viii PREFACE

cular writers ; the last attempts to illustrate and defend
the views I have advanced in what would seem to be the
most effective way—namely, by an actual application of the
alternative methods to a concrete issue of general interest,
the problem of temperamental types.

Part I is itself the outcome of an experiment. Two
years ago, I found myself compelled at short notice to take
over a course of lectures on factor-analysis to a group of
students, who, like most psychological students in this
country, had little mathematical or statistical knowledge.
Technicalities were therefore reduced to a minimum ; and
the main argument was based on a logical rather than a
mathematical approach. Presented in this light, the
fundamental ideas of the statistical factorist became at once
clearer and more interesting both to the mathematical and
the non-mathematical; at the same time, they illuminated
questions in psychology and other sciences, not generally
regarded as factorial.

In arranging these lectures for the press I have also kept:
in mind the needs of the research-worker. I have dis-
cussed in some detail the fallacies that invalidate so many
theses on the subject; and, since research students enter
psychology by such different paths, I have inserted a wide
variety of proofs and illustrations, in the hope that one at
least may be intelligible to every reader. For the same
reason, it seemed necessary to expatiate at greater length on
what may be called the present growing-points—particular
issues on which in the near future research is likely to con-
centrate, such as the problem of metric units, of mental
energy as a factorial concept, of correlating persons instead of
tests, and of the various hierarchical criteria. To avoid
breaking the general thread, these side-issues, and the
supplementary illustrations and interpolated proofs, have
been printed in smaller type. They are sections which the
general reader, at any rate on a first perusal, will merely
skip or skim.

To the advanced mathematical psychologist, this mode of
presentation may seem alike prolix and inconclusive. To
forestall this criticism, I had originally intended to append a
systematic summary, giving rigorous algebraic proofs of the