The corresponding condition of Semitic speech, its foundation of triliteral roots, is to no small extent restorable; but we have seen that these roots are themselves the products of a strange and highly perplexing development, beneath which their actual origin is not yet discernible. Among the different great branches of the Scythian family, the recognizable radical coincidences are hardly sufficient, if they are sufficient, to establish their unity as proceeding from the same stock: a reliable basis for comparison with other families is certainly not furnished us here. Nor was the Scythian the only family in establishing whose unity we were obliged to add the evidence of morphological structure to that of material correspondences: there were at least two, the monosyllabic in south-eastern Asia and the American, which were founded almost solely on accordance of type. And the former of them is a striking illustration of the power of phonetic corruption to alter and disguise the bare roots of language, without help from composition and fusion of elements. If we cannot find material correspondences enough between the pure radicals of Chinese, Siamese, and Burmese to prove these three tongues akin, but must call in, to aid the conclusion, their common characteristic of monosyllabism, what hope can we possibly entertain of proving either of them akin with Mongolian or Polynesian, for example, with which they have no morphological affinity? Who will be so sanguine as to expect to discover, amid the blind confusion of the American languages, where there are scores of groups which seem to be totally diverse in constituent material, the radical elements which have lain at the basis of their common development? Apparent resemblances among apparent roots of the different families are, indeed, to be found: but they are wholly worthless as evidences of historical connection. To the general presumption of their accidental nature is to be farther added the virtual certainty that the elements in which they appear are not ultimate roots at all, but the products of recent growth. There is nothing, it may be remarked, in the character of ultimate roots which should exempt them from the common liability to exhibit fortuitous coincidences, but rather the contrary. The system of sounds employed in the
