The grinding teeth underwent much more radical and striking changes. At first (_†Mœritherium_) they were small, very low-crowned and of simple pig-like or quadritubercular pattern, making two interrupted cross-crests; all were in use simultaneously and the succession of milk-teeth and premolars was by vertical replacement, as in normal mammals generally. In _†Palæomastodon_ there were three pairs of tubercles on the molars and in _†Gomphotherium_ these coalesced into ridges, but in all the †mastodons there was more or less distinctness of the conical tubercles. In one or more phyla the three-ridged plan persisted for a long time, one such phylum terminating in the Pleistocene _†Mastodon americanus_. In the other series the number of ridges increased, first to four, then to five, six and more (_†Stegodon_); the crowns of the teeth became much larger and higher, and the ridges, as their number increased, became much thinner, and the valleys between them were filled with cement, and finally, in the true elephants, with their fully hypsodont, many-crested teeth, were thickly covered all over with cement. The vertical succession of milk-teeth and premolars was retained in _†Gomphotherium_, at least in some species, but the large molars, which could not find room to be exposed while the premolars were in place, came in successively from behind. This horizontal mode of succession is the only one to be seen in the true elephants, in which but one tooth, or parts of two, on each side of each jaw are in simultaneous use and the premolars have entirely disappeared, but the milk-teeth are retained. The changes in the skull, which amounted to a reconstruction, were very largely conditioned by the great increase in the length and consequent weight of the tusks, in the size of the grinding teeth and the development of the proboscis. In the earliest known type (_†Mœritherium_) the skull had little about it that would, at first sight, suggest proboscidean affinities; it was long and narrow, with sagittal crest and occiput of normal type, very long cranial and very short facial region. The nasal opening was directed forward and the nasal canal was relatively long and horizontal in direction, but the nasal bones were already much shortened, indicating that the proboscis was probably in an incipient stage. The symphysis of the lower jaw was procumbent and somewhat elongated, but to only a comparatively slight degree. While the skull of _†Mœritherium_ was not obviously proboscidean, that of its successor, _†Palæomastodon_, was unmistakably so, yet retained several primitive features, which were lost in all of the subsequent genera, such as the sagittal crest, the relatively low cranium and moderate thickening of the cranial bones, the forward direction of the nasal opening, etc.; the symphysis of the lower jaw was very greatly prolonged.