It is easy to say that all this is modern. In one sense no doubt it is. The Vedic literature, the most ancient of the whole Aryan race, presupposes a succession of intellectual strata which no chronology can measure. The language of the Veda is a work of art which it must have taken generations to build up. But is it reasonable to expect anything less modern in the history of the human race? And is there not a continuity in language and thought which allows us to see even in these literary remains, call them as modern as you like, something of the first dawn of human life. French is a very modern language, but in _chien_ we still hear the Sanskrit _ṥvan_; in _journal_ we recognise the old Vedic deity _Dyaus_. In the same way we can go back from what is common to Sanskrit, Greek, and Latin, to what was the common language of the Aryans before they broke up in different nationalities. In that common Aryan vocabulary, again, we can distinguish between what is radical and primitive and what is formal and secondary. Thus we may go back beyond all so-called historical limits to a stage of primitive thought, represented by a small number of radical concepts, and a still smaller number of formal elements. And is not that enough? Is it not more historical and more trustworthy, at all events, than all _à priori_ speculations? and have we not at least a right to demand this from our _à priori_ friends, that, in running their tunnel from the other end, they should take care that when it emerges into the daylight of history it should meet the tunnel which comparative philology, mythology, and theology have carefully dug out on the opposite side through the solid rock of facts? It will never do for _à priori_ theories to run counter to _à posteriori_ facts. It is a fact, for instance, proved by historical evidence, that fetichism represents a secondary stage in the growth of religion, and that it presupposes an earlier stage, in which the name and the concept of something divine, the predicate of every fetich, was formed. It would be fatal, therefore, to any system of _à priori_ reasoning if it placed fetichism before that phase in the development of human thought which is represented by the first formation of divine concepts. It would be a real hysteron-proteron. Again, it is a fact, proved by historical evidence, that all the words of the Aryan languages are derived from definite roots, expressive of definite concepts. It would therefore be fatal, again, to any system of _à priori_ reasoning if it attempted to derive words direct from more or less inarticulate cries or imitations of cries, and not from that small number of roots which has been proved to supply all that is really wanted in explanation of all the facts of Aryan speech.