As we pass from older forms to their more recent representatives there is as a rule a progressive development of the form of the teeth. This is especially marked among the Ungulata. The extremely complicated type of tooth found in such a form as the existing Horse can be traced back through a series of stages to a tooth in which the crown is marked by a few separated tubercles or cusps. Arrived at this point, the differences between the teeth of ancestral Horses and ancestral Rhinoceroses and Tapirs are hard to distinguish with accuracy; and the same difficulty is experienced in attempting to give a definition of other large orders by the characters of the teeth, such as will apply to the Eocene or {56} even earlier representatives of these families. Fig. 36 (p. 51) illustrating a series of mammalian teeth will illustrate the above remarks. That there is such a convergence in tooth structure shows that it is, theoretically at least, possible to determine the ancestral form of the mammalian tooth. Practically, however, the difficulties which beset such theorising are great; that there are such divergent and such strongly-held antithetical views is sufficient proof of this. Two main views hold the field: one, which has found most favour in America, and is due chiefly to the labours and persuasiveness of Professors Cope, Scott, Osborn, and others, is known as "trituberculy."[30] The alternative view, as urged by Forsyth Major, Woodward, and Goodrich, attempts to show that the dentition of the original mammal included grinding teeth which were multicuspidate or "multitubercular." There is much to be said for both views, and something to be said against both. [Illustration] FIG. 38.--Molar teeth of A, _Phenacodus_, and B, the Creodont _Palaeonictis_. _End_, endoconid; _hld_, hypoconulid; _hyd_, hypoconid; _med_, metaconid; _prd_, protoconid. (After Osborn and Wortman.)