BIOLOGY APR , ,2?b '» » ^eM^/vc .• 2J^f^ ^^^^ The person charging th.s ma enaH re^ sponsible for its return to the library ^"-o^ which it was withdrawn on or betore the Latest Date stamped below. for dUtlpllnary action and may the University. ^ Li61_O-10<)6 THE GIANT PANDA A MORPHOLOGICAL STUDY OF EVOLUTIONARY MECHANISMS D. DWIGHT DAVIS M'- ''y^ ^'^PU "'//: ^r ^ FIELDIANA: ZOOLOGY MEMOIRS VOLUME 3 Published by CHICAGO NATURAL HISTORY MUSEUM DECEMBER 7, 1964 t-L-i) Ob FIELDIANA: ZOOLOGY MEMOIRS VOLUME 3 NATURAL HISTORY" MUSEUM lTllU««;H!f V l"i ! !l ^ HOTWmilll . I CHICAGO NATURAL HISTORY MUSEUM CHICAGO, U.S.A. 1964 I/. 3 Cop''? THE GIANT PANDA A MORPHOLOGICAL STUDY OF EVOLUTIONARY MECHANISMS Photograph by Waldemar Meisler THE GIANT PANDA MEI LAN Chicago Zoological Park, September, 1952 THE GIANT PANDA A MORPHOLOGICAL STUDY OF EVOLUTIONARY MECHANISMS D. DWIGHT DAVIS Curator, Division of Vertebrate Anatomy FIELD lANA: ZOOLOGY MEMOIRS VOLUME 3 Published by CHICAGO NATURAL HISTORY MUSEUM DECEMBER 7, 1964 "The field of macrotaxonomy ... is not directly accessible to the geneticist . . . Here the paleontologist, the comparative anatomist, and the embryologist are supreme." Richard Goldschmidt Edited by Lillian A. Ross Patricia M. Williams Edward G. Nash Publication costs defrayed in part by National Science Foundation Grant GN-116 Library of Congress Catalog Card Number: 6i-8995 PRINTED IN THE UNITED STATES OF AMERICA BV CHICAGO NATURAL HISTORY MUSEUM PRESS 6fO a ^3 J f S' ' ^' r PREFACE This study of the anatomy of the giant panda was originally intended to determine the taxo- nomic position of this species. As the dissection progressed, other questions of rather broader in- terest developed, and the scope was widened to embrace them. In studies of this kind the customary procedure is to compare structures with those of supposedly related organisms, and estimate relationships of organisms from these comparisons. In the back- ground are the broader questions of the phylogeny and fundamental uniformity of vertebrate struc- tures, which have long been the core problems of comparative anatomy. But superimposed on the underlying pattern of uniformity there is a be- wildering array of differences, mostly adaptations to special ways of life. Phylogeny — continuity of ancestry — explains the uniformities in vertebrate structure. It cannot explain the differences, which represent the active creative aspect of evolution. Yet we cannot pretend to explain the history of vertebrate structure without rational theories to account for the differences as well as the uni- formities. The existence of an underlying uniformity in vertebrate structure is now so well documented that it is practically axiomatic, but comparative anatomists have scarcely begun to seek similarly adequate explanations for the differences in verte- brate structure. At this stage I believe it is of crucial importance to ask whether comparative anatomy can undertake to explain, in causal-ana- lytical terms, the structural differences that char- acterize taxa among vertebrates. If it cannot, then I would agree with the statement once made by D. M. S. Watson, that comparative anatomy is a term "now obsolescent." Such an extension of the goal of comparative anatomy assumes that the genetic backgrounds for the kind of morphological differences with which anatomists are concerned are so simple that they can be estimated with reasonable certainty by in- ferring causes from results, without resort to breed- ing experiments. For some of the primary differ- ences at the generic level this appears to be true. Evidence is steadily accumulating that, in verte- brates, a quite simple change in epigenetic mecha- isms may have a profound and extensively different end result. Moreover, the result is an integrated oi'ganism. This suggests that in favorable cases, and at low taxonomic levels, the comparative anat- omist may properly seek the mechanisms behind the differences he observes. In many ways the giant panda seems to be al- most ideally suited to a test of this thesis. I do not, of course, believe that I have explained com- pletely how the morphology of the giant panda arose from the morphology of the bears, or that everyone will accept my interpretations. I ask only that this study be regarded as a first approxi- mation, a first attempt to explain the structural differences between a derived and an ancestral organism in terms of causal mechanisms, an at- tempt to identify the raw materials on which natural selection acted. I am indebted to several institutions and in- numerable individuals for assistance in this study. On several occasions the United States National Museum allowed me to study skeletons housed there, and lent embalmed and osteological mate- rials for detailed study in Chicago. The American Museum of Natural History and Carnegie Mu- seum permitted me to study and measure skeletons in their collections. Much of the material on which the woi-k was based, including all the em- balmed giant panda material, originally came from the Chicago Zoological Park. Observations on liv- ing carnivores were made at both the Chicago Zoo- logical Park and the Lincoln Park Zoo. Over the years so many individuals have con- tributed to this study in various ways that it is impossible to thank them individually. I have profited particularly from numerous discussions with Dr. Harry Sicher, Dr. E. Lloyd DuBrul, Dr. Rainer Zangerl, Pi'ofessor Bryan Patterson, and Dr. Carl Cans. Dr. Zangerl made many X-ray pho- tographs for me. My late colleague. Dr. Karl P. Schmidt, repeatedly interrupted his own work to help me translate difficult German passages. In a work of this kind the artist tends to become almost a collaborator. I have been particularly fortunate in the several artists who worked with 6 PREFACE me from time to time: the late John C. Hansen, of the finer blood vessels and nerves in addition to who made most of the bone drawings; John J. making most of the drawings of the soft anatomy; Janacek; Miss H. E. Story, who dissected out most Miss Phyllis Wade; and Mrs. Edward Levin. D. D. D. CONTENTS PAGE List of Tables 9 Introduction 11 Goals and Methods of Comparative Anatomy 11 Material and Methods 13 History 14 Distribution 17 Habits and Behavior 20 External Characters 28 Description 28 Measurements 31 Growth 31 Proportions 31 Conclusions 40 Skeleton 41 The Skeleton as a Whole 42 Measurements 44 The Skull 46 The Skull as a Whole 47 Cranial Sutures and Bones of the Skull 62 Hyoid 64 Review of the Skull 65 Summary of Skull 74 The Vertebral Column 74 The Vertebral Column as a Whole 74 Descriptions of Vertebrae 78 Review of the Vertebral Column 84 Conclusions 85 The Thorax 85 Ribs 85 Sternum 87 Review of the Thorax 88 The Fore Leg 88 Bones of the Fore Leg 88 Review of the Fore Leg 100 The Hind Leg 102 Bones of the Hind Leg 102 Review of the Hind Leg 120 7 8 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 PAGE Discussion of Osteological Characters 122 Conclusions 124 Dentition 125 Description 125 Discussion of Dentition 127 Conclusions 130 Articulations 131 Articulations of the Head 131 Articulations of the Fore Leg 132 Articulations of the Hind Leg 140 Review of Joints 145 Muscular System 146 Muscles of the Head 149 Muscles of the Body 158 Muscles of the Fore Leg 172 Muscles of the Hind Leg 183 Discussion of Muscular System 196 Conclusions 198 Alimentary System 199 Mouth 199 Salivary Glands 199 Tongue 202 Pharynx and Esophagus 204 Stomach 207 Intestines and Mesentery 208 Liver and Gall Bladder 212 Pancreas and Spleen 215 Discussion of Digestive System 216 Conclusions 218 Urogenital System 219 Urinary Organs 219 Male Reproductive Organs 221 Female Reproductive Organs 225 Discussion of Reproductive Organs 225 Conclusions 228 Respiratory System 229 Larynx 229 Trachea 235 Lungs 236 Conclusions 237 Circulatory System 238 Heart 238 Arteries 245 Veins 280 CONTENTS 9 PAGE Ductless Glands 288 Hypophysis 288 Thyroid 288 Parathyroid Bodies 288 Thymus 288 Nervous System 289 Brain 289 Discussion of Brain 297 Cranial Nerves 298 Cervical Plexus 305 Nerves of the Fore Limb 306 Thoracic Nerves 310 Nerves of the Hind Limb 311 Sympathetic System 315 Special Sense Organs 317 Eye 317 Middle Ear 318 Comparative Anatomy and evolution— An Evaluation of the Test Problem The Relationships of Ailuropoda 322 Morphogenetic Mechanisms in the Evolution of Ailuropoda 323 Conclusions 326 References 328 Index 335 LIST OF NUMBERED TABLES PAGE 1. Limb segment ratios in carnivores 35 2. Limb proportions in carnivores 36 3. Weight in grams of dry skeleton 42 4. Weight ratios in dry postcranial skeleton 43 5. Surface areas of limb bones 43 6. Measurements of carnivore skeletons 45 7. Cranial capacity of carnivores 46 8. Skull proportions in generalized and specialized carnivores 66 9. Vertebral counts in carnivores 75 10. Relative proportions of divisions of the vertebral column in carnivores 75 11. Measurements and indexes of pelvis in carnivores 103 12. Relative mass of masticatory musculature 154 10 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 PAGE 13. Relative weights of masticatory muscles in carnivores 155 14. Relative weights of muscles of the shoulder and arm in carnivores 183 15. Relative weights of muscles of the hip and thigh in carnivores 195 16. Myological characters in arctoid carnivores 197 17. Intestinal length in arctoid carnivores 210 18. Liver weight in mammals 214 19. Percentage differences from control animals in gut measurements of pigs raised on herbivorous and carnivorous diets 217 20. Number of renculi composing kidney in bears 220 21. Dimensions and proportions of kidneys in arctoid carnivores 220 22. Kidney weights in mammals 221 23. Heart structure in arctoid carnivores 243 24. Branches of aortic arch in arctoid carnivores 277 25. Composition of lumbosacral plexus in carnivores 315 INTRODUCTION It is my intent to make this study a test, based on the anatomy of the giant panda, of whether the comparative method can yield information that goes beyond the customary goals of comparative anatomy. It is evident, to me at least, that more than fifty years ago comparative anatomy reached a stalemate that can be broken only by seeking answers to new and different questions. I believe it must shift its major emphasis from the conserva- tive features of evolution to its radical features, from the features that organisms under compari- son have in common to those they do not have in common. It must seek rational explanations for these differences, drawing on data from other fields where this is necessary and possible. In this study of the giant panda the structural differences be- tween it and the bears, and the ways in which these differences arose, will be our primary concern. The original problem that motivated the work — the proper taxonomic position of Ailuropoda — was soon settled; Ailuropoda is a bear and therefore belongs in the family Ursidae.' The further prob- lem of attempting to infer the causal mechanisms involved in the origin of Ailuropoda from its ursid ancestors requires some discussion of goals and methods. GOALS AND METHODS OF COMPARATIVE ANATOMY The classical goal of comparative anatomy was to demonstrate the existence of an essential and permeating uniformity or "ordering" in the struc- ture of vertebrates. This goal has been reached. Details of the picture remain to be filled in, but the unifying concept itself is now so well docu- mented that it is no longer open to serious debate. Phylogeny, the genetic relatedness of all verte- brates, provides an explanation for the uniformity. This aspect of the history of vertebrate structure cannot be expected to give rise to further concepts. > This conclusion is not based on one or a few characters, but on a host of similarities, many of them subtle, through- out the anatomy. I tried to present the data on the affinities of Ailuropoda before going on to other considerations, but this became so difficult that I gave it up. Therefore one of the primary conclusions is assumed throughout the text. We may well ask where comparative anatomy is to go from here. From the evolutionary standpoint the structural differences among vertebrates are just as impor- tant as the structural uniformities; these two are, in fact, the obverse and reverse of the phylogenetic picture of vertebrate structure. Years ago W. K. Gregory distinguished them as "habitus" and "her- itage" characters. We cannot claim to have ex- plained the particular structure of an organism if we explain only its heritage characters and offer no explanation for its habitus characters. An "expla- nation" must account for the differences in terms of evolutionary mechanisms, not merely relate them to the functional requirements of the organism — in other words, explain them in the same causal sense that common ancestry explains the heritage characters. Classical comparative anatomy tended to con- centrate on the major features of vertebrate struc- ture— the differences that characterize orders and moi-e often classes. Such, for example, are the homologies of the gill arch derivatives, of the ele- ments of the mammalian middle ear, of the cranio- mandibular muscles. There was practically no in- terest in how and why such changes came about, and the morphogenetic and selective mechanisms involved in these massive alterations are prob- ably irretrievably lost in the vast reaches of time anyway. Structural differences at about the generic level are usually far less profound and more recently evolved, yet they often represent a level of mor- phological differentiation to which the methods of comparative anatomy can be applied. In this re- spect they differ from the characters with which the geneticist customarily deals. At about the generic level we may hope to decipher the mech- anisms responsible for the observed differences in structure between two or more related forms. A procedure designed to yield such information is followed in this study. The procedure may be divided into a series of steps. These are: (1) Identification of the structural differences between Ailuropoda and its structural ancestor, Ursus. At the outset nothing was known of pos- 11 12 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 sible pleiotropic effects, allometric relationships, morphogenetic patterns, or obscure functional re- lationships. Therefore all differences were tabu- lated uncritically, without attempting to evaluate them. For the same reason the entire anatomy of the organism was covered so far as practicable. (2) Correlation of the observed structural differ- ences between Ailuropoda and Ursus with dif- ferences in habits or behavior. This is the first step in sorting out the adaptive features peculiar to the anatomy of Ailuropoda — features that pre- sumably represent the modifications of the ursid morphology resulting from natural selection. This step results in two categories: (a) those differences that can be correlated with differences in habits or behavior, and (6) those that can not. The differences in category b may be conspicu- ous, and their presence must be accounted for. They may be genetically related to an adaptive modification but not themselves adaptive. They may reflect the results of an inherited differential growth rate, whereby the proportions of a mor- phological unit may change with the absolute size of the unit. A classical example of this effect is the antlers of deer. They may merely reflect re- laxed selection pressure on certain functions. The decision as to whether a condition is or is not adap- tive is often very difficult, requiring considerable knowledge of mechanics and engineering, as well as intimate knowledge of the habits and behavior of the animal. (3) Separation of the adaptive features that are genetically determined from those that are only indirectly related to the genetic substrate. Many conspicuous features in the skeleton depend only on the capacity of bone to respond to extrinsic forces. Many soft tissues have a considerable ca- pacity to accommodate their form to the molding action of extrinsic forces. The caliber of blood vessels varies with the demands of the tissues they supply, even during the life of the individual; if one kidney is removed, the remaining one hyper- trophies. Such conditions are adaptive, but they are not primary results of selection; they are the ex- ogenous adaptations of Waddington (1953). They reflect the action of natural selection at second or third hand, so to speak. If we are seeking to iso- late the structural features on which natural selec- tion acted directly, these secondary and tertiary effects must be discounted. These three steps have presumably isolated the morphological features in Ailuropoda that (1) dif- fer from those in its structural ancestor, Ursus, (2) are functionally correlated with differences in habits and behavior, and (3) are genetically deter- mined. They are the direct results of natural se- lection in the step from bear to giant panda. As will appear in the sequel, these features seem to be surprisingly few; we are not interested here in minor polishing effects, but only in decisive dif- ferences. We do not yet know the materials on which natural selection acted to effect these changes. One final step remains: (4) Determination of the morphogenetic mech- anisms that were involved in effecting these changes. This should be an experimental prob- lem, but obviously experimentation is impossible in the vast majority of cases, including this one. Fotunately, morphogenetic processes appear to be remarkably uniform among mammals. By a judi- cious combination of the comparative method with the known data of mammalian epigenetics I be- lieve it is possible to infer, with varying degrees of confidence, the true mechanisms behind many of the major structural differences that distinguish Ailuropoda from the true bears. Many of the "unit characters" involved appear to be sizable morphological units, although it does not neces- sarily follow that the shift from bear condition to panda condition was made in one jump, or that such morphological units are controlled by simple genes. It is clear, however, that they are geneti- cally controlled as units. It would be futile to at- tempt to reconstruct the history if major adaptive differences represent accumulations of numerous small mutation effects. To the extent that these four steps are carried out successfully, the differences between the giant panda and the true bears will be explained rather than simply described. Almost without exception, students of the higher taxonomic categories have been reluctant to believe that the kinds of morphological differences they observe represent accumulations of small muta- tion effects such as the geneticist customarily deals with. The once-popular solution — invoking un- known imminent forces to explain systemic differ- ences— is no longer common. Modern students have sought genetic mechanisms capable of pro- ducing phenotypic differences of the magnitude they believed were involved. Goldschmidt (1940), for example, emphasized (among other things) the massive co-ordinated differences that can result from acceleration and retardation of gene-con- trolled developmental processes. Rensch (1960) listed pleiotropy, allometric growth rates, and compensatory correlations — among the agents ac- cessible to natural selection — as capable of pro- ducing extensive generalized effects on the organism as a whole. It is now generally recognized that gi'owth is es- sentially a process of multiplication of cells. Multi- DAVIS: THE GIANT PANDA 13 plicative rates differ in different parts of the body, and in the same part at different times during its growth period. Regional growth rates may inter- fere with each other, resulting in negative interac- tions and in extreme cases even in deformation of the entire growth profile of the body. Correlation studies show clearly that both regional and gen- eral growth rates are genetically controlled as units. These insights stem chiefly from Huxley's Prob- lems of relative growth, which in turn grew out of the earlier On growth and form of D'Arcy Thomp- son and Goldschmidt's Physiologische Theorie der Vererbung. They provide a mechanism capable of producing plastic deformation of a common pat- tern, which is what the comparative anatomist seems to see when he compares homeomorphic or- ganisms. A bridge between genetics and compara- tive anatomy was sought in vain during the first third of this century; it now seems to have been found. Partly because evolution is a cumulative and non-repetitive process, and partly because growth fields in vertebrates have proved refractory to ex- perimental techniques, their role in the morphosis of animal form has been deciphered almost exclu- sively by morphological methods. The primary tool is demonstration of correlations; the method is comparative. Whether subtle correlations are sought by sophisticated statistical methods (as in recent studies of mammalian teeth) , or more obvi- ous correlations by means of coarser but no less rigorous comparative methods (as in the present study), the goal is the same. It is to identify and circumscribe the material bases for differences among homeomorphic organisms. This is a proper field for the comparative anatomist. MATERIALS AND METHODS This study is based largely on the embalmed and injected body of a giant panda that lived in the Chicago Zoological Park from February, 1937, to April, 1938. The panda was popularly known as Su Lin. Unless otherwise stated, all statements relating to the soft anatomy are based on this specimen. Su Lin was a subadult male (teeth fully erupted). His age (estimated) was 16 months at death. He was in excellent condition and weighed 132 pounds. Preserved portions of the carcass (head, fore and hind limbs, heart, genitalia) and the skeleton of an adult male giant panda (known as Mei Lan) were available. Mei Lan was esti- mated to be 15 years old at death. He was much emaciated, and weighed 205 pounds after autopsy. The following skeletal material of Ailuropoda was available for detailed study: CNHM 31128 id' ad.) Szechwan: Yehli. Complete skel- eton. CNHM 36758 ( 9 ad.) Szechwan: Dun Shih Goh. Com- plete skeleton. CNHM 34258 (- ad.) Szechwan: Mouping Dist. Skull, lower fore legs, fore and hind feet. CNHM 74269 ( (f ad.) (zoo animal: Mei Lan). Complete skeleton. CNHM 39514 (- ad.) Szechwan: Dun Shih Goh. Skull. USNM 259076 ( 9 jv.) Szechwan: Wen Chuan. Skull. USNM 259027 ( c? ad.) Szechwan: Wen Chuan. Pelvis. USNM 259403 ( 9 ad.) Szechwan: Wen Chuan. Pelvis. Most of the data on the soft anatomy of bears came from the following captive animals that died in the Chicago Zoological Park: CNHM 48304 (d" ad.) Ursus thibetanus, embalmed and injected body. CNHM 49061 ( & juv.) Ursus americanus, embalmed and injected body. CNHM 57267 ( 9 ad.) Ursus americanus, embalmed head, fore leg, and hind leg. CNHM 57200 ( 9 ad.) Tremarctos ornatus, embalmed head, fore leg, and hind leg. The following bear skeletons were used for most of the detailed osteological data: CNHM 43744 (- ad.) Ursus ardos; Iraq. CNHM 47419 (- ad.) Ursus arctos; Iraq. CNHM 44725 (cf ad.) Ursus americanus; (zoo animal). These three skeletons were supplemented with numerous skeletons and partial skeletons of bears, representing several genera and species, in the col- lections of Chicago Natural History Museum. Partial dissections were made of several procyo- nids, all embalmed zoo animals, representing the genera Procyon, Nasua, Bassariscus, Potos, and Ailurus. Numerous skeletons of these genera, from both wild-killed and zoo animals, were available. Linear measurements up to 150 millimeters were made with Vernier calipers graduated to 0.1 milli- meter. Lengths beyond 150 millimeters were meas- ured with large calipers and a meter stick. Weights up to 2 kilograms were determined with a small Ohaus triple beam balance. Larger objects were weighed on a large Ohaus beam balance with a capacity of 21 kilograms. In weighing preserved soft tissues the usual precautions of removing ex- cess surface liquid by blotting were taken. HISTORY The synonymy of Ailuropoda melanoleuca may be summarized as follows : Ailuropoda melanoleuca (David) Ursus melanoleuciis David, 1869, Nouv. Arch. Mus. Hist. Nat., Paris, Bull. 5, p. 13. Ailuropoda melanoleuca Milne-Edwards, 1870, .\nn. Sci. Nat., Paris, (5), Zool., 13, art. 10. Pandarctos melanoleucus Gervais, 1870, Nouv. Arch. Mus. Hist. Nat., Paris, 5, p. 161, footnote; 1875, Jour. Zool., Paris, 6, p. 87. Ailuropus melanoleucus Milne-Edwards, 1871, Nouv. Arch. Mus. Hist. Nat., Paris, Bull. 7, p. 92. Aeluropus melanoleucus Lydekker, 1891, in Flower and Lydekker, Mammals living and extinct, pp. 560-561, fig. 256. During his stay in Mouping on the second of his three expeditions to China, the noted French ex- plorer and naturalist Pere Armand David learned of the existence of a curious black and white "bear." This animal, called pei-hsuing ("white bear") by the natives, aroused David's interest, and he em- ployed hunters to capture specimens of it for him. After almost a month of unsuccessful hunting a young female was brought to him on March 19, 1869, and two weeks later he acquired an adult of the same sex. Although erroneously believing it to be a bear, David immediately recognized the animal as a novelty to science. He drew up a con- cise but adequate description under the name Ur- sus melanoleucus and despatched it to Alphonse Milne-Edwards at the Paris Museum with an ex- planatory note requesting its publication. David's letter, which was duly published in the Nouvelles Archives of the Paris Museum, introduced to sci- ence the animal now known as the giant panda. The subsequent history of the giant panda can best be presented in chronological form. 1870. Milne-Edwards, after e.xamining David's material, noted that its osteological characters and dentition "clearly distinguish" the giant panda from the bears and approach those of the lesser panda and raccoons. He erected the genus Ailuro- poda to receive it. Gervais, on the other hand, concluded from a study of an intracranial cast that its brain structure allies it to the bears. Gervais considered it worthy of generic distinction, how- ever, and proposed the name Pandarctos. 1871. David published a few brief notes on the habits of the giant panda, and even today sur- prisingly little can be added to these original ob- servations. David recorded that it is restricted to high altitudes, that it is herbivorous, and that it does not hibernate. Only one of his statements has not been substantiated : "It is said that it does not refuse meat when the occasion presents itself; and I even think that this is its principal nourish- ment in winter." Milne-Edwards, believing that the generic name Ailuropoda was preoccupied by Gray's use of the name Aeluropoda for his "Section I. Cat-footed Carnivora" in the Catalogue of Carnivorous, Pachy- dermatous and Edentate Mammalia in the British Museum (1869, p. 3), proposed the name Ailuropus for the giant panda. 1868-74. Milne-Edwards, in the Recherches des Mammiferes, gave a detailed description of the skin, skull, and dentition. His re-examination led him to the conclusion that Ailuropus should be placed between the bears and the [lesser] panda. 1875. Gervais, after an examination of the skeleton of David's panda, reasserted his former opinion that the giant panda is an aberrant bear. 1885. Mivart, in his careful review of the clas- sification of the arctoid carnivores, concluded that Ailurus is a procyonid and that Ailuropus is allied to Ailurus and therefore is a procyonid, too. Mi- vart thus set the pattern that, with few exceptions, has been followed by British and American auth- ors to the present day. His conclusion is based on the usual agreement of skull architecture and den- tal morphology that was to be stressed repeatedly by later authors. 1891. Flower and Lydekker, in their Mammals Living and Extinct, placed "Aeluropus" in the Ur- sidae and "Aelurus" in the Procyonidae. Their emendation of Milne-Edwards' generic name Ailu- ropus, appearing in an authoritative work, resulted in considerable confusion in subsequent literature. 1895. Winge regarded the giant panda as a very close relative of the extinct Hyaenarctos [- Agriotherium of recent authors], these two gen- era forming a separate branch of the ursine stem. 14 DAVIS: THE GIANT PANDA 15 Ailurus, on the other hand, he considered a pro- cyonid. Winge's views have been adhered to with- out exception by continental European authors. 1901. Both Lankester and Lydekker, after in- . / dependently studying the skull and limb bones, concluded that Aeluropus and Aeliirus are closely allied, that they are procyonids, and that the Pro- cyonidae should be subdivided into two subfam- ilies, the Procyoninae and the Ailurinae. ' This, of course, is merely a re-affirmation of the earlier views of Mivart. They emphasized the procyno- nid-like presence of both protocone and hypocone on the lingual border of P^ (the protocone is absent in the Ursidae), the presence of an entepicondylar foramen, and numerous "minute coincidences" in the structure of the skull and long bones of the limbs. Lankester and Lydekker deemed it desirable that Aeluropus, which hitherto had been called the "parti-coloured bear," should henceforth be called the "great panda." This appears to be the first published reference to Ailuropoda as a panda.' 1902. Beddard, in his Mammalia, followed Flower and Lydekker in placing "Aeluropus" in the Ursidae and "Aelurus" in the Procyonidae. 1904. Weber, in the first edition of Die Sduge- tiere, followed Winge in considering Aeluropus as an ursid closely related to Hyaenarctos and refer- ring Ailurus to the Procyonidae. 1913. Bardenfleth made a detailed study of the dental and osteological characters of Ailuro- poda and concluded that its resemblances to Ailu- rus are due to convergent development of the molar teeth based on herbivorous diet, and that its closest affinities are with the extinct ursids of the Hyaenarctos group. 1915. Woodward described the well-preserved skull of a Pleistocene giant panda, which he named Aelureidopus baconi, from Burma. This was the first proof that the giant panda once had a more extensive range than it has at present. 1921. Pocock, in a review of the classification of the Procyonidae, concluded that both Ailuro- poda and Ailurus represent distinct and separate families. This view he re-affirmed in 1929 and also in his article "Carnivores" in the fourteenth edi- tion of the Encyclopaedia Britannica, where no fewer than 13 families (compared with 7 of other authors) and 29 subfamilies (18 of other authors) of living fissiped carnivores are recognized. Po- cock's "families" correspond roughly to the gen- era of other authors. ' The word "panda," which had been applied to the lesser panda (Ailurus) since the time of Cuvier, is "said to be a Nepal name." (Oxford Universal English Dictionary.) 1923. Matthew and Granger described giant panda material, under the name Aeluropus fove- alis, from Pliocene deposits in eastern Szechwan, thus farther extending the former range of the giant panda. 1928. Weber, in the second edition of Die Sdu- getiere, retained his views of 1904 as to the ursid affinities of Ailuropoda. 1929. Theodore and Kermit Roosevelt shot a giant panda at Yehli, Sikang Province. This in- dividual, said to be the first giant panda shot by a white man, was mounted, together with a sec- ond skin purchased from natives, in a habitat group in Chicago Natural History Museum. The ensuing publicity started a cycle of "giant panda expeditions" that have gi'eatly increased our knowl- edge of the distribution, habits, and morphology of this animal. 1936. Gregory examined the skull and denti- tion of Ailuropoda, Ailurus, and various fossil and recent procyonid and ursid carnivores. He con- cluded that Lankester and Lydekker were correct in referring Ailuropoda and Ailurus to the Procy- onidae. Raven, in the same year, studied the viscera of a giant panda, which had been preserved in the field by an American Museum expedition. He listed six points of agreement between Ailuropoda and Ailurus, and concluded that resemblances be- tween the former and the bears "are an expression of convergence in size and food habits." 1937. Mrs. Ruth Harkness, of New York City, succeeded in bringing a living baby giant panda to the United States. This individual, named Su Lin, lived for 16 months in the Chicago Zoological Park. It formed the basis for the present monograph. The fanfare that surrounded the life and death of Su Lin started a new series of expeditions for living pandas. At least a dozen have since been exhibited in the United States and Europe. 1943. Segall made a study of the auditory re- gion in the arctoid carnivores. The structure of the bony auditory region and auditory ossicles led him to associate the Ailuridae (Ailurus and Ailuro- poda) with the Ursidae. 1945. Simpson, in his Classification of Mam- mals, adhered to the classical view of Mivart in grouping Ailurus and Ailuropoda in the subfamily Ailurinae of the family Procyonidae. 1946. Mettler and Goss, after studying the topography of the brain of an adult giant panda, concluded that "the configuration of the brain of Ailuropoda melanoleuca is identical with that of the bear." 1956. Leone and Wiens reported that compari- sons of serum proteins by means of precipitin tests 16 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 V. "clearly indicate that the giant panda belongs in the family Ursidae." An examination of this history of research is instructive. There can be no doubt that the giant panda occupies a more or less isolated position among living carnivores, and that the features usu- ally relied upon by mammalogists for determining affinities are masked by si>ecializing adaptations in this form. Two conclusions may be drawn from these historical data. 1. Quite different conclusions have been reached by a succession of capable investigators on the basis of the same data. This indicates that the data employed are not sufficient to form a basis for an objective conclusion, and that opinion has been an important ingredient in arriving at con- clusions. 2. Opinion as to the affinities of Ailuropoda is divided almost perfectly along geographic lines, which shows that authoritarianism i-ather than ob- jective analysis has really been the determining factor in deciding the question. After the pioneer- ing work of Milne-Edwards and Gervais, the first attempt at determining the affinities of Ailuropoda was made by Mivart in England. Mivart's con- clusion— that both the giant and the lesser panda are procyonids — has been echoed by every British and American author down to 1943, except for the short-lived dissenting opinion of Flower and Ly- dekker.' In the meantime, on the continent, Winge in 1895 relegated Ailuropoda to the Ursi- dae and Ailurus to the Procyonidae, and every subsequent continental authority has followed in his footsteps. Such a cleavage of opinion along geogi^aphical and linguistic lines cannot be due to chance. It is apparent that the relationships of Ailuro- poda will never be decided on the basis of the data afforded by the skeleton and dentition. Thus the fii-st task of this study was to examine data not previously available, with a view to determining the much-discussed affinities of this carnivore. > Beddard (1902) merely copied Flower and Lydekker. DISTRIBUTION The giant panda apparently has a very re- stricted distribution in the high mountains of western Szechwan and eastern Sikang in western China. This is the area of the extremely complex mountain escarpment that sharply separates the Min River Valley from the Tibetan highland to the west. Localities where or near which specimens have been collected are shown on the accompanying map (fig. 1). The localities given on many mu- seum specimens obviously represent the city where the skin was purchased (e.g., Mouping, Ya-chou) rather than the locality from which the specimen actually came. Localities given in the literature ("Moupin," David, 1869; "mountains of Mou- ping," Gervais, 1875; "Wassu mountains," "moun- tains east of Min valley," Jacobi, 1923a) are often very indefinite. Thus the localities that can be plotted with any certainty on a map are relatively few, although none of the unplottable localities ex- tends the known range of this species. The dis- tance between the southernmost record (Yehli) and the northernmost (25 miles west of Wen- chuan) is only about 175 miles. All records, ex- cept Yehli are on the slopes of the Chuing-lai mountains surrounding the valley of the Min River. Yehli, where the Roosevelt brothers shot their panda, is on the slopes of the Ta-liang Moun- tains south of the Tung River. Pen (1943) reported Ailuropoda horn "the up- per source of the Yellow [Yangtze] River where it connects the two lakes, the Tsaring Nor and the Oring Nor, near the central part of Chinghai prov- ince" at 34° 7' N. Lat. Pen refers, without cita- tion, to a record by Berozovski at 34° N. Lat., but I have been unable to find such a reference. Pen collected no specimens, but there seems to be no reason for doubting his identification of the ani- mals he saw. Even allowing this provisional ex- tension of range, the north-south distribution amounts to only about 470 miles. Sowerby (1932) has suggested even greater ex- tensions of the range of Ailuropoda. He writes: "The range of the giant panda is now admitted to be much more extensive than formerly supposed. . . . We came across indisputable evidence of the giant panda in the Tai-pei Shan region of South- western Shensi, where the local takin hunters de- scribed its appearance to us accurately and also showed us its droppings and the places where it had torn up the culms of bamboos for food. From this region it ranges southward throughout all the wilder mountainous areas at least to the Yunnan border, eveiywhere being known to the native hunters by its native name, pei-hsiung." Sowerby (1937a) later defined the range as "more or less restricted localities from the Tsing Ling range of mountains in southern Shensi and eastern Tibet to northern Yunnan." Others have emphasized the unreliability of reports by native hunters, how- ever, even after being shown pictures of the ani- mal, and it seems best to await more positive evidence before accepting Sowerby's broad exten- sions of range. Ailuropoda had a much more extensive distri- bution in comparatively recent geological times, as is shown by the two fossil records. Smith- Woodward (1915) described a Pleistocene panda under the name Aelureidopus baconi, from Mogok, Northern Shan States, Burma. This is in the Irra- waddy River drainage and is more than 500 miles southwest of the southern limit of the panda's range as now known. Granger (in Matthew and Granger, 1923) found giant panda material, which was named Aeluropus fovealis, in Pliocene deposits near Wan-hsien in eastern Szechwan. Wan-hsien is situated on the Yangtze River (of which the Min is a tributary), about 250 miles due east of Chengtu. Vertical Distribution The vertical distribution of Ailuropoda is as limited as its geographic distribution. All who have studied its habits agree that this animal is sharply limited to the bamboo zone, which lies between about 5,000 and 10,000 feet. Limited to the Si-fan region at altitudes of 1600 to 3300 m., consequently to the region of almost impenetrable bamboo jungle on the steep slopes. Here it forces tunnels through the thickets, which are IJ-^ to 5 m. high and are often matted by snow pressure. (Jacobi, 1923b, p. 72.) ... in the bamboo jungles in altitudes varying between six and fourteen thousand feet. We came to the conclusion that it could safely be assumed that where there were no 17 Fig. 1. Western Szechwan and eastern Sikang provinces, showing locality records for Ailuropoda melanoleuca. 18 DAVIS: THE GIANT PANDA 19 bamboo jungles, there were no beishung. (Theodore and Kermit Roosevelt, 1929, p. 261.) The limits of the giant panda's altitudinal range is deter- mined largely by the extent of the bamboo growth. Two exceptions to this statement were observed, however. In one case we found unmistakable panda droppings high on the Chen Lliang Shan range, 1000 feet above the rhododen- dron forest, and probably 1500 feet above the nearest bam- boo. It was interesting to find that on occasion the panda must travel above its regular habitat to the bare grasslands of the blue sheep country. In another instance I saw where a giant panda had climbed a small pine tree just above the village of Tsapei on Chengou River. It was located 300 feet above the river bottom on an open slope, with the nearest bamboo across the valley. (Sheldon, 1937.) The vertical distribution of the bamboo bear, which avoids the hot arid canyons as well as the high alpine zones, extends on the high levels between 1500 and about 4000 m., where it is closely confined to the moist, subtropical bamboo zone. (Schiifer, 1938.) Pen's sight record of a giant panda at the upper source of the Yangtze River was on the open steppe of the Tibetan plateau. He speculates that these animals may have reached the plateau country by migrating north and west along the bamboo zone of the mountains, and that there is here an annual summer migration onto the plateau, with a winter retreat into the less rigorous environment of the mountains. HABITS AND BEHAVIOR Because of the inaccessible and rugged nature of its habitat, there has been httle field observation of the giant panda. Various authors have re- corded information, beginning with the original notes of David, and the observations are in close agreement. Details of behavior are known only from observations on captive individuals (Schnei- der, 1939; Haas, 1963). HABITAT The giant panda appears to be closely confined to the moist bamboo zone on the slope of the high mountains. The bamboo culms, which are slender (up to an inch and a half in diameter) and grow to a height of 10 to 12 feet, form dense impene- trable thickets that are often matted by snow pres- sure. The bamboo jungle is associated with forests of fir trees, and at higher altitudes the bamboo gives way to rhododendron, into which the panda does not wander. The mountain slopes "under the influence of the summer-like monsoon rains, exhibit a comparatively mild subtropical climate." (Schafer, 1938.) The panda shares this habitat with such other large mammals as the golden monkey (Rhinopithe- cus), leopard (Panthera pardus), red dog {Cuon al- pinus), black bear (Ursus thibetanus), wild pig (Sus cristatus), barking deer (Muntiacus), serow (Capricornis), and takin {Budorcas). Only the leopard and the red dog would be likely to attack the giant panda, and such encounters would be uncommon.' Thus the giant panda is practically without natural enemies — an important point in estimating the selection pressures to which this species is subjected. Wilson (1913) described the vegetation on the mountain Wa Shan as follows: At one time a dense forest of Silver Fir covered the moun- tain. . . . Some of these Firs could not have been less than 150 feet in height and 20 feet in girth. . . . Besides the Silver Fir (Abies Delayayi), the only other conifers are Tsuga yun- nanensis, Juniperus formosana, and Picea complanata. Rho- dodendrons constitute the conspicuous feature of the vege- ' Seton (Lives of game animals, 2, 1929) lists the grizzly bear and the mountain lion as enemies of the American black bear, an animal about the same size as the giant panda. tation. . . . They begin at 7500 feet, but are most abundant at 10,000 feet and upwards. In the ascent I collected 16 species. They vary from diminutive plants 4 to 6 inches high, to giants 30 feet or more tall. . . . One of the common- est species is R. yanthinum. . . . Above this [7200 feet], for 500 feet, comes a wellnigh impenetrable thicket of Bamboo scrub. The species (Arundiruiria nilida) is of remarkably dense growth, with thin culms, averaging 6 feet in height. Next above this, till the plateau is reached, is a belt of mixed shrubs and herbs, conspicuous amongst which are Syringa Sargentiana, Hydrangea anomala, H. villosa, Neillia affinis, Dipelta ventricosa, Ribes longeracemosum, var. Davidii, Enki- anthus deflexus, Styrax roseus, Deutzia (2 spp.), Rubus (5 spp.), Viburnum (4 spp.), Spirea (4 spp.), Acer spp., Malus spp., Sorbus spp., Meconopsis chelidonifolia, Fragaria filipendulus, Lilium giganteum, and the herbs of the lower belt. A few Rhododendrons occur chiefly on the cliffs. The plateau (8500 feet) is about half a mile across, marshy in places, and densely clad with shrubby vegetation and Bamboo scrub. . . . From 10,000 feet to the summit of the mountain Rhododendron accounts for fully 99 per cent of the ligneous vegetation. FOOD All observers (except Pen, see below) agree that in its native state the giant panda subsists exclu- sively on bamboo. McClure (1943) identified the bamboo native to the haunts of the giant panda as Sinariindinaria sp. "Its food seems to consist exclusively of bamboo shoots, but by no means merely the young shoots, which even man himself eats with relish, but also those as thick as a finger. In winter, in fact, only strongly woody and silicified stalks are available. All this can be ascertained from fresh drop- pings, which consist almost exclusively of chewed-up stalks, often as long as a finger joint, whether in the middle of July or in the beginning of January." (Jacobi, 1923a.) Not only is the giant panda entirely herbivorous, but it is known to live on the dwarf bamboo of the northeastern spur of the Himalayas to the exclusion of all other vegetable matter. . . . The food supply in the mountains of west Szechuan is inexhaustible. . . . We found giant panda eating not only the bamboo shoots, but the stalks and leaves of fully mature sprouts, often an inch and one-half in diam- eter." The author followed a fresh morning trail and found "that at an average of every hundred yards there were from one to three large droppings (4 to 6 inches long and 2 inches thick, tapering at each end). At a conservative estimate there were 40 droppings. . . . Below the resting place was a pile of at least 30 more droppings, making a total of 70 ex- creted between early morning and 9 a.m These droppings 20 DAVIS: THE GIANT PANDA 21 Fig. 2. Sitting posture and use of fore paws in Ailuropoda. A-C, "Happy" eating bamboo in Leipzig Zoo (from Schneider, 1939). D, Mei Lan eating green cornstalks in Chicago Zoological Park. emerge almost totally undigested. It seems logical to assume that an animal of such large proportions must have to eat tremendous quantities to secure the nourishment that it requires. ... I estimate that they would have to spend from 10 to 12 hours a day feeding. (Sheldon, 1937.) The bear [Ailuropoda] prefers the young and succulent bamboo shoots to the woody stems. For this reason, in the main district of bamboo-bears I found no bamboo shoots in the spring, since they had been systematically 'browsed' by bears. The bulk of its nourishment consists, however, of stone-hard bamboo stems thicker than a finger. With its powerful molar teeth the bear bites off the 3 to 6 m. long stems about 20 to 40 cm. above the ground, lays them down and eats the middle part up to the beginning of the leaves, while it regularly rejects the lower, hard part and lets it lie. Such chewed places are not particularly hard to find, although they are always concealed in the middle of the jungle. Usually they are not larger than one to two square meters. In these places perhaps 15 to 20 stems are bitten off, and the rejected parts cover the ground. (Schafer, 1938.) McClure (1943) listed nine species of bamboo that are palatable to the giant panda, expressing astonishment at the range of its tastes. Sowerby (1937a) stated that a half-grown pet giant panda that wandered at will on a Chinese farmer's land "ate grass and other plants." Pen (1943) stated that a giant panda he ob- served at a distance of 2000-3000 meters on the 22 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Fig. 3. Use of fore paws in pandas. A, Ailuropoda (Mei Lan) using both fore paws to manipulate food; Chicago Zoological Park, September, 1952. B-D, Lesser panda (Ailurus fulgens) using fore paws to manipulate bamboo; Lincoln Park Zoo. Tibetan plateau was eating plants of various kinds, "principally gentians, irises, crocus, Lycium chi- nense and tufted grasses." Unfortunately it is not clear from his description how careful his observa- tion was, and this is the only reported field obser- vation of the giant panda's eating anything other than bamboo. Captive specimens of Ailuropoda have eaten — in addition to various bamboos — porridge, green corn stalks and ears, stalks of celery, carrots, and other vegetables. They refuse meat in captivity. Thus in nature the giant panda lives immersed in its food supply. It has practically no natural enemies, does not pursue prey, and does not need to wander in search of food. Demands on loco- motor efficiency are absolutely minimal. FEEDING AND MANIPULATION OF FOOD The manner of eating bamboo was well described by Schneider (1939), who carefully observed a 200-pound female temporarily exhibited in the Leipzig Zoo. The animal always sat or lay when eating bamboo, thus freeing the fore feet (fig. 2). Only the stalks were eaten; the leaves were re- jected. The bamboo stalks were held in the fore foot and carried to the mouth. The tough outer layer was quickly and skillfully stripped off with DAVIS: THE GIANT PANDA 23 the incisors, in which case the stalk was inserted transversely into the mouth, or with the canines and anterior premolars, in which case it was shoved lengthwise between the upper and lower tooth- rows. The stripped outer layer was torn off with a twisting movement of the fore foot coupled with a lateral turning of the head. The peeled stalk was then placed crosswise in a corner of the mouth, at the level of the large cheek teeth, where it was bitten off and chewed up. The giant pandas in the Chicago Zoological Park manipulated green corn stalks, celery stalks, and carrots in a similar manner. The animals invari- ably sat down, or stood on their hind legs with one fore leg braced against the bars of the cage, when eating such food. They often sat with a piece of corn stalk or a carrot in each fore paw. Items were carried to the mouth in the fore paw, inserted transversely between the large cheek teeth, and bitten off. Chewing was a succession of ver- tical chopping movements. Field observers (Weigold in Jacobi, 1923a; Shel- don, 1937) have emphasized the poorly chewed and undigested condition of pieces of bamboo in the droppings of the giant panda. The skill and precision with which objects are grasped and manipulated by the fore feet is aston- ishing. I have observed animals in the Chicago Zoological Park pick up small items like single straws and handle them with the greatest pre- cision. Small disks of candy less than an inch in diameter were handled deftly and placed in the mouth. Objects are grasped between the radial pad and the palmar pad and are held in the shal- low furrow that separates these two pads. The actions of the fore paw suggest a human hand grasping through a thumbless mitten but are less clumsy than this comparison would indicate. Bears and raccoons, of course, can grasp objects with their fore paws. In this action the digits, aligned side by side, are closed over the object, which is thus held between the digital pads and the transverse palmar pad. This is a quite differ- ent mechanism from the grasp of the giant panda. The lesser panda (Ailurus) grasps objects almost as skillfully as the giant panda, and apparently in a similar way (fig. 3). Diets of Other Carnivores It is remarkable that the food habits of none of the bears have ever been adequately studied. Cottam, Nelson, and Clarke (1939) analyzed the contents of 14 stomachs of black bears (Ursus americanus) killed in early winter, and found that fruits and berries, mast, and foliage accounted for 93 per cent of the bulk and vertebrates for 4 per cent. Brehm (1915, Tierleben, Saugetiere, 3, p. 394) states that "more than the rest of the carnivores, the bears appear to be omnivorous in the fullest sense of the word, to be able to nourish themselves for a long time from the plant king- dom alone." Seton (Lives of Game Animals, 2, (1), 1929) emphasizes the omnivorous nature of the diet of each of the species of North Amer- ican bears. No quantitative study of the diet of Bassariscus has been made. Grinnell, Dixon, and Linsdale (Fur-bearing Mammals of California, 1, p. 179) state that "mice and other small rodents consti- tute the largest part of the food eaten by the ring- tailed cat. Small birds and berries are the other two most important items found in the stomachs examined. . . . Their jaws and teeth were so strong that they could chew up the leg bones of chicken without any trouble." The seasonal or annual diets of several other American arctoid carnivores have been determined quantitatively through large-scale analysis of stom- ach contents and scats. These, of course, provide the only reliable data on the diet, as opposed to what may be eaten under exceptional circum- stances, of any animal that is not positively re- stricted to a single food item. The diet of Procyon is more than 50 per cent (by bulk) vegetable (fruits, berries, nuts, and grains). Among the Canidae, the fall and winter diet of the red fox (Vulpes) is about 20 per cent herbivorous (fruits, grains, grasses), the winter diet of the gray fox (Urocyon) about 20 per cent herbivorous, and the annual diet of the coyote (Canis latrans) only 2 per cent her- bivorous. Many mustelids (Mustela vison, Taxi- dea, Lutra) are exclusively carnivorous or nearly so, but the skunks {Mephitis, Spilogale) may in- clude up to 50 per cent of plant material in their diets. From these data it is evident that the closest living relatives of the giant panda (the Ursidae) are, next to Ailuropoda itself, the most herbiv- orous of living carnivores.' If the diet of Procyon is typical, the Procyonidae are likewise heavily herbivorous, though less so than the bears. The dogs and foxes are true carnivores, including only relatively small amounts of plant material in their diets. Thus Ailuropoda is a member of a group of carnivores (the procyonid-bear branch) that is already heavily herbivorous, and it is most closely related to the most herbivorous element of this group. The exclusively herbivorous diet of the ' Unfortunately, no information, beyond vague general statements, is available on the diet of the lesser panda (Ailurus). Sowerby (1936a) says it feeds largely on bamboo leaves, and specimens in the Lincoln Park Zoo in Chicago ate green bamboo ravenously. 24 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Fig. 4. Postures of Ailuropoda: standing (Mei Lan, Chicago Zoological Park) and climbing ("Happy," Leipzig Zoo). giant panda is merely an extension, via an inter- mediate stage (the Ursidae), of a non-carnivorous dietary trend already present in the group from which this species was derived. POSTURE The postures of Ailuropoda are similar to, but by no means identical with, the corresponding pos- tures of Ursus. The normal standing posture is similar to that of bears. Both fore and hind feet are fully planti- grade but are toed in more sharply than in Ursus. The prominent shoulder hump of bears is much less conspicuous in Ailuropoda, and the hind quar- ters are somewhat higher. As in bears, there is relatively little angulation at elbow and knee. The head is carried low, and the tail is clamped tightly against the body. The panda has a stocky appear- ance, less dog-like than that of bears. The animal often sits on the hind quartei's with the fore feet free of the ground. This posture is almost invariably assumed during eating, since it frees the fore feet for manipulating food (fig. 2). The panda does not normally sit erect, as bears often do, with the weight resting on the ischial surfaces. Instead, the back is curved like the let- ter C, and the weight appears to rest on the pos- terodorsal surface of the pelvis. In this posture the hind legs are thrust forward, their lateral sur- faces resting on the ground, with the knees slightly bent and the soles of the hind feet turned inward. Bears sometimes sit with their hind legs similarly extended, although more frequently the legs are drawn up in dog fashion. Ailuropoda often rests, half sitting and half re- clining, in the crotch of a tree. The back is then arched sharply, the weight resting on the lower part of the back rather than on the ischia. Like bears, Ailuropoda readily stands erect on its hind legs (fig. 4). This posture is assumed both in the open without any support for the fore feet and, more frequently, with the fore feet resting against the bars of the cage. The hind feet are nearly fully plantigrade, the femur and tibia in a straight vertical line. The zoo animals show no DAVIS: THE GIANT PANDA 25 • • 17 Ursus AUuropoda Fig. 5. The eight phases of the slow diagonal walk, with its footfall formula, of AUuropoda and Ursus americanus. Tracings from motion picture film taken at 16 f.p.s. Numerals are frame numbers in the sequences. more tendency to stand erect than bears do. I have never observed a panda walking in the erect position. "Bears are able to stand erect on their hind legs, and to walk a short distance in an un- steady but not particularly awkward movement." (Brehm.) LOCOMOTION The normal gait of the giant panda is a "fast diagonal walk" (figs. 5, 6) in A. B. Howell's termi- nology. Howell (1944) states that this gait is reg- ularly employed by nearly all mammals. It is used by bears and raccoons. When moving more rapidly the panda breaks into a clumsy trot. Whether it is capable of galloping at still higher speeds is not known. The walk of AUuropoda is bear-like, but less smooth and graceful. The head is carried well below the shoulder line, and the tail is closely ap- pressed against the body. The stride is consider- ably longer than in bears, and as a result the gait is more rolling, with much more lateral rotation of the shoulders and hips than in Ursus. This gives a pronounced waddling character to the locomo- tion. The heavy head is swayed from side to side. The sole of the fore foot is fully apposed to the ground, but the heel of the hind foot does not touch the ground. Indeed, the panda appears to be incapable of flexing the ankle joint enough to permit plantigrady (p. 144). In this respect AUu- ropoda contrasts with Ursus, in which the sole is naked to the heel and the foot is fully plantigrade. During the recovery phase of the stride the fore feet are directed inward much more than in Ursus, and this "pigeon-toed" position of the foot is main- tained during the support phase. During the re- covery phase the hind feet are rotated medially so that the soles are directed medially. During the support phase, when the hind foot is resting on the ground, the toes point inward. At the end of the support phase the feet roll off the ground with the lateral toes receiving the major thrust. 26 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Fig. 6. Two types of walking loco- motion in the giant panda Mei Mei. The top figure is the fast diagonal walk, cor- responding approximately to no. 19 in figure 5. The bottom figure is a slow walk. In captivity the giant panda is a persistent climber when young (fig. 4). The movements are often astonishingly clumsy but successful. In climbing vertical or near vertical tree trunks the movements are bear-like. The animal embraces the tree, with the soles of all four feet pressed against the bark, and progresses by a series of "caterpillar" movements. The animal takes ad- vantage of branches or other projections to hoist itself up. It descends tail first, unless the slope is gentle enough to allow it to walk down head first. The claws appear to be of less importance in climbing than the friction of the soles against the bark, although the claws are used, especially if the animal slips unexpectedly. In this type of climb- ing, called "bracing" or "prop" climbing {Stemm- klettern) by Boker (1935), the portion of the body not supported by the hind legs is suspended from the fore legs. DISPOSITION Young individuals are active and playful, and thousands of zoo visitors have been entertained by their clownish antics. As they grow older they become much less active. Some individuals, at least, become siu'ly and dangerous in captivity. The giant panda "Mei-Lan," while in captivity in the Chicago Zoological Park, mauled one of his keepers so severely that an arm had to be amputated. Sheldon (1937), who hunted Atluropoda, wrote: "My experience convinced me that the panda is an extremely stupid beast. On one occasion at a distance of 350 yards I obsei-ved two individuals on the edge of a bamboo jungle. Driven out by four dogs and warned by several high-powered bullets whistling about them, neither animal even broke into a run. The gait was a determined and leisurely walk. Again, Dean Sage and I observed another panda pursued by four dogs. In this in- stance he walked to within eight feet of Dean and was stopped only by bullets. He gave absolutely no evidence that he saw either of us, and seemed completely to disregard both the shots and the loud talking and shouts of a few minutes previous." SUMMARY The giant panda is confined to the moist bam- boo zone on high mountain slopes, where the leop- DAVIS: THE GIANT PANDA 27 ard and the red wolf are its only potential natural enemies. Its natural diet consists exclusively of bamboo, with which it is always surrounded. Se- lection pressure for locomotor efficiency is abso- lutely minimal. Bamboo stalks are consumed in enormous quantities, but are poorly chewed and poorly digested. The fore feet are constantly used to manipulate the food. Objects grasped in the fore paws are held between the radial pad and the palmar pad. This grasping mechanism differs from that used by bears and raccoons but is sim- ilar to that of the lesser panda {Ailurus). Ailuropoda is a member of a group (the bear- raccoon line) of carnivores whose diet is more than 50 per cent herbivorous. Its closest living rela- tives (the bears) appear to be more than 90 per cent herbivorous. Posture and locomotion are similar to those of bears. Locomotion is less efficient. Ailuropoda climbs clumsily but persistently when young. EXTERNAL CHARACTERS The general habitus of Ailuropoda is ursine. The head and fore quarters are heavy and power- ful, the hind quarters relatively weak. The build is much stockier than that of bears of comparable size. I. DESCRIPTION The pelage is thick and woolly, as befits an ani- mal frequenting high altitudes. The characteristic parti-colored pattern is shown in figui-e 9. This pattern is unique among carnivores, although it is approached by the ratels {Mellivora}, and by the lesser panda (AiluriLs) except that the areas that are white in Ailuropoda are for the most part red- dish-brown in Ailurus. The coloration of Ailuro- poda is certainly a "constitutional" pattern rather than a "biological" pattern conditioned by nat- ural selection. The most unusual feature of the hair arrange- ment is found in the nasal region. The short hair on the top of the rostrum, from a point just in front of the eyes down to the muzzle (a distance of about 55 mm.), is directed straight forward. Two whorls are formed, 35 mm. apart, in front and mesad of the eyes, from which the hair radiates. Attention was first drawn to this character, which is unique among arctoid carnivores, by Kidd (1904). Kidd's later suggestion (1920), that this reversal of hair stream resulted from rubbing the hair toward the muzzle in cleaning it, cannot be taken seriously. It is noteworthy that a similar reversal occurs in other short-nosed carnivores (e.g.,Fefe). The facial vibrissae (fig. 7) are rather feebly de- veloped, although not so poorly as Pocock (1929) concluded from an examination of prepared skins. The superciliary tuft is represented by about three moderately long hairs over the eye. There is a relatively heavy growth of mystacial bristles along the upper lip, extending back almost to the angle of the mouth. On the lower lip they extend as far as the angle of the mouth. These bristles are much worn and broken on the specimen at hand, so that their length cannot be determined. They cer- tainly do not reach any great length, however. Inter-ramal and genal tufts are absent. The rhinarium, as pointed out by Pocock, is hairy above, with a well-haired infranarial area on either side of the midline below. The naked area roughly resembles an inverted triangle and is con- tinued ventrally into a short, grooved philtrum. There is also a V-shaped notch between the nos- trils dorsally. The transverse groove below the nostrils referred to by Pocock is not evident on the fresh animal. The nostrils are transverse. The external ear is erect, relatively larger than in bears, arising from a curiously constricted base. The margin is rounded, as in bears. The ear is well haired internally far down into the meatus. There is no bursa. The height of the pinna in Su Lin is about 85 mm., its breadth about 80 mm. The eai-s are set higher on the head and closer to- gether than in bears — a consequence of the enor- mously developed masticatory musculature. The fore foot (fig. 8) is short and powerful. The digits are enclosed in the common skin of the foot up to the base of the digital pads. Examination of the fresh animal corrects several errors made by Pocock. All the pads are thick and cornified. The digital pads are elUptical in outline, those of the second, third, and fourth toes approximately equal in size. That of the fifth toe is slightly smaller, and the pad of the poUex is the smallest of all and is joined to the palmar pad by a narrow isthmus of naked skin. The palmar pad extends as a nar- row strip across the entire foot. There is no evi- dence of its breaking up into interdigital pads. The outer end of the pad is expanded slightly, and its inner end curves proximally to join the prominent radial lobe, from which it is separated by a transverse furrow. The radial lobe is smaller than the outer carpal lobe. This lobe is wanting in bears. It is ellip- tical in outline, the long axis running anteroposte- riorly, and is hemispherical in cross section. It is associated with the prominent radial sesamoid bone, which hes directly beneath it; Pocock was not sure that it represents the missing inner carpal lobe. Objects held in the hand lie in the furrow between the radial lobe and the inner end of the palmar pad and are grasped between these two pads. The outer carpal lobe is large and roughly cir- cular in outline and is situated somewhat farther 28 Fig. 7. Side view of head of Ailuropoda, showing pattern of vibrissae and hair-slope. 29 Fig. 8. Ventral surfaces of left fore and hind feet of Ailuropoda melanoleuca (A, B) and Ursus americanus (C, D). Ursus after Pocock reversed. 30 DAVIS: THE GIANT PANDA 31 proximally than the radial lobe, lying about a third of its own width behind the palmar pad, much closer than in Ursus. The remainder of the palmar surface is densely covered with long hair. The hind foot (fig. 8) is slightly narrower than the fore foot and is remarkable for the limited ex- tent of the cornified hairless areas. The absence of the posterior lobe of the plantar pad is associ- ated with the inability of Ailuropoda to flex the foot beyond 45° from the vertical (fig. 80). The digits are enclosed in the common skin of the foot nearly to the bases of the digital pads. The digital pads are elliptical in outline, and all are approxi- mately the same size. The pad of the hallux is joined to the plantar pad by a narrow isthmus of naked skin similar to that on the pollex. The plantar pad is a narrow transverse cushion, feebly convex anteriorly and very faintly divided into five lobes (not four as Pocock stated). The pad lies beneath the metatarso-phalangeal articulation. It is somewhat wider at the outer end than at the inner, and the lobe under the hallux is more clearly indicated than the others are. Metatarsal pads are absent; the remainder of the sole is densely covered with long woolly hair. The claws on all the digits are strongly com- pressed and taper from a wide base to a sharp tip. The upper edge of the claw describes almost a per- fect quadrant of a circle; the lower edge is sinuous. The tail is relatively small but longer and con- siderably heavier than that of any of the bears. It measures 115 mm. in length in Su Lin (the cau- dal vertebrae measure 203 mm. in the skeleton of an adult) and tapers abruptly from a heavy base. The base of the tail is flattened dorsoventrally; its width is about 35 mm. while its depth is only about 25 mm. (see p. 83). The entire organ is densely clothed in long, coarse hairs. There are two pairs of nipples, one pair pectoral and the other abdominal. The pectoral pair lies over the seventh rib, the abdominal pair 200 mm. behind the posterior end of the sternum. The bears have three pairs of mammae. The external structures in the perineal region are described on page 221. II. MEASUREMENTS No flesh measurements of an adult giant panda are available. The following measurements were made on the mounted skeleton of the adult male killed by the Roosevelt brothers. Flesh measure- ments of an adult female black bear, quoted from Seton (1929, Lives of Game Animals, 2 (1), p. 119) are given for comparison. Ursus Ailuropoda americanus mm. inches mm. inches Snout to tail tip 1422 56 1613 63.5 (along curve) Tail 203 8.5 127 5 Height at shoulder. .. . 635 25 648 25.5 Approximate mean pounds pounds weight of adult 275 250 The female "Happy" (weight 223 pounds), meas- ured by Schneider (1939), had a shoulder height of about 660 mm. No actual weight figures for adult giant pandas exist. Schafer estimated that an adult male would weigh 275 pounds; Ailuropoda is fully grown at 4-5 years. The adult male Mei Mei weighed 205 pounds at death but weighed 296 pounds some months earlier. The weight of the male Mei Lan was estimated by zoo officials at 300 pounds when he was six yeai's old. Skeletal measurements (Ta- ble 6, p. 45) show that Mei Lan was much the largest panda on record. A male at the St. Louis Zoo weighed about 280 pounds at eight years of age, and a female 240 pounds at five years. Thus it appears that the adult weight of the giant panda is 250-300 pounds, which is close to the average for the American black bear. The giant panda Su Lin weighed 132 pounds at death. The snout- vent length of this individual was 1195 mm. III. GROWTH Weight increments for about the first 18 months of life are available for three individuals. These figures are, of course, for captive animals and do not include the first month or two after birth. Figures for "Pandah" and "Pandee" were kindly supplied by Dr. Leonard J. Goss of the New York Zoological Society. Weight figures are shown in the accompanying graph (fig. 9). The average monthly gain was 9 pounds. IV. PROPORTIONS Measurements of the linear dimensions of ana- tomical structures serve two different purposes. The simpler of these is as a means of expressing relative sizes of homologous parts in two or more organisms. Thus, if femur length is 75 mm. in A and 60 mm. in B, we say that the femur is longer in A, or is 15 mm. longer, or we may express the difference as a percentage and say that femur length in B is 80 per cent of femur length in A. Such simple manipulations are much used in tax- onomy and comparative anatomy. They rarely present serious difficulties as long as the organisms being compared are fairly closely related. On the other hand, attempts to compare pro- portions between two or more species or genera 32 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 often present serious difficulties. If A and B rep- they are not). This difficulty has plagued corn- resent different species, the fact that the femur of parative anatomists from the beginning and has A is longer than that of B may reflect the fact never been satisfactorily resolved. 7 8 9 MONTHS Fig. 9. Growth curves of Ailuropoda. that A is a larger organism than B, or that the femur is relatively longer in A or is relatively shorter in B, or a combination of all of these fac- tors. The difficulty in determining what is in- volved arises from the fact that there is no com- mon standard to which the variable (in this case femur length) can be related; for practical pur- poses all measurements on an organism must be treated as independent variables (although in fact Many structures in mammals function as lever systems. Interpretation of the mechanical advan- tage of one lever system over another does not depend on knowing how the differences in propor- tions were achieved, but a true understanding of the morphology of the organism obviously does. Index figures, obtained by dividing one dimension (e.g., tibia length) by another larger dimension from the same individual (e.g., femur length) and multi- DAVIS: THE GIANT PANDA 33 Fig. 10. Body outlines of representative arctoid carnivores to show posture and proportions. All drawn from photo- graphs of living animals (not to scale). Top: Wolverine {Gulo luscus), a generalized mustelid; cacomistl (Bassariscus astutus), a generalized procyonid. Middle: Raccoon (Procyon lolor) and les.ser panda (Ailurus fulgens). Bottom: Black bear {Ursus americanus) and giant panda (Ailuropoia melanoleuca) . plying by a constant (commonly 100), ai-e widely used because they are independent of the absolute size of the original figures and therefore directly comparable between individuals of the most di.s- parate sizes. Uncritical comparisons of such index figures may, however, lead to grossly ei-roneous conclusions. In the present study the femoro- length tibia tibial index X 100 for a group of length femur badgers happened to be identical with the corre- sponding index for a series of giant pandas, 76 in both cases. Analysis of the figures for femur and tibia length, using a third dimension (length of 3 vertebrae) as a common standard, revealed that the tibia is abnormally short and the femur about normal in the badgers, whereas in the panda the reverse is true: the femur is abnormally long and the tibia about normal. These relationships may be of no importance in comparing the limbs as lever systems, but they are of the utmost impor- tance in interpreting the morphology, and partic- ularly the phylogeny, of the limbs. They could not have been detected from the dimensions of femur and tibia alone, but required the use of a third dimension as a common standard. Body Proportions Comparative proportions of the body in a series of animals may be expressed by equating spine length to 100 and expressing the dimensions of other body parts as percentages of spine length (Hildebrand, 1952). These proportions are shown pictorially (fig. 10) and graphically (fig. 11) for a series of carnivores. 34 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 23.5 76.5 Gulo luscus Potos flovus 21.5 27 73 \ o CO CO 00 00 5 ^ \ § fO to o 00 00 CD Ursus arclos Ailuropoda melanoleuca Fig. 11. Body proportions in representative carnivores (based on one specimen of each). In each case pre-sacral vertebral length was equated to 100, and lengths of other parts were indicated as percentages of vertebral length. Limb length is the "functional limb length" of Howell (lengths of propodium + epipodium + metapodium). The wolverine {Gulo) represents a generalized terrestrial carnivore, in which length of hind limbs exceeds that of fore limbs by about 10 per cent, the epipodial segments (radius and tibia) are slightly shorter than the propodials (humerus and femur), and the metapodials (metacarpals and meta- tarsals) are long. In an arboreal carnivore (Potos) the hind limbs are elongated and the metapodials slightly shortened. In canids, which are typically cursorial runners, the legs are relatively long, espe- cially the epipodial and metapodial segments. These all represent rather obvious adaptations for locomotor efficiency. Adaptation is less obvi- ous in certain other carnivores. The bears, which are mediportal ambulatory walkers (p. 38), have legs relatively as long as the cursorial canids and the proportion between length of front and hind limb is about normal for carnivores. The bears and the giant panda are remarkable among carni- vores in having a long femur associated with a short tibia, without corresponding reduction in radius length; this condition is characteristic of heavy graviportal mammals (A. B. Howell, 1944). In Ailuropoda the spine has been shortened by elimination of lumbar vertebrae, a condition other- wise unknown among carnivores. The trunk in Ailuropoda is relatively shorter than in any other known carnivore; the index "length thoracics 10- 12/length thoracolumbar vertebrae X 100" is 18 and 22 for two pandas, whereas it is 14 (13-15) for all other carnivores examined except a specimen of Mellivora, for which it is 16. This exaggerates apparent leg length, but the legs actually are rela- tively long (Table 2). Length of fore and hind legs is subequal in Ailuropoda; this condition is otherwise encountered among carnivores only in the hyenas, although the proportions of the limb segments in hyenas are quite different from those of Ailuropoda. Limb Proportions In studies on small rodents, body length (meas- ured on the freshly killed animal) is often used as the independent variable. This is impractical in work on skeletons of large mammals, for which measurements of body length are rarely recorded. Hildebrand (1952) used length of the vertebral column in his work on body proportions of the Canidae. Length of vertebrae probably varies as little as any convenient linear dimension, but for material as heterogeneous as the whole Order Car- nivora it is desirable to eliminate the lumbar re- gion, which, like the limbs, is intimately involved in the mechanics of locomotion and would there- DAVIS: THE GIANT PANDA 35 Table 1.— LIMB SEGMENT RATIOS IN CARNIVORES No. Canis lupus 4 Canis lalrans 3 Chrysocyon brachyurus 2 Bassariscus astutus 4 Bassaricyon 3 Nasua 3 Procyon lolor 4 Potos flavus 3 Ailurus fulgens 3 Ursus americanus 2 Ursus arctos 2 Ailuropoda 7 Gulo luscus 3 Martes pennanti 2 Taxidea taxus 3 Mellivora 1 Lutra canadensis 3 Enhydra 2 Viverra tangalunga 5 Paradoxurus 4 Herpestes 1 Felis onca 2 Felis leo 4 Felis tigris 1 Total 71 humero- Femoro- Femoro- Tibio- Inter- radial humeral tibial radial membral index index index index index 100.6 89.8 98.4 90.7 98.5 104.4 87.8 99.6 92.0 89.9 108.1 91.0 107.8 91.4 91.2 79.0 89.8 97.2 72.8 81.4 74.8 88.5 101.7 65.4 76.9 85.1 82.4 91.0 76.9 79.8 100.9 85.3 100.8 83.6 83.5 80.6 89.2 94.9 74.9 84.2 74.7 94.9 94.2 75.3 85.5 81.1 86.4 72.2 96.9 90.8 81.5 84.2 68.5 100.2 90.6 77.1 98.4 76.1 98.5 99.3 78.9 94.3 90.2 82.5 88.8 76.0 90.4 99.0 69.4 79.9 76.2 98.2 76.1 98.1 98.1 79.9 90.2 75.5 95.5 92.5 71.5 98.9 111.4 63.5 80.2 75.7 95.9 112.4 64.6 79.7 90.1 81.3 96.2 76.1 78.8 77.0 91.7 90.9 77.5 84.9 76.5 82.9 90.3 70.3 76.9 77.6 87.0 80.9 85.4 86.0 90.3 86.5 84.6 92.3 89.2 81.3 83.6 82.2 82.7 83.2 fore be expected to bias the results. A group of three thoracic vertebrae is convenient to measure and yields a linear dimension of convenient size. The combined length of thoracics 10-12 has there- fore been used as the independent variable in the present study. An obvious disadvantage of using this measure as the independent variate is that it is the least accurate of all the measures in the set, and errors of measurement in the independent variate will bias the results, even though the errors are random. Furthermore, length of centrum is itself a vari- able; simple inspection shows that vertebrae are relatively longer in Mustela than in Ursus, for ex- ample. Therefore, index figures derived from this common standard have no absolute value for pur- poses of comparison. They are only approxima- tions, their reliability depending upon the range of variation in relative vertebral length within the sample. Reliability is certainly great enough to demonstrate gross deviations from the norm. A further problem in interpreting these data is the selection of a norm against which the index figure can be evaluated. Femur length cannot be judged "short" or "long" unless it is shorter or longer than some standard femur length for the Carnivora. Probably the best that can be done is to use the index figure for the least specialized representative of the Carnivora as a norm. In Table 2 the figures for the wolverine (Gulo), whose locomotor habits are as generalized as those of any living carnivore, are used as a norm, the figures being rounded off to the nearest multiple of 5. From the table it appears that arm length is the most conservative among the four limb segments and foi-earm length the most variable. These indexes correlate quite well with what is known of the locomotor habits of the animals. There are puzzling non-conformities (e.g., long proximal segments in Ailurus, short arm in Vi- verra, long fore arm associated with long thigh in Felis leo, etc.) that cannot be explained on the basis of existing knowledge. Disregarding these exceptions, limb proportions appear to correlate with locomotor types as follows in the Carnivora: Ambulatory walking norm Running all segments long, especially forearm Arboreal climbing Type A hind legs long Type B forearm short, other segments norm Digging di.stal segments short Swimming all segments very short, especially forearm The bears and the giant panda, in which a short tibia is associated with length in the other three seg- ments, do not fit any of these categories, and this combination is difficult to justify on a mechanical basis. Elongated limbs are generally associated with running, where a long stride is advantageous. The limbs are also long in graviportal animals (e.g., elephants, titanotheres), although the me- chanical factors involved are unknown. The bears Table 2.— LIMB PROPORTIONS IN CARNIVORES' N Canis lupus 4 Canis latrans 3 Chrysocyon 1 Bassariscus 4 Nasua 1 Procyon 3 Polos 3 Ailurus 2 Ursus 4 Ailuropoda 2 Gulo 2 Maries pennarUi 1 Maries flavigularis 1 Taxidea and Mellivora 4 Lutra canadensis 3 Enhydra 2 Viverra langalunga 5 Paradoiurus 3 Herpesles 1 Croeula 1 Hyaena 1 Felis onca 2 Felis leo 4 Felis tigris 1 ' V=Iength of thoracics 10 ifcll to 20. Extremely long or short= ±21 or more V L. humerus V L. radius V L. femur V L. tibia (norm =40) (norm=50) (norm = 40) (norm = 40) 35 35 31 32 36 long 34 verj- long 32 long 32 long all long; forearm very lot 27 very long 25 extremely long 25 ven,- long •■>•> ven.- long all very long; forearm extremely long 43 norm 54 si. short 38 norm 40 norm forearm slightly short 37 norm 43 long 31 long 24 ver>' long hind legs long; all distal segments long to very long 34 long 34 ver>' long 29 ven,- long 28 verj- long all very long, except humerus long 39 norm 48 norm 34 long 36 long hind legs long 36 slightly long 48 norm 34 long 37 norm proximal segments long 32 long 39 very long 27 very long 39 norm all long-very long, except tibia norm 34 long 44 long 34 long 4S short all long, except tibia short 40 50 38 42 all norm norm norm norm norm 41 norm 73 extremely short 37 norm 37 norm forearm extremely short 45 short 60 short 40 norm 41 norm forelegs short 42 norm 55 short 40 norm 53 very short distal segments short; tibia very short 68 very short 95 extremely short 67 extremely short 60 very short all very short, forearm extremely so 77 extremely short 102 extremely short ( 74 extremely short 66 extremely short all extremely short, especially forearm 47 51 37 39 arm short short norm norm norm 42 norm 54 slightly short 39 norm 42 norm forearm slightly short 50 short 65 very short 41 norm 46 short all short-ver>' short, except femur 33 long 32 very long 27 very long 36 long all long; forearm and thigh very long 33 long 30 verj- long 30 long 34 long all long; forearm very long 40 norm 51 norm 35 long 43 norm long thigh 38 norm 42 long 33 long 39 norm long forearm and thigh 43 norm 52 norm 36 long 43 norm long thigh 2. Norm = ±3 from norm. Lor ig=-4 to -10. Short= +4 to +10. Verj- long or short 36 SUMMARY OF LIMB SEGMENT RATIOS IN CARNIVORES Ambulatory walking . Running Half-bound (cats) . . . Climbing Digging Swimming Mediportal types Ursus Aibiropoda Humeroradial Femorohumeral Femorotibial Tibioradial Intermembral 80 radius short 95 subequal 90 + tibia short 75 + radius shorter 90 + hind legs long 100 + equal 90- femur longer 98 + equal 90 + radius short 90 + hind legs long 80-90 radius short 85 + femur longest 85- tibia shorter 85 + radius shorter 85 + hind legs longer 80- radius shorter 90- femur longer 95 + subequal 75 + radius shorter 85- hind legs longer 80- radius shorter 90 + femur long 75+ tibia shortest 95 + subequal 92 + hind legs long 75 radius shortest 95 + subequal 110 + tibia longest 65- radius shortest 80 hind legs longest 80 + radius shorter 85 femur much longer 70 tibia very much shorter 96 + subequal 90 hind legs long 77 radius much shorter 98 equal 76 tibia much shorter 99 equal 99 equal 400T ? 300 0 Ursus omerlcanus D " arctos A " gyos • Ailuropoda 200 150 Ailuropoda Y= 37.3 + 0.63 X Ursus Y= -6.5 -1- 0.83 X 200 300 Humerus Length 400 500 Fig. 12. Scatter diagram, with fitted regression lines, showing length of radius and length of humerus in panda and bears. (Dashed line=slope of 1.) 37 38 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 / 400- 300-- 200- 150 0 Ursus omerlcanus o " orctos a " gyas • Ailuropoda / / 200 300 Femur Length 400 500 Fig. 13. Scatter diagram, with fitted regression lines, showing length of tibia and length of femur in panda and bears. (Dashed line=slope of 1.) and the panda are relatively slow-moving ambu- latory walkers and lack the elongation of the meta- podials that characterizes runners. Shortening of the distal segments characterizes digging animals, in which the mechanical advantage of increasing effective power at the distal ends of the limbs is obvious. Gregory (in Osborn, 1929) noted that among ungulates the tibia shortens with gi-avi- portal specialization, whereas relative radius length either remains stationary or shortens to a less de- gree than tibia length. This is exactly the situa- tion in the bears and the giant panda, whose limb proportions are those of mediportal or graviportal animals. Intramembral Indexes Ratios of limb segments with respect to each other reflect the same pattern as ratios derived from an independent variable. They have the ad- vantage over the preceding ratios of greater math- ematical reliability and of widespread usage (see A. B. Howell, 1944). Limb segment ratios of rep- resentative carnivores are given in Table 2. These figures are associated with locomotor types as shown in the following summary. Several forms (e.g., Procyon, Ailurus, Viverra, Herpestes) do not fit well into any of the categories, and again it must be assumed that unknown factors are in- volved in determining the limb proportions of such forms. Ratios for the bears agree with those of medi- portal or graviportal ungulates. Furthermore, this agreement is associated with other mediportal adaptations, such as flaring ilia and relatively slight angulation of the limbs at elbow and knee. The peculiar i-atios in Ailuropoda do not occur in any other known mammal, and they often differ from the corresponding ratios in Ursus. They are most closely approached by those of the burrowing mustelids. Functional lengths of humerus and femur are equal in a very few scattered forms {Tamandua, Icticyon, Dolichotis; A. B. Howell, 1944). Equality in length of radius and tibia is more common but follows no pattern. Equality in the intermembral index occurs elsewhere among terrestrial mammals only in a few aberrant forms (giraffe, hyenas, the extinct forest horse Hippidi- um; A. B. Howell, 1944). I conclude that limb proportions in Ailuropoda are attributable to fac- tors other than mechanical requirements — that DAVIS: THE GIANT PANDA 39 400 -- 5 300 -- 200 0 Ursus americanus D " arctos A >■ a gyas • Ailuropoda / / Ailuropoda Y= -89.1 -i- 1.21 X Ursus Y= -58.6+ 1.I6X 150 200 300 400 500 Pelvis Length Pig. 14. Scatter diagram, with fitted regression lines, showing breadth and length of pelvis in panda and bears. (Dashed line=slope of 1.) selection for mechanical efficiency has been over- ridden by some other factor or factors. Allometry Examination of linear measurements of the limb bones of Ailuropoda (Table 6, p. 45) shows that proportions vary with the absolute size of the bones. When pairs of measurements for all indi- viduals are plotted on scatter diagrams, clustering of observations along a line that deviates from a 45° angle is evident for nearly all limb proportions. This indicates that limb proportions conform to the well-known allometric equation y = a + bx, where z and y are the two measurements being compared, and a and b are constants. Regression lines were fitted to the data by the method of least squares (Simpson and Roe, 1939). For the limb bones of Ailuropoda the plotted points are somewhat scattered (figs. 12, 13), indi- cating considerable individual variation in pro- portions. The slopes of the regression lines diverge from unity, indicating an allometric relationship between proximal and distal segments of the legs; radius and tibia become increasingly short relative to the proximal segments as total organism size increases. Conditions in Ursus are similar, although allom- etry is considerably less for the radius than in Ailuropoda. The plotted observations for all pro- portions cluster much more closely ai'ound a straight line, indicating relatively little individual variation. The deviations of the regression lines from unity are not statistically significant for either Ailuro- poda or Ursus. The close clustering of the values, especially for Ursus, suggests that they would be significant in a larger sample. Similar analyses of data on limb proportions in other cai'nivores are available only for the domes- tic dog. Lumer (1940) found a close correlation, but only a very slight deviation from unity in the slopes of regression lines, in both humeroradial (6=1.098) and femorotibial (6=1.090) proportions in an analysis of data from a wide variety of breeds of dogs. 40 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 The limb girdles in the panda and bears are less consistent than the limb segments. In the scapula of the panda there is little correlation between height and breadth (r=0.45, N=9). In Ursus, on the contrary, there is a very close correlation be- tween height and breadth of scapula (r=0.98, N=9), but only a slight indication of allometry (6=0.94). The pelvis shows a high correlation in total length/breadth across ilia in both Ailuropoda and Ursus. There is also a strong allometric rela- tionship (6=0.75 in Ailuropoda, 6=0.57 in Ursus), the iliac breadth becoming increasingly great as size of pelvis increases (fig. 14). The "law of allometry' has been tested by many workers in a wide variety of cases, and found to be a valid empirical representation of ontogenetic growth relations. We may therefore postulate that the allometric relations demonstrable in Ailu- ropoda and Ursus reflect genetically determined processes that are as characteristic of the species or genus as are any morphological feature, repre- senting what Lumer has called "evolutionary al- lometry." The intensity of expression of such size-dependent relationships is a function of or- ganism size. Therefore the proportions at any particular phylogenetic stage (strictly, at any par- ticular organism size) may not be, and in extreme cases certainly are not, directly related to the re- quirements of the organism. If selection has fa- vored increased organism size, then proportions may become increasingly grotesque until a point is reached where the disadvantages of mechani- cally unfavorable proportions balance the advan- tages of further increase in organism size. V. CONCLUSIONS 1. The external characters of the giant panda are basically similar to those of Ursus. Differences from the bears are for the most part conditioned by more fundamental differences in underlying structures. 2. The absolute size of the giant panda is al- most identical with that of the American black bear. 3. Body proportions of the bears and the giant panda differ from those of all other living carni- vores. They resemble the proportions of medi- portal or graviportal animals, although the mass of the smaller bears and of the giant panda is less than that of mediportal ungulates. It is also less than that of the larger cats, which show no medi- portal specializations. 4. The trunk in the giant panda is relatively shorter than in any other known carnivore. 5. Limb proportions in the giant panda resem- ble those of bears, but differ in some important respects. In neither the panda nor the bears can they be explained on the basis of functional re- quirements. 6. Limb proportions in the panda and the bears show indications of allometry, the distal segments being relatively shorter in larger individuals. Pel- vic proportions are also allometric, but scapular proportions are not. 7. Body proportions in the pandas and bears are not the result of selection for mechanical effi- ciency. Rather they reflect pleiotropic correla- tions with other features that have been altered through natural selection. SKELETON Most of the literature on the mammalian skele- ton is purely descriptive, with no real considera- tion of the soft parts to which the bones are intimately related in form and function, of the functions of the bones themselves, or of the fac- tors responsible for observed differences between species. Comparisons are often unreal, for bones are compared as if they were inanimate geometri- cal forms rather than artificially segregated parts of living organisms. As a result there has been little attempt to evaluate differences in other than purely quantitative terms. Even the descriptions are often inadequate because the observer described only what he saw. The primary objectives have been to find "characters" on which a classification of mammals can be based, or to reconstruct the phylogenies of organisms or of structures. These are important but severely limited goals. The gross features of the skeleton are deter- mined by heredity, conditioned by events in the remote past; mammals have one bone in the thigh and two in the leg because they inherit this pattern from their remote ancestors — not because it is par- ticularly suited to the needs of mammals. Within the limits set by this inherited framework, the pri- mary function of the skeleton is support, and the form and architecture of bones reflect primarily the stresses and strains associated with this func- tion. Each bone is also subjected to an assort- ment of constantly varying localized stresses and strains resulting from the action of muscles and ligaments. Besides these mechanical factors, the skeleton also serves as a store for calcium salts. Consequently the architecture of a bone is far more complex than is generally assumed, and at- tempts to analyze bones from the engineering standpoint have not been entirely successful (see Wyss, 1948). In the individual the basic features of the skel- eton, including accumulated adaptive features acquired during phylogeny, are determined genet- ically. We cannot go far beyond this obvious gen- eral statement, although Stockard (1941) and Klatt (1941-43) made a beginning at discovering the nature of this genetic control, and Sawin (1945, 1946) and his co-workers demonstrated gene con- trol of morphogenetic fields in the skeleton. Scott (1957) concluded that growth and differentiation of the skeleton depend on two distinct processes: (a) a length-regulating process controlled by con- version of cartilage into bone (interstitial growth), and (b) a robustness-regulating process that deter- mines the thickness of the limb bones, the size of the vertebrae, etc., and involves the activity of the subperiosteal cellular tissue (appositional growth). It is likewise obvious that the inherited features of the skeleton are modified, within limits, by the activities of the individual. This is seen, if proof is needed, in the vertebral column of Slijper's bi- pedal goat (Slijper, 1946), in the adaptations to pathological conditions described by Weidenreich (1926, 1940), and in the experiments of J. A. Howell (1917), Washburn (1947), Wolffson (1950), Moss (1958), and others. This non-hereditary' factor is of unknown, but probably considerable, importance in determining the morphology of the bones. Howell, for example, found that in the bones of the fore leg of the dog most or all growth in diameter (appositional growth) is dependent on extrinsic mechanical factors, whereas growth in length (interstitial growth) is largely independent of mechanical factors. Finally, it is reasonable to assume that the ca- pacity of the individual skeleton to respond adap- tively to specific functional demands is inherited, and that this capacity varies with the age of the individual. The description of the skeleton of the giant panda here presented is somewhat unorthodox. The customary detailed description of each bone has been largely omitted; the illustrations should supply such information. The relations between bones and muscles, blood vessels, and nerves has been emphasized; and mechanical factors, which seem to have been of more than usual importance in molding the morphology of the giant panda, have been treated to the best of my ability. I have aimed not merely to describe and compare, but so far as possible to interpret. ' The muscles and other soft parts that act on the bones, as well as the psychology that directs the basic activities of the animal, are presumably gene-controlled. Thus even this factor is hereditary, at second hand, so to speak. 41 42 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Fig. 15. Skeleton of Ailuropoda melanoleuca (CNHM no. 31128, adult male). L THE SKELETON AS A WHOLE The skeleton (fig. 15) resembles in general ap- pearance that of a bear of similar size. The massive skull and short vertebral column give a somewhat non-ursid aspect to the skeleton. As in Ursus, surface modeling on the limb bones is prominent. The mass of the skeleton is greater than that of a black bear of similar size. This is largely but not entirely due to the much heavier skull (Table 3). Table 3.— WEIGHT IN GRAMS OF DRY SKELETON Skull as percentage CNHM Sex Total Skull of total 36758 Ailuropoda — 5550 1581 29 31128 Ailuropoda cT 6055 1583 26 44725 Ursus americanus a' 5029 818 16 18864 Ursus americanus — 3690 694 19 47419 Ursus arctos. . . . — 11018 1923 18 65803 Ailurus fulgens . 9 269.6 67.5 25 49895 Procyon lolor . . . d' 384.5 67.8 18 54015 Canis lupus . . . . 9 2013 377.5 19 46078 Hyaena striata . . — 2083 465 22 18855 Crocuta crocuia. — 3947 864 22 For the giant panda and black bear these figures represent about 4 per cent of total body weight. Further analysis of weight figures shows (Ta- ble 4) that percentages of total postcranial skele- ton weight formed by the trunk, fore limbs, and hind limbs are very similar in giant panda and bears. These ratios vary considerably among the other carnivores. It is evident that, except for the skull, the rela- tive proportions of total skeleton weight formed by each of the major regions of the skeleton in Ailuropoda do not differ significantly from those of Ursus. This is not true of the skull, which is extraordinarily dense in the giant panda. The skull-postcranial ratio is quite constant at 16-19 per cent in other carnivores examined, except Ailu- rus and the hyenas, in which the masticatory appa- ratus is likewise exceptionally powerful. The weight of the bones of the fore limbs is rela- tively greater in Ailuropoda, Ursus arctos, and the hyenas than in the other carnivores (Table 4). Klatt and Oboussier (1951) found this likewise true of bulldogs compared with greyhounds, al- though the disproportion (bulldog 69 : 31, grey- hound 61 : 39, on fresh bones) was greater than in any of our material. Klatt and Oboussier found a comparable disproportion in total weight (i.e., including soft parts) of the limbs, and an even greater disproportion for the head. They con- cluded that the bulldog proportions result from a DAVIS: THE GIANT PANDA 43 Table 4.— WEIGHT RATIOS IN DRY POSTCRANIAL SKELETON Percentage of Total Postcrania! Fore limbs : Hind limbs Trunk (incl. pelvis) CNHM 36758 Ailuropoda 44 31128 Ailuropoda 46 44725 Ursus americanus 46 18864 Ursus americanus 42 47419 Ursus arclos 46 65803 Ailurus fulgens 47 49895 Procyon lotor 47 54015 Canis lupus 40 46078 Hyaena striata 43 18855 Crocuta crocuta 45 Skeleton Fore limbs Hind limbs 31 25 55 45 31 23 57 43 29 25 54 46 30 28 52 48 32 22 59 41 27 26 50 50 23 30 43 57 32 28 53 47 34 23 59 41 32 23 59 41 generalized regional effect, centered in the head but affecting the whole forequarters. Taylor (1935) has shown that the relative mass of the skeleton increases, whereas relative bone area decreases, with increasing body size in a series of mammals. He presented data for a series of forms ranging in size from the albino rat to the domestic cow. Surface areas of a humerus and a femur of an adult male giant panda and an adult male black bear were measured according to Tay- lor's method. Each bone was carefully covered with adhesive tape. The tape was then removed and weighed (the number of square centimeters per gram of tape having been determined). This method yielded highly consistent results on our material. The data are given in Table 5. In the giant panda the surface area of the hu- merus exceeds that of the femur by 6 per cent, whereas in the bear the reverse is true and the area of the femur is 6 per cent greater than that of the humerus. The surface area per gram of bone in the bear is exactly the same as the figure for man, as computed by Taylor; in the panda it is slightly less, because of the greater thickness of the walls. Taylor found that this ratio decreases with increasing organism size from 10.6 square centimeters per gram of bone in the rat to 0.69 in the domestic cow. The bear falls in about its proper place in his table; in the giant panda the long bones are heavier than would be expected in a mammal of its size. Thickness of the walls of long bones was meas- ured at the center of the shaft on X-ray photo- graphs. The walls are notably thicker in Ailuropoda than in a bear of comparable size; the walls of the humerus are about 30 per cent thicker, those of the femur about 60 per cent thicker (Table 5). The diameter of the medullary cavity is corre- spondingly decreased in the panda, showing that the abnormal cortical thickness results from a slowing down of resorption rather than from in- creased osteoblastic activity. The ulna is about 20 per cent thicker in Ailuropoda, and the tibia about 27 per cent thicker. Such increased cortical thickness cannot be attributed to mechanical re- quirements; it must instead reflect a pleiotropic effect or important differences in mineral metab- olism.. Indeed, it is well known that thickening the walls of a tube internally adds very little to the strength of the tube, whereas adding the same quantity of material to the outer surface does in- crease its strength significantly. Increase in mus- cle mass leads only to increase in the surface area of bone, not to an increase in thickness (Weiden- reich, 1922). Table 5.— SURFACE AREAS OF LIMB BONES Bone Bone Surface Area per weight length area gm. of bone gms. cm. cm. 2 cm.' Ailuropoda Humerus 268.9 27.8 368.1 Femur 251.3 28.2 344.6 Total 520.2 712.7 1.36 Ursus Humerus 214.6 26.3 344.6 Femur 239.8 31.6 364.9 Total 454.4 709.5 1.56 Thickness of wall at center of shaft mm. 6.5 8 5.5 5 44 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Fig. 16. Ground sections of compacta from middle of shaft of femur of Ailuropoda (left) and L'rsus gyas (X 100). These measurements also indicate the existence of regional differences in rate of bone deposition or resorption. The walls are significantly thicker in the hind leg than in corresponding bones of the fore leg, and the proximal segments are relatively thicker than the distal. The histological structure of the compacta of the long bones shows no differences between Ailuro- poda and Ursus (fig. 16). The bone is typically lamellar, with well-developed Haversian systems. Partly destroyed Haversian systems are numer- ous, and osteocytes are present in normal numbers. There is no evidence of retarded internal reorgani- zation of the bone. Mineral metabolism involves the skeleton. The normal diet of Ailuropoda contains quantities of certain minerals (especially silicon) that are ab- normal for a carnivore. It therefore seemed desir- able to determine the relative amounts of minerals in the bone. The following semi-quantitative spec- trochemical analysis of bone samples from wild- killed animals was made by the Spectrochemical Laboratory of the University of Chicago. Obvi- ously there is no significant difference between them. In summary, the skeleton of Ailuropoda is more dense throughout than that of Ursus, due to Ursus americanus Ailuropoda CaO --48^^ ^45^c MgO 0.9% 0.95% SiOj x> .6 X X' Sr --1200ppm ^1200 ppm Ba -- 300ppm --300ppm ' Working curve not available, but SiOj is less than 1 %, probably about 0.1-0.4%. Ailuropoda has less SiOa than Ursus by a factor of 0.6. greater thickness of the compacta. This is partic- ularly true of the skull. The increase in quantity of compacta cannot be attributed to mechanical requirements. Regional differences in relative thickness of compacta indicate that rate of bone deposition or resorption is not uniform throughout the skeleton. There appears to be a gradient in which relative thickness of compacta decreases distally. II. MEASUREMENTS Most of the bone measurements used in this study, except for those of the pelvis, are given in Table 6. These include all measurements used in calculating ratios and proportions for the most im- portant of the species used in this study. Lengths of the leg bones are not greatest over- all length, but the much more meaningful "func- tional length" recommended by Howell. Func- tional length is the distance between the termina Table 6— MEASUREMENTS OF CARNIVORE SKELETONS' AMNH= American Museum of Natural History; CM=Carnegie Museum; CNHM = Chicago Natural History Museum; USNM = United States National Museum Ailuropoda melanoleuea ^ g m CNHM w ^~ 31128 c? 278 34258 285 36758 9 267 39514 277 47432' cfj 264 74269' cf 308 CM 18390 284 AMNH 110451 9 275 110452 9 265 110454 280 USNM 258423 274 259027 d' 295 259074 cf 282 259401 cf 266 259402 cf 290 259403 9 268 259076 9 238 258984 213 259400 9 243 132095 234 259075 cf 273 258834 cf 273 259029 cf 304 258836 d" 276 258425 d^ CNHM Ursus arctos 43744 321 47419 360 84467 9j 241 Ursus ggas 49882= 9 358 63802 d' 450 27268 440 27270 293 63803 9 Ursus americanus 18864 256 44725' & 273 Ailurus fulgens 65803' 9 112 57193' d' 57211' d' Procyon lotor 49895 d' 116 49227 d' 115 49057 d" 120 47386 d' 114 Gulo luscus 57196 9 158 74056 & 79409 d 167 Canis lupus 21207 9 246 51772 & 263 51773 9 253 54015 9 238 Skull Spine Fore Leg Hind Leg C3 ^ I _ C bo o c 0,2 250 257 246 289 2-c NJ2 252 131 206 132 129 130 144 206 207 180 152 168 149 153 132 ii be O C 685 665 CO >= o MJ3 taj3 M J3 96 164 184 92 164 160 826 105 ea c 73 bOc« fe.S OS'S -1 II o 2 ►J E 279 209 50.5 282 212 53 224 51.2 216 55.9 277 214 53.0 273 210 57 282 222 58 314 254 131 210 154 251 133 202 143 245 126 207 145 . 170 184 266 212 276 211 254 130 193 141 .. 173 180 267 211 271 206 253 134 199 142 160 157 260 199 260 202 267 135 213 164 737 192 186 280 222 53 290 225 57 256 130 206 167 680 164 164 272 ... 57.5 280 ... 247 122 200 150 261 142 213 159 794 162 185 256 ... 52.5 285 215 58 240 125 202 159 . 229 123 172 120 . 203 106 142 100 128 129 239 124 181 124 . 224 120 167 113 . 255 138 206 145 . 253 133 142 202 215 145 . 165 255 136 149 . 131 n 163 273 204 276 203 320 152 190 155 92 !2 93 2( )9 192 304 247 73.5 355 248 78 330 172 227 172 9; )0 01 2: ;6 220 312 255 73.5 377 253 78 234 128 132 95 . .. n !6 117 204 162 249 179 332 180 .. 2f 53 222 330 283 90 402 280 93 433 225 267 232 352 336 415 345 109 519 355 116 390 208 267 221 . 3 1 301 386 305 105 464 315 107 285 153 158 125 . 2; )3 230 346 276 ... .. 24 17 221 327 268 91 390 275 96 242 130 139 106 7< 51 81 1. 50 151 244 191 65.4 276 197 67 270 133 155 124 81 8 86 1' 72 157 268 225 70.5 318 232 73 102 53 76.5 38.5 3' r5 40.5 ( 51 57 107 83.2 115.5 107 .. 3( )6 40 ( 40.5 ( 54 58 105.2 83 55 59 111 80 30 115.5 104 30.5 116.5 104.5 38 38.5 115 70. 5 75.5 3( 50 38.0 ' r5 70 110.8 110.3 134.7 136.4 113 70 76 3< 2 36.5 ' 12 65 109 112 .. 130.5 127.5 116 72 76 3( 13 33 1 52.5 59 96 95.5 113.5 116 113 71 79.5 61 32 !7 35 ( 59 59.5 99 101 30 118.5 121 37.5 149 82 53.5 ! 56 80 139 107.5 .. 145 131.5 145 79 104 90 5i :0 57 ; iS 84 139 112 47 148 134 56 154 83. 5 110 95 c )2.5 90 141 111 47.5 151 135 56 234 122 129 77 74 1; )7 107 209 213 94 234 233 107 248 135 139 81.6 75 )5 80 ll 54 108 228 227 255 246 109 241 130 133 80.5 7f 53 76 IE )7 110 216 210 96 241 233 104 225 120 133 76 . 69 U )4 100 209 207 95.5 230 232 108 • Pelvic measurements on p. 103. ' Zoo specimen. 45 46 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 articular surfaces of the bone. In most instances the appropriate point on the articular surface is either the same as that used for greatest over-all length or can be fixed with equal precision. In a few instances — both ends of the radius, and the distal end of the tibia — the shape of the articular surface makes it impractical to fix exactly the proper point from which to measure, and conse- quently the corresponding measurements are less precise. I have measured from the approximate center of such oblique articular surfaces. In a study of the present kind the advantages of com- paring functional lengths outweigh any disadvan- tages resulting from slightly lessened precision. For metacarpal and metatarsal length the long- est bone was measured, regardless of which one it happened to be. For Ailiiropoda this is meta- carpal 4 and metatarsal 5; for all other species in the table it is metacarpal 4 and metatarsal 4. In measuring the scapula, height was measured along the spine, from the glenoid cavity to the ver- tebral border. Breadth is the distance between two lines that are parallel to the spine and intersect the anterior and posterior borders of the scapula. Length of the vertebral column was measured from the anterior border of the ventral arch of the atlas to the posterior border of the centrum of the last lumbar. The column of the smaller species was still articulated by the natural ligaments, and length was measured along the cui-ves of the artic- ulated spine. For the larger species, in which the bones were disarticulated, the vertebrae were laid out in proper sequence on a flat surface, following the natural curves of the backbone. Length was then measured along the cui'ves. All measurements are in millimeters. Cranial Capacity Cranial capacity was measured by filling the cranial cavity with dry millet seed and then meas- uring the volume of the millet seed in a gi'aduated cylinder. Ten trials were made for each skull, and the trial that gave the highest reading was re- garded as the closest approximation to the true cranial capacity. The difference between the low- est and highest reading averaged less than 4 per cent for all skulls, and in no case was it greater than 6 per cent. In cranial capacity, as in other basic size charac- teristics, the giant panda resembles the American black bear very closely. III. THE SKULL The skull of Ailuropoda is characterized by its great density and by extreme development of the sagittal crest and expansion of the zygomatic Table 7.— CRANIAL CAPACITY OF CARNIVORES Ailuropoda melanoleuca CNHM C.C. 31128 d" 320 36758 9 288 39514 — 282 Mean 297 Ursus americanus CNHM 16027 — 280 18146 — 261 18151 — 310 18152 d' 313 51641 cf 312 68178 d' 327 Mean 300 Ursus arctos CNHM 25713 — 412 81509 — 335 arches in comparison with other arctoid carni- vores. These features are associated with very powerful dentition and masticatory musculature. The cranial skeleton and to a lesser extent the facial skeleton are profoundly modified by the de- mands of mastication. The cranium gives the impression of having been subjected to plastic de- formation by the temporal muscle, which has at- tempted, so to speak, to achieve maximal volume. Expanding to the limit in all directions, the tem- poral muscle has displaced and compressed sur- rounding structures to the mechanical limit on the one hand, and to the limits of functional tolerance on the other. The face, on the contrary, is rela- tively unmodified except where it is hafted to the cranium, and in the expansion of the alveolar area in association with the enlarged cheek teeth. The sutures between bones are almost com- pletely obliterated in adult skulls. The bones of the cranium are much thickened. In the parietal region total thickness is 5 mm. (two individuals), whereas in a skull of Ursus arctos the bone in the same region measures 2.3 mm. and in a skull of Ursus americanus only 1.7. The increased thick- ness in the panda involves only the outer lamina of the bone; the inner lamina is no thicker than in the bears. This is likewise true of the basicranial region: in a sectioned skull of Ailuropoda the outer lamina of the sphenoid is 2.6 mm. thick below the sella, whereas in a skull of Ursus americanus it is only 0.9 mm. The difference is similar in the man- dible; at the level of the posterior border of M2 the body is 12.2 mm. thick from the mandibular canal to the external surface of the bone in Ailuropoda (3.6 mm. in Ursus americanus), and 5 mm. from DAVIS: THE GIANT PANDA 47 the mandibular canal to the inner surface (3.4 mm. in Ursus americanus) . The bones of the face, on the contrary, are little if any thicker in Ailuropoda than in Ursus. Ailurus agrees more or less closely with the giant panda in skull proportions. As was pointed out by the earliest investigators, there is also a super- ficial resemblance to the hyenas, associated with similar masticatory requirements. In the following description the skull of the Euro- pean brown bear {Ursus arctos) is used as a basis for comparison. Four adult skulls of Ailuropoda in the collection of Chicago Natural History Mu- seum were available for detailed examination. One of these (no. 36758) was bisected in the sagittal plane and cut frontally through the right auditory region. None of these skulls shows the sutures; these were determined on a young female skull bor- rowed from the U. S. National Museum (USNM No. 259076). A. The Skull as a Whole (1) Dorsal View In dorsal view (norma verticalis) the skull of Ailuropoda is dominated by the tremendously ex- panded zygomatic arches. These form nearly a perfect circle, compared with the triangular out- line in Ursus and other carnivores. The primary result of this expansion is to increase the volume of the anterior third of the temporal fossa. The muzzle appears to be shortened and has often been so described. This is not true, how- ever; the pre-optic length is nearly identical in Ailuropoda and Ursus. The muzzle is no wider anteriorly than in Ursus; its borders divei'ge pos- teriorly instead of being nearly parallel as in Ursus, but this merely reflects the broader cheek teeth of the panda. The postorbital process on the frontal is scarcely indicated, and in one skull it is absent. The alveolar pocket of the tremendous second up- per molar is conspicuous immediately behind the floor of the orbit; this is invisible from above in Ursus but is equally prominent in Ailurus and Procyon. The interorbital diameter is not greater in the bears than in the giant panda, but the post- orbital constriction is more pronounced in the panda, and this increases the volume of the ante- rior part of the temporal fossa. This constriction is reflected in the form of the brain, which in Ailu- ropoda is much narrower anteriorly, in both trans- verse and vertical diameters, than in Ursus. The maximal cranial diameter is about 10 per cent greater in Ailuropoda, and this, together with the greater postorbital constriction, gives a character- istic hourglass outline to the skull in dorsal view. Thus the volume of the anterior part of the tem- poral fossa has been increased by expansion both laterally and medially, whereas the volume of the posterior part of this fossa has been far less affected. The skull of Ailurus exhibits a similar increase in the volume of the anterior part of the temporal fossa. In the hyenas, in which the volume of the temporal fossa is also notably increased, it is the posterior part of the fossa that is expanded by pos- terior extension. The reasons for this difference between herbivorous and carnivorous forms are discussed later (see p. 155). The horizontal shelf formed by the posterior root of the zygoma is not wider in Ailuropoda than in Ursus, but it is carried farther forward along the ventral border of the arch, thus increasing the ar- ticular surface of the glenoid cavity on its inferior surface and the area of origin of the zygomatico- mandibular muscle on its superior surface. There are conspicuous muscle rugae, barely indicated in Ursus, on the inner face of the posterior half of the zygoma. The sagittal crest appears to have a conspicuous sagittal suture, but the juvenile skull shows that this is actually the first suture to close, and that the "suture" in the adult results from secondary up- growth of the frontals and parietals. The smoothly curved outline of the lambdoidal crest contrasts with the sinuous crest seen in Ursus, Ailurus, and Procyon; it reflects the posterior expansion of the temporal fossa. (2) Lateral View In norma lateralis (fig. 17) the skull of the panda contrasts sharply with the bears in the facial angle as measured from the Frankfort horizontal. In Ursus the toothrow is depressed from the Frank- fort horizontal at an angle of about 22°, whereas in Ailuropoda these two lines are nearly parallel. Reference to the ventral axis of the braincase re- veals, however, that the angle formed by the tooth- row is nearly identical in Ailuropoda and Ursus. Actually the position of the orbit is depressed in Ailuropoda, as a part of the over-all expansion of the temporal fossa, and therefore the Frankfort horizontal is misleading in this animal. The strongly convex dorsal contour of the skull increases the area of the temporal fossa dorsally. At the same time the vertical diameter of the mas- seteric fossa of the mandible is much greater than in Ursus. Thus the whole postorbital part of the skull appears expanded, and the skull has a trape- zoidal outline when viewed from the side. The margin of the nasal aperture in the panda curves sharply dorsally, its dorsal third lying at a right angle to the long axis of the skull. Behind 48 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Crista orbitalisi sup. For. ethmoideum For. aptiaim Fissura orbiialis M. temporalis Prof postorbitalis IfrotUalis] Fossa musculans Fossa lammalis For. injTaorbitalis I M pterygoideus int M pterygoideus ext For. ovale Can. palatina posl. mitior Crista orbitalis inj. For. spkeuopatatinum * Can. pterygopalatinum Meatus aruxtieus exlernus Proc. paroccipitalis Proc. masloideus For. poslgtenoideum Fig. 17. Skull of Ailuropoda seen from left side (norma lateralis). the nasal aperture the surface of the nasal and premaxillary bones shows a pattern of shallow grooves, in which lie the terminal ramifications of the infraorbital and external nasal vessels, and small foramina through which nutrient twigs from these vessels entered the bone. The infraorbital foramen is small and less elliptical in cross section than in Ursus. Below and in front of the orbit the anterior root of the zygomatic arch bulges forward conspicuously. The postorbital process of the jugal is less prominent than in the bears, in which it reaches its maximal development among the Arc- toidea. The temporal fossa in Ailuropoda is relatively enormous, in keeping with the size of the temporal muscle. Its anteroventral boundary, separating it from the orbit, is well marked by the superior or- bital ridge. Anteroventrally the fossa is provided with about three well-developed muscle ridges, paralleling the superior orbital ridge; in Ursus cor- responding muscle ridges are present, but scarcely more than indicated; in Ailurus there is a single ridge in old adults. In the upper posterior part of the fossa, near the juncture of the sagittal and lambdoidal crests, is a conspicuous nutrient fora- men; a similar foramen is present in the bears but is lacking in other arctoids. In Ailuropoda the infratemporal fossa is sepa- rated from the orbit above by the well-marked inferior orbital ridge throughout most of its length. Behind the orbital fissure it is separated from the temporal fossa by an indistinct elevation extend- ing from the superior orbital ridge in front of the orbital fissure to the anterior lip of the glenoid fossa. The infratemporal fossa is relatively small. The anterior half of the infratemporal fossa con- tains the entrance to the infraorbital foramen, the common foramen for the sphenopalatine (spheno- palatine artery and nerve; nasal branches of sphen- opalatine ganglion) and pterygopalatine (descend- ing palatine artery and nerve) canals. These exit by separate foramina in Ursus and other carni- vores, but are combined in Ailurus; they have undoubtedly been crowded together in the two pandas by the enlarged maxillary tuberosity. The posterior half of the fossa, from which the ptery- goid muscles arise, exhibits muscle rugosities. The areas of origin of the pterygoid muscles are sharply marked on the bone. The area of pterygoid origin is much reduced, both vertically and horizontally, as compared with Ursus. In Ailuropoda the foramen rotundum (maxillary branch of trigeminus) is confluent with the orbital fissure, although the identity of the two openings is usually indicated by a low ridge and on one side DAVIS: THE GIANT PANDA 49 of one skull there is a paper-thin partition sepa- rating them. This is a feature in which Ailuro- poda differs from all other canoids; it is associated with the general crowding together of non-masti- catory structures in the skull. Ailuropoda also lacks an alisphenoid canal, which is present in Ur- sus. In forms having an alisphenoid canal (Cani- dae, Ursidae, Ailurus) the foramen rotundum is situated within the canal; in Ailurus it is sepa- rated from the orbital fissure only by a thin sep- tum, but the two are some distance apart in the dogs and bears. In forms lacking an alisphenoid canal (Procyonidae, Mustelidae), the foramen and the orbital fissure are separated by a thin septum. In Ursus the vertical diameter of the infratem- poral fossa is much greater than in Ailuropoda. This is also true in Cants but not in the procyo- nids, in which the relatively much larger orbit encroaches on it. Reduction of the infratemporal fossa in Ailuropoda is correlated with the more ventral position of the eye, and thus secondarily with the ventral expansion of the temporal fossa. The tremendously enlarged maxillary tuberosity, associated with the enlargement of the molar teeth, further reduces the volume of the fossa. The Orbit. — The orbit in Ailuropoda, as in other arctoids, is poorly defined on the skull ; only the medial wall is entire. The orbit is an elongate cone with the base formed by the incomplete bony ring of the eye socket (completed by the orbital ligament), and the apex by the orbital fissure. On its medial wall the dorsal and ventral boundaries, separating the orbit from the temporal fossa above and the infratemporal fossa below, are well marked by the superior and inferior orbital ridges. These ridges are less prominent in other arctoids. Else- where the boundaries of the orbit are poorly marked on the skull; because of the feebly devel- oped postorbital processes on both frontal and jugal, even the anterior limits are poorly indicated in Ailuropoda as compared with those of other arctoids. The orbit is rotated slightly ventrad as com- pared with that of Ursus. Its long axis (from the orbital fissure to the center of the eye socket) forms an angle of about 10° with the long axis of the skull in Ursus, whereas in Ailuropoda the axes are parallel. At the ventral boundary of the or- bital opening there is a prominent crescent-shaped depression, which in life lodges a cushion of extra- ocular fat. The lacrimal fossa, which lodges the lacrimal sac, is a large funnel-shaped pit at the antero- medial corner of the orbit. The nasolacrimal canal opens into the bottom of the fossa. The canal is only a millimeter or two long, opening almost at once into the nasal cavity, immediately beneath the posterior end of the maxilloturbinal crest. Ur- sus is unique in having the nasolacrimal canal open into the maxillary sinus. Immediately behind the lacrimal fossa is a shallow pit, the fossa muscu- laris, in which the inferior oblique muscle of the eye arises; the thin floor of this pit is usually broken through on dry skulls, and then resembles a fora- men. In Ursus and other arctoids the lacrimal fossa is much smaller than in Ailuropoda, but otherwise similar. The fossa muscularis in Ailu- rus is very similar to that of Ailuropoda; in Ursus it is relatively enormous — as large as the lacrimal fossa and several millimeters deep. The fossa muscularis is completely wanting in the Canidae and Procyonidae. Three foramina in a row, about equidistant from each other, pierce the medial wall of the posterior half of the orbit. Each leads into the cranial fossa via a short canal directed posteriorly, medially, and ventrally. The most anterior is the ethmoi- dal foramen, which conducts the external eth- moidal nerves and vessels into the anterior cranial fossa. Behind this is the optic foramen (optic nerve, ophthalmic vessels), and most posteriorly and much the largest is the combined orbital fissure and foramen rotundum (oculomotor, trigeminal, trochlear, and abducens nerves; anas- tomotic and accessory meningeal arteries; orbital vein). Except for the confiuence of the orbital fissure and foramen rotundum, which is peculiar to Ailuropoda, the pattern of these three foramina is similar in all arctoids. Most variable is the eth- moidal foramen, which differs in size among the genera and may be characteristically multiple (e.g., in Canis). The foramen ovale, in forms in which it is separate from the orbital fissure, trans- mits the third (mandibular) branch of the tri- geminus and the middle meningeal artery. The zygomatic arch functions in the origin of the temporal fascia from its superior border, the temporal and zygomaticomandibular muscles from its internal surface, and the masseter from its in- ferior surface. Its anterior root lies over the first upper molar (over the second molar in Ursus), its posterior root over the glenoid fossa; the arch is therefore important in resolving the forces gener- ated during mastication. As pointed out above, the anterior part of the arch is expanded laterally, which increases the volume of the anterior third of the temporal fossa. In lateral view the arch is straighter than in Ursus and other arctoids. Its posterior half is much extended dorsally, which increases the available area of origin for the zygo- maticomandibularis muscle. The whole structure 50 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 For. nuiritium For. palatinum ant. med. Fossa palatina For. palatinum ant. Sulcus palatinus For. palatinum post. For. palatinum post For. palatinum Spina nasalis post: Fotsa nasopharyngea Semican. M. tensor- tl/mpani Semican. tubae „ , . audilirae Can. chordae tympani- For. postglenoideum For. lacerum post. Proc. mastoideus M. masseter M. z>'gomatico- mandibularis ■Incisura palatina M. pterj'goideus ext. For. ovale Fossa }nandiimlaris Hamulus plerygoideus Proc. postglenoideus -.^featiis acusticus at. M. stemomastoideus For. slylomasloideum Fossa hyoidea Proc. paroccipitalis M. digastricus For. hypoglossum For. mastoideum Capsula articularis M. longus capitis M. rectus capitis ventralis Fig. 18. Skull of Ailuropoda seen from below {norma ventralis). is extraordinarily massive. The anterior i"oot is bulky but relatively thin-walled, since it is exten- sively excavated internally by the maxillary sinus. It bulges forward anteriorly, and posteriorly forms the floor of the orbit for a short distance before passing into the alveolar pocket of the second molar; the infraorbital canal is thus considerably lengthened posteriorly. The posterior root of the arch is expanded posteriorly to accommodate the large mandibular (glenoid) fossa; it has encroached considerably on the space between the postglenoid and mastoid processes, in which the external audi- tory meatus lies, and the meatus is consequently much compressed. (3) Ventral View In ventral view {norma ventralis, fig. 18) the facial region is dominated by the massive denti- tion, the cranial region by the immense mandib- ular fossae. It has often been stated that the palate extends farther posteriorly in Ursus than in Ailuropoda, but this is an illusion created by the enlarged teeth of the latter. In relation to the anterior end of the braincase, the palate actually extends farther pos- teriorly in the panda. The lateral borders of the palate are parallel, as in Urstis; in other arctoids they diverge posteriorly. The anterior palatine foramina, which transmit nerves, vessels, and the DAVIS: THE GIANT PANDA 51 incisive duct, are situated in the posterior part of the large palatine fossa as in other arctoids. There is a median nutrient foramen between the fossae anteriorly, and a small median anterior pala- tine foramen (large in Ursus and procyonids) opening into a minute canal that arches back through the anterior part of the bony septum, lies between the fossae posteriorly. A shallow gi'oove, the sulcus palatinus in which the anterior pala- tine artery lay, connects each anterior palatine foramen with the posterior palatine foramen, which is situated at the level of the first molar and represents the outlet of the pterygopalatine canal. Immediately behind the posterior palatine foramen, at the level of the second molar, is a much smaller opening, the foramen palatina posterior minor. In other arctoid carnivores this foramen (often several) connects directly with the pterygopalatine canal, but in Ailuropoda, be- cause of the immense development of the second molar, its canal comes to the surface briefly as a groove on the lateral wall of the choana (fig. 20), then re-enters the bone and finally emerges several millimeters behind the entrance to the pterygo- palatine canal (fig. 18). A shallow groove, not seen in other arctoids, passes posteriorly from the posterior palatine foramen to the palatine notch (occasionally closed to form a foramen). As in other arctoids, the posterior border of the palate bears a prominent median spine. The choanae (posterior nasal apertures) are separate, the bony septum formed by the vomer extending to (dorsally beyond) the posterior bor- der of the palate. There is much variation in the posterior extent of this septum in arctoids. In Ursus, representing the opposite extreme from Ailuropoda, the septum ends far anteriorly at about the juncture of the middle and posterior thirds of the palate, and the posterior third of the nasopharyngeal meatus is accordingly undivided. Other genera are intermediate between Ailuropoda and Ursus in the posterior extent of the septum. The nasopharyngeal fossa, situated behind the choanae and between the pterygoid processes, is absolutely and relatively wider than in Ursus. The anterior half of the roof of the fossa bears a prominent median keel, the presence and degree of development of which varies with the posterior extent of the septum. The pterygoid processes present nothing unusual. The mandibular (glenoid) fossa is the key to other modifications of the skull in Ailuropoda. The transverse cylindrical mandibular articulation, limiting jaw action to a simple hinge movement vertically and a very restricted lateral move- ment horizontally, is a carnivore heritage that is ill-adapted to the feeding habits of this animal. In Ailuropoda the transverse diameter of the fossa is much greater than in other arctoids. This di- mension amounts to 30 per cent of the basal length of the skull, while in other arctoids it ranges be- tween 15 and 20 per cent, only slightly exceeding 20 per cent even in Ailurus. The increase in the length of the fossa in Ailuropoda has taken place wholly in the lateral direction; the medial ends of the two mandibular fossae are no closer together than in Ursus. The articular surfaces of the medial and lateral halves of the fossa in Ailuropoda are in quite dif- ferent planes. In the medial half the articular surface is almost wholly posterior (against the an- terior face of the postglenoid process), while laterally the articulation is wholly dorsal (against the root of the zygomatic arch). Transition be- tween these two planes is gradual, producing a spiral fossa twisted through 90°. The form of the fossa is similar, though less extreme, in Ursus and other arctoids. The mechanical significance of this arrangement is discussed below. The Basioccipital Region.— The basioccipital region in Ailuropoda, like other parts of the skull not directly associated with mastication, is com- pressed. It is somewhat shorter (about 5 per cent) anteroposteriorly than in Ursus, and since in addi- tion the postglenoid process is expanded posteri- orly and medially, the structures in this region (foramina, auditory bulla) are considerably crowded together. It is noteworthy that the areas of attach- ment of the rectus capitis and longus capitis mus- cles have maintained their size, partly at the ex- pense of surrounding structures. The foramen ovale (mandibular branch of tri- geminus; middle meningeal artery) occupies its usual position opposite the anterointernal corner of the mandibular fossa. There is no foramen spinosum, since as in carnivores in general the middle meningeal artery passes through the fora- men ovale; the foramen spinosum is sometimes present in Canis (Ellenberger and Baum, 1943). A small foramen situated dorsomedially at the mouth of the foramen ovale opens into a canal that runs medially and anteriorly through the cancellous bone of the basicranium to a point beneath the hypophyseal fossa, where it meets its mate from the opposite side. This canal apparently contained a nutrient vessel; its counterpart was found in Ursus, but not in other arctoids. A single large opening, the entrance to the ca- nalis musculotubarius, is situated at the ante- romedial corner of the bulla. The canal is partly divided by a prominent ventral ridge into a lateral semicanalis M. tensoris tympani and a medial 52 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 semicanalis tubae auditivae. The foramen lacerum medium, which normally lies just medial to the musculotubular canal, is usually wanting in Ailuropoda.^ Laterad of the musculotubular canal, at the medial border of the postglenoid process, is an irregular longitudinal slit, the canalis chordae tympani (canal of Hugier), which transmits the chorda tympani nerve. The position of this open- ing is the same as in Ursus (and arctoids in gen- eral), but in Ailuropoda it is somewhat deformed by the enlarged postglenoid process. The foramen lacerum posterior, which in Ailuropoda includes the carotid foramen, is situ- ated at the posteromedial corner of the bulla. It transmits the ninth, tenth, and eleventh cranial nerves, the internal carotid artery, and veins from the transverse and inferior petrosal sinuses. The posterior carotid foramen, through which the internal carotid enters the skull, is situated in the anterior part of the lacerated foramen ; this is true also of the Ursidae and Ailurus. In other carni- vores (Procyonidae, Mustelidae) the carotid fora- men is removed from the lacerated foramen, lying anterior to the latter along the medial wall of the bulla. Segall (1943) found the positional relations of the posterior carotid foramen to be consistently correlated with recognized family groupings among the Arctoidea. The postglenoid foramen, in the posterior wall of the postglenoid process near the external auditory meatus, connects the temporal sinus (in- tracranial) with the internal facial vein (extra- cranial). The foramen is smaller and more later- ally situated than in Ursus. Laterad of the posterior lacerated foramen, and bounded by the bulla anteriorly and medially, the mastoid process laterally, and the paroccipital process posteriorly, is a pit. This pit, a conspic- uous element of the basicranium, is not present in man and does not seem to have been named. I propose to call it the hyojugular fossa (fossa hyojugularis) . The stylomastoid foramen (fa- cial nerve, auricular branch of vagus nerve, stylo- mastoid artery) lies at the anterolateral comer of the fossa; a conspicuous groove, which lodges the ' In carnivores the foramen lacerum medium (anterior of some authors) transmits chiefly a venous communication between the pharyngeal veins extracranially and the caver- nous sinus intracranially. It also carries an anastomotic twig between the ascending pharyngeal artery (extracranial) and the internal carotid; this anastomotic artery is of con- siderable size in the cats, but in the pandas, bears, and pro- cyonids it is minute or absent. In Ursus the foramen lac- erum medium is larger than the canalis musculotubarius, and two openings, the outlet of the carotid canal posteriorly and the entrance to the cavernous sinus anteriorly, are vis- ible within it. facial nerve, runs laterad and ventrad from the foramen to pass between the postglenoid and mas- toid processes. The hyoid fossa, at the bottom of which the hyoid articulates with the skull, lies in the fossa immediately behind and mesad of the stylomastoid foramen, from which it is separated by a thin wall. Farther posteriorly (sometimes on the crest connecting the paroccipital process with the bulla) is a foramen that transmits a branch of the internal jugular vein that passes to the infe- rior petrosal sinus. The hyojugular fossa is almost identical in Ur- sus, except that it is deeper and more extensive posteriorly. In Ailurv^ it is widely open poste- riorly, between the mastoid and paroccipital proc- esses. The fossa tends to disappear when the bulla is gi-eatly inflated (in procyonids, except Nasua), but it is present in Cants. The hypoglossal (condyloid) foramen (hypo- glossal nerve, posterior meningeal artery) lies be- hind and slightly mesad of the foramen lacerum posterior. In Ursus it is usually connected with the foramen lacerum posterior by a deep groove. A similar groove is present in Ailurus but not in other arctoids. The mastoid process functions in the insertion of the lateral flexors of the head on its posterior surface, and in the origin of the digastric muscle on its medial surface. The process closely resem- bles the corresponding structure in Ursus but pro- jects much farther ventrally than in the latter. It is a powerful tongue-like projection, directed ventrally and anteriorly, extending far below the auditory meatus. The process is strikingly similar in Procyon but is much smaller in other procyo- nids. It is also small in Ailurus and Canis. The paroccipital process, which functions in the origin of the digastric muscle, is much smaller than the mastoid. As in Ursus, it is a peg-like projection connected by prominent ridges with the mastoid process laterally and the bulla antero- medially. In forms with inflated bullae (e.g., Pro- cyon, Canis) the bulla rests against the anterior face of the paroccipital process. The bulla is described in connection with the auditory region (p. 318). (4) Posterior View In posterior view (fig. 19) the outline of the skull has the form of a smooth arch; the constriction above the mastoid process seen in Ursus and other arctoids is not evident. To this extent the nuchal area is increased in Ailuropoda. The posterior sur- face of the skull serves for the insertion of the elevators and lateral flexors of the head and bears the occipital condyles. M. clavotrapezius M rhomboideus M. splenius M. rectus capitis dorsalis medius- Crista lamboidi Mm. biventer cervicus et complexus M rectus capitis dorsalis major For. masioideum Proe. muloideus Proe, paroccxpilalis' M. rectus capitis dorsalis minor M. cleidomastoideus Membrana atlantooccipilalis poelerior Capsula articuiaris M. obliquus capitis anterior M. stemomastoideus M. rectus capitis lateralis M. longissimus capitis 'Membrana teclaria \ ^^SIP' fcaput ventralis) ~^M. stemomastoideus 'M. digastricus Fig. 19. Skull of Ailuropoda seen from rear. Sinus 1 Far. efhmoideum Fossa eerebralis Sinus 2 Fossa olfactoria iMmina eribrosa Sinus I, Tentorium otaeum Fossa cerebelli Sinus sagitUUit Elhmoturbinalia inus transKTSut (pars supj Sasoturlnnale MaxillolUTbtnale For. paUuinum ant. For. palalinum med. anl. Sinus temporalis Sinus transtersus (pars tn/J minor Fossa hvpophyseos' Dorsum seUae ' 'iij- alare For. condyloideum Porus aeusticus int. Fig. 20. Sagittal section of skull of Ailuropoda slightly to left of midline. 53 54 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 In Ailuropoda the peripheral area of muscular attachment is sharply set off from the central con- dylar area by a ridge that runs dorsad from the medial border of the paroccipital process and then curves mesad above the condyle. This ridge marks the attachment of the atlanto-occipital articular membrane; it is not so well marked in other arct- oids. A median nuchal line, prominent in most arctoids, runs vertically from the foramen mag- num to the junction of the lambdoidal and sagittal crests, separating the nuchal area into right and left halves. The area of muscular attachment is rugose, and is punctured with numerous nutrient foramina. A conspicuous scar near the dorsal midline, seen in all except the smallest arctoids, marks the insertion of the biventer cervicis and complexus muscles. The mastoid foramen (meningeal branch of pos- terior auricular artery; vein from transverse sinus) lies directly above the paroccipital process. The condylar area is relatively smooth, and the condyloid fossae present an excavated appearance because of the posterior position of the paroccipital and mastoid processes. The occipital condyles are more obliquely placed than in Ursus, their long axis forming an angle of about 45° with the vertical compared with about 25° in Ursus. The condylar area is interrupted at the ventral border of the foramen magnum, as it is in Ailurus. This condition is usual, but not invariable, in Ursus. In procyonids and canids the condylar area is al- ways carried across as a narrow isthmus below the foramen magnum. In Ailuropoda the form of the foramen magnum varies from a transverse oval to almost square. (5) Internal View A sagittal section through the skull of Ailuro- poda (fig. 20) reveals the nasal cavity, the sinuses, and the cranial cavity. Nasal Cavity.— The nasal cavity is high, nar- row, and elongate in the arctoid Carnivora. This is especially evident in the Ursidae. In Ailuropoda the nasal cavity is slightly higher (index .14) than in Ursus (index .10-.12), and slightly shorter (in- dex .41 vs. .45-.51). In Ailurus the relative height is the same as in Ailuropoda, but the cavity is shorter (index .37). The structures of chief interest in the nasal cav- ity are the turbinates, consisting of three elements: the maxilloturbinals, the nasoturbinals, and the ethmoturbinals. These complex structures were described in detail for various Carnivora by Paulli (1900), and again by Anthony and Iliesco (1926). In some respects, particularly with reference to the ethmoturbinals, it is difficult to reconcile these two studies. Paulli worked chiefiy from frontal sections of the skull, made immediately anterior to the cribriform plate, while Anthony and Iliesco apparently worked from sagittal sections of the skull. The maxilloturbinal (fig. 20) is situated in the anterior part of the nasal cavity, which it nearly fills. It is kidney-shaped, much higher (45 mm.) than long (30 mm.), and its vertical axis is inclined posteriorly at an angle of 20°. It lies entirely an- terior to the ethmoturbinals. The maxilloturbinal is attached to the lateral wall of the nasal cavity by a single long basal lamella, which runs antero- posteriorly in a slightly sinuous line about parallel to the long axis of the skull. The line of attach- ment extends on the premaxilla and maxilla from near the anterior nasal aperture to a point several millimeters caudad of the anterior border of the maxillary sinus. The basal lamella promptly breaks up into an extremely complex mass of rami- fying branches that make up the body of the maxilloturbinal. In the Ursidae, according to Anthony and Ili- esco, the maxilloturbinal is characterized by its great dorsoventral diameter and its extremely rich ramification; Ailuropoda exceeds Ursus in both. According to these authors the Mustelidae resem- ble the bears in the height of the maxilloturbinal and its degree of ramification, although it may be added that in these the upper ethmoturbinals over- hang the maxilloturbinal. In the Canidae and Procyonidae this element is much longer than high, is less complex, and is overhung by the upper ethmoturbinals. In Ailurus it is high (height/ length ratio 1) as in Ailuropoda and the Ursidae but is overhung by the ethmoturbinals; its lamina of origin differs from that of all other arctoids in curving ventrad at a right angle to the axis of the skull, reaching the floor of the nasal cavity at the level of PMl The nasoturbinal in Ailuropoda (fig. 20) is, as in other arctoids, an elongate structure situated in the dorsal part of the nasal cavity. It arises from the upper part of the anterior face of the crib- riform plate and extends forward, above the maxil- loturbinal, to within a few millimeters of the ante- rior nasal aperture. The ethmoturbinal (figs. 20, 21) is very sim- ilar to that of Ursus. As in other carnivores it is composed of a medial series of plate-like out- growths (endoturbinals, internal ethmoturbinals) from the anterior face of the cribriform plate, and a similar more lateral series (ectoturbinals, exter- nal ethmoturbinals), that together fill the posterior part of the nasal cavity. The whole structure constitutes the ethmoidal labyrinth. The rela- DAVIS: THE GIANT PANDA 55 VOME, Ailuropoda Ursus Nasua Fig. 21. Frontal section through turbinates, just anterior to cribriform plate. Roman numerals refer to endoturbinals, Arabic numerals to ectoturbinals. (Diagrams for Ursus and Nasua from Paulli.) tions of these elements are best seen on a frontal section made immediately in front of the cribri- form plate (fig. 21). The endoturbinals number four, the typical number for all Carnivora except the Procyonidae. In the latter, according to Paulli, the fourth endo- turbinal has split into three to produce a total of six. It is impossible to decide, on the basis of the section available to me, how many olfactory scrolls the endoturbinals divide into in Ailuropoda. It is apparent, however, that the complexity is greater than in the Ursidae, in which there are seven. The ectoturbinals number nine, as in the Ursi- dae and Procyonidae. Except for Meles, in which there are 10 (Paulli), this is the largest number known for any carnivore. Ailuropoda further re- sembles the Ursidae and differs from the Procyoni- dae in having the first eight ectoturbinals situated between endoturbinals I and II, and in having the ectoturbinals arranged in a median and an exter- nal series, a long one alternating with a short one to produce the two series. Anthony and Iliesco state that there are seven or eight endoturbinals and that "on peut estimer que les Ours possedent plus de 40 ethmoturbinaux externes." These figures are obviously based on a quite different, and I believe less careful, inter- pretation than Paulli 's. Paranasal Sinuses.— The paranasal sinuses are evaginations of the nasal cavity that invade and pneumatize the surrounding bones of the skull, remaining in communication with the nasal cavity through the relatively narrow ostia. The cavities lying on either side of the dorsal midline are separated by a vertical median septum. The occurrence, extent, and relations of the individual sinus cavities vary greatly among mammals, often even among individuals, and hence topography is an unsafe guide to homologies. The cavity in the frontal bone of many mammals, for example, is not always homologous, and therefore cannot be indiscriminately referred to as a "frontal sinus." Paulli found that the relations of the ostia to the ethmoidal elements are constant, as would be ex- 56 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 pected from the ontogenetic history, and he there- fore based his homologies on these. He rejected the descriptive terminology of hviman anatomy as unusable in comparative studies, and substituted a system of numbers for all except the maxillary sinus. His terminology has been followed here. The dorsal sinuses are enormous in Ailuropoda (fig. 20), far exceeding those of any other carnivore. At the dorsal midline they separate the relatively thin true roof (inner lamina) of the cranial cavity from a much heavier false roof (outer lamina) situ- ated above it. Intrusion of the sinuses into the supracranial area greatly increases the area of the temporal fossa. The relations of the ostia to the ethmoidal ele- ments cannot be determined without cutting the latter away. The single bisected skull available to me could not be mutilated in this way, but simi- larity between the sinuses of Ailuropoda and Ur- sus is so close that there can be no doubt as to nomenclature. As in Ursus, there is no communi- cation between the sinuses. Sinus I, which occupies the frontal region, is much longer, higher, and wider than in Ursus. It is responsible for the characteristic convex fore- head of the giant panda. The posterior wall of the cavity lies at the level of the postorbital proc- ess, as in Ursus, and from here the sinus extends forward into the base of the nasals. Its lateral wall is formed by the outer wall of the skull. The large oval ostium in the floor of the cavity opens into the nasal cavity just anterior to the first endo- turbinal. None of the ectoturbinals extends into this cavity. In Ursus the corresponding cavity is narrower, the maxillary sinus lying laterad of it, and a leaf of the first ectoturbinal projects through the ostium into the cavity. Sinus 1 is a small cavity, measuring only about 15 mm. in length by 20 mm. in height, lying above the olfactory fossa some distance behind sinus I. It is surrounded by sinus 2 on all sides except ven- trally. The small round ostium is situated in the floor. In the skull that was dissected this cavity is asymmetrical ; it was present on the right side only. Sinus 2 is by far the largest of the sinuses. It begins at the level of the postorbital process and extends back through the frontal and parietal bones nearly to the occiput. It is very irregular, with numerous out-pocketings and partial septa. The long slit-like ostium lies in the extreme ante- rior part of the cavity, and as in Ursus a leaf of one of the ectoturbinals projects through the os- tium into the sinus. Sinus IV (sphenoidal sinus of authors) is a large, irregular cavity in the presphenoid. The ostium is situated in its anterior wall, and as in Ursv^ the posterior end of the last ectoturbinal projects through the ostium into the cavity. The maxillary sinus lies almost entirely in the maxillary root of the zygomatic arch, a condition that is unique among carnivores. It is situated farther laterad than in Ursus and other arctoids. This hollowing out of the zygomatic root makes possible a considerable increase in bulk without adding appreciably to its weight. The sinus is an irregular cavity lying directly above the posterior end of the fourth premolar, the first molar, and the anterior end of the second molar. It opens into the nasal cavity, immediately behind and be- low the crest of the maxilloturbinal, by a much smaller ostium than in Ursus. Thus there are five pairs of pneumatic cavities in the skull of the giant panda. Although these greatly exceed the corresponding cavities of Ursus in size, the arrangement and relations are very similar. Ursus has an additional small cavity in the roof of the skull; in Ailuropoda the area it occupies has been taken over by sinus 2, and this enormous sinus has almost absorbed sinus 1. In other arctoids pneumatization of the skull is much less extensive in number of sinuses and in the extent of the individual sinuses. In the Mus- telidae only the maxillary sinus is present, but other arctoids also exhibit at least some pneumati- zation in the frontal region. Ailurus has the same cavities as Ailuropoda, but sinus 2 is much less extensive, extending back only to the level of the optic foramen. Paulli generalized that the extent of pneuma- ticity is dependent on the size of the skull, and pointed out that this is borne out in large vs. small breeds of dogs. Another over-riding factor obvi- ously has operated in the pandas. In Ailurus the absolute size of the skull compares with that of Procyon, but the sinuses are more extensive. In Ailuropoda the skull is about a third smaller than that of Ursus arctos, but the dorsal and lateral sinuses are much larger. The secondary factor in pandas is a mechanical one. It is well known that the sinuses develop as evaginations of the walls of the nasal cavity, and that with increasing age these out-pocketings grad- ually invade the surrounding bone. The process is called "pneumatic osteolysis," but the nature of pneumatic osteolysis is unknown. In Su Lin (age 16 ± months, all permanent teeth in place) sinus 2 had not yet invaded the parietal; it termi- nated at about the fronto-parietal suture. In this animal, sinus I in the nasofrontal region also falls short, by about 20 mm., of its adult anterior ex- tension. The vertical height of both these cavi- ties, on the other hand, is as great as in the adult. DAVIS: THE GIANT PANDA 57 Thus considerable peripheral growth takes place in the larger sinuses after essentially adult skull size has been attained. Cranial Cavity.— The cranial cavity (fig. 20) is a mold of the brain, and in the panda it differs far less from the typical arctoid condition than do other parts of the skull. The cavity is divided into the usual three fossae: olfactory, cerebral, and cerebellar (anterior, middle, and posterior of human anatomy). The olfactory fossa is much reduced in diam- eter as compared with that of Ursus, but is other- wise very similar. It houses the olfactory bulbs. The floor of this fossa is on a higher level than the remaining cranial floor. In the midline of the floor a prominent ridge, the crista galli of human anatomy, extends nearly the entire length of the fossa. The cribriform plate, forming the ante- rior wall, is perforated by numerous foramina for filaments of the olfactory nerve. These foramina are larger and more numerous at the periphery of the plate. In the lateral wall of the fossa is a larger opening, the ethmoidal foramen. The cerebral fossa, much the largest of the cranial fossae, houses the cerebrum. As in the bears, a vertical ridge (the site of the sylvian fissure of the brain) separates a larger anterior fronto- parietal region from a smaller posterior temporal- occipital region. This ridge is less obvious in the smaller arctoids. The walls of the fossa bear nu- merous ridges and furrows that conform to the gyri and sulci of the cerebral cortex of the brain. A conspicuous groove immediately in front of the sylvian ridge lodges the middle meningeal artery; a smaller groove, which houses a branch of this ar- tery, lies in the posterior region of the fossa (fig. 22). In Ursus and other arctoids the groove for the middle meningeal artery lies in the posterior re- gion of the fossa. The cerebellar fossa is largely separated from the cerebral fossa by the tentorium osseum, which forms most of its anterior wall. The ten- torium is exceptionally well developed in the bears and pandas. The cerebellar fossa communicates with the cerebral fossa via the tentorial notch, a large opening that in Ailuropoda is much higher than wide; in f7rsMS it is more nearly square. The tentorium slopes backward at an angle of only about 10° in Ailuropoda, while in Ursus this angle is about 25°. The slope is much greater in other arctoids (about 45°). The walls of the cerebellar fossa are grooved and perforated by various venous sinuses (see p. 281) ; otherwise they conform to the shape of the cere- bellum. The medial face of the petrosal is visible in the wall of this fossa. As in Ursus and Ailurus, the tentorium is in contact with the petrosal along the entire petrosal crest, and covers the part of the petrosal anterior to this line. In Canis and the procyonids, in which the tentorium is not so well developed, an anterior face of the petrosal is also exposed in the cerebral fossa. The enlarged ten- torium in the bears and pandas has also crowded out the trigeminal foramen — the large opening in the petrosal near the apex that is so conspicuous in canids and procyonids. In the ursids the root of the trigeminal nerve passes over, instead of through, the apex to enter the trigeminal fossa. In Ailuropoda the most conspicuous feature on the medial face of the petrosal is the internal acous- tic opening, leading into the internal acoustic meatus. Immediately behind this opening is a smaller foramen, the aquaeductus vestibuli, overhung by a prominent scale of bone. Just above and behind the acoustic opening is a bulge in the surface of the petrosal, the eminentia ar- cuata, caused by the superior semicircular canal. In all other arctoids examined (except Procyon) there is a deep pit, larger than the acoustic meatus and situated directly above it, that houses the petrosal lobule or "appendicular lobe" of the cere- bellum; this pit is wanting in Ailuropoda and Pro- cyon. The inferior border of the petrosal is grooved for the inferior petrosal sinus, and the superior angle is crossed by the groove for the transverse sinus. The floor of the cerebral and cerebellar fossae exhibits several features of interest (fig. 22). The dorsum sellae marks the boundary between the cerebral and cerebellar spaces. Most anteriorly, near the middle of the cerebral fossa, is the open- ing for the optic nerve. It leads into a canal, nearly 25 mm. long, that opens in the orbit as the optic foramen. This canal is of comparable length in Ursus but is short in other arctoids. Behind the optic opening is a prominent sulcus for the optic chiasma, of which the canal itself is a con- tinuation. The sella turcica lies in the midline at the posterior end of the cerebral fossa. Of the components of the sella, the tuberculum sellae is wanting anteriorly, but the anterior clinoid processes at the anterior corners are well devel- oped; these processes, to which the dura is attached, are often wanting in arctoids. The posterior clinoid processes are plate-like lateral extensions of the dorsum sellae, overhanging the cavernous sinuses laterally. These processes, to which the dura also attaches, are well developed in all arc- toids examined except Canis, where they are want- ing. The hypophyseal fossa is a well-bounded pit in all arctoids except Canis, in which there is no anterior boundary. 58 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Impressio A. meningea med. For. ovale Fossa trigem Hiatus canalis facialis For. carot.ant. Siniis petrosus inf. Meatus acusticus interims Aquaedudus restibuli For. lacerum post. Sinus sigmoideiis Can. hypoglossi (condyloideum) For. mastoideum Sinus transversus For. opticum Fissura orbitalis + For. rotuiidum Proc. clinoideus ant. Sella turcica Sinus carer nosus T — Dorsum sellae Proc. clinoideus post. Clirus Pars basilaris occipitale For. magnum Fig. 22. Left half of basicranium of Ailuropoda, internal view. On either side of the sella turcica is a wide longi- tudinal sulcus, extending from the orbital fissure anteriorly to the petrosal bone posteriori}', in which the cavernous sinus Hes. Anteriorly the sulcus opens into the orbit through the large open- ing formed by the combined orbital fissure and foramen rotundum; fusion of these two foramina is peculiar to Ailuropoda. A ridge on the floor of the sulcus marks the boundary between the orbital fissure (medial) and foramen rotundum (lateral) of other arctoids. In the posterior part of the sulcus, just in front of the apex of the petrosal, is a deep narrow niche, the trigeminal fossa, which lodges the semilunar ganglion of the trigeminal nerve. The foramen ovale (third and fourth branches of trigeminus) opens into the floor of the niche an- teriorly; in Ursus, in which the trigeminal fossa extends farther anteriorly, both the foramen ro- tundum (^second branch of trigeminus) and the foramen ovale open directly into it. A small roimd opening at the posterior end of the trigeminal fossa is the outlet of the hiatus canalis facialis, through which the great superficial and deep pe- trosal nerves enter the cranial cavity. Imme- diately above this is a smaller opening (more conspicuous in Ursus), the foramen petrosum superior, the anterior outlet of the superior pe- trosal sinus. The anterior carotid foramen Hes at the an- terior corner of the petrosal, directed anteriorly and medially. In Ailuropoda, in which there is no foramen lacerum medium, the internal carotid artery passes from the carotid canal directly into the cavernous sinus, and the anterior carotid fora- DAVIS: THE GIANT PANDA 59 men is thus intracranial. In Ursus, the artery, after leaving the carotid canal, passes ventrad into the foramen lacerum medium, where it immedi- ately doubles back upon itself to pass nearly ver- Sinus cavernosus' Sinus petrosus iiif/ sinus runs nearly vertically, connecting the sagit- tal sinus above with the vertebral vein below. It is sharply divided into inferior and superior parts. The inferior section, much larger in caliber, lies Sinus sagillatis sup. Sinus rectus Sinus transversus (pars sup:) Sinus temporalis .V mastoidea Sinus transversus (pars inf) V verlebralis Sinus sigmoideus To V jugularis int.fvia for. lac. post J To V facialis inl.[via for. postglenj Sinus petrosus sup. Fig. 23. Sinuses and diploic veins. Right half of skull of Ailuropoda, internal view (semi-diagrammatic). tically into the cavernous sinus. Thus in Ursus the foramen in the floor of the cavernous sinus is the internal opening of the foramen lacerum me- dium, and the anterior carotid foramen is visible only externally within the foramen lacerum me- dium. The situation in Ailurus and Procyon is similar to that in Ursus. Obliteration of the fora- men lacerum medium and of the associated flexure in the internal carotid artery in Ailuropoda is un- doubtedly correlated with the general crowding of non-masticatory structures in this region and is therefore without functional or taxonomic signifi- cance. Cams, as usual, is quite different from either the Ursidae or Procyonidae. The inferior petrosal sinus lies just mesad of the petrosal, largely roofed over by a lateral wing of the clivus. The sinus is continuous anteriorly with the cavernous sinus and posteriorly with the sigmoid sinus, which name it assumes at the fora- men lacerum posterior, at the posterior corner of the petrosal. The superior petrosal sinus is re- duced to thread-like caliber in Ailuropoda and Ursus as a result of the great development of the tentorium. It opens into the trigeminal fossa via the superior petrosal foramen, at the apex of the petrosal. From here the sinus arches posteriorly around the petrosal, enclosed in the temporal bone, and enters the temporal sinus. The transverse in an open groove behind the petrosal, the upper part of the groove crossing the petrosal. The mastoid foramen and several diploic veins open into this part of the sinus. At the dorsal border of the petrosal the sinus gives off the large tem- poral sinus, which descends as a closed canal to open extracranially via the postglenoid foramen. The superior section of the transverse sinus con- tinues dorsad as a closed canal, much reduced in caliber, to open into the sagittal sinus at the dorsal midline. The sagittal sinus is visible for a vari- able distance as a shallow groove along the midline of the roof of the cerebral fossa. The short sig- moid sinus runs posteriorly from the foramen lacerum posterior, meeting the transverse sinus at a right angle about 5 mm. behind the posterior bor- der of the petrosal. Beyond the confluence of the inferior petrosal and transverse sinuses a groove, which houses the vertebral vein, continues caudad through the lateral corner of the foramen magnum. The vertebral vein lies in a similar groove in Ur- sus, while in all other arctoids examined (including Ailurus) the groove is roofed over to form a canal. From the dorum sellae the floor of the basi- cranium slopes backward and downward as the clivus. This region is basin-shaped to conform to the shape of the pons, and is separated by a transverse ridge from the basilar portion of the 60 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. pterygoideus ext For. matidibularis M. pterygoideus int. M. temporalis (prof.) Proc. morytnw M. digastrii M. temporalis superf.+ M. zygomaticomandibularis — Fossa masseterica Proc. angutaris ^_ _ ^, M. masseter For. merUaiia Fig. 24. Left mandible of Ailuropoda: external surface lower right, internal surface upper left. basioccipital bone lying behind it, which supports the medulla. The whole plate-like floor of the basicranium lying behind the dorsum sellae is shorter and wider in Ailuropoda than in Ursus. The hypoglossal (condyloid) foramen pierces the floor of the basilar portion in a lateral and slightly anterior direction, just anterior to the foramen magnum. Mandible. — The mandible of Ailuropoda is no- table for its extreme density. Its weight is more than twice that of the mandible of a much larger bear. The two halves of the mandible are firmly fused at the symphysis, with no trace of a suture, in all adults examined. This is contrary to the condition in Ursus and other arctoids. Fusion is nearly complete in a young adult Ailuropoda, in which most skull sutures are still open. The length of the symphysis is also remarkable. It is relatively nearly twice as long as in Ursus, and extends to the anterior border of the first molar instead of the third premolar. In Ailurus, by contrast, the symphysis is short (barely reaching the first premolar), and the two halves of the man- dible do not fuse. The body of the mandible, viewed from the side, tapers from the ramus forward, whereas in Ursus (but not in other arctoids) the height of the body is quite uniform. Among several mandibles of Ailuropoda the inferior border is curved in vary- ing degrees, reaching its nadir below the second molar; in one mandible this border is nearly as straight as in Ursus. The body is less high ante- riorly than in Ursus, and higher posteriorly, and this is probably correlated with the relatively feebly developed canines and the large molars. The up- per or alveolar border of the body lies about 30 mm. below the level of the articular condyle, whereas in Ursus these are at very nearly the same level (fig. 25). There are typically two mental fora- mina, as in arctoids in general. These are sub- equal in size. The more anterior foramen is often broken up into several smaller foramina. Throughout its length the body is more than twice as thick as in Ursus, and viewed from below the body arches abruptly laterad at the posterior end of the symphysis, giving a Y-shape to the ven- tral outline of the jaw. DAVIS: THE GIANT PANDA 61 Ailuropoda 31128. Basal skull length 235 mm. Ursus 21859. Basel skull length 303 mm. Fig. 25. Outlines of posterior ends of mandible of Ailuropoda (solid line) and Ursus horribilis (broken line) superimposed. Note (1) the excavation of the posterior border of the coronoid process, (2) the much deeper masseteric fossa, and (3) the depressed occlusal plane in Ailuropoda. The ramus, which is that part of the mandible lying posterolaterad of the last molar, differs from that of Ursus in several important respects. Be- sides bearing the mandibular condyle, the ramus functions chiefly for the insertion of the muscles of mastication. The areas where these muscles attach are large, well marked, and rugose in Ailuropoda. The masseteric fossa, in which the zygomatico- mandibular muscle inserts, is larger than in Ursus in both vertical and transverse diameters. The vertical diameter in particular has been increased relative to Ursus (and other arctoids) by exten- sion ventrad. It is also deeper, for the edges have been built out. The surface of the fossa is ex- tremely rugose, and is marked by several promi- nent transverse ridges (cristae massetericae) for the attachment of tendinous sheets in the muscle. The coronoid process, into which the masseteric fossa grades imperceptibly, functions in the in- sertion of the temporal muscle on both its lateral and medial surfaces. This process is similar to that of Ursus, except that its posterior border is eroded away, giving it a scimitar-like form and greatly reducing the area available for temporal insertion (fig. 25) . The angular process is a small but conspicuous prominence on the posteromedial border of the ramus, below the condyle. It pro- jects medially and posteriorly, instead of posteri- orly as in other arctoids. This process characteris- tically provides insertion for part of the masseter on its outer surface and part of the internal ptery- goid on its inner surface; none of the masseter fibers reach it in Ailuropoda. In Ursus and other arctoids (including Ailuru^) the angular process is large and tongue-like, with well-marked muscle scars for both the masseter and the internal ptery- goid. In Ursus a conspicuous marginal process (Toldt's terminology) on the inferior border of the ramus, anterior to the angular process, provides the main insertion for the digastric muscle. This process is wanting in other arctoids. In Ailuro- poda the insertion of the digastric is more diffuse than in Ursus, and the marginal process, while present, is less clearly marked and is situated on the medial surface of the mandible immediately in front of the internal pterygoid scar. Hypertrophy of the jaw-closing muscles in the giant panda is reflected in the relatively larger areas of attachment on the skull. The total area of insertion of the masseter and temporal on the lateral surface of the mandible was calculated roughly by plotting on millimeter paper. In Ailu- ropoda (basal skull length 252 mm.) this area amounted to 5368 mm.-, while in a much larger 62 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 LACRIMALE ORBITOSPHENO/D PTERYGOID Fig. 26. Lateral view of juvenile skull of Ailuropoda (USNM 259076), showing sutures. Ursus horribilis (basal skull length 303 mm.) it was only 4774 mm.- The medial surface of the ramus exhibits con- spicuous scars marking the attachment of several muscles. A rugose area occupying most of the medial surface of the coronoid process marks the insertion of the deep layer of the temporal muscle. The anterior border of this area sweeps back be- hind the last molar, leaving a triangular space (about one-fourth of the total medial coronoid surface) free of muscle attachment. The ventral border of the temporal area is a prominent hori- zontal crest at the level of the alveolar border, extending back immediately above the mandibular foramen; this is the level to which the temporal insertion extends in other arctoids. Immediately behind this crest, on the dorsal surface of the con- dyle, is the extraordinarily conspicuous, pock-like pterygoid depression that marks the insertion of the external pterygoid muscle. A much larger scar, below the condyle and extending back onto the angular process, marks the insertion of the in- ternal pterygoid. A triangular rugose area in front of this, beginning posteriorly at the marginal proc- ess, marks the insertion of the digastric. The mandibular foramen, for the inferior alveolar vessels and nerve, is circular instead of oval in cross section. It lies immediately above the mar- ginal process. The condyloid process has the transverse semi- cylindrical form characteristic of the Carnivora, but in Ailuropoda this region is an exaggeration of the usual arctoid condition. The neck support- ing the capitulum is short, flattened, and twisted through 90° — the typical carnivore arrangement. As a result, the medial half of the capitulum is buttressed anteriorly but unsupported below, while the lateral half is buttressed below but unsup- ported anteriorly. In all arctoids the articular sur- face tends to conform to this support pattern, the medial half facing posteriorly and the lateral half more or less dorsally. In Ailuropoda this tendency reaches full expression, and the articular surface is a spiral track rotated through more than 90°, "like a riband wound obliquely on a cylinder," as Lydek- ker stated. To some extent at least, this spiral form is correlated with the large size and dorsal position of the pterygoid depression, which in Ailu- ropoda occupies a part of the area of the articular surface of other carnivores. The width of the capitulum much exceeds that of any other carnivore. The index basal skull length /width capitulum is .27 to .31 for Ailuropoda, while for Ursus it is only .15 to .17. Ailurus is intermediate, with an index of .22 to .23, while all other carnivores examined were below .18 ex- cept an old male zoo specimen of Tremarctos or- natus, in which it was .21. The long axis of the capitulum is oriented at nearly a right angle to the axis of the skull in both horizontal and verti- cal planes. As in carnivores in general, however, the medial end of the axis is tilted slightly caudad and ventrad of 90°. B. Cranial Sutures and Bones of the Skull As was mentioned above, the sutures disappear early in Ailuropoda, and nearly all are completely obliterated on fully adult skulls. The following account of the bones of the skull is based on a young female skull, with a basal length of 213 mm.. DAVIS: THE GIANT PANDA 63 PREMAXI LLA •PER I OTIC CPars mastoidta) ^ occ\P^^ Fig. 27. Ventral view of juvenile skull of Ailuropoda (USNM 259076), showing sutures. on which all but a few of the sutures are still open (figs. 26, 27). This skull is intact, so that only surface features could be examined. For the most part, the relations of the bones differ so little from those of Ursus that there is no point in a detailed description. The exact po- sitions of the sutures are shown in the accompany- ing drawings. The premaxilla is essentially similar to that of Ursus. The maxilla is modified to accommodate the enlarged cheek teeth. The posterior part of the bone forms an enormous maxillary tuberosity that supports the second molar. The tuberosity carries the maxilla back to the level of the optic foramen, whereas in Ursus it extends only to the pterygo- palatine foramen. In the juvenile skull this pos- terior extension of the maxilla has a remarkably plastic appearance, as if the bone had flowed back over the vertical plate of the palatine, squeezing the pterygopalatine and sphenopalatine foramina upward against the inferior orbital crest. A sec- tion through this region (fig. 21) shows that the maxilla lies outside the palatine — that the latter is not displaced backward. 64 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 As in Ursus, at the anteromedial corner of the orbit the maxilla is wedged in between the lacri- mal and jugal, forming a part of the anterior, all the lateral, and a part of the medial boundaries of the lacrimal fossa. The anterior zygomatic root contains a lateral extension of the maxillary sinus, not seen in any other carnivore. The nasals, as in Ursus, are short and their lateral borders are not prolonged forward as in other arctoids. The lacrimal closely resembles the correspond- ing bone in Ursus, which Gregory characterized as "much reduced, sometimes almost vestigial." It is a minute plate, about 5 mm. wide and 12 mm. high, withdrawn entirely from the anterior rim of the orbit, and forming only a small part of the medial surface of the lacrimal fossa. The lacrimal of Ailurus is essentially similar. It is slightly better developed in the procyonids. The jugal (malar) does not depart in any essen- tial respect from the typical arctoid pattern. The palatine, except for the superficial modi- fication produced by the posterior prolongation of the maxilla over the pars perpendicularis, is sim- ilar to that of other arctoids. The pars horizontalis extends forward on the palate slightly anterior to the first molar. The vomer differs from that of Ursus and most other arctoids in the great posterior extent of its pars sagittalis. Otherwise its relations are similar to those of Ursus. The frontal, parietal, squamosal, and oc- cipital have all suffered more or less change in form with the remodeling of the skull to accom- modate the enormous masticatory musculature. Except for the morphologically insignificant dif- ferences resulting from this remodeling, the rela- tions of these bones are typical. The frontoparietal suture, which is relatively straight and about at a right angle to the axis of the skull in Ursus and other arctoids, here arches forward to the level of the optic foramen. At the dorsal midline a narrow tongue of the frontal pro- jects posteriorly between the parietals for about 15 mm., i.e., to about the level of the whole fronto- parietal suture in Ursus. This suggests that in Ailuropoda the parietal has increased anteriorly at the expense of the frontal. The interparietal suture is obliterated, and a secondary upgrowth of bone is approaching the site of the future sagittal crest. On the skull examined, the basioccipital-supra- occipital suture was still open, but the exoccipital- supraoccipital suture was closed. The mastoid portion of the periotic is exposed, as is usual in arctoids, on the posterior side of the mastoid process, where it is wedged in between the squamosal and the occipital. The suture be- tween the periotic and the tympanic disappears early in all arctoids, and was gone in the skull of Ailuropoda studied. The tympanic, in so far as it is visible exter- nally, differs considerably in shape from the cor- responding bone in Ursus. It is obvious, however, that this bone has merely been crowded by the surrounding structures, particularly the postglenoid process. The relations of the tympanic are almost exactly as in Ursus, and posterior expansion of the postglenoid process as seen in Ailuropoda might be expected to alter the form of the tympanic pre- cisely as it has. (This region is described in detail on p. 319). The sphenoidal complex has been affected rela- tively little by the remodeling of the skull and is very similar to the corresponding region in Ursus. In the skull examined, the four elements constitut- ing the complex (basisphenoid, presphenoid, alisphenoid, orbitosphenoid) are still distinct. They differ only in the most trivial respects from the corresponding elements in a young Ursus skull. The pterygoid is completely fused with the sphenoid, and this is one of the very few sutures of the skull that have been obliterated at this age. This condition contrasts sharply with Ursus at a comparable age, in which the pterygoid is still en- tirely separate. The ethmoid is not visible on the surface of the skull. The following sutures are closed in the young skull examined: tympanic-periotic, exoccipital- supraoccipital, pterygoid sphenoid, interparietal. The first two fusions are characteristic of carni- vores at this stage of development. The last two are not, and represent departures from the car- nivore pattern. C. Hyoid The hyoid (fig. 28) differs little from that of bears and other arctoid carnivores. It is composed of the usual nine rodlike bony elements, suspended from the basicranium by a pair of cartilaginous elements, the thyrohyals. The hyoid fossa, at the bottom of which the thyrohyal articulates with the skull, lies in the hyojugular fossa. The hyoid consists of a transverse body and two horns (cornua), an anterior composed of three pairs of bones plus the cartilaginous thyrohyals, and a posterior composed of a single pair of bones. Like all other bones of the skeleton, the hyoid bones of Ailuropoda exhibit more pronounced scars DAVIS: THE GIANT PANDA 65 Slylohyal Cornu anterior Epihyal Cornu anterior Cornu posterior Thyrohyal Cornu posterior Corpus Ceratohyal Fig. 28. Hyoid of Ailuropoda, lateral and ventral views. for muscle attachments than they do in Ursus, although the bones themselves are no more robust. In both the giant panda and the bears the body is a transverse rod, less plate-like than in other arc- toids. The ceratohyal is also less expanded than in other arctoids, and in Ailuropoda it has a distinct longitudinal furrow on the dorsal surface. The epihyal presents nothing noteworthy. The stylo- hyal is flattened and plate-like, with an irregular outline, in Ailuropoda. The thyrohyal is slightly curved and rodlike. D. Review of the Skull The skull and teeth of Ailuropoda were described in some detail by A. Milne-Edwards (1868-1874), Lydekker (1901), Bardenfleth (1913) and Gregory (1936). Each of these made point by point com- parisons with the Ursidae on the one hand and with Ailurus and the Procyonidae on the other, in an attempt to determine the affinities of Ailur- opoda. Conclusions were conflicting; the only legitimate conclusion is that the skull and denti- tion of the giant panda are so modified that the affinities of this animal cannot be determined from these structures alone. I have therefore used other characters in deciding the affinities of Ailur- opoda, which are unquestionably with the Ursi- dae. Here the only important consideration is that no skull or dental character shall point un- equivocally to relationship with any other group of carnivores. The demands of the masticatory apparatus in Ailuropoda have resulted in such extensive and permeating modifications in the skull that many elements have been modified beyond the limits of inter-generic or even inter-family differences with- in the Carnivora. Among those not so affected are the pattern (but not the extent) of the para- nasal sinuses, the turbinates, and the middle ear — all intimately associated with primary sense organs and not affected by muscle action. Each of these structures is very similar to the correspond- ing structure in Ursus. Klatt (1912) has shown that the extent of the frontal sinus is determined by the mass of the temporal muscle, as would be expected, because the sinus lies between the outer and inner lamina of the cranium. The temporal attaches to the outer lamina, whereas the inner lamina encapsulates the brain. Aside from its function of encapsulating the brain and sense organs, the generalized carnivore skull is designed primarily for seizing and cutting up prey. Skulls of omnivorous or herbivorous carnivores are secondary modifications of this pri- mary predatory type. Consider the skull of a generalized carnivore, such as Canis or Viverra, as a construction. How does such a construction compare with those of other generalized mammals in architecturally or mechanically significant ways? 1. The skull is elongate and relatively slender (see Table 8). Elongation of the head is a primi- tive mammalian feature that has been retained in 66 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Table 8.— SKULL PROPORTIONS IN GENERALIZED AND SPECIALIZED CARNIVORES N = SKULL LENGTH: Condylobasal length Length thor. vert. 10-12' FACE LENGTH: Gnathion-ant. end braincase Condylobasal length Preoptic length Condylobasal length SKULL DEPTH: Vertex-inf . border mandible Condylobasal length SKULL BREADTH: Zygomatic breadth Condylobasal length Least diam. braincase Condylobasal length • See page 35. Generalized Flesh- eating Carnivores Canis Viverra lupus tangalunga Predominantly Herbivorous Carnivores Procyon Ailurus lotor fulgens Ursus Extremely Powerful Jawed Carnivores Ailuropoda Hyaena 3 5 3 3 5 3 4 3.2 2.9 3.1 2.9 3.2 2.7 3.5 (3.08-3.23) (2.75-2.95) (3.04-3.24) (2.53-3.21) (2.99-3.44) (2.63-2.72) 51 42 38 36 50 49 49 (50-51) (40-43) (37-39) (34-38) (46-52) (47-53) (48-50) 45 33 30 23 33 31 32 (42.1-46.6) (31.8-33.6) (29.1-30.1) (22.4-24.1) (31.4-35.5) (29.2-32.0) (30.5-33) 49 43 59 68 52 71 71 (48-51) (40-47) (56-62) (67-69) (47-57) (70-72) 57 50 68 75 63 82 71 (55-60) (49-52) (65-70) (71-79) (57-69) (81-82) (66.5-75) 18 15 21 22 24 18 18 (17-19) (14-16) (20-22) (22-23) (22-26) (15-20) (16-20) the Carnivora; the skull was elongate in the creo- dont ancestors of the carnivores and is characteris- tic of generalized mammals. As a tool for seizing and cutting up prey an elongate skull (particularly an elongate face) has certain inherent mechanical advantages and dis- advantages. Speed of jaw closure at the level of the canines is achieved, though at the cost of power. But production of useful force at the sec- torial teeth is mechanically very unfavorable, since more than twice as much disadvantageous force is developed at the mandibular articulation (see p. 69). Preoptic length is a useful measure of face length for our purpose, since it approximately divides the tooth-bearing anterior part of the skull from the posterior muscle-attachment part. Calculated in this way, the face is long in Canis, moderately long in Viverra. Both fall within the known range of the Paleocene Arctocyonidae, the oldest and most primitive of all carnivores: Deltatherium 31 per cent, Eoconodon 38 per cent, Loxolophodon 45 per cent.' Depth and breadth of skull, both intimately as- sociated with mechanics of the jaw, are moderate in both Canis and Viverra. The civet is more slender in both dimensions. 2. Two areas dominate the dental battery: the enlarged dagger-like canines anteriorly, and the en- > Calculated from illustrations in Matthew (1937). larged scissor-like carnassials (P^ and MO poste- riorly. The remainder of the dentition is more or less degenerate. These two specialized areas of the dentition are the key adaptation of the Car- nivora. All other modifications of the skull away from the generalized mammalian condition are effectors of these seizing and cutting tools. These modifications are as follows: 3. The mandibular articulation is a transverse cylinder rotating in a trough-like fossa that is strongly buttressed above and behind. This ar- rangement permits only a hinge movement of the mandible, plus limited lateral shifting of the man- dible; the two may be (and probably normally are) combined in a spiral screw movement. The two halves of the mandible are not fused at the symphysis, which indicates that they are capable of at least some independent movement. 4. The mandibular articulation is at the level of the occlusal plane, and therefore upper and lower toothrows operate against each other like the blades of a pair of shears. 5. The canines interlock and act as a guide for the anterior part of the mandible as the jaws ap- proach closure (and the carnassials begin to func- tion). This is very evident from the wear areas on the canines. The interlocking restricts lateral movement and guides the two blades of the shear very precisely past each other. Xo such arrange- ment exists in such generalized marsupials as the opossum or in generalized insectivores. DAVIS: THE GIANT PANDA 67 ^\ Z i 4 5 67 89 (0 Ailuropoda Fig. 29. Differences of skull proportions in Ursus horribilis and Ailuropoda melanoleuca shown by deformed coordinates. 6. The temporal fossa is large, providing space, and particularly attachment surface, for the large temporal muscle (see p. 150). This fossa is simi- larly large in generalized primitive mammals. The masseteric fossa does not differ significantly from that of primitive mammals. The pterygoid fossa is small or wanting. This fossa is well developed in primitive mammals; its reduction in the Carni- vora is associated with the reduced size and im- portance of the pterygoid muscles. 7. The zygomatic arch is strong and forms a smooth uninterrupted curve in both the sagittal and frontal planes. The anterior buttress of this arch system lies directly over the primary cheek teeth, the posterior buttress over the mandibular fossa — the two sites where pressure is applied dur- ing mastication. The zygomatic arch represents the "main zygomatic trajectory" of Starck (1935) ; it is the principal structure within which are re- solved the disintegrating forces generated by the powerful jaw muscles. The arch is well constructed and extremely powerful in Didelphis. In general- ized insectivores, by contrast, the arch is structur- ally weak: the curvature is interrupted (Erina- ceus), parts of the arch are almost threadlike {Echinosorex, Talpidae), or the central part of the arch is missing (Soricidae). Support for the canines, by contrast, is relatively weak in generalized Carnivora. The main element of this support system is the "vertex trajectory," which in generalized carnivores is weak and often interrupted at the glabella. What, now, has happened to this basic carnivore construction in herbivorous carnivores, and par- ticularly in the purely herbivorous giant panda? The skull is still elongate, but slightly less so than in Canis or Viverra (Table 8). In Ursus the skull is even slightly longer than in Canis or Vi- verra. There is, in fact, little variation in relative skull length among all arctoids examined. Face length in the giant panda is only slightly less than in Viverra, and in the bears it is prac- tically identical with Viverra. Proportions vary among other herbivorous carnivores: the face is very short in Ailurus, of normal length in Procyon. Face length is extremely variable among the Car- nivora in general, and the significance of this vari- ability has not been explored. The face varies independently of the cranium in mammals (p. 72). We may conclude that Ailuropoda and Ursus show no significant differences from the generalized carni- vore condition in longitudinal proportions of the skull. 68 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Ailuropoda Fig. 30. Difference.s of skull proportions in Canis lupus and Ailuropoda melanoleuca shown by deformed coordinates. Depth and breadth of skull, on the contrary, in all herbivorous carnivores depart significantly from the generalized condition (see figs. 29 and 30, and Table 8). Among these, depth is least in Ursus, in which it scarcely exceeds that of the wolf. Depth of skull in Ailuropoda is equaled among carnivores only in Hyaena; in both the panda and the hyena, depth is achieved largely by develop- ment of a high sagittal crest, the inner lamina of the skull roof remaining unaffected. The skull is typically deep in all arctoids that have forsaken a purely carnivorous diet. Increase in depth in- volves only the external lamina of the cranium and adjoining parts of the mandible — not the face or the direct housing of the brain. The vertical height of the posterior half of the zygomatic arch, the area from which the zygomaticomandibularis takes origin, is also greatly increased in Ailuropoda. DAVIS: THE GIANT PANDA 69 Zygomatic breadth is consistently greater than in generalized flesh-eaters, and once again this is maximal in Ailuropoda and least in Ursus. Breadth in the powerful-jawed Hyaena is equal to that in most herbivorous carnivores, but is con- siderably less than in Ailuropoda. We may conclude that breadth and depth of skull are increased in all herbivorous carnivores, and that these reach a maximum in Ailuropoda. Increased breadth and depth of the cranium produce increased volume of the temporal fossa. In Ailuropoda the volume of this fossa has been further increased, especially anteriorly, by crowd- ing the orbit downward from its normal position, by carrying the temporal fossa anteriorly at the expense of the postorbital process and the poste- rior part of the frontal table, and by decreasing the anterior breadth of the braincase. The vol- ume of this fossa probably approaches the maxi- mum that is compatible with normal functioning of surrounding structures. Besides providing space for a greater volume of craniomandibular musculature, increased depth of skull greatly improves efficiency for production of pressure at the level of the cheek teeth. Worth- mann (1922) uses a simplified system of vector analysis to compare relative masticatory efficiency in man and several carnivores. He represents the action of the masseter and temporal muscles by straight lines connecting the midpoints of origin and insertion areas. The axis of the masticatory system is represented by a straight line connecting the center of rotation of the mandibular articula- tion with the last molar tooth. From the structural standpoint, greater depth of skull increases the magnitude of vertical forces that the skull is capable of withstanding. Comparison of masticatory efficiency in a gen- eralized carnivore iCanis) and in the purely her- bivorous Ailuropoda by Worthmann's method re- veals a striking improvement in the panda (fig. 31). In the wolf the axis of the masseter (m) intersects the masticatory axis GK at a point about 30 per cent of the distance from G to K. Thus force at the joint (G) would be to force at the cheek teeth (K) as 7 : 3; in other words joint force is about 2.5 times as great as useful chewing force. In the panda, by contrast, k : fir = 55 : 45 approximately. Similarly for the temporalis k:g = 28:12 for Canis, whereas k : g = 47 : 53 for Ailuropoda. In the cheek-tooth battery emphasis has shifted from the sectorial teeth to the molars (p. 128), and the anterior buttress of the zygomatic arch now lies over the first (Ailuropoda) or second (Ursus) upper molar. This shift, by shortening the resist- ance arm of the jaw lever, increases the mechanical efficiency of the system for production of pressure. The form of the mandibular articulation has not changed — it is still a transverse cylinder rotating in a trough. The extensive horizontal movements of upper molars against lower that characterize other herbivorous mammals are therefore limited to a slight lateral displacement in herbivorous carni- vores. Because of the interlocking canines at the anterior end of the system, no lateral shifting is possible with the teeth in full occlusion.' In Ursus the mandibular articulation is at the level of the occlusal plane as in generalized flesh- eating carnivores. In Ailuropoda the articulation lies considerably above the occlusal plane. Lebe- dinsky (1938) demonstrated that elevating the articulation above the occlusal plane imparts an anteroposterior grinding movement at the occlusal plane, even when the mandible is swinging around a fixed transverse axis. Lebedinsky's interpretation may be analyzed further. Figure 32, A, represents a mandible with the mandibular articulation (0) at the level of the toothrow. A point x on the lower dentition travels through the arc x~x' when the mouth is opened. The tangent to this arc at point x is perpendicular to the occlusal plane, and therefore there is no anteroposterior component in the movement of x with respect to the axis AO, and an object placed between the upper and lower dentitions would be crushed or sheared. This would likewise be true at any other point on the axis AO. Figure 32, B, represents a mandible with the mandibular articulation (0) elevated above the level of the toothrow. A point x travels through the arc x-x' when the mouth is opened, but in this case the tangent to the arc at x forms an acute angle with the occlusal plane, A-B, and there is a very definite anteroposterior component in the movement of x with respect to the axis AB. The angles formed by successive tangents along AB become increasingly acute as B is approached, until at B there is no longer any vertical compo- nent at all. Thus, as Lebedinsky pointed out, any object placed between the upper and lower denti- tions would be subjected to anteroposterior forces even with pure hinge movement of the jaw. More- over, the anteroposterior force becomes increas- ingly great as a point (B) directly beneath the articulation is approached. In Ailuropoda, there- fore, an anteroposterior grinding action is achieved by elevating the articulation, and its effectiveness is increased by extending the toothrow posteriorly. ' In Ailurus fulgens a lateroventral shifting of more than 2 mm., with the cheek teeth in complete occltision, is possible. This is true grinding, otherwise unknown in the Carnivora. Canis Ailuropoda Fig. 31. Relative masticatory efficiency in a generalized carnivore (Canis) and the giant panda (Ailuropoda). The line KG, representing the masticatory axis, connects the center of rotation of the mandibular joint (G) with the midpoint of the functional cheektooth area (K) (boundary between P' and Mi in Canis, anterior quarter of M' in Ailuropoda). The line m represents the axis of the masseter. The line /, the axis of the temporalis, connects the approximate center of origin (T) of the temporalis with the approximate center of insertion (C). The line / may be projected beyond C to K, since a force acting on an immovable system may be displaced in its own direction without altering the result. True masticatory force is represented by k, articular pressure by g. 70 DAVIS: THE GIANT PANDA 71 / / / ; / / / I X a W 1 1 / I \ / \ \ / ( / Fig. 32. Occlusal relations in a mandible with mandibular articulation at level of toothrow (A), and elevated above level of toothrow (B). The lines AO and ABO represent the mandible in occlusion, A'O and A'B'O its position when the mouth is opened. The points x and x' represent the positions of a cusp on one of the lower cheekteeth. (C) Occlusal relations in AUuropoda in biting down at point x on an object 25 mm. in diameter (see text). Stocker (1957) has calculated for the elephant the anteroposterior displacement of a point on the occlusal surface of a lower molar when the jaw is lowered. A similar calculation may be made on a panda skull (fig. 32, C). A point x at the anterior end of the iirst lower molar is 125 mm. from the center of rotation of the mandible, 0. The line xO was found to form an angle, X, of 7° with the oc- clusal plane, x B. The panda is known to chew up bamboo stalks up to 38 mm. in diameter (p. 20) ; to be conservative let us assume a bamboo stalk 25 mm. in diameter. An object 25 mm. in diameter placed between the upper and lower teeth at the level x displaces point X on the lower molar to position x'. The two lines x 0 and x' 0 form an angle, a, of 11° 30'. The horizontal displacement, j M, of x with re- spect to the occlusal plane may be calculated as a;M=a:OcosX-a:Ocos (X-|-a) = xO [cos X -cos (X-|-a)] Substituting the values given above, this equation gives a value for x M of 5.4 mm., which is the horizontal distance through which a point x on the lower molar travels as the teeth are brought into occlusion. This represents the anteroposte- rior grinding component that would be brought to bear on the bamboo stalk. 72 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 The mandibular symphysis remains unfused in Ursiis and othei- herbivoi'ous carnivores, although the two halves interlock so intimately that no movement is possible. Its fusion in Atluropoda reflects the general increase in bone tissue that characterizes the skull as a whole. We may conclude that the skull of Ailuropoda represents an attempt to adapt the carnivore type of skull — already highly adapted for seizing and cutting — to the radically different requirements of grinding siliceous plant fibers. Efficient grinding requires horizontal movements, but these are al- most completely inhibited by the cylindrical man- dibular articulation and the interlocking of teeth during occlusion, although Ailurus shows that effective horizontal grinding can be achieved in a carnivore. A compromise solution was to replace the unattainable mechanical efficiency seen in true herbivores with more power. This attempt to achieve maximal power in the masticatory equip- ment is the key to the architecture of the panda skull. The skulls of other more or less herbivorous car- nivores except Ailurus exhibit most of the modi- fications seen in Ailuropoda, but to a much less extreme degree. Thus the skull of Ailuropoda may be considered an ultimate expression of adaptation for herbivory within the Carnivora. What can be deduced of the morphogenetic mechanisms whereby these results were achieved — in other words, the mechanism through which natural selection had to operate? To what extent does the skull of Ailuropoda as compared with that of Ursus merely reflect extrinsic mechanical fac- tors arising from the massive musculature, and to what extent intrinsic factors, other than the ability of the bone to respond to mechanical stress? Some anatomists have recently attempted to re- examine the mammalian skull from an analytical rather than a purely descriptive standpoint (see Biegert, 1957, for a review). In these studies the skull is regarded as merely the bony framewoi-k of a major functional unit, the head. During on- togeny and phylogeny there is a complex interplay among the various organs making up the head, and the skull adapts itself to the changing spatial, mechanical, and static demands. In a given phylo- genetic sequence one of the head organs (e.g., brain, feeding apparatus, eyes) typically comes to dominate the whole and sets the pattern, so to speak, for further evolution within the group. Changes in the skull are thus not simply additive, but are a function of changes in other head organs, which in turn may be functionally irreversible and therefore in effect "fix" the pattern of further evo- lution within the group. The causal factors that determine changes in skull form are interpreted as an interplay between the inherited basic plan of the skull and the demands of other head structures extrinsic to the skull itself. This approach iso- lates some of the forces modeling the skull, but in the end it does little more than describe struc- tural correlations. It fails to come to grips with the problem of the mechanics of evolution. Correlation studies have shown that the facial part of the skull varies as if it were genetically distinct from the cranium, as it is in fact phylo- genetically (Cobb, 1943; and especially Starck, 1953, for a review). This genetic independence, and the further independence of the mandible, have been proved in breeding experiments on dogs (Stockard, 1941 ; Klatt, 1941^3) . Such independ- ence means that a genetic factor affecting the ontogenetic growth rate of the cranium (or a com- ponent of the cranium) need not affect the face, and vice versa. The union between face and cra- nium, however disparate these structures may be, is maintained by mutual accommodation during growth. Genetic control of growth rates in dental fields is well known to be distinct from that of any other part of the skull. Numerous observations (e.g., Cobb, 1943) indicate that the alveolar areas of the skull accommodate directly to the space re- quirements of the teeth during the gi-owth process. The mammalian skull, in short, is a mosaic of independent morphogenetic units that are fitted into a functional unit partly by natural selection acting on their several time-tables of gi'owth and differentiation, and partly by accommodation to extrinsic forces. The extent of the morphogenetic units may vary with time during ontogeny: the earlier in ontogeny a genetic effect is manifested, the more extensive its target is likely to be. A beginning has been made at identifying and iso- lating these morphogenetic units (Starck, 1953; Landauer, 1962), but they are still inadequately known. Thus, in considering the morphosis of the skull, two sets of factors must be kept in mind. These are the location and extent at any moment during ontogeny of the morphogenetic units of which the skull is composed (intrinsic to the skull), and the modeling effects on the skull of other head struc- tures (extrinsic to the skull as such). In the skull of Ailuropoda the increase in quan- tity of compacta is clearly limited to two major morphogenetic units, the cranium and the man- dible, and absent in a third, the face. The hyper- trophy of bone substance affects not only the skull, but all compacta in the body in a gradient falling off from the dorsal body axis, and including struc- tures such as the tail and the proximal ends of the DAVIS: THE GIANT PANDA 73 ribs where hypertrophy can scarcely represent structural adaptation. We do not know the time- table of mammalian ontogeny in enough detail to know whether these effects could have been pre- dicted and delimited a priori. The additional bone substance certainly strengthens the skull, although it is not distributed along trajectory lines of the skull as it should be if it were primarily functional. We cannot say whether increased bone substance in the skull of Ailuropoda was a primary target of natural selection, whether it is genetically linked with increase in the mass of the masticatory mus- cles, or whether it simply reflects disturbed meta- bolic or endocrine relations. Cephalization in bulldogs is in some respects similar to but less extreme than in Ailuropoda. Klatt and Oboussier (1951) found that all struc- tures of the head (skull, masticatory musculature, brain, eyeballs, hypophysis) are heavier in bull- dogs than in "normal" dogs. These authors con- clude that the bulldog condition results from an increase in the growth rate of the anterior end of the embryo. More likely it represents a temporary intensification of the general growth rate of the embryo during the period when the head region is undergoing its most rapid growth. The effects are less generalized in Ailuropoda; here the brain and eyeballs (and the internal ear) are of "nor- mal" size, a condition that would result if the ontogenetic growth rate were increased after the central nervous system and its sensory adnexa had experienced their period of most rapid growth. The condition in the panda is, in fact, the reverse of the condition in man, where the brain is en- larged while all other cranial (but not facial) structures are of "normal" size. As interpreted by Weidenreich (1941), in man the ontogenetic growth rate is temporarily intensified during the period when the brain is undergoing its most rapid growth, and returns to normal before the rapid growth period of other cranial structures is reached. It is known from comparative studies that sur- face relief of the mammalian cranium is deter- mined chiefly by the craniomandibular muscles (Weidenreich, 1922). The developing cranium is, as Anthony (1903) put it, molded between the brain and the masticatory musculature. Direct evidence of the role of the cranial muscles in de- termining skull form in mammals is limited to the effects of unilateral paralysis or removal of mus- cles in young rats, rabbits, guinea pigs, and dogs. Unilateral paralysis of the facial muscles (Wash- burn, 1946a), removal of one masseter (Horowitz and Shapiro, 1955, and earlier workers), of one temporal (Washburn, 1947, and earlier workers), or of neck muscles (Neubauer, 1925), all resulted in asymmetrical development of the skull, with failure of associated bony crests and ridges to form. Removal of the temporal was followed by resorp- tion of the coronoid process but did not alter the internal form of the braincase. No one has re- moved simultaneously the temporal, zygomatico- mandibularis, and masseter from one side to deter- mine the part played by these major muscles in determining the form of the zygomatic arch; it is very probable that bizygomatic breadth is inti- mately related to these muscles. These experiments were performed far too late in ontogeny to provide the intimate knowledge of the factors of embryogenesis we have for the limb bones of the chick (Murray, 1936). So far as they go, the experiments strongly reinforce the observa- tional data of comparative anatomy. Practically nothing is known of the development of the form of the skull, but from what is known of develop- ing limb bones in vertebrates (Murray, 1936; La- croix, 1951) the primary form of both dermal and cartilage bones of the skull is probably determined by intrinsic growth patterns, whereas modeling is determined by pressures and tensions extrinsic to the bones, created by musculature, brain, sense organs, vessels and nerves, and mechanical inter- action between the developing bones themselves. We may assume that, except for differences result- ing from increase in volume of bone tissue, the considerable differences in form between the skull of the panda and that of the bears are largely, perhaps almost entirely, dependent on such ex- trinsic factors — that of the cranium on the muscu- lature, and that of the face on the dentition. The only features for which intrinsic factors must be postulated appear to be the tremendous increase in the bone substance making up the skull (by proliferation of connective tissue) and the ele- vation of the mandibular articulation (by prolifer- ation of cartilage). Elevation of the articulation enhances horizontal movements of the mandible. It occurs in some degree in all herbivorous mam- mals and surely is a direct result of natural selec- tion operating on the skull. The morphogenetic mechanism whereby it is achieved is unknown, but the fundamental similarity to the acromegalic mandible suggests that it is simple. We may conclude that no more than four, and perhaps only three, factors were involved in the transformation of the ursid type of skull into that of Ailuropoda. Two of these — hypertrophy of jaw musculature and dentition — are extrinsic to the skull and therefore involve only the ability of the bone to respond to mechanical stress. Two — general hypertrophy of bone substance and ele- vation of the mandibular articulation — are intrin- sic to the skeleton but involve different growth 74 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 mechanisms. Thus only two factors acting di- rectly on the skull itself may distinguish the skull of Ailuropoda from that of other ursids. Natural selection has no doubt had additional minor polish- ing effects, although the whole morphology of the giant panda indicates that the morphological in- tegration produced by such refined selection is at a relatively low level. E. Summary of Skull 1. The skull of Ailuropoda is basically similar to that of Ursus. Agreement is particularly close in structures relatively unaffected by masticatory requirements: the turbinates, the paranasal sinuses, the middle ear, and the inner lamina of the cranial cavity. 2. The outer lamina of the cranium and the mandible are remarkable for the thickness and density of the bone. This greatly exceeds mechan- ical requirements, and therefore is not directly adaptive. 3. All parts of the skull associated with the masticatory apparatus are greatly expanded. The volume of the temporal fossa in particular, espe- cially its anterior third, has been increased at the expense of surrounding structures. Similar adap- tive changes appear convergently in Ailurus and, in slightly altered form, in hyenas. 4. From the genetic standpoint these adaptive changes are probably extrinsic to the bone itself, involving only the ability of the bone to respond to mechanical forces during ontogeny. 5. The only obvious intrinsic factors are the great increase in bone tissue in the cranium and mandible, and the elevation of the mandibular articulation above the occlusal plane. 6. Thus only two major factors acting directly on the skull itself may distinguish the skull of Ailuropoda from that of Ursus. 7. Certain features usually regarded as diag- nostic of the Ursidae (e.g., by Flower, 1869) have been obliterated in Ailuropoda by the expansion of the masticatory apparatus. Among these are postorbital processes on frontal bones, presence of alisphenoid canal, non-confluence of foramen ro- tundum and orbital fissure, and presence of fora- men lacerum medium. Such secondary differences cannot be used as evidence of non-relationship be- tween the panda and the bears. IL THE VERTEBRAL COLUMN A. The Vertebral Column as a Whole The vertebral column of the giant panda is in many respects the most remarkable among living carnivores. Slijper (1946) showed that the archi- tecture of the developing column is responsive to the mechanical demands of posture and locomo- tion. Morphogenetically the mammalian column behaves like other homiotic structures (Kiihne, 1936; Sawin, 1945, 1946). Therefore it is prefer- able to consider the column as a whole, rather than as a chain of independent units. The analytical study of the vertebrae of the Carnivora made by Stromer von Reichenbach (1902) showed that the morphological details of individual vertebae ex- hibit no important features consistently correlated with the major categories, and are therefore of little systematic importance. For this reason no detailed description and comparison of individual vertebrae of Ailuropoda is presented here. The number of presacral vertebrae is extremely constant in carnivores. The normal number of thoraco-lumbars in all living Carnivora is twenty, and individual variations rarely exceed one above or below this figure. The giant panda is conse- quently remarkable in having only eighteen trunk vertebrae; in one of nine skeletons this number was further reduced to seventeen, and in one there were nineteen (Table 9). The number of lumbar vertebrae in Ailuropoda is five in 50 per cent of the cases, and four in the remaining 50 per cent; in Ursus it is six in 79 per cent, and five in the remaining 21 per cent. (Other genera of the Ursidae appear to differ from Ursus, but the samples are too small to permit conclu- sions.) The modal number of lumbars is either four or five in Ailuropoda, and six in Ursus; the mean is 4.5 and 5.8, respectively, indicating that the lumbar region has been reduced by more than one vertebra in Ailuropoda. The thoracics show a similar but somewhat more limited tendency toward reduction: the mean is 13.5 in Ailuropoda, 14.2 in Ursus. There was evidence of disturbance at the cervico-thoracic boundary in one individual (p. 85). Thus in the column as a whole there is an anterior displacement of the boundaries of the several regions in Ailuropoda, and this displace- ment shows a gradient decreasing in intensity from the sacrum toward the head. A remarkable feature of the column in Ailuro- poda is its variability. Of nine skeletons examined, the thoraco-lumbar juncture was asymmetrical on the two sides of the body in three, and four differ- ent vertebral formulae are represented among the remaining six individuals (Table 9). This varia- bility is greater than was found in any of the nu- merous arctoid and ailuroid carnivores examined. The proportions of the three main divisions of the column in Ailuropoda differ from those in other carnivores, as shown below. These proportions also show a far greater range of variation than in DAVIS: THE GIANT PANDA 75 Table 9.— VERTEBRAL COUNTS IN CARNIVORES Number of indi- viduals Cants latrans 15 Canis lupus / 9 Vulpes fulva / 9 Uroeyon cinereoargenteus I , Bassariscus astutus J j 1 Nasua narica 1 Nasua nasua 5 fll Procyon lotor J 2 I 1 f ^ Bassaricyon alleni J 1 I i Ailurus fulgens^ 5 Ursus (various species)^ 7 Ursus^ 2 C7rsus' 1 (2 Ailuropoda melanoleuca . 1 3 ? ' One record from Flower (1885). ' Three records from Flower (1885). any other carnivore examined. The cervical re- gion is shorter in Ailuropoda than in Ursus but is only slightly shorter than in Ailurus and Nasua and no shorter than in Procyon. The thoracic re- gion is relatively longer than in any other arctoid carnivore, resembling that of burrowing mustelids. The lumbar region is short in both Ailuropoda and Ursus. The proportions of the vertebral colurnn Thoracics Thor- + lum- acics Lumbars bars 13 7 20 13 7 20 14 7 21 13 7 20 13/14 7/6 20 13 7 20 13 6 19 13 7 20 13 6 19 13 5 18 15 5 20 15 5 20 14 6 20 15 0 20 14 0 19 13 6 19 13 7 20 14 7 21 14 6 20 14/13 6/7 20 14 6 20 14 6 20 15 5 20 14/15 6/5 20 14 5 19 14 4 18 14/13 4/5 18 13 5 18 13 4 17 of the giant panda are similar to those of the an- thropoid apes and man, and to those of such bur- rowing carnivores as Taxidea, Meles, and Mellivora — columns designed to withstand anteroposterior thrust. The vertebrae of Ailuropoda are heavier than in Ursus; the weight of thoraco-lumbar vertebrae is about 16 per cent greater in a specimen of the panda than in a black bear of comparable size. The Mechanics of the Vertebral Column The vertebral column of mammals, with its as- sociated muscles and ligaments, is an extremely complex mechanism that has never been satisfac- torily analyzed. Yet it is only on the basis of its functioning that the differences, often extremely subtle, exhibited in this region from animal to ani- mal can be intelligently considered. Slijper (1946) made a painstaking comparative study of the col- umn in mammals in an effort to correlate mor- phology and function. Many of his findings are relevant in the present connection. Slijper rejects former comparisons of the verte- bral column with an arched roof, a bridge with parallel girders, or a cantilever bridge, and com- pares it with a bow flexed by a bow-string (the sternum, abdominal muscles, and linea alba). Vertebral Bodies. — Slijper points out that the principal static function of the column is to resist bending, chiefly in the sagittal plane, and that differences in the size and shape of the vertebral bodies reflect the forces acting on them. He used as a criterion of the stress to which any part of the column is subjected the moment of resistance to bending, which he computed for each vertebral body by using the formula: breadth of articular face of body X square of height of body (bh-). Plotting these data for the entire column in a series of mammals yields characteristic curves of the moments of resistance at successive points along Table 10.— RELATIVE PROPORTIONS OF DIVISIONS OF THE VERTEBRAL COLUMN IN CARNIVORES' N Cervical (%) Thoracic (%) Lumbar (%) Canis 2 30 39.5(39-40) 30.5(30-31) Vulpes 2 28.8 (28-29.5) 39.8 (39.5-40) 31.5 Bassariscus 1 24 42 34 Nasua 1 22 45 33 Procyon 4 21.4 (21-22) 47.7 (47-48) 30.9 (30-31.5) Ailurus 3 22 (21.3-22) 47 (47-47.5) 31 (31.0-31.2) Ursus 3 26.2 (25.5-27.4) 45.9 (45.6-46.3) 27.9 (26.3-28.9) Ailuropoda 6 22 (21-23.2) 55 (51.7-59) 23 (20-26) Taxidea 2 23 50.5 (50-51) 26.5 (26-27) Meles 1 25 53 22 Mellivora 1 25 56 19 ' Ursus and Ailuropoda determined on disarticulated skeletons. 76 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 ITh — r- IL B ITh 1^ IL Fig. 33. Diagrams of moments of resistance in the vertebral columns of various mammals: A. Moments of resistance in a beam supported at one end. B. Slijper's Type 16 curve of moments of resistance in the vertebral column of mammals with an erect or semi-erect posture (compare with A and fig. 34). C. Theoretical moments of resistance in quadrupedal mammals, in which the vertebral column is compared to a bow, with a beam supported at one end attached to the cranial (left) end of the bow. D. Slijper's Type II curve of moments of resistance in the vertebral column, characteristic of carnivores other than bears and Ailuropoda (compare with C). the column. Slijper divides these curves into three major types, each with several subtypes. Of the few carnivores examined by Slijper (Ca- ms, Vulpes, Ursus, Felis, Panther a), all except Ursus yielded curves of Type II, characterized by a hump in the posterior cervical region, and a flat anterior thoracic region, followed by a rise in the posterior thoracic and lumbar regions (fig. 33, D).' For Ursus the curve slopes upward gradually from the anterior cervical region to about the tenth thoracic, then abruptly breaks more steeply up- ward, sloping downward again in the posterior lumbar region. This is Slijper's Type lb curve, characteristic of bipedal animals, including man (fig. 33, B). The curve for Slijper's bipedal goat was also modified in this direction. This type of curve agrees closely with the diagram of the theo- retical moments of resistance if the column is re- garded as an erect or semi-erect beam supported at one end (fig. 33, A). The curve of the moments of resistance for Ai- luropoda was plotted for two individuals, which showed only minor differences (fig. 34) . This curve is very similar to that for Ursus, differing chiefly ' Slijper lists the domestic cat (along with the bear and the anthropoid apes and man) as having a Type lb curve. This is obviously a mistake. I have measured and plotted a disarticulated cat column, and find that it has a typical Type II curve. in its more even slope without the sharp upward break at the level of the diaphragmatic vertebra (eleventh thoracic in Ursus, eleventh or twelfth in Ailuropoda). In this respect Ailuropoda resem- bles the anthropoid apes and man more closely than Ursus does. It is evident that the vertebral axis in the bears, and especially in the giant panda, is constructed to withstand anteroposterior thrust. Neural Spines. — The length and angle of in- clination of the neural spines do not depend upon the static demands made upon the column, but upon the structure and development of the epaxial muscles that attach to them (Slijper). Thus the structure of the spines is ultimately determined by posture and locomotion, plus such secondary fac- tors as absolute body size, length of neck, and weight of head. Both length and inclination of a spine are resultants of the several forces exerted by the muscles attaching to it, the spine acting as a lever transmitting the muscle force to the ver- tebral body. Plotting the lengths of neural spines as percent- ages of trunk length permits comparison of the resulting curves for various animals. These curves apparently follow a common pattern in all mam- mals, although the relative lengths of the spines DAVIS: THE GIANT PANDA 77 SOr 40 30 SO bh« n4«IO* 36759 ■ Ailuropoda melanoleuca 3«I0< 2«I0* C'3 Th-l 10 L-l Fig. 34. Curve showing height (h), breadth (6), and moments of resistance (b/i') in the vertebral column of Ailuropoda. D = diaphragmatic vertebra. vary greatly from species to species. The spines are longest on the anterior thoracic vertebrae (at- tachment of cervical muscles and ligaments), de- crease in length back to the anticlinal or diaphrag- matic vertebra, and are slightly longer again on the posterior thoracic and lumbar vertebrae (at- tachment of longissimus and spinalis muscles). Both Ursus and Ailuropoda exhibit this type of curve, although in both forms the spines are rela- tively short along the whole length of the column (fig. 35). The inclination of the spines conforms less closely to a common pattern than does the height. Ac- cording to Slijper the direction of a given spine tends, for mechanical reasons, to be perpendicular to the most important muscle inserting into it. The spines of the pre-anticlinal (or pre-diaphrag- matic) vertebrae are inclined posteriorly in all car- nivores, as they are in all mammals. Among the arctoid Carnivora the post-diaphragmatic spines are inclined anteriorly in the Canidae and Procy- onidae, are variable among the Mustelidae (from an anterior inclination of 45° in the martens to a slight posterior inclination in the skunks and badgers), and are posteriorly inclined or at most vertical in the Ursidae. In Ailuropoda all the post-diaphragmatic vertebrae are posteriorly in- clined, the minimum inclination in two skeletons being 20° (fig. 36). According to Slijper the direc- tion of the post-diaphragmatic spines in Carnivora and Primates is determined chiefly by the length of the vertebral bodies, because the angle of at- tachment of the multifidus muscle depends upon this length. The bodies of the lumbar vertebrae are short in both giant panda and bears, but they are not notably shorter in Ailuropoda than in Ur- sus, although the posterior inclination of the spines is much greater. Thus, other factors must be in- volved in Ailuropoda. It is at least suggestive that among the primates and burrowing mustelids a pos- terior inclination of the post-diaphragmatic spines is associated with anteroposterior thrust along the column. 78 FIELDIAXA: ZOOLOGY MEMOIRS, VOLUME 3 % of length of trunk 13 LENGTH OF NEURAL SPINES \Canis (from Slijper) "Ailuropoda 2 3 Vertebrae K) n 12 13 Fig. 33. Curves showing lengths of neural spines in AUnropoda, Ursus arHof, and Canig familiaris. B. DE:scRipnoNS of Vertebrae 1. Cervical Vertebrae The cervical vertebrae in Ailuropoda are remark- able for their breadth, which gives the cervical region a compressed appearance, especially when viewed from below. Transverse broadening is evi- dent on all vertebrae including the atlas and epi- stropheus, and greatly exceeds that in any other land carnivore. The vertebrae are shorter antero- posteriorly than in the long-necked Ursus, but are no shorter than in Proeyon and Ailurus. There are seven cervicals in each of the eight skeletons examined. Except for the distortion resulting from broad- ening, the cervicals differ little from those of other carnivores. The atlas is similar to that of Ursus in the arrangement of foramina; in both there is an alar foramen < vertebral artery and vein>, in- stead of a mere notch as in other arctoids, into which open the atlantal foramen dntervertebral of authors; transmits first spinal nerve and verte- DAVIS: THE GIANT PANDA 79 Degrees 30 INCLINATION OF NEURAL SPINES Ailuropoda I 2 3 4 5 6 7 Thoraco Lumbar Vertebrae 20 21 Fig. 36. Curves showing inclination of neural spines in Ailuropoda, Ursus arctos, and Canis familiaris. bral artery) and transverse foramen (vertebral ar- tery and vein). The foramina on the atlas are crowded together as compared with Ursus (fig. 37). The transverse diameter across the wings is greater than in Ursus, but the wings are narrower antero- posteriorly. The third to sixth cervicals are notable chiefly for the conspicuous, backwardly directed hypera- pophysis (Mivart) atop each postzygopophysis; these are barely indicated in Ursus, and are want- ing in other arctoids. The spines are nearly obso- lete on the third, fourth, and fifth cervicals, but are of normal length on the sixth and seventh. 2. Thoracic Vertebrae The thoracic region in Ailuropoda is notable for its length. Since the number of thoracic verte- brae averages about one less than in Ursus, the gi-oater thoracic' length must be attributed to longer centra on individual vertebrae, but I have been unable to demonstrate this satisfactorily. There is, of course, no anticlinal vertebra in Ailuropoda, since the neural spines all slope in the same direction. A true anticlinal is also wanting in Urstis for the same reason. The diaphragmatic vertebra is that transitional vertebra on which the prezygapophyseal facets look upward (horizontal), while the postzygapophyseal facets look outward (vertical or oblique) . The diaphragmatic vertebra is the eleventh thoracic in one specimen of Ailu- ' This length of thorax is approached or even exceeded in some burrowing mustelids, e.g., Taxidea, Mephitis, Melli- vora. In these forms, however, the thoracic region has taken over the anterior lumbars, and the thoracic count is 1 6 or 1 7. For. atlantis For. alare Ala atlantis For. Iransversarium B For. transversarium' Corpus epistropheus Ailuropoda Ursus americanus Fig. 37. Cervical vertebrae of Ailuropoda and Ursus. A, atlas from below; B, epistropheus and third cervical from left side. anterior lateral Fig. 38. Fifth thoracic vertebra of Ailuropoda. posterior 80 DAVIS: THE GIANT PANDA 81 ropoda, the twelfth in another. It is the eleventh in Ursus. It is uniformly Th. 10 in the Canidae. The Procyonidae vary: Bassariscus, Th. 10; Bas- saricyon, Th. 10; Ailurus, Th. 11; Procyon and Nasua, Th. 12. There are fewer lumbar vertebrae (an average of 4.5 in the eight skeletons examined) than in any other arctoid carnivore. ^ The lumbar spines all slope posteriorly; this is not encountered in any other arctoid, but is approached in Ursus. Ailuropoda UrsKs americanus Fig. 39. Third lumbar vertebra of Ailuropoda and fourth lumbar of Ursus, seen from the left. There are few significant differences in morpho- logical details. The intervertebral foramina (spi- nal nerves and vessels) are conspicuously larger than in Ursus, owing chiefly to the larger size of the posterior vertebral notch. The width across prezygapophyses and postzygapophyses is much greater in Ailuropoda than in Ursus and other arctoids, which should contribute to the stability of this region. The spines are capitate, especially on the anterior vertebrae. Their posterior bor- ders are less produced than in Ursus, and their lateral surfaces present prominent muscle rugo- sities that are lacking in other arctoids. 3. Lumbar Vertebrae The lumbar region is shorter than in any other arctoid carnivore examined. It is short in burrow- ing mustelids (Meles 22 per cent, Mellivora 19 per cent, but Taxidea 26-27 per cent) and hyenas (18- 20 per cent). The length relative to the total col- umn is not much greater in Ursus than in Ailuro- poda (see Table 10) , but because of the long neck in bears this does not properly reflect the true short- ness of the lumbar region in Ailuropoda. The absolute length of the lumbar region in Ailuropoda is only 165-180 mm. (32-33 per cent of thoraco- lumbar length), while in a bear of comparable size (Ursus americanus) it measures 233 mm. (38 per cent of thoraco-lumbar length). The form of the vertebrae is similar to that in Ursus. The centra are very short in both. As with the thoracics, the intervertebral foramina are larger, and the pre- and postzygapophyses are wider than in Ursus. The lumbar spines in both the giant panda and the bears are short and stumpy, and are either ver- tical (Ursus) or posteriorly inclined (Ailuropoda). Slijper believes that the vertical position of the spines in Ursus is correlated with the shortness of the lumbar centra, which results in greater me- chanical efficiency in the longissimus and multifi- dus muscles attaching to them. The transverse proces.ses are not well developed in either Ailuropoda or Ursus. In both they are relatively short, and directed transversely instead of anteriorly as in other arctoids. These processes provide attachment for the ilio-costal and quad- ratus lumborum muscles, which function in exten- sion and flexion of the column and hence are important in movements of the back during run- ning. Anapophyses (accessory process of Reighard and Jennings and Baum and Zietzschmann) are pres- ' In some of the burrowing mustelids (Arctonyr, Cone- paius, Mellivora) four is apparently the normal number of lumbars. In these, however, the number of thoracics is cor- respondingly increased, and the thoraco-lumbar count is 20 or 21, the typical carnivore formula. The curve of the moments of resistance is also altogether different. 82 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 ent on the first two lumbars, are barely indicated on the third, and are obsolete on succeeding verte- brae. Ursus is practically identical. These proc- esses are present on all lumbars except the last in Procyon and Nasua, and on all but the last two in Four pelves of Ailuropoda were available for de- tailed examination. Three full vertebrae are in- volved in the sacro-iliac joint in two, and two and a part of the third are involved in two individuals. In one sacrum articulating by three full vertebrae. Proc. spinosus Praezygapophysis MC— Postzygapophysis Anapophysis Proc. tramrersus Ailuropoda Ursus Fig. 40. Second lumbar vertebra of Ailuropoda and Ursus americanus, seen from the rear. other procyonids and Canis. They provide inser- tion for the tendons of the longissimus muscle, which functions in extension and flexion move- ments of the vertebral column. 4. Sacral Vertebrae The sacrum is composed of five fused vertebrae in all eight skeletons of Ailuropoda examined. As will be seen from the accompanying table, Ursus is remarkably varied in this respect, although the most frequent number is likewise five. In all other arctoid carnivores the normal number of sacrals is three. (Sacrals are reckoned, according to the definition of Schultz and Straus, as "the vertebrae composing the sacrum and possessing interverte- bral and sacral foramina ringed completely by bone in the adult.") Number of Sacral Vertebrae Canis latrans Canis lupus Vulpes fulva Urocyon cinereoargentetis . Bassariscus astutus Nasua narica Nasua nasua Procyon lotor Bassaricyon alleni Ailurus fulgens* Ursus sp.** Ailuropoda melanoleuca. . S 14 10 10 5 7 1 5 13 6 5 1 * One record from Flower. ** Six records from Flower. the first sacral has the appearance of a transformed lumbar — well-formed pre- and postzygapophyses, enormous sacral foramina, incomplete fusion of the centra ventrally — although on the basis of the total column it is numerically equivalent to the first sa- cral of the second individual. This is of interest in connection with the reduced number of thoraco- lumbars in Ailuropoda, and the extraordinary in- stability of the thoraco-lumbar boundary. It is further evidence of the genetic instability of the posterior part of the vertebral column in this species. In the primary condition in arctoids, as seen in Canis, Bassariscus, and Nasua, the sacro-iliac ar- ticulation is restricted almost entirely to a single vertebra, the first sacral. In Procyon and Urstts the articulation is more extensive, including the first two sacrals, while in Ailuropoda it reaches its maximum among the arctoid carnivores with the third vertebra participating more or less com- pletely. It is interesting and suggestive that the increase in length of sacrum and extent of sacro-iliac artic- ulation among the Carnivora is paralleled among the Primates. The figures given by Schultz and Straus (1945) show that the number of sacrals in- creases abruptly in the anthropoid apes and man over the number found in other Primates (except the aberrant Lorisinae). Examination of a series of primate skeletons shows that the extent of the sacro-iliac articulation is likewise increased in the bipedal apes and man. DAVIS: THE GIANT PANDA 83 Praezygapophysis Arnis rertebrae — Proc: transrersus PostzygapophysU lsl,2nd & 3rd Caudals Isl.Znd & 3rd Caudals lst,gnd & Srd Caudals 6th Caudal 6th Caudal 6th Caudal 1st Caudal (anterior) Ailuropoda 1st Caudal (anterior) Ursns 1st Caudal (anterior) Procyon Fig. 41. Caudal vertebrae of Ailuropoda, Ursus americanus, and Procyon lotor. First three caudals, dorsal view; sixth caudal, dorsal view. The morphology of the sacrum in Ailuropoda is similar to that of Ursus but differs in a number of respects. The long axis of the bone is nearly straight in the panda, while in the bears it is slightly curved ventrad. In the panda the sacrum, like the remainder of the vertebral column, ap- pears to be expanded laterally and depressed dorso- ventrally. The spines are fused to form a contin- uous median sacral crest, which forms a peak on the first sacral and becomes nearly or quite obso- lete on the fifth. The intervertebral foramina are minute, irregular, and nearly obliterated. There are four pairs of dorsal sacral foramina (dorsal divisions of sacral nerves, branches of lateral sa- cral arteries). The first two pairs are irregular, often small and almost obliterated as a result of bone growth in connection with the sacro-iliac an- kylosis. The last two pairs are larger and more regular. The four pairs of ventral sacral foramina (ventral divisions of sacral nerves, branches of lateral sacral arteries) are much larger and more regular than the dorsal foramina. 5. Caudal Vertebrae The tail is short and almost vestigial, but neither as short nor as degenerate as in the bears. Nowhere is the shortening and dorso-ventral flattening of all the vertebrae of Ailuropoda more apparent than in the tail. All the caudals are heavy and stocky; even those toward the tip of the tail lack the slender rod-like form characteris- tic of other carnivores. This is undoubtedly to be interpreted as a gratuitous extension of the factors influencing the remainder of the column, since in Ailuropoda, as in the bears, the tail is functionless. The tail is composed of eleven vertebrae in the one specimen in which it is complete. This is within the range of variation of Ursus, in which there are eight to eleven or more vertebrae. Other arctoids have much longer tails, with from eighteen to twenty or more vertebrae, each of which is rela- tively much longer than in Ailuropoda or Ursus. The first two caudals are well formed in Ailur- opoda, with complete neural arch but no neural spine, wide transverse processes, and prezyga- pophyses; postzygapophyses, which are present in other arctoids except Ursus, are wanting. On the first vertebra the transverse processes extend the entire length of the centrum, and even ante- riorly beyond the centrum onto the prezygapophy- sis. There are no chevron bones. In Ursus, in 84 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 contrast, the neural arches are wanting on all caudals (U. americanus) or are present on only the first vertebra, and the transverse processes are almost completely obsolete even on the first cau- dal. Chevron bones are wanting in the bears. Viewed from the front, the first caudal exhibits to a striking degree the dorso-ventral flattening of the vertebrae (fig. 41). The remaining caudals are short and stocky, ex- hibiting less of the typical rod-like form than is seen even in Ursus. The broadening effect is evi- dent at least back to the seventh vertebra, the transverse processes becoming entirely obsolete on the eighth. C. Review of the Vertebral Column The contrast between Gadow's explanation of the evolution of the vertebral column (1933) and that of Slijper (1946) is a measure of the altered point of view with respect to this complex struc- ture. To Gadow the column is a series of discrete entities, each with its own almost independent phylogenetic history. A lumbar vertebra is fun- damentally a lumbar, regardless of whether it has been "transformed" into a thoracic in one instance or a sacral in another. The functioning of the column, as well as mechanisms by which observed differences could have been achieved, are ignored. The goal is to discover the "true homologies" of elements — a goal that, with respect to the verte- brae, we now know is largely a will-o'-the-wisp. This is the classical outlook of many of the older comparative anatomists. Slijper, on the other hand, has regarded the col- umn, along with its muscles and ligaments, as an architectural construction responsive to the me- chanical demands of posture and locomotion. He has tried to determine correlations between struc- ture and function under varying conditions. Ho- mologies are not considered. His work is essentially an engineering study. Neither Gadow nor Slijper considered the ques- tion of how, from the standpoint of evolutionary mechanisms, the differences they observed could have been brought about. Studies by Kiihne (1936) and others on the inheritance of variations in the human vertebral column showed that dif- ferentiation of the column, like that of other homi- otic structures, is genetically controlled as a series of fields or gradients of differentiation and growth. These fields correspond to the thoracic, lumbar, and sacral regions of the column. The anlage of a vertebra is indifferent; its differentiated form depends on its position in a particular field. There is also a general cranio-caudal gradient of differen- tiation; so increasing the tempo of development would shift the boundaries of all regions cranially, and vice versa. Kuhne emphasized that all dis- placements were always in the same direction in a given individual. Moreover, "besides the trunk skeleton, the field of action embraces the periph- eral nervous system (limb plexuses), musculature, blood vessels, and a large part of the organs of the thoracic and abdominal cavities" (Kiihne). Kiihne concluded that all the variations he observed could be explained by assuming a single pair of alleles, "craniad" and "caudad." These deductions were verified experimentally by Sawin (1945, 1946), who concluded from breeding experiments on rab- bits that displacements of the boundaries of verte- bral regions are determined primarily by a single pair of genes. Among the arctoid carnivores the thoracolum- bar boundary is shifted caudad in the Procyonidae (except the primitive Bassariscus) and Ursidae. The functional significance, if any, of this shift is unknown. It did not affect the number of thoraco- lumbar segments, which remain at the typical 20. In Ailuropoda the thoracolumbar boundary is vari- able, but obviously has been shifted cranially from its position in the Ursidae. The lumbosacral boundary has likewise been shifted cranially two to three vertebrae from its typical position in arc- toid carnivores. Thus in Ailuropoda, as in the higher primates, there is a general cranial displace- ment in the regional boundaries of the column. In both the panda and the higher primates this cranial shift is associated with intense differentia- tion in the anteriormost part of the body axis — the head. In both cases this "cephalization" rep- resents an increase in the tempo of differentiation or growth, although very different tissues are in- volved. Because of the axial gradient, the cepha- lization is accompanied by a cranial shift in the boundaries of the regions of the column. Conse- quently, shortening of the column and displace- ment of its regional boundaries in Ailuropoda (and probably also in the higher primates) are not themselves adaptive, but are consequential results of a process of cephalization. In bulldogs, which are likewise characterized by cephalization, Klatt and Oboussier (1951) reported malformations of the vertebral column (but no reduction in num- ber of vertebrae) in about 80 per cent of their specimens. The vertebrae are also broadened and depressed in Ailuropoda in comparison with Ursus and other carnivores. There is no way of determining how much this is due to secondary postnatal factors extrinsic to the bone itself, although there is no evidence that the condition is adaptive. The facts DAVIS: THE GIANT PANDA 85 that it is markedly evident in the tail, where the static influences of posture and locomotion do not exist, and that the same effect is evident on the proximal ends of the ribs, strongly suggest that this is a part of the field effect involving the entire axial region of the body. Homiotic variability in the column of Ailuro- poda is greater than in any other carnivore exam- ined. This indicates that the mechanism regulating differentiation of the column is not yet stabilized around a new norm, which in turn suggests an absence of strong selection pressure on this region. Thus the vertebral column of Ailuropoda differs from that of Ursus in several respects. The dif- ferences are not random, but rather form some kind of pattern. We must assume as a working hypothesis that the differences are adaptive — that they are a product of natural selection. We then seek answers to two questions: (1) what is their functional significance, and (2) what morphoge- netic mechanism, intrinsic to the bone tissue, lies behind them? It has been noted repeatedly throughout the description that the column of Ailuropoda resem- bles columns designed to withstand strong thrust forces acting anteroposteriorly in the direction of the sacrum. Among terrestrial mammals such forces, and correlated modifications of the column, occur only in fossorial and bipedal forms. The work of Slijper shows that the mammalian column responds adaptively to such forces, even non- genetically. Ailuropoda is, of course, in no way fossorial; and it is no more bipedal than the bears, in which the column shows slight — almost trivial compared with that in Ailuropoda — convergence toward the column of truly bipedal forms. The column of Ailuropoda cannot be explained on the basis of mechanical requirements, and therefore the differences from Ursus cannot be attributed to natural selection acting on the column. The seem- ingly adaptive modifications must be "pseudo- adaptations." The data of Sawin and Hull suggest an alterna- tive explanation. All those areas of a tissue that are in a state of competence at a given moment during ontogeny are known to be affected by a genetic factor operating at that moment. Thus the lumbosacral peculiarities of Ailuropoda may reflect an accident of ontogenetic timing rather than the action of selection on the lumbar region. If the differentiating lumbar region were compe- tent at the same moment as some other region on which selection was acting strongly (e.g., the skull), then in the absence of strong selection against the induced lumbar modifications, such modifications would be carried as a pleiotropic effect. If they were strongly selected against they would presum- ably be buffered out. The extraordinary homiotic variability of the lumbosacral region in Ailuropoda supports this interpretation, as does the otherwise unintelligible modification of the pelvis (p. 113). On the basis of the available evidence it must be concluded that primary differences between the column of Ailuropoda and that of Ursus are not adaptive, but represent a pleiotropic effect result- ing from an accident of ontogenetic timing. The genetic basis for such an effect is probably very simple. D. Conclusions 1. The vertebral column of Ailuropoda differs from that of Ursus (and other arctoid carnivores) in several important respects. (a) The regional boundaries are shifted crani- ally in a gradient that decreases in intensity from the lumbosacral boundary (greatest) to the thoracocervical (least). (b) All vertebrae are broadened and depressed. (c) Homiotic variability exceeds that known in any other carnivore. 2. The differences are not adaptive. 3. The differences are associated with intensi- fied growth at the anterior end of the body axis — the head. Similar correlations are evident in pri- mates and in bulldogs. 4. The characteristic basic features of the ver- tebral column in Ailuropoda are a pleiotropic effect resulting from an accident of ontogenetic timing. V. THE THORAX The thoracic region, as pointed out above, is relatively longer in Ailuropoda than in any other arctoid carnivore. This is true when the extent of the thorax is measured dorsally, along the ver- tebral column. On the other hand the ventral length of the thorax, measured along the sternum, is notably less than in any other arctoid carnivore. A. Ribs The number of ribs varies between 13 and 14 pairs in the eight skeletons examined, with a high proportion of asymmetries on the two sides of the same animal (see Table 9, p. 75). On the basis of the available material it is impossible to determine which is the typical number. In one skeleton (31128), which shows other gross abnormalities, the first rib on the left side is short, not reaching the manubrium, and the tubercular head is pathological. The second rib resembles the first of the opposite side, but its sternal end is 86 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 bifurcate and attaches to the manubrium by a wide bifurcate costal cartilage. of the sternum in this animal. In Su Lin two pairs of the false ribs are floating. Ailurvpoda ^"«* Fig. 42. Tenth rib, lateral view. Above, posterior views of heads of same ribs. In two skeletons there are nine pairs of true ribs, which is the normal number for arctoid carnivores. The eighth and ninth pairs are not attached to sternebrae in Ailuropoda, however; instead, the ends of the sternal cartilages of each pair meet at the ventral midline, ventral of the xiphoid carti- lage. This is obviously a result of the shortening The first costal cartilage is about 20 mm. long, the ninth about 230 mm., in Su Lin. In an adult the costal cartilages are very heavily calcified, with coarse granular deposits appearing on the surface. The ribs are very similar in length and curvature to those of bears of comparable size ( Ursus ameri- canus). All the ribs are remarkable, however, for DAVIS: THE GIANT PANDA 87 0 a 0 f \ I ° ff / Fig. 43. Approximate area of maximal increase in thickness of cortical bone in Ailuropoda. the immense bulk of their vertebral ends (fig. 42). The transverse diameter of the neck of a given rib in Ailuropoda is at least twice the diameter in Ursus americanus. The disparity becomes in- creasingly less toward the sternal end of the rib, until the sternal third is no larger in the panda than in the bear. It is at least suggestive that the maximum broadening is in that part of the rib closest to the vertebra, where, as we have seen, a pronounced broadening effect is apparent, and that the width gradually decreases to normal as we move along the rib away from the vertebra. B. Sternum The sternum is composed of a short body and an extremely long xiphoid cartilage. The body is about 55 per cent of the length of the thorax in 88 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Ailuropoda, while in other arctoids it is from 75 to 100 per cent. There are six sternebrae (including the manu- brium) in each of three skeletons of Ailuropoda examined. In other arctoids there are nine, ex- cept in the Canidae, which usually have only eight. All the sternebrae are short. The manubrium is short, compared with that of Ursus and other arctoids, and is relatively wider transversely. In other arctoids this bone is pro- duced anteriorly into a point, so that the outline is similar to a spear head. This point is much less evident in Ailuropoda, and in one of three speci- mens is totally lacking so that the anterior border of the manubrium is truncated. A single pair of costal cartilages articulates with the manubrium. The remaining sternebrae, five in number, are short and spool-shaped, rectangular in cross sec- tion. The first four measure about 25 mm. in length, the fifth about 20 mm. The xiphisternum is a remarkably long (120 mm.) cartilaginous rod, tapering gradually to a point. It provides attachment for the sternal part of the diaphragm and the posterior elements of the transverse thoracic muscle. Elongation of the xiphisternum appears to be a compensation for the shortness of the bod^'^ of the sternum, since the origin of the sternal part of the diaphragm is thus brought into line with the origin of the costal part of this muscle. In the Canidae and Procyonidae the xiphister- num is composed of an ossified rod ending in an expanded flattened cartilage. In the Ui*sidae it is a cartilaginous rod, with an ossicle of variable size embedded in the anterior end. In the Procyonidae the last stemebra is only about half the thickness of those preceding it, pro- ducing a "step" in the sternum. The last costal cartilages meet their fellows beneath this bone, in- stead of inserting into its lateral edges as they normally do. A similar condition is often seen in bears, in which this stemebra may be entirelj* un- ossified. The posterior end of the sternum seems to be undergoing regression in this group. C. Review of the Thorax Two points are of interest in the bones of the thorax: the extraordinary expansion of the prox- imal ends of the ribs, and the shortening of the sternum. No mechanical advantage can be assigned to the rib condition. It is most easily explained as an extension of the morphogenetic field effect that is oi>erating on the adjoining vertebrae, and hence without functional significance as far as the ribs are concerned. Thus a region of increased bone deposition extends the entire length of the head and body and extends laterally over the proximal two-thirds of the rib cage (fig. 43). Since the cor- tex of the long bones is also thickened, the effect is general over the entire skeleton though reduced peripherally. An astonishingly similar condition is seen in the ribs of the Triassic marine nothosaur Pachypleurosaurus (1931, Peyer, Abh. Schweiz. Paleont. Ges., 51, pi. 25; 1935, Zangerl, op. cit., 56, fig. 23). In the reptile, enlargement of the proximal ends of the ribs is associated with pachy- ostosis; there is no evidence of this in Ailuropoda. The extreme shortening of the sternum seen in Ailuropoda is foreshadowed in the related procyo- nids and bears, in which a tendency toward re- duction from the rear forward is evident. There is no obvious mechanical advantage to this shift, which is inversely correlated with elongation of the thorax in these animals. The sternum has been shortened repeatedly in various mammalian lines, but to my knowledge this has never been studied from the standpoint of animal mechanics. We may conclude, provisionally, that (1) the broadening of the vertebral column has extended mor- phogenetically to the proximal ends of the ribs in Ailuropoda, and (2) the shortening of the sternum is the final expression of a trend, of unknown signifi- cance, seen in related forms. VI. THE FORE LEG In the giant panda, the bears, and the procyo- nids the fore legs are used for manipulating objects, especially during feeding, to a far greater extent than in other carnivores. This requires a wider range of movement, particularly of abduction of the humerus and rotation of the fore arm, than in tj-pical carnivores. All these forms are also more or less arboreal, and in the heavier forms at least this has profoundly altered the architecture of the shoulder and fore leg (Davis, 1949i. Such uses of the fore limb are secondary ; in the primary carni- vore condition the fore leg is modified for cursorial locomotion, and the structure of the limb in all carnivores has been conditioned by this fact. A. Bones of the Fore Leg The clavicle is vestigial or absent in all Car- nivora, never reaching either the acromion or the sternum when a clavicle is present. Among the Arc- toidea it is normally absent in Canis, exceptionally being represented by a small nodule of cartilage or bone (Ellenbei-ger and Baum). It is present as a small spicule of bone embedded in the cephalo- humeral muscle in Bassariscus, Procyon, and Ailu- rus. It is completely wanting in the Ursidae, and M. rhomboideus M. rhomboideus capitis M. rhomboideus M. infraspinatus M. acromiotrap. + M. spinotrap. M. teres major M. subscapularis minor ■M. supraspin. M. triceps longus M. teres minor/ M. spinodeltoideus'' M. biceps M. atlantoscapularis M. acromiodelt. Fig. 44. Right scapula of Ailuropoda, lateral view. A, right scapula of Ursus areios. M. rhomboideus M. subscapularis M. coracobrachialis M. levator scapulae + M. serratus ventralis M. teres major M. triceps longm Fig. 45. Right scapula of Ailuropoda, medial view. 89 90 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 there is no indication of a clavicle in Ailuropoda. The clavicle is less degenerate in the Feloidea. 1. Scapula It has been stated repeatedly that the scapula is influenced by muscular action probably to a greater degree than any other bone in the body. Dependence of scapula shape on muscle function has been demonstrated experimentally for rats (Wolff son, 1950). The forces involved in molding the scapula are extremely complex, no fewer than 17 muscles arising or inserting on the scapula in carnivores, and interpretation of differences in scapular form is difficult. No adequate study of the I'elation between form and function of the mammalian scapula exists, although such a study was attempted by Reinhardt (1929). The scapula of the giant panda appears at first glance to be quite strikingly different from that of any other arctoid. This is due to the unorthodox outline of the bone (fig. 44). Actually, all the features that distinguish the scapula of Ursus from other arctoids are also present in Ailuropoda, al- though the large postscapular fossa of the bears is reduced in the panda. These ursid features are: prominent postscapular fossa, large table-like acromion with poorly differentiated metacromion, breadth of neck exceeding long diameter of glenoid fossa, well-defined spiral groove on axillary border, and narrow glenoid cavity. There can be no doubt that the scapula of the giant panda is basically a bear scapula. I have tried to show (Davis, 1949) that the shoulder architecture of bears, and hence the form of the scapula, is adapted to resist pulling forces (the opposite of the thrust associated with normal locomotion) developed in connection with climb- ing, the morphological effects of which are exag- gerated because of the size of the animal. The tremendous postscapular fossa, from which the subscapularis minor muscle arises, is the most con- spicuous feature associated with this reversed force direction; it is even larger in such powerful diggers as the anteaters and armadillos, in which similar pulling forces are involved. The posterior angle (and thus the scapular in- dex) is influenced chiefly by the posterior part of the serratus ventralis muscle. This part of the serratus is a posterior rotator of the scapula, and is used in protraction of the arm (A. B. Howell, 1926). The posterior part of the serratus is well developed in Ailuropoda, and this may account, at least in part, for the pulling out of the poste- rior angle. Morphology. — The scapula of Ailuropoda is more fan-shaped than the almost rectangular scap- ula of Ursus. Of the three borders, the coracoid border is produced anteriorly in some individuals (fig. 44) to form a sharp angle that marks the an- terior limit of the insertion of the rhomboideus, which is remarkable for the length of its insertion line. In other individuals this angulation is miss- ing. The scapular notch, which is at best poorly developed in nearly all carnivores, is almost oblit- erated in Ailuropoda and Ursus. The vertebral border forms a smooth, gentle curve, with no clear indication of the juncture of the coracoid and ver- tebral borders (the anterior angle; median angle of human anatomy). This blurring of the ante- rior angle is characteristic of Carnivora. The pos- terior extent of the vertebral border is determined by the serratus ventralis; the rhomboids appar- ently have no influence in determining the position of the posterior angle. The axillary border, from which the long triceps arises, is relatively straight and clearly defined. Its juncture with the verte- bral border (the posterior angle; inferior angle of human anatomy) marks the juncture of the ser- ratus ventralis and teres major muscles, and is clearly defined. In the Carnivora the ouiline, and hence the major indices, of the scapula are determined by two muscle groups related to the vertebral border: the rhomboids, and the levator scapulae + serratus ventralis. The lateral surface is slightly concave, and is divided by the spine into the supraspinous and infraspinous fossae. The infraspinous fossa con- siderably exceeds the supraspinous in area, and is relatively much larger than in the bears. This increased size is due to an extension posteriorly of the axillary border, as is shown by the angle formed by the axillary border with the spine; this is 38-40° in Ailuropoda, 20-30° in Ursus. The floors of both fossae are marked by vermiculate rugosities simi- lar to those seen in the giant anteater, and there is a nutrient foramen in each above the glenoid cavity. The coracoid border of the supraspinous fossa is sometimes raised and sometimes not, a variation also found in bears. In some individuals of Ailuropoda it is raised, so that the fossa is con- cave in cross section, while in others it is depressed, producing a prominent convexity in the fossa. The axillary border of the infraspinous fossa is influ- enced by the triceps longus, whose origin in the bears and giant panda extends nearly or quite to the posterior angle. This border is sinuous in Ailu- ropoda, straight in the bears. The teres major process lies behind the axillary border at the pos- terior angle. The teres major muscle arises from its posterior border. The lateral surface of this process is excavated into the postscapular fossa, from which the subscapularis minor muscle arises. DAVIS: THE GIANT PANDA 91 M. acromiodelt. M. biceps M. subscapularis minor Fig. 46. Ventral view of right scapula of Ailuropoda (left) and Ursus arctos (right). In Ailuropoda the postscapular fossa is well marked, but has been much reduced by the pos- terior extension of the infraspinous fossa so that it is much less conspicuous than in Ursus. The postscapular fossa is continued toward the glenoid cavity as a wide trough that extends the en- tire length of the axillary border, separated from the medial surface of the blade by a prominent ridge, and from the lateral surface by the infe- rior scapular spine. This trough (fig. 46), which lodges the subscapularis minor muscle, is twisted through 180°. The glenoid cavity is pear-shaped, with the apex anteriorly, as it is in other carnivores and in mam- mals generally. The notch that appears in the mar- gin opposite the spine in certain carnivores {Canis, Felis) is wanting in Ailuropoda and most other car- nivores. In Ailuropoda the cavity is narrower (in- dex length X 100 breadth = 645, mean of two specimens) than in any other carnivore. It is also narrow in bears (index 670, mean of 6 specimens), and gen- erally narrower in arctoids than in aeluroids. The cavity is shallow in both Ailuropoda and Ursus. The neck is notable for its great anteroposterior diameter, although this is slightly less than in Ur- sus. The supraglenoid tuberosity, for the origin of the tendon of the biceps, is a prominent scar immediately above the anterior border of the gle- noid cavity. Above and mesad of it is a slight elevation, the coracoid process, bearing on its me- dial surface a scar from which the tendon of the coracobrachialis arises. The infraglenoid tuber- osity, from which the anteriormost fibers of the long triceps take tendinous origin, is much less prominent than in Ursus. It is merely a rough- ened triangular area above the lip of the glenoid cavity that continues without interruption into the axillary border. 92 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. infraspinatus M. brachialis + M. triceps lateralis " M. triceps medialis tcaput longum) M. teres minor M. acromiodelt Crista deltoidea M. brachialis M. cephalohumer. Crista pectoralis M. ext. carpi radialis longus et brevis Crista epicondyltts lat. M. anconaeus M. ext. dig. comm. ct. lat M. ext. carpi ulnaris M. supraspin. Tuber, majus M. stemohumer. prof. M. pect. superf. M. brachialis M. brachioradialis M brachialis Epicondylus lateralis Fig. 47. Lateral view of right humerus of Ailuropoda. The spine is slightly twisted, as it is also in bears, reflecting the action of the deltoid and trapezius muscles. The line formed by the crest of the spine is convex posteriorly, in some individuals markedly so (reflecting the pull of the acromiotrapezius?) . The inferior part of the spine, just above the acro- mion, is inclined slightly anteriorly, while the pos- terior part is vertical or inclined slightly posteri- orly. The lateral (free) border, again as in bears, is squared in cross section. The spine is continued ventrally into a heavy acromion process, which functions in the origin of the acromiodeltoid and levator scapulae ventralis muscles. The meta- cromion, the process on the posterior border from which the levator scapulae ventralis arises in most carnivores, is not indicated in Ailuropoda and is scarcely more prominent in Ursus. The lateral surface of the acromion is flat and table-like in both bears and panda. In summary, the scapula of Ailuropoda agrees with Ursus in all features that distinguish the bear scapula from that of other carnivores. The most notable difference between the panda and the bears is the posterior expansion of the infraspinous fossa in Ailuropoda, which seriously encroaches on but does not obliterate the typically ursid postscapular fossa. The infraspinous fossa is associated with the infraspinous and long triceps muscles, which DAVIS: THE GIANT PANDA 93 Tuber, minus M. supi'aspin. j. M. subscapularis Tuber, majus M. triceps medialis (caput longum) M. coracobrachialis brevis M. pect. prof.- Crista peclorali M. teres major M. latissimus dorsi M. pect. supei f M. triceps medialis (caput intermedium) M. eoracobrachialib longus M. anconaeus Fossa olecrani M. flexor digitorum prof. (4) M. flexor digitorum prof. (2) Epicondylus medialis M. pronator teres M. flexor carpi radialis M. flexor digitorum prof. (1) M. palmaris longus M. flexor carpi ulnaris Fig. 48. Medial view of right humerus of Ailuropoda. are involved in fixation and flexion of the shoulder joint. 2. Humerus The humerus in the Carnivora serves for the origin or insertion of 28 muscles. Of these, 12 be- long to the shoulder joint and 16 to the elbow joint or lower arm and manus. The form of the humerus is determined largely by these muscles. In Ailuropoda the humerus is longer than the radius, as it is in all arctoid carnivores except Pro- cyon and most dogs. The mean ratios (length of radius X 100/length of humerus) for various gen- era are as follows: Humeroradial index * N Bassaricyon 1 72.7 Ailurus 3 74.7(72.1-77.8) Ailuropoda 7 77.1 (74.7-79.7) Bassariscus 4 79.0 (77.9-79.5) Ursus (various species) 6 82.3 (78.3-85.8) Nasua 2 85.5 (82.7-88.2) Canis lupus 4 100.6 (98.1-102.9) Procyon 4 100.9 (99.5-102.5) * In generalized mammals the radius length is about 85 per cent of the humerus; this is true in such generalized ter- restrial insectivores as Echinosorex, Erinaceus, and Soleno- don. A. B. Howell (1944) states that in the generalized condition the humerus and radius are about the same length, but this is obviously not true for mammals at least. For simple mechanical reasons the radius tends to lengthen with cursorial locomotion, but reasons for shortening this bone are not so clear. In man (European) the index is about 74. 94 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 The humerus of Ailuropoda (figs. 47, 48) does not differ notably from that of other arctoid carni- vores. It is slightly convex anteriorly. IVIuscle scars are extremely prominent, and the area above the olecranon fossa, where the anconeus muscle arises, is marked by vermiculate rugosities similar to those on the scapula. The angulation in the profile at the inferior end of the deltoid ridge, char- acteristic of bears, is wanting in the giant panda. The head is offset posteriorly from the shaft; a line drawn through the center of the shaft just touches the anterior edge of the head. This is similar to other arctoids, except Ursus in which the head lies almost on top of the shaft.' The ar- ticular surface greatly exceeds the opposing sur- face on the scapula in area. The head in transverse section forms a perfect arc of about 170°, thus nearly a semicircle. In frontal section it forms a much smaller sector (about 65°) of a circle nearly twice the diameter, so that the head appears flat- tened when viewed from the rear. In Ursus the transverse section of the head is nearly identical with that of Ailuropoda, but the frontal section forms a slightly larger sector (78°-93°) of a circle only slightly larger than that formed by the trans- verse section. In other words, in the bears the humeral head represents a part of a nearly perfect hemisphere, while in Ailuropoda it tends toward the almost cylindrical structure seen in such highly cursorial forms as the horse. The anatomical neck is scarcely indicated, ex- cept posteriori}'. The tubercles are low and very bear-like. The greater tubercle scarcely rises above the level of the head. It is sharply defined anteriorly, where it continues into the pectoral ridge; its posterior boundary is almost obliterated by the infraspinatus impression. The supraspi- natus impression extends almost the entire length of the dorsal lip of the greater tubercle. There are several large nutrient foramina between the greater tubercle and the head. The lesser tubercle is prominent; the well-marked subscapularis impres- sion covers practically its entire medial surface. The intertubercular (bicipital) groove between the two tubercles is wide and deep. In life it is bridged over by the transverse humeral ligament to form a canal. The groove lodges the tendon of the biceps and transmits a branch of the internal circumflex artery. There are a number of nutrient foramina in the floor of the groove. The shaft is triangular in cross section, because of the several prominent crests. The single nutri- ent canal that is prominent on the posterior surface of the shaft in other arctoids is represented by sev- ' In other ursids {Thalarcios, Melursus, Helarctos) the head is offset. Tremarctos is similar to Ursus. eral minute foramina in Ailuropoda. The pec- toral ridge (crista tuberculi majoris, BXA), on the anteromedial surface, extends from the greater tubercle nearly down to the distal end of the shaft. It is a very prominent crest that provides inser- tion for the superficial and deep pectoral muscles. The deltoid ridge begins immediately below the posterior end of the greater tubercle, on the pos- terolateral surface of the shaft; near the middle of the shaft it arches across the anterior surface of the shaft and joins the pectoral ridge just below the middle of the humerus. The deltoid ridge pro- vides origin for the long head of the brachial mus- cle and insertion for the cephalohumeral. Midway between the pectoral and deltoid ridges there is a third ridge, which marks the medial boundary of the insertion of the cephalohumeral. Mesad of the pectoral ridge, on the flat medial surface of the shaft, is a prominent elongate scar 40-50 mm. long that marks the insertion of the latissimus dorsi and teres major. Distally the shaft bears the tremendous wing- like expansion of the lateral epicondylar ridge on its posterolateral surface. This ridge extends proximad nearly to the middle of the shaft. It provides origin for the short head of the brachialis, the brachioradialis, and the extensor carpi radi- alis longus and brevis. These are all forearm flex- ors, although the extensor carpi radialis is chiefly an extensor of the hand. The lateral part of the anconeus arises from its posterior face. This ridge is well developed in all procyonids, in some of which (e.g., Nasua) it is as prominent as in Ailu- ropoda. It is about as well developed in bears as in the giant panda. It is likewise present in mus- telids, and is extremely well developed in bun'owers such as Taxidea and Meles. It is scarcely indicated in the cursorial dogs. The distal end of the shaft is thinner antero- posteriorly but wider than it is farther proximally; it is relatively slightly wider and much thinner than in bears. The trochlea ( = capitulum -f- troch- lea of human anatomy) is almost identical with that of Ursus, except that it is somewhat wider. The trochlea is divided into lateral and medial parts by a faint ridge that runs spirally postero- laterally to terminate in the ridge bordering the olecranon fossa. The lateral part of the trochlea, with which the radius and a small part of the ulna articulate, forms a semi-cylinder with only a very faint anteroposterior groove. The medial part of the trochlea, which forms the major ulnar articu- lation, forms a trough-shaped spiral path extend- ing posteriorly well into the olecranon fossa. This spiral trough forces the ulna to shift medially 5 mm. or more as the elbow is flexed. The poste- DAVIS: THE GIANT PANDA 95 rior part of this trough has an extremely prominent external lip on which the articular surface faces medially. The coronoid fossa, above the troch- lea anteriorly, is entirely wanting, as it is also in bears. The olecranon fossa, above the trochlea posteriorly, is deep and relatively wider than in Ursus. ratio, length pelvis/length radius is 130.3 (126.8- 132.7) in Ailuropoda, 110.3 (107.3-118.4) in Ursus, 100.9 (95.5-103.3) in Procyon, 108.2-108.8 in Ailu- rus, 110.9 (105.3-113.9) in Bassariscus, and 78.4 (76.3-80.1) in Canis. The significance of the re- duced radius length in Ailuropoda is discussed be- low (p. 102). In both panda and bears the radius Ailuropoda Ursus Canis Procyon Fig. 49. Distal ends of humeri of Ailuropoda, Ursus americanus, Canis lupus, and Procyon loior. The medial epicondyle is more prominent and more vertically compi'essed than in Ursus. It pro- vides origin for the pronator teres, flexor carpi ra- dialis, flexor digitorum profundus, palmaris longus, and flexor carpi ulnaris. These are all flexors of the hand, except the pronator teres, which pronates the forearm. The entepicondylar foramen, which transmits the median nerve and median artery, was present in all specimens of Ailuropoda examined. This foramen is absent in the Ursidae (except Tremarctos ornatus) and Canidae, present in the Procyonidae, in Ailurus, and in most Mus- telidae. Its presence in Ailuropoda and Tremarc- tos is probably a secondary condition correlated with the large size of the epicondyle in these two genera. The lateral epicondyle is less prominent than in Ursus, and is considerably narrower. It pro- vides origin for the extensor digitorum communis and lateralis and the extensor carpi ulnaris. These are all extensors of the manus, although the ex- tensor carpi ulnaris chiefly abducts the hand ulnar- ward. It has no direct genetic basis, and in this instance cannot be used as a "character." The humerus of Ailuropoda is so similar to that of the bears, especially to such forms as Tremarctos and Melursus, that Lydekker's statement (1901) to the contrary is almost incomprehensible. 3. Ulna and Radius The ulna is slightly heavier than in a bear of comparable size, while the radius is slightly more slender. The radius is shorter in relation to pelvic length than in any other carnivore measured. The lies almost entirely laterad of the ulna at the elbow joint. The radius is slightly more dorsal in Pro- cyon and Ailurus, and in the narrow elbow joint of the cursorial dogs it lies almost in front of the ulna. The form of the ulna is very similar to that of Ursus. The olecranon, measured from the center of the semilunar notch, averages 14 per cent of the length of the humerus;' this is likewise true for Ursus, Procyon, and Ailurus, while in Canis it is longer (19 per cent). The olecranon, which pro- vides insertion for the triceps complex and the flexor carpi ulnaris, is a heavy knob-like extension of the ulna, bent slightly medially. The medial surface is concave and is devoid of muscle attach- ments; the lateral surface provides attachment for parts of the triceps and anconeus. Anteriorly the olecranon forms the prominent anconeal process, which interlocks with the olecranon fossa of the humerus and forms the posterior part of the semi- lunar notch. The semilunar notch, bounded anteriorly by the coronoid process and posteriorly by the an- coneal process, is almost a perfect semicircle in profile. It is arched in cross section, lacking the median guiding ridge seen in dogs. The anconeal process has an extensive external face that rides against the external lip on the posterior part of the trochlea, and the coronoid process an internal face that rides against the inner wall of the troch- lear groove. This arrangement effectively locks the elbow joint and prevents any medial shifting ' Calculation as percentage of ulna length gives mislead- ing values in forms with elongated fore arm, such as Procyon and Canis. 96 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. anconaeus Incimira seiniliotari Proc. coroiioideus. M. brachialis M. biceps M. supinator M. flexor digitorum prof. 3 M. pronator teres M. pronator quadratus M. triceps Olecra}ion M. flexor carpi ulnaris M. flexor digitorum prof. 5 M. pronator quadratus Proc. .^Iiiloidviis Fig. 50. Right ulna and radius of Ailuropoda, posteromedial view. of the distal end of the ulna; there is no such pro- tection against lateral shifting. The radial notch is a shallow depression on the lateral side of and immediately below the coronoid process, in which the head of the radius rotates. The shaft tapers gradually toward the distal end. It is slightly bowed, with the convexity out- ward. The bone is wider anteroposteriorly than it is from side to side. Immediately below the coronoid process, on the anterior surface of the bone, there is a prominent ovoid depression that marks the insertion of the brachialis tendon. In most specimens a wide rugose ridge along the mid- dle third of the lateral surface of the shaft marks the attachment of the interosseous ligament. The distal end of the ulna is slightly expanded. Dorsally it bears a circular, much-rounded artic- ular facet for the radius. Beyond this the shaft is continued into the short peg-like styliform process, which bears a rounded facet for the cu- boid and pisiform on its anteromedial surface. The radius is curved in both planes; it is slightly convex anteriorly, and forms a long S-curve in the lateral plane. This complex curvature of the ra- dius is seen to some degree in all Carnivora except the cursorial dogs. The capitulum of the radius is set off by a very distinct neck. It is an elliptical disk, the long diameter running from anterolateral to postero- medial. The ratio of long to short diameter is DAVIS: THE GIANT PANDA 97 about 10 : 7, and this ratio is about the same as in Ursus. In burrowing forms (Taxidea, Meles) the capitulum is even more ovate, whereas in ar- M. triceps cumference of the head; the medial one-fourth, where the capitular eminence is situated, has no articular surface. M. anconaeus M. abductor poll, longus M. ext. indicus proprius M. ext. dig. lat. M. ext. carpi ulnaris Emiiieiilia capllulorum M. abductor poll, longus M. supinator M. pronator teres M. abductor poll, longus M. ext. dig. comm. M. ext. carpi radialis longus M. ext. carpi radialis brevis Fig. 51. Right ulna and radius of Ailuropoda, anterolateral view. boreal forms {Procyon, Nasua, Polos) it is more nearly circular. The capitular depression, which articulates with the lateral part of the trochlea of the humerus, is very shallow. On its anteromedial circumference it bears a low elevation, the capitular eminence, that forms the anterior lip of the radiohumeral articulation in all positions of the radius, and acts as a stop that limits the excursion of rotatory movements of the radius. The articular circum- ference, which articulates with the radial notch of the ulna, is not continuous around the entire cir- The shaft of the radius is triangular in cross sec- tion, the base of the triangle forming the flat ven- tral surface of the bone. The radial tuberosity, for the insertion of the biceps tendon, is on the ventro- medial surface immediately below the neck. Oppo- site this, on the anterior aspect, is a scar marking the attachment of the lateral collateral ligament. The interosseous crest, for the attachment of the interosseous ligament, begins below the radial tu- berosity as a wide, roughened scar for the heavy proximal part of the ligament. A little above the middle of the bone it changes abruptly into a ridge-like crest. Sesamoid, rad. Trapezoid Trapezium Scapholunatum Magnum Unciforme Cuneiforme Pisiforme Fig. 52. Right carpus and metacarpus of Ailuropoda, dorsal view. Fig. 53. Right carpus and metacarpus of Ailuropoda, ventral view. 98 DAVIS: THE GIANT PANDA 99 The distal end of the radius is expanded and bears two articular surfaces, the large concave car- pal surface for articulation with the scapholunar, and laterally the small flat ulnar notch for articu- lation with the ulna. The carpal surface is nar- rower from side to side but wider anteroposteriorly than in Ursus, thus providing a less trough-like articulation for the carpus. The prominent saddle shape of the articular area on the styloid process that is seen in Ursus is scarcely indicated in Ailu- ropoda. Also the medial end of the articular sur- face is in Ailuropoda deflected proximally toward the ulnar notch. The styloid process is a blunt projection on the medial side; a deep furrow on its dorsolateral surface lodges the tendon of the ab- ductor poUicis longus. Just laterad of this, on the dorsal surface of the styloid process, is a shal- low furrow for the tendon of the extensor carpi radialis longus, separated by a ridge from the fur- row for the extensor carpi radialis brevis. Another shallow furrow near the lateral border lodges the tendon of the extensor digitorum communis. 4. Carpus The carpus (figs. 52, 53) is very similar to that of bears, except for the tremendous development of the radial sesamoid and the modifications of the scapholunar associated therewith. The carpus-fore- arm articulation is largely between the scapholunar and the radius, which form an almost ball-and- socket joint permitting very extensive excursion. The styloid process of the ulna, as in bears and procyonids, is lodged in a widely open notch formed by the cuneiform and pisiform. The carpus is dominated by the scapholunar. This bone greatly exceeds any of the other carpals in size, and articulates with all the other carpal bones except the pisiform, and with the radius and the radial sesamoid. The articular surface for the radius occupies almost the entire dorsal and poste- rior surfaces of the bone, forming an ovate articula- tion that in some individuals is in contact anteriorly with the articular surface for the trapezium. This is more extensive than in any other carnivore, al- though in Ailurus and Potos it is closely approached. In Ursus the lateral part of this surface has a dimple-like depression, to receive the saddle on the distal end of the radius; this depression is com- pletely wanting in Ailuropoda and in Ailurus and Potos. The anteromedial end of the bone is pro- duced into a stout hook-like process, directed ven- trally, that bears a prominent articular surface for the radial sesamoid on its anteromedial surface. This articular surface is an elongate oval, its long axis vertical, and is convex in both planes. The anterior surface of the scapholunar bears three ir- regular shallow excavations for the trapezium. trapezoid, and magnum, and the lateral surface bears articular facets for the magnum and unci- form. The cuneiform is very similar to the corre- sponding bone in Ursus, but relatively slightly larger. It articulates with the scapholunar, the pisiform, and the unciform. The pisiform is, next to the scapholunar, the largest bone in the carpus, and is very similar to the corresponding bone in Ursus. It articulates with the cuneiform, forming with it a shallow V- shaped notch dorsolaterally, in which the styliform process of the ulna articulates. The bone extends posteriorly, ventrally, and slightly laterally from the carpus, its expanded tip embedded in a large fibro-fatty pad that underlies the lateral carpal pad. Five muscles and five ligaments attach to the bone. The tendon of the flexor carpi ulnaris attaches to the posterior surface, the opponens and abductor digiti quinti and palmaris brevis to the anterior surface, and the flexor digiti quinti to the inner border. A prominent scar near the tip on the anteromedial surface marks the attachment of the transverse carpal ligament, and another scar on this surface proximally marks the attachment of the pisometacarpal ligament. In the distal row the trapezium and trapezoid are very small, articulating distally with meta- carpals 1 and 2 respectively. The magnum is larger, and articulates with metacarpal 3. The unciform bears metacarpals 4 and 5. The radial sesamoid (fig. 54) is the most ex- traordinary bone in the fore foot. It is about 35 mm. in length, and lies in line with the meta- carpals, closely resembling a sixth metacarpal on the medial border of the hand. It underlies the accessory lobe of the carpal pad. The bone is com- pressed from side to side, measuring about 15 mm. in height by only 6 or 7 mm. in thickness. The distal end hooks sharply inward toward the first metacarpal. The radial sesamoid articulates ex- tensively with the enlarged medial process of the scapholunar, and is in contact with the medial border of the first metacarpal. The articular sur- face for the scapholunar is ovate with the long axis dorsoventral, and is concave both laterally and dorsoventrally. The contact surface with the first metacarpal is dorsomedial, and is not cartilage covered. A large depression on the outer surface of the radial sesamoid near the base marks the attachment of the tendon of the abductor pollicis longus. The abductor pollicis brevis and opponens pollicis arise from its medial surface. A sizable radial sesamoid articulating with the scapholunar is present in all the other arctoid car- nivores, and a corresponding bone exists in many 100 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 (^ B d Ailuropoda Tremarctm Ursus Ailurus ProcyM. adductor sM. gracilis M. rectus abdominis Fig. 57. Male pelvis of Ailuropoda, lateral view. (Inset, A, pelvis of Ursus arcios.) muscles and the anterior end of the inguinal liga- ment, is thick and heavy. It lies farther anterior than in Ursus, and the iliac crest is correspond- ingly shorter and less curved. The posterior supe- rior iliac spine is also relatively heavy. The anterior and posterior inferior iliac spines are not even indicated. The dorsolateral surface of the ilium, which provides origin for the middle and deep gluteals, is a shallow elongated trough, the gluteal fossa. It is devoid of surface modeling ex- cept for a faint vermiculation near the iliac crest. The area of the gluteal fossa is about 5700 and 7500 mm." in two specimens of Ailuropoda, 7200 mm.- in a specimen of Ursus americanus, and 11,900 mm.- in a specimen of Ursus arctos.' The ventro-medial surface of the ilium (fig. 58), which provides origin for the iliacus, quadratus lumbo- rum, and sacrospinalis muscles, is slightly convex along both its axes. A faint longitudinal ridge, not always evident, divides the surface into a lat- eral iliac area and a medial sacrospinal area; this is called the pubic border by Flower, Straus, and ■ See p. 43 for method used in measuring areas on bones. other anatomists. A low but prominent elevation near the middle of the ridge is associated with the origin of the sacrospinalis. A large foramen-like opening at the posterior end of the ridge, and lying in the sacroiliac articulation, is filled with fat and connective tissue in life; it is present but is usually less foramen-like in Ursus, and apparently repre- sents the separation between the dorsal and ven- tral elements of the embryonic transverse processes of the first sacral. The corpus is short and heavy, only slightly lat- erally compressed as in Ursus. Its superior border bounds the greater sciatic notch, which has been crowded posteriorly by the posterior extension of the sacroiliac union. The inferior surface is rounded, without crests or ridges. The iliopectin- eal eminence is a low elevation, much less promi- nent than in Ursus, on the inferior surface just anterior to the acetabulum. The inferior gluteal line, separating the gluteal and iliac surfaces of the ilium, is scarcely indicated on the corpus. Im- mediately in front of the acetabulum it passes into the iliopubic eminence, which is likewise much less 106 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. iliocostalis . transversus abdominis M. sartorius M. pectineus M. rectus femoris Fig. 58. Male pelvis of Ailuropoda, ventral view. prominent than in Ursus; it marks the attachment of the rectus femoris. The articular surface of the ilium (fig. 59), which articulates with the auricular surface of the sa- crum, resembles that of Ursus but is relatively longer and narrower. It is an elongate horseshoe, open anteriorly, with a very irregular surface, the irregularities interlocking closely with correspond- ing irregularities on the sacrum. The narrow space enclosed by the horseshoe is filled with fibrocarti- lage. The extensive articulation, intimate dove- tailing, and partial fusion of the sacroiliac joint contrast sharply with the relatively smooth and much smaller auricular surface of other arctoids. The pubis is the most delicate bone in the pel- vis. It is more lightly built than in Ursus, and much more so than in the cursorial dogs. The corpus, which forms the ventral part of the ace- tabulum, is the heaviest part of the bone. The acetabular ramus is very slender and elongate; it had been fractured bilaterally in one specimen ex- amined. The reduction in the length of the sym- physis has taken place anteriorly, and the angle formed by the acetabular ramus with the symphy- DAVIS: THE GIANT PANDA 107 Canis lupus lycaon Procyon lotor Ursus arctos AiluropodQ melonoleuca Fig. 59. Articular surface of left ilium in representative arctoid carnivores. sis in the sagittal plane is about 45° instead of 25- 35° as in Ursus, and the acetabular ramus is cor- respondingly longer. The length of the symphyseal ramus cannot be determined, since no available specimen is young enough to show the suture be- tween the pubis and the ischium. It is obviously very short, however, and is relatively much wider than in Ursus. The external surface of the sym- physeal ramus provides origin for the anterior parts of the gracilis, adductor, and external ob- turator muscles; the internal surface provides ori- gin for the anterior part of the internal obturator. The ischium is not directly involved in the sup- port function of the pelvis, except during sitting; it functions chiefly as anchorage for the posterior thigh muscles. The ischium does not differ much from that of Ursus or Procyon. It is composed of a stout acetabular ramus and a more slender de- scending ramus (tabula ischiadica of veterinary anatomy), and a heavy symphyseal ramus. The acetabular ramus is relatively shorter than in Ur- sus, and is ovate in cross section. Its shaft is almost free of muscle attachments; only the tiny gemelli arise from it. The sciatic spine, which separates the greater and lesser sciatic notches, is a short prominent transverse ridge as in Ursus. A small scar immediately anterior to the spine marks the attachment of the anterior gemellus, and immedi- ately behind the spine there is a smooth area, cov- ered with cartilage in life, over which the internal obturator rides. The saddle-shaped area between the sciatic spine and the ischial tuberosity is the lesser sciatic notch. It is converted into a fora- men by the sacrotuberous ligament, and transmits the distal end of the internal obturator muscle and various vessels and nerves. The ischial tuberosity is by far the most promi- nent feature of the ischium, and most of the mus- cles attaching to the ischium are inserted on or near it. The tuberosity is knob-like, about 35 mm. in diameter, with a much roughened posterior sur- face It has no inferior boundary, but continues directly into the roughened swollen posterior edge of the descending ramus, which narrows gradually as it descends and terminates abruptly about 40 mm. above the symphysis. The muscle attach- 108 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 ments are around the periphery of the tuberosity; the major part of its roughened posterior face Hes directly beneath the skin. The tuberosity is simi- lar, but more sharply bounded inferiorly, in Ursus. The lower third of the descending ramus, below the swollen area just described, is much the slen- derest part of the ischium ; it is no heavier than the acetabular ramus of the pubis. It provides attach- ment for the posterior ends of the adductor and gracilis externally, and for the internal obturator internally. The descending ramus forms an angle of about 55° with the sagittal plane. This angle is similar in other arctoids examined except in Canis, in which it is only about 20° (fig. 61). The symphyseal ramus, forming the posterior part of the symphysis pelvis, is broad and thick; the minimum transverse diameter of the entire symphysis (from obturator foramen to obturator foramen) is 40-50 mm. in Ailuropoda, whereas in a bear of comparable size it is 20-30 mm. In dor- sal view the sciatic arch, which is often non-existent in bears, is relatively deep. The acetabulum, composed of a horseshoe- shaped articular portion embracing a non-articu- lar fossa, differs little from that of Ursus and other arctoids. It looks slightly more laterally, forming an angle with the vertical of 11° and 14°, respec- tively, in two individuals, 15° in three specimens of Ursus. The acetabulum looks more ventrally in the cursorial wolf, forming an angle of 29° (26- 31) in three specimens of Canis lupus. The acetabulum is situated farther dorsad in Ailuropoda than in Ursus, its dorsal border lying well above the margin of the greater sciatic notch. The entire rim of the acetabulum is extremely heavy. The acetabular notch is almost twice as wide as in a bear of comparable size; the anterior boundary of the notch has been shifted forward to produce this increased width. The acetabular fossa is also relatively wider, and has increased its diameter by encroaching on the anterior arm of the articular portion, which accordingly is nar- rower than in Ursus. The obturator foramen is triangular in out- line, rather than ovate as in Ursus. Architecture and Mechanics. — The mam- malian pelvis is an extraordinarily complex struc- ture, subject to varied and often subtle forces. Moreover, it has had a long history, and treating the mammalian pelvis as if it were engineered de novo leads to difficulties and often even to absurdi- ties. Mijsberg's work (1920) was one of the first attempts to analyze the architecture and mechanics of the non-human mammalian pelvis. Other such studies have been made by Elftman (1929), Rey- nolds (1931), Kleinschmidt (1948), and Maynard Smith and Savage (1956). The mammalian pelvis serves three dissimilar purposes: (1) to provide support; to transmit thrust from the legs to the vertebral column, and from the column to the legs; (2) to provide attachment surfaces and lever arms for hip and thigh muscles; and (3) to transmit the terminal parts of the diges- tive and urogenital canals, especially important being the birth canal. Each of these has partici- pated in molding the pelvis, but the basic archi- tecture was largely determined by the support function. Elftman believed that the pelvis is "roughly modeled so as to fit the viscera and with finer detail so developed as to provide optimum support against gravity and leverage for loco- motion." As a supporting structure the pelvis is a complex system of arches and levers designed to provide strength and elasticity. Absorption of shock re- sulting from impact between the feet and the ground seems to have been a major factor in the design of limbs and girdles in mammals. The ar- chitecture of the mammalian pelvis, which is far less rigid than that of their reptilian ancestors, is otherwise unintelligible. In the frontal plane (fig. 60, B) the pelvis is com- posed of two round arches meeting at the acetab- ular a heavy dorsal arch composed of the two ilia and the sacrum, and a much lighter ventral ilio- pubic arch. Only the dorsal arch is directly in- volved in the support function of the pelvis; the ventral arch is concerned with the structural sta- bility of the pelvis. The dorsal arch is loaded both from above (weight of body, W) and from below (upward thrust of legs, T). In addition to bend- ing and shearing stresses, the loaded arch develops horizontal thrust which reaches a maximum at the base (the acetabula. A, A) whether loading is from above or below. The sole function of the iliopubic arch, aside from providing a base for muscle at- tachment, appears to be as a bottom tie for the dorsal arch, to counteract this horizontal thrust. Viewed from the side (fig. 60, D) the pelvis is not a simple arch as it is in reptiles. The acetab- ulum lies well behind the sacroiliac articulation, and upward thrust through the acetabulum is translated into a vertical rotational force around the sacroiliac articulation as a center; the coxa is cantilevered to the sacrum. The sacroiliac articu- lation is not normally fused in quadrupeds, but it is practically immovably fixed by the sacroiliac ligaments, often augmented by interlocking den- ticulations on the two articular surfaces. Thus, under loading, shearing forces are developed along the neck of the ilium — the axis connecting acetab- A. Alligotor Tronsverse iliosocrol orch of quadrupedal mammals CB). similar to orch Aliigotor Thrust T through acetabulum is transmitted directly to socroilioc joint 0^. The iliosocrol orch fufKtions OS simple orch , H-< >H B. Conis, Upword thrust T, T' through ocetobulo is resolved in transverse iliosocrol orch. This orch is also loaded from obove by the weight of the body, Vi_. Horizontal thfust, H, H, developed in the tronsverse orch by both T_ and W, is counteracted by the ventral iliopubic orch oct- ing OS a tie. Conis. Thrust T through acetabulum is translated into rototionol force R oround socroilioc joint 0 os o cen- ter. This produces o shear along the oxts 0-A, as indicated by x - xv Horizonol thrust H is developed dur- ing locomotion. Conis. Upword thrust T is tronsmitted directly to vertebral column through ilium and sacroiliac joint. A shear is pro- duced at ttw socroilioc joint and com- pression in the r»eck of ttw ilium. The socroilioc orch functions as o simple orch. Fig. 60. Forces acting on the pelvis in quadrupeds. A, transverse arch in a reptile, anterior view; B, transverse arch in a mammal, anterior view; C, transverse arch in a reptile, lateral view; D, cantilevered transverse arch of a mammal, lateral view; E, forces acting on mammalian pelvis in erect posture, lateral view. 109 110 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 ulum and sacroiliac articulation — and this is by far the most destructive force to which this part of the arch is subjected. The rotational force acting on the sacroacetab- ular axis produces a powerful rotational shear or torque on the sacroiliac articulation, similar to that on a bolt being tightened by a wrench. This force would tend to displace the anterior part of the articulation downward, the posterior part up- ward. The posterior upward force of the couple is counteracted by the firm union of the auricular surfaces of the sacrum and ilium. The anterior downward force is met by the shape of the sacrum, which is wedged between the ilia like an inverted keystone (fig. 60, B, a). This angle is about 15° in Canis, and rises to 40° or more in the Bovidae. In the bears and Ailuropoda, in which the articu- lation is synostotic, the angle approaches zero, and this is also true in the giant anteater {Myrmeco- phaga), where the joint is fused. During locomotion the sacro-iliac articulation is also subjected to momentary horizontal thrust (fig. 60, D, H) that tends to displace the ilium anteriorly on the sacrum. This force results from the anterior thrust of the hind legs, and is espe- cially evident during galloping or leaping, when the femur is nearly or quite in line with the sacro- acetabular axis, as is evident in Muybridge's (1957) photographs of horses and dogs. This force is counteracted by the wedge shape of the sacrum in the frontal plane : the bone is wider anteriorly than posteriorly. The plane of the auricular surface forms an angle with the mid-sagittal plane of 11-14° in Canis, Ursus, and Ailuropoda, and in a specimen of Bison this angle amounts to 34°. Forces on the Pelvis in the Erect Posture If a quadruped stands erect on its hind legs the forces acting on the pelvis are approximately dou- bled, since the pelvis then bears the entire weight of the animal. They are also significantly altered in direction. The transverse arch still functions as before, but the ilia are no longer cantilevered to the sacrum. The thrust is now along the sacro- acetabular axis (fig. 60, E, T). Instead of shear- ing forces along the sacroacetabular axis there is now compression. The rotational shear at the sa- croiliac articulation is converted into a simple shear, which is largely or entirely counteracted by the wedge shape of the sacrum. This is a stronger construction than in the quadrupedal posture, but most of the elasticity is gone; if the sacroiliac ar- ticulation fuses there is virtually no elasticity in the pelvis. Horizontal forces, i.e., forces approximately par- allel to the sacro-acetabular axis, predominate in burrowing animals that use their hind legs for brac- ing while digging. Thus the dominant forces act- ing on the pelvis in such forms are very similar to those in the erect posture, and this is reflected in a I striking similarity in pelvic architecture. Examination shows that seven features charac- terize the pelvis in mammals in which forces par- allel to the long axis of the pelvis predominate, i.e., those that stand erect and those that use their hind legs for bracing while digging. These are: 1. The wings of the ilia tend to shift into the frontal plane. 2. The pelvis is short anteroposteriorly. 3. The sacroiliac articulation is strengthened by includ- ing additional sacral vertebrae (increased area) and/or by strengthening the joint through interlocking bony processes, synostoses, etc. 4. The lateral diameter of the corpus of the ilium is increased, and it tends to become circular in cross section. 5. The pubo-ischiadic symphysis is greatly shortened. This reduction is in the anterior part of the symphysis. 6. The total number of sacral vertebrae is increased. 7. The tail is usually, but not always, shortened. In marsupials Elftman (1929) attributed the shape of the wing of the ilium anterior to the sacro- iliac joint chiefly to the "sizes of the three muscle masses whose areas of origin form its three borders — the erector spinae mesially, the gluteus medius and gluteus minimus dorso-laterally, and the ili- acus ventro-laterally." Waterman (1929) con- cluded that the form of the ilium in primates is largely determined by muscles. Elftman believed that in Vombatus, however, the width of the trunk is partly responsible for the lateral flare of the an- terior part of the ilium. In the bears and Ailuropoda the mass of the middle and deep gluteals is relatively no greater than in the cursorial dogs and cats (see Table 15). Even in man the relative mass of these muscles is no greater than in cursorial carnivores. The ilio- psoas in Ailuropoda is slightly heavier than in bears and dogs but it is smaller than in the lion, which has a notably narrow pelvis. In the lion the great size of the iliopsoas (almost identical with man) is associated with leaping. If the relative masses of the large muscles at- taching to the wing of the ilium are nearly con- stant, then differences in size, shape, and slope of the iliac wing must be attributable to other causes.' The most consistent character of the iliac wing in ' The long iliac crest (= broad iliac wing) characteristic of bears must be attributable to pecularities, still unknown, in the abdominal wall muscles and iliocostalis that attach to this crest. Elsewhere among carnivores the crest tends to be short in climbing and aquatic forms, "normal" in ter- restrial forms. Iliac Crest Descending Rannus Ischium Cams lupus Gulo luscus Procyon lotor Ursus orctos Ailuropoda melanoleuca Fig. 61. Anterior views of pelves of carnivores, to show angle of inclination of iliac and ischiadic planes. Ill 112 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 mammals in which forces parallel to the long axis of the pelvis predominate is that the wing tends to shift into the frontal plane (fig. 61). The iliac crest forms an angle with the frontal plane of 20 21° in Ailuropoda, 22° in a Mellivora, 28° in a Meles, and only 12° in a Vombatus. In the bears and American badgers the slope of the crest is about normal for terrestrial carnivores, 45-50°. In the cursorial wolf the slope approaches the verti- cal, 70-80° (fig. 61). The main advantage of a frontal position of the wing of the ilium is leverage; in both the erect and the burrowing posture the gluteals and iliacus are in an increasingly favorable position to stabilize the pelvis and vertebral column as these muscles approach the frontal plane. Waterman (1929) has discussed the relation between erect posture and the muscles attaching to the iliac crest in primates. The muscles attaching to this crest in Ailuropoda are shown in figs. 56-58; the corresponding rela- tions in other carnivores are unknown. Shortening of the pelvis is symmetrical, affect- ing the preacetabular and postacetabular regions about equally. The pelvis is almost as short in bears (index 36) as in the panda, and is only slightly longer in Meles (41) and Taxidea (41). Mellivora is a striking exception (index 50). The norm for terrestrial carnivores is about 46. The advantage of reduction in pelvis length with increased hori- zontal forces on the pelvis is not clear to me. Strengthening of the sacroiliac articulation with increase in horizontal forces on the pelvis is so ob- viously functional that it requires no comment. It reaches a maximum in the Myrmecophagidae, in which the sacroiliac articulation is supplemented by a strong sacroischiadic articulation occupying the normal site of the sacrotuberous ligament. In- creased diameter of the body of the ilium is like- wise associated directly with increased horizontal thrust; relative diameter of the body reaches a maximum in the Old World badgers. Shortening of the symphysis is invariably corre- lated with increased horizontal thrust on the pel- vis. It is seen in the wombat (Marsupialia), the extinct ground sloths and the anteaters (Edentata), the anthropoids (Primates), and in badgers and Ailuropoda among the carnivores. The symphysis is also short in aquatic forms: in the otters and particularly so in the seals. Various attempts, all more or less speculative, have been made to explain reduction in length of symphysis. All explicitly or implicitly regard sym- physis length as proportional to the forces the symphysis must withstand. Weidenreich (1913) attributed shortening of the symphysis in primates to the weight of the viscera and the pull of the sacrotuberous and sacrospinous ligaments drawing the pubic rami apart. Mijsberg (1920) suggested that vertical forces acting on the pelvis in quad- rupeds produce exorotation of the coxa around the sacrum, and that this exorotation is resisted by the symphysis, whose length is proportional to the exorotatory force. Mijsberg's interpretation is supported by the fact that the seals (Phocidae), in which vertical forces acting on the pelvis are negligible or absent, have no true symphysis. Elftman (1929) accepted Mijsberg's explanation, but suggested further that in Vombatus shorten- ing of the symphysis posteriorly is necessary to provide a proper outlet for the pelvis. Nauck (1938) believed he could detect a correlation be- tween dorsal shifting of the acetabulum — which he maintains would reduce the exorotatory forces on the pelvis — and reduction in symphysis length. Nauck's correlation exists only in selected cases, and obviously is not a general explanation. All investigators' agree that the iliopubic arch functions primarily as a tie to counteract horizon- tal thrust ("exorotatory forces") developed in the dorsal iliosacral arch. All agree further that re- duced symphysis length is somehow associated with reduced tensile stresses in the iliopubic arch. The resolution of vertical vs. horizontal forces within the pelvis has not been demonstrated ex- perimentally, however, and consequently all expla- nations are conjectural. A correlation between increased force parallel to the long axis of the pel- vis and reduced symphysis length remains as an empirical fact. Increased sacral length behind the sacroiliac ar- ticulation is associated with increased horizontal thrust on the pelvis both in forms that stand erect and in those that use their hind legs for bracing while digging. Extending the sacrum posteriorly increases the attachment area for the multifidus and sacrospinalis muscles. The main action of both of these muscles is to extend the vertebral column when acting on the vertebrae, or to extend the pelvis when acting on the sacrum. These ac- tions are obviously important for spinal fixation both in the erect posture and in burrowing. It seems likely that reduction in tail length is a consequence of increased sacral length, although critical data are lacking. If sacral length is in- creased to provide additional area for the spinal erectors, this area could be provided only at the expense of the basal tail muscles. The special cases of long sacrum associated with long tail in the anteaters and aardvark suggest a fundamental difference in either the spinal erectors or the caudal ' Braus (1929, p. 456) interprets the human pelvis as a ring under spring-like internal tension. DAVIS: THE GIANT PANDA 113 muscles in these forms, but pertinent data are lacking. The Pelvis of Ailuropoda. — The pelvis of the giant panda is notably different from that of the bears, which it resembles no more closely than it does the pelvis of several other arctoid carnivores. The bear pelvis, in turn, is unique among arctoids in its combination of long iliac crest, very broad iliac wing with normal slope in the transverse plane, and extremely long symphysis. The pelvis of Ailuropoda exhibits, to a far gi-eater degree than any other carnivore, the seven features that characterize the mammalian pelvis when forces parallel to the body axis predominate (p. 110). These forces predominate during burrowing, and when the animal stands erect on its hind legs. Ailuropoda is not a burrower, nor does it stand erect to any greater extent than do the bears. There is, in fact, no reason for believing that hori- zontal forces on the pelvis in Ailuropoda are gr-eater or more sustained than in Ursus or other carni- vores. This indicates that some other (non-adap- tive) factor is responsible for the form of the pelvis in Ailuropoda. The pelvis adjoins the lumbosacral region of the body axis. In this region in Ailuropoda the axial skeleton, the urogenital system, and the circula- tory system all show non-adaptive deviations from the norm. The most plausible explanation for the pelvic form in Ailuropoda is that it reflects the serious disturbance in the axial gradiant that is associated with cephalization (p. 84). 2. Femur The femur in the Carnivora serves for the origin or insertion of 22 muscles. Of these, 15 belong to the hip joint and 7 to the knee joint or lower leg and foot. In the Carnivora the form and archi- tecture of the femur are determined largely by the static requirements of support, to a far greater degree than for the humerus. Except for the tro- chanters, the external form of the femur is scarcely modified by the muscles that attach to it. It was found (Table 2) that if femur length is calculated against the length of three thoracic ver- tebrae, the femur in Ailuropoda is longer than the norm for carnivores but not so long as in Ursus. Relative femur length of the panda is similar to that of the cats, whereas the bear femur is among the longest known for the Carnivora, equal to Crocuta and exceeded only by Chrysocyon. If the position of the acetabulum remains rela- tively constant (as it does among arctoid carni- vores; see Table 11), then a long femur would re- sult in fast but weak movements of the femur around the acetabulum, as compared with a short femur.' From the standpoint of locomotor effi- ciency, the ratio between femur length and tibia length is much more significant than is femur length relative to pelvis length. The femur of Ailuropoda (fig. 62) is similar in form to that of Ursus and the Procyonidae, with a low greater trochanter and a straight shaft. As in most arctoid carnivores, the bone shows little torsion.^ In two wild-killed pandas the torsion angle is — 1° and — 3°; in a third, reared in cap- tivity, it is — 13°. The mean of twelve wild-killed arctoids is about 1°, extremes — 3 to +14. Four wild-killed Ursus range from — 2 to -t-14, mean +2°. The greatest torsion among arctoids is in the Procyonidae: 10 and 14 in two individuals of Procyon, 6 in a Nasua. Torsion in two cage- reared Ailurus is 3 and 12. In Ailuropoda the head of the femur is hemi- spherical, about 38 mm. in diameter, slightly larger than in a bear of comparable size. The fovea, for the ligamentum teres, occupies the same position as in Ursv^, but is wider and deeper. The neck is distinct, and forms an angle of about 130° with the shaft; it is slightly more angulated than in Z7rsMS (134-138°) or Proc|/ow (135°). Angu- lation of the neck is 125-140° in arctoid carnivores in general. The neck is narrower anteroposteri- orly but slightly wider dorsoventrally than it is in Ursus. The greater trochanter, which provides at- tachment for the middle and deep gluteals and the piriformis, does not differ significantly from that of Ursus. It is a broad knoblike structure scarcely rising above the level of the neck. Its anterior border is continued distally as a low crest that terminates at the level of the lesser trochanter in a prominent scar, the gluteal tuberosity, mark- ing the insertion of the superficial gluteal muscle. The trochanteric fossa, which receives the ten- dons of the obturator muscles, is deep and well defined. The lesser trochanter, on which the iliacus and psoas major muscles attach, is a low conical eminence projecting posteromedially, as in other arctoid carnivores. A crescent-shaped trans- verse scar extending across the posterior surface of the bone, from the lesser trochanter nearly to the gluteal tuberosity, marks the attachment of the quadratus femoris. ' Disregarding differences in tension and velocity of con- traction of muscles. See Maynard Smith and Savage (1 956) for methods of calculating relative mechanical advantages in limbs. * Torsion was measured by the method given by Schmid (1873). My figures do not always agree with his, and I sus- pect this is because many of his skeletons were from zoo animals. 114 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Lig. teres Jeiiioris Fossa Irochaiile Capauta arlictilaris ^***"^\ " TiiH-haiiler minor. M. iliacus & psoas major M. quadratus femoris M. vastus nied. M. adductor Capsitla articuhiris M. gastroc. (cap- nied.'i Lig. criic. M. pjTiformis Trochanter major M. glutaeus medius . M. obturator int. M. glutaeus prof. , M. obturator ext. M. glutaeus superf. M. pjTiformis Tuber, glulaea M. vastus lateralis M. adductor pars post M. adductor pars ant Epicomlijliis lateralis M. plant. & gastroc, (cap. lat.) Capsula arlicularis Colltim femoris M. iliacus & psoas major M. vastus intermedius Lig. coll. fihujare Plica synorialii patellaris Lig. cnic. ;)<)S( \M. poplitcus M. gastroc.(cap. med.) Lig, coll. libiale Capsula articttlaris M. e.\t. dig. longus Fig. 62. Right femur of Ailuropoda, posterior and anterior view. The shaft of the femur is nearly or quite straight ; it is convex anteriorly to a greater or lesser degree in Ursus and other arctoids. The anterior surface is faintly reticulated. As in other arctoids the shaft is wider transversely than anteroposteriorly: the ratio is about 80. The linea aspera on the poste- rior surface of the shaft is scarcely indicated, even less so than in Ursus. Slight roughenings on the proximal two thirds of the shaft mark the attach- ments of the pectineus and adductor muscles; these are wanting in the distal third of the bone where the femoral vessels are in contact with the bone. The anterior, medial, and lateral surfaces of the shaft are overlain by the vastus muscles, and are devoid of any modeling. The inferior end of the femur differs in details from that of Ursus. The condyles are roller-like, rather than ball-like as in Ursus and other arc- toids, and the intercondyloid fossa (in which the cruciate ligaments attach) is relatively broader. The lateral condyle is wider and longer than its fellow and its articular surface is more oblique. DAVIS: THE GIANT PANDA 115 The lateral epicondyle contains a large crater-like depression in which the lateral collateral ligament attaches; the plantaris and the lateral head of the gastrocnemius arise from the prominent superior rim of the crater. A dimple-like depression imme- diately below the crater marks the attachment of the popliteus. The median condyle is much narrower than the corresponding condyle in Ur- sus, as a result of encroachment by the intercon- dylar fossa. The medial epicondyle contains a large depression for the medial collateral ligament; its anterodorsal rim is elevated into a prominent tubercle on which the medial head of the gastro- cnemius arises. The patellar surface does not dif- fer from the corresponding area in Ursus and other arctoids. The femur of Ailuropoda thus differs from that of Ursus chiefly in details of modeling, torsion, and angulation — features that certainly represent post-natal adaptive adjustments. The only fea- ture that cannot be so interpreted is the relative length of the femur in Ailuropoda, which probably demands a genetic basis. I can find no mechan- ical explanation for the shortening of this bone in Ailuropoda relative to Ursus; the matter is dis- cussed further on p. 38. 3. Patella The patella (fig. 63) is ovate, about 37 mm. long by 32 mm. wide. It is relatively wider and more disk-shaped than the corresponding bone in Ursus, but is otherwise very similar. The anterior sur- face bears longitudinal striae. The ai-ticular surface is broader than high, and the lateral and medial articular facets are not clearly marked. The scar for the attachment of the quadratus femoris ten- don is prominent on the superior and lateral sur- faces, as is the attachment area for the patellar ligament on the anterior surface at the apex. 4. Tibia and Fibula The tibia and fibula are very short. These bones are also short relative to other limb segments in Ursus, and are very short in badgers (Table 2). Short distal segments result in relatively powerful but slow movements in the distal part of the limb. Hence the advantage of a low femorotibial index in graviportal animals and in digging forms that use the hind legs for bracing. The tibia (fig. 63) is basically similar to that of Ursus. It differs chiefly in being shorter and more compact, and in the greater torsion of the distal end. The head, which measures 65-70 mm. in transverse diameter, is relatively broader than in Ursus. The lateral condyle is about the same size as the medial, as in the bears. A crater-like depression on its lateral side, for the attachment of the lateral collateral ligament, is larger but shal- lower than in Ursus. The lateral articular surface is ovate, its anteroposterior diameter greatest; it encroaches on the anterior intercondyloid fossa more than in Ursus. The medial condyle pro- jects medially some distance beyond the border of the shaft. The articular facet for the head of the fibula lies farther posterior than in Ursus, but is otherwise similar. The medial articular surface is almost circular in outline. Both the anterior and the posterior condyloid fossae are wider than in the bear. The tibial tuberosity, on which the patellar ligament attaches, is prominent in Ursus but is scarcely indicated in Ailuropoda. The shaft of the tibia is almost straight. It is bowed very slightly medially, as in Ursus, and this bowing appears to be (but is not) exaggerated by the medial extension of the proximal and distal ends of the bone. This latter circumstance greatly increases the interosseous space between tibia and fibula, and the total width across the leg (from medial border of tibia to lateral border of fibula). The shaft is most slender near the middle, flaring somewhat both proximally and distally. The an- terior crest, which is associated with the insertions of the gracilis, sartorius, biceps, and semitendino- sus, is well marked, especially proximally; it con- tinues distally into the medial malleolus. The interosseous crest on the lateral surface of the shaft, on which the interosseous membrane at- taches, is a prominent ridge beginning below the lateral condyle and extending down to the distal fibular articulation. On the posterior surface of the shaft several ridges mark the boundaries be- tween the flexor hallucis longus, the tibialis poste- rior, and the popliteus (fig. 63). The distal end of the tibia is very similar to that of Ursus, except that it is rotated farther on the shaft; the torsion angle of the transverse axis of the distal end against the bicondylar axis of the proximal end is 35°-48° in Ailuropoda, whereas in Ursus it is only about 20°. The transverse axis also is inclined more obliquely with respect to the long axis of the bone: about 120° in Ailuropoda, about 105° in Ursus. The medial malleolus is short and wide anteroposteriorly. A deep groove, the sulcus malleolaris, on its posterolateral surface lodges the tendon of the posterior tibial muscle; a similar groove is present in Ursus. The inferior articular surface, which articulates with the as- tragalus, is ovate, wider medially and narrower laterally than in Ursus. It bears a median ridge, bounded on either side by a depression, that fits a corresponding surface on the astragalus. At the lateral end of the articular surface is a small 116 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. vastus lato^is M. rectus femoris Lig. paleUae M. tibialis a(it Condylus lot M. peronaeus longus M. soleus Lig. eoll. fibulare M. flex, hallucis longus M. peronaeus tertius M. tibialis ant. M. vastus med. M. sartorius Lig. patellae M. semimembranosus Condylus med. M. peronaeus brevis M. ext. hallucis longus M. peronaeus longus CapstUa artieularis M. popliteus M. flex, hallucis longus M. tibialis post. M. gracilis Sulcus malleolaris Malleolus med Capsula artieularis M. tibialis post. M. flex. dig. longus M. soleus M. tibialis post. M. flex, hallucis longus M. peronaeus bre\TS Capsula artieularis Malleolus lot. M. tibialis post.-/^ M. peronaeus brevis' vM. peronaeus tertius Fig. 63. Right patella, tibia, and fibula of Ailuropoda; anterior and posterior views. obliquely situated articular facet for the distal end of the fibula. The fibula (fig. 63) is slightly heavier than the fibula of Ursus, and is bowed slightly laterally, which further exaggerates the transverse diameter of the leg. It articulates with the tibia by a syno- vial joint at each end and therefore, as in Ursus, represents the mobile type of fibula. The head is an expansion of the proximal end differing from that of Ursus only in minor details. The articular facet is a flat ovate surface, set obliquely and directed medially and posteriorly. No scar marks the attachment of the lateral col- lateral ligament on the lateral surface immediately below the head. The shaft is triangular in cross section throughout most of its length, but is con- siderably flattened distally. Almost its entire sur- face provides attachment for muscles, of which seven arise from the shaft, and roughened longi- tudinal elevations on the shaft mark the attach- ments of aponeuroses and intermuscular septa separating many of these muscles. The most con- spicuous crest, on the medial surface, is the inter- osseous crest to which the interosseous membrane attaches. The distal end of the fibula is an irregular ex- pansion, larger than the proximal expansion, that forms the lateral malleolus. It is relatively larger and heavier than the lateral malleolus of Ursus, but is otherwise comparable. The lateral malle- M. tibialis ant- Os cuneiforme 1 Os cuneiforme 2. Os naviculare Os sesamoid, tib. M. flex. dig. quinti brevis peronaeus brevis Os cuneiforme 3 M. abductor dig. quinti Os cuboideum M. tibialis post Capsula articularis Calcaneus Fig. 64. Right tarsus and metatarsus of Ailuropoda, dorsal view. M. flex. dig. quinti brevis M. abductor dig. quinti M. peronaeus longus M. quadratus plant. Tendo m. plantaris M. soleus & gastrocnemius M. peronaeus longus M. tibialis ant. M. flex, hallucis irevis M. abductor dig. quinti Tendo m. plantaris Fig. 65. Right tarsus and metatarsus of Ailuropoda, plantar view. 117 118 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 olus does not project so far distad as the medial malleolus, and its articular surface is less extensive anteroposteriorly. The lateral surface bears a prominent elevation, the processus lateralis fibu- lae (new name), that separates the peroneal ten- dons into two groups; the long peroneal tendon lies immediately anterior to the process, while the tendons of the brevis and tertius lie immediately behind it. In Ursus this process is a sharply pro- jecting knob-like structure, and in other carnivores (except the Canidae) it forms a hook that arches backward over the tendons of the peronaeus brevis and tertius. The medial surface of the malleolus bears two articular surfaces: a smaller proximal one facing proximally and medially that articu- lates with the tibia, and a larger distal one facing distally and medially that articulates with the astragalus. 5. Tarsus The tarsus (figs. 64, 65) is in general more con- servative than the carpus. The tarsus of living procyonids actually differs little from that of the more generalized Paleocene creodonts, in which it is adapted to arboreal life ( Matthew, 1937, p. 317) ; and the tarsus of modem bears is strikingly sim- ilar to that of the Middle Paleocene creodont Claenodon. In the bears the ankle shows a char- acteristic shortening and broadening of all the tareal bones; this is also evident, though less pro- nounced, in Ailuropoda. The tarsus of Ailuropoda is, in fact, moi-phologi- cally more "primitive" than that of Ursus. This is seen in the less pronounced broadening of all the tarsal bones, in the presence of a large astragalar foramen, and particularly in the form of the two transverse ankle joints — the transverse tarsal and tarsometatai^sal joints (fig. 65). In Ltsws both of these joints are essentially continuous across the ankle (as in man), whereas in Ailuropoda and gen- eralized carnivores both joints consist of two or more transverse segments offset from each other. The ursid-human form of these joints is a second- ary adaptation to plantigrade walking, whereas the interrupted joints seen in more generalized car- nivores increase the lateral stability of the tarsus. A unique feature of the tarsus of Ailuropoda is the extraordinarily loose fit between the astragalus and calcaneus. The lateral and medial articular surfaces of the two bones cannot be brought into congi'uence at the same time, but only alterna- tively by sliding the astragalus sideways over the calcaneus. In association with this, the diameter across the two articular surfaces on the astragalus gieatly exceeds the diameter across the correspond- ing surfaces on the calcaneus. The astragalus (BXA: talus) (fig. 66) is rela- tively larger than in Ursus, but differs chiefly in its longer neck and narrower head, and in the presence of a large astragalar foramen. The troch- lea is broader than long, and is characterized by a very shallow gi-oove and relatively small malleolar surfaces; the upper tarsal joint is less secure and permits greater lateral rotation than in Ursus. The superior articular surface is not continued posteriorly over the posterior process as in pro- cyonids. The arc of the trochlea is thereby re- duced by about 35°; it measures about 165° in Ailuropoda and Ursus, and about 200° in procy- onids. The medial malleolar surface, which in Ursus extends over the neck nearly to the margin of the head, reaching as far distad as does the lateral malleolar surface, is much shorter in Ailuropoda, ending at the base of the neck. The lateral mal- leolar surface is similar to that of Ursus except that it is flatter. Immediately posterior to the trochlear gi-oove there is a large astragalar fora- men in all specimens examined. This foramen, characteristic of creodonts, occurs sporadically among generalized modem procyonids iBassaris- cus) and mustelids {Gulo, Taxidea); I also find a small astragalar foramen in one specimen of Ursus americanus. Behind the trochlea a deep groove for the flexor hallucis longus tendon is present in Ursus and other carnivores; this groove is want- ing in Ailuropoda. On the inferior surface the lateral (posterior in human anatomy) and medial articular surfaces, articulating with con-esponding articular surfaces on the calcaneus, resemble those of Ursus. They are oblong, relatively shallowly concave areas sep- arated by a deep astragalar gi-oove. The lateral is more extensive than the medial, and in Ailu- ropoda their axes diverge slightly distally. As in other camivores, the medial articular surface lies mostly beneath the neck of the astragalus. Of the accessory facets (Davis, 1958) only the anterior marginal facet of the medial articular surface is represented. It is a narrow extension of the medial surface, continuous anteriorly with the na\-icular articular surface, and it rests on the cuboid. The head and neck, on the contrary, resemble those of procyonids and generalized mustelids and viverrids more closely than they do those of Ur- sus. The neck is relatively long, nan-ower than in Ursus, and deflected toward the medial border of the foot, foiTning an angle of about 98° with the transverse axis of the trochlea. The head bears two articular surfaces, as in other camivores: an oval convex area anteriorly and medially for the navicular, and a small triangular area inferiorlj' DAVIS: THE GIANT PANDA 119 Ailuropoda Ursus Fig. 66. Opposing surfaces of right astragalus and calcaneus of Ailuropoda and Ursus arctos. and laterally for the cuboid. On the inferior sur- face, immediately behind the navicular articular surface, a deep pit marks the attachment of the talocalcaneal interosseous ligament. The calcaneus (fig. 66) is longer and more slen- der than in Ursus. On the superior surface the lateral articular surface is an elongate oval, ex- tending farther posteriorly than in Ursus. As in bears, it describes a continuous spiral track: an- teriorly it faces slightly laterally, while its poste- rior end is almost vertical, facing medially. This articular surface is scarcely curved in cross section, and the curvature along the long axis is relatively slight; in this flatness the bears and panda differ sharply from other carnivores. The medial artic- ular surface is a flat discoidal area on the superior surface of the sustentaculum. As in Ursus, the posterior end of this articular surface is deflected sharply downward, forming an angle of almost 90° with the main articular surface. This arrange- ment, which is present in Nasua and indicated in Gulo but is wanting in other carnivores, increases stability of the lower tarsal joint at the expense of mobility. The medial articular surface is continued ante- riorly into a narrow accessory facet that extends forward to the anterior border of the calcaneus, articulating with the anterior marginal facet of the astragalus. This accessory facet, which increases the stability of the lower tarsal joint, is present in most, but not all, carnivores. 120 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Opposite the sustentaculum the lateral surface of the calcaneus is produced into a prominent projec- tion, the coracoid process (Baum and Zietzschmann, 1936), from which arise the extensor digitorum brevis and quadratus plantae muscles. In Ursus the coracoid process is a long shelf-like structure extending posteriorly to the posterior border of the lateral articular surface, while in other carni- vores it is less extensive. The cuboid articular surface is more oblique than the coiTesponding surface in Ursus but is other- wise similar. The posterior end of the calcaneus is expanded into a knob-like structure. Almost the entire posterior face is occupied by a large de- pressed scar that marks the attachment of the tendo Achillis and its associated bursa. The navicular articulates with the astragalus, the cuboid, and the three cuneiforms, as in Ursus and other carnivores. The posterior surface is composed almost entirely of a large ovate concave articular facet that receives the head of the astrag- alus (fig. 64). The anterior surface is convex, its superior part indistinctly divided into three facets for the three cuneiform bones; inferiorly it is roughened at the attachment site of the plantar naviculari-cuneiform ligaments. On the medial surface a smooth prominence marks the articula- tion site of the tibial sesamoid. A narrow articu- lar facet on the inferolateral surface articulates with the cuboid, and immediately mesad of this on the inferior surface is a rounded prominence, the navicular tuberosity. The cuboid resembles that of Ursus in shape, but is relatively longer and narrower.' Its poste- rior surface presents a rectangular convex articular surface for the calcaneus (fig. 64) ; its anterior sur- face bears a slightly concave surface for the fourth and fifth metatarsals, a faint ridge dividing the two areas. Its medial surface presents two artic- ular surfaces, a vertical surface posteriorly that articulates with the head of the astragalus, and an irregularly shaped surface that articulates with the navicular and the third cuneiform. The inferior surface bears a prominent transverse ridge for the attachment of the long plantar ligament. The cuneiform bones articulate with the na- vicular posteriorly and the first three metatarsals anteriorly. The first is the largest ; the tibial sesa- moid articulates partly with its posteromedial cor- ' The tarsus and pes are relatively broader in bears than in other carnivores. It is interesting and suggestive that the relative breadth of the cuboid increases with absolute size in the genus Ursus. The ratio breadth 'length X 100 in a series of bears is: Ursus americanus 81, U. arctos 92, U. gyas 95. The only available specimen of U. spelaeus is a shade smaller than my very large U. gyas and has a ratio of 94. The corresponding ratio for Ailuropoda is 64. ner. The third cuneiform articulates laterally with the cuboid. The tibial sesamoid is relatively much larger than in Ursu^ (fig. 54); it measures 20 mm. in length by 13 in breadth. As in other carnivores, it articulates with the navicular and first cunei- form. The bone is flattened from side to side. The tendon of the posterior tibial muscle inserts on its posterior border, and a part of the flexor hallucis brevis muscle arises from its medial face (fig. 102). 6. Pes The metatarsals decrease in length from the fifth to the first; in Ursus the fourth is the longest, and in procyonids the third and fourth are sub- equal. As in Ursus, the metatarsals are short. The fifth is relatively heavier than in Ursus, but the others are of comparable size. As in other carnivores, the proximal end of the fifth metatar- sal bears a prominent lateral process to which the tendon of the peroneus longus and brevis and the abductor digiti quinti attach. j As in the manus, the distal articular surfaces of ' the metatarsals are narrower than in Ursus, and the median ridge is more prominent. The phalanges are similar to those of Ursus, relativelj- shorter than those of the procyonids. In the proximal row a pair of elevations on the inferior surface of each bone, near the distal end, marks the attachment of the interosseous muscles. A conspicuous pit-like excavation on the inferior surface of each bone of the middle row, immedi- ately behind the trochlea, receives the large plantar process of the terminal phalanx. A pair of sesamoid bones is present beneath the metatarsophalangeal articulation of each digit. There are ten in all. B. Review of the Hind Leg The bones of the hind leg of Ailuropoda, like those of the fore leg, agree with the corresponding bones of Ursus in all essential respects. As in the fore leg, differences in details of modeling, torsion, and angulation probably represent postnatal re- sponses to stresses extrinsic to the bone tissue itself. Relative lengths of limb segments agree with the proportions in graviportal animals. This suggests that limb proportions in Ailuropoda are broadly adaptive, although the animal is much too small to be truly graviportal and the adaptive signifi- cance, if any. of the limb proportions is not clear. Short distal segments result in relatively powerful but slow movements in the distal part of the limb. Hence the advantage of a low femorotibial index to heavy graviportal animals and to digging forms that use the hind limbs for bracing. Length of Cloenodon cofruqatus Potos flavus Ursus arctos Ailuropodo melonoleuca Fig. 67. Right tarsus and pes of representative carnivores. The small inset to the left of Claenodon corrugatus is Claenodon montanensis (Bull. U.S. Nat. Mus., 169). 121 122 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 long bones is certainly gene-controlled, but the mechanism of such control is unknown. Fuld (1901) demonstrated a slight but signifi- cant increase in tibia length in dogs that had been bipedal since puppyhood. Colton (1929) found that in the rat, on the contrary, bipedalism results in a slight increase in femur length. In neither the dogs nor the rats was the difference anywhere near as great as the difference between the relative lengths of these bones in Ailuropoda and Ursus. The hypertrophied tibial sesamoid is a product of natural selection, but of selection acting on the radial sesamoid. The fact that the tibial sesamoid has hypertrophied along with the radial sesamoid shows that these two bones are homeotic from the genetic standpoint as well as serially homologous from the morphological standpoint. Thus only one (presumably) adaptive feature in the bones of the hind leg — the relative lengths of the long bones — appears to result directly from natural selection acting on the bones themselves. Even this does not appear to be adaptive and may be a pleiotropic effect. VII. DISCUSSION OF OSTEOLOGICAL CHARACTERS It is evident from the foregoing description that the skeleton supplies abundant and convincing evidence that Ailuropoda is much closer to the Ursidae than to any other group of living carni- vores. Missing "ursid characters" have been partly responsible for disagreement among mammalogists as to the affinities of Ailuropoda (e.g., Mivart, 1885a; Weber, 1928). It is now obvious that these missing characters have been obliterated by phylo- genetically recent factors that for the most part are extrinsic to the skeleton. The most important of these extrinsic factors is hypertrophy of the skeletal musculature. Yet, despite close similarity in all essential respects, the panda skeleton differs from the bear skeleton in a number of very puz- zling ways. It is the interpretation of these differ- ences that is pertinent to our central problem. The panda skeleton resembles the bear skeleton in all essential respects. The bear skeleton itself differs from the general- ized carnivore condition in a number of features that cannot be interpreted as adaptive, and I am certain that many ursid characters represent what Griineberg (1948) has called "subordinated gene effects" — effects that are genetically, physiologi- cally, or even mechanically connected with a pri- mary gene effect on which natural selection has operated, without themselves being adaptive. Such non-adaptive characters might persist indefinitely if selection against them is less intense than selec- tion for the primary effect. Among the most con- spicuous of these in the bears are limb proportions, curve of moments of resistance in the vertebral column, shortness of lumbar region, shortness of tail, length of sacrum, form of pelvis. Ailuro- poda shares most of these characters with Ursus, and has superimposed additional features, likewise mostly non -adaptive, on the ursid pattern. Many of the differences between panda and bear skeletons are adaptive, but their cause is extrinsic to the bone itself; that is, they merely reflect the response of the bone tissue to external pressures, stresses and strains, and other purely mechanical factors. In the absence of the appropriate stimu- lus such characters fail to appear. Among such features are the surface modeling of bones, tor- sions, form and extent of articular areas, and size and position of foramina. These are characteristic features of the skeleton of Ailuropoda, and they may be clearly adaptive in the sense of promoting the efficiency of the organism, but they are epi- genetic to the bone and therefore are not the result of natural selection on the skeleton. The most conspicuous way in which the skeleton of Ailuropoda differs from that of Ursus is in a general increase in the quantity of compact bone throughout the entire skeleton. Except for mas- ticatory requirements, no differences from the habits of bears would demand such increased thick- ness of compacta for mechanical reasons. A gen- eralized effect of this kind, involving an entire tissue and with sharply localized advantage to the organism, would almost surely have a single cause. Comparable generalized effects, involving the whole skeleton and localized in a single genetic factor, are well known in laboratory and domestic ani- mals (Stockard, 1941; Griineberg, 1952; Klatt and Oboussier, 1951). Wherever they have been ana- lyzed, it has been found that such effects are me- diated through the endocrine system. We may postulate that in the panda, because of mastica- tory requirements, selection strongly favored in- creased thickness of compacta in the skull. This increase was actually achieved via a process that results in generalized thickening of the compacta throughout the skeleton. The functionally un- necessary increase of bone tissue in the postcranial skeleton is no great disadvantage because of the non-predatory habits of this species, which places no premium on speed and agility. The most significant feature in the panda skeleton is a generalized, increase in the quantity of compact bone. This probably has an extremely simple genetic base. The increased thickness of compacta is advan- tageous only in the skull. DAVIS: THE GIANT PANDA 123 Many proportions in the skeleton of the panda — and to a lesser extent in the skeleton of bears — are a mixture of those seen in bipedal, in burrow- ing, and in graviportal forms. In part these pro- portions are mutually contradictory — adaptations associated with bipedalism are not the same as those associated with graviportalism — and in part they are not contradictory, since adaptations for withstanding anteroposterior thrust are similar in bipedal and burrowing forms. Still other propor- tions in the panda, particularly in the limbs, cannot be reconciled with any mechanical requirements, and appear to represent disharmonious relations of the "subordinated gene effect" variety. The fact is that the panda does not burrow, it is bi- pedal only to the extent that, like many other mammals, it occasionally stands erect for short periods, and it is not heavy enough to qualify as graviportal. These facts show that the ill-assorted features distinguishing the postcranial skeleton of the panda from that of Ursus are not truly adap- tive, and that where they agree with conditions that presumably are adaptive in other specialized forms (bipedal, fossorial, graviportal) such agree- ments are either fortuitous or based on something other than functional demands. Thus we are confronted with a highly modified and strongly adaptive skull associated with a con- siderably modified postcranial skeleton in which the departures from the "ursid norm" appear to be completely non-adaptive, even inadaptive to the extent of producing a disharmonious organism. From what is known of the genetics of acromegaly, achondroplasia, and other pathological conditions of the skeleton in dogs and mice (Stockard, 1941; Griineberg, 1948) the most economical interpreta- tion, consistent with all known facts, of the syn- drome of non-adaptive features in the skeleton of Ailuropoda is that they are associated pleiotropi- cally with the one definitely adaptive feature. It is even highly probable that the whole complex has a very simple genetic base. The persistence of such morphological dishar- monies in a natural population is unusual but not unique, and might in fact be anticipated in highly specialized forms whose adaptive niche places a low premium on all-around mechanical efficiency. Similar disharmonies are clearly evident in the hyenas, which like Ailuropoda are highly special- ized for masticatory power but do not need speed or agility either to escape from enemies or to cap- ture prey. It is suggestive that bipedal, fossorial, and gravi- portal mammals are all characterized by local strengthening of the skeleton (i.e., by increase in quantity of compacta). The changes in form and proportions associated with such local strength- ening are presumptively adaptive, and in some instances it can be shown unequivocally that they are — the moments of resistance in the vertebral column of bipedal forms, for example. In other instances attempts at a functional explanation have been unsuccessful; for example, pelvic archi- tecture in bipedal and burrowing forms. In many instances no functional explanation has even been attempted; for example, limb proportions in gravi- portal forms. If it can be demonstrated that cer- tain features in the skeleton are correlated with increased quantity of compacta rather than with other functional requirements, then an association between such features and a particular functional requirement is merely a chance association. At- tempts to read adaptive significance into such associations are, of course, based on a false as- sumption and can only lead to false conclusions. The existence in the panda skeleton of numerous ill-assorted conditions convergent with conditions in bipedal, fossorial, and graviportal forms sug- gests that such spurious correlations with func- tional requirements may be more common than has been assumed. Much more data are required to prove this suggestion. Numerous ill-assorted disharmonies in the post- cranial skeleton of the panda are connected pleiotrop- ically, as subordinated gene effects, with the increase in quantity of compacta. One other feature in the skeleton demands atten- tion: the specialized and obviously functional ra- dial sesamoid. It was concluded (p. 183) that all that would be required to derive this mechanism from the radial sesamoid of Ursus is simple hyper- trophy of the bone. This symmetrical increase in the dimensions of a single bone is quite a different thing from the hypertrophy of the compacta seen elsewhere in the skeleton. The localized remodeling seen in the sesamoid surely has a specific genetic base, as is strongly indicated by the "sympathetic" hypertrophy of the tibial sesamoid. The parallel and non-functional hypertrophy of the tibial sesa- moid also indicates that the genetic mechanism is a very simple one, perhaps involving no more than a single gene. The highly specialized and obviously functional radial sesamoid has a specific, but probably very simple, genetic base. Disregarding any minor polishing effects of nat- ural selection, aimed at reducing disharmonious relations, it appears that the differences between the skeleton of Ailuropoda and that of Ursus could be based on no more than two gene effects. There is, of course, no way of proving that the situation actually was so simple, but mechanisms capable 124 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 of producing comparable eflfects on the skeleton have been demonstrated experimentally in other mammals. The alternative explanation — numer- ous small gene effects screened by natural selection — postulates a vastly more complex process, and leaves unexplained the many clearly inadaptive features in the skeleton. We could, of course, assume that these several inadaptive features ap- peared one by one during the evolution of Ailuro- poda, and persisted simply because there was little or no selection against them. But if each of these is unconnected with any of the other gene effects, then any selection pressure would have eliminated them. Obviously there is some selection against any inadaptive feature; no feature is truly adap- tively neutral. Therefore it seems to me that probability strongly favors a single gene effect as the causal agent for all the hereditary differences between the skeleton of Ailuropoda and Ursus, except in the radial sesamoid. The major features distinguishing the skeleton of Ailuropoda from that of Ursus may depend on as few as two gene effects. These are: (a) Generalized hypertrophy of compacta. (b) Specific hypertrophy of the radial sesamoid. VOL CONCLUSIONS 1. The skeleton of Ailuropoda resembles the skeleton of Ursus in all essential respects. 2. Many skeletal differences between Ailuro- poda and Ursus are epigenetic to the bone tissue, and therefore do not result from natural selection on the skeleton. 3. The most significant feature in the panda skeleton is a generalized increase in the quantity of compact bone. This probably has an extremely simple genetic base. 4. The increased thickness of compacta is ad- vantageous only in the skull. 5. Numerous ill-assorted disharmonies in the postcranial skeleton are connected pleiotropically, as subordinated gene effects, with the increase in quantity of compacta. 6. The highly specialized and obviously func- tional radial sesamoid has a specific, but probably very simple, genetic base. 7. Thus, the major features distinguishing the skeleton of Ailuropoda from that of Ursus may depend on as few as two genetic factors. These are: (a) generalized hypertrophy of compacta; (b) specific hypertrophy of the radial sesamoid. DENTITION I. DESCRIPTION The classification of mammals has depended more on the dentition than on any other single feature of morphology. The teeth of the giant panda have repeatedly been studied and discussed in great detail (Gervais, 1875; Lydekker, 1901; Bardenfleth, 1913; Gregory, 1936; McGrew, 1938). These studies have led to the most divergent views as to the homologies of the various cusps, and in- ferences as to the affinities of Ailuropoda based on such homologies. I conclude that the cheek teeth of Ailuropoda are so modified from those of any other known carnivore that interpretations based on them have been largely subjective. The dental formula of Ailuropoda is If CI Pi M§=42, which is the primitive form for the recent Carni- vora. The formula is the same in small species of Ursus, but various additional teeth have been lost in large species of Ursus and in other genera of the Ursidae. In the Procyonidae and Ailurus the third lower molar has disappeared, giving the formula U C{ Pi Mt=40. The incisors are in no way remarkable in Ailu- ropoda. As in carnivores in general, in both jaws they increase in size from the first to the third. As in Ursus (much less so, if at all, in other arc- toids), the third incisor in both jaws is abruptly larger than the second, and in the upper jaw is less chisel-shaped and more caniniform than the two more medial incisors. The third incisor is relatively larger in Ailuropoda than in Ursus, and is separated from the canines by a very short dia- stema. The shortness of the diastema is the only evidence of crowding in the anterior dentition. The incisors are, of course, single rooted. The six incisors in each jaw are closely crowded, their combined occlusal surfaces forming an essen- tially continuous, slightly arched, scraper or chisel edge. The resulting tool, lying between and often slightly in front of the canines, is one of the most characteristic features of the dentition of the Car- nivora. The canines are more robust than in Ursus, in both long and transverse diameters. Their rela- tive length is almost identical in bear and panda, however, and this gives the canines of Ailuropoda a relatively stumpy appearance. In the unworn dentition there is a vertical ridge on both anterior and posterior surfaces of the upper canine, and on the posterior surface only of the lower. Similar ridges are seen in other arctoids (e.g., Procyon), but not in Ursus. Ailurus, along with the pro- cyonids Bossoncyon and Potos, has vertical grooves on its canines. The phylogenetic and functional significance, if any, of these surface sculpturings is unknown. The upper canine in Ailuropoda projects for- ward at an angle of about 30°. The same tooth forms an angle of about 15° in Ursus, while in other arctoids examined it does not deviate more than a couple of degrees from the vertical. The premolars increase in size from the first to the fourth, as in all arctoids. The first premolar is degenerate and peg-like in both jaws, and is often missing. In size and structure it contrasts sharply with the remaining premolars. The re- maining three teeth are crowded, and in both up- per and lower jaws P2 is rotated at an angle of about 30° from the axis of the tooth row. In the upper series, P'^ is tri-lobed, two-rooted, and with no indication of internal (lingual) cusps. P ^ is very similar to P \ except in size. The fourth upper premolar, the upper carnassial of the Car- nivora, has been the chief object of discussion and speculation in the dentition of the giant panda. It is the largest of the premolars, but is neverthe- less considerably smaller than the two upper mo- lars. The tooth exhibits five prominent cusps arranged in two longitudinal rows. The three on the labial side are considerably higher than those on the lingual, with the central one the highest of all. These have been homologized, from front to rear, with the parastyle, paracone, and metacone. The two cusps on the lingual side are regarded as the protocone (anteriorly) and the hypocone (pos- teriorly). There are no cingula. The tooth has three powerful roots, arranged in the form of a triangle. The anteriormost root supports the para- 125 126 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Paracone MettKone AtUero-uiternal cusp Protocovid Hypocotiid Protoconid tii- liypoconid h h I, Protoconid- Hypoconid- Fig. 68. Occlusal views of unworn right upper and left lower dentitions of Ailuropoda (XI). style, the protocone, and the anterior half of the paracone. The posteroexternal root supports the metacone and the posterior half of the paracone. The posterointernal root supports the hypocone. Five cusps, occupying similar positions, are found on the upper carnassial of Procyon and Ailurus. The mode of origin of the Procyon crown pattern from the primitive three-cusped shearing carnas- sial is well known and documented by fossil mate- rial (McGrew, 1938). Morphologically the crown pattern of Ailuropoda is very similar to that of Procyon, but the relation of the cusps to the roots is quite different and essentially nothing is known of the history of this tooth in the panda. The form of the crown in Ailurus is like that of Pro- cyon and Ailuropoda, and the relation of cusps to roots is like that of Procyon, not like that of Ailuro- poda. It has been assumed by Lydekker, Gregory, and McGrew that the morphological similarity in cusp arrangement between the raccoon and the giant panda denotes homology and hence common ancestry. Winge and Bardenfleth, on the con- trary, argued that the different cusp-root relations show that the similarity in crown pattern is the result of convergence. In the Ursidae, by contrast, P^ is degenerate by normal carnivore standards. It is relatively and absolutely small, abruptly smaller than the mo- lars, and its crown usually bears only three cusps: the paracone, metacone, and protocone; in small individuals of Ursus (e.g., U. americanus) there may be a minute parastyle. The relation of cusps to roots is identical with Ailuropoda. The lower premolars are simpler and more uni- form than the upper, but otherwise follow the trend of the latter closely. As in the upper jaw, the first premolar is small and peg-like and con- trasts sharply with the following teeth. It is often DAVIS: THE GIANT PANDA 127 missing. Pj,, like its antagonist, is rotated at an angle of about 30° from the axis of the tooth row. P„ , increase in size from front to rear, but P, * is much smaller than the following M,. All are trilobed, with three conical cusps in series along the axis of the tooth. P^ has a small postero- internal cusp, sometimes subdivided into several f small tubercles. Thus the posteriormost part of the lower premolar series is slightly broadened. All the lower premolars, except Pj, are two-rooted. The upper molars are enormous and richly cuspidate, dominating the entire upper tooth row. They are abruptly and conspicuously larger than the upper premolars, and are closely crowded. M ' is almost square, slightly broader than long. It bears two prominent conical cusps, the para- cone and the metacone, on the labial side. Lin- gually and directly opposite these there is a second pair of smaller and lower cusps, the protocone and hypocone. A third pair of poorly defined cusps is situated in the valley between the outer and inner rows of cusps; the homology, if any, of these cusps is unknown. The internal border of the tooth forms a broad shelf-like cingulum whose occlusal surface is very regularly serrate. M^ is divided into two subequal parts, a trigonid anteriorly and a large talonid posteriorly. The anterior part of the tooth is very similar to M Mn form and arrange- ment of cusps and cingulum. The occlusal surface of the talonid is richly tuberculate, with a long blade-like cusp medially (immediately behind the protocone, and perhaps representing an elongate hypocone) and a narrow cingulum. There are three roots on M ' , two lateral and one medial, as is typical of the Carnivora. The medial root is greatly expanded anteroposteriorly, and is partly divided by a groove into two pillars that lie be- neath the two medial cusps. M'-, in addition to the usual three roots, has a fourth large root sup- porting the talon. The upper molars of Ailuropoda are fundamen- tally similar to those of Ursus, but they differ in two seemingly important respects: their relatively larger size, especially their greater breadth; and the rich development of secondary tubercle-like elevations. The extinct European cave bear, Ur- sus spelaeus, reached a larger size than any other known member of the genus Ursus, and hence had the largest molars. It is therefore extremely sug- gestive that the molars of the cave bear, while retaining their ursid outlines, exhibit the same rich development of secondary cusps and tubercles as is seen in the giant panda. The similarity of the molars in these two forms, except for the broaden- ing of the crown in Ailuropoda, is quite astonishing. The lower molars are simpler and less broadened than the upper. M^, the lower carnassial, has lost its sectorial character and is quite similar to the corresponding tooth in both Ursus and Pro- cyon. There are five cusps, which retain the prim- itive arrangement (fig. 68). The facing slopes of the entoconid and hypoconid exhibit low tubercle- like elevations similar to the medial row of cusps on M ' , but these are lacking between the proto- conid and metaconid. There is a poorly defined cingulum externally. M ^ is more tuberculate than Mj, and the cusps are less sharply defined. The paraconid, which is prominent on Mj, cannot be identified with certainty on M ^ . This cusp is often almost completely coalesced with the protoconid in Ursus. It is also associated with the proto- conid in Ailurus, but there is no indication of it in Procyon. M., has a rounded triangular outline in Ailuropoda, and the cusps are almost completely obliterated on its flattened crown. The occlusal surface, which opposes the talon of M'^, is thrown up into a complex pattern of low tubercles. The outline and crown pattern of M 3 in Ailuropoda are quite different from the more typically molariform Mg of Ursus. It is noteworthy, however, that Rode (1935, pi. 7) illustrates, as "abnormal" ex- amples, several lower third molars of the gigantic Ursus spelaeus and these are almost exactly like M., of Ailuropoda. II. DISCUSSION OF DENTITION It has long been the custom of systematists to regard individual teeth, and even individual cusps, as the basic units of the dentition. Thus, by im- plication, these units are construed as individually gene-controlled and therefore subject to individual selection. The tooth as a whole, to say nothing of the dentition as a whole, would then be a mosaic of individually derived elements, each of which survives or perishes according to the way in which it functions in the dental activities of the animal. Similarities between adjacent teeth are ascribed to convergence resulting from selection. Such a view naturally places great emphasis on "homol- ogies" between cusps and similar elements as in- dicating affinities between animals. Furthermore, the minute structure of each tooth is perforce directly correlated with function. In practice, the teeth are minutely scrutinized and compared, element by element, for similari- ties in structure. Identity or near identity in architecture is construed as an infallible indicator of relationship, and vice versa. Certain teeth (P^ in the Carnivora) are often assumed to be better indicators of affinities than others. This method has worked in the majority of cases because in 128 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 closely related forms the dentitions, like other morphological features, usually are closely similar. There is a considerable residue of forms with spe- cialized dentition, however, whose relationships cannot be resolved by any amount of peering at the teeth as discrete entities. As Bateson re- marked more than half a century ago, "the attri- bution of strict individuality to each member of a repeated series of repeated parts leads to ab- surdity." No better example of the limitations of this method could be asked than the giant panda. On the basis of a mechanical point-by-point comparison, the teeth of Ailuropoda are in some respects more similar to the teeth of the Procy- onidae than to those of the Ursidae. The whole premolar series is strikingly degenerate in the bears. Pl-3 are peg-like vestiges, often missing in part. Even P* (the carnassial), normally the larg- est tooth in the upper battery in carnivores, is greatly reduced in all bears. In Ailuropoda, in striking contrast, only PI is degenerate and the remaining premolars are robust functional teeth. P^ is large, with five well-formed cusps occupying the same relative positions on the crown as they do in Procyon and Ailurus (though they are dif- ferently related to the roots). On the other hand, the molars of Ailuropoda are far more bear-like than procyonid-like, especially in the presence of a large talon on M- and in the retention of M.,, which is lacking in the Procyonidae. It was the opinion of Lydekker, Gregory, and McGrew, how- ever, that the premolar features are "more im- portant" than the molar. Two recent concepts have greatly changed our ideas of the evolution of dentitions. Structures such as teeth or vertebrae are serially repeated (homeotic) elements. It has been found that such structures are at least partly controlled by genes exerting a generalized effect over a region com- prising several adjacent elements, rather than on each isolated element. This is the field control concept. Sawin (1945, 1946) and Sawin and Hull (1946) have so interpreted hereditary variations in vertebral formula in rabbits. Butler (1939, 1946) has applied the field concept to the teeth of mam- mals, arguing that they are homeotic structures that have evolved as parts of a continuous mor- phogenetic field rather than as isolated units, and that a common morphogenetic cause must have acted on more than one tooth germ to account for the close similarity between adjacent teeth. The second concept is that of differential growth, which was developed chiefly by Huxley (1932). Accord- ing to this theory, now voluminously documented, various structures may have a different growth rate from that of the organism as a whole. Thus, with increase in the size of the organism during phylogeny, structures may attain a relative size or degree of differentiation that is not directly determined by the action of selection on the struc- ture itself. The classic examples of the mandibles of lucanid beetles and the antlers of deer are well known, but it is not so well understood that this principle may apply also to the teeth of mammals. How do these concepts relate to the dentition of Ailuropoda'! In the primitive carnivore den- tition, as represented hy Canis, the dental gradient of the upper cheek teeth centers in P' and M', falling off steeply on either side of this center. More specialized carnivore dentitions exhibit a shifting of this center anteriorly or posteriorly along the tooth row, and expansion or contraction of the center to embrace one or several teeth (fig. 69). The Ursidae differ from other Carni- vora in that the center lies wholly in the molar region, falling off abruptly at the boundary be- tween molars and premolars. The molar empha- sis is further reflected in the conspicuous posterior extension of M- in the form of a large talon. In Ailuropoda the whole premolar- molar battery has been secondarily enlarged, but there is still the same molar emphasis as in the bears. The dental gradient is quite distinctive and different from that of the Procyonidae. Enlargement in Ailuropoda begins abruptly at the boundary between the first and second pre- molars; the teeth anterior to this line (first premo- lar, canine, and incisors) are no larger than in Ursus, whereas teeth posterior to the line are all enlarged to approximately the same degree. These correspond almost exactly to the canine and in- cisor fields and the molarization field, respectively, of Butler. An astonishingly close parallel to this condition is seen in the fossil anthropoid Paran- thropus robustus (Broom and Robinson, 1949), in which the premolar-molar series is so much and so abruptly larger than the canine-incisor series that it is difficult to believe they belong to the same individual. The data of Rode (1935) on the dentition of fossil and recent bears present a clear picture of changes directly correlated with skull size in the genus Ursus. Such changes are the result of dif- ferential growth rather than of direct selection on the dentition, and are only secondarily (if at all) related to the functioning of the teeth. The pre- molar dentiton is reduced in all members of the genus, no doubt as a result of selection, but de- terioration becomes progressively more pronounced with increased skull size. In small forms ( Ursus americanus) the formula is typically Pi; among the medium-sized species it is f in U. arctos and DAVIS: THE GIANT PANDA 129 Canis Felis Mustela Procyon Ailurus Ursiis Ailuropoda Fig. 69. Upper cheek teeth of representative carnivores to show varying gradients in the premolar-molar field. f in U. horribilis, but in the huge U. spelaeus it is { or even ^.' Thus there is an inverse correla- tion between skull size and premolar development in Ursus, and reduction of the premolars is a fea- ture of the growth pattern of this genus, its ex- pression becoming increasingly pronounced with increased skull size. It is probable, furthermore, that the growth pattern was established early in bear phylogeny, in animals of relatively small size, in adapting the primitive carnivore dentition to the requirements of the bear stock. The almost total suppression of the premolars in large species would then be merely an expression of the gi'owth pattern of the bear stock, a direct result of selec- tion for larger size, not of selection on the dentition itself. If an individual American black bear grew ' The Alaskan brown bear (Ursus gyas), with a basal skull length up to 405 mm., may rival U. spelaeus in size. The cheek teeth of gyas are the same absolute size as in the grizzly, however, showing that a new and different factor (probably resulting from direct selection on the dentition) has affected the teeth in gyas. The premolar formula is typically f . to the size of a cave bear, we should expect its pi'emolars to resemble those of a cave bear. With respect to the molars, Rode's data show a direct correlation between tooth size and elabora- tion of the crown sculpture in the form of second- ary wrinkles and tubercles. The cingula also become wider and better defined with increased tooth size. Both reach a peak in Ursus spelaeus. Thus, elaboration of the molar crown pattern is directly correlated with tooth size, and is an ex- pression of the growth pattern of the bear stock. The condition seen in U. spelaeus results from the absolutely larger teeth, not from selection on the teeth themselves. The consequences of differential growth thus re- veal two significant features of the dentition of bears. These probably could not have been de- tected, and certainly could not have been verified, at the stage when they were under the active in- fluence of natural selection. The later effects seen on larger individuals, by exhibiting the results of the pattern in exaggerated form, leave little doubt. 130 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 The two ursid features are: (1) almost total shift of emphasis to the molar region of the cheek-tooth field, with the great talon suggesting a tendency to carry the center as far posteriorly as possible, and (2) elaboration of the crown sculpturing of the molars. The basal skull length of Ailuropoda is slightly less than that of the American black bear. The molar teeth are disproportionately large, their ab- solute length agreeing with the much larger gizzly. But the whole tooth row of Ailuropoda is also enormously broadened, and the molars equal (M-) or exceed (M') those of the cave bear in width. The broadening affects the premolars as well as the molars (i.e., it extends over the whole cheek- tooth field), and the disparity between premolar and molar size is not as obvious as in the bears. Nevertheless the molar dominance is still evident in Ailuropoda. Broadening of the premolars in Ailuropoda is associated with the development of an internal row of cusps, as it is in Procyon and Ailurus, the other two arctoids in which the cheek teeth are broadened. These cusps are, of course, conspicu- ously wanting in the reduced premolars of Ursus. Now, their presence in Ailuropoda may (1) indi- cate affinities with the procyonids, or (2) be an expression of the common genetic background of the arctoid carnivores. As will appear in the se- quel, there can be no doubt of the ursid affinities of the giant panda, and therefore the second of these alternatives is correct. The internal row of cusps in Ailuropoda is the result of broadening of the premolars. The surface sculpturing of the molars is much more elaborate in Ailuropoda than in the small and medium-sized bears, but it is almost identical with that of the gigantic Ursus spelaeus. If, as pointed out above, elaboration of sculpturing is a function of absolute tooth size in the bear stock, then this is exactly what we should expect in the huge molars of Ailuropoda. Any relation between the "bunodont" character of the molars of Ailu- ropoda and its diet is fortuitous. It is only the enlargement and broadening of the teeth that are so correlated. Thus, given the morphogenetic pattern of the bear stock, only two (perhaps only one) important new factors have appeared in Ailuropoda. In the ursid stock the morphogenetic field is concentrated in the molar region, with the premolar field essen- tially vestigial. In Ailuropoda the ursid pattern has been further altered by two' simple morpho- genetic factors: (1) secondary enlargement of the whole cheek-tooth field, and (2) secondary broad- ening of the whole field. Note that both of these factors affect the cheek-tooth field as a whole (ex- cept PI, which is vestigial), with no indication of a gradient that did not already exist in the bears. It is these two factors that represent adaptation in the dentition of Ailuropoda, and not the de- tailed architecture of each individual tooth. III. CONCLUSIONS 1. In Ursus the expression of the dentition is a function of skull and or tooth size. Elements in the premolar field degenerate progi'essively with increasing skull size among the species of Ursus, whereas the molar crown pattern becomes increas- ingly elaborate as absolute tooth size increases. 2. The dentition of Ailuropoda is an ursid den- tition in which all elements in the premolar-molar field (except PI) have become uniformly much en- larged and broadened. The result is a disharmo- nious relation between skull and dentition, which is reflected in the displacement of P2. 3. The molar crown pattern of Ailuropoda closely resembles that of the gigantic Ursus spe- laeus. The molar crown pattern of Ailuropoda is therefore a function of tooth size, not of selection for a particular pattern. 4. Successive disappearance of premolars, which accompanied increased skull and or tooth size in Ursus, is not evident in Ailuropoda, although the ursid proportions between premolar and molar size are retained. 5. In Ailuropoda, selection was for increased cheek-tooth size. Selection operated on the mor- phogenetic field of the cheek-tooth battery as a whole rather than on individual units. The result is that all units in this field are enlarged to the same relative degree. 6. The genetic mechanism behind this trans- formation is probably very simple and may in- volve a single factor. ' Increased tooth length may, of course, be merely a sec- ondary result of broadening, in which case only a single new factor would be involved. ARTICULATIONS Descriptions of the joints of mammals other than man are very few, and are incomplete for even the common domestic mammals. This is unfortunate, since no mechanism— the mastica- tory apparatus or the hand, for example— can be understood unless the structure and functioning of the joints are known. Comparative arthrology, the comparative anatomy of the joints, cannot be said to exist as an organized body of knowledge. No attempt is made here to describe all the joints of the giant panda. Those forming parts of mechanisms that are much modified in Ailu- ropoda — the mandibular, wrist, and ankle joints — were studied in detail and compared with the corresponding joints in the bears. A few other joints, chiefly those important in locomotion, are also described. I. ARTICULATIONS OF THE HEAD Mandibular Joint The mandibular joint is a sliding hinge joint, as in all carnivores. The two joint surfaces are very closely congruent, as they are also in Ursus; they are less so in some other carnivores. The joint in Ailuropoda is not quite transverse, its axis in the frontal plane forming an angle of 5 10° with the transverse axis of the skull. This compares with a range of 5-20° in a series of Ursus. In the trans- verse plane the axis is depressed toward the mid- line at an angle of about 10°, compared with about 2° in Ursus. These deviations from the transverse axis represent sectors of two circles, one in the transverse and one in the frontal plane, whose common center lies some distance in front of the canines. They reflect the fact that the canines interlock as they come into occlusion, checking transverse movement at this point and causing the canines to act as a point of rotation. Since the canines are conical rather than cylindrical, the actual point of rotation lies in front of the canines. The mandibular joint is remarkable for its mas- siveness, its relative size exceeding that of any other carnivore. It is also displaced dorsally and posteriorly relative to its position in Ursus. This displacement increases the mechanical efficiency of the jaw apparatus for crushing and grinding (p. 69). The articular capsule is a close-fitting sac, heavy posteriorly but much thinner anteriorly, where it is intimately associated with fibers of the temporal muscle. The capsule is attached to the margin of the mandibular fossa all around, and to the mar- gin of the articular surface on the capitulum of the mandible. There is no thickening at the lateral end corresponding to the temporomandibular liga- ment of human anatomy. The articular disk is almost paper-thin and is imperforate. It increases slightly in thickness from anterior to posterior and is not notably thinner at the center than at the periphery. The disk is firmly attached to the capsule throughout its en- tire periphery, and is more tightly attached to the fossa than to the condyle. None of the external pterygoid fibers insert into it. A single ligament is associated with the man- dibular joint (fig. 70). This apparently represents the stylomandibular ligament combined with the posterior end of the sphenomandibular ligament; in Ursus these two are separate and distinct and attach at the normal sites. The ligament is a band about 5 mm. wide, attached anteriorly to the an- gular process of the mandible. From this attach- ment it runs posteriorly, dorsally, and medially beneath the outer end of the postglenoid process. Here it bifurcates, one branch going to the inferior surface of the bony auditory meatus (the normal attachment of the sphenomandibular ligament) and the other to the inferior surface of the carti- laginous auditory meatus (the normal attachment of the stylomandibular ligament). The absence of the anterior part of the sphenomandibular liga- ment, which normally attaches at the entrance to the mandibular foramen, is probably associated with the great size of the postglenoid process. Movement in the mandibular joint consists, as in all carnivores, of two components: hinge move- ment around an approximately transverse axis, in which the cylindrical head rotates in the trough- like fossa; and sliding movement, in which the head shifts transversely in the fossa. These two movements are combined into a spiral screw move- 131 132 FIELDIAXA: ZOOLOGY MEMOIRS, VOLUME 3 Proe. augularit mandibulat Capsula articularif Prof, mastoideuf Meatus acusticus ext. CorpMS monuHbulat M. pter\-goid. int. Capsula articularU Ljg. stytomandibulare sphenomandibulare Hamulus pterfgoidens Proe. posi- glenoid. Fig. 70. Right mandibular joint of Ailuropoda, external view. ment, as is evident from the wear surfaces on the teeth. Measured on the dr>- skull, the lateral com- ponent amounts to about 6 mm. after the cheek teeth first begin to come into occlusion. The cor- resix)nding lateral component is about 3 mm. in a specimen of Ursiis aretos. In summary-, the mandibixlar joint of Ailuro- poda differs from that of I'rsus chiefly in being larger and more robust, and in being displaced dorsally and posteriorly. These are all directly adaptive modifications. They can scarcely be attributed to extrinsic factors, but probably rep- resent the results of selection operating on intrinsic (hereditan.-) factors. It is even plausible that the increase in quantity of bone tissue in the skull as a whole reflects the generalized working of the morphogenetic machiner>- whereby increased size of the bony elements of the mandibular joint was effected. II. .\RTICULATIOXS OF THE FORE LEG A. Shoulder Joint The shoulder joint is an extremely simple joint, as it is in all mammals that lack a clavicle. The only ligament is the loose articular capsule, and consequently the joint is held in position solely by muscles. .\s pointed out by Baum and Zietzsch- mann for the dog, the powerful tendon of the in- fraspinatus laterally, and that of the subscapularis medially, functionally represent collateral liga- ments of the shoulder joint. In addition to their function of retaining the joint in position, these tendons must also tend to limit adduction and abduction of the humerus, and thus to i-estrict movement to a pendulum-like flexion and extension. The glenoid cavity of the scapula is remarkable for its narrowness in comparison with other carni- vores. The articular surface of the head of the DAVIS: THE GIANT PANDA 133 humerus, in contrast, is broader than in other car- nivores. The fibrous glenoid lip is inconspicuous except along the posterior border of the glenoid cavity, where it projects a couple of millimeters beyond the edge of the bone. The articular capsule is a loose sac enclosing the shoulder joint on all sides. It extends from the prominent rough surface around the margin of the glenoid cavity of the scapula, to the head of the humerus. On the humerus the capsule is attached to the roughened area at the periphery of the head. In the intertubercular area it is pro- longed distad into the intertubercular sheath that encloses the tendon of the biceps. The posterior (superficial) fibers of the triceps medialis separate from the anterior (deep) fibers at their origin, and arise from the inferior surface of the capsule instead of from bone. Contraction of this muscle would consequently exert traction on the capsule. A very few of the posteriormost tendon fibers of the triceps lateralis are also at- tached to the joint capsule. B. Elbow Joint The elbow joint (figs. 71, 72) depends for its strength and security on bony structures rather than on the number, strength, or arrangement of its ligaments, as is the case with the knee. In the giant panda and bears the elbow joint is a screw joint rather than a simple hinge joint as in other carnivores. The spiral trough formed by the me- dial half of the trochlea (fig. 49) forces the ulna to travel medially 5 mm. or more as the elbow is flexed. With the foot in the normal position of pronation, this would throw the foot medially as the elbow is flexed, and would account, at least in part, for the rolling motion characteristic of the fore feet in bears and the giant panda. The capsule is a large and capacious sac to which the collateral ligaments are inseparably united. The supinator and a small part of the abductor pollicis longus muscles arise directly from the capsule. The bony attachments of the capsule are as follows: (1) on the humerus it encloses the vestigial coronoid fossa anteriorly and the ole- cranal fossa posteriorly; laterally and medially it attaches to the sides of the trochlea and the distal ends of the epicondyles; (2) on the ulna it attaches to the edges of the semilunar notches; (3) on the radius it attaches just distad of the articular facet. The lateral collateral ligament arises from the lateral epicondyle and runs distad across the radiohumeral articulation. At the annular liga- ment it is interrupted by the origin of the supi- nator muscle, beyond which it continues distad to its attachment on the anterolateral surface of the radius about 30 mm. below the head. A prom- inent scar marks its radial attachment. There are two lateral ligaments in the dog (Baum and Zietzschmann) and cat (Reighard and Jennings), one going to the ulna and the other to the radius. The medial collateral ligament is stronger and better marked than the lateral ligament. On the humerus it is attached to the area in front of the medial epicondyle. The nearly parallel fibers pass across the joint and attach on the ulna in the conspicuously roughened area immediately distad of the semilunar notch. In both the dog and the cat the medial ligament is double, consisting of radial and ulnar heads. The oblique ligament is a slender band run- ning diagonally across the anterior (flexor) surface of the lateral epicondyle. Distally it attaches to the distal lip of the semilunar notch. In the dog the oblique ligament divides distally to embrace the tendons of the biceps and brachialis (Baum and Zietzschmann). Parsons (1900) says it is absent in Ursus, and Reighard and Jennings do not men- tion it in the cat. C. Union of the Radius with the Ulna The radius and ulna are united at three places: a proximal and a distal radioulnar articulation, and a mid-radioulnar union via the interosseous ligament. The proximal articulation is composed of the radial notch of the ulna and the smooth circum- ference of the head of the radius that rotates in it. Two ligaments are special to the joint. The lat- eral transverse ligament (fig. 71) is a short diag- onal band extending from the annular ligament just below the lateral collateral ligament to the border of the semilunar notch immediately behind the radial notch. This ligament is absent in the dog (EUenberger and Baum, 1943) but is present in the bears. The annular ligament of the ra- dius is a well-defined band of strong fibers about 15 mm. wide, encircling the head of the radius. It forms about 60 per cent of a ring, which is com- pleted by the radial notch of the ulna. The an- nular ligament is thickest over the notch in the head of the radius. It is strongly attached at either end to the margins of the radial notch, and is much more feebly attached by loose fibers to the neck of the radius below the epiphyseal line. Since the head of the radius is elliptical in out- line, it acts as a cam and imparts an eccentric motion to the radius during movements of prona- tion and supination. The cam action can easily be felt through the annular ligament when the radius is rotated on a ligamentary preparation. This eccentric motion has the effect of permitting lumerus Lag. transversum laterale Capsula articularis Fig. 71. Right elbow joint of Ailuropoda, bent at right angle, lateral \-iew. Foreann halfway between pronation and supination. \Ulna Fig. 72. Right elbow joint of Ailuropoda, bent at right angle, medial view. Forearm halfway between pronation and supination. 134 J DAVIS: THE GIANT PANDA 135 a certain amount of rotation of the radius witiiout stretching the interosseous ligament. The range of movement in the proximal radio- ulnar articulation appears to be severely limited in Ailuropoda. The pronation-supination range was about 40° (compared with 120-140° in man) on a ligamentary preparation when the radius was ments lying just distad of the radioulnar articula- tion. The dorsal radioulnar ligament (fig. 73) is a rope-like band attached at one end to a pit- like depression on the neck of the styloid process of the ulna, between the radioulnar articulation and the head. The other end attaches to the ra- dius immediately below and in front of the radio- Lig. radioulnaris dors. Bursa m. ext. y-<^ -^ carpi ulnaris ^^ ~ ^ ^t^ Comp;irtment for Radius ^facics artic. carpeae) M. ext. dig. com. Capsula articularis Proc. styloideus ulnae Lig. radiocarpi volari Septum artic. (cut) Fig. 73. Proximal articular surfaces of right antebrachiocarpal joint of Ailuropoda. rotated by grasping its distal end and manipulat- ing it by hand. Further rotatory movement was checked by the capsule of the proximal radioulnar articulation, by the interosseous ligament, and by the distal radioulnar ligaments. The interosseous ligament (figs. 71, 72) is a thick tract of glistening fibers extending between the ulna and the radius except for the proximal quarter of the interosseous space. The ligament is heaviest in the middle third of the interosseous space, becoming thin and almost membranous in the distal third. Most of the fibers run diagonally distally from the radius to the ulna, but on the anterior surface a large group of proximal fibers runs in the opposite direction. The interosseous ligament is so heavy that it binds the ulna and radius firmly together, permitting very little move- ment between them. Nothing comparable to the oblique chord of human anatomy is present in Ailuropoda. The distal radioulnar articulation (fig. 73) op- poses a flat, almost circular surface on the radius to a slightly convex, almost circular surface of the ulna. The surface on the radius is parallel to the midline of the radius (which curves toward the ulna in its distal quarter), whereas the surface on the ulna lies at an angle of about 45° to the long axis of the ulna. The articulation is enclosed in a capsule. This articulation, which closely resem- bles that of Ursus, permits the distal end of the radius to roll around the ulna in a limited arc. In Ailuropoda the distal ends of ulna and radius are held together by two strong transverse liga- ulnar articulation. The volar radioulnar liga- ment attaches at one end to the neck of the styloid process and at the other to the border of the distal articular surface of the radius, near the radioulnar articulation. It lies mostly deep to the volar radio- cai-pal ligament. D. Hand and Intercarpal Joints The range of movement of the hand as a whole is very great in primitive carnivores. All the pos- sible angular movements— rotation, flexion and extension, and abduction and adduction, together with combinations of these — can be carried out. One of the most important and extensive of these movements, rotation (inversion and eversion), is scarcely a function of the hand joint, but results almost entirely from movements of pronation and supination of the forearm and rotation in the shoulder joint. The essential hand joint for movement of the hand as a whole is the antebrachiocarpal joint (the radiocarpal joint of human anatomy). In all the other joints movement is extremely restricted, consisting only of a slight gliding of one bone upon another, which serves to give elasticity to the carpus. In a ligamentary preparation of Ailuro- ropoda, movement in the intercarpal and carpo- metacarpal joints is almost non-existent, whereas in a similar preparation of the bear Tremardos there is considerable movement in these joints, particularly in the direction of adduction and ex- tension. Ailuropoda RADIAL Add. ULNAR Tremarctos Anterior View (adduction - abduction) Abd. Ext. Flex. -12° Ext. Lateral View (flexion - extension) Flex. Fig. 74. Diagrams showing ranges of movement in the left antebrachiocarpal joint in ligamentary preparations of a giant panda and a spectacled bear. See text. 136 DAVIS: THE GIANT PANDA 137 Antebrachiocarpal Joint A double joint, consisting of the radius-scapho- lunar articulation medially, and the ulna cunei- form and pisiform articulation laterally. The joint cavity is partly divided into radial and ulnar com- partments by an incomplete septum of fibro- cartilage (fig. 73). This septum, the "triangular fibro-cartilage" of Parsons, is attached proximally to the radial side of the neck of the styloid process of the ulna; distally it passes into the notch be- tween the scapholunar and cuneiform and attaches to the scapholunar. Along its volar edge the sep- tum is continuous with the joint capsule, thus closing off the radial and ulnar compartments, but dorsally it stops abruptly at the level of the dorsal radioulnar ligament, leaving the radial and ulnar compartments in communication with each other. The distal articular surface on the radius is broader anteroposteriorly than in Ursus, and lacks the conspicuous saddle over the styloid process. The opposing articular surface on the scapholunar is smoothly ovate, lacking the depression into which the saddle fits in Ursus, and is about a third more extensive than the radial articular sur- face. Thus this part of the joint is an almost per- fect ellipsoid articulation, capable of extensive movements of flexion, extension, abduction, and adduction. Of these, only abduction is seriously restricted by the styloid process of the ulna and the ulnar collateral ligament, which also inhibits rotation almost completely. Range of the other movements is greatly facilitated by the disposition of the antebrachiocarpal ligaments. The ulnar-carpal part of the antebrachiocarpal joint is notable for the extent and flatness of the articular surface on the cuneiform-pisiform com- plex. Instead of forming a socket into which the styloid process of the ulna fits, as in Ursus, in Ailuropoda there is an extensive articular area over which the styloid process can wander. This articular area faces laterally, and therefore cannot transmit thrust from the carpus to the fore arm as it does in Ursus. Thus this part of the ante- brachiocarpal joint in Ailuropoda has the function of steadying the radio-scapholunar part of the joint. The following measurements of ranges of move- ment in the antebrachiocarpal joint were made on an embalmed adult panda and an adult spectacled bear. All muscles and tendons crossing the carpus were removed, but all ligaments were left intact. The fore leg was immobilized and the manus ma- nipulated from the distal end, the operator taking care not to force the manus beyond its normal limits or to induce movements in intercarpal or carpometacarpal joints. Angulation was read off directly on a protractor, two or mo;e readings be- ing made for each position. The long axis of meta- carpal 3 was used as the axis of the manus (see fig. 74). Ailuropoda Tremarctos Abduction— adduction 29° 22° Abduction (from radial axis=0). . . 4° 9° Adduction (from radial axis=0). . . 25° 13° Flexion-extension 78° 55° Flexion (from radial axis=0) 59° 67° Extension (from radial axis=0) .... 19° — 12° These figures indicate that the position of the manus in relation to the fore arm in the panda is quite different, in both planes, from its position in the bear. The axis of the radius is not the true axis of the fore arm, but it is close enough to show that in the "rest" position the hand of Ailuropoda is adducted whereas that of Tremarctos is abducted, and that the metacarpus is more strongly flexed in Ailuropoda than in Tremarctos. The figures also indicate that the range of movement in the ante- brachiocarpal joint is greater in the panda than in the bear, particularly movements of extension. The figures confirm the statement of Lips that the bears are incapable of extending the metacarpus beyond the long axis of the fore arm. Ligaments of the Carpus The carpal ligaments have not been described for any generalized carnivore. In the present study the ligaments of an adult spectacled bear {Tremarctos ornatus) were dissected, for compari- son, at the same time as those of Ailuropoda. The only significant differences were the presence in Tremarctos of stout dorsal radiocarpal and radial collateral ligaments. The absence of these liga- ments in Ailuropoda contributes greatly to the mobility of the antebrachiocarpal articulation, par- ticularly to the range of dorsal flexion. Antebrachiocarpal Ligaments The volar radiocarpal ligament (figs. 73, 76) is a thick flat band of fibers with a predominantly transverse direction. It is attached medially to the radius above the styloid process, and laterally to the neck and base of the pisiform ; its deep sur- face presumably attaches to the scapholunar and cuneiform. The proximal border of this ligament is thick and sharply defined; distally it continues into the transverse carpal ligament. The dorsal radiocarpal and radial collateral liga- ments of human anatomy are absent in the panda. Instead there is a roomy, tough-walled articular capsule enclosing the radiocarpal articulation dor- sally and laterally (fig. 73). The capsule attaches to radius and scapholunar near the margins of their articular surfaces. Lig. carp«)sesam(i Tendo m. abd. poU. longus Lig. basal is Fig. 75. Dorsal carpal ligaments of Ailuropoda. Os pisijorme Lig. pisometacarpeum. Lig. pisocuneiform. lat. Tuberc. ossis ameift Tendo mm. inter ossei Tuberc. ossis magmon. Ligg- carpometacarp. vol. Lig. radiocarpeum volare Lig. carpi transvereum Tendo m. flex, carpi rod. Tuberc. ossis scapholunaris Tendo m. abd. polUcis lonffus lig. carposesamoideum volare Os sesamoid, rod. \ ^■^— Lip. carpo - sesamoideum transv. Ligg. basium interossea vol. Fig. 76. Volar carpal ligaments of Ailuropoda. 138 DAVIS: THE GIANT PANDA 139 The ulnar collateral ligament of the wrist (fig. 75) is a heavy band of fibers extending from the latero- dorsal surface of the styloid process of the ulna to the distal end of the pisiform, where it attaches to a prominent scar on the posterior surface of the bone. Intercarpal Ligaments The transverse carpal ligament (fig. 76) is an ex- tensive tract of transverse fibers, continuous proxi- mally with the volar radiocarpal ligament. The band is cupped to form a trough for the tendon of the deep digital flexors. Attachment medially is to the ventral process of the scapholunar, laterally to the base of the pisiform. Its deep surface pre- sumably attaches to the ventral processes of the magnum, unciform, and cuneiform. The pisohamate ligament is a short band on the lateral aspect of the carpus. It attaches to the pisiform near the margin of the articular surface of the cuneiform, and to the lateral surface of the ventral process of the cuneiform. A system of short dorsal intercarpal ligaments (fig. 75) ties the carpal bones together. These are all short bands passing across from one bone to its neighbor. Ligaments of the Pisiform Bone Two ligaments connect the pisiform with the cuneiform. A volar pisocuneiform ligament passes from the volar surface of the pisiform to the volar surface of the cuneiform, median to the tubercle. It is inseparable from the pisometacarpal ligament throughout most of its length. A short lateral pisocuneiform ligament passes from the lateral sur- face of the pisiform, directly beneath the articular surface, to the tubercle of the cuneiform (fig. 76). A strong pisometacarpal ligament (fig. 76) ex- tends from the volar surface of the pisiform to the base of the fifth metacarpal. Carpometacarpal Joints The distal surfaces of the distal row of carpals present a composite articular surface for the four lateral metacarpals. In Ursus and most other carnivores the otherwise smooth contour of this composite articulation is broken by a wedge-shaped projection of metacarpal 2 that thrusts back be- tween the trapezium and trapezoid. This wedge is absent in Ailuropoda, and the transverse con- tour of the composite joint is therefore uninter- rupted. Otherwise the joint is similar to that of Ursus. The proximal articular surfaces on the metacarpals are convex dorso-ventrally, with a very slight transverse concavity on metacarpals 2-4 that produces a modified saddle joint. The saddle joint is most pronounced on metacarpal 4, and is wanting on metacarpal 5. The first metacarpal articulates with the tra- pezium by a saddle joint. The transverse curva- ture of the saddle is shallow, as in the lateral metacarpals. It is relatively deeper in Ursus, in- dicating a greater range of adduction-abduction movement. Carpometacarpal Ligaments Volar carpometacarpal ligaments are associated with digits 3, 4, and 5 but are wanting on digits 1 and 2 (fig. 76). These are short stout bands arising from the deep surface of the tendinous plate by which the digital adductors take origin — thus eventually attaching to the magnum and unciform — and inserting asymmetrically into the metacarpals near their bases. The ligament to digit 5 attaches to the radial side of the bone, those to digits 3 and 4 to the ulnar side. A short dorsal carpometacarpal ligament extends between the base of each metacarpal and the dor- sal surface of the adjoining carpal bone (fig. 75). Carposesamoid Joint The articulation between the radial sesamoid and the scapholunar is a true diarthrosis, capable of quite extensive movements of abduction and adduction, but probably incapable of dorsal and volar flexion. On a ligamentary preparation this bone could be manipulated through a range of about 20° of abduction-adduction, but was practi- cally immobile in the direction of flexion-extension. The radial sesamoid in Ursus has no such diar- throdial articulation, but the bone occupies the same positions relative to the scapholunar. Ligaments of the Radial Sesamoid Four strong and well-marked ligaments are as- sociated with the radial sesamoid bone. A short volar carposesamoid ligament (fig. 76) passes from the volar surface of the tubercle of the scapholunar to the volar surface of the sesamoid bone. A broad lateral carposesamoid ligament (fig. 75) passes from the lateral surface of the scapholunar tuber- cle to the lateral surface of the sesamoid, where it attaches proximad of the insertion of the tendon of the adductor poUicis longus. A transverse carpo- sesamoid ligament (fig. 76) passes from the lateral (ulnar) surface of the sesamoid into the transverse carpal ligament. On the dorsal side a dorsal basal ligament (fig. 75) connects the base of the sesamoid with the adjacent base of the first metacarpal. In Tremarctos the ligaments of the radial sesa- moid are similar to, but smaller than, those in Ailuropoda. 140 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Lig. patellae Lig. menisci med. ant Lig. cruciatum ant.. Tuberculum intercondyloid. med.. Meniscus med Lig. cruciatum post.. Lig. menisci lat. ant. Capsula articularis Meniscus lat. Lig. menisci lat. post. Bursa Fig. 77. Joint structures on head of right tibia of Ailuropoda. Review of Hand Joint Parsons (1900) reviewed very briefly the major carpal ligaments of the Carnivora in relation to those of other mammals, and concluded that the wrist joint in carnivores is modified to permit a "moderate amount" of supination. Lips (1930) described in great detail the structure and function- ing of the hand joint in Ursus arctos in compari- son with other arctoid carnivores, unfortunately without considering the ligaments. Lips concluded that the hand joint of Ursus represents a "univer- sal" (we would say unspecialized) type among the arctoid carnivores, capable of many-sided move- ments. The hand joint of Ailuropoda is very similar to that of Ursus, but the panda has gone beyond the bear in the range of movement possible in the ante- brachiocarpal joint, particularly extension. This is accomplished by extending and reshaping artic- ular surfaces, and by eliminating or reducing liga- ments that would restrict dorsal flexion. Such minor remodeling reflects the action of the mus- cles that operate these joints (largely the carpal extensors and flexors), and demands little or no morphogenetic action on the bones and ligaments themselves. Even the diarthrodial joint of the radial sesamoid requires only the well-known ca- pacity of bone to produce true joints wherever movement occurs. in. ARTICULATIONS OF THE HIND LEG A. Knee Joint The knee joint (fig. 77) is an incongruent com- pound joint involving the femur, the patella, and the tibia. The incongruence between the roller-like condyles of the femur and the relatively flat supe- rior articular surface of the tibia is compensated by the menisci. The internal ligaments of the knee joint of the horse, cow, pig, and dog were described by Zimmerman (1933). The structure of this joint in bears and procyonids is unknown. [ The menisci are unequal in size; the medial meniscus is larger than the lateral and its struc- ture is typical. The lateral meniscus has a promi- nent ridge on the femoral side that separates a medial articular area from a lateral non-articular area. The non-articular part of the meniscus ter- minates posteriorly at the entrance to a large bursa, which is situated above and immediately behind the fibular articulation. Each meniscus is attached to the capsule throughout its entire circumference, and each is also held in place by its own system of ligaments. Each meniscus is tightly attached to the head of the tibia at one end and more loosely attached at the other, which gives to both a certain freedom of movement on the tibial head. The lateral meniscus is continued into a liga- ment at each end. The anterior ligament passes mesad beneath the anterior cruciate ligament, to attach to the medial wall of the anterior intercon- dyloid fossa; the posterior one runs mesad and dorsad, to attach to the intercondyloid fossa of the femur. The medial meniscus is continued into a ligament only at its anterior end; the posterior end is tightly attached to the medial lip of the pos- terior intercondyloid fossa. The anterior end of the medial meniscus has no direct attachment to DAVIS: THE GIANT PANDA 141 the tibial head; it is continued into a powerful ligament that passes across immediately in front of the anterior cruciate ligament to attach to the anterior intercondyloid fossa in the area in front of the lateral condyle, laterad of the attachment of the anterior cruciate ligament. There is no transverse tract uniting the menisci anteriorly, corresponding to the transverse ligament of human anatomy; Zimmermann states that this tract was demonstrable in 94 per cent of the dogs that he studied. The cruciate ligaments are strong and rope- like. The anterior cruciate ligament is attached to the medial half of the anterior intercondyloid fossa of the tibia, near the medial intercondy- loid tubercle. It runs upward, backward, and slightly laterad to the medial surface of the lateral condyle of the femur, where it attaches. The pos- terior cruciate ligament is considerably longer than the anterior. It attaches to the tibia on a promi- nence at the extreme posterior end of the posterior intercondyloid fossa. From here it passes upward and nearly straight forward, crossing the posterior horn of the medial meniscus, and attaches to the femur in the medial half of the intercondyloid fossa. The only significant difference in the internal ligaments of the knee joint between Ailuropoda and the mammals described by Zimmermann is the less tight fixation of the menisci, especially the medial meniscus, in the panda. The resulting greater freedom permits more extensive pronation and supination in Ailuropoda. B. Ankle Joint The essential joints for movements of the foot as a whole in the primitive carnivore are the upper ankle joint, the transverse tarsal joint, and the lower tarsal joint. Each of these joints is primarily involved in a particular movement. In the upper ankle joint, movement is a hinge movement in the sagittal plane (flexion and extension of the foot). In the transverse tarsal joint, movement is rota- tion around the longitudinal axis of the foot (in- version and eversion of the foot). In the lower ankle joint, movement is an oblique gliding be- tween the astragalus and calcaneus (largely ab- duction and adduction of the foot) . None of these joints acts entirely independently of the others, and only the upper tarsal joint is confined to a single fixed axis. The resulting combined move- ments are extremely subtle and complex. The small bones of the distal tarsal row are probably mechanically unimportant. They func- tion chiefly to break shocks and to increase the general flexibility of the foot. Upper Ankle Joint (talo-crural) A perfect hinge joint between tibia and fibula proximally and astragalus distally. Axis runs trans- versely through trochlea of astragalus. Movement is restricted to dorsiflexion and plantar flexion of the foot. Lower Ankle Joint (subtalar) An incongruent gliding joint between astragalus and calcaneus. No definite axis can be fixed; Fick called the movement in this joint in man a "com- promise" movement consisting of the summation of successive rotations around a great number of momentary axes. In the bears and giant panda the congruence is less than in man, and it seems impossible to determine even a "compromise" axis. In procyonids the congruence is close and the movement is a screw movement. Movement is in general oblique: abduction coupled with ever- sion and dorsal flexion of the foot, or adduction coupled with inversion and plantar flexion (Sivers, 1931). X-ray photographs (fig. 79) show that movement in this joint is relatively slight in Ailu- ropoda and Ursus. Transverse Tarsal Joint (Chopart's articulation) A combination of rotatory and sliding joints, between the head of the astragalus and the navic- ular (rotatory) and the calcaneus and cuboid (glid- ing). The axis of rotation runs longitudinally through the head and neck of the astragalus and the approximate center of the navicular; the cal- caneus glides over the cuboid in an arc. Move- ment, which involves compensatory adjustments between the astragalus and calcaneus, is inversion and eversion and /or abduction and adduction of the foot. Dorsiflexion and plantar flexion of the foot, which is the main movement of this joint in man, is very slight. X-ray photographs (fig. 79) show that in Ailuropoda and Ursus rotatory move- ments in this joint are extensive, though less ex- tensive than between the navicular and the distal tarsal row. Most students of the comparative anatomy of the tarsus in quadrupeds (Tornier, 1888; Sivers, 1931; Schaeffer, 1947) have emphasized the trans- verse tarsal and lower ankle joints, dismissing the upper ankle joint as a simple hinge. In the tarsus of the generalized carnivores the most conspicuous difference is the relation of the axis of the upper ankle joint to the remainder of the ankle and foot. This difference is not apparent unless the astrag- alus is examined in situ, with the foot lying flat on the ground (fig. 78, A). Then the position of the axis with relation to the surrounding structures shows that the relation of the foot to the lower leg differs significantly from species to species. Angles 142 FIELDIANA: ZOOLOGY :MEM0IRS, VOLUME 3 "' ^y ^ Y^ 1 \^r J^ y^ \j -'- C^ PoboS Ursus Ailuropoda Fig. 78. Dorsal (A) and anterior (B) views of right astragalus and calcaneus of Polos flavus (an arboreal forml and Ursus arctos and Ailuropoda (terrestrial forms), to show differences in the angulation of the axis of the upper tarsal joint. In the dorsal views the horizontal line is drawn at right angles to the long axis of the foot. The anterior views are drawn with the foot flat on the ground, the horizontal line representing the horizon. The diagram associated with each drawing does not show the normal position of the foot, but the indicated position of the foot if the tibia were oriented (A) with the trans- verse axis of the inferior articular surface of the tibia parallel to the transverse axis of the body, and (B) with the long axis of the tibia vertical. C, proximal articular surfaces of navicular and cuboid in the same positions as B. were measured with a protractor on dried liga- mentary preparations with the foot in normal un- strained position. In dorsal view the axis is nearly transverse to the long axis of the foot in Ursus and Ailuropoda; actually it is rotated slightly counter- clockwise ( — 6° to —8°), so that the foot would have a slight tendency to toe out. In Claenodon, a primitive Paleocene creodont, the axis is rotated counterclockwise about 22°. In Potos and other procyonids, on the contrary, the axis is rotated clockwise (22° in Potos, 15° in Procyon, 15° in Ailu- rus), so that the foot would tend to toe in. In anterior view (looking at the distal faces of astragalus and calcaneus (fig. 78, B), there are similar though less extreme differences. In Ursus . and Ailuropoda the axis is tilted clockwise 15-20°, which would tend to produce moderate inversion of the foot. This tilting is greater in procyonids (50° in Potos) and would tend to produce strong DAVIS: THE GIANT PANDA 143 Fig. 79. Tracings from X-ray photographs of the right foot of the panda Mei Lan, to show areas in which joint move- ment takes place. A, medial view, foot abducted and inverted (solid line), superimposed on tracing of foot adducted and everted (shaded) ; the tibia, fibula, and calcaneus were superimposed in tracing. In abduction-eversion the calcaneus is rotated mesad on its long axis (note decreased width across sustentacular process — trochlear process), in addition to sliding laterad and proximad. Note, however, that the major movements of eversion-inversion and abduction-adduction take place in the transverse tarsal joint and the more distal parts of the ankle. B, dorsal view, the foot adducted and inverted (solid line), superimposed on tracing of foot abducted and everted (shaded). The calcaneus has rotated mesad on its long axis (note position of sustentacular process and decreased width across sustentacular process — trochlear process), in addition to sliding laterad and proximad. Note that the major movements of eversion-inversion and abduction-adduction take place in the transverse tarsal joint and the more distal parts of the ankle. inversion of the foot. The angle is about 45° in Claenodon. Sivers pointed out that the lateral and medial facets on the astragalus and calcaneus are more convex (or concave) in Mustela and Gulo, and that the facets are inclined toward one another. It may be added that the articular surface of the astragalar head is very extensive, and only part of it contacts the concavity of the navicular at any one time. This is likewise true of Procyon and Potos. These conditions permit a considerable range of inversion-eversion movement, wherein the astragalus rotates in a screw movement on the calcaneus, which remains relatively stationary with respect to the cuboid (movement in the intertarsal joint), while the astragalar head rotates extensively in the concavity of the cuboid (movement in the medial half of the transverse tarsal joint) . Exten- sive inversion and eversion are obviously associ- ated with the arboreal habits of these animals. It is functional eversion that permits these animals to apply the sole to a flat surface, as in standing on the ground. In Ursus and Ailuropoda, on the contrary, the lateral and medial facets are flatter and are less inclined toward one another, and the area of the astragalar head exceeds the area of the concavity of the navicular only slightly. This signifies a less extensive range of movement (particularly of ever- sion and inversion) in the ankle. Moreover, as Sivers pointed out for Ursus, movement between the astragalus and calcaneus (the lower ankle joint) is largely horizontal — rotation around a vertical axis running through astragalus and calcaneus; this is affirmed by our x-ray photos (fig. 79). This would increase the stability of the ankle, and would favor abduction and adduction rather than inver- sion and eversion. It also explains the fact that in the bears and panda the combined diameter 144 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 across the lateral and medial facets on the astrag- alus exceeds the diameter of those on the calcaneus. The following measurements of ranges of move- ment were made on the fresh unskinned hind leg (except Ailuropoda, which was skinned). The Ursus americanus was about a quarter grown, the other two fully adult. The tibia was placed in a vise and the foot manipulated by hand by grasp- ing the base of the metatarsals, the operator taking care not to force the foot beyond its normal limits. Angulation was read off directly on a protractor, two or more readings being made for each position. Prccjon Abduction-adduction. . . . 32° Flexion-extension 130-135° Flexion (from right/ 0). -40--43'' Extension (fromrightZO) -|-90--|-92° Eversion-inversion 87-89° Eversion 58-62° Inversion 27-29° iluropoda Ursus 28-29° 38° 45° 67-69° -1-45° -32° -1-90° -f35° 48-50° 42-50" 0° 17° 48-50° 25° Differences in abduction-adduction are negligi- ble among these three animals. Otherwise the total range of movement is notably greater in Procyon than in the bear or panda, and this presumably reflects the arboi-eal habits of Procyon. In Ailuropoda and Ursus not only is the range of flexion-extension more restricted than in Pro- cyon, but the pattern is different both from Procyon and from each other (fig. 80). In Ursus flexion and extension, measured from a line at right angles to the tibial axis, are about equal. In Ailuropoda the whole range of flexion-extension lies completely outside the range in Ursus, and well below the 90° axis; i.e., the foot in Ailuropoda is permanently extended on the tibia. The situation is comparable, although less ex- treme, for eversion-inversion (fig. 80). In Procyon eversion exceeds inversion. The reverse is true of Ursus, which also has a much more restricted range of movement. In Ailuropoda the range of movement is similar to that of Ursus, but is all in the direction of inversion; the foot cannot be everted on the tibia at all. I can find no differ- ences in the transverse tarsal joint of these two forms that would account for the differences in eversion-inversion. The difference probably lies in the torsion angle of the tibia (p. 115) and in- clination of the upper ankle joint. No detailed dissection of the tarsal ligaments of Ailuropoda was made. In summary, the ankle joint in the bears is a relatively unspecialized structure, combining mod- erate flexibility with adequate support (Davis, 1958) ; it is neither as flexible as the ankle of ar- boreal forms, nor as stable as the ankle of cursorial forms. The ankle joint of Ailuropoda, so far as Eversion Fig. 80. Diagrams of ranges of movement in the ankle joint of carnivores. A, eversion-inversion. B, flexion-extension. ^ (See figures in adjoining table). ji DAVIS: THE GIANT PANDA 145 known, is very similar to that of the bears. Cei'- tainly the resemblance is closer than in the hand joint. IV. REVIEW OF JOINTS In the developing individual the primary gross model of a joint is determined by intrinsic (heredi- tary) factors, but the further shaping of the joint depends almost wholly on extrinsic (non-heredi- tary) mechanical factors (Muiray, 1936). The importance of mechanical factors in determining joint form is heavily underscored by the pseudar- throses (joint-like structures in places where nor- mally there should be no joint) that have been described in the literature. Failure of a fracture to heal may, even in the fully mature adult, lead to the formation of a "a structure so exactly mim- icking a normal joint that the first half of the word 'pseudarthrosis' does it less than justice" (Mur- ray, 1936). Such pseudarthroses may involve joint-like expansions of the apposed ends of the bones, cartilage-covered articular surfaces, a cap- sule, ligaments, and synovial fluid. If only the gross model of an articulation is in- herited, then natural selection can act directly only on the gross model. The articulation is, of course, a part of a total functional mechanism that is sub- jected to selection. The articulation's response to such extrinsic factors as posture and movement may therefore, by limiting the range of possible functional mechanisms, limit or channel the gene- controlled changes in other elements of the total mechanism and thus indirectly play an active role in natural selection. In seeking a causal explana- tion for the differences between two closely related organisms, however, we must assign a passive role to differences in the articulations. This will not be true if we are comparing distantly related or- ganisms (perhaps above the family level), where differences in the gross model, attributable to in- trinsic factors, are likely to be involved. Nor will it be true for grossly adaptive differences, such as those in the mandibular articulation of the panda, if these involve differences in the gross model. The chief value of the joints in comparisons between closely related forms is, then, as extremely sensi- tive indicators of differences in other elements that are related mechanically to the joints. Except for the mandibular joint, the joints of Ailuropoda, so far as they have been studied, differ little from those of Ursus. Such differences as there are tend to increase the range of movement in the joints. None of these differences seems to depend on intrinsic factors other than the capacity of the joint to respond to extrinsic factors. THE MUSCULAR SYSTEM The muscles of the Carnivora are comparatively well known, but even for this order our knowledge is at a primitive level. Descriptions are incom- plete and inaccurate, often doing little more than establish the fact that a given muscle is present in species dissected. Even for the domestic carni- vores— the dog and the cat — the standard reference works are full of inaccuracies and are inadequately illustrated. Most of the genera of Carnivora have never been dissected at all. Within an order as compact as the Carnivora there are few differences of the "present" versus "absent" variety (see Table 16, p. 197 1, and ques- tions of muscle homology' are of no importance. There has, however, been a good deal of adaptive radiation within the Carnivora, as is obvious if the agile predaceous cats are compared with the lumbering semi-herbivorous bears, or the cursorial cheetah with the burrowing badgers. Such dif- ferences in habit are reflected in differences in the muscular system. These muscular differences — their nature, their directions, their limitations — are important elements of the over-all problem of evolutionarj- mechanisms. They show what has happened (and what has not happened i to the muscle pattern inherited by the Carnivora from creodont ancestors. Such empirical data form the basis on which the nature of mammalian evolution at the sub-ordinal level must be judged. How can such differences be detected and eval- uated? Certainly not on the basis of existing descriptions and illustrations. DATA OF COMPARATIVE MYOLOGY Observation indicates that within a gi-oup of related organisms a muscle is responsive, within limits, to mechanical demands in (li relative size, and (2) position most favorable for the required lever action. Limits are set, on the one hand, by the heritage of the group; the cephalohumeral of the Carnivora, for example, has never reverted to the original deltoid and trapezial elements from which it arose, no matter how mechanically ad- vantageous such a course might be. On the other hand, the structures surrounding a muscle defi- nitely limit the range of adaptive change of a muscle. No alteration can continue to a point where it interferes with the vital activities of other structures. A remarkable instance of this type of limitation is seen in the temf>oral muscle of the giant panda (see p. 69). A few generalizations as to the mode of phylo- genetic alterations of muscles at the sub-ordinal level may be listed. These have been derived em- pirically from direct observation. 1. The bony attachments of a muscle may wan- der almost at random (within the limits of its area of embryonic origin i, provided they do not en- croach on some other vital structure. This is seen throughout the muscular system. It is particu- larly apparent, for example, in the origin of the triceps in carnivores (fig. 81). 2. Phylogenetic decrease in the volume of a muscle presents no problem, since surrounding structures simply move in and occupy the vacated space (e.g., loss of the short head of the biceps in carnivores ) . The power of a given muscle is usu- ally increased phylogenetically by increasing its area of cross section (i.e., increasing the number and or diameter of fibers i . In muscles with dif- fuse origin this involves increasing the area of origin, and this is accomplished in various ways: (a) The bone surface may be increased, as in the temporal fossa of the giant panda, or the postscapular fossa on the scapula of bears. (b) Flat muscles may be reflected, like folding a sheet of paper, to increase the total length of origin without increasing the over-all linear extent on the bone. This is seen in the deep pectoral of the bears and giant panda compared with those of more primi- tive carnivores. (c) Accessory origin may be gained from super- ficial aponeuroses or from a tendon sheet embedded in the muscle, as in the temporal muscle of carnivores. (d ) Surrounding muscles may be displaced from their bony attachment, and arise or insert instead on the fascia of other muscles. This is seen in the deltoids of the giant panda. 3. It has long been known that muscles may become more or less completely transformed into I 146 DAVIS: THE GIANT PANDA 147 Canis Felis Ailuropoda Fig. 81. Medial view of humerus of Cant's (after Bradley), Felis (after Reighard and Jennings), and Ailuropoda to show variation in the origin of the medial head of the triceps. tendons during phylogeny, and Haines (1932) has demonstrated that tendons increase at the expense of muscle substance during ontogeny in man. He suggests that "tendon is lengthened by metamor- phosis of muscle tissue in response to a limitation of the range of possible contraction determined by the nature of the attachment of the muscle." Confirmation of this thesis is seen in the zygo- maticomandibularis of the dog, where two layers cross at an angle and the deeper layer is devoid of muscle fibers exactly to the boundary of the more superficial layer that partly overlies it. A similar situation exists in the trapezius muscles of the giant panda; muscle fibers are wanting exactly as far as the border of the scapula (fig. 88). In both of these examples pressure has limited the range of contraction of part of a muscle, and in the areas subjected to pressure, muscle tissue is replaced by tendon. Haines' further suggestion, that "it is no longer necessary to postulate complex co-ordinating mech- anisms to govern the sizes of the muscles, nor a vast series of genes to suit muscles to their work," is an over-simplification. In cursorial mammals, for example, the limb muscles are concentrated near the center of limb rotation, resulting in long terminal tendons. This is for the obvious mechan- ical reason that such an arrangement reduces the moment of inertia of the limb, not because of any limitation of the range of possible contraction. The tendons are already greatly lengthened in a fetal horse. Degree of tendinization may be (1) an active mechanical adaptation, or (2) a reflection of limi- tation of range of contraction resulting from (a) pressure from surrounding tissues or (b) simple degeneration, as in the short head of the biceps. Tendinization of type (2) is probably an individ- ual response to local conditions, not dependent upon gene action. 4. The relation between muscle attachment and bone relief at the site of attachment was reviewed by Weidenreich (1922, 1926) and Dolgo-Saburoff (1929, 1935). It is well known that the surface relief of bone is attributable almost entirely to the muscles and their adnexa, and the ligaments. The nature of this relationship is not well understood. Weidenreich emphasized that ridges and tuberosi- ties represent portions of tendons or ligaments that have ossified under tension and are then in- corporated into the underlying bone. The extent of this ossification tends to be directly proportional to the mass of the musculature, and thus to the force to which the connective tissue is subjected. 148 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Where a muscle mass is enlarged beyond the available attachment surface on the bone, attach- ment is extended onto the adjacent fascia; conse- quently the size of a muscle cannot always be judged from its mark on the bone (Weidenreich, 1922). Beautiful examples of this phenomenon are seen in the limb musculature of Ailuropoda. Transgression onto the fascia may lead to ossifi- cation of the fascia and its incorporation into the skeleton, as is easily seen in the development of the sagittal crest in many mammals. DATA OF MORPHOGENESIS We know almost nothing of the genetic basis for the differentiation and development of muscles, of the relative roles of intrinsic (genetic) vs. ex- trinsic (non-genetic ) factors, or of the parts played by generalized and localized gene effects. The ex- tensive catalog of genes in the laboratoiy mouse compiled by Griineberg (1952) does not contain a single reference to the muscular system. This al- most total ignorance contrasts sharply with the considerable body of such knowledge for the skele- ton and joints, and makes it almost impossible to postulate the nature of the machinery involved in producing adaptive differences in the muscular system. The differentiation and growth of muscle in the indi\-idual were reviewed by Scott (1957). There is an intimate relation between differentiation and gi-owth of a skeletal muscle and the neive supply- ing it, and the nei-ve seems to be the determining agent in this relationship. Initial differentiation of muscle fibers and their gi-ouping into individual muscles can take place in the absence of any ner\-e connection; that is, muscles have a certain capac- ity for self-differentiation. But without nen^e- muscle connections the muscle fibers do not de- velop beyond a certain stage and later undergo degeneration. Yet Pogogeff and Mun-ay (1946) and others have maintained adult mammalian skeletal muscle in vitro for months, without inner- vation of any kind, and during this time the tissue regenerated and multiplied. The developing mus- cles in the individual are at first independent of the skeletal elements, to which they gain attach- ment only later; a muscle develops normally even in the absence of the skeletal elements to which it normally gains attachment. Independence of the musculature from a factor affecting the skeleton was demonstrated in achondroplastic rabbits by Crary and Sawin (1952), who found the muscles of normal size whereas the bones with which they are associated were shorter. The muscles had to "readjust their bulk and area of attachment to the new bone shapes." During early ontogeny, skele- tal muscles grow by di\'ision of developing fibers or by differentiation of additional muscle-forming cells, but during later ontogeny, gi-owth is believed to be exclusivelj' by hj'pertrophy of individual fibers. Growth of muscles in bulk, even in the adult, seems to be controlled at least in part by the nerv- ous system. In man, disease of peripheral nerves (such as pohTieuritis) may be followed by abnor- mal nei-\-e regeneration and associated h\-pertrophy of the related muscles, and hypertrophy of the masseters is often associated with evidence of dis- order of the central nei-\'Ous system (Scott, 1957). Such gi-owth is by hypertrophy of individual mus- cle fibers. Muscular hypertrophy as a hereditan,- condition has appeared in various breeds of domestic cattle (Kidwell et al., 1952). In this condition the mus- cles are enlarged, and most authors (but not Kid- well et al. ) describe duplication of muscles. The effect is typically localized in the hind quarters and loin (Kidwell et al. state that in their stock the muscles of the withers and brisket were also somewhat hypertrophied ) . All authors describe the muscles as coarse-gi-ained, and mention a gen- eral reduction in the quantity of fat, both sub- cutaneous and intra-abdominal. Kidwell et al. concluded from breeding exjjeriments that the con- dition "appears to be inherited as an incomplete recessive with variable expressivity." In other words, a simple genetic mechanism capable of pro- ducing a generalized effect on the musculature has been demonstrated. The data of Fuld (1901) reveal differences from his control animals in the relative mass of certain muscles of the hind limb in dogs that were bipedal from puppyhood. Most of the limb muscles were unaffected, but foiu* showed differences of more than 5 per cent in their mass relative to the total mass of hip and thigh muscles. These were the gluteus medius (7.6 per cent heavier), quadriceps extensor (6.4 per cent lighter), biceps femoris (8.2 per cent lighter), and adductors (9.4 per cent heavier). Two of these differences (middle glu- teal and biceps) are in the direction of the weight relations found in man, whereas the other two are in the opposite direction. The dogs were said to hop rather than to walk on their hind legs, how- ever, and the differences from the control animals may well have been adaptive, or at least reflected differences in the demands made on the muscles. Under any circumstances they certainly were not hereditary. These scanty data provide few significant clues to the nature of the morphogenetic machinery in- volved in the evolution of adaptive differences in the musculature. DAVIS: THE GIANT PANDA 149 ABSOLUTE VS. RELATIVE MUSCLE MECHANICS Attempts to study muscle mechanics have dealt almost wholly with absolute values — absolute con- tractile force per unit of muscle cross section, lever actions of individual muscles or groups of muscles, or direct measurements of the power of an organ, such as a limb. This approach has yielded indif- ferent results because of the complexity of even the simplest bodily movement, and the still ob- scure relation between nerve impulse and the in- tensity of muscle reaction. A. B. Howell attempted to determine the rela- tions between various locomotor specializations (cursorial, saltatorial, aquatic) and musculature by comparing various representatives of such locomo- tor types regardless of their taxonomic affinities. This approach to muscle mechanics is indirect, and involves no mechanical analysis or estimate of forces. The intent is simply to discover a con- sistent correlation between a particular function and a particular modification of the muscle pat- tern. It may be confidently assumed that any such correlation is mechanically significant, even though no engineering analysis is made. Howell himself repeatedly expressed his disappointment at the meager results of this method. It is appar- ent that because of the diversity of genetic back- ground in so heterogeneous an assemblage of more or less remotely related forms, only the crassest morphological convergences would be evident. The lower the taxonomic level the more homo- geneous the genetic background that lies behind the muscle pattern. Among representatives of a superfamily or family we may focus more sharply on divergences from the basic muscle pattern of the group, for differences at this taxonomic level are not likely to represent the accumulated load of in- numerable earlier specializations in different an- cestral lines. Here any departure from the norm may be assumed to be adaptive, even though the mechanics are too complex or too subtle to ana- lyze. For example, in a series of carnivores rang- ing from most carnivorous to most herbivorous the relative masses of the external masseter and zygo- maticomandibularis vary reciprocally, whereas all other elements of the masticatory musculature re- main constant (Davis, 1955). Even without ana- lyzing the complex and subtle functioning of the masticatory complex we may be sure that in this instance the mechanically significant alterations are localized in these two muscles. Bringing rep- resentatives of other orders, with their different heritage, into this comparison would have ob- scured this relation, which is valid only within the masticatory pattern of the Carnivora. Besides mass or area of cross section, the relative values of force diagrams and leverage systems may be compared among closely related forms in the same way. Thus an insight into the functioning of a muscle or a group of muscles may be had at second hand, without the actual direct mechanical analysis, or determination of absolute forces, that has so far proved impossible to achieve. The possibilities of this method of assessing rela- tive muscle mechanics have not been explored. It will be used here, so far as existing data permit. NOMENCLATURE AND ARRANGEMENT The nomenclature used here is the BNA, with such obvious modifications as are necessary be- cause of differences from human anatomy. There is, of course, no "proper" sequence in which muscles can be arranged, and various sys- stems have been advocated. The arrangement adopted here is that of Howell's Anatomy of the Wood Rat, which is largely topographical. It may be suggested that the index is a more satisfactory means of locating a given description than at- tempting to find it via some system of arrangement. Innervation of muscles is given only in special cases, since the nerve supply of carnivore muscles is given in any standard anatomy of the dog or cat. Perhaps the most important consideration in evaluating muscle (and skeletal) differences within an order or family is an accurate picture of the bony attachments. This cannot be obtained from verbal descriptions alone; only carefully drawn maps will do. The exact areas of attachment of all muscles (except axial and a few others) in Ailuropoda have therefore been carefully plotted on the bones, and appear in the section on the skeleton. Unfortunately, comparable data for other carnivores exist only for the dog (later edi- tions of Bradley) and cat (Reighard and Jennings). I. MUSCLES OF THE HEAD A. Superficial Facial Musculature M. platysma is much reduced. It extends as a band of rather uniform width from a point above and behind the auditory meatus to the corner of the mouth. A few of the dorsal fibers swing up- ward in front of the ear, to lose themselves in the superficial fascia. Anteriorly a few of the most dorsal fibers are separated from the main mass, arising over the zygoma. M. buccinator (figs. 82, 84) is a heavy flat muscle sheet that forms the foundation of the cheek. It is not divisible into buccal and molar 150 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 parts as it is in most mammals. Instead, the muscle forms a uniform sheet of fibers that con- verges partly into the mucosa of the lips near the angle of the mouth, and partly into a horizontal raphe running back from the angle of the mouth. The dorsal fibers arise from the alveolar surface of the maxilla just outside the last upper molar, beginning at about the level of the middle of this tooth. The line of origin runs caudad onto the rugose triangular area immediately behind the tooth. Ventrad of this area, fibers arise from the pterygomandibular ligament, which extends cau- dad across the inner face of the internal pterygoid muscle. The ventral fibers arise from the alveolar surface of the mandible, just outside the molar teeth, beginning behind the last lower molar and extending as far forward as the middle of the first lower molar. The remaining superficial facial muscles were damaged in removing the skin and were not dis- sected. B. Muscles of the Ear M. levator auris longus (cervico-auricularis- occipitalis of Huber) is a fan-shaped sheet arising from the dorsal midline just behind the posterior end of the sagittal crest. There is no division into two parts. The posterior half of the muscle in- serts on the pinna. The anterior half is continued forward over the top of the head. M. auriculus superior is a narrow band lying behind, and partly covered by, the levator auris longus. Arising from the midline beneath the levator auris longus, it inserts on the pinna just caudad of that muscle, and separated from it by the insertion of the abductor auris longus. M. abductor auris longus lies immediately anterior to, and partly above, the auriculus supe- rior, and has approximately the same width. Dis- tally it emerges from beneath the levator auris longus, and inserts on the pinna just behind it. M. auriculus inferior lies wholly beneath the levator auris longus, and has the same general re- lations. It is more powerfully developed than the auriculus superior or the abductor auris longus, and is more than twice as wide. M. abductor auris brevis is the most caudal of the auricular muscles. Its origin is beneath that of the levator auris longus, but the belly of the muscle emerges and inserts low on the posterior face of the pinna. M. adductor auris superior (auricularis ante- rior inferior of Huber) is a narrow band arising from the posterior end of the scutiform cartilage. It inserts on the anteromesal face of the pinna. M. adductor aris medius arises from the ex- treme posterior end of the scutiform cartilage, be- neath the origin of the superior. It extends as a narrow band back to the posterior surface of the pinna, where it inserts proximad of the abductors. C. Masticatory Musculature The masticatory muscles, which are chiefly re- sponsible for the characteristic skull form of Ailuropoda, are remarkable for their enormous development. Otherwise they do not differ much from the typical carnivore pattern. In all Carni- vora the temporal is the dominant element of the masticatory complex, forming at least half of the total mass of the masticatory muscles. The in- sertion tendon of the temporal extends into the substance of the muscle as a tendinous plate, into which most of the muscle fibers insert. Thus the temporal is a bipennate (or if several such tendi- nous plates are present, a multipennate) muscle, in which the functional cross section per unit of volume is much greater than in a parallel muscle such as the masseter (Pfuhl, 1936). In carnivores, because of the form of the mandibular articulation, fast snapping movements of the jaws depend largely on the masseter, whereas slower and more powerful cutting and crushing movements depend largely on the temporal. The masticatory muscles arise ontogenetically from the mandibular arch, by condensation about the peripheral end of the mandibular nerve. Other muscles arising from the mandibular arch, and likewise supplied by the third branch of the tri- geminal nerve, are the anterior belly of the digas- tric, mylohyoid, tensor tympani, and tensor veil palatini. M. temporalis (figs. 82, 83) is enormously de- veloped, filling the greatly expanded temporal fossa except for a small area behind the orbit that is occupied by fat. In an old, badly emaciated male (Mei Lan) this muscle weighed more than twice as much as in a black bear of comparable size, and the temporal and zygomaticomandibularis together nearly three times as much. The muscle is cov- ered externally by a tough deep temporal fascia, more than half a millimeter thick, that arises from the sagittal and lambdoidal crests and postorbital ligament and extends to the superior border of the zygomatic arch. A few superficial fibers of the temporal muscle attach to the zygomatic arch im- mediately behind the temporal fascia and insert into its inferior edge, thus forming a tensor of the temporal fascia. The external face of the temporal muscle is cov- ered with an extremely heavy tendinous aponeu- rosis, the deep temporal fascia, from which the j DAVIS: THE GIANT PANDA Planum tendineum temporalis 151 Lig. postorbitale M. temporalis M. buccinator; p. buccalis (sup.)' M. buccinator; p. buccalis (inf.) Raphe tendinosa Fig. 82. Masticatory muscles of Ailuropoda, seen from the left side. The temporal and masseteric fasciae have been re- moved, and a window cut in the temporal muscle to expose the tendinous plane that separates the superficial and deep layers of the temporal muscle. The superficial and deep layers of the masseter are inseparable anteriorly. Note that the in- sertion of the superficial masseter does not extend posteriorly onto the angular process of the mandible. superficial fibers of the muscle take origin. As is usual in carnivores, the muscle is divided into superficial and deep parts, separated by a heavy tendinous plate, the insertion tendon of the mus- cle, that extends between the sagittal crest and the superior and posterior borders of the coronoid process. Muscle fibers attach to both surfaces of this tendinous plate. Additional tendon sheets embedded in the substance of the muscle insert into the coronoid process (fig. 83), making this complex a truly multipennate muscle composed of innumerable short fibers. These additional ten- don sheets do not occur in Ursus (Sicher, 1944, fig. 13; Schumacher, 1961a), and the temporal is therefore a simpler and less powerful muscle in the bear. The superficial part arises from the whole deep surface of the tendinous aponeurosis except for a small area near the orbit, and, at the periphery of the muscle, from the edges of the temporal fossa. The fibers converge to insert on the external face of the coronoid process of the mandible and into the external surface of the tendinous plate. Along its inferior border this muscle is incompletely sep- arable from the zygomaticomandibularis. The deep part of the temporal is much thicker than the superficial part and its structure is more complex. A tendinous sheet extends between the prominent crest running obliquely upward on the floor of the temporal fossa, some distance above the superior orbital crest, and a crest on the coro- noid process above the mandibular foramen. This sheet separates the anterior part of the deep tem- poral into superficial and deep parts. Additional smaller tendon sheets, embedded in the substance of the muscle, eventually attach to the inner face 152 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. temporalis Aponeurosis temporalis Fascia temporalis post ArcUS ryfjninnJirilft / 1 M. zygomaticomandibularis M. massctericus prof. M. massetericus superf L Canalis mandibularis Sinus I Tendo M. temporalis Fossa olfactoria Crista orbitalis sup. Foramen opticum Crista orbitalis inf. Pars nasalis pharyngis A. maxiltaris int. Tendo accessorius M. temporalis M. pter\'goideus int. Gl. sublijigualis M. genioglossus M. mylohyoideus V.Jacialis ex I. M. digastricus Proc. angularis mandib. Fig. 83. Frontal section through head of an old emaciated male Ailuropoda (Mei Lan). The section passes through the coronoid process of the mandible (see inset). of the coronoid process. Muscle fibers arise from tiie whole floor of the temporal fossa, and from the deep surfaces of the several tendon sheets. Some of the fibers insert into the surface of the coronoid process, the insertion area extending ventrad as far as the mandibular foramen. Other fibers in- sert into the superficial surfaces of the several ten- don sheets. The temporal is an elevator of the mandible. Because of its multipennate structure it produces slow but very powerful movements. M. zygomaticomandibularis (fig. 83) is rela- tively larger than in any other carnivore examined. It is completely hidden beneath the masseter and zygomatic arch, and fills the masseteric fossa. Origin is from the whole internal face of the zygo- matic arch. The fibers converge toward the mas- seteric fossa, into which they insert by both muscle and tendon fibers. Tendon sheets embedded in the muscle near its insertion attach to crests on the floor of the masseteric fossa, and these tendons increase the available insertion area. The flber direction of the zygomaticomandibularis is down- ward, mesad, and slightly backward. In the sag- ittal plane the fibers are almost vertical, forming an angle of about 80° with the occlusal plane. In the frontal plane the angle is about 75° with the transverse axis of the head. In both planes the angles become increasingly vertical as the jaw is opened. The zygomaticomandibularis is primarily an elevator of the mandible. The muscle of one side of the head, in conjunction with the pterygoids of the opposite side, shifts the mandible transversely toward the side of the contracting zygomatico- mandibularis. This motion is the grinding com- DAVIS: THE GIANT PANDA 153 ponent of the jaw movements in Ailuropoda and other carnivores. M. masseter (figs. 82, 83) is powerfully devel- oped. It is more or less divisible into the usual two layers, although these are fused and insep- arable anteriorly. The pars superficialis is a thin sheet covering all but the posterior part of the profunda. More than the proximal half of the external face of the superficialis is covered with a heavy tendinous aponeurosis (aponeurosis 1 of Schumacher, 1961a), which is continuous posteriorly with the aponeu- rosis of the profunda. The muscle arises by this aponeurosis and by underlying muscle fibers from the anterior half of the inferior border of the zygo- matic arch. The fibers run backward and down- ward at an angle of about 45° with the occlusal plane, to insert non-tendinously into the inferior edge of the mandible, immediately below the coro- noid fossa, the insertion extending back as far as the angular process. At its insertion the muscle forms a tendinous intersection with the internal pterygoid . The posteriormost fibers do not extend beyond the angular process at the posterior end of the mandible to insert into the stylomandibular ligament, as they do in Ursus and other carnivores. The internal face of the superficialis is in veiy intimate contact with the underlying profunda, the two layers being inseparable anteriorly. The pars profunda is covered by the superfcialis, except for a narrow area along its posterior edge. It arises by fleshy and tendon fibers from the en- tire inferior border of the zygomatic arch, back to within 10 mm. of the mandibular fossa. The fibers have a slightly more vertical direction than do those of the superficialis. A tendon sheet em- bedded in the posterior part of the profunda, attaching to the zygomatic arch, partly divides the muscle into superficial and deep layers. The external face of the mandibular half of the pro- funda is covered with a heavy glistening aponeu- rosis (aponeurosis 2 of Schumacher, 1961a). In- sertion is made by means of this aponeurosis into the mandible along the inferior border of the coro- noid fossa. The fibers run backward and down- ward at an angle of about 55° with the occlusal plane. The masseter is an elevator of the mandible. Because it is composed of long parallel fibers it produces quick snapping movements, relatively less powerful than those of the temporal muscle. M. pterygoideus internus (figs. 70, 83, 84; lateralis of authors) is a rectangular group of par- allel fibers arising from the ventral edge and outer side of the perpendicular plate of the palatine, pterygoid, and sphenoid bones. The muscle is thin and delicate posteriorly, and is relatively smaller than in any other known carnivore. It shows a tendency to break up into three or more subequal elements. Insertion is into the promi- nent fossa on the inner side of the lower border of the ramus of the mandible, extending onto the angular process. A few of the delicate posterior fibers insert into the anterior end of the stylo- mandibular ligament. The internal pterygoids acting together elevate the mandible. Unilateral contraction simultane- ously elevates the mandible and shifts it toward the contralateral side. M. pterygoideus externus (figs. 83, 84; medi- alis of authors) is much shorter, but considerably thicker, than the internal pterygoid muscle. Its lateral end lies dorsad of the internal pterygoid, and its medial end posterior to it. Origin is by two heads, which are separated by the buccinator nerve. The more ventral head arises from the outer side of the pterygoid plate at its posterior end, extending as far back as the combined fora- mina ovale and rotundum. The other head con- tinues this origin up onto the skull, behind the optic foramen. The two heads fuse, and the re- sulting muscle extends straight laterad to its in- sertion, which is into the prominent pit on the anteromedial end of the condyle of the mandible. The two external pterygoids are antagonistic. Unilateral contraction shifts the mandible toward the contralateral side. Discussion of Masticatory Muscles We have seen (p. 72) that the skull in Ailuro- poda, and in herbivorous carnivores in general, is designed to promote the production of maximum forces at the level of the cheek teeth by (a) im- proving lever advantages, (b) increasing the space available to muscle tissue, and (c) resisting dis- integrating forces. The active forces themselves are of course sup- plied by the craniomandibular muscles. These may further enhance the efficiency of the mastica- tory apparatus in three purely morphological ways: (a) generalized increase in mass of contractile tis- sue, (b) selective increase in mass, involving only those elements that produce the forces involved in pressure and grinding movements, and (c) increase in functional' cross section. Each of these is evi- dent in the masticatory musculature of the giant panda. 1 The functional cross section is a section at right angles to the fibers. The anatomical cross section is a section at right angles to the long axis of the muscle. In a parallel- fibered muscle these two sections may coincide; in a pennate muscle they never do. 154 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 Lif. ptervgomandib. {cut & reflected) M. ptcrygoideus int. M. pterygoideus ext. "Hamulus pterygoideue Capsula orttcuijris Lig. stylomandib. (cut) Proe. angukais M. mylohyoideus Fig. 84. Masticatory muscles of Ailuropoda, medial view. Generalized Increase in Mass.— I have used brain weight as a standard for computing an index of the relative mass of the total masticatory mus- culature of one side of the head. The data are given in the accompanying Table 12. The weights are all from zoo animals, and consequently the values for the musculature are undoubtedly low, although all except the panda were in good flesh at time of death. The panda (Mei Lan), in addi- tion to his years in captivity, was much emaciated at the time of his death. Nevertheless these fig- ures show that the relative mass of the masticatory musculature in Ailuropoda is at least twice as great as in bears of comparable body size. That this increase is truly generalized is shown by the fact that the mass of the digastric, a muscle not involved in jaw closure, equaled 30 per cent of brain weight in Ailuropoda, whereas in the bears it was less than 10 per cent of brain weight. It is impossible to determine whether both bellies of the digastric are equally hypertrophied; certainly the anterior belly is involved. The masticatory musculature, except for the posterior belly of the digastric, is derived from the mandibular arch of the embryo. Also derived from this arch are the mylohyoid, tensor tympani, and tensor veli palatini. The mylohyoid is in no way involved in jaw closure, yet in Ailuropoda it is hypertrophied like the craniomandibular mus- cles (p. 157). I was unable to decide from inspec- tion whether the tiny tensors were relatively larger than in the bears. It is evident, however, that what is enlarged in the panda is not a functional unit, but a morphological unit — the muscular de- rivatives of the mandibular arch. The fact that all are hypertrophied shows that, in this instance at least, the morphological unit is also a genetic unit. Indeed, hypertrophy extends in a decreasing gra- dient, beyond the derivatives of the mandibular arch, to the entire musculature of the anterior part of the body (p. 182). The morphogenetic mechanism involved in the hypertrophy is prob- ably very simple. Selection undoubtedly favored an increase in the mass of the jaw-closing muscles Table 12.— RELATIVE MASS OF MASTICATORY MUSCULATURE Masticatory Musculature Digastric (gms.) (gms.) Ailuropoda melanoleuca ( d" ad.) 890 92 Ursus americanus ( 9 ad.) 322 26 Thalaretos marilimns ( cf ad.) 910 86 • Mean of two brain weights (489 gms., 507 gms.) given by Crile and Quiring (1940). The brain of the polar bear from which I dissected the muscles was not weighed. Brain Index: Masticatory Musculature (gms.) Brain 277 3.2 238 1.4 498* 1.8 I DAVIS: THE GIANT PANDA 155 in the panda, but the results extend far beyond other mammals, including man) the insertion ten- the functional unit. don of the temporal muscle continues into the Selective Increase ln Mass.— Relative masses muscle substance as a broad tendon sheet. Fibers of individual components of the masticatory com- of the temporal muscle insert obliquely into both plex may be compared by reducing each to a sides of this tendon sheet, and the temporal is percentage of the mass of the total masticatory therefore a pennate muscle. In Ailuropoda the complex (Davis, 1955). Data are given in the temporal has been converted into a multipennate accompanying table. muscle by tendinization of numerous fascial planes Table 13.— RELATIVE WEIGHTS OF MASTICATORY MUSCLES IN CARNIVORES (Including data from Davis, 1955) Ailuropoda Tremarctos Ursus Procyon Thalarctos Canis Felis [Mei Lan| ornatus americanus* lotor maritimus familiaris onca Wt.ingms. % % % % % % % Masseter superf 44 5 7.5 10 5 1 f 15 21 12 Masseter prof 60 7 2.5 2 3 J [ 3 2.5 Zygomaticomand 188 21 14 11 13 7 6 2.5 Temporalis 477 54 58 62 63 66 58 59 Pterygoideus internus 18 2 7 5 6 4 7.5 6.5 Pterygoideus externus 11 1 1 1 1 1 0.5 0.5 Digastric 92 10 10 9 9 10 9.5 8 Totals 890 100 100 100 100 100 99.5 100 •Means of two specimens; data for one individual from Starck (1935). All other figures from one individual each. I have pointed out elsewhere (Davis, 1955) that in the Carnivora the masses of only two muscles, the superficial masseter and the zygomaticoman- dibularis, appear to vary significantly with differ- ences in food habits, and that these two muscles vary reciprocally. A large superficial masseter appeared to be associated with carnivorous habits, a large zygomaticomandibularis with herbivorous habits. The additional data presented here con- firm this relation. Moreover, in Ailuropoda the superficial masseter is relatively smaller (except in Procyon, where it is equally small) and the zygomaticomandibularis larger, than in any other carnivore examined. The masseter, because it is composed of long parallel fibers, is particularly effective in producing quick snapping movements of the mandible — a movement obviously important to predaceous car- nivores. There is an important horizontal compo- nent in the action of the zygomaticomandibularis. This muscle, which in bulk far exceeds the more horizontally situated but tiny external pterygoid, is primarily responsible for lateral shifting of the mandible — a movement important to herbivorous carnivores. Thus, in addition to the generalized increase, there is a selective increase in mass among the masticatory muscles, and the results conform to the requirements of differing dietary habits. Increase in Functional Cross Section. — In the temporal muscle of all carnivores (and of many in the substance of the muscle, with muscle fibers attaching to both surfaces of these tendon sheets. What are the mechanical advantages of penna- tion in a muscle? A pennate fiber is the diagonal of a parallelogram of which one component repre- sents force along the axis of the insertion tendon while the other component tends to pull the inser- tion tendon toward the origin. Only the first of these two components represents useful work. The second is waste effort, whose magnitude varies with the angle of pennation but in all cases repre- sents an important fraction of the total energy of the contracting fiber. There is no such waste of energy in a parallel-fibered muscle, which is there- fore more efficient than a pennate muscle. Some advantage must offset the inefficiency of the pen- nate structure. Eisler (1912) suggested maximum utilization of attachment area as a factor in the pennation of muscles. He pointed out that powerful mus- cles are pennate in situations where available at- tachment area is limited, whereas other powerful muscles remain parallel-fibered in situations where the attachment area can be expanded. Eisler com- pared the multipennate human deltoid, with its anatomically restricted areas of attachment, with the parallel-fibered gluteus maximus, which has been able to expand its areas of attachment un- hindered. Available attachment area is obviously a limiting factor in the temporalis of Ailuropoda. 156 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 The temporal fossa has been expanded in all direc- tions, apparently to the limits that are compatible with other vital functions of the head (p. 46). The mass of the muscle, particularly its area of origin, cannot be increased further to achieve additional power. Pfuhl (1936) attempted to work out the mechan- ics of pennate muscles. The work (a) of a muscle is expressed in two terms: force (F), and the dis- tance (rf) through which the force is exerted : a=F .d (1) The force of a muscle may be expressed by the equation F = k . q (2) where q is the functional cross section and A; is a constant representing the unit of muscle power.' Thus for any value of a in equation (1) the propor- tion oiF can be increased by increasing the func- tional cross section of the muscle, that of d by increasing its length. For a given mass of muscle tissue, maximum force would therefore be achieved by arranging the muscle as a series of minimally short parallel fibers, which would give maximum functional cross section. Such an arrangement would usually produce architectural difficulties, since areas of origin and insertion would become unduly large. An alternative is the arrangement of the fibers in pennate fashion between more or less parallel sheets of bone or tendon. This loses a portion of the total energy of the muscle, as shown above, but enormously increases the func- tional cross section and therefore the power per unit of mass. Thus pennation is a device permit- ting maximum production of force in a minimum of space, and utilizing limited attachment area on the skeleton. This effect is multiplied by multi- pennation. The craniomandibular musculature of Ailuro- poda represents an extension of conditions in the bears, which in turn are a modification of condi- tions in more generalized carnivores. Indeed, in Tremarctos, the most herbivorous of the bears, the craniomandibular musculature appears to be about intermediate between Ursus and Ailuropoda. As will appear in the sequel, the generalized in- crease in the mass of the craniomandibular mus- cles of Ailuropoda is associated with a generalized hypertrophy of the skeletal muscles of the shoulder region, and probably has a very simple genetic basis. The morphogenetic basis underlying the other two adaptive modifications — increase in rela- ' The unit of muscle power is the tension produced by a muscle with a functional cross section of 1 cm'. For pur- poses of calculation it is assumed to be 10 kg. tive mass of individual muscles, and increase in functional cross section — is unknown. D. Interramal Musculature These three muscles form a topographic, but not a morphological, unit. Ontogenetically they are derived from two different sources: the anterior belly of the digastric and the mylohyoid (from the mandibular arch) are supplied by the trigeminal nerve; the posterior belly of the digastric and the stylohyoid (from the hyoid arch) are supplied by the facial nerve. At least the elements derived from the mandibular arch are hypertrophied like the craniomandibular muscles derived from this arch. Of the elements derived from the hyoid arch, the stylohyoid is absent in Ailuropoda and there is no way of determining whether hypertrophy of the digastric involves the fibers of its posterior belly. M. digastricus (figs. 82, 83, 85) is a powerfully developed muscle, triangular in cross section, with the base of the triangle ventrad. The muscle has a thickness of 22 mm. The mass of the muscle is shot through with powerful longitudinal tendon fibers. Origin is from the paroccipital process and the ridge connecting this process with the mastoid process. The muscle is covered with a tendinous aponeurosis at its origin; there is also a small ac- cessory tendinous origin from the mastoid process. Insertion is into the inner surface of the mandible, from a point opposite the second molar tooth back as far as the mandibular foramen. A fine tendinous inscription runs across the belly of the muscle near its middle, marking the juncture of the anterior and posterior bellies. The digastric is relatively much larger than in the bears (Table 12) , but there is no way of deter- mining whether both bellies share in this hyper- trophy. Certainly the anterior belly is enlarged. M. stylohyoideus is absent. This muscle is tjT)ically composed of two parts in carnivores, a superficial slip external to the digastric and a deeper part internal to the digastric. Either may be absent, although there seems to be no previous record of both being absent simultaneously. Noth- ing corresponding to either part could be found in the specimens of Ailuropoda dissected. M. mylohyoideus (figs. 83, 84, 85) is a thick sheet that fills, with its fellow, most of the space between the rami of the mandible. Anteriorly a small space exposes the end of the genioglossus. The muscle arises from the medial surface of the mandible just below the alveoli of the teeth, from a point opposite the first molar to the angular process. The general direction of the fibers is transverse, although anteriorly and posteriorly DAVIS: THE GIANT PANDA 157 M. geniogloesu*' M. mylohyoideus' / M' geniogloasus M geniohyoideus litifualis M. styloglossus M. pterygoid eus int. M. pterygoid eus ext. M. thyreopharyngeua: constr. phar. post M. stemothyreoideus oc. mastoideus 0» stylohyale Proe. paroceipitalis N. hypogU>ssus M. constrictor pharyngis medius M. hyogloosus M. thyreohyoideus M. cricothyreoideus pars recta M. cricothyreoideus pars obliqua Fig. 85. Muscles of the head of Ailuropoda, ventral view. they diverge to the mandibular symphysis and the hyoid, respectively. Insertion is made in the usual way into a median raphe with the opposite muscle, and posteriorly into the hyoid bone. Medially the inner surface of the mylohyoid is almost insepa- rably united to the geniohyoid. The mylohyoid is much thicker, particularly near its origin (fig. 83), than is the mylohyoid of bears. E. Muscles of the Tongue The extrinsic muscles of the tongue show none of the hypertrophy that characterizes the cranio- mandibular muscles. Ontogenetically these tongue muscles arise from the ventral portion of the occip- ital myotomes. They are innervated by the hypo- glossal nerve. M. styloglossus (fig. 85) takes extensive origin from the stylohyal segment of the hyoid appara- 158 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 tus. The fibers diverge over the ventrolateral sur- face of the tongue before they disappear into the substance of the tongue itself. M. hyoglossus ifig. 85) arises from the inferior surface of the body of the hyoid, except for the area occupied medially by the origin of the genio- hyoideus, and the proximal part of the posterior horn. The fibers run straight anteriorly for a short distance before they penetrate the tongue, behind and laterad of the genioglossus and mesad of the styloglossus. M. genioglossus (figs. 83, 85) is a narrow band arising from the sjTnphysis just laterad of the mid- line. The origin of this muscle is ventral and lat- eral to the origin of the geniohyoideus. The muscle runs posteriorly, separated from the ven- tral midline by the geniohyoideus, and enters the tongue partly anterior to and partly medial to the hyoglossus. IL MUSCLES OF THE BODY A. Muscles of the Neck 1. Superficial Group M. sternomastoideus (fig. 86) is a heavy flat band about 40 mm. wide at its widest part (near its insertion). It arises, partly tendinously and partly fleshily, from the anterior border of the manubrium and the proximal end of the first costal cartilage. The muscle widens somewhat at its insertion, which is made on the lateral and ven- tral borders of the mastoid process. There is no indication that the sternomastoideus fuses with its mate at the midline. M. cleidomastoideus (fig. 86) arises from the dorsal edge of the stemomastoid, at a point about 70 mm. anterior to the origin of the latter muscle. With a maximum width of only 25 mm., it is con- siderably narrower than the sternomastoideus. The two muscles run forward side by side, the cleidomastoideus inserting on the lower part of the lambdoidal crest as a direct continuation of the in- sertion of the sternomastoideus, although the two muscles remain completely separate. 2. Supra- and Infrahyoid Group M. omohyoideus (figs. 86, 89) is a narrow rib- bon, about 16 mm. wide, arising from the coraco- vertebral angle of the scapula. It runs forward and downward, passing between the scalenus and the stemohyoideus. Near its insertion it divides into two bellies. The larger of these inserts on the hyoid, deep to the insertion of the stemohyoideus. The other belly inserts aponeurotically on the ven- tral face of the digastric, near its medial border. M. stemohyoideus figs. 86, 87, 89, 90) arises from the anterodorsal surface of the manubrium, a few of the most lateral fibers reaching the costal cartilages. It runs craniad as a narrow, flat band, in contact with its mate of the opposite side near its origin, but diverging from it farther anteriorly. Insertion is made on the thyrohj-al element of the hyoid. M. stemothyroideus (figs. 85, 87, 89, 90) is inseparable from the sternohyoid at its origin and as far forward as a tendinous intersection which crosses the common mass of these two muscles about 40 mm. in front of the manubrium. Ante- rior to this p>oint the sternothyroid lies partly above (dorsal ) and partly lateral to the sternohyoid. It inserts on the thyroid cartilage, just above the insertion of the sternohyoid. M. thyrohyoideus (figs. 85, 87, 89) is a wide, flat band on the ventrolateral surface of the thy- roid cartilage. Arising from the posterior border of the thyroid cartilage, just laterad of the mid- line, the fibers nm anteriorly to their insertion on the posterior border of the thyrohyal and the body of the hyoid. M. geniohyoideus (fig. 85) is a narrow band running from the symphysis mandibuli to the body of the hyoid, closely applied to its fellow of the opi>osite side. Arising from the s>"mphysis deep to and laterad of the genioglossus, it inserts on the anteroventral surface of the body of the hyoid, just laterad of the midline. 3. Deep Lateral and Subvertebral Group M. scalenus (figs. 86, 89) is divisible into the usual longus and breris. The short division Ues mostly beneath the much more powerful long divi- sion. The scalenus longus arises by short, stout tendons from the third to seventh ribs, its origins interdigitating with the sraratus anticus. The longus is subdivided into a dorsal part, which arises from the second to the fifth ribs, and a medial part from the sixth and seventh ribs. The bre\Ts arises fleshily from the first rib near its junction with the costal cartilage. The two divisions unite in the cervical region, and the resulting common mass j inserts on the transverse processes of the last five cervical vertebrae. M. longus colli arises from the ventral surfaces of the bodies of the first six thoracic vertebrae and from the ventral sides of the transverse processes of the sixth to third cervical vertebrae. The usual simple distinction of the thoracic and cervical parts of the muscle because of difference in their fiber directions is scarcely possible on the present speci- men. The fibers arising from the thoracic verte- brae are gathered into a tendinous band that in- a 3 2 I >> o Xi o t •3 2 s o ^ 3 . E oo !2 a U) c C^ o ■§ T3 ►J 159 160 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 serts into the transverse process of the sixth cer- vical. The fibers from the cervical vertebrae have the customary insertion into the next vertebra craniad of the one from which they arise, and into the ventral surface of the arch of the atlas. M. longus capitis is a prominent subcylindri- cal muscle, somewhat flattened dorsoventrally. It arises by fleshy fasciculi from the tips of the trans- verse processes of the sixth to the second cervical vertebrae. Insertion is into the prominent scar on the ventral side of the basioccipital. M. rectus capitis ventralis is a very slender muscle lying mesad of the longus capitis, and in contact with its mate of the opposite side at the midline. It has the customary origin from the ventral surface of the body of the atlas, and in- sertion into the basioccipital mesad and caudad of the longus capitis. B. Muscles of the Trunk 1. Muscles of the Thorax M. panniculus carnosus is rather feebly de- veloped; the dorsal division is represented only by an almost insignificant vestige. The borders of the ventral division do not reach the midline either dorsally or ventrally. A few fibers arise on the inner surface of the thigh, and the sheet then broadens as it passes anteriorly, reaching its great- est width over the posterior ribs. At this point it is approximately 170 mm. from the dorsal midline and 80 mm. from the ventral. The sheet then gradually decreases in width as it passes craniad. At the point where it passes under the pectoralis it is only about 50 mm. wide. The ventral fibers insert on the bicipital arch, the dorsal ones on the inner face of the pectoralis profundus. The dorsal division is represented only by two narrow ribbons, lying immediately dorsad of the ventral division, that run up onto the shoulder for about 50 mm. and insert into the epitrochlearis immediately below the latissimus. M. pectoralis superficialis (fig. 89). As in the bears, the superficial pectoral sheet is a compound muscle composed of the pectoralis superficialis an- teriorly and the reflected posterior edge of the pro- fundus posteriorly. Fusion is so intimate that the boundary between superficialis and profundus can- not be determined, but as in the bears the posterior part of the superficial layer is innervated by the medial anterior thoracic nerve. Along its posterior border the superficial sheet is folded sharply under and continued forward as a deeper layer (the profundus) immediately be- neath the superficial one. Thus a very deep and well-marked pectoral pocket, open anteriorly and closed posteriorly, is formed. The superficial sheet arises from the entire ma- nubrium and from the corpus sterni back to the level of the eighth sternal rib. The fibers converge toward the humerus, and insert into the pectoral ridge in a narrow line along the middle half of the bone. In other carnivores (including the bears) insertion is into the deltoid ridge. In Ailuropoda the proximal end of the insertion line deviates slightly from the pectoral ridge toward the del- toid ridge, but by no means reaches the latter. Probably the tremendous development of the del- toid and lateral triceps in the panda has crowded the superficial pectoral off the deltoid ridge and forward onto the pectoral ridge. M. sternohumeralis profundus is a narrow band anterior to the superficial sheet. It arises from the anterior end of the manubrium, increases in width as it passes toward the shoulder along the anterior border of the superficial sheet, and inserts on the lateral surface of the humerus immediately below the greater tuberosity, in a line that con- tinues proximad from the insertion of the super- ficial sheet. The lateral anterior thoracic nerve and its accompanying blood vessels pass through the split between this muscle and the supei-ficialis. M. pectoralis profundus (figs. 89, 133) lies mostly beneath the supei-ficialis, although as stated above its posterior edge is folded forward and fused with the posterior border of the superficialis. It is by far the widest element of the pectoral complex. It is not divisible into anterior and posterior parts. Origin is from the corpus sterni posteriorly, deeper fibers arising from the sternal cartilages, from the eighth forward to the third. At the anterior level of the third and fourth sternal cartilages the mus- cle arises wholly from the cartilages, none of the fibers reaching the sternebrae. The most poste- rior fibers are joined on their under side by the panniculus. Insertion extends almost the entire length of the humerus, beginning proximally on the greater tuberosity at the edge of the bicipital groove, and continuing distad on the pectoral ridge to within 60 mm. of the distal end of the humerus. M. pectoralis abdominalis (fig. 89) is a nar- row thin band lying posterior to the profundus. It arises from the rectus sheath at the level of the costal arch, passes beneath the posterior edge of the profundus, and inserts with the panniculus on the deep surface of the profundus, not reaching the bicipital arch. The abdominalis is degenerate. M. subclavius is entirely wanting. M. serratus ventralis (magnus or anterior of some authors) and M. levator scapulae (fig. 86) form a perfectly continuous sheet, so that the boundary between them cannot be determined. The common muscle arises from the atlas and all DAVIS: THE GIANT PANDA 161 succeeding cervical vertebrae, and by fleshy fibers from the first nine ribs. The sHp arising from the fifth rib lies over the scalenus; those farther for- ward lie beneath it. Insertion is made along the inner surface of the whole vertebral border of the scapula. Mm. intercostales externi (figs. 87, 89). The fibers of these muscles run craniodorsad as far back as the eleventh rib. Between the eleventh and fourteenth ribs they run nearly horizontally. The muscles reach the costal cartilages of all but the first two ribs, although the intercostales interni are exposed medially as far back as the seventh rib. The part of the muscle between the ribs is fleshy anteriorly, becoming quite tendinous poste- riorly. Between the costal cartilages this ai'range- ment is reversed, the muscles being tendinous an- teriorly and fleshy posteriorly. A small group of fibers arises from the first costal cartilage near the manibrium and inserts on the inner face of the tendon of the rectus. The dorsal edge of the muscle forms a raphe with the inter- costal fibers lying dorsad of it, and the fiber direc- tion is more vertical than that of the intercostales. It is not known whether this represents a part of the intercostalis internus or not. Mm. intercostales interni (figs. 87, 90) are, as usual, more extensive than the external inter- costals. They occupy all the space between the ribs and the costal cartilages. The fibers take the usual forward and downward direction. M. supracostalis (fig. 86) is a narrow band arising from the fourth rib. Running anteriorly closely applied to the ventral edge of the scalenus, it swings ventrad to insert on the costal cartilage of the first rib. M. transversus thoracis (fig. 90) is a thin sheet, more or less divisible into separate bands, that occupies the space between the third and eighth sternal cartilages on the inner side of the thoracic wall. Origin is from all the sternal seg- ments except the first two and from the anterior third of the xiphoid cartilage, and insertion is made on the sternal cartilages and aponeurotically on the fascia covering the inner surface of the in- ternal intercostals. A narrow ribbon of muscle arises from the third sternal segment and passes forward to insert apo- neurotically into the fascia of the intercostals. It is not known whether this represents a part of the transversus thoracis or not. M. diaphragma (fig. 90). Pars lumbalis is di- vided into three crura. Crus laterale, which is the largest of the three, has a double origin. The lat- eral fibers arise by means of a stout tendon from the ventrolateral surface of the third lizmbar ver- tebra. Medial fibers arise, at the level of the sec- ond lumbar vertebra, from the lateral edge of a long tendon that runs cephalad from the ventral surface of the fourth lumbar vertebra. This ten- don runs forward along the medial border of the pars lumbalis as far as the aortic notch, and gives rise to all the remaining fibers of this part of the diaphragm. On the deep surface of the lateral crus some of the fibers also arise directly from the second lumbar. Crus intermedium is very narrow. It is separated from the lateral crus throughout almost its entire length by a branch of the phrenic nerve, while its medial border slightly overlaps the lateral border of the medial crus. It arises from the medial tendon mentioned above, at the level of the anterior border of the second lumbar vertebra, its origin being continuous with that of the lateral crus. Crus mediale arises from the me- dial tendon at the level of the posterior border of the first lumbar vertebra, its origin being continu- ous with that of the intermediate crus. The me- dial crus fuses with its fellow of the opposite side cephalad of the hiatus aorticus, which is situated below the thirteenth thoracic vertebra. Pars costalis arises from the ninth to the eleventh costal cartilages by a series of interdigitations with the transversus abdominis. These interdigitations do not correspond perfectly in number with the ribs, some costal cartilages receiving more than one digitation each; nor do the digitations corre- spond exactly on either side of the sternum. Pars sternalis arises from the lateral border of the posterior part of the elongate xiphoid process. It is a narrow band that promptly joins the adja- cent medial border of the pars costalis. 2. Muscles of the Abdomen M. rectus abdominis (figs. 86, 87, 89, 91) ex- tends as a thin, rather narrow, band from the pel- vic symphysis to the first costal cartilage. It reaches its greatest width of 100 mm. at about the level of the sixth sternal cartilage. Tendinous in- scriptions are absent. The muscle arises by fleshy fibers, covered by a heavy aponeurosis, from the posterior part of the pelvic symphysis, the origin extending anteriorly along the ventral midline. A few of the fibers nearest the midline insert into the linea alba just behind the xiphoid cartilage. Suc- cessive slips farther laterad insert on the fifth, sixth and seventh costal cartilages, and slightly less than the lateral half of the muscle is continued forward, to insert by a wide tendon on the first costal carti- lage. This tendon begins at the level of the third costal cartilage. The rectus does not participate in the formation of the inguinal canal. M. atiantoscapularis (cut) M. acromiotiap. (cut) M. levator scapulae vent. M. cephalohumer. M. acromiodclL M. brachialis M. atiantoscapularis M. triceps lateralis ■M. triceps longus ■. dca-so-epitrochlearis •M. spinodeltoideus 'M. acromiotrap. M. spinotrap. M. obliquus abdom. extemus M. vastus lateralis M. qxiadratus femoris M. adductor M. semimembranosus M. semitendinosus M. tenuissimus Fascia lumbodorsalis superf. M. glutaeus superf. M. tensor fasciae latae M. semimonbranoeus M. biceps femoris Fig. 88. Dorsal view of body musculature of Ailuropoda, superficial layer on right, deeper layer on left. 162 M. omohyoideus M. stemocleidomastoideus ' M. cephalohumer. M. stemohumer. prof, M. pect. superf, Hyoid M. th}rreohyoideus M. cricothyreoideus M. stemothyreoideus M. stemohyoideus M. rectus abdominis (cut) M. vastus med.- M. sartorius M. adductor M. semimembranoeus M. gracilis M. aemitendinosus M. intercost. ext M. obliquus intemus M. tensor fasciae latae M. iliopsoas M. adductor M. vastus med. M. rectus femoris v — M. adductor M. aonimembranoeut M. aemitendinosus Fig. 89. Ventral view of body musculature of Ailuropoda, superficial layer on right, deeper layer on left. 163 164 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. sternohyoideus + M. stemothyreoideus M. stemomastoideus A. & V. mammaria int. M. transv. thoracis M. intercost. int, M. diaphragma, pars sternal is Proc. xiphoideus M. diaphragma^ pars costal is M. transversus abdominis Fig. 90. Ventral wall of thorax of Ailuropoda, internal view. M. obliquus abdominis externus (figs. 86, 88, 89, 91) arises by short tendons from the fourth to the ninth ribs, and by fleshy fibers from the tenth to the thirteenth. Apparently none of the fibers reach the dorsal fascia. Posterior to the serratus ventralis the obliquus attaches to the ribs (10-13) immediately behind the origins of the latissimus dorsi. It is difficult to determine whether the fibers dorsal to the origins of the latissimus rep- resent continuations of the obliquus or whether they are external intercostals, as the fiber direc- tion is exactly the same. When the dorsal border of the obliquus is lifted, however, the muscle sheet dorsal to it is found to be perfectly continuous with the intercostals lying beneath the obliquus. Insertion: the muscle fibers slightly overlap the lateral edge of the rectus before giving way to the tendinous aponeurosis that extends over the rectus to the linea alba at the ventral midline (the rec- tus sheath). In the inguinal region the aponeu- rosis expands into a large triangular sheet, the abdominal tendon (see below), which inserts into the posterior third of the inguinal ligament. DAVIS: THE GIANT PANDA 165 M. obliquus abdom. ext. M. aartoriua Vagina m. red. abdom. M. obLabd. int (cut &■ red.) Eminentia Uiopeetinea M. obL abd.ext. [Tendo abd] (cut & rea) Aimulus iniuimU int. M. adductor Tertdo praepubieu* Tendo praepubicus gracilis Fig. 91. The inguinal region of At'/Mropoda. The dotted line shows the position of the internal inguinal ring. The arrows pass through the lacuan musculo-vasorum (lateral) and inguinal canal (medial). M. obliquus abdominis internus (fig. 87, 89, 91) is much less extensive than the externus. It is rather sharply divided into two parts: an ante- rior division (pars costalis) that inserts on the last ribs, and a more extensive posterior part (pars abdominalis+pars inguinalis) that inserts aponeu- rotically into the ventral belly wall. These two divisions are separated by a considerable gap ven- trally. The anterior division arises from the crest of the ilium from the anterior superior iliac spine mesad nearly to the middle of the crest, and from the iliac end of the inguinal ligament, and inserts on the last three ribs. The posterior division arises exclusively from the inguinal ligament. Posteri- orly the fibers run almost vertically downward, or may even run slightly ventrocaudad ; anteriorly they run diagonally forward and downward. The muscle terminates in a tendinous aponeurosis that participates in the formation of the rectus sheath (see below). This aponeurosis is more extensive anteriorly, where the muscle fibers fail by 40 mm. to reach the edge of the rectus. Posteriorly the muscle fibers extend to the edge of the rectus. In the inguinal region the internal oblique is perforated by the inguinal canal. M. transversus abdominis (figs. 87, 89, 90, 91) arises from the cartilages of the last six ribs, interdigitating with the origins of the diaphragm. Additional origin is taken from the lumbodorsal fascia, from the tip of the ilium, and from the an- terior end of the inguinal ligament. The muscle terminates in a tendinous aponeurosis that fuses with the inner layer of the aponeurosis of the in- ternal oblique to form the inner sheath of the rec- 166 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 tus. The posteriormost fibers insert into the lateral third of the iliac crest. M. cremaster (fig. 91) arises as a fine tendon from the inguinal ligament 25 mm. anterior to the internal inguinal ring. The tendon takes accessory origin from the transverse fascia on its way to the inguinal canal. As it enters the canal the tendon fans out into a band of muscle fibers that passes through the canal dorsad of the spermatic cord, and expands to form the cremasteric fascia around the tunica vaginalis of the testis. M. quadra tus lumborum (fig. 100) is a com- plex muscle arising from the last three thoracic vertebrae and ribs and the transverse processes of all the lumbar vertebrae. Insertion is into the transverse processes of the lumbars and the in- ternal lip of the iliac crest for about its middle third and the adjacent inferior surface of the ilium. 3. The Inguinal Region. Figure 91. The structures in the inguinal region are some- what modified in Ailuropoda, in comparison with related carnivores, because of the extremely short pelvic symphysis. The abdominal tendon [Bauchsehne+Becken- sehne of German veterinary anatomists] is the in- sertion aponeurosis of the external oblique muscle. Anteriorly the aponeurosis of this muscle passes into the outer rectus sheath, while in the inguinal region it forms a large triangular sheet that fills the angle between the linea alba and the inguinal ligament. The aponeurosis is perforated by the inguinal canal ; the part anterior to this perforation is the "abdominal tendon," the part posterior to it the "pelvic tendon" of the German anatomists. The aponeurosis inserts into the posterior third of the inguinal ligament, from the level where the femoral vessels emerge back to the symphysis. The lamina femoralis, which in the dog and other domestic quadrupeds splits off from the abdominal aponeurosis at the lateral border of the inguinal ring and runs onto the medial surface of the thigh, appears to be wanting in Ailuropoda. The prepubic tendon is a heavy, compact liga- ment extending from the iliopectineal eminence back to the anterior border of the pelvic symphy- sis, where it meets its mate of the opposite side. The tendon is more or less continuous with the in- guinal ligament anteriorly. It lies superficial to the pectineus muscle, and arises chiefly from the origin tendon of that muscle. Where it passes over the origin tendon of the gracilis near the symphy- sis, the prepubic tendon is inseparably fused with the tendon of that muscle. The tendon provides attachment for the linea alba and the posterior- most fibers of the internal oblique. The inguinal ligament lies at the juncture of the medial surface of the thigh and the wall of the abdomen. It extends from the anterior iliac spine to the iliopectineal eminence. Beyond the eminence it is continued posteriorly as the prepubic tendon. As in other quadrupeds, the inguinal ligament is poorly defined in Ailuropoda. Anteriorly it is little more than a fiber tract from which the posterior fibers of the internal oblique take origin. Poste- riorly, where it bridges over the lacuna musculo- vasorum, it is a heavier and more sharply defined ligament. Between the inguinal ligament and the ventral border of the pelvis there is a large gap, the lacuna musculovasorum (lacuna musculorum + lacuna vasorum of human anatomy; the iliopectineal liga- ment, which separates these in man, is wanting in quadrupeds). Through this opening the iliopsoas muscles and the femoral vessels and nerve pass from the abdominal cavity onto the thigh. In Ailuropoda (as in the dog) the femoral vessels lie ventrad of the iliopsoas, rather than posterior to it, and no true femoral ring can be distinguished. The lacuna is about 50 mm. long. The inguinal canal is very short, its length being little more than the thickness of the abdom- inal wall. It is about 12 mm. long, and is directed posteriorly and slightly medially. It is situated about 30 mm. in front of the pelvic symphysis. The inlet to the canal, the internal inguinal ring, is formed by a hiatus in the internal oblique mus- cle; the anterodorsal border, between the limbs of the opening in the muscle, is formed by the in- guinal ligament. The rectus abdominis does not participate in forming the medial border of the ring, as it does in the dog. The internal ring meas- ures about 30 mm. in long diameter. The outlet, the external inguinal ring, is associated with the abdominal tendon of the external oblique. In the inguinal region this sheet splits to form the lateral and medial limbs of the ring. The fibers of the lateral limb radiate into the origin tendon of the pectineus and the prepubic tendon, while the fibers of the medial limb pass into the rectus sheath. The ring is completed posterodorsally by the pre- pubic tendon; i.e., the two limbs do not re-unite posteriorly, but merely form a ventral arch around the spermatic cord. The sheath of the rectus abdominis is formed externally by the aponeurosis of the external oblique fused with the ventral layer of the apo- neurosis of the internal oblique. Internally the sheath is formed by the dorsal layer of the aponeu- rosis of the internal oblique fused with the apo- neurosis of the transversus abdominis. Thus the rectus muscle is embraced between the dorsal and DAVIS: THE GIANT PANDA 167 ventral layers of the internal oblique aponeurosis. In the dog the inner layer of the rectus sheath ". . . is formed for the most part by the terminal aponeurosis of the transversus abdominis . . . and in the anterior portion in addition by an inner layer of the terminal aponeurosis of the obliquus abdominis internus." (Baum and Zietzschmann.) The inguinal region of Ailuropoda differs from that of the dog (Baum and Zietzschmann; the only other carnivore in which this region is known) in several respects. The following peculiarities of the giant panda may be mentioned: (1) The rectus does not participate in the for- mation of the inguinal canal. (2) The rectus inserts into the posterior part of the symphysis. (3) The cremaster does not arise from the pos- terior border of the internal oblique. (4) The abdominal tendon of the external oblique does not form the entire circum- ference of the external inguinal ring. I^. Muscles of the Back Superficial Secondary Back Muscles.— M. cephalohumeralis (= clavodeltoideus +clavotra- pezius) (figs. 88, 134) is powerfully developed. Near its insertion it has a thickness of about 20 mm. Its origin, which is continuous with that of the acromiotrapezius, extends on the lambdoidal crest from the level of the dorsal border of the zygoma to the dorsal midline, then by aponeurosis from the ligamentum nuchae for 90 mm. along the mid- line of the neck. The anterior border is slightly overlapped by the temporalis. The fibers converge over the anterior border of the shoulder, and insert fleshily into the lower half of the deltoid ridge and the area between this ridge and a second ridge midway between the deltoid and pectoral ridges. At its insertion the muscle forms a partial raphe with the acromiodeltoid laterally and with the pectoralis superficialis and profundus medially. The clavotrapezial part of the cephalohumeral is innervated by the spinal accessory, and the clavodeltoid part by the axillary nerve. Action: Chief extensor of the fore leg. M. acromiotrapezius (figs. 88, 134) is a thin, rectangular sheet arising from the dorsal midline by a long, broad aponeurotic sheet; fleshy fibers appear as the muscle crosses the scapular border. The muscle is thus sharply divided into two parts, a fleshy part lying over the scapula and an aponeu- rotic part between the vertebral border of the scapula and the dorsal midline. Its origin is con- tinuous with the aponeurotic origin of the cephalo- humeral anteriorly, and extends a distance of 110 mm. along the dorsal midline. The fleshy part of the muscle has a length of only 70 mm. Insertion is made for a distance of 105 mm. into the humeral half of the scapular spine. M. spinotrapezius (figs. 88, 134) is triangular in outline. The anterior border is sharply con- cave, so that a portion of the underlying rhom- boids and supraspinous fascia is exposed between this muscle and the acromiotrapezius. The pos- terior edge is concave and thin, but the muscle becomes quite heavy anteriorly. Origin is from the spinous processes of the thoracic vertebrae for a distance of 160 mm. The anterior border is over- lapped slightly by the acromiotrapezius near the midline. The fleshy part of the muscle stops abruptly at the posterior border of the scapula, and the muscle continues forward and downward across the scapula as a wide, heavy aponeurosis that inserts into the superficial fascia of the infra- spinatus. Thus the condition in the spinotrapezius is the reverse of that in the acromiotrapezius, where the part of the muscle lying over the scap- ula is fleshy and the part beyond the scapula is aponeurotic. The relations of fleshy and aponeurotic parts of the acromio- and spinotrapezius to the underlying scapula in Ailuropoda appear to be pressure phe- nomena. Similar conditions are known from human anatomy, e.g., the digastric. It is noteworthy, however, that the trapezius is almost exactly the same in the Ursidae (verified in our specimens of Selenarctos and Tremarctos), and is surprisingly similar, considering the difference in body size and proportions, in Ailurus. The development of these extensive aponeurotic sheets is even indicated in Bassariscus and Procyon. The dogs, on the other hand, show nothing comparable to it, nor do other carnivores, including such large forms as the hyenas and lion. Action: The trapezius muscles elevate the scap- ula and rotate it counterclockwise. M. latissitnus dorsi (figs. 88, 134) is very pow- erfully developed. It has the customary triangu- lar form. The anterior border is overlapped by the spinotrapezius. It arises mostly by aponeuro- sis from the mid-dorsal line, fleshy fibers reaching the midline only at a point just behind the spino- trapezius. Ventrally and ventro-posteriorly the muscle takes origin from the seventh to eleventh ribs. Origin from the seventh rib is limited to a very few fibers, but the origin from each successive rib increases in length until on the eleventh it ex- tends over 95 mm. The fibers converge toward the axilla, and insertion is made by two heads. The smaller head inserts chiefly into the inner face of 168 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 the panniculus carnosus, a few of the most poste- rior fibers reaching the epitrochlearis. The main mass of the muscle forms a powerful raphe with the epitrochlearis, and these two muscles make a common insertion into the tendon of the teres major. Action: Chief flexor of the arm. M. rhomboideus (figs. 86, 88, 92, 134) is more or less divisible into two parts. The muscle is elongate triangular in outline, and arises in a con- tinuous line from the lambdoidal crest at about the level of the dorsal border of the zygoma up to the dorsal midline, then back for 270 mm. along the midline of the neck. The muscle may be separated, particularly near its insertion, into anterior and posterior masses, of which the posterior is much the more extensive. Insertion is made into the dorsal half of the coracoid border and entire ver- tebral border of the scapula. The anterior edge of the posterior part lies partly over that of the anterior. Action: Draws the scapula toward the verte- bral column. M. occipitoscapularis (rhomboideus anterior or capitis of authors) (fig. 134) is a narrow band arising from the lambdoidal crest. The muscle runs backward, separated from the rhomboideus by the dorsal branch of the A. and V. transversa colli, to insert on the coracovertebral border of the scapula, beneath the insertion of the anterior part of the rhomboideus. Action : Draws the scapula forward. M. atlantoscapularis (levator scapulae ven- tralis of authors; omo-cleido-transversarius of Carl- sson) (figs. 86, 134) is a narrow, heavy band arising from the transverse process of the atlas. For a short distance it is inseparable from the first digi- tation of the levator scapulae, with which it has a common origin. Immediately distad of its origin it is easily separable into two subequal parts, which embrace a branch of the fourth cervical nerve be- tween them. This separation loses its identity near the insertion, which is made, by means of a short fine tendon, into the metacromion of the scapula, at the juncture of the acromiodeltoideus, the spinodeltoideus, and the acromiotrapezius. M. serratus dorsalis anterior (fig. 86) arises by fleshy slips from the posterior borders of the fifth to tenth ribs. The fibers from these six ori- gins more or less unite to form a continuous sheet that inserts aponeurotically into the dorsal fascia. M. serratus dorsalis posterior (fig. 86) is lim- ited to two slips. The more anterior of these arises from the twelfth rib; the posterior from the thirteenth, with a few fibers coming from the four- teenth. The fibers run straight dorsad, to insert independently of one another into the dorsolum- bar fascia by means of aponeuroses. Deep Intrinsic Back Muscles.— M. splenius (figs. 86, 87, 92) is very powerfully developed, par- ticularly along its lateral border, where it attains a thickness of 15 mm. Posteriorly the muscle arises by a wide tendinous aponeurosis from the dorsoliunbar fascia at about the level of the fifth thoracic vertebra; this aponeurosis lies beneath the origin of the serratus posterior superior. Origin, by a similar aponeurosis, is taken along the mid- line as far forward as the lambdoidal crest of the skull. This medial aponeurosis has a width of 15-20 mm. Insertion is made on the lambdoidal crest, just beneath the insertion of the rhom- boideus, and from the mastoid process down to its tip. Tendinous intersections are absent. The usual undifferentiated muscle mass occu- pies the trough formed by the spines and trans- verse processes of the lumbar vertebrae. At the level of the last rib it divides to form three mus- cles: the iliocostalis, the longissimus, and the spi- nalis. The medial part of the muscle mass is covered with a heavy aponeurosis, which gives rise to many of the superficial fibers of all three muscles. M. iliocostalis (figs. 87, 88, 92) is the most lat- eral of the superficial back muscles. It gives off a tendinous slip to each of the ribs near its angle and to the transverse process of the last cervical vertebra. The more posterior tendons pass over one rib before inserting, those farther forward over two. Slips from all the ribs except the first four join the muscle as it runs craniad. M. longissimus (figs. 87, 88, 92) is the middle one of the three superficial back muscles. There is no demarcation between the pars dorsi and pars cervicis of human anatomy. On the other hand, the muscle is sharply divided into a lumbar part (M. ilio-lumbalis [Virchow], Pars lumborum m. longissimus dorsi [Winckler], M. longissimus lum- borum [Eisler]), arising from the ilium and covered by the heavy deep layer of the lumbar fascia; and a thoracic part. The thoracic part arises from the lumbar fascia, and farther anteriorly from the fascia between itself and the spinalis. There is the usual double insertion: medially by fasciculi into the anapophyses of the lumbar and thoracic vertebrae, and laterally by long tendons into all but the last four ribs and into the transverse proc- esses of the last six cervical vertebrae. M. longissimus capitis (fig. 92) arises from the transverse processes of the last three cervical ver- DAVIS: THE GIANT PANDA 169 M. multifidus Vertebra thoraaUis I Nn. cervicales doraales M. splenitis (cut) M. rectus capitis dorsalis major (cut) w ,,. ... . \x -^"- biventer cervicus I M. rhomboideus (cut) M. oWiquus capitis post. \\ et complejcus ' M. multifidus cervicis \ \\ m. rectus capitis^ Axis-Proc. spiMlis\ \ \ dorsalis mediusN Coital M. rectus capitis lateralis' M. rectus capitis dorsalis minor M. obliquus capitis ant. Fig. 92. Deep muscles of neck and anterior thorax of Ailuropoda, right side. tebrae. It is composed of two very slender heads. One of these joins the ventral border of the sple- nius in the usual way, and thus inserts into the mastoid process. The other head, which comes from the anterior fibers of the common origin, lies deep to the splenius along the ventral border of the complexus, inserting with it into the occipi- tal bone. M. longissimus atlantis (fig. 92) is slightly larger than the combined heads of the longissimus capitis. It arises from the articular pi-ocesses of the third, fourth, and fifth cervicals, and inserts into the tip of the wing of the atlas. M. spinalis dorsi (figs. 87, 88) is the most me- dial and most extensive of the superficial back muscles. It is present only in the thoracic region. Origin is from the anterior edge of the deep lumbar fascia, and farther anteriorly from the fascia be- tween itself and the longissimus. The fibers run diagonally craniad and mesad, and insert, by ten- dons that become progressively longer, anteriorly into the tips of the spinous processes of all the thoracic and the first cervical vertebrae. M. semispinalis is represented only by the capitis, which is separable into a dorsal biventer cervicis and a ventral complexus. M. biventer cervicis (fig. 92) has three diagonal tendinous in- tersections. The muscle begins at the level of the fifth thoracic vertebra, arising posteriorly from a wide aponeurotic fascia that covers the underly- ing muscles. Additional origin is taken by means of tendinous fasciculi from the tips of the spines of the fourth, third, and second thoracics, and an- terior to this from the ligamentum nuchae, as well as from the transverse processes of the second to fifth thoracics. Insertion is fleshily into the occipi- tal crest near the dorsal midline. M. complexus lies beneath the biventer cervicis posteriorly. It begins at the level of the second thoracic vertebra, arising posteriorly from an aponeurotic fascia sim- ilar to that of the biventer. Additional origin is taken from the transverse processes of the first two thoracic and last four cervical vertebrae. Inser- tion is made, by mingled fleshy and tendon fibers, into the medial half of the occipital bone. The muscle lies partly deep to the biventer cervicis at its insertion. M. multifidus (fig. 92) is continued craniad from the extensor caudae medialis. In the lum- bar region it is deep to the spinalis. The muscle is, as usual, best developed in the lumbar region, where it is not separable into individual fasciculi; at the anterior end of the deep lumbar fascia it is fused with the spinalis. In the thoracic region the multifidus is more or less separable into fasciculi, which arise by mingled tendon and muscle fibers from the transverse processes of the vertebrae and pass forward over one vertebra to insert on the spinous process of the next. M. multifidus cervicis is well developed, consisting of three bundles of longitudinal fibers extending between the articular processes and the spines of the cervical vertebrae. M. rectus capitis dorsalis major (fig. 92) is a rather thin triangular muscle arising from the an- 170 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 terior two-thirds of the crest of the spine of the axis, and inserting into the occipital bone below the lambdoidal crest. The muscles from either side diverge as they leave the axis, so that a tri- angular cavity, bounded ventrally by the atlas and filled with fat, remains between their medial borders. M. rectus capitis dorsalis medius (fig. 92) is apparently represented by a few fibers, super- ficial to the medial fibers of the rectus minor and with a less oblique fiber direction, that arise from the anterior tip of the spine of the atlas and follow the border of the triangular cavity described above, to insert with the rest of the rectus on the skull. M. rectus capitis dorsalis minor (fig. 92) lies partly beneath and partly laterad of the medius. It is a large muscle with the usual origin from the anterior border of the dorsal arch of the axis, and inserts into the occipital bone beneath the major and medius. M. rectus capitis lateralis (fig. 92) is a rela- tively small muscle lying along the ventral border of the obliquus capitis anterior. Origin is from the ventral surface of the tip of the wing of the atlas, deep to the origin of the rectus capitis ven- tralis. The muscle expands somewhat toward its insertion, which is made into the posterior surface of the mastoid process near its outer edge. M. obliquus capitis anterior (fig. 92) is also relatively small. It is triangular in outline, arising from the tip of the wing of the atlas and insert- ing into the back of the skull just above the mastoid process. The dorsal edge of the muscle is overlain by the second head of the longissimus capitis. M. obliquus capitis posterior (fig. 92) greatly exceeds the anterior in size. Origin is from the entire spinous process of the atlas. The fiber direc- tion is nearly vertical. Insertion is into the wing of the atlas. 5. Muscles of the Tail. Figure 93. M. extensor caudae medialis is the posterior continuation of the multifidus, and is in contact with its mate along the dorsal midline. Origin is from the spinous processes of the last two lumbar vertebrae and from the spine of the sacnun. Inser- tion is into the prezygapophyses (on the anterior vertebrae) and dorsal surfaces (posterior vertebrae) of the caudals from the second on, by tendons that unite with the tendons of the extensor caudae lat- eralis. M. extensor caudae lateralis arises from the deep surface of the deep lumbar fascia, from the fused transverse processes of the sacral vertebrae, and from the transverse processes (or bodies, where these are absent) of the caudal vertebrae. Long tendons extend posteromesad over three vertebrae, uniting with the tendons of the extensor caudae medialis. M. abductor caudae externus arises from the dorsal surface of the fused transverse processes of the sacrum, from the fascia surrounding the base of the tail, and from the transverse processes of the first four caudals; there is no attachment to the ilium. Insertion is into the transverse processes (or the sides) of the three following vertebrae. M. abductor caudae internus is a relatively small fusiform muscle lying ventrad of the exter- nal abductor. Origin is by a rather wide, flat tendon that splits off from the tendon of the ilio- caudalis, thus coming from the medial surface of the ilium. Insertion is into the transverse proc- esses of the first six caudals, in common with the insertions of the external abductor. M. iliocaudalis is a thin triangular sheet. Ori- gin, by means of a wide tendinous sheet externally and fleshy fibers internally, is from the medial sur- face of the iliimi caudad of the sacro-iliac articula- tion. A long terminal tendon from the fusiform part of the muscle joins a tendon of the medial division of the flexor caudae longus, to insert into the ventral side of the sixth caudal. The remain- der of the muscle inserts fleshily into the trans- verse processes of the posterior sacral and first two caudal vertebrae. M. pubocaudalis is a very wide, thin sheet ly- ing immediately external to the levator ani. The dorsal fibers arise from the tendon of the iliocau- dalis, the ventral fibers from the dorsal (inner) surface of the symphysis pelvis. Insertion is into the ventral surfaces of the fourth and fifth caudals. M. flexor caudae longus is composed of two sets of fasciculi, which are separated proximally by the iliolumbalis. The lateral division consists of successive fasciculi arising from the posterior end of the sacrum and from the transverse processes (or sides) of the caudal vertebrae. The strong ter- minal tendons pass over three vertebrae before inserting into the transverse process (or side) of the fourth succeeding vertebra. The medial divi- sion arises just mesad of the lateral division. It extends from the anterior end of the sacrum to the third caudal vertebra, and its ventral edge is partly united to the adjacent edge of the iliocau- dalis. It is composed of three successive fasciculi, each of which terminates in a tendon. The tendon of the most anterior fasciculus joins the much stouter tendon of the middle fasciculus; together they insert with the pubocaudalis into the ventral surface of the fifth caudal. The tendon of the most posterior fasciculus joins a tendon of the long flexor, and inserts into the ventral side of the sixth caudal. u ■4-> B > a 3 O" o ■§ o a o •s I ■s o s 171 172 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. flexor caudae brevis consists of short fas- ciculi lying along the ventral midline from the fifth caudal on. Origin is from the ventral surface of the vertebra, and the fibers pass over one vertebra to insert into the next. 6. Muscles of the Perineum M. levator ani is a thin triangular sheet of muscle lying deep to the coccygeus, and over the lateral surfaces of the rectum and urethra. Its fiber direction is at right angles to that of the coc- cygeus. Origin is chiefly by means of a thin apo- neurosis from the medial surface of the ascending ramus of the pubis; some of the posterior fibers are continued from the retractor penis, and some are blended with the sphincter ani externus. Insertion is into the centra of the anterior caudal vertebrae. M. sphincter ani externus is a narrow ring of muscle fibers sun'ounding the anus. The two halves of the muscle meet below the anus and immedi- ately behind the bulbus urethrae; some of the fibers are continued into the suspensory ligament of the penis, which attaches to the posterior end of the symphysis; others attach to the bulbus urethrae and ischiocavernosus. M. ischiocavernosus is a very short muscle arising from the posterior border of the ischium, 25 mm. above the symphysis. It is closely applied to the posterior wall of the corpus cavernosum penis, and terminates by spreading out over this structure. M. bulbocavernosus is a thin layer of diag- onal muscle fibers surrounding the bulbus iirethrae. The two muscles arise from a median raphe on the ventral side, and insert into the posterior part of the root of the penis. M. sphincter urethrae membranaceae is a delicate layer of transverse muscle fibers surround- ing the urethra proximad of the bulb. It encases the urethra for a distance of 30 mm. M, retractor penis is a pale muscle arising as a continuation of fibers from the levator ani. It meets its mate from the opposite side just below the rectum, and the two muscles run side by side to the base of the glans penis, where they insert. A few fibers split off and insert into the side of the radix penis. M. caudorectalis is a prominent unpaired mus- cle lying along the midline in the anal region. It is distinctly lighter in color than the surrounding musculature. Origin is from the dorsal side of the rectum in the midline. The fibers pass backward and upward as a fusiform mass, to insert on the ventral surface of the sixth caudal vertebra. III. MUSCLES OF THE FORE LEG A. Muscles of the Shoulder Girdle M, supraspinatus (figs. 88, 95, 96, 133) is cov- ered externally by the usual heavy tendon-like fascia, which cannot be detached without cutting into the muscle substance. This tendinous fascia is continued diagonally downward to insert on the acromion process, immediately behind the origin of the acromiodeltoideus; the fascia over the distal end of the muscle is normal. The muscle occupies the whole of the supraspinous fossa, overlapping the cephalic border. It is powerfully developed, having a maximum thickness of 50 mm. Insertion is by fleshy fibers into the greater tuberosity of the humerus. Action: Extends the arm on the scapula. M. infraspinatus (fig. 95) arises from the entire infraspinatus fossa. It is covered with a tendinous aponeurosis down to the origin of the spinodel- toideus. The muscle is divisible into two parts, the one nearest the glenoid border of the scapula being slightly the smaller. The insertion tendons of the two parts are more or less distinct, but are fused where they are in contact. Insertion is into the prominent infraspinatus fossa on the greater tuberosity of the humerus. Action: Chief lateral rotator of the arm. Its tendon acts as a lateral collateral ligament of the shoulder joint. M. acromiodeltoideus (figs. 88, 95, 134) is powerfully developed, having a thickness of 23 mm. at its posterior edge. It is covered with tendinous fascia superficially. The muscle arises, partly flesh- ily and partly tendinously, from the whole tip of the acromion. It is bipennate, to two halves of approximately equal width. Insertion is by two heads, which correspond to the halves of the bi- pennate muscle. The anterior half inserts on the shaft of the humerus immediately above the in- sertion of the cephalohumeral, anterior to the del- toid ridge. The posterior part inserts partly on the lateral head of the triceps, posteriorly forming a strong raphe with the spinodeltoid. Action: Chief abductor of the arm. M. spinodeltoideus (fig. 88) arises almost wholly from the fascia of the infraspinatus; only its anterior tip reaches the scapular spine. Most of its fibers meet the acromiodeltoideus in a tendi- nous raphe, although a few insert on the triceps lateralis. Action: Flexes the arm. M. teres minor (fig. 95) is a small muscle, closely applied to the inferior border of the infra- I DAVIS: THE GIANT PANDA 173 spinatus, from which it is inseparable at its origin; it is not attached to the long head of the triceps. It arises by heavy aponeurotic fibers that are firmly attached to the underlying infraspinatus on the deep surface, from a small area on the axillary M. subscapularis (figs. 96, 133) is composed of three main divisions. The two anterior subdivi- sions are composed of numerous bipennate units, whereas the posterior one is made up of units with parallel fibers. Insertion is into the proximal end Caput humeri M. eoracobrachialis brevis M. eoracobrachialis longus M. biceps (caput longus) Epicondylut med M. biceps (caput brevis) Fig. 94. Right arm of bear (Ursus amerieanus) to show short head of biceps. Medial view. border of the scapula just proximad of the middle. Insertion is made by a short stout tendon into the head of the humerus, immediately distad of the insertion of the infraspinatus. Action: Flexes the arm and rotates it laterally. M. teres major (figs. 95, 96) is powerfully de- veloped. It arises from the usual fossa at the distal end of the glenoid border of the scapula, and from a raphe that it forms with the subscapularis on one side and the infraspinatus on the other. In- sertion is made, by means of a powerful flat tendon 30 mm. in width, common to it and the latissimus dorsi, on the roughened area on the medial surface of the shaft of the humerus, distad of the bicipital groove and immediately mesad of the pectoral ridge. An extensive bursa (Bursa m. teretis major of human anatomy) is inserted between the ten- don and the shaft of the humerus. Action: Assists the latissimus dorsi in flexing the arm, and the subscapularis in medial rotation of the arm. of the humerus, immediately below and behind the lesser tuberosity. The insertion tendon of the first (crania) unit is superficial to those of the other two units. Action: Chief medial rotator of the arm. The upper part of the muscle acts as an extensor of the arm. B. Muscles of the Upper Arm M. biceps brachii (figs. 96, 97, 133) is a fusi- form muscle that, in the position in which the arm was fixed, is rather sharply flexed at the site of the bicipital arch. The muscle displays a rather curi- ous structure. It arises by a single (glenoid) head, but in the proximal two-thirds of the muscle a narrow anterior group of fibers is more or less sep- arable from the main mass of the muscle. These fibers, which are particularly conspicuous because they lack the glistening tendinous covering of the rest of the muscle, arise from the origin tendon of the biceps as it passes through the bicipital groove and insert extensively into the anterior surface of 174 FIELDIANA: ZOOLOGY MEMOIRS, VOLUME 3 M. supraspinatus -Caput humeri -M. acromiodelt.(cut) ■M. infraspinatus -M. teres minor M. triceps longus M. triceps lateralis M. anconaeus M. ext. carpi uln. M. ext. dig. lat. Caput uln., m. flex dig. prof. M. ext. indicis proprius M. ext. dig. com. Lig. carpi dorsale icuti Fig. 95. Muscles of the right fore leg of Ailuropoda, lateral aspect. the main mass of the biceps, as far distad as the bicipital arch. There was no indication of a short head in two specimens dissected. The biceps arises from the bicipital tubercle at the glenoid border of the scapula, by a long, flat- tened tendon that runs through the bicipital groove, enclosed in the joint capsule, onto the ante- rior surface of the humerus. The tendon is contin- ued into an extensive area of tendinous aponeurosis on the external surface of the belly of the muscle, and a more limited area of similar tissue on the in- ternal surface. The most medial (superficial) fibers of the biceps terminate in a well-defined lacertus fibrosus, which is continued into the fascia over the pronator teres. The tendon of insertion begins mid- way on the deep surface of the muscle and continues distad as a distinct tendinous band on the deep surface of the muscle; this band does not form a longitudinal furrow as it does in the dog. The mus- cle fibers insert into it along its length at a very oblique angle, so that the biceps is a pennate mus- cle rather than a parallel-fibered one as in man. This tendinous band is continued into a short, very stout, flattened tendon, 12 mm. in width, that passes between the brachioradialis and pronator teres to insert into the prominent bicipital tuber- cle of the radius. Action: Flexes the forearm. The biceps is normally, but not invariably, two- headed in the bears, a degenerate short head usu- ally arising from the coracoid process with the brachioradialis (Windle and Parsons, 1897, p. 391). I have dissected the biceps in a young black bear, with the following results (fig. 94). The long head is similar to that of Ailuropoda except that the small group of accessory fibers coming from the origin tendon lies along the posterior border of the muscle, and the tendon of insertion does not begin far proximad on the deep surface of the muscle. The short head begins as a slender flattened ten- don arising from the fascia of the coracobrachialis just below the head of the humerus. At about the middle of the humerus the tendon begins to form a slender muscle belly that lies against the poste- rior surface of the long head. A few of the most superficial fibers insert via a lacertus fibrosus into the fascia over the pronator teres, but most of this belly inserts with the long head. The biceps was similar in an adult Tremarctos ornatus dissected by me. Windle and Parsons found a "very feebly DAVIS: THE GIANT PANDA 175 M. abd. poll, brevis Tendo m. ab