transition in the existing hare, which while at rest supports itself on the whole hind foot after the manner of a plantigrade animal, but when running does so upon the ends of its toes, after the manner of a digitigrade animal. It is of importance for us to note that this transition from the original plantigrade to the more recent digitigrade type, has been carried out on two slightly different plans in two different lines of mammalian descent. The hoofed mammals--which are all digitigrade--are sub-classified as artiodactyls and perissodactyls, i. e. even-toed and odd-toed. Now, whether an animal has an even or an odd number of toes may seem a curiously artificial distinction on which to found so important a classification of the mammalian group. But if we look at the matter from a less empirical and more intelligent point of view, we shall see that the alternative of having an even or an odd number of toes carries with it alternative consequences of a practically important kind to any animal of the digitigrade type. For suppose an aboriginal five-toed animal, walking on the ends of its five toes, to be called upon to resign some of his toes. If he is left with an even number, it must be two or four; and in either case the animal would gain the firmest support by so disposing his toes as to admit of the axis of his foot passing between an equal number of them--whether it be one or two toes on each side. On the other hand, if our early mammal were called upon to retain an odd number of toes, he would gain best support by adjusting matters so that the axis of his foot should be coincident with his middle toe, whether this were his only toe, or whether he had one on either side of it. This consideration shows that the classification into even-toed and odd-toed is not so artificial as it no doubt at first sight appears. Let us, then, consider the stages in the evolution of both these types of feet. Going back to the reptile _Chelydra_, it will be observed that the axis of the foot passes down the middle toe, which is therefore supported by two toes on either side (Fig. 78). It may also be noticed that the wrist or ankle bones do not interlock, either with one another or with the bones of the hand or foot below them. This, of course, would give a weak foot, suited to slow progression over marshy ground--which, as we have seen, was no doubt the origin of the mammalian plantigrade foot. Here, for instance, to all intents and purposes, is a similar type of foot, which belonged to a very early mammal, antecedent to the elephant series, the horse series, the rhinoceros, the hog, and, in short, all the known hoofed mammalia (Fig. 80). It was presumably an inhabitant of swampy ground, slow in its movements, and low in its intelligence. [Illustration: FIG. 80.--Fossil skeleton of _Phenacodus primavus_. (After Cope.)]