The skull was very like that of †Hoplophoneus, but was still longer and somewhat different in shape, owing to the higher forehead and lower occiput. The primitive features of the cranial base, such as the foramina, the imperfectly ossified tympanic bullæ, etc., were repeated in †Dinictis, but the lower jaw had much less prominent flanges for the protection of the tusks. The limbs differed considerably from those of †Hoplophoneus in being relatively longer and more slender and retaining more primitive features, such as the larger third trochanter of the femur. The five-toed feet were decidedly small and weak, and the claws, though retractile, were less so than in the other genus and were not hooded. The gait was probably plantigrade or semi-plantigrade.

The relationships of †Dinictis and †Hoplophoneus are rather puzzling; none of the known species of the former could have been ancestral to the latter, for the two genera were contemporaneous. †Dinictis was apparently the somewhat modified survivor of the ancestral stage and represented very nearly the common starting point of both the feline and †machairodont subfamilies. Dr. Matthew has propounded the bold theory that this genus was the actual ancestor of the felines, continuing the series through †Archælurus and †Nimravus of the John Day to the unmistakable felines of the middle Miocene. This view runs contrary to the supposed “law of the irreversibility of evolution,” a rule which many authorities look upon as well established. The theory postulates a different mode of development from anything that we have so far encountered in the series previously described and supposes that the upper canine first lost its original form, becoming a thin, elongate and scimitar-like tusk, while the lower canine was reduced almost to the proportions of an incisor and the lower jaw acquired a straight, flat chin and inferior flanges for the protection of the tusks. Then, after specialization had advanced so far, it was reversed and the original condition regained. This interesting hypothesis may possibly turn out to be true, though personally I cannot accept it, and, should it do so, it would necessitate a thoroughgoing revision of current opinions as to the processes of mammalian development.

The only John Day cat which was assuredly derived from †Dinictis was the large †Pogonodon, previously mentioned.

Also in the John Day stage lived †Archælurus and †Nimravus, which, as was noted above (p. 249), have been called the “false sabre-tooths,” for in them the upper canine was not much larger than the lower and the latter, though smaller than in the felines, was yet very much less reduced than in the true †machairodonts. The skull closely resembled that of †Dinictis, but the lower jaw was without flanges. The limbs were long and slender and the feet long and digitigrade. The pes had only four digits, of which the median pair was elongated and the lateral pair shortened, so as to produce considerable resemblance to the pes of the dogs, and the claws were partially retractile. The proportions of the body, limbs and feet were suggestively like those of the Cheeta, or Hunting Leopard (Cynælurus jubatus) of India, the generic name of which means “dog-cat,” and it is quite possible that the Cheeta may have been derived from some member of this “false †sabre-tooth” series, though the connecting links are unknown. These cursorial cats quite displaced the leaping †machairodonts of the †Hoplophoneus type, at least in the Oregon region at a time when, it will be remembered, that region had a remarkable variety of dogs. In other parts of the continent, of which we have no record, the true †machairodonts must have been thriving, as may be inferred from their comparative abundance in the later formations.

Concerning the habits of these cursorial cats, Professor Merriam says: “When the canines are not developed to the dagger-like form for stabbing, the premolar teeth serve a more definite purpose in the destruction of prey and would be less subject to reduction. The view suggested above finds support in that such evidence as we have indicates that during the deposition of the Middle John Day beds this region was in the main a country of open plains, offering advantages to running types of carnivores, and that during this epoch the Archælurus-Nimravus type of feline was by far the most common form [i.e. of cats].” The derivation of these cats is still obscure, but their likeness to certain forms of the European Oligocene suggests that they were immigrants.

The true cats of the subfamily Felinæ include the great variety of living forms, large and small, from the Lion and Tiger at one extreme to the Domestic Cat at the other. There is great difference among naturalists with regard to the nomenclature of the Recent cats; some make a considerable number of separate genera, while others include all the species, except the lynxes and the Cheeta, in the genus Felis. For the purposes of this book the latter practice is the more convenient and will be followed. In Felis the dental formula is: i 3/3, c 1/1, p 2-3/2, m 1/1, × 2 = 28-30; the canines are large and strong, of oval section, and the upper one is but little larger than the lower; there are two large and functional premolars in each jaw, and an additional very small one may or may not be present in the upper jaw. The upper sectorial has a large shearing blade, with well-developed anterior accessory cusp, and the inner cusp, which in †Smilodon had almost disappeared, is quite large and carried on a separate root. The lower sectorial is composed of two cusps only, all traces of the heel and of the inner cusp having disappeared. The single upper molar is very small and usually concealed by the sectorial. The skull is very short and broad, and the shortening of the jaws gives great power to the biting muscles, because of the more favourable leverage. The zygomatic arches are very stout and curve out boldly, contributing much to the rounded shape of the head; the orbits are almost encircled in bone. The large tympanic bullæ are two-chambered and there is no alisphenoid canal, but in several other respects the base of the cranium differs markedly from that of †Smilodon. The lower jaw is without flanges and there is no angle between front and sides.

The neck is short, the body long and the tail is long in most of the species, but short in the lynxes. The limbs are relatively longer and less massive than in †Smilodon, and there are five toes in the manus, four in the pes; the claws are hooded and retractile.

The western hemisphere at the present day contains none of the very large species, the Puma and Jaguar being the largest; but this was not true of the Pleistocene, where a huge cat (Felis †atrox), surpassing the Lion in size, ranged over the southern half of North America. Enormous cats also lived in the lower Pliocene and upper Miocene of the Great Plains region, but are not sufficiently well known for reference to either subfamily.

The history of the true felines has been but partially deciphered, and can, as yet, be traced back only to the middle Miocene, the genus †Pseudælurus representing the series both in Europe and North America. In this genus the dental formula was nearly the same as in Felis, but there was frequently an additional small premolar in the lower jaw and the sectorials were more primitive, the upper one having the accessory anterior cusp in a merely incipient stage and in the lower one there was a vestige of the heel. The upper canine was considerably longer than the lower, thinner and more blade-like than in Felis, which, so far as it goes, is in favour of Dr. Matthew’s theory (p. 541). What little is known of the skull and skeleton of †Pseudælurus agrees with the modern cats.

While it is not feasible to trace the series of true felines to an earlier stage than the middle Miocene, there can be no doubt that the subfamily was derived from the same stock as the †machairodonts and it is probable that the White River †Dinictis nearly represents the common starting point for both series; the resemblances between †Dinictis and such primitive dogs as †Daphœnus are suggestive of a common origin.

3. Procyonidæ. Raccoons, etc.

An almost exclusively American family of Fissipedia is that of the raccoons, which includes not only the latter (Procyon), but also the coatis (Nasua), curious animals, with long, flexible, pig-like snouts, the cacomistles (Bassariscus) and kinkajous (Potos). In addition to these American forms, there is an outlying Asiatic genus, the Panda (Ælurus) of the southeastern Himalayas, the last of a series which goes back to the European Pliocene.

The Procyonidæ are animals of small and moderate size, largely arboreal in habits and subsisting upon a mixed diet of fruit, eggs, insects and the like; the teeth are adapted to this diet and the sectorials have mostly lost their shearing form and the molars are tuberculated for crushing and grinding. The species generally have long tails, except in the raccoons proper, in which the tail is of medium length, and five-toed, plantigrade feet, with naked soles. Fossil members of this family are very rare in Tertiary formations and its history is therefore but scantily known; in the lower Pliocene have been found fragmentary remains with less specialized teeth, which appear to belong to the direct ancestor of Bassariscus. The upper Miocene genus †Leptarctus was an undoubted member of the family, and, while it would seem not to have been in the direct line of any of the modern forms, it was near to the common ancestry of the American genera, so far as the imperfect specimens enable us to judge.

By far the most primitive representative of the family yet discovered is the lower Miocene genus †Phlaocyon, which connected the Procyonidæ with the Oligocene genus of dogs, †Cynodictis (p. 529). The dentition resembled that of the latter, with several differences, which were all changes toward the Procyonidæ. All the cusps were lower and blunter than in †Cynodictis; the premolars were small, thick and closely crowded together and the upper sectorial, while still trenchant, had a postero-internal cusp, which is found in none of the Canidæ and was a first step toward the tuberculated pattern of the raccoons, and the lower sectorial had a very low cutting blade and large heel; the other molars of both jaws were low, wide and of subquadrate shape. The skull was short and broad, with the face as much shortened and the orbits as far forward as in Procyon, but the brain-case was narrower, less capacious, and the lower jaw had the curved form and much the same character as in the modern genus. The limbs were relatively more slender than in the latter and the five-toed feet were more canine than procyonine in the proportions of the digits.

The discovery of †Phlaocyon by Dr. Matthew was an event of capital importance, as showing the highly probable derivation of the raccoons from †Cynodictis and thus bringing another fissipede family into relationship with the dogs.

4. Ursidæ. Bears

The present distribution of the bear family is all but exclusively northern, as there is but one African species, confined to the northwestern corner of that continent, and one in the Andes of Peru and Ecuador, all the others belonging to Eurasia and North America.

Structurally, the family is very distinct and the dentition is quite peculiar. The incisors and canines resemble those of other Fissipedia; the three anterior premolars are very small, single-rooted and often shed early; the carnassials have lost their trenchant character; and the molars, which are usually longer than wide, are tuberculated, somewhat resembling those of pigs. Almost all the bears live principally upon vegetable food, and even the Polar Bear, which feeds upon fish and seals, will eat grass and berries in the brief Arctic summer; thus, the shearing teeth of the strictly carnivorous types are unnecessary to these animals. The skull is not unlike that of the dogs in shape, but the tympanic bullæ are much flattened and the entrances to them are long, bony tubes, while the cranial foramina are nearly as in the dogs. The body is very heavy and the tail always short. The limbs are short and thick; the humerus has lost the epicondylar foramen in all existing species except the South American Spectacled Bear (Tremarctos ornatus). The plantigrade feet have naked soles (except in the Polar Bear) and each foot has five well-developed and functional digits, armed with very long, sharp and non-retractile claws.

The Pleistocene representatives of the family in America included species of the true bears (Ursus) and of the very large †short-faced bears (†Arctotherium) which ranged over both North and South America. In †Arctotherium the dentition was less modified; the larger premolars were very closely crowded together and the molars were nearly square; the lower jaw was almost as much curved as in the raccoons. The humerus retained the epicondylar foramen. The family, which was of Old World origin, may have reached America in the lower Pliocene, but was rare until the late Pleistocene. †Arctotherium has not been found in the eastern hemisphere, but that, of course, is no proof that the genus was not an immigrant from Asia. On the other hand, it may have been a peculiar American development from Pliocene immigrants. In the Old World, bears were first distinguishable in the upper Miocene, and may be there traced back to forms which were unmistakably derivatives of the early dogs.

5. Mustelidæ. Mustelines

The last fissipede family, which has, or has had, representatives in the western hemisphere is that which includes a great variety of small carnivores, such as minks, martens, skunks, badgers, otters, etc., and was likewise of Old World origin, though now of universal distribution, except in Australia and Madagascar. These are fierce and bloodthirsty beasts of prey, most of them strictly carnivorous and often killing in mere wantonness more than they can devour. Though now quite numerous and varied in North and South America, they are decidedly less so than in the eastern hemisphere and comparatively few peculiar types have originated here. Owing to the small size and fragility of the skeletons, they have not been well preserved as fossils, and little can be done as yet in tracing out the genealogy of the various phyla.

The mustelines have shortened jaws and a reduced number of teeth, the molars being 1/2 or even 1/1 and the premolars varying from four to two, though three in each jaw is the usual number. The cranium is generally very long and the facial part of the skull short, but the soft snout may add considerably to the length of the face. The tympanic bullæ are single-chambered and little inflated, and the lower lip of the entrance is extended; the hard palate is usually continued well back of the teeth. The body is very long and the tail variable and, in most of the genera, is short rather than long. The limbs are short, the feet, except in one genus, five-toed and plantigrade or semi-plantigrade, and the claws are non-retractile. Terrestrial, arboreal, burrowing, aquatic and marine forms are all represented in the family.

So far as North America is concerned, it is scarcely practicable to do more than catalogue the genera of the successive geological epochs. Pleistocene mustelines were very modern in character, differing little from those now inhabiting the continent, though in some cases with different ranges, according to climatic fluctuations. Badgers, martens, skunks and others occurred then very much as they do now and the Boreal Wolverene extended down to Pennsylvania. Little is known of Pliocene mustelines, the Blanco having yielded fragments of only one genus of uncertain affinities and though several genera occurred in the lower Pliocene, but one, a marten (Martes), can be identified. Unquestionably, North America had many more Pliocene members of the family, but the conditions of preservation were unfavourable.

Much the same is true of the Miocene stages. In the upper Miocene there were a marten (Martes), a weasel (Mustela) and two otters (†Potamotherium and the modern Lutra), of which the marten and the more primitive otter went back to the middle Miocene. In the lower Miocene were several mustelines quite different from any now existing. One of those, †Megalictis, was truly gigantic, with a skull nearly as large as that of a Black Bear and having heavy, pointed claws. This and a similar genus, †Ælurocyon, were related to the Ratel (Mellivora) of India and Africa and, more closely, to the Wolverene. †Oligobunis, a much smaller animal, was apparently of the same group. This genus was also in the upper Oligocene, but there represented by a larger species, which was as large as a badger.

The White River beds have yielded but a single genus, †Bunælurus, which was the most primitive of American mustelines and had four premolars and two molars in each jaw, though the second upper molar was extremely small. The face was much less shortened than in the modern weasels and the tympanic bullæ were short and strongly inflated and had no tubular entrance, and were thus canine rather than musteline in form. The bony palate was not extended back of the teeth as it is in the modern genera. The same primitive group was much more abundant in the European Oligocene, migrating probably from Asia into Europe as well as into North America.

SOUTH AMERICAN FISSIPEDIA

The history of the South American carnivores is a comparatively brief one; the southern continent has representatives of the same five families as the northern, but most of the genera are different, the time since the great southward migration having been sufficient for the development of peculiar forms in the new environment. Among the dogs, there are to be noted the curious, close-haired, long-bodied and short-legged Bush-Dog (Icticyon) and the fox-like wolves (Cerdocyon), but there are no true foxes. Of the cats, the Puma differs little from that of North America, and the Jaguar (Felis onca) and Ocelot (F. pardalis) also range into the northern continent, but several small cats are confined to South America, which has no lynxes. There is but one bear (Tremarctos ornatus) of Andean range. Of the Procyonidæ, the northern Procyon lotor is replaced by the Crab-eating Raccoon, P. cancrivorus, while the coatis (Nasua) and kinkajou (Potos) are chiefly Neotropical. Except for the otters, the genera of Mustelidæ are nearly all different; there are no badgers and a different genus of skunks (Conepatus) replaces the northern Mephitis; the Grison (Grison), Tayra (Tayra) and the Patagonian Lyncodon are peculiar.

Even less can be done to trace the evolution of the South American genera than for the forms of the northern continent, whence migrated the more or less different ancestors of the former. The Pleistocene has yielded most of the modern genera, both existing and extinct species. An example of the latter was Procyon †ursinus from the Brazilian caverns, a truly gigantic Raccoon, as large as a bear. The †sabre-tooth tigers (†Smilodon) and short-faced bears (†Arctotherium) were shared with North America. In the Pliocene a bear, a raccoon and a dog were the only known fissipedes, and in the Miocene none have been found, their place being taken by flesh-eating marsupials.

While the history of the Fissipedia, as outlined in the preceding pages, is sadly incomplete as compared with that of many ungulates, it is nevertheless highly suggestive. In each family the advance of specialization and adaptation to a narrow range of habits may be followed; generally speaking, the teeth were diminished in number and increased in size and were either simplified by the loss of parts, as in the cats, or complicated by the addition of new elements, as in the bears and raccoons. The brain grew larger and more convoluted and the cranium more capacious; in most of the families, the face was shortened, notably in the cats and mustelines, while in others, especially the dogs, it was elongated. In all of the early types there was a long and heavy tail, but in most series it underwent more or less reduction. There was little reduction of digits, and no fissipede has less than four. In modern dogs and cats there are five digits in the manus and four in the pes and the hyenas have four in each, as has one genus of mustelines; other modern genera throughout the suborder are pentadactyl.

It is significant that the more ancient members of the various families differed less than do the modern ones; the various groups, as they are traced back in time, would seem to be converging to a common ancestry, of which the lower Oligocene dogs were the least changed representatives, and it is probable that all the families of the Fissipedia were derived, directly or indirectly, from a single Eocene group of primitive flesh-eaters. The families, none of which is extinct, are not all of equal antiquity. So far as now appears, the dogs and viverrines are the most ancient, having become distinct in the upper Eocene; in the Oligocene were added the mustelines and cats; the raccoons branched off from the dogs in the lower Miocene, as did the bears in the upper Miocene. Finally, the hyenas appeared in the lower Pliocene, seemingly derived from the viverrines. The dogs passed through the greater part of their development in North America, where, during the Oligocene and Miocene, they were very abundant and varied, while at the same time they were comparatively rare in Europe and belonged chiefly to the phylum of the †bear-dogs. On the other hand, the remaining four families are of Old World origin, the bears and mustelines migrating to America, while the viverrines and hyenas did not.

Suborder †Creodonta. †Primitive Flesh-eaters

This group long preceded the Fissipedia in time, for they began their recorded history in the Paleocene and became extinct in the Oligocene. Through one family, the †Miacidæ, the †creodonts were broadly connected with the fissipedes, and it seems probable that that family was the ancestral stock from which all the fissipede families were derived. The other †creodont families died out without leaving descendants.

There is some difference of practice as to the number of families to be admitted; the table contains those listed in Professor Osborn’s book and also adopted by Dr. Schlosser. I should prefer a somewhat larger number of family groups, but the matter is one of secondary importance. Many genera are omitted.

The †Creodonta were an extremely varied assemblage, of carnivorous, omnivorous and presumably insectivorous habits, so that few statements, not subject to exceptions, can be made of them all. Only seven genera are known from skeletons, and several more from skulls, but most are represented only by jaws and teeth; limb- and foot-bones, however, give us a conception of the general structure of a considerable number. As a rule, the dentition was complete, according to the formula, i 3/3, c 1/1, p 4/4, m 3/3, × 2 = 44, but the first premolar or the last molar may be lost. The canines were always large, as was befitting for beasts of prey. In only one family, the Miacidæ, were the carnassial teeth confined to a single pair and those the same as in the Fissipedia, the fourth upper premolar and first lower molar; in all the other families there were either no sectorial teeth, or else there was more than one pair. In the Fissipedia the first is the largest of the lower molars, while in the †Creodonta (except the †Miacidæ) it was usually the smallest. The premolars were generally simple, compressed-conical teeth and the molars, with all their great variety, may be reduced to a common plan; those of the upper jaw were primitively tritubercular, with a triangle of two external and one internal cusps, and those of the lower jaw were in two distinct parts, an anterior, elevated triangle of three cusps and a low heel of two.

The skull was almost always very large in proportion to the size of the animal; the cranium, though long, was of small capacity and the face varied much in length in the different families. Primitively, the face and jaws were short in correlation with the small size of the teeth, and this primitive condition was modified in two opposite directions; in one the face and jaws were elongated, as the teeth enlarged, and in the other they were shortened still further. The zygomatic arches were stout and curved out strongly from the sides of the skull, making very wide openings, and, in almost all cases, the sagittal and occipital crests were very high, as would be necessary from the combination of powerful jaws and small brain-case (see p. 63). The tympanic bullæ were not ossified. The brain was extremely small, especially in the more ancient genera, and the convolutions were almost always few and simple, which indicates a low grade of intelligence and very marked inferiority to the Fissipedia.

In all the genera of which sufficient material has been obtained the body was long and had 19 or 20 trunk-vertebræ: in the lumbar and posterior part of the dorsal regions the processes by which the successive vertebræ were articulated together (zygapophyses) were cylindrical and interlocking, as in the artiodactyl ungulates (p. 360). To this general statement, the †Miacidæ formed a partial exception. The tail was very long and heavy in all the forms of which the caudal vertebræ are known, and this was probably true of all. The limbs were short and generally heavy; the femur had the third trochanter and the humerus, save in a few of the later genera, the epicondylar foramen, and the manus could, in nearly all, be freely rotated. Except in the most advanced forms of one family, the †Mesonychidæ, the feet were five-toed and plantigrade, or semi-plantigrade, and of decidedly primitive structure. The scapho-lunar bone of the Fissipedia (see p. 519) was not formed, its three elements, with very few exceptions, remaining separate. The astragalus nearly always had a shallow groove, or none at all. The claws were thick and blunt and the ungual phalanges cleft at the end, except in the †Arctocyonidæ and †Miacidæ, which had sharp claws and uncleft phalanges.

From this brief description, it is obvious that the †Miacidæ occupied a very isolated position among the †creodonts and, in my judgment, it would be better to transfer that family to the Fissipedia and include the others in a separate order.

Throughout the Paleocene and Eocene epochs the †Creodonta were numerous and varied, the first of the Fissipedia appearing in the upper Eocene. Till then the †creodonts were the only predaceous mammals in North America and Europe, and they were especially abundant in the former. Most members of the suborder and all the Paleocene forms were of small or moderate size, but some of the Eocene species were very large. In the Uinta the †creodonts were greatly decreased in numbers and in the White River there were only two genera of one family, the †Hyænodontidæ, and since the Oligocene the suborder has been extinct.

1. †Miacidæ. Fissipede-like †Creodonts

It is unfortunate that no member of this family is known from a complete skeleton, but the material collected is sufficient to give a fairly adequate conception of these most interesting animals. These were the only †creodonts with a single pair of carnassials, the fourth upper premolar and first lower molar, but in some of the genera the carnassials did not differ greatly from the other teeth. In the various genera the skull differed considerably in length and in the proportions of cranium and face; the brain-case was larger than in most other †creodonts and the brain more advanced, though smaller than in the fissipedes, and the sagittal and occipital crests were very prominent; the tympanic bullæ were not ossified. The humerus had the epicondylar foramen and the femur the third trochanter; in the wrist the scaphoid, lunar and central were separate, almost the only important difference from the Fissipedia and merely the primitive stage of the latter. The feet were pentadactyl and the digits were arranged in spreading fashion; the claws were small, sharp and partially retractile and the ungual phalanges not cleft at the tip.

Within the family several different phyla may be distinguished, one of which (†Miacis—†Uintacyon) led to the dogs, another to the †bear-dogs, or †amphicyons. A third phylum (†Didymictis—†Viverravus) is by several authorities regarded as ancestral to the civet family, or viverrines, of the Old World, and a fourth (†Oödectes, †Vulpavus) as the forerunner of the kinkajous (Potos). Except for the connection with the dogs, the hiatus in time between the supposed ancestors and descendants is too great to permit any confident statements. It seems very probable, however, that the †Miacidæ represented the common stock, from which the fissipede families were all derived, directly or indirectly, though for most of them the details of the connection remain to be learned.

We find thus a group separating itself from the other †creodonts in the older Paleocene and gradually assuming fissipede characteristics, at the same time dividing into several phyla. In the upper Eocene this group passed almost imperceptibly into the Fissipedia, more obviously into the dog family, which, as we have seen, represents the central line of fissipede development.

2. †Mesonychidæ

This family displayed, in certain respects, the highest degree of specialization attained by any †creodonts, for they were the only ones which acquired cursorial limbs and feet. The †mesonychids were prevailingly, but not exclusively, a North American family and their range in time was through the Paleocene and Eocene.

The teeth, in the more advanced genera, had a curious mingling of primitive and specialized characters and none were sectorial in the proper sense of the word. The incisors were small, the canines large and bear-like and the premolars simple. The upper molars were very primitive, retaining the original tritubercular pattern, except that the two outer cusps were joined together, but the lower molars had lost all the internal cusps, which gave them a carnassial appearance; they were not sectorial, however, for their cusps wore directly against the upper teeth, not shearing past them, and were greatly blunted and worn down by use.

The last of the family was †Harpagolestes, of the Uinta and Bridger, one of the largest of the †creodonts. The skull, which was of disproportionate size, exceeded that of the Grizzly Bear; the upper profile of the skull had considerable resemblance to that of a bear in the steep forward descent at the fore head. The teeth were more reduced than in the other members of the family through the loss of the second premolar and third molar of the upper jaw. The skeleton is little known, but the humerus had a long and prominent deltoid crest and an epicondylar foramen.

In the middle Bridger stage were closely allied and very similar genera, †Mesonyx and †Dromocyon (Fig. 139, p. 269), which were like small, big-headed wolves, for the skull was as long as that of a Black Bear. Though the cranium was very long, the brain-chamber was very small and the sagittal crest enormously high, to afford surface for the attachment of the powerful jaw-muscles. The tympanic bullæ were ossified and had quite long, tubular entrances, a feature which has been found in no other †creodont skull. The face and jaws were also elongate, giving the head quite a wolf-like appearance. The neck and body were of moderate length, but the tail was extremely long, slender and whip-like.

The limbs and feet were more specialized than in any other †creodont and the changes were all in the direction of adaptation to swift running. The humerus was very smooth, with low ridges, and, alone among †creodonts, had in these genera no epicondylar foramen, though the femur retained the third trochanter. The radius was broad and so interlocked with the humerus as to prevent any rotation of the manus. The feet were four-toed and much resembled those of the modern dogs and hyenas. In each foot the metapodials were closely appressed and parallel, not spreading, but arranged in two symmetrical pairs, a longer median and shorter lateral pair, much on the artiodactyl plan; the ankle-bone (astragalus) also had an artiodactyl look, with its deeply grooved surface for the tibia and pulley-like lower end. The ungual phalanges were so short and broad as almost to suggest hoofs rather than claws. It is clear that the gait was as fully digitigrade as in a modern wolf and these were the only †creodonts of which this is known to be true. These were somewhat puzzling animals; the whole structure of the limbs and feet was that of cursorial types, but the broad, blunt claws do not suggest the running down and capture of prey, nor were the teeth those of savage killers. The speed may have been defensive, to escape from enemies, and the food may have been largely vegetable.

Ancestors of these Bridger genera have not been found yet in the Wasatch, a time when the family was represented by †Pachyæna, some of the species of which were very large, rivalling †Harpagolestes, which was descended from one or more of them. †Pachyæna had extremely massive teeth and was not improbably a carrion-feeder of hyena-like habits, and it retained the epicondylar foramen of the humerus and pentadactyl feet.

Much more primitive was †Dissacus, of the upper Paleocene, which was very probably the direct ancestor of both the Wasatch and the Bridger genera. The upper molars were substantially as in the latter, but the lower molars had the internal cusp of the primitive triangle, though the heel was trenchant, and had lost its inner cusps. The feet had five well-developed digits, which were arranged in spreading fashion, and the gait was plantigrade. The claws were longer, more pointed and much less hoof-like than in the Bridger genera. The Puerco genus †Triisodon may or may not have been directly ancestral to †Dissacus; at all events, it was very nearly what the desired ancestor must have been. The teeth were much less specialized than in †Dissacus; the tritubercular upper molars were broader and their external cusps were more separated, while in the lower molars the anterior triangle was made up of three nearly equal cusps and the heel was low and basin-shaped. The skull had an extremely narrow brain-case and a long, heavy sagittal crest.

The most interesting feature in the history of the †Mesonychidæ is the demonstrable derivation of the cursorial, digitigrade, four-toed and almost hoofed Bridger genera from the plantigrade, five-toed Torrejon genus, which had sharp claws. To all appearances, this family was the †creodont analogue of the hyenas.

3, 4. †Arctocyonidæ and †Oxyclænidæ

This second †creodont family which had no carnassial teeth has received the not very happily chosen name of †Arctocyonidæ, or “bear-dogs,” though they were not related to either bears or dogs. The family was a very ancient one and has been found only in the Paleocene and lower Eocene (Torrejon and Wasatch) of North America and Europe. The molar teeth were very low-crowned and quadritubercular, with numerous small tubercles in addition to the four principal cusps, a pattern which was rather pig-like than bear-like. The Wasatch genus †Anacodon, known only from jaws and teeth, had reduced premolars, both in size and number, while in the Torrejon genus, †Clænodon, the premolars, though small, were present in full number. The skull was like that of †Mesonyx in the relative lengths of cranium and face, the very small size of the brain-case and the great prominence of the occipital and sagittal crests. The feet were pentadactyl and plantigrade and the claws were long, thin and pointed, and the ungual phalanges were not cleft at the tip, the only †creodont family, except the †Miacidæ, of which this was true.

Of the †Oxyclænidæ, very little is known and they may not have been †creodonts at all. They were quite small animals, with sharp-cusped tritubercular upper molars and lower molars with high anterior triangle and low heel. This is the type of dentition from which all the divergent †creodont types were doubtless derived. The family was Paleocene.

5. †Hyænodontidæ

This was the last of the †creodont families to survive, being quite common in the lower Oligocene of North America and Europe and in the upper Eocene of the latter also. The family became extinct in the upper Eocene of North America and the White River genera were not of native origin, but migrants from the Old World. One of the more abundant predaceous genera of White River times was the European †Hyænodon; it was represented by several species which ranged in size from a fox to a Black Bear. In this genus the dentition was somewhat reduced, the incisors often numbering 2/2 and the molars constantly 2/3; there were three pairs of carnassial teeth on each side, of which the pair formed by the second upper and third lower molar was the largest and most efficient, the other pairs being the first upper and second lower molar, the fourth upper premolar and first lower molar, the latter the smallest of the three. The upper molars had lost the internal cusp and the remaining, external portion consisted of a flattened-conical anterior cusp and a posterior trenchant ridge; the milk-teeth of †Hyænodon, as well as the permanent dentition of the ancestral genera, show that the anterior cusp was composed of the two external cusps of the primitive tritubercular tooth fused into one and that the trenchant ridge was a superadded element. The fourth upper premolar was a sectorial like that of the Fissipedia, but of an unfinished, ineffective sort. The third lower molar was very similar in shape to the carnassial of the cats and was composed of only two large, thin and trenchant cusps, which made a shearing blade, having lost the inner cusp of the primitive triangle and the heel. The first and second molars were like the third except in size and in retaining a vestige of the heel. The premolars were large and massive, almost hyena-like, which suggested the generic name. The canines were prominent and strong.