Whitney, W. D. (1875). Other families of language: Their locality, age, and structure. In W. D. Whitney, The international scientific series: Vol. 16. The life and growth of language: An outline of linguistic science (pp. 228-264). New York, NY, US: D Appleton & Company.
http://dx.doi.org/10.1037/13300-012
We have called a certain body of languages a family, the Indo-European. The name "family," we saw, was applied to it by strict analogy with the use of the same term elsewhere: the languages in question had been found, on competent examination, to show good evidence of descent from a common ancestor. We had, however, to confess that the limits, even of this best-known of families, cannot be traced with absolute precision; one or another tongue, not now thought of, or else doubtfully regarded, as Indo-European, may one day make good its title to a place with the rest. We have also seen that, by the operation of completely comprehensible causes, no language on earth exists in a state of absolute accordance through the whole community that speaks it; it is a group, even if a very limited one, of related dialects. This being the case, it is the first task of the comparative study of languages to divide all human speech into families, by recognizable signs of relationship: only thus can there be made any such examination of their character and history as shall lead the way to the other results which the science seeks to attain. And such a classification has in fact been made. It is, of course, in parts only a tentative and provisional arrangement, held liable to rectification, both by addition and by the giving up of what is now held even with a fair degree of confidence: for it not seldom happens that lines which in a half-light appear definite and fixed dissolve away when full illumination is turned upon them. The cautious philologist combines only so far as trustworthy evidences take him, leaving the rest to be settled when more knowledge is won. (PsycINFO Database Record (c) 2016 APA, all rights reserved)