212 THE FACTORS OF THE MIND

([56], pp. 61 f.). Similarly, Alexander declares: " We do not
seek functional entities (i.e. entities that are mere mathemat-
ical functions), but psychological entities ; our purpose is to
resolve abilities into their true psychological factors " ([82],
p. 3). And Stephenson : " A factor must be a real and
tangible entity, defined in terms of psychological needs, not
a mere statistical artefact, however elegant the procedure
by which it is reached." Even Thomson describes the
purpose of factor-analysis to be measuring " the factors of
the mind," and objects to Spearman's system because it
does not " give the causes " ([87], p. 64: his italics). For
the rest most writers fall back on tacit identifications or
equivalences. Thus Thurstone, Alexander, Holzinger, and
many others treat the words * factor ' and c ability ' as
synonymousl : but such an interpretation would rule out
all application of factor-analysis to anything besides cogni-
tive traits or tests2 ; others refer to them more broadly as

* elementary ' or c unitary traits of personality ' (Kelley), or
as £ the fundamental dimensions of the mind ' (Guilford).
Few, if any, explain why some factors are c meaningful'
and others merely c statistical/ what makes one ability more

* fundamental? or more ' elementary ' than the others, how
to distinguish ' true psychological factors ' from the rest,
or the ' causal' from the merely ' descriptive.' Neverthe-
less, nearly all appear agreed that the factors sought by the
factorist, however else they may be characterized, are at once
real and causal. Thurstone, indeed, does not hesitate to

1 E.g., [84], p. 54. Usually they speak of * underlying abilities' (Holzinger,
[106], p. 5), or of * simple/ * primary,' or ' fundamental abilities ' (Thurstone,
[122], pp. 1,2,4), again using the different adjectives as equivalent. Holzinger,
however, incidentally makes an alternative identification, speaking of ' sorting
mental tasks into distinct categories or factors ' and of * classifying traits accord-
ing to correlation clusters' ([106], p. 4), which comes much nearer to my own
view.

8 In 1912, when I first applied methods of factor-analysis to assessments of
emotional and temperamental tendencies, it was necessary to defend such an
extension against a vigorous band of critics. But of late, particularly in
America, there have been numerous attempts to factorize non-cognitive
traits of personality. Hence it is surprising to find so many factorists still
writing as though factor-analysis had been applied to nothing but tested
abilities.