The Deseado Formation of Patagonia Frederic Brewster Looniis, Ph.D. Professor of Comparative Anatomy Amherst College EIGHTH AMHERST EXPEDITION 1911 PUBLISHED UNDER THE AUSPICES OF THE TRUSTEES OF AMHERST COLLEGE 1914 Copyright, 1014 BY FREDERIC B. LOOMIS THE RUMFORD PRESS CONCORD ' N H ' r CONTENTS CHAPTER PAGE I. Organization of the expedition history of the work done in the Deseado formation I II. Description of the Amherst locality age of the overlying beds age of underlying beds age of Deseado 6 III. Table of the animals study of the feeding habits character of the habitat the origin of the elements of the Deseado fauna 19 IV. Systematic arrangement the Litopterna, Eoproterotherium, Notodiaphorus, Deuterotherium, Protheosodon, Conioptero- therium, Tricoelodus, Proadianthus 28 V. Typotheria, Archaeohyrax, Plagiarthrus, Prohegetotherium, 53 Prosotherium, Propachyrucos, Phanophilus, Archaeophylus, Eutrachytherus, Argyrohyrax, Isoproedrium VI. Rhynchippidae, Toxodontia, Rhynchippus, Morphippus, Eu- geniops 86 VII. Leontinirdae, Leontinia, Ancylocoelus 108 VIII. Nesodontidae, Proadinotherium, Pronesodon, Coresodon, Inter- hippus, Nesohippus 122 IX. Isotemnidae, Trimerostephanus, Pleurocoelodon, Lophocoelus, Henricofilholia 129 X. Homalodontotheria, Asmodeus 134 XI. Astrapotheria, Parastrapotherium 142 XII. Pyrotheria, Pyrotherium IS 6 XIII. Rodents, Cephalomys, Scotamys, Litodontomys, Asteromys, Eosteiromys 185 XIV. Edentata, Proeutatus, Prozaedius, Stenotatus, Proeuphractus, Peltephilus, Palaeopeltis, Glyptatelus, Hapalops, Octodonto- therium, Orphodon 197 XV. Marsupialia, Pharsophorus, Notogale, Proborhyaena, Palaeo- thentes, Pilchenia, Callomenus, Pseuhalmarhippus, Parabderi- tes : 210 XVI. Birds, Physornis, Loxornis 225 PREFACE The results of the Amherst Patagonian Expedition were divided into two parts, the general features, to- gether with the narrative, were reported in a separate volume entitled, "Hunting Extinct Animals in the Pata- gonian Pampas," published in 1913. For this volume has been reserved the description of the material found and such conclusions as are directly derived from that ma- terial. The material on which this work is based has been prepared out and placed on exhibition at Amherst College. The material here described forms a unified body of data, which adds materially to our knowledge of the com- plete animals of the Tertiary period in Patagonia. There are beside this some small collections which offer some isolated new facts, but the working up of these has been reserved for the future for small articles, as the work may come to maturity. The field has only been touched and a vast amount of further work can be profitably done on the horizons im- mediately preceding and following the one described in this volume, after which an interesting study can be made on the evolution of a fauna which developed in a consid- erable degree of isolation. F. B. LOOMIS. March 18, 1914. THE DESEADO FORMATION OF PATAGONIA CHAPTER I INTRODUCTION THE material described and the conclusions drawn in the following pages are the results of the Amherst Ex- pedition to Patagonia in 1911; an expedition organized and sent out by the Class of '96 as a part of their fifteenth reunion. The party consisted of Frederic B. Loomis '96, Phillip L. Turner 'n, Waldo Shumway '12, and William Stein of St. Joe, Wyoming, and left Amherst July I, 1911, returning the first of February the ensuing year, having spent its time collecting in the early Tertiary beds of Patagonia, as exposed in the Territories of Chubut and Santa Cruz, the aim being to secure from the earlier periods a fuller knowledge of the vertebrate animals, such as the Princeton Expeditions obtained for the Patagonian and Santa Cruz formations. The narrative of the expedition has been told in " Hun ting Extinct Animals in the Pata- gonian Pampas." Material was found in various beds, from the Creta- ceous up to the Lower Miocene; but the major part of the fossils, and most of the facts new to science came from the work in the Deseado Formation. The collections from the horizon were so complete and interesting that this report of the expedition has assumed the form of a monograph of the Deseado Formation, otherwise known as the Pyrotherium beds. The first work in this formation was done by Carlos Ameghino who at various times between 1889 and 1894 collected for his brother, Florentine Ameghino, the latter studying and describing the collections of Carlos, whose 2 I I II-; DKSEADO FORMATION OF PATAGONIA trips covered the country from Chuhut down to the Straits of Magellan, and the various formations from the Lower Cretaceous to (he Pampean or Pleistocene. Carlos Amcghino and his brother, Florentine), for years explored in Patagonia, going summer after summer at their own expense, and in the meantime maintaining a small book and stationery store in La Plata, the profits of which gave the two brothers a living and furnished the funds for the continual expeditions. In the back of the store was the workshop from which came the continuous stream of knowledge in regard to these strange faunas. One of the best pieces of work done by the brothers was the collect- ing and describing of the fauna of the Pyrotherium beds the bulk of which is contained in two papers entitled, Premiere Contribution a la Connaissance de la Fauna mammalogique des Couches a Pyrotherium, and Mammi- feres Cretaces de 1'Argcntine, Deuxieme Contribution, etc., both published in the Boletin del Institute Geografico Ar- gentino, tomes 15 and 18 respectively. These two papers give names to most of the forms which we found, but the genera and species are based on very fragmentary and in- complete material. It has been a pleasure to find the accu- racy with which these descriptions were made; and our part has been chiefly to supplement and increase the knowledge of the various forms, and to determine from the more complete material the relationships of these strange forms. In some cases we have been able to assemble all the parts of the animals, and in the others to add more or less to the completion of the knowledge of the forms. There is one perculiarity of Ameghino's descriptions, namely the ab- sence of data as to the localities where the forms were found. About 1900 Tournier, in the interests of the Paris Mu- seum, made a series of expeditions (5) to Patagonia, on some of which he found a Pyrotherium, or as he has termed it Deseado, locality just south of the Deseado River, LOCALITIES 3 from which he gathered a considerable collection which has been described by Albert Gaudry in various papers mostly in the Annales de Paleontologie. These two collections and their collaborations represent all the work thus far done on the Deseado beds and fauna. Our collection is the first one of any considerable size to be brought to North America, and it seems to be by far the most complete, the various animals being represented by more complete skeletons than in any of the previous col- lections. The beds were first designated as the Pyrotherium beds, and are always so referred to by F. Ameghino. Tournier and Gaudry, feeling the prejudice which is fairly general among Palaeontologists against names based on any con- tained animal (which may or may not be present at other localities, which may extend through more than one period, and whose name may be changed as a result of further knowledge) used the term Deseado formation, as his collections came from the neighborhood of this river. This is a geographical name and avoids the chance for confusion; so I have adopted it throughout this paper, it being understood as an equivalent of the term Pyrotherium beds. Ameghino never gave the exact, or anywhere near the exact, localities from which his Deseado specimens came. It was not until 1906, when his Formations Sedimentaires* appeared, that any localities were designated, and there on a sketch map he indicates as Deseado exposures, about a dozen points, scattered between the upper part of the Chubut River to some 25 miles south of the Deseado River. These are included in an oval area some 500 miles long by 150 miles wide. Ameghino also suggests on this occasion that the Deseado formation originally extended over at least the whole of this area. As will be seen in the next chapter, I believe that the deposits of this age * Anal. Mus. Nac. Buenos Aires, ser. 3, t. 8, p. 99. 4 THE DESEADO FORMATION OF PATAGONIA and character have always been local and isolated. We sought for several of these localities and failed to locate them, especially those near Mazaredo, and the northern one on the Gulf of St. George. The point where we did find our material I believe was one of Ameghino's localities, though the settlers of that region had never heard of any- one hunting for fossils there; but the settlement had been practically all within the previous six years, which was much later than the time when Carlos Ameghino worked in the region. Beside the foregoing, an exposure of this age is reported by A. A. Romero, just above the fork of the two branches of the Rio Negro, w^hich is some 500 miles north of the first group of localities mentioned. Ameghino also refers to another locality in the Province of Misiones which would be 1,500 miles north of the typical localities. The collections made by Tournier for Gaudry came chiefly from an exposure south of the Deseado River, some 15 miles above the mouth of the river.* Our collection came from the Chico Branch of the Chubut River, about three miles east of the river, and almost due west of Puerto Visser. As mentioned above on account of the close coincidence of the various species and because Ameghino indicates a locality in the neigh- borhood, I think that our locality is the same as one of his, I should judge it the one from which he obtained a considerable part of his types. This is of importance; for, if in Ameghino's type locality, the determination of the species, as the same as those of Ameghino's, is much more certain. In the accompanying map I have indicated the locali- ties given by Ameghino, those of Tournier, and our own. *Bul. Soc. Geol. France, ser. 4, t. 3, 1903, p. 468. Fig. i. Map of Patagonia showing localities of Deseado beds. CHAPTER II AGE OF THE DESEADO FORMATION THE locality worked by the Amherst party is situated about three miles east of the Chico River, just across the line of the homestead of D. J. Venter as plotted on the Piano de la Gobernation del Chubut, 1910, by A. Lefrancois. This would be 45 10' S., and 67 32' W. (or as on the map 9 15' W. of the meridian of Buenos Aires). The exposure is on all sides of an elongated hill about a sixth of a mile long, averaging 200 feet wide, and constricted in the middle to a narrow neck. Figure 2 shows a section of the hill, made along the north side, and indicates the varied charac- ter of the stratified deposits. The material varies from brown sandy clay shales, to yellow sandy clay with concretions, and is capped with a varying layer of greenish sand, which, in some places, is coarse and irregular, in others fine and uniform, and in still other places is mixed with considerable quantities of volcanic ash. In it are many mud balls and also bits of bone which have been worn round, others but slightly worn, and finally bones and skeletons which apparently have been buried where they fell. This green sand is mostly covered with a layer of two feet of hard sandstone of the same composition as the rest of the bed, but cemented into a dense layer. Above the green sand is a layer of fine grey sand, prettily crossbedded, and of varying thick- ness, but without fossils. Remains of vertebrate animals occurred in the brown clay, the yellow clay and the green sand, and in all the cases fossils were of unusual abundance so that in this limited locality we collected over 300 speci- mens. Above the Deseado (layers 2 to 5) lies the Patagonian in its typical development, filled with Ostrea ingens, 8 THE DESEADO FORMATION OF PATAGONIA Turritellas, Brachiopoda, sharks' teeth, etc. It is separated from the Deseado by a marked unconformity, one of the finest examples of unconformity I have ever seen. Evidently the upper surface of the Deseado was fairly new at the time of the transgression, or k is much disturbed by the transgression, the upper layers in places being broken up into sort of blocks and the crevices filled with Patagonian sands with the contained shells; just as I saw the beds on the seashore being disturbed by the waves of today. Then too in the basal foot of the Patagonian I found material which without question came from the underly- inp Deseado beds, various fragments of mammal bones bored by seashells, and with the Patagonian barnacles on them, but these were never more than a few inches up in the Patagonian. The contact was not horizontal, but in the middle of the hill dipped down so that it came there onto the yellow beds of clays, and it was at this point only where we found bones had been washed out by the Pata- gonian sea. In the section the Deseado consists of layers 2 to 5, the white sandy clay below belonging to the St. George series and being Cretaceous. The contact below was also an unconformity, clearly marked for the white sandy clays were all horizontally bedded, while the Deseado is crossbedded in every direction, and has a distinct color. These white sandy clays of the St. George series are simi- lar to the same beds as shown in sections A and B (figures 3 and 4), and extend in all directions for miles. Going down toward the Chico River one passes into the green shales that make up the upper part of the Salamanca and had similar invertebrate fossils. About ten miles to the north was another bed of fossil trees similar to the one to be described on the Puerto Visser side of the pampa. The character of the material making up the Deseado deposit, its variations in size and material, the presence of worn pebbles and bits of bone, show these layers to be a A RIVER DEPOSIT 9 water deposit. The absence of any marine fossil in a bed otherwise rich in fossils indicated that it was a fresh water formation. The crossbedding, the irregularity of the de- posits and the mud balls, prove that it was the work of a river. As there are no aquatic forms in the fauna I further conclude that it was the deposit of a temporary or inter- mittent stream, such as occur in arid and semiarid coun- tries. The layer could hardly be interpreted as a part of a flood plain; for it is very limited in extent, there being bluffs on three sides of our exposure, but in them no trace of the Deseado was found, nor was I able to pick up the formation again across the Chico River. Then the bed- ding is very irregular, much more so than is typical of flood plain deposits. The conclusion I reach then is that this Deseado pocket represents the bottom of an ancient stream, which flowed over a land surface made up of the white sandy clays of the St. George age. The age then of the Deseado beds must be older than the Patagonian, and younger than the white sandy clays of the St. George. As to the age of the Patagonian two very divergent positions have been taken, which may be best indicated by the diagram on page 10. Without going into the history of the various positions which different authors have taken, and which will be found given in detail in Wilckens' paper, or in less detail in Ortmann's, we will consider the positions of the most recent students of the question. Ameghino postulates a marine and a continental series of deposits being laid down more or less simultaneously. In the marine series below the Deseado, which is grouped as Guarantic, he places the Luisa, the Roca and the Salamanca, followed by a hiatus, then the Sehuen, which in turn is followed by an- other hiatus and the end of the Cretaceous is reached. The Patagonian is his Eocene. Parallel to the marine series is the terrestrial, where the Casamayor (= Notos- 10 THE DESEADO FORMATION OF PATAr.ONIA * Formations Sedementaires, p. 498, Anal. Mus. Nac. Buenos Aires, ser. 3, t. 8. f Neues Jarhbuch fur Mineralogie. bd. 21, p. 193. | Princeton Expeditions Reports, vol. 4, p. 303. tylops) is contemporaneous with the Salamanca, the Deseado with the Sehuen, and the Colpodon, the Noto- hippus and Astrapothericulus with the Patagonian, thus making the Deseado of Cretaceous age. After a very detailed study of a large series of Patagonian fossils, Ort- mann concludes that the Patagonian is of Lower Miocene age. This is the most detailed study which has been made. Wilckens coincides with this view, though feeling that the Patagonian may have extended down a trifle into the last of the Oligocene. This latter author finds a long gap be- tween the Upper Cretaceous and the Patagonian, a period when Patagonia was above water. It was during this AMEGHINO'S SECTION II interval that the Casamayor, the Deseado and possibly other beds were deposited on the continent. I have gone over Ortmann's argument, and studied a large collection of Patagonian fossils, both vertebrate and invertebrate, of my own ; and while there are some places where we would like further data, I can come to no other conclusion but that these Patagonian beds are Lower Miocene, the exact relationship with beds in North America and Europe, being as yet not definitely settled, nor will this be possible until a study of the migrations of the elements of the Patagonian fauna has been made. As to the beds underlying the Patagonian, I am sure that a considerable study of the marine series is still requisite to determine the relationships of the beds in different parts of Argentine, and their relative positions as compared with beds in other countries. Ameghino appended to his paper on the Formations Sedementaires a section of the strata exposed on the coast of Patagonia from Rio Negro to Cape Virgenes, on which from above Punta Atlas south to below Pico Salamanca, the Casamayor ( = Notostylops) beds fill the interval from the Salamanca formation up to the Patagonian. On the strength of this map I followed these beds the whole distance looking for vertebrate fossils of Casamayor age. Nowhere did we find a Casamayor fossil. Instead at several points we did find marine fossils. I can not but feel that these beds are plotted as Casamayor, because of their resemblence in color and general texture to the beefs carrying the Notostylops fauna at Casamayor. Of several sections of these beds I pick out two as typical, and also because they are near the locality which we worked for the Deseado fauna. On the map they are indicated as A and B. The former passes through a bed of green sands which is, I think, the locality indicated as his north- ern locality for the Pyrotherium fauna. From Punta Atlas to Pico Salamanca, Ameghino plots at or just below sea level a bed known as the Salamanca, A Fig. 3- Section at A on map page 5, showing strata from sea level up to the Patagonian. Ib IfJ ? ' ^^^/^^.^ : % ^^^4*^^' '//'l'///^'''//.'/'''-'-'' ////// /, / -\ y . '//// w hi \ S a hd u C CcMj / alf J ', '''*''' Viv.v' Ve llovv, T?ed, G hiy:' 1 ;';?: ;.v SiaUS^^ fc. m * ~ ~ ^ r=r !z,^ vgil Ohales^ L-imes+0n fS ^2 > > ' ' Fig. 4. Section B on map, page 5, showing strata from sea level up to the Patagonian. 14 THE DESEADO FORMATION OF PATAGONIA being typically developed opposite Pico Salamanca. In this in the neighborhood of Pico Salamanca we found the fauna typical of this horizon. Ostrea rostigera v. Ih. Ostrea riongrensis v. Ih. Ostrea ameghinoi v. Ih. Oilamys salamanca v. Ih. Rostellaria striatissima v. Ih. Rostellaria sp. Cytherea calcedonica H. Discinia sp. Diplodon sp. This Salamanca formation is considered by \Yilckens as the equivalent of the Roca as exposed on the Rio Negro, and to the Luisa as exposed on the Rio Coyle. All agree that the Salamanca is Upper Cretaceous and a period when Patagonia was covered by the ocea'n. In section B we found the above fauna in layer I which is just above sea level here. In layer 2 we found casts of delicate marine shells (30 to 40 in number), representing four or five species and as yet undescribed. They seem to represent a deeper water facies of the Salamanca. In fact all the shales represented by layers I to 5 evidently belong to the Salamanca. Layer 5 was distinguished by having in it at a point some 200 yards north of the section line a quantity of turtle shell fragments. Layer 7, consisting of coarser sandstones, was at the point of the section, simply filled with a vast quantity of fossil wood, most of it agatized, though some was carbon- ized, and representing some eight species, mostly pines and palms, the latter much scarcer. The tree trunks, hundreds in number, lay scattered in all directions; but all were lying horizontal, and there was no indication of stumps in place; so I consider that the wood was driftwood. It is common in the series of beds of this general horizon along the Gulf of St. George. In the other layers up to the Patagonian THE SALAMANCA FORMATION 15 we found no fossils. The contact with the Patagonian was unconformahle, in some places being 50 feet higher than in others near by. In section A the typical Salamanca is below sea level, and the lower parts of the section are made up of the white sandy clay shales, so typical all along the Gulf of St. George. In the midst of these clays at the level indicated as 2 oc- curred a layer of concretions. On breaking these we found two specimens of Nautilus valencienni H., clear evidence that they were of marine origin. Layer 5 was filled with hundreds of the very characteristic oyster, described as Ostrea (Gryphaea) pyrotheriorum. Though in earlier papers suggesting that O. pyrotheriorum represented a horizon of marine sediments corresponding in age to the Deseado ( = Pyrotherium) formation, in his Formations Sedimentaires, Ameghino places this fossil in the Sala- manca fauna, though it here occurs at least 275 feet above the typical Salamanca fauna. I believe the layer should be distinguished. It is later than the typical Salamanca, though belonging to the same transgression of the sea over Patagonia. In layer 7 we found still another marine fauna consisting of Ostrea guarantica H. Venericardia sp. Corbula sp. Aporrhais. Patomides. Oxyrhinca. Milobates. Fragments of the limbs of a crab in abundance. This seems to be the same fauna as that described by Ameghino as the Sehuen developed on the RioSeheun. In layer 8 we found large quantities of gypsum, occur- ring mostly in balls of radiate structure. Layer n was a coarse green sand, and in it we found some fragments of some sort of a bone. I think this layer is what Ameghino 16 THE DESEADO FORMATION OF PATAGONIA designated as a Deseado exposure; and it has the same general appearance and color which is found in the green sands of the Deseado pocket on the Rio Chico. However it is conformable interbedded with the underlying and over- lying marine beds and I consider it a part of the marine series. Above it come more white sandy clays that are characteristic of the most of the section. Wilckens takes all of this series, from the base of the Salamanca, up to the unconformity below the Patagonian, and makes of it his St. George Period, a transgression epoch, lasting to the end of the Upper Cretaceous. I be- lieve it is all marine, and is all a part of the Upper Creta- ceous transgression of the sea over Patagonia. However the Salamanca is a clear cut deposit and I feel it should be retained as a distinct horizon. The overlying light colored (white, grey, brown, yellow, or green) sandy clay shales represent a deeper water and later facies, which is characteristically developed on the Gulf of St. George, and may well be distinguished as the St. George epoch or series, but I should use the term only in this more limited way. It is the same series which Ameghino has plotted as the Notostylops beds on his section of the coast of Patago- nia. This last it certainly is not. The unconformity between these white (or light) sandy clays and the Patagonian represents a regression period, during which Patagonia was not only above water, but extended an unknown distance further to the East. It was during this interval of time between the Upper Cretaceous and the Lower Miocene (Patagonian) that the limited and local land deposits known as Casamayor ( = Notostylops), the Astraponotus, and the Deseado ( = Pyrotherium) and probably other beds were laid down. In each case the age must be determined for the individual bed by its contents mostly; for as far as I know none of them overlap anywhere. DINOSAUR QUESTION 17 In regard to the discussion as to whether Dinosaurs were contemporaneous in South America with the fauna of the Deseado, I can only say, we found no trace of a din- osaur or any other Cretaceous animal in the Deseado beds which we worked. As the Cretaceous beds lie as high as the Deseado and are also practically horizontally striated, dinosaur remains might be found on the same level. I think the assigning of any such material to these beds was due to failing to recognize the unconformity under the Deseado beds. As to the Notostylops fauna and dinosaurs being contemporaneous, I only worked the Notostylops beds at Mazaredo, but there I found nothing to indicate the contemporaneousness of these two groups. As I have shown above, Ameghino's idea of the extent of the Noto- stylops or Casamayor beds was mostly at fault, and very much of that which he has designated as of Notostylops age is Upper Cretaceous. It is in these Upper Cretaceous beds that dinosaurs do occur and this seems to me to be the basis of the confusion. This Upper Cretaceous series is a field where consider- able work may profitably be done, in straightening out the relationships of the various layers to each other, their extent, and their relationship to the Salamanca and other Upper Cretaceous formations in other parts of Argentine. As to the age of the Deseado deposit which we worked. It is under the Patagonian, and therefore must be as old as the Oligocene. On the other hand it must be as young as the Eocene, lying as it does above the Upper Cretaceous. Of the three general faunas described it is clearly more ad- vanced than either the Casamayor, or the Astraponotus; so should be put as the youngest of these three. The Colpodon, the Astrapothericulus and the Notohippus, faunas are said to be inters tratified with the Patagonian and therefore of the same age. The amount of advance- ment from the Casamayor to the Deseado is considerable and the relationships of the Deseado are fairly close with 1 8 THE DESFADO FORMATION OF PATAGONIA the various genera of the Santa Cruz; so that I should put the Deseado as far up as possible toward the Santa Cruz. The Santa Cruz is above the Patagonian, and I think that the Deseado should be put just before the Patagonian; that is in the Oligocene, but just what part of the Oligo- cene can only be determined when the other faunas have been further studied. CHAPTER III THE DESEADO FAUNA THE exposure of the Deseado, which the Amherst party worked, yielded 293 specimens, each presumably repre- senting an individual. (There were besides these a few that were indeterminate and are not therefore included.) The consideration of the fauna as a whole suggests certain ideas as to the country in which the animals lived, and also certain comparisons with the fauna of the preceding and later faunas. The first striking feature is the presence of so many excessively large animals, as Asmodeus, Parastrapotherium, and Pyrotherium, in each case forms larger than a rhinoce- ros. Further than that they are in each case the largest members of their family, even larger than the representa- tives in the later Santa Cruz. This would indicate a period in which living conditions were at a high grade, sug- gesting both abundance of food and a moderate climate. The following table will give a good idea as to the range of species, and their relative abundance in the fauna, also a suggestion as to the class of food they used ; and from that an idea as to what sort of country they occupied : PER NUM- SPECIES FOOD COUNTRY CENT BER 3 Hegetotherium shumwayi 7 Prosotherium garzoni 17 Prosotherium triangulidens I Eutrachytherus grandis 4 Eutrachytherus spegazzinius i^JSpJJ } Plains 1 Isoproedrium solitarium 2 Phanophilus dorsatus 4 Argyrohyrax proavus I Plagiarthrus clivus 14% 40 TYPOTHERIA 20 THE DESEADO FORMATION OF PATAGONIA 2 3 2 I 5 8% 23 i i 5 2 3% 10 3% ii 100% 293 Proeutatus lageniformis Prozaedius planus Prozaedius depressus Proeuphractus setiger Peltephilus unclulatus Palaeopeltis inornatus Indeterminate EDENTATA Plichenia lucina Epanorthus chubutensis Callimenus praecursor Pharsophorus tenax Pharsophorus mitis MARSUPILAIA BIRDS Insects and leaves Open country Insects and flesh Open country THE EDENTATES 21 In our collection, all from one point, there are thirty- nine different species. Beside these Ameghino has de- scribed a considerable number of species, some of which in time will probably turn up at our locality; but others and I think the majority will be found to be representative of other localities which he worked. It is to be expected that a difference of locality will make a little difference in the fauna. Further I expect that no two localities represent exactly the same period of time, though they may do so approximately; but some of these local deposits must have been begun earlier, and others probably lasted to a later period. Thirty-nine species of mammals and land birds is a fairly varied fauna for one spot; and the time element involved in laying down the 50 feet which separated the bottom from the top of the Deseado deposit is not probably very long; for the material of which the deposit is com- posed is of a character which would have been laid down fairly rapidly. Of this fauna only 8 per cent belongs to the edentates; and if any element were disproportionately represented it would be this one, for the armadilloes have in addition to the skeleton the hundreds of tiny plates of the carapace, and several of the forms are represented by one or two plates only. When compared with the condition in the Santa Cruz this 8 per cent is strikingly small, for in that later bed, fully 50 per cent of the finds represent edentates. Are the Edentata just originating? Or, was the country less favorable to their habitation? The edentates which we did find are only slightly less advanced in their develop- ment than those of the Santa Cruz. Also, though in- frequent, all of the families of the Santa Cruz are repre- sented. It would seem therefore that the origin of the edentates was much earlier than the Deseado; and this relative paucity of edentates is also characteristic of the Casamayor and Astraponotus beds; but they are there, and in considerable variety, though small numbers. It 22 THE DESEADO FORMATION OF PATAGONIA would seem then that the country for some reason was less adapted to edentates, and that in some other part of South America they were flourishing and evolving. In the Deseado the rodents appear for the first time in South America. They are all Hystricomorpha and in a relatively primitive stage of development, but they are typically developed already. Did they migrate in from some other locality, or were they evolved on the spot? Ameghino believed that they were developed from some such form as Promysops or Propolymastodon of the Casa- mayor, and that these forms were ancestral to rodents all over the world. If my interpretation of the age of these beds is anywhere near correct, this last at least is impos- sible, for in North America and Europe typical rodents are present in the Eocene. Then as to even the hystri- comorphs being developed in Patagonia, I am very skep- tical, for the material offered in evidence of this is very insufficient, especially in the region of the incisors; and may be interpreted in other more probable ways. I am confident that either just before the beginning of the Des- eado, or at the beginning, the rodents of these beds mi- grated, either from some other continent, or at least from some other section of South America into this Patagonian region. Some idea of the type of country and the climate of the Deseado period in Patagonia may be obtained by ana- lyzing the fauna as to the character of its teeth as indicative of the food; and by studying the feet as indicative of the ground on which they were used. The Typotheria with their chisel-like front teeth, lack of canines, and their permanently growing grinders evi- dently ate a hard type of vegetation. Deep and permanently growing molars are characteristic of the eaters of grass, a form of vegetation which is especially hard on the grinding teeth, on account of the silica in the stems and leaves. This however would scarcely necessitate the development THE TOXODONTS 23 of permanently growing incisors. They are typical of gnawing animals, eaters of bark, twigs, and possibly also leaves, the wood and bark being also a hard type of vegeta- tion to grind. In the case of these forms I believe they were feeders on grass and bark. Their feet are developed either for running or hopping and would suggest hard ground for their habitat. The Litopterna are typically plains animals, paralleling in their development the horses. The cropping teeth and the grinding molars become progressively longer. The limbs are progressively elongated, the animals walking more and more on the tips of the toes. With this, the metapodials especially and the other limb bones to a less degree, are progressively lengthened. At the same time the side toes are progressively reduced. The teeth indi- cate grass eating; the limbs life on the plains. The Rhynchippidae, while not as advanced as the Litop- terna, show cropping front teeth, and the molars develop- ing in depth. The locomotion is semidigitigrade, the feet small, and the number of toes reduced to three. They too must be interpreted as grazing or grazing and browsing animals, living on hard ground. The Leontinidae are heavier forms, but with much the same features as Rynchippidae, though less specialized. On account of the broad upper molars and the less special- ization of the dentition, I should feel that these forms were browsers and lived among bushes, but the feet were three toed and semidigitigrade and they seem to have walked on hard ground. The Nesodontidae belong to the same type of adaptation as the foregoing family, but have the grinding teeth more complicated, indicative of a more advanced adaptation to hard vegetation. The feet were also adapted to hard ground. The Homalodontotheria, the Astrapotheria, and the Pyrotheria were all very large animals, known mostly by 24 THE DESEADO FORMATION OF PATAGONIA their dentition, which is adapted to browse. Whether they lived on soft or hard ground is not known, as the feet are not known in any case but the Homalodontotheridae, where they are five toed and adapted to soft ground. Such large animals were probably inhabitants of some river bank. The rodents do not contribute much in the determina- tion as to the type of the country, for they could have lived in the open or in the wooded country, but their relative abundance is rather typical of open country. The birds are all running birds, and indicative of the country having been an open one. Of our fauna 1 1 per cent were flesh or insect eating, and for the purpose of determining the type of country may best be omitted. The rodents could have been either forest or open country forms. Of the remaining 54 per cent, the typotheres, the litopternas, the Rhynchippidae, the Leon- tinidae, the nesodonts and the birds (46 per cent) were distinctly adapted to live on hard ground; the other 8 per cent being evidently suited to living near a river. All 54 per cent ate either grass or browse. The litopternas are grass eaters; the typotheres were specialized to eat grass or bark; nesodonts, Leontiniidae, and Rhynchippidae are grass and browse eaters. Even the Pyrotherium has a pair of gnawing tushes. The picture arising from these con- siderations is a bush covered prairie, a country not unlike the upland bush pampas of Patagonia today. There is not an aquatic form (fish or turtle) in the whole list, so it is evident that the stream which deposited these Deseado beds was not abundantly inhabited. To me it looks like so many of the streams in an arid country, dry through a considerable part of the year, and so uninhabited. In the whole list I see nothing to indicate forests or swamps. The arid bush covered plain alone seems to suit the re- quirements. As I see this fauna it is composed of several distinct elements, representing different invasions and an ele- VARIOUS INVASIONS 25 ment which arose in situ. The reasons for the affinities expressed in the different groups will be found in the intro- ductory paragraphs of the systematic discussion of each group. The Notungulata, including the Typotheria, the Toxo- dontia, the Litopterna, the Homalodontotheria, and the Astrapotheria are a group with apparently a common an- cestry. In Patagonia they have specialized into the various subdivisions as we find them in the Deseado. This group was in Patagonia as early or earlier than the Casamayor. Their relationships appear to me to be with the Hyracoidea which are generally credited with originating in Africa. The Pyrotheria are related to the early elephants which also arose in Africa, but it seems to me that this form came to Patagonia at least at a later period, making its first appearance in the upper part of the Astraponotus period. Ultimately the elephants and Hyracoidea had a common origin in Africa. The Rodentia are all hystricomorphs and appear in South America for the first time in the Deseado. They also occur in the Oligocene of Europe and the Fayum of north Africa. They never reached North America so must have come to South America by some southern route. The Edentata are an element of the Casamayor fauna and as there is no evidence of their originating anywhere else it would seem that they were indigenous to South America, where they later flourished and developed the greatest variety and profusion of numbers. The group of marsupials is an element the origin of which presents a most difficult problem. Some belong to the oppossum series which could well have been developed from some remnant of the Mesozoic marsupial fauna that had a world wide distribution; but the presence of dipro- todonts, which are characteristic of Australia, and of the Borhyaenidae which are closely related to the Thylacinidae of Australia, suggests a migration from that continent as 26 THK DESEADO FORMATION OF PATAGONIA late as Tertiary times; but to my mind this involves a connection which is most too difficult to postulate. There is no evidence that they came to South America in com- pany with other faunas, for they have not been found associated with any other fauna outside of Southern Pata- gonia. The explanation of the affinities of the Patagonian marsupials with the Australian marsupials is a problem which is not yet in position to be settled. The birds probably came from Africa with the invasion of the ancestors of the Notungulates. The idea of an invasion from Africa in Upper Cretaceous times, and possibly another at a later time is correlated with the other evidence of a land bridge between these two continents, as deduced by students of other groups. Eigenmann, working on the freshwater fishes,* Lydekker, studying the hystricomorphs,f Von Ihering, studying the freshwater mussels, J Ortmann, studying the freshwater crabs, not to mention several others studying mullocks, insects, plants, etc., have all postulated a land connection from Brazil to northern Africa during Cretaceous time to ex- plain the distribution of their various groups. The diver- gence is in the time when this land bridge sank, some be- lieving it to have lasted into Tertiary times, most feeling that it sank in Upper Cretaceous times. Another body of evidence is presented to show that a land bridge con- nected the West Indies with the Mediterranean regions.!! There was presumably but one such transatlantic connec- tion. Its position further to the south would seem to me to explain the distributional facts found in the West Indies, but the striking resemblances between the faunas of Africa * Princeton Expeditions to Patagonia, vol. 3, p. 310, 1905- n. t History of Mammals, p. 127, 1896. J Archhelenis and Archinotis, p. 125-145, 1907. Proc. Amer. Philos. Soc. Philadelphia, vol. 41, p. 350, 1902. || See Scharff, Distribution and Origin of Life in America, Ch. 11, 1912. SOUTH AMERICAN AFRICAN BRIDGE 27 and South America require a connection from the South Upper American Continent and Africa. It was along this land bridge which the ancestors of the Notungulata traveled, and when in South America, due to their isolation, developed all the peculiarities of the group. This must have been not later than the latter part of the Cretaceous. Either this bridge remained until into the early Ter- tiary; so the Pyrotheria and Hystricomorpha made their migration later, or these two groups did not reach the isolated Patagonian section until later than the first inva- sion. I am inclined to believe in the migration being at a later period. This bridge does not explain the presence of the edentates, for which there is every reason to believe that they developed in situ. The Marsupial invasion must have been from some other direction, or their presence in Africa has not yet been discovered. CHAPTER IV UNGULATA The systematic arrangement of the South American ungulates is of such a nature that scarcely two students of these forms have agreed. I feel that the Pyrotheridae are proboscideans as did Ameghino, but there my agreement ends. The other varied groups I believe have a common ancestry, their great divergencies being due to adaptations to the greatly varied characters of the country they occu- pied. In spite of the great variation they have certain features in common so that I agree with those who have developed the term Notungulata to include them all. From what source they originally came is not clear, but it seems to me that these notungulates have more in com- mon with what we know of the African fauna of the Fayum than with any other fauna; so that my feeling would be that these two faunas had a common ancestry at least, and possibly the South American ungulates are derived from the African. The lophiodont upper dentition, the bicres- centric lower molars with a "pillar" in the posterior crescent, the development of the tympanic bulla with the extension of the inflated cavity up into the squamosal bone, the development of the post-tympanic portion of the squa- mosum, and the general arrangement of the basi-cranial foramena indicate in my mind that these notungulates have all risen from the same stock, and that that stock had much in common with the hyracoids. I should therefore arrange the various groups as follows.* * The following references discuss in detail the arrangement of these forms. Ameghino, 1906, Formations Sedimentaires, Anal. Museo Nac. de Buenos Aires, ser. 3, t. 8, p. 287-498: Roth, Los Ungulados Sudamericanos, Anal. Mus. La Plata, t. 5, 1903, p. 1-36: Scott, Princeton Patagonian Expeditions, vol. 6, p. 287-299, 1912: Gregory, Bui. Amer. Mus. Nat. Hist., vol. 27, p. 273-285, 1910. SYSTEMATIC ARRANGEMENT 29 NOTUNGULATA Order I. Upper molars composed of an external longi- tudinal crest and two transverse crests, the posterior the less developed ; lower molars composed of two joining cres- cents with a "pillar" in the posterior crescent; structure of the feet and limbs varying. Suborder i. Litopterna: teeth brachydont to hypsodont; lower molars with the anterior and posterior cres- cents subequal; squamoso-periotic region not in- flated; limbs elongated; pes unguligrade; digits 3-3 or i-i. Suborder 2. Typotheria: teeth hypsodont lower molars with the anterior crescent shorter than the poste- rior; squamoso-periotic region inflated; limbs elongated in varying degrees; pes plantigrade to semi-plantigrade; digits 5-4 or 4-4. Suborder 3. Toxodontia: teeth brachydont to hypso- dont; lower molars with the anterior crescent shorter than the posterior; squamoso-periotic region in- flated; limbs short; pes semidigiti grade to digiti- grade; digits 3-3. Suborder 4. Homolodontotheria : teeth brachydont; lower molars similar to those of Toxodontia; limbs moderately elongate; pes semidigitigrade ; digits with large curved claws, 5-5. Suborder 5. Astrapotheria : teeth brachydont to mod- erately hypsodont; canines enlarged into tushes; molars similar to those of Toxodontia; limbs greatly elongated ; feet unknown. PROBOSCIDEA Order II. (see page 68) Suborder i. Pyrotheria: incisors developed into tushes; molars bilophodont; limbs short, especially the lower element; feet digitigrade. 30 THE DESEADO FORMATION OF PATAGONIA LITOPTERNA This order of South American ungulates is less abun- dantly represented in the Deseado formation than in the Santa Cruz, but most of the genera of this latter formation have representatives in the Deseado so that they seem to have diverged still earlier. By Scott the order is divided into two families, the Proterotheriidae and the Macrauchenidae, the less known Adiantidac being placed under the latter family until better known. 1 feel that I should prefer to retain the Adiantidac for the present, until they can be shown to be subordinate to another family, so that in this paper the three families are retained. The striking features of the two larger families may be best brought out by a compari- son of their chief features as follows. Proterotheriidae Macrauchenidae Upper inc. 2 and lower inc. 3 enlarged Incisors, canine, and premolar I and tush-like, growing from per- simple, compressed, subequal in sistent pulps. size, and rooted. Nasals normal Nasals shortened indicating a pro- boscis. \'c< k short. Neck long. Feet with median digit enlarged, lat- Feet with all three digits subequal in eral digits reduced. size. Proterotheriidae Ameghino In the Deseado, this family is scantily represented as compared with the rich fauna, both as to species and num- bers of individuals in the Santa On/, but of the four chief genera of the Santa Cruz, three have been found, though the remains are very fragmentary. They are the genera Eoprototherium, belonging to the Prototherium series, D ciiler other ium belonging to the Thaotherium series, and Notodiaphorus representing the Diadiaphoms series. EOPROTEROTHERIUM . The following table will give what is known in comparing the two series. Proterotherium PERIOD Santa Cruz UPPER MOLARS NASALS PES metaconule present normal tridactyl protoconule and protocone separate metaconule present protoconule and protocone separate metaconule present normal tridactyl metaconule present short tridactyl protoconule and protocone fused Deuterotherium Deseado metaconule lacking protoconule and protocone separate metaconule lacking protoconule and protocone separate tridactyl normal monodactyl Eoproterotherium Ameghino Eoproterotherium Amegh., 1904, Anal. Mus. Nac. B. A., ser. 3, t. 3, p. 441. The genus is founded on single teeth of the upper molar series, which, except for size, are very like those of Proterotherium. Limbs, etc., are unknown, so that this genus is simply a carrying back of the Proterotherium line into the Deseado. We found no teeth of this form, but one species has been described, E. inaequifacies, of which Ii /* | , c after Ameghino. reproduce Ameghino s figure com- pared with Proterotherium, which shows this species to have the metaconule better developed. Notodiaphorus gen. nov. The basis of this genus is particularly a hind limb found associated which is much less developed than the Santa Cruz genus Diadiaphorus to which it is most nearly related. These two genera are unique in having the ectal facet on 32 THE DESEADO FORMATION OF PATAGONIA the astragulus developed in two planes so that it appears as a deep notch. In the case of the new genus the toes are almost equal in size, giving us a stage in the develop- ment of this three-toed form which is much more primitive than the \vell-known Santa Cruz genus. Notodiaphorus crassus sp. nov. The specimen selected as type is number 3287 of the Amherst Collection, consisting of a complete pes, tarsus, lower end of the tibia, and the femur, from the Deseado on the Chico del Chubut River, west of Puerto Visser. Beside this, there are seven other speci- mens, mostly parts of hind limbs, but others having also the lower end of the humerus, the radius and ulna, metacar- pals, and some phalanges. The species is distinguished by its large size, being larger than the species of the Santa Cruz, and, at the same time, the three Fi g . 6. Distal end of right toes of both the pes and the manus are humerus 1/2 natural size. t i subequal in size. The distal end of the humerus associated indirectly with this species is moderately heavy, with fair-sized epi- condyles, and no entepicondylar foramen. The supra tro- chlear fossa is moderately deep, the anconeal very deep, the two being connected by a small foramen, as is typical for this family. The trochlearis, slightly oblique to the long axis of the shaft, has a simple pulley-like articular end without ridges of division, the internal border being narrower and higher than the external. MEASUREMENTS, SPECIMEN 3201 Humerus, greatest diameter of the distal end 5 8 mm - width of trochlea on the anterior side 37 mm. width of trochlea on the posterior side 28 mm. NOTODIAPHORUS CRASSUS 33 Fig. 7. Right radius and ulna, distal end of ulna from specimen No. 3275 1/2 natural size. 3 Fig. 8 Left femur posterior side 1/2 natural size 34 THE DESEADO FORMATION OF PATAGONIA The radius and ulna were from another specimen which, however, was associated with a typical astragulus. The two bones are long, slender, strongly curved, and in con- tact with each other throughout their entire length, so that there could have been no rotary movement of the fore- arm. The radius is a slender bone with the proximal articular facet relatively small, the facet being slightly concave, of ovoid outline and with the transverse diameter the greater. There is but a tiny band-like facet for the ulna situated on the posterior side near the inner margin. Distally, the radius widens into a heavy end with a rugose area on the outer side for contact with the ulna, and with two distal facets, a larger for the scaphoid, and a smaller for the lunar, the two being separated by a low ridge. The ulna is heavier above, with a strong backwardly directed olecranon process. The sigmoid notch makes almost a semicircle^ the articular surface being broad and extending well onto either side of the bone. The facets for the radius are tiny. The distal end of this bone is wanting. MEASUREMENTS, SPECIMEN No. 3275 Radius, length 251 mm. greatest width at proximal end 28 mm. greatest width at distal end 36mm. least diameter of shaft 16 mm. The femur belongs to the type specimen which is about 5% larger than the other specimens. This bone is long and rather slender, with the greater trochanter rising well above the head, which is rounded, on a short neck, and has the ligamentary pit on the posterior margin. The thick, rugose, greater trochanter bends in over the head at its upper end. The lesser trochanter is relatively small, and prolonged into a ridge. Unfortunately the third tro- chanter is broken off in my specimen. The digital fossa is extremely large and deep. Proximally the shaft is flattened, but becomes rounded distally. Just above the NOTODIAPHORUS CRASSUS 35 condyles there is a deep rugose pit for the plantaris muscle, and on the anterior side the suprapatellar fossa is well marked. The condyles are placed a trifle obliquely; the internal one being shorter and with a rounded articular face, the external condyle being longer, and with a flat- tened articular face which slopes obliquely inward. Of the tibia, only the distal end is preserved. This in- dicates a rather slender bone, with a shal- low, fairly wide concavity for the external astragular trochlea, and a narrower and deeper concavity for the internal astragu- lar trochlea. O/i the internal side of the tibia there is a rugose surface for the fibula. An isolated lower end of a fibula indi- cates a slender bone, enlarged distally where it comes in contact with the tibia. The fibula carries on its inner face a mod- erately large facet for the external side of the astragulus, and on the distal end a wider one for contact with the calcaneum. The tarsus is compactly built, wider than that of Diadiaphorus, because the external digits are not as much reduced. This especially shows in the greater de- Fig. 9. Distal end of teft t c , 1-1 i , i tibia */ 2 natural size. velopment of the cuboid and the meso- cuneiform, but in other features it is similar to that of its descendant. The astragulus is a very characteristic bone. The trochlea is asymetrical, the external condyle rising higher than the internal, and the median groove being wide and shallow. On the nearly vertical outer face of the astragulus, there is a semicircular band-like facet for the fibula. The trochea extends well around the top of the bone, allowing a wide movement of the foot. The neck of the astragulus is long and wide, carrying a broad flattened head, with its con- vex facet for the navicular, covering the entire end. On THE DESEADO FORMATION OF PATAGONIA Fig. 10. Left pes, dorsal side, ungual phalanx from specimen No. 3275 1/2 natural size. NOTODIAPHORUS CRASSUS 37 the plantar side are the most marked features. The ectal facet is in two planes, the anterior portion being bent down to nearly right angles with the posterior, which seems to be characteristic of this Diadiaphorus series. The susten- tacular facet also is characteristic, being gently rounded and extending clear to the navicular facet on the head, in Diadia- phorus becoming actually confluent with c , Fig. n. Leftastragulus the naVlCUlar facet. Just at the edge Of Plantar side: a, ectal facet J 1/2 natural size; b, sus- this sustentacular facet is a tiny surface foSord. facet; C(facet where the astragulus rubs on the cuboid, the only case, as far as I am aware, where this occurs in any Litopterna. The calcaneum is long and slender, the tuber being but slightly enlarged, its sustentacular facet being a broad oval surface, while the ectal facet is in two planes to cor- respond to that on the astragulus. The facet for the cuboid is at the distal end, but is unusually oblique, its inner margin sloping up almost to the sustentacular facet. It is this slope which brings the cuboid in contact with the astragulus. The navicular is broad and low, with a prominent hook behind. On its upper face there is only the broad facet for the astragulus head; on the lower face are three facets, externally, a large, more or less triangular area, for the ectocuneiform; medianly a smaller similar facet for the mesocuneiform; and on the internal side, sloping up onto the internal face, a small facet for the reduced endocunei- form. On the external face of this bone there is a tiny beveled facet for the cuboid. The endocuneiform is a large scale-like ossicle articulating on the lateral internal face of the navicular, and over- lapping markedly the inner surface of Metatarsus II. The mesocuneiform is considerably reduced in size, carrying a broad flat facet on the upper surface for the 38 THE DESEADO FORMATION OF PATAGONIA navicular, and a shallow saddle-like one below for Mt. II, which is entirely carried by this bone. The ectocuneiform is considerably larger than the meso- . cuneiform, resting above on the navicular, and carrying below the whole of Mt. III. On its inner side are two facets which rub against the upper end of Mt. II. The cuboid is a nodular bone, its upper surface occupied by the facet for the calcaneum, the lower face occupied by the facet for Mt. IV, while on the external side there is a tiny beveled facet for the vestige of Mt. V, and with a small boss on the inner surface which carries two tiny facets, the upper Fig. 12. cuboid one for the ectocuneiform and the lower for the internal side to navicular. On this same inner side, near the for navicular ^c top there is a second small boss, which carries tilr & tiny facet to rub on the astragulus, and fecetfar below that a second tiny facet for the navicular. mesocunieforms r^i . r , .. . . . . i/2 natural I he pes consists of three digits, with a vestige of Mt. V. Of the developed digits, the median one is the largest, but the two lateral digits are only a little smaller and were functional, so that this form was truly three-toed, comparable in the digital reduction to Meso- hippus. Mt. II is flattened above but soon broadens into a rounded shaft of considerable length, on the end of which is the articular trochlea, with the carina extending onto both the upper and lower surface, being, however, higher on the lower surface. Proximally this bone is overlapped by the endocuneiform, is carried by the small mesocunei- form, and also articulates on the inner side of the ecto- cuneiform. Mt. Ill is also compressed at the upper end, broadens below, and carries an articular trochlea similar to that of Mt. II, except that the carina does not extend so far onto the upper surface. Like Mt. II, Mt. IV is carried high on the tarsus, and therefore, though nearly as long as Mt. Ill, it does not have the same effective length. NOTODIAPHORUS CRASSUS 39 Proximally it articulates entirely on the cuboid; distally it has a trochlea similar to that of Mt. II, the carina extend- ing onto the dorsal surface. While Mt. V is lacking, it is clearly indicated that a vestige of it should have been present, as there is a tiny articular surface for it on the cuboid, and a rugose surface on the outside of Mt. IV. The phalanges are long and have the articular ends swollen somewhat as in camels. The phalanges of the first row are nearly equal in size, each with the proximal trochlea deeply notched for the carina of the metatarsus; and with the distal trochlea simple, though slightly concave from side to side, and reflexed well onto the dorsal surface. The phalanges of the second row are shorter and simpler, and somewhat depressed distally. The ungual phalanges are flattened from top to bottom, of moderate size, some- what longer than wide, and without any indications of a cleft. MEASUREMENTS, SPECIMEN No. 3287 No. 3275 Femur, length 289 mm. diameter across gr. trochanter 80 mm. diameter of middle of shaft 32 mm. diameter of distal end 7 mm. Tibia, diameter of shaft 28 mm. 24 mm. diameter at distal end 3 8 mm - 3 6 mm. Calcaneum, length 103 mm. 96 mm. width 3 6 mm. 35 mm. Astragulus, length 4 8 mm. 44 mm. width 3 8 mm. 35 mm. Metatarsus II, length 114 mm. 105 mm. Metatarsus III, length 122 mm. 114 mm. Metatarsus IV, length no mm. 101 mm. Phalanx I of digit III, length 49 mm - Phalanx 2 of digit III, length 27 mm. Phalanx 3 of digit III, length 29 mm. D cut er other ium Ameghino Deuterotherium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 633. Deutorotherium Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 452. This genus was first founded on a clacaneum and a bit of the mandibular symphysis, to which were added, later, 40 THE DESEADO FORMATION OF PATAGONIA both the upper and lower premolar and molar teeth. As far as it is known, it is distinguished by the upper molars lacking the metaconule entirely, and being approximately like those of Thaotherium. The dental formula is given by Ameghino as 2043* i, liol. .\cad. Nac. Cienc. Cordoba, 1. 16, p. 374. The general characters of this species are given under the generic description and I will here give only Ameghino's measurements which go with the figure: Skull, length over all Skull, length of the palate Upper dentition, length the inc. I to Diameter of the palate opposite inc. Diameter of the palate opposite in. 3 Height of mandible under m. I 210 mm. 120 mm. 120 mm. 37 mm. 75 mm. 33 mm. Kin. "7- M. imbricatus 1/2 natural size; A, upper dentition; /', louei ilni- tition after Ameghino. Fig. 68. K. latirostris, palatal view, after a pho- tograph of the type 1/2 natural si/e; the cross hatched an-a lepiesents matrix not yd removed. EURYGENIOPS 1 07 Eurygeniops Ameghino Eurygenium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 655. Eurygeniops Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 464. The name first given this genus was found to be preoc- cupied, and therefore changed. It is a clear cut genus, differing from the others in the family in the expansion of the front of the muzzle, and by the heavy broad character of the skull. Eurygeniops latirostris Ameghino E. latirostris Amegh., loc. cited above. This is the type species and is based on a muzzle which has never been figured, but which I figure, the drawing being made from a photograph taken by Professor Scott and kindly furnished me. The characters of the species are those of the genus, with the following measurements for specific determination, quoted from Ameghino: Palate, length 130 mm. Palate, width between incisors 3 41 mm. Palate, width between premolars 2 33 mm. Palate, width between molars 3 5 6 mm. Upper dentition, length from pm. R. to m. 3 82 mm. Upper premolar 4, length 1 1 mm. Upper premolar 4, width 19 mm. Eurygeniops normalis Ameghino E. normalis Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 466. This second species is described as being much smaller than the preceding, the length from pm. 4 to molar 3 being 65 mm. CHAPTER VII LEONTINIDAE THIS family was established to include a group of large, heavily built ungulates, not unlike rhinoceroses in form, which have rooted teeth, the molars being similar to those of Rhinchippidae, i. e., composed of a wall and an anterior and posterior lobes, but with the cristae either lacking or little developed ; and with the second upper, and the third lower incisors developed into tush-like caniniform teeth. Two genera are especially abundant, Leontinia of the Des- eado beds, and Colpodon of the Colpodon beds, the former with the formula 3^3, the latter with f^-j. In many ways, the family suggests Nesodontidae, and undoubtedly belongs to that series, if not directly ancestral. The lower molars are distinctly of the same type as in all the other toxodonts, but show a tendency to become hyposodont. The following genera have been assigned by Ameghino to the family. Some of them are based on very scant mate- rial and I have ventured to suggest in each case what dis- position I have felt to be the proper one. Leontinia, the type genus, is described in detail on pages 109-115. Scaphops is based on a mandibular symphysis, which is wider than usual for Leontinia, and on a second upper incisor which is compressed. The species in the genus Leontinia show a marked degree of variability, and I can see in this only individual variability, so that I place Scap- hops under Leontinia and S. grypus, as a synonym of L. gaudryi. Steniogenium is based on a mandibular symphysis with roots only of the teeth. The incisors are proclivous and inc. 3 small. I consider this also as Leontinia, and the species S. sclerops as a synonym of L. oxyrhynca, which I think is the female of L. gaudryi. LEONTINIDAE 109 Ancylocoelus is a valid genus, differentiated by its dental formula f^fy, the loss of the upper canine and the lower canine and first premolar distinguishing it from either Leon- tinia or Colpodon. Rodiotherium is based on a mandibular symphysis which would indicate an animal with the same formula as the foregoing genus, differing only in that lower incisor 3 is large. This, to my mind, does not make a generic char- acter, and at most the species, R. armatum, can only be considered an independent species belonging to the genus Ancylocoelus. Loxocoelus is a very questionable genus, based simply on an upper molar, which "is similar to that of Homolo- dontotherium, but more squared." I feel that in regard to this genus it should stand as unknown until more material is found. In our collection, over twenty skulls and jaws belonging to this family turned up, but all clearly belong to two types, the typical Leontinia gaudryi, and some others in which the caniniform teeth are not so well developed, which are either L. oxyrhynca or, as I believe, the females of L. gau- dryi. It is this uniformity of the material which leads me to doubt the validity of the considerable number of genera which Ameghino has established, for I found on sectioning the teeth that between the little worn crown and the much worn one there was a marked difference in the appearance of the infoldings and in the development of the pits. Leontinia Ameghino Leontinia Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 647. Leontinia Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 469. Scaphops Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 475. Steniogenium Amegh., 1895, Inst. Geog. Argen., t. 15, p. 654. Steniogenium Amegh., 1897, Inst. Geog. Argen., t. 18, p. 475. Colpodon Gaudry, in part, 1906, Anal. Palaeontologie, t. I, p. 30. The formula is 3 3 ] \\ ; type species L. gaudryi. Of all the animals in the Deseado, this is the most abundant. 110 THE DESEADO FORMATION OF PATAGONIA I 'sing this species as a basis, the following are the charac- teristic features. The first upper incisor is a small crop- ping tooth, with a well-developed cingulum high up on the inner face, which, when the tooth is worn down to the proper level, unites with the main part of the crown and incloses for a short time a small pit. On the external face there is also a feeble cingulum near the base of the enamel. The second incisor is greatly enlarged into a caniniform tush. In the species L. oxyrhynca, this tooth is much smaller but as this reduction of the tushes is the only dif- ference, I consider these forms as the female. The third incisor is again small, and has a well-developed cingulum on both the front and inner faces. The canine is similar to inc. 3. The first premolar is much reduced in size, with a weak cingulum on the outer face, and probably another on the inner side (the tooth is too much worn in my specimen to be sure). Beginning \\ilh premolar 2, the upper teeth are molariform. The premolars are rectangular in outline, each being much wider than long, and each having a cingu- lum on the outer side. On the inner side, along the anterior half of each premolar, there is a high cingulum, which, though interrupted at the anterior corner, continues around onto the anterior face of the tooth. On a worn tooth this anterior cingulum unites with the grinding surface, and leaves a small pit in the anterior internal corner, which is very suggestive of Rhynchippidae. In the middle of the grinding surface, there is an oblique pit, the remains of the basin in young teeth. The molars continue to increase in vsize toward the rear. They have a vestige of a cingulum on the external side, no cingulum on the inner side, but on the anterior side for about one third the distance there is a cingulum similar to that on the premolars. In the middle of the grinding surface is an elongated oblique pit, similar to that in the premolars, but a little more advanced, there being a trace of the development of cristae. LEONTINIA III In the lower dentition, the first two incisors are small cropping teeth, with the anterior face flattened and having a trace of a cingulum; while on the inner face the cingulum is well developed. Incisor 3 is developed into a tush cor- responding to inc. 2 in the upper dentition. As in the upper teeth, there are two types, that of L. oxyrhynca with the tush only about twice the size of an incisor, and that of L. gaudryi with it much larger. All the premolars are molariform and of the typical tox- odont character, consisting of two crescents with a pillar and septum in the posterior crescent. The septum, how- ever, does not appear until on pm. 3 and on all succeeding teeth, and is usually indicated by a tiny pit. From the front to the back, the premolars are progressively larger, each having a cingulum on both the internal and external faces. The molars continue to increase in size progres- sively, and have the same characters as the premolars, except that the crescents are mqre elongated, and the cin- gula are gradually becoming smaller toward the rear. The skull is low and heavy, with a low sagital crest, and with the lambdoidal crests continuous with the upper mar- gin of the zygomatic arches. The nasal bones are short and wide, and are markedly raised above the nasal cham- ber. On the outer margin of each is a low boss, somewhat as on the nasals of the rhinoceros, Dicer atherium, which would indicate that this form had a small pair of nasal horns.* The frontal bones are broad, projecting laterally in strong postorbital processes, which, with those from the jugals, almost close the orbit behind. The premaxillae are pecu- liar in having a median crest on the upper surface, the top of the crest being rugose, as though in life it had continued upward as a cartilage septum. The maxillae rise well up * Scott has restored the head of Leontinia gaudryi with a single median horn, but no specimen in my collection would indicate anything but a pair of nasal horns. See Scott, Mammals of the Western Hemisphere, fig. 138, 1912. 112 THE DESEADO FORMATION OF PATAGONIA on the sides of the skull, bounding the lower part of the orbit, and having a short zygomatic process. The small lachrymal is but little exposed on the exterior surface of the skull, the lachrymal pit being well inside the orbit. The zygomatic arches are broad and heavy, and composed mostly of the wide jugal bones. The palate is highly arched and relatively narrow, the crowns of the premolars and molars projecting inward over it, thus narrowing it still more. It extends back well beyond the last molar. The large tympanic bullae are hollow, and the cavity in the squamosum seems to be reduced in size, as compared with Rhynchippidae or Nesodontidae. The occipital condyles are set well apart and are sessile; and the paroccipital proc- esses are long and slender. The atlas, axis and cervical 3 are associated with the skulls. The atlas is short, heavy, and has the anterior cotyles broad, deeply excavated and wide apart; while the posterior cotyles are nearly flat, and high as well as wide. The transverse processes are only moderately wide, but are very heavy, especially along the posterior margin. The centrum of the axis is flattened, the neural canal, wider than high, and the neural spine of moderate height. The anterior cotyles are broad and moderately convex, and the odontoid process is a stout peg-like process, somewhat higher than wide. Slender transverse processes project sharply from the centrum, and have at their bases a large canal for the vertebral artery. Cervical 3 is shorter than the axis, has a less depressed centrum, a small neural spine, and short wide transverse processes. Though I have skulls and jaws to represent some twenty- five individuals, no limb material was found in direct asso- ciation with any of them. However we did find a humerus, radius and ulna on the same level and about fifteen feet front one of the skulls, and as it corresponds in size, and as humeri of this type are the most abundant skeletal bones found (as is also the case with the skulls), I have considered LEONTINIA II 3 it proper to associate these fore limb bones with these skulls. The humerus is a stout bone, of medium length, with a large sessile, and but little rounded head. The external tuberosity is wide, thick and projects a little above the head, while the internal tuberosity is so small as to be almost negligible. The shaft is flattened laterally at the upper end, but distally is compressed in the antero-pos- terior direction. The supratrochlear fossa is shallow, the anconeal deep, but there is no foramen connecting them. The external condyle is small, the internal much larger. The trochlea is narrow, with a swollen articular area for the radius, and a wider saddle-like one for the ulna. The ulna is a stout, nearly straight bone, slightly longer than the humerus. The olecranon process, though large, is not excessive. The sigmoid notch makes a deep semicircular cavity, with the articular facets expanding on either side. It was closely fitted to the radius so as to allow little or no rotary motion of the forearm. The facet for the radius is a narrow band-like area just below the sigmoid notch. The shaft is almost rectangular in section. Distally the ulna contracts sharply into a heavy styloid process, on the end of which is a large convex facet for the pyramidal, which merges without interruption into the facet for the pisiform. The radius is a slenderer bone, with a relatively small prox- imal head, but distally expanded into a much larger articu- lar end. My specimen is considerably weathered, but shows a wide shallow articular facet for the humerus, and a band-like facet for the ulna, but otherwise it gives little more than the length. Of the hind limb, Gaudry* figures the astragulus and the calcaneum, the former short and with a low trochlea, the latter also short and with a broad facet for the fibula. Gaudry also states that the foot was tridactyle and planti- grade, but I am doubtful of the plantigrade feature. * Anales Palaeontologie, 1906, t. I, p. 28. 114 THE DESEADO FORMATION OF PATAGONIA Ameghino has made six species of this genus, L. gaudryi, L. fissicola, L. lapidosa, L. oxyrhynca, L. stenognatha, and L. garzoni. All of the first five are described as of the same size as L. gaudryi. L. garzoni is a smaller, about 60 per cent, of the size of the others. Of the first five listed, the first three have the large incisor and I consider them all L. gaudryi. L. oxyrhynca and L. stenognatha are described as having small canines and I believe that this is a sexual difference only, so have considered these two species as also belonging to L. gaudryi, but females. I have made a careful comparison of L. gaudryi and L. oxyrhynca and find them identical in all the features except in the region of the canines where the latter is weaker, and can see no more than sexual differences. Usually with this weakness of the canine goes a smaller or lighter build of the lower jaw which is what would be expected. The points by which the various species were differentiated were, beside the size of the canine, the presence or absence of pit 3, and the vari- ation in the foldings on the outer sides of the lower molars, which I find on sectioning a tooth appear deeper or shal- lower according to whether the tooth was more or less worn. Leontinia gaudryi Ameghino L. gaudryi Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 648. L. gaudryi Amegh., 1897, Bol. Geog. Argen., t. 18, p. 472. Scaphops grypus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 629. Scaphops grypus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 475. Steniogenium sclerops Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 654. Steniogenium sclerops Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 475. Leontinia fissicola Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 474. L. (Senodon) lapidosa Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 649. Females L. oxyrhynca Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 472. L. stenognatha Amegh., 1897, Bol. Inst. Geog. Argen., t. 15, p. 474. Colpodon gaudryi Gaudry, 1906, Anal. Palaeontologie, t. I, p. 30. This species is represented in the Amherst Collection by five more or less complete skulls, and over twenty jaws, LEONTINIA GAUDRYI being by far the commonest fossil of the Deseado beds on the Chico del Chubut River, west of Puerto Visser. As mentioned above, there are two types, the first, with larger Fig. 69. Right upper dentition 1/2 natural size. Fig. 70. Lower dentition of male 1/2 natural size. Fig. 71. Lower den- tition of female 1/2 natural size. canines and heavier mandibles, designated by Ameghino as L. gaudryi, which I consider males; second, those with smaller canines, and lighter mandibles, slightly smaller in size, which Ameghino designated L. oxyrhynca and I con- n6 THE DESEADO FORMATION OF PATAGONIA sider females. Practically all of the other species are based on mandibular symphyses varying in details from the above, but in no case sufficiently for me to see a specific variation. The general features have been discussed under the generic description. Incisor i has a long crown and a long root, and is greatly crowded by the tushes. Incisor 3 and pm. I have the same crowded appearance. In giving the measurements I have used a skull which is typically L. gaudryi, a male, and parallel to it have put another skull, which is typically L. oxyrhynca, the female. By compar- ing the two sets of figures, the shortening, of which Ame- ghino speaks, will be seen to be all in the region of the tushes. SPECIMEN 3290 SPECIMEN 329 ix MALE FEMALE Upper dentition, length from inc. I to m. 3 Upper dentition, length from pm. I to m. 3 Incisor I, length Incisor 2, length Incisor 3, length. Canine, length Premolar I, length Premolar, I width Premolar 2, length Premolar 2, width Premolar 3, length Premolar 3, width Premolar, 4, length Premolar 4, width Molar i, length Molar i, width Molar 2, length Molar 2, width Molar 3, length Molar 3, width Lower dentition, height of mandible under m. i Incisor I, length Incisor 2, length Incisor 3, length Canine, length Premolar i, length LEONTINIA GAUDRYI Premolar I, width Premolar 2, length Premolar 2, width Premolar 3, length Premolar 3, width Premolar 4, length Premolar 4, width Molar, i, length Molar i, width Molar 2, length Molar 2, width Molar 3, length Molar 3, width Fig. 72. Top view of skull of L. gandryi (female) 1/4 natural size. In the skulls there is considerable variation in size in the different individuals, but the proportions remain very n8 THE DESEADO FORMATION OF PATAGONIA much the same throughout. In the female the snout is relatively a little shorter, and in general the female skulls are from 5 to 10 per cent, smaller throughout. The fol- Fig. 73. L. gandryi, view of case of the skull, female (L. oxyhynea) 1/4 natural size; Tym- pamic bullae broken open. lowing two sets of figures illustrate the comparative sizes of the two sexes. Skull, greatest length front to back Skull, greatest width Skull, length of nasal bone Skull, length of palate SPECIMEN 3335 SPECIMEN MALE FEMALE 420 mm. 252 mm. 115 mm. 235 mm. 392 mm. 236 mm. 102 mm. 230 mm. LEONTINIA GAUDRYI 119 The atlas associated with skull No. 3335 is a decidedly heavy bone in all its proportions. The axis and the third cervical were associated with skull No. 329 ix, and are likewise heavy bones. The following are typical meas- urements: Atlas, greatest length 86 mm. Atlas, greatest width 170 mm. Axis, length of centrum and odontoid process 132 mm. Axis, length of odontoid process 34 mm. Axis, width across anterior cotyles 98 mm. Cervical 3, length of centrum 66 mm. Cervical 3, width of posterior end of centrum 55 mm. Fig. 74. Atlas seen from below 1/4 natural Fig. 75. Axis and cervical vertebra, No. 3 1/4 size. natural size. While there are other vertebrae in the collection, which probably belong to this species, I have not cared to make the association without some evidence of a definite char- acter. However, in the case of a fore limb, which was found fairly near one of the skulls, is of proper size, and because this humerus occurs with something like the fre- quency of the skulls, I have been convinced that it belonged to this species, and so described it under the genus. This specimen consists of the two humeri, the radius and the ulna, No. 3328. Humerus, greatest length 314 mm. Humerus, diameter of head 77 mm. Humerus, transverse diameter of the shaft 43 mm. Humerus, width of distal end 116 mm. The ulna lacks some 60 mm. in the middle of the shaft, but when fitted to the radius its length can readily be ob- 120 THE DESEADO FORMATION OF PATAGONIA tained. The radius is considerably weathered so that measurements of the distal end are only approximate. Ulna, length over all Ulna, transverse diameter of distal end Radius, length over all 430 mm. 58 mm. 310 mm. Fig. 76. Right humerus from the posterior side 1/4 natural size. g. 78. mal end of right radius 1/4 natural size. Fig. 77. Right ulna from exter- nal side 1/4 natural size. ANCYLOCOELUS 121 Leontinia garzoni Ameghino L. garzoni, Amegh., 1896, Bol. Inst. Geog. Argen., t. 15, p. 650. L. garzoni, Amegh,, 1897, Bol. Inst. Geog. Argen., t. 18, p. 474. We were not fortunate enough to find this species, but as described by Ameghino it is about 60 per cent, of the size of L. gaudryi. The type is a lower jaw, for which the following figures are given : Lower dentition, length from pm. i to m. 3 120 mm. Lower dentition, length from pm. I to pm. 4 45 mm. Lower dentition, length of pm. 4 15 mm. Lower dentition, length of m. 3 39 mm. Ancylocoelus Ameghino Ancylocoelus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 652. Ancylocoelus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 475. Rodiotherium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 653. Rodiotherium Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 476. This genus differs from Leontinia in its formula, ~-~ but, except for the loss of these canines and the lower pre- molars, is very similar. In this it seems to approach the line which gave rise to Colopdon. The premolars and molars are also narrower in proportion than in Leontinia. I have placed Rodiotherium also under this genus as I can not see a generic difference in the descriptions. However we were not fortunate enough to find these forms. Ameg- hino has described four species as follows: A. frequens, 1895, Bol. Inst. Geog. Argen., t. 15, p. 475. Upper dentition, pm. i to m. 3 150 mm. Lower dentition, pm. 2 to m. 3 150 mm. Upper molar 3, length 39 mm. A. lentus, 1901, Bol. Acad. Nac. Cordoba, t. 16, p. 407. Upper molar 3, length 48 mm. A. minor, 1901, loc. cit. Upper molar 3 34 mm. A. (Rodiotherium) armatum, see cit. above. Based on an imperfect mandibular symphysis, in which incisor 3 is very large. CHAPTER VIII NESODONTIDAE THIS family is characterized by the teeth being hyp- sodont, the second upper incisor and the third lower in- cisor being enlarged into caniniform teeth, the upper molars complicated by the development of cristae, limbs short, feet tridactyl and semidigitigrade. Fig. 79. A, upper and a lower molars 2 of Proadinotherium ; B, upper and b lower molars 2 of Coresodon; C upper and c lower molars 2 of Neudon -1/2 natural size. In the Santa Cruz, the family is represented by the two genera Nesodon and Adinotherium. In the Deseado we find Proadinotherium evidently ancestral to Adinotherium and very little differentiated from it. Ameghino has de- scribed a genus, Pronesodon, which is evidently ancestral to Nesodon. I have referred Coresodon to this family be- cause the molars of the upper and lower jaws are very close to those of Adinotherium. Ameghino has also de- scribed two genera, Nesohippus and Interhippus, based on upper molars which are very similar in pattern to Adi- notherium and which I believe belong to this family, if they prove to be valid genera, of which I have some doubt, feeling that they will prove to be the deciduous upper pre- PROADINOTHERIUM 123 molars of Proadinotherium or some similar form. The genus Senodon, which Ameghino also places in this family, I feel will prove to be worn teeth of Leontinia. Proadinotherium Ameghino Proadinotherium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 625. The dental formula is -fi-Jy, as in Adinotherium, the chief difference being that the teeth are less hypsodont Fig. 80. P. leptognathus, rear portion of skull 1/2 natural size; shaded areas are matrix. than in the Santa Cruz genus. Little is known as yet of the skeleton, but when more is known probably more dis- tinctive characters will appear. Ameghino made two species, P. leptognathus which we also found, and P. angus- tidens a much smaller form. 124 THE DESEADO FORMATION OF PATAGONIA Proadinotherium leptognathus Ameghino P. leptognathus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 625. P. leptognathus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 467. Of this species we found on the Chico del Chubut River, west of Puerto Visser, three specimens; the back of a skull as far forward as molar 2, and two lower jaws. In general, the species is very similar, even to size, to Adinotherium ovinum of the Santa Cruz. The upper molars are strongly hypso- dont, curved teeth. On the upper surface, the basin is subdivided by two strong cristae into three smaller bays. In an early stage of wear, the second crista unites with the posterior lobe, convert- ing bay 3 into a pit. On the posterior margin of the tooth, the cingulum is de- veloped so as to appear like a third crista, ^ which inclosed bay 4, and when the tooth 1/2 natural size. 11 i is worn, bay 4 becomes a pit also. In my lower jaw incisor 3 is developed into a strong caniniform tush. Most of the teeth are lacking, but lower molar 2 is a strongly compressed, hypsodont tooth, surrounded by a thick layer of enamel. This tooth rises 22 mm. above the well-developed roots, and is already considerably worn down. The pillar is prominent as a strong fold in the middle of the posterior crescent. In this specimen there is no trace of the usual pit (3) indi- cative of the septum, but I should expect to find it in a younger specimen. The mandible broadens in front into a scoop-like anterior end, and the alveoli of the first two in- cisors would indicate that they were proclivous. The alve- oli for the other teeth are aranged as in Adinotherium. PRONESODON 125 MEASUREMENTS Skull, width across the zygomatic arches 148 mm. Skull, width across opposite m. 3 (outside) 73 mm. Upper dentition, molar 2, length 25 mm., width 13 mm. Upper dentition, molar 3, length 23 mm., width 12 mm. Lower dentition, incisor 3, length 13 mm., width 7 mm. Lower dentition, molar 2, length 20 mm., width 7^ mm. Proadinotherium angustidens Ameghino P. angustidens Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 467. This is based on a single lower tooth, which is considered either pm. 4 or m. I, and measures 13 mm. long by 4^ mm. wide. Pronesodon Ameghino Pronesodon Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 626. The genus is said to resemble Proadinotherium, but with the caniniform incisors proportionally much shorter. An associated calcaneum is shorter than that of Adinotherium and longer than that of Nesodon. Two species are described. Pronesodon cristatus Ameghino P. cristatus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 626. P. cristatus Amegh., 1897, Bol. Inst. Geog. Argen., t. 15, p. 467. This species is very imperfectly known, is characterized by a large external anterior style, molars said to be 15 mm. wide. Pronesodon robustum Ameghino P. robustum Amegh., loc. cit. above. This is a larger species, of which the three lower molars are known, and which measure 16, 22, and 30 mm. in length respectively, while they are 9-10 mm. wide. 126 THE DESEADO FORMATION OF PATAGONIA Coresodon Ameghino Coresodon Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 630. Coresodon Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 459. Coresodon Gaudry in part, 1908, Anal. Palaeontologie, t. I, p. 46. In this genus, the pattern of the upper molars is essen- tially the same as in Proadinotherium, and they are of the same hypsodont character, and with roots. I can now find only the fact that in Coresodon the teeth are more compressed and somewhat more hypsodont, as a feature by which to distinguish this genus from Proadinotherium. Gaudry figures the front of a lower jaw under the name Coresodon which lacks the caniniform incisors. I have doubted the association, but should it prove correct, then this genus would be markedly different in that respect. Two species have been described, C. scalpridens, and C. cancellatus, both of which I consider the same. Coresodon scalpridens Ameghino C. scalpridens Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 630. C. scalpridens Amegh., 1897, Bol. Inst., Geog. Argen., t. 18, p. 459. C. cancellatus Amegh., 1901, Bol. Acad. Nac. Cienc. Cordoba, t. 16, p. 374. Of this species we found two specimens, one containing the three lower molars, the other the second lower molar O o C D Fig. 82. Sections of second lower molar; A , top; B, 4 mm. down; C, 10 mm. down; D, 1 8 mm. down natural size. only. In establishing C. cancellatus, Ameghino says it is of the same size as C. scalpridens, but distinguished by the basin in the upper molars being narrower, the internal fold not being bifurcated, and by the absence of islets of CORESODON 127 Fig. 83. Molars i to 3 natural size enamel. All these features seem to me to be the results of wear. While the pattern of the upper molars is the same as in Proadinothcritim, these teeth are about as wide as they are long. The lower molars, however, are more com- pressed, with the ante- rior crescent occupying about a third of the tooth, and having in the early stages a deep pit, which disappears when the tooth is worn down. In the middle of the basin of the posterior crescent is a large pillar, and between this and the median horn of the crescent is a tiny septum, which early unites with the pillar, leaving a tiny pit (3) which soon disappears entirely. In fact, in an old tooth, the second and fourth bays, having become pits, may even be lost also. MEASUREMENTS Upper dentition, molar I, length Upper dentition, molar i, width Upper dentition, m. i to m. 3, length Lower dentition, premolar 2, length Lower dentition, premolar 3, length Lower dentition, premolar 4, length Lower dentition, molar I, length 18 mm., width Lower dentition, molar 2, length 19 mm., width Lower dentition, molar 3, length 20 mm., width Interhippus Ameghino Interhippus Amegh., 1904, Anal. Mus. Nac. Buenos Aires, ser. 3, t. 3, p. 183. Interhippus Amegh., 1904, Anal. Soc. Cienc. Argen., t. 56, p. 34 of reprint. This genus was established on isolated teeth which closely resemble those of this family, though the genus was placed among the Rhynchippidae by Ameghino. The teeth de- scribed as molars are much elongated and have the cristae greatly developed, and in one species there is a style rising about the middle of the inner side of the molar. Another feature emphasized as characteristic of this and the next 128 THE DESEADO FORMATION OF PATAGONIA genus is, that the crowns are expanded much wider than the roots. While there is not yet enough direct evidence to prove it, I feel that this and the next genus will prove to be deciduous teeth, of either Proadinotherium or some related genus. Two species of this genus have been de- scribed, both from the upper Deseado. I. phorcus Amegh., loc. cit. above. This species is characterized by its size, the last upper molar (so called) measuring 16 mm. long by 14 mm. wide. I. deflexus Amegh., 1904, Anal. Mus. Nac. B. A., ser. 3, t. 3, P. 183. This species is based on a worn tooth designated as molar I (probably d. pm. 3) 14 mm. long by 19 mm. wide. Fig. 84. I. phorcus. "Upper molars" natural size, after Ameghino. Fig. 85. I. deflexus, " Upper molar i " natural size, after Ameghino. Fig. 86. N. insulattis, "Upper molar i" natural size, after Ameghino. Nesohippus Ameghino Nesohippus Amegh., 1904, Anal. Soc. Cienc. Argen., t. 56, p. 34 of reprint. Nesohippus Amegh., 1904, Anal. Mus. Nac. B. A., ser. 3, t. 3, p. 218. This genus is described as very like the foregoing, but differs in having a strong perpendicular style on the ante- rior external face of the upper molars. As in the preced- ing genus, the crown is considerably expanded above the roots. I feel that this genus will also prove to be the milk teeth of some one of the genera of this family. One species is described, based on a single tooth. N. insulatus Amegh., 1904, loc. cit. under the genus. The species is just as described under the genus, the last upper molar measuring 24 mm. long by 16 mm. wide; given as from the upper Deseado. CHAPTER IX ISOTEMNIDAE THIS family is distinguished by the formula -f-j-J-f-, by the incisors, canine and premolar I all being of subequal size, by all the teeth being brachydont, and by the cres- cents of the lower premolars and molars being modified. On these lower premolars and molars the anterior crescent is longer than the posterior, and the short posterior cres- cent on the exterior of the tooth; so that its anterior horn, instead of uniting with the posterior horn of the anterior crescent, comes in back to about the middle of the anterior crescent. Then the pillar, which in the other families is situated in the posterior crescent, is opposite the posterior horn of the posterior crescent. The small animals which represent this family are rare in the Deseado beds, much more abundant in the Casamayor. The family seems to have died out in the Deseado as no forms are referred to it in later epochs. We found no specimens belonging to the family; but to make this discussion complete, I will give a digest of Ameghino's descriptions, with reproductions of such figures as he has given. All of the genera and species are based on very fragmentary material. The genera assigned to the family are Trimerostephanos, Pleurocoelodon, Lophocoelus and Henricofilholia. Trimerostephanos Ameghino Trimerostephanos Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 646. Trimerostephanos Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 483. This genus is based on upper and lower teeth, and dis- tinguished by the premolars and molars having a weak style on the anterior corner, and by the anterior lobe being considerably larger than the posterior. Four species have been described. 1 3 o THE DESEADO FORMATION OF PATAGONIA T. scabrus Amegh., loc. cit for genus. This is the type species, originally based on the third lower molar, to which was later added, upper prernolar 4 Fig. 87. T. scabrus natural size; /i, upper premolar 4; B, lower molars i and 2. The following measure- and the molars, and lower molar 2. ments are given: Upper premolar 4, length 15 mm., width 21 mm. Upper molar 2, length Upper molar 3, length Lower molar 2, length Lower molar 3, length T. scalaris Amegh., 1897, Bol. Tnst. Geog. Argen., t. 18, p. 483, is based on lower pm. 2 to m. 2, a somewhat smaller species than the preceding, the series as given measuring 53 mm. 15 mm. 31 mm. 35 mm. 20 mm. 29 mm. Fig. 88. T. scalaris, premolar 2 to molar 2 natural size. T. angustus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 484. This species is described without a figure, as smaller than T. scalaris, pm. 2 to m. 2 being 59 mm. The mandible is also slenderer. PLEUROCOELODON 131 T. biconus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18. p. 484. ^ This species is based on two lower premolars, said to be the same size as T. angustus, but with the pillar larger. Pleurocoelodon Ameghino Pleurocoeloclon, Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 645. Pleurocoelodon, Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 484. This genus is distinguished by the absence of the style on the anterior external margin of the upper molars, in- Fig. 89. P. wingei natural size; A, first molar; B, third molar. stead of which the external face is excavated medianly. Two species are described, based on isolated upper teeth. P. wingei Ameghino P. wingei Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 645. P. wingei Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 485. This species is founded on a couple of isolated molars, probably belonging to the same individual. The following measurements are given: Upper molar I, length 22 mm., width 26 mm. Upper molar 3, length 24 mm., width 29 mm. P. cingulatus Amegh., loc. cit. above, is based on an in- complete upper molar, probably the second, which is dis- tinguished by having the internal cingulum excessively developed. It measures 30 mm. in length. 132 THE DESEADO FORMATION OF PATAGONIA Lophocoelus Ameghino Lophocoelus Amegh., 1904, Anal. Soc. Cient., Rep. Argen., t. 58, p. 245. Lophocoelus Amegh., 1904, Anal. Mus. Nac., ser. 3, t. 3, p. 352. The genus is founded on a single upper third molar from Mazaredo, which is distinguished by a feeble style on the external face, by the anterior lobe being obliquely placed, and by the presence of a small secondary bay on the posterior side of the great internal basin. L. macrostomus Amegh., loc. cit. above. This species is the only one described, and has the generic features, the upper m. 3 being 21 mm. long, by 25 mm. wide. Henricofilholia Ameghino Henricofilholia Amegh., 1901, Bol. Acad. Nac. Cordoba, t. 16, p. 404. The type species is H. cingulata, based on a single upper molar. In general the upper molars are similar to those of Leontinia, but more brachydont, and with the internal cingulum well developed and tending to be crenulated. Four species have been made, all based on isolated upper molars. Henricofilholia cingulata Ameghino H. (? Parastropotherium) cingulata Amegh., Bol. Inst. Geog. Argen. t. 15, p. 640. H. (? Parastropotherium) cingulata Amegh., 1897, Fig. 90. ILdulata. upper Bo1 ' Inst ' Ge g" Ar S en ' 4 - l8 P' 45<>. molar r natural size, after H. cingulata Amegh., igoi, Bol. Acad. Nac. Cienc. Ameghino. ^ , , Cordoba, t. 16, p. 404. This is based on an upper molar I of which I reproduce Ameghino's figure. It measures 28 mm. long by 29 mm. wide. HENRICOFILHOLIA 133 H. lustrata Amegh., 1901, Bol. Acad. Cienc. Cordoba, t. 1 6, p. 405. This species is smaller than the preceding, and is based on an upper molar i and a last lower molar. The measure- ments are as follows: Upper molar I, length 25 mm., width 25 mm. Lower molar 3, length 25 mm., width 12 mm. Fig. 91. H, inaequilatera, upper molars 3 and 4 natural size, after Ameghino. H. inaequilatera Amegh. loc. cit. above. This species is larger than the preceding with the in- ternal cingulum more developed. Upper molar 2 measures 30 mm. long by 29 mm. wide. H. circumdata Amegh., loc. cit. above. This is a still larger type, with the internal cingulum enormously developed. Upper molar I measures 42 mm. long by 36 mm. wide. CHAPTER X HOMALODONTOTHERIA THE forms making up the Homalodontotheria are char- acterized by a dentition which is clearly a derivative of that of Toxodontia, but is distinguished by the teeth being brachydont, by the canines being the teeth which tend to become tush-like, though not advancing to a marked degree. But the distinctive feature of the suborder is found in the feet, which are clawed, the ungual phalanges being deeply cleft; and further, the animals seem to have walked on the sides of the foot, suggesting the Ancylopoda; but there does not seem to have been a phylogeretic re- lationship, rather it is a case of parallel development. Most of the forms found are of considerable size, and they are relatively scarce in all the formations. The representatives of the group in the Deseado all belong to the genus Asmodeus, which seems to be directly ancestral to the Santa Cruz genus Homalodontotherium, which seems to be the last representative of the series, no specimens, referable to the suborder having been found in later beds. Asmodeus Ameghino Asmodeus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 643. Asmodeus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 476. The formula is S T I f , the upper incisors have pits in the crowns; the canines are moderately enlarged; the upper premolars and molars consist of an external wall, with an anterior and posterior lobe, the lower premolars and molars are typically like those of toxodonts. Two species have been distinguished, a larger, A. osborni, and a smaller, A. scotti. Our collection contains seven speci- mens, all of which should apparently be assigned to A. osborni. ASMODEUS OSBORNI 135 Asmodeus osborni Ameghino A. osborni Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 644. A. osborni Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 478. Homalodontotherium osborni Gaudry, 1906, Anal. Palaeontologie, t. I, p. II. The type of this species is a calcaneum and astragulus, to which Ameghino later assigned the rear part of a mandi- ble with pm. 4 and the three molars; also a milk dentition, this last I think wrongly, for it is too small. I should interpret this specimen as deciduous inc. 2 to deciduous pm. 4, plus permanent molar i, in which case the permanent molar corresponds to that of A. scotti and it is not necessary to discuss "the remarkable bicuspid canine," as Ameghino Fig. 92. Molars 1-3 of the left side 1/2 natural size. does. Gaudry had some of this material, upper molars, the lower end of the humerus, the ulna, calcaneum and astragulus, and he referred the genus as the same as Homoladontotherium. With this last, I can not agree. We found the three upper molars, the lower end of the humerus, part of the radius, the tibia, and two phalanges, all on the Chico del Chubut, west of Puerto Visser. While brachydont, the external faces of the molars are high, and each has a tiny cingulum along the base of the crown. There is also a strong cingulum around the anterior, internal, and posterior faces of the crown, which on the posterior margin flares out, making a marked and characteristic ridge. The grinding surface, with its ex- ternal wall and two transverse lobes, is very similar to the molar of a rhinoceros. When the tooth wears down, the 136 THE DESEADO FORMATION OF PATAGONIA inclosed basin becomes a large pit. Between the posterior lobe and the flaring cingulum on the posterior margin, there is also a small posterior bay, which, in an old tooth, will also appear as a pit, but being shallow, it does not last long. The lower molars, as figured by Ameghino, are of the same type as those of the toxodonts, consisting of two cres- cents with the pillar in the middle of the posterior cres- cent, but the crescents and pillar are very plump; so that with wear they form broad grinding surfaces; and the bays, instead of becoming pits, first appear as notches, then disappear entirely. Each premolar and molar has a cingulum on the internal and external sides. Fig. 93. Premolar 4 to molar 3 1/2 natural size, after Ameghino. MEASUREMENTS, SPECIMEN Upper dentition, molar i, length Upper dentition, molar 2, length Upper dentition, molar 3, length Lower dentition, from Ameghino's measurements Lower dentition, premolar 4, length Lower dentition, molar I, length Lower dentition, molar 2, length Lower dentition, molar 3, length 3179 46 mm., width 50 mm. 51 mm., width 55 mm. 50 mm., width 51 mm. 28 mm., width 23 mm. 34 mm., width 24 mm. 46 mm., width 24 mm. 76 mm., width 23 mm. Only the distal end of the scapula has been found; and this shows a shallow glenoid cavity, which is much longer in the antero-posterior direction, than in the transverse. The spine rises close above the rim of the glenoid, and is unusually heavy. The lower half of the humerus is present, and character- ized by very wide epicondyles, a shallow supratrochlear fossa, a moderately deep anconeal fossa, no foramen, and a wide shallow trochlea. The ulna, according to Gaudry, is a long, heavy, nearly straight bone, with a shallow sig- ASMODEUS OSBORNI 137 moid notch, and with a large olecranon process which is not bent backward to any marked degree. The proximal Fig. 94. Humerus, anterior side i/5 natural size. Fig. 95. Ulna anterior side i/5 natural size, after Gaudry. end of the radius has a broad doubly curved articular surface to fit the full width of the humeral trochlea. Its ulna facet is a short broad area just below the margin of the bone, and would indicate little or no rotary motion of 138 THE DESEADO FORMATION OF PATAGONIA the fore arm. Most of the shaft is lacking but what is present indicates a very slender bone. The tibia is also a rather light bone of moderate length, and is strongly curved inward, the inner margin being Fig. 96. Upper end of radius, ulnar side 1/5 natural size. Fig. 98. Astragalus, dorsal aspect 1/2 natural size, after Ameghino. Fig. 97- Left tibia, posterior side i/s natural size. Fig. 99. A, Ungual phalanx, No. 3; B, Ungual phalanx, No. 5 i/2 natural si/e. especially concave. On the wide proximal end, the inner condyle is concave, the outer convex, the two being sep- arated by a prominent bifid spine. The shaft is slender, with a deep groove down the anterior face especially at the ASMODEUS OSBORNI 139 upper end, while on the posterior face, there is a large interosseus crest, which starts just below and external to the spine, and extends in a sigmoid curve three-fourths of the length of the shaft, ending on the internal border Distally the tibia is flattened antero-posteriorly, and the internal margin extends as a wide process down to the level of the navicular face of the astragulus. The articular facet for the astragulus is a rectangular depression, being about half as wide in the an tero- posterior direction as in the transverse. This facet is only slightly concave and the inner and outer portions are not separated by an inter- trochlear ridge. The fibula has not been found, but the tibia shows no indication of its having been fused to it. Ameghino has figured the astragulus as very low, with the trochlea flattened, the internal condyle being wider and flatter, while the external condyle is narrower and somewhat raised. The trochlea is peculiar in that its proximal margin is deeply notched by a depression in which there is a large perforation. The neck is prolonged and carries a large convex head articulating with the navicular only. The measurements given are, length 116 mm., width 75 mm. Gaudry figures a calcaneum, showing a long narrow tuber, and the facet for the fibula as a wide shelf which projects strongly on the external side. The size as given by Ameghino is 240 mm. long, by 120 mm. wide. I have two associated ungual phalanges, one of which corresponds to that figured by Ameghino as the third. It is high, laterally compressed, has a very rugose surface on either side, and a deep cleft in the end. This is 68 mm. long. The second ungual is very asymetrical, also laterally compressed, and with the point curved inward. I take it to be the fifth. The tibia, the tarsus, and the phalanges strongly suggest that this animal walked on the side of its foot. 140 THE DESEADO FORMATION OF PATAGONIA Asmodeus scotti Ameghino A. scotti Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 643. A. scotti Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 477. This species is not represented in our collection, but I reproduce Ameghino's figure of the type, and of the milk Fig. TOO. Upper and lower incisors, canines, and premolars 1/2 natural size, after Ameghino. dentition. Unfortunately his type figure is from the side and does not give all the desired information. In the upper dentition, the small incisors, pitted on the crown, increase regularly in size toward the rear; and each has an external cingulum around the base. The canine is about twice the size of the adjacent incisor, and also has pm '/ Fig. 101. Milk incisors, canine, and premolars and permanent m. i 1/2 natural size, after Ameghino. an external cingulum. The premolars increase regularly in size and also have at least an external cingulum. Figure 101 shows a dentition which Ameghino described as the milk set of A. osborni. At the same time he remarks the unusual character of the deciduous canine in being two- cusped. I think this set of teeth should be interpreted as deciduous inc. 2 to deciduous pm. 4, plus the permanent ASMODEUS SCOTTI 141 molar I. With such an interpretation, we find the incisors normal, the canine normal though not as large as in the permanent set, and the two-cusped tooth is the first milk premolar. The last tooth in the series is considerably different from the premolars and is evidently permanent molar I , which is about the size and character of this tooth in A. scotti, much too small to belong to A. osborni. This set of milk teeth differ from the permanent teeth in that the premolars do not have the anterior, inner and posterior cingulum, characteristic of the permanent dentition. The following measurements are taken from Ameghino: Upper dentition, inc. I to pm. 4 Upper dentition, premolar 2, length Upper dentition, premolar 3, length Upper dentition, premolar 4, length Upper dentition, molar I, length Upper dentition, molar 2, length Upper dentition, molar 3, length 104 mm. 18 mm. width 25 mm. 20 mm. 23 mm. 28 mm. 37 mm. 50 mm. width 28 mm. width 35 mm. width 39 mm. width 44 mm. width 48 mm. CHAPTER XI ASTRAPOTHERIA THIS group is composed of large, long limbed creatures, with a highly specialized dentition, in which the canines of the upper jaw are developed into great curved tushes, resembling those of Pyr other ium; while the canines of the lower jaw are compressed in the antero-posterior diameter and protrude laterally, like those of pigs. Upper pre- molars I and 2 are reduced or lacking, while pm. 3 and 4 are also reduced, but usually retained. The upper molars are brachydont, and have a crown very like that of the molars of homalodontotheres. The lower incisors are small, proclivious, and set at intervals around the broad semicircle of the front of the fused lower jaws. The lower canines are permanently growing teeth, smaller than the upper canines, project laterally, and have the tips recurved. Premolars i and 2 are usually lacking, pm. 3 more or less reduced, and pm. 4 is a normal, short, molariform grinder. The lower molars have the same basal pattern as in Toxodonta, the crown carrying two crescents with a plump pillar in the basin of the posterior crescent, the pillar, however, being situated far forward near the anterior horn of the rear crescent. Lydekker made an order Astrapotheria including the Astrapotheria and Homalodontotheria, but as the dentition of the two groups is so different, because of the enormous enlargement of the frontal region, and because of the reduction of the premolars, I am convinced that these two groups represent totally divergent lines of develop- ment; and I have therefore made each of the groups a separate suborder. PARASTRAPOTHERIUM 143 Ameghino has described several genera, which make a progressive series and show a constantly progressive variation as far as they are known. GENUS Albertogaudryi Astraponotus FORMATION Casamayor Astraponotus FORMULA - Post, inner and post, median, cusps isola- ted. ? i ? 3 3- Post, nner cusp, un- ted with wall making small lobe. Parastrapotherium Deseado and Colpodon Astrapothericulus Astrapothericulus Astrapotherium Santa Cruz ? * 2 3 Post, lobe large, also 23 a strong crista. ? i 2 3 3123 3123 3113 In the Deseado beds, beside Par astro pother turn, Ame- ghino has described Liarthrus, based on an upper second premolar and part of another tooth, but I can see no structural variation from Parastropotherium or indeed from P. holmbergi; so I consider this genus as a synonym. As to the genus Traspoatherium, I can not see in it any reason for making a genus separate from Parastrapotherium. Parastrapotherium Ameghino Parastrapotherium Amegh., 1895, Bol. Inst. Ceog. Argen., t. 15, p. 636. Parastrapotherium Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 449. Liarthrus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 641. Liarthrus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 451. Traspoatherium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 641. Traspoatherium Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 450. The genus, in general, is similar to Astrapotherium, so that Gaudry considered it the same, but Ameghino has distinguished it by the tushes being relatively of smaller size, the lower incisors larger, and by the presence of pm. 3 144 THE DESEADO FORMATION OF PATAGONIA in the lower series. The Deseado forms are also of con- siderably larger size than the Santa Cruz. Our material includes a pair of lower jaws, two scap- ulae, the humerus, and the lower end of the femur. No skull has been found in the Deseado. Those from the Santa Cruz are enormously swollen over the orbits, the massive bone making a skull wholly unique. The lower jaws are similar to those of Astrapotherium, except that the rami are deeper. The front ends are fused and expanded making the anterior much enlarged, and causing the incisors to stand at intervals as in Coryphodon. The symphysis is massive and prolonged backward nearly to premolar 3. The rami are plump and unusually thick. Fig. 102. Upper dentition of Astrapothericulus iheringi 1/2 natural size. Of the upper dentition, Ameghino figures only the first molar and the canine. I have given Ameghino's figure of the upper dentition of Astrapothericulus, to indicate what this would be like, for the variation is only slight. The canine is a great tush, not unlike the incisor-tush of Pyrotherium, oval in cross section with the greater diameter from front to back. The first and second premolars have disappeared. Premolars 3 and 4 are greatly reduced. The molars are very like those of Asmodeus, large brachy- dont grinders, composed of an outer wall, and an anterior and posterior lobe. The external cingulum is a trace only, and the internal cingulum is developed in varying degrees. The basin is deep and subdivided by a crista which rises from the external wall, and as the surface is worn off unites with the anterior lobe, cutting off a small PARASTRAPOTHERIUM 145 pit. Behind the posterior lobe is a small basin, bounded in the rear by a second crista from the rear end of the external wall, which, as the tooth is worn down, unites with the posterior lobes, cutting off a small posterior pit, suggestive of that of homalodontotheres. The three lower incisors are expanded at their ends into thick shovel-like crowns, each with a strong crescentic cingulum on the posterior face, and with a shallow furrow on both the front and back faces. Relatively the incisors are much larger and longer than in Astropotherium. The lower canine is flattened on the upper face, so that its cross section is close to semicircular making a typical permanently growing rooting implement. This tooth is relatively shorter and smaller than in Astrapotherium. Premolars I and 2 are wanting, a long diastema occupy- ing the interval between the canine and pm. 3. Premolar 3 is greatly reduced in size, and in my specimen has fallen out, being represented by a small alveolus. I judge that in old individuals it falls out. The fourth premolar and the molars are typically those of Toxodontia. The young show two plump crescents, with a low plump pillar, sit- uated near the anterior horn of the posterior crescent, which pillar, as the tooth wears, unites with the anterior horn. The scapula is a remarkably heavy and elongated bone, greatly arched where it lay over the ribs. The spine is high and heavy, with the upper margin developed into a thick ridge like a banister rail, which is prolonged in front to, or a little beyond, the level of the glenoid fossa, this distal portion being expanded into a broad plate more than half as wide as the widest portion of the blade of the scapula. The glenoid fossa is relatively small, oval in outline, and with the long axis parallel to the long axis of the body. The anterior margin of the articular surface is reflexed, apparently to come in contact with the base of the greater tuberosity of the humerus. This glenoid cavity 146 THE DESEADO FORMATION OF PATAGONIA is only large enough to actually eover about half of the head of the humerus, and fits so that, in a position of rest, the glenoid covered the outer part of the humeral head, and only articulated on the inner part of the humerus head when the limb was bent inward. The blade of the scapula is narrow, with the proximal end prolonged and ending in a thick rugose mass. The anterior and posterior mar- gins are rugose and thickened, the great thickness of the proximal end being due to the convergence of these thick- ened margins and the heavy spine. Lastly, this thick proximal end is peculiar in having on its posterior side a large rugose cavity, which was apparently to receive mus- cular attachments. For such a heavy animal, the humerus is extraordinarily long and slender. The sessile head is strongly compressed from side to side, very convex, and much larger than the glenoid fossa, its articular surface extending onto the base of the greater tuberosity. This tuberosity is heavy and thick, but does not project above the head. The powerful deltoid ridge extends from the tuberosity two-thirds of the way down the shaft. The shaft is unusually slender. Distally it expands laterally to make the two large epicon- dyles, of nearly equal size. The trochlca is relatively narrow, the internal surface being the narrower, and rising to a high margin; while the external portion is wider, rounded, and has a low margin. The supratrochlear fossa is moderately deep, the anconeal fossa somewhat deeper, but there is no connecting foramen. Gaudry* figures a radius and ulna, both relatively long bones, and closely apposed; so that there was no possibility of a rotary motion of the forearm. The proximal end of the radius is expanded, so that its articular surface is in contact with the full width of the humeral trochlea on the anterior side. Below, the bone contracts to a moderately slender shaft, and then expands distally into *Anal. Palacontologic, t. I, p. 5, 1906. PARASTRAPOTHERIUM 147 a heavy club-like distal end. The ulna has a short but heavy olecranon process, with a prominent coronoid process. The sigmoicl notch is shallow, but the articular surface expands on both sides, so that it covers the full width of the humeral trochlea on the posterior side. Dis- tally the ulna is not so heavy as the radius. Under the name Pyrotherium romeri, Ameghino* figures a carpus and metacarpus, which Tournierf however assigns to Parastrapotherium, probably P. herculeum; and figures a carpus and metacarpus of the same type, but smaller, which he attributes to Parastrapotherium. I, however, can not see how such a small foot can belong to so large an animal, and feel that, until evidence of direct association is given, it is best not to consider these feet as belonging to Parastrapotherium, but rather to Pyrotherium. Of the femur I have only the distal end, which, however, corresponds completely with the one figured by Gaudry. It is a long bone, slightly shorter than the humerus, with a small head, set on a short and poorly outlined neck. The greater trochanter is wide and rugose, rising to about the same height as the head. The lesser trochanter is not distinguishable. About the middle of the shaft there is a powerful third trochanter, which continues as a narrow ridge upward to the greater trochanter, and downward in a similar narrow ridge almost to the outer condyle. At the proximal end the shaft is greatly flattened, but in the central and lower parts becomes almost circular in section. The two condyles are set wide apart, project considerably behind the posterior face of the shaft, and and are only slightly convex. The trochlea is of moderate width, short, and shallow. Gaudry outlines a short, heavy, rugose calcaneum which has but a short tuber; a flat navicular; a small cuboid; and an astragulus with only a slight convexity of the * Bol. Inst. Geog. Argen., t. 18, p. 442, fig. 25, 1897. t Bui. Soc. Geol. France, ser. 4, t. 5, p. 305, 1905. 148 THE DESEADO FORMATION OF PATAGONIA trochlea, and with the navicular facet directed obliquely forward, making an angle of 127 with the plane of the trochlea, which, as he says, would indicate a semidigiti- grade position of the pes. The following species are distinguished by Ameghino as coming from the Deseado beds: P. holmbergi, P. Irons- sarti, P. lemoinei, P. ephebicum, P. martiale, P. super abile, P. insuperabile. The various species are known from the same parts in but a few cases. Their relative sizes are indicated from the following compilation of the measure- ments given by Ameghino: P. holmbergi is the type species, and of considerable size, and to it I have assigned my material. In such a large animal, variations in size are to be expected. P. troussarti, as described, is a tenth smaller than P. holmbergi, the only structural character differentiating it being the isolation PARASTRAPOTHERIUM 149 of the pillar in the lower molars, which is a character due to youth; so I have considered it a synonym of P. holmbergi. P. ephebicum is a much smaller and distinct species, with which I should associate the single upper molar to which the name P. lemoinei has been given. P. martiale is a large species, distinguished by the strong development of the cingulum on the internal side of the upper molars, and on the inner side of the lower molars; and by lower premolar 3 being well developed with two roots. P. superabile is of the same size as the foregoing, but has the cingulum on upper premolar 4 (the only tooth known) less developed. I should therefore consider it a synonym of P. martiale. P. insuperabile is the largest species, and is distinguished by the excessive development of the cingulum. Liarthrus is founded on an upper pm. 4 with a part of pm. 3, but, as far as I can see, does not differ in character or size from P. holmbergi. Traspoatherium is based on upper premolars which are distinguished by the roots being fused from side to side. I think it is an age character and for the present would consider it the same as P. holmbergi, probably the tooth being pm. 3. / Parastrapotherium holmbergi Ameghino P. holmbergi Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 636. P. holmbergi Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 449. P. troussarti Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 638. P. troussarti Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 449. Liarthrus copei Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 641. Liarthrus copei Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 451. Traspoatherium convexidens Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 641. Traspoatherium convexidens Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 450. Parastrapotherium holmbergi Tournier, 1905, Bui. Soc. Geol. France, ser. 4, t. 5, P. 305. Astrapotherium holmbergi Gaudry, 1906, Anal. Palaeontologie, t. I, p. 5. To this, the type species, I have assigned all the material we found on the Chico del Chubut, west of Puerto Visser, 150 THE DESEADO FORMATION OF PATAGONIA as enumerated under the generic description. The lower jaws belonged to an old individual. The humerus and scapulae were found associated, the femur isolated. Of the upper dentition, the only available measurements are those of Ameghino for the first molar, and the canine. Upper dentition, canine, length 256 mm. Upper dentition, molar I, length 57 mm., width 57 mm. Fig. 103. Upper molar i of the left side natural size, al'trr Ameghino. The measurements of the complete pair of lower jaws which we found are Lower dentition, length from inc. I to m. 3 455 mm. Lower dentition, incisor 2, length 22 mm., width 22 mm. Lower dentition, canine, ant. post. diam. at alveolus 52 mm. Lower dentition, canine, trans, diam. at alveolus 26 mm. Lower dentition, diastema from c. to pm. 3 116 mm. Lower dentition, premolar 4, length 28 mm., width 26 mm. Lower dentition, molar I, length 43 mm., width 28 mm. Lower dentition, molar 2, length 58 mm., width 32 mm. Lower dentition, molar 3, length 70 mm., widlh 36 mm. Height of mandible under molar 3 83 mm. The scapula is a very long heavy bone, with a narrow blade, and a high spine which has its upper margin thick- ened so as to appear like a banister rail. We found one complete scapula and a second incomplete one associated with it, which corresponded in all ways to the first one. PARASTRAPOTHERIUM HOLMBERGI Fig. 104. Lower jaws 1/5 natural size. Fig. 105. Dorsal view of right scapula 1/5 natural size. 152 THE DESEADO FORMATION OF PATAGONIA Fig. 106. Right humerus, posterior aspect i/5 natural size. Fig. 107. Left femur, posterior side 1/5 natural size; outline of upper portion after Gaudry from Astrapotherium magnum. PARASTRAPOTHERIUM HOLMBERGI 153 The following measurements are taken from specimen No. 3328: Scapula, greatest length 694 mm. Scapula, greatest width 283 mm. Scapula, glenoid fossa, ant. -post, diameter 130 mm. Scapula, glenoid fossa, transverse diameter 90 mm. Scapula, height of spine 120 mm. Scapula, width of enlarged margin of spine at the lower end 170 mm. Scapula, width of margin in middle 45 mm. The humerus was associated with the two scapulae mentioned above, and is complete. For such a large animal, its length is excessive, greater than that of the species assigned by Gaudry to P. herculeum which species has a skull larger than that of P. holmbergi. Humerus, greatest length 720 mm. Humerus, greatest width across proximal end 248 mm. Humerus, least diameter of shaft 78 mm. Humerus, width across the epicondyles 220 mm. Humerus, width of trochlea on distal end 125 mm. The femur which Gaudry figures as belonging to Astra- potherium magnum corresponds, as far as the distal end will admit comparison, with the one which we found in the Deseado beds, so that in restoring the outline of the missing parts, I have based it on this A. magnum. Femur, length (estimated) 4 8 mm - Femur, width of distal end 135 mm - Femur, width of trochlea 57 mm - Parastrapotherium ephebicum Ameghino P. ephebicum Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 639. P. ephebicum Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 449. P. lemoinei Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 640. P. lemoinei Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 450. Astrapotherium ephebicum Gaudry, 1904, Mem. Soc. Geol. France, t. 12, p. 15. Ameghino based this species on a portion of the mandible of an old individual with molars I and 2. Its chief dis- tinction lies in its small size as compared with P. holmbergi. Gaudry assigned to this species some upper teeth. We 154 THE DESEADO FORMATION OF PATAGONIA found no specimens of this species. The following are the measurements of the type according to Ameghino. Lower dentition, molar I, length Lower dentition, molar 2, length 31 mm., width 16 mm. 42 mm., width 21 mm, Parastrapotherium martiale Ameghino: P. martiale Amegh., 1901, Bol. Acad. Nac. Cordoba, t. 16, p. 402. P. superablie Amegh., 1901, Bol. Acad. Nac. Cordoba, t. 16, p. 402. The species seems to have been founded on abundant material, representing an animal of larger size than P. holmbergi, which is dis- tinguished by the greater width of the crowns of the incisors, by straight canines diverging but little, by a strong cingulum on the outer side of the lower molars and the inner side of the upper molars, and by the narrow symphysis of the lower jaws. P. superabile was distinguished by a difference in the arrange- ment of the roots of upper pm. 4, but as the pattern of the crown is the same, as is also the size, I feel that this differ- ence is simply an individual variation. The following measurements are given by Ameghino: Fig. 1 08. Upper molar i of the left side after Ameghino. -natural size, Upper dentition, length from pm. 3 to m. 3 Uppt-r dentition, premolar 4, length Upper dentition, premolar 4, width Upper dentition, molar 2, length Upper dentition, molar 2, width Lower dentition, length from inc. I to m. 3 240 mm . 30 mm. 43 mm, 82 mm. 83 mm. 550 mm. PARASTRAPOTHERIUM 1 55 Paras trapotherium insuperabile Ameghino P. insuperabile Amegh., 1901, Bol. Acad. Nac. Cordoba, t. 16, p. 403. This is the largest of all the species, and is distinguished by the enormous development of the cingulum on the anterior, inner, and posterior sides of the upper molars, the same being uninterrupted and so elevated, that the internal crests seem to rise out of a basin. The following measurements are taken from Ameghino: Upper dentition, premolar 4, length 37 mm., width 48 mm. Upper dentition, molar 3, length 100 mm., width 80 mm. Lower dentition, premolar 4, length 43 mm., width 23 mm. CHAPTER XII PROBOSCIDEA Suborder Pyrotheria THIS suborder was established by Ameghino to receive the peculiar genus Pyrotherium and related forms. These animals are of large size, massive build, with narrow elongated skulls, in which the nasal opening is situated far back, as in animals with a proboscis; with a tiny brain case surrounded by cellular spaces; with the maxillae, palatines, pterygoids and alispenoids developed downward, so that the palatal plane makes a strong angle with the basi-cranial plane; and with the occipital condyles high up on the back of the skull. Then the first and second upper incisors and the second lower incisors are developed into enormous tushes with enamel on the anterior sides only. The remaining incisors, the canines, upper premolar i, and lower premolars I and 2 are wanting; the remaining premolars and the molars being developed into great quadrilateral grinders, each with two transverse crests. The neck is short, the limbs massive and short, especially the lower members of each limb, and the feet were probably five- toed. The relationship of these forms has been the subject of extended discussion: Ameghino seeing in this genus the ancestors of the Probiscidea, and comparing them with Palaeomastodon, Dinotherium and Barytherium, even finding resemblances to Diprotodon; Gaudry concludes that they are not proboscidians; and others have suggested that they were specialized toxodonts. I have prepared the following table of comparisons with Palaeomastodon, a toxodont, and Diprotodon. PYROTHERIUM 157 158 THE DESEADO FORMATION OF PATAGONIA Still other forms like Amblypoda and Arsinotherium have been suggested as having characters in common with Pyrotherium, and it is clear that, with such a variety of forms, some of the characters must be parallelisms due to a common adaptation, and only one of these varied groups can be the one to which Pyrotherium is related. For myself, I have made comparisons with the Amblypoda and Arsinotherium, and feel that such features as the mas- sive limbs, shape of individual bones, etc., are due simply to the fact that all these are massive animals. In the case of Arsinotherium, there are some characters which are also common to hyracoids and elephants, like the position of various basicranial foramena, the prolongation backward of the jugal and the shape of the palatines. My conclusion is that Pyrotherium is related to the pro- boscideans, and came from the same stock which gave rise to hyracoids, elephants and Arsinotherium. I think further that Pyrotherium belongs definitely to the pro- boscidean line. Referring back to the foregoing table. The develop- ment of tushes may be an adaptive character; but in the elepnants it is inc. 2 of the upper and inc. 2 of the lower PYROTHERIUM 159 jaw which are so developed. In Pyrotherium, in the upper dentition, it is also inc. 2 which makes the tush, and inc. I is enlarged as in Moeritherium, and, so far as we know, has not been reduced in later forms as it was in the elephant line. In the lower jaw we have no final evidence which will show whether it is inc. I or inc. 2 which makes the tush; but the lower tush bites against upper inc. 2 and I have considered it to be incisor 2. The loss of the teeth behind the tushes is a character to be expected in the development of tushes and gives no data. The bilophodont character of the back teeth has oc- curred many times in the animal kingdom and while it may be the inheritance of the early elephants it can not be used as an argument. The position of the nasal opening looks very much like that of elephants, but again is coincident with the devel- opment of a proboscis. However, this has not occurred a great number of times in the animal kingdom, and where it has, it takes a variety of forms of modification. In Py- rotherium, the modification is of the type in elephants, and elephants only. A very striking feature is the development of the dental region downward so that the basi-cranial axis is bent upward, makmg an angle of about 140 degrees. There are other cases of the bending of the basicranial axis; but in the other ungulates it is a bend downward, the reverse of what we find here and in elephants. To adjust the posterior part of the nasal chamber to this, the ptery- goids and the alisphenoids are developed into great wing- like plates on either side. I find this modification of the basicranial axis and of the palatal, pterygoid and alis- phenoid bones in no other group but the elephants. In Palaeomastodon it has been developed to a degree so that the angle is about 155 degrees. The back of the palatine bones is also characteristic, for the^e begin as narrow pointed bones and behind the 1 60 THE DESEADO FORMATION OF PATAGONIA last molar expand into wide plates, just as in Palaeomas- todon (and in no other groups), having the postpalatine foramen opposite or behind the last molar. The posttympanic region of the squamosum is modified so that this process unites with the anterior squamosal region to crowd out more or less completely the tympanic bone where it should surround the auditory meatus. This feature is common to the elephants, the hyracoids, and the toxodonts, so that I consider it a primitive feature indicative of the ultimate common ancestry of these groups. The tympanic bulla can be compared with that of elephants closely, and has much in common with that of toxodonts, but in this last group the tympanic is much more highly developed. The premaxilla bone in Pyrotherium is crowded out, so that it makes no part of the palate, which is a character of elephants, and in contrast to toxodonts or other groups which have been mentioned. There are two antorbital openings as in elephants, and a feature not common, though not unknown. On either side of the brain case are cellular spaces with intercellular lamellae, which are so characteristic of elephants; a confirmatory feature, though in itself not conclusive. The foramena on the base of the cranium are similar to those on the base of the cranium of elephants, though there are some variations, as for instance, the exoccipital foramen, is isolated in Pyrollierium, but fused with the posterior lacerum foramen in elephants, and other slight variations in position; but, on the whole, the foramena of Pyrolherium are much closer to those of elephants than of any other group. There is also much in common with toxodonts and with hyracoids, as would be expected if they have a common ancestry. There is no suggestion of a marsupial arrangement as would be necessary if related to Diprotodon. PYROTHERIUM l6l The atlas of Pyrotherium is peculiar in having a marked hypophysis which is unusual, but is a feature of the atlas of Palaeomastodon and Moeritherium. The axis is peculiar in that the odontoid is flattened on the upper side and very short and wide. In this the form is unique. The continuation of the articular surface on the lower sicle of the odontoid with the articular surfaces of the ant. cotyles is a feature also of elephants. The remaining cervicals are greatly shortened almost to plates, which is elephantine again, though this short neck is approximated by Dipro- todon, some Amblypods and Arsinotherium, so that it must be in general looked upon as an adaptive feature, though in its detail it shows again an elephant character. The upper members of the limbs are longer than the lower, which is common to many massive animals. The humerus is tremendously flattened from front to back, even more so than in any of the animals used for compari- son, though flattening is a feature of them and of the elephants the most so. With the flattening, the deltoid ridge is prolonged enormously making a crest along the outer side of the bone, which at the lower end rises in a prominent process, as in elephants (also in Diprotodon but in this case the rest of the bone is very different). In addition to this, the supinator ridge is prolonged up- ward until it almost meets the deltoid, ending in a sharp spur at the top. This spur is more marked in Pyrotherium than in elephants, although they show the same develop- ment of the supinator ridge. The femur has the head much higher than the greater trochanter, which is a feature common to elephants, Diprotodon, Arsinotherium, etc., so that it must be looked upon as an adaptation. The third trochanter has disap- peared, and in elephants, it is lost in the advanced forms, remaining however as a trace in Palaeomastodon. The tibia is very short and massive and hardly gives any suggestions of relationship, except that it is not fused 1 62 THE DESEADO FORMATION OF PATAGONIA with the fibula at the upper end, in which it is in strong contrast to the toxodonts. While in the table of comparisons numbers I, 2, 3, 4, 8, 9, 10, 14, 17, 19, 20, and 21 may be, in part, or wholly, interpreted as adaptations, and alone would not be at all conclusive of relationship to elephants, numbers I, 5, 6, 7, 8, n, 12, 13, 15, 18, and 21 point toward the elephants as the close relatives of the Pyrotheria. In the first series of points there are none which mitigate against associating these two groups, while if the attempt is made to associate Pyrotherium with any group other than Proboscidea there are strong points, and a number of them, which would prevent this association. As a result of the foregoing, together with a feeling which continued handling of the specimens has given me, I can come to no other conclusion than that the Pyrotheria should be placed under Probos- cidea. In his Linea Filogenetica de los Proboscideos, Ameghino assigns to this suborder, or at least puts into the phylo- genetic tree, a considerable number of forms from the Casamayor beds, all of them genera with bunodont mo- lars, usually known by but one or two teeth, such as Asmith- woodwardi, Nephracodus, Cephanodus, Paulo gervaisia, and the better known genera, Carlo ameghinia, and Dido- lodus, all of which he makes ancestral to Pyrotherium. So far as known, however, these forms show none of the peculiarities of the Pyrotherium skull or dentition, so that it is difficult for me to see any reason for including them even in the suborder. The genus Carlozittelia, from the upper Casamayor, is in a different position, having an enlarged upper incisor (found isolated) and molars of the bilophodont type. I should include this in the family Pyrotheridae and none of the others. PYROTHERIUM 163 Pyrotheriidae Ameghino All the forms assigned to this family are supposed to be closely related to Pyrotherium and to have much the same structure. Ameghino has proposed the following genera, Pyrotherium, Parapyr other ium, Richardowenia, Archaeolo- phus, Propyr other ium. Parapyr other ium is based on a small molar and a tush which Ameghino first described as Pyrotherium planum, later elevating the species to a genus, designated as Para- pyrotherium, differentiated by the transverse crests being low and the valley at either end being blocked by an intertubercular ridge. Gaudry considered that this genus represented either the milk teeth of P. romeri, or a variation of that species. I can not see the basis of a new genus in the material. The genus Richardoweni is based on half of a molar, which has the transverse crest interrupted in the middle. Too little is known of this form to base a valid genus or even to associate it with Pyrotherium. Archaeolophus is founded on a small tush and part of an upper molar, also inadequate material for a genus. It is probably Pyrotherium. Pro Pyrotherium is a smaller form from the Astraponotus beds, apparently a good genus; the type species being P. saxeum, of considerable smaller size than any of the Deseado species. The distinctive features of the genus can not be given until more material is known. Carlozitteli is based on a small form from the Casamayor with narrow molars. An incisifoVm tush is associated with the molar, which, if correctly associated, would indicate a wide deviation from Pyrotherium, and would probably be an ancestral form. A second species is reported from the lower Deseado beds, but I am a little skeptical as to the horizon. 1 64 THE DESEADO FORMATION OF PATAGONIA Pyrotherium Ameghino Pyrotherium Amegh., 1889, Adas Acad. Nac. Cienc. Cordoba, t. VI, p. 617. Pyrotherium Lydekker, 1894, Anal. Mus. La Plata, Palaeontologia Argentina pt. 3, p. 4. Pyrotherium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 609. Pyrotherium Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 441. Pyrotherium Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. I, p. 19-43. Pyrotherium Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. i, p. 223-4. Pyrotherium Gaudry, 1909, Anal. Palaeontologie, t. 4, p. 1-28. Parapyrotherium Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. i, p. 29. The type species of the genus is P. romeri, which is however a rare species, most of the material and the best known belonging to P. sorondoi. The Amherst Collection contains a skull, complete except that the top of the brain case is crushed in and the parietals lost; a second skull with the full upper dentition but lacking the cranium; four lower jaws; two isolated tushes; the atlas, axis, and crevicals 3 and 4; the humerus; the proximal end of the femur; and part of the front foot; all from the Chico del Chubut west of Puerto Visser. Gaudry had upper and lower dentition and the fore and hind limbs except the feet. Ameghino described the upper and lower dentitions and a fore foot, so that with our material we now have a dasis for a fairly complete discussion, the vertebral column being the major part which is still lacking. The first striking feature is the dental formula. As formerly given, it is inaccurate, there being two great tushes on either side of the upper jaw, instead of one, as described. At first sight, I thought it might be a meristic variation, but both of my skulls show the same arrange- ment on both sides, and these are the first two skulls which have been found complete to the front end, and neither is by any means a young individual. The dental formula would then read ' 3 2 1 . Upper incisor I is a rootless, permanently growing tush about a fourth smaller than inc. 2, but of the same PYROTHERIUM 165 character, being oval in cross section and having enamel on the front face only. These first incisors are directed downward, so that their ends stand between and very slightly in front of the second incisors. The end of each is worn bluntly round in contrast to the beveled end of inc. 2. The second incisor is larger, rootless, and perma- nently growing, with a hollow base, enamel on the front face only, and oval in cross section. Both these teeth have a layer of cement on them, extending some distance beyond the alveolus. The tips are worn in a long bevel on the posterior side, very much as is the case on the incisors of rodents. The third upper incisor, the canine, and premolar I are lacking, a long diastema occupying their place, out of which they have been crowded by the development of the enormous root of inc. 2, which extends 150 mm. and more back into the jaw. P. romeri is distinguished from the others by pm. I being present. The teeth of the upper premolar-molar series have their crowns expanded, and the two series of either side have moved toward each other; until in front they are in con- tact while in the rear they are only 50 mm. apart, narrowing the palate in a unique manner, and giving the impression that the palate is mostly a grinding surface. The pre- molars are completely molariform and the whole series is at an advanced stage of specialization. Premolar 2 is three-rooted, with a triangular crown on which are three mamma-like tubercles, the larger one in front, and two behind. As the crown is worn, these unite into a flat, grinding surface, surrounding a central pit which opens behind. Premolars 3 and 4 and all the molars are large, four- rooted, quadrilateral teeth, each with two transverse crests running clear across the crown, and with a small cingulum across the anterior margin. Before they are worn, the top of each crest is tuberculated, and the cingulum is crenulated. In wearing, the anterior face 1 66 THE DESEADO FORMATION OF PATAGONIA of each crest is ground down; so that instead of the crown wearing to a level surface, it retains throughout life two oblique grinding surfaces. The lower dentition is more reduced than the upper. When in position, the tips of the two lower tushes diverge, so as to come in contact with the tips of the second upper tushes, from which I conclude that the lower tush is the second, rather than the first incisor, the latter having been lost when the second became enlarged as was the case in elephants. This lower tush has the same oval cross sec- tion, enamel on the front face only, and beveled tip as the corresponding upper incisor; but, in the same individual, is somewhat longer and slenderer. When isolated, how- ever, it is difficult to tell whether one is handling a small upper tush or a larger lower one. The remaining incisors, the canine, the first and the second pre molars are wanting, and their place is taken by a small diastema. The lower premolars and the molars are similar to those of the upper jaw, except the cingulum is on the posterior margin, and the wear is on the posterior face of each transverse crest. The skull is very long and narrow, with wide and deep zygomatic arches. The nasal opening is moved back from the front of the snout to just opposite the orbit, leaving a long, narrow, but heavy snout, made up mostly of the premaxillae, on the anterior end of which is an ovul boss, which must have seiVed as an attachment for muscles. With the tushes developed so as to bite against each other, as in a gnawing animal, I can not see any possibility for the development of a pendant proboscis, but think that the snout must have been developed more like that of a pig, but probably to a greater degree. The premaxillae are long and heavy, and prolonged backward to contain the roots of the great tushes; but these bones are not developed on the palatal side of the snout at all. The maxillae are also massive, carrying the premolars and molars, PYROTHERIUM 167 and extending forward to the bases of the tushes. They have developed downward so as to carry the plane of the palate far below the plane of the basicranium, and causing the upward bend in the basicranial axis, which is so characteristic of elephants. This bend leaves the occipital condyles a full foot above the plane of the teeth. The maxilla extends upward so as to bound the major part of anterior margin of the nasal opening, and of the orbit, which latter opening is small and directed forward. The zygomatic process is large and makes a considerable portion of the arch. The jugal is a broad flat bone making up most of the zygomatic arch and extending back so as to take a small part in making the glenoid fossa, as in elephants. The top of the brain case was crushed in before the burial of the animal, the anterior part being present, and about 40 mm. below its normal position, but the parie- tal region having been loose, exposed the brain cavity, the ear chamber and some of the cellular vacuities. The nasal bones are long and light in build, and are pushed back so that they lie between the postorbital processes of the frontals. The frontals were united medianly, and pro- longed on either side of the nasals to make the postorbital processes. The back margin of the frontals is broken away. The parietals are lost, but it is apparent that there was a short sargittal crest. From the middle, high lamb- doidal crests extend to either side, and become continuous with the upper margins of the zygomatic arches. The posterior face of the skull slopes back from the lambdoidal crests for a considerable distance, down to the moderate- sized foramen magnum. The squamosum is a large bone, with the lambdoidal crest and the extension of the zygomatic arch on its upper surface. It carries the major part of the glenoid fossa. Behind the auditory meatus is a large post-tympanic portion which extends down and unites with the pretym- 1 68 THE DESEADO FORMATION OF PATAGONIA panic portion, completely inclosing the opening of the ear and crowding the tympanic from being exposed on the side of the skull. There is a very short paroccipital process, and this posterior portion of the squamosum is the part which resembles that of elephants, hyracoids and, to some extent, Toxodontia. There is, however, no cavity in the squamosum as in toxodonts generally. The tympanic bulla is small, but little swollen, and hollow. It is quite exactly like that of probocideans. The basi- occipital is fused to the exoccipitals. The occipital con- dyles are very high above the plane of the teeth, are set wide apart, and are cylindrical bosses which would not allow a free movement of the head laterally, but only in the up and down direction. This last is again a feature of the elephants. The pterygoid bone is greatly enlarged to compensate for the bend in the basicranial axis, and the pterygoids, together with the alisphenoids, make broad plates bounding either side of the posterior nasal chamber, exactly as in Palaeomastodon. The palatal bones are slender in front, and broaden toward the rear, again, as in elephants. On the interior of the brain case is the cavity for the brain which indicates that this organ was of diminutive size, measuring about 150 mm. in length by 50 mm. in width at the widest part. It indicates a brain with very small cerebral hemispheres, which, however, had a swollen posterior margin, a larger cerebellum, and a wide medulla oblongata. The impression which I obtained of this brain is strikingly like that given for Palaeomastodon. On either side of the brain cavity are a couple of vacuities, apparently for lightening the weight of the skull. At the inner end of the auditory meatus is a large ear chamber, divided into a smaller anterior or cochlear portion, and into a larger posterior ear chamber proper. In figure 109, I have placed a diagram of the base of the skull of Pyrotherium, along beside that of Palaeomas- todon, for comparison of the basicranial foramena. The PYROTHERIUM 169 A 1 70 THE DESEADO FORMATION OF PATAGONIA skull of Palaeomastodon is somewhat more elongated, especially in the posterior part. In both, there are two antorbital foramena; the postpalatal foramena of Pyro- therium are a trifle further back, but this palatal region in both is of the same type which is peculiar to elephants and Pyrotherium. In Pyrotherium the condylar foramen is separate, while in elephants it is fused in with the fora- men lacerum posterior. This latter foramen in both cases is situated just back of the tympanic, and in Pyrotherium is of considerably larger size than in Palaeomastodon. The foramen lacerum medium is in front of the tympanic and in Pyrotherium appears considerably larger, mostly because it is under the margin of the tympanic in Palaeo- mastodon. The foramen for the internal common carotid in Palaeomastodon pierces the tympanic bone just to the inside of the middle line, while in Pyrotherium it is on the outer margin of the tympanic. The Eustachian canal is on the external border of the tympanic in both cases, but in Pyrotherium it is further back. The foramen ovale of Palaeomastodon is in the posterior part of the alisphenoid bone, but with the shorter alisphenoid of Pyrotherium^ this foramen is pushed back to the posterior margin of the bone. In both cases, the alisphenoidal canal starts under the base of the fused alisphenoid and pterygiod, and opens into the orbit. The stylornastoid foramen of Pyrotherium is situated further out than in the case of Palaeomastodon. The fusion of the postympanic portion of the squamosum is, in Palaeomastodon, much further advanced than in Pyrotherium, so that the passage to the ear is not apparent in the basal view of the former, but makes a considerable notch on the under side of the skull of Pyrotherium. The mandibles are excessively thick and heavy, being united at the symphysis, which extends back to the front of the second molar. The ascending rami are prolonged back- ward, but do not rise above the level of the articulation. PYROTHERIUM 171 The atlas is a massive vertebra with the anterior cotyles deeply excavated, especially on the upper side, so that, as Gaudry suggested, the head must have been carried low. The flat posterior cotyles face obliquely downward. The neural arch is light and without a spine or an opening for the vertebral artery. The basal portion of the bone, how- ever, is excessively heavy and thick; the socket for the odontoid process not reaching to the middle of the basal bar. The neural canal is oval in section, being a good deal wider than high, and of small size. The transverse proc- esses are short, heavy projections, adapted to receive heavy muscles. On the ventral surface there projects from the posterior margin a strong hypophysis, whiqh, as Gaudry has pointed out, is unusual, but which is a character of the atlas of the Palaeomastodon. The axis is a short, heavy bone, with the anterior cotyles facing obliquely upward, a small neural arch, no spine, and with a thick odontoid process, which has the form of a quarter of a hemisphere set onto the front of the centrum. Cervicals 3 and 4 are very short vertebrae with light neural arches and no spines. The neural canal is fully three times as high as wide. Thus it is entirely evident that the neck of Pyrotherium was extremely short, as is the case with elephants, which alone would not be sig- nificant, but coincides with many other elephant features. Gaudry described a lumbar vertebra which is also a short, heavy bone. Otherwise the vertebral column of Pyro- therium is unknown. The distal end of the scapula is described by Gaudry as indicating a short, heavy bone, with the glenoid cavity compressed so as to be about twice as long as it is wide. The coracoid is a short, blunt process. The spine was broken off, but enough remained to indicate a moderately high spine, prolonged toward the humerus, and bent some- what forward. 172 THE DESEADO FORMATION OF PATAGONIA The humerus is a very characteristic bone, short and stout, but greatly flattened from front to back. It has a large sessile head, which is strongly convex, and projects internally over the margin of the shaft. The external tuberosity is large and rugose but does not project above the level of the head. The deltoid ridge is shifted to the external side of the bone, and makes a long, muscular ridge, while on the opposite external margin is a second ridge, and between the first and second ridges a long furrow or trough is inclosed. These terminate just below the middle of the bone in roughened bosses, which all but meet. The epicondyles are large and give the excessive width to the bone. The external condyle is prolonged upward and ends in a spur. The trochlea is of moderate width and gently undulated. The supratrochear fossa is only slightly depressed, and the anconeal fossa is likewise shallow. The bone has no exact counterpart, but is simi- lar to that of Moeritherium and Palaeomastodon, but in each case is more flattened and has the external ridges more developed. Gaudry describes the radius and ulna. They are ridiculously short, and very massive. The ulna is stout with a massive olecranon which is directed well toward the rear. The sigmoid notch is shallow, the coronoid process short, and the articular area expanded so that the ulna covers the whole of the posterior of the trochlea of the humerus. The upper end of the radius is compressed antero-posteriorly, but distally it expands into a heavy bone. Its upper articulation is expanded, so that it comes in contact with the full width of the anterior portion of the trochlea of the humerus. The carpus and front foot are of questionable associa- tion. Ameghino described a front foot as P. romeri, and later Tournouer assigned this foot to Astrapotherium. However, I have seen no reason to think it belongs to Astrapotherium, being far too small, and so would for PYROTHERIUM 173 the present consider it as belonging to Pyrotherium. We found a couple of metapodials evidently belonging to the foot as described. This carpus is of the primitive type, the sca- phoid and luna being large and receiving the radius; while the pyramidal is smaller, low and broad and received the ulna. The trapezium is larger than usual, being elongated and standing out from the trapezoid, and support- t Fig. no. Left carpus and metacarpus, ing a reduced first metacarpUS. outlines after Xournouer 1/5 natural The trapezoid is also large and almost square in outline. The magnum is smaller and con- siderably flattened. The unciform is very large. These last three mentioned carpals carry the three medium meta- carpals which are quite normal and seem to have carried most of the weight of the animal. Metacarpus V articulates on the outer side of the unciform. It is a massive nodular bone with but a tiny articulation for the phalanx, which seems on this toe to have been reduced. Metacarpals IV, III, and II are short, stout bones, flattened from front to back, and enlarged at either end. On each, the trochlea extends well onto both the dorsal and palmar surfaces, thus giving the toes a considerable range of movement, and indicating at least a semidigitigrad mode of walking. Of the pelvis, the ilium is known as a broad, heavy bone with the acetabulum facing down. The hind limb is considerably longer than the front, and approximately pillar-like. The femur, as compared with the humerus, is quite a little longer, though, as femora go, it is not a long bone. The rounded sessile head stands high above the blunt, thick greater trochanter; the digital fossa is barely indicated; the rotular trochlea is short; the two condyles are subequal in size and set close together. The patella is short and nodular. The tibia is short and very 1/4 THE DESEADO FORMATION OF PATAGONIA heavy. The fibula is free its entire length and is a rather heavy bone. The astragulus is a lens-like bone with the trochlea but slightly convex, and the navicular facet directly below it, indicating a rectigrade foot. Ameghino established the following species, P. romeri, P. sorondoi, P. giganteum, P. crassidens, P. trilophodon, P. pluteum. This is a very considerable number of species of such a large type to occupy a limited region. Gaudry has lumped them all under species P. romeri. This, I think, is too drastic and T would find at least two species. It is true that there is great individual variation in such large animals, due to age, food supplies and individual vicissitudes; but where there is a difference of dental formula or a structural divergence I should consider these as specific in character. The type species is P. romeri (later spelled romeroi) which was based on a first and second upper premolar and an upper tush, all of smaller size than P. sorondoi and differ- ing from all the others described in having pm. I present. Gaudry suggests that this may be the milk dentition but there is no evidence as yet to settle this, so I have left this species standing. Most of the material found by Ameghino, by Gaudry and by our party belongs to the type described as P. sorondoi, which is somewhat larger than P. romeri, and lacks pm. I in the upper jaw. This then is the usual species and to it belongs most of what is known. It varies some in size but the characters are very uniform. P. giganteum is based on the root of a large tush, 90 by 70 mm. in cross section, which Lydekker associated with P. romeri, and which Ameghino later took as the type of a new species. I can see in this only a large individual of P. sorondoi. P. crassidens is based on a last lower molar 90 by 80 mm. in diameter. It seems to me to be an upper molar and no larger than m. 2 in either of my skulls. P. trilophodon is based on a lower pm. 3, which, in every way, resembles the corresponding tooth PYROTHERIUM SORONDOI 175 in lower jaw of P. sorondoi. P. pluteum is based on three lower teeth of smaller size than the typical P. sorondoi, but the difference is small and there are no structural features accompanying it; so I consider it simply a smaller individual of P. sorondoi. In the generic discussion, Parapyr other ium planum was also assigned to P. sorondoi. Pyrotherium romeri Ameghino P. romeri Amegh., 1889, Act. Acad. Nac. Cicnc. Cordoba, t. VI, p. 618. P. romeri Lydekker, in part, 1894, Anal. Mus. La Plata, Palaeontologia Argen., t. Ill, supplement, p. 4. P. romeroi Amegh., 1894, Bol. Inst. Geog. Argen., t. 15, p. 612. P. romeroi Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 442. P. romeroi Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. i, p. 32. P. romeri Gaudry, in part, 1909, Anal. Paleontologie, t. 4, p. 21. This species is characterized by the presence of pm. I which is absent in other species. The tooth is of fair size, two-rooted, narrow in front and has a narrow rim of enamel around it; and measures 22 mm. long by 14 mm. wide. The second premolar is 30 mm. long by 33 mm. wide. A lower tush is also associated with these two teeth and is of smaller size than in the following species, being at the alveolus border 40 mm. by 29 mm. in cross section. Pyrotherium sorondoi Ameghino P. sorondoi Amegh., 1894, Bol. Inst. Geog. Argen., t. 15, p. 613. P. sorondoi Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 443. P. sorondoi Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. I, p. 30. P. sorondoi Amegh., 1906, Anal. Mus. Nac. Buenos Aires, ser. 3, t. 8, p. 331. P. romeri Lydekker, in part, 1894, Anal. Mus. La Plata, Paleontologia Argen- tina, t. Ill, supplement, p. 4. P. romeri Gaudry, in part, 1909, Anal. Paleontologie, t. 4, p. 21. P. giganteum Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 447. P. crassidens Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. i, p. 34. P. planum Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 446. Parapyrotherium planum Amegh., 1902, Anal. Mus. Nac. B. A., ser. 3, 1. 1, p. 29. P. trilophodon Amegh., 1902, Anal. Mus. Nac. B. A., ser. 3, t. i, p. 33. P. pluteum Amegh., 1901, Bol. Acad. Nac. Cienc. Cordoba, t. 16, p. 386. P. pluteum Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, 1. 1, p. 29. The species varies in size and in the proportionate development of the tushes as would be expected in a large 176 THE DESEADO FORMATION OF PATAGONIA Fig. in. Top of the skull 1/5 natural size; Mate to in i on the left side is represented by an alveolus but the tooth had fallen out before the death of the animal; a.c., ear chamber; B, brain case; V, vacuities in the bone. PYROTHERIUM SORONDOI 177 1 78 THE DESEADO FORMATION OF PATAGONIA and slow-growing animal. Most all of the material found by any of the collectors falls clearly into this species. Our specimens all came from the Chico del Chubut, west of Puerto Visser, which point is about 250 miles north of the locality where Gaudry's material was found. The general features of the skull have been given under the generic description. The difference between this and the preceding species lies in the absence of pm. i, and the somewhat larger size. Both of our major speci- mens are mutilated in places; and as the better skull was found with only the zygoma tic arch exposed, we conclude that the parts missing were lost before burial. On the lower jaw the edges of some of the teeth were cracked off, and both the ascending rami are lacking, both things having happened before burial, this specimen being found well in the bank and never exposed to weathering. It appears as if the carcasses of the animals lay some time before being buried by the sediments. The scattered condition of all the finds indicates the same thing. The head of our femur is still marked by the teeth which cleaned the meat from the bone. The frequency of isolated tushes indicates that many jaws originally containing them have been either chewed to pieces or weathered away before burial. I do not think all the tushes found originally belonged to the lower jaw (Gaudry reports 18 tushes, all lower); for the upper and lower tushes when isolated are so much alike that it is difficult to distinguish them. The size of the skull is indicated by the following measure- ments: SPECIMEN No. 3207 Length from inc. I to occipital condyles 720 mm. Length from front of premax. to nasal opening 225 mm. Length of boss on premaxilla 77 mm. Length of nasal opening at top 80 mm. Width of nasal opening at top 88 mm. Length from premax. to lambcloidal crest 540 mm. Width across zygomatic arches 35 mm - Width across frontal bones 90 mm. Transverse diameter of the snout 115 mm - PYROTHERIUM SORONDOI 179 Fig. 113. Base of the skull i/S natural size; i.i, incisor the right side grown over toward the center as its mate is wanting; i.2, in- cisior 2; Pal., palatinum; Ft., pterygoid. l8o THE DESEADO FORMATION OF PATAGONIA The lower jaws were associated with the above skull, and are complete to behind the third molars on both sides. They are very short and heavy, especially in the anterior portion, the symphysis extending back to opposite the middle of the second molar. The height under molar 3 is 150 mm. See frontispiece. From the upper dentition, I give the measurements of my two chief specimens, together with those given by Ameghino for his type of P. sorondoi, and the figures given by Gaudry; from which may be seen the amount of variation which individuals may show. UPPER DENTITION SPECIMEN SPECIMEN AMBGHINO'S GAUDRY'S No. 3207 No. 3250 TYPE SPECIMEN Total length, inc. I to m. 3 530 mm. Inc. I, length above alveolus 133 mm. Inc. i, antero-postcrior diam. 49 mm. Inc. I, transverse diam. 37 mm. 42 mm. Inc. 2, length above alveolus 174 mm. Inc. 2, antero-posterior diam. 59 mm. 77 mm. 65 mm. Inc. 2, transverse diam. 40 mm. 52 mm. 41 mm. Premolar 2, length 45 mm. 52 mm. 48 mm. 40 mm. Premolar 2, width 36 mm. 40 mm. 30 mm. 29 mm. Premolar 3, length 46 mm. 49 mm. 48 mm. 40 mm. Premolar 3, width 57 mm. 57 mm. 46 mm. 48 mm. Premolar 4, length 47 mm. 51 mm. 46 mm. 43 mm. Premolar 4, width 63 mm. 64 mm. 58 mm. 57 mm. Molar I, length 55 mm. 55 mm. 57 mm. 55 mm. Molar I, width 68 mm. 69 mm. 61 mm. 61 mm. Molar 2, length 68 mm. 77 mm. 70 mm. 57 mm. Molar 2, width 85 mm. 93 mm. 75 mm. 68 mm. Molar 3, length 75 mm. 68 mm. 83 mm. 64 mm. Molar 3, width 86 mm. 87 mm. 82 mm. 88 mm. For the lower dentition, I give the figures which Ame- ghino records for his P. sorondoi, those given by Gaudry for his P. romeri, and the measurements of specimen No. 3207. The tushes in the lower jaw I would designate as incisors 2 ; because when the jaws are closed these diverge and their tips bite against the tips of the upper incisors 2. The first incisor seems to have been gradually lost and PYROTHERIUM SORONDOI 181 no space left for it in the front of the mandible, just as it was reduced and lost in the development from Moeritherium to Polymastodon or equivalent types. Tm.9 Hit Fig. 114. Lower dentition 1/5 natural size; edges of teeth broken off in my specimen are indicated in outline. LOWER DENTITION Incisor 2 to molar 3, length Premolar 2 to molar 3, length Inc. 2, length above alveolus Inc. 2, antero-posterior diam. Inc. 2, transverse diam. Premolar 3, length Premolar 3, width Premolar 4, length Premolar 4, width Molar i, length Molar i, width Molar 2, length Molar 2, width Molar 3, length Molar 3, width Four cervical vertebrae were preserved with the skull number 3207, of which only about a third of the atlas is represented, but fortunately we found a complete atlas isolated and of the same size. The measurements for the atlas are taken from this separate specimen. While my skull, especially the teeth, seems to have been larger than the skull Gaudry described, the cervicals are a little smaller. I give the measurements of the cervicals which we found, comparing them with the figures given by Gaudry. Figures taken from illustrations. 1 82 THE DESEADO FORMATION OF PATAGONIA Atlas, antero-posterior length (without hypophysis) Atlas, greatest width Atlas, height Axis, antero-posterior length Axis, greatest width Axis, width of odontoid process Axis, length of odontoid process Cervical 3, length of centrum antero-posterior Cervical 3, width of centrum Cervical 3, height of centrum Cervical 3, width of neural canal Cervical 3, height of neural canal Cervical 4, length of centrum antero-posterior Cervical 4, width of centrum Cervical 4, height of centrum 47 mm. 45 mm 125 mm. 145 mm 105 mm. 100 mm 75 mm. 22 mm. 45 mm. 132 mm. 105 mm. The humerus is flattened from front to back in a striking manner, so that, seen from the side, it looks most slender; while in reality it is a very broad bone, nearly straight, and with marked rugosities for the attachment of the muscles. We found but one specimen of the humerus, an isolated bone a little smaller than that described by Gaudry. SPECIMEN GAUDRY'S No. 3218 SPECIMEN Humerus, total length 470 mm. Humerus, width at proximal end 232 mm. Humerus, width at middle of shaft 165 mm. Humerus, antero-posterior diam. of shaft 61 mm. Humerus, width across epicondylcs 230 mm. 240 mm 500 mm. 232 mm. 170 mm. We found neither the radius nor the ulna, but Gaudry figures both, giving the following measurements. Radius, length 245 mm. Radius, proximal diam. 127 mm. Ulna, length (calculated) 280 mm. Ulna, width of olecranon 140 mm. PYROTHERIUM SORONDOI At. Fig. 115. At., atlas; Ax. .axis; .3, third cervical; C. 4, fourth cervical i/S natural size; apophysis of atlas is restored after Gaudry and the neural arch of cervical 4 is restored from the opposite side. Fig. 116. Axis i/5 natural size, front view. Fig. 117. Anterior face of the humerus i/S nat- ural size. 1 84 THE DESEADO FORMATION OF PATAGONIA For the hind limb I give some of the figures which Gaudry gives accompanying his illustrations of the hind limb. Femur, length 630 mm. Femur, greatest proximal diameter 240 mm. Femur, distal diameter 170 mm. Tibia, length 370 mm. Tibia, greatest proximal diameter 164 mm. Tibia, greatest distal diameter 114 mm. Astragulus, antero- posterior diam. 123 mm. Astragulus, transverse diam. 114 mm. Astragulus, height 65 mm. CHAPTER XIII RODENTIA WHILE all of small size, numerically the rodents make about a third of our collection, the number of genera and species being, however, relatively small. All are hystricomorphs with the pattern on the crowns of the teeth relatively simple. While the incisors are typically rodent-like, permanently growing teeth, the molars are all rooted, some being entirely brachydont, others begin- ning to show hypsodont features. So far as yet known, the rodents make their first appear- ance in South America, in this Deseado formation. Were they, as Ameghino thought, developed there from such a form as Propolymastodon or Promysops of the Casamayor formation? Or did they migrate into Patagonia from some other section? For the former proposition to be convincing to me, it would require more complete material of the forms suggested than now exists.* Other groups of hystricomorphs occur in the Theridomyidae of the European Oligocene, and from the Oligocene of the Fayum.f Either the old world forms are descended from the South American forms, or vice-versa. The two African lower jaws are very much like those of Cephalomys, and my feeling is that the Patagonian forms are derived from some immigrant reaching that section before Deseado times. The Deseado genera are not widely different from each other, but it is evident that they are the representatives of at least two families, and my expectation is that other families will be found eventually to be already represented. * I have a lower jaw of Propolymastodon which, though not complete in front, gives me no suggestion that the incisor was rodent-like, and I am inclined to think that the incisor associated with the type of P. carlo-zitelli is a mistake. fOsborn, Bui. Amer. Mus. Nat. Hist., Vol. 24, p. 265. 1 86 THE DESEADO FORMATION OF PATAGONIA Our material does not permit the discussion of the skeleton or even of the skull as a whole, for the specimens occur only as isolated jaws, palates, or even as isolated teeth. In a few cases, the upper and lower dentitions are associated, but in no case was skeleton material clearly associated with the teeth. The remains look very much like such as are often found today in the western United States under a hawk's nest or below the roosting place of owls. I think most of our specimens passed, before burial, through the stomach of birds or carnivors. Ameghino puts most of the forms in the family Cepha- lomidae, which he considers ancestral to Hystricomorpha in general. I feel, however, that it is better to assign the Deseado genera to the families which have persisted until recent times, as Scott and Ameghino, in another place, have done. There are six living families, four of which Scott found already represented in the Santa Cruz. Two of these clearly may be continued back into the Deseado, the Erethizontidae, and the Chinchillidae, nothing as yet having been found to represent the Santa Cruz families Cavidae and Octodontidae. Chinchillidae In the Deseado, this family is represented by the genera Cephalomys, Scotamys, and possibly Litodontomys. Cepha- lomys is very abundant and seems to be ancestral to Peri- mys of the Santa Cruz; Scotamys is relatively rare but seems to be ancestral to Scotaeumys; while Litodontomys is also rare and as far as I can see without a successor. Cephalomys Ameghino Cephalomys Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 494. This is the common genus of the Deseado, over three- fourths of the specimens of rodents found belonging to one of its three species. Its dental characters mark it CEPHALOMYS 187 clearly. All the premolars and molars are rooted, though the crown is incipiently hypsodont, as much so as in any rodent of this period. The incisors are moderately large with the anterior face slightly convex, and the antero- posterior diameter comparing with the transverse diameter as 3 does to 2. The interval between the incisor and premolar 4 is moderate, indicating a short snout. Each lower molar consists of two transverse laminae separated from each other by an internal and an external infolding, both of which approach the median line but do not meet, a narrow, longitudinal bar separating the folds and connecting the anterior and posterior laminae. On the inner side, the posterior lamina has a furrow extending to the middle of the tooth, but only sinking into the crown about a fourth of its height, so that, with wear, it appears first as a bay, later as a pit, and finally disappears. In general it will be found only on molar 3, and may be wanting there on old individuals. On an unworn tooth, there occurs, on the inner side of the anterior lamina, a rudimentary pit corresponding to the one .on the posterior lamina, but of much less depth, so that it is only occasion- ally seen, and that only on a very slightly worn tooth. The premolar differs from the foregoing in having a small median column on the anterior face of the anterior lamina. In three cases we found the deciduous fourth premolar (see fig. H9A), a complicated tooth, consisting primarily of three laminae in which furrows have developed until there are four folds or furrows on the internal side, sepa- rating five crests; while on the external side there are three furrows and four crests. Ameghino's figure of this tooth in C. prosus has four laminae running clear across the tooth. I think the difference is due to his having an unworn decid- uous premolar whereas mine are all worn considerably. At first glance, the upper teeth appear strikingly differ- ent, resembling those of Perimys to which genus they are 1 88 THE DESEADO FORMATION OF PATAGONIA probably ancestral. Each molar consists of two laminae, separated by a deep internal fold which extends almost to the external margin. On little worn teeth each lamina shows, on the external side, a shallow furrow extending to about the middle of the tooth, but these furrows early become pits and then disappear with further wear, being preserved on not over a fourth of our specimens. The fourth upper premolar consists of two laminae, but in this case, the separating fold is on the external side and extends nearly to the internal margin, so that this tooth appears to be reversed in its position in the jaw. As in the molars, there is, on. the external side of either lamina, a furrow, the one in the anterior lamina shallow and seldom seen, that in the posterior lamina deep and present in all but the most worn teeth. While the upper and lower molars appear so different they may be readily derived from such a tooth as the lower molar, as both have the two laminae and separating furrows in common. In the upper molars, however, the internal fold is prolonged until the external fold is merely indicated or -lacking. On upper premolar 4, on the con- trary, it is the external fold which is prolonged. The fur- rows in the external portions of the laminae of the upper molars correspond to those on the internal portions of the same laminae of the lower teeth, reversed, as is typical of all teeth. Ameghino distinguished three species of Cephalomys, which are based primarily on size, the other characters which he gave being inconstant. We found these three and no others. C. arcidens C. plexus C. prosus CEPHALOMYS ARCIDENS 189 Fig. 118. Right upper premolar, molar series x 4/1. Cephalomys arcidens Ameghino C. arcidens Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 494. This, the type species, is by far the commonest of the rodents, in fact of all the species in the Deseado, and we found forty-seven specimens on the Chico del Chubut River, west of Puerto Visser. In the species there is con- siderable variation in size for a rodent, but as there are intermediate specimens all the way between the extremes, and as the variation is mostly in the size of the fourth premolar, it does not seem proper to separ- ate the material into more than one species. In general, the form has relatively plump teeth, relatively heav- ier anct~ thicker than in the other species. Usually the fourth premolar is but little larger than the molars, Fig. 119. Left lower premolar, molar series; A, decodu- l-tnf- iti fliic r-Viot-a/^foi- ous premolar 4; B, a little worn mola i ; C, series about l JL 1U l Cl ' ^^^^^i^^^^tS&l^^. there is considerable variation. The fol- lowing measurements give the range of size on the upper jaws: Upper premolar 4 to m. 3 Upper premolar 4, length Each molar, length Each molar, width 190 THE DESEADO FORMATION OF PATAGONIA The lower jaw is low, heavy and rather short, the posterior part of the ramus being very thin, while the portion carrying the teeth is thick and heavy. A strong ridge extends along the inner side from just behind molar 3 to the base of the symphysis. As in the upper dentition, there are smaller and larger forms. SPECIMEN SPECIMEN AMEGHINO'S 3089 3058 TYPE A SMALL A LARGE INDIVIDUAL INDIVIDUAL Lower premolar 4 to m. 3 Lower incisor I to pm. 4 Height of mandible under pm. 4 Length of deciduous pm. 4 Cephalomys plexus Ameghino C. plexus Amegh., 1897, Bol Inst. Geog. Argen., t. 18, p. 494. In general, this species is similar to the foregoing, but is smaller in size and slenderer in proportions. Both the upper and lower fourth premolar tend to be considerably larger than the molars. The species was not nearly as abundant as C. Fig. 120. "Right paiaiate showing P re- arcidens, occurring but sixteen molar, molar series; external furrows i . appear as pits on molars 2 and 3, x 4/1. timCS in OUr Collection. Fig. 121. Left mandible, external side, x 4/1. CEPHALOMYS PLEXUS 191 MEASUREMENTS Upper dentition, pm. 4 to m. 3 Upper dentition, pm. 4, length Upper dentition, each molar length Upper dentition, each molar width SPECIMEN 3091 9. mm. 2.75 mm. 2.1 mm. 2.25 mm. AMEGHINO 's TYPE 9.5 mm. The lower jaw is slender, the incisor being relatively both smaller and slenderer than in C. arcidens; the back part of the ramus light and thin, the coronoid process being a tiny spur, and the articular condyle of small size, and on a level with the teeth. MEASUREMENTS Lower dentition, pm. 4 to m. 3 Lower dentition, in. i to pm. 4 Lower dentition, pm. 4, length Lower dentition, each molar, length Lower dentition, each molar, width Height of mandible under pm. 4 int. ext. Fig. 122. C. plexus, left lower premolar, molar series; A, of young individual; B, of old individual; inf., internal side; ext., external side, X4/I. SPECIMEN AMEGHINO'S 3005 10 mm. 6 mm. 3-5 mm- 2.5 mm. 2 mm. mm. TYPE 10.5 mm. 6.5 mm. 4.8 mm. Fig. 123. C. prosus, left upper premolar ,'_molar series, x 4/1. Fig. 124. C. prosus, premolar 4 and molar 2, x4/i. Cephalomys prosus Ameghino C. prosus Amegh., 1902, Bol. Acad. Nac. Cienc. Cordoba, t. 17, p. 37. This is the tiniest species of the genus, and least fre- quently found, probably because on account of the small size it was more frequently destroyed before burial, and also because it is hard to find such tiny specimens; so that 1 92 THE DESEADO FORMATION OF PATAGONIA the sixteen which we found would hardly represent the real proportion of the species in the fauna. The jaws are not only small, but also slender and deli- cately built, with the premolar about the same size or slightly larger than the molars. The drawings represent the proportions accurately so I will give but a few measure- ments. SPECIMEN AMEGHINO'S 3009 TYPE Upper premolar 4 to molar 3 8,5 mm. Lower premolar 4 to molar 3 95 mm. Scotamys gen. nov. A lower jaw, with premolar 4 and molars I and 2, from the Deseado beds on the Chico del Chubut River, west of Puerto Visser, indicates a genus of hystricomorph rodents not previously reported. The lower molars suggest those of Perimys, but premolar 4 is similar to that of Scotaeumys from the Santa Cruz. Scotamys differs, however, from Scotaeumys, in that its molars do not have the third lobe found in the Santa Cruz genus. I have, therefore, made a new genus which appears to be ancestral to Scotaeumys. Scotamys antiquus sp. nov. This, the type species of the above genus, is based on specimen 3063, a lower jaw with the incisor, premolar 4 and the first two molars. The incisor is fairly large and heavy, the anterior face slightly convex, and the anterio- Fi g .T 2S . Lo^er pTei^iar 4 posterior diameter greater than the transverse diameter. Premolar 4 has a deep external fold, dividing the crown into an anterior and posterior lamina, the former being then subdivided by another external fold, making the tooth three-lobed. Just internal to the median fold is a tiny pit, apparently the last vestige of an internal fold. Each molar consists SCOTAMYS 193 of two laminae separated by a deep external fold, around the inner end of which the laminae are connected by a narrow bar. In the present condition of wear there is no indication of secondary furrows. The premolar is smaller than the molars. MEASUREMENTS SPECIMEN 3063 Lower dentition, in. I to pm. 4 8 mm. Lower dentition, premolar 4 length 3 mm., width 2.5 mm. Lower dentition, molar I length 2.5 mm., width 2.5 mm. Lower dentition, molar 2, length 2.75 mm., width 3 mm. Height of mandible under pm. 4 5.5 mm. Fig. 126. Left mandible external side, x 4/1. Litodontomys gen. nov. One set of lower teeth found by the Amherst party shows a simplicity of pattern found in no other genus of South America; and this is, therefore, named Litodontomys. The teeth are brachydont, the premolar and the molars each being divided into two laminae by an external and an internal fold, the distinctive generic feature being in that this fold is narrowest at the margin of the tooth and expands internally. In connection with the expanded folds, the ends of the laminae are curved toward each other, so that in a worn specimen they would meet on the margins of the tooth, and leave the folds to appear as pits. No indication of a furrow is evident on either lamina. 194 THE DESEADO FORMATION OF PATAGONIA Litodontomys chubutensis sp. nov. The type is number 3086 of the Amherst collection, from the Deseado beds on the Chico del Chubut River, west of Puerto Visser, and consists of the lower premolar-molar series. Premolar 4 is elongated, the Fig. 127. Right jower / prcmoiar 4 -moiar anterior lamina being consider- ably longer than it is wide, whereas the laminae of the other teeth are wider than they are long. The following measurements with the figure give the specific details. Lower dentition, premolar 4, to molar 3 10.5 mm. Lower dentition, premolar 4, length 3.5 mm. Lower dentition, molar I, length 2. mm. Lower dentition, molar 2, length 2.5 mm. Lower dentition, molar 3, length 2.5 mm. Lower dentition, width of molars 2 . mm . ERETHIZONTIDAE Asteromys Ameghino Asteromys Amegh., 1897, Bol. Inst. Geog. Argcn., t. 18, p. 495. The genus contains small forms, with brachydont teeth. The upper molars consist of two laminae, separated by an internal and an external fold, and each lamina having, on the internal half, a deep furrow, which is but little shallower than the median fold; so that the outer side shows three furrows, folds, or pits, more or less completely separating four lobes; while on the inner side of the tooth there are but two lobes. On the lower teeth the external median fold is deep, while the internal median fold is shallower, usually appearing as a pit. Both the anterior and posterior laminae are subdivided by wide internal furrows which extend to the median line. These characters associate the genus with Acaremys of the Santa Cruz, from which genus it is not easy to separate ASTEROMYS 195 Asteromys; but as we know only the teeth from the Deseado beds, it is probable that, when the skull is found, larger differences will be recognized. Asteromys appears to be the direct ancestor to Acaremys. The species are all tiny, the following three being dis- tinguished by Ameghino. LENGTH OF LOWER MOLAR SERIES A.punctus (Bol. Inst.Geog. Argen., 1897, t. 18, p. 495) 12 mm. A. annectens (Bol. Acad. Nac. Cienc. Cordoba, 1902, t. 1 7, p. 37) 1 1 mm. A. prospicuus (Bol. Inst. Geog. Argen., 1897, t. 18, p. 495) each molar 1.6 to 1.8 mm. Of these three we found only the last. Asteromys prospicuus Ameghino A. prospicuus Amegh. loc. cit. above. The species is rare, only three specimens turning up in our collections. The upper molars are as described in Fig. 128. Right tipper premolar 4 Fig. 129. Left lower molar 3, x 4/1. molar 2, x 4/1. the generic discussion, but premolar 4 is simpler than the molars, the posterior lamina being small and without any sort of furrow. In the upper molars the anterior lamina is larger than the posterior, and the anterior furrow wider than the posterior. The following measurements, with the figures, indicate the character of the species. Upper dentition, premolar 4 to molar 3 8.75 mm. Upper dentition, premolar 4, length 2. mm. Upper dentition, each molar, length 2.3 mm. Upper dentition, each molar, width 2 . mm. Lower dentition, molar 2, length 2 -75 nun. Lower dentition, molar 2, width 1 . 75 mm. 196 THE DESEADO FORMATION OF PATAGONIA Eosteiromys Ameghino Eosteiromys Amegh., 1902, Bol. Acad. Nac. Cienc., Cordoba, t. 17, p. no. The genus was established by Ameghino for forms similar to Steiromys of the Santa Cruz, but antedating that time. I confess I can see but little difference between Steiromys and Eosteiromys, but the latter is as yet known only from isolated teeth and as in general it would be expected that there should be a generic difference, we may let this genus stand representing rather a prophecy than the facts as yet known. The upper teeth are brachydont, the crown being on the same general plan as in the foregoing genus, i. e., it is divided into an anterior and posterior lamina by a deep external median fold and by a shorter oblique internal me- dian fold. The anterior lamina is subdivided by two ex- ternal furrows, a lesser anterior and a larger posterior; while the posterior lamna is subdivided by a single external furrow; so that this tooth has four folds, furrows, or pits on the external side separating five lobes; while on the inner side there is but the one oblique fold separating two lobes. Eosteiromys medianus Ameghino ? E. medianus Amegh., 1902, Anal. Mus. Nac. Buenos Aires, ser. 3, t. 2, p. 129. The genus is based on a single, though entirely characteristic, upper molar. We found just one tooth of the species, also an upper molar. It is described above, under the genus. The measurements are : Upper Fig moL R 2 i8 xV/T Per molar 2 > ^ngth 4 mm., width 5 mm. CHAPTER XIV EDENTATA THE scarcity of edentates in the Deseado beds is in striking contrast to their abundance in the Santa Cruz formation. Whereas in this latter horizon over half of the finds are edentates, in the Deseado only eight per cent, of the total collection belong to this group, and this is doubt- less a larger proportion than these animals jepresented in the fauna; for the hundreds of small plates in a carapace, when scattered greatly, increase the chance that some part of an individual will be found, and most of the eight per cent, of finds are single plates. Most of the plates found represent armadilloes, our collection containing but one plate of a glyptodont, and no gravigrades. Ameghino's collections present about the same relations, but in the repeated trips he found a few more traces of glyptodons and a very few gravigrades. This scarcity of edentates can not be taken to mean that they were not developed, for they are a peculiarly South American group, and as they were developing somewhere into their great complexity, I take it to mean that the climatic conditions were unfavorable in this particular section. As noted above, all previous finds have been isolated plates. We were fortunate enough to find one specimen consisting of a carapace with ten rows of movable plates in place, and parts of four rows of the pelvic buckler to- gether with over fifty isolated plates. A second specimen had some fifty associated plates which were mostly from the pelvic buckler. Dasypoda The representatives of this group are so poorly known in the Deseado beds that Ameghino has, in general, used 198 THE DKSRADO FORMATION OF PATAGONIA the generic names of the Santa Cruz for their description, and, so far as known, they are little differentiated from those genera. There is as yet no material which shows the association of skeletal parts with the carapaces. There- fore, in this paper, comparisons are made wholly on the carapace, with the expectation that the skeleton, when found, will correspond. The Deseado species are but little less specialized than the Santa Cruz, the carapace consisting of movable over- lapping bands of plates both in the anterior and body portions, while over the pelvic region the plates are fixed, do not overlap, and form a pelvic buckler. Ameghino has described a considerable number of genera based on isolated plates, to which I refer later. The chief genera which occur in these beds are also found in the Santa Cruz, and the distinguishing features are as follows: Proeutatus Ameghino Eutatus Amegh., in part, 1887, Bol. Mus. La Plata, t. I, p. 25. Proeutatus Amegh., 1891, Revista Argen. d. Hist. Nat., t. I, p. 327. Thoracothcrium Mercetat, 1891, Revista Mus. La Plata, t. 2, p. 42. Eutatus Lydekker, in part, 1894, Anal. Mus. La Plata, t. 3, p. 62. Proeutatus Scott, 1903-5, Reports Princeton Patagonian Exp., vol. 5, p. 40. This is the most frequently occurring genus in the Deseado, but is as yet represented only by isolated plates. PROEUTATUS 199 The genus is distinguished by thick, relatively long and narrow, movable plates, each overlapped by about a third of its length. The plates of the pelvic buckler are shorter and thicker, the exposed surface of each being ornamented by a figure compared by Ameghino to a flask (see fig. 131), which figure is more distinct on the rear, fading away toward the front. On the plates of the pelvic buckler this figure is more accentuated, and from it, on either side, radiate two furrows dividing the surface into several (4 to 5) areas. The entire surface of each plate is irregularly punctate. Proeutatus lagenaformis Ameghino P. sp? Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 660. P. lagenaformis Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 507. On the Chico del Chubut River, west of Puerto Visser, we found nine specimens of this species, all fragmentary, though one consists of over fifty more or less broken plates, mostly from the pelvic buckler. This is the only species of the genus from the Deseado, and corresponds to the description above. A movable _P plate generally measures about 28 F i g . 131. A , movable P iate ; Band II i * i C, plates from pelvic buckler nat- mm. long by 10 mm. wide, and has urai T size. four large piliferous pits on the posterior margin. A plate of the pelvic buckler varies greatly in size, but is always thick and has two to eight piliferous holes on the posterior margin. A typical plate measures 20 mm. long by 10 mm. wide. Prozaedius Ameghino Zaedius Amegh., in part, 1889, Act. Acad. Nac. Cordoba, t. 5, p. 867. Prozaedius Amegh., 1891, Revista Argen. Hist. Nat., t. I, p. 327. Dasypus Lydekker, in part, 1894, Anal. Mus. La Plata, t. 3, p. 55. Prozaedius Scott, 1903-5, Reports Princeton Patagonian Exp., vol. 5, p. 69. Of this little genus, which is so strikingly like the living Zaedius , we found a carapace with ten rows of movable f 200 THE DESEADO FORMATION OF PATAGONIA plates in place, parts of four rows of fixed plates from the pelvic buckler, and some caudal vertebrae. The genus is distinguished by its thin plates, there being fourteen bands of movable plates, and eight rows in the pelvic buckler. The movable plates are narrow, each overlapped about a fourth of its length, and have a faint ornamentation, with no piliferous pits except on the posterior margin. The fixed plates are similar, except that they are shorter, and have the ornamentation more accentuated, with radial grooves. Ameghino has described three species as follows: P. impressus, sculpture little accentuated, post, piliferic pits rudimentary. P. planus, sculpture more accentuated, post, piliferic pits lacking. P. tenuissimus, very small. In my specimen, the two anterior rows of movable plates lack the marginal piliferous pits, on the next two rows they are rudimentary (which is also true of the lateral plates even further back) , while on the bulk of the movable plates and on those of the pelvic buckler there are two, three or four good-sized piliferous pits on the rear. I can therefore recognize but two species, P. impressus and P. tenuissimus. Prozaedius impressus Ameghino P. impressus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 508. P. planus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 509. Our specimen was found on the Chico del Chubut River, west of Puerto Visser, and preserves over two hundred plates, and eight caudal vertebrae. The anterior rows of plates of the carapace consist of thin plates overlapping about a fourth their length. Just behind the overlap, there is, on each, a group of small punctations, and the exposed part of the surface is divided by two shallow furrows, making three more or less equal ridges which die out toward the rear, leaving the posterior part of the plate plain. These most anterior plates are bent to one side and have no piliferous pits on the rear margin. The plates of the third and fourth rows are not bent, and have the sculpture more distinct, the extreme lateral plates having no piliferous PROZAEDIUS IMPRESSUS 201 Fig. 132. Portion of carapace natural size; unshaded plates are from cast; a and b plates from pelvic buckler. 202 THE DESEADO FORMATION OF PATAGONIA pits, the median lateral plates with rudimentary piliferous pits, and the dorsal ones with well marked posterior pits. In each succeding row toward the rear, the plates are more distinctly ornamented and have larger posterior marginal pits. I have no marginal plates. The plates of the pelvic buckler do not overlap, are shorter, have a very distinct figure, and, in addition to the longitudinal furrows, have a couple of radial furrows on either side, which divide the plate into four or five areas (see fig. 132 a and b). The caudal vertebrae are short and thick, indicating a short tail. I found no plates which would indicate a caudal shield, which coincides with the experience among the Santa Cruz specimens. The figures are to scale and give most of the measurements. There are ten rows of movable plates, probably two to three rows lacking. There are twenty + plates to a row. A typical movable plate measures 17 mm. long by 6 mm. wide. There were at least four rows in the pelvic buckler, probably eight as in the Santa Cruz. A typical fixed plate measures 10 mm. long by 5 mm. wide. Prozaedius tenuissimus Ameghino P. tenuissimus Amegh., 1902, Bol. Acad. Nac. Cienc. Cordoba, t. 17, p. 66. This species is characterized by Ameghino on account of its small size. The movable plates have two furrows which converge toward the front, and between which is a median crest. In the furrows are two rows of perforations. A movable plate measures 9 mm. long by 4 mm. wide. Stenotatus Ameghino Euphractus Ameghino, in part, 1887, Bol. Mus. La Plata, t. I, p. 26 of separate. Dasypus Amegh., in part, 1889, Act. Acad. Nac. Cienc. Cordoba, t. 5, p. 864. Stenotatus Amegh., 1891, Revisla Argen. Hist. Nat., t. I, p. 253. Dasypus Lydekker, 1894, Anal. Mus. La Plata, t. 3, p. 55. Prodasypus Amegh., 1894, Bol. Acad. Nac. Cienc. Cordoba, t. 13, p. 172 of separate. Stenotatus Scott, 1903-5 Princeton Patagonian Exped., vol. 5, p. 80. The genus is very like Prozaedius but differs in having thicker and wider movable plates, in having more rows of PROEUPHRACTUS 203 plates in the pelvic buckler (n), and in details throughout the skeleton. We found no representatives of the genus, but Ameghino has described a species (no figure), S. (Prodasypus) ornatus* based on isolated plates. A mov- able plate measures 18 mm. long by 6-7 mm. wide, while a fixed plate measures 9 mm. long, by 6-7 mm. wide. Proeuphractus Ameghino Proeuphractus Amegh., 1886, Bol. Acad. Nac. Cienc. Cordoba, t. 9, p. 208. This genus is seldom found, but is distinguished by Ame- ghino by the absence of a pelvic buckler, all the plates of the crapace being movable. From the Deseado beds, Ameghino describes two species, P. setiger and P. laevis, both based on isolated plates; the former distinguished by having no pilif- erous perforation in the furrows surrounding the central figure, and with well-developed pits on the posterior margin; while the latter has small piliferous perforations in the fur- rows and only rudimentary ones on the posterior margin. These features do not seem to me to distinguish species. In addition to the foregoing, Ameghino has made a series of genera and species, f Archaeutatus, Amblytatus, Isutaetus, Sadypus, Hemiutatus, Anutaetus, all based on isolated plates, and distinguished by variations in the central figure and the piliferous pits. I am unable to find a satisfactory basis for distinguishing the genera or species, and feel that, until more complete material is known, it is impossible to say which are valid genera, or species. Peltephilus Ameghino Peltephilus Amegh., 1887, Bol. Mus. La Plata, t. I, p. 25 of separate. Cochlops Amegh., in part, 1889, Act. Acad. Nac. Cienc. Cordoba, t. 5, p. 792. Gephyranodon Amegh., 1891, Revista Argen., Hist. Nat., t. I, p. 119. ? Anatiosodon Amegh., 1891, Revista Argen. Hist. Nat., t. I, p. 327. Peltephilus Scott, 1903-5, Princeton Patagonian Exped., vol. 5, p. 88. While rare, this genus is well known from the Santa Cruz, and is characterized by the curious development of *Bol. Inst. Geog. Argen., t. 18, p. 508, 1897. t Bol. Acad. Nac. Cienc. Cordoba, t. 17, p. 56-66, 1902, no figures. 204 THE DESEADO FORMATION OF PATAGONIA the head shield, which consists of nineteen or twenty-one definitely arranged head plates, the anterior ones being developed into horn-like projections. The plates of the carapace are wide, thin, and unique in each having two to four wide shallow pits on the exposed surface. We found the genus rare, only two isolated plates turning up. From the Deseado material Ameghino has made three species: P. protervus, of very large size; P. undulatus, of moderate size, with the median figure accentuated and ending in two pits and with piliferous depressions on the margin; and P. depressus, of the same size as the foregoing, with a faint central figure, often four pits on the exposed surface and no piliferous pits on the margin. We found but one species, one plate of which combines characters of both the last two as described, so that I feel that there should be but two species, P. protervus and P. undulatus. Peltephilus undulatus Ameghino P. undulatus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 509. P. depressus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 510. One of the plates we found has the rough surface, obscure figure, two pits on the median part of the surface, and marginal piliferous pits, of which the first two features are characters of P. undulatus, the last is the feature of P. depressus, so I have combined the Fig. 133. TWO movable two species. A second plate does not have the marginal pits but is otherwise the same. I expect considerable variation in the pattern on plates from different regions of the carapace. Peltephilus protervus Ameghino P. protervus, Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 509. This species, of which we found no representative, is very large. The plates of the type have two pits on the PALAEOPELTIS 205 anterior part of the exposed surface and none on the margin. A movable plate measures 41 mm. long by 22 mm. wide. One of the horn-like plates from the cephalic shield is 35 mm. long, by 30 mm. wide, and has a height of 44 mm. GLYPTODONTIA This suborder is most sparingly represented, apparently on account of unfavorable habitat. Ameghino has de- scribed a few fragments of the carapaces of these forms, making the genera, Palaeopeltis, and Glyptatelus , both pre-Santa Cruz genera. Palaeopeltis Ameghino Palaeopeltis Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 659. The basis of this genus is a few plates of a glyptodon- like animal of considerable size, but the plates are without ornamentation. This form Ameghino considers inter- mediate between glyptodonts and armadilloes. I feel that there is too little of the skeleton known- to justify this conclusion, especially as glyptodonts of a considerably higher grade of specialization are contemporaries of this form. Fig. 134. P. inornatus: a single plate natural size. 2O6 THE DESEADO FORMATION OF PATAGONIA Palaeopeltis inornatus Ameghino P. inornatus Amegh., 1895, Bol. Inst. Geog. Argcn., t. 15, p. 659. P. inornatus Amegh., 1897, Bol. Inst. Gcog. Argcn., t. 18, p. 506. The species is founded on four plates which are without ornamentation, and externally smooth except for numerous vascular perforations. They are of considerable size and entirely characteristic. The one such plate which we found is shown in fig. 134. Glyptatelus Ameghino Glyptatelus Amegh., 1897, Bol. Inst. Geog. Argen., 1. 18, p. 507. The plates of the carapace are similar to those of Pal- aeohoplophorus, the O-figure being, however, nearer the rear of each plate, and the number of radial furrows being smaller, usually six. We found no specimens of this inter- esting form. Ameghino has made two species, G. tatusinus and G. malaspinensis. Glyptatelus tatusinus Ameghino G. tatusinus, Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 507. I reproduce Ameghino's figure of this species which shows all that is known of the form. - I3S- Four7i>lalcs natural size, after Ameghino. GLYPTATELUS 207 Glyptatelus malaspinensis Ameghino G. malaspinensis Amegh., 1902, Bol. Acad. Nac. Cienc., Cordoba, t. 17, p. 50. This species is described (no figure) as about the same size as the preceding, but with subordinate figures in the central O-figure, and also outside of it. A dorsal plate measures 26 mm. long by 20 mm. wide. GRAVIGRADA Remains of this suborder are almost as rare as those of the glyptodonts, and apparently for the same reason, unfavorable habitat. We found no remains of this group, but Ameghino has described a skull and some teeth as belonging to this group; so, in order to present a complete view of the Deseado fauna, I give a digest of his descrip- tions. Hapalops antistis Ameghino H. antistis Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 505. The species is based on a skull, not figured, of which Ameghino says: the size is small, the molars are compressed from front to back, and gives the following measurements: Length of cranium from front of max. to occ. condyles 140 mm. Length of four post, molars 27 mm. Distance from front of max. to back of last molar 48 mm. Octodontotherium Ameghino Octodontotherium Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 656. The genus is based on isolated teeth, each a mass of dentine surrounded by a thin layer of cement. The an- terior tooth of the upper jaw is caniniform, the first molar ovoid in section, the last molar is bilobed, corresponding to Pseudolestodon. The first tooth of the lower jaw is also caniniform, but is two-faced as a result of wear. The intermediate upper and lower molars are rectangular prisms resembling those of Chlamdotherium. 208 THE DESEADO FORMATION OF PATAGONIA Octodontotherium grandis Ameghino O. grandis Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 656. O. grandis Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 505. In addition to the above, Ameghino simply gives the following measurements: First upper tooth, ant. -post. diam. 20 mm., trans, diam. 13 mm., height 80 mm. First lower tooth, ant. -post. diam. 21 mm., trans, diam. 16 mm., height 80 mm. Last lower tooth, ant. -post. diam. 28 mm., trans, diam. of ant. lobe 18 mm. Last lower tooth, trans, diam. of post, lobe 16 mm., median diam. 7 mm. Fig. 136. A, lower molar, side view natural size; B, lower molar, cross section natural size; C, upper molar, cross section natural size, after Ameghino. j __, Octodontotherium crassidens Ameghino O. crassidens Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 505. This second species is based on isolated teeth of larger size than the preceding, with measurements as follows: Upper molar, ant.-post. diam. 26 mm., trans, diam. 18 mm. Lower molar, ant.-post. diam., 26 mm., trans, diam. of ant. lobe 21 mm., trans, diam. of post, lobe 16 mm. ORPHDDON 209 Orphodon Ameghino Orphodon Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 658. Orphodon Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 504. The teeth of this type are a mass of dentine, each sur- rounded by a thin layer of cement, and each tooth subcy- lindrical in section, with the crown worn to two apposed oblique planes. The genus resembles Ortotherium. Fig. 137- Type natural size, after Ameghino. Orphodon hapaloides Ameghino O. hapaloides Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 658. O. hapaloides Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 505. In addition to the above, Ameghino gives a figure, here reproduced, and the following measurements: Tooth, greatest diam. 12 mm., lesser diam., 10 mm. 14 CHAPTER XV MARSUPIALIA IN OUR collection, the marsupials are represented, un- fortunately, by but a few specimens; though this Deseado fauna included, as is shown by the fragmentary remains, a wide range of forms from Pilchenia, the size of a mouse, up to the bear-sized Proborhyaena. The small forms were probably insectivorous, while the larger forms took the place of the carnivors, the absence of true Carnivom being one of the striking features of the fauna of South America during earlier Tertiary times. The treatment of these forms has been as varied as their sizes. Ameghino, with his idea that the Casamayor and Deseado beds were Cretaceous in age, groups the larger forms as a suborder, Sparassodonta, and considers them ancestral to the Creodonta; while the small forms make up his Sarcobora which he considered ancestral on one side to the rodents, on the other to the diprotodont marsupials. Sinclair, after showing the marked similarity of the Spar- assodonta to the polyprotodont marsupials, especially the genus Thylacynus, abandons that term and puts them in the family Thylacynidae along with the Australian forms; the Microbiotheridae he finds similar to opossums and puts in the family Didelphidae; while the remaining small diprotodont forms he associates with Caenolestes, and using Ameghino's families as subfamilies makes three divisions of the family, 'Palaeothentinae, Garzoninae, andAbderitinae. Matthew finds the sparassoclonts to be true marsupials, and without phylogenetic relationship with the creodonts. Gregory diagrams the sparassodonts as coming from gener- alized didelphids and derives them from the same line as the Australian poly pro todonts; while the small caenoles- MARSUPIALIA 211 toids represent a line of descent from some still earlier generalized poly pro todonts and a separate stem from the Australian diprotodonts. Sinclair has had the most complete material on which to work, and with his general grouping I have come to agree. This recognizes three divisions of South American Marsupials, the Didelphidae, representatives of which have not yet been found in the Deseado, though occurring in both the earlier and later formations; the Caenolestidae represented today by Caenolestes, the only survivor of the South American diprotodonts; and the Borhyaenidae ( = Thylacynidae of Sinclair this name having been used to indicate a much nearer relationship to the Australian Thy- lacynns than I feel is warranted), which includes a large range of medium to large sized animals ranging from the Casamayor formation throught the Santa Cruz beds. The locality from which these marsupials emigrated to South America and the time of their arrival is not yet agreed upon, and can not be settled until much more com- plete material is discovered in the Casamayor formation. I feel, however, that the three groups were separate when they entered South America. Borhyaenidae Ameghino has grouped in this family a considerable number of genera of powerful, wolf-like carnivorous marsupials, characterized by a dental formula 4 3 3 \ * ** heavy heads, short limbs with usually five semidigitigrade toes. The genera are mostly distinguished by the relative development of the protocone on the upper molars and the * There is a discussion as to the homologies of the premolars of marsupials and placental mammals, the one proposition being that marsupials have three premolars and four molars, the other that they have four premolars and three molars as in placentals. The evidence is not conclusive as to either proposi- tion, but in this paper I have designated these teeth along the latter line of thought. 212 THE DESEADO FORMATION OF PATAGONIA talonid on the lower ones. Figure 138 gives a typical mar- supial upper molar 2 and a lower molar 2 to show the sense in which these terms are used. The Santa Cruz genera are the best known and I therefore use them as a basis for comparison with the less known Deseado forms, of which \ve found but the one genus Pharsophorus at all abundant. In addition to this, Ameghino has reported a gigantic form designated Proborhyaena. The following table indicates the relationships of the best known genera. From the foregoing, it will appear that Pharsophorus approaches Borhyaena and Prothylacynus in the structure of its upper molars, being, however, nearer to the former, and the same is true of the structure of the talonid; but Pharsophorus differs markedly from both in retaining the metaconid as a small cusp on the side of the protoconid on all of the lower molars; also in the extremely small size PHARSOPHORUS 213 of the talonid of the lower molars, which in Pharsophorus have no basin and consist of a single cusp; and, lastly, in the symphysis of the lower jaws being ligamentous, whereas in the two preceding genera, it is fused. Pharsophorus is probably ancestral to Borhyaena. In the case of Probor- hyaena, only a mandible, with the canine and premolars 3 and 4 intact, has been found. The fourth premolar is more reduced than in other genera, but, until more teeth are known, its affinities can not be at all closely determined. U.m. Fig. 138. Diagram of a generalized upper molar, U.m., and a lower molar, L.m., of Borhynidae; a.s., ant. style; hid., hypoconulid; ml., metacone; mtd., metaconid; pa., paracone; pad., paraconid; Pr., protocone; prd., protoconid; Id., talonid. Pharsophorus Ameghino Pharspohorus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 502. The genus was founded on a lower jaw with premolar 3 to molar 3 in position. We found beside the above an upper jaw with premolar 3, and molars 2 and 3 complete while premolar 4 and molar I are more or less fragmentary ; from which the following generic characters may be made out. The incisors are tiny; the canine very large, equal to that of Borhyaena; the upper and lower premolars pro- gressively smaller from front to back. Upper premolar 3 is a simple two-rooted tooth, the crown consisting of a single blunt central cusp. On the upper molars the proto- cone is not developed as a cusp, though the third inner root 214 THE DESEADO FORMATION OF PATAGONIA is present and carries a rounded shelf. The paracone is the chief cusp, and is developed as a high central pointed denticle. The metacone is not developed as a cusp, but is represented by a long slanting ridge to the rear, the apex of which has been fused to the paracone. The last upper molar is better developed than in most Santa Cruz genera, consisting of a high median cusp, the paracone; a small anterior cusp, the anterior external style; and a shelf-like posterior cusp, the protocone. Lower premolars 1-3 are simple two-rooted teeth, each carrying a single cusp on the crown. The fourth premolar carries a well marked paraconid in front, a large median protoconid on the rear of which is a tiny metaconulid ; and a tiny talonid or heel w r hich is without a basin and consists of a single tiny cusp. The molars are all of the same character as the last pre- molar. The lower jaws are united by a ligamentous sym- physis. Ameghino distinguished four species, P. lacerans, P. tenax, P. mitis, and P. tennis, in the order of their size. The last two are but little known but are quite certainly another genus. Pharsophorus lacerans Ameghino P. lacerans Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 503. The species was founded on a lower jaw with the roots of the incisors, canine, and first two premolars, and with the remaining teeth intact. We did not find the species, Fig. 139. Left mandible 1/2 natural size, after Ameghino. PHARSOPHORUS 215 so I have reproduced Ameghino's figure and give his measurements. Lower dentition, length incisor I to molar 3 114 mm. Lower dentition, length premolar I to molar 3 90 mm. Lower dentition, height of mandible under pm. 4 38 mm. Pharsophorus tenax Ameghino P. tenax Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 504. The species was based on a fourth premolar which was 10 mm. long as compared with 13 mm. in P. lacerans. We found on the Chico del Chubut, west of Puerto Visser, both an upper and lower jaw belonging to this species, which give us the knowledge of the upper den- tition for the genus. The species is distinguished by the Smaller Size, relatively heavy jaws, and plump teeth, indicating a heavier built animal than P. lacerans. The following measurements distinguish the species. Fig. 140. Lef t upper jaw 1/2 natural size; A, molars 3 and 4 from above; B, molars from ex- Fig. 141. Right mandible 1/2 natural size. SPECIMEN 3192 Upper dentition, length premolar I to molar 3 Upper dentition, premolar 3, length Upper dentition, molar I, length Upper dentition, molar 2, length Upper dentition, molar 3, length 76 mm. 10.5 mm., width 6 mm. 11.5 mm., width 8.5 mm. 12 mm., width 9 mm. .55mm., width 12 mm. 216 THE DESEADO FORMATION OF PATAGONIA SPECIMEN 3004 Lower dentition Distance from premolar I to molar 3 76 mm. Molar i, length II mm., width 6 mm. Molar 2, length 13 mm., width 6 mm. Molar 3, length 13 mm., width 7 mm. Height of mandible under premolar 4 30 mm. Notogale gen. nov. This genus is proposed for the species designated ?Phar- sophorus mills by Ameghino (should probably include ?Phar- sophorus tennis which however never having been figured and not found in our collection I cannot definitely place). While the upper teeth have the same general character as Pharsophorus, they are much more compressed and tren- chant. Upper molar 2 is similar to that of Pharasophorus in having the protocone reduced, and the metacone repre- sented by a long sloping ridge. The last molar is also similar in having the antero-external style, the developed paracone, but the protocone is much less developed, appear- ing only as a ridge. In the lower teeth, however, there is a marked difference, in that the metaconid is lacking on molars, while the talonid is developed into a small basin with a single cusp on the posterior margin. This genus seems to be closest to the Santa Cruz genus Cladosictis. Notogale mitis Ameghino ? Pharsophorus mitis, 1897, Bol. Inst. Geog. Argen., t. 18, p. 504. Ameghino briefly describes, without a figure, a species in which premolar 4 and molar 3 together measure 14 mm. o p'C^F^ I have assigned to this two specimens, the {./ (js P one with pm. 2 incomplete, pm. 3 complete, Fig. 142. upper and m. 2 also complete. These teeth meas- molars 2 to 4 nat- ure the same as Ameghino' s and I think are the same.. There is also a fragment of the upper jaw with molars 2 and 3, though imperfect. From these it appears NOTOGALE 2I 7 that we have to do with an animal not only smaller than the preceding, but on much slenderer lines. The following are the measurements of the two specimens. SPECIMEN 3117 Upper dentition, molar 2, length Upper dentition, molar 3, length SPECIMEN 3060 Lower dentition, premolar 3, length Lower dentition, molar 2, length 8 mm., width 2 mm., width 6 mm., width 7 mm., width 6 mm. 6 mm. 2.5 mm. 4 mm. Fig. 143. Left mandible natural size. Notogale tenuis Ameghino ? Pharsophorus tenuis Amegh., 1897, Bol. Inst. Geog., Argen., t. 18, p. 504. This species was founded by Ameghino on a single lower premolar 3 which is 3 mm. in length. No further descrip- tion is given and no figure. Proborhyaena Ameghino Proborhyaena Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 501. The genus is founded on a large lower jaw carrying the canine and premolars 3 and 4 and roots or alveoli for the other teeth. It is the largest carnivor recorded from Patagonia, and as large as a small bear. It is not possible 2l8 THE DESEADO FORMATION OF PATAGONIA to place its exact relationships, for the most essential teeth are wanting, but it is certainly a distinct genus as indicated by the reduced size of premolar 4 and the plump character of the teeth. Proborhyaena gigantea Ameghino P. gigantea Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 501. We found no specimens of this great carnivor, so I am reproducing Ameghino's figure and measurements. The Fig. 144. Right mandible 1/2 natural size, after Ameghino. heavy canine is channeled on the sides and much worn on the posterior face. Premolar 3 is a plump tooth, its crown consisting mostly of a single median cusp, but with a small heel behind, and, strikingly enough, premolar 4 is a smaller and simpler tooth with a single cusp. MEASUREMENTS Lower dentition, canine, antero-posterior diameter 30 mm. Lower dentition, canine, transverse diameter 20 mm. Lower dentition, premolar I to molar 3 145 mm. Lower dentition, height of mandible under pm. 4 60 mm. CAENOLESTIDAE 219 Proborhyaena antiqua Ameghino ? Borhyaena antiqua Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 655. Proborhyaena antiqua Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 5012. This species is known only by a single canine 100 mm. long, of which but 15 mm. belongs to the crown. Its antero-posterior diameter is 14 mm., the transverse 12 mm. It seems to me very doubtful whether this is a valid species. Caenolestidae This family, based on the living genus Caenolestes, is represented in Tertiary times in Patagonia by three sub- divisions, Palaeothentinae, Garzoninae, and Abderitinae. While diprotodonts, as far as known, the family is in strong contrast to the Australian diprotodonts in that there is no sign of syndactylism in the pes. The American forms are characterized by four subequal upper incisors, a normal canine, the first three premolars vestigal, while the fourth is either normal or enlarged into a sectorial tooth. The three molars are progressively smaller from the front back. The first lower incisor is greatly enlarged and procumbrent, the remaining incisors, the canine, and the anterior pre- molars being vestigal though usually present. Premolar 4 is enlarged and sectorial in most genera, and the molars as in the upper jaw progressively smaller. For the practical purposes of this paper the subfamilies are distinguished as follows: Caenolestinae, lower pm. 4 not developed into a sectorial tooth. Palaeothentinae, lower pm. 4 is developed into a sectorial tooth. Abderitinae, lower pm. 4 is developed into a sectorial tooth and striated. Palaeothentinae Sinclair. ( = Epanorthidae Ameghino) This group or subfamily was established to hold several genera of tiny marsupials with the dental formula jTrf-; the lower fourth premolar enlarged into a sectorial tooth; 220 THE DESEADO FORMATION OF PATAGONIA and the molars small and buno-lophodont. From the Deseado beds but one genus of this subdivision has been found, Palaeothentes, designated by Ameghino first Epan- orthus ^ then later Palaepanorthus, but as I can see no reason for distinguishing the Deseado species of the genus from those of the Santa Cruz, I have retained the name Palaeo- thentes. The genera of this subfamily are distinguished as follows: LOWER THIRD PREMOLAR Palaeothentes 2-rooted, fairly large, equals pm. 4 in height. Pilchenia 2-rooted, moderate size nearly equals pm. 4 in height. Callomenus 2-rooted, small size much lower than pm. 4. Decastris, i -rooted, vestigal. Palaeothentes (Moreno) Ameghino Palaeothentes Moreno, 1882^ Patagonia, Resto de tin Continente hoy sub- mergido, p. 22, (nomen nudum). Palaeothentes (Moreno) Ameghino, 1887, Enum. Sist. Espesies Mamif. Fos, Patagonia, p. 5. Epanorthus Ameghino, 1889, Act. Acad. Nac. Cienc. Cordoba, t. 6, p. 271. Epanorthus Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 500. (nudum). Palaepanorthus Amegh., 1901, Anal. Soc. Cienc. Argen., t. 51, p. 77, (nomen). Palaepanorthus Amegh., 1902, Bol. Acad. Nac. Cienc. Cordoba, t. 18, p. 123. Palaepanorthus Amegh., 1903, Anal. Mus. Nac. Buenos Aires, t. 9, (ser. 3, t. 2) p. 239. Among the Santa Cruz specimens, this genus is distin- guished by having in the lower jaw the large first incisor, then five vestigal teeth, followed by a two-rooted, though somewhat reduced, third premolar, next the enlarged fourth premolar, making the sectorial tooth, and lastly three buno-lophodont molars. There is considerable confusion as to the use of the gen- eric name. Moreno designated the first specimen, Palaeo- thentes, without a description ; then Ameghino used this term describing the species; later Ameghino thinking that the name Palaeothentes was the same as Palaeothentis proposed the name, Epanorthus, using this for the first description PALAEOTHENTES 221 of the Deseado species. Later, however, he changed this for Palaepanothus. As I can see no generic differences between the Deseado and Santa Cruz species, I shall follow Sinclair in using the generic term Palaeothentes. Palaeothentes chubutensis Ameghino Epanorthus chubutensis Amegh., 1897, Bol. Inst. Geog. Argen., t. 18, p. 500. Palaepanorthus chubutensis Amegh., 1901, Anal. Soc. Cienc. Argen., t. 51, p. 77- Palaepanorthus chubutensis Amegh., 1902, Bol. Acad. Nac. Cienc. Cordoba, t. 18, p. 123. Palaepanorthus chubutensis Amegh., 1903, Anal. Mus. Nac. B. A., t. 9 (ser. 3, t. 2) p. 239. The species is founded on a tiny mandible with premolar 3-molar 3, on which the third premolar, while reduced, Fig. 145. Right mandible 2 times natural size, after Ameghino. has two roots and reaches the height of the fourth premolar, being in about the same stage of development as the Santa Cruz species. As we found no specimens of this species I reproduce Ameghino's figure and measurements. Lower dentition, premolar 3 to molar 3 19 mm. Lower dentition, height under premolar 4 12 mm. Pilchenia Ameghino Pilchenia Ameghino, 1903, Anal. Mus. Nac. Buenos Aires, ser. 3, t. 2, p. 128. Pilchenia Ameghino, 1904, Anal. Soc. Cienc. Rep. Argen., t. 58, p. 259. This genus was founded on a single lower molar which, in the light of the specimen we found, I take to be the third 222 THE DESEADO FORMATION OF PATAGONIA or last. Our specimen shows pm. 3 and 4 and the three molars. The third premolar is a small two-rooted tooth with a simple crown and no heel. Premolar 4 is an en- larged sectorial tooth, the anterior part consisting of two cusps, closely set near the median line, with an incipient cusp on the inner face of the large anterior cusp. The posterior part of this tooth is arranged as a typical talonid, with one internal and two external cusps on the margin of a shallow inclosed basin. On the rear of the tooth is a small crescent-like cingulum, which occurs in the same place on molars I and 2, but is lacking on molar 3. This is a characteristic feature of the genus. On the anterior part of the molars is developed a sort of trigonid of small size, and the cusps are indistinct. The posterior portion of each molar is a large talonid with a shallow basin surrounded by a low wall on which are three tiny cups (the entoconid, hypoconid, and hypoconulid). Pilchenia lucina Ameghino P. lucina Amegh., 1903, Anal. Mus. Nac. B. A., ser. 3, t. 2, p. 128. P. lucina Amegh., 1904, Anal. Soc. Cienc. Rep. Argen., t. 58, p. 259. In the Deseado beds, on the Chico del Chubut River, west of Puerto Visser, we found a single specimen of this Fig. 146. Left mandible with premolar 3 to molar 4 4 times natural size. species which agrees with the single tooth figured by Ameg- hino as the type, and which I interpret as molar 3. The description is given under the genus, the measurements are as follows: CALLOMENUS 223 SPECIMEN No. 3110 Lower dentition, distance from premolar 3 to molar 3 14 mm. Lower dentition, premolar 3, length 2 mm., width .75 mm. Lower dentition, premolar 4, length 5 mm., width 2.5 mm. Lower dentition, molar I, length 3 mm., width 2 mm. Lower dentition, molar 2, length 2.5 mm., width 2 mm. Lower dentition, molar 3, length 2 mm., width 1.75 mm. Lower dentition, height under pm. 4 5 mm. Callomenus Ameghino Callomenus Amegh., 1891, Neuvos Restos Mamif. Fos. Patagonia Austral, p. 20. Callomenus Sinclair, 1901-6 Princeton Patag. Expeditions, vol. 4, p. 434. This genus has not been previously reported from the Deseado beds, but we found a tiny lower jaw with three teeth to represent it. The genus is distinguished by pre- molar 3 being two-rooted, but so small as not to attain the height of premolar 4. Callomenus praecursor sp. nov. The type is specimen No. 3020, a fragmentary mandible with premolars 3 and 4 and molar i in place. Pm. 3 is Fig. 147. Left mandible with premolar 3 to molar 2 4 times natural size. Fig. 148. Left mandible internal side 4 times natural size. two-rooted, but so small as to be entirely overshadowed by the succeeding pm. 4, hardly reaching a half the height of that tooth. On the last premolar and the first molar, the cusps are arranged in a trigonid in front and a talonid behind, the cusps being joined by thick ridges, making two connecting crescents. 224 THE DESEADO FORMATION OF PATAGONIA MEASUREMENTS, SPECIMEN 3020 Lower dentition, premolar 3, length i mm. Lower dentition, premolar 4, length 5 mm., width 2 mm. Lower dentition, molar i, length 4 mm., width 2 mm. Lower dentition, height under pm. 4 6.5 mm. Caenolestinae The subfamily is distinguished by the formula 4133, pm. 4 not being enlarged, and the lower molars being (ul)crculo-sectorial. In the Descado formation this group is only represented by a single species, based on a single tooth found by Ameghino. Pseudhalmarhiphus guaraniticus Ameghino P. guaraniticus Amegh., 1903, Anal. Mus. Nac. Buenos Aires, ser. 3, t. 2, p. 83- Based on a single tooth, similar to those found in the Santa Cruz. Abderitinae The subfamily is distinguished by the formula * \ ? 2 * 3 , the fourth premolar being enlarged into a sectorial tooth on the sides of which are vertical striae. The molars are buno-lophodont. The Deseado has yielded only a tiny form, designated Parabdcrites, which differs from the Santa Cruz genus Abderiles in pm. 4, the same shape, by with few to no striae on its sides. Parabderites minusculus Ameghino P. minusculus Amegh., 1902, Bol. Acad. Nac. Cienc. Cordoba, t. 17, p. 43. The species as described by Ameghino is based on a lower jaw with pm. 3 to m. 3. The specific character is the lack of striae on pm. 4. No figure is given but the following measurements indicate the size. Lower dentition, premolar 3 to molar 3 9 mm. Lower dentition, height of mandible under pm. 4 4 mm. CHAPTER XVI BIRDS IN THE Deseado beds, birds occur in small numbers, Ameghino having described four species. The remains are generally found as isolated bones, and it is hard to as- sociate the separate finds one with another. Beside this there are very few birds of the early Tertiary so known, as to make separate bones indicate the family or generic relationships. In the overlying Patagonian beds, a considerable number of species have been found, mostly of penguin-like birds, the various genera and species being based on the tarso- metatarsus. On the upper surface of the Deseado, we found several bones of this penguin-like type, but in all cases they were washed out, so that I have considered them as having come from the Patagonian. However, we found eight specimens of birds in place in the Deseado, most of which are clearly land birds and belong to genera which are closely related to genera of the Santa Cruz, especially the two genera Phororhacus and Pelecyornis, and of sizes equal to the largest represen- tatives of the two genera. Phororhacus Ameghino Phororhacus Amegh., 1887, Bol. Mus. La Plata, t. I, p. 24. Phororhacus Amegh., 1889, Act. Acad. Nac. Cienc. Cordoba, t. VI, p. 659. Phororhacus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 10 of separate. This is a group of large land birds, comparable in size to the great moasof New Zealand which apparently arose, flour- ished, and died out in South America. In the Santa Cruz they were abundant, the best known form being P. inflatus, a bird some six feet high ; while the largest, P. longissimus, 15 226 THE DESEADO FORMATION OF PATAGONIA had a head nearly twice as long and limb bones half again as large as this species; so that it represented a bird nine to ten feet high. Previously but one specimen of this type, a part of a mandible, has been found in the Deseado beds. We were fortunate enough to find the greater part of a femur, indicating a bird equal to the largest of those in the Santa Cruz. There are also toe bones of Phororachus of a size about the same as P. inflatus. A host of names, generic and specific, have been given to the individual bones of the birds of this type, but Ameghino, in studying the birds of the Santa Cruz, brought them all together under the single genus Phororhacus. (See Bol. Inst. Geog. Argen., 1895, t. 15.) Referring to the single bone in the Deseado, however, he gave it a new generic name Physornis, which differs from Phororhacus only in the lower jaw being more convex, but should stand until better material has been found to establish whether it differs enough to be entitled to generic independence. Physornis fortis Ameghino. P. fortis Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 576. Under this specific name Ameghino describes a part of the lower jaw 150 mm. long which he says equals in size Phoror- hacus longissimus, and differs only in the greater convexity of the mandible. Our specimen is a femur, apparently of the same bird, being of the type of Phororhacus and about the size of P. longissimus; so I have placed it in this species. This femur is of large size, moderate length, and has a shaft subcylindrical in section. The distal end is expanded and the condyles are flattened, the inner one being the wider, the outer condylc being narrower and the external margin projecting to make a high ridge. The pit on the posterior side of the shaft just above the condyles is unusually deep and of large size. On the anterior side there extends from either condyle a low marginal ridge which soon fades into PHYSORNIS FORTIS 227 Fig. 149. Right femur, back view 1/2 natural size. 228 THE DESEADO FORMATION OF PATAGONIA the contour of the shaft. Between these ridges there is a wide shallow furrow which also loses itself above in the con- vex surface of the shaft. MEASUREMENTS Femur, least diam. of the shaft Femur, diameter across the condules 58 mm. 148 mm. Physornis sp.? Two phalanges of a size too small to belong to the above species represent a second smaller bird of this type, about equal in size to Phororhacus infiatus. I give a figure of one toe but would wait for more typical material before establishing a species. Fig. 150. Proximal phalanx of Physornis sp? natural size. Fig. 151. Loxornis cliyus, lower end of tibio-tarsus, after Ame- ghino natural size. Loxornis Ameghino Loxornis Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 595. Another group of bones, which we found with consider- able frequency, have the same features as Pelecyornis of the Santa Cruz. Ameghino has described but the lower end of a tibio-tarsus which can be associated with these bones and to it gave the name Loxornis. I can not find LOXORNIS 229 much variation from Pelecyornis except that the coracoid is considerably shorter and wider, and there is a slight variation in the lower end of the ti bio- tarsus. These then are the bases of the generic name. Loxornis clivus Ameghino L. clivus Amegh., 1895, Bol. Inst. Geog. Argen., t. 15, p. 595. Under this name Ameghino has described the lower end of a tibio-tarsus, a figure of which I reproduce here. This is of a size to complete the tibio-tarsus which we found, lacking the lower end, and agrees in size with the other Fig. 152. Humerus Fig. 153. Sternum, thin parts lack- Fig. 154. Coracoid 1/2 natural size. ing 1/2 natural size. 1/2 natural size. bones which we found, so that I shall describe my material under this name. The species is in size comparable to Pelecyornis tubulatus with which it agrees closely. We found the upper four-fifths of a tibio-tarsus, associa- ted with part of the fibula, the sternum, the humerus, and the coracoid; a second specimen consisting of a complete tarso-metatarsus, and fragments of the pelvis, vertebrae and wing bones; a third specimen consisting of part of the tibio-tarsus, and various fragments; a fourth consisting of a femur, and lastly two toes; all evidently representing one species, which in most respects is almost identical with Pelecyornis tubulatus. These all came from the Chico del Chubut, west of Puerto Visser. 230 THE DESEADO FORMATION OF PATAGONIA The humerus has a large head but is considerably flat- tened at the proximal end. The internal side is deeply excavated, the shaft is slender and light as though the wing were quite reduced, though not so much as in Pelecyomis and not nearly as much as in Phororhaciis. The sternum had a moderate keel but both this and body of the bone are very thin, so much so, that in my specimen, much is broken away, giving the figure the appearance of the bone being fenestrated, which w r as not the case. In general the sternum is similar to Pelecyornis. The coracoid is a decidedly stout bone, \vith a wide dis- tal end for articulation of (he sternum. The proximal end has a long articular facet for the scapula. This bone is heavier than the corresponding one in Pelecyornis. The femur has a small rounded head on a short neck, the articular surface spreading over the entire proximal end of the bone. Thus the trochanter is abbreviated and does not rise above the top of the head. The 4 shaft is of considerable length and fairly heavy. The tibio-tarsus has a wide flaring end to receive the articulation of the femur. The bone is very long as in Pelecyornis. On the external side is a long ridge along which the fibula was attached by cartilage or by ligaments, but was not fused to the tibio-tarsus. The shaft is appn >\i- mately cylindrical in section and fairly heavy. The distal end is missing, but if I have associated correctly the speci- men figured by Ameghino, the condyles are ilattened, the inner being the flatter, and the outer rising in a narrow margin. Figure 157 shows a fibula which would have occupied the position indicated along the side of the tibio-tarsus and corresponds entirely with the same bone in Pelecyornis. The tarso-metatarstis is long and slender, almost exactly the counterpart of the same bone in Pelecyornis. The bone has a triangular upper end, with (wo shallow articular LOXORNIS CLIVUS 231 concavities, separated by a median spine. The shaft is rectangular in cross section, has a shallow depression on the anterior face extending from the upper end to below the middle of the shaft; while on the posterior surface is a Fig. 155. Femur 1/2 natural size. Fig. 157- Fibula 1/2 natural size; outline from impression in ma- trix. Fig. 156. Tibio-tarsus 1/2 natural size; fib- ula indicated in out- line. Fig. 158. Tarso-met- atarsus, front view 1/2 natural size. similar furrow, which is however bounded by a higher ridge on the external margin. The distal articular condyles are almost bilaterally symmetrical, the middle one being about half again as large as the two lateral ones. Just above the cleft between the condyles for digits III and IV there is a moderate sized perforation. THE DESEADO FORMATION OF PATAGONIA Of the phalanges, I have two unguals which are narrow curved claws. These were not found in association with any of the foregoing bones, but correspond in size and general character to those of Pelecyornis, and so I consider them as belonging to this genus and species. Fig. 159- Ungual pha- lanx 1/2 natural size. Fig. 1 60. Femur of unknown bird natural size; special No. 3217. Ameghino has suggested that the genus was related to ducks, but with the more complete material it seems, in general build, much closer to the aberrant land birds of the Tertiary of South America, Pelecyornis and Phororhacus; and I am not in position to say what their derivation may have been. Beside the above species there are several more or less complete but isolated bones indicating the presence of other and much smaller birds. I figure such a femur natural size. U.C. BERKELEY LIBRARIES