sure indication of race. Since the dawn of time, those among whom individuals were born, of whom they learned how to express their mental acts, have been usually of their own blood. Nor do these difficulties place linguistic evidence at any marked disadvantage as compared with physical. They are, to no small extent, merely the effect, on the side of language, of the grand fact which comes in constantly to interfere with ethnological investigations of every kind: namely, that human races do not maintain themselves in purity, that men of different descent are all the time mingling, mixing their blood, and crossing all their race-characteristics. Fusion and replacement of languages are impossible, except when men of different native speech are brought together as members of the same community, so that there takes place more or less of an accompanying fusion of races also; and then the resulting language stands at least a chance of being a more faithful and intelligible witness of the mixture than the resulting physical type. That the modern French people, for example, is made up of a congeries of Celtic, Germanic, and Italican elements is to a certain extent—although only the aid of recorded history enables us fully to interpret the evidences—testified by the considerable body of Celtic and Germanic words mixed with the Latin elements of the French language; but no physicist could ever have derived the same conclusion from a study of the French type of structure. The physicists claim that there may be a considerable infusion of the blood of one race into that of another, without perceptible modification of the latter's race-type; the intruded element, if not continuously supplied afresh, is overwhelmed and assimilated by the other and predominant one, and disappears: that is to say, as we may interpret the claim, its peculiarities are so diluted by constant remixture that they become at last inappreciable. In any such case, then, traces discoverable in the language may point out what there is no other means of ascertaining. It is true that, on the other hand, the spread and propagation of a language may greatly exceed that of the race to which it originally belonged, and that the weaker numerical element in a composite community may be the one whose
