PHILOLOGY 777 stronger than the diversifying ; and with culture at its full height, and spread equally to every land and race, one universal language, like one universal community, is not an absurdity or theoretic impossibility, but only a Utopian or millennial dream. Dialectic variation is thus simply a consequence of the movements of population. As the original human race or races, so the divisions or communities of later formation, from point to point through the whole life of man on the earth, have spread and separated, have jostled and inter fered, have conquered and exterminated or mingled and absorbed ; and their speech has been affected accordingly. Hence something of these movements can be read in the present condition of languages, as in a faithful though obscure record more, doubtless, than can be read in any other way, however little it may be when viewed absolutely. Dialectic resemblances point inevitably back to an earlier unity of speech, and hence of community ; from what we know of the history of speech, they are not to be accounted for in any other way. The longer the separation that has produced the diversity, the greater its degree. With every generation, the amount of accordance decreases and that of discordance increases ; the common origin of the dialects is at first palpable, then evident on examination, then to be made out by skilled research, then perhaps no longer demonstrable at all ; for there is plainly no limit to the Families possible divergence. So long, now, as any evidence of of original unity is discoverable we call the languages speech. u related dialects," and combine them into a "family." The term " family " simply signifies a group of languages which the evidence thus far at command, as estimated by us, leads us to regard as descended by the ordinary pro cesses of dialectic divarication from one original tongue. That it does not imply a denial of the possibility of wider relationship is obvious from what has been said above. That there is abundant room for error in the classification represented by it is also clear, since we may take purely accidental resemblances, or the results of borrowing, for evidence of common descent, or may overlook or wrongly estimate real evidences, which more study and improved method will bring to light. Grouping into families is nothing more than the best classification attainable at a given stage in the progress of linguistic science ; it is in no small part provisional only, and is always held liable to modification, even sweeping, by the results of further research. Of some families we can follow the history by external evidences a great way back into the past ; their structure is so highly developed as to be traced with con fidence everywhere ; and their territory is well within our reach : such we regard with the highest degree of con fidence, hardly allowing for more than the possibility that some other dialect, or group, or now-accepted family even, may sometime prove its right to be added on. But these are the rare exceptions ; in the great majority of cases we have only the languages as they now exist, and in more or less scanty collections, of every degree of trust worthiness ; and even their first grouping is tentative and incomplete, and involves an adjournment of deeper ques tions to the day of more light. To complete and perfect the work of classification by relationship, or the establish ment of families and their subdivisions, is the first object of the comparative study of languages. No other classifi cation has a value in the least comparable with it ; that by grade of structure is a mere recreation, leading to no thing ; that by absolute worth is of no account whatever, at any rate in the present state of our knowledge. On genetic relationship, in the first place, is founded all investi gation of the historical development of languages ; since it is in the main the comparison of related dialects, even in the case of families having a long recorded history, and elsewhere only that, that gives us knowledge of their earlier condition, and enables us to trace the lines of change. In the second place, and yet more obviously, with this classification is connected all that language has to teach as to the affinities of human races ; whatever aid linguistic science renders to ethnology rests upon the proved relationships of human tongues. That a classification of languages, to which we have Recapi now to proceed, is not equivalent to a classification of ulatior races, and why this is so, is evident enough from the principles which have been brought out by our whole discussion of languages, and which, in their bearing upon this particular point, may well be recapitulated here. No language is a race-characteristic, determined by the special endowments of a race ; all languages are of the nature of institutions, parallel products of powers common to all mankind the powers, namely, involved in the application of the fittest available means to securing the common end of communication. Hence they are indefinitely trans ferable, like other institutions like religions, arts, forms of social organization, and so on under the constraining force of circumstances. As an individual can learn any language, foreign as well as ancestral, if it be put in his way, so also a community, which in respect to such a matter is only an aggregate of individuals. Accordingly, as individuals of very various race are often found in one community, speaking together one tongue, and utterly ignorant of any other, so there are found great communi ties of various descent, speaking the dialects of one common tongue, Avhich at some period historical circumstances have imposed upon them. The conspicuous example, which comes into every one s mind when this subject is discussed, is that of the Romanic countries of southern Europe, all using dialects of a language which, 2500 years ago, was itself the insignificant dialect of a small district in central Italy ; but this is only the most important and striking of a whole class of similar facts. Such are the results of the contact and mixture of races and languages. If language- history were limited to growth and divarication, and race- history to spread and dispersion, it would be 9 compara tively easy task to trace both backward toward their origin ; as the case is, the confusion is inextricable and hopeless. Mixture of race and mixture of speech are coincident and connected processes ; the latter never takes place without something of the former ; but the one is not at all a measure of the other, because circumstances may give to the speech of the one element of population a greatly disproportionate preponderance. Thus, there is left in French only an insignificant trace of the Celtic dialects of the predominant race-constituent of the French people ; French is the speech of the Latin conquerors of Gaul, mixed perceptibly with that of its later Frankish conquerors ; it was adopted in its integrity by the Norse conquerors of a part of the land, then brought into Britain by the same Norsemen in the course of their further con quests, this time only as an element of mixture, and thence carried with English speech to America, to be the language of a still further mixed community. Almost every possible phase of language -mixture is traceable in the history of the abundant words of Latin origin used by American negroes. What events of this character took place in pre historic time we shall never be able to tell. If any one chooses to assert the possibility that even the completely isolated dialect of the little Basque community may have been derived by the Iberian race from an intrusive minority as small as that which made the Celts of Gaul speakers of Latin, we should have to admit it as a possibility yet without detriment to the value of the dialect as indicating the isolated race -position of its speakers. In strictness, language is never a proof of race, XVIII. 98
