refore,
to suppose that the Celtic-speaking invaders wiped out the previous
inhabitants of the land to a corresponding extent. Races, in short,
mix readily; languages, except in very special circumstances, hardly
at all.

Disappointed in its hope of presiding over the reconstruction of the
distant past of man, the study of language has in recent years tended
somewhat to renounce the historical--that is to say,
anthropological--method altogether. The alternative is a purely
formal treatment of the subject. Thus, whereas vocabularies seem
hopelessly divergent in their special contents, the general apparatus
of vocal expression is broadly the same everywhere. That all men alike
communicate by talking, other symbols and codes into which thoughts
can be translated, such as gestures, the various kinds of writing,
drum-taps, smoke signals, and so on, being in the main but secondary
and derivative, is a fact of which the very universality may easily
blind us to its profound significance. Meanwhile, the science of
phonetics--having lost that "guid conceit of itself" which once led
it to discuss at large whether the art of talking evolved at a single
geographical centre, or at many centres owing to similar capacities
of body and mind--contents itself now-a-days for the most part with
conducting an analytic survey of the modes of vocal expression as
correlated with the observed tendencies of the human speech-organs.
And what is true of phonetics in particular is hardly less true of
comparative philology as a whole. Its present procedure is in the main
analytic or formal. Thus its fundamental distinction between isolating,
agglutinative and inflectional languages is arrived at simply by
contrasting the different ways in which words are affected by being
put together into a sentence. No attempt is made to show that one type
of arrangement normally precedes another in time, or that it is in
any way more rudimentary--that is to say, less adapted to the needs
of human intercourse. It is not even pretended that a given language
is bound to exemplify one, and one alone, of these three types; though
the process known as analogy--that is, the regularizing of exceptions
by treating the unlike as if it were like--will always be apt to
establish one system at the expense of the rest.

If, then, the study of language is to recover its old pre-eminence
amongst anthropological studies, it looks as if a new direction must
be given to its inqui