FINAL CONCLUSIONS 429

soon become familiar—e.g. those characteristic of the over-sexed
child or the anxiety-neurotic.

B. DISTRIBUTION OF TYPES

We come now to our final problem—to examine the
frequency-distribution of the temperamental tendencies as
thus assessed within the population as a whole and within
the chief samples of it that are conveniently accessible to
character-assessment.

Writers who have announced type-distinctions like those between
the extravert and the introvert, or the cyclothyme and the schizo-
phrene, have commonly spoken as if such types must be mutually
exclusive, or at any rate as if the amount of overlapping were all
but negligible., so that nearly every person in the population could
be allocated to one type or the other. Earlier psychologists indeed
held such mutual exclusion to be an essential requisite of a sound
logical classification. So far from believing that ' the commonest
type is the mixed or intermediate type/ they quote with approval
Kant's dictum : " Also gibt es keine zusammengetzten Tempera-
mente." x In his recent discussions of the problem Dr. Stephenson
supports this traditional view; and his main criticism against
" writers from Professor Burt, on the one hand, to Clarke Hull, on
the other " is that, on their assumptions, " types at most can only
be regarded as extreme cases, tail ends of normal and continuously
graded distributions." Any such deduction he emphatically
rejects.2 According to his own theory, " factors in Q-technique will
usually not be universal: as Stern puts it, not everyone is a pick-
pocket ; but we can determine how many are of one factor or type.
Some persons would be of one type ; some of another ; some of
neither ... A person who has a zero saturation coefficient and
correlates with the type factor by o-oo will have none of that type
in his make up" ([98], pp. 357-8). And in the last group (of 21
students) that he factorizes—though not in the first—there is a
remarkable bunching of saturations for each of his two types first
over the high values at the upper end of the scale and secondly
over values in the neighbourhood of zero.3 Such a bimodal distribu-
tion, I readily agree, is the very reverse of the frequency-diagram
given in the paper of mine from which he quotes.

This notion of clear-cut types has been accepted by several
psychiatrists; but it runs quite contrary to the current teaching

1 Antbropologie in Praktiscber Hinsicbt, II, § 87.

3 Character and Personality, IV, iv, pp. 295,296. 8 Ibfd., pp. 298-9,302.