I am tempted here to borrow Sir Herbert Risley's definition of caste. But it is a highly abstract definition, and one that cannot be easily carried in the head, even by those who have a practical and familiar acquaintance with members of Indian castes. Roughly a caste is a group of human beings who may not intermarry, or (usually) eat, with members of any other caste. There are also sub-castes which are also endogamous. Very frequently, especially in the parts of India where caste is already an institution of immemorial antiquity, a caste has allotted to it a profession or occupation. Before we discuss castes properly so called, it is convenient to speak of the tribes of India, since tribes have a tendency to become castes when they come under the pervasive influence of Hindu social ideas. In the south of India are Dravidian tribes, of which the best example are the tribes of Chota Nagpore. These are divided into a number of exogamous groups or clans, calling themselves by the name of an animal or plant, which may be regarded as their totem. The Khonds of Orissa, who once bore an evil name for their practice of human sacrifices to propitiate the earth-goddess, are divided into fifty _gochis_ or exogamous clans, each of which bears the name of a village, and believes itself to be descended from a common ancestor. These _gochis_ are the nearest known approach to the local exogamous tribe which Mr McLennan and the French sociologists believe to be the earliest form of human society. The Mongoloid tribes of Assam are much of the same kind, but in many cases, as among the head-hunting Nagas, live at perpetual warfare with one another. In such cases they usually capture their wives in war. It is interesting to note that when population grows too dense for the profitable pursuit of the chase, their principal means of livelihood, such a tribe breaks up into two or more "villages," which immediately begin waging war with one another, which is quite what a French sociologist would expect them to do. I can tell of a case within my own experience in which the headman of a parent village invited the chief of a colony village (his own nephew) to a feast and palaver with his young warriors. The guests were all treacherously put to the sword, as a means of acquiring heads and concubines. I could not get the headman to see that he had been guilty of an atrocious crime. For him, it was lawful strategy. And indeed Naga warfare is merely a series of artfully planned ambushes in which not a few of our own officers perished before we undertook the direct administration of the Naga Hills. Sir H. Risley remarks of this group of tribes that "no very clear traces of totemism have been discovered among them." Subsequent enquiries, however, show that totemistic clans do exist in some of the Assam tribes. [Illustration: _Plate III_ Dharkārs (_Mirzapur district_)]