VARIOUS USES OF FACTORS 17

(4) But the c practical necessity' (to use Thomson's
phrase) is not the only motive. The more obvious reason
for expressing our description in terms of factors is the
theoretical requirement that dictates the mode of descrip-
tion almost everywhere pursued in science—the desire,
namely, for the increased logical clearness and cogency
attained by using a few, permanent, and independent terms
of reference, instead of a large, indefinite number of casual
and semi-dependent concepts, changing from one problem
to another. Whether there are four chemical elements,
or ninety, or only two, the description of material substances
is greatly simplified if the chemist can analyse them all into
a limited number of independent and unaltering constituents.
So in psychology. The traits we can observe, the tests we
can apply, are numberless ; those actually selected vary
greatly from this investigator to that ; and yet, as the cor-
relations show, they are to a large extent functionally
dependent on each other : in short, they overlap. To
convert these correlated measurements of arbitrary and
changing traits into terms of uncorrelated components,
appearing and reappearing in successive investigations,
would not only effect an enormous economy of thought,
but (what is far more important) would greatly enhance
the precision and the validity of our logical arguments.

But, before going farther, let us note that the comparison
of psychological analysis with chemical analysis, though often
invoiced by earlier writers, has sometimes proved uninten-
tionally misleading. When we analyse table salt into
sodium, chlorine, and a residuum of impurities, we effect
an actual physical separation ; and we consequently infer
that the component atoms or elements are as concrete as
the particles of salt. With some such analogy in his mind,
the student of factor-analysis in psychology is tempted to
reify the factors named, and to visualize a logical analysis
as a physical separation, tacitly assuming that, if distinct
abilities are ever to be discovered, they will be concrete and
separable * organs,' like the heart or the lungs, and that the
* mental mechanisms' which form them will be localized in
separate brain-centres or cortical areas.

In psychology, however—and, personally, I should add in