Before we can decide with confidence in any case, a great mass of evidence must be brought into court. So long as we remained on Aryan ground, all went smoothly enough, because all the external evidence was in our favour. We knew at the outset, that the Aryans inherit a common language and a common civilization, and therefore we found no difficulty in accepting the conclusion that they have inherited, among other things, a common stock of legends. In the barbaric world it is quite otherwise. Philology does not pronounce in favour of a common origin for all barbaric culture, such as it is. The notion of a single primitive language, standing in the same relation to all existing dialects as the relation of old Aryan to Latin and English, or that of old Semitic to Hebrew and Arabic, was a notion suited only to the infancy of linguistic science. As the case now stands, it is certain that all the languages actually existing cannot be referred to a common ancestor, and it is altogether probable that there never was any such common ancestor. I am not now referring to the question of the unity of the human race. That question lies entirely outside the sphere of philology. The science of language has nothing to do with skulls or complexions, and no comparison of words can tell us whether the black men are brethren of the white men, or whether yellow and red men have a common pedigree: these questions belong to comparative physiology. But the science of language can and does tell us that a certain amount of civilization is requisite for the production of a language sufficiently durable and wide-spread to give birth to numerous mutually resembling offspring Barbaric languages are neither widespread nor durable. Among savages each little group of families has its own dialect, and coins its own expressions at pleasure; and in the course of two or three generations a dialect gets so strangely altered as virtually to lose its identity. Even numerals and personal pronouns, which the Aryan has preserved for fifty centuries, get lost every few years in Polynesia. Since the time of Captain Cook the Tahitian language has thrown away five out of its ten simple numerals, and replaced them by brand-new ones;and on the Amazon you may acquire a fluent command of some Indian dialect, and then, coming back after twenty years, find yourself worse off than Rip Van Winkle, and your learning all antiquated and useless. How absurd, therefore, to suppose that primeval savages originated a language which has held its own like the old Aryan and become the prolific mother of the three or four thousand dialects now in existence! Before a durable language can arise, there must be an aggregation of numerous tribes into a people, so that there may be need of communication on a large scale, and so that tradition may be strengthened. Wherever mankind have associated in nations, permanent languages have arisen, and their derivative dialects bear the conspicuous marks of kinship; but where mankind have remained in their primitive savage isolation, their languages have remained sporadic and transitory, incapable of organic development, and showing no traces of a kinship which never existed.
The bearing of these considerations upon the origin and diffusion of barbaric myths is obvious. The development of a common stock of legends is, of course, impossible, save where there is a common language; and thus philology pronounces against the kinship of barbaric myths with each other and with similar myths of the Aryan and Semitic worlds. Similar stories told in Greece and Norway are likely to have a common pedigree, because the persons who have preserved them in recollection speak a common language and have inherited the same civilization. But similar stories told in Labrador and South Africa are not likely to be genealogically related, because it is altogether probable that the Esquimaux and the Zulu had acquired their present race characteristics before either of them possessed a language or a culture sufficient for the production of myths. According to the nature and extent of the similarity, it must be decided whether such stories have been carried about from one part of the world to another, or have been independently originated in many different places.
Here the methods of philology suggest a rule which will often be found useful. In comparing, the vocabularies of different languages, those words which directly imitate natural sounds--such as whiz, crash, crackle--are not admitted as evidence of kinship between the languages in which they occur.
