tone and spirit requisite to give the ballad full effect. But if, as I believe, mingling of racial elements is an important factor in the evolution of our ballads, unity of language was necessary to their inception and production. The Welsh Marches, like the Border, were a wild mountain tract, full of feuds and forays, offering the same raw material, the same favourable conditions, for working it up into ballad form as the Northern Border. One of the conflicting races possessed a rich traditional romance, a highly organised literary class. Nothing comparable to the Northern ballad poetry was the result. Why? I can only conjecture that the explanation lies in the absence of a language common to friend and foe, to victim and oppressor, to Romeo and Juliet. The difference of language served as a bar to the communication of artistic conventions and formulas, the sagas remained local and provincial, and, on the English side of the Marches at least, died out because there were no folk-singers qualified to present them in a form truly popular and yet artistic because exactly suited to their nature.
I have now briefly indicated the problems that confront the discriminator of racial elements in our folklore, in so far as these elements are known to us historically, and belong, linguistically at all events, to the Aryan-speaking group. I may be held to have over-emphasised the complexities, to have minimised the possibilities of successful solution. I do not think so. In any case it will be agreed that, great as are the difficulties of discrimination when dealing with Aryan-speaking communities, they increase tenfold when we essay to determine the share of our pre- Aryan forefathers. In the former case we are guided and controlled by recorded history, by recorded literature; in the latter we are working almost entirely in the dark, with, at the best, the dubious, scanty, and conflicting evidence provided by the archaeologists. The very method by which such an inquiry must be prosecuted
