a — a' = 0, and in the only remaining case, namely a < a' [see A5)],
it follows from C6) that a — a' is the inverse —b of a natural number b.
For natural numbers b, b' we have ^ 7^ 0, — ^ 7^ 0 (since ^ + 0 7^ 0),
and b 7^ —b' (since b -\- b' :^ 0). Consequently, there exist exactly three
kinds of integers: the natural numbers, zero, and the inverses of the
natural numbers. The latter are called negative integers (or numbers of
negative sigri^"^), and then, in contrast, the natural numbers are called
positive integers.
In view of the above remarks, it would also be possible to extend the set of
natural numbers to the module of integers in the following way: For every
natural number n we introduce a new symbol —n, and also the new symbol 0,
for which —n = — m is defined as n = m and there are no other equalities
except 0 = 0. Addition is then defined as follows:
0 + 0 = 0, 0 + /w = /w + 0 = /w, 0 + (-/j) = (-/j) + 0 = -n,
i-m) + i-n) = -{m + n),
m + i — n) = {-n) + m = k, with n + k = m for n < m,
m + (—n) = ( —«) + /w = —k, with m + k = n for m < n,
m + {—m) = {—m) + m = 0.
This procedure is conceptually much simpler but has two serious disadvantages:
proofs of the rules for calculation must be divided up into many special cases
and thus become much lengthier, and addition must be required to satisfy
not only F), G), A1) but also A2), A5), which, in contrast to A3), A4), cannot
be deduced** from F), G), A1) alone.
2.4. Multiplication
In order to define multiplication, we adopt a plan which may at first
sight seem like a detour but has essential advantages over other methods.^^
Our task is to define multiplication of integers in such a way that it
satisfies the distributive law and, in the subdomain of the natural numbers,
agrees with multiplication as already defined. The distributive law
a{x -\- y) = ax -\- ay will be regarded as a property of multiplication
by a; that is, as a property of the mapping jc -^ ax. So let us first examine
mappings with this property [see C8)]. We begin with a discussion of
multiplication of the natural numbers from this point of view.
*' But this terminology readily gives rise to the common error that —a (for an arbitrary
integer a) is always a negative integer.
** Thus we cannot use this procedure in §3.1.
*' Two other possible procedures are described on p. 119 (in small print).
114 PART 6 ARITHMETIC AND ALGEBRA
By the distributive law A0) and the commutative law as proved in §1.6,
the mapping/of the set of natural numbers into itself defined by
C7) fix) = ax
has the property
C8) f(x-hy)=f(x)-hf(y)
and is therefore a homomorphism with respect to addition. If/ and g are
two such homomorphisms, it follows from/(I) = ^A) that/(A:) = g(x)
holds for all natural numbers*" x, and thus/ = g; foTf(x) = g(x) implies,
since C8) holds for g in place of /, that f(x-\-l) = f(x) +/A) =
g(x) + ^A) = g(x + 1). Thus
C9) /-/(I)
is a one-to-one mapping of the set of homomorphisms in question onto
the set of natural numbers, so that in particular each of these
homomorphisms has the form C7):
D0) f(x)=f(l)x.
Thus we have obtained a description of multiplication which is very
suitable for extension to the domain of integers. In what follows,
lowercase italic letters will refer to arbitrary integers or, when the argument is
applicable to modules in general, to arbitrary elements of a given module.
Homomorphisms (with respect to addition) of a module M into itself
are called endomorphisms of the module. Since we wish to use these
endomorphisms, as suggested by D0), in defining multiplication for the
module of integers, let us first examine in a general way the set of
endomorphisms of a module.
If / and g are endomorphisms of the module M^ the mapping