442 THE FACTORS OF THE MIND

the distribution of innate temperamental tendencies may
be treated as normal. Provisionally, and until more exten-
sive surveys have been carried out, we might perhaps
accept the converted z-distribution (last line of Table VI)
as indicating the theoretical distribution of temperamental
variations. If, however, the variations are measured by a
coefficient of correlation (or its equivalent) then we must
be prepared to accept distortions of the kind described above.

C. SUMMARY

1. Reasons have been advanced for believing (a) that the
temperamental factors deducible from assessments for
groups of normal adults and children are essentially the same
as those previously deduced from assessments for neurotic
and delinquent children referred for clinical examination,
and (b) that, whether persons or traits are correlated, the
main type-factors (i.e. all significant factors except the
first) remain virtually the same throughout.

2. Explicit calculation further shows that this equiva-
lence of * person-factors' and ' trait-factors' still holds
good, even (a) when the correlations are derived from a
relatively unselected group of normal persons, differing
in general or average emotionality, instead of from a
selected group, relatively homogeneous as regards this
general factor ; (b) when correlations are used instead of
covariances; and (c) when the simple summation method
is employed instead of the more laborious method of least
squares. Criticisms of this so-called ' reciprocity prin-
ciple ? are shown to be inconsistent with the critic's own
experimental results.

3. Explicit calculation verifies (what had previously been
proved algebraically) that with the summation method the
correlation of a person's measurements with the average
measurements for the type is virtually identical with his
saturation coefficient. It follows that, for practical pur-
poses, in order to estimate the degree to which the several
persons approximate to a given type, it is no longer necessary
to calculate all the intercorrelations and then to perform
a systematic factor-analysis ; it is sufficient to correlate the