METAPHYSICAL STATUS OF FACTORS 223

ing conception that can be adopted in regard to that
structure. What is more, the particular conception thus
suggested would appear to be, not merely consistent with,
but actually confirmed by, recent views on the general
character of mental and neural phenomena themselves.

The first assumption is, as Keynes points out, a generaliza-
tion of the principle known to mathematicians as the
' superposition of small effects.' It maintains that the
universe and its processes may be treated as consisting of
quasi-atomic elements, so that " a change of total state
may be considered to be compounded of a large number of
smaller separate and independent changes/31 A mental
change, for example, might be regarded as the resultant of
numerous all-or-none discharges of certain nerve-cells.
The relevance of Keynes' postulate to current statistical
reasoning is plain. In factor-analysis it supports the de-
mand for an ultimate analysis into independent or * ortho-
gonal ' factors; and it would seem to form the implicit
basis on which Thomson's arguments for a ' sampling view '
of factorial problems must really rest. It is an assumption
that colours the whole outlook of psychologists who belong
to the analytic and determinist school. Behaviourists like
Watson, associationists like Herbart, Titchener, and Thorn-
dike, determinists like Freud, have explicitly invoked some
such atomistic principle—usually in what would now be
regarded as a crude and untenable shape. The factorist
has similarly been accused by the Gestalt and Intuitionist
schools of clinging to an atomistic view of mental process
that " inevitably disrupts the personality into separate
bits."2 However, if, in Keynes' formulation of his
principle, we substitute the word * distinguishable' for
c separate,' such criticisms lose much of their force ; in any
case, neither analysis nor inductive inference as ordinarily
stated seem able to dispense with some such postulate.

1 Keynes, loc. cit., p. 249. In his recent Tamer lectures Eddington main-
tains that the mode of analysis in physics rests on what would seem to be a
very similar principle—viz. ' the atomic concept or the concept of identical
structural units.' He regards this as a necessity of the logical framework of
our scientific thinking (The Philosophy of Physical Science, 1939, p. 122).

2 Cf. Vernon, Character and Personality, IV, pp. l-io, and Spearman's
reply, ibid, pp. Il-i6.