PRINCIPLES OF LINGUISTIC SCIENCE. 113 



of the whole complicate structure of the languages classified ; it must, above 

 all, be historical, holding together, and apart from others, those groups which 

 give evidence of genetic derivation from a common original. 



On reviewing this division of the families of language, any one will be struck 

 by its non-agreement Avith the divisions based on physical characteristics. This 

 brings up the important question as to the comparative value of linguistic and 

 physical evidence of race. A reconciliation of their seeming discordance must 

 be sought and finally found, for the naturalist and linguist are both trying to 

 work out the same problem — the actual genealogical history of human races — 

 and they cannot disregard each other's results. Their harmonious agreement 

 can only be the result of the greatly advanced and perfected methods and con- 

 clusions of both. Nothing more can be attempted here than to note certain 

 general considerations bearing upon the subject. 



In the first place, language is no certain evidence of descent. As was shown 

 in the first lecture, language is not inherited, but learned, and often from 

 teachers of other blood than the learner. Nor does mixture of language prove 

 mixture of race. The Latin part of our vocabulary was brought us by men of 

 • Germanic descent, who learned it from Celts and Germans, and they from a 

 mixed mass of Italians. These defects of linguistic evidence have always to 

 be borne in mind by those who are drawing conclusions in linguistic ethnology. 

 But their eftect must not be exaggerated ; nor must it be overlooked that physi- 

 cal evidence has quite as important defects. The kind and amount of modifi- 

 cation which external circumstances can introduce into a race-type is as yet 

 undetermined. Many eminent naturalists are not unwilling to allow that all 

 existing differences among men may be the effect of processes of variation, and that 

 the hypothesis of different origins is at least unnecessary. Hence, as a race 

 may change its language, and not its physical type, it may also do the con- 

 trary. Language may retain traces of mixture uudiscoverable otherwise. Lan- 

 guage may more readily and surely than physiology distinguish mixed from 

 transitional types. In many respects linguistic evidence has a greatly superior 

 practical value ; differences of language are much the more easily apprehended, 

 described, and recorded. Individual diff'erences, often obscuring race-diff'erences 

 of a physical charactei , disappear in language. Testimony coming down from 

 remote times is much more accessible and authenticable in language. Discord 

 between the two, or question as to relative rank, there is none, or ought to be 

 none. Both are equally legitimate and necessary modes of approaching the 

 solution of the same difficult and, in its details, insoluble problem, man's origin 

 and history. Each has its notable limitations, and needs all the aid it can get 

 from the other and from recorded history to supply its defects and control its 

 conclusions. But the part which language has to perform in constructing the 

 ethnological history of the race must be much the greater. In laying down 

 grand outlines, in settling ultimate questions, the authority of physiology may 

 be superior ; but the filling up of details, and the conversion of a barren classi- 

 fication into a history, must be mainly accomplished by linguistic science. 



Another important question is, what has the study of language to say re- 

 specting the unity of the human race 1 This question can already be pretty 

 confidently answered, but the answer must be a negative one only. Linguistic 

 science can never hope to give any authoritative decision upon the subject. To 

 show that it can never pretend to prove the ultimate variety of human races is 

 very easy. It regards language as something which has grown by degrees out 

 of scanty rudiments. It cannot assume that these rudiments were produced 

 by any other agency than that Avhich made their after combinations. It cannot 

 say how long a time may have been occupied in the formation of roots, or how 

 long the monosyllabic stage may have lasted ; and it must confess it altogether 

 8 s 



