The Postcranial Skeleton of the Giant Permian Pelycosaur Cotylorhynchus romeri

J. WlLLlS STOVALL, LLEWELLYN I. PRICE, AND ALFRED SHERWOOD ROMER

Museum of Comparative Zoology, Harvard University

HARVARD UNIVERSITY VOLUME 135, NO. 1 CAMBRIDGE, MASSACHUSETTS, U.S.A. SEPTEMBER 22, 1966 THE POSTCRANIAL SKELETON OF THE GIANT PERMIAN PELYCOSAUR COTYLORHYNCHUS ROMERI

J. WlLLlS STOVALL, LLEWELLYN I. PRICE, AND ALFRED SHERWOOD ROMER

Introductory noie. In 1937 fragmentary paper may contain, I have felt that the remains of a gigantic new pelycosaur from paper should be considered as a joint effort, the Oklahoina Permian were brought to the to give credit to Dr. Stovall for his work attention of Dr. Stovall of the University of in the collection and supervision of prep- Oklahoma. ~7hoshortly after published a aration of the material, and to Mr. Price for prelinlinary account of the animal, as Coty- his invaluable notes and drawings of the lorhynchus romeri. aided by notes and draw- Oklahoma materials. ings by Llewellyn I. Price and myself (Sto- -Alfred Shenvood Roiner vall, 1937). Subsequently, many additional specimens of this form were found: these MATERIALS n-ere prepared under Dr. Stovall's direc- tion. and studies and drawings were made Although a broad band of continental of much of the material by Mr. Price. It Pelmian beds extends across Oklahoma was planned that the material be described from the Kansas border to the Red River, in a joint paper by Stovall and Price, but relatively few vertebrates have been &s- owing to a variety of circumstances, this covered in the Lower Permian of the state was never 14 ritten. Nearly thirty years have -quite in contrast to the situation to the passed and except for a brief notice and south, in northern Texas. This relative pau- figures of the skull in the "Review of the city of finds appears to be due to the gen- Pelycosauria" (Romer and Price, 1940: erally higher rainfall and consequent better 419421). no further account of Coty- vegetative covering in Oklahoma, and to lor17ynclzu.s has appeared. hgeanwhile Dr. the more level topography of Oklahoma, in Stovall has died, and h4r. Price is now contrast to the Texas area where the pres- resident in Brasil. Adding to the need for ence of a number of limestones and heavy description is the fact that Dr. E. C. Olson sandstones makes for a more rugged to- has found additional remains of Coty- pography and the development of poten- Io~l~yi~cl~ns.and a number of related forms tially fossiliferous "breaks." have been discovered in a variety of locali- The discovery of abundant remains of a ties and horizons in Texas aild even in large new type of caseid pelycosaur, Coty- Russia (Olson. 1962: 24-47). In ths situa- lorhy~zcl~usron~eri, was in consequence a tion, it has been agreed that I should pub- major event in the history of paleontological lish a general account of the postcranial work in Oklahoma. As noted in Stovall's anatomy of Cotylorhynclzus. But although paper (193'7: 308): the first find of Coty- I alone should be held responsible for any lo1.11ynchz~smaterial was made by Dr. W. S. inaccuracies or misinteil~retationsthat this Strain (then a graduate assistant at the Bull. Mus. CcImp. Zool., 135(1): 1-30, September, 1966 1 2 Bulletin Museum of Con~parativeZoology, Vol. 135, No. 1

University of Oklahoma). The site lay in the Hennessey shales, 4% miles west of Kavina, Logan County. The type specimen Specimen No. Humerus Radius Femur Tibia consisted of an incomplete skull and jaw. a 4-0-6 4-0-2 (1249) front foot, and interclavicle. Shortly after. 4-0-4 a nearly complete postcranial skeleton was CNHM 272 found at about the same horizon in the AhslNH 7517 Hennessey Formation. but close to the city 1250 (?) 4-0-16 of Norman and hence some 50 miles or so USNM from the region of the first discovery. and MCZ 3416 during the years 1937-1939 a very con- 4-0-13 siderable number of further Cotylorhyncht~s 4-0-3 specimens were collected in the Hennessey shales of the Nom~andistrict and prepared by laborers under Works Progress Admini- man area can be assigned to a single species stration funds. Most of this material is now as far as morphological characters are con- in the Stovall Museum at Norman. but cernedS1 There is however, considerable specimens were also obtained for the mu- variation in size. In reptiles generally. early seums in Chicago, New York. TVashington. growth is rapid. followed by a slower if and Cambridge. Preparation of the mate- steady growth; the natural result is that the rials at Norman was carried out under the greater part of specimens recovered in the general direction of Dr. Stovall. and notes case of fossil forms should represent rela- and figures on much of the material were tively young mature animals. with a small made by Mr. Price. In the preparation of percentage of overly large specimens repre- the present description. Romer has seen senting individuals which were exception- all this material and has utilized Price's ally long-lived and consequently exception- notes and figures in addition to his own ob- ally large: and a small percentage of young servations. A word of caution must be and immature individuals. The present ma- gi\ en, however. Although preparation was terials tend to show a rather greater size carried out under Dr. Stovall's direction. range than is common. As a rough index to it is not improbable that. with a consider- size, I list the length (in mm) of major limb able number of workers and with a number elements in a number of the better speci- of specimens undergoing simultaneous prep- mens ( Table 1) . aration, a certain amount of inaccuracy These specimens are deposited in the may have occurred in the restoration and following institutions: Chicago Natural association of materials. Further. in the History h4useum ( CKHM ) : American Mu- time that has since elapsed, some of the seum of Natural History ( Ahlh'H) ; United specimens have been moved several times States National hluseum (USNM): Museum for storage and re-cataloging and further of Comparative Zoology, Harvard ( MCZ ) . chances of error have crept in. Other specimens (numbered) are in the The reptilian remains from the Hennessey University of Oklahoma collection. shales of the Norman region appear. curi- As can be seen from this table. the first ously, to pertain almost entirely to Coty- five specimens listed appear to be of rela- lorhynchus. with very few remains of other tively uniform large size. Below this there reptiles and amphibians-a situation in is a sharp drop to three specimens which are 20 per cent or so smaller than the first strong:u contrast to the usual mixed assem- blages generally present in typical Texas Dr. Olson infornls rile that there is a possible redbeds deposits' As far as be seen, specific difference between the Norman material the Cotylorlzynchus materials from the Nor- and the type from Navina. COTYLORHYNCHUSSKEI~ETOK Stovall, Price, and Romer 3

group but appear to be essentially mature; of structures given here are based on a below. with the MCZ specimen as a tran- synthesis of a number of specimens. Wher- sition there is a drop to small and seem- ever possible the illustrations are based on ingly immature specimens such as 4-0-3. specific examples. although frequently with Possibly there is a specific difference be- the addition from other individuals of de- tween the first and second groups. Equally tails missing or obscured in the specimen possible. however the contrast is a sexual primarily utilized. one: in Dimetrodon limbattu. for example. It became apparent. even from the ma- there is a seemingly clear-cut size difference terial described in Stovall's preliminary of 10 per cent between the sexes (Romer paper, that Cotylo~hyiichuswas a @ant rel- and Price, 1940: 341342). ative of Casea. a small Clear Fork Texas In typical Texas redbeds localities. fossil pelycosaur described by Williston ( Willis- reptile remains are most conlmonly found in disarticulated con&tion. and when found ton. 1910: 590-592: 1911: 111-131, etc.; articulated appear to have undergone de- Romer and Price. 1940: 412419). Despite position in a variety of poses. One gets the cranial differences, Casea and a number of impression that this is the result of stream other Texas pelycosaurs show such a large action, and that many of the "bone pockets" number of diagnostic postcranial similari- represent back eddies in streams in which ties to Eduphosaurus that Romer (Romer cadavers brought donnstream tended to and Price, 1940: 21. 366-378) felt justified collect (and decompose). But in the Hen- in including these forms with Edaphosaurus nessey shales of the Norman region. many in a common suborder Edaphosauria. As in of the specimens show clearly that the Casea, the Cotylorhynclius postcranial skele- Cotylorhyizchus individuals were generally ton agrees in almost every particular with buried in articulated and undisturbed the characteristics assigned to the Edapho- fashion, right side up, with the limbs spread saulia as a whole. In consequence. in the outward at the sides. This strongly suggests description which follows. comparisons are. that we are here dealing with entombment in general, with Cnsea or other edapho- of quite another sort. South African Per- saurian types. mian pareiasaurs, equally large and clumsy herbivores, are typically preserved in simi- lar fashion (Watson. 1913). There are two VERTEBRAL COLUMN possible interpretations. Watson, in the The general characteristics of the Coty- case of the pareiasaurs. implies death from lorlzynclzus vertebral column are almost ex- starvation, and suggests a covering of the actly those cited in the "Review of the cadavers in situ by loess-like materials. I Pelycosauria" for the suborder Edaphosauria would favor an alternative interpretation in as a whole: both cases-that the animals were bogged down and entombed in swamps. "The number of presacral vertebrae is sometimes reduced. The dorsal vertebrae are moderately to Study has been greatly handicapped by greatly elongated; the cervicals are small in all the nature of the specimens. Due in part. dimensions. The dorsal centra are spool-shaped, perhaps, to a rather spongy structure of the with rounded bottoms, and lack any trace of ven- skeletal elements and to the nature of the tral keel or lip. Intercentra are small. The dorsal transverse processes are moderately elongated and sediments in which they were embedded, rise from a more antero-dorsal position on the arch much of the material has been subjected to than in other pelycosaurs. The zygapophysial crushing and distortion. u7itll a consequent surfaces are large, moderately tilted, and extend limitation of the accuracy with which res- farther laterally than in spkenacodontids. The neural arches are not excavated above the dia- toration and illustration can be made. In pophyses. . . . The atlas centrum reaches the ven- great measure description and illustration tral surface of the column." 4 Bulletin .?.lzrse~,mof Comparative Zoology, T701. 135: No. 1

Only a very few qualifications on ths terminal. are present in seemingly articu- definition need be made to fit Cotylorhyn- lated fashion. However. there is a suspi- clzzi.s. (1) It was stated that the presacral ciously sharp drop in the size of centra count is "sometimes" reduced. At the time following element 27, and a photograph of this was written, the figure was known the specinlen made during preparation only in Casea where apparently 24 or 25 shows that the last 16 vertebrae were not presacrals were present (Romer and Price. part of the block containing the main part 1940: 417). It is now known that Edapho- of the material. although they may well saurus boalze~,geshas likewise reduced the pertain to the same individual. Measure- primitive pelycosaur number from 27 to 23 ments suggest that there is a gap here of presacral vertebrae (Shuler and Witter, about 10 vertebrae. to raise the probable 1942). It is thus a reasonable inference that total count to about 53. In the CNHM presacral reduction was general in the sub- specimen, 37 caudals are present in articula- order, and reduction is present in Coty- tion with the trunk; following a gap of lorl~yncl~us.(2) "Intercentra are small." about 23 cm in the slab there are eight Few are present in any Edapl~osaurusma- further. much smaller. vertebrae. the last terial; only a single intercentrum was found apparently terminal or sub-terminal. The in the Casea material (in the sacral region); gap is of such length as to suggest that 10 none are known in Cotylorhynchz~sexcept vertebrae are missing here. giving a prob- for the atlas-axis. Small gaps between the able total length of 55 vertebrae. This is a ventral edges of the centra in all three reasonable pelycosaurian number. genera suggest that tiny intercentra were Dorsal ueriebrae. Such a dorsal vertebra present in cartilaginous form. (3) "The as. for example. vertebra 12 of 4-0-6 (Figs. atlas centrum reaches the ventral surface of 1C. 2C), exhibits a typical pelycosaurian the column." This appears to be true of and. further, edaphosaurian character. The Ednphosati~t~s,and Williston (1911: pl. centra are edaphosauroid-spool-shaped XIV) restores this centrum as reaching the structures without any of the tendency seen ventral surface in Casea. As described be- in ophiacodontoids and sphenacodontoids low, however, it does not reach this surface for develoument of a ventral keel. The in Cotylorlzynchus, and Williston's material centrum. as in Casea. is relativelv short and of Casea was obviously imperfect and open stout, its length, averaging about 60 mm in to misinterpretation. large individuals, being but about 1% times The Co-iylorl~ync12z~sromeri column in- the end height of 38-40 mm. The ventral cludes definitely 25 and probably 26 pre- surface is broad, essentially flattened: but sacral vertebrae, 3 sacrals, and about 55 slightly convex; at the margins of this area caudals. It has roved difficult to deter- the sides curve upward sharply and, above mine the presacral count in Cotylorhynclzus. this point, are slightly in-pinched. There There are several specimens in which dor- is no visible suture between arch and cen- sals, lun~bars,and sacrals are present in trum. On the sides of the centrum. at about well-connected series: generally, however; the level where the base of the neural arch the cervicals are poorly preserved or absent. pedicel would be expected, there is a low I11 4-0-6, however, the cervicals are present. longitudinal ridge, rounded at its summit, There are definitely 25 well-preserved pre- running lengthwise from a point well below sacral vertebrae, and apparently one further the upper margin of the centrum anteriorly poorly preserved one. Several specimens back to a somewhat higher position pos- sho\s a sacrum of three vertebrae. The teriorly. Above this point the upper part of tail is preserved in but few cases. In the the surface of the centrum and the lower mounted slab of 4-0-2 ( 1249): a hind leg lateral surface of the neural arch form a and tail, 43 caudals, the last apparently longitudinal depression extending length- COTYLORHY~~CHUSSKELETON Sto~all, Price, and Rome?. 5

Fig. 1. Lateral views of representative vertebrae, in side view. A, atlas-axis, X 2/3; B, vertebra 6, X >I2;C, vertebra 12, X 1/3; D, vertebra 21, X '/3; E, proximal caudal, X '1,. wise beneath the base of the transverse length. In edaphosauroid fashion the trans- process. verse processes arise from a far anterior The ends of the centrum are of a typical position on the neural arch, and their front pelycosaurian nature-essentially circular in margins extend almost directly outward from outline. with thickened edges. At either the level of the prezygapophyses, There is end this circle contains a cone-shaped de- in this region no ventral expansion of the pression for the notochord, the two cones process. which is thin dorsoventrally but connected by a small foramen. Both an- broad anteroposteriorly: there is thus no teriorly and posteriorly the rim of the cen- continuum in articular areas between ca- trum is somewhat thickened laterally for pitulum and tuberculurn. The transverse the articulation of the capitulum. which process here is directed almost straight lat- was thus inter-central in position. erally, with. however. a slight upward and The transverse processes are highly de- folward tilt, rather than the downward sIant veloped in this region. extending outward of most pelycosaurs (Casea excepted). The on either side to a distance from the mid- base of the process is very broad. extending line approximately equal to the vertebral nearly the whole length of the vertebra. 6 Btilletin M~ise~imof Con~paraticeZoology, Vol. 1.35, No. 1

Fig. 2. Posterior views of the same vertebrae as those of Figure 1.

The broad articular surface for the tuber- pletely preserved the rugose end of the culum faces ventrally and only slightly lat- spine is expanded laterally on either side erally on the under surface of the process. and indented medially. to give a somewhat Since the tuberculum is little raised; the bifurcate appearance. much as in. for ex- proximal portion of the rib continues out- ample, some parts of the Eryops column. ward in the line of the transverse process. Ce1.2jicals. Progressing forward through The zygapophyses are normally constructed, the anterior dorsals into the cervical series with a verj7slight median tilt of the articular the vertebrae are increasingly lightly built. surfaces. The neural arches, as in edapho- as tends to be true of pelycosaurs generally. saurs generally, lack the lateral excavation but more especially of edaphosauroids. such seen in sphenacodontoids above the trans- as Cotylwl~ynchus,in which the head is verse processes. The neural spines, as in relatively small. The centra decrease steadily Casea, are slender transversely but long in diameter and length anteriorly. and the anteroposteriorly, and are low (although width tends to equal the length. In the not as 10~1,relatively, as in the last-named 6th vertebra of 4-0-4. for example (Figs. genus). In such mid-dorsals as are com- 1B. 2B), the length is about 50 mm, the COTYLORHYNCHVSSKELETOS Sto~al/, Price, and Romer 'i

width about 45 mm. The flattened ventral not reach the surface here; possibly ths gap surface of the centra persists into the cer- may have been filled by a ventral car- vical region. In the cervicals the transverse tilaginous continuation of the bone. processes are shorter and lightly built. in Intercentrum 2 is very well developed correlation with the decreased size of the and bears a distinct rib facet. The axis cen- ribs. Anteriorly, the cleft in the summit of tlum and arch are built in typically pely- the neural spines disappears, although some cosaurian-and especially edaphosaurian- distal expansion may persist; the spines be- fashion. The transverse process slants come shorter and somewhat more rounded strongly downward and to a slight degree -ovoid-in section. Proceeding fonvard, backward. The neural spine is, for an axis, the direction of the transverse processes rather short. but is stout. long anteropos- gradually shifts. In a typical dorsal it points teriorly, and expanded in a 3-pronged ar- somewhat forward and upward, but in the rangement posteriorly toward the summit. cervicals the processes become somewhat Poste~iordorsals. Passing backward along ventrally and posteriorly directed, in cor- the dorsal series. the vertebral centra in- relation with the direction of the ribs in the crease somewhat in length to about verte- relatively slender neck region. Even as far brae 17-19, beyond which a progressive forward as vertebra 3, the capitular facet is reduction takes place so that beyond this high up on the rim of the centrum, showing point (as is also the case anteriorly) width little of the tendency to descend anteriorly, exceeds length. In 4-0-4, for example, the seen in most pelycosaurs. The cervical length of vertebra 21 (Figs. 1D. 2D) is 51 zygapophyses are, like the dorsals, widely mrn. the breadth about 55 mm. An increase separated and with essentially horizontal in central width, however, persists. width articular surfaces. reaching its maximum in the sacral region, Ath-axis. The atlas-axis complex is pres- and the posterior dorsals are the most mas- ent and well preserved in 4-0-6, and is sive vertebrae in the column. In the more present also in 4-0-4 (Figs. lA, 2A). A posterior presacrals the neural spines are facet on the atlas neural arch indicates the increasingly long fore-and-aft and increas- presence in life of a proatlas. The atlas ingly narrow in transverse diameter, al- arch is bipartite and typically pelycosaurian, though the arches are broadly expanded at with a long posterior process articulating the spine bases. In the posterior dorsal with the prezygapophysis of the axis and series the two members of each pair of with a spur running farther back along the prezygapophyses are joined by a horizontal side of the axis neural arch. There is a ridge of bone bridging the space between short but distinct transverse process for rib them and forming a strengthening trans- attachment, turned strongly downward and verse element whch crosses the front end backward. The atlantal intercentrum is of the vertebra between the ribs of either highly developed for cranial articulation; side. Progressing backward. the transverse laterally there is a distinct facet for the processes are increasingly stout but decrease rib capitulum. As is well known, the atlas in length. In the last members of the series centrum is variable in development in pely- the processes are fused with the ribs. In cosaurs, reaching the ventral surface in the posterior vertebrae the area of attach- Dimetrodon, for example, but excluded from ment of the rib capitulum moves upward this surface by the second intercentrum in and backward to an oval area on the an- Ophiacodon. The situation in Cotytorhyn- terodorsal surface of the side of the cen- chus is not too clear. In both specimens trum below the transverse process. Here. in which the atlas-axis is preserved, there as in the transverse process. fusion with the is a distinct ventral gap between inter- rib takes place in the last few members of centra 1 and 2; but the atlas centrum does the dorsal series. 8 Bulletill A4zrsezrin of Conzparative Zoology, Val. 1.3,5, No. 1

Fig. 3. Left, ventral view of last presacral, the three sacrals, and first two caudal vertebrae, X 11,. Right, above, lateral view of mid-caudal vertebrae, X 'I2. Right, below, posterior and ventral views of o mid-caudal vertebra and a mid- caudal chevron, X I/>.

Sacrak (Fig. 3). Three sacral vertebrae structure. The neural spines of the first two are present, as noted in the description of sacrals are similar to those of the "lumbars"; the ribs. The centra are broad but appear that of sacral 3 is shorter fore-and-aft and to be shallow dorsoventrally (an effect per- less compressed from side to side. haps due to crushing). In one specimen. at Caudals (Figs. 1E; 2E, 3). The most least, the centra of the first and second anterior caudals resemble the last sacral sacrals appear to be fused; whether the last closely. The breadth of the centrum is two were fused is uncertain. On the first greater than the length or height: the rib sacral the zygapophyses are still well sep- attachment areas are short but greatly ex- arated, but on the third vertebra the two panded, with a groove, presumably for an members of each pair are much closer to- intersegmental artery, marking the line of gether. a situation transitional to that in distinction between tubercular and capitu- the tail. There appears to be no ossification lar components. More posteriorly. the trans- of apposed zygapophyses. The transverse verse processes become greatly reduced in processes-firmly fused to the ribs-are size (with a concomitant reduction in rib short. but very stout, and are colltinuous size), and process and rib have essentially with a large area 0x7 the centrum represent- disappeared by vertebra 12, although a ing the capitular attachment. A depression slightly projecting ridge is present as far as which represents the point of juilctioil of vertebra 20. Meanwhile. the centra have tubercular and capitular areas is seen in begun a change in their proportions. so that some specimens; it can~lotbe determined by mid-length of the tail the length is con- whether an arterial foramen perforates the siderably greater than the width and the COII.I.LORHI~A.CHUSSKELETOK . Stocall, Price, und Ronzer 9

Fig. 4. Above, right ribs 1-8, seen from the posterior aspect, X '1,. Below, the heads of righi and left ribs from the same mid-dorsal segment of a single individual, to show contrasts in post-mortem distortion, X '/.

11eight also exceeds the width. The zygapo- normal. the length decreases posteriorly. phg-sial pairs have come to lie close to each although but slo\vly. In one specimen. as in other near the mid-line, and their articular pelycosaurs generally (and in Casea) , there surfaces have becoine sharply tilted to a is. in the proximal elements. a basal inter- plane close to the vertical. There is. further. central component conilecting the tw-o anns a gradual reduction in height and stoutiless of the chevron: but in the CKHM specimen of the neural spines. so that by the time not even the most proximal chevrons have the mid-caudals are reached. the spine is a proper intercentruin. there being merely a low nubbin projecting a short distance an enlargenlent of each of the two proximal above the postzygapophyses. The neural ends of the chevron structure. Although arch as a whole is much reduced, and in the most posterior part of the tail the material both of the adjacent centra are bevelled shon-s little evidence of any structure above for reception of a chevron. the centruin the centra. anterior to the chevron has the most ob- The first of the chevrons appears. in the vious articulations: it bears ventrally a pair CNHhl specimei~(No. 272). between the of well developed facets, each supported by 3rd and 4th caudals: they continue back a longitudinal ridge 011 the centrurn. In the as far as vertebra 36. at least in one speci- tail the bottom of the centruln sho\vs a inen. The first elements are stoutly built, longitudinal depression, bounded on either with a leilgth equal to two centra; as is side by these ridges. 10 Bulletin Museum of Comparatiz;e Zoology, 1'01. 1.35, No. 1

Fig. 5. The proximo1 port of right ribs 9-21, seen from the posterior aspect, X '1,. COTYLORHYNCHUSSKELETOK Stouall, Price, and Ronzer 11

and vertebrae gives. on the average. a pic- ture closely comparable to that given by As in all known pelycosaurs. ribs were will is to^^ (1911: pl. XVII, fig. 1) for a present on every vertebra from the atlas to Casea mid-dorsal. and shows the presence and including the proxin~al caudals. For of a very broad. barrel-like trunk. with a most parts of the series. specimel~s4-0-6 probable transverse diameter of about 60 and CNHM 272 furnish the best material. cm-i.e.. about two feet. Do~.sal7.ibs (Fig. 5). In a typical dorsal Except for the rather flattened capitular rib the shaft extends outward (and in life region, typical dorsal ribs are essentially somewhat upward) from the capitulum, oval in section, and so tilted in life that the the articular surface of which is an oval, long axis of the oval slants downward and narrow anteroposteriorly. occupying the en- forward well over 45",with the two major tire head of the rib. The tuberculum in the surfaces essentially anterior and posterior mid-dorsals does not project to any marked in position. The anterior margin is in all degree from the shaft; it apposes to the typical dorsals smoothly rounded. On the transverse process a concave articular sur- posterior surface a ridge develops at a point face with an oval shape, broader distally. somewhat distal to the tuberculum and the length about twice the width. rather toward the ventral margin. This ridge The proximal end of the rib appears to extends outward along the straight lateral run distally slightly upward to and past the segment of the rib. gradually approaching tubercular region. Shortly beyond ths point the ventral margin of the rib and, decreas- the rib curves markedly outward and down- ing in prominence, disappears at about the ward to encase the flank. The degree of point where the median curvature begins. curvature in life is difficult to determine, There is little expansion of typical dorsal due to variable post-mortem crushu~gand ribs at any region. The diameter of the rib compression; see, for example. the marked is greatest a short distance distal to the contrast between the two members of a rib- tuberculum: beyond this region there is pair from the same segment of a single in- generally a very gradual decrease to a mini- dividual shown in Figure 4. In most in- mum of about two-thirds of the maximum. stances there is a considerable segment of Progressing anteriorly, from such a typi- the rib. lying beyond the point of proximal cal dorsal as that described, to rib 9. a curvature. which is nearly straight; this in- series of gradual changes may be observed. cludes about half the total rib length. This The length decreases, rib 9 in 4-0-6, for segment appears to have been directed di- example, being but 430 mm in length. i.e. agonally outward and downward. Distally, about 60 per cent as long as the longest rib. beyond this straight section. the rib curves and the ribs are more lightly built. In an- gradually inward toward the mid-line. Al- terior dorsals, in contrast with mid-dorsals. though seldom is a complete rib length the tuberculum projects markedly above preserved, it would appear that the greatest the general outline of the dorsal rib margin, length of a rib (measured from the tuber- and, in contrast, the capitulum is sharply culum) is about the length of 13 dorsal turned downward medially from the line of centra: two mid-dorsals of CNHM 272 mea- the shaft. There is here only a slight curva- sure 721 and 718 mm. In such dorsals as are ture beyond the tubercular region; when completely presened the distal end of the the rib is articulated with the vertebra. it is rib exhibits a cupped termination with obvious that its shaft runs more directly which the cartilaginous section of the rib downward than in the typical dorsals, in- presumably articulated. dicating a narrowing of the body in the Despite the handicap noted as due to "chest" region, as the level of the shoulder post-mortem distortion, a~ticulatioi>of ribs girdle is approached. Distal to the straight 12 Bulletin iMlrse~r17zof Coinparatitic Zoology, Vol. 1.35, KO. 1

descending portion of the shaft, there is a most anterior ribs indicate that these ribs final segment curving some\vhat in~~ard.were shorter still. In correlation with the The total distance from the tuberculum to wide separation of the two points of attach- the beginning of the median curvature is ment to the vertebrae, tuberculum and little less than in rib 14, indicating that the capitulum are strongly divergent, giving a "chest" is nearly as deep as the belly, al- V-shape to the proximal part of the rib. though narrower. The slender distal portions of the ribs ap- The longitudiilal ridge found on the pos- pear to be sonlewl~atcompressed antero- terior rib surface changes in position as one posteriorly. progresses forward. It has shifted upward, I11 the cervicals, as in the most anterior so that its proxiinal portion lies along the dorsals. the rib shaft runs straight distally dorsal margin of the rib, and its distal end from the tuberculum, indicating a narrow I lies at about the middle of the posterior neck region; the tuberculum, as far as pre- ,I surface, rather than to\vard its ventral mar- served in these ribs, retains the somewhat gin. It has: further, moved proximally, so distinct character seen in the most anterior that its proximal end has attained the distal dorsals. On rib S the longitudinal ridge margin of the tubercular projection, and its retains the character seen in the rib follow- distal end extends only part-way down the ing for much of the proxiinal half of the vertical segment of the rib. Distally the shaft, forming a sharp dorsal margin of the anterior dorsal ribs remain sub-circular in rather thin rib. but it fails to reach the section. Proximally, however, the posterior tuberculum. On rib 7. the ridge is present surface becomes much hollowed out. with on the middle third of the shaft; on the the proximal part of the longitudinal ridge more anterior ribs it has disappeared. It forming a very prominent projection dor- is in a sense replaced by a thin dorsal flange i sally. -4s far as can be determined in the extending distally on rib 7 a short distance usually crushed con&tion of the specimens, outward froill the tuberculum. but not con- the most anterior dorsal ribs are somewhat tinuous distally with the typical dorsal ridge. expanded distally-notably rib 9 in 4-0-6. This flange is present in reduced form on Ce.i.z;ical ribs (Fig. 4). In the few pre- rib 6; more anteriorly, as far as can be seen viously known pelycosaurs in which the from the material, no noticeable structures 1 distal ends of the ribs have been ~ellpre- are present on the rib shafts. Ribs 6-8 served, it appeared that rib S was stout appear to have lain beneath the scapula; ! distally and presumably had a sternal con- these reduced flanges. presumably func- nection. establishing 7 as the probable num- tioning for the origin of serratus muscula- ber of non-sternal. i.e., cervical, ribs. In ture. contrast with the much greater flange 4-0-6, however: rib 8; although elongate development seen in many early tetrapods. (with a length of about 360 mm) and Posterior dorsal ~ibs(Fig. 6). Proceed- resembling the dorsal rib following it in ing backward along the dorsal series, the most regards, tapers to a point distally. It posterior dorsal ribs as far as about rib 20 thus failed. obviously: to reach the sternum appear to be essentially similar to more and must be considered technically to be a anterior dorsals in character, and show little cervical. decrease in length; however. the proximal In 4-0-6, remaills of all the cervical ribs end of the longitudinal ridge described for are present (but the third is poorly pre- the anterior dorsals retreats distally to a served), Although the rib-tips are incom- sinall extent in this region.- Rib 20 becomes plete, it is clear that all \yere slender and broader proximally; in the posterior ribs pointed distally. Froin rib S forward to rib there is a gradual approximation of tuber- 4 there is a sharp and steady dimin~~tionin cular and capitular heads so that, from lengtl~,and the preserved portions of the about rib 22back, capitular and tubercular COTVLORHY~VCHUSSKELETOK Stocall, Price, and Romer 13

Fig. 6. left, posterior presacral ribs of the right side, X '12. R~ght, left clavicle, external and internal views, X '1. areas are essentially fused into a single teriorly to a maxiilluln in rib 24, which is articular sulface. A decrease in length of very broad in proportion to its length. ribs is not marked until approximately rib Snc~alribs (Fig. 3). In correlation with 23, which is notably shorter than that pre- the width of the trunk and pelvic region. ceding it (with a length in CNHM 272. the sacral ribs are longer than in sphena- as preserved. of 413 mm) and ribs 211-26 codonts and o~hiacodonts. Three sacral are increasingly short, the last being about ribs are present. These are tightly fused 158 mnl long in CNHM 272. As in pely- to their vertebrae over a broad area; this cosaurs generally, the freedom of rib artic- includes the short transverse processes whicb ulation with the vertebrae decreases pos- arise from much of the lateral surface of the teriorly; the last 3 short ribs appear to be centra. No clear sutures are seen. but dor- well fused and im~novableand hence are to sally a rugose anteroposterior ridge marks be considered as lumbars, and the next pre- the line of fusion of rib and transverse proc- ceding may also have been imnlovable in ess. The rib Bead is deep dorsoventrally, life. although in 4-0-4 a suture between rib as well as anteroposteriorly. Distinction be- and vertebra was seen during preparation tween capitular and tubercular attachment for the fourth presacral. is indicated by depressions on both anterior All the more posterior ribs tend to have and posterior surfaces about half way down; shafts which are relatively broad and thin these depressions have not been excava- for most of their length. the last 5. however, ted. but may have been connected by a definitely taper to a point distally and thus canal for the intervertebral artery. lack a sternal connection. and the same The first sacral rib is short but stout. may be true of rib 21. As far as segment Beyond the head it contracts somewhat in 23 the ribs continue to be curved, turning width and extends outward horizontally and doun\vard and backward in life. Ribs 24- slightly posteiiorly to terminate in a rela- 26. however, are nearly straight and di- tively thin expanded blade extending do\vn- rected laterally. their length being little ward and apposed laterally to the i~lner more than that of the proximal nearly face of the ilium. The second rib extends straight segment of rib 23. The breadth of directly laterally; it is siiuilar to the first the proximal part of the rills increases pos- but slightly less developed. The third rib 14 Bulletin hluseum of Conlparatio~Zoology, Vol. 135, No. 1

Fig. 7. left, external, and right, internal views of the left scapulocoracoid. Broken lines on the internal view outline the portion preserved in No. 4-0.6. Csnter, dorsol, right lateral, and anterior views of the interclovicle. X is essentially accessory in nature; it is more ward curvature of the distal end is pro- slenderly built and curves forward to but- nounced. These ribs show a steady decrease tress the second rib as well as apposing the in length and stoutness. Rib G is notably ilium with an only slightly expanded tip. shorter, with little development of a distal The first two sacral ribs are fairly com- curved segment, and from ths point back parable to those of Casea. In that genus, the caudal ribs, fused to the transverse proc- however, sacral rib 3 is well developed. and esses, are laterally projecting, tapering, Coty1orhynchu.s here exhibits a condition and pointed structures which decrease to which is less advanced. Edaphosaurus small nubbins and disappear, except for shows an intermediate condition in the de- low rugosities, beyond vertebra 11. In gen- velopment of sacral rib 3. eral the caudal ribs are comparable to those Caudal ribs. As in pelycosaurs gener- of Casea. ally. ribs are present in the proximal tail region. The first five show a fused attach- GIRDLES ment to transverse process and centrum with a pattern similar to that of the sacrals, al- SI.l.oulder girdle. A nearly complete scap- though with a steadily diminishing attach- ulocoracoid, as seen from the inner side, ment area; the heads are pierced antero- is present in 4-0-6, and the figure is based posteriorly by canals for the intervertebral primarily on this specimen. The outer sur- artery. ~1~~ first caudal is to face is not available in this specimen. but the third sacral in most regards, but it ex- several others show this aspect. The scapu- tell& directly laterally, with a slight distal lar blade, illcomplete in 4-0-6, is better posterior curvature; to terminate in a pointed preserved in other specimens; its exact tip. Posterior to this, all the caudal ribs: as height, however, is difficult to determine, in pelycosaurs generally, continue to extend due to the lack of distinctive features in this outward in a horizontal plane and to taper area. The blade in large specimens may to distal extremities. In caudals 2-5 a back- have been somewhat more developed. In COTYLORHYNCHUSSKELETOK . Stouall, Price, and Ronzer 15 all available specimens the dorsal margin of in contrast to the gentler curvature found the blade shows an unfinished surface. so in pelycosaurs generally. so that the sur- that there may have been a considerable faces apposed to the clavicles are essen- cartilaginous suprascapula. No sutures be- tially in a vertical plane. The head of the tween the presumed three elements could intei.clavicle is short and T-shaued.A, in con- be determined. trast to the diamond-shaued head of many The scapulocoracoid (Fig. 7) is edapho- pelycosaurs, and is nearly completely oc- saurian in general- character. It is short dor- cupied by the pair of crescent-shaped sur- soventrally. broad anteroposteriorly. with faces for the clavicles. These surfaces, which much the proportions of Lupeosaurzrs extend far laterally, are covered by rugose (Casea. presumably in relation to its smaller striations; the areas on the clavicles which size, has a girdle of more slender build). meet them are similarly rugose, and set The scapular blade is very short and broad, in distinctly below the general level of the its breadth distally due to a backward flare '6inner" . surfaces of that bone. Obviously of the posterior margin greater than is seen the contact between clavicles and inter- in any other pelycosaur. Below the point clavicle was an intimate one. of clavicular attachment the margin of the Pelcic girdle. Pelvic girdle material scapula curves out widely ant&iorly, to (Fig. 8) is relatively poorly represented. give great breadth to the lower part of the The Cotylorhynclzus skeletons were nearly blade. as in Edapl~osaurusand Lzipeosatlrus. all buried with the dorsal surface upward; In contrast to Edaplzosaurus and Ltrpeosau- the ilia in consequence have been generally rus, but in agreement with Casea, Nito- subject either to damage by crushing dur- saurus, and 3dycterosau~us. there is no ing entombment or to erosion prior to dis- supraglenoid foramen. There is, as in all covery, and this element is well preserved pelycosaurs. a screw-shaped glenoid cavity; in only a few cases. The iliac blade is it is here. as in other edauhosaurians. re- moderately high, as in sphenacodonts and markably deep anteriorly. Below its anterior other edaphosaurians: and in contrast to end is the external opening of the supra- ophiacodonts, and agrees with other eda- coracoid foramen. Posteriorly. the coracoid phosaurians in that (in contrast with sphen- region shows little development of a tu- acodonts) there is little posterior elonga- bercle for originu of the coracoid head of the tion. There is a narrow but well-defined triceps. On the inner surface. the upper longitudinal area for muscle attachment portion of the subcoracoscapular fossa is at the top of the inner surface. above the shallow, in correlation with the absence of areas for the sacral ribs. The first sacral a supraglenoid foramen. appears to have been in contact with a de- No &thrum is present in the available pressed area at the anterior margin of the material. Much of the clavicles and the inner surface of the blade. the second pre- interclavicle are present in 4-0-6 and in the sumably apposed to a flat area posterior to type (Figs. 6; 7). In contrast to Edapho- this; still farther posteriorly. a well-marked saurus and sphenacodonts. there is little internal depression received the small distal ventral exuansionA of the clavicle (the clavicle end of the third sacral. of Casea is unkno\ml). Its upper portion is The acetabular region is of a typical pely- well grooved posteriorly to clasp the anterior cosaurian nature, with the usual primitive edge- of the scapula. The blade of the in- dorsal buttress. In most suecimens sutures terclavicle is unusually broad and short. between the three pelvic elements are not and there is no development of the longi- clearly seen; in one specimen part of the tudinal ridge- seen on the ventral surface of sutures could be made out on the external the shaft in most pelycosaurs. As preserved, surface, and in several instances lines of the head turns sharply upward on the shaft, striae on the inner surface indicated the 16 Bullefin Alrlsetcin of Cornpamti~eZoology, Vol, 1.35, No. 1

Fig. 8. Right pelvic girdle in dorsal ond loterol views, X y3

areas of fusion of ilium with pubis and right angles from the front end of the pubic ischium. symphysis, of 173 inm. There is but a slight The puboischiadic plate is of very large indication in the material of the develop- size, with an anteroposterior length in 4-0-4, ment of a pubic tubercle, such as is found for example, of 380 111111, a depth below and in Edaplzosau~us,Nitosau7.u~. and Casen. iilterilal to the base of the ilium of 112 mm The great developinent of the puboischiadic and a breadth of the pubis, measured at plate is comparable to the t)ye of structure COTYLORHYXCHUSSKELETOX . Stouall, P~ice,and Ronlei 17

seen in Edaphosazr~us and Casea. and is and well preserved in a single specimen. associated with bodily breadth, present in The pose was obviously that colnnlon to all the pelvic region as well as farther forward. pelycosaurs, with humerus and femur pro- The plate was strongly tilted outward at jecting nearly straight outward horizontally. somewhat more than a 43" angle. so that it and with the lower limb segment essentially is not seen to any great degree in side view. vertical in position. Front and hind legs Internally. the puboischiadic plate ex- appear to be nearly equal in length, but hibits, as in pelycosaurs generally. a gently with the humerus slightly shorter in over- hollowed out area. in which the bone is all length than the femur. I11 all pely- relatively thin, along the middle third of its cosaurs the lower limb is much shorter than length. Anterior and posterior to this. on the proximal segment, but this relative pubis and ischium respectively. thickened shortness is very marked in Coty70~7~yi~- bony areas-essentially supporting struts- chus. Here radius and tibia are only ap- descend from the iliac region to the sym- proximately 60 per cent as long as humerus physis. The presence of these thickened and femur, respectively. In Oplziacodon the areas is reflected in the symphysis. \vhich radius is about 77 per cent of the humeral is thickened in both pubic and ischiadic length, the tibia about 83 per cent of the regions. with a relatively thin internlediate length of the femur; in Dinzet~*odonthe zone. Anterior to the ridge descending the comparable figures are 82 and 83 per cent. pubis is a very broad area of origin for The Cotylo~lzyncizusproportions, however, puboiscl~iofemoralisinternus, which is but are comparable to those in other edapho- slightly turned outward from the general sauroids; in .Edapizosaurus boanerges we plane of the internal surface of the plate. find figures of 62 and 37 per cent. Like Within this area is the usual internal open- other edapl~osauroids;Cotylo~hyizclzzrs was ing of the obturator foramen which opens "low-slung." externally below the p~~bicportion of the Hzrmerus (Fig. 9). The humerus is of acetabulum. There is a slight indication the primitive tetrahedral type, although the in certain specimens of a small area along shaft is stouter than in typical (and smaller) the dorsal margin of the ischium for an pelycosaurs. As figured, the ''twist" of the ischiotrochantericus origin. ends on one another is extreme; this, how- ever, appears to be due to crushing of the LIMB ELEMENTS (none too well preserved) specimens upon wllicll the drawing is based; specimen 4-0-3 I11 almost all instances the limb elements sl~o\vsthe "twist" to have been one of ap- have been crushed and flattened dorsoven- proximately 90". As is sphellacodonts and trally, so that they are essentially two-di- cdaphosauroids generally. the proximal ar- P mensional-a situation making correct re- ticular surface curves posteriorly well construction difficult. down onto the ventral surface of the head. ii The major elements differ markedly from I11 its very considerable proximo-distal those of little Casea. and to a lesser extent breadth the entepicondyle is edaphosauroid. from those of Edaplzosazrrtls and Lupeosau- The entepicondylar foramen appears to ~zu.in the tact that they are very stoutly have been unusually large. built. This is, of course, a feature associated The region of the ectepicondyle and the with the large size and great neight of supinator process is imperfect in most speci- Cotylo~7~y~1clzzrs.I12 most regards, llov,~e~er,mens. In small specimens. such as Universit). they are not only typically pelycosaurian but of Oklalloina specimens bearing the num- S~IOM~a nu~nberof distinctive edaphosauroid bers 4-0-3 and 23-38, in which the distal end features. In few instances do we find all, or is well preserved, the supinator process is most, of the major limb ele~n~ntspresent broad and close to the ectepicondyle but 18 Bulletin hfzlseum of Con1paratit;e Zoology, Vol. 1,35, KO. 1

Fig. 9. Ri5;ht humerus, viewed in dorsal ond ventral aspects in the plone of the distal end, X '1,

separated from it by the notch typical of short. with a length rather less than three- pelycosaurs. This separation, however, is fifths that of the humerus. Although the obviously a growth stage only, for in 4-0-16 effect is in all specimens accentuated by and a further University of Oklahoma speci- crushing, the bone was obviously relatively men of uncertain number the notch is closed thin dorsoventrally. as in pelycosaurs gener- and an ectepicondylar foramen present. The ally. As in the case of other limb bones of foran~enis otherwise found in pelycosaurs Cotylo~~lzynchus.the radius is broad in pro- only in Edaphosau~.us,and it is stated to be portion to its length. absent in Casea. In Casea, however, the The proximal articular surface. where gap between the tip of the supinator proc- preserved. has the appearance of an oval, ess and the ectepicondyle is small, and thin dorsoventrally: presumably it was sub- may well have been bridged in cartilage. circular in life. The dorsal (extensor) sur- It is obvious that the ectepicondylar fora- face of the shaft is convex in section: the men has developed more than once, in ventral surface apparently was flattened. A parallel fashion, in early reptiles; its pres- rugose area for ligament or muscle attach- ence here may perhaps be correlated with ment is visible on the lateral edge of the changes in limb mechanics and muscula- dorsal surface just below the head of the ture, due to increased size, rather than attlib- bone. From the head the bone (when un- uted to inheritance from a basal edapho- crushed) constricts to a somewhat thimer sauroid ancestor. shaft. A short distance below the head. Radius (Fig. 10). The radius is pre- however, a ridge arises on the medial sur- served (although not too well preserved) face of the bone. Proximally it begins some- in several instances. As noted above, it is wh:it toward the ventral surface; it ex- COTYLORHYMCHUSSKELETON Stoz;all, Price, and Romer 19

Fig. 10, left, left ulna in extensor and flexor aspects. Right, comparable views of the radius. Right, below, proximo1 and distal surfaces of radius (dorsal aspect above). X I/j. pands, however, to attain the lateral margill due to crushing; very probably the head and runs downward much of the length in life was as thick as in Edaphosazwus. As of the bone. Distally, the ridge is absorbed in the case of the radius, the bone is very in a distal expansion of the bone leading short compared with the humerus. It is. toward the terminal articulation. As in further. exceedingly broad: the breadth of pelycosaurs generally. the distal end of the the distal end in one complete specimen bone is somewhat curved ventrally, so that measures about 40 per cent of the length of the oval distal articular surface (for the the bone as measured from the lower mar- radiale) faces son~ewhatventrally as well gin of the sigmoid notch; and the proxin~al as distally. The lateral margin of the shaft width, across the notch is about 50 per cent is also rather thin. but there is no develop- of the length. These figures are far in ex- ment of a projecting ridge. Medioventrally cess of those of other groups, in which the there is an abrupt out-turning of the lateral highest figures available to me are 29 per margin above the articular surface. cent and 39 per cent for a specimen of Ulna (Fig. 10). The olecranon appears Oplziacodon. to have become well developed at a rela- Femur (Fig. 11). The femur is typically tively early stage of growth, for it is nearly edaphosaurian in nature, closely compar- conlplete (although with a small unfinished able in every major u7ayto a well-preserved terminal surface) in University of Okla- Lupeosazuus femur in the Harvard collec- homa specimen N-7-37. a small specimen. tion and likewise comparable. except for its As preserved, the head of the ulna is thin stouter build, to the iemora of Casea and u-here seen, but this is presumably an effect Edaphosaznzis. -4s in other edaphosaurs. 20 Bulleti~~.Ilzcse~iii~ of Con1pa1.atz.t.c Zoology, Vol. 13.5, KO. 1

Fig. 11. Right femur, in ventral and dorsal views; at right, proximal and distal views [dorso! aspect above). X

the curvature of the shaft characteristic of shaft, there is sometimes seen, toward the sphenacodonts is absent. Particularly char- posterior margin. a rugose area for muscle acteristic is the ventral trochanteric system. attachment. Unique is the presence at the There is a well-developed internal tro- anterior margin of a very distinct rugose chanter from wliicl~,in contrast to sphena- ridge, about 25 mm long likewise presum- codoiits and ophiacodonts, a ridge descends ably for muscle attachment: this is clearly the under side of the shaft diagonally to- seen in two specimens. The external condyle, ward the external condyle. There is little as in other edaphosauroids. projects very indication of a distinct fourth trochanter markedly beyond the internal (medial) along this ridge. and likewise little develop- one, and in a well developed specimen the ment of the posterior proximal branch of tip of this condyle markedly overhangs the the Y-shaped ridge system. tlie iiitertro- articular surface for the tibia below it. as chanteric fossa thus being shallow posteri- it does in other edaphosauroid femora in orly. In spl~enacodonts. and to a lesser which ossification is ~velladvanced. degree in some opliiacodonts, the proximal Tibia (Fig. 19)). The tibia is, as noted articular surface extends along tlie proximal earlier, relatively short. with a le~igtl~only portion of the posterior margin of the shaft; three-fifths or less that of tlie femur. As here. as in other edaphosaurs. this surface with other liliib bones. the tibia is very is confined to the proximal end of the bone. broad, ilotably its head. The width of the Proxilnally. on the dorsal surface of the head in one specimen is ahout 57 per cent COTYLORH~-KCHUSSKELETOS Stocull, Price, and Rome? 21

Fig. 12. Letf, extensor aspect of right iibia; center, extensor and flexor aspects of right fibula; righi, obove, proximal l aqd d~sialsurfaces of tibia, ond below, distal sarface of fibula, extensor surface above. X ~ the length of the bone. and the distal width fibula. with distal widths in Dimet~.odon nearly 33 per cent of the length. The closest and Ophiacodol~.for example, of 20 per approach to these proportions is in Edapho- cent and 29 per cent of the length. I snzrrus: in which these tn7o figures approxi- mate 30 per cent and 30 per cent. In all FEET 1 I other known pelycosaurs the figures are 3la11u.s (Figs. 13. 14). As noted above. mucll lower-Dimetrodon limbatus, for ex- the specimens of Cotylolhynchus are not I I ample. giving figures of 38 per cent and 22 infrequently found in articulated fashion, l per cent: Ophiacodon 37 per cent and 26 and the feet are sometimes well preserved. i per cent. The two articular surfaces of the T47e may note; for example. well preserved ! head are distinctly separated and set off front feet jn 4-0-1. 4-0-6 and 4-132: hind from one another at a considerable angle. feet in 4-0-10. 4-0-2 (1249): and both front I As in other opl~iacodonts;the lateral femoral and hind feet in the Chicago skeleton. Be- articular area is relatively narrow dorso- tween the \7arious specimens nearly all fea- ventrally. The cnemial crest is little de- tures of carpus, tarsus and digits are seen. I veloped. As in Edaplzosazrru,s. and in con- As in the case of the major limb bones. the i trast with most other non-edaphosaurian feet are broad and short: and thus differ ! pelycosaurs, the distal articular surface considerably at first glance from those of curves strongly to\varcl the lateral side of most pelycosaurs. Study, however: shows 1 the bone. that, apart from questions of proportions Fibula (Fig. 12). As in the case of the related to the size of the animals the feet tibia, the fibula is very short as compared are typically pelycosaurian and, despite the with the femur, and is very broad distally. contrast in shape, resemble closely those The inean breadth here in three specimens of Cusea. is 38 per cent of the length. This figure is The manus has the usual pelycosaur ele- comparable in Ednplzo,~nul.us, but pely- ments, including a pisifonne. two centralia cosaurs generally have a ~nuchsliminer and a series of five distal carpals. The 22 Brilletin .?fuseurn of Cornparatice Zoology, VO/. 135, No. 1

Fig. 13. Lefl, left monus of No. 4-0-6; rlght, restored left rnonus. X '1.:. Abbreviations for Figs. 13-16: a, astragaus; c,

centraio; col, colconeurn; F, fibula; i, interrnedjum; p, pisiforme; R, rodiur; r, radiale; T, tib~a;U, ulna; u, ulnore; 1-5, distal carpals or tarsals; I-V, digits.

radiale has a very deep, essentially square, face. short proximodistally. broad mediola- proximal articular surface for the foot of terally. The articulated feet suggest that it the radius. The intermedurn is short and was placed well in toward the center of the broad. with well developed processes on manus, ~ithan unossified gap between both radial and ulnar sides at mid-height, radiale and distal carpal 1. and a broad proximal articular surface for Distal carpal 1 appears to be essentially the ulna. The ulnare is far shorter than in a simple rectangle in dorsal outline, short most pelycosaurs. but comparable in pro- proximodistally but broad mediolaterally, portions to that of Caseu. The proximal end covering the entire width of the head of is much less convex in outline than in most metacarpal I. Element 2 is longer but less pelycosaurs: it fonns a nearly continuous broad. its width less than the overall width articular surface for the very broad ulna: of its metacarpal: its lateral border is this surface extends medially from a con- straight. its proximal and medial borders tact with the intermedium to a lateral facet a continuous curve. Element 3 is about

for the pisiforme. This last element. as as broad as 2. but longer- proximodistally.- usual. is a thin plate. I11 4-0-6 the bone is As in pelycosaurs generally. 4 is by far the curved sharply toward the ventral surface largest of the distal series. with a width at its outer margin; this may, however. be doible that of element 3 and a somewhat an effect of crushing. The medial or proxi- greater length proximodistally. Its proxi- mal centrale is again relatively short. The mal end has. as in pelycosaurs generally. usual arterial gap is present between proxi- two articular faces at somewhat of an angle- mal centrale, intermedium and ulnare. The to each other. a laterally tilted surface for lateral centrale is not too well preserved; it apposition to the ulnare. a shorter nledially is. as i11 pelycosaurs genel.ally, a small ele- tilted face for the proximal centrale. Distally ment when viewed from the exterior sur- the hone articulatrs broadly with the ex- COTYLORHYNCHUSSKELETON . Sto~;all, Price, and Romer 23

Fig. 14. Lett, left rnonus of No. 4-1.52: right, left pes of No. 4-0-10. X y3 Abbreviations as in Fig. 13. panded head of metacarpal I\'. Element 3 margin is much more proximal than the is narro\i7 proximodistally but is expanded medial. mediolaterally to meet the entire breadth of The phalanges. like the metapodials. are inetacarpal \'. Proximomedial and proxi- short and massive in appearance. The na- molateral surfaces meet element 4 and the ture and structure of the articulatiolls of the ulnare, respectively. elements is in general of a typically pely- In the metapodials (and ill the phalanges) cosaurian type (Homer and Price, 1940: the shortness and breadth of the elements 16'7-169). From the massive build of the and the almost complete absence of a dis- feet and the presumed herbivorous habits tinct shaft region give the foot a clumsy ap- of the animal. one would expect the toes to pearance. There is a steady increase in ternlinate, like those of diadectids or parei- length from metacarpal I to metacarpal IV. asaurs. in hoof-like structures. Instead. and metacarpal V is. exceptionally. some- however, there are (as in Casea) long and what longer still, Metacarpal I appears to powerful bony claw supports. suggesting have a broad. flat head: in metacarpals I1 that in life the animal did considerable dig- and 111 the proximal articular surface is a ging for its food supply. Pelycosaurs. in which good articulated concave area not occupying the full width feet are known. typically have a phalangeal of the bone. In metacarpal I\' the proximal formula in the manus of 2-3-4-5-3.' Coty- articulation is a concavity, but a very broad 1o~hynchzl.s.as is definitely proven by the one. In metacarpals I1 and I11 there is a material, has the surprisingly low formula pronounced expansion of the head totyard of 2-2-3-3-2-even lolver than in typical the lateral side. and in 1V this lateral ex- therapsids and rivalled for reduction among tension is very pronounced. In inetapodial \7 the proxiinal articulation does not appear Edal~llo.r.uurtrs,formerly in doubt, is no\v kno~vn to be cupped; it is tilted so that the outer to have this forn~ulaalso. 24 B~illetinA4ziselcn.i of Compai.aticc Zoology, Vol. 135, No. 1

Fig. 15. Dorsal ond ventrol views of r:ght per of CNHM 272; the clawed toes are strongly f!exed and bent under the torsus. X I/:. Abbreviations as in Fig. 13.

Paleozoic reptiles only by the pareiasaurs. attempted to restore the foot of Casea with In T17illiston's material of Casea. the manus the primitive formula. But. as his descrip- was nearly complete but for the most part tion suggests. little of the material was actu- disarticulated. Tl7illiston. not expecting re- ally articulated and it seems reasonable to duction, utilized the material available to believe that Casea had a reduced pha- give a manus with the typical reptilian for- langeal fonnula in the pes similar to that in mula (1911: fig. 13). To do this he was forced the malus: if so, Tll1illiston's foot material to assume that a number of elements were was nearly con~plete. missing from the toes. However. new ma- terial described by Olson (1954) shows that ABDOMINAL RIBS the fonnula of the manus was 2-3-3-4-3-a In agreement. it would seem: with the definite reduction. although not as marked fact that in the Edaphosauria generally the as in its giant relative. The series of ele- gastralia are Little developed, no trace of ments present in Williston's specimen was, abdoininal ribs m7as discovered during prep- thus: actually nearly complete. aration of the materials except in one in- Pe.9 (Figs. 14, 15: 16). Although the stance. Here there were found numerous material of the hind foot is not as good as slender elements with tapering ends. The that of the manus, nearly all the structure lnaxilnum length as preserved was 65 mm: can be clearly made out. There is a series the widths 3 to 5 mm. The material was of typical tarsal elements-astragalus and not sufficient to determine their arrange- calcaneum proximally, and five distal tar- ment, although they were presumably ar- sals; there is a lateral centrale, but whether rayed in the usual series of V-shaped seg- a small medial ceiltrale was present is un- ments along the course of the abdomen. certain. The proximal tarsal elements are relatively short. as compared with those RESTORATION of most other pelycosaurs except for the A lateral view of a restoration is shown ophiacodonts; this presumably in relation to in Figure 17, based on the series of larger ponderous build. The astragalus appears specimens. The general appearance is com- to have had a relatively flat facet for the parable to that shown in restorations of its tibia. There is a typical arterial notch be- smaller relative. Casea (Williston, 1911: tween artragalus and calcaneum. frontispiece: Roiner and Price, 1940: fig. 4s in the manus. the reduced phalangeal 71 ). except for the.son~ewl~atmore massive formula of 2-2-3-3-2 was present. T.l7illiston build of Cotljlo~l~ynchusassociated 1t7ith its COTYLORHYNCHUSSKELETON Stouall, Price, and Rome~ 25

Fig. 16. Restored left pes, X 'I-. Abbreviations as in Fig. 13. larger size, and the absurdly small size of the head. The lateral view does not, of course, give proper emphasis to the great breadth of the barrel-like trunk. commented on previously, and well shown in Wil- liston's photograph of the mounted Casea. The build of Cotylo~hynclzusis in agree- ment with the portrait of a generalized edaphosaurian given by Romer and Price (1940: 377) : "We find a tiny head armed with a powerful battery of blunt teeth, the trunk a large, broadly rounded barrel, the legs spread out broadly, but tl~elower segments so short that the belly cannot have been far clear of the ground. Such an ani- mal was obviously not a carnivore, and, in the discussion of habits, we have already cited data supporting Williston's belief that these reptiles were herbivores. The enormous storage capacity of the abdomen further suggests that the food was probably of a bulky, watery nature, low in nutri- tive value, so that it was necessary for the animal to ingest large quantities. The curiously small head of the advanced edaphosaurs is matched among later plant-eating reptiles by the sauropods, whose food may have been of a comparable type." Cotylo~~hy~zchzis~omeri exceeds in bulk any of the known pelycosaurs froin the typical Texas Wichita and Clear Fork red- beds deposits, and is exceeded only by its presumed descendant, C. hancocki from the 26 Bulletin Museum of Conlpa~atiueZoology,

San Angelo (Olson and Beerbower. 1953; The caseids are the last major group of Olson. 1962: 28-45), I11 the study of pely- pelycosaurs to appear in the geological cosaurs generally. Romer (Romer and Price. record. Casea b~qoilii,the earliest acknowl- 1940; Romer. 1948) used as an index to edged member of the family. only appears relative size-and weight-an "orthon~etric at about the Arroyo-Vale boundary in the linear unit" based on the dimensions of Clear Fork group, other fornls occur in the dorsal vertebrae. At the time of publication later Clear Fork formations and the roughly of the "Review- of the Pelycosauria" such equivalent Hennessey of Oklahoma. or the information as was then available concenl- still later Pease River group of Texas and ing C. romeri suggested that this unit the Russian Kazanian. Although some mil- was approximately 8.32. and this appears lions of years must be allowed for the de- still to be a reasonable figure. Closest to velopment of caseid specializations, it is I Cotylorhynchus ronleri among Wichita and quite possible for the family to have evolved I Clear Fork pelycosaurs were the large ter- from some more generalized group during I minal member of the Ophiacodon phylum, Wichita and early Clear Fork times. Until

1 0. major. with an orthometric linear unit of recently clues as to caseid ancestry were 7.37. and the large tenninal Clear Fork few. Trichasaurus of the Arroyo Fomla- Dimetrodon grandis, at 7.61. Since the tion and Glaucosaurus of the Clyde Forma- Casea unit is but 3.30. the average linear tion of Texas were suggested by Romer and measurements of elements of Cotylorhyn- Price (1940: 421423) as presumed eda- chzis should be approximately two and one- phosauroids possibly related to the caseids. half times that of Casea. and the weight In the first-named genus the skull is un- more than 15 times as great. With an esti- known.l The second is represented only mated weight of about 331 kg-roughly by a single small skull which is extremely about one-third of a ton-Cotylorlzynchus short-faced and with an isodoilt dentition; was the giant of its times. this suggests possibilities of relationship to the caseid pedigree. As to a more remote RELATIONSHIPS ancestry, it was suggested by Romer When Casea was first described it oc- (Romer. 1937: Romer and Price, 1940: 405- cupied an isolated position among pely- 412) that hlycterosaurus and h'itosaurus. cosaurs-so isolated, indeed, that Watson small early Permian pelycosaurs which (1917: 173) suggested that it was not a seemed to be primitive in many ways but pelycosaur at all. Possibly Trichasaurus (cf. showed definite edaphosaurian characters Romer and Price, 1940: 422-423) is a rela- in the postcranial skeleton. might represent tive, but until the discovery of Cotylo- the ancestral stock of the caseids and per- rlzynchzrs no further members of the family haps of the edaphosaurians as a whole. Caseidae were recognized. In recent years: In recent years several new finds have however. Olson (1962: 24-47; and earlier added somewhat to the picture. Vaughn papers) has added a number of new fomls (19%) has described as Cobbomycter an to the group from the middle and upper imperfect skull from the Fort Sill quarry Clear Fork formations and the lower part (an Arroyo equivalent in Oklahoma). and of the Pease River group, including further considers. reasonably. that its characters in- species of Casea and Cotylorhynchus, and dicate that "it provides a good structural if the new genera Caseoides. Caseol~sis,and not an actual ancestor for the family Casei- Angelosaztrus: further, the caseids are llour dae." Fox (1962) has described as Delo- known to have ranged widely, for Ennato- saurus of the Russian Kazanian is clearly a A toothplate plovisionally referred to this genus caseid (as is possibly the poorly known (Ro~nerand Price 1940: 423) 1s now known to per- Pl~reatophasma). tain to the cotylosaur Labidosaurikos. COTYLORIIYNCHUSSKELETOS . Stouall, Plice, and Roi~zer 27

rhynchtis three isolated maxillae froin this sntirzis had more teeth than Oedaleops. In same quarry; the bone itself is caseid-like. the latter genus the dentary is not known but the dentition is primitive and hence from associated material. I11 the two max- Fox classes it amongst the presumably an- illae of Ocdaleops found by Langston, the cestral Nitosauridae rather than placing it tooth counts appear to be 18 and 16; in the in the Caseidae. As noted by Langston incomplete Nitosaurus maxilla 13 teeth and (1963) there is little to distinguish Delo- alveoli are present. and the total count was r11ynchus from its quarry-mate Colobomyc- probably about 18. (4) There appears to tel.. Langston (1965) has recently described be no significant difference in the height of as Ocdaleops a small pelycosaur from the the maxilla between Oedaleops and Nito- New Llexican Penmian represented by two saurus-particularly if the obvious crushing skulls and other fragmentary material. As undergone by the Oedaleops skull be taken Langston points out, the Oedaleops skull illto account. is of a type quite surely expected in a caseid The one possibly valid generic distinction ancestor, but differs from proper members lies in the greater isodonty of the maxillary of that family in that, for example, the face dentition in Oedaleops. In the type there is is not as abbreviated and: most especially, a modest development of a "canine" pair the dentition is primitive; as proper for a at maxillary positions 2 and 3; in a second primitive pelycosaur of any sort. the teeth specimen the third tooth is large; in the are sharp-pointed and somewhat recurved Nitosnzirus maxilla a maximum is not gained and with a modest development of a maxi- until we reach teeth 4 and 5. Considering mum tooth size in the canine region, in one the constant tooth replacement characteristic specimen: at least. As possibly attributable of reptiles and the consequent continual to Oedaleops: Langston describes a number changes in the aspect of a dentition, this one of small postcranial elements found in the feature seems hardly safe ground for generic same quarry. Of these, the ilium, as Langs- distincti0n.l ton notes: is of a very primitive type, cer- Although the material of the genera dis- tainly not expected in a pre-caseid. Other cussed above is quite incomplete, in most elements, most notably the scapulocoracoid instances, it appears that in this we have (lacking, significantly, the supraglenoid at least the beginnings of a phyletic series foramen). are comparable to those of Nito- leading from such a primitive but edapho- saurus and caseids. They may well pertain sauroid pelycosaur as Mycterosaurus up- to hTitosaurzu. ward toward the caseid condition. A com- Moreover, may not Oedaleops and Nito- plicating and confusing factor in the sit- saurus. contemporaries from the same region uation, however, was introduced by Watson and horizon: be identical? Of Oedaleops (1954: 356) with his suggestion that Eotlyris we have no certain knowledge of postcranial might be related to caseid ancestry. Both material; of Nitosaurus we have no skull Vaughn and Langston have adopted this material except of maxilla. Langston briefly point of view, placing such forms as Col- mentions this possibility but says that h7ito- obomycter and Oedaleops in the Eothy- snurzis "had longer jaws, a more slender rididae, and Langston goes to the extreme dentary, considerably more teeth (sub- of excluding the Nitosauridae from any re- isodont in form) and a higher maxillary lationship to the Caseidae: despite the num- bone." But (1) wre do not have a complete erous and surely significant postcranial re- jaw in Nitosaurzis; (2) the seeming slender- semblances between the hvo. ness of the h7itosaurus dentary is probably The basic reason for considering Eothyris due to loss of the thin lower margin (cf. Romer and Price: fig. 70, and Langston: A subordinal distinction, actually, in Langston's fig. 2a : ( 3) there is no evidence that Nito- chart, page 43. 28 Bulletin J4ziseum of Comparatice Zoology, Vol. 135, No. 1

as a possible relative of the caseids is, of either case. With regard to "enlarged nares course, the fact that Eotlzyris, like the and obtuse rostrum:" the nares in Eothyris caseids, is remarkably short-faced. This in do not appear to be any larger, propor- itself is no more a valid reason for associat- tionately, than in many other pelycosaurs: ing them than would be the association of and the rostrum does not appear to be any the sphenacodont Secodontosazrrus with the more obtuse than in pelycosaurs generally. ophiacodont T'aranosaurus because they The Eothyris prefrontals are said to be "in- are both extremely long-snouted. If the flated (i.e., sonleu-hat expanded). but this Eothyris-caseid relationship is to be sub- does not seem significant. stantiated, more positive reasons must be There is thus little positive reason to developed. associate Eothyris with caseid ancestry, and Langston (1965: 21) cites fourteen points one very strong objection-the dentition. in which Oedaleops and Eothyris are in In all early pelycosaurs-indeed, in nearly partial or complete agreement. As he says, all primitive tetrapods generally-there is a this seems to be, at first sight, an imposing trend for the development of somewhat list. Included, of course, is the fact that the enlarged teeth near the front of the maxilla face is short, not necessarily meaningful, as an incipient "canine" region. In sphena- and the fact: of no systematic value, that codonts and their therapsid descendants both are small. Correlated with small size. this trend is accentuated; in edaphosaurians and hence likewise without other necessary -both Edaphosaurus and the caseids- significance, is the relatively large size of there is an opposite trend toward isodonty. the orbits and of the pineal foramen. A In such forms as Oedaleops there is but a number of other common features are such mild, essentially primitive: development of as are liable to be present in any relatively a canine "maximum" such as might be ex- primitive pelycosaur, including: (1) rela- pected in any relatively primitive pely- tively flat skull; (2) jaw articulation on a cosaurs. In Eotl~yris,on the other hand, we level with the toothrow, as in the Ophia- find the greatest exaggeration of canine tusks codontia, the primitive sphenacodont Var- to be found in any pelycosaur. In their anops, and Alycterosazirus; (3) outward dentition, E0thyri.s and the caseids have slope of cheek plates (contrasting with evolved in such diametrically opposite spheilacodonts): (4) a primitive long lacri- directions that it is difficult to believe that mal; (5) normal relationship of roofing they are at all closely related. bones; (6) an unusually large supratemporal; But even if (as is not too probably the (7) a long, tapering postorbital (as: for ex- case) Eothyris should prove to be related to ample, in the ophiacodont T'a~anosaurus, the caseids: it is dangerous to base hypoth- and the sphenacodont Varaizops ) ; (8) eses of broader relationships on the "family some indications in the tabular-supratem- Eothyrididae," as has been done by Langs- poral region of the otic notch of ancestral ton and, to a lesser degree: by Vaughn. types: (9) a differentiated dentition. as in As I have pointed out (Romer and Price, most pelycosaurs except Edaplzosaurus and 1940: 247; Romer, 1936: 676); this family caseids. is a purely provisional one, set up to receive The two genera, thus, are short-faced, forms, res sum ably highly predaceous; which small in size: and have both retained various have marked canine development but are primitive characters. Little remains of the not members of the Sphenacodontia. Any original fourteen points which can be con- unity the group might have is based on strued as positive indication of relationship. this dental feature-which is, of course, Langston cites "relative position of orbits the one point in which all of them notably and pineal opening." but there does not differ from the trend toward isodonty ex- appear to be any uilusual condition here in pected in caseid ancestors. There is 110 COTYLORHI-~-CHUSSKELETOY Sto~all, Price, and Rome7 29

ekidence that any foiin assigned to this or anotller. of this suggestion. There Lvas a fainilv. other than Eothtiris.., . was short-faced. lateral temporal opening, but the cheek is Such postcranial material as can be as- poorly preserved. and \vl~etl~eran addi- sociated n.it11 any of the genera included tional upper opening \vas present canilot be in this provisional fanlily lacks any features determined. .Apart froill the possible but indicative of caseid I-elationships, and such unproven diapsid nature of the temporal indications of svstematic nosition as are region. there is 110 reason to assign Petrolu- cosazrlvs to the Eosuchia. Peabodv voints , A and ~al&cinonzr.ssuggest the 0~;11iacodon: out a number of colnlnon features of the tia. LVit11 regard to Eotlzyris. the lack of skulls in Petrolacosazrrt~s and Prolacerta ~ostci-anialdata is a stumbling block. but these are essentially prinlitive characters On the assumption that the Opl~iacodontia \vhich could have been inherited by both re~resentthe basal stock of the Pelyco- froin captorhinomorpl~ cotylosaurian an- sauria. it is a reasonable assumution that cestors. The time of avvearance of Petrolu- A A the caseids are of ultimate ophiacodontian cosaurz~sis one at which an eosuchian is derivation but that any of the "eothyrids" hardly to be expected. There are no traces are connecting links is very dubious. of any diapsid in the Lower Permian. and Roiner and Price (1940: 366-376) pointed no sure evidence in the Middle Permian; the out ilunlerous skeletal features whicll Eda- first certain diapsid is Upper Permiail in ~~hosaurusand Casea. as is170 extremes, have age-a full period after Petrolacosaurus- in common. and advocated their being- and M'atson ( 1957) has argued that diapsids placed in a cornilloil suborder. There is, were only then evolving from millerettid however. a seeining difficulty in assuming a cotylosaur derivatives. common ancestry. because of chronological If. then. we abandon the ~ossiblebut factors. The caseids seem quite surely to improbable suggestion that Petrolucosau~us have specialized from primitive ancestors was a precocious diapsid. all the features at a late date. for no foinl attributable to of this little reptile agree with the assump- this stock is knov~ilearlier than fairly early tion that we are dealing with a pelycosaur. Wichita times; on the other hand. Edapho- and several characters point strongly to the snzrrtrs had already evolved by the late suggestion that it is a primitive edapho- Penns\rlvanian, If both Eda~~1zo.saui~u.sand saurian. The skull is primitive and gen- the caseids evolved from essentially primi- eralized. as it is in such archaic opl-~iacodonts tive edaphosaurians. such as the nitosaurs. as Clepsydrops and Varanosamrzis, such a this must have been. as Langston (1965: 58) sphenacodontian as Varanops, and such a notes. a very bradytelic group and nitosaurs nitosaurian as Alz~cterosaurtls. Diagnostic- should have been in existence in the Penn- features. however. can be found in the sylvanian. postcranial skeleton. (1) The postcervical This appears to have been the case. It vertebral centra are rounded ventrally as seeins highly probable that Pefrolucosaurus. in ophiacodollts and edaphosaurs, in con- from the Pennsylvanian Garnett shales of trast to the keeled sphenacodonts. (2) In Kansas (Lane. 1945: Peabody. 1949. 1952). the carpus the ulnare is short. in contrast is a primitive edaphosauroid. to advanced sphenacodonts and as in ophia- Lane. and Peabody at first. concluded codonts and edal~llosaurs (however. Var- that this small reptile was a pelycosaur. The anops ainong the sphenacodonts also has a latter author. llo\vever, ilnpressed by cranial short ulnare). (3) 111 the tarsus there is a resemblances to Prolacei~tct ( with which he broad, highly developed. medial centrale was familiar). argued that Pet~~olacosaurus forming the sole coni~ectionbetween astrag- \I as an eosuchian-a primitive diapsid. alus and distal tarsals 1 and 2; edaphosaurs There is no morpllological proof, one way and sl~l~enacodontsare similar. but opllia- 30 Bulletin Aluseum of Coinparatifie Zoology, Vol. 1.35, KO. 1

codollts have two small centralia of sub- from the Lo\ver Permian of New Slerico, and the family Eothyrididae. Bull. Texas Mem. equal size in this position. (4) 011 the hlus., 9: 1-46. femur. the ventral system of trochanters OLSOK, E. C. 1954. Fauna of the Vale and and ridges is one found among pelycosaurs Choza. 7. Pelycosauria: family Caseidae. in edaphosaurs only (cf. Ronler and Price. Fieldiana, Geology, 10: 193-2004, 1940: fig. 37). Distal to the internal tro- . 1962. Late Permian terrestrial verte- chanter an adductor crest slants diagoilally brates, U.S.rl. and U.S.S.R. Trans. Alner. Philos. Soc., (N.S.) 52 (part 2): 1-196. across the bone towards the extenlal tro- OLSON,E. C. AND J. R. BEERBOWER,JR. 1933. chanter. 111 sphenacodonts there is a pro- The San Angelo Formation, Permian of Texas, nounced fourth trochanter but no adductor and its vertebrates. Jour. Geol., 61: 381-423. crest; in ophacodonts the crest descends PEABODY,F. E. 1949. .\lid-Pennsylvanian pely- the external margin of the bone. (5) Both cosaurs from Kansas. Bull. Geol. Soc. Amer., 60: 1913. Lane and Peabody ascribe to Petrolaco- . 1952. Pett.olacosaurus kansensis Lane, a sauyus a pelvis (University of Kansas 110. Perinsylvanian reptile from Kansas. Univ. 1425) which is very distinctively edapho- Kansas, Paleont. Contrib. Vertebrata, 1: 1-41. saurian. It is of a type strongly contrasting Rohi~n,A. S. 1937. New genera and species of pelycosaurian reptiles. Proc. New England with that of any other reptilian group- Zool. Club, 16: 89-96. most notab1)- in the shovel-shaped iliac , 1948. Relative growth in pelycosaurian blade. tall but ~rithouta marked posterior reptiles. Robert Broom Commen~orativeVol- projection. Were this pelvis definitely as- ume. Pp. 45-55. sociated, Petrolacosazl~z~scould be assigned . 1956. 0steolog)- of the reptiles. Chi- cago, Univ. Chicago Press, mi + 772 pp. to the Edaphosauria without hesitation. It ROMER,A. S. AID L. I. PRICE. 1940. Review of ~7as.however. found isolated. and since a the Pelycosauria. Geol. Soc. her., Spec. Pap. specimen of Edapho.sau~ushas been found No. 28 : 1-538. in the quarry. this pelvis may pertain to SRULER,E. W. .~XDR. V. WITTER. 1942. The mounted skeleton of Edaphosaurzis boanerges that genus. Several pelves with low, long Roilier at Southern Llethodist University. ilia of ophiacodont type are also present in Field and Laboratory, 10 (2): 140-144. the material: but these. too. lack association. STOVALL,J. W. 1937. Cotylorhynchus romeri, and may belong to an ophiacodont. also a new genus and species of pelycosaurian present in the Garnett material. reptile from Oklahon~a.Arner. Jour. Sci., ( 5) 34: 308-313. There are. thus, in the postcranial skele- VAUGHK,P. P. 1958. On a new pelycosaur froill ton of Pet~olacosuurzrsa number of features the Lower Permian of Oklahoma, and on the which strongly indicate that this genus be- origin of the family Caseidae. Jour. Paleont., longs to a group of archaic edaphosaurians 32: 981-991. from \vhich both and, at a WATSOS, D. hl. S. 1913. The Beaufort Beds of Edaplzosazi~us the Karroo System of South Africa. Geol. much later time. the caseids may have Mag., (5) 10: 388-393. arisen. , 1917. A sketch chssificatior~ of the Preparation of inaterial for this paper pre-Jurassic tetrapod vertebrates. Proc. Zool. was aided by a grant (No. GB 500) from Soc. London, 1917: 167-186. the National Science Foundation. . 1954. On Bolosaurt~s and the origin and classification of reptiles. Bull. hlus. Comp. Zool., 111 : 295449. LITERATURE CITED . 1957. On A.lillerosaurus and the early Fox, R. C. 1962. Two new pelycosaurs from the history of tlie sauropsid reptiles. Philos. Trans. Loxver Permian of Oklahoma. Univ. Kansas Roy. Soc. London, (B) 24.0: 325400. Publ. hlus. Nat. Hist., 12: 297-307. \VLLLISTOS,S. \f'. 1910. New Permian reptiles: LANE,H. H. 1935. New mid-Pennsylvanian rep- ~~l~acl~itorno~~svertebrae. Jour. Geol., 18: tiles from Kansas. Trans. Kansas Acad. Sci., 5855600. 47: 381-396. . 1911. .qmerican Permian vertebrates. LAKGSTOK,\T'. 1965. Oeclnleo~~scampi (Rep- Chicago, Univ. Chicago Press, 145 pp. tilia: Pelycosauria ). A new genus and species ( Receiced 27 Septen~ber1965.)