Amphilbian,S AN~D'reptiles'

Home , Crurotarsal, Mystriosuchus, Protorosauria, Sauropterygia

THE M&RPHOLOGICAL AND FUNCTIONAL EVOL'UTION OF THE "TARSUS IN' AMPHIlBIAN,S AN~D'REPTILES'

BBOBB SCIHAEFFER,

.,BUI-LETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY

VOL. LX'XVIII, ART. -VIpp 395-472 New~ York I8aued November 7, 1941 4' '4 '','L'

'4X's~~''1, 'p ,-

a ' , ,,, X, X. A" ,; Ses

t 4, .4i: t , r ;ts8

, {j ii f4 Waa rNt; /0n ;) f1 iba

4' 4';'j

.4 44, 4

'4, 49 44

'4" ,"4 '4v, ;; ,;Kf,'#''t< i; t;2s,9,,th~~~~~~'1Q'b'K;oe!sI Article VI.-THE MORPHOLOGICAL AND FUNCTIONAL EVOLUTION OF THE TARSUS IN AMPHIBIANS AND REPTILES BY BOBB SCHAEFFER'

FIGURES 1 TO 21 CONTENTS PAGE INTRODUCTION ...... 396 THE AMPHIBIAN TARSUS ...... 397 The tarsal pattern of the primitive tetrapods-labyrinthodonts...... 397 The tarsus of the lepospondyls...... -...... 403 The tarsus of the caudate amphibians ...... 403 Comparison with the primitive tetrapod pattern ...... 403 The centrale complex ...... 406 Comparison of the caudate and anuran tarsus ...... 407 The relation of the tarsus to the origin of the Amphibia ...... 408 Amphibian locomotion ...... 409 Pressure distribution in the unspecialized amphibian foot and an analysis of the forces acting on the limbs during locomotion ...... 415 The extrinsic and intrinsic musculature of the amphibian hind-foot and its relation to

locomotion ...... 420

THE REPTILIAN TARSUS ...... 425 The origin of the primitive reptilian tarsus ...... 425 The cotylosaurian tarsus and its function ...... 428 The seymouriamorph tarsus ...... 428 The captorhinomorph tarsus ...... 428 The diadectomorph tarsus ...... -...... 430 The procolophonid tarsus ...... 432 The pareiasaurian tarsus ...... 433 The structural and functional derivatives of the unspecialized cotylosaurian tarsus...... 434 Type I: The structure and function of the type found in the Mesosauria, Ichthyosauria, Protorosauria, and Sauropterygia ...... 434 Type II: The structure and function of the type found in the Chelonia and the Eosuchia and the reptilian orders derived from the latter; the elaboration of the mesotarsal joint 435 The chelonian tarsus ...... 435 The eosuchian tarsus and its derivatives ...... 436 The rhynchocephalian and lacertilian tarsus ...... 438 The thecodont tarsus...... 441 The crocodilian tarsus...... 442 The locomotion of the alligator...... 443 The tarsus of dinosaurs, pterosaurs, and birds...... 446 Type III: The structure and function of the pelycosaurian tarsus and of the types derived from it; the elaboration of the crurotarsal joint...... 447 The pelycosaurian tarsus and its derivatives ...... 447 The therapsid tarsus...... 449 The dromasaurian tarsus...... 450 The dinocephalian tarsus...... 451 The dicynodont tarsus...... 452 The gorgonopsian tarsus ...... 452 The method of phalangeal reduction as illustrated by the Gorgonopsia ...... 454

The therocephalian tarsus...... 454 The cynodont tarsus...... 454 The bauriamorph tarsus ...... 454 The functioning of the therapsid foot...... 455 Astragalar superposition ...... 455 The extrinsic and intrinsic musculature of the reptilian hind-foot and its relation to locomotion...... 457 SUMMARY...... 466 REFERENCES ...... 468 1 Department of Zo6logy, Columbia University. 395 396 Bulletin American Museum of Natural History [Vol. LXXVIII INTRODUCTION The possibility of tracing satisfactorily motion in recent types in the form of mo- the phylogeny of any part of the vertebrate tion pictures, together with experiments on skeleton is naturally limited by the for- the action of individual muscles and tuitous discovery of the proper material groups of muscles that tend to act to- that will enable one to define the critical gether. Such information is of great value stages in the transition from a more primi- in interpreting and supplementing the con- tive morphological type to a more ad- clusions drawn from the fossil material. vanced. Although the true nature of the Whereas such data have been determined tarsus of many extinct tetrapod groups is for the caudate amphibian and the alliga- either unknown or at best but poorly tor, it is realized that comparable data known, either due to lack of ossification or should also be gathered for a typical preservation, there has become available lizard, a bird, and a primitive mammal in within recent years a wealth of material connection with further studies on this which makes it practicable to follow the general topic. general evolution of this region in consider- The evolution of the tarsus and its ef- able detail. fect on the method of locomotion are ob- The purpose of this paper is to describe viously intimately associated with the the morphological and the functional evolu- evolution of the crus, the femur, and the tion of the tarsus in the amphibians and pelvis. Although it has not been possible reptiles, and to present a synthetic picture within the space of this paper to integrate of the steps leading to the various defini- all of this information, the more important tive types of tarsal structure. While the changes in the crus, the femur, and the morphological discussion is based as much knee-joint are briefly considered. as possible on actual material, it has been As might be suspected, the literature necessary to resort to the literature for relating to the tetrapod tarsus (excluding several important structural stages that the mammalian) is very large. Most of are only represented by material in collec- the accounts are purely descriptive, some tions in Europe and South Africa. Al- comparative, while very few make any at- though conclusions drawn from published tempt at discussing the functional im- descriptions and diagrams may not be as plications. In the descriptive category, accurate as those based on actual speci- the work of Boonstra, Broom, Gregory, mens, it is felt that constant cross-check- Haughton, Holmgren, von Huene, Rabl, ing through papers on the same subject Romer, Schmalhausen, Sewertzoff, and but by different authors has reduced to a Watson is of the greatest importance. minimum most of the inaccuracies and From the functional point of view, the misconceptions. contributions of Elftman, Gray, Gregory, The functional aspects have been treated Hirsh, von Huene, Haughton and Boonstra, in considerable detail for the primitive Morton, Nauck, and Romer are the most tetrapod tarsus and many of the types illuminating of the very few ventures into derived from it. Some of the functional this field. The study of the mechanics of interpretation is admittedly more or less locomotion in the lower tetrapods has superficial, as it has not been possible with been sadly neglected and any information the time and the material available to in this field is very desirable. treat each important functional change The writer wishes to acknowledge his with the same degree of thoroughness. deep gratitude to Professor W. K. Gregory In studying the functioning of the tarsus for suggesting the problem and for his con- and its relation to the movements of the stant interest and guidance throughout the hind-limb and body as a whole, it is highly investigation. The writer is also in- desirable to have experimental data on the debted to Mr. H. C. Raven and to Professor angulation and movement of the various A. S. Romer for helpful suggestions, and to limb segments and on the method of loco- Mr. Howard Pearson for lending his knowl- 1941 ] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 397 edge of mechanics to many discussions ance in making the motion pictures. The on locomotion. He also wishes to thank drawings have been made by Mrs. Helen Curators Barnum Brown and Walter Granger for permission to use the vertebrate Zisa witha herusa sil canthe wie paleontological collections in the American wishes to thank her for her cooperation and Museum, and Dr. A. J. Ramsay for assist- patience.

THE AMPHIBIAN TARSUS

THE TARSAL PATTERN OF THE PRIMITIVE TETRAPODS-LABYRINTHODONTS The tetrapod tarsus may be defined as son, it is here possible to describe a recently that portion of the posterior appendage collected and very complete lower leg and that lies between the tibia and fibula on tarsus of Trematops milleri (Fig. 1). The the one hand and the metatarsals on the phalanges are unfortunately missing. The other. It is a region that has always been bones were preserved in their natural associated with the filexion and extension positions with the exception of the third of the pes on the lower leg and is clearly centrale which was slightly crowded under constructed for such movements. Due the medial border of the fourth tarsale. to this function and because of its position Because of the completeness of this relative to the body, the tarsus has been specimen a revision of Williston's descrip- constantly subjected to varying tension tion (1909) is warranted. His figures are and compression forces throughout its somewhat diagrammatic owing to the fact history. These forces, the result of the ef- that the borders of the bones of his speci- fect of gravity and of locomotion, are in- men were not well preserved and, also, be- timately associated with the modifications cause the tarsus is lying across several of that have occurred throughout its evolution. the vertebrae, somewhat distorting the Although the oldest known amphibian natural position of the elements. Some tarsus already possessed the basic tetrapod of the corrections appear significant when pattern, it is quite evident that it retained the functioning of the foot is considered. the fundamental features of the rhipidistian The fibulare is proximodistally elongated pelvic fin. The convergence of all the with its greatest transverse width at the tarsal elements toward the fibula, except level of the articulation between the inter- those in line with and more distal to the medium and the fourth centrale. It is tibia, is without question a heritage char- narrowest at the point of articulation with acter of great significance, derived from a the fibula. The distal border is somewhat similar convergence of all the radials, ex- v-shaped, forming separate facets for con- cept the first preaxial one, toward the sec- tact with the fourth and fifth tarsalia. ond mesomere (Gregory and Raven, 1941). The medial and lateral borders are thick- A consideration of the primitive am- ened and rounded. phibian tarsus is a necessary prerequisite The intermedium, also, has its greatest- for the interpretation of this region in all length proximodistally rather than the re- higher tetrapods. Although the earliest verse as indicated in Williston's figures known tarsus is that of the embolomere (1909, 1913). The central portion of the Pholidogaster (Watson, 1926b) from the dorsal surface is depressed. The distal Lower Carboniferous of Scotland, it is by half of the medial border is elevated above no means completely preserved and for the rest of the bone and is rounded, giving that reason will be discussed later. At every indication of having made a ligamen- present, the only completely known Paleo- tous attachment with the lower end of the zoic amphibian tarsus is that of the tibia. The proximal half of this border rhachitome Trematops. is free and rounded and is not in contact Through the kindness of Dr. E. C. 01- with the tibia. (Williston considered the 398 Bulletin American Museum of Natural History [Vol. LXXVIII only with the tibia, and distally, with the )19'$ll:\\->first$ centrale and possibly the proximo- 1 hll))ymediall \g corner of the second centrale. The fourth centrale (proximal centrale lV1/, of Schmalhausen and others) is, as Willis- (l ton points out, one of the largest bones in F \~\XTl the tarsus. It is transversely widened and l i 1 \ !A not rectangular as previously figured. The proximal border is very slightly concave for articulation with the intermedium. The proximomedial corner is rounded, and probably it contacted the tibia. Laterally, centrale four articulates with the fibulare, and medially, with the tibiale. The distal edge of the bone is decidedly v-shaped, dividing this articular surface into two distinct parts. The lateral surface articulates both with the centrale three and a /JrIqr-cornerPt of tarsale four, while the medial /t{}7'

A b T q22u-St Tntarasec&s Ambystoma Id Ambystomoiea

OermeynaX { / | Cryptobranchas Pleyalobatrachus

Ranoon //ynob&s /

Cryptobranc/zo:iaecz CAUDATA /~~~~~~I,SALLEV/V/A tOoML.j \dSsaurasoru\L\

2527ff IA Microsauria Nectrndza \LEFOSROADLYL/

Archegosaas Trematops ZABYVINTHODOA/TA AknZ;sa Wet Fig. 2.-The phylogeny of the amphibian tarsus. Stippled areas, cartilage; unstippled areas, bone. 402 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 403 (Fig. 2). The chart is otherwise constructed Boyden and Noble (1933), and Romer according to the views of Noble (1931), (1939).

THE TARSUS OF THE LEPOSPONDYLS The Lepospondyli seemingly may be, at size of the tibiale and its gradual migration present, divided into three suborders to a position directly under the reptilian (Romer, 1940): the Nectridea, Micro- astragalus. There is no real evidence, sauria, and Aistopoda, the latter lacking however, either embryological or paleon- limbs completely. But one nectridian tological, that any of the bones of the tar- tarsus is known, that of Scincosaurus sus have migrated to new positions, the (Jaekel, 1909). It is completely ossified. one exception to this possibly being the The fibulare and the intermedium are about raising of the astragalus onto the cal- the same size and are the largest bones in caneum. Furthermore, there is every rea- the tarsus. The tibiale has apparently son to believe that the navicular is com- suffered a relative decrease in size. The posed of one or more centralia, and as a centrale series has presumably completely corollary the latter could not be absent in disappeared, as has the fifth tarsale. the reptilian ancestor. Broom is of the opinion (1921) that the A single ossified microsaurian tarsus is foot of Scincosaurus (Fig. 2) is structurally known, that of Hyloplesion (Steen, 1938). ancestral to the primitive reptilian foot. It is not possible, from the published fig- There is, however, very little evidence to ures of the specimen, to make a restora- support this conclusion, particularly since tion (Fig. 2). It is obvious, however, that the nectridians are apparently aberrant the foot bears a close resemblance to the and not ancestral to any later group. primitive pattern, including the presence Broom believes that the mammalian na- of centralia. It might very well be an- vicular has developed from a displaced cestral, therefore, to the caudate amphibian tibiale, and he constructs a series to show tarsus, which in turn, as will be seen, is the supposed progressive reduction in the very similar to that found in Trematops.

THE TARSUS OF THE CAUDATE AMPHIBIANS COMPARISON WITH THE PRIMITIVE TETRAPOD PATTERN On the whole the caudate amphibian Baur (1885, 1886, 1888) studied the ele tarsus has changed but little from the ments of the caudate tarsus in great detail, primitive ground pattern. In some forms including the few fossil forms known at the there has been a considerable degree of time. Some attention is paid to varia- fusion between the various tarsal elements, tion. He first recognized the presence of but in most cases the embryology reveals four centralia (1886) as the primitive num- the distinct origin of the parts concerned. ber. Much work has been done on the caudate Emery (1898) discussed the morphology foot, and the following papers appear most of the carpus and tarsus at great length. significant in this connection. He realized the significance of the Arche- Gegenbaur (1864, 1876) believed that gosaurus foot in interpreting that of the the primitive tetrapod tarsus contained caudate amphibia. Emery believed that one or two centralia, basing his conclusion the prehallux (or prepollex) is homologous on a study of recent caudate amphibians in the Amphibia, Reptilia, and Mammalia. only. He pointed out for the first time He made a distinction between the propo- the tendency of all the tarsal elements, ex- dium, made up of elements in line with, cept those running out from the tibia to and distal to the tibia, and the mesopo- the first digit, to converge toward the dium, consisting of the elements converg- fibula. ing on the fibula. 4044Bulletin American Museum of Natural History [Vol. LX-XVIII

Zwick's (1898) thorough work includes fourth and the fifth tarsalia uniting to form comparative and embryological studies. the cuboid and the two centralia forming Archegosaurus is reconsidered, but not too the head of the astragalus, the body of the successfully, since Trematops was then un- latter being composed of the intermedium known. In the discussion on the centrale and the tibiale. Although such conclu- complex, it is pointed out that there is no sions may ultimately prove to be correct, paleontological evidence of more than two Tornier, by ignoring the paleontological centralia (except abnormally). Unfor- evidence as to the ancestral and the inter- tunately, the Hynobiidae were not investi- mediate stages between the primitive and gated. the more advanced types, lacks perspec- Sewertzoff (1908) made a very compre- tive with regard to which elements were hensive investigation and survey of the fused, when and under what conditions literature on the carpus and tarsus of both they fused, and which elements were lost. recent and fossil amphibians and reptiles. Steiner's first work (1921) is a lengthy He pointed out the individual variation discussion of the phalangeal formulae of that occurs in the centrale region of the various recent Amphibians, some of which caudate tarsus and of the resemblance of are incorrect (Gregory, Miner, and Noble, this type to the tarsus of Archegosaurus. 1923). But little attention, however, is Rabl's (1910) lengthy treatise, so far as paid to the tarsus itself. The fact that the amphibian foot is concerned, is simply variation in the phalangeal formula is a review of previous work. The reptilian recognized is an important contribution. tarsus is covered in greater detail. He In his 1934 paper, Steiner compares one of maintains his previous conclusion (1901) the variants of the Cryptobranchus tarsus, that the Proteus tarsus, consisting of three having four centralia, with that of Tre- elements, is the ancestral type. This is, matops. As pointed out before, such com- of course, now entirely rejected since the parisons, while suggestive, are not valid. degenerate tarsi of Proteus and of Am- Dollo (1929) discussed the forms having phiuma are obviously associated with the various numbers of centralia and reached degenerated condition of the limbs in these the conclusion that four is the primitive exclusively aquatic types and since their number. He claimed that if the primitive derivation from more normal salamanders pentadactyl tetrapod limb is considered in is now generally accepted. the light of a numerical progression, the Schmalhausen's study (1910, 1917) of presence of four centralia is inevitable. the development of the tarsus of a primi- This is also the case when the segments of tive member of the Caudata (Hynobius), the limb are considered as alternating. together with a study of the adult tarsus There is no rational basis for applying such of Ranodon in the light of the Trematops a mathematical analysis to the hind limb. foot indicates, without doubt, that the The arrangement of the tarsal elements ini latter is at least structurally ancestral to the primitive tetrapod is obviously not in the caudate tarsus. accordance with such a plan because they Tornier (1927) is one of the few to at- are not arranged in transverse rows as the tempt a functional explanation of the evolu- theory requires but, excepting the tibiale, tion of the foot. He points out that the the first centrale, and the pretarsale, all main functional tarsal joint in the Am- the elements converge toward the fibula. phibia lies between the metatarsals and Howell (1935b) points out the very tarsalia, and that most of the tarsus is significant fact that there is, up to the pres- functionally part of the lower leg. He ent time, no proof that the prepollex and considers fusion of the centralia to be a postminimus of the manus ever occurred characteristic terrestrial modification, con- as true digits. The same holds for the stant in higher terrestrial animals. Tor- prehallux and postminimus of the pes. nier sees in certain fusions in the recent He makes the excellent suggestion that urodele foot the origin of several of the these terms are misleading and that pre- elements in the mammalian tarsus; the carpale and postcarpale (pretarsale and 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 405 post-tarsale) are much more accurate, in those forms in which there has been that is, the pretarsale of Trematops. fusion in the proximal tarsal row, that is, Holmgren's work (1935, 1939) is impor- in Necturus, Ambystoma, Proteus, and Am- tant mainly in that it supports the theory phiuma. of the dual origin of the Amphibia, a These changes, however, are not neces- thorough discussion of which is beyond the sarily significant, for they do not alter the scope of this paper (see p. 408). Of more structure of the tarsus so profoundly that pertinent value here is his emphasis on the it suffers any change in function. They fact that the caudate amphibian tarsus is are simply genetic differences which are subject to considerable variation and that actually subject to much specific variation. many separate rudiments are concerned in The extensive fusion in the tarsus of Pro- the ontogeny of the tetrapod tarsus. teus and Amphiuma is most probably as- The sporadic occurrence of ossification sociated with the reduction in the number in the amphibian tarsus is of great in- of the digits. Terrestrial locomotion in terest. But few of the well-known tarsi, these exclusively aquatic forms consists fossil or recent, are completely ossified in mainly of lateral undulations of the body, the adult stage, in most cases being partly the legs being relatively too small to be ef- or completely cartilaginous. There is no fective propellers. reason for believing that the amount of ossi- The phylogeny of the amphibian tarsus fication is related to the aquatic or the demonstrates very well that the primitive terrestrial habits of the animal or to its pattern has been retained among the cau- size or weight, but rather that it is en- date amphibia with but little modification. tirely genetic. This problem is worthy of This, together with the fact that the pat- further investigation with regard to the tern is similarly modified in distantly re- variation in the amount of tarsal ossifica- lated forms, for instance, Proteus and Am- tion within a species, the constancy of ossi- phiuma, demonstrates that it would be fication of the same tarsal elements within impossible to build up a phylogeny of the a species, and finally the possibility of ex- Amphibia on the basis of the tarsus alone. perimentally altering the extent or degree Noble (1931, p. 466) has pointed out that of tarsal ossification. a common mode of generic evolution in the The tarsus of the more primitive of the Amphibia may be associated with the de- caudate amphibian families (Hynobiidae velopment of parallel modifications in the and Cryptobranchidae), as well as of some derived genera of a widely dispersed stock. of the members of the more specialized This hypothesis is certainly supported by Salamandroidea and Ambystomoidea, re- the modifications in the tarsus. Of course, sembles the rhachitome tarsus in having the anuran tarsus is strikingly -different, the fibulare and intermedium articulate while the reduced tarsi of Proteus and Am- along their proximal borders only with the phiuma are clearly derivatives of the fibula, which, in turn, is much shorter than Trematops pattern. On the other hand, the tibia. The shorter fibula is, as was the tarsus of such forms as Triturus and previously pointed out, a primitive char- Ambystoma, even allowing for variation, acter. are closely similar to the Permian am- The articulation of the intermedium phibian pattern. with the tibiale is primitively weak or non- As pointed out before, the proximal and existent, as in Trematops. Among the distal regions of the tarsus are the most caudate amphibia, this articulation be- constant. The tibiale, intermedium, and comes much stronger, as in Cryptobranchus fibulare are always present, the interme- or Salamandra. In some cases, however, dium fusing with the fibulare in some forms, there is a relative increase in the size of the such as Amphiuma and Necturus. While tibiale accompanied by a loss of all but the intermedium, as will be discussed one member of the centrale series. This later, gives every evidence of having fused results in the leveling off of the joint be- with a centrale and possibly the tibiale in tween the lower leg and the tarsus, as in the reptiles, it never does so in the caudate Triturus. The same condition also exists amphibia. 406 Bulletin American Museum of Natural History [Vol. LXXVIII THE CENTRALE COMPLEX fibula with the fibulare and the interme- --There has been much speculation as to dium, and the convergence of. the-tarsal ele- the exact number of centralia in the primi- ments toward the fibula. Since the num- tive tetrapod tarsus. As far as urodeles ber of centralia present in a given tarsus are concerned, Zwick, Rabl, and Holmgren has no apparent effect on the locomotion, are of the opinion that two is the basic it would appear that there has been no number. The latter claims, on embryo- selection toward a constant number. logical evidence, that in all cases never Whether this variation also existed in the more than two rudiments develop and lepospondyl or rhachitome tarsus is, of that extra centralia merely represent buds course, unknown, but it may have been from these rudiments. more extensive than we now have reason Not only is there variation in the num- to believe. ber of tarsal elements present in the feet All available evidence indicates that the of the different caudate amphibian sub- Hynobiidae are the most primitive of the orders and families, but there is a signifi- caudate amphibians. One member of this cant amount of variation in a number of family, Ranodon, has, in the adult, three different species. This variation is particu- and sometimes four centralia that ap- larly common in the centrale region, as parently have independent rudiments in Holmgren (1939) has pointed out, and the embryo (Schmalhausen, 1917). Holm- there may be as few as one or as many as gren attributes these "extra" centralia to four or six centralia in certain species; a larval adaptation for broadening the specifically, Ambystoma sp.? (Siredon pisci- foot, and, as in the other cases, fails to see formis), Ambystoma opacum, Hynobius any atavistic tendencies. He does not kayserlingii, Cryptobranchus alleganiensis, take into account, however, the phylo- and Megalobatrachusjaponicus. genetic position of the Hynobiidae. Steiner, as pointed out elsewhere, con- There can be little question that the siders the supernumerary centralia to be tarsus of Ranodon is very close to that of phylogenetic importance, since in cases found in Trematops. The element labelled where they are present the foot resembles "im" by Schmalhausen, "y" by most that of the labyrinthodonts. Holmgren, workers, and by some the prehallux or pre- on the other hand, claims that this simi- tarsale, is clearly homologous with the so- larity exists in only a few cases and hence called first centrale as labelled by Williston lis of no atavistic significance. He agrees for Trematops, and is possibly present as a with Schmalhausen that the supernumerary small nodule in Archegosaurus. Although centralia are separated rather late in em- the third centrale and fourth tarsale are bryonic development from normal rudi- usually fused, individuals are known in ments, and always on the fibular side (Fig. which they are separate, resulting in four 12,A). Holmgren further believes that centralia, just as in Trematops. A speciali- the primitive condition for urodeles is that zation independently acquired in the Hy- in which there are two centralia (in some nobiidae is the fusion of the first and sec- cases fused into one). ond tarsalia. The principal criticism of Holmgren's The fact that only two centrale rudi- work, and for that matter Steiner's, is the ments appear in the tarsus of most uro- great emphasis placed on structural detail deles is no indication that there has been in comparing the caudate and labyrintho- a secondary segmentation of the centralia dont tarsus. A more valid comparison in the individuals having a greater number. It is just as reasonable to suppose would seem to be centered about the deep- that Trematops had two rudiments from seated resemblances in the tarsal pattern which the "supernumerary" centralia in the Caudata and the Labyrinthodontia, budded. In other words, throughout the regardless of whether one or four centralia Amphibia but two primary centrale rudi- are present. The two most striking char- ments may appear, and in those forms -acters are the constant articulation of the which always or occasionally have more 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 407 than two, the condition may be brought to many false conclusions and worthless at- about by budding. The only reasonable tempts to homologize the bones and mus- alternative is that certain centralia have cles of the hind- and fore-limbs. There is indistinguishably fused with each other or no reason to suppose that a first centrale with other tarsal elements, a condition must be present in the tarsus in order to existing in the earliest embryological stages. permit the development of the second and The development of a two-rowed tarsus third centralia, or that these streams of on the medial side of the foot in certain prochondral cells "issue" from the first caudates and lepospondyls is quite possibly centrale. They simply represent concen- associated with the position assumed by trations of cells that develop in situ in a the foot during propulsion. Through the manner resembling their development in action of certain muscles at the beginning the manus. The element called the first of propulsion, the long axis of the foot ap- centrale by Holmgren, the fourth centrale proximately parallels the long axis of the by Williston, may still be regarded as body. This position tends to cause the principally the tibiale. There is no con- tarsal elements on the medial side to be vincing evidence of the presence of the crowded together and those on the lateral first centrale. Schmalhausen and Holm- to be separated (Fig. 3). Such crowding gren agree that the fourth tarsale and the would produce strain on the medial side, first centrale (C4 of Williston?) fuse with tending to favor both the elimination of the fibulare. In Raina; according to the first centrale and possibly the other Holmgren, there is a fifth'tarsal&'-thatJ so modifications in the centrale region. fuses with the fibiulare. A centrale apparently fuses with the third'tarsale, while COMPARISON OF THE CAUDATE AND the first and second tarsalia ossify sepa- ANURAN TARSUS rately. With the y-element of Holmgren The embryology of the anuran foot has and others equal to Williston's first cen- been investigated by Tschernoff (1907), trale, there are, according to this author, Schmalhausen (1908a), Steiner (1921), five centralia in Anura and certain of the and Holmgren (1933). All agree that higher forms. The interpretation and there is no sign of the intermedium, al- identification of the anlagen in the anuran though the latter suggests that it may be embryonic tarsus are very difficult in view present as a bridge between the proximal of the obvious specialization from the ends of the tibiale and fibulare. With re- very moment at which the rudimen's are gard to the centralia, Holmgren makes laid down. what would appear to be some question- This discussion suggests that all the able homologies between the fore- and tarsal elements may be accounted for in hind-limb. In the manus of Pelobates the the Anura, except the intermedium, first centrale anlage has attached to it two Steiner being the only one possibly demon- streams of prochondral cells in which the strating its presence (1921). Thee early second and third centralia develop, respec- indistinguishable fusion of this element tively. In the pes (Fig. 12,B) similar with others must represent a profound and streams arise from the distal end of the ancient modification since its persistence tibiale, also giving rise to the second and along with the tibiale and fibulare is so third centralia. Holmgren argues that characteristic of all the other amphibian these streams could not arise from the first orders. The Lower Triassic preanuran centrale in the hand and from the tibiale in Protobatrachus (Piveteau, 1937) unfortu- the foot, and, therefore, that the tibiale nately sheds no light on the subject. Two must really include the first centrale. In elements, which are very presumably the fact he considers this element to be mostly tibiale and fibulare, are somewhat elon- the first centrale. gated, foreshadowing the anuran condi- Needless to say, the remarkable parallel- tion, but there is no indication that they ism in development and adult morphology were united, although they may have been between the manus and pes has given rise through proximal and distal cartilages. 408 Bulletin American Museum of Natural History [Vol. LXXVIII

There are two small ossicles between the THE RELATION OF THE TARSUS TO THE crus and the elongated elements that are ORIGIN OF THE AMPHIBIA thought to be sesamoids by Piveteau. The comparison between the caudate The possibllity must also be considered and the anuran tarsus quite naturally that they are the last remnants of the in- leads to the question of the diphyletic termedium and proximal centrale. origin of the Amphibia. Considerable Without going into the matter in greater evidence has been amassed (Graham- detail, there is little question that the Kerr, 1907, 1932; Save-Soderbergh, 1934; Trematops pes is very unfit for jumping Kindahl, 1938; and Wintrebert, 1910, and that much consolidation must have 1922), which has been interpreted as in- occurred during the evolution of the an- dicating that the Caudata are closely re- uran tarsus to create a lever arm that lated to the Dipnoi, while the Anura and could act as a single unbending unit. the extinct amphibian orders are sepa- The loss of an independent intermedium rately derived from the Crossopterygii. If and the losses and fusions of the centralia this be true, the Trematops type of tarsus and tarsalia are further expressions of cannot be considered as ancestral to that this consolidation. It would appear that of the urodeles, but only to that of the they are of no functional importance, ex- Anura and Reptilia. The plan of the cept possibly as "bearings" in the tar- tarsus is so very similar in the labyrintho- sometatarsal joint. Noble (1931, p. 243) donts and the Caudata, particularly the considers the reduction in the distal ele- Hynobiidae, that the possibility of this nients of the tarsus a, consequence of the resemblance being due to parallelism is very elongation of the tibiale and fibulare. difficult to visualize. If this is so, it is The persistence of these free elements on even more remarkable than the parallelism the tibial side and their fusion on the between the manus and the pes. fibular may be simply an expression of Holmgren (1939, p. 120) lists nine "most their mechanical importance on the medial striking" differences between the urodele side. and the anuran hand and foot. It is The anuran tarsus is a striking example practically impossible to evaluate these so- of uniformity in development brought called differences, first, because there are about as a result of localized function at so many gaps in the evolutionary record the tarsometatarsal joint. While the tar- of the foot, and secondly, because of the sus has been greatly modified and re- lack of certain experimental evidence. duced, there is very little variation in the How significant, for example, are hetero- elernents present. In other words, the chronic differences in the development of anuran tarsus is a region of great mova- the manus and pes as a whole, or in the bility with each element serving a defi- order of development of the fingers. Such nite mechanical need. As a result, there differences are to be expected in groups so tends to be a lack of great structural varia- far apart as subclasses, and in view of the tion. This is the situation in most rep- specialization of the Anura it would appear tiles and all mammals and for the same rea- precarious to consider these differences of son, the normal variation being of little functional significance. such great basic importance. This sort of The resemblance between the early em- reasoning also holds true when consider- bryonic stages of the anuran and caudate ing the fact that the urodeles always have foot is very striking. In both cases the four digits in the manus instead of five. tibiale and prehallucial elements arise to- The ontogeny of both the urodele and gether in a separate ray. In any case, if the anuran tarsus definitely foreshadows the anuran tarsus appears to be derived the adult condition. The modification of from the labyrinthodont type, as many the anuran tarsus for a leaping habitus is now believe, there is certainly more evi- clearly seen early in ontogeny; for in- dence for believing the same for the stance, the two proximal tarsal elements caudate amphibian tarsus. are elongated when they first appear and 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 409 very possibly the more ancestral-like demonstrated on other grounds and, if stages, which might have appeared during generally accepted, will of course modify development, have been greatly condensed the theories on the origin and the evolution or even eliminated. That the ontogeny of the tetrapod limb. Romer (1940) has of the anuran tarsus could be so modified stated the more reasonable view that the is very probable when the hundreds of Caudata are descended from the Lepo- millions of years at the disposal of the spondyli, more specifically from the gym- group for its evolution are considered. narthrids. He considers the great number It appears quite evident that the diphy- of characters common to all Amphibia as letic origin of the Amphibia will have to be ample proof of their common origin.

AMPHIBIAN LOCOMOTION Neither a study of the osteology nor of assumed by the Triturus limbs during loco- the myology of the amphibian hind-limb motion. A model of the pes of Trematops will suffice to describe exactly how the has also been made with the bones loosely limbs are used during progression, or set joined together with rubber cement and the stage properly for evaluating the latex to simulate the binding action of the changes that took place during the origin ligaments and muscles. By these means, and evolution of the reptilian foot. Romer it can be demonstrated with a reasonable (1931, 1940) has made the observation degree of certainty exactly how the recent that the oldest known tetrapod footprints caudate amphibia-and the labyrinthodonts point forward, as they do in all modern operated the entire hind-limb. Amphibia, during locomotion (Fig. 7). The extrinsic flexor muscles of the foot This important point has not been gen- arise, with minor exceptions, from the erally realized, as most mounts of Permo- fibular shaft and insert in the tarsal region Carboniferous tetrapods will testify. on the tibial side. Their action is not only The error is readily understood as there is to flex the tarsus, but to turn the foot for- nothing in the bony construction of the leg ward (adduct) to a slight extent just before or particularly in the foot which would it is placed on the ground at the end of the indicate such a position. recovery phase and to maintain -it in this With the evidence of the footprints position throughout the propulsion phase. demonstrating beyond doubt that the At the beginning of propulsion, however, labyrinthodont feet, like those of the the forward position of the crus auto- caudates, were pointed forward during matically places the foot in a forwardly locomotion, the disposition of the crural directed position. This is a function of and tarsal elements during propulsion be- the muscles operating on the crus rather comes of interest as the true solution to than on the foot. Hence it would appear the problem, particularly since it is very that the extrinsic foot flexors, as will be evident that, because of its terminal head, pointed out, are more concerned with the femur can only move in a horizontal maintaining this position during propul- plane. By manipulating the limbs of sion than in actually initiating it. cleared and stained specimens of, for in- In order that the foot may remain in stance, the caudate Triturus, such observa- this position while the femur is moving tion is possible. The diagram (Fig. 3) backward, the tibia is forced to move represents an attempt at a comparison of around the fibula in such a manner that it the skeletons of a typical caudate and crosses diagonally in front of the fibula Trematops posed as if in motion. The (cf. relations of tibia and fibula in Watson's posture of the limbs of Trematops is based restoration of Diplovertebron, 1926a, Fig. on casts of the fore- and hind-limb ele- 31). The obvious result of this action is to ments and girdles of Eryops, which were shorten the distance from the knee to the posed, without difficulty or violation to foot on the tibial side by throwing, so to the articular surfaces, in the same positions speak, the tibia out of line, into a diagonal A

Fig. 3.-A comparison of a typical caudate and a labyrinthodont (Trematops), showing the positions of the limbs during locomotion. The left fore-limb and the contralateral hind-limb are at the beginning of the propulsive phase, the other two about ready to enter the recovery phase.

410 1941 ] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 411 instead of a vertical position, permitting, This forward position serves an impor- in turn, the foot to remain forwardly di- tant purpose in that it orients the long axis rected. There is also some separation of of the foot so that it is parallel to the long the tarsal elements on the fibular side and axis of the body and hence to the direction a close approximation of the elements on of movement, thus allowing the muscles the tibial side and also some folding of the of the lower leg and foot to operate under preaxial tarsal series laterally (cf. discus- the greatest possible mechanical advantage. sion of flexion in Trematops tarsus). Only with the foot in this position can the The position assumed by the tibia pro- muscles impart power to the propulsive duces considerable torsion on the medial stroke. side of the foot, particularly at the place Morton (1926), from his study of the where it articulates with the tarsus. As footprint of Thinopus, believed that the the femur moves backward, it attempts to hind-foot of a primitive tetrapod was di-

Fig. 4.-Diagrams (after Morton, A, B, 1935; C, 1926) showing the supposed distribution of stresses in the primitive tetrapod tarsus if the long axis of the foot is at right angles to the long axis of the body. There is a shift from the postaxial border (A) at the beginning of propulsion to the preaxial border (B) at the end. In C the stress is considered to be localized along the postaxial border throughout propulsion.

rotate the tibia counter-clockwise in the rected outward during the greater portion case of the left limb. This action is op- of the propulsive stroke (Fig. 4). The posed, on the other hand, by the action of chief evidence, however, for this assump- the extrinsic flexors and the weight of the tion is possibly the footprint of Thinopus body bearing down on the foot, both of itself. Morton believed that the footprint which tend to maintain the forwardly di- belonged to a pentadactyl form which rected position of the foot. maintained its hind-limb in the above posi- In some of the cleared specimens of tion during locomotion. The resultant Triturus it is apparent that the torsion is so pressure on the fibular side, he believed, great in the proximomedial region of the forced the second and third digits into an tarsus that the first tarsale has been forced elevated position so that they would not plantarward, out of the plane of the other make prints. This ingenious explanation tarsal elements, thus narrowing the medial is, however, only based on one footprint. border of the tarsus. That this footprint is exceptional is indi- 412,412 Bulletin American Museum of Natural History [Vol. LXXVIII cated by the presence of hundreds of early quence of limb movement is right fore- tetrapod footprints (although from a later limb, left hind-limb, left fore-limb, and horizon) of normal tetrapod type. In a right hind-limb. Gray states that during more recent publication, Morton (1935) rapid locomotion the protractor and re- is of the opinion that the primitive position tractor phases take about the same period of the foot during propulsion caused the of time, but that in slow progression the functional stresses to be concentrated at time required for the retractor phase is first along the postaxial border and then relatively increased. toward the end of this phase to be trans- The movement of the left hind-limb may ferred transversely across the tarsus to the now be described in detail. The retraction preaxial border. or propulsion phase begins immediately on The propulsion and recovery movements the contact of the plantar surface of the of the lower leg and foot have been care- foot with the ground. - As will be explained fully investigated for this study in several later, the flexor musculature pushes the salamanders by means of motion pictures. foot against the ground, thus making of the Ambystoma maculatum, a relatively terres- crus and pes a kind of stationary pivot on trial salamander, and Triturus pyrogaster, which the femur rotates, and the net result almost exclusively aquatic, were used as of hind-limb movement is to move the subjects for the motion pictures. It was body forward. The body is always more found that there is no major difference in or less concave on the side on which the the locomotion of the two, even as far as hind-limb is just beginning the propulsive the details of limb movement are concerned, phase. This curvature actually decreases the only noticeable difference being the the length of the body on that side and, greater lateral undulation of the body in hence, increases the length of the stride Ambystoma because of its relatively greater of the hind-limb (see Watson, 1926a, length. Noble (1931, p. 92) has observed p. 202). that there are no functional changes in the As the body moves forward, the leg, locomotion of salamanders which have lost relatively speaking, moves backward, prin- the outer digit of the hind-foot. cipally through the action of the coccygeo- The pictures were made at 64 f.p.s. with femoralis complex. As this occurs, the the subjects walking over a thin sheet of foot itself gradually rolls off the ground, plate glass. A mirror was set up at an flexure occurring at each one of the joint angle of 45 degrees to the glass so that the planes mentioned earlier. The foot re- animal could be photographed simultane- mains in contact with the ground until the ously from two aspects. This arrangement last phalanx of the longest digit has been made it possible to study the positions of turned plantar-side up, thus producing the the limbs in detail at any stage during the greatest possible propulsive effort. Once locomotor cycle both with regard to the the tarsal region is off the ground, the range of lateral and of vertical movements. dorsal surface of the foot begins to turn The series of diagrams (Fig. 5) has been laterally until, by the time all the digits carefully drawn from one of the films by are free, the foot is in a vertical position. projection. It represents selected stages The recovery phase of the limb is com- during one-half of a normal locomotor paratively complicated, as the limb not cycle for Triturus. A complete cycle would only has to sweep through almost 180 de- show the propulsion and recovery phases grees but must also bring the foot itself for each limb. into the proper position to meet the ground The pattern of movement here illus- for the next propulsive stroke. The thigh trated is found in most tetrapods (Gray, musculature, in particular the pubo-ischio- 1939). When the right fore-foot is for- femoralis internus, is concerned with the ward the left hind-foot is also in that posi- forward movement of the entire limb. The tion, and vice versa, when the left fore- crural and intrinsic foot musculature brings foot is forward the right hind-foot is also the foot itself into position. The upper in this position; in other words, the se- and lower portions of the leg are brought 0

0~~~~~~

CA)

\i.K (('6>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i0 I'.~~~~~~~~~~~ 414 Bulletin American Museum of Natural History [Vol. LXXVIII

ment of the upper leg, plus the flexure of the crural portion, the limb is very rapidly brought into the forward position by a sort of crawl-stroke movement (Fig. 6). At about the stage when the knee is flexed, the foot extensors begin to operate and the foot is unrolled. Other muscles, which will be discussed shortly, act on the pes, bringing it into a position parallel with the body. With the foot brought forward, the digits are abducted and the foot is very rapidly lowered to the ground. Through- out the protraction and retraction phases, the extreme suppleness of the foot is very noticeable. This allows for some variation in the placing of the foot during the cycle, particularly when the animal is moving slowly. Gregory and Camp (1918) have pointed out that the Permian reptiles had permanently crooked knee-joints. This also appears to have been true for the labyrinthodonts (Fig. 3). The distal articular surface of the femur of Eryops, for instance, is more ventral than terminal. In fact this surface is so located that it was impossible to extend the lower leg beyond making an angle of about 135 degrees with the femur. This condition of permanent flexure must have been maintained by the flexor muscles of the crus inserting on the tibia and fibula, namely, the pubo-tibialis, the pubo-ischio-tibialis, the flexor tibialis internus and externus, and the ilio-fibularis. The permanent flexure would modify the movement of the limbs during locomotion somewhat from the above description. The length of the stride would not Fig. 6.-The movement of the hind-limb of a as and limb as a could caudate amphibian during propulsion (1-2) and be great the whole recovery (3-6). not reach as far forward relatively at the end of the recovery phase. The length of forward as a unit until they are just be- the propulsion phase would also be shorter yond a position at right angles to the body. because the leg could not straighten out At about this position there is flexion at the toward the end of the phase. Lastly, the knee. arc described by the leg during protraction Up to this point, the foot has remained would be much closer to the body and, of directed backward since there has been no course, much closer to the ground instead active participation on the part of the ex- of rising above the acetabulum, as in the tensors. The digits are adducted as soon Caudata. The pes itself must have moved as the foot leaves the ground and remain so in much the same manner as in the fore- until just before the beginning of protrac- going description. In a heavy animal tion. With the continued forward move- such as Eryops, however, the tarsal ele- 19411] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 415 ments are much more cuboidal than in ground. This is also true of the material Trematops. This would indicate a more from the Grand Canyon (Gilmore, 1926, extensive articulation between the bones 1927, 1928), with the exception of Batra- and consequently a reduction in the flexi- chichnus obscurus (idem, 1927, p. 41, bility of the foot. Thus, it follows that Fig. 17). In this case there is a distinct the lighter labyrinthodonts were able to path, representing the drag of the body, move their limbs, especially the pes, very between the prints of the right and the much as do the Caudata, while the heavier left limbs. Nevertheless, there is distinct types had restricted movement with less evidence that most of the heavy forms effective propulsion. It is interesting to must have been able not only to overcome note in this connection, however, that the the inertia of the body with the limbs, but int3rmedium of Eryops has a rounded tib- also to overcome gravity to the extent of ial facet (Romer, personal communication)l actually raising the body away from the thus increasing the flexibility and fore- ground during locomotion as in the Cau- shadowing the condition in the cotylosaurs. data. The body-dragging stage, if it ever Although the femur of the primitive existed, must represent a part of the lost tetrapods undoubtedly had a greater de- record between crossopt and true am- gree of freedom than the humerus, it is phibian. doubtful that it was directed relatively Gray (1939, P1. iv) has demonstrated further forward than the humerus at the by placing anurans, in which the spinal beginning of propulsion. During retrac- cord was cut just behind the brain, in con- tion, in order to maintain the body on an tact with a slowly revolving kymograph even keel, the humerus and femur must drum, that they exhibit the same sort of move through essentially the same dis- ambulatory movement as the caudates, tance or, in other words, swing more or apparently showing no inclination to less in phase with each other. Hence, leap. This has also been observed by the although the right hind-limb in the Tre- writer in s'owly moving normal toads. It matops restoration (Fig. 3) could be ex- is now becoming evident that the coordi- tended further forward without disloca- nated movements of locomotion, while they tion, it has been placed in a position more are under the control of the central nervous in agreement with the contralateral fore- system, are initiated by peripheral stimula- limb. tion. It is rather tempting to assume in Carman (1927) has estimated that in the this case, but open to criticism, that the case of Baropus hainesi and Ancylopus nervous mechanism inducing the primitive ortoni the length of the stride and the size tetrapod locomotor pattern has not been of the tracks indicate an animal weighing completely dominated or obliterated by close to 400 pounds and measuring six to that producing the leaping mode. The eight feet in length. In spite of the weight mechanics of the leaping method of loco- and size, there is absolutely no indication motion in the Anura have been thoroughly that the body was dragged along the investigated by Hirsh (1931).

PRESSURE DISTRIBUTION IN THE UNSPECIALIZED AMPHIBIAN FOOT AND AN ANALYSIS OF THE FORCES ACTING ON THE LIMBS DURING LOCOMOTION A knowledge of the distribution of pres- Most caudate amphibia weigh so little sure or weight in the amphibian foot is of that the most delicate apparatus would be value in interpreting more exactly the na- required to measure the weight distribu- ture of the locomotor process. It also tion of the hind-foot. In view of this gives some clue as to the functioning of the difficulty, an investigation of fossil tet- various elements of the foot and finally is rapod footprints has been made. Through helpful in working out the functional the kindness of Mr. C. W. Gilmore casts changes. have been obtained of the better preserved I WI,11 MA \1,/,

~ -'-4"

I~~~~~~~~~~~~~I

punX.~~~~~~~-:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~M-1 1('Y~~~~~~~-Z~~~~\zZ. I~~~ti

Fi.Thooprns fAmobtacu tubtn Spifraio,Pnslain rn Canyon)drawnfrom positive and negative cat'fteoiialtak.Te)r cnitnl ep eroteinesd,atiuaryinte iint o hefrs hretos Telegh fth trd,p bu 3.2ichs,wareatvey hoterthn ha /o rmaop-,hedoteline'connects the hin-fo

prints.~~ K-.-

416 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 417 footprints found in the Carboniferous and footprints of the front feet have a path Permian formations in the Grand Canyon running in a similar direction, essentially (Fig. 7). from the posterior to the anterior inner Elftman (1934) has criticized the use of border. When the foot is applied to the footprints made in any soft, yielding ma- ground at the beginning of propulsion, the terial because under such conditions they weight of the body proportioned to it is are a record of the shape of the foot rather distributed throughout the foot to its than of the pressure distribution. This is maximum extent. As the foot is rolled to a large extent true, particularly when off the ground, the area supporting this very accurate results are desired. It is weight is decreased and consequently, also true, however, that if the foot itself with the weight remaining constant (it is relatively expansive and supple at least may actually vary slightly), there is an in- a rough indication of the weight distribu- crease in weight per unit area of foot still tion should be indicated by the footprint. in contact with the ground. As a result, That this is so is demonstrated by the fact the print is deepest in the vicinity of the that different parts of all the well-preserved first three toes of a pentadactyl foot, which footprints show variations in their relative is the last part to be raised from the depth and that these variations are present ground. Although it is not possible by in a definite sort of pattern. It has been this means to determine the path of the argued that the variations are due partly resultant of pressure as Elftman and to the presence of plantar pads. While this Manter have done for the human and may be true, it would appear significant chimpanzee foot (1935b), it is possible to that the pads all show the same sort of trace the path of relatively greatest pres- arrangement, indicating a response to pressure or weight distribution in the above sure. manner and to derive some conclusions It is of course impossible to ascertain therefrom. with certainty which footprints are am- Nauck has analyzed (1924) the forces phibian and which are reptilian. In either acting on the amphibian hind-limb when case the great similarity between all of the body is at rest and in motion. He has them indicates that the type of locomotion postulated, in the resting position (Fig. was essentially the same. The deepest 8,A), the presence of a raising component part of the print is usually at the base of in each hind-limb b and b', and also the the first two digits, extending and becom- components c and c' that draw the limbs ing progressively shallower for a short way toward the midline. The latter counter- along the inner border of the foot. This act, along with friction, the forces w and depression then continues to the posterior w' (gravitational effect) which cause the portion. foot to slip sideward. The resultant At the beginning of the propulsive phase forces of the two sets of components, bb' the entire foot is placed upon the ground and cc', he designates as a and a'. They at the same time, that is, from the tips of cause the body to be raised, counteracting the toes to the tarsal flexure. As men- the downward effect of gravity. tioned earlier, the gradual sloping of the During locomotion one hind leg is al- posterior part of the print is due to the ways in some stage of the recovery phase wide curvature of the tarsal region. At and the other in some stage of the propul- no time during this phase is any pressure sion phase. The recovering leg is not in transmitted through the proximal tarsal contact with the ground and hence only c elements directly to the substratum. As (or c') is acting, causing the body to move the body moves forward and the foot rela- toward the c' (or c side). Under these tively backward, the pressure increases, conditions Nauck claims the propulsive reaching a maximum in the region of the leg will have a forward component, so he first, second, and third metatarsals and calls forth a new force m (Fig. 8,A') that corresponding digits and decreasing later- equals c plus a forward component. ally toward the fifth metatarsal. The Since the propulsive leg only is responsible 418 Bulletin American Museum of Natural History [Vol. LXXVIII

Y'rcvity4,v/

AlauekJnalysis

Fig. 8.-The forces acting on the limbs during locomotion. A. Nauck's analysis of the forces acting on the hind limb at rest (A) and in motion (Al). B. Diagram of forces acting on the fore- and hind-limbs during locomotion, as interpreted here. for raising the posterior part of the body, angles to the body during locomotion. the recovering leg being off the ground, Since this is not the case, a partial revision there is another component 1 that equals of his analysis is in order (Fig. 8,B). b plus b'. Nauck concludes that the re- Regardless of the number of limbs on sultant of these components is a force k the ground at any one time (two, three, or that moves the body forward and also four) either at rest or during movement, sideward. The contralateral fore-limb has the raising force exerted by the limbs must a resultant moving the body sideward in be greater than the force of gravity, if the the opposite direction, with the result that body is to be raised from the ground at all. the animal moves only forward. Nauck The morphological basis for Nauck's com- errs in assuming that the hind-limbs and ponents b and b', which are supposed to particularly the feet are directed at right raise the body off the ground in the resting 1941] Schaeifer, Evolution of the Tarsus in Amphibians and Reptiles 419 position, has not been determined. Ro- It would appear that Nauck is correct mer (1922, p. 568) has pointed out that in in stating that the forces w and w' are the primitive tetrapods the ventral mus- largely counteracted or balanced by the culature of the femur, mainly the pubo- friction that exists to a greater or lesser ischio-femoralis externus arising from the degree between the sole of the foot and the pubis and ischium, was responsible for ground, but more importantly by the ex- actually raising the pelvic girdle. The trinsic flexors of the crus, which, through downward pull exerted by this muscle mass their action, create the forces c and c'. on the femur was translated into a force The relationship between these forces raising the pelvis, as the movement of the and the structure of the crus and foot re- femur in this direction was limited by the mains to be considered. The tibia makes crus. Hence, b and b' might be considered the most extensive articulation with the as the effect of a system of forces operating femur and is certainly the weight bearing around the pelvis. As pointed out pre- axis of the upper portion of the crus. At viously, even the very heavy labyrintho- the crurotarsal joint, however, the fibula donts were apparently able to raise the is the weight bearing axis, as it, rather than body off the ground. the tibia, is in a position to distribute the As already noted, the long axis of the weight to the tarsal elements and the front- and hind-feet is parallel to the path digits. It appears evident, therefore, that of progression during the entire propulsive there must be a weight transfer from the phase. The forward force exerted on the tibia to the fibula. This transfer is ac- ground through the movement of the hu- complished by certain muscles, particularly merus and the femur is the sum of the for- the pronator profundus and the interosseus ward force fi exerted by the fore-limb plus cruris. This situation resembles the man- the forward force f2 exerted by the con- ner in which forces are transmitted along tralateral hind-limb, which, in turn, would the human forearm when it is supporting be equal to m minus c in the case of Nauck's analysis. Again, it should be stressed part of the body weight. The forces are that the active participation of the foot in transmitted from the hand mostly to the the propulsive effort is practically non- radius and then across the interosseus existent in the primitive tetrapods. The membrane to the ulna and hence to the main function of the pes is to increase the humerus (Grant, 1940, p. 74). area through which the tractive effort of the The forward force f1+2 is therefore trans- femur and its associated muscles is applied ferred from the femur mainly to the tibia, to the ground. The crus, on the other then across to the fibula and from the fibula hand, acts as a sort of pivot with an ex- it is distributed to all parts of the tarsal panded base, the foot, on which the femur swings in propelling the body forward. region and the digits. A greater propor- The extrinsic and intrinsic flexors of the tion of the weight appears to be thrown on foot are principally concerned with orient- the inner side of the foot, as is demon- ing and maintaining the most advanta- strated by the footprints and also by the geous position of the foot during propulsion, fact that during propulsion the lateral as well as with giving it rigidity. toes are raised first. 420 Bulletin American Museum of NaturafHistory [Vol. LXXVIII - THE EXTRINSIC AND INTRINSIC MUSCULATURE OF THE AMPHIBIAN HIND-FOOT AND ITS RELATION TO LOCOMOTION The well-differentiated musculature of all the details of origin and insertion be de- the lower leg and foot of the amphibian scribed, as there are numerous papers on when compared with the relatively undif- this subject (Eisler, 1895; Osawa, 1902; ferentiated musculature that must have Wilder, 1912; Francis, 1934). existed in the crossopt pelvic fin (Gregory Dissections were made of the lower leg and Raven, 1941) is an indication of a pro- and foot of Batrachuperus, a hynobiid, found change and increase in complexity Cryptobranchus, Necturus, and Triturus. of function. The range of movement of The diagrams show (Fig. 9) the generalized the amphibian hind-limb has been de- condition of the musculature of a typical scribed and it is now necessary to deter- caudate. Howell's contention (1935a) that mine the role of specific muscles in produc- it is not possible to restore accurately the ing the movements. musculature of an extinct form is valid in Muscular differentiation, whatever its certain cases, specifically where there are cause, is simply an indication of an in- no tuberosities or other indications on the creasingly complicated range of move- bones for the origin and insertion of the ment of the part concerned. Le Gros muscles in question. In this case, it Clark (1939) has pointed out that muscles should be noted that the structure of the are capable of progressive variation. skeleton of such forms as Trematops indi- That this progressive variation or, in cates that the mode of locomotion must other words, differentiation, is associated have been essentially the same as in the with the origin of specific and more local- Caudata. Hence, it follows that the ar- ized movements is well brought out by the rangement and subdivision of the muscula- example cited by Clark. In man the flexor ture must have been very nearly the same longus pollicis is separated from the com- in labyrinthodonts and caudates. mon flexor of the fingers and there is a partial separation of the flexor of the index finger. This differentiation can only be Extensor Series associated with the functional importance The origin of all the extrinsic extensors of these two digits in man in contrast to from the lateral condyle of the femur is the other primates. probably a primitive feature, still persistent Very few attempts have been made to in the Caudata. In Eryops the lateral assign definite actions to the leg muscles condyle is exceedingly well developed as is of the caudate amphibians, the most recent also the case in Trematops. The proximal designations being those of Francis (1934). portion of the lateral border of the tibia is These are in most cases correct but fre- depressed and somewhat convex, making quently the action of a muscle is much a channel for the extensor mass to reach more complicated than he describes. The the lower portion of the crus. The origin characteristic locomotor movements de- on the femur causes these muscles to ex- scribed above are produced by the coopera- tend the two lower segments of the leg tion of muscles and for this reason simple instead of just the foot alone. statements of their action are not always Another important feature of theextrinsic correct. extensors is that their tendons of insertion The details of movement during pro- do not run to the digits. The tibial and pulsion and recovery form a basis for a dis- fibular extensors insert on the shafts cussion of the details of the musculature. of their related bones and on the tarsus, There is considerable variation in the sub- while the medial digital extensor inserts by division of the larger muscles among the a number of tendons on the bases of the various caudate families. This variation, metatarsals. Mechanically, such an ar- however, is of no functional importance rangement produces rapid extension but is and hence will not be discussed, nor will not very efficient. The short digital ex- lI . \1

7lexor exyt,sor -primOrdidhA' Vhiulari,F commUnMs

I/aentar // . Z) aponearosts/ ofext comm. I/ //1 /Y/ flexor nt,perA7cialis GS I

7eVxor -accessor,us pronaator lateral/s protfindas

abductor mre tafarsaeA - dzqihi r fiexor e\6;-2,M nJh]gJentes _ loreptA'Drorundasf

C D

Fig. 9.-The cruropedal and pedal musculature of a typical caudate amphibian, based on dissections of Batrachuperus, Cryptobranchus, Necturus, Triturus. A, extensor musculature; B, C, D, successively deeper layers of flexor musculature. 421 422 Bulletin American Museum of Natural History [Vol. LXXVIII

Inner border of foot Outer border of foot raised, outer border raised, inner border lowered ='supination" lowered ="pronation" Midsaqgttal Axis of Foot

a b

Fig. 10.-A diagrammatic representation of the action of the extrinsic pedal musculature of the caudate amphibian. The shaded area represents the tarsus as a whole in which there is a slight amount of movement between elements. a, axis of tarsometatarsal joint; b, non-functional joint between the crus and the tarsus. tensors, of course, do have tendonous inser- ondly, it cooperates to some extent with tions on the distal phalanges. the flexor accessorius and pronator pro- All the caudate amphibia have a large fundus in aiding to maintain the forwardly combined abductor and extensor digiti I, directed position of the foot during the the part acting as an abductor being called propulsive phase. Both of these actions by some, for instance, Wilder, the supinator are peculiar to the amphibian type of pedis. It has a diagonal course across the locomotion. Needless to say, these muscles tarsus, arising from the head of the fibula, must have been present in the labyrin- the intermedium, and a part of the centrale thodonts to maintain the same positional complex and inserting on the base of the effect. first metatarsal. The role of this muscle has not been generally realized. The Flexor Series larger abductor portion is first of all con- The plantar aponeurosis developed as an cerned with the slight "supination" of the extensive tendinous sheet for the common foot at the end of the propulsive phase, thus insertion and origin of both intrinsic and bringing it into the position it assumes for extrinsic flexors. Its relatively great size the greater part of the recovery phase. Sec- in the primitive tetrapods is undoubtedly 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 423 associated with the fact that the posterior sist in such a migration in that an entirely border of the plantar surface of the foot is new attachment with the bone would not not sharply defined by the presence of an have to develop during the shift. ankle-joint. The plantaris profundus of Wilder or The origin of the flexor primordialis com- the flexor accessorius of Francis presents munis is subject to variation. In Nec- some difficulty. Francis divides it into a turus it arises entirely from the lateral lateralis and a medialis on the basis of in- condyle of the femur; in Salamandra, from nervation, with the contrahentium longum the fibular shaft; and in Cryptobranchus, running between the two divisions. Wilder from both. This difference in origin states that it is divided into two masses causes some slight variation in action. by the latter muscle but does not give them In Necturus it is only a flexor of the foot, separate names. In any case, both parts while in Salamandra, because of the exten- of the muscle have a similar action, that is, sive fibular origin, it also adducts the slight flexion of the tarsus. When the foot foot toward the midline. As Francis is in contact with the ground during propul- points out, this is due to the inequality of sion these muscles then act as adductors, the fibers on the tibial and the fibular sides moving the foot toward the midline of of the muscle. It is impossible to deter- the body. Both parts arise from the fibular mine which of these conditions is the more shaft and lateral tarsals, inserting on the primitive. In the more terrestrial Cau- plantar aponeurosis. data, this muscle undoubtedly assists in The pronator profundus varies but little the adduction of the foot. Since this in its origin and insertion, running from movement probably existed in the Per- the shaft of the fibula to the head of the mian types, at least a partial origin on the tibia and the tibiale. Its fibers run in the fibula seems probable. Romer (1922) same direction as those of the plantaris, believes that it arose from the femur, and both muscles having the same action. the roughened ventral surface of the lateral The role of the flexor accessorius and of condyle of Eryops indicates that it at least the pronator can be demonstrated in a partly did so. The role of the flexor prim- negative fashion by transecting these mus- ordialis communis in locomotion is dis- cles in the living animal. When this is cussed elsewhere (pp. 419 and 462). done, the animal is no longer able to adduct The caput longum musculorum con- the foot, making the retraction of this limb trahentium is the only crural flexor in very ineffective. The flexors cannot oper- which the fibers run in a proximodistal ate properly as the propulsive stroke is direction only. It functions as a flexor of now a sort of abduction rather than flexion. the tarsus and is a narrow, comparatively This would indicate the necessity of joint weak muscle in the Caudata. There is no action on the part of the extensors and basis for determining whether or not this flexors located along the lateral border of muscle was present in the labyrinthodonts. the entire leg in a fashion which Morton It appears to have split off from the com- believes is indicated by the Thinopus mon flexor mass as a separate muscle with footprint. It appears evident that the only flexor action, possibly during the arrangement of the musculature of the stage in which the flexor primordialis com- entire leg does not favor the view that the munis was developing an extensive fibular hind leg was at any time sprawled out at origin. It might be reasoned from this right angles to the body during locomo- that it must have been present in the primi- tion. tive types. This muscle may represent a It is obvious that the primitive tetrapod case in which the insertion has shifted to- tarsus does not allow pronation, or for that ward the postaxial side of the foot. It matter supination, of the foot on the does arise from the fibula, in common with lower leg. In fact, as will be pointed out the other flexors. The ligamentous band later, so-called pronation and supination on which it inserts, along with the origins can only occur in the mammalian foot. To Qf the wontrahentes digitorum, might as- name the above muscles pronators and 424 Bulletin American Museum of Natural History [Vol. LXXVIII supinators by no means sets forth their would have had an extensive tendinous true function. insertion. As this mass subdivided, the That the interosseus cruris simply acts tendinous sheet seemingly became the as an elastic ligament, as Francis believes, plantar aponeurosis. This conclusion is is rather doubtful. The fact that it is a supported by the common direction of the muscle rather than a ligament indicates a fibers of all the crural muscles except the more active role. Its function may be contrahentium caput longum. that of approximating the tibia and fibula, which undoubtedly have a tendency to The Plantar Musculature separate during the propulsive stroke, be- The intrinsic musculature of the pes is cause of the relative positions they are concerned with various movements of the forced to assume. The interosseus cruris toes. The flexores breves superficiales is also of importance in the weight transfer are subdivided into a number of slips for between the tibia and fibula in that it each toe. They arise from the plantar binds these bones together as a functional aponeurosis and insert on the metatarsals unit. The fibers run in the same direc- and proximal phalanges. The propulsion tion as those of the flexor accessorius and phase is carried out until all the toes have pronator. their dorsal surfaces turned plantarward In Necturus there is a muscle named by in the order of their length. When the Wilder the "flexor tibialis." It arises third and fourth toes, and finally the fourth from the lateral condyle of the femur and alone, are the only portions of the foot in inserts on the tibia and tibiale. It is not contact with the ground, the added length present in the other Caudata investigated of the tendons of insertion (to the second and may simply represent a deep portion most distal phalanges), plus the action of of the pronator that has separated as a dis- the interphalangeal muscles of these digits tinct muscle. Morton (1935) has postu- give the digits rigidity. lated the presence of a tibial and a fibular The contrahentes digitorum all arise flexor as the primitive tetrapod condition. from a ligamentous band in the region of Unless these names are substituted for tarsalia three and four. Hence, as Francis certain other differently named muscles, points out, the slips to the first and fifth that is, flexor primordialis communis, digits act primarily as adductors, second- flexor accessorius, and caput longum con- arily as flexors, while those of digits two, trahentium, it is not possible to confirm three, and four act as flexors only. They his conclusion after carefully examining all insert on the heads of the proximal the extrinsic flexors of the foot in the cau- phalanges. dates in which this region was dissected, The flexores digitorum breves profundi nor can any record be found in the litera- are also flexors except for those on digit one ture of the presence of tibial and fibular and digit five where they are also adduc- flexors (properly so-called) in any caudate tors. That on digit five may be strongly amphibian. developed with some fibers arising from It appears more likely that the most the distal end of the fibula. primitive condition was a single, large The intermetatarsales, arising and in- flexor mass, the fibers of which arose mainly serting on adjacent metatarsals at dif- from the lateral condyle of the femur and ferent levels, act as adductors of the digits. inserted on the head of the tibia and the They function mainly during the recovery tarsus at various points, including the phase when the digits are drawn close to- bases of the digits. Such a muscle-mass gether. 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 425 THE REPTILIAN TARSUS The reptilian tarsus, in contrast to the integration, the locomotor methods of amphibian, has evolved into many dis- each group will be discussed along with the tinct morphological and functional types particular type of tarsus, as well as in the (Fig. 11). For the sake of more complete section on reptilian foot musculature.

THE ORIGIN OF THE PRIMITIVE REPTILIAN TARSUS The primitive reptilian tarsus contained to reinvestigate the ontogeny of the rep- eight or nine bones. There are but two tilian tarsus, since a survey of a fairly ex- in the proximal row, the astragalus and the tensive literature shows that there is more calcaneum, and they are relatively larger or less general agreement as to the identi- than any of the elements in the proximal fication and disposition of the anlagen row of the primitive amphibian tarsus. present, except in the case of the tibiale, The origin of the astragalus has caused the presence of which is still open to ques- much speculation. Gegenbaur and Williston tion. believed that it consists of a fused tibiale and As in the embryonic amphibian tarsus intermedium, while Baur, Broom, Jaekel, (Fig. 12,A), the anlagen of all the tarsals Romer, Watson, and most other paleon- converge toward the fibula, with the inter- tologists are of the opinion that the as- medium and the fibulare making up the tragalus is made up of the intermedium proximal row. In most cases the tibiale is only. Schmalhausen (1908b) was one of conspicuous by its absence (Fig. 12,C,E), the first to consider the astragalus as con- although Howes and Swinnerton (1901) sisting of the intermedium and the fourth and Rabl (1910) claim that there is a sepa- centrale. Dollo (1929) carried this point rate tibiale in Sphenodon (Fig. 12,D) as a of view further by including the tibiale. very small concentration of cells at the Recently Holmgren (1933) has rein- base of the tibia that fuses very early vestigated the entire problem and has con- with the centrale. The evidence, as pre- firmed quite convincingly the belief that sented, is not conclusive, although, in its the astragalus of reptiles and mammals favor, the centrale does extend under the is the product of fusion of at least two and tibia, suggesting the fusion of the tibiale possibly three anlagen. If the ontogeny and the centrale. The other case of a of the reptilian tarsus is considered in the supposed separate tibiale is in the croco- light of de Beer's restatement of the Bio- dilian foot (Fig. 12,F). Although the genetic Law (1940), the problem can be anlage identified by Rabl as the tibiale is considered more intelligently in its proper in the position of the centrale, both Steiner perspective. According to this interpre- (1934) and Holmgren (1933) present more tation, ontogenetic events of the ancestor convincing evidence of its presence. In may be repeated in the descendant, but all cases, the very close proximity of this embryonic characters of the descendant do element to the distal end of the tibia makes not represent those characters of the adult its existence open to question. ancestor. The degree of resemblance be- The intermedium is conspicuously pres- tween the embryonic tarsi of amphibians ent and is always located between the and reptiles and mammals is great enough tibia and the fibula. Its relatively large to permit identification of homologous ele- size has suggested that the tibiale may be ments. While these anlagen remain sepa- indistinguishably fused with it (Sewert- rate in the adult amphibian, there is, of zoff, 1908) for which there is apparently no course, considerable fusion (and elimina- real evidence pro or con. It is evident, tion) before the adult reptilian (and mam- however, that the centrale (there is a malian) stage is reached. single centrale rudiment in most cases) In connection with the present investiga- fuses with the intermedium during ontog- tion it has not been considered necessary eny. Hence, it may be stated with con- AfeSosauras /MCCr)YL,VJICAAJfU11.1/iIf/)MA Labzdosauras COTYZOSAI[R/A kkZisak a Fig. 11.-The phylogeny of the reptilian tarsus. Original drawings of specimens, with the exception of Youngina and Chasmatosaurus (modified after Broom), Protorosaurus (after Peyer), and Pteranodon (after Eaton and Williston). 426 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 427 siderable assurance that the astragalus is mammal Claenodon, Matthew (1937) has composed of the intermedium and one described the presence of a small, laterally member of the centrale complex of Am- compressed bone which he suggests may phibia. be the remains of the tibiale (Fig. 11), the Sj6gren (1940), investigating the de- astragalus being made up of the inter- velopment of the tarsus of Chamaeleon, has medium only. That this bone and the arrived at a novel conclusion. He found "accessorium" are the same thing must be no evidence of the intermedium, but con- considered as a distinct possibility. sidered an anlage that formed near the There is no real indication of the loss distal end of the tibia to be the tibiale. of the tibiale in the labyrinthodonts. This anlage then migrated into the position Broom's arrangement of types (1921) to of the intermedium at a later stage. He support his theory, as pointed out earlier,

-C t1+2

r,3lu &4~ Iff .Jf CZ E F G -tq Fig. 12.-Embryonic stages of the tetrapod tarsus. The stage selected in each case shows the greatest number of anlagen present at any one time during ontogeny. A, Hynobius (after Schmal- hausen); B, Pelobates (after Schmalhausen, Holmgren); C, Chelonia (after Rabl); D, Sphenodon (after Howes and Swinnerton, Rabl); E, Tarentola (Ascalobotes) (after Sewertzoff); F, Crocodilus (after Holmgren, Rabl, Steiner); G, Mus (after Schmalhausen, Holmgren). also found that the fibulare and centrale would appear to be an illogical assemblage, fused at an early stage. Needless to say, although it supposedly demonstrates a it is very difficult to reconcile these con- progressive reduction in the size of this bone clusions with the other results. and a shift in its position, making it homolo- Both Schmalhausen (1908b) and Holm- gous with the mammalian navicular. As far gren (1933) are agreed that the mammalian as paleontological evidence is concerned, it astragalus (Mus) is made up of the inter- must be concluded that the tibiale simply medium and the centrale (Fig. 12,G). disappears in the Reptilia. Hartmann- They consider a condensation at the distal Weinberg's attempt (1929) to distinguish all end of the tibia to be a tibiale and that it the primitive tarsal components except the forms the "accessorium" to which the pre- tarsalia in the single pareiasaurian proximal hallux is attached. In the Paleocene tarsal element, according to the arrange- 428 Bulletin American Museum of Natural History [Vol. LXXVIII ment of the spongiosa and compacta, is a compound bone phylogenetically, com- not convincing. Haughton and Boonstra posed of the intermedium, a proximal cen- (1930) also made sections of this element, trale, and possibly the tibiale. This is but were unable to determine by this based mainly on embryological evidence means which elements are included. The and on the lack of any really contradictory presence of three proximal tarsal elements paleontological evidence. The constant in the cotylosaurs Seymouria (Fig. 13,A) position of the foramen for the perforating and Limnoscelis may be explained by the artery between the fibulare and the inter- fact that, in these forms, the tarsus is still medium in the amphibians, and between on the way toward becoming truly rep- the calcaneum and the astragalus in the tilian. Its presence, or supposed pres- reptiles, does represent paleontological evi- ence, in a number of the secondarily dence that the fibulare and- the inter- aquatic reptiles may be considered as due medium are present as the calcaneum and to a secondary fragmentation of this re- part of the astragalus, respectively. It is gion, or to a neotenic condition. not evidence, however, in the case of the The position here taken is that the as- astragalus, that only the intermedium is tragalus of reptiles and mammals is really present.

THE COTYLOSAURIAN TARSUS AND ITS FUNCTION The cotylosaurian tarsus evolved in the The tibia and the fibula of Seymouria Permian into a number of distinct types are of about the same length and their (Fig. 13). In fact, the extent of the al- distal ends must have been extensively teration from the primitive condition capped with cartilage in life. The three found in Seymouria is probably the most proximal tarsal elements are hence all on extreme to be found in any single reptilian the same transverse level; with the inter- order. medium articulating about equally with the tibia and fibula. According to White THE SEYMOURIAMORPH TARSUS (op. cit., p. 384) the proximomedial corner In Seymouria (Fig. 13,A), a member of of the tibiale is relatively thickened dorso- the most primitive cotylosaurian suborder, ventrally, foreshadowing the rounded tibial the Seymouriamorpha, the tarsus still facet on the astragalus of pelycosaurs. possesses three proximal elements, an ob- This may be true, if it be admitted that the vious holdover from the primitive tetrapod tibiale is included in the astragalus. condition. These are the fibulare, inter- For the first time in tetrapod evolution, medium, and tibiale (Watson, 1918; White, there may have been definite flexure be- 1939). The tibiale is not reduced in size, tween the crus and the tarsus, but the nor does it show any tendency to migrate articular surfaces give no evidence of it, as into the position of the navicular. On they were finished in cartilage. The the other hand, there is a mesial centrale greatest flexure must still have been be- lying just distal to the tibiale, which tween the tarsalia and the metatarsals, is certainly homologous with the na- and the mechanics of locomotion hence es- vicular of Labidosaurus. This is further sentially the same as in the labyrintho- evidence against Broom's theory (1921) donts. The distal articular surface of the that the navicular is simply the tibiale that femur faces ventrally, indicating a per- has migrated distally. The exact number manently flexed knee-joint, while the proxi- and the disposition of the unossified ele- mal articular surface is terminal, allowing ments cannot be safely determined. It only horizontal movement. would appear reasonable to assume that more than the one centrale was present, THE CAPTORHINOMORPH TARSUS particularly if the thesis that the astraga- Although the Captorhinomorpha are lus is always composed of a centrale and considered to be advanced cotylosaurs in the intermedium is believed correct. several characters, the limbs are quite 1941 ] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 429 primitive. The tarsus of Limnoscelis plete specimens by Case (1899, 1911) and (Williston, 1911a) is almost identical with Williston (1908). Complete preparation that of Seymouria, in that there are three of Case's material (A.M.N.H. No. 4883) separate proximal elements. The fibulare for the present study has resulted in a and the intermedium are, incidentally, the tarsus that does not resemble very closely only ossified elements. As in Seymouria, any of the previous restorations. The the intermedium articulates about equally fibula was mistaken for the tibia, thus giv- with the tibia and fibula, the latter two ing the restorations a rather strange ap- bones being of the same length. With the pearance. The fibula always articulates tibia extensively articulating with the in- with both the astragalus and the cal- termedium, it was in a position to support caneum, but never with the tibia, as was a large share of the body weight directly, supposed in this case. thus removing the necessity of a weight The astragalus is well preserved and can transference from the tibia to the fibula as be compared in some detail with the pely-

A X B 110

II\'

F Fig. 13.-The cotylosaurian tarsus. A, Seymouria (after White and specimen); B, Labidosaurus (from specimen); C, Tuditanus (from photographs); D, Diadectes (after Romer and specimen); E, Nyctiphruretus, plantar view (after Efremov); F, Scutosaurus (after specimen). in Trematops. In support of this conclu- cosaur astragalus. It belongs without sion, the distal end of the tibia is relatively question to the left foot rather than to more expanded than in the labyrintho- the right, as figured by Case (1911). donts. This is undoubtedly the most primitive The tarsus of the captorhinomorph astragalus known and it is, therefore, of Labidosaurus (Fig. 13,B) is the only coty- great interest. It has the L-shaped ap- losaur tarsus known that might be con- pearance of the pelycosaur astragalus, al- sidered as representative of the primitive though the vertical arm, in particular, is definitive reptilian type, in which there relatively very much shorter than in the are two distinct proximal elements, an typical pelycosaurs. The tibial facet on astragalus, and a calcaneum. The diadec- the horizontal arm is not rounded as in the tomorph tarsus, as will be pointed out, is pelycosaurs, and it is located more on the clearly specialized. The tarsus of Labido- lateral than the dorsal surface. The saurus has been described from two incom- fibular facet is simply the proximal sur- 480 Bulletin American Museum of NVatural History [Vol. LXXVIII face of the vertical arm, that in turn is lieve that the functional ankle-joint was dorsoventrally compressed, in contrast still in its old location between the tarsalia to the condition in pelycosaurs in which the and the metatarsals. vertical arm is more oval in cross section. The affinities of Tuditanus Cope (Eosaur- There is a decided notch on the medial avus Williston) have long been a difficult border for the passage of the perforating problem (Romer, 1930). Romer (1925) artery. has pointed out the reasons for consider- The calcaneum is not preserved in this ing it a lepospondyl, although later (1933) specimen, and is only represented by a he concluded it was a cotylosaur. Al- fragment in Williston's specimen. From though reptiles are known having three this, and the shape of the space it occupied, proximal tarsal elements, no primitive am- it seems evident that it must have re- phibians are known with but two. Fur- sembled that of the pelycosaurs or eo- thermore, Tuditanus has the typical rep- suchians rather closely. tilian phalangeal formula. The tarsus is There is but a single centrale or navicular not distinctive enough to be definitely that is very much compressed proximodis- cotylosaurian (Fig. 13,C), although it does tally. Its proximal surface is slightly con- definitely suggest one of the primitive rep- vex and fits into the slightly concave distal tilian orders. There are six distal tarsal surface of the astragalus. The fragment elements, with two situated at the base of considered by Case and Williston to be the fifth metatarsal in the fossil. The the centrale gives every appearance of be- tarsus has been restored with a single ele- ing the distomedial corner of a part of the ment at the head of each metatarsal, and astragalus that broke off during preserva- an elongated bone, as the navicular, which tion. There is apparently no sign of the was displaced in the original specimen. navicular in Williston's specimen, unless No known reptile, fossil or recent, has two it was not separated from the astragalus tarsalia at the head of any of the meta- during preparation. tarsals, excluding a pretarsale. On the Tarsale four is the largest of the tarsalia other hand, the largest tarsale is usually and has a very characteristic square shape the fourth, here it is considered as the resembling that found in the pelycosaurs third, and, furthermore, the navicular as and the eosuchians. There are four other restored, only lies proximal to the inner two tarsalia, subequal in size. tarsalia. That the tarsus of Tuditanus The tarsus of Labidosaurus is ideally belongs to a primitive reptile there is little suited to be the ancestor of that found in question, more than that is not known. either the pelycosaurs or the eosuchians or the more aberrant types. In the pely- THE DIADECTOMORPH TARSUS cosaurs the L-shaped astragalus was fa- The tarsus of the Diadectomorpha is vored, while in the other groups it was sup- specialized for supporting a fairly large pressed. The only feature which might and heavy-bodied animal, and it is dis- be considered as specialized, or rather tinctive for each of the three families in pointing more in the direction of the pely- this suborder. cosaurs than the other groups, is the loca- The diadectid tarsus (Fig. 13,D) has tion of the tibial facet. Being slightly distal been described fully by Romer and Byrne to the fibular facet, it automatically directed (1931) for Diadectes and for both Diadectes the foot forward as in the pelycosaurs. and Diasparactus on incomplete material (Cf. Fig. 21,B.) There is no indication, how- by Case (1911) and by Williston (1913). ever, that the ankle-joint was crurotarsal in The two proximal elements are relatively Labidosaurus as in the pelycosaurs. The much larger than those of Labidosaurus tibial facet being essentially a plane sur- and are considerably thickened dorsoven- face, although directed dorsomedially as trally. All the articular surfaces were well as proximally, it allowed very little finished in cartilage and it is not possible movement between the tibia and the as- to determine their exact extent, or shape. tragalus. Hence, there is reason to be- The astragalus at hand is the same speci- 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 431 men described by Case, although, as formed largely by ligaments and connective Romer and Byrne have pointed out, it be- tissue, after the manner of the mesotarsal longs to the left rather than the right side. joint in the typical lizards. The fibular facet is a plane surface facing With the proximal tarsals making an proximally and somewhat dorsally. The angle of 45 degrees with the crus, the tibial facet, which is separated by a notch plantar pad, which was undoubtedly pres- from the fibular, is quite convex and faces ent, may not have covered these elements more medially than dorsally or proximally. completely proximally, thus giving them The orientation of these facets indicates, a certain amount of independence over the as Romer and Byrne have stated, that the remainder of the tarsus. Hence, the flex- astragalus was set at an angle to the plane ure of the crus on the foot may have been of the crus, although by no means ap- distributed between the crurotarsal and proaching a right angle. the intertarsal joints, with the greater The calcaneum is a circular nodule of amount of movement at the latter. It bone, and the fibular and astragalar articu- seems quite impossible, from the nature of lar surfaces are not distinct and must the articular surfaces, that all the move- have been finished in cartilage. Accord- ment could have occurred at the crurotarsal ing to Romer and Byrne, there are three joint. ossified distal tarsal elements, the navicular Romer and Byrne have further pointed and the first and second tarsalia. It may out that the forwardly directed position be supposed that there were three more of the foot on the leg caused a shortening cartilaginous tarsalia. of the preaxial border of the tarsus, mainly The astragalus and calcaneum com- by the loss or shifting of the tibiale and a pletely dominate the tarsus, both struc- raising of the postaxial border from the turally and functionally. Romer and ground, or, in other words, a tilting of the Byrne are of the opinion that the astragalus tarsus from the outer to the inner side. and calcaneum were set at an angle of about As pointed out in the discussion of the 45 degrees with the crus, which would ap- Seymouria tarsus, it seems more likely pear to be correct. Furthermore, they are that the first step in the shortening of the of the opinion that a considerable amount tarsus occurred with the fibulare, inter- of movement was possible between the crus medium, and tibiale taking up positions and tarsus and that the rounded tibial on the same transverse level, and the fibula facet on the astragalus was of importance becoming the same length as the tibia. in permitting the tibia to rotate on the This was probably initiated through a astragalus as propulsion progressed. change in the centrale region, specifically Although it must be admitted that the through a reduction in size of the proximal rounded tibial facet is undoubtedly asso- centrale or a fusion of some or all of the ciated with the torsion and resultant stress other centralia to form the navicular. A produced on the inner side of the foot, and comparison of the Seymouria and Labido- that some movement of the tibia on the saurus tarsi would seem to indicate that astragalus occurred, there is no indication there has been no shortening of the inner from this material that the crurotarsal border of the tarsus through the loss of joint was the main functional ankle-joint. the tibiale, as the astragalus of Labido- The possibility must also be considered saurus takes up relatively about the same that there was a considerable degree of amount of space as the intermedium and flexure between the proximal and more tibiale of Seymouria. The shortening ef- distal tarsal elements, forming a meso- fect was produced in the crus by the rotat- tarsal joint as in the pareiasaurs. This ing and tilting of the tibia in the primitive view is supported by the extensive rounded tetrapods. The necessity for this effect articular surface on the distal border of the was eliminated by the various groups of astragalus and also on the calcaneum. reptiles in different ways, as will be pointed These surfaces might very well have out; in the case of pelycosaurs, for in- fitted into a transversely elongated socket stance, by the rounded tibial facet on the 432 Bulletin American Museum of Natural History [Vol. LXXVIII astragalus and the more distal articulation case, and the tibial and fibular facets on of the tibia than the fibula with the crus. the astragalus and calcaneum were on the same level, the foot could only be for- THE PROCOLOPHONID TARSUS wardly directed through a rotation of the The complete procolophonid tarsus is tibia in front of the fibula as the femur known for the genera Nyctiphruretus moved backward. The shaft of the femur (Efremov, 1940),Procolophon (Broom, 1903, is twisted somewhat as in the lizards, in- 1921; Watson, 1920), and imperfectly for dicating that the knee was held fairly Telerpeton (von Huene, 1920a) and Ano- close to the ground. The position of the moiodon (von Huene, 1939). foot throughout propulsion cannot be ac- In the case of Nyctiphruretus (Fig. 13,E) curately determined; it may have been the tarsus is very similar to that of Labido- twisted laterally to some extent. saurus, with one important exception, It is not possible to make out the de- namely, that although only the plantar tailed morphology of the Procolophon tar- surface of the foot is figured, it appears sus from the illustrations of Broom and quite certain that the tibial facet is re- Watson, particularly with regard to the stricted to the proximal border of the articular facets on the astragalus and cal- rectangular astragalus. The calcaneum caneum. The astragalus has a rather dis- is more or less round and probably dorso- tinctive shape, which might have been ventrally compressed. The navicular lies derived from the type found in Diadectes. proximal to the first and second tarsalia It is dorsoventrally compressed and some- and contacts the proximomedial portion what constricted transversely in the region of the third tarsale, suggesting the condi- of the notch for the perforating artery. tion in the restored tarsus of Tuditanus. The tibial articular surface was apparently The fourth tarsale is, as usual, the largest restricted to the proximal border with no of this series and, as in Labidosaurus, ar- extension on to the dorsal surface. The ticulates both with the astragalus and cal- fibular facet is of the same nature and is caneum. The fifth tarsale is relatively not tilted forward as in Diadectes. The larger than in Labidosaurus. calcaneum is also dorsoventrally flattened It is evident that there was no ap- and, according to Watson, articulates di- preciable movement at the crurotarsal rectly with the fifth metatarsal and the joint, and Efremov has stated that the entire lateral border of the fourth tarsale, functional tarsal joint was mesotarsal. the fifth tarsale being absent. Watson He considered the reductions in the size considers the loss of the navicular to be of the first three tarsalia and the "beginning an advance in the structure of Procolophon. of growth" of the fourth and fifth tarsalia It might better be considered as a speciali- as evidence of increased flexibility within zation, as both the eosuchian and pely- the tarsus and the first step toward the cosaurian tarsi possess this element. formation of a mesotarsal joint. In According to von Huene, the astragalus every case, however, the development of and the calcaneum of Telerpeton are fused. the mesotarsal joint, as will be pointed out, This element bears a marked resemblance was associated with a reduction in the size to the transversely elongated astragalus and number of the tarsal elements distal of Procolophon and, since it is the only to the astragalus and the calcaneum and a preserved element in the tarsus, as figured, functional union of these elements with the possibility of a separate calcaneum the metatarsals. There is no appreciable must be considered. modification of this region from the condi- The articulation between the crus and tion found in Labidosaurus, and hence it the tarsus must have been a simple hinge- would appear more logical to consider the joint capable of only a limited range of functional joint as tarsometatarsal, al- movement. It would appear that the though there undoubtedly was some move- principal plane of flexure was at the primi- ment at the mesotarsal joint and possibly tive location between the tarsalia and the some at the crurotarsal. If this were the metatarsals. With the lack of a rounded I1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 433 tibial fa'cet on the astragalus, the method Hence, they conclude, and correctly, that of locomotion must have been very much the functional joint must have been meso- as in the primitive tetrapods, with the tarsal in position. tibia forced into a diagonal position as They point out that the astragalocal- propulsion progressed. The head of the caneum must have been tilted forward in femur is terminal and the distal articular life, and being thus orientated, it could surface is directed downward, inferring a only have borne the weight of the body permanently flexed knee-joint. through the development of some sort of a cartilaginous heel-process along the dis- THE PAREIASAURIAN TARSUS toventral border, thus giving the tilted The structure and function of the pareia- elements a firm base. The same situa- saurian tarsus (Fig. 13,F) has been tion is present in the tarsus of Diadectes, studied in detail by Haughton and Boon- with the astragalus and calcaneum in- stra (1930). It is the most specialized of clined at about the same angle. It would the cotylosaurian tarsi, having the as- appear unnecessary to postulate the pres- tragalus and calcaneum fused into a ence of a cartilaginous heel when it is single, massive element and the more distal realized that the proximal element was tarsals reduced to from three to five poorly well imbedded in the extrinsic flexor mus- ossified nodules. The tarsi of the Russian culature and, undoubtedly, in a very thick form Scutosaurus (A.M.N.H. No. 5148) plantar pad. If such a process were pres- and the South African Embrithosaurus ent, it would be a very unusual structure (A.M.N.H. No. 2450) have been available indeed. for this study. As Haughton and Boonstra point out, The astragalocalcaneum is essentially the femur is of great interest, for the pareia- the same in both the South African and saurs are the most primitive tetrapods the Russian pareiasaurs. Viewed from known in which the head of the femur is in front, it is a rectangular mass of bone oriented at an angle with the long axis of with the lateral half of the dorsal face the shaft, amounting to about 30 degrees. flattened. The tibial facet is but poorly Hence, for the first time, the femur moved ossified in the South African forms and backward during propulsion in a more must have been capped with a rather vertical plane and not the horizontal plane thick layer of cartilage to articulate with of the other cotylosaurs with a terminal the tibia at a reasonable angle. In Scuto- femoral head. Furthermore, during the saurus, however, this articular surface is recovery phase the femur could swing for- better ossified and is convex. In both, it ward in almost the same plane. The more extends on to the dorsal surface, implying detailed implications of the medially di- a considerable range of movement. The rected femoral head will be considered in fibular facet is restricted to the proximal the discussion of the locomotion of the surface and is slightly concave. Haugh- alligator. ton and Boonstra have pointed out that Haughton and Boonstra consider the the articulation between the fibula and tibia to be the weight bearing element of the astragalocalcaneum was a very exten- the pareiasaurian limb, and the fibula to sive and intimate one, allowing no ap- be the regulator of the rotary movements preciable movement between them. of the foot. They do not elaborate fur- The above-mentioned authors stress ther. It is certainly true that, as in the the following points: The distal articular primitive tetrapods, the tibia makes an surface on the femur is directed more ven- extensive articulation with the femur, trally than distally and the crus could while the fibula merely articulates with a only have moved through an arc of 45 de- small and more laterally directed facet on grees. Furthermore, the non-functional the outer condyle of the femur. The tibia articulation between the fibula and the is thus situated to bear by far the greater proximal element must have prevented proportion of the body weight since, in any flexure at the crurotarsal joint. addition, in the cotylosaurs it obtained for 434 Bulletin American Museum of Natural History [Vol. LXXVIII

the first time an articulation with the tar- cal plane, and the fibula and tibia to- sus, which permitted weight distribution gether offered a double support on which to the greater part of the foot. The the femur could rotate, which was further fibula, however, still must have borne some strengthened by the fusion of the proximal of the weight, probably through a trans- tarsal elements. That the fibulotarsal ference from the tibia to the fibula, which articulation limited the abduction and as pointed out earlier, must have been the adduction of the foot in the pareiasaurs case in the labyrinthodonts. The fibula may be accepted as true, although there must, furthermore, have had a steadying must have been some rotary motion at the influence on the entire leg, in that its ex- femorofibular joint. The somewhat me- tensive and intimate articulation with the dially directed head undoubtedly compen- astragalocalcaneum would tend to prevent sated for the inability of the tibia to assume any displacement of the latter in the verti- the diagonal position during propulsion.

THE STRUCTURAL AND FUNCTIONAL DERIVATIVES OF THE UNSPECIALIZED COTYLOSAURIAN TARSUS The cotylosaurian tarsus, of a sort re- a paddle. There is a tendency toward the sembling that of Labidosaurus, gave rise convergence of the tarsal elements to roughly to three types of tarsi; ignoring, simple rectangular or circular nodules and of course, the numerous variations of each an increase in length, or multiplication in of these types. They may be briefly the number of phalanges. listed as follows: The least modified tarsus within this (1) The type found in the Mesosauria, group is that of the Mesosauria (Fig. 11). Ichthyosauria, Protorosauria, and Sauro- That of Mesosaurus resembles very closely pterygia-a modification of the capto- the tarsus of Procolophon, except that the rhinomorph pattern for an aquatic habitat former has retained a fifth tarsale. Both -to a greater or lesser degree depending on agree in the loss of the centrale and the the order, the protorosaurian tarsus being general configuration of the astragalus. the most conservative, as this group is The astragalus also bears a resemblance to mostly terrestrial. that of Labidosaurus and to the L-shaped (2) The type found in the Chelonia astragalus of pelycosaurs. The tibial facet and Eosuchia and the reptilian orders con- is decidedly more distal than the fibular sidered to be derived from the latter. All but is not rounded as in the pelycosaurs. the specializations in the tarsus in this The fibular facets on the astragalus and group are built around one central theme, calcaneum are also plane surfaces, and the development of a mesotarsal or inter- there is no indication of extensive move- tarsal joint. Although the pareiasaurian ment at the crurotarsal joint. There was tarsus might be included in this category, undoubtedly movement between the tar- it is more convenient and logical to include salia and the metatarsals, but, with the it in the foregoing discussion of the coty- foot functioning as a paddle, the greatest losaurian tarsus. movement must have been distal to the (3) The type found in the Pelycosauria, metatarsals. In swimming, the tarsal Therapsida, and finally culminating in the region, in fact most of the foot, must have Mammalia. This group specialized in been held rather rigid during the propul- the development of a crurotarsal joint. sion phase. Although the head on the femur is terminal, the foot was forwardly TYPE I directed since the tibia articulated with This rather heterogeneous, assemblage the foot at a more distal level than the fibula. of orders has, with few exceptions, aquatic The tarsus of the Ichthyosauria (Fig. tendencies and the tarsus has become 11) requires little comment. The tarsus modified from the Labidosaurus pattern was clearly not an independent functional with the usual end result that it simulates unit, but rather the entire limb must be 1941I Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 435 considered as such. There was undoubt- simple nodules. In the plesiosaurs (Fig. 11) edly some movement between the pha- the tarsus is completely ossified and must langes, but little in the tarsal region. have functioned very much as in the The tarsus of the Protorosauria (Fig. 11) ichthyosaurs. There was obviously no along with that of the Mesosauria, under- localized plane of flexion, but a slight went but little modification from the cap- amount of flexion between each row of ele- torhinomorph type. That of the ter- ments, producing the gradual curve neces- restrial form Araeoscelis is, as Williston sary for sculling movements. (1914) pointed out, unusual in a number of respects. The astragalus is a cube of bone with a flat tibial facet. The cal- TYPE II caneum is dorsoventrally compressed and TIHE CHELONIAN TARSUS of unusual shape. Williston refers to a lateral calcaneal process as a heel and The chelonian tarsus (Fig. 11) is sub- point of insertion for the "Achilles ten- ject to considerable variation, principally don," which cannot be the case, for, as due to the fusion of the proximal elements. will be pointed out, the true tuber cal- That of Chelydra (Fig. 11) is of particular canei first appeared in the thecodonts and interest in that it has retained the primi- therapsids. From the structure of the tive pattern both with regard to the num- proximal tarsals, it is evident that the ber of elements and their relative dis- ankle-joint could not have been cruro- position. The calcaneum is relatively tarsal but tarsometatarsal. The femoral much reduced in size, while the astragalus head is terminal and the tarsus has none is relatively larger than in the Cotylosauria. of the specialization of the lizard tarsus, There is a separate centrale in the Chely- hence there is no reason for assuming dridae and in most of the pleurodires that the feet were oriented at right angles (Zittel, 1932). That the tarsus might to the body during propulsion as Williston readily be derived from the Labidosaurus has done in his restoration (1914, p. 402). type is self-evident. Furthermore, the The tarsus of Protorosaurus, as figured, chelonian tarsus gives some clue as to the is likewise unusual (Peyer, 1937). The probable mode of origin of the mesotarsal astragalus and calcaneum are apparently joint. The first step was undoubtedly a dorsoventrally compressed and of un- flattening of the tibial and fibular articular usual shape. There was but slight move- surfaces on the astragalus and calcaneum, ment at the crurotarsal joint and again resulting in a total loss of movement at most of the action must have occurred at the crurotarsal joint. This was followed the tarsometatarsal joint. As this form by a tendency toward very close articula- was at least partially aquatic, movement tion and finally fusion between the as- in the tarsus was undoubtedly restricted. tragalus and calcaneum and a functional The tarsus of the Sauropterygia is dif- union of this proximal element with the ferent for almost each suborder. That crus. The latter was probably favored as of the terrestrial form Tanystrophaeus, is a means of overcoming the disrupting in- poorly ossified. The astragalus and cal- fluence of the torsion produced in the proxi- caneum are kidney-shaped elements, dor- mal tarsal region during propulsion. That soventrally compressed, and were probably this was the case, is supported by the fact surrounded by cartilage in life. The other that the completely marine cryptodires diagnostic features suggest that the foot secondarily lost the localized plane of was forwardly directed during locomotion flexure in the tarsus. In this type, the and that the functional ankle may have tarsus is made up of a number of rounded been partly crurotarsal, partly mesotarsal. nodules of bone (Archelon) with no indica- The other members of this order were ex- tion of a tendency toward fusion of the clusively aquatic. In the nothosaurs and astragalus and calcaneum or fixation of placodonts the tarsus was at least partly the crurotarsal joint. ossified, the bony portion consisting of The tarsalia are four in number, with 436 Bulletin American Museum of Natural History [Vol. LXXVIII the fourth exceeding the others in size as sus to be forwardly directed during propul- in the cotylosaurs. There is no fifth tar- sion. This difficulty is partly overcome sale in the adult nor any indication of it by the fact that the metatarsals are for- embryologically (Fig. 12,C); it may or wardly directed and are hence orientated may not be indistinguishably fused with at an angle of about 60 degrees to the long the fourth throughout ontogeny. axis of the tarsus and crus. The tarsus is Both the tibial and fibular facets on the narrower on the preaxial side than the astragalus, and the fibular on the cal- postaxial, mainly because the first tarsale caneum, are somewhat concave and are is compressed proximodistally, while the restricted entirely to the proximal sur- fourth tarsale is cubical in shape. This faces. There is therefore no possibility condition further aids the metatarsals, of movement at the crurotarsal joint. By particularly one and two, in orienting sectioning the hind-limb in the transverse themselves in a forwardly directed posi- plane, with the foot fixed in a flexed position. tion, it can be demonstrated quite con- clusively that the functional joint is mesotarsal in position. In the turtles the fusion THE EOSUCHIAN TARSUS AND ITS of the proximal elements has occurred in DERIVATIVES many types, such as Testudo, and where it The captorhinomorph tarsus is without has not, as in Chelydra, the astragalus, cal- question at least the structural ancestor caneum, and centrale are very intimately of the eosuchian tarsus; the number of associated with no possibility of movement elements, relative disposition, and many between them. of the morphological features being es- The manner in which the limbs operate, sentially the same. The tarsal pattern is in spite of the presence of the shell, is known in a number of eosuchians, speci- probably, as Romer has pointed out (1933, fically, Youngina (Fig. 14,A), Palaeagama p. 134), as close to the primitive tetrapod (Broom, 1926), Howesia (Broom, 1906a) method of locomotion as can be found (Fig. 14,B), and Saurosternum, and, except among living Reptilia. The femoral head for minor variation, the pattern is identical is very well developed and approaches the in all of them. perfection of the mammalian condition, The astragalus of Youngina differs from even to the presence of a neck between the that of Labidosaurus in lacking a well-dif- head and the femoral shaft. The orienta- ferentiated tibial facet. This articular tion of the head relative to the shaft is, surface is apparently not rounded (Broom, however, quite unique, for it is not medi- 1924), but is located as a shallow concavity ally but dorsally directed. The articular on the proximodorsal surface. It is en- portion, however, has its center of curva- tirely possible that its location is a herit- ture pointing in a proximodorsal direction age character, held over from the capto- and the acetabulum is correspondingly rhinomorph position. The fibular facet directed ventrolaterally. on both the astragalus and calcaneum is Functionally, therefore, the head may restricted to the proximal borders. The be considered as terminal, for the move- calcaneum is dorsoventrally compressed, ment of the femur is entirely restricted to but, according to Broom, the medial por- the horizontal plane. The well-differen- tion is somewhat thickened for the fibular tiated head and the orientation of the ace- facet and the articulation with the as- tabulum may be looked upon as specializa- tragalus. The primitive nature of the tions associated with the restrictions of Youngina tarsus is further attested by the shell and the fact that it prevents any the fact that the fifth metatarsal is not marked movement in the vertical plane. hooked, although it is hooked in all other With the movement of the femur re- eosuchians known. The tibial facet on stricted largely to the horizontal plane, and the astragalus of the other known eosuchi- the proximal tarsal elements at least func- ans is confined to the proximomedial tionally united, it is impossible for the tar- border with no overlapping on the dorsal 19411] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 437 surface. The nature of the articular sur- would restrict considerably the ability faces at this point would seem to indicate of the tibia to assume the diagonal posi- greatly restricted movement in a dorso- tion (cf. discussion of flexion in Trematops plantar direction, if indeed, there was any tarsus). It is evident, however, that the movement at all. The principal plane of tibial facet is still somewhat more distal flexure must have been tarsometatarsal than the fibular. Whether this slight dif- in position. ference in level was great enough to per- The femur is well preserved in Palaea- mit the foot to remain forwardly directed gama, and although the head is terminal, throughout propulsion is impossible to the shaft has a very slight s-curve, sugges- determine. The nature of the femoral tive of the condition found in the crocodiles head plus the lack of a rounded tibial facet

C D E Fig. 14.-The eosuchian tarsus. A, Youngina, plantar view (after Broom); B, Howesia (after Broom). The thecodont tarsus. C, Chasmatosaurus (modified after Broom); D, Euparkeria (from photo- graph, after Broom); E, Aatosaurus (after von Huene).

and lizards. In spite of the slight curve, on the astragalus are evidence for believing the head is not directed medially to any that the hind feet may have been forced appreciable degree, and the femur must into a moderate outward position as pro- have had its movement during propulsion pulsion progressed. largely confined to the horizontal plane. This is a rather crucial point in the evolu- This being the case, and with the tibia tion of this type of tarsus, for in all and fibula articulating with the tarsus on the orders derived from the eosuchians the same horizontal plane, it is very dif- the tarsus had a functional mesotarsal ficult to see how the foot could have been joint. The functional factors favoring its forwardly directed to the same extent as appearance are difficult to determine. it has been in all the previously discussed This much may be said: all the more ad- terrestrial tetrapod groups. The lack of vanced tetrapods arising from the cotylo- ,movement between the proximal tarsals saur stock tended to localize the functional 438 Bulletin American Museum of Natural History [Vol. LXXVIII ankle-joint proximal to its primitive posi- although not co6ssified. The distal tar- tion between the tarsalia and the metatar- salia are much reduced in size and there is sals. This migration resulted in producing no evidence of a separate centrale. The a highly mobile and complex joint by the functional ankle-joint was intertarsal. It fusion or very close articulation of the ele- seems clear that the intertarsal joint was ments on one or both sides of the joint, as preceded by the loss of the centrale as a pointed out in the discussion under the separate element and a re(luction in the chelonian tarsus. By this means, com- number and the size of the tarsalia. The plicate(d joint surfaces were perfected, fact that there is but one distal element definitely restricting movement in one in the tarsus that articulates with the as- particular direction (excepting in the mam- tragalocalcaneum is an expression of the mnalian foot), and, furthermore, in produc- consolidation necessary to form a diarthro- ing much greater stability at the joint. dial joint. The tarsus of Homaeosaurus is Such a process was never favored with the about as close to an intermediate condi- joint in its primitive location, for the ob- tion as can be found between the eosuchian vious reason that fusion of the tarsalia or on the one hand and the rhynchocephalian the metatarsals would greatly hinder loco- and lacertilian tarsus on the other. motion. The development of a meso- The tarsus of the Upper Triassic rhyn- tarsal joint in one group and the develop- chocephalian Stenaulorhynchus (von Huene, ment of a crurotarsal joint in another group 1938) has, according to this author, a may be looked upon as simply two dif- separate tibiale. That such is the case is ferent ways of producing the same general again extremely unlikely. It would appear result, a consolidated tarsus acting at right more reasonable to consider this element angles to the crus and giving support to a as the astragalus, and the extra element on diarthrodial joint. the lateral side as a persistent fifth tarsale, The consolidation of the proximal por- although the presence of a hooked fifth tion of the tarsus was accomplished in two metatarsal and a fifth tarsale in the same ways in the eosuchian descendants, either tarsus is unique. If this assumption is cor- by fusion or very intimate articulation of rect, the tarsal pattern is exactly like that the astragalus and calcaneum side by side found in Youngina and the joint must have in the same positions they assumed in the been tarsometatarsal. Furthermore, there primitive tarsus, or, in one groul), the is then reason to believe that the fifth meta- Crocodilia, by a rather complicated intertarsal became hooked before the fifth tarsale locking that still permits movement be- was lost, supporting Goodrich's contention tween them. The latter simulates some- (1916) that the hamate process was not what in form and function the superposi- produced by the fusion of the tarsale and tion of the astragalus upon the calcaneum the metatarsal. in the mammalian tarsus. The rhynchoceplialian and the lacertilian tarsus may be characterized as consisting THE RHYNCIIOCEPHALIAN AND LACERTILIAN of an astragalocaleaneum that is dorsoven- TARSUS trally compressed, and extended medially The rhynchoceplhalians and the lacertilian and laterally beyond the tibia and fibula to are considered to be directly descended a slight extent. Its proximal border has from the eosuchians, and this conclusion two shallow concave facets for articulation is borne out by the tarsus. The tarsi of with the tibia and fibula. The crurotarsal the other orders all show specializations joint is rendered completely immobile by that were acquired first in the thecodonts, ligaments that connect the crural bones as will be pointed out shortly. with the proximal element very snugly at The Upper Jurassic rhynchlocephalian every point. There is a particularly strong Homaeosaurus (Lortet, 1892; Broili, 1925) system of ligaments running from the an- has a tarsus resembling that of Sphenodon terior face of tlle very distal portion of the (Fig. 15,A). The astragalus and cal- tibia to the lateral face of the astragalar caneum are very intimately articulated, portion of the proximal tarsal element. 10413 SchaefJer, Evolution of the Tarsus in Amphibians and Reptiles 439

This branching of the ligamentous fibers obviously very specialized tarsus such as indicates relatively greater stress in this in the geckos, the distal surface of the as- region than in any other. tragalocalcaneum possesses two rather The detailed morphology of the as- prominent grooves, with the elevations tragalocalcaneum is somewhat different in between the grooves rounded. The groove Sphenodon (Fig. 15,B) and the lizards. between the medial and lateral elevations In the former, the distal border of the runs diagonally, or proximomedially, to proximal tarsal element has a slightly con- the center of the anterior surface. The cave articular surface that fits over the lateral wall of this groove is relatively somewhat convex proximal surface of the high, forming a kind of lip that projects fourth tarsale. The second and third tar- anteriorly. The groove between the cen- salia and metatarsal five articulate with tral and medial elevations does not extend the dorsal surface of the astragalocal- on to the anterior face, and its axis is di- caneum. Hence, the intertarsal joint is rected anteroposteriorly. The distal por- located laterally between the distal border tion of the tarsus is still more reduced; of the astragalocalcaneum and centrale the two remaining elements are considered four, while medially it is located between to be the tarsale three and the tarsale

"1 Iv A B 11 'I'l \11\\A B \ p'h\e'nd Fig. 15.-The rhynchocephalian tarsus. A, Homaeosaurus (after Broili, Lortet); B, Sphenodon (from specimen). The saurian tarsus. C, Iguana (from specimen). the dorsal surface of the astragalocal- four (by some four plus five). It is now caneum and the more distal elements. generally agreed that the hooked portion This arrangement tilts the astragalocal- of metatarsal five does not include tarsale caneum at an angle of about 40 degrees five. Tarsale four is the only member of with the horizontal plane of the digits, the distal tarsalia that forms a true joint with the result that the crural portion of with the astragalocalcaneum. Its proxi- the leg, particularly when the femur is at a mal surface possesses a groove running in right angle to the body axis, is tilted for- a diagonal direction, dorsomedial to ven- ward at least 45 degrees from the vertical trolateral. Into this groove fits the cen- plane. This condition undoubtedly arose tral elevation of the proximal element, along with the twisting of the femoral thus forming a tongue-and-groove joint. shaft, the distal portion having been ro- Because of the diagonal direction of the tated clockwise (or the proximal counter- elevations and the groove, the crus is di- clockwise), in the case of the right femur, rected medially in relation to the long axis about 50 degrees. of the leg. The nature of this articulation In the iguanids (Fig. 15,C) and other plus the twisting in the femur accounts in members of the Kionocrania, excepting an a large part for the characteristic pose of 440 Bulletin American Museum of Natural History .[Vol. LXXVIII the hind-limbs of the saurians, with the hooked fifth metatarsal may be of im- knee close to the ground and the long axis portance in the classification of the rep- of the foot at an angle with the long axis tilia, it is not associated with any particular of the body. type of locomotion. The hook is present While all the metatarsals but the first in the Chelonia and the Crocodilia, and articulate with the dorsal surfaces of the the mesotarsal joint is differently con- two tarsalia, the first metatarsal articulates structed in each. directly with the dorsal surface of the as- The characteristic manner in which the tragalocalcaneum. A meniscus is present hind-limbs move during the locomotion between metatarsal one and the astragalo- of Sphenodon and the typical lizards is calcaneum, where, in the primitive types, mainly due to the structure of the femur. the torsion was the greatest. It is not ex- In Sphenodon and all the saurians extensively developed and is embedded in amined, the transverse axis of the femoral the ligamentous capsule that arises from head is essentially parallel to the ground the anterolateral surface of the astragalo- throughout most of the propulsive stroke. calcaneum and covers the base of the first As pointed out earlier for Sphenodon, how- metatarsal. It acts as a sort of shock- ever, the distal portion of the femur is absorber and creates a true articular sur- twisted around the long axis of the shaft, face between these elements. from 40 to 50 degrees in a clockwise direc- In the varanids the saddle-joint between tion in the case of the right femur. the astragalocalcaneum and tarsale four is The articular surface for the tibia is modified into more of a concavoconvex thus directed ventrocaudally, forcing the articulation, with considerably restricted crus to be likewise ventrocaudally di- movement. In Varanus komodoensis the rected. This situation, as pointed out foot is permanently flexed on the crus, not above, accounts for the fact that in many only because all the metatarsals articulate lizards the knees almost touch the ground with the anterior surface of the tarsalia during the propulsive stroke. but also because of the relative immobility Besides the twisting in the femoral shaft, of the intertarsal joint. the head is directed about 30 degrees away The transverse tarsal arch, so character- from the long axis of the shaft, medially istic of the mammalian foot, is present to a and in the horizontal plane. With the slight degree in the saurian and the rhyn- femur moving in essentially the horizontal chocephalian foot. The bases of the first plane, this curvature has very little func- three metatarsals are broader dorsally tional effect, except at the beginning of than ventrally, resulting in an arch when propulsion when it enables the shaft almost they are side by side in their natural posi- to parallel the long axis of the body. tions. The origin of a transverse arch The structure of the tarsus and the equal cannot be attributed to the effect of the length of the tibia and fibula, however, as transversely applied stresses in the primi- has been pointed out, do not permit the tive tetrapod foot, as Morton believes, for foot to be forwardly directed unless the reasons previously stated. The fact that crus is forwardly directed at the same time, it is associated with the two types of di- as at the beginning of propulsion. Hence, arthrodial ankle-joint suggests that it has as propulsion progresses, the foot is forced developed along with, or as a result of, the to assume a position at right angles to the tarsal consolidation in these types but in long axis of the body. The position of the different ways. foot at the beginning of propulsion is sub- Metatarsal five, having no dirdet con- ject to much variation. By permitting tact with the tarsus has, however, a large, Anolis to walk over a rough, blackened somewhat circular flange extending under surface that has been covered with a thin metatarsal four, which articulates with film of white powder it can be shown that, tarsale four, and through which it receives although the foot may be almost forwardly its share of the body weight. Goodrich directed at the beginning of locomotion, (1916) is of the opinion that while the it is rotated into the right-angled position 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles >441 as the femur rotates backward. The The next stage in the evolution of the body, as is the case in the caudates, is thecodont tarsus, and of the mesotarsal raised from the ground during locomotion, joint, may be represented by Euparkeria although the relatively very long tail is (Broom, 1913a). The calcaneum still has dragged. Hence, during propulsion the a well-developed processus lateralis (Fig. stresses are for the first time mainly trans- 14,D) and, according to Broom, no indica- mitted transversely across the tarsal region tion of a heel. The centrale has disap- as propulsion progresses. peared as a separate element, however, and there are four (?) reduced tarsalia. THE THECODONT TARSUS There is definite evidence that the joint in The thecodont tarsus is of considerable this case was mesotarsal. Apparently the significance as the Thecodontia are an- astragalus and the calcaneum were very cestral to the pterosaurs, crocodiles, dino- closely associated, and the reduced tarsalia saurs, and birds. The most primitive the- could have had no effect in maintaining codont tarsus known is that found in the plane of flexure in the primitive posi- Chasmatosaurus. The tarsus has most tion between them and the metatarsals. probably two elements in the proximal There is every reason for believing that the row, although Broom (1932b) believes mesotarsal joint developed independently there are three, with a separate inter- in the archosaurs and lepidosaurians, as medium and tibiale. After considering it was not developed in either the eosuchians the tarsus of other closely related types, or the most primitive thecodonts. it would appear evident that the inter- Broom has pointed out (1932b) that the medium and tibiale of Broom are actually tarsus of Chasrnatosaurus agrees very the astragalus and the centrale. It is closely with that of Howesia (Fig. 14,B), difficult to 'believe that a separate tibiale a Triassic eosuchian, and that it also agrees appeared sporadically in various unre- with that found in Youngina, which, in lated genera of terrestrial fossil reptiles, as turn, agrees with the tarsal pattern found some paleontologists would have us be- in Labidosaurus. In other words, the lieve. It appears more reasonable to as- primitive cotylosaurian pattern is still sume that the so-called tibiale is really a present in the eosuchians and the theco- centrale that was displaced during preser- donts, showing the first sign of modifica- vation, or had shifted its position in the tion in Euparkeria. tarsal pattern in a group of genera of Although Saltoposuchus (von Huene, which we have only one or two representa- 1921) is considered to be a primitive theco- tives of the tarsus. It is certainly too pro- dont, there is a well-developed mesotarsal found a modification to be considered only joint, and the calcaneum, while lacking the as a generic difference. In the phylogeny processus lateralis, does have a well-de- of the reptilian tarsus (Fig. 11), that of veloped tuber (Fig. 21,D). The latter is a Chasmatosaurus has been reproduced as distinct advance over the condition found figured by Broom, while in Fig. 14 the in Euparkeria. centrale has been placed in its probable Aetosaurus (von Huene, 1920b), a some- natural position. Assuming the tibiale to what aberrant thecodont (Fig. 14,E), be actually the centrale, this tarsus has which has a tarsus very similar to that of the same number and disposition of ele- Saltoposuchus, has been figured for com- ments, lacking one tarsale, present in the parison with Chasmatosaurus and Eu- foot of Youngina. As far as can be de- parkeria. The tibial and fibular facets termined from -Broom's sketch, the cal- are restricted to the proximal surfaces of caneum has a well-developed processus the astragalus and calcaneum. The proxi- lateralis but no tuber. The functional mal elements are more cubical than the ankle-joint, as in Youngina, was un- same elements in Euparkeria and were im- doubtedly between the tarsalia and the movably articulated with the tibia and the metatarsals as none of the modifications fibula. With the remaining two tarsalia resulting in a mesotarsal joint are present. reduced to mere nodules, there. is little 442 Bulletin American Museum of Natural History [Vol. LXXVIII question but that the functional joint was but poorly ossified. Camp (1930) has mesotarsal (intertarsal). described an astragalus for Mystriosuchus The thecodont femur has a well-de- adamanensis which resembles that of AMto- veloped and medially directed head, a saurus rather closely. condition that first appeared in this order, although, as pointed out previously, the THE CROCODILIAN TARSUS medial bending of the head was in an in- The tarsus found in the Crocodilia (Fig. cipient stage in the eosuchians. The fe- 16,B) is in many ways unique and is not mur was thus able to swing in the vertical approached in any other reptilian group. plane during propulsion and recovery, and The tarsus of the precrocodilian Proto- the feet were brought into a forwardly suchus (Brown, 1933, 1934) is very similar directed position without any torsion at to that of Adtosaurus. The calcaneum has the ankle-joint as the tibia and fibula were a well-developed tuber and the astragalus able to remain parallel throughout propul- is a nodule cuboidal in shape. Although the sion. tarsal region has not been completely Of the groups derived from thecodonts, worked out of the matrix, there is no indi- the crocodiles and one primitive family of cation of any tarsalia. The functional theropod dinosaurs are the only ones re- ankle-joint is definitely mesotarsal in posi- taining the tuber calcanei. The loss of the tion. In the Upper Jurassic Alligatorellus tuber was likely associated with the perfec- (Lortet, 1892) the tuber is reduced (Fig. tion of bipedalism in both orders of dino- 16,A), but otherwise the astragalus and the saurs. Since none of the crural flexors in- calcaneum are very simlilar to the same serted entirely on the tuber, the secondary elements in Protosuchus. There are three loss of the tuber would have but little ef- ossified tarsalia in contrast to the two fect on the musculature. The r6le of the found in recent crocodiles. Apparently foot in propulsion in the thecodonts and the movable interlocking of the astragalus dinosaurs is discussed in the section on the and the calcaneum occurs only in the reptilian musculature (p. 462). Crocodilia. The importance of making a distinction Morphological features of the crocodile between a true tuber calcanei and the pro- tarsus have been considered in detail by cessus lateralis of the calcaneum has been Rabl (1910), with some attention paid to stressed by Tornier (1927). The tuber al- its function. He has pointed out that the ways points in a posterior direction and is tarsal joint is mesotarsal on the tibial side always associated with the superficial ex- and crurotarsal on the fibular side, a fact trinsic flexors of the foot. The processus that can be easily demonstrated in- a pre- lateralis, on the other hand, is definitely a served specimen. lateral extension of the calcaneum around The astragalus and the calcaneum (Fig. which the tendons of the peroneal muscles 16,C) are entirely separate and articulate pass. The lateral process is always best de- with each other by means of a modified, veloped in a calcaneum that is dorsoven- double ball-and-socket joint. In the pos- trally compressed, as in the Cotylosauria, terolateral articulation the astragalus has a Rhynchocephalia, Lacertilia, Pelycosauria, convex surface and the calcaneum a con- and some Therapsida. cave, whereas in the anteromedial articula- It is evident that the process is not as- tion the opposite condition exists. This sociated with any particular type of loco- arrangement permits the calcaneum to ro- motion. In the forms with a mesotarsal tate on the astragalus through about 45 joint the process projects laterally, while degrees. In other words, as the foot is in the cotylosaurs and the pelycosaurs it is dorsiflexed on the crus, the astragalus re- directed more proximolaterally. This may mains immobile and functionally united be- of functional significance. The forms with the crus while the calcaneum moves having a well-developed tuber always tend with the foot (some movement is possible to have a reduced lateral process. between the calcaneum and the adjacent The phytosaur tarsus was apparently tarsale). The fibula is held immobile by 19411] Schaeffer, Evolution of the Tarsus Tn Amphibians and Reptiles 443 its immovable articulation with the as- tarsale and the calcaneum is a plane sur- tragalus. face, prohibiting any movement between The fibular articular surface on the as- these elements, and there is likewise no tragalus is on the lateral extension, which movement between the tarsalia and the places the distal end of the fibula imme- metatarsals. diately above the middle of the fibular articular surface on the calcaneum. THE LOCOMOTION OF THE ALLIGATOR Hence, as in the advanced therapsids and The locomotion of the alligator (Alligator mammals, the weight is passed to the foot mississippensis) has been investigated in in essentially a vertical plane. This the present study by means of motion method of weight distribution is apparently pictures, employing the same methods always associated with a ginglymoid ankle- used for the caudate amphibians (Fig. 17). joint with well-developed, concavoconvex The peculiar structure of the tarsus in articular surfaces. Furthermore, when addition to the somewhat medially di- this type of ankle-joint is present, allowing rected head on the femur would seem also the foot to bend at right angles to the crus to give some clue as to the type of locomo-

Fig. 16.-The crocodilian tarsus. A, Alligatorellus (after Lortet), lateral view of calcaneum showing tuber on left; B, Alligator (from specimen); C, Astragalus and calcaneum of Alligator, slightly separated to show the nature of the articular surfaces.

(Fig. 21,E), and when the functional ankle- tion found in the thecodonts, dinosaurs, joint is between the calcaneum and the lacertilians, and even the advanced therap- crus, at least on the fibular side, there is a sids. The locomotion of the crocodile tendency for the calcaneum to develop a (Osteolamus) has been investigated by von tuber. The origin of the tuber will be Huene (1913) by direct observation and by discussed in the section on the therapsid footprints made in wet clay. The croco- tarsus, although the discussion applies diles have two methods of locomotion, one equally well for the thecodonts and croco- for rapid progression, the other for slow. diles. When moving slowly, the body is dragged The distal tarsalia are two in number. along the ground and the limbs are moved There is, of course, no free centrale, and in a very lizard-like manner. When Rabl considers it to be fused with the as- moving rapidly, however, the body is tragalus (Fig. 12,F). There is a meniscus raised from the ground and the femur between the medial tarsale and the base of swings backward in more of a vertical the fifth metatarsal and the calcaneum, plane. The ability of the crocodiles to which forms a socket in which the astragalus raise the body relatively far from the rotates. The contact between the lateral ground when compared with a lizard may 0 40

/ ~~~~~~~~~~~~~~~~~~bo

'I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- 1941 ] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 445 be associated with the extensive development than the same joint in the lizards, ment of the pubo-ischio-femoralis ex- functionally approaching the condition ternus and the adductor femoris as de- found in the higher therapsids and mam- scribed by Romer (1923). mals. The typical tetrapod pattern of A direct comparison between the rapid movement is present and the other details locomotion of the alligator and the loco- of locomotion common to all tetrapods are motion of an advanced therapsid such as also exhibited, as described for the am- Bauria appears plausible when the mor- phibian. phological similarities producing similar The movement of the hind-limb will functional effects are considered. It will be now be described in detail. At the be- recalled that the terminal head on the am- ginning of propulsion the crural portion phibian and on the primitive reptilian femur of the leg is almost vertical in position in allowed the latter to move only in the contrast to its forwardly directed position horizontal plane, no appreciable vertical in the caudate amphibians. This is a func- component of movement being present. tion of the flexibility of the ankle-joint and In the form under consideration, as in of the position of the femoral head, and is certain of the ancestral thecodonts, how- associated with the fact that during re- ever, the position of the femoral head per- covery the limb is simply swung forward mits a vertical component. The femoral into the proper position rather than being shaft is twisted in the same manner as in brought forward "overhand," as is the case the Lacertilia, the proximal and the distal in the amphibian. ends being rotated about 50 degrees out of When the foot is placed on the ground line with each other. The proximal end, at the beginning of the propulsion phase, furthermore, is directed away from the its long axis is usually almost parallel with long axis of the shaft also at an angle of the body, but the exact position is by no about 50 degrees in a dorsomedial direc- means constant, as the diagrams demon- tion. When the body is held close to the strate. Von Huene states that the foot is ground, the femur moves in the horizontal directed forward during propulsion in the plane as in the lizards. When it is raised crocodile, with the second toe parallel to from the ground, the medial bend in the the body axis, while the third and fourth proximal portion of the femur becomes ef- toes are nearly always bent more or less fective, permitting the femur to swing in a sideways. As can be readily observed, plane about 40 degrees off the vertical. this is true for both slow and rapid pro- In the dinosaurs, birds, some therapsids, gression. The long axis of the foot, how- and mammals the horizontal component ever, remains almost parallel with the longi- is, of course, reduced to zero. tudinal body axis, by the end of propul- In both the advanced therapsids and sion being oriented at an angle of about crocodiles the tuber on the calcaneum in- 20 to 30 degrees to this axis. Likewise in creases the leverage action of the foot by the therapsids there is every reason for be- increasing the effective distance through lieving that the feet were directed forward which certain of the flexors, which partially throughout propulsion as they were in or entirely insert on the tuber, act (Fig. the primitive tetrapods. In other words, 21,E,H). Hence, the foot is able actively the laterally directed foot of many rep- to participate in the propulsion phase. Von tiles is entirely secondary and not a herit- Huene points out that when the crocodile age character as many have supposed. is moving very rapidly, the metatarsals and Under these conditions, stresses are ap- phalanges appear to snap plantarward, plied transversely across the foot but not pushing the body forward. This motion he to the same degree that they would be ap- states, is favored by the tuber on the cal- plied in Morton's hypothetical tetrapod caneum, giving the extrinsic flexors of the in which the feet are at right angles to the foot a "lever grip." Finally, the ankle- body. These stresses may be partly re- joint, with its concavoconvex articular sponsible for the low transverse arch in surfaces, allows greater freedom of move- the lepidosaurian foot, although there is 446 Bulletin American Museum of Natural History [Vol. LXXVIII no sign of such an arch in the alligator foot. tween the foot and the ground is required Morton (1935) is of the opinion that the for a successful thrust, and, furthermore, transversely applied stresses by producing successful propulsion requires a firmly the superposition of the astragalus on the consolidated lever which, in this case, is calcaneum were responsible for the crea- simply the elongated metatarsals. In the tion of the transverse tarsal arch in the case of the amphibian, as pointed out therapsids and mammals, which in the earlier, the foot, although it may be con- light of the evidence here presented would sidered as a compound lever system as far appear improbable. as its intrinsic movements are concerned, During rapid propulsion, the femur is but a flexible base giving support to the moves backward at an angle of about 60 propulsive movement of the leg. degrees with the vertical plane. This has The recovery phase of the reptilian and the effect of increasing the average angle mammalian limb is relatively simple com- between the femur and the crus during pared to the amphibian. That is to say, locomotion. The analysis of the forces at in all tetrapods having a terminal head on work on the amphibian limbs during loco- the femur the recovery phase is very similar motion is again applicable to reptilian to, or some modification of, the recovery locomotion. The component c (Fig. 8) described for the caudates. As pointed which draws the limbs toward the body, out several times previously, the medially is, however, not nearly as large with the directed head allows the femur to move in femora in the vertical position. Although a more vertical plane. Hence, during the values of the various forces involved recovery the crus is flexed on the thigh may differ from those for the amphibian, and the foot on the crus, and the entire there is no radical change in the nature of limb simply swings forward. As recovery the forces until the femur assumes the progresses, the segments are extended so vertical position and the effect of gravity that the foot is parallel with the ground is largely overcome by the position of the and the crus is almost vertical by the time limbs under the body. Under these con- the foot contacts the ground at the begin- ditions, the force w, which tends to cause ning of propulsion. Although therapsid the foot to slip sideward, is practically locomotion differed from this only in de- non-existent, and the force b requires but tail, the causes for this difference are of little muscular exertion as most of the great importance (cf. section on therapsid body weight is transmitted directly through foot). the bones in the vertical plane. It is obvious that the alligator foot does THE TARSUS OF THE DINOSAURS, PTEROSAURS not slip off the ground as does the am- AND BIRDS phibian foot, but that it is lifted from the The final stage in the perfection of the ground immediately after the weight is mesotarsal joint is found in the Saurischia, transferred to the heads of the metatarsals. Ornithischia, Aves, and Pterosauria. The- reptilian foot lacks the suppleness of They have paralleled one another in pro- the amphibian foot, particularly in the ducing a tarsus in which the astragalus and case of the digits. Furthermore, in the the calcaneum tend to be proximodistally amphibian foot the metatarsals, phalanges, compressed, and in which the number of and the distal portion of the tarsus are the the more distal tarsal elements is greatly only parts making effective contact with reduced. the ground, while in the alligator the en- In the primitive members of both the tire foot contacts the ground almost simul- Saurischia and the Ornithischia (Fig. 11), taneously at the beginning of propulsion, the tarsus consists of a number of distinct as von Huene also noticed. The proximal elements that have been identified as the portion of the foot is then raised by flexion astragalus, calcaneum, and two to four between the metatarsals and the phalanges. tarsalia. A tuber calcanei was retained With the foot participating in the propul- in the Hallopodidae, but was lost in all sive effort, a certain minimum contact be- other dinosaurs, Its implications are dis- 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles AA7 cussed in the section on reptilian muscula- tibia, the fibula being entirely absent. ture (p. 462). In the more advanced There are but two tarsalia, the lateral one, members of both orders, there is a tendency undoubtedly the result of fusion of several for the astragalus and the calcaneum to of the elements in this series, is the larger. articulate very intimately with the crus These were firmly united by ligaments to and for the crural elements to do the same each other and to the metatarsals. The with each other, preventing any possible metatarsal joint is again concavoconvex. movement between them. The distal Exactly how the pterosaurs used their surfaces of the astragalus and calcaneum hind-limbs is still a moot question and the are rounded off and form a continuous matter will not be speculated on here, ex- articular surface. This transversely cept to say that movement must have been rounded surface, in the living animal, largely restricted to the anteroposterior must have fit into a concave surface com- plane as in the birds. Furthermore, the posed of menisci and ligaments, the ossified presence of a well-developed and medially tarsalia forming all or part of the floor of directed head on the femur indicates that this concavity. the entire hind-limb moved in the same The tibia was the main weight bearing plane. axis, the fibula being a much thinner bone. In the birds (Fig. 11), the consolidation Apparently, whenever the femur develops has been carried a step further, with the a medially directed head, the tibia becomes astragalus and the calcaneum fusing to- the main weight bearing axis of the crus. gether and ossifying with the tibia. The This is true of all the members of this fibula has been reduced to a mere vestige. group, and of the advanced therapsids and The remaining tarsalia indistinguishably mammals as well. Accompanying the fuse with the remaining metatarsals, which dominance of the tibia, there is always a have likewise fused. This extensive fu- consolidation of the proximal part of the sion of the tarsal elements not only makes tarsus. In the dinosaurs, this was ac- for rigidity at the tarsal joint, but also has companied by the development of a flange the effect of preventing movement in more on the distal portion of the tibial shaft that than one plane. The restriction of move- extended laterally around the fibula, for ex- ment to the anteroposterior plane is ac- ample, in Tyrannosaurus, Trachodon, Proto- tually due to a combination of factors. ceratops, and Stegosaurus. This tibial The tibia and the tarsometatarsus are flange was undoubtedly firmly attached to firmly united by ligaments, forming a the fibula by ligaments, thus reenforcing capsulated, diarthrodial joint. Secondly, the ligamentous attachment of the as- the proximal end of the tibia and the base tragalus with the calcaneum. Further- of the tarsometatarsus are transversely more, as is well known, the astragalus de- widened, and the former is convex and the veloped a dorsal process in some members latter is concave. These factors, together of both orders; for instance, Tyranno- with the arrangement of the insertion ten- saurus and Trachodon, which attached it dons of the extensor and the flexor muscles, even more securely to the tibia. That prevent any movements of one element on this flange should have developed on the the other in the lateral plane. anterior rather than the posterior surface of the astragalus may be associated with TYPE III the fact that in the bipedal forms the THE PELYCOSAURIAN TARSUS AND ITS digits, meeting the tarsus at an angle DERIVATIVES might have tended to rotate the astragalus The morphology of the pelycosaur tarsus anteriorly during propulsion. The dorsal has been very thoroughly described by process, being firmly attached to the tibia, Romer (1940), including a discussion of counteracted this tendency. its functional aspects. The complete tar- The pterosaur tarsus (Fig. 11) is quite sus of Varanosaurus (A.M.N.H. No. 4174) dinosaur-like. The astragalus and the and parts of the tarsus of Dimetrodon, to- calcaneum are ossified and fused with the gether with a cast of an entire foot of the 448 Bulletin American Museum of Natural History [Vol. LXXVIII latter, have been available for study. articulation required that the tibia assume This tarsus has retained to a greater de- a diagonal position as the femur moved gree than did the eosuchian the unmistak- backward. This difference in level, how- able mark of the older captorhinomorph ever, did decrease the amount of tilting pattern (Fig. 18). There are two proximal required of the tibia. elements, the astragalus and the calca- The L-shaped astragalus, as already neum; two centralia; and five distal tar- noted, first appeared in the cotylosaurs, salia. The presence of two centralia in specifically the captorhinomorphs. It was the pelycosaurs is a unique condition apparently secondarily lost in the other among reptiles. Labidosaurus has but cotylosaurian types except possibly in one and but one is present in Dimetrodon, Procolophon. It will be recalled that it Youngina, and other primitive reptiles. persisted in the Mesosauria and the more As Romer points out, the astragalus is primitive Protorosauria. There is every L-shaped, the vertical arm of the L articu- reason to believe that this form of as- lating proximally with the fibula by a tragalus served the same purpose in all hinge-joint. The horizontal arm of the these types, that of directing the long axis L, however, articulates with the tibia by of the foot forward and relieving the stress means of a well-rounded articular surface, in the proximomedial region of the tarsus permitting gliding movements for the at the same time. first time between these elements. In the The calcaneum of the pelycosaurs is types derived from the eosuchians, as dorsoventrally compressed and only thick- just pointed out, the torsion on the medial ened where it articulates with the fibula, side of the foot was relieved in several astragalus, and tarsalia four and five. cases by the shortening of the tarsus on the The more distal elements are in most medial side, but mostly by the develop- cases merely rectangular or rounded, and ment of a medially directed femoral head. must have had cartilaginous peripheries in In the pelycosaurs, this same torsion prob- life. It would appear that Romer is cor- lem was partly alleviated by the formation rect in stating that there was no movement of a rounded articular surface on the as- between these elements. tragalus, permitting the tibia to move The locomotor methods of the pelyco- freely over the astragalus as propulsion saurs have been discussed by Romer. progressed. The movements of the femur with its The long axis of the foot must have been terminal head, as he deduces them, were directed forward during locomotion. The almost identical with those described for ability to do this, when the proximal ends the primitive tetrapod. Actually, the of the tibia and the fibula were directed only real advance is the organization of the laterally, he attributes to the ability of the crurotarsal articulation on two different freely movable astragalotibial joint to allow levels, and the rounded astragalotibial the necessary rotational movement. There articulation relieving the stress produced is one other feature, however, which must by the torsion in the medioproximal re- not be overlooked, namely, that the tibial gion of the tarsus. While this may be articulation is on a more distal level than considered as an advance, it has actually the fibular (Fig. 21,B). Because of this, modified the locomotor movements very the foot has to be oriented with its long slightly, the principal difference being, as axis directed forward, otherwise the tibia just pointed out, that the tibia was no and the fibula could not articulate with longer required to assume a diagonal posi- the foot at the same time. Although the tion during the last part of the propulsion tibia articulates with the tarsus at a more phase. distal level than the fibula in the primitive According to Romer, the femur has a tetrapods, the pes was not directed for- greater forward than backward swing. ward throughout propulsion because of this At the beginning of propulsion, the femur condition. As already pointed out, the was lower distally than proximally, and relative immobility of the tibiotarsal the crus was directed anteriorly and some- 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 449 what outward from the knee. At the end THE THERAPSID TARSUS of propulsion, the femur was raised dis- According to Romer, the pelycosaurian tally and was rotated counter-clockwise family Sphenacodontidae, which includes (on the left side), resulting in the tibial Sphenacodon and Dimetrodon, is most articulation facing directly downward, probably ancestral to the therapsids. Al- while the crus was in an almost vertical though this may be true, it is rather dif- position. ficult to reconcile the relative specializa- The method of weight transfer during tions of the Dimetrodon tarsus with the propulsion was, even at this stage, typical rather primitive tarsus found in the Dro- of the crurotarsal foot. At the beginning mosauria, and even the Dicynodontia and of propulsion, the body weight was trans- the Dinocephalia. (The tarsus of Hapto- mitted to the ground through the astragalus dus is, however, more primitive.) As far and the calcaneum by way of the heel as the tarsus alone is concerned, that of pads. As soon as the body moved for- Ophiacodon appears better suited to be ward and the posterior part of the foot was an ancestral type, as Romer (1940, p. raised from the ground, the weight was 180) implies in noting the resemblance transferred to the heads of the metatarsals between the tarsus of Ophiacodon and Labi- or the ball of the foot. The wide tarsus, dosaurus. plus the length of the metatarsals, must The therapsid tarsal pattern is again the have imposed tension-compression stresses same as that found in the pelycosaurs, ex- of considerable magnitude on this region cept for the loss of the medial centrale when the heel region was off the ground. (Fig. 18). This centrale tends to be dis- These stresses could have been counter- placed medially in the advanced pelyco- acted only by the efforts of the plantar saurs and is finally lost (Romer, 1940). ligaments and the intrinsic muscles of the The remaining centrale is the homologue foot. Not until the mammalian stage of the mammalian navicular and will hence- was reached, was a longitudinal tarsal forth be called by that name. arch created to counteract the stresses. The tarsus of the therapsids evolved Even in the mammals, the integrity of this along several distinct lines, as can be arch is only maintained by the action of readily seen on the accompanying chart the ligaments and the intrinsic muscula- (Fig. 18). The Dromosauria, Dinoce- ture (Jones, 1941). phalia, Dicynodontia, and Theriodontia The tibial articular surface on the pely- probably represent four distinct evolu- cosaur astragalus shows one further point tionary lines and the differences in the of interest. As Romer points out, this sur- tarsus tend to support this conclusion. face is directed not only medially but also There is some indication that the L- dorsally. In other words, it has migrated shaped astragalus was carried over into somewhat on to the dorsal surface of the the therapsids. The astragalus of the astragalus (Fig. 21,B). This tendency, gorgonopsians Lycaenops and Aelurogna- again associated with the relief of the tor- thus is somewhat of this shape. There was sion on the medial side, must have first ap- a definite tendency in all the suborders peared in the cotylosaurs, and there is defi- to transform the astragalus into a more nite evidence of it in Labidosaurus. In rounded or cubical element, with its tibial the advanced therapsids, as will be pointed articular surface either on the proximal out shortly, not only has the tibial articular border, as in the Dromosauria, Dinoce- surface migrated entirely on to the dorsal phalia, and Dicynodontia, or on the dorsal surface of the astragalus, but the fibular or as in articular surface has done likewise on both anterior surface the Gorgonopsia the astragalus and the calcaneum (Fig. and the groups derived from this infraorder. 21,H). In the more specialized pelycosaurs, In all of them the crurotarsal joint tends to the tibial articulation is more dorsally lo- be on the same level proximodistally, rather cated (Dimetrodon) than in the primitive than the tibial articulation being more distal types (Ophiacodon). than the fibular. 450 Bulletin American Museum of Natural History [Vol. LXXVIII

Alhailsi Zficf^dos vaurwt Efr,c,o/crcertar 7heroce/ha/ia Cynodontza Baartavorp7ha

Mipp,osauras 6oryon'wpsaz/

PI, oJrkerlz 6alechzras dinocehabaha Dromasauraz Dicynodom' Z& THERAPSMIA

I'aranosaurus DELZYCOS.AURIA Fig. 18.-The therapsid taxsus. (Compiled from the literature and specimens.) This leveling off in the crurotarsal articu- body without the necessity of the tibia lation had no great effect on the position crossing in front of the fibula as was the of the foot relative to the body during lo- case when the head was terminal. comotion. In all therapsids the femur has a head that is medially directed to a greater THE DROMASAURIAN TARSUS or lesser degree, but minus the twisting of The Anningiamorpha having been dem- the ends in opposite directions, depending onstrated to be pelycosaurs (Romer, on the group and the extent of specializa- 1940) and the position of Galesphyrus, tion in the direction of the mammals. which Broom referred to this group, now This undoubtedly compensated for the open to question, the dromasaurian tarsus leveling off, and permitted the long axis of represents the earliest specialized offshoot the foot to parallel the long axis of the from the primitive therapsid stock. The 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 451 tarsus of Galechirus (A.M.N.H. No. 5516) THE DINOCEPHALIAN TARSUS is the only well-preserved dromasaurian The dinocephalian tarsus is very poorly tarsus known (Fig. 18). The astragalus known considering the amount of ma- has lost the L-shape and has become a terial belonging to this suborder that has triangular element, with a rounded sur- been uncovered. Apparently, the only face for articulation with the tibia. The titanosuchid tarsus known is that of calcaneum is disk-shaped and dorsoven- Jonkeria (Broom, 1939) and it is incom- trally compressed. As in the pelycosaurs, plete, the astragalus and the calcaneum its proximal and lateral borders are being the only elements preserved. It is thickened for articulation with the fibula quite possible that the titanosuchid tarsus and the astragalus, respectively. Its cen- generally was but poorly ossified. The ter is slightly concave, a characteristic of astragalus of Jonkeria (Fig. 18) is rounded all therapsid calcanei, even including the in much the same fashion as in Galechirus bauriamorphs. and undoubtedly there was movement be- The navicular is directly distal to the tween it and the tibia. The tibial articu- astragalus in the typical mammalian posi- lation was at a somewhat more distal tion. It articulates with the three medial level than the fibular, a probable heritage tarsalia that are in turn homologous with character derived from the pelycosaurs. the three mammalian cuneiforms. The As the illustration shows, however, the cuboid is a triangular element that barely astragalus is not L-shaped. The fibular articulates with metatarsal five. It is facet on the astragalus appears to be also not possible to determine whether the somewhat rounded. The calcaneum is fifth metatarsal actually articulated with again a disk-shaped element, compressed the calcaneum in life. dorsoventrally and somewhat concave on The composition of the therapsid (and its dorsal surface. It is thickened along mammalian) cuboid bone presents the same the proximolateral and medial borders sort of problem as that of the astragalus. where it articulated with the fibula and the Both Schmalhausen (1908b) and Holmgren astragalus. (1933), studying the embryology of the The only tapinocephalid tarsus that has tarsus of Mus (Fig. 12,G) and Sus, are of been described is that of Moschops (Greg- the opinion that the cuboid is composed of ory, 1926). Since it was assembled from tarsale four and tarsale five. Dollo (1929) disassociated elements the distal portion was also of this opinion. Most paleon- is regarded as largely conjectural (Fig. 18). tologists, on the other hand, consider it the The element thought to be the astragalus homologue of the fourth tarsale only. is much more dorsoventrally compressed Again, it would appear impossible to solve than that of Jonkeria. The tibial articu- this matter with paleontological evidence lar surface is but slightly rounded as com- only, and its dual origin must be considered pared with Jonkeria and is restricted to as a distinct possibility. the proximal border. The calcaneum is The ankle-joint in the dromasaurians in all respects similar to that of Jonkeria. must have been crurotarsal in position Although the distal tarsals could not be and must have functioned in the same identified with certainty, there is every manner as that of the pelycosaurs, with a reason to believe that there was the same freely movable joint between the tibia and number of elements, arranged in the same the astragalus, and a hinge-joint between manner as in Galechirus. the fibula and the astragalus and the cal- The location of the ankle-joint in Mos- chops is somewhat of a problem. The caneum. articular surfaces on the astragalus and the The Dromasauria possessed the mam- calcaneum are relatively restricted in size, malian digital formula of 2-3-3-3-3. indicating but little movement between This must have been acquired indepen- the crus and the tarsus. It would appear dently, as many of the more advanced probable that the functional joint was ac- therapsids have the old reptilian formula. tually -located more distally between the 452 Bulletin American Museum of Natural History [Vol. LXXVIII tarsalia and the metatarsals or between the tarsus is not ossified, except for two the proximal row and the next more dis- very small distal nodules in the Endo- tal elements. This point cannot be settled thiodon tarsus (Fig. 18). unless a foot is found preserved in place. THE GORGONOPSIAN TARSUS THE DICYNODONT TARSUS The most primitive gorgonopsian tarsus The tarsus of the dicynodonts varies known is that of Lycaenops (Fig. 18). A greatly in the extent of ossification. In complete and perfectly preserved specimen the case of Emydopsis (Broom, 1906, has been studied in great detail (A.M.N.H. 1932) and Kannemeyeria (Pearson, 1924), No. 2240) and a model has been con- the tarsus is fully ossified, while those of structed in order to restore the tarsal ele- Lystrosaurus (Watson, 1913a), Dicynodon, ments in their probable natural positions. and Endothiodon are but partly so. The The astragalus is more or less rectangu- lack of ossification in this group has gen- lar in shape and is well rounded on its erally been associated with a semiaquatic or dorsal surface. This area is divided into a fully aquatic habitus, but again, to some two parts or articular surfaces by a shallow degree at least, the cartilaginous state may groove running in a proximodistal direc- be genetic, as it appears to be in the cau- tion. The medial articular surface for date amphibia. the tibia is directed in a medial and dorsal The tarsi of Emydopsis (Fig. 18) and direction in the same manner as is the Kannemeyeria are very similar to each homologous surface on the. pelycosaur other and to that of Lycaenops. The astragalus. The lateral fibular articular astragalus has a rounded dorsal surface surface is relatively smaller and is in line for the articulation of both the tibia and with the rounded proximal border of the the fibula and the calcaneum also appears calcaneum, which, of course, also articu- to have a rounded fibular facet on its lates with the joint. The astragalus over- dorsal surface. The location and relative laps the calcaneum to a slight extent proxi- size of these facets indicates a functional mally and the calcaneum overlaps the crurotarsal joint with the foot assuming a astragalus to a greater extent distally. semiplantigrade or plantigrade position There is no evidence that this is due to during propulsion. The femoral head was crushing as the elements have been clearly medially directed to a marked degree, preserved in almost their natural posi- giving the hind-limbs a pose clearly re- tions, with no sign of mediolateral compres- sembling that indicated in Pearson's res- sion. In Broom's drawing of a Lycaenops toration of Kannemeyeria. tarsus (1930) there is, however, no indica- The single specimen available (A.M.N.H. tion of this interlocking. This method of No. 5635) of the foot of Dicynodon platy- articulation absolutely precluded any possi- ceps would seem to indicate that the as- bility of movement between the astragalus tragalus and calcaneum are fused, although and the calcaneum. As will be seen this is not certain. The tibial facet is shortly, the development of a medial proc- rounded and dorsally located, as is the ess on the calcaneum was favored for the fibular facet, indicating that the foot could same reason in the bauriamorphs. assume the same position as that of Kanne- The calcaneum is very similar to that meyeria. Except for an unidentifiable found in Galechirus; in other words, it still splinter, the remainder of the tarsus is un- resembles the primitive therapsid type, ossified. being disk-shaped and concave in the The astragalus and calcaneum of Endo- middle of the dorsal surface. Its medial thiodon (A.M.N.H. No. 5613) and Lystro- border is notched for the reason men- saurus (Watson, 1913) are, in contrast to tioned above. The proximal border is their form in the genera just mentioned, relatively thicker and the curved surface dorsoventrally compressed with the tibial more extensive than in Galechirus. and fibular articular surfaces restricted to In Lycaenops there is evidence for the the proximal borders. The remainder of first time that the fibula was able to rotate 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 453 somewhat on the calcaneum in the vertical The tarsus of Hipposaurus (Fig. 18) is plane. This slight amount of movement of great importance as it is the first known reduced still more the torsion at the ankle- example of a gorgonopsian tarsus in joint, which must have increased somewhat which the calcaneum possesses a true when this joint leveled off in the primitive tuber. In fact, the resemblance between therapsids. the calcaneum of Hipposaurus and Bauria The cuboid is transversely elongated, is remarkably close. In both cases the tapering laterally to almost a point. The body of the calcaneum is still basin- other elements do not require special com- shaped on its dorsal surface as in the more ment, except that the navicular is oval primitive types. The fibular facet, how- in outline and barely contacts the ento- ever, consists of a rounded knob that is cuneiform. elevated above the surface of the body. When the Lycaenops foot is properly The tuber projects in both cases backward oriented, the midtarsal axis passes either just lateral to the facet, and is relatively along the third digit or slightly to the more robust in Hipposaurus than in fibular side of that digit and then approxi- Bauria. mately between the astragalus and the The nature of the astragalus cannot be calcaneum. If this axis is raised into a made out from Boonstra's description. vertical position, it is found that a rather He considers it to be a composite element, well-developed transverse arch supported the "conjoined" intermedium and tibiale, by plantar pads was required in order that with each part containing a knob-like proc- all the heads of the metatarsals could touch ess. It is quite possible that further prepa- the ground at one time. It is highly prob- ration would reveal that there is but a able that the foot was semiplantigrade single element, the astragalus, with two and not digitigrade as figured by Broom rounded surfaces, the more medial one (1932, p. 129), although he has described for the tibia and the more lateral one for it as plantigrade (1929, p. 35). In fact, the fibula, as in Bauria. there is no reason for believing that any The more distal portion of the tarsus is of the therapsids were digitigrade, although also very difficult to interpret. Boonstra they have been commonly restored as is of the opinion that the medial, proximal such. This matter has been fully dis- mass may also contain the navicular cussed in a previous paper (Schaeffer, which is absent as a distinct element. In 1941). any case, the illustration shows but four Two more gorgonopsian tarsi have re- distal elements instead of the usual five, cently been described by Boonstra (1934), and the presence of a fifth may be strongly belonging to Aelurognathus microdon and suspected. Hipposaurus boonstrai. In both cases The method of locomotion must have Boonstra considers the proximal row of the been almost mammalian. The femoral tarsus to contain three elements: the head is medially directed almost to the tibiale, the intermedium, and the fibulare. same extent as in the mammals. As the Boonstra does admit, however, that his head became more medially directed, the interpretations may be incorrect in both importance of having the tibia and the cases, which is very probable. fibula articulate on the same level is ob- The tarsus of Aelurogrcathus (Fig. 18) vious. If the pelycosaur condition were undoubtedly contains all the elements still maintained, the foot would be in- possessed by that of Lycaenops. The verted and almost supinated to the extent element considered by Boonstra to be a of causing the animal to walk "pigeon- tibiale would appear, however, to be the toed." The axis of the ankle-joint was at slightly displaced navicular. It has the all times at approximately right angles to same oval shape and relative size as the the long axis of the body. This relation- Lycaenops navicular. His intermedium is ship imposed the least amount of lateral then, of course, the astragalus. strain on the hip-, knee-, and ankle-joints. 454 Bulletin American Museum of Natural History [Vol. LXXVIII

THE METHOD OF PHALANGEAL REDUCTION AS saurus. Apparently the fibular facet is ILLUSTRATED BY THE GORGONOPSIA not a knob-like elevation as in Hippo- The question of the reduction from the saurus and Bauria but is simply a part of reptilian to the mammalian phalangeal the body. The astragalus does not have formula has been discussed by Broom its dorsal surface divided into facets and (1913b). In the manus of Scymnognathus in this respect resembles Whaitsia. The there figured, and also in the manus of navicular articulates with all three cunei- Lycaenops (Broom, 1929), digits one and forms. The entocuneiform tends to simu- two and five have the mammalian for- late the first metatarsal in its proxi- mula, while digit three has four phalanges, modistal elongation. This tendency, and digit four has five phalanges. In which is also evident in Bauria and Emy- digit three, the second most distal phalanx, dopsis, has been noted by Romer (1941) and in four, the second and the third, are in the tarsus of the pelycosaurs. One mere disks. The second most distal pha- curious feature is the apparent presence lanx of the third digit of the Hipposaurus of a fifth tarsale. This is apparently the pes is also of this sort. It would appear only case among the therapsids where this that certain of the phalanges were reduced element is separately present. If it does to disk-like proportions and finally either not turn out to be simply a part of the fused with the adjoining unreduced ones fourth tarsale or cuboid, it presents strong or were eliminated altogether. The cyno- evidence for believing that the mammalian dont Thrinaxodon also shows this condi- cuboid is a compound bone. The pres- tion, having exactly the same arrangement ence of a cartilaginous fifth tarsale must of the disk-like elements. also be considered as a distinct possibility in the case of Bauria. THE THEROCEPHALIAN TARSUS Broom figures (1932, p. 76) an unidenti- Whether or not the gorgonopsians will fied scaloposaurid that apparently also prove to be the ancestors of the thero- has a tuber on its calcaneum. The foot cephalians and the more advanced the- has, however, unfortinately been restored rapsid groups, it seems evident that the as digitigrade, a criticism that also applies tarsus of Lycaenops is clearly, at least to Watson's restoration of the bauria- structurally, ancestral to the tarsus of morph Ericiolacerta. these forms. Of particular importance is the rounding off of the tibial and fibular THE CYNODONT TARSUS articular surfaces on the astragalus and The cynodont tarsus has not been com- the calcaneum, and of the migration of pletely described, although it is known in these surfaces on to the dorsal surface of Cynidiognathus, Thrinaxodon, and Diade- these elements. modon. That of Thrinaxodon (Fig. 18) as The tarsus of the therocephalian Whait- figured by Broom (1932, p. 270) has an sia (Broom, 1930a, Fig. 4,B) is very similar astragalus and calcaneum very similar to to that of Lycaenops. The calcaneum that found in Lycaenops. The calcaneum is still disk-like with no indication of a is oval in shape and lacks a tuber. Broom tuber (Fig. 18). The astragalus is very quotes Watson as saying that a tuber is similar except that its dorsal surface is present on the calcaneum of Diademodon not divided into distinct tibial and fibular (1930b, p. 133)., The other tarsal ele- facets as is that of Lycaenops. The navicu- ments resemble very closely those of Ictido- lar is relatively wider transversely and suchoides. As pointed out above, the articulates with the entocuneiform, but phalanges appear to be in a state of reduc- barely contacts the ectocuneiform. tion toward the mammalian formula. The tarsus of Ictidosuchoides (Broom, 1938) is of great interest in that the cal- THE BAURIAMORPH TARSUS caneum possesses a tuber (Fig. 18). The The tarsus of the bauriamorph Bauria body of the calcaneum is transversely (Fig. 18) has been described by Broom widened as in the gorgonopsian Hippo- (1937), and Boonstra (1938), and further 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 455 prepared and redescribed by Schaeffer had more than one axis of rotation, produc- (1941). Its more important features may ing dorsoplantar movement at the crurotar- be listed here. The calcaneum has a well- sal joint. As will be seen later, the compli- developed tuber and has lost its disk-like cated movements of the mammalian foot or oval outline, being more rectangular in are directly associated with the superposi- shape. The fibular facet is in the form tion of the astragalus on the calcaneum. of a rounded knob that is directly opposite Supination and pronation, or any rolling the fibular facet on the astragalus. The movement of the foot from side to side, astragalus has two distinct articular sur- must have been negligible as they could faces divided by a shallow groove much as only have occurred by an actual disloca- in Lycaenops. It has a rather extensive tion of either the tibio- or the fibulotarsal articulation with the cuboid, a condition joints. It is evident therefore that the usually found in the ungulate mammals. locomotion of the therapsid, no matter how The navicular articulates only with the mammal-like the construction of the hind- meso- and the ectocuneiform, barely con- limb, lacked the fine degree of postural ad- tacting the entocuneiform. The latter, justment that the relatively unspecialized as pointed out above, simulates the first mammalian foot makes possible. Move- metatarsal. The cuboid only articulates ment, particularly over uneven terrain, with the fourth metatarsal, the fifth has no must have been relatively slower as more apparent bony support whatsoever. This effort had to be expended in preventing a matter is fully discussed in a previous shifting of the center of gravity during paper (Schaeffer, 1941). The plantar progression. With the body raised well surface of the calcaneum has a short off the ground and the legs swung under medial process that extends under the the body, this factor was of greater im- lateral border of the astragalus, and it has portance than in the primitive tetrapods. been suggested that this may be the begin- It would thus appear that superposition ning of the sustentaculum tali. The cu- was favored when the hind-limbs assumed boid also has a medially directed process the mammalian position, as a means of extending under the navicular for a short vastly improving the mechanism of pos- distance, which is probably a specialized tural adjustment. feature for additional consolidation. In other types in which the femur as- The one other baurimorph tarsus that sumed a vertical position, such as the dino- has been described is that of Ericiolacerta saurs and the birds, a relatively smaller (Watson, 1931). The astragalus and the portion of the foot touched the ground calcaneum (Fig. 18) are very similar to at any one time, and also the toes were those found in Bauria. The more distal capable of a certain amount of indepen- tarsals were apparently poorly ossified. dent movement. Furthermore, the archi- The navicular, ento- and mesocuneiforms tecture of the foot with the extrinsic are represented by nodules, and there is no flexors inserting directly into the plantar sign of the ectocuneiform and the cuboid. surface would assist in necessary adjust- To date, the ictidosaurian tarsus is un- ments. known. If this group is as close to the mammals as is now supposed, the cal- ASTRAGALAR SUPERPOSITION caneum should have a well-developed tu- The superposition of the astragalus in ber, and possibly the superposition of the its relationship with the calcaneum is a astragalus upon the calcaneum is in an distinctly mammalian characteristic and incipient stage. the one major difference between the mammalian and the bauriamorph foot. It is THE FUNCTIONING OF THE THERAPSID FOOT the final step in the transformation of the In spite of all the advances in the con- therapsid foot into the mammalian. The struction of the foot toward the mammalian construction of the tarsus among mammals condition which appeared in the higher is subject to but relatively minor varia- therapsids, there is no evidence that the foot tion associated with the posture of the 456 Bulletin American Museum of Natural History [Vol. LXXVIII foot. This is in sharp contrast to the great weight was transferred both to the as- variation in the structural design of the tragalus and the calcaneum through the tarsus among the reptiles. tibia and the fibula, although most of the The superposition was brought about at weight must have been carried by and first not so much by an actual overriding transferred to the former. Thus, it seems of the calcaneum by the astragalus as by evident that while the calcaneum trans- the development of a medial process on mitted the greater part of the muscular the calcaneum that finally extended under force, the astragalus transmitted the the astragalus. This process is present greater part of the body weight to the more in an incipient stage in the foot of Bauria. distal portion of the foot. As a result of As pointed out in the description of the the superposition, a certain proportion of Bauria foot, the tuber calcanei appeared both forces was transmitted to both proxi- long before the astragalus became super- mal elements and then through them to imposed on the calcaneum, and although the more distal portions of the foot, spe- the former may be of arboreal significance, cifically, directly to the heads of the meta- the development of a tuber definitely is not. tarsals. The fulcrum was then located at The extension of this medial process of the talocrural joint. the calcaneum, which may be called the Finally, the appearance of the susten- sustentaculum, produced three important taculum was undoubtedly followed by a functional changes in the therapsid foot. certain amount of actual superposition. First, the fibula lost its important func- It was this superposition that actually tional contact with the tarsus as a weight brought about the loss of the calcaneofibu- bearing element and the tibia became the lar articulation. With the development main weight bearing axis of the crus. A of the extensive articular surface between single weight bearing column for each leg the astragalus and the calcaneum in a is a less stable means of support than a plane almost paralleling the plane of the double column, for the latter tends to foot, the functional subastragalar joint stabilize the crus on the foot in the trans- (talocalcaneal) was established, permitting verse plane. The two remaining changes for the first time so-called pronation and are associated with maintaining stability supination. between the crus and the foot. The effect of the superposition of the Superposition caused the astragalus astragalus upon the calcaneum has been and the calcaneum to act as a functional discussed by W. Abel (1930). He believes unit, at least so far as movement of the that it caused two major changes in the foot, foot in the sagittal plane is concerned. one in the creation of a two-layered condi- The development of the tuber changed tion in the tarsus, and the other, a result of the foot into a lever of the first order the first, an alteration in the axis of motion with the fulcrum at first at the joint of the foot. Along with the superposition formed by the tibia and the fibula on the one of the astragalus, the navicular was hand and the astragalus and the calca- pushed above the cuboid, forming, along neum on the other. This must have been with the astragalus, what Abel calls the a rather inefficient system for, with the upper layer. The cuneiforms, no longer astragalus and the calcaneum side by side, able to remain in the same horizontal most of the force exerted on the tuber by plane as the cuboid, were thrown into an the triceps surae must have been trans- arcuate arrangement. As the cuboid was mitted directly through the calcaneum to tilted to complete the arch, it resulted in the distal portion of the foot, with very the creation of the transverse arch of the little being transferred to the foot via the mammalian foot. Morton (1935, p. 17) astragalus. There was, therfore, an un- also is of the opinion that the transverse equal distribution of the muscular force arch was created by the superposition of applied at the tuber. This arrangement the astragalus. had no appreciable effect on the weight In order that the subastragalar joint distribution at this stage, as the body might be functional, or in other words, in 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 457 order that the foot might pronate and su- (1935a). They are of the opinion that the pinate, the astragalus assumed a functional longitudinal arch of the human foot is the role as part of the crus. This made it result of the transverse tarsal joint being necessary for movement to occur between locked in a plantar-flexed position. the astragalus and the navicular. This If the longitudinal arch be defined as an situation undoubtedly favored the de- arch resulting from permanent plantar- velopment of a rounded head on the as- flexion of the transverse tarsal joint, tragalus and a concave proximal surface on strictly speaking, man may possibly be the navicular. Abel's contention that the only animal possessing it, although this the tarsus may be divided into two layers point will bear future investigation. This can, it would seem, only apply to the proxi- joint reaches its extreme in mobility in the mal elements of the tarsus, in that the artiodactyl where it is just as functional astragalus moves on the calcaneum. The as the upper ankle-joint. In this case the more distal portion of the tarsus, although arrangement of the joint planes allows it is thrown into an arc, cannot be con- dorsiflexion only. By definition, therefore, sidered as either structurally or functionally there is no true longitudinal arch in the two-layered. therapsids or any of the other reptilian The one remaining joint that is char- groups. Furthermore, there is no indica- acteristic of the mammalian foot is the tion that a transverse tarsal joint was ever transverse tarsal joint. It is situated be- developed in any of the therapsids with- tween the astragalus and the calcaneum out the support of plantar pads, as in the on the one side and the cuboid and the Lycaenops foot. The articular surface be- navicular on the other. The function of tween the astragalus and the navicular is a this joint in man and the chimpanzee has simple, plane surface that allowed no been discussed by Elftman and Manter movement.

THE EXTRINSIC AND INTRINSIC MUSCULATURE OF THE REPTILIAN HIND-FOOT AND ITS RELATION TO LOCOMOTION Although there are some variations in shift in origin undoubtedly occurred when the cruropedal and pedal musculature in the femorotibialis and the iliotibialls be- the existing orders of limbed reptiles, it is came the principal extensors of the crus. possible, as the work of Ribbing (1909) The insertion on the base of the first meta- demonstrates, to observe fundamental tarsal in all reptiles indicates that its ac- similarities. The musculature of this re- tion has always been to dorsiflex the foot, gion in Iguana (Fig. 19) is quite typical particularly the medial border. and represents about the maximum degree The extensor digitorum longus (com- of differentiation found in this class. In munis) has retained its primitive point of order to clarify a number of points dissec- origin on the lateral femoral condyle. In tions were made of the cruropedal and the Lacertilia and Sphenodon and also in the pedal musculature in Chelydra, Sphenodon, Crocodilia its insertion has been limited to Iguana, Varanus, and Alligator. The the base of the second and third metatar- papers of Osawa (1898), Sieglbauer (1909), sals, while in the Chelonia it inserts on all and Rabl (1916) were found helpful in this but the first metatarsal. In all recent connection. reptiles the tendonsof insertion are attached to the base of the metatarsals, while in birds Extensor Series and mammals they have migrated out to In the reptiles, the tibialis anterior (ex- the distal phalanges. In the latter groups tensor tibialis of amphibia) arises from the this muscle functions not only as an ex- proximal portion of the tibia, while in the tensor of the foot as a whole, but also as an Caudata, Anura, and presumably in the extensor of the digits. In other words, labyrinthodonts, this muscle arises from the extrinsic extensor of the digits has the lateral condyle of the femur. This taken over more complete control, supple- 11 t. ),

- .,,II peroneas g,9srocneluas(/u erarheaa/ - /ony9L4 6-p.pcsterifor9| // Y>/ frtrocnemis u4 peroneud hra/Aeaz( 1 ) (-p.cznttrA~~~r)a le', iS i tiOnterioria/is apo/zewyrsi.r

tarsomentatar-ya/s j ~i&/ irevs t s \fw \\ Ireiv dtensoresy ~gitorteY IrepBes I tovZdon.lrofI /Zexor dlitoru ,4re;,i. anda 11 /umhrica/e,y v A B

Fig. 19.-The cruropedal and the pedal musculature of Iguana. A, extensor musculature; B, C, D, successively deeper layers of the flexor musculature. 458 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 459

menting the action of the extensor digi- the lateral condyle of the femur and in- torum brevis. The migration to the distal serts into the ligamentum laciniatum and phalanges was probably favored by the also by a strong tendon into the hooked arboreal habitus of the earliest birds and portion of the fifth metatarsal. Hence, as mammals as a means of effectively and Tornier (1927) has pointed out for Varanus, quickly extending the toes after relaxing it runs along the ventral surface of the the hold on a branch. processus lateralis. Being posterior to the Although the reduction in the number axis of rotation of the mesotarsal joint, of tendons of insertion of the extensor the peroneus longus functions as a flexor. digitorum longus is evident in the lizards This change in function was probably and crocodiles, there is no evidence of such brought about by the development of the reduction in the birds and mammals in processus lateralis, separating the brevis which the tendons of insertion run out to and the longus near their insertion. the distal phalanges. The nature of the As Morton has stated (1935, p. 75), the insertion cannot be ascertained in any of changes in the course of the tendons of the the extinct orders with reasonable cer- longus and the brevis, as they appear in tainty. It would appear logical to assume mammals, must be associated with the that the tendons of insertion were at- superposition of the astragalus upon the tached to the proximal ends of the meta- calcaneum. Superposition had the im- tarsals in the cotylosaurs, the eosuchians, mediate effect of narrowing considerably the thecodonts, and possibly the pelyco- the posterior portion of the foot, and, as- saurs and at least the more primitive the- sociated with this narrowing, the tendons rapsids. It is probable that they migrated of both the peroneus longus and brevis out to the distal phalanges in the dino- slipped behind the lateral malleolus. saurs and pterosaurs. This shift placed the line of pull of these One of the most interesting problems in muscles behind the axis of rotation of the the evolution of the foot musculature is crurotarsal joint instead of in front, as in that of the changes in the peroneal mus- all amphibians and most reptiles. Under culature that occurred in the lizards, these conditions the muscles must neces- crocodiles, birds, and undoubtedly in the sarily act as flexors rather than extensors. most advanced therapsids. By an altera- Jones (1941) has pointed out, however, tion in its topographical relationships, one that the chief action of the peronei during or both divisions are changed from an ex- propulsion is the distribution of weight tensor into a flexor. As in the caudate am- among the metatarsals. phibians, the peroneus (extensor fibularis) Morton (ibid, p. 75) is of the opinion of the Chelonia is divided into several parts that the separation of the primary fibular and this is also the case in the lizards and extensor mass into the peroneus longus the alligator. In Sphenodon, on the other and brevis did not occur until after the hand, the muscle appears to be undivided. change in the course of the tendon of inser- Its origin on the fibular shaft and its in- tion. In every type examined, except sertion on the lateral surface of the fifth Sphenodon and Varanus, however, as metatarsal just anterior to the axis of the pointed out above, the peroneus is divided, mesotarsal joint causes it to extend the and was certainly divided at the stage rep- foot to a slight extent and to elevate the resented by the baurimorphs. Morton lateral border. states that the superposition of the as- In the lacertilians, as in the crocodiles, tragalus upon the calcaneum so altered the the peroneus has differentiated into brevis position of the latter relative to the fibula, (anterior) and longus (posterior) portions. that the displacement of the tendon oc- The brevis arises from the fibula and in- curred. Possibly the most important step serts on the dorsal and the lateral surfaces in the shift was the slipping of the tendons of the fifth metatarsal. It dorsiflexes the behind the lateral malleolus, and this foot to a slight extent and elevates the must have occurred at the time when the lateral border. The longus arises from superposition caused a certain amount of 460 Bulletin American Museum of Natural History [Vol. LXXVIII forward displacement in the articulation and inserts on the first and sometimes on of the tibia and the fibula relative to the the second digits. It functions almost en- tarsus. The peronei, on the other hand, tirely as an extensor of the first and second maintained their old insertion on the base digits. of the fifth metatarsal, with the result that The presence of the separate extensor the fibula was displaced relatively to a hallucis longus in birds and mammals, position in front of the tendons. arising from the tibial shaft and inserting The fibula of Bauria does not have a on the terminal phalanx of the first digit, well-developed, lateral malleolus, although is again a specialization associated with it is possible that the peronei could have the arboreal habitus. It must have split functioned as flexors with the longus and off from the extensor digitorum longus. possibly also the brevis passing along the The extensor digitorum brevis is usually plantar surface of the lateral process of the divided into a number of slips and two calcaneum. The lateral process on the layers. The divisions arise for the most Bauria calcaneum is certainly not a pero- part from the proximal tarsal elements neal tubercle nor is it necessarily homolo- and their tendons insert into the bases of gous with that tubercle on the mammalian the distal phalanges. In mammals all calcaneum, but is, more likely, an extension the parts tend to converge toward an of the entire lateral border, homologous origin on the calcaneum, although some with the processus lateralis. The peroneal parts still arise from the fibula (i.e., Caeno- tubercle is simply a projection on the lateral lestes). This condition must have arisen surface that must have developed along in the therapsids, and- very probably with or after superposition. It is of great Bauria had a completely mammalian ar- importance, however, in directing the ten- rangement of these slips. dons more ventrally and thus increasing It is highly probable that the dorsal mus- the flexing power. culature did not differ extensively in any In the alligator the peroneus brevis in- of the orders of reptiles in spite of the dif- serts for the most part on the fifth meta- ferent trends in tarsal evolution. The tarsal, but to some extent on the distal most important change was the shift in outer surface of the calcaneum. It func- the insertion of the peronei in the lacer- tions as an extensor (dorsiflexor). The tilians, crocodiles, birds, and presumably relatively much larger peroneus longus, in the advanced therapsids, causing these on the other hand, inserts on the outer sur- muscles to act as flexors rather than as ex- face of the tuber, posterior to the axis of tensors. It is not possible to determine the ankle-joint and is, hence, a flexor. Its whether such a shift occurred in the dino- migration into this position was accom- saurian orders or in the pterosaurs, al- plished simply by a posterior migration of though it is quite possible in both cases. the broad tendon of insertion. In the birds, the peroneus longus arises Flexor Series from the upper half of the dorsal surface At some early stage in reptilian evolu- of the tibia and inserts mostly into the tion the flexor primordialis communis tibial cartilage (a block of fibrocartilage split into two layers, the superficial be- situated on the ventral surface of the tibia coming the gastrocnemius, and, the deeper, between the condyles), and the tendon of the flexor digitorum longus (flexor pro- the flexor of the third toe. Thus, by vir- fundus). The mechanical causes favoring tue of its insertion posterior to the axis of such a division are obscure, although in a the ankle-joint, it functions as a flexor of thick muscle-mass, such as the flexor pri- the tarsometatarsus. mordialis communis, certain regions are The abductor and extensor hallucis, or undoubtedly more concerned with the the so-called supinator of the Amphibia, complete flexion of the digits, while other is well developed in the Chelonia, Lacer- regions are more concerned with flexing tilia, and Crocodilia. It still arises from the foot as a whole. This differential ac- the distal end of the fibula and the fibulare tion of groups of fibers within the same 1941]11 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 461

Inner border oF foot Outer boirder oF foot raised, otLter border raised, irnner border lowered --'stpination" lowered --"pronation" Midsaqittal Xxis of Foot

Fig. 20.-A diagrammatic representation of the action of the extrinsic pedal musculature in a lacertilian (Iguana) and an alligator. Note the change in the action of the peroneus longus from the primitive condition found in the chelonians and Sphenodon. muscle due to slight differences in origin, the lateral condyle of the femur. It in- insertion, or in direction, could very well serts for the most part into the plantar result in such a subdivision. fascia, although it has a supplementary The gastrocnemius was apparently split insertion on the fifth metatarsal. In the into tibial and fibular portions soon after Crocodilia, the tendon of the relatively it separated from the primitive flexor mass, much larger fibular portion passes directly or it may have separated off into two parts. over the tuber and is partially attached to The origin of the tibial portion is subject it, although the actual insertion is into the to some variation, arising either from the plantar aponeurosis. The tibial head femur or tibia or both. It inserts, for the joins the fibular at the tuber, as does also most part, into the plantar aponeurosis, the long tendon of the flexor tibialis. The but in many cases with a small slip to the latter, through this tendon, amplifies the base of the fifth metatarsal, as in Iguana action of the gastrocnemius and peroneus and Sphenodon. In the Recent Reptilia, longus to a slight degree. the fibers of the tibial head run diagonally The usual function of the gastrocnemius across the crus with the lateral border of is stated to be plantarflexion of the foot in the muscle inserting on the fifth meta- association with propulsion. Another im- tarsal. The femoral head of the gastro- portant phase of its action, particularly in cnemius springs from the lateral surface of the more primitive reptiles, is to prevent 462 Bulletin American Museum of Natural History [Vol. LXXVIII

the foot from sliding forward as the body- It would seem, in the case of the cotylo- weight is shifted to one or the other hind- saurs and in the terrestrial forms grouped limb at the beginning of propulsion. This under "Type I," that the foot could have is essentially the same function performed exerted relatively little forward thrust dur- by the flexor primordialis communis in the ing propulsion. There is no evidence in- caudate amphibian foot. dicating that the gastrocnemius was rela- It was stated in the section on amphibian tively larger in these groups than in the locomotion (p. 419) that the primitive tet- labyrinthodonts, although the body-weight rapod foot takes a very small part in the was certainly as great in many cases. propulsive effort. This was probably the Furthermore, the gastrocnemius was in- case when the very supple nature of serted into the plantar aponeurosis, which the foot is taken into account, as well as in turn was practically parallel to the plane the body-weight compared with the size of the foot, resulting in a very small angle of the flexor primordialis communis, and of application. A similar situation must the angle of application of this muscle have existed in the eosuchians and pelyco- (discussed below). In the Anura, on the saurs (Fig. 21,B), and in the therapsids other hand, the work of Hirsh (1931) dem- lacking a tuber calcanei. Romer (1940, onstrates that the elongated tarsal ele- p. 165), has stated that a portion of the ments, motivated by the homologue of the gastrocnemius was inserted on the thinned flexor primordialis, usually called the proximolateral border of the pelycosaur plantaris longus, plays a very active role calcaneum. This being the case, the angle in propulsion. The plantaris longus, how- of application would indeed be very small. ever, is relatively very large and powerful In the lizards and Sphenodon there is compared to the size and weight of the evidence of the active participation of animal and the size and length of the seg- the foot in propulsion to at least a limited ments moved by it. degree. The axis of the fibular head of It would appear that the ability of the the gastrocnemius, when continued dis- flexors to enable the foot to participate tally, passes through the hooked-portion actively in the propulsive effort is depend- (hamate process) of the fifth metatarsal. ent on the size and weight of the animal, This portion of the gastrocnemius, further- on the relative size of the flexors concerned, more, has a very strong insertion on the on the degree of consolidation within the elevated and thickened distomedial border tarsus, and, finally, but of great impor- of the hook, as pointed out before (Fig. 21, tance, on the angle of application (the C). This arrangement has the effect of angle between the axis of the muscle and increasing the angle of application, and the axis of the part moved). Steindler the ventrally extended portion of the hook (1935) has pointed out that if the angle of may be looked upon as the functional ana- application is under 30 degrees the force logue of the tuber calcanei. In Varanus exerted by the muscle has a greater stabiliz- komodoensis, this portion of the hook ing than rotary component. He figures the amounts to a ventrally directed tuber. condition in the human knee, a type of The position which the lizard's foot is joint basically similar to the mesotarsal forced to assume during the latter half of type of ankle-joint. In the former the propulsion, however, is not favorable for angle of application is increased by the its active participation as the leverage ac- presence of the patella, forcing the applying tion of the foot cannot be employed to full tendon to make an angle in its course advantage. instead of running in a straight line from In the more advanced thecodonts and, the femur to the cnemial crest. With all as mentioned earlier, in one primitive these variables that must be considered, family of theropod dinosaurs, the Hallo- it is very difficult to determine, with a podidae, the calcaneum possessed a tuber, reasonable degree of accuracy, the im- although the calcaneum was functionally portance of the foot in the propulsive united with the crus. This tuber did, effort in many of the fossil reptiles. however, influence the orientation of the 19411] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 463

D E

Fig. 21.-Medial views of the lower leg and foot of A, Trematops (from model); B, Varanosaurus (after Romer and specimen); C, Iguana (from specimen); D, Saltoposuchus (after von Huene); E, Alligator (froin specimen); F, Plateosaurus (from specimen); G, Gallus (from specimen); H, Bauria (from model); I, Taxidea (from specimen). The heavy black lines indicate the course of the gastrocnemius (the flexor primordialis communis in the case of Trematops). 4644Bulletin American Museum of Natural History [Vol. LXXVIII tendons so that they inserted at a slight nemius subdivided into three parts, the angle to the plane of the foot, thus pre- largest retaining the same name, the other sumably enabling the gastrocnemius to two becoming the soleus and the plantaris. exert a more effective pull on the foot by The soleus undoubtedly represents a por- increasing the angle of application. In tion of the gastrocnemius externus that view of its functional importance, it is migrated on to the tibia and the fibula and rather puzzling that this tuber should was separated from the gastrocnemius have been lost in all the other descendants through differential action. The plan- of the thecodonts, except the crocodiles. taris, on the other hand, while arising from With the exception of the Hallopodidae, the lateral condyle, maintained its ancient neither the saurichian nor the ornithischian insertion into the plantar fascia. The dinosaurs have any indication of a struc- latter gives every indication of being the ture that would increase the angle of ap- only part of the gastrocnemius that did plication. In view of this fact, the ex- not develop a new insertion on the tuber. trinsic flexors had to be of relatively great The other derivative of the flexor pri- size to permit the active cooperation of the mordialis communis, the flexor digitorum foot. The same reasoning would seem to longus, has in all reptiles several heads of apply to those therapsids in which the origin: from the lateral condyle of the fe- tuber calcanei is lacking. mur, from the upper third of the fibula, With the appearance of the tuber in the and one or two transverse tendinous bands advanced therapsids, and with its reten- from the distal end of the fibula or the tion in the crocodiles, in both cases as- fibulare and the fifth metatarsal. These sociated with a calcaneum, which, in turn, divisions unite in the lower third of the is functionally united with the foot, the crus to form a broad, flat tendinous sheet mechanical setting is much clearer. The covered by the plantar aponeurosis. The foot is, under these conditions, definitely former then splits into five, or sometimes a lever of the first class, as was pointed out less, tendons that extend to the distal earlier, and the muscular force, being ap- phalanges, where they become attached. plied at almost a right angle, goes almost In the primitive tetrapod foot, and also entirely into the rotary component. In in the types with a metatarsal joint, the other words, the foot definitely partici- flexor digitorum longus has a relatively pates in the propulsive effort and, assum- broad plantar tendon. It becomes nar- ing other things to be equal, the longer the rowed somewhat in the Sauria as it passes tuber, the less the amount of power re- through a shallow groove on the plantar quired of the gastrocnemius, but the slower surface of the astragalocalcaneum. In the action. Alligator, the presence of the tuber has In the birds, the angle of application of forced the tendon toward the medial side the gastrocnemius is increased by the pres- as it has in the mammals, and undoubtedly ence of the hypotarsus and by the tibial did in the therapsids having this structure. cartilage. Both structures are grooved to In birds, the digital flexor musculature has guide the tendons of the gastrocnemius reached an extreme stage of differentiation, over the ankle-joint and, therefore, they each digit being flexed by separate mus- bring about the active participation of the cles, a condition obviously associated with foot in propulsion. The gastrocnemius prehension. has an attachment on the hypotarsus and The pronator profundus (sometimes finally inserts on the sheath of the digital called the tibialis posterior in reptiles) is a flexor tendons. A soleus arises from the very powerful muscle. In all cases it tibia and inserts on the tibial cartilage. arises from the entire length of the fibular With the development of the tuber in shaft and inserts by means of a strong ten- the therapsids, the gastrocnemius lost its don into the base of the first metatarsal, insertion into the plantar aponeurosis and and in some cases, also, the second meta- inserted directly on the tuber. Also at tarsal. The pronator profundus func- this stage, the fibular head of the gastroc- tions to some extent as a flexor, but it is 19411 Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 465

also an adductor of the hallux. That such flexor digitorum brevis and the lumbricales. a powerful muscle should function mainly The former has its origin for the most as an adductor seems unlikely. A more part from the plantar aponeurosis, although important function would appear to be, a deeper layer may arise from the apo- as in the primitive tetrapod, that of slightly neurosis of the flexor digitorum longus, and pronating (in the lacertilians) and adduct- inserts into the proximal phalanges. It ing the foot at the end of the recovery is not only concerned with flexing the digits, phase. During propulsion its contraction but also with amplifying to a slight ex- can only evert the knee. tent the action of the gastrocnemius. There has been considerable disagree- With the development of the tuber in the ment regarding the origin of the mam- therapsids this continuity was broken, malian flexor fibularis. Without discuss- the gastrocnemius inserting on the tuber ing the matter in detail here, the conclu- and the plantar aponeurosis, and the flexor sion of Howell (1939) appears most rea- digitorum brevis arising from the plantar sonable, namely, that it has been separated surface of the tuber. from the flexor digitorum longus (flexor The quadratus plantae of mammals is tibialis) and has no relationship with the considered by Howell to be a remnant of pronator profundus, which is probably the extrinsic musculature interposed be- the homologue of the mammalian tibialis tween the superficial and deep layers. posterior. This observation would appear tenable in The deepest layer of the extrinsic series view of the fact that the quadratus plantae of muscles is divided into proximal and has no homologue in the amphibian in- distal parts. The proximal portion, the trinsic musculature. It may represent a interosseus cruris or fibulotibialis supe- distal muscular portion of the flexor digi- rior, arises from the medial side of the fibula torum longus, which was left behind on the and the interosseus membrane and in- sole during the alteration in the orienta- serts on the proximal half of the tibia at a tion of the distal portion of this muscle more distal level. The distal portion, the necessitated by the development of the fibulotibialis inferior, arises from the distal tuber. half of the fibula and inserts on the lower Among recent reptiles there is no tend- third of the tibia. This layer is concerned ency for the superficial layer to develop with the maintenance of the normal rela- independent abductors of the marginal tions of the tibia and fibula during pro- digits; they are very probably associated pulsion. with the arboreal ancestry of the mammals, where the independent action of the The Plantar Musculature first and fifth digits was favored. The plantar musculature was presum- The lumbricales arise from the dorsal ably but little altered from the primitive surface of the plantar aponeurosis of the tetrapod plan until the tuber calcanei ap- flexor digitorum longus and insert into peared in the advanced therapsids, or the the tendons of the flexor digitorum brevis. highly specialized tarsometatarsus de- They are likewise flexors of the digits, am- veloped in the birds, favoring its reduction. plifying the action of the flexor digitorum It is composed of two basic layers (Howell, longus to some extent. 1939), each being subdivided into several The deep layer is made up of the con- minor layers. It is not required here to trahentes digitorum and the various divi- discuss this musculature in detail, or to sions of the interossei. The contrahentes trace its homologies. Briefly considered, digitorum has maintained its identity from it is concerned with the flexion of the digits, the amphibian stage and is present in all in some cases amplifying the action of cer- recent reptiles. In the caudates it arises tain of the extrinsic flexors, and, in some mostly from a tendinous band attached to mammals, with the maintenance of the the second to the fifth tarsalia, while in longitudinal arch. the reptiles it arises from the fifth meta- The superficial layer consists of the tarsal. In the reptiles it has, therefore, 466 Bulletin American Museum of Natural History [Vol. LXXVIII shifted and concentrated its origin on the The plantar musculature in the extinct lateral side of the foot. The contrahentes orders must have been very similar to are mainly adductors of the digits, partic- that just described, except in the aquatic ularly of digits one, two, and three. If a forms and in those types in which it was slip is present for the fourth digit, it can reduced. In such types as the ichthyo- only act as a flexor. The adductor of the saurs and pleisosaurs there must have been fifth digit, and possibly also the abductor, a reduction and secondary simplification are probably derived from this layer. of the extrinsic and intrinsic musculature, The interossei (flexor breves profundi), as the more distal portion of the paddle having their origin on one digit and insert- could only be flexed to a very limited de- ing on the adjacent digit at a more distal gree. In the less specialized aquatic level, act as adductors and flexors. They forms, such as some of the protorosaurs, it are usually divided into several layers. was probably not reduced.

SUMMARY 1.-Both the number and the disposi- caudate tarsus is discussed and its marked tion of the elements in the tarsus of the resemblance to the primitive tetrapod rhachitome Trematops indicate that it is tarsus is stressed by description and by a the most primitive completely preserved pictorial phylogeny. The sporadic oc- tetrapod tarsus known. This tarsal pattern currence of ossification in the amphibian is dominated by a convergence of the ele- tarsus, excepting the anuran, is considered ments, excepting those along the preaxial to be genetic and not related to the habitus border, toward the fibula, a condition in- of the animal. herited from the rhipidistian pelvic fin 5.-The variation in the centrale region and persisting into the caudate amphibian of the tarsus is discussed. It would ap- tarsus. Although movement could occur pear that the number of centralia present between most of the tarsal elements in is of no functional importance and hence varying degrees, the important plane of that there has been no selection toward a flexure during locomotion was between the constant number. tarsalia and the metatarsals, as is the case 6.-Previous work on the embryology in the caudates. and phylogeny of the anuran tarsus is 2.-The known tarsi of other rhachi- critically reviewed. It is concluded, on tomes, the embolomeres, and stereo- embryological evidence, that the two spondyls are incompletely and variably os- elongated proximal tarsal elements are sified. The available evidence indicates a the fibulare and the tibiale plus several general agreement with the Trematops centralia and possibly the intermedium. pattern, although there may be a reduc- There are no paleontological data to sup- tion in the number of elements present. port or refute this view. The relation- 3.-The tarsus of the Lepospondyli, an ship between specialization in the tarsus order now considered to be ancestral to and the lack of variation is discussed. the Caudata, is known in only two forms. 7.-The basic resemblance between the That of the nectridian Scincosaurus can- caudate and the labyrinthodont tarsus is not be ancestral to the reptilian tarsus, as evidence against the theory that the cau- Broom believes, nor is there any evidence dates were independently derived from to support his view that the tibiale is the Dipnoi, and the Anura from the Laby- homologous with the mammalian navicular. rinthodontia. The very early embryonic The tarsus of the Microsauria, considered stages of the tarsus are very similar in the to be directly ancestral to the caudates, is Caudata and the Anura, although the known only in the case of the poorly pre- specializations of the anuran tarsus are served although completely ossified tar- evident as soon as the anlagen of the tarsal sus of Hyloplesion. elements are formed. 4. -The pertinent literature on the 8.-All the evidence indicates that the 1941 ] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 467 labyrinthodonts, like the caudates, had tarsus of Seymouria is essentially am- their hind-feet forwardly directed during phibian, while that of the diadectomorphs locomotion, although the femoral head is is specialized, culminating in the pareia- terminal and there are no diarthrodial saurian condition in which the astragalus tarsal joints. The necessary adjustments and calcaneum are coossified, and the func- allowing this were made in the crus. The tional ankle-joint is mesotarsal. tibia assumed a diagonal position, lessening 14.-The reptilian tarsus originating the functional length of the inner side of from a Labidosaurus-like type evolved the crus, and thus permitting the foot to roughly along three different lines: be forwardly directed throughout propulsion. TYPE I.-Retention of the primitive pattern or a modification of it for an aquatic habitus, to 9.-By means of motion pictures of the a greater or lesser degree, in the Mesosauria, locomotion of Triturus and Ambystoma Ichthyosauria, Sauropterygia, Protorosauria. it is possible to describe in detail the move- TYPE II.-Modifications leading to the meso- ment of the hind-limb during propulsion tarsal joint as the functional ankle-joint, in the Chelonia and the orders descended from the and recovery. With this information, Eosuchia. The joint probably developed inde- deductions are made concerning the move- pendently in the archosaurs and lepidosaur- ments of the hind-limb in the labyrintho- ians, as it was not developed in the eosuchians donts. and possibly not in the most primitive thecodonts. TYPE III.-Modifications leading to the cruro- 10.-The pressure-distribution of the tarsal joint as the functional ankle-joint in the amphibian foot, as deduced from fossil pelycosaurs, therapsids, and mammals. footprints, is discussed, together with the forces acting on the hind-limb during loco- 15.-In the first type the functional motion and their relation to the structure ankle-jolint remained tarsometatarsal un- of the crus and tarsus. The foot probably less the foot degenerated into a paddle with did not enter actively into the propulsive most of the flexion in the digits, as in the effort to any great extent in the primitive ichthyosaurs and plesiosaurs. tetrapods, but functioned mainly as an ex- 16.-In the second type the mesotarsal panded base through which the tractive joint was elaborated in various ways. In forces centered around the femur were ap- all the orders in this category, excepting plied to the ground through the crus. the Crocodilia, the astragalus and cal- 11.-The extrinsic and intrinsic mus- caneum are functionally part of the crus. culature of the amphibian hind-foot is In the latter there is movement between described with particular emphasis on the the astragalus and the calcaneum and be- functional grouping of the muscles in pro- tween the latter and the fibula. In every ducing the characteristic movements of case there is a marked reduction in the locomotion. There is no true pronation number of distal tarsal elements, a modifi- and supination in the amphibian foot, cation that occurred along with formation these movements occurring only in a limited of the mesotarsal joint. degree in the mammalian foot. 17.-The locomotor movements of the 12.-The origin of the reptilian tarsus hind-limb are determined as much by the is critically examined. Both ontogenetic morphology of the femur as by the struc- and phylogenetic evidence is presented. ture of the tarsus. Regardless of the The position is here taken that, although plane in which the femur moves through- the calcaneum is only the fibulare, the out propulsion, the foot is directed for- astragalus is compound in origin, being ward, unless, in the case of the Sphenodon made up of the intermedium, a centrale, and the lizards, the twisting of the femoral and possibly the tibiale. shaft and the structure of the tarsus forces 13.-The evolution of the tarsus within it to rotate laterally as retraction pro- the Cotylosauria resulted in a number of gresses. distinct types. That of Labidosaurus is 18.-The crurotarsal joint, characteristic considered to be representative of the of the third type, first appeared in the primitive, definitive reptilian type. The pelycosaurs. Although the femoral head 468 Bulletin American Museum of Natural History [Vol. LXXVIII

is terminal and the knee-joint permanently plane approaching ninety degrees. The crooked, the rounded tibial facet on the well-rounded tibial and fibular facets elimi- astragalus, plus the fact that the tibial nated the torsion in the tarsus and formed articulation is on a more distal level than a highly mobile hinge joint. the fibular, permits the foot to be for- 22.-The tuber calcanei developed inde- wardly directed. Although the fibular pendently in the Thecodontia and the facet is restricted to the proximal border, Therapsida. Of the orders derived from the tibial has migrated somewhat on to the thecodonts, the tuber was retained the dorsal surface. only in a primitive family of theropod 19.-The development of the meso- dirtosaurs, the Hallopodidae, and in the tarsal and the crurotarsal joints may be Crocodilia. looked upon as two different ways of 23.-At some stage between the the- treating the stress in the proximomedial rapsids and the mammals the astragalus region of the primitive tarsus. In the case became superimposed upon the calcaneum. of the former the joint is constructed to This alteration relieved the fibula of its resist the stress, in the latter to eliminate articulation with the calcaneum and the it. tibia became the weight bearing axis of the 20.-The therapsid tarsus evolved along crus. The functional implications of the several distinct lines. In the dromasaur- superposition are discussed, the most im- ians, dinocephalians, and dicynodonts, the portant being the creation of the subas- calcaneum remained dorsoventrally com- tragalar joint, permitting so-called supina- pressed and the astragalus tended to lose tion and pronation for the first time, and its rounded facet. In the gorgonopsians, the formation of the transverse tarsal arch. therocephalians, cynodonts, and bauria- 24.-The extrinsic and intrinsic mus- morphs the rounded tibial facet on the culature of the reptilian foot is described astragalus was retained, and the tibial and with emphasis on function. The role of fibular facets on the calcaneum tended to the foot in propulsion is evaluated, partic- migrate completely on to the dorsal sur- ularly in relation to the angle of applica- face. This migration freed the proximal tion of the gastrocnemius. In this con- border of the calcaneum, permitting a nection, the ventrally directed portion of tuber to develop. the hook on the fifth metatarsal of lizards 21.-The therapsid femur early de- is considered to be the functional analogue veloped a medially directed head without of the tuber calcanei, in that it increases any accompanying twisting in the shaft, the angle of application of the fibular head and the femur swung more and more in a of the gastrocnemius.

REFERENCES ABEL, 0. 1888. "Beitriage zur Morphogenie des Carpus 1919. "Die Stiimme der Wirbeltiere." Ber- und Tarsus der Vertebraten. I. Ba- lin and Leipzig. 914 pp. trachia." Gustav Fischer. Jena. 86 ABEL, WOLFGANG PP. 1930. "Beitrage zur Kenntnis der Anpas- DE BEER, G. R. sungserscheinungen der distalen Hin- 1940. "Embryos and Ancestors." Oxford terfussmuskulatur der Saugetiere bei University Press. London. 108 pp. einem Wechsel der Lebensweise." BOONSTRA, L. D. Morph. Jahr., LXIV, pp. 558-635. 1929. "On the parieasaurian pes." Ann. So. BAUR, G. Afr. Mus., XXVIII, Pt. 1, pp. 113-122. 1883. "Der Tarsus der Vogel und Dinosaur- 1934. "A contribution to the morphology of ier." Morph. Jahr., VIII, pp. 417- the Gorgonopsia." Ann. So. Afr. 456. Mus., XXXI, Pt. 2, pp. 137-174. 1885. "Zur Morphologie des Carpus und 1938. "On the South African mammal-like Tarsus der Reptilien." Zool. Anz., reptile, Bauria Cynops." Paleobio- VIII, pp. 631-638. logica, VI, pp. 164-183. 1886. "Die alteste Tarsus (Archegosaurus)." BOYDEN, A., AND NOBLE, G. K. Zool. Anz., IX, pp. 104--106. 1933 "The relationships of some common 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 469

Amphibia as determined by serological of Bauria cynops Broom." Amer. Mus. study." Amer. Mus. Novitates, No. Novitates, No. 946, pp. 1-6. 606, pp. 1-24. 1938. "On a nearly complete therocephalian BROILI, F. skeleton." Ann. Transvaal Mus. XIX, 1925. "Beobachtungen an der Gattung Homo- Pt. 2, pp. 257-261. eosaurus H. v. Meyer." Sitzungs- BROWN, B. berichte Bayerischen Akademie der 1933. "An ancestral crocodile." Amer. Mus. Wissenschaften: Math.-Natur. Abt. Novit., No. 638, pp. 1-4. VII, pp. 81-121. 1934. "A change of names." Science, BROILI, F., AND SCHR6DER, J. (N. S.) LXXIX, p. 80. 1935. "Beobachtungen an Wirbeltieren der CAMP, C. L. Karrooformation. XIII. Uber die 1930. "A study of the phytosaurs, with de- Skelettreste eines Gor gonopsiers aus scriptions of new material from West- den unteren Beaufort-Schichten." er n North Amer ica." Mem. Univ. Sitzungsb. Bayer. Akad. Wissen., Calif., X, pp. 1-174. Abt., pp. 279-330. CARMAN, E. J. 1937. "Beobachtungen an Wirbeltieren der 1927. "Fossil footprints fIom the Pennsyl- Karroofor mation. XXV. Uber vaniia System in Ohio." Bull. Geol. Micropholis Huxley." Sitz. Bayer. Soc. Amer., XXXVIII, pp. 385-396. Akad. Wissen., pp. 19--38. CASE, E. C. BROOM, ROBERT 1899. "A redescription of Pariotichuis incisi- 1903. "On the iemains of Procolophon in the vyes Cope." Zool. Bull., II, pp. 231- Albany Museum." Rec. Albany Mus., 245. I, No. 1, pp. 8-24. 1910. "New or little known reptiles and am- 1906a. "On the South African diaptosaurian phibians from the Permian (?) of reptile Howesia." Proc. Zool. Soc. Texas." Bull. Amer. Mus. Nat. Hist., London, XL, pp. 591-600. XXVIII, pp. 163-196. 1906b. "On the arrangement of the epiphyses 1911. "A revision of the Cotylosaur ia of of the mammalian metacatrpals and North America." Carn. Inst. Wash., metatarsals." Anat. Anz., XXVIII, No. 145, pp. 1-122. pp. 106-108. CASE, E. C., WILLISTON, S. W., AND MEHL, 1913a. "On the South African pseudosuchian M. G. Eoiparkeria and allied genera." Proc. 1913. "Permo-Carboniferous vertebrates Zool. Soc. London, Pt. 3, pp. 619-633. from New Mexico." Carn. Inst. 19131). "On the origin of the mammalian digi- Wash., No. 181, pp. 1-81. tal foImula." Anat. Anz., XLIII, pp. CLARK, W. E. LE GROS 230-232. 1939. "The tissues of the bodly." Oxfor d 1914. "On the origin of mammals." Croon- Univ. Press. 372 pp. ian Lecture. Philos. Trans. Roy. Soc. CREDNER, HERMANN London, (B) CCVI, pp. 1-48. 1881-93. "Die Stegocephalen uind Saurier 1921. "On the structure of the reptilian tar- aus dem Rothliegenden des Plauen'- sus." Proc. Zool. Soc. London, pp. schen Gr unides bei Dr esden." Zeit. 143-155. der Deutschen Geol. Gesell., XXXIII, 1924. "'Further evidence on the strtucture of 1881-XLV, 1893, and Naturwis- the Eosuchia." Bull. Amer. Mus. Nat. sensch. Wochenschrift, V, 1890. Hist., LI, Art. 3. pp. 67--76. DoILo, L. 1926. "On a nearly complete skeleton of a 1929. "Carpus und tarsus." Roux's Archiv, new Eosuchian r-eptile Palaeagama CXX, pp. 272-298. vie!haoeri." Proc. Zool. Soc. London, EATON, G. F. Pt. 2, p1). 487-491. 1910. "Osteology of Pteranodon." Mem. 1929. "On the carnivorous mammal-like Conn. Acad. AIrts and Sciences, II, reptiles of the family Titanosuchi- pp. 1-38. dae." Ann. Transvaal Mus., XHII, EISLER, P. Pt. 1, pp. 9-36. 1893-95. "Die Homologie der Extremi- 1930a. "On the structure of the mammal-like thten." Abh. d. naturf. Gesell. zu reptiles of the suiborder Gorgonopsia." Halle, XIX, pp. 89-344. Philos. Trans. Roy. Soc. London, (B) EFREMOV, J. A. CCXVIII, pp. 345-371. 1940. "Die Mesen-Fauna der per misehen 1930b. "The origin of the human skeleton." Reptilien." Neues Jahrb., LXXXIV, Witherby, London, 164 pp. Aht. B, pp. 379-466. 1932a. "The mammal-like reptiles of South ELFTMAN, H. Africa." Witherby, London. 376 pp. 1934. "A cinematic study of the distriblution 1932b. "On some South African pseudo- of pressure in the human foot." Anat. suchians." Ann. Natal. Mus., VII, Rec., LIX, pp. 481-487. Pt. 1, pp. 55-59. ELFTMAN, H., AND MANTER, J. 1937. "On the palate, occiput and hind-foot 1935a. "The evolution of the human foot, with 470 Bulletin American Museum of Natural History [Vol. LXXVIII

especial reference to the joints." Nat. Hist., XXXVIII, Art. 15, pp. Jour. Anat., LXX, pp. 56-67. 447-563. 1935b. "Chimpanzee and human feet in bi- GREGORY, W. K., MINER, R. W., AND NOBLE, pedal walking." Amer. Jour. Phys. G. K. Anthrop., XX, pp. 69-79. 1923. "The carpus of Eryops and the struc- EMERY, C. ture of the primitive chiropterygium." 1898. "Die Fossilen Reste von Archego- Bull. Amer. Mus. Nat. Hist., XLVIII, saurus und Eryops und ihre Bedeutung Art. 10, pp. 279-288. fur Morphologie des Gliedmassen- GREGORY, W. K., AND RAVEN, H. C. skelets." Anat. Anz., XIV, pp. 201- 1941. "Studies on the Origin and Early Evo- 208. lution of Paired Fins and Limbs." FRAAS, E. Annals N. Y. Acad. Sci., in press. 1889. "Die Labyrinthodonten der schwiib- HARTMANN-WEINBERG, A. ischen Trias." Paloeontpgr., XXXVI, 1929. "Uber Carpus und Tarsus der Pareia- pp. 1-158. sauriden." Anat. Anz., LXVII, pp. FRANCIS, E. T. B. 401-428. 1934. "The anatomy of the salamander." HAUGHTON, S. H. Oxford Univ. Press. 381 pp. 1924. "Reptilian remains from the Karroo FRITSCH, A. Beds of East Africa." Quart. Jour. 1889. "Fauna der Gaskohle und der Kalk- Geol. Soc., LXXX, Pt. 1, pp. 1-11. steine der Permformation B6hmens." HAUGHTON, S. H., AND BOONSTRA, L. D. II, Prag. 114 pp. 1930. "Pareiasaurian studies. Part VI. GEGENBAUER, C. The osteology and myology of the loco- 1864. "Untersuchungen zur vergleichenden motor apparatus. A. Hind-limb." Anatomie der Wirbelthiere. Erstes Ann. So. Afr. Mus., XXVIII, Pt. 3, Heft. Carpus und Tarsus." Leipzig. pp. 297-366. Pp. 1-127. HIRSH, W. 1876. "Zur Morphologie der Gliedmassen der 1931. "Zur physiologischen Mechanik des Wirbeltiere." Morph. Jahrb., II, pp. Froschsprunges." Zeit. vergl. Phys- 396-420. iol., XV, pp. 1-49. GILMORE, C. W. N. 1926. "Fossil footprints from the Grand HOLMGREN, Canyon." Smiths. Misc. Coll., 1933. "On the origin of the tetrapod limb." LXXVII, No. 9, pp. 1-41. Acta Zool., XIV, pp. 185-295. 1927. "Fossil footprints from the Grand 1939. "Contribution to the question of the Canyon: Second contribution." origin of the tetrapod limb." Acta Smiths. Misc. Coll., LXXX, No. 3, pp. Zool., XX, pp. 89-124. 1-78. HOWELL, A. B. 1928. "Fossil footprints from the Grand 1935a. "Morphogenesis of the shoulder archi- Canyon: Third contribution." tecture. Part III. Amphibia." Smiths. Misc. Coll., LXXX, No. 8, pp. Quart. Rev. Biol., X, pp. 397-431. 1-16. 1935b. "Primitive carpus." Jour. Morph., GOODRICH, E. S. LVII, pp. 105-112. 1916. "On the classification of the Rep- 1936. "The phylogenetic arrangement of the tilia." Proc. Zool. Soc. London, (B) muscular system." Anat. Rec., LXVI, LXXXIX, pp. 261-276. No. 3, pp. 295-316. GRANT, J. C. B. 1938. "Morphogenesis of the architecture of 1940. "A method of anatomy." 2nd ed. hip and thigh." Jour. Morph., LXII, Williams and Wilkins Co. Baltimore. pp. 177-218. 794 pp. 1939. "Gross anatomy." D. Appleton- GRAY, J. Century Co. New York. 403 pp. 1939. "Aspects of animal locomotion." HOWEs, G. B., AND SWINNERTON, H. H. Proc. Roy. Soc., (B) CXXVIII, pp.28- 1901. "On the development of the skeleton 62. of the tuatara, Sphenodon punctatus." GREGORY, W. K. Trans. Zool. Soc. London, XVI, pp. 1912. "Notes on the principles of quadru- 1-86. pedal locomotion." Ann. N. Y. Acad. VON HUENE, F. Sci., XXII, pp. 267-294. 1912. "Die Cotylosaurier der Trias." Pale- 1926. "The skeleton of Moschops capensis ontogr., LIX, pp. 69-102. Broom, a dinocephalian reptile from 1913. "Beobachtungen ilber die Bewegung- the Permian of South Africa." Bull. sart der Extremitiaten bei Krokodilen." Amer. Mus. Nat. Hist., LVI, Art. 3, Biol. Centralbl., XXXIII, pp. 468- pp. 179-251. 472. GREGORY, W. K., AND CAMP, C. L. 1920a. "Ein Telerpeton mit gut erhaltenem 1918. "Studies in comparative myology and Schiidel." Centralbl. f. Min., etc., No. osteology. III." Bull. Amer. Mus. 11, u. 12, pp. 189-192. 1941] Schaeffer, Evolution of the Tarsus in Amphibians and Reptiles 471

1920b. "Osteologie von Aetosaurus ferratus 0. punctata." Arch. f. Mik. Anat., LI, Fraas." Acta Zool., I, pp. 465-491. pp. 481-690. 1921. "Neue Pseudosuchier und Coeluro- 1902. "Beitrage zur Anatomie des Japanis- saurier aus dem Wurttembergischen chen Riesensalamanders." Mitteil Keuper." Acta Zool., II, pp. 329-403. Med. Facultat Kais-Japan Univ., 1938. "Stenaulorhynchus, ein Rhynchosau- Tokio, V, pp. 221-427. ride der ostafrikanischen Obertrias." PEARSON, H. S. Nova Acta Leopoldina, (N. F.) VI, 1924. "A dicynodont reptile reconstructed." No. 36, pp. 83-121. Proc. Zool. Soc. London, pp. 827-855. 1939. "Ein neuer Procolophonidae aus dem PEYER, B. deutschen Buntsandstein." Neuen 1937. "Die Triasfauna der Tessiner Kalkal- Jahrbuch fur Min., etc., LXXXI, pp. pen: XII. Macrocnemus bassanii 501-511. Nopcsa." Abh. Schweiz. Palaeont. JAEKEL, 0. Gesell., LIX, pp. 1-140. 1909. "Ueber die ialtesten Gliedmassen von PIVETEAU, J. Tetrapoden." Sitz. d. Gesell. Natur. 1937. "Paleontologie de Madagascar. Freunde, X, pp. 587-615. XXIII.-Un Amphibien du Trias In- JONES, R. L. f6rieur. Essai sur l'origine et l'6volu- 1941. "The human foot: An experimental tion des Amphibiens anoures." Ann. study of its mechanics and the role of d. Paleont., XXVI, pp. 135-178. its muscles and ligaments in the sup- QUENSTEDT, F. A. port of the arch." Amer. Jour. Anat., 1861. "Bemerkungen zum Archegaosaurus." LXVIII, No. 1, pp. 1-39. Neues Jahrb., pp. 294-300. KERR, J. G. RABL, C. 1907. "The development of Polypterus sene- 1910. "Bausteine zu einer Theorie der Ex- galus Cuv." Budgett Mem. Vol., tremitilten der Wirbeltiere." Leip- Cambridge University Press, pp. 195- zig. 290 pp. 284. 1916 "Uber die Muskeln und Nerven der 1932. "Archaic fishes-Lepidosiren, Protop- Extremitaten von Iguana tuberculata terus, Polypterus-and their bearing Gray." Anat. Hefte, LIII, pp. 683- upon problems of vertebrate mor- 789. phology." Jena. Zeit. Naturw. RIBBING, L. LXVII, pp. 419-433. 1909. "Die Unterschenkel-und Fussmuskula- KINDAHL, M. tur der Tetrapoden." Act. Univ. 1938. "Zur Entwicklung der Exkretion- Lund., (N. S.) V, pp. 1-158. sorgane von Dipnoern und Amphi- ROMER, A. S. bien." Acta Zool., XIX, pp. 1-190. 1922. "The locomotor apparatus of certain LORTET, L. primitive and mammal-like reptiles." 1892. "Les reptiles fossiles du Bassin du Bull. Amer. Mus. Nat. Hist., XLVI, Rhone." Arch. Mus. d'Hist. Nat. de Art. 10, pp. 517-606. Lyon, V. 139 pp. 1925. "Permian amphibian and reptilian re- MATTHEW, W. D. mains described as Stephanospondy- 1937. "Paleocene faunas of the San Juan lus." Jour. Geol., XXXIII, pp. 447- Basin, New Mexico." Trans. Amer. 463. Philos. Soc., (N. S.) XXX. 510 pp. 1933. "Vertebrate Paleontology." Univ. MEYER, H. v. Chicago Press. 491 pp. 1939. "Notes on Branchiosaurs." Amer. 1858. "Reptilien aus der Steinkohlen-For- Jour. Sci., CCXXXVII, pp. 748-761. mation in Deutschland." Cassel. 126 1940. Review of "Die Fossilen Amphibien," PP. by 0. Kuhn, Berlin, 1939, Jour. MORTON, D. J. Paleont., XIV, pp. 389-391. 1926. "Notes on the footprint of Thinopus ROMER, A. S., AND BYRNE, F. antiquUs." Amer. Jour. Sci. XII, pp. 1931. "The pes of Diadectes: Notes on the 409-414. primitive tetrapod limb." Paleobio- 1935. "The human foot." Columbia Univ. logica, IV, pp. 25-48. Press, 244 pp. ROMER, A. S., AND PRICE, L. W. NAUCK, E. T. 1940. "Review of the Pelycosauria." Geol. 1924. "Die Beziehungen zwiscben Becken- Soc. Amer. Special Papers, No. 28, 538 stellung und Gliedmassenstellung bei PP. tetrapoden Vertebraten." Morph. SCHAEFFER, BOBB Jahr., LIII, pp. 1-47. 1941. "The pes of Bauria cynops Broom." NOBLE, G. K. Amer. Mus. Novitates, No. 1103, pp. 1931. "Biology of the Amphibia." McGraw- 1-7. Hill Co. New York. 577 pp. SXVE-S6DERBERGH, G. OSAWA, G. 1934. "Some points of view concerning the 1898. "Beitriige zur Anatomie der Hatteria evolution of the vertebrates and the 4.72 Bulletin American Museum of Natural History [Vol. LXXVIII

classification of this group." Arkiv. 1913b. "On the primitive tetrapod limb." Zool., XXVI-A, No. 17, pp. 1-20. Anat. Anz., XLIV, pp. 24-27. SCHMALHAUSEN, I. 1914. "Procolophon trigoniceps, a cotylosaur- 1917. "On the extremities of Ranidens sibiri- ian reptile from South Africa." Proc. cuS Kessl." Rev. Zool. Russe, II, pp. Zool. Soc. London, pp. 735-747. 129-135. 1926a. "The evolution and origin of the Am- SCHMALHAUSEN, J. J. phibia." Philos. Trans. Roy. Soc. 1908a. "Die Entwickelung der Sekelettes der London, (B) CCXIV, pp. 189-257. hinteren Extremitat der Anuren Am- 1926b. "The Carboniferous Amphibia of Scot- phibien." Anat. Anz., XXXIII, pp. land." Paleo. Hung., Budapest, I, pp. 337-344. 221-252. (Issued in 1929.) 1908b. "Zur Morphologie des Saugetier- 1931. "On the skeleton of a bauriamorph fusses." Anat. Anz., XXXIII, pp. reptile." Proc. Zool. Soc. London, Pt. 373-378. 3, pp. 1163-1205. 1910. "Die Entwickelung des Extremitiaten- WETTSTEIN, 0. VON skelettes von Salamandrella Kayser- 1931. "Ordnung der Klasse Reptilia: Rhyn- lingii." Anat. Anz., XXXVII, pp. chocephalia; Handbuch der Zoologie, 431-446. Siebenter Band-Sauropsida." Berlin SEWERTZOFF, A. N. und Leipzig. 224 pp. 1908. "Studien uber die Entwickelung der WHITE, T. E. Muskeln, Nerven und des Skeletts der 1939. "Osteology of Seymouria baylorensis Extremitaiten der niederen Tetra- Broili." Bull. Mus. Comp. Zoo6., poda." Bull. Soc. Imp. Nat. Moscou, LXXXV, pp. 325-409. (N. S.) XXI, pp. 1-430. WIEDERSHEIM, R. SIEGLBAUER, F. 1876. "Die altesten Formen des Carpus und 1909. "Zur Anatomie der Schildkroten- Tarsus der heutigen Amphibien." extremitalt." Arch. f. Anat. u. Phys., Morph. Jahrb., II, pp. 421-433. Anatom. Abt., pp. 183-280. WILDER, H. H. SJ6GREN, S. J. 1912. "The appendicular muscles of Necturus 1940. "Die Entwicklung des Hand- und maculosus." Zool. Jahrb., Suppl. XV, Fuss-Skelets bei Chamaeleon bitae- II, pp. 383-424. niatus Ellioti." Morph. Jahrb., WILLISTON, S. W. LXXXV, pp. 243-260. 1908. "The Cotylosauria." Jour. Geol., STEEN, M. C. XVI, pp. 139-148. 1938. "On the fossil Amphibia from the Gas 1909. "New or little-known Permian verte- Coal of Nyrany and other deposits in brates: Trematops, new genus." Czechoslovakia." Proc. Zool. Soc. Jour. Geol., XVII, No. 7, pp. 636-658. London, (B) CVIII, Pt. 2, pp. 205-283. 1911a. "A new family of reptiles from the STEINDLER, A. Permian of New Mexico." Amer. 1935. "Mechanics of normal and pathologi- Jour. Sci., XXXI, pp. 378-398. cal locomotion in man." Thomas, 191 lb. "American Permian vertebrates." 424 pp. Univ. Chicago Press. 145 pp. STEINER, H. 1914. "The osteology of some American 1920-21. "Hand und Fuss der Amphibien, Permian vertebrates." Jour. Geol., XXII, No. 4, pp. 364-419. ein Beitrag zur Extremitiatenfrage." 1925. "The osteology of the Anat. Anz., LIII, pp. reptiles." 513-542. Cambridge. Harvard Univ. Press. 1934. "Ueber die Embryonale Hand-und 300 pp. Fuss-Skelett-Anlage bei den Crocodi- WINTREBERT, P. liern, sowie uber ihre Beziehungen zur 1910. "Sur le determinisme de la m6tamor- Vogel-Flutgelanlage und zur ursprutng- phose chez les Batraciens: XVIII. lichen Tetrapoden-Extremitat." Rev. L'origine des Urodeles." C. R. Soc. suisse Zool., XLI, pp. 383-396. Biol., LXIX, pp. 172-174. TORNIER, G. 1922. "La vou'te palatine de Lysorophus." 1928. "Zur Phylogenese des Landwirbeltier- C. R. Soc. Biol., LXVII, pp. 928-930. Hinterfusses und des Menschen." ZITTEL, K. VON Sitz. Gesell. Natur. Freunde zu Berlin, 1932. "Text-book of palaeontology." Second Jahr. 1927, pp. 22-35. English Ed., Vol. II. Macmillan and TSCHERNOFF, N. D. Co., London, 464 pp. 1907. "Zur Embryonalentwickelung der hin- ZWICK, W. teren Extremitaten des Frosches." 1898. "Beitrage zur Kenntnis des Baues und Anat. Anz., XXX, pp. 593-612. der Entwicklung der Amphibien- WATSON, D. M. S. gliedmassen, besonders von Carpus 1913a. "The limbs of Lystrosaurus." Geol. und Tarsus." Zeitschr. Wissen. Zool.. Mag., (N. S.) Dec. V. X, pp. 256-258. LXIII, pp. 62-114.

-- :~

4Al d -tA ., 7

*1,- 'f

5<,/f .i '!' , .'

Ili ','i:4t:X JrX 4 4

W/?A jij' 4 4

~~44<~~~~~4$ ~ 4

7tV-""k'

X '4 -' 4

''~~~~"'~~Al -4 o