P-, Q-, AND R-TECHNIQUES 191

concrete nature of the factors themselves. In doing so, we
must carefully distinguish between the first or positive
* general factor ' and the secondary c type-factors.' It is
with the latter that Q-technique is primarily concerned.
Accordingly, let us glance at the type-factors first.

i. The Type-factors.—" Q-technique," we are told, "is
pre-eminently a method of isolating types. ... In the
past, type psychology could not be subjected to factor-
analysis " ([98], p. 366). Now, as I shall endeavour to show
later on, in virtue of the principle of reciprocity, the type-
factors obtained by correlating persons are, in theory at any
rate, virtually identical with those obtained by correlating
traits. But we need not invoke a disputed principle ; let
us turn to actual results. A comparison of Stephenson's
conclusions with those of previous investigators1 would, I
think, show that all the type-factors that he discovers are
analogous to those that have been discovered by R-tech-
nique and by P-technique. Take, for example, his very first
experiment [92] in which he proposed to test artistic
appreciation or taste for the same kind of objects as Bulley,
Felling, and myself had used (" colours, vases, cloth materials,
furniture, pictures, literature, and so forth"—with the
addition of " perfumes" used by Beebe-Center ([92],
p. 23). His two-factor procedure leads to two group-fac-

1 Unfortunately many of the earlier efforts with so-called P-technique
remain buried in degree and diploma theses; even when the final con-
clusions have been published, the cost of printing tables of correlations and
saturations for a hundred or more persons has apparently been prohibitive.
Da vies, however, has brought together a summary of the chief problems
attacked in this country and America ; her review appears to include all
except a few tentative experiments by junior workers ([130], cf. [101]). On
the issue raised in the text, fuller evidence will no doubt be shortly available
in the further results of research-students who have recently been working
jointly under Dr. Stephenson and myself: as soon as we found that we
differed so widely over the uses of correlating persons, a programme of work
was drawn up covering the main fields in which divergences of opinion had
arisen, with a view to setting investigators to examine each particular pro-
blem simultaneously by both procedures—by his method and by my own.
Some of the preliminary results are quoted in Stephenson's articles; and the
further data so far available—those of I. Cohen, P. C. Hu, M. Hill, and
E. Knowles—would appear to strengthen the view that the discrepancies are
more apparent than real.