sequent sections is fully developed in the earliest texts known.
It is customary to postulate a long development which is sup-
posedly necessary to reach a high level of mathematical insight.
I do not know on what experience this judgment is based.
All historically well known periods of great mathematical dis-
coveries have reached their climax after one or two centuries of
rapid progress following upon, and followed by, many centuries
of relative stagnation. It seems to me equally possible that Babyl-
onian mathematics was brought to its high level in similarly
rapid growth, based, of course, on the preceding development
of the sexagesimal place value system whose rudimentary forms
are already attested in countless economic texts from the earliest
phases of written documents.
16. The mathematical texts can be classified into two major
groups: "table texts" and "problem texts". A typical representative
of the first class is the multiplication table discussed above p. 16.
The second class comprises a great variety of texts which arc all
more or less directly concerned with the formulation or solution
of algebraic or geometrical problems. At present the number
of problem texts known to us amounts to about one hundred
tablets, as compared ",ith more than twice as many table texts.
The total amount of Babylonian tablets which have reached
museums might be estimated to be at least 500,000 tablets and
this is certainly only a small fraction of the texts which are still
buried in the ruins of Mesopotamian cities. Our task can there-
fore proPerly be compared with restoring the history of mathe-
matics from a few torn pages which have accidentally survived
the destruction of a great library.
t 7. The table texts allow us to reconstruct a small, however
insignificant, bit of historical information. The archives from
the city of Nippur, now dispersed over at least three museums,
Philadelphia, Jena, and Istanbul, have given us a large percentage
of table texts, many of which are clearly "school texts", i. e.,
exercises written by apprentice scribes. This is evident, e. g.,
from the repetition in a different hand of the same multiplication
table on obverse and reverse of the same tablet. Often we also
find vocabularies written on one side of a tablet which shows
mathematical tables on the other side. These vocabularies are
 Babylonian Mathematics 31

the backbone of the scribal instruction, necessary for the mastery

of the intricacies of cuneiform writing in Akkadian as well as in
Sumerian. Finally, many of our mathematical tables are combined
with tables of weights and measures which were needed in daily
economic life. There can be little doubt that the tables for multi-
plication and division were developed simultaneously with the
economic texts. Thus we find explicitly confirmed what could
have been concluded indirectly from our general knowledge of
early Mesopotamian civilization.
18. Though a single multiplication table is rather trivial in
content, the study of a larger number of these texts soon revealed
unexpected facts. Obviously a complete system of sexagesimal
multiplication tables would consist of 58 tables, each containing
all products from 1 to 59 with each of the numbers from 2 to
59. Thanks to the place value notation such a system of tables
would suffice to carry out all possible multiplications exactly
as it suffices to know our multiplication table for all decimal
products. At first this expectation seemed nicely confirmed except
for the unimportant modification that each single tablet gave
all products from 1 to 20 and then only the products for 30,
40, and 50. This is obviously nothing more than a space saving
device because all 59 products can be obtained from such a
tablet by at most one addition of two of its numbers. But a more
disturbing fact soon became evident. On the one hand the list
of preserved tables showed not only grave gaps but, more discon-
certingly, there turned up tables which seemed to extend the
expected scheme to an unreasonable size. Multiplication tables
for 1,20 1,30 1,40 3,20 3,45 etc. seemed to compel us to assume
the existence not of 59 single tables but of 3600 tables. The
absurdity of this hypothesis became evident when tables for the
multiples of 44,26,40 repeatedly appeared; obviously nobody
would operate a library of 60S = 216000 tablets as an aid for
multiplication. And it was against all laws of probability that we
should have several copies of multiplication tables for 44,26,40
but none for 11, 13, 14, 17, 19 etc.
The solution of this puzzle came precisely from the number
44,26,40 which also appears in another type of tables, namely,
tables of reciprocals. Ignoring variations in small details, these
tables of reciprocals are lists of numbers as follows