METAPHYSICAL STATUS OF FACTORS 225

blob. . , . What we find is not a regular distribution of all
the actual states among all possible states, but a bunching
together of instances in the neighbourhood of certain sorts
of states.'l Factor-analysis might well be described as a
method of locating such * blobs.'

Thomson, as we have just seen, and with him many of
the more cautious writers (Thorndike, for example, and
M. S. Bartlett), have been led to assume that the mind
must be a relatively structureless aggregate of similar
elements independently acting and unlimited in number.2

1 Mind, loc. cit,, p. 25. It may be noted that Keynes' first principle was
intended to rule out the possibility that * natural law might be organic and
not atomic ' (loc. cit., p. 249) ; but the second seems to allow us to introduce
6 organic ' relationships. Hence, I should prefer to say that all we need is to
assume that natural law, even if organic, may be treated as atomic by way of
a first approximation. However, what is still more interesting is the fact that
Broad seems to demonstrate successfully that, once we have used the first
principle (in the form of elementary generators) to reach the second, we can
drop it out, and take the second (in the form of mutually exclusive coherent
sets of observable characteristics) as alone being fundamental. " The
hypothetical generating factors can now be regarded as no more than
convenient parameters : they may exist, but it is not necessary to suppose
that they do. . . . It must therefore be possible to eliminate them, and to
state the case wholly in terms of observable characteristics and their relations"
(loc. cit.y p. 39). This, as we have seen, is precisely what the matrix formula-
tion of the problem enables us to do in detail for any particular instance.

2 The empirical reasons for this conception are apparently that the nervous
system is built up out of innumerable cellular units and that behaviour is built
up out of innumerable reflex bonds. If some kind of structural organization
appears later within the mind, that, it is argued, is "probably because
education and vocation have imposed a structure on the mind which was
absent in the young." In considering this argument two points seem
pertinent. First, even within the cortex the cellular elements are not innately-
identical in form or function, nor have they equal and unlimited connexions
with all parts of the nervous system (including its sensory and muscular
appendages). Secondly, from the very outset the young organism reacts as a
whole to its environment as a whole: the specialized responses, though in
part innately determined, mature gradually within this integral mode of
behaviour ; and the traditional * reflex action ' of the textbook is not the
unit out of which behaviour is built up, but a comparatively late feature in
maturation (cf., for example, Coghill, * The Structural Basis of the Integra-
tion of Behaviour,* Proc. Nat. Ac. $ci*, XVI, p, 637 f.)- This conception is
an keeping with the correlational results obtained on testing children of differ-
ent ages : as I have pointed out elsewhere, " the relative influence of the
general factor is greater in earlier years as contrasted with later." " Group-