As is the case with other groups, the Amblypoda commenced existence as a sub-order with relatively small forms such as _Pantolambda_, the most ancient type known, which is in many respects a transition between the later forms and other groups of mammals such as the Creodonta.[127] The race culminated and ended in the giant _Dinoceras_ and _Coryphodon_, and spread into the Old World. In spite of their smooth and diminutive brain, these mammals were able to hold their own and to multiply into many species and genera; in this they were perhaps aided by their formidable tusks and by the horns which many of them possessed. The teeth seem to imply an omnivorous diet, which was quite possibly an additional advantage in the struggle for existence. It does not seem to be necessary to divide off the Dinoceratidae into a sub-order equivalent to the Coryphodontidae as was done {207} by Professor Marsh; the numerous points in common possessed by the members of both families forbid their separation more widely than as families. The earliest types of Amblypoda belong to the genus _Pantolambda_, of which the species _P. bathmodon_ was about four feet in length. As restored it seems to have had proportionately short fore- and hind-limbs, and it had a long tail. It was apparently plantigrade, and would have had not a little likeness to a carnivorous type. The skull has no air cavities, such as are developed in the later types from the Lower Eocene, e.g. _Coryphodon_; _Pantolambda_ is from the basal Eocene. The frontal bones show no trace of the horns that are developed in subsequent forms; the nasals are comparatively long; the zygomatic arch is slender. The molar teeth are in the primitive form of trituberculy, and the premolars, as is so often the case with primitive animals, are unlike the molars in form, being less markedly selenodont. As to the vertebral column, the dorsal vertebrae appear to have had short spines, which argues, as it does also in the case of the larger and heavier _Coryphodon_, a feebleness in the development of ligaments and muscles supporting and moving the head. The scapula seems to have the same peculiar leaf-like form that it has in the later _Coryphodon_.[128] This primitive type shows an entepicondylar foramen in the humerus. It is interesting to observe that the posterior border of the ulna is convex, as in the Creodonts, and in the early Condylarthrous form _Euprotogonia_. In the subsequently-developed Amblypoda, as in the later Condylarthra, that bone acquires a concave outer border. In the carpus the os centrale is distinct. In the femur the third trochanter is well formed; it gradually dies out in later Amblypoda. The fibula articulates with the calcaneum. This species, according to Osborn, "typifies the hypothetical Protungulate, being more primitive than either _Euprotogonia_ or _Phenacodus_."[129] [Illustration] FIG. 115.--Skeleton of _Coryphodon radians_. x 1/10. (After Osborn.)