CHAPTER XIII

HISTORY OF THE †LITOPTERNA AND †ASTRAPOTHERIA

Besides the four well-defined groups which make up the †Toxodontia (or †Notoungulata) there are two other extinct orders of indigenous South American ungulates, which remain to be considered. These did not have the exceptional development of the auditory region of the skull which characterized the †Toxodontia. The best known and most important genera of the †Litopterna are listed in the following table:

†LITOPTERNA. †Litopterns

I. †MACRAUCHENIDÆ.

_†Macrauchenia_, Plioc. and Pleist. _†Scalibrinitherium_, Paraná. _†Theosodon_, Santa Cruz. _†Cramauchenia_, Patagonian. _†Protheosodon_, Deseado.

II. PROTEROTHERIIDÆ.

_†Epitherium_, Monte Hermoso. _†Diadiaphorus_, Santa Cruz and Paraná. _†Proterotherium_, do. _†Thoatherium_, Santa Cruz. _†Deuterotherium_, Deseado. † _Prothoatherium_, do.

III. DIDOLODIDÆ.

_†Didolodus_, Casa Mayor. _†Lambdaconus_, do. _†Notoprogonia_, do. _†Proectocion_, do., etc., etc.

Only one of the families of this suborder survived into the Pampean stage, where it was represented by a single genus, _†Macrauchenia_. Like all the other large Pampean mammals of distinctly South American type, this was a grotesque creature, from the modern point of view. The genus was first discovered by Darwin, who says of it: “At Port St. Julian, in some red mud capping the gravel on the 90-foot plain, I found half the skeleton of the Macrauchenia Patachonica, a remarkable quadruped, full as large as a camel. It belongs to the same division of the Pachydermata with the rhinoceros, tapir, and palæotherium; but in the structure of the bones of its long neck it shows a clear relation to the camel, or rather to the guanaco and llama.”[12] The views upon classification and relationship here expressed have been superseded, but the passage is an important one in the history of scientific opinion.

_†Macrauchenia_ (Fig. 120, p. 216), as Darwin says, was as large as a camel; it had an unreduced dentition of 44 teeth and in each jaw the teeth were arranged in continuous series and were quite decidedly hypsodont. Both in the upper and the lower jaws the incisors formed a nearly straight transverse row and have a “mark,” or enamel pit, like that seen in the horses; the canines were but little larger than the incisors and did not form tusks. The premolars were smaller and simpler than the molars. The upper molars had two concave and crescentic external cusps, connected by a median ridge, as in several families of perissodactyls; two transverse crests and several accessory spurs and enamel-pockets gave to the grinding surface, when somewhat worn, the appearance of considerable complexity. The lower molars had the two crescents, one behind the other, which recurred in almost all the South American types of ungulates; the vertical pillar which so generally in these types arose in the inner concavity of the posterior crescent was wanting in the permanent teeth of _†Macrauchenia_, but present in the milk-premolars.

No part of this remarkable animal was more curious than the skull, which was quite small in proportion to the rest of the skeleton. It was long, narrow and low, sloping and tapering forward to a blunt point at the end of the muzzle, though there was a slight broadening here to accommodate the transverse row of incisors. The sagittal crest was replaced by a short, narrow and flat area; the cranium was shortened and the face elongated, the orbits, which were completely encircled in bone, having been shifted behind the line of the teeth, as in the modern horses. The nasal bones were reduced to a minimum, a mere vestige of their original length, the anterior nasal opening being directly over the posterior, making the nasal passage vertical. Such an arrangement is an almost positive proof that in life the animal had a flexible proboscis, a conclusion which is confirmed by the presence, on the top of the head and behind the nasal opening, of deep pits for the attachment of the proboscis-muscles. A very curious feature of this skull was that the bones of the upper jaw, the maxillaries and premaxillaries of the opposite sides, united in the median line, making a long, solid, bony rostrum in front of the nasal opening, a character not found in other land mammals.

The neck was almost as long as in a camel and its vertebræ agreed with those of the latter in the very exceptional character of having the canal for the vertebral artery passing longitudinally through the neural arch, instead of perforating the transverse process. As Darwin says in the passage quoted above, “it shows a clear relation to the ... guanaco and llama,” but this is founded on the postulate that such a likeness must, of necessity, imply relationship. As was shown in the chapters on the Artiodactyla and Perissodactyla, it is the general rule among long-necked ungulates that the odontoid process of the axis assumes a spout-like shape, but _†Macrauchenia_ was an exception and had an odontoid which retained its primitive and peg-like shape; it was, however, relatively very short and in cross-section was no longer circular, but oval. This may be regarded as a step toward the assumption of the spout-like form, but the extinction of the family put an end to further changes in that direction.

The body was rather short and the limbs very long, giving the animal a stilted appearance, while the feet were relatively short. The proportionate lengths of the different limb-segments was unusual; the upper arm was short, the fore-arm very long, the thigh long and the lower leg quite short. The humerus was very heavy; the ulna and radius, which were firmly coössified, formed a very long compound bone, which was broad transversely and thin antero-posteriorly. The long femur had only a small and inconspicuous third trochanter and the shaft was broad and thin, being flattened, or “compressed” antero-posteriorly. The tibia and fibula were united at both ends; the former was very heavy at the upper end, but diminished downward in width and thickness, and the fibula articulated with the calcaneum, as in the artiodactyls. The feet were tridactyl and had mesaxonic symmetry; that is to say, the median digit, or third of the original five, was symmetrical in itself and was bisected by the middle line of the foot, while the lateral toes (second and fourth), each of which was asymmetrical, formed a symmetrical pair. It is this perissodactyl character of the foot to which Darwin refers when he says that _†Macrauchenia_ “belongs to the same division of the Pachydermata with the rhinoceros, tapir and palæotherium.” On the other hand, the very significant structure of the ankle-joint was radically different from that of the Perissodactyla; not only did the calcaneum have a special facet for articulation with the fibula, but the lower end of the astragalus was a convex “head,” resting only on the navicular, as in the †Toxodontia, †Condylarthra, Hyracoidea and other very primitive groups of hoofed animals and in clawed mammals generally. Such a combination of characters is not known in any of the perissodactyls and precludes the reference of the †Litopterna to that order, though such a reference is strongly maintained by several authorities. The ungual phalanges were small and appear to suggest the presence of pads on the feet.

The appearance of _†Macrauchenia_ in life must have been sufficiently strange. The small head with its proboscis and the long neck and legs should probably be regarded as indicative of browsing habits, though the hypsodont teeth show that grazing was at least an occasional mode of feeding. The long limbs and short feet gave to the extremities an appearance unlike that of any existing hoofed animal. The form and size of the ears and the character of the hairy coat are, of course, conjectural.

In the later Pliocene the family was represented by forms which differed so little from the Pampean _†Macrauchenia_ as to call for no particular notice, but in the presumably lower Pliocene of the Paraná stage, occurred several genera, all unfortunately but imperfectly known, which are of interest as being less specialized than _†Macrauchenia_ and as showing the way in which some of the peculiarities of the latter were acquired. In _†Scalibrinitherium_, which may be taken as an example of these genera, the teeth were brachyodont; the upper molars were rather less complex than those of _†Macrauchenia_, while the lower molars had the pillar in the concavity of the posterior crescent, which the Pampean genus retained only in the milk-teeth. As we have repeatedly found, the milk-dentition is often conservative and retains primitive or archaic features which have been lost in the permanent teeth, and _†Macrauchenia_ is another illustration of the same principle. In the skull of _†Scalibrinitherium_ the nasal bones, though very short, had not suffered such extreme abbreviation as in the succeeding genus, the nasal opening was farther forward and the maxillaries united in the superior median line for only a short distance, while the premaxillaries were fused together for their whole length. The orbit had not been shifted entirely behind the teeth, but was above the third upper molar.

[Illustration: FIG. 244.—Santa Cruz †macrauchenid (_†Theosodon garrettorum_) and predaceous marsupial (_†Borhyæna tuberata_). Restored by C. Knight from skeletons in the museum of Princeton University.]

Next in the ascending series, to use the genealogist’s term, came the genus _†Theosodon_ of the Santa Cruz, of which almost all the skeletal parts are known and thus make possible a full comparison with _†Macrauchenia_, which assuredly was its direct descendant. In view of the great lapse of time involved, the differences between the two genera were less than might have been expected, though the more ancient animal was in all respects the more primitive. _†Theosodon_ was, in the first place, considerably smaller, not much exceeding a llama in size; the teeth had lower crowns than even those of _†Scalibrinitherium_ and the incisors were arranged in line with the grinding teeth, not in a transverse row, but curving inward slightly, so that those of the opposite sides nearly met in front. The incisors, canine and first premolar were simple, sharply pointed, conical teeth, which gave an almost reptilian expression to the anterior part of the skull. The upper molars were on the same fundamental plan as those of _†Macrauchenia_, but in a less advanced stage of development, the transverse crests being incomplete and the internal cusps had a certain degree of separateness from the crests and from each other. It is evident that the upper molars were derived from the quadritubercular type. The lower molars had the vertical pillar in the concavity of the posterior crescent very prominently developed.

The resemblance of the skull to that of _†Macrauchenia_ is obvious at the first glance, but it was less specialized and departed less from the ordinary ungulate type. The cranium was longer and the face shorter, the orbit, which was incompletely closed behind, extending over the second molar. There was a sagittal crest, the length of which differed much in the various species; the nasal bones were already very short, though decidedly longer than in the subsequent genus _†Scalibrinitherium_, and the anterior nasal opening was extended forward as a long, narrow slit, because the maxillaries did not come into contact with each other in the superior median line, and the premaxillaries touched each other, but were not coössified. The nasal canal, though very short, was horizontal, not vertical. The skulls of the three genera thus displayed three successive stages in the backward shifting of the orbit and of the anterior nasal opening, in the shortening of the nasal bones and in the formation of a solid rostrum by the fusion of the upper jaw-bones. No doubt also the living animals exhibited a corresponding gradation in the development of the proboscis.

[Illustration: FIG. 245.—Development of the skull in the †Macrauchenidæ, side views. _A_, _†Theosodon_, Santa Cruz. _B_, _†Scalibrinitherium_, Paraná. (After Ameghino.) _C_, _†Macrauchenia_, Pampean. (After Burmeister.) _n._, nasal bones.]

[Illustration: FIG. 246.—Development of the skull in the †Macrauchenidæ. _A_, _†Theosodon_. _B_, _†Scalibrinitherium_. (After Ameghino.) _C_, _†Macrauchenia_. (After Burmeister.)]

The neck of _†Theosodon_ was even longer proportionately than in _†Macrauchenia_ and the transference of the canal for the vertebral artery from the transverse processes to the neural arch had already taken place, except in the first, sixth and seventh vertebræ, and was thus less complete than in the Pampean genus, in which all the vertebræ of the neck, save the seventh, had the canal in its exceptional position. The odontoid process of the axis was less modified than in the latter, being relatively longer and more conical. The body was rather short, and the spines of the trunk-vertebræ were proportionally higher and more prominent. No caudal vertebræ have been found, but, from the shape of the sacrum, it is evident that the tail was short.

The limbs were long, but more slender and less elongate than in _†Macrauchenia_, in which the growth of the neck did not keep pace with that of the limbs, the lengthening of the proboscis probably compensating for this. The shoulder-blade had two conspicuous metacromia, very much as in the contemporary †toxodont, _†Nesodon_, but shorter and more widely separated. The humerus was short and quite slender and the fore-arm bones, which were much longer, did not coössify. The femur had a more slender and rounded shaft than in _†Macrauchenia_ and a much larger third trochanter; the leg-bones were also separate from each other. The tridactyl feet were so like those of the Pampean genus, that no particular account of them is necessary, and the proportions of the limb segments were similar in both genera, short upper arm and lower leg, very long fore-arm and thigh, and short feet.

[Illustration: FIG. 247. Left manus of _†Theosodon_. _S._, scaphoid. _L._, lunar. _Py._, pyramidal. _Tm._, trapezium. _Td._, trapezoid. _M._, magnum. _U._, unciform. _V._, rudimentary fifth metacarpal.]

The appearance of the living animal, as shown in the restoration, was no doubt somewhat like that of _†Macrauchenia_, but less bizarre. That there must have been some sort of a proboscis or prehensile upper lip, is indicated by the greatly shortened nasal bones, but this may not have been longer than in the existing Moose or Saiga Antelope. The long neck, short body and tail and long limbs suggest an animal not unlike a Guanaco, but larger and heavier. The hair may or may not have had the woolly character given to it in the drawing; upon such a point there can be no certainty.

In the older formations preceding the Santa Cruz, the †macrauchenids are known only from fragmentary material, though something of their history may be made out even from these fragments. _†Protheosodon_, of the Deseado stage, was considerably smaller than the Santa Cruz genus and had more primitive upper molars, in that the internal cusps and intermediate cuspules were isolated and conical, not forming transverse crests. Still smaller were the several genera (_†Lambdaconus_, etc.) related to the †macrauchenids found in the Casa Mayor Eocene, which have been referred, perhaps correctly, to the †Condylarthra. In these the formation of the external wall of the almost bunodont upper molars was in progress, by the fore-and-aft extension and transverse thinning of the external cusps; the internal pair of cusps and the cuspules were separate and conical. With much confidence, it may be inferred that in these little animals the skull was normal, the nasal bones were long and that the feet were five-toed, but demonstration is lacking.

* * * * *

The second family of the †Litopterna, the †Proterotheriidæ, were remarkable for their many deceptive resemblances to the horses. Even though those who contend that the †Litopterna should be included in the Perissodactyla should prove to be in the right, there can be no doubt that the †proterotheres were not closely related to the horses, but formed a most striking illustration of the independent acquisition of similar characters through parallel or convergent development. The family was not represented in the Pleistocene, having died out before that epoch, and the latest known members of it lived in the upper Pliocene of Monte Hermoso. In the still older Paraná formation more numerous and varied forms occurred, but only from the Santa Cruz have materials been obtained of sufficient completeness to furnish a full account of the structure of these extraordinary animals. Not that this remarkable character was due to grotesque proportions; on the contrary, they looked far more like the ordinary ungulates of the northern hemisphere than did any of their South American contemporaries; it is precisely this resemblance that is so notable.

In Santa Cruz times the family was represented by a large number of species, which have been grouped in four or five genera, which differed sufficiently to require generic separation, yet were closely similar. In all of them the dental formula was: _i_ 1/2, _c_ 0/1, _p_ 4/4, _m_ 3/3, × 2 = 36. Except in one genus (_†Thoatherium_) a pair of small tusks was formed by the enlargement of the second upper and third lower incisors, as in the †toxodonts, but the first upper and lower and the third upper incisors, which were retained in the †toxodonts, were lost in this family, as was also the upper canine, and the lower canine was very small, of no functional use. The teeth were brachyodont and, except the small tusks, displayed no tendency at any time toward the acquisition of high crowns. The premolars were less complex than the molars, though the last one approximated the molar-pattern. The upper molars had two crescentic outer cusps, meeting in a vertical ridge and together forming the outer wall; the transverse crests were imperfect, especially the hinder one which was often merely the intermediate cuspule, and did not fuse with the external wall. The lower molars had the two crescents, one behind the other, which recur in the †macrauchenids, all the suborders of the †Toxodontia, except the †Pyrotheria, and other South American ungulates, but the pillar in the posterior crescent, which was so characteristic of the groups named, was reduced to very small proportions and sometimes suppressed altogether. It should be noted, however, that this was the loss of an element which was formerly present.

The skull had a long cranium and rather short face, with long, high sagittal crest. The neck was short, the odontoid process of the axis peg-shaped, and the canal for the vertebral artery was in its normal position. The body was rather short, like that of a deer or antelope; the number of trunk-vertebræ is not definitely known in any of the genera, but was very probably 19 or 20, and the tail must have been short.

[Illustration: FIG. 248.—Skull of _†Diadiaphorus_, Santa Cruz. American Museum.]

The limbs were slender and of moderate length; there was no coössification between the bones of the fore-arm or the lower leg. The feet were three-toed, except in one genus (_†Thoatherium_) in which they were single-toed, and nearly or quite the whole weight was carried upon the median digit, the laterals being mere dew-claws. The shape of the hoofs and the whole appearance of the foot were surprisingly like those of the three-toed horses, but there were certain structural differences of such great importance as, in my judgment, to forbid the reference of these animals, not merely to the horses, but even to the perissodactyls. In studying the †Litopterna, one is continually surprised to note the persistence of archaic and primitive characters in association with a high degree of specialization.

[Illustration: FIG. 249.—Three-toed †proterothere (_†Diadiaphorus majusculus_), Santa Cruz. Restored by C. Knight, from skeletons in the American Museum and Princeton University.]

The largest Santa Cruz representatives of the family were the species of _†Diadiaphorus_, animals considerably taller than a sheep and of heavier build. Their appearance was not unlike that of a short-necked, hornless antelope, but with the feet of the three-toed horses! These feet were, however, merely superficially like those of the horses, differing in points of fundamental significance. In the horses, the reduction of the digits was accompanied by a readjustment of the carpal and tarsal articulations, so that, in proportion as the median toe was enlarged and the laterals reduced, the weight was shifted more entirely upon the former. This is the method of digital reduction which Kowalevsky called “adaptive” and is exemplified in all existing artiodactyls and perissodactyls and by none more perfectly than by the monodactyl horses. In “inadaptive reduction,” the method followed by _†Diadiaphorus_ and the other genera of this family, there was no readjustment, or a very imperfect one, of the articulations, the lateral digits, however small and rudimentary, retaining the connections which they had when they were of full size and function. This distinction may seem to be unimportant, but its significance is shown by the fact that not a single ungulate with inadaptively reduced feet has survived to the present time.

[Illustration: FIG. 250.—Left pes of _†Diadiaphorus_, from specimens in Princeton University and the American Museum. _Cal._, calcaneum. _As._, astragalus. _N._, navicular. _Cn. 3_, external cuneiform. _Cb._, cuboid.]

In still another respect the feet of _†Diadiaphorus_ deviated markedly from those of the horses, viz. in the great proportionate length of the phalanges, especially of the first one, and the shortness of the metapodials, the three phalanges of the median digit together exceeding in length the metacarpal or metatarsal, while in the horses this proportion is reversed. The skull of this genus was short, deep and with an anterior taper; it had a long sagittal crest, but a brain-chamber of good capacity, considering its geological date. The nasals were quite short, though the degree of shortening was not such as to suggest the existence of a proboscis. In general appearance the skull recalls that of one of the larger †oreodonts (p. 372) of the North American Oligocene.

To the genus _†Proterotherium_, the type of the family, belonged a great number of Santa Cruz species, for at that time the genus was in a state of most vigorous development and the species were so variable that satisfactory discrimination of them is exceedingly difficult. They were all much smaller and slighter animals than the species of _†Diadiaphorus_, but did not differ from them in any important structural character. The skull in this genus closely resembled that of the one last named, save for its smaller size and lighter and more slender proportions; the nasal bones were considerably longer and the occiput was somewhat wider.

[Illustration: FIG. 251.—Skull of _†Thoatherium_, Santa Cruz. Princeton University Museum.]

A more isolated position was held by the genus _†Thoatherium_, which was very clearly demarcated from all of the other genera of the family. Its species were the smallest of the commoner Santa Cruz members of the order and were of very light and graceful form. The dental formula was the same as in the other genera, but there were no tusks; the single upper and two lower incisors were of nearly the same size and simple, chisel-like form. The upper molars had the same elements as in the preceding genera, but somewhat differently connected, the two internal cusps and the anterior intermediate cuspule being united into a nearly longitudinal ridge. The skull was light, slender and pointed; the nasals were shortened, though less than in _†Diadiaphorus_; the sagittal crest was shorter than in the latter and the occiput was far narrower. The neck was short, the body of moderate length and the tail short. The limbs and especially the feet were proportionately more elongate and slender than in any other known genus of the family, giving quite a stilted appearance to the skeleton. The fore-arm bones were not coössified, but the ulna was much more reduced than in any of the other genera of the family, and the same is true of the fibula, which, though very slender, showed no tendency to unite with the tibia. The limb-bones, especially the femur, had a decided resemblance to those of _†Mesohippus_, the lower Oligocene tridactyl horse of North America, with the smaller species of which, _†M. bairdi_, _†Thoatherium_ agreed well in size. Most remarkable of all were the feet, _which were more strictly monodactyl than those of any other known mammal_. The single functional digit, the third, had on each side of its upper end a very small, scale-like nodule of bone, the last vestiges of the lateral digits, corresponding to the immensely larger splints of the horse. Despite the unrivalled completeness of digital reduction which _†Thoatherium_ displayed, the mode of reduction was inadaptive and the rudimentary metapodials retained the same carpal and tarsal connections that they originally had in the pentadactyl manus, a very great difference from the horses. The ankle-joint also was of the same primitive character as in the other †Litopterna. The feet were relatively longer and more slender than in the other †proterotheres and the metapodial of the single functional digit longer in proportion to the phalanges.

[Illustration: FIG. 252.—Single-toed †proterothere (_†Thoatherium minusculum_), Santa Cruz. Restored by C. Knight from a skeleton in the museum of Princeton University.]

The appearance of the living animal, aside from the character of the hair, colour-pattern, etc., may be closely inferred from the skeleton. It was a much smaller and more graceful animal than its contemporary and relative _†Diadiaphorus_, as light and agile as a gazelle. The head had some resemblance to that of a small horse, but the neck was much shorter than in the horses; the body also was shorter than in the latter, and the proportions of the trunk and limbs were quite as in the smaller antelopes. But these likenesses to horses and antelopes were, it must again be emphasized, superficial; the fundamental characteristics of structure were more primitive than in the most ancient known artiodactyls and perissodactyls.

[Illustration: FIG. 253.—Left pes of _†Thoatherium_. Princeton University Museum. Letters as in Fig. 250.]

With the aid of the fragmentary material which alone represents the †proterotheres in the formations preceding and following the Santa Cruz in time, it is not practicable to trace the development of the various phyla in a satisfactory manner. Two of the Santa Cruz genera, _†Diadiaphorus_ and _†Proterotherium_, continued into the lower Pliocene (Paraná), and two additional ones have been named, but little is known about them. The latest known member of the family so far discovered is a genus (_†Epitherium_) from the upper Pliocene of Monte Hermoso, a tridactyl form like _†Diadiaphorus_. It is a noteworthy fact that the most advanced and specialized genus of the entire family ended with the Santa Cruz, while the less differentiated types survived till a considerably later period. Possibly, it was the incoming of the highly efficient Carnivora from North America that led to the extermination of the last †proterotheres.

Turning backward from the Santa Cruz, the family may be traced without any question to the Deseado stage of the Oligocene, though nothing but teeth has yet been obtained, while in the Eocene it would appear to have become merged in the same group of small, †Condylarthra-like animals with quadritubercular molars, as those which are regarded as the probable ancestors of the †macrauchenids. However likely this conclusion may seem to be, its confirmation must await the discovery of much more complete specimens than are now available.

ORDER †ASTRAPOTHERIA. †ASTRAPOTHERES

In the Santa Cruz another group of peculiar South American ungulates, the †Astrapotheria, made its last recorded appearance. Though not at all uncommon in that formation, no complete or even partial skeleton has yet been found, but merely the skull and a few bones of the limbs and feet. For this reason there is much doubt as to the systematic position and relationships of these animals, which were among the most curious of the many strange mammals which made up the Santa Cruz fauna. They were mentioned in connection with the †Amblypoda (p. 456) as possible representatives of that order in South America, but, as will be seen later, this is an improbable conclusion, and the group appears to have been indigenous in the southern continent, in which, at all events, it had a very long history. It has not been found in any formation later than the Santa Cruz, unless the Friasian fauna, which contains it, should be removed from that stage, of which it apparently forms the latest division.

I. †ASTRAPOTHERIIDÆ.

_†Astrapotherium_, Santa Cruz and Patagonian. _†Astrapothericulus_, Patagonian. _†Parastrapotherium_, Deseado. _†Astraponotus_, Astraponotus Beds. _†Albertogaudrya_, Casa Mayor.

II. †TRIGONOSTYLOPIDÆ.

_†Trigonostylops_, Casa Mayor. _†Edvardocopeia_, Astraponotus Beds.

The genus _†Astrapotherium_, which was the only well-defined representative of its family and order in the Santa Cruz stage, contained several species, some of them the largest animals of their time, as well as the most grotesque in appearance. The dentition differed in some important respects from that of all the other South American ungulates, the formula being: _i_ 0/3, _c_ 1/1, _p_ 2/1, _m_ 3/3, × 2 = 28. The upper incisors had completely disappeared, but the lower ones were large and, what was an exceptional character, they were partially divided into two lobes, somewhat as in the Eocene †uintatheres of North America (p. 446). The canines were very large and formidable tusks, which grew throughout life and apparently formed no root; the upper tusk was nearly straight and was obliquely truncated by the strongly curved and sharp-pointed lower tusk. This arrangement was very unusual among South American hoofed mammals, many of which had no tusks at all; and in those which possessed them, such as the †toxodonts (p. 468), they were mostly incisors. Only in the †astrapotheres and †homalodotheres were there canine tusks, and in the latter group they were small and of limited growth. All the teeth, except the canines, were brachyodont and, though rather high-crowned, formed roots before coming into use. The premolars were small and greatly reduced in number (2/1), and in pattern were simpler than the molars. The upper molars were constructed on essentially the same plan as in the †Toxodonta; indeed, the first specimen of this genus collected was referred to a large species of _†Nesodon_ by Owen. On the other hand, the resemblance to the rhinoceros teeth is very decided, and has led several writers to postulate a relationship between the †astrapotheres and the rhinoceroses. The lower molars were of the bicrescentic pattern so frequently met with already; these teeth were very narrow in proportion to their length and strongly suggest those of _†Metamynodon_, the supposedly aquatic rhinoceros of the North American Oligocene (p. 346). It may be confidently inferred that so small a number of premolars was due to reduction from a full series, and this is confirmed by the milk-dentition, in which the premolars were 4/3.

[Illustration: FIG. 254.—Head of _†Astrapotherium magnum_. Santa Cruz. Restored from a skull in the museum of Princeton University.]

The skull was extremely peculiar, more so than in any other of the contemporary genera of hoofed animals. The toothless premaxillaries were quite small, but thick, and must have supported an elastic pad, against which the lower incisors could effectively bite in cropping herbage. The nasal bones were extremely short and there must have been a proboscis or greatly inflated snout, probably the former; the immense development of sinuses in the frontal bones elevated the whole forehead into a great, dome-like convexity, a feature which is not equalled in any other known mammal. The orbits were open behind and the brain chamber was small, so that the sagittal and occipital crests were very high and strong, to afford sufficient surface for the attachment of the great temporal muscles. The horizontal portion of the lower jaw was shallow vertically, but very thick and massive, and the symphyseal region was broad and depressed.

Unfortunately, the skeleton is still very incompletely known. Of the vertebræ, only the atlas and axis have been recovered, and these resembled those of the Santa Cruz †toxodont _†Nesodon_, on a larger scale. The scapula had a very thick spine, without the projections which were found in most of the Santa Cruz ungulates. The limb-bones were long and comparatively slender, and the processes for muscular attachment were singularly small and weak; the bones of the fore-arm and lower leg did not coössify and were proportionately elongate, the tibia being but little shorter than the femur. The latter had the flattened shaft which recurs in nearly all of the very heavy ungulates, but retained a remnant of the third trochanter. If the feet found isolated in the Santa Cruz and Deseado stages have been correctly referred to this order, then the genus was five-toed and the feet were broad, short and heavy, quite elephantine in appearance, especially the fore foot. The ankle-joint was very peculiar and the calcaneum had no articulation with the fibula, which it had in all the other indigenous South American ungulates.

Incomplete as the material is, it is yet possible to form some general conception of this extraordinary animal when in life. The head was short, broad and deep, rounded and very probably furnished with a proboscis; the neck was of moderate length, so that the mouth could not reach the ground without a straddling of the fore legs. The body was no doubt long, the limbs long and rather slender, giving the animal a stilted appearance, the feet very short, broad and columnar. Several species of the genus are known, which differed much in size, the largest (_†A. giganteum_) probably exceeding any modern rhinoceros in height and length, and the smallest (_†A. nanum_) not much larger than a Wild Boar.

_†Astrapothericulus_, of the Patagonian stage, was smaller than the average species of the Santa Cruz genus, and had teeth of the same number, but the canines were not capable of indefinite growth, and the lower molars had the pillar in the posterior crescent so characteristic of the South American hoofed animals. In the Deseado stage, on the contrary, the †astrapotheres were of larger size, and in the commonest genus, _†Parastrapotherium_, the grinding teeth had lower crowns and the premolars were more numerous, at least 3/2. In the still more ancient _†Astraponotus_, which gives its name to the upper Eocene (or lower Oligocene) of Patagonia, the premolars were present in full series. In the Casa Mayor the order was abundantly represented by still more primitive genera, which assuredly had an undiminished number of teeth, though this has not been proved. One of these genera, _†Albertogaudrya_, was the largest animal of its time and the highly probable ancestor of the series leading to the Santa Cruz _†Astrapotherium_.

The second family of the order, the †Trigonostylopidæ, did not survive beyond the Eocene and is so imperfectly known that any account of it would be to small profit.

As stated above, the †Astrapotheria were an isolated group and their relationships are problematical and are likely to remain so pending the discovery of much more complete specimens of the various genera which made up the series. I am inclined to the opinion, however, that all of the indigenous groups of South American ungulates, which inhabited that continent before the great immigration from the north, were derivatives of the same stock and more nearly related to one another than to any of the orders which lived in other regions.

* * * * *

In looking over the labyrinth of ungulate history, as recorded by the fossils, certain facts stand out clearly, while others are still very obscure. It is like trying to trace the plan of vast and complicated ruins, which here are deeply buried in their own débris, there are fully exposed and in another place are swept away so completely that hardly a trace remains. But the problem is far more complex than any which can be presented by buildings, for the factor of repeated migrations from continent to continent comes in to obscure the evidence. Had each of the great land areas received its original stock of early mammals and then been shut off from communication with any other, many of the difficulties would be removed, but the story would lose half its interest.

Within the limits of the family, giving to that group the broad and elastic definition which has hitherto been employed, we have repeatedly found it feasible to construct a phylogenetic series which very nearly represents the steps of structural modification as they occurred in time. Much less frequently is it possible to trace allied families to their common starting point, and, so far as the hoofed animals are concerned, in no case have we yet succeeded in doing this for the separate orders. The obstacle lies in the fact that the ordinal groups were already distinct, when they made their first appearance in the known and accessible records, and the hypothetical ancestors common to them all, or to any two of them, are to be sought in regions of which we know little or nothing. Nevertheless, certain legitimate inferences may be drawn from the available evidence. It remains to be proved whether the assemblage of hoofed mammals, as a whole, was of single or multiple origin. Have all ungulates been derived from a common stock, or did they arise independently from several groups of clawed mammals? While the records cannot be followed back to the point, or points, of origin of the various orders, yet it is a noteworthy fact that, between several of them, the differences grow less marked as the more ancient members are reached, as though they were converging to a common term; others again show little such approximation, and the most probable conclusion from the evidence now at hand is that the ungulate assemblage is composed of several independent series.

One such series is that of the Hyracoidea and Proboscidea, to which Dr. Schlosser has given the name “Subungulata,” and has pointed out its relationship to the †Condylarthra, which, however, is not a close one and may be illusory. Another apparently natural group is that of the peculiarly South American forms, the †Toxodontia, with its four suborders, the †Litopterna and the †Astrapotheria, which all appear to be traceable to closely allied families in the Eocene, whose teeth strongly suggest derivation from the †Condylarthra; but the material does not permit any positive statements. The Artiodactyla and Perissodactyla have so many similarities that they have always been regarded as closely related groups, but the distinction between them was almost as sharply drawn in their most ancient known members as it is to-day, and there was no distinct tendency to converge back into a common stem. Their mutual relationships are thus obscure, but the Perissodactyla, at least, seem to be derivable from a †condylarthrous ancestry.

The †Condylarthra, as a whole, were by far the most primitive of the ungulates, which they connected with the clawed mammals. None of the genera yet discovered can be regarded as ancestral to any of the higher orders, but it is entirely possible that in the upper Cretaceous period the †Condylarthra were spread over all the continents, except Australia, and that from them the other ungulate orders arose in different regions. At all events, the †Condylarthra show how the transition from clawed to hoofed types may have occurred and perhaps actually did so, but it would be premature to affirm this.