In the Uinta stage of the upper Eocene lived the most ancient and primitive member of the family yet discovered, the genus †Protoreodon, which is in every respect what the ancestor of the White River genus should be. The functional transformation of the lower canine into a fourth incisor and the replacement of the canine by the first lower premolar had already taken place, but the molars were much more primitive than those of the White River and succeeding genera; the crescents were thicker and less complete, plainly indicating their derivation from conical cusps, and a small fifth cusp was present between the anterior pair of the upper molars, as in the †anthracotheres and other European families of the Artiodactyla †Primitiva. Before the discovery of †Protoreodon, the character of its molars was predicted by Dr. Schlosser, of Munich. The skull resembled that of the White River genera, except that the eye-socket was open behind, and there was no glandular pit in front of the eye. The skeleton is but partially known, but it has been ascertained that there were five toes in the manus and probably also in the pes.

Nothing has yet been discovered in formations older than the upper Eocene which can be regarded as ancestral to the †oreodonts, and this is not surprising in view of the extremely meagre and unsatisfactory nature of our information regarding the artiodactyls of the Bridger. On the whole, however, it seems rather more probable that the Uinta genus was an immigrant (whence, we cannot say) than that the Bridger will ever yield the desired ancestral forms. So long as the early Tertiary mammals of northern and central Asia remain unknown, this and many similar problems can find no definitive solution. The question of relationship with other families is bound up with that of the origin of the †oreodonts; many characters point to a connection with the †anthracotheres and, from the standpoint of present knowledge, that appears to be the most probable affinity; but, on the other hand, there are structural features which suggest relationship with the primitive camels. Between these and other alternatives, only the recovery of the middle and lower Eocene forms can finally decide.

Reviewing the long history of the oreodont family from the evolutionary point of view, we find a course of development which differs in several respects from that exemplified by most of the families previously considered:

(1) There was a general increase in size, though it was far from steady, and almost every genus had larger and smaller species, and in some of the phyla the species were far larger than in others. The members of the †Leptauchenia phylum were very small and no member of the family ever attained to more than moderate size.

(2) The upper molars early lost the fifth cusp, and after that there was little change in the dentition, except that in the †Merychyus and †Leptauchenia phyla the grinding teeth became hypsodont.

(3) There was great variety in the modifications of the skull, each phylum having its own peculiarities. The orbit, which was open behind in the Uinta †Protoreodon, was closed in the White River and all succeeding genera. In the †Merycochœrus series, the skull first enlarged, with little change in proportions, then elongated the facial region, then shortened the face and so reduced the nasals as to indicate the presence of a proboscis, culminating in the grotesque, ape-like skull of †Pronomotherium. In the †Leptauchenia phylum the skull became depressed and flattened and the face was invaded by great openings, or vacuities; the tympanic bullæ were enormously inflated and the orbits and ear-openings raised, presumably in adaptation to an amphibious mode of life. These were the extremes of change within the family; the other phyla need not be considered.

(4) At an early stage the digits were reduced from five to four, first in the pes and then in the manus, and there reduction ceased; though in †Merychyus, especially in the upper Miocene species, the lateral digits were very slender and, had this series survived, it would probably have led to didactyl forms.

In other respects there was very little difference in the skeletons of the various phyla and herein lies the peculiarity in the history of the family, great variety in the form of the skull, and, relatively speaking, hardly any change in the body, limbs or feet. In the horses, rhinoceroses and †titanotheres the modifications of the successive genera affected all parts of the structure, but in the †oreodonts, except for the loss of one digit in manus and pes and variations in the length of the tail, the skeletons of the latest genera did not differ in any important respect from those of the earliest. Such a combination of mutability and plasticity in the skull with extreme conservatism in the remainder of the bony structure is an exception to the usual mode of development, though something of the same sort has already been pointed out in the case of the tapirs (p. 325) and will recur in that of the elephants (Chap. X).

5. †Agriochœridæ. †Agriochœrids

This family, one of the strangest and most aberrant of ungulate groups, was very closely allied to the †oreodonts and by many authorities is included in the same family. The history of the successive steps of discovery, by which the structure of these extraordinary animals was gradually made plain, is much the same as in the case of the even more peculiar perissodactyl family of the †chalicotheres (p. 356). The various parts, found scattered and at long intervals of time, had been referred to no less than three different mammalian orders! for, until the discovery of †chalicothere skeletons gave the clue, no one imagined that such discordant parts could belong to the same animal.

The †agriochœrids had a very much shorter career than the allied family of the †oreodonts, extending only through the upper Eocene and the Oligocene (Uinta to John Day, inclusive); and only two genera of the family are yet known, †Agriochœrus of the John Day and White River, and †Protagriochœrus of the Uinta. In the former the teeth were not in a continuous, closely crowded series, but there were open spaces behind the upper canine and first lower premolar; the same exceptional character of the lower teeth which was found in the †oreodonts was repeated in the present family, the canine assuming the form and functions of an incisor and the first premolar those of the canine; the upper incisors were extremely small and were shed in the adult, just as in the true ruminants. The molars had the selenodont pattern, but the upper molars were very different in shape from those of the †oreodonts, resembling rather those of the †anthracothere †Bothriodon (see p. 370). Another difference from the †oreodont dentition was that the last lower premolar had acquired the molar form and the last upper one nearly so, a very unusual feature among the artiodactyls. The skull was almost exactly like that of the White River †oreodonts, save in a few details; the face was somewhat longer, the orbit was open behind and there was no glandular pit on the face in front of the eye. The neck was short and the body long, and the backbone in the region of the loins very stout, the vertebræ of this region having much resemblance to those of the great cats, as though †Agriochœrus were an agile and powerful leaper. Another likeness to the cats was in the very long and heavy tail, which was much longer than in the †oreodonts, and its vertebræ were hardly distinguishable from those of a Leopard. The limbs were relatively longer than those of the †oreodonts and the separate bones had a suggestive likeness to those of carnivores, and, more specifically, of cats. The feet, save in one particular, were not only artiodactyl, but also characteristically †oreodont in structure and, as in the earlier members of that family, there were five toes in the manus and four in the pes. The exception was that, instead of narrow and slender hoofs, the feet were armed with sharp, though not very large claws, which were not comparable in relative size to the great claws of the †chalicotheres.

Altogether, a strange jumble of incongruous characters was united in this skeleton. Were only the skeleton known without the skull, one would be tempted to call it that of a carnivorous artiodactyl, but the teeth make such a suggestion absurd, since they could have been used only for masticating a diet of soft vegetable substances. No flesh-eater has, or ever had, teeth in the remotest degree like these, which were of characteristically herbivorous type. How such a creature lived and what were its habits, are questions to which no satisfactory answer has been found.

†Protagriochœrus of the upper Eocene is, unfortunately, known only from very imperfect and fragmentary specimens, which, however, are sufficient to determine some significant points. These remains show that, while the two families of the †agriochœrids and the †oreodonts were already distinct in the Uinta, they were decidedly nearer together than they became in the Oligocene. In other words, it is clear that the two groups were converging back to a common ancestry. This may be discovered in the Bridger, but it seems more probable that these forms were immigrants. Another fact concerning the Uinta genus, which is important, is that the upper molars possessed the fifth or unpaired cusp which also occurred in the contemporary †oreodonts, as well as in the †anthracotheres and other Old World families.

Suborder Tylopoda. Camels and Camel-like Animals

Existing Tylopoda are all included in a single family, the Camelidæ, and by several authorities no other family, even of extinct forms, is admitted to the suborder. My own preference, however, is to refer the problematical little †hypertragulids to this group, as will be shown subsequently.

6. Camelidæ. Camels and Llamas

Under modern conditions, no mammals could seem more completely foreign to North America than those of the camel family, which, now restricted to two well-defined genera, inhabit central Asia and the colder parts of South America. Yet, as a matter of fact, this family passed through nearly the whole of its development in North America and did not emigrate to the other continents before the late Miocene or early Pliocene, and it is this North American origin of the family which explains its otherwise inexplicable distribution at the present time. To all appearances, the whole family had completely disappeared from this continent in the later Pleistocene, but in the middle and earlier portions of that epoch both true camels and large llama-like animals were very abundant on the Great Plains and in California, while they seem to have avoided the forested regions.

In order to appreciate the changes through which the camels and llamas have passed, it will be necessary to consider briefly the skeletal and dental structure which characterizes the modern genera. In the true camels (Camelus) the first and second upper incisors have been lost, but the third remains as a large, sharp-pointed tooth, as are also the upper canine and first premolar; thus there are three pointed, spike-like teeth in a row, with spaces between them, constituting with the lower canine a very effective lacerating apparatus. Behind the first premolar is a long gap, the second being suppressed; the third and fourth are grinding teeth, but unusually small. The molars are selenodont and high-crowned, though not extremely hypsodont. The lower incisors are large and shovel-shaped, the canine large and erect and there are but two lower premolars. The dental formula thus is: i 1/3, c 1/1, p 3/2, m 3/3.

The skull is long, with the facial region much and abruptly narrowed, which gives a triangular appearance to the head when seen from above; the orbit is completely encircled with bone and the sagittal and occipital crests are very prominent. The tympanic bullæ are large and filled with spongy bone. The condyle of the lower jaw is hemispherical and not, as it is in most ungulates, semicylindrical, and a curious, hook-like angulation is on the posterior border of the bone. The neck is very long, and the vertebræ have the exceptional peculiarity that the canal for the vertebral artery runs through the side of the neural arch, instead of perforating the transverse process, and thus is invisible externally; the odontoid process of the axis is spout-like. The legs and feet are very long; the humerus has a double bicipital groove and the fore-arm bones are coössified, and the ulna is so reduced that the radius carries the whole weight; in the lower hind leg the tibia supports the weight, and of the fibula only the lower end remains as the malleolar bone. There are but two digits in each foot, the third and fourth, the metapodials of which have coalesced to form a cannon-bone, which differs from that of the true ruminants, or Pecora, in the curious way in which the lower ends, separated by a Λ-shaped notch, diverge from each other, and by the fact that the keels of the lower articular surfaces are confined to the posterior side, not visible from the front. The ungual phalanges are small and nodular, and the hoofs, which carry no part of the weight, are hardly more than nails. Under the other phalanges is a broad pad of elastic tissue, upon which the weight rests, and the separation of the toes is very partial. The peculiar external appearance of the camels is largely due to structures which leave no trace in the skeleton, and especially to the great humps, one or two according to the species, which are accumulations of fat; the ears are short and rounded and the hair is not woolly, but almost straight.

The teeth and skeleton of the llamas (Lama) are closely similar to those of the camels, but the absence of humps, the long, pointed ears, the woolly hair and the much smaller size and lighter build give to the living animals a more marked difference of appearance from the camels than one would expect from a comparison of the skeletons alone. The dental formula is: i 1/3, c 1/1, p 2/2, m 3/3. The remaining upper incisor, the third, is recurved, as is also the canine, but the spike-shaped first premolar of the camels is absent and the other premolars are much smaller than in the latter. In the skull the brain-case is larger, and the sagittal and occipital crests are much less prominent. The skeleton differs hardly at all from that of the camels, except for its smaller size and more slender proportions. The toes are more distinctly separated, each having its own pad. Thus, among the existing representatives of the family are two very well-defined phyla, each characteristic of a different continent.

The Blanco stage of the middle Pliocene, which has preserved but a meagre representation of the life of its time, has yielded a number of very large, llama-like species, not, however, ancestral to the modern species, for they had but one premolar in each jaw. From the lower Pliocene we have fuller information. In the Snake Creek stage the separation of the two modern phyla was complete, and there was a third one, now extinct, that of the browsing or “†giraffe-camels” (†Alticamelus) (see Fig. 127, p. 236), a term which must not be taken as implying any relationship with the giraffes, but merely a resemblance to them in proportions. These browsing camels were very large animals, but with relatively small heads and low-crowned teeth not suited for grazing; the neck was extremely long, made so by the great elongation of five of the vertebræ (second to sixth, inclusive), and the legs were also very long, fitting their possessors to browse upon trees. Much of the description of the appearance and habits of the Giraffe given by Flower and Lydekker would no doubt be applicable to these extinct camels. “To produce the extremely elongated neck the seven cervical vertebræ are proportionately long, which gives a somewhat stiff and awkward motion to the neck.... The Giraffe feeds almost exclusively on the foliage of trees ... for browsing on which its prehensile tongue and large free lips are specially adapted.”[8]

In teeth and skeleton the phyla of the true camels and of the llamas in the lower Pliocene did not differ very strongly from the living forms; the upper incisors were already reduced to one, but the premolars were not so small; the ulna and radius had coalesced and of the fibula only the lower end remained; the cannon-bones were completely formed, and that the pads of the feet had already been developed is shown by the phalanges, especially the irregular, nodular unguals.

The most ancient known camels of the Old World are found in the Pliocene of India, and the first llamas recorded in South America are also Pliocene. Since both camels and llamas existed together in North America, it may reasonably be asked why only one phylum migrated to Asia and only the other to South America. Why did not each continent receive migrants of both kinds? Without knowing more than we are ever likely to learn about the details of these migrations, it will not be possible to answer these questions, though plausible solutions of the problem suggest themselves. It is to be noted, in the first place, that a migration from the central portion of North America to Asia was by way of the far north and thus involved very different climatic conditions from those which must have been encountered in passing through the tropics to South America. It is perfectly possible that animals which lived together in temperate North America should have had very different powers of adaptation to heat and cold respectively, and the northern route may have been impassable to one and the southern route to the other. To this it might perhaps be objected that the llamas are cold-country animals, but this is true only of the existing species, for fossil forms are found abundantly in the Pleistocene of Ecuador, Brazil and Argentina. Another possibility is that both phyla did actually migrate to both continents and that only the camels succeeded in permanently establishing themselves in Asia and only the llamas in South America, though for this solution the fossils afford no evidence.

The camels of the upper Miocene did not differ sufficiently from those of the lower Pliocene to call for special notice other than to remark that the two phyla of the true camels and the llamas were hardly distinguishable and one genus (†Procamelus) may have been ancestral to both. In the middle Miocene the browsing camels (†Alticamelus) reached the acme of their importance and made no great progress subsequently. The generalized stock, from which the upper Miocene and lower Pliocene †Procamelus descended, was represented by †Protolabis and †Miolabis, smaller animals, which had a full set of upper incisors and premolars and the grinding teeth were not so high-crowned. In most of the species the metapodials had not fused to form cannon-bones and probably there were no pads on the feet, though †Alticamelus, the †Giraffe-Camel, had already developed both cannon-bones and pads.

In the lower Miocene the †giraffe-camels were represented by the genus †Oxydactylus, which was a considerably smaller animal than its successor †Alticamelus, of the middle Miocene and later formations, and had shorter neck and legs. The teeth, though brachyodont, were not very low-crowned. There was no cannon-bone, the two metapodials of each foot remaining separate. An especially noteworthy feature in this genus is to be observed in the character of the hoofs, which, as the ungual phalanges demonstrate, were narrow and pointed, like those of antelope and deer, and carried most of the weight. The member of the grazing series (†Protomeryx) was smaller in every way than its contemporary (†Oxydactylus) of the browsing line and had shorter neck and legs, though these were already long. The teeth were present in undiminished number, and the grinders, while not properly to be called hypsodont, showed a decided tendency to assume that character. The feet were in the same stage of development as in †Oxydactylus, that is to say, with two free digits and pointed, deer-like hoofs. We have thus the remarkable and most significant fact that, while the grazing and browsing camels of the lower Miocene were already distinctly separated, neither had yet attained to the type of foot-structure which both of them afterwards independently acquired. This is a very instructive example of parallel evolution in closely related series.

Of still another phylum of the camel family, the lower Miocene contains the only representatives yet discovered, the little “†gazelle-camels,” as they may be called. The single known genus (†Stenomylus, Fig. 131, p. 242) of this series was quite a small animal, much smaller than its contemporaries of the grazing or browsing series. †Stenomylus was an extremely slender, cursorial creature and had a very exceptional feature in its dentition in the apparent presence of ten lower incisors, five on each side, the canine and first premolar having assumed the form and functions of the incisors; the molars were low-crowned. The head was rather small and rounded, the neck long and light, the limbs and feet elongate and excessively slender. The feet had two digits each, which were separate, not forming a cannon-bone, and the hoofs were narrow, pointed and deer-like. These delicate and graceful little animals had but a brief career, which seems to have reached its close in the lower Miocene. Perhaps their complete defencelessness made it impossible for them to maintain themselves against their enemies, despite their evident capacity for swift running.

The camels of the upper Oligocene (John Day) are still incompletely known, but appear all to have belonged to the series of grazers which led up to the modern genera. Future discovery may bring to light in the John Day earlier members of the †giraffe-camel series, of which a possible member is found in the uppermost substage of the White River, or perhaps both phyla united in the upper Oligocene, a question which remains to be determined. At all events, in the middle substage of the White River, or lower Oligocene, there is no evidence of more than a single phylum, from which the others were almost certainly derived, branching off from the main stem at different levels. First was given off the branch of the †giraffe-camels, then (or perhaps even earlier) that of the little †gazelle-camels, and, finally, the main stem bifurcated into the two phyla of the llamas and the true camels. The point of origin of the †gazelle-camels is still uncertain.

The typical White River genus (†Poëbrotherium) included a series of species which increased in size from the earlier to the later portions of the stage, but showed no such structural changes as to call for special notice. The larger of these species was somewhat taller than a sheep, but of much lighter proportions, with small, pointed head, long neck and body and long, very slender limbs and feet. The teeth were present in undiminished number, 44 in all; the lower incisors were small, simple, nearly erect and chisel-shaped, very different from the large, procumbent and shovel-like teeth of the modern genera, and the trenchant canines were much smaller than in the latter. The first premolar had an isolated position, the second and third were trenchant and much extended antero-posteriorly, quite as in many other groups of primitive artiodactyls. The molars, which were typically selenodont, were low-crowned in the upper jaw, but in the lower showed an incipient tendency to hypsodontism. The skull, by its shape and the characteristic narrowing of the face, immediately suggests the modern type, but differed in many details of structure, the most obvious of which were the incompletely closed orbits, the shallow and slender jaws, and the very large, hook-like process from the angle of the lower jaw, which, in greatly reduced form, is present in both of the Recent genera. The neck was relatively long, though by no means so long proportionately as it subsequently became, and the vertebræ had already acquired the peculiarity found in all the succeeding camels, of the exceptional position of the canal for the vertebral artery, save in the sixth vertebra, where it pierced the transverse process, as in mammals generally; the odontoid process of the axis was neither spout-like nor peg-like, but of intermediate form, convex below and flat above. The body was long and light, and the ribs were much more slender than in the Recent genera. The fore and hind limbs, which were of nearly equal length, were very slender; the humerus had a single bicipital groove; the fore-arm bones were fully coössified and in the lower leg only the two ends of the fibula remained. The feet were already in the stage of development which persisted through the lower Miocene in all of the phyla, with two separate digits and nodular remnants of two others, and deer-like hoofs.

It would be of interest to compare this little White River camel with its contemporary genus of horses, †Mesohippus, and to observe in how many respects they have followed a parallel course, and how nearly †Poëbrotherium occupied the same position with reference to the modern camels and llamas as †Mesohippus did to the Recent horses; but such a comparison would involve too many technicalities to be profitably undertaken here. Suffice it to say that in many details there was a genuine parallelism in the progress of these two widely separated families from a polydactyl ancestry towards an extreme of digital reduction, ending in the horses in the single-toed and in the camels in the two-toed foot. The members of the two series kept nearly equal pace in their slow progress, with the camels a little in advance, since they were the first to attain the modern state of development in the height of the teeth and the structure of the feet, though eventually the horses surpassed them in both respects.

In the upper Eocene (Uinta stage) there were at least two kinds of camels, the time-relations of which to each other are not known, that is, whether they were contemporary or successive. The best-known genus, †Protylopus, may perhaps not be in the direct line of camel descent, but it so nearly represents the proper ancestral stage that, for all practical purposes, it will serve nearly as well. It was a much smaller animal than the smallest of the White River species, and was hardly larger than a “jack-rabbit.” The teeth of each jaw were in continuous series and the canines were but slightly longer than the incisors; the premolars had less antero-posterior extension than in †Poëbrotherium, and all the molars above and below were very low-crowned. The skull was almost a miniature copy of that of †Poëbrotherium, but more primitive in a number of details, the most important of which was that the tympanic bullæ were much smaller and hollow, not filled with spongy bone. The neck, concerning which it would be very desirable to have information, is almost the only part of the skeleton that is not known. The fore limb was considerably shorter than the hind, making the back slope downward from the rump to the shoulders; in the fore-arm the two bones were entirely separate and in the lower leg the fibula, though very slender, was still complete. In the manus there were four functional digits, the laterals not very much smaller than the median pair; but in the pes the lateral metatarsals were reduced to mere bony threads, to which small phalanges, in full complement, were attached, making tiny dew-claws.

With †Protylopus ends the genealogy of the camels so far as it can be definitively traced, but in the middle of the Bridger stage is found a genus, †Homacodon (family †Dichobunidæ), which is a probable member of the series. However, until the connecting link can be found in the upper Bridger, this conclusion cannot be demonstrated and †Homacodon itself is incompletely known. It was a very small animal, even less in size than †Protylopus, and had not yet acquired the selenodont molars. These teeth were quadritubercular, i.e. with four principal cusps arranged, in the upper molars, in a square, and with a minute cuspule between each transverse pair, while the lower molars were narrower and had only the four principal cusps. The cusps were not conical, as they are in the pigs, but angular and pyramidal, the first step toward the assumption of the selenodont form. The skull was not specifically cameline in appearance, but rather indifferent, as though almost any kind of an artiodactyl might have been derived from it. The feet were decidedly more primitive than those of the Uinta genus, having four functional digits each, perhaps five in the manus. While it cannot be positively stated that †Homacodon was the actual ancestor of †Protylopus, it nearly represents what we should expect that ancestor to be.

In the lower Eocene (Wasatch stage) lived a tiny creature, †Trigonolestes (family †Trigonolestidæ), smaller even than †Homacodon of the Bridger, and one of the most ancient and primitive of known artiodactyls, but, unfortunately, still represented only by very imperfect specimens, so that much which it would be highly desirable to learn must await the finding of better material. The upper molars were triangular and tritubercular, i.e. with three principal cusps arranged in a triangle, and are hardly to be distinguished from those of other early mammalian orders. From the teeth alone the artiodactyl nature of the animal would not have been suspected, and, in fact, they were, when first discovered, referred to primitive monkeys. The feet probably had five toes each, but this is not certain, and the femur had the third trochanter, the only known artiodactyl of which this is true. As this little Wasatch genus is so imperfectly known, it would be premature to claim it as the starting point of the camel family, and yet it may very well have been so. Better material of this genus and the links of the chain which belong in the upper Bridger and the Wind River respectively must be recovered before this earliest portion of the family history can be written in more than tentative fashion.

The mode of evolution displayed by the camels does not differ in any significant respect from that seen in the horses. There was the same increase in bodily stature and in the relative lengths of the limbs and feet, the same kind of diminution in the number of digits from the original five, the same reduction of the ulna and its coalescence with the radius and the loss of the fibula save for its two ends. There was also a similar development of the high-crowned, or hypsodont, grinding teeth, from the low-crowned, or brachyodont, type. In still another respect there was a similarity in the mode of development of the two families, namely, in the way in which the several phyla of each originated. For the earlier portion of their history there was in each but a single distinguishable series, though it is very possible that fuller knowledge and more complete material would enable us to distinguish more than one. This monophyletic condition continued through the Eocene and most of the Oligocene, but in the upper portion of the latter and, more markedly in the lower Miocene, the two families branched out, each in its own fashion.

Of course, there were differences in the development of the camels and horses, some conditioned by the fundamental distinction between artiodactyl and perissodactyl, such as the didactyl foot as the possible minimum and the formation of cannon-bones in the camels. Other differences are characteristic of the latter family, such as the great elongation of the neck and the peculiar structure of its vertebræ, the formation of pads on the feet and concomitant reduction of the hoofs. In a general way, the two families kept quite an even pace in their advance from the more primitive to the more specialized condition and, though the camels were the first to acquire certain modifications, the horses ultimately surpassed them.

Even more close was the parallelism in evolution between the camels and the true ruminants (suborder Pecora), and this case is of particular importance as clearly demonstrating the development, in two independent but related lines, of similar structures not derived from a common ancestry. This comparison must naturally await the description of the Pecora.

7. †Hypertragulidæ. †Hypertragulids

This was a very peculiar family, of exclusively North American distribution and of doubtful systematic position, the known history of which extended from the upper Eocene into the lowest Miocene and then abruptly terminated. None of its members attained to considerable size, the largest hardly surpassing a sheep, and some were extremely small. In view of its comparatively brief career, this family was surprisingly ramified, and no less than four phyla may be distinguished within its limits.

One of the phyla which persisted into the lower Miocene was there represented by a most fantastic creature (†Syndyoceras) with four horn-like outgrowths from the skull, one pair arising from the anterior part of the face and curving outward away from each other, and the hinder pair, which were placed over the eyes, curved toward each other at the tips and were shaped much like a cow’s horns in miniature. The shape of these bony protuberances makes it unlikely that they were sheathed in horn and probably they were merely covered with skin like the horns of the giraffes. This description applies only to the skull of the male; that of the female is not yet known, but there is good reason to believe that in that sex the horns were much smaller or wanting, as in nearly all existing deer. The skull was long, narrow and low; the orbits were small, completely enclosed in bone and unusually prominent; the nasal bones were exceedingly short, as though indicating the existence of a proboscis, but this can hardly have been the case, for the nasal opening was divided into anterior and posterior portions by the bony bridge which united the bases of the forward pair of horns. In no other known mammal does such a division of the nasal opening occur. The upper incisors had all disappeared, but there was a small upper canine tusk and another formed by the first lower premolar, while the real lower canine had gone over to the incisor series. This exceptional arrangement is a point of resemblance to the †oreodonts (see p. 372). The grinding teeth were brachyodont. The fore limb is not known, but the hind limb has been completely recovered; it was stout and not very long in proportion to the length of the head. The fibula was completely reduced, only the ends remaining, and the pes was didactyl, the two metatarsals uniting in a cannon-bone; the hoofs were like those of deer and antelopes.

No representative of this series has yet been found in the upper Oligocene; and it is not yet possible to say whether their absence from the John Day beds, as in several other cases already referred to, was due to an actual geographical difference in contemporary faunas, or whether it is merely one of the accidents of preservation and collecting. In the upper White River, however, was another most curious animal (†Protoceras), a forerunner, if not a direct ancestor, of †Syndyoceras. The exact relationship between the two forms can hardly be determined, until the genera, one or more, which once connected them shall have been recovered, though it is obvious that they belonged to the same series. †Protoceras was a smaller animal and, if anything, an even more bizarre-looking object, for the anterior protuberances were broad, prominent and everted plates of bone, not even suggesting horns in their form, and the posterior pair were short and club-shaped; in the female neither pair was more than indicated. The dentition was very similar to that of †Syndyoceras, except that the upper tusk was considerably larger and scimitar-shaped; the female had no tusks. In the fore-arm the two bones were just beginning to coalesce, but in the lower leg the fibula was completely reduced. The manus had four complete and functional digits, the laterals not very much shorter and thinner than the median pair; but the pes was already didactyl, though the metatarsals were separate, not fused into a cannon-bone; two long and pointed splints were the vestigial remnants of the second and fifth digits.