much importance to language as a test of race. Though they have not
wholly forgotten the facts which tell the other way, they have not
brought them out with enough prominence. But I can also believe that
many people have written and spoken on the subject in a way which cannot
be justified from a strictly scientific point of view, but which may
have been fully justified from the point of view of the writers and
speakers themselves. It may often happen that a way of speaking may not
be scientifically accurate, but may yet be quite near enough to the
truth for the purposes of the matter in hand. It may, for some practical
or even historical purpose, be really more true than the statement which
is scientifically more exact. Language is no certain test of race; but
if a man, struck by this wholesome warning, should run off into the
belief that language and race have absolutely nothing to do with one
another, he had better have gone without the warning. For in such a case
the last error would be worse than the first. The natural instinct of
mankind connects race and language. It does not assume that language is
an infallible test of race; but it does assume that language and race
have something to do with one another. It assumes, that though language
is not an accurately scientific test of race, yet it is a rough and
ready test which does for many practical purposes. To make something
more of an exact definition, one might say, that though language is not
a test of race, it is, in the absence of evidence to the contrary, a
presumption of race; that though it is not a test of race, yet it is a
test of something which, for many practical purposes, is the same as
race.

Professor Max Müller warned us long ago that we must not speak of a
Celtic skull. Mr. Sayce has more lately warned us that we must not infer
from community of Aryan speech that there is any kindred in blood
between this or that Englishman and this or that Hindoo. And both