64 THE FACTORS OF THE MIND

course, I do not seriously contend that there is anything like a
one-to-one correspondence. But it seems clear that, only by
following some such general plan—the method of £ progressive
delimitation/ as it was originally termed x—can the task be reduced
to manageable dimensions.

But now, be it noticed, the function of the factors is not so much
prediction as description : they aim at a first broad classification of
children and of occupations according to what we have termed their
vocational type. Moreover, any prediction or recommendation
derived solely from factorial specifications would be so wide and
mechanical that it would certainly need to be supplemented by a
qualitative study of a more intensive kind. Hence, as we argued in
our Report, once a tentative classification in terms of factors has been
made, " the problem in its final stage should be turned into one of
selection rather than of guidance " ; and here we must have recourse
to direct estimation as before.

Some of the considerations I have brought forward in the last
few pages may lead us to wonder whether after all the factors as
such may not be otiose, and to inquire if eventually they may not
altogether disappear from the psychological picture. Here perhaps
it will be useful to distinguish between factors regarded as c abilities '
and factors regarded as patterns of behaviour. To my mind, even
when our arguments are couched in numerical form, the essential
thing is not the supposed c ability,' regarded as a self-subsistent
quantity that can be measured like physical energy or power, but
the pattern of correlations between the various hypothetical
performances, or, if I may use more technical language, not the
' component' as such, but the i unit hierarchy' that describes the
component.2 These patterns, these hierarchical matrices, are in

1 [S3]> PP- 57 and 81-82. Thus, to take the factors in the order proposed
in the Report, if a particular child is found to be in the middle grade for
intelligence, in the upper grade for mechanical ability, and in the lowest
grade for sociability, we should at once have reduced the number of vocations
suitable for him from 700 to about 30 (chiefly skilled trades, not requiring
co-operative work in the workshop), A further finer grading of intelligence,
a further consideration of more specialized abilities and temperamental
qualities, and a glance at reports on physical health, home background,
previous training, and the like would probably diminish the number of
appropriate openings to half a dozen or less (cf. lof. cit>> pp. 82-5 and Tables
IV and XXXIV).

2 Cf. * The Unit Hierarchy and its Properties,' Psychometrika, III, pp. 151
et seq. It may be observed that, in discussing the practical applications of
factor-analysis (e.g. to vocational guidance), Thomson himself has described
' factors? as " unnecessary middlemen between the tests and the occupational