The story of the hoofed mammals, as sketched in brief outline in the preceding chapters (VIII-XIII), is a curious mixture of relatively full and satisfactory paragraphs, with scanty, broken and unintelligible ones, not to mention those which have not yet been brought to light at all. With all its gaps and defects, which inhere in the nature of things, the history of the various ungulate series is the best that the palæontology of mammals has to offer and constitutes a very strong and solid argument for the theory of evolution. For the Carnivora the story is less complete and for obvious reasons. Individual abundance was a very large factor in determining the chances of preservation in the fossil state for any given species, and, as a rule, whole skeletons are found only when the species was fossilized in large numbers. In any region the Carnivora are less numerous than the herbivora upon which they prey, and while most ungulates live in larger or smaller herds, the carnivores are mostly solitary.

The Carnivora are divisible into three well-marked suborders, called respectively the Pinnipedia, Fissipedia and †Creodonta. The Pinnipedia, seals, walruses, etc., which are almost purely marine in habitat, are not dealt with in this book, since so little can be learned of them from the fossils, and the †Creodonta, an extremely ancient and primitive group, will be treated separately. The Fissipedia are chiefly terrestrial, though they include the otters, and their subdivisions, so far as the American forms are concerned, are shown in the following table, which, it should be observed, omits several genera. Unless otherwise noted, the genera are North American.

Two families, the hyenas (Hyænidæ) and civet-cats (Viverridæ), are omitted from the table because they apparently never reached the western hemisphere. The bears, of Old World origin, invaded America at a very late period and are not certainly known here before the Pleistocene. The other four families were well represented in North American history, though the great weasel tribe (Mustelidæ) went through the greater part of its history in the Old World. None of the families is indigenous in South America, and all of the five families which it now shares with North America came in in the series of immigrations, of which the first recorded effects are found in the Pliocene and continued into the Pleistocene.

The Fissipedia are adapted to a great variety of habits and modes of life and consequently there is considerable diversity of structure among them, though they all form a homogeneous, natural group. The dogs (Canidæ) are terrestrial, neither swimmers nor climbers; some, like the foxes, are solitary, others, like the wolves, hunt in packs and nearly all are strong, swift runners. The cats (Felidæ) which have a remarkable range of size, are terrestrial or arboreal; they take their prey by stalking and leaping upon it, not by running it down. The bears (Ursidæ) are mostly omnivorous, not very often killing prey, and largely vegetarian in diet. The raccoons (Procyonidæ) are chiefly arboreal and omnivorous. The very large and varied weasel family (Mustelidæ) have different habits, though nearly all are fierce and bloodthirsty. Otters and sea-otters are aquatic and prey chiefly on fish; minks and fishers are semi-aquatic; martens are arboreal, skunks terrestrial and badgers fossorial.

While there is thus much diversity of habit with corresponding differences of structure among the Fissipedia, there is a certain unity of plan recognizable among them all. With but few exceptions, the incisors are present in full number and the canines are formidable lacerating weapons. Especially characteristic of the dentition are the “sectorial” or “carnassial” teeth, always the fourth upper premolar and first lower molar, which form a pair of shearing blades, the premolar biting outside. In the bears and most of the raccoons the teeth are tuberculated, in adaptation to the omnivorous habit, and the carnassials have lost the shearing form, though clearly derived from that type. The skull has powerful jaws, and the crests and ridges for the attachment of the jaw muscles are prominent except in very small animals, and the stout, boldly outcurving zygomatic arches are very characteristic. The face may be elongate, as in the dogs, or extremely short, as in the cats, or of intermediate length; the brain-case is relatively capacious, and the orbits, except in the cats, are widely open behind. The neck is never very long, but the body often is, and the tail varies greatly in length, as do also the limbs. There is great difference, too, between the various families in the prominence of the processes on the limb-bones for the attachment of muscles, as expressive of the muscular development of the limbs, and also in the extent to which the fore foot can be rotated and used for grasping. In all existing Fissipedia the femur has no third trochanter, but many extinct genera possessed it. The bones of the fore-arm and lower leg are always separate and uninterrupted.

In the wrist (carpus) there is always a large bone, the scapho-lunar, which is made up by the coalescence of three elements, the scaphoid, lunar and central, a feature which, though recurring in a few other mammals, is essentially characteristic of the modern Carnivora. The feet are armed with claws more or less sharp, which in some families, notably the cats, are retractile and may be folded back into the foot. The gait may be plantigrade, as in the raccoons and bears, or digitigrade, as in the dogs and cats, or intermediate in character.

Throughout the Paleocene and most of the Eocene, there were no Fissipedia, the flesh-eaters all belonging to the extinct †Creodonta, and the first clearly recognizable fissipedes occurred in the upper Eocene or Uinta.

1. Canidæ. Dogs, Wolves, Foxes, etc.

This family, which may with convenience be called simply dogs, is at present the most widely distributed of the families of Fissipedia, occurring in every continent, even Australia, and ranging through all climates almost from pole to pole. They are a singularly homogeneous family and show few differences of structure; such differences as there are affect chiefly the number and size of the teeth and external characters, such as the size of the ears, length and colouring of the hair, etc. The many domestic breeds are not here considered. Almost alone among the Fissipedia the dogs capture their prey by running it down, and they are endowed with remarkable speed and endurance. The entire organism, especially the limbs and feet, are adapted to cursorial habits.

For the purpose of comparison with the extinct genera of the family, some account of a wolf will suffice. The wolves, like most other members of the family, have a larger number of teeth than is usual in the suborder, as appears from the formula: i 3/3, c 1/1, p 4/4, m 3/3, × 2 = 42, that is to say, only the third upper molar has been lost from the typical number, though the third lower is very small and seemingly on the point of disappearance (Fig. 44, p. 93). The upper sectorial tooth, the fourth premolar, has its shearing blade made up of two sharp-edged cusps, one behind the other, and there is a small internal cusp carried on a separate root; the upper molars are triangular and tritubercular and are used for crushing. The lower sectorial, the first molar, has an anterior blade of two shearing cusps, with the remnant of a third, and a low, basin-like posterior “heel.”

The skull is characterized by the long face and jaws and by the structure of the auditory region; the tympanic bones are inflated into large oval bullæ, which are hollow and undivided, and the external opening of each is an irregular hole, without tubular prolongation. There is an alisphenoid canal for the passage of the internal carotid artery. The neck, body and tail are of moderate length and the vertebræ of the loins are not conspicuously large and heavy. There is no collar-bone. The limb-bones have a distinct, though superficial, resemblance to those of hoofed animals; the humerus has no very prominent ridges for the attachment of muscles and no epicondylar foramen, and the femur no third trochanter. The fore-arm bones are separate, but are so articulated together and with the humerus as to give the fore foot no power of rotation. The manus in all existing wild species has five digits, though the pollex or first digit is very small, a mere dew-claw; the four functional digits are arranged in two symmetrical pairs, very much as in the artiodactyls, a longer median pair, of which the metacarpals have a nearly square cross-section, and a shorter lateral pair (2d and 5th) of more trihedral form. All the metacarpals are closely appressed and almost parallel. The pes has four digits arranged in similar fashion. The claws are blunt and non-retractile, and are of little use in seizing or lacerating prey, but are useful in digging. The ungual phalanges have no bony hoods reflected over the base of the claw. All modern forms are digitigrade.

Materials are lacking for the construction of any such detailed phylogeny of the dogs as has been accomplished for many ungulates. Many of the extinct genera are known only from skulls, or even jaws, and the well-preserved skulls are too few to form distinctly defined and continuous series. On the other hand, there is every reason to believe that the canine genera of the successive geological stages did approximately represent the successive steps of development within the family, though it is difficult to distinguish between the phyla.

The Pleistocene dogs, for the most part, differed little from the Recent ones; there were some very large species like the Canis †dirus (Frontispiece) of the Mississippi Valley and the Pacific Coast. Two very peculiar genera have been reported. One (†Pachycyon), from a cave in Virginia, had remarkably short, stout and strongly curved limb-bones, which suggest otter-like habits; the other (†Hyænognathus), from California, had a very short face and extremely massive lower jaw and very heavy teeth; it was probably like a hyena in appearance.

As far back as the Blanco stage of the middle Pliocene, remains occur which are assigned to the modern genus Canis, though better preserved specimens would probably require their removal from that genus. In the lower Pliocene the phylum of the true wolves was represented by †Tephrocyon, which, so far as it is known, differed only in minor details from Canis, and †Tephrocyon went back to the middle Miocene. What would appear to be its direct ancestor is †Cynodesmus, of the lower Miocene, which, in view of the long lapse of time involved, differed less from the modern wolves than one would have supposed, but the differences are significant, as pointing back to a far more primitive type of structure. †Cynodesmus was a small animal, intermediate in size between a Red Fox and a Coyote. The dental formula was the same as in Canis, but the teeth were relatively smaller and more closely crowded, as the face and jaws were shorter and the cranium, though longer, had a less capacious brain-chamber. The cast of this chamber, which very perfectly reproduces the form of the brain, shows that the latter was not only smaller but less convoluted than in the modern animals, and this, in turn, denotes a lower grade of intelligence. The limb-bones were like those of wolves, but the feet were quite different. In the manus the first digit, or pollex, was much less reduced, though considerably shorter than the other digits, which were not in two symmetrical pairs, but were all of different lengths, not closely appressed, but arranged in radiating fashion; the metacarpals had not yet acquired the quadrate or trihedral form, but were more oval in cross-section. The pes was more modernized, but had five digits, which is not true of any existing member of the family. The claws were thin and sharp and were slightly retractile, a power which has been completely lost in all the modern canids. Such an animal could hardly have been preëminently cursorial.

Out of the crowd of dog-like creatures in the John Day Oligocene, it is not yet practicable to select one which is to be taken as the ancestor of the Recent wolves through †Cynodesmus, nor can this be done with better assurance of success in the White River, though the beginning (†Daphœnus) of the †bear-dogs in that formation probably closely represents the ancestral stage sought for. It is likely that several of the phyla into which the family was divided became blended in a common stock at that stage.

A second phylum, now entirely extinct, is that of the †bear-dogs, which is not certainly recorded later than the middle Pliocene, though some have been doubtfully reported from the older Pleistocene of the Great Plains and the remarkable Californian genus, †Hyænognathus, may have been an offshoot of the same stock. The phylum was characterized by the unusually large size of the molars and by certain other features, which, however, are not known to have persisted through the entire series from first to last. In the middle Pliocene lived some very large bear-dogs, of the genus †Borophagus, the teeth of which had a strong likeness to those of the hyenas and probably the animals had hyena-like habits, feeding largely upon carrion and crushing the stoutest bones with their massive teeth. The same, or a very similar, genus lived in the lower Pliocene, but none of the species of that date is at all well known. In the upper Miocene occurred several species which have been referred to the European genera, †Amphicyon and †Dinocyon. The latter was an enormous canid, equalling in size the largest of living bears, the great Kadiak Bear of Alaska, and, though probably having a long and heavy tail, was much like a bear in appearance. The teeth indicate a more exclusively carnivorous habit than that of the bears and these may well have been savage and terrible beasts of prey.

†Amphicyon, which had three upper molars, continued down through the middle Miocene, but was replaced in the lower by †Daphœnodon, which may or may not have been its direct ancestor. The uncertainty as to the exact relationship between the two genera will remain until more complete material shall have been obtained from the middle Miocene. †Daphœnodon was the largest dog of its time, the contemporary wolves (†Cynodesmus) having been hardly half so large, but was much inferior in size to the huge †bear-dogs of the middle and upper Miocene. The skull resembled that of a large wolf, but the tympanic bullæ were smaller and more loosely attached and the molar teeth were relatively much larger, a persistent characteristic of this phylum. The very long and heavy tail was a cat-like feature. The limbs were comparatively short and stout; the humerus had the epicondylar foramen and the femur retained a trace of the third trochanter, both of which are lost in the modern members of the family. The feet were not at all canine in type, but rather resembled those of the ancient and unspecialized flesh-eaters. There were five digits in manus and pes and were not arranged in parallel pairs, but diverging; the metapodials were of oval cross-section, not squared, and their lower ends, which articulated with the first row of phalanges, had hemispherical surfaces, not semicylindrical. The claws were sharp and a remnant of former retractility was to be observed. Such an animal could hardly have been a strong and enduring runner and its structure suggests that it captured its prey by stalking and leaping upon it. The wolf-like head, with cat-like body, tail and limbs, made a strange combination, not closely paralleled by any existing carnivore.

Through the Oligocene the phylum was carried back by the several species of †Daphœnus, assuredly the ancestor of †Daphœnodon and decidedly more primitive in many respects. The Oligocene genus was a much smaller animal than its lower Miocene successor, the larger species hardly equalling a Coyote; the teeth were smaller and more closely set, but the molars were proportionately large, while the carnassials were less finished and effective shearing blades. The skull was less distinctively dog-like and had a smaller brain-case, with very prominent sagittal and occipital crests, a longer cranium and shorter face; the tympanic bones were very small and so loosely attached to the skull that they are rarely found, a very striking difference from all existing dogs. The backbone was remarkable for the unusually large size of the lumbar vertebræ, a point of resemblance to the cats and suggesting that †Daphœnus had great powers of leaping; there was a long, heavy, leopard-like tail, and the caudal vertebræ were very like those of the long-tailed cats. The limbs and feet were similar in character and proportions to those of †Daphœnodon, but the astragalus was less grooved for the tibia, the claws were rather more retractile and the gait was probably more plantigrade. There were so many cat-like features in the skeleton of †Daphœnus, that the observer cannot but suspect that these resemblances indicate a community of origin, but, until the Eocene ancestors of the cats are found, the question of relationship must remain an open one.

The most ancient member of the bear-dog phylum yet discovered appears to be one of the †creodont family of the †Miacidæ, found in the Uinta Eocene.

A short-lived branch of the canine stock was that of the so-called “†hyena-dogs,” a peculiar American type, which abounded in the upper Miocene and lower Pliocene and then became extinct. Traced backward, this brief series of species would appear to have sprung from the true wolves (†Tephrocyon) of the middle Miocene. The upper Miocene and lower Pliocene genus †Ælurodon had several species, which differed considerably in size; the commoner of these were large wolves with very modern type of body, tail, limbs and feet, but having short and massive heads. The premolars were extremely thick and heavy, with such a resemblance to those of the hyenas, that these animals have sometimes been mistakenly regarded as ancestral to that family. The especial characteristic, however, of the series was in the form of the upper sectorial tooth, which was much more feline than canine in construction and has given occasion for the generic name which means “cat-tooth.”

A fourth phylum of the Canidæ, which would seem to be represented in the modern world by the Indian Dhole, or Wild Dog (Cyon), and perhaps by the Brazilian Bush-Dog (Icticyon), was characterized by the lower sectorial molar, the heel of which was not basin-like, as in the typical dogs, but trenchant and consisted of a single sharp-edged cusp, the external one of the primitive basin. Although there is no inherent improbability in the view that the Dhole and the Bush-Dog are derivatives of this phylum, no positive statement can yet be made, for the gap in the history is too great to be bridged with any assurance. The fossil members of the series did not come down later than the middle or upper Miocene and it is quite possible that the trenchant heel of the carnassial was developed more than once. The middle and lower Miocene members of the series are still very imperfectly known and it is only from the upper Oligocene (John Day) that well-preserved skeletons have been obtained. These pertain to an aberrant member of the phylum, the genus †Temnocyon, in which not only does the sectorial have a trenchant heel, but the second lower molar also was trenchant, having lost the two inner cusps, while the upper molars were as large as in the †bear-dogs.

†Temnocyon was a comparatively large animal and its skeleton had a mixture of primitive and advanced characters, the latter predominating, so that this genus was not only the largest but also the most specialized canid of its time. There was the long, heavy tail, which all of the known Oligocene carnivores possessed, but the limbs were long and the gait was, it would seem, thoroughly digitigrade. While the epicondylar foramen was retained by the humerus and the third trochanter by the femur, those bones were otherwise very modern in form. The feet were five-toed, but the functional metapodials were parallel, appressed and with something of the quadrate shape. In very notable degree, therefore, the feet of †Temnocyon anticipated the characters which the true wolves acquired considerably later. The less specialized †Mesocyon, which was smaller, was the ancestor of the Miocene forms and was, in turn, very probably derived from the White River †Daphœnus.

Still a fifth phylum, that of the †short-faced dogs (†Enhydrocyon), is very imperfectly known and has, so far, been found only in the lower Miocene and upper Oligocene. These also may have been descended from †Daphœnus, but the connection is not clear, nor has the relationship of the American genus to the extremely †short-faced dogs of the European Pliocene been determined.

Finally, so far as North America is concerned, there was a phylum of very small fox-like canids, which ranged from the lower Miocene to the upper Eocene and were very abundant, relatively speaking, in the White River and John Day. The dental formula was the same as in Canis and the skull was narrow and slender, though the brain-chamber was proportionately capacious, and the face was quite short. The tympanic bullæ were large and inflated. The body and tail were long and the limbs quite short and weak. The humerus had no epicondylar foramen and the femur no third trochanter. The five-toed feet had the spreading arrangement of the metapodials seen in the more primitive fissipedes generally and the claws were sharp. In proportions and appearance these animals must have been more like civets or weasels than like dogs and it is evident that they were not swift runners. The series had its earliest representatives (†Procynodictis) in the Uinta and was doubtless derived from the †creodont family †Miacidæ. The White River species are referred to the European genus †Cynodictis, those of the John Day and lower Miocene to †Nothocyon, and it has been suggested that this series gave rise to the foxes, a suggestion which may prove to be true, but the very long gap in time between these animals and the most ancient known foxes prevents any conclusion.

To determine the mutual relationships of the six phyla of Canidæ which, from the Eocene onward, inhabited North America in such numbers, is a task of great difficulty and only a tentative solution of the problem can be offered. The central stock would seem to be nearly represented by the White River †Daphœnus, leading through †Cynodesmus and †Tephrocyon, of the Miocene, to the wolves. A short-lived series, apparently given off from †Tephrocyon, was that of the †hyena-dogs, which flourished greatly in the upper Miocene and lower Pliocene and then became extinct. Another branch, that of the †bear-dogs, was derived from †Daphœnus, through †Daphœnodon to †Amphicyon, †Dinocyon and †Borophagus, the gigantic Miocene and Pliocene forms, ending perhaps in †Hyœnognathus of the California Pleistocene. A third branch, represented by †Mesocyon and †Temnocyon, is believed to be continued to-day by the Asiatic Dhole and the Brazilian Bush-Dog. The †short-faced dogs (†Enhydrocyon) are still very obscure. The last phylum, that of †Nothocyon, †Cynodictis, †Procynodictis, had become distinct in the upper Eocene and possibly gave rise to the foxes, but this is highly conjectural.

2. Felidæ. Cats

The only other fissipede group whose development in North America may be followed for a long period is that of the †Sabre-Tooth Tigers, the subfamily †Machairodontinæ, which have been extinct since the Pleistocene; the history of the True Cats (Felinæ) is much more obscure. In most respects the two subfamilies agreed closely and, as they became separate at least in the early Oligocene, they furnish instructive parallel series. The †sabre-tooth cats were terrible beasts of prey, which in most of the Tertiary period ranged over the whole northern hemisphere and in the Pleistocene or late Pliocene extended throughout South America.

The Pleistocene genus †Smilodon (Frontispiece) belonged to nearly the whole western hemisphere and its various species were distributed from California and Pennsylvania on the north, to the Argentine Pampas on the south. The most obvious and striking peculiarities of †Smilodon were in the teeth, which were much reduced in number, the formula being: i 3/3-2, c 1/1, p 2/2-1, m 1/1. The upper canine was a great, curved, scimitar-like blade, eight inches or more in length, with broad inner and outer faces, but quite thin transversely, and with finely serrate posterior edge. It is difficult to understand how these great tusks, which would seem to have blocked the entrance to the mouth, could have been effectively used, unless the creature could open its mouth much more widely than any existing mammal, so as to clear the points of the tusks, and would then strike with them as a snake does with its fangs. There are great anatomical difficulties in the way of accepting this explanation and the problem, which is the same as that presented by the †uintatheres (p. 446), is still unsolved. It is, however, quite certain that no arrangement which was disadvantageous, or even inefficient, could have persisted for such vast periods of time. The lower canine was much diminished and hardly larger than an incisor. The two upper premolars were the third and fourth of the original series; the third was small, but the fourth, the sectorial, was a very large and efficient shearing blade. In addition to the two external trenchant cusps of the blade, which are present in the Carnivora generally, the cats have a third small, anterior cusp which in †Smilodon was large; the internal cusp had almost disappeared. The single upper molar was very small and so overlapped by the great carnassial as to be invisible from the side. The third lower premolar was small and unimportant and most specimens had lost it, leaving only the fourth, which was larger and evidently of functional value. The single molar was the sectorial, a large, thin, flattened blade, consisting of only two cusps, one behind the other, the trenchant edges of which met at nearly a right angle, and there was no trace of a heel.

The skull was in appearance closely similar to that of one of the great modern cats, such as the Lion or Tiger; with extremely shortened face, heavy and widely expanded zygomatic arches and very prominent sagittal crest. The tympanic bullæ were large and inflated, each divided by a septum into two chambers, but were not visible from the side, being covered externally by very large processes, which served for the attachment of some of the great muscles of the neck. The short, rounded, bullet-head of the true cats was thus repeated, but there were in the skull several interesting differences of detail, which it is not worth while to enumerate here. Suffice it to say, that some of these differences were due to the retention of primitive characters in the skull of †Smilodon, which have been lost in the modern felines, and others to special developments, in which the true cats did not share. The lower jaw had on each side a small, descending flange for the protection of the tusks, which, however, projected well below these flanges when the jaws were shut. The neck was heavy and the structure of its vertebræ was such as to suggest the presence of unusually powerful muscles; the back and loins were also uncommonly stout, in the larger species heavier than in the Lion or Tiger, but, in marked distinction from those modern forms, the tail was short. The limbs were shorter and much heavier in relation to the size of the body than in the great existing cats and must have been extremely powerful. The humerus usually had no epicondylar foramen, which all the true felines possess, though it was sometimes present. The feet also were very stout and armed with large retractile claws; the base of each claw was covered by a thin bony hood, an outgrowth of the ungual phalanx, which is very characteristic of the entire family. The hind foot had five digits, whereas no existing cat has more or less than four. The appearance of these animals must have been very much like that of the Lion or Tiger, aside from the unknown factors of mane and colour-markings, but differed in the great tusks, the short tail and the shorter and more massive legs and feet.

On account of the very incomplete preservation of the material so far collected, little is known of the †sabre-tooth series in North America during the Pliocene and Miocene epochs. Remains of very large cats have been found in the lower Pliocene and upper Miocene, but it is uncertain whether they belong to the feline or the †machairodont subfamily. Some of the species have been referred to the genus †Machairodus, which ranged from the lower Pleistocene to the middle Miocene of Europe, and the reference may be correct, but is uncertain. However, the European representatives of that genus, which are much better known, will serve to show the developmental stage from which †Smilodon was undoubtedly derived. The dental formula was the same as in the American genus, though there were generally two premolars in the lower jaw and in †Smilodon generally but one; the individual teeth were formed on the same plan as in the latter, but were relatively smaller, and the very small, rudimentary upper molar was visible externally and was not overlapped and concealed by the great carnassial; the sabre-like tusk had not attained such great proportions. The skull of †Machairodus, the only part of the skeleton which is definitely known, was like that of †Smilodon on a much smaller scale, but more primitive in several respects. It was longer and had a less capacious brain-case and less prominent sagittal and occipital crests. The large tympanic bullæ were conspicuous in the side-view of the skull, as the processes for the attachment of the neck-muscles had no such development as in †Smilodon. The descending flanges of the lower jaw were larger than in the latter.

The upper Oligocene (John Day) contained a large variety of cat-like forms, of which no less than five genera have been described; one of them (†Pogonodon), nearly as large as a Lion, would seem to have died out here without descendants, and two others, to which we shall return later, so combined the characters of true felines and †machairodonts as to be of uncertain reference. Two other genera, which are much commoner and better known, from the White River, will be described from specimens of that stage.

The White River, or lower Oligocene, had three highly interesting genera of †machairodonts, two of them known from nearly or quite complete skeletons. One of these (†Hoplophoneus), which was, it can hardly be doubted, the direct ancestor of the later typical †machairodonts, had several species, which are found in the various levels of the White River beds. The largest of these species was considerably smaller than †Machairodus, and the smallest and most ancient was inferior to the modern Wild Cat. The number of teeth was variable, but normally greater than in the genera above described, being i 3/3, c 1/1, p 3-2/3-2, m 1/1, × 2 = 28-32. The foremost premolar in each jaw was very small and often absent. The upper canine was a long and curved, but very thin, scimitar, finely serrate on both edges, while the lower canine was but little larger than the incisors. The carnassial teeth had a significant likeness to those of other fissipede families; the upper one, the fourth premolar, was relatively smaller than in †Machairodus and its blade less effectively trenchant; the accessory antero-external cusp was present, though extremely small, and the internal cusp, which in †Smilodon had almost disappeared, was quite large. The lower sectorial, the first molar, though already cat-like and consisting of two thin, broad and trenchant cusps in line, yet had vestiges of the heel and sometimes of the inner cusp. These vestiges were a connecting link between the highly specialized sectorial of the cats and the type usual among the Fissipedia, which is exemplified by the dogs. The small upper molar was less reduced than in the Miocene and Pliocene genera and plainly consisted of a larger external and smaller internal cusp.

Compared with that of other Fissipedia, the skull was short and broad, but in comparison with that of the modern cats and of †Smilodon, it was decidedly longer and narrower and the face was less abbreviated; the resemblance to †Smilodon was very marked in the form of the cranium, but, of course, the skull of †Hoplophoneus was distinctly more primitive in many respects. Thus, the orbit was much more widely open behind, the tympanic bullæ were but imperfectly ossified, and the perforations, or foramina, in the base of the skull, by which the nerves and blood-vessels communicated with the brain-chamber, were quite different and had more resemblance to those of the ancient dogs (e.g. †Daphœnus). In the classification of the Fissipedia much stress is laid upon the number and arrangement of these cranial foramina, and it is very significant to find the primitive dogs and cats agreeing so much more closely than do the modern members of these families. The lower jaw was relatively much stouter than in †Smilodon and the anterior flanges much more prominent, projecting downward so far that, when the jaws were closed, the points of the tusks did not extend below the flanges. The animal could have made no use at all of the sabre-tusks unless the mouth could have been opened so widely as to clear their points.

With close general resemblance, allowing for the very inferior size, the skeleton of †Hoplophoneus had many significant differences from that of †Smilodon. The neck was shorter and the body, especially the loins, longer, lighter and more slender and the tail very much longer, equalling that of the Leopard in relative length and surpassing it in thickness. The limbs were much less massive and somewhat differently proportioned, the upper arm being shorter and the fore-arm longer. The humerus, though far more slender than that of †Smilodon, was remarkable for the great development of the deltoid and supinator ridges, the latter, together with the shape of the radius, indicating very free rotation of the fore paw. The very prominent internal epicondyle was pierced by a foramen, and the femur had a distinct remnant of the third trochanter. The five-toed feet were comparatively small, but the claws were as completely retractile and as fully hooded as in any of the subsequent genera.

That †Hoplophoneus was a fierce destroyer, is made evident by every part of its skeleton, and, like other cats, it no doubt subsisted upon warm-blooded animals, which it killed for itself, the size of the prey being determined by the size and power of the particular species of the †sabre-toothed genus. In view of the probable extent of the Oligocene forests, the restoration (Fig. 264) gives the animal a spotted coat and the general aspect is that of one of the modern spotted cats, but the protruding ends of the tusks and the relatively long head distinguish it from any existing cat. “The presence of long, knife-like canines is correlated with powerful grasping feet possessing highly developed retractile claws. With its powerful feet the animal clung to its prey, while it struck repeatedly with its thin, sharp sabres” (J. C. Merriam).

In the latter part of the White River stage lived one of the most highly specialized of the †machairodonts, so far, at least, as the dentition is concerned, for only the skull is known. This genus, †Eusmilus, which also occurred in the Oligocene of Europe, was apparently an example of premature specialization which led to nothing, for none of the subsequent genera could have been derived from it. The teeth were reduced to a minimum in number: i 3/2, c 1/1, p 2/1, m 1/1, × 2 = 24, one lower incisor and at least one premolar less in each jaw than had †Hoplophoneus. The canine tusk was very large and the flange of the lower jaw for its protection correspondingly elongated, being more prominent than in any other †machairodont. The American species, †E. dakotensis, was the largest carnivore of its time and not greatly inferior in size to the Lion.

Still another White River †machairodont, †Dinictis, differed in many interesting ways from its contemporary †Hoplophoneus, being more primitive and departing less from the ordinary fissipede type of structure. This is shown by the greater number of teeth, which was normally, i 3/3, c 1/1, p 3/3, m 1/2, × 2 = 34. The upper carnassial had a considerably larger internal cusp and the trenchant blade did not have the accessory anterior cusp, which is present in almost all other cats and was thus more dog-like than cat-like. The lower carnassial was more feline, but retained a remnant of the heel and of the inner cusp, but the latter was variable, being sometimes present in one side of the jaw and not in the other, a sign that it was on the point of disappearance. The upper molar was plainly a reduced form of the tritubercular tooth, in plan like that of the dogs, while the second lower molar was a very small, single-rooted tooth. No other American cat has such a primitive dentition as this, and, aside from the sabre-tusk, which was not nearly so long as in †Hoplophoneus, and the lower carnassial, it might almost as well have belonged to a dog or musteline.