186 THE FACTORS OF THE MIND

2. The measurements with which Q-technique is con-
cerned must be expressed in terms of an entirely new unit,
namely, * significance* This " concept of significance is
peculiar to Q-technique and new to systematic psycho-
metry. . . . No better example could be afforded of the
difference between Q-technique and R-technique " ([136],
P- 233/)- One trait is said to be more significant than
another if it is " more representative or characteristic of a
personality as an indivisible whole." Both P-technique
and R-technique worked with isolated traits, which were
measured, not by reference to the person who possesses
them, but in terms of artificial norms which " at bottom are
unscientific and unsound " ([92], p. 293) : " the new unit
of * significance ' will enable us to quantify qualities without
tearing them from their immediate context." " R-tech-
nique took the person to pieces; only Q-technique can put
him together again " ([98], p. 365, [96], p. 2O2).1

3. R-technique studied the trait as variable, with the
persons as a normally distributed population ; Q-technique
studies the person as variable, with trait-measurements as a
normally distributed population. Stephenson thus " draws
a sharp distinction between statistical variables and statis-

I should rather argue that the contrast between the fields to which the two
methods are appropriate cannot be pushed so far as Stephenson maintains,
since when we correlate persons the factor-saturations obtained for the indivi-
duals tested are virtually the same as the factor-measurements obtained for
the same individuals when we correlate traits.

1 Once again I doubt whether the contrast is as radical as is assumed.
' Significance ' seems to me merely a matter of weighting : if c anger ' is
more t representative' or * characteristic' of a given person than * fear,'
that means we must give it a larger weighting in specifying his temperamental
pattern. Moreover, in a research, dealing with say 60 traits and 200 persons
(like Thurstone's recent study), factor-patterns characteristic of the * whole
person ' can in theory be deduced just as well from correlations between traits
as between persons. Actually, in giving instructions to his observers who
are to rank each picture postcard for its significance to themselves and to
grade temperamental traits for their significance to the patient, Stephenson
uses almost exactly the same phrasing as the rest of us have done: his observers
are asked to say which postcard they " liked best/' or which mood or emotion
is ** most characteristic of the patient for its duration, prevalence, and its
intensity " : (these latter being the same criteria as I myself had used :
" intensity, frequency, duration, and after-effects " [30], p. 694).