the necessary distinctions. You must please regard this only as a suggestion. We need far more detailed evidence concerning the social status of different relatives among the peoples who use these descriptive terms. Such knowledge as we possess seems to point to the dependence of the Semitic and Sudanese terminology upon the social institution of the extended family, just as our own system depends on the social institution of the family in the narrow sense and the classificatory system upon the clan.
If this descriptive mode of nomenclature be thus the outcome of a social organisation of which the essential element is the extended family, I need hardly point out how natural it is that we should find this kind of nomenclature so widely in Europe. The presence of this descriptive terminology in Celtic and Scandinavian languages, in Lithuanian and Esthonian, would be examples of the persistence of a form of nomenclature which had its origin in the kindred of the extended family. On this view we must believe that, in other languages of Europe, this mode of nomenclature has gradually been replaced by one dependent on the social institution of the family in the narrow sense.
At this point I should like to sum up briefly the position to which our argument has taken us. I have first shown the dependence of a number of special features of the classificatory system of relationship upon special forms of marriage. Then I have shown that certain broad varieties of the
