FINAL CONCLUSIONS 441

intrinsic influences as well. Thus, our final conclusions
in regard to the distribution of temperamental tendencies
are in complete accord with those previously reached for
the distribution of intellectual capacities ([35], p. 34 ;
[41], p. 162). Just as age, social environment, and special
pathological influences disturb the symmetry and nor-
mality of the curves for intelligence., so they seem also to
disturb the expected curves for the distribution of tem-
peramental types. The deviations that result, however,
are comparatively slight. For most practical purposesl
(e.g. the construction of rating scales) it seems reasonable
to assume that, except where special causes grossly intervene,

Secondly, owing to his strong migratory propensities, the different varieties
thus arising have been continuously crossed and recrossed, so that any
existing group (at any rate in civilized countries) exhibits a wider individual
variability than any wild creature found in the same habitat. On Mendelian
principles we should therefore expect to find, not the sharply segregated
types and the bimodal distribution that Stephenson seems to anticipate in
mixed communities of human beings as well as in mixed communities of cats
and dogs, nor yet the more or less uniform blend that he seems to attribute
to me, but countless recombinations, ranging from rare examples of one
primitive type at one extreme to rare examples of the opposite type at the
other, with most individuals varying continuously about a common mode and
merely tending towards one extreme or the other.

However, I fully agree that it is high time that problems of mental inherit-
ance and sex-differences should be taken, up afresh by more up-to-date pro-
cedures* The psychological factorist will find it instructive to compare with
his own procedure the statistical methods independently worked out by the
geneticist for the detection and estimation of segregation, linkage, and
heterogeneity in biological work : a convenient summary with references is
available in' The Measurement of Linkage in Heredity,' by K. Mather [126].
Ah up-to-date resume of the literature on * Methods of Assessing Tempera-
ment and Personality* is to be found in Dr. C. J. C. Earl's excellent chapter
in Tfa Study of Society (ed. F. C. Bartlett et <z/.). The concluding sentences
of his section on ' Typology' (p. 234 a<L -fin) seem to me to express, very
clearly and succinctly, the right provisional conclusion.

1 As I have argued elsewhere, the best general distribution for obtaining
a close theoretical fit to the frequencies of mental traits is that given by the
hypergeometric series. The series given by the symmetric binomial (from
Which the normal curve is derived) is merely a special case of this more general
series. The latter distribution will include cases in which the contributory
causes are not independent and in which the chances are not equal; and thus
will enable us to introduce any requisite degree of asymmetry into the
theoretical frequencies : (see The Backward Child, pp. 664 et /<?£.).