AMERICAN MUSEUM NOVITATES Published by Number 1136 THE AMERICAN MUSEUM OF NATURAL HISTORY August 11, 1941 New York City

LARGE FELINES OF BY GEORGE GAYLORD SIMPSON

About thirty occurrences of true , ends can be gathered, but on present evi- felines, of the size of pumas or larger have dence it seems possible to establish the been reported in the Pleistocene of North following conclusions: America. Except for the specimens from 1.-Known large Pleistocene felines from the asphalt of Rancho La Brea and of Mc- North America suffice to demonstrate the pres- Kittrick, in , these are neverthe- ence of three, and only three, groups: pumas, less relatively rare and the specimens , and P. atrox. 2.-Although scattered from the Atlantic to are usually fragmentary. They have been the Pacific Coasts, the Pleistocene pumas do assigned by various students to about not, in the known parts, show much if any more fifteen different and their affinities variation than do recent pumas of one and have not been understood. and of more limited geographic distribution. They average a little larger than recent pumas Many have been placed in extinct, or sup- and show minor morphologic distinction of not posedly extinct, species with no definite more than specific value and possibly less. idea as to their relationships to other cats. The definitions of the several supposedly dis- A few have been recognized as pumas, or as tinct groups are not yet satisfactory. 3.-True jaguars specifically inseparable from related to pumas, but on the other hand onca, the living species, occur widely. some that are pumas beyond much doubt They may (doubtfully) average somewhat have been sharply separated from that larger than the largest living races and may be group. Others have loosely been called tentatively distinguished as P. onca augusta. No local differences within North America are "" or "" without clear demon- demonstrated by the known materials. stration of a reason for such explicit terms 4.-Panthera atrox, the so-called American and in spite of the fact that they are re- , is not a lion and might be called a giant lated only remotely and not in any exclu- , clearly distinct specifically from P. onca sive way to those Old World forms. There but nevertheless related to it. has indeed been a general tendency to com- The evidence for these views will be pare our larger felines with those of briefly summarized in this paper. the Eastern Hemisphere and to minimize As acknowledged in more detail else- or omit entirely comparisons with other where,' I am much indebted to Dr. W. J. American cats, in spite of the fact that the Cameron and Mr. W. E. Michael for the latter comparisons would seem more ob- specimens of fossil jaguars from Tennessee vious and are in most cases, perhaps in all, described in this paper. I am also indebted truer indications of affinity. It is particu- to Dr. C. L. Gazin for the opportunity to larly striking that (with one cryptic and study several types and other specimens unexplained partial exception, noted later) in the collection of the U. S. National Mu- no one seems to have reported jaguars in seum, for the loan of an important, unde- the Pleistocene of the United States and scribed specimen collected by J. W. Gidley few have seriously considered this rather in Florida, and for permission to describe obvious possibility. and figure this specimen here. The De- It is the purpose of the present paper to partment of Mammalogy of this Museum make a summary review of these discov- has provided a large number of recent eries and to attempt to establish some specimens for comparison. Dr. Anne Roe order in this confusion. As usual, many problems remain unsolved and not all loose 1 Amer. Muis. Novitates, No. 1131. 2 AMERIAV MVJSEU VYJVITATES [N:. 1136 .

Simpson has read the manuscript and The drawings in this paper were made by offered useful suggestions on several points. J. C. Germann.

NORTH AMERICAN FOSSIL JAGUARS The certain identification and determina- arrangement is still well visible. The right tion of the affinities of the large cats de- canine had apparently been broken off dur- mand associated dentitions, at least, snce ing life and the stump later much worn by even forms as dissimilar as jaguars and use-this is a common occurrence among pumas are much alike as regards isolated recent jaguars. The left canine had not teeth or many skeletal fragments. It is been broken, unless possibly at the tip, but probable that specimens described as long is also much worn, especially on the pos- ago as 1872 and at intervals since then are, terior side. in fact, jaguars, but only now is it possible This jaw considerably resembles Pan- to prove that these did live in thera atrox, indeed it probably has no mor- North America, well outside their present phological characters that do not occur as range, during the Pleistocene. Various variations in that species, which is very American fossils have been compared with jaguar-like in this part and most others. jaguars, as well as with lions, tigers, and It is, however, significantly smaller than other forms in reality more distantly re- any of the fine series of P. atrox jaws now lated, but a positive identification is now known, even if the relatively small " made possible for the first time by three imperialis" be included in P. atrox as Mer- relatively good specimens, Amer. Mus. Nos. riam and Stock suggest and as I also think 32633 and 32635 and U.S. Nat. Mus. No. proper. 11470. In every character, including size and These American Museum specimens were proportions, the fossil jaw closely ap- found in Craighead Caverns, near Sweet- proaches the largest living jaguars. Com- water, Tennessee. An account of their parisons were made especially with a series discovery and peculiar occurrence is being of eight recent specimens of Panthera onca given elsewhere and need not be repeated milleri, which is among the largest, but is here.' Amer. Mus. No. 32633 (Fig. 1) in- not the largest, of recent groups. All cludes the right ramus of the lower jaw, strictly morphological features of the fossil lacking the incisors, which had dropped can be almost exactly matched in this sub- out after death, and the coronoid process species. The fossil jaw is slightly deeper and medial portion of the articular condyle, (not longer), but not significantly more gnawed off by rodents. The symphysial deep in a statistical sense, than in large portion of the left ramus, with the canine males of the recent group. The canine is but without the incisors was also found. unusually stout, but again the difference The symphysis was open and the two sides is slight and not significant. The cheek separated, although they certainly belong teeth are not significantly wider, but are to one individual. Posterior to the ca- significantly longer both absolutely and nine the left ramus had been broken and relatively, although the difference from the the surface is gnawed. There is a slight largest milleri males is not great. They rim of calcareous material near this break comp3rr as follows: and it has been suggested that this repre- sents healing of a jaw fractured during d of Amer. life, but I think it more probably post- Mus. No. mortem and representative of contact with P. onca milleri 32633 from the surface of the dirt, the protruding part (N = 8) milleri having been gnawed away. The cheek Variate Mean a' mean dla' teeth show considerable wear but the cusp LP3 15.19 .99 3.01 3.0 LP4 21.26 .86 3.44 4.0 1 Amer. Mus. Novitates, No. 1131. LMl 21.16 1.25 3.44 2.8 9411 PLEISTOCENE FELINES OF NORTH AMERICA 3

3

Fig. 1. Lower jaws"of fossil and recent jaguars. A.M. No. 32633, Panthera onca augusta, yresumed male, subfossil, 'Craighead Caverns, Tennessee. A.M. No. 120997 (Dept. Mammalogy), Panthera onca milleri, male, recent, Matto Grosso. N. M. 11470, Panthera onca augusta, presumed female, fossil, Melbourne Formation, Melbourne, Florida. A.M. No. 120998 (Dept. Mammalogy), Panthera onca milleri, female, recent, Matto Grosso. All one-third natural size. 4 AMERICAN MUSEUM NOVITATES [No. 1136

I do not have an adequate series of the eluding their palatal processes, and the an- largest living subspecies, P. o. palustris, but terior part of the jugal and a fragment of the lengths of these teeth in the Tennessee the right nasal are also preserved. Near fossil are probably within the range for this and apparently belonging to the same that subspecies, surely so as regards M1 for individual were found the nearly complete which Cabrera (1934) has given ten meas- zygomatic process and a small part of the urements. For this series M = 22.80, a' articular process of the left squamosal and = 1.18, d for the Tennessee specimen = another piece with much of the right ear 1.8, and d/a' = 1.5, far from significant. region, including the inner half of the gle- On the available data I do not, indeed, noid , post-glenoid and post-tympanic find any way to distinguish this fossil from processes, adjacent parts of the squamosal, the large living males of P. o. palustris.1 outer and anterior parts of the tympanic, Amer. Mus. No. 32635 (Fig. 2) was also and the complete petrosal. found in Craighead Caverns and in the Like Amer. Mus. No. 32633, these parts same fissure as the jaw just discussed, al- agree very closely with recent jaguars and though some distance from it and higher in are in all respects near or within the known elevation. The teeth are less worn and range of variation of large males of P. onca

Fig. 2. Panthera onca augusta, restoration of right lateral view of , probably male. Parts in continuous lines are Amer. Mus. No. 32635, from Craighead Caverns, Tennessee. Zygomatic fragment is reversed from the left side. Parts in broken lines restored with reference to recent Felis onca . One-third natural size. do not occlude well, so that this specimen palustris or P. o. milleri. In the upper probably represents a second individual. dentition (unfortunately lacking the char- The right side of the face is represented, acteristic ) the only definite dis- with J1-3, C, and p2-3. The premaxilla tinction from our P. o. milleri series is the and are practically complete, in- greater length of p2 8.5 mm. as opposed 1 This is the subspecies usually called Panthera onea to 7.28 mean and 7.7 maximum observed paraguensis. On the nomenclature of this group see Cabrera, 1934. Incidentally Cabrera notes that the Throughout this paper all recent jaguars are con- whereabouts of the type of Felis onca antiqua Ame- sidered as belonging to one species, Panthera onca, ghino, 1889, is unknown to him and he discusses this following Nelson and Goldman, 1933, and other good form on the basis of a cast. The original was sold by recent authorities. The living pumas are similarly Ameghino to Cope and by Cope to this Museum, all referred to Felis concolor. In both these cases the where it is now permanently preserved as fossil mam- degree of divergence and the character of distribution mal No. 11107. I have examined it and agree with is typically that of geographic races and subspecies, Cabrera's conclusion that it is inseparable from P. not of several well-defined full species according to onca palustris. the best usage. The subspecific separation of P. onca milleri from Finally on the subject of nomenclature it is noted P. o. palustris is uncertain, although milleri does rep- that the name onca should be pronounced "onsa" resent a local group, race or stirps, with slightly and not "onka," but the otherwise preferable spelling different average characters. onca is not permissible because Linnaeus wrote onca. PLEISTOCENE FELINES OF NORTH AMERICA 5 in our recent series of eight specimens. a groove which may be deep or shallow, Although small, the difference is significant broadly rounded or narrow and fissure- (d/o' = 3.7) and the race represented by like. But all these variations may occur the fossil did have a larger p2 than does this (among jaguars, at least) not only in one recent race, both absolutely and relative subspecies but also in one interbreeding to the width of the same tooth. It may not local stock. Indeed in one of the female have differed significantly from P. o. milleri specimens this character is markedly palustris which has on an average a larger different on the two sides of the same skull. dentition than milleri and for which I do (The fossil has a broad groove with rounded not have an adequate series of measure- bottom and of moderate depth.) ments of p2. There are no other appar- Here again this part is basically jaguar- ently significant differences in size or struc- like throughout in the Craighead . ture of the dentition. The distance from the post-glenoid process The osteology of the preserved parts of to the posterior part of the mastoid process the face appears to be within the range of is shorter and the lateral exposure of the either of the recent subspecies mentioned mastoid process is smaller than in recent and, except for size, of recent jaguars gen- comparative specimens of similar size, but erally. An interesting feature is the pres- the difference is so slight and the variation ence of a shallow, irregular pit on the so great that this observation has no neces- maxillo-nasal suture immediately posterior sary significance. Another unusual but to the ascending process of the premaxilla. not unique feature of doubtful significance This apparently represents a puncture, is the clarity and size of a pit (foramen such as would be made by the canine of an- spinosum?) anterior to the external audi- other jaguar, healed during life. Such tory meatus and immediately medial to the scars are very common among recent jag- post-glenoid process, and the development uars, usuallyon males, to such an extent that from the tympanic of a prominent spine they could almost be considered as second- medial to this. ary (or tertiary) sexual characters. All Found near these skull parts was the of our wild-collected and definitely sexed articular end of a right , Amer. male milleri skulls have such head scars, in Mus. No. 32638, and a left second meta- one case (Dept. Mam. No. 36950) in almost tarsal, Amer. Mus. No. 32637. These also the same position as in the fossil. None of belong to a jaguar and probably to one of the definitely sexed or probable females has the individuals represented by the jaw and similar scars. Although obviously incon- skull parts (or possibly, but improbably, clusive, this is serious evidence that the all are of one individual), but association is fossil is a male, a matter of some taxonomic not demonstrated. The scapula has been importance as will appear. much gnawed by rodents. As far as can be The zygomatic process of the squamosal judged by what remains, which is charac- is not very characteristic and requires no teristic enough for fair identification, it more comment than to say that it is almost closely resembled the same part in a large exactly like the same part in large recent recent jaguar. The metatarsal is perfectly jaguars. The rugosities on the temporal preserved and is particularly valuable be- surface of the root are a little more promi- cause the metapodials are among the most nent than in my comparative specimens, characteristic parts of the jaguar and per- but this has no apparent significance. mit unequivocal identification at least as The auditory region, on the other hand, far as the species. This specimen is de- has great value in determining affinities cisively like P. onca in all important re- despite the fact that it shows a surprising spects. In comparison with my limited re- amount of variation. For instance, the cent material, the mesocuneiform articular degree of application of the mastoid proc- surface extends farther on the head in a ess to the bulla gives rise to striking dif- plantar direction, the process that abuts ferences. The two are sometimes smoothly against the first metatarsal is more promi- continuous and sometimes separated by nent and more distal, and the shaft is less 61 AMERICAN MUSEUM NOVITATES [No. 1136 curved. These small differences in variable like P. atrox, and quite different from the features are outweighted by the general . agreement, especially since the available Taken together, the Craighead Caverns comparative specimens are from zoo ani- specimens include many fully identifiable mals and the slight differences do not con- features and leave no possible doubt that stitute definite resemblances to other they do represent a jaguar inseparable known species. from Panthera onca, if that species be used The principal measurements of this to include all recent jaguars as is now gen- are as follows: erally done. Greatest length ...... 84.2 mm. Dorsoventral diameter on outer face of proximal This positive identification and spread of end...... 20.6 material for comparison makes possible the Greatest transverse diameter of articulation on more tentative identification of various proximal end ...... 12.2 Least transverse diameter of shaft... 11 .9 similar but less complete specimens found Transverse diameter of distal end of shaft previously. Of these by far the best is

...... 16.6 US. Nat. Mus. No. 11470, collected by

CRAIGHEAD RECENT AMER.MUS. RANCHO LA BREA MERRIAM & STOCK DATA)>

OORSOVENTRAL DIAMETER ON I OUTER FACE OF PROXIMAL END. GREATEST TRANSVERSE DIAMETER OF PROXIMALEND.

SECOND METATARSAL Fig. 3. Ratio diagram of dimensions of second metatarsal in various large felines, as labeled. Recent jaguar and puma, one specimen of each in American Museum. P. atrox, large Rancho La Brea sample, data from Merriam and Stock; horizontal lines show observed ranges. For method of construction and interpretation of this and other ratio diagrams (Figs. 6-10) see notes at end of text.

The accompanying ratio diagram, Fig. 3, the late Dr. J. W. Gidley at Melbourne, compares the proportions of this bone to Fla., in 1926 but not hitherto published. homologous elements in some other cats. It is not exactly labeled as to horizon but The construction and interpretation of doubtless came from the "No. 2 bed" or such diagrams are explained at the end of Melbourne Formation, considered by Hay this paper. It is noted there that (with the to belong to the first stage but arrangement used in this figure) closer ap- by most other students believed to be late proach to a single vertical of observations or latest Pleistocene. This specimen is on a form used for comparison means included in this study through the courtesy closer approximation of proportions to of Dr. C. L. Gazin. In a list of Melbourne those being compared, which are arbitrarily fossils prepared by Gidley and published set in a straight vertical line. The diagram by Hay (1927) the entry "Felis centralis?" therefore shows that the proportions of the appears without comment. "Felis cen- Craighead metatarsal are almost exactly as tralis" is the Central American jaguar, in the recent jaguar, somewhat, but less, Panthera onca centralis. Gidley's label 1941], PLEISTOCENE FELINES OF NORTH AMERICA 7 on this specimen is "Felis cf. veronis Hay," pared this specimen with the (Asiatic) but it is probably the basis for his queried and the jaguar and, while he did not defi- identification as F. centralis. Both iden- nitely commit himself, he seems to have felt tifications appear to me to be correct to the that it was closer to the tiger.' He said point of recognizing the specimen as a that it differed from the jaguar in the pres- jaguar and placing F. veronis also in that ence of a "preanterior tubercle," less re- group. duced protocone, relatively lower main The specimen is a right lower jaw with cusp, and greater size. These distinctions the canine and P3-M1, the central part of are all invalid. Within a single subspecies, the ramus somewhat broken, the incisive recent jaguars may lack the "preanterior region and tip of the coronoid broken away tubercle" on P4 or may have one notably and lost, and the ends of the articular con- larger than in the type of F. veronis. The dyle abraded. This certainly is a development of the protocone averages jaguar and is in every respect within the about as in F. veronis and may be greater or known range of living Panthera onca. It less. In the series of P. onca milleri skulls can be almost exactly matched among re- the relative height of the main cusp of P4 cent specimens. It is smaller and more is almost exactly as in F. veronis. Al- lightly built throughout than the Tennessee though Hay's type is at about the maxi- jaw and these characters, together with the mum known size for P4 in recent jaguars, it rather feeble carnassial, do make it re- is within the range of the large race P. onca semble the small race P. onca centralis as palustris, in which the length of this tooth Gidley's apparent identification implies. reaches at least 33 mm. (Cabrera), which Nevertheless it is possible and I think prob- is the length of the type of F. veronis (to able that it belongs to the same group as two significant figures). the notably more massive Craighead Cav- It is difficult and occasionally impossible erns cats. Recent jaguars, as well as the to distinguish recent feline species from iso- Old World species of Panthera, show lated P4 alone, so that demonstration that marked , the females there is no definite distinction between this being smaller and more lightly built than the tooth in F. veronis and in P. onca does not males. The differences between this Mel- absolutely demonstrate an identity. It is, bourne specimen and those from Craig- however, very suggestive. F. veronis is head Caverns are about what would be ex- significantly smaller than P. atrox and pected if the former were a female and the significantly larger than any known fossil latter a male. For instance, as is evident in or recent pumas. Such North American Fig. 1, the Melbourne specimen agrees specimens as are of comparable size and closely with rather large females of P. onca more definitely identifiable now prove to milleri and the Craighead Caverns speci- belong to P. onca. Moreover P. onca is mens similarly agree with large males of now shown to occur at the same horizon the same subspecies. Moreover, as will be as the type of F. veronis and at a nearby shown, "Felis veronis" is a form comparable locality. All these considerations make it in size with that from Craighead Caverns as nearly certain as may be without the and this supposed species certainly lived in discovery of more complete upper denti- direct association with this Melbourne cat. tions that F. veronis is a jaguar doubtfully These facts seem to me to admit no other or not separable from Panthera onca. reasonable interpretation than that U. S. The animal to which the type of F. veronis Nat. Mus. No. 11470 is a female of "F. belonged was of about the size of the large veronis." lower jaw from Craighead Caverns. In Felis veronis Hay, 1919, was based on an eight specimens of P. onca milleri the ratio isolated left P4 from Vero, Florida, in the length Ml: length P4 ranges from .74 to .79, same bed as the Melbourne jaw but at a mean .755. Ml in the Craighead cat is different locality. Since the jaguar speci- 24.6 mm. in length, so that its P4 was about not P4 mens hitherto discussed do include 1 In subsequent non-technical accounts he called it direct comparison is impossible. Hay com- a tiger, without other specification. 8 AMERICAN MUSEUM NOVITATES [No. 1136

33 mm. in length, with a possible range of rence in the of a species of the about 31 to 331/2. P4 of F. veronis is 33 recent Felis, even sensu latissimo, is mm. long, and hence suited to be the upper evidently anomalous but has not been the carnassial of an animal of the size of the subject of very explicit comment. It was Craighead cat as is also evident in Fig. 4. perhaps on this basis, and certainly not F. augustus Leidy, 1872, was based on a from examination of the specimen, that small fragment of upper jaw with P4 and Matthew (e.g., 1909, p. 116) referred the

MILLIMETERS 96 17 18 19 20 25 30 35 40 45 50 ... I.I .-. MAXIMUM LENGTHS OF CARNASSIAL F'. F".ox TEETH -IW = UPPER 1-- -IF -. A = LOWIE 16 is

1.- 16 3 .o.p/1J7'r " RECENT 8--I 1~~~~~~88 A.A> o.r;///; JAGUARS

V F. ve,'or igs FOSSIL v MELB0URIs4E A U.5. N. M.NO. 11470 A A,M.NO. 35653 JAGUARS A PORT KENNEDY _ _ _ 3 - RANGE IN FOSSIL JAGUARS 152----~~ ------1 c. O-eygonensis RECENT PUMAS A .,,,. .**v Pc/ad9gQtl-o A ...... w bVY-u.mrr7o9e A f, hawver; Fg /'7e)'ec7ewl'. v V F. j0r79iCPUS FOSSIL W "S.' rn oo,'ehe&o;' A . % U.S.N.MI.NO.11990 PUMAS v A.M.NO.25541 A A.M. NO. 12398

--__I - RANGE IN FOSSIL PUMAS I.20 1.20 9.25 930 1.35 L40 1.45 1.0 1.55 1.60 1.65 1.70 LOGARITHMS Fig. 4. Logarithmic plot of absolute lengths (not ratios) of upper and lower in various large felines, as labeled. Dots in carets represent single specimens, horizontal lines groups of specimens, their number given below the line. P. atrox, sample from Rancho La Brea, data from Merriam and Stock. P. o. milleri, living in Matto Grosso, American Museum specimens. F. augustuts, type. F. veronis, type. Melbourne, unnumbered specimen from that locality in U.S. Nat. Mus. U.S.N.M. No. 11470, also from Melbourne. A.M. No. 32633, the Craighead Cat. Port Kennedy, specimen from the locality referred to F. inexpectata by Cope. F. c. oregonensis, living in northwestern U5. S., includes some specimens referred to F. c. hippolestes, data from C. H. Merriam and from J. C. Merriam and C. Stock. F. daggetti, type and referred, data from Merriam and Stock. F. bituminosa, type and referred, data from Merriam and Stock. F. hawveri, type, data from Stock. F. inexpectata, type, data from Cope. F. longicrus, type. "S." mooreheadi, type. U.S.N.M. No. 11890, from Cumberland Cave, data from Gidley and Gazin. A.M. No. 23541, from Seminole Field, Florida. A.M. No. 23541, from Conard Fissure, referred to "F. " by Brown. (Logarithmic plotting is used because it represents equal relative variation by equal horizontal distances.) the middle and posterior parts of P3, col- species questionably to "Machaerodus," a lected by F. B. Hayden on the Niobrara genus to which it cannot possibly belong. River, Nebraska. It was described as from The apparent explanation of the anomaly the "Loup Fork" and as of Miocene age. is that the age assignment, not the gen- This age assignment appears to have gone eric assignment (sensu lato), is wrong. unchallenged ever since and has prevented Leidy's "Loup Fork" materials included the careful comparison of the specimen with specimens from several quite distinct hori- the large Pleistocene felines. The occur- zons, ranging from Miocene to Pleistocene 1941] PLEISTOCENE FELINES OF NORTH AMERICA 9 in age. The present specimen can be re- difference, still less to validate specific ferred to the Pleistocene with little ques- separation. Of course a difference in the tion, on this evidence: average characters of groups, correspond- a.-Some unquestionably Pleistocene speci- ing with local racial distinction, is not ex- mens were recorded in the same way. cluded, but the specimens now available b.-The preservation of this specimen is show that the two are very closely related distinctly unlike that of any truly Tertiary speci- mens from the same region and like that of and do not permit any valid taxonomic dis- some Pleistocene specimens. tinction, so that there is little choice but to c.-In the enormous collections subsequently consider the two names as synonymous. made, nothing at all like this animal has turned In the U.S. National Museum collection up in the Tertiary but there are now a number of specimens known to be from the Pleistocene (or there are two uncatalogued specimens of possibly early post-Pleistocene) that are specifi- P4, both broken, of an essentially similar cally indistinguishable from this. feline, collected at Melbourne, Florida, by Like the type of Felis veronis, this speci- J. W. Gidley. One is 29.6 mm. in length men is not very exactly determinable in it- and the other, on which this dimension can- self but a general comparison of all the not be exactly measured, was about 33 mm. materials establishes the probability that in length. Both have the anteroexternal it belongs to the same group as F. veronis cuspule vestigial or absent, thus tending to and hence that it is one of the large Pleisto- eliminate this apparent difference from cene North American jaguars. When Hay "F. augustus," and in one the protocone, described F. veronis he said that F. augustus although broken, evidently projected for- "differs in various ways from the Vero ward as in "F. augustus." The specimens specimen," without saying in what ways are from the same bed as the type of F. and obviously depending on the supposed veronis and from a nearby locality and they difference in age rather than on actual com- tend to support the synonymy of that name parison. Such comparison is necessary with F. augustus. now that the two are seen to be of approxi- Freudenberg (1910) described a left mately the same age. ramus with P4-Mi of a "jaguar-like felid" P4 iS of almost precisely the same length from presumably Pleistocene deposits at in the two types, 33.1 mm. in the type of Tequixquiac, . The specimen itself "Felis augustus" and 33.4 mm. in the type was lost before Freudenberg's study was of "Felis veronis," but the latter is wider, published, but a photograph was preserved 18.7 mm. as opposed to 16.7 mm. In and reproduced by him. His remarks are keeping with this lesser width, the proto- not explicit, but seem to imply that he had cone in the type of "F. augustus" is smaller handled the specimen before it was lost, al- and it is directed somewhat more ante- though his measurements are given as from riorly. The anteroexternal cuspule is the photograph and so are subject to error. slightly smaller. The anterior border is As Freudenberg said, the photograph shows emarginate between the protocone and the a jaw completely jaguar-like, as far as it parastyle.1 These and greater distinc- goes, except that the coronoid process is tions occur between different individuals of peculiarly hook-shaped. If this character a single race of living jaguars, and the two is real, it is far more likely to be pathologi- fossils are otherwise almost identical in cal or otherwise anomalous than to be a size and structure. In our very homogene- feature of taxonomic value. ous milleri series the observed range (for If the measurements from the photo- only eight specimens) of the length:width graph are reliable, they indicate an Ml index for P4 iS 173-200. The difference in about 24-25 mm. in length (Freudenberg this respect between "F. veronis," index of says 25, the published photograph suggests type 179, and "F. augustus," index of type a slightly smaller size), while P4 is about 198, is therefore not enough, in itself, to 231/2-24 mm. in length (24 according to demonstrate or suggest even a local racial Freudenberg). The length ratio P4 :Ml was therefore probably between .94 and l I tentatively follow Wood (1929) in the nomen- clature of these cusps. 1.00, with approximately .96 as the most 10 AMERICAN MUSEUM NOVITATES [No. 1136 likely figure. Not entirely distinctive, of F. inexpectata and the two could be of one these ratios yet strongly suggest a jaguar race only on the assumption that the lower or P. atrox in distinction from the pumas jaw is a male of maximum size and the type (see Fig. 5). The indicated size is smaller a female of minimum size, not impossible than specimens surely referred to P. atrox but improbable. Recently Gidley and or its probable synonym imperialis, and Gazin (1938) have described what prob- compares closely with the large (male?) ably is a large individual of F. inexpectata, specimens of P. onca augusta, such as that and its lower jaw and teeth are smaller from Craighead Caverns. It is at about the than and different from Cope's referred maximum for living jaguars belonging to P. specimen (see Fig. 3). They noted the onca palustris and is, as far as I know, larger disproportion of the Port Kennedy lower than any living Mexican or Central Ameri- jaw and the possibility that it represents a can jaguars. different species. In the same deposit there was found an The Port Kennedy lower jaw includes upper jaw with I3, C, and P3-4. This only M1, the posterior part of P4, and the is at about the minimum size for P. atrox ramus below these teeth and is not ade- ("P. imperialis") from California. Freu- quate for absolute determination in this denberg concluded that it belonged to an difficult group. It does, however, differ animal of about the same size as that just from any known pumas and does not vis- discussed. I think that it belonged to a ibly differ from recent jaguars or the group slightly but distinctly larger animal. P4 of fossil jaguars described here, so that there of the animal represented by the mandible is some probability that it belongs to the would be almost surely between 30 and 34 latter group. mm. in length, with 33 as the most prob- These various occurrences of jaws and able figure. The Tequixquiac P4 measures teeth thus demonstrate that jaguars for- 35 mm. As an isolated occurrence, this merly ranged widely over eastern United slight difference would not suggest any States. Although the specimens differ taxonomic distinction, but it happens that considerably in size and proportions, they 35 mm. is just within the well-established do not differ more than do some individuals range of the California P. atrox while 33 is belonging to a single subspecies of recent not and is exactly right for typical P. onca jaguars and they do not serve to demon- augusta, as shown in Fig. 6. These facts strate that more than one group is repre- tend to support the possibility that the sented, although of course this might prove Tequixquiac lower jaw is a large extinct to be true were more adequate series from -true jaguar while the upper jaw is a female the various localities at hand. No specific P. atrox. At the same time two other pos- differences from Panthera onca are defi- sibilities are not excluded by these unsat- nitely shown, and the specimens must be isfactory data: (1) that both Tequixquiac referred to that species. On the basis of specimens belong to a small Mexican race the present materials no distinction from of P. atrox, or (2) that both belong to a very P. onca palustris can be conclusively dem- large extinct Mexican race of P. onca. onstrated. It is, however, manifestly im- Except in size, there is no reliable difference probable that this subspecies centering between P. onca and P. atrox in the parts around the Chaco should have occurred in represented by these specimens. eastern United States. Moreover the few "Felis" hyaenoides Freudenberg, from North American fossils are in several re- the same deposit, is a sabertooth and not a spects near the limits for the recent sub- feline. species so that an adequate series might One other specimen that requires com- show a significant difference in average ment is the fragmentary lower jaw from characters, although no such difference is Port Kennedy, Pennsylvania, referred by yet shown. It therefore seems justified Cope to Felis inexpectata (Cope, 1899, Pl. and is surely convenient to recognize this XXI, figs. la, lb). This belonged to an group tentatively as a distinct subspecies. animal considerably larger than the type "Felis augustus" is the oldest name applied 1941] PLEISTOCENE FELINES OF NORTH AMERICA 11

JAGUARS AND 0. .s ox 13 f=.aerox/ A?o. rnilleri 8 x IOOX LP4 A.NM.NO.32633U.S.N.M.NO.114-70 X L M 12 P C. 0oreYof7e/7S/S ) ~~~~~xs. ^, Llr"IO'7sa PUMAS x e* 'gge*e ' X F. hevvver'i X A.M. NO. 12398 X U.S.N.M.NO. 11890

I ~~~. afr0x 15 JAGUARS AND * ~ ~~~~~~~I1 0. r iller; i 8 X A.M.NO.32633 F'0.2 rox X U.S.N.M.NO.11470 J IOOX LP3

core90'7ensisC. ,,If ---4 LP 4 . '61*ur7"0nos& x S. c/a99e '*/ X PUMAS P. hawver/ x ) X A.M.NO. 12598

60 65 70 75 80 85 90 95 100 105 110 Fig. 5. Diagram of two dental indices in various large felines, as labeled. Crosses represent single specimens, lines samples of the size indicated by the number below the line. P. atrox, Rancho La Brea, data from Merriam and Stock. P. o. milleri, recent, Matto Grosso, specimens in American Museum. A.M. No. 32633, the Craighead Cat. U.S.N.M. No. 11470, Melbourne, Florida. F. c. oregonensis, living, western U. S., data from Merriam and Stock. F. bituminosa, type, data from Merriam and Stock. F. daggetti, type, data from Merriam and Stock. F. hawveri, type, data from Stock. A.M. No. 12398, Conard Fissure, Arkansas. U.S.N.M. No. 11890, Cumberland Cave, Maryland, data from Gidley and Gazin.

P Oi7c.9 m;7//er- P: a1r0x ..04 _.02 .0 .02 .04 .06 .oa .10 .12 .14 .16 .18 .20 LOGDFFER.ENC-

x x LENGT;A Pt

x_TYPE F. AUGUSTU9

XX LENGTH iT TEQUIXQUIAC. -CRAI6HEPD CAVSC;S Fig. 6. Ratio diagram of lengfths of P4 and in various large felines, as labeled. Isolated crosses represent single specimens and lines represent larger samples with a cross at the mean value. P. onca milleri, Recent, Matto Grosso, specimens in American Museum. P. atrox, Rancho La Brea, data from Merriam and Stock. Tequixquiac, two specimens from that locality, in Mexico, data from Freudenberg. Craighead Caverns, the Craighead Cat. to a probable member of this group, so that occur in the North American Pleistocene it may be provisionally designated as Pan- and the establishment of this as a fact is thera onca augusta. welcome confirmation of the theory. The positive identification of Panthera It is also interesting that the most north- onca adds another to the list of animals now ern and northeastern race now known, this mainly or typically South American but extinct P. onca augusta, so closely resembles also typically North American in the Pleis- the most southern and southeastern living tocene. In view of the zoogeographic his- race, P. o. palustris. As far as now known, tory of the two continents, it has long been the former ranged in latitude from abou-t assumed that the jaguar did migrate to 200 (if the Mexican specimen belongs here) South American from North America. It or 280 (Florida) to about 430 north, while was therefore to be expected that it would the latter is known from about 200 to 380 12 AMERICAN MUSEUM NOVITATES [No. 1136 south (fossil only in the southern part)- The accompanying diagrams (Figs. 4-6) remarkable concordance. The jaguars be- show some of the important relationships of tween these extremes are all smaller in size and proportion in the North American size. A similar tendency for the extreme fossil jaguars here discussed, as well as in northern and southern races to be largest the fossil pumas to be mentioned in more is seen among pumas. In detail on later pages. Perhaps the most jaguars are also known fossil in higher lati- interesting feature of the largely self- tudes than in North America, nearly to 520 explanatory diagrams is their demonstra- south, and the form there found, P. onca tion (Fig. 5) that the index 1OOX lengthP3/ mesembrina Cabrera, 1934, is larger than length P4, although it does not distinguish the forms ranging through the thirties in either continent. Finally the ally known P. atrox from P. onca, does in most and from the most extreme latitude, about 650 perhaps in all cases distinguish these two north, is P. atrox, largest of all. (Of course from any of the fossil and living pumas. it also ranged into relatively low latitudes It is more distinctive than the index 1OOX during the Glacial .) length P4/length Ml, which has previously LOWER JAWS P. onca milleri (recent) P. onca auigutsta (fossil) Observed range in Amer. Mus. U. S. Nat. Mus. eight specimens No. 32633 No. 11470 (including both (male?) (female?) miales and females) Length of iamus 203 ca. 173 162-206 Length of aniterior border of symphysis 68 51-69 Depth below anterior end of P4 42 ca. 34 33-40 Depth immediately posterior to Ml 43 37 35-41 Thickness posterior to Ml 17.5 16.0 13.5-18.5 C-Mi, inclusive 105.4 95.0 87.1-102.1 P3-MI, inclusive 65.7 56.5 54.1-59.7 { length ca. 26 ca. 20 19-26 Canine alveolus width ca. 22 ca. 18 16-21 p | length 18.2 16.1 14.0-17.3 width 9.0 8.3 7.4-9.5 p | lerngth 24.7 21.0 19.8-22.9 width 12.4 ca. 11 9.8-12.8 24.6 20.7 20.0-24.0 Ml lengthwidth 12.7 11.1 10.2-13.0 UPPER JAWS P. onca augutsta P. onca milleri (recent) (fossil) Observed ranige in eight specimens Am. Mus. No. 32635 (both rrmale and female) (male?) It, imiaximnuin anteropostelior dimenision of crown 6.9 5.9-6.9 12, same 8.0 6.6-7.9 I3, samiie 11.0 8.8-11.2 l 21.8 20.1-26.2 (mlinimiiumri for kniowni males: 21.7) Caninle alveolus wndgth 16.9 15.4-18.6 (iryiinimumii for knowni males: 16.4) Diastema, C-P3, exclusive 13.7 10.0-17.5 (miiinimuml- foi krnown mriales: 11.3) p2 f length 8.5 6.8-7.7 width 5.8 5.2-5.8 p3 length 19.6 18.2-21.2 width 10.1 9.1-11.4 Oblique distanlec, pirosthioni to lowest poinit on orbital rim1 102 85-110

1 This odd and otherwise rather useless dimension is taken to give a rough idea of coiiiparative gross size because it is onie of the lonigest readily definable bone dimensions that cail be taken oni the fossil. 1941]. PLEISTOCENE FELINES OF NORTH AMERICQA 13 been used for this purpose and which does complete roster of the dimensions of the tend to distinguish jaguars and pumas but better specimens of North American fossil shows considerable overlap for values of the jaguars and a comparison of them with a index from 93 to 95, inclusive. series of recent jaguars of one subspecies The preceding measurements give a more -(or lesser local race).

NORTH AMERICAN FOSSIL PUMAS Unlike jaguars, it has long been recog- male of one of the larger living races of nized that pumas occur in the Pleistocene pumas, e.g., F. concolor hippolestes or of North America. I do not now have any oregonensis. As far as I know, the only important undescribed specimens of this previously referred specimens permitting group and the purpose of the present sec- close determination of affinities are those tion is to point out that certain species or described by Gidley and Gazin (1938) supposed species not generally recognized from Cumberland Cave, Maryland, in- as such are pumas and to suggest some pos- cluding P4 and a lower jaw with P4 and sible relationships and identities within the Ml. P4 iS larger than the type of F. inex- group. pectata, but the size is within the probable Teeth of approximately puma-size found range even for a local race and the charac- in the Pleistocene of the eastern states have ters are otherwise closely similar. The usually been referred to or compared with lower jaw is, as Gidley and Gazin remarked, Felis inexpectata, founded by Cope on an definitely puma-like and not jaguar-like, imperfect P4 from Port Kennedy, Penn- although slightly larger than most or any sylvania. In his definitive description recent pumas. The relatively shallow jaw, (1899) Cope referred the species to Uncia, the definition of the anterior end of the a genus used by him essentially as Pan- masseteric fossa and its position distinctly thera is in this paper and by most recent posterior to Ml, and, especially, the small authors, but including the pumas as well as size of P4 relative to Ml strongly suggest the lions, tigers, , and jaguars. the pumas and practically exclude close re- Cope then referred to "Uncia inexpectata" lationship to the jaguars. These specimens a lower jaw and various isolated teeth and are from near the type locality of F. inex- skeletal parts. From his statements that pectata (about 175 miles) and occurred in inexpectata is a "large species of Uncia," similar geological circumstances. Subject the smaller skeletal parts "equal the aver- to the reservations imposed by the frag- age size of a jaguar," the species as a whole mentary and scanty material, they strongly "as large as the jaguar," etc., he apparently suggest that the name F. inexpectata be- based his final views more on these referred longs to a puma and that most of the speci- specimens than on the type, which is de- mens so identified are pumas. cidedly below average size for a jaguar and Smilodontopsis mooreheadi Hay, 1920, smaller than any but small females of the was based on another isolated P4 found in smallest races of jaguar, while it is of nor- the same region, at Cavetown, Maryland, mal size for a robust puma. I have sug- under similar circumstances, and as far as gested above that his referred lower jaw such a specimen can be identified it appears probably does not belong to F. inexpectata to be the same as F. inexpectata. and may well be a jaguar. This is probably It was apparently the belief that this true also of some of the referred skeletal tooth had no protocone that led Hay to parts, which I have not had occasion or op- consider it a sabertooth rather than a feline, portunity to examine and restudy. with the additional evidence of the pres- From the type alone it is almost impos- ence of a "preanterior lobe" (prostyle of sible to determine the affinities of F. inex- Merriam and Stock). At the least, how- pectata beyond the fact that it is an ad- ever, there is a protocone swelling and this vanced feline about as large as an average and the (broken) third root were in the fe- 14 AMERICAN MUSEUM NOVITATES [No. 1136 line position, quite different from the ma- kansas, with an isolated P4** and several chairodontine. The preanterior lobe or skeletal parts referred. The latter will be prostyle is not a distinct cusp, as it almost mentioned on a later page. As regards the always is in machairodontines, but only a type tooth Brown noted the close resem- slight rudiment such as is frequently pres- blance to F. inexpectata and at first referred ent in felines. Moreover the protocone this material to that species, but he later region shows distinct abrasion, probably erected a new species because P4 iS more post-mortem in origin, and it is possible massive, the protocone larger and farther and probable that a definite feline proto- forward, the tooth narrower, the "para- cone was present before this occurred.' In cone" (parametacone of Wood) larger, the short, this is a feline and not a machairo- "metacone" (metastyle of Wood) project- dontine tooth. ing farther backward and at a more obtuse Hay noted a resemblance to the type of angle. As between the two type specimens, Felis inexpectata, but said that the latter these distinctions are all real, but compari- had a protocone and that on it the angle de- son of a series of recent pumas and of sev- scending to the protocone from the outer eral fossils found since Brown made his cusp was stronger. The former distinction is now seen to be nonexistent or unreliable study shows that none of them is outside and the latter is slight and has no taxo- the probable variation of a single species or nomic value in allied recent forms. Hay's perhaps subspecies. The local races could specimen is slightly larger, being 26.9 mm. well be distinct, but the isolated specimens in length (or 26.5, according to Hay), while in hand do not and cannot prove this. Cope gives 24 mm. for his type, but this The following measurements show some-

"F. longicrus" F. inexpectata "Smilodontopsis U.S.N.M. No. 25 specimens type type. After mooreheadi" 11890. After of F. concolor Cope type Gidley and Gazin oregonensis Length P4 27.3 24 26.9 26.6 20.2-25.3 WidthP4 13.4 (broken) (abraded) 12.2 10. 7-13.2 again is not sufficiently pronounced in it- thing of the normal range of variation in self to distinguish subspecies and still less this and allied forms. species. In the same publication Brown described The synonymy is further supported by a mandibular ramus that he referred ques- comparison with U.S. Nat. Mus. No. 11890, tionably to Felis cougar (i.e., Felis concolor the P4 from Cumberland Cave referred couguar). The specimen, Amer. Mus. No. to F. inexpectata by Gidley and Gazin. 12398, was found with the type of F. longi- This specimen is almost as large as the crus (although clearly a different individ- type of "Smilodontopsis mooreheadi," 26.5 ual) and includes the canine and P3-M,. mm. in length (26.6 according to Gidley It certainly belongs to a puma of some sort and Gazin), and in general it is almost ex- and can be closely matched among modern actly like Hay's type except for the differ- pumas except that it is slightly larger than ence in preservation, not in original struc- any individual with which I have compared ture, of the protocone. it, without beings quite beyond the probable Felis longicrus Brown, 1908, is another range for living forms. The small canine, supposed species not definitely distin- high cheek-tooth crowns, shallow ramus, guished from F. inexpectata by the known and particularly the length ratios of P3:P4 specimens and representing a species or and P4:M, are characteristically puma-like subspecies of puma. It was based on an (see Fig. 5). Ml in this jaw is 20.0 mm. in isolated P4 * from the Conard Fissure, Ar- length. In twelve recent specimens of F.

1 Drs. Gazin and Colbert also examined the speci- men and agree with this statement. ** Misprinted P3 and P3 in different places in the * Misprinted MI in the original description. original description. 1941] PLEISTOCENE FELINES OF NORTH AMERICA 15 concolor oregonensis, the index 100 X LM, :P4 ence or the absence of specific distinction varies from 73 to 85 and averages 78.4. P4 within the general fossil puma group. For of this individual would thus probably be convenience the specific name may be very about 25.5 mm. in length and may have tentatively retained, preferably in the form been anything from 24 to 27 mm. Thus, Felis (Puma) inexpectata, although the even without allowing for the undoubtedly supposed species has not really been de- considerable individual variation, this fined except for the very inadequate dis- lower jaw could belong to an individual of tinction that it probably attained a larger about the size either of the type of F. size than do recent specimens of F. (P.) longicrus or that of F. inexpectata. It is concolor. certainly most improbable that two dis- Various remains of fossil pumas, some of tinct puma-like felids of so nearly the same them relatively complete and excellently size occurred together in Arkansas, and it preserved, have been found near the Pa- seems almost certain that this is the lower cific coast. These have received three jaw of "F. longicrus." specific names, Felis daagetti Merriam, 1918, This specimen compares very closely Felis hawveri Stock, 1918, and Felis bitu- with the lower jaw referred by Gidley minosa Merriam and Stock, 1932. I have and Gazin to F. inexpectata, both in form not examined any original materials of and, as the following measurements show, these supposed species and can make no in size: additions to the excellent descriptions U.S. Nat. Mus. given by Merriam and Stock, singly and 11890. together, and cannot make a revision of the Amer. Mus. After Gidley somewhat less satisfactory taxonomy. It No. 12398 and Gazin must be noted, however, that the published ( leingth 13.7 data do not compare F. hawveri and F. P3 i bituminosa and do not seem to demonstrate t width 7.0 ( length 17.1 17.2 any valid taxonomic distinction between P4g them, and that no comparison has been width 8.9 9.1 made with any of the eastern specimens of length 20.0 20.9 m F. longicrus, F. inexpectata, etc., which, as t width 9.2 9.6 far as they go are very similar to these western materials and could well represent These comparisons tend strongly to sup- races of the same species. It wouldl also be port the propositions (a) that F. longicrus advisable to consider the possibility that and F. inexpectata are synonymous and (b) F. bituminosa (= F. hawveri?) is merely the that all these forms are pumas. female of F. daggetti, a p9ssibility that If all these eastern specimens do belong seems to be rather supported than opposed to the same group, it was one that ranged by the published data. In any case, as and doubtless also averaged larger than do Merriam and Stock have shown, these the living pumas, but one that intergraded forms are true pumas and they may be with the latter in size and in all other known referred to Felis (Puma), either as distinct characters. Our knowledge is hardly yet species or as subspecies of the closely allied adequate to demonstrate either the pres- recent F. (P.) concolor. 16 AMERICAN MUSEUM NOVITATES [No. 1136

PANTHERA ATROX This species has been so fully described perhaps be to recognize the and compared and beautifully illustrated Panthera (Jaguarius) and to place P. atrox by Merriam and Stock (1932) that no addi- in it. tions are possible or necessary. It is men- Although the type of P. atrox is from tioned here in order to express the opinion, Natchez, Mississippi, the species has not based mainly on their data and consistent otherwise been identified with certainty with their views but not so definitely ex- from anywhere east of the Rocky Moun- pressed by them, that it is distinctly closer tains. A few limb and doubtful to P. onca than to the lions and tigers of fragments suggest that it may have ranged the Old World and that it could, indeed, be in eastern United States, but it is in any called an extinct giant jaguar.1 Despite case a rare fossil in this region. Similarly this affinity, Merriam and Stock have well the large jaguars here called P. onca au- shown the P. atrox is sharply distinct in size range and in some characters of shape gusta have not yet been well identified from and proportion from P. onca (see, for in- west of Nebraska, although some very du- stance, Fig. 10), and I agree that it merits bious limb fragments suggest its possible specific distinction. The clearest taxo- presence in the farther west and it may nomic expression of these views would have occurred in Mexico.

NOTES ON LIMB BONES The preceding studies are based mainly land Cave. These elements are said to be on jaws and dentitions. The study of too big to belong to a (recent) jaguar or skeletal parts is more difficult and less con- puma, although smaller than any homolo- clusive because they are poorly known ex- gous P. atrox bones from Rancho La Brea. cept in P. atrox, because they are some- These various elements are not wholly diag- times less characteristic, and because ade- nostic as to affinities with living pumas, quate series of comparable recent speci- jaguars, and P. atrox, especially as they mens are difficult to obtain and have not evidently belong to some group distinct been available to me. Nevertheless scat- from any of these. They appear to me, on tered skeletal remains afford some dis- the whole, at least as puma-like as like the tributional data, or at least hints, that are other two groups mentioned. The third of interest and a few of them warrant men- metatarsal is stouter than in living pumas, tion. and hence is jaguar-like in this respect, Gidley and Gazin (1938) referred to but all the bones otherwise agree fairly "Felis near atrox" scapholunar, second well with puma except in size. Their pro- metacarpal, calcaneum, astragalus, and portions, not only of the bones singly but third and fifth metatarsals from Cumber- also in comparison with each other, sup- posing all to represent a single species, are 1 The habit of calling the machairodonts or saber- tooths "tigers" and P. atrox a "lion" in the vernacular distinctive but are more like those of a is misleading, to the point of confusing scientific thought as well as misinforming the non-scientific, puma than of a jaguar or P. atrox. and it should be abandoned. To most Americans, both South and North American, the American "lion" This is shown by the accompanying is the puma and the American "tiger" is the jaguar. P. atrox is certainly nearer the American "tiger" than ratio-diagram, Fig. 7, the construction and the American "lion." Many paleontologists and interpretation of which are easy to accom- laymen tend to think of it as similar to the African lion, and the usual restorations are simply drawings plish but the principle of which is some- of an African lion slightly modified in proportions. what to in is The known facts would equally well or better justify difficult explain words and restoration as a jaguar of suitably modified propor- discussed in the last pages of this paper. tions, which woulld produce a profoundly different picture and habit of thought about the animal. An essential point is that if measurements Of course the sabertooths were neither lions nor tigers, of one individual, or means for a species, a fact so well known to paleontologists that calling them "tigers" is not confusing to the well-informed, are set up in a vertical line, then series of although it certainly creates confusion among the non-scientific public. homologous measurements of other indi- 1941] PLEISTOCENE FELINES OF NORTH AMERICA 17 viduals with similar proportions will also semble the puma most and could well be- tend to fall into vertical lines, farther to the long to a close ally of F. concolor, somewhat right or left in proportion as the animal is larger than the living members of the spe- larger or smaller than that used as a stand- cies. They are of about the right size to ard of comparison. Here the Cumberland belong to F. inexpectata if, as is to be ex- Cave specimens are treated as standard and pected, that species was relatively small- their dimensions set in a single vertical headed, as is its ally F. concolor. line. Then the more nearly another set of If these are skeletal elements of F. measurements comes to vertical arrange- inexpectata, they tend to confirm the dis- ment, the more nearly its proportions ap- tinction of that species by showing that its proximate those of the Cumberland Cave feet may have had slightly different pro- material. None of the three comparisons portions than in F. concolor. The most gives close approach to a vertical, but the definite suggestions are that the third recent puma comes much nearer than do metatarsal (at least) was relatively shorter the jaguar or P. atrox, and hence shows and stouter in the extinct form and that

RECENT RECENT CUMBERLAND P. atrox JAGUAR PUMA CAVE SPECIMENS Ii -.,6 _.14 _.12 _.10_.08 _.06_.04 _.02 0 .02 04 .06 .0811.;.10 .12 .14 .16 .18I:.20 .22 .24 .26 .28 LOG DIFFERENCE SCALE LENGTH 0-~~ PROX.TRANS. DIAM. Mc r PROX. D.V. DIAM. I LENGTH '1 PROX.D.V.DIAM. }Mt 3 LENGTH MAX.PRFOX. DIAM. I M+

LENGTH ) WIDTH ON CALCANEUM I ASTR. FACETS) GREATEST DIAM. ASTRAGALUS LENGTH P 4

.70 .75 .80 .85 .90 95 1.00 IDS 1.10 US 1.20 1.25 ISO 135 1.40 1.45 150 1.60 1.70 1.80 1.90 200 RATIO SCALE Fig. 7. Ratio diagram of dimensions of foot bones and P' in various large felines, as labeled. Recent jaguar and puma, from one individual of each, in American Museum. Cumberland Cave specimens, data from Gidley and Gazin. P. atrox, large samples from Rancho La Brea, data from Merriam and Stock; horizontal lines show observed range. most nearly the proportions of the fossils. the second metacarpal (at least) had a Closer approach would hardly be expected relatively narrower head. in view of the probability that the fossils In describing F. longicrus, Brown re- are not all from one individual and are of a ferred to it an ulna, complete except for different species or subspecies from the the distal end, the distal end of a , living pumas. The diagram also shows that and various other isolated skeletal parts. the fossils are closer to living jaguars and The two principal specimens, figured by pumas than to P. atrox in size, and that Brown on his P1. XXIII, seem to me to be they have about the same average propor- correctly referred and to represent F. tions to a recent puma skeleton that the P' longicrus, which I believe to be the same as of F. inexpectata found in the same deposit F. inexpectata. They are very like F. con- has to P4 of the same recent puma. color and hardly differ except in being larger These bones do not belong to P. atrox, in and somewhat more elongate. They differ which the range of variation is well estab- more markedly from the jaguar or P. atrox. lished, and it is improbable that they be- In comparison with recent pumas as to long to P. onca. On the whole they re- proportions, they are appropriate in size 18 AMERICAN MUSEUM NOVITATES [No. 1136

CONAFRD .D, +rOx A.M. JAGUAR PUMA 11787A CALIFORNICUS _.e0 -.19 -16 -.14 .12 .10 .08 .D6 .04-02 0 .02 .04 06 .08 .10 .12 .14 .16 .18 .2 .22 LOG DIFFERENCE SCALE ~

6 LENGTH -- PROXIMAL TRANSVERSE l- . gDIAMETER. PROXIMAL DORSOVENTRAL 0 DIAMETER. LEAST TRANSVERSE DIAMETER OF SHAFT. -. GREATEST TRANSVERSE DIAMETER OF DISTAL END OF SHAFT.

FOURTH METACARPAL Fig. 8. Ratio diagram of dimensions of fourth metacarpal in various large felines, as labeled. Jaguar and puma, one specimen of each in American Museum. Conard, A.M. No. 11787A, one specimen from Conard Fissure, Arkansas. Smilodon californicus and P. atrox, large samples from Rancho La Brea, data from Merriam and Stock; horizontal lines show observed range. for F. inexpectata (or F. longicrus), as Fig. Smilodon californicus (on Merriam and 9B shows. Stock's data). Aside from these characters In the same collection, identified but not of size and proportion, however, there are described by Brown as F. longicrus, there is marked structural differences between the a fourth metacarpal of considerable inter- species here in question, and structurally est. Gidley and Gazin (1938, p. 53) men- the Conard Fissure bone is quite unlike tioned it and noted resemblance to a saber- Smilodon,1 as shown by the following table tooth. of some of these characters (again drawing The accompanying ratio diagram (Fig. on Merriam and Stock for data on Smilo- 8) shows that the proportions are indeed don and P. atrox). more like Smilodon than like any of the In these structural features P. atrox, P. large American true felines, notably in onca, and the Conard metacarpal IV are having the proximal and medial shaft trans- 1 In view of some previous misunderstandings re- verse widths relatively greater than in the garding metrical methods, it may be well to empha- size that these ratio diagrams do not directly or ade- latter, including even the stout-footed quately in themselves determine affinities. They show only what is puit into them: the comparative jaguar and P. atrox. In fact, of the five proportions of the parts measured. Similarity of pro- standard dimensions all but the distal portions is one of the important data for determining relationships, but it is not the only nor a sufficient width are within the observed range of datum.

Conard Fissure Smilodon P. atrox P. onca Specimen F. concolor External border of un- Straight As in P. atrox As in P. atrox Nearly straight ciform facet curved and P. onca Notch on internal Sharp and pro- Strong but less As in P. onca Rather obscure border of proximal nouncled sharp end obscure No dorsal indentation Deep, rlarrow in- Shallow, broad As in P. onca Shallow, narrow in- between unciform dentat"ion indentation dentation and MC III facets Unciform and MC III Continu(ous, almost As in P. atrox As in P. atrox Roughly parallel, facets continuous but sarrie Iplane and P. onca but MC III facet at a sharp angle sharply offset dis- tally relative to unciform facet Usually two distinct Facets usually con- More as in P. Nearly like P. Nearly like P. atro c facets for MC V, not tinuous, reaching atrox atrox or P. and P. onca reaching proximal proximal margin onca margin dorsally dorsally Palmar end of distal Less prominent, About as in P. More as in About interrne- keel pronounced, ending less atrox Smilodon diate in char-ac- ending abruptly abruptly ter 1941] PLEISTOCENE FELINES OF NORTH AMERICA 19 slight variants of the same type, while 13. At the same time it is larger than any Smilodon on one hand and F. concolor on the specimens now referred to P. onca augusta. other are markedly different. Relationship The fepaur from Harvey Co. reported by of the Conard specimen with either of the Hibbard as Felis cf. atrox in the same paper latter two is improbable. Even in its has the same size relationships, being Smilodon-like proportions it is definitely slightly smaller than Raincho La Brea P. more like P. onca than like F. concolor, atrox but larger than would be expected in and on the whole its general structure is P. onca augusta (in which this element is nearer to P. onca than to the very similar not known). P. atrox. It is also nearer to P. onca in size A fifth metacarpal from the same quarry and is roughly of the right size for a robust as the "Felis cf. imperialis" tooth seems, as male of P. onca augusta (see Fig. 9A). It described and figured by Hibbard, to be

RECENT JAGUAR 4. wtrox ISOLATED SPECIMENS OF FOSSIL JAGUARS 0 .02 .04 .06 .08 .10 .12 I*' .16 .18 .20 .22 .24 .26 .28 .30 LOG DIFFERENCE SCALE

CRAIGHEAD CAVERNS X LENGTH OF MANOIBULAR RAMUS WASHTUCKNA LAKE WIDTH OF DISTAL END OF HUMERUS

FOSSIL | A - LAKE Z' VVIIVFTH OF ULNA ACROgS 1IrMOID GLIRFACS X CONARD FISSURE i LENGTH OF FOURTH METACARPAL

x CRAIGHEAD CAVERNS LENGTH OF SECOND METATARSAL

RECEN1r ISOLATED SPECIMENS PUMA OF FOSSIL PUMIAS X CONARD FISSURE C F. /on,9CeuS ) WIDTH OF DISTAL END OF HUMERUS X CONARD FISSURE ("F. /on.lycru.s) WVVIDTH OF ULNA AT SIGMOID SURFACE B x RANCHO LA BREA, F. c/dg99efi LENGTH OF FEM-AUR x LENGTH OF SECOND METACARPAL

x CUMFBERELAND CAVE LENGTH OF THIRD METATARSAL LENGTH OF FIFTH MIETATARSAL x CUMBERLAND CaVE 'P. /o09gcfus' TYPE LENISGTH OF P i X FP. oU99**1. TYPE P. inPeI$OC*N a. TYPE Fig. 9. Ratio diagrams of dimensions of various scattered large fossil feline remains, as labeled, compared with recent jaguar, A, and puma, B. Recent jaguar and puma, one specimen each in American Museum. P. atrox, sample from Rancho La Brea, after Merriam and Stock; horizontal lines show observed range. Isolated specimens from localities shown, further specified in text and in Figs. 4, 5, and 7. may very well be a jaguar of this group, within the probable range of P. atrox and and it is very unlikely to belong to F. somewhat enhances the probability that longicrus or F. inexpectata. all these remains do belong to that species. Hibbard (1939) has recently reported A broken M1 evidently belonging to a puma various feline remains from Meade County, of normal size was found at the same place. Kansas. A left M1 was classed as "Felis cf. As far as they are known to me, none of imperialis" which according to Merriam the many discoveries in California suggests and Stock (whose study apparently was not the presence there of large felines other available to Hibbard) is indistinguishable than P. atrox and various pumas, but iso- from P. atrox. The tooth is said to meas- lated discoveries in Washington and Oregon ure 26 by 12.8 mm., hence is slightly smaller hint that a big true jaguar, like P. onca than the smallest Rancho La Brea speci- augusta, may possibly have occurred there. mens of P. atrox, for which Merriam and In the "Washtuckna Lake" collection from Stock give the minimum values 26.9 and Whitman County, Washington, there is an 20 AMERICAN MUSEUM NOVITATES [No. 1136

P. Onca P.p. "ON _.34..22...20 _.i.. .1. _.14_.12 ..10._O _XB..04..-01 0 .0I.0. .06 .08 .10 .12 LOG DIFFERENCE SCALE .a

- CONDYLOBASAL LENGTH OF SKULL BIZYGOMATIC BREADTH Of SKULL - IENGTH p - - LENGTH HUMERUS

-4 LENGTH ULNA

- LENGTH METACARPAL m

- LENGTH FEMUR

- LENGTH TIBIA

I LENGTH METATARSAL m11

.,6 .58 .60 .10 .80 .90 I.OO 1.05 1.10 1.20 1.30 lAO 1SO Fig. 10 Ratio diagram comparing various dimensions, as labeled, in a single recent jaguar and in the series of P. atrox described by Merriam and Stock. imperfect distal end of a humerus, prob- Mus. No. 8680, the proximal half of an ulna. ably the basis for Matthew's report of Morphologically this differs noticeably "Felix cf. imperialis" (1902). It is very from the puma but is almost exactly like a jaguar-like, as far as it goes, and on the recent jaguar except for its greater size. basis of jaguar proportions belonged to an The same is true of this part in P. atrox, animal of about the size of the type of P. which the fossil also resembles except for onca augusta (see Fig. 9A). smaller size. These relationships suggest In redescribing the Fossil Lake, Oregon, P. onca augusta and the ulna may belong material Elftman (1931) reported what he to or be allied with that group, although called "Felis sp. major" and "Felis sp. on the basis of jaguar proportions it is minor." The latter is smaller than the somewhat larger than would be expected puma and so, although of great interest, is for an animal as large as any definitely outside the scope of these notes. The identified as one of these great true jaguars former is known to me chiefly by Amer. (see Fig. 9A).

CONSPECTUS OF LARGE PLEISTOCENE FELINES OF NORTH AMERICA The following list includes most or all of 3.-Cumberland Cave, Md. Gidley and Gazin, 1938: Felis cf. inexpectata, P4 and the occurrences that have been published. lower jaw with P4-M1, and Felis near atrox, The numbers correspond with those of the isolated foot bones. The teeth almost surely accompanying map (Fig. 11). Each entry and the foot bones possibly belong to F. (P) includes the locality, principal reference inexpectata. The foot bones are not P. atrox. 4.-Craighead Caverns, Tenn. This paper: (often but not always the first published Panthera onca augusta, upper and lower jaws, notice), name used in that publication, na- etc. ture of the material, and the present iden- 5.-Vero, Fla. Hay, 1919: F. veronis, type tification or comments. P4. Probably a synonym of P. onca augusta. 1.-Port Kennedy, Pa. Cope, 1899: Uncia 6.-Melbourne, Fla. This paper: P. onca inexpectata, type P4, referred jaw with P4-Mi, augusta, lower jaw and various isolated teeth. various skeletal fragments. Type = Felis 7.-Seminole Field, Fla. Simpson, 1929: (Puma) inexpectata. Referred jaw perhaps Felis cf. veronis, several isolated teeth, and Panthera onca augusta? Felis cf. inexpectata, isolated P4. Probably P. 2.-Cavetown, Md. Hay, 1920: Felis onca augusta and F. (P.) inexpectata. couguar?, jaw fragment without teeth, and 8.-Sparta, Ill. Leidy, 1888: Felis con- Smilodontopsis mooreheadi, type P4. Both speci- color, cranium. Specimen not illustrated or re- inens probably belong to Felis (Puma) inex- studied. Possibly recent? pectata. 9.-Natchez, Miss. Leidy, 1853; Merriam 4-D

0rJ2=0

4 o;- 1 0 C3 Ca

-*D o

bo 22 AMERICAN MUSEUM NOVITATES [No. 1136 and Stock, 1932: Felis atrox, type, mandible getti, partial skull, jaws, foot bones. The pub- with broken canine and P3-Ml. Panthera lished data do not fully establish the presence (Jaguarius) atrox. of more than one species of puma, see above. 10.-Conard Fissure, Ark. Brown, 1908: 21. Samwell Cave, Calif. Furlong, 1906; Felis longicrus, type P4, referred P4 and lirnb Stock, 1918; Merriam and Stock, 1932: Felis fragments, and Felis coutgar (?), mandible with sp., hippolestes or near daggetti, skull. The C-Ml and end of humerus. Teeth and most specimen does not appear to have been fully limb material probably Felis (Puma) inexpec- described or exactly identified. tata. One or two limb bones may be P. onca 22. American Falls, Idaho. Gazin, 1935: augusta. Felis near atrox, metacarpal and radius. Some- 11.-Meade Co., Kansas. Hibbard, 1939: what small for P. atrox, affinities very uncertain. Felis cf. imperialis, Ml and fifth metacarpal, 23. Malheur Co., Oregon. Merriarn and and Felis cf. oregonensis, broken Ml. The big Stock, 1932: Felis cf. atrox, fragment of lower cat may be P. atrox, although somewhat aber- jaw without teeth. ranit. The smaller is apparently a puma, but 24. Fossil Lake, Oregon. Elftman, 1931: exact attiibution uncertain. Felis sp. majoIr, skeletal fragments. In part 12.-Harvey Co., Kansas. Hibbard, 1939. suggestive of a large jaguar, or peihaps small Felis cf. atrox, femur. Small for P. atrox but P. atrox, but exact identification very uncertaini. perhaps that species. 25. Whitman Co. (near Washtuckna Lake), 13.-Niobrara River, Neb. Leidy, 1873: Wash. Matthew, 1902: F. cf. imperialis, dis- Felis augustus, type, jaw fragment with P4, tal end of humerus, Felis cf. concolor, limb frag- humerus fragment. Panthera onca augusta, ments. The large cat may be a true jaguar, cf. Pleistocene, not Miocene as hitherto reported. P. onca aiugusta, but identification uncertain. 14.-Burnet Cave, N. M. Schultz and 26.-Fairbanks, . Frick, 1930: Felis Howard: 1935. Felis concolor hippolestes, limb atrox alaskensis, type. Not described. fragmeints. A large puma, but materials per- 27.-Tequixquiac, Mexico. Freudenberg, haps inadequate for subspecific identification. 1910: Felis imperialis, skull fragment with I3, Possibly post-Pleistocene. C, P3-4, Felis cf. imperialis, mandible with 15. Astor Pass, Nev. Merriam, 1915: Felis P4-M1, Felis atrox, upper jaw fragment without atrox, P4. teeth. The first and third specirnens probably 16.-Rancho La Brea, Calif. Merrliam and represent Panthera atrox, and the second is Stock, 1932: Felis atrox (including F. a. perhaps a jaguar, see above.' bebbi, type, and F. imperialis), all paits of skull 28.-San Luis, Mexico. Freudenberg, 1910: and skeleton, F. daggetti, type lower jaw, addi- Felis onca?, broken canine. Veiry dubious, tional jaws and limb bones, F. bituminosa, type perhaps a jaguar. skull anid jaws, additional skull, jaws, limb 29.-Mexico, exact locality unknowin. Freu- bones, F. concolor?, jaws. See discussion on pre- dernberg, 1910: F. concolor, mandible. The vious pages. idenitification is by Villada. Freudenberg had 17. McKittrick, Calif. J. R. Schultz, 1938: not seen the specimneni and it has niot beein figured Felis atrox, skulls, jaws, limib bonies, Felis dag- or described, nor are its age and geologic occur- getti, skull. rence known. The record has no scientific value 18.-Livermore Valley, Calif. Leidy, 1873; at present. Merriam and Stock, 1932: Felis imperialis, 30.-Grand Canal, Valley of Mexico. Os- type uppei jaw with P3. Female or small in- born, 1905: "laige true cat, puina-like," skull. dividual of P. atrox. This specimein has not been described or figured 19.-Potter Creek Cave, Calif. Sinclair, and the ideintificatiorn and age are too uncertain 1904; Bovard, 1907; Merrliaimi and Stock, 1932: to have positive value. Felis sp., atrox group, upper imiilk carnassial, 31. Herculaneum, Missouri. Olson, 1940: Felis sp. near daggetti, jaws, F. hippolestes, Felis cf. concolor, upper carnassial. skull, partial jaws. The situatioin regarding is to me fromyi 1 Felis hyaenoides Freudenberg, 1910, is usually the pumas nlot clear the published cited as another large Pleistocene feline, but it is data aind a single variety of F. concolor or of probaby machairodontine arid not feline. Freuden- F. daggetti (if that is valid) miay be present. berg apparenatly later recognized this probability be- 20. Hawver Cave, Calif. Stock, 1918; cause in other lists (e.g., 1922) he referred the species to SmilodoWtopsis. It is based oII an upper jaw frag- Merriiam and Stock, 1932: Felis hau veri, ment without teeth and is hardly identifiable in any type lowet jaw withi P3-Mi, Felis sp. near dag- case. 1941] PLEISTOCENE FELINES OF NORTH AMERICA 23

TAXONOMIC SUMMARY

The evidence reviewed on preceding Pleistocene fissure filling at Port Kennedy, pages suggests the following provisional Pennisylvania. Uncia inexpectata Cope, 1899. taxonomic arrangement of the forms dis- Felis longicrus Brown, 1908. Type, Amer. cussed in this paper. Mus. No. 11787, isolated P4, from Pleistocene fissure filling, Conard Fissure, Arkansas. Felis coutgar Brown, 1908, not Felis couguar Kerr, 1792. Smilodontopsis mooreheadi Hay, 1920. Type, U.S. Nat. Mus. No. 9212, isolated right P4, GENUS PANTHERA fromyi Pleistocene fissure fillinig at Cavetown, SUBGENUS JAGUARIUS Mar yland. Panthera (Jaguarius) atrox (Leidy) This species is inadequately distinguished from F. concolor. Felis atrox Leidy, 1853. Type, Phila. Acad. Nat. Sci. No. 12546, left lower jaw with brokeni C anid P3-M1, frorml presumably Pleistocene de- Felis (Puma) hawveri Stock posit at Natchez, Mississippi. Felis hawveri Stock, 1918. Type, Uniiv. Calif. Felis imperialis Leidy, 1873. Type, upper No. 10636, left lower jaw with P3-MI, from jaw with P3, from presurmiably Pleistocene gravel Pleistocene deposit in Hawver Cave, Californlia. at Livermoie Valley, California. Felis atrox var. bebbi Merriam, 1909. Type, This species is inadequately distinguished Univ. Calif. No. 14001, skull, from Pleistocene from F. inexpectata. asphalt pit at Rancho La Brea, Califoirlia. Varietal or subspecific distinction from typical P. atrox not demonstrated. Felis (Puma) daggetti Merriam Felis atrox alaskensis Frick, 1930. Nomen Felis daggetti Merriam, 1918. Type, Uniiv. nudum. Calif. No. 21572, left lower jaw with C and P3-MI, from Pleistocene asphalt pit at Rancho Panthera (Jaguarius) onca augusta La Brea, California. (Leidy) This species is inadequately distinguished Felis autgustus Leidy, 1872. Type, U.S. Nat. from F. inexpectata and may possibly be Mus. No. 125, upper jaw fragment with left P4 based on a robust male of the species to and broken P3, from "Loup Fork" (Pleistocerne which the type of F. hawveri belongs. not Miocene) of the Niobrara River, Nebraska. Felis veronis Hay, 1919. Type, U.S. Nat. Mus. No. 11411, isolated left P4, from Melbourne Felis (Puma) bituminosa Formation, Pleistocene, at Vero, Florida. Merriam and Stock Felis centralis? Gidley, in Hay, 1927, niot Felis centralis Mearns, 1901. Felis bituminosa Merriam and Stock, 1932. Type, Mus. No. X8628, essentially complete skull and jaws, from Pleistocene asphalt GENUS FELIS pit at Rancho La Brea, California. SUBGENUS PUMA This species is inadequately distin- Felis (Puma) inexpectata (Cope) guished from F. inexpectata or F. hawveri Crocuita inexpectata Cope, 1895. Type, Phila. and may possibly be based on a female of Acad. Nat. Sci. No. 52, isolated right P4, froImi F. daggetti.

EXPLANATION OF RATIO DIAGRAMS

F igs. 3 and 6-10 of this paper are constructed sent each of a number of anialogous observations on a principle that I have not seeni used else- by a single entry and to plot thern in such a way where and they require some explaniation. The that the horizontal distance between any two of method is onie that lends itself readily to several them will represent the ratio of either onie of useful types of graphic analysis and compari- those two to the other. A simple plotting of son and may therefore find wide application. calculated ratios, as in Fig. 5, has various and I have found it convenient as ain aid in dis- valuable pr operties, but it does not have the tinguishing species and determining affinities, in basic property sought for these more generalized sorting collections of bones, and in various ways ratio diagrams. It shows the ratios of various beyond those illustrated in this publication. observations to one fixed standard or between The basic purpose of the diagram is to repre- fixed single items in one series to similar single 24 AMERICAN MUSEUM NOVITATES [No. 1 136 items paired with these in a related series, but it A scale for reading ratios directly from the cannot show ratios between any two observa- diagram can easily be constructed and one tions among many. correct for the ratio figures in this paper is Consideration of the desired properties shows printed at the bottom of Figs. 7 and 10. By that the scale used must be logarithrnic and not copying this on a separate slip of paper, a mov- arithmetic. For instance, given absolute values able scale can be made and the diagrams have a= 1, b=2, and c=4, the distance plotted be- the property that if 1.00 on the ratio scale be tween a and b should be the sarne as between set at any specimen (whether the standar d or b and c, because a:b = b:c. This is true on not), the values of the ratios of all other speci- logarithmic but not on arithmetic coordinates. mens (set on the same horizontal) to this one Since it is desired to ignore absolute values and can at once be read on the scale. represent only ratios, the simplest approach is All the ratio diagrams are here reproduced to plot the logarithms of ratios. The logarithmn on the same scale, so that a scale traced froin of a ratio is the difference between the logarithms Figs. 7 and 10 is valid for all. In study it is of the two absolute measurements entering into convenient to adhere to a single scale as far as the ratio. The easiest method, then, is to plot possible. On study sheets, using arithmnetic the differences between logarithms. By using graph paper ruled in millimeters, I have found these differences as they are, not converting scaling log difference .02 as 10 millimeter s to be them to antilogs, a step may be saved arild also most convenient. ordinary arithmetic graph paper may be used, Points marked on a single horizontal line rep- since plotting logs on this gives the same result resent different values of one variate. In the as plotting antilogs on logarithmic paper. diagrams in this paper they represent homolo- For calculation, the direct measurements are gous dimensions of different specimens, but first conveited to their logarithms, three deci- any series of values of one variate can be used. mal places generally sufficing. Some one ob- The further and pelhaps greatest usefulness of servation is then taken as "standard," to repre- the method lies in the arrangement and inter- sent zero difference in logarithms which corre- pretation of diagrams in which several different sponds with the ratio 1.00. Observations variates are involved, each represented by one larger than this then fall to the right of it at horizontal series and these series arranged one distances determiinled by their ratios to it, and below the other. The horizontal single variate smaller observations similarly fall to the left. series are so placed that related points fall into a Although the differences are thus calculated single vertical linle. For instance, different from some one standard, the resulting diagram measurements on one individual may be placed shows not only ratios to that standard but also in a vertical line, as in Fig. 9, or mean values for ratios of any combinations of observations: a unified sample may be so placed, as in Fig. 10, once the diagram is made, the zero point, or or values for a group of specimens may be thus ratio 1.00 point, may be placed anywhere and arranged in order to test the hypothesis that they ratios of all other observations to that point represent one species and to see what com- will still be correctly represented. parable species they most nearly resemble in The arithmetic involved is much simpler than their ratios to each other, as in Fig. 7. calculating even one set of ratios, not to speak The diagrams so constructed have many of all possible sets, in spite of the fact that the different uses and properties, some of which resulting diagram does represent all possible will be mentioned arid others of which will be sets. For instanice, all the arithmetical calcula- seen as the method is used. tion for the top line of Fig. 7 is as follows: If a series of measurements involves the same

VARIATE: length of second metacarpal Measurement Difference from log of Specimen in mm. Log standard of comparison U.S.N.M. No. 12840 94 1.973 0 (This is the standard here used) Recent jaguar 68.3 1 .834 - .139 Recent puma 77.2 1.888 - .085 P. atrox from maxirimum 124.4 2.095 + .122 Rancho La Brea app. mean' 2.048 + .075 t minimum 100.4 2.002 + .029

These figures can then be plotted against the ratios as the standard of comparison, or, dealing log difference scale as shown. with two individual animals, if a second animal has the same proportions thioughout the parts 1 Data are from Merriam and Stock who do not give the mean or the figures from which it could he riceasur-ed as has the animal with which it is calculated. The range mid-point of the logarithrrms then this seiies is taken as a sufficiently close approximation to the being compaied, being conmpared logarithm of the mean. will also fall into a straight veitical line, regaid- 19411 PLEISTOCENE FELINES OF NORTH AMERICA 25

less of whether the two animals are of the same ing its Ml to the horizontal line on which P4 iS absolute size or not. The more nearly the graphed shows that P4 in this individual must comparative series approaches the proportions have been smaller than in P. atrox, larger than of the standard series, the nearer will the com- in P. onca milleri, and almost exactly the size of parative series come to falling into a vertical the type of "F. augustus." line. Thus in Fig. 7 the series representing Similarly the unknown P4 of the individual the puma, although not in a vertical line, is represented by the lower jaw from Tequixquiac obviously nearer to being in such a line than are (Freudenberg, 1910) was smaller than the P4 those for the jaguar and for P. atrox. It there- known from the same locality and was almost fore follows that the recent puma is, of these exactly the size of the type of "F. augustus." three, nearest to the proportions shown by the Or, working in the opposite direction, the dia- Cumberland Cave specimens. gram shows that the missing Ml of the type If, as is generally the case, it can be postulated specimen of "F. augustus" was larger than in that scattered and isolated specimens belonged P. onca milleri, smaller than in P. atrox, and to animals with about the same proportions as almost exactly as long as the lower carnassials some available standard of comparison, then known from Craighead Caverns and from the method makes it possible to compare the Tequixquiac. sizes of two or more animals known only by Finally when the sets of measurements used different specimens that are not homologous represent individuals or species that do, in fact, and that cannot be directed compared. Thus in have different proportions, such diagrams show Fig. 9 it is a permissible postulate that the fossil in a simple and immediately apparent way what jaguars had approximately the proportions of a these differences are. Thus Fig. 10 compares recent jaguar. The mandible and the meta- means and observed ranges for some of the tarsal from Craighead Caverns, the humerus most important dimensions in P. atrox with the fragment from Washtuckna Lake, and the meta- same dimensions in one specimen of recent P. carpal from Conard Fissure fall in the diagram onca. The comparison would, of course, have near a single vertical line. They therefore are been more reliable if averages for a series of shown to have belonged to animals of almost P. onc2 were used, but for the limbs no such the same size. The ulna from Fossil Lake be- series is available to me and the individual com- longed, on this postulate, to a larger animal, parison suffices for present purposes. From but the probable amount of variation (as shown, this diagram it is at once obvious that the re- for instance, by the comparable series for P. cent jaguar has' the skull broader relative to its atrox) is such that it still is of a size probable for length than in P. atrox, the distal limb seg- members of the same species. ments shorter relative to humerus and femur, A related use of the method is to estimate the hind-limb longer relative to fore-limb but meta- relative sizes of missing parts of specimens, a tarsals about the same relative to metacarpals, procedure illustrated in Fig. 6. In this diagram carnassial about the same relative to gross size the vertical alignment is on the means for of skull, being larger relative to skull length P. onca milleri. The fact that the means for and shorter relative to skull breadth, and so on. P. atrox also are almost precisely on a vertical The proportions of any two of these variates in line shows that in that species the average ratio P. onca as compared with P. atrox can be directly of length P4 to length Ml was the same as in observed in the chart without further, work. P. onca milleri. By plotting other specimens To make similar comparison by non-graphic either of P4 or of Ml, the most probable relative means would involve the arithmetic deter- size of missing carnassials of the same animals is shown by a position on a vertical line through mination, for this example, of at least 108 differ- the point indicating a known carnassial. Thus ent ratios, and when this labor was completed P4 iS unknown in the Craighead Caverns cat, the results still would not be as clear and usable but erecting a vertical from the point represent- as they are in this one simple diagram.

REFERENCES BOVARD, J. R. CABRERA, A. 1907. "Notes on Quaternary Felidae from 1934. "Los yaguares vivientes y extinguidos California." Univ. Calif. Pub. Dept. de la America austral." Notas Pre- Geol., V, pp. 155-170. lim. Mus. La Plata (), II, pp. BROWN, B. 9-39. 1908. "The Conard Fissure, a Pleistocene COPE, E. D. bone deposit in northern Arkansas: 1895. "The fossil vertebrates from the fissure with descriptions of two new genera at Port Kennedy, Pa." Proc. Acad. and twenty new species of ." Nat. Sci. Philadelphia, 1895, pp. 446- Mem. Amer. Mus. Nat. Hist., IX, 451. ("Crocuta" inexpectata: defini- Pt. IV, pp. 157-208. (F. longicrus, tive description in Cope, 1899.) etc.) 1899. "Vertebrate remains from Port Ken- 26 AMERICAN MUSEUM NOVITA TES [No. 1136

nedy bone deposit." Jour. Acad. LEIDY, J. Nat. Sci. Philadelphia, XI, Pt. 2, 1853. "Descr iption of arn extiriet species of pp. 193-267. (F. inexpectata, fullest Am-erican lion: Felis atrox." Trans. description and figures.) Airier. Phil. Soc., N.S., X, pp. 319- ELFTMAN, H. 0. 321. 1931. "Pleistocene mammals of Fossil Lake, 1872. "Remarks on somne extinct verte- Oregon." Amer. Mus. Novitates, No. brates." Proc. Acad. Nat. Sci. Phila- 481. ("Felis sp. major," possibly a delphia, 1872, pp. 38-40. (F. augus- jaguar.) tus, figured and mor-e fully described FREUDENBERG, W. in Leidy, 1873a.) 1910. "Die Saugetierfauna des Pliocans 1873a. "Conitiributions to the extinct ver te- und Postpliocrins von Mexico. I. brate fauna of the Wester n Terri- Carnivoren." Geol. Pal. Abb., N.F., tor ies." Rept. U. S. Geol. Surv., IX, pp. 195-231. (Including seveiral Hayden, I, pp. 14-358. (Pr incipal large felines.) reference for F. augustus.) 1922. "Die Sdiugetierfaujia des Plioc-inos 1873b. "Remar-ks on extiinct animals firom unid Postpliochins von Mexico. II Californiia." Proc. Acad. Nat. Sci. Teil: Mastodonten und Elephanten." Philadelphia, 1873, pp. 259-260. (F. Geol. Pal. Abh., N.F., XIV, Heft 3, imperialis.) pp. 1-76. (Lists some felines, without 1888. "On) a fossil of the puma." Proc. descriptions.) Acad. Nat. Sci. Philadelphia, 1888, FRICK, C. pp. 9-10. 1930. "Alaska's frozen fauna." Nat. Hist., MATTHEW, W. D. XXX, pp. 71-80. (F. atrox alasken- 1902. "List of the Pleistocenie fauna from sis, no technical description.) Hay Sprinigs, Nebr aska." Bull. Airier. FURLONG, E. L. Mus. Nat. Hist., XVI, pp. 317-322. 1906. "The exploration of the Samwell (Lists "F. cf. imperialis" from Wash- Cave." Amer. Jour. Sci., (4) XXII, tuckna Lake, Washington.) pp. 235-247. (Menitioins var ious fel- MERRIAM, J. C. ine3.) 1909. "The skull and deiltitioin of ani extinct GAZIN, C. L. cat closely allied to Felis atrox Leidy." 1935. "Annotated list of Pleistocene Mam- Bull. Dept. Geol. Univ. Calif., V, malia from Ainerican Falls, Idaho." pp. 291-304. (F. a. bebbi; see Merriam Jour. Washington Acad. Sci., XXV, and Stock, 1932.) pp. 297-302. (Felis near atrox.) 1915. "An occurrence of mammalian remainis GIDLEY, J. W., arid GAZIN, C. L. in a Pleistocene lake deposit at Astor 1938. "The Pleistocene vertebrate faunla Pass, near Pyr-amid Lake, Nevada." from Currilberland Cave, Marylaind." Univ. Calif. Pub. Dept. Geol., VIII, U. S. Nat. Mus. Bull. 171, pp. I-VI, pp. 377-384. (F. atrox.) 1-99. (F. inexpectata, etc.) 1918. "New puma-like cat from Rancho La HAY, 0. P. Brea." Bull. Dept. Geol. Univ. Calif., 1919. "Description of some mammalian X, pp. 535-537. (F. daggetti; see also and fish reiriains from Floiida, prob- Merriain and Stock, 1932.) ably of Pleistocene age." Proc. U. S. MERRIAM, J. C., and STOCK, C. Nat. Mus., LVI, pp. 103-112. (F. 1932. "The Felidae of Rancho La Brea." veronis.) Carnegie Inst., Washington, Pub. No. 1920. "Desciription of some Pleistocene 422. vertebrates found in the Uniited States." Proc. U. S. Nat. NELSON, E. W., and GOLDMAN, E. A. Mus., 1933. "Revision of the jaguars." Jour. LVIII, pp. 83-146. ("Smilodontopsis mooreheadi. ") Mammalogy, XIV, pp. 221-240. (Liv- 1923. "The Pleistocene of Noirth America ing forms only.) and its vertebrated animiials from the OLSON, E. C. states east of the Mississippi River 1940. "A fauna fromn Her- and from the Canadian provinces east culaneum, Missouri." Jour. Geol., of longitude 95°." Carnegie Inst., XLVIII, pp. 32-56. (F. cf. concolor.) Washington, Pub. No. 322. OSBORN, H. F. 1927. "The Pleistocene of the western 1905. "Recent veirtebr ate paleonitology. Fos- region of North Amer ica and its sil mammals of Mexico." Science, vertebrated animals." Carnegie Inst., (N. S.) XXI, pp. 931-932. (Mentions Washington, Pub. No. 322B. "Puma-like" skull.) HIBBARD, C. W. SCHULTZ, C. B., and HOWARD, E. B. 1939. "Notes on some mammals from the 1935. "The fauna of Burinet Cave, Guadalupe Pleistocene of Kansas." Trans. Kan- Moulntains, New Mexico." Proc. Acad. sas Acad. Sci., XLII, pp. 463-479 Nat. Sci. Philadelphia, LXXXVII, (F. cf. atrox, etc.) pp. 273-298. (F. concolor.) 1941] PLEIS7T'OENE FELINES OF NORTH AMERICA 27

SCHULTZ, J. R. Hist., LVI, pp. 561-599. (F. cf. 1938. "A late Quaternary fauna veronis and cf. inexpectata.) from the tar seeps of McKittrick, SINCLAIR, W. J. California." Carnegie Inst. Wash- 1904. "The exploration of the Potter Creek ington, Pub. No. 487, pp. 111-215. Cave." Univ. Calif. Pub. Amer. (F. atrox and Arch. Ethn., II, pp. 1-27. (Mentions daggetti.) various felines.) SIMPSON, G. G. STOCK, C. 1929. "Pleistocene mammalian fauna of the 1918. "The Pleistocene fauna of Hawver Seminole Field, Pinellas County, Cave." Univ. Calif. Pub. Dept. Florida." Bull. Amer. Mus. Nat. Geol., X, pp. 461-515. (F. hawveri.)