372 THE FACTORS OF THE MIND

definite example the particular view of factors and of factor-
analysis, which I have endeavoured to advance in the pre-
ceding pages : namely, that factors are essentially principles
of classification, and that factor-analysis is merely a device
for assigning an individual member, whether trait or
personality, to its appropriate class on the basis of an
average (or weighted sum) of a relevant set of assessments.

The procedure is simple enough in actual practice : but
the theoretical assumptions on which it rests have recently
become the subject of much criticism. Hence the first
two chapters will be concerned more with the discussion
of methods than with the demonstration of results. My
critics have rightly pointed out that, in my justification of
the general principles employed, 'several statistical diffi-
culties are involved which [my earlier exposition] passed
over rather lightly in order to present the main idea as
clearly and briefly as possible' ([132], p. 286). Students
working in the same specialized field have also raised a
number of incidental questions which could hardly have
been discussed at length in a journal for the general reader.
These questions and these difficulties are themselves by no
means devoid of theoretical interest; and I have therefore
thought it necessary to preface the summary of concrete
results by a more detailed defence of the statistical pro-
cedure.

My plan will be to take in turn the chief objections urged
against the proposals, and to indicate the ways in which
they can be met. This will entail a criticism of my critics.
But I should like to say at the outset that my object is not
to engage in controversy, but simply to use this dialectical
procedure as the clearest means of extracting the elements
of truth that are no doubt common to both the antithetical
approaches.

p. 418)—has proved particularly helpful in regularizing assessments for com- •
plex temperamental characteristics, where more objective norms, such as are
provided on the cognitive side by standardized tests, are no longer available.
The relation between what I called ' temperamental profiles ' and the con-
stants extracted by factorizing correlations between persons is briefly
indicated in [101], p.* 65. The full method of calculation is explained in my
paper on The Analysis of Temperament [114.].