222 THE FACTORS OF THE MIND

Nevertheless, some a priori postulate is avowedly requis-
ite. And the appropriate postulate or postulates may in
their turn throw light upon the fundamental nature of the
subject-matter about which we desire to reason. So far as
physical phenomena are concerned, most logicians would
probably prefer to invoke, in the place of the traditional law
of causation, something more akin to the two postulates
proposed by Keynes, which he terms the * principle of
atomic uniformity ' and the tf principle of limited inde-
pendent variety.'1

Strangely enough, the very philosophers who accept
these two principles for the physical sciences warn us that
they " do not apply to inductive generalizations about
mental phenomena ; so that with our present knowledge
we have no good reason to attach great weight to the con-
clusions of inductive argument on these subjects " 2—a
shattering verdict for those who seek to reduce mental
phenomena to something like an empirical science. Yet I
venture to affirm that these two postulates, or something
very like them, are precisely what are needed to justify the
inductive inferences of the factorist. And if that view is
right, we could, it seems to me, contend that factor-
analysis may in this way throw an illuminating beam on the
ultimate structure of mind, or at any rate on the best work-
universal causation was needed to help us out of the apparent fallacies that all
inductive arguments otherwise seem to involve, and further that it could do
so successfully. To cite the arguments brought by more recent logicians
against such a doctrine is hardly necessary here. Of the many refutations
perhaps the most readable and forcible is that set out by Russell in Mysticism
and Logic, chap, ix.

1 J. M. Keynes, A 1 realise on Probability, pp. 256 et seq, A modified version
of them is elaborated by C. D. Broad, whose exposition will probably be
found a little clearer and more suggestive to the factorial psychologist.
(* Principles of Problematic Induction,' Proc. Arist. Soc., N.S., XXVIII,
pp. 1-47, and * The Relation between Induction and Probability,' Mind,
XXVII, pp. 389 et seq.). A still more recent statement of these views in
Le problems logique de ^induction, by Jean Nicod (ably summarized by
R. B. Braithwaite in Mind, XXXIV, pp. 483 ft seq.\ contains several import-
ant revisions. But even if he hesitates over trying to master the current views
of inductive logic, every factorial worker should at least be familiar with
Keynes* Treatise.

2 Broad, loc. cit., p. 46.