define P as the set of jc with 0 < jc.

For then D4i) follows from A2), and

D42) is proved as follows: 0 :^ a^ Pimplies by A5) that a < 0, and thus

by A4) we have 0 = a + (—a) < —a, and therefore —a e P. As for

D43), we see that 0 < a, 0 < ^ imply by A4) that a < a -'r b and

consequently by A3) that 0 < a -\- b. Finally D44) is obtained from D5)

with a = 0. Since by A4) a < b means the same as 0 < ^ — a, the relation

< actually results in this way from P and is therefore the ordering that

corresponds to the domain of positivity P.

The assumption 0 < c in D5) is essential; for if c < 0, then 0 < — c,

so that —ac = a(—c) < b{—c) = —be and therefore be < ae; of course

for c ^ 0 we have ae = be (=0). This argument shows that for c 7^ 0

the endomorphism jc ^- jcc of the module of integers (by §2.4 every

endomorphism is of this form) is a monotone and therefore one-to-

one mapping^" (monotone increasing for c > 0 and monotone decreasing

for c < 0), whereas for c = 0 the endomorphism is not one-to-one.

A twofold application of the monotonic law A4) for addition shows that

if a < b and e <d, then a -\- e < b -\- e, b -\- e < b -\- d and therefore

by A3) a + c < 6 + </; thus, inequalities of the same kind may be added.

To obtain the same result for multiplication, we must make the additional

assumption that b, e > 0 or a, d > Q; since otherwise D5) would not be

applicable.

3. The Rational Numbers

3.1. Introduetion of the Rational Numbers

But now the integers are incomplete with respect to multiplication in the

same way as the natural numbers were incomplete with respect to addition :

not every equation jca' = a has a solution. However, multiplication is

associative and commutative in exactly the same way as addition, and

as a substitute for A1) we at least have: if ac = be and c 7^ 0, then a = b,

as follows from ae — be = (a — b)e because of the absence of divisors

of zero^^ (see §2.4). By the remark in small print at the beginning of §2.2

(for C we take the set of integers 7^0) these properties permit us to follow
the construction given there for extending a domain, provided we restrict

ourselves to pairs {a, a') with a' 7^ 0. Instead of a — a' we naturally use

the symbol aja' and replace addition by multiplication. Essential for our

present purpose is the following fact: in the product (a, a'){b, b') = (ab, a'b'),

defined in analogy with C0), the second number a'b' of the pair is also

*" Of course, unless c = 1 or c = —1, not every integer will be an image in this

mapping; we will have an isomorphism of the module onto a submodule.

*^ Consequently, the procedure we are about to describe may be carried out for any

arbitrary commutative ring that has no divisors of zero.

122 PART 6 ARITHMETIC AND ALGEBRA

7^0 because of the absence of divisors of zero. For the rational numbers

a/a' obtained in this way (they are also called fractions) the equation

a/a' = bjb' means the same as ab' = a'b, so that in particular

ac/a'c = a/a' for c 7^ 0.

Just as for the integers in §2.2, we now admit as statements about the

rational numbers a/a', b/b',... only such statements about the integers

a, a', b, b', ... as do not change their truth values when a/a', b/b' are

replaced by rational numbers that are equal to them (in the sense of

equality just defined). If, as is customary, we call a' the denominator of the

symbol a/a', then the statement "a/a' has the denominator a'" is not an

admissible statement about the rational numbers since, for example, 2/3

has the denominator 3 but 4/6 does not. Of course, there is no serious

objection to such statements, even though they are not "equality-invariant,"

provided we clearly understand their special position. But since it is

possible to avoid them altogether in mathematics, we will find it safer

and more convenient to exclude them on principle. If we do this, we can

still speak about the denominator a' of the rational number a/a' if we

assume, for example, that a' > 0, a ^ 0 and a, a' are relatively prime

(see IB6, §2.6).

On the other hand, it is customary to speak in school about the numerator

and the denominator of a fraction a/a' even when a and a' have common factors.
In this case (on account of the order in which the extensions are usually made in

school) a and a' are natural numbers. Moreover, it is a common habit to say

that fractions are equal only if they have the same numerator and the same

denominator. In this terminology the fractions are actually the pairs of numbers

(a, a').^'^ Then to obtain the rational numbers one says that the fractions a/a',

b/b' are equal in value if ab' = a'b. This equality of value is our equivalence

relation (denoted by = in the analogous developments in §2.2), For the rational

numbers one then uses the same symbol a/a as for fractions, but equality is

taken in the sense of equality of value. In contrast to our procedure, in which

pairs of numbers are denoted by symbols different from those for rational

numbers, the distinction between fractions and rational numbers is now taken

into account only in the diflFerent concepts of equality. This procedure, which is

permissible enough in itself, is obscured by the fact that only equality of value

actually appears in the formulas, so that the symbol "=" always means equality

of value, whereas the "original" equality of fractions (equality of numerator

and denominator) occurs only in informal statements. Here again (in analogy

with the use of residue classes or of the equivalence relation = mentioned

on p. 109) we may consider a rational number as the set of fractions that are

equal in value to a given fraction a/a; then the rational number is represented

by the fraction a/a' (or by any other fraction with the same value), and equality

of value of fractions means exactly the same as equality of the rational numbers

represented by them (in the sense of the definition of equality for