Up to the beginning of the Miocene, Africa was, as we saw, cut off from Europe and Asia by the sea which stretched from Spain to India. Then the land rose, and the elephant passed by the new tracts into the north. Its next representative, Tetrabelodon, is found in Asia and Europe, as well as North Africa. The frame is as large as that of a medium-sized elephant, and the increase of the air-cells at the back of the skull shows that an increased weight has to be sustained by the muscles of the neck. The nostrils are shifted further back. The tusks are from twenty to thirty inches long, and round, and only differ from those of the elephant in curving slightly downward, The chin projects as far as the tusks. The neck is shorter and thicker, and, as the animal increases in height, we can understand that the long snout--possibly prehensile at its lower end--is necessary for the animal to reach the ground. But the snout still lies on the projecting lower jaw, and is not a trunk. Passing over the many collateral branches, which diverge in various directions, we next kind that the chin is shortening (in Tetrabelodon longirostris), and, through a long series of discovered intermediate forms, we trace the evolution of the elephant from the mastodon. The long supporting skin disappears, and the enormous snout becomes a flexible trunk. Southern Asia seems to have been the province of this final transformation, and we have remains of some of these primitive elephants with tusks nine and a half feet long. A later species, which wandered over Central and Southern Europe before the close of the Tertiary, stood fifteen feet high at the shoulder, while the mammoth, which superseded it in the days of early man, had at times tusks more than ten feet in length. It is interesting to reflect that this light on the evolution of one of our most specialised mammals is due to the chance opening of the soil in an obscure African region. It suggests to us that as geological exploration is extended, many similar discoveries may be made. The slenderness of the geological record is a defect that the future may considerably modify.