Such, then, is one way in which it appears to me quite conceivable that the faculty of articulate sign-making might have taken the first step towards the formation of speech. But, not to go further than this first step, I can see another possibility as to the precise method of attainment, and one which I think is still more probable. It is the opinion of some authorities in anthropology that speech was probably, and comparatively speaking, late in making its appearance; so that our ancestors in whom it did first appear were already more human than simian, and as such deserving of the name _Homo alalus_.[315] Now, if this were the case, the course of our hypothetical history would be even more easy to imagine than it was under the supposition previously considered. For, under the present supposition, we start with an already man-like creature, erect in attitude, much more intelligent than any other animal, shaping flints to serve as tools and weapons, living in tribes or societies, and able in no small degree to communicate the logic of his recepts by means of gesture-signs, facial expressions, and vocal tones. Clearly, from such an origin, the subsequent evolution of sign-making in the direction of articulate sounds would be an even more easy matter to imagine than under the previous hypothesis. For, let us try to imagine a community of _Homo alalus_, considerably more intelligent than the existing anthropoid apes, although still considerably below the intellectual level of existing savages. It is certain that in such a community natural signs of voice, gesture, and grimace would be in vogue to a greater or less extent.[316] As their numbers increased (and, consequently, as natural selection laid a greater and greater premium on intelligent co-operation, as in the case of social insects),[317] such signs would require to become more and more conventional, or acquire more and more the character of sentence-words and denotative signs.[318] Now, where the signs were vocal, the only ways in which they could be developed so as to meet this need would be, (1) conventional modulations of intensity, (2) of pitch, and (3) of time-intervals. But clearly, neither modulations of intensity nor of pitch could carry improvement very far, seeing that the human voice does not admit of any great range of either. Consequently, if any improvement at all were to be effected—and it was bound to be effected, if possible, by natural selection,—it could only be so in the direction of modulating time-intervals between vocal sounds. Now, such a modulation of time-intervals is the beginning of _articulation_.