METAPHYSICAL STATUS OF FACTORS 221

because psychologists are unfamiliar with any other form
of law. Until recently, the idea of a functional law as
distinct from a causal law seems to have been quite foreign
to them ; the nature of correlation, for example, has nearly
always been explained to the novice in terms of causes
rather than of dependence. Yet, after all, the most that
science requires for its probable inferences, whether
theoretical or practical, are not so much laws of cause and
effect as laws of ground and consequence. What we need
to know is not the cause of an occurrence, but the reason
for a conclusion. When we say c A causes B,' all that we
mean is 6 A's existence at the time of determination implies
B's existence at the time to which our predictions refer,
provided other conditions do not appreciably interfere.'
Our questions therefore must be, not what conditions
compel such and such things necessarily to happen, but what
conditions enable us to infer that they will probably happen.
For this reason, as it seems to me, the notion of logical
dependence lies at the very root of factor-analysis in all its
forms and all its applications; and for this reason the more
special problems of linear dependence and statistical
dependence are always coming to the fore.

The Postulates of Inductwe Inference.—When this view is
put to the empirical psychologist, I find he is usually ready
to admit it for what he calls the deductive sciences ; but
he still insists that laws of causation are necessary for psycho-
logical science, because it is essentially inductive. Without
the fundamental premiss of causality, he supposes, there
can be no valid induction. The modern logician, however,
would at once point out that reasoning by causes is popular
reasoning, not rigorous reasoning, and that neither the law
of universal causation, nor yet (what has frequently been
identified with it) the axiom of the uniformity of Nature,
is necessary or even sufficient to guarantee the validity o£
inductive proof.1

1 The only logician quoted by factorists seems to be Mill. His law of
causation, his canon of induction, together with the principle of parsimony
or simplicity (wrongly attributed to Occam), are adduced again and again.
Until recently, indeed, not only empiricist logicians of the school of Mill,
but even rationalist logicians like Joseph, would have agreed that a law of