vocabularies was sufficient to bring out the most striking relationships. The classification of North American languages, that we owe to Major J. W. Powell, which will form the basis of all future work, was made by this method. Further progress on these lines is beset with great difficulties, that are common to America and to those continents in which we cannot trace the development of languages by means of historical documents. The results of the historical and comparative studies of Indo-European languages show very clearly that languages that have sprung from the same source may become so distinct, that, without documents illustrating their historical development, relationships are difficult to discover; so much so, that in some cases this task might even be impossible. We are therefore permitted to assume that similar divergences have developed in American languages, and that quite a number of languages that appear distinct may in a remote period have had a common origin.
Here lies one of the most difficult problems of research, and one in which the greatest critical caution is necessary, if we wish to avoid the pitfalls that are besetting the path of scientific inquiry. The method of investigation has to take into account possibilities of linguistic growth, in regard to which generalized data are not available. Modern languages have developed by differentiation. In so far as this is true, the establishment of a genealogical series must be the aim of inquiry. On the other hand, languages may influence one another to such an extent, that, beyond a certain point, the genealogical question has no meaning, because it would lead back to several sources and to an arbitrary selection of one or another as the single ancestral type. Our knowledge of linguistic processes is sufficiently wide to show that lexicographic borrowing may proceed to such an extent, that the substance of a language may be materially changed. As long, however, as the inner form remains unchanged, our judgment is determined, not by the provenience of the vocabulary, but by that of the form. In most Indian languages etymological processes are so transparent, that borrowing of whole words will be easily detected; and, on the whole, the diffusion of words over diverse groups does not present serious difficulties, provided the borrowed material does not undergo radical phonetic changes.
The matter is different when we ask ourselves in how far phonetics and morphological features may have been borrowed. In these cases our experience does not permit us to give a definite answer. The system of sounds of a language is certainly unstable; but in how far inner forces and in how far foreign influence mould its forms, is a question not always easy to answer. In America we can discern various areas that have common phonetic characteristics; like the areas of prevalence of nasalization of vowels, of glottalized fortes, of superabundant development of laterals, of absence of bi-labials or of labio-dental spirants, or of trills. These areas do not coincide with any morphological groupings, and are apparently geographically well defined. If we are dealing here with phenomena of late assimilation, a disturbing element is introduced that will make it more difficult to assign a language to a definite genealogical line, much more so than is the case in the borrowing of words. The conditions favoring such phonetic influence must have been much more numerous in primitive America than they were in the later development of European languages. The number of individuals speaking any given American dialect is small. Many women of foreign parentage lived in each tribe, and their speech influenced the pronunciation of the young; so that phonetic changes may have come about easily.
Still more difficult is the problem presented by the distribution of morphological traits. Even with our imperfect knowledge of American languages, it may be recognized that certain morphological types have a
