ARTICLE VII. (Plates XVII and XVIII.) THE OSTEOLOGY OF KLOTHERIUM. myn oe nner ) BY W. BY SCOTT. i (INVESTIGATION MADE UNDER A GRANT FROM THE ELIZABETH THOMPSON FUND OF THE A. A. A. S.) Read before the American Philosophical Society, February 4, 1898. Elotherium is one of the many genera of fossil mammals concerning which the growth of our knowledge has been exceedingly slow, and only of late has it become prac- ticable to give a complete account of its bony structure. The genus was named in 1547 by Pomel (47 a, 6) and shortly afterward renamed Entelodon by Aymard (748) from a better specimen, but for several years only the dentition was known and that imperfectly. _ Tn 1850, Leidy (’50, p. 90) described the first American species, but, not suspecting its generic identity with the European forms, he at first referred it to a new genus, Archo- therium. Leidy’s material enabled him to give a fairly complete account of the skull. Kowaleysky, in 1876, described an imperfect skull found in France and he further showed that the feet were didactyl, a very unexpected fact in view of the pig-lke char- acter of the dentition. In this country Profs. Marsh and Cope have added materially to our knowledge of this remarkable animal (Marsh, 773, 793, 94; Cope, °79) and the former has published a restoration of one of the species. In spite, however, of this list of workers who have, from time to time, occupied themselves with the study of EVothe- rium, much still remains to be learned regarding its structure, and its phylogenetic rela- tionships are eyen more obscure. In the summer of 1894, Mr. H. F. Wells discovered in the White River Bad Lands of South Dakota certain bones, which, with the expenditure of infinite pains and skill, were excavated from the rock by Mr. J. B. Hatcher, and which proyed to be a most remarkably complete skeleton of Hlotherium. This beautiful specimen (Princeton Mu- seum, No. 10885,) formed the subjecu of a preliminary communication which I made to the third International Zodlogical Cong.ess, at Leyden (Scott, 96), and will be more fully described in the following pages. Except for a single thoracie vertebra (and perhaps a 274 THE OSTEOLOGY OF ELOTHERIUM. - few caudals) and part of the hyoid apparatus, the skeleton is complete; it is represented in Pl. XVII, which will enable the reader to judge of its unusual state of preservation. Additional material, belonging to several species, will also be made use of for purposes of of comparison, but the description will deal almost exclusively with the White River forms. The Artiodactyla may almost be designated as the despair of the morphologist. So manifold are the forms which this puzzling group has assumed, and so variously are the characteristics of its minor groups combined, that the confusion seems hopeless. The only way in which this tangled skein can be unraveled and its many threads separated and made straight, is by the slow but sure method of tracing the phylogenetic develop- ment of each family step by step from its incipient stages. Many years must pass before sufficient paleontological material has been gathered to make this possible, but already some progress has been made in the work. Each successive form in a series, as soon as it is recovered, should be fully described and illustrated for the benefit of other workers, a necessity which must excuse the minuteness of detail into which the following descrip- tion enters. For the sake of convenience the entire bony structure of the animal will be described, including those parts which are already well known, in order that the reader may be spared the trouble of searching through many scattered papers, written in several languages. I. Tue DeEntirion. The teeth of Eotherium are already familiarly known and require but a brief account here. The dental formula is I 3, C4, P 4, M 3. A. Upper Jaw.—The incisors, three in number, increase regularly in size from the first to the third, the latter being much the largest of the series; it has a conical or some- what trihedral crown and resembles a canine in shape and appearance. In some indiyid- uals the crown of this tooth is worn in a peculiar manner, a deep groove or notch being formed on its posterior side, in a place where it cannot have been made by the attrition of any of the lower teeth. The other incisors have spatulate crowns, with blunted tips, the attrition of use wearing down the apices as well as the posterior faces of these teeth. This description applies more particularly to the larger White River species, such as EE. ingens and E. imperator ; in E. mortoni the upper incisors are of more nearly equal size and more conical shape. In all, the median incisors are separated from each other by a considerable notch, and the whole series is much more extended antero-posteriorly than transversely, the external incisor standing behind the second one. I 3 is separated by a short diastema from the canine and at this point the premaxillary border is quite deeply notched to receive the lower canine. The canine is a very large and powerful tusk, with a swollen, gibbous fang; the THE OSTEOLOGY OF ELOTHERIUM. 275 crown is long, massive, recurved, and bluntly pointed; it is oval in section, and has a prominent posterior ridge. The premolars are very simple in construction. The first three are well spaced apart and have compressed, but thick, conical crowns, without accessory cusps of any kind, and each is implanted by two fangs. In size, they increase posteriorly and p ® has a decidedly higher crown than any other premolar. P 4 is smaller than p * in every dimension except the transverse, this diameter being increased by the addition of a large internal cusp (the deuterocone) and the crown is carried upon three fangs. In the smaller species of the genus, such as £. mortoni, p ? and p 4 are placed close together, while in the larger forms these teeth are separated by a short space, and the diastemata between the other premolars and between p ! and the canine are relatively somewhat greater, the enlargement of these teeth hardly keeping pace with the elongation of the muzzle. In the European species, £. magnum, the arrangement of the premolars is somewhat different, p 2, ® and 4 forming a continuous series, while p + and ? are quite widely separated. The molars are relatively quite small; m ? is the largest and m * the smallest of the series. The crowns are low and bunodont, bearing six tubercles arranged in two trans- verse rows. The hypocone, though functionally important, is decidedly smaller than the protocone, and structurally is still a part of the cingulum. Schlosser is, however, mis- taken in supposing that there is any important difference between the American and the European species of Hlotherium with regard to the position of the protocone. In m 4, which has a more oval crown than the other molars, the sexitubercular pattern is obscured by the development of numerous small tubercles upon the hinder half of the tooth. The cingulum of the molars is quite strongly marked, especially upon the ante- rior and posterior faces. B. Lower Jaw.—The incisors resemble those of the upper jaw, except that they are of more nearly equal size and somewhat more spatulate shape; i = is little enlarged and is much smaller than the corresponding tooth in the upper jaw. The canine is a very large, recurved tusk, ike the upper one in size and shape; it bites between the upper canine and enlarged external incisor, the three teeth together making up a very formidable lacerating apparatus. An interesting hint as to the habits of this animal is given by a peculiar mode of wear of the lower canine which occurs in some well-preserved specimens. In these we find a deep groove on the posterior face of the tooth, beneath the enamel cap and close to the level of the gum. No other tooth can reach this point to cause such a mode .of attrition, and the grooye is doubtless due to the habit of digging up roots with the lower tusks; the pull of the roots, especially when covered with sand or other gritty material, would naturally wear such a groove.* The * This ingenious and highly probable explanation of a somewhat puzzling fact was suggested to me by my colleague, Prof. C. F. Brackett. 276 THE OSTEOLOGY OF ELOTHERIUM. same explanation applies to the curious notches sometimes worn in the external upper incisor. The numerous specimens examined do not indicate that there was any difference between the males and the females in the size of the canines, the tusks being invariably large and powerful. If, as here suggested, the canines served other purposes than those of weapons, the lack of any such sexual difference would be intelligible enough. The premolars are very simple and quite like those of the upper series in shape; their crowns are massive, compressed cones, without additional cusps. The cingulum is usually prominent, but varies in the different species. P , is much the highest of the series, especially in £. imperator, where it rises to the full height of the canine, and gives a very characteristic appearance to the lower dentition. P > has its posterior face flat- tened, forming an incipient fossa with a number of small tubercles in it. P and ; stand quite close together, and p + is separated by a short space from the canine, while pz is isolated by considerable diastemata both in front of and behind it. The lower molars are small in proportion to the size of the jaw and to the space occupied by the premolar series. In size they increase posteriorly, and they have a simple, quadritubercular pattern, the crowns surrounded by a strong cingulum. There is much variation in the development of the fifth or posterior unpaired cusp (hypoconu- lid); it is frequently absent and represented only by a strong cingulum, though some- times it is present as a distinct cusp on m > or mz. It is less commonly found on m x and only in the very large £. /eidyanum is it well developed. The Milk Dentition.—The temporary canines and incisors differ from the permanent ones only in size. It is uncertain whether the first premolar, in either jaw, has a prede- cessor in the deciduous series, none of the specimens distinctly showing such a predecessor. In one individual, however, the tip of p 1 is just visible in the centre of a large alveolus, from which a milk-tooth has apparently been shed. If this change does actually occur, it must take place at an early stage, and, on the whole, it seems probable that, at least in the upper jaw, the number of deciduous premolars is four. Dp 2 has a compressed, elongate, conical crown, without accessory cusps of any kind; it is carried on two widely separated fangs, and is isolated by diastemata both in front of and behind it. Dp 2 consists of three principal cusps. The antero-external cusp (protocone) is an acutely pointed pyra- mid, while the postero-external cusp (tritocone) is lover and smaller. The internal cusp (tetartocone) is posterior in position and placed on the same transverse line as the trito- cone, while between the two is a small conule. The cingulum is distinct on the front and hind faces, obscure on the outer and absent from the inner face of the crown. Dp 4 is molariform, but differs somewhat from the molar pattern in the fact that the postero- internal cusp is even more distinctly an elevation of the cingulum and that the posterior conule is double. THE OSTEOLOGY OF ELOTHERIUM. 277 The lower milk-premolars are eyen simpler than the upper; dp 5 and = are com- pressed and conical, without accessory cusps, but with serrate edges and sharply-pointed summit. Each of these teeth is supported upon two fangs, Dp ; is of the usual artio- dactyl type, consisting of three transverse pairs of cusps, of which the median pair is the largest, and the anterior pair the smallest. A small talon is formed by the eleyation of the cingulum in the median line, behind the posterior pair of cusps. This account of the milk dentition applies only to #. mortoni; I have not seen these teeth in the larger species, Measurements. No, 11156 No. 10885 | No. 11009 No. 11440 | Es ae ees Upper dentition, length I 1 to M3...... BARONE enue ene tA Ne | | | 20.270 SpeMOlATMSeTICS Mle MOL Da tetrmseeyiciint aia eiiictaasisaeee isi | 118 1 104 | 064 .065 He ORAM EN oaks We coc bbpboecoooubauboobegooooREdS 238% | 175 | 124 113 cose CANIN Crs AN te- POS URC IAME LET ein ict ees elerepoicteteladeestsreiezacicre .048* .046 .032 ss oG ELAN S\VeETSen diam eteryesere cer vace ase eee .03885* 042 | .022 IPA, Noy sonpavern-naossauckpoosmecsuonn aauutosdecouall: 9 @USIe Mee ADy hy | @2e0sh) EOP IO CNA ORL Uehara Ne ncih ower oce ties saat otaet moped AUS SS) | ets 028 GER TES BN AES SR UA ae DS REE AC Seat el ae aac Ra es | 088 028 [Lee P yA: GG (geood.acbo aon SES ODS RIE OTR AE EOE eo .035* .031 | .0195 -018 comer MOl len Otis case tepevolnistelsraretans aicicisieter tacaci acs eee pelseereeseis .035 083 | -020 Sa witli eee ee eee eee Auer es ne not ne Ea | | 036 | 019 Mo len ctheer eee nie Ore eso aoe see Ne | .042 085 025 023 Es LEP WAG lrceou ee sastcersietrio seats Gate SOD OMA A Hey ota | .039 0235 024 miss .029* 027 013 GE WL PHIIGIVRIN, sopmcoobosDSe OP sseaesnooG idhvecsuddautuads .0895* .035 .0225 se CSL WAl Cl tM teen tsavctey lay cole telal sjniovet stefajeiesieteneweicieleforsayeatray-tehcfoas -036* .030 -016 CS WLS ISMN S65 ooodob ode SnonsgusabeooanubtosusatoeddnoE .043* .039 .0245 es GG" A VAGAN so ggoneaeuceas CODED da DODS SanYalewisievar selveieye doe .O8T* .028 .016 *No. 11161. 278 THE OSTEOLOGY OF ELOTHERIUM. Il. Tue Sxvtt. The skull of Elotherium is one of the most remarkable features of this very curious animal. It is characterized by great length and slenderness, with the supraoccipital and nasal bones lying in the same horizontal plane. The muzzle is exceedingly long and narrow, and tapers somewhat anteriorly, though expanded by the sockets of the great tusks ; the orbit has been shifted far back, its anterior border being, in some species, over m ?, and in others above m °. The cranium is short and of absurdly small capacity, which, with the great temporal openings, gives an almost reptilian appearance to the skull when viewed from above or below. The sagittal crest is very high and thin, and the zygomatic arches, though rather short, are enormously developed. One of the most peculiar features of the skull is the great, compressed plate which is given off from the ventral surface of the Jugal and descends below the level of the lower jaw, and this gro- tesque appearance is further increased by two pairs of knob-like processes on the ventral borders of the mandible. The occiput (Pl. XVIII, Figs. 1,2) is high and very broad at the base, but narrowing rapidly to the summit; above the foramen magnum it forms a broad, flat projection of almost uniform breadth, with a very deep fossa on each side of it. The basioccipital is stout and rather short, keeled in the median ventral line and slightly contracted to receive the auditory bulla; at its junction with the basisphenoid it forms a pair of small, roughened tubercles. The exoccipitals are yery large bones, espe- cially in the transverse direction along the base of the occiput, dorsally they narrow fast. Above the foramen magnum they form the very broad, prominent and nearly square pro- jection which has already been mentioned ; this is thick and is filled with cancellous bone, the fossa for the vermis of the cerebellum making but a slight depression upon its internal face. On each side of the projection is a large and deep triangular fossa, which, how- ever, is not confined to the exoccipital, the periotic and squamosal both being concerned in its formation. The inferior part of the exoccipital extends widely outward, reaching to the line of the glenoid cavity, and ending in the large, prominent and massive, but not elongate paroccipital process. In this region the exoccipital is brought very close to the zygoma, but, ventrally at least, does not quite touch it, a narrow band of the tympanic inter- vening between them. The foramen magnum is strikingly small and of a transversely oval shape. The occipital condyles are relatively rather small, especially in the vertical dimen- sion, laterally they are well extended, and they are widely separated both above and below. In the very large 2. imperator the external angles of the condyles are abruptly truncated in a curious way, and bear flat articular surfaces, though in some individuals this trunca- tion is found only on one side; while in the smaller species the condyles are of the usual form. The supraoccipital is a large bone, widest at the base (7. ¢., the suture with the exoccipitals) and narrowing dorsally. Superiorly it is drawn out into two posterior wing- THE OSTEOLOGY OF ELOTHERIUM. 279 like processes, such as are found in Oreodon and other White River ungulates. Between these wings the hinder face of the bone is concave and at the bottom of this concavity are two small, but profound pits. The supraoccipital is continued over upon the roof of the cranium and forms a part of the sagittal crest. A considerable part of the periotic is exposed on the surface of the skull, at the bot- tom of the lateral occipital fossa, where it is enclosed between the exoccipital and the squamosal; it does not give rise to any distinct mastoid process. The occiput of the European species, 2. magnum, as figured by Kowalevsky (’76, Taf. XVII, Fig. 5), is different in many details from that which characterizes the Amer- ican species. It has more of an hour-glass shape, not so wide at the base, more contracted in the middle and more expanded at the top, but with much less conspicuous wing-like processes, and it has no such projection above the foramen magnum, nor such deep lateral fossee. The condyles are larger and of an entirely different shape, haying their principal diameter vertical, instead of transverse. The paroccipital processes are longer, more com- pressed and not so widely extended laterally. The foramen magnum is large and of more nearly circular outline. The basisphenoid is narrower than the basioccipital and is not keeled on the ventral surface, but is otherwise like that bone. So much of its course is concealed by the union of the palatines and pterygoids along the median line that its length cannot be deter- mined, while the presphenoid is nowhere exposed to view. The tympanic is very extensively developed (PL XVIII, Fig. 1). Part of it is inflated into an oval, somewhat flattened and rather small auditory bulla, which differs from that of Hippopotamus and of all existing suillines in being hollow and not filled up with spongy tissue. On the outer side of the bulla the tympanic is extended as a narrow strip, which broadens considerably between the squamosal and the exoccipital, with both of which it articulates suturally, as well as with the alisphenoid in front. The bulla itself terminates anteriorly in a blunt spine. The alisphenoid is small and forms very little of the side of the cranium. It is most elongate antero-posteriorly along the ventral line, but has hardly any distinctly developed pterygoid process. At the line of the sphenoidal fissure, which notches but does not per- forate the bone, the alisphenoid is narrowed, to expand again at its suture with the parie- tal and frontal. The orbitosphenoid is relatively rather large, but is low in the vertical dimension, and does not extend upward into the orbit proper. Two sharp ridges on the external face of the bone enclose a V-shaped groove, in which lie the optic foramen and foramen lacerum anterius. The parietals are very large proportionately to the size of the cranium, but quite small as compared with the entire length of the skull; they roof in most of the cerebral A. P. $.—VOL. XIX. 2 J. 280 THE OSTEOLOGY OF ELOTHERIUM. chamber, but toward the ventral side they rapidly contract, forming narrow strips between the squamosal and frontal. Throughout their length the parietals unite to form the very high, thin and plate-like sagittal crest, which is one of the most characteristic features of the skull. In the European species, £. magnum, this crest has a remarkably straight and horizontal course, but in the known American species it is gently arched from before backward. Large sinuses are developed in the parietals, so that the cerebral chamber is even smaller than it appears to be, when viewed from the outer side. These sinuses extend over the entire roof of the cerebral fossa, even invading the supraoccipital ; they appear to be traversed by numerous small trabeculae, the ends of which are seen, in the sagittal section, embedded in the matrix which fills the sinuses. The frontals are much larger than the parietals. In the postorbital region they are very narrow, in conformity with the very small size of the brain, but at the orbits they expand widely to form the broad, lozenge-shaped forehead, which is convex from side to side, though slightly depressed, or “dished” in the middle; the supraciliary ridges are very inconspicuous. _Anteriorly the frontals diverge to receive the nasals between them, sending forward long, pointed nasal processes, which, owing to the great elongation of the muzzle, are widely separated from the premaxillaries. The orbit is large and projects prominently outward ; it is completely encircled by bone, the long and massive postorbital process of the frontal uniting suturally with the shorter process of the jugal. The orbits do not rise above the level of the forehead, as they do in Hippopotamus, and present more anteriorly, less directly outward, than in that animal. Mention has already been made of a groove on the orbitosphenoid, which terminates below and behind in the fora- men lacerum anterius; this groove is continued upward and forward upon the frontal, steadily widening as it advances. The postero-superior ridge bounding the groove is the more prominent; it extends almost to the postorbital process, from which it is separated by a distinct notch, while the antero-inferior ridge dies away within the orbit. In most of the American species the forehead rises very gradually and gently behind to the sag- ittal crest, but in ZL. ingens the rise is much more sudden and steep. The frontal sinuses are large, giving the convex shape to the forehead which has been described ; these sinuses appear to communicate with those formed in the parietals. Except posteriorly, the syuamosal forms but little of the side-wall of the cranium, its suture with the parietal curving abruptly downward and forward; its compressed and prominent hinder margin forms nearly the whole of the lambdoidal crest, though a con- tinuation of it extends upward upon the supraoccipital, ending in the wing-like processes of that bone. The zygomatie process is enormously developed; it extends widely out- ward from the side of the skull as a massive, vertical plate, which is shaped much as in Aippopotamus, and is not continued forward as a broad, horizontal shelf, such as is found THE OSTEOLOGY OF ELOTHERIUM. 281 in Sus. The superior border curves upward into a great, hook-shaped process, which resembles that seen in Merycocherus, and gives a highly characteristic appearance to this region of the skull. That portion of the zygomatic process which is directed anteriorly is short and, though massive, is much less so than that which extends out laterally ; in front it is received into a notch of the jugal. The glenoid cavity is large, transversely directed and quite deeply concave, though the postglenoid process is not strongly devel- oped and is hardly more conspicuous than the preglenoid ridge. This disposition is unusual among the ungulates, but it occurs also in the Eocene genus Achawnodon and in the modern Dicotyles. The glenoid cavities of the two sides are very widely separated, their inner margins lying external to the line of the paroccipital processes. The posttym- panic process of the squamosal is small, and is closely applied to the paroccipital process. The shape of the zygomatic arches, together with the extreme narrowness of the cranium proper, causes the temporal openings to be very large and to appear widely open when the skull is viewed from above. These openings are, however, less extended transversely and more antero-posteriorly than in [ippopotamus, while in Sus they are hardly yisible from aboye. The jugal isa very remarkable bone and constitutes one of the most extraordinary features of the Elotherium skull. Posteriorly it is notched to receive the zygoma, and sends out a process along the ventral face of that bone, extending to the preglenoid ridge. The jugal forms the inferior half of the nearly circular orbit, and for this purpose its dorsal border is made deeply concave, giving off a stout postorbital process to meet that of the frontal, while anteriorly it is moderately expanded upon the face in front of the orbit, where it is wedged in between the lachrymal and the maxillary. The most pecu- liar feature of the jugal, however, is the immensely developed vertical plate, which descends from beneath the orbit downward and outward to below the level of the yen- tral border of the mandible, recalling the similar, but much less massive processes: found in certain edentates, ¢. g., Megatherium. These plates are laterally compressed, but quite thick, and when the skull is viewed from the front, they are seen to diverge quite strongly downward ; their shape varies in the different species. In the very large forms from the Protoceras beds, such as /. imperator, the process retains its plate-like form throughout, its free end being only moderately thickened. This appears to be true also of EL. mortoni, though my material is not sufficient to allow me to make this statement positively, but in the large species from the Titanotherium and Oreodon beds (2. ingens) it forms a club-like thickening at the tip, which in /. ingens is coarsely crenulate on the posterior border (see Pl. XVII). These processes are, so far as is yet known, quite unique among the hoofed mammals, and it is difficult to form eyen a conjecture as to what their functional significance may haye been. Some misunderstanding has arisen as to the spe- 282 THE OSTEOLOGY OF ELOTHERIUM. cies in which these jugal plates are found. Nothing is known concerning their presence or absence in the European representatives of the genus. Leidy’s material gave him no reason to suspect their occurrence in the species described by him, and he consequently restored the zygomatic arches without them (69, Pl. XVI). Marsh first discovered the processes in a skull of the species named by him £. crasswm, and it has sometimes been assumed that they were more particularly characteristic of that form. As a matter of fact, they have been observed in all of the American species of which well-preserved skulls are known, viz., 2. mortoni, E. ingens, and LL. imperator, and, in all probability, all the American forms, at least, possessed them. The lachrymal is a rather large bone and forms nearly half of the anterior boundary of the orbit. On the face it is expanded into quite a large plate, which articulates below with the jugal, in front with the maxillary, and above with the frontal, the long anterior process of which prevents any contact between the lachrymal and nasal. In Hippopota- mus the very short, broad frontal has no anterior process, and so the nasal and lachrymal are connected, as they are also in Sus. Within the orbit the lachrymal is but little extended ; the foramen is single, very small, and placed inside the orbital margin. The lachrymal spine is very low. The nasals are narrow, slender and very much elongated. Their greatest width is at the anterior end of the nasal processes of the frontal, and here is also their greatest transverse convexity; from this point they narrow and flatten, both in front and behind. Anteriorly they contract very gradually and terminate in sharp points, with their free ends quite deeply notched. In 2. ingens the nasals appear to be relatively shorter than in the other species. In //ippopotamus these bones have much the same shape as in Motherium, but they narrow more abruptly behind the point of greatest width, and their free ends are not notched. In Sus the nasals are truncated posteriorly and in front their free tips project far beyond the borders of the premaxillaries. The premaxillaries are yery large and heavy bones, the horizontal or alveolar portion especially so. Posteriorly, this portion is constricted, forming a groove for the reception of the lower canine, expanding again in front to carry the large incisors. The palatine processes are not much developed, the very large incisive foramina leaving but little space for them; the spines are long and slender, extending behind the canine alveolus. The ascending ramus of the premaxillary is low and rises gradually behind, and though broad at first, it rapidly becomes very slender, terminating behind in a fine point. Though these bones in Elotherium have a very different appearance from the immensely enlarged premaxillaries of Hippopotamus, vet both may have been formed by divergent modifications of a common plan. The maxillary is greatly extended antero-posteriorly, in correspondence with the THE OSTEOLOGY OF ELOTHERIUM. 283 elongation of the whole muzzle ; its facial portion is low, gradually diminishing in height forward, where its suture with the premaxillary forms a very gentle, sweeping curve. The longest suture of the maxillary is that with the nasal, the connection with the frontal being very short, owing to the extension of the lachrymal. Posteriorly, this bone pro- jects but little beneath the orbit, which has an imperfectly developed floor, and the pro- jection which it sends out to the jugal is much less massive than in Hippopotamus. The face gradually narrows forward, until it reaches the infraorbital foramen, expanding again in front of the foramen and swelling out into the prominent canine alveolus. The palatine processes of the maxillaries are long and narrow, and as the molar-premolar series of the two sides form almost straight and parallel lines, the bony palate is of nearly uniform width, slightly concave transversely, but almost plane antero-posteriorly. In front, these palatine processes are deeply emarginated by the large incisive foramina, and in the median line are still further notched to receive the long premaxillary spines. The palatines make up but very little of the bony palate, forming only a narrow strip in front of the posterior nares, and narrow bands along the sides. The palatal notches are small and shallow. The pferygoids are elongate, but quite low; there are no hamular processes or pterygoid fossee; the two bones meet suturally along the median dorsal line, completely concealing the presphenoid from view. The posterior nares are long, narrow and low, extending forward to the middle of m 2; the opening gradually contracts posteriorly, where it becomes very narrow, while the side-walls slope upward and die away upon the alisphenoids. Anteriorly the nares are divided by the very large vomer, Which is distinctly visible, and which at its hinder termination expands into a transverse plate, articulating with the palatines. The meeting of the two pterygoids forms a small canal, which appears to overlie the whole length of the posterior nares and to open forward into the nasal chamber on each side of the yomer. This is a very excep- tional arrangement, and I am unable to suggest what its functional meaning may be (see Pl. XVIII, Fig. 1, c). The cranial foramina are, in some respects, quite peculiar. The condylar foramen is large and conspicuous, being placed well in front of the condyle ; it is, however, smaller than in the specimen of /. magnum which Kowalevsky has figured. The close approxi- mation of the paroccipital and stylomastoid processes, and the outward extension of the tympanic between them, have given a somewhat unusual position to the postglenoid and stylomastoid foramina; they are crowded close together at the postero-external angle of the auditory bulla, and both of them perforate the enlarged tympanic bone. The fora- men lacerum posterius forms a long, narrow and curved slit at the postero-internal angle of the bulla, while the foramen lacerum medium and the opening of the eustachian canal occupy their ordinary position at the front end of the bulla. No distinct carotid canal is visible externally. 284 THE OSTEOLOGY OF ELOTHERIUM. Kowaleysky inferred from the study of his specimen that the foramen ovale “ nicht als selbstiindiges Foramen existirte, wie z. B. bei den Ruminanten, sondern mit dem For. lac. med. verschmolzen war, wie bei den heutigen Suiden und bei Hippopotamus” (’76, p- 433). This is probably a mistake; at all events, it is not true of the American species, in which the foramen ovale is a long, conspicuous opening, of oval shape, perfo- rating the alisphenoid. As in the ungulates generally, there is no separate foramen rotun- dum, that opening being fused with the foramen lacerum anterius. The latter is a large and somewhat irregular opening, which notches the anterior border of the alisphenoid, passing between that bone and the orbitosphenoid. The optic foramen is small and well separated from the foramen lacerum anterius, lying in front of and at a slightly higher level than the sphenoidal fissure ; it does not open so far forward as in E. magnum, and, in consequence, it does not form such a remarkably elongated canal as in the European species (see Kowalevsky, ’76, Taf. XVI, Figs. 1 and 5, dd), but, on the other hand, it is far from being a simple perforation of the orbitosphenoid, such as occurs in the recent ungulates. This elongation of the optic canal should probably be correlated with the very small size of the brain, which would seem to have been relatively smaller than in the ancestors of the genus. Though the orbits are far behind their primitive position, the backward shifting of the optic tract would seem to have kept pace with the change in the position of the orbits. The posterior palatine foramina are large and conspicuous openings, placed at the maxillo-palatine suture, and separating the two bones at these points; the palatine plates of the maxillaries are deeply grooved for some distance in front of the foramina. The incisive foramina are likewise large, invading both the maxillaries and the premaxilla- ries ; indeed, their size prevents the development of any considerable palatine processes on the latter bones. These foramina are in very marked contrast to those of Hippopota- mus, in which the enormously expanded and massive premaxillaries are perforated by two small and widely separated openings; in Sus also the incisive foramina are propor- tionately much smaller than in L/otherium. The infraorbital foramen is large and is separated from the orbit by a considerable interval, opening above the anterior border of p*. In front of the foramen a deep groove channels the outer face of the maxillary for a short distance. The canal itself is much elongated, in correspondence with the great length of the jaws, and its posterior orifice, within the orbit, is very large. The lachry- mal foramen, which is single, is quite small and is placed inside of the orbit. The supraorbital foramen is subject to some variation in the different species. In £. ingens, from the Titanotherium beds, these openings are of good size, are placed quite near to the median line, and have well-marked vascular channels running forward from them. In specimens of 2. mortoni trom the Oreodon beds, and in the very large species THE OSTEOLOGY OF ELOTHERIUM. 285 oO (£. imperator) from the Protoceras beds, the openings have become minute; they are shifted laterally and have no anterior grooves leading from them. The mandible is not the least curious part of this remarkable skull. The horizontal ramus is extremely long and nearly straight, with an almost horizontal inferior border. The depth and thickness of the ramus vary considerably ; even in skulls of the same length the mandible is decidedly more slender in some specimens than in others. The materials are, however, not yet sufficient to determine whether this difference is of a spe- cific, sexual, or merely individual character. A remarkable knob-like process is given off from the ventral border of the mandible, beneath p z, which is subject to much yari- ation in shape and elongation, in accordance with the age and size of the animal. In young individuals still retaining the milk-dentition, the process is a mere rugose eleya- tion, and in the adults of the smaller species it is hardly more than a knob, while in the large forms it becomes greatly elongated and club-shaped. No marked difference in this regard is observable between the species from the upper and those from the lower hori- zons of the White River formation, the process being relatively quite as long and promi- nent in £. ingens from the Titanotherium beds, as in /. imperator from the Protoceras beds, but in the huge John Day species it has become particularly long and heavy. The symphysis is quite long and very thick and massive; the two rami are indis- tinguishably fused together and laterally expanded, so as to somewhat resemble the sym- physis of Hippopotamus, though not attaining any such extreme degree of massiveness as in the modern genus. The chin is abruptly truncated and flattened, and rises very steeply from below; on each side, beneath or a little behind the canine alveolus, there arises from the ventral border a second club-shaped process, similar to, but much heavier and more prominent than the posterior process already described. These two pairs of knobs give to the jaw a highly peculiar and characteristic appearance ; they form another of the enigmatical features of the Elotheriwm skull, for it is difficult to imagine what part they can have played in the economy of the animal. The two inferior dental series pursue a nearly parallel course, diverging backward but little, but behind the molars the two rami turn outward and diverge rapidly, so that posteriorly they are very widely separated, in correspondence with the great interval between the glenoid cavities of the two squamosals. The angle of the mandible is prom- inent and descends below the ventral border of the horizontal ramus, much as in Hippo- potamus, though not to the same extent. The ascending ramus is not high, but of con- siderable antero-posterior extent. The masseteric fossa is quite small, but very deeply impressed, and is situated quite high upon the side of the jaw. The condyle is relatively little raised above the level of the molar teeth, and it is sessile, hence inconspicuous, oh it is large, transversely expanded, and stronely convex. The coronoid process though it is large, t h panded, and strongh Tl | proce 286 THE OSTEOLOGY OF ELOTHERIUM. is strikingly low and small; it is of triangular shape, erect and not at all recuryed, and is separated from the condyle by a very wide sigmoid notch. The mental foramen is small, single, and placed below p 35. Several of the hyoid elements are preseryed in connection with the skeleton of #. ingens which forms the principal subject of this description. The stylohyal is quite long and slender; its proximal portion is laterally compressed and very thin, but moderately broadened in the fore and aft direction. For the distal two-thirds of its length the bone is thicker and of a compressed oval section, expanding into a club-shaped thickening at the lower end, which is excavated for the connecting cartilage. The ceratohyal is con- siderably shorter than the stylohyal, but of quite similar shape; its proximal end bears a cup-shaped expansion, beneath which it becomes very thin and much compressed, but broadened antero-posteriorly ; the inferior part of the shaft is slender and oval in section, with another cup-shaped expansion at the distal end. The epihyal and basihyal have not been preserved. The thyrohyal is of remarkable length and slenderness, and obvi- ously was not codssified with the basihyal; the bone is of subcylindrical shape, with expansions at the proximal and distal ends. This hyoid apparatus does not resemble that of any artiodactyl with which I have been able to compare it. The elements of the anterior arch somewhat resemble those of Hippopotamus, but are more slender and elongate. In the modern genus, on the other hand, the thyrohyals are very short, and are ankylosed with the basihyal, a totally differ- ent arrangement from that which characterizes EVlotherium. From the foregoing description and accompanying figures it will be obyious that the skull of Hlotherium is an extremely peculiar one. Among recent animals that of Hippo- potam us approximates it most closely, and displays, with many striking differences, sey- eral decided and, it may be, significant resemblances. Some of these resemblances, such as the straight cranio-facial axis and the long sagittal crest, are of no particular import- ance, because they occur so very generally among the primitive ungulates of all groups. Other similarities, again, are not of this nature. The proportions of the cranial and facial regions, the degree of backward shifting of the orbits, the relations of the zygo- matic and paroccipital processes, the broadening of the muzzle, and the general plan of skull construction, are all similar in the two genera. On the other hand, each genus has certain peculiarities correlated with its manner of life. Thus, the elevation of the orbits and the backward displacement of the posterior nares in Hippopotamus are adaptations to its aquatic habits. Doubtless the extraordinary peculiarities of Elotherium, such as the dependent processes of the jugals and the great knobs on the mandible, are of a sim- ilar nature, though, in the absence of the soft parts, it is difficult even to conjecture what their use may have been. {THE OSTEOLOGY OF ELOTHERIUM. 937 Measurements. No. 11156, No. 10885, No. 11009. No. 11440. a Skull) extreme length on basal line...) 0... on eee ee | 0.803 20.648 70.450 «« width across zygomatic arches (behind jugal process)... 7.500 443, 297 -264 Semen Wil Lula taprsavelettepacteretelereteriieissisiereieieratetcr-toeis opogooudooDeal .183 140 089 082 Cranium, length to anterior border of orbit..................-- | .282 .288 | .198 193 Face, length to anterior border of orbit...............5....... | .518 2.378 -270 Occiputybreadthmof base mcters errr cistelveiceisleric sieicteereiesiaic er .281 202 .160 158 fe INGGANE Gog gonaquobbeonestoconebas sb aveoundaoEdecsEos | 158 120 Bony, palate; length in median line..-5..0.... 28005 -.ee ee eee | 2.376 | 247 A GOMEE OREN, NOMA os oagsdonpoovondasoedacdods doo aagues” 279 271 146 146 Descendinisprocess) Of; Jugal; Men thier. cree relic ey iajeiin aie sree | 330 256 .126 WynGhiaie, Tenshi osposasvane sdedondesaqna Sounocen ab ouseen sods -659* 608 ‘s heim htyaticoOronOldn processus state lee jokelaleelereieet = -253* 171 107 sf GUN Ain) seo seoan sguogonHooneoaUDuaS ScosbUmotooboE .183* 091 052 * No. 11161. Il. Tue Brat. Attention has been repeatedly called, in the foregoing description of the skull, to the extraordinarily small size of the brain-cayity. yen on viewing the skull externally, this smallness of the cranium proper strikes the observer immediately, and, in connection with the long, slender muzzle, gives the skull something of a reptilian aspect. When the cranium is sawn open in longitudinal section, it becomes apparent that the brain is even smaller than would be inferred from the external yiew alone, much of the space being, so to speak, wasted in the great frontal and parietal sinuses which overlie the whole cerebral chamber. In a large, full-grown skull this chamber will hardly contain an ordinary human fist. The olfactory lobes are yery large and are connected with the cerebrum by short thick olfactory tracts. The lobes are not at all overlapped by the hemispheres, but are entirely exposed for their whole length. The cerebral hemispheres are relatively small, though they are, of course, much larger than the other segments of the brain; so short are they that they do not extend over the olfactory lobes in front, or the cerebellum behind. In shape, they are low and wide, narrowing gradually forward, but with blunt anterior termination. The frontal lobe is very small, for the frontals take but little share in the roof of the cerebral chamber. The parietal lobe, on the other hand, is relatively large and forms the greater part of the hemisphere, for there is, properly speaking, no occipital lobe, the occipital bones not tak- ing any part in the formation of the cerebral fossa. The temporo-sphenoidal lobe is also quite large and prominent, but is short antero-posteriorly. The brain-cast shows that the A. P, S—VOL. XIX. 2 kK. 288 THE OSTEOLOGY OF ELOTHERIUM. hemispheres were conyoluted, but the conyolutions are so feebly marked that they are hardly worth description. It is obvious, however, that the gyri were fewer and simpler than in any of the modern ungulates. The cerebellum is rather small, though the cerebellar fossa has a vertical diameter not much less than that of the cerebral fossa. Antero-posteriorly the former is quite short and its transverse breadth is not great. This breadth is still further reduced by the relatively very large size of the periotic bones which extend freely into the fossa. IV. . THe VERTEBRAL CoLumMy. The vertebral formula is: C7, Th ? 18, L 6,8 2, Cd 15 + ; The atlas (Pl. XVIII, Fig. 3) is very wide transversely, and at the same time it is of considerable antero-posterior extent, a shape which recalls that of Anoplotherium, rather than that of the recent ruminants or suillines. The anterior cavities for the occipital condyles are deep and wide, but low and depressed. Dorsally, these cotyles are widely separated by a broad, but not very deep emargination of the neural arch, nor do they approximate each other yery closely on the yentral side, a notch of considerable width intervening between them at this point. The neural arch is thick and heavy, but short from before backward and quite narrow transyersely ; it is also low, not arching strongly toward the dorsal side, and nearly smooth, being free from any but the most obscurely marked ridges. The foramina perforating the arch for the first pair of spinal nerves are unusually large. The neural spine is rudimentary and forms only an inconspicuous tubercle. The neural canal is low and broad, forming a transversely directed ellipse. The inferior arch is considerably more elongated antero-posteriorly than the neural, and has but little transverse curvature, except laterally, where it rises to form the sides of the neural canal. The hypapophysis is represented by a small, backwardly directed tubercle, which arises from the hinder margin of the ventral arch, and occupies the same position as in the pigs, but is much less strongly developed. The articular surfaces for the axis are low and broad, and have a very oblique position, presenting inward toward the median line, almost as much as backward ; they have also a slight dorsal presenta- tion. In shape, they are yery slightly concaye and are surrounded by prominent borders. The facet for the odontoid is wide, and deeply concave in the transverse direc- tion, but quite short antero-posteriorly. This facet is connected at the sides with those for the centrum of the axis, but distinct ridges are formed along the line of junction. The transverse processes of the atlas extend out widely from the sides of the arch, attaining their greatest transverse breadth along the posterior line; they are also yery long in the fore-and-aft direction, reaching far behind the surfaces for the axis. For most of their course the transverse processes have thin borders, but posteriorly the THE OSTEOLOGY OF ELOTHERIUM. 289 margin becomes much thicker and more rugose. The vertebrarterial. canal, which is notably small, occupies much the same position as in Sus, opening posteriorly upon the dorsal side of the hinder border. The anterior extension of the transverse processes has conyerted into foramina (atlanteo-diapophysial) the notches for the inferior branches of the first pair of spinal nerves. On the ventral face of each process is a large fossa, enclosed between the side of the inferior arch and the greatly thickened posterior border of the process. The resemblance in shape to the atlas of Anoplotherium, to which atten- tion has already been called, affects more particularly the form of the transverse processes but they are more extended transversely than in that genus and are not so pointed at the " postero-external angles. The axis (Pl. XVIII, Fig. 4) is a short, but very massively constructed bone, which in general shape and appearance resembles that of Aippopotamus. The centrum is short, anteriorly very broad and depressed, but thickening posteriorly, and with a nearly circular and shghtly concave hinder face. A strong and prominent keel runs along the ventral face of the centrum, enlarging backward, and terminating behind in a trifid hypapophysis. The odontoid process is short, heavy and conical, with no tendency what- ever to assume the depressed and flattened shape which occurs in so many White River ungulates. The ventral articular surface of the odontoid seems like something super- added to the process itself, for it is clearly demarcated by a groove running all around it, and projects slightly in front of the body of the process. On the dorsal side of the centrum a broad and well-defined ridge runs backward from the odontoid along the floor of the neural canal. The atlanteal articular surfaces are very broad and low, not rising so as to enclose any part of the neural canal. They are very oblique with reference to the median line of the centrum, with which they form angles of about 45°. These surfaces are slightly convex in both directions, and ventrally they project much below the level of the centrum. The transverse processes are short, thin and compressed, much less massive and widely extended than in Hippopotamus ; they are perforated by very large foramina for the vertebral arteries. The pedicels of the neural arch are low and short, but very heavy ; they are not pierced for the passage of the second pair of spinal nerves, as they are in Hippopotamus and in some of the pigs. The neural canal is decidedly small, especially its anterior opening; behind, it enlarges somewhat, particularly in the dorso- ventral dimension, the posterior opening being high and narrow, while in Hippopotamus it is low and broad. The neural spine is a large plate which is very thin in front, but becomes thick and massive behind, ending in a broad rugosity. This spine resembles that of Hippopotamus, but is not produced so far backward and does not overhang the third cervical. The postzygapophyses are large, slightly concave, and present obliquely 290 THE OSTEOLOGY OF ELOTHERIUM. outward, as well as downward; their bases are separated by a broad and deep groove, which is continued upward upon the posterior side of the neural spine. The third cervical vertebra also bears a considerable resemblance to that of Hippopo- famus, differing only in some points of detail. The centrum is short, heavy and moder- ately opisthoccelous, depressed, but increasing posteriorly in vertical thickness. It bears a strong ventral keel, which terminates behind, as in the axis, in a trifid hypapophysis. The pedicels of the neural arch are not, as in the pigs, pierced by foramina for the spinal nerves; they are low and short, but very thick, and the neural canal is strikingly small. The dorsal side of the arch is short, broad and nearly flat. The neural spine is remarkably well-developed (when the anterior position of the vertebra is taken into account), rising as high as that of the axis. It is rather thin and compressed, although its base occupies the whole fore-and-aft length of the arch. From the base, however, it rapidly tapers upward and terminates ina small, rough tubercle. In Hippopotamus the third cervical has an even better deyeloped neural spine, not higher, but broader and less tapering than in Elotherium. The prezygapophyses are large, oblique and somewhat convex ; they are placed yery low, so that their inferior margins are separated from the centrum only by narrow notches. The posterior zygapophyses are much larger and more prominent than the anterior pair; they are also less oblique in position and are raised higher above the centrum, corresponding to the posterior elevation of the neural arch. The transverse process is a compressed plate, which has no great vertical height, but is well extended from before backward, exceeding the centrum in length; the pos- terior portion of the process is thickened and recurved, ending in a rugose hook. The absence of any distinctly marked diapophysial element distinguishes this vertebra from the corresponding one of Hippopotamus and Sus, and in the latter genus the inferior lamella is more slender and rod-like, while the spinal neryes make their exit through foramina in the pedicels of the neural arch. The fourth cervical vertebra is different, in many respects, from the third. The centrum is somewhat shorter and is less distinctly carinate on the yentral side, but is more decidedly opisthoceelous, The neural arch is remarkably short in the antero-posterior dimension, so that the articular faces of the postzygapophyses actually extend forward beneath those of the anterior pair, which gives to the pedicel of the neural arch, when seen trom the side, a curiously notched appearance. The neural spine is higher, but more slender and recurved than that of the third cervical. The transverse process is altogether different in shape from that of the latter. It has, in the first place, a very prominent diapophysial element, which projects outward as a heavy, depressed bar, thickened, rugose, and slightly upcuryed at the distal end, In the second place, the inferior lamella is much higher yertically, but decidedly shorter from before backward, THE OSTEOLOGY OF ELOTHERIUM. 291 In Hippopotamus and in Sus this vertebra is very similar to that of Elotherium, but the neural spine is notably heavier. The fifth cervical vertebra has an even shorter neural arch than the fourth and a much higher neural spine. The spine tapers rapidly from the base upward and becomes yery slender, but it is nearly straight and only slightly recurved. The neural canal is somewhat larger than in the fourth vertebra, but, as in all the cervicals, it is strikingly small as compared with the size of the vertebra as a whole. The diapophysis is strong and prominent, but more slender than on the preceding vertebra, while the inferior lamella, though relatively short from before backward, has attained great vertical height and is strongly everted. In lotherium the fifth vertebra is of the same type as the sixth, whereas in Hippopotamus it more nearly resembles the fourth. The sixth cervical is very like the fifth, but displays certain obvious differences. Thus, the neural arch is even shorter antero-posteriorly, and the neural spine is higher, heavier and much more strongly recurved. The postzygapophyses are decidedly smaller and are very characteristic in their markedly oblique position, for they rise steeply back- ward in a way that occurs in none of the other yertebree. The diapophysis is shorter but heayier than that of the fifth, while the inferior lamella is of similar shape, but larger, higher and with the free margin more thickened. In Hippopotamus this vertebra has much the same construction as in Elotherium, but the spine is shorter and more massive and the inferior lamella is much larger. In Sus the sixth cervical bears considerable resemblance to that of the White River genus. The seventh cervical is characterized by the height and thickness of the spine, which in these respects much exceeds that of the sixth. This spine tapers superiorly, but expands again at the tip into a rough tubercle. The posterior zygapophyses stand at a higher level than the anterior pair and are unusually concave. The peculiarities seen in the postzygapophyses of the sixth and seventh vertebree are to provide for the curvature of the neck, which changes its direction at this point. From the occiput to the sixth cervical the neck is nearly straight and inclines downward and backward, while the seventh vertebra begins the rise which culminates in the anterior thoracic region. This change in direction requires greater freedom of motion, which is supplied by the modifi- cation of the zygapophyses upon the vertebree mentioned. The transyerse process is, as usual, not perforated by the vertebrarterial canal; it is rather short, but heavy and much expanded at the distal end. On the posterior face of the centrum are large facets for the heads of the first pair of ribs. In Hippopotamus the neural spine of the seyenth cervical is relatively much longer and heayier than in Llotherium or in Sus. As a whole, the neck of Elotherium is short and massive, with yery strongly developed processes for muscular and ligamentous attachments, as are indeed necessitated 292 THE OSTEOLOGY OF ELOTHERIUM. by the immense weight and length of the head. Among recent artiodactyls Hippopotamus has cervical vertebree most like those of H/otherium, though there are many differences in the details of construction. The most apparent of these differences lies in the greater and more uniform height and thickness of the neural spines in the modern genus. Doubtless the eyen more exaggerated massiveness of the skull in the latter is the occasion of this increased deyelopment of the cervical spines. In Sus the perforation of the neural arches for the passage of the spinal nerves constitutes an important difference from Elotherium. The thoracic vertebra would appear to have numbered thirteen, though this point cannot, as yet, be determined with entire certainty, and while the thoraco-lumbar vertebree were, in all probability, nineteen in number, as is well-nigh universal among the artio-— dactyls, yet there were doubtless variations in the number of ribs, as is very frequently the case among existing animals. The first thoracic has a rather small centrum, with decidedly convex anterior and nearly flat posterior face; the facets for the rib-heads are very large and deeply concave. The transverse process is rather short, but very large, heavy and rugose, and bears an unusually large, concave facet for the tubercle of the first rib. The prezyga- pophyses are of the ceryical type, but present more obliquely inward than in the vertebrie of the neck, while the postzygapophyses are, as in the other thoracics, placed upon the ventral side of the neural arch. The neural canal is high and narrow and its anterior opening has assumed a cordate outline. The neural spine is inclined strongly backward, much more so than that of the seventh cervical, and though laterally compressed it is extremely high, broad and massive, greatly exceeding in all its dimensions that of the last neck vertebra. The anterior six thoracic vertebrae (see Pl. XVIII, Fig. 5) are very much alike in appearance. The first three have broader and more depressed centra, which in the others become deeper vertically and more trihedral in section. The transverse processes are very large and prominent and carry large, deeply concave facets for the rib tubercles. The neural spines are very high, thick and heavy, and are strongly inclined backward, with club-shaped thickenings at the tips. At the seventh thoracic begins a rapid reduc- tion in the length and weight of the spines, a process which reaches its culmination on the eleyenth vertebra, which has a remarkably short, weak and slender spine. This arrangement results ina great hump at the shoulders, somewhat as in Vitanotherium, though in a less exaggerated form. In both genera, the length of the anterior thoracic spines should be correlated with the great elongation and weight of the skull which requires immense muscular strength in the neck and shoulders. Hippopotamus has no such hump, but this is probably explained by its largely aquatic habits. THE OSTEOLOGY OF ELOTHERIUM. 295 A change in the character of the facets for the rib tubercles occurs simultaneously with the shortening of the neural spines ; they suddenly become much reduced in size and are plane instead of concave. The transverse processes, however, remain yery large and prominent as far back as the eleventh thoracic. In no case are these processes per- forated by vertical canals, such as occur in Sus. The twelfth thoracic is the anticlinal vertebra and has a nearly erect spine of lumbar type, though somewhat more slender than in the true lumbars. On the thirteenth the spine is quite like that of the lumbars and inclines slightly forward. Transverse processes are absent from the last two thoracic vertebrae, which display the feature, very unusual in an ungulate, of large and conspicuous anapophyses. As far back as the eleventh vertebra the zygapophyses are of the ordinary thoracic type; they are small, oval facets, the anterior pair on the front of the neural arch and presenting upward, the posterior pair on the hinder part of the arch and presenting downward. On the eleventh thoracic a change takes place ; the anterior zygapophyses are as before, but the posterior processes are fiat and present obliquely outward, rather than downward, the two together forming a prominent, wedge-shaped mass. The prezygapophyses of the twelfth vertebra are correspondingly modified ; they present obliquely inward and together constitute a cavity which receives the wedge-like projec- tion from the eleventh. Prominent metapophyses also make their appearance on the twelfth thoracic. The posterior zygapophyses of the latter and both pairs of the thirteenth are of the cylindrical, interlocking type characteristic of the lumbars. These processes are remarkably complex and in a fashion that does not occur in Hippopotamus, but is found in Sus and many of the Pecora. The complexity is occasioned by the development of large episphenial processes, which give an additional articular surface above the zy gapophyses proper ; in section these processes have an S-like outline, and they constitute a joint of great strength. The lumbar vertebre (Pl. XVIII, Fig. 6), almost certainly six in number, have rather short, but massive centra. In the anterior part of the region the centra are some- what cylindrical in shape, but they become more and more depressed and flattened as we approach the sacrum. The neural canal is broad and very low, especially in the pos- terior part of the region. The neural spines are inclined forward and are of moderate height ; they are broad antero-posteriorly, but thin and laterally compressed, except at the tips, where they are thickened. The spine of the last lumbar is a little different from the others in being more erect and slender. Episphenial processes are present on the first, second and sixth yertebree, but not on the third, fourth or fifth. These processes are apt to be somewhat asymmetrical and better deyeloped on one side than on the other, and it is probable that more extensive material would show them to be subject 294 THE OSTEOLOGY OF ELOTHERIUM. to much individual variation. Metapophyses are prominent only on the first and second lumbars, rudimentary on the third and absent from the others. The transverse processes are very feebly developed in proportion to the size of the yertebree. On the first lumbar they are short and straight, and gradually increase in length up to the fifth, but in all they are strikingly thin and slender. The last lumbar has transverse processes of unusual length, space for them being obtained by the sudden eyersion of the anterior ends of the ilia, but even here they are weak. The trunk-vertebree of Hippopotamus are much more massively constructed than those of Klotherium, the decrease in length of the thoracic spines posteriorly is more gradual, while the neural spines and transverse processes of the lumbars are much longer and in every way heavier. The thoraco-lumbar series of Sus bears considerable resem- blance to that of L/otherium, but in the former the transverse processes of the thoracic vertebrae are perforated by vertical canals, and those of the lumbars are much longer and stouter. The sacrum consists of two vertebrie only. The first has a broad, depressed centrum and very large pleurapophyses, which carry most of the weight of the ilia, though the second sacral has also a limited contact with the pelvis. On the first vertebra the prezygapophyses are very well-developed and have large episphenial processes to receive those of the last lumbar. The two neural spines are co(ssified into a high but short ridge. The second sacral has a yery much smaller and especially a narrower centrum than the first, and retains moderately complete postzygapophyses. In Hippopotamus and in Sus the sacrum is relatively much larger than in Llotheriwn, and consists of at least four vertebrae, sometimes eyen as many as six. Even in aged individuals of the White River genus I have not seen more than two vertebrae in the sacrum. The caudal vertebra (Pl, XVIII, Figs. 7, 8, 9), of which fifteen are preserved in association with one individual, indicate a tail of only moderate length, and present a number of peculiarities. The first caudal has somewhat the appearance of a miniature lumbar ; its centrum is short, broad and depressed, with quite strongly convex faces; the neural canal is relatively large and a distinct, though small, neural spine is present. The zygapophyses, especially the anterior pair, are large and prominent and _ project much in front of and behind the centrum. The transyerse processes are quite long and heavy, and are directed outward and backward. A pair of tubercles on the ventral side of the centrum represent rudimentary hemapophyses. The succeeding caudal yertebree resemble the first in a general way, but passing backward, the centra become more and more slender and elongate, while the neural canal diminishes in size, and the various processes are reduced. The hzemapophyses, on the THE OSTEOLOGY OF ELOTHERIUM. 295 other hand, increase in size and on the (?) fifth vertebra they curye toward each other, almost meeting and enclosing a canal, which continues as far back as the (?) eighth vertebra, behind which the hemapophyses are again reduced. The middle portion of the tail is composed of very long, cylindrical vertebrae, which in shape strikingly resemble those of the great cats, and which are proportionately much longer, though apparently less numerous than those of Anoplotherium. At the anterior end of each yertebra are six prominent, nodular processes, the zygapophyses, transverse processes and heemapophyses respectively. Posteriorly the centra become more and more slender, but are not much diminished in length, for what appesrs to be the penultimate vertebra is nearly as long as those in the middle region. The various processes are, however, reduced to very insignificant proportions. The last vertebra has its anterior portion shaped like that of its predecessor, but it rapidly tapers behind to a smooth, slender, compressed and subeylindrical rod, with a club-shaped thickening at the end. As I have seen but a single specimen of this curious vertebra, I cannot feel quite confident that its shape is a normal one and not due to some injury or morbid process. The tail of Aippopotamus is of about the same relative length as that of H/othe- rium, but the individual vertebree are very different, being all shorter and heayier, and diminishing in size more gradually to the end. In Sus the caudal vertebrae are somewhat more lke those of L/otherium, but none of them haye such slender elongate centra. Little is known concerning the caudals of Anthracotherium. Kowaleysky says of them: ‘“ Von _den Schwanzwirbeln liegt mir nur ein einziges vor. Obwohl seine Erhaltung sehr mangelhaft erscheint, kann man doch aus diesem kleinen Stiick den Schluss ziehen, dass der Schwanz bei den Anthracotherien kurz war und somit gar keine Aehnlichkeit mit dem sonderbar langen Schwanze der Anoplotherien hatte” (’73, p. 333; Taf. x, Fig. 36). The vertebra described by Kowalevsky is an anterior caudal and is much smaller and in every way more reduced than the corresponding ones of Elotherium. Among existing artiodactyls, it is the giraffe which most resembles the White River genus in the peculiar character of its caudal vertebree. Measurements. JXBSR ETSI, ccoseces odanccoda andsancsehodbe ogadbooodangIobceduinqDuGO0G0G00 oSausAsaHeBaRHbaccaotin Condado: osenndS 0.160 IAG] AS OLEALESUMW CME sestaseactenenaasececmeccide seascstscoteressrensodsseoretiis 94. Zittel, K. A. Handbuch der Paleontologie, Bd. IV, Miinchen and Leipzig. the EXPLANATION OF THE PLATES. Plate X VII. Skeleton of Hlotherium ingens Leidy, from the Titanotherium beds of South Dakota, about ;; natural size. Only eighth thoracic vertebra and the distal ends of certain ribs are conjectural. The tail may well have been considerably longer, as only the vertebre associated with the skeleton have been drawn. Plate X VIII. 1. Hlotherium mortoni. Basal view of skull, } nat. size. Ty, tympanic bone ; ¢, canal opening above and behind the posterior nares, 2. Hlotherium mortont. Occiput from behind, } nat. size. 3. Hlotherium ingens. Atlas, ventral side. 4. Hlotherium ingens Axis, left side. 5. Hlotherium ingens. Fifth thoracic vertebra, from the front. 6. Hlotherium ingens. Last lumbar vertebra, from behind. es, episphenial process. 7. Elotherium ingens. Anterior caudal vertebra, from above. 8. Hlotherium ingens. (?) Fifth caudal vertebra, left side. 9. Elotherium ingens. Posterior caudal. ig. 10. Hlotherium ingens. Right ulna and radius. ig. 11. Hlotherium ingens. Right manus. ii, second metacarpal (conjectural) ; v, fifth metacarpal. ig. 12. Hlotheriwm ingens. Right pes. ii, v, second and fifth metatarsals. Figs. 3-12 are approximately 1 nat. size and are of bones belonging to the skeleton figured in Plate XVIT. ( frug Pl Yq y gung IY (OMe Pid.0 Yosh xs é Viol iy GE GF, XV PT al snp ARTICLE VIII. NOTES ON THE CANIDA OF THE WHITH RIVER OLIGOCENE. BY W. B. SCOTT. (INVESTIGATION MADE UNDER A GRANT FROM THE ELIZABETH THOMPSON FUND OF THE A. A. A. S.) (Plates XIX and XX.) Read before the American Philosophical Society, February 4, 1898. The problems concerning the origin and mutual relationships of the various families into which the Carnivora Fissipedia are divided have not yet been satisfactorily solved, principally because of the rarity of well-preserved fossils representing the earler and more primitive members of the families. Especially obscure are the questions dealing with the derivation and systematic position of the Feld, a family which by many authorities is regarded as occupying an entirely isolated position, not directly connected with any of the other groups. Hardly less puzzling, however, are many of the facts of canine phylogeny, such as the relations between the two great series of the wolves and the foxes, and the connection between the many divergent genera of successive geological horizons. No satisfactory answer to these questions can be given until many complete phylogenetic series of the Carnivora shall have been discovered, for so long as the numerous wide gaps which now separate the known members of the various series remain unbridged, those series must continue to be largely conjectural. At any time, new discoveries may call for an entire readjustment of our views regarding the lines of descent of the different families. Recently, there has come into my hands some uncommonly well-preserved material for the phylogenetic history of the Canide and is the occasion of the present paper. This material was obtained for the museum of Princeton University by Messrs. Gidley and Wells, who in the summer of 1896 made a collecting trip through the Bad Lands of Nebraska and South Dakota. They had the good fortune to discover certain unworked localities where the exposures of the White River Oligocene proved to be richly fossiliferous and, in par- ticular, yielded many unusually complete specimens of primitive dogs. A study of this material has brought to light some very remarkable and unexpected facts, which, to the writer at least, seem to require a revision of some current views upon the phylogeny of the carnivorous families, and to throw some light upon the obscure and difficult problems relating to the origin of the cats. The most valuable of these specimens are referable to 326 NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. the genus Daphenus Leidy, which has long been known, though but very imperfectly, and several partially preserved skeletons permit an almost complete account of its osteology to be given. DAPHAENUS Leidy. Proc. Acad. Nat. Sci. Phil., 1858, p. 393. Amphicyon Leidy (non Pomel), aid. 1854, p. 157; Ext. Mamm. Fauna Dak. and Nebr., 1869, pp. 32, 359; Cope, Ter- tiary Vertebrata, pp. 894, 896. Canis Cope, Ann. Rep. U.S. Geolog. Surv. Terrs., 1875, p. 505. This genus represents nearly the most primitive type of dogs which has so far been determined from the Tertiary deposits of North America. It was originally described and named by Leidy, who afterward mistakenly referred it to the European genus Amphicyon, a reference which was also adopted by Cope. ‘Though more than forty years have thus elapsed since the first discovery of these animals, singularly little has been known about them, for the material obtained has been very scanty and very badly preserved. Fragments of jaws, a few very imperfect skulls and fewer Hmb-bones have hitherto been the only specimens found, in spite of long and careful search, and beyond the fact that Daphanus was apparently a primitive member of the canine phylum, little could be predicated of it. The new material gathered by Messrs. Gidley and Wells fortunately removes this difficulty and gives us information regarding nearly all parts of the skeleton of these curious animals. These skeletal characters are of a very surprising nature and their interpretation is by no means easy. Especially remarkable are the many points of resemblance which we find between the structure of Daphanus and the corresponding parts of ‘such primitive Machairodonts as Dinictis. Aside from the dentition and the shape of the mandible, these resemblances in structure between the primitive dogs and the early sabre-tooth cats are ubiquitous, and recur in the structure of the skull, of the vertebrae, of the limbs and of the feet. To bring out the full force of these remarkable characteristics, it will be necessary to enter into a detailed and somewhat tediously minute description of the osteology of Daphenus, so that the means of comparison may be com- pletely laid before the reader. I. Tse DeEntition. The dental formula of the genus is I 3, C +, P 4, M 3, the same as that of Amphi- cyon, a resemblance which caused the erroneous identification of the two genera already referred to. A. Upper Jaw (Pl. XIX, Fig. 2).—The incisors are closely crowded together and form a nearly straight transverse row; they are smaller and occupy less space both NOTES ON THE CANID# OF THE WHITE RIVER OLIGOCENE. D27 transversely and antero-posteriorly than in most recent species of Canis. As in that genus, the external incisor is much the largest tooth of the series, and forms with the upper and lower canines a formidable lacerating apparatus. The diastema between the incisors and the canine is somewhat greater than in Canis, and the premaxillary is quite deeply constricted at that point, forming a groove for the reception of the lower canine. The canine is of the usual compressed, oval section, but the compression is less decided than in Canis, the longitudinal diameter not so greatly exceeding the transverse. The fang of the canine is long and stout, producing a marked swelling upon the outer face of the maxillary; the crown is of only moderate length, but is both actually and proportionately heavier than in the coyote (C. latrans). The premolars are notably small and simple; they increase in size regularly from the first to the fourth, the sectorial being, of course, much larger than any of the others. The first premolar is implanted by a single fang, and has a small crown of compressed conical shape, with much less conspicuous internal cingulum than in the recent species of the Canide. The second premolar is decidedly smaller than in most of the modern dogs, and is separated by longer interspaces from both the preceding and the succeeding tooth ; it has a low, pointed, simple and much compressed crown, without the small posterior tubercles which are found in nearly all the recent species of the family. The third pre- molar is much longer and especially has a higher crown than p 2, but has a similar shape, without posterior basal tubercles, and, like p 2, is inserted by two fangs. ‘The sectorial (p *) is very primitive in character, as compared with that of the typical recent species of Canis. Certain modern members of the family, such as Ofocyon and Canis corsac, for example, have, it is true, even smaller and simpler sectorials than Daphenus, but as in these forms this is doubtless due to a secondary simplification, they need not be drawn into comparison. The primitive character of the sectorial in the White River genus is shown in the thick, pyramidal shape of the antero-external cusp (protocone) which is less compressed and trenchant than in the modern species, in the smaller size of the postero- external cutting ridge (trifocone) and in the unreduced internal cusp (deuterocone) which is very much larger and more prominent than in Canis, and is carried upon a larger fang. The position of this inner cusp with reference to the protocone is the same as in the recent genus. As a whole, the sectorial is small and gives to the dentition a decidedly microdont character. The premolar series of the two sides diverge quite rapidly posteriorly, each tooth, except p 1, being oblique in position, with reference to the long axis of the skull, thus giving the bony palate its greatest width at the hinder edge of the sectorials. The obliquity of the teeth and their divergence posteriorly are even more strongly marked than in most recent dogs. A. P. S—VOL. XIX. 2 P. 328 NOTES ON THE CANID£ OF THE WHITE RIVER OLIGOCENE, The upper molars are large and well developed, though the different species vary in this respect, D. vetus haying larger tubercular molars than..D. hartshornianus. ‘The first molar is, in general, like that of Cans, but differs in certain details. Thus, the two external cusps are more conical in shape, more nearly equal in size, and are not placed so near to the outer edge of the crown, resembling in this respect the upper molars of certain creodonts, such as Sinopa; the large inner crescentic cusp is much as in Canis, though hardly so prominent, especially in D. hartshornianus ; in D. vetus it is larger. The second molar is much like the first in shape and construction, but smaller and some- what simplified, the conules being minute or altogether absent. The third molar is very small and has a low, transyersely oval crown, in which separate elements are not distin- guishable. This tooth is rarely preserved and none of the specimens at my disposal “possess it, though the alveolus for it is almost always present; it is well figured by Leidy (69) Ploy Big. 5). & B. Lower Jaw (Pl. XIX, Figs. 5, 6,7). In none of the available specimens are the lower incisors sufficiently well preserved to be worth description. The canine is very much the same as in the recent members of the family. The premolars are somewhat more complex than those of the upper jaw. The first is very small and simple, while p. 3, g and , increase progressively in size and in the deyelop- ment of the posterior basal cusps. In the more ancient and primitive species ? D. dodgei, from the Titanotherium beds, the premolars are lower, thicker transversely and less acutely pointed, and haye larger posterior basal cusps than in the later species from higher horizons. In all the species these teeth are more widely separated than in the modern genera. The molars are yery characteristic of the genus, but well-marked specific differences may be observed. In ?D. dodgei the anterior triangle of the lower sectorial is of only moderate height and the heel is but slightly concaye, the outer and inner ridges (hypo- and entoconids) being very little raised. In D. hartshornianus the protoconid is high, narrow and pointed, and the talon is more concave than in the first-named species, and has more prominent internal and external cusps. In D. vetus the inner cusp of the talon (entoconid) is reduced and, as Cope has already pointed out (’84, p. 898), there isa tendency toward the formation of a talon with a single trenchant ridge, a tendency which is fully carried out in the genera Temnocyon and Hypotemnodon of the succeeding John Day horizon. In all the species of Daphenus the inferior sectorial is much more primi- tive than in the typical modern Canide, as is clearly shown by the higher and more conical protoconid, the lower and smaller paraconid and much less reduced metaconid. In fact, both the superior and inferior sectorials of Daphenus have a close resemblance to those of the creodont family MJiacide, from which this genus could hardly be separated upon the ground of the dentition only. > NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. 329 The tubercular molars are not preserved in the specimens of ?_D. dodgei ; in D. vetus they are proportionately larger than in D. hartshornianus. My is relatively large, especially in the antero-posterior diameter; it resembles the corresponding tooth of Canis, except for the presence of the small paraconid, thus giving to the tooth all the elements of a true sectorial, as is also the case in the creodont JMacide, though in the White River genus all the cusps are lower and more tubercular. Mg, is quite small, though both proportionately and actually larger than in species of Canis of similar stature, and is inserted by a single fang ; the crown is of oval shape and has an irregularly ridged surface, without distinct cusps. As a whole, the dentition of Daphenus is that of a primitive member of the Canidae and resembles the dentition of the recent members of the family in general plan and structure. Measurements. No, 11421. No, 11424. No. 10538. No. 11423. | No. 11425. | No. 11422. Upper dental series, length C to M 2.........ssseeecceeeees | 0.069 0.076 | ** incisors, transverse width | 014 O15 ‘canine, length .010 O15 3 “width | . .008 .008 Smal lene bhivecssieetsssiscctsviensteeecesss Speers | 005 005 .006 | = SoG BY Pa .008 .0085 .0095 i AGH Cates ean gees Prat Noahs .009 2.010 .009* | OE BETAS) OGL tas oe ee eR ry ee Ee ES 2.014 .0145 015 015 | | CE TOSS ashi ydacoe recat eeeasGaccBcODO SESH Seo eee eare .0085 | .009 .0105 50105) a | | “M1, length... .012 O11 O11 | | SAESIME ToSwrid bhitesereceatertec ace caer eases eager ns |)? .015 O15 015 016 | | Seas Mir) Sen prehi- vases scene teccseseessis acvcuseue osisseaveseen 0065 | ~— .007 .007 007 | | pay MgO wid Cetecnseeter sc ee ne hota near abate Wee eatreat aos 010 O11 O11 O11 Lower dental series, length C to M 3 .sesee..seeeseeseee lo O78) || .090% .090 “© premolar series, length -svssessesseeseesessseeseeeee: (ee0364 | | 041* | .040 0315 ‘“« molar series, length .. | .026 0245 | .031* | .030 PANT canto wlene theres ocak oA ioeelene tes fecoide | o11* | .012 | 010 ss co Wid CHmmeete oats sce Stactvate weteesee tise: oe ||) breadth: of proximal) end <<... :0-.c.0cc.cscceseccosesesncersencecessnassccasscnsscnscennaccunsssananases | -009 SG a 0) ‘Gistal! CnC vkcess ccoccccosscsscccccavecsccevecascvcenccscesedscasve cece ccceserensuascenseccmecemes .012 2 ms trochlea... -009 Metacarpal iii)breadth: of proximal endls..c...cc--cccescace.osececs (eave son -scnssntecdenecesedonpocssensnscanna-vaecoaas -0105 MWretacarpalstvye WON Oth awsaccc-cccecanccoracectccenctcsccractcnecc+aneusscesesscnssce-cessstensassenssttcesscassasencneaaseseas -050 * breadth of proximal end .....2.0.....0..0.sccecsccssoesccncnscnccenbeecnsansnssssnanensenacensnssasensce -0095 S Cistal EnW<.cccccessscncceeeccapberectscwscsssupccensccnssaasenecransasennessnscacasstsaeseasmrs .012 TTOCHICA ce vcvececantechuctisccon coast dacouecacedscscscuseacuos-asenteanascendsemeeraas | .010 NOTES ON THE CANIDE® OF THE WHITE RIVER OLIGOCENE. 349 VI. Tue Hrnp Lip. The pelvis is represented by several specimens belonging to D. vetus, D. hartshornianus and D. felinus, all of them incomplete, but so supplementing one another, that the shape of the os innominatum may be determined, with the exception of the anterior border of the ilium, which is unfortunately missing from all the individuals. So far as it is preserved, the pelvis is rather feline than canine in character, both in its general outlines and in its details of structure. The neck or peduncle of the ilium is wider and shorter than in Canis, narrower than in Felis; the anterior plate expands to its full width somewhat more abruptly than in the latter, but enough of the broken fossils remains to show that the iliac plate has the’ narrow form which is found in the cats and does not expand so much at the free end as in the modern dogs. The gluteal surface is not simply concaye, as it is in the two recent genera mentioned, but is divided into two unequal fossee by a prominent longitudinal ridge, such as occurs, though not so prominently developed, in certain viverrines. This feature is repeated in another White River dog, Cynodictis, and is almost duplicated in the contemporary sabre-tooth, Dinictis, another of the many correspondences between Daphenus and the early Machairodonts. The sacral surface is placed much less in advance of the acetabulum than in Canis, and occupies about the same relative position as in the cats. The ischial border of the ilium is, for most of its length, nearly straight and parallel to the acetabular border, but descends more abruptly than in either the recent dogs or cats, and follows a course more like that seen in Viverra. As in Canis, the acetabular border is more distinctly defined than in the true felines, and ends near the acetabulum in a long, roughened prominence, the anterior inferior spine. The pubic border is very short, and hence the iliac surface is not well defined. The acetabulum is of moderate size and has somewhat more elevated borders than in the cats. The ischium, which in the existing Canid@ is much shorter than the ilium, is very elongate, and is proportionately even longer than in the felines. The anterior portion of this element is straight, rather slender, and of obscurely trihedral section; behind the acetabulum the dorsal border is arched upward into a convexity, the spine of the ischium, terminated abruptly behind by the ischiadic notch, which is as conspicuous as in the cats, while in Canis it is very faintly marked. The posterior part of the ischium is expanded into a broad and massive plate, which is very rugose upon the external surface. This posterior portion is not so strongly evyerted and depressed as in the modern dogs, and there is no such stout and prominent tuberosity, which, again, constitutes a resemblance to the cats. The pubis is L-shaped and its anterior, descending limb is unusually long, broad and thin, much more so than in the felines or modern dogs. The obturator foramen is 350 NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. very large, forming an oval, with its long axis directed antero-posteriorly, in shape and size agreeing much more closely with the condition found in the cats than with that of the recent dogs. The femur (Pl. XX, Fig. 18) is stout, and long in proportion to the length of the fore-limb bones, but not very long as compared with the size of the animal. While not differing in any very marked fashion from the thigh-bone of Canis, it yet has some resemblances to that of the felines. The small, hemispherical head is set upon a longer neck than in recent dogs and has a smaller, deeper and more circular pit for the round ligament, than in the latter. As in Canis, the head projects more obliquely upward and less directly inward than in Felis. The great trochanter is large and has a very rugose surface, but it has no such antero-posterior extension, does not rise so high and is not so pointed as in the existing forms of Canidae. In consequence of this shape of the great trochanter, the digital fossa is smaller and much shallower than in the cats or recent dogs. From the great trochanter a sharp and prominent ridge, the linea aspera externa, descends along the external border of the shaft. Whether a third trochanter was present cannot yet be definitely determined, because in the only two femora preserved in the collection, the outer edge of the shaft is broken away at the point where the third trochanter would be, if present. In all probability, however, Daphenus did possess this trochanter, at least, in rudimentary form, as may be inferred from the analogy of the sabre-tooth Dinictis, and still more from the little contemporary dog, Cynodictis, which in many respects approximates the structure of the modern Canide more closely than does Daphenus. 'The lesser or second trochanter is larger, more prominent, and of more decidedly conical shape than in the recent species of either Canis or Felis. The shaft of the femur is long, slender and nearly straight, though slightly arched toward the dorsal or anterior side; it differs from that of the modern dogs in its lesser curvature, and in broadening and thickening more gradually toward the distal end, and from that of the true cats in being more slender and of more nearly cylindrical shape. The rotular trochlea is rather narrower transversely than in the true cats, or eyen than in Dinictis, but is characterized by the same shallowness, and resembles that of the latter genus in its shortness vertically and lack of prominence. Trans- versely, the groove is but slightly concave, and it has much less prominent borders than in the existing species of Canis ; these borders are slightly asymmetrical, the external one rising a little higher and being a trifle more prominent than the internal. A decided difference from both Canis and Felis consists in the fact that the trochlea hardly projects at all in front of the plane of the shaft, the anterior face of the latter gradually swelling to the level of the groove. In both of the recent genera mentioned, and especially in the canines, the trochlea projects prominently in adyance of the shaft. ea beh NOTES ON THE CANIDA OF THE WHITE RIVER OLIGOCENE. De The femoral condyles are feline rather than canine in shape; they are small and of nearly equal size, though the outer one is slightly the larger of the two, and project much less strongly behind the plane of the shaft than in Canis. They are also less widely separated and less expanded transversely than in the latter genus. As in so many features of the limb bones, the whole distal end of the femur is more like that of Dinictis than it is like the corresponding part of the modern dogs or cats. In Dinictis, however, the rotular groove is shorter proximo-distally and broader, and the condyles are even less prominent. The patella is very different from that of the recent Canida, in which group this bone is small, narrow and thick, but has more resemblance to that of Dinictis. It is quite broad, but very thin in the antero-posterior dimension; the anterior face is more roughened than in the Machairodont genus and the proximal end is more pointed, not so abruptly truncated. The facet for the rotular trochlea of the femur is, in correspondence with the shallowness of that groove, but slightly convex transversely and slightly concave proximo-distally. The tidia (Pl. XX, Figs. 19, 20) is relatively short and slender, and bears consider- able resemblance to that of Dinictis, more than to that of Canis. The proximal facets for the femoral condyles are small and but little concave ; the outer facet is somewhat larger than the inner, and projects farther beyond the line of the shaft, both posteriorly and laterally. On the distal side of the overhanging shelf thus formed is a facet for the head of the fibula, which is much larger than in the recent dogs and more rounded in shape than in Dinictis. The spine of the tibia is very low and is more distinctly bifid than in the Machairodont genus, though much less so than in Canis. As in the former, the cnemial crest is not very strongly developed; it is far less prominent than in the existing Canide and does not descend so far upon the shaft as in them. The tibial shaft is slender and nearly straight, not displaying the lateral and antero- posterior curyatures seen in Canis ; proximally the shaft is of trihedral section, becoming approximately cylindrical below and transversely oval at the distal end. The latter is shaped much as in Dinictis and is conspicuously different from that of Canis; the astragalar facets are less deeply incised, and the intercondylar ridge is less elevated than in the latter, but the facets are deeper and the ridge higher than in the Machairodont, in correlation with the deeper grooving of the astragalus. The large transverse sulcus, which in the recent dogs invades these astragalar facets, is not shown in Daphenus. The internal malleolus is very large and resembles that of Dinictis, save that its posterior border is more inclined and the process is thus distally somewhat narrower. ‘The sulcus for the posterior tibial tendon is very distinctly marked, more so than in Canis. The A oP) S.—— VOL. EX.) 2:8: 352 NOTES ON THE CANID# OF THE WHITE RIVER OLIGOCENE. distal fibular facet is quite large, bemg much as in Dinictis. and consequently much larger than in the recent Canide. The fibula (Pl. XX, Figs. 19, 20), which is greatly reduced in the modern dogs, is in Daphenus much stouter and has heavier ends, both proximal and distal. In Canis these ends have the appearance of being reduced and simplified from the condition seen in the White River genus. In the latter the proximal end of the fibula is relatively very large, especially in the fore-and-aft dimension, in which it considerably exceeds that of Dinictis, though the excess is principally due to a large tuberosity which projects from the hinder border, and which is present, though much less prominent, in the Machairo- dont. The facet for the head of the tibia is longer antero-posteriorly and narrower transversely than in the latter, forming a long, narrow, irregular oyal. The shaft of the fibula is slender, though very much thicker both actually and proportionately than in Canis, and has about the same proportions as in Dinictis ; it is laterally compressed, the principal diameter being the antero-posterior one, and of oval section, though its size and shape vary from point to point in an irregular fashion. The distal end of the fibula resembles that of Dinictis, though it is somewhat smaller, in proportion to the length of the bone. The enlargement is both antero-posterior and transverse and gives rise to a very stout outer malleolus, at the postero-external angle of which is a deep sulcus for the peroneal tendons. The distal tibial facet is rather larger than that of Dinictis, while the surface for the astragalus is somewhat smaller, the two together making a high narrow band. Measurements. No, 11421. No. 11424. No. 11423. Femur, length (fr. head) | 0.195 read thyon proxi llen esses seapeances cxtee sane = nae esens ecen ne senneatendansnesectate cm snastear 044 GISbA ONG eraccsccn scot eccceannecascenaccacanccsUacsenucsncsaascccincdarscsaeeaccasccens .038 TOCMIATH POW e sas oecenccaceassss:senccssusinndoneacanecsasee sate deresdacemestenssacas | .014 Tibia, length 149 | Preach ok proxiniall exh c .0055 .005 | 006 -005 ve .0065 007 006 0065 .006 4s .O10 | .0095 .0095 -010 .009 ts 005 | .005 005 005 | .0045 ui 003 | 008 | ~~ .0035 KEP S'— VO lp exeeXe Ue 368 . NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. Il. Tee Surin (Pl. MUX, Pigs, 12): The skull of Cynodictis is decidedly primitive and in general appearance resembles that of such viverrine genera as Paradovrurus, rather than that of the modern Canide. Among the latter the alopecoid series have skulls more resembling the type of Cynodictis than do the thooids, though the Brazilian bush-dog (Jcticyon) is, on the whole, most like the fossil in the proportions of its skull. In Cynodictis, as in Daphanus, the facial or preorbital region of the skull is very short and the cranial portion yery long. The occiput is low and the upper contour of the skull rises steeply from the inion to about the middle of the parietals, whence it descends in an almost straight line to the anterior nares, the only departure from straight- ness being a hardly noticeable concavity or “ dishing” of the nasals about midway in their length. In Vulpes the profile is quite similar, but the posterior rise from the occi- put is much shorter and less steep, and the dishing of the nasals is more conspicuous. The sagittal crest is low and weak, and in the John Day C. /emur, the smallest species of the genus, the crest is replaced by a lyrate sagittal area. The cranium, though slen- der, elongate and contracting anteriorly, is relatively fuller and more capacious than in Daphenus, and the postorbital constriction, though much deeper, is as near the orbit as in the modern foxes, and is, therefore, much farther forward than in Daphenus. The John Day specimens, which Cope has referred to C. gregarius (°85, Pl. LX VIII, Fig. 6), have an eyen fuller cranium and shallower postorbital constriction, which should, per- haps, be a reason for separating these animals specifically from the White River forms. The muzzle in Cynodictis is yery slender, but tapers gradually and is not so abruptly constricted at the line of the infraorbital foramina as in Daphenus. In the European representatives of the genus the skull is much like that of the American species, but is somewhat more primitive and like that of Daphanus. Thus, the muzzle is more abruptly constricted, and the postorbital constriction is deeper and occupies a more posterior posi- tion. A more detailed examination of the skull brings out the following facts : The occiput is low, very broad at the base and narrowing toward the summit less than in the large wolyes, but more than in Vulpes or Urocyon ; a well-marked median convexity is produced by the yermis of the cerebellum.- The crest of the inion is low and weak, much less prominent than in Daphenus. The foramen magnum differs some- what in shape in the different individuals, being in some low and broad, and in others of subcircular outline, a difference which may, in part, be due to a slight crushing. The dorsal margin of the foramen projects much more prominently than in the recent Canidae. The basioccipital is long, broad and of nearly uniform width throughout; it is NOTES ON THE CANIDEZ OF THE WHITE RIVER OLIGOCENE. 369 slightly concave transversely, but has a low median conyexity, with very feebly deyel- oped keel, the convexity being much less prominent than in Daphenus. The exoccipitals are low and wide and so convex in the median line that this por- tion projects much behind the sides. The condyles are low and depressed and are separated on the ventral side by a narrower, deeper and more V-shaped notch than in the modern wolves or foxes. The paroccipital processes are yery small and project almost directly backward, as if to avoid the auditory bulla, with which they are not in contact at any point. The supraoccipital isa large bone, both high and broad; dorsally it is reflected over upon the cranial roof, and in this region is thickened and diploétic. The mastoid is exposed quite extensively upon the occipital surface, somewhat more so than in the modern representatives of the family, and as the distance between the paroccipital process and the postty panic process of the squamosal is greater than in the latter, the mastoid occupies a rather more lateral position. The mastoid process is very small, almost obsolete. The sphenoid bones cannot be described, as none of the specimens allow the limits of these elements to be determined. The tympanic differs in very important ways from that of Daphenus. In the first place it is inflated into a very much larger auditory bulla, filling out the entire fossa and leaving no part of the periotic exposed ; and in the second place, the posterior cham- ber of the bulla is ossified and fused with the anterior chamber. The line of junction between the two elements which compose the bulla is very plainly marked by a groove upon the external surface, and shows the posterior chamber to be considerably the smaller of the two. Ihave not been able to detect any, even partial, septum between the two chambers, but such a septum as that of Canis may well haye been present. The bulla is relatively as elongate as that of Canis, but is much narrower and more compressed, and therefore has a less inflated appearance. The external auditory meatus isa very large, oval aperture, without any tubular prolongation, the borders being flat, except the ante- rior one, which forms a more prominent lip than in Canis and partially conceals the postglenoid foramen. The auditory bulla of Cynodictis is thus thoroughly cynoid in development and displays no resemblance to the characteristic viverrine type. The parietals are proportionately very large bones and make up the greater part of the sides and roof of the cranium. Throughout their length they unite to form a very low and weak sagittal crest, which becomes moderately prominent only at the concavity of the cranium formed between the occipital crest and the hinder wall of the cerebral fossa. Owing to the larger size and backward extension of the cerebral hemispheres, as well as to the lowness of the occipital crest, this concavity is shorter and much shallower 370 NOTES ON THE CANIDHZ OF THE WHITE RIVER OLIGOCENE. than in Daphenus. In some specimens, even aged ones, the anterior half of the parietals carries a very narrow sagittal area, rather than a crest, but only in the little C. lemur from the John Day does this area assume the lyrate form. This fact is of importance in determining the primitive or secondary nature of the sagittal crest, concerning which there has been some dispute. The frontals form relatively as much of the cranial roof as in Canis and have, when viewed from above, an hour-glass shape, which is due to the deep postorbital constriction, though the depth of this depression varies considerably in different individuals. The postorbital processes are very small and owe their prominence entirely to the constric- tion. The forehead is slightly conyex, both transversely and longitudinally, though in some specimens it has a narrow and shallow depression along the median line, such as is found, though much more distinctly, in modern species of both Canis and Vulpes. The forehead is bounded by the obscurely marked supraciliary ridges converging posteriorly to the sagittal crest, which is entirely upon the parietals, none of it being formed by the frontals. Anteriorly the frontals are emarginated to receive the narrow nasals, and send forward slender nasal processes, which are separated by short interspaces from the ascending rami of the premaxillaries. A noteworthy difference from Daphenus consists in the absence of frontal sinuses, in which respect Cynodictis agrees with the alopecoid series of the modern Canidae, as Daphenus does with the thooid series. The significance of this fact will be discussed in a subsequent chapter. The squamosal has a relatively small extension upon the side of the cranium, and this portion of it has a different shape from that seen in the modern dogs, the pari- etal suture descending very steeply forward from the occipital crest, while in the modern genera this suture pursues a nearly horizontal course. From the base of the zygo- matic process to the posttympanic process of the squamosal runs a projecting shelf, which overhangs the auditory meatus and is much wider than in Canis or Vulpes, though not so broad as in Cynodesmus, Hypotemnodon or Daphenus. The posttym- panic process is not larger than in Canis, but is made more conspicuous by the absence of any tubular meatus auditorius. The zygomatic process is relatively somewhat heavier than in Vulpes, and in shape and proportions much like that of the wolves, though not so strongly arched upward; anteriorly it extends to the postorbital process of the Jugal. The glenoid cavity is broad and the postglenoid process is proportionately heavier, more extended transversely and its distal end is more curved forward than in Canis. There is no preglenoid ridge. The jugal also resembles that of Canis, though it displays some differences. Thus, it is not quite so long as in the modern genus and does not extend so near to the glenoid cavity ; it has a less decided upward curvature, and the postorbital angle (it can hardly be called a process) is even less conspicuous ; the masseteric surface is broader, more lat- NOTES ON THE CANIDZ OF THE WHITE RIVER OLIGOCENE. 371 eral and less inferior in position, and is bounded above by a distinct crest; the antero- inferior, or maxillary, process is shorter, and the ascending, or frontal, process is narrower, but extends farther upward along the margin of the orbit. As a whole, the zygomatic arch is of nearly the same proportionate length as in Canis latrans, but has a straighter fore-and-aft course, being much less strongly arched upward, though curving outward quite as decidedly from the side of the skull. This comparative shortness of the arch, in association with the very elongate cranium, is due to the anterior position of the zygo- matic process: of the squamosal, which is placed much farther in advance of the occi- pital condyle than in the recent members of the family. The lachrymal forms but a very smail portion of the anterior rim of the orbit and ‘carries a rudimentary spine. Within the orbit the bone is relatively more extended and occupies a more elevated position than in the modern dogs, while the ascending or fron- tal process is much shorter ; the lachrymal foramen is large and is farther removed from the frontal suture. The nasa/s are short, narrow and slender, splint-like bones, which are convex trans- versely and very slightly concave antero-posteriorly ; their general shape is much the same as in Vulpes, except for the much less distinct fore-and-aft concavity and their lesser elongation. The premazillaries are small ; the alveolar portion is weak, in correspondence with the smallness of the incisors, and is not produced anteriorly in the spout-like form which characterizes Daphenus ; the groove for the reception of the inferior canine is much less deeply incised than in the latter. The ascending ramus is long and slender, but forms a wider strip upon the side of the muzzle than in the last-named genus. The anterior narial opening is small, oval in shape and more oblique in position than in either Canis or Vul- pes. The palatine processes of the premaxillaries are short and very narrow, and the incisive foramina are small. This portion of the palate has an entirely different appearance from that found in Daphenus ; the premaxillaries are not nearly so much extended in front of the canines, the incisive foramina are shorter and have no such grooves extending forward from them; the spines are very slender and much shorter, reaching only to the canines and not to the line of p+, as they do in the larger genus. In most of these respects Daphenus is nearer to Canis and Vulpes thaw is Cynodictis. The maaillaries are relatively very short, much shorter than in the existing genera, a statement which especially applies to the facial or preorbital portion. At the same time the vertical height is proportionately great. Except for the swelling produced by the root of the canine, the facial surface of the maxillary is simply conyex, there being no distinctly marked fovea maxillaris. Owing to the shortness and height of the facial portion, its superior and anterior margin, formed by the sutures with the frontal, nasal and premaxillary, is more strongly curved and descends much more steeply in front than OTe NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. in Canis. As in Daphenus, the infraorbital foramen is placed very near to the orbit, while in the modern genera it is much in advance of the orbit. The arrangement seen in Cynodictis is due chiefly to the anterior position of the orbit and in much less degree to the backward shifting of the foramen itself. The palatine processes of the maxillaries are short and narrow, corresponding to the shortness and slenderness of the muzzle, and they resemble those of Daphenus in being slightly concave transversely, with a faintly marked median ridge along the line of suture. The palatines have nearly the same shape and proportions as in Canis latrans (though they are relatively somewhat narrower) and extend forward to the anterior edge of p 4; the palatine notch is more deeply incised than in either Canis or Vulpes and is nearly as deep as in Urocyon. Only a single posterior palatine foramen is visible on each side. As a whole, the bony palate resembles that of Canis more than that of Daphanus in its much less abrupt narrowing at the level of the sectorials. The posterior nares have about the same shape and position as in Vulpes and have a similar median spine-like process on the anterior border. The plerygoids terminate in longer, more distinct and more thickened hamular pro- cesses than in the recent genera, some of which, like Uvocyon, have no vestige of such processes. rom the descending process of the alisphenoid is given off a prominent lateral spine, which, in Canis and Vulpes, is represented only by a low ridge. The mandible has a slender and compressed horizontal ramus, which tapers rapidly toward the anterior end; it forms a long symphysis with its fellow of the oppo- site side and curves very gently upward at the chin. The ventral border describes a somewhat sinuous course, curying downward beneath the sectorial, from which point it rises very gradually and regularly to the symphysis, while beneath the masseteric fossa it is concave. There is no trace whatever of the lobation which is found in so many of the existing Canidae, both alopecoids and thooids. The ascending ramus, which forms an obtuse angle with the horizontal, has a proportionately smaller antero-posterior width than in Daphenus, though a greater one than in the modern genera; the coronoid process, in particular, is much narrower than in the former, and the sigmoid notch is wider than in the living forms. The masseteric fossa is very deeply impressed, but it has no such definitely marked upper boundary and it does not extend forward so far beneath the molars as in Canis, features of resemblance to the alopecoids. The angle is formed by a short, slender and blunt, hook-like process. The condyle, which is not in any way peculiar, is elevated much more above the level of the molar teeth than in Daphenus. The cranial foramina are very minute and hence are often difficult to detect, save in exceptionally well-preserved specimens, a yery slight degree of crushing being often sufficient to obliterate them. In general, they may be described as characteristically NOTES ON THE CANID#Z OF THE WHITE RIVER OLIGOCENE. 373 cynoid. The condylar foramen is an opening, hardly larger than a pin-hole, which per- forates the ridge running mesially from the paroccipital process ; its position is just as in Canis. The foramen lacerum posterius is rather smaller than in existing representatives of the family, which is due to the greater proportionate elongation of the auditory bulla, and for the same reason the stylomastoid foramen is less conspicuously displayed. An important difference from Canis and Vulpes consists in the presence of a well-defined external opening of the carotid canal, which grooves the inner side of the auditory bulla somewhat behind the middle of its course ; it is much better shown in some specimens than in others. In the modern Canida, “ the carotid canal is complete and of tolerable dimensions ; but its external opening is not visible on the surface of the bulla, being deep in the foramen lacerum posticum” (Flower, ’69, p. 24). The other carniyorous families, however, have the carotid canal with visible opening, but varying in position in the different groups. The foramen lacerum medium and the Eustachian foramen are yery much as in Canis, but the glenoid foramen is somewhat concealed by the prolonged anterior lip of the auditory meatus. The foramen ovale is a narrow slit which may be readily over- looked, and is closed by even a slight distortion of the skull. An alisphenoid canal is present, and the other openings, the optic, anterior lacerated and round foramina, are as in the recent cynoids. The whole structure of the cranial basis and its foramina are thus canine in character, with only a single difference, the distinctness of the carotid canal. There is nothing to suggest relationship with the viverrines. Measurements. No. 10493. |No. 10513. | No, 10989. | No. 11012. | No. 11381. | No, 11382. No. 11432. | fon arh lisse rip Pa Eres ees oar c 3 Skull) length (fr. occ. condyles))..--....---.---scsee-ssss-reeose | 0.092 | 0.092 | 0.086 0.089 Cranium, length (occ. condyles to preorbital border)...-- .062 | 2.062 064 064 .059 -063 Hace; preorbitalolencthi.-:--+.s)o+s-ssesssnss-sccvesessecceerats a= 032 030 030 .028 Occiput, breadth across mastoid processes .......-.+sseeeeeeee LOSSEmme OS 034 .038 .035 032) ,083 Brain case, greatest breadth 031 | .032 | .032 | .035 | .033 | .033 | .033 Skull, width across ZyGOMAS ......-..ssecesscrssccnsessecceccees 052 | .055 Zygomatic arch, length............-.sesssscseesssescessesseercseess 042 043, 043 .043 | | .042 | O14 Eucel width latin; 2en a ees Seen es. Sesieaticaaes 026 | .026 | .026 | .030 | 025 a S MUCCHCANIIN Guaestersc. ste sacuisesvecstsceccscesscecscsceses 016 COL aaa 018 015 Mandible, length (fr. condyle) ........-..sseescseccnsecnseeceees 063 | .060 | “ Gepth atime aeccsccsccccssoscdcesacneseseacsesacscaremsne | .009 O11 O11 O11 | .010 | “ GE ay ed ere eee Ore EEE | 010 | .008 | 007 | cepa Miick nice xbamley -edecsteterescseseesesoass ese none 10045 | .0055 | .0055 | .005 005 Ue height of coronoid process (from ventral, | [820530 (378) keaceseneococecpoaoocnonoo doce Loox IECDECOUIG000 027 .029 | 2.027 | | .029 | height of condyle fr. angle........-+-ssseeeeseeeeees | .014 | .013 NOTES ON THE CANIDH OF THE WHITE RIVER OLIGOCENE. ise) ~T e 1, Dan Bram (Pl exe ict): The brain of Cynodictis has already been described by Bruce (’83, p. 41), but as I wish to consider it from a different standpoint, some account of it will be necessary. In this genus the brain is relatively smaller than in any of the recent Canide. ‘The olfac- tory lobes are large and are left exposed by the hemispheres, with which they are con- nected by short and thick olfactory tracts. The cerebral hemispheres are pear-shaped, broad behind, but tapering rapidly forward, where they decrease in vertical as much as in transyerse diameter. The frontal lobe is short, narrow and of small vertical depth, while the parietal lobe much surpasses it in every dimension ; a transverse depression marks the boundary between the two. The temporo-sphenoidal lobe is also quite well developed and adds materially to the dorso-ventral diameter of the brain in this region. Posteriorly the hemispheres slightly overlap the lateral lobes of the cerebellum (which appears not to be the case in Daphanus), but leave the vermis entirely uncovered. The shape of the cerebrum is thus alopecoid rather than thooid in character. In the former series the hemispheres are wide behind and taper anteriorly, with slight incurvations at the sylyian and presylyian fissures, while in the thooids the cerebrum is narrower behind and at the presylvian fissure the sides are abruptly incuryed almost at a right angle ; the frontal lobes are much larger relatively than in the foxes (see Huxley, ’80, pp. 245— 247). The hemispheres of Cynodictis agree well in shape with those of the alopecoids, and when compared with the brain of the later and more adyanced genus Cynodesmus from the John Day, the greater width of their posterior region is distinctly to be seen. The whole character of the skull makes it evident that Cynodesmus is a thooid, while both brain and skull structure approximate Cynodictis more to the alopecoids. The hemispheres are yery simply convoluted and the sulci are few, simple and short, though it should not be forgotten that the brain-cast yery probably fails to reproduce all of the fissures. In the recent Canide the conyolutions are numerous and complex, and the sulci pursue a remarkably curved course, giving to the conyolutions, when seen from the side, the appearance of a succession of U-shaped, concentric coils, grouped around the sylvian fissure as a centre. In Cynodictis, on the other hand, the visible sulci are few, shallow, short and nearly straight. On the dorsal surface of the hemisphere only two fissures are to be observed, the lateral and the suprasylyian, the former of which is short and almost straight, dying away before it reaches the hinder part of the parietal lobe. If the coronal sulcus is present at all, it is in the same fore-and-aft line as the lateral, and has not the outward sweep around the crucial fissure which is so characteris- tic of Canis. No trace of the crucial fissure is preserved in the brain-cast, and if it was present in the brain, it must have been short, as is indicated by the straight course of the NOTES ON THE CANIDEH OF THE WHITE RIVER OLIGOCENE. 379 lateral sulcus. The suprasylvian sulcus is likewise very short and but little curyed, and is not divisible into anterior and posterior portions. The sylvian fissure itself is but feebly marked upon the cast, but the rhinal sulcus, on the contrary, is yery distinctly shown and extends for nearly the whole length of the hemisphere. Making all due allowance for the fact that a cast of the brain-case can but imperfectly reproduce the features of the brain itself, yet it is clear that the cerebrum of Cynodictis was conyolu- ted in a much simpler way than in any of the existing Canide, and that it retains char- acteristics which among the modern dogs are embryonic and transitory. The cerebellum is rather large and is less overlapped by the hemispheres than is the case among the recent members of the family. The vermis is narrow, but prominent, and is quite clearly divisible into three lobes, corresponding apparently to the lobus cen- tralis, lobus monticuli and declivus of Canis. The vermis is less regularly curved in the antero-posterior direction than in the modern genus, the posterior surface forming nearly a right angle with the dorsal. The lateral lobes of the cerebellum have quite a different appearance from those of the recent Canide. Thus, the lobus quadrangularis is less extended transversely and narrows less toward the external side, while the lobus lunatus inferior is yery imperfectly developed, and the lobi semilunares appear not to be repre- sented at all, or, if present, they must be exceedingly small. This latter point is difficult to decide definitely, because a small fragment of the skull, which cannot be removed with- out danger to the specimen, covers the place where the semilunar would be if present. A small additional lobe, not represented in Canis, lies upon the dorsal surface of the lobus quadratus and near to the vermis. Complex as it looks, the cerebellum of Cyno- dictis is simpler than in the recent dogs. Measurements. No. 10513. Brain, length fr. cerebellum to olfactory lobe (incl.).....-...seseeseeeeeeeeeeteeceennsseceeceeeeeeeeeeteneees 0.045 Olfactory Jobe, fore and aft Ciameter.........--.sccccoseeeccneeceseeecneteectetcecoesctcecseseerenceccresreconseee -005 a CMSVCLtICALGCUAMELCLA a cchebe esc: circ oceee sadee sovetooreeaeet sonselnhaisetastocsbccvascvasessuacesias.« O11 Cerebrum, length insmedian Line:.:.....c....o...seeecnseneserecomeerercnntstecsteaseceaeeeeerccceeseestensensnots 030 ef height at temporo-sphenoidal lobe 025 i width ‘ sf ue Hf -030 Cerebellum, length in median line...........seesseeeeeeeeeeeeseeeeeeeeeeeeesesese naan ee eeseeeeseeeeseesersaeeneees -013 ss TUTTLILA 1 Scscob So0bedecSacRiobn Soba oqnesd000 oBeddadeoadenEseBocdcDcab os annoeaTO Bo Ec oD TODO RIaBODO DE AOCOSOO HES 024 MG Vertical height ......cseceeeeseeeccececceeccecceceeseceeeeeeessssssesseseasceeseseeeseeeeeeeesceneeeeeees 018 .012 Medulla oblongata, width IV. THe VERTEBRAL COLUMN. The backbone is not preserved entire in any of the specimens, but by the aid of the more complete individuals from the John Day, the numbers of the yarious categories of vertebrae may be inferred. ie, 12, Os OID VE 576 NOTES ON THE CANIDZ OF THE WHITE RIVER OLIGOCENE. The atlas (Pl. XIX, Fig. 13) is somewhat more canine in character than that of Daphenus, having a short and broad body and moderately developed transverse pro- cesses. The anterior cotyles are shallower and more depressed than in Canis ; the neu- ral arch is well extended in the antero-posterior direction and is quite smooth, without ridges or tubercles of any kind; it is very strongly convex, giving to the neural canal an almost circular shape. The inferior arch is very slender and has but a rudimentary hypapophysial tubercle. The posterior cotyles for the axis are somewhat more concave than in Canis and present more obliquely toward the median line. The transverse pro- cesses are rather small and are much less extended antero-posteriorly than in Canis, not reaching so far behind the surfaces for the axis, nor so far forward upon the neural arch ; in consequence of this, the atlanteo-diapophysial notch is less deeply incised. The pos- terior opening of the yertebrarterial canal presents backward, as it does in Daphenus, but has shifted a little more toward the dorsal side of the transverse process, thus show- ing a tendency to assume the position which is characteristic of the recent Canidae. The axis is not especially canine in appearance, but rather resembles that of Viverra. The centrum is long, narrow and yery much depressed anteriorly, becoming somewhat deeper vertically toward the hinder end, which has a transversely oval and nearly flat face for the third vertebra; the yentral keel is relatively better developed than in Daphenus. The articular surfaces for the atlas are low and wide, but project much less outside of the pedicels of the neural arch than they do in Canis, and are more convex than in that genus. The odontoid process is slender and elongate, more so than in Viverra, and the articular surface on its ventral side is not, as in Canis, continuous with the lateral facets for the atlas, but is separated from them by a feebly marked ridge. The transyerse processes, which are yery thin and compressed, are of no great length ; they are perforated by the vertebrarterial canal, which is relatively longer than in the recent dogs. The pedicels of the neural arch are short from before backward, but are quite high, and the neural canal is proportionately much larger in both dimensions than in the existing dogs. The neural spine, at least in the White River species, resembles that of Daphenus much less than it does that of Canis. It is long, not very high, and in front extends far in advance of the pedicels, but posteriorly it does not project behind the zygapophyses, as it does so conspicuously in Daphenus; as in the modern genus, the dorsal border of the spine is continued into the hinder margins of the neural arch. The zygapophyses are rather small and do not extend out so prominently from the sides of the neural arch as in Canis. The axis of the John Day species, C geismarianus, as figured by Cope (’85, Pl. LX Xa, Fig. 12), differs from that of C. gregarius in having a much higher neural spine, which is continued posteriorly into a pointed projection, similar to but shorter than that seen in Daphenus. NOTES ON THE CANIDEZ OF THE WHITE RIVER OLIGOCENE. O11 The third cervical vertebra is markedly different from that of Daphenus and quite like the corresponding vertebra of Canis. The centrum is moderately elongate (though shorter with reference to the axis than in most of the modern dogs), quite depressed and slightly opisthoccelous, and has a stout, prominent ventral keel, which is better developed than in Daphenus, or even than in Canis, and ends behind in a tubercle. The ante- rior face is broad, depressed, quite convex and very oblique in position with reference to the fore-and-aft axis of the centrum, while the posterior face is more nearly circular in outline. The transverse process is, in general character, quite like that of Canis, but has a relatively smaller extension from before backward, and is less obviously divided into anterior and posterior projections, the ventral margin of the process being nearly straight. The vertebrarterial canal is proportionately much longer than in Canis, being nearly as long as the entire centrum. The neural canal is relatively larger and especially wider than in the modern genus, while the neural arch is long and broad and but slightly convex on the dorsal surface. One noteworthy difference from Canis con- sists in the fact that the arch does not project over the sides, or pedicels, as an overhang- ing shelf, or does so but slightly. The neural spine is represented only by an incon- spicuous ridge. The zygapophyses are small and extend but little in front of and behind the neural arch, which constitutes a yery marked difference from Daphanus. In the latter, it will be remembered, the neural arches are deeply emarginated between each transverse pair of zygapophyses, so that when the vertebree are placed in their natural position, large vacuities occur between the successive neural arches. In Cynodictis, as in Canis, these interspaces are very narrow and in certain parts of the neck they are hardly at all visible. The fourth vertebra is somewhat shorter than the third, but is otherwise very much like it and also like the corresponding vertebra of Canis. The transverse process is some- what larger and heayier than on the preceding vertebra, and the greater antero-posterior extension of its outer portion makes the yertebrarterial canal relatively longer than in Canis ; the inferior lamella is very thin and light. The neural spine is short and slen- der, but is relatively better developed than in most of the modern representatives of the family. On the fifth cervical the neural spine is higher but more slender than on the fourth. The sixth is not preserved in connection with any of the specimens. The seventh cervical is almost a miniature copy of the same vertebra in Canis ; the neural spine is relatively higher, more slender and more pointed than in most species of the existing genus, and the transverse processes are proportionately longer and thinner, but otherwise the resemblance is very close and detailed. The number of thoracic vertebre cannot, as yet, be definitely stated, because in 378 NOTES ON THE CANID& OF THE WHITE RIVER OLIGOCENE. none of the specimens is the series preserved entire. Probably, however, these vertebree numbered thirteen, as is commonly the case among the recent representatives of the family. The specimen of C. geismarianus figured by Cope (85, Pl. IX Xa) has the posterior ten thoracics in place, and there must have been at least three additional ones. The anterior yertebre of this region have yery small, contracted centra, but long and prominent transyerse processes and neural spines which are relatively higher and more slender than in Canis, and are also inclined more strongly backward than in the latter. Posteriorly the centra become longer, broader and more depressed, and are quite distinctly keeled in the median ventral line. In addition to this median keel are two shorter and less prominent lateral ridges, which, however, terminate behind in distinct tubercles and thus give avery characteristic appearance to these vertebrae. The transverse processes become more and more shortened and the neural spines lower, less strongly inclined, but more compressed and broadened at the base (antero-posteriorly). The antepenultimate thoracic (presumably the eleventh) is the anticlinal vertebra, of which the neural spine is low, broad, compressed and erect. The penultimate (? twelfth) and last (? thirteenth) thora- cies are very much like lumbars in appearance and structure, but have no transverse processes, while in Canis these processes, though small, are quite distinct.on the twelfth and thirteenth thoracics. Large, heavy and prominent anapophyses and metapophyses are present on the last two thoracics. Of lumbar vertebra this genus probably possessed seyen, that many being preserved in position and in connection both with the thoracics and with the sacrum in Cope’s speci- men of C. geismarianus. In the White River material at my command not more than five lumbars haye been found in association with any one individual, but the series is obviously incomplete, and there is no reason to suppose that C. gregarius differed in this respect from the John Day species. The lumbar region is proportionately long and stout and the individual vertebrae are quite massively constructed (7. e. for so small an animal), indicating a powerful musculature in this region. The centra increase in length up to that of the penultimate vertebra, while the first and the last are the shortest of the series. These centra are broad and depressed, and bear distinct median ventral keels, while the lateral ridges and tubercles are present on the first two vertebrae, but not on the last three. The faces are kidney-shaped, slightly conyex in front and concave behind, and are placed obliquely with reference to the long axis of the centra. This obliquity is to provide for the curvature of the loins, which rise to the pelvis, the ramp standing considerably higher than the shoulders. The transverse processes, which are quite short on the anterior lumbars, increase steadily in length up to the sixth, where they become yery long; they are slender, depressed, pointed and curved forward. The neural spines are low, compressed and thin, broad at the base, narrow and pointed at NOTES ON THE CANIDA OF THE WHITE RIVER OLIGOCENE. 379 the tip, and are inclined forward rather more decidedly than in Canis. Anapophyses are quite prominent on the anterior lumbars, but diminish posteriorly, becoming rudi- mentary on the fifth, while the metapophyses are conspicuous in all. The zygapo- plyses are but moderately concave and convex respectively. The general aspect of the lumbar region is not canine in character, but rather resembles that of the civets and mustelines. The sacrum is quite short and consists of three yertebree, only the first of which has a contact with the ilium. The first sacral has a broad and much depressed centrum and large, expanded pleurapophyses, which give considerable width to the vertebra. The neural spine is a mere feebly marked ridge, while the spines of the second and third are higher and separate. The transverse processes of all the sacrals are fused into a continu- ous lateral ridge, but that of the third vertebra extends outward much farther than the others and ends in a point, an arrangement which gives to this sacrum an appearance quite different from that of Canis. The prezygapophyses of the first vertebra are large and conspicuous, but all the other zygapophyses of the sacrum are small. The neural foramina are remarkably small. The centrum of the last vertebra is almost as large as that of the first and the widely extended transverse processes make the sacrum nearly as broad behind as it is in front. The caudal vertebrw are not preserved entire in any of the specimens, nor, indeed, can all of them be recovered from all the individuals combined, so that the number of tail vertebree is, as yet, conjectural. However, enough remains to show the character of the tail and of the various elements which compose it. The tail was evidently very well developed, being relatively longer and stouter than in any of the recent Canidae, and much like that of some of the long-tailed viverrines, such as Herpestes. ‘The anterior caudal vertebrae have short, but heavy centra and very long, broad and depressed trans- verse processes, which extend out nearly at right angles with the line of the centrum. The breadth of the first caudal across the transverse processes about equals that of the last sacral. The zygapophyses of the anterior caudals are large and prominent. ‘The ante- rior caudals are succeeded by a number of vertebrae with very elongate centra, which resemble in miniature the corresponding vertebrae of Daphenus, having distinct remnants of the various processes. Toward the tip of the tail the vertebrae become very slender and of a cylindrical shape, the centra being slightly contracted in the middle and expanded at the ends. The ribs, so far as they are preseryed in the yarious specimens, are remarkable chiefly for their length and slenderness and for their subeylindrical shape. Tubercles appear to be absent from the twelfth and thirteenth pair. The sternum is of the usual carnivorous character, without being especially like that either of the dogs or of the 380 NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. civets. The manubrium is long, more so than in Canis, as well as narrower and more compressed. The first pair of ribs is attached to a pair of wing-like processes, which are unusually far from the second pair. In front of these processes the bone is com- pressed and very narrow. For much of its length the manubrium possesses a ventral keel. The segments of the mesosternum, so far as they are preserved in the various specimens, are more elongate, more slender and depressed and more contracted in the middle than in the recent Canide. Measurements. No. 10493. No. 11012. No. 11381. | No. 11382. | No. 11432. (Atilastplen other accccssscecsoccsssesecestocaccssestensasee ereenacesneanaesaseer 0.016 | | Hed | COME DT CAC tN ssasserscctecrsessseccrscciecsecsesdoccrcccresadccssessesercecocss 0384 | | | Axis, length (excl. of OGontoid) ..........0+:cseseeeeeeeeseeeeeeeeeeeeeees 019 | =a 020 | | e CHES OM OC ONCOIDsPLOCESS -c--bsee-caen-sevesstceacestesceressoesee- 005 | | Co ibread thiofian terior faCe-v-cs-cse-s-sesensscsvasevncpcesaeccser-soecees 0013S}; -0135 Third cervical, length............0.sccccsescevesccsescncccencsscoescsveracs -O11 013 | 012} } 013 Fourth ‘‘ Se | | -014 Fifth “ “ | | 013 Sixth sf 013 | | -012 Seyenth ‘ ee 011 | | .010 Anterior; thoracie.lenethh cssssccecse-ctececcstets soe cestotececcessceeeeeeoes | .008 009 | | .0085 Last thoracic, length 012 | 012: | -013 | First lumbar, .015 013 | Second * -017 -0145 | Fifth s .016 -018 016 | ; 66 © WIth POst. [ACO .r.ceveseeeoseosescenvecscecesccssesncssusns> -010 -O11 -009 | Six Un boeo meen eth -cesecassaesesstesseceesct cneecssfeacerensueneaancntacnts 015 017 014 | Seventh ‘‘ a Meteecnenccersceataserans sasusenccencacentosrccsremmeacese ce } .013 013 | -012 Saclum leno thccssscastssccsncnsestenancsstencnsrorussncreser=nesterastcrgenes -024 -026 First sacral, width across pleurap..........-0--eesseseeeeeneeeeeeeeeereee -024 -024 | Third ‘ " SCG CranSVey pics sessaceanbacraessoncsoueseeresssac -021 | hirstycandal glen ctiteswecssctessecesssscassearsresssocersscrnececssncoreenes -007 .008 2.010 | ‘* width across transy. pr... -021 -026 | Median caudal, length ...-.....00..0ss.ceccesccesvscccscanssecssenecncsseces rf Es WAdthVAM ACG s scsccsovea-csecsceccuccssbusseesscaeeces -005 | V. Tae Fore Live. The scapula is quite remarkable and is in character rather viverrine or raccoon-like than canine. The shoulder blade is rather low and broad and is divided by the spine into pre- and postscapular fossze of nearly equal breadth, while in the modern dogs the scapula is high, narrow and of subquadrate shape, and has the spine so placed as to make the postscapular fossa much the larger of the two. The glenoid cavity is moder- ately concave, and is elongate antero-posteriorly, but narrow transversely. The coracoid NOTES ON THE CANIDZ OF THE WHITE RIVER OLIGOCENE. 381 process is unusually large, forming an incurved hook, which, however, does not appear prominently when the scapula is viewed from the external side ; in the recent Canida the coracoid is reduced to much smaller proportions. A resemblance to the shoulder-blade of Canis is to be found in the broad neck of the scapula and in the absence of any well- defined coraco-scapular notch. The coracoid border is slightly concave at the neck, but then curves forward and upward, giving great width to the prescapular fossa; the gle- noid border is, as usual, straight and is steeply inclined, so that the postscapular fossa, which is very narrow distally, becomes very broad proximally. The spine is high and ends in a very long and prominent acromion, which descends below the level of the gle- noid cavity, which suggests that in this genus the clavicles were much better developed than in the existing dogs. A very large metacromial process is also present. The meta- cromion may be observed in most of the existing families of Carnivora, but it is seldom so large and so prominent as in Cynodictis ; perhaps, the nearest approach to it among modern genera is in Arctictis. The humerus is much more suggestive of viverrine than of canine affinities. As compared with the bones of the forearm, or even with the femur, the humerus is elon- gate, but it is short in proportion to the length of the back or loins. The head is strongly convex and projects farther behind the plane of the shaft than in the modern dogs ; the external tuberosity is a heavy, but low ridge, which barely conceals the head when the bone is viewed from the front; a large, irregularly circular area near the hinder end of this ridge plainly indicates the insertion of the infraspinatus muscle. The external tuberosity is both lower and shorter than in the modern dogs, but the inter- nal one is rather more prominent, and the bicipital groove is more widely open, more internal in position and more of it is visible from the anterior side. The shaft is rather long, and, when seen from the side, exhibits a sigmoid curvature, which is somewhat better marked than in Canis. For most of its length, the shaft is laterally compressed and has but a very short cylindrical portion before expanding laterally at the distal end. Most of the ridges and prominences for muscular attachment are well developed, more so than would be expected in so small an animal. The deltoid ridge is much more prominent than in the recent dogs, and is more like that of the cats and yiverrines ; the supinator ridge is likewise very much more prominent than in Canis, in correlation’ with the power of rotation of the radius, which Cynodictis appears to have retained in almost undiminished degree. On the other hand, the rough ridge, which runs down from the head upon the outer side of the shaft (spina humeri) and serves for the attach- ment of the teres minor, anconeus externus and brachialis internus muscles, is much fainter than in Canis and the linea tuberculi minoris is very feebly marked. The supra- trochlear fossa is very shallow and the anconeal fossa is much smaller and shallower than in the modern representatives of the family, there being no perforation of the shaft 382 NOTES ON THE CANID® OF THE WHITE RIVER OLIGOCENE. at this point. The internal epicondyle is much. more prominent and more massive than in Canis, and a conspicuous epicondylar foramen is present, in the form of a long, nar- row slit. The external epicondyle, on the contrary, is rather smaller than in the recent genus. The humeral ¢froch/ea has a much smaller proximo-distal diameter than in the exist- ing Canidae, in which respect it preserves a primitive character and resembles the troch- lea of such viverrine genera as Cynogale and Viverra. The radial surface is small and simply conyex, while the ulnar facet is much larger than in the recent dogs; the inner flange of the ulnar facet is also more produced distally and forms a sharper edge than in the latter. The radius is not at all suggestive of canine affinities, but rather resembles the cor- responding bone of the cats and viverrines. The capitellum is small and of subdiscoi- dal shape ; while it is somewhat more extended transversely than in Fe/is, it is much less so than in Canis ; its articular surface is moderately concave and is slightly notched on the anterior border. The proximal facet for the ulna is a simple, convex band, separated from the humeral surface by a distinct angle and entirely resembling that of Daphenus. The character of the articulation at the elbow-joint and the large development of the supinator ridge on the humerus would seem to imply that in Cynodictis a considerable degree of freedom in the rotation of the manus had been preserved, though probably less than in the cats and in many viverrines. The bicipital tubercle is prominent, but occu- pies a more posterior position than in either the cats or the recent dogs, and is not visible when the radius is looked at from the front. The shaft of the radius is relatively short, slender and rounded, very different from the broad, oval and antero-posteriorly compressed shaft seen in Canis; it has a slight double curvature, arching anteriorly and externally, and is of almost uniform thickness throughout its length, except at the distal end, where it broadens considerably. A very striking difference from Canis consists in the very great size and prominence of the sty- loid process, which forms a relatively enormous tuberosity ; it is even much larger pro- portionately than in the cats or civets and is as large as in Je/livora, though of a differ- ent shape. In Daphenus, as we have already learned, the styloid process is very promi- “nent and of a generally feline appearance, but it is proportionately smaller than in Cyno- dictis. The radius figured by Schlosser (’89, Taf. VII, Fig. 8) and by him attributed to one of the European species of the latter genus has a styloid process in the form of an enormous, recurved hook, much longer and much more slender than in the American species and of an entirely different appearance. The distal tendinal sulci are not very well marked, though that for the abductor and extensor muscles of the pollex is a deep groove. The distal facet for the ulna is smaller and less deeply impressed than in Canis. The carpal facet is small and slightly concave, narrowing toward the internal side; it NOTES ON THE CANID® OF THE WHITE RIVER OLIGOCENE. 383 does not extend over upon the styloid process, from which it is separated by a broad and deep notch. The wlna is, in its way, as peculiar as the radius. The olecranon is quite typically fissipede in character and differs from that of the creodonts in its comparative shortness and breadth ; though proportionately somewhat longer than in Canis, it is hardly so long as in Daphenus, and the suleus for the tendons of the anconeal muscles is more distinct than in the former. The sigmoid notch is hardly so deep as in Canis, and, in particular, the internal facet for the humerus projects less in front of the plane of the shaft, and the external process is very feebly developed. The radial facet is narrower and less deeply concave than in the modern Canide, but has a somewhat greater vertical diameter. The shaft of the ulna is decidedly less reduced than in the recent representatives of the family, and for most of its length is little or not at all more slender than that of the radius. In its proximal portion the shaft is much more compressed laterally and thicker antero-posteriorly than in Canis, in which genus this portion of the shaft is trihedral. The middle and distal portions are of triangular section, none of it having the subcy- lindrical shape which characterizes the distal one-third of the shaft in the recent genus. The distal end has quite a different shape from that seen in Daphenus, a difference which ‘is due to the much greater prominence of the radial facet in the latter. In Cynodictis this facet is almost sessile and projects but little more than it does in Canis, The car- pal facet is very small and quite simply conyex. Measurements. No. 11381. | No. 11382, | No. 11432. No. 10493. No. 11012. Scapulawleng thieeresstee-osteeecseteeee senceasccscccaseasecesaneccnmneccst 0.054 oo Mmm OTeALCSHWIO bNwrereteneccncere sence stecsnceeereeceetenan taccnwm ens 2.049 | at GWAC EHYO Len EC kotecatsonectrecesnentccoscereecoccecuscesecess retecioes 013 | ant. post.-diameter of glenoid cavity ..... | -012 -0095 -0095 sf transverse sf | 008 .007 007 Humerus, length... 20-27.-.02..0c.0cccresceccnssecececcerseceeccenstreneceeres 075 070 070 af ant. post. diam. prox. Nd ....---.sseseeeeeeesseeeeeeceeeeee NOL OLS re 2 O19 019 -O15 “ — transv. veal cegensce SMIE E.SUN 8 eat LAs O14 016 O13 0125 4 breadth of distal end... (eae sO1G8 Failte 2020 015 sa i Glia ROve] NESC reciga uD e Oo RIc ECE aT SOE LEE Secor SO ScEERCEE Hee Ou wee Ol45ipale Old. Radius, length .....---.ceessscesercssscnecesccsecerescensercecersceesecencceees 057 O6L ‘ant post. diam. prox. Nd. -.-----0......ceeceecenecnersesen cree -005 006 =| .007 .005 -005 «« transy. f§ fs SNe ie rag HosSoeDaGa Cobo A aEcoadcookeodoecs 007 - .007 |. +: 009 007 007 ** breadth of distal end O12 -013 .013 | -009 fe Crema Hoar altace we veccessires set cwsseeeeecon sun testcseees 0055 | .CO6 .007 .0055 Wilnae Vengthieess.-s2-cccscncscecseecesecerccnstrnseccnsecsncscrcnessnsitecessnes -072 OG CEO PA OLCCTATIONS sacc reset co cece ceeeis colocle o- cadlateeosistse ae aleeelevle 007 010 0095 .009 “* thickness of olecranon .010 | .008 .008 ie es SG, Sad, Wy, 384 NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. VI. THe Manus (PI. XX, Fig. 23). By a fortunate discovery of Mr. Hatcher’s, Iam enabled to give an account of an almost complete carpus belonging to Cynodictis, which has hitherto been entirely unknown. A scapho-lunar is present, formed by the coalescence of the scaphoid, lunar and central, which distinguishes Cynodictis from the creodonts. This bone resembles that of Canis in general character, but displays quite a number of differences in points of detail, and these differences are, at the same time, approximations to the structure found in Daphenus. 'The scapho-lunar has a very small vertical (proximo-distal) diameter, especially on the radial side, where it thins away to a mere edge, the facets for the radius and the trapezium almost meeting. As compared with the corresponding carpal of Canis, this bone has a somewhat greater transverse and smaller dorso-palmar diameter. The radial facet is simply convex both transyersely and antero-posteriorly, and has not the saddle-shaped extension at the interno-palmar angle which is found in the recent dogs. This facet descends quite low upon the dorsal side of the bone, as is also the case in the modern plantigrade and semiplantigrade carnivores. The hook-like process which arises from the postero-internal angle of the scapho-lunar is much shorter and less mas- sive in eyery dimension than that of Canis. Another difference from the modern genus | consists in the absence of any distinct articular surface for the pyramidal, the facet for the radius and that for the unciform almost coming into contact along the ulnar side of the bone. On the distal side of the scapho-lunar are four facets, for all the carpal elements of the distal row. That for the unciform is relatively smaller than in Canis, and is con- fined to a narrow strip near the ulnar border; the magnum facet is much the same as in the modern genus, but is somewhat more oblique in position. The surface for the tra- pezoid is fairly large and keeps more nearly parallel with that for the magnum than in the recent dogs, while the trapezium facet is small and of almost circular shape. The pyramidal is a very different-looking bone from that of the modern dogs, being broad, depressed and scale-like in shape; its vertical (or proximo-distal) diameter is very small and relatively much less than in Canis, and there is no such process from the ulnar side of the bone as in the latter, in which the pyramidal articulates with the head of the fifth metacarpal by a much more extensive facet than in Cynodictis. The recent yiverrines haye the pyramidal shaped very much as in the White River genus. The proximal surface is divided into two narrow and somewhat concave facets for the ulna and pisiform respectively, of which the latter is slightly the larger. On the distal side is a single large and coneaye facet for the unciform, and posterior to this NOTES ON THE CANIDA OF THE WHITE RIVER OLIGOCENE. 385 avery narrow surface which appears to be destined for articulation with the head of the fifth metacarpal. The pisiform differs very decidedly in shape from that of Canis. This carpal is small and light; its proximal (7. e., articular) end is greatly depressed, but much extended transversely (in the existing genus the principal diameter of the proximal end is the vertical one) and the facets for the pyramidal and ulna are correspondingly broad- ened transversely and narrowed vertically. The pyramidal facet is the larger of the two and is quite deeply concave, while that for the ulna is small and nearly plane; the two facets together form an acute angle and are separated only by an inconspicuous ridge. The distal end of the pisiform is moderately expanded, but in the vertical dimension, so that the proximal and distal expansions are almost at right angles with each other. Between the two expansions the body of the bone is much contracted and very slender, which is in marked contrast to the shape seen in Canis. A so-called “radial sesamoid” appears to haye been present; at least, there occurs in the same block of matrix through which the carpals of one individual were scattered, a small, irregularly wedge-shaped bone, to which I can give no other interpretation. Assuming that this reference is correct, we find in the relative size and shape of this bone another resemblance to such yiverrine genera as Herpestes, Cynogale and Paradoxurus, etc. The radial sesamoid also occurs in Canis, at least in certain species, but is very minute. The trapezium is very small and differently shaped from that of Canis; its princi- pal dimension is the dorso-palmar, while the transverse diameter is the least. The sur- face for the scaphoid, which in Canis is a very oblique, convex facet, is in Cynodictis entirely proximal in position and nearly plane, and there is no such large concave facet for the trapezoid on the ulnar side as in the modern genus; the distal facet for the head of the first metacarpal is less distinctively saddle-shaped than in the latter. In view of the well-developed pollex, the small size of the trapezium is somewhat surprising. The trapezoid is shaped very much as in the existing dogs, but with certain minor differences, especially noticeable in the very small vertical diameter and in the thinning of the bone to an edge on the ulnar side. The proximal end bears a simply convex facet for the scapho-lunar, while the distal facet, for the second metacarpal, is very slightly saddle-shaped ; on the palmar side the trapezoid contracts to a point. The magnum is small and that portion of it which is visible from the dorsal side, when all the carpal elements are in their natural positions, is minute, especially in its proximo-distal dimension. In shape the magnum does not differ materially from that of the recent dogs, but the proximal surface is narrower and rises more abruptly to the “head,” and on the palmar side the bone broadens out in a fashion not repeated in Canis. 386 NOTES ON THE CANID® OF THE WHITE RIVER OLIGOCENE. The unciform facet is large and plane and does not rise so high upon the head as in the modern genus. On the radial side we find no distinct facet for the trapezoid, which, as already mentioned, thins to a mere edge toward the magnum, but there is a well-defined facet for the projection from the head of the second metacarpal, which is proportionately larger than in Canis. On the distal end of the magnum is a narrow facet for the third metacarpal, a facet which is less concave in the dorso-palmar direction than in the case of the last-named genus. The wnciform is viyerrine rather than canine in character, being much narrower in proportion to its vertical height than in the recent dogs. The facet for the scapho- lunar, which in Canis has an almost entirely proximal position, is in Cynodictis much more nearly lateral. The pyramidal facet is also decidedly more steeply inclined than in the existing genus, the two articular surfaces meeting at a very acute angle and mak- ing the proximal end of the unciform narrow and wedge-shaped. On the radial side is a large facet for the magnum and a small one, confluent with it, for the extension from the head of the third metacarpal. The distal facets, for the fourth and fifth metacarpals respectively, are narrower than in Canis, contracting especially toward the palmar side. The metacarpals, five in number, are remarkably short, slender and weak and have but little resemblance to those of the recent dogs. The first metacarpal is yery small, but is, nevertheless, proportionately much less reduced than in Canis, taking the length of me iii in each genus as a standard of comparison. The head is thicker and relatively heavier than in Canis and on the radial side, internal to the trapezium facet, is a tubercle for the attachment of the lateral liga- ment. The facet itself is much less deeply concaye transyersely than in Canis, but more convex in the dorso-palmar direction. The shaft is short, slender, arched toward the dorsal side, antero-posteriorly compressed and of oval section, tapering considerably toward the distal end. The distal trochlea is very small, but formed entirely like those of the other metacarpals ; it is strongly conyex, almost hemispherical and bears a dis- tinct carina upon the palmar face, just as in Daphanus. In Canis, on the other hand, this structure is of an entirely different character, forming an asymmetrical hemicy- linder, with a broad shallow groove placed somewhat internal to the median line, and thus resembles the trochlea of a phalanx rather than that of the other metacarpals. The second metacarpal is represented in the collection only by a single imperfect specimen, consisting of the proximal end. This shows a much stouter shaft than me i, being of about the same diameter as the corresponding portion of me iy, and more slen- der than that of me iii. The head is narrow and bears a saddle-shaped facet for the trapezoid, but sends out a projection which rises more above the head of me iii than in Canis and articulates with the magnum by a larger facet than in that genus. NOTES ON THE CANIDA OF THE WHITE RIVER OLIGOCENE. 387 The third metacarpal, though short and slender, is somewhat the longest and heay- iest of the series. The proximal articular surface for the magnum is shaped very much as in Canis, but is slightly broader in proportion and rather more concave transversely ; on the radial side of the head is a large facet for me ii, which has a more oblique position than in the modern genus. On the ulnar side is a small projection which abuts against the unciform and is relatively larger than in Canis. The shaft, and indeed the whole metacarpal, has a yiverrine rather than a canine appearance ; it has not acquired the prismatic, quadrate shape which is so characteristic of the modern dogs, but is of oval section and is of almost uniform width throughout, but broadens slightly at the distal end. The distal trochlea, though much lower in the vertical diameter, is yet of decidedly more canine character than is that of Daphenus, being broad and hemi- cylindrical in shape instead of subspherical. The pit above the trochlea, which is absent in Daphenus, is distinctly marked and the lateral processes for ligamentous attachment are much less prominent. All of these conditions are approximations to the conditions seen in Canis. The fourth metacarpal is not completely preserved in any of the specimens, but it appears to have been of about the same length as me ii and to have formed with it a symmetrical pair, although the two metacarpals are not so closely appressed as in Canis, but diverge slightly toward the distal end. The head has a simply convex facet for the unciform and is somewhat narrower proportionately than in the existing members of the Canide, owing to the overlapping of the head by me ii, in order to reach the unciform. So far as it is preserved, the shaft is rather more slender than that of me in and of a more cylindrical, less compressed shape. The fifth metacarpal is remarkably short, much more so in proportion to the length of me ili than is that of Canis. ‘The head is less broadened and thickened than in the latter genus, and carries a simple, convex facet for the unciform. In the modern genus there is likewise a large facet for the pyramidal, which extends down over the unciform and comes into contact with me vy. In Cynodictis there appears to be a facet of a simi- lar kind, but if so, it is very small and obscurely marked and may be regarded as in only an incipient stage of development. The shaft is slender proximally and broadens dis- tally, the reyerse of the proportions which obtain in Canis, and the distal trochlea is small and is of somewhat more spherical, less cylindrical, shape than in the existing members of the family. The phalanges. It is unfortunate that in all of the specimens in the collection the phalanges are in such a fragmentary state that only an incomplete account of them can be given, and some important questions must be left unanswered for the present. The proximal phalanx of one of the median digits is short, slender and straight, and is rela- 388 NOTES ON THE CANIDE OF THE WHITE RIVER OLIGOCENE. tively broader but more depressed than in Canis. As in Daphenus, the proximal articu- lar surface is somewhat more deeply concaye and presents more obliquely toward the dor- sal side than in the recent genus. The distal trochlea likewise resembles that of Daphe- nus in having a deeper median groove and in being more confined to the palmar aspect of the bone than in Canis, which has the distal trochlea reflected well over upon the dor- sal side of the phalanx. Of the second phalanx only the proximal half is preserved in any of the specimens, and I have so far failed to find even a fragment of the distal end. So far as can be judged from the material at hand, Cynodictis would appear to have differed from Daphe- nus in the very important respect that the claws were not at all or only very imperfectly retractile. In Daphenus the asymmetry of the second phalanx is clearly displayed even in its proximal portion, while in Cynodictis the proximal end is quite symmetrical and does not possess any depression or excavation upon the ulnar side. However, a certain resemblance to Daphenus and difference from Canis may be observed in the greater con- cayity and more marked separation of the two pits into which the proximal facet is divided, as well as in the greater prominence of the beak-like process which rises from the dorsal margin and fits into the median distal groove of the first phalanx. In the absence of the distal end of the second phalanx, it cannot be positively stated that Cynodictis had lost (or had never possessed) all trace of the retractility of the claws, but it does not seem unlikely that such was the case. Measurements. No. 10193. No. 11012. Carpusaiheightyinrmediany line crscsacscecssccecacescererececescccsscaesunenceestonesncanecresstsensieerses=arastatccalacsess 0.006 SS PDTOAC EN acccscaccca ccc cetncscesesceesiserserssscccuveressdectosess sedstenateusantoosereceretortetreratiiecsstsec entree ee O11 Metacarpallilenpthicscss-.pecsest A. P. S.—VOL. XIX. 3 A. LAL 2 C Soe ¢ Ay pee ~S¢EMaM J Gnr.TDptl. ARTICLE IX. CONTRIBUTIONS TO A REVISION OF THE NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. (Plates XXI-XXV.) BY SAMUEL N. RHOADS. Read before the American Philosophical Society, May 6, 1898. An unusually fine series of the skins and skulls, with reliable data and measure- ments, of the beavers, otters and fishers of the United States and Canada having lately come into the custody of the writer, it is thought advisable to publish the results of a study of the various nominal forms of these mammals and briefly discuss the nomencla- ture involved. Owing to a lack of specimens from some regions whose faunal condi- tions are known to produce in many other mammals well-recognized geographic varia- tions, this paper must be considered rather as a contribution to the subject, and in no sense a complete synopsis. The area covered by this study comprises solely that part of North America north of Mexico, no attempt being made to discuss the relationships of the tropical species. To Mr. Outram Bangs the author acknowledges his gratitude for a most valuable loan of skins and skulls of nearly every species and race recorded in these pages. To the kindness of Mr. F. W. True, of the National Museum, is due the loan of a series of skulls of the Alaskan otter. The North Carolina Department of Agriculture has courteously loaned two skins and four skulls of beavers recently killed in Stokes county of that State through the kind offices of Mr. H. H. Brimley, the Curator of the State Museum. Aid has likewise been generously given by Dr. J. A. Allen, Dr. C. Hart Merriam, Dr. T. 8. Palmer, Mr. Gerrit 8. Miller, Jr., Dr. M. W. Raub and Mr. C. 8. Brimley. THE BEAVERS OF NORTH AMERICA. Contrary to evidence which must eventually be accepted by all zodlogists, the Ameri- can beaver, Castor canadensis Kuhl, is still considered by many eminent authorities as 418 CONTRIBUTIONS TO A REVISION OF THE specifically the same as the Castor fiber Linnzeus of Europe. In 1897, Dr. E. A. Mearns described* a subspecies of the typical Canadian animal, naming it Castor canadensis Jrondator and assigning its habitat to the ‘southern interior area of North America, ranging north from Mexico to Wyoming and Montana.” This appears to be the first attempt in literature to formally subdivide the American beaver, a species whose con- stancy of characters over the vast and varied habitat which it frequents had hitherto been unquestioned. There can be no doubt as to the tenability of Dr. Mearns’ “ Broad-tailed Beaver ” as distinguished from the Hudson bay animal, whose habitat Kuhl designated as “ad fretum Hudsoni” in his original description of canadensis. It is probable that the beavers inhabiting the Carolinas, Georgia, Alabama, Missis- sippi and Tennessee are equally entitled to subspecific rank. So rare has the beaver become in these States, however, it would probably be impossible to verify such a predic- tion with specimens now in our museums.+} From what we know of the relationships of the representatives of our eastern species inhabiting the Pacific slope, we are led to expect that the beaver of that region would also prove separable from canadensis. A yery complete series of skulls, with three adult and three young skins from the Cascades of Washington and Oregon, shows this to be the case. Fortunately the synonymy of the American beaver is not involved and requires no elucidation in this connection, as is shown by reference to Dr. J. A. Allen’s Monograph of the North American Rodentia. A synopsis of the American forms is herewith pre- sented. Canavan Braver. Castor canadensis Kuhl. Plate XXI; Fig. 3. Plate XXII; Fig. 3. Castor canadensis Kuhl, Beitr. Zool., 1820, p. 64. ?“ Castor americanus F. Cuvier, Hist. des Mam. du Mus., 1825” ( fide Brandt in Kennet. Sdugt. Russl., 1855, p. 64). Castor fiber americanus Richardson, Faun. Bor. Amer., I, 1829, p. 105. Castor fiber var. canadensis J. A. Allen, Monog. N. Amer. Rod., 1877, Pp: 444. Type Locality— Hudson bay (“ad fretum Hudsoni” Kuhl). Geographic Distribution.—Northeastern North America, from the northern limit of trees south to the United States and west to the Cascade mountains ; intergrading east of the Mississippi river into subspecies carolinensis, south-centrally into subspecies fron- dator and westwardly into subspecies pacificus. * Proc, Nat. Mus, Vol. XX (adv. sheet, March 5, 1897). T As will be seen later, such specimens have since come to hand and are described as Castor canadensis carolinensis. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 419 Color.*—Winter pelage, above, including sides, dark bay or blackish brown, tip- ped with chestnut or russet, becoming pure chestnut on top and sides of head and on chin, jaws and sides of neck. Rump and thighs purer chestnut. Ears black. Hair of feet, legs and under parts seal brown. Anatomical Characters.—Size, smallest of the American forms. Scaly portion of tail more than twice as long as wide; hind foot with claw about 175 mm. Skull wide for its length ; maximum size of skull 136 by 99 mm. ina New Brunswick example, No. 31, collection of E. A. and O. Bangs. Rostrum and nasals relatively short and wide, the nasal bones averaging more than half as wide as long and extending but little behind the premaxillaries. Upper molar dentition wide and heavy, the crowns oblique, triangular and very wide anteriorly. Measurements.—Of a large, typical, adult male specimen from Quebec, No. 5825, collection of E. A. and O. Bangs (measurements made by collector from newly killed specimen). Total length, 1130 mm.; tail vertebree, 410 mm.; scaly portion of tail (dry meas. from skin), 263 by 122 mm.; hind foot, 176 mm.; length of skull, 132 mm.; breadth of skull, 95 mm.; length of nasal bones, 46 mm.; breadth of nasals, 21.4 mm.+ Remarks.—The above diagnosis is taken mainly from the Quebec specimen, because of the authentic measurements and superior condition of the skin and pelt. The aver- age beaver from the Hudson bay regions, however, is somewhat lighter colored than this specimen, which, in its darkness and richness of shade, rivals the best examples of paci- ficus. In size, and ratio of length to width, the skull of the Quebec specimen is typical, but the nasals are too narrow to serve as a standard for canadensis, whose nasals average wider than pacificus and narrower than frondator. In general terms, canadensis differs from frondator in smaller size, narrower tail, much darker coloration and narrower nasals. It differs from carolinensis in smaller size, narrower, longer nasals and somewhat darker coloration. From pacificus it differs in smaller size, lighter coloration, wider nasals and broader skull. Subspecies pacificus differs from frondator in larger size, greatly nar- rowed and lengthened tail-paddle, rostrum and nasals, and in its dark coloration. In color frondator is decisively and uniformly lighter than eastern canadensis and carolinen- sis and western pacificus, but darkened canadensis (not melanistic) are nearly as dark as pacificus. In size, pacificus is much the longest of the three, with very long hind foot and tail. Its skeleton is slenderer and weaker in eyery part as compared with the massive frame of canadensis and frondator of same age. Carolinensis is nearly of the color of * Ridgway’s Nomenclature of Colors is the standard used throughout this paper. + The narrow nasals of this specimen are an exception, the average of several east Canadiau specimens showing the ratio of length to breadth as less than two to one. 420 CONTRIBUTIONS TO A REVISION OF THE lighter hued canadensis, but agrees with all the other characters of frondator, to which it seems most nearly allied in cranial and caudal characters. Specimens Examined.-New Brunswick, 1 skull; Quebec, 1 skin with skull ; Canada (?), 3 skulls, 1 skeleton, 2 mounted skins; Ft. Simpson, N. W. T., 1 mounted skin; Idaho, 1 skin with skull. CarotiniAN Beaver. Castor canadensis carolinensis, subsp. nov. Plate XXIII; Figs. 1 and 2. ’ 5 Type Locality—Dan river, near Danbury, Stokes county, North Carolina. Type No. z.607, old ad. 3, in the collection of the North Carolina State Museum, Raleigh, N.C. Collected by a trapper in flesh for the Museum, April, 1897. Geographic Distribution —Carolinian fauna, south into the Austroriparian. Color.—Of type and topotype: Overhair of upper head, neck, back and sides, bright hazel. Underfur of same parts, seal brown. Hinder back and rump lightening from hazel to cinnamon rufous and then to tawny olive near base of tail. Vent and under base of tail, dark, rich burnt umber. Ears pale blackish. Sides of head below eyes light hair brown, shaded with pale cinnamon rufous. Feet bistre. Below, from throat to vent, dark broccoli brown with wood-brown tips to overhair. Anatomical Characters. —Size large, larger than canadensis, with relatively much broader tail, as in frondator. Skull large and broad, with very short, broad nasals. In the type the base of nasals does not reach back to the line connecting the anterior walls of the orbits. Ros- trum very short and broad. Audital bulle remarkably contracted laterally, with a strongly developed osseous column on the outer wall and the transyerse diameter less than the longitudinal. Incisors weak, narrowed; molars large, with triangular crowns. Pelage short and harsh as compared with canadensis. Measurements. —Of the type, from carcass: Total length, 1130 mm.; scaly portion of tail, 279 by 158 mm.; hind foot, 184 mm.; ear, from crown, 21 mm.; length of skull, 148 mm.; breadth of skull, 107 mm.; length of nasals, 43.5 mm.; breadth of nasals, 29 min. Of the topotype (ad. %): Total length, 1080 mm. ; scaly portion of tail, 260 by 146 mm.; hind foot, 174 mm.; ear from crown, 23 mm. Remarks.—The two skins and four skulls upon which the above diagnosis of carol- nensis is based were secured, just before the completion of this paper, from the authorities of the State Museum of North Carolina. They are intended to form a group exhibit in the State Museum, and have been carefully measured by the curator, Mr. H. H. Brimley, while yet in the flesh, The old male which forms the type had lost one of its fore feet, NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 421 apparently in a trap, some years previous to its final capture, but its evident health and great size show that it had suffered little inconvenience from the loss of the member. The strong cranial and caudal affinities which this beaver shows to frondator as dis- tinguished from canadensis indicate that it is more closely related to the western form. In color, however, it shows a nearer approach to canadensis, as, in fact, do many other animals of similar distribution and racial differences. The Mississippi and Louisiana beavers are undoubtedly, from what I can hear from the furriers, the darkest and thin- nest pelted of our American beavers, but their separability from what I have named carolinensis is not probable. They may be considered as belonging to carolinensis rather than to frondator. 7 Specimens Examined.—Stokes county, North Carolina, 4. Sonoran Beaver. Castor canadensis frondator Mearns. Plate XX1; Fig. 2. Plate XXII; Fig. 2. Castor canadensis frondator Mearns, Proc. U. 8. Nat. Mus., XX, ady. sheet, Mar. 5, 1897. Type Locality—San Pedro river, Sonora, Mexico, near monument No. 98, of the Mexican boundary line. Geographic Distribution.—Southern interior of North America from Mexico to Wyoming and Montana, intergrading northwardly into canadensis, southeastwardly into the trans-Mississippian carolinensis and westwardly into pacificus. Color—Much paler than canadensis or carolinensis. ‘“ Above russet, changing to chocolate on the caudal peduncle above and to burnt sienna on the feet ; toes reddish chocolate. Below grayish cinnamon, brightening to ferruginous on the under side of caudal peduncle. Sides wood brown enlivened by the tawny-olive color of the over- hair.”* A specimen from Red Lodge, Montana (No. 32, collection of E. A. and O. Bangs), taken in November, is wood brown above and below, the longer overhair of upper pelage washed with pale rusty. Anatomical Characters.—Size large, exceeding average of Hudson bay beaver, with a longer foot and broad tail. Scaly portion of tail less than twice as long as wide, hind foot with claw about 185 mm. Skull massive, large, with short rostrum and very wide, short, tumid nasal bones, the average skull probably exceeding canadensis in size, cer- tainly exceeding it in relative width to length and in the relative breadth of the nasals. Upper molar dentition as in canadensis. Measurements.—Of the type: Total length, 1070 mm.; tail vertebree from anus, 360 mm.; scaly portion of tail, 290 by 125 mm.; hind foot, 185 mm.; length of skull, 153 * Quoted from Dr. Mearns’ original description (/. c.) of type. 422 CONTRIBUTIONS TO A REVISION OF THE mm.; breadth of skull, 99 mm. Maximum length of old males, measured by Dr. Mearns, 1150 mm.; of the tail paddle, 285 by 155 mm. Remarks. Dr. Mearns’ comparisons of frondator with canadensis were evidently not made with the largest specimens of the latter, as I have examined some whose cra- nial and body measurements are about equal to the maximum recorded by him for frondator. Nevertheless, there is little doubt that the larger size of average frondator is well established. Its long hind foot, broad tail and light coloration distinguish it immediately from canadensis. Its approach to pacificus is solely along the line of great size as indicated by the length of body and hind foot, but in cranial characters, as also in color, it is farthest removed from that race. The close anatomical relation of frondator to carolinensis has been mentioned. Specimens Examined.—Montana, 1 skin with skull; Wyoming, 1 skull. Pactric BEAVER. Castor canadensis pacificus, subsp. nov. Plate XXI; Fig. 1. Plate XXII; Fig. 1. Type Locality.—Lake Kichelos, Kittitass county, Washington ; altitude about 8000 feet. Type, No. 1077, ad. 2, in the collection of S. N. Rhoads; collected in April, 1895, by Allan Rupert. Geographic Distribution.—Pacific slope, of America, from Alaska to California. Color.—Above with very uniform, dark and glossy reddish chestnut overhair, almost concealing along dorsum the seal-brown underfur. Top of head like back ; sides of head, throat, rump, thighs and vent not decidedly lighter than back and belly as in the other forms, these parts paling to walnut brown. Ovyerhair of sides and under parts, between seal brown and broccoli brown ; under fur of belly drab gray at the roots ; hind feet dark seal brown ; fore feet and limbs, dark wood brown. Ears black. Anatomical Characters.—Size, largest of the canadensis group, but of more slender build, the skeleton throughout being of much greater longitudinal and lesser lateral dimensions than in the other forms. Tail and hind foot relatively long. Skull large, relatively narrow, with long, narrow rostrum and nasals, the latter with outer margins nearly parallel and reaching basally decidedly beyond the premaxillaries. Upper molar dentition weak, the crowns of molar teeth rectangular. Measurements.—Of the type from carcass: Total length, 1143 mm.,; tail vertebra, 330 mm.; (from relaxed skin) scaly portion of tail, 295 mm. by 122 mm.; hind foot, 185 mm.; length of skull, 142 mm.; breadth of skull, 101 mm.; length of nasals, 53.6 mm.; breadth of nasals, 24 mm.; average length and breadth of five skulls from Tacoma and Lake Kichelos, Washington, 144 mm. by 99 mm.; average nasal length and breadth of same, 54 mm. by 23 mm. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 423 Remarks.—Reliable measurements of only one adult skin specimen (the type) of pacificus were accessible. An adult mounted specimen from Josephine county, Oregon, in the Wagner Institute, Philadelphia, confirms the color and measurements of the type so far as the latter can be ascertained from the stuffed animal. Pacificus, like its associates, Mustela americana caurina and M. canadensis pacifica of the Pacific slope regions, is distinguishable by its rich and deep coloration from its darkest trans-Cascadian representatives. No specimens have come to hand from Alaska, but undoubtedly, from what we know of other species found there as well as from the accounts of trappers and furriers, the Alaskan coast beaver represents the maximum of size* and the greatest richness and depth of fur coloration seen in American beavers. Specimens Examined.—Washington, Tacoma, 1 skeleton, 1 skull; Lake Kichelos, 1 adult skin with skull, 3 young skins with skulls, 1 skeleton, 12 separate skulls ; Ore- gon, Josephine county, 2 mounted specimens ; British Columbia, (?) Sumas, 1 skull; + Victoria, 1 skull. THE OTTERS OF NORTH AMERICA. As Mr. Oldfield Thomas has shown in his “ Preliminary Notes on the Species of Otter,” published in 1889 in the Proceedings of the London Zoilogical Society, the charac- ters and nomenclature of the North American species are in great need of study. Dr. Elliot Coues has elucidated with sufficient clearness, in his Monograph of the Mustelide, the habits and characters, and, to some extent, the synonymy of the typical Canadian otter, Lutra hudsonica Lacépede. Its relations, however, to other nominal species, especially to the otters of the Pacific slope of America from California northward, demand investigation. As in the case of the American beaver, just treated, this paper has to do solely with one central Canadian type and its subspecies found in America north of Mexican terri- tory. Avoiding a general preliminary discussion of the rather perplexing questions of nomenclature and geographic variations and distribution, I will present these in order in the more formal and detailed synopses which follow. * Dr. Allen’s measurements of Alaskan skulls, page 447 of the Monograph of N. A. Rodentia, do not indicate unusual size, but as we have no precise locality given they may not have come from the coast region, and, therefore, do not represent pacificus. { Thisskull (No. 5545, 3, coll. of E. A. and O. Bangs) is the largest of which I find any record, measuring 154 by 108mm. The next in size is No. 2146, U. S. Nat. Mus., from Nebraska, recorded by Baird. Its size was 147 by 105.5 mm. Unlike all my pacificus specimens, No. 5545 has very wide convex nasals. A. P. S.—VOL. XIX. JB. 424 CONTRIBUTIONS TO A REVISION OF THE Hupsonran Orrer. Lutra hudsonica (“ Lacéptde,” Desmarest). Plate XXIV ; Figs. 1 and 2. Mustela lutra Linn., canadensis Schreber, Sdugt., III, Pl. CX XVI, B. (dated 1778 on title-page, but, according to Sherborn, the text of Vol. III was published in 1777 and this plate in 1776). Mustela (lutra) canadensis Kerr, Linn. An. Kingd., I, 1792, p. 173 (see Thomas, Proce. Zool. Soc. Lond., 1889, p. 197, and Allen, Bull. Amer. Mus. N. Hist., VIL, 1895, p. 188). “ Mustela hudsonica Lacép.[ede],” Desmarest, Nouv. Dict. d’ Hist. Nat., XIII, 1803, p. 384 ; (Nouv. Hd.) 1817, p. 219. Lutra canadensis J. Sabine, App. Frankl. Jour., 1823, p. 653, and of nearly all subse- quent authors (not L. canadensis F. Cuvier, Dict. Sci. Nat., 1828, p. 242; see O. Thomas, J. ¢., p. 197). Lutra hudsonica F, Cuvier, Suppl. Buff., 1, 1831, p. 194; Merriam, NV. Amer. Fauna, No. 5, 1891, p. 82. Lataxina mollis Gray, List Mamm. Brit. Mus., 1843, p. 70. Lutra destructor Barnston, Canad. Nat. and Geolog., VIII, 1863, p. 147, Figs. 1 to 6. Type Locality.—* Ou la trouve au Canada sur les bords de la mer.” Geographic Distribution—Northern North America from the Arctic ocean south- ward into the United States and from the Atlantic ocean to the Cascade mountains ; intergrading southeastwardly into subspecies dataxina F. Cuyier and vaga Bangs, south- centrally into subspecies sorone Rhoads, and westwardly into subspecies pacifica Rhoads.* Color (taken from two specimens in the Bangs collection, No. 5638, yg. ad. 3, Annapolis, Nova Scotia, November 23, 1896, and No. 4190, ad. 2, Upton, Me., Octo- ber 25, 1895).—Aboye, dark seal brown from nose to tip of tail, darkest posteriorly, below from breast to tail between broccoli and vandyke brown in the Nova Scotia speci- men and between seal and yandyke brown in the Maine specimen. Head and neck below a line running from nose to lower base of ear and base of foreleg light Isabella color anteriorly darkening on lower neck to wood brown in the Nova Scotia animal. In the Maine specimen the neck is Prout’s brown. Feet, legs and tail corresponding to darker shades of upper and lower body. A summer specimen from New Brunswick is dark, vandyke brown, but little paler below than on back, and darker than winter specimens of dataxina from Maryland. * The o{ters of Louisiana and Mississippi are stated by furriers to be very dark and light-pelted, resembling South Florida and Gulf-coast skins. No specimens having been examined, they are referred to oaga. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 425 Anatomical Characters.*—Size, medium (exceeded by vaga, sonora and pacifica). Tail relatively short. Inferior webs of feet and interspace between posterior and ante- rior callosities of manus, densely haired. Hind foot with claw about 125 mm. in old adults ; but so variable as to have little diagnostic value. Total length rarely exceeding 1100 mm. Skull—size, medium (greatly exceeded by vaga and pacifica). Teeth large, crowded longitudinally upon each other and obliquely overlapping. Postorbital neck of frontals relatively short and wide, its superior ridge on a plane with nasals and occi- pital crest. Mastoid width much less than zygomatic width. Postorbital processes short and stout. Audital bulle large, tumid, rising abruptly from the sides of basioccipital. Measurements.—See tables. Remarks.—V ariations in the size of adult otters from apparently the same region seem remarkable at first sight, but I find that these are not always to be attributed to sex (for the female otter sometimes reaches near to the average size of the males), but to environment. The otters of the Alleghany mountain streams are uniformly smaller than those of the tide-water creeks and rivers of the Atlantic seaboard. This rule applies from Labrador to Florida and is undoubtedly the result of the relative difficulty of obtaining food and securing shelter from enemies in the two kinds of habitat. On the other hand, this difference lies wholly within the limitations of individual variation and in no sense affects the well-defined cranial and other characters which distinguish the races and species hereafter defined. It has to do solely with size, not with propor- tions. In a letter from Mr. C. 8. Brimley, of Raleigh, North Carolina, the same feature is alluded to where he states: ‘A trapper of our acquaintance says that otters from the saltmarshes of eastern North Carolina average considerably larger than the otters of the small streams of the central part of the State.” There is rarely to be found a case in mammalian nomenclature more puzzling than that of the first tenable name of the Hudsonian otter. Its synonymy involves that of the mink and the fisher as well as the questions of priority of publication of Erxleben’s and Schreber’s great works on the Mammalia, and the tenability of plate names. I have consulted Drs. C. H. Merriam and T. 8. Palmer at length on these questions and have accepted their ruling as to the first tenable name of the Hudsonian otter being Luéra hudsonica Lacépede and that of the northeastern mink to be Putorius vison Schreber. In regard to the name of the fisher, however, I prefer to abide by Canon XLIII of the Code of the American Ornithologists’ Union, which accepts, under certain conditions, the names of species originally published on plates, which Drs. Merriam and Palmer and Mr. Sherborn do not accept. Returning now to the abstract of synonymy as given above for the Hudsonian otter, the case may be concisely stated thus: JJustela lutra * The diagnostic value of the nose pad has no significance in this study of the relationships of a monotypic group. 426 CONTRIBUTIONS TO A REVISION OF THE canadensis Schreber is a plate name published (jide Sherborn) in 1776, and is the ear- est applied to this otter. It would stand (A. O. U., Canon XLII) were it not unques- tionably applied and intended by Schreber merely as a geographic name without refer- ence to its specific relations to “ J/ustela lutra Linn.” For this reason alone it should be discarded. Furthermore, the name J/ustela canadensis was used by Schreber on a pre- vious plate in the same volume (Pl. No. 126) in the specific sense for the fisher. This plate was also (fide Sherborn) published in 1776, one year before the text, which was published in 1777, and the bound yolume of text and plates were dated 1778. In 1777, Erxleben published a description of the fisher and named it Justela pennantii, by which name it has been since designated by authors generally. As this name is antedated by — the tenable plate-name J/ustela canadensis of Schreber by one year, I adopt it as the name of the fisher of Pennant from the northeastern United States. Erxleben pub- lished in the same work a description of an animal which he named Iustela canadensis, and which Baird and Coues have considered applicable to the mink, and the accept- ance of the dates on the title-pages of Schreber’s (1778) and Erxleben’s (1777) works would give priority to Erxleben’s name and displace Justela vison of Schreber. But Sherborn’s emendation of these dates makes J/. canadensis of Erxleben for the mink untenable, it being preoccupied by Schreber’s plate-name J/. canadensis for the fisher, as stated above. Besides this fact, Dr. Merriam considers that Erxleben’s description of M. canadensis also applies to the fisher and the marten in such a way as to make it untenable for any species. Returning to the search for a first name for the otter, we find Kerr’s name, Jf. cana- densis of 1792, to be unayailable because he placed it under the old genus Mustela. Next in order appears to be the name hudsonica, which is accredited to Lacépéde, in an article on the Canadian otter in the first edition of the Nowvelle Dictionaire d Histoire Natur- elles, which is signed ‘‘ Desm.” I haye not examined this reference personally, but am indebted to Dr. J. A. Allen for a transcript of these facts from the only known copy of the work in America which appears to be available, belonging to the library of the American Museum of Natural History. In agreement with my previous rendering of manuscript names, and on the supposition that Desmarest was the real author and pub- lisher of this name and description of hudsonica, I cite it as Lutra hudsonica (“ Lacé- pede,” Desmarest). I agree with Dr. Merriam that this name should stand for the otter of eastern Canada. Frederick Cuyier seems to have been the first to place this animal in the genus Lutra under the Lacépéde-Desmarest name hudsonica in 1831. The Lataxina mollis of Gray and the Lutra destructor of Barnston are no doubt synonyms of hudsonica. Specimens Examined.—Labrador, Okak, 1 skull; Grand river, 1 skull ; New NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 427 Brunswick, Restigouche river, 1 skin; Nova Scotia, Annapolis, 1 skin with skull; Maine, Upton, 1 skin with skull; Bucksport, 1 skull; Massachusetts, Kingston, 1 skin with skull; Westford, 1 skull; Canton, 1 skull; Missouri, 1 skull; British Columbia, Vernon, 1 skull; Alaska, Tanana river, 1 skull. Caronintan Otter. Lutra hudsonica lataxina (F. Cuvier). Plate XXIV; Fig. 4. LIutra lataxina F. Cuvier, Dict. des Sci. Nat., 1823, p. 242. Type Locality.—South Carolina. Geographic Distribution —Carolinian faunal region, intergrading through the Tran- sition region northward with hudsonica and southward through the Austrariparian into vaga of southern Florida. Color.—Much lighter than hudsonica. Above (from a specimen taken at Liberty Hill, Conn., No. 4252, ad. #, Noy. 19, 1895, collection of E. A. and O. Bangs*), dark vandyke brown, tipped on upper head, neck and shoulders with wood brown, darkening posteriorly. Upper feet and limbs dark bistre. Below, from lower breast to end of tail, between Prout’s brown and broccoli brown. Head, neck and breast, including ears, below a line connecting nose, upper eyelid, upper ear and upper base of fore leg, grayish wood brown, lightest on head, darkening posteriorly to color (/. c.) of breast. The ayver- age Carolinian winter specimens from Maryland southward are somewhat lighter and some are Prout’s brown above, the wood brown of lower head and neck becoming a pale grayish buff. : Anatomical Characters.—Size, smallest of the hudsonica subspecies. Inferior webs of feet and interspace between callosities of manus, sparsely haired. Hind foot with claw about 120 mm. Total length rarely exceeding 1100 mm. Skull relatively small, with very large teeth, and weak postorbital processes. In other respects like the hud- sonica type. Measurements.—See tables. Remarks.—The relations of this subspecies to northern Audsonica on the one hand and to the southern vaga on the other are rather peculiar. It is without question a nearer ally to hudsonica than vaga in the territory between Connecticut and South Caro- lina, but, as Mr. Bangs has implied in his remarks on vaga, there is a tendency in the Georgia (and we may infer in the South Carolina) otter to the large size and peculiar * This specimen comes from the northern edge of the Carolinian region. No equally good skins from more southern localities being available, it is used as typical of the Curolinian race. It corresponds closely to two fine 1897-8 winter pelts of Maryland otters, examined through the courtesy of Mr. 8. E. Shoyer, of Philadelphia. 428 CONTRIBUTIONS TO A REVISION OF THE skull and color characters of the south Florida animal. There is so much evidence of the intergradation of /ataxina both north and south that the specific separation of vaga from it is not permissible. On the other hand it is impossible to ignore the decided racial differences of the Carolinian otter from the Hudsonian type. Cuvier’s original description of /ataxina gives “Caroline du Sud” as the locality where the type was taken ; it is, therefore, permissible to restrict this name to the Caro- linian form as typified in the otters found in the Carolinian lowlands of the eastern United States from south of the “Transition Zone” of Dr. C. Hart Merriam, as far as middle South Carolina, Alabama and Mississippi, where it merges into vaga of the Gulf or southern “ Austroriparian Realm ” of Dr. J. A. Allen. I know of no restricted synonyms of /ataxina. Dr. Coues quotes in his Pur-bear- ing Animals a“ Latax lataxina Gray, Ann. Mag. N. H., I, 1837, p. 119.” The work referred to contains no such name. Cuvier’s description of dataxina gives its color as “dark blackish brown, a little paler beneath. Cheeks, temples, lips, chin and throat pale brownish gray, and under side of tail grayish brown, the hair tips reddish.” He compares the skull of dataxina with his Lutra enudris, “ Loutre de Guiane ” of the pre- ceding page and remarks on the “straight line, even concave or depressed,” joining the nasals and occiput. This is significant, as one of the peculiarities separating vaga from lataxina and hudsonica is the convexity of the frontal plane in the former. Specimens Examined.—Connecticut, Liberty Hill, 1 skin with skull; Pennsylva- nia, Clinton county, 2 mounted specimens; Monroe county, 3 skulls; New Jersey, Tuckerton, 1 skull; Mickleton, 2 disarticulated skeletons ; Maryland, 2 fresh cased — winter furs; North Carolina, Raleigh, 2 skulls. Frorma Orrer. Lutra hudsonica vaga Bangs. Plate XXV; Fig. 2. Lutra hudsonica vaga Bangs, Proc. Bos. Soc. Nat. Hist., XXVIII, 1898, p. 224. Type Locality—Micco, Brevard county, Florida. Geographic Distribution Florida, southeastern Georgia and the Gulf regions of Alabama, Mississippi and Louisiana, intergrading (?) northwardly into lataxina. Color.—Dark ; less black than hudsonica, darker and redder than /atazina. Breast and belly nearly unicolor with back. Paler area of head and neck, scarcely reaching breast. Above and below, dark, rich chestnut, scarcely paler on belly. Lower head and anterior throat below line from nose to and behind ears, strongly tipped anteriorly with tawny Isabella color darkening to raw umber on throat, the underfur darker than over- fur, instead of lighter as in Jataxina. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 429 Anatomical Characters.—Size, large. Tail relatively long (fide Bangs). Inferior webs of feet and interspace of palms nearly naked. Hind foot with claw reaching maximum (No. 4998 Bangs Coll., yg. ad. 3, Citronelle, Florida) of 130 mm. Total length (maximum of No. 4998, é@ ¢., 1285 mm.) exceeding 1200 mm. Skull large, teeth relatively small, not crowded longitudinally. Postorbital neck of frontals long and narrow, suddenly constricted at base. Frontal plane strongly upraised above a line con- necting occipital crest with base of nasals and above the level of postorbital processes. Mastoid width nearly equaling the zygomatic width in very old specimens, in young adult skulls the mastoid width is the greater. Wings of mastoid processes strongly developed and flattened laterally. Audital bullee as in hudsonica and lataxina ; well developed, tumid at basioccipital margins. Postorbital processes relatively weak and_ slender. Underfur short, sparse. Measurements.—See tables. Remarks.—This subspecies just described by Mr. Bangs in his most valuable paper on Florida and Georgia mammals is, as already noticed, quite different from lataxina, its nearest geographic ally. In color it comes nearer hudsonica intermediates from New England. In size and color and lack of hair on the webs and palms it shows approach to the remote pacifica, but its peculiar long-waisted and broad-based skull dis- tinguishes it from all other American forms except, perhaps, those of the northern Cen- tral American and South American otters which I have examined. The yellowish and reddish shades of south Florida vaga suggest affinity with what we find published of the characters of the otters of the Caribbean coasts. In essential respects Mr. Bangs’ diagnosis of this animal is very good. He, however, used the skull of a young adult male for cranial comparisons, and while it is true that the ratio of the mastoid to the zygomatic width is much greater in vaga than hudsonica it is not as great as would appear by Mr. Bangs’ figure. In crania of old adult vaga in my collection the mastoid and zygomatic widths are about equal, the latter slightly wider. In hudsonica, however, the excess of zygomatic width and slight development of the mastoid wings is marked. Specimens Examined.—F lorida, Tarpon Springs, 1 adult pelt, 8 young skins with skulls and 2 extra skulls; Salt Run, St. John’s river, 1 skull. Paciric Orrer. Lutra hudsonica pacifica, subsp. nov. Plate XXIV; Fig. 3. Plate XXV; Figs. 1 and 3. Intra paranensis and aterrima Thomas, P. ZS. l.¢.,p.199; Trouessart, Catal. Mamm., 1897, pp. 286, 287 (not of Pallas, Zoogr. Ross. Asiat., 1811, p. 81). Lutra californica Baird, Mamm. N. Amer., 1857, p. 187 (not of Gray, Jag. Nat. Hist., I, 1835, p. 580, which is LZ. felina ; see Thomas, /. ¢., p. 198). 430. CONTRIBUTIONS TO A REVISION OF THE Type Locality.— Lake Kichelos, Kittitass county, Washington ; altitude about 8000 fect. Type No. 616, yg. ad. 3, in the collection of 8. N. Rhoads ; collected in fall or winter* of 1892—93, by Allan Rupert. Geographic Distribution —Pacific slope of North America, from Alaska to Cali- fornia. Color.—Of type: Lighter than hudsonica, with a browner cast, approaching nearly to lataxina. Average of coast specimens from Puget Sound northward, ruddy seal brown, sometimes very dark in Alaskan coast specimens. Lower parts from breast to end of tail much lighter (Mars-brown) than back. Ventral region conspicuously lighter. Lower head, neck and breast very pale wood brown, almost dirty gray. Anatomical Characters.—Size, very lorge.t Tail normal. Inferior webs of feet and palmar interspaces nearly naked. Hind foot not recorded in type, the caleaneum missing ; no measurements of other specimens available. Skull largest of the North American otters (reaching a maximum of 119 mm. in occipito-nasal length and 83 mm. in zygomatic expanse in an Alaskan coast example) ; teeth relatively weak, less crowded longitudinally than in hudsonica. Interorbital width relatively very great, nearly 12 times postorbital constriction ; postorbital processes long and stout. Mastoid and zygo- matic proportions as in hudsonica. Audital bulle remarkably flattened. Measurements—See tables. Remarks.—The type specimen, though taken in the mountains and not fully mature, is large and has a skull which would have, perhaps, eventually equaled the maximum size recorded aboye for an Alaskan specimen of much greater age. A very old female skull from the vicinity of Puget Sound confirms fully the diagnostic characters of pacifica as given. In treating of the otters of the Pacific slope of America we are confronted with two nominal species to which they have been doubtfully referred by authors. In point of time the first to be considered is the Viverra aterrima of Pallas,{ described from a hunter’s skin, lacking skull and feet, taken in northeast Siberia, “between the Uth and Amur rivers.” Schrenck and Middendorff listed this animal in their works on Siberian Zodlogy with the remark that they were unable to verify its existence or clear up the mystery of its strange characters as given by Pallas. Mr. Thomas (P. Z. 8, /. c., p. 199) queries, on the basis of a mistaken suggestion of Dr. Coues, whether it may * The season of capture was not recorded, but the pelt indicates that it was taken in full winter fur. {Ihave no measurements of Alaskan otters, but judging by the great size of the skulls from there they must greatly exceed any known species of Lutra. On the basis of the skull they must attain a maximum length of over 1400 millimeters. t Zoog. Rosso. Asiat., l. c. ——— NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 431 not prove to be the same as the so-called Lutra paranensis Rengg. which he assumed might occur throughout the whole Pacific coast regions of America. The close relation- ship of our Pacific coast otters to hudsonica will effectually remoye them from any com- plication with paranensis, but as regards aterrima we must devote sufficient space to show the impossibility of referring the Alaskan land otter to that animal, as Trouessart has lately done.* A careful study of Pallas’ original description, together with the fact that no later author or explorer has been able to explain or rediscover the animal, convinces me that it is either unidentifiable or will prove not to belong to the Lutrine but to the Musteline. Pallas states it to be intermediate in size between the European otter and the European mink. He states the length of the skin to be 19 inches, 3 lines, and of the tail 5 inches with a brush of 12 inches! The color of the animal is said to be very black and shin- ing, except the sides of the head between the eyes and ears, which change from black to “subrufescent.” The absurdity of applying such a description to the animal which I have named pacijica, or, indeed, to any member of the genus Lutra, is certainly evident. So far as any animal now known to zodlogists is concerned, the Viverra aterrima of Pallas should be consigned to oblivion. Another name which has given trouble to those who had to deal with the Pacific coast otter is the Lutra californica of Gray. Fortunately, Mr. Thomas has effectually exposed the history and at the same time the inapplicability of that name to a North American animal of the hudsonica type. He has shown in his paper in the Proceedings of the Zoblogical Society (l. ¢., p. 198) that Gray’s type of californica did not come from California, but most likely from Patagonia, in which case he makes it a synonym of Lutra felina Molina. Specimens Examined.—Washington, near Tacoma, 3 skulls ; Lake Kichelos, 1 skin with skull, 1 skull; Oregon, 1 skull; British Columbia, Sumas, 1 skull; Alaska (coast?), 3 skulls; Kodiak Island, 2 skulls; Mission, 1 skull; Queraquinay Island, 1 skull. Sonoran Otter. Lutra hudsonica sonora, subsp. noy. LIutra canadensis Mearns, Bull. Am. Mus. Nat. Hist., 111, 1891, pp. 258-256. Type Locality.—Montezuma Well, Beayer creek, Yayapai county, Arizona. Type, ad. 2, No. 3712 in the collection of the American Museum of Natural History. Col- lected December 26, 1886, by Dr. Edgar A. Mearns. * Catalogus Mammalium, 1. c. } It is conjectured that this skull came from the North Pacific. It has Capt. T. J. Turner’s name on it. I cannot find an island of this name on the maps. EPS —=V. Ol xox, 11: 432 CONTRIBUTIONS TO A REVISION OF THE Geographic Distribution —Arid southern interior of North America, from Mexico, probably to Wyoming. Color.—Of type, fide Mearns, /. c.: “ Above dark brown, without reddish tinge ; this color changing gradually to a light grayish brown below, being palest (almost whitish) upon the sides of the head below the level of the eyes and upon the under side of the head and neck as far back as the fore limbs. . . . . The long hairs of the lighter por- tions of the body are pointed with yellowish gray and upon the upper surface of the head and neck the tips of the hairs are yellowish brown, giving a paler cast to that part of the dorsum.” Anatomical Characters.—Size, large, with a very long hind foot, the body length measurements exceeding those of any other specimen of North American otter exam- ined or recorded.* Webs of feet not densely haired beneath. Hind foot,145 mm. Total length reaching 1300 mm. Skull—size, large, nearly as great as in largest Alaskan pacifica, but small for the great relative length of: body, “less massive, broader, with more eyenly rounded zygomatic arches and with the brain case more conyex or bulging in its outlines.” “ Arizona skulls differ from all others in the slender, attenuated postor- bital processes and in the greater height of the lower jaw from angle to condyle, or to summit of coronoid process. From its geographically near neighbor, L. felina of Cen- tral America, it presents many cranial and dental differences; in fact, skulls of the lat- ter are so very distinct [in their inferior concavity, frontal depression, short muzzle, ‘ narrow postorbital constriction and absence of the heel in front of the antero-internal cusp of the last upper molar] from any known specimens from North America, north of Mexico, as to be distinguishable from them at a glance.” Measurements.—Oft type: “ Total length, 1300 mm.; head and body (measured from tip of nose to anus), 815 mm.; tail measured from anus to end of vertebrae, 472 mm. ear, height above crown, 15 mm.” No skull measurements given. Remarks.—I have accepted Dr. Mearns’ very full and satisfactory diagnosis of the Arizona otter, given in the Bulletin of the American Museum of Natural History, as conclusive evidence of the existence of a recognizable race in arid interior America, south of Montana. Its great size and light color together form a combination not found in any other known or named otter. It has been thought unnecessary to examine the type, as, owing to the author’s removal from Philadelphia during the completion of this paper, such an examination would have caused a greater risk to the type specimens than the facts warranted. * The great size of the type, as compared with an adult male also recorded by Dr. Mearns from Arizona, indicates that the sex of the type may have been wrongly determined. If correct, the size to be expected of a full-grown male sonora would be extraordinary. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 433 Newrounpianp Orter. Lutra degener Bangs. . Plate XXIV; Fig. 5. Lutra degener Bangs, Proc. Biol. Soc. Wash., XII, 1898, p. 36. Type Locality —Bay St. George, Newfoundland. Geographic Distribution —Confined to Newfoundland (?). Color.—Of type, ad. 3, taken April 22, 1897: Above, black with seal brown reflec- tions. Ears, seal brown. Lower head and neck areas grayish wood brown, becoming seal brown on breast; the remainder of lower parts nearly as dark as back. Tail uni- color. Feet seal brown and densely haired on under side of webs and palmar interspaces. Anatomical C haracters—Size, much smaller than any of the hudsonica group. Hind foot small, with claw averaging about 112 mm.* long in the two specimens exam- ined. Total length about 1000mm. Tail relatively short. Skull very small, narrowed, weak and fragile; the brain case wide anteriorly ; the frontal and interorbital widths narrow and the postorbital processes weak and slender, strongly grooved on their supe- rior face. Sagittal crest not developed even in old specimens. Interorbital constric- tion about equal to postorbital constriction. Teeth weak, with normal cuspidation. Audital bulle normal. Measurements.—See tables. Remarks.—Vhe type specimens of degener, so generously loaned to me by Mr. Bangs, when compared with the large series used in the preparation of this paper, con- vince me that this depauperate insular form has no intercourse with the larger typical hudsonica of Labrador and New Brunswick. A skull from Grand river, Labrador, shows no approach to the degener type, and another from Okak, Labrador, agrees in the same differences. A young adult skull and skin of hadsonica from Nova Scotia, and an adult summer skin from New Brunswick, show that the maritime otter of the mainland some- times attains a size nearly one-third larger than the largest known specimens of old, adult degener. Specimens Examined.—Newfoundland, Bay St. George, 2 skins with skulls, 1 extra skull. THE FISHERS OF NORTH AMERICA. Apology must be made for the inferior series of skins and skulls which form the basis of the subjoined remarks on the Pekan. They serve, however, to elucidate some * The collector’s measurement of the hind foot of type is given on label as ‘126 mm.”’ This is certainly incorrect, as the length determinable by feeling the caleaneum in the dry skin could not haye exceeded 115mm. This accords with the small size of the hind foot and the length of other specimens of degener. 454 | CONTRIBUTIONS TO A REVISION OF THE questions sure to be soon brought up in the active advance of monographic work in American mammalogy. The synonymy of Pennant’s Fisher has already been discussed under Lutra hud- sonica, and I have there given reasons for my adoption of the plate-name canadensis of Schreber as having priority over the long-accepted name pennanti of Erxleben for this animal. Pennant’s Fisuer. Jlustela canadensis Schreber. Mustela canadensis Schreber, Saugt., I, p. 492, Pl. CX XIV. Text published in 1777, plate in 1776 (fide Sherborn). Mustela pennantii Erxleben, Syst. An., 1777, p. 470. Mustela melanorhyncha Boddaert, Elench. An., 1784, p. 88. Viverra piscator Shaw, Gen. Zodl., 1, 1800, p. 414. Mustela nigra Turton, ed. Linn. Syst. Nat., 1, 1802, p. 60. Mustela godmani Fischer, Syn. Mamm., 1829, p. 217. Type Locality —* New York and Pennsylvania,” Pennant. Geographic Distribution —Northern North America, east of the Cascade moun- tains, from the northern limit of trees to Colorado and North Carolina in the mountains. Intergrading on the Pacific slope into subspecies pacifica, and probably in the southern Rocky mountain region into a paler race. Probably represented in the Hudsonian faunal region by a subspecies.* Color.—F rom an adult, male, winter specimen taken near Lancaster, Pa., March 11, 1896, and in the possession of Dr. M. W. Raub, of that city, who furnished the description : “ Head and one-half of the length of body, gray and black mixed, gray predominating ; throat darkest, with snout from tip to line of eyes dark brown. The hinder half of body gradually darkens into a deep chocolate color until it reaches the tail, which is almost black with a tip entirely black. Hind legs and tail, viewed at a distance of six feet, look very dark, almost pure black. The fore legs are black but not so deep. Tips of ears, darkest.” Two specimens from the Bangs collection, one from Moosehead lake, Maine, the other from Idaho county, Idaho, seem to answer closely the above description. The light upper and forward portions of body are a grizzled grayish brown, the long hairs black tipped. The basal half of hairs of anterior back are hair brown. I can discover no color characters to separate the Idaho specimen from the one from Maine, nor do the skulls indicate any reliable differences. The Maine skin (of an animal two-thirds grown) * Typical canadensis must be restricted to the Alleghenian form. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 435 has white patches on lower fore leg, breast and vent, and an immature specimen of paci- fica has white spots on throat, arm-pits and vent. The four adult specimens examined are not thus pied. Dr. Coues, in his Hur-bearing Animals, says that the fisher is an exception to the marten, mink and weasel in not having these patches. They may dis- appear with age in the fisher, but they do not in the other species. Anatomical Characters.—Size, smaller than subspecies pacifica. Skull small; nasals relatively short, less elongate at basal apex. Posterior upper molar relatively small, its inner lobe not greatly developed longitudinally so as to only slightly exceed the breadth of outer lobe ; neck of crown of same tooth but slightly constricted. Measurements.—Of Dr. Raub’s Pennsylvania specimen, old ad. 3, /. ¢.: Total length, from end of nose to end of tail hairs, 965 mm.; tail vertebrae, 318 mm.; hind fod foot, 115 mm.; ear from crown, 27 mm. A mounted specimen, No. 507, Academy Natural Sciences, adult 3, from “ Pennsylvania,” has a total length of 1000 mm., with tail (minus brush), 390 mm., and hind foot, 112 mm., taken from the dry mount. The Idaho specimen, No. 6964, young adult 3, coll. of E. A. and O. Bangs, is 978 mm. long, with tail, 369 mm., and hind foot, 117 mm. Skull of No. 7437, yg. ad. #, Greenville, Me., total length, 117 mm.; zygomatic width, 63 mm.; mastoid width, 54 mm.; mesial nasal length, 22 mm. ~ Remarks.—The characters of the Pennsylvania fishers above enumerated, so far as they are based on reliable measurements and color diagnoses, may be considered as repre- senting typical canadensis, based on Pennant’s original notice of the animal. Whether a series of Alleghenian fishers will show the Hudsonian animal to be separable is an interesting question probably to be decided in the affirmative. The Idaho and Maine specimens examined, though not contrasted by me with Dr. Raub’s specimen, must be very close to it. No skulls of Pennsylvania fishers have been examined, but the close resemblance of the Idaho skull to those from Maine, as indeed to pacifica also, strongly indicates that no cranial differences exist between the east American fishers of the north and south. The “saturated” color characters of pacifica are alone sufficient to distin- guish it from all fishers found east of the Cascades. Specimens Examined.—Pennsylvania, 1 mounted specimen (fide Dr. Raub, 1 mounted specimen) ; Maine, Mooseland lake, 1 skin with skull; Greenville, 2 skulls; Lincoln, 1 skull; Idaho, Idaho county, 1 skin with skull. Other specimens from east- ern North America, 1 mounted, 2 old ad. skulls. Paciric Fisner. Justela canadensis pacifica, subsp. nov. Type Locality.—Lake Kichelos, Kittitass county, Washington ; altitude about SO00 436 CONTRIBUTIONS TO A REVISION OF THE feet. Type, No. 1074, old ad. 9, in the collection of 8. N. Rhoads; collected in the fall or winter of 189295, by Allan Rupert.* Geographic Distribution.—Pacific slope of America, from Alaska to California. Color—Ahbove, from between eyes to middle back, grizzled, grayish ochraceous heavily lined with black, becoming hazel black on hind back and dark black on rump, thighs and tail. Whole head, behind eyes clove brown basally, strongly grizzled with dirty white. Snout to eyes blackish seal brown. Chin, throat, breast and belly between dark chestnut and hazel, obscured with black. Legs and feet black, the fore legs show- ing the vandyke brown bases of hairs. Basal half of hairs of anterior back are Prout’s brown as contrasted with the hair brown of canadensis. Anatomical Characters.—Size, large, skull very large, with relatively long nasals. Posterior upper molar large, with spreading inner lobe much wider longitudinally than outer section of same tooth; the crown suddenly constricted at the middle. Measurements—Of type from relaxed skin: Total length, 1090 mm.; tail, 350 mm. without brush ; hind foot not determinable, as the bones are missing. Measure- ments of a specimen two-thirds grown, No. 295, coll. S. N. Rhoads, from near Tacoma, Wash.: Total length (relaxed skin), 970 mm.; tail, 400 mm.; hind foot, 112 mm.; ear from crown, 21 mm. Skull of type: Total length from hinder end of sagittal crest to front end of premaxille, 125 mm.; zygomatic expansion, 73 mm.; mastoid expansion, 54 mm.; interorbital constriction, 28.5 mm.; postorbital constriction, 20 mm.; mesial length of nasals, 27 mm. - Remarks.—The dimensions of the type skull, when we consider it was from a female, show that the fishers of the Cascade mountains attain a much greater size than those of the Appalachian chain. Young adult skulls of the same age from western Washington and Maine show the same distinctions. The younger specimen from Tacoma, while approaching nearer to Idaho and Maine specimens in grayer color, is very much darker than they, the difference in shade between the anterior and posterior dorsal areas of the former being slight, while in the latter it is striking. The tawny suffusion so deeply marked in the type of pacifica and which separates it at a glance from canadensis is also noticeable in the Tacoma specimen. Specimens Examined.—Washington, Lake Kichelos, 1 skin with skull, 2 skulls ; near Tacoma, 1 skin, 1 skull ; British Columbia, Sumas, 1 skull. * Mr. Rupert, whose business is hunting and trapping, first sent me the fresh skull of a very old Q fisher, which was entered in my catalogue as No. 621. I wrote him immediately that I would like to have the pelt belonging thereto, and in a later shipment the skin, which forms the type of pacifica, was sent on without label. As it is also from a female and a very old animal, I consider the skin and skull as belonging to the same individual. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. Skull Measurements of North American Otte (in millimeters) | | | | | s35e)2 1, 12 ls leal es BS | | ee S é - - iu ss Collection. EE Sex. | Locality. Species. as ok 38 28 e8 ae 2 & 3 Remarks. oA wes] 8 | 2° | ge| Se] 28] By = Ress iS |e |e |e | as] $3 = : hae ol SS E a | ; a A ee ae E. A. and O. Bangs} 5638 | yg. ad. ¢\| Nova Scotia, Annapolis L. hudsonica (‘‘La-/ 113.5] 72 | 68 | 27.7; 23 | 35 | 15 | Large, coast form. cép.,’’? Desm.) do. 7431 old ad. | Labrador, Okak do. 74.5) 67 | 23 | 19 | 35 | 18.5] Coast form. Acad. N, Sci. Phila.} 3150 old ad. | Labrador, Grand River do. 105 2.5) 65 | 20.8) 20 | 29 | 10.5) Inland form. Smithsonian Inst. | 21483 old ad. | Alaska, Tanana River | do. 102 72 | 63.5) 24 | 18 | 32 | 12.5) Inland form. E. A. and O. Bangs, 4238 | old ad. | Maine, Bucksport do. 109 | 73.5] 66 | 25.5] 21.5137 | 14 | Coast form. do. 4188 | old ad. 3) Massachusetts, Canton do. 112 | 76 | 69 | 26 | 22 | 38 | 15. | Intermediate. Acad. N. Sei.Phila.) 3569 | oldad. | Pennsylvania, Monroe Co. | Lh. lataxina(F.Cuyv.)| 100 | 69.5 65 | 22.8) 20 | 31 | 13 | Inland interm., prob. 2. 8. N. Rhoads 1840 g.ad. do. do. 104.5 | 68 | 61 | 21.5) 19 | 28.6) 12 | Probably . do. 1565 yg. ad. | New Jersey, Tuckerton do. 104 70 | 63.5} 24.5! 23 | 33.5] 11 do. 3896 g.ad. | New Jersey, Mickleton |: do. 107 70 | 63 | 23 12 E. A. and O. Bangs) 3537 | old ad. (| North Carolina, Raleigh | do. 104 PAG G2 ea 622 yb 22 ESS sales: do. 3538 | yg. ad. 9 do. do. 103 65.5] 61 | 21.5} 21 | 30.5) 11 do. 5749 | yg. ad. 3) Florida, Micco L. h. vaga Bangs 108 | 71 | 71.2) 24 | 18.6 35 | 16 | Type (fide Bangs). do. 4995 | ad. Q | Florida, Roseland do. [101] | 70.3] 67 | 21.8} 17.8) 30 (fide Bangs.) WagnerlInst., Phila) —— ad. Florida, St. John’s Riy., Vol- do. 105 | 72 | 67 | 24 | 22 | 34 | 18.2 usia Co. S. N. Rhoads 1580 | old ad. °3'| Florida, Tarpon Springs | do. 116 | 79 | 76.5) 27 | 20.5) 39.5) 20 do. 616 | yg. ad. | Washington, L. Kichelos L. h. pacifica Rhoads | 115.5) 72.5) 69 | 25 | 20 | 36.5) 12 | Type. do. 303 | old ad. Q | Washington, near Tacoma do. ALO evi ale OMe OIE | PAS eral G Smithsonian Inst. | 8686 old ad. | Alaska (coast ?) do. 115.5 | '74.5| 70.4] 27.3] 24 | 41 | 16 | Col. by Dr. T. T. Minor. do. 8687 | old ad. do. do. 119 | 83 | 76 | 84 | 25 | 49 | 14 do. do. 8688 old ad. do. do. 110 78 | 73 | 27 | 18 | 41.5) 15 do. E. A. and O. Bangs) 6965 | yg.?ad.-f'| Newfoundland, Bay St. George | L. degener Bangs 101 66 | 60 || 22 | 19.5) 32.5) 11.5) Type. do. 6966 | old ad. ° do. do. [98] | 70 | 63 | 22.8) 19 4) 33.6 Topotype (fide Bangs). do. 3755 | yg. ad. 2 do. do. 93 | 64 | 56 | 19 | 18.8) 25.8) 10 | Topotype. CONTRIBUTIONS TO A REVISION OF THE 438 ‘ad {yodo J, © ad ky, sqrt 2T6r ‘Gusta ‘ad ay, “mu04a -[OYS SNOJUIMVSTT WOT “SvdTL “ad LT, ‘SUT OT “PUS10.AN ‘SQL AT USIO AA 'Sq[ GT JUSTO AY “VWI PIUL.L9} UT ‘uoads poyunout ‘addy pur(uy “ad A} parpuy ‘o4VIpIULoJ Uy ULYS payNgys ULOTZ 19 YVY “SvOT *“ULIOF ysvoo ‘OSIe'T “SYIVULOY, OIL “q004 PUT “erqoqzaA TEL, | L8P OOF OV 066 866 OOET G60L LIIT S861 OOTT O8IT PPIL 990T S60T 9TOT SOOT 89TT O6IT O60T “yysuaT [eo], ‘op sSuvg tlouesop Sproyyy V.LOUOS "Y ‘op sproyyy voytord “ty ‘op ssuvg vsva ‘ty ‘op ‘op ‘op ‘op (ang “q) vurxegry “YT ‘op ‘op “op (wsocy |. dgov'y,,) vormospny y “satoadg “Al “1 | ‘op asto9y 49 ‘gq ‘puv[punoyaonr vuozity “OO wdeanx ‘39019 IOAvegT ease avout ‘HOP ALYSe A, SOLAYOLY oyeryT “uoyBarysvAy aT[OUOIPD ‘CplLO[ puvypasoy “epiopy ‘op ‘op YSopRyy ‘wurpoaND YIALON [He Aqaoquy “norjoaumoy "oD uoqUTD ‘ermvapfsaua,t (Sosvqury 'T) uojdyQ ‘oureyy UOIsTUL yy ‘SPJosnpoUssv “ART OtpNos sayy ‘YOIMsuUNIG MON srpodeuy “14099 BAON “ABT *(suagauyyv Uv) SMONO UDIIW UPON fo SPUIMWAMNSDA TT 3 ‘pu £ “pe d “pr Pope as P ‘pv ‘sf Pru & ‘pe P “pu 4 P ‘pe ax hpog P “pu ‘ds | 9969 ‘op S969 ssuvg "OQ pur y “7 aa YSU UN ‘SHY “ouLy oe op 919 spvouy “N's 8661 OP. CUbP | sduvg ‘OQ pur *y “7 —_|| ‘op &SP ‘Op Top Aoptttg 9g "9 pur "EL TT ecep | Ssuvg *O pur sy "gq ONE “BUM d “109 "N “pvoy O6LP | ‘op | GBIP ayy OP ge9¢ | sung “O puny “a Ze | =e-4 Be *M01}99[ [09 &3 oO Figs. 1 and 1. Figs. 2 and 2. c=} Figs. 3 and 3. Figs. 1 and 2. NORTH AMERICAN BEAVERS, OTTERS AND FISHERS. 459 EXPLANATION OF PLATES. Plates X XT and XXII. (Scale slightly less than two-thirds natural size. ) Castor canadensis pacificus Rhoads. Topotype ; No. 1865, col. of S. N. Rhoads; old adult 3, from Lake Kichelos, Kittitass county, Wash. Superior and inferior, vertical aspects of same skull. Castor canadensis frondator Mearns. No. 32, col. of E. A. and O. Bangs; young adult 2, from Red Lodge, Mont. Superior and inferior, vertical aspects of same skull. Castor canadensis Kuhl. No. 31, col. of E. A. and O. Bangs ; old adult (probably <), from New Bruns- wick. Superior and inferior, vertical aspects of same skull. Plate X XI. (Scale four-fifths natural size.) Castor canadensis carolinensis Khoads. Type; No. Z. 609, col. of State Museum of N. Carolina ; old adult