AMERICAN MUSEUM NOVITATES Published by Number 1123 THE AMERICAN MUSEUM OF NATURAL HISTORY June 11, 1941 New York City

THE SPECIES OF HOPLOPHONEUS BY GEORGE GAYLORD SIMPSON Recent expeditions in charge of Albert clear away some of the existing confusion Thomson have added several fine specimens and has some aspects of general interest. of Hoplophoneus to the American Museum A summary of its results is therefore pre- collections from the "Oreodon beds," lower sented in the following comments which Brul6 or Orella, of the Big Badlands of are merely notes, omitting most of the de- South Dakota. Routine identification of tailed data on which they are based and these specimens for cataloguing revealed making no pretense of monographic treat- that none of the supposed species of this ment or conclusive value. genus has really been defined in a distinc- tive, valid way and led to extensive re- The morphology of the genus has been search on the literature and on our speci- so well described and illustrated in the mens. This study did not result in a defini- studies cited at the end of this paper that tive revision of the genus, but it does no illustrations need be given here.

TAXONOMIC HISTORY Taxonomy, especially that of fossil unduly high rank (although this may also vertebrates, tends to go through four be done), but that of recognizing too many stages of increasing sophistication. In the groups. Under this system it is unlikely first stage, now entirely past, taxonomic that two different groups will be falsely categories were used very broadly. Work- united, but it is inevitable that single ers were more concerned with resemblances groups will be falsely divided. Cope, than with differences and tended to place Williston, Adams, and Thorpe attacked together all animals that resembled each the taxonomy of Hoplophoneus while in other to any noteworthy degree. Students this historical stage, adding to Leidy's two then commonly failed to make real distinc- species seven others of which I believe tions, but they seldom made unreal ones. only one to be (somewhat doubtfully) In the history of Hoplophoneus this stage valid. involves only Leidy's work. He described A third stage, now predominant in most two very different species now referred to paleontological work, is a natural reaction this genus, and these, incidentally, appear from the second, with a tendency to swing to me to be the only species of the six de- back toward the first, which, however, scribed before 1920 that are surely valid. cannot now be reached again because of- In the second stage of taxonomy, which the great accumulation of impedimenta still persists as a lively historical hangover, inherited from the second. The concept of students are fascinated by differences, to variability here becomes basic in taxonomic which alone they pay much attention. work. Differences are noted, as before, They seem to assume that the taxonomic but unless they are very pronounced or, rank and reliability of the morphologic especially, qualitative as well as quantita- characters of a specimen are in direct pro- tive, they are commonly denied taxonomk portion to the magnitude of their devia- value, or it is believed that their taxonomic tions from those of allied specimens. This value cannot be determined. A defeatist leads to splitting in a special sense, not nec- attitude regarding the recognizability of essarily that of giving taxonomic groups real minor groups is common and it is often 2 AMERICAN MUSEUM NOVITATES [No. 1123 felt and sometimes said that paleonto- Leidy described a second, much larger logical 'species have no particular value or species as Drepanodon occidentalis. In cannot be recognized objectively and are 1873 Cope described a species Machaerodus convenient artifices rather than expressions oreodontis and in the following year re- of natural truths. In the history of Hop- moved it to a new genus Hoplophoneus, of lophoneus, Sinclair, Scott, and Jepsen have which it became the type. The original exemplified this stage, Jepsen formerly sug- definitions of both species and genus were gesting that the Linnaean system may based on errors or misconceptions but in be hopelessly unsuited for use on such ma- his definitive work of 1885 Cope gave an terial and Scott and Jepsen virtually aban- essentially correct description and placed doning the attempt to revise species and Leidy's two species in his genus (along considering only genera as really distinc- with a species from the John Day, not re- tive. viewed in the present paper). In 1895 The fourth stage, which is perhaps now Williston briefly described another Oligo- being reached in vertebrate paleontology, cene saber-tooth as Dinotomius atrox, approaches the subject from a combination making no comparison of his new genus of these earlier points of view and takes an with Hoplophoneus except to say that the essentially new attitude in the interpreta- coronoid process was much as in the latter. tion of their data. Resemblances and dif- The first general revision of the genus ferences are equally studied and stressed was that of Adams in 1896. He recognized and the concept of range of variation within five White River species: Leidy's two, a group is accepted as basic. Minor natu- Cope's one and two then described as new, ral groups, including species and even, H. robustus and H. insolens. He noted the at times, subspecies and races, are believed unquestionable synonymy of Dinotomius to be recognizable from paleontological with Hoplophoneus and placed Dinoto- materials and the method of their recogni- mius atrox in the synonymy of H. occiden- tion and study are to consider the collected talis. This synonymy was accepted by specimens as samples from which inferences Riggs (1896) in his detailed description of as to the probable limits of the natural Williston's material and was also accepted population are to be made by orderly by Williston, himself, and by all later stu- methods. Objective estimates of popula- dents. The only valid characters used by tion variation developed from the theory Adams to define the five species were those of probability and sampling are made. of size. He also mentioned differences in It is recognized that the differences be- the vertical or overhanging character of the tween two real groups may be less than occiput, but later study has shown that those within one of them and that it is not these differences, as far as real and not due simply the magnitude of differences that to distortion, are simple functions of size. determines their significance but the varia- Adams also mentioned the presence or bility involved and the associations be- absence of p2, but showed it to be variable tween different characters and different and later students have found that it is not types of data. The present paper does not clearly associated with size or other char- succeed in taking the taxonomy of Hoplo- acters and has no evident taxonomic value. phoneus fully into this stage, but it at- Of H. occidentalis he added that it had no tempts to direct study toward it. posterointernal cusp on the lower sectorial. The first species now placed in Hoplo- Riggs (1896) noted that this was an error phoneus was named by Leidy as Mach- as regards "D. atrox" and that it might airodus primaevus in 1851. He later placed validate that species if confirmed in H. it in Drepanodon, under the (mistaken) occidentalis. In fact this part of the tooth impression that this was the valid name is broken away in both the other specimens for European saber-tooths.' In 1869 known to Adams and the supposed charac- ter was simply a mistake. I Scott and Jepsen (1936) believed that Leidy's was the first use of this generic name and that he Drepanodon had previously been used at least once in therefore validated it as antedating Hoplophoneus in a different sense and that Hoplophoneus is the valid this sense. They have since found (pers. com.) that name of the genus here under discussion. 1941] THE SPECIES OF HOPLOPHONEUS 3

In 1920 Thorpe studied the group and Sinclair (1924) expressed dissatisfaction added three more species, H. latidens, H. with the large number of supposed species, marshi, and H. molossus. These are all most of them apparently living in the same well within the size range of the previously region at the same time, and showed that named species and were defined mainly on most of these supposed species intergraded the basis of various supposedly distinctive and that they did not differ more than do proportions and indices in the type skulls individuals of one species of living cats. and jaws. He explicitly reduced H. latidens to the The species so far mentioned were all synonymy of H. primaevus but otherwise from the lower Brul6 or its equivalent. made no attempt at revision. Jepsen Some later forms (John Day) were referred, (1933) again emphasized the difficulty of but these seem best placed in different specific definition in a thoughtful, pessi- genera and are not discussed here. The mistic digression on the unsuitability of first Hoplophoneus found in the older Linnacan taxonomy for paleontological Chadron was H. mentalis, described by use. He was then studying Eusmilus and Sinclair in 1921, and in 1926 Jepsen named did not discuss the species of Hoplophoneus another Chadron species, H. oharrai. This except to say that too many species were completed the list of supposed species of currently recognized. (Oligocene) Hoplophoneus so far proposed, In their description of the genus, Scott which thus includes: and Jepsen (1936) did not attempt specific revision, but said that the number of pro- 1.-H. primaevus (Leidy, 1851). posed species was assuredly too large. 2.-H. occidentalis (Leidy, 1869). They did list H. robustus, insolers, and 3.-H. oreodontis (Cope, 1873). 4.-H. atrox (Williston, 1895). atrox as synonyms of H. occidentalis, with- 5.-H. robustus Adams, 1896. out discussion, and omitted mention of H. 6.-H. in8olen8 Adams, 1896. latidens and marshi.l The other proposed 7.-H. latidens Thorpe, 1920. species were listed as if valid and H. oreo- 8.-H. marshi Thorpe, 1920. 9.-H. molossus Thorpe, 1920. dontis, primaevus, occidentalis, and oharrai 10.-H. mentalis Sinclair, 1921. were said to be "most distinctly marked 11.-H. oharrai Jepsen, 1926. as separate."

SIZE DISTRIBUTION AND VARIATION IN THE H. OREODONTIS-INSOLENS SERIES The first step in attempting the deter- stages comparable with the other specimens mination of probable true specific limits is in question and because the type and one to bring together as large and as homogene- of the referred specimens are not from ous a sample as possible. For this purpose South Dakota. Adult specimens from specimens from the lower Brul6 of the Big South Dakota referred to H. oreodontis are, Badlands of South Dakota referred or re- however, included in the present section. ferable to H. oreodontis, H. primaevus, H. Thorpe's three species are omitted from robustus, and H. inrsolens, as these species this basic sample for comparison, because have hitherto been distinguished, afford the they are not from South Dakota and be- best available basis for study. These form cause they are defined mainly by charac- a nearly continuous sequence in size. The ters of proportion requiring separate dis- still larger H. occidentalis proved, in the cussion. H. mentalis and oharrai are from course of this analysis, to be distinctly lower horizons and are therefore also kept separable from this series and so is omitted separate. from the analysis given here and is dis- The area covered by these specifications 1 It is probably a lap8us that they gave a copy of cussed on a later page. The type and two Thorpe's figure of the type of H. latidens over the important referred specimens of H. oreo- legend "Drepanodon moloasus" (their Fig. 7). This skull does not show the characters that they give as dontis are likewise omitted here because, as distinctive of H. rnolossus and so they probably did not mean to reduce H. latidens to the synonymy of the will be shown later, they are not in growth latter. 4 AMERICAN MUSEUM NOVITA TES [NO. 1123

OBSERVED RANGES IN LowE BRULE ADULT SPECIMENS OF Hoplophoneus FROM SOUTH DAKOTA, 'R EXCEPT H. occidentalis H. H. H. H. "oreodontis" "primaevus" "robustus" "insolens" Totals Variate N IR N R N R N R N R Prosthion-basion 3 1301--137 8 142-156 5 163-179 2 177-179 18 1301k-179 Prosthion-inion1 3 144-163 1 169 5 184-200 2 206-216 11 144-216 Prosthion-anterior edge of P3 3 36-41 6 41-47 7 49-55 2 51-55 18 36-55 Maximum diameter of upper canine alveolus 3 11.7--13.9 9 13.6-15.2 6 14-18.5 1 17.5 19 11.7-17.5 P3_M1 2 29.6--30.5 10 30.5-37.5 7 34.5-41.5 2 37.3-43.5 21 29.6-43.5 Length P3 3 9.7-10.1 9 9.8-12 6 11-12.8 2 12-14 20 9.7-14 Length P4 2 17.3--17.6 9 17.7-20.9 7 17.5-21.5 2 20.5-21.3 20 17.3-21.5 Palatal length 3 62-75 8 72-77 7 78-90 2 86 20 62-86 Breadth on post- orbital processes 2 56-57 2 4 63-70 4 67-76 2 76-81 12 56-81 Breadth of post- orbital constric- tion 2 28 8 23-34 5 31-36 2 35-36 17 23-36 Diastema, C_P3 1 11.5 7 11.5-14.5 4 13-16.3 0 12 11.5-16.3 Condylobasal length 2 135-1145 6 149-165 4 172-182 1 190 13 135-190 Orbit to alveolar margin 1 23 5 21" -26 3 23-261 0 9 211-261 Infraorbital fora- men to poste- rior edge of zy- gomatic process of maxilla 1 21 5 22-27 3 251--291 0 9 21-291-2 Length of face 1 47 5 47-58 5 53-58 1 67 12 47-67 Length of craniUm2 1 98 4 102-105 4 111-125 1 123 10 98-123 Length of man- dibular ramus 2 106-1112 5 110-123 6 130-137 1 142 14 106-142 Depth of flange 1 38 4 39-47 5 41-50 1 57 11 38-57 Diastema C-P4 2 28-25 5 29-37 5 34-39 1 391- 13 28-391- Depth under P4 2 19.4--21.5 5 20-24 5 23-26.6 1 26.2 13 19.4-26.6 Depth posterior to Ml 2 20-2(0.6 5 18-21 5 21.5-26 1 24.6 13 18-26 Coronoid height above condylar- 91 alveolar level 2 17-1' 2 4 161-20 4 15-20 1 17- 11 15-20 Length P4 2 9-11. 5 11-12.8 6 11-13.5 0 13 9-13.5 Length Ml 2 15.7--16 5 15-18 6 15.8-20.0 0 13 15-20 is only about 200 square miles, incompar- tically impossible that any geographic ably smaller than the average range of a varieties were developed in such an area, single species of cats today and, indeed, and unlikely that any would be preserved far smaller than the average for a single in it. It is quite possible that different subspecies or geographic race.3 It is prac- allied taxonomic groups might have oc- 1 Or, maximum diameter of skull. curred in different ecologic stations in the 2 The skull is taken in lateral projection, a line area, but this can be from drawn from prosthion to most distant point of con- only inferred the dyle and a vertical dropped to it from the most existence of such groups, not used as a anterior point of the orbital rim. The length anterior to this vertical is the facial length and that posterior basis for their separation and recognition. to it the cranial length. The measurement is dif- The geologic time covered by the lower ficult and subject to observational error, but it pro- vides important data not as well shown by any other Brule is considerable, since a thickness of as dimensions. 3 For instance, a single species of jaguar, perhaps much as four hundred feet of sediments is the most nearly comparable of recent American cats, ranges over more than five miillion square miles and apiece. Figures for the puma are still larger, and its local races, even iii the most split classification, those for the medium-sized to large Old World cats average over three hundred thousand square miles have the same order of magnitude. 1941] THE SPECIES OF HOPLOPHONEUS 5 included. A priori there is no reason why follow either those published by competent distinct species may not have followed authorities who recognized the four species each other here at different times. In fact, as separate or were made by me using their however, there is now no evidence that this criteria. It will later appear that they did occur and some evidence that it did probably, do not correspond with true tax- not. All four of the supposed species here onomic species. considered (and also H. occidentalis) have These observed ranges give the most im- definitely been recorded within a vertical mediately comprehensible picture of the distance of less than forty feet of conform- general situation for a practical worker. able strata and all the data suggest that For the present purpose, however, they are they were absolutely contemporaneous. It of less value than are actual distributions, is fairly well established that a succession of to be studied for evidences of bimodal or invading, sharply distinct species does not multimodal character, and statistical data occur. It is probable that slow evolution providing valid means of comparison of in situ did occur, but if so the known facts the relative amounts of variation of each almost preclude evolution of more than sub- variate. specific rank. This would be shown only Of the various distributions, some have by a secular shift in group means, and the no very clear mode, some have one, some material for its possible detection is not seem obscurely to have two, and some now in hand. clearly have two. It would take many Twenty-eight different continuous vari- pages and it is not necessary to give all ates of skull and jaws have been studied in these in full. The following examples detail. By combining trustworthy meas- cover the sorts of patterns shown: urements in the literature (particularly A.-Breadth at postorbital constriction Sinclair, 1924, and Jepsen, 1933) with (Mode obscure) others taken by myself and with one set (on the type of H. robustus) kindly taken Frequencies H. H. H. H. for me by Dr. C. L. Gazin, I have been able "oreo- "pri- "ro- "in- to obtain from nine to twenty-one indi- don- mae- bus- 801- vidual values for each of these variates. Value tis" Vu8" tus" ens" All The fact that measurements have been 23.0-24.9 12 1 taken by different hands introduces another 25.0-26.9 0 27.0-28.9 2 1 3 element of variation, but this is clearly too 29.0-30.9 1 1 insignificant to have affected materially 31.0-32.9 3 2 5 the results based on these data. Some 33.0-34.9 2 2 4 variates, such as breadth across the zygo- 35.0-36.9 1 2 3 mata, were discarded because the material 17 afforded too few reliable values, and others, B.-P3-Ml like the distance from prosthion to bregma, (One well-defined mode) were given little emphasis because they proved to have little practical significance Frequencies or not to be comparable in different in- H. H. H. H. stances.' Value A"o.'" "p." "r.", "i." All 28.0-29.9 1 1 The preceding table gives the observed 30.0-31.9 1 2 3 ranges for most of the important variates 32.0-33.9 3 3 considered. The "specific" determinations 34.0-35.9 4 3 7 36.0-37.9 1 2 1 4 1 1 The bregma, for instance, is the intersection of 38.0-39.9 1 sagittal and coronal sutures. Its position is subject 40.0-41.9 1 1 to great non-significant variation from unimportant 42.0-43.9 1 1 fluctuations of these sutures. Moreover in adult specimens the coronal suture frequently cannot be ii exactly followed. In young individuals it crosses the midline near or at the anterior end of the sagittal 2 This remarkably small value, from Sinclair, may crest, but in old individuals it is commonly at some possibly be a misprint, although it is not definitely distance and usually an indeterminable distance beyond the possible range. In either case it makes no posterior to this point. important change in the general situation. 6 AMERICAN MUSEUM NOVITA TES [No. 1123

C.-Condylobasal length onomic terms. I think that they are simply (Two obscure modes) the males and females of the same species, Frequencies on the following grounds: H. H. H. H. 1.-From the field data it appears that these Value 0o." "P. "r.' .,I'i." All two groups lived for a considerable period in the 135-144 1 1 same region at the same time. They are com- 145-154 1 4 5 monly found in close association. It is almost 155-164 1 1 impossible for two species so closely similar in 165-174 1 2 3 all characters, actually intergrading in most, 175-184 2 2 to remain distinct when occupying the same 185-194 1 1 area. 13 2.-The two groups are about equally abun- dant, such discrepancy in numbers as exists in D.-Length of mandibular rarnus collections being far within the probable limits for chance sampling of a population in which (Two well-defined modes) they did occur in equal numbers. Male and fe- Frequencies male cats of one species tend to be about equally H. H. H. H. All abundant. When two distinct cat species oc- Value ". p. r. cupy one area, one is generally much more 105-109 1 1 abundant than the other in collections. 110-114 1 1 2 3.-The differences between the two groups 115-119 3 3 are closely similar in kind and in degree to those 120-124 1 1 between males and females of one species of cat 125-129 o in populations where the sex difference is known. 130-134 1 1 They do not correspond, or are not so closely 135-139 4 4 analogous, to differences usually existing be- 140-144 1 1 tween two species of cats occupying one area.1 - -4.If these are distinct species, then one or both of these two groups presumably contains both males and females.2 But if this were true, Comparison of these and of other dis- it would follow that these two species would tributions of linear dimensions (and also show less sexual dimorphism and less variation those of _proportions and qualita6tive char- within the species (sexes combined) than is yet acters, studied with equal care) show the known for homologous characters of any species following facts beyond much que Lstion: of felids, recent or fossil. 1.-The adult specimens referred to H. oreo- These considerations are in themselves dontis are in no way clearly distingiiished from almost conclusive. In order to test whether the H. primaevus group, and they a]ppear to be the amount of variation shown by this merely the two or three smallest in(lividuals of species, on the hypothesis here expressed, H. primaevus. 2.-The specimens placed in H. fnsolen8 are is consistent with that normal for a single similarly indistinguishable from the general species of cats, I have compiled and ana- sample identified as H. robustus an d probably lyzed a large number of data on recent are merely the largest individuals of the latter. lions, leopards, jaguars, and pumas and on 3.-The H. oreodontis-primaevus Smilodon the H. robustus-insolens group apy)roach each the extinct species californicus other closely in all characters and Itheir ranges and Panthera atrox. The results show con- widely overlap in many characters (inicluding clusively that the amount of variation all tooth dimensions). ; 4.-These two groups tend, neve rtheless, to shown by the whole H. oreodontis-pri- have distinctly different modes in m any charac- maevus-robustus-insolens series is of the ters, especially longitudinal bone dimensions same order of magnitude as in most single (and in proportions or indices that cein be shown species of cats. It is, if anything, rather to be correlated with these), and b3 comparng less than might be expected. These ex- all of these and the positions of indi- mens in the different distributions itidiua1Pele-t is possible~ tensive and complex data are not here pub- to divide the whole series into these two groups, lished in full. The following extract from leaving only one or two specimens c f somewhat doubtful position. 1 These statements are based on detailed study of many data on several species of recent and some of There are thus two, but not four, dif- fossil cats. 2 My records include about a dozen members of ferent groups included in thiu 3seriesseries Ofof each group. The chances that each group would be The next to consider is all of one sex and yet that the two should not be the specimens. thing two sexes of one species are so small as to be quite ,what these groups are in biologi1c and tax- negligible. 19411 THE SPECIES OF HOPLOPHONEUS 7

COMPARISON OF VARIABILITY IN Hoplophoneus primaevus (INCLUDING H. oreodontis, H. robustus, AND H. insolen8) WITH Two OTHER FELID SPECIES N R M v 1.-Condylobasal length of skull: a.-Mixed samples: H. primaevus 13 135-190 162.3 ^^4.4 15.9 ^3.1 9.8 1.9^ Smilodon^aliforrticu8 25 271-344 308.6 14.7 23.4 3.3 7.6 +1.1 Pantheraparduschui 8 179-245 206.0 ^9.1 25.7 ^^6.4 12.5 - 3.1 b.-Males only: H. primaevu8 6 165-190 177.5 =3.3 8.1 i2.3 4.6 1.3 P. p. chui 13 206-245 221.8 =i3.7 13.5 d=2.6 6.1 =1.2 c.-Females only: H. primaevus 7 135-157 149.4 ^ 2.6 6.9 ^1.8 4.6 ^ 1.2 2.-Length P4 a.-Mixed samples: H. primaevu8p a.30^20 17.3-21.5 19.53 1.34 .21 6.9 1.1 Smilodoncalifornicus ^^^.53^^.9^22 33.4-46.0 40.15 2.47 .37 6.1 Pantherappardu8 chui 8 22.4-27.9 25.19 .64 1.82 - .45 7.2 +1.8 b.-Males only: H. primaevus =^^.19^^^.279 19.0-21.5 20.43 .82 4.0 .9 P. P. chui 13 24.5-27.9 25.86 ^^.30 1.07 ^^.21 4.1 ^^.8 c.-Females only: H. primaevus 11 17.3-20.9 18.78 ^^.37 1.23 ^^.26 6.6 ^^1.4 them is sufficiently typical of the whole, made on Hoplophoneus.1 On the whole using one variate (skull length) in which this recent group is actually more variable sexual dimorphism is likely to be pro- than the combined Hoplophoneus sample, nounced and one (length of upper carnas- although one would expect the greater sial) in which it is usually more obscure. heterogeneity and the varying preservation The H. primaevus sample has been sexed of the fossils to exaggerate their variability. as explained above, considering H. "oreo- The P. p. chui sample is very homogeneous, dontis" and most H. "primaevus" (of all belonging not merely to one species but authors) as females, and one H. "pri- also to one local, interbreeding geographic maevus" (which according to its combined race living under almost completely uni- characters should probably have been form ecologic conditions. This leopard identified as a small "robustus") and H. sample happens to include only four adult "robustus" and H. "insolens" as males. females and in order to test the hypothesis The Smilodon californicus sample is that as to H. primaevus here advanced, a sample of Merriam and Stock (1932). It is, on the balanced as to sex was made by combining whole, less variable than would be ex- these with four males taken at random (by pected-there are several possible reasons shuffling and drawing numbered cards) for this, but it is not here important be- from the full series of thirteen adult males. ,cause, little variable as it is, the difference It is interesting that the pronounced from the whole Hoplophoneus sample is dimorphism of most of the larger felids has not probably significant. The Panthera smaller and darker, and to differ in various propor- pardus chui sample comprises seventeen tions. They were therefore distinguished as a sub- adultadultspcspecimensesrfrom thethpecies.Belgian Congo, "but the seriesThe procedureare highlyis,variableof course,andentirelythe meancorrect,dif- ferences are so slight that they are not statistically classified and annotated by Allen (1924). significant in any case. These series of specimens, On these and nineteen other recent leopard which include all of Allen's hypodigms of both species and the type of P. p. thus do not demon- skulls and associated Jaws, I have made all strate that the forest populationiturensis, is really distinct. A larger series might show this or might obliterate the the measurements homologous with those supposed difference. Incidentally, Allen exaggerated the difference by discarding the smallest old male of 1 An accidental result of the study of variation in P. p. chui on the grounds that it is a dwarf. The the Congo leopards was to show that the supposed specimen is not abnormal as far as any objective data subspecies Panthera pardus iturensis J. A. Allen, show and is merely a normal small variant. His 1924, is not demonstrably distinct from P. p. chui largest specimens of this series could with equal pro- (Heller, 1913). Allen sorted his specimens by habitat priety have been discarded as "giants" making P. p. into those of the rain forest (Ituri) and those of the iturensis obviously the same as chui. Even his small- adjacent savannah and brush country. The latter est iturensis adults are well within the probable were referred to chui and the former were found to be range of chui as far as his material shows. 8 AMERICAN MoISEUM NOVITAYTES [No. 1123 commonly led to the belief both popular riam and Stock (1932, pp. 165-167) have and scientific that two distinct sorts or remarked on the supposed occurrence to- species usually occur in one area. Thus gether of two species of Pleistocene felines, Allen (1924, p. 261) refers to what Pocock, Panthera atrox and Panthera imperialis, who inclined to believe them coriect, called over a large area including the Mexican "the oft-repeated statements of sportsmen plateau and California and they show that that two kinds of Leopards, larger ones these, again, are probably based in large called Panthers and smaller ones called part on sexual differences in one species. Leopards, occur in the same localities." The available evi(lence seems to me to Allen shows beyonid much doubt that the point to the conclusion that the difference "panthers" are males and the "leopards" between H. primaevus and H. robustus is females of the same race. Similarly Mer- simply another example of this sort.

PROPORTIONS AND INDICES The conielusionis of the preceding section with the results of the latter study, which were reached after the characters of rela- will now be briefly mentioned. Fifteen tive, rather than absolute, size had also different indlices, each of which has been been studie(d and they are in agreement used in the taxonomy of this genus or of OBSERVED RANGES OF INDICES IN SIUPPOSED SPECIES OF Hoplophone?is H. H, H. H. "oreodontis" " primaevIts" "robatstas" "insolens" Total Index N R N IR N R N R N R A. Prosthion-inion 3 110-119 1 116 5 112-116 2 116-121 11 110-121 Prosthion-basion B. Face 4 46-55 4 42-52 1 54 10 42-55 Cranium C. Breadth on Postorbital processes 4 41-47 4 39-46 2 43-45 12 39-47 Prosthion-basion 2 42-43 D. Breadth of Postorbital constriction 2 20.4-2 1.3 7 18.1-23.2 5 18.6-21.1 2 19.8-20.1 16 18.1-23.2 Prosthion-basion E. Length C alveolus 3 8.5-10 7 9.1-11.1 5 8.6-11.1 1 9.9 16 8.5-11.1 Prosthion-basion 1.1 F. Length P4 3 12.9-13.3 7 12.1-14.1 5 10.7-12.6 2 7.44-11.6 17 7.4-14.1 Prosthion-basion G. Flange depth 1 33.9 4 31.6-40.2 5 31.3-36.8 1 40.1 11 31.3-40.2 Ramus length H. Depth under P4 2 17.3-20.3 5 18.1-20.5 6 16.9-19.8 1 18.5 14 16.9-20.5 Ramus length I. Coronoid above condyle- level alveolar 2 15.2-118.4 4 14.7-16.2 4 11.2-14.7 1 12.3 11 11.2-18.4 Ramus length J. Length Ml 2 14.3-14.8 4 12.6-14.9 6 12.2-14.6 0 12 12.2-14.9 Ramus length 1941] THE SPECIES OF HOPLOPHONEUS 9 related forms, were studied carefully and To check this, the relations of propor- several others more summarily examined. tions to size in Hoplophoneus have been The observed ranges of ten indices for the examined in some detail and compared supposed species H. oreodontis, primaevus, with growth changes in Hoplophoneus and robustus, and insolens are given in the ac- with growth series and adult series in companying table. (In each case the index Smilodon californicus, Panthera atrox, Pan- is understood to be 100 times the ratio as thera onca, and Panthera pardus. Phleger labeled.) (1940) in an important paper on this sub- It is evident that all these indices are ject, has studied relative growth involved in highly variable and also that no one (or several analogous indices of the first three indeed no combination) of them would suf- of these species. For a rough determination fice to classify an individual specimen as of relative growth in Hoplophoneus, Amer. belonging to one and only one of these Mus. No. 9764, an immature animal with supposed species or even to place it surely the permanent dentition just starting to be as male or female, assuming that the four used, was compared with old individuals of "species" really represent two sexes of one comparable size as judged by the cheek species as concluded in the previous sec- teeth. From all these data the following tion. This is good confirmatory evidence conclusions were reached as to each of that these specimens do really belong to these ten indices: one species. Groupings based on size, as A.-These two dimensions are closely these "species" were, cannot also be de- correlated as would be expected from their fined by proportions and groupings based close structural relationship and in all the on proportions cannot also be defined by series examined their index has relatively size. In most cases there are no significant little variation, with no evidence of signifi- differences in indices between the various cant trend of change with growth or adult size groups and the group of which the most size. In Hoplophoneus one of the largest observations happen to be available often specimens happens to have the largest in- shows nearly or quite the whole observed dex, but this is probably a chance result. range for all. The series as a whole shows no significant There are, however, indices for which correlation of this index with size or other there is a fairly regular change in the characters. averages for smaller and for larger indi- B.-The facial: cranial index is highly viduals, even though the great variation variable but in Panthera pardus (both and the overlapping ranges make this ob- chui and the perhaps synonymous ituren- scure or undeterminable on the basis of a sis) there is a slight but real tendency for few specimens. For instance, index D of the index to increase with age and with this list is really and definitely smaller for size. This is, moreover, a widespread H. "insolens" (or for large males) than for tendency among mammals (and even H. "oreodontis" (or small females). Such among many lower vertebrates). The face changes may have taxonomic value or grows more rapidly than the cranium, so they may be correlated with size regardless that older and larger animals have rela- of taxonomic differences. It is now a well tively longer faces. The young A.M. No. established fact that proportions change 9764 also has a slightly shorter face, rela- during growth and that they may be cor- tive to the cranium, than do the most related with size, even among adults, in nearly comparable adults. In the adult such a way as to be determined by the size specimens the correlation is obscured by and not to be independent genetic charac- great variation and it is not statistically ters. Thus a difference in index accom- significant for the small available series of panied by a difference in size has no taxo- values (ten), but the visible trend is in the nomic significance unless it can be shown same direction. Its relatively short face to be different in kind or degree from what has been given as a distinctive character of normally occurs in a single species of the H. "oreodontis," but in the first place the group under consideration. index is within the wide range of variation 1() AMERICAN MUSEUM NOVITATES [No. 1123 for the larger supposed species, and in the for instance, and comparable adults k is second place the fact that it is below the not significantly different from 1.0, which average for the larger forms is just what would imply either that the canine grows would be expected within a single species. harmoniously with the skull or that this C.-This index shows no clearly signifi- animal is not really comparable but be- cant tren(I in any of the material examined. longs to a group with decidedly smaller It is not correlated with size, as far as can canines. Even among the mature speci- be seen from small samples, and it fluctu- mens the differences are not what would be ates erratically. It is the resultant of two expected and what occur with other tooth opposite factors and the ascendancy of one indices. The material of various growth over the other is not definite or predictable. stages is insufficient to explain the anom- The general width of this part of the skull aly, but it suggests the possibility that increases much more slowly than does the the canine alveolus (which is what is skull length, but on the other hand in measured, rather than positively homolo- middle stages of growth, at least, there is gous points on the canine itself) does in- an acceleration of the development of the deed grow in this group. The canine per- postorbital processes tlhemselves since these haps continues to move downward for some are barelv developed in young and small time during adult life, so that the part at individuals and are prominent in ol(1 and the alveolus changes an(l becomes pro- large ones. gressively larger. D.-In spite of the erratic overlap of the This peculiar situation, whatever its observed ranges, this index in every felid true cause, tends to obscure real andl examined shows a strong, indubitably definitive differences between the size of significant negative correlation with the the canine in different mature groups of gross size of the skull, or in other words the Hoplophoneus. Thus young males may breadth of the postorbital construction has tend to have relatively smaller canines (or strong negative heterogony. Phleger at least canine alveoli) than o0l females. found values of the lheterogony coefficient, There is, nevertheless, some difference, the k, of .42 for Panthera onca, .55 for Smilodon average index (E) for five specimens re- californicus, and .56 for Panthera atrox. garde(d as males being 10.3 while for nine Between our immature Hoplophoneus, regarded as females it is 10.0, but this is a A.M. No. 9764, and the type of H. robustus somewhat sinaller difference than is usual k is .41 and between this immature animal in cats an(l is not clearly significant. Tak- and the likewise comparable Princeton No. ing only in(livi(luals in late maturity or 12957 k is .45. Larger growth series of senility, in which further protrusion of Panthera pardus show that the interorbital canines is unlikely, would probably give breadth grows almost harmoniously for a higher averages for both males and females short time after birth but that by the time and probably show a greater mean differ- the full deciduous dentition is well in use ence between them, but I have only one its growth slows down abruptly and that it probable male specimen in this condition thereafter grows very slowly or hardly at (which (loes happen to have a very high all. The same tendency has been observed index, 11.1). These (lata are consistent in other mammals and it is probably gen- with the sexing of Hoplophoneus speci- eral. mens as I have carried this out, but they E. The size of the canine tends to be do not strongly confirmn its correctiness or positively, but only loosely, correlated with help in its performance.' adult size. It would be expected to show F. The length of p4, measured on the strong negative heterogony with growth, tooth itself, certainly (loes not grow sig- on the principle that the canine dloes not, nificantly after eruption. Its in(lex against in fact, grow while the skull (loes. It is 1 Eveni in r-ecenit cats that show ver'y stirong sex surprising that such a tendency is not dilnlorphisnl and decidedly different averages for this visible in our material of Hoplo- index in the two sexes, the ranges for the index often really overlap so that sex cainniot be suoely determined frorn phoneiis. Between Amer. Mus. No. 9764, the irndex itself. 19411 THE SPECIES OF HOPLOPHONEUS 11 skull length therefore becomes rapidly do overlap between these two values. The smaller as the individuals become larger in same tendency for large individuals to have a growth series, and the same decisively relatively small coronoid processes is seen significant negative correlation with skull in Smilodon californicus but I do not detect size is shown within the adult series not it in Panthera. It is probably related to only in these Hoplophoneus specimens but the great difference in jaw musculature be- also in Panthera atrox and P. pardus, in tween the machairodont and feline groups. which I have measured the correlation, and J.-This index is analogous with F and probably in all cats. In these animals the remarks about the latter are also true of males and females and small and large this. individuals of the same sex differ less in the Thorpe (1920) has used an index for the size of the cheek teeth than they do in skull relationship of the anteroposterior dimen- dimensions. The situation in this Hoplo- sion of the anterior zygomatic pedicle (from phoneus series is what would be expected infraorbital foramen to posterior margin of by analogy in a single species. the zygomatic process of the maxilla) to its G.-The index for relative depth of vertical dimension (from the orbital rim flange, although highly variable, shows no to the alveolar margin). He says that this evident trend in our material. As far as has considerable value in specific determi- shown, the flange has great individual vari- nation and uses it in defining H. latidens, ation but grows more or less in halmony said to have an index of 117 as against with the growth of the skull. This again 162.5 for H. robustus' and 143 for H. suggests that the upper canine grows (or is primaevus. This implies that the index is progressively more protruded) throughout relatively invariable within one species and much of the animal's life, for the canine tip is positively associated with other charac- never projects beyond the flange and yet ters of possible taxonomic value, but the flange is not relatively longer in young grounds for these assumptions are not than in old animals. given. I have investigated the point in H.-Contrary to expectation, this index fourteen specimens of Hoplophoneus and shows a low, suggestive but not conclusive, in twenty-two of Panthera. The index is negative correlation with length of ramus. highly variable within any one species (no There may thus be a tendency for larger matter how narrow the species be made, animals to have relatively more slender beyond placing every specimen in a dis- jaws, although this is too variable to be tinct species) and it does not show evident certain on the basis of a small series of association with any other possibly signifi- animals. P. pardus seems to show the same cant characters. Thus for specimens re- tendency, but here again the evidence is ferred by competent authorities (including not fully conclusive. Thorpe, himself) to H. primaevus and I.-This index shows a statistically surely referable to that species even if H. significant negative correlation with length oreodontis, H. robustus, etc., be kept dis- of mandible. As in all cases, there is much tinct, the index varies in known specimens variation, but large individuals of Hoplo- from 88.5 to 143. Instead of being distinc- phoneus unquestionably tend to have rela- tive in this respect, the type of H. latidens tively smaller coronoid processes. This happens, by accident, to be almost exactly may also be in part a sexual difference at the average for H. primaevus. since the difference between presumable Nor is there in my data any clear trend males and females may (doubtfully) be a or association of the values of this index, little more pronounced than would be ex- even in the loose way that the facial-cranial pected from size alone. Although it was index is associated with size. It is possible not primarily used for that purpose, the that the index tends to become smaller index permits sexing most specimens, an I But there is a lapsus somewhere, because the di- index (for an adult) greater than 15 usu- mensions that he gives would make the index 180.5. one less than which is so much above other known indices as to be ally belonging to a female and improbable. Even 162.5 is the highest known. to me, 14.5 uisually to a male. The distributions although not outside the probable range. 12 AMERICAN MUSEUM NOVITA TES [No. 1123 with increasing size (or age), but if so the measure of the variability of the character. correlation is very low and many more The sizes and proportions of limb bones specimens would be required to demon- and other skeletal parts have not been strate its reality (i.e., statistical signifi- studied in great detail. Although it is the cance). As it is, the four smallest speci- rule rather than the exception for skulls of mens for which I have data cover almost Iloplophoneus to be associated with skele- the entire known range for the genus, tal material, the number of homologous from 91 to 150. Their mean value, 120, post-cranial measurements available to me happens to be greater than the mean for the is not great enough to establish surely the four largest specimens, 103, but this could significance of deviations and the trends of be a mere accident of sampling. Even if associations. It is, however, evident that real, its taxonomic value would be more the general size grouping is consistent with than doubtful, for the differences bear no that of the skulls and can be interpreted relationship to the supposed species as in the same way. The characters some- determined by Adams, Thorpe, etc. For times considered of taxonomic value are instance, the types of latidens and molos- those of length of limb bones relative to sus and authoritatively identified, thor- the skull and of the stoutness of the bones oughly representative specimens of robus- relative to length. In recent cats (and in- tus, primaevus, and oreodontis all show in- deed in most mammals) these are both cor- dices within the relatively narrow range related with size within a species. Small 108-120. Moreover, indices in Pleistocene individuals tend to have relatively more Smilodon and even in recent leopards also slender limb bones but these bones tend occur in this range. The index seems to to be shorter relative to the skull, i.e., have less taxonomic value than almost any their width grows with positive heterog- other. ony relative to their length and the length This is not said in criticism of Thorpe, with positive heterogony relative to the but to emphasize a serious defect in most skull. What data are available for Hoplo- paleontological taxonomy, including much phoneus simply agree with this general of my own. We often tacitly assume that trend and suggest no basis for specific dis- any great, clear morphological distinction tinction. For instance, in the smaller has taxonomic value. In fact, as this ex- specimen Princeton No. 13628, referred to ample shows and as could be shown by H. "oreodontis," the index of distal width many more examples, this principle is against length of femur is 17 and that of thoroughly fallacious. A very slight mor- length of femur against condylobasal skull phological difference may have taxonomic length is 113. In the larger Amer. Mus. value and a very great difference may have No. 1406, referred by Matthew to H. none. The valid criterion is not the great- "robustus" and by Scott and Jepsen to ness of the distinction, but its association H. "primaevus,"'l both indices are larger, with different samples defined on other as would be predicted, 22 and 118, respec- grounds, taking into account some good tively.

QUALITATIVE CHARACTERS The only clear-cut qualitative differences with age since p2 happens to be quite ab- observed within Hoplophoneus are the pres- sent in one of the rare immature specimens ence or absence of p2 and of Il. Although and to be present in our most nearly senile Adams suggested that the presence of p2 individual. In specimens referred to each had some significance, he showed that it of the four supposed species here under was variable and later students have not consideration p2 is present in some and given it taxonomic value. It is more often present than not, but its absence has no 1 This is the specimen referred to in the first column with size or other of their table, 1936, p. 139. The number given there, clear association possible 1405, is a misprint and should be 1406. I believe that taxonomic characters, or even, apparently, this animal is a male. 1941] THE SPECIES OF HOPLOPHONEUS 13 ab-sent in others. Moreover, in specimens broken off in this specimen, (b) in some referred to H. primaevus (sensu stricto) and cats there is a stage of replacement wvhen to H. robustus, at least, it is present on one only two lower incisors are functioning and siide andI absent on the other. It also this is an immature animal, and (c) the -aries greatly in size and in position, some- otherwise very closely similar Amer. Mus. times a fairly large, two-rootedl tooth, No. 5338 has three lower incisors anid the sometimes a one-rooted vestige, sometimes most probable explanation of the present niear the caniine, and sometimes almost specimen is that it has anl individual overlapping P3. anomaly. WA-ood (1927) iefers to the speci- Amer. Mus. No. 9764, an immature in- men and notes that I, may be absent in dlividual placed by Matthew in H. oreodon- two known specimens of H. mentalis, but fis, appears to have only two incisors. As that this is somewlhat doubtful as to fact regards the upper jaw (Matthew, 1910, and if true is of doubtful significance. Fig. 3A), this is an error. The side figured is restored, and further preparation and ii is relatively reduced in Hoplophonteus stti(ly of the other side shows three incisor and there may have been some tendency alveoli. In the lower jaw only two incisor toward its loss, but this is not established roots are present but (a) I, has a short root and has no present significance for tax- in this genus and its alveolus is possibly onomy.

THE TYPES OF H. ROBUSTUS AND H. INSOLENS A(lams did not explicitly designate types tions three specimens and says that o)f sp)ecies and his descriptions, basedl on "either" (any?) of them are "typical": various specimens, are somewhat ambigu- a skeleton determined by Osborn and otUs in this respect. Although I believe Wortman as H. occidentalis, Princeton No. both these species to be synonyms of H. 11022, and Princeton No. 11372. Although primaevus, the designation of their types is not mentioned by number, either by Os- important. Adams' description of H. born and XVortman or by Adams, the first robustus mentions Amer. Mus. No. 650 specimen mentioned is almost certainly and Princeton No. 10647. He says that Amer. Mus. No. 655. The three specimens "tthe species is represented" by the latter are evidently syntypes in Adams' publica- but that the former, better specimen is tion. Princeton No. 11372, which includes "its most perfect type." Although he does a good skull, appears to be the most exactly not seem here to be using "type" in a tech- determinable specimen and is therefore iiical sense, there is little doubt that Amer. designated as lectotype. There is some Mklus. No. 650 was essentially his type and doubt whether Amer. Mus. No. 655 really it should be, and hereby is, designated belongs in the same group, but its skull is lectotype if both of these specimens are so poorly preserved that any conclusion considered syntypes. Dr. Matthew over- based on it is shaky. The lectotype, as looked the fact that Amer. Mus. No. 650 represented by measurements given by was the type or the better syntype of H. Jepsen (1933), is included in the preceding robustus and he exchanged it with the IJ. S. discussions of variation in the H. pri- National Museum, where it is now pre- maevus group. Amer. Mus. No. 655 is not, served. because it provides so few reliable measure- Adams' description of H. insolens men- ments anid is so uncertain in general. 14 AMERICAN MUSEUM NOVITATES [No. 1123

H. OREODONTIS The specimens discussed above and remain so until a series of specimens shows shown to be small females of H. primaevus the specific association of dmn4 with the per- do not include the type or two important manent dentition. The type does not war- referred specimens of H. oreodontis. rant the conclusion that H. oreodontis is a The type of this species, now Amer. small species, since its small size is purely Mus. No. 5337, is a fragment of the right juvenile. lower jaw with two unerupted incisors, It is extra-legal but is, I think, justified and with dM4 in place. P4 is forming in its to accept Amer. Mus. No. 5338 as neotvpe crypt, but so little of it is yet calcified that of this species.1 It was, Cope says, found its definitive form and size cannot be de- at the same horizon and locality as the type, termined. DM4, the most characteristic it cannot be shown to be of a different spe- part of the specimen, is not known to me cies, it was referred to this species by its in any other specimen of Hoplophoneus. original describer, and the definition that Cope's emended, indeed totally differ- has ever since been accepted as actually ent, definitive description of H. oreodontis basic for the species, even though it is not in 1885 was not based on the type but on tile earliest definition (which involved a another specimen, now Amer. Mus. No. thoroughgoing error), was based on this 5338, which includes the lower jaws and specimen and not on the type. If this is the facial part of the skull, with the per- accepted as neotype, then the species can manent dentition. The hypodigm of the be shown beyond much doubt to be syn- species as Cope left it (1885) and as sub- onymous with H. primaevus. sequent students have accepted it thus in It differs from representative specimens reality excludes the type and for Cope was of H. primaevus as follows: essentially this one referred specimen. All bone dimensions are absolutely smaller. Adams practically ignored both the type The canine base or alveolus dimensions are and absolutely smaller. Cope's hypodigm and used as his The diastemata are also relatively smaller. stated basis for comparison, i.e., as his The flange depth is smaller. hypodigm, Princeton No. 10515. rhe The ramus depth is smaller. same specimen was apparently Sinclair's The ratios of cheek tooth dimensions to almost hypodigm (1924). Jepsen (1933) gave any bone dimension are larger. measurements of this specimen and an- If Amer. Mus. No. 9764 is considered to other, Princeton No. 13628, and these two be the same as this specimen, whatever it seem to have been the principal if not the may be-an almost unescapable conclu- only members of the hypodigm of Scott sion-then the following can be added: The diameter of the postorbital constriction is and Jepsen (1936). relatively larger. To these specimens may be added an- The ratio of face to cranium length is smaller. other, Amer. No. 9764, which includes al- The index of the anterior zygomatic pedicle is most complete skull and jaws and is closely smaller. comparable with Cope's second specimen, This is an impressive array of characters Amer. Mus. No. 5338. Matthew identi- and withouit analysis might appear to vali- fied this specimen as H. oreodontis and ap- date the species beyond any question, but parently based his concept of the species it does not really do so. It is demonstrable mostly on it, although he did not say so in that most of these characters of proportion print. He figured the palate of this speci- are correlated with gross bone size among men (1910, Fig. 3A), but he labeled it the cats in general and, with little doubt, simply Hoplophoneus, without specific in Hoplophoneus in particular, regardless name. His measurement on a later page of species. Smaller and larger specimens (p. 313) was probably of this specimen. It referred to H. primaevus by authorities has the peculiarity that only two incisors who recognized H. oreodontis as distinct seem to be present, as discussed above. 1 Using "neotype" broadly for any substitute type If one goes back to Cope's type, then H. replacing either a lost type or, as in this case, a type that cannot be placed in any possible current hy- oreodontis is now unrecognizable and must podigm. 1941] THE SPECIES OF HOPLOPHONEUS 15 tend to show the same sort of differences, The canines are the only apparent argu- only less in the degree to which the size ment against this conclusion. It is possible difference is less. Differences equal both that they are really smaller than in most in kind and in degree are demonstrated in a specimens of H. primaevus, and in such a series of leopard skulls of a single local case this could and, on the available data, race. probably would be merely a variation. It is necessary, then, either to show that There is, however, inconclusive but prob- these proportions are different from what able evidence that in Hoplophoneus the they would be in a H. primaevus skull of canines continued to move downward and THIS SIZE, or that the size itself is signifi- hence that their alveoli continued to en- cantly different from H. primaevus. As large for some time after the permanent far as one can extrapolate from the avail- cheek teeth were in place. In that case, able specimens of H. primaevus, the pro- the small diameters of these alveoli are also portions are within the probable range for simply characters of youth. In either case that species at about this gross size. As no reliable specific difference from H. pri- for the size itself, Cope's 1885 hypodigm maevus is demonstrated. specimen or neotype and Amer. Mus. No. There are specimens in the collections 9764 are the youngest specimens available that as adults are distinctly below the size (except the useless H. oreodonlis type). usually assigned to H. primaevus and that Sutures closing or closed in all the other have been called H. oreodontis. These specimens are here completely open and include Princeton Nos. 13628 and 10515 the permanent teeth have just come into (Scott and Jepsen hypodigm of H. oreodon- place and are barely beginning to wear. tis) and our new specimen Amer. Mus. These are therefore small animals because No. 38980, which is one of the oldest but they are young. one of the smallest specimens in the series. What size they would have when fully Combining my measurements with Jep- grown can best be judged from the only sen's, the tooth dimensions homologous dimensions available on them that are not with those given below are for this group positively correlated with age, i.e., tooth of three specimens- dimensions. The table below affords a summary comparison. LC 11.7-14.3 Thus the cheek teeth are not only within P3_M1 29.6-30.5 LP3 9.7-10.2 the range of H. primaevus, sensu striclo, but LP4 17.3-18 are on the whole rather large for that LP4 ca. 9-11.4 "species" and are also within the range of LM, 15.7-ca. 16 (the probably synonymous) H. "robustus." On this basis there is every reason to be- These specimens thus are really small, lieve that these specimens far from repre- and not merely young. These are the speci- senting a separate, small species would, if menls that I have considered to be small they had lived to full maturity, have been females of H. primaevus. The present data rather large individuals of H. "primaevus" show that even if they arc assigned to a dis- or rather small individuals of H. "robus- tinct species, that species is not H. oreodon- tus." tis. Referred to H. robustvs Amer. Mus. Amer. Mus. Referred to H. primaevus by Matthew and by Jepsen No. 5338 No. 9764 by Matthew and by Jepsen (including type) LC 12 1/2 12 13.6-16.4 14-18.7 P3_M1 33.3 31.6-35.7 33.6-41.5 LP3 11 9.8-12 11-13 LP4 {r. 20.2 19.71 18.2-19.9 17.5-21.5 t 1. 19.8 19.7J fr. 11.7 12.21 11-12.8 11-13.5 LP4 l1. 11.9 12.01 fr. 17.7 17.91 15.8-20.0 LMi 11. 17.8 17.7J 16.1-18 16 AMERICAN MUSEUM NOVITA TES [No. 1123

H. MARSHI Thorpe pointed out that this species is served or safely inferable range of large based on a submature individual and he individuals of the H. primaevus-robustus compared it chiefly with H. oreodontis group: large incisive alveoli, long pointed (i.e., presumnably with Amer. Mus. No. palate, and large molar alveoli. The 5338, not the type), which is also sub- fourth, peculiar shape of canines, appears mature. The various skull proportions are to me to be wholly due to the fact that they near those of H. oreodontis if based on this are only partly protruded and would be specimen, in other words, near those of far larger and more recurved in a fully immature H. primaevus. Thorpe's type is adult animal. We do not have enough perhaps somewhat older than A.M. 5338 and a somewhat larger individual, i.e., a growth stages to judge the systematic male that would be referred to "H. robus- position of an immature animal with entire tus" when fully grown. Three of the confidence, but I do not see any clear-cut characters on which Thorpe mainly based character that would exclude this from the species appear to be within the ob- being a robust young male of H. primaevus.

H. LATIDENS Sinclair (1924, p. 113) has already re- into which Thorpe's specimen fits as a marked that he saw no reason for refer- normal member not peculiar in any im- ring H. latidens elsewhere than to H. portant way. The numerous small differ- primaevus. With this I entirely agree. ences in absolute dimensions and in pro- Thorpe was, of course, quite correct in portions noted by Thorpe are all well saying that his specimen is on the whole within the ranges now demonstrated for nearest to H. robustus, with some more H. primaevus. It is entirely possible primaevus-like characters and some be- that a form like this, being from a different tween the two. It was doubtless this ap- parently equivocal mingling of the char- locality (in Colorado) and possibly a some- acters of two supposed species that led what different horizon from typical H. Thorpe to create a third, but the two spe- primaevus, really does represent a different cies of reference intergrade and are (in my race, but the two known specimens do not opinion) synonymous, making a group show this to be the case.

H. MOLOSSUS This species was described by Thorpe age males of H. primaevus (or of H. "robus- on the basis of a single specimen from Ne- tus"). This at once suggests that the braska. Most of the dimensions are within animal is an immature male and some of its the observed range of H. primaevus (de- apparently distinctive indices are in fact fining that species to include H. "oreo- closely approached by those of Amer. Mus. dontis" and H. robustus) and all are within No. 9764 which is very probably an iinma- the theoretical range (i.e., the range that ture male of H. primaevus, but others are the available sample shows to be probable not and Thorpe's data plainly show that for the whole population). Nevertheless his type is not really immature. The the proportions are peculiar and are prob- facial : cranial index (B of the preceding ably not those normal for H. primaevus, discussion) has an observed range of 42- sensu lato. In general the bone dimen- 55 in H. primaevus and is 44 in No. 9764, sions of skull and jaws are small, about but is 56 in H. molossus. This is not surely those of small females of H. primaevus significant in itself, but it is striking in (or of H. "oreodontis") while the tooth view of the fact that small individuals dimensions are large, about those of aver- tend to have smaller indices while this, 1941] THE SPECIES OF HOPLOPHONEUS 17 one of the smallest known specimens of as was at first thought. Scott and Jepsen the genus, has the largest known index (1936, p. 135) add that the constriction is (except for H. occidenta7is, an animal not remarkably anterior in position. I cannot far from twice this size). judge the significance of this rather subtle A similar difference, not surely signifi- character from the data now available. cant in itself but perhaps becoming so Along with some of the other characters, when related to the size of the skull, is seen it is made questionable by Thorpe's state- in the index length of canine alveolus: ment that the only known specimen shows prosthion-basion (E), which is 11.7 in the some post-mortem deformation. type of H. molossus. This is above the ob- Thorpe's statement that H. molossus served range in H. primaevus and is strik- has heavy, massive skeletal bones also sug- ing because in the latter species this index gests a distinction since small specimens of may be larger in large than in small in- H. primaevus tend to have relatively slender dividluals. The postorbital constriction is some- bones. what below average diameter for H. pri- Given only one specimen and the ab- maevus, but does not differ noticeably from sence of absolute distinctions, it is possible those of the skulls of that species most com- that H. molossus is based on an abnormal parable in other dimensions, so that it is specimen of H. primaevus, but the evi- not really distinctive (nor is its index, D), dence suggests that it is distinct.

H. OCCIDENTALIS Only four specimens surely of this species purely morphological grounds. It is, how- have been recorded in the literature and I ever, unlikely on the following grounds: know of no others. The type is a lower jaw 1.-If H. "robustus" and H. "insolens" do not fragment, lacking both ends, with P4. belong to H. primaevus then that species (in- cluding the material referred to H. "oreodontis") Williston's type of Dinotomius atrox is a was much less variable and showed less sexual nearly complete skull and skeleton, and dimorphism than any comparable cats, machai- another closely similar specimen was found rodont or feline, as far as known. with this. Amer. Mus. No. 1407 is a com- 2.-As demonstrated above, H. primevut8 and plete mandibular ramus with the crowns "oreodontis," on one hand, and H. "robustus" and "insolens," on the other, nearly or quite inter- of P4 and M1, correctly referred to this grade in all their characters and show all the ex- species by Adams. These specimenis are pected and probable resemblances and distinc- much larger in every dimension than are tions normal for the two sexes of a single species. those of H. primaevus, with its probable It is highly improbable that the females of one species would so closely resemble and intergrade synonyms, and equally distinctive from with another species living in the same region at H. molossus, the third probably valid the same time or that the morphological rela- Brul6 species. The difference in gross size tionships of the females of one species to the is unquestionably significant and the addi- males of another would so exactly correspond to tion of the H. occidentalis specimens to the the normal relationships of males to females of a one species. H. primaevus series would indicate species 3.-It is very improbable that of two closely much more variable than is warrantedl by related species one would show great variation any valid analogy. The only possible and extreme (even though not abnormal) sexual question is whether, as Scott and Jepsen dimorphism, as would a species combining H. occidentalis with H. "robustu8" and "insolens," (1936) implied by placing them in the while the other showed abnormally little varia- synonymy of H. occidentalis, the specimens tion and no evident dimorphism, as would a called H. robustus and H. insolens might species including only H. primaevus and "oreo- not be females of this species rather than dontis." males of H. primaevus. From the point of 4.-The H. occidentali8 specimens may be all of one sex, or may include three females and one view of the sexual dimorphism and vari- male. This sex ratio in the sample and this ability implied, this is not impossible and degree of variation are well within limits of perhaps it cannot be quite disproved on probability if these are the only specimens 18 AMERICAN MUSEUM NOVITATES [No. 1123 known of the species. But if robustus and in- other teeth and of the lower jaw show no solens are the females of the species, then all four specimens of H. occidentali8 are males (and striking peculiarities. The small brain, highly variable as such) and a sample of about high emarginate sagittal crest, produced twenty includes only four males. This is not inion, and other peculiarities of Williston's impossible, but it is improbable. specimen are also about what would be ex- Most of the proportions of these large pected either in a large variant individual specimens are about as in H. primaevus. or in a large species. Those that are more or less distinctive are, H. occidentalis thus appears to be the as a rule, correlated with size and show in only species known in the lower BrulM of H. occidentalis about what would be ex- South Dakota that is surely distinct -fom pected because of its large size. In other H. primaevus. The occurrence together words, although the proportions are in of these two sharply different allied species some cases distinctive, give the species a is entirely normal. A definite ecological characteristic appearance, and are aids in distinction is suggested. H. primaevus recognizing it, they may be merely conse- is abundant, H. occidentalis rare, also a quences of its greater size and not separate normal situation when two allied cats genetic characters. Thus the facial-cranial occur together. The area of deposition index (B) is about 63 (calculated from Wil- was apparently mostly inhabited by the liston's apparently accurate figure), a very smaller species. The larger species may large figure well above the known range for have been everywhere less common, but H. primaevus, but in the latter this index may also have lived mostly in a somewhat is highly variable and apparently tends to different environment and be represented be larger for larger individuals. The index here only by strays. One possible analogy of the postorbital constriction (D) is rela- is afforded by the occurrence of leopards tively small, but within the H. primaevus and tigers in Asia. range and this index is generally smaller in It will be recalled that this fauna also larger animals. The canine index, E, is includes Dinictis, probably with two valid very large in Williston's specimen, 14.0, species, and Eusmilus, with one. These and probably does go beyond the limit of are adaptively quite distinct from Hoplo- normal variation or of probable regression phoneus and complete a picture of a nor- on size for H. primaevus, but the correlation mally varied cat assemblage with five or in the latter is not close enough to show six different sorts each occupying its own this beyond doubt. The proportions of the place in the general regional fauna.

AN INTERESTING NEW SPECIMEN Amer. Mus. No. 38804, found by Mr. of H. primaevus, used in the broadest sense Thomson in the lower Brule in 1938, in- to include "robustus" and "insolens," or so cludes skull, jaws, and most of the skeleton near it that the deviation has no clear of a Hoplophoneus smaller than H. occiden- significance. The only exception is that talis but well above the average for H. the upper canines appear to be abnormally primaevus. The prosthion-basion length, large, both relatively and absolutely. 181 mm., is slightly above that recorded Crushing and fissuring of the specimen with for any Brule specimen except Williston's subsequent filling of cracks with matrix "Dinotomius atrox" (= H. occidentalis), make this not exactly measurable, but the although the difference of only 2 mm. from anteroposterior diameter of the canine the largest specimens referred to H. alveoli was probably about 22 mm. and the "insolens" and "robustus" is hardly sig- index, E, about 12. The largest values in nificant. Although an unusually big the H. primaevus series are 17.5 mm. and, animal, with the development of crests, for the index, 11.1. The difference is not etc., that goes with size, almost all its di- clearly significant in view of the imperfect mensions and proportions are within the preservation of the specimen and it is, observed range for the largest individuals after all, most likely that this is merely an 1941] THE SPECIES OF HOPLOPHONEUS 19

extreme variant of H. primaevus, about the ways hard to evaluate on such imperfect most robust male known. The very frag- material, among them the probable pos- mentary skull of Amer. Mus. No. 655, a syntype but not the lectotype of H. in- session of upper canines absolutely and solens, also belongs to a very large animal relatively very large, as in this new speci- for this group and is peculiar in various

CHADRON SPECIES The described Chadron species are H. Chadron specimens will be required to show mentalis Sinclair and H. oharrai Jepsen. the probable average distinctions -corre- Their authors have described them fully sponding with changes of level within this and I have not re-examined any of the group. specimens so that I can add nothing except A new specimen extending the geographic the suggestion that the two may be syn- range is Amer. Mus. No. 27798, found by onymous as already hinted but not defi- A. C. Silberling in beds of Chadron age on nitely advanced by Wood (1927). The the west side of the Long Pine Hills in specimens rather closely resemble the most Carter Co., Montana. This consists of the robust males of H. primaevus, but are cranium posterior to the postorbital con- probably correctly distinguished from that striction, part of the upper jaw with all six species. It is just possible that the dubious incisors and the left canine, and a frag- specimens mentioned above, Amer. Mus. ment of the left lower jaw with P4-M1. Nos. 38804 and 655, represent a survival These remains do not permit exact identi- of this group, as a separate phylum, into the Brul6, but this is not probable. fication, but the large canine and the size Clark (1937) has reported the discovery of the cranium and production and shape of of a specimen referred to H. robustus its crests suggest H. oharrai. The lower jaw in the upper part of the Chadron, but has fragment is perhaps closest to H. pri- given no detailed description or measure- maevus, but the parts preserved show no ments. A considerably larger sample of good distinction between the two species.

TAXONOMIC CONCLUSIONS The data summarized in preceding 5.-H. robustus Adams, 1896. Probably pages includes most of the males of H. primaevus, of suggest the following conclusions which it is thus a probable synonym. Very regarding the eleven proposed White River unlikely to be a synonym of H. occidentalis. species of this genus: 6.-H. insolens Adams, 1896. One or two 1.-H. primaevus (Leidy, 1851). Surely specimens so identified may just possibly, but valid. As bitherto used, in a strict sense, the not probably, be distinctive. Other specimens, name has probably included most of the females including the lectotype, appear to be merely the and one or two small or young males of a natural largest males of H. primaevus. Unlikely to be species. synonymous with H. occidentalis. 2.-H. occidentalis (Leidy, 1869). Surely 7.-H. latidens Thorpe, 1920. Probably valid. A second, larger, ecologically different, based on slight variants of H. primaevus. They less abundant species. could represent a different race from most of 3.-H. oreodontis (Cope, 1873). Now inde- the material, but the available specimens do not terminate but possibly determinable at some establish this as probable. future time on the basis of the type. Cope's 8.-H. marshi Thorpe, 1920. Similarly a hypodigm of 1885 or neotype is probably an possible race, but not demonstrated to be such. immature H. primaevus and is above rather than At present not distinguishable from immature below the average size for that species. The H. primaevus. small mature specimens later referred to H. 9.-H. molossus Thorpe, 1920. Apparently oreodontis are thus different. They are probably valid, but data do not yet quite preclude the the smallest females of H. primaevus. possibility that the known specimen is a deviant 4.-H. atrox (Williston, 1895). A synonym or abnormal H. primaevus. of H. occidentalis. 10.-H. mentali Sinclair, 1921. Probably 20 AMERICAN MUSEUM NOVITATES [No. 1123 valid and almost surely so if H. oharrai is a any named BrulI species. Possibly synony- synonym. mous with H. mentalis but this is not demon- 11.-H. oharrai Jepsen, 1926. Distinct from strated.

REFERENCES ADAMS, G. I. 1854. "The ancient fauna of Nebraska." 1896a. "On the species of Hoplophoneus." Smithsonian Cont. Knowledge, VI, Amer. Nat., XXX, pp. 46-51. Art. VII, pp. 1-126. ("Machairo- 1896b. "The extinct Felidae of North Amer- dus" primaevus.) ica." Amer. Jour. Sci., (4) I, pp. 419- 1866. "Drepanodon (Machairodus) occiden- 444. talis." Proc. Acad. Nat. Sci. Phila- 1897. "On the extinct Felidae." Amer. delphia, 1866, p. 345. (Nomen Jour. Sci., (4) IV, pp. 145-149. nudum.) ALLEN, J. A. 1869. "The extinct mammalian fauna of 1924. "Carnivora collected by the American Dakota and Nebraska." Jour. Acad. Museum Congo Expedition." Bull. Nat. Sci. Philadelphia, (2) VII, pp. Amer. Mus. Nat. Hist., XLVII, pp. 1-472. ("Drepanodon" primaevus and 73-281. (Source for comparative occidentalis.) data.) MATTHEW, W. D. CLARK, J. 1910. "The phylogeny of the Felidae." 1937. "The stratigraphy and palaeontology Bull. Amer. Mus. Nat. Hist., XXVIII, of the Chadron Formation in the Big pp. 289-316. Badlands of South Dakota." Ann. MERRIAM, J. C., AND STOCK, C. Carnegie Mus., XXV, pp. 261-350. 1932. "The Felidae of Rancho la Brea." (H. "robustus" in the Chadron.) Carnegie Inst. Washington, Pub. No. COPE, E. D. 422. (Source for comparative data.) 1873. "Synopsis of new Vertebrata from the NELSON, E. W., AND GOLDMAN, E. A. Tertiary of Colorado, obtained during 1933. "Revision of the jaguars." Jour. the summer of 1873." Separate Mammalogy, XIV, pp. 221-240. printed as from 7th Ann. Rep't U. S. (Source for comparative data.) Geol. Surv. Terr., but not in that OSBORN, H. F., AND WORTMAN, J. L. volume. ("Machaerodus" oreodon- 1894. "Fossil mammals of the lower Miocene tis.) White River beds. Collection of 1874. "Report on the vertebrate palaeon- 1892." Bull. Amer. Mus. Nat. Hist., tology of Colorado." Ann. Rep't VI, pp. 199-228. (Brief notes on Hayden Geol. Geog. Surv. Terr. for Hoplophoneus.) 1873, pp. 427-533. (Hoplophoneus PHLEGER, F. B., JR. oreodontis; original definition of the 1940. "Relative growth and vertebrate genus.) paleontology." Amer. Jour. Sci., 1880. "On the extinct cats of America." CCXXXVIII, pp. 643-662. (Some Amer. Nat., XIV, pp. 833-858. data on heterogony in Smilodon and 1885. "The Vertebrata of the Tertiary For- Panthera.) mations of the West." Rep't Hayden Geol. Surv. Terr. for 1884, Vol. III. RIGGS, E. S. (Cope's definitive treatment of Hoplo- 1896. "Hoplophoneus occidentalis." Kansas phoneus.) Univ. Quart., V, pp. 37-52. JEPSEN, G. L. ScoTr, W. B., AND JEPSEN, G. L. 1926. "The oldest known cat, Hoplophoneus 1936. "The mammalian fauna of the White oharrai." Black Hills Engineer, S. D. River Oligocene-Part I. Insectivora Sch. Mines, XIV, No. 2, pp. 87-92. and Carnivora." Trans. Amer. Phil. 1933. "American eusmiloid sabre-tooth cats Soc., (N.S.) XXVIII, pp. 1-153. of the Oligocene epoch." Proc. SINCLAIR, W. J. Amer. Phil. Soc., LXXII, pp. 355- 1921. "A new Hoplophoneus from the Ti- 369. (Also many measurements of tanotherium beds." Proc. Amer. Phil. Hoplophoneus.) Soc., LX, pp. 96-98. LEIDY, J. 1924. "The faunas of the concretionary 1851. (No title.) Proc. Acad. Nat. Sci. zones of the Oreodon beds, White Philadelphia, V, pp. 329-330. ("Ma- River Oligocene." Proc. Amer. Phil. chairodus" primaevus.) Soc., LXIII, pp. 94-133. (Many 1852. "Description of the remains of ex- data on Hoplophoneus.) tinct Mammalia and Chelonia from THORPE, M. R. Nebraska Territory." Rep't D. D. 1920. "New species of Oligocene (White Owen Geol. Surv., pp. 534-572. River) Felidae." Amer. Jour. Sci., ("Machairodus" primaevus.) (4) L, pp. 207-224. 1941] THE SPECIES OF HOPLOPHONEUS 21

WILLISTON, S. W. WOOD, H. E., II 1895. "New or little known extinct verte- 1927. "Hoplophoneus mentalis, and cusp brates." Kansas Univ. Quart., III, homologies in cats." Jour. Mam- pp. 165-176. ("Dinotomius atrox.") malogy, VIII, pp. 296-302.