240 THE FACTORS OF THE MIND

To these questions several answers may be offered. The
first, and perhaps the most final, were it only convincing,
would be one we have already encountered, namely, that the
simplest functions have the highest a priori probability.
Indeed (though their views are never expressed precisely
in these terms), that would seem to be the answer favoured
by the majority of those writers who incline towards
Spearman's methods or towards methods developed out of
his—for example, Thurstone and Thouless.1 " The desire
to find * realities? behind the phenomena," says Thomson,
" appears to be strong in Thurstone : his belief that, when
c simple structure' is achieved, the factors have more
significance than that which attaches to mere coefficients is
of the greatest interest," and seems to imply a " refreshing "
faith that " mathematical elegance is bound to correspond
to physical or mental entities or actualities." z The same
attitude, too, would, I suppose, be taken by those mathe-
matical physicists whose familiar c principle of simplicity*
or ' overdeterminism' Thurstone takes to be axiomatic
for every branch of science. " Of the laws that fit the
data," we are told, " the simplest is most likely to be cor-
rect " 3 ; and " of all functions by which our laws can be
expressed, linear functions are the simplest." If we pre-
sented our tables to a worker of experience in the field of
applied mathematics, without stating that they were based
on psychological rather than physical measurements, he
would assuredly advise us to seek the simplest formula ; and
if he gave any reason (other than the mere saving of labour)
it would no doubt be that " the simplicity of a formula is
a better guarantee of probability than accuracy of fit." 4
Yet, when dealing with the complex phenomena of psycho-
logy as distinct from the simple phenomena of physics, I
myself find it difficult, as I have already confessed, to believe
that the simplest explanations can always claim the highest
a priori probability.

1 Cf. the discussion of much the same issue between Thomson and
Thouless, Human Factory IX, 1935, p. 3,
* Loe. «*., 1938, pp. 4-5.
8 Jeffreys, H., Scientific Inference, 1931, p. 38,
< Jeffreys, loc* cit^ p. 40 ; cf. Johnson, Logic, Ft. II, chap, xi, p. 248.