206 THE FACTORS OF THE MIND

tests or abilities as causes, and individual performances as
their effects, then, as in physics, our equations should in
general become reciprocal or reversible. Within the
limits imposed by errors of measurement and sampling, we
should be able to deduce trait-measurements from factor-
measurements by the means of the same equations that are
used to deduce factor-measurements from trait-measure-
ments. And once again we shall be led to identify person-
factors and test-factors ; for our conclusion implies that the
factor-measurements derived from observations on the
same sample of traits in the same or similar samples of the
population should in principle be equivalent whether we
proceed by grouping traits to correlate persons or by
grouping persons to correlate traits. To suppose that the
two sets of factors will completely change when we turn
from one alternative procedure to the other would be rather
like announcing that Boyle's law will hold good if you
consider the effect of changing temperature upon the volume
of a certain gas, but that a different formula will be required
when you treat the change of volume as cause and the
change of temperature as effect.

Cautions for Research.—The foregoing conclusions lead at
once to several obvious corollaries of immediate importance
for the research worker. More especially, we see at almost
every point how dependent the emerging factors will be, not
only upon the initial choice of persons to be tested or observed^
but also upon the initial choice of tests or traits. The early
factorists believed that the factors they were discovering
were clear-cut, concrete entities, forming part of the well-
defined structure of the mind : it scarcely mattered what
tests were picked for correlation, the general factor and the
group-factors (if any) would surely disclose themselves.
And their later followers have further maintained that
" no matter what sample of the population we take, our
technique will enable us to classify them into types as
distinct as primroses and buttercups." These two beliefs
are more in keeping with the views of the scholastic logicians
than with those of the present day. The first belief seems
implicitly to take for granted the old traditional notion that
the behaviour botfr of the mind and of material substances