I do not think that this is the most fruitful way of stating the problem. I believe it to be more important to ask in what other respects lovers of dogs and of cats differ from one another. The habit, where one difference has been detected, of seeking for the associated differences, will prove extremely useful not only to pure morphology and to the science of character, but ultimately to physiognomy, the meeting-point of the two sciences. Aristotle pointed out long ago that many characteristics of animals do not vary independently of each other. Later on Cuvier, in particular, but also Geoffrey St. Hilaire and Darwin made a special study of these “correlations.” Occasionally the association of the characters is easy to understand on obvious utilitarian principles; where for instance the alimentary canal is adapted to the digestion of flesh, the jaws and body must be adapted for the capture of the prey. But association such as that between ruminant stomachs and the presence of cloven hoofs and of horns in the male, or of immunity to certain poisons with particular colouring of the hair, or among domestic pigeons of short bills with small feet, of long bills with large feet, or in cats of deafness with white fur and blue eyes--such are extremely difficult to refer to a single purpose. I do not in the least mean to assert that science must be content with no more than the mere discovery of correlations. Such a position would be little better than that of a person who was satisfied by finding out that the placing of a penny in the slot of a particular automatic machine always was followed by the release of a box of matches. It would be making resignation the leading principle of metaphysics. We shall get a good deal further by such correlations, as, for instance, that of long hair and normal ovaries; but these are within the sphere of physiology, not of morphology. Probably the goal of an ideal morphology could be reached best not by deductions from an attempted synthesis of observations on all the animals that creep on the land or swim in the sea (in the fashion of collectors of postage stamps), but by a complete study of a few organisms. Cuvier by a kind of guess-work used to reconstruct an entire animal from a single bone: full knowledge would enable us to do this in a complete, definite, qualitative and quantitative fashion. When such a knowledge has been attained, each single character will at once define and limit for us the possibilities of the other characters. Such a true and logical extension of the principle of correlation in morphology is really an application of the theory of functions to the living world. It would not exclude the study of causation, but limit it to its proper sphere. No doubt the “causes” of the correlations of organisms must be sought for in the idioplasm.