VARIOUS USES OF FACTORS 31

effective means towards certain desirable ends. How is this
transition bridged ? The logical link is indicated in the
writer's introductory account of his statistical procedure.
Here he describes correlation as " a method of measuring
the agreement between two variables " ; and then, like
other writers before him, quotes (or, rather, significantly
misquotes) Mill's canon : " The method of agreement
assures us that, if two or more instances of a phenomenon
have one factor in common, that common factor is the cause
of the given phenomenon." According to the axiom of
universal causation, therefore, we tacitly assume that a com-
mon factor is a common cause. I shall later consider how far
factorial arguments necessitate some such causal postulate.
Meanwhile, let us note how the method of agreement is
used in the problem before us. It furnishes what the in-
vestigator calls a c decisive verification.' " Our generaliza-
tion," he writes, " appears confirmed beyond dispute, when
we examine the correlations of each test and of each school
subject with the common factor, i.e. as they are called,
their loadings or saturations, which indicate the extent of
their agreement with that common factor. All the satura-
tions are positive, and, with one exception, all are significant.
Thus high achievement at school and high ability in the fac-
tor everywhere coincide, . . . We are consequently able to
explain all intellectual progress at school ass the effect of
one simple noegenetic law." 1

But is this the only conceivable explanation ? I observe
that the instructions to each of the writer's group tests
were given in abstract verbal terms; and, from my own
experience of group-testing, I suspect that the understanding
of the instructions presented far greater difficulties to the
young examinees than the solution of the actual test-prob-
lems. Accordingly I might argue with equal plausibility :
" the understanding of abstract verbal instructions is a
factor common to all the tests ; it is therefore the general
cause of success in both the psychological and the educa-
tional problems."

1 The writer has previously argued that " the third noegenetic law " (the
eduction of correlates) " really depends on the second " (the eduction of
relations) and that " the first'. . . is not a noegenetic law at all."