This article was downloaded by:

On: 30 April 2011

Access details: Access Details: Free Access
Publisher Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK

International Journal of Mathematical Education in Science and

Technology
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713736815

Themes in the evolution of number systems

S. Avitala; I. Kleinerb
a
Technion, Israel Institute of Technology, Haifa, Israel b Department of Mathematics and Statistics,
York University, North York, Ontario, M3J 1P3, Canada

To cite this Article Avital, S. and Kleiner, I.(1992) 'Themes in the evolution of number systems', International Journal of
Mathematical Education in Science and Technology, 23: 3, 445 — 461
To link to this Article: DOI: 10.1080/0020739920230314
URL: http://dx.doi.org/10.1080/0020739920230314

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
 INT. J. MATH. EDUC. SCI. TECHNOL., 1992, VOL. 2 3 , NO. 3, 445-461

Themes in the evolution of number systems

by S. AVITAL
Technion, Israel Institute of Technology, Haifa, Israel
and I. KLEINER
Department of Mathematics and Statistics, York University,
North York, Ontario, M3J 1P3, Canada

(Received 31 May 1991)

This paper offers suggestions for introducing various mathematical concepts
and topics related to, and often originating in, the study of number systems
(broadly construed). The material is organized around eight themes which may
serve as source material for courses of varied degrees of sophistication. The
themes deal with algebraic, analytic, geometric, number-theoretic, set-theoretic,
cultural, and philosophical issues. In many cases we sketch the historical origin of
the mathematical ideas involved. We believe the paper will be especially useful in
courses for teachers.
Downloaded At: 22:54 30 April 2011

Number systems have been a fruitful source of ideas, concepts, results, and
theories in the evolution of mathematics. In fact, it has been suggested that much
even of modern mathematics has its roots in the study of number and shape (see e.g.
[1,2]). Using 'number' as an organizing principle, one can introduce students to
algebraic, analytic, set-theoretic, geometric, and number-theoretic ideas in a
relatively concrete setting. As illustrations we present in this article eight 'themes'
related to the study of various number systems. The themes vary in detail, difficulty,
and sophistication, and can serve as source material for courses at various levels and
be addressed to varied audiences (e.g. teachers, mathematics majors, 'liberal-arts'
enthusiasts). Although the themes are interconnected, they can be read independ-
ently. No attempt is made to be thorough, but references to an extensive
bibliography are provided throughout. Readers are invited to come up with their
own themes to suit their interests, needs, and objectives.
Our notion of 'number' is broad—from the natural through the complex
numbers and beyond, to transfinite, p-adic, and hyperreal numbers. The approach
within each theme is, when appropriate, historical (or rather genetic). Consideration
of the origin of mathematical ideas, of the burning questions which the originator of a
concept or result tried to answer, lends a useful perspective to the teaching and
learning of mathematics. For example, cardinal numbers can (of course) be studied
as a mathematical topic without reference to history, but when viewed in an
historical setting as the resolution of centuries of gropings for the meaning of the
infinite in mathematics they acquire special significance. T h e historical approach
also enables us to raise various philosophical issues which lend themselves to
classroom discussion (with any student audience and at almost any level). Possible
issues are:
(a) T h e roles of problems and paradoxes in the genesis of mathematical
concepts, results, and theories.
(b) Internal versus external motivation in the evolution of mathematics.
(c) T h e nature of mathematical existence and proof.

0020-739X/92 $3.00 © 1992 Taylor & Francis Ltd.

 446 S. Avital and I. Kleiner

(d) The roles of intuition versus logic in the creation of mathematics.

(e) The relation of mathematics to the physical world.
Now to the themes.

1. Beyond the complex numbers

This is a good theme with which to begin since it leads rather quickly, and more
or less naturally, from the familiar (to students) to the less familiar.
A brief review of the 'standard' number systems, from N (the natural numbers),
through Z (the integers), Q (the rationals), and R (the reals), to C (the complex
numbers), sets the scene for what is to come. We focus in this summary on their
historical evolution and especially on gains and losses at each transition stage in the
evolutionary process. For example, in going from 1R to C one gains algebraic closure
but loses order, and in going from Z to Q one gains division but loses divisibility. (Q,
but not Z, is closed under division). On the other hand, the notion of divisibility,
while fundamental in Z, is trivial in Q.) The transition from Q to R is accompanied
by a loss of 'innocence': while Z, Q (and even C) can be built up from their
predecessors by finitary operations (e.g. Z can be viewed as consisting of equivalence
classes of pairs of elements of N), R requires infinite processes for its construction
(e.g. infinite decimals, Cauchy sequences, Dedekind cuts). A point to highlight in
Downloaded At: 22:54 30 April 2011

this discussion is solvability of polynomial equations: in each number system we can

solve more equations than in its predecessor (e.g. 2x = 6 can be solved in Z but 2x = 7
only in Q).
In expounding these ideas one can thus introduce in a 'natural' manner a number
of fundamental mathematical notions, such as: mathematical induction, closure
under an operation, denseness, commensurability, completeness, order, and
algebraic closure. See [3-8] for details.
The introduction of the quaternions (H) follows. U = {a + bi+cj
+ dk:a,b,c,deU}, where i, j , k are arbitrary 'units' which satisfy the relations
i2 =y 2 = k2 = ijk = — 1. (From these the other properties of these units, such as
ij=k= —ji, can be deduced.) All the usual laws of numbers, except for commutativ-
ity under multiplication, hold in H (technically: H is a skew field or division ring).
Moreover, just as in the case of C, a polynomial equation over D-0 has a root in H.
However, in contrast to the situation in C, an equation of degree n over HI may have
more than n roots in HI. E.g. x2 + \=0 has infinitely many roots: bi+(\—b2)ll2j,
where 0«S&^ 1. See [9,10].
Hamilton's invention (in 1843) of the quaternions is well documented. This is a
rare instance of the evolving process of mathematical discovery, and students can
readily be led through it [11,12]. Although the quaternions did not live up to
Hamilton's expectations as a creation rivalling in its applicability the infinitesimal
calculus, they proved important in helping 'liberate' algebra from arithmetic (i.e.
from its dependence on the laws of arithmetic), and in helping liberate geometry
from its restriction to three dimensions. As Poincare and Hamilton, respectively, put
it:
Hamilton's quaternions give us an example of an operation which presents an
almost perfect analogy with multiplication, which may be called multipli-
cation, and yet it is not commutative.... This presents a revolution in
arithmetic which is entirely similar to the one which Lobachevsky effected in
geometry ([13], p. 29).
 Themes in the evolution of number systems 447

There dawned on me the notion that we must admit, in some sense, a fourth
dimension of space for the purpose of calculating with triples ([6], p. 191).
See [3,6,11,14,15,16,105] for details on quaternions.
Is there a 'natural' extension of the quaternions? Cayley and Graves (in 1844)
gave an affirmative answer by introducing the octonions (Cayley numbers) K: 8-
tuples of reals which form a non-commutative and non-associative, but alternative
((aa)b = a(ab), a(bb) = (ab)b) division ring. The problem (as for quaternions) was to
define multiplication in K. The difficulty 'disappears' if we first reconsider
multiplication in H. Write a + bi + cj+dk as (a + bi) + (c + di)j=w + zj, where w,zsC,
and j2 = — 1. The quaternions are thus viewed as pairs of complex numbers. Now
define (wt + z1j)(w2 + z2j) = (w1w2 — z2zl) + (z1W2 + z2w1)j, where 2 denotes the
conjugate of z. (It is important to have the wt and zt appear in precisely this order.) It
is easy to verify that this product in HI is the same as the usual product defined in
terms of i,j, &. If we now let K = {a + fie : a, /?eHJ}, where e is an arbitrary unit with e2
= — 1, and define the product in K analogously, as (<x1 + P\e){a2 + p2e)= ( a i a 2 ~~ P2P1)
+ (Pl5i2 + P2ct1)e ( t n e conjugate a of the quaternion <x = a + bi+cj+dk is a — bi—cj
— dk), then IK becomes an alternative division ring. It is easy to verify that (say)
(ij)e¥=iUe)- See [3,6,15] for details.
It is tempting to continue in this fashion by defining a 'number-system' of 16
Downloaded At: 22:54 30 April 2011

units consisting of pairs of octonions. That this is doomed to failure was shown
(independently) by Frobenius and by C. S. Peirce (ca 1880):
Theorem. The only n-tuples of real numbers which form an alternative division
ring (which need not be commutative or associative) are the reals, the complex
numbers, the quaternions, and the Cayley numbers.
See [15] for a not very demanding proof.
Incidentally, it is easy to show directly (once we know what we want to show) that
triples of real numbers do not form a division ring extending C For if they did, then ij
would have to be of the form a + bi+cj (a,b,ceU). Then i(if) = i(a + bi + cf).
Multiplying and collecting terms (using i2 = — 1), we get c2 +1 =0—a contradiction
(see [17]). (There is an elementary proof v/hich shows that, for oddn, a division ring of
n-tuples of reals is possible only for w = l; see [6], p. 190.) There is, of course, an
important product defined on triples, namely the vector product (a1i + a2j
+ a3k) x(bii + b2j+b3k) = (a2b3 — a3b2)i + {a3bx — a1b3)j+(a1b2 — a2bY)k. Moreover,
the vector product ( X ), the quaternion product (O), and the scalar (inner) product
(•) of 3-dimensional vectors are related: ax P = ocQP + Of /? (see [15], p. 29). (The
only other Euclidean n-space in which a 'cross product' can be defined is n = l\ see
[18].)
Having defined various number systems with differing properties, one ought to
raise the question 'what is a number?', which should quickly lead to the (in this
setting, more appropriate) question 'how do numbers "behave"?' (The idea here is
that, in general, we study not objects but relations among objects. The other
important point to make is that the notion of number has evolved over time.) In the
context of this first theme, the answer leads to the definition of ring, field, and
division ring. This, in turn, leads to:
(a) The consideration of other number-like objects such as integers modulo n,
matrices, (boolean) rings of sets, polynomials, rational functions, power
series, and extended power series. Such examples were instrumental in the
emergence of the abstract concepts of ring and field. See [16, 19, 20].
 448 S. Avital and I. Kleiner

(b) Axiomatic characterizations of Z, Q, R, and C; that is, doing for these

structures what Euclid and Hilbert have done for the Euclidean plane. (For
example one can characterize Z as the smallest ordered integral domain and U
as a complete ordered field.) This was but one instance of the re-emergence
of axiomatics at the turn of this century following a 2000-year dormancy.
This topic provides a good opportunity to discuss issues of completeness,
independence, and consistency of an axiomatic system. As we keep introduc-
ing various properties (axioms) of (say) the integers, we ask: do we have
enough such properties to characterize Z? (completeness); are there now too
many? (independence); perhaps we have picked 'incorrect' axioms? (consis-
tency). See [3, 6, 7, 21] for details.

2. The algebraic-transcendental dichotomy

While the first theme explored number systems beyond C, this one explores such
systems (initially fields) between Z and C. We easily show that there is no field
between Z and Q (Q is the smallest field contained in C), nor between 1R and C (if a
field contains IR and any (non-real) complex number, it must be all of C). The latter
observation entails the notion of adjunction of an element a to a field Fto obtain a field
F(a) containing F— a fundamental idea in field theory. (It was used by Galois in his
Downloaded At: 22:54 30 April 2011

development of Galois theory; see [20].) When applied to F=Q and a = ^j2 (say), it
yields the field Q(y/2) of polynomials in y/2 with coefficients in Q, Q(yj2) =
{a + byj2 : a, beQ}. What if a = 7t? In this case polynomials in n will not do—we need
rational functions:

These two examples highlight the difference between algebraic and transcend-
ental numbers (e.g. ^ 2 and n, respectively). In fact, we have found this algebraic
framework to be a good way to motivate the introduction of these two concepts. That
the algebraic/transcendental dichotomy is not unbridgeable, is indicated by the
remarkable relation exp(7ii) + l =0. See [4,21-23].
The set of all algebraic numbers forms a field A. (For a nice proof based on an
elementary result in linear algebra see [23], p. 83.) Since it consists of all roots of
polynomial equations with integer coefficients, A is a natural generalization of Q
(which can be viewed as the field of roots of linear equations ax + b = 0, with a, beZ).
Probably the first to define transcendental numbers (i.e. those (complex) numbers
which are not roots of polynomial equations over Z) was Euler (in the 18th century),
although he never proved their existence. (The distinction between algebraic and
transcendental/ttMC*«OM.s was made by Leibniz in the 17th century.) Since A is a field,
it follows that if t is transcendental and a algebraic (a^O), then ta is transcendental.
Hence there exist infinitely many transcendental numbers if there exists one. The
first to prove the existence of transcendental numbers was Liouville (in 1844), who
showed that 1/1O1! + 1/1O2! + 1/1O3! + ... is transcendental. (For a nice proof using
some basic ideas of calculus see [24], p. 735.) The transcendence of n, hence the
impossibility of squaring the circle, was proved by Lindemann in 1882. (For a proof
see [25]. For a history of n, which involves geometric, analytic, and computational
ideas, see [26] and [27].)
 Themes in the evolution of number systems 449

An important mathematical event occurred at the second International Congress

of Mathematicians in Paris in 1900. Hilbert proposed 23 problems which proved to
be instrumental in generating much research during this century (see [28], Ch. 27
and [29]). The seventh problem was to determine which (complex) numbers of the
form a? are transcendental (see [29], p. 242). Gelfond and Schneider proved
(independently) in 1934 that a? is transcendental if a and jS are algebraic, a ^ 0,1, and
P is not rational (it may be complex). It follows, of course, that 2^2 is transcendental,
but also that so are (for example) exp (ft) and log10 2. First, one of the values of the
infinite-valued expression i~21 is exp (ft) (use exp(fti)= —1). Second, Iog102 is not
rational (let log102 = a/6, then 10a = 2b); if Iogi02 were algebraic, then 10loglo2 = 2
would be transcendental. See [23,30].
A related topic for investigation is to determine which (real) algebraic numbers
are rational; namely, to determine the rational roots among the zeros of f(x) = a0
+ a^x +... + aYlx\aleZ'). If sjt is such a root with s and t relatively prime, it follows
readily that 5 divides a0 and t divides an (see [30], p. 58). As a corollary we get that ^k
is irrational unless k = un, ueZ (since x" — k has no rational roots), and that (for
example) cos 20° is irrational (since it satisfies the equation 8x3 — 6x — 1 = 0 which has
no rational roots). See [30].
For algebraic numbers which are irrational, it has been found important to
examine how closely they can be approximated by rational numbers (with given
Downloaded At: 22:54 30 April 2011

conditions). It turns out that such algebraic numbers cannot be approximated 'too
well' by rational numbers. Specifically, if a is an irrational algebraic number which is
a root of an irreducible polynomial over Z of degree n > 1, then there exists a positive
real number c such that for all p, qeZ with q>0, \ct—plq\>clqn. It was this result
which enabled Liouville to prove the transcendentality of
00

y io~ m!
(see above). See [4, 24, 30] for details.
The ideas used by Liouville and by Gelfond and Schneider have recently been
generalized (by Roth, Baker, et al.). Although these new results have not resolved the
question of the transcendentality (or even the irrationality) of such numbers as fte,
ft + e, 7re, ee, ft", 2", 2e, y (the Euler constant), they have played a crucial role in the
study of a wide variety of diophantine problems. See [31], volumes 1 and 2 and [23].
Algebraic numbers arise naturally, interestingly, and importantly in the solution
of diophantine equations. In fact, this is where much of their importance lies. For
example, to find integer solutions of x2+y2 = z2, x2 + 2=y3, x3+y3 = z3, we
factor the left side of each equation to obtain, respectively, x2 + y2 = (x+yi)(x —yi),
x2 + 2 = (x + y/2i)(x — yj2i), and x3+y3 = (x+y)(x+yw)(x+yw2), where
w = ( — 1 +yj3i)l2. For x, y, zeZ, the right sides are algebraic numbers—in fact,
algebraic integers. These are roots of monic polynomials with integer coefficients; they
generalize the ordinary integers (which can be viewed as roots of monic linear
polynomials a + x = 0 over Z).
The consideration of diophantine equations in domains of algebraic integers
often facilitates their solution. Take the case of x2+y2 = z2. Since (x+yi)(x—yi)
= z2, the product of the two 'relatively prime algebraic integers' x+yi and x— yi is a
square, hence (as for ordinary integers) each is a square (of an algebraic integer). In
particular, x +yi = (a + bi)2 = (a2 — b2) + 2abi (a, beZ), hence x = a2 — b2,y = 2ab, from
which we get z = a2 + b2. This is the formula that yields all Pythagorean triples.
 450 S. Avital and I. Kleiner

The above argument takes place in the domain D of algebraic integers called
'Gaussian integers', D = {a + bi: a, beZ). The general idea is that in order to answer
questions dealing with the integers one must sometimes enlarge the domain of
integers (in this case to the domain of Gaussian integers) so as to provide for more
'elbow room'. This is a familiar theme in mathematics, going back to Euler in the
18th century. In this example, the essential divisibility properties of Z carry over to
D, making it a unique factorization domain and enabling us to justify the above
heuristic reasoning. See [32] or [33] for details. Similar considerations apply to the
equations x2 + 2=y3 and x3+y3=z3. (See [34] or [35]).
The consideration of unique factorization in various domains of algebraic
integers and its failure in some such domains led to the creation of algebraic number
theory—the study of number-theoretic problems using algebraic tools. Algebraic
number theory was an important source for the introduction into mathematics of the
concepts of ideal, ring, field, and module. See [31 volume 2, 32, 33,34,35] for details.

3. Transfinite numbers
The infinite! No other question has ever moved so profoundly the spirit of man;
no other idea has so fruitfully stimulated his intellect; yet no other concept
Downloaded At: 22:54 30 April 2011

stands in greater need of clarification than that of the infinite (Hilbert [36],
p. vii).
The clarification sought by Hilbert was initiated by Cantor beginning in the
1870s. This theme deals with some of Cantor's work and some of its consequences. A
brief historical survey of manifestations of the infinite in mathematics prior to
Cantor's work sets the scene for his ideas. In Greek antiquity, Zeno's paradoxes
dealing with the infinite divisibility of space and time confounded contemporary
thinkers. Aristotle viewed the natural numbers as a potential but not an actual
infinite. Medieval scholastic speculations on the nature of the infinite included a
discussion, without resolution, of the 'paradox' that two concentric circles have
unequal perimeters but an equal number of points. Galileo pondered the inherent
contradiction in comparing (for 'size') the positive integers and their squares: on the
one hand the former contain the latter, but on the other one can match up the two
collections in a one-one manner. He concluded that the difficulties arise because
We attempt, with our finite minds, to discuss the infinite, assigning to it those
properties which we give to the finite and limited; but this ... is wrong, for we
cannot speak of infinite quantities as being the one greater or less than or equal
to another ([37], p. 5).
Even the great Gauss protested 'against the use of an infinite quantity as an actual
entity'. He claimed that 'this is never allowed in mathematics', adding that 'the
infinite is only a manner of speaking' ([14], p. 160). See [36-38] for details.
The revolution in our understanding of the infinite was brought about by Cantor,
almost single-handedly, in the space of a few years. There were philosophical and
theological underpinnings to Cantor's creation [39]. The mathematical origins of his
ideas on set theory had to do with the representation of functions in Fourier series
and the consideration of those sets of points for which unique representation fails. For
a thorough analysis of these matters Cantor realized that he needed a rigorous theory
of real numbers. He proceeded to found it using Cauchy sequences. See [39, 40].
 Themes in the evolution of number systems 451

Among the standard but thought-provoking topics for discussion associated with
and resulting from Cantor's ideas are: cardinal and ordinal arithmetic, the existence
of a non-denumerable infinity of transcendental numbers, the continuum hypo-
thesis, the paradoxes of set theory and the resulting axiomatizations of 'naive' set
theory, the axiom of choice and some of its consequences, various philosophies of
mathematics, and Godel's theorems. Among the unusual implications for the
student are:
(a) The giving up of the fundamental tenet that 'the whole is greater than any of
its parts'. This was one of the 'common notions' in Euclid's axiomatization of
geometry. The reluctance to part with it stood in the way of progress in the
study of the infinite, as shown by the examples of Galileo and of others cited
above. See [14, 36, 37, 38, 41].
(b) The existence of 'arithmetics' in which the additive and multiplicative
cancellation laws, the commutative laws of addition and multiplication, and
the (right) distributive law fail. The first two fail in cardinal arithmetic (e.g. 1
+ Kn = 2+ Xnbutl ^ 2 , 3 - Ko = 4- No but 3 9^4) and the last three in ordinal
arithmetic (e.g. 1+co^co + l, 1-co^arl and, (1 + l)-co^l -co). See
[37, 42].
(c) The existence of an infinity of infinities (cardinals, ordinals) of different
Downloaded At: 22:54 30 April 2011

'sizes'. We have in mind here the infinitely increasing sequence of infinities

which is obtained (with A=Z say) from the result |^4|<|P(^4)|, where \A\
denotes the cardinality of the set A, P(A) the set of all subsets of A. If A is the
set of all sets then P(A)^A, hence \P(A)\ < \A\, and the above result gives rise
to the so-called Cantor paradox. If A = Z, then P(A) = M, and since the
algebraic numbers are denumerable, this implies the existence of nonde-
numerably many transcendental numbers. See [41, 42].
(d) The fact that one can have two equally consistent mathematical theories
contradicting one another. This is a relatively recent (19th century)
phenomenon in mathematics. Cohen's proof in 1963 of the independence of
the continuum hypothesis from the Zermelo-Fraenkel axioms of set theory
gave rise to Cantorian and non-Cantorian set theories, in which the
continuum hypothesis and its negation, respectively, hold (see [43]). The
other major example is, of course, Euclidean and non-Euclidean geometries.
(e) The idea that 'simple' assumptions can have very surprising consequences.
The simple assumption we have in mind here is the axiom of choice. Among
its surprising consequences are the Hausdorff and Banach—Tarski para-
doxes. The latter says that any three-dimensional object can be cut into a
finite number of (ca. 1050) pieces and reassembled to produce two objects,
each identical to the original one (see [44, 45]). In a recent (1988) result along
these lines, it was shown that using the axiom of choice a circle can be
'squared'; that is, it can be dissected into a finite number of pieces (not the
kind that can be cut out of paper with scissors!) and reassembled to yield a
square equal in area to that of the given circle. See [45—47].
(/) The pluralistic nature of mathematics, namely the fact that mathematicians
can differ fundamentally in their views of and approaches to mathematics.
This phenomenon is more prevalent than one would suppose, given the
seemingly 'deterministic' nature of mathematics. For example, Leibniz was
strongly opposed to Descartes' use of algebra in dealing with geometric
 452 S. Avital and I. Kleiner

matters. Hermite (in the 19th century) 'turn[ed] away with fright and horror
from this lamentable evil of functions without derivatives' ([48], p. 973)
which were introduced by Riemann and Weierstrass earlier in the century.
In the beginning of our own century there has been a fundamental dispute
between the formalists and the intuitionists. Witness the radically different
views of Hilbert and Poincare, respectively, on Cantor's set theory:
No one shall expel us from the paradise which Cantor created for us
([48], p . 1003).
Later generations will regard Mengenlehre [Set Theory] as a disease
from which one has recovered ([48], p. 1003).
See [49-51] for details.
(g) The notion that mathematics and theology may have a closer affinity than
meets the eye. Godel's Incompleteness Theorem showed that the consis-
tency of many of the common axiom systems (e.g. for number theory or set
theory) cannot be formally established. This implies an act of faith on the
part of mathematicians in their pursuit of the consequences of axiomatic
systems. See 'Lecture Thirty-Eight' of Eves [14] entitled 'Mathematics as a
branch of theology.'
Downloaded At: 22:54 30 April 2011

4. The personality of numbers*

Theorem. All [natural] numbers are interesting.
Proof. If not, let n0 be the least uninteresting number; but that makes n0 very
interesting. Q.E.D.
Further testimony (if any were needed) to the above result may be found in
[52-57]. (For example, the number 5 is the fifth Fibonacci number; it divides at least
one of the numbers of every pythagorean triple; it is the smallest integer n for which
the general polynomial of degree n is unsolvable by radicals; it is the second Fermat
prime; it is the smallest positive integer d for which the ring of integers of the
algebraic number field Q(y/ — d) is not a unique factorization domain; it is the only
number of the form 4x* +y4 which is prime; it is the only prime p for which p — 2,p
and p, p + 2 are twin primes; and it is the number of regular polyhedra.)
Of course, some numbers are more interesting than others. Moreover, it is
collections of numbers rather than individual numbers which often command
special attention. Among these the primes must be singled out as the building blocks
of all numbers. Their distribution among the integers is a deep and fascinating story.
See [58-60], and theme 8.
A much more elementary classification of the integers than into primes and
composites is into even and odd integers. Yet even that simple idea made possible the
profound discovery by the ancient Greeks of (what we refer to as) the irrationality of
yj2. See [28, 48].
In early Greek mathematics, prior to the 'crisis' resulting from the proof of the
incommensurability of the diagonal and side of a square, geometry and al-
gebra/arithmetic cohabited amicably (cf. theme 7). An example of this cooperative
relationship was the introduction by the Pythagoreans of the polygonal numbers. For

The phrase 'the personality of numbers' was coined by P. Davis [52].

 Themes in the evolution of number systems 453

example, the triangular numbers are 1, 3, 6, 10, ... (The nth triangular number is
1 +2 + 3 + .. . + n = n(n+1)/2.) Their geometric representation is:

The square numbers 1, 4, 9, ... (in general, 1+3 + 5 + .. . + (2n —1) =n 2 )

have geometric representation as squares, and, in general, the &-gonal numbers
n + (n2 — n)(k — 2)/2 (« = 1, 2, 3, ...) as (regular) k sided polygons. (The inclusion of
1 among the polygonal numbers is recent.) There are many interesting relations
involving polygonal numbers, some obtained from their geometric configurations
(see [61-63]). The major result, stated in the 17th century by Fermat, is that every
integer is a sum of ^ 3 triangular numbers, of <4 square numbers, and, in general,
of < k &-gonal numbers. The case k — 4 (every integer is a sum of 4 squares) was
proved by Lagrange in the 18th century, and the general case by Cauchy in the 19th.
Downloaded At: 22:54 30 April 2011

See [64].
Other 'personable' collections of numbers one might investigate are:
(a) Perfect and amicable numbers. These, too, date back to the Pythagoreans and
their numerological notions. See [48, 62, 65] for details.
(b) Fibonacci numbers and their connection with the golden ratio. See [66, 67].
(c) Large primes, factorization of large numbers into primes, and the relation to
public-key cryptography. This recent topic belies the notion of the
'uselessness' of number theory. See [68, 69].

5. One, two, many

We have mainly in mind here the cultural history of numbers. (There is
anthropological evidence that counting in prehistoric civilizations was, at one time, of
the 'one-two-many' variety; see [70].) Among topics to study are number-words in
various societies, number-mysticism, symbolization, the emergence of the abstract
notion of number, and notation—both additive and positional (in different bases).
(Boyer [71] claims that it is an exaggeration to regard positional notation as a
fundamental accomplishment of civilization.) See [5, 40, 70, 72-75] for
details.
A related topic for investigation is how various cultures computed (i.e. performed
the four algebraic operations and root extraction): fingers, pebbles, abacus,
algorithms, logarithms, calculating machines, and computers. See [74-79,106] for
details.
A culminating idea for this theme might be to consider mathematicians' views of
the natural numbers in the 19th century: from Kronecker's 'God made the integers,
all the rest is the work of man', through Dedekind's, Frege's, and Peano's 'The
integers are man-made', to Godel's 'Man-made axioms for the integers are
incomplete*. See [37, 40, 48, 80, 81].
 454 5 . Avital and I. Kleiner

6. Discovery (invention), use, understanding, justification

The sequence of the title is often the order of evolution of mathematical ideas. It
is the reverse of the usual pedagogical order. (Is there a lesson here for pedagogy?) It
is especially evident in the evolution of the various number systems: natural,
negative, irrational (real), and complex. Each is an absorbing tale. While the previous
theme suggests describing the story for the natural numbers, this theme attempts to
sketch it for the other number systems. We will give here only a very brief outline of
the evolution of the negative numbers. That of the real and complex numbers is
readily available in the mathematical literature (e.g. in [5, 38, 40, 48, 82]). These
'stories' provide very good vehicles for raising such issues as the role of paradoxes,
internal versus external motivation, and intuition versus logic as factors in the
development of mathematics. See [49, 82, 83].
Negative numbers entered mathematics as subtrahends (as in a —b), as distinct
mathematical entities, as coefficients in polynomial equations, and as roots of such
equations. At various times, mathematicians accepted or rejected negative numbers
in one or another of these contexts.
The ancient Babylonian, Egyptian, and Greek civilizations disallowed negative
roots of equations and, in general, avoided explicit use of negative numbers although
they were, at least implicitly, aware of the rules governing their manipulation (e.g. the
Greeks had established geometrically the law (a — b){c — d) = ab — bc—ad+bd for
Downloaded At: 22:54 30 April 2011

a > b and c> d; see [28,48]). The Chinese (ca. 200BC) used negative numbers freely in
their calculations. The Hindus (ca. 620 AD) gave the first explicit rules for operations
with negative numbers (e.g. that negative x negative = positive) and allowed for
negative roots of equations (e.g. they stated that every positive number has two
square roots).
European mathematicians of the 16th and 17th centuries were ambivalent about
negative numbers and about their use as coefficients or as roots of equations. The
Greek tradition of using geometry to solve algebraic problems (their 'geometric
algebra'—see [28, 48]) no doubt had an impact. Thus Cardan in 1545 considers
x3 — ax + b and x3 + ax = b (a, b > 0) as distinct equations (replacing these two cubics
with the single cubic x3 + ax + b = 0 would have meant introducing negative
coefficients). Descartes in 1637 determines the number of 'true' (positive) and 'false'
(negative) roots of an equation (his well known 'law of signs'—see [48]), and avoids
the use of negative coordinates in his development of analytic geometry. Pascal
regards the subtraction of 4 from 0 as yielding zero(!). Wallis in 1655 'proves' that
negative numbers are greater than infinity: since a/0 = oo, a/a negative no. > oo, for if
one decreases the denominator of a fraction one increases its value. Arnauld (17th
century) objects to —1/1 = 1/ — 1: how can the ratio of a smaller to a greater quantity
equal the ratio of a greater to a smaller? Leibniz agrees there is a difficulty here but
argues that one should tolerate negative numbers because they are useful and lead to
consistent results. See [28, 48].
There were exceptions to the above misgivings. Girard in 1623 admitted negative
(and complex) roots of equations, and was thus able to state clearly the relation
between the coefficients and roots of an equation, as well as the result that every
equation of degree n has n roots (no proof was given). Stifel in the 16th century used
negative numbers as exponents, and Hudde in the 17th allowed literal coefficients in
an equation to represent any real number—positive or negative. This was a most
important idea, since it permitted a unified treatment of polynomial equations of a
given degree (recall Cardan having to consider x3 + ax = b and x3 = ax + b as distinct
equations). See [40, 48].
 Themes in the evolution of number systems 455

'Models' of abstract concepts are (and were) an important aid in their

understanding and accommodation as bona fide mathematical entities. In the case of
negative numbers, the Chinese viewed them as black rods (using red rods for positive
numbers), the Hindus thought of them as debts, Fibonacci (early 13th century) as
losses, and Girard (1620s) anticipated their representation on a number-line by
noting that 'the negative in geometry indicates a retrogression, where the positive is
an advance' ([28], p. 343). These models aided in justifying addition, but not
multiplication, involving negative numbers.
Textbooks of the 18th century (e.g. Euler's Algebra) continued to give detailed
rules for manipulation of negative numbers, but some resistance to the very notion of
negative numbers persisted to the early 19th century (see [48], p. 593 and [84]).
During this period, however, mathematicians also began to ask why such rules
should hold. Euler himself tried to justify the rule (— a)( — b) = ab by noting that since
( — a)b equals — ab, ( —a)( —6) must equal ab (!) (see [84], pp. 31-32). In the first half
of the 19th century Peacock (and others) made a bold (but not very successful)
attempt to give such justifications by creating symbolical algebra. This was a precursor
of modern axiomatics in algebra—symbols taking on a life of their own, independent
of meaning. For example, ( — a)( — b) was 'decreed' to equal ab, by virtue of its form
rather than its content (see [16], [28], chapter 26, and [84] for details). Finally, in the
Downloaded At: 22:54 30 April 2011

latter part of the 19th century Weierstrass (and independently Peano) give a formal,
abstract definition of negative numbers (integers) as ordered pairs of natural
numbers (see [48], p. 987 ff.). (Hamilton, about half a century earlier, had defined the
complex numbers as ordered pairs of reals; see [11].) Proofs of ( — a)( — b) = ab and
similar laws from the axioms of a field or a ring (as done in today's abstract-algebra
courses) were given in the early 20th century during the emergence of the axiomatic
movement in algebra.
See [48, 78, 84, 85] for further details on the evolution of the negative numbers.
To summarize the above account, a formal, logical justification of negative
numbers came only in the late 19th century, although such numbers were (in one
form or another) used for over two millennia. Mathematical needs played at least as
important a role in their evolution as practical utility, and formal manipulations,
more often than genuine understanding, guided mathematicians in formulating the
rules of operation with negative numbers.

7. Numbers and geometry

Arithmetic and geometry seem, at first sight, to be antithetical. The one deals
with the discrete, the other with the continuous. The relations between the two are,
however, deep though often hidden. The tensions between number and geometry,
and between the related analytic and synthetic approaches to mathematics, have
been very beneficial for the development of the subject. See [38, 50, 86].
Although the relation between arithmetic and geometry is fundamental, it has
not always been amicable. The early Greek harmony between number and shape,
given expression in (among other things) the arithmetical development of the
Pythagorean theory of similarity, was shattered by the Greek crisis of in-
commensurability (i.e. proof of the existence of incommensurable magnitudes; see
[48]). Geometry reigned supreme for roughly the next two millennia (with the
notable exceptions of Chinese, Hindu, and Islamic mathematics). The two joined
again in the 17th century through the emergence of analytic geometry and the
calculus. With the arithmetization of analysis in the latter part of the 19th century (cf.
 456 S. Avital and I. Kleiner

theme 8), arithmetic gained the upper hand, at least in analysis. But the harmony
persisted in other areas (e.g. algebraic geometry). We now give several examples,
from different periods, of the collaborative relationship between number and
geometry.

(a) The notions that geometric relations can be expressed by numbers and that,
conversely, relations among numbers have implications in geometry, have
their (implicit) roots in ancient Babylonian and Egyptian mathematics (ca.
1 5 0 0 B C ) in the computation of areas and volumes of various geometric
figures and in the (apparent) construction of right-angled triangles from
relations such as 3 2 + 4 2 = 5 2 (the Egyptian 'rope-stretchers'; see [38], p. 1).
The formal expression of the latter idea was the (geometric) statement and
proof of Pythagoras' theorem and the (arithmetic) determination of all
Pythagorean triples, both coming to fruition in ancient Greece (the latter
likely even earlier in Babylon). The geometry suggested number-theoretic
questions. Thus Fermat in the 17th century showed that there are no
Pythagorean triangles whose areas are squares and hence that x4 + y4 = z4 has
no non-trivial integer solutions (see [38], p. 141 ff., [65], p. 199 ff.).
On the other hand, geometry was used to answer number-theoretic
questions, especially to solve diophantine equations. For example, finding
Downloaded At: 22:54 30 April 2011

integer solutions of x2+y2 = z2 is equivalent to finding rational solutions of

u2 + v2 = l; that is, finding all points with rational coordinates ('rational
points') on the unit circle. If (a, b) is one such point, then the line with
rational slope t passing through (a, b) will cut the unit circle at another
rational point. Conversely, if (c, d) is a second rational point on the unit
circle, then the slope of the line joining (a, b) and (c, d) is rational. Thus one
obtains all rational points by letting t take on all rational values. Hence, if we
let (a, b) = (— 1,0) (a point on the unit circle), the line through ( — 1,0) with
slope t is v = t(u +1). Finding its point of intersection with u2 + v2 = l gives all
rational points on the unit circle: M = (1 -« 2 )/(l +t2), v = 2t/(l+t2) (see [38],
p. 4). From this we obtain all solutions of x2+y2 = z2 :x=p2 — q2, y = 2pq,
z=p2 + q2. This idea of 'embedding' a diophantine equation in Euclidean
space and using the geometry of that space to facilitate the solution of the
equation is a fundamental method in the study of diophantine equations,
apparently already employed in embryonic form by Diophantus in the 3rd
century AD (see [38, 87]). (Cf. theme 2 in which we indicated how the
'embedding' of diophantine equations in algebraic domains facilitates their
solution.)
(b) Another problem with roots in Greek antiquity is constructions with
straightedge and compass. Its resolution, over two thousand years later,
depended on the 'arithmetization' of the problem—its transformation to the
problem of the 'construction' of real numbers. The constructible numbers
form a field—a subfield of the field of algebraic numbers; since n is
transcendental, one cannot 'square the circle'. Neither can one 'double a
cube' nor 'trisect an angle'. See [4] for a nice, elementary proof using very
little field theory, and [28, 48] for historical background.
(c) The one-one correspondence between the points on a line and the real
numbers 'represents a remarkable link between something which is given by
our spatial intuition and something that is constructed in a purely logico-
 Themes in the evolution of number systems 457

conceptual way', says Weyl ([86], p. 159; see also theme 8(a)). This
correspondence forms, of course, the basis for analytic geometry, developed
by Descartes and Fermat (independently) in the early 17th century. It is a
striking example of the fruitful synthetic-analytic tension in mathematics
and it suggests the coordinatization of other geometries (e.g. projective). The
idea to pursue here is the assignment of number-like objects to the elements
(points, lines) of various geometries. This idea leads to such algebraic
systems as ternary rings, Veblen-Wederburn systems, division rings, and
fields, and to the study of various geometric properties via an analysis of the
corresponding algebraic systems. For example, the only known proof of the
result that in a finite projective plane Desargues' theorem implies Pappus'
theorem follows from the corresponding algebraic result that a finite division
ring is a field. See [19, 88] for details.
The geometric-arithmetic correspondence between the points on a line
and the real numbers was also used by Hilbert in reducing the question of the
consistency of Euclidean geometry to that of the consistency of the real
number system, subsequently dealt with by Godel. See [38 p. 73, 48 chapter
42, 89]).
(d) The important role of the complex numbers in algebra and analysis is well
known. (Their role in number theory was noted in theme 2 and will be further
Downloaded At: 22:54 30 April 2011

explored in theme 8 (c).) The prominent part complex numbers play in

geometry—in the solutions of elementary problems and the formulation of
fundamental principles—is perhaps less familiar. This can be explored at
various levels via the study of Euclidean, hyperbolic, and algebraic geometry.
For details see [14, 38, 90, 91].

8. Numbers and analysis

In this theme we will indicate, on the one hand, the roles of the real and hyperreal
number systems in the study of the foundations of analysis, and, on the other, the role
of analysis (real, complex, and p-adic) in the study of (the divisibility properties of)
the number system Z of integers.
(a) The real numbers are in the foreground or background of much of analysis,
yet they were not well understood until the late 19th century. The calculus
was grounded largely in geometry in the 17th century and in algebra in the
18th century. Even in the first half of the 19th century the proofs of several
fundamental results in analysis were based on geometric intuition (especially
of the real numbers). Among these were:
• the existence of the definite integral of a continuous function;
• the convergence of a Cauchy sequence;
• the Intermediate Value Theorem.
The realization that the foundations of the calculus are to be based on an
arithmetical (rather than a geometric) foundation for the real numbers was
due mainly to Dedekind and Weierstrass who (along with Cantor et al.) gave
'arithmetic' constructions of the reals founded on the rationals. In
particular, the above results could now be established rigorously using
(in one form or another) the completeness property of the real numbers.
 458 S. Avital and I. Kleiner

The real numbers were subsequently shown to be constructible from the

positive integers, hence analysis was shown to depend logically only on the
properties of the natural numbers. This program of building up analysis
from the natural numbers was later (1895) called by Felix Klein the
'arithmetization of analysis'. Since (Euclidean) geometry was also shown to
be logically dependent on the properties of R (and hence of AT), and much of
19th-century algebra was dominated by real or complex algebra, it seemed at
the end of the 19th century that two and a half millennia after the
Pythagoreans' dictum that 'all is number', mathematics had come full circle.
As Eves put it, 'the great edifice of mathematics was shown to be like an
enormous inverted pyramid delicately balanced upon the natural number
system as a vertex' ([14], p. 132).
The purpose of this subtheme is to impart some insight into these—both
technical and conceptual—ideas. See [14, 48, 81, 92-94] for details.
(b) It became evident in the early stages of the development of the calculus that
infinitely small and infinitely large 'numbers' are very useful. The intent of
this subtheme is to trace their usefulness, beginning with Archimedes,
through Cavalieri, Leibniz, Euler, and Cauchy, and culminating (quite
recently) in Robinson's nonstandard analysis. The focus is on indicating how
the hyperreal numbers (Robinson's formalization of infinitesimals) have
Downloaded At: 22:54 30 April 2011

given rise to a new foundation of analysis (in the sense of giving alternative
answers to the fundamental questions of the subject), and thus noting that
'the epistemological foundation of mathematical analysis is far from settled'
(Steen [95], p. 92). For details see [49, 93, 96-99].
(c) Analysis—be it real, complex, or p-adic—has in the 19th and 20th centuries
played a very important role in number theory, providing yet another
example of the interplay between the continuous and the discrete
(cf. theme 7):
(i) The famous Pell equation x2 — Ny2 = 1 can be solved by expanding yjN
in an (infinite) continued fraction—an analytic process. See [33, 62].
(ii) In 1845 J. Bertrand postulated (having shown empirically for all
n<6xlO 6 ) that there is a prime between n and 2n for all positive
integers «. The proof (given by Chebyshev in 1850) uses analytic
methods (see [31], volume 1, p. 108).
(iii) Dirichlet's famous result on the infinitude of primes in an arithmetic
progression is proved using complex analysis (e.g. Dirichlet's L-series).
See [31, volume 2, 35, 58, 100].
(iv) The study of the distribution of primes among the integers (e.g. the
prime-number theorem, the zeta function, the Riemann hypothesis)
uses ideas from both real and complex analysis. See [31, volume 2, 35,58].
(v) As the above four examples indicate, to study the arithmetic of Q (i.e.
number theory) it is useful (at times essential) to 'complete' Q to R. Now
the field Q of rationals has, for each prime/), also a 'p-adic' completion
Qp ('completion' in the sense of analysis). In fact, U and Qp are all the
completions of Q (but the Qp are non-archimedean fields). Moreover,
just as the hyperreal and real numbers can be regarded on an equal
footing in analysis (this notion will undoubtedly not go unchallenged),
so the p-adic and real numbers are on a par in number theory (this is
 Themes in the evolution of number systems 459

noncontroversial). For example, it is a fundamental result that the

diophantine equation/(x) = 0 is (for some / ) solvable in Z if and only if it
is solvable in R and in Qp for all primes/). The idea here is similar to that
mentioned in theme 7(a): while there one embedded a diophantine
equation in Euclidean space in order to bring the geometry of W to bear
on the study of the equation, here one embeds the diophantine equation
in the metric spaces R and Qp in order to bring the analysis of these
topological fields to bear on the relevant equation. See [100-104] for
details.

References
[1] MACLANE, S., 1981, Amer. Math. Monthly, 88, 462.
[2] MACLANE, S., 1986, Mathematics: Form and Function (New York: Springer-Verlag).
[3] ARTMAN, B., 1988, The Concept of Number: From Quaternions to Monads and Topological
Fields (New York: Wiley).
[4] COURANT, R., and ROBBINS, H., 1941, What is Mathematics? (Oxford: Oxford University
Press).
[5] DANTZIG, T., 1967, Number: the Language of Science, fourth edition (Free Press).
[6] EBBINGHAUS, H.-D., et al., 1990, Numbers (New York: Springer-Verlag).
[7] STEWART, I., and TALL, D., 1977, The Foundations of Mathematics (Oxford: Oxford
University Press).
Downloaded At: 22:54 30 April 2011

[8] YESHURUN, S., 1989, Theta, 3, 28.

[9] NIVEN, I., 1941, Amer. Math. Monthly, 48, 654.
[10] NIVEN, I., 1942, Amer. Math. Monthly, 49, 386.
[11] HANKINS, T . L., 1980, Sir William Rowan Hamilton (Baltimore: The Johns Hopkins
University Press).
[12] VAN DER WAERDEN, B. L., 1976, Math. Mag., 49, 227.
[13] MACDUFFEE, C. C., 1944, Scripta math., 10, 25.
[14] EVES, H., 1983, Great Moments in Mathematics: (a) before 1650 and (b) after 1650 (The
Math Assoc. of Amer).
[15] KANTOR, I., and SOLODOVNIKOV, A. S., 1989, Hypercomplex Numbers (New York:
Springer-Verlag) (translated from the Russian by A. Shenitzer).
[16] KLEINER, I., 1987, L'Enseign. Math., 33, 227.
[17] MAY, K. O., 1966, Amer. Math. Monthly, 73, 289.
[18] MASSEY, W. S., 1983, Amer. Math. Monthly, 90, 697.
[19] BURTON, D. M., 1970, A First Course in Rings and Ideals (Reading (Mass.): Addison-
Wesley).
[20] VAN DER WAERDEN, 1985, A History of Algebra (New York: Springer-Verlag).
[21] HILLMAN, A. P., and ALEXANDERSON, G. L., 1983, A First Undergraduate Course in
Abstract Algebra, fourth edition (Belmont (Calif.): Wadsworth).
[22] KASNER, E., and NEWMAN, J. R., 1967, Mathematics and the Imagination (New York:
Simon & Schuster).
[23] NIVEN, I., 1956, Irrational Numbers (Washington: Math. Assoc. of Amer.).
[24] SIMMONS, G. F., 1985, Calculus with Analytic Geometry (New York: McGraw-Hill).
[25] NIVEN, I., 1939, Amer. Math. Monthly, 46, 469.
[26] BECKMANN, P., 1971, A History of π (New York: St. Martin's Press).
[27] CASTELLANOS, D., 1988, Math. Mag., 61, 67-98 and 148-163.
[28] BOYER, C. B., 1989, A History of Mathematics, revised by U. C. Merzbach (New York:
Wiley & Sons).
[29] BROWDER, F. E. (ed.), 1976, Mathematical Developments Arising from Hilbert Problems,
two volumes (Providence (R. I.): Amer. Math. Soc).
[30] NIVEN, I., 1961, Numbers: Rational and Irrational (New York: Random House).
[31] L E VEQUE, W. J., 1956, Topics in Number Theory, two volumes (Reading (Mass.):
Addison- Wesley).
[32] POLLARD, H., and DIAMOND, H. G., 1975, The Theory of Algebraic Numbers (Washington:
Math. Assoc. of Amer.).
 460 5 . Avital and I. Kleiner

[33] STARK, H., 1978, An Introduction to Number Theory (Cambridge (Mass.): M.I.T. Press).
[34] ADAMS, W. W., and GOLDSTEIN, L. J., 1976, Introduction to Number Theory (Englewood
Cliffs (New Jersey): Prentice-Hall).
[35] GROSSWALD, E., 1984, Topics from the Theory of Numbers, second edition (Boston:
Birkhauser).
[36] MAOR, E., 1987, To Infinity and Beyond (Boston: Birkhauser).
[37] RUCKER, R., 1982, Infinity and the Mind: The Science and Philosophy of the Infinite
(Boston: Birkhauser).
[38] STILLWELL, J., 1989, Mathematics and its History (New York: Springer-Verlag).
[39] DAUBEN, J. W., 1979, Georg Cantor: His Mathematics and Philosophy of the Infinite
(Cambridge (Mass.): Harvard University Press).
[40] CROSSLEY, J. N., 1987, The Emergence of Number (New Jersey: World Scientific).
[41] VILENKIN, N . YA., 1968, Stories about Sets (New York: Academic Press).
[42] HALMOS, P. R., 1960, Naive Set Theory (New York: Springer-Verlag).
[43] COHEN, P. J., and HERSH, R., 1967, Sci. Amer., 217, 104.
[44] FRENCH, R. M., 1988, Math. Intell., 10, 21.
[45] WAGON, S., 1985, The Banach-Tarski Paradox (Cambridge (Mass.): Cambridge
University Press).
[46] CIPRA, B., 1989, Science, 244, 528.
[47] GARDNER, R. J., and WAGON, S., 1989, Notices of the Amer. Math. Soc., 36, 1338.
[48] KLINE, M., 1972, Mathematical Thought from Ancient to Modern Times (New York:
Oxford University Press).
[49] DAVIS, P. J., and HERSH, R., 1981, The Mathematical Experience (Boston: Birkhauser).
Downloaded At: 22:54 30 April 2011

[50] KLEINER, I., and MOVSHOVITZ-HADAR, N., 1990, Interchange, 21, 28.
[51] MOORE, G. H., 1982, Zermelo's Axiom of Choice: Its Origins, Development, and Influence
(New York: Springer-Verlag).
[52] DAVIS, P. J., 1961, The Lore of Large Numbers (New York: Random House).
[53] AVITAL, S., 1986, Arithm. Teacher, 34, 42.
[54] HENRY, B., 1985, Every Number is Special (Palo Alto (Calif.): Dale Seymour).
[55] L E LIONNAIS, F., 1983, Les nombres remarquables (Paris: Hermann).
[56] RICHARDS, S. P., 1982, A Number for Your Thoughts (New Providence (N.J.): S. P.
Richards Publ.).
[57] WELLS, D., 1986, The Penguin Dictionary of Curious and Interesting Numbers (London:
Penguin).
[58] APOSTOL, T . M., 1976, Introduction to Analytic Number Theory (New York: Springer-
Verlag).
[59] DUDLEY, U., 1983, Math. Mag., 56, 17.
[60] RIBENBOIM, P., 1989, The Book of Prime Number Records, second edition (New York:
Springer-Verlag).
[61] BEILER, A. H., 1964, Recreations in the Theory of Numbers (New York: Dover).
[62] BURTON, D . M., 1989, Elementary Number Theory, second edition (Dubuque (Iowa):
Wm. C. Brown).
[63] EDWARDS, A. W. F., 1987, Pascal's Arithmetical Triangle (Oxford: Oxford University
Press).
[64] NATHANSON, M. B., 1987, Proc. Amer. Math. Soc, 99, 22.
[65] ORE, O., 1948, Number Theory and its History (New York: McGraw-Hill).
[66] HUNTLEY, H. E., 1970, The Divine Proportion (New York: Dover).
[67] VOROBOV, N . N., 1961, Fibonacci Numbers (New York: Blaisdell).
[68] HELLMAN, M. E., 1979, Sci. Amer., 241, 146.
[69] POMERANCE, C., 1982, Sci. Amer., 247, 136.
[70] WILDER, R. L., 1968, Evolution of Mathematical Concepts: An Elementary Study (New
York: Wiley).
[71] BOYER, C. B., 1944, Isis, 35, 153.
[72] CLOSS, M. P. (ed.), 1986, Native American Mathematics (Austin: University of Texas
Press).
[73] GARDNER, M., 1985, The Magic Numbers of Dr. Matrix (Buffalo (N.Y.):Prometheus
Books).
[74] IFRAH, G., 1985, From One to Zero (New York: Penguin).
 Themes in the evolution of number systems 461

[75] MENNINGER, K., 1969, Number Words and Number Symbols: A Cultural History of
Numbers (Cambridge (Mass.): M.I.T. Press).
[76] FLEGG, G., 1983, Numbers: Their History and Meaning (London: Andre Deutsch).
[77] GOLDSTINE, H. H., 1972, The Computer from Pascal to von Neumann (Princeton (N.J.):
Princeton University Press).
[78] KARPINSKI, L. C , 1965, The History of Arithmetic (New York: Russell and Russell).
[79] SWETZ, F. J., 1987, Capitalism and Arithmetic (La Salle (Illinois): Open Court).
[80] GILLIES, D. A., 1982, Frege, Dedekind, and Peano on the Foundations of Arithmetic (Assen
(The Netherlands): Van Gorcum).
[81] GRATTAN-GUINNESS, I. (ed.), 1980, From Calculus to Set Theory, 1630-1910: An
Introductory History (London: Duckworth).
[82] KLEINER, I., 1988, Math. Teacher, 88, 583.
[83] WILDER, R. L., 1981, Mathematics as a Cultural System (New York: Pergamon Press).
[84] ARCAVI, A., BRUCKHEIMER, M., and BEN-ZVI, R., 1982, For the Learning of Math., 3, 30.
[85] GARDNER, M., 1977, Sci. Amer., 236, 131.
[86] GARDINER, A., 1982, Infinite Processes: Background to Analysis (New York: Springer-
Verlag).
[87] BASHMAKOVA, I. G., 1981, Historia Math., 8, 393.
[88] BLUMENTHAL, L. M., 1961, A Modern View of Geometry (San Francisco: W. H.
Freeman).
[89] HILBERT, D., 1959, The Foundations of Geometry (La Salle (Illinois): Open Court).
[90] SCHWERDTFEGER, H., 1979, Geometry of Complex Numbers (New York: Dover).
[91] YAGLOM, I. M., 1968, Complex Numbers in Geometry (New York: Academic Press).
Downloaded At: 22:54 30 April 2011

[92] BOTTAZZINI, U., 1986, The Higher Calculus: A History of Real and Complex Analysis From
Euler to Weierstrass (New York: Springer-Verlag).
[93] EDWARDS, C. H., 1979, The Historical Development of the Calculus (New York: Springer-
Verlag).
[94] FRASER, C. G., 1989, Arch, for Hist, of Exact Sci., 39, 317.
[95] STEEN, L. A., 1971, Sci. Amer., 225, 92.
[96] DAVIS, M., and HERSH, R., 1972, Sci. Amer., 226, 78.
[97] HARNIK, V., 1986, Math. Intell., 8, 41, 63.
[98] KEISLER, J., 1986, Elementary Calculus: an Infinitesimal Approach, second edition
(Boston: Prindle, Weber & Schmidt).
[99] VAN OSDOL, D. H., 1972, Amer. Math. Monthly, 79, 355.
[100] BOREVICH, Z. I., and SHAFAREVICH, I. R., 1966, Number Theory (New York: Academic
Press).
[101] AGNEW, J., 1972, Explorations in Number Theory (Belmont (Calif.): Wadsworth).
[102] LEWIS, D. J., 1969, in Studies in Number Theory, edited by W. J. Le Veque (Washington:
Math. Assoc. of Amer.), pp. 25-75.
[103] MACDUFFEE, C. C., 1938, Amer. Math. Monthly, 45, 500.
[104] BACHMAN, G., 1964, Introduction to p-adic Numbers and Valuation Theory (New York:
Academic Press).
[105] CROWE, M. J., 1968, A History of Vector Analysis (Notre Dame (Ind.): University of
Notre Dame Press).
[106] ZASLAVSKY, C , 1973, Africa Counts (Boston: Prindle, Weber and Schmidt).