58 THE FACTORS OF THE MIND

want to know, not merely what factors are general, but how far
those general factors are stable. In each case the problem is
primarily one for correlation and statistical analysis. The precise
methods that may be employed, and the provisional results so fax-
secured, I have already discussed in early writings.1 Here, there-
fore, I need do no more than emphasize this somewhat neglected
aspect of factorial work.

Again, in examining individual children we often seek, not only
to predict each child's general progress, but also to analyse out
the special weaknesses of this child or of that. For such a purpose
the heterogeneity of the Binet tests brings with it an added ad-
vantage ; for they often throw an incidental light on these more
specialized characteristics. As time goes on we shall come more and
more to rely on the results of factorial work.3 But in these cases the

1 t1^]? P- I7° f-> [22]j P* IS t-> [23]> P- 25° ^ ^n ^1C morc popular formula-
tion of my conclusions I freely availed myself of biological terms, and spoke
of these tendencies as £ inborn/ as often ' hereditary,' and as identifiable on
the conative side with * instinctive * trends. This lias led to occasional
criticism, which is, I fancy, directed rather against the associations which
those terms suggest, than against the ideas I intended to convey. I fully
agree that the distinction between what is innate and what is acquired is an
abstract and artificial distinction : so is the distinction between a projectile's
* state of uniform, motion ' and the c external forces' that compel it to deviate
from a straight line. I equally agree that the physiological and biological
evidence for innate neural dispositions, and for genes governing those dis-
positions, is at present inconclusive: yet, so far as it goes, it lends some
support to inferences from correlational work, and even throws some tentative
light upon the nature of the more permanent factors. Nor does the weakness
of the biological evidence destroy the correlational evidence, which is the
chief thing on which the practical psychologist ought to rely. Only by
correlational studies shall we be able to determine how far we can safely
predict the future performances of a child at a subsequent age from his
performances when he is first tested or, it may be, from the performances of
his parents, his brothers and sisters, and his other relatives. Only by experi-
mental checks—removing a child to a new environment, improving his
physical health, altering the way he is taught—shall we be able to assure
ourselves whether the factors we presume to be permanent—intelligence,
for example, as tested by this or that set of tests—are really permanent
and unchangeable (see The Backward Child, pp. 540-1).

2 A misstatement in the interesting article by Cattell, just cited, needs
correction, namely, that (except for a study of his own dealing with tests for
the ages below eight) " the test-items (in the Binet scale) have never been
validated by statistical treatment " (loc. cit.9 p. 115), In the work that he
quotes and criticizes (Mental and Scholastic Tests) a separate correlation co-
efficient (based on a simplified tetrachoric formula) indicating the * validity *
of every test-item is given (Table XXXI, p. 205), In an earlier t Annual