188 THE FACTORS OF THE MIND

5. As a result, " Q-technique opens out an entirely new
field, namely, typology," Types are defined as " any persons
who satisfy the conditions for a non-fractional factor in
Q-technique." And whereas R-technique was confined
solely to c abilities ' and c tendencies? Q-technique turns
away from these artificial concepts and deals with the
analysis of * types'—a form of analysis (as Stephenson now
holds) that could not be undertaken by the older methods.

Alleged Differences in the Logical Nature of the Techniques.
—The whole theory is evidently full of interesting and
original suggestions. Here, however, the question that we
have to decide is not the validity, utility, or relative merits
of P-, Q-, and R-technique, but the logical nature of the
methods and the logical status of the factors that they yield.
Does the device of correlating persons, or do the conditions
under which it appears legitimate and fruitful, involve
principles that are logically different from the procedures
we have so far reviewed ?

My own opinion is decidedly that it does not. Some
slight modifications are no doubt requisite (e.g. in regard to
the minor but difficult problems of sampling error and the
like). But for the rest I am wholly opposed to any sharp
division of correlational data into two completely separate
branches, each with its own stereotyped * technique.'
Always, and throughout every field of work, we are con-
cerned in the last resort with relations between individual
persons with their characteristic traits, on the one hand,
and the various test-situations, on the other, and these
test-situations themselves may equally arise out of the per-
formances of other individual persons. Indeed, as often as
not, it is a mere incident of the investigator's standpoint
whether he regards a particular set of measurements as
describing a * person ' or as describing a ' test.'*

1 Stephenson objects strongly to this view, protesting that the same row
of figures cannot' change chameleon-like' from a * population * into a ' vari-
able ' ([136], p. 21). However, his original exposition of Q-technique was
based on this very assumption ([98], p. 354). Moreover, in some cases the
material contrast between the rows and columns of the initial matrix may
vanish almost entirely. A curious example arises in paired comparison. In
contrasting the reliability or f objectivity ' of judgments on pictures and on
weight differences respectively (somewhat along the lines of Lyman Wells),