190 THE FACTORS OF THE MIND

is " methodologically as different from the older techniques
as chalk from cheese."

But are the alleged differences so insuperable as he
avers ? Do they really oblige us to " break with the old
R-technique " ?

First, let us be clear where precisely the differences are
said to lie. Too often it has been supposed that the differ-
ences referred to are not logical or psychologicalx but merely
mathematical or statistical. Yet the two-factor theorems
by which the calculations are to be carried out are not in
themselves new theorems, nor even c inversions' of the old.
Both the formula and the proofs are the same as in the case
of R- and P-technique ; all that is different is the inter-
pretation of the symbols, or alternatively the symbols
themselves. And, of course, these formulae or their
equivalent had often been used before to factorize cor-
relations between persons.2

As Stephenson himself repeats, it is not a new arithme-
tical procedure, but a c new methodology, opening up a new
field of psychology,' that he is anxious to advocate. The
fundamental conceptions are different. Thus, for him
* typology ' requires us to envisage a * fresh kind 9 of factor,
related to the traits on which it is based * in a statistical and
no longer in a mathematical fashion.' And this in turn
introduces new logical principles into our demonstrations
([136], p. 280).

Alleged Differences in the Nature of the Factors.—It would
take us too far afield to discuss these arguments in detail
here. Perhaps the best criterion will be to look at the

1 This is not Stephenson's own view. " In Burt's work on person-factors
there is statistical novelty, but it is psychologically unimportant. In
Q-technique there is nothing statistically novel. On the other hand, on the
-psychological side Q-technique opens up an entirely new field " ([138], p. 280,
his italics).

2 All the equations deduced in the papers on Q-technique ([97], [98])
were included in my earlier memorandum [93] as applicable to correlations
between persons, though admittedly they were there depicted as narrower
and less useful than the more general formulae, based on the method of
weighted summation. In the worked example ([93], p. 294) the saturation
coefficients were calculated for comparison by both methods; and it was
shown that the differences were all less than -005.