I do not propose here to bring forward a number of new observations on the highly intelligent actions which animals are capable of performing. Mr. Romanes has given us a most valuable collection of anecdotes on the subject in his volume on "Animal Intelligence." It is more to my purpose to discuss some of the more remarkable of these, and endeavour to get at the back of them, so as to estimate what are the mental processes involved. In doing so, the principle I adopt is to assume that the inferences are perceptual, unless there seem to be well-observed facts which necessitate the analysis of the phenomena, the formation of isolates, and therefore the employment of reason (_as I have above defined it_). In doing this, I shall _seem_ to differ very widely from Mr. Romanes and other interpreters of animal habits and intelligence. But I believe that the divergence is less wide than it seems. I believe that it is largely, but I fear not entirely, a question of the terms we employ. Why, then, rediscuss the question under these new terms? Because I believe that such rediscussion may place the matter in a fresh and, perhaps, clearer light. The question of the relation of animal intelligence to human reason is one upon which there is a good deal of disagreement, and one that has been discussed and rediscussed. I seek to put it in a somewhat new light. I have endeavoured to define carefully and accurately the terms I use, and the sense in which I use them. I have coined for my own purposes unfamiliar terms such as "construct," "isolate," and "predominant," that I might thereby be enabled to avoid the use of terms which, from the different senses in which they are employed by different writers, have become invested with a certain ambiguity. I trust, therefore, that even those with whom I seem most to disagree will allow that my aim has not been mere disputation, but scientific accuracy and precision in a difficult subject where these qualities are of essential importance.