*fg'?0 HARVARD UNIVERSITY Library of the Museum of Comparative Zoology BULLETIN OF THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD COLLEGE, IN CAMBRIDGE VOL. 1^20 CAMBRIDGE, MASS., U.S.A. 1959 The Cosmos Press, Inc. Cambridge, Mass., U.S.A. CONTENTS PAGE No. 1. — The Herpetology op Southern Rhodesia. Part 1. Snakes. By Donald G. Broadlev. (6 plates). March, 1959 .' . . . . ' . . . 1 No. 2. — Studies on the Comparative Embryology of the Reptilian Nose. By Thomas S. Parsons. (7 plates). March, 1959 .' 101 No. 3. — The Rodents of the Deseadan Oligocene of Patagonia and the Beginnings of South Amer- ican Rodent Evolution. By Albert E. Wood and Bryan Patterson. May, 1959 . . . . 279 No. 4. — The Types of Corbiculidae and Sphaerhdae (]\[ollusca: Pelecypoda) in the Museum of Comparative Zoology, and a Bio-Bibliographic Sketch of Temple Prime, an Early Specialist op the Group. By Richard I. Johnson. (8 plates ). May, 1959 429 No. 5. — Stitdies on the Ant Fauna of Melanesia. V. The Tribe Odontomachini. By Edward 0. Wilson. (2 plates). May, 1959 .' 481 Bulletin of the Museum of Compcorative Zoology AT H A E V A R D C O L L E (i K Vol. 120, No. 1 THE HERPETOLOGY OP SOUTHERN RHODESIA PART 1. SNAKES By Donald G. Rroadi.ey Ildiioniiy Kt'cpi-r of Ilcrpetology, National Miiseii f Southern Rhodesia With Six Plates CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE M U S E IT M March, 1959 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD COLLEGE Bulletin (octavo) 1863 — The current volume is Vol. 120. Breviora (octavo) 1952 — No. 101 is current. Memoirs (quarto) 1864—1938 — Publication was terminated with Vol. 55. JoHNSONiA (quarto) 1941 — A publication of the Department of Mollusks. Vol. 3, no. 38 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 22 is current. Proceedings of the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained on application to the Director of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Of the Peters "Check List of Birds of the World," volumes 1-3 are out of print ; volumes 4 and 6 may be obtained from the Harvard Uni- versity Press; volumes 5 and 7 are sold by the Museum, and future volumes will be published under Museum auspices. Bulletin of the Museum of Comparative Zoology AT HAEVAED COLLEGE Vol. 120, No. 1 THE HERPETOLOGY OF SOUTHERN RHODESIA PART 1. SNAKES By Donald G. Broadley Honorary Keeper of Herpetology, National Museum of Southern Rhodesia With Six Plates CAMBEIDGE, MASS., U.S.A. FEINTED rOE THE MUSEUM IVIarch, 1959 No. 1 — The Herpetology of Southern Rhodesia. Part 1. Snakes. By Donald G. Broadley CONTENTS Page Introduction 3 Acknowledgements 5 Index to List of Species Dealt With 6 Systematic Discussion 8 Bibliography 81 A Key to the Snakes of Southern Rhodesia 88 Alphabetical Index of Localities in Southern Ehodesia 95 INTRODUCTION This is the first comprehensive Check List of the Snakes of Southern Rhodesia to be published. The paper has been ex- panded to include much original ecological material and with the addition of a systematic key it should serve as a sound basis for the serpentology of the colony. This work has been undertaken at the suggestion of Dr. E. E. Williams of the Museum of Comparative Zoology, after I had .submitted a brief outline of the herpetology of Southern Rho- desia to Mr. Arthur Loveridge at that Museum. The material examined consists of all the specimens in the National Museums of Southern Rhodesia at Bulawayo and Salis- bury, plus a collection made by the Umtali Branch of the Her- petological Association of Rhodesia. I have also obtained the data for all Southern Rhodesian specimens in the Transvaal Museum and the British Museum (Natural History). Finally, I have examined a large number of snakes which were too badly decomposed or damaged to preserve. Altogether I have collated the data for 1385 specimens representing 61 species or races. Chubb 's 1909 list contained 30 species from Matabeleland, and subsequent additions brought tlie total up to 52. The following 4 BULLETIN : MUSEUM OF COIMPARATIVE ZOOLOGY forms are recorded from Southern Rhodesia for the tirst time : Leptotyphlops longicauda (Peters) Lycodonomorphvs rufidus ynlanjensis Loveridge Meizodon semiornata semiornata (Peters) Philothamnus ornatus Boeage Prosymna sundevallii siindevallii (A. Smith) Dromophis Uneatus (Dumeril and Bibron) Psammophis angolensis (Boeage) Aparallactus lumdatus lunidatus (Peters) Naja melanoleuca Hallowell There are very few taxonomic alterations. Naja nigricollis niossamhica Peters is revived for certain light-coloured cobras with 21-25 midbody scale rows, which are found in Southern Rhodesia and Mozambique, extending north into Northern Rho- desia, Nyasaland and Tanganyika, and south into the Union of South Africa. Matabeleland specimens of Aparallactus capensis are regarded as intermediates between the typical form and the western race hocagei. Most specimens of Atractaspis hihronii from Southern Rhodesia prove to be intermediates between the typical form and the northern race rostrata. Systematic Discussion. In this section is presented all the in- formation at present available on the snakes of Southern Rho- desia. The subjects covered are best discussed under the headings employed. Citations of literature. The original description of each form is given in full and these references are consequently omitted from the Bibliography on pp. 81-84. This is followed chronolog- ically by every reference to Southern Rhodesian material in the herpetological literature that I have been able to trace. Native Nannies . The present generation of Africans use very few names for snake species and the few that I have recorded were gleaned from the older men. There is much confusion among the younger generation. In Matabeleland, Pimpi is the name correctly applied to the Spitting Cobra (Naja n. mossani- bica), but it is now often applied to any grey or brown snake, even a female Boomslang (Dispholidus typus) . A full list of English names is included in the systematic key on pp. 88-95. BROADLEY : SNAKES OF SOUTHERN RHODESIA 5 Size. Where a specimen is referred to by a registered museum number, the followin»>- prefixes are used : NM — National Museum of Southern Rhodesia, Bulawayo. SM — Queen Victoria Museum, Salisbury. UM — Umtali Museum, Umtali. TM — Transvaal Museum, Pretoria. MCZ — Museum of Comparative Zoology, Harvard. BM — British Museum (Natural History) . Variation, Colouration, Size, Sexual dimorphism. Breeding, Diet, Parasites, Enemies, Defence, Venom, Hahits and Hahitat. These data are based on Southern Rhodesian material only. Localities. Under this heading are listed : all Southern Rho- desian localities found in the literature ; the localities of all specimens in the six museums listed above and the additional material examined by myself or members of the Herpetological Association of Rhodesia ; a few localities are based on personal positive sight records. Key to the Snakes of Southern Rhodesia. This is based purely on local material and the couplets do not necessarily hold good for other regions. Plates and. Text Figures. Live subjects were used for all the photographs. The text figures illustrating the key (Figs. 7-10) were personally drawn from local specimens. ACKNOWLEDGEMENTS The opportunity is taken of thanking Mr. Arthur Loveridge and Dr. E. E. Williams of the Museum of Comparative Zoology for their constant encouragement and advice leading to the production of this paper. I am indebted to Mr. Loveridge and Dr. Vivian FitzSimons of the Transvaal Museum for identify- ing specimens, answering numerous queries and supplying me with data for Southern Rhodesian material in their charge. I would also express my thanks to Mr. J. C. Battersby and Captain C. R. S. Pitman for supplying me with information on the rele- vant material in the British Museum (Natural History). I am most grateful to Mr. R. H. Smithers and the staff of the National Museum of Southern Rhodesia for their help and co- operation. Much valuable material has been collected by the Herpetological Association of Rhodesia, particularly W. Arm- 6 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY itage, D. K. Blake and S. Warren of Umtali Branch and A. H. Siemers of Salisbury. Father K. Tasman, S.J., very generously supplied me with much useful information from his notes, com- piled during many years of collecting in Rhodesia. Thanks are also due to Dr. G. Theiler of the South African Veterinary Department for identifying the ticks, and Dr. F. Zumpt of the South African Institute for Medical Research for identifying the mites. Index to List op Species Dealt With TYPHLOPIDAE Page Typhlops delalandii Schlegel 8 Typhlops schlegelii mucruso (Peters) 8 LEPTOTYPHLOPIDAE Leptotyphlops conjioncta (Jan) 9 Leptotyphlops scutifrons (Peters) 10 Leptotyphlops longicanda (Peters) 10 BOIDAE Python sebae (Gmelin) 11 COLUBEIDAE Lyeodonomorphus rufulus rufulm (Lichtenstein) 12 Lycodonomorphus rufulus mlanjensis Loveridge 13 Boaedon fuliginosus fuliginosus (Boie) 14 Lycophidion capense capense (A. Smith) 15 Mehelya capensis capensis (A. Smith) 16 Mehelya nyassae (Giinther) 18 Natriciteres olivacea olivacea (Peters) 18 Natriciteres olivacea uluguruensis (Loveridge) 20 Meizodon semiornata semiorimta (Peters) 21 Philothamnus hoplogaster (Giinther) 21 Philothamniis ornatus Bocage 22 Philothamnus irregularis irregularis (Leach) 23 Philotliamnus semivariegatus semivariegatus (A. Smith) 24 Prosymna sundevallii sundevallii (A. Smith) 27 Prosymna lineata (Peters) 28 Prosymna ambigua stuhlmanni (Pfeffer) 29 Pseudaspis eana (Linne) 30 Buberria lutrix rhodesiana Broadley 31 Telescopus semiannulatus semiannulatxis A. Smith 32 Crotaphopeltis hotamboeia hotam,boeia (Laurenti) 33 BROADLET : SNAKES OF SOUTHERN RHODESIA 7 Chamaetortus aulicus aulicus Giinther 35 Dispholidus typus (A. Smith) 35 Thclotornis kirtlandii oate.sii (Giinther) 37 Thelotornis kirtlandii capensis A. Smith 38 H emir hag errhis nototaenia nototaenia, (Giinther) 41 Amplorhinus multimaoulatus A. Smith 42 Psammophylax tritaeniatus tritaeniatus (Giinther) 42 Rhamphiophis oxyrhynchus rostratits Peters 44 Dramophis lineatus (Dumeril and Bibron) 44 Psammophis sibilans sibilans (Linne) 45 Psammophis subtaeniatus subtaeniatus Peters 47 Psammophis jallae Peracea 49 Psammophis crucifer (Daudin) 49 Psammophis angolensis (Socage) 51 Calamclaps unicolor miolepis Giinther 52 Calamelaps ventrimaculatus websteri FitzSimons and Brain 53 Xenocalamus bicolor blcolor Giinther 54 Chilothinophis gerardi gerardi (Boulenger) 54 Aparallactus lunulatus lunulatus (Peters) 55 Aparallactus guentheri Boulenger 56 Aparallaetus cnpensis capeuMs A. Smith 56 Aparallactus capensis capensis X boc-er, 1913, p. 162; FitzSinions, V. F., 1939, p. 22; Tasman, 1953, p. 35; Rose, 1955, p. 93; Isemonger, 1955, p. 70. Chlorophis neglectus Boulenger, 1910, p. 507; FitzSinions, F. W., 1912, p).. 86, 87; Eose, 1955, p. 93; Iscniouger, 1955, p. 70. Philothaimius hoplngastcr Broadley, 1957a, p. 53. Variation. (30 specimens.) Midbody scale rows 15; ventrals 148-160; anal divided; subcaudals 77-103; upper labials 8, the fourth and fifth entering the orbit (an Umtali snake has only 7 labials on one side, the third and fourth entering the orbit) ; 22 BULLETIN : MUSELTM OF COMPARATIVE ZOOLOGY lower labials 9-11, the first four or five in contact with the anterior sublinguals; preocular 1; postoculars 2; temporals 1+1. Tail length .25 to .30 of the total. Colouration. Light blue-green to grass-green above, olive-green when about to slough ; white below. Some specimens have up to a dozen black blotches on the neck. Size. Largest (B.M. 02.2.12.87), 945 (685+260) mm. from Mazoe. Smallest (SM/R. 19) 234 (164+70) mm. from Odzi. Breeding. A 618 mm. 9 from Cleveland Dam laid 4 eggs on 18.xi.55. A 484 mm. 9 from the Umvumvumvu River laid 4 eggs measuring 28 x 9 mm. on 20.i.57. Diet. A Cleveland Dam snake fed readily in captivity on frogs {Bana spp. ; Hyperolius spp.) and took a small skink [Mabuya v. varia) . Enemies. A Spitting Cobra {Naja n. mossamhica), captured beside the N'sese River at Irisvale, disgorged the tail of a South- eastern Green Snake. At Selukwe I was given an account of a crab killing and devouring a small green snake, probably refer- able to this species. Defence. Unlike the next two species of Philotliamnus, this snake does not inflate the throat or try to bite when captured. Habitat. At Salisbury hoplogaster frequents the reedy vleis to the west of the city, extending east to Cleveland Dam. In the more open sandveld streams farther east it is replaced by P. i. irregularis. The two species occur together in the Eastern Dis- tricts and at Selukwe. Distribution. Common in Mashonaland and the Eastern Dis- tricts. Very scarce in Matabeleland. Localities: Eldorado; Trelawney; Mazoe; Mount Hampden; Salisbury ; Cleveland Dam ; Odzi ; Nyamaropa ; Imbeza ; Vumba Mountain ; Umvumvumvu River ; Haroni-Lusitu Junction ; Chi- rinda Forest ; Selukwe ; Driefontein ; Ilmshagashe River ; Irisvale ; Lumane ; Mount Darwin. PlIILOTHAMNUS ORNATUS BocagC PJiiIof]uiiu)it(s o)natu.^ Boeage, 1872, Jour. Sci. Lisboa, vol. 4, p. 80. Data of unique specimen. Midbody scale rows 15 ; ventrals 161 ; anal divided ; subeaudals 96 ; upper labials 9-10, the fourth, fifth and sixth (normally the third, fourth and fifth ) entering the orbit ; BROADLEY : SNAKES OF SOUTHERN RHODESIA 23 lower labials 10, the first five in contact with the anterior sub- linguals: preocular 1 ; postoculars 2 ; temporals 1 + 1. Tail length .30 of the total. Colouration. Emerald green above, with a red-brown dorsal stripe three scales in width, narrowly edged with yellow. The anterior dorsal scales edged with black. Labials and belly white Avith a bronze tint. Size. $ (NM/M.621) 599 (420+179) mm. from Reitfontein, Salisbury. Discussion. I found this specimen freshly killed on the road where a bridge spans a vlei on the outskirts of Salisbury. It agrees well with a series of nine ornatus collected by Monsieur H. J. Bredo in the Mweru-wa-Ntipa area of Northern Rhodesia and examined by me through the courtesy of the Musee Royal d'Histoire Naturelle de Belgique. The Salisbury snake differs only in having the fourth, fifth and sixth upper labials entering the orbit instead of the third, fourth and fifth as in all nine of Bredo 's specimens. The head of this species is more rounded in profile than that of irregularis and resembles hoplogaster more in this respect. In any case ornatus must be restored to specific status ; its range in the Rhodesias overlaps that of P. i. irregularis, so it can no longer be regarded as a race of the latter. Distribution. Uncertain. Localities. Reitfontein, Salisbury. PhILOTHAMNUS IRREGULARIS IRREGULARIS (Leach) Coluber irregularis Leach, 1819, in Bowdich, Mission to Ashantee, p. 494. Chlorophis irregularis Chubb, 1909a, p. 595; 1909b, p. 35; Boulenger, 1910, p. 508; FitzSimons, F. W., 1912, p. 87; FitzSimons, V. F., 1939, p. 22: Bogert, 1940, p. 53; Eose, 1955, p. 94; Isemonger, 1955, p. 70. Philothamnus irregularis irregularis Broadley, 1957a, p. .~)3. Variation. (35 specimens.) Midbody scale rows 15; ventrals 154-169; anal divided; subcaudals 94-115; upper labials 9 (10 on one side only of two snakes), the fourth, fifth and sixth (fifth, sixth and seventh on one side of a Nyamashatu River snake) entering the orbit; lower labials 9-11, the first five (rarely six) in contact with the anterior sublinguals ; preocular 1 ; postoculars 2 (3 on one side of an Imbeza snake) ; temporals 1+1 or 1+2. Tail length .27 to .33 of the total length. 24 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Colouration. Brilliant emerald green above, some scales on anterior part of body black edged, but not forming regular cross-bars. Below, light green to yellow green. Eye with a hand- some golden iris. Size. Largest (NM/M. 736) 978+ (898+80+) mm. from Umzilizwe River. Largest perfect specimen (NM/M. 1471) 1106 (810+296) mm. from Umzilizwe River. Rate of groivth. A Domboshawa S grew from 698 to 763 mm. in 12 months of captivity, during which time he consumed 19 frogs (Rana) ; 9 toads (Bufo) and a pigmy mouse (Leggadasp.) . A Chishawasha S grew from 779 to 798 mm. in 14 months of captivity, during which time he consumed 14 frogs and 13 toads. Diet. Captive specimens fed readity on frogs {Rana spp. ; Hyperolius spp.; Phrynohatrachus natalensis) and toads {Bufo regularis; Bufo carens) ; one took a pigmy mouse {Leggada sp.). Defence. When captured or molested, this snake inflates its throat vertically and strikes fiercely. A large Chishawasha $ drew blood on my top lip as I was admiring my capture, thoi she left six rows of bleeding tooth marks on my proffered hand. Habitat. The Western Green-Snake prefers open, free running streams with plent}^ of shade. At Chishawasha it was abundant where a small stream flowed between high banks with plenty of overhanging trees. Four of these snakes were killed in the foundations of the demolished bridge at Glen Lome. One Chisha- washa snake was living in a crab hole beside a culvert, where many natives came to wash their clothes. The species is plentiful in reedbeds along the Umzilizwe River and elsewhere in the Eastern Districts. Like P. hoiilogaster it often rests in small trees overhanging the water. Distribution. Mashonaland and the Eastern Districts, extend- ing southwest to Selukwe. Localities: Victoria Falls; Glen Lome, Salisbury; Dombo- shawa ; Chishawasha ; Odzi ; Imbeza ; Umtali ; TTmvumvumvu River ; Umzilizwe River ; Cliirinda Forest ; Selukwe ; Moonies Creek ; Pungwe River. 2200' ; Nyamashatu River. Philothamnus semivariegatits semivariegatus (A. Smith) Dendrophis (Philothamnus) aemivariegata A. Smith, 1840, Til. Zodl. S. .^fricii, Eept., pis. lix, Ix, Ixiv, figs, la, lb. BROADLET : RXAKES OF ROTTTHERN RHODESIA 25 Philothamnus semivariegatus Boulenger, 1902, p. 17; 1910, p. 508; Hewitt and Power, 1913, p. 162; Tasnian, 19r)3, p. 35; Rose, 1955, p. 94; Isemon ger, 1955, p. 71. PhilotJmmnus semivariegatus semivariegatus FitzSimons, V. F., 1939, p. 2'J ; Bioadley, 1957a, p. 53. Native name of Variegated Bush-Snake. N'dlondlo (Sinde- bele), hut is mistaken for a young Boomslang (Dispholidus typus). •GATOOMA*" ^ J \ / / On ''-.. A ^* A ^ AA- SUtAWAYO , FoRt *) I • < i » '•'7 ^A'Ja; ;>^^=a!ipopo a.. Pig. 2. Eecoi'ded localities for Philothamnus. • Philothamnus hoplogaster (Giinther) ■ Philothamnus irregularis irregularis (Leach) ▲ Philothamnus semivariegatus semivariegatus (A. Sniitli) Variation. (31 specimens.) Midbody scale rows 15; ventrals 179-204; anal divided; subcaudals 121-142; upper labials 9, the fourth, fifth and sixth entering the orbit (10, the fifth, sixth and seventh entering the orbit, on one side only of a Bembesi snake) ; lower labials U) (rarely 9), thr- first five (rarely 4) in contact 26 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY with the anterior sublinguals; preocular 1; postoculars 2^ ; tem- porals 1+2 or 2+2. Tail length .28 to .34 of the total. Colouration. Head blue-green, anterior third of the body blue- green with narrow black cross-bars, fading out to uniform bronze on the posterior of body and tail. Chin white, throat bright yellow, rest of underside cream. Eye with a golden iris. Size. Largest 1108 (780+328) mm. from Umshagashe River. Breeding. On 26.i.56 the huge Umshagashe $ laid 8 eggs, measuring 26 X 10 mm. Diet. Captive specimens at Umtali feed freely on dwarf geckos {Lygodactylus c. capensis). The Umshagashe 9 swal- lowed two frogs {Rana sp.), but later disgorged them. Defence. This snake is extremely truculent and is slow to settle down in captivity. The record specimen was found crawling slowly through the grass beside the Umshagashe River. She made no attempt to dash for cover, but inflated her throat vertically, at the same time lifting her head and neck to display the yellow patch and hissing fiercely. She struck viciously and repeatedly when picked up, drawing blood several times. Habits. This species is more persistently arboreal than the two previous ones. Tasman (m litt.) states, "I have seen one {P.s. semivariegatus) go some way up the somewhat sloping trunk of a gum tree and later make quite a good jump to a lower stump or branch." The Bush-Snake has the same build and strongly keeled ventrals as the "Flying Snakes" {Chrysopelea) of Southeast Asia and it may be that it shares their ability to glide from tree to tree. Loveridge received reports of green snakes which behaved in his manner while he was collecting in Tanganyika, where P.s. semivariegatus is common. Habitat. Most plentiful along shady streams w^ith plenty of trees and bushes. This species is less dependent on water than the other species in the genus, feeding as it does mainly on geckos. I was called to remove a young Essexvale snake from the engine of a lorry, while another is recorded as having been found on the steps of the Bulawayo Public Library ! Distribution. Connnon throughout Southern Rhodesia, but most abundant in the eastern districts. 1 One on both sides of a Bembesi Snake (NM/>r. 2060). BROADLEY : SNAKES OF SOUTTIERN RHODESIA 27 Localities: 8iiioia; Eldorado; Mazoe; Salisbury; Monte Cas- sino ; Kondo ; Odzi ; Umtali ; Nyamaropa ; Zambezi-Sebimgwc Junction ; Wankie ; Fatima ; Moliem Mine ; Que Que ; Selukwe ; Bulawayo ; Khanii Dam ; Essexvale ; Irisvale ; Empandene ; Uni- sliagashe River; Devuli River Bridge; Beitbridge ; Umvuma ; Bembesi. PrOSYMNA SUNDEVALLli SUNDEVALLll (A. Smith) Plate 3, upper figure Temnorhynclius .'inall3^ described by (liin- tlier from a specimen collected on the Zambezi by Sir John Kirk. F. W. FitzSimons included Southern Rhodesia within its range in his "Snakes of South Africa." As Chamaetortus is known from Mozambique and N. E. Transvaal it may well occur in the eastern districts of Southern Rhodesia. East African specimens are usually associated with bamboos or palms. DisPHOLiDus TYPus (A. Smith) Bucephalus tijpus A. Smith, 1829, Zool. Journ., vol. 4, p. 441. Bispholidus typus Boulenger, 1902, p. 18; 1910, p. 515; Gough, 1908, p. 33; Chubb, 1909a, p. 596; 1909b, p. 36; FitzSimons, F. W., 1912, p. 127; FitzSimons, V. F., 1939, p. 23; Tasman, 1953, p. 29; Eose, 1955, pp. 105-107; Isemonger, 1955, p. 73. Native Name of Boomslang. N'dlondlo (Sindebele) ; Coracimdu (Cheshona). Variation. (87 specimens.) Midbody scale rows 19, rarely 17 (one Salisbury snake) or 21 (3 specimens); ventrals 171-196; anal divided ; subcaudals 104-130 ; upper labials 7, rarely 8, the third and fourth (rarely fourth and fifth or fourth only) enter- ing the orbit; lower labials 8-12, the first four (rarely three or five) in contact with the anterior sublinguals; preocular 1; post- oculars 3, rarely 4; temporals 1+2, rarely l-fl or 2+2. Tail length .25 to .30 of the total. Colouration. Juveniles: Head uniform brown or blackish above, upper labials and chin white, uniform or more often speckled with black. Body blackish above, with numerous small blue spots, situated at the tips of dorsal scales and arranged in ])airs vertically. These spots normally only appear as "warning colouration" when the body is inflated in anger. Sides of body greyish, passing to white below, very heavily peppered with dark maroon or grey to resemble a lichen covered branch. Iris of eye brilliant emerald green. Adult males : TTsually bright leaf green above, ^\^th black- edged scales and ventral scutes. A Fatima specimen had the head vermiculated with black. Some specimens from the eastern 36 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY districts are uniform leaf green or blue-green. Several males from the Essexvale area are dark olive-green above and pale blue below. A single male from Essexvale was olive-brown. Iris of eye grey. Adult females: Usually blackish-brown to olive, pale "biscuit" brown or light grey above. Below, dirty white to pale olive. Iris of eye grey or brown. Size. Largest $ (UM/R.21) 1750 (1290+460) mm. from Umtali. Largest 9 (NM/M.413) 1625 (1210+415) mm. from Essexvale. Breeding. A pair of captive Boomslangs were found in coitu on 29.i.56. Diet. The full stomachs examined all contained chamaeleons {Chamaeleo d. dilepis) except two, which contained three fledg- lings. A black 9 Boomslang, just over five feet in length, lived for 21 months in captivity. In that time she consumed 25 chamaeleons, 3 agamas {Agama h. distanti and Agania Mrkii fitzsimonsi), 1 dead snake (Psammophylax t. tritaeniatus) , 6 fledglings. Only one snake, a $ captured swimming in a small stream at Odzi, would take frogs {Rana spp.). A dark olive S from Balla Balla took dead rats readily, but rodents were ignored by eight other Boomslangs in the cage. Birds' eggs were taken readily when offered. Other lizards taken by captive snakes were: Mahuya s. striata; Mahuya q. margaritifer and Agama cya-nogaster (juvenile). Defence. The Boomslang is normally good natured, but when roused to anger first the throat, then the whole body, is inflated with air, making the snake appear twice its normal size. If further molested, the snake strikes with gaping jaws almost in one plane, so although the fangs are situated below the eye, it is not difficult for an adult snake to bring them into action. The fangs are approximately i/4" in length, longer than in a cobra of the same size. Although this species is probably the common- est snake in Southern Rhodesia, bites are extremely rare. The Boomslang dashes to the top of the nearest tree at the least sign of danger and is seldom encountered at close quarters except by the herpetologist. Venom. An adult snake normally takes about 20 minutes to kill a full-grown chamaeleon, longer than does a Vine Snake BROADLEY : SNAKES OP SOUTHERN RHODESIA 37 {Thclotornis) . The venom may be more effective against warm- blooded prey, as fledglings die quickly and the effects on a human are very severe if not fatal. Habits. Mainly arboreal, often staying in the same tree, or group of trees, for several days. I have taken several along stream banks, where they were presumably searching for frogs. Large numbers are killed on the roads, which indicates the abundance of the species. Distribution. Abundant throughout Southern Rhodesia. Localities: Trelawney ; Mazoe ; Mount Hampden; Salisbury; Lake Mcllwaine ; Odzi ; Old Umtali ; Umtali ; Que Que ; Gatooma ; Selukwe ; Lukosi ; Fatima ; Turk Mine ; Bulawayo ; Empandene ; Essexvale ; Balla Balla ; Irisvale ; Stanmore ; Beitbridge ; Mtao Forest ; Dadaya ; Shabani ; Lundi River ; Nyaratedzi River ; Birchenough Bridge; Mount Silinda; Chirinda Forest; Mount Darwin. Thelotornis kirtlandh oatesh (Giinther) Dryophis Oatesii Giinther, 1881, in Gates, Matabeleland and the Victoria Falls, App., p. 330, col. pi. D: Matabeleland, Southern Khodesia. Thelotornis Mrtlandii (not Hallowell) Bouleuger, 1896, p. 185; 1910, p. 515; Chubb (part), 1909a, p. 596; 1909b, p. 36; FitzSimons, F. W. (part), 1912, p. 126; Hewitt and Power, 1913, p. 164; Isemonger (part), 1955, p. 78. Thelotornis lirtlandii capensis Loveridge (part), 1944, p. 154 (generic revision) ; Broadley (part), 1957c, p. 297. Thclotornis Icirtlandii oatesii Loveridge, 1953, pp. 277-279. Native name of Gates' Vine-Snake. Kotikoti (Sindebele). Variation. (11 specimens.) Midbody scale rows 19; ventrals 163-17-4 ; anal divided ; subcaudals 140-159 ; upper labials 8, the fourth and fifth entering the orbit ; lower labials 11-12, the first four in contact with the anterior sublinguals ; preocular 1 ; post- oculars 3, rarely 4; temporals 1+2, rarely l-|-3. Tail length .36 to .38 of total. Colouration. Top of head pale green speckled with pink and black, this speckling is normally restricted to a Y-shaped mark- ing, whose stem lies along the interparietal suture and its arms extend across the posterior portion of the frontal to the supra- oculars (in a Matopos specimen the speckling extends along the frontal to the internasals). A band of pink, black-edged, scales 38 BULLETIN : MUSEUM OF COMPARATH^E ZOOLOGY runs from the nostril through the orbit and lower temporals to the back of the head. Upper labials are white with a black streak extending from the eye to the full width of the sixth labial ; there is usually a row of black spots on the lii:>s. Chin white, the lower labials heavily speckled with black. Body light grey with diagonal cross bars of whitish blotches ; the sides are adorned with scattered touches of pink or orange and black. A'entrum pinkish-white heavily mottled with dark grey. On the sides of the neck are one or two vivid black blotches. Size. Largest (NM/M.972) 1417 (900+517) mm. from Ga- tooma. Diet. A Matopos snake contained an immature Plated Rock- Lizard {Gerrhosaurus v. validus). The large Gatooma snake recorded above lived for 6 months in captivity before being killed by a large Boomslang {Disi)holidus typus) in a dispute over a chamaeleon. In that time she consumed 11 Chamaeleo d. dilepis, an Ayatna h. distanti and two bird's eggs. Defence. When molested the Vine-Snake inflates its throat enormously to display the vivid black and white markings. If further tormented it strikes viciously with gaping jaws. Venom. See under Thelotornis k. capensis. Distribution. Northwestern parts of Southern Rhodesia, ex- tending south to a line Matopos-Bulawayo-Gatooma-Norton which more or less follows the principal watershed. Localities: Trelawney; Donnington Farm. Norton; Gatooma; BulaM-ayo ; Khami Dam ; Matopos ; Karoi. Thelotornis kirtlandii capensis A. Smith Thelotovnis c-apensis A. Smith, 1849, 111. Zool. S. Africa, 3, App., p. 19. Thelotornis kirtlandii (not Ilallowell) Chubb (part), 19(l9a, p. 596; 19091), p. 36; ritzSinious, F. W. (part), 1912, p. 126; FitzSimons, V. F., 1939, p. 23; Tasman, 1953, p. 29; Isemoiigor (part), 1955, p. 78. Thelotornis hirtlandii capensis Loverid^e (part), 1944, p. 154 (yenerif revision) ; Broadley (part), 1957e, p. 297. Thcletornis (sic) hirtlnndii, Eose, 1955, pp. 114-119. Native name of Gape Vine-Snake. Kotikoti (Sindebele) ; Kuni- t umuti { Cheshona ) . Variation. (38 specimens.) Midbody scale rows 19; ventrals 146-163 ; anal divided ; subcaudals 127-166 ; upper labials 8, BROADLEY : SNAKES OE SOUTHEKN RHODESIA 39 rarely 9, the fourth and fifth entering the orbit; lower labials 11-12, the first four in contact with the anterior sublinguals ; pre- ocular 1 ; postoculars 3, rarely 2 ; temporals 1-1-2, rarely 1+3. Tail length .33 to .40 of total.' Colouration. As in Thelotornis k. oatesii, except that in speci- mens from the south (Balla Balla-Lumane) the speckling extends over the whole of the top of the head. However, specimens from the Eastern Districts have either uniform green heads or the markings are reduced to a few spots arranged in the Y-shape typical of oatesii. The pink and black band on the side of the head is replaced by a uniform dark brown streak. Tanganyika specimens are similar. Because of this confusion in head mark- ings the two forms can only be distinguished by their ventral counts 1. Size. Largest 1440 (910+530) mm. from Umtali. Smallest (NM/M.613) 637 (410+227) mm. from Selukwe. Diet. An Irisvale snake was swallowing a young Tree Agama {Agama cyanogaster) when captured. Captive specimens fed readily on lizards {Mabuya s. striata; Mahuya q. margaritifer; Agama h. distanii: Agama. h. armata; Agama cyayiogaster ; Platy- saurus g. rhodesianus) , chamaeleons {CJiamaeleo d. dilepis) and frogs {Rana spp.). Although birds and their eggs were occa- sionally taken, cold-blooded prey seems to be preferred. Defence. See under Thelotornis k. oatesii. Venom. On 1.x. 57 I spotted a pair of Vine-Snakes mating in a tree at Lumane. As 1 tried to get both snakes into the bag at the same time, while perched at the top of the tree, the larger snake fastened on to the middle finger of my right hand. I de- scended to the ground and had some difficulty in disengaging the snake's fangs. The time was 3.30 p.m. I sucked the bite and then went after the second snake which I had been obliged to release. Although I failed to dislodge it from a thick bush I finally captured it two days later in the original tree. The finger was slightly swollen after half an hour and there was some slight haemorrhage from the fang punctures by 5 p.m. By 9 o'clock the finger was very .swollen and discoloured at the joint. There was persistent haemorrhage from the fang punctures and teeth 1 In 8 specimens of caprnsis from Zululand, which may be considered topotypes, the head speckling is confined to the Y-shape characteristic of oatesii'. 40 BULLETIN : MUSEUM OF COMPARATR^ ZOOLOGY marks, also from numerous scratches on my legs (received while climbing- the thorn tree after the snakes) and small shaving cuts. There was no pain whatsoever. The haemorrhage continued all night and all the next day. The blood was very slow to clot and I left pools of blood wherever I went. There were purple patches round all cuts, etc. By 9 p.m. on the 2nd the haemorrhage w^as easing off and confined to the scratches on my legs. The bleeding had stopped altogether by the next morning although the finger ]'? • SINOIA ■ 0 g •SALtSBoR.y ■ 'fiATOOftAA • S*S luMTAUl •Qwtuo • • V \ BULAWAYO . FORT ; ^ • YicroRjA. ^ J ••••• Fig. 3. Kecorded localities for Thelotornis. ■ Thelotornis Tcirtlandii oatesii (Giinther) • Thelotornis Icirtlandii capensis A. Smith was still swollen and the hand puffy. There was slight bleeding from the fang punctures about 7 p.m. There was no haemorrhage on the 4th and the hand started to return to normal the follow- ing day. I think that the numerous cuts and scratches on my body acted as safety valves and prevented the dreadful internal haemorrhage BROADLEY : SNAKES OF SOUTHERN RHODESIA 41 which was a prominent feature when F. J. de R. Lock died from a Thclotornk bite in Tanganyika i. The snake concerned in the latter case was a juvenile 2'5" in length ; 1 was bitten by an adult of just over 4 feet. Ilahits. This species is abundant in the dry Mopani bush at Lumane where there is very little undergrowth. It is also plenti- ful at Irisvale and in the Eastern Districts, where the vegetation is more varied and provides better cover. The snakes are usually found in bushes or on dead tree stumps not far from the ground where they can spot any lizards or frogs passing below. They remain motionless even when passed within a few inches. Distrihution. Southeastern districts of Southern Rhodesia, ex- tending as far north as a line Balla Balla-Selukwe-Salisbury. Localities: Salisbury District ; Odzani ; Odzi ; Umtali ; Selukwe ; Balla Balla ; Irisvale ; Sinkukwe ; Lumane ; Mount Silinda ; Pungwe River, 2400'. The specimens recorded from Empandene by Chubb (190i)b) probably belong to this race, but are now miss- ing from the National Museum collection. Hemirhagerrhis nototaenia nototaenia (Glinther) Coronella nototaenia Giinther, 1864, Proc. Zool. Soc. London, p. 309, pi. xxvi, fig. 1. Amplorhinus nototaenia Hewitt, 1913, p. 481. HemirliagerrJiis nototaenia Isemonger, 195.3, p. 7.1. Variaiion. (6 specimens.) Midbody scale rows 17; ventrals 164-168 ; anal divided ; subcaudals 72-83 ; upper labials 7 or 8, the third and fourth or fourth and fifth entering the orbit ; lower labials 9, the first four in contact with the anterior sublinguals; preocular 1; postoculars 2; temporals 1+2, rarely 1+3. Tail length .23 to .27 of the total. Colouration. Dark ash grey or grey-brown above; top of head black, continuing as a vertebral stripe about three scales in Avidth, which is black on the neck, but less well defined on the rest of the body. A row of black spots merges with the vertebral stripe on either side. These may be opposed to form cross-bars or alternated to form a zigzag. A dark streak on either side of the liead passes through the eye and fades out on the neck. Below, mottled in ash grey or grey-brown and dirty white. 1 See Loveridpre (1956). pp. 12-1."?. 42 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Size. Largest (UM/R.332) 870 (280+90) mm. from Mount Darwin. Habitat. A Bark-Snake was taken at 11 a.m. as it was crawling on the ground under Mopani trees with no undergrowth. This was in the Wedza Reserve, between the Maeheke and Sabi rivers (W. Armitage). Distribution. Found in the low-lying river valleys of Southern Rhodesia. The species seems to be closely associated with dry Mopani bush. Localities: Zambezi River, 40 miles east of Chirundu; Matetsi; Macheke-Sabi Junction, Wedza Reserve ; Devon Farm, Odzi River ; Ramaquabane River ; Beitbridge ; Mount Darwin. Amplorhinu8 multimaculatus a. Smith Aviplorhinus nmltimactdatvs A. Smith, 1847, 111. Zool. S. Africa, Eept. pi. Ixii. FitzSimons, V. F., 1958, p. 209. Variation. (2 specimens.) Midbody scale rows 17; ventrals 140-141 ; anal entire ; subcaudals 58-75, the anterior five single, the remainder paired ; upper labials 8, the fourth and fifth enter- ing the orbit ; five lower labials in contact with the anterior sub- linguals ; preoeular 1 ; postoculars 2 ; temporals 2-[-2. Tail length .22 of total. Colouration. Dark green to olive green above, with a paler dorsolateral and a longitudinal series of elongate black spots on either side ; scattered scales narrowly edged wdth bluish white, especially over anterior half of body ; upper labials each bearing a yellowish-white spot or irregular vertical streak. Chin yellow to yellowish-white, scales edged with bluish grey; underside of body and tail uniform bluish grey (V. F. FitzSimons). Size. Largest (T.M.22407) 488 (383+105) mm. from Pungwe River Causeway. Distrihntion. Mountains on the eastern border of Southern Rhodesia. Localities: Nyamaziwa; Pungwe River Causeway. Psammophylax tritaeniatus tritaeniatus (Giinther) RhagerrJtis tritaeniatus Giinther, 1868, Ann. Mag. Nat. Hist., ser. 4, vol. 1, p. 423, pi. xix, fig. 8. Coronelln frifarnia, Giinther, 1881, in Gates, IMatabeleland, p. 329, i>l. C. BROADLEY : SNAKES OF SOTTTHERN RHODESIA 43 Trimerorhiiuts tritaeniatus, Bouleuger, 1896, p. 139; 1902, p. 18; 1910, p. 512; Chubb, 1909a, p. 596; 1909b, p. 35; FitzSimons, F. W., 1912, p. 121; Hewitt and Power, 1913, p. 163; Tasman, 1953, p. 33; Rose, 1955, p. Ill; Iscniongei", 1955, p. 76. rs(i))iinopln/la.v tritiuiiiatu,s, Broadlcy, 1956, p. L'15. Native name of Three-lined Grass-Snake. N'shwazi (Sinde- hele), but also applied to Psmnmophis s. suhtaeniatus. Variation. (56 specimens.) Midbody scale rows 17; ventrals 150-168; anal divided; subcaudals 54-67; upper labials 8, the fourth and fifth entering the orbit; lower labials 9-11, the first five, rarely four or six, in contact with the anterior sublinguals; preocular 1; postoculars 2; temporals 2-[-3, rarely 2+2 1. Tail length .19 to .22 of the total. Colouration,. Top of head light brown ; vertebral scale row dark l)rown, the superior halves of the scales flanking it are black, forming a sharp-edged vertebral stripe 2 scales wide ; this is flanked by a pale brown, grey or yellowish stripe 3 scales wide followed by another dark brown, black-edged stripe 3 scales wide, which begins at the snout and runs through the eye ; outer li/> scale rows white, with a broken orange or pinkish line run- ning through the outer row. Upper labials, chin and throat white ; underside white, cream or lemon yellow, with some salmon or pink flecking at the ends of the ventrals. Size. Largest (SM/R.70) 851 (680+171) mm. from Salisbury. Smallest 172 (140+32) mm. from Essexvale. Breeding. A captive 9 from West Nicholson, 728 mm. in length, laid 4 eggs between 27th and 30th November, when she died with 10 eggs still in her ovaries. Diet. The huge Salisbury specimen, recorded above, contained a partially digested rat. Captive specimens took mice (Rhah- (lomys and Leggada sp.) ; lizards (Chamaeleo d. dilepis juv. ; Mahuya s. striata; Mahuya v. varia ; Mabuya q. margaritifer : Agama h. distanti), and frogs (Kassina .■. 16; Boulenger, 1902, p. 18; 1910, p. 516; Gough, 1908, p. 33; Fitz- Simons, F. W., 1912, p. 3 28; FitzSimons, V. F., 1935, p. 323; 1939, p. 24; Tasman, 1953, p. 33; Eose, 1955, p. 119. BROADLEY : SNAKES OF SOUTHERN RHODESIA 57 Aparallactus capensis capensis Loveridge (part), 1944, p. 205 (generic revision) ; Witte and Laurent, 1947, p. 122 (generic revision) ; Fitz- Simons, V. F., 1958, p. 210. Variation. (32 specimens.) Midbody scale rows 15; ventrals 137-170; anal entire; subcaudals 30-51 ; upper labials 6, the third and fourth entering the orbit, the fifth largest and in contact with the parietal ; lower labials 5-6, the first three in contact with the anterior sublinguals, the first pair not in contact behind the f Fig. 5. •X- o • •^AToolKVA ■O o • • • / SWELXJ i ^evuwAYo Recorded localities for ChilorhinopMs and Aparallactus. Chilorhinophis gerardi gerardi (Bonlenger) Aparallactus lunulatiis lunulatus (Peters) Aparallactus guentJieri Boulenger Aparallactus capensis capensis A. Smith Aparallactus capensis capensis X bocagei mental ; preocular 1, in contact with the nasal ; postocular 1 ; temporals 0+1+1. Tail length .13 to .20 of total. Colouration. Top of head and neck black, descending on the sides of the neck to form a half ' ' collar ' ' ; sometimes a pair of 58 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY light elongate spots extending back from the ends of the parie- tals; sides of head black from snout to anterior edge of the fifth labial. Bod,y light grey-brown to bright red-brown, uniform, or with a narrow darker vertebral line, or with five evenly spaced narrow dark lines. Below, uniform white. Size. Largest (NM/M.1190) 331 (268+63) mm. from Odzi. Diet. A three-inch centipede recovered from the stomach of a Gatooma snake. Enemies. The tail of a Cape Centipede-eater was disgorged by a Burrowing Adder {Atractaspis hihronii intermediate) taken at the Mchingwe River, Belingwe. Habits. Taken under stones in many different types of coun- try. Distribution. Common throughout Mashonaland and the East- ern Districts. I regard Matabeleland specimens as intermediates between the typical form and the race hocagei described from Angola (vide infra). Localities: Mazoe ; Trelawney ; Mount Hampden; Salisbury; Ilunyani ; Domboshawa ; Chishawasha ; Kondo ; Odzi ; Umtali ; Imbeza ; Nyamaziwa ; Vunil^a Mtn. ; Gatooma ; Gwelo ; Mount Silinda. ApARALLACTUS CAPENSIS CAPENSIS X BOCAGEI Apaiallactus capensis Chubb, 1909a, p. ."396; 19091), p. 3(). Aparallactus cape7isis capensis Loveridge (part), 1944, p. 20.1 (gt'iioiio ro- vision). Variation. (17 specimens.) Midbody scale rows 15; ventrals 156-181 ; anal entire ; subcaudals 44-63. Lepidosis otherwise as in the typical form. Tail length .17 to .22 of total. Colouration. As in the typical form except that no specimens have the bright red-brown colouring found in some Mount Hamp- den snakes. Most specimens have the pair of light spots behind the parietals. Size. Largest (NM/M.495) 348 (285+63) mm. from Tuli Hill. Diet. Captive specimens took small centipedes. Distrib ution. Matabeleland. Localities: Victoria Falls; Fatima: Gwaai ; Bulawayo; Mato- ])os Dam ; Bambata Cave, Matopos ; Tuli Hill ; Plumtree ; Essex- vale ; Balla Balla ; Irisvale ; Lumane ; Syringa. BROADLEY : SNAKES OF SOUTflERN RHODESIA 59 Discussion. Analysis of the data for 49 specimens of Aparal- lacfus capoisis from Southern IMiodesia shows a (h'finite increase in ventral and subeaudal counts fi'oni east to west. The lowest ventral counts are 137 and 138 for two Vumba Mountain snakes. At the other extreme are two snakes from the Matopos with 178 ventrals. The Matabeleland snakes, while not approaching the high ventral count of bocagei (175-191), do not fall within the accepted range of the typical form, and are best regarded as intermediates. The average counts for the material examined are: A. c. capensis . . . ventrals 157, subcaudals 43. ^1. c. capensis X hocagei . . . ventrals 167, subcaudals 50. DASYPELTINAE Dasypeltis scabra (Linne) Cohibcr scaler Linne, 1758, Svst. Nat., ed. 10, 1, ]\ 223. Dasypeltis scabra Boulenger, 1894, p. 354; 1902, r- 17; 1910, p. 509; Chubb, 1909a, p. 595; 1909b, p. 35; FitzRimons, F. W., 1912, pp. 90-91; Tas- man, 1953, p. 35; Eose, 1955, pp. 98-104; Isemonger, 1955, p. 72. Variation. (27 specimens.) Midbody scale rows 21-27; ven- trals 192-248; anal entire; subcaudals 45-70; upper labials 7, rarely 6, the third and fourth entering the orbit ; no loreal ; pre- ocular 1 (2 on both sides of a Zambezi snake and one side of a Bulawayo snake) ; po.stoculars 2; temporals 2+3; 2+4 or 3+4. Tail length .11 to .17 of the total. Colouration. Two phases occur. The commonest is the rhombic phase : Light brown or greyish above with a dorsal row of dar)\ elongated blotches and a lateral series of dark vertical bars. A broad V-shaped mark on the neck is usually preceded by a nar- i-ower V on the head (sometimes two). A Fatima snake has a double row of coalescing dorsal blotches. Some specimens from Mashonaland and the Eastern Districts are uniform red -brown; this phase seems to predominate around Salisbury. Ventrum white, usually with some brown flecking at the ends of the ventrals. Size. Largest (UM/R.6) 724 (630+94) nun. from Umtali. Diet. Captive specimens would take only birds' eggs. Hatch- lings consistently refused fresh gecko eggs, wliich seem quite suitable fare. 60 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Distrihuiion. Found throughout Southern Rhodesia, but scarce. Localities: Zambezi River, 40 miles east of Chirundu; Mazoe; Salisbury ; Chishawasha ; Musami ; Odzani Falls ; Imbeza ; Odzi ; Umtali ; Tandaai ; Fatima ; 25 miles north of Bulawayo ; Bula- wayo ; Springvale ; 9 miles south of Bulawayo ; Mount Silinda ; Mount Darwin; Bembesi. ELAPIDAE AspiDELAPs scuTATus scuTATUS (A. Smith) Plate 4, lower figure Cyrtophis soutatus A. Smith, 1849, 111. Zool. S. Africa, Rept., App. p. 22. Axpidelaiis scutatus Chubb, 1909a, p. 597; Boulenger, 1910, p. ril9; Fitz- Simons, F. W., 1912, pp. 165-166; Tasmaii, 10.13, p. 24; Isemonger, 1955, p. 85. Aspidelaps scutata Chubb, 1909b, p. 35. Variation. (9 specimens.) Midbody scale rows 21; ventrals 113-123 ; anal entire ; subcaudals 23-31 ; upper labials 6, the fourth entering the orbit ; lower labials 7, the first three or four in contact with the anterior sublinguals ; preocular 1 ; postoculars 2-3 ; temporals 2+4, 2+5 or 2+6, the lower anterior temporal very large, lying between the fifth and sixth labials (in a Kezi snake this shield reaches the lip, excluding the fourth labial). Tail length .14 to .17 of the total. Colouration. Head black, chin and throat white, with black intriLsions at the angle of the jaw ; a broad black band, approxi- mately 12 ventrals wide, encircling the neck. Body bright orange flecked with brown, a dorsal series of brown blotches. Ventrum white. Size. Largest 6 ; Hewitt ninl Power, 1913, p. 164. .\aja hale Eose, 1955, p. 132. Native rtamcs of Egyptian (Johra. Pimpi (Siiidebele), but properly applied to Naja n. mossanibica, sometimes confused with Dendroaspis p. polylepls and called Imaniha; Maliurc or Mungu (for the black phase) (Cheshona). The Banded Cobra (var. cmnulifera of Peters) is known as Lume in Sindeliele and Nyama- fingu in Cheshona. Variation. (79 specimens.) Mid body scale rows 19 (17 in three cobras from Chishawasha, Bulawayo and Essexvale; 18 in one Essexvale snake); ventrals 186-203; anal entire; subcaudals 53-66; upper labials 7, the sixth largest (rarely 6, the fifth largest), normally excluded from the orbit by the suboculars (entering the orbit in three Bulawayo snakes: (a) third labial on each side; (b) second on one side and third on the other; (c) third on one side and third and fourth on the other) ; preocular 1 ; suboculars 2-3 ; postoculars 2, rarely 3 ; temporals 1+2 or 1+3. Tail length .14 to .18 of the total. Colouration. Juveniles : Head brown, body dull yellow with a broad black band encircling the neck, ventriim bright yellow. Hatchlings belonging to the variety annulifera have barely dis- cernible light and dark yellow bands. The first yelloAV band is clearly visible against the black on the back of the hood. Adults: Head dark brown to black, body grey-brown (most Mashonaland snakes) to ashy black (most Matabeleland snakes). Below, yellow more or less mottled with brown, a broad (ca. 10 ventrals wide) purplish-brown band on the throat. In many Matabeleland cobras the belly gradually darkens from the tail towards the head, adults often becoming uniform lilaek al)ove and below with only the chin left yellow. Occasional specimens retain the lighter colouring of the juvenile and become an attractive orange-brown, with pink interstitial skin. In the variety aM7iulifera Peters the yellow livery of the juvenile is partially retained, while the rest of the body becomes even darker than usual. The normal colouring is : Head dark brown, body blue-black with from 7 to 11 bright yellow or creamy white cross bands, which are normally about half tlie width of the black interspaces. The belly is bright yellow, uniform, or more often blotched with black where the cross bands M'ould continue. BROADLEY : SNAKES OF SOITTIIERN RHODESIA 63 In some specimens the body is completely ringed in black and yellow. The first band, in the centre of the "hood," is very narrow and is often broadened in the centre Avith a black median spot, reminiscent of the hood marking of the "monocellate" variety of Naja naja. One Umtali snake had a series of yellow dorsal blotches instead of bands. Another Umtali snake had a .single yellow band just before the vent, the rest of the body being uniform black. Size. Largest S (NM/M.1378) 2285 (1905+380) mm. from Nyamandhlovu. Largest 9 (NM/M.393) 2238 (1900+338) mm. from 7 miles north of Bulawayo. Smallest (T.M.) 302 (250+52) mm. from Amandas. This species appears to reach a length in excess of 10 feet. Discussion. Naja haje var. annulifera Peters is represented by 26 of the 79 specimens for which data is available. The only difference in scale counts is a slightly higher average for sub- caudals in annulifera. All the specimens of annulifera that I have sexed have been males, but some typical haje are also males, so there is no clear-cut sexual distinction. Unlike most species, the males seem to grow at least as large as the females. The record Nyamandhlovu specimen is an annulifera, as are two other l)ig males of 2118 and 1943 mm. This handsome variety has not been recorded north of the Zambezi, but is known from Mozam- bique (Tete, type locality), Transvaal and Bechuanaland. Diet. The largest 9 contained a 9 Bitis a. ariefans 2'5" in length, which in turn contained 19 fully formed young. x\ 61/^-foot annulifera from Redbank had also swallowed an adult puffadder. A 5-foot cobra, captured on the Umzingwane River at Essexvale, disgorged five toads {Bufo regularis). Stomachs examined usually contained toads or amphibian remains. Cobra are persistent raiders of poultry runs. A 6-foot cobra killed at Irisvale contained two well-grown chickens. The Rev. K. Tas- man, S.J. records {in lilf.) a <'ol)ra of 7'1" (killed in a poultry run) which contained 12 eggs. Another snake contained five ducklings. Tasman reports that 7 out of 17 stomachs wath recog- nisable prey contained warm-l)looded animals (rats and duck- lings). Captive specimens took toads {Bufo regularis; Bufo carens) ; frogs {Rana spp.) ; chamaeleons {Chamaeleo d. dilepis) s 64 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY and snakes {Boaedon f. fuliginosus; Psammophis s. siMlans; Naja n. mossambica ; Causus defilippii) ; one took a rat. Parasites. Many specimens harbour ticks {Aponomma latum). The largest male was full of nematodes. Enemies. While collecting along the N'sese River on the edge of the Matopos, I disturbed a fine Martial Eagle (Polemaetus bellicosus), which rose with a dead six-foot cobra in its talons. Defence. Normally a cobra tries to escape when disturbed, but if taken by surprise or cornered, it rears and spreads a broad "hood," but does not strike unless molested. If left alone, it soon drops to the ground and tries to escape. Occasional speci- mens will attack if provoked ; twice, cobras have turned and come straight at me when I attempted to catch them ; both were annulifera. Some specimens sham death after capture. The first cobra I ever captured was a 4'9" Naja h. haje which "played possum" very convincingly. I measured the "corpse," took all the scale counts, removed numerous ticks, examined the fangs and washed some sand out of the mouth. The cobra seemed quite lifeless, which puzzled me as I had not been rough while catching it. While I was getting out my skinning knives, the "corpse" came to life and started to glide across the floor ! Venom.. The glands of an adult cobra contain a large quantity of the powerful neurotoxic venom. The only bite received personally was from a two-foot juvenile, which quietly started to chew my finger while I was handling it. I ligatured the finger at the base, cut and sucked the punctures and had no symptoms of poisoning whatsoever. Habits. Cobras do most of their hunting at night, but may often be found basking near their holes during the day, partic- ularly in the early morning. The usual lair is a disused termita- i-ium, but rat holes and mole runs are sometimes used. I have no records of this species taking to the water or climbing trees. Habitat. This species does not share the Spitting Cobra's pref- erence for waterside localities, but I found it abundant near Mount Hampden in an extensive vlei which is inundated during the rains. The cobras lived in the numerous large termitaria, the only dry spots. Distribution. Common throughout Southern Rhodesia. Localities: Trelawney; Horseshoe Block; Mazoe ; Amandas; BROADLEY : SNAKES OF SOUTHERN RHODESIA 65 Mount Hampden ; Salisbury ; Chishawasha ; Marandellas ; Monte Cassino; Odzi; Umtali; Vumba Mountain; Zambezi-Sebungwe Junction; Gwelo; Nyamandhlovu ; Redbank; Bulawayo; Mato- pos ; Plumtree Balla ; Irisvale shagashe River Springvale; Essexvale; N'eema Dam; Balla Stanraore; Tod's Hotel, West Nicholson; Um- Bubye River ; Umvuma ; Cyrene ; Figtree. Naja haje ancheetae Bocage Naja anchieiae Boeage, 1880, Jour. Sci. Lisboa, vol. vii, pp. 89, 98. Discussion. This race is distinguished from the typical form by having only 17 midbody scale rows. Two cobras from Chi- shawasha and Bulawayo have 17 scale rows, as does an annulifera from Essexvale, but as they occur in the midst of a population of typical Jia^e with 19 rows I regard them as aberrant specimens. As Naja haje anchieiae has been recorded from Livingstone in Northern Rhodesia and Kabulabula in N.E. Bechuanaland, both on the Southern Rhodesian border, this race probably occurs in the northwest corner of the colony. Naja nigricollis mossambica Peters Plate 6, upper figure Naja mossambica Peters, 1854, Monatsb. Akad. Wiss. Berlin, p. 625. Naianigricollis (not Reinhardt) Boulenger, 1902, p. 18; 1910, p. 518; Chubb, 1909a, p. 597; 1909b, p. 36; FitzSimons, F. W., 1912, pp. 164, 165; Hewitt and Power, 1913, p. 65; Tasman, 1953, p. 22. Naja nigricoUis (not Eeinhardt) Eose, 1955, pp. 128-132; Isemonger, 1955, p. 84. Native Name of the Mozambique Spitting Cobra. Pimpi (Sin- debele), correctly applied to this species, but often applied to other brown, grey or blackish snakes. Variation. (79 specimens.) Midbody scale rows 21-25 (68 snakes have 23) ; ventrals 182-203; anal entire; subcaudals 54-70; upper labials 6-7, the third, rarely the fourth, entering the orbit ; preoculars 2; postoculars 3; temporals 2+4, 2-{-5, 3+4, 3+5, 3+6 or 3+7. Tail length .15 to .19 of the total. Colouration. Above, head light brown, l)ody light grey to grey- brown, scales black tipped. Below, salmon-pink to yellowish, with an irregular series of black cross-bands and blotches on the throat. 66 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Size. Largest (NM/M.973) 1543 (1285+258) mm. from Essex- vale. Smallest (NM/M.765) 292 (240+52) mm. from Chirinda Forest. This is a smaller species than Naja h. Jiaje, averaging less than 4 feet in length. Discussion. The species of the genus Naja are sorely in need of revision, none more so than Naja nigricollis. Two well-defined races occur in Rhodesia. Naja nigricollis crawshayi Giinther ranges through Northern Rhodesia as far south as Lusaka and Fort Jameson. Naja nigricollis mossamhica Peters occurs in the south of Northern Rhodesia, Nyasaland and S.W. Tanganyika, ranging south through Southern Rhodesia and Mozambique into the Union of South Africa. These races may be separated as follows : Midbodj^ scale rows 17-21 (usually 19) ; above, dark brown to black ; below, yellowish to grey, with a single broad black band on the throat . . . Naja nigricollis crawshayi Giinther. Midbody scale rows 21-25 (usually 23) ; above, light grey or brown with black-tipped scales ; below, pinkish or yellowish white, with a series of black bands and blotches on the throat . . . Naja nigricollis mossamhica Peters. Diet. A cobra captured beside the N'sese River at Irisvale disgorged the posterior half of South-eastern Green Snake (Philothamnus hoplogasier) . An Imbeza snake disgorged a Slug- eater {Duherria I. rhodesiana). Numerous stomachs examined contained amphibians. This species also raids poultry runs, a 4-foot Irisvale cobra contained four small chickens. Captive specimens took toads {Bufo regularis: Bufo carens) ; frogs {Rana spp.) ; dead snakes. This species is truly cannibalistic, for a 4-foot cobra swallowed a 12" juvenile when it was placed in the same cage. Parasites. Most specimens carry a few ticks, but an aestivating cobra, killed when a culvert was demolished near Bulawayo, yielded 21 adults and 35 larvae of Aponomma latum; many more escaped. Defence. Although it invariably tries to escape from man if possible, this cobra needs little provocation to malve it start "spitting." The snake rears and spreads a lung narrow "hood," very different from that of Naja h. hajc. Occasionally a cobra will rear higher and higher until it is supported by little more BROADLEY : SNAKES OF SOUTHERN RHODESIA 67 than its tail. In these circumstances the snake will repeatedly overbalance, but a cobra with no more than a third of its length on the ground can balance itself beautifully, recoiling from a strike like a piece of sprung steel. When "spitting," the cobra draws the head back, opens the mouth, then, as the venom is forced down the fangs and through the bend at the tip, the head is thrown forward and a blast of air from the glottis assists the twin jets of venom to reach the target. The cobra aims for the eyes and is very accurate. A four-foot cobra has a range of from six to eight feet; although drops of venom travel farther than this, the range is too great for them to reach eye level. A cobra can continue to spit almost indefinitely ; I have never managed to exhaust the supply of venom except by persuading the snake to discharge its venom repeatedly for several days. Even then the supply of venom is replenished in a day or two. This species seems to rely primarily on blinding an aggressor with venom and rarely tries to bite in the normal way. Habits. This cobra is very plentiful along streams in Mata- beleland. It often takes to the water when disturbed, swimming strongly on the surface with head down. Although often found living in termitaria, this is the only snake that I regularly find in rock crevices. Whenever a stream floAvs past fissured rocks, there will be found Naja n. mossamhica, or at least the tell-tale sloughs. A well-populated fissure, in a granite outcrop at Iris- vale, contained two four-foot Spitting Cobras, a dozen geckos {P achy dactyl us hibronii), a few other lizards and a big scorpion {Opistopthalmiis) ! This species sometimes climbs trees. One evening, while in camp at Beitbridge, I found a Spitting Cobra climbing the tree that I was reclining against ! I have also taken a juvenile on a branch ten feet from the ground at Irisvale. Distributio7i. Abundant in south Matabeleland, where it is probably the commonest snake. The species is less common else- where in the colony and is extremely scarce around Salisbury. Localities: Sinoia; Eldorado; Trelawney; Mazoe ; Bindura; Salisbury District ; Penhalonga ; Tmbeza ; Odzi ; Umtali ; Hunter's Road; Wankie ; Deka : Ntabezinduna; Nyamandhlovu : Bula- wayo; Syringa; Erai)andene; Essexvale ; Balla Balla ; P^ilabusi ; 1 risvale ; Sinkukwe ; Glass Block ; Mazeppa Mine, Gwanda ; Beit- bridge ; Umzilizwe River ; Chirinda Forest ; Mount Darwin ; Umvuma. 68 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Naja melanoleuca Hallowell Plate 6, lower figure Naja haje var. melanoleuca Hallowell, 1857, Proe. Acad. Nat. Sci. Phila- delphia, p. 61. Data of unique specimen. Midbody scale rows 19 ; ventrals 215 ; subcaudals 67 ; upper labials 7, the third and fourth enter- ing the orbit, the sixth largest and in contact with the postoculars ; preocular 1; postoculars 3; temporals 1+2 or l-|-3. Tail length .17 of the total. Colouration. Light grey -brown above, freely speckled with black, tail dark brown. Belly bright yellow, heavily spotted with black, no bands on the throat. Upper labials yellowish, with only faint traces of the back sutures characteristic of the species. Size. 1690 (1410+280) mm. from Mount Silinda. Diet. The only food taken while in captivity consisted of two dead rats and two Blind-snakes (TypJilops s. mucruso). Parasites. Numerous ticks found on this cobra were identified as Aponomma latum by Dr. G. Theiler. Defence. This is a much faster species than the other local cobras. When cornered it rears and spreads a long narroM' hood, similar to that of Naja n. mossanibica. It is a formidable reptile to capture. Temperament. In captivity, this specimen is much quieter and less nervous than the other cobras ; it never spreads a hood or strikes at the glass when I pass the cage, as specimens of Naja Ji. haje invariably do. Habitat. This specimen was taken as it emerged from a hole a yard from the edge of a strip of forest, where it borders mealie lands on the summit of Mount Silinda. On an earlier expedition 1 disturbed a huge cobra about eight feet in length, which was basking on a mat of floating grasses on the Umzilizwe River, below Mount Silinda. The cobra slid to the edge of the vegetation and dived to the bottom of a deep pool. Distribution. Liable to be found in suitable localities any- where along the Eastern Border of Southern Rhodesia. Localities: Umzilizwe River; Mount Silinda. BROADLEY : SNAKES OF SOUTHERN RHODESIA 69 Dendroaspis angusticeps (A. Smith) Naia angusticeps A. Smith (part), 1849, 111. Zool. S. Africa, Kept., pi. Ixx. Dendroaspis angusticeps Loveridge, 1950, p. 251. Data for unique specimen. Midbodj^ scale rows 17 (usually 19); ventrals 214; anal divided; subcaudals 120; upper labials 8-9, the fourth entering the orbit. Tail length .25 of the total. Size. $ (M.C.Z. 29182) 1702 (1275+427) mm. from Mount Silinda. Distribution. The Green Mamba may be found in any of the forested areas of the Eastern Districts. Localities: Mount Silinda. Dendroaspis polylepis polylepis (Giinther) Dendraspis polylepis Giinther, 1864, Proc. Zool. Soc. London, p. '310. Dendraspis angusticeps Boulenger (part) 1910, p. 520; FitzSimons, F. W. (part), 1912, pp. 169, 170. Dendroaspis angusticeps Tasman, 1953, p. 20. Dendroaspis polylepis Eose, 1955, pp. 143-147; Isemonger, 1955, p. 81. Native Name of the "Black" Mamba. Imamha (Sindebele) ; N 'zayo ( Cheshona ) . Variation. (24 specimens.) Midbody scale rows 21-25 ; ventrals 256-275; anal divided; subcaudals 115-131; upper labials 8-9, the fourth 1 entering the orbit; preoculars 3; postoculars 3-4 (1, through fusion, on one side of a Selukwe snake) ; temporals 2-[-3 (usual), 2+4 or 3+3. Fusion of head shields common, particu- larly the sixth labial and low^er anterior temporal. Tail length .20 to .22 of total. Colouration. Above, very dark olive-green when freshly sloughed, rapidly becoming dark brown, grey-brown or olive, sometimes mottled with blackish-brown towards the tail. Below, dirty white or greenish white, often with dark mottlings pos- teriorly. Size. Largest (NM/M.372) 2875 (2280+595) mm. from 20 miles north of Bulawayo. Detailed measurements are available for only 10 specimens, the rest consisting of skins and heads. The species reaches a length of 14 feet. Parasites. Ticks from an Odzi specimen were identified as 1 Fifth on one side of a Bubye River snake. 70 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Aponomma latum by Dr. G. Theiler. Most mambas have a fevp of these ticks on the neck. Defence. Normally the mamba receives ample warning of the approach of man and quietly glides into cover. If taken by sur- prise or cornered, the mamba usually attempts to intimidate the enemy. This was well demonstrated by a huge mamba I found in a gravel pit near Balla Balla. The snake was at least 12 feet in length and was basking between a large termitarium and a track. I cut off the mamba from its antliill and pushed a noose in front of it as it made for home. As it reached the noose, the mamba reared up to the level of my face, spread a broad ' ' hood ' ' and opened its mouth, displaying the black interior and formid- able fangs. AVith that terrible head only 18" from my nose, I lost much of my enthusiasm and recoiled. Thereupon the mamba dropped to the ground and streaked down a hole in the ter- mitarium ^ . Habitat. Mambas are often found on granite kopjes, where they live in rock crevices. Tn open thorn-buph they usually occupy disused termitaria. Dint ril}ui ion. Throughout Southern Rhodesia, excluding the highlands over 5,000 feet. Most plentiful in the low-lying river valleys. Localities: Inyazura ; Odzi ; Grand Reef ; Umtali ; Hot Springs ; Umvumvumvu River ; Hartley ; Que Que ; Wankie ; West Se- bungwe; Patima to Bulawayo ; Turk Mine; Heany; Selukwe ; Inyati ; Nyamandhlovu ; Plumtree ; Matopos ; Balla Balla : Beit- bridge; Shabani; Devuli River; Bubye River; Umvuma. YIPERIDAE Atractaspis bibronii bibronii a. Smith Atractaspis bibronii A. Smith, 1849, 111. Zool. S. Airit-a, Eent., p!. Ixxi. Variation. (7 specimens.) Midbody scale rows 21; ventrals 218-236; anal entire; subcaudals 19-26, single; upper labials 5. the third and fourth, rarely fourth only, entering the orbit ; lower labials 5, the first pair in contact behind the mental, the first three in contact with the anterior sublinguals, the third much the larg- 1 Enemies — a 2C<\-2" mamba was fliss:i>i'frf^rt by a Psnmmophis s. ftilt:in)>s rap tiii-ed at Uimuma by D. S. Rider. BROADLEY : SNAKES OF SOT^TnERN RHODESIA 71 (^st; preociilar 1 ; postocniar 1 ; temporals 1+2. Tail length .05 to .07 of the total. Colouration. Uniform purplish black above and below. Size. Largest (SM/R.52) 538 (510+28) mm. from Umtali. Venom. At 5 p.m. on 20.viii.55 I was bitten by a 318 mm. $ Burrowing Adder. I turned over a stone on an Umtali hillside and exposed the snake, which promptly dived down its hole. 1 seized the tail and pulled steadily, but the snake reversed sud- denly and bit me twice on the left index finger and once on the right before I got it into a bag. As the adder was only a juvenile 1 decided to let the venom run its course and observe the effects. After 15 minutes the left finger M-as slightly swollen and a feeling of depression was observed. By 9 p.m. the left hand was swollen hut not verj^ painful and the depression had gone. During the night the hand and wrist became excessively swol- len ; the index finger was twice its normal size and painful. The right index finger, having received a smaller amount of venom, was only slightly swollen, but the glands under lioth armpits were swollen. 1 was persuaded to go to the hospital, and was admitted at 7.30 a.m. 1 was given 10 cc. of Polyvalent Serum and also penicillin injections every six hours. The left hand was poulticed and this helped to bring the hand down to normal size after four days. A large blood blister formed at the fang punctures ; this was cut open and drained for several days. On the 28th August a rash appeared on my limbs and spread throughout my body before disappearing. 1 was discharged from the hospital the following day, but exercise brought on a severe recurrence of the serum reaction. This time, in addition to the rash, I suffered l)aralysis of the knees and fingers. 1 returned to the hospital, where calcium injections loosened the joints after 24 hours. 1 had violent recurrences ol" the rash for another three days. The dead flesh around the fang punctures sloughed away, leaving a pit 1/4 inch deep. I finally left the hospital on September 3rd. Habits. This fossorial species is usually found under stones ; a burrow often extends downwards for a foot or so, providing tlic snake with a more secure retreat. Distribution. The eastern districts of Southern Rhodesia. The rest of the colony is inhabited by a population of intermediates. Localities: Odzi ; Imbeza; Umtali. 72 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY AtRACTASPIS BIBRONI X ROSTRATA Atractaspis hihronii (not A. Smith) Chubb, 1909b, p. 36; Tasman, 1953, p. 27; Isemonger, 1955, p. 90. Yariation. (23 specimens.) Midbody scale rows 21-23 ; ventrals 217-252 ; anal entire ; subeaudals 20-27, mostly single ; upper labials 5, the third and fourth entering the orbit (4, the second and third entering the orbit, on one side of an Essex vale snake) ; lower labials 5, the first pair in contact behind the mental, the first three in contact with the anterior sublinguals, the third much the largest; preocular 1; postocular 1; temporals 1+2 (l-|-3 on one side of a Chishawasha snake). Tail length .05 to .07 of the total. Colouration. Uniform purplish-black above and below. Five snakes from Nyamandhlovu, Plumtree, Irisvale and Mchingwe River have the outer two roAvs of dorsal scales and the whole of the ventrum white. A Chishawasha snake has the sides and belly brownish with white blotches. Size. Largest (NM/M.720) 553 (524+29) mm. from Chisha- washa. Diet. A Bulawaj'o snake contained a Nucras intertexta holubi with a head and body length of 75 mm. A Plumtree snake dis- gorged another lizard of the same species. A small specimen taken near the Mchingwe River disgorged the tail and partially digested body of a Cape Centipede-eater {Aparallactus c. capen- sis). Venom. On 30.ix.55 I was bitten on the tip of the second finger of my left hand by the large Burrowing Adder from Chisha- washa, recorded above. As I pinned the snake down, it twisted round and struck with one fang. This species does not open its mouth to bite ; instead the lower jaw is contracted and the fangs brought down on each side of it. I ligatured the digit at the base and made some longitudinal cuts before squeezing out as much venom as possible. The finger became swollen during the night, with the characteristic throbbing pain. By the following day the pain had almost gone and I was fit enough to capture a ten- foot python. Habits. As for the typical form. One specimen was taken at night as it was crossing a road. BROADLEY: SNAKES OF SOUTHERN RHODESIA 73 Distribution. These snakes are intermediate between the typi- cal form of South Africa and the race rostrota found north of the Zambezi. Whereas typical bihronii normally has 21 midbody scale rows and the race rostrata 23 rows, throughout most of Southern Khodesia snakes with 21 and 23 scale rows occur side by side. Localities: AVest Sebungwe; Trelawney; Chishawasha; SaAV- mills ; Nyamandhlovu ; Bulawayo ; Plumtree ; Essexvale ; Balla Balla ; Irisvale ; Mchingwe River, Belingwe District ; Bembesi. Causus rhombeatus (Lichtenstein) Sepedon rhombeatus Lichtenstein, 1823, A'erz. Doubl. Mus. Berlin, p. 106. Causus rhombeatus Boulenger 1902, p. 18; 1910, p. 521; Gough, 1908, p. 38; Chubb, 1909b, p. 36; FitzSimons, F. W., 1912, pp. 215, 216; Hewitt and Power, 1913, p. 165; FitzSimons, Y. F., 1939, p. 24; Tasman, 1953, p. 27; Eose, 1955, pp. 166-167; Isemonger, 1955, p. 87; Broadley, 1957b, p. 115. Native names of Rhombic Night-adder. Changwa (Sindebele) ; Cheewa (Cheshona). Variation. (48 specimens.) Midbody scale rows 17-18, rarely 19 ; ventrals 138-151 ; anal entire ; subcaudals 24-32 ; upper labials 6, excluded from the orbit ; preoculars 1-3 ; suboculars 1-2 ; post- oculars 1-2 ; temporals 2+3. Tail length .09 to .12 of the total. Colouration. Above, pinkish, grey-brown or grey-green, with a darker forward -directed "V" on the head and a series of large, white or yellow-edged, dorsal rhombs. These markings may be very faint, often only visible when the body is inflated in anger. A Mount Hampden 6 was uniform pinkish brown without a trace of markings. A series of dark lines radiate from the orbit. Ventrum, mother of pearl white to pinkish grey or black. Size. Largest (UM/R.13) 813 (735+78) mm. from Odzani. Smallest (NM/M.806)' 224 (203+21) mm. from Essexvale. Breeding. A 578 mm. snake laid 14 eggs on 8.x. 57. Diet. Toads (Bufo rcgularis and Bnfo carens) form the greater i)art of the night -adder's diet. Frogs {Rana spp.) are also taken. Enemies. A Selukwe snake was killed by a chicken. Defence. The normal reaction to a disturbance is to coil up and inflate the body with air, at the same time hissing loudly. Occa- 74 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY sionally a specimen will flatten its neck after the manner of a cobra and make off slowly. Venom. W. Armitage of Umtali was bitten on the tip of his left index finger by an adult nightadder, which penetrated with only one fang. A burning pain was felt at the fang puncture. A ligature was applied and the bite cut open; it proved impossible to induce bleeding. After half an hour sporadic pains were ex- perienced as far as the wrist. Patient hot and flushed. The fol- lowing day found the victim depressed and suffering from a headache and sore throat. Another 24 hours brought pains in the kidneys and the next day saw the patient confined to bed with a fever which broke the same evening, 72 hours after the bite. Habits. This species mainly hunts its amphibian prey at night, although it may often be found basking during the day. A large uniform grey male, taken on Vumba Mountain, was engaged in swallowing a large Bufo regularis at 3 p.m. on a sunny day. Longevity. A Mount Hampden snake lived for 27 months in captivity and grew from 560 mm. to 720 mm. in total length. Distribution. Widely distributed throughout Southern Rho- desia. Common in Mashonaland and tlie Eastern Districts but rather scarce in Matabeleland. Localities: Trelawaiey; Mazoe ; Mount Hampden; Salisbury; llunyani; Prince Edward Dam; Chishawasha ; Marandellas; Odzi; Inyanga ; Odzani ; A'umba Mtn. ; Chirinda Forest; Ga- tooma : Selukwe ; Bulawayo ; Tuli Reservoir ; Hope Fountain ; jMatopos; Cyrene ; Empandene; Essexvale. Causus defilippii (Jan) Hetciodon Dc FiUppii Jan, 1862, Zool. Anat. FisLol., 2, [). 22."i. Cav.ms defilippii Boulenger, 1902, p. 18 ; 1910, p. 521 ; Chubb, 1909a, p. 597 ; 1909b, p. 36; FitzSimons, F. W., 1912, pp. 215, 216; FitzSimons, V. F., 1939, p. 24; Tasman, 1953, p. 27; Broadley, 1957b, p. ll;"i. CaiisuN dcphillippil {sic) Rose, 1955, p. 167. Causus drfillippii (sic) Isemonger, 1955, p. 88. Native name of Snouted Night-adder. Changiva (Sindebele), but generic. Variation. (53 specimens.) Midbody scale rows 17, rarely 16 or 18; ventrals 112-127: anal entire; subeaudals 10-19; upper lHl)ials 6-7, excluded from the orbit (3rd labial enters orbit on BROADLET : SNAKES OF SOUTHERN RHODESIA ( ) one side of a Chirinda Forest snake) ; preoeulars 2; siiboculars 1-2 (rarely 0 or 3) ; postoenlai's 1-2: temporals 2-\-'A. Tail lenj^th .05 to .09 of the total. Colouration. Above, pink, mauve or pale brown, with a broad, slightly darker, dorsal stripe and a vertebral series of large, dark brown, blotches or backward-directed chevrons. A well-defined " \'" on the back of the head has its apex on the frontal. A dark band passes through the orbit and the labial sutures are black. Below, glossy black in juveniles, l)ecoming lighter in adults. Size. Largest 410 (380-)-30) mm. from TTmtali. Smallest (NM/M.715) 102 (95+7) mm. from Bulawayo. Sexual dimorphism. In 24 males the range of ventrals is 112- 117 ; range of subeaudals is 14-19, and the tail length is .07 to .09 of the total. In 22 females the range of ventrals is 120-127 ; range of subeaudals is 10-16, and the tail length is .Oo to .07 of the total. Breeding. A captive 9 laid two eggs on l.i.56 and was seen to mate the following day. Many hatchJings are killed while basiling on the roads in February- and March. Diet. Captive specimens took small toads {Bufo reguJaris; Bufo carens) and frogs (liana dclalandii; Phrynohatrachus nata- lensis). A captive specimen at Umtali is reported to have seized and swallowed a small Boaedon f. fuliginosus in mistake for a frog while being fed. The bulky meal was later disgorged (D. K. Blake). Defence. These small snakes behave like a puffaddei- when disturbed, inflating the liody with air and hissing fiercely. They seem rather reluctant to bite. Venom. W. Armitage and D. K. Blake were both bitten on the hand by Umtali specimens within two hours. No treatment was given. In each case the hand became swollen to the wrist and a dull throbbing pain was observed. The swelling sid^sided after three days, leaving a feeling of stiffness. There was no sloughing away of flesh. Habits. The prominent upturned rostral suggests fossorial habits, l)ut the habits of this smaller species do not seem to differ from those of C. rhomheatus. Distribution. Widely but rather patchily distributed through- out Southern Rhodesia. This species is locally abundant at Umtali, Avhere it seems to completely replace C. rhombeatus. At 76 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Odzi, only 20 miles away, the latter species is common and defilipini very scarce. Localities: Karoi; Trelawney ; Mazoe; Salisbury; Musami; Umtali; Vumba Mountain; Cliirinda Forest; 165 miles north of Bulawayo ; Gwaai ; Turk Mine ; Bulawayo ; Kliami ; Matopos ; Figtree; Empandene; Essexvale; Balla Balla; Irisvale; Mount Darwin. BiTis ABiETANS ARiETANS (Mcrrem) Cohra lacliesis Laurenti, 1768, Syn. Eept., p. 104. Vipera (Echidna) arietans Merrem, 1820, Vers. Syst. Aniphib., p. 152. Bitis arietans Boulenger, 1902, p. 18; 1910, p. 522; Gough, 1908, p. 39; Chubb, 1909a, p. 597; 1909b, p. 36; FitzSimons, F. W., 1912, pp. 216, 217; Hewitt and Power, 1913, p. 165; FitzSimons, V. F., 1939, p. 24; Tasman, 1953; p. 25; Eose, 1955, pp. 156-161; Isemonger, 1955, p. 88. Native names for the Puffadder. Ihululu (Sindebele) ; Chiva or M'vumbi (Cheshona). Variation. (52 specimens.) Midbody scale rows 31-36; ven- trals 126-141 ; anal entire ; subcaudals 17-36 ; smooth ; upper labials 12-15. Tail length .05 to .16 of the total. Colouration. Dark grey to reddish brown above, with back- ward-directed yellow or white and black chevrons. Yellow below with black markings. Size. Largest $ (NM/M.958) 915 (800+115) mm. from Khami. Largest 5 990 (920+70) mm. from Umtali. New-born young measure just over 200 mm. in total length. Sexual dimorphism. In 23 males the range of ventrals is 126- 136 ; range of subcaudals is 30-36, and the tail length is .12 to .16 of the total. In 27 females the range of ventrals is 131-141 ; range of subcaudals is 17-25, and the tail length is .06 to .09 of the total. Breeding. Captive specimens observed mating on 26.viii.56. A 30" 9 produced a brood of 35 young on 2.xii.57 at Essexvale. Another Essexvale specimen, 2'8" in length, contained 37 well- developed young at the beginning of November. Diet. All stomachs containing food held rats. Captive speci- mens took rats readily. "When hungry, some specimens would take toads (Bufo carens; Bufo regidaris) rather reluctantly, other consistently refused to take anything but rats and starved to death rather than swallow a toad. On the other hand, juve- BROADLEY : SNAKES OF SOUTHERN RHODESIA 77 niles fed readily on small toads and frogs {Bana delalandii). One took a shrew {Crocidura sp.). Enemies. On two occasions large cobras (Naja h. haje) were found to contain adnlt puffadders. Both this species and Naja n. mossamhica readily devoured new-born puffadders in captiv- ity. Defence. Although it may sometimes try to escape when ap- proached, more often than not the puffadder remains motionless and its colouration makes it difficult to see in dry grass. As it often lies on footpaths through long grass, waiting for an unwary rat, this species is responsible for the majority of the cases of snakebite. Often the puffadder gives warning of its presence by inflating the body with air and hissing fiercely ; at the same time the head is drawn back over the body in a striking coil. Venom. As puffadder bites are invariably received on the limbs, prompt treatment can usually localise the venom and pre- vent death. The effects are unpleasant, local haemorrhage often causes an area around the bite to turn black and slough away and sloughing of the skin often recurs for several years after- wards. Hahits. The Puffadder does most of its hunting at night, but is often found basking during the day. It frequently becomes a road casualty. Distribution. Common in Matabeleland but less plentiful in other parts of the colony. This species seems to be very scarce in Salisbury District. Localities : Sinoia ; Mazoe ; Hunyani ; Gatooma ; Salisbury ; Marandellas ; Odzi ; Umtali ; Mount Silinda ; Chikore ; Victoria Falls ; Zambezi-Sebungwe Junction ; Shangani River ; Gwaai ; Bulawayo ; Khami ; Essexvale ; Balla Balla ; Irisvale ; Glass Block ; Gwanda ; Tod 's Hotel ; Beitbridge ; Mount Darwin ; Bubye River ; Umvuma. BiTis GABONicA GABONicA (Dumeril and Bibron) Echidna gahonica Dumeril and Bibron, 1854, VII, p. 1428, pi. Ixxxb. Bitis gabonica Isenaonger, 1955, p. 88; FitzSimons, V. F., 1958, p. 211. Variation. (4 specimens.) Midbody scale rows 39-46; ventrals 134-136 ; anal entire ; subcaudals 19-22 ; upper labials 16-17. Tail length .06 to .09 of the total. All four specimens are females. 78 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Colouration. Head buff, a dark brown triangular wedge, run- ning from the orbit straight down to the lip and diagonally back to the angle of the jaw ; this marking is divided by a narrow light line or spot. A vertebral series of sharply defined buff rectangles are connected by hour-glass shaped rich brown markings. The lateral markings consist of a complex geometrical pattern, com- posed of bold triangles in shades of buff, purple, brown and pink. The ventrum is buff with dark grey infuseations. Size. Largest (UM/R.25) 1266 (1196+70) mm. from Dzoroka, Chipinga Dist. Distribution. This massive viper ascends the river valleys from Mocambique and is liable to be found anywhere along the Eastern Border. It is reported to be not uncommon in the Inj^anga Tea Estates. The Stapleford specimen was taken on the edge of a plantation at 6,000 feet. Localities : Pungwe Valley ; Hondi Valley ; Stapleford ; Dzor- oka. BiTis CAUDALis (A. Smith) I'ipera caudalis A. Smith, 1839, 111. Zool. S. Africa, Rept., pi. vii. r.iiis caudalis Chul)h, 1909a, p. n97; 1909b, p. 3o; Boulenscr, 1910, p. i>:lo\ FitzSimons, F. W., 1912, pp. 217, 218; Isemonger, 1955, p. 89. Bitis candaUs (niispiint) Tasmaii, 1953, p. 27. Variation. (8 specimens.) Midbody scale rows 24-27; A'entrals 120-13-1-; anal entire; subcaudals 16-26; upper labials 11-18. Tail length .07 to .11 of the total. Colouration. Males — Head light red-brown passing to grey on the temples, a yellow line connecting the supraocular "horns," dark lines radiating from orbit to mouth, a U-shaped dark red- brown marking on back of head extending as a pair of bars on the neck. Body blue-grey dorsally, with a series of dark red- brown, yellow-edged, blotches, large and more or less oval in shape. Laterally light red -brown with a series of oval markings, which are grey above and dark red-brown, edged with yellow, below. Ventrum white, tinged with orange laterally. Females — Light sandy brown with indistinct rows of dorsal and dorso- lateral blotclies, whieh are oidy slightly darker than the ground colour. Ventrum butt". BROADLEY : SXAKES OF SOUTHERN RHODESIA 19 Size. Largest 6 (NM/M.857) 301 (270+31) mm. from Beit- bridge. Largest 9 (NM/M.93f)) 322 (300+22) mm. from Bula- wayo. Defence. A very truculent and' "explosive" little viper. Hisses very loudly for so small a snake and strikes fiercely at anything within range. Habits. A Horned Viper from Bulawayo could not be induced to "sidewind." Local conditions seem unsuitable for a species which has become specialised for life in loose sand. It would seem that these Bulawayo specimens have spread out of their normal environment. Distribution. Matabeleland. Localities: Insiza ; Bulawayo; We.stacre ; Beitbridge. BiTis ATROPOs (Linne) Coluber atropo.s Linni;, ITo-i, ^Miis. Ad. Frid., p. 22, pi. .\iii, rig. 1. Bitis atropos Lsemonger, 1955, p. 90; FitzSimons, Y. F., 19."8, p. 210. Variation. (7 specimens.) Midbody scale rows 29-31 ; ventrals 121-134; anal entire; subcaudals 18-25; upper labials 11-12. Tail length .07 to .11 of the total. Colouration. Above, grey-brown with a double series of dark dorsal blotches arranged in pairs on either side of the vertebral line : these abut against an ill-defined lighter dorsolateral line, below which there is a similar series of dark blotches. There are some dark markings on the top of the head. Below, chin white or cream with some sharply-defined black markings on the lower labials, ventrum white to dark grey. Size. Largest $ (NM/M.1702) 347 (310+37) mm. from Chimanimani Mts. (5.500'). Largest ? (NM/M.1704) 288 (268+ 20) mm. from Chimanimani Mts. (8,000'). Diet. A juvenile from Inyanga Xorth, in captivity at Umtali, gorges himself on toads {Bufo r((/ulari. ^•iiJMrALi ( • ••euLAWAyo 2V* .FoRT VICTORIA %^ '£' Rivea^ Eecorded localities for Bitis. Bitis arictans arietans (Merrem) O Bitis gabonica gahonica (Diimeril and Bibroii) ■ Bitis oaudalis (A. Smith) A Bitis atropos (Linne) 10 minutes later, now only semiconscious and staggering. He had by now lost all sense of taste and smell. The patient was given 20 cc. of polyvalent serum in the buttocks and put to bed. By 11.30 the hand was swollen up to the wrist; the pupils were dilated and showed no reaction to light. By noon the patient was again fully conscious. BROADLEY : SNAKES OF SOUTHERN RHODESIA 81 Armitage's condition showed no change on the following day, followed by a slight improvement on the 14th. By the 15th, the hand was back to normal, the thumb remaining swollen. The pupils were normal, but the left eyelid remained closed, the right being half open. There was still no sense of taste or smell. Both eyes were fully open the next day and Armitage was discharged from the hospital. His sense of smell returned on the 17th, but focusing of the eyes was still slow. The sense of taste returned the following day. The site of the bite was numb, but there was no sloughing of flesh around the fang punctures, which is a normal feature of viperine bites. Hahitat. The Univei^sity of Cape Town sent an expedition to the Chimanimani Mountains on the Eastern Border of Southern Rhodesia in February 1958. A member of the Expedition, Mr. J. R. Grindley, collected 4 specimens of Bitis atropos and pre- sented them to the National Museum. His field notes on these Mountain Vipers are given verbatim: (NM/M.1701) . . . "In long grass by stream on floor of Bundi Valley at 5,300 ft." (NM/M.1702) ... "On path to Martin's Falls in open grassland at 5,500 ft." (NM/M.1703) ... "In grassland above Martin's Falls at 5,000 ft." (NM/M.1704) ... "In grass near summit of Point 71 at 8,000 ft." Distribution. The Eastern Districts of Southern Rhodesia. Not entirely restricted to the mountains, for the Inyanga North specimen came from an altitude of approximately 2,500 feet. Localities: Inyanga North (ca. 2,500') ; Pungwe River Cause- way (ca. 5000')'; Chimanimani Mountains (5,000' to 8,000'). BIBLIOGRAPHY BOGERT, C. M. 1940. Herpetologieal Results of the Vernay Angola Expedition. Bull. Amer. Mus. Nat. Hist., vol. 77, pp. 1-107, figs. 1-8, pi. i. 1943. Dentitional Phenomena in Coltvas and Other Elapids Avith Notes on Adaptive Modifications of Fangs. Bull. Anier. I\Ius. Nat. Hist., vol. 81, pp. 285-360, figs. 1-73, maps 1-3, pis. xlviii-li. BOULENGER, G. A. 1893. Catalogue of Snakes in the British Museum (Natural History). (London), vol. 1, pp. xiii -\- 448, figs. 1-26, pis. i-xxviii. 1894. Catalogue of Snakes in the British Museum (Natural History). (London), vol. 2, pp. xi + 382, figs. 1-25, pis. i-xx. 82 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY 1896. Catalogue of Snakes in the British Museum (Natural History). (London), vol. 3, pp. xiv + 727, figs. 1-37, pis. i-xxx. 1902. A list of the Fishes, Batrachians and Reptiles collected by ISIr. J. ffoUiot Darling in Mashonaland, with descriptions of new species. Proc. Zool. Soc. London, 1902, vol. II, pp. 13-18, 4 plates. 1910. A Revised List of the South African Reptiles and Batrachians Avith SjTioptic Tables, special reference to the specimens in the South African Museum and Descriptions of New Species. Ann. S. Afr. Mus., vol. V, pp. 455-538. Broadley, D. G. 1956. Snakes of Southern Rhodesia 1. The Striped Skaapsteker or Grass Snake (Psammophylax t. tritaeniatus) . African Wild Life, vol. 10, no. 3, pp. 215-216. 1957a. Snakes of Southern Rhodesia 2. The Northern Green Snake {PhilotJiamiius i. irregularis). African Wild Life, vol. 11, no. 1, pp. 53-55. 1957b. Snakes of Southern Rhodesia 3. The Night-adders . . . Genus Causus. African Wild Life, vol. 11, no. 2, pp. 115-118. 1957c. Snakes of Southern Rhodesia 4. The Southeastern Vine or Twig Snake {Thelotornis h. capensis). African Wild Life, vol. 11, no. 4, pp. 297-300. 1958. Serpentes, Colubridae : Duherria lutrix rhodesiana subsp. nov. Occ. Papers Rhod. Mus., vol. II, Part 22B, pp. 215-216. Chubb, E. C. 1909a. On the Batrachians and Reptiles of Matabeleland. Proc. Zool. Soc. London, pp. 590-597. 1909b. List of the Rhodesian Batrachians and Reptiles in the Rho- desian Museum Collection. Rhod. Mus. Bulawayo, 8th Ann. Report, pp. 34-36. COTT, H. B. 1928. Report of the Zoological Society's Expedition to the Zambesi, 1927. Proc. Zool. Soc. London, pp. 923-961, text fig., map, 4 plates. FiTZSlMONS, F. W. 1912. The Snakes of South Africa; their Venom and the Treatment of Snake Bite. Cape Town. Pp. xvi -f 547, figs. 1-162, A-Z. FiTZSlMONS, V. F. 1935. Scientific Results of the Vernay-Lang Kalahari Expedition, March to September 1930. Reptilia and Amphibia. Ann. Transvaal Mus., vol. XVI, pp. 295-397, figs. 1-30, pis. x-xi. BROADLEY : SNAKES OF SOUTHERN RHODESIA 83 1939. An Account of the Reptiles and Amphibians collected on an Expedition to South-eastern Rhodesia durmg December 19157 and January 1938. Ann. Transvaal Mus., vol. XX, pp. 17-46. 1946. Notes on some South African Snakes, including a description of a new species of Xenocalamus. Ann. Transvaal Mus., vol. XX, pp. 379-393. 1958. On a Small Collection of Reptiles and Amphibians from the Inyanga District, S. Rhodesia.. Occ. Papers Rhod. Mus., vol. II, part 22B, pp. 204-214. FiTzSiMONS, V. F. and C. K. Brajn 1958. Serpentes, Colubridae: Calamelaps ventrimaoulatus websteri subsp. nov. Occ. Papers Rhod. Mus., vol. II, part 22B, pp. 202- 203. GouGH, L. H. 1908. Catalogue of the South African Snakes in the Collections of the Transvaal Museum, Pretoria, the Albany Museum, Grahamsto'WTi, and the State Museum, Bloemfontein. Ann. Transvaal Mus., vol. I, pp. 17-45. Hewitt, J. 1912. Xotes on the Specific Characters and Distribution of Some South African Ophidia and Batraehia. Rec. Albany Mus., vol. 2, pp. 264-285. 1913. Description of Hdeophryne matalerisi-s, a New Batrachian from Natal; and notes on several South African Batrachians and Reptiles. Ann. Natal Mus., vol. 2, pp. 475-484. Hewitt, J. and J. H. Power 1913. A List of South African Lacertilia, Ophidia and Batraehia in the McGregor Museum, Kimberley, with field notes on various species. Trans. Roy. Soc. S. Afr., vol. Ill, pp. 147-176. ISEMONGER, R. M. 195."). Snakes and Snake Catching in Southern Africa. Cape Town. Pp. xii -f- 105, pis., figs. LOVERIDGE, A. 1939. Revision of the African Snakes of the Genera Mehelya and Gonicmotophis. Bull. Mus. Comp. Zool., vol. 86, pp. 131-162, figs. 1-2. 1940. Revision of the African Snakes of the Genera Dromophis and Psammophis. Bull. Mus. Comp. Zool., vol. 87, pp. 1-70. 1944. Further Revisions of African Snake Genera. Bull. Mus. Comp. Zool., vol. 95, pp. 119-247. 84 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY 1950. The Green and Black Mambas of East Africa. Jouni. East Afr. Nat. Hist. Soc, vol. 19, no. 89, pp. 251-252. 1951. On a Collection of Reptiles and Amphiljians taken in Tanganyika Territory by C. J. P. lonides, Esq. Boll. Mus. Comp. Zool., vol. 106, pp. 177-204. 19.33. Zoological Results of a Fifth Expedition to East Africa. III. Reptiles from Nyasaland and Tete. Bull. Mus. Comp. Zool., vol. 110, pp. 143-322, figs. 1-4, pis. 1-5. 1955. On a Second Collection of Reptiles and Amphibians taken in Tanganyika Territory by C. J. P. lonides, Esq. Journ. East Afr. Nat. Hist. Soc, vol. 22, no. 5 (97), pp. 168-198. 1956. On a. Third Collection of Reptiles and Amphibians taken in Tanganyika Territory by C. J. P. lonides, Esq. Journ. Tan- ganyika Soc, pp. 1-19. 1957. Cheek List of the Reptiles and Amphibians of East Africa (Uganda; Kenya; Tanganyika; Zanzibar). Bull. Mus. Comp. Zool., vol. 117, pp. 153-362. Parker, H. W. 1927. Parallel Evolution in some Opisthoglyphous Snakes, with the Description of a new Species. Ann. Mag. Nat. Hist., ser. 9, vol. 20, pp. 81-86. Pitman, C. R. S. 1938. A Guide to the Snakes of Uganda. Kampala, pp. xxi + 382, pis. i-xxviii, col. pis. A-Q, diags. I-II, maps. Rose, W. 1955. Snakes, Mainly South African. Cape Town. Pp. xvi + 213, figs. 1-89. Schmidt, K. P. 1923. Contributions to the Herpetology of the Belgian Congo based on the Collection of the American Museum Congo Expedition, 1909-1915. Part II. Snakes. Bull. Amer. Mus. Nat. Hist., vol. 49, pp. 1-146, pis. I-XXII, maps 1-19, figs. 1-15. Tasman, K. 1953. Every Man's Hand against Him. The Rhodesian Graphic Annual, Nov. 1953, pp. 17-37. VViTTE, G. F. DE and R. Laurent 1947. Revision d'un Groupe de Colubridae Africains. Genres Cala- melaps, Miodon, Aparallactiis et formes af&nes. Mem. Mus. Roy. Hist. Nat. Belgique, ser. 2, vol. 29, pp. 1-134, figs. 1-132. BROADLEY : SNAKES OF SOUTHERN RHODESIA 85 Glossary of Scientific Terms. Anal, the shield immediately preceding the vent. Chin-shields, see sublinguals. Frontal, see Fig. 10. Internasal, see Fig. 10. Loreal, see Fig. 10. Labial, one of the shields bordering the mouth. Lower labial, see Fig. 10. Mental, see Fig. 10. Midbody scales, those, other than ventral shields, encircling the body at a point midway between snout and vent. Nasal, see Figs. 8, 9, 10. Ocular, see Figs. 8, 9. Parietal, see Fig. 10. Postocular, see Fig. 10. Prefrontal, see Figs. 8, 9, 10. Preocular, see Figs. 8, 10. Eostral, see Figs. 8, 9, 10. Subcaudals, the series of scales beneath the tail, which may be single or (more often) paired. Sublinguals, see Fig. 10. Subocular, see Fig. 10. Supraocular, see Figs. 8, 9, 10. Symphysial, see Mental. Temporal, see Fig. 10. Upper labial, see Figs. 8, 9, 10. Ventrals, the series of broad plates on the belly. Fig. 7. From left to right, dorsal scales, smooth, dorsal scales, keeled, ven- tral scales, smooth, ventral scales, keeled. 86 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY H06TRAU pftgoCOLAfl HA.SA ftOSTRAU VPPtfl LA6IAU SOPRAOCUUAR ( PREFROMTALl Fig. 8. Head shields of TypJilops schlegelii mucruso (Peters). ROSTRA U SyP«AOCUUAR,_^ NASAL fU»9T«AU OCUUAR. P«£P»toWTV FR.ONTAU NVSAU Rostral NASALS l_p£tAL OCOUA^.. lA&lAU NASAL WfBfi. LA61AU5 Fig. 9. Head shields of Leptotyphlops longicauda (Peters). BROADLEY : SNAKES OF SOUTHERN RHODESIA 87 Fig. 10. Head shields of Causus rhombeatus (LichtensteiB), 1. rostral ; 2. anterior nasal ; 3. posterior nasal; 4. internasal; 5. prefrontal; 6. frontal ; 7. supraocular; 8. parietal ; 9. loreal; 10. preocular; 11. subocular; 12. postocular; 13. anterior temporal ; 14. posterior temporal; 15. upper labial; 16. mental ; 17. lower labial; 18. anterior sublingual ; 19. posterior sublingual ; 20. ventral ; 21. poison fang. 88 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY A Key to the Snakes of Southern Rhodesia A. Key to Families 1. Body encircled by small scales more or less uniform in size; 3 or 4 scales immediately preceding vent; eye minute, when distinguishable, beneath a shield 2 Body not encircled by small scales owing to the presence on the belly of a longitudinal series of transversely enlarged plates, known as ventrals; 1 or 2 scales immediately preceding vent; eye clearly visible beneath a transparent "watchglass" scale 3 2. Ocular shield not bordering mouth; 18 or more scales round middle of body; tail as long as or only slightly longer than broad; size small to moderate TYPELOPIDAE (B) (Blind-Snakes) Ocular shield bordering mouth; 14 scales round middle of body; tail much longer than broad; size vei-y small, wormlike LEPTOTYPHLOPIDAE (C) (Worm-Snakes) 3. Ventral shields much narrower than body; midbody scale rows more than 75; vestigial limbs present, discernible as a pair of "claws" before the vent BOIDAE (D) (Boas and Pythons) Ventral shields as broad as, or nearly as broad as body; midbody scale rows less than 50 ; no vestigial limbs present 4 4. No enlarged poison fangs at fro7it of jaw^ COLUBRIDAE (E) (Typical Snakes) One or more pairs of enlarged poison fangs at front of jaw 5 5. Poison fangs immovable, not enclosed in a sheath of membrane ELAPIDAE (F) (Cobras, Mambas, etc.) Poison fangs movable and very large, so folded back when not in use, encased in a sheath of membrane VIPEEIDAE (G) (Adders and Vipers) B. Key to the TYPELOPIDAE (Blind Snakes) of SoutJicrn Ehodesia Midbody scale rows 28; diameter into length 41-50 times Typhlops delalandii (Delaland's Blind-Snake) Midbody scale rows 30 or more; diameter into length 21-53 times Typlilops s. mucru.so (Zambezi Blind-Snake) ^Warning: Snakes of the genus PsammopMs have a pair of greatly enlarged fang-like teeth below the anterior border of the eye, the grooved fangs being situated below the posterior border of the eye. BROADLEY : SNAKES OF SOUTHERN RHODESIA 89 C. Key to the LEPTOTYPHLOPIDAE (Worm-Snakes) of Southern Rhodesia 1. Eostral in contact with supraocular; black or grey in colour 2 Eostral separated from supraocular by the nasal; flesh pink in colour Leptotyphlops longicauda (Long-tailed Worm-Snake) 2. Eostral very large, more than twice width of nasal Leptotyphlops scutifrons (Peters' Worm-Snake) Rostral narrow, less than twice width of nasal Leptotyphlops conjnncta (Jan's Worm-Snake) D. Key to the BOIDAE (Pythons) of Southern Rhodesia Only one species occurring south of the Zambezi Python sebae (African Python) E. Key to the COLVBRIDAE (Typical Snakes) of Southern Rhodesia 1. Subcaudals mostly in pairs 2 Subcaudals single 39 2. A loreal shield present between postnasal and preoeular 3 No loreal shield present 35 3. Pupil round (horizontal only in Thelotornis) AND anal divided . 12 Pupil vertically sub-elliptic AND/OE anal entire 4 4. Dorsal scales smooth . . .5 Dorsal scales strongly keeled in 15 rows at midbody 11 5. Snout rounded; prefrontals and internasals paired G Snout with angular horizontal edge; prefrontal usually single 22 6. Midbody scale rows 25-33 Boaedon f. fuliginosus (Common House Snake) Midbody scale rows l.')-21 7 7. Midbody scale rows 15; a single postocular Dubcrria I. rhodesiana (Ehodesian Slug-eater) Midbody scale rows 17 or more; two or three postoculars 8 8. - No pair of enlarged grooved fangs situated lielow the posterior border of the eye 9 A pair of enlarged grooved fangs situated below the posterior l)order of the eye, separated by an interspace froui the preceding teeth ... 24 9. Midbody scale rows 19-21; subcaudals more than 50; iu)stril directed upwards from between two nasals; semi-aquatic 10 Midbody scale rows 17; subcaudals less than 50; nostril directed out- wards from a single nasal; tcrrestial Lycophidion c. capense (Cape Wolf Suake) 90 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY 10. Midbody scale rows 19 Lycodonomorphus r. rufulus (Brown Water Snake) Midbody scale rows 21 Lycodonomorphus r. mlanjensis (Mlanje Water Snake) 11. Ventrals 195-220; subcaudals 44-58; vertebral scale row and ventrals ivory white Mehelya c. capensis (Cape File-Snake) Ventrals 173-184; subcaudals 60-68; vertebral scale row blackish, ventrals brown Mdlielya nyassae (Nyasa File-Snake) 12. Midbody scale rows 25-27; snout pointed Pseudaspis cana (Mole Snake) Midbody scale rows 11-21 13 13. Pupil round 14 Pupil horizontal; body extremely slender and vine-like 27 14. Scales smooth 15 Scales strongly keeled Dispholidus typus (Boomslang) 15. A broad dark vertebral band with numerous short cross-bars or stag- gered to form an irregular zig-zag line ; underside dark grey flecked with brown ; not exceeding 450 mm. in length Hemirhagerrhis n. nototaenia (Eastern Bark-Snake) Markings and colouration not as above 16 16. No pair of enlarged grooved fangs situated below the posterior border of the eye 17 A pair of enlarged grooved fangs situated below the posterior border of the eye, separated by an interspace from the preceding teeth 28 17. Midbody scale rows 21; ventrals 175-204 Meizodon s. semiornata (Semiornate Snake) Midbody scale rows 19 or fewer 18 18. Midbody scale rows 19; ventrals 140-149 Natriciteres o. olivacea (Olive Marsh-Snake) Midbody scale rows 17 or fewer 19 19. Midbody scale rows 17 (rarely 15); ventrals 132-141; 3 postoculars; colour in life not green Natriciteres o. uluguruensis (Montane Marsh-Snake) Midbody scale rows 15; ventrals 148 or more; 2 postoculars; colour in life green 20 20. Ventrals 148-169; subcaudals 77-115, rounded or angular but without keels 21 Ventrals 179-204; subcaudals 121-142, angular and strongly keeled like the ventrals PhiloiJiamnus s. semivariegatus (Variegated Bush-Snake) BROADLEY : SNAKES OF SOUTHERN RHODESIA 91 21. Usually two labials entering orbit; subcaudals 77-103 Philothamnus hoplogaster (Southeastern Green-Snake) Usually three labials entering orbit; subcaudals 94-115 Fhilothamnus i. irregularis'^ (Western Green-Snake; 22. Internasal single; snout horizontal Prosymna a. stuhlmanni (Eastern Shovel-snout) Internasals paired ; snout upturned 23 23. Internasals forming a median suture; habit moderate Prosymna lineata (Peters' Shovel-snout) Internasals separated by rostral ; habit slender Prosymna s. sundevallii (Sundevall's Shovel-snout) 24. Midbody scale rows 19 25 Midbody scale rows 17 26 25. Ventrals 202-241 ; salmon pink above with black dorsal blotches Telescopus s. semiawnulatiis (Tiger Snake) Ventrals 154-168; black to olive above, flecked with white Crotaphopeltis h. hotamhoeia (White-lipped Snake) 26. Ventrals 172-195; eye large with vertically elliptical pupil Chamaetortiis a. aulicii. mis. NE of (iatoonia. Headlands, Makoni Dist., 70 nils. SE of Salishnry. Heany, Unizingwane Dist., 15 mis. NE of Bulawayo. Hondi Valley, Inyanga Dist., 30 inls. N of Unitali. Hope Fountain, Umzinswane Dist., 8 nils. SE of Diilawayo. Horseshoe Block, Umvukwes Range, 40 nils. XW of Salisl.iiry. Hot Springs, Molsettei- Dist., 50 nils. S of Umtali. Hunter's Road, Gwelo Dist., 20 mis. N of Gwelo. Hunyani, now site of Lake Mcllwaine Dam-wall, 20 mis. W of Salisbury. Hunyani River, nfHiient of the Zaiiihezi River, .Maslionaland. Imbeza, Umtali Dist., 5 mis. N of Umtali. Insiza, Tnsiza Dist., 50 nils. NE of Bulawayo. Inyanga North, Inyanga Dist., 100 mis. N of Umtali. Inyati, Bubi Dist., 35 mis. NNE of Bulawayo. Inyazura, ?kIakoni Dist., 9.') nils. SE of Salisbury. Irisvale, Umzingwane Dist., 40 mis. SE of Bulawavo. Karoi, Urungwe Dist., 110 mis. NW of Sulisbui.s . Kariba, Urungwe Dist., Zambezi River, 35 mis. S of Chirundu. Kazungula, Wankie Dist., on Zambezi River, where N. Rhodesia, S. Rhodesia, Caprivi Striji and Bechuanaland meet. Kezi, Matobo Dist., '>5 mis. S of Bulawayo. Khami Dam, Bulawayo Dist., 10 mis. VV of Bulawayo. Killarney Mine, Tnsiza Dist., 5 mis. E of Filabusi. Kondo, 45 mis. WSW of T^iitali. Kutania, Lomagundi Dist., 45 mis. W of Salisbury. Iv Lake Mcllwaine, 20 mis. W of Salisbury. Legion Mine, Matobo Dist., 90 nils. S of liuhiwayo. Leopard Rock. T^'iiitali Disi., on summit of \'umb;( .Moinit.nin. 98 RTTIjLETTN : MUSETTM OF OOMrARATIVE ZOOLOGY Lukosi, Wankie Dist., 10 nils. S of Wankie. Lumane, Gwanda Uist., 12 nils. N of Gwandu. Lundi River, affluent of the Sabi Kiver, Matahelel.-ind. Lupane Valley, Nkai Dist., 80 nils. N of Bulawayo. M Maedieke, Maiandellas Dist., ,15 nils. SE of SalisVjury. Makunibi, Goronionzi Dist., 25 nils. NK of Sallslnuy. Marandellas, Maiandellas Dist., 40 nils. HE of .Salisluu-.v. Matetsi, Wankie Dist., 25 nils. S of Victoria Ealls. Matopos Dam, Matolio Dist., 18 nils. S of Bulawayo (Schist). Matopos Hills, Matobo Dist., 25 mis. S of Bulawayo (Granite). Maviiradona Mountains, 20 nils. N of Mount Darwin. Mazeppa Mine, Gwanda Dist., 10 nils. ESE of Gw;im(I;i. Mazoe, Mazoe Dist., 25 mis. N of Salisbury. Mchingwe River Bridge, Beling'we Dist., 20 mis. W of Shalmiii. Melsetter, Melsettei' Dist., 00 nils. S of Tlnitali. Miami, TTrunj>we Dist., 120 nils. NW of Salisbury. Mohem Mine, near Bembesi. Monte Cassino, 5 nils. SE of Maelieko. Moonies Creek, 5 mis. S of Selukwe. Mount Darwin, Darwin Dist., 80 nils. NNE of Salisbury. Mount Hampden, Salisbury Dist., 10 nils. NW of Salisliury. Mount Silinda, Ghipinga Dist., 100 mis. S of TTmtali. Mrewa, Mrewa Dist., 50 nils. ENE of Salisbiny. Mtao Forest, Chilimanzi Dist., 10 nils. SE of [Tnivuiii.n. Mtoko, Mtoko Dist., 80 mis. ENE of Salisbury. ISfusami, Mrewa Dist., 50 mis. ENE of Salisbury. N Nampini, Wankie Dist., Zambezi River, 50 nils, aliove Victoria Palls. N 'cenia Dam, Unizingwane Dist., 5 uils. SI<1 of P^sscwnle. Norton, Hartley Dist., 25 nils. W of Salisbury. N'sese River, affluent of the Unizingwane River, flowing into it at a point 10 nils. SE of Balla Balla. Ntabezinduna, Bubi Dist., 20 mis. NE of Bulawayo. Nyamashatu River, 12 mis. SSW of Umtali. Nyamandhlovu, Nyamandhlovu Dist., 30 mis. NW of Bulawayo. Nyamaropa, Inyanga Dist., 75 nils. N of Umtali. Nyamaziwa, Inyanga Dist., 5 mis. E of Rhodes Estate, 5-0,000 ft. Nyaratedzi River, Chibi Dist., IS mis. NE of Shabani. UKOADLKV : SXAKKS OF SOITIIKK.X K 1 lODKSIA DO o Odzi, Umtali Dist., I'o mis. \V of r?iit;ili. Odzi Bivpr, affluent of the Salii Kiver. Odzani Falls, Umtali Dist., 15 mis. X of I'liitali. Old Umtali, Umtali Dist., 5 mis. NW of Umtali. Penhalonga, Umtali Dist., G mis. X of Umtali. Plumtree, Biilalima-Mangwe Dist., .jo m's. W'SW of Bulavvayo. Prince Edward Dam, Salisbury Dist., Id iids. S of Salislniry. Pungwe River Causeway, Inyaiiga Dist., ii) mis. N of Umtali. Q (.^uc Que, (Jwclo Dist., 3.") mis. N of (iwelo. R l>;iiiia(|ualiaiu' River, P.ulalima -MaUKwe Dist., on the IJecliuaiia la nd horder, S of Plumtree. Redbank, Xyamandhlovu Dist., :^ii nils. X\V of Bulawayo. Howa Division, Umtali Dist., 10 m's. S of Umtali. Husape, Makoni Dist., 90 nds. SE of Halisbury. S Salisbury, Federal and Territorial Capital. Sawmills, Xyamandhlovu Dist., 55 mis. XW of Bulavvayo. Sebungvve River, affluent of the Zaml)e/.i River, flowing into it :it a point 90 mis. E of Victoria Falls. Selukwe, Selukwe Dist., 120 mis. 8E of (4welo. Shabani, Belingwe Dist., 100 mis. E of Bulawayo. Shamva, Mazoe Dist., 50 mis. NE of Salisbury. Shaiigani River, affluent of the Gwaai Rivei-, Matabeleland. •Shangani (Siding), Insiza Dist., 55 nds. NE of Bulawayo. Shawanoe River, Mrewa Diet., 40 mis. ENE of Salisbury. Shiloh, 25 mis. of Bulawayo. Sinkukwe, Umzingwane Dist., 10 nds. S of Balla Balla. Sinoia, Lomaguudi Dist., 65 nds. NW of Salisliury. Sipolilo, Sipolilo Dist., 85 mis. XNW of Salisbury. S|)iingvale, Umzingwane Dist., 16 nils. SE of Bulawayo. Stanniore, (iwanda Dist., 15 nds. S of Balla Balla. ]0() BULLETIN: MUSEU.AI OF COMPARATIVE ZOOLOGY Stapleford, Umt;ili Dist., 20 iiil«. XNE of Uintali. .Sun Ynt Sen AFinc, Matolio Dist., near Ke/.i. Syringa, Bulalinia-^langwe Dist., 4.5 nils. S\V of I5ulawayu. T Tandaai, Melsetter Dist., 45 nils. S of Umtali. Tanoanda River, afflnent of tlie Sal)i River, V2 nils. SE of I'in-lieiiouKli 1> ridge. Tanganda, Tea Ivstate, ('liiidiiga l>ist., -0 mis. SK of Bin-lii'iiougli l',riist., 5 nils. N of ^Mount Siliuda. Valindie, .Matobo Dist., 10 mis. «\V of Bulawayo. \'ictoria Falls, Zamhezi River. Viimba Mountain, Umtali l>ist.. Ill nils. Sp] of Umtali. W Wankie, Waiikie Dist., 5~) mis. SE of \'ictoria Falls. VVedza Reserve, Marandellas Dist., 00 mis. \V of Umtali. Westaere, Mato1)o Dist., 18 mis. SW of Ibilawayo. West Nicholson, Gwanda, Dist., 25 nils. ESE of Gwanda. Woodvale, Bulawayo Dist., 10 mis. N of Bulawayo. Z Zambezi-Selmugwe . I unction, Sclmngwi' Dist., Wtt nils. E of N'ictoiia h'alls Zezani, (iwanda Di t., on the Umzingwane River near Beitliridge. Zimlialiwe, X'ictoiia Dist., 15 mis. SSE of Fort \'ictoria. PLATES PLATE 1 Map of Soulin'in Hliodesiii. Li JM.A'l'K PLATE 2 Vpyer, Nyasa File Snake {MeJieh/a nyassae). This specimen came from Fatima Mission, between Bulawayo and the Victoria Falls. The photo Krapli clearly shows the liicarinate vertebral scale row. The tail is much more slender than in Mehelya c. capensis, and can be seen coiled beneath the body, on the right. Photo by D. G. Broadlev Lower, Mole Snake {Pseudaspis cana). This four-foot Bulawayo snake is the largest so far recorded from Southern Khodesia. The stout body and the relatively small head readily identify this powerful constrictor. Photo bv D. G. Broadlev fiW '^^^v^' ' ' PLATE 2 PLATE 3 Upptr, Suudevall's Shovel-snout {Prosymna .s. sundevaUii). This adult female was taken under a stone by the roadside at Essexvale. The dorsal stripe is bright orange against a purplish brown background. Apart from the colouring, this species is distingiiished from the lower one by its more slender body. Photo by U. a. Broadlev Lower, Peters' Shovel-snout {Prosymna lineata). This adult female was collected near Bulawayo. It is pale brown with darker markings and is more stoutly built tliaii the one pictured al)0ve. Photo bv D. G. Broadlev PLATE PLATE 4 l'l>l«'i', Xyasa Purple-glossed Snake (CaUniuIaii.s itnictilur iiiioJciiis) . This eighteen inch male eanie from a farm 15 miles nortli of Bulawayo. The iridescence of the glossy black scales can lie clearly seen. This snake preys U])on other Inirrowing sjiecic^s, |)a rticularly Ti/plihips and Lcptotiiplilops. Photo liv D. a. Broadlev Lower, Shield Snake (Aspidelaps s. scutati(s). This tine female came from near Beitln-idge. The enormous rostral shield, detached at the sides and in broad contact witli the prefrontals, is a good distinguishing feature. The dorsal scales on the i)Osterior part of the l)ody and the tail are strongly keeled. PI 10 to bv 1). C. B roadie V PLATK 4 PLATE 5 Upper, Egyptian Colna (Naja h. liajr). This massive cobra, seven feet in length, was captured at Irisvale. This is the all black phase commonly found in Matabeleland. Compare the Ijroad ''hood" of this species with the long narrow hoods spread by Naja n. mnsmmbica (PI. 6, upper) and Xaja meJanolriira (PI. (5, lower). Photo l)v D. (\. Broadlev Loirir, Banded Cobra (Xaja li. Iiajc, var. aniiulifcKi) . I pulled this fine cobra out of a teimitariuui adjoining my camp at Irisvale. This .- ■■ -', ■ '■ ':'#*?-^pS5 f M '*?:. . . ■'A / -S! ' ^ -f*^ r"«^% :,,<»■ '--«JL>> r--' ^■« - . •. i - ^ ">I'A. -•• , t .. PLATE G Bulletin of the Museum of Comparative Zoology AT II A K V A K I) ( ; O [j L E G E Vol. 120, No. 2 STUDIES ON THE COMPARATIVE EMBRYOLOGY OF THE REPTILIAN NOSE By Thomas S. Parsons Harvard Biologicil I^;rl)oratories Witli Seven Plates CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM March, 1959 Publications Issued by or in Connection WITH THE MUSEUM OP COMPARATIVE ZOOLOGY AT HARVARD COLLEGE Bulletin (octavo) 1863 — The current volume is \'nl. 120. Breviora (octavo) 1952 — No. 101 is current. ]\Iemoirs (quarto) 1864-1938 — Puhlicatioii was terminated with Vol. 55. JoHNSONiA (quarto) 1941 — A publication of the Department of Mollusks. Vol. 3, no. 38 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, uo. 22 is current. Proceedings of the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained on application to the Director of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Of the Peters "Check List of Birds of the World," volumes 1-3 are out of print ; volumes 4 and 6 may be obtained from the Harvard Uni- versity Press; volumes 5 and 7 are sold by the Museum, and future volumes will be published under Museum auspices. Bulletin of the Museum of Comparative Zoology AT HARVARD COLLEGE Vol. 120, x\o. 2 STUDIES ON THE COMPARATIVE EMBRYOLOGY OF THE REPTILIAN NOSE By Thomas S. Parsons Harvard Biological Laboratories With Seven Plates CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM March, 1959 No. 2 — Studies an the Comparative Embryology of the Reptilian Nosr By Thomas S. Parsons TABLE OP CONTENTS Page Introduction Purpose and scope 104 Acknowledgements 105 Materials and methods 106 Terminology 106 Observations Order Chelonia General 108 Adult anatomy 110 Early embryologj- 117 Later embryology 125 Jacobson 's organ 145 Order Ehvnehocephalia Adult anatomy 152 Early embryology 156 Later embryology 1 56 Order Squamata General 160 Adult anatomy ] 61 Early embryologj- 166 Later embryology 174 Jacobson 's organ 197 Order Crocodilia Adult anatomy 200 Early embryology 205 Later embryology 214 Jacobson 's organ 214 Nasal glands General 221 Glandula nasalis externa 221 Glandula nasalis medialis 225 Bowman 's glands 226 Other nasal glands 227 Olfactory nerves Early development of the olfactory nerve 228 Later development of the olfactory nerve 231 104 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Olfactory bulbs 236 Nerviis terminalis 241 Discussion Major subdivisions of the nasal cavity 241 Jacobson 's organ 248 Early embryology of the nasal cavities 254 Phylogenetic considerations 260 Literature cited 261 Explanation of figures 275 INTRODUCTION Purpose and Scope. The following paper is a survey of the embryological development and adult anatomy of the nasal cavities and their epithelial walls in the living orders of the Reptilia ; the embryology is considered in greater detail than the anatomy. Although the investigation was started as a study of Jacobson 's organ, it soon became apparent that the remainder of the nasal organ would also have to be considered, if any valid conclusions were to be reached. Despite this widening in the scope of the study, special emphasis has still been placed on Jacobson 's organ. There is a very large body of literature on the nasal area of reptiles, a fact made apparent by the size of the bibliography at the end of this work. However most of the previous studies have been rather limited in their scope. Thus there remain many lacunae in our knowledge of the nasal embryology of most groups, and much disagreement persists over the interpretation of the observed structures, especially concerning the presence and nature of Jacobson 's organ in turtles and crocodilians. In the following descriptions and discussions, an attempt has been made both to collect the various observations by earlier investi- gators, and to fill in some of the major gaps in our knowledge by studying a large number of reptilian embryos. Naturally, a detailed study of the whole nasal area in the entire class was impossible. One major limitation is imposed by the availability of material. Thus only four embryos of Sphenodon were seen, so that no intensive investigation of the nasal embry- ology of this form was possible. The suborder Lacertilia has been studied far more carefully than any other comparable group, and therefore is not discussed in detail here. Similarlv the skeleton PARSONS: NASAL EMBRYOLOGY 105 of the nasal area, although important for a complete under- standing of the olfactory organs, is not treated here, since it is better known than the soft parts of the region. However, some knowledge of the nasal glands and olfactory nerves is essential for a consideration of Jacobson's organ, and they are therefore discussed briefly. Acknowledgements. This investigation was carried out as a thesis problem under the supervision and guidance of Professor Alfred S. liomer ; to him I wish to express my deepest gratitude for his careful and constructively critical reading of the manu- script, and constant aid with the many problems which arose in its preparation. Also to Dr. Ernest E. Williams, I extend sincere thanks for his assistance on countless points during the course of this study. Many other people have patiently listened to my problems, and rendered valuable assistance ; to all of them, 1 am most grateful. I am indebted to numerous people for the loan of most of the material studied. The late Dr. G. B. Wislocki and Dr. R. 0. Greep of the Harvard Medical School have kindly allowed me to borrow large numbers of slides from the Minot Collection of embryos. Further embryological slides were loaned to me by Dr. A. M. Keese of West Virginia University, Dr. F. J. Ryan of Columbia University, and Dr. E. E. Williams of Harvard University. My thanks are also due to Dr. A. Fleminger for an adult Crotalus, and to Dr. P. P. Vaughn for an immature Alligator and an adult Pseudemys. Among the many people who assisted me in other ways, I should like especially to thank the librarians at the Museum of Comparative Zoology. Their constant help in locating many, often obscure, works for me is greatly appreciated. A portion of this investigation was carried out during the tenure of a National Science Foundation Predoctoral Fellowship. Finally, my deepest gratitude is due to my wife, Margaret C. Parsons. Besides making all the drawings for this work, she has assisted in countless other ways, including reading much of the manuscript, typing large sections of it, and making several of the dissections. Without her aid the work could not have been completed. 106 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY Materials and Methods. Embryos representing all four orders of recent reptiles were studied. These embryos were borrowed from the Minot Collection of the Harvard Medical School, from Columbia University, and from Drs. E. E. Williams and A. M. Reese. Most of the series were stained with alum cochineal, borax carmine, cochineal, hematoxylin, or iron hematoxylin, and counterstained with eosin, Lyon's blue, orange G, picric acid, or saifranin. The following forms were available: Chryscmys (80 embryos), Emys (9), Chelydra (9), Trionyx (3), Sphenodo7i (4), Laccrta (59), Aristelliger (32), Iguana (5), Sphaerodacty- lus (2), Anniella (1), Lygosoma (1), Thamnophis (70), Oxyhelis (3), and Alligator (31). Material representing all of the recent rei^tilian orders except the Rhynchocei)halia was dissected for this study; normally the nasal cavities were opened sagittally, as shown in the figures of the dissections. The specimens dissected were: Pseudemys (1 specimen), Chelonia (1), Drymarckon (1), Crotalus (1), and Alligator (3). Besides this material, one series of sagittal sec- tions through the head of an adult Storeria was studied. All of the specimens which are mentioned above were studied in detail, but no reconstructions were made. Although models or other reconstructions would have been most useful in certain cases, the time involved in their preparation would have pre- vented the examination of as many specimens as were actvially studied. In view of the comparative nature of this investigation, the use of as large a sample as possible appeared more desirable than the more concentrated study of a smaller number of embryos. Most of the drawings were made by tracing photographs of the sections or dissections. The eye, brain, integument, and nasal epithelium are colored black Avhile other tissues are stippled in the figures showing sections through the nasal areas of the embryos. In most cases the nerves are not indicated. Such a method gives an extremely accurate picture of the area, although in some cases the boundary between the nasal epithelium and the surrounding mesoderm is not, on the actual specimen, as sharp as the figures would indicate. Terminology. The differences in nasal anatomy between the various orders of reptiles are of sufficient magnitude to necessi- PARSONS : NASAL EMBRYOLOGY 107 tate the use of a different system of terminology for each order. Therefore discussions of the anatomical terms are most con- veniently included in the sections on tlie various orders. In general, an attempt has been made to follow either the terminol- ogy in most general usage, or that proposed in the most recent {)aper dealing with any specific group ; however in some cases, none of the available terms appeared to be suitable, and new ones are introduced. In most cases the forms mentioned are identified only to genera, and specific names are not cited. The classification is, unless otherwise noted, that of Romer (1956). In all discussions of the literature, the generic names are those suggested by Romer; the following list includes the synonyms which are used in the papers cited or on the slides studied. The first name is that used else- where, and the second the one employed in the present work. .Spelling variants, such as Crocodilus for Crocodylus, are not in- cluded, and specific names are considered only when necessary. Alligator cynocephalus =Cainian Alligator sclcrops = Caiman Aspidonectes = Trionyx Boa constrictor = Constrictor Caiman niger = Melanosuch iis Callopeltis == Elaphe Chersydrns = Acrochordus Chelonia cauana= Caretta Ch clone imhricata. = Eretmochelys Cist lido = Terrapene Coluher aesculapii = Elaphe Coluber natrix = Natrix Crocodilus gangeticus = Gavialis Dryophis = Ahaetulla Enhydris hardtwickei = Hydrophls Eutaenia = Thamnophis (jlavialis schlegelii = Tomistoma natteria= Spherwdon Pelias == Vipera Platurus = Laticauda Sphargis = Dermochelys Tropidonotus = Natrix 108 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY OBSERVATIONS Order Chelonia General. Despite a fairly extensive literature on the anatomy of the nose in turtles, only a few genera have been adequately studied. The most important papers are those by Seydel (1896), Nick (1912), and Fuchs (1915). Hoffmann "(1879-1890) de- scribes a large series of forms, but never in detail. The other Avorks cited below are of lesser importance, although they are mainly on forms not considered in the major papers. The most careful recent review is that of Matthes (1934). Terminologj^ poses a major problem in chelonian nasal anat- omy. Most authors have followed Seydel (1896) in this regard, although Nick (1912) expresses doubt on the propriety of some of the terms, and Fuchs (1915) uses an essentially new system. In the following descriptions the terms adopted are mainly those of Seydel, while Nick is followed for some of the specialized structures of sea turtles. However, in some cases it is felt that none of the available terms ai'e appropriate, and new ones have been made. The most important new usage concerns the partes olfactoria and respiratoria of Seydel. In turtles these terms have come to be used in a purely topographic sense while in mammals they refer to the histological nature of the epithelium. The pars olfactoria is the dorsal or posterodorsal portion of the nose, typically separated from the remaining portions by a pair of horizontal ridges, one lateral and one medial, the Grenzfalten of Seydel. As far as is known, its walls are always almost entirely covered by olfactory epithelium ; it thus corresponds closely to the pars olfactoria in the mammalian sense despite its definition on topographic rather than histological criteria. However, the pars respiratoria of Seydel, essentially the remainder of the cavum nasi proprium, bears not only non-sensory respiratory epithelium, but also has certain sensory zones. Since by defini- tion a pars respiratoria is non-sensory, this term is obviously an undesirable one. However, the only inajor paper since 1896 which has not fol- lowed Seydel in the use of these terms is that bj^ Fuchs (1915). There, the pars olfactoria of Seydel is named the recessus PARSONS: NASAL EMBRYOLOGY 109 superior posterior. The latter term, while descriptive of the condition found in sea turtles, is inappropriate for most forms. Fuchs gives no corresponding term for SeydeFs pars respiratoriu which is treated as a series of separate parts. In the absence of an}- other available terms, Seydel's pars olfactoria is here designated rcgio olfactorla and his pars respira- toria as regio intcrmedialis. The former at least preserves the familiar term in part, and the latter is descriptive of the general position in the nose as a wiiole for this rather variable region. Another set of new terms apply to the furrows in the surface of the regio intcrmedialis which bear sensory epithelium. Seydel and most later workers refer to them as partes of Jacobson's organ. Since in the present paper this sensory epithelium is not considered to constitute a Jacobson's organ, other names are necessary. The furrows are here called sulci. Thus Seydel's pars anterior is the sulcus anterior, his pars ventralis the sulcus ven- tralis, and his pars dorsalis the sulci mcdialis and lateralis. Two other terms require comment. Concha is used by different authors in very different seuses. Gegenbaur (1873) and Solger (1876) proposed a restricted definition, but this has not been generally accepted. A possible homolog of the concha of other reptiles is the laterale Grenzfalte of turtles; however a lateral outpocketing of the wall of the regio olfactoria, which is analo- gous to the saurian concha in tliat it increases the surface covered by olfactory epithelium, occurs in some turtles and is sometimes referred to as the concha (e.g. Plate, 1924). To avoid confusion, Seydel's term 3Juschclu'ulst is here used for such an outpocket- ing, and concha not at all in reference to turtles. Finally, ductus nasopharyngeus is used instead of ductus cJioanalis. Fuchs (1915) objects to the former term on the basis of embryological differences between the structure in mammals and in turtles, but as it is most often used in a general sense for any duct-like con- nection of the nasal cavity with the mouth, it is here retained. For descriptive purposes, the turtles may be split into four groups, only two of which are known in any detail. These are the ordinary land and pond turtles (Chelydridae and Testudini- dae including Emydinae) and the sea turtles (Cheloniidae and Dermochel,vidae). Two groups about which much less is known are the Trionychidae and the Pleurodira. Apparently nobody has studied the nose of the Dermatemvdidae or Carettoehelvidae. 110 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Morphologically, turtles can be arranged in a series showing a greater or lesser degree of adaptation of the nose for an aquatic life. The greater this adaptation, the more important and com- plex is the regio intermedials and the smaller is the regio olfac- toria. Using those genera which have been most studied, such a sequence would be, in order of increasing aquatic adaptation : Testudo; Chclydra; Emys, Chrysemys, and Pseudemys; Pelo- medusa; Trionyx; and Chelonia, Caretta, Eretmochelys, and Dermochelys. Naturally such a sequence is not phylogenetic, but represents the end points of various lines. Adult Anatomy. The Chelydridae and Testudinidae display the simplest arrangement of the nasal cavities found in turtles, and are probably closer to the primitive ehelonian condition than any other living groups. Chclydra may be taken as a possible starting point, with Testudo somewhat specialized for terrestrial and the emydines for aquatic life. Chelydra has been carefully studied by Nick (1912), and the following description and Figure 1 are taken from his paper with only slight modifications in terminology. Figure 1. The nasal cavity of Chelydra (after Nick, 1912). A, lateral view of the right nasal cavity to show the medial wall. B, medial view of the left nasal cavity to show the lateral wall. Figure 2. The nasal cavity of Chelonia. A, medial view of the left nasal cavity to show the lateral wall. B, lateral view of the right nasal cavity to show the medial wall. PARSONS : NASAL EMBEYOLOGY 111 DNP 112 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY The vestihulum is fairly short and straight. In Chelydra it runs ventrally as well as posteriorly from the naris externus, but this is probably a specialized condition; in most turtles it runs directly posteriorly. In transverse section it is circular anteri- orly, becoming higher and oval posteriorly. The vestibulum is separated from the cavum nasi proprium by a slight ridge which is most prominent laterally (the Grenzwall of Nick, 1912, and .Seydel, 1896, here referred to as the postvestibular ridge). The cavum nasi proprium is divided into regiones olfactoria and intermedialis by prominent mediale and laterale Grenzfalten. These ridges both start on the dorsal wall of the cavum nasi proprium just posterior to its juncture with the vestibulum and run posteroventrally to the dorsal margin of the ductus naso- pharyngeus. The mediale Grenzfalte is slightly ventral to the laterale and divides posteriorly, with the smaller dorsal arm meeting the dorsal border of the ductus as described above, and the larger ventral one continuing along the medial wall of the ductus nasopharjaigeus. The regio olfactoria is a large chamljer lying posterodorsal to the regio intermedialis. It is roughly semicircular in sagittal, and oval in transverse section. Its ventral surface is open to the regio intermedialis except at its posteriormost extremity. The posterior half of its lateral wall projects slightly into the nasal cavity forming a distinct Muschelwulst whose ventral margin is part of the laterale Grenzfalte. The regio intermedialis forms an oval tube with the dorso- ventral axis greater than the mediolateral. It runs postero- ventrally from the vestibulum to the nasal end of the ductus nasopharyngeus so that all three structures make one continuous and straight tube. The walls bear low ridges which form the boundaries of three shallow grooves. Seydel (1896) and most later workers term these grooves Jacobson's organ. The first ridge forms a circle in the wall of the regio intermedialis just posterior to the boundary with the vestibulum, thus forming a circular groove, the sulcus anterior, between it and the post- vestibular ridge. A second ridge runs posteroventrally from the first along the lateral wall approximately halfway between the floor of the regio intermedialis and the laterale Grenzfalte. It becomes less pronounced posteriori}^, disappearing near the start PARSONS: NASAL EMBRYOLOGY 113 of the ductus nasopharyngeus. The groove between this ridge and the laterale Grenzfalte is the sulcus lateralis while the ventral and medial walls of the regio intermedialis are the sulcus ventralis et medialis. Nick w^as unable to distinguish differences in the epithelium of the ridges and sulci. The ductus nasopharj-ngeus continues posteroventrally from the regio intermedialis, from which it is not sharply divided. The ductus boars low columnar epithelium which gradually merges with the thicker sensory epithelium of the cavum nasi pro- prium. As previously mentioned, the mediale Grenzfalte con- tinues posteriorly as a low ridge along the medial wall of the ductus. Lateral and somewhat dorsal to the anterior part of the ductus is a small blind sac opening posteriorly into the ductus, the 7rccssus ducti nasopharyngci (the sinus maxillaris of von Mihalkovics, 1898, and Nemours, 1930). Nick does not describe the choanae. Kinosternon is the only other genus of the Chelydridae whose nose has been studied, and it is, judging from the diagrammatic figures and brief description in Hoffmann (1879-1890), entirely unlike that of any other turtle. Until this genus is restudied in greater detail, it is impossible, profitably, to compare the nasal cavity of Kinosternon with that of other turtles. Among the Testudininae (Testudinidae) the only genus which has been studied is Tesfudo, and its anatomy is known from a brief accouut by Hoffmann (1879-1890) and the detailed work of Seydel (1896). In this genus the regio olfactoria is large with a very well developed Muschelwulst. The mediale Grenzfalte is more prominent than in Chelydra. On the other hand, Testudo possesses a simpler regio intermedialis. There is only one small ridge which runs posteroventrally from the ventral part of the postvestibular ridge along the medial wall of the nose. Be- tween this ridge and the mediale Grenzfalte, there is a very shallow sulcus medialis. Seydel found vomeronasal epithelium only in this sulcus, with the remainder of the regio intermedialis bearing non-sensory respiratory epithelium. Few genera of the Emydinae (Testudinidae) have been studied; all seem very similar and quite like Chelydra or Testudo. The best known form is Emys, which has been briefly noted by Solger (1876) and described in detail by Seydel (1896), who 114 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY also studied Chrysemijs. McCotter (1917) has written on '"Chry- semys punctata.'' It is uncertain just what form this is ; however, it is almost certainly either Chrysemys or Clemmys. Finally, Nemours (1930) liriefly describes Pseudcmys. An adult Pseudemys was dissected for the present investiga- tion and found to agree closely with most of the previous studies, especially that of Seydel. Emydines differ from Chelydra pri- marily in the reduction of the size of the regio olfactoria and increased complexity of the regio intermedialis. The Muschel- wulst, if present at all, is very small. In the regio intermedialis there are ridges similar to those found in Chelydra, l)ut there is also a third ridge along the medial wall, separating the sulci ventralis and medial is. According to Seydel, all four sulci bear vomeronasal epithelium, with non-sensory respiratory epithelium being found on the Grenzfalten and ridges. Figure 31, which shows a late embryo of Chrysemys, illustrates the relationships of the sulci. The sea turtles show extreme adaptation to the aquatic en- vironment in the great development of the regio intermedialis and the comparative reduction of the regio olfactoria. For this study an adult CJielonia was dissected as is shown in Figure 2. The vestibulum is a short tubular structure without any marked increase in height posteriorly. Its juncture with the cavum nasi proprium is not marked by any postvestibular ridge, but is sharply defined by the sudden increase in cross-sectional area of the cavity. The cavum nasi proprium is divided into regiones olfactoria and intermedialis by tlie laterale and mediale Grenzfalten as in the forms already described, but the olfactoria has become re- stricted to the posterodorsal third of the cavum. The complex regio intermedialis is divided into several sections whicb do not seem comparable to the sulci of other turtles. Unfortunately most of these sections have received several names, many of which are rather confusing. The regio intermedialis may be thought of as a roughly tubular structure with several anterior outpocketings. The central tube is not generally given a separate name ; however it corresponds to the anterior half of the ceUa media plus the pars posterior of the recessus inferior of Fuchs (1915). This tube runs from the vestibulum to the ductus nasopharyngeus so that the three form I'ARSONS : NASAL EMBRYOLOGY 115 a straight passage from the naris externus to the choana. Tri the floor of this central portion of the regio interniedialis there is a shallow groove running anteroniedio-posterolaterally. The first outpocketing to be considered is the recessus ven- tralis. It is the untere Aushnchtung of Gegenbaur (1873), the recessus inferior of Nick (1912), and the pars anterior of the recessus inferior of Fuchs (1915). This deep recessus is narrow anteroposteriorly, but lateroniedially it is wider than the central part of the regio intermedialis due to its extending into the medial wall as a shallow groove. This groove pushes postero- medially into the septum, reaching dorsally to the level of the top of the vestibulum where a small ridge separates it from the recessus dorsalis (see Fig. 2B). It also reaches slightly lateral to the cella media of Fuchs (1915) where it expands as a wider but shallower pocket. The ventral portion of the recessus ven- tral is curves anteriorly so that its deepest point is the anterior- most part of the cavum nasi proprium. The recessus dorsalis (the ohere Aushuchtung of Gegenbaur, the recessi superior and medialis of Nick, and the recessus su- perior anterior of Fuchs) is somewhat more complex. The main part of it is a deep groove similar to the recessus ventralis in its anteroposterior narrowness and lateromedial width. Medially, it is separated from the recessus ventralis by the narrow ridge already mentioned; lateralh', the recessi are widely separated. The medial section of the recessus dorsalis extends posteriorly in the septal wall for a short distance, and this portion is often, as by Nick, described as a separate recessus medialis (see Fig. 2A). It is not open into the remainder of the regio intermedialis ventrally, but only into the central part of the recessus dorsalis. and is therefore here considered merely a part of that recess, as suggested by Fuchs. The regio olfactoria (the innere Riechgruhc of (icgenbaur and the recessus superior posterior of Fuchs) forms another large recess which is roughly two-thirds of a sphere in shape. It lacks a Muschelwulst. The laterale Grenzfalte is a small but distinct ridge running from the ventral end of the bar of tissue separat- ing the recessus doi-salis from the regio olfactoria to the juncture of the cavum nasi i)roprium and the ductus nasopharyngeus. This ridge is the concha of Gegenbaur and Fuchs. The mediale Grenzfalte is less marked, but since the regio olfactoria extends 116 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY farther medially than the adjoining section of the regie inter- medialis, there is a distinct angle in the wall of the nose to represent it. The ductns nasopharyngeus is a long simple duct, not sharply set off from the regio intermedialis. The mediale Grenzfalte continues along the dorsomedial wall as a low ridge which slowly disappears posteriorly. Chelonia lacks a recessus ducti naso- pharyngei. Along the lateral margin of the choana there is a series of about twelve small stiff papillae; these are described in a separate paper (Parsons, 1958). There is a moderately extensive literature on the nasal cavities of sea turtles ; all appear to be very similar, although the similar- ity has been somewhat obscured by the variation in terminology. Chelonia has been studied by Solger (1876) and Nick (1912), Caretta by Gegenbaur (1873), Eretmochelys by Hoffman (1879-1890) and Fuchs (1915), and Dermochelys by Hoffmann (1879-1890) and Nick (1912). Deraniyagala (1939) and Par- sons (1958) discuss the variation in the choanal papillae. All the forms Avhich have been studied possess the same basic pattern of recessi dorsalis and veutralis in the regio intermedialis, al- though in Dermochelys the entire nasal cavity is shorter and the recessus ventralis smaller than in the members of the Cheloniidae. According to Nick, the sensory (vomeronasal) epithelium of the regio intermedialis is restricted to the recessi in Dermochelys, and to the recessi plus a narrow midventral band in Chelonia. The nose of the highly specialized genus Trionyx has been but little studied. The most complete account is that of Ploffmann (1879-1890) ; Seydel (1896) compares it to Emys, but unfortu- nately Avithout any detailed description or figures. The general pattern appears to resemble that found in Emys, but Avith re- duction of the regio olfactoria and greater development of the regio intermedialis. Hoffmann figures some irregularities in the su'rface of the latter regio, but these do not appear to be com- parable to the sulci found in Emys. Little can be said concerning the nose of pleurodires. The only description aA^ailable in the literature appears to be that of A'an der MerAve (1940) on Pelomcdtisa, in Avhieh he is mainly con- cerned with the skull and treats the nasal caAdties only briefly. Both recessi and sulci are said to occur in the regio intermedialis, but their relationships are not made clear. PARSON'S: NASAL EMBRYOLOGY 117 Early etnbnjology. There have been few studies made on tlie early develojnnent of the nose in turtles. Sevdel (1896) describes a small series of Chryscmys embryos, and Voeltzkow (1903) con- siders the general embryology of the face in Eretmochelys, giving excellent figures of the external appearance of the nose in his embryos. Dohrer (1912) discusses the nose briefly in his paper on Chelydra. IMore recently, Loew (1956) considers Emifs, mainly in relation to the homologies of Jacobson's organ. The other papers which must be considered, Fuchs (1907 and 1915; Emys) and Thiiter (1910; Chrysemys) , are concerned primarily with the development of the palate and therefore give little detail on the eai'lier stages. Since the findings of all these investi- gations are similar, at least insofar as these stages are concerned, and in essential agreement with the present study, no extensive review is necessary. The youngest Chryscmys embryos used in the present study which display any nasal differentiation have well developed nasal placodes without any conspicuous inpocketing (see Figs. 3 and 1). These placodes are located ventrolaterally on the heads, extending from the anterior end of the optic vesicles almost to the tip of the snout. They are nearly circular, with the ventral margin slightly flattened. In none of the series studied are the placodes in actual contact with the wall of the brain; there is always a small amount of undifferentiated mesenchyme inter- posed betAveen the two. As in most of the early stages, there is as yet no visible nuclear differentiation, but the ectodermal tissues, both neural and integumentary, are distinguished from the intervening mesoderm by their darker-stained cytoplasm. The placodes are three to four times as thick as the simple cuboi- dal epithelium of the head with which they gradually merge without any sharp boundaries. The thickest part of the placode is posterodorsal to its center. In the youngest series, the nasal epithelium is simple columuar, but in a slightly more advanced embryo it is either stratified or pseudostratified, probably the latter since Voeltzkow (1903) simply describes it as "hohe Cylinderepithel" in a somewhat later stage in Eretmochelys. In the placode the nuclei tend to be basally situated, instead of being central as in the general body epithelium. 118 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY The nasal placode is, in the next stage, slightly indented, as shown in Figure 5. Its position is much as in the preceding stage, but it does not now reach as close to the anterior end of the snout as formerly, and it is separated from the brain by a greater thickness of mesenchj-me. The nasal epithelium has be- come noticeably thicker, being up to five nuclei in deptli. Whether it is stratified or pseudostratified could not be told from the available material ; this stage appears, however, slightly less iidvanced than Voeltzkow's (1903) embryo mentioned above. The tendency for the nuclei to be basal is most clearly marked in the region of greatest indentation. However, there are more mitotic figures peripherally than basally. The greatest inpocketing is in the thickest region of the placode, posterior and dorsal to its center. Thus, although there is nowhere a clear separation between nasal and general body epithelium, the transition is most abrupt posterodorsally. The indentation in the placode continues to deepen to form a simple pit. x\t this stage the indentation is still strongest dorsally, so that the dorsal wall is almost horizontal and the ventral wall vertical. However, now the deepest part is central when seen in frontal section instead of being posterior (see Fig. 6; Fig. 7 shows this stage in sagittal section). The nasal anlage is still ventrolateral and subterminal on the snout. Voeltzkow's (1003) figure 1 on plate 28 shows the external appearance of this Figure 3. Cliryscmys 1049 (4.8; MC). Frontal section through the nasal area (section 33). 160x. Figure 4. Chiytscniy.t IUjU (4.8; MC). Transverse section through the nasal area (section 13). ICOx. Figure 5. Chrysemys 105.5 (6.2; MC). Transverse sections through the nasal area. A, section 74; B, 7 sections posterior to A; and C, 6 posterior to B. 90x. Figure 6. Chrysemys 1060 (5.6; MC). Frontal sections through the nasal area. A, section 90; B. 4 sections ventral to A; and C, 6 ventral to B. 90x. Figure 7. Chrysemys 1481 (o.O; MC). Sagittal section through the nasal area (section 16). 78x. PARSONS: NASAL EMBRYOLOGY 119 120 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY stage ; the pit is circular with its margins slightly raised, ap- ])arently more so in Eretmochclys than in Chrysemys. The thick- ened wall still merges gi-adnally with the general body epithel- ium, which is one or occasionally two cells thick. The transitiou NPl Figure 8. Chrt/acDiiis 1061 (7.6; MC). Transverse sections throvigh the nasal area. A, section 87; B, 6 sections posterior to A; and C, 9 posterior to B. 60.\. Figure 9. Clirijsciin/s 1(162 Hj.5; MC). Sagittal section through the nasal area (section 23). 60x. Figure 10. Chrysemys 2132 (7.0; MC). Transverse sections through the nasal area. A, section 189; B, 11 sections posterior to A ; C, 11 posterior to B ; and D, 6 posterior to C. 42x. is most gradual ventrally. The placode appears but little thicker than in the preceding stage and is histologically much the same, although the basement membrane is now very distinct, showing low gentle undulations, and the peripheral concentration of mitoses is more marked. PARSONS : NASAL EMBRYOLOGY 121 The pit increases in depth, growing dorsally I'rum the original placode (see Fig. 8). In some of the more advanced series of this stage its dorsal end is slightly lateral to the naris. The nasal pit is no longer round, but has become elongate so that the naris is oval, with the long axis anteroposterior. Anteriorly, the naris is wide ; it narrows as it goes posteriorly and slightly ventro- laterally. The pit is deepest in the anterior half, especially in the more advanced embryos, where it is approximately as high as long (see Fig. 9). At the posterior end it becomes very Ioav and is no more than a small band of thickened epithelium before it finally blends with the general body epithelium. The lumen is widest anteriorh'. At this stage there appears the start of differential thickening of the nasal epithelium ; it is thickest dorsally and medially, especially in the anterior two-thirds of the pit. Seydel (1896), however, in a Chrysemys embrj-o of similar stage noted no differences in thickness between various regions of the nasal epithelium. His description is otherwise in agreement with that given above. Histologicall.y there are few changes. There is still no visible nuclear dift'erentiation. The previouslj' noted concentration of nuclei basally and mitotic figures j^eripherally is still marked near the apex of the pit, but is not clearly seen elsewhere. The nasal epithelium appears to be up to six nuclei deep, but the cell shape and arrangement could not be distinguished. Seydel reports high columnar epithelium. The basement membrane tends to disappear dorsally, so that the boundaries of the nasal epi- thelium and the wall of the brain with the intervening mesoder- mal tissues are hard to distinguish. However, there appears to be no direct contact between the brain and nasal tissue. As before, the epithelia of the nose and of the rest of the head are continuous and not marked by any sharp separation. The final stage before the fusion of the lateral and medial nasal processes (Nasenfortsdtze of the German literature) is that of a deep, dorsally-extending pit. The shape is basically as in the ])receding stage. Transverse sections of this stage are shown in Figure 10. A slightly more advanced embryo is shown in Figure 11, while Figure 12 shows frontal sections of a similar embryo. The naris is widely open anteriorly, but becomes a long narrow slit posteriorly. It is most lateral near its center, curving medially at both ends. Its anterior end is well dorsal to the 122 I5ULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY posterior. Above the naris, the pit is dorsally and slightly later- ally directed, with a wide lumen anteriorly and ventrally and a very narrow slit-like one dorsally, especially in the posterior half. Figure 11. Chrysemys 1073 (9.5; MC). Transverse sections through the nasal area. A, section 171; B, 5 sections posterior to A; 0, 5 posterior to B ; and D, 8 posterior to C. 40x. Figure 12. Chrysemys 1075 (8.4; MC). Frontal sections through tlie nasal area. A, section 7.3 ; B, 6 sections ventral to A ; and G, 10 ventral to B. 40x. Figure 13. Chrysemys 1074 (8.6; MC). Sagittal sections through the nasal area. A, section 121 ; and B, 11 sections medial to A. 40x. l^ARSONS: NASAL EMBRYOLOGY 123 The pit is deepest in the antei-ior half as is shown most ch^arly in sagittal section (see Fig. .13). The posterior Avail is now nearly vertical, bnt in the more advanced embryos of this stage the naris continues anteriorly as a shallow, almost directly dorsally di- rected groove which runs a short distance anterior to the deej) l)art of the pit. Anteriorly, the lumen turns laterally at the naris, thus forming a concavity in the ventromedian wall. This is shown best in Figures 12B and C. It lies at approximately the place where a Jacobson's organ might be expected to develop. There are, hoAvever, no signs of any differentiation of this area, and were a concavitj' not specifically looked for here on the basis of the nasal embryology of other forms, it Avould hardly be con- sidered worthy of any note. The position of the naris on the snout is the same as in earlier stages, although due to an increase in the mesodermal tissue of the head, it now reaches anterior to the brain. The differential thickening of the epithelium in different parts of the nose has become very marked, as is shown in the several figures of this stage. In general the Avails are very thick an- teriorly and dorsally, and thin in the entire posterior third and ventrally. They are thicker medially than laterally, and the much thickened part reaches farther ventrally on the median Avail. It usually becomes thinner in the region of the concavity mentioned above so that the possible Jacobson's organ has Avails of very variable thickness in different embryos — a condition markedly different from that in forms Avhere such an organ is typically dcA'cloped. IIoAveA^er, in some of the series there does appear to be some definite thickening of the epithelium of this region. The histology shows fcAV changes. The basement membrane of the nasal epithelium is not at all clear dorsally, but the nose appears not to touch the AA'all of the brain at any point, although they are in A'ery close proximity. Near the apex of the pit the distribution of nuclei and mitotic figures previously described is still found, but farther ventrally there are fcAV mitoses visible and the nuclei seem rather evenly distributed in the epithelium or even concentrated toAvards the lumen. There is still no sharj) sei)aration of nasal and general body epithelium. Besides the Chrysemys material, a fcAv other turtle embryos of this stage Avere examined. An embryo of Chelydra shoAved almost 124 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY no differences from the descriptions already given. The only noteworthy point is that in this embryo the ventromedian con- cavity has (juite thick walls. Two Trionyx embryos also possess the same pattern. Some details show slight differences from the Chrysemys studied, but these are not basic or even obvious. Most of the papers dealing with the early embryology of the nasal region figure this stage. However, for the most part they show very little. Seydel (1896) shows a Chrysemys, sliglitly more advanced than any specimen available for the i)resent study, in which the still single naris has become mucli constricted and is almost closed near its anterior end, thus separating the naris externus from the primitive choana. The only other dif- ference from the embryos described above is that tlie ventro- median indentation is more pronounced than in any of the material used in the present study, or than figured by other workers except Loew (1956), although it is still far less con- spicuous than the Jacobson's organ of a snake or lizard of a similar stage. Thater's (1910) figures of a Chrysemys are too diagrannnatic to show more than that it is of tliis stage, and Fuchs' (1907) Emys agrees with the Chrysemys examined ex- cept that the nasal pits seem to lean farther laterally in his ma- terial. The external appearance in Eretmochelys is clearly shown by Voeltzkow (1903), whose figure 4a is reproduced in Figure 14. This needs no comment except to call attention to the prom- inent nasal processes which in section merely apiiear as the lateral and medial walls of the nasal pit. Other papers showing the external appearance at this stage are Dohrer (1912, Chely- (Ira), Parker (1880, Chelonia), and Deraniyagala (1932 and 1939, Bermochelys) . Thus the nasal pits of all turtles so far studied, and very probably of all cryptodires at least, are ex- ceedingly similar at this stage. Unfortunately, there were no series available for the present study which show the actual fusion of the medial and lateral nasal processes separating the nares externi from the primitive choanae. X'oeltzkow (1903) summarizes the process in Eretmo- chelys as follows (p. 187) : "Wie wir sahen, erfolgte der Ver- sehluss der Nasenrinne durch Aneinanderlegen des lateralen und medialen Nasenfortsatzes, wobei jedocli eine primitive Choana I'ARSONS: NASAL EMBRYOLOGY 125 vora Verschhiss ausfieschlossen blicb. Durcli Verwaclisiinjj: bcidcr Portsatze and Vei"S(*hinp]znn<>' (]vv hindcacwcliijien (T]-un(]la. Sagittnl seetioiis tliroiiiili the iias:il area. A, section 80; B, S sections medial to A; and C, 10 medial to I>. 18x. Figure 19. Chnjsf mvN 1079 (9.(i; MC). Transverse sections through the nasal area. A, section 240; B, 7 sections posterior to A ; C, 10 jiosterior to B ; D, 6 posterior to C ; and E, 5 posterior to D. 18x. Figure 20. Chryscmys 1080 (9.6; MC). Frontal sections through the nasal area. A, section 80; B, 17 sections vential to A; and C, 22 ventral to B. 18x. ventrolateral portions, but there are no sharp boundaries be- tween the regions. HoAvever, the thicker portion appears to be well innervated, while no branches of the olfactory nerve could be traced to the thinner parts, and tlie former shows a frreater 128 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY concentration of mitotic figures near the lumen. There is a slight hint of a thinner line approximately two-thirds of the way dorsal on the medial wall which could indicate the future line of separation between the regiones olfactoria and intermedialis, but since this division is so vague and is absent in some older embryos, this seems unlikely. There is as yet no distinct vestibu- lum or ductus nasopharyngeus. The next group of embryos shows a very similar condition but with the line of fusion of the nasal processes no longer marked by a band of epithelial cells, although in some embryos traces of one may still remain. Figures 17 and 18 illustrate this stage. The naris externus is at the anterior tip of the snout and leads into a short and rather indistinct vestibulum. This chamber has a distinctly oval lumen in contrast to later stages in which it becomes more nearly circular. Histologically, the epithelium of the vestibulum shows something of a transition between the nasal epithelium and the epidermis, although it resembles the former more closely in being columnar rather than cuboidal. The cavum nasi proprium remains essentially as in the preced- ing stage. The dorsal half, the presumptive regio olfactoria, possesses thicker walls and is somewhat lateral to the more ventral anlage of the regio intermedialis, but the separation between the regiones is not well marked. Such a division is most clearly seen in sagittal section (see Fig. 18B). The epithelium is generally slightly thicker than before, but otherwise unchanged. A ductus nasopharyngeus is not yet present; the cavum nasi proprium opens directly into the mouth cavity by a slitlike choana along the posterior half of its ventral margin. The lumen is, at this time, wider here than in any other part of the nasal cavity. At this stage the small lateral groove along the choana reaches its greatest development. The next stage is represented by fourteen embryos of Chry- semys. During the period covered by these series many changes occur, but the order in Avhich they take place is not constant, so that further division into shorter stages is impractical. The description, therefore, is general rather than based on a few selected embryos. Figures 19, 20, and 21 illustrate this stage. The vestibulum is more distinct than previously, but still tends to merge gradually with both the cavum nasi proprium and the PARSONS : NASAL EMBRYOLOGY 329 external covering of the snout. It may be either circular or oval in transverse section. Throughout this stage it increases in length to the condition shown in Figure 21C. The presence of a moderatel.y large lumen differentiates this from the following stages in which the vestibulum becomes virtually solid. As would be expected from the shape of the snout, the naris externus is the most uKHlial part of the nose. The cHvum nasi proprium is becoming considerably mure com- plex. In most of the series the regiones olfactoria and inter- medialis can be distinguished, although this is not universally true. The lateral border is marked by the transition from tiic thicker epitlielium dorsally to thinner ventrally ; at the end of CNP — NE Figiue 21. Chrysemys 1085 (9.0; MC). Sagittal sections through the nasal area. A, section 91 ; B, 10 sections medial to A; and C, 8 medial to B. 23x. this stage the transition is quite abrupt. Medially, the entire wall of the cavum is thick, but a line of demarcation is starting to develop between the regiones. Most commonly this is simply a horizontal line of thinner epithelium appearing as a groove along the basal surface of the epithelium. Occasionally there are two such grooves wdth a third parallel groove between them in the superficial edge of the epithelium (see Fig. 20B). There is also a developing tendency for the lumen to be wider in the regio intermedialis than further dorsally. The regio olfactoria reaches farther ventrally on the lateral than the medial wall. The cavum is in general closer to the midline anteriorly than posteriorly. This is most marked dorsally where the lumina of 130 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY the reg'iones olfaetoriae of opposite sides are at approximately rioJit angles to each other; ventral to the nares externi the two sides are nearly parallel. Seen in transverse section the regio interniedialis is almost vertical. At the start of this stage the i-egio olfactoria appears to lean slightly laterally, but by its end this is true only posteriorly. Anteriorh% the dorsal part of the cavum becomes slightly median to the ventral. In frontal sections the posterior part of the regio olfactoria can l)e seen to turn laterally, thus producing- a convexity in its lateral vail ; this is the tirst appearance of the Muschelwulst. Histologically, there is very little regional ditt'erentiation visi- ble in the available material, aside from the thickness of the epi- thelium. In some of the oldest embryos of this stage the thicker and presumably sensory portions show a clear cytoplasmic zone peripherally Avhicli has very few nuclei, while in the thinner portions, including the vestibulinn and ductus nasopharyngeus, the nuclei are evenly distributed throughout the epithelial layer. However, this distinction is not marked. The pattern of thicken- ing is Avell shown in the figures. Witli the continued posterior growth uf the primary palate, the choanae become more restricted, and a short ductus naso- pharyngeus is formed (see Fig. 21B). This possesses a very wide lumen. The ductus runs slightly medially as well as ven- trally from the cavum, l)ut the choanae, which open into shallow grooves on the palate, are widely separated. The next stage (see Fig. 22) displays few but important changes. The A'estibulum is longer and generally quite circular. Figure 22. Oirytuviyn 2136 (8.S; 'S\V ) . Tiaiiswrse sections tliioutjli tlu' nasal area. A, section 346; B, 10 srt-tions i)osterioi' to .\ ; (', 20 jiostevioi to B; D, 16 posterior to C; E, 20 posterior to D; F, 11 posteiior to E; ;iiiil G, 12 posterior to F. 22.\. Figure 23. Chryscmys 1-172 (10.8; MC). Transverse sections through the nasal area. A, section 313; B, 16 sections posterior to A ; C, 12 posterior to B; D, 9 posterior to C; E, 6 posterior to D; and F, 8 posterior to E. 22x. PARSONS: NASAL EMBRYOLOGY 131 132 BIXLETIN : MUSEUM OF COMPARATIVE ZOOLOGY The most conspicuous development is that its Avails have greatly tliickened so that its lumen is much reduced in diameter. The cavum nasi jiroprium is exactly the same shape as described aljove, although the lumen is generally narrower and the regio olfactoria proportionately larger. However, the regio intermedi- DNP RDN 25 F RDN Figure '2i. (lirij.sciinjs 1473 (9.8; MC). Frontal sections through the nasal area. A, section ]83; B, 16 sections ventral to A; C, 18 ventral to r. : T), ." ventral to C; and E, (i ventral to D. 21x. Figure 2"i. Cltry.sc)nys 2141 (10.2; MC). Transverse sections through the nasal area. A, section 452; B, 17 sections posterior to A; C, 13 posterior to B; D, 20 posterior to C; E, 32 posterior to D; F, 24 posterior to E; and G, 24 posterior to F. 13x. alls noAv hears thickened epithelium not only medially, but also veutrally and ventrolaterally. This thickening appears to pro- ceed progressively around the ventral margin from the medial wall. Posteriorly there is far less of the thick epithelium on the PARSONS: NASAL EMBRYOLOGY 133 lateral wall than there is anteriorly. Tlie cii-cular ductus naso- |)liaryn<;eus is considerably lonjier tlian i)i-evious]y so that the clioanae are noM" completely i)()sterior to the cavnin nasi pro- prium. The following stage (see Pigs. 23 and 24) is unfortunately represented by only three embryos. The vestibulum is essentially as in the preceding stage, but the lumen is even smaller; in at least one series it appears to be completely solid for a short distance. The cavum nasi })roprium is also relatively unchanged. The regio olfactoria is generally slightly medial to the more ventral regio intermedialis, but the cavum is very close to vertical. The cava of opposite sides are now parallel. The Muschelwulst is highly developed as can be seen in Figure 24B. Surprisingly, the older embryos of this group seem to have less of the thickened epithelium ventrolaterally than in the preceding stage. There is, however, still some anteriorly. Histologically, the epithelium appears uniform except for the previously mentioned tendency for a i)eripheral zone lacking nuclei in the thickened areas ; such zonation is not found in the areas of thinner epithelium. Figure 75 illustrates this feature. Posteroventrally the cavum becomes ver}^ narrow in the region of its junction with the ductus naso- pharyngeus. It is in the ductus that the greatest changes have occurred. The lumen is large and circular except at its anterior or nasal end where it connects with the cavum by a thin slit along its medial edge. The anterolateral end continues slightly anteriorly as a small blind pocket best shown in Figure 24D. This is the first stage in the development of the recessus ducti nasopharyngei. The next stage, as shown in Figures 25 and 26, displays in- ci-eased complexity in all regions, but most especially in the vestibulum. This is a quite long tubular structure with the anterior three-fourths circular in transverse section. Posteriorly, it l)ecomes oval with the long axis vertical. The walls are very thick, and the lumen very restricted or completely obliterated. The histological character of the epithelium now reaches the condition which persists throughout the later development. There is an outer layer or two of columnar cells surrounding a thicker zone of cuboidal cells Avhich gradually merges with the 134 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Figure 26. Chryfsemys 1088 (11.0; MC). Sagittal sections through the nasal area. A, section 180; B, 6 sections medial to A; C, 6 medial to B; and D, 5 medial to C. 14x. Figure 27. Chryseinys 1092 (1(3.7; MC). Transverse sections through the nasal area. A, section 30; B, 13 sections posterior to A; C, 7 posterior to B; D, 11 posterior to C; E, 11 posterior to D; F, 13 posterior to E; G, 16 posterior to F; H, 17 posterior to G; and I, 11 posterior to H. 22x. PARSONS: NASAL EMBRYOLOGY ]35 stratified s(inamous epithelium actually lining- and often filling the lumen. Such a pattern is clear in Figure 76 although the embrj^o figured is considerably more advanced. The nares externi are somewhat dorsal on the snout and very close to the midline. More important are the appearances of two outgrowths of the vest ibul urn. The first to develop is the anlage of the glandula uasalis externa. This arises as a short solid prong growing dorso- laterally and slightly posteriori}^ from the dorsolateral surface of the vestibulum, very near its posterior end (see Fig. 25B). B}- the end of this stage there may be signs of a lumen within the prong, but this is variable. The subsecjuent development of the gland is described separately. From the same level or slightly further posteriorly another solid process may be seen medially on the dorsal edge of the vestibulum in the later embryos of this stage. This structure is found in all the older series, but has apparently not been previously described. Neither its ultimate fate nor its function are known; it is here termed the "dorsal recess of the vestibulum." Within the cavum nasi proprium the changes are slightly less marked but important. The regiones olfactoria and inter- medialis are divided as described above. Anteriorly, the former regio reaches farther ventrally on the lateral than on the medial wall, but posteriorly the reverse is true. The regio olfactoria is in general vertical, with the dorsal part somewhat medially directed near the anterior end. The more ventral regio inter- medialis is also vertical in the anterior two-thirds of the nose, although with a strong medial twist at its ventral end in the anterior one-third. Posteriorly-, this regio is more laterally directed, as shown in Figure 25. Thus the regio olfactoria is mostly medial to the intermedialis. The Muschelwulst is well developed. The cavum has a narrow lumen at this stage. Histologically, there is little to add to the earlier descriptions. The regio intermedialis tends to have slightly thinner epithelium than the regio olfactoria, but this distinction is not always shown. The regio intermedialis possesses much thickened epi- thelium medially and, in the anterior two-thirds, ventrally and laterally as w-ell. Although the anteromedial curve gives some indication of their future development, the sulci are not yet differentiated. 136 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY The ductus nasopharyngeiis has a very small lumen anteriorly, but widens into a large circular tube posterior to the entrance of the recessus ducti nasopharyngei. The recessus is larger than in the preceding stage, but its walls are thicker so that the lumen is but little increased. It lies lateral and slightly dorsal to the anterior end of the ductus. Posteriorly, the ductus again nar- rows and becomes a small horizontal oval. The choanae are now closer to the midline, and those of opposite sides open into a single median groove in the palate by the end of this stag'e. The embryos of the next stage start to assume the adult form with the development of the complex pattern of sulci in the regio intermedialis. Figures 27 and 28 illustrate this stage. The vestibulum shows further differentiation with the de- velopment of the gland ula nasalis medialh. It appears as a solid rod growing posteroventrally from the medial surface of the vestibulum, just anterior to the border of the regio inter- medialis. It is further discussed in the section on nasal glands. The dorsal recess of the vestibulum is somewhat larger and, in the oldest embryos of this stage, is beginning to acquire a lumen. In frontal section it is oval as shown in Figure 28C. There are no visible histological changes, and the vestibulum remains essen- tiallv a thick-walled horizontal tube. Figure 28. Chrysemys 1090 (16.8; MC). Frontal sections through the nasal area. A, section 204; B, 16 sections ventral to A ; C, 20 ventral to B; D, 16 ventral to C; and E, 10 ventral to D. 22x. Figure 29. Chrysemys 1096 (27.0; MC). Transverse sections through the nasal area. A, section 52; B, 27 sections posterior to A ; C, 23 posterior to B; D, 28 posterior to C; E, 20 posterior to D; and F, 29 posterior to E. 17x. PARSONS: NASAL EMBRYOLOGY 137 *^ 5^ ^^ ..1 RO 138 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY The regie olfactoria is quite simple. Its walls are everywhere very thick with a sharp boundary to the thinner epithelium of the regio intermedialis, except posteroventrally where they thin gradually. The lumen is wider than in the precedino- stage, especially anteriorly. The regio olfactoria is virtually' vertical, but posterolaterall.y there is a very small Muschelwulst so that this regio is slightly concave laterally in either transverse or frontal sections. The posterior ends are closer to the midline than the anterior. Grenzfalten are not yet formed, although there are faint signs of the medial one. The regio intermedialis now possesses all four sulci. These are present as narrow grooves rather than as the adult condition of shallow troughs separated by ridges. Histologically, the epi- thelium shows no differentiation. It is all thinner than that of the regio olfactoria, most markedly so along the dorsal margins where the Grenzfalten will develop. There is fre(]uently a clear peripheral zone lacking nuclei in the thinnest portions by this stage, so that such a zone can no longer be used as a criterion for presumptive sensory areas. The shape of this regio as a whole can be seen in the figures. The regio is most medial anteriorly where the vestibulum enters its anterodorsal wall. In this region it is J-shaped with the hook directed medially. Posteriorly, it soon becomes vertical, and then the ventral end is turned away from the midline. The posterior ends of either side converge towards the midline so that in frontal section it appears concave medially. Throughout most of its length it is lateral to the regio olfactoria. The regio inter- medialis slopes posteroventrally at an angle of approximately twenty-five degrees to the horizontal. The pattern of the sulci is best shown in Figui-e 27. The sulcus anterior is ventromedian in the anterior quarter of the cavum, thus forming the hook of the J mentioned above. Starting at its posterior end and imperfectl.y separated from it, the sulcus ventralis runs along the ventral margin of the regio to its junction with the ductus nasopharyngeus. The smaller sulcus lateralis is prominent in the anterior three-quarters of the cavity, while the slightly more ventral sulcus medialis is present only in the posterior third of the regio intermedialis where it reaches to the start of the ductus. PARSONS: NASAL EMBRYOLOGY 139 The ductus nasopharyng-eus shows fewer changes. It is very narrow anteriorly, but increases in diameter posterior to the recessus. The latter is now dorsolateral and larger, with an extensive lumen. Posteriorly, the ducti converge slightly to enter a medial groove in the palate. There is only a very slight pos- Figure :{0. Chry.s-emys 1098 (28.1; MC). Frontal sections through the nasal area. A, section 133; B, 26 sections ventral to A; C, 11 ventral to B; and D, 29 ventral to C. 15x. Figure 31. Chryscmys 16.")3 (32.0; MC). Transverse sections through the nasal area. A, section 0-5 ; B, 14 sections posterior to A ; C, 9 posterior to B ; and D, 19 posterior to C. 13x. terior thinning at this stage and the caudal portion of the ductus is triangular (see Fig. 27H) rather than oval as in previous stages. A final stage is represented by nine Chryscmys embryos. The pattern is very close to that of the adult, although in general the 140 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY various ridges are less marked in the embryos. Figure 29 shows transverse sections of an embryo of this stage, and Figure 31 the cavum nasi proprium of a slightly older one. Frontal and sagittal sections are shown in Figures 30 and 32. Gf.'M Figure 32. Chrysemys 1099 (i^Ci.-l; MC). Sagittal soetions through the nasal area. A, section 520; B, 11 sections medial to A; and (", 0 medial to P>. 15x. The vestii)ulum is very like that in tiic j)reeeding stage. Anteriorly it is a long circular tube with very thick walls and a small lumen which is i)robal)]y not continuous in most of the series. Histologically, its epithelium still shows a transition from columnar cells basally to s(|uamous cells centrally as shown in Figure 76. Posteriorly, the vcstibulnm is, in transverse section. PARSONS: NASAL EMBRYOLOGY 141 uval with the loiiji' axis vertical. Here the epithelium shows a transition to that of the eavum nasi i)ropriuni — the term ves- tibnlnm is tluis used in a topo<'i-aphie sense without im])lyiniven previously. Seydel's Chrifscnn/s embryo IV ((•arai)ac'e 4 mm.) lias a some- what more prominent Muschelwulst than -was seim in any of the material studied foi- the present work, hut Ihc differenee is slifi'ht. His older Emys embryos ajiree tdosely with those already de- scribed, except that he shows tlie sulcus antei'ior (the Pars anterior des Jacobsou'sehes Organs of Seydel) as more restricted, with the sulcus ventralis reaching farther anteriorly. Loew (1956) also describes a series of Emys embryos in a very recent paper; since his figures are very diagrannuatic and his text is concerned primarily with the problem of Jacobson's organ, it adds comparatively little here. In general his account agrees with that already given and with fSeydel's, but there is one very important differenee. Loew calls attention to the early develop- ment of a ventromedian gi-oove which he considers to be Jacob- son's organ. This is present in all of liis embryos in which the nasal processes are completely fused, and seems to correspond to the medial part of the sulcus anterior and the sulcus medialis of the present work. However, in the earliest stages in which he describes it, these are not distinct, and his groove appears to be merely the ventromedian corner of the regio intermedialis. Four other papers figur(\ but do not discuss in any detail, the nasiU cavities of emydine emln-yos. Fuchs (1907 and 1915) studied Emj/s in connection with his work on the palate, figuring two early stages. Kunkel (1912) in describing the eml)ryology of tlie skull has drawings of two later Enu/s embryos. Finally, Thiitcr (1910) gives highly diagrannnatic figures of Cliryseniy.s embryos. None of these accounts adds any information to that already given. The only other turth's whose nasal eml)ryology has even been figured are the Cheloniidae ; no descriptions are available. Parker (1880) shows several sections tiirough the nasal area in his paper on the skull of Chrlonin, and Fuchs (1907, 1911, and 1915) figures Eref))iocheh/s in his discussions of the p^date. The total information which can l)e gained from these sources is, however, far too meager to permit any description of the various stages or comparisons wit I) the emydines. Jacohsoii's Oryaii. Thei-e has been much disagreement in the literature concerning the nature of tliis organ in turtles. In the 146 RULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY present discussion the various theories are presented, and then considered in the light of the observations made in the present study. Due to the marked differences in their nasal anatomy, the literature on the marine turtles is separately described. The first paper mentioning' Jacobson's organ in turtles is Seydel's 1896 monograph, which remains the most detailed and influential account of the nose in this order. His thesis is that all of the sensorj' epithelium of the regio intermedialis (the pars respiratoria of his paper) represents pJacobson's organ. A large l)art of his paper is devoted to an attempt to prove the homology of this sensory area with the supposed Jacobson's organ in the amphibians which lie had previously studied. However, since the exact homologies in that group are also disputed, his argu- ments concerning them need not be given here. Other evidence which he presents in support of his theory includes the innervation of his Jacobson 's organ by branches of the medial division of the olfactory nerve and the absence of Bowman's glands throughout the regio intermedialis. Seydel recognized that, according to this theory, Jacobson's organ would essentially equal the entire ventral part of the nose, rather than being restricted to a pocket in the medial wall as in the other araniotes. However, he puts forward two points which he believes should eliminate this as a problem. First, in Testudo the sensory epithelium of the regio intermedialis is restricted to the medial wall, and he assumes that this is the primitive condition in turtles. Second, in Chrysetnys embryos the first areas of thickened and presumptive sensory epithelium, which he observed ventrally, were medial, and only in later stages did they spread to the ventrolateral wall. The latter fact considered with the innervation from the medial nerve trunk strongly suggests a primitive medial position for this sensory area. Seydel's theory received strong support from McCotter (1917), who studied the nervous connections of the regio inter- medialis in " Chrysemys punciata." McCotter describes a dis- tinct accessory olfactory bulb receiving the fibers of the medial olfactory trunk. This bulb he considers to be the homolog of the accessory bulb in other amniotes in which it receives the nerve fibers from Jacobson's organ. The evidence presented in these two papers has been thought convincing by man}- later workers, PARSONS: NASAL EMBRYOLOGY 147 and Spydel's theory that tho entiro sensory area of the reoio intermedialis represents Jaeobson's organ is accepted in many general works including the important review by Matthes (1934). The next theory to be put forward is that Jaeobson's organ is present as a small tubular structure in the septal wall of the nose, and has at its deep end a prominent gland. This is the structure considered by Seydel and other workers to be the duct of the glandula nasalis medialis. Such a theory was first proposed by von Mihalkovics (1898), who calls attention to the gross simi- larity between this duct in Emys and the Jaeobson's organ of some mammals such as the mouse, which he also studied in some detail. He further states that a ])ortion of the duct is lined by sensory olfactory epithelium and is strongly innervated bj^ the medial trunk of the olfactory nerve. The only other worker to accept this theory is von Navratil (1926) who agrees with von Mihalkovics in all respects. Other investigators, such as Zucker- kandl (1910a) who also studied Emys, were unable to find any sensory epithelium in the duct, and thus could not consider it to be Jaeobson's organ. The most recent theory on the nature of this organ in turtles is that of Loew (1956), who studied a series of Emys embryos. In these he finds a rather small but distinct horizontal groove in the ventromedial wall of the nasal cavity — apparently the medial part of the sulcus anterior plus the sulcus medialis in the terminology used in the present paper. This groove first appears shortly after the fusion of the nasal processes has separated the naris externus from the choana. Although Loew notes that this is a relatively advanced stage for the appearance of Jaeobson's organ, he considers it quite similar in general form to the anlage of that organ in Natrix, which he used for comparison. The groove remains relatively constant in the later development of Emys and is present in the adult. In denying that the lateral portion of the regio intermedialis is also part of Jaeobson's organ, he states that only the medial wall of the nose is innervated by the medial division of the olfactory nerve. Thus he flatly con- tradicts the main evidence that Seydel put forward to support his theory. The other workers who have studied the nose in turtles have generally refused to commit themselves on the problem of Jaeobson's organ, although some tend towards the belief that it 148 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY is completely absent in this order. The latter point of view enjoys a popularity eq\va\ to that of Seydel's theory, in p^eneral textbooks and reviews. The first major work whieli casts donbr on l)otli theories previ- ously proposed is Peter's (1901) chapter in Hertwio's Handhuch. Although he did not actually study any turtles, his paper is so widely cited and important that it must lie considered. Seydel's theory is discussed, l)ut, in the absence of any transitional forms, Peter could not consider as probable the homology of the Jacobson's organ of lizards and snakes with a structure so radically different in pattern as the regio intermedialis of turtles. The relatively late embi-yonic development of Seydel's Jacobson's organ is also considered a major objection to such a homology. Von Milialkovics' theory is mentioned, but. in the absence of any embryological evidence, Peter refuses to speculate on its validity. Zuckerkandl (1910a) briefly describes the relationsliip of the olfactory nerve to the various parts of the nose in E)n\is. lie definitely denies von Milialkovics" conclusions, as stated above, l)ut although lie reviews Seydel's theory and states that he was able to confirm that author's anatomical findings, he nowhere states any opinion on the homology of Jacobson's organ. In another paper Zuckerkandl (1910b) again reviews the previous literature, and concludes that further study of a greater number of forms is necessary for any solution to tliis problem. Nick (1912), in describing the nose of Chclydra. states that he considers Seydel's identification of Jacobson's organ probably correct if turtles have any sucli organ at all. However, he refuses to commit himself further on this point. In his study of Pcloinedusa, van der Merwe (1940) could not find any structur(> which he would consider to represent Jacob- sou s organ. Finally the literature on sea turtles must l)e noted. The first paper to consider Jacobson's organ in this group is Seydel's (1896). His conclusions arc based entirely on the study of published figures; he had no actual material. It is his l)elief that only the i)oster()dorsal recess of the nasal cavity represents the regio olfactoria and that all the remaindei' of the cavum nasi |)roprium is the regio intermedialis. This division has found general acceptance and is used in the jiresent paper. Thus, by PARSONS: NASAL EMBRYOLOGY 149 Seydel's theory, any sensory epithelium to be found in tlic recessi dorsalis or ventralis or the areas between them wouhl constitute Jacobson's organ. He was, of course, unable to tell which areas were actually sensory. Nick (1912) studied both Chdonia and Deniiochelijs. Al- though he mentions Jacobson's organ onlj- in connection with the former, he considers them (luite similar in nasal anatomy, so presumably his comments would apply to botli genera. He found that the sensory epithelium of the regio intermedial is is restricted to the various recesses, but hesitates to give any opinion on the homologies of the area. He states only that if any Jacobson's organ is present, it is either the total sensory area of the regio intermedialis as postulated by Seydel, or else it is found in the medial part of the recessus dorsalis (the recessus medialis). The latter possibility is suggested by the gross form and topographical position of that recessus. The last paper on sea turtles is that of Fuchs (1915) which is based on Eretmochchjs and contains an extensive discussion of the problem. First, he gives four criteria which he considers essential for a Jacobson 's organ .- independence from the nasal epithelium, connection with the mouth rather than the nose in the adult, innervation ])y the medial division of the olfactory nerve, and a medial and ventral position. Clearly, turtles possess no organ fulfilling these requirements. However, on the medial wall of the recessus ventralis there is ventromedian sensory epi- thelium innervated by the medial trunk of the olfactory nerve. This Fuchs considers the probable homolog of Jacobson's organ, and for it he suggests the term pars vomeronasalis. The other sensory areas of the regio intermedialis, including the recessus medialis of Nick, are tentatively rejected as pi-obable homologs of that organ on topographical grounds despite his own demonstra- tion that all the sensory areas of this regio are medially inner- vated. In so doing Fuchs points out that it is probably incorrect to assume that the entire medial division of the olfactory nerve innervates only Jacobson's organ in other reptiles, so that there is no reason to assume that it does in this group. However, he also admits that, in the absence of any embryological data, these ideas remain speculative. Finally, he l)riefly comments on the phylogenetic importance of his ideas concerning Jacobson's organ 150 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY in turtles, preferring to consider their condition primitive rather than derived from some form already possessing- a distinct Jacob - son's organ. The observations previously described in this paper and those in the subsequent sections on the nasal glands and nerves shed considerable light on certain aspects of this problem. First, it now seems possible to deny the validitj^ of von Mihalkovics' theory on several grounds. In none of the available material is there any evidence of a sensory area in the duct of the glandula nasalis medialis ; rather, its epithelium appears uniform through- out, as reported by Zuckerkandl (1910a). Equally important is the veiy late embryonic appearance of the gland, which first forms as a solid rod of cells exactly similar to the anlage of the glandula nasalis externa, and totally unike that of Jacobson's organ in any other group. The evidence of the innervation is less clear. The gland itself is definitely supplied by fibers from the trigeminal nerve ; however, it is impossible to show^ that no fibers from the olfactory nerve reach it. Even ^^^th the innerva- tion questionable, the embryological argument appears sufficient to eliminate the duct as a possible Jacobson's organ. Loew's theory seems almost as improbable as von Mihalkovics'. In the material used in the present study, the sulci anterior and medialis do not appear as early as w^as reported by him, and the thickened epithelium in which they form appears to be part of the single area which gives rise to all the sulci. Besides these embryological considerations, the innervation of the area does not bear out his assertions. The present work confirms the de- scriptions of Seydel (1896), Zuckerkandl (1910a), and McCotter (1917) rather than that of Loew in that all the sensory epi- thelium of the regio intermedialis receives fibers from the medial division of the olfactory nerve. Thus, this entire sensory area would appear to be a single unit rather than at least two distinct sensory regions as postulated by Loew. That leaves only Seydel's theory. As mentioned in the pre- ceding paragraph, his evidence from the pattern of innervation is confirmed by the present study. Even more important, the relationship to a distinct accessory olfactory bulb, first reported by McCotter (1917), is also confirmed. It is this neurological evidence that has convinced such workers as Matthes (1934) of the correctness of Seydel's identification. To deny its validity, PARSON'S: NASAL EMBRYOLOGY 151 it would appear necessary to deny the true homology of the accessory olfactory bulb of turtles with that of other forms, and to consider them independent developments, a step no one has advocated. Seydel's evidence from the absence of Bowman's glands remains unchallenged. In all tetrapods, except certain neotenous urodeles and possibly some aquatic snakes, such glands are present in the sensory olfactory epithelium ; they are absent in Jacobson's organ, the regio intermedialis of turtles, and the various ventral sensory areas in modern Amphibia. Next, what are the main arguments against Seydel's theory? The basic one is that the adult form and position of his Jacobson's organ in turtles are radically different from those in other forms possessing a typical Jacobson 's organ — that is, snakes, lizards, and most mammals. However, neither form nor position can be considered necessarily accurate guides to liomology. Further- more, embryonically the epithelium of the regio intermedialis first becomes thickened on the medial wall, hence in the position where Jacobson's organ develops in other groups. Fuchs (1915) states that a connection with the mouth by a duct is characteristic of that organ. This is true of most forms, but in some mammals it retains its primitive connection with the nasal cavities, thus making an oral connection invalid as a criterion. Fuchs also cites independence of Jacobson's organ from its parent tissue: "diese Selbstiindigkeit ist morphologisch eigentlich ja im Begriffe 'Organ' gefordert. Denn was zu seinem Mutterboden in Un- selbstandigkeit beharrt, ist eben kein Organ ftir sich, sondern nur Teil eines Organes" (p. 169). This rather semantic argument seems stronger, but it attacks only the use of the name Jacobson 's organ, not the acceptance of the homology as proposed by Seydel. It seems, therefore, that the sensory epithelium of the regio intermedialis of turtles is indeed to be considered homologous with the Jacobson 's organ of other amniotes, but in turtles it does not assume the definitive form. Hence the term is best not used in describing tlie nasal cavities of members of this order. Fuchs (1915) proposed the term pars vomeronasalis for the area of sens- ory epithelium which he believed to represent Jacobson's organ. This term could be useful if taken in a histological sense as opposed to pars olfactoria and pars respiratoria, the last two referring to normal olfactory epithelium and non-sensory epi- 152 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY theiium, respectively, as is the case in inammalian anatomy. Plowever, since in turtles the partes have generally been used in a topographical sense, as by Seydel, this usage would now be most confusing. The sensory areas of the regio intermedialis are in discontinuous patches of variable pattern so that any term would necessarily be rather vague with histological but no morphologi- cal significance. Such a term would, at best, have questional)le value, and therefore none is proposed here. The acceptance of the homology advocated in the preceding ]>aragraph is of great phylogenetic importance. Two alternate hypotheses seem possible. First, it is possible that primitive tetrapods had considerable areas of sensory ei)ithelium in the ventral half of their nasal cavities, and that the pattern of restricting this to a small area on the medial wall (which embryo- logically forms a distinct pocket, the definitive Jacobson's organ) did not arise until after the turtles were already established as a group. The other possibility is that turtles arose from forms having a typical Jacobson's organ, but during their evolution modified its development greatly to reach their present condition. Consideration of these alternatives is deferred to the general discussion on phylogeny. Order Rhynchocephalia Adult Anaiomi). No adult specimens of Sphoiodon were avail- able for the present study so the following review is based entirely on the literature. The earliest careful account of the nose of this genus is that of Osawa (1898). Later workers, in- c-luding Broom (1906), Fuchs (1908), Hoppe (1934), and Malan (1946), studied very late embryonic material which shows essentially adult conditions. Finally, Pratt (1948) very briefly describes the nasal area in his paper on lizards. The most detailed paper is that by Hoppe, and the following description is based jirimarily on his work. P^igure 33 shoAvs the lateral wall of the nasal cavity of an im- mature i^phenodon . The naris externus leads into a small vestibu- liim which is an essentially tubular chamber with its axis run- ning lateromedially. Its medial end is somewhat expanded and separated from the cavum nasi proprium by a slight fold in the PARSONS : NASAL EMBRYOLOGY 153 nasal wall. The ducts of the glandulae nasales externa and medi- alis both empty into the vestibulum on the posterior and ventral walls, respectively. In describing the complex cavum nasi proprium, Hoppe uses in part new terms and in j)art the terminology of Beecker (1903) and Fuchs (1908). The area anteroventral to the conchae which lies between the vestibulum and the choana is termed the AOR- CP- VM Figure 33. Medial view of the left nasal cavity of an immature Sphenodon to show the lateral wall (after Hoppe, 1934). Choanenga)i(j. It is a relatively simple, non-sensory zone and extends dorsally ventrolateral to the concha anterior as the dorsale Tasche des Choanengangs. Most of the remainder of the cavum consists of the complex Muschelzone containing the con- cliae. Posterior to this is the small Antorhitalraum. Iloppe distinguishes two conchae, the concha anterior and concha posterior (the vordere and hintere Muschcln). Both are 154 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY simple lamellar projections of the lateral nasal wall containing cartilaginous processes from the nasal capsule. The anterior runs posteroventrally from the posterodorsal corner of the vestib- ulum, thus forming a diagonal ridge across the anterior half of the cavum nasi proprium. It projects medially and ventrally into the nasal cavity. The posterior end of the concha anterior turns dorsally and is continuous with the anterior end of the concha posterior. From there the latter structure extends posteriorly and somewhat dorsally nearly to the posterior end of the nasal cavity. Its free end is slightly inrolled so that in section the concha appears J-shaped. The space dorsal and lateral to the concha posterior is termed the dorso-lateraler Muschelraum by Hoppe ; it corresponds to the Sahter of Beecker (1903) and Fuchs (1908). Ventral to this concha there is another recess, the ventraler Muschelraum of Hoppe. This is considered a part of the Choangengang by Beecker and Fuchs. There is no palatal fusion posterior to the premaxillae in Splienodon and the choanae run almost the entire length of the cavum nasi proprium. They are widely open throughout their length. However, they appear to enter the mouth from the side rather than leading directly ventrally from the nasal cavity, due to the large Vomer polster (Busch, 1898) which forms a horizontal plate at the ventral end of the nasal septum. The lateral margin of the choana is marked by a prominent ridge extending medially from the boundary between the nasal and oral cavities. This Choanenfalte is ventral to the Vomerpolster. Jacobson's organ retains its primitive position, opening into the anterior end of the choana. In Sphcnoclon it is a tubular struc- ture which lies along the ventromedial wall of the nose. Accord- ing to Iloppe the anterior end of the organ turns laterally to enter the anteroventral end of the Choanengang on the lateral side in common with the lachrymal duct. The posterior end of Jacob - son's organ is slightly dorsal to the anterior. Only the dorsal wall bears thickened epithelium, and it lacks the mushroom body {= pilzenformige Wulst or concha of Jacobson's organ) found in the Squamata. In general, most of the other descriptions of the nose in Sphenodon agree closely with that given above. However, there PARSONS: NASAL EMBRYOLOGY 155 arc two points which require further comment. The first of these concerns the conchae. Osawa (1898), Fuchs (1908), and Pratt (194S) all mention a single concha. This certainly corresponds to Hoppe"s (1934) concha posterior; whether the concha anterior is considered part of this or is not considered to be a conchal formation is not clear from their papers. Malan (1946) calls attention to this (Utferent-e, but does not express any opinions on the problem. DeBeer (1937) states that Sphenodon lacks a con- cha completely, but he apparently restricts the term to out- pocket ings of the capsule while Sphenodon has only lamellar ])rojections from the surface of the capsule. CHG ;^LD Figure 34. Sphenodon 1491 (7.9; MC). Transverse section through the nasal area (section 198). 46x. Figure 35. Sphenodon 1507 (17.4; MC). Frontal sections through the nasal area. A, section 144; and B, 43 sections ventral to A. 16x. The second debated point concerns the connection of Jacobson's organ with the nasal cavity. The description above is based on Hoppe (1934). Previous worker.s, Osawa (1898), Broom (1906), and Fuchs (1908), had stated that the very short duct of Jacob- son's organ ran laterally from the center of the tubular organ to enter the medial wall of the anterior end of the choana. The opening of the lachrymal duct was reported to be just opposite or slightly posterior to it on the lateral wall. Hoppe claims that their opinions result from studying only transverse sections and that his models show the true situation. However, Malan (1946), Pratt (1948), and Bellairs and Boyd (1950) all corroborate the earlier workers. 156 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Early Emhryology. The early nasal embroyology of ^plicno- don is poorly known, but appears to resemble that of other reptiles very closely. For the present study two series sliowinp' nasal i)its were available. In both, the pits are moderately shal- low, but already dorsal!}' directed. The older of the two is shown in Figure 34. As in the other groui^s, the epithelium is thickest dorsally and medially, but no histological differentiation is vis- ible, and the nasal and epidermal epithelia are not distinctly separate. At this stage the pit possesses a wide lumen throughout its extent and its dorsal end leans slightly laterally. The naris forms a long, widely-open slit which is on the latei'al surface of the snout anteriorly, but becomes ventral posteriorly. There is a ventromedial band of thickened epithelium which is presum- ably the first stage in the development of Jacobson's organ; it is lacking in the younger series. At this stage there is still no real inpocketing for that organ. Only three papers mention such early stages. Schauinsland (1903) shows the nose in his figures of the general external ap- ])earance of several Splicnodon embryos, but gives no descrip- tions. The nasal placode is shown as a roughly circular depres- sion with the margins somewhat raised. This rim is first seen dorsally, but soon completely surrounds the nasal tissue. In slightly older embryos the placode has become a nasal pit, and the naris is elongated witli the posteroventral end pointed and the anterodorsal round. Iloppe (1934) considers one embryo which is slightly more advanced than the series described above. In his specimen the anlage of Jacobson's organ is more noticeably indented, but it is very small and the general appearance of the nasal pit is the same as in the younger embryo. The pit extends slightly anterior to the naris, thus forming a small apikale Bli}idsacl-. The pattern of epithelial thickening shows no changes, and the nasal epi- thelium is still not distinctly set off from the epidermis. Finally, Fuchs (1908) studied early stages of SphciiorJov, but states only that they closely resemble lizard embryos. Later Emhryology. The most detailed study of the later stages in the development of the nose of Sphotodon is that of Hoppe (1934) who constructed wax models of three embryos. Other pai)ers, such as Broom (1906) and Malan (1946), figure only PARSONS: NASAL EMBRYOLOGY 157 single very late stages and have already been considered. Fuchs (1908) deals with the embryology of the palate, but does not treat the general nasal development in any detail. The following description is based on Iloppe and a .study of the two older Sphenodon embryos available for the present work. lloppe's stage 2 embryo already has both conchae well formed. In this and in a slightly older series studied in the present investigation (see Figs. 35 and 80), they are parallel ridges in the lateral wall of the cavum nasi proprium wliicli run dorsally and somewhat posteriorly from the anteroventral corner of the cavum. As in the adult, the ventral ends of the two conchae are joined so that they form a narrow U with the slit-like dorso-lateraler Musclielraum between them. The ventraler Muschelraum and Antorbitalraum are only weakly developed, but the dorsale Tasche des Choanengangs is present. Iloppe shows the choana opening beneath only the anterior half of the nasal cavity ; in the series which I studied it extends nearly the entire length of the cavum. A more important point of disagreement concerns the connection between Jacobson's organ and the nose. In all of his models Hoppe shows the anterior end of that organ passing laterally to enter the anterolateral corner of the Choanengang in company with the lachrymal duct. This is not the case in tlie series shown in Figure 35B. There the center of Jacobson's organ opens into the anteromedial corner of the nasal cavity and tlie lachrymal duct into the anterolateral with the latter slightly the more posterior. In his two later eml)ryos Iloppe shows a gradual development towards the adult condition. The concha anterior becomes vertical and then continues to rotate so that its ventral end is posteroventral to the dorsal. This rotation, coupled with a proportional lowering of the posterodorsal end of the concha posterior, results in a great enlargement of the dorso-lateraler Muschelraum which becomes a wide cavity rather than a narrow .slit. The ventraler Muschelraum, Antorbitalraum, and dorsale Tasche des Choanengangs also show increases in size relative to the nasal cavity as a whole, and the choana is now figured as .stretching more nearly the length of the cavum nasi proprium. The conchae increase in length, but do not thicken in pro]iortion to the general growth of the area, .so that they appear lamellar 158 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY as in the adult, while in the younger embrj'os they are wide ridges in the nasal wall. An older series, which is shown in Figure 36, resembles closely Iloppe's stage 4 embryo. The vestibulum runs medially and slightly posteroventrallj^ from the naris externus to enter the cavum nasi proprium. As in other forms, the external end of -CNP '— CHG 36C Figure 36. (First half). Syhenodon 1490 (25.2; MC). Transverse sec- tions through the nasal area. A, section 43; B, 13 sections posterior to A; C, 7 posterior to B; D, 6 posterior to C; and E, 11 posterior to D. 16x. the vestibulum is a solid cylinder of epithelial cells at this stage. The cavum nasi proprium is very similar to that of the adult Splienodon although the ventraler Musehelraum is not yet fully developed. As is shown in the figures, the anterior end of the concha posterior is projected anteriorly from its line of attach- PARSONS: NASAL EMBRYOLOGY 159 ment, so that in transverse section it appears to lie free in the nasal cavity. The choana is ventral to the central part of the cavnm nasi proprium and approximately two-thirds its length. Both Choanenfalten and Vomerpolster are well developed and resemble the adnlt structures. Figure 36. (Second half). Spltowdon 1490 (25.2; MC). Transverse sections through the nasal area. F, 7 sections posterior to E; G, 8 posterior to F; H, 7 posterior to G; I, 11 posterior to H; and J, 10 posterior to I. 16x. Jacobson's organ also appears essentially adult in this series. It ai)parently enters the nasal cavity medially and quite sep- arately from the lachrymal duct (see P'igs. 36C and D). The structure of this region thus appears exactly as described and figured by Broom (1906), Fuchs (1908), and Bellairs and Boyd (1950). Despite careful study of both available series, I am unable to see any evidence for Iloppe's statements on the junc- 160 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY ture of the lachrymal duct with a duct from the anterior end of Jacobson's organ. This cannot be caused by the use of only transverse sections ; one of the series studied is cut in the frontal plane. The questions on the nature and homologies of the two conchae are more difficult to answer. Certainly there seem to be two distinct lamellae, as described by Iloppe. Although they are, in all stages which have been studied, joined ventrally, their position in the embr^yos (see Fig. 80) makes it appear extremely unlikely that S'phenodon possesses a single concha comparable to that of the Squamata. In its general position and structure the concha posterior resembles the saurian concha very closely and their homology seems probable. However, the concha anterior of Sphenodon is less easily explained ; this problem is discussed in a subsequent section. Order Squamata General. The Squamata include a majority of the living rep- tiles and display a far greater range of morphological diversity than any of the other orders. It is also the order which has been most studied witli respect to nasal anatomy and embryology. However, the preponderance of the earlier work treats the sub- order Lacertilia ; the Ophidia are less well known. Therefore, since snakes will serve as well as lizards as examples of the Squa- mata for comparison with members of the other orders, only the Ophidia are described in the following account. The major works on the nasal anatomj^ of lizards may be briefly noted. Leydig's (1872a) great monograph of the German species forms the starting point for all the more recent studies of the anatomy and histology of the nasal region. The classic works on the nasal embryology of the Lacertilia are those of Born (1879) and Peter (1900). Two other early papers, Beecker (1903) and Fuchs (1908), made major contributions to our knowledge of both the adult form and the development of the nasal area, and established the terminology now generally used for all mem])ers of this order. There are many recent works treating the nasal cavities and Jacobson's organs of lizards. Especially noteworthy are the many papers on cranial anatomy published by the South African PARSONS: NASAL EMBRYOLOGY 161 school of anatomists, de Villiers, clu Toit, and their students. The following- list includes only major works which consider a variety of forms and which, in most cases, have extensive bibli- ographies: Matthes (1934), Malan (11)46), Pratt (1948), Steb- bins (1948), and Bellairs and Boyd (1950). For detailed descriptions and many further references, the reader is referred to these papers. Adult Atiatomif. The gross form of the nasal cavities is relatively simple in adult snakes. Anteriorly, there is typically a very small tubular vestibulum, running medially from the naris externus to the anterior end of the cavum nasi proprium. The posterior end of the cavum is connected to the oral cavity by the rather short and wide ductus nasopharyngeus. Certain authors, most notably Fuchs (1908), would prefer to restrict the term ductus nasopharyngeus to the crocodilian or mammalian condi- tion, in which a well formed secondary palate is present ; how- ever, as in the case of turtles, the term is here retained and used in a topographic sense. Since this problem is discussed in the section on turtles, the arguments need not be reviewed again. The cavum nasi propium is a large chamber, with its greatest height and wddth generally near the posterior end. It is thus somewhat ovate in either frontal or sagittal section. The concha is a prominent ridge along the posterior two-thirds of the ventro- lateral wall of the cavum. Beecker (1903) proposed a series of terms for the various parts of the cavum. These were adopted by Fuchs (1908), and are now in fairly standard usage. There- fore, they are employed in the present work, except in one case where Beecker 's term appears to be rather confusing. It must be emphasized at the start that these subdivisions of the cavum do not have any definite boundaries, but, despite their vagueness, they are most useful descriptively. First, the cavum nasi proprium is divided into three main zones. The largest of these is the conchal zone ; all of the cavum anterior to the concha is here termed the anterior caval zone, and the smaller part posterior to it is the Antorbitalraum. Sec- ondly, the conchal zone is subdivided into three parts. The por- tion of the cavum lateral and dorsolateral to the concha is the Sakter. This is continuous with the dorsomedial Stammteil, while the ventromedial third of the cavum is termed the Choanengang. 162 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY The relationship of these three parts of the conchal zone is best seen in transverse sections such as are shown in Figure 58. Beecker and Fuchs term the anteriormost of the three main divisions of the cavum the Nasenvorhof ; however, since other workers, such as Kathariner (1900), have used it to designate the vestibulum, the term is not employed in the present work. The only other difference in usage is that the ductus nasopharyngeus of the present paper is considered to be a part of the Choanengang by Beecker and Fuchs. In the descriptions of Jacobson 's organ, the anteroventral pro- jection into its lumen is here termed the mushroom body. This structure is generallv known bv that term or a variant such as rS Figure 37. Medial view of the left nasal cavity of Drymarchon to show the lateral wall. Figure 38. Medial view of the left nasal cavity of Crotalus to show the lateral wall. The bristle passes through the naris externus. pilzenformige AVulst or fungiform eminence, but is sometimes referred to as the concha of Jacobson 's organ. Three genera were studied in the present investigation. Dis- sections were made of the heads of Drymarchon corals and Cro- talus adamant ens, and slides of an adult Storeria dekayi were also examined. Unfortunately, the last is an incomplete series and many of the sections are not in ideal condition, so that the bulk of the description is based on the first two forms. Figure 37 shows the lateral wall of the nasal cavity of Dry- marchon. The naris externus is a large, nearly circular opening PARSONS: NASAL EMBRYOLOGY 163 very near tlie tip of the snout. From it, the very short vestibuluni runs medially and slij^-htly anteriorly to enter the anterior end of the eavum nasi proprium. On the anteromedial wall there is no macroscopically visible boundary between these two regions, but there is a marked, though rather moderate swelling along the lateral part of the anterior margin of the eavum. The duct of the glandula nasalis externa opens well dorsally on the posterior wall of the vestibulum, approximately halfway from the naris to the eavum. The shape of the eavum nasi proprium is well shown in the figure. Its anterior half is a simple, nearly circular tube, with its dorsoventral diameter slightly greater than the lateromedial. Posteriorly, the eavum enlarges quite markedly. The concha is a prominent ridge attached along its anterior and ventrolateral edges. The dorsal margin turns slightlj^ laterally, especially posteriorly. The Sakter is a quite large space, but does not extend anterior to the anterior line of attachment of the concha. Thus in transverse section the conchal zone is shaped rather like a question mark, and its three parts, the Sakter, Stammteil, and Choanengang, form a continuous cavity without any visible boundaries between them. The Antorbitalraum is very small. Posteroventrally the Choanengang is continuous vrith the rather short but very wide ductus nasopharyngeus. The latter Ls a circular tube which enters the oral cavitj' at the anterior end of the large median palatal trough. Although they are not joined together further anteriorly, the ducti of opposite sides thus have a common opening into the mouth. Along the w^alls of the ductus there are many very small, but macroscopically visible grooves, probably of a glandular nature. These grooves continue pos- teriorly into the midpalatal trough, and also, though less obvi- ously, anteriorly into the Choanengang. In all regions they run parallel to the long axis of the nasal cavity. Jacobson's organ is, as in all snakes, well developed as a roughly spherical structure. In Drymarchon it lies ventromedial to the Choanengang. The mushroom body forms a large hem- ispherical projection from the anterior and anteroventral third of the wall of Jacobson's organ, thus restricting the lumen of that organ to a rather narrow, bowl-shaped space (see Fig. 37). From the anteroventral part of Jacobson's organ, the extremel}^ 164 RULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY narrow duct runs ventrally to enter the mouth well anterior to tlie choana. The other snake which was dissected, Crotalus, is shown in Figure 38. Although the naris externus closely resembles that of Drymarchon, the vestibulum is considerably longer and better developed, and runs anteroventrally as much as medially. The latter is, however, still a very simple tube. No trace of the duct of the glandula nasalis externa could be found. As in the case of Drijmarchon, the vestibulum and the cavum nasi pro- prium are not maeroscopically distinguishable anteriorly. Pos- terolaterally, however, their lioundary is sharply marked by a very large ridge in the anterolateral wall of the cavum. This ridge, which is well shown in Figure 38, represents the greatest difference in nasal anatomy between Drymarchon and Crotalus. It runs more than half the length of the cavum nasi proprium, from the anterodorsal corner of the lateral wall to the ventral caval wall and the posteroventral part of the conchal attachment. There is a rather shallow groove in the lateral wall of the cavum ventral to this ridge and a somewhat deeper one dorsal to it ; thus in transverse section the anterior portion of the cavum is more C-shaped than circular. In its gross form the ridge is somewhat reminiscent of the concha anterior of Sphenodon; however, its embryology is completely unknown, so that it is impossible to determine whether the two structures are actually the same. The ridge found in Crotalus appears to be a great elaboration of the small postvestibular swelling mentioned in Drymarchon. In the absence of any sectioned material of Crotalus, it is not possible to tell if any special struc- ture is contained within the ridge, although it can be seen that the facial pit is well posteroventral to the area in question. Posteriorly and dorsally the cavum nasi proprium of Crotalus is very similar to that of Drymarchon. The concha of Crotalus is slightly shorter and more dorsally situated, but is essentially the same shape. Its posterodorsal margin is laterally directed. The Antorbitalraum is considerably larger than in Drymarchon, but it remains simple and, as far as can be seen from gross dissection, unmodified in any way. The ductus nasopharyngeus and choana of the two genera are also very similar, as is shown by a comparison of Figures 37 PARSONS : NASAL EMBRYOLOGY 165 and 38. The only difference worth notinj; is the oecurrence of a small ridg:e, separatinp- the ductus from the cavum nasi proprium, which is found in Crotalus hut not Drjfmarchou. No grooves could be seen in the walls of the ductus in the former genus, but the preservation of the specimen is not good enough to permit a denial of their presence there. The only noticeable difference between the Jacobson's organs of Drymarchnn and Crotalus is the smaller mushroom bodj^ and correspondingly larger lumen of the latter. However, the duct could not be dissected out in the rattlesnake studied, so its structure and exact location are not known. Unfortunatel}^, the slides of Stoi'eria show little of the gross anatom}^ of the nasal cavity. The cavum nasi proprium appears to be approximately the same shape as in Drymarchon, with a well developed concha containing part of the glandula nasalis externa. Anteriorly, the sections are almost all broken, so that nothing can be determined concerning the form of the vestibu- lum. The ductus nasopharyngeus is proportionately longer than in Drymarchon, but not otherwise different. Jacobson's organ is large, and appears exactly like that of Drymarchon. The dorsal end of its duct leaves the posteroventral corner of that organ ; the ventral end including its connection with the lachrymal duct is missing on the available slides. There is a moderately sizable body of literature on the nasal anatomy of adult snakes, but the structure in many major groups is totally unknown ; even in the commoner and larger families, very few forms have been studied. The only embryological paper which contributes anything to a comparative study of the nasal cavities of snakes is that by Pringle (1954) ; it is discussed in the subsequent section on the later nasal embryology. Major review papers considering this group include Hoffmann (1879-1890), and Matthes (1934). Zuckerkandl fl910b) revicAvs the earlier literature on Jacobson's organ. The following genera have been described or figured : Typhlops (Zuckerkandl, 1910a), Xenopeltis (Bellairs, 1949), Python (Sol- ger, 1876; and Dieulafe, 1904-1905), Constrictor (Gegenbaur, 1873), Nafrix (Leydig, 1872b; Solger, 1876; von Mihalkovics, 1898; Kathariner, 1899 and 1900; Broman, 1920; and von Navratil, 1926), Thamnophis (Macallum, 1884), Elaphe (Ne- 166 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY mours, 1930), Acrochordus, Cerberus, Hydrophis, Pelamis, Lati- cauda (the last five all by Kathariner, 1899 and 1900), Vipera (Solger, 1876; and Bellairs, 1942), and Crotalus (Solger, 1876). Finally Bellairs and Boyd (1950) investigated the relationship between the lachrymal duct and Jacobson's organ in a large series of forms including representatives of a majority of the families. Almost all these descriptions appear essentially similar, and no detailed review is necessary. There is some variation in the at- tachment of the concha with the wall of the nasal cavity; in Python it is attached dorsally for the anterior two-thirds of its length, and hence the Sakter forms a large blind pocket, while in Drymarchon the concha is connected to the nasal wall only ventrolaterally. More conspicuous variations are found in the aquatic snakes studied by Kathariner (1900). In such forms the concha is reduced or absent, the sensory area restricted, and the vestibulum modified so as to prevent the entrance of water Avhen the animal is submerged. Jacobson's organ is well de- veloped in all snakes thus far studied. Otherwise, there are no basic differences reported within this suborder. Early Embryology. The early stages in the nasal embryology of snakes have been little studied. Five papers mention the sub- ject, all of them dealing with Natrix natrix and all written before 1890. The first is Rathke's (1839) classic monograph in which the nasal placodes and pits are described and the origin of Jacobson's organ (Nasendriise of Rathke) is noted. In 1878 both Parker and Fleischer published papers which add little to Rathke's account. Born (1883) is concerned primarily with the development of the lachrymal duct and thus does not treat early stages in great detail. Finally Beard (1889) observed the embry- ology of the nose in connection with his studies on the olfactory nerve and Jacobson's organ. Specific points from these papers will be noted in the descriptions given below ; therefore, no de- tailed review of the earlier literature is necessary. The following account is based primarily on the study of a large series of Thamnophis embryos. At least two species, T. radix and T. sirtalis, are represented, but there does not appear to be au}^ difference between them, and therefore they are here treated as one. Three embryos of Oxybclis were also available and serve to fill in one gap left in the Thamnophis series. PARSON'S: NASAL EMBRYOLOGY 167 The first sign of nasal differentiation is the formation of the nasal placode which is, in its earliest stapes, nothing more than a thickened area of epithelium without any indentation. It is 40B 40C Figure 39. Thamnophis 1559 (3.7; MC). Frontal sections through the nasal area. A, sei-tion 52; B, 6 sections ventral to A; and C, 5 ventral to B. 42x. Figure 40. Thamnophis 1592 (3.3; MCj. Transverse sections through the nasal area. A, section 123; B, 7 sections posterior to A; and C, 5 posterior to B. 42x. Figure 41. Thamnophis 1589 (3.5; MC). Transverse sections through the nasal area. A, section 94; B, 3 sections posterior to A; and C, 5 posterior to B. 42x. roughly circular and located on the ventrolateral surface of the snout as shown in Figures 39 and 40. The greatest thickening, up to ten times that of the unmodified epidermis, occurs some- 168 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY what dorsal to the center of the placode so that the transition is most abrupt dorsally and most gradual ventrally, but there is nowhere a visibly distinct boundary between the nasal epithelium and the surrounding' epidermis. The transition tends to be slightly more abrupt posteriorly than anteriorly. In some of the older embryos of this stage the placode exhibits a marked central thickening with an only slightly thickened ring surround- ing it (see Fig. 39) ; in other embryos there is no such differentia- tion within the placode. Although there may be a small anlage of the olfactory nerve present, the placode is, at this stage, generally well separated from the brain by an intervening layer of mesenchyme. The epithelium of the nasal placode is high columnar centrally, becoming low columnar near the periphery where it merges with the simple cuboidal epidermal epithelium. It is probably simple throughout ; however, the available slides do not show cell boundaries so that it is impossible to be certain about this point. The nuclei are generally concentrated in the basal half of the epithelium, but mitotic figures appear to be most numerous near the lateral or external surface. In many of the embryos there is a very distinct basement membrane separating the nasal epi- thelium from the underlying mesenchyme. In some series, prob- ably slightly more advanced, this disappears beneath the centei- of the placode, but is still present toward the edges. When it is present, the basement membrane is, for the most part, not smooth, but shows gentle undulations. The first major change in the structure of the placode is the development of a small indentation in the area of greatest thickness (see Figs. 41 and 42). The placode still extends over a large part of the ventrolateral surface of the snout, but the inpocketing is almost directly lateral, xinother change which occurs at approximately the same time concerns the establish- ment of a distinct cellular anlage of the olfactory nerve; the structure and further development of this are discussed in a subsequent section. As in the preceding stage, the nasal placode is continuous with, and indistinctly set off from, the epidermis. The transition in thickness remains most abrupt dorsally and most gradual ventrally. PARSONS: NASAL EMBRYOLOGY 169 Histologically, there are no essential differences between the unindented and indented nasal placodes. Throughout its early development the nasal epithelium slowly becomes thicker, but the general picture of high columnar epithelium with somewhat NIPT Figure 42. ThamnopMs 1643 (3.4; MC). Frontal sections through the nasal area. A, section 71 ; B, 3 sections ventral to A ; and C, 4 ventral to B. 42x. Figure 43. OxyhclxH l.";28 (6.9; MC). Transverse sections through the nasal area. A, section 166; B, 8 sections posterior to A; and V, 5 posterior to B. 39x. Figure 44. Thamnophia A2 (16.0; EEW). Sagittal section through the nasal area (section 26). 39x. basally located nuclei does not change. The basal position of the nuclei is most marked in the area of indentation, but, as in the previoiLs stage, mitotic figures are concentrated along the external surface of the placode. 170 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY The invagination of the placode continues so that a nasal pit is formed. Early stages showing such a pit are poorly represented in the available Thamnophis material, but the three Oxybelis em- bryos make detailed description possible (there appear to be no important differences between these two genera in the younger stages ; thus the Oxyhelis material may be inserted into the series of Thamnophis embryos without breaking the continuit.y of the description). As is shown in Figure 43, the pit is, at this stage, relativel}^ shallow and possesses a wide lumen ventrally. It extends dorsomedially from the dorsal part of the original pla- code. The inpocketing occurs along the horizontal length of the nasal epithelium so that the naris is formed as a slit whose an- terior end is slightly dorsal to its posterior. At this stage the naris extends the full length of the nasal pit although the anterior wall of tlie nose is very nearly vertical. The ei)itlielium of the pit is thickest dorsally and medially, but the regional differences are very slight and the histological appearance is exactly as in the preceding stage. The most important development during the early nasal pit stage is the appearance of Jacobson's organ. Parker (1878). Born (1883), and Beard (1889) all report the presence of a small depression representing this organ while the nose is still in the placode stage, but none could be seen in am- of the embryos used in the present study. Rathke (1839) also seems not to have found Jacobson 's organ until after the start of nasal invagi- nation, although his description is not perfectly clear on this point. In the youngest Oxyhelis embryo, Jacobson's organ is a very slight ventromedian groove in the wall of the nasal pit (see Fig. 43B). It is somewhat posterior to the center of the naris. This groove graduall}'' lengthens and becomes very large, reaching a size equal to that of the main nasal pit. With further growth of the nose, Jacobson's organ comes to lie medial to the center of the naris. Histologically, it does not differ in any way from the main portion of the nasal pit. The next stage in the development of the nasal pit of Tham- nophis is shown in Figures 44, 45, and 46. The pit has become moderately deep, with its long axis essentially vertical when seen in transverse section. As in the previous stage, the naris is a long slit with the posterior end ventral to the anterior. PARSONS: NASAL EMBRYOLOGY 171 The lumen is widest anteriorly and ventrally, but it is generally far less extensive than in younger embryos. Anteriorly a small portion of the pit extends anterior to the naris, thus forming a rather indistinct recess, the apikale Blindsack of Peter (1901). Posteriorly, the naris continues slightly posterior to the nasal pit as a shallow groove in the palate. The posterior part of the nose is slightly lateral to the anterior, but there is as yet no indi- Figure 45. Thamnophis K (19.0; EEW). Frontal sections through the nasal area. A, section 52; B, 8 sections ventral to A; and C, 5 ventral to B. 38x. Figure 46. Thamnophis 1348 (5.8; MC). Transverse sections through the nasal area. A, section 154; B, 4 sections posterior to A ; C, 6 posterior to B; D, 9 posterior to C; and E, 6 posterior to D. 38x. cation of a concha laterally. Jacobson's organ is, at this stage, relatively huge, having a size equal to that of the entire re- mainder of the nasal organ. It is a roughly spherical mass on the ventromedial wall of the nasal pit. The large lumen of this organ is widely open into the nasal pit medial to the center of the naris. 172 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY Histologically, the nasal epithelium shows rather few changes. It is still mainly high columnar and merges graduall}' with the epidermal covering of the snout. In the available material it is impossible to tell whether any of the epithelium is stratified or whether it is all simple. Considerable differential thickening has now occurred. The thickest epithelium is that of Jaeobson's organ, while in the nasal pit the epithelium is thicker anteriorly, especially dorsomedially, than posteriori}'. Throughout the markedly thickened areas the nuclei are still concentrated basally and mitotic figures are commonest toward the lumen; however in the thinner zones both are evenly distributed throughout the epithelium. The final stage before the fusion of the lateral and medial nasal processes resembles closely that described above. Considerable growth has occurred and the extension of the lateral nasal process causes the naris to lie along the edge of the ventral surface of the snout rather than on its lateral face (see Pigs. 47 and 48). The lumen is generally quite restricted, but expands ventrally near the naris and anteriorly where the apikale Blindsack is very strongly developed. The most striking change in the form of the nasal pit is the appearance of the concha as a projection of the lateral wall. This is most clearly seen in Figure 48D. Un- fortunately, there is some distortion in the series figured; how- ever, the drawings show the general form of the nose at this stage. With the formation of the concha, that portion of the nasal pit dorsal and lateral to it becomes recognizable as the Sakter. There appear to be no histological changes in the nasal epithelium at this stage. Jaeobson's organ continues to enlarge. It is still an essentially spherical organ opening into the ventromedian part of the nasal cavity, now slightly anterior to the center of the naris. The con- nection between its lumen and that of the main nasal pit has become slightly constricted, but is still very wide. All of the epithelium of Jaeobson's organ is markedly thickened, but ven- trolaterally it is somewhat thinner than elsewhere. In most of the series of this stage, the nasal pit has increased in size more rapidly than has Jaeobson's organ so that it is considerably larger than the latter (see Pig. 47) ; however in one specimen (see Fig. 48), this is not the case, and Jaeobson's organ is still fully as large as the nasal pit. PARSONS : NASAL EMBRYOLOGY 173 NPT 48 D NPT JO k48B ■MNP Figure 47. Thamnoplds 13(31 (0.0; MCj. Frontal sections through the nasal area. A, section 99; B, 7 sections ventral to A ; C, 7 ventr.-il to B; and n. .■; ventral to 0. 43.\. Figure -48. ThamnophiN G (33.0; i^EW). Transverse sections through the nasal area. A, section 162 ; B, 4 sections posterior to A ; C, 4 posterior to B; D, (5 posterior to C; and E, 2 jiosterior to U. 43x. 174 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY None of the available embryos show the stages immediatelj' preceding the actual fusion of the lateral and medial nasal processes. However, some of those to be described in the follow- ing section on the later embryology are very close to that stage, and therefore the process is best discussed when they are con- sidered. Later Embryology. The following account is based on the study of a large number of Thamnophis embryos. As in the case of the earlier stages, these are of two species, T. radix and T. sirtalis. However, since there appear to be no differences be- tween the two, only one description is necessary. Unfortunately, material of other genera was not available. As far as could be told from the available material, the fusion of the lateral and medial nasal processes follows a slightly differ- ent ])attern in snakes from that described for turtles. Un- fortunately, certain critical stages were not represented, but observations on lizards (mainly Aristelliger) and the accounts in the literature, to be discussed subsequently, appear to sub- stantiate the impression gained from the Thamnophis embryos. In snakes the processes meet throughout almost all of the length of the anterior third of the nasal pit, leaving an open slit-like tlioana beneath the posterior two-thirds. There is, however, a very small naris externus, the lumen of which is almost immedi- ately obliterated. The processes very quickly fuse and the ap- posed epithelial layers disappear throughout most of their extent, but anteriorly no such fusion occurs. There the cells clearly retain their epithelial character, and form a .solid tubular vestibu- lum. In turtles the nasal processes fuse and, at a slightly later stage, the naris externus is closed as the vestibulum becomes a solid tube ; in snakes the two events occur almost simultaneously. With the fusion of the lateral and medial nasal processes, the nose takes on the form shown in Figure 49. The vestibulum is a short, nearly tubular structure, although its diameter is slightly greater dor.somedially than ventrolaterally and it is slightly flattened dorsoventrally. The naris externus is ventro- laterally placed very near the anterior end of the snout, and the vestibulum runs dorsomedially from it to the anterior end of the cavum nasi proprium. PARSONS: NASAL EMBRYOLOGY 175 iM9C Figure 49. Thamnophis 1350 (7.4; MC). Transverse sections through the nasal area. A, section 174; B, 11 sections posterior to A; C, 10 posterior to B ; D, 6 posterior to C; and E, 4 posterior to D. 28x. Figure 50. Thamnophis 1372 (headlength, 4.4; MC). Sagittal sections through the nasal area. A, section 116; B, 11 sections medial to A; C, 4 medial to B ; and D, 7 medial to C. 28x. 176 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY The latter chamber has already attained a considerable degree of complexity. Anteriorly, it is an oval tube which is nearly vertical except at the extreme anterior end where the dorsal portion leans somewhat medially. At the level of the anterior end of Jacobson's organ, the dorsal end begins to curl laterally and then ventrally around the concha. At the same level the ventral margin of the cavum extends venti-all}^ and ventromedialh^ to enter the oral cavity at the choana (see Fig. 49C). In this region the cavity can be roughly divided into the three zones, described by Beecker (1903) and Fuchs (1908); that section lateral and dorsolateral to the concha is the Sakter, the dorsomedial third is the Stammteil, and the ventromedial segment leading to the I'hoana is the Choanengang. As in the later stages, there are no definite boundaries l)etween these three regions. The concha is a large inpocketing of the wall of tlie cavum which is attached anteriorly and ventrolaterally. Posteriorly, it may project freely a short way into the lumen, but this is variable. Although the concha does not reach the posterior end of the cavum, the Antorl)italr-aum is very small and not in any way a distinctly separate region. The clioana extends to the jiosterior end of the cavum, and even slightly ])osterior to it as a groove along the surface of the palate, but there is still no ductus nasopharyngeus. Posteriorly, the Choanengang l)ecomes mure ventromedially di- rected as the dorsal part of the cavum nasi proprium comes to lie farther from the midline. Thus, in frontal section, the nasal cavities of opposite sides show a marked posterior divergence dorsally and anteriorly, but are almost pai-allel posteroventrally. The lumen of the cavum is generally very narrow. There are however three areas where it expands slightly. The first of these is the anteriormost part, anterior to Jacobson's organ and the concha. Even here the lumen is narroAver than it would appear in transverse sections, as its long axis is transverse. Another expansion occurs in the anterodorsal part of the Choanengang lateral to the connection between Jacobson's organ and the nasal cavity. The third and last is at the posterior end of the concha, that is, in the Antorbitalraum. Despite the general restriction of the lumen, the choana is open throughout its length. During this stage there are important changes in the form of Jacobson's organ. When the nasal processes first fuse, it is PARSONS : NASAL EMBRYOLOGY 177 still a large spherical organ opening- into the medial wall of the nasal eavity at the level of the anterior end of the ehoana. Very soon thereafter, however, the anteroventral wall of the sphere pnshes into the large central Ininen, thus forming the mushroom hody (see Fig. 49C). With this development the lumen of Jaeobson's organ is considerably reduced, although it is still (luite sizable, and the connection with the main nasal cavity, although open, becomes greatly narrowed. Jaeobson's organ still extends as far posterior as does the cavum nasi proprium. Histologically, the picture appears to remain essentially simi- lar to that in the earlier stages. The vestibulum possesses one or two laj^ers of moderately low columnar cells basally and is filled Avith a mass of irregular, probably roughy isodiametric cells, the Fiillgewebe of Weber (1950). The remainder of the nasal eavity is lined by two epithelial types which are not dis- tinctly separable, but gradually merge along their boundaries. First, there is a conspicuously thickened and presumably poten- tially sensory epithelium. This occurs anteriorly and dorsally in the cavum nasi proprium and throughout in Jaeobson's organ except on the surface of the mushroom body. It possesses a clear peripheral zone containing many mitotic figures and a thicker basal zone with many rows of circular nuclei. The thinner pos- teroventral area of the cavum and the mushroom body are covered by simple or lulaminar low columnar epithelium. The distribution of simple versus stratified epithelium is difficult to determine in the available matei-ial, ])ut does not appear to follow any definite pattern. At the ehoana this non-sensory epithelium merges with the simple cuboidal epidermal epithelium without any visible break. The details of the pattern of thickening are clear in Figure 49 and i-equire no description. Slightly older embryos show very few changes ; two such series are shown in Figures .30 and 51. The naris externus is .still very well ventral, and the vestibulum therefore runs dorso- medially to its juncture with the cavum nasi proprium. How- ever, its dorsoventral diameter is somewhat increased, especially superficially, so that the vestibulum is now slightly flattened in the transverse plane. The basic shape of the cavum luisi propi-ium is unchanged, Init some minor differences are noticeable. Anteriorly, it has ex- tremely thick epithelium covering the small transverse segment 178 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Figure 51. Thamnophis 1364 (10.0; MC). Frontal sections through the nasal area. A, section 135; B, 10 sections ventral to A; C, 13 ventral to B; D, 11 ventral to C; and E, 5 ventral to D. 22x. Figure 52. Thamnopkia 1415 (12.0; MC). Transverse sections through the nasal area. A, section 213; B, 5 sections posterior to A ; C, 11 posterior to B; D, 15 posterior to C; E, 5 posterior to D; and F, 12 posterior to E. 22x. PARSONS : NASAL EMBRYOLOGY 179 just anterior to Jacobson's orf?an. However, the lumen is now as narrow there as in most of the conchal zone. The thickness of the epithelium anteriorly causes that wall of the cavum to project slightly anterior to the vestibuluin. The concha is slightly larger than previously, but its shape remains the same, although the connection between the Sakter and Stammteil extends further ventrally posterior to tlie concha. Another change affects the somewhat expanded area of the Choanengang opposite Jacobson's organ. An anteroventral ridge has developed which results in the partial formation of a duct of Jacobson's organ. This runs slightly posterodorsally as well as laterally before entering the cavum (see Fig. 51D). However Jacobson's organ still opens into the anteromedial part of the Choanengang. Posteroventral to Jacobson's organ, the cavum becomes a rather small oval slit leading to the choana. Although this slit is not yet distinctly separated from the remainder of the Choanengang, it may noAv be considered the anlage of the ductus nasopharyngeus. The choana is still a moderately long slit which continues ]>osteriorly along the palate as a shallow^ groove. The mushroom body of Jacobson's oi*gan increases in size so that the lumen of that organ becomes (juite thin. However, there is no change in its shape. The thinning of the lumen reaches its extreme at the entrance to the short duct which is now virtually, if not actually, closed in several of the embryos. On its basal or external surface, the thickened epithelium of Jacobson's organ presents an irregular surface, with many small liemispherical projections into the surrounding mesodermal tis- sue. These represent the first stage in the development of the epithelial columns. Their later development and significance are considered in the section on Jacobson's organ, and no attempt has been made to indicate them in the various drawings of the embryos. Histologically, these embryos appear exactly like those of the preceding stage. In most places the non-sensory epithelium probably is two cells thick, but the available material does not permit any definite statements on this. The next stage is shown in Figure 52. Further differentiation is noticeable in almost all regions of the nose, but the changes are probably most marked in the vestibulum. During the period 180 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY represented by the several embryos of this stage, the naris externus starts to move further dorsal ly on the snout, although this process of movement is far from complete, even at the con- elusion of this stage, and the vestibulum still runs slightly dorsally, as well as medially, to reach the cavum nasi proprium. The posteroventral portion of the vestibulum is extended pos- teriorly as a small flange. From the dorsal surface of this pro- jection, a solid rod-like process reaches a short distance dorso- laterally and posteriorly; this process is the anlage of the glan- dula nasalis externa, and is considered in the section on glands. The anterior end of the cavum nasi proprium has also de- veloped considerably. It remains an oval tube with the long diameter dorsomedial to ventrolateral. The walls are very thick except at the ventrolateral end where the cavum possesses small dorsolaterally and ventromedially directed extensions. Thus in transverse section this region resembles an inverted T. The medial edge of the posteroventral flange of the vestibulum is continuous with the end of the ventromedial caval extension (see Fig. 52B). This pattern persists posteriorly to the level of the anterior end of Jacobson's organ where the dorsal end of the cavum turns laterally dorsal to the concha and the ventral processes are lost. A slight trace of the more lateral of these extensions may, however, reach posteriorly into the conchal region. The concha is very large, nearly filling the posterior two-thirds of the cavum. As shown in Figure 52, its shape is very similar to that of the preceding stages, although the ventrolateral line of attachment is proportionately narrower. The small area pos- terior to the concha, the embryonic Antorbitalraum, is un- changed. Near the posterior end of Jacobson's organ, the Choan- engang becomes nearly vertical, and its ventral end turns sharply medially to reach the slitlike choana. The posterior extension of the choana as a palatal groove is now becoming shorter. In most of the series of this stage, the lumen is almost totally obliterated in the ventral portion of the C'hoanengang. When one is present, it is typically largest near the posterior end; this fact would support Fuchs' (1908 and 1911) theories on the formation of the choanae, which are discussed below. PARSONS : NASAL EMBRYOLOGY 181 Jacobson's organ shows no real changes, but its duet has de- veloped considerably. It starts as a semicircle around the postero- ventral end of the nuishroom body and runs ventrally to the choana. At the beginning of this stage the duct enters the anterior wall of the Choanengang well dorsallj^ but by the end of the period it Joins the ductus nasopharyngeus at the latter 's ventral margin, that is, actually at the choana. The duct is solid. Although not yet connected directly to Jacobson's organ, the lachrymal duct now reaches to the anterolateral wall of the Choanengang. This area thus displays a pattern closely com- parable to that found in the adult Sphenodon. The non-sensorj' portions of the cavum nasi proprium typically have a basal layer of Ioav columnar cells with central to basal nuclei, weakly staining cytoplasm, and a very distinct basement membrane. Usually there is a thinner layer of cuboidal or squamous cells facing the lumen, but in some areas, especially posteromedially, this inner layer appears to be lacking. Such epithelium is found at the ventral end of the anterior oval sec- tion of the cavum, throughout the Choanengang, on the postero- ventral third of the Sakter, and on the posterior wall of the nasal cavity. Tlie vestibular and sensory epithelia appear exactly as in the younger series, except for some increase in thickness. As shown in Figure 52, the sensory epithelium is thickest in Jacob- son's organ; in the cavum nasi proprium, it is thicker dorsally and anteriorly than further ventralh' or posteriorly. In slightly older embryos, there is further growth but almost no change in form (see Fig. 53). The posteroventral flange on the vestibulum is slightly more marked, extending from the naris externus to the cavum nasi proprium as a thin shelf. Within the cavum there is great variation .in the extent of the lumen ; in some embryos it is clearly open throughout, and even wide in some areas such as the antei-ior (quarter, but in others it is as restricted as in the previous stage. This same variation is found in the choana and duct of Jacobson's organ, although most commonly both of these are almost entirely without a lumen in the material studied. The concha is now recognizable further posteroventrally than l)efore, extending ventral to the level at which tlie jiosteroventral part of the Choanengang (or anterior 182 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY part of the ductus nasopharyngeus ; the two are not really dis- tinct) turns posteromedially towards the choana (see Fig. 58C). Jacobson's organ has very nearly attained its adult structure. It is a large roughly spherical organ, Avith the ventral surface rather flattened. The mushroom body forms a large hemispheri- cal process which extends inwards from the anteromedial corner of its ventral surface and nearlv fills the lumen. Thus the cavitv Figure 5a. Thamnophis 1365 (10.2; MC). Frontal sections through the nasal area. A, section 105; B, 10 sections ventral to A; (", 16 ventral to B; 1), 6 ventral to C; and E, 9 ventral to D. 23x. is the shape of an inverted bowl. The duct of Jacobson's organ runs ventrally from the posterior part of the organ, and is semi- lunar in frontal section, with the convex side posterolateral. In its ventral course the duct is slightly spiraled so that the convex side comes to lie posteriorly and finally, in this stage, postero- medially. The available embryos show some variation in the degree of separation between this duct and the choana; gen- PARSONS: NASAL EMBRYOLOGY 183 erally the duet enters the roof of the mouth at the anterior end of the choanal groove. Although very close to the duct of Jacobson's organ, the lachrymal duct still appears to enter the choana. All of these openings occur along the dorsolateral corner of a large median trough in the palate ; at this stage the trough is comparatively wide and sliallow. The histology appears to be very similar to that of earlier stages. In the presumably non-sensory areas of the cavura nasi proprium, the epithelium most commonly appears to be two or even three cells thick, but some areas still seem to have simple columnar epithelium. One embryo possesses small projections of the dorsolateral sensory epithelium which could be the first appearance of Bowman's glands; however the other embryos lack them, and their nature cannot definitely be determined in the series in (juestion. Some embryos of this and earlier stages lack a distinct basement membrane under parts of the sensory epithelium, and the epithelial boundary becomes very vague in such areas. The nasal cavity continues slowly to assume its adult form without any really marked changes. Figure 54 shows an embryo slightly more advanced than those described above. Here the vestibulum has become perfectly horizontal, and the naris ex- ternus is directly lateral on the snout, quite near, but not at, its anterior end. The posteroventral vestibular flange is very well developed, as shown in Figure 54A. Histologically, the vestibu- lum cannot always be differentiated from the non-sensory ventral portion of the cavuni nasi proprium, but in some series minor differences can be found which indicate that the former region forms onty the short and solid horizontal tube. Anterior to Jacobson's organ, the cavum nasi proprium still resembles an inverted T when seen in transverse section. The dorsal end is well medial to the ventral, so that the long central arm of the T forms an angle of approximately 30 degrees with the midline (see Fig. 54B). Around the ventral projections, the epithelium is thin, but the central ])ortion possesses the thickest epithelium of any part of the cavum, even as far ventral as the vestibulum. Posteriorly, in the conchal zone, the ventro- medial projection disappears, but the ventrolateral one reaches almost to the posterior end of Jacobson's organ. The Sakter is now very large. It extends farthest anteriorly near its ventro- 184 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY lateral end ; that is, the concha is attached dorsally as well as ventrally at its anterior end. For most of its length, the ventral end of the Sakter is turned slightly medially where it lies dorsal to the ventrolateral extension of the Clioanengang. Thus the line of attachment of the concha is a (inite thin lamellar band (see Fig. 54D). \'enti-ally the epithelium is non-sensoi'v in type; dorsally it remains thick. ■*-n— OB 54G CHG -LD Figure 54. Thamnophis 1399 (12.1; MC). Transverse sections through the nasal area. A, section 273; B, 12 sections posterior to A ; C, 14 posterior to B; D, 12 posterior to C; E, 13 posterior to I); F, 6 posterior to E; and G, 5 posterior to F. 23x. Near the level of the posterior end of Jacobson's organ, the I'hoanengang becomes more nearly vertical, thus making the attachment of the concha ventral rather than ventrolateral. The Antorbitalraum is still not well developed; it forms a small vertical slit which is continuous ventrally with the ductus naso- pharyngeus. The latter turns ventromedially to reach the PARSONS: NASAL EMBRYOLOGY 185 ehoana, which is a quite short and narrow slit opening into tlie dorsolateral corner of the large median trough in tlie palate. A small groove still runs a short way posterior from the end of the ehoana. As can be seen in Figure 54, the epithelium is thicker dorsally than ventrally or posteriorly. Jacobsou's organ is unchanged, but its duct is now clearly separated from the ehoana. The duct is the same shape as in the preceding stage and remains solid. It leads ventrally from Jacobson\s organ to the palate. Like the ehoana, it enters the dorsolateral corner of the median trough, but is well anterior to that structure. The lachrymal duct enters the roof of the mouth between the ehoana and the duet of Jacobson 's organ ; it is slightly closer to the latter. The histology remains basically the same. A single basal layer of coluuniar cells is found in the vestibulum, with the center filled by rather irregular, but essentially isodiametric cells. The thickened sensory epithelium shows no changes at all. It is darkly staining and contains many nuclei except in the thin clear zone bordering the lumen; there, mitotic figures are common. In the thinner non-sensory zones, there is a basal layer of low columnar cells with central nuclei and weakly staining cytoplasm. Between these and the lumen are one or two laj-ers of very thin irregular cells. Jacobson 's organ has sensory epithelium re- sembling that of the cavum, except in its greater thickness and in the columns found at its basal surface. The nuishroom body is covered by bilaminar cuboidal epithelium. Slightly older embryos are shown in Figures 55 and 56. As in the previous stages, the changes are not great. The posteroventral flange of the vestibulum remains well' developed, but is now most conspicuous laterally. Otherwise that region is unchanged. Elongation of the snout has resulted in a corresponding elon- gation of the cavum nasi proprium, especiallj- anteriorly. There, the segment anterior to Jacobson 's organ no longer appears as a transverse slit in frontal section, but now runs anteromedially to posterolaterally. Both anteroventral extensions are present, with the lateral one tending to turn dorsally at its lateral end, and the medial one extending ventrally as much as medially. The latter does not reach posteriorly as far as the conchal zone, but the former extends nearly to the posterior end of Jacobson 's 186 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY organ as part of the Choanengang. There is little change in the conchal region, except for the further narrowing of the attach- ment of the concha. As in the other stages, that structure may possess a small posterior end projecting freely into the Inmen ; most commonly, however, it is connected with tlie ventrolateral wall of the nasal eavitv thronghont its length. Figure 55. Thamnoplm 1366 (12.0; MC). Frontal sections through the nasal area. A, section 105 ; B, 12 sections ventral to A ; C, 20 ventral to B ; D, 14 ventral to C; E, 14 ventral to D; and F, 8 ventral to E. 26x. During this stage, the Antorbitalraum becomes a somewhat more distinct and complex region. The posteromedial part of the cavum w^hich leads ventrallv to the choana increases in height, PARSONS : NASAL EMBRYOLOGY 187 Figure 5(5. ThamnophUi 1395 (10. -4; MC). Sagittal sections thiough the nasal area. A, section 59; B, 8 sections medial to A ; C, 11 medial to B; D, 6 medial to C; E, 7 medial to D; F, 8 medial to E; and G, 5 medial to F. 20x. 188 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY SO that it now extends dorsally into tlie oonchal zone (see Fig. 55C). By the end of the stage, the posteroventral portion of the Antorbitalraum starts to develop a small lateral pocket and a larger medial one. However, it still merges gradually with the more anterior concha zone and ventral ductus nasopharyngeus. The ductus is a moderatelj^ narrow tube which runs medially as well as posteroventralty to the choana. Although narrow, it typically possesses an uninterrupted lumen at this stage. The choana is much lik(^ that of the preceding stage, Init it continues to shorten and become more posteriorly located. During this period it is a small oval opening, with only a short groove ex- tending posteriorly from it. The duct of Jacobson's organ, now well separated from the choana, is slightly more spiraled than before. At its ventral end the convex side is medial. The lachrymal duct becomes more closely associated with that from Jacobson's organ, and by the end of this stage, typically enters the palate at the posterior end of the latter duct. Both of these are still solid structures, at least at their palatal terminations. Dorsally the duct of Jacol)- son's organ may possess a narrow lumen. Only two comments on the histology are necessary. Bowman's glands are becoming more common in the posterodorsal part of the Sakter, but they are still very rudimentary and alisent from the remaining sensory areas. Secondly, in the non-sensory epi- thelium of the cavum nasi proprium, the basal columnar cells now have the nuck^i towards the lumen. In the vestibulum, the nuclei remain central ; thus the two regions may be separated on histological grounds, although the transition between them is often ([uite gradual. With further growth, the na.sal cavity attains the condition shown in Figure 57. The vestibulum remains a very short solid tube running directly medially from the lateral naris externus. A prominent flange still extends posteroventrally from the ven- tral margin of the vestibulum, especially near its lateral end. The most conspicuous change in the cavum nasi proprium is a general increase in the size of the lumen. However the ductus nasopharyngeus, and frequently also the posterior conchal region, remains as narrow as in the preceding stages. The increase is most marked near the anterior end of the cavum, where it results PARSONS : NASAL EMBRYOLOGY 189 in a relative decrease in the size of the anteroventral extensions. Slightly posterior, however, these extensions reach their maxi- mum development (compare Figs. 57B and ('). The anterior end of the concha is at approximately the level of the center of Jacobson's organ, and, as in previous stages, it is attached to the 57H '57 iL/^. Figure '>'. Tliamnophis 1328 (12.0; MC). Transverse sections through the nasal area. A, section 182; B, 12 sections posterior to A; C, 17 posterior to B; D, 24 posterior to C; E, 8 posterior to I); F, 15 posterior to E; G, 11 posterior to F; H, 7 posterior to G; and I, 15 posterior to II. ISx. Avail of the cavum dorsally as well as ventrally ; thus the Sakter possesses a small anterior pocket. There is no change in the shape of the concha. From the posterior end of its line of attach- ment, a slight ridge continues posteriorly and curves around the 190 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY back Avail of the cavum. The medial end of this ridge is ventral to the lateral part. Tims there is now a moderately definite line of demarcation between the Antorbitalraum and the ductns naso- ])haryngeus. The former chaml)er is still a vertical slit of rather small extent ; ventrally it possesses two small extensions, a verti- cal slitlike opening in the medial wall and a small lateral pocket. The ductus nasopharyngeus continues to increase in length. It forms a narrow tube running from the posteroventral corner of the cavum to the small, nearly circular choana. The ducti of opposite sides converge posteriorly, so that the choanae are very close to the midline. Betw^een them a narrow vertical slit is developing in the palate. With the increased length of the ductus, the choana is now Avell ]iosterior as well as ventral to the cavum nasi ]n"oprium. Jacol)son's organ is relatively little changed. The mushroom body arises from the anteroventral wall, slightly farther anterior than previously, but retains the same form. Thus the entire organ appears to have been tilted slightly posteriorly from its former l)osition. By this stage the lachrymal duct joins the ventral end of the duct of Jacobson's organ, and the two enter the oral cavity together. Tlie former runs anteromedially from the eye to a point posteromedial to the duct of Jacobson's organ, and then turns anterolaterally to enter the antei-omedial corner of the latter duct. Histologically, the vestibulum is unchanged. The sensory areas of the cavum have slightly thicker epithelium than in earlier stages, but show no real dilferem-es, although Bowman's glands are becoming commoner and more widely distributed. Sensory epithelium is found in the dorsomedial two-thirds of the region anterior to the concha, the anterodorsal half of the conchal zone, and the dorsal Avail of the Antorbitalraum. HoAvever, in the non- sensory caval regions a major change is starting to occur. Much of the thin epithelium, especially anteroventrally in this early stage, is becoming folded to produce shalloAv grooves. The cells of these small crypts, Avhich are presumably glandular in the adult, appear to be identical Avith those of other non-sensoiy areas ; they shoAV no changes from the jireceding stages. Figure 58 shoAVS a slightly older embryo in Avliich the posterior part of the nasal cavity approaches closely the adult condition. PARSONS : NASAL EMBRYOLOGY 191 AOR-^ Figure 58. Thamnophis 1326 (headlength, 8.0; MC). Transverse sections tlirough the nasal area. A, section 3G9; B, 24 sections posterior to A ; C, 19 posterior to B; I), 20 posterior to C; E, 21 posterior to D ; F, 15 posterior to E; G, 10 posterior to F; H, 14 posterior to G; and I, 16 jiosterior to II. 21x. 192 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Anteriorly, the anatomy is essentially the same as in the preced- ing stage ; comparison of Figures 57 and 58 renders description unnecessary. The Antorbitalraum is now a large open cavity posterior to the concha. Ventrolaterally, there is a prominent ridge continuing posteriorly from the line of attachment of the concha, and dorsally the end of that structure projects freely into the lumea. Farther posteriorly the cavity narrows and its dorsal wall slopes posteroventrally. The two posteroventral ex- tensions are present, and the medial one is quite large, so that in transverse section the posterior portion of the Antorbitalraum appears triangular (see Fig. 58G). The ductus nasopharyngeus is an oval tube with the long diameter dorsomedial to ventrolateral. As in earlier stages, it runs posteromedially to the clioana. The latter has, however, changed considerably. A narrow cleft rises dorsally from the center of the large trough in the palate. The medial sides of the ducti of either side open into this cleft. At their posterior ends, the ventral walls of the ducti tend to disappear too, so that the ehoanae open in part directly into the mouth, but at this stage their lumen is very small. Thus the choauae have attained their adult pattern, in that there are no longer any conspicuous grooves continuing posteriorly from them. The liistology is little changed. Figure 81 shows the vestibulum at this stage ; there is a single basal layer of columnar epithelium with basal nuclei. The remainder of the region is filled by an irregular mass of more or less isodiametric cells. In the sensory epithelium of the cavum nasi proprium (see Fig. 82), the Bow- man's glands are still rather sparsely scattered and small. The epithelium itself appears to be in three indistinct layers. First, there is still a thin inner enucleate zone with occasional mitotic figures, although these are becoming quite rare by this stage. The central zone, which is intermediate in thickness, is filled with verj- darkly-staining, oval nuclei. Basally there is a very thick area filled wdth circular nuclei. Although these nuclei do, in some cases, appear to be aligned in radial rows, there are no indica- tions of the columnar structure found in Jacobson's organ. Slight irregularities do occur in the basement membrane, but these show no real resemblance to even the earliest stages in the development of the columns. The only change in the non-sensory epithelium of the cavum is the greater development of the folds and small PARSONS: NASAL EMBRYOLOGY 193 crypts described above. At this stage they are found in all the non-sensory areas of the cavum, but are still most common and highly developed anteroventrally (see Fig. 83). The nuclei of the basal columnar cells are typically near the surface facing the lumen. In a few sections some of this epithelium appears to be simple, but in most places it is stratified, as shown in the figure. In the oldest available embryo, the vestibulum shows no changes in its gross form, but its histology indicates that the embryo must have been very close to hatching. It possesses thin epithelium, two or three cells thick, with some indications of cornification. Within it are loose masses of irregular cells, the remains of the solid epithelial plug of earlier stages. The naris extornus is not present on the slides; thus it cannot be told if the A-estil)ulum is actually open or not. The lumen of the cavum nasi proprium is greatly eidarged in all i-egions. Anteriorly, this results in a considerable decrease in the size of the ventral extensions. With the increased growth, the anterior part of the cavum retains its oval shape, but the greatest diameter is nearly horizontal in transverse section, with the former ventral end now lateral. The concha is little changed : it is a shelf -like horizontal projection of the lateral nasal wall which expands into a large cylindrical mass dorsomedially. The free posterior end is very small. Posterior to it there is a large Antorbitalraum which is quite similar to that of the preceding stage except in its increased volume. From the posteroventral corner of the Antorbitalraum, the ductus nasopharyngeus ex- tends posteroventromedially as before. The histology of the sensory epithelium shoAvs no changes, although it appears to be proportionately thinner, due to the great growth of the nasal area; its distribution is also like that described for the earlier stages. However the non-sensory areas are markedly different. There the epithelium is either simple or of a very low stratified type with the basal layer of cells low columnar. The development of the small crypts has progressed to the point where almost all of the non-sensory epithelium is thrown into folds. At the bases of these, and often elsewhere, the epithelium is simple and appears to consist entirely of goblet cells with basal nuclei and clear peripheral cytoplasm. Such crypts are very marked anterolaterally, as in the preceding stage, 194 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY but now reach their maximum development in the posterior por- tion of the Antorbitalraum. Shallower, but basically similar jirooves occur in the walls of the ductus nasopharyngeus. The choanae are also somewhat further developed. As in the preceding stage, the ducti nasopharyngei converge towards the midline, and there is, between them, a very narrow median cleft in the palate. However the anterior end of this cleft is no longer open ventrally to the oral cavity. The two ducti and the cleft join ventrally to produce a median tube, the whole resembling the letter E in transverse section (with the open side dorsal). A short distance posterior to the juncture of the ducti nasopharyn- gei, this tube opens into the mouth by a narrow median slit. This choanal slit Avidens posteriorly, and the nasal and oral cavities possess a moderately wide connection by this stage. The literature on the later stages in the nasal embryology of snakes is surprisingly meager. Rathke's (1839) great Avork on Natrix deals only briefly with the development of the nasal cavities. There are three other early papers based on studies of series of Natrix embryos. Fleischer (1878) and Beard (1889) are concerned primarily with the nature of Jacobson's organ, and neitiier author gives any detailed descriptions, so that their papers need no further comment here. However, Born (1883) does consider the general nasal embryologj^ at some length, and his work has long been the classic in this field. It remains so, not only because of its early date and its quality, but also be- cause there appear to be no more recent studies of a comparable nature. In general, Born's account agrees well with the description given above. There are no differences concerning the develop- ment of the duct of Jacobson's organ and its relationship to the nasal cavity and to the lachrymal duct, points discussed in some detail in his paper. Born gives shorter descriptions of the cavum nasi proprium in most of his stages; a review of his findings would add nothing to the preceding section of this paper. How- ever, Born's material included stages showing the fusion of the lateral and medial nasal processes more clearly than any avail- able for the present study. He states (pp. 194 and 195) that: "... der lappen- oder vorhangformige iiussere Xasenfortsatz . . . legt sich mit seiner Spitze und mit den grossten Theile seines PARSONS: NASAL EMBRYOLOGY 195 Vorden-andes an die Aiissenfliiche des iniieren Nasenfortsatzes vor den obereii Hillfte des zum Jaeobson'sclien Orj^an fiihrenden Loclies an und vorst'hmilzt mit derselben. Xur oben bleibt oine g-anz kleine piinktformige (Jft'nung, die Apertura externa, frei und dies aiieh nnr auf knrze Zeit, denn sehr bald legen sich aueh hier die Epithelflachen an einander nnd die Offnimg wird ver- legt. " The clioanae are described as broad slits which rnn oblicpiely posterolaterally. Along the line of fusion of the nasal processes. Born found some traces of their original epithelial covering, but these quickly disappear. In the embryos used for the present study, there is virtually no trace of such remnants. Several other ])apers mention or figure individual stages in the nasal embryology of snakes, although generally without detailed descriptions. The majority of these works are based on Natrir. Dieidafe ( 1904-1 !)05) describes a single embryo of 40 mm. total length (uncoiled) ; it is of the stage innuediately following the fusion of the nasal i)rocesses, and is, judging by his figure and discussion, jirobably identical in structure with the Thomnopliis embryo shown in Figure 49. Three pajiers considering the palate and choanae. Fuchs (1908 and 1911) and Tliater (1910), figure sections through the nasal cavities of Nafri.r embryos; they ap- pear to be in agreement with the observations recorded previously in this paper. Only one earlier paper deals with any embryonic material of TJiamnophis. Macallum (1884) studied a specimen of 6 mm. headlength, but gives almost no information on its anatomy. Apparently only two papers contain even a passing mention of other forms ; both of them are concerned primarily with the development of the skull, but do figure sections through the nasal area. Peyer (1912) studied Vipera. Judging by his figures, moderately late embryos of this genus resemble verv closelv those of Thamnopliis. Finally Pringle (1954) describes the ehondrocrania of four South African snakes, Lamprophis, Dasy- peltis, Causiis, and Hemachatiis. In each case he studied a series of embryos, but has few figures of the nasal region, generally all of one stage. All four genera appear relatively' similar to Thamnopliis, with the greatest differences occurring in Hema- chatus. In that genus the concha is attached dorsally as well as ventrally for an unstated distance at its anterior end ; thus the 196 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY anterolateral portion of the Sakter extends anteriorly as a con- spieuons pocket. Posteriorly the form of the concha shows an- other point of difference. There is a lateral projection of the concha which results in an increase in the medial twist at the ventral end of the Sakter. Although such a curving of that region is also seen in Thamnophis (see Fig. 54D), in Hemachatus the lateral conchal projection is supported by a simple lamellar extension from the conchal cartilage; no such support is present in Thamnophis. As was the case with turtles, there is considerable literature on the embryology of the ducti nasopharyngei and palates of snakes. Early papers considering this problem include those of Rathke (1839) and Born (1883) ; their conclusions are sum- marized in the later papers and need no comment here. The views of Born are, in the main, substantiated by Fuchs (1908). An- other viewpoint is put forward by Thater (1910), and this last paper is answered I)y Fuchs (1911) who reiterates his former stand. Much of the last two papers is polemic in nature, and the arguments are here reviewed only briefly. For details the reader is referred to the papers cited, especially Fuchs (1911). Fuchs' (1908) observations agree very well with those given above. When the choana is first formed, it stretches almost the entire length of the conchal zone of the cavum nasi proprium. During the later stages, Fuchs believes that fusion occurs in the ventral portion of the Choanengang, and Itetween the palatal "processes" lying on either side of the primitive choana. The anteroventral corner of the choana remains open as the duct of Jacobson's organ, but posterior to this the fusion progresses nearly to the end of the choana. Thus he considers the situation in the adult snake to resemble closely that in turtles, and uses the same terms in both groups (choanae reliquae and tegmen oris primarium commutatum). Since the ductiLS nasopharyngeus of the present paper is formed from the posterior part of the nasal cavity rather than from a part of the oral cavity, Fuchs suggests that it is better termed part of the Choanengang. Thater 's (1910) observations do not differ markedly from those of other workers; however his interpretations are quite radically opposed to those of Born and Fuchs. The only descrip- tive diff'erence requiring note is Thater 's denial that the choanae are long slits, even at a stage in Avhich Jacobson's organ is still PARSONS: NASAL EMimVOLOGV 197 connected to the medial wall of the nasal cavity. Rather ho states that they are very small apertures. Tlius Thater believes that there is no marked shortenino' of the choanae in the embryo, and denies that fusion has occurred. He explains the changes in the position of the duct of Jacobson's orjian by "eine aktive Umformung ... so dass das Jacobsonsche Organ durch Wachstum oral vom Choanengang weggelagert wird" (p. 489). Fuchs' (1911) answer adds little to the discussion. He points out that Thater 's youngest embryos are larger than those in which he had previously described the long choanal slits, thus denying the validity of much of Thater 's argument. The observations made in the present study seem to support Fuchs' contentions. Although in later stages the choanae are very small, they appear to be long slits in younger embryos. As reported by Fuchs (1911), they become not only relatively, but absolutely shorter as they develop. Therefore some fusion seems to be required. However, there is almost certainly some differen- tial growth involved in the early changes in the position of the duct of Jacobson's organ. One final paper must be considered here. Weber (1950) describes in detail the histological changes involved in the closure of the vestibulum and naris externus. His observations were made on a large series of Natrix embryos. The epithelial lining of the vestibulum and the central plug of tissue, the Fiill- (jewehe of Wel)er, are at first very similar. However, the former develops into columnar epithelium and becomes cornified, while the Fiillgewebe remains a mass of more or less isodiametric cells which eventually degenerate. They are sloughed away after hatching. Thus, a review of the literature shows that in only one genus of snakes, Natrix, has the later nasal embryology been studied in any detail. As might be expected from their close relation- ship, it resembles almost perfectly TJiamnophis, as here described. The amount of information available does not permit any con- clusions concerning the other groups of snakes, or generalizations for the suborder as a whole. Jacobson's Organ. Since the adult anatomy and embryology of this organ have been discussed in previous sections of this paper, only two points need to bo considered hero. There has 198 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY been little question of the homology of Jacobson's organ in the Sqnamata, and the history can be reviewed very briefly. The problem of the nature of the cellular columns forming the basal three-fourths of the sensory epithelium also merits some note. The first description of Jacobson's organ in a member of this order appears to be Rathke's (1839) account of the embryology of the structure which he termed the Nasendriise in Natrix. In 1854 Stannius noted the similarities between this "gland" in snakes and the Jacobson's organ of mammals, but not until 1872 was their homology definitely suggested. In that year Leydig described the organ both in lizards (1872a) and snakes (1872b), and stated that it was an accessory olfactory organ. Although in 1878 Parker referred to the structure as a nasal gland, Leydig 's proposed homology has been accepted by all later workers, and has, to my knowledge, never been questioned. The epithelial columns found in the dorsal wall of Jacobson's organ have long been known, and their nature was, for a con- siderable time, much discussed in the literature. More recently, however, they have received no attention, altliough many papers do figure them (e.g. von Navratil, 1926, and Pringle, 1954). Such columns are also known in certain lizards, and are con- sidered in that group by Beard (1889) and von Mihalkovics (1898). Figures 84, 85, 86, and 87 show stages in the development of these columns in ThaninopJiis; all are equally magnified. The sensory epithelium of Jacobson's organ becomes very greatly thickened as soon as the organ is formed as a distinct medial pocket. During the later development, the epithelium continues to thicken, but at a much reduced rate. Early in its develop- ment, the basal surface of the epithelium becomes irregular, and the columns are first formed as small bumps pushing into the surrounding mesodermal tissue. A network of capillaries ap- pears between these anlagen of the columns, and is found at their base in all subsequent stages. With further growth, the columns increase greatly in length, and the undivided epithelium lining the lumen of Jaco])son's organ becomes thinner, l)oth relatively and absolutely. Another change which occurs is that the columns become thinner in cross-section, and hence more numerous. How this occurs is not known ; none were seen dividing or groov- ing out between already existing columns in the present study. In PARSONS : NASAL EMBRYOLOGV 199 cross-section the columns appear polygonal. Tlioy do not possess a lumen. The material used in the present work is not suitable for de- tailed histological studj'. At all stages there is a very narrow zone in the sensory epithelium facing the lumen in which there are few nuclei, but in Avhich mitotic figures are common. The remaining portion of the epithelium contains many densely- packed circular nuclei in the early stages. Later a small central zone with more darkly stained oval nuclei develops between the thick basal and clear peripheral portions. The entire length of the columns lies completely within the basal zone as does part of the undivided epithelium. Precise tracing of nerve fibers was not possible in the available material. Certainly most of these leave the ends of the columns, but whether or not some run between them after leaving the epithelium at their bases could not be determined. Thus the histological picture is identical with that of the olfactory epithelium of the nose, except that the epithelium of Jacobson's organ is somewdiat thicker, and its basal portion is divided into these cellular columns with capillaries and other nu^sodermal tissue l)etween them. The earlier literature contains two theories on the nature of these columns. The more generally accepted one was that they are formed of ganglionic cells. Leydig (1872a) first proposed such a theory, and Wright (1883), Macallum (1H84), and Beard (1889) all supported it. The other theory is that the columns are simple tubular glands. Born (1883) proposed such an explanation on the basis of his embryological studies, but no other workers have concurred with his opinion. The important paper by Retzius (1894) should have settled the controversy ; however, it seems generally to have been over- looked. By a study of Golgi preparations of the epithelium of Jacobson's organ, Retzius determined that the columns are an integral part of the epithelium which is completely comparable with that of the sensory areas of the nose. It contains two cell types, sustentacular and sensory. The former are long thin cells extending the entii'c thickness of the epithelium. Their nuclei lie in a zone near the lumen, thus corresponding to the darker oval nuclei mentioned previously in the description of the material used in the present study. The sensory cells have round nuclei 200 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY located at various levels in the epithelium. One process from each cell runs to the lumen, and another process emerges from the opposite surface of the epithelium and continues to the brain as a fiber of the olfactory nerve. There is no sign of any glandular structure, and Retzius mentions no ganglion cells. A few later papers, including Leydig (1897), von Mihalkovics (1898), and Zuckerkandl (1910b), further discuss the problem without reference to Retzius' discoveries, although his work may be cited in their bibliographies. Their works add no new informa- tion on the columns. Thus there is no evidence for the presence of any secretory cells in the Jacobson's organ of snakes, and the occurrence of ganglionic cells in any numbers would be most surprising in view of present knowledge of the development and structure of the olfactory (including the vomeronasal) nerve. The observations made in the present study add only one point of interest. A capillary network is present at all stages in the development of the columns, not, as was stated by Beard (1889), only after tliev are well formed. Whether or not there is anv causal rela- tionship between this network and the form of the columns, as suggested by Wright (1883), remains unknown. Order Crocodilia Adult A7iatomy. The crocodilians possess by far the most complex nasal cavities of any reptiles. For the present study three immature specimens of Alligator were dissected ; one of these is shown in Figure 59. Since the observations agree well with the accounts found in the literature, only one description is necessary. Major papers considering this group include those of Kathke (1866; Alligator, Caiman, Crocodylns, and Tomisto- nia), Gegenbaur (1873; Alligator), Solger (1876; Crocodylus), Hoffmann (1879-1890; Crocodylus), Nemours (1930; Alligator), and Bertau (1935; Alligator, Mclanosuchiis, and Crocodylus). Matthes (1934) and von Wettstein ( 1954) present rather detailed summaries of earlier work. As with the other groups which have been described, there are many problems concerning terminology. In the following de- scriptions tliat of Bertau (1935) will be followed unless other- wise noted. Tlie diagram shown in Figure 60. which is based in PARSONS : NASAL EMBRYOLOGY 201 3 o ^ =■ > ^ OS 2 ■r; CO > OS ''I ti P is — . o S p. CS o t-l ^H o c3 ^ ^ ^ c be »; 5 1 B .2 202 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY large part on Bertau's Abbildiing 18, includes most of the regions to be described. The nose may be divided into tliree main parts comparable to those in other reptiles : an anterior vestibulum. a large cavum nasi proprium, and a ductns nasopharyngeus leading to the mouth. The short vestibulum (Vorhdhle of Bertau) is a vertical tube leading from the naris externus to the anterior end of the long horizontal cavum nasi proprium. The ducts of the glandula nasalis externa enter its posterior wall well ventrally. The cavum nasi proprium is a large complex chamber, con- taining several concha-like projections of the lateral wall, and including a series of well-defined recesses or sinuses. It has often been treated as two somewhat separate parts, one lying anterior to the nasal end of the ductus nasopharyngeus and one lying posterior. The former has sometimes been considered as part of the vestibulum, as by (iegenbaur (1873). Bertau points out the incorrectness of this identification and treats the entire cavum as one region which he refers to as the Hanpthohle ; this interpre- tation is followed here. There are three projections of the lateral wall into the cavum which have been referred to as conchae. CJegenbaur (1873) and Solger (187()) proposed a restricted definition of this term, and considered only the middle projection as a true concha, call- ing the other two pseudoconchae. Almost all the more recent workers have used the terms preconcha, concha, and postconcha. The preconcha is a quite pronounced convexity of the lateral wall running from a short distance posterior to the vestibulum to the start of the ductus nasopharyngeus, thus making the anterior half of the cavum crescentic in transverse section. The concha (^ middle or true concha) is more complex, con- sisting of a thin projection from the lateral wall which extends medially and then posteriorly and ventrally. It stretches from the region of the ductus nasopharyngeus approximately two- thirds of the distance to the posterior wall of the cavum. In the Alligator material dissected for the present study the anterior two-thirds is essentially horizontal and the posterior third turns dorsally (see Fig. 59). In a late embryo of this genus, Bertau (1935) shows the posterior portion as larger and more distinctly separated from the anterior by a groov(» across the surface of the concha. He also shows a horizontal groove along the medial PARSONS: NASAL EMBRYOLOGY 203 face of the anterior portion. In Crocodylus, the concha is simph'r in havinf; a fairly straiprht line of attaehnient which runs diag- onalh' posterodorsally from the posterior end of the preconcha (Bertau). In both genera the anteroventral end of the concha is connected to and continuous Avith the posteroventral corner of t he preconcha. The post concha is a very prominent convexity in the postero- lateral wall of the nose. Its anterior end, which lies partly ventral to the posterior part of the concha, is pointed while its posterior end is rounded so that the projection is wedge-shaped in medial view. Crocodilians possess a large number of recesses and sinuses connected to the nasal cavity. The first of these, the recessus preconchalis, enters the cavum nasi proprium betw'een the pre- concha and the concha just dorsal to their connection. It expands, especially anteriorly, under cover of the posterior end of the preconcha. The larger cavity lateral and ventral to the concha is the recessus extraconchalis which opens into the cavum by a diagonal slit between the concha and postconcha. From the ventrolateral corner of the i)osterior wall of this recessus a small duct-like passage, the sinus postturbiualis, runs posteriorly to the posterolateral corner of another large cavity. The last lies entirely within the postconcha and is termed only the Hohle in der Postconcha by Bertau. All of these chambers lie within the cartilaginous nasal capsule. Three more sinuses lie outside the nasal capsule. The most anterior of these is the sitiiis maxiUaris which enters the cavum ventral to the anterior end of the preconcha and expands into a longitudinal cavity in the maxilla ventral to the preconcha. The recessus caviconchalis is a rather digitiform sinus lying just anterior and parallel to the recessus extraconchalis. Its opening to the cavum is ventral to the anterior portion of the concha. In Crocodylus, Bertau found a small connection between this recessus and the Hohle in der Postconcha ; such a connection is not present in the other genera which have been studied. Finally, ventral to the postconcha lies the hintere Interale Nchenhohle of Bertau. It possesses a small connection with the cavum ventral to the postconcha and is of considerable extent ventrolateral to the nasal capsule. This sinus is well developed in AUigator; Ber- 204 liULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY tau also found it in Melanosuchus, but reports its absence in Crocodylus. The lachrymal duct also enters the cavum nasi proprium. Its opening is ventral to the center of the preconcha. In crocodilians this duct is much enlarged and bears an anterior extension, the saccus lacrimalis, which resembles another small paranasal sinus. The ductus uasopharyngeus is a long simple tube leading from the floor of the cavum in the region of the anterior portion of the concha to the posterior end of the extensive secondarj^ palate. Its position is well shown in Figure 59. The ducti of opposite sides are separated by a thin septum throughout their entire lengths. The pattern just described appears to be relatively constant in all the living forms that have been studied. Crocodylus has a longer snout than Alligator, and thus the anterior parts of the nasal cavity are relatively longer, but the differences are very slight. Melanosuchiis closely resembles Alligator in all regards. The nasal anatomy of Gavialis is known only from a figure in Briihl (1886) ; it appears to have the same pattern as the other genera. However, the crocodilian pattern is quite different from that of any other group, and the homologies of several of the structures are debated. Thus the crocodilian sinus maxillaris is a sinus within the maxilla, but is almost certainly not phylogenetically related to the mammalian sinus maxillaris. Both the other extracapsular sinuses, the recessus caviconchalis and the hintere laterale Neben- liohle (e.g. Bertau), are also within the maxilla and have been proposed as the true homolog of the mammalian sinus. However, since paranasal sinuises are not of frequent occurrence in reptiles and are absent in pelycosaurs (A. S. Romer, personal communi- cation), it seems almost certain that these cavities arose inde- pendently in crocodilians and mammals. Therefore, there are probably no strict homologies between the sinuses of these two groups. The hintere laterale Nebenhohle of crocodilians has also been compared to the orbital sinus of birds, but here again there is no general agreement (see Matthes, 1934). The homologies of the conchal formations will be considered in the general discus- sion at the end of this paper. A final consideration is the extent of the sensory areas of the crocodilian nose. Since the onlv relatively mature material used PARSONS: NASAL EMBRYOLOGY 205 in the present study was grossly dissected, no original observa- tions were made on this point. In general, as in the other reptiles, the posterodorsal half of the cavum nasi proprium is said to be sensory and the anteroventral part is not (Bertan, lf)85, and others). Early Embryulogy. The only previous i)aper on the early embryology of the crocodilian nose is that of Voeltzkow (1899), who describes a series of embryos of Crocodylus niloticus (C. madagascariensia of Voeltzkow) in some detail. Unfortunately, the sizes of the embryos are not given. Rough approximations can be obtained from the figures of the entire lieads, dividing the length of the figures by the stated magnifications ; such are used in the following discussion. V^oeltzkow, in a series of excellent figures (his plate IX), shows the external appearance of the nasal placodes and pits, in the stages of headlengths 2.1, 2.4, and 4.0 nun. they appear as small circular depressions with a slightly raised rim, first on the lateral and then the ventrolateral surface of the snout, fairly near, but not at, its anterior end. In larger embryos (6.0 and 4.4 mm. headlength^), the nose has become a slit which is rounded anteriorly, and posteriorly passes into the developing mouth cavity. It is ventrolateral and nearly terminal in position. Embryos with headlengths of 5.8 and 6.5 mm. have the nares externi and choanae separated by the fusion of the lateral and medial nasal processes. The nares externi are nearly vertical slits which lean slightly towards the midline at their dorsal ends. At this stage the secondary palate has not yet formed. In the text, Voeltzkow describes briefly the early nasal embry- olog}', using very diagrammatic drawings of transverse sections of several stages. These stages agree quite closely with those seen in AUigator in the present study, and will be mentioned further in the following descriptions. The early embryology of the nasal cavities of Alligator, which has to my knowledge never before been studied, is here described in some detail. In the youngest embryos displaying any nasal differentiation, there are distinctly thickened nasal placodes, covering most of the sides of the snout anterior to the eyes. 1 The former of these embryos seems very large for its stage of development ; either there is considerable variation, or Voelt/.kow's figure 70 on plate IX is mislabeled as to magnification. 206 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY f— NPL I— AON Figuio Gl. Alligator 059 (?; AMR). Frontal section tliiouo;]! the nasal .lien (section 15). 115x. Figure 612. Alligator 1-7 (?; AMR). Frontal sections through the nasal area. A, section 394; and B, 10 sections ventral to A. 77x. Figure ()8. Alligator 1-14 ( .' ; AMR). Transverse sections through the nasal area. A, section 18(5; I>, (i sections i)Osterioi' to A; and C, 5 posterior to B. 77x. Figure G4. Alligalor L-'20 (?; AMR). Transverse sections through the nasal area. A, section 264; B, 9 sections posterior to A; and C, 8 posterior to B. 77x. PARSONS: NASAL EMBRYOLOGY 207 1'liey are still convex, conforming to the general outline of the head without any indentation. The placode is op to three nuclei in depth in one euil)ry(), and tive in a second, Avhile the general body ('i)itli('liuni has but a single layer. There is, however, no histological distinction between the i)lac<)dal and general body e[)itheliuni other than the tliickness, and Ibc two mei-ge without any sharp boundary. Anteriorly, the transition is siiarper than elsewhere. The epithelium jiossesses a clear basement membrane throughout. The major difference between the placodes of the two available series of tlds stage is in their relation to the brain : in the younger the anterior half of the placode is in direct contact with it (see Fig. 61), while in the older, mesodermal tissue has pushed forward and separated these two structures. In the hitter embryo a small bud of cells is formed which runs from the center of the placode towards the l)rain. it represents the anlage of the olfactory nerve, and its form and sid)sequent dv- velopment are described in the section on that nerve. The placode next starts to invaginate to form the nasal pit. Early stages in this process, however, are better described as indented i)]acodes than pits (see Figs. 62 and 68). Tlie indenta- tion is a sliallow symmetrical cone. The placodes are slightly more restricted in area than in the preceding stage, being ventro- lateral and subterminal on the snout. Voeltzkow's (1899) earliest embryo (2.1 mm. headlength ; his fig. 1, p. -13) is of this or possibly the following stage. There is still no histological differentiation of nasal versus general body epithelium except in thickness. The indented sec- tion of the i)lacode, somew'hat dorsal to its center and hence lateral in position, shows five or six layers of nuclei, while most of the remainder of the placode is two or three layers ; the general body epithelium remains a single layer. Thus the transition is sharpest dorsally and most gradual ventrally, with the anterior sharper than the posterior; however, in no case is there any definite boundary. The basement membrane is distinct under the placode, but elsewhere it disappears in the older series. This membrane is not smooth, but has low gentle undulations. The next stage is that of a small conical nasal pit which is still quite symmetrical without any noticeable dorsal extension (see Figs. 64 and 65). Its depth is equal to its width in the youngest series and, again, in the oldest ; it is less in the inter- 208 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY mediates, due to a greater growth in area than in depth for a short period. Although the placode does not greatly increase in size, the much thickened and indented portion comes to assume a larger percentage of the total area, with only the ventral NPT Figure 65. AUigalor X-no. 4 (?; AME). Frontal sections through the nasal area. A, section 132; B, -4 sections ventral to A; and C, 7 ventral to B. 42x. l"Hgure 66. Alligator F-7 (?; AME). Frontal sections through the nasal area. A, section 107; B, 3 sections ventral to A ; C, 7 ventral to B; and D, 8 ventral to C. 31x. Figure 67. Alligator F-9 (?; AME). Transverse section through the nasal area (section 118). 31x. third remaining slightly thickened. Histologically, there is still little differentiation. All the nuclei appear the same, but the cytoplasm stains more densely in the epithelium, both nasal and PARSONS : NASAL EMBRYOLOGY 209 epidermal, and the brain than in the interveninjj mesodermal tissue. The walls of the nasal pit are everywhere the same, with a slight tendency for the concentration of nuclei toward the deep surface, but with more mitotic figures periplierally. The nasal pit next starts to extend dorsally, probably in i)art by further invagination, and certain!}^ in part by the antero- ventral extension of the lateral nasal process (seitliche Nasen- fortsatz of the German literature). Figure 66 shows this stage in frontal section and P'igure 67 gives some idea of the height of the pit, although the section is not perfectly transverse. The pit now extends slightly anterior as well as dorsal to the naris so that in frontal section it appears as a parallelogram with the acute angles approximately 45 degrees. At the ventral end, the posterior wall tends to ])ecome more confluent with the epithelium of the head than anterodorsally, and the nasal pit thus appears more as a simple notch when seen in section. The naris is lateral anteriorly, but at its posterior end is on the ventral surface of the snout ; the pit therefore appears deepest posteriorly. Histologically, there are few changes. The walls of the pit are evenly thickened and are, in one case, up to ten nuclei deep. The boundaries of the nasal epithelium are fairly sharp, based on thickness, but are marked by no other changes. Posterior to the nasal pit, the epidermal epithelium is now two nuclei thick; elsewhere it has a single layer. Tlie growth of the pit and also of the brain is reflected in the approximation of these two organs, with considerably less intervening mesodermal tissue than in the preceding stage. With the further growth of the nasal pit and the lateral nasal process, the pit becomes a narrow but deep, dorsally directed groove on the side of the snout (see Fig. 68). Voeltzkow's (1899) CrocodyJiis embryo of headlength 2.4 mm. (his fig. 2, p. 43) is of this stage. In Alligator the pit still has its apex pointed; it is more rounded in Voeltzkow's Crocochjlus. The naris is widely open at its anterior end where it is lateral and subterminal. but posteriorly it becomes a narrow slit along the lateral edge of the ventral surface of the snout. Anteriorly and posteriorly, the walls of the pit are very nearly vertical, as is the lateral wall, while the medial slants slightly latcralh'. Tlic anterior portion of the pit is slightly curved when seen in transverse section, with the convexity medial. The lumen is 210 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY widest ventrally and anteriorly, but is, in most cases at least, open throughout. Histologically, there is some differentiation, but it is not great. The epidermal ejMthelium of the head is now thicker, having two or three layers of nuclei. In at least one series, it has a distinct basement membrane. There is thus no sharp boundary between it and the nasal ei>ithelium, although the latter is every- where considei-ably the thicker. A major change is the develop- Figuie 08. Allifiutor X-1'6 (?; AMK). TraiLsverse sections through the nasal area. A, section 4 on slide M; B, 9 sections posterior to A; and C, 11 posterior to B. 3(ix. Figure {)9. Alligator H-22 (?; AMEj. Transverse sections through the nasal area. A, section 335; B, 4 sections posterior to A; C, 21 posterior to B; I), 14 posterior to C; and E, 11 posterior to D. 36x. ment of differences between the sensory and non-sensory portions of the nasal epithelium. The former, the dorsal third of the pit. especially anteriorly and medially, is thicker and lacks a base- ment meml)rane. The non-sensoi-y portion has such a membrane and shows a tendency for the concentration of its nuclei toward the exposed, primitively lateral surface. This tendency is also PARSONS : NASAL EMBRYOLOGY 211 present, but far less marked, in the sensory areas, in the older series of this sta(;e tliere is a second area of much thickened epithelinin : a hi\nd runnin*;: alonan typically develops and apparently represents the earliest aida^-e of that organ. However, at this stage it shows no indentation ; in fact, it may bulge out into the lumen of the pit. Another embryo represents essentially the same stage, but the development of Jacobson's organ renders separate descri])tion necessary, and is shown in Figure 69. The pit is as in the preced- ing stage, although the lumen is slighth' wider than before. The NPT Figure 7U. Tniiisverse section througli the lui.sal ;ire;i o!' a Croroil i/l ii-- eiiihiyo (after Voeltzkow, 1899). 2(ix. size of the lumen seems to vary from series to series and may represent either very small differences in distortion in the sec- tioning or a somewhat variable feature of the nasal embryology in this species. Histologically, there are no changes, thus still no nuclear difterentiation. In the nasal epithelium, the layer of nuclei closest to the lumen shoAvs many mitotic figures. Jacob- son's organ is represented by a distinct, though very shallow, groove along the medial wall of the nasal i)it in the same position as the thickened band of epithelium noted above (see Fig. 6nB). It is apparent only in tlie anterior half of the nose. Since it is in the proper location, shows the expected form, and gives rise to many nerve fibei-s, there seems little reason to doubt that this groove is the .lacobson's organ of AUigaior. The literature on 212 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY this problem is reviewed separately, however, and further dis- cussion is included there. Voeltzkow (1899) describes two Crocodylus embryos, head- len^^ths 4.0 and 6.0 mm. (his figs. 3 and 4, p. 44; see second para- graph of this section), which are of a comparable stage. The first of these possesses a Jacobson's organ very like that just described, but much more conspicuous (see Fig. 70). Whether (_CNP Figure 71. Alligator 24: (10.0; AMR). Frontal sections through the nasal area. A, section 367; B, 32 sections ventral to A ; C, 15 ventral to B; D, 12 ventral to C; E, 12 ventral to D; F, 11 ventral to E; and G, 7 ventral to F. 42x. this difference in size is a generic one between Crocodylus and Alligator or merely due to Voeltzkow 's having obtained an embryo showing more nearly the maximum development of the organ than any of those used in the present study, I am unable to say. In the very slightly more developed series shown in his figure 4, Jacobson's organ seems to have disappeared almost completely, although there is still a very slight indentation in the PARSONS : NASAL EMBRYOLOGY 213 same area. It appears, however, to liave thinner epithelium than in the preceding series. Two other differences between these series and the Alligator material are : the greater development of the lateral nasal process in Crocodyliis, and the development of a dorsomedial outpocket- ing of the nasal epithelium in the posterior part of the nose in Voeltzkow's material. There is nothing corresponding to the latter in the Alligator embryos studied. This outpocketing pro- duces an apparent medial pocket; however, it seems dorsal to Jacobson's organ. Whether or not the two are connected is not mentioned and cannot be told from the figures. On the basis of comparison with the Alligator material and Voeltzkow's figures of later stages, it appears unlikely that they are related. The last series to be described here is shown in Figure 71. Unfortunately the plane of section is rather oblique, so that in all the figures the lateral edge of the section is slightly ventral to the medial. The nose is still a fairly simple pit, but the lateral and medial nasal processes have fused to separate the naris externus from the primitive choana. This is apparently a very recent occurrence; a double layer of epithelium with no inter- ^'ening cavity extends about half way ventrally from the nasal cavity toM'ard the roof of the mouth to mark the line of fusion. Both nares and choanae are widely open and well ventral. The concha (= middle concha) is starting to form laterally; it is directed somewhat posteriorly so that the nasal cavity, especially dorsally, has a lateral twist at its posterior end. The nasal pits are farther apart posteriorly than anteriorly (the choanae are widely separated at this stage), and each is slightly curved with the convexity lateral. There is no Jacobson's organ visible in this series ; a transverse section might show a small remnant, but this seems unlikely, since there is no increase in the width of the lumen in the region where it would be expected. The lumen is widest about midway dorsal at its posterior end, but there is no sign of any dorsomedial projection into it at that point. Histologically, the only real differentiation within the nasal epithelium remains that of thickness. It is thickest dorsally and medially, and thin- nest at the choanae. Voeltzkow's (1899) 4.4 and .3.8 to 6.5 mm. headlength stages (his figs. 5 and 6, pp. 45 and 46) are comparable. The former is slightly less developed than the series described above; in his 214 JJULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY figure, the nasal processes do not appear to be fused as yet, although the lumen is obliterated ventrally between the nares extern! and the choanae. The second embrj'O seems to be at almost exactly the stage shown in Figure 71. Both Crocodylus embryos appear to have a somewhat narrower lumen than that of the Alligator studied, but this is a very minor difference. A far greater one concerns the development of the dorsomedial outpocketing mentioned above. In Voeltzkow's figure 5b it has become very prominent so that the nasal cavity is Y-shaped. Such a structure is not figured in his succeeding stage. The medial arm of the Y appears to be the pocket considered to be Jacobson's organ by Shiino (1914) in his Crocodylus embryo of headlength 5 mm. (see his fig. 24, p. 319) ; this is an older embryo and is discussed in the section on Jacobson's organ. However, in Voeltzkow's figure 5 there is no sign of the more anterior and ventral groove here considered to represent that organ. AVhatever this outpocketing may represent, it is appar- ently present only for a very limited time in Crocodylus and probably not at all in Alligator, although it could be formed in the latter genus for so short a period that none of the embryos used in the present study possessed it. Later Embryology. Tlie later stages of the nasal embryology have been very well described in several crocodilians, and no descriptions are necessary here. Early accounts include those of Rathke (1866), Meek (1893 and 1906), Reese (1901), and Shiino (1914). More important are the detailed studies by Meek (1911; Crocodylus) and Bertau (1935; AlUgaior, Mclanosuchus, and Crocodylus) ; both of these papers include excellent figures of models of several stages. Finally, Fuchs (1908) described the formation of the ductus nasopharyngeus and secondary palate of Crocodylus in detail. For the present investigation, eleven Alligator embryos were studied; they did not differ appreciably from those figured and described by Bertau (1935). The later stages shoAV a gradual development of the complex pattern of conchae and recesses and the long ductus nasopharyngeus found in adult crocodilians. Jacobson's Organ. The literature concerning Jacobson's organ in the crocodilians is quite extensive. However, there is much disagreement, especially concerning the earlier stages, as to Avhat structures, if any, represent the rudiment of this organ. PARSONS: NASAL EMBRYOLOGY 215 The earlier descriptions of crocodilian noses do not mention Jacobson's organ at all, and neither the texts nor the fif?ures give any indication of one. Snch works include those of Rathke (1866) who studied a variety of forms, Gegenbaur (1873) who studied a young- Alligator, and Solger (1876) who employed Crocodylus. Beard (1889) was unable to find any trace of Jacob- son's organ in an embryo crocodile, and Hoffman (1879-1890) cites its absence as a major difference between crocodilians and lizai'ds. However, in 1891 Howes started a great search for this organ by reporting a small vesicular structure (which he thought might represent a vestigial Jacobson's organ) at the anterior end of the vomer in a young Alligator. He was led to consider the possibility of finding such a vestige by a study of the skull of Melanosuchus in which the vomers extend farther anteriorly than in the other crocodilians and contain a small cavity. From a comparison with mammals, Howes believed that this cavity should contain Jacobson's organ. Unfortunately, he lacked preserved material of Melanosuchus in which to check this. The first description of Jacobson's organ in embryonic croco- dilians is that of Sluiter (1892). In his 20 mm. Crocodylus, the nasal cavity was, amazingly, exactly like that of a lizard with a very well developed Jacobson's organ opening into the roof of the mouth just anterior to the choana. In a younger stage (16 mm.) he was also able to find the anlage of that organ. Meek (1893), in the first of a series of papers, presents a rather confusing picture, based on a series of Crocodylus embryos, in which he admits that, had it not been for Sluiter 's paper of the preceding year, he would consider Jacobson's organ absent. In his youngest stage (5.0 and 5.75 mm. headlength), he considers a very small depression in the anteroventral Avail of the ductus nasopharyngeus to represent a rudimentary Jacobson's organ, although he gives no reasons for this conclusion. In a slightl.y larger embryo (7.0 mm. headlength), he describes what appears to be an entirely different structure as Jacobson's organ. It is a slight fold enclosing a small space on the medial wall of the cavum nasi proprium. An embryo of headlengtli 7.75 mm. showed no trace of this, nor was he able to find any remnant of the organ in still larger series. He concludes that at least some trace of Jacobson's organ is pi-obably present in all croco- 216 15ULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY dilians, but that in Crocodylus it is present only for a very short period during development. Rose (lS93a) clarifies the situation somewhat. His rather extensive Crocodylus material included much of that used by Meek and Sluiter. By studying organs other than the nose, he was able to show that the critical embryo (20 mm.) of Sluiter 's was, in fact, not a crocodile at all, but a lizard. In the Crocodylus embryos, he recognized the structure which Meek (1803) con- sidered to be Jacobson's organ, but denied that it had any rela- tion to that organ. Rose, instead, describes a small ])it on the posterodorsal wall of the ductus nasopharAaigeus (which was noted in passing by Meek) as the true anlage of Jacobson's organ. It is shown in Figure 72 (his fig. 1). This pocket was found iu embryos of Figure 72. Sagittal section through the nasal area of a •" mm. headlength Crocodylus embryo, showing the pocket which Eose (lS93a) considers to represent Jacobson's organ (after Rose, 1893a). from 5.0 to 12 mm. headlength and was also macroscopically visil)le in a young crocodile (headlength 42 mm.). No reason is stated for his designation of this })it as Jacobson's organ other than its general position near the primitive choana. Von Mihalkovics (1898), having no material of crocodilians available, used Rose's (1893a and b) descriptions, but reached very different conclusions. lie calls attention to the dorsomedial point of entrance of the duct of the glandula nasalis externa into the nasal cavity, and postulates that the nasal end of this duct may be the true rudiment of Jacobson's organ in this group. In his detailed study of the early embryology of the crocodilian nose, Voeltzkow (1899) never mentions Jacobson's organ at all. PARSONS: NASAL EMI5KYOLOGY 217 However, an early stage of Crocodylus, shown in Figure 70 (his fig. 3a), has a distinct concavity in the ventral part of the median wall of the nasal pit which many later workers consider to be the anlage of Jacobson's organ. In later stages, after the nasal j)rocesses have fused, this small medial deepening appar- ently disappears ; it is certainly not evident in Voeltzkow 's figures of later stages. Probably the most generally cited and accepted ideas on Jacobson's organ in the crocodilians are those of Peter (1901). Ilis descriptions are based primarily on Voeltzkow 's (1899), although most of the early works are cited. Peter was the first to call attention to Voeltzkow 's figure 3a Avliich he believed to show this organ "deutlich in typischer Lage und Form" (p. 43). Sluiter's (1892) and Rose's (1893a) theories are definitely denied and some doubt is expressed concerning Meek's (1893), although in the last case Peter could not be certain. His doubt is based in large part on the absence of sensory epithelium in the area, which he considers an important, if not diagnostic, characteristic of Jacobson's organ. Iveese (1901) describes very briefly one moderately well de- veloped stage in Alligator, unfortunately without giving the size. He apparently accepts Rose's (1893a) identification of Jacob- son's organ which is here described as a tubular mass of cells, without a lumen for most of its length. This so-called Jacobson 's organ is made up of "typical mesoblast cells" (p. 462) similar to those of the ductus nasopharyngeus. It runs posteroventrally medial to the nasal cavity and enters the dorsal surface of the ductus. From the figures it would seem that the point of juncture is well posterior to the separation of the ductus nasopharyngeus from the cavum nasi proprium — thus in the portion which, ac- cording to Fuchs (1908), represents part of the primitive palatal rather than the nasal wall. The length of this organ seems far greater than that reported by Rose (1893a) or that of the embryos used in the present study. Dieulafe (1904-1905) had available only an adult Crocodylus; he could find no trace of a Jacobson's organ. In his review of the literature he ventures no opinion on the various suggested em- bryonic anlagen. In his paper on palatal embryology, Fuchs (1908) makes only passing reference to the absence of Jacobson's organ in croco- 218 BULLETIN: MUSEHJM OF COMPARATIVE ZOOLOGY dilians, although he does mention the possibility of an anlage of this organ in early stages, citing Voeltzkow's figure men- tioned above. His material was mainly late stages of Crocodylus from Voeltzkow's collection. Zuckerkandl (1910a), in connection with his studies on the olfactory bulbs, searched for Jacobson's organ in both Alligator and Crocodylus; he found no trace of it nor of the associated nerves and accessory olfactory bulb. However, he did not have early embryonic stages. In another paper, Zuckerkandl (1910b) discusses the various earlier accounts briefly, dismissing all re- ports except that of Peter (1901) based on Voeltzkow's figure. He considers further investigation to be necessary before croco- dilians can be thought definitely to possess even a transitory anlage of Jacobson's organ. The next major paper is that of Meek (1911) ; the main points of this paper are also present in abstract form in his note of 1906. In this paper he presents Ijetter figures of the same stages as were used in his earlier (1893) paper and rather cautiously reiterates his former conclusions. The figures of his 5.0 mm. headlength stage (his nos. 5 and 6, p. 363) show a small pocket in the floor of the cavum nasi proprium, just anterior to the ductus nasopharyngeus, which he calls Jacobson's organ. He then rather qualifies his conclusion (p. 364) : "At all events, whether this be the representative or not of the primitive median moiety of the nose, there is nothing else at this stage which can be identified with a Jacobson's organ." It is in the non -sensory area. However, the same figures also show a small pocket in the anterior wall of the ductus nasopliaryngeus which appears to me more like the one described and figured in his earlier paper. He could find no trace of the organ in embryos of headlength 12.0 mm. or larger. Meek supported Peter's (1901) conclusions, and admits that further study is needed to show whether or not the pit described by liim has any relation to that figured by Voeltz- kow. Shiino (1914) is concerned primarilj^ with the development of the skeletal structures in the head, but does include a little information on the nose. In a 5.0 mm. headlength Crocodylus he briefly describes and figures a medial pocket which he believes to be that figured by Voeltzkow (1899) and mentioned by Peter (1901). Shiino considers it the true Jacobson's organ. The I'AKSONS : NASAL EMBRYOLOGY 219 orientation of liis fiynre (no. 24, p. ;J19) is rather confusing, but the embryo is considerably more advanced than Voeltzkow's (the nasal processes have already fused), and the pocket appears to be the more i)C)sterior one shown l)y Voeltzkow and discussed in the section on the early embryolooy of the nose. Shiino was apparently unable to find any sign of Jacobson's organ in more advanced embryos. Nemours (1930) studied a young (155 mm. in length) Alli- gator and found no trace of a Jacobson's organ. He mentions its absence in both young and adult material although he cites only the single specimen. Matthes (1934) very briefly reviews all the earlier literature. Jacobson's organ is said to be definitelj^ absent in the adult croco- dilians. After disagreeing with the opinions of Sluiter (1892), Meek (1893), Rose (1893a), and Reese (1901), Matthes presents Peter's (1901) and Shiino 's (1914) work, concluding (p. 921), "Danach liisst sich nicht mehr bezweifeln, dass sehr junge Krokodilembryonen ein Jacobsonsches Organ in typische Weise anlegen." However, he describes as still unknoAvn the fate of this anlage in later embryos. Similar views are expressed in other recent reviews, e.g. von AVettstein (1954) and Gerard (1954). Finally Bertau (1935) was unable to find Jacobson's organ in any of his material (late embryos of Alligator, Melanosuchus, and Crocodylus). He disagrees with the ideas of most earlier workers, and cites Peter's (1901) opinions without comment, concluding that it is doubtful that crocodilians possess any Jacob- son's organ. Of the Alligator embryos used in the present study, only one possesses an anlage of Jacobson's organ (see Fig. 69), although two slightly younger series show a slight thickening of the epi- thelium in the same area. This anlage, as described in the section on early nasal embryology, is definitely the same as that figured hy Voeltzkow (1899), and first considered to be Jacob- son's organ by Peter (1901). There is no trace of it in later stages. Since it is in the usual position, has the usual form, and gives rise to a large part of the olfactory nerve in early embryos, there seems to be no reason to doubt the homology. The organ is thus formed in early stages of Alligator, as in most other amniotes, as an inpocketing of the medial wall of the nasal pit, but it never develops to any great extent, hi fact it very rapidly 220 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY disappears. The situation would appear to be the same in Crocodylus, judging from Voeltzkow's (1899) figures. There, too, it is not shown after the lateral and medial nasal processes have fused. None of the other structures described as Jacobson's organ in crocodilians appears to have any relation to that organ. The small vesicular structure found by Howes (1891) is reported only by him and from nearly adult material ; what it may repre- sent I do not know, but it seems unrelated to any of the various pits described in embryonic material. Meek (1893, 1906, and 1911) presents what seem to be several structures. The fold in the medial wall of the nose (his 7.0 mm. headlength stage) has been reported by no other workers, and it is possibh' if not probably formed by folding or breakage of the sections. A small depression does occur, in some series, in the ventral surface of the cavum nasi proprium or ductus nasopharyngeus, but none is present in the stages immediately following the disappearance of tlie typical anlage, so that aside from its being in the general area where Jacobson's organ might be expected, there is nothing to support such an identification. The same argument holds for the dorsal pocket in the ductus nasopharyngeus described by Rose (1893a) and Reese (1901). This last also appears to be posterior to tlie primitive nasal tissue of placodal origin, which makes the suggested homology most unlikely. Von Mihalkovies' (1898) suggestion that the medial duct of the glandula nasalis externa may represent the organ in question is not supported by embryological study, for, aside from its lack of relation to the early anlage, the duct and, in fact, nasal glands in general are formed quite late in the development of the nose while Jacobson's organ is typically present in very young stages. Also the duct arises from the wall of the vestibulum, not the cavum nasi proprium as would be expected if it represented Jacobson's organ. Finallj', the posteromedial pocket which Shiino (1914) de- scribes must be considered. As mentioned in describing Voeltz- kow's (1899) Crocodylus embryos, this pit does not seem to be connected with the structure here considered to be Jacobson's organ. It does not occur in the Alligator material studied. While PARSONS : NASAL EMKRYOLOGY 221 it is possible that this indentation may i-epresent a part of that organ, it seems unlikely on the basis of Voeltzko-w's fifrnres; if it should, Crocodylus embryos i-etain tlie finlage of Jjicobson's organ longer than AUiyaior, but still lose it completely early in their development. Nasal Glands Geyieral. All terrestrial vertebrates possess various glandular outgrowths of the nasal epithelium. Fish typically, if not uni- versally, lack such glands (Broman, 1939, believes that lungfish may possess an embryonic rudiment of a nasal gland). Those tetrapods which are secondarily adapted for an aquatic, and especialh^ for a marine, life tend to lose the glands, as is seen in many neotenous urodeles, marine turtles, some sea snakes, and whales. Birds, however, do not follow this pattern; in that class, typically, the nasal glands are most developed in marine forms, and salt water can be shown to stimulate the development of the major gland (Schildmacher, 1932; Marples, 1933; and Technau, 1936). A recent note by Schmidt-Nielsen et al. (1957) states that in Phalacrocora.r this gland actually functions to excrete excess salt. The glands are very variable, even within orders and smaller taxonomic units, and since they do not relate closely to the problem of Jacobson's organ, no detailed treatment of them will be given here. However, in view of some theories as to the homology of their ducts with Jacobson's organ, in certain forms (von Mihalkovics, 1898), some consideration of their nature and development is necessary. Glandida Namlis Externa. The glandula nasalis externa (= glandula nasalis lateralis, g. n. dor.salis, g. n. superior, or sometimes simply glandula nasalis) is, in all four Recent orders of reptiles, the largest of the nasal glands, and generally the first to appear in the ontogeny of the animal. It is here termed the external nasal gland as it lies outside of the carti- laginous nasal capsule, although generally within the dermal skull roof, in most forms; it may be either dorsal or lateral (or even, rarelj% posterior) to the nasal cavity. In all the forms which were studied for the present investiga- tion, this gland arises as a solid rodlike outgrowth of the pos- 222 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY terior or dorsolateral wall of the vestibiilum. Although it is the first of the nasal glands to develop, there is never any indica- tion of the anlage nntil well after the fusion of the nasal processes. The exact stage at which it first appears has already been noted in the sections on the nasal embryology of the various groups. The rodlike anlage of the gland increases in length, growing posteriori}- and, in most cases, somewhat dorsally. Its further development is primarily a matter of repeated branching near its posterior end, thus forming a mass of tubules which make up the body of the gland. These findings are in agreement with the observations of Born (1883) and Rose (1893b) who studied the development of the glandula nasalis externa in Natrix and Grocodyhis respectively; there appear to be no other embryological studies on this gland in reptiles. Histologically, tlie earlier stages in the development of the gland resemble those of the vestibulum. In older embryos, the duct is ]-ather variable, but generally possesses rather low columnar epithelium; although most commonly simple, it may he low stratified. The distal tubules have simple cuboidal epi- thelium Avitli very large circular nuclei. In very late embryos, the epithelium of the secretory tubules becomes high columnar, and the nuclei tend to be basally located. There is little lumen in either the gland or its duct in most of the embryos, and never in the material studied was the duct open throughout its length. Presumably the lumen is formed and the gland becomes active only at the time of hatching. In most adult turtles, the glandula nasalis externa forms a single middorsal plate lying between the nasal capsule and the skull roof; in sea turtles, both cheloniids and Dcrmochelys, it is completely absent (Nick, 1012, and Fuchs, 1915). Judging by the few forms which have been studied, this gland is larger in ter- restrial forms, such as Testudo, than in aquatic genera, such as Emys. The gland possesses, on either side, a single duct which enters the dorsolateral Avail of the vestibulum well posteriorly, hence in exactly the same ])osition as in the embi-yonic material which has been studied. Studies of the adult structure of the gland include those of Seydel (1896; Tcshido and Emys), Nick (1912; ChcJydra), and van der Merwe (1940; Pelomedusa). Iloff^mann (1S79-1890) also mentious this gland in most of the PARSONS: NASAL EMBRYOLOGY 223 forms Avhich he studied, but liis findin<>s with respect to nasal jrlands do not always agree with those of more recent Avorkers. The only description of the external nasal gland of ISplienoclon is that by lloppe (1934). lie states that the duet runs postero- dorsally from the posterolateral surface of the vestibulum, very close to the boundary of that area with the cavum nasi projirium. Posteriori}', its few branches form a fairly compact mass em- bedded in the connective tissue lateral to the concha anterior. Pratt (1948) mentions only that it is nuich less highly developed in Sphcjiodon than in most lizards. In snakes, the prominent glandula nasalis externa is an oval mass lying lateral to the nasal cavity, between the nasal capsule and the skull roof. It is almost universally present in this sub- order; of the over fifty genera which have been examined, only one is said to lack the gland completely. This is Pelajuis (Kath- ariner, 1900), a hydrophid, but its al)sence is not typical of the sea snakes since Kathariner (1900) and Smith and Bellairs (1947) have found it in three other genera of that family. Papers considering the structure of this gland in snakes include almost all of those cited in the discussion of the nasal anatomy of snakes and that of Smith and Bellairs (1947). Other works mention it, but do not add much information. There is considerable variation in the size of the external nasal gland, but the variation does not show any obvious relationship to either habits or phylogenetic groups. When the gland is large, it typically has a lobe lying Avithin the caAaty of the conchal cartilage ; this lobe is lacking in forms Avith smaller glands. The most detailed description of the liistology of this gland in snakes is that by Ileese (1925). Although he considers it to be acinar, other Avorkers, such as Baumeister (1908), call it tubular. This difference is almost certainl}^ one of terminology only ; Leydig (1873) describes the gland as consisting of tubides AA'hose ends are slightly sAvollen. In most snakes, a single duet runs anteriorly from the body of the gland to enter the posterolateral AA'all of the short vestibulum. exactly as in Spheuodon. IIoAvever, Baumeistei- (1908) states that in Rhinophis the duct enters the anteromedial glandular area Avliich he terms the median nasal gland, and Smith and Bellairs (1947) mention, Avithout listing genera, that there may be more than one duct from each external nasal gland. 224 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Although the glandula nasalis externa of croeodiliaiis was described in 1848 by Stannius, it has been little studied, and two authors (Hoffmann, 1879-1890, and Dieulafe, 1904-1905) have even denied its existence. The three papers which do treat it are those of Rose (1893b), Reese (1925), and Bertau (1935). As in other forms, the gland is an oval body lying dorsolateral to the nasal cavity between the nasal capsule and the skull roof. There may be either one or two ducts on each side which open into the vestibulum close to its junction with the cavum nasi proprium. One duct is generally posterolateral, and the other, Avhich is apparently present in most cases, enters the posteromedial vestibular wall. The latter duct, to reach the medial side of the nasal cavity, passes dorsal to the anterior end of the cavum. The histology of the gland is much like that found in other reptiles (Reese, 1925). An apparently similar glandula nasalis externa is found in most other tetrapods. It is present in all three orders of Recent amphibians, with the body of the gland forming a mass of tissue within the fenestra narina and with one or more ducts entering the lateral wall of the vestibular region (Matthes, 1934). The most detailed description of this gland in any amphibian is that given by Schucli (1934) who studied Triturus. In the birds, this gland is best developed in marine forms, as was noted above. It most often lies superficially in the orbital region (see Marples, 1933, and Technau, 1936, for detailed descriptions). There are two ducts on each side, one entering the lateral and the other the medial wall of the vestibulum, except in the Galliformes which possess onh' the medial duct. Although this pattern of two ducts is somewhat similar to that found in crocodilians, the two groups have probably developed the medial one independenth- ; in birds the medial duct passes ventral to the cavum nasi pro- prium. The homology of the glandula nasalis externa in amphib- ians, reptiles, and birds is suggested not only by its position, but also by its innervation. ({au])p (1888) found that in all cases this is by the nervus externus narium of the ophthalmic ramus of the trigeminal nerve. In mammals, Steno's gland is the major lateral nasal gland, and has generally been considered the homolog of the glandula nasalis externa as first postulated by Jacobson (1813). Peter PARSONS : NASAL EMBRYOLOGY 225 (1901) disagreed because lie believed that Steno's gland de- veloped from the caviini nasi i)ropriiiin rather than from the vestibulum. Broman (1921) and Kangro (1928), ho-\vevei', state tliat the gland does arise from the vestibulum. Thus, although Steno's gland typically lies Avithin the cartilaginous capsule, the homology as suggested by Jacobson seems probable. Glandula Nasalis Medialis. A distinct median nasal gland is found in most turtles and Sphenodoyi; it is absent in all members of the S(iuamata and Crocodilia as far as is known. In the embryos of Emys and Cliryseiinjs which were used in the present study, the glandula nasalis medialis arises at ap- proximately the same time as does the glandula nasalis externa or slightly later. Its anlage is first a solid rodlike process of cells projecting posteriorly or posteroventrally from the medial wall of the nasal cavity. In all the embryos it appears to arise from the vestibulum, but it is very close to or at the boundary of that region with the cavum nasi proprium. The pattern of further development and histology of this gland is identical to that already described for the glandula nasalis externa. Studies of the adult anatomy of the median gland include those of Seydel (1896), Nick (1912), and van der Merwe (1940). It is absent in sea turtles (Nick, 1912, and Fuchs, 1915), but in the other forms which have been studied it forms a mass of tubules lying between the medial wall of the regio intermedialis and the nasal septum. In Testudo, where it is very highly developed, it may even reach ventral to the nasal ca])sule (Seydel, 1896) ; however, the gland is normally entii-ely contained within the capsule. Van der Merwe (1940) reports multiple ducts enter- ing the nasal cavity in Pelomedusa, but other authors have found only one. In the adult the duct is said to enter the extreme anterior end of the regio intermedialis of the cavum nasi pro- prium rather than the vestibulum. The glandula nasalis medialis of SpJienodon was first described by Iloppe (1934) and is also mentioned by Pratt (1948). In both Iloppe's material and the embryos used in the present study, the gland is an unbranched rod of cells running ventrally and slightly posterolaterally from the vestibulum. It joins the antero- ventral surface of that structure well medially. Iloppe sug- gested that this rod was the homolog of the chelonian glandula 226 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY iiasalis medialis, and the embryology would appear to support such a view. Gaupp (1888) hoinologized the glandula nasalis medialis of turtles with the median nasal gland of amphibians on the basis of their innervation by the nervus septi narium, a branch of the oph- thalmic trunk of the trigeminal nerve. This homology has been generally accepted. In amphibians this gland normally lies ^•entromedial to the nasal cavity within the nasal capsule, with one or more ducts entering the unterer Blindsack (see Wieder- sheim, 1879; Seydel, 1895; Schueh, 1931; and Matthes, 1934). Birds lack a median nasal gland, although, as has already been noted, the glandula nasalis externa possesses a duct entering the medial wall of the vestibulum. Mammals, too, have no gland which would appear homologous to the glandula nasalis medialis of lower tetrapods, but many forms, especially rodents, do possess medially located glands, often in association with Jacobson's organ (Broman, 1921). Bowman's Glands. Bowman's glands or glandulae olfactoriae are typically present in the sensory olfactory epithelium of all tetrapods except certain neotenous urodeles. The only reported exceptions to this rule occur in the hydrophid snakes in which Kathariner (1900) states that they are greatly reduced in num- ber in Pelaniis and totally absent in Hydrophis. In all the forms studied for the present investigation, Bow- man's glands appear only in very late stages in the development of the embryos. They are, in all cases, simple fiask-shaped glands. The body of the gland is composed of simple cuboidal epithelium, while the A'ery narrow duct is formed by squamous cells (see Figs. 77 and 82). Similar descriptions based on adult material have been given l)y Seydel (189(); Emys and Testudo), OsaAva (1898: Sphenodon), and Macallum (1884; Thamnophis), al- though the latter two authors also report finding a few branched Bowman's glands. Hoffmann (1879-1890) also discusses these glands, mainly in turtles, but his descriptions do not agree well with those of other workers. One very important point concerning the distribution of Bowman's glands was first made by Sej^del (1896) : they are common in the regio olfactoria of turtles but never found in the I'cgio intermedialis. This observation has been confirmed by all PARSONS : NASAL EMBRYOLOGY 227 later workers, and forms a major line of evidence for the homology of tlie ehelonian regio intermedialis with the Jacob- son's organ of other amniotes. Previously, Seydel (1895) had shown that the nnterer Blinclsaek of ami)hibians, which he also considers the homolog of Jacobson's organ, lacks these glands. Bowman's glands have never been reported from the Jacobson's organ of the Sqnamata or mammals. Bowman's glands similar to those of reptiles have been de- scribed in amphibians (Sehnch, 1934) and birds (Ishihara, 1932). In the last class they are usually simple, but may be branched in certain forms. However, in mammals they are typically rather long and In^anched tubules which are quite contorted (Dogiel, 1886, Paulsen, 1886, and many others). Oilier Nasal Glands. A variety of other small glands are associated with the nasal cavities in many forms. Such glands may occur in any part of the nasal cavity; for example, Ishihara (1932) reports goblet cells in the sensory epithelium of an owl, Athene, and Maeda (1954) describes numerous glands in the vestibulum of mammals. The great variety is best shown by Broman (1921) who studied mammals. Among the reptiles, glands other than those described above have been found in members of the orders Chelonia and Squamata. Seydel (1896) found large numbers of pyriform and tubular mucous glands {Schleimdriisen) in all non-sensory parts of the nasal cavities of Testudo and Emys. They may occur singly or in clusters. None of the embryos studied for the present work possessed these tubular glands Avhich presumably develop only when the animal is hatched. However, as was noted in describ- ing the nasal embryology, there are areas in which the epithelium does contain massed goblet cells. It is possible that these areas invaginate to foi-m the tubules reported by Seydel; it is also possible that they are merely another type of glandular develop- ment not previously reported in turtles. In snakes there is even greater development of glandular epithelium in the non-sensory portions of the nasal cavities. Late embryos of Thamnophis possess many folds or shalloAV crypts lined by goblet cells which have already been described. Macal- lum (1884), von Mihalkovics (1898),' and Kathariner fl900) report similar structures in A-arious colubrids and hydrophids. 228 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY At least certain adult snakes display slightly more complex glandular formations. In the Storeria which was used in the present study, many of the sections are unfortunately broken and difficult to interpret, but anteriorly there appear to be glandular tubules as well as crypts entering the cavum nasi proprium. Kathariner (1900) reports a similar condition in Natrix. Finally, Baumeister (1908) describes a saccular out- growth at the anteromedial corner of the nasal cavity of Rhi- )iophift. This he terms a median nasal gland or Internasaldriise, witli the latter term intended to differentiate it from the glandula nasalis medialis of turtles. In view of the variation in structure and the small sample which has been studied, no generalizations can be made concerning these glands, and further discussion is not profitable. Olfactory Nerves Early Development of the Olfactory Nerve. A large number of workers have studied the embryology of the olfactory nerves, but there is still much disagreement over many points. However, since none of the material used in the present investigation was specifically stained for neurological study, little can be added here. Thus, the following description is concerned primarily with those points which appear to be necessary for an under- standing of the morphological significance of the various parts of the nasal cavity. The embryology of the olfactory nerve can conveniently be divided into two main stages, which will be discussed separately. The first extends from the time of the initial appearance of the nerve until approximately tlie period at which the nasal processes fuse ; during this stage the various elements forming the nerve become differentiated, or at least recognizable, and the basic histological pattern of olfactory innervation is established. Sub- setpiently, during the remainder of the embryonic period, the nerve gradually assumes the gross anatomical pattern found in the adult. In all vertebrates there appear to be two distinct elements in the anlage of the olfactor}^ nerve ; both originate from the epi- thelium of the nasal placode and migrate inwards toward the PARSONS : NASAL EMBRYOLOGY 229 anterior end of the brain. First there are nerve fibers Avhich {i'row out from cells remaining within the einthelium, and sec- ondly there are cells which move medially along- the course of the developing nerve. These two parts may be called the fibrous and cellular anlagen of the nerve, respectively. Since the appear- ance of Disse's paper of 1897, the fibrous anlage has almost uni- versally been considered the source of the adult olfactory nerve. The fate of the cellular anlage is still disputed. Disse and many later workers (e.g. Pearson, 1941b) have believed that they become Schwann cells, but others, including Groth (1938), deny this. A further source of confusion has been the nervus terminalis. This nerve also develops from the epithelium of the olfactory pla- code by inward migration. However, it is a ganglionated nerve, and cells migrate inward and then form fibrous processes.^ Therefore, it is now generally assumed that any ganglion cells which are found to migrate inward from the nasal epithelium are to be considered part of the terminalis. As has been pointed out by Belogolowy (1909-1910) and Simonetta (1933), this means that birds possess at least a histologically recognizable remnant of the nervus terminalis even though it cannot be dis- tinguished grossly — both these authors as well as Disse (1897), Tello (1923), Locatelli (1927), and Ishihara (1932) have re- ported ganglion cells along the course of the olfactory nerve hi birds. In the three reptiles studied in the present investigation, the olfactory nerve appears either just before the nasal placode first becomes indented {Thamnophis and Alligator) , or immedi- ately thereafter (Chrysemys) . In all three forms there appears to be considerable variation in the timing of the development of this nerve relative to that of the nasal epithelium ; although this is probably caused in part by the non-specific methods of stain- ing, it is almost certainly partly natural. The cellular anlage of the nerve forms as a massive columnar ingrowth of cells from the nasal placode in Alligator. It extends to the brain, but is, in all the material studied, sharply separated from the brain by a distinct basement membrane (see Figs. 63B and 64B). Due 1 See Pearson (lt)4l!ij tor a di.sciission of the embryology of the nervus ter aiinalis. 230 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY to this darkly staining cellular mass, it is veiy difficult to see any nerve fibers in the earliest stages, although they are probably present as soon as the cellular anlage is formed. In Chrysemys, on the other hand, the fil)rous anlage is the first formed. At all times there are nuclei among the fibers; presumably at least some of these have migrated inward from the nasal epithelium, but there is never a distinct cellular anlage separated from the surrounding mesodermal tissue by a basement meml)rane. Tham- no'phis shows an intermediate condition in which the cellular anlage is well formed (but not as large as in Alligator'), and in which nerve fibers may be seen in most embryos. After the placode has formed a nasal pit, the nerve appears predominantly fibrous throughout its length in all three forms. However, there are always abundant nuclei among the fibers ; these are most abundant in AUigato)- and least in Chrysemys. The nerve arises from the medial surface of the nasal pit at or just anterior to its apex. It runs anterodorsally as well as medially to reach the wall of the telencephalon near the anterior end of that structure; the olfactory bulbs are not yet distinct. All the fibers of the nerve appear to enter the telencephalon at one point which is marked hy the absence of any nuclei. This restricted enucleate zone gives the first indication of the myelo- spongium, the outermost layer of the adult olfactory bulb. The subsequent changes in the structure of the nerve in the stages preceding the fusion of the nasal processes are concerned primarily with an increase in the area of the olfactory e]iithelium which gives rise to fibers. In Chrysemys the nerve arises from most of the median and dorsal walls of the ]nt, that is from the most conspicuously thickened areas of epithelium, and forms several small trunks which do not unite until they have run approximately half the distance to the brain. As soon as the anlage of Jacobson's organ is formed in TJiariinopliis, two por- tions of the olfactory nerve may be distinguished. The first arises from the dorsomedial wall of the nasal pit, especially anteriorly, and the second from the dorsal surface of Jacobson's organ. By the end of this pei'iod a few fibers may be seen eutei-- ing the first trunk from the dorsolateral surface of the nasal i)it. Finally, in Alligaiur the fibers arise from the dorsal half of the medial wall of the pit and, as in Chrysvyuys. may form several PARSONS : NASAL EMBRYOLOGY 231 small trunks. Of especial interest is the fact that the small rudimentary anlage of Jacobson's organ, in the one embryo in which it is clearly present, gives rise to many nerve fibers, although in most cases few if any fibers may be found this far vent rally. In none of the embryos studied could a nervus terminalis be recognized, nor is there any convincing evidence of specializa- tion of the nuclei of the cellular anlage of the olfactory nerve. In some Chryscmys embryos these nuclei appear oval rather than circular, but such is not always the case. As was noted l)y Beard (1889) who studied Natrix, the nuclei may, in Tham- nophis, be more abundant in that portion of the nerve which arises from Jacobson's organ; however this is not a constant feature, but shows considerable variation. Later Developnient of the Olfactory Nerve. The major points of interest in the later developmental stages of the olfactory nerve concern the pattern of branching and the areas innervated ; these must be described separately for the various orders. However, a few general statements may be made first. With the growth of the snout region, the olfactory nerve, which previously i-an anteromedially, l)ecomes medially and finally posteromedially directed in all forms. The nuclei found associated with the fibers of the nerve gradually change from a circular to a long oval or even fusiform shape, with their long axes paralleling the course of the nerve fil)ers. Their ultimate fate could not be determined from the available material. Finally, in the absence of specially prepared material, ganglion cells could not be recognized. On the basis of the studies on birds mentioned above, their presence may be suspected, but none could be demonstrated in the present investigation. In Chrysemys embryos the pattern of distribution of the olfactory nerve remains very simple until the stage Avhen the sensory epithelium of the regio intermedialis spreads onto the lateral wall. All the thickened portions of the epithelium, that is. the dorsal half of the lateral and entire medial wall of the nasal cavity, give rise to olfactory nerve fibers, while the thinner- walled ventrolatei-al portion does not. The fibers tend to be collected into two main divisions, the so-called lateral and medial trunks. Each of these "trunks" is generally made up of several 232 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY small bundles, and frequently the separation between the two is not clear. The branches of the lateral trunk innervate the dorso- lateral portion of the cavum nasi proprium, and those of the medial innervate the ventromedial portion. The dorsomedial area appears, in most cases, to be innervated by the larger lateral trunk. However, the lack of a distinct separation between the two trunks renders a definite generalization impossible ; in many cases the medial trunk also appears to receive fibers from the dorsomedial area. As the nerve approaches the olfactory bulb, or in earlier stages the surface of the telencephalon, the bundles become more closely associated, and may even form one massive trunk. In general, the fibers of the medial trunk first lie medial to those of the lateral, and then gradually assume a dorsomedial and finally a dorsal position as the bulb is reached. Thus the lateral trunk tends to enter the main olfactory bulb, while the medial leads to the accessory bullx Despite references in the literature (e.g. McCotter, 1917) to the lack of any fiber interchange between the two trunks, the rather netlike appearance of the nerve as it nears the bulb makes such a definite statement impossible, at least in the material available for the present study. A further problem with the nerves of this area concerns the nervus terminalis. The literature on this nerve is considered subsequently, but since the terminalis could not be definitely identified in the available materal, it is mentioned here. In most of the series there is a small band of fibers along the medial edge of the olfactory nerve which contains many nuclei. This band appears to run to the posteromedial end of the accessory ol- factory bulb. Not until the latest embryonic stages were any fibers seen running posteriorly along the medial surface of the bulbs to the cerebral hemispheres. It appears probable that these fibers form the nervus terminalis, but I cannot be certain. In more advanced embryos in which the lateral wall of the regio intermedialis bears thickened and presumably sensory epithelium, there is only one major change in the pattern of innervation. The lateral trunk of the olfactory nerve appears to be exactly the same as before; however, the medial now extends ventral to the nasal cavity, and then turns dorsally along its lateral margin, thus innervating the newly thickened ventrolateral areas. In onlv one case was there any indication PARSONS: NASAL EMBRYOLOGY 233 that any branches of the lateral trunk might arise from the regio intermedialis ; even in that embryo the fibers could not be traced with certainty, so that, in view of the situation in other embryos, such innervation seems very unlikely. This pattern was observed in both Emys and Chelydra as well as in Chrysemys. There is some disagreement in the literature over the degree of distinctness between the lateral and medial trunks of the olfactory nerve. Seydel (1896) and Zuckerkandl (1910a) do not appear to consider the distinctness as marked, while Ogushi (1913) denies its existence. However, Johnston (1913), McCotter (1917), and Hanson (1919) describe them as completely separate at all levels. All these works are based on late embryos or adults of testudinid turtles except for those of Ogushi and Hanson which treat Trionyx and Chelydra respectively. The most important differences in the previous statements on this nerve concern the course of the fibers arising from the lateral wall of the regio intermedialis. The pattern described above, in which the medial trunk receives branches which pass ventral to the nasal cavity from this area, was first reported by Seydel (1896). His findings were confirmed by Zuckerkandl (1910a). Johnston (1913), Ogushi (1913), and McCotter (1917). Such a pattern is denied by Hanson (1919) and Loew (1956). The former states, without further amplification, that the entire medial trunk arises from the medial wall of the nose. Loew studied Emys, and reports that the lateral portion of the regio intermedialis is innervated not bj' the medial trunk, but by other parts of the olfactory nerve, hence presumably the lateral trunk. The descriptions given by Loew are thus in marked dis- agreement with the observations made in the present study. Finally, the olfactory nerve in a sea turtle has been described but once. Fuclis (1915) found that in Erctmochelys there are lateral and medial trunks comparable to those in other forms. The former innervates only the regio olfactoria (recessus su- perior posterior of Fuchs), while all the sensory epithelium of the recesses of the regio intermedialis sends olfactory fibers to the bulb through the medial trunk. The latter is composed of two main branches, one of which courses posterodorsally along the septum from the recessus ventralis; the othei- runs posteriorly from the recessus dorsalis. 234 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY The few available specimens of Sphenodon do not permit any detailed consideration of that form. In late embryos the olfac- tory nerve appears to arise by many bundles from the dorsal half to two-thirds of the cavum nasi proprium. These findings agree closely with those of Osawa (1898) and Hoppe (1934). Olfactory fiV)ers also arise from the dorsal surface of Jacobson's organ ; these fibers join bundles from the medial wall of the nasal cavity, so that a distinct vomeronasal nerve is not present. Wyeth (1924) was similarly unable to recognize a vomeronasal nerve in a late Sphenodon embryo (stage Q of the Dendy collec- tion). In Thannnophis, at the time of the development of Jacobson's organ, the olfactory nerve becomes divided into two trunks, or more exactly groups of bundles, of approximately equal size. One division of the nerve innervates Jacobson's organ, and is generally termed the vomeronasal nerve. However, the second part, which innervates the cavum nasi proprium, is referred to as the olfactory nerve. Since the vomeronasal nerve is now almost universally considered to be an integral part of the olfactory nerve, the latter term is used in two quite different senses. In order to avoid ambiguity, the nerve trunk innervating the cavum nasi proprium is here termed ''the olfactory nerve proper." In the younger embryos, the division into trunks is not very clear cut. The olfactory nerve enters the anteroventral surface of the brain as a rather irregular collection of bundles. The more lateral of these run dorsally from the anterodorsal portion of the cavum nasi proprium, while the medial bundles run somewhat anterodorsally to reach the olfactory bulb. Most of the medial division originates from Jacobson's organ, but many small l)ranches from the postei-omedial wall of the cavum nasi proprium run to, and enter the lateral side of the vomeronasal nerve. Since the bundles are, in the jjroximal third of their extent, frequently contiguous, it is im])Ossible to determine Avhether there is any interchange of fibers between them. In later stages, the two trunks become more distinct, with the olfactory nerve proper entering the anteroventral end of the olfactory bulb as a circular ring of from five to ten bundles. The vomeronasal nerve joins the accessory bulb posterodorsal to the main bulb; this iiei-ve typically consists of only one to three PARSONS : NASAL EMBRYOLOGY 235 bundles proximally. There is, during development, a gradual decrease in the number of small branches of the vomeronasal nerve which arise from the cavum, and they disappear almost com- pletely with the formation of the capsule of Jacobson's organ. Distall}', fibers of the olfactory nerve proper arise from the entire sensory area of the cavum nasi proprium — that is, all of the regions previously described as having thickened epithelium. In early stages, the majority of the fibers are found antero- dorsally, but they are soon present over the entire dorsal half of the cavum. The concha comes to be especially well innervated. These fibers are irregularly collected into larger and larger groups until the bundles are formed. The entire posterodorsal surface of Jacobson's organ gives rise to many nerve fibers, all of which run posterodorsally and pass out of the capsule of that organ at the posterior end of the septomaxilla. One very large and rather loose bundle is formed, which then runs dorsally and only slightly posteriorly. It therefore crosses the medial surface of the olfactory nerve proper. At the level of the anterior end of the olfactory bulb, the vomeronasal nerve turns posteriorly, and, as two or three bundles, runs along the medifil wall of the olfactory bulb to reach the accessory bulb. Despite a careful search for it, no trace of a nervus terminalis could be found. While it is possible that some of the fibers and nuclei within the olfactory nerve actually represent the termin- alis, none of the fibers run posteriorly along the medial wall of the telencephalon posterior to the accessory olfactory bulbs. Therefore, although there could be cellular rudiments of it, there is no morphologically distinct nervus terminalis. This is in agree- ment with the findings of Bellairs (1942 and 1949), who reports the absence of that nerve in Vipera and Varanus. The litera- ture appears to contain no other references to the terminalis in any member of the Squamata. The literature on the olfactory nerve of adult snakes in general agrees well with the observations made on late embryonic ma- terial in the present study. Important papers include those of Leydig (1872b, Natrix; and 1897, Natrix, Coronella, and Vipera), Lee (1893, Typhlops), Macallum (1884, Thanmophis) , Zucker- kandl (1910a, TiipMops), McCotter (1912, "snake"), and Bell- airs (1942, Vipera). Most of llic disagreement which does exist 286 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY concerns the degree of separation between the two trunks, and whether or not the vomeronasal nerve supplies all the fibers innervating Jacobson 's organ and no others. Since the proximity of the two trunks makes fiber interchange a possibility, this does not appear to be a very important point. The available embryos of Alligator do not permit any detailed study of the later developmental stages of the olfactory nerve. In the series studied, nerve fibers appear to arise from the greater part of the surface of the cavum nasi proprium, even well ventrally and anteriorly. However, Bertau (1935) found sensory epithelium only in the posterodorsal third of the cavum in the oldest of his embryos of Alligator, Melanosuchus, and Crocodylus. Shiino (1914) and Nemours (1930) describe intermediate con- ditions. Thus it appears probable that the olfactory nerve fibers are originally formed by a large part of the caval epithelium, l)ut during the course of later development become much re- stricted in their areas of origin. Further study is necessary before any detailed accounts of olfactory nerve distribution in crocodilians would be profitable. Olfactory Bulbs. Although any detailed consideration of the olfactory bulbs is outside the province of this study, their struc- ture and development are of importance with respect to ques- tions on the homology of Jacobson 's organ. Therefore, a brief treatment is included here. Special emphasis is placed on the occurrence and nature of the accessory olfactory hulh {^area vomeronasalis or Nehenbulbus) which typically receives the nerve fibers from Jacobson 's organ. As in the other parts of this paper, the snakes are taken as examples of the Squamata, and the lizards are not discussed. However, the extreme variation found in the degree of develop- ment of the accessory bulb in the latter group, and its correlation with the size of Jacobson 's organ are of great interest; the lizards present convincing evidence for the relationship of Jacob- son 's organ and the accessory bulbs. Major papers which con- sider the lizards include those of Zuckerkandl (1910a), Krabbe (1939), and Crosby and Humphrey (1939). Terminology, as usual, poses certain problems. The term olfactory ])ulb is frequently used in a topographic sense as a synonym of rhiuencephalon ; snch is done in the present paper. PARSONS : NASAL EMBRYOLOGY 237 When it is necessary to distinguish those areas which are histo- logically the olfactory bulb — that is, the areas possessing olfac- tory glomeruli, the term bulbar formation is used. The region receiving fibers from the vomeronasal nerve or its homolog, when such is a separate zone, is the accessory olfactory bulb ; the region receiving fibers from the olfactory nerve proper is here termed the main bulb. AOB Figure 73. Diagram of tho olfat-tory and accessory olfactory luillis uf a reptile in frontal section to show the various cell layers mentioned in tlie text. Detailed structure is shown on the right side only. Typically the olfactory bulb of the adult displays a rather definite layered structure ; Figure 73 shows, in very diagrammatic fashion, those zones mentioned in the text. In no reptile are these zones as distinct as they are shown in the figure ; in most cases the outer granular layer, especially, is not separate, but represented 238 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY by periglomerular cells within the glomerular layer. The termin- ology used is based on the papers of Krabbe (1939), Crosby and Humphrey (1939), and Allison (1953). The last of these presents the most recent review of the nature and functions of the various layers, and of the structure of the central portions of the olfactory system in general ; the reader is referred to that work for further discussions and an extensive bibliography. There have been few studies on the embryology of the olfactory bulbs in reptiles. Krabbe (1939) considers the development of their gross form, but does not discuss the various cellular layers in any detail. His descriptions are similar to the observations which were made on Chrysemys, Thamnophis, and Alligator in the present study. In all cases differentiation of the bulbs starts before they are morphologically distinct from the telencephalic hemispheres. The region to become the olfactoiy bulb is first on the ventrolateral surface of the telencephalon, near but not at its anterior end. With the growth of the brain it gradually as- sumes its characteristic terminal position. Soon after the begin- ning of liistological differentiation, the presumptive bulbar area forms a small protrusion of the wall of the telencephalon; it contains a narrow, digitiform extension of the lateral ventricle. Further development is essentially a matter of the elongation of the olfactory bulb with the formation of a thinner olfactory tract between it and the hemisphere. The extension of the lateral ventricle also elongates, and extends to the center of the bulb at all stages. In all three forms studied, the development of the bulbar lamination is very similar. At the .start of its differentiation there are two layers i-ecogni/ahle : an outer layer comjiosed almost entirely of nerve fibers with almost no nuclei, and an inner zone with many small circular nuclei. The latter zone soon divides in two with the innermost half remaining the same and the outer lialf now containing fewer nuclei. At a later stage, the outermost enucleate /one can be seen to form an outer, still enucleate, zone the nijidosponginm of the adult, and an inner zone with scattered nuclei. Simultaneously, the central layer with few nuclei becomes differentiated into a thin outer zone with many nuclei and a thicker inner one which is almost entirely fibrous; these are the mitral rrlJ and iniifr plexiform layers respectively. Finally two PARSONS: NASAL EMBRYOLOGY 239 more divisions of zones occur. The myelospongium is unchanged, but inside that the zone with scattered nuclei forms a glomerular layer containing the olfactory glomeruli and scattered periglo- merular cells, and a fibrous outer plexiform layer. Continuing inwards, the mitral cell and inner plexiform layers remain as before, and the innermost, densely nucleate zone is divided into a thick inner granular layer and a thin layer of ependymal epi- thelium actually lining the ventricular cavity. In adult turtles, it lias long been recognized that the olfactory bulb may be divided into two major areas, a larger anteroventral one which includes the tip of the bulb and a smaller postero- dorsal one. This division is described by Ilaller (1900), Zucker- kandl (1910a), Johnston (1915), McCotter (1917), and Crosby and Humphrey (1939). The last two of these papers consider the smaller posterodorsal area to be an accessory olfactory bulb receiving the nerve fibers fi'om Jacobson's organ or its homolog. The older Chrysemys embryos studied in the present investiga- tion correspond exactly with these earlier descriptions, with the absence of the outer la.yers of the bulbar formation, especially of the glomerular zone, marking the boundary between the main and accessory bulbs. As far as I am aware, onlj^ one author has denied this pattern. Schepers (1948b), describing Testudo geometrica (= Psammo- l)ates oculifer of Loveridge and Williams, 1957), states that there is no definite sepai-ation into dorsal and ventral parts in either the olfactory nerves or bulbs. In an earlier paper Schepers (19-18a) describes a deep invagination of the posterolateral wall of the olfactory l)ulbs in this form which greatly increases the circumferential area of the bulb, and results in the virtual obliteration of the liulbar part of the lateral ventricle. No other worker has reported such an invagination. In his second paper, Schepers suggests the possibility that this invagination may be somehow related to the vomeronasal nerve, but at the same time states that he was unable to identify any such nerve. Although the olfactory bulbs of Sphenodon have been described bj' several workers, including Osawa (1898), Unger (1914), Cairney (1926), and Christensen (1927), only one previous paper, that of Wyeth (1924), mentions the accessory bulb. Unger may also have recognized it. but his description does not make 240 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY this clear. In the oldest embryo used in the present study, as in the specimens described by Wyeth, the accessory bulb is a small circular or oval area on the dorsomedial wall of the posterior part of the olfactory bulb. As in the case of turtles, it is separated from the main bulb by a narrow region lacking the outer layers of the bulbar formation. In the snaites, an accessory olfactory bulb has long been known and is universally recognized. Previous descriptions include those by Herrick (1893), Rabl-Riickhard (1894), Leydig (1897), Zuckerkandl (1910a), and Crosby and Humphrey (1939). The main bulb forms, in the adults studied by these authors, a small cap on the olfactorj^ bulb. Posterior to it, the medial wall is enormously thickened and forms a very large accessory bulb. Due to its size, the ventricle of the bulb is displaced laterally, and often the surface of the accessory bulb contains an indentation, presumably to increase the surface area for entering nerve fibers. One exception has been described : in Typhlops the accessory bulb is posterodorsal rather than posteromedial to the main bulb (Zuckerkandl, 1910a). Embryos of Thamnophis present an essentially similar picture. There is no indentation in the wall of the accessory bulb in any of the series studied, but this could easily form after the animal is hatched. One interesting point is that when the accessory bulb first appears in the embryo, it is dorsal, as in Typhlops, rather than medial. In later embryos it gradually becomes dorsomedial in position. Presumably it does not become truly medial until the time of hatching ; it is still somewhat dorsal in the most mature embryos used in the present study. There are no reports of an accessory bulb in any crocodilian. None was present in any of the Alligator embryos studied in the jn-esent investigation, thus confirming the studies of Herrick (1890), Zuckerkandl (1910a), Unger (1911), and Crosby (1917), and the statements in more general works by Kappers (1934), Kappers, Huber, and Crosby (1936), Crosby and Humphrey (1939), and Allison (1953). Herrick, and later Crosby, describe an "olfactory fossa" in which a thickening of the glomerular layer results in an unfolding of the inner zones of the bulbar formation. According to Crosby (1917), this lies in the ventro- medial wall of the bulb near its anterior end. However, it does PARSONS: NASAL EMBRYOLOGY 241 not appear to bear any relation to an accessory bulb, and could not be found by any of the other workers who studied croco- dilians. Nervus Terminalis. Although, as was previously noted, a nervus terminalis could not be definitely identified in any of the material used in the present study, it should be mentioned here since it has often been confused with the vomeronasal nerve (e.g. Kappers, 1934, p. 457). As far as is known, the terminalis is completely lacking in crocodilians (Haller, 1934; and Kappers, Huber, and Crosby, 1936) and in squamates (Bellairs, 1942 and 1949). The only report of its occurrence in Sphenodon is that by Wyeth (1924) ; he admits that he is unable to be certain about the identification, and describes only the proximal end of the structure in question. However, in turtles, three workers have been able to locate the terminalis: Johnston (1913, Emys) ; Hanson (1919, Chely- (Ira) ; and Larsell (1919, Chrysemys) . According to Johnston and Larsell, the nerve is associated with the medial trunk of the olfactory nerve and forms a plexus in the median wall of the nasal cavities. Larsell believes that this nerve is probably pri- marily autonomic, but its function is not understood at all. DISCUSSION Major Subdivisions of the Nasal Cavity. Throughout this work the nasal cavity has been considered to consist of three main parts — an anterior vestibulum, a posterior ductus naso- pharyngeus, and, between them, a cavum nasi proprium. The vestibulum has, unfortunately, l)een defined in three dif- ferent ways : on the basis of embryological development, of his- tological structure, and of gross anatomy. Theoretically, the first of these would be preferable in a consideration of the homologies of the nasal area. According to the embryological definition, the vestibulum is that part of the nasal cavity whose walls develop from essentially unmodified ectoderm rather than from the nasal ])lacode, and are, at some stage in ontogeny, in- vaginated to form a tubular connection l)etween the naris ex- ternus and the tissue of ])lacodal origin. This definition, while fre(|uently cited, is but rarely actually followed ; it is rendered 242 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY vague and impractical by the lack of any distinct boundary be- tween the placode and the surrounding epidermal epithelium, at least in reptilian embryos. In mammalian anatomy, the vestibulum is most commonly de- fined by its histology, as that portion of the nasal cavity whose walls bear stratified squamous epithelium comparable to the skin, rather than columnar epithelium as in the remainder of the nasal cavity. However, in reptiles, the transition between these epi- thelial types can occur at very different parts of the nasal cavities, even in members of the same family (see Eckart, 1922, on agamids), so that a histological definition also appears to be un- satisfactory. The third definition of the vestibulum, that l)ased on the gross form, is used in the present paper. In most reptiles, it forms a rather distinct anterior tubular region. In the orders Rhyncho- cephalia and S(iuamata it is typically separated from the cavum nasi proprium by a small ridge on the lateral wall of the nasal cavity. Turtles may, as in the Testudinidae, possess such a post- vestibular ridge, but in others, such as the Cheloniidae, no ridge is present. However, the great and abrupt increase in cross- sectional area at the posterior end of the vestibulum provides a sharp boundary in all turtles. Crocodilians do not display so obvious a division between the vestibulum and cavum ; the former region is the vei-tical segment of the nasal cavity, while the cavum is horizontal. Throughout the reptiles, the duct of the glandula nasalis externa enters tlie nasal cavity in the posterior part of the vestibulum (as here defined), thus providing evidence that the postvestibular boundary is closely comparable in all groups. Although such a morphological definition does not appear to be satisfactory for all other classes of tetrapods, it is the most useful in any discussion of the reptilian nose. In almost all reptiles, the vestibulum forms only a small portion of the nasal cavity. However, in certain lizards, such as some iguanids and agamids, it may form a long and variously curved tube. Such elongation is most probably a specialization for desert life in many cases ; the longer and more complex structure would tend to prevent i)articles of sand from becoming lodged within the sensory areas (Stebbins. 1948). Modifications enabling the animals to close the nares externi are found in certain aquatic PARSONS : NASAL EMBRYOLOGY 243 groups, both among the snakes (Kathariner, 1900) and croeo- dilians (Bellairs and Shute, 1953). Otherwise, the vestil)nlum is a simple, essentially tubnlar structure, without conspicuous variation among the reptilian orders, although its length and diameter are markedly ditt'erent in different forms. As stated in the section on turtles, the term ductus nasopharyn- geus is used in the present paper for any tubular structure lead- ing from the nasal cavity to the choana, although Fuchs (1908) proposes the name ductus choanalis for those cases in which no secondary palate is formed. The distinction made by Fuchs — that a ductus choanalis is a part of the nasal cavity, while a ductus nasopharyngeus is a part of the buccal cavity which is separated from the remainder of that cavity bj^ a secondary palate — is useful in considering the embryology, but appears to be rather arbitrary as far as the adult anatomy is concerned. A practical problem is that Fuchs' definitions depend on the recog- nition of an exact boundary of the nasal tissues ; such an exact boundarj^ is not generally recognizable. For a further discussion of this problem, the reader is referred to the works of Fuchs, especially his paper of 1908. Among the recent reptiles, there is considerable variation in the length of the ductus Ufisopharyugeus. In Splifnodou, the cavum nasi proprium opens directly into the mouth cavity, so that there is no ductus. However, in the Squamata, tiie ductus is present, although generally short, especially in lizards. Turtles possess a well-developed ductus, which Fuchs considers to be a ductus choanalis similar to that of snakes, although it is typically somewhat longer in the Chelouia. The crocodilians have a highly developed secondary palate and an exceedingly long ductus naso- pharyngeus. In all cases, the ductus is a very simple tube ; many turtles, such as Testudo, Emys, and CJinjscmys, possess a small dorisolateral diverticulum near the anterior end of the ductus, the recessus ducti nasopharyngei. but otherwise there are no gross structural modifications reported iu reptiles. The third major section of the nasal cavity is termed the cavum nasi proprium. It can be delimited only as the region lying be- tween the vestibulum and the ductus nasopharyngeus, and thus its definition varies with those of the other ]iortions of the nasal cavity. In most cases, the cavum is a large cavity, bearing sensory 244 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY olfacton' epithelium on a part of its walls; such sensory epi- thelium is not found in the vestibulum or ductus. Since the form of the cavum varies greatly among the reptilian orders, no generalizations can be made concerning its form, other than that there are usually one or more projectioiLs of its lateral wall ex- tending into the cavity. Many names have been applied to the different parts of the cavum ; all of these are defined in terms of the lateral projections, generally termed conchae, which must, therefore, be considered next. The term concha has been used in several different senses. Most commonly it is applied to any process of the lateral wall of the nasal cavity, or the extension of the nasal capsule into such a process. This general usage is followed in the present paper. However, at least two more restricted definitions have been pro- posed, or at least used. Gegenbaur (1873) and Solger (1876) apply the term concha only to simple lamellar processes ; all other projections of the wall are termed pseudoconchae. Hence a struc- ture such as the concha of most squamates, which is U-shaped in section and frequently contains a portion of the glandula nasalis externa within its center (the cavum concliale), is not considered to be a true concha by these workers. On the other hand, deBeer (1937) applies the term concha only to those projections which do contain a cavum conchale. The problems of conchal homologies among the various amniotes have received mucli attention in the literature, but there is, as yet, no general agreement. In the following discussion only the epithelial relationships are considered, and the nasal capsules are not treated. This limitation prevents a complete discussion of the problem, but several aspects can still be satisfactorily dealt with. Turtles do not possess a concha, although the convexity of the dorsolateral wall (the Muschelwulst) has sometimes been referred to by that term. The latter structure, however, is well developed only in the Testudininae (Testudo), and is small or absent in most other groups including the Emydinae. Since the testudinines are believed to have been derived from emydine ancestors (Loveridge and Williams, 1907), it seems probable that a prominent Muschel- wulst is a specialization within that subfamily. Gegenbaur (1873) and Fuclis (1915) consider the laterale Grenzfalte of sea turtles to be the homolog of the saurian concha. The Grenz- PARSONS: NASAL EMBRYOLOGY 245 falte is in approximately the position where a concha might be expected, but its form is not at all like that of the concha of any other group. It cannot, of course, be demonstrated whether the absence of a concha in turtles is jirimitive, or whether that structure has been secondarily lost within the order ; certainly there is no evi- dence which would tend to cast doubt on the former theory. None of the modern Amphibia possess any conchae. Although these forms are very specialized in many respects, their nasal regions are, according to Jarvik (1942), very similar to those of the crossopterygians. Thus, in the absence of paleontological evidence, there is no reason to assume that any amphibians or primitive reptiles possessed a concha, or that its absence in turtles is not primitive, although both are entirely possible. All other amniotes have at least one concha. Among the re- maining reptilian orders, the Rhjaichocephalia possess two, the Squaraata one, and the Crocodilia three. However, in the last case, the preconcha and concha develop as a single projection of the lateral wall Avhich subsequently divides into two parts, so that there are only two elements whose homologies must be considered. In the birds there are three conchae ; the most anterior of these, the covcha vestihulae (or "preconcha"), is on the wall of the vestibulum, and thus quite distinct from all other conchal formations. Mammals typically possess a maxillo- turbinal, a nasoturbinal, and a series of ethmoturbinals ; Peter (1901) also describes a small concha obtecta in the posterodorsal l)art of the nasal cavity. The most commonly accepted system of homologies of these various conchae is that proposed by Peter (1901). He considers the concha of squamates, concha media of birds, and maxillo- turbinal of mammals to be primary conchae laterales anteriores, while the postconcha or Riechhligel of birds, nasoturbinal of mammals, and possibly the postconcha of crocodilians are termed secondar}^ conchae laterales anteriores. All the other conchae are believed to be independently developed and not related one to another. More recent authors, such as Fuchs (1908) and Bertau (1935), have added the concha (= the concha posterior?) of Sphenodo)i and the preconcha plus concha of crocodilians to the list of primary conchae laterales anteriores. Matthes (1934) 246 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY and other workers have questioned the homology of the post- concha of crocodilians and birds with the mammalian nasotiir- binal, but have, for the most part, accepted Peter's ideas. However, at least one recent paper presents a quite different scheme. Hoppe (1934) recognized two conchae, anterior and pos- terior, in Sphenodon. The latter he considers to be the homolog of the concha of squamates ; the former he compares to what he terms the "Lippe am Choanengang" of lizards. This lip is a small ridge along the lateral side of the choana described by Beecker (1903), who attached no importance to it. Hoppe also compares Sphenodon to crocodilians and birds, equating the concha pos- terior of the first with the postconcha, and the concha anterior with the middle concha of crocodilians and birds. Thus, he believes the concha of squamates to be homologous to the post- concha of crocodilians. Hoppe 's proposals do not appear to be very satisfactory. The concha posterior of Sphenodon does resemble closely the concha of the Squamata, and their homology seems probable. However, the similarities between the concha anterior and the lacertilian Lippe am Choanengang are slight, and are found only in the adults. The rhj-nchocephalian concha anterior is first formed as a ridge which extends dorsoventrally along the lateral wall of the cavum nasi proprium ; during its embryonic development, it rotates to the more nearly horizontal position typical of the adult (see Fig. 33, and the figures in Hoppe, 1934). The Lippe of lizards, on the other hand, forms in place as a horizontal ridge along the Choanengang at the nasal end of the ductus naso- pharyngeus. With regard to crocodilians, Hoppe 's suggestions seem improbable. On the bases of both gross form and position within the cavum, the squamate concha appears to be far more readily comparable to the crocodilian middle concha than to the postconcha. Admittedly neither form nor position are infallible guides to homology; however, the position of the postconcha, posterior to the nasal end of the ductus nasopharyngeus, and thus in the region corresponding to the squamate Antorbitalraum, is radically different from the location of the concha of am- squa- mate. Therefore, it seems most probable that the concha posterior of Sphenodon, the concha of squamates, the preconcha plus concha PARSONS : NASAL EMBRYOLOGY 247 of croeodilians, the concha media of birds, and the maxillotnr- hinal of mammals are all to be considered as homologous. It may he noted that this is typically the first of the conchae to appear in ontogeny in all groups, although in some mammals (e.g. Lepus, Peter, 1901) the nasoturbinal may also arise at an e(iually early stage. The only other proposed conchal homology which must be considered is that of the postconchae of croeodilians and birds with the mammalian nasoturbinal. The latter structure is typi- cally dorsal or anterodorsal to the maxilloturbinal, while the postconcha is posterior or posterodorsal to the concha, so that their positions are not readily comparable. In form, the post- concha is merely a marked outpocketing of the lateral wall, while the nasoturbinal is most commonly a more elongate and fre- quently lamellar formation. Actually, there appears to be no convincing evidence for their homology. It therefore seems more probable that, in the course of their evolution, both archosaurs and synapsids independently developed a second lateral concha. Unfortunately, there appears to be little paleontological evidence on the origin of conchae. Watson (1913) reports the presence of ethmoturbinals in at least two genera of therapsids ; in 1951, he states that the nasoturbinal and maxilloturbinal were also present, but gives no details on their structure. Thus a single concha was probably developed at a very early stage in the evolution of reptiles, before the separation of the line leading to mammals. The Squamata retain this primitive condi- tion, but other groups have independently acquired other conchal structures. This theory is supported by the lack of evidence for further homologies ; the idea of the gradual acquisition of various projections of the nasal wall in different groups seems to be a more reasonable working hypothesis than the supposition that a rather complex pattern appeared suddenly in primitive reptiles and has persisted, although often much modified, to the present. Conclusive evidence can come only from paleontology, and will probably never be available. There is no need to make further comparisons of the nasal cavities of reptiles with those of other tetrapods, since all the problems relating to the cavum nasi proprium are basically concerned with the conchae, and the other regions are generally simple. Matthes (1934) gives excellent descriptions of the nasal 248 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY cavities of all groups, considering the embryology as well as the adult anatomy ; the reader is referred to his paper for further information and references on these topics. Jacohson\^ Organ. Any discussion of the evolution of Jacob- son's organ must necessarily start with the amphibians; un- fortunately, there is no general agreement on tlie intci'j)retatioii of the nasal cavities of that class, despite the existence of a very large body of literature on all aspects of the organ's form and development. In the following review only a few of the major papers are cited. Many more references can be found in the review by Matthes ( 1!)34). Special attention should also be called to Helling 's (1938) study of the Anura. Although there is considerable variation in nasal anatomy among the three living orders of amphibians, representatives of all three possess what may be called the unterer Blindsack, a conveniently vague term tirst employed by Born (1876). In most adult urodeles, ignoring the neotenous forms, this unterer Blind - sack forms a small groove along the lateral margin of the main nasal chamber with which it is connected throughout its length. The remaining two orders possess a more complex structure. In the Gymnophiona, the Blindsack is a tubular inpocketing of the nasal wall which extends anteriorly from the anterolateral mar- gin of the choana and lies ventral to the posterolateral portion of the main nasal cavity. The nasal anatomy is more variable among the Anura, but there is typically a series of quite distinct saccular extensions of the main nasal chamber. The unterer Blindsack lies ventral to the main chamber, and the two are joined by a narrow vertical slit which runs nearly the length of tlie nasal cavity. At the ventral end of this slit, the Blindsack extends laterally and, in its anterior half, also medially, so that in transverse section it resembles an inverted letter T. Seydel (1895) considered the untere Blindsacke of all three orders to be liomologous, but the radical differences in their forms and posi- tions caused many workers to doubt this. However, subsequent embryological studies, such as that of Hinsberg (1901) have tended to support Seydel's opinions, and the proposed homology is now commonly accepted (see Matthes, 1934). In all three orders, the wall of the unterer Blindsack bears an area of sensory epithelium: this area was termed the Jacobson's organ by many early workers, most notably Seydel (1895). PARSONS : NASAL EMBRYOLOGY 249 Other investigators (iiiestioned the liomology of the Blindsack with the Jaeobson 's organ of anniiotes, and tliere is still no general agreement on the subject. Matthes (1934) states that there is now little doubt that the structures are homologous, but Francis, in the same year, claims that "there is evidently a growing con- sensus of opinion . . . against Seydel's view" (p. 303). In the urodeles the sensory epithelium lines the posterior half of the Blindsack, while in anurans it is found in the anteromedial por- tion only. According to Wiedersheim (1879), the entire Blindsack of the Gymnophiona is lined by sensory epithelium, but Sarasin and Sarasin (1887-1890) report that onlj^ its ventral surface is sensory. Seydel's arguments in favor of the proposed homology include the innervation of the sensory areas of the Blindsack bj^ the ventral branch of the olfactory nerve, which corresponds to the medial trunk of that nerve in reptiles. He further cites the absence of Bowman's glands in this area, and states that, in all eases, both in amphibians and amniotes, Jaeobson 's organ is used to smell the contents of the mouth. A final major point is the presence of glandula nasalis medialis associated Avith the Blind- sack in all three groups of amphibians. The major objections raised against Seydel's identification of Jaeobson 's organ in amphibians are based on the adult topog- raphy of the nasal area. The lateral position of the Blindsack in Solamandra caused von Mihalkovics (1898) to consider it to be the homolog of the maxillary sinus rather than of Jaeobson 's organ, although he accepts Seydel's conclusions with respect to Rana in which the sensory portion is medially situated. That the sensory area of the urodele Blindsack has the same function as the Jaeobson 's organ of amniotes, is not denied by von Mihalkovics; however, as he points out, analogy is an unreliable guide to homology. Supporters of Seydel's theory argue that the embryological development of the Blindsack demonstrates its primitive medial position, even in the urodeles. In Triturus {= Triton of authors), the Blindsack is said first to be present as a small inpocketing in the ventral half of the medial w^all, and to assume its adult position through a rotation of the wall of the nasal cavity around its long axis. Tbe first major paper describing this mode of development is that of llinsberg (1901 ) ; more recently. 250 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Scliucli (1934) has confirmed his results. However, the early embryology of the amphibian nose is still not at all well under- stood (see the discussion in the subsequent section of this paper), and thus Hinsberg's findings are subject to some doubt. Another objection which is frequently raised is that the am- phibian Blindsack does not typicallj^ develop until after the formation of the choanae, although at a stage before the lumen of the nasal cavity is continuous with that of the mouth. In amniotes, on the other hand, Jacobson's organ generally appears at a much earlier stage. However, in Xenopus, Foske (1934) finds a small ventromedial inpocketing, the anlage of the Blind- sack, while the nose is still in the placodal stage. The significance of this case, which appears to be exceptional, is, of course, open to question. Seydel's evidence from the glands remains virtually unchal- lenged. Almost all workers seem to accept the homology of the giandula nasalis medialis throughout the Amphibia, which is one of Seydel's main arguments, and Matthes (1934) states that Bowman's glands are never found in the sensory epithelium of the unterer Blindsack. As in the case of the turtles, the neurological evidence is of prime importance ; however, interpretations vary. The first major paper on the olfactory bulbs of the amphibians is that of Zucker- kandl (1910a). lie recognized the main and accessory bulbs, but was generally unable to determine the relationship between the subdivisions of the bulb and specific areas of sensory nasal epi- thelium. In Rana, the ventral nerve trunk is thought to enter, in part, the accessory bulb. Zuckerkandl stresses that in all am- phibians the olfactory fibers enter the anterior and lateral sur- faces of the bulb; none are present medially. The accessory bulb lies at the posterolateral end of the main bulb, and may be either dorsolateral {Ilypogcophis) or ventrolateral (Bana). In amniotes, on the other hand, many olfactory fibers enter the medial surface of the bulb and the accessory bulb typically lies posteromedially. The difference is so great that Zuckerkandl believes that two somewhat parallel lines are present, but that there is no homolog^y between the accessory bulbs of amphibians and those of amniotes. Therefore, he does not accept Seydel's theory on the Jacobson's organ of the Amphibia. PARSONS: NASAL EMBRYOLOGY 251 Wliether the medial position of the accessory bulb is actually of basic importance appears to be debatable. In turtles and most niamnials it is more dorsal than medial, and in at least two maunnals, Blarina and Scalopns, Crosby and Humphrey (1939) report a dorsolateral position. Therefore, Zuckerkandl's argu- ments cannot be considered conclusive. McCotter (1917) studied Rana, and describes a vomeronasal nerve as arising from the medial portion of the unterer Blind- sack. This nerve runs posteriorly along the medial wall of the nasal cavity, and then turns laterally, passing ventral to the remainder of the olfactory nerve. Finally, it runs posteriorly along the lateral surface of the olfactory bulb to reach the accessory bulb. McCotter, like most later workers, reports some intermingling of fibers from the two parts of the olfactory nerve, but states that it is slight. He accepts Seydel's theories. The most important work on the urodeles is that of C. J. Herrick ; of his many papers which consider this problem, those of 1921, 1924, and 1948 will suffice for the present discussion. In both Amhystoma and Nccturus, he identifies an accessory olfactory bulb at the posterolateral end of the main bulb, but notes that the two are only indistinctly separated. The accessory bulb is said not only to receive the nerve fibers from the unterer Blindsack, but also many from the main nasal chamber. Herrick therefore concludes that a Jacobson's organ is becoming differ- entiated within the Urodela, and reaches its definitive condition in the Anura. Like Seydel, he appears to consider the modern Amphibia to be a phylogenetic series, an idea which invalidates his evolutionary conclusions. Thus the amphibians appear to possess certain areas of sensory epithelium within the untere Blindsacke which are, in many respects, comparable to the Jacobson's organ of amniotes. How- ever, there are also important differences in embryology, anatomy, and neurological connections. That the unterer Blindsack is, at least in part, homologous to Jacobson's organ seems probable, but the use of the latter term does not appear to be advisable. The Jacobson 's organs of the four Recent reptilian orders have been discussed in previous sections, and only a short summary is necessary here. In the turtles, the ventral or anteroventral half of the cavum nasi proprium (the regio intermedialis) contains areas of sensory epithelium which are thought, on the basis of 252 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY neurological and other evidence, to be homologous with Jacobson 's organ. However, as in the case of the amphibians, these sensory areas are quite different from the definitive Jacobson 's organ found in mammals and squamates, and tlie use of tliat name is therefore not reconnnended. In all other reptiles, a depression is formed in the ventromedial wall of the nasal pit ; this depression, which is the anlage of Jacobson 's organ, appears well before the nasal processes fuse to separate the naris externns from the choana. The anlage rapidly disappears in the crocodilians, in which Jacobson 's organ is lacking in adult animals, but in the Rhynchocephalia and Squa- mata it typically enlarges to form a tubular or spherical pocket off the median nasal wall. In some lizards, such as Chamaelco (Haas, 1947), Jacobson 's organ may be much reduced, but some trace of it appears to persist in all members of these two orders. Jacobson 's organ remains as a pocket off the medial wall of the choana in the adult Sphenodon, while in squamates its duct opens into the mouth cavity, and is separated from the nasal cavity by the closure of the anterior half of the primitive choana. Adult birds do not possess a Jacobson 's organ. Correlated with this is the complete lack of an accessory olfactorj- bulb in the forms studied b.y Crosby and Humphrey (1939) ; even in Apteryx, in which the nasal cavity is far more complex than in other birds, Craigie (1930) reports the total absence of that structure. However a very small, though apparently typical anlage of Jacobson 's organ has been found in embryos of at least three species. Cohn (1902) first reported such an anlage in chicks of 5.6 to 5.9 mm. headlength in which the nasal cavity is a very simple pit; in these, Jacobson 's organ forms as a small groove in the ventromedial wall of the nasal pit, but quickly dis- appears. No trace of it was found in any later stages. Cohn's observations have more recently been confirmed by Zuckerkandl (1910b; Vanellus) and Schiiller (1938; Sterna). Two other workers reached very different conclusions on the basis of adult structure. Ganin (1891; also published in abstract in 1890) suggested that the medial duct of the glandula nasalis externa might represent, in part, the Jacobson 's organ of birds. This duct enters the anteromedian wall of the nasal cavity. His suggestions received support from von Mihalkovics (1898), but more recent workers have not accepted the theory. The argu- PARSONS : NASAL EMBRYOLOGY 253 ments against it are the same as those already given against von Mihalkovies' proposed homology between tlie dnct of the chel- onian glandiila nasalis medialis and the Jacobson 's organ of other amniotes, and need not be reviewed here. A Jacobson 's organ is typically present in mammals, although in some groups, such as the Chiroptera, Cetacea, and higher Primates, it is absent. The anatomy of this organ has been de- scribed in a large series of forms by many different workers, starting with Jacobson (1950; the original papers were published in 1811 and 1812). In the adult, its relationships are somewhat variable, but it is always ventromedially located within the nasal area. Most often the organ becomes associated with the naso- palatine duct (= incisive or Steno's duct), which represents the anterior end of the primary choana ; thus it lies in a position closely comparable to that in Sphenodon. In almost all cases the organ is a tubular structure whose anterior end opens into the nasal cavity or nasopalatine duct, and whose posterior end terminates blindly along the surface of the nasal septum. Herz- feld (1888) describes the range of variation found within this class. The early embryology of the mannnalian Jacobson 's organ is virtually identical with that of the Squamata. Its anlage forms as a small pit or groove in the medial nasal wall, at approximately the center of its ventral margin. This indentation appears at a very early stage, when the nose is represented by a simj^le nasal pit. Jacobson 's organ grows larger and soon becomes a tubular structure. Subsequent changes in its structure are concerned primarily witli the establishment of the various adult conditions, and do not need further comment. Thus in all amniotes, except the turtles, Jacobson 's organ arises in a similar fashion, and there is no reason to question its strict homology in all these forms. The modern Amphil)ia display a variety of conditions in the adults, and the euibryology of the nose is poorly understood. In the Gymnophiona and especially in the Anura, the unterer Blindsack does appear to he very similar to the Jacobson 's organ of amniotes; however, the variation among the Amphibia and the very different structure of the chelonian nasal cavities raises the possibility that some of the resemblances may be the result of jjarallel evolution. 254 BITLLETIN : MUSEITM OF COMPARATIVE ZOOLOGY Among the amniotes in which Jacobson's organ is unqnestion- ably present, it generally becomes associated with the buccal rather than the nasal cavity of the adult. However, this is not universally true ; rodents and lagomorphs possess well-developed Jacobson's organs whose luniina are joined to the nasal cavities only. It therefore appears to be impossible to assume, as some workers have, that the organ always functions in the testing of the mouth content. Since the Jacobson's organ of Sphenodon retains its embryonic position in the wall of the nasal cavity, it is generally assumed that the oral connection of that organ in mammals and sciuamates is a parallel development rather than a trait inherited from a common ancestor ; this assumption ap- pears to be supported by the differences in the embryology of that connection in the two groups. However it is very difficult to conceive of the rodents and lagomorphs being the only mam- mals to retain the primitive condition ; in these cases, a suspicion that the nasal connection is a secondary retention of an embrj^onic character would appear to be justified. Early Embryology of the Nasal Cavities. In all vertebrates, the earliest stage in the differentiation of the nasal cavities is the formation of a thickened area of ectoderm, the nasal placode, on the lateral surface of the snout. The placode soon becomes iupocketed to form the nasal pit in most forms ; in amphibians there is still dispute concerning the mode of origin of the pit. Reports vary slightly on the sharpness of the boundary between the nasal epithelium of the placode and pit and the epidermal epithelium surrounding them. In all the reptilian embryos studied in the present investigation, there is a gradual transition, without any definite line of division. Such an absence of a bound- Avy is probably characteristic of the youngest stages in amniotes generally; Cohn (1902) found none in the chick until the nasal pit was well developed. At that time, ridges appear at the margins of the nasal epithelium, and a teloderm or superficial epithelial layer is formed over the epidermal epithelium. No teloderm was seen in the reptilian material used in the present study. In early mammalian embryos, various workers, such as Seydel (1899) and Frets (1912), have described a transition between nasal and epidermal epithelia. However, a majority of the investigators who have studied amphibians report a quite sharp boundary between the two areas. PARSONS : NASAL EMBRYOLOGY 255 Another difference between the Amphibia and the amniotes concerns the histological structure of the placode. In the former class, the epithelium in which the placode forms is generally said to be two layered (Hinsberg, 1901 and 1902; Kurepina, 1931; Schuch, 193-1; ct al.) ; the thin outer layer is termed the Decl- srhichf, and the thicker inner layer the Swnesschicht. The thicken- ing which forms the placode occurs entirely within the latter inner zone. Hinsberg, Kurepina, and other workers believe that the Deckschicht disappears in the area which will become sensory, and that the first indentation of the nasal pit is formed by the loss of this lawyer. However, Schuch considers it to be more likely that the layers fuse in the presumptive sensory area; the inpocketing would then he the result of active growth of the placodal epithelium. At least one investigator, Marcus (1930), has denied the existence of two distinct layers within the epithelium on the basis of his studies of Hypogeophis; his obser- vations are questioned by most other workers. The nasal pit enlarges and, in all amniotes, becomes a deep groove which extends dorsally or dorsolaterally from a rather slit-like naris along the ventrolateral margin of the snout. During this period of growth, many mitotic figures are seen in the nasal ejiithelium, and it is believed that the increase in the depth of the pit is primarily the result of the growth of the epithelium ( Peter, 1900, Lacerta; and Frets, 1912, mammals). The development of lateral and medial nasal processes on either side of the original pit also aids in this iiici-eased deepening. As in most questions on nasal development, the situation in the Amphibia is unclear : Schuch (1934) i'e])oi'ts a similar formation of the pit through epithelial growth, but Hinsberg (1901) states that the cavity is formed by a ''Dehiscenz der Zellen der Geruchsplatte " (p. 476). The next major step in the nasal embrj^ology of all tetrapods is the development of separate anterior and posterior openings, the nares externi and choanae respectively. Since the reptiles are the only group which was actually studied in the present investi- gation, and since they are believed to display the primitive method of choanal formation, tiiey are here considered first. In all the reptiles which have been studied, the single primitive naris is a long slit whose anterior end is near the tip of the snout and whose posterior end lies within the roof of the develop- ing mouth cavity. The portion of the head lateral to the nasal 256 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY pit grows ventrall}^ and anteriorly to form the lateral nasal process ; the median nasal process develops simultaneously at the ventromedial margin of the pit. The ventral ends of these two processes meet and become fused ventral to the central portion of the nasal pit. However they do not join at either the anterior or l)osteri()r ends of the ])rimitive naris; thus the nasal cavity remains open at either end, and becomes tubular. The smaller anterior opening lies on the ventrolateral surface of the snout and is the naris externus, while the larger posterior one forms the primary choana Avhieh lies on the roof of the mouth. The bar of tissue between the naris externus and the choana, which is com- posed of the fused ends of the nasal processes together with the maxillary process, is the primary palate. In birds the formation of the choanae occurs in precisely the same manner as in reptiles (von KoUiker, 1860; Cohn, 1902; and many other authors). Certain modifications of this method of choanal formation occur in the mammals, but the process is still generally considered to be quite comparable to that described for reptiles and birds. The earlier descriptions of mammalian nasal embryology, e.g. von KoUiker (1860), report the fusion of the nasal processes as taking place exactly as described above. However, Hochstetter (1891) states that in the rabbit and the cat the nasal and maxillary processes fuse ventral to the entire posterior portion of the nasal pit, leaving only a single anterior opening, the future naris externus. He also appears to deny that the posterior end of the nasal ])it was continuous with the mouth cavity before the fusion of the processes, l)ut liis paper is not absolutely clear on this point. Ventral to the central portion of the pit, there is formed a primary palate similar to that in reptilian embryos, hut further posteriorly there is only a thin ectodermal membrane, the membra no bncco-nasalU, separating the buccal and nasal cavities. This memln-ane soon ruptures and the primary choana is thus formed. In the following year, 1892; Hochstetter extended his observations to man, and Keibel (1893) found a similar development of the clioanae in tlie pig and other mammals. Since 1893 Hochstetter 's views have been virtually universally accepted. However, Seydel (1899) was unable to find the membrana bucco-nasalis in Echidna embryos, and believes that the condition there resembles that in reptiles. Others workers, such as Peter (1901), have questioned Seydel 's conclusions, PARSONS: NASAL EMBRYOLOGY 257 doubting' that liis series of enibiyos was complete enough to enable him to j-eaeh any definite decision on the absence of the membrana. Unfoi'tuiiately, most of the subseciuent discussions of the choa- nal formation in mammals are not entirely clear on one major point. If the pattern observed in mammals is a modification of the reptilian condition, then one should expect that the membrana bucco-nasalis is composed of the epithelium of the posterior parts of the nasal processes, and that the primary choanae formed by its rupture lie in the same position as did the posterior ends of the primitive nares before the fusion of the nasal processes. Such expectations are fulfilled according to the observations of His (1901; human embryos) and Frets (1912; mice). Thus the only modification is that the primary choana is secondarily closed by a membrane for a short period. However, other workers, including Hochstetter (1891 and 1892), Peter (1901), and Burns (1925), deny that the primitive naris extends posteriorly as far as the mouth cavity; Peter summarizes Hochstetter 's conclusions as follows (p. 55) : "Die primaren Choanen der Sauger . . . entstehen also erst durch Eroffnung des hinteren Endes des Nasenblindsackes, nach Durchreissen der Membrana bucconasalis, und es existiert bei Saugern keine primare Verbindung der Nasenhohle mit der Mundhohle, keine Mundnasenrinne." Should this statement be correct, then it would be very difficult to con- sider the pattern of choanal formation in mammals as a modifica- tion of the process observed in reptiles. Rather, the mammalian condition would appear to be basically difiPerent, and any com- parison of the primary choanae throughout the amniotes would be open to doubt. Most general texts, including that of Matthes (1934), do not differentiate between the two possibilities outlined above, or give any clear indication of which interpretation they believe to be correct. Concerning the Amphibia opinions differ, and it is impossible to reach any general conclusions about the formation of the primary choanae in this class. The Gymnophiona show a pattern somewhat comparable to that seen in amniotes, and are therefore considered first. In Hijpogeophis, Hinsberg (1902) describes a small furrow running from the nasal pit posteroventrally to reach the roof of the mouth. Epithelial cells resembling the epi- dermal rather than the nasal type grow inwards to form a ridge 258 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY along the basal surface of the epithelium of the snout under- neath this furrow. As the embryo develops, mesodermal elements separate this ridge from the remainder of the epithelium so that it forms a narrow and solid cylinder of epithelial cells which runs from the nasal pit to the roof of the mouth. The primitive choana is developed by the creation of a lumen within the epithelial cylinder. Thas in the larvae of Hypogeophis there is a rather long tubular N asengawmengang connecting the main nasal cavity with the mouth ; in the adult, the tube is shorter, and the sensor}- region of the nasal cavity reaches to the choanal region. Marcus (1930) also studied Hypogeophis, and his obser- vations agree well with those of Hinsberg, except that the former author recognizes a small lumen within the Nasengaumengang at the time of its formation. Thus in gymnophionans, the groove observed by Hinsberg appears to represent the posterior end of the nasal pit of amniotes, and the tissue on either side of it, the nasal processes. However, in both urodeles and anurans, Hinsberg (1901) and most other early workers report a very different situation. In Bana and Tritnrus, Hinsberg could find no groove connecting the nasal pit with the roof of the mouth. Instead a solid process of epithelial cells from the wall of the pit grows posteroventrally through the intervening mesoderm to reach and fuse with the buccal epithelium. The choana is formed, as in the Gymnophiona. by the extension of the nasal cavitj' into the solid epithelial process, and finally by the breakdown of the epithelium of the roof of the mouth at the point where the epithelial process has fused with it. According to Hinsberg, this point of fusion is pos- terior to the pharyngeal membrane; therefore the primitive choana lies Avithin the endodermal portion of the mouth, while in all amniotes and gymnophionans it is the ectodermal region. Such a development is so radically different from that found in other forms that Peter (1901) has questioned the homology of the amphibian choana with that in amniotes; thus he also ex- pressed doubt that the unterer Blindsack can be homologized with Jacobson's organ because of the general incomparability of the nasal cavities of amphibians and amniotes. Several more recent authors, most notably Kurepina (1931), have disagreed with most of Hinsberg 's observations. In a series of six species of anurans and urodeles, Kurepina reports the PARSONS : NASAL EMBRYOLOGY 259 presence of a small oro-nasale Furchen connecting the nasal pit and tlie moutli cavity. This groove is small, and is present only in very early larvae. The choanae are formed by a strand of epithelial cells Avhich becomes separated from the oro-nasale Furchen in a manner precisely similar to that described by Hinsberg (1902) for Hypogcophis; they lie within the ectodermal portion of the mouth. Thus Kurepina believes that the amphibian nasal cavities develop by means of the same processes which have been de- scribed in reptiles. There is still no general agreement concerning these problems. Foske (1934) reports the formation of the choana in Xenopus from an oro-nasale Furchen similar to that described by Kure- pina, but in the same year Schuch repeats Hinsberg's major conclusions on the basis of his study of Tr it urns. Much of the solid epithelial strand between the nasal pit and the roof of the mouth is believed to be formed from endodermal cells of the latter area by Schuch, while Hinsberg considers it to be primarily an outgroAvth of the nasal pit, but this difference is relatively slight. Thus the Gymnophiona, lieptilia, and Aves all display a similar mode of choanal formation. If the observations of Kurepina (1931) on urodeles and anurans and those of Frets (1912) on mammals are to be credited, then this mode is common to all tetrapods. However, if Hinsberg (1901) and Peter (1901) pre- sent a more accurate picture of amphibian and mammalian de- velopment, then the choanae are formed in a very different man- ner in these two groups; in fact, three somewhat different pat- terns would be present if the amphibian choanae are actually in the endodermal portion of the palate. Further comparison of these patterns does not appear to be profitable until greater agree- ment has been reached concerning the choanal embrj^ology of the Urodela, Anura, and Mammalia. There is no need for any discussion of the later embryonic stages. In all groups, the nasal cavities slowly assume their adult configuration by the development of the various conchae and other structures already mentioned. The reptilian orders have already been described. Papers previously cited wiiich consider later stages in amphibians include those of Hinsberg (1901 and 1902), Foske (1934), and Schuch (1934); Schiiller (1938) presents the best account of the nasal embryology in a bird. 260 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY References to the literature on mammals may be found in the recent paper by Keilbach (1954) which, although concerned primarily with the nasal skeleton, contains an excellent bibli- ography on nasal anatomy and eml)r.yology in all mammals. Phylogenetic Considerations. In conclusion, a discussion of the phylogenetic implications of this study is in order. However, since I am presenting these ideas in some detail in a separate paper (Parsons, in press), the discussion here will be very brief. The most important point is the general similarity of the nasal anatomy and embryology in all groups of amniotes with the ex- ception of turtles. Sphenodon, lizards, snakes, crocodilians, birds, and mammals typically possess at least one concha (the maxillo- turbinal of mammals) which appears to be homologous in all these groups : it is absent in turtles and also in amphibians. Similarly all amniotes (except turtles) possess, in early embryonic stages, a distinct ventromedial inpocketing of the nasal epithelium. This inpocketing is the anlage of Jacobson's organ. However in turtles and amphibians the structures which are probably homo- logous to that organ do not exhibit the same mode of develop- ment. In the absence of any evidence to the contrary, it seems most probable that the ancestors of turtles had separated from the basic reptilian stock at a \evy early stage, before the diver- gence of the lines leading to the other modern amniote groups; before the latter division had occurred, a concha was formed and a typical Jacobson's organ developed. Naturally, the theory stated in the preceding paragraph can- not be accepted as proven. It is possible that some of the re- semblances between various amniote groups are due to parallel evolution or that the turtle ancestors possessed a concha and a typical Jacobson's organ. However the embryology of the nasal organ, as described in this paper, does not appear to support either of these possibilities. Since the early history of the reptiles has been much debated in the literature, it is most interesting to find another anatomical complex, the nasal cavities and Jacobson's organ, which may give some evidence concerning this problem. PARSOXS : NASAL EMBRYOLOGY 2()1 LITERATURE CTTED Allisox, a. C. 1953. The uioipliology of the olfactory sj'steni in the vertebrates. BioL Rev., vol. 28, pp. 195-244. Baumeister, L. 1908. Beitriige zur Anatomie und Physiologie der Rhinopliiden. In- tegument, Driisen der Mundliohle, Augen und Skeletsysteni. Zool. Jahrb., Abt. Anat., vol. 26, pp. 423-52G. Beard, .1. 1889. Morphologk'al studies. Nr. 4. The nose and Jacobson 's organ. Zool. Jahrb., Abt. Anat., vol. 3, pp. 753-783. Beecker, a. 1903. Yergleichende Stilistik der Nasenregion bei den Sauriern, Vogelu und Saiigethieren. Morph. Jahrb., vol. 31, pp. 565-619. de Beer, G. R. 1937. The development of the vertebrate skull. Oxford, xxiv+ 552 pp. Bellairs, a. d'A. 1942. Observations on Jacobson 's organ and its innervation in Fipera ierus. J. Anat., vol. 76, pp. 167-177. 1949. Observations on the snout of Varanus, and a comparison with that of other lizards and snakes. J. Anat., vol. 83, pp. 116-146. Bellairs, A. d'A., and J. D. Boyd 1950. The lachrj-nial apparatus in lizards and snakes. II. The anterior part of the lachrymal duct and its relationship with the palate and with the nasal and vomeronasal organs. Proc. Zool. Soc. London, vol. 120, pp. 269-310. Bellairs, A. d'A., and C. C. D. Shute 1953. Observations on the narial musculature of Crocodilia and its innervation from the sympnthetic system. J. Anat., vol. 87, pp. 367-378. Belogolowy, J. 1909- Zur Entwickelung der Kopfnerven der Vogel. Ein Beitrag zur 1910. Morphologie des Nervensystems der Wirbeltiere. Bull. Soc. Imp. Nat. AIoscou, n.s., vol. 22, pp. 177-537. Bertau, M. 1933. Zur Entwickluugsgescliichte des Geruclisorgans tier Krokodile. Zeitschr. Anat. Entwickl., vol. 104, pp. 168-202. 262 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Born, G. 1876. Ueber die Nasenhohlen und der Thraneniiasengang dor Amphi- bien. Morph. Jahrb., vol. 2, pp. 577-646. 1879. Die Nasenhohlen und der Thriinennasengang der amnioten Wir- belthiere. I. Morph. Jahrb., vol. 5, pp. 62-140. 1883. Die Nasenhohlen nnd der Thrjinennasengang der amnioten Wirbelthiere. III. Morph. Jahrb., vol. 8, pp. 188-232. Broman, I. 1920. Das Organon vomero-nasale Jacobsoni — ein Wassergeruchy- organ! Anat. Hefte, Abt. I, vol. .'i8, pp. 137-191. 1921. Ueber die Entwickelung der konstanten grossercn Nasenhoh- lendriisen der Nagetiere. Zeitschr. Anat. Entwickl., vol. 60, pp. 439-586. 1939. Ueber die Entwicklung der Geruchsorgane bei den Lungen- fischen. Morph. Jahrb., vol. 83, pp. 85-106. Broom, R. 1906. On the organ of Jaeobson in Sphenorlo7i. J. Linn. Roc. London, Zool., vol. 29, pp. 414-420. Bruhl, C. B. 1886. Zootomie alio Thierklassen filr Lernende, uaeh Antopsien, skiz- zirt. Vienna, Lief. 35. Burns, B. I. 1925. The development of the nose in the spermophile, with particular reference to the part played by the primitive choanae in the formation of the nasal cavity in mammals. Anat. Rec, vol. 31, pp. 27-42. Busch, C. II. 1898. Beitrag zur Kenntniss der Gaumenl^ildung liei den Eeptilien. Zool. Jahrb., Abt. Anat., vol. 11, pp. 441-500. Caibney, J. 1926. A general survey of the forebrain of Sphenodon punctatnm. J. Comp. Neurol., vol. 42, pp. 255-348. Chbistensen, K. 1927. The morphology of the brain of Sphenodon. Univ. Iowa Stud. Nat. Hist., vol. 12, pp. 1-29. COHN, F. 1902. Zur Entwicklungsgeschichte des Geruchsorgans des Hiihnchens. Arch. Mikr. Anat. Entwickl., vol. 61, pp. 133-150. PARSONS : NASAL EMBRYOLOGY 263 Craigie, E. H. 1930. Studies on the brain of the kiwi (Apteryx auMralis). J. Comp. Neurol., vol. 40, pp. 223-357. Crosby, E. C. 1917. The forebrain of Alligator mississippiensis. J. Comp. Neurol., vol. 27, pp. 325-402. Crosby, E. C. and T. Humphrey 1939. Studies of the vertebrate telencephalon. I. The nuclear con figuration of the olfactory and accessory olfactory formations and of the nucleus olfactorious anterior of certain reptiles, birds, and mammals. J. Comp. Neurol., vol. 71, pp. 121-213. Dj.raniyagala, p. E. p. 1932. Notes on the development of the leathery turtle, Dermochelys coriacea. Ceylon J. Sci., Sect. B (Spolia Zeylanica), vol. 17. pp. 73-102. 1939. The tetrapod reptiles of Ceylon. Vol. 1. Testudinates and croco- dilians. Columbo, xxxii + 412 pp. DiEULAFE, L. 1904. Les fosses nasa.les des vertebres (morphologic et embryologie). 1905. J. de I'Anat. Physiol., vol. 40, pp. 268-298 and 414-444, and vol. 41, p]). 102-112, 300-318, 478-560, and 658-680. DiSSE, J. 1897. Die erste Entwickelung des Eiechnerven. Anat. Hefte, Abt. I, vol. 9, pp. 255-300. 1)UGIEL, A. 1886. Ueber die Driisen der Rcgio olfactoria. Arch. Mikr. Anal., vol. 26, pp. 50-60. Dourer, J. 1912. Die Metamorphose der Mundrachenwand der Schildkrote Chely- (Jra serpentina. Morph. Jahrb., vol. 44, p. 661-705 (Fortsetzung 9 of A. Fleisehmann, Die Kopf region der Amnioten.). ECKART, PI. 1922. Das Geruchsorgan einiger ceylonischer Eidechsen (Agamiden). (Fauna et Anatomia ceylanica, Bd. II, Nr. 1.). Jena. Zeitschr. Naturwis., vol. 58, pp. 271-318. Fleischer, R. 1878. Beitrage zu der Entwicklungsgeschichte des Jacobson'scheu Organs und zur Anatomic der Nase. Sitzungsber. Phys.-Med. Soe. Erlangen, Hefte 10, pp. 8-19. 264 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY FoSKR, H. 1934. Das Geruclisorgan von X(noj)ii.s locvix. Zoitsdir. Anat. P^ntwickl., vol. 103, pp. 519-550. Francis, E. T. B. 1934. The anatomy of the salamander. Oxford, xxxi + 381 pp. Frets, G. P. 1912. Beitrage zur vergleit-liendon Anatomic und Oiitogenie der Nase der Primaten. I. Beobaehtungen und Bemerkungen zur Entwick- liing der Nase bei einigen catarrhinen Affen, SJiugern und dem Menschen. Morph. Jahrb., vol. 44, pp. 409-4fi3. FUCHS, II. 1907. I'nteisuihungt'n iiljer Ontogenie und Phylogenie der Gaunien- l)ildungcn l)ei den Wirbeltioren. Erste Mitteilung. Uebcr den Gaumen der Sehildkroten und seine Entwickelungsgeschiehte Zeitschr. Morph. Anthrop., vol. 10, pp. 409-463. 1908. Untersurhungen liber Ontogenie und Phylogenie der Gaumen- Itildungen l)ei den Wirbeltieren. Zweite Mitteilung. Ueber das Munddaeh der Rhynchoeephalen, Saurier, Schlangen, Krokodile und Sanger und den Zusammenhang zwischen Mund- und Nasen- hohle Ijei diesen Tieren. Zeitsehr. Morph. Anthrop., vol. 11, pp. 153-248. 1911. Bemerkungen iiber das Munddaeh der Amnioten, insbesoudere der Sehildkroten und Schlangen. Anat. Anz., vol. 38, pp. 609- 637. 191.3. Ueber den Ban und die Entwicklung des Sehiidels der Chclonc imbricata. Ein Beitrag zur Entwieklungsgeschichte und ver- gleiehenden Anatomic des Wirbeltierschadels. Erster Teil: Das Primordialskelett des Neiirocraniums und des Kieferbogens. I>n : Voeltzkow's Reise in Ostafrika in den Jahren 1903-1905, Wissensch. Ergebn., vol. 5, pp. 1-325. Ganin, M. 1890. Einige Thatsachen zur Frage iiber das dacobson'sche Organ der Vogel. Zool Anz., vol. 13, pp. 285-287. 1S91. Quelques faits a questions sur 1 'organ de Jaeobsou chez les oiseaux. Trav. Soc. Nat. Univ. Imp. Kharkow, vol. 25, pp. 1-40 (text in Russian). Gaupp, E. 1888. Anatomische Untersuehungen iiber die Nervenversorgung der Mund- und Nasenhohlendriiscu der Wirbelthiere. ^forph. Jahrb., vol. 14, pp. 436-480. PARSONS : NASAL EMBRYOLOGY 265 (iF.GENBAUR, C. 1873. Ueber die Nasennmsclichi der Vfifjcl. .lenn. Zcitscln-. X.-if iiiwis., voL 7, pp. 1-21. (Jerard, p. 1954. Orgaiie olfactif. In: Grasse's Traite de Zoologie, vol. ll*, ]>\). 522-552. (iROTH, \V. 1938. Der Ursprung der Riechzelleiineuroblasten und iliii' erste Ent- wic'klung Ins zur Ausliilduiig der Riechneivenanlage beim Kaiiin (lien. Zeitschr. Mikr.-Anat. Fors(di., vol. 43, pp. 207-234. Haas, G. 1947. Jaeobson's organ iii tlio chameleon. J. Moiph., vol. SI, pp. 195-207. Haller, B. 1900. Vom Bau des Wirbelthiergehirns. II. Theil. Emya. Morph. Jahrb., vol. 28, pp. 252-34ri. llALLER, v., Graf von Hallersteix 1934. Kranialnerven. In: Bolk, et al., Handbucdi der vergleichenden Anatomie dor Wirlieltiere, vol. 2, part. 1, pi). 541-684. 11 AX SON, F. B. 1919. The auterioi' cranial nerves of Chelydra serpentina. Washington Univ. Stud., Ser. Sci., St. Louis, vol. 7, pp. 13-41. Helling, H. 1938. Das Geruchsorgan der Anuren, vergleichend-morphologiseh be- traehtet. Zeitschr. Anat. Entwickl., vol. 108, pp. 587-643. llERRICK, C. J. 1921. The connections of the vomeronasal nerve, accessory olfactory l)ulb and amygdala in Amphibia. J. Comp. Neurol., vol. 33, pp. 213-280. 1924. The amjihibian forebrain. II. The olfactory bulb of Amhhj- utonia. J. Comp. Neurol., vol. 37, pp. 373-396. 1948. The brain of the tiger salamander. Chicago, viii + 409 pp. llERRlCK, C. L. 1890. Notes upon the brain of the alligator. J. Cincinnati See. Nat. Hist., vol. 12, pp. 129-162. 1893. Contributions to the comparative morphology of the central nervous system. II. Topography and histology of the brain of ccrtMin reptiles. .J. Comp. Neurol., vol. 3, pp. 77-106. 266 BULLETIN: MUSEUM OP COMPARATIVE ZOOLOGY TTerzfeld, p. 1888. Ueber das Jacobson'sche Organ des Menschen uud der Sauge- tliiere. Zool. Jahrb., Abt. Anat., vol. 3, pp. ^51-574. HiNSBERG, V. 1901. Die Entwicklung der Nasenhohle bei Amphibien. Theil I und II: Anuren imd Urodelen. Arch. Mikr. Anat. Entwickl., vol. 58, pp. 411-482. 1902 Die Entwicklung der Nasenhohle bei Amphibien. Teil III: Gymnophionen. Arch. Mikr. Anat. Entwickl., vol. 60, pp. 369- 385. His, W. 1901. Beobachtungen zur Geschichte der Nasen- und Gaunienbildung beim menschlichen Embryo. Abhandl. Kon. Sachs. Gesel. Wis- sensch., Math.-Phys. CI., vol. 27, pp. 349-389. HOCHSTETTER, F. 1891. Ueber die Bildung der iuneren Nasengange oder ])rinutiven Choanen. Verh. Anat. Gesel., Vers. 5, pp. 145-151. 1892. Ueber die Bildung der primitiven Choanen lieim Menschen. Yerh. Anat. Gesel., Vers. 6, pp. 181-183. Hoffmann, C. K. 1879- Eeptilien. I)i: Bronn's Klassen und Ordnungen des Thier-Reichs. 1890. Leipzig, vol. 6, pt. 3, 3 vols., 2089 pp. HOPPE, G. 1934. Das Geruchsorgan von Eatferia punctata. Zeitsehr. Anat. Ent- wickl., vol. 102, pp. 434-461. Howes, G. B. 1891. On the probable existence of a Jacobson 's organ among the Crocodilia; with observations upon the skeleton of that organ in the Mammalia, and upon the basi-maudibular elements in the Vertebrata. Proc Zool. Hoc. London, 1891, pp. 148-159. ISHIHABA, K. 1932. Zur Kenntnis des Nasenhohlenorgans der A'ogel. Zeitsehr. Anat. Entwickl., vol. 98, pp. 548-.177. Jacobson, L. 1S13. Sur une glande conglomeree appartenante a la cavite nasale. Nouv. Bull. Sci. Soc. Philomat. Paris, vol. 3, pp. 267-269. 1950. Ouvrages sur 1 'organe vomero-nasal. Edited by 0. C. Hollnagel- .Tensen and E. Andreasen. Copenhagen, xxxi -f 175 pp. PARSONS : NASAL EMBRYOLOGY 267 Jarvik, E. 1942. Oil tli(> stnicture of the snout of crossopterygiaiis ancl lower gnatliostoiiu's in general. Zool. Bidrag, vol. 21, pp. 235-675. Johnston, J. B. 1913. Nervus terminalis in reptik's and mammals. J. Conip. Neurol., vol. 23, pp. 97-120. 1915. Tlie cell masses in the forebrain of the turtle, Cistudo Carolina. J. Comp. Neurol., vol. 25, pp. 393-468. Kangro, C. 1928. Beitrtige zur Frage dcs Vorkommens und der Bedeutung der Steuoschen lateralen Nasendriise der Siiugetiere sowie der Driisen des Sinus maxillaris. Zeitsehr. Anat. Entwickl., vol. 85, pp. 376-399. Kappers, C. U. a. 1934. Feinerer Bau und Bahnverbindungen des Zentraluervensystems. In: Bolk, et ah, Handbueh der vergleichenden Anatoniie der Wirbeltiore, vol. 2, part 1, pp. 319-486. Kappers, C. U. A., G. C. Huber, and E. C. Crosby 1936. The comparative anatomy of the nervous system of vertebrates, including man. New York, 2 vols., xxviii + 1845 pp. Kathariner, L. 1899. Anatomische Eigentiimliehkeiten im Bau der Nase der im Wasser lebenden Schlangen. Bull. Soe. Fribourg. Sei. Nat., vol. 7, pp. 186-195. 1900. Die Nase der im Wasser lebenden Schlangen als Luftweg und Geruchsorgan. Zool. Jahrb., Abt. Syst., vol. 13, pp. 415-442. Kkibel, F. 1893. Zur Entwickelungsgeschichte und vergleichenden Anatomie der Nase und des oberen Mundrandes (Oberlippe) bei Vertebraten. Anat. Anz., vol. 8, pp. 473-487. Keilbacii, E. 1954. Vergleichend-anatomische Studien iiber die Saugernasc mit besonderer Beriicksiehtigung des Knorpelskelcttes. Wissensch. Zeitsehr. Univ. Greifswald, vol. 3 (Festschr. Wegner), pp. 87-130. VON Kolliker, a. 1860. Ueber die Eutwicklung des Geruchsorganes beim Mensehen und beim Hiihnchen. Wiirzburger Med. Zeitsehr., vol. 1, pp. 425-435. Krabbe, K. H, 1939. Studies on the morphogenesis of the luain in reptiles. Coi)en- hagen, 87 pp. 268 BULLETIN : MUSEUM OF COMPARATlVTi ZOOLOGY KUNKEL, B. W. 1912. The development of the skull of Ennis httarin. J. ^forph., vol. 23, pp. 693-780. KUREPINA, M. 1931. Die Entwicklung des Gei-uchsorgaiis der Amphiliion. Zool. Jalul)., Al)t. Anat., vol. 54, pp. l-.i4. Larsell, O. 1919. Studies on the nervus tenninalis: turtle. J. Conip. Neurol., vol. 30, pp. 423-443. Lee, S. 1893. Zur Kenntniss des Olfaotorius. Ber. Xaturforsch. Gesel. Frei burg i. B., vol. 7, pp. 179-192. Leydig, F. 1872a. Die in Deutschland lebenden Arten der Saurier. Tiiliiiigen, vii -f 262 pp. 1872b. Zur Kenntuiss der Sinnesorgane der i^t-ldangen. Arch. :\[ikr. Anat., vol. 8, pp. 317-357. 1873. Ueber die Kopfdriisen einheiniisehcr Ophidier. Arc-h. Mikr. Anat., vol. 9, pp. 598-652. 1897. Zirbel und Jacobsou'sehe Organe einiger Rei)tilien. Arch. Mikr. Anat. Entwickl., vol. 50, pp. 385-418. LOCATELLI, p. 1927. Sur la structure du neif olfactif. Arch. Ital. Biol., vol. 77, pp. 208-213. LOEW, J. 1956. Beitrage zur Entwicklungsgesehichte des Jacobsonsclun Organs bei Emys europaea. Acta Zool., vol. 37, pp. 61-85. LovERiDGE, A., and E. E. Williams 1957. Revision of the African tortoises and turtles of the suborder Cryptodira. Bull. Mus. Conip. Zool., Harvard Coll., vol. 115, pp. 163-557. Macallum, a. B. 1884. The nasal region in Eni(uni(i. Proc. Canadian Inst., vol. 1, pji. 390-404. Maeda, M. 1954. Die vergleichend-histologisehe Untersnchnng der sog. Glandulae vestibulares nasi bei den Siiugetieren. 1. Mitteilung. Ueber das Vorkommen der Driise und ihre Verteilung und Grosse. Okajinias Fol. Anat. Jap., vol. 25, pp. 229-233. PARSONS : NASAL EMBRYOLOGY 269 Malan, M. E. 1946. Contributions to the comparative anatomy of the nasal capsule and tlie organ of Jacobson of the Tjacertiiia. Ann. Univ. Stellenbosch, vol. 24, sect. A, pp. 69-137. Marcus, H. 1930. Ueber die Bildung von Geruchsorgan, Tentakel und Choanen bei Hypogeophis, nebst Vorgleieh mit Dipnoern und Polypterus. Beitrag zur Kenntnis der Gymnopliionen XIII. Zeitschr. Anat. Entwickl., vol. i)l, pp. 657-691. Marples, B. J. 1933. The structure and development of the nasal glands of l)irds. Proc. Zool. Soc. London, 1932, pp. 829-844. Matthes, E. 1934. Geruchsorgan. In: Bolk, et al., Handbueh der vergleiehenden Anatomic der Wirbeltiere, vol. 2, jiart 2, pp. 879-948. McCOTTER, R. E. 1912. The connection of the vomeronasal nerves with the accessory olfactory Inilb in the opossum and other mammals. Anat. Eec, vol. 6, pp. 299-318. 1917. The vomero-nasal apparatus in Chrysemyf< punrtnfn ;uid liana cat('.'ihin7ui. Anat. Eec, vol. 13, pp. 51-67. Meek, A. 1893. On the occurrence of a Jacobson 's organ, with notes on the development of the nasal cavity, the lachrymal duct, and the liarderian gland in Crocodilus pnro.su.s. J. Anat. Physiol., vol. 27, pp. 151-160. 1906. The olfactory organs of the crocodile and the homologies of the ethmoid region. Proc. Univ. Durham Phil. Soc, vol. 3, pp. 1-2. 1911. On the morphogenesis of the head of the crocodile. J. Anat. Physiol., vol. 45, pp. 357-377. MeRWE, N. J., VAN DER 1940. J^ie skedelmorfo\ogie van Pelomrdiisd f/alr(t I a (Wagler). Tydskr. Wetens. Kuns, vol. 1, pp. 67-S5. MlHALKOVICS, v., VON 1898. Xasenhohle und Jaeobsonsches Organ. Eine iiKiiipliologische Studie. Anat. llefte. Abt. 1, vol. 11, i)p. 1-107. Xavratil, D., von 1926. Ueber das Jacobsonsche Organ (U'r Wiihcltiere. Zeitschr. Anat. Entwickl., vol. 81, pp. 648-6.-;6. 270 BULLETIN : MUSETTM OF COMPARATRrE ZOOLOGY Nemours, P. E. 1930. Studies on the accessory nasal sinuses: the comparative morphol- ogy of the nasal cavities of reptiles and birds. Ann. Otol. Ehinol. Laryngol., vol. 39, pp. 1086-1108. Nick, L. 1912. Das Kopfskelet von Dermochelys coriacea L. Zool. Jahrb., Abt. Anat., vol. 33, pp. 1-238. Ogushi, K. 1913. Zur Anatomic der Hirnnervcn und des Kopfsympathicus von Trionyx japonicus nebst einigen kritischen Bcmerkungen. Morph. Jahrb., vol. 45, pp. 441-480. OSAWA, G. 1898. Beitriige zur Lehre von den Sinnesorganen der Hatteria punctata. Arch. Mikr. Anat. Entwickl., vol. .12, pp. 268-366. Parker, W. K. 1878. On the structure and development of the skull in the common snake {Tropidonotus nntrix). Phil. Trans. Eoy. Soc. London, vol. 169, pp. 385-417. 1880. Eeport on the development of the green turtle (Chclone viridis, Schneid.). Eept. Sci. Ees. Voyage H.M.S. Challenger, Zool., vol. 1, part 5, 58 pp. Parsons, T. S. 1958. The choanal papillae of the Cheloniidac. Breviora, Mus. Comp. Zool., no. 85, 5 i)p. In press. Nasal anatomy and the phylogeny of reptiles. Evolution. Paulsen, E. 1886. Ueber die Driisen der Nasensehleimhaut, besonders die Bow- man'schen Driisen. Arch. Mikr. Anat., vol. 26, pp. 307-321. Pearson, A. A. 1941a. The development of the nervus terminalis in man. J. Comp. Neurol., vol. 75, pp. 39-66. 1941b. The development of the olfactory nerve in man. J. Comp. Neurol., vol. 75, pp. 199-217. Peter, K. 1900. ]Mittheilungen zur Entwicklungsgesehichte der Eidechse. I. Das Waehsthum des Eiechgriibchens. Ein Beitrag zur Lehre vom embryonalen Waehsthum. Arch. Mikr. Anat. Entwickl., vol. 55, pp. 585-617. PARSONS : NASAL EMBRYOLOGY 271 1901. Die Entwickelung des Geruchsorgaiis iiiul Jakobson 'sclioii Organs in der Eeihe dor Wirbeltioie. Bilduiig der iiiisseren Nase und des Gaumens. In: Ilertwig's Haiidbucli der vergleic-henden und experimentellen Entwickelungslehre der Wirbeltiere, vol. 2, part 2, pp. 1-82. Peyeb, B. 1912. Die Entwicklung des Schiidelskeletes von Fvpera aspis. Mor})li. Jahrb., vol. 44, pp. 563-621. Plate, L. 1924. Allgemeiue Zoologie und Abstammungslehre. Zwciter Teil : Die Siniiesorgane der Tiere. Jena, ix + 806 pp. Pratt, C. W. McE. 1948. The morphology of the ethmoidal region of Sphenodon and lizards. Proc. Zool. Soc. London, vol. 118, pp. 171-201. I'RIXGLE, J. A. 1954. The cranial development of certain South African snakes and the relationshijj of these groups. Proc. Zool. Soc. London, vol. 123, pp. S13-86.J. Rabl-Ruckhard, H. 1894. Einiges iil>er das Gehiru der Riesenschlange. Zeitschr. Wis- sensch. Zool., vol. ."58, pp. 694-717. Rathke, H. 1839. Entwickelungsgeschichte der Natter (Coluber natrix). Konigs- berg, iv -\- 232 pp. 1866. Untersuchungen iiber die Entwickelung und den Korperbau der Krokodile. Braunschweig, vii -j- 275 pp. Reese, A. M. 1901. The nasal passages of the Florida alligator. Proc. Acad. Nat. Sci. Philadelphia, vol. ')'6, pp. 457-464. 1925. The cephalic glands of AUigaior mvisissippiensis, Florida alli- gator, and of AgTciatrodon, copperhead and moccasin. Biol. Generalis, vol. 1, pp. 482-500. Retzius, G. 1894. Die Riechzellen der Ophidier in der Riechschleimhavit und ini Jacobson 'sclicn Organ, liiol. Untersuchungen, n.s., vol. 6, pp. 48-51. ROMER, A. S. 1956. Osteology of the reptiles. Chicago, xxi + 772 pp. 272 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Rose, C. 1893a. Ueber das rudinientiire Jacobson'sche Organ der Crocodile und des Menschen. Anat. Anz., vol. 8, pp. 458-472. 18931). Ueber die Nasendriise und die Gauniendriisen von Crocodihts paluslris. Anat. Anz., vol. 8, pp. 74.')-751. Sarasin, p., and F. Sarasin 1887- Zur Entwicklungsgeschichte und Anatomie der ceylonischen 1890. Blindwiihle Ichthyophis glutinosus, L. Ergelmisse naturwissen- seliaftlicher Forst-hungen auf Ceylon. Wiesbaden, vol. 2, 263 pp. Schauinsland, H. 1903. Beitriige zur Entwic-keluugsgescliichte und Anatomie der Wir- beltiere. I. Sphcnodon, Callorhynchufs, Chamaleo. Zoologica, vol. 16, no. 39, p. 1-98. SCHEPERS, G. W. H. 1948a. A dorso-lateral invagination of the olfactory l)ulb in Tesfiido gcomctrica. S. Afr. Sci., vol. 2, pp. 911. 1948b. Evolution of the forebrain. The fundamental anatomy of the telencephalon with special reference to that of Tesfudo geo- metrica. Capetown, 212 pp. ScHILDMACHER, H. 1932. Ueber den Einfluss des Salzwassers auf die Entwieklung der Nasendriisen. J. OrnithoL, vol. 80, pp. 293-299. Schmidt-Nielsen, K., C. B. Jorgensen, and H. Osaki 1957. Secretion of hyi)ertonic solutions in marine birds. Fed. Proc, vol. 16, pp. 113-114. SCHUCH, K. 1934. Das Geruchsorgan von Triton ulpe^triti. Zool. Jahrb., Abt. Anat., vol. 59, pp. 69-134. SCHtfLLER, H. 1938. Die Entwieklung des Geruchsorgans bei der Sturmmove und der Seeschwalbc. Zeitsclir. Anat. Entwickl., vol. 109, pp. 75-98. Seydel, O. 1895. Ueber die Nasenhohle und das Jacobson'sche Organ der Amphi- bien. Morph. Jahrb., vol. 23, pp. 453-543. 1896. Ueber die Nasenhohle und das Jacobson'sche Organ der Land- und Sumpfschildkroten. Festschr. 70 (Tcburtst. Gegenbaur, vol. 2, pp. 385-486. PARSONS : NASAL EMBRYOLOGY 273 1S99. T'i'I)or Entwit'kelungsvorgansc ;iii dcr Nasciihohle uiid am Muii(lliolilon(lai-he von Echidiw m'hst Bi'itriisi'ii zur Moipliolofjio lies poripheren Geiuehsorgans und dcs Gauiiiciis der Wirbeltliioro. Denksehr. Med.-Naturwis. Gescl. Jena, vol. (5, pp. 445-532. Shiino, K. 1914. Rtudion zur Keantnis des Wirbeltierkopfes. I. Das Choiidro- c-ranium von CrocodiJns niit Beriicksiclitigung der (ieliii'nnerven und der Kopfgefassc. Anat. Hefte, Abt. I, vol. 50, pp. 253-382. SiMONETTA, B. 1933. Kieerdie suU'origine e sulio sviluppo del iiervo olfattivo negli uccelli. Esistano negli uecelli il nervo terminale e I'organo di Jacobson? Arch. Ital. Anat. Eniliriol., vol. 31, pp. 396-424. Sia'ITER, C. P. 1892. Das Jacolison 'sehe Organ von Crocuililu.s p(iro.\us (Helm.). Anat. Anz., vol. 7, pp. 540-545. Smith, M. A., and A. d'A. Bellairs 1947. The head glands of snakes, with remarks on the evolution of the parotid gland and teeth of the Opisthoglypha. .1. l.inn. Soe. London, Zool., vol. 41, pp. 351-368. SOLGER, B. 1876. Beitrage zur Kenntniss der Nasenwaiidung, und besonders der Nasenmuseheln der Reptilien. Morph. Jahrb., vol. 1, pp. 467- 494. Stannius, H. 1848. Vergleichende Anatomie der Wirbelthiere. In: von Siebold and Stannius' Lehrlnich der vergleichenden Anatomie. Berlin, part 2, 482 pp. . 1854. Handbuch der vergleichende Anatomie der Wirbelthiere. Frank- fort (not seen in the original). Stebbins, E. C. 1948. Nasal structure in lizards with reference to olfaction and con- ditioning of the inspired air. Anier. J. Anat., vol. 83, pp. 183- 221. Technau, G. 1936. Die Nasendriise der Vogel. Zugleich eiu Beitrag zur Morpholo- gie der Nasenhohle. J. Ornithol., vol. 84, pp. 511-617. Tello, J. F. 1923. Les differenciations neuronales dans I'embryon do poulet, pendant les premiers jours de 1 'incubation. Tiav. Lab. Reeh. Biol. Univ. :Nradrid, vol. 21, pp. 1-93. 274 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY T HATER, K. 1910. Das Munddach der Schlangen und Sehildkroten. Morph. Jahrl)., vol. 41, pp. 471-518 (Fortsetzung 5 of A. Fleischnianii, Die Kopf region der Amnioten.). 1912. Der Einspruch von Hugo Fuehs. Morph. Jahrb., vol. 44, pp. 707-711. Ungeb, L. 1911. Untersuchungen iiber die Morphologie und Faserung der Rep- tiliengeliirns. II. Das Yorderlurn des Alligators. Sitzungsber. Kais. Akad. Wissensch. (Wien), Math.-Naturwis. Kl., vol. 120, pt. 3, pp. 177-202. 1914. Untersuchungen iiber die Morphologie und Faserung der Kep- tiliengehirns. III. Das Vorderhirn der Hatteria punctata {Sphenodon punctatum). Sitzungsber. Kais. Akad. Wissensch. (Wien.), Math.-Naturwis. Kl., vol. 123, pt. 3, pp. 293-318. VOELTZKOW, A. 1899. BeitrJige zur Entvricklungsgeschichte der Reptilien. Biologie und Entwicklung der jiusseren Korperform von Crocodiliis madagas- cariensis Grand. Abhandl. Senckenb. Naturf. Gesel., vol. 26, pp. 1-150. 1903. Beitriige zur Entwicklungsgcschichte der Eeptilien. YI. Gcsicht- sbildung und Entwicklung der iiusseren Korperform bei Chelo7ic imbricata Schweigg. Abhandl. Senckenb. Naturf. Gesel., vol. 27, pp. 179-190. Watson, D. M. S. 1913. Further notes on the skull, brain, and organs of special sense in Riademodon. Ann. Mag. Nat. Hist., ser. 8, vol. 12, pp. 217-228. 1951. Paleontology and modern biology. New Haven, xii + 216 pp. Weber, E. 1950. Transitorische Verschliisse von Fernsinnesorganen in der Em- bryonalperiode bei Amnioten. Eev. Suisse Zool., vol. 57, pp. 19-108. VON Wettstein, 0. 1954. Sauropsida : AUgemeines — Eeptilia. In: Kiikenthal's Hand- buch der Zoologie, vol. 7, part 1, pp. (2) 321-424. WiEDERSHEIM, E. 1879. Die Anatomic der Gymnophionen. Jena, viii + 101 pp. Wright, R. E. 1883. On the organ of Jacobson in Opliidia. Zool. Anz., vol. 6, pi). 389-393. PARSONS : NASAL EMBRYOLOGY 275 Wyeth, F. J. 1924. Tho development aiul neuromciy of tlie fore-brain in Split nodon punctatus, with special inference to the presence and neuronierie significance of certain paired rnetameric diverticula of the central cavity of the fore-brain. Proc. Zool. Soc. London, 192-4, pp. 923-959. ZUCKERKANDL, E. 1910a. Ueber die Wechselbeziehung in der Ausliildung des Jacobson- schcn Organs und des Rieehlappens nel)st Benierkiingen iiber das Jacobsonsche Organ der AniphiJiieii. Anat. Plefte, Alit. T, vol. 41, pp. 1-7-1. 1910b. Das Jacobsonsche Organ. Ergebn. Anat. Entwickl., vol. 18, pp. 801-843. EXPLANATION OF FIGURES Unless it is otherwise noted in the caption, all text fioures are oriented as follows : transverse sections with the dorsal snrface toward the top of the page and the midline to the left ; frontal sections with the anterior end tow^ard the bottom and the midline to the left ; and sagittal sections or views with the dorsal snrface toward the top and the anterior end to the right. In all fignres of the embryological series studied in the present investigation, the number of the series is given and then, in l)arentheses, its size in millimeters (if known) and its source. The abbreviations for the last are : AMR, Dr. A. M. Reese ; EEW, Dr. E. E. Williams ; and MC, the Minot Collection of the Harvard Medical School. The abbreviations used on the figures are as follows: ABS — apikale Blindsack AOB — accessory olfactory bulb AON — anlage of the olfactory nerve AOR — Antorbitalraum B — brain C — concha CA — concha anterior CH — choana CHF — Choanenfalte CHG — Choanengang CHP — choanal papillae 276 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY CNP — cavuiii nasi proprium CP — concha posterior D JO — duct of Jacobson 's organ DLM — dorso-lateraler Muschelraum DNP — clnctiis nasopharyngens DRY — dorsal recess of the vestibnliim DTC — dorsale Tasche des Choanengangs E — eye EP — ependyma GL — glomerular layer GNE — glandnla nasalis externa GNM — glandnla nasalis medialis TILN — hintere laterale Nebenhohle HPC — Holde in der Postconcha IGL — inner granular layer IPL — inner plexiform laj'er JO — Jacobson 's organ LD — lachrymal duct LG — laterale Grenzfalte LNP — lateral nasal pi'ocess LVB — lateral ventricle of the brain MB — mushroom body MCL — mitral cell layer MG — mediale Grenzfalte MNP — medial nasal process MS — myelos])ongium MW — Musehelwulst N — iiaris NE — naris externus NPL — nasal placode NPT — nasal pit OB — olfactory bulb OGL — outer granular layer ON — olfactory nerve OPL — outer plexiform layer PRO — preconcha PTC — postconcha PVR — postvestibular ridge RCC — recessus eaviconchalis PARSONS : NASAL EMBRYOLOGY 277 IvD — reeessus doi-salis of tlic rcti'io intci-inodialis RDX — reeessus ducti iiaso])haryngei REC — recossus extraconelialis RT — rojiio iiiteiniuHlialis RJO — -lat'obson 's or<;'an of Rose (,1898a) RO — reg'io olfaotoria RFC — reeessiis precoiiehalis RV — reeessiis vciitralis of the regio intermedial is S — Sakter SA — sulcus anterior 8L — sulcus lateralis SM — sulcus medialis SMX — sinus m axillaris SPT — sinus postturhinalis STT — Stanunteil SV — sulcus ventralis SVM — sulcus ventralis et medialis Y — vestibulum VM — ventraler ]\Iuschelraum VNN — vomeronasal nerve VP — Vomerpolster PLATE 1 Figure 74. (' jijiir /i;/ iir<. Clirtiscnii/.s lOliJ (10.4; MO. Transverse section sliuwinjj- the fnsed ends of the lateral and medial nasal processes. Tlie nasal cavity is to the rig^ht and the oral cavity to the left (section 256; see Fig. Ki). 380.\. Fi>;ure Zo. Loin r /i;/iin. (_'lir!/s( in i/s '21'.-'>7 (8.9; MC). Transverse section ::howing' (right) the lateral hand of thin epithelium between the dorsal and ventral areas of thickened nasal epithelium. The doi-sal surface is toward the bottom of the page (section 350). 380.\. l.»<" r- - ri VW.'.t- .^^ bW-M^L^A.^ *-v TLATE 1 PLATE 2 Figure 7G. Upper fliiare. Clirysemys 1653 (32.0; MC). Transverse section through tlie vestilniluiu. Tlie dorsal surface is to the right and the midline toward tlie top (section 25). 275x. Figure 77. Loiccr figure. Chr ii.se iin/s 1U9S (2S.1 ; MC). Frontal section througli the sensory epithelium of tlie regio olfactoria showing a Bowman's gland. Tlie anterior end is to the left and the midline toward the top (section lo'2 ) . 275x. --^, y>.<>^c»'*' 1 % • ■•'»-■ 6 . ». .-~5^ : .t '» -'a «■ ^.* ,, S 4 . , • -3 '■/'^^^^^^.^^ #/ -c. ,<>.X^ -■ • > -^,^, ■.v-'-.-vx:^ . ^ V . ' - ^nSffBa.' •■*,''•' .■>? . ;' '•* '- • . . ~ C?^%t , •. ' \ i «. ' „ '. '• '. • ■'■*V^ »' •*'..«' ' ' ^ ■#' "7). 275x. J'^igure 7!». Loin r fi/nin\ Clir/isrnnis 11-7 (32.0; M('). Transvei'se section through the sensoi-y epitheliuui of the- regie iiiterniedialis. The dorsal surface is to the right and the midline towai'd the top (section 7(i). 275.\. ^^-d!0&':ik^^^ ^^j^^m^At^. :«'^™1lt!5*8«t, l^^ls^s^, «.'.v^a<^ .:XS»S?_v**^^ PL ATI-: 3 PLATE 4 Figure 80. Upper fgiirr. Sphenodon ln07 (17.4; MC). Frontal section througli the nasal area to show the relationships of the conchae anterior and posterior. The anterior end is to the left and the midline toward the top (section 1.^.5). 35x. Figure 81. Lower figure. Tliamnophis 132() (headleugth, 8.0; MC). Transverse section through the vestibuluni showing the Fiillgewebe. The dorsal surface is to the left and the external surface of tlie snout toward the lidttdin (section .STO"). '27.1x. jn.ATK I PLATE 5 Figure 82. Upper figure. Tliamiiophis U2G (headlength, 8.0 ; M(^). Tnms- verse section through the sensory epitlieliuin of the Snkter showing Jiow- man's ghinds. Tlie iloisal surface is to the lower riglit and the midline to tlie uiiper riji'lit (section -toS). i^Z-lx. Figure 83. Loicer figure. Thiuunop]ii.<< WVH^ (headlength, 8.0; MC). Transverse section through the non-sensory epithelium of the anteroventral |i(iition of the cavum nasi iir()i)rium. The dorsal surface is to the light and the iiiidliiu' toward the top (section 388). 273x. :i^.r';^^^f?^P<..:^ I'LATE 6 J^'igiii-c 84. Upper fir/urr. Thamnophis 1:^50 (7.4; MC). Transvei'se section through Jacobson 's organ at a stage immediately preceding the formation of the columns. The dorsal surface is to the right and the midline toward the top (section 194). 275x. Figure 8.1. Lnwcr figiiri . Thanrdopliis 1351 (10.0; MC). Transverse section through Jacolison's organ showing the earliest stage in the forma- tion of the columns. The dorsal surface is to the right and the midline toward the top (section 291). 27.1x. •^5 ^'!'^^k« r>#! PLATE 7 Figui-e 86. Uppi r fiz/iirc. Thanniopliis 1M:23 (11.0; MC). Transverse sec- tion tlirouoh .Tacoljson 's oroan sliowhis' the continued growth of the cohnnns. The dorsal surface is to the riglit and the midline toward tlie top (section •1-lQ). 27ox. Figure 87. Lower fii/iirc. Tliphis l.'llMi (headlength, 8.0; MC) . Trans- verse section through Jacobson's organ showing the fully developed columns of the later embryonic stages. The dorsal surface is to the right and the luidlinc toward the top (section 438). 275x. "■'^^^^•'^# - > ^4i rj.ATi: 7 Bulletin of the Museum of Comparortive Zoology AT HARVARD COLLEGE Vol. 120, No. li THE RODENTS OF THE DESEADAN OLIGOCBNE OF PATAGONIA AND THE BEGINNINGS OF SOUTH AMERICAN RODENT EVOLUTION By Albert E. Wood and Bryan Patterson CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM May, 1959 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVAPtD COLLEGE Bulletin (octavo) 1863 — The current volume is Vol. 120. Breviora (octavo) 1952 — No. 106 is current. Memoirs (quarto) 1864—1938 — Publication was terminated with Vol. 55. JoHNSONiA (quarto) 1941 — -A publication of the Department of Mollusks. Vol. 3, no. 38 is current. Occasional Papers op the Department op Mollusks (octavo) 1945 — Vol. 2, no. 22 is current. Proceedings op the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing publications are issued at irregular intervals in num- bers which may be purchased separately. Prices and lists may be obtained on application to the Director of the Museum of Comparative Zoology, Cambridge 38, Massachusetts. Of the Peters "Check List of Birds of the World," volumes 1-3 are out of print ; volumes 4 and 6 may be obtained from the Harvard Uni- versity Press; volumes 5 and 7 are sold by the Museum, and future volumes will be published under Museum auspices. Bulletin of the Museum of Comparative Zoology AT HARVARD COLLEGE Vol. 120, No. 3 THE RODENTS OF THE DESEADAN OLIGOCENE OF PATAGONIA AND THE BEGINNINGS OF SOUTH AMERICAN RODENT EVOLUTION By Albert E. Wood and Bryan Patterson CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM iViAY, 1959 No. 3 — The Rodents of the Deseadan OUgocene of Patagonia and the Beginnings of SdiitJi American Rodent Evolution By Albert E. Wood and Bryan Patterson CONTENTS Page Introduction 282 Molar terminology and measurement 286 Taxonomy and Morphology 289 Suborder Caviomorpha 289 Superfamily Oetodontoidea 294 Octodontidae 294 Platypittamys Wood 296 Platypittamys hradlxyodon Wood 296 Echimyidae 300 Deseadomys gen. nov 303 Deseadomys aramhourgi sp. nov 303 Deseadomys loomisi sp. nov 310 Superfamily Chinehilloidea 312 Chinchillidae 313 Scotamys Loomis 313 Scotamys antiqims Loomis 313 Dasyproctidae 323 Cephalomys Ameghino 329 Ceplialomys arcidens Ameghino 351 Cephalomys plexus Ameghino 354 Litodontoviys Loomis 358 Litodontomys chuhiite7isis Loomis 358 Dasyproctidae, gen. et sp. indet 362 Superfamily Cavioidea 365 Eocardiidae 365 Luantinae subfam. nov 368 Asteromys Ameghino 368 Asteromys punctus Ameghino 369 Eocardiinae 373 Chubutomys gen. nov. 373 Chuhuiomys simpsoni sp. nov 373 Eocardiidae gen. et sp. indet 375 Superfamily Eretliizontoidea 377 Erethizontidae 377 Protosteiromys gen. nov 377 282 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Protosteiromys medianus (Ameghino) 382 Protosteiromys asmodeophilus sp. nov 383 Caviomorpha indet 385 Discussion 387 Bearing of the rodents on the age of the Deseadan 387 Eelationships of the Deseadan rodents to later forms and to each other 388 The origin of the Deseadan rodents 396 Ancestry of the Caviomorpha 401 Problems of rodent phylogeny 407 Classification of lodents 415 Summary 421 Eef erences 422 INTRODUCTION Several years ago, one of us (Wood, 1949) undertook the study of two excellent rodent skeletons from the Deseadan of Patagonia in the collections of The American Museum of Natural History. This work emphasized the need for a review of the Deseadan rodent fauna as a whole, a project that we had had in mind for many years and for which each of us had been taking notes in somewhat desultory fashion. A first draft, based on material in the Amherst collection, was written by Wood. This was re- vised and extended by Patterson on the basis of specimens in Chicago Natural History Museum, The American ]Museum of Natural History, and the Museum National d'Histoire Naturelle. A second draft, on which we collaborated, was then prepared. After this had been completed, the opportunity to convert the review into a monograph was afforded by the award of a John Simon Guggenheim Memorial Foundation Fellowship to Patter- son for work in Argentina. Study of the types and other spec- imens in the Ameghino Collection revealed so much that was in- teresting and unexpected that the second draft had to be ex- tensively revised. We then collaborated on the writing of a third draft. This was finally rewritten to include a discussion of the series of important papers on rodent phylogeny that have appeared during the last eight years. The work has indeed been a long time in preparation, but we believe that its present com- pleteness fully justifies the delay. The number of specimens available is surprisingly large. There are approximately 70 (mainly isolated teeth) in the Museum WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 283 National d'Histoire Naturelle; 54 (including many complete jaws or upper dentitions) in the Amherst College Museum; 27 in the Ameghino Collection, now preserved in the Museo Argentino de Ciencias Naturales; !) in Chicago Natural Ilistoiy Museum; !) in the Yale Peabody ^Museum ; 5 in the Museum of Comparative Zoology; and 11 in The American Museum of Natural History; a total of over 185. Loomis (1914, p. 20) mentions 102 rodent specimens, all but 6 being Cephalomys, collected by the Am- herst Expedition (not by the Princeton Expeditions, as stated by Landry, 1957a, p. 53 — the Princeton Expeditions, to Hatcher's sorrow, never saw the Deseado). In his text, however, (pp. 189, 190 and 192) Loomis gives totals of specimens under individual species of Cephalomys that add up to 79, for a total rodent col- lection of 85. We have found 74 numbered specimens collected by Loomis (54 in Amherst, 9 in the Yale Peabody Museum, 5 in the Museum of Comparative Zoology and 6 in the American Museum of Natural History). Some of these consist of groups of isolated teeth. We do not know whether this is Loomis' entire collection, or whether there are, in other museums, additional specimens, or lots, that we have not been able to locate. The locality known as Cabeza Blauca on the Rio Chico del Chubut, Province of Chubut, has yielded the largest number of specimens. All of the specimens in Amherst, Yale and the Mu- seum of Comparative Zoology, all but two of the Chicago spec- imens, and all but three of the American Museum specimens are from here. To judge from the color, preservation, adhering matrix and one of two labels preserved with them,^ all but one 1 This labt'l reads "Grao .vacimieiitu del Pyroterio-Rio Chico," wliich can only mean Cabeza Blanca. Other labels clearly referring to this locality read "U-r yacim. Pyroth. Rio Chico," i.e. the first Deseadan (Pyrotherium beds) locality encountered on the way up tlie valley of the Rio Chico del Chubut. It might be assumed from the wording that this was the first Deseadan locality found liy Carlos Ameghino, but such is not the case. Isolated remains of Deseadan mammals from Neuiiuen (I'yrolherimn romeri, Tracliytlnrus spegazzinianus, Parastra- potherium ephebiviim) reached Florentine Ameghino during 1885-1889, and Carlos had come across Deseadan localities in Santa Cruz and Chubut during 1888-1892, but was prevented l)y unfavorable circumstances from making collections. It was not until his seventh expedition to Patagonia, 18'.):i-4, that Carlos was able to collect a Deseadan fauna and to prove that the beds lay beneath the Patagonian formation. The material collected during this expedition formed the basis of Florentino's first paper on this fauna, his "Premiere Contribution" (Ameghino, 1895, pp. 603-COt)). Judging from the forms represented, the color and preserva- tion of the specimens, the adhering matrix and such labels as are present, there can be no doubt that the locality or area was that now Ivnown as La Flecha in Santa Cruz Province, situated a little distance to the south of the inlet of the Rio Deseado. This, then, should be considered the type locality. By a happy coincidence, most of the Deseadan material, collected by Tournouer and studied by Gaudry, who proposed the name Deseado, came from this locality. The localities at Cabeza Blanca and Lake Colhue-Huapi were not discovered by Carlos Ameghino until 1894-1896, and material from them was first described by Florentino Ameghino in his "Deuxi&me Contribution" (1897c). 284 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY of Ameg'hino 's specimens are also from this locality. The Platypit- taniys material in the American Museum is from the Scarritt Pocket, Kinconada de los Lopez, Chubut. All of the material in the Museum National d'Histoire Naturelle is from La Flecha, Santa Cruz, as are the specimens not from Cabeza Blanca in Chicago Natural History Museum and in the Ameghino Collec- tion. The specimens in the Ameghino Collection were, of course, collected by Carlos Ameghino; those in the Museum National d 'Histoire Naturelle by Andre Tournouer ; those obtained by the Amherst College expedition were collected by Frederick B. Loomis, William Stein, Waldo H. Shumway and Phillip L. Tur- ner; in Chicago Natural History Museum by Elmer S. Riggs, John B. Abbott and George F. Sternberg; and those from the Scarritt Expedition, in The American Museum of Natural His- tory, by George Gaylord Simpson, Coleman S. Williams and Justino Hernandez. Tournouer was the only collector who found rodents in numbers at La Flecha. He very probably happened upon a small pocket containing them. The fact that Carlos Ameghino, who clearly had a magnificent eye for minute specimens, found only one isolated molar there is sufficient proof that they are really rare at the locality. The Chicago specimens from there consist of a ramus, and of an isolated molar that was found, during preparation, in the matrix surrounding a large mammal. To date, seven genera have been described from the Deseadan : Cephalomys, Asteromys and Orchiomys Ameghino 1897, Eostei- roniys Ameghino 1903, Scotamys and Litodontomys Loomis 1914, and Platypittamys Wood 1949. The Deseadan species medianus, referred hy Ameghino to Eosteiromys (type species E. liomo- genidens Ameghino 1902 from the Colhue-Huapi) is here made the type of a distinct genus, Protosteiromys, and two other new genera, Deseadomys and Chuhntomys, are proposed. Orchiomys is shown to be a synonym of Cephalomys. Eight genera and twelve species of Deseadan rodents are now known, but frag- mentary remains too incomplete for formal description show that we are as yet far from a complete knowledge of the fauna. The phylogenetic relationships are \evy interesting and, for the most part, quite determinable. Six families, representing all four of the currently recognized superfamilies of South American WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 285 rodents, are represented. Despite this apparent diversity, the Deseadan is not far from showing the basic stage from which the indigenous Neotropical rodent fauna has been derived. Each of the adequately known forms has been compared with its possible ultimate ancestors, particularly the North American Paramyidae and Sciuravidae, the European Theridomyidae, and the African forms included by Wood (1955, p. 172) in the Phiomyidae, although these comparisons may not always be dis- cussed in detail. In no case does the material here considered warrant the statement that any known form from the rest of the world is definitely ancestral to the Deseadan rodents. It is pos- silile, however, to arrive at what we consider to be reasonable inferences as to the source of the South American rodents, and we have gone into this matter at length after the description of the fossils. In addition to the Guggenheim Foundation Fellowship, this study has been assisted by grants to Wood from the Marsh Fund of the National Academy of Sciences, by tenure of a Cutting Traveling Fellowship from Columbia University by Wood, which afforded the opportunity for first hand study of numerous ther- idomyids, by tenure of a Carnegie Corporation Grant-in-aid for Travel awarded to Patterson by the American Association of Museums, which made possible a preliminary examination of the Deseadan material in the Museum National d 'Histoire Naturelle, and by a grant from the National Science Foundation to Wood, which permitted, incidental to other work, the statistical analysis of the material belonging to Ccphalomys. We wish to express our sincere thanks to Professor Camille Arambourg, ]\Iuseum National d 'Histoire Naturelle; Dr. George Gaylord Simpson, The American Museum of Natural History ; Dr. J. T. Gregory, Yale Peabody Museum; and Drs. Agustin E. Riggi, Noemi Cattoi and Jorge L. Kraglievich, Museo Argentino de Ciencias Naturales, for their kindness in permitting the study and de- scription of material in their charge. We are especially obliged to Professor Arambourg for sending the Tournouer Collection to us for detailed study. The statistical studies were made by Frances W. Wood, for whose assistance in this and in many other ways we are deeply appreciative. Regression lines were cal- culated by ]\Ir. Craig C. Black. The aid of Mrs. William D. Turn- bull and of Mrs. Elizabeth E. Wareham in preparing several 286 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY drafts of the manuscript is gratefully acknowledged. Figures 21, 24 and 26 were drawn by Senorita Estela Leseano, Figure 1 by Mrs. Dorotliy Marsh ; the remainder are by Wood. The following abbreviations are employed: A. CM., Amherst College Museum; A.M.N.H., The American Museum of Natural History; C.N.H.M., Chicago Natural History Museum; M.A.C.N., Museo Argentino de Ciencias Naturales; M.C.Z., Museum of Comparative Zoology: M.N.H.N., Museum National d'Histoire Naturelle; and Y.P.M., Yale Peabody Museum. MOLAR TERMINOLOGY AND MEASUREMENT As discussed in detail on later pages, we believe that all in- digenous South American rodents were derived from ancestors whose molars had four transverse crests on each tooth, both uppers and lowers. Adopting the terminology suggested by Wood and Wilson (1936), we employ the following terms for these crests : auteroloph, protoloph, metaloph and posteroloph in the upper teeth ; anterolophid, metalophid, hypolophid and posterolophid in the lowers (Fig. 1). There is some uncertainty, due to a tendency to shift position, as to whether the terms protoloph and metaloph in the upper teeth, and metalophid and hypolophid in the lowers, are strictly applicable throughout, or whether such terms as protolophule II or hypolophulid I should be employed in some cases. The crests are strictly homologous throughout this group of rodents, however, and we have there- fore thought it best to employ a uniform nomenclature rather than to introduce new terms whenever shifts occur in the relation of, say, the protoloph to the protocone or of the hypolophid to the hypoconid. For the valleys between these crests, we have borrowed terms from Stirtoii's nomenclature for castorid teeth (1935), as mod- ified by Black and Wood (1956) . In the uppers, these are : para- flexus, mesoflexus and metaflexus (external) and hypoflexus (in- ternal) ; in the lowers: hypoflexid (external) and anteroflexid, mesoflexid and metafiexid (internal). Whenever a valley is eon- verted into a lake, either by wear or by the complete union before wear of the extremities of the flanking crests, the endings -flexus and -flexid become -fossette and -fossettid, e.g. parafossette, meta- fossettid (Fig. 1). In general, our terminology agrees with that WOOD AND PATTERSON : OLIGOCENE RODENTS OP PATAGONIA 287 of Fields (1957, Fig. 2). We do not, however, agree with his use of parafossettid and paraflcxid in the lower teeth since we can- not accept the implied suggestion that these teeth incorporate a paraconid, and we therefore prefer the terms anterofossettid and anterofiexid. A second point of disagreement concerns the ter- Fig. 1. Key to terminology applied to crests and valleys of molar teeth. A and B, RM^ and EMi of Deseadomys aram'bourgi gen. et sp. nov.; A, A.C.M. no. 3163; B, type, M.N.H.N. no. 1903-3-1. C, LM^, reversed, of Protosteiromys medianvs (Ameghino), A.C.M. no. 3014. D, RM>5 of Ercthizon dorsatum epixanthnm Brandt, M.C.Z. no. 36718. A-C, x 6; D, x 3. Abbreviations: afd., anterofossettid; al., anterolopli ; aid., anterolophid ; eld., ectolophid; /)/., hypoflexus; ///(/., hypoflexid ; hid., hypolophid ; ml, metaloph; mid., metalophid; msf., mesoflexus; msfd., mesoflexid; mtf., metaflexus or metaf ossette ; mtfd., metaflexid; mii., nnirc ; nf., neofossettc; nf.^, lingual neofossette; nl., neoloph; pf., paraflexus; pi., postcroloph; jdd., posterolophid ; prl., protoloph. 288 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY minology in the lingual side of the upper teeth, where Fields calls the valley that separates the anteroloph from the protocone the hypoflexiis, instead of using that term, as did Stirton and as do we, for the valley between the protocone and the hypocone. There is a rather common tendency among South American rodents for the development of a fifth crest in the upper molars. Less commonly, a fifth crest may arise in the lower molars. These crests arise from the posteroloph above and from the anterolophid below. As pointed out farther on (pp. 334-336), they have noth- ing to do with the mesoloph and mesolophid of the Theridomyidae and other rodents ; such crests do not occur in the permanent molars of caviomorphs. For descriptive convenience, we employ the terms neoloph and neolophid for these fifth crests, and neo- fossette and neofossettid for the lakes isolated by them. Fields' entoflexus and entofossette are the same as our neoflexus and neofossette. We prefer definite names for these crests and valleys to such more general terms as first, second, etc. anticlinals and synclinals employed by Stehlin. The latter have the merit of simplicity but the defect of implying homology wherever used. A five-crested caviomorph molar and a five-crested theridomyid molar, for example, do not, we believe, have all their crests and valleys homologous. The molar terminology is applied to structures on premolars that occupy the same position, relative to the tooth as a whole, as those on molars. It is recognized, however, that some of them, at least, may have had a different history. At the present time, the evolutionary sequence in rodent premolar development is often clouded. The development of the premolars in the Cavio- morpha is a case in point. The tooth measurements given are the maximum diameters, in millimeters, of the various areas measured. Depending on the nature of the specimen, they may be either diameter of the wear surface, maximum diameter of the extra-alveolar portions of the teeth, or the maximum diameter whether extra-alveolar or intra- alveolar. The second group of these measurements has been marked t and the last *. The diameters given are averages of two to four measurements. There is considerable variation in some of the high-crowned teeth as to the location of the maximum WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 289 diameter. The measurements made from figures will not neces- sarily agree with those listed. The drawings have been made looking down perpendicularly on the wear surface. Tn many specimens, however, the wear surface is oblique to the long axis of the tooth, so that the drawing will show, for example, a con- siderable expanse of the lower part of the tooth on the lingual side of the crown, without indicating that there is an extensive over- hang on the buccal side. TAXONOMY AND MORPHOLOGY Suborder CAVIOMORPHAWood and Patterson 1955 (in Wood, 1955) The Caviomorpha may be defined as follows : rodents with en- larged infraorbital foramen, through which progressively passes M. masseter medialis, pars anticus; infraorbital foramen fre- quenth' approaching orbit in size ; angle of jaw of hj'stricognath type ; malleus and incus fused or separate ; dental enamel of inultiserial type but derivable from an ancestral stock with enamel of pauciserial type ; cheek teeth derived from an ancestral stock with four transverse crests on both upper and lower molars ; permanent molars without mesoloph or mesolophid ; metaloph and metalophid frequently unstable, disappearing in some groups; neoloph and neolophid arising in some groups from posteroloph and anterolophid respectively ; restricted to the New World. Known distribution : Oligocene to Recent, South Amer- ica; Pleistocene to Recent, North and Central America and An- tilles. In addition to these characteristics, several supplementary myological features may be cited. M. palmaris longus generally arises from the olecranon or is lost, though in the erethizontids and caviids it arises in the primitive manner from the medial epicondyle, as it does in the Old World hystricomorphs. M. trapezius is generally divided into two or three separate elements, although it is a continuous sheet in the erethizontids, as in Aplodontia and Ilystrix; in the African Petrornus there are three elements. M. omohyoideus is progressively lost, whereas it is present in the Old World Ilystrieomorpha. M. sternoscapu- laris is present in its entirety, but the fibers of the two halves 290 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY are separated at the clavicle, as in Petromus, whereas they are continuous in Hystrix and Thryonomys. There is a tendency to lose the inner head of M. brachialis. The insertion of M. pronator teres has shifted from the distal to the proximal part of the radius. Perhaps these muscle variants are of no great importance in themselves, but it appears to us that they are as significant in separating the Caviomorpha from the Old World Hystricomorpha as are myological characters sometimes cited as favoring the union of the two groups. Of all the characteristics that have been used to classify rodents, there are none so subjective as variations in muscles, no two authors ever seeming to give the same description for a given group of muscles. There are also few systems in the order where individual variation is less Avell understood. Some caviomorphs possess five-crested molars, but the fifth crest (neoloph, neolophid), which was evidently acquired after the group had reached South America, is not homologous with the mesoloph or mesolophid of the Theridomyidae nor (probably) of the Hystricomorpha. In dm4, of certain echimyids {Prot- acaremys,Prospamomys) and erethizontids {Erethizon and Coen- doii) and in P4 of some Santacruzian octodontids (Scianiys), a short crest may develop from the ectolophid posterior to the metalophid. Topographically, this occupies the position of the mesolophid in theridomyids and other rodents, but it is certainly an independent acquisition and does not extend posteriorly to the molar series. Within Sciamijs, there is evidence that this and other minor crests posterior to the anterolophid developed es- sentially at random in P4 (Patterson and Kraglievich, ms.). On the basis of serology, Mood}^ and Doninger (1956) conclude that there is no indication of special affinity between the Hystri- comorpha and the Caviomor])ha, based on studies of Hystrix, Erethizon, Cavia and Dasyprocta. Still another line of evidence that may be cited is that of ectoparasites, Vanzolini and Gui- maraes (1955 a and b) pointing out that there is no special re- lationship between the lice of caviomorphs and of hystricomorphs. We believe that the indicated relationships warrant the sep- aration of the Soutli American rodents as a distinct suborder, the Caviomorpha. AVe believe that it can be considered as demon- strated that the South American forms were not descended from WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 291 the European Theridomyidae, whether or not any of the Old World forms were. There does not seem to be any possibility that the Old and New World rodents f>'enerally placed in the "Ilystrieomorpha" (serisu lata) could have had a common ancestor later than the Early Eocene members of the Paramyidae. The similarities between the Old and New World forms must then all be examples of parallelism. Cabrera (1927) has reached broadly comparable conclusions, although without presenting supporting evidence. Lavocat (1951b) expressed a similar opinion, although he continued to place the Old and New World forms in a common group, based on jaw structure, and distinct from that in which the Paramyidae are placed. In another paper (1951a), he questioned there being any special relationship be- tween the two groups, although later (1956), as discussed below, he returned to the possibility of special, though not close, relation- ships. Schaub (1953 a and b), Viret (1955), and Landry (1957a) have supported affinity between the Old and New World "hystri- comorphs, ' ' which we do not accept for a variety of reasons given in detail below. Since there are morphologic grounds on which the Old and New World "hystricomorphs" can be separated from each other, and since the paleontologic and particularly the paleogeographic evidence is against the two groups having a common ancestor later than the Early Eocene, we feel entirely justified in establish- ing a suborder for the South American forms, even though the morphologic criteria for doing so may appear to be neither excep- tionally strong nor exclusive to the group. Similar situations will almost certainly be encountered as knowledge of rodent phyl- ogeny improves. The classic subordinal divisions do not, it seems to us, stand on a secure foundation, and, in view of the abun- dantly demonstrated prevalence of parallelism in the order, can hardly be expected to stand unmodified indefinitely. The microstructure of the dental enamel provides an interest- ing case in point, and one with a direct bearing on the present question. Over a hundred years ago, in his classic work on the microstructure of rodent enamel, Tomes (1850) showed that the crossing enamel layers in the incisors of "hystricomorphs" were each composed of several laminae of enamel prisms, a character that set them apart from other rodents. Here, seemingly, was a character as fundamental as the zygomasseteric structure for 292 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY uniting Old World and New World " hystricomorplis " in the same suborder. Korvenkontio's magnificent work on the same subject (1934) suggests another interpretation, however. The "hystricomorphs" are indeed characterized, as Tomes showed, by crossing enamel layers composed of several laminae, Korven- kontio's multiserial type, in contrast to "sciuromorphs" and ' ' myomorphs, " which have the crossing layers composed of a single lamina, the uniserial type. Korvenkontio shows, however, that early, very j)riniitive rodents have a type of enamel struc- ture, which he calls the pauciserial type, that is intermediate be- tween the two extremes, having neither as many laminae to a layer as in the multiserial nor as few as in the uniserial. The pauciserial type occurs both in paramyids and in some theri- domyids." It seems clear to us, therefore, that enamel of mul- tiserial ty]ie could have been evolved independently by the an- cestors of both the Hystricomorpha (sensu stricto) and the Cavio- morpha; the evidence from the histology of the teeth does not now conflict with other evidence derived from gross morphology and from distribution. Sections of an isolated incisor from the Deseadan, probably of Scotamys, have been prepared. These show that, as might have been anticipated, multiserial enamel had already been acquired by this time. The relations of the malleus and incus provide a similar case. Doran (1878, p. 418) stated that in the "Hystricomorpha" these ossicles were almost invariably fused in adults. Tullberg (1899, p. 69) gave such fusion as a character of his Hystricognathi, and Landry has recently (1957a, p. 16) given it great prominence as an item of evidence in favor of the unity of the Hystricomorpha {sensu lato). It must be noted, however, that Cockerell and Miller {in Cockerell, Miller and Printz, 1914, p. 372) found these ossicles to be separable in Proechimys.^ We find that in Echimys armatus, - Because some theridoni.vids have the pauciserial type of enamel and some have the uniserial type, Landry (1957a, pp. 27-28) following Korvenkontio, considers that the Theridomyidae are a composite group, with members belonging to two suborders. This appears to us to be a misinterpretation of the data, which merely indicate that, in this feature, the Theridomyidae are in process of passing from a pauciserial to a multiserial type. The discovery of such transitional forms should, of course, come as no surprise. 3 These authors also questioned one of Doran's exceptions, a young specimen referred by him to Octodon, stating that this looked very sciurine to them. They were quite right ; a malleus and an incus of 0. degtis extracted by us from a skull are typically caviomorph in structure and bear no close resemblance to the bones so labeled by Doran. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 293 Octodon degus, Spalacopus poepigii and Aconaemys sp. the mal- leus and incus are not fused. Since the Octodontidae and Echi- myidae are the most generalized of the caviomorph families, we strongly suspect that the lack of fusion in their auditory ossicles indicates that such fusion was acquired independently within the Caviomorpha.* The condition of the malleus and incus of Pedetes is of some interest in this connection, as showing that very close union is not confined to caviomorphs and hystri- comorphs. Landry, following Tullberg, states that in this form the ossicles are separate, and that ' ' this character, therefore, has proved useful in separating dubious groups from the Ilystri- comorpha" (op. cit., p. 16). Doran had stated that these ossicles were fused in Fedetes. Upon investigation we have found them to be separable but with opposing surfaces complexly inter- locking in both the vertical and horizontal planes. There is clearly even less possibility of movement between the two bones than in octodonticls and echimyids. This anomalous rodent thus exhibits a " sciurognath " mandible, a " hystricomorph " zygomas- seteric region, "hystricomorph" enamel, a " sciurognath " ptery- goid fossa, and a malleus and incus at least as "hystricomorph" in degree of union as are some of those occurring in caviomorphs. Pedetes is, in fact, a standing warning against too dogmatic statements as to what does and what does not characterize major groups of rodents. Our suborder Caviomorpha is not a new concept. The word "caviomorphs" was used by Simpson (1950, p. 376) for exacth' the same group that we are including in the Caviomorpha, but without being formally proposed as a subordinal term, and without definition. Lavocat (1951b, p. 72) proposed a number of divisions of the rodents including one, the " Orthohystrico- gnathes, " for the forms we include in the Caviomorpha. Schaub (1953a) established an Infraorder Nototrogomorpha for the same group. Wood (1955, p. 180) proposed the suborder Cavio- morpha, crediting it to the present authors. For reasons which we elaborate below, we feel that both Lavocat 's and Schaub 's 4 As Fields (1957, p. 347) implies, the auditory ossicles of rodents merit more study than they have received. As a minuscule contribution to this subject, we may record the following additional data as to fusion : Ahrocoma cinercus, juv., separable ; Abrocoma bennetti murrayi, adult, inseparable ; Ctenomys magcllani- cus, adult, inseparable ; Thrt/onomys swindcrianua angolaCj juv., inseparable, area of fusion involving part of crua longus of incus. 294 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY terms contain phylogenetic implications that we distrust, and we are therefore continuing to use Caviomorpha, which, we feel, carries the minimum of phylogenetie implications involv- inp: other groups. The rules of priority do not apply to subordinal units, but it is possible to make a legalistic (although not very impressive) argument, if one wanted, that Simpson's usage establishes the priority and that neither Lavocat nor Schaub provided diagnoses or definitions. In any event, we feel that the entire problem of the subordinal and superfamilial arrange- ment of the rodents is at present in a state of flux, and that the next decade will see a considerable alteration in our knowledge of rodent relationships as indicated by subordinal terminology, so that the precise terminology now employed makes little dif- ference. Superfamily OCTODONTOIDEA Simpson 1945 Family o'CTODONTIDAE Waterhouse 1839 Acareniyinae Anieghiuo 1902 Aearemyidae Wood 1949 The most surprising result of a revision of the Acaremys- Sciamys group and a comparison of these forms with Pliocene and Recent octodontids (Patterson and Kraglievich, ms.) was the realization that the former could not be separated from the latter, either as a family or as a subfamily. Winge (1924, pp. 73, 78) would appear to have been right in his reference of "Aca- jvmys" and " Sciomys" to his Octodontini, Octodontes. A gap exists in our knowledge of the Octodontidae between the Santa- cruzian and the Huayquerian, but Sciamys on one side of this gap and the Pliocene forms on the other are sufficiently close to each other in dental and cranial characters as to permit no doubt on this score. Sciamys has rooted molars but, in contrast to those of the contemporary Acaremys, these were well on the way toward attaining the hypsodont condition. This trend was carried to completion, and cement added, during the hiatus in the history of the family. The later forms are nearly all hypsel- odont, but the Huayquerian to Hermosan PJitoramys does de- velop roots in old age, and the changes in crown pattern that it displays are very similar to those seen in Sciamys and, more fleetingly, in Acaremys. The postcranial skeleton of the Deseadan WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 295 Platijpittam.ys (Wood, 1949) is amazingly similar to that of Octodon itself (with which Wood did not have an opportunity to compare it), considering the vast time interval between the two. The chief differences lie in the ])roportions of the hind limb elements, and even these are not great. The tibia and fibula and the metatarsals are somewhat longer in the Recent form. The intersegmental indices of Octodon are: humerus/radius = 1.06 ; femur /tibia = 0.83 ; interseg*mental index, R + H X 100 = 67.3 ; T + F the corresponding figures for Platypittamys (Wood, 1949, pp. 42-43) are 1.04, 0.99 and 67.0. The tibia and fibula remain sep- arate in Octodon as in Platypittamys, as indeed they do in many caviomorphs. In old individuals in several families, the expanded proximal portion of the fibula fuses with the tibia, but, in all skeletons we have examined, the distal ends are separate. We interpret fusion of either end of these bones, then, as a progres- .sive character of the modern Caviomorpha (whatever it may have been in the Old World forms). We feel that the distribution of proximal fusion of these bones within the suborder very clearly demonstrates that the two bones were separate in the ancestral stock, a conclusion diametrically opposed to that reached by Landry (1957a, p. 19). It would appear that, apart from the hypselodont cheek teeth, Octodon and other nonfossorial octo- dontids are among the most generalized of living caviomorphs. It must also be noted (see below, p. 300) that the Acaremyinae of Ameghino, Scott and others (Wood's Acaremyidae) included a number of forms that are in reality members of the very closely related Echimyidae. We may also note in passing that the sup- posed Colhuehuapian octodontid "Eoctodon" is also an echimyid (see p. 302), and not an "acaremyine" as stated by Landry (1957a, pp. 39 and 56). On the basis of his study of Platypittamys, Wood concluded that the ''Acaremyidae" were ancestral, at least structurally, to all rodents here included in the Caviomorpha, with the possible exception of the Erethizontidae. This conclusion stands un- changed, except that the word "Acaremyidae" must be replaced liy "primitive Octodontidae." 296 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Platypittamys Wood Platj/pittamys Wood, 1949, p. 5. Platypittamys brachyodon Wood P. brachyodon Wood, 1949, pp. 6-43, Figs. 1-6, 8. P. brachydon (sic), Landry, 1957a, pp. 92, 93. Horizon and locality. Deseado formation; Scarritt Pocket, Rinconada de los Lopez, Chubut. This geiuis has been described in detail by Wood, and there is no need to repeat his account. This opportunity is taken to point out that he erroneously labeled Figure 3C as left instead of right and Figure 3D as right instead of left. The M'ork reported on here and studies of later octodontids (Patterson and Kraglievich ms.) necessitate some discussion of the previous interpretation of P^. The pattern of this tooth in Platypittamys is such that it could not have been derived from that of any described paramyid except by considerable reduction or degeneration. A recently discovered but still undescribed lower Gray Bull paramyid. however, has a premolar that could readily have given rise to that of Platypittamys : this suggests the possibility that there has been no secondary reduction in P^ of the Caviomorpha. By comparison with later members of the family, a terminology of the parts of the premolar has been adopted Avith the avowed intention of making the premolar terminology of the later octo- dontids agree with that of the molars, with the full understand- ing that the homologies suggested may be incorrect. This may or may not be phylogenetieally justifiable, but is certainly con- venient. On this basis, we now interpret the premolar of Platypittamys as having a lingual protocone, and a buccal paracone and meta- cone, widely separated, the former incorporated in the protoloph and the latter in the posteroloph. No trace of a separate metaloph is present. There is a well-developed anterior eingulum, ap- proaching the stage where it would be called an anteroloph. The Santacruzian Sciamys shows that the metaloph in octodontids was a later addition to the premolar crown that arose as a low, inconspicuous ridge on the anterior slope of the posteroloph. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 297 Availcible evidence strongly suggests that in all caviomorplis the metaloph is the last of the premolar crests to come into existence, being formed in some groups by a progressive division of the posteroloph that begins with the development of a minute fos- sette in this crest. Wood regarded the single posterior crest of P4 of Platypit- tamys as the hypolophid. Conditions in Sciamys tend to confirm this interpretation. In that genus, some examples of P4 occur in which the talonid is as simple as that of Platypittam.ys; others show various stages in the formation of a posterolophid by de- velojiment of a cingulum posterior to the internal half of the hypolophid. In one individual (M.A.C.N. no. A 1879), however, the entoconid is connected to the ectolophid by a crest anterior to, and obviously formed later than, that joining the entoconid to the h^vpoconid. Development of a notch posterior to the en- toconid in this specimen would have converted what is clearl}^ the equivalent of the hypolophid in other individuals into a posterolophid. Such differences are trivial but they do show that the two talonid crests of the caviomorph P4 could have arisen in different ways in various groups. Conditions in P4 of the Deseadan cavioid Asteromys in fact suggest that the hypolophid arose in a manner comparable to that of the anterior crest in the aberrant Sciamys just discussed. On the other hand, there still remains the possibility that there is no direct connection between the undescribed Gray Bull form and Plafypittamys, in view of the very considerable time gap. A gap of this size is certainly too large to permit phylogenetic conclusions to be drawn with any confidence. The possibility should not be lost sight of that P'* of Platypittamys is secondarily simplified. Such a simplification would not be entirely unexpected, and the possible sequence of events would be as outlined in the following paragraphs. Among various groups of rodents, there has quite obviouslj' been reduction of the premolars. In no known rodents are there more than two upper and one lower premolars. No one doubts that this is a reduction from the primitive placental formula. A premolar formula of f is retained by only a few rodents, in- cluding most of the typically Eocene Ischyromyoidea, and the Sciuridae. Many other rodents show varying traces of reduc- 298 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY tion, to PJ or PJ, , before -we reaeli the ericetid and miirid condition of PJJ . We explicitly accept this dental formula for these families, without any prejudice as to other families where dm;^ may have been retained and M^lost, which is surely not true of the Muroidea. The Caviomorplia retain one upper and one lower premolar. They must, therefore, have lost the third upper premolar, which is present in the paramyids. Obviously, there must have been a reduction of the premolars, in the Paleocene rodents, o-iving us a paramyid condition by the late Paleocene. There was then a halt in the reduction, until middle Eocene or later, when a number of g-roups continued the reduction of premolars, first losing P'', then P4, and finally P"*. It seems evident to us that this tendency to lose premolars must have been accompanied by a simplification of the pattern before the loss took place. According to this interpretation, the premolars of Plaiypit- farui/s (especially P^) would represent a simplification from the Paramys pattern, which was followed by a secondary redevelop- ment of complexity in this tooth in the later Caviomorpha. While this would be a ease of reversibility of an evolutionary trend, it nevertheless is still an entirely logical explanation of Avhat happened among the Caviomorpha. Platypittamys is extraordinarily primitive, for a caviomorph, in the small size of the infraorbital foramen. The two available specimens are both badly crushed but, despite this, the structure of the zygomasseteric region can be determined with reasonable assurance. The foramen can hardly have been appreciably larger than shown in AVood's figure (1949, Fig. 2). Landry (1957a, p. 9o), reports that he was unable to confirm Wood's account of the size of this structure, stating that "if this is true, Wood has here an exact intermediate in zygomasseteric structure l)etween the ischyromyids and one of the three advanced groups, something no one else has ever been able to find." An indepen- dent check of the specimens by both of us, however, has con- vinced us that Wood was correct in his interpretation. Perhaps some of Landry's difficulty in visualizing what has happened in the crushing of this part of the fossil may be explained by the fact that he has confused M^ and P^. As a result, a shifting of the palate to restore the break and to put M^ in its proper WOOD AND DAT'IKKSON : OLIGOCENE RODENTS OF PATAGONIA 299 place would move the lateral wall oi:" the infraorbital foramen mncli less than would have been the case if Landry had been eoi-rect and the tooth had been P^. The fact that there is no definitely marked area on the mandible for the insertion of M. masseter medialis, pars antieus, is in accord with the small size of the foramen, since it indicates that this division of the muscle (which is the one that passes through the foramen) was at best poorly developed. A recheck of the specimen again confirms Wood's statement that the masseter was apparently limited to the ventral surface of the zygoma, since there is no suggestion that any part of the muscle passes through the foramen. As Landry states (1957a, p. 93), the specimens are so crushed that it cannot be told whether the muscle had invaded the orbit. The contemporary Ccphalomiis, the only other Deseado rodent for which we have any knowledge of the skull, has an enormous in- fraorbital foramen, relatively larger than in its much later relative Neoreomys. It is very possible therefore that Platy- piitanuis was a persistently primitive, little-modified survivor, in this respect at least, of the basic immigrant stock. In the Colhuehuapian species of Acaremys there is a well-defined area of insertion on the mandible for M. masseter medialis, pars antieus, indicating that this part of the muscle, and hence pre- sumably the foramen through which it passed, was nearlj^ if not quite, as developed as in the Santacruzian species. Landry (1957a, pp. 93-94) questions whether it is likely that the en- largement of the foramen took place independent of the move- ment of the muscle. We agree that this independence of develop- ment may seem improbable, but the evidence seems to indicate that it took place. As stated by Wood (1949, p. 29), there is an entepicondylar foramen on the humerus, and the proximal ends of the til)ia and fibula are not fused (op. cit., p. 3(i), which makes it difficult to accept Landry's thesis that the primitive "hystricomorphs" were characterized by proximal fusion of the tibia and fibula and by loss of the entepicondylar foramen (Landry, 1957a, pp. 19, 20V The suggestion has been made that the Scarritt Pocket, in which Platypittujuys occurs, may be somewhat older than the other Deseadan local faunas of Patagonia, but this does not noAV 300 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY appear to be the case. The evidence strongly suggests that these faunas were all very nearly of the same age, since there is no indication of evolutionarj" change from one to another. Platypiitamys is the only known Deseadan octodontid. The family is actually a rare one in the Oligoeene and Miocene. Only the Santacruzian Sciamys is reasonably common, and there are but four known specimens from the earlier horizons. The rel- atively abundant Colhuehuapian Protacarcmys, as noted below, is not an octodontid, but an echimyid. A tentative approxima- tion of a phylogeny of the Miocene and earlier forms is shown in Figure 2. SANTACRUZIAN SCIAMYS ACAREMYS COLHUEHUAPIAN \^ ACAREMYS DESEADAN PLATYPITTAMYS Fig. 2. Tentative phylogeny of the Oligoeene and early Miocene Octodou- tidae. Family ECHIMYIDAE Miller and Gidley 1918 The earliest Echimyidae are very similar to the earliest Octo- dontidae in molar structure, so much so that several of them liave been i)laced in the latter family (in the guise of ''Acare- myinae") ])y Ameghino, Scott and others. The similarities are so close that we have no hesitation in referring them to the same superfamily. Landry (1957a, p. 56) separates the two families widely, placing the Echinwidae in his Erethizontoidea. We feel that this disagreement is to some degree more apparent than real, since he, in common with most other authors, refers the earlier octodontids, as "Acaremyinae," to the Erethizontidae, and between these early octodontids and the early echimyids there is a very close affinity. Landry includes Capromys and WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 301 its relatives as a subfamily of the Echimyidae and separates the nutrias (Myocastor) as a distinct family, which he places in the Octodontoidea. The eehimyids and capromyids seem to be detinitely related, and we believe that the nutrias also belong in the same major group, although we have no strong feelings as to the precise taxonomic rank that should be given them. Study of post-Deseadan members of the Echimyidae and Octodontidae (Patterson and Kraglievich, ms.) has revealed that the Colhue- huapian Protacareinys and '^ Acaremys'" prcminutus, and the Santaeruzian "Acarcniys" minutua and "A." mmutissimus and " Sciamys" tennissimus are all eehimyids. The molars of these forms may be distinguished from those of octodontids by the deeper anterior and posterior fossettes and fossettids, the some- what more oblique lophs and lophids, the somewhat deeper para- flexus and metaflexid, a marked tendency toward reduction and loss of the mctalophid (starting with M3), and by the somewhat greater length as compared to width of unworn crowns. The most obvious difference in the dentition lies, however, in the re- tention of dm^ and the suppression of P^ . This was first pointed out by Friant (1936a), who noted that in certain Recent and Pleistocene forms the anterior cheek tooth is invariably more worn than the molars and that there is no evidence that it was ever replaced. The abundant material examined by Patterson and Kraglievich shows that this is also true for all Tertiary forms from the Colhuehuapian on. From one Santaeruzian specimen, they were even able to extract the crown of a premolar, of the Acaronys tj^pe, from deep within the ramus, below^ the tirst lower cheek tooth. It is virtualh' certain that this tooth would never have erupted. Four Deseaclan specimens, representing two distinct species of a ncAv genus, Descadomys, are certainly referable to this family on the basis of molar structure. The anterior cheek tooth is preserved in two of them and in both it is clearly an octodontid- like premolar and not a milk molar. This suggests that retention of dm] in the Echimyidae came about between Deseadan and Colhuehuapian time. The phjdogem' of the Echimyidae was evidently very complex, and the available material is insufficient to trace it in any great detail. A tentative phylogeny, however, is given in Figure 3, 302 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY adapted from Patterson and Kragiievich (ms.). Deseadoniys seems to be ancestral to the comparatively specialized Adelphomys group, composed of the Santacrnzian Adelphomys and Stichomys and an undescribed Colhuehuapian genus, but not including Pro- tadelphomys. The molars of Deseadomys are actually higher crowned than those of the Colhuehuapian Protacaremys and Prospaniomys. These last are sufficiently similar to suggest that their common ancestors may have lived in Deseadan time. The primary radiation of the Echimyidae was evidently in progress PROECHIMYS- DACTYLOMYS ECHIMYS SPANIOMYS ADELPHOMrS EUMYSOPS GR GROUP GROUP GROUP GROUP SANTACRU7IAN GEN.NOV.^ COLHUEHUAPIAN PROTADEL- PHOMYS DESEADAN PROTACAR-- EMYS SPAN!OMYS STICH- ADEL- ■? t OMYS PHOMYS /' DESEADOMYS OGTODONTIDAE Fig. 3. Tentative phylogeny of the Oligocene and early Miocene Echi- myidae. during the Deseadan and presumably had begun not long prior to it. Derivation from the primitive octodontid stem is clearly indicated. The Colhuehuapian Eoctodon Ameghino is neither valid nor an octodontid. The type species, E. securiclatus, is a synonym of Protacaremys prior and the larger E. crassiuscidus is a synonym of Prospaniomys priscus. Both these forms are echimyids. The Santacrnzian Spaniomys is highly variable, and Gyrignophus complicatus and Graphimys provectus appear to have been based on extreme variants of 8. modestus.^ 5 J. L. Kragiievich (19.")7. p. .37) believes the "Eumi/gops group" (cf. Fig. :'.) to be part of the subfamily Heteropsoniyinae Antbony. The Pliocene genera /'ronthvnira and Prongiiti of Ameghino cannot be separated from JUivn/KOi)-^. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 303 DeSEADOIMYS gen. nov. Asteromys Loomis, 1914, pp. 194-195 (in part), nee Ameghino 1897c; Wood, 1949, pp. 4, 15, 16, 18-20; Schaub in Stehlin and Schaub, 1951, p. 60. Type species. D. ayainhoiirgl sp. nov. Distribution. Deseadan; Patagonia. Diagnosis. Differing from all other known ecliimyids in re- placement of dm^ by P] ; cheek teeth mesodont, lower crowned than in Adelphomys and Stichomys, higher crowned than in Prot- acaremys and Prospaniomys, with nnilateral hypsodonty in uppers but not in lowers; P^ (known only in type species) not molariform, without nietaloph, protoloph concave anteriorly; M^ (known only in type species) more circular in outline and with hypocone less internal than in Adelpliomys and Stichomys; P4 (known only in type species) not molariform, without anter- olophid, metaconid separated from protocouid by narrow cleft, metalophid rudimentary; lower molars more advanced in meta- lophid reduction than in Protacaremys and Prospaniomys. A second species, D. loomisi sp. nov., somewhat more advanced than the type, is referred to the genus. The two species appear to have beeu trending toward the two Santacruzian genera, the type toward Stichomys and D. loomisi toward Adelphomys. Deseadomys arambourgi" sp. nov. Figs. 4-6 ^t,>- Asteromys pro.^ pic tins Loomis, 1914, p. 195 (in part). Fig. 128 (nee Ame- ghino 1897c) ; Schaub in Stehlin and Schaub, 1951, p. GO, Fig. 72. Type. M.N. H.N. no. 1903-3-1, right mandible with I, P4-M0. Hypodigm. Type and A.C.M. nos. 3163, right P-^-M^^ (de- scribed and figured by Loomis as Asteromys prospicnus)'' and 3071, left M-\ 6 Named for Professor Camille Animliour:,', as a slight return for his kinilni'ss in entrusting to us the Doseado rodents in his charge. 7 These three teeth are isolated witli no trace of conuectiug hone. Tl:cy were. however, associated in the field (Loomis, personal communication to Wood, Fel>ruarv, 10:^.")) and presumably represent a single individual. Loomis' liguic shows four teeth. The one figured liy him as M- has a wear facet on liotli the front and tlie rear, proving it to be one of the central molars. On the basis of wear, it is interpreted as M2 and Loomis" M- as Mi. The tooth figured by him as Mi is not in the Amherst collections, and was api>arently lost sometime Ix-tween Loomis' description of the material in ]!)14 and Wood's visit to Amherst in lit.'!.). As the specimen was mounted, there was a space between P* and Loomis' M- (our Ml), where a tooth obviously had been, and whence it had been lost. 304 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Horizon and localities. Deseado formation; La Fleclia, Santa Cruz (type), Cabeza Blanca, Chubut (Amherst specimens). Diagnosis. JMetalophid absent on M3, interrupted on M2, pres- ent on Ml ; teeth somewhat lower crowned than in B. loomisi. Loomis' reason for referring A.C.M. no. 3163 to Asteromys, whose upper dentition is not certainly known, presumably lay in his identification of an isolated cheek tooth (A.C.M. no. 3054). shown in his Figaire 129, as M2 of A. prospicuus. This species, as pointed out below, is actually a synonym of Cephalomys plexus; A.C.M. no. 3054 is not referable to it. The specimen appears to be an upper cheek tooth of an eocardiid, perhaps, although not surely, of Asteromys punctus, which is without question a mem- ber of this family (see below, p. 376). Since A.C.M. no. 3054 does show a certain resemblance to the lower molars of A. punctus, which were figured by Ameghino, Loomis' generic identification of it was as good as he could have made. As regards the specific identification, he should have noted that his tooth was much larger than Ameghino 's syntype specimens of "A. prospicuus." The small, ephemeral mesofossette of A.C.M. no. 3054 definitely excludes the specimen from the Echimyidae and Octodontidae. The reference of the upper cheek teeth, A.C.^I. no. 3163, to the same species as no. 3054 was presumably made chiefly on the basis of size, general similarity, and occurrence in the same deposit. It was, however, definitely erroneous. Having made it, Loomis went on to observe the close general resemblance between the cheek teeth of A.C.M. no. 3163 and those of " Acaremys" mi7iutus and minutissimus^ figured by Scott, and to conclude that "Asteromys appears to be the direct ancestor of Acaremys." Unfortunateh', as already noted, "A." minutus and minutis- simus are not octodontids ("acaremyines"), as Ameghino and Scott supposed, but echimyids. As a result of this cumulative series of errors, Asteromys has been, since 1914, listed as a member of the ' ' Acaremyidae " or Erethizontidae, "Acare- myinae. ' ' A.C.M. no. 3163 is certainly an echimyid, possessing every one of the molar characters mentioned above that separate the early members of this family from the early Octodontidae. Our s Loomis did luit specifically mention these species, Init they are the only ones anionfr those referred to Acareiiiys by Scott for which crown details are shown, and it is to Scott's publication that he would obviously have turned. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 305 reasons for referring the specimen to Descadomys aramtourgi are as follows: the teeth are higher-crowned than those of the Colhuehnapian Protacarcmys and Prospaniomys, and in this feature and in crown structure resemble the Adelphomys group ; the lower molars of Descadomys definitely resemble those of the Adelphomys group in both lieight and structure; within this group A. CM. no. 3163 is closer to Stichomys than to Adelphomys in molar structure; the same is true of the lower molars of D. aramhowrgi; A. CM. no. 3163 and the type of D. aramhourgi agree perfectly in size. The two specimens are from different localities — Cabeza Blanca and La Flecha, respectively — but this is no obstacle to regarding them as conspecific, since many species, among them rodents, are common to both. A B Fig. 4. Deseadomys aramhourgi gen. et sp. nov. x 10. A, EP^-M-, A.C.M. uo. 3163; B, L]\I3, A.C.M. no. 3071. P^ is considerably smaller, simpler and more oval in outline than the molars ( Fig. 4 A ) . The anteroloph runs outward from the large protocone for about two-thirds of the way across the anterior face of the tooth, falling well short of the paracone. The protoloph is slender and somewhat concave anteriorly. As in Plat ypitt amy s, there is no trace of a metaloph. The poste- roloph forms the posterior third of the tooth. This crest differs from that of Platypittamys in that there is a definite enlargement at the postero-internal corner, so that we may consider that there is a hypocone as well as a metacone in the posteroloph. The para- flexus is rather shallow, the mesoflexus deep. The protocone and 306 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY hypocone are only very slightly separated, there being a faint g-roove on the lingual face indicating incipient separation. It is too faint to be called a h3q:)oflexus. Near the middle of the tooth, there are crests extending into the mesoflexus from both the protoloph and the posteroloph, which partially dam the meso- fiexiis (and would split it in half at an advanced stage of wear). These crests seem to suggest the initial stages of a mure, as in the molars, but it seems clear that the molar jiattern was not developed in this manner, but rather by a deepening of the hypo- flexus until the connection between the protocone and hypo- cone came to lie near the center of the tooth. The tooth is def- initely more advanced than P^ of Platypittamys, but the ad- vances were of no significance for the later history of the family, the premolars being destined to disappear from the functional dentition. M^"- (Fig. 4A) have the usual four transverse crests which are longer and more lophate than in PlatypiUamys. The ant- eroloph is better developed than in P^^, but does not extend as far externally as does the protoloph. The latter crest is thin and inclined anteriorly, markedly so on RP. It is united with the anterior end of the mure rather than directly with the protocone, to which it is connected by a thin isthmus. The metaloph and posteroloph unite externally at an early stage of wear to enclose a metafossette that is worn away before the mesoflexus is isolated to form a fossette. The paraflexus is more widely open than in early octodontids, but would be isolated to form a lake while the deep lingual end of the metafossette was still present. The meso- flexus is widely open, but would ultimately be converted to a mesofossette. The hypoflexus is of nearly uniform depth through- out so that it might never be converted to a fossette. The proto- cone is elongate anteroposteriorly. These teeth closely resemble those of Sticliomys and Adclphomys, and agree with those of SticJiomys in all characters in which they differ from those of Adelphomys. These differences are : anterolopli not extending as far externally as the protoloph; paraflexus notably shallower than mesoflexus; metafossette relatively large and persistent; hypoflexus deep and persistent. The hy])oflexus does not extend as far externally into the base of the protoloph in all molars of Dcseadomys as it does in the Santacruzian forms. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 307 iVP (Fig-. 4B) is slightly smaller and has a rounder outline than its predecessors. The metaloph and posteroloph are just in l)roeess of uniting:, so that decision as to whether there is a meta- Hexus or a metafossette is purely a matter of terminology. There was clearly a metaflexus at an earlier stage of wear. The hypo- cone is quite external in position, more so than in Platypittamys, and the posterior portion of the tooth is notably smaller than the anterior. Adclphoinys has a smaller, less elongate M^ than Stichoni!/^, a point of resemblance to Deseadomys and presum- ably a primitive feature. SticJioniys resembles the Deseadan form in nearly all details of crown structure, including the presence of a metaflexus. B Fig'. 5. Deseadomy.s- aramhourgi gen. et sp. iiov. x 10. Type, M.N. H.N. no. 1903-3-1. A, EP4-M3; B, cross-section of KIi, posterior view. The lower premolar {Fig. 5A) is very similar to that of Platy- pittamys (Wood, 1949, Fig. 3 C, D), but is slightly less molar- iform. The protoconid is connected with the center of the hypo- lophid by a nearly straight, slightly obliquely directed, and nearly centrally located ectolophid, as in Platypittamys. The hypolophid forms the posterior margin of the crown, uniting the prominent anteroposteriorly compressed hypoconid with the very large and more rounded entoconid. As in Platypittamys, there is 308 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY no suggestion that more than one crest is present at the rear of the tooth. Anteriorly, the protoconid sends a blunt spur linguad — the rudiment of the metalophid. The metaconid is a stout, four-sided cusp, separated from the protoconid by a narrow, fairly deep mesoflexid. With extreme wear, the central part of this would form a mesofossettid, as it does in a much earlier stage of wear in Plat yiJitt amy s. The posterior corner of the meta- conid is formed by a very short, blunt spur that projects in the direction of the tip of the metalophid rudiment. There is no anterolophid, nor any indication of a mesoconid or mesolophid. External and internal valleys are long and of nearly equal length. The lower molars are longer relative to width than those of Platypitiamys and are considerably more specialized in the re- duction and loss of the metalophid, which is progressive from Ml to M.s (Fig. 5A). On Mi, this crest is short, inclined more anteriorly than in Platypittamys and joined to the internal rather than to the posterior face of the metaconid. On M2, it is interrupted, being represented only by a short, blunt spur from the protoconid, which would unite with the metaconid only after extreme wear. On M3, the metalophid is extremely reduced, and barely present at all. It would never reach the metaconid. A tendency toward reduction and loss of the metalophid occurs sporadically in caviomorphs. It is characteristic of echimyids, occurs in the later members of the Luantinae among eoeardiids, in the Caproynys group of capromyids, and may be seen in a few individuals of the octodontid Sciamys. This crest is the weakest of the four on the lower molars (as Wood noted [1949, p. 21], the anterolophid provides the main connection between protoconid and metaconid), and this no doubt accounts for the tendency toward loss. Curiously enough, there is a compensatory tendency in the Proechimys-Eumysops group of echimyids for a neolophid to arise in place of the metalophid subsequent to its disappearance. The deep hypoflexid is much less oblique than in Platypittamys — a point of resemblance to Stichomys and Adelphomys. The internal flexids are wider, due in part to the reduction of the metalophid and in part to the greater length of the teeth relative to width. The mesoflexid is vertically the deepest of the valleys, being slightl}' deeper than the hypoflexid, and is of essentially WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 309 even depth throughout its course. The narrower metaflexid is shallower but deepens sharply Iniccad, so that a metafossettid would l)e isolated by further wear. The posterolophid is made up of the hypoconid and the posterior cingulum. There does not seem to be any swelling of this crest that could reasonably be considered to he the entoconid. The hypolophid therefore must contain the entoconid at the lingual margin of the tooth. In the upper Eocene paramyid Rapamys (Wood, 1950, Fig. 2B), there is a hyi^olophid extending from the entoconid to the anterior corner of the hypoconid at the point where the eetolophid arises. Apparently, with the gradual reduction of the metalophid, the entoconid and particularly the buccal end of the hypolophid have swung forward, giving the diagonal direction to this crest that characterized Dcscadomj/s, where the hypolophid joins the eetolo- phid rather than the hypoconid. AVith the reduction of the metalophid, there is a corresponding increase in the width of the flexids. As in Stichomys and AdelpJiomj/s, there is no ten- dency toward unilateral hypsodonty. The depths of the hypo- flexid and mesoflexid and the shallowness of the metaflexid relative to the mesoflexid are points of resemlilance to SticJiomys. In this genus, there is no trace of a metalophid on M2-3, but in approximately a third of the specimens in the Ameghino Col- lection this crest is present, although very small, on Mi, connect- ing the protoconid with the junction of the anterolophid and metaconid. The incisor (Figs. 5B, 6) is moderately stout and extends back to a point beneath M^. The anterior face is nearly flat, but has a very faint longitudinal sulcus down the center, which is prob- ably an individual peculiarity, since such faint sulci are very variable in many groups of rodents, including both the Para- myidae and the Santacruzian Echimyidae. The enamel extends for over a third of the way around on the outer side and is barely reflected over to the inner. In cross-section, the pulp cavity is elongate. The incisor of Stichomys has a much more curved anterior face. The mandible (Fig. 6) is robust, deep through the masseteric fossa and symphysis, and the symphyseal region is elongate, ex- tending back to a point beneath P4. The genioglossal pit is poorly marked. The masseteric fossa begins beneath the posterior 310 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY ])art of Ml and the ventral margin is markedly everted, so that the fossa is deep ; there is almost no dorsal bonndary. The spee- imen is broken in the region of the angle and it is therefore impossible to say whether or not tliis was inturned, althongh it presumably was. The coronoid arises near the front of Mg. The mental foramen is about a third to a half of the way down the side of the mandible, al)ont the middle of the diastema. The chin is slightly indicated. The symphysis shoAvs very minor irregularities, and it seems probable that some movement be- tween the two halves of the jaw was possible and that a small .Al. trausversus mandibnlae was present. In nearly every respect, tlie jaw closely resembles those of SticJiomys and Adelphomys. Fig. 6. Dcseadomys aramhourgi gen. et sp. iiov. x 3. Type, M.N. II. N. no. 190,3-3-1, right ramus, lateral view. Apart from slight differences in the general shape of the in- cisor and in the contour of its anterior surface, D. aramionrgi resembles Stirhoniys very closely. Since development and loss of shallow sulci and changes in the shape of the incisor may occur frequently in rodents, it is even possible that this species may be in the direct ancestry of the Santacruzian genus. For measurements, see Table 1. Deseadomys loomisi^ sp. nov. Fig. 7 Type. A.CM. no. 3087, right mandible with T, Mi-o, alveoli of P4, Mo. Hypodigm. Type only. y Named for the latp Professor Pretlerick B. Looniis. who organized and led tlic Amherst Patagonian Expedition of 1911-12. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 311 Horizon and locality. Deseado formation; Cabeza Blaiica, Chubut. Diagnosis. Metalophid entirely absent; teeth slightly larger and molars somewhat hig'her-erowned than in D. aromhowgi ; mesoflexid deeper than in D. aramhourgi; metafiexid fairly shal- low, so that lingual ends of hypolophid and posterolophid ap- proach more closely than in D. aramhourgi. The cheek teeth (Fig. 7 A) are medinm-crowned, but, as in Adclphomys and Stichoniys, show no indications of unilateral hypsodonty. As in the type species, Mo is the largest tooth. M^ (as determined from the alveolus) was the smallest of the series. P4 appears to have been about the size of Mi, which is an ad- vanced character. There is no trace in either molar of any rem- B Fig^. 7. Deseadomys loomisi sp. nov. x 10. Type, A. CM. no. 3087. A, RM1-2; B, cross-section of EJi, posterior view. uants of the metalophid, so that the teeth are entirely three-lobed on the lingual side. The mesoflexid and hypoflexid are about equally deep, and much deeper than the metafiexid. With wear, tlie metafossettid would persist until after the formation of the mesofossettid and hypofossettid, but would be worn away sliortly thereafter. The hypofossettid is slightly deeper than the mesofossettid. The anterior face of the lower incisor is straight for the median two-thirds and gently curved on the buccal third (Fig. 7B). There is no suggestion of a sulcus. The enamel extends nearly halfway around the buccal side of the tooth, but only a short distance onto the lingual face. The two sides of the tooth are nearly parallel, as in D. aramhourgi. The pulp cavity is long 312 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY and narrow, but its details could not be determined. The incisor extends back to beneath Mo, thus not as far as in B. aramhourgi. The lower incisor in Adelphomys is relatively larger, wider, and has a plane anterior surface. The mandible differs from that of the type species in that the anterior end of the masseteric crest is beneath P4 instead of Mi. This is a primitive character encountered in octodontids, Prota- caremijs, Frospaniomys, etc. The sjanphyseal surface is essen- tially plane. As far as the available evidence goes, there is nothing to oppose the view that D. loomisi was in or near the ancestry of Adel- phomys. The tooth measurements are given in Table 1. Table 1 Tooth measurements (in mm.) oiBcseadoynys Upper Teeth P-*, anteroposterior width protolopli width posteroloph ^ 1, anteroposterior width protoloph width metaloph ]\I-, anteroposterior width protoloph width metaloph D. aramhourf/i A. CM. no. 3163, right 2.09 2.03* 2.12* 2.25 2.25* 2.04* 2.42 2.61* 2.41* A.C.M. no. 3071, left Lower Teeth D. aramhourgi M.N. H.N. no. 1903-3-1, right Incisor anteroposterior 1.76 transverse 1.31 P4 anteroposterior 2.12 anterior width 1.68* width hypolophid 1.90* Ml anteroposterior 2.41 width metalophid 1.92* width hypolophid 1.96* M2 anteroposterior 2.52 Avidth metalophid 2.05* D. loomisi A.C.M. no. SOSl right ca. M--. anteroposterior 2.14 width hypolophid 2.18* 2.35* Ms anteroposterior 2.32 1.88* width metalophid 1.90* width hypolophid 1.89* * = Maximum diameter of tooth. anteroposterior width protoloph width metaloph 1.9 1.29 2.63 2.10* 1.92* 2.64 2.47* 2.29* Superfamily CHINCHILLOIDEA L. Kraglievich 1940 As discussed below under Dasyproctidae (p. 327), we now sus- pect that a number of caviomorph families should be included in this superfamily. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 313 Family CHINCHILLIDAE Bennett 1833 ScoTAMYS Loomis 1914 Scotamys Loomis, 1914, p. 192; Stehlin and Schaub, 1951, p. 260. Type species. Scotamys antiquiis Loomis 1914. Distribution. Deseadan; Patagonia. Emended diagnosis. Teeth very high-crowned, apparently rootless and ever-growing ; crown pattern rapidly lost ; hypo- flexus and hypoflexid very deep; small third lobe on unworn upper molars ; enamel missing from anterior part of lower molars and posterior part of uppers after little wear; cement present; incisors proportionately small. Scotamys antiquus Loomis 1914 Figs. 8-10 Siioiamys antiquus Loomis, 1914, p. 192-193, Figs. 125-126; Stehlin and Sehanb, 1951, Fig. 434, p. 260. Type. A.C.M. no. 3063, a left lower jaw with I, P4-M2, from Cabeza Blanca, Chubut. Hypodigm. Type and a series of isolated teeth in collection of Museum National d'Histoire Naturelle (3 P-^, 15 M'"'~, 3 M^, 3 P4, 9 Ml or 2 > 1 ^^3 and 11 incisors), all from La Flecha, Santa Cruz. Diagnosis. As for the genus ; tooth measurements as shown in Table 2. It is often very difficult to determine with certainty the posi- tion in the tooth row of the isolated teeth. The patterns of P*, M^ and P4 are sufficiently distinctive so that these teeth can be readily identified. Mo can, we believe, be identified by the narrow talonid. M^"- cannot be separated from each other, as is also the case with Mi -2. The upper molars have a greater curvature of their vertical axes, both anteroposteriorly and transversely, than do the lowers, and the two have been separated on this basis. The interruption of the enamel in upper molars is generally the mirror image of that in the lowers. No unworn cheek teeth are available. The general resemblance between Scotamys and the better known CepJialomys is suffic- iently close, however, so that it seems justifiable to interpret such structure as can be seen in the former in terms of the latter. 314 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY P* (Fig. 8A) has a very large combined aiiterolopli and protocone that, togetlier, forms all of the anterior and much of the external and internal faces of the tooth. This curving crest is separated from the paracone and its crest by a cleft, the hypo- flexns, that extends nearly to the buccal margin of the tooth, though, in contrast with the situation in Ceplialomys, it is open at the lingual end. This lingual opening does not extend far down the tooth, however, so that, ^\^th wear, the valley becomes a fossette, as in C( phalnmys. The paracone is central in position and continues into a short crest, the protoloph, that describes a D E Fig. 8. Scofami/s antiquus Loomis x 5. A, LP^, M.N.H.N. no. 1903-3-8; B, LMi<"^-, M.N.H.N. no. 1903-3-9; C, Llf i " 2, M.N.H.N. no. 1903-3-10; B, LMioi--, M.N.H.N. no. 1903-3-11; E, EM3, M.N.H.N. no. 1903-3-12; F, LM3, M.N.H.N. no. 1903-3-13. C, T) and F sliow crown outlines at wear surface. curve similar to that of the much larger anteroloph. The meta- cone is posteroexternal and united by wear to the paracone. A narrow crest, the posteroloph, runs internally from the metacone, forming the posterior margin of the tooth and joining the pro- toloph posterointernally. A small, shallow mesofossette is isolated between protoloph and posteroloph. Metacone, mesofossette and posteroloph are situated on a posteriorly overhanging portion of the crown, so tliat all trace of these structures disappears after WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 315 a little wear, leaving a bilobed crown that then persists until after the closure of the hypoflexus. There is no indication that a metaloph was ever present. The enamel is interrupted early in life at the postero-buecal corner of the tooth, and then along the rest of the buccal side. The least worn M^"'- in the series (M.N.H.N. no. 1903-3-9, Pig. 8B) shows a pattern consisting of two lobes separated by a narrow hypoflexus that extends about three-quarters of the way from the lingual to the 1)uccal margin. Buccad of the hypoflexus is a small fossette and in the buccal portion of the posterior lobe are two more. All three are shallow. The anterior is the smallest, the middle the deepest, and the posterior the largest. From these remnants, the unworn crown pattern can readily be recon- structed. It was clearly very similar to that of the unworn CepJwIomys molar without the fifth crest, consisting of a large anteroloph, a stout protoloph, a metaloph that Avas more trans- verse than in Cephalonujs, a posteroloph and a hypoflexus that undermined the paraflexus (i.e., the paraflexus extended more linguad on the unworn croA\Ti than was the case farther down in the tooth, so that, as wear proceeded, the paraflexus diminished in length and the hypoflexus increased). Thus interpreted, the three fossettes in M.N.H.N. no. 1903-3-9 are the parafossette, mesofossette and metafossette. A basic similarity to the Platy- pittamys-Deseadomys pattern is evident. The posteroloph pro- trudes beyond the main body of the tooth and is soon eliminated by wear. In M.N.H.N. no. 1903-3-9, the enamel is complete all the way around the croAvn, although it would be interrupted at the anterior and posterior buccal corners after very little addi- tional wear. In the next least-worn tooth (M.N.H.N. no. 1903- 3-10, Fig. 8 C), which only shows a single fossette (the mesofos- sette) in the posterior lobe, the enamel is interrupted antero-buc- cally and posteriorly, as is the case with most of the other teeth. The pattern shown in Figure 8D (M.N.H.N. no. 1903-3-11) is quicldy attained, and persists indefinitely as far as can be told from the available material. Of the three specimens believed to be M'^ two are complete and one is broken. All three, however, show a characteristic three- lobed pattern, the posterior lobe being smaller than in Perimys. One tooth (M.N.H.N. no. 1903-3-12, Fig. 8E) is relatively little 316 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY worn. In this specimen, the enamel is interrupted at the antero- buccal corner and the extreme posterior end of the tooth. The two anterior lobes are sharply angulate, and the third lobe is quite small. A small mesofossette is present. In a more worn specimen (M.N.II.N. no. 1903-3-13, Fig. 8F), the mesofossette is lost, the second lobe is larger, and the angulation of the tooth is reduced. The enamel is interrupted along the whole buccal surface of the third lobe. As far as the base of the prism, there is no further reduction of the enamel. To judge from the struc- ture of the anterior molars, the posterior lobe is probably formed from the posterior cingulum. The differences between the teeth shown in Figures 8B and 8F are entirely due to wear. B D Fig. 9. Scotamys antiquus Loomis x 5. A, LP4-M0, type, A.C.M. no. 3063 ; B, LP4, M.N.H.N. no. 1903-3-4; C, LP4, M.N.H.N. no. 1903-3-5; D, RP4, M.N.H.N. no. 1903-3-6; E, RMs (incomplete anteriorly), M.N.H.N. no. 1903-3-7. B \o E show crown outlines at wear surface. The central valley of all lower cheek teeth curves antero- linguad. At least in the stage of wear represented by most of the lower molars, each lobe is simple, all fossettids having been worn away. In unbroken lower molars, the prism may be over 7 mm. high, with no change in tooth pattern in this distance. The least worn P4 is that of the type. This tooth is bilobed, but there is a small antero-buccal valley in the anterior lobe and a minute fossettid, evidently a metafossettid, postero-linguad of the head of the central buccal valley (Fig. 9A). The metafos- WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 317 settid, liowever, has almost been eliminated by wear, and the anterior valley would be converted to a fossettid by a very slight additional amount of wear. This valley is directed nearly for- ward, partly because the protoconid has shifted forward at its buccal margin, and partly because the anterior cingulum is short at the lingual margin of the tooth. The anterior valley is filled with cement, as is the central part of the hypoflexid. This anterior valley is probably that between protoconid and antero- conid. None of the three Museum National d'Histoire Naturelle si^ecimens possesses an anterior valley, even as a fossettid. They apparently show a progressive transverse narrowing of the tooth with wear (Figs. 9B-D). On M.N.H.N. no. 1903-3-6, the hypo- flexid ends about half way down the crown. On the lower half of the tooth, another valley appears buccally, as well as one on the lingual side, so that the pulp end of this tooth shows a figure-eight outline. This is the most worn lower premolar avail- able. In the area with the figure-eight outline, the enamel was not laid down on the anterior face or on most of the lingual face, being present on the latter side only in the lingual valley. In addition to those of the type, there are nine first or second lower molars in tlie M.N.H.N. collection. These all agree in pattern with those of the type (Fig. 9A), showing the enamel absent along the entire anterior face of the tooth and at the posterointernal corner. The central part of the hypoflexid is filled with cement, which extends part way along the sides of the flexid. Cement is also present on the posterolateral corners of the teeth. There is little or no change in pattern in the molars once the stage sho\vn in Figure 9A has been reached. Although no unworn lower molars are preserved, there is no reason to sup- ])0se that the pattern of the lower molars was not a mirror image of that of the uppers. M;5 (M.N.H.N. no. 1903-3-7, Fig. 9E) is similar to the anterior molars except that the talonid is appreciably narrower than the trigonid, at least at the wear surface. The difference is much less marked after further wear. The tooth is the smallest of the lower cheek teeth. The crowns, particularly of the upper teeth, change in outline as wear proceeds. The internal cleft in P^ disappears with wear, which gives the tooth an old Ccphalomys-Neoreomys aspect, though arrived at in a different manner. The two sides of this 318 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY cleft approacli each other rather closely in old specimens of Perimys (e.g. A.M.N.H. no. 29673, Colhue-Iiuapi). Upper and lower molars of Scotamys are more ragged looking than in Peri- mys; they lack the neat, regimented appearance of the teeth of that form. Three fragments of upper incisors are present in the Museum National material. These are essentially identical in pattern with the lower incisors, but the anterior face is slightly more rounded and the pulp cavity is more nearly a straight slit than a dumb-bell shaped cavity (Fig. lOA). The lower incisor extends back to a point below M2 . It does not run beneath the cheek teeth, but rather medial of them, due to their excessive hypsodonty. The anterior face is slightly curved, and the enamel rather thick. The lateral face is markedly rounded. The enamel extends only a short distance onto the lateral side, and nearly as far on the median side (Fig. lOB) . The pulp cavity is quite unusual, having a dumb-bell shape near the tip of the incisor, though it comes closer to the shape of the incisor itself several millimeters nearer the pulp end. Both upper and lower incisors are stocky but very small in comparison to the size of the cheek teeth, a combination also true of Cephalomys. On this basis, Scotamys appears to have been a very poor gnawer. Coupled with the powerful masseter muscle and the hypselodont cheek teeth, this suggests an approach toward a grazing adaptation. The mandible (Figs. IOC, D) is very massive and thick, with a broad, rather flat shelf on the ventral surface. The large mental foramen is in the ventral half of the jaw beneath the anterior end of P4 ; tAvo minute foramina are present above it. The chin is heavy and the symphyseal area large ; the ventral margin forms a straight line terminating posteriorly in a strong mental process. The anterodorsal extemity of the symphysis extends upward above a line extended forward from the outer alveolar border of the cheek teeth. The pit for the genioglossal is pronounced. The masseteric crest is very strong, slopes gently downward and extends laterally for a considerable distance. On the dorsal side of the crest, within the masseteric fossa, there is a series of interrupted grooves. The cheek teeth are inclined laterally, as in all chinchillids and their open pulps are external WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 319 to the incisor. The latter extended at least as far back as the end of the molar series, and the snrrounding bone forms a shelf on the inner face of the ramus (Fig. lOD). Externally, the alveolar border of M;5 is broken away ; the ascending ramus may have begun to arise from this region, as in other members of the family. For a member of the Chinchillidae, this is a primitive jaw. In the living forms the masseteric crest is reduced to the B C D Fig. 10. Scotainys antiquus Loomis. Cross-sections of incisors x 5. A, right upper, anterior view, M.N. H.N. no. 1903-3-14; B, left lower of type, anterior view, A.C.M. no. 3063. Left mandible of type, A. CM. no. 3063 X 2 ; C, lateral view ; D, ventral view. vanishing point; in Pcrimys this is much less salient than in Scolamtjs. The anterodorsal extremity of the symphysis extends as far upward in Periniys as in Scotcnnys, and nearly as much so in Chinchilla and Lagidiuni, l)ut is low relative to the alveolar l)order of the cheek teeth in Lagostomns. The incisor extends back to M;; in Lagostomus, to Mo in Chinchilla and Lagidium and 320 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY to varying distances in Perimys. The mental foramen is in the same position in Perimys as in Scotamys; mnch higher on the mandible in the living forms. As regards hypselodonty, Scotamys is the most specialized Deseaclan rodent known. Loomis (1914) placed this genus in the Chinchillidae and considered it to be ancestral to Scotaeumys o£ the Santa Cruz.'" The upper molars of Scotaeumys (Scott, 1905, PI. G8, fig. 22) could have been derived from those of Scotamys, but this would have involved a reversal in the direction in which the evolution of Scotamys had evidently been proceeding. Scott states that in P4 of Scotaeumys (the only lower tooth known) there is an anterior third lamina. However, he indicates a dis' tinct change in pattern of the upper molars with wear. From this, it would appear that Scotamys is appreciably more advanced in its hypsodonty than is Scotaeumys. Scotamys is therefore probably too specialized in its own direction to be ancestral to Scotaeumys. On the other hand, Perimys could have descended from Scotamys if the posterior cingulum of M^ - w^ere eliminated and the incisor shortened. P'* of Perimys could be derived from that of Scotamys by an emphasis of the separation between the paracone and the anteroloph. This latter would result in the reversal of the drainage of the valley, and the formation of a buccal, rather than a lingual, cleft in partly worn teeth. In P4, elimination of the anterior cleft, not a persistent feature in Scotamys, w^ould give a Perimys-like pattern. Scotamys, then, could be a collateral ancestor of Scotaeumys of the Santa Cruz, and perhaps, but probabl}^ not, an actual ancestor of Perimys oP the Colhue-Huapi and Santa Cruz. Prolagostomus and Pliola- gostomus are at present of unknown ancestry, although descent from Scotamys, or from a form generally similar to it, is not impossible. The relationships, then, might be as shown in Figure 11. Landry (1957a, p. 59) refers Scotamys to the "Cephalomyi- dae, " although from his text {op. cit., p. 54) it is obvious that lie feels there are suggestions of relationship to Perimys. We do not differ greatly from him on this point, Init we feel that the 10 Landry's areoiint nf the taxonomic history of Scotamys (1957a, p. Ti4) is sonu'what confused. He states that it was considered to be a cephalomyid by Anieghino and Loomis. However, Aniegliino had died before the genus was described, and l/ooniis, as stated, referred it to the Cliinchillidae. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 321 similarities to Perimys are indicative of phylogenetic relation- ships, and that the similarities to Cephalomys are indicative of the close approach of the two distinct lines in the Deseadan. Landr3% combatting an early idea of Ameghiuo's that Perimys was directly ancestral to Lagostomits, has also separated some of the earlier ehinchillids as a family, the Perimyidae, on the following grounds: cheek teeth with less appressed laminae, and with folds opening lingually on upper molars and labially on upper premolars ; ventral maxillary zygomatic root not anterior to dorsal maxillary zygomatic root ; masseteric ridge on mandible strongly developed {op. cit., p. 59). As regards these characters, we may observe: 1), that Scotamys, which on other grounds SANTACRUZIAN SCOTAEUMYS PERIMYS PROLAGOSTOMUS PLIOLAGOSTOMUS COLHUEHUAPIAN X PERIMYS ^ ^ -^ DESEADAN DASYPROCTIDAE \ \ SCOJAMYS rrODONTIDAE Fig. 11. Tentative phylogeny of the Oligocene and early Miocene Chin- chillidae. would be a " perimyid, ' ' has the fold in the upper premolar open- ing in the same direction as in the molars; and 2), that the ventral maxillary zygomatic root is just as far anterior to the dorsal root (the upper portion of the preorbital bar) in Perimys as it is in the living forms (Scott, 1905, PI. 68, fig. 4, and original specimens). We are thus left with tlie more open folds and the masseteric ridge as the characters of the new family. As noted above, the masseteric ridge, or crest, is considerably less prominent in Perimys than in Scotamys, which suggests it was undergoing reduction through time. The folds in the upper cheek teeth of Prolayostomns, which Landry places in the Chin- chillidae, are not as tightly appressed as in the living forms. It would seem that narrowing of the flexi and flexids was progres- sive in the family. Perimys may well have been, in these respects, a persistently primitive form, and may prove referable to a 322 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY distinct subfamily when the phylogeny is better known, but we do not think that recognition of a family Perimyidae is at present helpful. Table 2 Tooth measurements (in mm.) of Scotamys antiquiis Uppee Teeth P^ Ml »r M3 M, Mo M.N.H.N. No. 1903-3-8 X left M.N.H.N. 0. 1903-3-47 right anteroposterior 2.53* 2.70* ■width anteroloph 2.57* 2.21* width posterior lo )e 1.73* 1.64* M.N.H.N. Xo. 1903-3-9 N left M.N.H.N. 0. 1903-3-10 left M.N.H.N. No. 1903-3-11 left anteroposterior 2.58* 2.49* 2.20* width protoloph 2.33* 2.41* 2.53* width metaloph 2.15* M.N.H.N. No. 1903-3-12 X right 2.18* M.N.H.N. 0. 1903-3-13 left 2.41* anteroposterior 2.84* 2.68* widtli protoloph 2.12* 2.61* width metaloph 2.23* 2.38* M.N.H.N. M.N.H.N. No. 1903-3-72 No. 1903-3-71 left left M.N.H.N. No. 1903-3-14 right anteroposterior 2.19 2.07 2.25 transverse 1.40 Lower Teeth 1.60 1.48 T.vpe A.C.M. M.N.H.N. No. 3003 Xo. 1903-3-4 N left left M.N.H.X. o. 1903-3-5 left M.N.H.N. No. 1903-3-0 right anteroposterior 2.62 2.68* 2.45* 2.97* width metalophid 1.74 1.75* 1.86* 1.41* width hypolopliid 2.75 2.58* 2.38* 2 ^^9* M.N.H.N., variou« specimeu.s, left ri, No. No. No. No. No. Ml aii.l M.. ?lit No. No. 1903- 190::- 1903- 3-66 3-68 3-67 1903- 1903- 3-65 3-62 1903- 1903- 3-63 3-64 anteroposterior 2.64 2.77* 2.49* 2.15* 2.76* 2.53* 2.25* 2.23* width metnlophid 2.38 2.48* 2.57* 2.40* 2.56* 2.63* 2.48* 2.43* width hypolopliid 3.08 2.45* 2.56* 2.52* 2.73* 2.99* 2.73* 2.74* anteroposterior 2.78 width metalophid 2.62 width liypolophid 2.85 WOOD AND PATTERSON : OLIGOCENE EODENTS OF PATAGONIA 323 M:; M.N.H.N. No. 1903-3-7 right iiiiteroposterior 2.09* width metalopliifl 2.20* wiiltli hyjioloiiliid 2.08* Type No. A.C.M. 3063 M.N.H.N. No. 1903-3-80 M.N.H.N. No. 1903-3-1 nnteroposteriov 1.98 1.98 1.90 transverse l..-)2 1.50 1.51 * = Maximum diameter of tooth. Family DASYPROCTIDAE Smith 1842 Cephaloniyidae Ameghino 1897. Two previously described Deseadan genera are here referred to this family, Ccphalomys Ameghino and Lifodontomys Loomis, and there is a third unnamed form that appears to belong with them. There has been much uncertainty concerning the position of Cephalomys and a reevaluation of its affinities leads to certain changes in the content of superfamilies, not only as envisaged by others (e.g. Simpson, 1945) but also as envisaged by ourselves at the stage of this investigation reported by Wood (1955). Ameghino, in establishing the family Cephaloniyidae, stated that "On ne pent les placer dans aucune des families connues de preference aux autres ... lis presentent un assemblage de earacteres propre aux Eriomyidae [=Chinchillidae], Caviidae, Echinomyidae [=Echimyidae], Hystricidae, etc." (1897e, p. 494). At that time and subsequently, Ameghino regarded them as the basal stock for the Hystricomorpha {sensu laio). Loomis (1914, p. 186) put Cephalomys in the Chinchillidae, as ancestral to Perlmys of the Santa Cruz.^^ Miller and Gidley (1918, p. 445) placed Cephalomys and Litodontomys in their subfamily Octo- dontinae (our Octodontidae plus Ctenomyidae) of the family Echimyidae. L. Kraglievich (1940a, p. 441) considered the Cephalomj-idae to have been ancestral to the Eocardiidae and, through them, to the Caviidae and Hydrochoeridae. Simpson's 11 Laudry (l'J57a, p. 54) states that Ameghino and Loomis placed Cephalomys ill the Caviidae and that Simpson referred it to the Chinchillidae. However, Ameghino never considered it as anything but the type of his family Cephalomyi- dae, Loomis referred it to the Chinchillidae, and Simpson accepted Ameghino s Cephalomyidae. 324 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY classification (1945, p. 94) placed them in a similar position. We formerly referred them to the Capromyidae (in Wood, 1955). Landry (1957a, pp. 54 and 59), considers the Cephalomyidae to he a basal family of the Chinchilloidea, from which he excludes Dasyprocta. It appears to ns that almost every allocation except the right one, or so we now think, has been made. There are no very striking resemblances between Cephalomys and the Cavioidea in general, or the Eocardiidae in particular, other than tliose common to most of the South American rodents. Some similarities exist in the general outlines of worn teeth, but the details of the tooth structure of Cephalomys are very different from those of Eocardia (cf. figures given here with those in Scott, 1905, PI. 69). The resemblance to Perimys, pointed out by Loomis and by Landry, is definite, but again is most prominent in worn teeth, and is largely confined to the upper molars. The reversal of pattern between worn upper premolars and molars is present in both of these genera, however, and this is of interest and is discussed below. The cheek teeth are far more advanced than those of the contemporary octodontids and echimyids. Now that unworn teeth are available, it has become apparent, as pointed out below, that there is great similarity, although not identity, in the dentition between Cephalomys and Neoreoniys of the Santa Cruz, which we think indicative of affinity, and this is supported by what is known of skull structure. Since Scott's study (1905) of the Santacruzian rodents, the view that Neoreo- mys and apparently related genera — Scleromys and Olenopsis — were capromyids or myocastorids has been rather generally held. Accepting this, without more ado we considered Cepha- lomys and, tentatively, Litodontomys to be capromyids. Further, noting the close resemblance in cheek tooth structure between Cephalomys and the contemporary chinchillid Scotamys Loomis, we transferred the Capromyidae from the Octodontoidea of Simpson's classification to the Chinchilloidea (Wood, 1955). It would now appear that our acceptance of Neoreomys as a capro- myid and the allocations based on this acceptance were in error. We had noted the replacement of dm| by P| in Neoreomys, a difference from the Capromyidae, in which, as in post-Deseadan Echimyidae, there is no such replacement, but had supposed that retention of the milk teeth in the family had come about in post- WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 325 Santacruzian time.^- However, reinforcing this difference, there is also a cranial character that sets Neoreomys sharply apart from the Caprom.vidae. In the latter group there is a process from the siipraoccipital (the lateral process of Tullberg, 1899, p. 69) that projects downward parallel and anterolateral to the paroccipital process, from which it is separated by a narrow notch. This process is very large in Myocastor and of more modest dimensions in the other capromyids. It also occurs in echimyids, ranging in size from incipient to small, and in erethizontids. The structural base for such a process is present in Santacruzian {Scianiys, Scott, 1905, PL 67, fig. 9) and Recent octodontids, and in abrocomids and ctenomyids, in which a strip from the supraoccipital extends down to a point above the attach- ment for the stylohyal but does not form a process. In the Deseadan Platypittamys, however, so far as can be determined from the crushed material, there is not even a rudiment of such a strip. This suggests that the strip — and the process that could arise from it — was not a part of the original caviomorph heritage but arose independently more than once. The primitive octodontids could have given rise to groups with and without this character, and the possibility of its later rise in other groups cannot be excluded. Landry (1957a, pp. 74-75) correctly points out that this feature is widespread and sporadic among rodents. In one very young specimen of Erethizon seen by us the "proc- ess" starts as an independent ossification, separate from the supraoccipital. The complete absence in Neoreomys of even the rudiment of a supraoccipital strip (see Scott, 1905, PI. 54, figs. 6, 6c), not to mention a lateral process, coupled with re- placement of dm| , however, would appear to exclude this form from the Capromyidae. The latter family, as probable echimyid derivatives, should be returned to the superfamily Octodontoidea." The combination of lateral process and un- replaced dm| so neatly reinforces other characters that suggest a close relationship between the Echimyidae and the Capromyi- dae that it is almost a pity to have to introduce a note of caution. 12 The supposedly Santacruzian Paramyocastor Ameghino (1904) is actually from the Pliocene Hermosan (J. L. Kraglievich, personal communication). 13 We are greatly indebted to Drs. Ernest E. Williams, Karl Koopman and Samuel B. McDowell for their kindness in pointing out to us the important lack of a lateral process in Neoreomys. 326 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Rodents being what they are, however, we are compelled to observe that the two characters do not always go together in the Caviomorpha. Elasmodontomys has a lateral process second only to Myocastor in point of size, yet its milk molars are shed," a feature that scarcely admits it to either the Capromyidae or the Echimyidae. Neoreomy." and Cephalomys being excluded from the Capro- myidae, the question of their familial position comes to the fore. Fortunately we do not have far to seek for an answer, no farther in fact that the work of Scott, who (1905, pp. 387-399) very clearly pointed out the many resemblances of Neoreomys to Dasyprocta and Cnniculus. Quite obviously, he felt that Neoreo- mys could almost equally as well be referred to the Dasyproctidae as to the Capromyidae. Following the hint contained in his work, and in agreement with Miller and Gidley (1918, p. 447) and Winge (1924, pp. 66. 77), we now refer Neoreomys and its rela- tives to the Dasyproctidae. Scott's work, moreover, contains a further hint of interest in this connection :".... the likenesses of Neoreomys to the Dasyproctidae and Dinomyidae are also very suggestive and indicate that this genus is not far removed in structure from some earlier and more generalized form, which was the common ancestral stock of several distinct families" (1905, p. 387). We are in complete agreement with Scott that Neoreomys suggests some degree of relationship between Dasyproctidae, Cuniculidae and Dinomyidae. With Schaub {in Stehlin and Schaub, 1951, pp. 369-370), and Landry (1957a, pp. 44, 57-58), we now believe that these families should be removed from the Cavioidea of Simpson's classification, thus limiting that superfamily to the Eocardiiclae, Caviidae and Hydrochoeridae. We agree with Schaub that the Cuniculidae and Dinomj'idae should be associ- ated with the Chinchillidae, and to these we would add the Dasyproctidae, which he left as inccrtae sedis together with the ''Cephalomyidae" (Schaub, 1953a, pp. 396-397). The West Indian Elasmodontomys group may also be included here, at least 14 "Heptaxodon" is in fact based on young individuals of Elasmodon- tomys with dm ^ . The fine material now available, much of it obtained since Anthony's work on these forms, includes enough of the growth stages to relieve all doubt on this head. This question will be discussed in a forthcoming study of West Indian rodents b.v Mr. Clayton E. Ray. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 327 provisionally. For this assemblage the earliest available name is Chinchilloidea L. Kraglievieh 1940. We agree with Schaub, Stir- ton (1947), Landry (1957a) and Fields (1957) that the extinct Eumegamyinae belong in the Dinomyidae and not in the EI asm o- dontomys gronp. Landry (1957a, pp. 57-59) groups the Myocastoridae (includ- ing iVeoreomi/s), Cuniculidae {mcluding Dasyprocta), Dinomyi- dae, Octodontidae and the Elasmodontomys group in the Octo- dontoidea. As we have indicated, we consider Neoreomys to be a dasyproctid. Eumysops, which Landry refers to the Myocasto- ridae,^° is an echimyid. We feel that the distinction between the Cuniculidae and Dasyproctidae is sufficient to justify their fa- milial separation, and that, as we have just indicated, they should be referred to the Chinchilloidea, while Myocastor surely belongs with the Echimyidae and Capromyidae in the Octodontoidea. The Chinchillidae, as represented by Scotamys, and the Dasy- proctidae, as represented by Cephalomys, are very similar in most details of crown structure in the Deseadan. Both forms have molars in which the structural details are rather slialloAv, in the former somewhat shallower than in the latter. This, we suspect, foreshadows a marked difference between later repre- sentatives of these two rapidly diverging families. In the Dasy- proctidae, the evolutionary trend was toward the preservation and increase of crown complexity, the crown growing down toward the root, as it were. Chinchillid molar evolution pro- ceeded in almost precisely the opposite direction, eliminating all the minor features of the cheek tooth crown, deepening only the clefts between the lobes. Cephalomys, in view of its resemblance to Neoreomys and to the contemporary chinchillid Scotamys, seems surely to be a member of this group, but its exact position within the super- family is uncertain. As is emphasized below, the upper molars of Neoreomys and Cephalomys are not identical ; they differ in the structure of the neoloph and in the extent of enamel deposition. There are also differences in the skull and mandible, which are generally, but not entirely, due to Cephalomys being the more primitive. Although the two forms are, we believe, related, 15 A deduction from the position occupied by the genus in L. Kraglievich's list (1034, p. 30) of Argentinian Pliocene rodents (Landry, pers. comm.). 328 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY they are most certainly not ancestor and descendant. In fact, Cephalomys cannot at present be brought into direct relationship with any later rodents, which is rather curions in view of its great abundance in the Deseadan. For the moment it can only be regarded as representing a very successful although short- lived phylum. Litodonfomys may be related to the type species of Olenopsis although this is uncertain. The third Deseadan form, represented only by an isolated upper cheek tooth in the Ameghino Collection, was, on this slender evidence, perhaps close to the Santacruzian Scleromys, and may conceivably, therefore, have something to do with the ancestry of this form. The on\j known Colhue-IIuapi form possibly referable to the Dasyproctidae is represented by an isolated upper molar intermediate in some respects between those of Neoreomys and Scleromys. Divergence was clearly under w^ay during the Deseadan, but presumably had not begun very long previously. The phylogeny of these forms was certainly complex. We undoubtedly know only a small fraction of the forms that once existed. It will be evident that in a situation such as this taxonomic assignments can only be tentative. There is no available evidence that any of the Santacruzian and earlier forms are members of the Cuniculidae. There is a possibility that some of them may be dinomyids. Very recently. Fields (1957) has published a valuable paper on late Miocene rodents, especially Dinomyidae, from Colombia. He refers the species of this family represented among his material to Scleromys (in wdiich he includes Lomomys) and to Olenopsis, the type species of which are Santacruzian. There is no question as to the familial position of the Colombian species, but we feel some uncertainty as to the generic references. As pointed out below under Lifodontomys, there is doubt as to whether the Santacruzian species of Olenopsis is congeneric with tlie Colombian one described by Fields. AVe reserve judgment, l^ending further knowledge of the type species of both genera. The Colhuehuapian material (including Stirton's 1953, p. 611, record of teeth of Scleromys 1 from Colombia) is so fragmentary as to permit no certainty as to familial position. Continued recog- nition of the family Cephalomyidae for the reception, primarily, of one seeminglv aberrant genus is liardlv warranted. Our WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 329 present, very tentative, conception of affinities within this very complex and little understood group is shown in Figure 12. SANTACRUZIAN COLHUEHUAPIAN DESEADAN CHINCHILLIDAE NEOREOMYS SCLEROMYS OLENOPSIS LITODONTOMYS OCTODONTIDAE Fig. 12. Tentative phjiogeny of the Oligocene and early Miocene Dasyproctidae. Cephalomys Ameghino, 1897 Cephalomys Ameghino 1897c, p. 494. Loomis 1914, pp. 186-188; Stehlin and SehauL 19.11, p. 61 ; Schaub, in Stehlin and Schaub 1951, pp. 245-246. Orchiomys Ameghino 1897c, p. 495. Asteromys Ameghino 1897c, p. 495; 1902b, p. 37 (in part, not including type species). Type species. Cephalomys arcidens Ameghino 1897. Referred species. Cephalomys plexus Ameghino 1897. Type species of synonym.. Orchiomys prostans Ameghino 1897. Distribution. Deseadan, Patagonia. Emended diagnosis. Teeth high-crowned but rooted, with uni- lateral hypsodonty, particularly in upper molars; no cement on cro^^'ns; crown pattern of unworn cheek teeth essentially re- sembling that of Neoreomys, also basically similar to that of Platypittamys but with P^ much more advanced than in latter ; pattern disappearing fairly rapidly with wear, much less per- sistent than in Neoreomys; enamel interrupted on lingual and anterior sides of lower teeth and buccal and posterior sides of upper teeth after considerable wear; rostrum shorter than in Platypittamys, narrower than in Neoreomys ; fossa for M. masse- tericus medius pars anticus present on lateral surface of rostrum but smaller than in Neoreomys; infraorbital foramen 330 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY very large, larger tlian in Neoreoniijs; mandibular condyle low, nearly on level of cheek teeth. The synonymy is disenssed l)olow nnder tlio species. Dentition All the cheek teeth of this genus are divided into two lobes. In the upper molars, these lobes unite, after wear, on the buccal side of the crown; in the lowers, at the center of the tooth. The molars are made up of five cross crests above and four below, but the valleys l)etween the members of the anterior and posterior sets of crests are shallow and evanescent. Several specimens in the Amherst, Buenos Aires, Chicago and Paris collections show unworn or essentially unworn cheek teeth, from which the details of the pattern can be determined for nearly all of the teeth. /-^, Fig. 13. Cephalomys arcidens Ameghino, LP"* x 5. A and B, crown and posterior views of A.C.M. no. 3122; C, croAvii view of A.C.M. no. 3064. Two unworn and two little worn P' are available: A.C.M. no. 3122 (Fig. 13A and B), C.N.H.M. nos. P15241 and P14652 (Fig. 14A) and M.N.H.N. no. 1903-3-23. These agree closely in essen- tials. There is a long narrow blade extending across the anterior and lingual sides of the crown to unite posterointernally with the straight, transverse posterior margin of the tooth. In the middle of the buccal side is a large cusp that sends a curved crest posterad to the posterior margin. A little distance anterior to the point of junction, a short oblique ridge runs anteroexternally between this crest and the posterior margin, isolating a shallow, WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 331 posteromedian fossette on three of the teeth. The interpretation of this structure seems clear. The large, curving anterior and internal crest is certainly the anteroloph plus pi-otoeone; the large buccal cusp is the paracone; and tlie crest curving pos- teriorly from it is a protoloph that has lost contact with the protocone. The transverse, posterior crest may be interpreted as the posteroloph with the metacone incorporated in its external portion, precisely as in Platypittamys and Dcseadomys. The short, oblique crest connecting the posteroloph and the protoloph appears to be the serial equivalent of the corresponding addi- tional crest (neoloph) of the molars (see below) rather than a metaloph. A hypocoue, as such, is not differentiated and there is no trace of a hypoflexus. P^ of the contemporary chinchillid Scotamijs, Avliich is a basically similar tooth, has a very shallow hypoflexus. This suggests that the hypocone may have been slightly differentiated, or that a tendency existed toward its dif- ferentiation, in the common ancestry, perhaps to the extent seen in Platypittamys. The Santacruzian Scleromys, which has a deep hypoflexus and a well defined hypocone in P^, also suggests that this was the case and that, furthennore, divergence took place among these forms, Ccphaloniys and Ncorconiys obliterat- ing, the Santacruzian Scleromys accentuating the hypoflexus. According to Fields' figures (1957), the Colombian species he refers to Scleromys agree in this respect with Ceplialomys and Neoreomys. However, his figures indicate that P"* in his forms is more molariform than in Ceplialomys. Minor variations in pattern occur in all specimens. In M.N.H.N. no. 1903-3-23, which is very slightly worn, there are irregularities in the buccal margin of the paracone, not indi- cated in the other specimens. The paracone of the unworn C.N.H.M. no. P15241 is broad and quadrilateral, instead of being a thin blade. There is a clearly marked depression on its top, surrounded by ridges of thick enamel and floored with a thin veneer of enamel. There are three buttresses of enamel running down the buccal margin of the cusp, which would give its outer border a fluted pattern if the tooth were slightly worn. There are two small knobs on the posterior face of A. CM. no. 3122. These differences in the details of the crown (seen also in the unworn molars) are not considered to have any taxonomic sig- nificance. Variation of this sort (which would be considered of 332 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY specific or even generic value in brachyodont teeth) is a very common feature in the portions of the crown pattern of extremely hypsodont teeth that have not taken part in the increased hypsodonty, and is of no taxononiic significance whatever (Wood, 1940, p. 300), though it may give clues to the phylogeny of the animals concerned. B C D Fig. 14. Cephalomys, upper cheek teeth, x 5. Cephalomys arcidens Ameghino. A, LP^-M^, C.N.H.M. no. P 14652; B, LP-*-M3, A.C.M. no. 3099 (M3 reversed from right side of same specimen). Cephalomys sp. C, LP^- M^, A.C.M. no. 3109. Cephalomys plexus Ameghino. D, LP-i-lVP, A.C.IM. no. 3085. The first development during wear of P^ is the isolation of a small lake between the knobs on the posterior surface, when these are present (Fig. 14A). Further wear produces a trident-like pattern (Fig. 14B), and then eliminates all the pattern except for the valley behind the protocone-anteroloph blade, which re- mains as a lake reaching almost to the roots (Fig. 13C). The WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 338 sequence of pattern changes with wear (at least in C. arcidcna) is as follows : 1. elimination of minor irregularities and fonnation of pos- terior lake (Fig. 14A) ; 2. elimination of postiM-ior lake and develoi)ment of trident pattern (Fig. UB) ; 3. conversion of posterior external valley to a lake (Fig. UC) ; 4. elimination of posterior external lake (Fig. 13C) ; 5. conversion of anterior valley to a lake (Fig. 13C) ; 6. interruption of enamel on posterior face. In Cephalomys plexus, it seems that there may be variability in the sequence, with some specimens agreeing with that listed above (Fig. 14D), whereas in others stage 6 precedes stages 4 and 5 (Fig. 22B), as is also true in Cephalomys sp. (Fig. 14C). Tlie upper premolars of the dasyproctid Cephalomys and of the chinchillid Scofamijs are very similar, differing only in that the latter apparently lacks a neoloph, and has an internal opeuing of the paraflexiLs. Superficially, they appear to be quite different from those of the early octodontids, yet closer inspection reveals that all the ingredients for them are represented in P* of Platy- pitfamys and Deseadomys. To convert premolars of this type into the Cephalomys-Scotamys P* would require only the pos- terior growth of the protocone, loss of contact between the paracone and the protocone, and a posterior shifting of the lingual end of the paracone crest as the protocone shifted to the rear. This paracone shift might have been facilitated by capture of an incipient mure of the type seen in Deseadomys (Fig. 4). An entirely unworn left upper molar (probably M^) of C. plexus, M.N.H.N. no. 1903-3-3, reveals every detail of the crown (Fig. 15A). Four crests are present, the third crest running obliquely to the posteroloph in the more buccal part of its course. This crest we take to be the lingual part of the metaloph plus a diagonal neoloph crossing the metafiexus. The protoloph, lingual part of the metaloph, neoloph and the buccal part of the postero- loph are the highest parts of the crown, and are practically at the same level, the anteroloph being slightly lower, especially buccally. The lingual portion of the posteroloph is considerably lower and is bowed posteriorly at the occlusal surface, though not at a lower level of the crown. At the posteroexternal angle of the tooth, and 334 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY somewhat below the level of the rest of the crown, is a small cuspule, from which a short crestlet runs antero-linguallj^ By comparison of this tooth with the slightly worn first or second upper molars of C. arcidens shown in Figures 15B and C, this cusp would appear to be the metacone, and the crestlet running from it to be the last stage in the degeneration of the buccal part of the metaloph. In the specimens shown in Figures 15B and C, the metaloph is complete and five crests are present. This may be a specific character, but is probablj^ an individual vari- ant, since, in upper molars of Neoreomys australis (Fig. 17A), and especially in M'^, the metaloph exhibits various stages of dis- integration. The neoloph in Cephalomys has split the metafos- sette into two parts, the posterior one (which may be termed the A B Fig. 15. Cephalomys, uuworn aud little worn upper molars. Cephalomys plexus Ameghino. A, LM^, M.N.H.N. no. 1903-3-3, x 10. Cephalomys arci- dens Ameghino. LM^ or 2^ x 5; B, M.N.H.N. no. 1903-3-22; C, M.N.H.N. no. 1903-3-20. lingual neofossette) being large, irregularly trumpet-sliaped, but decreasing rapidly in size with depth. Only a moderate degree of wear would eliminate both parts of the metafossette entirely, and no suggestion of the presence of the neoloph would remain. The parts of the metafossette are worn away in all other available material except C.N.H.M. no. P14652, where the lingual neo- fossette is preserved on ]VP (Fig. 14A). This peculiar development of a diagonal neoloph directed an- terointernally might at first glance seem to separate Cephalomys widely from all other caviomorphs. However, investigation of unworn teeth of various meml^ers of the suborder shows that it is by no means a unique feature, appearing in several forms as an WOOD AND PATTERSON : OLIGOCENE RODENTS OP PATAGONIA 335 individual variation. For example, it is also present in some specimens of Ercthizon dormtum cpi.rantJiKm (Fig. ID). In Coendou hrandti we have seen the neolopli extending to the hypo- eone. Landry ligures an interesting case in Dasyprocta fuJirjinosa (1957a, PI. 5, fig. a) in which the neolopli seems to have split into two parts, one diagonal and one transverse, and we have seen the same thing in I), aguti. Evidently this part of the npper molars of caviomorphs is (piite unstable. The angulations at the ends of the neoloph on M.N.II.N. no. 1903-3-3 would be eliminated by wear about the same time the lingual neofossette was lost (Fig. loA). The great depths of the hypofiexus, mesoflexus and jiaraflexus are strikingly apparent, the hypofiexus being the deepest, followed by the mesoflexus and Fig. 16. Cephalomys arcidens Ameghino. A. CM. uo. 3160, LM^ ""^ -, x 5 ; A, crown, and B, anterior views. l^araflexus in that order. The anteroloph and protoloph unite at the buccal margin of the tooth a short distance below the apex of the crown so that the parafossette is isolated soon after wear begins. The protoloph and metaloph (or its vestige) unite eon- sitlerably farther down, the mesofossette becoming isolated about the time the lingual neofossette disappears (ef. A.M.N.II. no. 29558 or C.N.II.M. no. P14652, Fig. 14A), or later (Fig. 16A). The mure is slightly angulate, but shows no thickening that would suggest the presence of a mesocone. In addition to giving us the crown pattern of an upper molar of a member of this genus, these teeth are interesting because of the similarity — apart from the neoloph — to the unworn molars of Nroreomys (Fig. 17A). In the latter, there is also a fifth crest, but this seems to be a normal neoloph, a development from the posterolophid, as in Protosteiromys, since it is connected at both 336 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY ends with that crest, and has no connections whatever with the hypocoue. There is a good metacone in Neoreomys, which is sometimes partially isolated from the lingual part of the metaloph as in Cephalomys, and which thus looks like part of a mesoloph, which it certainly is not. This difference between the manner of development of the fifth crest in Cephalomys and Neoreomys merely emphasizes the great plasticity of rodent cheek tooth pat- terns, and the fact that variants may (and generally do) arise in a wide variety of different manners. Perhaps this discrimina- tion of minor types of neoloph variation may seem to he ultra- fastidious but one of the difficulties in unraveling rodent evolu- tion has been the tendency to overlook such differences in origin of a pattern, or to assume that these differences are of no « B Fig. 17. Xcoreomys australis Anieghino. C.N.H.M. no. P 13164, x 5; ^, L]M2 ; B, LMo. phylogenetic significance. The development of a neoloph, of various types, seems to be characteristic of the caviomorphs, and to distinguish them from the hystricomorphs, where the fifth crest appears to be a mesoloph. The primary difference in pattern between the premolars and the molars is that, in the latter, the hypocone is a large and inde- pendent element, which it is not in the premolar. This presum- ably means that the teeth represent specializations of something like those of Platypittamys, and that, in the premolar, the proto- cone grew backward along the lingual margin of the tooth, taking over the functions of both protocone and hypocone in the molars. An additional difference is tliat the main valley of the premolar WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA .'^37 is deeper buecally than lingually, so that, \vit)i wear, it opens on the buccal side of the tooth, as in chinchillids. With wear, the molars pass through the following stages : 1. crown formed of anteroloph, protoloph, metaloph or rem- nant, posteroloph and neoloph, the last dividing the metafossette into two parts (Figs. 15B-C) ; 2. paraflexus converted to a parafossette (Fig. 14A) ; 3. elimination of all parts of metafossette ; 4. mesoflexus converted to mesofossette (Fig. 14A) ; 5. elimination of parafossette (Fig. 14B) ; G. interruption of enamel on buccal surface (Fig. 14B) ; 7. elimination of mesofossette ; 8. interruption of enamel on posterior surface (Fig. 14C). A B Fig-. 18. Cephalomys arcidcns Ameghiiio. EP4 x 5. A, A. CM. no. 3108; B, M.X.H.N. no. 1903-3-19. The little worn lower premolar (Figs. 18A, 19 A, and 19E) shows a pattern rather distinct from that of the lower molars. The trigonid consists of a metaconid, a protoeonid and an antero- conid. All of these elements are exceedingly narrow and com- pressed on the unworn surface, showing little or no suggestion of separate cusps. The protoeonid curves forward at its buccal end, and the anteroconid extends straight forward, so that, in unworn or partly worn teeth, there is an anterior lobule (Fig. 18A) giving this tooth a distinctive appearance. Various speci- mens show notable variation in the anteroconid, which is in part intraspecific (Figs. ISA, 18B, IDA, C. arcidens) and is in part associated with the fact that two distinct species are represented. In the unworn P4 of C. arcidcns the tip of the anteroconid is con- nected with the anterolophid, whereas in unworn C. plexus it stands free (A.C.M. no. 3113, Fig. 19E, and M.A.C.N. uos. 338 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY A 52-103, and A 52-106). The talonid in slightly worn teeth consists of a single lobe, connected diagonallj^ with the middle of the metalophid. This diagonal crest arises from the central part of the talonid. Unworn teeth, however, show that the lingual margin of the talonid is actually subdivided into two crests, the hypolophid and the posterolophid. The division extends neither very far across the tooth nor very far down the crown. The entoconid is considerably higher than the posterior cingulum. The enamel over the latter is very thin, sometimes appearing almost not to have been deposited. The apparent pinching at the waist of A.C.M. no. 3108 (Fig. 18A) is due to the lower part of the crown being buried in the jaw. If this tooth were fully erupted, the outline would presumably be the same as that shown in Figures 18B and 19A. The crown pattern is so shallow and so rapidly worn away that there is really no sequence in its loss. All three lower molars show essentially the same pattern, which may be very clearly seen in M1.2 of M.N.H.N. no. 1903-3-2 (Fig. 19B) and in M3 of A.C.M. nos. 3162 and 3113 (Figs. 19C and E). This pattern is obviously the same as that of Neoreomys (Fig. 17B). Again there is variation among the specimens re- ferred to the two species of the genus. The anterolophid extends across the anterior face of the tooth, reaching the lingual margin of the crown, uniting with the metaconid either at or a short distance below the level of the crown. It gives the appearance of having been prevented from growing forward by the presence of another tooth in front, whereas the absence of such a tootli permitted forward growth in the premolar. The anterofossettid is quite ephemeral, though not as short-lived as the metafossettid of P4. The hypolophid curves slightly to the rear, swinging forward to join the metaconid a short distance below the crown surface. In the crown view of an unworn tooth, the entoconid appears to have its relationships with the trigonid rather than with the talonid (Figs. 19C, E). However, the valley between this crest and the metaconid is actually much deeper than that between it and the rest of the talonid. The latter consists of a long lobe, believed to be formed of the hypoconid and postero- lophid. In C. arcidens, the posterolophid is incipieutly subdi- vided near its lingual margin in one specimen (Fig. 19C), in a manner that might be the initial stage in the development of an WOOD AND PATTERSON : OLIGOCENE RODENTS OP PATAGONIA 339 additional lobe. The roots have not yet formed on these unworn third molars, thonph they probably soon would have developed since tliey have formed on the second molars of the same speci- mens. B D Fig. 19. Cephalomy.s, lower cheek teeth, x 5. Ceplmlomys arcidens Ameghino. A, RP4-M2, A.C.M. no. 3161; B, LMi-o, M.N.H.N. no. 1903-3-2; C, LM.s, A.C.M. no. 3162; D, RM1-.3, A.C.M. no. 3078. Cephalomys plexus Ameghino. E, LP4-M3, A.C.M. no. 3113; F, RP^-Ma, A.C.M. no. 3006. 340 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY In another specimen of C. arcidens (A.C.M. no. 3144), there is a slightly more worn M.-,. In this tooth, thongb damaged, the metaconid and anterolophid have nearly united lingnally. In C. plexus they may (A.C.M. no. 3113, Fig. 19E) or may not (e.g. M.A.C.N. no. A 52-103) be united lingnally on unworn teeth. The entoconid is shorter, on the crown surface, in A.C.M. no. 3144 tlian in the previously described specimens, and the talonid is broken. The valley between the metaconid and ento- conid is 5.0 mm. deep at the lingual edge of the tooth, extending within 0.6 mm. of the base of the enamel on the crown. The roots are formed in this tooth, so that the crown height can be deter- mined to have been approximately 5.6 mm. on the lingual side and 7.5 mm. on the buccal side. The sequence of loss of crown pattern in the lower molars is as follows : 1. wear to produce a flat surface with loss of minor irregu- larities ; 2. conversion of the anteroflexid (in C. arcidcns)to a lake, as in unworn C. plexus (Fig. 19E) ; 3. loss of the anterofossettid (Fig. 19A) ; 4. conversion of hypoflexid to a lake (Fig. IDA) ; 5. loss of hypof ossettid ; 6. interruption of enamel on anterior face (Fig. IDD) ; 7. conversion of mesoflexid to a lake; 8. interruption of enamel on ling-ual face. This sequence of events is based on C. arcidens. There may be slight differences in the sequence in C. plexus, but they are not very significant. In general, through stage 6, each tooth is about one stage ahead of the next one behind it. Four specimens in the Amherst and two in the Ameghino collections show dm^. There is a certain amount of variation in the anterior part of the crown, which is not unusual among rodents (Figs. 20A, B). The talonid seems to be similar to that of the permanent teeth, except that the valley between the ento- conid and the posterolophid is as deep as any other valley in the crown. The talonid is connected to the center of the straight metalophid. Anterior to this is what seems to be a tripartite anteroconid, connected to the protoconid or to the protoconid and metaconid. There are two buccal, one anterior, and four lingual valleys. The tooth is replaced after Mi has reached wear stage WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 341 3 or 4, and M2 wear stage 2 or 3. and yovy shortly after M.-? has come into use. The crowns of all the teeth were covered b}' a complete enamel cap, even though it was removed very quickly by wear. That on the upper surface of the talonid of P4 was so thin as to appear absent (Fig. 19E). Its thinness in this specimen cannot be due to wear, however, unless the wear was against the roots of dm4, since we personally removed the roots from over this part of the crown in A.C.M. no. 3113. The enamel extends to varying distances down the sides of different parts of the cro^vn of the cheek teeth, due to the asymmetric development of hypsodonty. It is interrupted, after the tooth is about half worn down, along the anterior faces of A B C Fig. 20. CepJialomys arcidens Ameghino, lower teeth, x 5. A, Rdm^, A.C.M. no. 3013; B, Ldm4, A.C.M. no. 3011; C, cross-section of EIi, anterior view, A.C.M. no. 3155. the lower molars (Figs. 19D, F). Before this level is reached, the thickness of the enamel is considerably reduced by inter- dental wear (Fig. 19A). The enamel is next lost on the posterointernal corner of the entoconid, beginning with Mi. This occurs much earlier in C. plexus (Fig. 19F) than in C. arcidens. A layer of cement is deposited around the basal portions of the crown, particularly in C. plexus, which sometimes makes it difficult to determine the exact points at which the interruptions occur. In the upper molars, since the teeth are high-crowned lingually and low- crowned buccally, the initial point of loss of enamel is the buccal surface of M^, followed in turn by M- and M-^. The enamel is then lost on the posterior faces of the same teeth. Before the level of no enamel is reached, the buccal enamel becomes con- siderably thinned, which is obviously not due to interdental wear. 342 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY These interruptions are certainly primarily due to non-deposi- tion, and not to interdental wear, as may be seen by looking at the anterior or posterior faces of slightly worn teeth (Pig. 16B), where the uneven ends of the enamel may be seen on the intra- alveolar portions of the teeth. Conditions appear to be identical with those described for the Lagomorpha (Wood, 1940, pp. 356-357). After the portion of the tooth with interrupted enamel has reached the occlusal surface, however, interdental wear becomes quite prominent, and the shapes of the teeth may change considerably and quite rapidly, due to the wear of the enamel of one tooth against the dentine of the adjacent tooth (11. E. Wood, 1938). The crowns of the teeth are quite high, the height in unworn upper premolars and M3 reaching twice their maximum diameter. Tlie difference in pattern between slightly worn lower pre- molars and molars is believed to be due to two factors: 1) the anterolophid of P4 extends anterad, whereas in the molars it is squeezed parallel to the metaconid ; and 2) the entoconid of P4 is closely united with the posterolophid, on the wear surface as well as farther down the crown. However, the difference in pattern between both upper and lower premolars and the molars indicates that the molarization of the premolars in this, as Avell as in other Deseaclan genera in which the premolars are known, occurred at a time when the molars had already acquired the main elements of their pattern, while the premolars were relatively simple. It would therefore be in accord with the derivation of Cephalomys from something quite similar to Platypittamys. The lower incisor has a nearly flat but gently curved anterior face (Fig. 20C). The enamel face bears a series of fine lines, nearly parallel to the long axis of the tooth. The enamel extends over about a third of the lingual side and about half the buccal side of the tooth. The incisors are relatively short and narrow in proportion to the size of the cheek teeth. The pulp cavity is roughly triangular in outline. No upper incisors are present in the jaws, but two such incisors are apparently associated. They appear to be similar to the lower incisors, but have not been figured because of the lack of positive association. The course of the alveolus is shown in M. A.C.N, no. A 52-89. It is curved to about the same extent as in Neoreomys, WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 343 but does not extend as far posteriorly, ending a short distance behind the maxillo-premaxillary suture and thus well in front of the anterior root of the zygoma, its posterior extremity descends farther ventrally, liowever, than in the Santa Cruz form. Fig. 21. Cephalomys arcidens Ajiieghino. Facial region of skull, M.A.C.N. no. A 52-89; A dorsal, B ventral, C right lateral view, x 2. Crosshatching indicates matrix; position of infraorbital foramen shown in dashed out- line in C. 344 bulletin : museum of comparative zoology Skull and Mandible Various specimens, M.A.C.N. no. A 52-89, A.C.M. nos. 3066, 3085, 3091, 3099, 3274 and A.M.N.H. no. 29558, combine to fur- nish some information concerning the skull. By far the most complete of these is M.A.C.N. no. A 52-89, consisting of the greater part of the facial region. This is the "skull" mentioned by Ameghino in his original description (1897c, p. 494) and the following account is largely based on it. The rostrum (Fig. 21) is relatively narrow and nearly parallel- sided; it is shorter, but otherwise rather similar to that of Platypittamys, and decidedly narrower than the rostrum of Neo- reomys. The nasals are missing. They extended as far posteriorly as in Neoreomys, and the suture between them and the frontals is nearly as transverse; the only notable difference between the two forms in this area is that the frontal in Cephalomys sends forward a small wedge l-etween the premaxillary and the nasal. The position of the suture relative to the dorsal root of the zygoma was clearly very similar in both. The frontals are extremely flat, even those parts above the orbits not curving downward to any marked extent. The maxillo-frontal suture, transverse in Neoreomys, is here almost longitudinal in direction, due to a narrow triangular projection from the frontal that runs forward between the premaxillary and maxillary. Posterior to this projection, the edge of the frontal shows a somewhat ribbed ventral surface, which suggests the sutural area for the lachry- mal. If this indication is correctly interpreted, the size and rela- tions of the latter bone must have been much as in Neoreomys. Behind this supposed sutural surface, the upper margin of the orbit is more sharply excavated than in Neoreomys. In all probability a postorbital process was present, as in Caviomorpha generally, but the specimen ends at a point just anterior to where this structure would begin. The premaxillary is similar in general to that of Neoreomys. Anteriorly, it sends forward a thin median projection in advance of the incisor alveoli, a feature Ijarely indicated in the Santa Cruz form. The dorsal process is stout, and extends back almost as far as the level of the fronto-nasal suture. The lateral surface of the bone, in decided contrast to Neoreomys, is not involved in the masseteric fossa on the side of the rostrum. The ventral WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 345 surface is about as wide, proi)ortionately, as in Neoreomys, the expansion of the rostrum in the latter having taken place in the dorsal portion, evidently pari passu with the forward extension of the masseteric fossa. The incisive foramina are wade, much wider than in Neoreomys, and lie together in a fossa that deepens posteriori}' in its maxillary portion, the posterior wall sloping sharply posteroventral to the palate. The fos,sa extends back almost to the level of P"" and slightly beyond that of the anterior root of the zygoma (Fig. 22B). Precisely similar conditions do not appear to exist in any Santa Cruz rodents. Fig. 22. Cephalomys. A, Cephalomys plexus Ameghino, dorsal view of left maxilla, A.C.M. no. 3085; B, Cephalomys sp., ventral view of palate, A.C.M. no. 3109 (detail between tooth rows restored from A.C.M. no. 3091, C. plexus), X 3. C, Cephalomys arcidens, lateral view of mandible, A.C.M. no. 3058 (parts in outline from A.C.M. no. 3005, C. plexus), x 2. 346 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY The palate is very compressed, being narrowest between the premolars and gradually widening (Fig. 22B), as in Myocastor and Olenopsisf (Fields, 1957, Fig. 15). The posterior widening is much more marked than in Neoreomys but less so than in Perimys. A deep notch, extending as far forward as the middle (C arcidens) or even the front (C plexus) of M^, marks the posterior border of the palate. Such a deep notch is not present in Neoreomys nor in any other Santa Cruz rodent figured by Scott except Stichomys (Scott, 1905, PI. 65, fig. 17) ; there is no posterior median projection of the palate. The palatine ex- tends parallel to the margin of this notch, its anterior limit being near the front of M^ (Figs. 21B, 22B). The posterior palatine foramina lie very close to the midline beside the anterior tip of the palatines, which they may indent very slightly (Fig. 22B). In this, they differ from Neoreomys, in which the foramina notch the palatines so deeply "as to be almost entirely enclosed in them" (Scott, 1905, p. 394). The greater part of the zygomasseteric region is preserved, and what is missing may be inferred with confidence from what is present (Figs. 21-22). The anterior root of the zygoma is somewhat variable in size in some specimens of C. arcidens, ex- tending from well in front of P'' to a point external to M^, but in some of C. plexus not extending behind P*. The ventral surface of the arch is gently concave, with a sharply crested external margin and a rounded internal one that becomes less pronounced posteriorly. Anterointernally there is a small but clearly defined ventral projection, on the posterior slope of which is a very slightly roughened area, prominent in some specimens of C. plexus, that is set off a little from the surrounding bone. It seems very likely that this marks the place of origin of the tendon of M. masseter superficialis. M. masseter lateralis clearly originated from the concave under surface of the arch, and M. masseter medialis, pars posticus, probably had its origin along the medial border. The relations of these portions of the muscle mass would appear to have been essentially similar to those of Chinchilla (Wood and White, 1950, pp. 552-554, Figs. 2-3). The dorsal side of the ventral root is very robust, arching strongly upward from front to back. Its posterior surface passes medially into a raised area on the side of the maxillary above P^-M^ The arch is preserved to a point opposite M- in M.A.C.N. no. A 52-89, WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 347 and at its posterior extremity the dorsal margin begins to turn upward, thus marking the beginning of the ventral part of the lateral wall of the infraorbital Foramen (Fig. 21C). In Neoreo- mys the upturn begins opposite M^ The fenestra, therefore, was relatively larger in the Oligocene form than in the Miocene one, a fact of some interest, as will appear. The infraorbital nerve and blood vessels evidently ran forward medial to the raised area on the side of the maxillary, where there is a groove (Fig. 22A), and reached the rostrum via the ventral part of the infraorbital foramen. The lateral surface of the rostral portion of the maxilla is occupied by the masseteric fossa mentioned above, which extends anteriorly as far as the maxillo-pre- maxillary suture (Fig. 21C). Within the fossa, the naso-lachry- mal canal is large and prominent, similar in general to that of Neoreomys and Dasyprocta. Dorsally, the fossa is roofed, as in most caviomorphs, by a thin lateral extension from the upper part of the maxillary, which progressively increases in width poster- iorly. This extension is broken off at the level of the naso-frontal suture, but obviously it continued downward and backward to join the upturned part of the zygomatic portion of the maxillary and thus formed, together with the lachrymal, the lateral wall of the infraorbital foramen (Fig. 21C). The fossa clearly served for the reception of M. masseter medialis, pars anticus. Wood (1949, pp. 13, 49-50) has suggested that enlargement of the infraorbital foramen did not accompany but preceded the forward migration of the anterior portion of the deep masseter to the rostrum. Conditions in Flat y pit t amy s, in which there is a foramen of moderate size, much smaller than in other cavio- morphs, and no indication of a fossa in the side of the rostrum, certainly appear to indicate that this was the case. Cephaloniys is clearly a more advanced form. As noted above, the infraorbital foramen of Cephalomya is relatively larger than that of the later Neoreomys. Despite this, the fossa in the rostrum is smaller, extending anteriorly to the maxillo-premaxillary suture, whereas that of the Santa Cruz form extends far on to the premaxillary (Scott, 1905, PL 64, fig. 6a). It seems evident that the part of the masseter that had its origin in the fossa in Cephalomys must have been smaller than the corresponding part in Neoreomys. As in riatypittamys, it is therefore probable that enlargement of the foramen preceded, or at least more than merely kept pace 348 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY with, the forward migration of the muscle. This is not the only interesting inference to be drawn from the available facts. Platy- pittamys, in which the masseter had not begun, or at most had just begun, to pass off the ventral surface of the zygoma, is an octodontid ; Cephalomys, in which the migration is in full prog- ress, is a dasyproctid. It would therefore appear that in the ancestral caviomorph stock : 1) there was an infraorbital foramen of moderate proportions; 2) M. masseter medialis, pars anticus, was of protrogomorph type; 3) this combination provided the structural basis for the acquisition of the caviomorph zygomas- seteric structure; and 4) the advanced stages of this structure were acquired independently in the various groups that rapidly diverged from the ancestral stock. It is now, we believe, virtuallv certain that the resemblances between caviomor])hs and hystrico- morphs in zygomasseterie structure are the result of parallel evolution. Several mandibles, one or two quite complete, are known. The ventral part of the mandiljle is broadened, as in Cavia, Ciiniculus, Neoreomys, Dasyprocta and Lagostomus, but is quite different from the conditions in Chinchilla, Cocndou or Erethizon. This thickening is lateral to the roots of the cheek teeth, and is ap- parently related to the development of a more efficient insertion for the masseter. The lower incisor lies on the lingual side of the roots of the cheek teeth, its posterior end being below or nearly below M.. The diastema is rather short. The mental foramen is high on the ramus, nearly at the alveolar border of the diastema, and well forward of the cheek teeth. There is some uncertainty as to the coronoid process. M.A.C.N. no. A52-104, C. plexus, the specimen figured by Ameghino, shows a thin ridge of bone running from a point external to the posterior half of M3 to the base of the condyle and scarcely rising to the level of the molars. There is no trace of an elevation on the ridge, although a small process could have been broken off. Loomis (1914, p. 191, Fig. 121) figures and describes a small coronoid. His figure was apparently based on A. CM. no. 3005, C. plexus, but this specimen is now broken off behind M3. A. CM. no. 3058, C. arciclens (Fig. 22C), suggests that a coronoid may have been present. The mandibular foramen is large. ^'' The condjde is oval, 16 Ameghino shows a miuute foramen beneath the mandibular foramen, within the pterj-goid fossa. Examination of the figured specimen (M.A.C.N. no. A 52-104) does not support this. The jaw is craciied in this region, and the supposed fora- men appears to be a local enlargement within the craclc. Other specimens show no foramen in this position. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 349 laterally compressed, faces dorsally, and is but little above the level of the cheek teeth. Below it, and continuing from its pos- terior margin, is a delicate right-angled process. The structure of this part of the mandible is rather different from that of Neoreo- mys,^^ in which the condyle is higher. Although the angle is broken off in all available material, the beginning of it is present in A. CM. no. 3059, showing it to have been inflected. The sym- physis is very extensive, running the entire length of the receding chin, as in Erethizon, but is rather corrugated, indicating a rela- tively firm union of the two mandibles. This varies between the species. The geniohyal pit is very weak. Near the anterior end of the masseteric fossa is a pit for the attachment of the masseter. The chin region bears a considerable number of nutritive fora- mina, as in Eutypomys (Wood, 1937a, PI. 29). Species The species of Cephaloniys are not as readily distinguishable as would appear at first glance. Due to their hypsodonty, the details of the tooth pattern are rapidly lost with wear, and only the generic pattern remains. A few possible characters have been mentioned above. Ameghino's distinctions were based largely on size. Initially we thought that three species were represented. However, on plotting the greatest length against the greatest width for each tooth, two size groups appeared. These are quite separate except for one individual, initially referred to C. plexus (A. CM. no. 3109, a palate with all eight cheek teeth; Figs. 14C, 22B, Table 6), which falls in the smaller group for P*, occupies an intermediate position as regards M^ -, although somewhat nearer the smaller group, and falls in the larger group for M^. A few other specimens that appear to be intermediate in size are clearly very old individuals of the larger species, C. arcidens, the small size being due to the fact that the widest parts of the crown have been worn away. The statistical analysis for the larger species, C. urcidcns (Tables 3-4) shows that it is a homogeneous population (V usually in the range 6-10). This is true even 17 While on the subject of the jaw of Neorcomijs, we may mention that this genus does not possess the "wide, shelflike masseteric ridge extending back along the angular process" (Landry, 19o7a, p. 45) which might be expected of it if it were closely related to Myocastor. Neoreomys is actually very much like Dasy- procta in this respect, the ridge in both being of modest proportions. 350 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY though the measurements of maximum tooth diameter, maximum extra-alveolar diameter, and diameter of wear surface were lumped (Tables 3-4). An analysis of two of the three types of measurements taken separately and all three lumped (Table 3) showed that their means and standard deviations coincided within the overlap of their standard errors. Similar investigation of the less numerous, smaller species, C. plexus (Table 5) showed about the same situation for the lower teeth, but gave extreme variation in the upper teeth (with the SD reaching 20 per cent of the mean, and V reaching 22.07) when A.C.M. no. 3109 was included. When this specimen was excluded, the statistics of the upper teeth of C. plexus correspond with those of the other sets of teeth. It is therefore concluded that there are two valid species, C. arcidens and C. plexus, and that A.C.M. no. 3109 is an anomalous individual that requires separate treatment. Its measurements are given in Table 6. Regression lines were computed for the length, x, versus meta- lophid width, y, of Mo. These gave regression coefficients for the two species which showed no significant difference, the only dis- tinction in the regression lines being the size of the species. The regression coefficients are : C. arcidens, bvx = .658, b^v = .667 ; C. plexus, byx = .765, bxy = .703. It might be worth mentioning that the coefficient of variation (V) of Cephaloniys is much greater than that computed for brachyodont rodent teeth (particularly a series on members of the Paramyidae), which is certainly related to the much greater hypsodonty of Cephaloniys and the corresponding inability to measure with certainty the maximum diameters, except on isolated teeth. As indicated, the variation in tooth measurements is partly due to wear. The widest point of the tooth is near the middle, vertically. Therefore, old individuals show smaller measure- ments of maximum transverse width than do younger ones. Very young individuals also show smaller measurements if the teeth are in jaws, since tlie widest point may be within the alveoli. In the anteroposterior measurement, there is extensive inter- dental wear after the enamel is interrupted, and the teeth become markedly shortened with further use. Processes of this sort, together with long-continued growth, rather than tooth growth anywhere except at the base of tlie crown, are, we feel certain, WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 351 the basis for the anomalous results reported by Fields (1957, pp. 854-355) for late Miocene dinoniyids. The possibility has been considered, and can by no means bo ruled out, that these two groups represent the two sexes. In general, however, sexual size differences among rodents do not appear to be as great as the differences between these two groups. Moreover, it would seem reasonable to expect more nearly equal representation of the two sexes in a sample of this magnitude than is the case with the two size groups. It seems probable to us that these differences are of taxonomic significance and that two forms are present. The presence of the unusual individual, A. CM. no. 3109, rather complicates the I)icture, in that it shares size features with both groups. This could be exjilained as merely an extreme in the variation of either group except for the statistical studies. It might also be a hybrid, if specific crosses were possible in Cephalomys. The possi- bility also exists that A. CM. no. 3109 is an isolated representative of a third species, occupying a different habitat from that in which the other two species lived and were buried. Cephalomys arcidens Ameghino 1897 Figs. 13, 14A-B, 15B-C, 16, 18, 19A-D, 20 and 21 Ccphulomys arcidens Ameghino 1897c, p. 494; 1906, p. 414, Fig. 292; Loomis 1914, pp. 189-190, Figs. 118-119; Schaub, in Stehlin and Schaub 1951, Fig. 396, p. 246. Cephalomys prorsus Ameghino 1903a, p. 93, Fig. 11, p. 98, Figs. 14, 16, p. 99 (in part, not including the type). Orchiomys prostans Ameghino 1897c, p. 495. Syntypes. M.A.CN. no. A 52-88, right and left palatal frag- ments wnth P*-M'', left mandible with P^-M.-i and root of I ; M.A.CN. no. A 52-89, facial region of skull with alveoli of R and L I, alveoli of LP*, M'' and LM'"^-. M.A.CN. no. A 52-88, Ameghi- no's figured specimen, is here designated as lectotype. Type of Orchiomys prostans. M.A.CN. no. A 52-96, fragment of right mandible Avith po.sterior part of dni4, Mi_o. 18 !■< Ameghino (1897c, p. 495) referred to the two coiniik'te teeth as the fourth and fifth molars, i.e. as P4-M1, which i.s certainly not the case. Pearlier (1S9H. 1897b), he had declared that the Ueseadan rodents had Ave lower cheek teeth, a statement that he soon corrected (1897c, p. 494, footnote), di'claring that In- had based it on a specimen in wliich the anterior part of the milk molar was preserved together with the corresponding premolar. The type of 0. prustanfi may have been the specimen in question, with "anterior end of the milk molar" l)eing a lapsus for posterior end, and >Ii mistaken for I'4 : certainly there is no other specimen now in the Ameghino Collection on which the error could possibly have been based. 352 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Hypodigm. The types listed above, and an additional series of good specimens : six in the Ameghino Collection, twenty-eight in Amherst, twelve in the Museum National d'Histoire Natu- relle, three in Chicago Natural History Museum, six in The American Museum of Natural History, five in the Yale Peabody Museum and two in the Museum of Comparative Zoology. The more important of these are mentioned in the account of the morphology or identified in the figure captions. In addition there are numerous isolated first and second molars and incisors that have not been used in the statistical investigation. Horizon and localities. Deseadan. The Museum National specimens are from La Flecha, Santa Cruz ; those in the Amherst, Chicago, Yale, Museum of Comparative Zoology and American Museum collections are from Cabeza Blanca, Chubut. Ameghino gave no localities for his specimens, but to judge from their ap- pearance, it is probable that all of them are from Cabeza Blanca. One lot of isolated teeth in the Ameghino Collection, M.A.C.N. no. A 52-97, has an accompanying label reading "Gran yaci- miento del Pyroterio — Rio Chico, ' ' which surely indicates this locality. Table 3 Comparison of measurements (in mm.) fur occlusal diameter and maximum diameter of lower cheek teeth of C. arcidens N M SD V P4 anteroposterior diameter of occlusal surface 24 4.33±.0G .:;i±.04 7.1fi±1.03 maximum diameter of crown 12 4.45±.09 ..'^Ort.Ofi (•..74±1.3S Total, all specimens 37 4.38±.0r. .32rt:.04 7.31:1: .8.", widtli hvpolophid diameter of occlusal surface 14 3.24±.09 .32zt.0G 9.88±1.8< maximum diameter of crown 16 3.46±.12 .47±.08 13.58±2.4() Total, all specimens 34 3.38±.07 .42di.05 12.43±1.51 Ml width hvpolophid diameter of occlusal surface 10 3.07±.04 .12±.03 3.91± .8< maximum diameter of crown 17 3.12±.08 .32±.05 10.26±1.76 Total, all specimens 29 3.10±.0r) .26±.03 8.39±1.10 Ma width metalophid __ ,„ diameter of occlusal surface 10 3.00±.00 .19±.04 6.33±1.42 maximum diameter of crown 10 3.10±.0S .26±.06 8.23±1.84 Total, all specimens 24 .•!.ll±.or. .24±.03 7.72±1.11 Statistics not run on specimens where measurement was ' ' greatest extra-alveolar diameter, ' ' since in all cases there were fewer than 10 specimens. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 353 5£) ■i- •}- O C] t^ 00 CO o •i— 00 Ci CI CO •i— 1—1 I— 1 o in O O T-t 05 Oi O •i— O t)< rH (M 'i<< M ■^ CO tH CO ■* CO •* -* OJ (M CO CO CO IM (M CO a. <4J o tM ■^ c a • hj ^ Cfi f^^ eO O) CO lO Oi IC (MO t^ >— tCO(M^HCOi— |CC|QOtJ< i-H rH r-i ,-H "^ r-I r-i "^ rH > +1 +1 +1 +1 +1 +1 II +1 +1 'X!O-*0iC;rHir5O t>^o6wt>^o6o6cooooo»-05oio -t<^'-OCO-^COCOCOTtra «3 I— I t— I CO n t^ «o m OOOOOOOOOOOOOO +1 +1 +1 II +i +1 II +1 +1 +1 +i +i +i +i CO 00 in 00 in CO o in o -^ 00 -H i-< m ^ 00 Ci 00 cj c] o o ir: in CI 1— I ^ o~- Oi -f tOCOmOClOO'^ClOirHt^OCiirJcO'— ICOCli— iT-HCOCOCOOi-ll^i— ICit- A 00 „ _j» v^ W>aJ "i*! ^^ V*'i v^ >^^ ^' l^ J (^4 T— 1 I.-" ^^ (.fri jr^ W'.J ' — I *'W '^^* ' — I ' — I W'^ I.'*; l.i.' ^-^ I — I I™ 1 — I ^^ "- ~ C'J C 1— icocococo'cocococococici5cicii--J'HT)5cococococococococococic5rH-^c^^ o +^ OS ■r cic;.— lOcioit^tDcioocoOit^ooincit^cooocicotocot^coi^t^oinooo'-C ?^ o) ^ cj i>- 00 CO ca CO CO to in ci ci ci Oi C] ^ Oi in C] oo i~ '-o ?D t^ c^ to to CI o CO t^ CO cs C ^i-H -^ CO tJh CO TjH CO CO Tj! CO CO -^ CO c] ci 1— I Tji co" Tt5 co' CO CO CO co' CO CO CO ;. = to o t^ oi CO t^ rH CO in to 00 to CI in OS ai bL CO ■*! c] ■'ti to to o oi ^^ o o to c'] o in S*, — ii g^^' r-i^ 00 o in o !>. 00 ci oi in 00 c] -^ t>- ^ to Ln o to t- ■*_ o o o to t^ tjh oi •>-. i>: ca ^ — ^ CO ci CO ci ci ci ci co' ci ci ci ci a rA i— i co' ci ci ci ci ci ci co ci ci ci ci t-h i-J ■•^' ci tr ^'t^oinmtOTjt-^caoooiocoococoL' '^ OJ.-ll-ll-H.-lrHrHi-l,-l t^ O -^ O 1— ( 05 d CI •* CJ TjH Tt< CO in C3 CO ?, CO CI CO CI CI OJ CJ C\l rH CI CI 1— I r-l r-l g O o o o ?? o o *-^ O O o O C o C .Ji ^ o .ii ^ o .ij ^ o g ^ o .i; 2 -ii rt c c: O^^tiQJMSjitc;- o ^ ^ o C3 So?: p ^* ^' C ^^ ^"^ p ^"^ ^ o ^ ^ p > ' •*-* -^ S'^'*-' C'-*-^"^ ^"*-'"*-' O >-" H ^ & CT ^ -^ ^ "k j: fe 5 -3 ^3 rO 'Ti -r t- ^^^ ^-^-^ i.-^-? i,-5-5 ^ ° o o .2 o ^-2 5" o .2 o 0.2 K O ^ K C 03 'S a O o « ^iL-t^ a^L-i- ^^-^ di^-^ c--ii ^ -^ c< p'-*-'?'-'-^? -T^p ^_rP > p «"c, E!'-*^-*- ^-^-*-- ^^^ r^^ 7\ ^ s^c c: E^-gS-g-S P !a % - r^ o-Q. c>i a S3 354 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Diagnosis. Larg-est known species of the genus (Tables 3-4) ; anteroconid of P4 a rounded enlargement in middle of anterolo- phid when unworn. In addition, several possible differences based on one or two individuals have been mentioned above in the description of the genus. Ameghino's reference of two specimens of this species to Cepha- lomys prorsus (a synonym of C. plexus) was probably a lapsus. The specimens in question — M.A.C.N. no. A 52-94, a mandible, and M.A.C.N. no. A 52-95, an isolated P4 and dm4 — fall within the size range of C. arcidens, and this is evident not only from our very large series but also from the much smaller one that was available to Ameghino. M.A.C.N. no. A 52-94 is, in fact, approxi- mately the same size as the lectotype of C. arcidens. "Orchiomys prostans" was based on a young individual of C. arcidens, in which Ml is at stage 3 and M2 at stage 2. The lectotype of C. arcidens happened to be an old individual with Mi at stage 6 and M2 at stage 5. Ameghino did not visualize the differences result- ing from wear; his identification of Mi_2 of ''0. prostans" as P4-M1 may indeed have prevented him from doing so. This is the most abundant Deseadan mammal (Loomis, 1914, p. 189), being represented by over 100 individuals, counting isolated teeth, in the collections we have studied. In the unworn P4, the anteroconid is connected primarily with the middle or the lingual part of the metalophid. This may be its only connec- tion (A. CM. no. 3161, Fig. 19A and no. 3162), or there may be another one with the protoconid (A. CM. no. 3108, Fig. 18A). The enamel, as already pointed out, is interrupted on the anterior face of the lower molars (Fig. 19D) and the buccal and posterior faces of the upper molars after considerable M'ear. There do not appear to be any other interruptions. Cephalomys plexus Ameghino 1897 Figs. 14D, 15 A, 16E-F and 22A Cephalomys plexus Ameghino 1897c, p. 494; 1906, p. 421, Fig. 314; Loomis, 1914, pp. 190-191, Figs. 120-122; Stehlin and Schaub, 1951, Fig. 76, p. 61 ; Schaub, in Stehlin and Schaub, 1951, p. 245, Fig. 394. Cephalomys prorsus Ameghino, 1899, p. 560, Fig. 6; 1902b, p. 37; 1902d, p. 425, Fig. 6; 1903a, p. 82, Fig. 1, p. 96, Fig. 12, pp. 122-123, Figs. 41-42; Schaub, in Stehlin and Schaub, 1951, Fig. 395, p. 245. Cephalomys prosus [sic] Loomis 1914, pp. 191-192, Figs. 123-124 (lapsus for C. prorsus Ameghino). WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 355 Asteromys prospicuus Ames^hino 1897c, p. 495. Asteromy.s annectens AnieKliino 1902h, p. 37. Syntupes. M.A.C.N. no. A 52-99, portion of right maxillary with P*-M''; M.A.C.N. no. A 52-100. portion of left maxillary with P^-^P ; M.A.C.N. no. A 52-101, portion of rjoht mandible with P4-M0; M.A.C.N. no. A 52-102, portion of left mandible with P4-M2. M.A.C.N. no. A 52-99, the only one of the series subse- quently figured by Ameghino, is here designated as the lectotype. Types of synonyms. Cephalomys prorsus. M.A.C.N. no. A 52- 103, portion of right mandible with P4 not yet fully erupted and Mi_3.''* Asteromys prospicuus. M.A.C.N. no. A 52-107, four iso- lated upper molars, one LM^, one LM^°''-, one KM^'"'- and one fragment. LM^, the only tooth that actually comes within the size range given by Ameghino, is hereby designated as the lecto- type. Asteromys annedens. M.A.C.N. no. A 52-108, portion of the left mandible with I, P4-M3, and M.A.C.N. no. A 52-109, palate with R and L P*-M'^; M.A.C.N. no. A 52-108 is here desig- nated as the leetotype. Hypodigm. The types listed above and a series of good addi- tional specimens, three in the Ameghino collection, seventeen in Amherst, one in the Museum National d'Histoire Naturelle, one in Chicago Natural History Museum, two in The American Museum of Natural History and four in the Yale Peabody Museum. As in the case of C. arcidens, there are, in addition, numerous isolated teeth. Horizon and localities. Deseadan. The Museum National specimens are from La Flecha, Santa Cruz ; those in the Amherst, Chicago, Yale and American Museum collections are from Cabeza Blanca, Chubut. Again, as in the case of C. arcidens, it is prob- able that Ameghino 's specimens were obtained at Cabeza Blanca. Diagnosis. Approximately tw^o-thirds the size of C. arcidens (Table 5) ; apex of anteroconid of P4 distinct from anterolophid when unworn. Cephalomys "prorsns" was distinguished from C. plexus on the basis of slightl}- smaller size and on the following structural 19 C. prorsus was formally proposed as new by Auicghino in 1902b, p. 37, and the diagnosis there given was based on this specimen and on M.A.C.N. nos. A 52-104 and A 52-10."'>. Previously, however, he liiid described and tigured M3 of M.A.C.N. no. A 52-10?, under this name (18!»!», p. 500, Fig. 0). The earlier account is sufficient to date the name from ISO'.I and to Hx the tvpe as M.A.C.N. no. A 52-103. 356 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY i_:>5 o o oj c^ o 00 o t~ CI w (M ©q T— 1 00 O i-H r-l »0 St. Si. 5i Si. O a si o o H CO cq +1 rH CD o" 00 rt lO CO I— I ci +1+1 tH CO o oo 00* o i-i in Ol l-H +1+1 t^ oo Oi p oo' o CO t~ iH oi +1+1 '-0 00 eg "-H lo CO ■* lO r-H t^ + 1 CO l-H eg l-H rH +1 +1 +1 lo o CO CO to Ol sg th +1+1 00 to ca o cn I-H l» «> O 00 50 -^lO-^inMin ■^io=ort■ 03 T). CO p c5 th p CI -* CI 05 rH o lo Oi eg eg -^ CO eg CO CO CO rH rH «>• irs CI q oo' ci ci CI ci cd ci ci ci ci r-I ci rH o cj ci ci ej ci ci eg' ci ci eg' eg' tA rA tA c "-H aj o 00 "S .— OOO':t;0i'*C]TtHC0^iq^rHt>-0000Ol0C0kOC0piOC0 aS j;; ^ci c-i ci co ci ei ei ci ci ci ci ci rH ^^ co ci co ci ci ci ei ci ci ci ci rn rH r-! "£ S; 5 o o o c'l e'l t^ ci co t-^ oo c'l cV-© 'o >n -^i rH -4* o o o lo o oo "-i c'j ^i o u ^"jjLOrHprHOOOt-OOOOt^OO^O-^t-OOpOlOOCgoCOOt^^CO-^rH O C t^ C^i rH ci rA rA tA ci ci r-J rH rH >— ' OS ci 1-i tA ci rH rA Og ci ci rH rH rH rH rH £ c A^ ;7TtlO'OC5COO]t^rHrH:OrHrH'n-*Clrt*Oi'?-lt^egOo'-'^'-HrHt^OO^C '^rH r-i i-t T-\ T^ T-\ T-t ^^ t^ ,-i ,-t ,-< r-\ %V t- CS rt a; ■^ "^ rc ro •^ r^ '^r 'C o f 1 o ,2 o o ^ o ri o "ft "5 o *r-( .|-» o "ft "a o & ft 5 '-H •r-* o o •rl O O •- _o ^ •H. O O a 2 2 -2 rt 2 i 2 rt ?t3 O'rt ^-^ (-H ^ • c Oj ■2 £=" OS f lis i-( CI n ^_| CI «. ■H^ eiH^ wood and patterson : oligocene rodents op^ patagonia 357 Table 6 Tooth measurements (in mm.) of Cepholomys sp., A. CM. no. 3109 Lrjt IiU)ht P4-M3 11.0 P4 anteroposterior 2.42 2.40 width, protoloph 2.53 2.27 width, nietalopli 2.86 2.51 Ml anteroposterior 2.33 2.42 width, protoloph i\99 i!.89 width, metaloph 2.33 2.47 M? anteroposterior 2.39 2.55 width, protoloph 2.82 2.78 width, metaloph 2.37 2.44 M? anteroposterior 3.10 3.19 width, protoloph 2.89 2.76 width, metaloph 2.09 2.33 characters : 1 ) anterior lobe of lower molars more compressed anteroposteriorly and more i)ointed at the extremities, 2) pos- terior lobe of lower molars with a more convex posterior face, and 3) upper molars with small, isolated enamel fossettes. The difference in size is not significant. The first two features are due in part to age and in part to individual variation ; speci- mens structurally intermediate exist, even in the Ameghino Collection. Item 3) has no significance whatever, being entirely an age character; these fossettes may also be seen in Ameghino 's syntype series of C. plexus. " Aster omys prospicnus" was not, of course, contrasted with C. plexus by Ameghino, and the only character he gave to sep- arate it from Asteromys punctus was the small size of the upper molars, 1.6 to 1.8 mm. in length. Since ^i. punrtvs was based on a mandible, no real basis for comparison existed. The syn- types of ^'A. prospicuus" are upper molars of Cephalomys at stage 5. Their sizes (length of LM-^ 1.8, of LM^ - 2.0, and of jj]y/[ior2 2.4 mm.) are within the range of C. plcxiis and there can be no doubt as to the synonymy."*^ -f In the same labeled box with the molars of .1. itrospicitns was another upper tooth, M.A.C.N. no. A 52-87 (described here on p. 376), that Is very different from the syntypes and does show some resemblance in crown structure to the lowei' molars of A. punctus, although much lower-crowned. This tooth may have decided Ameghino that lie was dealing with an Axti rDtniix, but it cannot lie selected as the lectotype and the dtlier nudars excluded from the species. It is over ;i mm. in length, and this is too great a ditt'erence from tlie measurements given by Ameghino to be attributed to the cruder measuring devices and optical aids available in 1897. 358 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY "Asteromys annectens" was described as l)eing transitional be- tween Asteromys and Cephalomys. In reality, the syntypes are indistinguisha])le from C. plc.rus either in size or in structure. In the lectotype mandible, M1-2 are at stage 4, and M^ at stage 2. P* is at stage 2, M^- at stage 5, and M'^ at stage 4. The differences, other than size, separating C. plexus from C. arcidcns are difficult to detect. The distinction in the antero- conid of P4 can be seen only in unworn teeth. With wear, there are interruptions of the enamel at the posterointernal corners of the lower molars, which occur before the interruptions on the anterior face (Fig. 19F). In the upper cheek teeth there is some variation. In most specimens, the enamel is interrupted first along the buccal side and subsequently along the posterior face, as in the anomalous specimen, A. CM. no. 3109 (Fig. 14C). In one specimen, A. CM. no. 3085 (Fig. 14D), which is at the lower limit of the size range, the interruption appears to take place at a much later stage in the wear, as indicated by the difference in pattern of P^. In this specimen, however, the enamel along the buccal and posterior margins of the teetli is considerably thinner than elsewhere. LiTODONTOMYs Looniis 1914 Litodontomys Loomis 1914, p. 193. 'Type species. L. chuhutensis Loomis 1914. Distribution. Deseadan, Patagonia. Emended diagnosis. Cheek teeth high-crowned, cement present in flexids, enamel not extending to bases of teeth on anterior and lingual faces ; P4 not molariform, elongate ; para- and metaflexids shallow (obliterated by wear on only known specimen), meso- flexid not as deep as hypoflexid, converting to fossettid with deep wear. Litodontomys chubutensis Loomis 1914 Fig. 23 L. chuhutensis Loomis 1914, p. 194, Fig. 127. Type. A. CM. no. 3086, fragmentary right mandible with I, P4-M:,. Jlypodigm. Type only. Horizon and locality. Deseadan, Cabeza Blanca, C'hubut. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 359 Diagnosis. As for the genus ; for measurements see Table 7. The teeth are evidently not typically brachyodont, as Loomis believed, but are fairly well along the road toward high crowns. They appear low, however, due to the excessive wear that they have undergone. This interpretation is based on the fact that cement is present, that the enamel is interrupted on the anterior and lingual sides of the teeth, and that there is a con- siderable difference in the crown height of Mi and P4, on the one hand, and M3, on the other. In many respects, this form appears to us to be one of the more specialized of the Deseadan i-odents. The hypoflexids are directed just posterad of the meso- flexids. A B Fig. 23. Litodontomys cJnibutensis Loomis, type, A.C.M. no. 3086, x 10. A, RP4-M:j; B, cross-section of EIi, anterior view. The premolar has what appears to be a simple talonid (al- though a shallow metaflexid or -fossettid may have been present originally) connected with an anteroposteriorly expanded trigo- nid by an ectolophid that widens near its anterior end (Fig. 23A). The mesoflexid is considerably smaller than the hypo- flexid, both trigonid and talonid sending arms along the lingual margin of the tooth, which meet, closing off a cement-filled meso- fossettid at the present stage of wear. The hypoflexid is much 360 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY larger, and extends low down on the side of the tooth, as a broad, open emliayment. It is partially blocked by a low cingnlum, and is partially filled with cement. Enamel is absent over the anterior face of the tooth and from the lingual surface of the talonid, in both eases obviously due to original non-deposition near the base of a high-crowned tooth (Wood, 1940, pp. 356-357), and not due to interdental wear (H. E. Wood, 1938). The free end of the enamel plate on the lingual margin of the tooth apparently is embedded in the dentine. The molars are made up of two nearly equal lobes, although the talonid is somewhat the larger in each case. The flexids are of more nearly the same size than in P4, though the hypoflexids are somewhat larger than the mesoflexids in each case. In Mi, both flexids are nearly filled with cement. The mesoflexid is closed, as in P4, by arms from the internal cusps, whereas the hypoflexid appears to slope to the base of the tooth, being filled with cement. Enamel is absent along the anterior and lingual faces of the tooth and the lingual part of the posterior face. The mesoflexid of M2 appears to be of about the same depth as that of Ml of this specimen. Since the tooth is less worn than Ml, this means that the mesoflexid was actually shallower. The hypoflexid is quite deep. The lingual fold extends as a groove down the side of the tooth, the cingulum closure not being quite at the lingual margin of the valley. Enamel is missing from that part opposite the enamel-bearing portion of Mi and is pres- ent on that part opposite the enamel-free section. It is also missing on part of the lingual border of the talonid, as in geomyoids (Wood, 1937b). A peculiar feature is the apparent encirclement of the free end of the posterior enamel plate by dentine. M3 is more elongate and narrower than the other molars, so that the two flexids are wider (especially at their open ends) than in the anterior molars. This is partly due to the lesser amount of wear on this tooth. The enamel has only just been interrupted on the anterior face of the tooth, but is broadly absent on the lingual face of the talonid. The lingual side of the tooth is grooved, as in M2, but the mesoflexid does not seem to be dammed. This is not certain, however, due to the cement filling. The hypoflexid contains little cement, and is not dammed WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 361 except at the very base of the crown, where there is a prominent cuspule. The absence of cement here is very possibly due to its removal during preparation. Loomis' statement (1014, p. 193) that the folds are narrowest at the margin of the teeth and expand toward the center is true of the mesoflexids, due to their having been dammed by the cingular outgrowths of the trigonid and talonid, but is definitely not true of the hypoflexids. A section of the right lower incisor is associated with this specimen as well as a fragment of the mandible. The tooth is egg-shaped in cross section, with a round, wide, enamel-covered anterior face and a tapering posterior side (Fig. 23B). The lingual margin is nearly straight. The enamel extends about a third of the way around the lateral surface and about a fifth of the distance around the lingual face. The enamel is smooth, and is rather thin. The pulp cavity is very small and is essentially circular near the tip of the tooth. The tooth is quite long, with a large radius of curvature. The affinities of Litodontomys appear to be with the Dasyproc- tidae, but its position within the family is uncertain. There are some similarities to Cephalomys in the cheek teeth and in the incisor, and also in crown height and the tendency toward non- deposition of enamel, but there are also differences. The pre- molar outlines are very different, for example, and Cephalomys lacks cement. There are resemblances to the type material of Olenopsis^^ that may indicate some degree of relationship. P4 of 0. uncinus agrees in general outline, and, what is more im- portant, in the presence of cement at the base of the crown and in the hypoflexid. The mesoflexid has been converted into a mesofossettid, very possibly in much the same manner as in Litodontomys. It is not likely that the two forms stood in an ancestor-descendant relationship. Olenopsis uncinus has a more 21 The type species, 0. uncinus Ameghiiio from the Santa Cruz, was based on three syntypes : M.A.C.N. no. A 17, a fragmentary left mandible with the base of I, dm4-M2 (M2 unenipted) ; M.A.C.N. no. A 1613, an isolated RP4 ; and M.A C.N. no. A 1614, an isolated Ldm-i (Amoghino 1889, pp. 145-146. PI. 6, figs. 14-16). M.A.C.N. no. A 17 is unfortunately not to be found in the Ameghino Collection. M.A.C.N. no. A 1613 has a deposit of cement. M.A.C.N. no. A 1614 lacks all trace of this substance and agrees very closely with dm* of Neorcomj/s, to which it may well be referable. BMelds (1957, p. 325), in his description of the material he identifies as Olenopsis aeguatorialis and places in the Dinomyidae, states that there is no evidence of cement in the teeth of that form, which raises the question as to whether or not this species is correctly referable to Olenopsis. 362 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY persistent crown pattern and the enamel was more extensive and differently distributed. In addition to the metafossettid, there are three small anterior fossettids, possibly remnants of a larg:e anterior fossettid with an irregular floor or with spurs of enamel projecting into it, and a long, narrow metafossettid. The enamel extends down to the roots everywhere except along the posterior two-thirds of the lingual face, whereas in Litodontomys it is in- complete anteriorl}^ as well. The two genera may or may not have had a common, pre-Deseadan ancestry, but the resemblances between them at least suggest some degree of affinity. There are no particular resemblances between Litodontomys and the ma- terial Fields described as OJenopsis aequatorialis. Table 7 Tootli measurements (^in mm.) of Litodonto)nys chuhutensis, A.C.M. no. 3086 P4 Ml anteroposterior 3.21 M2 anteroposterior 2.21 width, metalophid 1.21* width, metalophid 2.08* width, hypolophid 1.90* width, hypolophid 2.03* anteroposterior 1.97 M:- anteroposterior 2.27 width, metalophid 1.85* width, metalophid 1.71* width, hypolophid 2.09* width, hypolophid 1.59* anteroposterior 1.88 transverse 1.44 VDasyproctidae gen. et sp. indet. Fig. 24 M.A.C.N. no. A 52-113, an isolated, unworn upper cheek tooth, presumably from Cabeza Blanca, Chubut. This tooth shows a very high degree of unilateral hypsodonty, comparable to that of Cephalomys. It differs from that genus and from Litodontomys in the distribution of the enamel, which persists evenly all around the tooth as far as the beginning of the roots. Fields' figures of the Colombian material referred by him to Sclcromys suggest a similar distribution of enamel (Fields, 1957, Figs. 10, 14). The protocone and hypocone are approxi- mately subequal and sharply angulate internally. They are separated by a prominent hypoflexus that progressively dimin- ishes and disappears about half way down the crown. The WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 368 protocone-anteroloph crest curves forward and outward, losing altitude progressively, so that the extremity is far below the level of the apex of the paracone. The paraflexus deepens lingually, and would eonvei-t to a long, narrow parafossette. Frotoloph and mure form a curving crest connecting the pai-acone and liypocone. Opposite the protocone, there is a short blunt projec- tion that, with wear, would unite the two crests at this point. The paracone is joined to the external part of the posteroloph by a gently curving crest that bears two minute cuspular elevations. Frotoloph, mure, posteroloph and this external crest isolate a large, deep fossette. From the external crest, opposite the second cuspular elevation, a prominent spur projects into this valley. 24. fDasyproetidae, gen. ct sp. indet. Left upper cheek tooth, T A T O 1 1 9 ,. « Fig M. A.C.N, no. A 52-113, x 6 A slight swelling is present on the outer surface of the mure. The region of the hypocone is the highest part of the unworn crown. No trace of- cement is present. We are in some doubt as to whether this tooth should be in- terpreted as P* or as a molar. It agrees in outline with P^ of Santacruziau SeJe)-omys as figured by Scott, but not with Fields' (1957) referred material. It differs from both in the presence of the large spur in the posterior valley. This structure appears to be a remnant of a metaloph, the swelling on the mure repre- .senting the aliorted inner end of the crest. So molariform a premolar would be decidedly exceptional in the Deseadan. If the spur and swelling be regarded as a metaloph in process of formation, then such a mode of origin would be exceptional for P* among caviomorphs. Agreement with the upper molars of Santacruziau Scleromys is close. These teeth lack the metaloph entirely, and, since this is clearly a secondary condition, the 364 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY structure in M.A.C.N. no. A 52-113 could be regarded as indi- cating a molar in process of losing this crest. This we believe to be the more likely interpretation. It is supported by the structure of the only known Colhuelmapian dasyproctid, also an isolated upper molar (M.A.C.N. no. A 52-163)," in which the metaloph is interrupted by a deep notch in its outer half. We believe that this specimen is probably a dasyproctid. It is clearly distinct from Cephalomys and there is no basis for reference to Litodontomys. As regards the even distribution of enamel on the crown, it is more normal, with respect to the later members of the family, than either of its contemporaries. The tendency toward loss of the metaloph suggests affinities with the Santacruzian Scleromys rather than with Neoreomys, in which complete loss of this crest occurs only on M^, where it is represented merely by a small metacone behind the paracone (Scott, 1905, PI. 65, figs. 1-2). M.A.C.N. no. A 52-113 shows no indication of a neoloph, however. This structure does occur in the Santacruzian Scleromys, as is revealed by a minute remnant of the floor of the neofossette in M^ of M.A.C.N. no. A 10142 and by the presence of a neofossette in M^ of M.A.C.N. no. A 4361. (The isolated Colhuehuapian molar, in which the neoloph and neofossette are fully developed and the metaloph is in process of reduction, demonstrates that these are neofossettes and not metafossettes.) If M.A.C.N. no. A 52-113 was in the ancestral line leading to the Santacruzian Scleromys, then a neoloph must have arisen in post-Deseadan time. This is con- ceivable, but more and better material is clearly needed before any conclusion can be reached. Table 8 Tooth measurements (in mm.) of M.A.C.N. no. A 52-113 Anteroposterior 3.10t Greatest width 3.83 1 Height, buccal side 2.80 Height, lingual side over 6.4 t = Greatest extra-alveolar diameter. 22 This tooth is one of Ameghino's two syntypes of Luantus initialis (see below, p. 365). WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 365 Superfaniily CAVIOIDEA L. KraglieYich 1930 Family BOCARDIIDAE Ameghino 1891 Two cavioid genera, Asteromys Ameghino and Ckuhuiomys gen. nov., that differ rather strikingly from each other, occur in the Deseadan. So different are they that we once thought that Chubutomijs should be considered as an aberrant dasyproctid. Study of the Colhuehuapian and Santacruzian material in the Ameghino Collection, however, has revealed that there are two clear-cut groups within the Eocardiidae, one typified by Luan- tus,^^ the other by Eocardia and Schistomys. Asteromys is clearly referable to the first of these, and Chuhntomys is as certainly an early representative of the second; both these genera possess small, ephemeral mesofossetids, a typical eocardiid character. The two groups may be distinguished as follows : LuANTiNAE subfam. nov. Cheek teeth mesodont to hypsodont; hypoflexus and hypoflexid extending approximately half way across crowns ; external f ossettes and internal f ossettids relatively persistent; metalophid lost on M2-3 of later forms; enamel ex- tending equally far down all sides of cheek teeth ; no cement. Distribution. Deseadan to Santacruzian, Patagonia. Oenera. Asteromys Ameghino 1897, Deseadan ; Luantus Ame- ghino 1899, Colhuehuapian to Santacruzian. 23 Ameghino, subsequent to his description of the Colhuehuapian L. initialis (1902e. pp. 114-115), consistently referred Luantus to the Capromyidae (Myo- castoridae of his usage). L. initialis was based on two syntypes, a portion of a left ramus with the base of the incisor and P4-M2 (M.A.C.N. no. A 52-164), and an isolated right upper molar (M.A.C.N. no. A 52-lOH) — not a premolar, as he supposed. The latter represents a hitherto unknown dasyproctid and as such has been briefly referred to above (p. 364), but it clearly has nothing to do with the jaw fragment, which we designate as the lectotype of the species. The two specimens were presumably picked up and packed together, a circumstance that Ameghino all too frequently considered to be sufHcient proof of natural associa- tion. There is a general resemblance between the lower molars of the Santa- cruzian species of Scleromys and those of Luantus, but the latter retains the metalophid in Mi and has a short, typically eocardiid incisor that does not extend back beyond M2. Scott (1905, p. 388), evidently on the basis of the lectotype jaw, placed 'Luantus witli Eocardia and Sichistomys, an assignment followed by Simpson (1945, p. 94) and Landry (1957a, p. 43). There can be no doubt that this is the correct position of the genus, although we cannot agree with Landry that "a form only a little less specialized than Luantus . . . may have been the common ancestor of the Superfamilies Octodonoidea, Cavioidea, and possibly Chinchilloidea" (op. cit., p. 44). The Deseadan Asteromys is such a less specialized form, and it is unmistakably a cavioid. Schaub (in Stehlin and Schaub 1951, p. 369) considers that Luantus is widely separated from the Eocardiidae, and should be referred to the "Acaremyinae" (1953a. p. 399) because the second and third inner "synclinals" of the lower cheek teeth are united. This condition, wliich is due to loss of the metalophid in M2-3, we cite as a character of the new subfamily Luantinae. We, too, regard it as a character of importance, though we cannot agree that it requires the removal of Luantus from the Eocardiidae. 366 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY EocARDiiNAE. Cheek teeth hyiisodont to hypselodont ; hypo- flexus and hypoflexid, after slight wear, extending more than half way across crowns ; external fossettes and internal f ossettids becoming shallow and ephemeral ; metalophid retained ; enamel of crowns not extending equally far down all sides of cheek teeth, and lacking, after slight wear, on external sides and antero- and posteroexternal corners of uppers, on internal sides and antero- and posterointernal corners of lowers ; cement present in some later forms on sides of crowns and in hypoflexus and hypo- flexid. Distribution. Deseadan to Santacruzian, Patagonia. Genera. Chuhutomys gen. nov., Deseadan; Eocardia, Schisto- mys, Phanoniys'^ Ameghino 1887, Santacruzian. Both in the luantines and in the eocardiines with rooted teeth, the hypoflexus and hypoflexid decrease in extent toward the base of the crown and disappear entirely before the root is reached. The known luantines may well have formed a direct phyletic series. Luantus has not hitherto been recorded from the Santa Cruz proper, but there are from this horizon two good speci- mens, a right mandible (M.A.C.N. no. A 2018) and a palate (M.A.C.N. no. A 2025), and a number of isolated teeth in the Ameghino Collection that are certainly referable to the type species, L. prophcticus Ameghino 1898.""' The type specimen of this species was found in the Pinturas, or AsfrapotJtericuhis, beds, supposed bj^ Ameghino to be older than the Santa Cruz. Examination of the scanty, and fragmentary, Pinturas fauna reveals, however, that there is really no good evidence for coii- 24 This is a very poorly knowu genus. The syntypes of P. midtus, tlie type species, are not now in the collections of the Museo de La Plata, and must he presumed lost. Material in the Ameghino Collection described later by Amegliiiio (18S9, p. 217, PI. 10, tigs. 12-25) agrees well with the brief original description. Unfortunately, these "neosyntypes"' (M.A.C.N. no. A 2022) are all isolated teeth, as are the syntypes of P. vctiilus Ameghino (M.A.C.N. no. A 2024). The best specimen o} P. mixtus, the maxillary fragment with r4-M2 figured in 18S!» (PI. 10, fig. 21) and in 1906 (Pig. SI.'!), cannot be found in the collection. The available material shows that Phanomys is a valid form with rooted molars but with cement ; hypoflexid and distribution of enamel are as in Eocardia and Schistomys. Hcdymys, also described by Ameghino in 1887, is a nomcn ratiiim (see below). 25 P4 and M3 of M.A.C.N. no. A 2018 are so similar to the types of Ameghino's Eocardia itriaca and Luantus pro/jhrticus as to show that these species were based on KM-^ and RP4. respectively, of one species, perliaps even of one indi- vidual. The former has line priority, which we disregard. Luantus prophcticus was described by Ameghino as the type species of the genus, and the fact thai Eocardia prisca was described two lines above would liave been an altogetlier inadequate reason for violating common sense and presenting a new combination /{. prisca is here relegated to the synonymy of L. prophcticus. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 3(i7 sidering it to be anything but a Sautacruzian local fauna. The palate, M.A.C.N. no. A 2025, a young specimen with dm^-M'^ pre- served, reveals a very interesting feature of Luantiis that must be recorded liere. M- has a swelling in the center of the postero- loph, which on ]\P becomes a minute fossette, partially open posteriorly on the left side. This is a rudimentary neofossette, the first stage in the formation of a neoloph such as occurs in various caviomorphs and it is of tlie greatest interest to find a tendency toward this condition occurring in the family. This specimen has been figured by Scott (1905, PI. 68, fig. 27) and by Ameghino (1906, Fig. 311, p. 419). The former figure is very inaccurate, and shows neitlier the swelling nor the minute fos- sette; the latter shows the swelling but not the minute fossette and errs in depicting the enamel as missing from the outer face of JVP. Scott called the specimen '^ Eocardia sp. (referred to TIcdymys integrus)'^ while Ameghino labeled it "Hediniys in- tegrus," and thereby hangs another lengthy footnote."*' -^ HediinniH integrus (Ameghino, 1SS7, p. 14; 1S89, PP- 217-218 — the uujusti- fled ememlation to Hedlmys dates from 1804, p. 331) was based on a single upper molar, measuring 3 mm. in length and breadth, with a large internal and two small external r((ots, a deep hypoflexus and no enamel on the external face. The tooth was evidently only moderately worn since it measured G mm. high on the internal side and 4 mm. on the external. The type is not now in the INIuseo de La I'lata and is almost certainly lost. From the description, it is clear that the specimen was au eocardiine, but beyond that it is impossible to go. Jledymys hitegrun Ameghino 1887 is a nomcn raninn in the fullest sense of that useful term. Scott (1905. p. 472) stated that in his opinion the type was a milk molar of Eocardia, but there is no real evidence for this view. In the Ameghino Collection, however, the palate, M.A.C.N. no. A 2025 is labeled as "Hedlmys intcyrus tipo.'' This label, and other.s like it purporting to identify the types of species described in 1887, has an interesting history. After Ameghiuo"s resigna- tion from the statf of the Museo de La Plata in 1888, he was denied access to the collections by .Moreno, the Director. Consequently, he had no opportunity to compare Sautacruzian specimens obtained at a later date with those he had described in his paper of 1887. It then became his custom to label as types (and his concept of a type was not that now current) specimens in his own collection that he had referred, without benefit of direct comparison, to species ilescribed in that publication, frequently selecting nuiterial figured in his great work of 1889. The present specimen was not there figured, and it is difiicnlr to understand why he made the identification, liecause the teelh considerably exceed the measurements he gave for H. inteynm. How thoroughly Ameghino had come to base his conception of }Iedymyn on this newly estalilisheil "type'" is shown by his subsequent description of 'Hedimys" yruiilis (1900, Fig. 312, p. 420 — the figure caption is the only mention in his writings of this species although it is not there stated to be new). "//." gracilis was based on two isolateil cheek teeth, M.A.C.N. no. A 4481, an upper nujlar (designated P* by Ameghino) and dnij. The upper molar is here designated as the lectotype ; the dm4 is probaldy referable to Eocardia. The molar is smaller than any of the cheek teeth of the new "type" of //. integrus and agrees with them in pattern, allowing for the difference in wear, but the measurements (L. 3.0, W. 2..") mm.) do not suffice to separate it from the real type (now lost) of //. iiiteyrns. The degre<' of wear is almost exactly the same as in this lost specimen (external height 0.0, internal 3.3 mm.), however, and this reveals that "//." gracilis is not //. integrus, since Ameghino's description of the latter stresses the absence of enamel on the outer face and uuiUes no mention of fossettes on the grinding surface, whereas there is enamel and there are fossettes on the lectotype of gracilis. The species must be listed as Luantiis gracilis Ameghino. 368 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY The absence of eocardiines in the Colhuehuapian is certainly an accident of collecting; the family is very rare prior to the Santa Cruz, being represented by only three identifiable speci- mens from the earlier deposits. The Colhuehuapian Archaeocar- dia is not an eocardiid. The two proposed species, "Palaeocar- dia" mater Ameghino 1902c and Archaeocardia mustersiana Ameghino 1904, both fall into the synonymy of the echimyid Protacaremys prior Ameghino 1902 (Patterson and Kraglievich ms.)- Simpson (1945, p. 94) inadvertently credits ^rc/iaeocarcZia to Ameghino. The name was proposed by Cossmann in 1902"^ to SANTACRUZIAN LUANTUS PHANOMYS EOCARDIA SCHISTOMYS \ ■ ' ' COLHUEHUAPIAN LUANTUS DESEADAN ASTEROMYS CHUBUTOMYS OCTODONTIDAE Fig. 25. Tentative pliylogeny of the Eocardiidae. replace Palaeocardia Ameghino 1902 nee Hall 1867. The hypselo- dont Eocardia and Schistomys, both adequately known from Scott's work, are the only common eocardiids. Both forms, and also the rooted Phanomys, may well have been derived from Chuhutomys. The phylogeny of the family may be tentatively expressed as shown in Figure 25. Subfamily LUANTINAE AsTEROMYS Ameghino Asteromys Ameghino, 1897c, p. 495; 1898, p. 176. Looniis, 1914, pp. 194- 195 (in part). Stehlin and Schaub, 1951, p. 244. Type species. A. punctus Ameghino, 1897. Distribution. Deseadan, Patagonia. Emended diagnosis. Cheek teeth mesodont, considerably^ lower- crowned than in Luantus initialis; P4 not molariform, anterior -'■ In an unsigned footnote on page 17G of the Ilevue Critique de P.ileozoologie, sixigme ann6e. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 369 face with deep vertical groove, minute fossettid on side of eeto- lophid, mesoflexid large, widely open internally; metalophid present on molars; M2-3 not notably larger than P4-Mi. AsTEROMYS FUNCTUS Ameghino Fig. 26 A. punctus Ameghino, 1897e, p. 495; 1906, p. 412, Fig. 287. Stehlin and Sehaub, 1951, Fig. 390, p. 244. Tyi^e. M.A.C.N. no. A 52-110, fragment of right ramus with base of I, P4-M3, part of anterior root of dm4. Fig. 26. Asteromys functus Ameghino. RP4-M3, type, M.A.C.N. no. A 52-110. A, crown, B, labial view, x 5. Hypodigni. Type only. Horizon and locality. Deseado formation; evidently Cabeza Blanca, Chubut. Diagnosis. As for the genus ; for tooth measurements see Table 9. The cheek teeth are moderately high-crowned and exhibit a strong degree of unilateral hypsodonty. The enamel is con- tinuous all around the crowns. P4 differs somewhat from those 370 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY of other Deseadan rodents. It is less curved and hence exhibits a lesser degree of unilateral hypsodonty than the molars. The crown tapers toward the apex both transversely and antero- posteriorly; the maximum area of grinding surface would not be attained until wear had proceeded about half way down the crown. The upper portion of Mi curves forward over the obliquely sloping upper part of the posterior face of the premolar, so that with wear the relative proportions of the two teeth would change considerably (Fig. 26). The talonid of P4 is essentially molariform, the only difference being that the metaflexid is shallower; the hypolophid arises near the center of the tooth and is directed slightl}- posteriorly, thus suggesting that it may have arisen anterior to the primary posterior crest as in the specimen of Sciamys referred to above (p. 297). The trigonid is Yevy different from that of the molars. The anterior face is deeply grooved and the groove extends back between the proto- conid and metaconid, which thus appear as though situated at the ends of crests that diverge from the ectolophid. Immediately behind the point of divergence, there is a minute fossettid on the inner side of the ectolophid. The differences from the tri- gonid of P4 of Platypittamys are striking but apparently not profound. The divergent crests running to the protoconid and metaconid appear to be merely the two wings of the anterolophid, here deeply concave due to the anterior groove, and the inner wall of the minute fossettid is best interpreted as a rudimentary metalophid connected, again due to the presence of the groove, to the metaconid wing of the anteroloiDhid. The crown elements present in both thus appear to be comparable. In Luantus, the anterior groove has been eliminated, the short anterolophid is transverse and the metalophid is at first small and freely pro- jecting (L. initialis), later becoming larger and uniting with the inner extremity of either the anterolophid or the hypolophid {L. propheticiis). The hypoflexid extends about half way across the crown and persists nearly to the root. The mesofiexid is very large, wide above and tapering downward; it persists for about two-thirds of the height of the crown. The metaflexid is a mere nick in the rim of the shallow metafossettid basin. The molars are ver}- similar in size, a notable difference from Luantus, in which M2-3 are larger than P4-M1. The only apparent WOOD AND PATTERSON : OLIGOCENE RODENTS OP PATAGONIA 371 structural difference Avitliin the series is the greater depth of the metaflexid in M3, which is not entirely due to the difference in wear. The protocouid and hypoconid form sharp external anples and the lingual border is very straight in all. The metalopliid is present and complete on all molars, whereas in Luantns witialis it is interrupted on M2 (M3 unknown) and in L. proplieiicus absent on Mo..-,. The hypoflexid extends approximately half way across the crown, diminishing in size and ending near the roots. The metaflexid is the deepest of the internal folds, and the others, if present at all, must have been ephemeral. The minute mesofossettid is the smallest and shallowest of the fossettids, as in all eocardiids. The parafossettid is considerably larger and somewhat deeper, and the metafossettid is the largest and deepest of the three. The metaconid is a large cusp and the metalophid unites with its posterior extremity. In Luantns initialis, the inner extremity of the metalophid appears to have lost its con- nection with the metaconid and to have become secondarily en- larged on Mo. The incisor is small in comparison with the size of the cheek teeth, and there is not much difference between the two diameters. The anterior face is gently convex. The tooth lies ventromedial to the cheek tooth row, relatively lower than in later forms, and evidently did not extend posteriorly beyond M2. The internal face of the horizontal ramus is gently shelving above the incisor, as in other eocardiids. The lateral surface of the mandible is not preserved in this specimen and is incomplete in the lectotype of the Colhuehuapian Luantus initialis. It is fortunately pre- served in one of Ameghino's specimens of L. propheticus, which reveals a typically eocardiid structure. Ameghino placed Asteromys in his family Cephalomyidae, which he regarded as broadly ancestral to all caviomorphs, and indeed to all rodents. In 1898, he shot very close to the mark with his statement that the genus "... parece ser el antecesor de los cavinos" (1898, p. 176). By 1906, however, he appears to have receded from this conclusion for we find him making com- parisons between Asteromys and Alactaga. Later authors, misled by the upper cheek teeth erroneously referred to" A. prospicuus" 372 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY by Loomis, have placed the genus in the Erethizontidae (as "Acaremyiuae"). There can be no doubt whatever that Ara- eghino 's views of 1898 were very nearly correct ; Asteromys is unquestionably the most primitive known cavioid. It is equally clearly a member of the Luantinae, and none of its known fea- tures would seem to bar it from the ancestry of Lnanius. The contemporary, and much higher-crowned, Chuhutomys shows that the eocardiid dichotomy began in pre-Deseadan time, but it is nevertheless probable that in Asteromys we have a comparatively little-modified descendant from the common an- cestry. The lower molars are of the usual, four-crested, early caviomorph type. Apart from the higher crowns and the re- duction of the mesofiexid-mesofossettid, a striking familial char- acter that was presumably established earh^ in the history of the group, there is no important structural difference from the molars of Platypitfamys. The premolar, as Wood (1949, p. 20) has pointed out, is advanced over that of Platypittamys in talonid structure, and is superficially different in the trigonid, as noted aboA^e. These differences appear to be relatively trivial. The premolars of Santacruzian octodontids, especially the lowers, were extraordinarily A^ariable in superficial detail (Patterson and Kraglievich ms.). Within Sciamys principalis, P4 may or may not possess a grooved anterior face and a mesolophid and a fossettid on the side of the ectolophid, the talonid may or may not be molariform, and all combinations of these characters occur at random. If earlier octodontids were similarly variable — and it may be noted that the premolars of the two known specimens of Platypittamys are far from being exactly similar — derivation of the ancestral eocardiids from an octodontoid stem is altogether likely. As stated elsewhere in this paper, caviomorphs do not possess, nor were they derived from forms possessing, a mesolophid in the lower molars. The "mesolophid" of P4 of certain specimens of Sciamys principalis was an independently acquired premolar character. It crops up as one of a number of variants resulting from the essentially random variation of the talonid crestlets, which dift'er in position and in attachments to each other and to the metaconid, protoconid and ectolophid. wood and patterson : oligocene rodents of patagonia 373 Table 9 Tooth measurements (in mm.) of Asteromys punctus, M.A.C.N. no. A 52-110. Ii anteroposterior 1.52 transverse 1.45 P4 anteroposterior 3.50t transverse 2.51t Ml anteroposterior 2.97 transverse 2.58t Mo anteroposterior 3.10 transverse 2.77t M3 anteroposterior 3.17t transverse 2.30t Greatest extra-alveolar diameter. Subfamily EOCARDIINAE ChUBUTOMYS gen. nov. Type species. C. simpsoni sp. nov. Distribution. Deseadan, Patagonia. Diagnosis. Check teeth much higher-crowned than in Aster- omys, approaching hypselodonty, higher than in Cephalomys and Litodoniomys; lower molars with hypoflexids extending across approximately two-thirds of crown diameter; enamel in- terrupted after wear on anterior face and posterointernal corner of teeth, more persistent on internal sides than in later eocar- diines; cement absent. Chubutomys simpsoni-'* sp. nov. Fig. 27 Type. A.M.N.II. no. 29557, fragment of left ramus with Mo-s- Hypodigm. Type only. Horizon and locality. Deseado formation; Cabeza Blanca, Chubut, Argentina. Diagnosis. As for the genus; for tooth measurements see Table 10. The teeth are very high-crowned. As in Cephalomys, the enamel extends to variable distances down the sides of the teeth, 28 Named for Dr. George Gaylord Simpson, leader of the Scarritt Patagonian Expeditions and finder of the type. 374 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY SO that it is interrupted after wear on the anterior face (begin- ning at the lingual side) and at the posterointernal corner. The large hypofiexid, dividing the teeth into two equal halves, extends about two-thirds of the way across the crown. At the lower levels of the crown, this fold gradually becomes shallower, disappearing just above the roots, as in the luantines and in other eocardiines with rooted cheek teeth. The anterofossettid is a narrow oval, di- agonally placed on the crown and extending well down into' the tooth. On M3, the posterior end of the metaflexid still opens lingually. On M2, it had just been converted to a metafossettid, which also extends deep into the crown. No trace is present of the ephemeral mesofossettid, the smallest and shallowest of the folds in all eocardiids. The buccal tips of the protoconid and hypoconid are partially separated from the rest of the tooth by faint vertical furrows (Fig. 27), which Scott (1905, p. 462) Fig. 27. Chuhniomys simpsoni gen. et sp. nov. LM2-3, tvpe, A.M.N.H. no. 29557, x 5. observed in Eocardia, and which also occur, although not con- stantly, in Luantus. Enough of the bone is present on the lateral side of the ramus fragment to reveal the presence of a rather weak masseteric crest of eocardiid type. There are several points of resemblance to Scotainys: the ephemeral nature of the lingual flexids; the linguad extension of the hypofiexid; and the disposition of the enamel. Among other features, however, the two genera differ in that the hypo- fiexid is shallower, the antero- and metafossettids persist longer, and there is no cement in Chuhutomys. There are also resem- blances to Cephalomys and Litodontomys, but the present form is more hypsodont and the crown is divided into two lobes by tlie hypofiexid instead of by that fold and the mesoflexid, whicli WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 375 disappears quickly in Chuhutomys. As in Neoreomys, the hypo- flex id becomes progressively more anterior in position as the root is approached and the outline of the tooth becomes less elongate and more quadrangular. These various similarities to other high-crowned caviomorphs are merely such as would be expected to occur among forms that had diverged from a common ancestry not long prior to the Deseadan. The several characters in which Chuhutomys re- sembles the eocardiines seem definitely to indicate close affinity. The obviously small and ephemeral mesoflexid-mesofossettid, already worn away in the type specimen, the very high crowns, the very extensive hypoflexid, the general shape of the crowns, and the sharply angulate protoconids and hypoconids with their tendency toward constriction of the tips by faint vertical furrows — all these present a combination of characters typical of this subfamily. The distribution of enamel is more extensive in- ternally than on the corners. The bearing of Chuhutomys and other high-crowned Deseadan forms on the time of origin and initial radiation of the Caviomorpha is discussed in the conclud- ing section. *o Table 10 Tooth measurements (in mm.) of Chuhutomys simpsoni, A.M.N.H. no. 29557. Ml anteroposterior 3.42 width metalophid 2.65 width hypolophid 2.80 M2 anteroposterior 3.58 width metalophid 2.73 width hypolophid 2.65 EoCARDiiDAE gen. et sp. indet. Fig. 28 A.C.M. no. 3054, EM^, figured by Loomis (1914, Fig. 129) as M2 of Asteromys prospicuus Ameghino. Cabeza Blanca, Chubut. M.A.C.N. no. A 52-87, R dm4 (the tooth, referred to above, p. 357, that was in the same box with the syntypes of "Asteromys prospicuus"). Pre- sumably Cabeza Blanca, Chubut. 376 BULLETIN : MUSEUM OF COMPAEATIVE ZOOLOGY The Amherst specimen (Fig. 28) has a wear facet at only one end, showing that it cannot have been Mo, as Loomis supposed. There was a single internal root, indicating that the tooth is an upper. The lack of divergence of the roots and the rather high crown suggest that it is a permanent tooth. We consider it most probably to be M^. The tooth is elongate, quite high-crowned, despite considerable wear, and show^s marked unilateral hypso- donty. The parafossette is larger and deeper than either the mesofossette or the metafossette. Of the three, the mesofossette is the smallest and the metafossette is the shallowest, and hence the most ephemeral. There is a partial dam across the parafossette, so that it might be split into two fossettes briefly at the right stage of wear. These characters combine to suggest eocardiid affinities. Fig. 28. Eocardiidae gen. et sp. indet. EM^?, A.C.M. no. 3054, x 5. M.A.C.N. no. A 52-87 lacks the posteroloph and is not quite complete externally. It is not as worn as the Amherst tooth — the paraflexus is still open - — but is very similar in structure. There is also one large internal root and two small external ones. We suspect this tooth to be dm'*. These two teeth certainly represent the same form. The fact that they agree with the lower molars of Asteromys puncius in possessing a similar complete investment of enamel inclines us to suspect that they may i)ossibly be referable to that species. Cer- tainty, of course, can only come with discovery of more complete material. Table 11 Tooth measurements (in mm.) of Eocardiidae indet. A.C.M. no. M.A.C.N. no. - _ 3054. RM3 A 52-87, dm4 V Anteroposterior 3.05 — AVidth protoloph 1.70 1.91 Width metaloph 1.73 — WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 377 Siiperfamily ERETHTZONTOIDEA Simpson 1945 Family ERETIIIZOXTIDAE Thomas 1897 Subfamily ERETIIIZONTINAE Thomas 1897 The New World porcupines are represented in the Deseadan by various isolated teeth that indicate the presence of one genus, Protostciromys nov., with two species. These fragmentary re- mains reveal that the typical, rather simple molar pattern of the family was then already fully established ; in this character there has been almost no subsequent change. The Recent erethi- zontids are rather distinct in their myology, serology and lice from the remaining caviomorphs, and there are considerable differences between the Santacruzian representatives of the fam- ily and other contemporary forms in the structure of the skull and the postcranial skeleton. The skull structure of the Colhue- huapian erethizontids, so far as known, is generally similar to that of the Santacruzian ones. It seems likely that the cranial and postcranial specializations characteristic of the family were already under way by Deseadan time. At the least, it would be unsafe to assume that the similarity in molar structure between the erethizontids and the earliest octodontids and echimyids indicated that the Erethizontidae had diverged from the an- cestral stock only slightly prior to the Deseadan. PrOTOSTEIROMYS gen. nov. Ty2)e species. Eosteiromys medimius Ameghino 1903. Distribution. Deseadan, Patagonia. Diagnosis. Molars lower-crowned than in later forms other than Eosteiromys: upper molars with four main crests, rudi- mentary neoloph; mesofiexus shallower than in Eosteiromys; hypocone on same anteroposterior line as protocone on M^'^; lower molars with four crests, antero- and metafossettids nearly or fully formed. The taxonomic history of this genus is rather confused. In 1901 Ameghino (pp. 76-78) gave a faunal list of the Santa- cruzian including Steiromys principalis and 8. annectens, and indicated that the Coendidae (=Erethizontidae), to which he referred them, were present in the Colhue-Huapi {=Colpodon beds). The following year (1902c, p. 110), he described Eosteiro- 378 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY mys homogenidens, nov. gen., nov. sp., from the Colhue-Huapi, but marked it as having been mentioned in the 1901 paper, which was not the case. A year later (1903a, p. 129), Ameghino dis- cussed Eosteiromys homogenidens and E. uniformis from the Colhue-Huapi, and E. medianus from the Deseado {=Pyrother- ium beds). Although he did not describe E. medianus as new in this paper, his brief discussion and figure (Fig. 51) are sufficient to date the species from this paper. The next year (1904, p. 249), he formally described E. medianus as a new species from the Deseado, referring it with some doubt to the Colhuehuapian Eo- steiromys, and cited his previous reference of 1903. This confu- sion was no fault of Ameghino 's, but was the result of papers being published in different journals. Also in 1904, he described Parasteiromys imiformis, nov. gen., nov. sp., from the Colhue Huapi, referring back to the 1903 paper where he had figured and described it as Eosteiromys uniformis. In view of this rather confused situation some comment on the Colhuehuapian erethizontids is necessary. Three genera have been recorded : Eosteiromys, Parasteiromys and Steiromys. The first of these is a relatively primitive form, but little advanced over Protosteiromys, and with a non-molariform P^*. Para- steiromys was originally diagnosed by possession of P^, reduction of the mure (resulting in continuity of the meso- and hypoflexus) , and the presence of a neofossette. The type specimen of P. uni- formis is a very young individual in which dm^-M^ are preserved, not P^-M^ as Ameghino supposed. The evidence for a tooth an- terior to dm"* consists of the perforations in the maxilla shown in his figure and interpreted as alveoli (1903a, pp. 126-7, Fig. 47). At the same time, he stated that a fifth upper cheek tooth also occurred in young individuals of Steiromys detentus, presum- ably on similar evidence. Whatever these perforations may be, and they are probably vascular, they are not alveoli. Similar ones occur in young Recent erethizontids and in these no cheek tooth anterior to dm^ is ever present. A neofossette occurs in several erethizontids, and is fully as well developed in tlie Santacruzian Steiromys duplicatus as it is in the type of P. uniformis. The reduction of the mure scarcely seems to be a character of generic significance, and we therefore refer uni- formis to Steiromys. The species Steiromys axiculus, S. tab- WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 379 lilatus and S. nectus, the last two synonyms of the first, represent a new and very distinct genus, Hypsosieiromys (Patterson, 1958a), also with a non-molariform P^ The Deseadan species, medianus, is generically distinct from the forms from the Colhue-Huapi. Direct comparison with the type of E. homo genid ens reveals that the latter has a con- siderably deeper mesoflexus, and a hypocone more external in position on M^ -. We therefore have no alternative but to propose a generic name and have selected Protosteiromys. A second, and more primitive species, P. asfnodeophiUis, is described below. This genus may be characterized as including very low-crowned erethizontids with the four principal crests of the upper molars usually still quite distinct and with incipient division of the posteroloph to form a neoloph. In the lower molars, the fosset- tids are already nearly or quite formed. Variability in homo- logous teeth is considerable, as in later forms, which warns against too rigid a reliance on diagnoses based on one or two specimens. The upper premolar is unfortunately unknown, but there is no reason to doubt that only three crests, the anteroloph, proto- loph, and posteroloph, were present. This tooth is not yet molariform in the Colhuehuapian Eosteiromys hnmogcnidens, in which a small fossette in the posteroloph marks the rudiment of a fourth crest. Five isolated upper molars are known: Ameghino's type of medianus; A.C.M. no. 3014, described by Loomis (1914); and three specimens in the Museum National d'Histoire Naturelle. All are referred to P. medianus. The type upper molar, M^^''^ (Fig. 30A), is a little wider than long, and shows a slight degree of unilateral hypsodonty; there is a wear facet at each end. The protocone and anteroloph form a curving, lunate, crest that extends almost to the paracone, only a very shallow paraflexus intervening. The protocone is connected to the an- terior end of the short mure by a well developed isthmus, the protoloph continuing externally from this point in a very gentle curve, convex forward. A slight terminal swelling marks the position of the paracone. The thin, straight and nearly transverse metaloph arises at the posterior end of the mure, both loph and mure diverging from the anterior arm of the hypocone. A very shallow metaflexus intervenes between the metaloph and the anteriorly curving external extremity of the posteroloph. Both 380 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY para- and metaflexus would be converted to fossettes with very little additional Avear. The mesoflexus is deeper and it is doubt- ful if a mesofossette would ever have developed. All three flexi are rather shallow relative to the height of the crown. The pro- tocone is moderately elongate anteroposteriorly, more so than in most erethizontids. Probably in correlation with this, the hypo- cone is decidedly internal in position, on the same anteroposterior line as the protocone. Posterointernally, it is produced into a thin, prominent crest. The hypoflexus is widely open in the un- worn or little worn crown but rapidly decreases in size upward, due to the forward slope of the anterior face of the hypocone. It extends for approximately half the height of the internal face of the tooth. On the anterior slope of the external half of the posteroloph there is a small, fifth crest, the neoloph, which has isolated, with the posteroloph, two small neofossettes, a minute external one and a slightly larger internal one. This crest is further discussed below (p. 391). The other first or second molars (Fig. 30B) differ from the type, and among themselves, in a few respects. Only one small neofossette is isolated by the posteroloph and the neoloph, the anteroloph may be less curved and the paraflexus and metaflexus slightly deeper. If we may judge from Steiromys, in which the molars show a considerable degree of variability in these features, such minor differences have no taxonomic significance. The last molar is represented by M.N.II.N. no. 1903-3-15, (Fig. 30C), a practicallj" unworn specimen. As in M^ of all erethi- zontids, the hypocone is more buccad in position than the proto- cone. The protocone is not united with the protoloph ; spurs from the protocone and the lingual end of the protoloph ap- proach eaeh other but would not have united until a considerably more advanced stage of wear had been reached. This feature also occurs in some specimens of Jlijpsosteiromys, and may crop up in molars of Steiromys and the Recent forms. Anteroloph and protoloph are united buccad, an advanced feature not found in the other specimens. The metaloph is not connected to the hypocone. The neoloph is rudimentary and the neofossette very small. Five lower molars (Figs. 31, 32) are also known, of which three are of appropriate size for P. mcdionus and are accordingly WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 381 referred to that species. The remaining two, which differ from the rest in large size, relative narrowness and other features, con- stitute the hypodigm of P. asmodeophilus described below. As in later erethizontids, the lower molars are relatively broad com- pared to their length. The four lophids are very distinct, stout and narrowing rapidly toward their summits, as are those of the upper molars. When unworn they tend to have a delicate ap- pearance, which is rapidly lost with wear. The cusps are almost indistinguishable from the crests, only a few traces remaining in P. medianus. The hypolophid and posterolophid are united lingually to isolate the metafossettid, even in unworn teeth. The anterofossettid is similarly isolated in the type of P. asmodeo- philus; in the other available teeth a slight and shallow notch intervenes between the anterolophid and metalophid. The meso- flexid is the deepest of the lingual folds. Only after great wear, if at all, would a mesofossettid be formed. The hypoflexid is deeper than the mesoflexid and extends into the crown surface for slightly over a third of the total width of the tooth. A thin, prominent spur extends anteroexternally from the site of the hypoconid, but, in contrast to some later forms, there is no corresponding spur from the site of the protoconid. There is no trace in the anterofossettid of any cuspules or crestlets, such as may occur in later forms (cf. Steiromys duplicatus, Scott, 1905, PI. 66, fig. 2a). The upper and lower molars of this earliest known erethizontid genus show no particular resemblance to those of the Hystricidae, in which the unworn crowns have massive, tuberculate and closely appressed crests (Friant, 1935, p. 231). Protosteiromys seems clearly to be related to Eosteiromys of the Colhuehuapian and to Steiromys of the Colhuehuapian and Santaci-uzian, to which it may well have been directly ancestral. The Colhuehuapian Hypsostciromys is sufficiently specialized to suggest that its ancestry may have been distinct in the Deseadan. The only known later Tertiary form is the relatively gigantic N eosteiromys homhifrons Rovereto from the Pliocene of Cata- marca, the largest known erethizontid (condylobasal length 151 mm.) ; this was probably a descendant of Steiromys. The living erethizontids appear readily derivable from an ancestor such as Protosteiromys. This form, then, as far as known, may be con- 382 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY sidered as representing essentially the basal stock of the Erethi- zontidae, as indicated in the phylogenetic chart of the Oligocene and Miocene members of the family (Fig. 29). SANTACRUZIAN COLHUEHUAPIAN OESEADAN HYPSOSTEIROMYS STEIROMYS STEIROMYS EOSTEIROMYS PROTOSTDROMYS Fig. 29. Tentative phylogeny of the Oligocene and early Miocene Erethi- zontidae. Protosteiromys medianus (Ameghino) Figs. 30-31 lEosteiromys medianus Ameghino, 1903a, Fig. 5129, p. 129. Eosteiromys medianus Ameghino, 1904, p. 249; 1906, p. 413, Fig. 288; Loomis, 1914, p. 196, Fig. 130; Schaub, in Stehlin and Schaub, 1951, pp. 40-41, Fig. 38. Type. M.A.C.N. no. A 52-111, RM^ °^ -. Hypodigm. Type and the following specimens: A. CM. no. 3014, LM^"^- {not right, as Loomis stated); M.N.H.N. nos. 1903-3-83, LMi"r2. 1903-3-84, broken LMi"'-^; 1903-3.15, unworn LM3; 1903-3-16, LMioro; 1903-3-17, RMi.,r2; 1903-3-85, RMior2. Fig. 30. Protosteiromys medianus (Ameghino). Upper molar teeth x 5. A, EMI, type, M.A.C.N. no. A 52-111; B, LMi »' 2^ A.C.M. no. 3014; C, LM3, M.N.H.N. no. 1903-3-15. 29 Ameghino's figure is actuall.v about four times natural size, and not three times as stated. It is inaccurate as regards siiape and in showing one fossette too many at the rear of the tooth. Loomis' figure of A.C.M. no. 3(>14 is likewise inaccurate in shape and in most details of crown structure. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 383 Horizon and localities. Deseado formation ; the Amherst speci- men is from Cabeza Blanca, Chubut; the Museum National d'Histoire Naturelle specimens are from La Flecha, Santa Cruz. Amep:hino (1904, p. 249) stated that the type was from the "py- i-otheriense del Deseado." This was the Ameghinos' designation, variously written as "Deseado," "Rio Deseado" or "Sur del Deseado," for the La Fleeha locality. A B Fig. 31. Protosteiromys medianus (Ameghino). Lower molar teeth x 5. A, LMi, M.N.H.N. no. 1903-3-16; B, EM2, M.N.H.N. 1903-3-17. Diagnosis. Smaller and relatively slightly higher-crowned than P. asmodcophllus (Table 12) ; lower molars shorter relative to width, fossettids thus more compressed anteroposteriorly, meta- lophid and hypolophid slightly convergent buccad. Protosteiromys asmodeophilus^" sp. nov. Fig. 32 Type. C.N.II.M. no. 15061, RMior^- Ilypodigm. Type and M.N.H.N. no. 1903-3-18, a worn and broken tooth, perhaps RM;{. Horizon and locality. Deseado formation, La Flecha, Santa Cruz. Diagnosis. Larger than /'. medianus (Table 12) ; crowns of same actual height as in F. medianus and hence proportionately lower; fossettids, particularly the metafossettid, rounder and less compressed anteroposteriorly ; mesoflexid with sides slightly divergent rather than slightly convergent lingually, lingual ex- 30 Thp typp was founrt in the laboratory in matrix surrounding a femur of AsmodeuH osborni Ameghino. 384 BULLETIN : MUSEUM OP^ COMPARATIVE ZOOLOGY tremity deeper; central portion of ectolophid less oblique; no trace of distinct cusps. In every character except the complete incorporation of the cusps into the crests, this species is more primitive than P. medianus. An early trend in dental evolution in the Erethizon- tidae was toward rather wide lower molars. Nearly every point in which P. asmodeophilus differs from P. medianus reflects a tooth that is relatively narrower in tliis diameter. Table 12 Tooth measurements (in mm.) ot Protosteiromys P. medianus 2; CO o ■* u^; %6''- • c. M c- w ®'^ M o-^ HcK < =^ S a~ S a J S C-; anteroposterior 5.21 4.70* 4.48* 4.25* ca. 4.6i width protoloph 5.48 5.00* 5.07* 4.81* width metaloph 4.75* 4.92* 4.09* height of crown, protocone 3.01 2.80 worn 2.77 .CC sis lA 00 ^•2S .CO Zco " anteroposterior 4.88* 5.19* 5.18* width trigonid 4.51* 4.62* 4.62* width talonid 4.58* 4.84* 4.93* height of crown, protoconid 2.47 worn 2.44 P. asmodeophilus Type C.N.II.M. no. P 150G1 RMi orn M.N.H.N. no. 1903-3-18 RM3? anteroposterior 6.75* ca. 7.20 width trigonid 5.52* ca. 4.45 width talonid 5.60* 5.01* lieight of crown, protoconid 2.50 (unworn) 2.47 (worn) WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 385 The referred specimen is sonie-vvhat narrower than the type. As in Ms of Hypsosteiromys, the trigonid is appreciably narrower than the talonid, which is the chief reason for identifyinj? tliis tootli as M;5, since the break at the rear of the tooth makes it impossible to determine whether or not there was a wear facet there. The anterofossettid is still a little open lingually, in contrast to the closed condition in the type. A B Fig. 32. Protosteiromys asmodeophilus sp. nov. Lower molar teeth x 5. ./, EMi or 2, type, C.N.H.M. no. P 15061; B, RM3, M.N.H.N. no. 1903-3-18. Caviomorpha indet. A single left lower molar (A. CM. no. 3039, Cabeza Blanca, C'hubnt) cannot be allocated to any of the described forms. It is high-crowned but rooted, comparable to Cephalomys and lAtodoniomys and of about the same size as L. chuhutensis and A B Fig. 33. Caviomorpha, gen. et sp. indet. x 5. A, left lower molar, A.C.M. no. 3039; B, LM^?, M.N.H.N. no. 1903-3-21. C. plexus. It does not have any cement, and the enamel is con- tinuous around the crown almost, if not quite, to the roots. There is a hypoflexid that extends about half way across the crown, buccal to which is a small mesofossettid (Fig. 33A). Neither 386 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY antero- nor metafossettid is present. This form is clearly distinct from all described species in the uniform extension of enamel and the persistence of the mesofossettid. It shows some resemblance to the tooth described above (p. 362) as Dasyproctidae indet., to which it may conceivably be related. An isolated upper molar (M.N. H.N. no. 1903-3-21, La Flecha, Santa Cruz; Fig. 33B) appears to represent still a different structural type. There are three transverse crests, although a small anteriorly directed crestlet from the posteroexternal corner suggests the last remnant of a metaloph. There is a continuous ridge of enamel around the margin of the tooth, so that the buccal valleys were fossettes from the start. In the reduction to three crests, this tooth is reminiscent of the lower molars of Deseadomys, but no such reduction occurs in the uppers referred to that form. In addition, the present tooth is much too high- crowned to be referred to the Echimyidae. It is possible that the animal represented by this molar may also be a dasyproctid. The degree of hypsodonty is also comparable to that occurring in Cephalomys and Litodonfomys, and the absence of a metaloph and the presence and direction of the short spur running from the metacone are reminiscent of conditions in the unworn upper molar of Cephalomys plexus (cf. Fig. 15A). There is a slight elevation near the middle of the posteroloph and another postero- internal to the protoloph on the mure. These two elevations occupy positions comparable to those occupied by the two ends of the neoloph in C. plexus. It is obviously inadvisable to name either of these forms until material is available that would permit their relationships to be better understood. They clearly show how very far we are from a complete knowledge of the Deseadan rodents. Table 13 Tooth measurements (in mm.) of Caviomorpha, indet. anteroposterior anterior width posterior width M.N.H.N. 0. 1903-3-21 LM3? A.C.M. no. 3039 KMl or 2 2.75* 2.06 2.10* 1.91* 2.00* 1.94* WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 387 DISCUSSION Bearing of the Kodents on the Age of the Deseadan The Deseado is sufficiently oldei- than the Colhue-IIuapi so that all genera of rodents, and practically all genera of other mammalian groups, are different in the two horizons. If the criteria on which these South American rodent genera are based are comparable to the criteria distinguishing their contemporane- ous North American relatives, the gap between the Deseado and Colhue-Huapi is much greater, for example, than that between the early Oligocene Chadronian at Pipestone Springs, Montana, and the middle Oligocene Brule of the "Oreodon Beds" of South Dakota. Of the seven rodent genera from Pipestone, Prosciurus, f.scJn/roniys and Farad jidaumo are represented in the Brule; TitanoihcriottwjH probably is; and Adjidaumo is found in con- temporaneous deposits not in South Dakota. This leaves Pseudo- cylindrodon and Cylindrodon as the only rodent genera present at Pipestone but not in the "Oreodon Beds," and the former is represented only by a single specimen. This comparison sug- gests that the gap between the Deseado and the Colhue-Huapi may represent most of the Oligocene. Since, however, no rodents are known from South America earlier than the Deseadan, we are convinced (as pointed out below) that they reached that continent during the long post- Mustersan hiatus. Once members of this order reached the relatively virgin field of South America, they certainly under- went a very rapid initial radiation, and this would have lasted until the rodents had occupied most of the available niches. Therefore, such an evolutionary development would have in- volved less time than a similar amount of evolution in North America, where the niches were already filled. Intercontinental correlations Avould be greatly aided if we knew the detailed derivation of the Caviomorpha. Of the known possible ancestors, however, we believe Rapamys to be the most probable candidate (see l)elow). This rodent, known from the late Eocene Duchesnean of North America, could be distantly ancestral to all the Deseadan rodents. If so, the morphologic change was certainly much greater than that between the Deseadan and Colhuehuapian rodents, being on the subordinal 388 BULLETIN : MUSEUM OK COMPARATIVE ZOOLOGY or familial rather than the g-eneric level. Of course, if the argument for rapid post -Desea dan evolution, outlined above, is valid, that for rapid pre-Deseadan evolution would be even more so. A combination of these lines of reasoning- would suggest that the Deseadan is early Oligocene, but probably not earliest Oligocene, being perhaps contemporaneous with typical C'hadron in North America; the Colhuehuapian is perhaps early Whit- ney an. If the Deseadan eaviomorphs were not derived from Rapamys or a close relative, all other known possible ancestral forms would be middle Eocene or earlier. Such an ancestry would allow the age of the Deseadan to be shifted back toward the beginning of the Oligocene, but probably not as far as into the late Eocene. Therefore, no matter what the ultimate source of the Cavio- morpha, it would seem probable that we are here dealing with an early Oligocene fauna. Schaub (m Stehlin and Schaub, 1951, p. 41) has expressed a comparable opinion. The Relationships op^ the Deseadan Rodents TO Later Forms and to Each Other The problem of the relationships of these rodents is a two- fold one. On the one hand, is the question of their affinities to each other and to those that follow them in time ; on the other, is the broader question of the origin of the indigenous Neotropical rodent fauna as a whole. The second part of the first question — the relationships of Deseadan to later forms — has been gone into in the discussions on tlie preceding i)ages. We believe that it has been satisfactorily demonstrated that Flatypiftamijs is very near the direct line of the Octodontidae and is structurally very close to the ancestry of all the rest of the fauna except Protosteiromys; that the two sj)ecies of Deseadomys are essentially ancestral to the Addplio- mys group of the Echimyidae ; that Ceplialnmys and lAtodon- tomys may be referable to the Dasyproctidae, although their precise relationships to later members of the family cannot be decided at present : that Scotamys is a broadly ancestral member of the Chinchillidae ; that Astcronujs and Chiihutomys are an- cestral members of the two subfamilies of the Eocardiidae ; and WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 389 that Protosteiromys is on or very near the direct line of the Erethizontidae. It would appear at first X) eopiXujouin — "'^«»^ !c drt^^^^ O ^ V "■"aDPMn^!^"^ .---'■ :;«n-voi»'«"" s "^'X •''°^ ^ ^ \ ° aopirADO O \ c 0) > • t. ■■•■ Cavioid / - -!^ 00^ ' 3DP!JaOL|30JpAn ««^^ t» ^^^ »» ^^ # ^ ^ « ^v • ^v • X. • X. • X t X • > 1 aopi^Luouajo ♦ 1 % ■ o ^^^^^ \y. -••' o aopuuoDo •30 c o ^^"■^^ ■o .••' / o ^^^^^^^^^^ 9 ^_ CJ ■ ,„ ^^^^^ f . , 1 1 1'> ^ "^ • .•' ° aopiAUJ'.^ ■(■'3 * aopiXoiojdoQ * • • y 0 • o ^^ y^ • c lA 1/) V □ o "o n! 396 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY by the considerable degree of advance that took place in all lines between Deseadan and Colhiiehnapian time. The parallel case of late Pliocene and Pleistocene evolution of hypsodonty in the raicrotines can be cited in this connection. This point of view is a considerable contrast to that of Landry, who considers that the absence of any paramyids, as well as theridomyids and hystricids, in South America "suggests the possibility that no known rodent is ancestral to the hystricomorphs and that their intercontinental dispersal may have taken place considerably before the time usually postulated, perhaps as far back as the Early Paleocene" (1957a, pp. 81-82). This is, naturally, a possi- bility, but the hypothesis involves the assumption that pre- Descadan rates of evolution in caviomorphs were extremely slow, and for this we see no evidence. Furthermore, as pointed out in more detail below, it also requires us to believe that in the Paleocene a relatively advanced group of rodents was in existence in the Northern Hemisphere, and that this group then disap- peared from that hemisphere to be replaced (or displaced) in the early Eocene by the less advanced paramyids. This we regard as unlikely in the extreme. The Origin of the Deseadan Rodents The relative uniformity of the Deseadan caviomorphs leaves little doubt that the entire South American stock represents a development from a single invasion. A number of problems still remain to be considered, liowever. These are: (1) from what region did the ancestors of the caviomorphs come ; ( 2 ) at what time did the invasion occur; and (3) what group of rodents was ancestral to the caviomorphs. These questions will be discussed in order. The similarities between South American and African rodents tempted many authors to derive the former from the latter by a direct transatlantic invasion via a land bridge. Current opinion seems practically unanimous that there could not have been direct connection between Africa and South America, either by bridge or by former continental juxtaposition, at any time when rodents could have utilized it, and that therefore the immediate source of the South American rodents must have been North WOOD AND PATTERSON : OLIGOCENE RODENTS OP PATAGONIA 397 America (see for example Lavocat, 1951a, 1951b, p. 72; Scliaub, 1953a, p. 391; Simpson, 1950, p. 375; Wood, 1950).'- Landry (1957a, p. 91) suggests that the ancestral caviomorphs may have reached South America by way of the Greater Antilles, and tluis ultimately be of North American derivation. On present evidence, this route is nearly (though not quite) as logical a postulate as the alternative route through the then Central American archi- pelago. Landry's second hypothesis, that " hystricomorph " ro- dents reached the Antilles in the Paleocene is not a necessary part of his migration route hypothesis, and is discussed in more detail below. Viret (1955, pp. 1563-1564) is more cautious on the whole problem, and concludes : "Le probleme du peuplement de I'Ameri- que du Sud par les Rongeurs reste une enigme qui ne peut etre resolue dans I'etat actuel de nos connaissances. " Simpson (1950, p. 375) and Wood (1950) have discussed the question of arrival of rodents over a land bridge or by rafting. Both reached the conclusion that the rodents must have entered South America by the latter method, because both interpreted the evidence as showing that the invasion took place in the latter part of the Eocene, a time at which no other mammals entered South America from North America, with the possible exception of the primates. We fully agree with this conclusion. Recently, however, Schaub has suggested that the source of the Cavio- morpha is to be found in rodents with a Theridomys-like tooth pattern tluit reached South America in the Eocene, Paleocene, or even earlier (1953a, p. 393). Lavocat (1956, p. 55) goes part way with Schaub, suggesting that the similarities between the Caviomorpha and the Hystricomorpha (or Nototrogomorpha and Palaeotrogomorpha) could be explained on the assumption that the two groups were remnants of a homogeneous Paleocene 32 That the ancestral South American rodents might have been transporteri by natural rafts directly from Africa to South America, or, contrariwise, the ancestral hystricomorphs (scnsii stricto) from South America to Africa, in the latter part of the Eocene are views seldom advanced. We do not false either possibility very seriously, particularly not the former. Hystricomorphs (scnsu stricto) do not appear to be raft-prone : they have failed to reach such islands as Madagascar and the Philippines proper, not to mention New Guinea and the continent of Australia. We wish to point out, however, that, in the light of current knowledge of the paleontology and zoogeography of the Rodentia, raftings of this sort are the only alternatives left to those who insist on direct caviomorph- liystricomorph relationships. Otherwise, as we indicate below, there seems to l)e no escape from the conclusion that the ancestors of the South American rodents lived in North America during the latter part of the Eocene, and that, in this continent at that time, they could only have been protrogomorphs and, in all probability, paramyids. 398 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY rodent population of the Northern Hemisphere that had pene- trated South America and Africa. On the other hand, he had previously concluded (1951a, p. 38) that any resemblances be- tween New and Old World forms must be pure parellelism. Landry (1957a, pp. 90-91) strongly supports the caviomorph- hystricomorph distribution as indicative of a Paleocene radia- tion. In view of the weight of this continued support for close relationships between South American and Old World forms, the problem of the time of orig'in of the caviomorphs must be recon- sidered. There was certainly an early immigration from North America to South America, in late Cretaceous and/or early Paleocene time, which provided the ancestors of most of the indigenous South American Tertiary mammalian fauna (Simpson, 1950, pp. 368-373). These immigrants may well have reached the con- tinent by rafting (Darlington, 1957, p. 364; Patterson 1958b, pp. 11-13). This, apparently, would be the source envisioned by Schaub and Landry, although, as already mentioned, Landry (1957a, p. 91) hedges a bit by suggesting that perhaps the "hystricomorphs" reached the Greater Antilles in the Paleo- cene, but did not reach South America until early Oligocene. Lavocat specifically suggests (1956, p. 55) that, after making such a Paleocene invasion, the rodents could well have lived in parts of South America whose Tertiar}^ history is still unknown. Although this is undoubtedly a hypothetical possibility, it does not seem at all probable to us, for several reasons. There are actually three problems to be discussed here : whether there could have been any rodent invasion much before the end of the Eocene; whether there could have been an invasion by any Paleocene rodents (Lavocat, 1956, p. 55) ; or whether there could have been an invasion by Paleocene tlieridomyids. Establishment of the improbability of any one of these would establish the im- probability of any of the others. Since the Deseadan rodents are, as pointed out above, a closely related stock, and since they were at this time near the beginning of a rapid differentiation which continued through Colhue- hua])ian into Santacruzian time, it seems to us almost certain tluit there was a single common ancestor of tlie Deseadan rodents which existed not too long before Deseadan time. All previous WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 399 history of South American rodents must have involved a rela- tively small amount of evolutionary progress, with little or no adaptive radiation. No trace of a rodent has been found in the Paleocene Kio Chico of Patagonia, in the Paleocene Itaborai fauna of southern Brazil, or in the Eocene Casamayoran and Mustersan of Pata- gonia. It could be supposed that these animals might have been living in northern South America during the earlier part of the Tertiary, but we regard this as extremely unlikely. The Pleisto- cene faunal history' of South America and the Recent faunal liistor-y of Australia reveal what happens when forms titted for life in a given adaptive zone are suddenly confronted by in- vaders higher in the scale of evolutionary progress that happen to be fitted for the same zone ; the forms lower in the scale do not linger long and the invaders spread with the greatest rapid- ity. Had any rodents reached South America before late Eocene time, we believe it to be essentially axiomatic that their descend- ants would have swept all before them within their adaptive zone, which was then only marginally occupied (by polydolopids, groeberids, typotheres and hegetotheres), and would surely have l)ut in an api)earance in Patagonia. Moreover, we see only one l)robable explanation, other than their recent arrival in South America, for the sudden burst of the rodents just l)efore Desea- dan time. This would be the rapid development of new structures that enabled them to compete much more successfully with their non-rodent rivals. The only such structures of which we are aware would be the multilophate cheek teeth and the enlarge- ment of the masseter muscle and its expansion through the infra- orbital foramen. While we believe that such changes did take place in the latter part of the Eocene, their development at this time in the ancestors of the caviomoi-phs, had these been in the continent at a much earlier date, would establish that these an- cestors were non-hystricomorphous in teeth and masseter ; would eliminate the possibility that any Old World hystricomorph-like rodent had any special relationship with them ; and, we believe, would lead by a process of elimination to our interpretation of the late P^ocene caviomorph ancestor as being a paramyid or sciuravid. It appears to us, however, virtually certain that rodents did not reach South America luitil after the close of 400 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY the Mustersan, which, according to the best recent estimate (Simpson, 1940), is approximately of mid-Eocene age. They therefore arrived sometime during a great hiatus in South American fannal history — the gap in our knowledge that ex- tends from Mustersan to Deseadan, from approximately mid- Eocene to early Oligocene. The precise time of their arrival within this interval cannot now be determined. Rejection of the possibility of a Paleocene rodent invasion of South America requires consideration of the time of origin and of early geographic radiation of the rodents. This is something about w^hich virtually nothing is known, except that rodents do not become numerous until the early Eocene and that only a single earlier species, Paramys atavus, is known from the late Paleocene (Jepsen, 1937). How much earlier the rodents were in existence we do not know. Wilson (1951) has argued that the order, as such, probably arose about late Paleocene time, agreeing with Jepsen (1949) that the abrupt replacement of multituber- culates by rodents at the beginning of the Eocene argues against the latter having been in existence for more than a short time previously. We believe that if rodents had entered South Amer- ica in the Paleocene they would surely have been widespread in North America at that time and hence would be represented in our collections from that continent, as well as from the Eocene of South America. This, of course, is arguing from negative evidence, and anyone who wishes to hold contrary views is free to do so. However, it is surely unwarranted to conclude that such a hypothetical group of rodents were so much more highly specialized than any of the known early Eocene rodents that they must have been hystricomorphous, and to assume, further, that they acquired a world-wide (or nearly world-wide) distribution is piling hypothesis upon hypothesis, and is logically indefensible. We do not wish to argue the question as to the date of the separation of the rodents from the insectivores, or as to the rates of pre-Eocene rodent evolution, because we know of no new evidence bearing on these subjects. The possibility of a theridomyid having reached South America in the Paleocene, as envisioned by Schaub (1953a, p. 393), has additional arguments against it. As Viret has observed (1955, p. 15tJ4, footnote 1), the Theridomys-psittern was not present, even in Europe, in the Paleocene. As pointed out above, we believe WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 401 that we can trace the development of five-crested eaviomorphs from four-crested ones, and not vice versa. If we are correct in this interpretation, the theridoiiiyids again could not have been ancestral to the Caviomorpha, and we should have to rely i'or ancestors on four-crested or incipiently four-crested North American forms, of which a number are known. Our arguments on this point are involved in the whole question of the basic pattern of rodent teeth and the primitive stock of the rodents, which are discussed below. But we feel very strongh'- that, what- ever the Paleocene rodents were like, they had not yet attained the complexity of structure that would permit them to be called theridomyids, or indeed to be referred to any currently recog- nized groups except the Paramyidae. We therefore feel that the most probable explanation of the known liistory of South American rodents is that they arri^-ed, by island -hop jiing, from North America, sometime in the latter part of the Eocene. Ancestry of the Caviomorpha Having attempted to answer the questions as to the geographic source of the Caviomorpha and the time of their arrival in South America, we come to the question of the group of rodents that gave rise to them. Stehliu and Schaub (li)51) have presented a major and ex- ceedingly valuable study of the development of the clieek teeth in rodents, which we will discuss more fulh' below. But, on the basis of this study, Schaub has strongly supported the Theri- domyidae as being the ancestral stock of all multi-crested rodents. We have tried to show above that the ancestral eaviomorphs were not pentalophate, as is Thcridomys, but tetralophate, and that the development of a fifth crest, when it occurs, is a strictly local phenomenon in South America. It would be advisable, however, to review in this connection the possible origins of each of the known Deseadan rodents. As indicated above, Platijp'tiamiis nnist be very close to the liasic type from which all the rest of tlie eaviomorphs were derived. It clearly could not be descended from the Theridomyi- dae, but shows numerous structural similarities to such North American rodents as an undescribed lowest Eocene paramyid 402 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY and the Duchesnean Rapamys, as far as the latter is known. The pattern of the teeth (particularly of ?■*) of Deseadomys is very different from that of the Theridomyidae. Without exerting what we consider to be an excessive strain on the imagination, the only known source from which it could ultimately have been derived would be a paramyid or sciuravid. Deseadomys shows no close relationships to any Old World forms, and particularly not to the theridomyids or to Phiomys. It would be possible to imagine its descent from such forms, by the loss or fusion of crests, but there is no evidence for this and it seems clear to us that derivation w-as from the octodontids as represented by Plat ypitt amy s. It is therefore much simpler and more reasonable to assume the ultimate derivation of Deseadomys from North American paramyid ancestors. Landry (1957a, p. 94) states, "Wood's proof that South American hystricomorphs could not be related to Old World hystricomorphs consists entirely of reasons why the teeth of Platypittamys could not be derived from those of the therido- myids." There really was a little more to the story, but leaving this aspect of the matter aside we must point out that, in 1949, derivation of Ilystricomorpha {seusu lato) from the theridomyids was a liypothesis very nuich to the fore and Wood had no choice but to discuss it. The idea is still current — in greatly expanded form (cf. Stehlin and Schaub, 1951) — and we still think it necessary to discuss it. As we have shown above, the third crest in caviomorph upper molars is a metaloph, and, as may be seen from Stehlin and Schaub 's figures (16, 21, 26, 29) showing portions of the sequence (which we believe to be phylogenetic or nearly so) running from Paramys to Decticadapis to Adclomys to Thcridomys, the third crest in theridomyids is a mesoloph. The considerable superficial resemblance notwithstanding, the two molar types are (juite different, despite Landry's remark {op. cit., p. 95) that "A comparison . . . shows that exactly the same elements are present in both. . . . The only difference is in the assignment of names to them." As regards the Hystrico- morpha {sensn stricto), we suspect the third crest of the upper molars to be a mesoloph. Ilystricomorph history is very poorly known, however, and since this homology has not yet been proven phylogenetically, the possibility still exists that the third and WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 403 fourth crests of hystricomorph molars and of such caviomorph molars as possess five crests may be comparable. Even if this should pi'ove to be the case, it would not affect the evidence indicatiuf; that the fifth crest of caviomorphs came into existence after arrival of the group in South America, and thus quite in- dependent of its origin in the Ilystricomorpha. We may state our full afjreement Avith Landry — and with Lavocat before him — as concerns his observation that the seiuroo-nath type of jaw characteristic of the Thcridomvidae tends to bar the family from the hystricomorph — and caviomorph — ancestrj". Cephalomys could likewise have been ultimately derived from a North American Eocene paramyid, perhaps something like Rapamys, through intermediate stages similar to Plat y pit t amy s. In view of the pattern of P"*, it would seem almost certain that it was not descended from forms with quadrilophate upper pre- molars. This would eliminate most forms that have been sug- gested as being the ancestral stock, and particularly the therido- A B Fig. 35. Cheek teeth of TJieridomys x .1. Theridomys rotundidens Sfhlosser, LP-t-M^, Munich no. 1879-xv-182 a-d; B, Theridomys speciosus Schlosser, Eclm4-Mi, Munich no. 1879-xv-183 a-b. myids (Fig. 35A), which have a completely molariform upper ])remolar. The theridomyids also show no particular similarity in the structure of P4. The presence of a mesoloph in the theri- domyid upper teeth is a notable difference from the pattern of CepJialomys. The patterns of the lower molars are similar in a very general way in the two groups, but by no means sufficiently so to indicate special relaLionslii])s. The pattern of dm4 appears i-ather similar at first glance, but it is not identical. The postero- lophid and hypolophid of the two seem closely comparable. The mesolophid in Theridomys (Fig. 35B) lies well behind the an- terior part of the tooth, and the metalophid is connected with both ends of the anterolophid whereas in Cephalomys (Figs. 404 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY 20A, B) there appears to be no mesolophid, and the three anterior cusps, presumably derived from the anterolophid, connect with the middle of the metalophid. Although these are rather notable differences, how much weight should be attached to them is uncertain, due to our general lack of knowledge of the evolution of rodent deciduous molars. Scotamys, which, as has already been shown, is fundamentally similar to CepJialomys, is also derivable from the octodontids. The same applies to the two cavioids, Asteromys and Chuhutomys. Protosteii'omys, as already noted, is the only Deseadan form that shows any appreciable similarity to the theridomyids, and much of the resemblance is due to the presence of a neoloph, giving an incipient five-crested pattern. Even if the similarities in tooth pattern had originated in the same way, which, with Friant (1936b), we do not believe, it would still lie simpler to postulate an origin of Profosteiromys from the Octodontidae, rather than to assume that one Deseadan rodent originated from an Old World group while the others had a different ancestry. Since the last crests to develo]), the neoloph (fourth crest) in Protosteiromys and the mesoloph (third crest) in the theri- domyids, are not homologous, we believe the octodontid relation- ship to be the only possilile one. These arguments, of course, would have no weight Avith anyone who adopts Landry's point of view that "the pattern of the occlusal surface of almost any tooth could be derived from any other" (1957a, p. 89), but this does not appear to us to detract from the validity of our argu- ment. Since the Old World theridomyids must, it would seem, be ruled out of the caviomorph ancestry, the only possible source of this ancestry is some North American rodent, unless, with Landry, one prefers hypothetical ancestors with hypothetical molar structure, or fancies transatlantic rafting. Such an an- cestral form would have reached South America from North America by island-hopping. Since all the South American forms could hardly have been derived from anything else, and since it seems virtually certain that the Deseadan forms Avere derived from a single source, this apjiears to be the only logical interpre- tation of the available data. The similarities between the Hystricomorpha and the Caviomorpha must then be parallelisms. We feel ver^^ strongly that parallelism is a dominant character of WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 405 rodent evolution, and particularly of rodent dental evolution (Wood, 1937b, 1947, 1950), and that very similar structures may appear over and over again Avitliin the order. Lavocat (195(5) argues this same point of view. Foi- all these reasons, we do not feel that any known theridomyid, or any ancestral rodent which was sufficiently advanced so it Avould l)e called a theridomyid, could be ancestral to the Caviomorpha, and the possibility of such relations between theridomyids and hystricomorphs (soisn strict o) also seems to be ruled out. The resemblances of the thryonomyids and petromurids to certain caviomorphs have been cited as indicative of relationsliips. As pointed out by Wood (1950), if there were a close relation- ship, which is most unlikely, it would indicate that the African forms more probably had a South American ancestr.y, rather than vice versa. Landry (1957b) has recently demonstrated that the petromurids do not resemble the octodontids to any significant extent. There are only two known groups of North American rodents that could be ancestral to the Caviomorpha, on the basis of their tooth structure.^"' These are the Sciuravidae and the Paramyidae. The Sciuravidae show a basic four-crested pattern that could have been ancestral to that of the Caviomorpha. There are a number of variant types within the sciuravids, which have not been thoroughly studied. But none of the knoAvn material shows any suggestion of an enlargement of the infraorbital fora- men, and none that we have seen suggests the inflected angle of the hystricognath type found in caviomorphs. For the present, at least, we regard the sciuravids as a possible source for cavio- morphs, but not a probable one. The review of the Paramyidae, in progress by AVood, has revealed that a four-crested pattern develops within this family from a basic tribosphenic ancestral type. Structures of this sort :■:! L.infh'.v (1957.1, pp. 88-89) olijpcts to the use of iltnital aiiatoni.v In studyini; roceeds to ignore it over and over again in his discussion of hypotlietical evolutionary trends in other structures where parallelism is just as comi)lex as it is in tootlipattems. 406 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY have arisen several times within the family in different lines, in each of which the temporal se(|uence is sufficiently complete so that it is obvious tliat this is what was taking place. In one line, leading through Reithroparamys to the Ducliesnean Rapa- mys, acquisition of this dental pattern is associated with an enlargement of the infraorbital foramen. Unfortunately, the known material of Rapamys is very fragmentary, and it is im- possible to be sure of the exact size of the foramen. Nothing is known of the structure of the angle in the genus, although, as pointed out by Landry (1957a, p. 82), the angle of Reithropa- ramys suggests an incipient stage toward a hystricognath jaw. Therefore, all that can be stated with certainty is that Rapamys seems to be closer to the caviomorph type than does any other currently known North American rodent. An undeseribed earliest Eocene paramyid in the Amherst College collections, represented by a partial skull, a jaw, and fragments of the skeleton, has some similarities to the Deseadan rodents, particu- larly in the pattern of F\ The infraorbital foramen is that of a tyjncal paramyid, but the angle appears to l)e arising lateral to the incisive alveolus, making this a hystricognath paramyid. It is possible that this form may bear some relationship to the caviomorph ancestry. Geographically, then, we believe that the Caviomorpha were derived ultimately, and perhaps directly, from North American rodents, whicli. taxonomically, would be referred to the Paramyi- dae. The sequence is, of course, far from complete. The South American rodents were not descended from immigrants from Wyoming, but rather from i-odents that lived in some part of Middle America or southeastern United States, regions from which the Eocene nuunmalian faunas are essentially unknown. When and if these are discovered, we feel that tliey will include paramyids, probably related to Rapamys, which will show a number of pre-caviomorph specializations. The discovery of sciuravids at Guanajuato, Mexico (Fries, Hibbard and Dunkle, 1955), suggests that there are deposits of the critical age in this area that may eventually produce additional rodents that will help to solve this important problem. wood and patterson : oligocene rodents of patagonia 407 Problems op Rodent Phylogeny The discussion above on the origin of the eaviomorphs appears logical to us on the basis of our ideas as to rodent evolution. Not everyone is in agreement with these, and we must therefore explain why our opinions ditifer from those of some other stu- dents. Stehliu and Schaub (1951) present a detailed analysis of rodent tooth patterns and their interrelationships. This work was unfinished by Stehlin at the time of his death in 1941, and was completed, with signed sections and unsigned insertions within brackets, by Schaub. Both authors have an unsurpassed background of study on fossil rodents, so that it is with some hesitation that we express our serious objection to a number of points they raise. One of the main points of this major contribution, and the one of chief concern for the present study, is Stehlin 's concept of a " Theridomys-Trechomysplan" in molar structure. This "plan" is believed by him to have been structurally derivable from the "trigonodonten Grundplan," represented by such forms as Marmota, and to have been structurally capable of giving rise to a wide variety of derivatives. The pattern found in the early Eocene rodents referred by us to the Paramyidae is not consid- ered by him to represent the primitive rodent tooth pattern, in spite of its being the pattern of the earliest known members of the order, and in spite of the fact that it shows considerable similarity to the Marmota pattern. Nor is the paramyid structure considered to be intermediate between the "trigonodonten Grund- plan" and the "Thcridomijs-Trechomysplnn," which leaves practically no known forms occupying such a position. Evi- dently considered as close structurally (Verwandten) to Therido- mys are various African and European forms (Phiomys, Sciuro- mys, etc.) and Eosieiromys (including our Protosteiromys) and "Parasteiromys." The morphological derivates of the "Theridomys-Trechomysplan" include a wide variety of groups: llystricidae, Thryouomyidae and Bathyergidae, all the Cavio- uiorpha of this paper (derivable morphologically, it would seem, via the Erethizontidae), Castoridae and Eutypomyidae, Rhizo- inyidae and Spalacidae — to give but a partial list. 408 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY It must be emphasized that, as Stehlin states {op. cit., p. 351), this is a morphological study pursued essentially without refer- ence to taxonomy and with but little reference to relative ages of the forms concerned. The work is in fact an undertaking carried out partially, but only partially, along the lines of what Davis (1949, p. 65) has dubbed the neoclassical school of com- parative anatomy, although without reference in the bibliography to works of any of the chief exponents of this school, e.g., Naef, Kalin. It is no deprecation of the methodology characteristic of this approach to state that, in our opinion, its application, as partially carried out by Stehlin, to a single system of an order notorious for the parallelism that has occurred within it is not at this time very helpful. The "Theridomys-Trecliomysplan'^ is a morphological abstraction that has, we believe, a limited phylogenetic — and also morphologic — application, and may well contribute to future misunderstanding. As Stehlin himself emphasizes, only an acquaintance with more than one structural system of a group through time can place our knowledge of taxonomy and phylogeny on a sound basis. For this very reason we are inclined to regard the inclusion of all forms that appear to show four or five transverse crests on their cheek teeth, what- ever other variations may occur in the i*est of their anatomy, among the structural derivatives of this "plan" as being a step away from a classification based on phylogeny and back toward a convenient pigeon-holing with no basis in the evolution of the rodents. A rather large part of our objection to Stehlin 's procedure is nomenclatural. To label this morphologic abstraction the "Theridomys-Trecliomysplan" is, we believe, dangerous. To do so is to apply terms relevant in one frame of reference, the phylogenetic-taxonomic, to another that is admittedly non- taxonomic. Althougli it may not have ])een Stehlin 's intention, no one can read this paper without feeling that there are at least implications that the arrangement in accordance with this "plan" is a phylogenetic one. This point of view has since been expressly adopted by Sehaub (1953a, p. 394 et scq.), and the usage has spread to at least one standard reference work (Grasse and Dekeyser, 1955). WOOD AND PATTERSON : OLIGOCENE RODENTS OP PATAGONIA 409 In neo-classical comparative anatomy, if it is to serve its pur- pose, interpretations (in this case anythin^r that smacks of formal taxonomy) must be rigidly excluded from the methodoloj?ical process and introduced only in the terminal interpretive phase. That the "Theridomys-Trechomysplan" was derived from the "trigonodonten Grundplan" is a reasonable assumption from the viewpoint either of neoclassical comparative anatomy or of interpretive phylog-eny, but to assume that it is itself a single entity from either viewpoint is hardly warranted. Nor does it appear to us to be reasonable to assume that all the lower Eocene rodents are highly specialized and that none of the primitive ones appear until considerably later. Neither does it seem rea- sonable to assume that a family, whose known distribution is restricted to Europe, gave rise to a wide variety of forms living in all the rest of the world. At the very least, this is contrary to what has been observed in other orders of mammals. Without entering into a detailed neoclassical comparative anatomical analysis of rodent molars with "Aussen- und Innenantiklinalen und- synklinalen, " it would seem to us that any such analysis should surely postulate, or at least entertain the possibility of, an ancestral stage ("plan") without a fifth crest; and that, in the interpretive ]ihase of the work, the possibility that various forms with this crest may have passed independently through such a stage should be carefully considered. If the "trigonodonten Grundplan," with anterior and pos- terior cingula, protoloph and metaloph, but no mesostyle or mesoloph (Stehlin and Sehaub, 1951, Fig. 1) gave rise to the "Tkeridomys-Trechomysplan," it would seem almost (though perhaps not quite) necessary to assume that a four-crested pat- tern preceded the five-crested one. At the very least, a primitive four-crested stage should be considered as an alternative possi- bility to the postulate that the four-crested pattern is a speciali- zation formed by reduction from a five-crested one. This is one point where we feel that the * ' Theridomys-Trecho- mysplan" has been transferred from a morphologic to a phylo- genetic concept by Stehlin and Sehaub. For example, dealing with Deseadomys (" Asteromys"), they state: "Dass schon in der Deseadoformation Sudamerikas stark spezialisierte Formen vorliegen, beweist Asteromys prospicuus. . . . Die brachyodon- 410 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY ten Zahne besitzen niir noch drei Aussensynklinalen ; es f ehlt auscheinend, wie bei Acaremys, die vierte" (ojy. cit., p. 60). The transformation of one type of rodent tooth into another has clearly occurred independently on numerous occasions. For example, the change from brachyodont to mesodont or hypsodont teeth can be followed independently in the Geomyidae, Hetero- myidae, Eorayidae, Theridomyidae, Sciuravidae, and Cricetidae, to name but a few cases, and the same was true, we believe, for the Caviomorpha. In each case, the original, low-crowned mem- bers of the group have rounded, plump cusps, or distinct traces of cusps, with a minimum of connecting crests. In each case, subsequent evolution accentuates the crests, giving rise to rather similarly appearing multi-crested forms. As far as we are able to interpret the evolutionary picture of the rodents, these ani- mals represent a number of lines that evolved, independently, from ancestors that were primarily cuspate. On the basis of the time of occurrence of the earliest members of these groups, the ancestors must be sought not later than late Eocene for the geomyids, heteromyids, cricetids and caviomorphs ; not later than middle Eocene for the eomyids and theridomyids ; and not later than earliest Eocene for sciuravids. This, again, suggests to us that these various groups have developed independently of each other. We believe that the early Eocene paramyids are the most primitive known rodents. The data supporting this point of view will be included in the review of the Paramyidae in prepa- ration by Wood. Unfortunately, most published illustrations of the teeth of primitive paramyids are inadequate, but a few good ones are available (Stehlin and Schaub, 1951, Figs. 16, 17, 156, 157, 305, 306, 465 and 466). Within the Paramyidae, a series of lines can be traced, leading to greater specialization and gradual development of lophate teeth. Some of these lines we believe can be considered to lead to other, and more advanced, families (Paramys — Decti cad apis — Adelomys — Theridomys; Paramys — Sciuravus; Paramys — later paramyids — sciurids; etc.), whereas others were doubtless sterile offshoots. But we believe that this early and middle Eocene radiation of the Paramyidae provided the source for the subsequent evolution of the rodents. WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 411 It is in the lack of allowance for the known parallelism that is found among the rodents, it seems to lis, that Stehlin's analysis fails of being a completely neo-classical one, just as it is admit- tedly not a i)hylogenetic one. And it is here that the employment of the term " Theridomys-Trechomysplan," rather than a de- scriptive, non-committal name, introduces an unnecessarily complicating nomenclatural factor. This term is associated with a well-known phylogenetic hypothesis (derivation of Hystrico- morpha sensu lato from the Theridomyidae) that should never have been introduced into the morphological analysis. This phylogenetic hypothesis would be applicable at the very most only in part (Theridomyidae-Hystricomorpha sensu stricto), and even so limited an application as this now seems unlikely, since Lavocat has clearly indicated that this cannot be a true phylogenetic series for some of the Old World Hystricomorpha, and he has recently (1956, p. 54) suggested that the Theridomyi- dae are related to no known rodents except the Pseudosciuridae. We further believe that the analysis of the differences between the theridomyid and caviomorph cheek teeth given here destroys the validity of the " Theridomys-Trechomysplan" as a non- phylogenetic concept also, certainly so far as it applies to the caviomorphs. The somewhat critical tone of part of the foregoing should not be interpreted as an expression of our opinion concerning Stehlin and Schaub's work as a whole. The volume is one of the most useful publications on rodents that has ever appeared, and will, we believe, rank with Tullberg as a classic that must always be consulted, particularly by anyone interested in rodent teeth. The clear and lucid descriptions, the magnificent figures and the stimulating taxonomic discussions by Schaub ensure for it a great and enduring value. Part of our difference of opinion with Stehlin and Schaub over the primitive nature of the Paramyidae lies in the interpre- tation of the anterior end of the lower molars. This will be discussed at length by Wood in his forthcoming review of the Paramyidae, but we ])elieve that the paraconid was lost before the first appearance of any rodents in the fossil record. There is, in the paramyids, a weak crest along the anterior face of the lower molars. This might be the last remnant of the paraconid- 412 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY protoconid crest, or it might be a iieomorph, but there certainly' is no paraconid. When a cusp appears at the anterior end of lower molars later on in rodent history, we believe that it is clearly a neomorph, best called an anteroconid. This is the structure Schaub (1953b, pp. 8-9 and Figs. 11-12) calls a paraconid, and which he considers an important indication that Oligocene squirrels are the most primitive known rodents. The recent discussion of " hystricomorph " relationships by Landry (1957a) falls into an entirely different category. This starts out as a comparative-anatomical study of rodents that have been classified as hystricomorphs at one time or another, and concludes, on the basis of a number of similarities of various kinds, that they are related. The resemblances exist, and in the absence of a fossil record would indeed justify retention of the Old and New World forms in one suborder. As matters stand, however, we are compelled to differ and to regard the re- semblances as due to parallelism, not to special affinity in the sense of derivation from a common ancestral stock that was already "hystricomorph." Furthermore, we have to take excep- tion to some of Landry's reasoning. We cannot accept his view that structures appearing late in evolutionar.y history, such as the multicusped pattern of the molar teeth in some groups, are primitive. When a specialized character, which he considers as basic for the ' ' Hystricomorpha, ' ' does not occur in some early members of the group, he considers that these members have secondarily reverted to the primitive rodent condition, a hypothesis reciuiring a degree of evolutionary reversal which appears improl)able. For example, Landry con- siders the loss of the entepicondylar foramen to be a basic charac- ter for the "Hystricomorpha," stating that "... the fact that the entepicondylar foramen is almost never found in hystrico- morphs indicates that it was lost early in the history of this group" (Landry, 1957a, p. 20). He adds that the only form he knows that had such a foramen Avas the Pliocene Lagostonwp- sis, in which L. Kraglievich found it in 14 out of 16 specimens. This he interprets as meaning that the lost entepicondylar foramen was reacquired by Layostomopsis, and subsequently lost once more by its descendants. In this connection, however, he overlooked Wood's record (1949, p. 29) of the presence of this WOOD AND PATTERSON : OLIGOCENE RODENTS OF PATAGONIA 413 foramen in Plat y pit t amy s, which we consider as evidence that the foramen was present in the basic caviomorph stock, and in- dependently lost in a number of lines. It should be emphasized that this form is the only pre-Santacruzian caviomorph for which we have any knowledge of the postcranial skeleton. As another example, Landry considers that fusion of the upper ends of the tibia and fibula ' ' may have been present in ancestral hystricomorphs " (p. 19), overlooking the fact that there is no evidence of such fusion in Plat ypitt amy s, nor in the Santacruzian Neoreomys and Steiromys (Scott, 1905, pp. 397, 415). Having decided that a proximal fusion of the tibia and fibula may have been present in the ancestral "hystricomorphs," and accepting that this may be a fossorial character, he suggests that the an- cestral "hystricomorphs" may have been fossorial animals. He then states that the fossorial bathyergids possess some primitive rodent characters, such as a separate scaphoid and lunar, and that therefore they may be modified descendants of the ancestral " Hystricomorpha " (op. cit., pp. 19-20). This we think is some- what tenuous. His arguments in favor of the primitive nature of a multiplicity of cusps, of the value of the milk teeth in de- termining the primitive nature of the molar pattern, of a secondary decrease in hypsodonty in the teeth of the Bathyergi- dae, and his refusal to accept reasonably well documented evolu- tionary lines as being indicative of what really happened, seem to us to reflect a basic unfamiliarity with the fossil record and with the methods of study of fossils, without which no classifi- cation can hope to achieve a firm phylogenetic base. Landry is unwilling to accept the results of detailed phylo- genies of rodents based on tooth structure, stating ". . . it is nevertheless true that where one observer sees similarities in teeth, another sees differences. The matter is so subjective that I believe that classifications of rodents based on similarities of the occlusal surface of the teeth are useless" {op. cit., p. 89), and he cites the fact that the same type of tooth pattern has evolved many times in the rodents. He is perfectly willing, how- ever, to accept the evidence of structures where there is, and presumably always will be, no actual evidence of evolutionary sequence, such as the sacculus urethralis in the penis {op. cit., 414 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY pp. 16-17).^* He believes that the presence of multiple cuspules in isolated instances among modern "hystricomorphs" is a clue to the ancestral pattern, but that the underlying structure of the teeth is not. He believes the structure of the angular process but not that of the infraorbital foramen to be fundamental; e.g., he places the Bathyergidae in the Hystricomorpha and elimi- nates the Pedetidae. He believes that the histologic structure of the incisor enamel is important, except presumably for such forms as Pedetes that do not fit his classification. He considers the fact that there are spines on the penis or a forward opening of the pterygoid fossa in "hystricomorphs" as indicative of relationship among these forms, but disregards these characters when they occur in non-hystricomorphous rodents (e.g. a for- ward opening of the pterygoid fossa in geomyoids), or passes over the absence of one of them in a group regarded as "hys- tricomorph" (e.g. lack of a forward opening of the pterygoid fossa in Ctenodactylidae). The evidence from auditory ossicles that he advances has been discussed above (p. 292). He believes that cranial foramina are worthless in rodent classification be- cause there is considerable variability as to which nerves and blood vessels follow which paths, whereas this fact may well indicate that the cranial foramina and their contents are po- tentially very useful although at present poorly understood. In other words, in spite of his protestations to the contrary, we feel that Landry is setting up a key classification instead of a phylo- genetic one. This is justifiable when the phylogenetic data are not available, and is justifiabk^ as an adjunct to the use of data from phylogeny, but is not warranted when it is contrary to the data from the study of fossil rodents, as is the case with Landry 's proposed classification. Finally, it seems to us, Landry is guilty, on a number of occasions, of assuming that a particular con- dition held true for the hypothetical ancestral ' ' hystricomorphs, ' ' 34 We fully share Landry's interest in the reiiiarkiible distribution of the sacculus urethralis within the Kodentia. althoiif,'h we are unable to see quite eye-to-eye with him regarding the over-riding iniixirtauee he attributes to it as a proof of the unity of Hystricomorpha (.vc«a(/ latu). We note its occurrence in the r.athyergidae and the presence of a vestige (or rudiment V) of it in the Ctenodactylidae, groui)s which, with Lacovat, we would exclude from the Hyslricomorpha (or Palaeotrogoniorpha ) . Landrv does not mention Pocock's Hnding (li)22, pp. 412, 425) that the sacculus is" lacking in Lagoxtomiis — pre- sumably a secondary loss, since it is present in ( hinchiUa. This absen.^a: 1862, PZS, p. 3; 1864, ALNHNY, 8, p. 82, fig. 32 (Hab.? [Java]) Holotype MCZ 176910. [Is Batissa jayann Lea, teste Prime.] solidula Prime, Corhicula : 1860, PANSP, p. 273 [nomen nudum]; 1861, Ibid., p. 127; 1864, ALNHNY, 8, p. 81, fig. 31 (Hab. ?) Holotype MCZ 152907. solidula Prime, Cyclas: 1852, PBSNH, 4, p. 158 (Ohio, Adams, Agassiz) ; 1865, Smithsonian Misc. Coll. No. 145, p. 30, fig. 27 Holotype MCZ 19438. [Is Sphaerium striatlnum Lamarck, teste Prime.] somersetii Prime, Helix: 1853, Bermuda List, p. 55 [tiomen nudum]. soriniaiia Ileude, Corhicula: 1880, Conch. Fleuv. Nanking, pt. 10, pi. 8, fig. 49 (Ruisseau des montagnes dans la province de Kouang-tong, M. I'abbe Soi-in) 2 Paratypes MCZ 167280 ex Musee Heude. sphaerica Heude, Corhicula: 1880, Conch. Fleuv. Nanking, pt. 10, pi. 7, fig. 37 (Ruisseaux du Kieu-te Men sud) 21/0 Paratypes MCZ 167299 ex Musee Heude; 2 Paratypes MCZ 154845 from Heude to Morelet ex E. W. Roper coU'n. sphaerica 'Anthony' Prime, Cycla.s: 1853, PBSNH, 4, p. 275, text fig. (Black River, Loraine Co., Ohio) Cotypes MCZ 19524. sphaerica Prime, Cyrena: 1861, J. de Conch., 9, p. 354; Ibid., 10, p. 386, pi. 13, fig. 2 (Hab.?) Holotype MCZ 176901. JOHNSON : CORBICLTLIDAE AND SPIIAERIIDAE TYPES 473 sphaerwula Prime, Batissa: 1860, ALNHNY, 7, p. 155 [nornen mtdum] ; 1862, PZS, 22, p. 2 (Panimbang River, near Pardana, Java) Holotype MCZ 17691. [Is Batissa jayana Lea, teste Prime.] sphaericum Sterki, Pisidium : 1912, Nautilus, 26, p. 8 (Westbrook, Maine, A. H. Norton; Saco, Maine, H. M. Winkley [Cotypes MCZ 34662] ; Lynnfield, Mass., II. M. Winkley). splendidulum Sterki, Pisidium: 1898, Nautilus, 11, p. 113 (Barren Brook, Caribou, Aroostook Co., Maine, 0. 0. Nylander [Cotypes MCZ 20616 ex V. Sterki]; Saco, Maine, H. W. Winkley; Old Orchard, Maine, J. B. Henderson; Grand Eapids, Michigan, L. H. Streng [Cotypes MCZ 112574] ; Potomac River, Washington, D.C.) splendidum 'Parreyss' Prime, Pisidium: 1859, ALNHNY, 7, p. 102 [nomen nudum] "Is a Nticula," (Prime). squalida Deshaj'es, CorUoula: 1854, [1855] PZS, 22, p. 342 (Hab.? [PMlip- pines], H. Cuming). The Cotype figured by Prime 1866, ALNHNY, 8, p. 218, fig. 47 is, here selected, Lectotype MCZ 187443 ex H. Cuming. squalida Heude, Corbicula: 1880, Conch. Fleuv. Nanking, pt. 10, pi. 8, fig. 43 (Les lacs de Me-keng, Tong-lieou, sur la rive droite et sur la rive gauche, le lac Teh'ao et ses dependences) 2 Paratypcs MCZ 167303 from le lac Tch'ao ex Musee Heude. Since this name is preoccupied by Deshayes, 1854, we propose the name Corbicula clenchii. stabileii 'Schmidt' Prime, Pisidium: 1859, ALNHNY, 7, p. 97. [nomen nudum]. Listed as a synonym of Pisidium caste^-tanum Bourguignat. staminea Conrad, Cyclas: 1834, Amer. Jour. Sci., 25, p. 342, pi. 1, fig. 5 (small streams in South Alabama) Holotype Acad. Nat. Sci. Phila. 10186. Paratype MCZ 19434 ex Prime coll 'n. The Paratype is figured by Clench and Turner, 1956, Bull. Florida State Mus., 1 (3), pi. 6, fig. 5. sieenbuchi Moller, Cyclas: 1842, Naturhist. Tidsskrift, 4, p. 93 (Greenland). Figured by Prime 1878, Bull. MCZ, 5, p. 45, pi. 2, fig. 1, a-c. 21/2 Cotypes MCZ 19845 ex Prime coll 'n. sicrTcianum Pilsbry, Pisidium: 1897, PANSP, 49, p. 291, pi. 6, figs. 1-4 ("Prado" Montevideo, Uruguay) Cotypes MCZ 78076 ex B. Walker. stimpsoniana Prime, Corbicula: 1866, ALNHNY, 8, p. 222, fig. 54 (Hab.?) Holotype MCZ 152908. strengi Sterki, Pisidium: 1902, Nautilus, 15, p. 126 (Perch Lake, Michigan, L. H. Streng and Dr. Kirkland ; Reed Lake [Cotypes MCZ 154927 ex 474 BITLLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Grand Eapids Public Mus. coll'n.] and Little Bostwiek Lake, Michigan; Base Lake, Indiana, L. E. Daniels; Meyer's Lake, Ohio, V. Sterki; Little Lakes, New York, J. Lewis). striatella Deshayes, Corbicula: 1854 [1855], PZS, 22, p. 344 (Pondicherry [India], H. Cuming). It is doubtful if the specimen figured by Sowerby [in] Reeve, 1875, Conch. leonica, 20, Cyrena, pi. 11, fig. 49 is a type, since it is from Hanley. The largest CotjT)e, therefore, is, here selected, Lectotype, British Museum (Nat. Hist.) 1956.12.3.20, PI. 7, fig. 3, also Paratype BM(NH) 1956.12.3.21, 5 Paratopes MCZ 152920 from H. Cuming ex Prime coll'n. striatella 'Ferussac' Prime, Cyclas: 1865, Smithsonian Misc. Coll. No. 145, p. 54 [nomen nudum]. Listed as a synonym of Sphaerium modioli f or mia Anton. strictwm Normand, Sphaerium : 1854, Cyclades, Dept. du Nord., p. 3 (habits les fosses de forets de Kaismes et de Vicoigne, et ceux des marais d'Helesmes [France]) 1 Cotype MCZ 19516 labeled "France," ex Prime coll'n. sublobata Deshayes, Cyrena: 1854, PZS, 22, p. 18 (Hab.? [New Caledonia], H. Cuming) Holotype figured by Sowerby [in] Reeve, 1875, Conch. Iconica, 20, Cyrena, pi. 10, fig. 39 [40]. British Museum (Nat. Hist.) 1956.12.3.22. "Some of the erosion pits have been left out" (Wilkins). Also Paratype BM(NH) 1956.12.3.23. 5 Paratypes MCZ 176898 (labeled Louisiade Archiapelago) from H. Cuming ex Prime coll'n. subquadrata Heude, Corbicula : 1880, Conch. Fleuv. Nanking, pt. 10, pi. 8, fig. 45 (La Houai moyenne [Hong-tse] ) 4 Paratypes MCZ 167262 ex Musee Heude. 2 Paratypes MCZ 154839 from Heude to Morelet ex E. W. Roper coll'n. subradiata 'Kerr' Prime, Corbicula: 1861, PANSP, p. 127; 1864, ALNHNY, 8, p. 75, fig. 23 (Agra, India) Holotype MCZ 152914. subrotundum Sterki, Pisidium: 1906, Nautilus, 20, p. 19 (New England, Anticosti Island to Michigan. Jupiter River, Anticosti Id., Albany River, Ontario, W. Mclnnes ed., both sent by J. F. Whiteaves; *Vme- yard, " Pawtuxet, R. I., J. F. Perry; several places in Aroostook Co., Maine [Mud Lake, West Manland], 0. 0. Nylander, [Cotypes MCZ 20615 ex 0. 0. Nylander] ; Pine River, Marquette Co., axid Oswald Lake, Michigan, B. Walker; . . .) subtransversum Prime, Sphaerium : 1860, PZS, 28, p. 322 (^ Tohasco, Mexico, IL Cuming) Holotype British Museum (Nat. Hist.) No. 1952.8.20.41 JOIIXSON : COKBICULIDAE AND SPIIAERIIDAE TYPES 475 figured by Sowerby, [in] Reeve, 1875, Conoli. leonica, 20, Sphaerium, pi. 4, fig. 38. sulcatina Deshayes, Corhicula: 1854 [1855], PZS, 22, p. 345 (Hab.?, II. Cuming). The Cotype figured by Prime, 1864, ALNHNY, 8, p. 70, fig. 28 [China] is, here selected, Lectotype MCZ 187458 ex II. Cuming, since it precedes, Sowerby [in] Reeve, 1875, Conch. Iconica, 20, Cyrena, pi. 14, fig. 70. sulcatum 'Parreyss' Prime, Pisidiwm: 1859, ALNHNY, 7, p. 95 {^tiomrn nudum]. Listed as a synonym of Pisidium amnicum Muller. siilculosa ' Charpentier ' Prime, Cyclas: 1860, PANSP, p. 300 [nomrn nudum]. Listed as a synonym of Si)haerium fahale Prime. sumatrensis 'Ferussac' Prime, Cyclas: 1869, AJC, 5, p. 164 [nomen nudum]. sumatrensis Prime, Glauconome : 1862, J. de Conch., 9, p. 384, pi. 14, fig. 3 (Sumatra) Lectotype, here selected, MCZ 176956. 3 Paratypes British Museum (Nat. Hist.) 1952.8.20.38-40. ienpJhnn Gould, Pisidium: 1850, [in] Agassiz, Lake Superior, p. 245 (Fort William, Michipicotin, Ontario). lu comparison to Pisiduni duhium Say, Ciould says, "They are smaller, more elevated, less sulcated, and the hinge is less robust. I had designed to apply to them the specific name P. tenellum, but unfortunately the specimens were mislaid before I could examine them with sufficient care to give the characters with the requisite precision." Idiotype MCZ 19828, ex Prime coll'n. tenuis Prime, Cyclas: 1852, PBSNH, 4, p. 161 (small stream in the vicinity of New Bedford, Mass.); 1853, Ibid., 4, p. 285 (Mass., Stimpson ; Androscoggin River, Maine, Girard) ; 1865, Smithsonian Misc. Coll. No. 145, p. 47, fig. 44. ' ' It was discovered some years since by Mr. Girard, from whom I obtained my specimens." 1 Cotype MCZ 19474, from tiirard ex Prime coll'n. [Is Sphaerium occidentaJe Prime, teste Prime.] tenuistrata Prime, CorUcula: 1860, PZS, 28, p. 322 (Hab.? H. Cuming); 1861, J. de Conch., 9, p. 40, pi. 2, fig. 3. The type was figured by Sowerby [in] Reeve, 1875, Conch. Iconica, 20, Cyrena, pi. 15, fig. 80. Holotype British Museum (Nat. Hist.) 1952.8.20.34. tenuistrata Prime, Cyclas: 1852, PBSNH, 4, p. 156 (Tennessee). Prime states: (1853, PBSNH, 4, p. 272) "Besides the specimen T possess (Alabama, Ingalls, Idiotype MCZ 19429), I have seen but one other, which was kindly sent to me for description by Prof. C. B. Adams." Holotj'pe MCZ 177106 ex C. B. Adams coll'n. [Is Sphaerium striatinnm Lamarck, teste Prime.] 476 BLTLLETIN: MUSEUM OF COMPARATIVE ZOOLOGY ierveriana Diipuy, Cyclas: 1849, Cat. Extram. Galliae Test. No. 87; 1852, Moll, de France, p. 674, pi. 29, fig. 9. (les mares des environe d'Auch, a la Boubee et a la Hourre, Dept. de Gers, France) 5 Cotypes MCZ 19;158 ox Prime coll'n. [Specific locality not mentioned.] tetragonum Normand, Fisidium: 1854, Cyclades, Dept. du Nord, p. 5 (Habite les fosses des marais a Valenciennes, . . . aussi dans la foret de Raismes) 3 Cotypes MCZ 19925 from Valenciennes ex Prime coll'n. thermale Diipuy, Fisidium : 1849, Cat. Extram. Galliae Test. No. 238 (les eaux thermales dans les Pyrenees, a Bagneiures de Bigorre, . . . a Cauterets, pres de la RaDlere et de Mahourat [Dept. de Creuse, France]) 2 Cotypes MCZ 19900 ex Prime coll'n. [Specific locality not mentioned.] titicacense Prime, Fisidmm: 1895, Cat. Corbiculadae in MCZ, p. 55 (Lake Titieaca, Peru) Cotypes MCZ 19853 [nomen nudum]. toxcutcnsis Kruimcl, Corhicida: 1918 [in] Abendanon, Geol. eu Geog. door- kruisingen van Midden-Celebes, 4, p. 1645, pi. 27, fig. 5 (Towuti-See, Sud-Ost-Celebes, Abendanon) 2 Paratypes MCZ 53503 ex Zoological Museum, Amsterdam, Netherlands. trapezoideum Sterki, Fisidium: 1896, Nautilus, 9, p. 12.'i (. Michigan: many places in the Upper and Lower Peninsula; Soutlieru ^Minnesota; Pedan River, Canada; Philadelphia, Pa.; Adamsville, New Jersey [Cotypes MCZ 112579 ex E. W. Roper coll'n] ; Comal Co., Texas). triangularis Deshayes, Corbicida: 1854 [1855], PZS, 22, ]). 345 (Hab. ?, H. Cuming). Not figured by Sowerby [in] Reeve, 1875, Conch. Iconica, 20, Cyrena. Lectotype, here selected, British. Museum (Nat. Hist. ) 1957.6.25.22, PI. 7, fig. 2; 1 Paratype MCZ 176923, from H. Cuming ex Prime coll'n. tribunalis Prime, Cyrena: 1869, AJC, 5, p. 148 [nomeji nudum]; 1870, ALNHNY, 9, p. 300 ([Tecames River], Ecuador) Holotype MCZ 176888, PI. 5, fig. 4. trigona Deshayes, Corbicida: 1854 [1855], PZS, 22, p. 344 (Pondicherry [India], H. Cuming). The eotype sent to Prime by Cuming, and figured, 1866, ALNHNY, 8, p. 221, fig. 53 is, here selected, Lectotype MCZ 152922, smce it precedes Hanley & Theobald, 1866, Conch. Indica, p. 60, pi. 155, fig. 7 and Sowerby [in] Reeve, 1875, Conch. Iconica, 20, Cyrena, pi. 11, fig. 43. triquctra Deshayes, Batissa: 1854, PZS, 22, p. 13 (Philippine Islands; Australia, H. Cuming). The eotype sent to Prime by Cuming from the Philippine Islands and figured, 1866, ALNHNY, 8, p. 230, fig. 61 is, JOHNSON : CORBICULIDAE AND SPIIAERIIDAE TYPES 477 here selected, Lectotype MCZ 119013 since it precedes the type figured by Sowerby [in] Eeeve, 1875, Conch. Iconiea, 20, Cyrrva, pi. 2, fig. 3. Paratypes British Museum (Nat. Hist.) 19.17.().2o.l8 19. tumidn Deshayes, Corhicula: 1854 [1855], PZS, 22, p. 343 (Borneo, H. Cuming). The Cotype sent to Prime by Cuming and figured, 18fi6, ALNHNY, 8, p. 219, fig. 50 is, here selected, Lectotype MCZ 187444. Since it precedes the type figured by Sowerby, [in] Reeve 1875, Conch. Iconiea, 20, Cyrena, pi. 17, fig. 92. tumida Prime, Cyrena: 1860, PANSP, p. 286. New name for Cyrena angulata Deshayes 1854, non Roemer 1835. uUramontanum Prime, Pisidium: 1865, Smithsonian Misc. Coll. No. 145, p. 75, fig. 85 (Canoe Creek, California) Cotj-pes MCZ 19847. ■ttncinulata Heude, Corhicula: 1880, Conch. Fleuv. Nanking, pt. 10, pi. 2, fig. 13 (Canaux des districts de Li-yang, Ti-hing) 2 Paratypes MCZ 167290 ex Musee Heude. 1 Paratype MCZ 154847 from Heude to Morelet ex E. W. Roper coU'n. iinioniformis Prime, Batissa: 1860, PZS, 28, p. 319 (Hab.? H. Cuming) Holotype British Museum (Nat. Hist.) 1952.8.20.18 figured by Sowerby, [in] Reeve 1875, Conch. Iconiea, 20, Cyrena, pi. 3, fig. 9. Idiotype MCZ 176909 from the Fiji Islands. [Is Batissa violacea Deshayes, teste Prime.] variaUle Prime, Pisidium: 1852, PBSNH, 4, p. 163; 1852, BJNH, 6, p. 351, pi. 11, figs. 7-9 (Fresh Pond, Cambridge, Mass.) Cotypes MCZ 19805. varia/ns 'Carpenter' Prime, Cyrena: 1865, Smithsonian ^lisc. Coll. No. 145, p. 22 [nomen nudu^n]. Listed as a synonyn of Cyrena mexicana Sowerby. variegata Heude, Corhicula : 1880, Conch. Fleuv. Nanking, pt. 10, pi. 8, fig. 44 (La Ilouai moyenne) 2 Paratypes MCZ 167256 ex Musee Heude. variegata D'Orbigny, Cyrena: 1835, Mag. de ZooL, 5, p. 44 "Synopsis Terr, et Fluv. Moll. Amer. Merid. " (fluniina rcpub. Uruguayensis orientalis, nee non Paranaense flumeu, a provincia Buenos Ayres ad Missionum provinciam) ; 1846, Voyage Amer. Merid., p. 567, pi. 82, figs. 15-17. 1 Paratype MCZ 143097 from D'Orbigny (former locality) "fide H. Cimiing ex Prime coll'n. veatleyii C. B. Adams, CycJas: 1849, Cont. to Conch., p. 44 (Jamaica) Cotype MCZ 73920 from Adams ex Prime coll'n. venesuelonsis Prime, Cyclas: 1860, PANSP, p. 294 [nomen nudum]. Listed as a synonym of Sphaeriam maculatum Morelet. Later, Prime, 1865, 478 P.ITLLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Smithsonian Misc. Coll. No. 145, p. 54, listed this name under Sphaerium modioliforme Anton. ventricosa Prime, Corhicula: 1860, PANSP, p. 274 [nomen nudum]; ISG.!, Smithsonian Mise. Coll. No. 145, p. 3. Listed as a synonj-m of CorhicuJa convexa Deshayes. ventricosa Deshayes, Cyrena: 1854, PZS, 22, p. 16 (Philippine Islands. Australia, H. Cuming) Holotype figured by Sowerby [in] Eeeve, 1875, Conch. Tconica, 20, Cyrena, pi. 9, fig. 35, British Museum (Nat. Hist.) 1956.12.3.24; also Paratype BM(NH) 1956.12.3.25; 1 Paratype MCZ 176897 (Philippines) from H. Cuming ex Prime coll'n. ventricositm Prime, Pisidium: 1851, PBSNH, 4, p. 68; 1852, BJNH, 6, p. 355, pi. 11, figs. 16-18 (small stream running out of Fresh Pond, Cambridge, Mass.) Cotypes MCZ 19814. venustula Prime, Corbicula: 1864, ALNHNY, 8, p. 73, fig. 21 (Manilla, Philippine Islands) Holotype MCZ 135636. [Is Coriicula cumingii Deshayes, teste Prime.] vermontana Prime, Sphaerium: 1861, PANSP, p. 128 (Vermont, C. B. Adams) ; 1865, Smithsonian Misc. Coll. No. 145, p. 42, fig. 37 (Lake Chaniplain and Memphremagog) Holotype MCZ 19440. [Is Sphaerium striatinum Lamarck, teste Prime.] vicina Heude, CorhicuJa: 1880, Conch. Fleuv. Nanking, pt. 10, pi. 3, fig. 15 (La riviere de San-ho, district de Ho-fe, Lu-tcheon-fou) 4 Paratypes MCZ 167259 ex Musee Heude. vintaense Call, Sphaerium : 1886, Proc. Davenport [Iowa] Acad. Nat. Sci., 5, p. 8, fig. 4 (Lake in the Vinta Mountains, Utah, elevation 10,500 feet) Holotype MCZ 4859 ex E. E. Call coll'n. violacea Prime, Corbicula: 1860, PANSP, p. 274 (Hab.?) [nomen nuduvi]; 1861, PANSP, p. 128 (Hab.?) Measured specimen, here selected, Lecto- type MCZ 152899. [Is Corbicula trigona Deshayes, teste Prime.] viridante 'Morelet' Prime, Sphaerium: 1865, Smithsonian Misc. Coll. No. 145, p. 57, fig. 59 (Pointe a-Pitre, Guadeloupe) Cotypes MCZ 73919. It cannot be determined which is the figured specimen. vulgaris Prime, Corbicula: 1866, ALNHNY, 8, p. 223, fig. 55 (Hab.?) Holotype MCZ 152906. [Is Corbicula largillierti Deshayes, teste Prime.] walTceri Sterki, Pisidium: 1895, Nautilus, 9, p. 75 (Kent County, Michigan, B. Walker; Utica, Michigan; Grand Eapids, Michigan, L. H. Streng; Columbia, Pa., B. Walker; Mohawk, New York, in E. W. Eoper's coll'n, JOHNSON : CORBICULIDAE AND SPHAERIIDAE TYPES 479 from E. E. Mayo [Cotypes MCZ 112551] ; Clearwater and Mississippi Rivers, Minn., H. E. Sargent). vatsoni Paiva, Pisidiitm: 1866, J. de Conch., 14, p. 340, pi. 9, fig. 3 (Madeira) P:iratype.H MCZ 20023 ex Ic Baron de Castello de Paiva. irillcinsii Johnson, Cyrena: (new name for Cyrcna simllis Deshayes 1854, non Gray 1834) see under: similis Deshayes, Cyrena. winldeyl Sterki, Musculium: 1909, Nautilus, 23, p. 66 (Old Orchard, Maine; Danvers, Massachusetts, H. W. Winkley). Sterki selected the former place as the type locality. Cotypes MCZ 84751, ex H. W. Winkley coll'n. Specimens from the latter locality also ex H. W. "Winkley are MCZ 34745. woodiana lica, Cyrena: 1834, Trans. American Phil. Soc, 5, p. 110, pi. 18, fig. 55; 1834, Obs. Genus Unio, 1, p. 222, pi. 18, fig. 55 (Canton [China], W. W. Wood). An Idiotype MCZ 152936 was figured by Prime, 1866, ALNHNY, 8, p. 227, fig. 59. zeteJci Pilsbry, Polymesoda : 1931, Nautilus, 44, p. 85, pi. 7, fig. 2-2a (near Chama, Panama, J. Zetek) 5 Paratypes MCZ 81073 ex J. Zetek. zonatum Prime, Pisidium: 1852, PBSNH, 4, (Fresh Pond, Cambridge, Mass.) ; 1852, BJNH, 6, p. 364, pi. 12, figs. 17-19. Cotypes MCZ 19831. [Is Pisidium ahditum Haldeman, teste Prime.] zonatum Prime, SpJiaeriuyn : 1860, PANSP, p. 301 (New Zealand) [nomen nudurn]. PLATE 1 Portrait of Temple Prime taken from his own copy of his complete works on Mollusca. This volume is now in the library of the Museum of Compara- tive Zoology. PLATE 1 PLATE 2 Fig. 1. Batissa elangata Prime. New Caledonia. Holotvpe BM (NH) 1952.8.20.26 (Natural size). Fig. 2. An original Prime label. The number in the upper left hand corner is a reference to his ' ' Catalogue of the species of Corbiculadae in the collection of Temple Prime, now forming part of the collection of the Museum of Comparative Zoology, Cambridge, Massachusetts" published in 1895. v^. S^ri ^ PLATE 2 // PLATE 3 rig. 1. Corbicula semisulcata Deshayes. Victoria River, Australia. Holo- type BM (NH) 1956.12.3.19 (2 x). Fig. 2. Batissa corbiculoides Deshayes. New Guinea. Leetotype BM (NH) 1957.6.25.4 (Natural size). Fig. 3. Cyrena mactroides Deshayes. Hab.? Holotype BM (NH) 1956.12.3.33 (Slightly reduced. Actual length 98 mm.). PLATE 3 PLATE 4 Fig. 1. Cyrena hnninca Prime. Hab.? TTolotype BM (NH) 1952.8.20.32 (Natural size). Fig. 2. Cyrena sinuosa Deshayes. Paningbang River, Java. Lectotype BM (NH) 1957.6.25.6 (About % natural size. Actual length 71 mm.). Fig. 3. Cyrena lauta Deshayes. Hal).? Lectotype BM (NH) 1956.12.3.11 (About % natural size. Actual length 68 mm.). Fig. 4. Cyrena nitida Deshayes. Borneo. Holotype BM (NH) 1956.12.3.34 (About % natural size. Actual length 74 mm.). PLATE 4 PLATE 5 Fig. 1. Cyrenu oiHfonnis Ttenhiiyes. Port Essington [Australia]. Ilolotype BM (NH) 1956.12.3.36 (Natural size). Fig. 2. Cyrena arctata Deshayes. [Lake] Maracaibo [Venezuela]. Lectd- type BM (NH) 1956.12.3.28 (Slightly enlarged. Actual length 30 mm.). Fig. 3. Cyrena affinis Deshayes. Australia. Holotype BM (NH) 1956.12.3.26 (About % natural size. Actual length 55 mm.). Fig. 4. Cyrena trihimalis Prime. [Tecames Eiver], Ecuador. Holotype MCZ 176888 (About % natural size. Actual length 55 mm.). Fig. 5. Glauconome jayana Prime. Australia. Lectotype American Museum of Natural History, New York, 31537 (About 1% x. Actual length 65 mm.). PLATE 5 PLATE 6 Fig. 1. Corbicula bcnsonii Deshayes. Bengal. Leetotype BM (NH) ]!)n(i.ll^.3.31 (2 x). Fig. 2. Corbicula gubernatoria Prime. Saigon, Cambodia. Holotype BM (NH) 1893.2.4.1-408 (2 x^. Fig. 3. Cyrena similis Deshayes. Manila, Philippines. Holotype BM (NH) 195ti.6.25.3 (Natural size). Fig. 4. CnrbicuJa amasonica 'Anthony' Prime. Amazon Eiver, Brazil. Holotype MCZ 175641 (IVo x. Actual length, 16 mm.). Fig. 5. Corbicula consularis Prime. Malacca. Holotype MCZ 152911 (IV2 X. Actual length 17 mm.). Fig. 6. Cyrena chilina Prime. Chile. Holotype MCZ 176883 (l^/i x. Actual length 52 mm.). PLATE 6 PLATE 7 Fig. 1. Corhicula angasi Prime. Murray River, South Australia. Leoto- typeMCZ 176917 (5 x). Fig. 2. Corbicula triangularis Deshayes. Hab. ? Lectotype BM (NH) 1957.6.25.22 (2 x). Fig. 3. Corhicula straitella Deshayes. Pondiuherry [India]. Lectotype BM (NH) 1956.12.3.20 (IVa x. Actual length 19 mm.). Fig. 4. Spliaerium cosiaricense Prime. Lake Yuriria, West Costa Rica. Lectotype United States National Museum 37251 (6 x). Fig. 5. Cyrena flava Prime. Hab.? Holotype BM (NH) 1952.8.20.30 (1^/4 X. Actual length 45 mm.). Fig. 6. Cyrena aeqidlateralis Deshayes. Cayenne [Mazatlan, Mexico]. Lectotype BM (NH) 1956.12.3.1 (About 1% x. Actual length 30 mm.). PLATE 7 PLATE 8 Fig. 1. Pisidium retusum Prime. Honduras. Leetotype MCZ 19848 (13 x. Actual length 3.25 mm.). Fig. 2. Cyrena germane I'lime. Panaco River, Tampico, Mexico. Holo- type MCZ 176944 (About l-'f, x. Actual length 43 mm.). Fig. 3. Corhicnla rotunda Prime. Surinam River [Dutch (uiiauaj. Holotype MCZ 176913 (5 x). Fig. 4. Corhicula colonialis Prime. Java. Holotype MCZ 135633 (ly^ x. Actual length 23 mm.). Fig. 5. Batissa compie.s.'ia Prime. Hurneo. Holotype BM (NH) 1952.8.20.20 {ji natural size. Actual length 68 mm.). Fig. 6. Corhicula baronlali.s Piime. Port Morton [Morton Bay] Australia. Holotype MCZ 176936 (1% x. Actual length 17 mm.). I'lA'l'K S Bulletin of the Museum of Comparative Zoology AT HARVARD COLLEGE Vol. 120, No. 5 STUDIES ON THE ANT FAUNA OF MELANESIA V. THE TRIBE ODONTOMACHINI By Edward 0. Wilson Biological Laboratories, Harvard University With Two Plates CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM May, 1959 Publications Issued by or in Connection WITH THE MUSEUM OF COMPARATIVE ZOOLOGY AT HARVARD COLLEGE Bulletin (octavo) 1863 — The current volume is Vol. 120. Breviora (octavo) 1952 — No. 109 is current. Memoirs (quarto) 1864-1938 — Publication was terminated witli Vol. 55. Johnsonia (quarto) 1941 — A publication of the Department of Mollusks. Vol. 3, no. 38 is current. Occasional Papers of the Department of Mollusks (octavo) 1945 — Vol. 2, no. 22 is current. Proceedings op the New England Zoological Club (octavo) 1899-1948 — Published in connection with the Museum. Publication terminated with Vol. 24. The continuing' publications are issued at irregular intervals in num- bers Avhich may be purchased separately. Prices and lists may be obtained on application to the Director of tlie Museum of Comparative Zoology, Cambridge 38, Massachusetts. Of the Peters "Check List of Birds of the World," volumes 1-3 are out of print ; volumes 4 and 6 may be obtained from the Harvard Uni- versity Press; volumes 5 and 7 are sold by the Museum, and future volumes will be published under Museum auspices. Bulletin of the Museum of Comparative Zoology AT H A K \' A li I ) CO L L K (i K Vol. 120, No. 5 STUDIES ON THE ANT FAUNA OF MELANESIA V. THE TRIBE ODONTOMACHINl By Edward O. Wilson Biological Laboratories, Harvard University With Two Plates CAMBRIDGE, MASS., U.S.A. PRINTED FOR THE MUSEUM May, 1959 Xo. 0 — Studies on the Ant Fauna of Melanesia V. The Tribe Odontomaehinl^ Jiv Edvvakd ( ). WiLSOX Riolo^ic;il r>;ili(ir;it()li('s, i l;i iv;i iil riiivci'sitv Till' oiloiituiuacliiiu' fauna of .Melanesia is the larji'est and most diversified of any comparable ai-ea of tlie world. Ei-zhteen s|)p- oies of OdoiitniiKirlnis and five of Anochetus are known IVoiii Xew (luinea alone. The evolutionary history of Odontoniaelius seems to liave i)ui'sued a pattern eommon for the Formieidae in tliis part of the woi'ld : initial eolonization from southeastern Asia, followed by extensive radiation on Xew Guinea and eoneurrent penetration l)y some of the more successful stocks into outer Melanesia and Australia.- Three species, aeictdafus, eeplialotes, and saevissiniu.s, ap|)ear to have be^un a secondary ex))ansion out of X>w Guinea aci'oss the ^Moluccas; rephalotes has in addition entered Queensland. The Papuan species of Anochetus are for the most part peculiar to this region, with no clear affinities to any known Oriental or Australian stocks. It is uncertain whether they have been derived ultimately from (>arly invaders from southeastern Asia, as is the ease with most jireeinctive Odonto- inaeJius species. The Odontonuirhus saccissinius group is the largest and most varied odontomachine species group in Melanesia. Of the eleven species found on X^'ew Gninea, two (niald/nus, papuanus) are closely related to or conspecitic with widespread Oriental species; the remaining nine are truly preeinctive. 0. )naJignus is un- usual among the odontomachines, and Melanesian ants generally, in its preference for littoral habitats. Possibly as a result of this exceptional ecological distribution, it is one of the most widely dispersed of all Indo-Australian ants, ranging from SaraAvak in the west to Aru and Santa Cruz in the east. New Caledonia is occupied by only tAvo odontomachine spe- cies, Odontuniachus siiuillinius and Anochetus e/raeffei. Both of these species very likely have been introduced onto the island by 1 I'l-cvinus p.-irts dl" this series were piildislnMl in tlic Bulletin nf tlie >nisemii (if Coinpariitive Zoology, lis :l(il-1.-.:i. iiiid 1 1'.i :;!ii:i-:{71 (I'.tns). -Wilson, K. (I. lit.")!!. Ailiiptive sliil't .iinl dispersal in .1 tropicil ant fauna. ICvolution, in press. 484 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY luaii, since they are widespread in other parts of the Pacific, and on New Caledonia occur mostly in disturbed habitats. The author wishes to express his gratitude to the following persons for making possible the examination of many of the critical type specimens : AV. Forster, Zoologische Staatssamm- lung, Munich ; D. Guiglia, Museo Civico di Storia Naturale. Genoa; G. E. J. Nixon, British Museum (Natural History): E. S. Ross, California Academy of Sciences, San Francisco; E. Tayh)r, Oxford University Museum. The Genus OdONTOMACHUS Latreille List of the Melanesian and Moluccan Odontomachus, Ar- ranged in Species Groups and Including New Synonymy. Group of 0. saci'issimus Fr. Smith angulatus Mayr aMimosKS Fr. Smith emery i Mann gressitfi Wilson impcrator Emery latissinius Viehmeyer linae Donisthorpe malign us Fr. Smith = Odontomachus tuherculatus Koger montanus Stitz opaculus Viehmeyer 2)ap nanus Emery = Odontomachus papuanus var. concentricus Emery rufi t h 0 ra x E me r y saevissinnis Fr. Smith == Odontomachus impcrator var. taucrni Stitz = Odontomachus transversostriafus Donisthorpe Group of 0. simiUimus Fr. Smith aciculatus Fr. Smith = Odontomachus ruficcps ccphalotcs var. vcrticillatus Stitz aeneus Emery cephalotes Fr. Smith WILSON : ANTS OP MELANESIA. V. 485 = Odonto))iaclnis rKJiccps c< phalotcs var. criienta Emery = Odontomachus ruficcps ccpJudotcs var. fusca Emery -= Odontomachus rujic(})s cepJudotes var. tcrnatensis Forel = Odontomnchus n(ficcps cephalotes var. tamensis Stitz = OdontontacliHs ntficcps subsp. aruamis Karawajew = Odontomachus riificcps cephalotes var. lonisimu.'< and ximiUiviUH groups: this tiTiii is m)t ust'd to refer to any tini'r sculptural details such as rugae or striae. The standardized measurements used have alreadv been delined in au earlier paper (Bull. Mus. Comp. Zool., 116:3.55, VJ'u). 486 lUTLLETIN : MUSEt'M OI'^ COMPARATIVE ZOOLOGY Body of a different cdlor 3 3. Entire liead l)lackisli lirown; the remainder of the liody testaceous, with local infuseations on the gaster nigriccpn Fr. Smith At most only the central portion of the head dark in color, and this area ranges from light brown to blackish brown; the remainder of the body testaceous, with local infuscations on the gaster . testaceus Fr. Smith 4. Central portion (disc) of tirst gastric tergite striate 5 Entire first gastric tergite completely smooth or partly shagreened 7 5. Much of body surface showing pronounced metallic reflections aeneuft Emery Body surface completely lacking metallic reflections (i (i. Sculpturing of first gastric tergite in addition to central striae con- sisting chiefly of either coarse shagreening or striation ; if the latter, then the individual striae are stiongly cuived and many run longi- tudinally for variable distances ccphalotes Fr. Smith Sculpturing of first gastric tergite consisting of striae that are at most slightly curved and always transverse ariculatii.s Fr. Smith 7. Most of the pronotum, including all of its dorsal surface exclusive of the anterior "neck," comidetely lacking striae, its surface either shagreened or smooth and shining 8 Most or all of the pronotum covered by striae; in occasional specimens a limited transverse strip may lack striae aiul l)c smooth and shining, but in these individuals the greater part of the ijronotum is still striate 12 8. Occiput bearing distinct tumosities on either side of the median line of the head one-third the distance from the median line to the occipital corners; color uniformly yellowish orange inalic/iius Fr. Smith Occiput lacking tumosities; color other than described above 9 9. Posterior border of basal portion of ]ietiolar spine viewed exactly from the side strongly convex, contrasting markedly with the weakly con vex anterior border (Fiji Islands) anguJatus Mayr Posterior border of basal portion of petiolar spine viewed exactly from the side only weakly convex, appearing very similar in this respect to the anterior border (New Guinea) 1'* 10. Seen exactly from the side, the anterior face of the petiolar node forms an angle of ll!(i^-l.';0° with the doisal node face (for a similar con dition, see drawing of impuann.'s in Fig. 1) hitissimus Yiehmeyei Seen exactly from the side, the anterior face of the petiolar node forms an angle of approximately l(ii>" with the dorsal node face (see draw- ing of impcrntur in Fig. 1) 11 11. Kxtraocular furrows at least partly striate; head and gaster black, alitrunk ;ind petiole red rufithorax Emery WILSON : ANTS OF MELANESIA. V. 487 Extraocular furrows couipk'tt'ly lacking striai'; liody uniformly dark reddish brown impcrator Emery 12. Striae covering almost the entire dorsum of the head 13 Striae covering at most the interocular depression and jiarts of the ocular ridge and e.xtraoeular furrow, and strongly developed only in the interocular depression 15 13. Entire body concolorous blackish brown (widespread over all Melanesia and the Moluccas) simUlimus Fr. Smith Head, alitrunk, and petiole dull dark reddish yellow; gaster dark red- dish brown (New Guinea only) 14 Fig. 1. Side views of petioles of worker syntypes of Odontomachus im- perator Emery (A) and 0. papuaitu.s Emery (B) ; based on specimens in the Emery Collection, Genoa. 14. .Mesepisternum comi)letely smooth and shining except for the anterior sixth of its length, which is vertically striate; smaller species, IIW of unique type 2.08 mm (Manokwari, Xeth. New Guinea) animosus Fr. Sniitli Mesepisternum completely covered by dense vertical striae; larger spe- cies, HW of syntype examined 2.*i4 mm (Sepik Watershed, N-p]. New Guinea) montanus Stitz 15. Extraocular furrows partly striate, although occasionally the striae arc very feeble and limited to the inner one-fifth of the furrows 16 Extraocular furrows comjiletely smooth and shining 18 488 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY 16. Head, alitrunk, and petiole light reddish brown; gaster medium red- dish brown limae Donisthorpe (pnrtim) Body concolorous dark to blackish brown 17 17. Striae limited to upper fifth of extraocular furrows; anterior and anterodorsal faces of petiolar node grading into one another thrcmgli an even curve without any sign of an intervening angle papitanus Emery (partivi i Striae covering approximately the upper half of the extraocular fur row ; anterior and anterodorsal petiolar node faces meeting in a distinct angle of about 100° opaculus Viehmeyer 18. Anterodorsal surface of petiole, exclusive of the spine, transversely striate; color of head and alitrunk yellowish to light reddish brown, with the possible exception of the mesonotum, which is occasionally (in gres»itti) medium to dark reddish brown 19 Either the anterodorsal surface of the petiole is completely smooth, or the color of the head and alitrunk is dark reddish brown, or both 20 19. Head, alitrunk, and petiole uniformly light reddish brown; Inrgei species, HW of two specimens examined 2.52-2.54 mm linae Donisthorpe (partiin i Either head, mesonotum, and gaster dark reddish brown, contrastiuf; with the yellowish brown alitrunk (exclusive of mesonotum) and petiole (Central Highlands of New Guinea), or body concolorous yellowish brown, the mesonotum a shade darker than the rest (Guadalcanal) ; smaller species, HW of two type specimens 2.1G mm gressitti Wilson 20. Posterior face of petiole, extending from the tip of the spine to the posterior peduncle, evenly concave when viewed from the side saevissimus Fr. Smith (partim > Lower half of posterior face of petiole appearing distinctly convex when viewed from the side 21 21. Mesepisternum almost completely smooth; body and antennne dark reddish brown, legs yellowish brown (New (iuinea) papnaniift Emery Mesepisternum completely striate; coloration nut as above 22. Head and gaster dark reddish brown; alitrunk and petiole light yel- lowish red, with the mesonotum lightly infuscated (Solomon Islands) emeryi Mann Gaster medium reddish brown, only slightly darker tli.-iii the .'ilitnnik (head color unknown) (New Ireland variant) saevissimus Fr. Smith (pdrtiiii i ).) WILSON : ANTS OF MELANESIA. V. 489 Odontomaciius aciculatus Fr. Smith (Fig-. 4, no. 12) OdontoiiMchus aciculatus Fr. Smith, 1863, J. Proc. Linn. Hoc. London, ZooL, 7:19, worker. Type, locality. Misol. Mayr, 1876, J. Mus. Godeff. Hamburg, 12:86; worker. Donisthorpc, 1940, Entomologist, 70:109; worker. Odontomaehus cephalotes var. aciculatus, Emery, 1887, Ann. Mus. Civ. Stor. Nat. Genova, 4:428, worker, dist. Odontoirmchus ruficeps var. aciculata, Emery, 1911, Gen. Ins., 118:115. Odontomaciius ruficeps cephalotes var. verticillatus Stitz, 1912, Sitzber. Ges. Nat. Freunde Berlin, 9:501-503, figs. 5, 6, worker. Type locality: New Guinea (ex Kaiserin-Augustafluss Exped.). NEW SYNONYMY. (Syntype examined — MCZ.) Material examined. NEW GUINEA: verticillatus syntype. Odontomaciius aeneus Emery Odontomaciius aeneus Emery, 1910, Nova Guinea, 5(4) Zool.: 534, worker. Type locality: Wendesi, Neth. New Guinea. Known from only the unique type worker. This species shouhl be easily disting-uished from all other Melanesian Odontomachus through its remarkable metallic coloration: "La tete est bronzee avec des reflets violaces, le corselet est de la meme teinte, mais plus foncee, le petiole et le gastre noir-bronze, les pattes sont brun fonce." Odontomachus angulatus Mayr Odontomachus angulatus Mayr, 1866, Sitzber. Akad. Wiss. Wien, 53:500, fig. 10, worker. Type locality: Ovalau, Fiji Islands. Mann, 1921, Bull. Mus. Comp. Zool., 64:427, fig. 11, queen, ecology, distribution. Material examined. FIJI ISLANDS: Nadarivatu, Viti Levu (W. M. Mann) ; Vunidawa, Viti Levu (N. L. II. Krauss) ; Andu- bangda, 300-500 m., Ovalau (E. C. Zimmerman). Odontomachus animosus Fr. Smith (Fig. 4, no. 3) Odontomachus animosus Fr. Smith, 1860, J. Proc. Linn. Soc. London, Zool., 4 (suppl.): 102-103, worker. Type locality: Dory (Manokwari), Neth. New Guinea. Donisthorpe, 1932, Ann. Mag. Nat. Hist., (10)10:461. (Holotype examined — Oxford University Museum.) 490 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY Through the courtesy of Mr. Ernest Taylor of Oxford Uni- versity I have recently had the opportunity to study the unique type of this most enigmatic of Odontomachus species. 0. ani- mosus proves to be a very distinct species belonging to the sacvissimus group. In petiole shape and general habitus it is convergent (or annectant) to the species of the simillimus group, in particular simillinius itself, and may in fact represent a trvie phylogenetic link between the two groups. Unfortunately, I have been unable to find any more material of this species in recent collections. Below is a brief reclescription of the holotype. Holotijpc worker. HW 2.08 mm, HL 3.08 mm, SL "^3.09 mm, PW 1.19 mm, distance from the basal line of the petiolar node to the tip of the petiolar spine 1.34 mm, distance from the pos- terior margin of the petiolar spiracle to the tip of the spine 1.21 mm. Shape of petiole similar to that of 0. papuanus Emery, differing in that the anterodorsal face of the node seen from the side forms an almost perfectly straight line from the spine itself to the anterior collar, thus lacking the rounded angle that sepa- rates the anterior and dorsal faces in other members of the sacvissimus group {animosus shares this character with some species of the simillimus and iyrannicus groups; see Fig. 4). Entire dorsal surface of head striate, the striae becoming feeble in the occipital region. Ocular ridge striate posterior to the eye, smooth anterior to it. Sides of head ventral to the inner border of the eye, including the ventral half of the extraocular furrow, smooth and shining. Pronotum, mesonotum, and propo- deum covered by dense transverse striae. Anepisternum verti- cally striate; katei)isternum almost completely smooth and shining. Anterior face of petiolar node very faintly and trans- versely striate ; remainder of node smooth and shining. Gaster entirely smooth and shining. Coloration uniformly yellowish brown (but the specimen is undoubtedly faded; Frederick Smith described it in 1860 as "ferruginous"). Odontomachus cepiialotes Fr. Smith (Fig. 4, no. 11) OdontonKU-hus ccphalotes Fr. Smith, 1863, J. Proc. Linn. Soc. London, Zool., 7:19, worker. Type locality: Ceram. Crawley, 1922, Ann. Mag. Nat. Hist., (9)9:-l-41, fig. 6; redescription of holotype. WILSON : ANTS OF MELANESIA. V. 491 Odontomachus ruficeps subsp. cephalotes, Emery, 1911, Nova Guinea, 9(2) Zool.: 250-251, diagnosis, variation, distribution. Odontomachus ruficeps cephalotes var. omenta Einerj-, 1911, ihid., p. 251, worker, queen. Original localities: Merauke and Etna Bay, Neth. New Guinea. NEW SYNONYMY. Odontomachus ruficeps ceplialotes var. fusca Emery, 1911, ibid., p. 251, worker. Type locality: Merauke, Neth. New Guinea. NEW SYNONYMY. Odontomachus ruficeps subsp. cephalotes, Forel, 1911, Sitzber. Bayer. Akad. Wiss., (1911), p. 252, worker, variation. Odontomachus ruficeps cephalotes var. ternatensis Forel, 1911, ibid., p. 252, worker. Type locality: Ternate. NEW SYNONYMY. Odontomachus rtificeps cephalotes var. tamensis Stitz, 1912, Sitzber. Ges. Nat. Freunde Berlin, 9:503, fig. 7, worker. Type locality: Tami Islands, N-E. New Guinea. NEW SYNONYMY. Odontomachus ruficeps subsp. aruanus Karawajew, 1925, Konowia, 4:295, fig. 14, worker. Type locality: Wammar Island, Aru Archipelago. NEW SYNONYMY. Odontomachus ruficeps cephalotes var. longitudinalis Donisthorpe, 1940, Entomologist, 73:108-109, fig. 1, worker. Type locality: Camp Nok, Waigeo, 800 m. NEW SYNONYMY. (Syntype examined — MCZ.) Material examined. WAIGEO: Camp Nok: (longitudinalis Donisthorpe syntype). NETH. NEW GUINEA: Merauke (MCZ). PAPUA: Karema, Bro^vn R. (Wilson, nos. 545, 553, 579); Bisianmmi, 500 m. (Wilson, no. 615). N-E. NEW GUINEA: Sepalakambang, 1920 m., Saruwaged Ra. (E. J. Ford). AUSTRALIA: several series from various localities in North Queensland (MCZ). Taxonomic note. This species shows considerable variation in several external characters. The sculpturing of the first gastric tergite usually consists of whorled, coarse striae, but in a minor- ity of series from Queensland the striae are replaced by heavy shagreening. The pattern of pronotal sculpturing is also very variable. The petiolar node varies in shape from a form re- sembling that of 0. papuanus to one resembling that of 0. simil- limus. The body color is typically blackish brown but is occasionally replaced locally by a lighter reddish brown. Ecological notes. At Karema a colony was found nesting in a small rotting log on the floor of primary lowland rain forest. Workers from other colonies were found at the same locality foraging in leaf litter during the day. 492 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY Odontomachus emeryi Mann, n. status (Fig. 4, no. 5) Odontomachvs imperator subsp. emeryi Mann, 1919, Bull. Mus. Conip. ZnoL, 63:303, fig. 12, worker, queen, male. Type locality: Maliali, riorida, Solomon Islands. (Syntypes examined — MCZ.) Material examined. SOLOMON ISLANDS: Maliali, Florida (syntypes) ; Fulakora, Santa Isabel (syntypes) ; Torokina R., Bongainville (B. D. Valentine) ; Kokure, 690 m., Bougainville (E. J. Ford); Boku, Bougainville (Ford). Taronomic notes. This species is very similar to tlie wide- spread 0. saevissimns of western Melanesia and in fact may be no more than a geographic variant of it. Emeryi differs in its distinctive coloration and convex posterior border of petiolar node. A single worker of saevissimns from New Ireland ap- pears to be both geographically and morphologically interme- diate. It has a petiolar node like that of emeryi, but the body coloration is typical of saevissimns. Unfortunately, the head of this interesting specimen is missing. Ecological notes. Mann made the following observations on the type colonies : ' ' They were in dense forest ; the nests were in the ground among the roots of trees and contained large numbers of workers. The workers are less active than haemotoda [= simil- limns] and not as aggressive." E. J. Ford, Jr., collected winged queens from a nest at Kokure on June 12, 1956. Odontomachus gressitti Wilson, n. sp. (Fig. 4, no. 8) Diagnosis. A small, slender species belonging to the saevissi- mns group and most closely resembling papuanns Emery. It differs from pap nanus by its distinctive coloration, presence of transverse striae on the anterodorsal face of the petiolar node, and more slender petiolar spine. It bears a superficial re- semblance to liyiae Donisthorpe but differs markedly from that species in its smaller size, distinctive coloration, and " papuanus- type" petiolar node. Holotype worker. HW 2.16 mm, HL 3.48 mm, SL 3.43 mm, PW 1.35 ram, length of petiolar node 1.00 mm, distance from l^osterior margin of petiolar spiracle to tip of petiolar spine 1.42 mm. WILSON : ANTS OF MELANESIA. V. 493 Cephalic striae entirely limited to frontal lobes and interoeular depression ; remainder of head entirely smooth and shining. En- tire alitrunk transversely striate, the striae becoming very weak in the center of the pronotnm and even failing entirely in a lim- ited area just 1.32 mm posterior to the anterior margin of tlie pronotal "neck." Entire anterodorsal and lateral faces of petiolar node, exclusive of the spine and most of its supporting cone, transversely striate. Gaster completely smooth and shining. Head and gaster dark reddish brown. Pronotal "neck," pos- terior margin of pronotum, entire mesonotum, and propodeal dorsum posterior to the level of the propodeal spiracles medium reddish brown. All of these areas contrast with the remainder of the alitrunk and the petiole, which are a much lighter shade of brownish yellow. Type locality. Nondugl, 1750 m., Ahl Valley, N-E. New Guinea (J. L. Gressitt). The single worker from this locality has been returned to Dr. Gressitt for deposit in the B. P. Bishop Museum, Honolulu. Paratype ivorher. A single worker from Gold Ridge-to-Suta (Jonapau), 1100 m., Guadalcanal (Gressitt) has been deter- mined as this species. It differs from the holotype in its overall much lighter coloration (body light brownish yellow, the pro- notal neck and mesonotum a shade darker than the rest), slightly thicker petiolar spine, and presence of numerous oblique hairs on the spine and cone (standing hairs completely lacking in holotype). Further collecting may show the Solomons form to rank as a distinct species. The single Solomons specimen has been deposited in the Museum of Comparative Zoology. Odontomachus imperator Emery (Fig. lA) Odontomachus imperator Emery, 1887, Ann. Mus. Civ. Stor. Nat. Genovn. (2)5:429, pi. 1, fig. 1, worker. Type locality: Andai, near Manokwari, Neth. New Guinea. (Lectotype examined — Emery Coll.) Through the courtesy of Dott. Delfa Guiglia, I have been able to examine a syntype worker, which, with her permission, is here- in designated as lectotype. Lectotype worker. HW 2.59 mm, HL 4.20 mm, SL 4.20 mm, PAY 1.55 mm, petiolar node length 1.13 mm, distance from the 494 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY level of the basal line of the petiolar node to the tip of the petiolar spine 1.55 mm. Striae of the head limited to the trans- verse depression just anterior to the ocular rid^e, and not ex- tending onto the ridge itself ; remainder of head smooth and shining. Anterior "neck" of pronotum transversely striate, re- mainder of pronotum smooth and shining. Entire mesonotal and propodeal surfaces transversely striate, the mesonotal striae much more feeble than those on the propodeum, becoming obsolescent anteriorly and medially. Anepisternum striate ; katepisternum striate only along its dorsal margin, the remainder of its surface smooth and shining. Petiolar node and gaster entirely smooth and shining. Body uniformly dark reddish brown, appendages medium reddish brown. Odoxtomachus latissimus Viehmeyer Odontomacltus latissimus Viehmeyer, 1913, Arch. Naturgos., 79A(1L'):31, fig. 5, worker. Type locality: Sattelberg, N.-E. New Guinea. Known from type material onlj'. Odontomachus linae Donisthorpe, n. status (Fig. 4, no. 6) Odontomachus saevissimus var. linae Donisthorpe, 1940, Entomologist, 73:107, worker, queen. Type locality: Mt. Lina, 1200 m., Cyclops Mts., Neth. New Guinea. (Syntype e.xamined — MCZ.) Material examined. NETH. NEW GUINEA : Mt. Lina (syn- type). N-E. NEW GUINEA: Kumur, 1000 m., upper Jimmi Valley (J. L. Gressitt). Taxonomic notes. The only observable difference between this species and opaculus Viehmeyer is in the color characters cited in couplet 16 of the key. But the two forms have widely overlap- ping ranges and occur at similar elevations. The wisest course for the present would seem to be to treat them tentatively as distinct biological species. The Kumur specimen differs from tlie syntype cited in its weaker body sculpturing. Striae are very feeble in the center of the pronotum and completely lacking from the extraocular fur- rows and mesopleural centers. WILSON : ANTS OP MELANESIA. V. 495 Oix)NTo:MACiirs imalignus ¥v. Smith (Figs. 3; 4, no. 9) Odontomachus maUgniis Fr. Smith, 1859, J. Proc. Linn. Soi'. Loudon, Zool., liilAi, worker. Type locality: Aru. Emery, 1887, Ann. Mas. (^v. Stor. Nat. Genova, 4:429; distribution. Kutter, 1934, Mitt. Schweiz. Ent. Gcs., 15:472. Donisthorpe, 1940, Entomologist, 73:107, redescri]iti(m of holotype. Odontomachus tubercidatuft Soger, 18G1, Berl. Eut. Zeitsehr., 5:28-30, worker. Type locality: unknown. NEW SYNONYIMY (provisional). Donisthorpe, 1940, Entomologist, 73:107, worker. Odontomachus malig'nus subsp. tuherculatus, Mann, 1919, Bull. Mus. Conip. Zool., 63:305-306, fig. 13, worker. Material examined. SANTA CRUZ: Graciosa Bay (W. M. Mann). Emery (1887) records this species from Sarawak, Ce- lebes, and New Guinea (Sorong), while Kutter (1934) records it from Jacquinot Bay, on the southern coast of New Britain. Taxonomic note. Roger's tuherculatus is probably conspecific with maJignuii. The only difference that can be gleaned from the original description is in the orientation of the mesonotal stria- tion, which is said to be horizontal in malignus and longitudinal in tuherculatus (Mann, 1919). However, the distinction is doubt- ful. In Mann's single nest series of '"tuherculatus" from Santa Cruz the orientation of mesonotal striae actually varies widelj^, from longitudinal to oblique. Ecological notes. This species, which ranges from Borneo to the Santa Cruz Islands, appears to be a normal inhabitant of the littoral zone. Mann says of his Solomons collections, "I found this species only once, at Graciosa Bay, where workers were mov- ing in and out of the crevices of a large block of coral on the beach. Mr. Harry Hall, who brought me specimens from Simoli on South Malaita, states that he found it nesting there under the same conditions." According to Kutter (1934), H. Hediger found workers of malignus at Jacquinot Bay, New Britain, foraging 100 meters out in the intertidal zone during low tide ! These individuals were running among the coral rocks and far from the normal foraging ranges of other ant species. Odontomachus montanus Stitz, n. status (Fig. 4, no. 7) Odontomachus imperator var. montanus Stitz, 1923, Sitzber. Gos. Nat. Freunde Berlin, p. 116, worker. Type locality: Lordbcrg, iiiiddh' Scpik 496 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY region, N-E. New Guinea. (Syntype examined — MCZ.) Known from type material only. Odontomaciius nigriceps Fr. Smith Odontomachus nigriceps Fr. Smith, 1860, J. Proe. Linn. Soc. London, Zool., 4 (suppl.) : 103, pi. 1, fig. 9 (this figure is also referred to by Smith as 0. saevissimus) . Type locality: Dory (Manokwari), Neth. New Guinea. Mayr, 1876, J. Mus. Godeff. Hamburg, 12:85, distrilnition. Donisthorpe, 1932, Ann. Mag. Nat. Hist., (10)10:461, worker. Odontomachus angulatus subsp. praefectus Forel, 1911, Sitzber. Akad. Wiss. Munich, p. 251, worker. Type locality: New Guinea. (Holotype ex- amined— Zoologische Staatssammlung, Munich.) NEW SYNONYMY. Material examined. PAPUA: Kokoda, 350 m. (L. E. Chees- man) ; Dobodura (P. J. Darlington). Taxonomic notes. This species is closely related to testacea Emery, from which it can be distinguished by its distinctive coloration and much denser body pilosity. The two species oc- cur together at Dobodura. The holotype of 0. angulatus subsp. praefectus Forel is a typi- cal worker of nigriceps. It has relatively convex anterior and posterior nodal borders, resembling the Kokoda specimen rather more closely than workers from Dobodura. There is no evident reason for Forel's decision to associate it with angulatus Mayr, a widely dissimilar member of the saevissimus group. Odontomachus opaculus Viehmeyer, n. status Odontomachus imperator subsp. opaculus Viehmeyer, 1912, Abh. Zool. Anthrop. Mus. Dresden, 14:6, fig. 6, worker. T^^)e locality: TorricuUi Mts., 640 m., N.-E. New Guinea. (Sjoitype examined — MCZ.) Material examined. NETH. NEW GUINEA : Wendesi-Majosi (S. Issiki). N.-E. NEW GUINEA: Torricelli Mts. (syntype). Taxonomic note. A single worker examined from Wendesi- Majosi differs from the opaculus syntype in its larger size (HW 2.77 mm vs. 2.32 mm), different orientation of pronotal striae (concentric vs. straight-transverse), and feebler striation within the extraocular furrow. Odontomachus papuanus Emery, n. status (Fig. IB) Odontomachus rixosus var. papuanus Emery, 1887, Ann. Mus. Civ. Stor. Nat. Genova, (2)5:429, worker. Type locality: Aru. (Syntype ex- amined — Emery Coll.) . WILSON: ANTS OF MELANESIA. V. 497 Odontomachus papuanus var. concentricus Emery, 1897, op. clt., 38:5r)7, worker. TjT)e locality: Moroka, Papua. NEW SYNONYMY (pro- visional). Material rxamincd. NETII. NEW GUINEA: Maffin Bay (E. S. Ross). N-E. NEW GUINEA: lower Busii Kiver (Wilson, nos. 901, 923) ; Biilolo, 730 m. (E. J. Ford) ; Sattelberg-Maroriio, 800-900 m. (Wilson, no. 724) ; Maroruo, 900 m. (Wilson, no. 729) ; Ng-anduo, 1000 m. (Wilson, no. 733) ; Ebabaang, 1300- 1400 m. (Wilson, no. 828) ; Wamnki, 800 m. (Wilson, no. 850). ARU: syntype worker. Taxoiwmic notes. 0. papuanus is closely related to the Orien- tal species rixosus Fr. Smith, differing chiefly in its longer petiolar spine and more rounded apical mandibular teeth. Its recognition here as a distinct species is a provisional measure only. Significant geographic variation occurs within the range of papuanus on New Guinea. Workers from the lowlands (Maffin Bay, low-er Busu River) are smaller and lighter in color than those from the mountains of the Huon Peninsula. The syntype from Aru is light in color but as large as the montane New Guinea workers. Ecological notes. This species has been collected in both pri- mary lowland rain forest and primary and secondary mid- mountain rain forest on New Guinea. Workers were found foraging singly on the ground during both the day and night. At the Busu River, a nest was found on a steep forested hillside. It consisted of a single shaft, five centimeters wide, extending horizontally into the soil beneath a tree root for a distance of approximately 45 centimeters. The colony was a small one, con- taining a single (pieen and about twenty workers, and may have been incipient. Odontomachus rupithoeax Emery, n. status Odontomachus imperaior subsp. rufithorax Emei-y, 1910, Nova Guiaea, 5(4)Zool.: 534, worker. Original localities: "Tawarin" and "Ti- mena," Neth. New Guinea. Known from type material only. Odontomachus saevissimus Fr. Smith (Fig. 4, no. 4) Odontomachus saevissimus Fr. Smith, 1858, Cat. Ilym. Brit. Mus., 6:80, pi. 5, fig. 10, worker. Type locality: Ceram. Fr. Smith, 1860, J. Proc. 498 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Linn. Soc. London, Zool., 5(suppl.): pl. 1, fig. 9, queen. Mayr, 1867, Tijdschr. Ent., 10:47, worker, distribution. Karawajew, 1925, Kono- wia, 4:290-292, fig. 12, worker, queen, male. (Holotvpe examined — BMNH.) Odonto7nachus vmperator var. tauerni Stitz, 1923, Sitzber. Ges. Nat. Freunde Berlin, p. 116, worker. Type locality: Ceram. (Syutype ex- amined—MCZ.) NEW SYNONYMY. Odontomachus transverso-striatus Donisthorpe, 1941, Trans. Roy. Ent. Soe. London, 91:53, worker. Type locality: Mt. Baduri, 300 m., Japen I., Neth. New Guinea. (Holotype examined — BMNH.) NEW SYN- ONYMY. Material examined. MOLUCCAS: Ceram (holotype; sj^iitype of tauerni). NETH. NEW GUINEA: Japen I. (transversostria- tus holotype) ; Doormanpad (W. C. van Heurn). N-E. NEW GUINEA: Korop, 1300 m., Upper Jimmi Valley (J. L. Gres- sitt). NEW BRITAIN: St. Paul's, 350 m., Bainings Mts. (Gressitt). NEW IRELAND: "Camp Bishop," 12 km. up Kait River, 240 m. (E. J. Ford) (tentative determination; see below). Taxonomic notes. This is one of the most widespread of the Papuan-based Odontomachus. Mayr (1867) records it from Celebes, while as noted elsewhere the form emeryi Mann may be nothing more than a geographic variant from the Solomon Islands. Notable geographic variation is shown in the sculptur- ing of the mesepisternum. Workers from New Guinea have katepisterna completely striate and the anepisterna smooth ex- cept for feeble striation along the dorsal and posterior mai'gins. A single worker examined from the offshore island of Japen has essentially simihir sculpturing, except that on the katepisternum striation is confined to the posterior margin. Workers from more peripheral localities both to the east and west (Ceram, New Britain, New Ireland) have completely striate mesepisterna. Thus geographic variation in this character appears to exhibit a concentric "central -peripheral" pattern. A single headless worker from New Ireland differs from otJier material in having heavier alitruncal sculpturing and a convex posterior border of petiolar node, in these characters approaching the Solomons form emeryi. Future collecting may show that saevissimus and emeryi are connected by other morphologically intermediate populations and hence must be considered conspecific. If so, then alitruncal sculpturing and coloration clearly show discordant patterns of geographic variation. WILSON : ANTS OF MELANESIA. V. 499 Odontomachus simillimus Fr. Smith (Fig. 4, no. 10) Odontcymacliiis simillimus Fr. Smith, 1858, Cat. Hyin. Brit. Mus., (i:80, pi. 5, figs. 8, 9, queen. Original localities: Fiji Islands, Ceylon. Odontomachus liacmatodus, div. auct. {nee Formica haematoda Linne, 1758, Syst. Nat., 10th ed., 1:582). Odontomachus hacmatodus var. fuscipennis Forel, 1913, Zool. Jahrb. Syst., 36:19, worker, queen, male. Original localities: Peradeniya, Ceylon; Bahsoemboe, Sumatra. NEW SYNONYMY. Material examined. MOLUCCAS: Amboiua (H. Smith). NETII. NEW GUINEA: Doormanpad (W. C. van Heurn) ; Biak I. (G. E. Bohart). N-E. NEW GUINEA: Goroka, 1600 m., Asaro Valley, Central Highlands (J. J. H. Szent-Ivany) ; Mt. Misim (H. Stevens) ; Wareo; Nadzab (Wilson, no. 1100) ; Bubia (N. L. H. Krauss) ; lower Busu River (Wilson, no. 944) ; Finschhafen (Wilson) ; Bolingbangeng, 900-1000 m. (Wilson, no. 728) ; Ziiigzingu, 1200 m. (Wilson, no. 763). PAPUA: China Strait (W. J. Eyerdam) ; Dobodura (P. J. Darlington) ; Laloki R., near Port Moresby (Wilson, no. 528). NEW BRIT- AIN: St. Paul's, Bainings Mts., Gazelle Pen. (J. L. Gressitt). NEW IRELAND: "Camp Bishop," 12 km. up Kait R., 240 m. (J. L. Gressitt). SOLOMON ISLANDS: Kungana Bay, Rennell (M. Willows, Jr.) ; Bellona I. (Willows). SANTA CRUZ : Vani- koro (AVillows). NEW HEBRIDES: Vila, Efate (N. L. H. Krauss) ; Aore I. (W. L. Nutting) ; Ratard Plantation, near Luganville, Espiritu Santo (Wilson). FIJI ISLANDS: nu- merous series from throughout the islands, from Viti Levu to the Lau Archipelago, collected chiefly by W. M. Mann. This species is also widespread through Micronesia and Polynesia. Taxonomic note. AV. L. Brown (pers. commun.), who is cur- rently studying the New World species of 0 clout omachus, in- forms me that the true 0. haematodus is probably a species indigenous to the Amazon-Orinoco Basins and not conspecific with the Pacific similUmus. According to Linne 's original de- scription, haematodus possesses the following color characters : "Abdomen nigricans . . . Pedes flavi . . . Corpus nigrum." The Melanesian species identified here as simillimus (the next oldest name applicable to Indo-Australian populations) has me- dium brown legs and dark brown head and alitrunk. 500 BULLETIN : MUSEUM OP COMPARATIVE ZOOLOGY Ecological notes. In eastern New Guinea simillhnus is common everywhere in clearings and second-growth forest. Colonies apparently nest in the soil, and workers can be found foraging in leaf litter during both day and night. At the Laloki River in Papua a colony was found nesting in accumulated soil and vegetable debris in the primary fork of a tree a little less than two meters from tlie ground. Near Luganville, in the New Hebrides, a large colony, containing alate queens and males, was found in early January beneath a rotting log on the floor of Ioav- land rain forest. Odontomachus testaceus Emery, n. status (Fig. 4, no. 2) Odontomachus tyrannicvs var. testacea Eiiierj-, 1897, Ann. Mus. Civ. Stor. Nat. Genova, 38:.")57, worker. Type loeaJity: "Haveri, " New Guinea. (Syntype examined — Emery Coll.) Odontomachus gulosus Emery, 1902, Termeszetr. Fiiz., 25:160, worker. Type locality: Sattelberg, N-E. New Guinea. NEW SYNONYMY. (Syntype examined — Emery Coll.) Odontomachus gulosus var. nubila Emery, 1911, Nova Guinea, 9(2)zool.: 250, worker. Type locality: Etna Bay, Nctli. New Guinea. NEW SYNONYMY (provisional). Odontomachus nigrifrons Donisthorpe, 1940, Entomologist, 73:106, worker. Type locality: Hollandia, Neth. New Guinea. NEW SYNONYMY. (Syntype examined — BMNH.) Material examined. NETH. NEW GUINEA: Hollandia {nigrifrons Donisthorj^e s^yntype). N-E. NEW GUINEA: lower Busu River (Wilson, nos. 904, 944) ; Lae (N. L. II. Krauss) ; Boana, 1100 m. (Wilson, no. 1115) ; Finschhafen (N. G. L. Wagner) ; Lambaeb, 900 m., Saruwaged Ra. (E. J. Ford) ; Foria River to Zingziugu, ca. 1000 m. (Wilson, no. 757) ; Zingzingu, 1200 m. (Wilson, no. 761) ; Gemeheng, 1300 m. (AVilson) ; Tum- nang, 1450-1600 m., a single male (Wilson). PAPUA: Dobodura (P. J. Darlington) ; Karema, Brown R. (Wilson, no. 595) ; Bisianumu, 500 m. (Wilson, nos. 607, 623; J. L. Gressitt). Taxonomic note. Workers from the Iluon Peninsula have a paler ground color than those from Papua, and their cephalic and gastric patches are reduced to faint infuscations. They in- clude the type specimens of Emery's synonymous form gulosus. WILSON : ANTS OP MELANESIA. V. 501 Ecological notes. Many nests of this species were found by the author during liis field studies in New Guinea. In most cases they had been excavated in the soil and were marked externally by a single wide, vertical entrance shaft surrounded by a ring of coarse pellets of excavated earth. Often they were located be- tween the buttresses of forest trees. At Karema a single colony was in the rotting center of a branch of a fallen tree, in a termi- nal portion raised off: the ground. At Bisianumu a dealate queen was found isolated with five eggs in a cell under the bark of a rotting log. Mature colonies contain several hundred workers. Alate queens and males were taken in a nest at Karema on March 10, 195"), and a lone male was collected at light at Tum- nang on April 14, 1955. The workers are unusually aggressive and capable of delivering a shocking sting. These traits, com- bined with the large size of the w^orkers, make them among the most formidable ants to be found anywhere in the world. Odontomacuus tyrannicus Fr. Smith (Fig. 4, no. 1) Odontomachiis tyrannicus Fr. Smith, 1859, J. Proc. Linn. 8oc. London, Zool., 3:455, worker. Type locality: Aru. Viehmeyer, 1912, Abli. Zool. Anthrop. Mus. Dresden, 14:6. Nee Odontomaclius tyrannicus Fr. Smith, 1861, J. Proc. Linn. Soc. London, Zool., 6:44, pi. 1, fig. 4 (see 0. gladiator Douisthorpe, 1932, Ann. Mag. Nat. Hist., (10)10:467). Odontomachus tyrannicus var. ohsolescens Donisthorpe, 1940, Entomologist, 73:106-107, worker, male. Original localities: Kokoda, Papua; Camp Nok, Waigeo. NEW SYNONYMY. (Syntypcs examined — MCZ.) Material examined. WAIGEO : Camp Nok (syntypes). NETII. NEW GUINEA: Mamberamo River (W. G. van Ileum). N-E. NEW GUINEA: Finschhafen (2 series; N. G. L. Wagner, E. S. Ross) ; Lambaeb, 900 m., Saruwaged Ra. (E. J. Ford) ; Sattel- berg, 660 m. (Wilson, no. 722) ; Sattelberg-Maroruo, 800-900 m. (Wilson, no. 724) ; Bolingbangeng-Nganduo, 900-1000 m. (Wil- son, no. 731). PAPUA: Mt. Lamington (C. T. McNamara) ; Bisianumu, 500 m. (Wilson, nos. 659, 667). NEW BRITAIN: St. Paul's, 350 m., Bainings Mts., Gazelle Pen. (J. L. Gressitt) ; Ti, Nakanai Mts. (Ford). Taxonomic note. The depth of pronotal striation varies greatly, as noted by Donisthorpe in his description of var. ohso- lescens, but this variation is apparently non-geographic. Both 502 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY extremes as well as intermediates are included in material from a single locality, Finschhafen. Ecological note. This species was relatively common in second - growth forest along the native trail between Sattelberg and Nganduo. Both in this area and at Bisianumu, Papua, workers were found foraging on the ground during the day. A note accompanying van Heurn's Mamberamo series (MCZ) states that this collection was made from a small colony nesting in the soil beneath a fallen palm trunk. The Genus ANOCHETUS Ma^^r List of the Melanesian and Moluccan Species, Arranged into Species Groups and Including New Synonymy Group of A. cato Forel cato Forel = Anochctas cato var. suhfasciatus Mann = Odontomachus rossi Donisthorpe isolatus Mann semmigcr Donisthorpe splendens Karawajew Group of A. chiricJtinii Emery chirichinii Emery fricatus Wilson Group of A. graeffei Mayr graeffei Mayr = Anochctus amati Karawajew ^ Anochctus niijiutus Karawajew = Anochctus punctiventris Mayr = Anochctus punctiveyitris subsp. oceanicus Emery Group of A. variegatus Donisthorpe variegatus Donisthorpe Incertae Sedis filiconiis (Wheeler) WILSON : ANTS OF MELANESIA. V. 503 Kcii It) the Anociieti's Species of Melanesia and the Mohicra.<;, based on the Worker Caste 1. Masticatory border of mandible with a ])rominent l)liint tooth located at midlenjjfth; dorsolateral i)ropodeal corners tuberculate ; dorsal niarg;in of petiolar node concave when node is viewed anteropos- tcriorly 2 ^[astieatory border of mandible lacking a prominent tooth at mid- longth; dorsolateral propodeal corners rounded or obtusely angnlate; dorsal margin of petiolar node convex to acute in anteroposterior view 3 2. Central portion of pronotum striate and subopaque; dorsolateral cor- ners of petiolar scale forming angles of 80° or more frieaius Wilson Central portion of pronotum comjjletely smooth and shining; dorso- lateral corners of petiolar scale drawn out into spine-like processes of which the apices form angles of 60° or less chiricliinii Emery 3. Central portion of pronotum coarsely rugose and subopaque ; propodeum angulate when viewed from the side; petiolar scale broad and mod- erately convex in anteroposterior view ; anterior half of first gastric tergite often punctate graeffei Mayr Central portion of pronotum smooth and shining ; propodeum rounded in side view; petiolar scale narrowed dorsally, its crest strongly convex to acute; anterior half of first gastric tergite always com- pletely smooth and shining 4 4. Intercalary tooth of apical mandibular fork located on the inner border of the ventral tooth about two-thirds the distance from the angle of the fork to the tip of the ventral tooth (position of the median tooth is measured from the center of its base) ; petiolar scale taper- ing dorsally into a spine variegatus Donisthorpe Intercalary tooth of apical mandibular fork located on the inner border of the ventral tooth about half way between the angle of the fork and the tip of the ventral tooth ; petiolar scale tapered somewhat dorsally but not forming a spine 5 5. Cephalic striae covering most of the dorso-central surface of the head as well as the frontal area cato Forel Cephalic striae limited to the area between the frontal carinae (isolatus superspecies) 6 G. Head and alitrunk black, gaster and appendages yellowish brown (Waigeo) seminigcr Donisthorpe Head and alitrunk at most dark reddish brown, gaster and appendages dark yellowish brown to reddish brown 7 7. Head and alitrunk dark reddish brown, petiole and gaster dark yel- lowish brown (eastern Solomons and Santa Cruz) . isolatus Mann 504 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Head and alitrunk light yellowish brown, petiole and gaster light red- dish brown (Am) splendens Karawa.iew Anochettts cato Forel (Fig. 2) Anodhetus Cato Forel, 1901, Mitt. Zool. Mus. Berlin, 2(1, b) ; 6, worker. Type locality: Lowon Valley, near Ealum, New Britain. Anochetus cato var. suhfasciatus Mann, 1919, Bull. Mus. Comp. Zool., 63:301, worker, queen, male. Type locality: Malapaina I., Three Sisters Group, Solomons. (Syntypes examined — MCZ.) NEW SYNONYMY. Odontomachus rossi Donisthorpe, 1947, Ann. Mag. Nat. Hist., (11)14:186- 187, worker, queen. Type locality: Maffin Bay, Neth. New Guinea. Nee Anochetus rossi Donisthorpe, 1949, op. cit., (12)1:747. (Syntypes examined — CAS, MCZ.) NEW SYNONYMY. Anochetus cato group Fiji Urands Samoo o- Fig. 2. Maximum known distributions of species of the Anochetus cuin group, a, cato, New Guinea to Solomon Islands; h, isolatus, eastern Solo- mons and Snnta Cruz ; 6', splendidultis, Palau ; h", semi/niger, Waigeo ; // ", splendens, Aru. Material examined. NETH. NEW GUINEA : xVlaffin Bay (syn- types of Odontomachus rossi and one additional nest series). WILSON : ANTS Oi<^ MELANESIA. V. 505 N-E. NEW GUINEA: Boliiigbangeng-Ngaiuluo, 000-1000 m., alate queen (Wilson, no. 731). PAPUA: Bisiannnni, 500 m. (Wilson, nos. (m, 660, 667). NEW BRITAIN: Koravat, 60 m., Gazelle Pen. (J. L. Gressitt). SOLOMONS: Reiulova; Mala- paina; Fnlakora, Santa Isabel; Pawa, I^gi; Auki, Malaita; Wai-ai, San Cristoval (all W. M. Mann). Taxonomic notes. The available material of this species shows noteworthy geographic variation in color, which can be outlined as follows. Nciv Britain and Rendova: body and appendages concolorous reddish 3'ellow. Ugi: body medium reddish brown, legs yellowish brown. San Cristoval: body dark reddish brown, legs yellowish brown. Malapaina and Santa Isabel: body very dark reddish brown, nearly black, the legs medium brown. Malaita: body and appendages intermediate in shade between the San Cristoval and Malapaiua-Santa Isabel series. Bisianumu, Papua: body very dark reddish brown, almost black, legs yel- lowish to medium brown. Bolinghangeng, N-E. New Guinea: head, alitrunk, and petiole very dark reddish brown, almost black, gaster somewhat lighter in shade, appendages medium brown. Maffin Bag, Neth. New Guinea: head very dark reddish brown, almost black, alitrunk medium to moderately dark red- dish brown ; petiole and gaster contrasting dark yellowish brown ; legs light reddish brown. Of particular interest is the possibility revealed in the above data of the existence of graded inter-island variation in the east- ern Solomons. As more material becomes available from over its entire range, this species should prove an especially fruitful subject for a thorough analysis of geographic variation. Ecological notes. At Bisianumu a small colony of this species was found nesting in a large, " passalid-stage " log on the floor of second-growth foothills rain forest. Workers from other nests were found foraging on the forest floor during the daytime. On Rendova, Mann (1919) also found a colony nesting in a rotting log, presumably in lowland rain forest. Winged queens were collected in a nest on September 9, 1944, at Maffin Bay by Dr. E. S. Ross. Anochetus cniRicniNii Emery Anochetus Chirichinii Emery, 1897, Termeszetr. Fiiz., 20:597, pi. 15, figs. 46, 47, worker. Type locality: Hansemann Mts., N-E. New Guinea. (Syntype examined — Emery Coll.) 506 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY Material examined. N-E. NEW GUINEA: Hansemann Mts., near Madang (syntype) ; Nadzab (Wilson, nos. 1088, 1101, 1107) ; lower Busu River (Wilson, nos. 999, 1051, 1112). Ecological notes. The collections so far recorded indicate that A. chirichinii is relatively abnndant in both rain forest and monsoon forest. At the Busu River and Nadzab, workers Avere encountered on several occasions foragiiifj during the day on the forest floor. A male tentatively determined as chirichinii was collected at light on May 16, 1955, at the Busu River. Anochetus fricatus Wilson, n. sp. Diagnosis. Closely related to .1. chirichinii Emery, differing primarily in the following characters : (1) In fricatus the entire pronotum is covered by longitudinal striae, and its surface is subopaque. In chirichinii striae are limited to the declivitous surfaces and posterior margin of the pronotum, and the entire central portion of the sclerite is smooth and shining. (2) In fricatus the striae of the central portion of the cephalic dorsum extend all the way back to the occipital border ; in chiri- chinii they reach only to within about 0.09 mm of it. (3) In fricatus the posterolateral corners of the petiolar node are much less attenuated than in chirichinii, forming a blunt angle of 80° or more (as opposed to 60° or less in chirichinii). (4) The body pilosity of fricatus is considerably denser over- all than in chirichinii. When the mesonotum is seen in exact side view the number of hairs projecting beyond its dorsal mar- gin is 15-19 in the fricatus types and 5-10 in the Museum of Comparative Zoology series of chirichinii. Holotype ivorker. HW 1.06 mm, HL 1.22 mm, SL 1.00 mm, CI 87, SI 94, PW 0.58 mm, petiole height 0.36 mm. Worker paratype variation. HW 1.02-1.12 mm, HL 1.17-1.30 mm, SL 0.98-l.io mm, CI 85-87, SI 94-98, PW 0.56-0.62 mm, petiole height 0.33-0.35 mm. Material examined. PAPUA: Karema, Brown River (type lo- cality), holotype and two paratype workers (Wilson, no. 541). N-E. NEW GUINEA : Bubia, a single paratype worker (Wilson, no. 683). Ecological note. The holotype nest series was collected from a large rotting log on the floor of primary lowland rain forest. WILSON: ANTS OP MELANESIA. V. 507 Anoc'Hetus graefkei Mayr Anochetus Graeffei Mayr, 1870, Vorli. zool.-Iiot. (!cs. Wicii, 2n:9()l, worker. Type localitj': Upolu, Samoa. Anochetus punctiventris Mayr, 1878, op. cit., 1^8: l.l-K), worker. Type lo- cality: Calcutta area, India. NEW SYNONYMY. (Syatypo exaniiiicil — Emery Coll.) Anochetus p^unctivtnlris subsp. occanicus Euiery, 1897, Tei'iueszetr. Fiiz., 20:597, worker. Original localities: Friedrieh-Wilhemshafen (Aitape), N-E. New Guinea. NEW SYNONYMY. (Holotype examined — Emery Coll.) Anochetus aniuti Karawa.jew, 1925, Konowia, 4:285, fig. 8, queen. TyjK' locality: Wammar I., Aru Archipelago. NEW SYNONYMY (pro- visional). Anochetus minutus Karawajew, 1925, ihid., pp. 288-289, fig. 10, worker, queen. Type locality: Segamat, Johore, Malaya. NEW SYNONYMY (provisional). Anochetus minutus, Yasuinatsu, 3940, Annot. Zoo). Jap., 19:31.3, fig. 2, worker, distribution. Material examined. N-E. NEW GUINEA: Aitape (occanicKs Emery holotype); Didiman Creek, Lae (Wilson, no. (i90). SOLOMON ISLANDS: Maravo Lagoon, New Georgia (W. M. Mann) ; Auki, IMalaita (Mann) ; Wainoni Bay and Pamua, San Cristoval (Mann). XEW HEBRIDES: Ratard Plantation. Lnganville, Espiritu Santo (Wilson, no. 332). FIJI ISLANDS: Lasema, Vanna Levn (Mann); Somo Sonio, Taviuni (Mann). NEW (CALEDONIA: Chapeau Gendarme (Wilson, no. 62); Mt. Mou (Wilson) ; "S.E. New Caledonia" (N. L. II. Krauss). This species is found through large parts of southeastern Asia, Melanesia, Polynesia, and Queensland. Very probably it has been distributed through part of this range through the inad- vertent agency of man. Taxonomic notes. During the course of the ])resent study the author has examined large amounts of Asian, Melanesian, and Australian material determined by earliei- specialists variously as graeffei and punctiventris, without being al)le to distinguish two species on the basis of Mayr's original diagnostic characters or any other characters. There appears to be but a single rela- tively variable species. In particular, sculj)turing of the first gastric tergite, traditionally stressed in earlier descriptions, varies gradually from smooth and shining (extreme "graeffei" form) to coarsely punctate and subopaque {"punctiventris" 508 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY form). The same conclusion has been reached with respect to the supposed distinction between puncHventris and its subspecies oceanicns. A few differences observed between types of the two forms in the Emery Collection, involving, especially, total size and petiolar node shape, are connected in other series by inter- gradient forms. A. amati Karawajew was described without reference to A. graeffei, but its characters seem to fit the latter species in detail. A. minutus Karawajew was compared with amati in the original description; it is supposedly distinguished from that species by its smaller size and proportionately shorter mandibles. There seems to be no reason to suppose that it is anything more than an infraspecific variant of graeffei. Ecological notes. Both the author's New Guinea and New Hebrides collections consist of stray workers taken during the day from the floor of lowland rain forests. At Chapeau Gen- darme, New Caledonia, a small colony was found nesting beneath a rock in a clearing. Anochetus isolatus Mann, n. status (Fig. 2) Anochetus cato subsp. isolatus Mann, 1919, Bull. Mus. Comp. Zool., 03:302, fig. 11, worker, male. Type locality: Graeiosa Bay, Santa Cruz. (Syn- types examined — MCZ.) Material examined. SANTA CRUZ: Graeiosa Bay (syntypes). Mann also recorded this species from Malapaina, Three Sisters Group, Solomons, on which island it occurs sympatrically with the closely related A. cato Forel. Taxonomic note. A. isolatus forms with A. splendens (Am), A. scminiger (Waigeo), and A. splendidnlus (Carolines), the "isolatus superspecies, " i.e., a tightly-knit group of cognate forms which seem sufficiently well differentiated to be good bio- logical species, but which are completely allopatric in distribu- tion. Actually, treatment of these four forms as species must be considered arbitrary until evidence is obtained of non-intergrada- tion in areas of overlap, if indeed such areas exist at all. The range of the isolatus superspecies forms a nearly complete circle around that of the related species cato. Brown (Quart. Rev. Biol., 32:271, 1957) has suggested that this unusual pattern WILSON: ANTS OF MELANESIA. V. 50!) may have resulted from the replacemoni of isolofits in New (lUiiiea, Bismarck Arehipelajro, aiul western Solomons by the more reeently evolved cato. ANOCHETrs SEMLViGER Donisthorpe (Fig. 2) Anochetus neminiger Donisthorpe, 1943, Ann. Mag'. Nat. Hist., (11)9:170, worker. Type locality: Camp Nok, 800 m., Waigeo. Known only from ty])e material. See discussion under A. Isold fus Mann. ANOCHETrs spLENDENs Karawajew Anochetiu^ splemJens Karawa.iew, 1925, Konowia, 4:289, fig. 11, (nieen. Type locality: Wammar I., Aru Archipelago. Known from type material only. See discussion under .1. isolat}(s ]\Iann. Anociieti-s variboatus I)onisthori)e, n. status Anochetufi cato var. variegatus Donisthorpe, 1938, Ann. Mag. Nat. Hist., (11)1:597, worker. Type locality: Mt. Nonio, south of Mt. Bougain- ville, Neth. New Guinea. (Syntype examined — MCZ.) Materml examined. NETII. NEW GUINEA: Mt. Nomo (syn- type). N-E. NEW GUINEA: Didiman Greek. Lae (Wilson, no. 711). Taxonottiic note. The Lae specimens differ from the MGZ syn- type in havino- cephalic striae limited to the area between the frontal carinae ; in the type, striae extend laterally beyond the carinae to a point midway between tlie carinae and the com- pound eyes, and posteriorly to within 0.20 nun of the anterior- most point of the occipital border. The Lae s])ecinuMis also have somewhat more acute petiolar spines. Ecological notes. At Lae, two workers were found during early evenino; foraging on the lower part of a tree trunk at the edge of rain forest. Species Inquirendae Anochetus filicornis (Wheeler) Mifnnapatetes filicornis Wheeler, 1929, Amer. Mus. Novitates, no. 349: ti, fig. 3, male. T^'pe locality: Larat I., Taninihar. (Holotype exainincil — MCZ.) .lnochetu.