ANALYSIS OF ILLUSTRATIVE GROUP 407

saturation coefficients are partly positive, partly negative. It is
thus a bipolar factor. The saturations are positive for sociability,
assertiveness, anger, joy, and sex ; negative for submissiveness,
tenderness, disgust, fear, sorrow, and curiosity. Both the twofold
allocation and the relative size of the figures agree pretty closely
with what was obtained by the simpler method here described (last
column of Table III). We may infer that positive values of the
factor indicate the aggressive, unrepressed, or extraverted type, and
negative values, the inhibited, repressed, or introverted type.
Incidentally, for purposes of practical diagnosis it is instructive to
note that, of the more special characteristics assessed in the original
schedule, talkativeness still shows a higher correlation with the
second factor than any other.1

To the figures for the third factor little importance can be
attached. In this group it contributes barely 10 per cent, to the
total variance. The saturation coefficients are positive for tender-
ness, submissiveness, joy, sex, curiosity, assertiveness, and socia-
bility ; negative for sorrow, fear, anger, and disgust. The first
list includes all the emotions that are attended with pleasure ; the
latter comprises the only four that are definitely unpleasant, the
coefficient for sorrow being the largest of all. Put in this way the
results are consistent with the suggestion that the third factor
makes for cheerful or optimistic moods when positive, and
melancholy or pessimistic moods when negative : if we prefer a

1 With the homogeneous group it was -78 ; with the present group,
slightly lower, tfji. Statistically the difference is hardly significant, but it is
in keeping with the complex nature of this symptom, as I have implied else-
where. The importance of an assessment for talkativeness lies in the fact
that the qualities most reliably assessed at an interview are those which the
interview itself elicits (cf. [53], p. 64); but it may indicate one or more of at
least three underlying tendencies. Speech and voice are no doubt influenced
quite as much in their quality by specific emotions as facial expression is;
but they are also affected in their quantity (or fluency) by (i) general
emotionality, (2) special inhiUtive tendencies, and (3) the verbal factor,
usually regarded as a specific intellectual capacity, but as such probably of
relatively small importance. Thus, the tongue-tied, taciturn child may be
either unemotional, or inhibited, or (less likely) of poor ability in the use of
words; the loquacious, garrulous child is usually both emotional and
uninhibited (i.e. of the so-called extraverted or cyclothymic type). Now
in the homogeneous group, differences in general emotionality were elimin-
ated by the mode of selection; consequently, talkativeness, when present,
appeared as a well-marked symptom of extraversion: hence the higher
correlation. (Cf. The Backward CMld, pp. 401, 547. The bearing of this
on the use of performance instead of verbal tests with such children has been-
repeatedly emphasized, but is still too often overlooked.)