224 THE FACTORS OF THE MIND

Nor does it stand alone. We need, as we have seen, a
second principle. Briefly this may be expressed as follows :
" the almost innumerable observable characteristics of any
object may be treated as deducible from a smaller and finite
number of independent generating factors " : so that " the
objects in the field, over which our generalizations extend,
do not have an unlimited number of independent qualities :
i.e. their qualities, however numerous, cohere together in
groups." l Thus stated the principle may be regarded as a
refinement of the traditional doctrine of Natural Kinds,
Real Species, or Types.2 It is complementary to the pre-
ceding ; and warns us that the postulate of atomic unifor-
mity must be qualified by observing that the atomic elem-
ents are not absolutely isolated, but join to form systematic
sets or wholes, each more or less continuous within itself,
and not always completely discontinuous from others.
Neural cells are integrated into patterns or systems:
mental traits seem to hang together in clusters ; persons can
be classed together to form what used to be called types ;
sensations coalesce to form Gestalten, or rather (for here it
is fallacious to put the part before the whole) sensations are
simply distinguishable aspects in complex and continuous
patterns. Together, the two principles allow us to treat
the phenomena to be factorized as consisting, structurally,
of systems of correlations superimposed upon a background
of non-correlation or c chance' that is essentially structure-
less. The upshot is that the apparently unlimited number of
determinates that are actually observable may be regarded
as arising from the relations between a comparatively
limited number of determinables.

A c Natural Kind,' says Broad, c is a region containing a

1 Keynes, loc. cit.> p. 25. I take it that the term * generator' has a
mathematical meaning and was perhaps suggested by Laplace's doctrine of
generating functions. In any case, as we shall see in a moment, the notion
of a * generating factor ' is not so essential to the principle as the notion of a
c coherent group/

* See Mill, Bk. I, chap, vii, pp. 134-6. BL III, chap, jpdi, p. 107.
" Among the uniformities of co-existence which exist in nature may be
numbered all the properties of Kind. ... A portion of them . . . are
independent of causation " ; but " it is impossible in any case to be certain
that they are." Cf. also ibui>9 pp. 260 et seq., and p. 69 above.