spread, they exhausted, or even now still exhaust, all collective science. Now, what they presuppose is the belief in the possible transformation of the most heterogeneous objects into one another, and consequently the absence more or less complete of definite concepts.
Among savages, that is to say, in the least evolved societies with which we are acquainted, the mental confusion is still greater. Here even the individual loses his personality. Between him and his external soul, between him and his totem, the in- distinction is complete. His personality and that of his fellow- animal only make one. Identification goes so far that the man takes the character of the thing or the animal with which he is thus brought together. For example, in Mabiuag the clansmen of the crocodile are supposed to have the temperament of the crocodile : they are proud, cruel, always ready to fight. Among certain Sioux there is a section of the tribe which is called red, and which comprises the clans of the mountain lion, the buffalo, and the moose-deer — animals all characterised by their violent instincts. The members of these clans are by birth warriors, while the husbandmen, people naturally peaceable, belong to clans whose totems are animals naturally pacific.
What applies to men in the savage mind, applies with even greater force to things. According to a remark of Von den Steinen, our determination of species by relation to one another, so that the one does not merge into the other, does not exist for the Indian of Central Brazil. Animals, men, inanimate objects have almost always been originally conceived as sustaining with one another relations of the most perfect identity.
So far indeed from man having exercised the classificatory function spontaneously and by a sort of natural necessity, at first the most indispensable conditions for its exercise were wanting. Let us analyse the idea of classification. A class is a group of objects. But objects do not present themselves to observation ready grouped. It is true we can perceive more or less vaguely their resemblances. But the sole fact of these resemblances does not suffice to explain how we are led to assemble the objects which resemble one another thus, to muster them in a sort of ideal milieu shut up within determinate limits which we call a genus or a species. Besides, to classify is not merely to constitute groups ; it is to dispose these groups according to very special