Page:Primitive Culture Vol 1.djvu/320

302 the practice of attributing sex to objects that have none is not easy to explain in detail, yet there seems nothing mysterious in its principles, to judge from one at least of its main ideas, which is still quite intelligible. Language makes an admirably appropriate distinction between strong and weak, stern and gentle, rough and delicate, when it contrasts them as male and female. It is possible to understand even such fancies as those which Pietro della Valle describes among the mediæval Persians, distinguishing between male and female, that is to say, practically between robust and tender, even in such things as food and cloth, air and water, and prescribing their proper use accordingly. And no phrase could be more plain and forcible than that of the Dayaks of Borneo, who say of a heavy downpour of rain, 'ujatn arai, 'sa!' 'a he rain this!' Difficult as it may be to decide how far objects and thoughts were classed in language as male and female because they were personified, and how far they were personified because they were classed as male and female, it is evident at any rate that these two processes fit together and promote each other.

Moreover, in studying languages which lie beyond the range of common European scholarship, it is found that the theory of grammatical gender must be extended into a wider field. The Dravidian languages of South India make the interesting distinction between a 'high-caste or major gender,' which includes rational beings, i.e. deities and men, and a 'caste-less or minor gender,' which includes irrational objects, whether living animals or lifeless things. The distinction between an animate and an inanimate gender appears with especial import in a family of North American Indian languages, the Algonquin. Here not only