246 THE FACTORS OF THE MIND

adequately definable by averages, with diagnostic characters
in perfect correlation, as though human races had remained
isolated and unmixed, like the varieties and the subspecies
of wild animals. On the contrary, throughout prehistoric
and historic times, human groups have freely migrated
and freely interbred, in a way no other creature has ever
done; in consequence, as recent genetic principles have
forced us to recognize, the notion that human beings are
still classifiable into a few racial types, with little or no
overlapping, becomes wholly untenable. Accordingly, as
I have elsewhere suggested, the statistical issues can no
longer be dealt with by the mere comparison of averages,
irrespective of variation and correlation, but must be
attacked by the analysis of variance and covariance—in
short, by a method which regards the so-called races as
fluctuating combinations of genes, resulting in relatively
stable patterns of characteristics, each pattern definable by
a factor.1

The differences between different social classes or groups
might also be studied along similar lines. Here I am
thinking, not so much of alleged innate differences in
intelligence or temperament, but rather of effective
differences in attitudes, preferences, or beliefs, as they influ-
ence the actions of different sections of the community in
our social, economic, and political life. Investigations on
such problems must necessarily be planned on a large scale ;
and that in turn introduces difficulties both in collecting and
in analysing the data. So far, the main field for extensive
surveys in social psychology has been the elementary
school. The wireless offers yet another easy avenue for
gathering rough facts on an exceedingly large scale. The

1 A preliminary trial of methods has been made by several of our students.
J. I. Cohen has obtained data for Jewish and non-Jewish persons, though
his figures and his inferences are perhaps open to some criticism (cf.
* Determinants of Physique,' J. Mental Science, May 1938, and Ph.D. thesis,
University of London Library). P. C. Hu has made a comparative study of
Chinese, English, and Anglo-Chinese children. Miss M. Davidson and
others have applied factorial methods to test-results obtained from English
and Welsh students and children. It is too early to pass any judgment
on the value of the conclusions emerging; but the applicability of the
procedure would seem at any rate to*be demonstrated.