kinds, evidence from arms, ornaments, pottery, modes of burial. But all
these are secondary; the primary ground of classification is the
physical conformation of man himself. As to language, the ethnological
method, left to itself, can find out nothing whatever. The science of
the ethnologer then is primarily physical; it is historical only in that
secondary sense in which palaeontology, and geology itself, may fairly be
called historical. It arranges the varieties of mankind according to a
strictly physical classification; what the language of each variety may
have been, it leaves to the professors of another branch of study to
find out.

The science of the philologer, on the other hand, is strictly
historical. There is doubtless a secondary sense in which purely
philological science may be fairly called physical, just as there is a
secondary sense in which pure ethnology may be called historical. That
is to say, philology has to deal with physical phenomena, so far as it
has to deal with the physical aspect of the sounds of which human
language is made up. Its primary business, like the primary business of
any other historical science, is to deal with phenomena which do not
depend on physical laws, but which do depend on the human will. The
science of language is, in this respect, like the science of human
institutions or of human beliefs. Its subject-matter is not, like that
of pure ethnology, what man is, but, like that of any other historical
science, what man does. It is plain that no man's will can have any
direct influence on the shape of his skull. I say no direct influence,
because it is not for me to rule how far habits, places of abode, modes
of life, a thousand things which do come under the control of the human
will, may indirectly affect the physical conformation of a man himself
or of his descendants. Some observers have made the remark that men of
civilized nations who live in a degraded social state do actually
approach to the physical type of inferior races. However this may be, it
is quite certain, that as no man can by taking thought add a cubit to
his stature, so no man can by taking thought make his skull
brachycephalic or dolichocephalic. But the language which a man speaks
does depend upon his will; he can by taking thought make his speech
Romance or Teutonic. No doubt he has in most cases practically no choice
in the matter. The language which he speaks is practically determined
for him by fash