Now, the higher animals unquestionably do understand the meanings of words; idiots too low in the scale themselves to speak are in the same position; and infants learn the signification of many articulate sounds long before they begin themselves to utter them.[78] In all these cases it is of course important to distinguish between the understanding of words and the understanding of tones; for, as already observed, both in the animal kingdom and in the growing child it is evident that the former represents a much higher grade of mental evolution than does the latter—a fact so obvious to common observation that I need not wait to give illustrations. But although the fact is obvious, it is no easy matter to distinguish in particular cases whether the understanding is due to an appreciation of words, to that of tones, or to both combined. We may be sure, however, that words are never understood unless tones are likewise so, and that understanding of words may be assisted by understanding of the tones in which they are uttered. Therefore, the only method of ascertaining where words as such are first understood, is to find where they are first understood irrespective of the tones in which they are uttered. This criterion—so far, at least, as my evidence goes—excludes all cases of animals obeying commands, answering to their names, &c., with the exception of the higher mammalia. That is to say, while the understanding of certain tones of the human voice extends at least through the entire vertebrated series,[79] and occurs in infants only a few weeks old; the understanding of words without the assistance of tones appears to occur only in a few of the higher mammalia, and first dawns in the growing child during the second year.[80] The fact that the more intelligent Mammalia are able to understand words irrespective of tones is, as I have said, important; and therefore I shall devote a few sentences to prove it. My friend Professor Gerald Yeo had a terrier, which was taught to keep a morsel of food on its snout till it received the verbal signal “Paid for;” and it was of no consequence in what tones these words were uttered. For even if they were introduced in an ordinary stream of conversation, the dog distinguished them, and immediately tossed the food into his mouth. Seeing this, I thought it worth while to try whether the animal would be able to distinguish the words “Paid for” from others presenting a close similarity of sound; and, therefore, while he was expecting the signal, I said “Pinafore;” the dog gave a start, and very nearly threw the food off his nose; but immediately arrested the movement, evidently perceiving his mistake. This experiment was repeated many times with these two closely similar verbal sounds, and always with the same result: the dog clearly distinguished between them. I have more recently repeated this experiment on another terrier, which had been taught the same trick, and obtained exactly the same results.