names of Celtic deities, too, fortunately bear in their very forms unmistakable evidence as to their original character, so that we have an insight not only into the religious conceptions of the later and more developed stages of civilisation, but also into the earlier ideas from which they sprang. It is possible, also, to see in this way how variation in the degree of civilisation both locally and temporally is reflected in the forms of religion. In dealing with the religion of the Celtic world, it is of prime importance to bear in mind at the outset that Celtic civilisation was very far from being homogeneous in character, and that we must expect to see this absence of homogeneity reflected also in the religious evolution.
Unfortunately, the inscriptional evidence that has come down to us is associated almost entirely with districts that were highly Romanised. This, however, is not such a serious drawback as might at first have appeared, since we are thus enabled to see what districts during the Empire were those in which the conceptions of certain deities had been so developed as to make it possible to identify them with the gods of Rome. These, too, were the chief districts of trade and commerce among the Celts, and, though the earlier and the later trade-routes through Gaul, for example, were not the same, yet the routes which were important in Roman times had already risen into some prominence in the earlier period. Archæology cannot render greater service to the study of Celtic religion than by mapping out the distribution of Celtic civilisation. The religion of the rich is never quite the same as that of the poor: that of the farmer is never absolutely identical with that of the sailor: that of the townsman always differs in some degree from that of the countryman: the merchant and the craftsman do not usually worship quite the same gods as the soldier. The gods of fashionable health-resorts vary no less than their worshippers, and the study of the forms and degrees which civilisation assumed in Celtic lands may help to bring these variations more and more home to us.
In the present paper the larger problem of the Celtic
