20 THE FACTORS OF THE MIND

histological, physiological, and biochemical research is needed to
turn the empirical factorization into an intelligible factorization,
and even (as with the duplicity theory) to correct the inferences
from mere quantitative analysis. And if we know so little about
the physiology of colour vision, we know still less about the physi-
ology of intellectual aptitudes and temperamental traits.1

(II) Prediction.—So long as the correlations on which our
descriptive terminology is based remain mere statistical
statements of co-existence (or tendencies to co-existence),
and are not yet fully explicable from a functional stand-
point, any inference based upon factorial results must be
subject to all the limitations of an empirical induction. In
consequence, however much they may be manipulated or
transformed in the course of factorial analysis, the set of
measurements derived from the study of a single sample
remains in itself nothing more than a description of the
performances of the particular pupils tested with the par-
ticular tests employed. That holds true even if the cor-
relations which lead to, or are deduced from, the hypo-
thetical factors are made to bridge an interval of time, and
take the form of a prediction.

Suppose, for example, we apply a series of educational
tests to a hundred nine-year-olds, wait five, ten, or twenty
years, and then correlate the results of our early tests with
the children's later progress at school, or at the university,
or in after-life : the correlations still remain descriptive
of the group that has been followed up, and of that alone.
They would enable us to reconstruct what we already know,
namely, the subsequent achievements of these particular
pupils; but in themselves they can tell us little or nothing
about other groups or other individuals.

There is, however, a widespread notion that, even
though the discovery of a correlation by itself may not
justify general predictions, nevertheless the discovery of

1 In passing, may I point out that factorial work was attempted in
psychology long before factor-analysis as such was recognized as involving a
special correlational technique. The topics instanced in the text provide
familiar examples. Thus, the work on the colour-'triangle was essentially a
problem in factor-analysis; and the reader will be able to provide numerous
other examples for himself (see below, p, 84).