ih CR oe ec a en ee er ee on, et ea a A tes lg ree eee “ene . - ee ese eed, si ie beet ee en mt i phim 7S es a ne _ a Bee canara ueraearartie . ge that anaes thet » 3 i ¥ 1 Bho net Se gle arina ie cP aR ni a the Co Tn Renton gl Di Se A (6a Dig Dir bn lem Vin tae Dm Few Sg Op PM = 7 = _ “ Ue Hem ttle Ge te PK fom iB ee iy a Fo a i ete rae tie fin ae Neh. Ln pom Sa TONG Me Gr Ali pe te Hote iene “s y - ae em ee a pain mF Se Sti Ga Pe 244 Cae Um Liane Same fhe Phe Shas ge th ew seh Et | gd pe inline cw Gertie Bi Mee tm Fite i Fe a os p naar ioneeae a on ac te Dhan te oe Bhat EIR Fam es ef Fre Tm ina he ene hm or Mean 9 ag tin Os eesti enya Hote Ana iS ae 2 J “ A A . a im Meat Rela ee Ve Ye Ea iw oe Tan? wit Ti ar the Pir Bag te Pv Bie mE Ra Ri tas ie oy, aN I Ee nt abs Hoe Rite Mane titer hai Pat Rani! ” NAEP ARTA | Ow RI Nae Pr aE SR i TE Rasen titt : “ mht ~ hes ig DnB Nit RTA th Then te tin and Eiger em the yo Rh fn ag He ills He an Be finan canteens Tarte Bo Netin in Ads aOR Shirt Be fea te pedal Or gaat nine ciate fede Se Fy hain ih . a ssl ee ces a Re a att ae ee ca ce a De eae ae ace i ea Vibe Be om t= PNB hae Bo ti Fo an A> i le ie ae Poe er he Vm tg Fee Mi And Rhy ones dh ey de Pome he i he ha toh hag Palo TP tat ch pms om inl Ne Be athe PS Ae he ate mst lin athe iS = Rha Tomi hae on eRe ek eet © 9 it~ ay rN a ig Hr Fn" War a > fs a he a nh SS Yaw er ety Hi i: te Ry My em: a i oe a et age FO Ph the fe Ginn eRe ir hhc i actin thay Hy, . - Bn oe tp di Ry aS Min ete in _ pe i Fes Rep Pan Ai 0 Rison Foi Dain ie he Ree Dp nn Bm eo Figs ag th hh tn Megs Mle Ha Does Hm hn Wan i Bae Fa Vm ihe Ra Opeth Sym a at hg Se Ren i th pte hae NE ile he BE: Ra Bh ot cae lite liewtn FiO : ~ ' . s e = Ponte Ra ie ethan aS | fen ae bel Phage ne arth Ree i ea oa Bani in mRNA hm ll =A AT MEER pe Ask AE A a MART TI Nig AP ere Rut il Fem pili, SE layer Kinch et Ben a tions Mow ane Kaya Che hee Ae Hm Pop Mes agro Seren nie Tn he tS aloe ge Be nl = Fa Rn tn Mp Re ei anh Sn an i inns Dea dingsh Mpa ey 2 Ce i Bo Tg on a fe He STs Gin Wim Be lvl gir the a . mes , se The .. Sie iil Sate’ hee te Bs ne 2 Salinity =o ¥ ar 2 s ck hy Bing Bech te he ; ~ me ne ene > 3 : ne “ : - AT tig ae Mince ake hath 3 en btn Anshan MiP Be “ Ses, 4s Ta Abin iene NAAM Sct Giri mie Rasch atone ‘ “ ‘ + ie ew i eee ee eee Ce ean ne = tien, Site enti Set ht Ty i ie eo Res Hee FCAPS ie a tg tn ee a patlNS ~< a as Se ee es 5% - Re Saale Nae Fyn Shad een ttl Acct > see Sa ie ei ee Rein eae atm C= tee Bee Ce f . ” etdlaeatecty — _ Papert nc e rE an = AF te Re RR en Ae rete sehr PR Tag — eset % at ” ~ : 7 ett a“ oe a ae | -% eee Ba - oe ther eas - ~ ve a " i “i ‘. . ~ * edie Sis = 7 " A “ it Poe Pe aie Sie ee : — ; La Tntetn atthe macho tp veintincs yaar en . ™ pene " aera ihe . By eee iether - > ‘ f o a z < x i es Ne “ ‘ a 2 ie b ae ws eS ee cre es = neh = phe oo eh — > c= DPS he alone mm Fag RN Tale Ragone Dw oni a om the tT ibn, “raitiom CoP % SR ee lh Beis Se ee ote ee fr roi Nie 2 ~ si ~—a “a : a eee ee ee « we onal Teta aie het ai itp an 8 Bie Bom a Shrine hte ee ent athe ei esipo— inl 2: eng = Hae Be ae - oa, i Ae Fob ' - ‘Meg pt ftp ne en ae sh Sie oe oe orale thn Proag ee ee High ba tah Seg She iiehre ¥ Get 3. itp tat ire Fo Fle Re he 4 ed —s =. ate { f ANNALS: OF THE SOUTH AFRICAN MUSEUM VOLUME XXXI vv ee ee 2 ee rey =e TE —— | ee es eer ers ANNALS OF THE SOUTH AFRICAN MUSEUM VOLUME XXXII PRINTED FOR THE TRUSTEES OF THE SOUTH AFRICAN MUSEUM AND THE GEOLOGICAL SURVEY OF SOUTH AFRICA BY NEILL AND CO. LTD., 212 CAUSEWAYSIDE, EDINBURGH. 1934-1950. ao]. be TRUSTEES OF THE SOUTH AFRICAN MUSEUM. Professor R. S. Apamson, F.R.S.S.Afr. The Rev. S. J. Fort. Professor D. L. ScuHoutz, D.Sc. C. J. SIBBETT, J.P. S. H. Sxaire, Ph.D., F.R.S.S.Afr., J.P. SCIENTIFIC STAFF OF THE SOUTH AFRICAN MUSEUM. KepreL Harcourt BaRNnarD, M.A., D.Sc., F.L.S., Director; in Charge of Fish and Marine Invertebrates. ALBERT JOHN Hesse, B.Sc., Ph.D., F.R.E.S., Assistant in Charge of the Ento- mological Department. Miss G. Joyce Lewis, B.A., Assistant in Charge of the Botanical Department. LizUWE Dirk Boonstra, D.Sc., Assistant in Palaeontology. A. J. H. Goopwin, M.A., Honorary Keeper of the Ethnological and Archaeological Collections. Miss K. Marcaret Suaw, B.A., Assistant in Charge of the Ethnological Collections. LIST OF CONTRIBUTORS. R. S. ADAMSON. Fossil Plants from Fort Grey, near East London L. D. BoonstTRa. Pareiasaurian Studies. The Cranial Date eer The Dermal Armour : The Vertebral Column and Ribs A Contribution to the Morphology of the Gorgonopsia : Additions to our Knowledge of the South African Gorgonopsia preserved in the British Museum : A Contribution to the ee ee of the Mammal-like Reptiles of the Suborder Therocephalia_ . L. CAYEUX. The Phosphatic Nodules of the Agulhas Bank H. B. 8. Cooke. A Critical Revision of the peli aay baeisaact of Southern Africa : ; : : S. H. Haveuton. On some Karroo Fishes from Central Africa See also J. V. L. Rennie J. V. L. RENNIE. On Placocystella, a New Genus of Cystids from the Lower Devonian of South Africa ' : Lower Cretaceous Lamellibranchia from Northern Zululand (with an Account of the Geology of the Cretaceous Beds and a Preliminary Analysis of the Associated Ammonite Fauna by S. H. Haughton) . LIST OF NEW GENERA AND SUBGENERA PROPOSED IN THIS VOLUME. Aelurosauroides (Gorgonopsia, Reptilia) Boonstra Arctognathoides (Gorgonopsia, Reptilia) Boonstra Ischnolepis (Pisces) Haughton . Megacucullaea subg. (Arcidae, Mollusca) Ronnie Placocystella (Cystoidea) Rennie ; Sphenotrigonia subg. (Trigoniidae, M oluxea) Rennie : 393 97 269 277 190 143 97 305 269 365 PLATES XVIII XXX. XXXI. XXXII XXXII. XXXIV. XXXV. XXXVI. XXXVII —LY. LIST OF PLATES. Pareiasaurian skulls. Pareiasaur teeth. Bradysaurus baini. Bradysaurus seeleyt. Bradysaurus seeleyt and vanderbyli. Bradysaurus vanderbyli. Bradysuchus whaitsi. Nochelesaurus alexandert. Nochelesaurus strubent. Nochelesaurus strubent and Dolichopareia angusta. Dolichopareia angusta. Brachyparewa rogersi. Brachypareia watsont. Embrithosaurus schwarz. XIX. \ Pareiasuchus peringueyt. XX. XXI. XXII. XXIII. XXIV. XXYV. XXVI. XXVII. XXVIII. XXIX. Pareiasuchus nasicornis. Pareiasuchus nasicornis and Anthodon serrarius. Anthodon serrarius and Pareiasurus serridens. Dermal Armour of Pareiasaurians. Bradysaurus seeleyt. Fossil Karroo Fishes. Map of Agulhas Bank Area. Phosphatic Nodules. Devonian Cystids. Lower Cretaceous Lamellibranchia. DATE OF ISSUE OF THE PARTS. Part 1, pages 1-136, April 1934. Part 2, pages 137-268, July 1934. Part 3, pages 269-392, August 1936. Part 4, pages 393-480, December 1950. Vil ; : , ate ty = ts mae ag; ry ; i ; cn colon i cae ee nyt ty s : F 6 \ ' h ae = " a Ak marci bind i, aN ea g ‘ , ; n 4 ‘sf t ie ‘ + i, ef en | ye ~ 7 - rm ; " a ' - 7 ¥ . i 7; x s r i h i 7 { ; \ " fo » Vee ‘ , ‘ t ( ; - - j 7 a 1 " i § A * D4 PN Mig . in i | i 4 ® “gerald ae,’ ‘3 4 i) jan -_ ." ba , tee by't fy dno yr wae x = i ‘ ae | r2 ae Ra ae . ‘ | | } . | ; > | | Bh ody . | \ | 2 f | | ‘ ] Z i by = i? ke ks} | } ) iN = j . - ) } ey i . ; . ~~ 1} | I! co | ’ ' i i te = ‘ : i it e t es FA ik ms INDEX OF GENERA AND SUBGENERA. PAGE A Aelurognathus . é : 2 143 Aelurosauroides ’ : ye Le) Aelurosaurus . ; : 2 ile’) Alectryonia : : ; -al6 Alopecodon : : B +. 216 Anthodon t 35, 42, 65 Arctognathoides : : . 148 Arctognathus 142, 203 Arctops : : 5 se dls) Asinus . : : : 451 Atherstonia : 3 ; lo? Aucellina . : : ‘ «ol Avellana . : ‘ : . 388 B Brachypareia 31, 41, 64 Bradysaurus 24, 39, 63 Bradysuchus . , ‘ Sea C Cardium ...-> *. : : i 382 Ceratotherium . : ; . 404 Cerdodon . 4 ; : : ez Cerdognathus . ; : . 211 Cucullaea . ; ; 3 . 304 Curtisia . : : : 2388 Cynariognathus ‘ : . 216 Cynariops ‘ : i = Lo Cyniscodon : ; : ee NILES) Cynodraco 158, 210 Cyphoxis : : ; . 3804 D Diceros : : ; : . 401 Dicroloma ‘ , . 388 Dolichopareia 1x EB Elginia . : Embrithosaurus Equus Euchambersia Kurygnathohippus Exogyra Gervillia Glycymeris Goniomya Gorgonops Hipposaurus Hyorhynchus Icticephalus Ictidosuchoides Idonearca Inoceramus lotrigonia Ischnolepis Isognomon Koalemasaurus . Kolpohippus Kraterohippus Leptotrachelus . 29, 64 | Lopha PAGE BF red, 29, 42, 63 431 261 423 316 310 310 378 18] 154, 158 231 186 316 x Index of Genera and Subgenera. PAGE Lycaenodon . : : . 192 | Ptychomya Lycaenodontoides . : . 141 | Pygopterus Lycaenops : , : . Ld4 Lycideops : : ; . 235 Lycosaurus - . : . 202 Rhinoceros M Rutitrigonia Megacucullaea . é : . 305 Megatrigonia . ; : - idol Metaschizotherium . : ss g ed Moschorhinus . : : - 238 | Scabrotrigonia . Scutosaurus N Scylacops i Scylacosaurus Neithea . : ; . » anad .: m 9137 a 3. med? * “sf 9168 re «5. mde Durban Museum, Pa » low. Bradysaurus vanderbylt (Haughton and Boonstra). (Pls. I-V.) There are an intermediate number of teeth (15-16), which do not overlap, have 7 cusps, and are moderately swollen. The posterior border of the “cheek” is very smooth, and so is the general surface of the skull. The maxilla and lacrymal are slightly bulging. The dorsal and lateral surfaces are not sharply separated as the ridge is low and rounded. The postparietal is small and is longitudinally rectangular. The anterior border of the interpterygoid vacuity is concave. The tabular boss is prominent. The following skulls are assigned to this species :— Type, South African Museum, No. 3718, Tapinocephalus zone, low. 6 . 8941 _ 35. pene i " 9169 5 sof, lands Pareiasaurian Studies.—Part IX. 7a || Genus BrapysucHus Haughton and Boonstra. Bradysuchus whartst (Broom). (Pls. I-II.) As I have not had the opportunity of seeing the type specimen in the American Museum of Natural History, this diagnosis had to be based on photographs kindly supplied by that institution and also on information received from Dr. R. Broom. The skull of the single specimen of this form is large; the teeth are badly preserved, but there were probably about 16 teeth with few (less than 9) cusps, probably arranged as in the genus Bradysaurus; the interorbital width is appreciably less than the “tabular width.” The posterior border of the “‘cheek”’ carries well-defined bosses. In general the skull is ornamented with pits and rugae. The maxilla and lacrymal are only slightly bulging. Both behind and in front of the orbit the dorsal and lateral surfaces are separated by a strong ridge. Medially the pterygoids are fused far posteriorly to form a long palate. The quadrate is inclined forwards. The brain-case and the supraoccipital pillar appear to be high. The tabular boss is prominent. Except for Broom’s statement that the lower jaw carries two bosses on the angular, and the fact that the snout appears to be somewhat pointed, this genus is very similar indeed to the forms included in the genus Bradysaurus, particularly to Br. seeleyt. Type, Amer. Mus. Nat. Hist., No. 5567, Tapinocephalus zone, low. Genus NocHELESAURUS Haughton and Boonstra. This genus includes forms with fairly large skulls; forms with an intermediate (15-17) number of teeth—these teeth may have a slight or a considerable overlap; all forms have teeth with few cusps (less than 9) arranged as follows: 3 anterior, 3 median, and 2 posterior; they are slightly swollen. The interorbital width is approximately equal to the “tabular width.” The posterior border of the “cheek”’ is smooth, or carries very strong bosses. The snout is rather pointed and the “cheek” deep; the latter is vertical or somewhat diverging. The skull is in general very smooth or strongly ornamented, but with little “crocodilian” pitting. The maxilla and lacrymal are moderately or slightly bulging. The dorsal and lateral surfaces are not separated by a very strong ridge. The postparietal is either transversely or longitudinally rectangular. The pterygoids are fused 28 Annals of the South African Museum. far posteriorly so as to form a long palate. The border of the inter- pterygoid vacuity is concave. There is hardly any vault between the medial rows of pterygoid teeth. The quadrate is much (55°), or slightly (85°), inclined forwards and does not stretch far medially. The brain-case is fairly low; so is the supraoccipital pillar. The tabular boss is low. The maxilla is fairly deep and the quadratojugal moderately deep. The quadratojugal just meets the maxilla or is separated from it by the jugal. Nochelesaurus strubeni (Broom). (Pls. I-V.) The teeth are slightly overlapping; the whole skull is smooth; there is a single nasal boss; the posterior border of the “cheek” has hardly any development of bosses. The snout is higher and more pointed than in N. alexanderi, and the “cheeks” are more vertical. The maxilla and lacrymal are slightly bulging. The postparietal is longitudinally rectangular. The quadrate is slightly inclined forwards (85°). The quadratojugal just meets the maxilla. From his descrip- tion and figures it appears that von Huene is right in assigning the lower jaw in the Tiibingen collection to this species. The following skulls are assigned to this species:— Type (lower jaw): Amer. Mus. Nat. Hist., Tapinocephalus zone, low. South African Museum, No. 1207, Tapinocephalus zone, low. e 4352 5 9 Ws 2: - 5019 3 aie B 2 5590 53 es Geol. Pal. Inst. Tibingen, a 3 ee Nochelesaurus alexanderi Haughton and Boonstra. (Pls. I-V.) The teeth are considerably overlapping; the whole skull is heavily ornamented with bosses and rugae; there are two nasal bosses; there is a great development of bosses on the posterior border of the “cheek” approaching the condition shown in Pareiasuchus perin- gueyt. The snout is lower and broader than in WN. strubeni, and the “cheek” is more flattened out. The maxilla and lacrymal are moderately bulging. The postparietal is transversely rectangular. Pareiasaurian Studies.—Part IX. 29 - ‘The quadrate is much inclined forwards (55°). The jugal is inter- calated between the quadratojugal and the maxilla. The following skulls are assigned to this species:— Type, South African Museum, No. 6239, Tapinocephalus zone, mid. 6239A s r 39 3) 8944 99 3) 99 Genus EMBRITHOSAURUS Watson. Embrithosaurus schwarz Watson. (Pls. I-V.) In this genus the skull is quite large; there are an intermediate number of teeth (15-16), which are considerably overlapping; there are an intermediate number of cusps (9) arranged as follows: 3 anterior, 3 median, and 3 posterior; they are slightly swollen. The interorbital width is approximately equal to the “tabular width.” The posterior border of the “cheek” carries pronounced bosses. The snout is rather pointed and the “cheek” deep. In general the skull is ornamented with pits and rugae. The maxilla and lacrymal are only very slightly bulging. The postparietal is transversely rectangular. Medially the pterygoids fuse far posteriorly to form a long palate; the border of the interpterygoid vacuity is convex. The quadrate has a moderate forward inclination (75°), and it does not stretch far medially. The brain-case and supraoccipital pillar are fairly low. The tabular boss is prominent. The maxilla is fairly deep, but the quadratojugal is shallow and small. The jugal is intercalated between the quadratojugal and the maxilla. The mandible and teeth with 9 cusps described and figured by Broom must be assigned to this species if we are correct in maintaining that the nature of the cusping is a feature of taxonomic importance. The following skulls are assigned to this species :— Type, South African Museum, No. 8034, Tapinocephalus zone, mid. Brit. Mus. Nat. Hist., 3 5 dae, (Watson’s Hottentots River Specimen.) Genus DoLticHoPaREIA Haughton and Boonstra. Dolichopareia angusta Haughton and Boonstra. (Pls. I-V.) In this genus the skull is long and fairly narrow; there are many teeth (19-20), which are slightly overlapping; there are few cusps 30 Annals of the South African Museum. (less than 9) arranged as follows: 3 anterior, 3 median, and 2 pos- terior; they are slightly swollen and decrease much in size in antero- posterior direction. The interorbital width is approximately equal to the “tabular width.’ The posterior border of the ‘“‘cheek”’ carries strong bosses. The snout is very pointed and the “cheek” deep. In general the skull is moderately ornamented with pits and slight rugae. The maxilla and lacrymal are only slightly bulging. Pos- terior to the orbit a strong ridge separates the dorsal from the ventral surface of the skull. The postparietal is irregularly shaped and its posterior border is notched. Medially the pterygoids are fused fairly far posteriorly so as to form a moderately long palate. The border of the interpterygoid vacuity is convex. The quadrate is much inclined (60°) and is not directed medially but anteriorly (see Pl. III, fig. 14). The brain-case and supraoccipital pillar are fairly high, the supraoccipital is not excluded from the foramen magnum. The tabular boss is fairly prominent. The maxilla is fairly deep and so is the quadratojugal. The quadratojugal just fails to meet the maxilla. The following skulls are assigned to this species:— Type, South African Museum, No. 6238, Tapinocephalus zone, mid. 3717 9 ee: 33 39 Genus KoALEMASAURUS Haughton and Boonstra. Koalemasaurus acutirostris (Broom). (Pl. III and text-fig. 5.) Through the kindness of Mr. J. Hewitt, of the Albany Museum, I have been able to study Broom’s type, which consists of a little more than the half of the mandible with a very imperfect skull (parts of the cheek, upper jaw, palate, and basis cranu alone are preserved in a bad condition). As is the case in so many of Broom’s types, this type is really very bad, and it is with reluctance that an attempt is made to place it in this system of classification. In our preliminary classification we accepted Broom’s identification of the specimen in the American Museum of Natural History and based our diagnosis on both specimens. Through the courtesy of the American Museum I have photographs of the specimen in that institution and am able to confirm Broom’s identification. Although both skulls are not very well preserved the position of the orbit and the nature of the angular boss justifies such an identification. The skull in this genus is of moderate size; there are an intermediate number (15-16) of Pareivasaurian Studies.—Part IX. Bil teeth with a fair amount of overlap; the teeth have few cusps (less than 9) arranged as follows: 2 anterior, 3 median, and 1 or 2 pos- terior. The interorbital width is greater than the “tabular width.” The posterior border of the “cheek” carries low but distinct bosses. The snout is high and pointed (but in the American Museum specimen it appears to be truncated). In general the surface of the skull is moderately ornamented. The maxilla and lacrymal are moderately bulging. The dorsal and lateral surfaces are only distinctly demar- cated by a ridge behind the orbit. The postparietal is transversely rectangular. The pterygoids are fused far anteriorly so as to form a short palate; the border of the interpterygoid vacuity is convex. The quadrate is inclined forwards (75°) and is situated far forward in the skull. The brain-case and supraoccipital pillar are fairly high. The tabular boss is fairly prominent. The maxilla is fairly shallow Fic. 5.—External view of lower jaw of Koalemasaurus acutirostris. x 4. Type in Albany Museum. Ang. =angular. Dent. =dentary. Art. =articular. Sp. =splenial. Cor. =coronoid. Sur Ang. =surangular. and the quadratojugal deep. The orbit is in the anterior half of the skull. As the angular boss is such a characteristic feature a figure of the lower jaw is included (fig. 5). The following skulls are assigned to this species :— Type, Albany Museum, Tapinocephalus zone, high. American Museum, No. 5568, ks a 2 South African Museum, No. 4345, Br op Litiealay Genus BRacHYPAREIA Haughton and Boonstra. Both species of this genus have medium-sized skulls; an inter- mediate (15-16) number of teeth, which are slightly overlapping; there are few cusps (less than 9) arranged as follows: 3 anterior, 3 median, and 2 posterior—they are slightly swollen. The inter- orbital width is appreciably greater than the “tabular width.” The snout is either rounded or rather pointed; the “cheek” is broad and 32 Annals of the South African Museum. low, and its posterior border is rather smooth or has fairly marked bosses. The orbit does not lie in the anterior half of the skull. In general the skull is fairly smooth, with a slight indication of pitting and rugae. The maxilla and lacrymal are slightly bulging, more so in rogerst than in watsont. Medially the pterygoids fuse far anteriorly, thus producing a short palate; in one species the overhang of the maxillaries and premaxillaries increase this appearance of shortness. The quadrate is not inclined much forwards (80°) and it does not stretch far medially. The brain-case and supraoccipital pillar are low. The tabular boss is low. The maxilla and quad- ratojugal are fairly deep. The jugal is intercalated between the quadratojugal maxilla. Brachypareva rogers: (Broom). (Pls. I-III, V.) The snout is rather rounded and depressed; the posterior border of the “cheek” is rather smooth. The maxillaries and premaxillaries overhang the anterior part of the palate, so that in ventral view the “comma-shaped” vacuities are partly obscured. The following skulls are referred to this species :— Neotype, South African Museum, No. 5012, Tapinocephalus zone, low. 9 +B) 4350 e) 9 mid. 9 oy) 9095 3 9 low. ? 9 9098 ” ” mid. Brachypareia watsont Haughton and Boonstra. (Pls. I-II, IV—V.) The snout is rather pointed and high; the surface of the “cheek” is rugose and carries fairly prominent bosses on its posterior border. There is hardly any overhang of the maxillaries and premaxillaries. Type, South African Museum, No. 6240, Tapinocephalus zone, low. Genus Propappus Seeley. Propappus omocratus (Seeley). Propappus parvus (Haughton). (PIV) In Propappus omocratus only a fragment of maxilla and premaxilla is known. The lower jaw has two angular bosses; there are an Pareiasaurian Studies.—Part IX. ao intermediate number of teeth (14), which do not overlap; there are many cusps (10-11) arranged evenly in a semicircle. Judging from the rest of the skeleton, the skulls must be of medium size in omocratus and rather small in parvus. Type, South African Museum, No. 1058, Endothiodon zone. Genus ParErasucHus Broom and Haughton. In this genus the skull is of medium size; there are few teeth (14), which do not overlap; there are many cusps (more than 9) arranged evenly around the edge of the crown. The interorbital width is approximately equal to the “tabular width”; the snout is very broad, depressed, and rounded; the “cheek” is deep and carries strong or very strong bosses on its posterior border. In general the skull is rugose and pitted, more so in peringueyi than in nasicornis. The maxilla and lacrymal are bulging. The postparietal is small and transversely rectangular. The dorsal and lateral skull surfaces are not separated by aridge. Medially the pterygoids fuse far forward so as to produce a short palate. The quadrate is inclined much forwards (55°-65°) and does not stretch far medially. The brain-case and supraoccipital pillar are low; the opening of the Vth nerve is low. The supraoccipital forms the dorsal border of the foramen magnum. The tabular boss is low. The maxilla is shallow, but the quadrato- jugal is large and deep; it does not reach the maxilla. Pareiasuchus peringueyt Broom and Haughton. (Pls. I-V.) There are very strong bosses on the posterior border of the “cheek,”’ whose surface is also rugose. The quadrate is very much inclined forwards (55°). The angle of the “cheek” is situated far forward. The nostrils are visible in dorsal view. The following skulls are assigned to this species :— Type, South African Museum, No. 2337, Cistecephalus zone, low. - ¥ 2367 ‘ x s 10668 a LOM 29 29 mid. Pareiasuchus nasicorms Haughton and Boonstra. (Pls. I-III.) There are strong bosses on the posterior border of the “cheek,” whose surface is fairly rugose. The quadrate is inclined forwards VO. .xXxt, PART i. a 34 Annals of the South African Museum. (65°). The angle of the “cheek” is situated far back. Owing to the forward direction of the nasal bosses the nostrils are not visible in dorsal view. Type, South African Museum, No. 3016, Endothiodon zone. Genus PAREIASAURUS Owen. Pareiasaurus serridens Owen. (Pls. I-V.) The skull is medium sized; there are few teeth (14), which overlap considerably; the number of cusps varies—9 on the anterior and up to 11 on the posterior teeth; these cusps are regularly arranged around the crown and are of equal size. The interorbital width is less than the “tabular width.” The “cheek” is very deep and carries strong bosses. The snout is high and pointed. In general the skull is fairly rugose and is considerably pitted. The maxilla and lacrymal are only slightly bulging. The dorsal and lateral skull surfaces are not separated by a ridge but pass into each other in rounded curves. The tabular and postparietal have flange-like developments overhanging the occiput. Medially the pterygoids are fused far anteriorly so as to form a short palate. The border of the interpterygoid vacuity is concave, and the parasphenoidal rostrum is partly visible through it. The quadrate is nearly vertical, the forward inclination being 85°; it also stretches far medially so that in anterior view it nearly obscures the posterior pterygoid ramus altogether. The brain-case and the supraoccipital pillar are high, the fenestra ovalis is small, and the opening for the Vth nerve is situated high up on the side wall. The supraoccipital forms the dorsal border of the foramen magnum. The maxilla and quadrato- jugal are deep, and the jugal is intercalated between the maxilla and quadratojugal. The following skulls are assigned to this species :— Type (cast), Brit. Mus. Nat. Hist., No. R4063, Cistecephalus zone, mid. Neotype, University of Stellenbosch, Cistecephalus zone, high. As only a part of the lower jaw and some teeth are known of the specimen, which Seeley named Parezasaurus russouwi (Brit. Mus., No. R1996), it is not possible to include it in our scheme with certainty. The nature of the teeth is, however, so similar to that of Parevasaurus serridens that there appears to be little doubt as to the generic Pareiasaurian Studies.—Part IX. 35 name, but it appears advisable to retain Seeley’s specific name— thus Pareiasaurus russouwi. Recently I collected another mandible (S.A.M., No. 10667) fairly high up in the Cistecephalus zone which I refer to the genus Pareiasaurus. Genus ANTHODON, Owen. Anthodon serrarvus, Owen. (Pls. I-III, V.) The skull is small; there are few teeth (11-14), which do not overlap; there are many cusps (8-15) of equal size arranged regularly around the crown. The interorbital width is less than the “tabular width.” The posterior border of the “cheek” has small and low bosses. The snout is fairly high and rather pointed; the “cheek”’ is quite deep. In general the skull is moderately ornamented with pits and rugae. The maxilla and lacrymal are only very slightly bulging, but there is a small prominent tubercle on the maxilla. The dorsal and lateral surfaces are not sharply separated by a ridge. Medially only the anterior half of the pterygoids fuse, so that pos- teriorly a long interpterygoid slit remains open. The quadrate is moderately inclined forwards (75°), but it extends far medially to meet the lateral pterygoid flange so that in anterior view the posterior pterygoid ramus is altogether covered by the quadrate. The brain- case and supraoccipital pillar are high. The tabular boss is low. The maxilla and quadratojugal are deep. The jugal separates the maxilla from the quadratojugal. The pineal foramen is very large. The following skulls are assigned to this species :— Type, Brit. Mus. Nat. Hist., No. 47337, Cistecephalus zone, mid. Amer. Mus. Nat. Hist., No. 7001 as 54) Taal South African Museum, No. 4020 ms ss Lowe 45 a 7841 i a. low: 10074 a lowe Owen’s type of serrarius is an imperfect skull from which Watson has, however, been able to determine a number of features. I have described a lower jaw of a specimen in the South African Museum (S.A.M., No. 4020) as of this species. This specimen has also a partial skull of which I have been able to determine the whole palatal structure. At the end of 1930 Broom published two photographs and a short description of a specimen in the American Museum, which he maintains is certainly a distinct species—gregoryi. In April 1931 36 Annals of the South African Museum. I found an impartial skull and some elements of the postcranial skeleton of an Anthodon (S.A.M., No. 10074) at Dunedin. The pelvis and part of the hind-limb have been described without a specific determination, as I then doubted the validity of Broom’s new species. Measurements of these four specimens show a uniform progression from a small form (Owen’s), through two slightly larger forms (Broom’s and 8.A.M., No. 4020), to another slightly larger form (S.A.M., No. 10074). But they do not differ in any essentials of struc- ture, and in Broom’s description of his new species he gives no reasons for believing his specimen to differ from Owen’s. He simply states that it certainly is a distinct species. In view of these considerations it is best to refer them all to Owen’s Anthodon serrarius, and refer the slight variation in size to a variation within the species or, maybe, an age variation. A complete mandibular ramus in the Geological Museum of the University of Stellenbosch is also assigned to this species. Anthodon minusculus, from Tanganyika, is only known from the distal ends of two humeri. Genus ScurosauRuS Hartmann- Weinberg. Scutosaurus karpinskw Amalitzky. (Pls. I, III-IV.) The skull is very large; there are a variable number of teeth with many cusps (9-17) arranged regularly around the crown. The interorbital width is much less than the “tabular width.” The posterior border of the “cheek” carries well-defined bosses. The snout is fairly broad and deep, and the “cheek” is deep. In general the skull is sculptured with pits and rugae. The maxilla and lacrymal are bulging. The dorsal and lateral surfaces are not abruptly demar- cated by a ridge but pass into each other in rounded curves. Medially the pterygoids are only fused along their anterior half, so that pos- teriorly a long interpterygoid slit remains through which the para- sphenoidal rostrum is visible. According to Hartmann-Weinberg the quadrate is nearly vertical, but according to Sushkin inclines at 45°, and it does not stretch far in medial direction. The brain-case and supraoccipital pillar are high. The tabular boss is low. Type, mounted specimen, Geol. Mus. Acad. Sci., Leningrad. In Amalitzky’s memoir “diagnoses” are given of three further species, but as the figures are so poor, the text so meagre, and other Russian authors do not mention them, it would be foolhardy to Pareiasaurian Studies.—Part IX. BT attempt to establish their systematic position. It would, however, appear that one is justified in referring them all to the genus Scuto- saurus, for they manifestly cannot be included in the South African genus Pareiasaurus. Genus Exainta Newton. —Elginia mirabilis Newton. (Pls. J-II.) The skull is small; there are few teeth (12), which do not overlap: there are many cusps (9-10) of equal size arranged regularly around the crown. The interorbital width is less than the “tabular width.”’ The posterior border carries horns. The snout is high and fairly pointed, and the “cheek” is deep. In general the skull is greatly ornamented, carrying a great development of horns, in some respects of an analogous nature to that in the living Moloch and Phrynosoma. The maxilla and lacrymal carry protruding horns. The quadrate appears to be much anteriorly inclined (45°). The maxilla is deep. A long and strong horn is developed on the tabular, and represents the tabular boss. Type, Coll. Geol. Survey, Elgin Sandstone. LITERATURE CITED. AMALITZKY, V.—‘‘ Diagnoses of the New Forms of Vertebrates and Plants from the Upper Permian on North Dvina,” Bull. Acad. Sci. Russ., 1922. Boonstra, L. D.—‘‘Pareiasaurian Studies. Part III.—On the Pareiasaurian Manus,” Ann. 8. Afr. Mus., vol. xxviii, pt. i, 1929. Boonstra, L. D.—‘‘Pareiasaurian Studies. Part IV.—On the Pareiasaurian Pes,” Ann. S. Afr. Mus., vol. xxviii, pt. i, 1929. Boonstra, L. D.—‘‘A Contribution to the Cranial Osteology of Pareiasaurus serridens Owen,” Ann. Univ. Stellenbosch, vol. viii, Section A, No. 5, 1930. Boonstra, L. D.—‘‘Pareiasaurian Studies. Part VII.—On the Hind Limb of Two Little-known Pareiasaurian Genera: Anthodon and Pareiasaurus,” Ann. S. Afr. Mus., vol. xxviii, pt. iv, 1932. Boonstra, L. D.—‘‘Pareiasaurian Studies. Part VIII.—The Osteology and Myology of the Locomotor Apparatus.—B. Fore Limb,” Ann. 8S. Afr. Mus., vol. xxviii, pt. iv, 1932. Boonstra, L. D.—‘‘A Note on the Hyoid Apparatus of Two Permian Reptiles (Pareiasaurians),’ Anat. Anz., Bd. 74, 1932. Boonstra, L. D.—“ A Study in Evolution: The Phylogenesis of the Parevasauridae,”’ S. Afr. Journ. Sci., vol. xxix, 1932. Boonstra, L. D.—‘‘Paleobiologiese Beskouinge oor ’n Uitgestorwe Reptielgroep (Pareiasauridae),” S. Afr. Journ. Sci., vol. xxix, 1932. 38 Annals of the South African Museum. Broom, R.—‘On an Almost Perfect Skeleton of Pareiasaurus serridens Owen,” Ann. 8S. Afr. Mus., vol. iv, pt. ii, 1903. Broom, R.—‘‘ A Comparison of the Permian Reptiles of North America with those of South Africa,” Bull. Amer. Mus. Nat. Hist., vol. xxviii, pp. 197-234, 1910. Broom, R.—“On Four New Fossil Reptiles from the Beaufort Series, South Africa,’ Rec. Alb. Mus., vol. ii, 1913. Broom, R.—‘“‘ Further Observations on the South African Fossil Reptiles,’ Amer. Mus. Journ., vol. xiv, 1914. Broom, R.—‘‘Catalogue of Types and Figured Specimens of Fossil Vertebrates in the American Museum of Natural History. JJ.—Permian, Triassic, and Jurassic Reptiles of South Africa,’ Bull. Amer. Mus. Nat. Hist., vol. xxv, 1915. Broom, R.—“ Pareiasaurian Nomenclature,” Ann. Mag. Nat. Hist., ser. viii, vol. xvii, 1916. Broom, R.—‘On Some Points in the Structure of the Pareiasaurian Skull,” P.Z.S., vol. i, 1924. Broom, R.—‘‘ On a New Species of Anthodon (A. gregoryi),” Amer. Mus. Novitates, No. 448, 1930. Broom, R., and Haucuton, 8. H.—‘‘On the Skeleton of a New Pareiasaurian (Pareiasuchus peringueyt),’’ Ann. 8. Afr. Mus., vol. xii, pt. i, 1913. HarRTMANN-WEINBERG, A.—“‘Ziir Systematik der Nord Diina Pareiasauridae,”’ Palaeontologische Zeitschrift, Berlin, Bd. 12, 1930. Havucuton, S. H.—‘‘Pareiasaurian Studies. Part I].—Notes on Some Pareia- saurian Brain-cases,’’ Ann. 8. Afr. Mus., vol. xxviii, pt. i, 1929. Havucuton, S. H., and Boonstra, L. D.—‘‘ Pareiasaurian Studies. Part I.—An Attempt at a Classification of the Pareiasauria based on Skull Features,” Ann. 8. Afr. Mus., vol. xxviii, pt. i, 1929. HavueutTon, S. H., and Boonstra, L. D.—‘‘ Pareiasaurian Studies. Part V.—On the Pareiasaurian Mandible,” Ann. S. Afr. Mus., vol. xxviii, pt. ii, 1930. Hauvcuton, 8. H., and Boonstra, L. D.—*‘ Pareiasaurian Studies. Part VI.—The Osteology and Myology of the Locomotor Apparatus.—A. Hind Limb,” Ann. 8. Afr. Mus., vol. xxviii, pt. ili, 1930. Newton, E. T.—‘On Some New Reptiles from the Elgin Sandstones,” Phil. Trans. Roy. Soc. (Lond.), B, vol. clxxxiv, pp. 473-489, 1893. OweEN, R.—‘‘Catalogue of the Fossil Reptilia of South Africa, etc.,’’ Publ. British Museum, London, 1876. SEELEY, H. G.—“‘ Researches on the Structure, Organization, and Classification of the Fossil Reptilia. I1.—On Pareiasaurus bombidens (Owen), etc.,” Phil. Trans Roy. Soc. (Lond.), B, vol. clxxix, 1888. SEELEY, H. G.—‘“‘ Researches on the Structure, Organization, and Classification of the Fossil Reptilia. VII.—Further Observations on Pareiasaurus,” Phil. Trans. Roy. Soc. (Lond.), B, vol. clxxxiii, 1892. SusHxin, P. P.—‘On the Modifications of the Mandibular and Hyoid Arches and their Relations to the Brain-case in the Early Tetrapoda,” Pal. Zeitschrift, Bd. viii, Heft 4, 1927. VERSLUYS, J.—‘ Das Streptostylie-Problem und die Bewegungen im Schadel bei Sauropsiden,” Zool. Jahrb. Suppl., xv, pp. 545-716, 1912. Watson, D. M. S.—*‘ On the Skull of a Pareiasaurian Reptile, and on the Relation- ship of that Type,” P.Z.S., vol. i, 1914. Watson, D. M. S.—‘‘On the Nomenclature of the South African Pareiasaurians,” Ann. Mag. Nat. Hist., ser. viii, vol. xiv, 1914. Fig. EXPLANATION OF PLATES. Puate I.—Pareiasaur Skulls in Lateral View. x i. 1. Bradysaurus baini. This is drawn from the type in the British Museum, but was checked and augmented by reference to South African Museum specimens (Nos. 3533, 4347, 4999, 5002, 5127, 9104, 9105), and the specimen in the Tiibingen Museum. . 2. Bradysaurus seeleyi. Drawn from the type and co-type in the British Museum, but checked and elaborated by reference to 8.A.M., Nos. 5624, 9115, 9121, 9137, and 9168. . 3. Bradysaurus vanderbyli. Drawn from the type in the South African Museum (No. 3718) and two other specimens (Nos. 8941 and 9169). . 4. Bradysuchus whaitsi. Drawn from Broom’s published photograph of the type in the American Museum (No. 5567). . 5. Nochelesaurus alexandert. Drawn from the type in the South African Museum (No. 6239). . 6. Nochelesaurus strubenit. Drawn from a South African Museum specimen (No. 5019). . 7. Dolichopareia angusta. Drawn from the type in the South African Museum (No. 6238), but checked by reference to No. 3717. . 8. Brachypareia rogersi. Drawn from the neo-type in the South African Museum (No. 5012) and checked by No. 5340. g. 9. Brachypareia watsoni. Drawn from the type in the South African Museum (No. 6240). . 10. Embrithosaurus schwarzi. Drawn from the type in the South African Museum (No. 8034) and incorporating some details elucidated by Watson in the Hottentots River specimen. g. 11. Anthodon serrarius. Drawn from the South African Museum specimens (Nos. 4020 and 10074) and incorporating the details elucidated by Watson in the type in the British Museum. . 12. Pareiasuchus peringueyi. Drawn from the type in the South African Museum (No. 2337), with additional details from No. 2367. g. 13. Pareiasuchus nasicornis. Drawn from the type in the South African Museum (No. 3016). . 14. Pareiasaurus serridens. Drawn from the neo-type in the Geological Museum of the University of Stellenbosch. . 15. Scutosaurus karpinskii. Drawn from the published photograph by Hartmann-Weinberg. . 16. ‘ Pareiasaurus”’ horridus. Drawn from a cast in the British Museum. _17. Elginia mirabilis. Drawn from the published illustration of Newton. Newton’s determination of the sutures are doubtful. ir —troneale P. Orb. =postorbital. Jug. =jugal. Pr. Fr. =prefrontal. Lac. =lacrymal. Pr. Mx. =premaxilla. Mx. =maxilla. Qu. Jug. =quadratojugal. Na. =nasal. Sq. =squamosal. Pa. =parietal. Tab. =tabular. P. Fr. = postfrontal. i | Plato 1. Ann. 8. Afr. Mus., Vol. XXXT. Bia, 7. Fig. 13. Fro. 16, Neill & C0, Ly Fo. 3. eR Cs tie ANIA Raye! Payty Fy : a a er ee ee ee For details of specimens examined see explanation to Plate I. . 1. Bradysaurus baint. . 2. Bradysaurus seeleyi. . 3. Bradysaurus vanderbyli. . 4. Bradysuchus whaitsi. and a photograph by the American Museum. . Nochelesaurus alexanderi. . Nochelesaurus strubent. . Brachypareia rogersi. 5 6 . 7. Dolichopareia angusta. 8 9 . Brachypareia watsoni. . 10. Embrithosaurus schwarzi. . Ll. Anthodon serrarius. . 12. Pareiasaurus serridens. . 13. Pareiasuchus peringueyt. . 14. Pareiasuchus nasicornis. eg. 15. “‘Pareiasuchus”’ horridus. g. 16. Elginia mirabilis. Puate I1.—Pareiasaur Skulls in Dorsal View. x Drawn from a sketch kindly supplied by Dr. Broom Newton’s determination of the sutures are probably incorrect. Fr. =frontal. P. Orb. =postorbital. Jug. =jugal. Pr. Fr. =prefrontal. Lac. =lacrymal. Pr. Mx. =premaxilla. Mx. =maxilla. P. Pa. =postparietal. Na. =nasal. Qu. Jug. =quadratojugal. Pa. =parietal. Sq. =squamosal. P. Fr. = postfrontal. Tab. =tabular. Fia. ll. eo yeah \ SecccZ Tae is | SIN \ wy crate NTI Tak MO Lor A AONE, ~ Ann. 8. Afr. Mus., Vol/ XN XT. Plato If. HGH i, Fig, 11. Fro. 16. Neill & 00, ld, oon oa»ark WH | =) ee Hm CW bo PuateE III.—Pareiasaur Skulls in Ventral View. x }. For details of specimens examined see explanation to Plate I. . Bradysaurus baini. . Bradysaurus seeleyi. . Bradysaurus vanderbyli. . Nochelesaurus strubent. . Brachypareia rogersi. . Embrithosaurus schwarzi. . Anthodon serrarius. . Pareiasuchus peringueyi. . Pareiasuchus nasicornis. . Pareiasaurus serridens. . Scutosaurus karpinskii. Details published by Sushkin are incorporated. . Koalemasaurus acutirostris. Drawn from the type in the Albany Museum. . Dolichopareia angusta. . Dolichopareia angusta. Dorsal view of the palate of the type (S.A.M., No. 6238). A. Pter. =anterior ramus of the pterygoid. B. Oe: = basioccipital. B. Sph. = basisphenoid. For. =foramina in the premaxillary processes. For. Qu. =foramen quadrati. Int. Pter. Shit. =interpterygoid slit. Jug. =jugal. lis ter. =lateral ramus of the pterygoid. Mix: = maxilla. Pal. = palatine. Pe Oe: = paroccipital. P. Sph. = parasphenoidal rostrum. Pear: = postparietal. Pre He: = prefrontal flange. Prix: = premaxilla. Pre V0: = prevomer. Ree Ot: = pro-otic. Pe Pter. = posterior ramus of the pterygoid. Qu. =quadrate. Qu. Jug. = quadratojugal. Sub. Orb. Vac. =suborbital vacuity. SOc =supraoccipital. Sq. =squamosal. Tab. =tabular. Trans. =transversum. CASt: =tubercle for attachment of stapes. 2 50 c0Soc0~- ot Ann, 8. Afr. Mus., Vol. XXXT. 3 ‘Plate U1. Boe ce [sooo Si igecee =a Neill & Co, Mids Puate IV.—Pareiasaur Skulls in Occipital View. x i. For details of specimens examined see explanation to Plate I. Fig. 1. Bradysaurus baini. Fig. 2. Bradysaurus seeleyi. Fig. 3. Bradysaurus vanderbyli. Fig. 4. Nochelesaurus alexanderi. Fig. 5. Nochelesaurus strubent. Fig. 6. Dolichopareia angusta. Fig. 7. Brachypareia waisoni. Fig. 8. Embrithosaurus schwarzi. Fig. 9. Pareiasuchus peringueyi (No. 2337). Fig. 10. — Be (No. 2367). After Haughton. Fig. 11. Scutosaurus karpinskii. After Hartmann-Weinberg and Sushkin. g. 12. Pareiasaurus serridens. From Annals of the University of Stellenbosch. B. Oc. —=basioccipital. B. Sph. =basisphenoid. Ex. Oc. =exoccipital. For. Mag. =foramen magnum. For. Qu. =foramen quadrati. P. Oc. =paroccipital. ED: = postparietal. P. Pter. =posterior ramus of the pterygoid. P.T.F. |=post-temporal fenestra. Qu. = quadrate. Qu. Jug. =quadratojugal. Sc: =supraoccipital. Sq. =squamosal. Tab. =tabular. . S. Afr. Mus., Plate IV.‘ Neill & Co., Lid, > Ann, 8, Afr. Mus., Vol. XXXT. Plato IV. * Neill & Co., Ltd, Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. fo) Fig. Fig. Fig. Fig. oe owtn or WN — = © i eo Hq GW bo C — oO . Bradysaurus baini. 5th left tooth. . Bradysaurus seeleyi. 1st left tooth. . Bradysaurus vanderbyli. 3rd right tooth. . Nochelesaurus alexanderi. 6th left tooth. . Nochelesaurus strubeni. 5th left (mandibular) tooth (after Broom). . Dolichopareia angusta. 8th left tooth. . Brachypareia rogersi. 8th right tooth. . Brachypareia watsoni. 14th right tooth. . Embrithosaurus schwarz. 9th right tooth. . Anthodon serrarius. 6th or 7th left tooth (after Owen). . Propappus omocratus. 3rd or 4th right tooth. . Pareiasuchus peringueyi. 5th right tooth. . Pareiasuchus nasicornis. 4th right tooth. . Pareiasaurus serridens. 6th right tooth. . “ Pareiasaurus” russouwt. 3rd right mandibular tooth (from Haughton). Puate V.—Pareiasaurian Teeth in External View. x 3. Ann. S. Afr. Mus., Vol. XX XI, Plate V. r ha ney Sve cf [fy \ Au ti \, al Fic. 6. nie! 72 Fic. 8. Fig. 9. Fic. 10. | lh | te y \ i : i 5 S 1 \ Nh Wiad ql ul Wy Fic. 11. Pie. 12: iG. Fic. 14. 1 acepee ays Neill d& Co., Lid. VOL. XXXI, PART 1. 4 Pirate VI. a Bradysaurus baini . —. +~S.A.M., No. 50 ; (a) Lateral view. ‘ oe | ee (b) Dorsal view. | . 2, aie m1 i eee eee eee Ann. S. Afr. Mus., Vol. XX XI. Plate VI. NEU dP COn, iLial. nd 7 a 5 oe ; 4%) a ‘el 7 _ {2 eel a ie “-2 of ; ; : Se s —_ Pirate VII. S.A.M., No. 9137. . (a) Lateral view. Bradysaurus seeleyi (6) Dorsal view. 4 § ’ ; q Ann. S. Afr. Mus., Vol. XX XI. Plate VII. Neill & Co., Ltd ya a ne Sa To Puate VIII. ' (a) Bradysaurus seeleyi eee : S.A.M., No. 9137. (b) Bradysaurus vanderbyli S.A.M., No. 3718. = Ann. S. Afr. Mus., Vol. XX XI. Elate. Vil Neill & Co., Lid. < [ 4 : é = ¥ - ie ; ee =) > ci wat pi . ‘4 ; ioe leo a he . i 7 2 f a 2 ‘ : 28 ab . * Ps Poe el ee a Puate IX. view. Bradysaurus vanderbyli (a) Lateral (6) Dorsal view. Ann. S. Afr. Mus., Vol. XX XI. Plate IX. Neill & Co., Ltd. Ue ey ne A ia Photograph merican view. Bradysuchus whaitsi . Lateral t a —_ a iS ee i iy ies | iene eee ag aha ms y- = ; - ' i F Ve aces 8S EF IS Soe ga Se ee ee ee aa a mit = arc Sts - —— Se ai += —————_———————————————— eee ——— ——_____—. — SS Ee ee ee Plate X. Ann. S. Afr. Mus., Vol. XX XI. va Puate XI. \ i No. 6239 S-AM ere Nochelesaurus alexand (a) Lateral view. (b) Dorsal view. Plate XI. Ann. S. Afr. Mus., Vol. XX XI. Neill & Co., Ltd. Rn < mY 1 24 , oe i ad . a = - n Nochelesaurus strubeni ‘ ; (a) Lateral view. (6) Dorsal view. Prats XI. S.A.M., No. 5¢ Ann. S. Afr. Mus., Vol. XX XI. Plate XII. Neill & Co., Lid. sae ee a > OA Se eee Puate XIII. (a) Nochelesaurus strubeni . S.A.M., No. 5019. (b) Dolichopareia angusta . S.A.M., No. 6238. ¢ of Ann. S. Afr. Mus., Vol. XXXI, Plate XIII. S Neill & Co., Ltd. Rep iis 5 A all i Cia q ae ay 4 LO: 5 , t a j - % tm ; fd ad P yt ee , y Ag tg ates i 4 7 . e ~ a ‘ ¥ ia er ’ 3 / a , “ va a j Ss | — a \ Ad 4 LATE XIV. | Dolichopareia angusta . + S.A.M., No. (a) Dorsal view. - (6) Ventral view. Plate XIV. Ann. S. Afr. Mus., Vol. XX XI. Neill & Co., Ltd, y dj ij m . pel a j ah F itary aed ; > t = ni 4 7 nn) oe | } , we y YM ~* e ; A i BVA: gaya Weg al ae ee be i i i a ‘ed 7 1 ~! 7 i : * aby ( At) ate { 9 - j eee BT , Mi ee? ¥ A id on) Ue i v7 ee NYO i Napa a a ; ie "4: iy Tee hs 4 r ‘ f| ; oui ; A 7 : , s . = An y ‘- ee |: | ~ hy : Se « ies ; Prat XY. Brachypareia rogersi Pee eo (a) Lateral view. . SAM, No, 5012.amm (b) Lateral view. . §S.A.M., No. 4350. s * : | | . | \ j i! si — | 2 : | . | | : 1 a ; ; : pe A { i ¥ . 1H = lb . . i) iy 7 hat Ve aan et : re (nae ‘ane l v a a ie 2 a PLATE XVI. Brachypareia watsoni . . S.A.M., No. 6240. (a) Lateral view. (b) Dorsal view. - , \ x | hs | i} < li | i = ¥ = = => = ' ~ 5 s x ¥ . yy , » } : y ' 7 fi 7 ad . ad * ; iy ws = 9 Ann. S. Afr. Mus., Vol. XX XI. Plate XOV I. Neill & Co., Lid. sie isees M A aAll re v vay | > f f rn ) y 4 wi * Wor ale i) : 5 A A r aan i a) 7 eri - 7 : . 4 5 : if : M i . iy 1 ‘ . i - . ‘ = ' 5 . ‘ ‘ ' : > i — ¥ a oa a . a : y ti | i ‘ in } . + { “ ¢ 1 \ . : ’ f ; oa) Mi is K te ms oN arte ete Gog: = iz © aa ; at (5 0 ee 7 Thue Aa OS ee sr CN ar. biol i 2; mae v : 7? ; ba ws ie =e) *, * j : 1 oa Ys , we j PLATE XVII. Embrithosaurus ge pnne . “S.AM, No: (a) Lateral view. ig) Sse (6) Occipital view. . x }.- 1 ‘ r . 1 se ™rk “s . ’ ‘ : i : A ioe?) oor ite Ann. S. Afr. Mus., Vol. XX XI. Plate XVII. Neill & Co., Ltd, a rh is . 4 } i Pi . ¥ . ; iy j a Das Are ‘ : ra hie i\ fr t x * i" i A i i f ; ae en r , f 4 : in ie a ct a) ? 2 y ; \ . ‘ - , F yi ‘ i 3 \ ‘ 1 r pi = | | { ‘ . ti i ‘ ” . r XXXI, PART 1. 7 oy . % r¢ - : 4 ; Puate XVIII. re Pareiasuchus peringueyi . S.A.M., No. 2337. « ea , (a) Lateral view. : - ’ (6) Occipital view. = i ei eee a ee ee ee ES Ann. S. Afr. Mus., Vol. XX XI. Plate XVIII. Neill & Co., Lid. nile Aas r 7 a rt iar PS es ‘ . ‘ 4 ’ F F bs 7 ' rs P ‘ y Zs - ti . ‘ r | Z 4 - ° ne | | PLATE XIX. Pareiasuchus peringueyi . S.A.M., No. 2337. (a) Dorsal view. _ (b) Ventral view. . a ae oe —= Ais hit ate i Ann. S. Afr. Mus., Vol. XX XI. Plate XIX. eill & Co., Ltd. SSS Poke "s Boy Lee ae ey me Me Sa 7 Hy te oF - ia A ‘eux , 4 be i] - \ ac? K ne a 7 7 i ae : » . v : ah : : oe - PS Rae ae i o “bu nt \ 7 \ ie Puate XX. ' Pareiasuchus nasicornis . S8S.A.M., No. 3016. (a) Ventral view. (6) Dorsal view. Ann. S. Afr. Mus., Vol. XX XI. Plate XX. Neill & Co., Ltd. = : = A i Ph a ¢ : 7 7 - - : " ! | ~ Z 7 ro 1 E eae ie : : , > é A = 7 i | a - ; | | { =~ 7 o ; ; 7 - ; | 3 4 ; 7 | | , & i = ; ‘ ‘> a a * Pas oe ee: “ee. a “ rar 2 . er : ms am | 7 a - ey rs 4 oe cy . in 7 7 Y; ci : : cal | : | 5 ‘ ae . Va * - . = = eee ee SF ae 4 ‘ . & «ees y i : a eee eee ee ee ee ‘a a . t © \ ’ ‘ a 4 - ’ . ’ . P a 1 a] ‘ xe by 1 4 Ly 5 . 5 a} " : m x n fi ‘ H : ee A Ky a pv YS aa oF] eee ae Trea aoe eo Cee ae tee i ‘Be A Oe i ieee iar aM | } u 7 ies ¢ ; : i" oa ae lh iG oe =~ i c ln Ne f ¢ ll lve wae E ek =e 9 E ae aan a ie E ies. see agg a 4 aay via y | 7 i é ’ [: / | | | | | Pe ~ 4 | | | : I 2 | | | Prarin XOX, (a) Pareiasuchus nasicornis §.A.M., No. 3016. Lateral view. | (b) Anthodon serrarius . §.A.M., No. 10074. Lateral view. ‘ Ann. S. Afr. Mus., Vol. XX XI. Plate X XI. Neill & Co., Lid. PuLaTE XXII. (a) Anthodon serrarius . Amer. Mus., No. 7001. Dorsal view. x 4. Photograph kindly supplied by the American Museum. (b) Pareiasaurus serridens . Geol. Mus. Univ. Stellenbosch. Lateral view. x #. Plate XXII. Amn. S. Afr. Mus., Vol. XXXI. i S Neill & Co., Ltd. (30m) Pareiasaurian Studies. Part X.—The Dermal Armour.—By Likuwr D. Boonstra, D.Sc., Department of Palaeontology, South African Museum. (With Plates XXITI-XX VII.) INTRODUCTION. THat the Pareiasaurs possessed a more or less extensive armour of bony scutes of dermal origin has been noted by a number of authors. Seeley gave the first and also the most complete account of the armature in the genus Bradysaurus. Broom described it in Pro- pappus and Brachypareia, and referred to it in Embrithosaurus and Anthodon; publishing jointly with Haughton, a brief reference was made to the scutes of Parevasuchus. Subsequently Watson used the nature of the armature as a generic character. Amalitzky and Hartmann-Weinberg have considered the nature of the scutes as of specific as well as of generic value in the case of the Russian Parelasaurs. In the body of this paper an attempt is made to gather together our knowledge of the Pareiasaurian armour in order to obtain a general idea of the appearance and the defensive abilities of the Pareiasaurs as a whole, and, secondly, to determine the specific and generic differences (if any), and, as a further development of the same theme, to see whether there is any progressive or retrogressive development of the armature. The large number of new specimens now in the South African Museum, together with those already partly known, fully justifies such an attempt being made. Wherever possible the verbal description is accompanied by photographs. Genus BRADYSAURUS. 1888. Seeley, H. G., Phil. Trans. Roy. Soc., vol. clxxix, B 22, p. 91. 1892. Seeley, H. G., Phil. Trans. Roy. Soc., vol. clxxxii, B 76, pp. 345, 346. 40 Annals of the South African Museum. 1908. Seeley, H. G., P.Z.8., vol. 1, pp. 605-607. 1914. Watson, D. M.8., Ann. Mag. Nat. Hist., vol. xiv. The dermal ossicles in the three species of this genus are very similar (Pl. XXIII; Pl. XXIV, fig. 1); the scutes preserved are roughly circular or oval in outline; the external surface is relatively smooth; in Br. vanderbyli they are practically devoid of any ornamentation; in Br. baini the scutes are convex, curving down from a thickened centre which in some scutes appears like an incipient boss; there are no distinct radiating ridges; in Br. seeleyr the central boss appears to be somewhat more distinct, and the surface somewhat pitted. There is also very little difference in the size of the scutes, viz. Br. baini (50 x 50-45 x12 mm.), Br. seeleys (57-35 x 50-30 x 10-7 mm.), Br. vanderbyli (48 x 47 x 7 mm..). The above facts appear to exhaust the dissimilarities. From the available facts the following composite account, which includes some of Seeley’s observations, of the arrangement of the armour appears to be correct for all three species. “No indications of large dermal bones are preserved. Three rows of scutes appear to have extended down the median line of the back.” “The median row was placed on the summits of the neural spines and above the interspaces between them. In contact with these, laterally, there is a pair of scutes extending transversely outward; so that in the antero-posterior direction three scutes may be counted corresponding to each vertebra. And the effect is a median longitudinal strip of close-set scutes, flanked by two lateral rows.” From the large number of specimens in the South African Museum it appears that the scutes were in more close contact above the scapulae and the sacrum than along the rest of the back and on the tail. Above the ribs, 7.e. on the flanks of the animal, the scutes seem to have been little developed, and this also applies to the upper portions of the limbs. No small stud-shaped or conical ossicles have been found in this genus, and they probably did not exist. In this connection it is perhaps significant to note that in the other large genera (Dolichopareia and Nochelesaurus) of the Tapinocephalus zone no scutes have been found. It must be admitted that of these genera we only know a few specimens, not very fully preserved, but the absence of scutes in the specimens known seems to indicate a lesser development of armour than in the more highly evolved forms of the younger horizons. Pareiasaurian Studies.—Part X. 4] Genus KOALEMASAURUS. No dermal scutes are preserved in association with the type skull in the Albany Museum. Associated with a skull and vertebral column referred by me to this genus there are, fortunately, a few scutes. Although it is impossible to ascertain the extent of the armature from this scanty material, the nature of an individual scute can be described. The scutes preserved (Pl. XXIV, fig. 2) are small (45 x32x7 mm.), oval in outline, smooth, with only a slight indication of a central knob and a hardly perceptible rugous radiation. The possibility exists that the scantiness of the number of scutes preserved is due to the fact that the specimen is immature, and to the same cause the smoothness of the scutes may be ascribed. Genus BRACHYPAREIA. ie broom. kh. Anns, Air. Mus, vol. vii, p. 323, 1914. Watson, D. M.8., Ann. Mag. Nat. Hist., vol. xiv. Unfortunately no scutes are known of Brachypareia watsoni—only the skull of the type specimen being preserved. The description that follows thus only refers to Brachypareia rogersi. A very large number of dermal scutes of various sizes and shapes are preserved in associa- tion with the type specimen (PI. X XIV, fig. 3). There is a perfect gradation from fairly large ovoid or circular ossicles, whose outer surfaces are ornamented with a central knob from which fairly strong ridges radiate, through a series of smaller circular ossicles with a strong central knob, but with only feeble ridges radiating from it, standing on a base and presenting the appearance of a depressed collar stud, to small globular ossicles about the size of large peas, and with apparently no surface ornamentation. In all the ossicles, with the exception of the globular ones, there is on the external surface a foramen entering the interior of the bone and cut into the side of the knob. It communicates with a small foramen opening on the internal surface. From the plate a better idea can be obtained of the relative sizes and shapes than from any verbal description. The number of ossicles preserved, in addition to this gradation in a series in point of size and ornamentation, allows one to draw the following conclusions, viz. that all along the back, from the neck to the proximal dorsal surface of the tail, there must have been rows of ossicles (two or three per rump segment) with contiguous edges. The flanks, withers, and quarters were apparently studded with the 42 Annals of the South African Museum. circular stud-shaped ossicles; the upper portions of the limbs, the extremity of the tail, and the lateral parts of the belly must have housed the smaller globular ossicles; the belly, throat, and those parts affected by the movements of the limbs would have been practically free of dermal armour. Genus EMBRITHOSAURUS. 1903. Broom, R., Ann. 8. Afr. Mus., vol. iv, pl. xvi. 1914. Watson, D. M.S., Ann. Mag. Nat. Hist., vol. xiv. Although Broom only mentioned the dermal ossicles of Embritho- saurus in his explanation to a plate, a series, above the 6th, 7th, and 8th presacral vertebrae, is very well preserved in situ and merits a fuller description. Above the spinous process a large (50 x 50 mm.) and thick (20 mm.) scute is situated (Pl. XXYV, fig. 1); it has a very large and strong central boss of irregular shape, but no ridges radiate from it towards the periphery. Both anteriorly and posteriorly this scute articulates with a smaller smoother scute, which overlies the interspace between the spinous processes; laterally there lie two rows of scutes, each row has two scutes per segment—one lying on the postzygapophysial ridge and the other in the interspace; these scutes articulate with each other and with the median row of scutes. These lateral scutes are smaller (50 x 40 mm.) and thinner (15 mm.), have a more circular boss from which a few ridges commence, but do not reach the periphery; theedgesareserrate. Astillsmaller scute (35 x 22 x5 mm.), apparently from the flank, is preserved. It is oval in outline, has a slightly thickened centre, rugose near the edges, and serrated edge. In summary one can thus state that above the withers of the animal there were five rows of articulating ossicles, that it is probable that some of these rows extended along the whole length of the back, and that the flanks were in all probability studded with smaller and smoother scutes. Genus ANTHODON. 1930. Broom, R., Amer. Mus. Novitates, No. 448. Although Broom has described the specimen of Anthodon in the American Museum of Natural History as a new species, I have seen fit to doubt that it is distinct from Owen’s species. The nature of the dermal armour in a specimen in the South African Museum, which Pareiasaurian Studies.—Part X. 43 certainly belongs to Owen’s species, is so similar to that of the American Museum specimen that my doubt as to the validity of Broom’s species appears to have further justification. Broom has stated that the back of Anthodon was “covered by a bony carapace formed by large articulating bony scutes.”’ From a number of dermal ossicles found overlying the sacral region of one of the South African Museum specimens, it becomes evident (Pl. XXV, fig. 2; Pl. XXVI, fig. 1) that at least some of the scutes are firmly articulated to each other, so that in ventral view (Pl. X XVI, fig. 1) the line of articulation is only indicated by a digitating suture similar to those found between some cranial bones. A small foramen is also to be noted opening on to the ventral surface. The scutes are of medium size (64-54 x 45-40 x 11 mm.), roughly circular in outline, and are strongly ornamented on the outer surfaces, the central knob stands out very distinctly—this being due to the fact that it does not grade into the base, but is partly demarcated from it by an encircling moat traversed by radiating ridges. Relative to this moat the peripheral edge is somewhat raised and thickened; at the articulation with adjoining scutes the edge is deeply serrate. Genus PAREIASAURUS. 1914. Watson, D. M.8., Ann. Mag. Nat. Hist., vol. xiv. Watson has shown that the pelvis described by Owen as that of Dicynodon tigriceps belongs to the type of Pareiasaurus serridens. With it are associated scutes which would appear to have formed an extensive armour. “The whole of the dorsal surface of the pelvis between the crests of the ilia is covered with very large scales which are of the Propappus type, but differ in their more definitely pitted ornament and much larger size.” Associated with the neo-type skull in the University of Stellenbosch there are a number of scutes preserved, and I also have a specimen in the South African Museum with a number of scutes (Pl. X XVI, fig. 2). Three sizes (67-53-47 x 58-45-40 x 25-20-15 mm.) of ossicles are preserved. Each ossicle has a strong central boss from which strong ridges alternating with deep grooves radiate towards the periphery. The foramen, entering the interior of the bone on one side of the boss, is constant. A small foramen also opens on the internal surface of the scute. The scutes of Pareiasaurus are very similar to those of Parevasuchus. There is evidence (Pl. X XVI, left bottom pair) that some of the 44 Annals of the South African Museum. lateral scutes overlying the ribs have their serrated edges interlocking, and if this is the case in a series of scutes a carapace of mosaic pattern would be formed, but it does not appear that a very great degree of rigidity would result as there appears to be a possibility of relative movement at the “sutures.” The scutes overlying the ribs have their longer diameter directed dorso-ventrally and the shorter antero-posteriorly. There is, further- more, reason to believe that, as in the case of the dorso-median scutes, the lateral scutes on the flanks were also regularly arranged in rows. Genus PAREIASUCHUS. 1913. Broom, R., and Haughton, 8. H., Ann. 8. Afr. Mus., vol. xu, p- 25. Dermal scutes are present in the type specimens of both species. Those of Pareiasuchus peringueyt have been described by Broom and Haughton in the following words: ‘Dermal ossicles occur all along the back, between the proximal parts of the ribs, especially in the neck, and around the pelvis. The scutes are smaller than those of Propappus.” Those of P. nasicornis have as yet not been described. Although only a limited number of bony scutes are preserved there is sufficient evidence available to enable one to state that fairly large (65-40 x 55-30 x 15-10 mm.) scutes were present, in at least three rows, all along the back, and that they appear to have been especially well-developed above the shoulder-blades and the sacrum; as preserved, some of the scutes are in contact along their edges, and it is thus legitimate to assume that the armour consisted of rows of scutes in contact with each other. As is apparent from the illustration (Pl. XX VII, fig. 1), the scutes vary in size, shape, and ex- ternal ornamentation. All the preserved scutes are rugose; they have a central knob from which strong ridges radiate; the foramen on the side of the knob is constant, but in one case the foramen is open to the periphery and forms a notch. Genus PROPAPPUS. 1908. Broom, R., Ann. 8. Afr. Mus., vol. iv, pp. 358, 359. 1914. Watson, D., M.S. Ann. Mag. Nat. Hist., vol. xiv. Although only a few scutes of Propappus parvus are preserved in the one known specimen, it appears that the armour differs sub- stantially from that of Propappus omocratus. Pareivasaurian Studies.—Part X. 45 The armour of Propappus omocratus was first described by Broom : “Propappus must have had the whole back and probably sides covered with a carapace of bony plates somewhat after the manner of the crocodile. They vary in size from plates 50 mm. in diameter to little bony nodules about the size of peas. The larger ones have a central boss and irregular radiating ridges. At the edges they overlap each other, and it seems probable that the carapace was not quite rigid. It extended at least as far out on the sides as 300 mm. from the middle line. There is no evidence of any plastron.”’ Some additional remarks may be appended (Pl. XXVII, fig. 2): there is evidence that along the median line in each rump-segment one scute overlay the spinous process, one lay between the spinous processes, and lateral to these one lay on the postzygapophysial ridge, and one in the interspace between this and the next ridge. Broom records an overlapping of the scutes, whereas to me there appears to be an articulation at the edges. Only a few scutes of Propappus parvus are preserved (Pl. X XVII, fig. 2). Judging from these the scutes appear to be more oval (60-50 x 37 mm.) and flatter (8 mm.) than those of Pr. omocratus; the central boss is weaker and the ornamentation much weaker—the radiating ridges being only distinct at the edges of the scute. Genus ScUTOSAURUS. 1922. Amalitzky, V., Bull. Acad. Sc. Russ. 1930. Hartmann-Weinberg, Pal. Zeitschr., Bd. xi. In 1922 Amalitzky, in his brief ‘“‘diagnoses,” mentioned the dermal armour of the Russian Pareiasaurs. In Scutosaurus Karpinski, “star-shaped dermal plates are situated above the spinous processes of the vertebrae, two or three rows of similar plates lie on the sides of the body. There are also plates distributed, without apparent order, over the whole body (fig. 8) (from the illustration these ossicles look very similar to those of Parevasaurus L.D.B.); the belly was covered with small conical bosses.” “ Pareiosaurus”’ tuberculatus has “three rows of star-shaped dermal plates on the sides of the neck; in the tail region the plates are boss-like.”’ : In 1930 Hartmann-Weinberg stated that the osseous scutes of Scutosaurus are of various shapes and sizes. The ornamentation is similar to that of the skull; they lie closely applied to each other to form a heavy carapace. 46 Annals of the South African Museum. Incerta sedis. 1908. Seeley, H. G., P.Z.S., pp. 607-610. In 1908 Seeley described and figured (text-figs. 125 and 126) a number of scutes which he had obtained from Steenkamps Poort. The evidence afforded by the nature of these scutes appeared to him to justify the creation of a new species, Parevasaurus steenkampensis. From his excellent description and figures it appears that these scutes are very similar to those of either Brachypareia rogersi or Pareiasuchus nasicornis, but it would be rash, lacking corroborative evidence from the rest of the skeleton, definitely to assign these scutes to either of these forms. These scutes are thus best referred to as simply Pareia- saurian, with the corollary that they definitely do not belong to forms of the genus Bradysaurus. GENERAL. It is possible, from the above detailed account, to determine the essential characters of the Pareiasaurian dermal armour. One of the most striking features is the apparent correlation that exists between the ornamentation on the dermal cranial bones and that on the dermal ossicles. Most of the dermal cranial bones have a more or less centrally placed boss from which ridges and grooves and bone- fibres radiate, and the sutures between adjoining bones is often of a zigzag nature. Practically the same features are found in the dermal ossicles. Moreover, in the early forms with moderately smooth skulls the dermal ossicles are found to be fairly smooth, and conversely, the rugose skulls of later forms are accompanied by highly ornamented and closely articulating dermal ossicles. It thus appears to be con- sistent with the facts to conclude that the Pareiasaurs during their evolutionary history, in addition to acquiring a greater cranial rugosity, developed an armour of more highly ornamented and more closely articulating dermal ossicles. It would, furthermore, appear that the development of the dermal armour began above the scapulae and above the sacrum, and then extended along the whole back; then along the flanks, and finally above the limbs and possibly also the belly. So also the articulation of the scutes, inter se, appears to have followed the same order of development. There seems to be little doubt that in general the dermal ossicles Pareiasaurian Studies.—Part X. 47 were arranged in regular rows running parallel and at right angles to the dorsal median line. The arrangement of a pair of ossicles per vertebral segment and also the position in relation to the ribs seems to indicate that the scutes were more firmly articulated along their lateral margins than along either the anterior or posterior margin, so that it is perhaps more correct to refer to transverse rows of scutes. I imagine the arrangement of the scutes to have been very similar to those of the living Zonurus. Mention has been made of a foramen entering each ossicle on the external surface on the side of the central bess, and also of a much smaller foramen opening on the internal surface. The former opens into a conical pit on the outer surface. In specially prepared sections it is seen that these foramina communicate; I have examined the scutes in the living Varanus and Zonurus, and find that in the former a nutritive foramen enters on the external surface, whereas in the latter it enters on the internal surface just under the spine. It would thus appear that both foramina in the Pareiasaur scute, in all probability, housed blood-vessels. Consisting of strong bony ossicles, the Pareiasaur armour was certainly of great biological value to these animals. As has been noted, it is first developed above the most vulnerable spots, and it is significant that the forms surviving up to the end of Lower Beaufort times show the highest development of armour. The last Pareia- saurs, although carrying a greater amount of armour, do not appear to have become more clumsy animals. On the contrary, our studies of the limbs and limb-musculature point to a more upright gait accompanied by greater agility. The development of armour in these animals was certainly an acquisition of a character of decided survival value, and, furthermore, does not appear to have eventually overreached itself. 48 Fig Fig. Fig. Fig. Fig. Fig. Fig. Annals of the South African Museum. EXPLANATION OF PLATES. Prate XXIII. . 1. Bradysaurus seeleyi (9168). x 3. Dermal ossicles above 5th, 6th, and 7th presacral vertebrae. The median and first lateral row of each side are represented. . 2. Bradysaurus baini (9001). x 2. Dermal ossicles of the first lateral row above the anterior dorsal vertebrae. PratrH XOX Ly: . 1. Bradysaurus vanderbyli (9169). x 1. Three articulating ossicles above the anterior dorsal vertebrae. 2. Koalemasaurus acutirostris (4345). x 3. Two isolated ossicles. 3. Brachypareia rogerst (8453). x 2. XO IE) Three specimens, obtained by Dr. Davis from the beds which yielded Ischnolepis bancroftt in the lower part of the Lunsempfwa Valley, North Rhodesia, give evidence of the presence of a large Palaeoniscid fish, although they are not complete enough to be generically identifiable. Each will be briefly described. Specimen 9351 (in coll. S. Afr. Mus.). Contains part of the head and pectoral fin. Operculum high and narrow, ornamented with irregular, closely set ridges and elongate On some Karroo Fishes from Central Africa. 103 tubercles. Suboperculum very much smaller, subpentagonal in shape, as wide as high, ornamented with a central mass of rounded tubercles surrounded by irregularly curving ridges and elongate tubercles. Branchiostegal rays numerous—23 being preserved ; rays much wider than high. Pre-operculum long and well developed, situated very obliquely. Bone slightly bent, rodlike posteriorly and expanded anteriorly. Maxilla large, presumably Palaeoniscid in form. Dentition (seen on mandible only) consists of stout conical teeth of two sizes, larger and smaller, rather widely spaced. Pectoral fin incomplete; large, consisting of about 21 rays, of which most show dichotomous forking; fulcra numerous, but small; articulation doubtfully present in anterior rays. Specimen 9350 (in coll. S. Afr. Mus.). Slab and partial counter-slab, showing most of the body, lacking the tail and dorsal fin, together with the badly preserved posterior part of the head, and the pelvic and anal fins. Tentatively, it is considered to belong to the same species as the foregoing, although it is rather smaller. Body fusiform. Depth of body about half the length from posterior edge of clavicle to root of tail. Scales rather small with non-denticulated posterior borders, and ornamented with irregular, rather weak, forked, and anastomosing ridges. Continuous series of dorsal ridge-scales. Pelvic fin rather closer to anal than to pectoral; rays nearly 30 in number, articulated, distally bifurcated; anterior rays longer than fin-base. Anal fin long-based; rays numerous, articulated. Position of dorsal fin not exactly determinable, but probably opposite to, or slightly in advance of, anal. It is certainly not behind the anal. Specimen 9353 (in coll. S. Afr. Mus.). The greater part of a tail, which probably belongs to this species. Tail strong and forked, upper and lower lobes about equal. Rays strong, jointed throughout their length, dichotomously forked. Dorsal lobe has ridge-scales of medium strength; ventral lobe has rather small fulcra. There are shown five strong haemal spines forming a support for the anterior part of the ventral lobe. The body squamation extends for a considerable distance up the dorsal lobe, and ends in a sharp point. On account of the uncertainty concerning the position of the dorsal fin, it is not clear whether these specimens belong to that 104 Annals of the South African Museum. group of the Palaeoniscidae that contains the genera Acrolepis, Gyrolepis, Atherstona, Myriolepis, and Oxygnathus, or to that con- taining Pygopterus and Urolepis. It differs from Acrolepis in the length of base of the anal fin and in the size of the scales. It agrees with Gyrolepis, particularly in the relative narrowness and depth of the operculum and in the nature of the fins; it is larger than most of the described species of that genus, and its scales are smaller and not so deeply overlapping. The dorsal ridge-scales are not so pro- nounced as in Atherstonia, and the body-scales are smaller; the ornamentation on the scales is somewhat similar. Myriolepis differs in the possession of smaller scales and in having a short-based anal fin. In Oxygnathus, too, the scales are smaller and the body is more elongate. The generic position of this form is thus rather doubtful. Of those cited, Acrolepis, Atherstonia, and “‘Oxygnathus”’ (Broom) are known from deposits in South Africa and Madagascar; and it would seem best temporarily to designate the forms under consideration simply as Cf. Atherstonia sp., recognising that there are certain points of difference that mark them off specifically at least from other members of the genus. Ann. S. Afr. Mus., Vol. XX XI. Plate X XIX. ISCHNOLEPIS BANCROFTI Gen. et Sp. N. Co-types: Two fish on Slab 9338. Neill & Co., Ltd. ee ee cm a Re RR oe FE a LN Ae NTN AD AEN GI IT LT LT OLLI LEELA OLD: TLL: Si EO ALLL ALG AL LOL ee ee ed maf Dade Se a a ai wee f y - ae = Plate XXX. Ann. S. Afr. Mus., Vol. XX XI. ‘0SE6 GRIS “USY Teed dg VINOLSYUAHLV fO Neill & Co., Ltd. | Ann. S. Afr. Mus., Vol. XX XI. ¢ Plate XX XI. eres Ojo: Max. Mand. Branchial Rays. Cf. ATHERSTONIA a. Slab 9351. b. Slab 9353. Neill & Co., Ltd. ( 105 ) 4. The Phosphatic Nodules of the Agulhas Bank. (A Study of Sub- marine Geology.)—By L. Cayrux, Membre de l'Institut, Professor at the Collége de France, Paris. (Translated into English by Dr. 8. H. Haveuron.) (With Plates XX XII-XXXV.) THE cruise of the Challenger was responsible for the discovery of phosphate of lime among the marine sediments which are accumu- lating in our day. According to J. Murray, A. F. Renard,* and L. W. Collet, ¢ the Challenger dredged nodules of phosphate of lime on the Agulhas Bank, south of the Cape of Good Hope, on the eastern sides of Japan and of Australia, on the coasts of the Argentine Republic, between the Falkland Isles and the estuary of the Plata, to the west of Chile, and on the coast of Spain. Further, phosphatic nodules have been recovered by the Blake { at numerous points on the coast of the United States between Florida Strait and Cape Hatteras, and in the North Pacific. Finally, the German cruises of the Gazelle and of the Valdiia,§ not to mention the important researches of the Department of Agriculture of the Cape, have dredged much phosphatic material on the Agulhas Bank, previously explored by the Challenger. In our present state of knowledge it can be said that the three great meridional oceans have each yielded phosphatic concretions. J. Murray and A. F. Renard (1891) and, above all, L. W. Collet (1905 and 1908) have given detailed descriptions of the nodules * J. Murray and A. F. Renard, “ Deep-Sea Deposits” (Report of the Scientific Results of the exploring Voyage of H.M.S. Challenger), 1891, Phosphatic Con- cretions, pp. 391-400, pl. xx. + L. W. Collet, “Les concrétions phosphatées de l Agulhas-Bank,” Proc. Roy. Soc. Edin., vol. xxv, 1905, pp. 862-893; “‘Les dépots marins, 1908,’ Les con- crétions phosphatées, pp. 194-213. { J. Murray, “Report on the Specimens of Bottom Deposits (Cruise of the Blake in the Atlantic, 1877-80), Bull. Mus. Comp. Zoology, vol. xii, 1885-86, pp. 37-61; A. Agassiz, Three Cruises of the Blake, vol. i, 1888, pp. 275-276, and vol. ii, 1888, pp. 281-282. § Sir John Murray and Professor E. Philippi, Die Grundproben der Deutschen Tiefsee-Expedition, 1898-99, Bd. xx, 1908, pp. 181-187, pl. xxii. 106 Annals of the South African Museum. recovered from the Agulhas Bank. Then Sir John Murray and Professor E. Philippi have devoted several pages to the concretions obtained by the Valdiia (1908). J. Murray had previously initiated these studies by a brief description (1885-86) of the phosphatic con- cretions dredged by the Blake on the coasts of South America. Although, however, phosphatic concretions have been dredged at a number of places, it would seem that, up to the present, but one true deposit has been discovered—that of the Agulhas Bank. To this reason for giving to the subject particular attention I add another —the very great interest which it presents from the theoretical standpoint. Thanks to the happy intervention of Dr. A. W. Rogers, formerly Director of the Geological Survey of the Union of South Africa, and to the kindness of Dr. EK. L. Gill, Director of the South African Museum, to whom I express my acknowledgments, I have had the good fortune to study some specimens of phosphatic nodules dredged, many years ago, by the Cape Government, under the direction of the late Dr. Gilchrist. The material examined came from nine different stations, ranging from 48 fathoms (88 m.) to 56-0 fathoms (1024 m.). A table of stations is given here.* caaaie Bearings of Station. ie ae Depth. i Lion’s Head, S. 82° E. 27 miles= 43-2Km.| 125 fathoms; 229 m. 2 ve INSGa eee 345) sues) Odea: 154 a DBO nn 3 Vasco da Gama Peak. New] hes.) =) s2e8-oiee 230 - aes 4 Hang Berg, N. 2? E. 29) et g0 om ck 48 ae Shee 5 Cape St. Blaize, N. x EH. 3? E. Yo ss) Gas 105 i 192i 6 Cape Point, N. 44° E. 38 «yy = 60-89) 1315-100 576-731 m. 7 ie N.E. 3 N. BO 5s Oa ee 560 ce 1024 8 a N.E. x E. 3 EK. 2S eee 300 ao 549 ,, 9 Lat. 36° 34’ S., long. 21° 32’ E. Me 240 af 439 ,, 10 Off Vasco da Gama (specimen DOM not 166 . 304 ,, from L. W. Collet, P.F. Stat. XI). Note.—Lion’s Head is a part of Table Mountain, Cape. Vasco da Gama is a headland near Cape Point. Hang Berg is found at Cape Hangklip, on the E. coast of False Bay. To this material is added a nodule of great interest, dredged from a depth of 166 fathoms (304 metres) at station 10 (Vasco da Gama), which has been given me by Mr. L. W. Collet (see Map). * The map (Pl. XX XII) indicates the points of origin of my samples, numbered 1 to 9. From my examination, I conclude that all the stations, except 4 and 5, are situated on the slopes S.W. of the Agulhas Bank. The Phosphatic Nodules of the Agulhas Bank. 107 The nodules recovered by the Challenger came from two very dis- tinct regions. One group was obtained at 98 fathoms (179 m.) and at 150 fathoms (274 m.) on the edge of the Agulhas Bank; the other at 1900 fathoms (3475 m.), and thus in very deep water, nearly 100 miles S.E. of that Bank. Those which were sent to the Challenger office by Dr. J. D. F. Gilchrist, and studied by L. W. Collet, were taken from depths varying from 80 fathoms (146 m.) to 800 fathoms (1460 m.). External Characters of Nodules.—The specimens furnished by the Aculhas Bank and its neighbourhood have a striking morphology, seldom comparable with that of the pebbles of our Albian green- sands. Of a form extremely variable in detail, generally rounded but occasionally very angular (Pl. XX XIII, figs. 1-3), the nodules are irregular, mammiliform, ornamented with protuberances, even deeply indented, and—according to L. W. Collet—perforated by numerous holes. Certain of my specimens show a tubercular and scoriaceous appearance. One of them, cut through the middle, shows irregular funnel-shaped cavities reaching to the centre, and representing from a quarter to a third of the total volume. Representative in an unknown measure of the deposit, these nodules are divisible into two categories. The majority are characterised by an irregular form, and have a colour varying from a blackish-grey to pure black: the others are angular and brownish-yellow in tint. The former are dull, and the latter have a polished and varnished appearance, remarkably pronounced. In other words, the one group resembles concretions, while the other gives the idea of remanié rock fragments. These latter, which have on a single face cavities measur- ing 2 cm. in diameter and 1 cm. in depth, show further, on the same side, tiny perforations resembling cupules, 2 mm. broad, filled with glauconite. These differences of facies do not appear to me at all a function of the bathymetric conditions of the surroundings whence the nodules have been dredged. In this regard, it should be noted that the lightest colours (brownish-yellow) are found at depths of 105 fathoms (192 m.) and of 230 fathoms (421 m.), while the blackish nodules have been dredged from 154 fathoms (281 m.) to 400 fathoms (731 m.) As to the scoriaceous pebbles, I can only say that the most typical of my specimens from this point of view came from the greatest depth, 560 fathoms (1024 m.). Many nodules carry incrustations of an organic nature, already recorded by L. W. Collet, who has identified corals, bryozoans, worms, 108 Annals of the South African Museum. alcyonaria, sponges, and foraminifera with this interesting fact—which I have not been able to observe—that the incrustations can determine the positions of the concretions on the sea-floor and their degree of burial. One part, black and shiny, corresponds to the portion em- bedded in the sediment, and another—grey—to the incrustations defining the upper zone protruding from the mud. The incrustations visible on my samples, mostly due to bryozoa, are seen sometimes over the whole surface—proof that the nodules have been rolled in every direction—and sometimes over a part only. Several lack incrustations. According to J. Murray and A. F. Renard, the dimensions of the concretions vary mostly from 1 to 3 cm., and exceptionally reach 4-6 cm. L. W. Collet states that they generally lie between 5 and 10 cm. The largest, dredged from 111 fathoms (304 m.) at 25 km. from the coast (St. 10, Vasco da Gama), measures 23 x 16x12cm. The largest of my specimens does not exceed 7:5 cm. in greatest diameter. From the descriptions and figures given by L. W. Collet, Sir John Murray and Prof. E. Philippi, it is concluded that the nodules are either isolated or cemented to form a true conglomerate. As far as I can judge from the few indications furnished by the gangue of the nodules, it is formed of greensand more or less consolidated. The latter authors have figured a curious grouping of perforated nodules.* It must be noted here that, whatever their shape. the nodules have acquired the consistency shown by the hardest pebble of the Albian of the Paris Basin. Chemical Composition of Nodules—J. Murray and A. F. Renard have published the following analysis of concretions dredged at Station 142, at 150 fathoms (274 m.), on the edge of the Agulhas Bank :— f PSOE : . 19-96 per cent. CO, Skt. SS 2 Eee SO, ye lesle Jducon’ leeagtiedae ere Si0, ; ; s6n sae Ghee ae ee) bo oe ee Oe MgO : : -,_ 0°6t Fe Fe,0, ’ ta eee ALOLG st Ae ure eae gaara Loss f : : ; ar Insoluble residue SVs: r 95-89 ny Phosphate of lime . 2 4 S-Di * Sir John Murray and Professor E. Philippi, op. cit., pl. xxii, fig. 1. + J. Murray and A. F. Renard, op. cit., pp. 392-393. {t Not estimated owing to an accident. The Phosphatic Nodules of the Agulhas Bank. An analysis of a nodule dredged by the Valdivia * gave: S10, . 26-70 per cent. Ca3(PO,). : ES Sc eee CacOs 7\- ST AO FOS y's: CasOn vr”. : : a WIEZ 6 anit, MeC@Osnerk « f : 4-67 i Fe,O, ' 8 es a Al,O3 ; : RRS ae 102-38 109 Two analyses of yellow nodules, made by L. W. Collet, showed respectively 37-79 and 46-69 per cent. of phosphate of lime, and 21-09 and 23-70 per cent. of Fe,O3;. Two other specimens studied by the same author, taken from one and the same station, yielded 48-70 and 52-05 per cent. phosphate of lime. Three specimens which I have analysed, without estimating all the elements present,{ are characterised by the following compositions :— (10). (7). (8). SiO, LS SR ee 21-90 LLOo a 3-00 3-00 Hee eee, 6-25 6-25 BeOmea ce 2 (Te lB 7-80 31-70 MgO Radtaee Ayia ety cs yt 1-16 1-80 POs Ree ils if nd 1125/1 dyallOo86 18-47 a Yor 1-19 0:97 ol hia 0:25 0-24 Sar re POR a 8-14 4-20 EO 2; yoy ee ae 5-75 7-20 99-50 95-73 13-60 2-10 6-30 38-40 1-44 18-60 1-35 0-32 7:93 6-50 96-54 (7) Dredged at 560 fathoms (1024 m.), Cape Point. (8) Dredged at 300 fathoms (549 m.), Cape Point. (10) Dredged at 166 fathoms (304 m.), Vasco da Gama. * Sir J. Murray and Professor Philippi, op. cit., p. 185. t+ L. W. Collet, “Les concrétions phosphatées de Agulhas Bank,”’ Proc. Roy. Soc. Edin., vol. xxv, 1905, pp. 868 and 870. { In particular, the alkalis ought to figure in a higher proportion in the complete analyses of the nodules from Stations 8 and 10, on account of the frequency of occurrence of glauconite. VO. mex, PART I. 13 110 Annals of the South African Museum. The composition of these three nodules leads to the conclusion that the phosphate of lime is a fluocolophanite, a mineral to which A. Lacroix attributes a formula susceptible of variation : * z[(CaFl),Ca,(PO,),_] + yCaCO, + zH,0. From this composition it can be inferred that, from the chemical point of view, the nodules of the Agulhas Bank are allied to the group of phosphates of the ancient sedimentary series. These analyses show a remarkable similarity of content of phosphate of lime to those of J. Murray and A. F. Renard, and of Sir John Murray and EK. Philippi. The figures of L. W. Collet (values of 46-69, 48-70, and 52-05 per cent. of phosphate of lime), whilst sensibly higher, do not invalidate the conclusion to be drawn from the other analyses, that the nodules of the Agulhas Bank fall within the category of poor phosphates. I would note, without stressing the point, that the figures of phos- phoric acid content are of the same order as those of the Albian greensands of the Paris Basin. MICROGRAPHIC STUDY. J. Murray and A. F. Renard have described and figured two types of concretions: (1) those found in the greensands of Station 142 of the Challenger, dredged at 150 fathoms (274 m.), compounded into aggregates, formed of minerals in the maximum proportion of 2/3, cemented by phos- phate of lime, with organisms absent; + (2) those found in the Globi- gerina ooze, and found at a depth of 1900 fathoms (3475 m.) (Station 143), containing a preponderance of Foraminifera.{ The occurrence of the latter being independent of the Agulhas Bank, I leave them provisionally aside, but not without remarking on the extraordinary bathymetric dispersion of the nodules in question. I define here briefly six varieties of nodules, and afterwards shall consider them together. (1) Calcareo-phosphatic nodules, very quartzitic. [No. 1, 125 fathoms (229 m.), Lion’s Head, and No. 8, 300 fathoms (549 m.), Cape Point (Pl. XXXIV, fig. 1).]|—Blackish nodules, the richest in detrital matter of the whole series. Very uniform in size and scarcely ever exceeding 9? comprising glauconite, quartz, and “silicates, * A. Lacroix, Minéralogie de la France, t. iv, 1890, p. 561. + J. Murray and A. F. Renard, op. cit., pl. xx, fig. 1. { J. Murray and A. F. Renard, op. cit., pl. xx, figs. 2-4. The Phosphatic Nodules of the Agulhas Bank. LTTE 0-05 mm. in diameter (No. 8); the quartz grains (q) are unequally scattered in the restricted space of a section, of which they form a third, a half, or even more. On the average, they represent less than 50 per cent. of the deposit. Contrary to the state of things in the nodules of the greensand, described by J. Murray and A. F. Renard, glauconite (g) may be rare, although one may count several dozen small grains in a preparation (No. 8). In the other nodules (No. 1), the mineral is thinly scattered. The fauna is characterised by the presence of some Globigerinae. The gangue is sparingly developed, and is calcareo-phosphatie. As can be seen, the rock is far from being a typical greensand. (2) Phosphatic or calcareo-phosphatic nodules, very glauconitic, with Globigerinae. [No. 4, 48 fathoms (88 m.), Hang Berg; No. 3, 230 fathoms (421 m.); and No. 7, 560 fathoms (1024 m.), Cape Point. ]— The nodules from Stations 4 and 7, blackish and encrusted with glauconite, show indentations filled with this mineral. The grains of quartz and of glauconite alternate, but the glauconite has the greater relative volume. In the specimens examined, the quartz forms grains of variable diameter, seldom greater than 0-13 mm., or elements whose diameter is always less than 0-10 mm. The glauconite is distributed in a most irregular manner, being totally absent sometimes, and at others forming by itself the greater part of the pebble. The Globigerinae are widely distributed, so much so that there is very little cement. Between this type of nodule and the preceding the differences are purely quantitative, and they are linked by transitional varieties. (3) Calcareo-phosphatic nodules with Bryozoa and Pulvinulinae. [No. 9, 240 fathoms (439 m.), 8. of Agulhas Bank.]—Nodules of very deep colour, in which the quartz, represented by grains measuring at the most 0:06 mm.—0-07 mm., is noticeably less. Glauconite is, on the average, less widely distributed or rare. Numerous very fragmentary remains of Bryozoa appear, of which at least a dozen can be counted in a section. The Globigerinae are considerably reduced in number, a reduction which is only partly compensated by a very marked increase in Pulvinulinae, which are generally char- acterised by a very thick test. The amount of gangue material is noticeably increased. The appearance is thus very distinct from that of the preceding types. (4) Calcareo-phosphatic nodules, quartzitic, and with abundant Glo- bigerinae. [No. 3, 230 fathoms (421 m.), Vasco da Gama (Pl. XXXIV, fig. 2).]—These nodules differ greatly from the preceding in their 112 Annals of the South African Museum. buff-yellow colour. The quartz-grains (q) are more numerous, but do not show the same degree of frequency as in the nodules of the first two categories. Fairly variable, the maximum diameter reaches 0-13 mm., while the minimum may be less than 0:05 mm. Glauconite is only an accessory mineral. The principal characteristic of this type is the abundance of Globigerinae and the presence of spicules of calcareous Sponges. Gangue material is very unimportant. The matrix of the nodules formed in this way is a passage-deposit to the true Globigerina mud. (5) Phosphatic and ferruginous nodules with Globigerinae. [No.5, 105 fathoms (192 m.), Cape St. Blaize (Pl. XX XIV, fig. 3).]|—These nodules have an angular shape and brown colour, are almost free of quartz (0-05 mm.—0-10 mm.), very poor in glauconite, and particularly rich in Globigerinae. Their cement, which is both phosphatic and ferruginous, has preserved the organisms in an exceptionally beauti- ful manner. Numbers of these are riddled with perforations filled with oxide of iron. This type of nodule offers most affinity, of all the series examined, with the Globigerina ooze. (6) Phosphatic and ferruginous nodules with large benthic Fora- minifera and Bryozoa. [No. 6, 105 fathoms (192 m.), Cape St. Blaize (Pl. XXXIV, fig. 4).]—Although coming from the same station as the foregoing, these nodules have almost nothing in common with them (compare figs. 8 and 4). The organic remains, often so abundant that they touch one another, give to the thin section a very pro- nouncedly coarse appearance. They consist principally of the remains of benthic Foraminifera with extremely thickened tests, accompanied by accessory pelagic forms, especially Globigerinae, in a very small proportion. There are also fragments of bryozoan colonies and of parts of Echinoderms (plates and spines), both fairly frequent, and some fragments of the shells of Mollusca and of Brachiopoda, to the absolute exclusion of detrital minerals and of glauconite. In this chaotic complex, the pelagic Foraminifera may be arranged in rows. Most of the organisms are riddled with perforations which are filled with phosphate of lime and oxide of iron. Many are in- complete and penetrated by the matrix. Amorphous phosphate of _ lime and iron peroxide play a very important part both in the fossilisation of the test and in the infilling of the cavities. To these six categories I add another, very different from the preceding, and constituted of nodules of phosphatic greensand, whose characters will be discussed later (Pl. XX XV, fig. 3). From the examples thus passed in review, which possibly give an The Phosphatic Nodules of the Agulhas Bank. 113 incomplete idea of the mode of occurrence, it is possible to judge of the variety of phosphate rocks represented in the deposit. These rocks include types obviously terrigenous, such as the calcareo-phos- phatic, very quartzitic, nodules and also the phosphatic and ferruginous nodules with Globigerinae, which resembles pelagic formations. Be- tween these two extremes is ranged a series of intermediate types whose terrigenous characters gradually disappear, step by step. The variety which is characterised by Bryozoa and numerous benthic Foraminifera appears to stand at the border of the complex. I see in it a representative of the numerous bottom deposits of organisms, formed in a terrigenous environment, which never figure in the classification of true marine sediments. In order to state fully the problem for solution, it is necessary to add that these phosphatic rocks have been dredged from a bottom of greensand, and that the specimens which most closely approach a Globigerina ooze—although deviating from the normal type by a very small proportion of quartz granules—have been recovered at a depth of only 105 fathoms (192 m.), 7.e. well outside the actual domain of the Globigerina oozes. As far as I know, no deposit of phosphate of lime in the older sedimentary series contains such a variety of types nor raises a problem of such importance. ANALYSIS OF THE CONSTITUENTS OF THE NODULES. The little that we know, up to now, of the constitution of the nodules scarcely allows us to suspect the great imterest which is attached to the elucidation of the general history of phosphatic concretions of all ages. Therefore I describe successively all the elements which go to form the nodules, and then discuss the structural details which give an unforeseen picture of the mode of origin. 1. Minerals.—From one end to the other of the series analysed, quartz shows a great variation in frequency, since the grains may be predominant in the first variety and completely absent in the last. These variations entail only slight changes in the quantity of the elements, without showing a modification in shape. All are angular, with a mean diameter which is always less than 0-10 mm.; hence nearly all the grains would float in water that was only feebly agitated. The angular form of the grains, already noted by J. Murray and A. F. Renard, is a consequence of this. From our point of view, one character of this mineral dominates all the others: the variations in 114 Annals of the South African Museum. size and in the degree of frequency of the quartz grains are independent of the depth at which the nodules were obtained. As to the “silicates” mentioned by J. Murray and A. F. Renard, they are undoubtedly related to rather rare felspathic elements, and comprise triclinic felspars. Glauconite.—It must be confessed that the numerous works devoted to this mineral and its history are far from exhausting the subject, as the following data show:— J. Murray and A. F. Renard have shown, in their description of the Agulhas Bank phosphates, that the glauconite, which has a cryptocrystalline structure like that of the glauconite of older deposits, is present in rounded grains independent of organisms, and as a product of the material filling the interior of numerous Foraminifera, in which it is very common. Their study has been very usefully completed by Messrs. L. W. Collet and G. W. Lee,* who discovered in these same phosphates the variety which I formerly designated tf as pigmentary glauconite. Bearing in mind their observations, I shall abstain, as far as possible, from repeating what is already known in this matter. An analysis by Giimbel { of glauconite from the Agulhas Bank gave the following results :— S10, : : . 46-90 per cent. Al,O; : 2. «a OGL eae FeO, ; + © 20-09" Shee FeO f : ; 3°60 y CaO : “20:20 MgO . 0:70 i K,O : : J Os 167 a Na,O : EDS ee H,O oe eee 99:24 wats In preparing for examination nodules from the Agulhas Bank, I have been able to gather new data, some of which concern the glau- * L. W. Collet, ““Les concrétions phosphatées de PAgulhas-Bank avec une note sur la glauconie qu’elles contiennent par G. W. Lee,” Proc. Roy. Soc. Edin., vol. xxv, 1904-1905, pp. 885-893; L. W. Collet and G. W. Lee, “Recherches sur la glauconie,”’ Proc. Roy. Soc. Edin., vol. xxvi, 1906, pp. 266-267. + L. Cayeux, “Contribution a l’étude micrographique des terrains sédimentaires,”’ Mém. Soc. Géol. Nord, t. iv, 2, 1897, pp. 165 and 175. t V. Giimbel, “‘Ueber die Natur und Bildungsweise des Glaukonits,” Sitz. d. Math. Phys. Clas. d. k. Akad. Wiss., Miinchen, t. xxvi, 1896, p. 545. The Phosphatic Nodules of the Agulhas Bank. 115 conite of the sediments whose age will be discussed later, and others the history of glauconite in general. (A) Despite the frequent occurrence of glauconite grains on the one hand and of Rhizopods on the other, it has not been possible to find a single glauconite element enclosing a Foraminifer. (B) The relations between glauconite and Foraminifera give material for interesting observations. In general, the mineral invades all the chambers, but it may avoid one or two, and it is also capable of associating there with phosphate of lime. The chambers once being filled, the glauconitic complex which results either ceases to develop—a state exhibited by absolutely intact shells of Foraminifera filled with glauconite—or the glauconite continues to form. If the latter, then a whole series of results is presented: (a) Exceptionally the material enters the pores, without any trace of metasomatism. (b) The glauconite fixes itself in the test, no matter where, in the form of minute globules, free or coalesced, which destroy the micro- structure of the shell. (c) The whole test becomes metasomatised by the glauconite, darker than that of the cells, which admirably preserves the structure, for, curious to state, the pores are not penetrated by it. Globigerinae, fossilised in this manner, yield beautiful tangential sections in the form of a network with very regular mesh, whose distinctness recalls sections of crinoids mineralised by ferruginous compounds. To these examples of pseudomorphosis of foraminiferal shells by glauconite—which, it must be added, are by no means rare—I add fragments of foraminiferal tests reduced to simple isolated arcs grouped in such a manner that their fragmentation was certainly anterior to their deposition. These too are metasomatised without an infilling of the pores, and yet do not show the slightest impression of glauconite on the matrix. In certain sections, these fragments of shells converted into glauconite are frequent. This is the first occasion on which I have been able to observe such a disposition to metasomatism on the part of foraminiferal remains without a resultant destruction of their individuality and of their microstructure. (d) In a variant of this, the whole test is metasomatised and the pores are invaded. Thanks to its darker colour, the surrounding material is perfectly differentiated from that within the shell. As in the preceding case, it is impossible to notice the slightest projec- tion into the matrix, which could lead one to suppose the existence of 116 Annals of the South African Museum. a passage between the glauconitic moulds and the grains independent of the Foraminifera. (ce) In a fifth mode of occurrence—which, as in the preceding, involves a substitution of glauconite for carbonate of lime in Rhizopod shells—these appear as moulds, without trace of test. But in many cases there can be seen a thin border, of a very pale yellowish-green colour, which, between crossed nicols, is distinguished by brighter tints than the rest. Actually, there has been metasomatism of the test, and orientation of the glauconite in such a manner as to differ- entiate it more or less in white hght and between crossed nicols. This example shows that it may be necessary to discriminate casts of Foraminifera which may simulate whole Foraminifera, test included. The absence of such differentiation, which was only observed in a few casts, leads one to assume the existence of elements which originated within the cells of the Foraminifera, and were subsequently extracted from them by a process of mechanical decortication. This is wholly uncertain, and it is possible that the whole question of the inter- pretation of the glauconitic casts of Foraminifera may be studied anew. (C) Foraminifera, which are often profusely distributed in the nodules, are not the only organisms more or less altered into glau- conite. Leaving aside the Foraminifera, which play the principal role in the fixation of the mineral by organisms, we can pass at once to the groups of Brachiopods, Bryozoa, and Mollusca, not pausing at groups which are without interest. In the Bryozoa, glauconite plays a part in the infilling of the chambers; but most of the specimens do not contain it. Some Brachiopod fragments are penetrated by glau- conite in the form of irregular globules, which may be free, in groups, or fused together. (D) Apart from the glauconite grains associated with organic remains, the mineral also acts as a cement. Of this type of occur- rence one example only can be cited, which was observed in the most quartzitic nodule of all. In it was found a small stony layer of quartz grains accompanied by felspars, all bound together by a very pure glauconite which encroached on the neighbouring calcareo- phosphatic cement, both as irregular prolongations and as small completely isolated particles. Considering the total amount of glauconite of altered appearance, it can be seen that the percentage of this mineral which acts as gangue material reaches a figure that is wholly unknown in the older rocks. Those nodules which are of a brown colour and usually The Phosphatic Nodules of the Agulhas Bank. Ly angular shape possess a brown cement of various shades in which the obvious calcareous islets form but a very small proportion of the whole. (EK) Another variety shows glauconite present in veinlets of very small size, similar to those that have been noted in nodules of Albian age.* It will be seen later that, under certain conditions, glauconite in this form may have a wide distribution. (F) Not one of the least interesting facts to record is the profoundly altered appearance of glauconite in certain nodules. All things considered, this alteration—real or apparent—falls into two cate- gories which appear to denote two distinct phenomena: (a) In the midst of a layer rich in grains of very pure glauconite, one may see, é.g., a single and unique grain which is brown at the periphery, or several grains laden with oxide of iron to varying depths, the alteration sometimes reaching the centre and leaving only insignificant traces of the original mineral. Decomposition may affect all, or nearly all, the glauconite grains in a layer, thus freeing oxide of iron which partly remains in place and partly masks the neighbouring cement. The same phenomenon is produced at the expense of glauconite included in the organisms. In both cases, the iron oxide invades the gangue material and produces within it tints of very variable colour, a proof that the decomposition has taken place on sztu. (6) This occurrence of glauconite carrying iron oxide, playing the role of cement, in angular brown-coloured nodules, is probably . connected with another phenomenon. Nota single layer has escaped what appears to be, at first sight, a very characteristic alteration which is developed in varying degree. Some parts, which remain greenish, are partly masked by a little oxide of iron; others have become deep brown, but the colour retains a greenish shade, without the cryptocrystalline structure, so characteristic of the mineral, being destroyed. Gradually the colour turns to dark brown. All this takes place, therefore, as if there were a production, on an un- paralleled scale, of glauconite playing the part of cement, followed by a no less general alteration of the mineral, liberating oxide of iron. The phenomenon should be important, since L. W. Collet has found up to 21-09 and 23-70 per cent. of FeO; in the yellow nodules. f That is probably an incomplete picture of the sequence of events leading to the present condition of things. * L. Cayeux, op. cit., p. 181. + L. W. Collet, op. cit., pp. 869 and 870. 118 Annals of the South African Museum. This opinion is founded on the existence in the apparently decom- posed glauconitic gangue of grains of absolutely unaltered glauconite. On an hypothesis of one alteration, it is necessary to suppose that glauconite of the second generation is deeply decomposed, whilst that of the first generation, corresponding to the grains, has re- mained unaltered in all the samples examined. The conclusion, therefore, seems inevitable that the glauconite of the second genera- tion has kept its present nature right from the time of its formation. In other words, for unknown reasons, certain reactions caused the mineral to produce an excess of oxide of iron, and the so-called altered glauconite was in reality, at the time of its formation, nothing but glauconite coloured by oxide of iron. This question was raised by L. W. Collet, who wrote concerning the yellow nodules of the station Vasco da Gama: “Une idée qui vient naturellement 4 l’esprit est que cette couleur des nodules jaunes pourrait étre due a de l’oxyde de fer provenant d’une décomposition de la glauconie, ou que les conditions n’étant pas satisfaites pour la formation de la glauconie, ou hydrosilicate de fer et de potasse, il se soit formé un hydrate ou un oxyde de fer.”” * Accepting the fact that a web of typical glauconite is found behind the veil of iron oxide, as is shown by examination between crossed nicols, it appears more logical to admit that the reactions produced glauconite surcharged with peroxide of iron. It is not impossible that this problem likewise affects certain glauconites of the older deposits that appear to have suffered general alteration. Whatever the solution of this problem may be, the development on a large scale of glauconite as a cement-mineral is certain. Conclusions.—As will shortly be learned, the details enumerated by no means exhaust all the facies adopted by this mineral in the nodules. However, three salient facts have been ascertained up to the present: (A) The mineralisation of foraminiferal shells by glauconite is a very rare phenomenon. The specimens analysed here in detail have furnished examples which are incomparably more numerous than those of the sedimentary formations that hitherto have been submitted to micrographic study. (B) In the light of what the nodules examined have shown, no proof is given that the free grains originate from glauconite primi- tively englobed in the chambers of Foraminifera. Of the authors who have previously occupied themselves with the problem of glau- * L. W. Collet, op. cit., pp. 874 and 876. - |= The .Phosphatic Nodules of the Agulhas Bank. 119 conite origin, some conclude that all the grains, of whatever kind, have originated in the interiors of organic cavities and, particularly, in foraminiferal chambers. This is, in particular, the opinion of Ehrenberg, W. J. Sollas, Bonney, J. Murray, and A. F. Renard, etc. Others, such as J. Bailey, A. E. Reuss, V. Giimbel, etc., believe in a dual mode of formation. The hypothesis—often propounded—that the innumerable elements whose shape does not suggest organic origin result from the irregular growth or fusion of casts of Foraminifera receives no support, in spite of the fact that the occurrence yields observations particularly decisive in this respect because of the large number of nodules and of free grains found side by side. No example of a cast passing into an irregular grain has been seen, and no example of an irregular grain enclosing a foraminifer with its chambers obliterated by glauconite. In every case, without exception, the two types of glauconite grains present are entirely independent; my absolute conviction is that, in this deposit, glauconitic casts remain glauconitic casts, and that grains, whatever their size and form, never originate from them. Many examples, taken from elsewhere, help to give this conclusion a general application. I add that the development of the epigenic glauconite of the cement and a mode of occurrence on which I shall lay stress later prove, once more, that glauconite can be formed on a large scale independently of organisms. (C) Incomplete as it is, the preceding study shows the existence of at least two generations of glauconite—one consisting of the free grains and casts of Foraminifera, and the other the epigenetic glauconite of the matrix and the rare veinlets. 2. Organisms.—Beginning with the highest forms, it is necessary only to mention the exceptional presence of splinters of bony tissue, which are usually absent. To the Brachiopoda are assigned a small number of fragments of tests and of entire shells which, on the average, number at the most one example in each section. Both have preserved their structure intact, and the canals may, or may not, be filled with phosphate of lime. Less rare than the Brachiopods, Bryozoa—represented by very fragmentary colonies—do not appear in all the thin sections; their maximum frequency is 20 individuals in one section—a notable proportion, having respect to the size of the débris. The Mollusca, which form the most important group after the 120 Annals of the South African Museum. Foraminifera, are chiefly found in the angular brown nodules. There occur some dozens of curved, fragmentary, and usually thin forms with their microstructure preserved. To the Echinoids are assigned some calcareous plates, with a single optical orientation, whose microstructure is almost entirely destroyed, as well as some exceedingly rare spines. All occur in the brown nodules, to the maximum number of a dozen plates in a section. The very rare Sponge spicules observed are calcified. The one example, unique up to the present time, proves that the solution of siliceous spicules and the replacement by calcite are not necessarily late phenomena posterior to the emersion of the sediments. The Foraminifera, however, easily surpass the preceding organisms in importance and numbers. Except in the brown nodules with Bryozoa and large benthic Foraminifera, Globigerinae predominate, to the point of forming by themselves the whole Rhizopod fauna. Generally speaking, the Foraminifera are whole and beautifully pre- served. With the exception of the brown nodules, rich in bottom- dwelling Foraminifera, the shells, both entire and fragmentary, show no corrosion, and it is impossible to imagine the fragmentation of the Globigerinae otherwise than by mechanical action. Boring organisms have left profound traces in the brown nodules in the form of numerous irregular tubes filled with phosphate of lime, ferruginous glauconite, or opaque oxide of iron, most of which are probably attributable to sponges. -Except for the sponge spicules and the Echinoid remains, all the groups have suffered from their action. 3. Cement.—From the point of view of the constitution of the matrix, the nodules form very dissimilar groups—the brown nodules on the one hand, and the blackish nodules on the other. (A) By their matrix the latter recall immediately, under the micro- scope, the phosphatised chalks of the North of France. Although on the average little developed, on account of the profusion of Foraminifera, the cement is generally calcareo-phosphatic, at places phosphatic and calcareous and, very exceptionally, formed of pure phosphate of lime in very restricted areas. In white light, the phosphate of lime is identified by a very pale yellowish tint, which becomes most pronounced as the phosphate content increases and, finally, attains a pale yellow colour. Under crossed ricols, as a rule the yellowish base becomes charged with very numerous small brillant and iridescent spots, representing the presence of carbonate of lime, The Phosphatic Nodules of the Agulhas Bank. 121 which are prominent when phosphatisation is feeble but disappear completely in the exceptional layers which are wholly phosphatic. In short, it is necessary to examine but a very limited number of nodules to observe a complete passage from a matrix exclusively calcareous to a cement which is entirely phosphatic. It follows that, on the average, these nodules are rather poor in phosphoric acid. (B) Actually, the brownish nodules—which differ considerably from the preceding—are far from forming a constant variety. Although limited to two specimens, my studies have shown profound differences between them. The one which is considered a partly phosphatised Globigerina ooze possesses a cement that is essentially glauconitic and more or less haematitised (Pl. XX XIV, fig. 3), enclosing some plates of Globigerina mud free from mineralisation, exceptionally converted into mostly crystalline calcite, and others—rather rare—which are on the way to being phosphatised. It is difficult to say if the matrix, which is strongly impregnated with oxide of iron, is not at the same time slightly phosphatised, since the brown base prevents the identifica- tion of phosphate of lime. The cement of those brownish nodules that are rich in large benthic Foraminifera has visible phosphate of lime in preponderance. Some- times this mineral surrounds uniformly all the perfectly preserved constituents in a yellowish base, which is undifferentiated and full of minute calcareous specks; sometimes it is associated with calcite, which is mostly crystalline, and forms a thin, pure yellow border to all the material which it serves to bind. In this variety the organic remains are corroded and are often almost unrecognisable. Itis rather the calcite which actually plays the part of cement. The same slide shows both these modes of occurrence, to which can be added a third, characterised by a ferruginous matrix (Pl. XX XIV, fig. 4) like that of the preceding specimen, which does not permit one to say that the oxide of iron is intimately associated with glau- conite, although this is probable. This ferruginous matrix may also enclose calcite in the form of irregular patches of varying size. The history of these nodules is thus complicated by a profound alteration of the original constituents of the ooze which have escaped phosphatisation and mineralisation, these constituents having now a crystalline condition comparable with that of the highly recrystallised ancient limestones. There are even true veinlets of calcite in the mass, like those found in ancient marbles. These are, it is true, few in 122 Annals of the South African Museum. number, and one cannot invoke dynamic alteration to account for them. Mode of Occurrence of Phosphate of Lime.—At the outset, phosphate of lime plays a double réle: (A) It fills the chambers of Foraminifera, the cellular fabric of echinoderm plates, the cells of Bryozoans, etc. Actually, this mode of occurrence 1s bound up with the following, for the phosphate of lime in the foraminiferal chambers retains traces of calcite which bear witness to epigenetic phenomena. The truth is that, despite appearances, phosphate of lime is not concentrated as such in these spaces. (B) It is produced on a large scale by alteration of the calcareous matrix of the foraminiferal muds and, on a reduced scale, in the tests of the organisms present. It follows that, in its two principal modes of occurrence, phosphate of lime is undoubtedly a substitution product. The phosphate of lime formed at the expense of the cementing material is optically amorphous and lacks any kind of morphological differentiation; it therefore belongs to the group of colophanite or, more exactly, of fluocolophanite. This is so for all the nodules ex- amined to date. It will be shown later, from a study of aberrant specimens, that a crystalline variety may also occur, but seemingly exceptionally. Granular phosphate of lime is extremely rare, when it is not completely absent. There are doubtless exceptions to this rule, for J. Murray, examining material dredged by the Blake, noted that a concretion, obtained in the Atlantic at Station 317, in lat. 31° 57’ S., contained phosphatic grains “similar to the grains found in the phosphate nodules dredged near the Cape of Good Hope and elsewhere by the Challenger.” * Finally, the alterations undergone by the parent rocks, converted into poor phosphates, are much less profound than those noted by J. Murray and A. F. Renard in the nodules from the Indian Ocean where, at the same time, are integral substitution, destruction of organic characters, development of zonary structure, and crystallisa- tion of the phosphatic material. * J. Murray, “Report on the Specimens of Bottom Deposits,” Bull. Mus. Comp. Zool., vol. xii, 1885-86, p. 43. + J. Murray and A. F. Renard, op. cit., pl. xx, figs. 3 and 4. The Phosphatic Nodules of the Agulhas Bank. 123 ANOMALIES OF COMPOSITION AND STRUCTURE OF . NODULES. Among the nodules of dark colour which arise from quartzitic Globigerina muds are some that are characterised by the presence of inclusions, traces, and incrustations of aberrant constitution, and some by the juxtaposition of two very dissimilar rocks. 1. Nodules with very Glauconitic Inclusions.—(A) Glauconite gives rise to narrow bands with well-defined contours, from which minerals and organisms are absent. These differ from the grains disseminated through the rock by a light brownish colour, indicative of an elimination of peroxide of iron. In this form the mineral presents its true appearance. This same variety serves as a cement to large grains of glauconite of a normal colour, strikingly different from those that form an in- tegral part of the nodules. Thus it is possible to see well-defined bands formed exclusively of glauconite grains of one shade (g), surrounded by a glauconitic matrix of an entirely different shade (Pl. XXXV, fig. 1). Thus there are present two generations of glauconite which have nothing in common, in point of age, with those concerning which question was raised in the study of nodules of normal type. These can, moreover, be seen in a given nodule. (B) The inclusions are rich in grains of ordinary glauconite, which may form the greater part of them, and often touch one another. Apart from insignificant exceptions, the mineral is very clear, except when it is obscured by the neighbouring plates. With few exceptions, the inclusions in this category show no trace of Foraminifera or other organisms. Their cement is entirely, or almost entirely, phosphatised and cryptocrystalline, i.e. it is a staffelite. In general, the phosphate is differentiated as a very clear border round grains of glauconite and of quartz. This differentiation is also observed, but not always, around the islets and bands. The relations of these inclusions to the neighbouring rock are illustrated by the two following examples :— In the first, there is a beautifully distinct band of glauconitic phosphate which is bounded on one side by a curved line traversing a large plate without deviation, cutting as with a knife the glauconite grains and phosphatic constituents which it encounters. On the other side, the boundary is irregular and roughened with protuber- ances and hollows in a manner that is immediately reminiscent of the phenomena of corrosion. Moreover, along this line of separation the ia | 124 Annals of the South African Museum. nodule is strongly impregnated with glauconite, just as is frequently found on old hard and corroded floors. But the relations between these two portions are susceptible of quite another interpretation. The cementing of the two rocks follows, then, a very irregular line, whose limits it is often impossible to fix, and the rocks merge into each other at their junction. From the study of a number of thin sections it is possible to obtain an idea of the form of this sort of inclusion. They represent, in fact, a product of the partial or complete infilling of cavernous or scoriaceous holes in the nodules. Such of these bands as can be followed through- out their whole development leave the bottom of a hollow in the form of a funnel and penetrate into the heart of the nodules. In short, bands and islets are really the same, and simply represent respectively longitudinal and transverse sections of cavities filled by greensands. The specimens which favour this type of observation correspond, I believe, to the concretions “avec fentes remplis d’une substance verdatre ’’? which appears to L. W. Collet “pouvoir étre attribuée a de la glauconie.”’ *- Nothing is more evident than that this material exhibits an inter- penetration of two fundamentally distinct types of rock—an older, not typical, Globigerina mud and a true greensand, both consolidated. The same contrast exists between the material of the nodules and their mineral incrustations, which do not differ at all from the islets and bands. Under these conditions it 1s impossible to escape from the con- clusion that the nodules, formed at the expense of non-typical Globigerina muds, have been removed from the locality in which they were formed and deposited in areas where greensands were being laid down. From this conclusion is derived another: In the process, the nodules have been transported from one sediment more pelagic than terrigenous into another plainly terrigenous, and from greater depths to more shallow areas. 2. Nodule formed of two differing Rocks (Pl. XX XIII, fig. 1, and Pl. XX XV, fig. 2).—Station 10 (Vasco da Gama), situated 25 km. from the coast, has yielded a set of “‘concretions,”’ among which is the largest known, as well as rolled pebbles of quartzites and schists. In one of the nodules which he has figured, L. W. Collet has clearly distinguished two portions separated by a black line 0-5 mm. thick, one of which is yellowish and characterised by numerous Foraminifera, the other blackish and rich in grains of glauconite.t This nodule, whose * L. W. Collet, op. cit., p. 867. + L. W. Collet, op. cit., p. 869 and fig. 3. The Phosphatic Nodules of the Agulhas Bank. 125 greatest diameter is a dozen centimetres, is formed of two distinctly different rocks, joined in such a manner that it would not be correct to say that one serves as a matrix for the other. On a polished surface (Pl. XX XIII, fig. 1) there can be seen a lower part of clear yellowish colour, darkened above over a width of about 1 cm., and an upper part of very deep tint in which, under the lens, a crowd of little black glauconitic grains can be distinguished. (A) The yellowish portion (Pl. XX XV, fig. 2, a), which was clearly the lower part of the specimen in place and which, according to L. W. Collet, was “‘formée presque entiérement de coquilles de Foramini- féres,’’ must have a variable composition. Actually, sections cut from the same specimen display but a small proportion of Foraminzfera and, it is interesting to note, an irregular distribution of the organic constituents—a fact which doubtless explains the discrepancies between our observations as to the quantity of Foraminifera present. However, as far as I have seen it, the yellowish rock contains a pre- ponderance of fragments of Bryozoa and, in addition, Foraminifera which are rare or relatively frequent and mostly Globigerinae. Echinoderms are represented by several plates and one or two echinoid spines, together with a single holothurian spicule. In addition there are some indeterminable fragments. Along certain widespread planes all the organisms are represented by hollow moulds. ~The matrix of the rock, which in each section is formed of lime- stone containing dozens of quartz granules, is more or less pene- trated by phosphate of lime and glauconite; it becomes more phos- phatic as one approaches the contact of the two rocks. At the point (b) and along the suture-line (c) it is wholly phosphate and glauconite. The phosphate of lime, tinted in yellow of differing shades, is clearly cryptocrystalline. In the sections showing the onset of de- calcification it is crystalline in the form of short scales bordering the cavities or arranged in very thin zones. Hxcept for a short distance from the line of contact (c) carbonate of lime is abundant, even in the most phosphatic portions, in the form of a very large number of small corroded fragments which are identifiable. The phenomenon of epigenesis is clearly marked. Everywhere glauconite accompanies the phosphate of lime as very scattered grains, and chiefly as pigment. Its partial alteration is con- comitant with the yellow coloration of the deposit. The fact that this does not extend to the glauconite of the upper part of the nodule, which is of more recent formation, shows that the decomposition of this mineral dates from far back and that it is of submarine origin. VOL. XXXI, PART I. «414 od 126 - Annals of the South African Museum. The dark zone (c), which marks the upper limit of the rock con- taining Bryozoa and Foraminifera, is characterised not only by the complete disappearance of carbonate of lime, but also by a develop- ment of pigmented glauconite on a scale which is unique in my experi- ence. The phosphate of lime and glauconite are intimately associated there. The latter mineral is concentrated in plates, pockets, and bands that are arranged very irregularly. Some clear areas are a kind of discontinuous network formed of anastomosing veinlets, and have no parallel in any known rock. In addition, glauconite metasomatises all the organisms in the zone, such as Foraminifera, Bryozoa, and Echinoderms, showing absolute freshness everywhere. It is here that the most beautiful examples of Foraminifera converted into glauconite can be seen. The same line of suture (c) can be clearly traced from one end to the other. It is characterised sometimes by a concentration of coloured glauconite whose density may or may not decrease lower down, and sometimes by a concentration of phosphate of lime marked by a particular coloration and transparency. There is no encroachment of phosphatised limestone beyond this line. Finally, its shape is marked by a number of irregularities which give it the characteristic shape of a line of corrosion. (B) Beyond this line an entirely different rock appears. This is coarse-grained, crowded with glauconite, much more quartzitic, lacking invertebrate remains, and has a phosphate matrix. The contrast is extremely striking. Here is no trace of pigmentary glauconite or of epigenesis. The mineral forms a crowd of very pure grains of the ordinary type, measuring on the average 1 mm. in diameter, and may preponderate in a good many of the sections (Pl. XX XV, figs. 2 and 3). In all characters this glauconite is identical with that of the incrustations and of the nodules previously described. In addition to the granules which represent quartz in the preceding rock, there are here large grains measuring 0-6 mm. and more (Pl. XX XV, figs. 2 and 3, q), of such a kind that, in one and the same section, the diameter of the grains may vary in the ratio of 1: 30 and even more. This is quite abnormal and necessitates a strongly agitated environ- ment. Organisms are represented only by one or two fragments of bony tissue, with very undulose extinction. The matrix, composed of phosphate of lime which is grey, pale yellow, or deep yellow, is isotropic, or cryptocrystalline, between crossed nicols and nearly always free from fine calcareous inclusions. In its characters this rock might be taken for certain pebbles of the greensands with Douvilleiceras mammillatum from the east of the Paris Basin. The.Phosphatic Nodules of the Agulhas Bank. 127 It can be noted here that this constitution is a replica of that of the nodules formed exclusively of phosphatised greensand, described with the six types already analysed (p. 112). This specimen yields good evidence of a great disturbance which has affected the sea-bottom at some time and produced an eroded surface. There are two fundamentally different rocks in contact, one arising from greensands and the other from a bryozoan fora- miniferal limestone, two deposits formed in different surroundings. In parenthesis, the term “concretion” does not appear the correct one to designate a rock detached from the junction of two different - formations and shaped into a nodule or, more properly, a pebble. Corrosion of Nodules.—The intrusion of glauconite of the greensands into certain nodules, discussed above, has arisen from true corrosion phenomena which, as a final product, have resulted in the formation of scoriaceous material. The cavities thus created are to-day either empty or have been more or less filled with glauconitic phosphate. The material at my disposal is of too restricted a nature to elucidate the problem of the corrosion of pebbles. Some nodules have escaped corrosion, but the proportion is unknown; this is the case for those which have been produced from greensand. Some have been affected over their whole surface, resulting in a scoriaceous appearance. Others have been worn on one side only; e.g. a brownish, very angular nodule resting on the bottom on a flat face has been corroded and finely perforated on its upper side only. The conclusion is reached that corrosion has affected material that was already detached from the parent rock, 7.e. the nodules themselves, in which the phenomenon has nothing in common with the grooving shown by one specimen. It cannot be questioned that all the corrosion is anterior to the penetration of the greensands into the cavities created by it; but it is no less certain that the wearing away was continued after the change in environment which caused the greensands to appear. As witness to this, there is the nodule formed of two distinct rock-types, of which the upper part, characterised by an old phos- phatic greensand (Pl. XX XIII, fig. 1), is itself corroded. At the moment this phenomenon of submarine corrosion requires elucidation, especially as to the nature of the environmental conditions that are favourable to it. Possibly an examination of a large number of specimens would throw some light upon it by supplying an answer to one important question—Does a single station provide, at one and the same time, corroded pebbles and pebbles that have escaped corrosion ? 128 Annals of the South African Museum. Nomenclature of the Phosphatic Material of the Agulhas Bank.—1 have purposely refrained from using the term “concretion,” since it was first necessary to obtain discriminatory knowledge of the mode of origin of the material. In its general acceptation the term nodule does not imply any given mode of origin, while that of concretion immediately suggests concentration around one or several points of attraction. Such is the idea that is held—often wrongly— of the mode of formation of flint. Among this material, the only true concretions are the phosphatic greensand nodules. All the rest are referred to a phenomenon of substitution of phosphate of lime for carbonate of lime, a phenomenon which does not obey the same laws as the former and which, moreover, is quite independent of their morphology. In short, all the phosphatic materials of the Agulhas Bank, without exception, are nodules, and an unknown fraction of these nodules falls within the category of concretions. INTERPRETATION OF THE FACTS, AND CONCLUSIONS. In describing the material dredged to the south of the Cape of Good Hope which furnished almost the only basis for their study of the mode of formation of phosphatic concretions, J. Murray and A. F. Renard laid emphasis on the important fact that the minerals and organisms of the nodules on the one hand, and of the sediments in which they are included on the other hand, are identical, both in the case of the concretions of the greensands and those of the Globigerina mud. This is to say that all the present-day phosphates, and those of South Africa in particular, have been formed in situ. In this connection there is no doubt that J. Murray and A. F. Renard studied material, formed in place, that perfectly displayed the characters of the greensands and of the Globigerina ooze. L. W. Collet agreed with their opinion, writing ‘‘ Les concrétions phosphatées sont en quelque sorte ’image du fond dans le quel on les trouve.’ * There are certain facts cited by L. W. Collet in his deserip- tion of the phosphatic concretions of the Agulhas Bank which support this contention. One is the observation that in the concretions “la quantité de coquilles de Globigérines augmente avec la profondeur de la mer.” It is pertinent here to state that the concretions studied by J. Murray * BE. W. Collet, Les dépots marins, 1908, p. 200. + L. W. Collet, Proc. Roy. Soc. Edin., vol. xxv, 1905, p. 879. | = The Phosphatic Nodules of the Agulhas Bank. 129 and A. F. Renard are not the same as the large number of specimens obtained by the Cape Government. Having stated this, it appears to me that a conclusion contrary to that given above is supported by some very convincing evidence and by arguments drawn from the interesting monograph of L. W. Collet on the phosphatic concretions of the Agulhas Bank. Station 12 (421 m.) of the Agulhas Bank yielded L. W. Collet some nodules rich in Globigerinae, which were frequently whole, and some which “se composent uniquement de grains de glauconie cimentés par le phosphate.” * Similarly, Station 7 (146-238 m.) } yielded some very ferruginous nodules of a yellow-brown colour and others “‘ressemblant extérieurement 4 de la ponce.” { The first, according to G. W. Lee, - are free from detritic minerals, and the second contain abundant quartz, etc. Lee concludes, logically enough, that “il y a une différence essentielle entre ces deux espéces de concrétions.” § My own observations show that the brownish nodules, dredged at one and the same place, are of very differing characters (p. 112); one type is a consolidated globigerinal mud, the other is formed from a deposit of large benthic Foraminifera with fragments of bryozoan colonies. These two very different rocks cannot have been formed at the same depth. The existence of cavernous nodules with Globigerinae carrying in- clusions from which these tests are excluded, as well as that of the nodule from Station 10 formed of two distinct rock-types, are further evidence on the same lines. From this assemblage of facts the following conclusions can be drawn :— (1) All the nodules used in this study and, without doubt, a portion of those studied by L. W. Collet have, at some period of their history, become remanié. We shall see later how it is possible to reconcile the presence of nodules formed on the spot with the existence of transported pebbles. (2) To this idea of submarine remaniement can be added another, also formulated for the first time, which concerns the present-day sea-bottom and can be called a submarine erosion. The composite nodule, described above, furnishes proof that the bottom of the sea has suffered denudation, and that this erosive action coincided with * L. W. Collet, op. cit., p. 876. j{ This Station 7 is not the same as that given in the table on p. 106. t L. W. Collet, op. cit., p. 877. § L. W. Collet, op. cit., p. 878. 130 Annals of the South African Museum. a radical change in the nature of the surroundings as well as in that of the sediments. The phenomena of erosion have acted on the cal- careous substratum in the same way as the changes of level of the supra- Cretaceous epoch on the chalky muds of the Paris Basin ; hardening of the bottom, impregnation by phosphate and by glauconite, etc., are present in both cases. It remains to determine the succession of events that can explain this deposit, a problem that the preceding discussion seems to have obscured, and which necessitates a departure from the domain of facts for a moment. In my opinion, the problem is essentially dependent on the following important fact: All the remanié material is derived from deposits that were formed at a greater depth than that at which they were dredged. - This fundamental observation leads to the conclusion that the region in question has been the seat of important bathymetrical changes. Without justifying their opinion, J. Murray and E. Philippi have ad- vanced the idea of an elevation of the sea-floor.* Unless I am in error, everything points to an uplift of great amplitude, which has raised up lumestones of varied types, and even Globigerina oozes. This change of level, which is to be measured in hundreds of metres, has terminated the deposition of limestones, has inaugurated the formation of green- sands, and has induced the phenomena of erosion, of disintegration, transportation, and re-cementation of the material. Phosphatisation enters into this complex of phenomena, of which we find an equivalent in the distant past, always in close connection with the manifestation of great changes. It is, too, from this period that the phenomenon of corrosion, which has shaped a portion of the nodules, dates. To be more precise, it must be added that the phenomenon is less simple than it appears, for it is complicated both by a delay in movement of the material and by a resumption or continuance of the corrosive action after the change in level, as is shown by the nodule of mixed composition. Nothing is easier to understand than that this change of level, with its consequences, can explain the duality of origin of phosphatic rocks. At the same time as the pre-existing materials have been metasomatised there were formed true concretions at the expense of the greensands that were in course of formation. I shall show later, in examining the more ancient phosphates, that such an association is not unique in the sedimentary series. Although I believe this to be the true succession of events, it must * J. Murray and E. Philippi, op. cit., p. 186. The Phosphatic Nodules of the Agulhas Bank. 131 be confessed that it is not wholly clear. One episode of the history in particular remains a puzzle. How did the nodules, or the limestone fragments from which they were formed, become concentrated in one area? This question is all the more important, seeing that rocks formed under differing bathymetric conditions are found to-day gathered together in one place. Without pretending to give to this problem the answer it deserves, it may be that the solution should be sought in the following direction :— One finds, on the deep-sea side of the Agulhas Bank, very steep slopes which sink rapidly to depths of 3000 metres and more (see Map, Pl. XXXII). Actually, all the concretions dredged by the Challenger, and most of those supplied to L. W. Collet by the Cape Government, came not from the Agulhas Bank proper but from the external slopes. Of the nine stations which supplied the nodules sent to me one only is on the actual Agulhas Bank. It is from this that were obtained the brownish nodules, one of which corresponded to a Globigerina mud, and the other to a calcareous clay enclosing numerous benthic Foraminifera and Bryozoa. To clarify the idea even a little, it can be presumed that the different types of limestone from which the phosphatic nodules have been derived outcrop in horizontal beds of the slopes of the plateau which is covered with a mantle of green- sands, which is probably very thin. Whatever may be the succession of the deposits it is natural that, following the change of level, an agglomeration will be produced in the talus on the outside of the ridge. Further, it must be supposed that this surface, profoundly eroded during uplift, would enable the dredge to explore different horizons of the complex forming the Agulhas Bank. Whether this view is acceptable or not, it is indisputable that the phosphatic nodules of the Agulhas Bank are not dependent on present- day phenomena, a conclusion contrary to the opinions of J, Murray, A. F. Renard, and L. W. Collet. Possibly the conclusion, stated thus, is too absolute. It includes all the nodules which do not result from the consolidation of green- sands. It applies equally to the nodule formed of two rocks fused together, one of which was derived from phosphatised greensand. Actually, one can show that the nodules formed from metasomatised limestones are not of modern origin, but it is not possible to show this for the concretions that are exclusively built up in consolidated greensands. It is not an objection that the remaniement of the fragment detached from the contact of the greensands and the 132 Annals of the South African Museum. underlying calcareous beds leads, naturally, to the supposition that those nodules which were produced solely from the greensands date from the same period as the others. If I do not state my conclusions as absolute, it is because I can picture an hypothesis which involves a prolonged period for the genesis of the greensand nodules. It is not possible to fix the age of the phosphatic materials of the Agulhas Bank from a knowledge of their mode of origin. In this connection it must be recalled that the dredgings have not furnished only phosphatic nodules. L. W. Collet taught that the nodules of the Agulhas Bank are “accompagnés de nombreux restes d’animaux,”’ “consistant en dents de poissons et de squales, os tympaniques de cetacés.”” * We know also from him that, at a depth of 421 metres (Station Vasco da Gama),{ there were dredged, in addition to the concretions, “un grand nombre de fossiles de Lamellibranches et de Brachiopodes, ainsi que des coquilles mortes,”’ and living representa- tives of several groups “qui témoignent d’une abondance de vie” ¢ in this region. Concerning the fossils, L. W. Collet notes that one finds “tous les modes de passage depuis la coquille morte, blanche, a celle complétement transformée en phosphate de couleur brillante brune.”’ § The same author records the presence of phosphate of lime in the dead shells. Unfortunately, none of the fossils are identifiable. Now that the problem has been placed on a new basis, the shells ought to furnish material for two interesting observations: (1) If they are corroded, they must date at least from the period of dis- turbance which determined a radical change in the nature of the sediments. (2) If showing no trace of corrosion, and if they are of the same size and thickness as the living shells or the unfossilised dead ones, then the hypothesis of remaniement (derivation) must be dis- carded, and cannot be used in fixing the age of the deposit. Sir John Murray and E. Philippi, whose brief account of the phosphates of the Agulhas Bank appeared three years after the publication of L. W. Collet’s memoir, wrote: “Es scheint aber sicher zu sein, dass diese Fauna nicht alter als jungtertiar ist. . .”’ || without furnishing any proof. The presence of a tooth of Carcharodon, probably belonging to the * L. W. Collet, Les dépdts marins, 1908, p. 207. + This station has furnished nodules rich in Globigerinae, as well as others that are solely composed of grains of glauconite cemented by phosphate of lime. t L. W. Collet, Proc. Roy. Soc. Edin., vol. xxv, 1905, p. 875. § L. W. Collet, zbid., p. 876. || Sir John Murray and Professor E. Philippi, op. czt., p. 186. The Phosphatic Nodules of the Agulhas Bank. 133 living C. Rondeletw, does not settle the question. Itis useful to note that Dr. A. W. Rogers, formerly Director of the Geological Survey of the Union, recognised among the material obtained by Dr. Gilchrist a shell of Aturia partially enclosed in the phosphate “which,” he wrote to me, “reminded me of the remaniés fossils in the Cambridge Greensand.” This observation would date the deposit in a singular manner, if the hypothesis of derivation (remaniement) could be dis- carded with certainty, for the most recent known forms of Aturia appear in the Miocene. Butif at least part of the material is derived, then any fixed conclusion is impossible. Whatever the age of the phosphatic deposits of the Agulhas Bank, it is indisputable that the major phosphatisation which produced them began before the present epoch and after the disturbance which put an end to the deposition of limestones and substituted greensands for them. The change of level is such that it is probably necessary to place it before the Pleistocene period. The fact, observed by J. W. Collet, that dead shells belonging to living species are encrusted with phosphate of lime does not lead to the conclusion that the phosphatisation has been prolonged until quite recent times. The absence of incrustations of phosphate on _ living shells rather leads to the supposition that deposition has not continued until now. Perhaps the problem of the age of the phosphatic nodules could be approached from an entirely different standpoint. Has the change in level shown by these submarine sediments, not far from the coasts, left traces on the dry land? If so, one can foresee the possibility of solving the problem. This aspect of it must be dealt with by those geologists who have such a wide knowledge of that portion of South Africa. From all that has been learned, it results that the submarine deposit of the Agulhas Bank constitutes in some way the last phase of an wmportant series of phosphatic deposits which are arranged almost from one end to the other of geologic tume. From this it follows that the deposit in question, that must be considered whenever one seeks to explain the origin of sedimentary phosphates, is far from presenting us with a clear picture of all the factors which play a part in their formation. As far as is known at present, it shows no essential difference from the older phosphatic formations. It is no exaggeration to say that the preceding study has been unexpectedly fruitful and full of novelty. Apart from the ideas of great changes of level, and of submarine erosion and derivation 134 ._ Annals of the South African Museum. which have clearly emerged from it, it has shown the existence of several generations of glauconite and of instructive conditions of fossilisation (calcified Sponge spicules and numerous Foraminifera with glauconitic tests). Moreover, it has made known phenomena, developed in a submarine environment, which ordinarily belong to continental areas, such as decomposition of glauconite, the sometimes widespread crystallisation of carbonate of lime which has escaped phosphatisation, the genesis of true veinlets of calcite, decalcification observed in the composite nodule, and corrosion of numerous nodules. With regard to the problem of genesis, this study has stressed the extreme rarity or absence of microscopic remains of fishes, organisms whose role is usually held to be preponderant in the genesis of phos- phate nodules. Finally, from the chemical point of view, it has shown the presence of fluorine, and thus created a close connection between the phosphates of the Agulhas Bank and those of earlier times. Thanks to the material placed at my disposal, this description has assumed the character of a study of submarine geology applied, not to an interior sea of reduced depth like the English Channel, but to one of sub-oceanic depths, with both pelagic and terrigenous sediments. From every aspect it is intimately bound up with the geology of emerged formations. If one separates phosphatic nodules of the Agulhas Bank from those of the present day for the reason that they are older—with the reservation made above in favour of the pebbles that are solely derived from the greensands—the distribution of phosphates of modern origin becomes very restricted. Hitherto, nearly all the evidence of the properties of modern phosphates has been furnished by the nodules of the Agulhas Bank. If these latter are considered as of an earlier origin, all that remains within the category of phos- phates of modern origin are the nodules dredged by the Blake and the Challenger in areas not within the Agulhas Bank; and the conclusions reached for the Agulhas Bank nodules are of such a nature as to suggest doubts as to the true age of the others. Moreover, the authors who have recorded the existence of phos- phatic nodules have scarcely given all the necessary information required. It is impossible to find statements as to their degree of frequency; nor do we know whether we are dealing with petro- graphic curiosities or with true pebble-beds—a matter of considerable importance. If the former—which is more probable—the existence of a few nodules throws no more light on the genesis of sedimentary The Phosphatic Nodules of the Agulhas Bank. 135 phosphates than does a study of the scattered nodules in the White Chalk of the Paris Basin clarify the problem of the origin of the older phosphates. Finally, almost nothing is known of the deposits of nodules other than that of the Agulhas Bank. What is known can only justify doubts as to their age. The single nodule, described by J. Murray, which was dredged by the Blake,* came from a hard bottom, 1.e. from unusual conditions of sedimentation. The question arises, Is not this hard bottom evidence of a change of level with all its conse- quences? That such a question can be asked is enough to show that the value of the nodules dredged by the Blake is not known. Contrary to expectation, the present-day formations throw but little light on the vast problem of the genesis of sedimentary phosphates. To find a solution it is necessary to pin our faith on the past. * J. Murray, Three Cruises of the Blake, vol. i, 1888, p. 276, fig. 189. 136 Annals of the South African Museum. EXPLANATION OF PLATES. Prate XXXII. Map of Agulhas Bank area. PuaTtE XXXII. Morphology of nodules, nat. size. Fig. 1. Rounded nodule, formed by two different rocks (see Pl. XXXYV, fig. 2). Fig. 2. Scoriaceous nodule, Stat. 7, Cape Point, 560 fath. (1024 m.). Fig. 3. Very irregular nodule, Stat. 6, Cape Point, 315-400 fath. (576-732 m.). PLATE XXXIV. Thin sections of nodules. Fig. 1. Calc-phosphate nodule, very quartzitic. Stat. 8, Cape Point, 300 fath. (549 m.). gq, quartz; g, glauconite. (x 50.) Fig. 2. Calc-phosphate quartzitic nodule, rich in Globigerinae. Stat. 3, Vasco da Gama Peak, 230 fath. (421 m.). 9g, quartz; g, glauconite. (x 45.) Fig. 3. Phosphatised and ferruginous nodule, full of Globigerinae. Stat. 5, Cape St. Blaize, 105 fath. (192 m.). (x 50.) Fig. 4. Phosphatised and ferruginous nodule, with large benthic Foraminifera. Stat. 5, Cape St. Blaize, 105 fath. (192 m.). (x 50.) PLatTE XXXV. Thin sections of nodules. Fig. 1. Glauconitic layer enclosing grains of glauconite (g) in the scoriaceous nodule of Pl. XXXITI, fig. 2. Stat. 7, Cape Point, 560 fath. (1024 m.). ( x 60.) Fig. 2. Section of nodule formed of two rocks, showing line of contact. Stat. 10, Vasco da Gama, 166 fath. (304 m.). a, lower yellowish portion; 6, band strongly impregnated with glauconite and phosphate, in neighbourhood of contact; c, line of contact; q, quartz of upper portion. (x 45.) Fig. 3. Upper part of same nodule, showing the constitution of a phosphatised greensand nodule. gq, quartz; g, glauconite. (x 50.) XX XII. Plate Ann. S. Afr. Mus., Vol. XX XI. -cL6L “SpOOqYS UBILILY Qqynoy 10 9G ize 8 vE avg 03] HIAGVZITA LXOd 96 VG ‘SUIOYIRY UT SAMOQUOD = *(SJABY AQ[RaTUpy pure Aydeasooyy yeorsdy “WOT, Up uo peseq) YNVA SVHTOOV AHL JO NOIWY GG 0G \P Cog vyuvpjvs aS 81 Ann. 8. Afr. Mus., Vol. XXXT. Plate: XX-XTE, Neill & Co., Lid. a : i “ ; pa iba. a ae Ltd & Co. 9 Neill Plate XXXV. ING, Bis 9 Fie ill & Co., Lid. Ne Ann. S. Afr. Mus., Vol. XX XI. att gt ees ante e Wine. Ie > A, i a ae dz d= ‘ Via a y ‘ = ‘ib ae > 2 ve at ka oils ee ihe ) Fz. t oe wd ’ ' — 7] » ~~ a 5 vA 3 . ' i . ’ ? ’ ’ ‘ ol f 4 , j pH ¥ ‘ n - / - eu . ‘ . ‘ - 7 ‘ rs ~ : = x . Ss > a 08 t 4 ” he oi ‘ aa * PARTS OF THE ANNALS PREVIOUSLY’ ISSUED— Vol. I.—Part 1, out of print; Part 2, out of print; ‘Part 3, out of eae se ah Vol. I1.—Part 1, out of print ; Part 2, 5/6; Part 3, outof print; Part4,3/-; Part 5, 1/6; Part 6, 3/—; Part 7, 1/6; Part 8, 3/-; Part 9, J ‘Part 10, 7/-; Part 11, 3/-; Index, Title, etc., 1/6. Vol. I1l.—Part 1, out of print; Part 2,.1/6; Part 3, 5/6; Part 4, 3/-; - Part 5, 5/6; Part 6, 7/—; Part 7, 1/6; Pate 8, 3/-3 Part 9, 1/63. Index, Title, etc:, 1/6. Vol. IV (containing Palaeontological papers published in ‘conjunction with the | Geological Survey).— Part 1, 11/-; Part 2, 7/-; Part 3, 4/6; Part 4, 4/6; Part 5, 2/6; Part 6, 4/6; Part 7, 14/—; Part 8, 8/-. ~— Vol. V.—Part 1, 4/6; Part 2, 8/6; Part 3, 2/6; Part 4, 1/6; Part 5 ie a Part 6, 5/-; Part 7, 3/-; Part 8, 4/6; Part 9, 4/6; Index. ~ = ae Title, etc., 1/6. Vol. VI.—Part 1, 13/6; Part 2, 4,6; Part 3, 3/6; Part 4, 30/-; Index, ea Title, etc., 1/6. Vol. VII (containing Palaeontological papers published in conjunction with the aie Geological Survey).— Part 1, 3/-; Part 2, out of print; Part 3, 5/-; Part 4, Ries <5 peee | Part 5, 5/6; Part 6, 1/6; Index, Title, etc., 1/6. Vol. VilI.—Complete, out of print. Index, Title, etc., 71 /6. Vol. 1X.—Part 1, 4/6; Part 2, 5/6; Part 3, 10/5 Part 4, 6/6; Part 6, 3/6; Part 6, 11/-; Part 7, 9/-; Tae Title, etce., 1/6. Vol. X.—Part I, 3/—; Part 2, 2/6;-Part 3, 2/—; Part 4,3/-; Part 5, 20/-; Part 6, 3/—; Part 7, 10/-; Part 8, 2/6; Part 9, 5/—; Part 10, 2/6; eee ; | Part 11, 20/—; Part 12, 7/-. Complete. Title, ete., contained in, Part 11. Vol. XI.—Part 1, 3/6; Part 2, 2/—; Part 3, 13/6; Part 4, 1/6; Part B, 1-3 Part6, 11/-; fadex, Title, etc., and Plate II, 3/-. Vol. XII (containing Palaeontological pavers published in conjunction with the 3 Geological Survey).— Part 1, 15/6; Part 2, 3/6 ; Part 8, 4/6; Part-4, 3/-; Part Part 6, 6/—; Part 7, 20/-; Part 8, 20/-; Index, Title, etc., 1/6. Vol. XIII.—Part 1, 6/—; Part 2, 2/6; Part 3, 3/-; Part 4, 8/6; Pate 5, 1/6; Part 6, 5/-; Part 7, 30/—; Part 8, 1/-; Index, Title, ete., 1/6. = Vol. XIV.—Part 1,8/6; Part 2, 8/-; Part 3, 6/-; Part 4, 21/-; Part 5, 5/-; Part 6, 9/-; Index, Title, ete., 1/6. pe Vol. XV.—Part 1, 17/-; Part 2,17/-; Part 3, 14/-; Part 4, 12/6; Part 5, 5/6; Part 6, 3/6; Index, Title, etc., 1/6. . Vol. XVI.—Part 1, 30/6; Part 2, 4/-: Dart 3 (with Title, Index, etc.), 25/-. Vol. XVII.—Part 1, 12/—; Part 2,9/6; Part 3,3/-; Part 4,17/-; Part 5, 17/—; Part 6, 2/6; Index, Title, etc., 1/6. Vol. XVIII.—Part 1, 20/-; Part 2, 7/6; Part 3, 30/-: Part 4, 12/6; Index, ie Title, ete., 1/6. Vol. XIX. Part 1, 29/_: Part 2, 17/6; Part 3, 11/-;, Part 4, 5/6; Index, Title, etc., 1/6. Vol. XX.—Part 1, 8/6; Part. 2, 12/6; Part 3, 4/-; Part 4, 10/—; Part 5, 4/5 Part 6 (with Title, etc.), 4/6. Vol. XXI.—Part 1, 25/-; Part 2 (with Title, etc.), 30/-. Vol. XXII (containing Palaeontologieal papers published in novsnnetion with the 5 Geological Survey).— x Part 1, 20/-; Part 2, 10/-; Part 3 (with Title, etc.), 3/6. Vol. XXIII.—Part 1, 12/6; Part 2, 8/-; Part 3 (with Index, Title, etc. ), 9/-. Vol. XXIV.—Part 1, 10/-. Vol. XX V.—Part 1, 12/6; Part 2. 10/-; Part 3 (with Index, Title, ete. ) ee Vol. XXVI Complete, 25/-. ; Vol. XXVII.—Complete, 25/-. Vol. XXVIII (containing Palaeontological papers publishes in conjunetion y with | the Geological Survey).— Part 1, 10/-; Part 2, 17/6; Part 3;.6/—: Part 4 (with Index, Title, etc.), 10/-. Vol. XXTX.—Part 1, 20/—-; Part 2 (with Index, Title, ete.), 20/-. Vol. XXX.—Part 1, 15/6; Part 2, 20/-; Part ee Sj. Vol. XXXI.—Part 1, 20/-. The Annals of the South African Museum will be issued at pie intervale, as matter for publication ts available. Copies may be obtained. oa Messrs. WHELDON & WESLEY, Lap., 2, 3, and 4 Arruur Street, New Ox#ouD STREET, oman, W.C. 2: or, The LIBRARIAN, Sovuru Arrican Musrum, Carr Town. in tee D res 5 Ne dled \atialiole ASG a ade ae , Be aay My Siasnedineing ays : 5 em ae is bia one Sat a ey a La aes 2 sii a Maas). wie rt ach gabe diem Ee ae) | Mes MMI TRY Wy ge Ria tis A OF THE VOLUME AKA. ‘DESCRIPTIONS o OF THE PALAEONTOLOGICAL MATERIAL “COLLECTED BY THE SOUTH AFRICAN Ss cn ee ee ee &. SEP 14 34 5. me Contribution to ‘the Morphology of the Gorgonopsia. By ee _ Lieuwe D. Boonstra, D.Sc., Palaeontologist to the South = ee - - African Museum. oe Et Text- -figures. ) 6. ‘Additions to our aencntletlie - the South African Soidinpsie: : preserved in the British Museum (Natural History). By a Lizuwe D. Boonstra, D.Sc., Palaeontologist to the South African Museum and Queen Victoria Scholar of the University | of cote (With 18 oe tenes) 7 A Cais to the Morphology of the Mammal-like Reptiles et of the Suborder Therocephaha. By LinuwE D. Boonstra, BES. Palaeontologist ‘to the South African Museum and Queen Victoria Scholar of the iaaees of Stellenbosch. ; (With 35 5 Text eee ) ISSUED JULY 1934. PRICE 13s, 6d. ny oor ; PRINTED FOR THE : TRUSTEES OF THE SOUTH AFRICAN MUSEUM BY NEILL AND CO} LTD Gy 212 CAUSEWAYSIDE, EDINBURGH. * zh =e, Gsm) 5. A Contribution to the Morphology of the Gorgonopsia.—By Lizuwe D. BoonstrRA, D.Sc., Palaeontologist to the South African Museum. (With 11 Text-figures.) ALTHOUGH the first Gorgonopsian was described as long ago as 1860, and notwithstanding the fact that since then twenty-seven reputed genera and thirty-six species have been recorded from the South African Karroo Beds, our knowledge of these animals was until quite recently practically confined to their cranial morphology. The reason is not difficult to find. The Gorgonopsia were agile beasts of prey that lived on the higher and drier parts of the great Karroo Basin during Upper Permian and Lower Triassic times, so that on death their remains, in order to be entombed in the sediments, had to be transported to the lower-lying areas. During transportation the comparatively slender and more fragile bones of the postcranial skeleton were destroyed, whereas the more compact skull was more often preserved. Hitherto very few parts of the postcranial skeleton have been found in association with skulls. Owen, in 1876, described at length and gave good figures of the humerus of Cynodraco. Nothing more was added to our knowledge of the postcranial skeleton until Broom and Haughton, in 1913, jointly described and gave unsatisfactory photographs of the cervical vertebrae, pectoral girdle, and fore-limb of Aeluro- gnathus tigriceps. In a later paper, 1913, Broom gave a fuller description, with a figure, of the manus of Aelurognathus, and in the same year a short account of the shoulder-girdle of Scylacops. In 1921 Watson figured, without describing, the anterior cervicals of Scymnognathus whaitst. Then, in 1927, Pravoslavlev published two important papers on the Gorgonopsia of the North Russian Karroo Beds. Some of these forms had been photographed by Amalitzky, but had never been described. Unfortunately Pravo- slavlev’s work is in Russian, so that only the text-figures and photo- graphs are intelligible to most non-Russian scientific workers. It would appear, however, that in two genera—Inostrancemia and Amalitzkia—a practically complete vertebral column, pectoral and pelvic girdles, hind- and fore-limbs lacking only the feet, are known. In 1929 Haughton had the good fortune to obtain from the Rev. VOL. XXXI, PART 2. 15 138 Annals of the South African Museum. J. H. Whaits a fairly well-preserved shoulder girdle and fore-foot of Lycaenodontoides. ‘These were fully described and figured. In a recent (1930) fairly comprehensive account of the Gor- gonopsia, Broom figured and described an imperfect vertebral column, pectoral and pelvic girdles, and the practically complete fore- and hind-limbs of Lycaenops. The known parts of Aelurognathus and Scylacops were also incorporated. During three collecting trips in the Southern Karroo I had the good fortune to collect a number of Gorgonopsians, which had considerable portions of the postcranial skeleton preserved. This material, augmented by the other previously described specimens preserved in the South African Museum, forms the basis of this communication. The collection consists of :— (a) S.A.M. Cat. No. 8950. Klein Koedoeskop, Beaufort West. Middle Tapinocephalus zone. A practically complete vertebral column, complete pectoral girdle, imperfect pelvic girdle, and fore- and hind-limb, lacking only some of the digits, are preserved. Unfortunately this specimen is preserved in a very intractable calcareous nodule, so that it has been very difficult to free the various skeletal elements from the matrix without more or less damaging the bone. The skull of this specimen has been described by Haughton under the name, Hipposaurus boonstrat. Collector, L. D. Boonstra. (6) S.A.M. Cat. No. 2342. Dunedin, Beaufort West. Low Cistecephalus zone. This specimen consists of a skull, seven cervical vertebrae, a com- plete though somewhat crushed pectoral girdle, a good humerus and radius, an imperfect ulna and a nearly complete fore-foot. This is the type specimen of Aelurognathus tigriceps described by Broom and Haughton. Collector, 8. H. Haughton. (c) S.A.M. Cat. No. 2348. Dunedin, Beaufort West. Low Cistecephalus zone. Associated with the skull there is an incomplete and crushed pectoral girdle, a good sternum and humerus, and some cervical vertebrae. This is the type of Scylacops capensis described by Broom. Collector, 8. H. Haughton. (d) S.A.M. Cat. No. 3329. Oudeberg, Graaff-Reinet. Cvste- cephalus zone. Associated with a snout, an imperfect pectoral girdle and manus ‘ are preserved. Thisis Haughton’s type, Lycaenodontoides bathyrhinus. Collector, J. H. Whaits. A Contribution to the Morphology of the Gorgonopsia. 139 (ec) S.A.M. Cat. No. 9344. Klein Bloemhof, Richmond. Ciste- cephalus zone. This specimen consists of a crushed skull; a continuous series of seven cervicals and seven dorsals (the rest of the posterior dorsals are lost), then a series commencing with the last presacral, three sacrals and five caudals; an excellently preserved pectoral girdle with a sternum; two good humeri, a good radius and ulna, and the proximal ends of the ulna and radius of the other side; an imperfect carpus; the left ilium and ischium and part of the right ischium; a good femur, tibia, fibula, tarsus, metatarsals, and some phalanges. For this specimen I propose the name—Aelurognathus microdon sp. nov. Collector, L. D. Boonstra. (f) S.A.M. Cat. No. 9345. Ou Plaas, Richmond. Cistecephalus zone. In this specimen there are preserved: a good, though slightly weathered and crushed, skull; a continuous series of seven cervicals and nine dorsals, together with some isolated dorsals and caudals; a nearly complete pectoral girdle with a sternum; two crushed humeri, a good right radius and ulna, and a crushed left radius and ulna; some isolated carpals, metacarpals, and phalanges. For this specimen I propose the name—Arctognathoides breviceps gen. et sp. nov. Collector, L. D. Boonstra. (g) S.A.M. Cat. No. 10188. Dunedin, Beaufort West. Low Cistecephalus zone. There is preserved a somewhat crushed skull with an excellent palate; a humerus, radius, and ulna; a tibia and fibula; and some crushed vertebrae. I have identified this as a specimen of Scylacops capensis, Broom. Collector, L. D. Boonstra. Skull. As the main object of this paper is primarily to note the increase of our knowledge of the postcranial osteology of the Gorgonopsia, no detailed account of the cranial morphology will be advanced. In the case of the three skulls of known species, only those points of structure which have hitherto not been fully understood will be mentioned. In the case of the two new forms the account will necessarily be somewhat fuller. Scylacops capensis (text-fig. 1, a, b, and c). The palate of the type (S.A.M. Cat. No. 3444) has been recon- structed by Broom (Mammal-like Reptiles, fig. 394). The anterior “1ep eI i" eT ; sidan ET rend SR EE eras is eee Height of mentum of ips jaw. 3 ; 2° Toke Length of molar series. 4 . ‘. whey Diastema : P ; : : : : Gwe Length of incisor series. 1G Type, B.M.N.H., R339, Gouph, Beaufort West, Oats Province. Aelurosauroides watsoni, gen. et sp. nov. (RionGs 5.) This specimen has been regarded by both-Lydekker and Broom as Aelurosaurus felinus. It consists of the anterior two-thirds of a small skull, and lacks the lower jaw; the outer surfaces are well exposed, and Watson has had the palate developed. I concur in Watson’s description and figures, except that, in dorsal view, the preparietal is oval and not squarish and, moreover, agree with him that it is generically distinct from Aelurosaurus. In addition to the points enumerated by him, I have been able to determine that the palatines meet in the median line, and thus the pterygoids do not meet the prevomers. South African Gorgonopsia preserved in the British Museum. 191 Aelurosauroides agrees with Aelurosaurus felinus in its dentition, size, and general shape, but differs in that the snout is rounder and less high. As Watson has already pointed out, the two forms differ very markedly in the relative size and shape of the palatal elements, although they agree in the essential morphological features. Watson’s point, that great care must be taken before assigning specimens of Gorgonopsians, which have the same outward appearance, to the same species, without knowing the nature of the palate, therefore deserves full emphasis. The differences in proportion are best understood by referring to the figures; the main are—the shortness and raised nature of the tooth-bearing ridge in Aelurosauroides; the relatively shorter prevomers; the great ventral extent of the ecto- pterygoids, and the more posterior extent of the internal nares. The main morphological features are: Primitive. Laterally directed orbits; snout slightly higher than broad; slight. preorbital depression; frontal apparently forms a large part of the orbital border. Advanced. Snout short and somewhat rounded in section; pterygoid not situated far posteriorly; tooth row slightly curved; sides not straight; deep maxilla; no step in alveolar border. Thus, although retaining some characters which are primitive, Aelurosauroides is more advanced than the forms from the Tapino- cephalus zone, and also than Arctops and Gorgonops. Chief measurements : Length from premaxilla to front of orbit . sy) (oO innrcay Interorbital width . ; i : : Asuka Pa Width of snout : : : : 4 yy Oot ee Height of snout : : : Roce oe Width of pterygoid flanges “ My OOr 5s Length of molar series. ates lige Diastema : : : : : : (Shite Length of incisor series. s : Spee ovals Type, B.M.N.H., R855, Gouph, Cape Province. An Unidentified Specimen. There is in the collection a specimen, labelled 855a, which lacks the external surface; but its palate can be exposed. Until this is done, it is not possible to state whether it must be referred to Aeluro- saurus or to Aelurosaurovdes. 192 Annals of the South African Museum. C. Forms from the Cistecephalus zone. Hight species from this zone are represented, viz. Arctognathus, Arctosuchus, Cerdognathus, Cynariops, Cynodraco, Lycaenodon, Lycosaurus, and Scylacops. Careful preparation has produced two good palates and the lateral surfaces of three excellent brain-cases. Lycaenodon longiceps, Broom. (Fig. 9.) The type-specimen consists of two-thirds of a skull; the whole parietal, occipital, and basicranial regions are lost; very little of the palate is present. A short original description was given by Broom in 1925, and in 1930 two figures were published. The dental formula isi. 5, c. 1,m.6; the very long molar series, measuring 28 mm., is remarkable, as in the Gorgonopsia the tooth row is in general very short. A sagittal fracture reveals some points of structure, which were figured by Broom; the dorsal unperforated keel of the pre- vomers is very strongly developed; the sphenethmoid is a large element, and its shape is probably as figured by Broom; on its antero- lateral surface there is a large foramen for the exit of the second nerve; Broom’s figure clearly shows that the palatines meet in the median line and thus prevent the pterygoids from meeting the prevomers. There can be no doubt that this is the usual Gorgonopsian condition, and that my interpretation of the palatal aspect of Arctops, Gorgonops, Arctognathus, etc., is correct, and that Watson misinterpreted these relations. Broom’s reconstruction of the basicranial portion of the sagittal section, as figured, is based on another specimen, and this will be referred to later. The detailed structure of the anterior part of the palate cannot be determined. Its general shape is, however, of interest; although the skull is a long one, the prevomers appear to be relatively longer, and the palatines do not stretch so far anteriorly, as 1s usually the case—their anterior borders lie well posterior to the plane of the canines; the median groove is deep, but unfortunately the ridges bounding it are not shown. The outer surface, although thoroughly cleaned and etched, shows few sutures, as the skull appears to be that of an old animal with closed sutures. The dorsal surface, with the posterior third recon- structed, is shown in the accompanying figure. The main morphological features are: Primitive. High, fairly narrow snout; extraordinarily long South African Gorgonopsia preserved in the British Museum. 193 straight tooth row, with a step anterior to the canine: preorbital depression; frontals apparently forming a large part of the orbital border; temporal openings apparently small, and parietal region probably very wide. Advanced. Orbits slightly anteriorly directed; sides of snout not upright; deep maxilla. 1 Sie tents nics 5 .: a aS Seo . ss. meet é - eras ‘ ' € 2202222 Qe =< Fie. 9.—Lycaenodon longiceps. Type, B.M.N.H., R5700. Restored dorsal view of the skull. x 4. As far as the skull of Lycaenodon is known, it appears to be very primitive for a form from the base of the Cistecephalus zone. The very long molar series is unique amongst all the known Gorgonopsia, and, if we are correct in considering this to be a primitive character, then Lycaenodon is in this respect the most primitive known Gor- gonopsian. The general contour indicates a very broad parietal region, with very small laterally directed temporal openings, and if such is actually the case, then Lycaenodon would be less advanced in this region than either Arctops or Gorgonops. 194 Annals of the South African Museum. Chief measurements : Length from premaxilla to front or orbit. . 100% mm. Interorbital width : y . : J Aes Width of snout. , . : , ; oD Width across lateral pterygoidal flanges . st) Boas Height of snout . ; : ; d : AG Length of molar series . J : ; : ee Diastema . : : 3 ; : 10, 2 Length of incisor series . f 4 : ; ise Type, B.M.N.H., R5700, Biesjespoort Station, Cape Province. “ Lycaenodon”’ sp. (Fig. 10.) In the collection purchased from Dr. Broom, in October 1932, there is included an isolated brain-case, parts of which were figured in 1930 by Broom, and included in his reconstruction of the sagittal section of Lycaenodon, and it is stated that it probably belongs to this species, presumably on the fact that it was found at the same locality. The difference in the state of preservation precludes the assumption that it belongs to the same individual, and, as this particular part is not preserved in the type, a comparison is not possible. Until further specimens are found, there is no means of proving that the brain-case is that of Lycaenodon; for convenience of reference it will be referred to “‘ Lycaenodon”’ sp. The whole posterior part of the brain-case and basicranial axis is perfectly preserved, and I have, with great care, completely exposed the brain-case on the left side and removed all but the dorsal extremity of the epipterygoid; the anterior border of the pro-otic and the pituitary fossa have also been freed of matrix. This specimen, before adequate development, was the basis of Broom’s (1930) figure showing the relations of the epipterygoid, pterygoid, and the two elements identified as “vomer”’ and ‘“‘basisphenoid”’ respectively, and also of the reconstructed posterior part of his sagittal section of Lycaenodon longiceps. The basisphenoid is a fair-sized bone with edge-like tubera, which underlie the processes sent downwards by the basioccipital, and which form the anterior border to the large irregular fenestra ovalis. From the dorso-anterior corner of the fenestra a slightly digitating suture runs anteriorly, in a plane practically horizontal, to enter the South African Gorgonopsia preserved in the British Museum. 195 lateral border of the pituitary fossa. From the tubera the basi- sphenoid extends, in anterior direction, in the form of a vertical sheet of bone, which laterally carries the horizontal basipterygoid processes; on the ventral surface this produces the median keel; dorsally, it forms the anterior border of the pituitary fossa. Here it is broken off, but, in life, it apparently stretched forwards and upwards as a median septum. In Arctognathus curvimola this tr i | Fic. 10.—‘‘ Lycaenodon.” B.M.N.H., R5746. Lateral view of the left side of the brain-case. x 4. structure is better shown, and this will be discussed further when describing that form. It is evident that this extent of bone cannot all be basisphenoid; the anterior part of the ventral keel, and the antero-dorsal vertical sheet of bone must, for its greater part, be the parasphenoid. The limits of the parasphenoid cannot, however, be determined; the parasphenoid—a membrane bone—appears to have become intimately connected with the basisphenoid, which 1s, of course, preformed in cartilage. As indicated in the figure, the flattened, horizontally situated basipterygoid processes of the basisphenoid are clasped on their dorsal and ventral surfaces by the posterior end of the pterygoid 196 Annals of the South African Museum. ramus, which has split into two thin horizontal plates—a long ventral one and a shorter dorsal one. Immediately behind the end of the basipterygoid process a Vidian foramen pierces the bone. It is on the latero-dorsal edge of the ventral pterygoidal plate that the epi- pterygoid stands; from this lateral edge a thin horizontal flange extends medially, and this acts as a base to the epipterygoid. From its base the epipterygoid stretches dorsally as a thin pillar, which is shown in broken lines in the figure. The pro-otic is a large bone; posteriorly, it is applied to the paroccipital process in the usual manner and forms the antero- dorsal part of the border of the fenestra ovalis; anteriorly, it stretches to the plane of the epipterygoid, which dorsally overlies its extreme antero-dorsal corner. The shape of the pro-otic is best understood from the figure; it forms nearly the whole of the lateral wall of the posterior part of the brain-case. A small foramen for the seventh nerve lies a short distance antero-dorsally to the fenestra ovalis; below it is a shallow depression for the geniculate ganglion; still further antero-dorsally the bone is pierced by a small rounded foramen for the sixth nerve, and under it is a depression; this foramen lies near the edge of a large opening through which the fifth nerve emerged; this opening appears to be a large oval foramen, whose anterior border is formed by the long anterior process of the pro-otic; it is not certain that this opening is bounded antero-dorsally by bone, as a part of the epipterygoid overlies this region; its general shape indicates a large oval foramen completely bounded by the pro-otic; the postero- dorsal border of this foramen is formed by a long process of the pro- otic, similar to the anterior one, which separates it from an equally large foramen lying dorsally to it; this is the large venous foramen. In anterior view, the edges of both pro-otics have been exposed and it is seen that, above the pituitary fossa, processes are sent inwards by both pro-otics, so as to form a bridge of bone over the pituitary fossa; ventrally, the bridge appears to be supported by a spur of the basisphenoid, which here forms the posterior border of the pituitary fossa; dorsally to this bridge, the pro-otics approach each other closely (3 mm.), so that the brain here emerged through a high and narrow slit. On the anterior face of the pro-otic, just above the bridge, is a small foramen facing directly forwards—this is for the exit of the fourth nerve. The sphenethmoidal region is unfortunately not preserved. The basioccipital has a well-developed condyle and has strong downwardly directed processes, on whose ventral surfaces the South African Gorgonopsia preserved in the British Museum. 197 basi-sphenoidal tubera are applied; it forms the postero-ventral border of the fenestra ovalis; on its ventral surface lie the two nutritive foramina in their usual position. The foramen for the tenth nerve is overhung by the exoccipitals in the usual manner. If we are right in our interpretation that in Arctops the anterior part of the pro-otic is not ossified, and that the foramen for the fifth nerve is only represented as a shallow notch, and that in Cynariops the notch is deeper, but still unenclosed, then “‘Lycaenodon’”’ is a more highly developed form, as here the pro-otic has extended forward to surround the foramen for the fifth nerve. The type of Lycaenodon was shown to appear primitive, and it would thus seem that there is an additional reason for regarding this brain-case as not belonging to the genus Lycaenodon. The strong basioccipital, much stronger than in Cynariops, is, however, a primitive character, which may well be found associated with the characters found in the type of Lycaenodon. Until we know more about the relative values of characters styled “primitive” and “advanced,” any further expression of opinion will serve no useful purpose. B.M.N.H., R5746, Biesjespoort Station, Cape Province. Cynariops robustus, Broom. (Figs. 11-14.) The type consists of a good skull, lacking the temporal arches, part of the occiput, and the premaxillaries. After some further preparation, the lateral and dorsal surfaces show the sutures well; the relations of the various elements are best understood by referring to the figures. | The dental formula isi.5,c. 1 ,m.4; on the right side there are three, on the left four, the fourth being small; there is a diastema between the last incisor and the canine, whereas the first molar follows immediately on the canine. Mr. L. E. Parsons, Technical Assistant in the Geological Depart- ment, has prepared the palate and the lateral surface of the brain- case for me, and after some further development and etching, a number of interesting features have become apparent. The structure of the palate is adequately exposed; ventrally, the prevomers have a strong median keel and two lateral flanges on their anterior part; the median palatal groove has deep overhanging ridges, and both the palatine and pterygoid portions bear teeth; the median convoluted interpalatine suture is beautifully shown, and so is the transverse 198 Annals of the South African Museum. suture separating the palatine and pterygoid in ventral view; the latter crosses the ridge at its lowest part and then continues in postero-lateral direction to meet the ectopterygoid suture. The posterior limits of the prevomers are not clearly shown, but probably are as indicated in the figure. Fic. 11.—Cynariops robustus. Type, B.M.N.H., R5743. Dorsal view of the skull. x 3. On the left side, the outer surface of the brain-case has been partially exposed; on this side, the epipterygoid is not preserved, but, on the right side, it is practically complete. The epipterygoid has a long base resting on the quadrate ramus of the pterygoid; it then extends dorsally as a very high, narrow, and flattened pillar to meet the parietal; it thus forms the side wall of the cavum epiptericum anterior to the pro-otic, with the Gasserian ganglion lying immedi- ately median to it; in lateral view, it lies in front of the lateral opening into the pituitary fossa and, further dorsally, overlies the anterior South African Gorgonopsia preserved in the British Museum. 199 | border of the large irregular notch for the fifth cranial nerve. The pro-otic is a fairly large bone, whose shape will be better understood from the accompanying figure than from a verbal description. Ventrally, it meets the basisphenoid in a fairly open suture; antero- La ie mice S B.Oc., B.Sph. Fic. 13.—Cynariops robustus. Type. Ventral view of the skull.. x &. ventrally, it forms the posterior border of the fenestra leading into the pituitary fossa; dorsally, it forms the ventral and posterior borders of the notch for the fifth nerve; the anterior pro-otic process is ap- parently not strongly developed, so that the notch is not deep; further dorsally, the pro-otic forms the ventral border of the large 200 Annals of the South African Museum. P Ovb. Melee wae Deis Fig. 12.—Cynariops robustus. Type. Lateral view of the skull. x 4. Fie. 14.—Cynariops robustus. Type. Lateral view of the left side of the brain-case. x 4. South African Gorgonopsia preserved in the British Museum. 201 venous foramen; but, unfortunately, the extent to which the supra- occipital and interparietal enter the side wall of the brain-case cannot be determined. The fenestra ovalis is seen to lie above and behind the basisphenoidal tubera, and appears to be fairly large. The foramen magnum is extraordinarily large and the basioccipital con- dyle is remarkably small. Unfortunately I have not been able to determine anything as to the presence of a parasphenoidal rostrum or the nature of the sphenethmoidal portion of the brain-case. The most striking feature about the brain-case of Cynariops is its great height and relative narrowness. In this respect it agrees with the fragmentary brain-case of “ Lycaenodon.”’ The main morphological features are: Primitive. Sloping occiput, not cupped; epipterygoid slender; preorbital depression; pro-otic of no great anterior extent. Advanced. Snout as broad as high and rounded; no step in maxillary border; orbits slightly anteriorly directed; pterygoid flanges anteriorly situated; temporal openings somewhat dorsally directed, but not large; parietal apparently enters brain-case; basi- occipital extraordinarily undeveloped; pro-otic larger than in Arctops; frontal forms only small part of orbital border; deep maxilla; short tooth row, but straight; parietal width reduced; paroccipital not massive. From this summary of characters, it is clear that Cynariops is much further advanced than Arctops, Gorgonops, Scymnognathus, and Leptotrachelus. In the nature of its brain-case it appears to occupy a position intermediate to Arctops and that represented by the brain-case referred to “‘Lycaenodon,” and it is in this respect com- parable to Scymnognathus and Leptotrachelus. These two latter genera are, however, in all other respects definitely more primitive. Chief measurements : From premaxilla to basisphenoidal tubera = 1A @ sani. From premaxilla to pineal foramen QUAL: From pineal foramen to edge of occipital plate . Ss From premaxilla to front of orbit . : é Gores. Interorbital width BOM 5: Intertemporal width . ; 2215; Width of snout . : Scans Height of snout . ; : Ae. Width across nierresidal anges BOP x Height of occiput . ; : ; : 45 ,, VOU. Pter. —— — =~. Type, B.M.N.H., 47339. Ventral view of the skull, modified after Watson. x 2. Qu. = quadrate. Qu.Ju. = quadratojugal. Fic. 15.—Arctognathus curvimola. St. =stapes. The basioccipital condyle and the greater part of the paroccipital have been removed by a fracture; their nature is probably as indicated in broken lines in the figure. | South African Gorgonopsia preserved in the British Museum. 205 The basisphenoid is present im toto. Ventro-posteriorly lie its tubera, which are thin and edge-like; they underlie the ventrally directed processes of the basioccipital and diverge in posterior direc- tion, so that the floor of the hinder end of the brain-case (for the cerebellum) is wide, and the fenestrae ovales far apart. Anteriorly, the basisphenoid extends as a thin vertical sheet of bone, which laterally carries the horizontal basipterygoid processes; the ventral part of this sheet forms the posterior part of the keel, which must further anteriorly be formed by the parasphenoid flanked by the pterygoids, but the exact relations of these elements cannot be determined as the fusion appears to be very intimate; the dorsal part of the basisphenoid is anteriorly confluent with the parasphenoid; its free anterior edge lies in the plane of the posterior border of the pituitary fossa and, in life, this edge must have supported the infundibulum; dorsally, the basisphenoid meets the pro-otic along a line running from the fenestra ovalis to the open side of the pituitary fossa. The outer surface of the pro-otic does not show up too well; posteriorly, it was apparently applied to the anterior face of the paroccipital in the usual manner; ventrally, it rests on the sides of the basisphenoid; anteriorly, its extent is not great; this is rather remarkable for a form otherwise showing many advanced characters; the anterior process of the pro-otic does not extend far, so that the exit for the fifth nerve appears to be only a shallow notch, whereas in “‘Lycaenodon”’ it is a large oval foramen; the outer surface of the anterior pro-otic process is deeply hollowed out, and this may have housed the Gasserian ganglion; dorsal to the opening for the fifth nerve, the hinder pro-otic process is also short, so that the venous notch has no extensive osseous ventral border. Dorso-anteriorly to the large fenestra ovalis lies a small foramen for the facial nerve, with a slight hollow under it for the geniculate ganglion. Anteriorly, the pro-otic forms the posterior border of the open side of the pituitary fossa and, posteriorly, forms part of the anterior border of the large and irregular fenestra ovalis. Dorsally, there are indications that both the parietal and inter- parietal sent down flanges, which contributed to the formation of the side-wall; the anterior part of this parietal flange articulated with the posterior part of the sphenethmoid; it was this relation that apparently led Watson astray, since, what he interpreted as a broad epipterygoid on the right side is, in reality, the outer surface of the sphenethmoid; the posterior part of the parietal flange articulated with the ascending epipterygoid. 206 Annals of the South African Museum. The epipterygoid is not preserved, but there is no reason to suppose it to be different from that of the other known Gorgonopsia; standing on the pterygoid, it would ascend as a thin bony rod in a plane lateral to the other bones of the brain-case, and acted as a lateral wall for the thalamencephalon, mid-brain, and Gasserian ganglion. I ft sitar Fic. 16.—Arctognathus curvimola. Type. Dorsal view of the skull. x 4. The parasphenoid, from where it is indistinguishably fused with the basisphenoid, stretches antero-dorsally as a vertical sheet of bone; at its dorsal edge it appears to split, and in this groove the ventral edge of the sphenethmoid rests; anteriorly, the parasphenoid tapers, and here appears as a vertical slip of bone applied to the lateral surface of the sphenethmoidal part of the interorbital septum. The sphenethmoidal part of the brain-case is beautifully shown, and it is now evident that the “sphenethmoid ”’ figured in Scymnognathus ee South African Gorgonopsia preserved in the British Museum. 207 and Leptotrachelus is only a part of that bone. The sphenethmoid in Arctognathus is a very large element; its ventral half represents a part of an ossified interorbital septum and rests on the parasphenoid, and, further forward, on the dorso-median keel of the pterygoid. Anteriorly, the interorbital septum is continued forward, but is separated by an unossified portion from the internasal septum. It is thus only the dorsal portion, which has split into two curved sheets of bone, that is preserved in Scymnognathus and Leptotrachelus. This housed the COP Fie. 17.—Arctognathus curvimola. Type. Lateral view of the left side of the brain-case. x 2. fore-brain. Dorsally, flanges from the frontal and parietal are sent downwards and these, meeting the dorsal edges of the sphenethmoid, complete the side-wall. Ventro-posteriorly, a large rounded foramen pierces the sphenethmoid; it is directed somewhat forward and out- ward, and from it may have emerged the ophthalmic branch of the fifth nerve, but the large size makes it more probable that it is really a venous foramen. Ventro-anteriorly, lies a downwardly-directed oval slit between the median septum and the side of the sphenethmoid, through which the second nerve emerged. Anteriorly, the sphe- nethmoid has a free edge, and anterior to this edge, the median septum 208 Annals of the South African Museum. is visible. The olfactory lobes were probably situated in this position. Antero-ventrally to the tapering end of the parasphenoid, the sphe- nethmoidal part of the interorbital septum is extraordinarily thickened and, in this part, rests on the dorso-median keel of the anterior ptery- goid ramus. Thethickening is solid, and I can make no suggestion as to its probable function. Posterior to this structure there is a gap, which thus lies between the sphenethmoid, pterygoid, and parasphenoid. The general plan of all these internal ossifications in the skull of Arctognathus is very similar to that which is known in the Dicynodonts from Pearson’s account. The main morphological features may be summarised :— Primitive. Preorbital depression, but very weak; septomaxilla with large facial exposure and large foramen; epipterygoid apparently rod-like; fenestrae ovales far apart; pro-otic relatively small; fenestra ovalis not situated high up in the brain-case. Advanced. Snout short and rounded in section; nostril large; orbits not wholly laterally directed; small postfrontal; preparietal small or absent; short, deep maxilla; tooth row short; dentigerous border with a distinctive, ventrally convex curve, with no step; pterygoid flange anteriorly situated; interorbital width reduced; intertemporal width reduced, and fairly large temporal openings; basisphenoidal tubera reduced. Arctognathus is thus clearly a Gorgonopsian well advanced in the direction of development taken by this group, but it still possesses a number of primitive features. It has been shown that a number of features noted by Watson as advanced were cited on invalid grounds, viz. the septomaxilla and its foramen are large as in other Gorgonopsians and there is no evidence that the epipterygoid is broad. The pro-otic is certainly not very advanced, and stands on the same developmental niveau as do Scymnognathus and Lepto- trachelus; in this respect “‘Lycaenodon”’ is much more advanced. The nature of the basisphenoid and its relations with the pterygoids and parasphenoid do not show very marked advances over the condition of the earlier Gorgonopsians that have been described above. Inthe attainment of a short rounded snout and the reduc- tion of the interorbital and intertemporal width, however, Arctog- nathus shows definite advances. The great amount of ossification in the median septum appears to be primitive and will probably be found in the earlier forms when better preserved material is examined. South African Gorgonopsia preserved in the British Museum. 209 Chief measurements: From premaxilla to basisphenoidal tubera ks arena. From premaxilla to pineal foramen my Ota, From pineal foramen to edge of occiput . ayy ps From premaxilla to front of orbit . : : gil). Length of lower jaw : LSS) Width across squamosals : ' ; 5 WAGY Interorbital width ; ; OMe Intertemporal width . ee En OOCn Width of snout . : G4. Width across lateral pervect anges ; : TA 96 Height of snout . : A SON ®. Height of occiput DOONEY es Mentum : ; OY ee Length of molar series . ; : Sil Diastema . He 5 Length of incisor series . TAB) ig Type, B.M.N.H., 47339, Kagaberg, Bedford, Cape Province. Arctognathus *curvimola, Owen. In the collection there is a preorbital portion of a skull with part of the lower jaw in position, but with the teeth badly preserved. Owen described it under the name Lycosaurus tigrinus; Broom, in 1911, showed that it could not be included in the genus Lycosaurus, and proposed the new name Arctosuchus, and, in 1932, considered it to be a Therocephalian. The specimen is a very bad type, but there is no doubt that it represents a Gorgonopsian, probably very close to Arctognathus. In 1921 Watson maintained that it was specifically identical with Arctognathus curvimola. I have etched the right side of the snout, and the maxillo-premaxillary suture now shows clearly; the relations of maxilla, premaxilla, septomaxilla, and septomaxillary foramen and the nostrils are typically Gorgonopsian and do not differ much from the condition in A. curvimola. There appear to be four incisors, although the roots of only two are actually preserved; the molars are badly preserved, but probably number five. All the features shown by the fragment thus agree very well with 4. curvimola and, as no differentiating features can be determined, there is no valid reason to distinguish it by name from A. curvimola. B.M.N.H., R1719, Mildenhalls, Fort Beaufort, Cape Province. 210 Annals of the South African Museum. Cynodraco serridens, Owen. Owen described a number of fragmentary snouts under this name, and another fragment presumably associated with a good humerus under the name of C. major. There is no evidence to warrant the separation into two distinct species. Moreover, so little can be determined of the structure that, except for the deep mentum in one specimen, there is very little evidence that the fragments are Gor- gonopsians at all. Owen’s lithographs show the general appearance of these specimens very well. They are: Type, B.M.N.H., 47084, Bovey’s Farm, Fort Beaufort, Cape Province. This is a bad piece of snout, which exhibits some features of the incisors; there are five upper incisors with fairly strong and coarse serrations, four lower incisors with similar serrations; the incisors are strong and protruding. B.M.N.H., 47086, Fort Beaufort, Cape Province. A mandibulary fragment showing the canine of both sides and also four incisors. B.M.N.H., 47085, Stylkrantz, Graaff-Reinet, Cape Province. Weathered and fragmentary snout; the fragmentary incisors show fairly coarse serrations on both anterior and posterior edges. The spur of the septomaxilla protruding into the nostril is shown on the right side. B.M.N.H., 47309, Mildenhalls, Fort Beaufort, Cape Province. This is Owen’s type of.C. major. It consists of a fragment of mandibulary symphysis with both upper canines attached. The roots of four large incisors are seen in section; the serrations on the posterior border of the very large canines are beautifully preserved. The symphysis is deep. B.M.N.H., 47310, Mildenhalls, Fort Beaufort, Cape Province. A very good left humerus showing both foramina, well illustrated in Owen’s lithograph. Chief measurements : Length , : > 260 mime Width across the epicondyles : A - Dahon Width across the proximal surface . 105 by 36 mm. Width across delto-pectoral crest . - \LhOjmam: Maximum width of shaft . : 4 iranSoune Minimum width of shaft ’ IMRU2G Length of delto-pectoral crest : L of DASE ES, South African Gorgonopsia preserved in the British Museum. 211 Cerdognathus greyi, Broom. The type is an imperfect dentary; the number of incisors cannot be ascertained, but four seems to be the most probable number; the canine and four molars are badly preserved as casts and, although no serrations can be seen, this evidence is simply negative; Broom is probably right in postulating a low coronoid process, as this would conform with the general contour. The very low, though upright, symphysis and the absence of a diastema stamps the dentary as a distinct type; but one wonders if such a fragment warrants the creation of a new generic name. Chief measurements : Probable length of dentary . ; a LOO? mm Height of symphysis. : : AM) og Length of molar series . : Si 4. Depth of dentary behind last molar ; i ae Type, B.M.N.H., R2892, Klippoort, Cradock, Cape Province. Scylacops capensis, Broom. This specimen consists of the middle third of a skull, which shows the greater part of the structure of that region. Watson’s description and figures appear to be correct in every detail. His attitude with regard to the specimen is to be highly commended. The snout is missing and the dentition is thus unknown and, as much of the classi- fication of the Gorgonopsia rests on this character, Watson, although recognising the fact that he had before him a representative of a new genus, refrained from naming it. Instead, he gave a good morphological account, which is of far greater value than a new generic name. Subsequent finds have shown the wisdom of such a procedure. We now have an excellent skull as the type of Scylacops capensis, named by Broom, but to Watson must go the honour of the first morphological description. B.M.N.H., 47098, ?Fort Beaufort, Cape Province. A Fore-foot of an Unnamed Form. (Fig. 18.) Lying on a series of vertebrae, there is a partial right fore-foot of a Gorgonopsian. The fourth digit is fully preserved; the third lacks 212 Annals of the South African Museum. only the point of the claw; the fifth has two phalanges preserved; parts of the third, fourth, and fifth metacarpals are present; the fourth and fifth distals are present as a single fused element, to which the distal half of the ulnare is articulated. The foot is typically Gorgonopsian (about two-thirds the size of that of Aelurognathus tagriceps), with the distinctive mammal-like epiphysial distal ends to the metacarpals, and with indications of an incipient reduction Fig. 18.—An unnamed Gorgonopsian. B.M.N.H., R3768. Dorsal aspect of the partial manus. x #$. U. =ulnare. 4+5 =fused fourth and fifth distal carpals. III-V =the third, fourth, and fifth digits. in the number of phalanges. The fourth and fifth metacarpals articulate with the fused fourth and fifth distals; the second phalanx of the fourth, and, in particular, of the third digit, is very much shortened; the third phalanx is robustly developed. As preserved, the phalangeal formula is—?, ?, 4,4, 3. If the first and second digits possessed two and three phalanges respectively, the structure of the digits would be exactly as in Aelurognathus tigriceps and, as in that form, dorso-ventral movement of the segments inter se is well developed with finely modelled articulatory faces. B.M.N.H., R3768, Oude Klip, Cape Province. South African Gorgonopsia preserved in the British Museum. 213 Short Discussion. In a series of papers on the Therocephalians, Gorgonopsians, Bauriamorphs, and Cynodonts, Watson has argued that one branch of the Therocephalians gave rise to the Bauriamorphs, and that the Gorgonopsians led on to the Cynodonts. With the first conclusion I can concur, but in this study of a limited number of Gorgonopsian skulls two important facts bearing on this question have been established, viz. the nature of the epipterygoid and the relations of the posterior end of the prevomers, and these seem to invalidate Watson’s second conclusion. In all the Gorgonopsians I have studied there is no evidence, whatsoever, of a tendency for the original slender rod-like epipterygoid to become widened and intimately incorporated into the side-wall of the brain-case; and in all, where the palate is known, the prevomers have their posterior ends tapering and are here clasped by the palatines and do not meet the pterygoids on the ventral surface. In the Therocephalia, on the other hand, there is very definite evidence of the progressive widening of the epipterygoid and, in this group, the posterior end of the prevomers is expanded and underlies the pterygoid in ventral view. This is also the condition in the Cynodonts. With respect to the nature of the prevomers and the epipterygoid, it seems that the Gorgon- opsians must be excluded from the direct ancestry of the Cynodonts; but at this stage it would be premature to maintain that these two characters indicate a Therocephalian ancestry. I am more inclined to think that no known Therocephalian indicates anything more than a parallel development in the two suborders, Therocephalia and Cynodontia. I do not intend to continue this discussion any further; this can be more profitably done when the various suborders of the Theriodontia are treated monographically, as I hope to be able to do in the near future. My thanks are due to the officers of the British Museum (Natural History) for permission and facilities to study the material in their charge. I am particularly grateful to Dr. W. E. Swinton for his constant readiness to facilitate my work, and to Mr. Parsons for doing some of the developmental work. Professor D. M. 8. Watson’s interest was very encouraging, and his critical advice always welcome. To the University of Stellenbosch I am indebted for a small scholar- ship, which has enabled me to undertake the visit to London. All but three of the figures are by my wife. eo ig Ag ( ee ey deit J ai =e, «re: . ta. oo “i a x 9 ire Se ea hy = < ae Ae i> ae el ay rt 3 ied ie ee Sta pot aan se caw ; ry ee ee Pei ie adaghage wide wy E Parr ty & bea prt wea ‘vt ited Sie : ed AD ate te aeeem ie bins hast atl = Ade meee , ae i -_ neem ( 215 ) 7. A Contribution to the Morphology of the Mammal-like Reptiles of the Suborder Therocephalia.—By Lizuwr D. Boonstra, D.Sc., Palaeontologist to the South African Museum and Queen Victoria Scholar of the University of Stellenbosch. (With 35 Text-figures.) Although many genera and species of Therocephalians have been described by Owen, Seeley, Broom, Watson, and Haughton since these forms became known about seventy years ago, very little is known of the cranial morphology. The outer surface of the skull has hitherto been the only structure adequately known; certain features of the palate of Scylacosaurus, Ictidosuchoides, Moschorhinus, Notosollasia, and Whaitsia have been described by Broom; Haughton has described the palate in Akidnognathus and Whaitsia, and a longitudinal section of the brain-case of Alopecognathus; Watson’s account of the palate and basicranial region of the Scaloposaurids has placed our knowledge of this family on a sound footing. In this paper the results of a study of the Therocephalians preserved in the British Museum (Natural History) are recorded. This collection consists of about twenty distinct types, and in a number of cases it has been possible to determine the structure of the palate and brain-case in addition to the dorsal, lateral, and occipital surfaces. As a number of these specimens are the historical types of Owen, Seeley, and Lydekker, a re-examination will serve as a useful introduction to a monographical account of the Therocephalia, which I hope to under- take in the near future. The suborder Therocephalia has hitherto been subdivided into five families, viz. Alopecopsidae, Ictidosuchidae, Pristerognathidae, Scalo- posauridae, and Whaitsidae; to these must now be added Huchamber- sidae and Lycideopsidae. The Pristerognathidae are represented in the collection by Alope- codon, Cynariognathus, Pristerognathus, Scylacosaurus, Scymnosaurus, and Trochosaurus; Hyorhynchus and probably also Theriodesmus may be included here. All these forms are from the Tapinocephalus zone, and represent the more generalised forms of the Therocephalia, which are on a definitely more primitive stage of development than the more advanced Scaloposaurids, Whaitsids, Huchambersia, and Lycideops. 216 Annals of the South African Museum. Alopecodon cf. priscus, Broom. A very much weathered and badly preserved snout with the anterior third of the mandible, showing only the dentition in a very unsatis- factory manner, was, in 1925, made the type of a new species— Alopecodon minor, Broom. Broom gave the dental formula i. 7, c.3, m.7 or 8. I find seven incisors as described by Broom; the two small teeth, which on account of their different direction may possibly be small canines, but may equally well represent two last incisors differing somewhat from the anterior ones, as there is no evidence that they are implanted on the maxilla; then a large canine, oval in cross-section; then a diastema of 8 mm.; then six closely packed molars followed by an empty socket, to which, medially, there lies a small tooth; the length of the molar series (7) is 24 mm. On none of the teeth are any serrations visible, but they may have been present in life. Both A. minor and A. rugosus are bad types. Broom’s specific characters comprise only the number of teeth, and as these are badly preserved, it is not a reliable criterion. In A. priscus the dental formula given by Broom isi. 7,¢.2+1,m.8?; in A. minori.7,c.2+1, m.7or8; in A.rugosusi.?,c. ?+1,m.7. Considering the nature of the preservation in these three specimens, the dental formulae approxi- mate so closely that, in the present state of our knowledge, it would appear advisable to consider both A. minor and A. rugosus as examples of Alopecodon priscus. It is true that they are considerably smaller forms, but this does not appear to be a sufficiently weighty reason for retaining three distinct specific names, each of which is only repre- sented by a very incomplete skull. B.M.N.H., R5750, Abraham’s Kraal, Prince Albert, Cape Province. Cynariognathus platyrhinus, Broom. (Fig. 1.) This specimen is a weathered snout, which Broom has determined as belonging to the same species as the type snout now in the American Museum. Sufficient of the teeth are preserved in a fair state of preservation to enable one to determine something of the dentition; two anterior and two posterior incisors and a space for another in between are present; there is one large canine, oval in cross-section; then a series of six closely set, fairly small molars; posterior to the sixth, there may have been one or more additional molars. In the lower jaw, a smaller canine followed by eight very closely set molars are visible; there is no diastema as in the maxillary Morphology of Mammal-like Reptiles of Suborder Therocephalia, 217 series; there does not seem to be room for more than three lower 1 2 incisors. The dental formula would thus be - ee nS ere * For ie 2,0 deanma8 the type, Broom has given the P 1,6, Go ls in. & Laie i.3,c.1,m.8 BY GC following formula : As both specimens are not well preserved, I believe it legitimate to assume that in life the denti- tions were probably similar. In the British Museum specimen the posterior border of the incisors, B. canines, and molars all carry fine serrations. Of the type, Broom stated, “There is no evidence of any serrations, and were they present the specimen would be A. expected to show some of them.” If Broom’s observations are cor- rect, there are thus points of Fic. 1. — Cynariognathus platyrhinus. B.M.N.H., R4097. Diagrammatic difference between the two speci- cross-sections through the snout. mens. Further evidence, based x 1. on more complete and better pre- ~ At the level of a kas : ; : ss ne ourth ,, served specimens, is, however, C. iM thind & necessary before a new species Mx.=maxilla. Pr.V. =prevomer. can legitimately be created. Pal.=palatine. Pter. =pterygoid. I have had three slices cut across the anterior part of the snout. These are reproduced here ina slightly diagrammatised form. It is seen that, posteriorly, the prevomers are fused, whereas, anteriorly, the fusion becomes less intimate and finally ceases; the palatine is seen to form a paired dorsal keel in the median line, which is supported by a girder of the anterior pterygoid ramus. Through the weathered right side the dorsal paired keel of the prevomers is seen; this supported the soft internasal septum. B.M.N.H., R4097, Uitkyk, Prince Albert, Cape Province. Pristerognathus polyodon, Seeley. The specimen in the collection is the type of the genus; it consists of a weathered snout, from which very little of morphological interest can be determined. As far as can be ascertained the dental formula WOllis LOO, INE AB 20 218 Annals of the South African Museum. i, Ole oa ema 3) Os lL ita OF fine serrations; in the upper incisors the crowns are lost, but one may assume that they also were serrated; the posterior border of the upper canine (and lower?) carry somewhat coarser serrations. The incisors are fairly large—so is the canine; the molar roots are of the same size as those of the incisors. There is a step on the maxillary border anterior to the canine. The splenial practically enters the symphysis on the ventral surface; further dorsally, it may actually enter it. Chief measurements: is : The posterior border of the lower incisors bears Length of the six upper incisors . : . 35 mm. Length of the three lower incisors ! -, | ae Width of upper canine ; is, See Width of the snout . : ; “+ 255i Height of the snout . ' ; Morr Type, B.M.N.H., R2581, Cypher, Tamboerfontein, Beaufort West, Cape Province. Under the number R5753 there is registered a badly preserved shoulder girdle, of which Broom has given good restored figures under the name ?Pristerognathus minor. The number R5752 refers to some cervical vertebrae of the same form. Scylacosaurus sclatert, Broom. (Figs. 2-5.) A nearly complete, though slightly weathered, skull, preserved in an extremely intractible matrix, was purchased by the British Museum, in 1912, from J. H. Whaits. This specimen has been referred to the above species. In point of size and arrangement of the external bones of the snout it agrees exactly with the type in the South African Museum. Broom’s account (1903) of the dentition of the type gave the dental formula 1. 6, c. 3, m. 7, incisors without serrations, canine probably with serrations. In 1932, he gave the formula 1. 6, c. 2, m. 7. In this specimen the incisor series is composed of five small, slender, and closely packed teeth; these are definitely on the premaxilla; behind them are three much smaller teeth; these appear to lie on the maxilla, as in lateral view the premaxillo-maxillary suture lies anterior to them; posterior to a short diastema lies the large canine (20x 7mm.); the molar teeth are of the same size and nature as the anterior incisors; on the right side, a closely set series of seven teeth sete Sere Fig. 2.—Scylacosaurus sclateri. xs B.M.N.H., R4055. Dorsal view of the skull. Abbreviations to this and the subsequent figures. Ang. =angular. Art. =articular. B.Oc. =basioccipital. B.P.P. =basipterygoid processes. B.Sph. =basisphenoid. Cor. =coronoid. Dent. =dentary. ict. Pter. =ectopterygoid. Ep.Pter. =epipterygoid. Ex.Oc. =exoccipital. Fen.Ov. =fenestra ovalis. BL. =foramen incisivum. FJ. =foramen jugulare. Ie. =frontal. I.Na. =internal nares. Ju. =jugal. La. =lacrymal. Mx. = maxilla. Na. =nasal, Pa. = parietal. Pal. = palatine. Jee dite = postfrontal. Pin. =pineal foramen. Pit. P.Oc. P.Orb. Pr. Art. Pr.Fr. Pr.Mx. Pr.Ot. Ven. = pituitary foramen. = paroccipital. = postorbital. = prearticular. = prefrontal. = premaxilla. = pro-otic. = prevomer. = parasphenoid. = pterygoid. =post-temporal fenestra. = quadrate. = quadratojugal. =surangular. =septomaxillary foramen. =septomaxilla. =supraoccipital. =suborbital vacuity. = splenial. =squamosal. = stapes. = tabular. = venous foramen. Roman numerals refer to the cranial nerves. 220 Annals of the South African Museum. occupy 23 mm.; on the left side, there are six less crowded teeth, with a gap for one additional tooth, occupying 27 mm.; the dental formula thus appears to be 1.5, c.1+3,m.7. All the teeth bear fine serrations on the posterior border. The dentigerous border curves upwards anterior to the canine, but there is no step. There are thus some differences in the dentition between the type and this specimen; allowing for the state of preservation and for a certain amount of individual variation, these differences should not restrain us from con- sidering them specifically identical. Mr. HK. L. Parsons, at my request, attempted to display the palate and occiput, but found that the bone was more fragile than the hard S.Mx. Na. Pelr Fro PFEr POrh Ja. Fic. 3.—Scylacosaurus sclateri. B.M.N.H., R4055. Lateral view of the skull. x ?. matrix, and had to abandon the attempt. On taking up the specimen again, I decided to make another attempt, and have succeeded in sufficiently exposing the left side of the brain-case and parts of the posterior half of the palate, so as to determine the general structure. In the accompanying figure, the anterior part of the palate has been drawn from the type-specimen in the South African Museum, as figured by Broom, and the posterior half from the British Museum specimen. The chief points of interest are: the widening of the basisphenoid and the bulge on the paroccipital, which characters are, in some respects, paralleled in the Scaloposaurids; the prevomers have a posterior shovel-shaped expansion, which underlies the ptery- goid; the anterior palatal fenestrae are relatively short; the paired tooth-bearing ridge lies wholly on the pterygoid, not, as in the Gor- Morphology of Mammal-hke Reptiles of Suborder Therocephalia. 221 gonopsians, partly on the palatine; the suborbital vacuities are large; an interpterygoid vacuity is present; the pro-otics enclose a broader space than in the Gorgonopsians; the auditory groove, formed by a ridge of the squamosal, is well developed. The brain-case does not show up very well, but is of considerable interest in that it shows that there is a certain amount of parallelism between this Pristerognathid and the Scaloposaurids, and Pr Me in that it differs considerably te from that of another Pristero- gnathid—Trochosaurus. The basisphenoid is a strong bone with widely sepa- rated, ridge-like tubera; lateral to the tubera proper, the basisphenoid extends still further laterally, and pon ye here forms a sharp distinct- Pee © Nowe ct. Phor ive ridge, which bounds AB : the fenestra ovalis antero- laterally and then continues posteriorly to meet the par- occipital (figs. 4, 5). This sharp ridge is not known in any other Therocephalian, but is present in Galesaurus andsome Anomodonts. An- teriorly, the basisphenoid carries a ventral keel and, : Bike anterior to the plane of the Qu pituitary fossa, carries the Fic. 4.—Scylacosaurus sclateri. B.M.N.H., horizontal basipterygoid FOS. Ventral view ofthe hl processes on its lateral face. Nothing can be determined of the relations of the basisphenoid with the parasphenoid. The lateral development of the basisphenoid, mentioned above, also occurs in the Scaloposaurids and Bauriamorphs, but if Watson’s account is correct, the relations differ somewhat, e.g. the flange does not bear a sharp ridge and, in Watson’s figure of Scaloposaurus, the fenestra ovalis lies anterior to the paroccipital and lateral to the widened end of the basisphenoid, whereas in Scylacosaurus it lies in 222 “Annals of the South African Museum. the normal Therapsid position between the basisphenoid, pro-otic, paroccipital, and basioccipital; in Scaloposaurus there does appear to be a hollow filled with matrix in the same position as the fenestra in Scylacosaurus, and, moreover, the depression figured by Watson as the fenestra ovalis in Scaloposaurus also occurs in Scylacosaurus, and here it definitely is not the fenestra ovalis. The pro-otic is strongly developed; its anterior margin is overlain by the epipterygoid so that the cavum epiptericum has not yet been included in the cranial cavity; the ventral part of this bone has been removed and it can now be seen that -the posterior border of the lateral opening into the pituitary fossa is formed by the basisphenoid Ply Prk. Fin. —_ | Nee = Fig. 5.—Scylacosaurus sclatert. B.M.N.H., R4055. Lateral view of the left side of the brain-case. Occiput, basipterygoid process, and the postorbital bar seen in section. Lower half of epipterygoid removed. x l. and pro-otic; in lateral view the epipterygoid obscures the perforation for the fifth cranial nerve; further dorsally les the large venous fossa found in all Therapsids; the dorsal limits of the pro-otic cannot be determined; here the side-wall is formed by flanges from the parietal, interparietal, and supraoccipital, but their exact relations cannot be ascertained. As far as can be determined, the anterior margins of the pro-otics do not approach each other as they do in some Gorgonopsians. The epipterygoid is a high bone standing with its elongated base on the quadrate ramus of the pterygoid in a plane immediately lateral to that occupied by the pro-otic, forming a cavum epiptericum; it is fairly narrow and flattened, but is proportionately twice as broad as in any known Gorgonopsian, and roughly half as broad as Morphology of Mammal-like Reptiles of Suborder Therocephalia. 223 in the Whaitsids and Cynodonts; dorsally, its inner surface is applied to the parietal, and its outer covered by a flange of the postorbital. The lower jaw is of the usual Therapsid nature, and its structure is best understood by referring to the figure, where it is shown in lateral view. Chief measurements : Length from premaxilla to the basioccipital condyle 2) doen Length from premaxilla to the pineal foramen MLNS) 38) Length from pineal foramen to edge of occipital plate . 25 ,, Length from premaxilla to anterior margin of orbit ue noo: o Length of the lower jaw : ; ; oa HOO! 5 Width across the squamosals 93. 5 Interorbital width : ; : 221 ae Intertemporal width . ! : : os Width of the snout. , : : Sob ee Height of the snout. ‘ : ; ; : S00 Height of the occiput . : ah ibe Oy as Width across the lateral pteryaoid ames OW Ais Height of mentum of lowerjaw . £4, EOD as, Length of the molar series . : . Zi. 5, Diastema . : ; d ; : " : Ss. Length of the incisor series . s : : ' 4 Die B.M.N.H., R4055, Fraserburg Road Station, Prince Albert, Cape Province. Scymnosaurus watsom, Broom. (Figs. 6-8.) The type-specimen consists of the greater part of a large skull, lacking the snout; some vertebrae and ends of limb-bones, from which very little can be determined. The palate was figured, in 1914, by Watson under the name Lycosuchus?. In 1915 Broom published outline drawings of the dorsal and lateral aspects, without indicating the constituent bones, and proposed the new name— Scymnosaurus watsont. In 1921 Watson published a fuller account; his figures and account of the internal aspect of the brain-case I find to be correct, and here he corrected his former error in the orientation of the articulatory surfaces for the lower jaw. I have refigured the palatal aspect, as my interpretation of its structure differs con- siderably from that of Watson. The most obvious error is that he maintained that a median bone, which he called the vomer, was 224 Annals of the South African Museum. present in the middle third of the palate, lying between the pterygoids and palatines, and posteriorly separated from the basisphenoidal Fia. 6.—Scymnosaurus watsoni. B.M.N.H., R4100. Dorsal view of the Skalle) = xt rostrum by the interpterygoid vacuity and the pterygoids. The long median slip of bone interpreted as ‘“‘vomer,” is really the fused Fia. 7.—Scymnosaurus watsoni. B.M.N.H., R4100. Lateral view of the skull. x 4. dorsal keel of the pterygoids, on which the median septum rested. The state of preservation of the specimen is, however, such that it is only in the light of our increased knowledge that we are able to Morphology of Mammal-hke Reptiles of Suborder Therocephalia. 225 establish the fact that no median “vomer” can possibly be present in this position. From my figure it can be seen that the relation of the prevomers, palatines, pterygoids, and ectopterygoids is typical as in all known Therocephalia. Attention may here be drawn to the points in which the Thero- cephalian palate differs from that of the Gorgonopsians. ee ae = sic Tub Fie. 8.—Scymnosaurus watsoni. B.M.N.H., R4100. Ventral view of the skull. x 4. (a) The internal narial opening is relatively shorter and, in the Whaitsids, is wholly or partly bipartite. (6) Posterior to the plane of the nares, the prevomers expand in shovel-shaped fashion and underlie the palatines and ptery- goids, whereas in the Gorgonopsians the prevomers end posteriorly as tapering slips of bone, laterally flanked by the palatines. (c) The suborbital vacuities are large, and consequently the ecto- pterygoids are reduced to form their posterior girder-like border. 220; Annals of the South African Museum. (d) The median dentigerous ridges, when present, are carried wholly by the pterygoids. (ec) The pterygoid has a long anterior ramus, which meets the prevomers; the palatines thus do not meet in the median line, whereas they do in the Gorgonopsians. (f) The quadrate ramus of the pterygoid has a straight outer edge, whereas in the Gorgonopsians it is concave. (g) The paroccipital bar is a much stronger element. (hk) The ventral basisphenoidal keel is very deep in nearly all Therocephalians. I had considerable difficulty in determining the structure of the dorsal and lateral surfaces, but, by a lengthy process of alternately grinding down with a file and etching with acid, the limits of the majority of elements are now visible, and are shown in the accom- panying figures. The main points of interest are: the deep preorbital depression; the two grooves on the frontal, which is short and cruciform with only a small entry into the orbital border; the large prefrontal and postfrontal; the deep maxilla; the peculiarly high slip of the jugal on the postorbital bar; the high, thin parietal crest; the somewhat anteriorly directed orbits; the small posttemporal fossae; the epipterygoid, with straight edges, proportionately twice as broad as in the Gorgonopsians. Chief measurements : Length from premaxilla to the basioccipital condyle . 240? mm. Length from premaxilla to the pineal foramen : . MSO Fare Length from pineal foramen to the occipital plate . « Oma Length from premaxilla to the front of the orbit . . Oe Length of the lower jaw : F , LOD aes Width across the squamosals ; : : : , 2a Interorbital width : ‘ : : : : 50m Intertemporal width . : : ‘ eae : homes Width of the snout. ; : : : Geb 5. Width across the lateral nterroeid Toe 5 SEINE Height of the snout. : ; 62a Height of the occiput : - LOD ea Length of the molar series : Doe Type, B.M.N.H., R4100, Uitkyk, Prince Albert, Cape I Province. Morphology of Mammal-like Reptiles of Suborder Therocephalia. 227 Trochosaurus major, Broom. (Figs. 9-11.) The type of this species is in the American Museum, and was described by Broom as Trochosuchus major. Haughton described a specimen in the South African Museum as Trochosaurus intermedius, Wc Fie. 9.—Trochosaurus major. B.M.N.H., R5747. Ventral view of the skull. x 4. and, in 1932, Broom referred his species to Haughton’s genus. The specimen in the British Museum formed the subject of the figures of the dorsal and lateral surfaces published by Broom in 1932. After Mr. Parsons had done the preliminary development, I succeeded in exposing the whole palate, occiput, and the greater part of the lateral surface of the brain-case on the right side. In the type, Broom found the dental formula to be 1.5, c. 2, m. 3%; 228 Annals of the South African Museum. the incisors have a strongly serrated posterior border. In the British Museum specimen the dental formula is identical, but here there is, in addition, definite evidence that the molars as well as the canines and incisors are replaced; it is not certain that the last incisor is implanted on the premaxilla. I find Broom’s figures of the dorsal and lateral surfaces correct; in this supplementary account only the palate, occiput, and lateral aspect of the brain-case will be figured and described. The skull had been fissured, and subsequently weathered at the plane of the lateral pterygoid flanges so that this region is not very well shown; the middle part of the palate is revealed by a frontal fracture; the anterior and posterior portions have been chiselled out. As shown in the figure, the palatal structure is typically Thero- cephalian; the paired prevomers separate two short and wide fenestrae—the anterior part of which housed the lower canine, and the posterior functioned as internal nares; the prevomers have a shovel-shaped posterior expansion, which underlies the palatines and pterygoids; the palatines do not meet in the median line; the suborbital vacuities are large, and the ectopterygoids appear to be beam-like structures forming the posterior and lateral borders of these vacuities; the extent to which the ectopterygoids participate in the formation of the transverse pterygoidal bar cannot be deter- mined; posteriorly, the deep median keel, formed by the pterygoids flanking the basisphenoidal rostrum, is a prominent feature; the quadrate ramus of the pterygoid has a nearly straight outer edge; the basisphenoidal tubera are very massive and rounded, and diverge greatly and have a deep hollow between them, with the result that the fenestrae ovales are very far apart and the stapes short; there appears to be no evidence of a “‘vomer”’ underlying the basisphenoidal tubera. The flat nature of the anterior part of the palatal roof is in strong contrast to the deeply vaulted antero-median part of the palate of the Whaitsids. In the occiput the structure is adequately shown, only the basi- occipital condyle being lost. In general appearance the low and wide occiput is very similar to that of the Whaitsids, but there are great differences in the proportions of the constituent elements. A curious step above the supraoccipital forces the interparietal on to the dorsal surface; the post-temporal fenestrae are situated very high up and are far apart; the paroccipitals are enormously developed, their distal ends being bipartite, the lower supporting the quadrate and the upper supporting the strongly developed auditory groove Morphology of Mammal-like Reptiles of Suborder Therocephalia. 229 of the squamosal. In posterior view, the quadrate and quadratojugal appear to be unfused, the quadrate is rounded and forms the main articulation, whereas the quadratojugal is exposed on the lateral surface and forms the small outer part of the articulatory surface; the stapes is short; the foramen magnum is small; in section it is seen that, at the plane of the foramen magnum, the dorso-lateral corners of the condyle are formed by the exoccipitals; the foramen jugulare is large; the basioccipital sends two processes ventrally to support the widely separated and massive basisphenoidal tubera; on this specimen I can see no evidence that there are three bones in this region; if, however, these ventrally directed processes really represent the basisphenoid intimately fused to the basioccipital, then the bone lying ventrally to them would represent a parasphenoid. Fig. 10.—Trochosaurus major. B.M.N.H., R5747. Posterior view of the occiput. x %. A successfully induced parasagittal fracture has enabled me to get at the brain-case from the right side; this section also reveals the structure of the occiput (fig. 11). The basisphenoid is seen to be a large bone with very strong posteriorly directed tubera, which rest on the basioccipital processes; the floor of the posterior part of the brain-case is thus very wide; laterally, it forms the ventral and anterior borders of the fenestra ovalis (the stapes is in situ), and its rostrum is flanked by the ptery- goids; the basipterygoid processes are horizontal and flattened, and are immovably clasped by the pterygoids; dorsally, it meets the pro- otic in a long suture and forms the lateral border of the opening leading into the pituitary fossa; anteriorly, it is continued as a median keel, but it has not been possible to determine where the transition into the parasphenoid occurs; the pituitary would le in the dorsal surface of the basisphenoid. 230 Annals of the South African Museum. The pro-otic is a large bone, whose anterior portion is flanked by the epipterygoid; posteriorly, it is applied to the anterior surface of the paroccipital, and in part to the squamosal; antero-dorsal to the fenestra ovalis lies a small foramen for the seventh cranial nerve, and under this there is a small depression for the geniculate ganglion; as, in lateral view, the anterior part of the pro-otic is overlain by the epipterygoid, forming a cavum epiptericum, the foramina for the fifth and sixth nerves cannot be seen; dorsally, the pro-otic forms the ventral border of the venous fossa. Dorsal to the pro-otic, the side-wall of the posterior part of the brain- case 1s formed by the supraoccipital, interparietal, and parietal; the ! ROD) Ep. Pler. Fic. 11.—Trochosaurus major. B.M.N.H., R5747. Lateral view of the right side of the brain-case. Occiput and postorbital bar seen in section; part of the epipterygoid removed to show the pro-otic. x 4. relations of these bones are shown in the accompanying figure and requires no verbal description. The epipterygoid is a large and broad plate of bone, which forms the side-wall of the cavum epiptericum; ventrally, it stands on the quadrate ramus of the pterygoid in the usual Therapsid manner: continuing upwards it obscures the lateral opening into the pituitary fossa and the foramina for the fifth and sixth nerves from lateral view; dorsally, it meets the parietal and is flanked by a thin sheet of the postorbital; the epipterygoid is thus seen to be relatively more than twice as broad as that of any known Gorgonopsian, and has the appearance of the relations hitherto known only in the Cynodonts, amongst the Therapsids. It differs from that known in Scymnosaurus, Scylacosaurus, and Ictidosuchoides in that it is not a relatively narrow plate of bone with straight edges, but has a shaft-like middle and Morphology of Mammal-like Reptiles of Suborder Therocephalia. 231 expanded upper and lower ends, a condition reaching its greatest development in the later Whaitsids. In the nature of the brain-case and the basicranial axis, T'rocho- saurus differs considerably from that of the other members of the family Pristerognathidae. An examination of further material will undoubtedly show that this family includes forms which will event- ually have to be arranged in more than one group. It is, moreover, clear that the Therocephalians are, already in the Tapinocephalus zone, a well-established group, in which a number of different evolutionary tendencies have been developed. Chief measurements: Length from the premaxilla to the basisphenoidaltubera . 225 mm. Length from the premaxilla to the pineal foramen . ot ROME S Length from the pineal foramen to the edge of the occipital plate : oy ae Length from the premaxilla tp étie Hone of He osbit 5 I Length of the lower jaw “ . , SheoiliGe us. Width across the squamosals 4 ; beh GOL Oe: Interorbital width : . 402 Intertemporal width . ‘ : : : ; 5 20m Width of the snout. . : Bb, Width across the lateral pterygoid flange : : PRO O Le Height of the snout . oe GOR: Height of the occiput . cihwaieO) Bes Length of the molar series. ; Area bali Length of the diastema : ae Length of the incisor series . : inte! 0 a B.M.N.H., R5747, Tapinocephalus zone ?, South Africa. Hyorhynchus platyceps, Seeley. The incomplete skull shows the structure of the middle third of the roof of the skull; the frontal is short and has a large entry into the orbital border; the postfrontal and postorbital are well developed; the pineal foramen is large; the parietal region is narrow. Although some bones are visible on the palatal surface, I do not quite understand their relations. Hyorhynchus is undoubtedly a member of the Pristero- gnathidae, and the few characters that can be determined seem to indicate a close relationship to Pristerognathus. B.M.N.H., R872, Gouph, South Africa. 232 Annals of the South African Museum. The Ictidosuchidae are represented in the collection by two speci- mens referred to two genera, viz. Cerdodon and Ictidosuchoides. Cerdodon, which is not well known, is from an uncertain horizon in the Lower Beaufort, whereas Ictidosuchoides is known from a good skull, and is from the Endothiodon zone. Cerdodon tenuidens, Broom. In 1878, T. Bain had collected a calcareous nodule, which contained part of a Therapsid skull; the anterior part of the left side was weathered .and showed a portion of maxilla, premaxilla, and dentary. So little could be determined that neither Owen, Seeley, Lydekker, or Watson named it, but, in 1915, Broom produced a generic description, together with a number of others equally valueless. It is a pity that some authors will persist in naming such fragments, which are of no morphological or phylogenetic value whatsoever. 1.5.%,¢. 1, m. T=82 y In 1915, Broom gave the dental formula, ———_—___—____, and in 1.5; €. lane 62s ce 1 mae) 1.4, c. i 4 1932, DOP ee . I find the formula to be, 1 Se The 1. te ol mee 1. 35 (Cs. aes posterior border of the lower molars, which alone have the crowns preserved, are serrated; the incisors are small and slender. I asked Mr. Parsons to attempt to have a parasagittal section made in order that something more definite may be known of this “‘type.”’ Unfortunately, this section was not cut true, with the result that very little of the structure could be determined; an additional slice was cut in an attempt to obtain a section along the median line, but also without success. The external weathered surface of the left side, figured by Broom, is thus all that can be determined in this new genus. Type, B.M.N.H., 49420, Gouph, South Africa. Ictidosuchoides longiceps, Broom. (Figs. 12-14.) In his original description Broom gave the dental formula, 1. ?, c. 3, m. 9, and maintained that anterior to the large canine there were an additional two teeth implanted in the maxilla. Unable to find these two teeth, and thinking that they may have been lost in transit and through handling, I ground down the anterior maxillary border in order to expose the roots. At the level of the maxillary edge two filled alveoli were found, but no crowns extended beyond the maxillary border; it would thus appear that these two teeth were in the process of eruption. No serrations are visible on the canine or the molars. Morphology of Mammal-like Reptiles of Suborder Therocephalia. 233 Broom found that due to the narrowness and depth of the anterior part of the palate the infilling matrix could not be removed without damaging the maxillary teeth. By careful preparation with a fine f=, - i-S 2 ‘ 7 ost i ee — ve 0 A Eee ny " ‘ " 1 ‘ Oo cee 7 “ soe = ; Ko at Ae “ , Ste Me oe Sac Jd =e Os 3 , ose s ee, 4 Fie. 12. — Ictidosuchoides longiceps. Fie. 14. — Ictidosuchoides longiceps. B.M.N.H., R5744. Dorsal view of B.M.N.H., R5744. Ventral view the skull. x 4. of the skull. x 3. Fig. 13.—Ictidosuchoides longiceps. B.M.N.H., R5744. Lateral view of the skull. x 4. chisel I have been successful in exposing the whole palate, but, un- fortunately, the premaxillaries had fallen out prior to fossilisation, and thus the nature of the junction of the palatal processes of the pre- maxillaries with the prevomers cannot be determined. An oblique fracture at the level of the last molar reveals that, on the dorsal surface of the unfused prevomers, a pair of ridges are developed, and the groove WO, VOOR, DAI A, 21 234 Annals of the South African Museum. in between them must have supported the soft internasal septum. The prevomers are of the usual Therocephalian type; anteriorly, the end is fan-shaped, and posterior to the level of the internal nares, they again expand and underlie the anterior rami of the pterygoids and the median edges of the palatines well behind the level of the last molar. The interpterygoid vacuity has its posterior margin somewhat indefinitely preserved, but it would appear that it was very similar to that of Scaloposaurus and Icticephalus. The suborbital vacuities are large, and the lateral border formed by the ectopterygoids are of the narrow beam-like character as in Scaloposaurus. The sweep of the quadrate ramus is typical of the Therocephalians, and nothing of the web, which in Scaloposaurus and Icticephalus connects the lateral and posterior pterygoid rami, can be seen. The basisphenoid, however, appears to be of the same broad nature as in the Scaloposaurids, and, although the specimen is not very well preserved in this region, there does appear to be some evidence that the pro-otic is applied to the sides of the basisphenoid in a manner very similar to that described by Watson in the type of Scaloposaurus. Only part of the epiptery- goid is preserved; this is very similar to that of Scylacosaurus and Scymnosaurus, 2.e. its dorsal and ventral ends are not expanded, and its sides straight; in this character I[ctidosuchoides is thus distinct from Trochosaurus and the Whaitsids, and more akin to the generalised Pristerognathids. In the nature of the dentition, the slight flaring-out of the maxillaries, the elongated interpterygoid vacuity, the absence of a postfrontal, the slender inferior temporal and postorbital bar, the apparently widened basisphenoid, and the large suborbital vacuities bounded by a beam- like epipterygoid, Ictidosuchoides approaches the Scaloposaurids. The narrow temporal region with its sharp crest, the comparatively long and large temporal fossa, the complete postorbital bar, the confine- ment of the interparietal to the occipital surface, the large pineal foramen, and the apparent absence of a web of bone connecting the lateral and posterior pterygoid rami, go to show that Ictidosuchoides is a form more akin to the more generalised Pristerognathids, but has some parallel characters, which are typical of the Scaloposaurids. Chief measurements: Length from premaxilla to the basisphenoidal tubera . 140? mm. Length from premaxilla to the pineal foramen . 0a Length from pineal foramen to edge of occipital plate . 18 ,, Length from premaxilla to front of orbit .. 15 ee Width across the squamosals «|, 16O2iie Morphology of Mammal-like Reptiles of Suborder Therocephalia. 235 Interorbital width . . ; 24 mm. Intertemporal width . ; ; selec lO ee Width of the snout . : : : ile wie Width across the lateral pterygoid anaes : She Pd Or onl Height of the snout . , sa wt hy ag Height of the occiput. : : : : wy an28 2 . Length of the molar series : ! : P : 2 ies eh Diastema ; : : ‘ ie 4 Length of the three canines. é ‘ 4 18 eee Type, B.M.N.H., R5744, Bruintjieshoogte, Somerset Hast, Cape Province. The family, Lycideopsidae, is instituted for the reception of a single crushed skull; this skull cannot be included in any of the existing families without affecting the homogeneity of these groups. The characters distinguishing this from the other families are—extremely long dorso-ventrally depressed skull, with an extremely long molar series (10 teeth), consisting of ill-developed teeth, which may be absent in old age. Lycideops longiceps, Broom. (Figs. 15-16.) In his original description of 1931, Broom gave little more than the dental formula, 1.5, c.1,m.8. In 1932, a slightly fuller description and a restoration of the anterior half of the left side was given without the constituent elements being indicated. Although considerably crushed and sun-cracked, it has been possible to clear the external surfaces and most of the palate of a considerable amount of matrix. The extensive cracking, however, makes it extremely difficult to determine the structure. The dental formula is ue 53. 2; mii 10 mere. m5 crowded, and in the lower jaw there are no incisors at all; in the upper jaw two long and slender canines are present, but it would appear that the posterior one is being replaced; in the lower jaw there is only one canine, situated very far anteriorly; the upper molars are exceedingly small teeth, occupying as much as 50 mm.; they are well separated, and appear to be rudimentary; in the lower jaw the molars are also very small, with large interspaces; they occupy 26 mm.; there is a diastema anterior as well as posterior to the canine. Broom is The incisors of the upper jaw are very small and rather 236 Annals of the South African Museum. probably right in thinking that the molars eventually disappear, and that Lycideops would in old age be molarless. The skull has lost the greater part of its posterior third; on the left side, however, the lower jaw is complete and is in articulation with the quadrate and quadratojugal; part of the squamosal is also preserved. The species is remarkable for its great relative length and con- sequent slenderness; the snout, in particular, is very long and slender, and was rounded in section; an antorbital depression, if any, must have been shallow; the orbit was large and rounded, directed as much outwards as upwards; the length of the molar series is very great. As far as can be determined, the palate is typically Therocephalian, with Ect Peer Poe ee Fig. 15.—Lycideops longiceps. B.M.N.H., R5695. Lateral view of the skull, with the distortion corrected. x 4. all the elements relatively lengthened; the transverse pterygoidal bar is deep; the suborbital vacuities fairly large; the anterior ramus of the pterygoid is very long; the vacuities for the internal nares are extremely short, absolutely as well as relatively; they are nearly rounded and the anterior part, which received the lower canine, situated very far anteriorly. The lower jaw 1s very long and slender, and the symphysis is extremely weak; there are no incisors, and the splenial, if present, must have been weak, especially at the symphysis. This form, with its incipient molar reduction, its depressed form and rounded snout, narrow, transverse pterygoidal bar, and the curved contour of the mandible, has much in common with the Whaitsids. Its great length and the presence of a long molar series, however, excludes it from the Whaitsids proper. Lycideops may thus be regarded as a form, which in certain respects foreshadows the Whaitsid structure. Morphology of Mammal-lke Reptiles of Suborder Therocephalia. 237 Fie. 16.—Lycideops longiceps. B.M.N.H., R5695. A. Inner view of the left mandible. x 4. B. Outer view of the left mandible. x 4 Chief measurements : Length from premaxilla to front of the orbit . 120 mm. Length of the lower jaw : ; 5 ello Interorbital width : : : : ; 32a Width of the snout. : : : LOW Width across the lateral pterygoidal flanges . 58 ,, Height of the snout. : : . BO 5, Height of the mentum. 4 ; , ; Dias Length of the molar series. hee D0: 3 Diastema . ; ; : 13 or 22 mm. Length of the incisor series . 24 mm. Type, B.M.N.H., R5695, Thaba ’Nchu, Orange Free State. The family, Whaitsidae, is particularly well represented in the collection in the British Museum, viz. a good anterior half of the type-skull of Moschorhinus kitchingi, a practically complete skull of the type of Notosollasia laticeps, the imperfect skull of the type of Theriognathus microps, the excellent type-skull, and an additional snout, of Whaitsia major, and the imperfect snout, which is the type of Tigrisuchus simus. All these forms fall into a homogenous group, 238 Annals of the South African Museum. having all the characters enumerated by Haughton in his original description of Whaitsia platyceps. The Whaitsidae may be defined as follows: Therocephalians with a broad, depressed, moderately short and rounded snout; large temporal openings; crested parietal region with a small pineal foramen; molars absent or obsolescent; width across the lateral pterygoidal flanges reduced; the anterior palatal vacuity greatly shortened, wholly or partly separated into an anterior foramen 1ncisivum, and a posterior internal narial opening; the anterior part of the palate deeply vaulted with an ingrowing of the maxillaries, which is sometimes met by an outwardly directed swelling of the prevomers; suborbital vacuity variable, absent, small or large; epipterygoid broad, with expanded dorsal and ventral ends; postfrontal absent; very strong paroccipital process; weak, but long, mandibular symphysis and distinctively curved dentary. Moschorhinus kitchingi, Broom. (Figs. 17-19.) Broom’s original account and figures are good; but, since I have developed and etched the skull, some additional features can be recorded. Although it is possible that in life six incisors were present, this is by no means “quite manifestly”? so; actually, on either side, only five are preserved; on both sides there are no functional molars, but sockets of two are preserved. The dental formula can thus be givenasi.5,c.1,m.2%. Inthe accompanying figures of the dorsal and lateral surfaces a number of additional features are shown; it is evident that the septomaxilla is a much larger bone than Broom thought—its posterior extent, in particular, is very large; the short frontals form only a small portion of the supraorbital border; there is no postfrontal; the lacrymal carries a distinctive tubercle; there is no step in the dentigerous border; the ectopterygoid meets the jugal. The palate of Moschorhinus is essentially similar to that of the other members of the family Whaitsidae, but there are some very interesting and illuminating differences in the less basic points of structure; the entire palate is relatively wider than in the other Whaitsids, and, due to its absolute greater width and less vaulted nature, creates an impression of still greater width; this is due to the fact that the maxillaries and, to a lesser extent, the palatines do not curve inwards as much as they do in the other Whaitsids; in Moschorhinus the alveolar edge is fairly straight, whereas in the other Whaitsids it sweeps inwards with a medially directed convexity; Morphology of Mammal-hke Reptiles of Suborder Therocephalia. 239 on the inner vertical surface of the palatines there is a sharp oblique ridge for the attachment of the soft palate; this is situated half-way up, whereas in Notosollasia, Theriognathus, and Whaitsia it lies on the ventral edge. The prevomers are of a distinctive shape; anteriorly, they are greatly expanded, and carry only a very low ventral keel on their middle third; a fracture reveals the fact that the premaxillaries Jw.— Orb, Po. Fie. 17.—Moschorhinus kitchingi. B.M.N.H., R5698. Dorsal view of the : anterior half of the skull. x 4. extend on to the dorsal surface of the prevomers; posteriorly, the prevomers underlie the pterygoids and palatines and present a shovel- shaped outline; they form the greater part of the posterior margin of the extremely short internal nares; there is no ventro-lateral swelling of the prevomers, which in Notosollasia and Whaitsia divides the anterior palatal vacuities in two unequal parts; in Moschorhinus these vacuities are short but wide, and slightly con- stricted to produce a dumb-bell shape; the whole functioned as internal nares, since the lower canine was received in a depression 240 Annals of the South African Museum. Pe, Mx. ¥, OAC ik ee £: a § <7 et 1 2 va . PAZ Fic. 18.—WMoschorhinus kitchingt. B.M.N.H., R5698. Ventral view of the anterior half of the skull. x 3. PO-b Fig. 19.—Moschorhinus kitchingi. B.M.N.H., R5698. Lateral view of the skull; posterior half reconstructed. x 4. Morphology of Mammal-like Reptiles of Suborder Therocephalia. 241 in the premaxilla, anterior to the vacuity and the upper canine, whereas in Whaztsia the anterior part functioned as a foramen in- cisivum, and the posterior as internal nares; the ectopterygoids are strongly developed; they form the lateral borders of the large sub- orbital vacuities and, posteriorly, send a strong flange to support the anterior surface of the transverse pterygoidal bar, but do not form the whole corner of this bar as they do in Whattsia, Theriognathus, and Notosollasia. The nature of the suborbital vacuities is more akin to that of the Scaloposaurids and the less advanced Pristero- gnathids, than to that of the other Whaitsids, where they are very small or absent; in the Gorgonopsians they are also very small or absent. The pterygoids are only partly preserved; owing to the size of the suborbital vacuities and the posterior extent of the prevomers, the pterygoids have, anteriorly, a much smaller palatal surface than in the other Whaitsids; as is shown in the figure, a very deep but thin median keel is developed on the ventral surface of the anterior rami of the pterygoids—a feature not found in the other Whaitsids; lateral to this there lies a deep groove, and then there lies the deep, thin, postero-medially directed flange of the lateral rami of the ptery- goids; there is no interpterygoid vacuity. Unfortunately, nothing is preserved of the posterior half of the skull. The skull of Moschorhinus kitchingi throws an illuminating light on the nature of the Whaitsids. The broad depressed snout, the apparently large temporal openings, the reduced width of the lateral pterygoidal bar, the absence of the postfrontal, the reduction of the anterior palatal vacuities, the slight inward swelling of the maxillaries and the palatines, and, finally, the reduction of the molar series, clearly stamps it as a Whaitsid. The persistence of two molars, the unipartite anterior palatal vacuities, the large suborbital vacuities, and the ectopterygoid not forming the corner of the pterygoidal bar, show this species to be less advanced than the Whaitsids—Therio- gnathus, Notosollasia, and Whaitsia, and probably indicates the road traversed by these more advanced forms. In an unnamed species of Moschorhinus, described by Broom, there are two per- sistent molars in a skull which is mature; and in the inadequately known Moschorhinus warreni there are three upper and four lower molars. We-thus appear to have a progressive series showing molar reduction. The relations of Moschorhinus, and thus all the Whaitsids, to the other Therocephalians are not very clear; the three specimens of 242 Annals of the South African Museum. Moschorhinus are manifestly considerably removed from the less advanced Pristerognathids, and only the discovery of some forms less advanced than Moschorhinus will enable one to trace the ancestors of the Whaitsids and also of the parallel family of Scaloposaurids. Chief measurements : Length from premaxilla to pinealforamen . 145 mm. Length from premaxilla to front of orbit 93.3 Interorbital width : : . ; 5G" aes Intertemporal width . : : : 1 Onno Width of the snout : : Sa ae Width across the lateral feeadal Hanon ; oO re Height of the snout. 1 DOG Length of the molar series . : 4 : ies Diastema . : i ; Ae gs Length of the incisor series . : A Type, B.M.N.H., R5698, Bethesda Road Station, Kea: Province. Notosollasia laticeps, Broom. (Figs. 20-24.) Broom founded this new genus in 1925, and in his description gave little more than the dental formula—i. 4,c.1,m.0. In 1932, he gave the formula—i. 5, c. 1, m. 0, and stated that a small secondary palate is formed by the union of the maxilla and a downward development of the prevomers; he could find no suborbital vacuity. I have spent a considerable time in removing masses of matrix and in etching the skull, which now shows a number of very interesting points of structure. In the upper jaw there are four well-developed incisors with longi- tudinal grooves, but with no serrations on either the anterior or posterior edges; these are followed by a very much smaller fifth incisor, which appears to be in the process of eruption; between the last incisor and the canine there is a space of 4 mm.; the canine is a curved tooth of medium size; on the left side there is another canine being absorbed, whereas on the right there is a filled-in alveolus behind the functional canine; posterior to the canines, there are no indications of molars whatsoever. Noftosollasia thus differs in this character from Moschorhinus, where two molar alveoli or even three teeth persist. The palate of Notosollasia is of some interest; anteriorly, the paired prevomers meet the palatal processes of the premaxillaries and form PO-b. -ann 7. Snr} Fic. 20.—WNotosollasia laticeps. B.M.N.H., R5699. Lateral view of the skull. x 3. Fic. 21.—Notosollasia laticeps. B.M.N.H., R5699. Ventral view of the skulle \' e. a) ~, ‘ S xo, y , XY S SS oe \ Wx) 600% > RR y \) ON Se we ©: 8) 86 OR S SAS S Ss XS ON °. 0°, 7° °, oo, mS 0: SSR LOY % Ne : OX . ROX SR So eS 2. S82 OO B.M.N.H., 47065. Fie. 26.—Theriognathus microps. A. Sagittal section, reconstructed from a number of partial frontal, parasagittal, sagittal, and cross-sections. x 4. B. Cross-section through the brain-case, just anterior to the pituitary fossa. x 1. 22 It is now clear that, basing his conclusions mainly on the molarless nature of the jaws and the structure of the mandible, Broom was VOL. XXXI, PART 2. 250 Annals of the South African Museum. right in referring this form to the family Whaztsidae. The general appearance and arrangement of the various palatal and occipital bones and the structure of the brain-case bear a very close resemblance to that of Whaitsia, Moschorhinus, and Notosollasia. The anterior end of the snout is missing and nothing can be ascer- tained as to the nature of the canines and incisors; as both alveolar edges have been perfectly exposed it is possible to state definitely that no molars were present; there are not even indications of alveoli. The dorsal and lateral surfaces have for their greater part been lost, and the occipital surface has also been much damaged. The palatal surface, although somewhat sun-cracked, is preserved in a sufficiently complete state for its whole structure to be deter- mined. The relations of the constituent elements are shown in the accompanying figure. Although only the posterior borders of the anterior palatal vacuities are preserved, it would appear that they were of the type of those in Moschorhinus rather than that of Notosol- lasia and Whaitsia, 2.e. the whole functioned as internal nares, the canines of the lower jaw being received in hollows antero-laterally to the nares. The prevomers are relatively long and form the median part of the roof of the deeply vaulted anterior half of the palate. On both the dorsal and ventral surfaces the prevomers carry high median keels. Lateral to the prevomers he the palatines, which form the deep lateral walls to the highly vaulted median part of the palate. The palatines extend laterally to meet the maxillaries at the edentulous alveolar edge; on their medial surface the palatines carry a well-defined ridge, to which in life the soft palate must have been attached; between the prevomers and the soft palate a spacious air-passage was thus developed. The ectopterygoids are remarkable in that they form the strong lateral corner of the transverse pterygoidal bar, whereas, in the normal Therocephalian condition, and also in Mos- chorhinus, they are only supporting flanges to the pterygoids, which themselves form the corners. A small suborbital vacuity is present; this is very similar to that of Notosollasia, whereas it is large in Moschorhinus and, by both Broom and Haughton, maintained to be absent in Whaitsia platyceps, as it isin Whaitsia major. The pterygoid is very well developed; it forms a large part of the middle portion of the palate; the lateral flanges are not very strong since the corners of the pterygoidal bar are formed by the ectopterygoids; the posterior rami form, in their anterior part, two diverging flanges, in between which lies a deep groove; continuing posteriorly, these flanges form a very deep keel and clasp between them a similar high unpaired keel Morphology of Mammal-like Reptiles of Suborder Therocephalia. 251 sent forwards by the basisphenoid (basisphenoidal rostrum); the quadrate rami stretch postero-laterally with a straight outer edge (as is typical for the Therocephalia in contradistinction to the Gorgonopsians, where it is curved) to meet the quadrate. The basioccipital (figs. 25-27) forms the well-developed condyle, and, in a fracture, the processes supporting the basisphenoidal tubera are very well shown; dorso-laterally the exoccipitals support the condyle; the medulla rested on its rather irregular grooved dorsal surface. The basisphenoid forms two widely separated tubera and, anteriorly, forms the deep vertical keel flanked by the pterygoids; the basipterygoid processes are horizontal, flat, and immovable; in a sagittal fracture no trace ofa median parasphenoidal process is seen extending dorsally, although both the anterior and posterior borders of the pituitary fossa can be seen; it is, however, possible that ais the anterior partofthe bone Fie. 27.—Theriognathus microps. B.M.N.H., labelled “ basisphenoid” re- eee Lateral nay of the left side of the ; rain-case. x #4. presents a parasphenoid indistinguishably fused with it; in a cross-section it is seen that the basisphenoid forms a broad floor for the hind-brain. The pro-otic is only partly exposed and, as far as can be ascertained, agrees with the condition in Notosollasia; it lies against the par- occipital; antero-dorsally, it forms the ventral border of the venous foramen; the foramen for the seventh nerve apparently lies in the usual position; that for the fifth is obscured by the epipterygoid. The epipterygoid, as in all the Whaitsids, is a large flat element with expanded dorsal and ventral ends; its base on the quadrate ramus of the pterygoid is very long and, as it lies lateral to the cavum eprptericum, obscures the foramen for the fifth nerve, whose nature has thus not been determined. The occiput is depressed; the post-temporal fenestrae are of medium size; the paroccipitals are very massive, with greatly expanded distal ends, which support the auditory ridge formed by the squamosals; they are situated in a plane considerably in advance of the par- occipital processes. 252 Annals of the South African Museum. The lower jaw is of the usual Therocephalian structure, but in shape is curved in the distinctive way seen more highly developed in Notosollasia and Whaitsia. The angular is a strong element with a large exposure externally; the surangular forms the usual curved girder intercalated between the dentary and the articular; the dentary bears no molars; the articular articulates with both quadrate and quadratojugal. A long slender bone, which is probably a ceratohyal, is imbedded in the matrix lying on the right side of the basisphenoidal keel. Chief measurements : Length from premaxilla to the basioccipital condyle . ‘ : . 205? mm. Length from eens to fhe Biviea foramen 150? ,, Length from pineal foramen to the occipital plate : ‘ 30: Length from premaxila to “the font a he orbit : : : : ; . . 90%ee Length of the lower jaw ; . 200 Far Width across the squamosals ‘ : Rc ® Interorbital width t : ; 5 , aoe Intertemporal width . . » AO sae Width across the lateral nicreeidal fanees 80) 25; Width of the snout. : ; . [igs Height of the occiput . 2 : .. 45-7 Type, B.M.N.H., 47065, Stylkrantz, Sneeuwberg, Cape Province. Whartsia major, Broom. (Figs. 28-34.) Five good skulls of this species are known. The type in the British Museum consists of a very good skull, which is, however, slightly distorted; associated are a right radius, ulna, and manus; the latter was described by Broom as a pes; a humerus is catalogued (R5755) as that of Whaitsia, and the nature of the matrix and of the fossilisa- tion is as in the skull; the femur described by Broom is not in the British Museum collection; a snout labelled as Anteosaurus minor is obviously that of another specimen of this species: its number is R5748. Through the courtesy of Dr. W. E. Swinton, Mr. Parsons was asked to do the preliminary clearing of the type skull; I have carried the — EEE = Morphology of Mammal-like Reptiles of Suborder Therocephalia. 253 Fic. 28.—Whaitsia major. B.M.N.H., R5694. Dorsal view of the skull. x 33. depp Dent. Fic. 29.—Whaitsia major. B.M.N.H., R5694. Lateral view of the skull. x j- 254 Annals of the South African Museum. development further so that it now shows the greater part of the detailed structure. The skull is of a large Therocephalian; it is broad and depressed; the intersquamosal width is great and the temporal fossae are, as in all the Whaitsids, thus very wide, but short; the posterior part of the mandibular ramus 1s very wide, but in anterior direction it curves inwards and forms a narrow, though long, symphysis. The dental formula is 1. 5, ce. 1, m.0; asin Notosollasia the incisors carry longitudinal grooves. The structure of the outer surface is best understood by referring to the figures of the dorsal and lateral surfaces; the parietal crest is thin and fairly high; the pineal foramen is a narrow slit; the frontal is broad, but short, with only a small entry into the supra- orbital border; there is no postfrontal; the dorsal and anterior borders of the orbit are raised into a sharp ridge, which is interrupted above the lacrimo-prefrontal suture; the lacrimal and prefrontal form a deep anterior face to the orbit; there is a small lacrimal foramen; the facial exposure of the septomaxilla is fairly large, and it has two distinctive spurs projecting into the nostril; the two depressions along the maxillo-septomaxillary suture, usually found in Therocephalians, are very well shown. The occiput is exposed exceedingly well; the basioccipital condyle is strong and kidney-shaped; it sends down two strong processes to support the basisphenoidal tubera; the exoccipitals support its corners without actually coming on to the articulatory surface, and form quite strong processes overhanging the large jugular foramina; the paroccipital processes are strong; the post-temporal fenestrae are of medium size, and are situated low down; the foramen magnum is small; the tabular is a large bone forming a considerable part of the post-temporal edge; the quadrate and quadratojugal le well below the level of the condyle. The palate has its posterior half well exposed in the type, where it is, however, distorted; the anterior half is beautifully shown in the duplicate specimen—R5748, and the accompanying figure is a composite one based on both specimens. In the anterior half the important point to notice is the bipartite anterior fenestra; the anterior part received the lower canine, and the posterior functioned as internal nares, posterior to which the palatines form a deep naso- phalangeal passage; the partition is brought about by a process of the maxilla growing in dorso-median direction and then overlying a ventro-laterally directed process sent out by the prevomer. In Morphology of Mammal-like Reptiles of Suborder Therocephalia. 255 Fig. 30.—Whaitsia major. B.M.N.H., R5694. Ventral view of the skull. x 45. } eres | iH an ) Pe X BOc. BNph. Exe St Fic. 31.—Whaitsia major. B.M.N.H., R5694. Occipital view of the glaullls 8 Bs 7 256 Annals of the South African Museum. Moschorhinus the swelling of the maxilla inwards has been mentioned, but here there is no outgrowth from the prevomer; in Notosollasia the maxilla grows inwards and the prevomer has a lateral process, which meets it so that the vacuity is bipartite, but here the process of the prevomer is in a plane considerably dorsal to the maxillary swelling; in Whzitsia the maxillary and prevomerine processes lie in the same plane, which les at an angle of about 45° to the horizontal. In this character these three forms form a morphological series, which is, however, certainly not a direct ancestral one. Posterior to the . ie <7 aA\e SE Per — PSph. Fit. ies Fie. 32.—Whaitsia major. B.M.N.H., R5694. Lateral view of the right side of the brain-case. x 3. internal nares, the prevomers expand in shovel-shaped fashion and underlie the palatines and pterygoids. On the anterior ramus of the pterygoid there are two large and deep depressions; I can offer no suggestion as to their significance. There is no suborbital vacuity. The ectopterygoids, as in Notosollasia and Theriognathus, form the corners of the transverse pterygoidal bar, whereas in Moschorhinus it is still formed by the pterygoid itself. Posteriorly, the pterygoids form a deep keel in the median line, and here clasp the anterior pro- longation of the basisphenoid, which here may have a parasphenoid fused on to its ventral surface. The quadrate ramus has a straight outer edge and is applied to the anterior face of the quadrate. The Morphology of Mammal-lke Reptiles of Suborder Therocephalia. 257 basisphenoid is not a very strong bone and its tubera are, for their greater part, formed by the overlying basioccipital processes; it is, however, possible that the bone forming the ventral surface of the tubera is really a parasphenoid, and that the processes supporting it are really basisphenoidal, and not basioccipital; the sagittal fracture in Theriognathus, however, shows a good suture between the basi- occipital and basisphenoid, with nothing to indicate that the ventral part of the basisphenoid is really an underlying parasphenoid very closely applied to the basisphenoid proper. In ventral view it is seen that the basisphenoid swells out and meets the pro-otic, which then continues dorsally. The quadrate forms the main articulatory Spl cart PB Are. Ang Fic. 33.—Whaittsia major. B.M.N.H., R5694. Inner view of the right mandible. x 4. surface, and the quadratojugal only the lateral corner, which, in lateral view, projects below the level of the squamosals. The brain-case is beautifully shown in lateral view on the right side; the epipterygoid is a large bone with greatly expanded ends and a constricted waist; it lies in a plane lateral to the pro-otic, and encloses a large cavum epiptericum; the pro-otic is well developed; ventrally, it meets the basisphenoid in a suture, which runs from the border of the fenestra ovalis to the lateral border of the pituitary fossa; dorsally, it meets a flange of the interparietal and, posteriorly, is applied to the paroccipital; it has a strong anterior process, which separates the lateral opening into the pituitary fossa from the opening for the fifth nerve; it forms part of the anterior border of the fenestra ovalis and, antero-dorsally to this, is pierced by a foramen for the seventh nerve; a posterior pro-otic process separates the foramen for the fifth nerve from the large venousforamen. From this structure 258 Annals of the South African Museum. of the lateral wall it is clear that the posterior part of the brain-case is fairly wide and not very high, with a mass of bone below its ventral A floor; its anterior extent is not great, and it housed only the hind-brain; the mid-brain and thalamen- cephalon lay median to the epipterygoid, and the whole cerebrum must have been housed by the ethmoidal part, which, in this speci- men, has not been exposed. The lower jaw is char- acterised by the great width across the condyles and then, anteriorly, by the in- wardly curved sweep to the long and narrow symphysis; there do not appear to be any incisors, and there are no indications of molars whatsoever; the splenial is very intimately fused to the dentary; the coronoid is small, and the surangular strong; the angular is large, but extremely thin. ee The fore-limb is repre- “sented by a good right humerus, which may be of the same animal as the right ulna, radius, and manus. The humerus is of Fic. 34.—Whaitsia major. a type distinct from that A. Dorsal view of the right fore-foot. known in any other Therap- B.M.N.H., R5694. x 3. sid; there is a very large cl—c3 =centrals. Ua tilnar : is =intermedium. ws =ulnare. Eons icondy lar, but no Ro) radius: 1-5. =distalia. ectepicondylar foramen; r. =radiale. I-V. =digits. the delto-pectoral crest is very long; distally, it merges into the shaft; it curves inwards and does not lie along the anterior border as in the Gorgonopsians, e.g. Morphology of Mammal-like Reptiles of Suborder Therocephalia. 259 Aelurognathus, Arctognathoides, etc.; the surface for the insertion of the deltoid muscle lies less on the dorsal surface than it does in the Gorgonopsians and Anomodonts; the distal end is greatly ex- panded, the ectepicondyle being particularly strong; the scar of the scapulo-humeralis posterior and subcoraco-scapularis is very strong, and so is the supinator and extensor crest; the large pit for the Fic. 34.—Whaittsia major. B. Ventral view of the right humerus. B.M.N.H., R5755. x C. Dorsal view of the right humerus. B.M.N.H., R5755. x ent. =entepicondyle. to be Ser brachialis indicates a strong muscle of that name; the intertrochanteric fossa 1s deep, and the coraco-brachialis must have been strong. The ulna is a very peculiar bone; there is no olecranon process, but the proximal end is greatly expanded preaxially in a manner unknown in any other Therapsid; the bone is somewhat flattened, and this expansion may represent a sigmoid process with a sigmoid notch into which the radius fitted. The radius has no distinctive characters; it is slightly shorter than 260 Annals of the South African Museum. the ulna; as the humerus is also short, the whole fore-limb was short, and the animal could thus not have been very agile. The carpus is not fully preserved, the intermedium and the greater part of the ulnare being lost; the radiale is a strong, oval-shaped bone, to which, distally, two bones are articulated; the preaxial one is undoubtedly a composite bone consisting of the fused first and second centralia; the central one is the third central; there are four distals; the postaxial one, which articulates with the fourth and fifth meta- carpals, represents the fused fourth and fifth distals; the meta- carpals have peculiar mammal-like epiphysial ends, also encountered in the Gorgonopsian, Aelurognathus; the phalanges have very well- modelled articulatory surfaces with a much better finish than those of Aelurognathus. The carpal formula is thus 3, 3, 4, and the phal- angeal 2, 3, 3, 3, 3, as in Cynodonts and primitive Mammals, and not 2, 3, 4, 4, 3, as in Aelurognathus. Chief skull measurements : Length from premaxilla to the basioccipital condyle . 305 mm. Length from premaxilla to the pineal foramen A .. BOOT Length from premaxilla to the occipital plate. : Ue Length from premaxilla to the front of the orbit . -: PROT Length of the lower jaw : : ~ Zoom Width across the squamosals : ; > 4 2p0' a Interorbital width : : : . : . : 65-5 Intertemporal width . : : : ; : ; 1: = Width of the snout. : : , SD os Width across the lateral Meee oid dunes : 507 Height of the snout . : : : ‘ F ; ro 6 hae Height of the occiput . : : : Ue Length of the incisor series . : Say Type, B.M.N.H., R5694, Thaba ’Nehu, One cene Free State. Tigrisuchus siumus, Owen. This very unsatisfactory type consists of a weathered snout, which shows the roots of three large incisors and a large canine on both sides of the upperjaw. This snout has always been considered to represent a species of Gorgonopsian, but recently Broom has recognised it to be a Therocephalian. In comparing it to the Whaitsid skulls in the collection, I was impressed by the similarity in the broad depressed snout, the general shape of the nostril and the septomaxillary, and the broad expanded anterior end of the prevomers. As there appears Morphology of Mammal-like Reptiles of Suborder Therocephalia. 261 to be no character which would exclude it from the Whaitsid family, this fragment had better be included in this group. The specimen is of some morphological interest in that a fracture shows the detailed relations of the premaxilla, septomaxilla, and the prevomers. Type, B.M.N.H., R1721, Stylkrantz, Graaff-Reinet 2, Cape Province. The family Huchambersidae is instituted for the remarkable skull described by Broom as Euchambersia mirabilis. The possession of the following characters clearly distinguish this new family from the other Therocephalian families, viz. jugal bar absent; incomplete postorbital bar; the presence of a very remarkable preorbital de- pression, which leads by a notch into the buccal cavity; the canine has a sharp ridge on its labial surface; there is no maxillary border posterior on the canine, and thus there are no molars. Euchambersia mirabilis, Broom. (Fig. 35.) This remarkable skull was first described by Broom in 1931, and figured in 1932. The skull, though crushed in an oblique direction, is fairly completely preserved; the squamosals, quadratojugals and quadrates have been lost, and the occiput cannot be satisfactorily cleared. The skull is unlike that of any other Therapsid, but the essential points of structure in the palate, dorsal and lateral surfaces con- clusively show that the animal is an aberrant Therocephalian. There are a number of remarkable specialisations, viz. the postorbital and infraorbital bars are incomplete; the infratemporal bar is absent, though it is likely that the squamosal formed an incomplete bar; there is a very large and deep preorbital depression lying in the external surface of the maxilla and the lacrymal; a notch leads from this cavity to the buccal cavity; there are no molars; the canine is of medium size, and has a unique ridge on its labial surface; no incisors are preserved, but, according to Broom, five were present. Broom has suggested that the preorbital depression housed a huge par-otic gland, and Nopcsa suggested that the gland was poisonous, and was connected with the ridge on the canine. Broom’s suggestion is untenable, as the depression is preorbital, and the par-otic gland is generally postorbitally situated. It seems more likely that this depression housed an enlarged labial or lacrymal gland. At this fen. Ov Fie. 35.—Euchambersia mirabilis. B.M.N.H., R5696. x 2. A. Dorsal view of the skull. B. Ventral e C. Lateral a 29 29 Morphology of Mammal-like Reptiles of Suborder Therocephalia. 263 point it is of interest to recall the fact that in the Gorgonopsians and Therocephalians there is generally a shallow preorbital depres- sion, which probably housed a gland of some kind; the gland in Euchambersia may be the result of the hypertrophy of such a normal supralabial or lacrymal gland. On the outer surface the important points are: the very large prefrontal, which forms the whole supraorbital border; the small frontal with no entry into the orbital border; the absence of a post- frontal; a minute pineal foramen; the rudimentary jugal forming a little of the preorbital border, but not extending backwards to meet the squamosal or postorbital; the short maxillary border, which does not meet the jugal; the large septomaxilla. The palate is essentially Therocephalian; the anterior palatal vacuities are unipartite and relatively long; the prevomers are, anteriorly, greatly expanded and, posteriorly, are shovel-shaped, and underlie the pterygoids and palatines; the ectopterygoid does not reach the corner of the lateral pterygoidal bar; the median vertical flanges of the posterior pterygoid ramus do not meet along the median line, and laterally flank the basisphenoidal rostrum; if my interpretation, that the basioccipital sends two strong processes ventrally, is correct, then the basisphenoid is a thin bone underlying these processes; on the other hand, if these processes are actually the basisphenoidal tubera, then the thin bone must be a parasphenoid ; I find no evidence for the latter view; the paroccipital bar is strong, and is distally grooved to receive the quadrate; dorsal to this the supraoccipital sends a thickened process laterally, and to this a tabular must have been attached; actually, no tabular is, however, preserved; the basioccipital condyle is fairly strong. The epipterygoid is a moderately slender element, as in the Pristerognathids, Scylacosaurus, and Scymnosaurus; the pro-otic is not sufficiently exposed to merit a description; the fenestra ovalis is large; there is a great depth of bone ventral to the foramen magnum. Euchambersia is thus a Therocephalian, which can be derived from an unspecialised Pristerognathid; its evolutionary direction has been along a line quite distinct from that of the Lycideopsids, Whaitsids, and Scaloposaurids; in the main its palate is quite primitive, and so is the epipterygoid; but the extreme specialisation in the occiput, the loss of the postorbital and jugal bar, the development of the peculiar preorbital depression, and the apparently correlated notch leading on to the ridged canine, are all characters with no parallel 264 Annals of the South African Museum. in any of the other Therocephalian families. It is of interest to recall that an analogous loss of postorbital and infratemporal bar connected with the development of a poison fang occurs in the Ophidia. Chief measurements : Length from the premaxilla to the basioccipital condyle . 120 mm. Length from the premaxilla to the front of the orbit . | Sao Interorbital width : : : : ; . 28a Intertemporal width . : : : : . ieee Width of the snout : ; : i ae Width across the lateral sae antes . ee Height of the snout. : ee Height of the occiput . : : : oaees Depth of bone below the foramen magnum . : Zee Type, B.M.N.H., R5696, Norval’s Pont, Cape Province. In the family Scaloposauridae seven genera have been included, viz. Akidnognathus, Choerosaurus, Icticephalus, Ictidognathus, Ictido- stoma, Scaloposaurus, and Simorhinella. Three of these are repre- sented in the collection of the British Museum, and they have recently formed the basis of an excellent description by Watson; with the exception of a few supplementary remarks on Icticephalus, it is thus unnecessary to go into any details. The family is of considerable interest in that it ranges throughout Lower Beaufort times without showing any very marked evolutionary changes; Icticephalus of the Tapinocephalus zone may even be cogeneric with Scaloposaurus of the Custecephalus zone. Although occurring as early as the Tapinocephalus zone, the Scaloposaurids are, in a number of points of structure, definitely distinct from any of the other Therocephalian families. The Whaitsids, Lycideopsids, and Euchambersids represent three lines of development which are completely divergent from that of the Scaloposaurids. The Pris- terognathids are apparently a retarded stock in which a number of primitive features are retained; in one form, Scylacosaurus, the nature of the basisphenoid suggests a development parallel to that of the Scaloposaurids. In the Ictidosuchids, Ictidosuchoides has been shown to possess some characters also met with in the Scalopo- saurids. It would thus appear that the Therocephalians of the Lower Beaufort represent the end branches of a phylogenetic tree, whose main stem must be sought for in pre-Beaufort rocks. Morphology of Mammal-lhke Reptiles of Suborder Therocephalia. 265 Icticephalus polycynodon, Broom. In his original description Broom made the following observations : dental formula, 1.6, c.3,m.11; all the teeth are unserrated; this new type differs from Scaloposaurus in having eleven, and not nine, molars, and in that the postorbital arch meets the jugal. Watson, in 1931, gave a much fuller account, and with this I am in agreement. I have, however, found that something more can be determined of the dentition. Although very badly preserved, Broom is probably right in assuming the incisors to be six in number; there is a large canine, with one tooth anterior and another posterior to it—all on the maxilla; then there is a diastema of 2 mm.; this is followed by nine molars, although between the fifth and sixth there is an interspace, which may have housed another tooth; the molars are small and progressively decrease in size in posterior direction. In the British Museum specimen the dental formula would thus be, i. 6, c. 3, m. 9-10, which is thus not very different from that of Scalopo- saurus. If Broom’s supposed distinguishing character in the dentition is thus invalid, the only other distinguishing character is the nature of the postorbital bar, and in neither of these two specimens is the nature of the postorbital bar known with absolute certainty. Watson has at great length drawn attention to the great similarity of these two skulls in nearly all, even the most detailed, points of structure. So that one can really only maintain that the type Icticephalus of the Tapinocephalus zone is represented in the Cistecephalus zone by a practically identical animal, Scaloposaurus. The older Icticephalus has a slightly larger skull, with its snout higher and thus relatively less broad; the orbits look as much outwards as upwards; the mentum of the lower jaw is fairly deep. In the younger Scalopo- saurus the skull is somewhat smaller; the whole skull, including the snout, is somewhat flattened, so that the snout is relatively broader than high; the orbits look more upwards than outwards; the mentum of the lower jaw is less deep and more sloping. Viewed from in front, Scaloposaurus has a more pointed snout and appears rounded in section, whereas in Icticephalus it is squarish. Type, B.M.N.H., R4096, Weltevreden, Prince Albert, Cape Province. Chief measurements: Scaloposaurus. Icticephalus. Premaxilla to basisphenoid . 50% mm. 65 mm. Premaxilla to pineal foramen . a D9 +55 VO “xX. PART 2. 23 266 _ Annals of the South African Museum. Scaloposaurus. Icticephalus. Premaxilla to front of orbit . 282% mm. 34 mm. Interorbital width . Eo OES ae IZ er Intertemporal width oT cee ee Width of the snout . isis 16.49 Width across lateral ieereoid flanges. : re OTs oe 2 Height of the Laois Kees 83 Lo Height of mentum of lowerjaw. 7? ,, Ow Length of the molar series : ee Lane SUMMARY. 1. The palate is figured and described in three Pristerognathids, one Ictidosuchid, four Whaitsids, and one Euchambersid; it can thus be claimed that the main morphological features of this region in the Therocephalians have been established. In the more generalised Pristerognathids, in the slightly more advanced Ictidosuchids, and in the specialised Euchambersids and Scaloposaurids, the palate is comparable in general plan to that of the less advanced Deino- cephalians and Pelycosaurs; even in the very specialised Whaitsids, the underlying ground plan is still the same. The Therocephalian palate differs from that of the Gorgonopsians, which occupy the same developmental niveau, but on a parallel plane, in the number of points mentioned; the chief being the nature of the prevomers. 2. The occiput is described and figured in one Pristerognathid and in two Whaitsids; it differs mainly from that of the Gorgonopsians in the very massive paroccipital bar and the large post-temporal fenestra (except in Scymnosaurus); in the Scaloposaurids a special process is developed on the paroccipital. 3. Parts of the brain-case are figured and described in two Pristero- gnathids, in one Euchambersid, and in three Whaitsids; it has become evident that in the Therocephalians there has, in respect of the epipterygoid, occurred a development, which has in some respects been paralleled by the Cynodonts. 4. The description of a complete fore-limb of Whaitsia adds to our knowledge of the postcranial skeleton of the Therocephalians, about which lamentably little is as yet known. 5. In the course of this paper it has been shown that the seven families, into which the suborder has been divided, represent distinct evolutionary trends, and that certain similarities in structure do not necessarily imply any close direct relationship, but may simply have Morphology of Mammal-like Reptiles of Suborder Therocephalia. 267 resulted as parallel developments from an original stock possess- ing common characters and_ potentialities. The Whaitsids, Kuchambersids, and Lycideopsids are families possessing some very -definite characters, separating them from each other and from the Scaloposaurids, Alopecopsids, Ictidosuchids, and Pristerognathids; the Scaloposaurids are also quite distinct, but may be a composite family, as there is a discrepancy in one character, viz. in some the postfrontal is present, whereas in others it is absent; finally, the Pristerognathids, although the most generalised group, does contain some forms which possess quite advanced characters, e.g. the epipterygoid of Trochosaurus. Some subdivision, as, for example, that of Williston, will in time have to be considered seriously. In conclusion, my thanks are due to the officers of the British Museum (Natural History), in particular to Dr. W. E. Swinton, for affording me the facilities I have enjoyed in my study of the material in their charge. Professor D. M. 8. Watson’s critical interest has been very encouraging. To the University of Stellenbosch I am indebted for a small grant, which has enabled me to come to London. My wife has been responsible for the execution of the greater number of the figures which illustrate this paper. ' 2 ra § , ‘ re Dal ‘ le fig , gp #* 7 5 t ¥ j ; Te , GI ie J Lyi ) ‘ roe : = io a j ‘ } j ' : 4 i + y | ee [ 4 \ ‘ ~ Heo 7) wot SAT ad Mura) ise mit Taya [ } ipa aa Pala oe - el si) tuyobeltd Oy had ae ‘Realise coat on Lak Give Pn ree, wil alnrpcendabv imate’ ia ot” poten Pee mre net PB LOY fi vat ee ‘ebsdad thiol iat bothont its oie 10st yh ethentis halle seal wi, a H yeul be ae seaides ee a , ieee tee VARA peg vey Ave Lah Te nik ea Arh As Od Die joan wt aah CWT ACL od Helo a here Lapwta ayotis | em PARTS OF THE ANNALS PREVIOUSLY ‘ISSUED Pe ee ee ee ae Vol. L—Part. 1, out of print; Part 2, out of print; Part 3, out of. Smee Vol. 11.—Part 1, out of print ; Part 2, 5/6; Part 3, out of print; Part 4, 3/-5_ d | Part 5, 1/6; Part 6,-3/—; Part 7, 1/6; Part’ 8, 3/—; Part 9, 1/6 Ree Part 10,7,—; Part lL, 3/-; Fader Title, etc., 1/6. : % | Vol. I11.—Part 1, out of print ; Port 2,.1/6; Part 3, 5/6; Part 4, 3/-3 Sees oS Part 5, 5/6; Part 6, 7/—; Part 7, 1/6; Part” 8, 3/-; Fart 9, 1/6; MF; ; Index, Title, etc., 1/6. BPS eee Vol. IV (containing Palacomtblocial papers published in conjunction with the 4 ee Geological Survey).— Part 1, L1/—; Part 2, 7/-; Part 3, 4/6; Part 4, 4/6; Part 5, 2/6; Part 6, 4/6; Part 7, 14/-; Part 8, 8/-— Fae Vol. V.—Part 1, 4/6; Part 2, 8/6; Part 3, 2/6; Part 4, 1/6; Part 5,2/-; — ~ el “Part 6, 5/—; Part 7, 3/-; Part 8, 4/6; Fart 9, 4/6; Be pee as pine Title, etc., 1/6. ‘ ie S/R ae a Vol. VI.—Part 1, 13/6; Part 2, 4/6; Part 3, 3/6; Part 4, 30/-; Index, Title, ete., 1/6. 2 + Vol. VIL (containimg Palaeontological papers published in conjunction with the Geological Survey ).— Part 1, 3;-; Part 2, out of print; Part 3, 5/—;- Part 4, 8/- 5 Part 5, 5/6; Part 6, 1/6; Index, Title, ete., 1/6. . Vol. Vill.—Complete, out of print. Index, Title, etc., 1/6. Vol. 1X.—Part 1, 4/6; Part 2,.5/6; Part 3, 10/-; Part 4, 6/6; Part 5, 3/6; | Part’ 6, 11/-; Part 7 , 9/-; Index, Title, etc., 1/6. : | Vol. X.—Part 1, 3/-; Part 2, 2/6 ; Part 3, 2/-; Part 4; 3/—; Part 5, 20/-;. Part 6, 3/—; Part 7, 10/-; Part 8, 2/6; Part 9, 5/-; Part 10, 2/6: Part 11, 20/-; Part 12, 7/-. Complete. ‘Title, etc., contained in Partstl? . Vol. XIl.—Part |, 3/6; Part 2, 2/-; Part 3, 18/6; Part 4, 1/6; Part 5, 17/-; Part 6, 11/- ; Index, ‘Title, etc., and Plate LI, 3/-. is <2 Vol. X11 (containing Palaeontological papers published in conjunction with the ; Geological Survey).— | eS Part 1, 15/6; Part 2, 3/6; Part 3, 4/6; Part4, 3/-; Part 5, 7/-; BOS ~ io Part 6, 6/-; Part 7, 20/- ; Part 8, 20/—; Index, Title, etc., 1/6. Ba Bay é : Vol. XI11.—Part. 1, 6/—; Part 2, 2/6; Part 3, 3/—; Part 4, 8/6; Part 5, 1/6; Part 6, 5/-; Part 7, 30/-; Part 8, 1/-; Index, Title, etc., 1/6. Vol. X1V.—Part 1, 8/6; Part 2, 8/-; Part 3, 6/-; Part 4, 21/-; Part 5, 5/-; Part 6, 9/—: Index, Title, ete., 1/6. Vol. XV.—Part 1, 17/-; Part 2,17/-; Part 3, 14/-; Part 4, 12/6; Part 5, 5/6; Part 6, 3/6; Index, Title, ete., 1/6. Vol. XV!.—Part 1, 30/6; Part 2, 4/-; Part 3 (with Title, Index, etc.) ), 25/-. Vol. XVII.—Part 1, 12/-; Part 2,9/6; Part 3,3/—; Part 4,17; Part 5, 17/-; Part 6, 2/6; Index, Title, etc., 1/6. | Vol. XVITI.—Part 1, 20/-; Part 2, 7/6; Part 3, 30/-; Part 4, 12/6; Index, | Title, ete., 1/6. | Vol. XIX.—Part 1, 29/_: Part 2, 17/6; Part 3, 11/-; Part 4, 5/6; Index, Title, Nee 1/6. | Vol. XX.—Part 1, 8/6; Part 2, 12/6; Part 3, aj Part 4, 10/—; Part 5, 4/-;. | _ Part 6 (with Title, etc.), 4/6. | : Vol. XXI.—Part 1, 25/-; Part 2 (with Title, etc.), 30/-. Vol. XXII (containing Palaeontological papers published in conjunction with the Geological | Survey ).— Part 1, 20/-; Part 2, 10/-; Part 3 (with Title, ete.), 3/6. : Vol. XXIIT.—Part 1, 12/6: Part 2, 8/—; Part 3 (with Index, Title. etc.), 9/-. Vol. XXIV.—Part 1, 10/-. Vol. XXV.—Part 1, 12/6; Part 2, 10/-; Part 3 (with Index, Title, ete.), 10/-. Vol. XXVI.—Complete, 25/-. Vol. XXVITI.—Complete, 25/-. Vol. XXVITI (containing Palaeontological papers -published i in conjunction with the Geological Survey).—- . Part. 1, 10/-; Part 2, 17/6; Part 3 » 6/5 Part 4 (with inde Title, a. 10/-. Vol. XXIX.-—Part 1, 20/-: Part 2 (with Index, “Title, etc.), zu Vol XXX.—Part 1, 15/6; Part 2, 20/—; Part 3, 8/-. Vol. XXXI.—Part 1, 20/-; Part 2, 13/6. The Annals of the South A frican Museum will be issued at hod intervals, as matter for publication ts available. Copies may be obtained from— Messrs. WHELDON & WESLEY, Lop., 2, 3, and 4 ArTHuR STREET, NEW Gancan STREET, Lonpon, W.C..2;- or, The LIBRARIAN, Souts Arrican Musrum, Carr Town. “Py ~ bins tie) * * _ a omy o y : ‘, 7. r ANNALS AFRICAN MUSEUM VOLUME XXXI. 5 Pac ocecln a New Genus of Cystids from the Lower evonian of South Africa. By Joan V. L. Rennie, M.A., )., Department of Geology, Rhodes University College, IT unstown, South Africa. aie Plate XXXVI.) CEL ch : PRINTED | oe THE ES OF THE SOUTH AFRICAN MUSEUM _, AND THE ( 269 ) 8. On Placocystella, a New Genus of Cystids from the Lower Devonian of South Africa.—By Joun V. L. Renntz, M.A., Ph.D., Depart- ment of Geology, Rhodes University College, Grahamstown, South Africa. (With Plate XXXVI.) The class Cystoidea was recorded for the first time from South Africa in 1925, when Dr. F. R. C. Reed described a single specimen from the Bokkeveld Series (Lower Devonian), collected by him at a roadside cutting near Bufielskraal, between De Doorns and Tunnel Siding in the Hex River Pass. The specimen was named Placocystis africanus sp. nov. (Reed, 1925, p. 30, pl. iv, fig. 1) and was compared with the Silurian (Wenlock) species Placocystis forbescanus de Koninck (see Bather, 1900, part iii, p. 51, fig. xiii) and other representatives of the Anomalocystidae, a primitive but specialised family of echino- derms rare in post-Silurian deposits. In 1928 Dr. L. D. Boonstra and the writer collected two specimens of a cystid in the First or Basal Shales of the Bokkeveld Series at the northern entrance of Gamka Poort in the division of Prince Albert. The writer was able to make an examination of Reed’s specimen in the Sedgwick Museum in Cambridge, and to compare directly with it the two specimens here described. The writer is indebted to the Research Grant Board of the Union of South Africa for financial assistance and wishes to express his thanks to Dr. John Hewitt, Director of the Albany Museum in Grahamstown, for the use of camera and dark room. Placocystella capensis gen. et sp. nov. The two specimens from Gamka Poort, here described as Placo- cystella capensis gen. et sp. nov,* are preserved as moulds, coated with reddish-yellow ferruginous matter, in a greenish micaceous mudstone. In each case what is presumed to be the convex external * This paper was read at a meeting of the Royal Society of South Africa in 1932, and the generic name appeared in the title of the paper published in the minutes of proceedings in 1934 (Trans. Roy. Soc. 8. Afr., xxii, p. lvi). The name Placo- cystella Rennie 1934 is a nom. nud. VOl, XXXI, PART 3. 24 270 Annals of the South African Museum. mould of the concave or right side of the theca is shown, the edge of the convex surface on the hand specimens being reflexed upwards if preserved at all; in the paratype (S.A.M. 9700) the adcolumnal margin is also reflexed upwards, but perhaps as the result of fracture; in the holotype (S.A.M. 9701) the lower margins of the two ad- columnal plates curve downwards, this feature apparently being an original one, and supporting the interpretation of the specimens as external moulds. The preservation is relatively superior, though the column and brachioles are not attached in either case. There is some indication, shown by irregularities of the surface, of shght relative movement of plates subsequent to the death of the animal, but on the whole there has been little compression of the thecae, the majority of the plates being in position and the sutures being very clearly marked. A number of the sutures are preserved as strong raised ridges, such as could have been formed by infilling; the fact that they are raised would also indicate that the specimens are external moulds. The Holotype of Placocystella (Plate XXXVI, figs. 4-6).—The arrangement of the plates is more distinct in one of the above (S.A.M. 9701) than in the other, and the specimen is accordingly taken as holotype. The moulds of the plates are arranged on a gently convex surface. The periphery is intact except along the adoral margin. The theca is subquadrate in outline, somewhat wider towards the columnal end than towards the oral end, the widest part situated ata little less than one-third of the length of the theca from the columnal end. The theca is distinctly longer than wide. The lateral margins are gently convex, but straightened towards the oral end. The adoral margin is not preserved. The adcolumnal margin is concave for the attachment of the stem. The margin of the lower left lateral marginal plate is abruptly reflexed upwards, the other lateral marginal plates not showing this feature. The concave side of the theca appears to have been composed of twelve plates, ten of them marginals, with two somatic plates en- closed by them, the twelve arranged symmetrically about a line drawn from front to back. At the columnal end are two large marginals, meeting along the middle line, truncate above against the larger somatic plate, in length somewhat less than one-third of the length of the theca, the sutures between them and the adjacent marginals reaching the lower angles of the theca; towards the lower angles of the theca these plates are truncate, while the columnal margin is fairly deeply embayed for the attachment of the stem, On Placocystella, a New Genus of Cystids. 271 with the edges of the plates curved downwards within the embayment. On each side are three marginals, subequal, the lower pair sub- triangular, the middle and upper pairs subquadrate. At the oral end two smaller marginals meet along the middle line; the one on the right side meets the uppermost right lateral in a suture which is apparently directed towards the right upper angle of the theca, the corresponding suture on the left side being only visible in part. Centrally placed is a large subquadrate somatic plate, bounded below by the adcolumnal marginals and laterally by the two lower pairs of lateral marginals. The second somatic plate is considerably smaller than the other, is subquadrate, and placed along the middle line. The sutures between the plates, as distinguished above, are for the most part visible as straight ridges or lines of whose nature there can be no doubt. This applies particularly to the sutures between the marginals. The sutures bounding the somatic plates are some- what disturbed, owing probably to a slight relative movement of the plates, but their positions can be determined with some certainty. A few of the plates are traversed by ridges, which are not regarded as sutures, since (i) they bear little or no relation to the symmetry of the theca, (11) they are in some cases much less evident than the sutures, (iii) they often pursue an irregular course across the plates, and (iv) similar features in the paratype are differently placed. Hach adcolumnal marginal is traversed by a faint ridge in the anterior-posterior direction, but the position of that on the right is quite different from that on the left, and that on the right is curved in an irregular manner. A similar ridge pursues a somewhat irregular course across the lower corner of the right adcolumnal plate and is continued across the lowermost lateral marginal roughly parallel to its outer margin, almost certainly a fracture from its appearance under the binocular microscope; a ridge is present in nearly the same position on the corresponding marginals of the left side. The smaller somatic plate is traversed by two faint ridges, one curved and almost along the middle line, the other across the upper left- hand corner. The remaining plates are free of these features. The larger somatic plate has a broad raised band placed obliquely across it. The three marginals on either side are traversed by irregular, relatively broad, wavy, or discontinuous bands placed obliquely to the outer margin of the theca. These may have been original features of the theca. 272 Annals of the South African Museum. The stem is absent, as well as the brachioles. The margin of the lowermost lateral marginal on the left side is curved upwards, 2.e. there is preserved a fragment of that part of the plate which bounds the rim of the theca; the curved fragment is demarcated by a groove (presumably secondary) and traversed by wavy bands. Dimensions: length 165 mm.; greatest width 13-5 mm.; length of adoral margin about 10 mm. The Paratype of Placocystella (Plate XXXVI, figs. 7—-9).—The second specimen from the same locality (S.A.M. 9700) agrees with the first in dimensions and structure and is taken as paratype of the species. The moulds of the plates are arranged on a gently convex surface and the periphery is well preserved. The outline agrees completely with that of the holotype, with the addition that the adoral margin is very slightly convex and the upper corners of the theca are rounded angles of about 115°. The sutures are for the most part well marked. The sutures between the marginal plates are clearly marked on both sides and correspond exactly with those of the holotype. At the oral end there are only two marginals, that on the left being clearly demarcated by a suture along the middle line and a suture joining the periphery just behind the left upper corner of the theca, the corresponding suture on the right being less defined. The suture between the large central somatic plate and the adcolumnal marginals is only partly seen, and that between the adcolumnals is not well marked. The suture between the two somatic plates is very evident. The theca is in various places traversed by faint incisions or low well- defined ridges, or broader bands, all of which are probably secondary features; noteworthy is the fact that none of these features corre- spond exactly in position with similar markings on the plates of the holotype. The margins of the lateral marginal plates are curved upwards, i.e. there is preserved the cast of the rim of the theca on either side. The lateral marginal plates are traversed by wavy bands roughly at. right angles to the periphery of the theca. In places these appear as short crescentic lines, sometimes continuous with one another. The stem is absent; the adcolumnal margin is reflexed upwards, but apparently as the result of fracture. The brachioles are not present in position, but there is a cast of a cylindrical spine-like body, 8 mm. long, emerging from beneath the theca on the right side. * Dimensions: length 17 mm.; greatest width 13-5 mm.; length of adoral margin 9 mm. On Placocystella, a New Genus of Cystids. 273 Affinities.—The specimens described above are similar to but somewhat larger than the type of Placocystis africanus Reed (1925, p. 30, pl. iv, fig. 1) from the same formation, from a locality about 100 miles distant. In the latter the column is present in position, and what is apparently a fragment of a brachiole is preserved near the right anterior corner. The theca has much the same proportions as Placocystella capensis, but the arrangement of the plates, as in- terpreted by Reed, is entirely different from that of the corresponding surface of either the new species here described or of Placocystis forbesianus de Koninck. Moreover, Placocystella capensis differs from de Koninck’s genotype in features of sufficient importance to warrant the erection of a new genus. The holotype of Placocystis africanus (see Plate XXXVI, fig. 1) is too poorly preserved to allow of a definite statement as to the re- lationship of that species with the new species here described. The figure published by Reed is a drawing of the test apart from the matrix on which it is situated, enlarged x24, giving the impression of an extraordinarily well-preserved specimen; the sutures in the drawing are very well marked and give the author’s interpretation of the test, which is described in some detail. The writer examined this specimen, which is in the Sedgwick Museum,* and came to the conclusion that only a few of the markings present could be inter- preted as sutures. The specimen is preserved on a piece of mudstone, and has a weathered flaky ferruginous surface. Presumably, from the few sutures present as well as from a comparison with the Gamka Poort specimens, the concave side of the theca is shown, but in a flattened condition. According to Reed this side is made up of some twenty plates arranged for the most part in five transverse series in four vertical rows. The writer came to the conclusion that the arrangement of the plates is to a certain extent very doubtful; that few of the markings accepted as sutures by Reed are either as clearly marked or as regular as his figure indicates; that several of these markings could equally well have been accepted as cracks through the plates or the matrix, or lines due to the partial removal of surface flakes by weathering. Indeed, the whole surface of the thecal portion of the specimen is in a most unsatisfactory condition, rough with ferruginous flakes, and marked by a multitude of relatively coarse and finer lines of a most irregular kind, clearly related to the weather- ing of the surface. There are, however, several relatively straight lines on the surface, which are quite unmistakably sutures, and it is * Recently numbered A. 3044. 274 Annals of the South African Museum. noteworthy that these correspond in position with the sutures observed in the Gamka Poort specimens. Sutures bounding three marginals on the right side are distinct, the plates having the same shapes and relative positions as in the new species. There is a well- marked suture between the right adcolumnal plate and the somatic plate anterior to it. The lowermost of the three right marginals is traversed by a crack which continues into the matrix, accepted as an additional suture by Reed. Indications of portions of a few other sutures are present. The writer was unable to find evidence sug- gesting a fundamental difference in structure between the two Bokkeveld species, but as far as the evidence goes, it would appear not more than probable that they belong to the same genus and possibly to the same species. In view of the very great imperfection of the holotype of Placocystis africanus, and the absence of any satis- factory evidence as to the arrangement of the plates in the middle and anterior part of the concave side of the test, 2.e. over about two- thirds of the exposed surface, the writer has thought fit to propose a new specific name for the specimens from Gamka Poort. Placocystis africanus must certainly be removed from the genus Placocystis on either interpretation of the specimen. Placocystella capensis is distinguished from Placocystis forbesianus chiefly by reason of its almost perfect bilateral symmetry. In P. forbesianus (see Plate XXXVI, fig. 2) there are thirteen plates on the concave side, including three adoral marginals; the larger somatic plate is situated on the anterior half of the theca, and the smaller somatic plate 1s placed towards the left upper corner. The perfect bilateral symmetry in the arrangement of the plates thus distinguishes Placocystella from the Silurian Placocystis as well as from the early Devonian Ateleo- cystis Billings and Anomalocystis Hall (for references see Reed, 1925, p. 31). According to Bather (1900, p. 49) the evidence suggests that the evolution of the family was towards greater bilateral symmetry, and therefore started from the usual sac-like form. Placocystella is thus both the most advanced and one of the last representatives of the Anomalocystidae. On Placocystella, a New Genus of Cystids. 275 REFERENCES. BatuHer, F. A. (1900). Anomalocystidae, in Lankester’s Treatise on Zoology, pt. ii, pp. 49-51. Hatt, J. (1859). Anomalocystites, in Palaeontology of New York, vol. iii, pp. 132, 133, 145, 146. Kirk, E. (1912). “The Structure and Relationships of certain Eleutherozoic Pelmatozoa,”’ Proc. U.S. Nat. Museum, vol. xli, pp. 1-137, pls. i-xi (Anomalo- cystidae, pp. 21-29). REED, F. R. C. (1925). “Revision of the Fauna of the Bokkeveld Beds,” Ann. of the S. African Museum, vol. xxii, pp. 27-225, pls. iv-xi (Placocystis africanus, p. 30, pl. iv, fig. 1). ScHucHERT, C. (1905). “‘On Silurian and Devonic Cystidae and Camarocrinus,”’ Smithsonian Misc. Coll., vol. xlvii, pp. 201-272, pls. xxxiv—xliv (Anomalo- cystidae, pp. 204-208). ScHUCHERT, C. (1913). Anomalocystidae, in Maryland Geological Survey, Lower Devonian, pp. 228, 229. Woopwaprp, H. (1880). ‘‘ Notes on the Anomalocystidae, etc.,’’ Geol. Mag., N.S., dec. 2, vol. vii, pp. 193-201, pl. vi. EXPLANATION OF PLATE XXXVI. Placocystis africanus Reed. 1. Holotype, x 2:5 approx., in the collection of the Sedgwick Museum (A. 3044), re-drawn by Miss E. T. Talbot under the writer’s direction. Placocystis forbesianus de Koninck. 2. Concave side of theca, after Bather, modified. Placocystella capensis gen. et sp. nov. 3. Concave side of theca, showing probable arrangement of plates. 4. Drawing of holotype, natural size. 5. Holotype, 8.A.M. 9701, x 2-6. 6. Another photo of holotype, x 2-9. 7. Drawing of paratype, natural size. 8. Paratype, S.A.M. 9700, x 2-6. 9. Another photo of paratype, x 2-5. : Pins. ‘ 7 yy i Ps... i gts Sy) aa ig a A a Date Ann. S. Afr. Mus., Vol. XXXI. Plate DVO. ANS | L. T. Talbot, del., et J. V. L. R., phot. Neill & Co., Lid. CYSTIDS FROM DEVONIAN OF SOUTH AFRICA. @ 2) 9. Lower Cretaceous Lamellibranchia from Northern Zululand.—By Joun V. L. Renniz, M.A., Ph.D., Department of Geology, Rhodes University College, Grahamstown, South Africa. Together with an Account of the Geology of the Cretaceous Beds and a Preliminary Analysis of the Associated Ammonite Fauna, by S. H. Haveuton, B.A., D.Sc., Director of the Geological Survey of the Union of South Africa. (With Plates XX XVII-LV.) CONTENTS. PAGE PAGE 1. INTRODUCTION : 5 . 277 | 3. PRELIMINARY ANALYSIS OF Historical . d 5 te AMMONITE Fauna (by SRE ene) on ery New Collections from N. 4. Tor LAMELLIBRANCH Fauna . 297 SETS os : ee Note on Certain Uitenhage Acknowledgments and Pre- Species . : 5 . 303 liminary Remarks . . 282 | 5. DESCRIPTION OF THE SPECIES. 304 6. REFERENCES . ; : . 388 2. GEOLOGY OF THE CRETACEOUS 7. TABLE SHOWING DISTRIBUTION Beps (by 8S. H. H.) . . 283 OF SPECIES . : at end 1. INTRODUCTION. Historical.—In South Africa marine deposits of Cretaceous age are well represented, and have provided the material for a formidable list of palaeontological publications. The deposits are all but con- fined to the south-eastern and eastern coastal belt, and occur inter- mittently for a distance of over 700 miles in the Union and extend northwards into Portuguese Hast Africa. General accounts of these deposits have been given in recent years by du Toit (1926, pp. 292- 330), Rogers and Haughton (1929, pp. 143-148), and of the occur- rences in Portuguese territory by de Andrade (1929, pp. 147-165). In the southern part of the Cape Province only the Neocomian is represented, in the Sundays River Beds of the Uitenhage Series, which have yielded a rich fauna, monographed by Kitchin (1908), and including ammonites which have enabled Spath (1930) to deter- _ 278 Annals of the South African Museum. mine the age more precisely as Upper Valanginian. The Neocomian stage has not hitherto been recorded from between the neighbourhood of Algoa Bay and Mogambique, a distance of well over 1600 miles. A few Neocomian species have been reported from the northern part of Portuguese Hast Africa near Mocambique (de Andrade, 1929, p. 158), and the well-known Tendaguru Formation in Tanganyika has yielded a fine Lower Cretaceous fauna, described by Lange (1914), Dietrich (1933), and other workers. Between Algoa Bay and Hast Pondoland, for a distance of about 280 miles, the Cretaceous System is practically absent, and the Karroo and Cape Systems outcrop at the coast. In Hast Pondoland the Umzamba Beds, almost confined to beach outcrops over some 25 miles of coast, including outcrops at Umpenyati in southern Natal, are characterised by an exceptionally rich fauna of Upper Campanian age, described by Woods (1906), Spath (1921 a, 1922), Rennie (1930), and others. The beds here rest directly on the pre- Karroo basement. The same horizon occurs as sub-outcrops farther north at Durban. Cretaceous beds reappear at the coast near Port Durnford, south of Richards Bay and about 80 miles north of Durban, and stretch continuously northwards through Zululand for about 150 miles, passing into Portuguese territory, where the system is widespread. A considerable portion of the Cretaceous succession is represented in the Cretaceous of Zululand, though much remains to be done before the sequence is thoroughly understood. The beds rest un- conformably on the Karroo System and on Old Granite, for a long distance directly overlying the Lebombo Volcanics, and are practically undisturbed, the beds dipping eastwards towards the coast at very small angles. Fossiliferous outcrops occur at certain favourable localities, the greater part of the area covered by the system being hidden beneath recent deposits. Nearly all the species described up to the present from the Zulu- land Cretaceous come from two areas, south of the Umfolosi River and north-west of Lake St. Lucia respectively, situated about 40 miles apart, the latter some 35 miles south of the big bend of the Pongola River. The Lamellibranchia described below come from localities north of the big bend of the Pongola River, and between that river and the Swaziland border. A rich fauna has been described by Etheridge (1904) and Spath (1921, pp. 224-272), with a few additions by Newton (1909) and Rennie (1930, pp. 236-250), from a limited area just south of the Lower Cretaceous Lamellibranchia from Northern Zululand. 279 Umfolosi River, including the well-known fossiliferous locality of Umkwelane Hill and the neighbouring localities Lake Itesa (Eteza, Isitesa), and “Railway Cutting between Umfolosi Station and Mill Halt.” The work of Spath (ibid., pp. 264-272) has made it clear that the great majority of, if not all, the ammonites are Campanian (and Maestrichtian?), though it is hardly safe to assume the complete absence of Lower Senonian horizons; there is no indication of pre- Senonian species. The beds here are approximately equivalent to the Umzamba Beds in Pondoland, and the transgression of the Senonian over older horizons is evident, for at Umkwelane Hill the Senonian rests directly on Lebombo basalts. A considerable succession of Cretaceous beds is known from the area north-west of Lake St. Lucia, where fossiliferous outcrops occur along the Umsinene River and its tributary the Manuan Creek (Munyuana), which flow into the north-west corner of False Bay, Lake St. Lucia. Marine fossils have been described by Etheridge (1907), Crick (1907), Newton (1909), Spath (1921, pp. 273-309), Rennie (1930, pp. 236-250), van Hoepen (1929, 1931), and Heinz (1930). According to Spath the ammonite evidence proves the presence of the Middle and Upper Albian, the Cenomanian, the Lower Senonian (Coniacian), and the Upper Senonian (Campanian, including Maestrichtian?) horizons. An account of the succession along the Umsinene River, from above its junction with the Manuan Creek to the shore of False Bay, has been given by van Hoepen (1926), who divided the sequence into seven beds, to which names were given as follows, with suggested correlations :— 7. Umzamba Beds . Upper Senonian. 6. Itweba Beds ; . Middle Senonian. 5. Peronceras Beds ; . Lower Senonian. 4. Munyuana Beds . Turonian. 3. Skoenberg Beds ; . Cenomanian. 2. Umsinene Beds . 1. Ndabana Beds | Albian The palaeontological evidence on which these correlations were made has only been published in part by van Hoepen (1929, 1931). The presence of the Turonian has not been recognised by previous workers (Spath, 1921, p. 268; du Toit, 1926, pp. 316, 319), its presumed absence being in conformity with evidence from Angola and else- where; the transgression of the Senonian on to pre-Cretaceous forma- tions at Umkwelane Hill and in Pondoland points to a break within 280 Annals of the South African Museum. the Upper Cretaceous of Zululand which has not yet been disproved. However, van Hoepen (1929 a) insists that there is a practically unbroken sequence from the Aptian to the Senonian, and Heinz (1930) has identified species of Inoceramus from the Umsinene River as Turonian. In a later publication van Hoepen (1929 a, p. 10) correlates the Ndabana Beds as Aptian, though without giving palaeontological evidence. A few Cretaceous species have been recorded from localities in Northern Zululand, between the Manuan Creek and the Portuguese border, but practically nothing was known of the succession in this stretch of country until the visits in 1933 of Mr. H. F. Frommurze and Drs. 8S. H. Haughton and A. W. Rogers, whose collections have afforded material for this paper. Spath (1921, p. 221) has recorded Albian and Senonian ammonites from the Mkusi (Mkuzi) River, and Heinz (1930, p. 685) has recorded from the same area a species of Inoceramus, said to be Upper Turonian; this locality, 20 miles north of the Umsinene, forms an interesting link between the Umsinene and Pongola areas. The writer (Rennie, 1930, p. 239) has placed on record the occurrence of the Albian species Trigonia cricki, Neithea quadricostata, and Veniella etheridge: at the Ingwavuma road drift on the Pongola River, as well as Nordenskjéldia natalensis and Pleuromya africana, Senonian species, at some locality west of the Pongola River. The Cretaceous of Zululand is continuous with that of the district of Lourengo Marques in Portuguese Hast Africa, where outcrops occur from the Zululand border northwards as far as the neighbourhood of the point of entry of the Limpopo River into Portuguese territory, a distance of about 300 miles. Outcrops occur on the Maputo (Usutu) River near Catuane, close to the Zululand border, and between Bela Vista and the east coast to the south of Delagoa Bay; Cretaceous beds are known to extend northwards along the eastern flank of the Lebombo Mountains as far as the Limpopo, where they rest on the Lebombo Volcanics and dip eastwards at small angles beneath recent deposits as they doin Zululand. The area is at present being investi- gated by the officers of the Department of Mines and Industries of Mogambique. It would appear that at least the Aptian, Albian, Cenomanian, and Senonian stages are present. A number of species of Aptian ammonites have been recorded by Kilian (1902), Krenkel (1910), and Spath (1921, pp. 309-318) from Catuane, Powell’s Camp, and other localities (de Andrade, 1929, p. 159) in the region south of Delagoa Bay; the Aptian has not hitherto been recorded from farther south in Zululand, unless van Hoepen (1929 a, p. 10) is Lower Cretaceous Lamellibranchia from Northern Zululand. 281 correct in correlating his Ndabana Beds at the base of the sequence on the Umsinene with that stage; according to Spath the Aptian beds at Catuane, etc., are probably Upper Aptian. From Catuane Spath (1925, pp. 180-196) has also described a number of Upper Albian ammonites, while Cox (1925, pp. 202-207) has described some gastropods from the same locality and horizon. The Senonian stage has been recognised at Incomanini on the Incomati (Komati) River, north of Lourengo Marques, the fauna containing a few species of Lamellibranchia identical with or allied to species in the Umzamba Beds of the Union (du Toit, 1926, p. 327); the Lamellibranchia have not been described, but Cox (1925, pp. 202, 208-215) has recorded a number of Gastropoda of somewhat Tertiary aspect; ammonites appear to be absent. Spath (1925, pp. 196-200) has recorded a Cenomanian ammonite and two Upper Senonian ammonites from Maputoland, the region south of Delagoa Bay. In the region between the Limpopo and the Zambesi the Cretaceous System is for the most part covered with recent deposits, but beds referable to the system appear here and there, both on the coast and inland. Newton (1924) has recorded both Senonian and Danian faunas from the Cheringoma (Sheringoma) Plateau adjoining the Urema trough, north of Beira; while the Upper Cretaceous species Lopha (= Alectryonia) ungulata was recorded by Newton (1896) from the neighbourhood of Sofala. Still farther north the Cretaceous is absent for a considerable distance, but Cretaceous beds reappear on the coast south of Mocam- bique, and can be traced into Tanganyika. From this region Choffat (1903) has recorded Cretaceous beds at Conducia, apparently largely Cenomanian; the Neocomian, Aptian, and Albian stages are said to occur in this region also (de Andrade, 1929, pp. 158-160). The New Collections from Northern Zululand.—The Lamelli- branchia described in this paper come from outcrops in Northern Zululand, situated for the most part on the Pongola River and in the region between the Pongola River and the borders of the Trans- vaal and Swaziland, not far to the south of the southern boundary of Portuguese East Africa. Prior to 1933 this tract of country was practically unknown from a geological point of view, though it was known that the Cretaceous beds exposed north-west of Lake St. Lucia extended northwards along the lower course of the Pongola and into Portuguese territory. In 1933 Mr. H. F. Frommurze, B.Sc., of the Union Geological Survey, while engaged in irrigation work in this area, collected from fossiliferous Cretaceous beds along the 282 Annals of the South African Museum. Mfongosi stream, a tributary of the Pongola, and the collections sent to Cape Town were deemed of such importance that the area was immediately visited by Dr. 8. H. Haughton, now Director of the Survey, and Dr. A. W. Rogers, formerly Director of the Survey. Cretaceous outcrops were located over a wide area and an attempt was made to work out the sequence by careful collections of fossils on a zonal basis. Subsequent to their visit further collecting was done by Mr. J. 8. Hutt, Surveyor to the Irrigation Department, without reference to localities or horizons; this collection was pre- sented to the Transvaal Museum. Acknowledgments and Preliminary Remarks. — The Lamelli- branchia described below form part of the collections obtained by Mr. H. F. Frommurze, Dr. 8. H. Haughton, and Dr. A. W. Rogers, and by Mr. J. 8. Hutt in 1933. The writer is very greatly indebted to Dr. 8. H. Haughton, Director of the Geological Survey of the Union of South Africa and Honorary Curator of the Palaeontological Collections in the South African Museum in Cape Town, for placing the Lamellibranchia at his disposal for description, and for much useful advice. Dr. E. L. Gill, Director of the South African Museum, and Dr. T. W. Gevers, formerly on the Union Geological Survey, have very kindly forwarded to the writer specimens and literature contained in the collections in Cape Town, while the writer is parti- cularly grateful to Dr. Gill for arranging for the publication of this paper and for help in various ways. Mr. L. R. Cox, of the Geological Department of the British Museum (Natural History), and Dr. K. H. Barnard, Assistant-Director of the South African Museum, have given invaluable help by looking up early references to Trigonia not accessible in Grahamstown, and giving their opinions on certain problems of nomenclature; to them the writer is very greatly in- debted. The accompanying plates were prepared by the writer, and his best thanks are due to Dr. John Hewitt, Director of the Albany Museum in Grahamstown, for granting every facility in the use of camera and dark room; and for much helpful criticism. The cost of photographic materials was defrayed by a grant from the Research Grant Board of the Union of South Africa, for which the writer expresses his thanks. . The collections obtained by Mr. H. F. Frommurze have been placed in the South African Museum in Cape Town; those made by Mr. J. S. Hutt in the Transvaal Museum in Pretoria; where individual specimens are referred to, “S.A.M.” or “T.M.” precede the catalogue numbers. The collections made by Dr. 8. H. Haughton and Dr. Lower Cretaceous Lamellibranchia from Northern Zululand. 283 A. W. Rogers are in the Cape Town office of the Geological Survey of the Union, at the South African Museum; these are simply labelled with the horizon letter and number, but types and figured specimens at least will be transferred to the collections of the South African Museum in due course. A few specimens from the Albany Museum (A.M.) are also referred to. In recording localities in the text the number of available speci- mens from each locality has been indicated by a number in brackets, e.g. Z19 (4) indicates that 4 specimens were collected at horizon Z19. References to published papers have been made by indicating the author’s name and the date of publication; a list of references will be found at the end of the paper. In preparing the descriptions that follow, reference to a number of badly preserved, tragmentary, or otherwise doubtful species of Lamellibranchia has been omitted. It was considered that no good purpose would be served by giving descriptions or figures of these species, either because the locality at which they were collected was not precisely recorded, or because they were found at one locality and horizon only, and therefore fail to be of any value in correlating horizons at one locality with those of another, or because of some uncertainty as to the identity of the genus to which they belong, or because of great imperfection in the preservation of the shells concerned. Included in this category are a number of specimens referable to the Veneridae, to Ostrea, to Pecten, and to a few other genera. | Trigomae belonging to the section Scabrae, which would be referred to the sub-genus Scabrotrigonia Deecke, or to the several sub-genera proposed by van Hoepen (1929), have not been specially dealt with here, although several species are represented in the abundant material available. Apart from this group, which will be worked out later, all the recognisable and significant species of Lamellibranchia con- tained in the collections are described below. A brief reference to the Scabrae is included, however. 9. An ACCOUNT OF THE GEOLOGY OF THE CRETACEOUS BEDs. (By S. H. Haughton, M.A., D.Sc., Director of the Geological Survey of the Union of South Africa.) The localities from which the fossils described in the following pages were collected lie in the coastal belt of Northern Zululand, in the magisterial districts of Ubombo and Ingwavuma. This belt is bounded to the west by the Lebombo mountain range, a dissected Mzinyeni Pan NATIVE RESERVE NO 15 ~ PongoloA a Mokotini A (Camp) Lower Cretaceous Lamellibranchia from Northern Zululand. 285 peneplaned ridge of Karroo volcanic rocks whose structure and features have been recently described by du Toit (Trans. R. Soc. S. Afr., 1929, vol. xvi, pp. 189-217). The range has an approximately north-south strike, and—in the area visited by the writer—its wonderfully even sky-line is broken only by the deep gorges of the Mkuzi, Pongola, and Ingwavuma rivers. Transecting the range in a west-east direction these streams begin to meander as soon as they leave the foothills of rhyolite and enter the flat coastal plain, the Mkuzi turning in a south-east direction to empty itself into False Bay, the Pongola—joined by the Ingwavuma—flowing northwards into Delagoa Bay. In addition to these major streams which cut through the Lebombo, a number of smaller tributaries rise on its eastern flanks. Where they, and the bigger streams, flow through the Cretaceous beds they form a number of disconnected deep pools in which hippopotami and crocodiles are to be found. The actual river-beds lie about 60 feet below the general level of the coastal plain. This plain is capped by a covering of gravel with, in parts, a consolidated calcareous grit of marine origin. Sand, too, is common. The fossiliferous Cretaceous beds are only exposed on the banks of the streams or sides of gullies. The localities in- vestigated are shown on the accompanying plans, and are—from south to north—the Myesa Spruit, the Mfongosi Spruit, the Pongola River with a tributary gully, the Msinyene Pan, and the Lombag- wenya Spruit. Small collections were also made south of Pongola Poort at Mokatini Camp and to the south of it. The investigation of this area by the writer and Dr. A. W. Rogers, F.R.S., was only made possible by the kindness of the Director of Irrigation, who kindly placed all his large-scale contour plans at our _ disposal, and of Mr. J. 8. Hutt, Surveyor-in-Charge of the Lower Pongola Irrigation Survey, whose camp at Otobotini provided us with every comfort, and who placed both his time and local knowledge enthusiastically at our service. Mr. Hutt, both before and after our visit, was active in the collection of fossils, and presented the bulk of his material to the Transvaal Museum. Some of it has been studied, and the results are incorporated in the palaeontological descriptions which follow. Description of Localities. 1. Myesa Spruit.—Very platy rhyolites with a dip of about 10°-12° in a direction 10° north of east form a cliff just below the drift on VOL. XXXI, PART 3. 25 286 Annals of the South African Museum. the southern branch of the stream. The contact between the rhyolites and Cretaceous beds is not seen, being overlain by alluvium and gravels. \ | North of the drift, across the northern branch, buff sandy beds containing gypsum are seen in a roadside hole and in a shallow donga. They contain a band of hard gritty quartzite, which is nodular; on Rag. We the weathered surface this shows masses of fibrous calcite, which may possibly be replaced wood. At locality M2 are unevenly bedded pebble beds and grits, buff in colour, with a dip towards the south-east. The exposures on the left bank of Myesa Pan (M3 and M4) are poor. They consist of buff soft sandstones with occasional small lenses of hard calcareous grit, which sometimes carry badly preserved shells. 2. Mfongost Spruit.—The most easterly outcrop of rhyolite along this spruit is seen at the point where the river leaves the foothills, some 3-8 miles in a direction west 30° north from the drift across the spruit on the main road. Here the rhyolite is platy and amygdaloidal, and dips at 12-15° in a direction slightly north of east. Lower Cretaceous Lamellibranchia from Northern Zululand. 287 Below this the river banks are low and gravel-covered. A few hundred yards downstream there is a short exposure of consolidated boulder beds and gravels which appear to have a slight dip down- stream. ‘There is no evidence available as to their age; but none of the recent conglomerates along the rivers appear to be so compacted as these. The most westerly outcrop of Cretaceous beds occurs on the south branch of the spruit at a point about 2 miles above the drift, measured in a straight line, bearing west 15° north from the drift. This con- sists of bufi-coloured, sandy pebble beds and sandy grits in which no fossils were seen. Just above the junction of the two branches of the spruit there is a good exposure on the south branch of a thickness of about 30 feet of buff sandstones, grits, pebble-washes, and conglomerate, with a dip of 3° towards the east. In places current-bedded bands show a much higher dip in the same direction. Included in the conglom- erate are many rhyolite pebbles, some of agate, and some of sedi- mentary rocks. The sandstones carry ferruginised badly preserved fossil wood, but no marine fossils were seen. At the big cliff, on the left bank of the stream above the drift, there is a succession of beds, mostly of buff sandstone with hard lenticular nodules about 55 feet thick. This locality was designated Za. The basal 5 feet contain six thin hard bands filled with fossils, which include Trigonia obesa, Trigonia haughtom, and Trigonia hennigi, separated by buff sandy joint clays. Two feet above this is a narrow band with the same Trigomia species and a Belemnite. The buff sandstones of the higher part of this succession outcrop at the first cliff above the drift on the right bank of the river. Here some of the gritty sandstone is rather more micaceous than at locality Za; the finer-grained sediments contain fragments of stems; and a few fossiliferous concretions with Trigonia occur. On the slope above the cliff (locality Zo) one specimen of a large uncoiled ammonite, Tropaeum cf. gigas, was discovered, and some fragments provisionally assigned to the genus Toxoceratoides. Below the drift the cliff on the left bank has at its western end sandy shales with rounded and irregular nodules containing a species of Acanthohoplites and Dicroloma sp. (locality Z1). Then follows a succession of sandy clays with thin bands of fossiliferous limestones (locality Z2) containing Trigonia obesa, Trigonia pongolensis, Cardium rogerst, and Dicroloma sp. The highest horizon in this cliff section (Z3) consists of buff argillaceous sandstones with thicker elongate (oes ies: LII7S COOP OOO? GOO! a Cosa Ss! See. ; Amgen vi SHY (PMYE LiMy HZ. (tee oe Se a) ——- 1S) = : ; N f > | . Tas . =) ww erty ce Au —x~_ OZ ‘i ye ia TI So AW, RY gs Z)_/ Ou sie Ne oN ; 9 Z i PUB) SOUPS Lf 1 i AS BEE \ é TONY \ Z; ae 1S. fees ‘ ae, v, we GP ree Re WSs cae S ---<5 Lower Cretaceous Lamellibranchia from Northern Zululand. 289 lenticles of limestone, and yielded Trigoniae of the Scabrae group, Trigoma obesa, Gervillia dentata, and Cardium rogers. On the right bank the first cliff is of buff sandy clays, with plant fragments and calcified wood with, at the top, bands of unfossiliferous septarian nodules. This cliff is succeeded by a slope, at the top of which is a fossiliferous limestone (locality Z4), with Gervillia dentata, Pseudavicula (*%) africana, Trigonia obesa, and Trigomae of the Scabrae group. Many of the shells here were broken before consolidation; further, the limestone carries large mud-pellets and some pebbles. This fossiliferous limestone dips at a low angle downstream at the top of the next cliff on the right bank. Near the lower end of the cliff it gives place gradually to fossiliferous soft sandstones, which yield a large number of small shells. The limestone is in reality a series of flat lens-like bodies occupying one horizon. Between the lenses the grey marly sandstone is fossiliferous. Shells of the same species as occur in the limestone are also found some 2 feet below it. Locality Z5 occurs on the slopes below a terrace farther down the right. bank of the river and shows outcrops of shelly lmestones definitely higher in the succession than those of Z4. From here were collected Gervillia dentata, Pseudavicula (%) africana, Trigonia obesa, Belemnites, and Acanthoplitids. Farther along the slope, opposite a cliff of the left bank, dirty bufi-coloured, weathering, shelly limestone bands, frequently with spheroidal weathering, become common. This is locality Z6. Logs of heavily bored calcified wood are numerous, and the fauna contains Gervillia dentata in large numbers, Inoceramus concentricus (*%), Trigonia frommurzer, Trigoniae of the Scabrae group, Panope gurgitis, Acanthohoplites spp., Toxocara- toides sp., and Ancycloceras (Australiceras) sp. The higher ground on the right bank of the spruit ends to the east in a narrow gravel-capped spur. Locality Z11 is on the south flank of this spur; locality Z12 on the north. The outcrops are of buff shales and nodular limestone, and the fossils collected include a Pseudodiadema, Acanthohoplites spp., several large ammonites, Cucullaea woodsi, Pseudavicula (*%) africana, Trigonia obesa, abundant Trigonia frommurzer, Trigoniae of the Scabrae group, Protocardia ct. sphaeroidea, and Panope gurgitis. On the left bank, locality Z7, is a cliff opposite the space between Z5 and Z6. A highly fossiliferous limestone band overlies sandy clays with limestone concretions. Gervillia dentata, Trigonia from- murzei, and large uncoiled ammonites are very common; Isognomon sp. and scabrous Trigoniae also occur. 290 Annals of the South African Museum. Above the limestone band which delimits the upper part of this succession is a slope formed of soft sandy shales and calcareous shelly concretions. This is locality Z8. Numerous bored logs, some of them 9 feet in length, occur here. Large loosely coiled Lythceratids, Acanthohoplites spp., Lytocrioceras, Cheloniceras, Diadochoceras sp. nov. aff. nodosocostatum, Tropaeum cf. bowerbanku, Helicancyloceras (%), and Sonneratia all occur in fair numbers, and the Lamellibranch fauna includes Trigoma frommurzei, Trigonia obesa, scabrous Trigoniae, Cucullaea woodsi, Glycymeris cf. griesbachi, Gervillia dentata, Isog- nomon sp. (2), Hxogyra conca, and Panope gurgitis. The band con- taining this fauna continues for several hundred yards downstream, and is overlain at Z9 by a series of outcrops of ball-like concretions that contain very few fossils, succeeded by outcrops of shelly lime- stone with minute shells, Trigonia frommurzei, Rhynchonella sp., Acanthohoplitids, and Diadochoceras. In these beds a log of calcified wood 30 feet long was seen. A gully running into the spruit at Z10 exposes soft buff shales with numerous logs of wood and a layer of ball-like concretions with sparse, large, flat, involute Beudanticeras—like ammonites, Gervillia dentata and Trigonia frommurzer. Below this gully the cliff shows bands of highly fossiliferous sand- stone and sandy limestone dipping at a low angle eastwards. The upper end of this cliff (Z13) has Pseudavicula (*%) africana, Trigonia frommurzei, Neithea quinquecostata, Protocardia cf. sphaeroidea, Douvilleiceras mammillatum, Beudanticeras, and Puzosia cf. stoliczkai; the lower end (Z14) yielded Isognomon sp., Inoceramus concentricus (*), and Panope gurgitis. Locality Z15, which is near the lower end of the Mfongosi Pan, yielded a possible new species of Beudanticeras. 3. Pongola River.—North of the mouth of the Mfongosi Spruit, between the Irrigation Survey points RAF and CRAFT, a short rather wide valley runs into the Pongola River. Along the sides of the valley is a continuous outcrop of a prominent thin band of limestone crowded with shells of Turritella manuanensis. This lime- stone has an apparent slight dip to the north-east. Below it are sandy buff clays with limestone concretions, which are highly fos- siliferous, Dipoloceratids, Mortoniceratids, and Desmoceratids being common, particularly at an horizon about 40 feet below the Turritella limestone. This limestone and the beds below were numbered Z16, whilst the beds above the limestone on the valley sides were numbered Z17. From Z16 were obtained Puzosia stoliczkar, Hysteroceras spp., Lower Cretaceous Lemellibranchia from Northern Zululand. 291 Dipoloceras quadratum, Deiradoceras spp., Pervinquieria spp., Gervillia dentata, Inoceramus concentricus, Scabrotrigonia spp., Pholadomya vignest, Gonomya sp., Veniella etheridgei, Exogyra conica, Ostrea sp., and Turritella manuanensis. The beds Z17 are less fossiliferous, but carry Adkinsites umsinenense, SZ Gastergpod Liumeslore. 4000 FEET Pervinqueria, Hysteroceras choffati, Dipoloceras cristatum, and Phylloceras velledae. North of the junction of this valley with the Pongola, on the farm Neser, there is a very fine long cliffi-section on the left bank of the river. The section is about 40 feet high, and its upper 5 feet is formed of Tertiary or Recent gravels and sands. The Cretaceous beds consist of a succession of clays and clayey sandstones with limestone bands and bands of nodules, and their apparent dip is at a low angle downstream. 292 Annals of the South African Museum. Near the base of the section (Z19) is a prominent thin pebble bed. Below this the clays are highly fossiliferous, Oxytropidoceras sp. nov., Nautiloids, Neithea quadricostata, Cucullea woodsi, Gervillia dentata, Trigoma cricki, Protocardia cf. sphaeroidea, Veniella etheridgei, Exogyra conica, Turritella manuanensis, and Avellana cf. incrassata being the commonest forms present. Above this pebble bed there are definite highly fossiliferous layers with the same Lamellibranchs as in the lower bed (with the exception of Veniella), together with small Dipoloceratids and the echinoderms Hemiaster zululandensis and Holaster vanhoepent. Towards the top of this section (Z19) Puzosia cf. bhima becomes the commonest ammonite, and Dipoloceras cristatum is found. Proceeding northwards the cliff section continues (Z20) and some 25 feet of buff sandy clays and limestone nodules carry abundant Exogyra, Vemella, Trigona, Neithea, and Nautiloids. Ammonites, however, become rather scarce, Myloceras, Puzosia, and Dipoloceras being the only genera collected. Still farther to the north the cliff becomes lower, and the post- Cretaceous grits and gravels sink considerably in level. The Creta- ceous beds are more sandy than to the south, and carry sporadic nodules of calcareous quartzite, some of which are fossiliferous. Fossils are scarce in the buff sandstones. From this locality (Z21) Holaster, Hemiaster, Inoceramus concentricus var. subsulcatus, Phola- domya vignesi, and Goniomya are recorded. 4. Lombagwenya Spruit.—The main road from Otobotini to Ndumu crosses this spruit to the west of the Kwambosi Pan, and rises up to the plateau past the Lombagwenya Store. On the slopes down to the spruit, west and north-west of the store, richly fossiliferous out- crops occur; and other exposures of Cretaceous beds occur a short distance below the drift. Three traverses were made across this fossiliferous slope. Immediately to the west of the store a thickness of about 125 feet of beds (locality L3) consists mainly of buff sandy marls with cal- careous nodules. Near the base are large open-coiled Crioceratids with depressed whorl section and prominent widely spaced ribs passing right across the venter, associated with Gervillia dentata and Trigonia frommurzei. Just above the bed most prolific in ammonites is a narrow band of sandy clays, below a thin limestone, that contains numbers of Rhynchonella and Terebratula. The upper 50 feet of beds carry several species of Douvillecceras including typical D. mamillatum, Phylloceras sp., Leymeriella sp.n., Acanthohoplites spp., and a new Lower Cretaceous Lamellibranchia from Northern Zululand. 293 genus of ammonite. At the top of the slope large Terebratulids occur. Following the slope to the north-west one encounters blocks of fossiliferous limestone; here and there are low benches formed of limestone outcrops. The buff shaley beds which are intercalated with these are only occasionally exposed in shallow dongas. The second traverse was made upwards from the southern end Iie, Ze of a cliff about 2300 yards north-west from Lombagwenya Store. The chff is composed of almost unfossiliferous buff sandy clays, showing some false-bedding with fossiliferous limestone nodules (locality L8). Phylloceras sp. and Hxogyra conica (%) occurred here. Just above the crest of the cliff occurred Megacucullaea sp., Trigonia hennigi, Tropaeum sp., and higher up the gully are limestones with large Ostrea, Panope gurgitis, Aconeceras cf. walshense, and Toxocera- toudes of the royert group (locality L9). The next outcrop up the sully (L7) is limestones with Pholadomya, Panope, Australiceras ct. lampros, Lytoceras cf. mikadiense, Cheloniceras (%), and numerous specimens of a sinistral Turrilites cf. emericianus. At the top of the slope are specimens of the loosely coiled Crioceratids generally similar to those that were found near the base of locality L3. 294 Annals of the South African Museum. The third traverse was made beyond the northern end of the cliff. The lowest beds seen here (L10) are buff sandy marls with large lenticles and rounded masses of slightly calcareous sandstone carrying fossil wood. The lenticles sometimes carry shells, especially a medium sized Chlamys. Above these beds is a band of shelly limestone (L11) with Panope gurgitis, Aconeceras cf. nisus, Acanthohoplites spp., Leymeriella spp. Slhghtly to the south-east, and at about the same height above the river, are shelly sandstones (L12) with Cucullaea woods (%), Megacucullaea sp., Trigonia hennigi, but no ammonites. Above these is a band of sandy shale with fragile Scabrotrigonia, Gervillia dentata, Ostrea sp., and Thetironia sp., followed by shales with limestone and very abundant fossil wood. Higher is a zone of buff shales with limestone concretions yielding enormous specimens of Gervillia, Trigonia hennigi, Ostrea sp., Nautilus, and Phylloceras, with fossil wood (L4). Above this is an outcrop of shelly limestone with large loosely coiled Crioceratids, whilst at the top of the slope are numerous weathered, rounded clay nodules (L3+) * of various sizes, many of which are fossiliferous and carry Isognomon sp., Lopha diluwana, Tropaeum cf. mozambiquense, Tropaeum cf. arcticum, Australiceras, Acanthohoplites of the roseanus group. Exposures of Cretaceous beds occur also below the drift. On the right bank of the spruit about 150 yards below the drift is an exposure of a thickness of about 30 feet of stiff sandy clays with one layer of hard fossiliferous limestones and some sporadic calcareous nodules (locality L6). These yield Phylloceras, Douvilleiceras, Scabro- trigonia, Terebratula, Exogyra cf. conica, and Gervillia dentata. On the north side of the spruit are low ridges above the alluvial — plain. On the slopes of these (L1) are occasional outcrops of weathered calcareous nodules that contain large specimens of Gervillia dentata, Scabrotrigonia, Panope gurgitis, Glycimeris, brachiopods, Douvilleiceras, and Oxytropidoceras (*) sp. juv. Farther down the spruit a low cliff above the water has buff clays which are highly fossiliferous, but retain the fossils only in the form of internal casts. Fragments of ammonites only were seen here (L.2), one small crushed Echinoid, and Scabrobrigonia sp. 5. Msinyene Pan (locality L5).—The low slopes on the western side of the south end of this pan show few outcrops; but some fossils were collected here which are of interest. Among them was a speci- men of a new ammonite belonging to the form already recorded * Locality L3t is distinct from L3 and may not be the same horizon. L37 is equivalent to [L3] in fig. 4. Lower Cretaceous Lamellibranchia from Northern Zululand. 295 from the upper beds of section L3, Leymeriella sp., and a species of Beudanticeras similar to one found in the highest beds of the Mfongosi Spruit. Scabrotrigonia sp. also occurs, and an imperfect shell which Rennie has tentatively assigned to Neithea quadricostata. 6. Mokatini.—The area to which this name has been given lies between the Pongola and Mkuzi Rivers. Apart from the rhyolite foothills of the Lebombo the country is almost entirely covered by soil and sand, and shows very few outcrops of the Cretaceous beds. The low ridge which runs south from the Mokatini Irrigation Survey camp, on the south side of the Enseleni road, carries an exposure of a band of very hard shelly limestone from which it is difficult to extract shells (locality Z24). The shells are arranged in layers separated by bands of hard, barren, brown-weathering, fine- grained calcareous quartzite. From here come Pseudavicula (?) africana, Scabrotrigoma sp., and Thetironia sp. About 4 miles to the east of this, and at an altitude about 150 feet lower, is a small exposure of a hard limestone with pebbles and shell fragments and blocks of fossil wood. A little downstream from this.a softer shaley bed (Z25) yielded Trigonarca cf. ligeriensis, Exogyra conica, Trigoma blanckenhorni, and Veniella etheridge. 3. A PRELIMINARY ANALYSIS OF THE AMMONITE FAUNA. (By 8. H. Haughton, M.A., D.Se., Director of the Geological Survey of the Union of South Africa.) Until the ammonite fauna is identified specifically exact determina- tion and correlation of the various horizons present will not be possible. Nevertheless, certain generalisations may be made which are of interest. 1. Mfongosi Spruit Succession.—It is estimated that a thickness of 800 feet of fossiliferous beds are exposed along this spruit, the lowest being exposed at locality Za and the highest at locality Z15. On the right bank ammonites were collected at, in ascending order, Zo, 25, 78, and Z12: on the left bank at Z1, Z7, Z8, Z9, Z10, Z13, Z15. The lowest horizon (that at Zo) yielded Tropaeum cf. gigas, and must lie some 30 feet below Z1, which carries Acanthohoplites sp. nov. The presence of Tropaewm would seem to indicate the presence of the top of the Lower Aptian or of the lower part of the martini zone of the Upper Aptian at Zo. The assemblage from Z8 includes forms that may be representative of more than one of the European zones. Tropaeum ct. bowerbanki 296 Annals of the South African Museum. is compared with a species that 1s representative of the middle sub- zone of the Tropaeuman of the Upper Aptian. On the other hand, Diadochoceras sp. nov. aff. nodosocostatum (ident. L. F. Spath) is compared with a species that is characteristic of the lowest zone of the Albian. It is possible, however, that this apparent mingling may be susceptible of interpretation, as the locality Z8 extends for nearly 1500 feet along the side of the spruit and contains a thick- ness of some 78 feet of beds. Further collecting may show that each of the critical forms is confined to a definite horizon, and that the intervening subnodosocostatum zone may also be present. It is of interest that a species of Acanthohoplites, comparable with A. aschiltaensis, comes from the lower beds at Z9; this species is characteristic of the upper part of the swhnodosocostatum zone. At locality Z9 Acanthohoplites bigourets is definitely indicative of the presence of the lowermost Albian (Clansayes horizon); and at Z13 Douvilleiceras mammillatum shows the presence of the base of the Middle Albian (mammullatus zone). Here again there is a hiatus in the faunal succession as worked out in Europe, corresponding to the Leymeriellan zone of the Lower Albian, which may be represented by the beds at Z10, which contain Beudanticeras sp. These beds cannot be more than a few feet thick. Since no stratigraphic break was visible in the rock succession, it may be accepted that along this section there is a succession of fossiliferous beds ranging from the lower part of the martini zone of the Upper Aptian at least to the mammullatus zone of the Middle Albian, and that the unfossiliferous beds in the upper reaches of the spruit are Lower Aptian and may range down into the Neocomian. A tentative classification of the beds, according to the locality numbers, is as follows :— Middle Albian . mammillatus zone INS; LAD: tardefurcata zone Z10 (2), Upper beds (Leymeriellan) of Z9 (2). ee nodosocostatum zone Z9. Upper part of (Acanthohoplitan) Z8 (2), Zz subnodosocostatum zone Lower part of Z9? (Parahoplitan) Lower part of Z8? Upper Aptian . i (2);) ZS eee ZT (2). martini zone Zo, Ll (2). (Tropaeuman) Lower Cretaceous Lamellibranchia from Northern Zululand. 297 2. Lombagwenya Spruit.—Beds of approximately similar age to the above occur in this area. The presence of the mammillatum zone is evidenced by abundant Douvilleiceras in the upper 50 feet of the L3 succession; these upper beds also yielded Leymeriella, so that it may be possible to see in them lower tardefurcata zone beds and upper mammillatum zone beds. The lowest beds in this section contain Crioceratids that are comparable with those of Z7 and ZS. The assemblage at L11 contains peculiar features, in that Aconeceras and Leymeriella apparently occur together. Hitherto Aconeceras has been considered to be confined to the Gargasian, with the exception of possible representatives from the Bedoulian, but is unrecorded from the Albian; Leymeriella, on the other hand, seems to be con- fined to the upper half of the Lower Albian. It is certain that L11 lies below L3y7, in which Gargasian Ancycloceratids and Acanthohoplites of the roseanus group occur. It may be remarked that the associa- tion of Tropaeum with Australiceras both at L3} and at Z6-Z8 tends to throw doubt on Whitehouse’s suggestion that the former was a descendant of the latter. 3. Pongola River.—In this area the fauna is definitely Middle and Upper Albian. The lowest horizon (basal part of Z19) contains Oxytropidoceras, whilst Manuaniceras occurs just above the pebble bed near the base of this section. Slightly higher were found Dipoloceras cristatum, Dipoloceras bouchardianum, and Puzosia cf. bhima. The Dipoloceratids of this assemblage are closely compar- able with those from the top of the Middle Albian of England, the cristatum sub-zone of the Lower Gault. Localities Z16 and Z17 contain forms that are indicative of the Upper Albian, such as various species of Pervinqueria and Deirado- ceras. The former ranges in Hngland from the orbigny: to the aequatorialis sub-zones. Hysteroceras, which is also fairly abundant here, is typical of the lower part of the Upper Albian. On the other hand, Dipoloceras cristatum has also been found at this locality, as has Adkinsites umsinenense. The presence of these probable members of the cristatum sub-zone fauna in the upper part of this section above the gasteropodous limestone is an anomaly that cannot at present be explained. 4. Tor LAMELLIBRANCH FAUNA. The faunas collected at the localities described above consist mainly of Ammonites and Lamellibranchs, with comparatively few Gastro- 298 Annals of the South African Museum. pods, and rare Hchinoids, Belemnites, etc. The Lamellibranchs are described below, and a brief notice of the associated Gastropods is appended. The assemblage of Lamellibranchs in the collections is remarkable for the high proportion of large, massive, and thick-shelled species, among which several species of Trigonia are conspicuous, e.g. T. obesa, T. hennigit, T. frommurzei, as well as Cucullaea spp. aff. kraussi, Gervillia dentata, Veniella etheridger, and Protocardia cf. sphaeroidea. From the systematic point of view, the chief interest in the faunas is the great diversity displayed by the Trigoniae, at least ten species of which are represented, referable to as many as seven sub-genera, including the very remarkable new species 7. frommurzei, for which the new sub-generic name Sphenotrigonia is proposed. A table showing the distribution of the species at the several localities is appended to this paper. Though in the last resort the precise identification of the ages of the several horizons must be based on the associated Ammonite fauna, the Lamellibranchs are of some considerable interest. On the Lamellibranch evidence alone, the beds would appear to range from Neocomian to not later than Cenomanian, and are thus largely Lower Cretaceousinage. The presence of the Neocomian is suggested by the occurrence of several species of Lamellibranchs with distinctly Neocomian affinities, but it appears from Dr. 8. H. Haughton’s preliminary analysis of the Ammonite fauna that these species are actually associated with Upper Aptian and even Lower Albian Ammonites at certain localities. The association of Trigoniae, belong- ing to groups hitherto regarded as exclusively Neocomian and highly characteristic of Neocomian deposits in Central and South America, the Uitenhage Formation and deposits in East Africa and India, with post-Neocomian ammonities, is a remarkable feature of the faunas. Asa result of his studies of the southern Neocomian Lamelli- branch faunas, Kitchin (1926, p. 467) has emphasised the fact that such Lamellibranch types as the Pseudo-quadrate Trigoniae, Trigoniae of the groups of T. conocarduformis and T. v-scripta, and large Cucullaeae of the C. kraussi group are confined to Neocomian deposits.* It is therefore somewhat surprising to find species that can be paired off with well-known Uitenhage (Valanginian) species from the Cape Province occurring in Zululand in association with Ammonites of later stages. Dr. Haughton finds it probable that on the Mfongosi “the unfossiliferous beds in the upper reaches of the spruit are Lower * But see p. 300. Lower Cretaceous Lamellubranchia from Northern Zululand. 299 Aptian, and may range down into the Neocomian,”’ but Neocomian Ammonites are wanting in the collections. It is just possible that the Neocomian is represented on the Myesa stream (M1 and M3) and at ZA on the Mfongosi, since the Trigonia v-scripta group is there represented by a new species that was not collected at higher horizons, and Ammonites were not found; but in view of the undoubted association of e.g. a Pseudo-quadrate Trigonia with Aptian-Albian ammonites on the Lombagwenya and of T. obesa with Lower Albian ammonites at Z12, the occurrence of a new species of Trigonia belong- ing to the 7. v-scripta group cannot be held to prove the presence of the Neocomian at the localities mentioned, and it is thus possible that the Neocomian is not present in Zululand at all. Cretaceous beds older than the Middle Albian were not known to occur in Zululand until after the publication by Spath (1921) of his paper on the Cretaceous Cephalopoda of Zululand, though it might have been suspected that the Aptian stage, which was known to outcrop just north of the Zululand border, extended farther south. In 1929 van Hoepen published a description of the Trigoniae obtained by him on the Umsinene River, and in the same year claimed (1929 a, p. 10) that his Ndabana Beds at the base of the succession at that locality represented the Aptian. The evidence on which that claim was based has not yet been published in full. From his Ndabana Beds van Hoepen (1929) has only described four species of Trigonia, and these have a distinctly Uitenhage aspect, being closely comparable with species occurring in and highly characteristic of Neocomian deposits in Central and South America, and of the Uitenhage Series in the Cape Province. The occurrence of no less than four species of Trigonia in the Ndabana Beds that can all be paired off with well-known species from the Sundays River Beds in the Uitenhage Series of the Cape is suggestive of a Neocomian age for the Ndabana Beds, since at least three of these species belong to groups of Trigoniae highly characteristic of the southern Neocomian, as Kitchin (1903, 1908, 1926, 1929) has urged, and since also the age of the Sundays River Beds has been determined very precisely by Spath (1930, p. 132) as undoubtedly Upper Valanginian, 7.e. well below the top of the Neocomian. The species concerned are as follows: Megatrigonia obesa van H. belongs to the group of T. conocarduformis (Krauss), named after a well- known Uitenhage species; JLotrigonia crassitesta van H. and J. in- constans van H. belong to the group of 7. v-servpta Kitchin, and are closely comparable with T. stow: Kitchin and 7. vau Sharpe 300 Annals of the South African Museum. respectively, both characteristic Uitenhage species; Pvrsotrigonia salebrosa van H. is a massive species of the group of the Scabrae distinctly reminiscent of the Uitenhage species T. kraussi Kitchin. The occurrence of the Neocomian in Zululand is further suggested by the new collections of Lamellibranchs from the Pongola area. In addition to T. obesa (van H.), a new species belonging to the T. v-scripta group (TL. haughtoni), and a large member of the group of the Scabrae strikingly like T. kraussa Kitchin, other character- istic southern Neocomian types are recorded. T. hennigi Lange is a very typical pseudo-quadrate Trigonia, first described from the T. schwarzi-Beds of the Tendaguru Formation in Tanganyika, the Neocomian age of which has been generally accepted (Kitchin, 1908, pp. 46-49; 1926, 1929), though Dietrich (1933, pp. 75-79) would correlate these beds with a succession ranging from Upper Valanginian to Lower Aptian; the species is also closely allied to two character- istic Uitenhage species, T. herzogi (Goldfuss) and T. holubi Kitchin, as well as to species in the Neocomian of South and Central America, and the Oomia Beds of Cutch. TZvrigonia pongolensis nov. is apparently very closely allied to T. krenkeli Lange from the T. schwarzi-Beds in Tanganyika. The massive radially costate Cucullaeae, collected on the Lombagwenya stream, belong to the group of which Cucullaea kraussi Tate from the Uitenhage Series is the type, and which has hitherto been regarded as confined to the southern Neocomian faunas, being reported also from the Neocomian of East Africa and of Madagascar, and from India. The Zululand species referred to in the preceding paragraphs are closely allied to species belonging to a widespread southern bivalve fauna which Kitchin (1926, p. 467) regarded as probably entirely Valanginian. The association of members of this fauna with Aptian and even Albian Ammonites is, however, recorded above by Dr. Haughton for Northern Zululand, while in India the work of Spath (Geol. Mag., 1935) and Cox (Pal. Indica, 1935) has shown that this fauna survives the Neocomian there also. 1. Lombagwenya Spruit—The whole of the succession on the Lombagwenya stream (L1—L14) carries Lamellibranchs of a distinctly Uitenhage aspect. This is based on the occurrence of the pseudo- quadrate Trigonia hennigi in LY, L12, and L14, and of Cucullaea spp. aff. kraussi in L9 and L12. The occurrence of Panope gurgitis in L1, L3, L7, L9, and L11 is not inconsistent with this comparison, since the species ranges from Neocomian to Albian in Europe. In a preliminary identification of species certain Albian or post-Albian Lower Cretaceous Lamellibranchia from Northern Zululand. 301 Lamellibranchs appeared to occur at these localities, but in each case the identification was based on doubtful material and no weight can therefore be attached to the record based thereon; Cucullaea woodsi from L12 was based on a single incomplete specimen very doubtfully referred to this species; ‘ Y ~ | i x q | 7 - | | ie C: r 7 fag Ate i > eu . Geeta nr” | . 8. Afr. Mus., f : Ann. S. Afr. Mus., Vol. XX XT Plate LIV. whoto: J. V. L. R. Neill & Co., Lid. LAMELLIBRANCHIA FROM THE UITENHAGE SERIES. Gervillia dentata. PuatEe LV. FIGS. PAGE 1, 2. Gervillia dentata Krauss . : 4 : : E ; : =< Sundays River Beds, Uitenhage Series, Cape Province. 1. Right valve from the Sundays River, collected by Dr. W. G. Atherstone, Albany Museum 827, reduced x 0:7 approximately, seen in lateral view. 2. The same, interior view. Ann. S. Afr. Mus., Vol. XX XI. Photo: J. V. L. R. LAMELLIBRANCHIA FROM THE UITENHAGE Gervillia dentata. SERIES. Plate LY. Neill & Co., Ltd. ee ee eee ERN ZULULAND. owenya and Msinyene Pan. 7. TABLE sHowING DistTRIBUTION oF Species 1N Lower CreTAckous Deposits, Poncora River AreA, NortHERN ZULULAND. Mfongosi. Ponpola, Myean Lombagwenya and Mainyene Pan, mae nd. 7u.)7Z1.\22.|23,)24.)20.|26.\27.\28,)29.)210,)211.|z12 \z13.|214,)z16.)z17.\z19.)220,| 221, M1. | M3.) U1. | 12.) 03.|L3t.| 14. |05. | u6.|o7. | os. | 9. joan.|c12,.n141z04.|290, LAMELLIDRANCHIA. (Gucullaen|(Cyphoxis) ocala) 2 . callie z ; 4 (Mlegnevculiaea) spp. | : : Arigonarea ot. ligeriensis me 2 GimmeiOeytien o 3 2 6 A F Gervillia dentata . : és 3 : Dab escn | eoee | een een | ese x x x x x x x x Jeognomonsp. . = - = - x| x x x Inoceramus concentricus . o 5 6 ? i) x Wl \concenricuavar-eubeulcatua) x : : Peeudaviculat africana... «|x F s Arcelie priitevites 7) ee 1 Erogyraconica =. wee t x x | x 1 t z Lophadiluviang 2 9. 0. ee 1 x inion » = Nell |pcleslells: x : x » _(lotrigonia) haughloni. . - || Sail lds (Rutitrigonia) pongolensis . x WN (eicurolrsgonsa) BUasceken Korn tae 3 = \(Steinmanella)kennigh 9s Nx * < x ralles » (Sphenotrigonia) frommurzei Set | sed | set | Se sc | ese x » (Scabrotrigonia)spp.. . . x| x S| sel | sel | sed | eset || sect} seal x |x |} sll Riel x Ba hee sen 33 Verickis x || = ear ( 3 Yoh brava. Neitheaquadricosfala =. . . . ! x || x ? y quinquecostata Pholatomyatiuyness : aape viens ls «Nt ae : alalls 1 Goniomya ap... we x x Ventdllatheridge) = 9. : mlles ? z Protocardia cf. sphaeroidea . . al| cele c Cardiumrogersi =. 2. 0 x |l x Rint so 2 o 6 all [eae Palle “ 2] x 2 : : x alts “WRG oo a 8 « x : 5 GASTROPODA. Turritelamanvanensia » - : “alls Avelanawiineneata «9 6 : : ’ Dicroloma sp. tes «on gee x| x a Vol. I. apa 13 out of ane ta 2 ’ eee eee Vol. I1.—Part 1, out of print; ae 3 ees _* “Part 6, 1/6; Part 6, °3/-;~ bare "Pan ee | Part 10, 7/-; Part ul, 3/-; Index, Title > ABs Nes b Vol. IIL.—Part 1, out of print; Part 2, 1/6; Part 3, Bibs Part 4, Resi Sent 5/6; Part 6, 7/-; Part | a: 1/65. Part 8, 3/-3. Part 9, eae at he Title, etc., 1/6. eRe Vol. IV (containing Palaeontological papers published in conjunction y it Geological Survey).— 2S : Part 1, 11/-; Part 2, 7/-; Part 3: 4/6; Part 4, 4/6; Part 5 | wae | Part 6, 4/6; Part 7, 14/—; Part 8, 8/-. Be Vol. V.—Part 1, 4/6; Part 2, 8/6; Part 3, 2/6; Part: 4, 1/6; Part 5, 5 Part 6, 5/-; Part 7, 3/-; Part 8, 4/6; Shed 9, 4/6; Index, etc., 1/6. Vol. VI.—Part 1, 13/6; Part 2, 4/6; Part 3, 3/6; Part 4, , 30/-; Index, 1 Big etc., 1/6. ime Vol. VIL (containing Palaeontological papers published i in conjunction vith Geological Survey).— er gee: Part 1, 3/~; Part 2, out of print; Part 3, 5/-; Part 4, 8-5 Part 5/6; Part 6, 1/6; Index, Title, etc., 1/6. Vol. VIII.—Complete, out of print. Index, Title, etc., 1/6. Vol. [X.—Part 1, 4/6; Part 2, 5/6; Part 3, eee Bey 4, 6/6; Part 5, * Part 6, 11/-; Part 7, 9/-; Index, Title, ete “‘ Vol. X.—Part 1, 3/-; Part 2, 2/6; Part 3, 2/-; Pest 4, 3/-3 Part 5, 2 a Part 6, 3/-; Part 7, 10/—; Part 8, 2/6; Part 9, 5/-; Part 10, 2 Part 11, 20/-; Part 12, 7/-. Complete. Title, ete., Part 11. ad . Vol. XI.—Part 1, 3/6; Part 2, 2/-; Part 3, 13/6; Part 4, 1/6; ‘Part 6,3 Part 6, 11/-; Index, Title, etc., and Plate Ill, 3/-. | . Vol. XII (containing Palaeontological papers published in conjunction with Re : Geological Survey).— : bore a: Part 1, 15/6; Part 2, 3/6; Part 3, 4/6; Part 4, 3/-; Part 5, 7/: . Part 6, 6/-; Part 7, 20/-; Part 8, 20/-; Index, Title, etc., 1/6. beaeS Haas Vol. XIII.—Part i 6/-; Part 2, 2/6; Part 3, 3/-;. Part 4, 8/6; Part ‘5, 1 Part 6, 5/-; Part 7, 30/—; Part 8, i/-3 Index, Title, etc., 1/6.~ Vol. XIV.—Part 1, 8/6; Part 2 8/-; Part 3, 6/-; Part 4, 21/-; Part 5, 5y- Part 6, 9/—; Index, Title, etc., 1/6. + Vol. XV.—Part 1, 17/-; Part 2, 17/-; Part 3, 14/-; Part 4, 12/6; Part 5, 5/65 Part 6, 3/6; Index, Title, etc., 1/6. >a Vol. XVI.—Part 1, 30/6; Part 2, 4/-; Part 3 (with Title, Index, etc.), 25/-, = Vol. XVII.—Part 1, 12/-; Part 2, 9/6; Part 3, 3/-; Part 4, 17/-; Part 5, 1 Part 6, 2/6; Index, Title, etc., 1/6. * Vol. XVIII. vied 1, 20/-; Part 2; 7/6; Part 3, 30/-; Part 4, 12/6; Index, Title, , 1/6. Fas Vol. XIX aes 1, 22/-; Part 2, 17/6; Part 3, 1l/-; Part 4, 5/6; Index, Title, etc., 1/6. . Vol. XX.—Part 1, 8/6; Part 2; 12/6; Part 3, 4/-; Part 4, 10/-; Part 5, a Part 6 (with Title, etc.), 4/6. z Sy Vol. XXI.—Part 1, 25/-; Part 2 (with Title, etc.), 30/—.. Vol. XXII (containing Palaeontological papers eabishod § in conjunction with the Geological Survey).— ‘Part 1, 20/-; Part 2, 10/-; Part 3 (with Title, etc.), 3/6. ~ Vol. XXIT.—Part 1, 12/6; Part 2, 8/-; Part 3 (with Index, Title, etc.), 9/-. Vol. XXITV.—Part 1, 10/-; Part 2. 2/6; Part 3, 4/6. aes Vol. XXV.—Part 1, 12/6; Part 2, 10/-; Part 3 (with Index, Title, se 12 Sot Vol. XXVI Complete, 25/-. Vol. XX VII.—Complete, 25/-. | : . Vol. XXVIII (containing Palaeontological papers — in conjuniotion ithe the. Geological Survey).— nes: ae 1, as Pat 2, 17/6; Part 3, 6/-; Part 4 (with Se Title, etc.), 10/— Vol. XXIX.—Part 1, 20/-; Part 2 (with Index, Title, ete. a 20/-. . Vol. XXX.—Part 1, 15/6; Part 2, 20/-; Part 3, 8/-; Part 4, prs Part 5, Cee Vol. XXXI.—Part 1, 20/-; Part 2, 13/6; Part 3; ae z Vol. XX XII.—Part 1, 2/6. The Annals of the South African Museum will be ivead? at regular i intervals, as matter for publication is avarlable. : pares may be obtained from— 3 = anes =< oe = ; Messrs. WHELDON & WESLEY, Lrp., : Sen ae 2, 3, and 4 Arrnur Street, New OxrorD oS Loxox, We. a4 The LIBRARIAN, Sourn Arrican paee Care ToS ak z= t; : t: yee . : % ae ‘ 7 a Poa a : 4 | ANNALS , OF THE : " VOLUME ae Jw. ; fe DESCRIPTIONS or toe PALAKONTOLOGICAL MATERIAL COLLECTED sy tut SOUTH AFRICAN MUSEUM anp THE GEOLOGICAL SURVEY OF SOUTH AFRICA. : eel IV, containing : a 10) A Critical Revision of the (pieces Perissodactyla of Southern Africa.—By H. B. 8. Cooke. (With 31 Text-figures.) Title Page and Index to Volume XXXI. ie ' | ISSUED DECEMBER 1950. PRICE 14s. PRINTED FOR THE | TRUSTEES OF THE SOUTH AFRICAN MUSEUM ‘ AND THE : * GEOLOGICAL SURVEY OF SOUTH AFRICA BY NEILL AND CO. LTD., 212 CAUSEWAYSIDE, EDINBURGH. ( 393 ) 10. A Critical Rersion of the Quaternary Perissodactyla of Southern Africa *.—By H. B. 8. Cooxz, M.A., D.Sc., F.G.S., F.R.Met.S., F.R.S.S. Afr. (With 30 Text-figures.) INTRODUCTION, THE study of fossil mammals in Southern Africa has been far from systematic and, with a few rare exceptions, writers have been more concerned with placing new species on record than with studying the fauna and revising our knowledge in the light of later discoveries. The first fossil mammal known to have been found in this region is the giant “Bubalus” bainw, whose horns and damaged skull were recovered in 1839 by the remarkable civil engineer and naturalist Andrew Geddes Bain from alluvial deposits of the Modder River, Orange Free State. The material was described only in 1891 by the British palaeontologist Seeley, and the next record of a fossil mammal appeared in 1906 when Dr. R. Beck described a mastodon tooth from the gravels of the Vaal River. In the following year another German scientist, Professor EH. Fraas, gave a further account of this tooth and commented on other remains from the gravels. Also in 1907, the celebrated American palaeontologist Professor W. B. Scott described a collection of fossil mammals from the coast of Zululand. Two years later Dr. Robert Broom made the first of his long series of contributions to mammalian palaeontology in South Africa with his descriptions of a new antelope from alluvial deposits at Caledon and of a new giant horse from a limestone fragment washed up on the beach near Maitland, in the south-western Cape. In 1913 Broom described an assemblage of mammalian fossils from the thermal springs at Floris Bad, and in later years he described several new mammals from the Vaal River gravels and various open sites. In the past twenty-five years the initiative in the description of South African material has passed from the hands of outside experts like Seeley, Scott and Fraas to those of local workers such as Broom, * Manuscript submitted in January 1946. Appendix added in October 1950. VOL. XXXI, PART 4. 34 FEB 7 195% 394 Annals of the South African Museum. Haughton, Dart, van Hoepen, Middleton Shaw and others. During this period a considerable body of material has been collected, but much of it has remained largely unstudied and undescribed unless something obviously new was noticed by the individuals through whose hands it passed. Even then it has been for the most part only the new genera and species which were described, and the fauna as a whole has received little attention. Van Hoepen has large collections from his site at Cornelia which still await description, the South African Museum at Cape Town and the McGregor Museum at Kimber- ley have hundreds of specimens collected over a long period, and other museums have smaller quantities of undescribed material. Since its inception in 1934 the Archaeological Survey of the Union has acquired notable collections as a result of the activities of several collectors, and little of this material has been described or considered as a whole. During the years 1935-36 a joint survey of the Vaal River basin was carried out by Messrs. P. G. Séhnge and D. J. L. Visser of the Union Geological Survey, and Professor C. van Riet Lowe, Director of the Archaeological Survey, and during the survey much fossil material was recovered from various horizons in the deposits. The results of the geological and archaeological investigations were published in 1937, under the title “The Geology and Archaeology of the Vaal River Basin”’, as Memoir No. 35 of the Union Geological Survey, and it was intimated in the letter of transmittal of this Memoir that the fossil material collected would be described at a later date. Through the courtesy of the Director of the Geological Survey all this material was placed in the hands of the present writer for examination and report.* It soon became apparent that a description merely of the material comprising this collection would be of little value, since it would not include all the species recorded from the deposits, and also because there occur in Pleistocene deposits other than those of the Vaal River basin many species which are likely at any time to be found within this area. For example, a tooth found at Christiana by a student and brought to the writer while this account was in course of preparation has been identified as belonging to a species hitherto recorded only at Cornelia in the Orange Free State. It was also only too obvious that the study of our fossil mammals had been far from systematic and that, with the notable exceptions of Haughton and Shaw, writers had been concerned more with placing new fossils on record than with comparative studies and revision of our knowledge in the light of * See Appendix. A Critical Revision of the Quaternary Perissodactyla. 395 further discoveries. There exists a very considerable confusion of nomenclature and a multiplicity of specific names which renders the task of description of additional material virtually impossible unless it is accompanied by an amount of concurrent specific revision which would obscure the value of any account of the faunal assemblages. Indeed, it appears that until the material already described has been reviewed and new assessments made of the described species, it is of little value to proceed with the many other problems which our fauna presents. In the course of his investigations on the cave deposits, Broom has to a certain extent reviewed and revised the Primates, Insectivora, Rodentia and Carnivora. Furthermore, these orders furnish the bulk of the cave fauna, and are virtually unrepresented in the material from open sites. Most of the fossils recovered from the Vaal River basin, surface deposits and other open sites belong to the Perisso- dactyla, Proboscidea and Artiodactyla. It has accordingly been decided that before the undescribed material can profitably be dealt with, the described species of each of these three important orders of mammals in Southern Africa must be critically reviewed. The present paper considers the first, and perhaps the most important, of these orders.* As much as possible of the material from the Vaal River basin and elsewhere in the possession of the various museums in Southern Africa has been obtained on loan and examined in addition to the large collection in the Archaeological Survey. Dr. Broom has also been kind enough to make his material available to the writer. With the exception of those specimens which are in other countries, the type specimens of every species have been studied and are figured in the present account. Many of these figures have been drawn by the writer from the original specimens where the published figures are considered inadequate or unsuitable; others are reproductions of the original figures. As far as is possible new fossil material has deliber- ately not been introduced in this paper, as the purpose is to revise the specific descriptions from type material or from such other specimens as can with reasonable certainty be identified with the types. These descriptions are for the most part new, and are based on a fresh assessment of the original specimens interpreted in relation to the wider assortment of material now available and considered against a background of comparative researches upon the characteristics and variability of related living forms. It is hoped that this revision and * See Appendix. 3 96 Annals of the South African Museum. correlation of scattered data may help to place our knowledge of these fossil mammals in Southern Africa on a firmer basis than has hitherto been the case, and serve to evaluate the characters of the revised species in a form which may facilitate future identification and comparison. MATERIAL. As a result of the conditions which prevailed in Southern Africa in the Quaternary, the fossil remains which have survived are mainly teeth, though skull fragments, loose bones and horn cores are some- times found under suitable circumstances. Almost all the described fossil mammalian species from this region have been named on teeth, and in the present account the descriptions of species generally give only their dental characters. Complete skulls are so rare that they need not be considered, and the identification of isolated bones is not yet possible, largely owing to the complete absence of whole or even partial fossil skeletons. The majority of the published descriptions have been rather un- systematic and often scanty. It has therefore been considered generally advisable not to quote the original account but to use it and the specimens in the preparation of a new description. Where quotations are given, however, the quoted material is indicated by the usual signs. The type specimens of each extinct species are figured and, in the descriptions given in the text, it has accordingly been possible to give an account of the features of the species which may be used for identification rather than a mere list of the character- istics of the particular specimen or specimens. An attempt has also been made here to give a definition of the generic characters of any extinct genus which has not been defined by its founder separately from the description of the genotype species. Where the genus is represented only by a single species, or perhaps by two species, this generic definition obviously may require considerable future revision, but some care has been taken to select as generic characters only the most outstanding features which differentiate the material from related types. The synonymy given for each species is as full as possible and, it is hoped, includes reference to all the specific designations given to fossil representatives of each described species. In the case of living species, however, the synonymy gives the reference to the type description only, and the further synonymy of the living forms can - A Critical Revision of the Quaternary Perissodactyla 397 be found in the ordinary zoological sources: * the names applied to petrified specimens now assigned to these living forms are, however, given as fully as possible. The accepted specific names of all species living within historic times are followed by an asterisk, to avoid their confusion with species extinct prior to historic times and known only as fossils. In the descriptions of dental characters the nomenclature of the Cope-Osborn scheme has been followed as far as possible, in accordance with the generally accepted information available. Since there appears, however, to be some lack of uniformity in the conclusions regarding the homologies of the cusps, and there is occasionally some doubt regarding the application of certain terms, the terms applied in the present account are shown in a diagram of a typical member of each family described, or are clearly defined in the text. For convenience and brevity the customary abbreviations are used to denote molar, premolar, canine and incisor teeth. Milk teeth are, on the whole, of little value for specific identification owing to their considerable variability and the uncertainty of the relationship between their characters and those of the permanent dentition. They are accordingly considered only when absolutely necessary. In some cases a species is regarded as unrecognisable owing to the inadequacy of the material upon which it is founded or defined, and thus becomes a species insuff. descr. aut inqguirenda. It may neverthe- less be the case that a specimen which has been regarded as in itself inadequate for the creation of a species may be capable of reference to more adequate material. The numbers assigned by the various museums to the type and other specimens mentioned in this text are given wherever possible, together with an abbreviation indicating the museum concerned, viz.: Arch. Sur. Archaeological Survey, University of the Witwaters- rand, Johannesburg. Dept. Anat. Department of Anatomy, Medical School, University of the Witwatersrand, Johannesburg. M.M.K. | McGregor Memorial Museum, Kimberley. Nas. Mus. Nasionale Museum, Bloemfontein. S.A. Mus. South African Museum, Cape Town. Tvl. Mus. Transvaal Museum, Pretoria. * The most up-to-date synonymy is that given in “A Checklist of African Mammals’’, by G. M. Allen, Bull. Mus. Comp. Zool. Harvard, vol. Ixxxii, 1939. 398 Annals of the South African Museum. ORDER PERISSODACTYLA. THE RHINOCEROSES. Amongst the fossil mammalia, the rhinoceroses are only very scantily represented in Southern Africa. Two supposedly extinct forms have been mentioned, each only from a single site, but petrified specimens indistinguishable from the two living species have been recovered from various superficial deposits. These latter specimens are probably not of any very great antiquity, but rhinoceros species are in any case not subject to rapid changes. The two living forms belong to different genera, and both their skulls and their teeth are quite distinct. The square-lipped or white rhinoceros is quite considerably larger than the hook-lipped or black rhinoceros, as can be seen from the drawings of their respective skulls (fig. 1). The lower jaws are sharply distinguished, that of the black rhinoceros having a deep compressed symphysis as compared with the depressed and rather spatulate symphysial region of the mandible in the white rhinoceros. The horns have been found isolated and again differ widely in form. Incisor and canine teeth are rudimentary or absent in both species. The cheek teeth in the rhinoceroses comprise four premolars and three molars arranged in a continuous series and having essentially the same structure, though the first premolar is considerably more simplified and is shed early. The lower third molar is also simple, and does not possess the third lobe so characteristic of the horses and most artiodactyls. The premolars are somewhat smaller than the true molars, the second premolar and first premolar particularly being smaller than the more uniform succeeding teeth. Structurally the teeth differ from those of the horse in being rather low crowned and in possessing strong, distinct roots, but their essential composition is similar to that of the equine cheek teeth. The normal order of eruption of the permanent dentition appears to be M!, Pm1, Pm?, M2, Pm’, Pm, and lastly M3, and is thus somewhat different from that of the horses. The cheek teeth of the rhinoceroses are lophodont in form, 7.e. the rows of cusps tend to become fused into ridges. In the upper teeth the two main outer cusps form a ridge known as the ectoloph, two anterior cusps form the protoloph and two posterior ones the metaloph. In the lower teeth three triangularly arranged cusps unite to form a crescentic metalophid, and posterior to this two cusps form an arcuate A Critical Revision of the Quaternary Perissodactyla. 399 oO (0 20 30 40 50 690 Centimet re Seale Fig. 1. Above: Lateral view of skull of Ceratotheriwm simum* (Burchell) and plan view of spatulate symphysial region of the lower jaw. (After Sclater.) Below: Lateral view of skull of Diceros bicornis* (Linnaeus) and plan view of the compressed symphysial region of the lower jaw. (After Owen.) 400 Annals of the South African Museum. hypolophid ridge. With wear the enamel is rapidly removed from the top of these ridges and an area of dentine surrounded by enamel is exposed. This can be clearly seen in fig. 2, in which typical upper and prefossette Metafiexid Entoflexid Fie. 2.—Molar elements (following Osborn) of the upper and lower cheek - teeth of the Rhinoceros group. Abbreviations. Upper teeth: pas, parastyle; pa, paracone; me, metacone; hy, hypocone; pr, protocone; pel, protoconule; mcl, metaconule. Lower teeth: prd, protoconid; hyd, hypoconid; pad, parastylid; med, meta- conid; end, entaconid. (Original.) lower first molars are shown indicating the nomenclature used for the cusps, folds and ridges (following Osborn). The valley between the protoloph and metaloph appears to have received no name, and is here termed the medivallum by analogy with the corresponding valley in horse teeth. For the two inlets in the lower teeth the terms ““metaflexid”’ and “‘entoflexid” are here suggested for convenience A Critical Revision of the Quaternary Perissodactyla. 401 in description, as the same terms have been proposed by Stirton (1941) in the lower teeth of the horses. Owing largely to a lack of sufficient material, it has unfortunately not been possible to gain any reliable idea as to the constancy or variability of the tooth characters within the series in the rhinoceroses. From the limited material examined, however, it does appear that while the essential structures are reasonably constant, the effect of attrition alters the pattern of the grinding surface to such a degree that identification may be made most difficult. As wear proceeds, the ridges widen and obliterate the intervening valleys, at first fairly slowly, but afterwards very rapidly, until ultimately a uniform tract of dentine may be produced. The crochet, antecrochet and crista which project into the medivallum are generally more prominent in early wear, and are reduced in size as this valley is narrowed. In some species the crochet and crista may unite and isolate the medifossette as an accessory valley, leaving the prefossette as the terminal portion of the medivallum. The postfossette may also become isolated by closure of the posterior enamel border. In the lower teeth the chief effect of attrition is to reduce the size of the two flexids, the metaflexid in particular tending to disappear with wear. Fusion of the meta- conid and entaconid may also lead to the complete isolation of the entoflexid as an accessory valley. Famity RHINOCEROTIDAE. Genus Dicrros Gray 1821. Genotype: Rhinoceros bicornis* Linnaeus. Diceros bicornis® (Linnaeus). Rhinoceros bicornis®* Linnaeus 1758. Syst. Nat. Ed. (10), i, p. 56. Opsiceros simplicidens (pars) Scott 1907. 3rd Rep. Geol. Surv. Natal and Zululand, pp. 258-259, pl. xvu, figs. 4, 5. Diceros whitei (pars) Chubb 1907. Geol. Mag., V, vol. iv, pp. 447-448. The horns of the black rhinoceros are almost invariably two in number, but exceptionally as many as five have been recorded. The anterior horn has a height of about forty-five to sixty centimetres on the average and has a basal diameter fifteen to twenty-five centi- metres. The rear horn is about one-third to one-half the length of the anterior one and has a diameter only a little less than its height. The record horn lengths are about double the average figures. Both horns are rather blunt and curve very slightly posteriorly. In the upper jaw the first premolar is very small and exhibits no 402 Annals of the South African Museum. structures which can be said to be recognisable as persistent. The third molar is triangular in form, the metaloph, being reduced to a small posterior prominence only, and this tooth is very variable in pattern in wear and is of little value for specific identification. The remaining three premolar and two molar teeth are generally more consistent, and are essentially similar in structure, though the pre- molars differ slightly from the molars. In the premolars the anterior wall of the protoloph is fairly straight and makes an angle of about 75° with the ectoloph, whereas in the molars the protoloph initially makes almost a right angle with the ectoloph, and then curves some- what posteriorly. The protoloph and metaloph are roughly parallel or slightly divergent and with the ectoloph give the appearance of the Greek letter 7. The ectoloph itself is not straight, but has an outer wall incurved or grooved between the paracone and metacone and also has a shallow groove behind the parastyle. The parastyle itself is commonly anteriorly grooved and projects very little in front of the protoloph. The antecrochet is apparently absent, and the crista is very small and disappears rapidly with attrition. A crochet is always present in the earlier stages of wear, and in the normal dentition increases progressively in size from the second premolar to the second molar. It tends to become rounded with increased wear, and may disappear completely before the medivallum is obscured. In no case has isolation of the medifossette been observed in this species except in the third molar. The postfossette is somewhat obliquely V-shaped, tending to be U-shaped with wear as a result of the expansion of the hypocone lobe of the metaloph, and then becomes isolated as an oval valley. The dimensions vary considerably with wear, the breadth across the grinding surface increasing as attrition proceeds. The height above the basal cingulum increases progressively with the successive teeth, and a typical second molar in early wear has a height of about 50-55 mm. The breadth at the base of the second molar is about 60 mm., but in normal wear the grinding surface measures only some 45 mm. transversely. The size and characters can be seen from the scale drawings in fig. 3. Two typical upper dentitions are shown, one in fairly early wear, the other well worn and lacking the first premolar. The lower teeth have little to distinguish them from the very generalised form of most rhinoceros teeth. The first premolar is greatly simplified in form, but the remaining teeth, including the third molar, are similar in structure. The anterior and antero- external walls of the metalophid are markedly flattened, and make an A Critical Revision of the Quaternary Perissodactyla. 403 < Fic. 3.—Two left upper dentitions (A and B) of Diceros bicornis® (Lin- naeus) and a right lower dentition (C) of the individual B. One-half natural size. (Original.) 404 Annals of the South African Museum. angle with each other of about 100° ora little more. A fairly marked groove separates the outer wall of the metalophid from the curved hypolophid walJl. The inner walls of the metaconid and entaconid are somewhat flattened. The metaflexid is a good deal smaller than the entoflexid and is rapidly reduced to a shallow V-shaped notch. The height of a normal second molar is about 50mm. The lower teeth of a typical specimen are shown in fig. 3, and belong to the same indi- vidual as the upper dentition figured immediately above it. Referred Material. Apart from the petrified specimens from superficial deposits in various parts of Southern Africa which obviously belong to this species, two of the teeth from Zululand which Scott (1907) very tentatively referred to his species Opsiceros simplicidens do not appear to warrant distinction from the living Diceros bicornis*. The type LM? of Scott’s species manifestly is not that of D. bicornis*, but the two heavily worn teeth (M! and Pm+*) do not differ appreciably from correspondingly worn teeth in old individuals of the living black rhinoceros. Scott himself realised the close similarity, and suggested that these two teeth did not actually belong to his new species. The specimens themselves have not been seen by the present writer, but natural size photographs kindly supplied to the writer by Professor Scott, together with the admirable description, form an adequate basis for the conclusion reached above. Genus CERATOTHERIUM Gray 1868. Genotype: Rhinoceros simus* Burchell. Ceratothervum simum* (Burchell). Rhinoceros stmus* Burchell 1817. Bull. Sci. Soc. Phil. Paris, F. 1, 2, 105 Sil Opsiceros simplicidens Scott 1907. 3rd Rep. Geol. Surv. Natal and Zululand, pp. 257-258, pl. xvii, fig. 3. Rhinoceros scotti Hopwood 1926. Occ. Papers No. 2, Geol. Survey, Uganda, pp. 16-17, fig. 3. The white rhinoceros possesses a long and rather slender anterior horn which attains a height of about ninety centimetres and, exception- ally,as much as a hundred and fifty centimetres.* The second or rear * Southern race: 624 inches. Rowland Ward’s Records of Big Game, 9th ed., 1928, p. 446. A Critical Revision of the Quaternary Perissodactyla. 405 horn 1s small and does not usually attain a height exceeding twenty-five centimetres. The anterior horn is normally more slender than that of the black rhinoceros. In the upper jaw the first premolar is small and is shed early, but the other premolars differ notably from the molars, for in the former the medivallum becomes rapidly enclosed by fusion of the proto- cone and hypocone, the form of the metaloph being consequently also affected. From Diceros bicornis* they differ most markedly in the arcuate, posteriorly curving protoloph, and in the early fusion of the well-developed crista with the crochet to form an isolated medi- fossette. Due to the posterior curving of both protoloph and metaloph the m-like shape of the teeth is very distorted and not nearly as noticeable as in D. bicornis*. The form of the ectoloph is also different, being rather more undulate in Ceratotherium simum* with a distinct outward bulge at the paracone. The parastyle is more pointed and is not apparently grooved, but there is a groove immedi- ately behind it. The postfossette becomes isolated with wear as a result of closure of the posterior enamel. As in D. bicornis* this isolation of the postfossette is not a constant feature and is generally less marked in the molars than in the premolars. The teeth are higher crowned than those of the black rhinoceros, a typical second molar in normal wear measuring about 75 mm. above the indistinct cingulum. Fig. 4 shows (half natural size) two upper dentitions, one in very early wear with the third molar only just erupting and the fourth premolar coming into use, the second dentition being in a more advanced state of attrition. The lower teeth are somewhat difficult to distinguish from those of the black rhinoceros. The anterior and antero-external walls of the metalophid make an angle with each other close to 90° instead of the obtuse angle found in D. bicornis*. The metaflexid appears to be more persistent in the white rhinoceros than in the black, and the enclosure of both metaflexid and entoflexid with advanced wear is a common feature. The height of a typical second molar above the cingulum is about 60mm. The lower dentition of the same individual as the more worn upper dentition figured is shown in fig. 4. Referred Material. Petrified specimens of this species have been found in various superficial deposits in the coastal region and in the interior. A portion of an anterior horn is also recorded from a cave deposit near Kuruman, Cape Province (Malan and Cooke, 1941). 406 Annals of the South African Museum. Fie. 4.—Two left upper dentitions (A and B) of Ceratotherium simum* (Burchell) and a lower right dentition (C) of the individual B. One-half natural size. (Original.) A Critical Remsion of the Quaternary Perissodactyla. 407 W. B. Scott reported in 1907 the discovery of a supposedly extinct species in fossiliferous marine clays from the Zululand coast, and he named this species Opsiceros simplicidens, with an unworn upper left second molar as the type. Scott compares this tooth with the corresponding one of Diceros bicornis*, and states that the differences are “clearly of specific value”. The distinctive characters are stated ‘to lie, enter alia, in the much stronger recurving of the protoloph and in the much better development of the crista, which “fuses with the anticrochet so as to enclose a small and apparently shallow fossette’’. (Scott here used the term ‘anticrochet’ in error for crochet, as his figure shows, and in this follows an error previously made by Osborn in describing the Perissodactyla of White River (Scott and Osborn, 1890).) The characters which Scott used to differentiate the species from D. bicornis* are exactly those which distinguish the white rhinoceros. A skull of the latter species in the South African Museum possesses a second molar in much the same state of development, and the dimensions and appearance of this tooth correspond very closely to the data and figure furnished by Scott. There can thus be little doubt that Opsiceros simplicidens is a synonym of Ceratothervum simum*, a fact which Scott would undoubtedly have realised had comparative material of this rather rare species been available to him. In 1926 Hopwood recorded an upper left second molar from the Kaiso beds of Uganda and assigned this specimen to Scott’s species. He also pointed out that the name R. simplicidens was preoccupied and proposed Rhinoceros scotti as a substitute. There seems no doubt of the correctness of the reference of this specimen, and equally there is little doubt of its similarity to teeth of the living white rhinoceros. Rhinoceros scotti is thus also apparently a synonym of Ceratotherium sumum*. DISCARDED SPECIES. Diceros white Chubb. Diceros whitei Chubb 1907. Geol. Mag., V, vol. iv, pp. 447-448. Diceros whitei Hopwood 1928. Rhodesian Man and Associated Remains. A supposedly new species of rhinoceros was described very briefly by E. C. Chubb in 1907 in a “List of Vertebrate Remains” from the Broken Hill Cave. It was founded on two limb bones (a right tibia and a right humerus) which had been excavated by Mr. Franklin 408 Annals of the South African Museum. White and presented by him to the Rhodesian Museum. This species was also mentioned by A. T. Hopwood in the British Museum memoir on Rhodesian Man, and is there said to be “closely allied to D. stmus”’. In view of the uncertainty of the generic position of the species whitet, the material was obtained on loan from the Rhodesian Museum, Bulawayo, and permission was obtained from Mr. Chubb to amplify his preliminary description and to figure the specimens. In his brief account Chubb remarks on the scantiness of the comparative material available to him, and this lack and the seeming association of the two bones appears to have resulted in an error in the distinction of the material. The tibia is certainly that of a rhinoceros, though com- parison with recent skeletons shows no notable differences in size or in other characters from the corresponding bone in the living Diceros bicornis*. The humerus, however, differs very considerably from both the living rhinoceroses, and it would appear that it is an artio- dactyl and not a perissodactyl humerus, the differences formerly regarded by Chubb as of specific distinctness being actually too great for that possibility to be upheld. The compressed narrow olecranon fossa is a normal artiodactyl feature unlikely to occur in a rhinoceros, and the deltoid ridge and deltoid tuberosity are also much more artiodactyl than perissodactyl. With these views Mr. Chubb now expresses his agreement. On comparison with various living artiodactyls, the closest resem- blance is found between the fossil humerus and that of the living Cape Buffalo. There is no great difference in length, but the fossil bone is somewhat more massive, with the attendant minor modifications consequent upon its greater weight-supporting requirements. Other- wise, however, there is a very close agreement in every character, and it seems highly probable that the fossil humerus belongs to a member of the Buffalo group. It may possibly belong to the extinct “ Bubalus”’ baini Seeley, or to “‘Bubalus” andersoni Scott. The species Diceros whiter appears, therefore, to have been founded on a humerus which is not that of a rhinoceros and on a tibia which does not warrant distinction from the living D. bicornis*, so that D. whitei must be regarded as incorrectly founded. THE Horsks. There have been described at various times from Southern Africa more than twenty-five species belonging to this family, some based on upper and some on lower teeth, but of these not more than half can be A Critical Revision of the Quaternary Perissodactyla. 409 regarded as valid. The position was first reviewed by Haughton in 1931 when the twenty then existing species were reduced to eleven, and in general the present writer is in agreement with these con- clusions. Haughton divided the members ascribed to the genus Equus into two groups, which he terms the “quagga”’ group and the “zebra” group on the parallelism of certain characters with those in the teeth of the two living species, the bontequagga and the mountain zebra. Unfortunately these characters in the recently extinct true quagga are very different from those in the living bontequagga and the two group terms must therefore be abandoned. The relationships indicated by Haughton’s work are of great interest and, in order the better to appreciate the definitive characters of the teeth in the extinct forms, the writer has carried out an extensive examination of skulls and teeth of the living forms and of the recently extinct true quagga. As a result of this work it is possible to dis- tinguish on dental characters from this material three undoubted species: Hquus zebra*, the living mountain zebra, Equus quagga*, the recently extinct true quagga, and Equus burchellii*, the living bontequagga or Burchell’s zebra. Since zoologists have been greatly at variance on the status of these forms, and since all three species occur in the fossil state, the results of the investigation have already been considered fairly fully (Cooke, 1943). These observations also throw some light on the morphological characters and variations encountered in equine species and are of great value in considering the fossil finds. To some extent they repeat and amplify the work of Gidley (1901), and in the present examination a general agreement was found with the conclusions outlined by him. For convenience of reference the nomenclature of the important elements of the molar teeth of the Equidae (following Osborn) is given here in diagrammatic form (fig. 5). The specimens figured are upper and lower fourth premolars, and show the appearance of the cusps on the unworn crowns and the enamel patterns of the teeth in normal wear. The two enamel islands in the upper cheek teeth have long been known as the pre- and postfossettes, but the partial islands, or inlets, in the lower teeth have until recently received no name. The terms advocated by Stirton (1941) are used here. They are re- spectively “‘metaflexid’”’ for the anterior and “entoflexid” for the posterior partial islands of the lower cheek teeth. (These terms have already been suggested for the analogous parts of the rhinoceros teeth.) It is also proposed here to call the posterior groove which lies between the hypocone and the hypostyle in the upper teeth the “hypoglyph”, VOL, XxXxt, PART 4, 3D 410 Annals of the South African Museum. and the groove anterior to the protocone the “protoglyph”. These terms are preferred to ‘“‘hypoconal groove” and “preprotoconal groove’ used by Stirton. Ectoloph a %, Medivallum pli pretossette pli postfossette protoloph pli hypostyle pli protoconule Hypoglyph pli caballin Metaflexid Entorlent. Fig. 5.—Molar elements (following Osborn) of the upper and lower cheek ARG RCDATIONS teeth of the Horse group. Upper teeth: pas, parastyle; mss, mesostyle; mts, metastyle; pa, paracone; me, metacone; hy, hypocone; pr, protocone; pel, protoconule; mcl, metaconule; hys, hypostyle. Lower teeth: prd, protoconid; hyd, hypoconid; med, metaconid; end, enta- conid; pad, parastylid; msd, metastylid; hld, hypoconnlid; ec.sd., ectostylid (fold or ridge); esd, entostylid. (Original.) Distinction between Zebra, Quagga and Bontequagga. The writer has examined a large number of skulls of these three species in the past few years, though skulls of the recently extinct ~ Sinai anal A Critical Revision of the Quaternary Perissodactyla. 411 quagga are rare and difficult to obtain. Owen (1869) figured one in a little known paper, there is a cast in the Transvaal Museum, one skull in the Kingwilliamstown Museum and several in the McGregor Museum, Kimberley. Based on an examination of this material the chief distinguishing features have been analysed (Cooke, 1943). From the point of view of classification it would appear that the three species are quite distinct. The bontequagga or Burchell’s zebra, while showing skull characters intermediate between the other two forms, is sharply distinguished in dental characters. There can thus be no doubt that the bontequagga is very distinct from the true zebra despite the many similarities of form and colouring which have caused much dispute amongst zoologists with respect to the relationship of the two forms. It differs to an equal degree from the true quagga, and can under no circumstances be regarded as a variety of this form. The specific designation guagga* must therefore be restricted to the historically extinct true quagga, and the bontequagga or Burchell’s zebra must receive the full specific name burchellii*, to which may be appended, if it is considered necessary, the varietal names wahlbergi*, transvaalensis*, etc., though these varieties cannot be distinguished on skeletal or dental grounds as far as the present writer is aware. The Degree of Constancy and Range of Variability of the Tooth Characters. Aside from the examination of a large number of skulls of the zebrine group of horses to determine their distinctive features, an attempt has been made to estimate the range of variation found within each species and hence to estimate the value of each possible factor in specific determination. A number of skulls of Equus caballus* and Equus asinus* have also been used for this purpose, and reference has been made in addition to many published figures of equine dentitions to ascertain how widely the generalised conclusions may be applied. The factors appear from this examination to follow certain definite trends, a knowledge of which greatly enhances the value of the specific identification of individual finds of isolated teeth. It must be stated however that, despite this knowledge, a fossil species which is named purely on dental characters may well be a “form” species only, as there is a certain amount of overlap in the extreme variations of certain species. The first upper true molar of Hquus quagga*, for instance, may be almost identical with that of Equus zebra* under certain conditions of variation. 412 Annals of the South African Museum. One of the difficulties which besets the worker on the fossil Equidae is the fact that almost all the finds consist only of isolated teeth. Complete dentitions are very rare and are consequently of immense value. It is of the utmost importance therefore to ascertain correctly the position occupied by the isolated teeth in the former jaw in order correctly to evaluate the determinative characters. By far the most satisfactory method is that of direct comparison with a known complete dentition. The angle of wear and degree of antero-posterior curvature are the chief guides in the estimation as to whether a particular tooth is, for example, a fourth premolar or a first molar. Fig. 6 gives an indication of the general shape of the teeth in Equus burchellii*, and figs. 6 and 19 show the form in some large extinct equines. The second premolar is at once distinguishable by being rather pointed anteriorly, and the third molar shows a posterior tapering. The last erupted tooth of the series at any given age shows this posterior taper, however, so that in early wear such a tooth might possibly be a second or even a first molar. The rather sharp curvature of the true third molar should, however, facilitate the distinction of this tooth. In the lower teeth the premolars typically have a larger entoflexid than the molars, and the outer groove between hypoconid and proto- conid extends further towards the inner groove between metaconid and metastylid in the true molars than it does in the premolars. In the upper teeth the first true molar normally has a rather narrow mesotyle, but otherwise the degree of curvature and angle of wear form the only guide to the position of the tooth in the series. The factors which affect the characters of each cheek tooth in an individual of a species are: (A) The position of the tooth in the series, (B) the degree of attrition, and (C) variability within the species. The factor of sex does not appear to exert any appreciable influence on the cheek teeth, though the canines, which are usually prominent in the male, are vestigial or absent in the female. The generalised con- clusions regarding the three main factors given above may therefore be discussed in turn. A. THE PosIrion oF THE TOOTH IN THE SERIES. (a) In the Upper Teeth. 1. The anterior and posterior teeth (second premolar and third molar) show decidedly different forms of enamel pattern due to their tapering nature and are very variable in character. They are of the least possible value in specific determination, as the prefossette Above: Below: Fic. 6. Lateral view of skull of Equus burchellii* (Gray) showing the shapes of the cheek teeth. (Original.) Lateral view of fragmentary remains of skulls of large extinct equines; Maxilla of Hquus fowleri Wells, and mandible of Equus plicatus (van Hoepen). (Original.) ~—eow @ 2 2 @ =a ~~. 414 Annals of the South African Museum. in the second premolar and the postfossette in the third molar are subject to considerable distortion. 2. In transverse breadth, measured from mesostyle to protocone, the third and fourth premolars are approximately equal, and the first and second molars usually slightly narrower (by perhaps 5-10 per cent.) than the preceding teeth. The second premolar and third molar are always narrower than the other premolar or molar teeth respectively, but bear no constant ratio of relative size. 3. The fourth premolar presents the most complex and the first molar the simplest enamel folding.* 4. The protocone is shortest (antero-posteriorly) in the second premolar and longest in the third molar. The length of the protocone either increases progressively in each succeeding tooth of the series or else the first molar has a smaller protocone than the fourth premolar, and the second molar has one about the same length as in the fourth premolar. 5. The mesostyle is generally reduced in width in the molars as compared with the premolars. 6. The pli-protoloph and pli-hypostyle may be lacking in the first molar though present in the remaining teeth. 7. The pli-caballin may be present in the premolars yet absent in the molars, and may occur in the fourth premolar only. 8. The hypoglyph is often absent in the third molar. (b) In the Lower Teeth. 1. As in the upper teeth, the second premolar and third molar are of sub-triangular form and show departures from the normal enamel pattern. The changes are not always so great as to prevent specific identification, but they are the least useful teeth of the series for this purpose. 2. In transverse breadth the premolars are wider than the molars by about 5-10 per cent. The second premolar may, however, be as narrow as the first molar. The third molar bears no reliable ratio of width to the other molars, and in the second molar the posterior breadth is less than the anterior due to the reduction in size of the hypoconid. * ‘The normal order of eruption of the cheek teeth is M2, M2, Pm?, Pm’, Pm4, and lastly M%, though the order of the last two may exceptionally be reversed. In a normal dentition, therefore, the first molar in addition to possessing an initially less complex character is also the most worn tooth. A Critical Revision of the Quaternary Perissodactyla. 415 3. The entoflexid has a larger lobe in the premolars than in the molars, this being generally of greatest antero-posterior length in the second premolars. 4. The outer groove between protoconid and hypoconid is deeper in the molars than in the premolars and sometimes meets the internal groove between metaconid and metastylid. (c) Relation between Upper and Lower Teeth. Of great interest and value is the hitherto apparently unrecognised fact that the breadth over the enamel in the upper teeth bears a fairly constant ratio to the breadth of the lower teeth over the enamel. This ratio is approximately 1: 0-6 for the third and fourth premolars to _1: 0-55 for the first and second molars, and is apparently subject to a variation of only about 5 per cent. B. Tut Decree or ATTRITION. I. Changes in Dimensions. The effects of the degree of wear (or age) on the cheek teeth have been well described by Gidley (1901), and the following selected quotations from his paper are illustrative of his findings in this regard. 1. “When a molar or premolar tooth first comes into use, the face, as well as the sides of the crown, is completely covered with enamel which folds in and out, and (though somewhat hidden by cement) presents the same general appearance as that seen in the much more primitive forms Anchitherium and Mesohippus. Soon the enamel on the tips of the cones and along the ectoloph wears through, and small patches and ridges of dentine surrounded by a border of enamel are exposed. As the tooth is further worn away, these patches and irregular ridges broaden and rapidly lengthen until when about one- half to three-quarters of an inch of the crown has been worn away they have all become united by narrow isthmuses, and the fundamental tooth pattern of the horse is presented.” “From this point... the triturating surface presents a gradually less complex pattern of enamel folding as the tooth crown is worn away, until in the very much worn tooth the simplest pattern of enamel folding is presented” tee 0) , 2. ! The antero-posterior diameter of the first premolar (p,)* remains about the same for the whole length of the crown, except that sometimes it narrows slightly near the roots.” * Actually the true second premolar according to present usage. 416 Annals of the South African Museum. Fic. 7.—Sectioned right upper fourth premolar of adult male Equus caballus* Linnaeus, illustrating changes in enamel pattern which would result from various stages of attrition. Natural size. (Original.) 3. ‘The antero-posterior diameter of the last molar (m3), however, is relatively small at first, and increases continually as the tooth is worn away.” 4. “The antero-posterior diameter of the grinding surfaces of all A Critical Revision of the Quaternary Perissodactyla. 417 the intermediate teeth are greater at the stage when the tooth has just fully come into use; ...from this point the antero-posterior diameter diminishes very rapidly for a short distance and then con- tinues to diminish more gradually to the roots of the tooth.” 5. After the tooth has just come fully into use “the transverse diameters of p* to m? inclusive remain about the same, diminishing slightly near the roots; p* gradually diminishes while m? increases in transverse diameter as the crown wears away.” 6. “The antero-posterior diameter of the protocone in all the teeth of the series remains the same for the whoie length of the crown.” 7. “Owing to the very slight variation of the transverse diameters of the crowns of p* to m? inclusive . . . and to the great shortening of their antero-posterior diameters, the ratio of these diameters is very different in old and in young individuals of the same species. Thus in the little worn condition . . . the antero-posterior diameter is always greater than the transverse. As the crown wears away... a stage is reached where the two diameters are about equal, then, as the antero-posterior becomes still more shortened, the transverse exceeds it. In every series this variation in ratio seems always to be more advanced in m! and m?.” The present writer has not observed any conflict with these views. The transverse diameter (or “breadth’’) of the crown and the antero- posterior diameter (“‘length’’) of the protocone do, however, decrease to a very slight degree with wear, though their ratio remains constant. Gidley made no comments on the precise effects of wear on the com- plexity of the enamel folds, and the present writer has therefore attempted by means of serial sections of a number of teeth to ascertain which folds are most affected by wear and which least. As a result, the following generalisations, though based on comparatively little material, appear to be possible. il. Changes in Enamel Complexity. (a) In the Upper Teeth. 1. The pli-protoloph may disappear completely when the tooth is as little as half worn, and is always the most affected of the fossette folds. 2. The pli-hypostyle and pli-prefossette suffer rapid reduction in size but have not been observed actually to disappear. 3. The secondary small plications in the regions of the pre- and postfossette folds proper disappear rapidly if they are of an angular or saw-tooth nature, but may survive a considerable degree of wear if they have the form of rounded loops. 418 Annals of the South African Museum. 4, The pli-protoconule and pli-postfossette are the least affected of the fossette folds. 5. The hypoglyph becomes shallower with wear but does not normally disappear. 6. The pli-caballin does not disappear with wear in the premolars unless initially very small, but may do so in the molars. 7. The halves of the ectoloph become slightly flattened with advanced wear. (b) In the Lower Teeth. 1. The folding of the entoflexid lobe may be very complex in the earliest stages of wear, but becomes rapidly simplified and does not normally survive the half-worn condition of the tooth. C. VARIABILITY WITHIN THE SPECIES. I. Transverse Diameter of the Crown. As Gidley has shown, the transverse diameters over the enamel of the teeth (excluding the second premolar and third molar) are subject only to relatively slight variation within a species. In a series of ten skulls of Equus caballus* ranging from a large draught horse to a small Texas pony, he notes a difference in the transverse diameter of the second molar of only 2-5 mm. or about 10 per cent. In the third premolar of the same individuals, however, his figures show a difference of 4-5 mm. or nearly 20 per cent., and this range of variation agrees with the present writer’s findings in the case of the zebrine group. In the many skulls of Equus burchellui*, Equus zebra* and Equus quagga* which have been examined, the extreme variations in trans- verse diameter amount to about 20 per cent. of the mean value, but at least 70 per cent. of the specimens in each species are very close indeed to this mean value. The transverse diameter of the cheek teeth is thus subject to a variation of about plus or minus 10 per cent. from a predominant mean value, and this value is therefore a useful characteristic of a species (except in the anterior and posterior teeth). The same relative figures apply to the lower teeth. II. Length and Form of the Protocone. (a) The form of the protocone, and particularly the relative anterior and posterior development or elongation, is a very constant character in a species. A Critical Revision of the Quaternary Perissodactyla. 419 (0) The absolute length of the protocone is subject within a species to a variation of as much as 30 per cent. between corresponding teeth. In general, the ratio of the length of the protocone to the transverse diameter of the crown is not affected by wear, and this ratio (which we may term the “protocone ratio”) is not quite as variable as the absolute length of the protocone. In Equus caballus* the average value of the protocone ratio length of protocone transverse diameter of crown is 0-48 in the fourth premolar; in the corresponding tooth of Equus asinus* 0-41; Hquus burchellii* 0-41; Equus zebra* 0-40; Equus quagga* 0-44. Though subject to a variation of plus or minus 15 per cent., and thus having no precise significance since values would overlap considerably, the protocone index might be a useful indication of the relative development of the protocone in a species. It must be remembered, however, that in addition to variation between corre- sponding teeth, the value of the protocone ratio will change from the second premolar to the third molar. In Hquus caballus* average figures are p* 0:38, p? 0-45, p4 0-48, m! 0:47, m? 0-52, m® 0-62. Ill. The Enamel Folds. A wide variation is found in the degree of complexity of the enamel folding within a species, but the following generalisations appear from the writer’s investigations to be applicable to the upper teeth. 1. The caballine fold is not infrequently absent in particular teeth of an individual of a species which normally shows a strong caballine fold. Even though in some species it is apparently never present, it has no precise value. 2. The pli-protoloph is of very variable development and unless very strong is of no specific value. 3. The degree of development of the pli-hypostyle and pli-prefossette, and the secondary plications of the latter, is variable and of slight value only. 4, The pli-protoconule and pli-postfossette appear to be of very constant development and their presence or absence can be considered as of diagnostic importance. 5. In some species the hypocone shows a consistent slight bulge into the medivallum, giving it the appearance of a partial isolation from the metaconule, and this character seems to be of some determinative value. 420 Annals of the South African Museum. 6. The flattening of the inner wall of the protoconule appears to be a character of minor diagnostic value in some species. 7. The depth of the hypoglyph is a moderately constant feature. IV. The Ectoloph and Styles. 1. Except in the second premolar and third molar, the nature of the halves of the ectoloph, more particularly the anterior one, is of fairly constant character, being either concave inwards as seen in Equus caballus* and Equus burchellii*, or else flattened or even convex as in Equus asinus*, Equus zebra* and Equus quagga*. 2. The three styles, parastyle, mesostyle and metastyle, show quite a considerable range of variation in detail within a species in addition to the difference shown between premolars and molars. The characters of the metastyle are of no specific value, and neither the absolute size nor the grooving of the parastyle and mesostyle are entirely constant characters though they may be useful ones. The degree of isolation of the mesostyle and parastyle from the walls of the ectoloph or their easy confluence with them, however, appears to be a valuable specific character. Owing to the normal relative reduction in size of the styles in the true molars, their isolation is less apparent in these teeth than in the premolars. V. The Lower Teeth. To quote from Gidley, “the characters of the lower teeth are, in general, affected in the same way as the upper, and seem to be of even less value in determining the species”. With these conclusions the present writer is in agreement, in that the very simplicity of structure of the lower teeth deprived them of many of the characters which might be of value in making distinctions. The range of normal variation within a species thus permits of considerable overlap between what are actually different species (as is only too clearly seen in the zebrine group of horses), and thus absolute size must be the chief determinative factor. Other useful, though not entirely constant characters, are (a) the shape of the outer walls of proto- conid and hypoconid, (6) the shape and degree of separation of metastylid and metaconid, (c) the size and form of the metaflexid and entoflexid lobes, (d) the relative size of the stylids and conids, (e) the development of the ectostylid fold* in the hypoconid. * Called “pli-caballinid’’ by Stirton (1941). A Critical Revision of the Quaternary Perissodactyla. 421 GENERAL CONCLUSIONS. As a result of the examinations made, second premolars and third molars cannot be regarded as providing adequate material for deter- mination, and no species founded on a second premolar or a third molar can be regarded as valid. If second premolars and third molars are known in a species from certain association with determinable teeth, corresponding teeth showing close agreement can then be assigned to the species with fair certainty. On general grounds of size and complexity, isolated second premolars and third molars may be referred to a species as “‘cf. species”’. In the remarks which follow, only the third and fourth premolars and the first and second molars are regarded as exhibiting characters of value in specific determination. The position of the tooth in the series affects two main characters as follows :— 1. The transverse diameter of the crown, measured over the enamel, is about 5 per cent. greater in premolar than in molar teeth. 2. The enamel pattern is somewhat more complex in the premolar than in the molar teeth, the fourth premolar showing the most complex and the first molar the simplest enamel folding. The mesostyle is somewhat reduced in size in the upper molars, and the entoflexid lobe is smaller in the lower molars than in the premolars. Bearing in mind the position of the tooth in the series, and the effect upon it of the above factors, the following generalisations may be made with regard to the definition and identification of equine species :— 1. The transverse diameter (“‘breadth’’) measured across the enamel is a character of considerable value in determination. The normal departure in size is not more than 10 per cent. and the maximum variation 20 per cent. If the transverse diameter of a tooth departs by more than 20 per cent. from the measurement of the type, wt cannot be regarded as belonging to the same species. If several teeth of a species are known, the possible range which would exclude a specimen will be further reduced and a departure of 15 per cent. may be regarded with suspicion. It must be remembered, however, that different species overlap in point of size, and that size alone is not a sufficient basis for determination. 422 Annals of the South African Museum. A. In Upper Cheek Teeth. The following factors may be regarded as of reasonable constancy and value in determination :— 1. The relative anterior and posterior development and form of the protocone. (The absolute size of the protocone is subject to a variation between premolars and molars of as much as 20 per cent.) 2. The form of the ectoloph and styles. 3. The presence or absence of the pli-protoconule and pli-postfossette and their degree of development. The following factors may be regarded as of confirmatory value :— (a) The partial isolation of the hypocone (or the lack of such isolation). (b) The protocone ratio (subject to a variation of 30 per cent.). (c) The presence of any or all of the following folds, though their absence cannot be regarded as significant :— (i) Pli-hypostyle. (ii) Pli-caballin (particularly in true molars). (iii) Pli-protoloph (least reliable). (d) The form of the inner wall of the protoconule. (ec) The depth of the hypoglyph. B. In Lower Cheek Teeth. The following factors may be regarded as of reasonable constancy and value in determination :-— 1. The shape of the outer walls of the protoconid and hypoconid. 2. The relative development, degree and mode of separation of the metastylid and metaconid. The other factors which, though variable, may be regarded as of confirmatory value are: (a) The size and shape of the entaconid. (6) The relative sizes of the stylids and conids (if unusual). (c) The development of the ectostylid ridge or fold in the anterior wall of the hypoconid. (d) The size and form of the metaflexid and entoflexid lobes. A Critical Revision of the Quaternary Perissodactyla. 423 Famity EQUIDAE. Genus EuRYG¢NaTHOHIPPUS van Hoepen 1930. Genotype: Hurygnathohippus cornelianus van Hoepen. Broad mandibular symphysis with the four first and second incisor teeth large, anteriorly flattened and arranged almost in a straight line, each incisor showing the cup or ‘‘Mark’’; the third incisors small and lying behind and in contact with the second incisor. (The cheek teeth are unknown.) Eurygnathohippus cornelianus van Hoepen. Eurygnathohippus cornelianus van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, IT, 2, pp. 23-24, figs. 20-22. Type: Anterior portion of lower jaw with incisor teeth. Nas. Mus., No. C.679. (Fig. 8.) | Locality: Uitsoek, near Cornelia, Orange Free State. Horizon: The “Cornelia” Beds of van Hoepen (19304A). The following is a translation of van Hoepen’s description :— “There is preserved the anterior portion of the lower jaw. The four incisors are long and broad. The teeth lie deep in the antero-central part of the jaw and their grinding surface is practically horizontal. The teeth are anteriorly flattened. The first incisor has an indistinct groove along the centre of its anterior or lower surface. The second incisor has two such grooves. On the inner surface each tooth has two surfaces, ribbed parallel to the height, which meet in a blunt corner and a thick ridge. The canine is approximately one-third the width of the other teeth; unfortunately both are broken off. The four big incisors each show a large cup or mark, entirely surrounded by enamel and also completely filled by cement. The enamel on the front of the tooth is thick, but on the back it is thin. The two small canines show no mark; they are broken off very low down.” The present writer conjectured on the possibility of the teeth, which van Hoepen regarded as canines, being in point of fact reduced third incisors, and this suggestion was independently put forward by Dr. L. H. Wells to the writer. The writer and Dr. Wells have had the opportunity, through the courtesy of Dr. van Hoepen, of examining the type specimen, and are of the opinion that this is probably the case and that the specimen is that of a female individual, consequently lacking canine teeth. This reduction of the third incisors is not quite such a startling supposition as would be their complete absence, 424 Annals of the South African Museum. though, as Dr. van Hoepen has said, it is but a step in that direction. The jaw thus clearly belongs to a horse, presumably a very large one, Fie. 8.—Upper, right lateral and lower views of anterior portion of mandible of type of Hurygnathohippus cornelianus van Hoepen. Two- thirds natural size. (From van Hoepen.) but there is some difficulty in absorbing it into Equus. Van Hoepen’s genus must stand in the mean time, and if “third incisor” be sub- stituted for “canine”’, his description cannot be bettered. A Critical Revision of the Quaternary Perissodactyla. 425 Genus Notouipparion Haughton 1932. Genotype: Notohipparion namaquense Haughton. Rather low-crowned heavily cemented hypsodont lower cheek teeth with an extra antero-external cingulum fold or column, either isolated or fused with the parastylid,* present in all the permanent cheek teeth except the second premolar, and a deep groove separating the strongly developed metaconid and metastylid. The upper dentition is unknown. Notohipparion namaquense Haughton. Notohipparion namaquense Haughton 1932. Ann. 8. Afr. Mus., XXvll, pp. 421-423, fig. 5. Cotypes: Series lower Pm,, Pm,, Pm,, M, and M, of the left side and lower M, ofthe right side. S.A. Mus., No. 9982. (Fig. 9.) Locality: “40 miles east of Springbok in Namaqualand... .” Honzon: “from a granite gravel 60 feet down in the surface limestone”’. Measurements : LPm, LPm, LPm, LM, RM, LM, Breadth . ; 16 mm. 17mm. 17mm. 16mm. 14:5 mm. 12-5 mm. Length . - 295mm. 27mm. 26mm. 23mm. 24mm. 29mm. Height . - 145mm. 175mm. 25mm. 23mm. 30mm. 34mm. The following is Haughton’s description of the material :— “The lower teeth upon which this new genus is founded, indicate a stage of equine evolution not hitherto discovered in South Africa. They consist of a series from pm? to m!, together with m? of the left side, and pm‘, m* and m? and a part of m® of the right side, obviously of the same individual. As preserved the teeth are all low, the height not being much greater than the length. The main features can be distinguished from the drawings given. In all the teeth the meta- stylid column is separated from the metaconid column to the base of the crown, and the former projects further inwards than the metaconid or the entaconid, whilst its posterior flange overlaps the anterior border of the entaconid. The antero-external cingulum fold [proto- conid fold] is prominent to the grinding surface, being fused with the parastylid in pm? and m!, but still separated from it in pm‘, m? and m®, In pm‘ of the right side (but not of the left), in m' and m?, there * As defined by Osborn. Stirton (1941) replaces this by “‘paralophid”’ and uses ‘“‘narastylid”’ for the antero-external column. Such reapplication of an existing name is too confusing to be accepted. Dr. L. H. Wells has suggested to the writer that this extra fold might be termed the “‘ protoconid fold”’. VOL. XXXI, PART 4. 36 426 Annals of the South African Museum. is an ectostylid pillar, which near to the root fuses with the wall of the hypolophid to form an ectostylid ridge. Enamel fairly wavy. Teeth heavily cemented” (pp. 421, 422). “Unfortunately nothing is known of the sequence of gravels and limestone encountered in the well from which Notohipparion nama- quense was obtained, and but little hght can be thrown upon its age. The valleys of Namaqualand seem to have suffered a progressive infilling with sand from Upper Cretaceous times onward; but the process was, in all probability, not a continuous one and further study will probably reveal breaks in the sedimentation. In so far as com- parison is possible, Notohipparion would seem to represent an early Pliocene stage of equine evolution” {p. 425). In view of the fact that teeth of another member of the Hipparion group occur in the Vaal River gravels associated with well-made stone implements, it would appear that these forms have survived in this region until a much later time than elsewhere. Though Notohipparion is quite possibly a Pliocene form, it need not necessarily be “early Pliocene” as Haughton suggests, and it is included here since its horizon is so uncertain and since it may have a bearing on the undoubted Pleistocene forms. Genus STYLOHIPPARION van Hoepen 1932. Genotype: Stylohipparion steytleri (van Hoepen). (=8S. hipkini van Hoepen). High-crowned rather narrow hypsodont lower cheek teeth with a strongly developed isolated pillar external to the ectosylid, possessing no external groove between hypoconid and hypoconulid, having narrow protoconid and hypoconid and small rather widely separated metaconid and metastylid. High-crowned upper cheek teeth with isolated oval protocone and possessing a small flange on the antero-internal side of the parastyle. Stylohipparion steytlert (van Hoepen). Hipparion steytlerr van Hoepen 1930. Pal. Nav. Nas. Mus. Bloem- fontein, II, 2, pp. 21-23, figs. 14-19. Stylohipparion hipkint van Hoepen 1932. Pal. Nav. Nas. Mus. Bloemfontein, IT, 3, pp. 31-32, figs. 14-17 and 18-20. Stylohupparion steyilery van Hoepen 1932. Ibid., pp. 33-35, figs. 21-23. A Critical Revision of the Quaternary Perissodactyla. 427 Fig. 9.—Series*of lower cheek teeth of the type of Notohipparion namaquense Haughton. Natural size. (From Haughton.) 428 Annals of the South African Museum. Type: Upper M! or M? of the right side. Nas. Mus., No. C.558. (Fig. 10.) Locality: Uitsoek, near Cornelia, O.F:S. Horizon: The “‘Cornelia Deposits”’ of van Hoepen. Measurements: Breadth . ; « 22 7am, Length . ; . 22mm. Height . . . 54mm. Paratypes: Upper M3 of the left side. Nas. Mus., No. ©.555. (Fig. 10.) Lower M, and M, of the left side. Nas. Mus., No. C.556. (Fig. 10.) Measurements: LM LM, LM, Breadth . . 18mm. 10(12) mm. 10-5(12-5) mm. Length . . 21 mm. broken 21 mm. Height . . 66 mm. 32 mm. 41 mm. (The figures in brackets include the accessory outer column.) Upper Teeth. The halves of the ectoloph are concave inwards though the posterior half may be flattened. The metastyle is small and the mesostyle and parastyle narrow, the latter having an unusual anterior flange or groove. The protocone is isolated and oval in shape with a somewhat flattened interior face. A small tongue projects into the medivallum towards the protocone. The hypocone is very small, being less than one-third the size of the protocone. A deep, sometimes double protoconule fold is present and, with the prefossette folds, tends to isolate the postero-internal corner of the prefossette. Protoloph, hypostyle and postfossette folds are present and are of moderate depth. Secondary plications are present in the region of the prefossette fold and appear to be persistent. Lower Teeth. Protoconid and hypoconid are rather narrow transversely and have slightly flattened or even concave outer walls. An ectostylid ridge or fold is present in the anterior wall of the hypoconid, and external to this is a stout oval pillar arising from the cingulum and reaching A Critical Revision of the Quaternary Perissodactyla. 429 Fie. 10.—Stylohipparion steytlert (van Hoepen). Top: Type upper right second (or third) molar; crown and anterior views. Left: Paratype upper left third molar; crown and outer lateral views. Right: Paratype lower left first and second molars; inner and crown views. All natural size. (From van Hoepen.) almost to the unworn crown. There is no external groove marking off hypoconid and hypoconulid. The entaconid is small, and the meta- conid and metastylid are rather small and widely separated by a rounded groove. The entoflexid lobe is only slightly longer than the 430 Annals of the South African Museum. metaflexid lobe, and folding in their walls is slight or lacking, a small notch in the anterior of the entoflexid lobe being sometimes present. Referred Upper Teeth. An upper second (or possibly third) molar of the left side has been recovered from the Vaal River deposits at Christiana, probably from Younger Gravels. This specimen (Arch. Sur., No. 113) is consider- ably worn, but retains all the characters shown by the less worn type and paratype. Referred Lower Teeth. Five lower teeth, being a series from LPm, to LM, from the type locality, were referred to this species by van Hoepen (1932), and the enamel pattern of four of them is shown in fig. 11. bat) \ (ina Ue a E BA OOOCERM Fic. 11.—Referred series of lower left cheek teeth of Stylohipparion steytlert (van Hoepen). Crown views. Natural size. (From van Hoepen.) The following are the dimensions of these teeth (as preserved) :— LPm, LPm, LM, LM, LM, Breadth . Samim: 12 (14) mm. 12 mm. 11-5 mm. 9 mm. Length ; . 25mm. 28 mm. 22 mm. 24-5 mm. 23 mm. Height ; . 64mm. 74 mm. 69 mm. 73 mm. 38 mm. A lower second molar from the type locality was described by van Hoepen in the same paper (1932) as the type of a new species, S. hipkin. The specimen (Nas. Mus., No. 0.797) is very little worn and thus exhibits an entirely uncharacteristic pattern. Its height is 81 mm., and 15 mm. below the exposed grinding surface the dimensions of the tooth are: breadth 12-5 mm., length 26-5 mm.; in the middle of the tooth the corresponding dimensions are 13 mm. and 22mm. A left second premolar is provisionally referred to this new species, but there does not appear to exist any valid reason for separating either of these teeth from the earlier species. Two lower teeth recovered by Broom from the Kromdraai Cave have been referred by the present writer to this species but they have not yet been described. A Critical Revision of the Quaternary Perissodactyla. 431 Genus Equus Linnaeus 1758. Genotype: Equus caballus* Linnaeus. Equus burchellii* (Gray). Asinus burchelliz* Gray 1824. Zool. Journ., 1, p. 247, peo, aigs, 1,2. Equus quagga wahlbergi* of most authors. Equus platyconus van Hoepen 1930. Pal. Nav. Nas. Mus. Bloem- fontein, IT, 1, pp. 4-5, figs. 3-5. Equus simplicossimus van Hoepen 1930. Ibid., p. 6, fig. 7. Equus simplicissimus van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, II, 2, p. 21, figs. 12, 13. Kraterohippus elongatus van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, II, 1, pp. 7-8, fig. 9. Equus lyle: Dreyer 1931. ‘‘New Fossil Mammals and Man”, p. 30, poy, we. 9: pl. vi, fies. 6,7, 8; pl: vil, fig) 8. The general dental formula is the same as in Equus caballus*, but the vestigial first upper premolar, which is normally lacking in £. caballus* and in most other members of the genus, is frequently present in £. burchellw* and is shed only comparatively late in life. The upper incisor teeth closely resemble those of H. caballus*, but the lower incisors all typically lack the cup or mark so characteristic of the commor horse. This cup is normally formed as a result of closure of two posterior folds of the tooth, and its incomplete formation is well seen in many specimens of LH. asinus*. In E. burchellii* it is completely absent in the lower I,, partially formed in I, and partially or completely formed in I,, but even if present in I, is so shallow as rapidly to disappear with wear. A typical specimen is shown in fig. 12. This feature is not characteristic of either H. zebra* or E. quagga*, though in the latter species the cup is often absent in the lower third incisor, and is not as deep as in EH. caballus*. The retention of the vestigial first premolar and the nature of the incisor teeth may both be regarded as rather primitive characteristics. The upper cheek teeth are a good deal smaller than those of HL. caballus*, and the dimensions of a typical complete dentition (Tvl. Mus., No. 173, fig. 13c) in normal wear are: Pm? Pm? eem* M! M2 M$’ Breadth . 22mm. 24mm. 24:5mm. 23mm. 23mm. 21 mm. Length ~. 85mm. 24mm. 25mm. 205mm. 2mm. 23mm. The breadth variation noted in Pm*-Pm* is 22-26 mm. and in M2-M? is 21-25 mm. 432 Annals of the South African Museum. Fic. 12.—Lower incisor teeth of Equus burchellii* (Gray) to show the characteristic absence or reduction of the cup or “mark”. Natural size. (Original.) A Critical Revision of the Quaternary Perissodactyla. 433 The halves of the ectoloph are concave from without inwards and curve smoothly into the styles, though the mesostyle may overhang slightly. The parastyle is commonly obliquely flattened anteriorly in the premolars but less commonly so in the true molars. This and the mesostyle are sometimes grooved externally for a small extent below the grinding surface. The metastyle is small and not prominent. The protocone is elongate oval in form (proportionally narrower than in H. caballus*), and the portion anterior to the junction with the ‘protoconule comprises about one-third of the total protocone length (except in Pm? where the anterior extent is very small in all species of Equus). The hypocone is not large and frequently shows a slight bulge into the medivallum, giving to it the appearance of a tendency towards isolation from the metaconule. The caballine fold is typically absent, but may be present in some individuals, particularly in the premolars. The fossette folds are generally small except for a well- developed pli-protoconule and a fairly good pli-postfossette. The pli-hypostyle and pli-prefossette are very variable in development, but the pli-protoloph is usually present and often fairly deep. The crown patterns of three typical dentitions are shown in fig. 138. The lower teeth are also a good deal smaller than those of E. caballus*. The dimensions of the lower teeth (fig. 14) in the individual for which the measurements of the upper teeth have already been given are: Breadth f ~ 125mm. 13-5mm. 14mm. 12mm. 11:-5mm. 11 mm. Length ; 2 228mm. 25mm. 24mm. 22mm. 2mm. 27 mm. The breadth variation noted in Pm,-Pm, is 12-15 mm. and in M,—M, is 11-14 mm. The outer walls of the protoconid and hypoconid are both flattened or even concave inwards, and the ectostylid fold in the hypoconid is small or even absent. The entaconid is small and somewhat quadrate inform. The metaconid is oval, and is separated from the metastylid by a fairly sharp-pointed groove. The metastylid itself is rather small, pear-shaped and bluntly pointed. The metaflexid lobe is smaller than that of the entoflexid, and the outer walls of the ento- flexid may show some folding which does not persist with advanced wear. Referred Fossil Material. Petrified specimens of this species are of common occurrence In cave deposits, in superficial deposits, in river gravels and elsewhere. Such A Critical Revision of the Quaternary Perissodactyla. 435 @ Fie. 14.—Right lower cheek teeth of Hquus burchellii* (Gray). The lower dentition C belongs to the individual figured in fig. 13, C. Natural size. (A, after Owen; B and C original.) 436 : Annals of the South African Museum. material does not always seem to be of very great antiquity and is quite possibly only of Upper Pleistocene Age. The species Equus platyconus van Hoepen and Kraterohippus elongatus van Hoepen have already been regarded by Haughton (1932) in his revision as probable synonyms of the living Burchell’s zebra, and there seems no reason to doubt the correctness of this conclusion. Kraterohippus elongatus, moreover, is founded on a broken upper Pm?, almost certainly incorrectly restored and is value- less as a type. Equus sumplicissimus van Hoepen also appears to fall well within the range of variation of the living species and does not warrant distinction. Equus lylet Dreyer, founded on well-mineralised material from the springs at Floris Bad, was regarded by Haughton as falling within the range of variation of the living species. This is certainly the case, most of the specimens from the type and neighbouring areas, however, lying close to the lower size limit of the species. Their enamel characters are quite typical and the material does not apparently require specific distinction, though a varietal or sub-specific separation may prove to be justified. Mineralised specimens of Equus burchelliv* occur in the deposits of the Vaal River basin. Equus zebra* Linnaeus. Equus zebra* Linnaeus 1758. Syst. Nat., Ed. 19, 1, The incisor teeth have been considered under LZ. burchellir* above. The upper cheek teeth differ very little in size from those of E. burchellu*, though on the average they are possibly a millimetre broader than the average bontequagga. The halves of the ectoloph are almost straight or even slightly convex, and the parastyle and mesostyle are somewhat abruptly marked off from the ectoloph walls. These features are less apparent in the true molars than in the pre- molars. The protocone is sub-triangular rather than oval, and the portion anterior to its junction with the protoconule is less than one- third of the protocone length. The parastyle is commonly flattened anteriorly but is not as obliquely directed as in #. burchellu*. The hypocone is only a little smaller than the protocone. A caballine fold is often, but not consistently, present. The fossette folds are all small, and even the pli-protoconule and pli-postfossette may be virtually lacking. The pli-protoloph is typically absent or extremely small. 438 Annals of the South African Museum. The lower teeth are also comparable in size with those of the bontequagga. The outer walls of the protoconid and hypoconid are convex and well rounded, and the ectostylid fold is small or lacking. The entaconid is rounded, as also is the metaconid, but the metastylid is pear-shaped and rather pointed. The groove separating metaconid and metastylid is somewhat pointed. The metaflexid and entoflexid lobes are simple. Two upper and a lower dentition are shown in fig. 15. Referred Fossil Material. The only petrified specimens certainly to be ascribed to this species were recovered from the Cango Caves near Oudtshoorn, but the remains of zebrine skulls are not uncommon in the older dune sands of the coastal belt in the Cape Province. Equus quagga* Gmelin. Equus quagga* Gmelin (Linnaeus 1788). Syst. Nat., Ed. 13, 1, p. 21. Equus quagga* Cooke 1941. S. Afr. J. Sci., xxxvui, p. 307. Equus quagga* Shapiro 1943. 8S. Afr. J. Sci., xxxix, p. 117. (non E. quagga quagga* Dreyer 1931. “New Fossil Mammals and Man”’, p. 33.) (non EF. quagga* var. Haughton 1932. Ann. 8. Afr. Mus., xxviii, p. 424.) Equus quagga* became extinct in 1872 before most museums had commenced the collection of skeletal material from Southern Africa, and only a cast in the Transvaal Museum, taken from a skull in the possession of the Stuttgart Museum, represents a specimen actually known to be that of the true quagga. This individual was very young, and the cast is not of great value for identification purposes. Fortunately the great Richard Owen, in the “Philosophical Trans- actions of the Royal Society of London” for 1869, gives natural size figures of the dentition of a good adult specimen at that time in the Royal College of Surgeons. Dr. Robert Broom, while in London many years ago, made drawings of the teeth of a quagga in the British Museum collections, and these agree with those given by Owen in all characters. Some old skulls and jaws in the McGregor Museum, Kimberley, and in the collections of the Kingwilliamstown Museum, agree closely with the figures published by Owen and with A Critical Revision of the Quaternary Perissodactyla. 439 the manuscript drawings given to the writer by Dr. Broom, and as these skulls were collected in a region certainly at one time overrun by quagga, this material may with fair certainty be taken to represent the species. Furthermore, the characters which Owen remarks as distinguishing the quagga from the other two South African species, are present in these skulls. From the limited material available, it would appear that the teeth of this species do not differ very markedly in size from those of the zebra and bontequagga. The largest and smallest upper dentitions in the McGregor Museum collection (fig. 16, B, C) have the following dimensions over the enamel :— Pm? me Pm* WE M? M3 Breadth 215mm. 225mm. 22mm. 205mm. 20:0mm. 19-5 mm. Length 305mm. 235mm. 215mm. 20:(0mm. 195mm. 20:5 mm. Breadth 23-0 mm. 270mm. 265mm. 245mm. 240mm. 19-5 mm. Length 325mm. 245mm. 245mm. 215mm. 215mm. 23-5 mm. Owen’s figure gives measurements only slightly smailer than the dimensions of the larger dentition, so that it would appear that the average quagga dentition was a little larger than the normal bonte- quagga, and the range in breadth is somewhat greater as well, being 22-27 mm. for Pm?—-Pm*# and 20-25 mm. for M!-M?. The halves of the ectoloph are almost straight or even slightly convex, and are abruptly marked off from the parastyle and mesostyle by an acute angle, though this character is suppressed in the third molar and not clear in the second premolar. The external face of the parastyle, unlike that in EL. burchelli* and EL. zebra*, is not directed in an oblique angle anteriorly but lies roughly parallel to the axis of the row of cheek teeth. The protocone is elongate oval or somewhat triangular in form, and commonly has an almost median indentation in its inner wall, giving a rather bilobate appearance. The junction of the protocone with the protoconule is only slightly anterior to the middle of the protocone. The hypocone is small and the hypoglyph is rather variable. The caballine fold is normally absent. The fossette folds are all small, even the pli-protoconule and pli-postfossette being commonly reduced to a small notch only. Nevertheless the pli-protoloph and pli-hypostyle are both normally visible, though the pli-prefossette is typically small or absent. The crown patterns of three upper dentitions are shown in fig. 16. The lower teeth are comparable in size with those of HL. burchelli*, two typical dentitions measuring: A Critical Revision of the Quaternary Perissodactyla. 441 Pm, Pm: Pm, M, M, M, Breadth 130mm. 140mm. 145mm. 125mm. 12:0mm. 11-5 mm. Length 275mm. 25-0mm.. 240mm. 225mm. 220mm. 26-0 mm. Breadth 130mm. 140mm. 140mm. 125mm. 12-0 mm. 11-0 mm. Length 315mm. 265mm. 260mm. 235mm. 240mm. 23-5 mm. Owen’s figure agrees very closely in dimensions with those of the latter species. The probable range in breadth may be estimated as 12-16 mm. for Pm,-Pm, and 11-15 mm. for M,-M,,. In character the lower teeth resemble those of bontequagga rather than zebra, for the outer walls of the protoconid and hypoconid are somewhat flattened, the latter more so than the former. The ectostylid fold is often present. The metaconid is rounded or oval, and is separated from a pear-shaped metastylid by a rounded groove. The entaconid is rounded to quadrate and notably larger in the premolars than in the molars. Three dentitions are shown in fig. 17. Probably largely in consequence of the confused ideas which have hitherto existed regarding the characters of the teeth of Equus quagga*, none of the teeth referred to by various authors as “Z. quagga”’ or “ E. quagga quagga”’ can be regarded as actually belonging to Gmelin’s species. These so-called “quagga” teeth are mainly those of £. burchellia* (Gray) but, remarkably enough, none of the fossil teeth described under other names can be regarded as belonging to E. quagga* either, so that it would appear that this recently extinct species has not hitherto been recognised to any notable degree in our fossil collections. This species is represented by material in the McGregor Museum, Kimberley, derived from surface deposits at Koffiefontein, Orange Free State. It has also been recorded from a cave near Kuruman (Cooke, 1941), from Bankies, O.F.S. (Shapiro, 1943) and appears to be present in the deposits of the Vaal River valley. Doubtless more material will be recognised in due course. Equus capensis Broom. Equus capensis Broom 1909. Ann. 8. Afr. Mus., vil, pp. 281-282. Equus capensis Broom 1913. Bull. Amer. Mus. Nat. Hist., xxxii, p. 437, fig. 1. Equus capensis Broom 1928. Ann.S8. Afr. Mus., xxii, p. 441, fig. 2, A. Equus cawoodi Broom 1928. Ibid., pp. 443-444, fig. 3, A. Equus gigas van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, IT, 1, pp. 2-3, fig. 1. Equus capensis Dreyer 1931. Ibid., pp. 36-87, pl. vii, fig. 5. VOR, XXXI, PART 4. 37 442 Annals of the South African Museum. < a = Fic. 17.—Right lower cheek teeth of Equus quagga* (Gmelin). Natural size. (A, after Owen; B and C original.) A Critical Revision of the Quaternary Perissodactyla. 443 Equus westphali Dreyer 1931. Ibid., pp. 36-37, text-fig. Equus capensis Haughton 1932. Ann. 8. Afr. Mus., xxviii, pp. 410- 412, fig. 2. Equus capensis Cooke 1939. 8. Afr. J. Sci., xxxvi, pp. 413-414, fig. 2. Equus capensis Cooke 1941. 8. Afr. J. Sci., xxxvii, pp. 308-311, “figs. 4, 5. Equus capensis Wells and Cooke 1942. Trans. Roy. Soc. S. Afr., XX1X, pp. 228-229, fig. 12. Equus capensis Shapiro 1943. 8. Afr. J. Sci., xxxix, pp. 117-178. (non #.cawoodi van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, II, 1, pp. 3-4, fig. 2.) (non £. cawoodi Dreyer 1931. “New Fossil Mammals and Man”’, pp. 26-29, pl. vi, figs. 8-12; pl. v, fig. 2.) (non £. capensis Broom and Le Riche 1937. S. Afr. J. Sci., xxxiil, pp- 769-770, fig. 1.) (non pars H. capensis Cooke 1939. S. Afr. J. Sci., xxxvi, pp. 413- 414, fig. 2a.) Type: Series lower left P,, P;, P,, M,, M, embedded in limestone and damaged on the inner sides. 8.A. Mus., No. 658. (Fig. 18.) Locality and Horizon; Beach debris, Yzerplaats, Maitland, Cape Province. Measurements: ibles LP, GP; LM, LM, Breadth (damaged) 2 215+ mm. 15+ mm. 14+ mm. 14+ mm. Breadth (as restored) . : 2 19-5 mm. 18-5-19 mm. 18mm. 17-5 mm. Length : . ?30 mm. ?35 mm. 34:5 mm. 31mm. 30mm. Height : a? (Og. 87 mm. 103 mm. 92mm. 94mm. Neotype (Haughton 1932): Lower left fourth premolar. §S.A. Mus., No. 2821. (Fig. 18.) Locality: Saldanha Bay. Measurements: LPm, Breadth Z é . 18-5 mm. Length : A - 33-0 mm. Height : ‘ . 92-0 mm. The type of this species was first described by Broom in 1909 but was not figured until 1928. It comprises a damaged series of teeth embedded in a slab of sandy limestone found on the beach at Yzer- plaats, Maitland district, Cape Province. It would appear that this limestone represents a former land deposit now lying below sea-level 444 Annals of the South African Museum. Fic. 18.—Equus capensis Broom. Crown view of the type series of left lower cheek teeth as embedded in Top: the limestone matrix. (Original.) Centre: Restoration of crown patterns of type series. (Original.) Below: Enamel pattern of neotype lower left fourth premolar. (From Haughton.) All natural size. ae A Critical Revision of the Quaternary Perissodactyla. 445 as a result of late crustal warping. While it is impossible to arrive directly at an estimate of the age of the original deposit, this must antedate the warping which itself is most probably the final crustal warping occurring between the development of the “ major emergence”’ and the “minor emergence” raised beaches. The dating of this event is difficult, but the writer has elsewhere (Cooke, 1941 ) suggested that it coincides approximately with the existence of the Stellenbosch industry and may be broadly Middle Pleistocene. - If this estimate is not far wrong, the Equus capensis bearing lime- stone belongs most probably to the Middle Pleistocene, and the general affinities of the teeth also support a Middle Pleistocene age. The teeth themselves are exposed on a fractured surface of the slab, and the inner portions of each tooth have been lost. The series belongs to the left lower jaw and lacks the third molar, but the second premolar is completely embedded and the anterior portion of the third premolar partly covered by limestone. The patterns of the grinding surfaces are consequently visible only in part of the third premolar, the fourth premolar and the first and second molars, but all the teeth are well exposed in their inner lateral aspects. Broom figured a restoration of these four teeth in 1928, but many features of the drawing are faulty, and the restored portions probably incorrectly given. Fig. 18 shows an accurate drawing of the enamel patterns as they appeared in the slab in 1940, though possibly the specimen is now less perfect than in 1928. Nevertheless, Broom’s restoration fails to show the distinct flattening of the hypoconid outer wall, gives in- correctly the shape of the visible portions of the entaconid and simplifies the irregularities in the entoflexid. As Haughton has pointed out: “In his restoration of the missing portions he has shown the metaconid, metastylid and entoconid as being rounded in outline, his restoration of the metastylid in particular differing con- siderably from the somewhat triangular or pointed form seen in Equus caballus.’’ It is impossible from the specimen to see any grounds for maintaining this restoration, and in the light of all the available evidence a new restoration is now given which is probably not far from the truth (fig. 18). In form these teeth resemble those of the true quagga rather than the other living species, but they are notable for their very large size, far exceeding the dimensions of any living type. The protoconid and hypoconid are well developed, and the outer wall of the latter is more markedly flattened than in the former. The protoconid projects, in the molars at least, beyond the outer wall of the hypoconid, giving a 446 Annals of the South African Museum. somewhat lopsided appearance to many of the teeth. The ectostylid fold in the hypoconid wall is small or completely lacking. The hypolophid is quite distinctly marked off from the hypoconid. The entaconid is quadrate to semi-lunate in form with a distinctly angular antero-internal corner (“entostylid”). The metastylid is pear- shaped and somewhat pointed (if the restoration or evidence from the neotype and other specimens can be assumed to be correct). The metaconid is oval and is separated from the metastylid by a pointed groove. The outer wall of the entoflexid is irregular or wavy in the premolars and slightly concave in the molars. The simplicity of the molars as compared with the premolars, and also the rather great disparity of form and size, is striking in some of the series available. The range in breadth for the premolars appears to be about 18-5 to 22 mm. and for the molars 17-5 to 21 mm. Amongst the large number of lower teeth referred to this species, the best preserved and most valuable series is shown in Fig. 19. These teeth come from a thermal spring at Vlakkraal near Bloem- fontein, Orange Free State, and have been described by Wells and Cooke (1942). Their individual dimensions are as follows:— Pm, Pm, Pm, M, M, Breadth . 5 : - 185mm. 19mm. 195mm. 17mm. 17-5 mm. Length . : ; : 38mm. 34 mm. 34mm. 29mm. 30-5 mm. Height (excluding roots) . 56mm. 62 mm. 74mm. 58mm. 51+ mm. The teeth are thus very slightly smaller than those of the type series, but there can be little doubt regarding their close agreement in all essentials. This series also illustrates very well the lateral aspects of the teeth and their normal angles of wear, and is very useful as an aid in determining the position of an individual tooth in a series. The upper teeth of Equus capensis are so far not certainly known, but from the normal relationships between upper and lower teeth the expected breadth of the uppers would be 29 to 35 mm. for the pre- molars and 27 to 33 mm. for molars. Broom (1913 6, 1937), Dreyer (1931) and Haughton (1932) have referred large upper teeth to £. capensis. New species within the correct size group have also been erected by Broom (1928), van Hoepen (1930), Dreyer (1931) and Wells (1941). All these finds have been on open sites where association is virtually valueless, but material described by the writer (Cooke 1941 a) and by Wells and the writer (1942) from sealed sites may elucidate the problem. In a large collection of equine remains from the Wonderwerk Cave is . cee . 5 Oyo of as ee ee ae ee 5 “st; Ae? ere diiteutelats Tinga al “9.%°% 0 < CC tas ae) “g¢ =. oie Cer ae iy eee ee oS ce e@* .? ° ar Fig. 19.—Fine series of lower left cheek teeth referred to Equus capensis Broom. Outer lateral and crown views. Natural size. (From Wells, Cooke and Malan.) 448 Annals of the South African Museum. in the Kuruman district, twelve teeth (upper and lower) were immedi- ately distinguished by their great size. Five are lower teeth which agree in pattern with those of H. capensis, though they are very slightly smaller than the type. The associated upper teeth can hardly belong to a different species, and these are closely comparable with the type of Equus cawoodi described by Broom (1928) from the Vaal River gravels. The type of EZ. cawoodi (which is most probably a Pm? and not a Pm‘ as Broom states) is little worn, and agree in almost every detail with a little worn LPm‘* from Wonderwerk. The Wonderwerk LPm¢ was sectioned, and on its sectional surface showed a very considerably simplified pattern agreeing very well with the uppers referred to EH. capensis by Broom (1913) and one referred to that species by Haughton (1932). Additional material from Wonderwerk described by Wells (1943) confirms this view. The large upper teeth associated with the fine lower series from Vlakkraal are very worn, but they agree in all essentials with EH. cawoodi and there can be little doubt that the two species are synonyms. Shapiro (1943) dissented from this view, but his argument was based chiefly upon the material erroneously referred to LH. capensis by Broom & le Riche and now falls away, as he himself would doubtless agree. The type RPm?® of HL. cawoodi (M.M.K., 3711) has the followine dimensions (fig. 20) :— Breadth, on grinding surface (which is Fi little worn) . 31-5 mm. Breadth, 3 cm. below grinding surface . : ; 32 mm. Length , ; : . : : : . 33 mm. Height ‘ é , : : : ; ‘ 80 mm. It may conveniently be taken as a neotype of the upper dentition of E. capensis. The form of the upper dentition of E. capensis (as now understood) is very similar to that of E. burchelli*, of which it is virtually an enlarged version. The halves of the ectoloph are concave inwards and curve easily into the styles, except for a distinct tendency for an anterior overhang of the mesostyle. The parastyle is obliquely flattened anteriorly in the premolars but less noticeably so in the true molars. The metastyle is small. The hypocone is moderately small but the hypoglyph may be rather deep. The caballine fold is sometimes present but may often be absent. The pli-protoconule and pli-postfossette are well marked, but the pli-protoloph and pli- hypostyle are very small and may disappear completely with wear. The pli-prefossette and secondary small postfossette are distinct in Upper left: Upper right: Lower left: Lower right: pau0ij9as Fie. 20.—Hquus capensis Broom. Crown and outer lateral views of neotype upper right third pre- molar. (Original.) Enamel pattern of two referred upper left fourth premolars from Saldanha Bay. (After Broom.) Enamel pattern of referred upper left second molar from Bloem- bosch, Darling. (After Haughton.) Original crown and sectioned enamel pattern of a broken, little worn, referred upper left fourth premolar from Wonderwerk. Sectioned surface approximately 2 cm. below crown. (From Cooke.) All natural size. 450) Annals of the South African Museum. very early wear but vanish rapidly with abrasion. The protocone is elongate oval in form, with rather more than one-third of its total length lying anterior to the junction with the protoconule. The range in size is, as far as observation goes, that deduced from the dimensions of the lowers. Fig. 20 shows the form and pattern of the neotype (previously the type of EH. cawoodi), the two teeth referred to the species by Broom (1913), that referred to it by Haughton (1932), and the sectioned damaged tooth from Wonderwerk. Fig. 21 gives the pattern of part of a series from Wonderwerk, two teeth from Floris Bad and one from the Vaal River gravels. These figures give a good idea of the characters and variations of the species. The series of milk molars and two molars described by van Hoepen (1930 a) as Equus gigas do not appear to warrant specific distinction and may be referred to EL. capensis. The upper teeth from Vlakkraal agree with those from Floris Bad to which Dreyer gave the name E. helmet, a species which Haughton (1932) regarded as a synonym of E. cawoodi. This is accordingly a synonym of £. capensis, as also is E. westphali, which Dreyer erected on lower teeth of large size from the Vaal River gravels at Pniel. The tooth which van Hoepen (1930 a) referred to EL. cawoodi and subsequently (1930 b) to a new species Z. louwi does not appear to be E. capensis, but rather E.kuhni Broom. The same incorrect identifica- tion appears to apply to Dreyer’s FZ. cawoodi (1931) and to the specimen described and wrongly figured by Broom and le Riche (1937). The smaller of the two right lower fourth premolars referred by the present writer (Cooke 1939) to #. capensis seems in the light of later knowledge to belong to EZ. harrisi Broom. Equus kuhni Broom. Equus kuhni Broom 1928. Ann. 8. Afr. Mus., xxii, p. 444, fig. 3, B. Equus cawoodi van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, II, 1, pp. 3-4, fig. 2. Equus louwi van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, II, 2, pp. 19-21, figs. 6-11. Equus kuhni Haughton 1932. Ann. 8. Afr. Mus., xxviii, p. 414. Equus kuhni Cooke 1941. 8. Afr. J. Sci., xxxvii, pp. 307-308, fig. 3. Equus capensis Broom and Le Riche 1937. S. Afr. J. Sci., xxxiii, pp. 769-770, fig. 1, A. A Critical Revision of the Quaternary Perissodactyla. 451 Fic. 21.—Hquus capensis Broom. Left: Part of incomplete series of referred left upper cheek teeth from Wonderwerk Cave. (From Cooke.) Referred left upper fourth premolar and second molar from Floris- bad. (After Dreyer.) Lower right: Referred upper right fourth premolar from Vaal River gravels at Austin’s Rush, Barkly West. (Original.) All natural size. Top right: 452 Annals of the South African Museum. Sectioned Fig. 22.—Equus kuhni Broom. Left: Type upper right fourth premolar. (Original.) Right: Neotype upper left fourth premolar from Wonderwerk Cave, showing pattern on little worn crown and on surface sectioned 2-5 cm. below the crown. (From Cooke.) Natural size. Type: Upper Pm‘ of the right side. M.M.K., No. 3929. (Fig. 22.) Locality: Pniel, Vaal River. 7 Horizon: % Vaal River gravels. Measurements: RPm#? 2 cm. below crown Breadth : : . w29:>)mm. 28-5 mm. Length : : . 431-0! 30-5 mm. Height : : + 273 -0imm.: A Critical Revision of the Quaternary Perissodactyla. 453 Neotype: Upper Pm!‘ of the left side. Arch. Sur., No. 143. (Fig. 22.) Locality: Wonderwerk Cave, Kuruman District. Horizon: Disturbed cave deposits. Measurements: LPm#4 2-5 cm. below crown Breadth : 5 . 27-5 mm. 28 mm. Length é 5 - 29-5 mm. 28-5 mm. Height : 2 - 77-0 mm. The type of this species, like that of E. cawoodi, is little worn; indeed the specimen is so little abraded that the protocone is not yet properly connected to the protoconule. The pattern it presents is thus not typical, and it has been necessary to create a neotype which has been sectioned to show the simplified pattern consequent upon more advanced wear. The marked grooving of the mesostyle does not extend far down the tooth. It is also apparent in the type that the smooth and deep concavity of the two halves of the ectoloph is a feature of the early state of attrition and that further wear would show these to he a little more flattened. The upper teeth are intermediate in size between those of E. capensis and EH. quagga*, the breadth in Pm?—Pm* being about 27-30 mm., and in M,-M, about 25-29 mm. The halves of the ectoloph are concave with a slight flattening of the inner face, and the parastyle and mesostyle are prominent and somewhat abruptly marked off as in £. zebra*. The anterior face of the parastyle is obliquely flattened and commonly slightly grooved in the premolars. The protocone is elongate oval with a flattened or a “bilobed” inner wall. Its junction with the protoconule is not much in front of the middle of the proto- cone. The hypocone is small and has the appearance of a tendency towards isolation, as the hypoglyph is unusually deep in many specimens and there is also commonly an indentation in the wall of the medivallum between the hypocone and the metaconule. The caballine fold is usually present and may be quite strongly developed. The pli-protoconule is well developed and often deep, as also is the pli-postfossette. The pli-postfossette is usually clearly shown and secondary plications are common. The pli-hypostyle and_pli- protoloph are very variable in development and the former is sometimes well marked. The lower teeth are not certainly known, but from the normal size relationship between upper and lower teeth they should range from 16-5 to 19 mm. for Pm? to Pm4 and 15-5 to 18 mm. for Mt to M?. The species, H. harrist Broom lies in the correct size group, and there is a 454 Annals of the South African Museum. strong probability that HE. kuhni is the upper dentition of that species. In 1937 Broom and le Riche described and figured certain teeth which they referred to #. capensis. The Bothaville material has not been seen by the present writer, but the Sterkfontein specimens were amongst those lent to him by Broom. The lower jaw described by Broom and le Riche from Sterkfontein have the incisors and a second premolar present, and in size these exceed those of the living horses, but are otherwise not specifically identifiable. They might belong to E. capensis, EZ. kuhni, E. harrisi or E. plicatus. The upper palate and teeth from Sterkfontein were not found in association with the lower jaw. This upper jaw is well embedded in the typical rather hard cave matrix, and it has proved impossible to clean the specimen properly in the normal manner. The occlusal surface of some of these teeth had been ground down by Broom to expose the enamel pattern, and the pattern of the first right molar was figured in the paper cited (fig. 1, A, p. 770) and measurements given. When the present writer examined the material, however, it was realised that the ground surface was very oblique to the plane of normal wear with a con- sequent distortion and exaggeration of breadth. The whole specimen was accordingly carefully ground down parallel to the palate and the dentition exposed as it would appear in early wear. The surface of the specimen (Tvl. Mus., No. 682) with the patterns now exposed is shown, natural size, in fig. 24. The dimensions are: RPm2 RPm? RPm#4 RM} RM? RM? Breadth . 29:0 mm. 229mm. ?30mm. 27:5 mm. 227mm. ?26 mm. Length . 365mm. 295mm. 30mm. 305mm. 295mm. ?28 mm. The whole character of the teeth places the specimen in the species E. kuhni, of which it is an excellent example. The molar originally referred by van Hoepen (1930 a) to E. cawoodt, and later (1930 b) cited as the type of a new species H. lowwi, is a very typical tooth of EH. kuhni, as also are the other teeth ascribed to E. louwt. Haughton (1932) has already suggested that this species is a synonym of £. kuhnt, and this view is confirmed. The teeth which Dreyer (1931) refers to as “Equus kuhni simplex” are difficult to refer with certainty, but the “wholly unfossilised fourth premolar dug up at Glen Craig, Grahamstown”’, looks remark- ably like H. zebra*, while the one from Pniel is too small to be E£. kuhni and agrees excellently with little worn teeth of H. burchellu*. The partly fossilised first and second molars “dug up by Mrz. S. (‘TeuIS1Q) ‘oeZIs TBMyBNY “POOL oy} Surystjod puv Surpuri3 Aq posodxe 3003 yooyo Joddn jo suseqqed joureue YyIM Puv XI1}eUT UT poppe ‘moog wyny snnby—'ss “Ol ure ‘oABD UloJUOFY10}g Woy UoMToods porloyod JO MOTA [VFL]CA 456 Annals of the South African Museum. Kinneard in Grahamstown”’, however, agree excellently both in size and character with #. kuhni and may certainly be assigned to that species. Fic. 24.—Hquus harrisi Broom. Top: Type left lower second molar and fourth premolar. (Original.) Centre: Referred right lower cheek teeth from Sterktontein Cave. (Original.) Below: Referred right lower fourth premolar from the Vaal River gravels at Pniel. (From Cooke.) All natural size. A Critical Revision of the Quaternary Perissodactyla. 457 Equus harrist Broom. Equus harrist Broom 1928. Ann.S. Afr. Mus., xxii, pp. 441, fig. 2, B. Equus (Sterrohippus) harrisi (pars) Haughton 1932. Ann. §. Afr. Mus., XXvill, pp. 412-413. Equus capensis (pars) Cooke 1939. 8S. Afr. J. Sci., xxxvi, p. 413, fig. 2, A. (non Equus harrisi Dreyer 1931. ““New Fossil Mammals and Man”, pp. 23-25, pl. vi, figs. 18, 14; pl. vii, figs. 11, 12.) _ Lectotypes: Left lower Mand Pm,. M.M.K., No. 3939. (Fig. 24.) Locality and Horizon: “Diamond gravels of the Middle Terrace at the bend near Barkly West.” Measurements: LM, LPm, Breadth . : ‘ : : . 16-5 mm. 17-5 mm. Length . : : : : . 29-5 mm. 30-5 mm. Height (excluding roots) . : . 37:0 mm. 45-0 mm. The species is a good deal larger than EF. burchellii*, which it some- what resembles, and it is smaller than E. capensis. The type speci- mens are very much worn, and this fact makes specific identification with less worn specimens rather difficult. A feature of the species is its relatively greater antero-posterior compression compared with corresponding teeth in EH. capensis. In the premolars, the entoflexid lobe is simpler and smaller than in £. capensis. There is also a marked tendency in the premolars for the indentation between hypoconid and hypolophid to be very small or absent. The outer walls of the well-developed hypoconid and pro- toconid are flattened or even slightly concave inwards, the ectostylid fold in the hypoconid wall being small or lacking. There is a strongly developed entaconid of rounded or quadrate form. The metaconid is relatively large and elongate antero-posteriorly, while the meta- stylid is smaller and pear-shaped. The valley between metaconid and metastylid is bluntly pointed. Good material of this species is difficult to find, but an excellent series discovered by Broom in the hard limestone of the famous Sterkfontein Cave was given to the writer for identification and is figured here (fig. 24) for the first time.* It is unfortunately impossible to remove these teeth from their matrix, but the available dimensions are: RPm, RPm, RM, RM, Breadth . 2 eb maim: 18 mm. 16-5 mm. 16 mm. Length ‘ . 33-0 mm. 32 mm. 28-5 mm. ?28 mm. * Also see Appendix. VOL. XXXI, PART 4. 38 458 Annals of the South African Museum. A lower right fourth premolar from Pniel was identified and figured in error by the present writer (Cooke, 1939, p. 413, fig. 2) as belonging to EF. capensis. This tooth (Arch. Sur., No. 5) is now assigned to FE. harrisi and is refigured here. Its breadth is 17 mm., its length 30-5 mm. and its height 74 mm. The lower right fourth premolar (M.M.K., No. 4066), figured by Haughton (1932, p. 415, fig. 3, B) and referred by him to £. simplex van Hoepen, probably belongs to this species, as also may the associated right first molar (M.M.K., 4047). Referred Upper Teeth. In his original paper Broom (1928) refers to this species a right upper Pm‘ from the same locality. Considering the nature of this deposit, the fact of an associated occurrence is only of very slight value in con- necting the specimens. On a basis of size relationship between upper and lower teeth, the upper Pm* corresponding to the type lower tooth of this species should be about 28 or 29 mm. broad. The associated specimen referred by Broom has a transverse diameter of 31 mm. which, though not certainly outside the possible range, is rather on the large size. In view of these considerations the reference of this upper tooth to #. harrist cannot be regarded as satisfactory, and the tooth is therefore provisionally made the type of a new species Hquus broom described below. Van Hoepen (1930 a) referred this specimen to a new genus and species Sterrohippus robustus which he created on a second premolar. ‘A second premolar cannot be regarded as forming adequate material for the type of a species, nor does there appear to be grounds for separating the genus from Equus. His specific name could not be retained as a species of this latter genus, as it is preoccupied by Equus robustus from the Pleistocene of Europe. Sterrohippus robustus is therefore regarded as incorrectly founded and invalid. The upper teeth of Equus harrisi must for the moment be regarded as unknown until they are found in a good association with recognis- able lower teeth. The rather poor lower teeth found in apparent association with upper teeth assigned to Equus kuhni Broom agree fairly well with those of H#. harrisi, but there is as yet insufficient material to warrant identification with that species. The upper teeth of HL. kuhni, however, are in precisely the right size group and it is very possible that EH. kuhni represents the upper dentition of E. harrist. The association of the two species in the Sterkfontein group of caves may also be significant. A Critical Revision of the Quaternary Perissodactyla. 459 Equus plicatus (van Hoepen). Kolpohippus plicatus van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, II, 1, pp. 8-10, fig. 10. Equus plicatus Dreyer 1931. “‘New Fossil Mammals and Man”, pp. 33-35. Equus plicatus Haughton 1932. Ann. 8. Afr. Mus., xxviii, p. 417. Type: Lower series Pm, to M, of the right side. Nas. Mus., No. C.425. (Fig. 25.) Locality: Tierfontein, near Port Allan, O.F-.S. Horizon: Not given in description; probably surface deposits. Measurements: The measurements cited in the type description do not agree with the text-figure, but it appears that the breadths given are total breadths and not just the width across the enamel. The dimensions given below give the breadth over the enamel, estimated by enlarging the illustration in van Hoepen’s paper to agree as well as possible with the various measurements he cites. The text-figure of the type series given here is the one derived in this way: RPm, RPm, RPm, RM, Breadth . . 17-5 mm. 17-5 mm. 15+ mm. 15-5 mm. Length . . 38:0 mm. 33-0 mm. 30 mm. 27-0 mm. Height . . 68-0 mm. 81-0 mm. 89 mm. 79-0 mm. + very little worn. The lower teeth are larger than those of #. quagga* and smaller than those of E. capensis, agreeing in size with those of H. harrisi, which they otherwise closely resemble. The probable range of breadth in Pm, to Pm, is about 15-5 to 19 mm., and in M, to M, 14-5 to 17-5 mm. The outer walls of the protoconid and hypoconid are flattened or even concave inwards in the premolars; in the molars the protoconid may be convex. A strong ectostylid fold is present in the anterior part of the hypoconid wall in the premolars, but this fold is small or even absent in molars. The metaconid is rounded, and is separated from the bluntly pointed metastylid by a pointed groove. The entaconid is quadrate and smaller than the metaconid. The lobe of the ento- flexid is complicated in the premolars by strong—or at least well- marked—crimping of the inner wall of the hypoconid. The molars are much simpler than the premolars and cannot be easily distinguished from those of E. harrist. Van Hoepen’s reference of the type teeth to a new genus does not Fic. 25.—Equus plicatus (van Hoepen). Left: Crown view of type series of right lower cheek teeth. (After van Hoepen; modified.) Right: Crown view of series of lower left cheek teeth from damaged lower jaw from Koffiefontein, Orange Free State. This jaw is shown at the bottom of fig. 6. (Original.) Natural size. A Critical Revision of the Quaternary Perissodactyla. 461 appear justifiable and even if the species is valid, it must be placed in Equus. A fine series of teeth in a damaged left lower jaw from Koffiefontein, O.F.S., may be referred to this species with considerable certainty. The jaw (M.M.K., No. 4845) is shown in fig. 6 in comparison with a skull of H. burchellii* and the enamel pattern of the teeth is shown in fig. 25. The inner side of the dentition is slightly damaged but the dimensions over the enamel are given below: LPm, LPm, LPm, LM, LM, Breadth : . L7 mm. 17-5 mm. 17-5 mm. 16:5 mm. 15:5 mm. Length ; . 38 mm. 33 mm. 33 mm. 29 mm. 29 mm. It would appear very probable that the differences between the dentitions ascribed to E. harrisi and to E. plicatus are not really of specific value. The type teeth of E. harrisi are very worn indeed, and it is difficult to assess the probable appearance of the teeth of this species in early wear. There can, however, be little doubt that teeth of L. plicatus in advanced wear would be indistinguishable from those assigned to #. harrist. The upper teeth are not known from a certain association. There is a strong probability that the upper dentition of a skull from the " same locality (Koffiefontein) as the lower jaw described above may be that of H. plicatus. This upper dentition has been described by Wells as E. fowler, and in the absence of a certain association the merging of these two species must remain an open question. Equus fowler: Wells. Equus fowleri Wells 1941. Trans. Roy. Soc. 8. Afr., xxvii, pp. 301- 906, fig: 1, pl. Iv. Cotypes: “Portions of skull and isolated tooth catalogued as No. 555 in the collection of the Department of Anatomy, University of the Witwatersrand, Johannesburg.” (Figs. 6 and 26, A. B.) Locality: Koffiefontein, O.F.S. Horizon: “At a depth of about 18 inches in surface soil overlying shale.”’ Material: This comprises “the greater part of the right maxilla and palatine bone with Pm?—-M? in position and perfectly preserved, the right maxilla containing an unerupted I® and part of the left maxilla lacking teeth”. ‘‘The presence of the socket for C sug- gests that the individual is a male.” “The remains are those of an individual not quite fully grown.” (‘STI9AA W017) ‘ozs [wIn4vU TTY ‘IB[OUL puodes 4ySt1 Joddn porsejor Jo ur944ed UMOID “CT ‘gq Ul uMOYsS ;uI'T edA409 Jo UMOIO MOTEq “UII YZ-GT SoVJINS pouoTyoos UO uI0}4ed JoULVUTT “OD “Aqteooy od44 oy} Woy rvejour ysay soddn 4yoy od4400 Jo usz90y4ed UMOID “g “ad 44 Jo yy004 yooyo zoddn 4ystr Jo usr9yyed uMOID “Ww "say wajnof snnby—'9z "oy a | Ss >) A Critical Revision of the Quaternary Perissodactyla. 463 Measurements: Skull: Length of upper jaw . f ‘ : . About 44-5 cm. Breadth across the first molars . : : . 16-5 em. Length of Pm?-M? : . 20-6 cm. Diastema between Pm? and I3 . Approximately 11 em, _ Teeth: em? Pm? Pm* M! M? M? Breadth . : 2 mime 3lmm. 31mm. 30mm. 29mm. 24mm. Length . : 44 mm. 309mm. 32mm. 32mm. 33mm. 32 mm. Length of OEOROLS 13mm. 145mm. 15mm. 135mm. 15mm. 15 mm. Height (ex. roots) ~. 56 mm. 66mm. 70 mm. 64mm. 74mm. 76 mm. The following account is based on the type description, much being directly or indirectly quoted from Wells’ paper. The dimensions of the skull exceed those of many specimens of E. caballus* and considerably exceed those of the zebrine horses. The lateral aspect of the maxilla in its upper portion, below the naso- maxillary suture, shows a marked hollowing. This region thus presented a form seen in the asses and in the quagga group, but not in the caballine horses nor, to any marked degree, in the mountain zebra. The teeth are large, being intermediate in size between those of EH. capensis (EH. cawoodi) and E. kuhm, both of which they somewhat resemble. The halves of the ectoloph are deeply concave inwards, and curve rather sharply into the styles with a tendency for the mesostyle to overhang anteriorly. Both parastyle and mesostyle are prominent and massive, especially in the premolars, and are more or less conspicuously grooved. The protocone is elongate, and its connection with the protoloph is established very near its anterior extremity, so that the elongation appears to affect chiefly the posterior lobe. The medial wall of the protocone is sinuous, with an ill-defined median groove. In Pm? and Pm‘ the protocone has a remarkably oblique direction, and these two teeth in the type show an unusual prominence of the hypostyle.* The caballine fold is well developed and may be duplicated. The enamel of the fossettes is irregularly crimped and its secondary plications are numerous but relatively coarse. The pli-protoloph and pli-hypostyle are well developed in early wear. The pli-protoconule is unusually complex and even in M! is duplicated. The pli-postfossette is also rather complex, but the pli-prefossette is simpler, though small secondary folding may be present. Another tooth found half a mile distant from the skull fragments at a depth of five feet in similar soil has been referred to this species. * This is almost certainly a feature of early wear. 464 Annals of the South African Museum. It is an upper left first molar and shows all the characteristics of the type, but is from a different individual, also young. The measure- ments of this tooth are: Breadth . : ; - ‘ . 30mm. Length : - : : ; . 32 mm. Height (excluding roots) : : . 65 mm. The pattern it presents is a little more complex in its secondary plication than that of the type (fig. 26, B). A section has been made through the tooth 15-20 mm. below the grinding surface, where the pattern is somewhat simplified but remains essentially the same (fig. 26, C.) In the McGregor Museum there occurs a right upper M? (M.M.K.., 4342) also from Koffiefontein which Wells refers to his species (fig. 26, D). It is more worn than the corresponding tooth of the type and is a little smaller, its dimensions being: Breadth . 3 } : P . 28-5 mm. Length . : ; - . 29-5 mm. Height (excluding roots) . ‘ . 76 mm. The styles are narrower than in the type and are very indistinctly grooved, and the pli-hypostyle is single though remarkably large. These differences, however, are not outside the expected range of variation of such a species and the tooth is referred to E. fowlert. As has been stated above, the lower jaw which agrees so well with the type of van Hoepen’s #. plicatus, was found also at Koffiefontein, and this dentition is of a suitable size to correspond with £. fowlert. There is thus a good possibility that these names are synonymous, but such a decision must await confirmation from a better association. It also seems very likely that H. fowlert is merely a somewhat abnormal variant of E. kuhni, and that, in fact, BE. kuhni, E. harrist, E. plicatus and E. fowleri represent only a single species.* Equus sandwitht Haughton. Equus sandwithi Haughton 1932. Ann. 8. Afr. Mus., xxvii, pp. 419-421, fig. 4. Cotypes: Upper right Pm? and Pm? and incomplete M°. Upper left Pm* and Pm+. Lower right M, and incomplete Pm. Lower left M, and M,. S.A. Mus., Nos. 6577 and 6578. (Fig. 27.) Locality: Usakos, South-West Africa. * Also see Appendix. A Critical Revision of the Quaternary Perissodactyla. 465 yt ge aa L Meo LM, Fie. 27.—Equus sandwitht Haughton. Cotype upper and lower cheek teeth. Natural size. (Original.) VOL. XXXI, PART 4. 39 466 Annals of the South African Museum. Horizon: In a bed of hard clay and sand 10 feet thick underlying 8 feet of surface limestone. Measurements : Upper RPm? RPm?® RM RPm? LPm* Breadth ; . . 265mm. 28 (?) mm. 245 mm. 29mm. 32 mm. Length : : 2/4 poo at 285mm. 29(?)mm. 29mm. 29 mm. Height (ex. roots) . - 56mm. 54 mm. 67mm. 53mm. 57 mm. Length of protocone . “Timm. 10-5 mm. llmm. 9mm. 10mm. Lower RPm, RM, LM, LM, Breadth (Est.) : : . 16-17 mm. 15mm. 16mm. 11-5 mm. (14-5) Length ; : ; 30mm. 295mm. 29mm. 29mm. Height (ex. roots) . 70 mm. 78mm. 65 mm. 77 mm. Haughton’s measurements differ slightly from those given above, the explanation lying in the fact that the breadth quoted here is measured directly across the tooth and not along the rather sloping surface of wear. The tooth which Haughton describes as a lower left third molar is here regarded as a lower left second molar in very early wear. The very narrow enamel surface is a feature of early wear, and the crown is 3 mm. wider half an inch below its present surface. The upper teeth of this species resemble those of Equus zebra* but are a good deal larger and present an even simpler enamel pattern. The halves of the ectoloph are flattened or even bulge outwards near their centres. The parastyle and mesostyle are strongly developed and well demarcated. The mesostyle is rather narrow, and the parastyle has an oblique flattened face which may be grooved. The protocone is abnormally small, being proportionally even shorter than in £. zebra*. This shortness is due to the very small develop- ment of the anterior prolongation. The fossettes are large and lunate. They are almost devoid of folding, only the pli-protoconule being distinct while the other folds are almost indistinguishable. The pli-caballin is absent. The hypoglyph is shallow. The protoconule shows marked antero-internal flattening. The lower teeth are unusual in shape, as they present an anterior face which is oblique to the axis of the tooth instead of directly transverse to it. This is due to the posterior deflection of the point of the parastylid and its anterior face. The walls of the protoconid and hypoconid are rounded or slightly flattened and the ectostylid fold is inconspicuous or absent. The entaconid is moderately small in the molars. The metaconid and metastylid are normally large and pear-shaped, but their attitude may be distorted by the relative compression of the interior side of the tooth. The anterior part of the entoflexid tends to be slightly or moderately folded. A Critical Revision of the Quaternary Perissodactyla. 467 The ratio of size of molars and premolars is rather unusual, as the molars appear relatively to be abnormally small and narrow both in upper and lower dentitions. The type LPm*‘ owes its unusual width at least in part to the abnormal projection of the mesostyle, but it appears that the expected range in breadth would be 28-33 mm. in Pm? and Pm* and 26-29 mm. in M! and M?, with the anterior and posterior teeth notably narrower. The same applies to the lower teeth, in which the range in Pm, and Pin, may be 15-18 mm. and in M, and M, 13-17 mm. Referred Material. Haughton (1932) assigned to H. sandwithi two lower teeth from the Vaal River gravels at Sheppard Island. These teeth were discovered by van Riet Lowe but were lost in the library fire at the University of the Witwatersrand in 1931. LPm, RP? RPm4 Fic. 28.—Equus sandwithi Haughton. Referred upper and lower cheek teeth from the Kromdraai Cave. Natural size. (Original.) Amongst the material collected by Broom in the Kromdraai Cave deposit there occur several teeth which appear to belong to this species. Most of the teeth, both upper and lower, are damaged and difficult to clean, but a selection of the better ones 1s shown in fig. 28. Equus poweri Cooke. Equus powert Cooke 1939. Q Afr. J. Sci., XXXVI, PP. paul fig. 1. 468 Annals of the South African Museum. Type: Upper right first molar. Arch. Sur., No.3. (Fig. 29). Locality: Pniel, Vaal River. Horizon: ?% Younger gravels of Vaal River. Neotype: Upper left first molar. Arch. Sur., No. 213. (Fig. 29.) Locality: 35-foot shaft, Lot 197, Windsorton. Horizon: Younger gravels of Vaal River. Fic. 29.—Equus poweri Cooke. Left: Type right upper first molar. Outer and crown views. Right: Neotype left upper first molar. Outer and crown views. Natural size. (Original.) Measurements: Type RM? Neotype LM? Breadth . ‘ . 27 mm. 26 mm. Length . - ‘ 24 mm. 23 mm. Height . : : 44 mm. 64 mm. This species resembles Equus sandwitht, but appears to be very slightly smaller and to differ in several particulars. The halves of the A Critical Revision of the Quaternary Perissodactyla. 469 ectoloph are moderately flattened or bulge outwards near their centres, and the parastyle and mesostyle are well demarcated. Though prominent, the mesostyle is not nearly as well developed as in EL. sandwith. The protocone is very small, but the anterior lobe is proportionately larger than in E. sandwithi and more like E. zebra* in this respect. The fossette are large and simple, the post-fossette being subrectangular in shape and not lunate asin H. sandwitht. The pli-protoconule is small and the hyposiyle and protoloph folds are present only as small notches. The medivallum is small and narrow and devoid of any trace of a caballine fold. The protoconule shows a well-rounded antero-internal face and is not flattened as in #. sand- with. The hypoglyph is very shallow and may disappear altogether towards the base of the tooth. The estimated dimensions fall a millimetre below those of E. sanduitht. Equus broom sp. nov. Equus harrisi (pars) Broom 1928. Ann. 8. Afr. Mus., xxii, pp. 442- 443, fig. 2 (B,). Type: Upper right fourth premolar. M.M.K., No. 3939 c. (Fig. 30.) Locality: The Bend, near Barkly West. Horizon: “Middle Terrace” of the Vall River gravels. Measuremenis : RPm* Breadth . : : : -) olemm. Length 2 : : : - o2mm. Height ‘ : : ‘ . 62mm. Length of protocone . ‘ : 9 mm. This tooth, which is badly battered, was assigned by Broom to the species E. harrisi on account of the association of the specimen with the two lower type teeth of that species. This association is not a good one in view of the very disturbed nature of the deposit, and the upper tooth seems somewhat large to be correctly associated with the lowers. Itis therefore given a new name here until certain associations settle the problems of the relations between the various species named on upper and on lower teeth. The resemblance of this species to Equus sandwithi is strong, almost the only marked difference, apart from the greater size, being the shape of the ectoloph, which is smoothly concave in both halves and is only slightly overhung by the parastyle. The protocone is very short and 470 Annals of the South African Museum. unusually rounded. The hypocone projects into the medivallum as a slight bulge but the hypoglyph is small. The fossettes show more folding than is the case in H#. sandwitht, the pli-protoloph, pli- postfossette and pli-protoconule being distinct though small, and Fie. 30.—Equus broomi sp. nov. Outer lateral and crown views of the type upper right fourth premolar of Equus broomi sp. nov.; formerly referred to Equus harrisi Broom. Natural size. (Original.) there is a certain amount of minor folding in the posterior wall of the prefossette. | The specimen on which van Hoepen created the new genus Sterro- hippus and species robustus might belong to this species, but van Hoepen’s specimen is a second premolar which cannot be regarded as forming adequate material for specific identification. There is no reason to support the generic status of Sterrohippus, which is indis- tinguishable from many species of Equus, and the specific name Hquus A Critical Revision of the Quaternary Perissodactyla. 471 robustus is preoccupied and cannot be retained. Sterrohippus robustus is therefore regarded as invalid. Incertae sedis. Equus simplex van Hoepen 1930. Pal. Nav. Nas. Mus. Bloemfontein, SI p25; fig. 6. Type: Upper-right first molar. (Nas. Mus., No. 284). Locality: Koffiefontein, O.F-.S. Horizon: % Surface. Measurements: RM1 ‘ Breadth . , . 380mm. Length . : - ag) naan, Height . : - _o9 mm. The type specimen has not been seen by the writer, and it is not possible from the drawing to assign the tooth with certainty to any particular species. If the dimensions given in the text of the type description are correct, the tooth may belong to FE. capensis as suggested by Haughton (1932). Otherwise from the text-figure it resembles #. kuhna more closely than H. capensis, so that the position (or the validity) of this species is uncertain. GuIDE TO EQUINE SPECIES. Incisors. A. Very broad mandibular symphysis with large, anteriorly flattened first and second incisors and reduced third incisors. (Upper incisors unknown.) Eurygnathohippus cornelianus v. Hoep. B. Incisors of the normal (i.e. Equus caballus*) type possessing the cup or “mark” in the upper jaw but lacking it in the lower jaw. Equus burchellic* (Gray). C. Incisors of the “normal” type with the cup or “mark” present at least in the first and second incisors of both jaws. 2 Hquus quagga* Gmelin. Equus zebra* Linn. ? Hquus capensis Broom. ? Equus kuhn Broom. 2 Equus fowlert Wells. 472 Annals of the South African Museum. D. Incisors unknown. Equus harrist Broom. Equus plicatus (v. Hoep.). Equus sandwitht Htn. Equus poweri Cooke. Equus broom: Cooke. (Equus simplex v. Hoep.) Notohipparion namaquense Htn. Stylohipparion steytlert (v. Hoep.). Upper Cheek Teeth. A. Protocone isolated; parastyle anteriorly flanged. Stylohipparion steytlert (v. Hoep.). (Pm. ? 19-24 mm. M. 18-23 mm.) * B. Protocone attached. 1. Breadth less than 27 mm. (a) Ectoloph halves concave; parastyle obliquely flattened anteriorly. Equus burchellu* (Gray). (Pm. 22-26 mm. M. 21-25 mm.) (b) Ectoloph halves flattened. Equus zebra* Linn. (Pm. 22-27 mm. M. 21-26 mm.) Equus quagga* Gmelin. (Pm. 22-27 mm. M. 21-26 mm.) 2. Breadth greater than 27 mm. (a) Ectoloph halves concave; parastyle usually flattened. Equus capensis Broom. (Pm. 29-35 mm. M. 27-33 mm.) Equus kuhni Broom. (Pm. 27-30 mm. M. 26-29 mm.) Equus broom: Cooke. (Pm. 28-33 mm. M. 27-32 mm.), Equus fowlert Wells. (Pm. 29-33 mm. M. 28-32 mm.) (6) Kctoloph halves flattened; parastyle prominent. Some #. kuhni premolars. * These dimensions are range of breadth. A Critical Revision of the Quaternary Perissodactyla. 473 Equus sandwithi Htn. (Pm. 28-33 mm. M. 26-29 mm.) Equus powert Cooke. (Pm. ? 26-29 mm. M. 25-28 mm.) . Position uncertain. Equus simplex v. Hoep. . Upper cheek teeth unknown. Equus plicatus (v. Hoep.). Equus harrist Broom. Notohipparion namaquense Htn. Eurygnathohippus cornelianus v. Hoep. Lower Cheek Teeth. . External ectostylid column present. 1. Strong protoconid fold or pillar; metaconid and metastylid strong. Notohipparion namaquense Htn. (Pm. 15-16 mm. M. 14-17 mm.) 2. Protoconid fold weak or lacking; metaconid and metastylid small. Stylohipparion steytleri (v. Hoep.). (Pm. 11-15 mm. M. 10-13 mm.) . No external ectostylid column. 1. Face of parastylid oblique; metaconid and metastylid strong. Equus sandwitht Htn. (Pm. 15-18 mm. M. 13-17 mm.) 2. Face of parastylid normal. (a) Breadth less than 15 mm. Equus zebra* Linn. (Pm. 12-15 mm. M. 11-14 mm.) Equus burchelli* (Gray). (Pm. 12-15 mm. M. 11-14 mm.) Equus quagga* Gmeiin. (Pm. 12-15 mm. M. 11-14 mm.) (Some molars of E. plicatus resemble LE. quagga*.) (b) Breadth greater than 15 mm. Equus harrisi Broom. (Pm. 17-20 mm. M. 16-19 mm.) 474 Annals of the South African Museum. Equus plicatus (v. Hoep.). (Pm. 15:5-19 mm. M. 14:5-17-5 mm.). Equus capensis Broom. (Pm. 18:5-22 mm. M. 17-5-21 mm.) C. Lower dentitions unknown. Eurygnathohippus cornelianus v. Hoep. Equus kuhni Broom. Equus fowlert Wells. Equus broom: Cooke. Equus powert Cooke. REFERENCES. ALLEN, G. N., 1939. ‘““A Checklist of African Mammals”, Bull. Mus. of Compar. Zool. Harvard, vol. lxxxiii, pp. 1-763. Brcx, R., 1906. “Mastodon in the Pleistocene of South Africa”, Geol. Mag., Decade V, vol. iii, pp. 49-50. Broom, R., 1909. “‘On the Evidence of a Large Horse recently Extinct in South Africa”’, Ann. S. Afr. Mus., vol. vii, pp. 281-282. ——, 1913 a. “Note on Equus capensis’’, Bull. Amer. Mus. Nat. Hist., vol. xxxii, pp. 437-439. ——, 19136. ‘Man Contemporaneous with Extinct Animals”, Ann. 8. Afr. Mus., vol. xii, pp. 13-16. ——, 1928. ‘On Some New Mammals from the Diamond Gravels of the Kimberley District”’, Ann. S. Afr. Mus., vol. xxii, pp. 439-444. —, 1937. ‘‘New Pleistocene Mammals from Limestone Caves of the Transvaal’’, S. Afr. J. Sci., vol. xxxiii, pp. 750-768. Broom, R., and Lz Ricuez, H., 1937. ‘The Dentition of Equus capensis Broom”’, 8. Afr. J. Sci., vol. xxxiii, pp. 769-770. Cuuss, E. C., 1907. “List of Vertebrate Remains” in “On an African Occurrence of Fossil Mammalia associated with Stone Implements’’, by F. P. Mennel and KE. C. Chubb, Geol. Mag., Decade V, vol. iv, pp. 447-448. Cooxg, H. B. S., 1939. “On a Collection of Fossil Mammalian Remains from the Vaal River Gravels at Pniel”’, S. Afr. J. Sci., vol. xxxvi, pp. 412-416. ——, 194la. “A Preliminary Account of the Wonderwerk Cave, Kuruman District. Section II.: The Fossil Remains”, S. Afr. J. Sci., vol. xxxvii, pp. 303-311. —, 19416. “A Preliminary Survey of the Quaternary Period in Southern Africa”, Bur. of Archaeology, Arch., Ser. iv, pp. 1-59. ——,1948. ‘Cranial and Dental Characters ofthe Recent South African Equidae’’, S. Afr. J. Sci., vol. xl, pp. 254-257. (See also WELLS, CooKE and Maan, 1942; Matan and Cooks, 1941.) A Critical Revision of the Quaternary Perissodactyla. 475 Dreyer, T. F., in Dreyer, T. F., and Lyuz, A., 1931. ‘New Fossil Mammals and Man from South Africa”, Dept. Zool. Grey. Univ. College, Bloemfontein. Fraas, E., 1907. » Pleistocane Fauna aus dem Diamantseifen van Siid-Afrika”’, Zeit. d. Deutsch. geol. Gesell., vol. lix, pp. 232-243. GIDLEY, J . W., 1901. ‘“‘Tooth Characters and Revision of the North American Species of the Genus Hquus”’, Bull. Amer. Mus. Nat. Hist., vol. xiv, pp. 91-141. Grecory, W. K., 1934. “A Half-Century of Trituberculy: the Cope-Osborn Theory of Dental Evolution”’’, Proc. Amer. Phil. Soc., vol. Ixxiii, No. 4, pp. 169-317. HaveutTon, 8. H., 1932. “The Fossil Equidae of South Africa”’, Ann. 8. Afr. Mus., vol. xxviii, pp. 407-427 (February 1932). Horwoop, A. T., 1926. “Fossil Mammalia” in “Geology and Palaeontology of the Kaiso Bone Beds’’, Occ. Paper No. 2, Geol. Survey of Uganda. —, 1928. ‘‘Mammalia” in Rhodesian Man and Associated Remains, London: British Mus. (Natural History), pp. 70-73. Le Ricus, H., 1937. See Broom, R., 1937. Linnaeus, C., 1758. Systema Naturae, 10th Edition. Lowe, C. vAN Rizt, 1937. See S6HNGE, VissER and Lowe, 1937. Maay, B. D., and Cooks, H. B.S.,1941, ‘‘A Preliminary Account of the Wonder- werk Cave, Kuruman District”, 8. Afr. J. Sci., vol. xxxvii, pp. 300-312. Matay, B. D., 1943. See Wetts, L. H., 1943. Ossorn, H. F., 1907. Evolution of Mammalian Molar Teeth. New York: Macmillan & Co. (See also Scorr and Ossory, 1890). Owen, R., 1869. “Description of the Cavernof Bruniquel and its Organic Contents. Part II. Equine Remains”, Phil. Trans. Roy. Soc. London, vol. clix, pp. 535-557. Romer, A. S., 1933. Vertebrate Paleontology. Chicago: University of Chicago Press. Scort, W. B., 1907. ‘‘Mammalian Remains from the Coast of Zululand’’, Third and Final Report, Geol. Survey Natal and Zululand, pp. 251-262. Scott, W. B., and Ossorn, H. F., 1890. ‘“‘Preliminary Account of the Fossil Mammals from the White River and Loop Fork Formation contained in the Museum of Comparative Zoology. The Perissodactyla”, by H. F. Osborn, Bull. Mus. Comp. Zool., vol. xx, No. 3. Szriey, H. G., 1891. “On Bubalus bainii (Seeley)”, Geol. Mag., vol. viii, pp. 199-201. Suaprro, M. M. J., 1943. “Fossil Mammalian Remains from Bankies, Kroonstad District, O.F.S.”, S. Afr. J. Sci., vol. xxxix, pp. 176-181. Suaw, J. C. M., 1938 a. “The Teeth of the South African Fossil Pig (Notochoerus capensis syn. meadowsi) and their Geological Significance”, Trans. Roy. Soc. S. Afr., vol. xxvi, pp. 25-37. ——, 1938. b.. “Growth, Changes and Variations in Warthog Third Molars and their Palaeontological Importance”, Trans. Roy. Soc. S. Afr., vol. xxvii, pp. 51-94. Sounce, P. G., Visser, D. J. L., and Lows, C. van Riet, 1937. “Geology and Archaeology of the Vaal River Basin”, Geol. Mem. 35, Geol. Survey of the Union of S. Afr. Srirton, R.A., 1941. “Development of Characters in Horse Teeth and the Dental Nomenclature”, Journ. of Mammalogy, vol. xxit, pp- 434-446. 476 Annals of the South African Museum. van Horpen, E. C. N.,1930a. ‘‘Vrystaatse Fossiele Perde’’, Pal. Nav. Nas. Mus. Bloemfontein, II, pp. 1-11. ——, 1930 b. ‘‘Fossiele Perde van Cornelia, O.V.S.”, Pal. Nav. Nas. Mus. Bloemfontein, II, pp. 13-24. —, 1932. ‘Die Stamlyn van die Sebras”’, Pal. Nav. Nas. Mus. Bloemfontein, IT, pp. 25-37 (September 1932). —, 1940. “Oor die tande van die Equinae: 1. Die snytande van die Onderkaak”’, Tydskrif vir Wetenskap en Kuns, vol. i, pp. 101-114. Visser, D. J. L., 1937. See S6uncE, P. G., 1937. WELLS, L. H., 1941. ‘A Fossil Horse from Koffiefontein, O.F.S.”, Trans. Roy. Soc. 8. Afr., vol. xxviii, pp. 301-306, pl. lv. WELLS, L. H., Cooxsn, H. B. S., and Maan, B. D., 1942. ‘*‘The Associated Fauna and Culture of the Vlakkraal Thermal Springs, O.F.S.’’, Trans. Roy. Soc. 8. Afr., vol. xxix, pp. 203-233. WELLS, L. H., and Maran, B. D., 1943. “A Further Report on the Wonderwerk Cave, Kuruman. Section I: Archaeology, by B. D. Malan. Section JI: Fauna, by L. H. Wells”, 8S. Afr. J. Sci., vol. xl, pp. 258-270. ZITTEL, K. A. von, 1925. Text-book of Palaeontology. Vol. III. ““Mammalia”’, ed. by Sir A. Smith Woodward. London: Macmillan & Co. A Critical Revision of the Quaternary Perissodactyla. 477 APPENDIX (With Fig. 31.) In the five years which have elapsed since this account of the Perisso- dactyla was written, certain additional information has come to light which, while it does not involve any important alteration of the views already expressed, nevertheless requires to be added in order to bring the work reasonably up to date. Firstly, the description of the fossil mammals of the Vaal River deposits, mentioned in the Introduction to the present account, has appeared as Memoir 35 (III) of the Geological Survey of the Union of South Africa and it contains a certain amount of stratigraphic informa- tion on the distribution of the Equidae in those deposits (Cooke, 1949 6). In addition, revisions of the Suina (Cooke, 1949 a) and of the Proboscidea (Cooke, 1947) have been published as companion studies to the present one. Secondly, a perissodactyl family new to this region has been recorded through the discovery of an undoubted Chalicothere in the cave breccia of the Makapan valley (George, 1950). The type isa left upper second molar and there are also a number of other cheek teeth and an ungual phalanx. The type molar and the phalanx are illustrated here (fig. 31). Chalicothere remains have previously been reported from the Kaiso beds of Uganda (Andrews, 1923) and from south Serengeti in Tanganyika (Dietrich, 1942), but the Makapan specimens appear distinct and were described as a new species tentatively placed in Metaschizotherum as M. transvaalensis. The resemblance to the Tanganyika species M. henmgi is fairly close and it is considered by Miss George that both species certainly belong to the same genus, whether or not this is really Metaschizotherium. Thirdly, a new species of fossil equine has been erected by Broom (1948) based on material from a cave breccia in the Sterkfontein area about half a mile south-west of the site known as “Bolt’s workings”’. The type of this species, Equus zietsmani, comprises an imperfect lower jaw with the incisors, most of the cheek teeth of the right side and some of those of the left side, and part of the left maxilla. The lower series from P, to M, can be well reconstructed and is figured by 478 Annals of the South African Museum. Broom, as also is the damaged upper series from P* to M (the latter tooth being a mere fragment). The lower series agrees very well with that shown in fig. 24 of the present text and referred to E. harrist. The upper series resembles that illustrated in fig. 23, which was referred to E. kuhni, and also shows features suggestive of #. fowlerr. In view of these resemblances and especially as the teeth shown in figs. 23 and 24 were also from the Sterkfontein group of breccias, it is likely that only a single species is represented by the three lots of specimens. It remains to be considered what that species is. Fic. 31.—Metaschizotherium (?) transvaalensis George. Top left: Anterior view of ungual phalanx. Lower left: Side view of ungual phalanx. Right: Crown view of type left upper second molar. All natural size. (From George.) It has already been suggested in the present revision (p. 464) that Equus fowlert and EH. kuhni may be synonyms and that the corre- sponding lower dentitions may be represented by F#. plicatus and E. harrist. This idea lacked confirmation due to the absence of associated upper and lower dentitions but Broom’s new material goes a long way towards correcting this deficiency and greatly strengthens the probability that all five species are identical. It may now be considered that EF. fowleri, E. plicatus, E. harrisi and E. zietsmani are synonyms of Equus kuhni. Some additional confirmation of the above conclusion can be obtained from a study of the dental characters of the living Grevy’s A Critical Revision of the Quaternary Perissodactyla. 479 Zebra of Kast Africa. It was suggested to the writer by Dr. L. S. B. Leakey that the South African Equus kuhni was, in fact, Equus grevyt. Through the courtesy of Dr. Leakey the writer had the opportunity, while in Hast Africa, of studying a small collection of Grevy skulls and it is clear that there is a close resemblance between the teeth of the living animal and the fossil material referred to E. kuhm. Individual teeth, but not whole dentitions, exhibited characters recalling H. fowler, EL. plicatus and EH. harrisi and the suggested identity of the four fossil species thus receives a measure of confirmation. There can be no doubt that the fossil HL. kuhni (and its presumed synonyms) is closely related to H. grevyi and the relatively small differences which have been noted could be dismissed as due to individual variation. For the present, however, it is felt that it will be better to retain the designation EH. kuhni for the fossil material until additional skeletal material is available to support the absorption of LE. kuhni into EF. grevyt. REFERENCES. AnpreEws, ©. W., 1923. In Nature, 10th November 1923. Broom, R., 1948. ‘‘Some South African Pliocene and Pleistocene Mammals”’, Ann. Tvl. Mus., vol. xxi, pp. 23-25. Cooxs, H. B. S., 1947. ‘‘ Variation in third molars of the living African Elephant and a Critical Revision of the Fossil Proboscidea of Southern Africa”’, Amer. J. Sci., vol. cexlv, pp. 434-457, 492-517. ——, 1949a. “The Fossil Suina of South Africa’’, Trans. Roy. Soc. 8. Afr., vol. xxxii, pp. 1-44. ——, 1949. ‘“‘Fossil Mammals of the Vaal River Deposits’’, Geol. Mem. 35 (IIT), Geol. Survey of the Union of 8. Afr., pp. 1-109. Dierricu, W. O., 1942. In Palaeontographica, Band xciv, Abt. A, 94. Gzrorcs, M., 1950. ‘‘A Chalicothere from the Limeworks Quarry of the Makapan Valley, Potgietersrust District”, S. Afr. J. Sci., vol. xlvi, pp. 241-242. \ ‘ f * lod ‘ ‘ y ‘ "4 \ ‘ , 4 ‘ 2 “ * - / ' ' ] . \ ‘ ‘ A \ \ j s The ANNALS OF THE SOUTH AFRICAN MUSEUM are issued in parts at irregular intervals as material becomes available. As far as possible each volume is devoted exclusively to a particular subject (Zoology, Botany, etc.). Two or more volumes may be in course of publication concurrently. Most of the Geological and Palaeontological papers are issued in conjunction with the Geological Survey of the Union of South Africa. - Some volumes and parts are out of print, and others are only sold as parts of a set, or volume, respectively. The prices of parts published prior to 1940 have been increased. Out of print: Vols. I, II, V (Part 9), VII, VIII, [X (Part 1), XII (Part 7), XXII, XXIV (Part 2). Vol. £ se. d. III. 1903-1905 Zoology 119 0 IV. 1903-1908 Palaeontology 3.0 6 V. 1906-1910 Geology, Palaeontology, Zool, Anthropology l ee VI. 1908-1910 Zoology 6 IX. 1911-1918 Botany arn Part 1) 216 0 X. 1911-1914 Zoology . +i XJ. 1911-1918 Zoology 2. + . oe XII. 1913-1924 Palaeontology met Gacions .. (excl. Part 7) 313 0O XIII. 1913-1923 me ied and preey 4 ' 3.18 XIV. 1915-1924 Zoology : ‘, 3 6 6 XV. 1914-1916 Zoology 45 0 XVI. 1917-1933 Botany 311 0 XVII. 1917-1920 Zoology 310 0 XVIII 1921 Zoology 45 6 XIX. 1924-1925 Zoology 3, e XX. 1924-1926 Zoology 212 0 XXII. 1925-1927 Zoology 3.64 XXIII. 1925-1926 Zoology ihe ye .. 1 oe XXIV. 1929-1938 sAnithropelory bse Ethnology .. (excl. Part 2) 2 9 6 XXV. 1927-1928 Zoology . 119 0 XXVI. 1928 Zoology 110 0 XXVIII. 1929 Anthropology 110 0 XXVIII. 1929-1932 Palaeontology 2.12 0 XXIX. 1929-1931 Zoology 2 SD XXX. 1931-1935 Zoology 313 6 INDEX of papers, authors, and subjects, published ’ in Vols. I-XXXx 0; tas XXXII. 1934-1950 Palaeontology ie : Jee oe 315 0 XXXII. 1935-1940 Zoology 3.9 6 XXXII. 1939 Zoology y a ae | XXXIV. 1938 . Zoology 2 8 0 XXXV. Reserved for conclusion of sieghapeinels in n Vk XXXIV. XXXVI. 1942-1948 Zoology sie Pe -