The science of Comparative Philology is, however, in too unripe a condition to allow us to speak with dogmatic assurance with regard to its inferences; even those which seem fundamental have been, and may again, be called in question. It is right here, therefore, to remind the reader that it is quite upon the cards that further research may end by upsetting the generally accepted theory of the growth of inflexions in language. Even now there is a school of philologists and anthropologists that denies the premise upon which this theory rests--the _radical_ origin of all language. This school maintains that, instead of speech beginning in monosyllabic root-sounds, as is generally supposed, it begins in extremely elaborate and complicated sounds which are in fact nothing else than sentences; that it is only by the wear and tear of use that the sentence has got split up into its component sounds, which have then taken the character of monosyllabic roots. This theory was first set on foot by a writer (Waitz) who is an anthropologist rather than a student of language, and it might be distinguished as the anthropological theory of the origin of speech. We have no space here for a full discussion of its merits. It will be enough to indicate some _à priori_ arguments in its favour. 1. It would make the language of primitive man analogous to a state of things which many people think they have discovered as typical of the most primitive savages--namely, a state of society which, in its customs, marriage laws, etc., differs from modern society in being not more simple, but infinitely more complex. 2. This supposed original expressive sentence and its subsequent analysis would have considerable analogy to what we ourselves have just seen is the history of writing, which begins with a more or less elaborate picture; then the parts of the picture are split up, and by the wear and tear of frequent use these parts are added together in separate items to form picture-_writing_, which is quite a different thing from picturing, and which is the immediate parent of writing as we know it. An analogy of this kind cannot be without weight. On the other hand, it must be pointed out that the strongest arguments in favour of this view are the _à priori_ arguments. True, we do not know enough of the languages of the world to speak with dogmatic assurance. But the history of all the languages which have been closely studied points away from the anthropological theory.