416 THE FACTORS OF THE MIND

The identifications which I have proposed could be proved or
disproved more conclusively if we had fuller information about the
nature of the factors discovered. As a rule, each is just given a
single descriptive name. In none of the experiments so far
mentioned is any quantitative specification of the factors
attempted, e.g. by computing factor-measurements for the several
traits (which are, of course, equally trait-descriptions for the
several factors). After all, to make up a group containing (a)
manic-depressives and (£) schizophrenes, and then to discover by
factor-analysis that the group comprises two antithetical types,
namely, the manic-depressives and the schizophrenes, yields of
itself no great addition to our theoretical knowledge—though no
doubt it may be of great practical interest as confirming the diag-
nosis of the patients, provided of course the doctors making the
diagnosis did not also make the assessments. The problem for the
psychologist is to ascertain what are the traits that more particularly
characterize the schizophrene and the manic-depressive respectively.
Are we, for example, right in equating them with the * schizothymic-
cyclothymic types ' of his preceding research ? Or again what
degree of relative predominance do the several Kretschmer traits
exhibit in each ? May we sum them up by saying that they show
the inhibitive or the aggressive emotions developed to excess ?

In a more recent research,1 Stephenson carries his analysis
one stage further : he applies the regression-equation, deduces
factor-measurements for the traits, and so provides material
for an answer to these questions. If what he terms my ' reciprocity
principle 9 is correct, these factor-measurements should correspond,
not only with my own factor-measurements for traits obtained by
correlating persons, but also with the saturations for traits
obtained by correlating the traits themselves. If, on the other
hand (as he declares), the principle " involves such arbitrary
foundations that it scarcely merits a moment's consideration,"
then there should be no resemblance whatever. He bases his
figures on correlations between 21 normal persons, calculated from
self-estimates for a modified list of 22 traits. He obtains a bipolar
table of correlations, and deduces two antithetical types and two
contrasted factors. Once again, the second factor is virtually the
same as the first with the signs reversed : indeed, if we eliminate
what I regard as the irrelevant e general factor,' the correlation
between the two sets of saturation coefficients rises to practically
— i»oo.

On turning to his list of factor-measurements we find that

1 Character and Personality, IV, iv, pp. 303—4.