The second phylum of the family was represented in the Uinta and Bridger stages by a group of small species, which were survivors of still more ancient and primitive progenitors of the family. In the typical genus, †Limnocyon, the dental formula was the same as in †Oxyæna: i 3/3, c 1/1, p 4/4, m 2/2, but the first upper molar had its two external cusps well separated and a much lower posterior cutting ridge, while the inner cusp was much larger. The second upper molar, though transversely placed, had all the elements of the primitive tritubercular tooth, the pattern from which all the varied types of †creodont upper molars were derived by the addition or suppression of parts. The two lower molars were very primitive, having a high anterior triangle of three cusps, forming an imperfect shearing blade, and a low heel. This dentition was on nearly the same plan as that of the small, contemporary †hyænodonts, but the emphasis of development, so to speak, was differently placed. In the †hyænodonts there were three pairs of sectorials and the best-developed pair was made up of the second upper and third lower molar; while in †Limnocyon the third molar was lost, and there were but two pairs of sectorials, of which the largest pair was the first upper and second lower molar, as was also true of †Oxyæna and †Patriofelis.
The skull of †Limnocyon had a much longer facial region, and more elongate and slender jaws than in the last-named genera, and the feet must have been quite different, with less spreading digits. †Limnocyon thus tends to indicate a common origin for the †oxyænids and †hyænodonts, though these common ancestors are still unknown.
A very interesting genus of this series, †Machairoides, of the Bridger, shows another imitation of the cats, the flanges of the lower jaw indicating sabre-like upper canines.
Another genus, †Palæonictis, of the Wasatch, found also in France, is sometimes referred to the †Oxyænidæ and sometimes made the type of a distinct family, but is too incompletely known for final reference. It had the same number of teeth as †Oxyæna, but the principal pair of carnassials was the fourth upper premolar and first lower molar, as in the Fissipedia, the first upper and second lower molar forming the subsidiary pair. The first upper molar was hardly sectorial at all; its two outer cusps were long, sharp-pointed cones, and the posterior cutting ridge was a mere tubercle. The skull had a short, cat-like face. The genus left no successors.
This concludes the long story of the Carnivora, so far as it has been recovered from the rocks. Incomplete as it is, and full of unsolved problems, it yet enables us to follow, somewhat vaguely, but with a general kind of accuracy, the development of the various modifications which characterized the different families and genera of the group.
The more ancient and primitive suborder, the †Creodonta, made its first recorded appearance in the lower Paleocene and was, no doubt, derived from Mesozoic ancestors, which cannot yet be distinguished among the very imperfectly understood mammals of that era. In the upper Paleocene, if not before, the †creodonts had spread over the northern hemisphere and had begun to diverge into a number of families, which continued to diverge more and more widely throughout the Eocene epoch, as they became more specialized and adapted to different habits of life. From the most primitive group, represented more or less accurately by the †Oxyclænidæ, may be traced the several lines of diverging adaptations incorporated in the various families, some of which had become distinctly recognizable in the lower Paleocene, others in the upper, while all were in existence in the lower Eocene. In one series, the †Mesonychidæ, the upper teeth underwent comparatively little change, while the lower ones lost the inner cusps, but no carnassials were formed. The face and jaws were elongated and the limbs and feet became adapted to cursorial habits, and the more advanced genera had four-toed, completely digitigrade feet, with blunt, almost hoof-like claws. A second series, the †Arctocyonidæ, likewise failed to develop sectorial teeth, the molars becoming quadritubercular, with many accessory tubercles, and assuming a bear-like or pig-like pattern, while the premolars were reduced in size. The pentadactyl feet had sharp claws.
In the †Oxyænidæ two pairs of carnassial teeth were formed, of which the larger and more effective pair were the first upper and second lower molar, the smaller pair the fourth upper premolar and first lower molar. The teeth were diminished in number, first by the loss of the last molar, then the suppression of the first premolar and finally by that of the third incisor and second upper molar; the remaining teeth were enlarged. The upper carnassial molar (the first) was formed by the approximation and partial fusion of the two external cusps and the addition of a trenchant ridge behind these, and by the reduction and eventual loss of the internal cusp, thus becoming more exclusively shearing in function. The second lower molar also lost the inner cusp and the heel, becoming remarkably cat-like in form; the first was similar, but less simplified. The face and jaws were greatly shortened, which, with the widely expanded zygomatic arches, gave the head a very cat-like appearance. The body and tail were long, the limbs short and thick, and the feet had spreading toes and blunt claws. Save for a notable increase in size and muscular power, the †oxyænids showed but little change within the family.
The †Hyænodontidæ differed from the †oxyænids in the retention of all or nearly all the teeth and in having three pairs of sectorials, of which the largest pair was the second upper and third lower molar, but resembled them in the mode of forming these sectorials and in the cat-like form of the inferior ones. Although the actual line of descent was not through these genera, the series, †Sinopa—†Tritemnodon—†Pterodon—†Hyænodon, extending from the lower Eocene into the Oligocene, displays perfectly the successive steps in the transformation of the teeth. The skull underwent a corresponding series of changes, ending in long-faced, long-jawed, wolf-like forms, with larger brain-case than in any other †creodonts. The elongated form of body was retained, but the tail was reduced to moderate proportions. The limbs and feet did not change greatly, except in size and in the greater bluntness of the claws.
The †Miacidæ, if not actually referable to the Fissipedia, at least anticipated them in the mode of carnassial development. The upper molars changed very little from the primitive tritubercular plan, but the fourth upper premolar was enlarged and acquired a trenchant ridge behind the original single outer cusp. The lower molars were at first all alike, except in size, the first being the largest; they had the primitive pattern common to the earlier members of nearly all the †creodont families, of an elevated anterior triangle of three subequal cusps and low, basin-like heel. The first molar grew larger in the successive genera and, by the enlargement of the two external cusps of the primitive triangle and reduction of the inner one, gradually became an efficient sectorial, the fourth upper premolar keeping pace with it. In proportion as the first lower molar was elaborated, the second and third were reduced in size and the anterior triangle was lowered to the level of the heel, these teeth thus becoming tubercular. All the †Miacidæ were small animals, none attaining the stature of a fox, though some had heads as large. From this family, as was pointed out above, probably arose all of the Fissipedia, the history of which it is needless to repeat.