We need not follow Plato’s investigations into the meaning of knowledge and the causes of illusion any further; especially as they do not lead, in this instance, to any positive conclusion. The general tendency is to seek for truth within rather than without; and to connect error partly with the disturbing influence of sense-impressions on the higher mental faculties, partly with the inherent confusion and instability of the phenomena whence those impressions are derived. Our principal concern here is to note the expansive power of generalisation which was carrying philosophy back again from man to Nature—the deep-seated contempt of Plato for public opinion—and the incipient differentiation of demonstrated from empirical truth. A somewhat similar vein of reflection is worked out in the _Cratylus_, a Dialogue presenting some important points of contact with the _Theaetêtus_, and probably belonging to the same period. There is the same constant reference to Heracleitus, whose philosophy is here also treated as in great measure, but not entirely, true; and the opposing system of Parmenides is again mentioned, though much more briefly, as a valuable set-off against its extravagances. The _Cratylus_ deals exclusively with language, just as the _Theaetêtus_ had dealt with sensation and mental imagery, but in such a playful and ironical tone that its speculative importance is likely to be overlooked. Some of the Greek philosophers seem to have thought that the study of things might advantageously be replaced by the study of words, which were supposed to have a natural and necessary connexion with their accepted meanings. This view was particularly favoured by the Heracleiteans, who found, or fancied that they found, a confirmation of their master’s teaching in etymology. Plato professes to adopt the theory in question, and supports it with a number of derivations which to us seem ludicrously absurd, but which may possibly have been transcribed from the pages of contemporary philologists. At last, however, he turns round and shows that other verbal arguments, equally good, might be adduced on behalf of Parmenides. But the most valuable part of the discussion is a protest against the whole theory that things can be studied through their names. Plato justly observes that an image, to be perfect, should not reproduce its original, but only certain aspects of it; that the framers of language were not infallible; and that we are just as competent to discover the nature of things as they could be. One can imagine the delight with which he would have welcomed the modern discovery that sensations, too, are a language; and that the associated groups into which they most readily gather are determined less by the necessary connexions of things in themselves than by the exigencies of self-preservation and reproduction in sentient beings.