PAR-E-IASAURS VER-SUS PLACO- DONTS AS NEAR Ancevstors :TO the TURTLES

PAR-E-IASAURS VER-SUS PLACO- DONTS AS NEAR ANCEvSTORS :TO THE TURTLES

WILL:IAM :KING- GREGORY

, .e... . .r ....7... --'''' '- "BULLETIN' -- ;OF THE -- AMERIC:AN MUSEUM OF NATURAL HISTORY fVOLUM-E---* 8:6:..ART-IC:LE...6v:: -0--NEW YORE: 1946; m PAREIASAURS VERSUS PLACODONTS AS NEAR ANCESTORS TO THE TURTLES

PAREIASAURS VERSUS PLACODONTS AS NEAR ANCESTORS TO THE TURTLES

WILLIAM KING GREGORY Curator Emeritus of Fishes and of Comparative Anatomy

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY VOLUME 86: ARTICLE 6 NEW YORK: 1946 BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Volume 86, article 6, pages 275-326, text figures 1-33, plates 34-35, tables 1-3

Issued May 8, 1946 CONTENTS

INTRODUCTION: THE APPROACH TO THE PROBLEM ...... 281 COMPARISON OF DIADECTID, PAREIASAUR, AND TURTLE SKELETONS ...... 285 General Body Form and Habits ...... 285

Skull ...... 285 Vertebrae and Ribs ...... 296 Pectoral Girdle and Limbs ...... 300 Pelvic Girdle and Limbs...... 307

COMPARISON OF CAPTORHINOMORPH AND TURTLE SKELETONS ...... 311

COMPARISON OF SEYMOURIAMORPH AND TURTLE SKELETONS ...... 313

COMPARISON OF PLACODONT AND TURTLE SKELETONS ...... 314

DISCUSSION: CONVERGENT EVOLUTION AMONG TESTUDINATE ANIMALS ...... 319

CONCLUSIONS ...... 321

REFERENCES ...... 323

279

INTRODUCTION: THE APPROACH TO THE PROBLEM THAT THE TURTLES, especially the sea turtles, called Otocoelus and Conodectes, in which the were more or less closely related to the region of the auricular meatus (otic notch) plesiosaurs has been the view of certain "resembled those of the Loricata and Testu- authors, including Baur (1887b, pp. 97-99), dinata." On top of the neural spines were but, according to Williston (1907, p. 487), placed "transverse osseous arches," which "It is chiefly because of the external resem- even overlapped the scapular arch (as does blances of form and similarity in mode of the carapace of turtles), while the temporal locomotion in the water that it has been region of the skull was "complex and with no generally and indefinitely assumed from longitudinal bars." Cope (1898) therefore Buckland's time to the present that the proposed a new order, the Chelydosauria, to plesiosaurs were related to the turtles." In include this family (Otocoelidae) and sug-

Diadectes FIG. 1. Skeleton of Diadectes phaseolinus Cope. X About 1/12. Data from American Museum material and from Case (1905, 1911). Length about 1.7 m. opposition to this view, Williston (loc. cit.) gested that "the order is probably ancestral pointed out many supposedly inadaptive to the Testudinata and the Pseudosuchia." differences between the plesiosaurs and the Eventually it was shown that the Otocoelidae turtles and concluded that "Whatever of (Dissorophidae) were rhachitomous laby- resemblance there may be in the form and rinthodonts, and Cope subsequently aban- habits of these two orders of animals has doned the name Chelydosauria, retaining the been due solely to parallel evolution, to similar name Cotylosauria for Diadectes and its aquatic adaptations. In their internal struc- relatives. ture they are really remote from each other Case (1905), restricting the "Chelydo- and neither could have been derived from the sauria" to the Diadectidae and Otocoelidae, other type, not even in a remotely ante- pointed out seven important points in which cedent stage." the diadectid skull indicates affinities to the A new phase in the development of this Testudinata and concluded: "In just the already old problem began with Cope's characters in which the Chelydosauria (Oto- discoveries in the Texas Permian (or Permo- coelidae, Diadectidae) approach the turtles Carboniferous) of the very primitive reptiles they are distinct from the Cotylosauria (fig. 1) which he named Diadectes, Empe- (Pareiasauridae and Pariotichidae) and so it docles, and Empedias and for which he erected seems very probable that we have in the (1880) the suborder Cotylosauria, of the Diadectidae forms very closely related to the "order Theromorpha." Case (1905, p. 127) ancestral stem of the turtles, which tell us states that in 1896 Cope described two other much regarding the development of the genera from the Permian of Texas, which he Testudinata directly from the Cotylosauria" 281 282 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 [sic ?Chelydosauria]. The conclusion of Cope general conclusion that the turtles have been and Case that the chelonians have indeed derived from the cotylosaurian order, with- been derived from the Cotylosauria (sensu out making any serious effort to find out lato) has been sustained by Williston (1904; which cotylosaurs, if any, were the nearest 1907; 1917; 1925, pp. 209, 210), Watson to the undiscovered pre-Triassic line that led (1914b), Hay (1905), von Huene (1936b), to Eunotosaurus (fig. 4) and the chelonians. and others. Nor did a cursory first review of the evidence The older view that the turtles are some- reveal anything further. The Procolophonia, how related to the Sauropterygia has been which have been ably reviewed by Colbert defended by Jaekel (1902, pp. 135-144) (1946), were a distinct suborder of cotylo- and independently by Broom (1924). Both saurs, running their course to their ex- these authors cite the turtle-like skull tinction in the Upper Triassic, retaining the characters of the Upper Triassic placodont great otic notch of Diadectes but remaining Placochelys (see p. 314), and Jaekel also quite different from the contemporary am- attempted to homologize certain parts of the phichelydian turtles. The captorhinomorphs, carapace and plastron of Placochelys with another suborder of cotylosaurs, had lost the those of chelonians. Jaekel (ibid., p. 143) otic notch and the curved quadrate which explained the contemporaneity of the heavily were characteristic of Diadectes (fig. 6A) and toothed placodonts and the virtually tooth- which would - be expected in an ancestral less beaked chelonians in Upper Triassic chelonian, nor did either the rest of the skull times by the assumption that the ancestors or the postcranial skeleton indicate a very of the chelonians had undergone a severe close connection with the turtles. For that transformation together with a change in matter, Diadectes itself, although quite turtle- food habits. The traces of some such trans- like in its large otic notch and curved quad- formation were indeed evident in the much rate, was well provided with highly special- later discovered skulls of Henodus chelyops ized jaws and dentition; its very large from the Lower Jurassic of Europe, described vertebrae (fig. 16) bore complex hyposphene- by von Huene (1936a, 1938). In this genus hypantral articulations, while a bony cara- only the last upper and lower crushing teeth pace and plastron were either lacking or not of the placodonts remained. The oral borders sufficiently coherent to be preserved. were evidently covered with a horny beak When I turned my attention to the which, on the evidence cited by von Huene, pareiasaurs (figs. 2, 3), I was surprised to may have borne parallel ridges (fig. 31) for find that apart from their gigantic size they sifting food from the water, as in cer- seemed on the whole to afford an excellent tain ducks. The carapace and plastron of starting point for the chelonian line,, both in Henodus in general form were remarkably their general construction and in many like those of certain turtles. Nevertheless features of the skull, vertebrae, ribs, girdles, von Huene showed that Henodus is only a limb bones, hands, and feet. highly specialized placodont and the last of Seeley (1887-1895) made many compari- its order. The earlier placodonts, including sons between pareiasaurs and anomodonts, Paraplacodus (Peyer, 1935), are far less pareiasaurs and protorosaurs, pareiasaurs turtle-like and, on the other side, I shall and mammals, and the like, but, strange to presently cite decisive evidence that in the say, neither his text nor his figures compare girdles and limbs as well as in the construc- pareiasaurs and chelonians. Haughton and tion of the temporal region the older placo- Boonstra's thorough systematic and com- donts were related to the nothosaurs and parative studies on South African pareiasaurs plesiosaurs. Von Huene, therefore, even in (1929, 1930, 1932, 1933) have supplied in- his latest phylogenetic diagrams (1936b, dispensable material for the present study, 1940) continues to place the Testudinata as but they do not touch upon the major prob- a derivative of the Cotylosauria, its line lem of the possible relationships of pareiasaur starting between the pareiasaurs and the to chelonian. seymouriamorphs. The present paper is in part a continuation Until recently I was content to accept the of the series on comparative osteology and 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 283

z(ore

Scutosaurus FIG. 2. Skeleton of Scutosaurus karpinskyi (Amalitsky). Based on American Museum skeleton mounted by Mr. Charles Lang under the direction of Dr. Barnum Brown. Length about 3.05 m.

Pareiasaurus karpinnskyi FIG. 3. Skeleton of "Pareiasaurus karPinskyi." Based largely on unpublished lithographic plates of Amalitsky's, with changes indicated by comparative study. The line along the middle of the back indicates the position of the neural or middorsal bony scutes. The transverse rows of scutes are not indicated. 284 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 myology by the author and his collabora- transformation of organic designs: a review tors, especially L. A. Adams (1919), Charles of the origin and deployment of the earlier L. Camp (Gregory and Camp, 1918), R. W. vertebrates" (1936). Miner (1925), G. K. Noble (Gregory and The skeletons of various chelonians in the Noble, 1924), A. S. Romer (1922), H. 0. American Museum have been studied with Elftman (1929), H. C. Waterman (1929), special reference to the skull, vertebral B. Schaeffer (1941), H. C. Raven (Gregory column, girdles, and limbs, especially in their and Raven, 1941), and F. G. Evans (1939), adjustments to the presence of a fixed cara-

A Eunofosauru-s FIG. 4. Sketches of thoracic skeleton of Upper Permian chelonian, Eunotosaurus See- ley. Based on description and figures by Watson (1914b). Of the 10 dorsal ribs, no. I was small and apparently had not yet overlapped the top of the scapula; nos. III-VIII were greatly widened, no. IX was less so, no. X was very small. The narrow pelvis was behind the carapace. The latter consisted of thin median and lateral ossified plates. The centra were slender and somewhat hour-glass-shaped, with reduced neural spines, low neural arches, closely interlocking zygapophyses. The prominent rib facets were near the front end of the centra in vertebrae II-VII but near or at the middle in VIII and IX, all essentially as in later chelonians. which dealt with the evolution of various pace and plastron and in comparison with the parts of the vertebrate skeleton on the corresponding parts of cotylosaurs, placo- diverging lines of ascent from the crossop- donts, sauropterygians, and other extinct terygian fishes to the amphibians, reptiles, reptiles. Some excellent pareiasaurian mate- and mammals. rial in the American Museum, including a The accepted general principles of evolu- skeleton of the Russian Scutosaurus, mounted tion and the methods of analysis and under the direction of Dr. Barnum Brown, synthesis applied in the present investigation and casts of the limb bones of the South are discussed in earlier papers, especially African Propappus, received from Dr. Rob- "The orders of mammals" (1910, pp. 105- ert Broom, have also been studied and are 107), "Fish skulls: a study of the evolution here figured. Numerous courtesies from Dr. of natural mechanisms" (1933, pp. 443-449), Edwin H. Colbert and Mr. Charles M. "Reduplication in evolution" (1935b), "The Bogert are gladly acknowledged. COMPARISON OF DIADECTID, PAREIASAUR, AND TURTLE SKELETONS GENERAL BODY FORM AND HABITS The bony rib plates in chelonians were not or THE CONSENSUS OF RECENT AUTHORS is that developed as such in diadectids pareiasaurs the most primitive turtles living today are and probably originated within the order, in not the flattened, wide, marine leathery part as lateral extensions of the bony peri- turtles (Dermochelys, Sphargis), or even the chondral husks of the ribs, within the widen- hawksbill and loggerhead sea turtles (Chelo- ing periosteal membranes (Goette, 1899, nidae) but the terrestrial tortoises (fig. 5A), p. 415), in part through an invasion of sur- with high, well-rounded carapace, short, face scutes into the intercostal spaces (see elephant-like feet, and strong blunt claws. fig. 3). These features were already foreshadowed The marginal teeth of pareiasaurs were in at least some of the pareiasaurs, which, small, laterally compressed, cuspidate, and according to Boonstra (1933b), were also close set, unlike the transversely oval cheek provided with a partial carapace of bony teeth and nipping incisors of Diadectes scutes along the midline of the back and sides. (fig. 7) and less restricted to the cutting of The outer surface of the skull of both diadec- tough, resistant substances; while their tids and pareiasaurs was often heavily powerful jaws and strong bony palates furrowed or wrinkled by the under side of a (fig. 8A) should have enabled them to smash massive integument. In certain diadectids and devour a variety of animal and vegetable (Chilonyx) and many pareiasaurs there were foods. The change to the horny beak and conical protuberances around the rear bor- powerful cutting and crushing jaws of turtles ders of the skull, which also indicate a great may be paralleled more or less among other development of an integumental skeleton. reptiles of carnivorous to omnivorous habits Clear traces of such a condition are seen also (e.g., endothiodonts among anomodonts, on the surface of the skull in Triassochelys Psittacosaurus among dinosaurs). (fig. 6C), as figured by Jaekel. In walking, the body of pareiasaurs A ventral armature of gastralia or ab- (figs. 2, 3) could be raised farther from the dominal ribs seems to be unknown among the ground than it was in the sharply bent- diadectids and pareiasaurs, as well as among limbed Diadectes (fig. 1), and both the fore- the chelonians, and it may be conjectured arm and the shank could be more widely that, in spite of analogies with the plesiosaur extended than in that primitive cotylosaur. gastralia (Owen, 1849), the plastron of The turtles, in spite of their extreme develop- chelonians, which was well developed in the ment of carapace and plastron, exhibit even Upper Triassic Triassochelys, arose from a greater mobility of the limbs than did the rapid spreading of bone-producing power of pareiasaurs and have advanced well beyond the integument on the ventral surface of the the diadectid grade. body across the integumental "bridges" between the humeral and femoral notches. SKULL The pareiasaurs, as far as known, were The general form of the skull in Diadectes gigantic animals, at least in their adult stage, suggests that of Triassochelys and the sea and the lack of a firmly knit carapace and turtle Chelone in all views (figs. 6, 7, 8, 9), but plated plastron apparently did not endanger retains many more primitive features, such their existence, But the turtles in their as the ascending rami of the premaxillae and beginnings were small animals which could the full complement of primitive reptilian easily be turned over by aggressive pelyco- elements. Of the latter, modernized turtles saurs and theriodonts, and for them a stout have lost the supratemporals, tabulars, plastron of broad plates must have been nasals, postorbitals, and ectopterygoids. a much needed addition to their armor. The dentition of diadectids is of course 285 286 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86

A Testudo elegans

Triassochetys

C Proganochelys FIG. 5. Modern tortoise and Mesozoic amphichelydians. A, After Lydek- ker; B, B 1, after Jaekel; C, after E. Fraas. Horny shields are here desig- nated by roman, bony plates by arabic, numbers. 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 287 highly specialized but, as noted above, the closely with the basic plan of this region in reduction of crushing teeth in the later pareiasaurs (fig. 8A). placodonts (Henodus), together with other The secondary bony palate was only evidence, suggests that the possession of beginning in Triassochelys (fig. 8B), and in specialized crushing teeth in Diadectes does fact the entire under side of the skull in this not necessarily exclude it from ancestry to form suggests that, apart from its develop- secondarily beaked forms without teeth. ment of horny upper and lower beaks, it was

B Dolichopareia

A Diadectes D Che/one FIG. 6. Skulls of diadectid (A), pareiasaur (B), and chelonians (C, D). A, Mostly from A.M.N.H. No. 4839, mandible from A.M.N.H. Nos. 4684 and 4353 and Welles' figure (1941, p. 426); B, after Boonstra (1933a); C, after Jaekel; D, original. However, it would presumably be easier to nearer to certain pareiasaurs than it was to get rid of the small marginal teeth of pareia- Chelone. In Diadectes there are curved palatal saurs than of the large crushing teeth of ridges (fig. 7) on the maxillae and palatines, Diadectes. And according to Jaekel (1918, but nothing like the massive secondary pp. 172, 195) Triassochelys retained clear palate of Chelone (composed of rolled-over vestiges of small teeth in the maxillae and flanges of the maxillae and palatines that dentaries, while it still bore a median patch arch over the choanae) is known either in the and four converging rows of small teeth on pareiasaurs or in Triassochelys. The massive- the pterygoids (ibid., p. 172), together with a ness and full development of these parts in long median streak on the prevomers be- Chelone (fig. 8C) may indeed be correlated tween the large choanae (fig. 8B), all agreeing with similar qualities of its beak and man- 288 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VIOL. 86

r\ " , r.-' \ I ~ J ,,4 II \itLjAj*~ o;P%.mx

1'. '.4, it /\1; * mx

7,7/ t,

(ioo I,.tI ,

\ 'll Io / I

,//,;/ / , / aI *.-:e q .op. jreceislcctu$lus for antica-scapifis Shoc Diadectes FIG. 7. Palatal aspect of skull of Diadectes. Reconstructed from A.M.N.H. Nos. 4352, 4839, and 4370, with data from Case (1905, 1911, 1912) and von Huene (1913). 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 289 dible, as well as with the huge size of the jaw muscles, which have pushed up the high muscles. The secondary palate of most other median occipital crest (cf. L. A. Adams, fossil and recent chelonians is much less 1919, pp. 88-89, and pl. 6, figs. 1, 2). In strongly developed than that of Chelone, and Diadectes the jaw muscles were confined to in some it may be lacking entirely. the dorsally shut-in temporal chamber (figs.

C Chelone

FIG. 8. Palatal and occipital aspects of skulls of Permian pareiasaur (A, D), Triassic chelonian (B, B 1) and recent sea turtle (C, C 1). A, D, After Boonstra; B, B 1, C 1, after Jaekel.

The occipital views (fig. 8) tell the same 7, 9) both above and behind, and the rear story. They also indicate that the tripartite wall of the occiput was occupied by nuchal condition of the occipital condyle in Chelone muscles (Evans, 1939, p. 58). Thus with is a secondary feature, but that its paired regard to these features of the occiput the posttemporal fenestrae have been inherited pareiasaur (fig. 8D) more clearly fore- from those of Triassochelys and that they in shadowed the chelonian and was well ad- turn are strictly homologous with those of the vanced beyond Diadectes. pareiasaurian. According to Sushkin (1928, In all three forms (figs. 6, 8) the massive p. 264) these fossae were already developed quadrates were monimostylic, that is, they in the cotylosaur Captorhinus, but they were were firmly fixed in place, medially by their not present in Diadectes. attachments to the great pterygoids, laterally In all the turtles these posttemporal holes by the quadratojugals, and dorsally by the were invaded by the large temporal jaw squamosals, especially by the massive paroc- 290 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 cipital processes of the opisthotic bones. according to Boonstra (1933a, pl. 1), the The opisthotics in Chelone (figs. 8C, 12C) otic notch in the side view is concealed by were suturally separate from the exoccipitals, the backward growth of the squamosal and as they were also in the pareiasaurs (Boon- quadratojugal, but in the rear view (fig. stra, 1933a, pp. 17, 18, and pl. 4, fig. 1) and 8A, D) the tympanic cavity was evidently Captorhinus (Sushkin, 1928, p. 264, figs. 1, medial to the squamosal and quadrate, the 2), but the separateness of the opisthotic was chief difference being that in Chelone (fig. less clear in Diadectes. According to White 8C 1), owing to the expansion of the post- (1939, p. 345) the opisthotic was distinct temporal fenestrae under the influence of the

fr-..

as A Diadectes Ai FIG. 9. Under side (A) of the temporal roof of Diadectes, with natural cast (A 1) or obverse of the same. The narrow braincase, indicated by the large pineal canal, lies between the large chambers that con- tained the temporal muscles. in Seymouria, as it was also in Kotlassia huge jaw muscles, the outer part of the (Bystrow, 1944, p. 385). In many later tympanic cavity has, as it were, been pressed reptiles, on the contrary, the opisthotic and downward below the level of the occipital exoccipital were not separate, so that here as condyle. It would not indeed be surprising if in other ways the turtles have inherited some part of the resemblance to Diadectes in the very early reptilian features. otic notch of Chelone were merely convergent, In Chelone and other Cryptodira the because the otic notch of Triassochelys was posterior side of the quadrate and the adja- narrow and less G-shaped than that of cent border of the squamosal (fig. 6C, D) Chelone. In either case the "meatal excava- were together curved into a G-shaped figure tion" and all its bounding ridges were, so to bordering the "meatal excavation" (Cope) speak, the work of the tubo-tympanal pouch, or tympanic notch. In Diadectes (fig. 6A) this which, as in the mosasaurs, moulded the notch is essentially similar, except that the surrounding bones into a G (cf. Jaekel, 1918, tabular (which is lost in modern turtles) p. 163, fig. 43). forms the upper border (cf. von Huene, 1913, The stapes (columella auris) of Diadectes p. 334, fig. 17). In pareiasaurs (fig. 6B), is not known but to judge from the stoutness 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 291 of the surrounding bones and the large size "the stapes was thus probably a massive of the crater-like depression which served curved bone with two lateral processes and instead of a true fenestra vestibuli, it must a medial process tapering nearly to a point

FIG. 10. Oblique rear view (A) of skull of Diadectes, with attempted restoration of middle ear parts and depressor man- dibulae muscle. (In A, for temp., read tymp.)

have had a large and thick shaft, as is further at the fenestra ovalis." Probably the very suggested by the large size of its extra- large stapes which may thus be predicated columellar contact with the quadrate (figs. 7, for Diadectes and pareiasaurs retained its 10). The stapes of pareiasaurs is also un- primitive piscine function of supporting the known, but after his thorough studies of the lower part of the hyoid arch and larynx. skulls, Boonstra (1933a, p. 21) concludes that Moreover, although the rim for the tympanic 292 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 membrane was well defined in Diadectes, we quadrate, but its evident homology with may doubt whether this membrane served that of Triassochelys is another feature that so much for transmitting sound waves tends to connect the turtles with the coty- through the stapes as for facilitating the losaurian stem. adjustment of the pressure inside the tubo- Parts of the hyobranchial complex of two tympanal canal to changes in pressure of the adult pareiasaurs, as described and named air or water. According to Jaekel (1918, by Boonstra (1932b), differ conspicuously

4rhy pro pe.pi A

cbrIpoli

DORSAL ASPECT A Bradysaurus B Emys.1 sE Chelone VENTRAL ASPECr

VENTRA L AS PECTC F Che/ys D Macrochelys FIG. 11. Hyoid bones of pareiasaur (A) and recent chelonians (B-F). A, After Boonstra (1932b); B, after Wiedersheim (1907); C, D, F, from American Museum specimens; E, after Ihle, Van Kampen, et al. p. 163) the columella auris of Triassochelys (fig. 11) from those of adult Testudo, Emys, (ibid., p. 164, fig. 44, Hm) likewise still Macrochelys, especially (1) in having an served as a "hyomandibular." The columella exceedingly short basihyal; (2) in apparently of the adult Chelone has become a very thin lacking an ossified processus lingualis (= "ba- curved rod, lodged partly in a special "in- sihyal" of Parker, or hypohyal of later cisura columellae" on the rear face of the authors); and also (3) in lacking an ossified C Chelone FIG. 12. The braincase of Diadectes (A, A 1), pareiasaur (B, B 1) and modern chelonian (C). A, Skull of Diadectes (A.M.N.H. No. 4839), with part of temporal roof cut away, showing large epipterygoid in contact with cranial roof. A 1, Paramesial section and right half of cranium of Diadectes. A.M.N.H. No. 4441, with data from Watson (1916). Shows the primitive sphenethmoid, large epiphyseal tube and low otic-occipital chamber. B, Lateral view of pareiasaur braincase, showing reduced epipterygoid. After Boonstra. B 1, Mesial section of pareiasaur braincase. After Watson. C, Paramesial section and right half of skull of Chelone, showing highly raised roof of braincase, great sagittal crest, and abortive epipterygoid. 294 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 second ceratobranchial. In an embryo Chelone basic plan and its peculiarities to descent figured by Parker his "basihyal" is relatively with modification from the cotylosaurian shorter and wider than it is in the adults of stock, with special affinity to the pareiasaurs. the other genera (fig. 11). Possibly these dif- The base of the cranium is essentially

D Rhinochelys

'2 quj dio soc A Diadectes FIG. 13. Dorsal aspect of skulls of Diadectes (A), pareiasaur (B), Triassochelys (C), Jurassic amphichelydian (D). B, After Boonstra; C, afterJaekel; D, from Zittel. ferences may be associated with (1) the cotylosaurian in its lack of cranial flexure and shortness and great width of the neck in strong development of the median basi- pareiasaurs; (2) the narrowness and flexible parasphenoid rostrum. Through the hyper- tubular form of the neck in chelonians. trophy of the jaw muscles, the roof of the Mesial sections of the skull (fig. 12) in turtle skull has been raised high above the Diadectes, pareiasaurs, and modern turtle low-roofed level of the diadectids and again show that the turtle skull owes both its pareiasaurs. The epipterygoid, which in 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 295

B Chelys

A1 B1 FIG. 14. Skulls of Pleurodira. Podocnemis (A, A 1), after Hay (1908, pl. 4); Chelys (B, B 1), from American Museum specimens. 296 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 diadectids is very large and helps support the where within limits of the Pareiasauria, but skull roof, is reduced in pareiasaurs (fig. 12B) not perhaps in any of the known gigantic and nearly aborted in Chelone, and the large forms of that group. pineal opening of the cotylosaurs is completely closed. VERTEBRAE AND RIBS In spite of the advanced stage represented In modernized turtles the expanded ribs, by Chelone in the construction of its sec- together with the bony neural plates, col- ondary palate and massive jaws, its dorsal lectively form a single immovable carapace, skull pattern nevertheless retains the general which effectively prohibits either dorso- type seen in Triassochelys, Rhinochelys (fig. ventral or lateral flexure of the greater part 13D), and other Mesozoic amphichelydians, of the vertebral column; thus the only parts which are all presumably more primitive of the column left free to move are the than any of the modern cryptodirans or cervical and caudal vertebrae. Hence it is not pleurodirans. In the leathery turtles (Der- surprising that the centra (pl. 34), especially mochelys, Sphargis) the skull is essentially of the mid-thoracic vertebrae, are greatly like that of Chelone, but with secondarily elongate besides being closely appressed to enfeebled jaws and a marked tendency for each other at both ends; or that the anterior the dried skull to fall apart easily, due to and posterior zygapophyses, which permit cartilaginous connections and poor sutures. dorso-ventral flexure in primitive reptiles, In the Chelonidae (fig. 6D) the shell of have here lost their transverse facets and bone over the temporal region remains but have become merged with the anterior and little excavated from the rear and rather posterior descending vertical prolongations closely resembles that of Triassochelys. Con- of the neural plates above the spinal nerve ceivably this condition may be secondary, as exits. True neural spines seem to be absent the, in general, more primitive Chelydra has except on the first and last thoracic and first a more deeply embayed temporal roof, while sacral vertebrae. The latter, at least in the typical tortoises retain only a post- Chelydra, retains a real neural spine and what frontal-quadratojugal bar. appears to be a postzygapophysis. Mean- Fenestration of the temporal roof from the while the tunnel for the spinal cord has rear is also seen in various stages in the increased considerably in relative size. Pleurodira (fig. 14), from the relatively Each of the widely spreading thoracic ribs primitive condition in Podocnemis, which is of Macrochelys is immovably fastened to the essentially like that of Chelydra, to the vertebral column by a sharply forking excessively specialized stage in the Matamata V-shaped proximal end; the thin lower (Chelysfimbriata), in which the emargination branch (capitulum), resembling an inverted of the temporal region from the rear has been flying buttress (fig. I5C), is suturally united supplemented by extensive transverse emar- intercentrally with a wide lateral projection gination, so that the dorsal sides of the broad from the centra of two successive vertebrae pterygoids are widely exposed. The skull in and from the base of the neural arch; the the trionychids is likewise specialized in still upper costal branch (pseudo-tuberculum), other ways from the primitive cryptodire which is continuous laterally only with pattern. the so-called dermal portion of the rib Thus there seems to be no great risk of complex, makes an extended sutural contact serious error either in assuming a sub- with the neural plate. stantial unity of origin of the entire order or In Triassochelys both the visceral surface in taking the skulls of Triassochelys and other and the cross section of the carapace, as Amphichelydia as representative of a primi- figured by Jaekel (1918, pp. 103, 129), tive chelonian type that was ancestral to conform closely to the foregoing description, those of modernized Cryptodira, Pleurodira, especially in the relations of the thoracic and related branches. The striking evidences centra to each other and to the wide rib of affinity of the skull of Triassochelys to the plates, so that in these features the Upper pareiasaurian type suggest that the origin Triassic chelonians were almost as modern- of the chelonian order is to be sought some- ized as their surviving descendants. 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 297 The thoracic vertebrae and ribs of pareia- wide, hinge-like joints, favoring dorso-ventral saurs (fig. 15A), while nearer to those of flexure but prohibiting lateral abduction or diadectids (fig. 16A), are also to some extent adduction. The zygapophyses favor the same modified in the direction of Triassochelys, flexures and are not radically different from as shown in table 1. the primitive types in their directions. In TABLE 1 VERTEBRAL CHARACTERS Pareiasaur Diadectes (Bradysaurus) Triassochelys Proportions of a mid- Very high Much lower, with very Low and elongate, with thoracic vertebra wide transverse process- transversely short trans- es verse processes Anterior and posterior Large, supported by The same Absent or fused with zygapophyses massive buttresses vertical flanges from neural plates Hyposphenes and epi- Highly developed Reduced Absent sphenes Thoracic ribs Curved downward a- Essentially as in Dia- Ribs arching upward, round a somewhat deep, dectes, but with very then outward and barrel-like thorax wide proximal ends con- downward around a necting with wide trans- wide, relatively low car- verse processes apace. Ribs greatly widened and flattened, connecting with horizontal transverse processes and with neural plates Lateral flexibility of tho- Marked Reduced None racic region

In short, the thoracic region of pareiasaurs the Pleurodira, on the other hand, the centra seems to afford a nearer source for that of develop ball-and-socket joints, the opposite the most primitive known chelonian, Trias- anterior zygapophyses and their supporting sochelys, than does that of Diadectes. pedicels are drawn in near to the midline, With regard to the numbers of presacral while their articular facets have slipped vertebrae: for Diadectes, Case (1911, pl. 14) around onto the front face of the zygapo- figures 22; for the pareiasaur Bradysaurus, physial process and there form a large, Boonstra (1933c, pl. 28) and Hartmann- strongly convex turntable for the equally Weinberg (1930) give 19; for Triassochelys, concave facet of the posterior zygapophysis Jaekel gives 18 (if we subtract the pro- of the preceding vertebra. This is a far higher atlas) and in modern Chelonia also there are specialization than that which is attained by 18 (Gadow, 1920, pp. 315, 316). the Cryptodira, but both are easily derivable The cervical vertebrae (figs. 17-19) of from the conditions described by Jaekel for modern Cryptodira and Pleurodira differ Triassochelys, while all the cervical vertebrae profoundly and condition the well-known of the latter are not very far beyond those differences in the method of retracting the of the pareiasaurs. head. In the former the centra, especially of In Chelydra and other Cryptodira the ninth the sixth and seventh vertebrae, develop vertebra (fig. 15C) forms the main pivot for 298 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 the sigmoid neck, from which it receives downwardly curved anterior zygapophyses upward and backward thrusts. Its strong, and their pillars, while wrenching strains forwardly produced neural spine, together are neutralized by inverted flying buttresses

a(-D2), ( Ob/hlue ver7ntr.On/Crior v;ew C Macroche/ys

anterior A 3cutosauru4 FIG. 15. Relations of anterior dorsal ribs and vertebrae in pareiasaur (A), Triassic (B), and modernized (C) chelonians. A, From American Museum specimen. with the well-braced bowl at the front end of formed by the capitula of its own ribs and its centrum, resists upthrusts. Anteroposte- those of its neighbor, the tenth vertebra. rior thrusts are met by its great smooth, These complexities are barely foreshadowed 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 299

FIG. 16. Dorsal vertebrae of Diadectes (A, A 1, A 2, B, B 1) and python (C, C 1, C 2). A, A 1, A 2, From A.M.N.H. No. 4354. Shows the contrast between the zygo- sphene-zygantrum articulations of serpents (C, C 1, C 2), which are dorsal to zygapophysial articular surfaces, and the hyposphene-episphene articulations of Diadectes, which are ventral to them. The supposed "hypantrum" on the anterior face (Williston, 1925, p. 91) is indeed a single median depression but, contrary to long-accepted statements, it does not bear articular facets and probably lodged median ligaments or muscles. The true accessory articular surfaces are furnished by the'paired dorsal faces of the zygosphenes and the paired ventral surfaces of the hitherto unnamed episphenes. 300 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 in Triassochelys and are -not to be expected position. In embryonic Istages of recent in a pre-chelonian ancestor before the con- chelonians (fig. 20) the rudiment of the solidation of the carapace. carapace appears well behind the root of the very large neck and gradually overspreads PECTORAL GIRDLE AND LIMBS the pectoral girdle, raising the anterior ribs The carapace of the Permian Eunotosaurus and swinging them far beyond the tops of the (Watson, 1914b) was already beginning to scapulae (Ruckes, 1929b, pp. 109-115, figs. overlap the upper ends of the pectoral girdle 1-30).

E Chelys LEFT SIDE VIEWS FIG. 17. Cervical vertebrae of pareiasaur (A), Trias- sochelys (B), and modernized chelonians (C, Dr E, E 1). A, After Boonstra; B, after Jaekel; C, D, E, E 1, from specimens.

(fig. 4), and in Triassochelys it projected The plastron develops in the floor of the considerably above the cleithra (Jaekel, pectoral body wall, extending from beneath 1918, p. -116, fig. 17). The astonishing retrac- the heart to beneath the an'terior end of the tion of the root of the neck beneath the pelvis; it probably replaces an earlier para- anterior ribs was made possible in part by an sternum. Even in the 6-mm. embryo of early transverse widening of the fore part of Chelydra (fig. 20A) the plastron ends sharply the carapace, involving the raising of the at the humeral and femoral notches. The anterior thoracic ribs into a nearly horizontal former defines the sharp and ancient bound- BULLETIN AMER. Mus. NAT. HIST. VOL. 86, PLATE 34

t. i1 i jFh . .:.. i

't' ;2

,,.: I:

Skeleton of Testudo pardalis BULLETIN AMER. Mus. NAT. HIST. VoL. 86, PLATE 35

A--.#.1:

"."/A' ..

Skeleton of Chelone 1946 GREGORY: PAREIASAURS-AS ANCESTORS TO TURTLES 301

A B C U PAREIASALUR AMPHICHELYDIAmN CRYPTODIRE PLEURODIRE Bradysaurus Triassochelys Macrochely4s Chelys fimbriata FRONT VIEWS

LEFT SIDE VIEWS FIG. 18. Cervical vertebrae in pareiasaur (A, A 1), amphichelydian (B, B 1), cryptodiran (C, C 1), and pleurodiran (D, D 1). Side and front views. From specimens.

A Macroche/ys B Chelys RIGHT SIDE VIEWS

At Bl roP VIEWS FIG. 19. Sixth and seventh cervical vertebrae of cryptodiran (A, A 1) and pleurodiran (B, B 1). Top views, showing different adaptations for bending the neck. 302 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 ary between the dermal pectoral girdle and origin from the dorsum of the neck and the neck; the femoral notch marks the place thorax onto the outer sides of the deep where the pelvic limb bud has, as it were, scapular muscles (fig. 21). These and other burrowed inward and upward. factors cooperated in the piling up of the As the plastron surrounds the interclavicle shoulder muscles into a pyramidal mass, and clavicles, it impresses them into itself which was supported at its core by the rod- and separates them from the coracoids, which like scapula, the sharply projecting acromion,

A Che/ydra "'u" Af FIG. 20. Embryo turtles, showing the origin and spread of the carapace and plastron. A, A 1, After Ruckes (1929b); B, B 1, after W. K. Parker (1880). thus appear as free blades in the dried and the inwardly directed, fan-like coracoid. skeleton (pl. 35). The coracoid and acromion are tied together The new spatial relations of the shoulder by a so-called episternal band (Huntington, girdle to the enclosing shell accompanied 1918, p. 372). The changes in the musculature radical changes in the movements of the of the pelvic limbs were less radical (fig. 22). limbs and a shift of the trapezius, latissimus The morphologic sequence of changes of dorsi, and other muscles that formerly took the parts of the pectoral girdle indicated 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 303

hvcid /t*m

1 A , ,, p 4J 4

Sphenodonr

A2 43--~~~~4 ---

I6 43

I - J. jj 4 x4c asiI 4"! .3 'o .i ., ZB MI Te s tudo B5 FIG. 21. Musculature of the pectoral girdle in a primitive reptile (A, A 1), and in a typical chelonian (B), lateral and ventral aspects. A, A 1, A 2, From Gregory and Camp, after Fiirbringer; B, B 1, after Gadow. For key to muscles and nerves, see Gregory and Camp (1918).

TABLE 2 CHARACTERS OF THE PECTORAL GIRDLE Diadecites Pareiasaur Trias sochelys Chelydra Scapula Flattened, blade- Higher; shorter an- High and very short Tall and rod-like like teroposteriorly Acromial process Not developed Conspicuous Large Very long, rod-like Glenoid articular Extended antero- Coracoid part ob- Extended obliquely Nearly vertical facet posteriorly lique downward and backward

Coracoid Plate-like, extended Shorter anteropos- Very short; extend- Fan-like or spatu- anteroposteriorly teriorly ed inward late 304 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86

A Testudo - B Alligator FIG. 22. Musculature of the pelvic limb in a chelonian (A) and a typical reptile (B). After Gadow. For key to lettering, see Gadow (1882).

B Embrithosauru.s C Triassochelys FIG. 23. Pectoral girdles of diadectid (A), pareiasaur (B), Triassochelys (C), and Chelydra (D). A, B, D, From American Museum specimens; C, after Jaekel. 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 305 in figure 23A-D may be summarized as in and other flexors of the forearm. The op- table 2. posite, or ectocondyle, although much shorter The humerus of Diadectes has strongly than the entocondyle, was very massive emphasized all the marks of a primitive and quite conspicuous. From its margin arose Permian tetrapod (Romer, 1922, pp. 554- the extensor carpi and extensor digitorum 559; Miner, 1925, fig. 19). For example, its longus muscles. Above the entocondyle was shaft (fig. 24A) was very short, and the a conspicuous supinator process for the extremely wide proximal end bore a long, humero-radialis and supinator longus. On flattened, spirally warped capital facet; this the huge deltopectoral crest were inserted permitted a limited twisting and rocking the pectoralis, the dorsalis scapulae, and the movement on the glenoid articular facet of clavicular portions of the deltoid. On the the scapulo-coracoid. In accord with the dorsal surface of the shaft of the humerus was

FIG. 24. Humeri of diadectid (A), pareiasaur (B), tortoise (C), and sea turtle (D). American Museum specimens. sharply angulated forearm, the large capitel- inserted the powerful anconeus medialis lum or radial facet was on the front or ventral (part of the triceps), while the ventral surface face of the shaft, facing downward, but was of the shaft gave rise to the brachialis in- continuous medially and dorsoposteriorly ternus. The entepicondylar foramen in primi- with the facet for the ulna. The very large tive reptiles serves for the transmission and distal end of the humerus was produced into protection of the brachial nerve and artery a huge, downwardly pointing entocondylar (Huntington, 1918, p. 390); in Diadectes it is process, the general plane of which was represented by a long and conspicuous twisted to about 80 degrees to the plane of fissure. The ectepicondylar foramen had not the proximal end. Miner (1925) and Romer yet appeared but, as shown by earlier authors (1922) agree that on and near the proximal and confirmed by Romer (1922) and Miner margin of this entocondylar process were (1925), arose later by the folding down of the inserted the great coracobrachialis muscles, supinator crest over the ectocondyle and the a flexor of the upper arm; from its antero- subsequent conversion of the groove into a ventral surface arose the flexor carpi radialis canal. 306 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 In the humerus of the pareiasaur Propap- movements by the pectoral and pelvic pus (fig. 24B) the primitive tetrapod charac- notches, the following changes have taken ters noted above are somewhat modified by place: the further expansion of the radial capitellum 1. Greater mobility at the shoulder was into a huge, partly globose prominence and attained by developing the caput humeri extension of the ulnar facet onto the dorsal into a ball-like protuberance pari passu with surface of the olecranal depression, so that the above change in the glenoid facet of the the forearm could be fully extended on the scapulo-coracoid and with the increasing humerus. Moreover, the supinator crest has ability to raise the body well off the ground. bent downward and joined the entocondyle, 2. The length of the stride was increased while the ventral process of the deltopectoral by lengthening the shaft, while the retraction crest has grown into a roundly triangular of the limb into the humeral notch was protuberance. facilitated by the marked narrowing of the

RIGHT MANUS

BBasrsI

k)s.Y *pfts A Uimnoscelis 5 8rad,ysaurus C Testudo FIG. 25. Forearm and hand of cotylosaur (A), pareiasaur (B), and tortoise (C). A, C, After Williston; B, after Boonstra. Although the humerus of Triassochelys distal end, with the elimination of both itself is not known, von Huene (1926, p. 510, entepicondylar and ectepicondylar projec- fig. 1) describes and figures a chelonian tions. humerus from the uppermost Trias of 3. Still greater extension of the lower arm Wuirttemberg which he refers, on good was obtained by widening the radial facet all evidence, to the Amphichelydia. This is across the distal end. already essentially chelonian in its sigmoid 4. Acceleration of growth rates on the flexure of the shaft in lateral view, its dorsad dorsal side of the distal half of the humerus turning of the globular caput humeri, the shaft caused the latter to curve downward presence of an ectepicondylar foramen and toward the top of the radius and ulna, while canal, and in the general form of the distal retardation of growth rates on the ventral end. But it is apparently more primitive side, just distal to the head, together with than the humerus of Testudo (fig. 24C) in its rapid increase of the lateral tuberosity and notably smaller medial tuberosity ("median still greater growth of the medial tuberosity, trochanter") and in the more globular form and other anisomeral growth, all imparted a and smaller size of the radial capitellum. sigmoid curvature of the bone as a whole The humerus of Testudo (fig. 24C) at first when viewed from either side. The result of sight differs much more from that of Propap- these and similar modifications was to permit pus than the latter does from that of Diadec- the opposite humeri to be swung forward tes. Further comparison, however, suggests in front of the humeral notches and parallel that with the fixation of the carapace and to the extended neck, and at the same time plastron and the limitation of the limb to twist distally down toward the ground, 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 307 so that in walking the elbows could be drawn D IV and suddenly to D V. The phalangeal forward like knees (fig. 5A and pl. 34). formula was reduced to 2, 3, 3, 3, 2. Even greater changes have occurred during The forearm and hand of Testudo (fig. the evolution of the strange humerus of the 25C) are basically the same as in the parei- sea turtles (fig. 24D), where, as part of the asaurs but with some anisomeral emphases, wing-like paddle (pl. 35), the humerus, losing e.g., the radiale has become very wide and its sigmoid curvature, has become seconda- may have absorbed the centrals, the meta- rily flattened, the medial tuberosity has at- carpals are excessively short, all the claws tained great size, while the deltopectoral are very large, and the phalangeal formula is crest has been greatly reduced. Thus the reduced to 2, 2, 2, 2, 2. The manus as a whole humeri of Diadectes, Propappus, the Upper is strikingly elephantine and adapted for a Triassic amphichelydian Testudo, and Che- vertical position of the forearm of a heavy lone fall into a well-knit morphological series animal. which reinforces much other evidence that Thus, as here noted in regard to many the chelonian order has been derived remotely other parts, the pareiasaur stage in the hand from the basic cotylosaurs and more im- appears to be structurally ancestral to the mediately from the stem pareiasaurs. chelonian. For the forearm and hand (fig. 25A) of a The hands of the sea turtles have become cotylosaur we may use Williston's figure of transformed into wing-like flippers (pl. 35). those parts in Limnoscelis. The radius and ulna are wide transversely, the distal end of PELVIC GIRDLE AND LIMBS the ulna extending well below that of the The pelvis of Diadectes (fig. 26A) has its radius. The olecranon is large. Part of the ventral, pubo-ischiadic plates well extended carpals were cartilaginous. There were four anteroposteriorly to give a long swing to the bones in the first row, the radiale being rather femora, while the dorsal border of its low small. By analogy with other primitive Per- ilium is extended anteroposteriorly for the mian tetrapods, Williston probably correctly longissimus dorsi and ilio-costalis muscles arranged all the digits in a series converging (Romer, 1922, pl. 44). The large acetabular toward the ulna. The wide metacarpals in- depression has a cardiform rim, its axis creased rapidly in length from Mtc I to Mtc pointing obliquely downward. It already III; Mtc IV was about equal to III. Mtc V permits considerable twisting and long was shorter and narrower than III and IV, anteroposterior swing of the caput femoris. and it diverged remarkably from Mtc IV. All these and other features of the pelvis of The primitive reptilian phalangeal formula Diadectes are associated with the sprawling was preserved intact, namely, 2, 3, 4, 5, 4. movements, with strongly bent and everted The distal phalanges indicate relatively small posture of the thighs and with the inability nail-like claws. to raise the pelvis high off the ground. In the forearm and hand of pareiasaurs The pelvis of pareiasaurs (fig. 26B) is much (fig. 25B), as figured from excellent material shorter ventrally and higher dorsally than by Boonstra (1932b, pt. 8, pls. 42, 43) the that of Diadectes, and the acetabulum has a olecranon is enormous, implying an extremely shorter anteroposterior axis, is relatively powerful triceps system. The radiale is very higher dorsoventrally, and on the whole faces large and well ossified, the intermedium more obliquely downward, so that the head small. Probably two centralia were present. of the femur thrusts more upward against Some of the inner carpalia may have been the overhanging dorsal rim of the acetabular small or cartilaginous, but carpale 4 was large depression. The space for the gluteal muscles and had contacts with Mtc IV and V. All the on the outer blade of the ilium has greatly metacarpals were very short and wide, that increased; this prevents sagging of the of the pollex being extremely wide and mas- opposite side when its foot is raised from the sive. The first digit was very large and di- ground (cf. White, T. E., 1939). rected inward, the others diminished to the In Triassochelys (Jaekel, 1918) the well- small fifth. The ungual phalanges were very fenestrated pelvis (fig. 26C) is anteroposte- large, the first digit decreasing gradually to riorly shortened, the ilium both elevated and 308 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86

A Diadectes B Scutosaurus C Triassochelys FIG. 26. Pelvis of diadectid (A), pareiasaur (B), and amphichelydian (C). A, B, From American Museum specimens; C, after Jaekel.

H1. z. N r C.. A w2Diadectes B Propappus FIG. 27. Right femur, medial aspect, of diadectid (A), pareiasaur (B), and tortoise (C). From American Museum specimens. 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 309 shortened to support the dome of the cara- In the pareiasaur Propappus (fig. 27B) the pace, while the acetabulum has been deep- head end is bent forward and upward on the ened dorsoventrally in accord with the shaft, the caput femoris is expanded into a globular development of the caput femoris wide oval, the prominent internal trochanter and now faces partly backward as well as bounds a large concavity, presumably for the outward to permit a longer sweep of the pubi-ischio-femoralis muscles, and the facets femur and a more direct thrust from the rear. for the tibia are rounder, permitting wider The pelvis of Testudo is remarkably close extension of the tibia on the femur than was to that of Triassochelys. In side view the possible in Diadectes. three branches, ilium, pubis, ischium, form In Testudo the caput femoris (fig. 27C) has a truss or kingpost for the support of the become globular, and its midpoint is inclined AIGHT

A Diadectes B Bradysaaras C Testudo FIG. 28. Right foot of diadectid (A), pareiasaur (B), and tortoise (C). A, After Romer and Byrne; B, after Haughton; C, after Williston. carapace (Ruckes, 1929a) especially in Pleu- nearly at a right angle to the axis of the rodira, where the pubis and ischium are upper half of the shaft. The sigmoid contour solidly based on the plastron. In aquatic and is due to the faster growth of the ventral marine turtles, as the carapace widens the side of the proximal half and of the dorsal ilium is directed backward as well as upward side of the distal half. Through this sigmoid (Ruckes, 1929a). In the soft-shelled turtles curve the lower leg and foot can be extended (trionychids) the carapace becomes very behind the carapace, and when the knee is broad and nearly flat, while the pubes send drawn forward, the shank and foot can be forward a pair of great anterior plates turned downward around the edge of the analogous with those of plesiosaurs and femoral notch. equally concerned with the protraction and Thus the typical chelonian femur aligns retraction of the rear paddles. itself with other parts in suggesting that the The femur of Diadectes (fig. 27A) is pareiasaur affords a structurally connecting adapted to the bent-kneed posture, in which stage on the path of its evolution from the the femur is directed outward and slightly diadectid cotylosaur. upward. The caput femoris is an elongate The lower leg and foot of Diadectes (fig. oval of slight convexity; the shaft is nearly 28A) conserve much of the primitive tetrapod straight, the distal end with large protuber- condition (Romer, 1922; Romer and Byrne, ances and two wide facets for the tibia 1931; Schaeffer, 1941, p. 428). Their principal separated by a deep notch for the extensor specialization is the enlargement of the tibialis. On the proximal ventral third there intermedium and fibulare to form an astraga- is a high internal trochanter and a median lus and a calcaneum, whose facets permit depression supported distally by a low, well- considerable rotation of the tibia on the rounded stem. astragalus. However, there was also a 310 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 mesotarsal joint as in pareiasaurs (Schaeffer, general it closely approaches the Testudo ibid., p. 431). The primitive anisomerism of type. the metatarsals and convergence of the In Testudo (fig. 28) the consolidation of the entire tarsus toward the fibula were retained, astragalus and calcaneum suggests derivation as was also the primitive reptilian phalangeal from massive pareiasaurs in which such formula (1, 2, 3, 4, 5, 3). consolidation had already taken place. Thus The pareiasaur pes (fig. 28) has advanced the primitive Triassochelys may be a dwarf beyond that of Diadectes, especially in the derivative of a much heavier race. All the enlargement and consolidation of the astrag- tarsals are now well ossified, and the phalan- alus and calcaneum and in the reduction of geal formula in chelonians is reduced to the phalangeal formula to 2, 3, 3, 4, 3. In 2, 3, 3, 3, 3 (Williston, 1925, fig. 154, p. 188). COMPARISON OF CAPTORHINOMORPH AND TURTLE SKELETONS IN SOME FEATURES the smaller cotylosaurs, advanced beyond the primitive reptilian especially the Upper Pennsylvanian and/or type and might well be almost immediately Lower Permian captorhinomorphs, would ancestral to the diadectid, and more remotely seem at first sight to make a better starting ancestral to the pareiasaurian, and eventually point for the turtles than the diadectids, to the chelonian type. which already had a specialized dentition. The entire palatal view of Limnoscelis For the most part, the captorhinomorphs (Williston, 1914) was also of primitive were rather small and lizard-like, the excep- reptilian stamp, especially in so far as the pm.ar po

tnx dni A Captca'rhinus

,Pd Pa~~~~F;o~fr .pasph ~-ectpr bptK.bsph

I O boc YZ At A2 Mz. FIG. 29. Skull of a captorhinid cotylosaur. Composite from American Museum specimens. Sutural pattern after Case, Williston, Broom, von Huene, Sushkin, Romer, and Price. tion being Limnoscelis; this reptile, as widening of the jaw muscles had bent the restored by Williston (1914, p. 20), was about pterygoids sharply inward against the basi- 2 meters long. Limnoscelis had a far more pterygoid processes of the parabasisphenoid; primitive dentition than Diadectes, with meanwhile under the influence of the same simple conical-tipped cheek teeth, not modi- capiti-mandibularis muscle mass, the ante- fied into crushing ovals. Its rather large front rior processes of the pterygoid were: (1) upper teeth pointed downward and a little bending outward and thus enlarging the in- backward and were for catching and piercing. terpterygoid vacuity; (2) developing sharply On the whole, the jaws and dentition, al- transverse tooth-studded bars; and (3) grow- though perhaps on the way toward a snipe- ing downward to form a descending process, billed, fish-catching stage, were not far or sliding surface, to guide the mandible 311 312 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 upward into correct occlusal relations with anterior nares to the orbit. Thus the skull, the upper jaw and teeth. Thus in Limnoscelis jaws, and dentition of the captorhinomorphs (Williston, 1911, fig. 4, p. 29) was perhaps are far more primitive than those of chelo- the earliest known example of the so-called nians, but they do not show such convincing rhynchocephalian palate, which is more evidences of close relationship to the chelo- delicately constructed in the smaller capto- nians as do those of the pareiasaurs. rhinomorphs and in the modern tuatara. The The vertebrae of Limnoscelis and the protean modifications of this pattern are smaller captorhinomorphs were essentially manifest in the diadectids, pareiasaurs, like those of Diadectes, except that they had chelonians, and in fact in all known reptilian not developed hyposphenes and hypantra. orders, while other and still greater trans- They differed from those of pareiasaurs in formations are seen in the birds and mam- having very short transverse processes. But mals. in the absence of an immovable plastron, the In the smaller captorhinomorphs (fig. 29) vertebral column as a whole retained con- the upper front teeth were pointed downward siderable flexibility, and the thoracic part had and overhung the somewhat procumbent not begun to be immobilized as it was in the lower front teeth, essentially as in Limno- chelonians. scelis, but the cheek teeth were small, It can also be said of the captorhinid ribs, numerous, and in clusters, and there were pectoral and pelvic girdles, and limbs, hands, small teeth on the anterior branches of the and feet, that while they retained many pterygoids and prevomers. But secondary primitive reptilian features, which at a palatal plates from the maxillae and pala- remote period the ancestors of the chelonians tines, such as are seen in the primitive must also have possessed, yet the capto- chelonians, had nrot developed. rhinids give no indubitable signs of being The smaller captorhinomorphs (Labido- direct ancestors of the turtles. Moreover, it sauris, Captorhinus), while sharing with seems likely that if the captorhinomorphs had Limnoscelis the elimination of the otic notch, possessed a vigorously developed dermal have advanced beyond Limnoscelis in the skeleton of bony plates embedded in a loss of the supratemporals, in the reduction leathery skin, some tokens of it would have of the tabulars to vestiges, and in the turning been discovered. On the contrary, the downward of the whole of the dermosupraoc- captorhinids present more marks of relation- cipitals onto the nuchal surface. Otherwise ship to the pelycosaurs and other forms their skull and jaws retain such primitive that were becoming swifter and more agile reptilian features as an unperforated tem- predators (Romer and Price, 1940). poral roof and a lacrimal extending from the COMPARISON OF SEYMOURIAMORPH AND TURTLE SKELETONS

THE THOROUGH MONOGRAPHIC STUDIES of by posttemporal fenestrae (White, 1939, T. E. White (1939) on Seymouria from the p. 334). The large otic notch and the con- Lower Permian of Texas and of Bystrow sequent curvatures of the quadrate, squamo- (1944) on Kotlassia from the Upper Permian sal, supratemporal, and tabular merely show of Russia have shown that these animals, that in both Seymouria and the chelonians whether regarded either as primitive reptiles this region was being moulded by a divertic- or as advanced amphibians, retain many ulum of the tubo-tympanal canal to support primitive tetrapod characters intermingled a tympanic membrane. with their own divergent specializations, but Although its carapacial covering, if any, neither of them shows any definitely pre- is unknown, in its girdles and limbs Sey- chelonian tendencies. mouria was on the whole far more primitive In a general way Seymouria remotely than the chelonians and had already adopted approaches the diadectid cotylosaurs in its the rocking, bent-legged gait which even- general habitus, especially in the vertebrae, tually contributed to the fixation of the which have large, high, transverse processes, column by the coalescence of the carapacial large pre-zygapophysial and post-zygapo- elements. On the whole the evidence suggests physial facets, rounded shoulder-like swell- that Seymouria was a sort of contemporary ings above the posterior zygapophysial great uncle to the diadectid cotylosaurs and facets, and well-developed intercentra, which perhaps through them could claim a remote carry facets for the capitula of the ribs. connection with the chelonians. But hyposphene-episphene articulations are Still further removed from the chelonians wanting. was the Upper Permian Kotlassia, the habitus The entire vertebral column of Seymouria of which was tending in the direction of such seems sufficiently primitive to have given secondarily aquatic urodeles as Crypto- rise to that of the chelonians, but the branchus. Although it had a well-developed evidence from other parts of the skeleton plated dorsal armor of flat quadrangular indicates decisively that it did not do so. bony scutes, doubtless surfaced with a In many other ways Seymouria remained leathery skin, this armor was more nearly on, or had drifted toward, an amphibian akin to that of the amphibian otocoelids than level and away from an immediately pre- to the chelonian type. Its wide flattened chelonian stage. Thus, its muzzle was broad, skull, truncated posteriorly, suggested the with flat amphibian-like palate, which as yet larval Branchiosaurus, retaining the primitive showed little if any tendency toward the intertemporals and supratemporals and dor- inrolling of palatal plates of the palatines and sally exposed tabulars and dermo-supraoc- maxillae to form a secondary palate (see the cipitals. Its mandible retained the amphibian cross sections figured by White, 1939, p. intercoronoid and two splenials, while its 340); nor was the occipital plate perforated tooth structure was also labyrinthodont.

313 COMPARISON OF PLACODONT AND TURTLE SKELETONS As NOTED ABOVE (p. 282), the view that the pattern than did the placodont. This is chelonians are nearly related to the turtle- equivalent to saying that solely on the basis like placodonts rested in part on the com- of skull pattern, the placodonts could more parison of modern turtle skulls with Triassic easily have been derived from the earliest

Triassoche/lys

C Placochelys FIG. 30. Skulls of pareiasaur (A), Triassochelys (B), and Placochelys (C). A, After Boonstra; B, after Jae- kel; C, after von Heune. placodonts. But if we compare (fig. 30) the known chelonians than the chelonians from Triassic turtle skull with those of Triassic the placodonts! It is true that both placodont , placodonts (Broili, 1912; von Huene, 1931) and chelonian quadrates and squamosals and a Permian pareiasaur, the turtle is seen have been curved or hollowed out by a dorsal to have retained a far older, less specialized dilatation of the tubo-tympanal system. But 314 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 315 so have the quadrates of mosasaurs, phyto- Henodus (fig. 31A 1) shows that one placo- saurs, and crocodilians and the temporal dont at least got rid of most of its teeth and complex of notoungulates, jerboas, and men. acquired a beak. But the beak of Henodus It is also truejthat the Lower Jurassic did not entitle it to claim close kinship either

Ai - A FIG. 31. The "false turtle," Henodus chelyops von Huene. A, Dorsal view; A 1, skull, palatal aspect; A 2, another specimen with one-half of mandible. After von Huene. 316 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 with shoveller ducks or the duck-billed muscles worked their way backward through platypus. In short, although there are several the posttemporal openings (fig. 8B 1), and striking resemblances between placodont and great pressure was exerted on the anterior chelonian skulls, the earliest chelonians still beaked part of the wide jaws. literally belonged in the Anapsida, along In the palatal view, the skull of Trias- with the cotylosaurs and pareiasaurs, since sochelys was extremely unlike those of neither the sides nor the top of their tem- placodonts, and such features as the latter poral roof had been perforated to form do show in common with Triassochelys may supratemporal or lateral temporal fenestrae; safely be regarded as part of the common whereas in the placodonts the dorsal (supra- heritage from the Permian stem reptiles. temporal) fenestrae were well developed. The carapace of Placochelys consisted of an

*1 Dorsat line 7 Z Up~pervlertebral 3 / 4ZyZgapoph,sialplpper rib 5 Lower vertebral. 6 Lower rib 7 Ventral 'VggPlacodus g-as

FIG. 32. The skeleton of Placodus. Diagram based chiefly on photographs of the mounted skeleton in the Senckenberg Museum, Frankfurt. With data from Drevermann (1933). Estimated total length about 7j feet. The placodont skulls, even with their aggregation of large and small conical bosses, massive crushing teeth and appropriate which, as restored by Jaekel (1907), were bracing for the powerful jaw muscles and arranged in a more or less regular reticulate strong jaws, are generally recognized as pattern. In Henodus, the last of the placo- being genetically related to the skulls of donts (fig. 31), the general configuration of lariosaurs and nothosaurs. To these they were the wide, low carapace (von Huene, 1936a, also functionally related, somewhat as the 1938) suggests that of the trionychids, but massive skull of the sheepshead (Archosar- the secondary polyisomeres of the shell in gus), with its crushing teeth and thick jaws, Henodus were multitudinous, whereas in the is related to the lightly built skull of the far older Triassochelys the main units of the snappers (Lutianidae). The dentition of carapace were much fewer and essentially Paraplacodus, as described by Peyer (1935), chelonian in pattern (fig. 5). is indeed strongly suggestive of the sheeps- The postcranial skeleton of Placodus, as head type and like the latter may have been described by Drevermann (1933) and von derived from an early predatory stage with Huene (1933) was profoundly different from caniniform marginal teeth. that of any chelonian, and in general contour In the lariosaurs, nothosaurs, plesiosaurs, (fig. 32) it rather suggested a manatee. and placodonts (fig. 30C) the main pull of the Lacking a chelonian type of carapace, its temporal muscles was evidently nearly di- vertebrae, ribs, and gastralia were likewise rectly dorsoventral, as it is in other lines with un-turtle-like. In its very short neck the a "snipe-bill" or fish-snatching rostrum, often vertebrae did not even remotely foreshadow with narrow parallel rows of marginal teeth. the amazing specializations of either the In Triassochelys, on the other hand, the jaw Cryptodira or the Pleurodira. Its pectoral 1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 317 girdle was definitely nothosaurian: the clav (fig. 33), had become reduced to a slightly icles formed a great transverse bar, in whic] curved rod and lacked the vigorous develop- the small interclavicle was incorporated ment of the lateral and medial tuberosities as in nothosaurs; the scapulae, low ano that is characteristic of chelonians. A similar recurved, contrasted widely with the tall rod contrast could be shown to be true of the like scapulae of chelonians; the expanded pelvis and hind limbs. plate-like coracoids suggested the plesiosau In short, Placodus is as un-turtle-like in its type. The humerus, completely nothosauria girdles and limbs as it is in its dentition.

M1 G H T H U M! VENTRAL V/EWS

P/a

Champsosaurus NothosauruS ambulatvfr laramiensis C A B

i~~~B

D OR SA L VIEWS co FIG. 33. Comparative series of right humeri, illustrating progression from a relatively primitive, semi-aquatic type, with little modified limbs (A, B) to highly specialized marine plesiosaurs. The placodont humerus (D) closely resembles that of the nothosaur (C). A, B, After Barnum Brown; C, D, after von Meyer; E, after Watson; F, after Welles. 318 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86

And even when Henodus, the last of the Placochelys in the Upper Triassic. placodonts (von Huene, 1936a, 1938) made, According to Jaekel (1907, pl. 7), the stout as it were, a belated attempt to become a "ventral ribs" of Placochelys were closely turtle, the best it could do was to take on a embraced by widened bony "gastralia," very wide, round, flattened shell, remotely which were analogous with the plates sur- like that of a "soft-shelled" turtle (Trionyx) rounding the dorsal ribs of chelonians. and suggesting the subcircular disc of Presumably then the bony plastron of the Potamotrygon, with a stout long tail, not placodont Saurosphargis and its allies (von unlike that of Macrochelys, combined with a Huene, 1936a, p. 145) arose through the wide bill like that of a duck-billed dinosaur consolidation of the parasternum, or ventral but very short (fig. 31A 2). Meanwhile, rib system. But such a stage is unknown in however, Henodus retained a substantially any chelonian or pareiasaurian, and in the placodont heritage in its skull, vertebrae, chelonians the plastron may be due to a new ribs, girdles, limb bones, hands, and feet growth of the bone-producing integument. (von Huene, 1936a, p. 127; 1938, pp. 107- This may have replaced a precociously 111). eliminated gastralia complex, which had been Eunotosaurus, with its quite tortoise-like highly developed in many early tetrapods form (fig. 4), more nearly approached the but was poorly developed, or unreported, in later chelonian type in the Permian than did seymouriamorphs and cotylosaurs. DISCUSSION: CONVERGENT EVOLUTION AMONG TESTUDINATE ANIMALS

THE SHELL OF ANY CHELONIAN is only a very brates, the Ordovician ostracoderms, the special kind of armor, and whether made by units of the cephalo-thoracic shield or test man or produced in the laboratory of nature, were polyanisomeres of extreme complexity, armor is composed of polyisomeres. These with stratified basal layer, trabeculated mid- are serially homologous or homogenous units dle layer, and glass-like or "enamel" surface (Gregory, 1934, 1935b) such as the links of layer, all punctured by minute lymph or chain armor or the bony polygons of the mucous canals and richly supplied with cells shells of glyptodonts, ankylosaurs, and and blood vessels that deposited bone-like Dermochelys. Such polyisomeres are often and dentine-like material. organized or integrated into larger units, and With regard to external form, among some these complex pieces may be either fused to- of the later gastropods the screw-like growth gether into a fixed test or carapace or of the mantle became greatly accelerated on movably articulated with each other, as the the outer side of the aperture, while the spire parts covering the arms and legs in mediaeval was variously reduced. In the cowries, armor, or the plates on the abdominal seg- Cypraea caputserpentis has developed a ments of a lobster. Examples of polyisomeres streamlined domed shell, flattened beneath, belonging to non-homologous sets are the with bulging basal marginal angles, all multitudinous conical bony pieces of the curiously suggesting the shell of a high- carapace of Placochelys and the conical horny backed tortoise. Thus an almost bilaterally shields that make up the surface armor of symmetrical, tortoise-like shell has been certain tortoises. evolved from a remote helicoid ancestor. And When the parts of an originally homo- the shell of the tortoise itself has reached its logous series of polyisomeres undergo re- modern forms only after passing through gional changes in dimension and contour, several major transformations. they are called anisomeres. Thus anisomeres All the available evidence suggests that the are but modified polyisomeres, and one earliest vertebrates were bottom-living forms, category grades into the other. Combinations more or less depressed, broadly tadpole- of polyisomeres of different sets are called shaped, with an exoskeletal shield covering polyanisomeres. the head and shoulder girdle, followed by a In the animal kingdom the skeletal locomotor system of many segments (cf. polyisomeres and/or polyanisomeres are de- Heintz, 1935). posited by membranous tissue, such as the Free-swimming predatory fishes came later, shell-producing mantle of mollusks or the through reduction in size of the head and integument of vertebrates. In both cases multiplication of the locomotor segments. the shell or exoskeleton may be composed of From the earliest of these, in turn, sprang several layers, horny on the outside, cal- the air-breathing fishes, one branch of which careous or bony in the deep layers. (the Crossopterygii) gave rise to the am- There are often many striking resem- phibians. blances even in detail between the minute Among the latter the Upper Pennsylvanian structure of the shells of such widely un- and Lower Permian Otocoelidae were fore- related animals as the king crab, Limulus, shadowing the turtles in developing a series and the Silurian ostracoderms (Patten, 1912, of transversely wide bony scutes above the pp. 289-301). Patten indeed maintained neural spines of the dorsal vertebrae, to- that "the distinction between the epidermal gether with a circular tympanic membrane skeleton of the arachnids and the dermal and "meatal concavity." At one time Cope one in vertebrates is therefore only one of regarded these forms, together with Dia- degree, not of kind." However that may be, dectes, as possible ancestors of the Chelonia, it is a fact that in the oldest known verte- but much additional collateral evidence 319 320 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 TABLE 3 CONVERGENT EVOLUTION AMONG TESTUDINATE VERTEBRATES

Name Bothriolepis Otocoelus Triassochelys Henodus Archelon Ankylo-saurus Glyptodon Geologic age Devonian L. Permian Up. Triassic L. Jurassic Up. Creta- Up. Creta- Pleistocene ceous ceous Class Placodermi Amphibia Reptilia Reptilia Reptilia Reptilia Mammalia Order Antiarchi Labyrintho- Chelonia Placodontia Chelonia Dinosauria Edentata dontia Local habitat Fresh water ?Swamps More terrres- Inshore; Marine ?Swamps Terrestrial trial shallow water Contour of test or thorax in: Dorsal view Oval Oval Oval Wide oval Wide oval Oval Oval Lateral view Domed Flat Low domed Flat Flat Domed Domed Skull fastened to: Pectoral girdle Yes No No No No No No Free neck No Yes Yes Yes Yes Yes Yes Skull with helmet or scutes Yes Yes Yes No Yes Yes Pectoral appendages as: Paddles Yes No No Partly yes Yes No No Feet No Yes Yes Partly yes No Yes Yes Carapace covering scapula Yes Yes Yes Yes Yes Yes Yes Scutes above neural arches Yes Yes Yes Yes Yes Yes Yes Bony scutes uniting with ribs No ribs No Yes Yes Yes Yes No Ventral surface with: Scales or tubercles No ?No No Yes No ?Yes ?No Gastralia No Probably No Yes No ?No No yes Flat bony plastron Yes No Yes Yes Yes No No shows that they were specialized amphibians, modern Amphiuma, Proteus, and others. which were merely converging toward the Similar changes transformed the moderately turtles in certain features. short, well-limbed lizards into snake-like When in Upper Devonian or Lower Mis- forms of several families. All these had very sissippian time certain air-breathing fishes long and flexible bodies and are at the op- succeeded in transforming their stout paddles posite end stage from that of the turtles and into fore feet and hind feet, various lines of similar testudinate animals; the latter have further evolution became open to them. reached their terminus after shortening and Some of them, by multiplying the thoracic immobilizing the thorax, retaining stout and caudal segments, rapidly diminished the and fle-xible limbs, and usually emphasizing ratio of limb length to total body length. the mobility of the neck. Similar end results The terminal stages of such lines were have been attained in certain dinosaurs limbless serpentiform types, such as the (Ankylosaurus) and among the mammals in Permo-Carboniferous Aistopoda and the the glyptodonts. (See table 3.) CONCLUSIONS 1. COMPARATIVE ANALYSIS of chelonian skel- been reached but not before great strength, etons by earlier authors are confirmed and especially in the limbs, had been developed. extended, as indicating quite definitely that 7. The stem chelonians may have been the Testudinata (Chelonia) are a very derived from some small pareiasaurs related ancient group of reptiles without close rela- to Elginia, which is known chiefly from a tionship to other surviving orders. skull about 15 cm. long. This skull retains 2. The Triassic chelonian Triassochelys, a pineal foramen and small marginal teeth as described by Jaekel, already possessed a and has several spikes projecting from its truly chelonian carapace and plastron but squamosal border; it is thus regarded as retained many primitive reptilian features, nearer to the typical pareiasaurs than to the especially in its cervical vertebrae and primitive chelonian Triassochelys. pectoral girdle. 8. The placodonts, as clearly recognized 3. The conclusion of earlier authors that by Drevermann and von Huene, were a the Chelonia were derived from the Upper progressively sluggish and finally turtle- Pennsylvanian and Lower Permain cotylo- like offshoot of the nothosaurian stem, as saurs is confirmed. shown by their retention of many primitive 4. But the present comparative osteologi- sauropterygian features, including dorsal cal study shows that, among the cotylosaurs, supratemporal fenestrae, monimostylic or the Permian pareiasaurs approach nearer fixed quadrates, an almost plesiosaur-like to the Triassic chelonians in many features pectoral girdle, and a well-developed gas- than did the Upper Pennsylvanian and tralia system. Their latest member, Henodus Lower Permian diadectids. chelyops, had evolved a perfected carapace 5. The smaller captorhinomorphs had al- and plastron and was amazingly turtle-like ready lost the otic notch and reduced the in general appearance but retained numerous tabulars ("supratemporals") to a vestigial traces of a remote nothosaurian derivation, condition, whereas these elements, called together with many significant differences supratemporals by Jaekel in Triassochelys, from the contemporary true chelonians were apparently well developed in the stem (von Huene, 1936a, 1938). chelonians, although lost in later chelonians. 9. The Lower Permian Seymouria, whether Moreover the captorhinids were relatively regarded as a progressive labyrinthodont slender-limbed reptiles, and they have been or a primitive cotylosaur, already possessed generally recognized as suitable ancestors a well-developed carapace, consisting of for the pelycosaurs and therapsids. They numerous polygonal ossicles which were apparently lacked carapacial nodules and doubtless covered by a tough leathery in- show nothing definitely chelonian in their tegument, but neither the limbs nor girdles vertebrae, girdles, limbs, hands, or feet. show nearly so much resemblance to those 6. Even the gigantic known pareiasaurs of turtles as do the corresponding parts of seem to present almost ideal conditions for pareiasaurs. Kotlassia Bystrow (1944), the the derivation of the primitive chelonian Russian relative of Seymouria, seems to be aharacters of the skull, carapace, vertebrae, converging toward such modern urodeles as ribs, girdles, limbs, hands, and feet. The Cryptobranchus and has no evident claim stout girdles and limbs of pareiasaurs are for near relationship to the turtles. well adapted to the lifting propulsion of their 10. Testudinate forms, that is, with a massive bodies; the vertebrae, girdles, and carapace, have been developed independently limbs of Triassochelys were also stoutly built. in the following vertebrate groups: Placo- Perhaps the first turtles were both pygmies dermi: Antiarchi (Pterichthys, Bothriolepis); and pycnics, a pug-like branch of the pareia- Amphibia: Labyrinthodontia (Otocoelus); saurs in which adolescence and maturity Reptilia: Chelonia, Placodontia (Henodus), came on long before the normal height had Dinosauria (Ankylosaurus); Mammalia: Ed- 321 322 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 entata (Glyptodon). In such cases the survival has acquired its sigmoid curvature, its value of a strong carapace has evidently been globular head, and generally femur-like sufficient to compensate for its adverse appearance partly to enable it to function as effect on rapid locomotion. if it were an anterior knee, which can be 11. Some of the disadvantages and limita- directed sharply forward; the sigmoid flexure tions imposed by the carapace and plastron and other features also facilitate the with- of turtles have been met and compensated in drawal of the arm into the humeral notch the following ways: between the carapace and plastron. Here a. Fixation of the thoracic vertebrae and again the Pareiasaurus humerus gives the ribs by the plastron have been accompanied necessary suggestions as to the rather com- by remarkable differentiations in the cervical plex way in which the humerus of the typical vertebrae. Among the Cryptodira, the ninth chelonian has been derived from a primitive vertebra, firmly supported both by the reptilian type. carapace and by the tenth vertebra, serves d. The pelvis of typical chelonians, al- as a base and pivot for the vertically retract- though overgrown by the posterior border able neck, its wide, rolling anterior zygapo- of the plastron, retains some of its primitive physial facets permitting ample vertical and reptilian stamp, save for the fenestration of considerable lateral movements of the neck. the pubo-ischiadic plate. The chelonian In the Pleurodira, the anterior zygapophysial pelvis is foreshadowed in that of the pareia- facets, especially of the sixth and seventh saurian, which shows an early stage in the vertebrae, have grown forward and, together shortening of the pubo-ischial plate, the with the cup-and-ball articulations of the elevation of the ilium and the enlargement centra, form well-turned pivots for sharp and obliquely dorsoventral direction of the lateral movements of the neck. Neither of acetabulum. The more posterior direction these specializations was developed in Trias- of the acetabulum in the chelonian gives the sochelys, whose neck vertebrae remained femur a more direct thrust from the rear. much nearer to the pareiasaurian stage. e. The sigmoid femur of chelonians has b. The pectoral girdle of chelonians, over- raised the head above the plane of the shaft shadowed by the anterior rim of the cara- and is likewise readily derivable from the pace, transmits part of the thrust of the pareiasaurian stage. fore limb to the under side of the carapace. f. The almost elephant-like hands and feet It has lost its clavicles and interclavicle to the of land tortoises are directly derivable from plastron and has assumed a triramous form, those of pareiasaurs. with a long columnar scapula, a strong rod- g. The plastron of chelonians appears to like acromial process, and a fan-like coracoid. be an entirely unique development within These three form the frame of a muscular the chelonian order and is not foreshadowed pyramid for the scapulo-humeral and sub- by the sternum and parasternum or gastralia scapular coraco-brachial and other arm complex of lower reptiles, which was pre- muscles. Every part of the chelonian pectoral sumably lost before the origin of the order, girdle has evidently been derived from a but is due to the spread of the process of pareiasaurian or pareiasaur-like ancestral ossification across the wide ventral in- stage. tegumental zone between the pectoral girdle c. The humerus of the typical chelonian and the pelvis. REFERENCES

ADAMS, LEVERETT ALLEN Hist., vol. 86, pp. 225-274, figs. 1-20, 1919. A memoir on the phylogeny of the jaw pls. 25-33. muscles in recent and fossil vertebrates. COPE, EDWARD DRINKER Ann. New York Acad. Sci., vol. 28, pp. 1880. The skull of Empedocles. Amer. Nat., 51-166, figs. 1-5, pls. 1-13. vol. 14, p. 304. [First use of name BAUR, GEORGE "Cotylosauria."I 1887a. On the morphogeny of the carapace of 1896a. The ancestry of the Testudinata. Ibid., the Testudinata. Amer. Nat., vol. 21, vol. 30, pp. 398-400. p. 89. 1896b. The reptilian order Cotylosauria. Proc. 1887b. On the phylogenetic arrangement of the Amer. Phil. Soc., (for the year 1895), Sauropsida. Jour. Morph., vol. 1, no. 1, vol. 34, pp. 436-457, pls. 7-9. pp. 93-104. [Stereosternum, Progano- 1896c. Second contribution to the history of the sauria, p. 103; turtles related to stem of Cotylosauria. Ibid., vol. 35, pp. 122-139, plesiosaurs, pp. 97-99.1 pls. 7-10. BOONSTRA, LIEUWE D. 1898. Syllabus of lectures on the Vertebrata. 1929-1933. Pareiasaurian studies. (See Haugh- Philadelphia, University of Pennsylva- ton, Sidney H., and Lieuwe D. Boon- nia, pp. 1-135, figs. 1-66. stra.) DREVERMANN, FR. 1930. A contribution to the cranial osteology 1924. Schadel und Unterkiefer von Cyamodus. of Pareiasaurus serridens (Owen). Ann. Abhandl. Senckenbergische Naturf. Ge- Univ. Stellenbosch, vol. 8, sect. A, no. 5, sellsch., vol. 38, pp. 291-309, pl. 23. pp. 1-18, figs. 1-3, pl. 1. 1933. Die Placodontier. 3.-Das Skelett von 1932a. The phylogenesis of the Pareiasauridae: Placodus gigas Agassiz im Senckenberg a study in evolution. South African Museum. Ibid., vol. 38, pp. 323-364, Jour. Sci., vol. 29, pt. 1, pp. 480-486, pls. 1-16. 1 fig. ELFTMAN, HERBERT 1932b. A note on the hyoid apparatus of two 1929. Functional adaptations of the pelvis in Permian reptiles (pareiasaurians). Anat. marsupials. Bull. Amer. Mus. Nat. Anz., vol. 75, nos. 4, 5, pp. 77-81, 3 figs. Hist., vol. 58, pp. 189-232, figs. 1-12, BROILI, FERDINAND pls. 9-14. 1912. Zur Osteologie des Schadels von Placo- EVANS, FRANCIS GAYNOR dus. Palaeontographica, vol. 59, pp. 1939. The morphology and functional evolu- 147-155, figs. 1-4, 1 pl. tion of the atlas-axis complex from fish BROOM, ROBERT to mammals. Ann. New York Acad. 1924. On the classification of the reptiles. Bull. Sci., vol. 39, art. 2, pp. 29-104, figs. Amer. Mus. Nat. Hist., vol. 51, pp. 1-15. 39-65, figs. 1-13. FTRBRINGER, MAX BYSTROW, A. P. 1922. Das Zungenbein der Wirbeltiere insbe- 1944. Kotlassia prima Amalitsky. Bull. Geol. sondere der Reptilien und Vogel. Ab- Soc. Amer., vol. 55, no. 4, pp. 379-416, handl. Heidelberger akad. Wiss., math. figs. 1-22. naturw. Kl., div. B, no. 11, xii +164 pp., CASE, ERMINE COWLES 12 pls. 1905. The osteology of the Diadectidae and GADOW, HANS their relations to the Chelydosauria. 1882. Beitrage zur Myologie der hinteren Ex- Jour. Geol., vol. 13, no. 2, pp. 126-159, tremitit der Reptilien. Morph. Jahrb., figs. 1-20. vol. 7, pp. 329-466, figs. 1-4, pls. 17-21. 1911. A revision of the Cotylosauria of North 1920. Amphibia and reptiles. In Cambridge America. Carnegie Inst. Washington natural history, eds. S. F. Harmer and Publ., no. 145, pp. 1-121, figs. 1-52, pls. A. E. Shipley. London, vol. 8, xiii +668 1-14. pp., 181 figs. COLBERT, EDWIN HARRIS GOETTE, A. 1946. Hypsognathus, a Triassic reptile from 1899. tJber die Entwickelung des Knocherneu New Jersey. Bull. Amer. Mus. Nat. Ruickenschildes (carapax) der Schild- 323 324 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 Kr6ten. Zeitschr. Wiss. Zool., vol. 66, sauridae. Palaeont. Zeitschr., vol. 12, pp. 407-434, figs. 1-3, pls. 27-29. pp. 47-59, figs. 1-9. GREGORY, WILLIAM KING HAUGHTON, SIDNEY H. 1910. The orders of mammals, parts I, II. 1929-1933. Pareiasaurian studies. (See Haugh- Bull. Amer. Mus. Nat. Hist., vol. 27, ton, Sidney H., and Lieuwe D. Boon- pp. 1-524, figs. 1-32. stra.) 1933. Fish skulls: a study of the evolution of HAUGHTON, SIDNEY H., AND LIEUWE D. BOON- natural mechanisms. Trans. Amer. Phil. STRA Soc., vol. 23, art. 2, vii+481 pp., figs. 1929-1933. Pareiasaurian studies, parts I-XI, 1-302, pls. 1, 2. as follows: 1934. Polyisomerism and anisomerism in cra- 1929a. Part I.-An attempt at a classification nial and dental evolution among verte- of the Pareiasauria based on skull brates. Proc. Natl. Acad. Sci., vol. 20, features. Ann. South African Mus., vol. no. 1, pp. 1-9, figs. 1-3. 28, pp. 79-87. 1935a. On the evolution of the skulls of verte- 1929b. Part II (Haughton, S. H.).-Notes on brates with special reference to heritable some pareiasaurian brain-cases. Ibid., changes in proportional diameters (ani- vol. 28, pp. 88-96, figs. 21-26. somerism). Ibid., vol. 21, no. 1, pp. 1-8, 1929c. Part III (Boonstra, L. D.).-On the figs. 1-3. pareiasaurian manus. Ibid., vol. 28, 1935b. Reduplication in evolution. Quart. Rev. pp. 97-112, figs. 27, 28, pls. 6-8. Biol., vol. 10, no. 3, pp. 272-290, figs. 1929d. Part IV (Boonstra, L. D.)-On the 1-12. pareiasaurian pes. Ibid., vol. 28, pp. 1936. The transformation of organic designs: 113-122, fig. 29, pls. 9-12. a review of the origin and deployment 1930a. Part V (Haughton, S. H., and L. D. of the earlier vertebrates. Biol. Rev., Boonstra).-On the pareiasaurian man- Cambridge Phil. Soc., vol. 11, no. 3, dible. Ibid., vol. 28, pp. 261-289, figs. pp. 311-344, figs. 1-10. 4-16, pls. 32-36. GREGORY, WILLIAM KING, AND LEVERETT ALLEN 1930b. Part VI (Haughton, S. H., and L. D. ADAMS Boonstra).-The osteology and myology 1915. The temporal fossae of vertebrates in of the locomotor apparatus. A-Hind relation to the jaw muscles. Science, new limb. Ibid., vol. 28, pp. 297-366, figs. 763-765. 1-60, pl. 38. ser., vol. 41, no. 1064, pp. 1932a. Part VII (Boonstra, L. D.).-On the GREGORY, WILLIAM KING, AND CHARLES LEWIS hind limb of the two little-known CAMP pareiasaurian genera: Anthodon and 1918. Studies in comparative myology and Pareiasaurus. Ibid., vol. 28, pp. 429- osteology, no. III. Bull. Amer. Mus. 435, figs. 1-7. Nat. Hist., vol. 38, pp. 447-563, figs. 1932b. Part VIII (Boonstra, L. D.).-The oste- 1-16, pls. 39-50. ology and myology of the locomotor GREGORY, WILLIAM KING, ROY WALDO MINER, apparatus. B-Fore limb. Ibid., vol. 28, AND GLADWYN KINGSLEY NOBLE pp. 437-503, figs. 1-41, pls. 42-44. 1923. The carpus of Eryops and the structure 1933a. Part IX (Boonstra, L. D.).-The cra- of the primitive chiropterygium. Bull. nial osteology. Ibid., vol. 31, pp. 1-38, Amer. Mus. Nat. Hist., vol. 48, pp. figs. 1-5, pls. 1-22. 279-288, figs. 1-4. 1933b. Part X (Boonstra, L. D.).-The dermal GREGORY, WILLIAM KING, AND GLADWYN KINGS- armour. Ibid., vol. 31, pp. 39-48, pls. LEY NOBLE 23-27. 1924. The origin of the mammalian alisphe- 1933c. Part XI (Boonstra, L. D.).-The verte- noid bone. Jour. Morph. Physiol., vol. bral column and ribs. Ibid., vol. 31, pp. 39, no. 2, pp. 435-463, figs. 1-13, 1 pl. 49-66, figs. 1-7, pl. 28. GREGORY, WILLIAM KING, AND HENRY CUSHIER HAY, OLIVER PERRY RAVEN 1905. On the group of fossil turtles known as 1941. Studies on the origin and early evolution the Amphichelydia; with remarks on the of paired fins and limbs, parts I-IV. origin and relationships of the suborders, Ann. New York Acad. Sci., vol. 42, superfamilies and families of testudines. art. 3, pp. 273-360, figs. 1-34, pls. 1-5. Bull. Amer. Mus. Nat. Hist., vol. 21, HARTMAN-WEINBERG, A. pp. 137-175, figs. 1-5. 1930. Zur Systematik der Nord-Duna-Pareia- 1908. The fossil turtles of North America. .1946 GREGORY: PAREIASAURS AS ANCESTORS TO TURTLES 325 Carnegie Inst. Washington Publ., no. JAEKEL, OTTO 75, pp. 1-568, figs. 1-704, pls. 1-113. 1902. Ueber Placochelys n.g. und ihre Bedeu- ,HEINTZ, ANATOL tung fur die Stammesgeschichte der 1935. How the fishes learned to swim. Smith- Schildkroten. Neues Jahrb. Min., vol. 1, sonian Rept. (for the year 1934), publ. pp. 127-144, pl. 2. 3316, pp. 223-245, figs. 1-12. 1907. Placochelys placodonta aus der Obertrias HUENE, FRIEDRICH VON des Bakony. Resultate der wissenschaftl. 1913. The skull elements of the Permian Erforschung des Balatonsees, vol. 1, pt. Tetrapoda in the American Museum of 1, palaeont. suppl., pp. 1-90, figs. 1-50, Natural History, New York. Bull. pls. 1-10. Amer. Mus. Nat. Hist., vol. 32, pp. 1914. t.ber die Wirbeltierfunde in der oberen 315-386, figs. 1-57. Trias von Halberstadt. Palaeont. Zeit- 1920. Scierosaurus und seine Beziehungen schr., vol. 1, pp. 155-215, figs. 1-34, zu anderen Cotylosauriern und zu den pls. 3-5. Schildkroten. Zeitschr. f. induktive 1918. Die Wirbeltierfunde aus dem Keuper Abstammungs- und Vererburlgslehre, von Halberstadt. Serie 2. Testudinata. vol. 24, pp. 163-166, 1 fig. Teil 1. Stegochelys dux, n.g., n.sp. Ibid., 1926. Einige Schildkrotenreste aus der ober- vol. 2, pp. 88-214, figs. 1-62, pls. 2-7. sten Trias Wiirttembergs. Centralbl. f. MINER, Roy WALDO Min., div. B, no. 14, pp. 509-514, figs. 1925. The pectoral limb of Eryops and other 1-4. primitive tetrapods. Bull. Amer. Mus. 1931. Erganzungen zur Kenntnis des Schadels Nat. Hist., vol. 51, pp. 145-312, figs. von Placochelys und seiner Bedeutung. 1-104. Geol. Hungarica, ser. palaeont., fasc. 9, OWEN, RICHARD pp. 1-18, pls. 1-3. 1849. On the development and homologies of 1933. Die Placodontier. 4. Zur Lebensweise the carapace and plastron of the chelo- und Verwandtschaft von Placodus. Ab- nian reptiles. Memoirs from the Trans. handl. Senckenbergische Naturf. Ge- Geol. Soc. London, vol. 11, pp. 151-171, sellsch., vol. 38, pp. 365-382, figs. 1-5. figs. 1-9, pl. 13. PARKER, WILLIAM KITCHEN 1936a. Henodus chelyops, ein neuer Placodon- 1880. On the development of the green turtle tier. Palaeontographica, vol. 84, div. A, (Chelone viridis, Schneid.). Report on pp. 99-148, figs. 1-37, pls. 9-13. the Scientific Results of the Voyage of 1936b. Kurze iibersicht Uiber die Geschichte der H.M.S. Challenger during the Years Vertebraten. Eine graphische Darstel- 1873-1876. Zoology. London, vol. 1, pt. lung. Palaeont. Zeitschr., vol. 18, pp. 5, pp. 1-58, pls. 1-13. 198-201, pI. 10. PATTEN, WILLIAM 1938. Der dritte Henodus, Erganzungen zur 1912. The evolution of the vertebrates and Kenntnis des Placodontiers Henodus their kin. Philadelphia, xxi +486 pp., chelyops Huene. Palaeontographica, vol. 309 figs. 89, div. A, pp. 105-114, figs. 1-6, pls. PEYER, BERNHARD 3,4. 1931. Die Triasfauna der Tessiner Kalkalpen. 1940. Die stammesgeschichtliche Gestalt der III. Placodontia. Abhandl. d. Schweizer- Wirbeltiere-ein Lebensablauf. Palaeont. ischen Palaeont. Gesellsch., vol. 51, Zeitschr., vol. 22, pp. 55-62, figs. 1, 2. pp. 1-25, pls. 15-17. HUNTINGTON, GEORGE S. 1934. Die Triasfauna der Tessiner Kalkalpen. 1918. Modern problems of evolution, varia- VII. Neubeschreibung der Saurier von tion and inheritance in the anatomical Perledo. Ibid., vols. 53, 54, pp. 1-130, part of the medical curriculum. Anat. figs. 1-14, pls. 31-41. Rec., vol. 14, no. 6, pp. 359-445, figs. 1935. Die Triasfauna der Tessiner Kalkalpen. 1-18, pls. 1-17. [Shoulder girdle of VIII. Weitere Placodontierfunde. Ibid., Chelydra, fig. 6, p. 372; morphology of vol. 55, pp. 1-26, pls. 42-46. the entepicondylar foramen, pp. 390- ROMER, ALFRED SHERWOOD 392.1 1922. The locomotor apparatus of certain IHLE, J. E. W., P. N. VAN KAMPEN, H. F. NIER- primitive and mammal-like reptiles. STRASZ, AND J. VERSLUYS Bull. Amer. Mus. Nat. Hist., vol. 46, 1927. Vergleichende Anatomie der Wirbel- pp. 517-606, figs. 1-7, pls. 27-46. tiere. Berlin, viii +906 pp., 987 figs. 1933. Vertebrate paleontology. Chicago, Uni- 326 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY VOL. 86 versity of Chicago Press, vii +491 pp., 1916. On the structure of the brain-case in 359 figs. certain Lower Permian tetrapods. Bull. 1945. Vertebrate paleontology. Second edi- Amer. Mus. Nat. Hist., vol. 35, pp. tion. Chicago, University of Chicago 611-636, figs. 1-11. Press, viii +687 pp., 377 figs. 1924. The elasmosaurid shoulder-girdle and ROMER, ALFRED SHERWOOD, AND FRANK BYRNE fore-limb. Proc. Zool. Soc. London, pp. 1931. The pes of Diadectes: notes on the primi- 885-917, figs. 1-12 tive tetrapod limb. Palaeobiologica, WELLES, SAMUEL PAUL vol. 4, pp. 24-48, figs. 1-9. 1941. The mandible of a diadectid cotylosaur. ROMER, ALFRED SHERWOOD, AND L. W. PRICE Univ. California Bull. Dept. Geol. Sci., 1940. Review of the Pelycosauria. Geol. Soc. vol. 25, pp. 423-432, figs. 1, 2. Amer. Special Papers, no. 28, x+538 1943. Elasmosaurid plesiosaurs with descrip- pp., figs. 1-71, pls. 1-46, tables 1-8. tion of new material from California RucKEs, HERBERT and Colorado. Mem. Univ. California, 1929a. Studies in chelonian osteology. Part I. vol. 13, no. 3, pp. 125-254, figs. 1-37, pls. Truss and arch analogies in chelonian 12-29 and frontispiece. pelves. Ann. New York Acad. Sci., vol. WHITE, THEODORE E. 31, pp. 31-80, figs. 1-14, pls. 4-7. 1939. Osteology of Seymouria baylorensis 1929b. Studies in chelonian osteology. Part II. Broili. Bull. Mus. Comp. Zool. Harvard The morphological relationships be- Coll., vol. 85, no. 5, pp. 325-409, figs. tween the girdles, ribs and carapace. 1-30, pls. 1-3. Ibid., vol. 31, pp. 81-120, figs. 1-30. WIEDERSHEIM, ROBERT SCHAEFFER, BOBB 1907. Einfuhrung in die Vergleichende Anato- 1941. The morphological and functional evo- mie der Wirbeltiere. Jena, Gustav lution of the tarsus in amphibians and Fischer, xxii +471 reptiles. Bull. Amer. Mus. Nat. Hist., pp., 334 figs. vol. 78, pp. 395-472, figs. 1-21. WILLISTON, SAMUEL WENDELL SEELEY, HARRY GovIER 1904. The temporal arches of the Reptilia. 1887-1895. Researches on the structure, or- Biol. Bull., vol. 7, no. 4, pp. 175-192, ganization and classification of the fossil figs. 1-17. Reptilia. Parts I-IX. Phil. Trans. Roy. 1907. The skull of Brachauchenius, with ob- Soc., London, vols. 178-186. servations on the relationships of the SUSHKIN, PETER P. plesiosaurs. Proc. U. S. Natl. Mus., vol. 1928. Contributions to the cranial morphology 32, pp. 477-489, pls. 1-4. of Captorhinus Cope (Reptilia, Cotylo- 1911. American Permian vertebrates. Chi- sauria, Captorhinidae). Palaeobiologica, cago, University of Chicago Press, 145 vol. 1, pp. 263-280, figs. 1-10. pp., 38 pls., 32 figs. [Limnoscelis, fore- WATERMAN, HARRIET CUTLER limb and hand, fig. 13, p. 39; skull, fig. 4, 1929. Studies on the evolution of the pelvis of p. 29.] man and other primates. Bull. Amer. 1914. Water reptiles of the past and present. Mus. Nat. Hist., vol. 58, pp. 585-642, Chicago, University of Chicago Press, figs. 1-10, pl. 22. 251 pp., 131 figs. [Limnoscelis, restored WATSON, DAVID MEREDITH SEARES skeleton, fig. 3, p. 20.] 1914a. On the skull of a pariasaurian reptile, 1917. The phylogeny and classification of rep- and on the relationship of that type. tiles. Jour. Geol., vol. 25, no. 5, pp. Proc. Zool. Soc. London, pp. 155-180, 411-421, figs. 1-5. figs. 1-7. 1925. The osteology of the reptiles. [Ed. 1914b. Eunotosaurus africanus Seeley and the William K. Gregory.] Cambridge, Har- ancestry of the Chelonia. Ibid., pp. vard University Press, xiii +300 pp., 1011-1020, fig. 1, pl. 7. 191 figs.