ISTITUTION NOIlfUllSN! NVlNOSHlIINS^Sa I d Vd 8 II LIB RAR i ES^SMITHSONIAN INSTITUT CO > CO ___ = CO & ~ GQ Vi'' v O VW _ Xjvav^ o _ 2 ’ X' cJ 2 __ _ aidVdan libraries Smithsonian institution noiioxiisni nvinoshxiws saidVd r- 2 r* 2 ISTITUTION NOIlfUllSNI NVINOSHXIIAIS S3 I dVd 8 81 LIB RAR I ES^SMITHSONIAN^INSTITUT CO 2 ____ CO 2 v N CO 2 > 3 lavaan >IB RAR 1 es smithsonian_institution NoiiniusNi^NViNOsaLiws^sa 1 ava SSffiSx “ .. /SSgsgSx “ /sSSSx i .JK-x 2 ° ISTITUTION NO I XflXIXSN I NVINOSHXIIAIS S3 I dVd 8 II^LI B RAR I ES^SMITHSONIAN^INSTITUT 2 __ 2 2 ^ r= 2 r= ° _ .w O m /f^\ 2 m ^ m x^osv^x g (ms 3 I H Vd 8 11 LIBRARIES SMITHSONIAN~INST!TUTION NOIXrUllSNl“ NVINOSHXIIAIS S3!dVH CO H CO MuvmA «o O MV '#?;/ x > j> '' ^ X^os’^/r > ' (STITUTION^NOlifUUSNI^NVINOSHlIlAIS^SB I ava a 11 LI B RAR I ES^SMITHSONIAnInSTITUT oi =5 > o> 5 « — w w U i 2 CO c H HI c o 5 “ ,c 3 I dVd 8 II LI B RAR I ES SMITHSONIAN INSTITUTION NOIinXIJLSNl””NVINO$HXIW$ ~~S3 S d Vd 2 r- _ 2 m pi zz, m CO ^ CO ± CO NSTITUT10N NOIXfUIXSNI NVINOSHXIIAIS S31d¥d8n LIBRARIES SMITHSONIAN INSTITUT CO 2 _ CO 2 5 '«r £ > S '^Siss-' > , 2 \RIE$ SMITHSONIAN INSTITUTION NOlifliliSNI NVINOSHilWS SSSHV^ail^ _ 5 ^ ^ . w 2 LIBRARIES S A? $-W3 .« ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. {nat. Hist.) VOLUME 8 • PARTS 1—14 December 1969 — December 1970 PUBLISHED JOINTLY BY THE APE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Cape Provincial Museums South Africa Albany Museum, Grahamstown Alexander McGregor Memorial Museum Kimberley East London Museum East London Kaffrarian Museum King William’s Town Port Elizabeth Museum, Snake Park and Oceanarium Port Elizabeth Editor C. F. Jacot-Guillarmod Assisted by R. M. Tietz Albany Museum, Grahamstown Printed by Cape & Transvaal Printers Ltd., Cape Town LIST OF CONTENTS New names proposed in this volume: Species iii Part 1. The Nothobranchius (Pisces, Cyprinodontidae) of Southern Africa and a new species from lake Chilwa, Malawi. Part 1 — R. A. Jubb 1-11 2. Revision of the Chiloglanis (Pisces: Mochokidae) of Southern Africa and des- criptions of two new species. Part 1 : Limpopo, Incomati and Pongola Rivers — R. A. Jubb and P. le Roux 13- 23 3. Variational Patterns and Races of the Clapper Lark Mirafra apiata (Vieillot) — C. D. Quickelberge 25- 37 4. Taxonomic notes on South African marine mollusca, 1 — R. N. Kilburn . . 39- 48 5. Four new forest Millipedes from Lesotho and the Eastern Cape — R. F. Law- rence . 49- 55 6. Aspects of adaptive radiation in Southern Africa Accipiters — R. A. R. Black and G. J. B. Ross . 57- 65 7. Tilapia Mossambica Peters, from Australia — R. A. Jubb and F. O. Petrick . 67- 71 8. The juvenile stadia of Anampses caeruleopuntcatus Ruppell (1829) — G. J. B. Ross 73- 78 9. The Southern limits of distribution of commercially important penaeid prawns in South Africa — D. A. Hughes 79- 83 10. A new freshwater gobi (Pisces: Gobiidae) from Lake Sibayi, Zululand, South Africa — F. L. Farquharson 85- 87 11. The Acheulian occupation at Amanzi Springs, Uitenhage district, Cape Pro- vince— H. J. Deacon 89-189 12. Plant remains from Amanzi Springs — M. J. Wells 191-194 13. The occurrence of Hector’s beaked Whale Mesoplodon hector i (Gray) in South African Waters — G. J. B. Ross 195-204 14. A new species of House Snake from Swaziland, with notes on the status of the two Genera Lamprophis and Boaedon — N. Schaefer 205-208 Author index 209 Subject index 210 The Annals of the Cape Provincial Museums are published jointly by the five Cape Provincial Museums situated at East London, Grahamstown, Kimberley, King William’s Town and Port Elizabeth respectively. The editorial headquarters are at the Albany Museum, Grahamstown. The Journal is in- tended to record the results of research at the Museums in the fields of pre-history, ethnography and natural history. The Editorial Board wishes to acknowledge the generous financial assistance it receives towards the cost of publication of the Annals from the Provincial Administration of the Cape of Good Hope. INDEX TO AUTHORS BLACK, R. A. R. and G. J. B. ROSS. Aspects of adaptive radiation in Southern African Accipiters 57-65 DEACON, H. J. The Acheulian occupation at Amanzi Springs, Uitenhage district, Cape Province 89-189 FARQUHARSON, F. L. A new freshwater gobi (Pisces: Gobiidae) from Lake Sibayi, Zululand, South Africa 85-87 HUGHES, D. A. The Southern limits of distribution of commercially important penaeid prawns in South Africa. . 79-83 JUBB, R. A. The Nothobranchius (Pisces, Cyprinodontidae) of Southern Africa and a new species from Lake Chilwa, Malawi. Part 1 1-11 and P. LE ROUX. Revision of the Chiloglanis (Pisces: Mochokidae) of Southern Africa and descriptions of two new species. Part 1: Limpopo, Incomati and Pongola rivers 13-23 and F. O. PETRICK. Tilapia Mossambica Peters, from Australia 67-71 KILBURN, R. N. Taxonomic notes on South African marine mollusca, 1 39-48 LAWRENCE, R. F. Four new forest Millipedes from Lesotho and the Eastern Cape 49-55 LE ROUX, P. see JUBB, R. A. and P. LE ROUX. PETRICK, F. O. see JUBB, R. A. and F. O. PETRICK. QUICKELBERGE, C. D. Variational Patterns and Races of the Clapper Lark Mirafra apiata (Vieillot) 25-37 ROSS, G. J. B. The juvenile stadia of Anampses caendeopunctatus Ruppell (1829) 73-78 The Occurrence of Hector’s Beaked Whale Mesoplodon hectori (Gray) in South African waters . 195-204 ROSS, G. J. B. see BLACK, R. A. R. and G. J. B. ROSS. SCHAEFER, N. A new species of House Snake from Swaziland, with notes on the status of the two Genera Lampro- phis and Boaedon 205-208 WELLS, M. J. Plant remains from Amanzi Springs 191-194 INDEX TO SUBJECTS Accipiter Spp. Southern African adaptive radiation wing proportions Accipitrinae adaptive radiation wing proportions Acheulian culture African Goshawk Amanzi Springs excavation formation occurrence plant remains Anampses caeruleopunctatus, juvenile stadia Annual fishes Anvils . . . 57-65 . . . 61-65 . . . 62-65 . . . 57-65 . . . 61-65 . . . 62-65 . . . 89-189 . . . 59 . . . 89-189 . . . 93-98 ... 92 . . . 91-93 114-115, 191-194 . . . 73-78 . . . 1-11 ... 101 Balmoral Member Barbels Beaked Whale, Hector’s Bifaces .... Boaedon Spp. Botanical Remains Buccinidae .... Bullia ancillaeformis . . . . 92,193 . . . 13-23 . . .195-204 . 104, 173-178 . . . 205-208 114-115, 191-194 ... 41 . . . 40-41 Cantharus subcostatus comb. nov. Chiloglanis bifurcus sp. nov. . Chiloglanis emarginatus sp. nov. Chiloglanis spp Key Cinysca nom. nov Clapper Lark Cleavers Cobble tools Colubrid snake Columbellidae Cores Crassispira hottentota, comb, nov, Croilia mossambica .... Cymatiidae Cymatium durbanense Cymatium labiosum .... Cymatium parthenopeum Cyprinodontidae ... 41 . . . 17-20 . . . 21-23 . . . 13-23 . . . 13-14 . . . 46-47 . . . 25-37 . 104, 167-172 . 102,155-156 . . .205-208 . . . 42-44 100-101, 153-154 ... 40 ... 85 . . . 45-46 . . . 45-46 ... 46 ... 45 . . . 1-11 Diplopoda 49-55 Drillia burnupi 39-40 Enghura Member 92, 191 Fasciolariidae 41 Fawn-coloured Lark 25 Flakes 105,181-189 Flappet Lark 30 Gabar Goshawk 61 Geelhoutboom site 117 G/ossogobius giuris 85 Gnomeskelus basuticus sp. nov. . 52 Gnomeskelus graemi sp. nov 51-52 Gnomeskelus montifelis, sp. nov 52-51 Gnomeskelus outeniqua sp. nov 49-53 Gnome skeins 49-55 Distribution 54 Gobi 85-87 Gobiidae spp 85-87 Gonopod, Gnomeskelus 50 Goshawk, African 59 Gabar 61 Handaxes 102-104,160-166 Hawks, accipitrine 57-65 adaptive radiation 61-65 wing proportions 62-65 Hector’s Beaked Whale 195-204 House Snake 205-208 Killifishes 1-11 Labrid fish . 73-78 Lamprophis spp 205-208 Lamprophis swazicus, sp. nov 205-208 Larks, Clapper 25-37 Sabota 26 Latirus filmerae comb, nov 41 Limopsidae 47 Lithic Industry 98-107 Artefacts 100-106 Raw material 98-99 Little Banded Goshawk 59 Melierax gabar Mesoplodon hectori Mesoplodon spp. . Skull characters. Metapenaeus monoceros Millipedes .... Mirafra apiata spp. Mochokidae Mollusca, marine Monilea ponsonbyi Mudjair, Ikan . Muricidae .... . . 57-65 . .195-204 . .195-197 196-198, 202-3 . . 79 . . 49-55 . . 25-37 . . 13-23 . . 39-48 . . 47 . . 67 . . 44-45 Nassariidae 40-41 Natica forata 46 Naticidae 46 Nothobranchius kirki sp. nov 4-8 Nothobranchius spp 1-11 distribution 2 key to characters 10-11 Palaemon pact ficus Penaeid prawns southern distribution . juveniles, scarcity .... origin Penaeus spp Philobrya limoides .... Pleistocene vegetation Prawns, penaeid southern distribution . juveniles, scarcity .... origin Pteropurpura incurvispina nom. nov, Pyrene floceata egg capsules Pyrene natalensis comb. nov. 81 . 79-83 79 1 82 79 i . 79-83 47 91,92,111 193 . 79-83 79 82 79 . 44-45 . 42-44 42 42 ■ Radiocarbon dating 92,112 Radulae 43 Rietheuwel Member 92,96,191-193 Sabota Lark 26 Shrimps see Prawns Silhouetted sibayi sp. nov 85-87 Springs, as sites 112-114 Tilapia mossambica 67-71 Australian specimen compared 68-70 Trochidae 47 Trochus nigropunctatus 47 Turbinidae 46-47 Turridae 39-40 Ziphiidae 195-204 NEW NAMES PROPOSED IN THIS VOLUME SPECIES ARACHNIDA: DIPLOPODA basuticus ( Gnomeskelus ) Lawrence, 1970 52 graemi ( Gnomeskelus ) Lawrence, 1970 51 montifelis ( Gnomeskelus ) Lawrence, 1970 52 outeniqua ( Gnomeskelus ) Lawrence, 1970 49 PISCES: CYPRINODONTIDAE kirki ( Nothobranchius ) Jubb, 1969 4 PISCES: GOBIIDAE sibayi ( Silhouetted ) Farquharson, 1970 85 PISCES: MOCHOKIDAE bifur cus ( Chiloglanis ) Jubb and Le Roux, 1969 17 emarginatus ( Chiloglanis ) Jubb and Le Roux, 1969 21 SERPENTES: COLUBRIDAE swazicus ( Lamprophis ) Schaefer, 1970 205 ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prow Mus. (Nat. Hist.) s VOLUME 8 . PART 1 29th DECEMBER 1969 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMS! OWN SOUTH AFRICA MAR ?. 0 1970 The Nothobranehius (Pisces, Cyprinodonlidae) of Southern Africa and a new species from Lake Chilwa, Malawi PART 1 R. A. JUBB Due to the attractive colours of mature males, Nothobranehius , endemic to Africa, are popular aquarium species in many parts of the world. Amongst aquarists they are generally known as “killifishes” or “annual fishes”. The term “annual” is used since it was originally thought that these fishes completed their entire life cycle within the course of a single year. In nature the aquatic habitats of these fishes usually dry up during seasonal periods of pro- longed lack of rain, and the adults, which have spawned by this time, together with all other post-embryonic forms, die. The ability of Nothobranehius populations to survive both cyclically recurring periods of dryness, as well as to escape extinction due to erratic climatic conditions, has been found by Wourms (1964) to be due to the ability of individual eggs to enter into and remain in a state of developmental arrest or diapause during their normal ontogeny. From a neat summary published by Klee (1965) no fewer than fourteen species of Notho- branchius have been described from the region embracing Tanzania, Mocambique and the north-eastern lowveld of the Republic of South Africa. These are: N. ernini Ahl, 1935. Single specimen. Kongoran Botto, Tanzania. N. guentheri (Pfeifer), 1893. Zanzibar, Tanzania. N. kuhntae (Ahl), 1926. Beira, Mozambique. N. mayeri Ahl, 1935. Single specimen. Beira, Mozambique. N. melanospilus (Pfeifer), 1896. Longo Bay, Zanzibar, Tanzania. N. mkuziensis (Fowler), 1934. Mkuzi River, Natal, South Africa. N. neumanni (Hilgendorf), 1905. North Ugogo, Tanzania. N. orthonotus (Peters), 1844. Quelimane, Mozambique. N. palmquisti (Loennberg), 1907. Tanga, Usambara, Tanzania. N. raehovii Ahl, 1926. Beira, Mozambique. N. robustus Ahl, 1935. Swampy bay of Tschangarra, North Usinja, Tanzania. N. taeniopygus (Hilgendorf), 1888. Lake Tshaya, Bubu River, Tanzania. N. troemneri (Myers), 1926. Based on single aquarium specimen from East Africa. N. vosseleri Ahl, 1924. Single specimen. Mom bo. Tanzania. The approximate positions of the type localities of the species listed above are shown in fig. 1. A number of these species have not been recognized since, and Klee’s (he. cit.) proposed list of valid species and synonyms for this region is as follows: N. guentheri N. melanospilus (Synonym N. seychellensis Ahl, 1935). N. mkuziensis N. neumanni N. orthonotus (Synonyms: kuhntae? troemneri? mayeri?), N. palmquisti (Synonyms: vosseleri , emini). N. raehovii N. taeniopygus (robustus). 1 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 1, DECEMBER 1969 To a large extent this list is supported by the distribution of type localities shown on fig. 1 but it can be trimmed even further. There is some confusion about the status of Pfeffer’s melanospilus ( vide Boulenger, 1915, p. 34; Pfeffer, 1896, p. 48; Ahl, 1935, p. 128) and I am inclined to regard melanospilus as a synonym of orthonotus, the name originally used by Playfair & Gunther (1866). It should be noted that Smith (1963) does not record Nothobrcn- chius from the Seychelles and, indeed, it would be surprising to find representatives of this genus there, unless transported and transplanted by man. Crass (1964) records N. orthonotus from the Ndumu Game Reserve, northern Natal. It was from the Mkuzi River, a short distance to the south of this, that Fowler’s mkuziensis was described. Bruce Turner, American Museum of Natural History, has examined the type specimen of mkuziensis and has informed me (in. litt.) that it is in poor condition. As the type locality has had its environment changed considerably by man for agricultural purposes the validity of mkuziensis will never be settled by actual specimens. It has been accepted that 2 JUBB: THE NOTHOBRANCHIUS OF SOUTHERN AFRICA mkuziensis is a synonym of the widely distributed TV. orthonotus, but, after examing specimens of TV. rachovii, as well as excellent slides supplied to me personally and illustrations published by Turner ( 1 964), and referring them to Fowler’s ( 1 934) illustration, the possibility of mkuziensis being a synonym of rachovii must not be ruled out. It is probable that I was mistaken in not paying attention to the concentric bands on the caudal fin, and the blotches on the dorsal and anal fins, so reminiscent of the colour pattern of a male rachovii (see jubb, 1967, fig. 180 and Turner, 1964, fig. 2), which are illustrated by Fowler. As regards the other proposed synonyms 1 can find no difference of opinion between Turner {in lift.) and Klee {Joe. cit). Whilst discussing orthonotus it is of interest to record the following remarks by Mr. E. J. Seymour, British Killifish Association, who was supplied with a colour slide of a male TV. orthonotus from the Kruger National Park: “here TV. mclanospilus is regarded as the same fish as TV. orthonotus so at least your colour slide clears that up.” The description of Hilgendorf’s (1905) neumanni states that there are 32—36 scales around the body in front of the ventrals, but, the excellent illustrations of the type specimens, male and female, indicate that this is more likely to be 22 — 26. If this assumption is correct then I propose that TV. neumanni , for this as well as geographical reasons, be regarded as a synonym of the widely distributed TV. tacniopygus. This latter species is recorded from the streams of Lake Victoria (Greenwood, 1966), Lake Bangwelu, Upper Zambezi and Kafue River systems (Bell-Cross, 1965). I have examined specimens from a flood pool, Kafue River, between Mazabuka and the Kafue River Bridge. Both Klee ( loc . cit.) and Bell-Cross ( loc . cit.) consider TV. brieni Poll, 1938, to be a synonym of TV. taeniopygus. Under the name brieni Tait (1965) has described in some detail the habits and appearance of this species. I am indebted to Mr. G. Bell-Cross for notes and a colour photograph of an adult male TV. taeniopygus from the Kafue River system. There is no difficulty about the identification of a living specimen of an adult male of Ahl's rachovii with its spectacular colour pattern, particularly the caudal fin. The known distri- bution of this species extends from Beira southwards along the lowveld region to pans in the Kruger National Park, Eastern Transvaal, situated between the Olifants River (Limpopo system) and the Nwanetzi River (Incomati system) (Pienaar, 1968). Pans near this site are where the first specimens of TV. orthonotus were collected in the Kruger National Park. Table 1 is a summary of the general colour patterns of mature male specimens of TV. taeniopygus , supplied by Bell-Cross, TV. guentheri and TV. palmquisti , supplied by Haas, TV. rachovii , supplied by Haas and Pafenyk, TV. orthonotus supplied by Pienaar and Rose, and an undescribed Nothobranchius, supplied by Kirk and Goldberg. The colour slides of this latter species, discovered by Kirk in the Lake Chilwa drainage system, Malawi, were of the first mature male discovered, as well as of mature males in America which were bred from fish sent via West Germany. Every attempt has been made to confine these patterns to those of mature adult males as there are considerable variations in colour patterns during the various stages of development of the male from the immature stage to the plumage of a mature adult male. As pointed out by Tait {loc. cit.) development of the colour pattern starts in males at a size of about 25 mm., but previous to this their colour and appearance is similar to that of the ! rather drab females. The inland species N. tacniopygus is distributed along the western section of the region under discussion and the colour pattern of mature adult males can be recognized. The two species N. orthonotus and N. rachovii inhabiting suitable waters along the Mozambique coastal plains and lowveld of the eastern Transvaal and northern Natal can also be recognized. It is not as easy to separate TV. guentheri and TV. palmquisti individually but they can be separated from the three species mentioned above without difficulty. Resembling TV. guentheri and TV. palmquisti to some extent, but, differing markedly in the colour pattern of the dorsal and anal fins of mature adult males, is the Nothobranchius species discovered by R. G. Kirk, 3 I ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 1, DECEMBER 1969 fish biologist, in the drainage system of Lake Chilwa, Malawi. From fig. 1 it will be seen that the type locality is practically in the centre of the Tanzania, Mozambique, northern Natal region. This Nothobranchius is now described as a new species. Nothobranchius kirki sp. nov. Holotype: An adult male, total length 45 mm., Std. length 37-5 mm., collected on the 21st July, 1966, by Mr. R. G. Kirk, fish biologist, Agricultural Research Services, Ministry of Natural Resources, Malawi, in a pool adjacent to the Likangala River which forms part of the Lake Chilwa endoreic drainage basin, Malawi. Registered No. P.F. 994, Albany Museum, Grahams- town. Fig. 2. Male paratype Nothobranchius kirki sp. nov. Std. length 35 mm., total length 41 mm. Description: This species has been described from the holotype and nine paratypes, No. P.F. 995, from the same locality, four males and five females. The values in parentheses are those of the type. In percentage of standard length: Total length 116-0 — 120-0 (120-0); depth 30-0 — 33-5 i (32 • 0) ; length of head 30 • 0 — 34 • 0 (32 • 0) ; snout to origin of dorsal 58 • 0 — 62 • 4 (61 • 5) ; snout to origin of pectoral 31-6 — 33-5 (32-0); snout to origin of ventrals 48-0— 53-5 (50-6); snout to origin of anal 60-6 — 68-8 (62-5). Ovigerous females gave the higher values. In percentage of length of head: Length of snout 19-0 — 23-0 (20-7); eye 20-0 — 26-0 (25-0); interorbital width 39-8 — 42-7 (41-6). Scales markedly deciduous in preserved material, 26 — 28 in longitudinal series, lateral line pores absent. Scales around body immediately in front of pelvic fins 22 — 24. Snout short, flat and broad. In mature males the snout is covered with numerous small tubercles which extend to forehead, edges of adjacent scales and margins of orbits. These tubercles also appear on rays of dorsal and anal fins. Mouth directed upwards, lower jaw projecting. Teeth in upper jaw conical, sharply pointed, those of outer series largest; teeth in lower jaw mostly coarse with flattened crowns. Dorsal fin 15 — 17 (16); anal 15 — 18 (16). Origin of dorsal fin over origin of anal except in some distorted ovigerous females where the anal fin is displaced slightly posteriorly. Mature males have tips of anal and caudal rays extended to beyond membrane of these fins, 4 JUBB: THE NOTH OB RANCHI US OF SOUTHERN AFRICA a feature not evident in female material available. In specimens of the same size there is no significant difference in size of the anal fins, but males have slightly larger dorsal fins, the pos- terior rays being longer. Colouration: For the description of the colours of living mature adult males I am indebted to Mr. R. G. Kirk and Dr. R. J. Goldstein. The former supplied a colour transparency of one of the first males discovered near Lake Chilwa (jubb, 1967), and the latter a description (Goldberg, 1968) and colour transparencies. Figure 3, a monochrome photograph of Kirk’s first mature adult male, has been used to facilitate the description of this beautiful fish. Basically in the illustration all the black may be regarded as some shade of crimson, and the white as some shade of turquoise. The detailed description is as follows: Scale centres iri- descent turquoise (2) edged crimson (1), edging on scales forming a reticular pattern when fully developed; ventrum crimson from midway between insertion of pectorals to and along base of anal fin (12); crimson region broadest between ventrals and anal fin; pectorals (14) pectorals practically transparent, pale olive; pelvics ventrals (13) crimson tipped with black; anal crim- son at base (12), then a clear band (11) colourless or pale turquoise, the rest of the anal (10), except for the extreme edge where the extended rays are black, being crimson. Fig. 3. Key to colour pattern: 1, crimson; 2, turquoise; 3, turquoise with crimson markings; 4, golden, tinted turquoise; 5, edge turquoise; 6, crimson to red-brown; 7, membrane pale turquoise or olive; 8, colourless transparent border; 9, crimson; 10, crimson; 11, colourless or pale turquoise; 12, crimson; 13, crimson tipped black; 14, membrane practically colourless or pale olive. Caudal crimson (9), darker at base, with narrow black band, or, as in some specimens, a colourless posterior border (8). In either pattern the tips of any extended rays are black. The dorsal fin has membrane pale olive or pale turquoise (7) with crimson to red-brown spots and irregular bands (6); membrane darker towards extremity with edge turquoise (5). The pupil of the eye has a narrow golden border, the iris being golden, tinted turquoise in places with traces of pigmented vertical dark bar through eye (4). Chin and throat pale golden olive, operculum (3) turquoise with crimson markings. 5 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 1, DECEMBER 1969 Specimens preserved in 5% formalin solution are practically featureless. Both sexes have numerous small black dots visible through the scales, males being generally darker as shown in fig. 2. The membrane of the dorsal fin is pigmented giving a blotchy effect. The pale band in the anal fin, No. 11 in fig. 3, becomes a pigmented band of small dots in preserved material. This pale band, shown broken (11) in fig. 3 is sometimes continuous as shown in fig. 4, a photograph taken in North America of a male N. kirki (third generation) from Malawi stock. The pale band at the posterior extremity of the caudal fin is also pigmented with minute dots, as are extended tips of the rays of the anal fin and articulations of the lepidotrichia of both fins. Fig. 4. Photograph by Dr. Goldstein of a male N. kirki , third generation, bred in America. Living female specimens of A. kirki are drab silvery-grey with pale olive or pale turquoise- tinted fins. It is of interest to note that male breeding colours have been taken as characteristic of a species by other workers. This has been accepted both by Greenwood and Trewavas (1966) ) in the case of Cichlids. Habitat: Lake Chilwa lies 16 miles east of Zomba, Malawi, and approximately 32 miles from the nearest point on the Great Rift Valley. The lake basin is approximately 1,800 feet t above sea level, and its catchment is, today, endoreic. When first discovered it was thought t to have evolved as part of Lake Malawi, but, recent investigations (Kirk, 1967) indicate that t at one stage it formed a single body of water with Lake Chiuta which overflows into the : Lugenda River, a major tributary of the Rovuma River system. It should be noted that on l some maps Lake Chilwa is marked as Lake Shirwa. At present Lake Chilwa has an open area of water of some 260 square miles, and sur- rounding marshes of about the same area. From observations recorded during historical i times it is evident that the area of this productive lake varies considerably both annually and over long periods. It lies in an area of variable summer rainfall, which accounts for its s annual rise and fall in water level, and it also lies in the path of occasional Indian Ocean i cyclones which cross the Mozambique Channel and move inland accompanied by heavy rains. These extraordinary rains account for the huge areas that get inundated every decade or so. The lake drainage system therefore provides a suitable habitat for Nothobranchius. 6 JUBB: THE NOTHOBRANCHIUS OF SOUTHERN AFRICA The water in the lake and surrounding marshes is saline but this varies greatly according to the season, being about -01 p/1000 after good rains, to about 1 p/1000 at the end of the dry season which is during the months August to November. The lake is extremely productive and according to Kirk ( loc . sit. & in. lift.) practically the entire shoreline of Lake Chilwa is obstructed by dense growth of marginal vegetation, mainly Typha, but with occassional tracts of Phragmites and Papyrus. At low water the marshes are particularly dense, particularly to the north and north-west, the beds of reeds being inter- spersed with lagoons which cover a large area. In short the whole habitat provides excellent cover for small species of fishes and this probably accounts for the fact that N. kirki has remained undiscovered for such a long period. Although Lake Chilwa has no outlet, several rivers flow into it, the three largest entering the lake from the south-east. These are the Sombaxini, Palombe and Likanagala rivers, the latter two draining the Mlange and Zomba plateaux respectively. It was in isolated pools adjacent to the Likanagala River, and near the main body of water of Lake Chilwa that N. kirki were first discovered. Lake Chilwa is shallow, the deepest water being about 20 feet, and the adjacent pools and lagoons are both shallow and warm. Observations made during August 1 966 at a pool containing N. kirki gave a water temperature reading of 3 1 • 0 C (88 • 0 F.), pH 8-0 and salinity 16-5 mg/litre. Breeding habits and behaviour: Apart from the fact that female N. kirki collected during July and August were ovigerous with eggs 1*0 — 1*3 mm. in diameter, translucent off-white in colour with large centrally-located oil droplets visible in the yolk, nothing is known of the breeding habits of N. kirki. These fish would be ready for spawning as August is the beginning of the dry season when isolated pools would start drying up. It is unlikely that the breedings habits of N. kirki differ materially from those of N. taeniopygus described so care- fully by Wourms {loc. cit.), or behaviour of N. brieni described by Tait {loc. sit.). N. taeniopygus is a substrate spawner. Wourms supplied a substrate of fine white quartz sand for his investi- gations. A spawning male drives a female onto the surface of the substrate. Eggs and sperm are deposited in a depression in the substrate formed by the joint action of the two fishes, now in close opposition to one another. The eggs are then covered over with the caudal fins and the pair move away from the spawning site. Affinities: Kirk {loc. cit.) collected the following species within the Lake Chilwa basin and drainage system: Gnathonemus macrolepidotus (Peters), Petrocepha/us catostoma (Gunther). Cyphomyrus discorhynchus (Peters). Alestes imberi Peters. Barbus trimaculatus Peters, B. paludinosus Peters, B. manicensis Pellegrin, B. tangandensis Jubb, B. innocens Pfeffer, B. toppini Boulenger. Beirabarbus radiatus (Peters). Labeo cylindricus Peters. Clarias gariepinus (Burchell), B. theodorae Weber. Pareutropius longifilis (Steindachner). Tilapia shirana chilwae Trewavas, T. sparrmanii, T. melanopleura. Haplochromis callipterus (Gunther), Hemihaplochromis philander (Weber). Except for Barbus innocens , which is closely related to the widely-distributed B. unitaeniatus Gunther, all the above species are found in the inland waters of the east coast of Africa, some of them being widely distributed in Africa. The pressence of Pareutropius longifilis is of particu- lar interest as it is known only from the Ruvuma, Rufigi and Kingani River systems. Tre- wavas (1966) has re-examined the syntypes of Eutropius longifilis Steindachner, 1916, and 7 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 1, DECEMBER 1969 has found that they have 3 soft dorsal rays only, and are conspecihc with Pareutropius micristius Regan, 1920. It would not be surprising, therefore, to find N. kirki associated with the Ruvuma and Rufigi River systems. In the colour pattern of an adult breeding male N. kirki comes nearest to the closely related N. guentheri and N. palmquisti , but differs entirely in the colour patterns of the dorsal and anal fins, as shown on Table 1. It also differs from its western rela- tives, N. rubroreticuiatus Blache and Miton, from the Chad basin, and the N. kiyawensis Ahl, N. gambiensis (Svennson) group. This latter species, considered by Klee ( loc . cit.) to be a synonym of N. kiyawensis , was first described from a single female specimen, but later Johnels (1954) described a mature male in full plumage. Acknowledgements: The taxonomy of the freshwater fishes of southern Africa is part of a research programme carried out at the Albany Museum, and which is sponsored by the Council for Scientific and Industrial Research, Pretoria. Collaborators in this investigation have been mentioned in the text. Illustrations in colour of the species N. orthonotus , N. rachovii and the new species N. kirki will be found facing page 121, Jubb 1967, and the former two species in Pienaar 1968, pages 60 — 64. These latter illustrations are of particular interest as they include the females of the species. Mrs. H. M. Jubb kindly prepared the drawing of the paratype, fig. 2. I wish to thank Mr. R. G. Kirk for supplying the material and data relating to N. kirki and Mr. C. F. Jacot Guillarmod, Director of the Albany Museum, for reading and publishing this paper. (Addresses of collaborators overseas: Dr. R. J. Goldstein, Department of Biology, Emory University, Atlanta, Georgia, 30322. Mr. R. Haas, Department of Zoology, University of California, Los Angeles, California, 91364. Mr. B. J. Turner, American Museum of Natural History, Central Park West, 79th Street, New York, 10024. Mr. E. J. Seymour, British Killifish Association, 12 A, Hoylake Court, Ardler, Dundee, Scotland.) Part 2 deals with additional material from Mozambique. JUBB: THE NOTHOBRANCHIUS OF SOUTHERN AFRICA REFERENCES Ahl, E., J935. Ober neue seltene afrikanische Zahnkarpfen der Gattungen Aphyosemion und Nothobranchius. Zool. Anz., 112 (5/6): 125—129. Bell-Cross, G., 1965. Additions and amendments to the Check List of the fishes of Zambia. Occ. Papers Dept. Game and Fisheries, Zambia, No. 3: 29 — 43. Boulenger, G. A., 1915. Catalogue of the Freshwater Fishes of Africa. British Museum, London, 3: 31 — 37. Crass, R. S., 1964. Freshwater Fishes of Natal. Shuter & Shooter, Pietermaritzburg, p. 103. Fowler, H. W., 1934. Natal fishes obtained by Mr. H. W. Bell-Marley. Ann. Natal Mus., 7: 411. Goldberg, R. J., 1968. Description of a new species of Nothobranchius Peters, from Malawi (Nyasaland), East Africa. Journ. Amer. Killifish Assn., 5 (1): 1 — 3. Greenwood, P. H., 1966. The fishes of Uganda. Uganda Society, Kampala, pp. 96 — 7. Hilgendorf, F., 1905. Fische von Deutsch- und Englisch-Ost Africa. Zool. Jahrb., 22: 417. Johnels, A. G., 1954. Notes on Fishes from Gambia River. Arkiv. Zool. (2) 17: 397 — 399. Jubb, R. A., 1967. Freshwater fishes of southern Africa. A. A. Balkema, Cape Town, p. 121, pis. 29, 30, 31. Kirk, R. G., 1967. The Zoogeographical affinities of the fishes of the Chilwa-Chiuta Depression in Malawi. Rev. Zool. Bot. Afr., 76 (3—4): 295—311. Klee, A. J., 1965. A quick review of Nothobranchius. Journ. Amer. Killifish Assn. 2 (2): 11 — 16. Pienaar, U. de V., 1968. The freshwater fishes of the Kruger National Park. Koedoe No. 11: 60 — 64. Playfair & Gunther., 1966. Fishes of Zanzibar. London, p. 118. Poll, M., 1938. Poissons du Katanga recoltes par le professeur Paul Brien. Rev. Zool. Bot. Afr., 4, 30 (4), p. 409. Regan, C. T., 1920. Three new fishes from the Tanganyika Territory. Ann. Mag. Nat. Hist., (9) 6: 104 — 105. Smith, J. L B. & Smith, M. M., 1963. Fishes of Seychelles. Dept, of Ichthyology, Rhodes University, Grahamstown. Steindachner, F., 1916. Bericht fiber ichthyologische Aufsammlungen der Brfider Adolf und Albin Horn wahrend einer im Sommer 1913 ausgeffihrten Reise nach Deutsch-Ostafrika. Denkschr. Akad. Wien, 92: 59 — 86, pis. I — V. Svennson, G. S. O., 1933. Freshwater fishes from Gambia River. Kungl. Svenska Vetenskapsakad. Handl., Stockholm, (3) 12 (3): 81, fig. 25. Tait, C. C., 1965. Notes on the species Nothobranchius brieni Poll (Cyprinodontidae). Occ. Papers Dept. Game & Fish., Zambia, No. 3: 125—131. Trewavas, E., 1966. A preliminary review of fishes of the genus Tilapia in eastward-flowing rivers of Africa, with proposals for two new specific names. Rev. Zool. Bot. Afr., 74 (3 — 4): 395. Trewavas, E., 1966. Fishes of the genus THapia with four anal spines in Malawi, Rhodesia, Mozambique and southern Tanzania. Rev. Zool. Bot. Afr., 74 (1 — 2): 58 — 59. Turner, B. J., 1964. An introduction to the fishes of the genus Nothobranchius. African Wild Life, 18 (2): 117—124. Wourms, J. P., 1964. Comparative observations on the early embryology of Nothobranchius taeniopygus (Hilgendorf) and Aplocheilichthys pumilis (Boulenger) with special reference to the problem of naturally occurring embryonic diapause in Teleost fishes. Ann. Rep. E. Afr. Freshwater Fish, res. Org., Jinja, 1964, pp. 68 — 73. 9 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 1, DECEMBER 1969 Table I Species Scales Dorsal Caudal N. taeniopvgus D 15—18 ' A 15—19 L.L. 27—32 Tr. 22—26 Edged carmine or brick-red, centres iridescent turquoise. Membrane liver-coloured or brick-red, turquoise spots or blotches. Dark red or brick-red 2 base, fading to brigf orange band with distinc black border. N. guentheri D 17—18 A 18—19 L.L. 27—30 Tr. 24—28 Edged carmine or dark red; centres iridescent turquoise. Membrane olive with dark red or liver-coloured spots, fin darker towards ex- tremity and bordered with white. Carmine to dark red wit paler band towards e> tremity, border jet-blac) N. palmquisli D 15—16 A 14—15 L.L. 26—28 Tr. 22—26 Edged carmine or dark red; centres iridescent turquoise. Membrane olive with small spots, reddish at base, be- coming liver-coloured to- wards extremity. Entirely carmine with pai? or transparent border. N. rachovii D 14—16 A 15—16 L.L. 25—27 Tr. 22—24 Edged carmine; centres iri- descent turquoise. Membrane turquoise with large liver-coloured spots or blotches forming irregular transverse bands, fin bor- dered faintly with white. Four to six concentric band:;! spotted or mottled tui quoise at base, then turquoise band follown| by brilliant orange ban which pales to yellow, thdi! a distinct black band fain ly edged with white. N. orthonotus D 14—16 A 14—17 L.L, 27—32 Tr. 22—27 Edged maroon or liver- coloured; centres iridescent blue, turquoise or dark green. Membrane olive with nume- rous small liver-coloured spots, extremity of fin being dark and often edged with white. Dark olive with rays, am sometimes small spot liver-coloured. N. sp. nov. D 14—16 A 15—18 L.L. 26—28 Tr. 22—24 Edged crimson; centres iri- descent turquoise. Membrane pale olive with large carmine or red-brown spots and irregular bands, extremity of fin darker and bordered with white or pale turquoise. Entirely crimson, the bai being darker and cxtremi colourless or edged win thin black border. 10 1 ** w 1 JUBB: THE NOTHOBRANCHIUS OF SOUTHERN AFRICA Anal Ventral Pectoral Ventral surface and eye ick-red at base followed by darkish band, the remaining two-thirds being bright orange, bordered ventrally by a black band. Carmine or orange- red with black tips. Transparent pale olive or orange membrane. Distinctly rufous especially lower parts of gill cover and mandible, scales on head also turquoise. Iris golden or tinted blue, pupil bordered gold, vertical dark bar through eye. embrane pale olive with car- mine or liver-coloured spots, the extremities of some rays black. Dark olive tipped black. Transparent Pale olive membrane. Golden-olive with some red on gill cover and lower part of mandible. Iris golden tinted blue, pupil bordered gold with darker vertical bar through eye. embrane olive with small car- mine to liver-coloured spots near base. Olive tipped with dark red. Transparent pale olive membrane. Golden-olive. Iris turquoise with traces of vertical dark band through eye. smbrane turquoise with large liver-coloured spots forming rregular bands, the fin being bordered ventrally with thin white band. Turquoise with large brick-red or liver- coloured spot at base of fin. Transparent olive membrane crim- son tinted in axil of fin. From vent to gill opening golden-olive, ihe sides of the head and lower jaw being carmine. Iris turquoise, pupil edged with gold, black or dark vertical bar through eye. imbrane pale olive wiih nume- rous liver-coloured or magenta ;pots, extremity mauve tinted with white border. Liver coloured, often tipped with white. Transparent pale olive. Golden, green tinted, with numerous small liver- coloured spots. Iris golden with dark vertical band through eye. { simson at base, then pale olive br pale turquoise band with re- mainder of fin crimson. Tips of in rays black. ii Crimson tipped black. Transparent pale olive or turquoise. From vent to beyond ventrals distinctly rufous, but throat and lower jaw olive. Iris golden, tinted turquoise in places, traces of darker verti- cal bar through eye. 11 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 1, DECEMBER 1969 JUBB: THE NOTHOBRANCHIUS OF SOUTHERN AFRICA Table I Species Scales Dorsal Caudal Anal Ventral Pectoral Ventral surface and eye TV. taeniopygus D 15—18 A 15—19 L.L. 27—32 Tr. 22—26 Edged carmine or brick-red, centres iridescent turquoise. Membrane liver-coloured or brick-red, turquoise spots or blotches. Dark red or brick-rti, .sSfred at base followed by base, fading to It darkish band, the remaining orange band with di. tw0-thirds being bright orange, black border. bordered ventrally by a black band. Carmine or orange- red with black tips. Transparent pale olive or orange membrane. Distinctly rufous especially lower parts of gill cover and mandible, scales on head also turquoise. Iris golden or tinted blue, pupil bordered gold, vertical dark bar through eye. TV. guentheri D 17—18 A 18—19 L.L. 27—30 Tr. 24—28 Edged carmine or dark red; centres iridescent turquoise. Membrane olive with dark red or liver-coloured spots, fin darker towards ex- tremity and bordered with white. Carmine to dark red paler band towards tremity, border jet-bi Membrane pale olive with car- mini or liver-coloured spots, the extremities of some rays bife. Dark olive tipped black. Transparent Pale olive membrane. Golden-olive with some red on gill cover and lower part of mandible. Iris golden tinted blue, pupil bordered gold with darker vertical bar through eye. TV. palmquisti D 15—16 A 14—15 L.L. 26—28 Tr. 22—26 Edged carmine or dark red; centres iridescent turquoise. Membrane olive with small spots, reddish at base, be- coming liver-coloured to- wards extremity. Entirely carmine with or transparent borde iemb min neat rane olive with small car- to liver-coloured spots base. Olive tipped with dark red. Transparent pale olive membrane. Golden-olive. Iris turquoise with traces of vertical dark band through eye. TV. rachovii D 14—16 A 15—16 L.L. 25—27 Tr. 22—24 Edged carmine; centres iri- descent turquoise. Membrane turquoise with large liver-coloured spots or blotches forming irregular transverse bands, fin bor- dered faintly with white. Four to six concentric h femb spotted or mottled livei quoise at base, the irreg turquoise band folk bore by brilliant orange whit which pales to yellow. 1 a distinct black band! 1 ly edged with white. rane turquoise with large -coloured spots forming ular bands, the fin being red ventrally with thin e band. Turquoise with large brick-red or liver- coloured spot at base of fin. Transparent olive membrane crim- son tinted in axil of fin. From vent to gill opening golden-olive, ihe sides of the head and lower jaw being carmine. Iris turquoise, pupil edged with gold, black or dark vertical bar through eye. TV. orthonotus D 14—16 A 14—17 L.L. 27—32 Tr. 22—27 Edged maroon or liver- coloured; centres iridescent blue, turquoise or dark green. Membrane olive with nume- rous small liver-coloured spots, extremity of fin being dark and often edged with white. Dark olive with rays, sometimes small s liver-coloured. lemb rous spot with rane pale olive with nume- iiver-coloured or magenta extremity mauve tinted white border. Liver coloured, often tipped with white. Transparent pale olive. Golden, green tinted, with numerous small liver- coloured spots. Iris golden with dark vertical band through eye. TV. sp. nov. D 14—16 A 15—18 L.L. 26—28 Tr. 22—24 Edged crimson; centres iri- descent turquoise. Membrane pale olive with large carmine or red-brown spots and irregular bands, extremity of fin darker and bordered with white or pale turquoise. Entirely crimson, the being darker and extr colourless or edged thin black border. rimsc or p mair fin r n at base, then pale olive ile turquoise band with re- der of fin crimson. Tips of ays black. Crimson tipped black. Transparent pale olive or turquoise. From vent to beyond venlrals distinctly rufous, but throat and lower jaw olive. Iris golden, tinted turquoise in places, traces of darker verti- cal bar through eye. 10 11 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (Nat. Hist.) VOLUME 8 • PART 2 29th DECEMBER 1969 PUBLISHED JOINTLY BY THE :APE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Revision of the Chiloglanis (Pisces : Mochokidae) of Southern Africa and descriptions of two new species PART 1 : Limpopo, Incomati and Pongola rivers R. A. JUBB Albany Museum, Grahamstown P. LE ROUX* Provincial Fisheries Institute, Lydenburg Until Crass (1960: 445 — 456) published detailed descriptions of Chiloglanis species found in the inland waters of Natal little was known of the representatives of this genus in southern Africa. Further systematic collecting has been carried out since that date and it is now possible to review the Chiloglanis of this region. This section is devoted to the Chilo- glanis of the Limpopo, Incomati and Pongola River systems collected by personnel of the Lydenburg Provincial Fisheries Institute, Dr. U. de V. Pienaar of the Kruger National Park, Mr. R. S. Crass and Mr. B. G. Donnelly. With abundant material available the authors have found that the Chiloglanis of this southern region can be divided into four distinct groups by the character and maximum number of mandibular teeth, the maximum number of mandibular teeth in each case being that of a complete row, whether functional or replacement teeth. This grouping can be ex- tended also to Zambezi River material, see Figure 1 1 , which will be discussed in a subsequent paper. Based on this grouping and other characteristics the following key to their identification has been prepared : Key to the identification of the Chiloglanis of the Limpopo, Incomati and Pongola River systems. |i 1. Mandibular teeth short, up to 14 in number widely spaced, width of band 40 — 50 % of width of inside of mouth; maxillary and mandibular barbels long; dorsal spine not serrated: C. swierstrai v. d. Horst. Fig. 1. * Director of Nature Conservation, Bloemfontein. 13 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 2, DECEMBER 1969 2. Mandibular teeth long, up to 8 in number widely spaced, width of band 25 — 40% of width of inside of mouth; dorsal spine not serrated : (a) Mandibular barbels short, caudal forked, the upper lobe often larger than the lower: C. bifurcus sp. nov. ( b ) Mandibular barbels long, caudal emarginate : C. emarginatus sp. nov. 3. Mandibular teeth long, slender, up to 12 in number closely spaced at base, width of band at base 20 — 30% of width of inside of mouth; dorsal spine serrated: C. paratus Crass. Fig. 3. 4. Fig. 4. Mandibular teeth long, up to 12 in number rj closely spaced, width of band 30 — 40 % of width i of inside of mouth; dorsal spine not serrated: (a) Mature males with median rays of caudal fin i elongated; females emarginate: C. anoterus Crass. (b) Males and females with emarginate caudal fins; dorsal spine length variable, 12 — 1 5 % of standard length in Limpopo populations, 5 — 10% of standard length in Incomatiij River populations: C. pretoriae v. d. Horst. 14 JUBB AND LE ROUX: REVISION OF THE CHILOGLANIS OF SOUTHERN AFRICA Chiloglanis swierstrai v. d. Horst, 1931. (Fig. 5). 1931: v. d. Horst, p. 249, Fig. 3. 1960: Crass, pp. 247, 451, text-fig. 3. C. engiops . syn. nov. 1964: Crass, p. 94, Fig. 23, F & G, C. engiops. 1967: Jubb, p. 147, Fig. 165. 1968: le Roux & Steyn, pp. 85 — 6. 1968: Pienaar, p. 53, Fig. 31. Crass’ C. engiops was based on material from the Pongola River only, no specimens from rivers to the north of this system having been collected at that time. Due to pollution of the upper reaches of the Crocodile, Pretoria District, it has not been possible to obtain specimens of C. swierstrai from the type locality, but, it has been possible to compare material from the Elands River, Limpopo system, and the Incomati River system with C. engiops from the Pon- gola River and the type specimen of C. swierstrai , Transvaal Museum No. 8655. This latter specimen is in poor condition but from this study it is considered that C. engiops is a synonym. Distribution: Usually below 3,000 feet in pools and backwaters of the Pongola and Incomati River systems, and those tributaries of the Limpopo River which rise in the Transvaal. Fig. 5. Chiloglanis swierstrai (X 2). Chiloglauis paratus Crass, 1960. (Fig. 6). 1960: Crass, pp. 452 — 456, text-fig. 4. 1964: Crass, p. 96, Fig. 23, D & E. 1967: Jubb, p. 145, Fig. 163. 1968: le Roux & Steyn, pp. 83 — 4. 1968: Pienaar, P. 54, Fig. 32. Crass' (1960) text-fig. 4 (d) illustrates clearly the arrangement and character of the mandi- bular teeth. Distribution: This species has been collected from the Incomati River system and the Pon- gola River system. Specimens have been collected by B. G. Donnelly from the Limpopo River near Tuli which is upstream of Beitbridge. I. G. Gaigher has also collected C. paratus from the Limpopo system but it is not common. 15 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 2, DECEMBER 1969 Fig. 6. Chiloglanis paratus (X 2). Chiloglanis pretoriae v. d. Horst, 1931. (Fig. 7). 1931 : v. d. Horst, p. 248, Fig. 2. 1931 : v. d. Horst, p. 250, Fig. 4: C. pumilus. syn. nov. 1960: Crass, p. 447, pp. 450 — 1. 1967: Jubb, p. 145, Figs. 160 & 162 (C. pumilus v. d. Horst). 1968: Pienaar, p. 55, Fig. 33. 1968: le Roux, pp. 83 — 86. A variable species throughout its distribution. Limpopo River populations have a dorsal spine length of 12 — 15% of the standard length, the dorsal spine length being shorter in specimens from the Blyde River, and very short in specimens from the Incomati River system where the dorsal spine is only 5 — 10% of the standard length. The type specimen of C. pretoriae cannot be found but a type specimen of C. pumilus , Transvaal Museum No. T.M. 8655, has been examined by the authors. It is considered that this is a half-grown specimen of C. pretoriae. Distribution: Tributaries of the Limpopo River system, both in Rhodesia and the Transvaal, as well as tributaries of the Incomati River. Fig. 7. Chiloglanis pretoriae (X 2-3). 16 JUBB AND LE ROUX: REVISION OF THE CHILOGLANIS OF SOUTHERN AFRICA Fig. 8. Chiloglanis anoterus (X 2). Chiloglanis anoterus Crass, 1960. (Fig. 8). 1960: Crass, pp. 446—450, text-fig. 2. 1964: Crass, p. 94, Fig. 23, A, B & C. 1967: Jubb, p. 145, Fig. 161, A & B. 1968: Pienaar, p. 51, Fig. 30. 1968: le Roux, pp. 83 — 84, distribution map. A species of the upper reaches of tributaries of the Pongola and Incomati River systems which is closely allied to C. pretoriae. It differs from this species chiefly in the form of the caudal fin, the median rays of the caudal fin of mature males being elongated to give the fin a pennant-like appearance. This elongation of the median rays is not always symmetrical and is barely noticeable in some specimens. Females of this species do not have the median rays elongated and it is not possible, without knowing the origin of the specimens, to separate female C. anoterus from female C. pretoriae. It is, however, possible to separate female C. anoterus from females of the Incomati River form of C. pretoriae as the latter tend to have shorter dorsal spines. Work by Gaigher, as yet unpublished, on the distribution of fishes in the Incomati River system, indicates that there may be some ecological separation between the two species as males of C. anoterus and the Incomati River form of C. pretoriae do not occur together. Both species show a preference for rocks in flowing water. Distribution: Parts of the Incomati and Pongola River systems. Chiloglanis bifurcus sp. nov. Fig. 9. Chiloglanis bifurcus sp. nov. Std. length 68 mm. Type specimen. 17 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 2, DECEMBER 1969 Fig. 9a. Enlarged ventral view of mouth of type specimen of C. bifurcus. Holotype: Male. Std. Length 68 mm., fork length 76 mm. Collected by I. G. Gaigher on 14th September, 1965, from the Crocodile River, Incomati River system, Montrose Farm, Nelspruit District. Lydenburg Provincial Fisheries Institute No. P.F. 348, Albany Museum i No. P.F. 996. Description: From type specimen and nine paratypes selected from Lydenburg Provincial I Fisheries Institute specimens Nos. P.F. 67/2B, Elands River, and 67/16B, Crocodile River, Incomati River system, collected by I. G. Gaigher, Albany Museum No. P.F. 997. All measurements taken direct using calipers or dividers, the head length being taken as $! from the top of the gill opening to the tip of the snout. The value in parenthesis is that of the type specimen. In percentage of standard length: Length of head 27 — 29 (29); length of snout 17 — 19 (19); snout to origin of dorsal fin 36 — 40 (40); snout to origin of adipose fin 68 — 73 (73); snout t to origin of anal fin 65 — 68 (65); posterior base of dorsal fin to origin of adipose fin 20 — 26 ■] (23); length of dorsal spine 12 — 16 (16); length of pectoral spine 16 — 20 (19); length of base of adipose fin 11 — 14 (12). In percentage of head length: Snout length 59 — 65 (65); length of maxillary barbel 25 — 41 (25); length of outer mandibular barbel 9—14 (9); eye 12 — 15 (15); interorbital width 27 — 31 (28). 18 30. Chiloglonis onoterus Crass, 1960. Penant-tailed dwarf catfish or rock catlet. Wimpelstert-suierbekbaber. 31. Chiloglonis swierstroi v.d. Horst, 1931. Synonym: C. engiops Crass, 1960. Bearded or Slender dwarf catfish. Langbaard- of Bont-suierbekbaber. 32. Chiloglanis parotus Crass, 1960. Spiny dwarf catfish or rock catlet. Swart of Gestekelde suierbekbaber. 33. Chiloglanis sp. cf. C. pretoriae v.d. Horst, 1931. Lesser dwarf catfish. Kleinste suierbekbaber. JUBB AND LE ROUX: REVISION OF THE CHILOGLANIS OF SOUTHERN AFRICA A ventral view of the mouth is shown in figure 9A of the type specimen. The paratypes, which are smaller specimens, have longer maxillary and mandibular barbels. The long, widely-spaced teeth, eight in number, which curve inwards are clearly visible. Compare the character of these teeth with those of C. pretoriae shown in figure 10, and those of a common Zambezi River species, C. cf neumanni Boulenger illustrated in figure 11. In this latter example it will be seen that the teeth are minute, the width of the band being approximately 10 — 13% Fig. 10. Ventral view of mouth of a specimen of C. pretoriae showing functional and replacement teeth. of the inner width of the mouth. The mouth of C. bifurcus is surrounded by a large circular, papillose lip, thickened anteriorly and posteriorly divided down the midline. The premaxillary teeth, on two large oval pads separated at the midline, are pointed, widely spaced and form three transverse series. The vomerine teeth, separated from the premaxillary teeth, are variable in number and position. Dorsal fin I 6, the spine, not serrated, 12 — 16% of the standard length. Anal fin III 8, Caudal fin forked, the upper lobe in many specimens being larger than the lower. 19 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 2, DECEMBER 1969 Colour: Dorsal surface olive light brown with light olive blotches, pale on ventral surface. There is a distinct pale mid-lateral line. Fins colourless except for caudal, anal and ventral fins which have some rays pigmented, these pigmented areas being more accentuated in speci- mens preserved in formalin. The skin is rugose and covered with minute protuberances which appear white in preserved material. Food: The large sucker-like mouth is well adapted for clinging to rock surfaces. Gut contents indicate that these small fishes feed on epiphytic flora and associated fauna found on rocks in running water, aquatic insect larvae being predominant. Some specimens contained the empty shells of small molluscs. Habitat: During his surveys Gaigher found this species to be associated with rapids in perennial streams, and to be absent from pools or annual streams. Breeding Habits: Maturity is reached at a size of about 25 mm. standard length. It is not known where these fish spawn but ripe females were collected during the month of January. Distribution: This new species has been found only in the Elands and Crocodile rivers, tributaries of the Incomati River. The specific name bifurcus refers to the shape of the caudal fin which is forked with large lobes. Fig. 11. Ventral view of mouth of a specimen of C. cf. neumanni , Middle Zambezi system. Note relatively narrow band of small teeth. 20 JUBB AND LE ROUX: REVISION OF THE CHILOGLANIS OF SOUTHERN AFRICA Chiloglanis emarginatus sp. nov. Fig. 12. Chiloglanis emarginatus sp. nov. Type specimen Std. length 57-5 mm. Fig. 12a. Enlarged ventral view of mouth of type specimen of C. emarginatus. 21 ANN, CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 2, DECEMBER 1969 Holotype: Female. Std. length 57-5 mm., fork length 65-0 mm. Collected by I. G. Gaigher on 15th May, 1967, from Lekkerloop River, tributary of the Komati River of the Incomati River system, on the farm Vergelegen, Carolina District. Provincial Fisheries Institute, Lydenburg, No. P.F. 532, Albany Museum No. P.F. 953. Description: From the type and nine paratypes, some of the latter taken from P.F. M/67/14 collected by I. G. Gaigher on the 12th January, 1967, from the Komati River, on the farm Vlakfontein, Carolina District. Albany Museum No. 954. In percentage of standard length, values in parentheses being that of the type specimen: Fork length 109 — 113 (113); head length 26 — 28 (28); snout length 15 — 18 (16); snout to origin of dorsal fin 36 — 38 (36); snout to origin of adipose fin 63 — 69 (69); snout to origin of anal fin 63 — 67 (65); posterior base of dorsal fin to origin of adipose fin 20- -23 (23); length of dorsal spine 15 — 19 (17); length of pectoral spine 17 — 19 (19); length of adipose fin base 16—19(17). In percentage of head length: Snout length 54 — 69 (56); pectoral spine 59 — 71 (69); maxillary barbels 37—57 (50); outer mandibular barbels 13 — 28 (20); width of eye 15 — 21 (16); inter- orbital width 23 — 32 (25). A ventral view of the mouth of the type specimen is shown in figure 12A. This illustrates the difference in the general shape of the mouths of these two species. The teeth, which are not as robust as those of C. bifurcus are widely spaced also, the maximum number in a com- plete number being eight and the band width 25 — 35% of the inner width of the mouth. The premaxillary teeth, on two large oval pads separated at the midline, are pointed, widely spaced and form three transverse series. The vomerine teeth, separated from the premaxillary teeth, are variable in number and position. Both the maxillary and mandibular barbels are longer than those of C. bifurcus. Dorsal fin I 6, the spine, not serrated, 15 — 19% of the standard length. Anal fin III 8, caudal fin emarginate. Colour: Dorsal surface olive light brown with light olive blotches, pale on ventral surface. There is a distinct pale mid-lateral line. Fins colourless except for caudal, anal and ventral I fins which have pigment on rays forming bars along the middle of each fin. The skin is rugose and covered with minute protuberances which appear white in specimens preserved in formalin. Food: Feed on fauna associated with epiphytic flora found on rocks in perennial streams. Gut contents were found to contain remains of epiphytic algae and aquatic insect larvae, , the type specimen being engorged with Simulium larvae. Habitat: Lives amongst rocks, usually associated with rapids, in perennial streams. Breeding Habits: Maturity is reached at a standard length of about 25 mm. It is not known i; where these fish spawn. Females collected by Gaigher in May 1967 have ovaries forming, but f those collected during January the same year were found to be ripe and ready for spawning. Distribution: C. emarginatus has been collected by R. Pott in tributaries of the Pongola r River system in the Paulpietersburg District, and a single specimen from the Pungwe River, Inyanga District, Rhodesia, was submitted by D. H. Plowes of Umtali. From this it would appear that C. emarginatus occurs in rocky, perennial tributaries of east-flowing rivers from i the Pongola northwards to the Pungwe River of the Inyanga Mountains, Rhodesia. The specific name emarginatus refers to the shape of the caudal fin of this new species. 22 JUBB AND LE ROUX: REVISION OF THE CHI LOG LAN IS OF SOUTHERN AFRICA Affinities: Both Dr. Max Poll, Musee Royal du Congo Beige , Tervuren, Belgium, and Mr. James Chambers of the British Museum (Natural History), London, have provided the authors with minute details regarding the dentition of type specimens of Chi log Ian Is housed in their respective museums. It has not been possible to correlate C. bifurcus or C. emarginatus with any species of Chiloglanis described from north of the Zambezi River system, or with material supplied by G. Bell-Cross, Game and Fisheries Department, Chilanga, Zambia. Beyond what is published it has not been possible to obtain further details about Pellegrin’s (1936) Chilo- glanis faseiatus described from the upper Okavango River, and no additional material is available from this area. In his description Pellegrin considers C. faseiatus , mandibular teeth 14, to be related to C. pretoriae. Both C. bifurcus and C. emarginatus can be separated from known Chiloglanis of the Limpopo, Incomati and Pongola rivers by their mandibular dentition. From one another C. emarginatus can be distinguished by its longer maxillary and mandibular barbels, its longer adipose fin base and emarginate caudal fin. In addition to these characteristics C. bifurcus is a more robust species with a broader pectoral girdle when specimens of a similar standard length are compared. ACKNOWLEDGEMENTS Research work carried out by the senior author, a Research Associate of the Albany Museum, Grahamstown, is sponsored by the Council for Scientific and Industrial Research, Pretoria. The authors are indebted to the Director, Department of Nature Conservation, Pretoria, for permitting material collected by Messrs. Gaigher and Pott, of the Lydenburg Provincial Fisheries Institute, to be supplied for this investigation, and to Mr. R. S. Crass, Senior Re- search Officer, Natal Parks, Game and Fish Preservation Board, Pietermaritzburg, for supplying material from Natal rivers. Distribution records were completed by material supplied by Mr. B. G. Donnelly, Rhodesia, Mr. D. H. Plowes through the Umtali Museum, Rhodesia, and Mr. G. Bell-Cross, Game and Fisheries Department, Chilanga, Rhodesia. Dr. Max Poll supplied taxonomical data relating to Angola and Congo species, and Mr. James Chambers data relating to type material in the British Museum (Natural History). This paper is published by courtesy of the Director of the Albany Museum, Mr. C. F. Jacot Guillarmod. The four colour plates appended were taken from Pienaars (1968) Fresh- water Fishes of the Kruger National Park by kind permission of the Director, National Parks Board, Pretoria, and the illustrations for this paper prepared by Mrs. H. M. Jubb, REFERENCES Crass, R. S., 1960. Notes on the freshwater fishes of Natal with descriptions of four new species, Amu Natal Mas., 14 (3): 405 — 458. Crass, R. S., 1964. Freshwater fishes of Natal, Shuter & Shooter, Pietermaritzburg. Jubb, R. A., 1967. Freshwater fishes of southern Africa A. A. Balkema, Cape Town. le Roux, P. & Steyn, L., 1968. Fishes of the Transvaal, S.A. Breweries Institute, Johannesburg. Pellegrin, J., 1936. Contribution a ITchthyologique de F Angola, Arq. Mus. Bocage Lisboa, 7: 45 — 62. Pienaar, U. de V., 1968. The freshwater fishes of the Kruger National Park, Koedoe, 11: 1 — 82. Poll, M., 1967. Contribution a la faune ichthyologique de F Angola, Publ. Cult. Comp. Diamantes Angola, vol. 75, Lisboa. v. d. Horst, C. J., 1931. Some South African Siluroid fishes, Ann. Transvaal Mus., 14 (3): 246 — 250. il 23 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mas. (Nat. Hist.) PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Variational Patterns and Races of the Clapper Lark Mirafra apiata (Vieillot) by C. D. QUICKELBERGE Ornithologist, East London Museum No complete revision of this highly variable and cryptic species has been attempted since J. D. Macdonald’s excellent account compiled some fifteen years ago. By this author’s own admission, his paper established only the first stepping-stone towards an accurate under- standing of the full extent of the variational pattern displayed by this lark. Since that time more concerted collecting by various South African museums has led to a considerable aug- mentation of the available specimen material in our collections. In the light of this additional material it is now possible to present a much clearer picture of the individual and geographical variation in this species, and while it has become evident that well-marked races do exist, knowledge gleaned during the course of this work has contributed most towards a fuller understanding of the amazing extent of this lark’s range of individual variation. Former ignorance of the true magnitude of this variability has been directly responsible for the gross additions to the list of described sub-species over the years. We may trace the path of such over-splitting if the races M.a. apiata (Vieillot), M.a. adendorffi Roberts and M.a. algoensis (Roberts) are taken as examples. If all the available skins of these races are examined collectively, irrespective of racial affiliation, a certain spectrum of variation will be noted. That this is only individual and not geographical is easily demonstrated if a com- prehensive series of skins collected from any point within the combined ranges of these three contiguous races is examined. It will be seen that any such series will show as wide a range of variation as that of the pooled body of material alluded to above, i.e., all three races combined. The danger then of describing sub-species from both limited material and number of localities, especially in highly variable species, becomes obvious. It could happen, for example, that the diagnostic characters of a race described from such an incomplete series may only be revealing one segment or a single facet of the total possible individual variation, thus creating the impression of the entity constituting a distinct taxon. The fuller series from individual locali- ties subsequently reveals the complete variational pattern and the necessity for dropping any such superfluous races. Such a series as the one assembled by the Durban Museum of topotypical specimens of M.a. apiata from Malmesbury in the western Cape has thus helped considerably in revealing that the minor characters which formed the basis upon which the races adendorffi and algoensis were proposed, completely disappear as only part of a wider character complex or mosaic, inherent in all populations of the three races under discussion. It is anticipated, therefore, that as our series of skins of various bird groups become fuller and more representative of their ranges it will be found necessary to dispense with many of the races burdening the literature. This is most likely to happen in groups with cryptic plumage coloration such as larks, which are usually highly variable and have thus been notorious for their accumulation of racial names. Recently, a start in this direction has been made by South African specialists, and Mr P. A. Clancey (1966), in a recent revision of the races of the Fawn-coloured Lark Mirafra africanoides Smith has reduced the number of races to half of those previously admitted by authors. In a similar revision of the South 25 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 3, DECEMBER 1970 African Sabota Lark Mirafra sabot a Smith, Clancey (1965) likewise found that no less than six races out of some thirteen proposed by systematists could not be maintained, and re- quired to be synonymized with other racial taxa. My findings in the Clapper Lark likewise show that the species has been over-split on inadequate evidence, and that the number of races requires to be reduced. GEOGRAPHICAL VARIATION From a study of over 200 specimens from most parts of the species’ range it was clear at the outset that all the forms show intergradation with each other, making their recognition as geographical variants of only one polytypic species imperative. The only possible exception is the apparent lack of complete intergradation between M.a. apiata and M.a. hewitti. How- ever, the rather close resemblance of some specimens of one form to variants of the other would militate against their being treated as specifically distinct. The three extremes in variation found in the Clapper Lark are, however, so diverse in appearance that it is hardly suprising that pioneer taxonomists working without a full range of specimens came to regard these peculiar forms as discrete species. There is, on the one Illustration depicting stages in the increasing development of rufous dorsal pigments in three races of Mirafra apiata. Each example represents the pigmented terminal portion of a feather, each from the same region of the upper-parts. Such feather patterning varies quite considerably through wear and individual variation. 26 QUICKELBERGE: VARIATIONAL PATTERNS AND RACES OF THE CLAPPER LARK hand, the very dark, black-and-grey-backed south-western form with a minimum of dark russet mottling over the dorsal surface, exemplified by the race Mirafra apiata marjoriae Winterbottom. North-east of this, mainly over the highveld, we have the very russet M.a. hewitti (Roberts), in which the dark greyish dorsal colours have virtually disappeared, leaving the back plain reddish-brown in some specimens. The third extreme is perhaps the most divergent of all in that russet dorsal colours fade out completely in what is now a north and westerly direction until the race M.a. nata Smithers is encountered in Botswana, this last named presenting a remarkable ashy-grey dorsal appearance. One could concede that as in various Cisticola forms, there are greater colour differences between M.a. apiata races than those separating some closely allied species in the Alaudidae , e.g., M. rufocinnamomea (Salvadori) and M. apiata. Of the dark, mottled, grey-backed southern populations only two sub-species are ad- missible, the one being the extreme form, i.e., M.a. marjoriae , which is found to have a wider range than formerly thought, stretching from the Cape Flats along the coastal strip to at least the Knysna area. The second is the nominate race which has a less southerly disposition than the last and occupies most of the inland areas of the south-west Cape, eastwards to about the Albany district, also stretching in a broad belt up the west coast to Little Namaqualand. Both M.a. marjoriae and M.a. apiata are closely linked in many ways and form an assemblage rather distinct from the others. In their range they mainly occupy the winter rainfall areas, and are never found to any great distance from the coast. In size males vary mainly between 80 and 89 mm. in wing length, seldom exceeding 90 mm. and the breast spotting is dark and bold, often extending well over the throat and almost invariably covering the under tail- coverts. A large distributional gap also separates the bulk of this group from their nearest neighbours situated to the north-east. This next group also has its components closely allied, and constitutes the very rufous-backed populations occupying an extensive area stretching from the eastern Cape through the Orange Free State to the Transvaal, northern Cape and Griqualand West, also penetrating the southern parts of Botswana and continuing up into the more northern parts of South-West Africa. The Botswana and South-West African popu- lations are somewhat paler dorsally and must bear the name M.a. deserti (Roberts), while the darkest rufous populations occurring to the south and east are known as M.a. hewitti. Both hewitti and deserti differ from the winter rainfall populations of Clapper Larks in being con- siderably larger with but few males measuring less than 90 mm. in the wing. Ventrally the breast spotting has a different character, being lighter and more diffuse, not extending so far over the throat. The under tail-coverts are, almost without exception, unspotted. In range this group avoids the littoral and inhabits the more elevated, better grassed inland areas of southern Africa. The paler, grey-backed forms are divisible into three races with M.a. nata as the eastern extreme. The other two were described more recently, i.e., M.a. reynoldsi Benson & Irwin from southern Barotseland and M.a. jappi Traylor which was found to occur a little to the north in the Kalabo district. In South-West Africa these pallid forms inhabit only the more northerly reaches from the Etosha Pan northwards but extending in Botswana much farther south, especially in the central parts. They occupy various types of savannah and open grass- land. About Lake Makarikari, M.a. nata constitutes a curious ecological race restricted to the white calcareous ground found in this area. Although there is in the National Museum at Bulawayo a large number of these grey-backed Clapper Larks from the Makarikari area in Botswana, there are relatively few specimens from the rest of their somewhat extensive range through South-West Africa and western Zambia. In addition they emanate from only a few widely scattered and often restricted localities with the result that any systematic arrangement of the constituent races at this stage cannot be regarded with any finality. There is the added difficulty that both M.a. damarensis Sharpe and M.a. reynoldsi have been described from 27 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 3, DECEMBER 1970 28 QUICKELBERGE : VARIATIONAL PATTERNS AND RACES OF THE CLAPPER LARK specimens in worn plumage. In this state the vital clues to racial allocation are often missing or plumage colour has changed drastically. This is clearly evident in specimens of M.a. nata where birds in older dress are seen to be warmer or more sandy in appearance over the upper- parts, and with the neck regions tinged yellowish to a more marked extent. One worn speci- men of M.a. jappi was found to be inseparable from those of the reynoldsi series and one naturally wonders to what further extent fresh specimens of reynoldsi will be found to resemble jappi , seeing the series of the latter race consist mostly of birds in fresh dress. Similarly, it would be interesting to examine specimens in good plumage from the type-locality of dama- rensis since specimens from the nearby Etosha Pan area and similar birds from the Gemsbok Pan vicinity in Botswana are a motley lot and evidently result from a degree of intergradation with the contiguous M.a. deserti. Hardly two are alike with every variation from a “ reynoldsi - like” specimen to some strongly buff-tinged and with broad reddish sub-apical barring to the dorsal feathers, thus showing approaches to M.a. deserti. M.a. damarensis is, therefore, still largely an unknown quantity and if we are to judge from present material, rather un- satisfactory. Occupying only two limited areas, i.e. the vicinity of the Etosha Pan and a sixty-mile radius of the Gemsbok Pan, there is every appearance of having to deal with birds of hybrid origin. A typical specimen of deserti has in fact been collected within twenty-five miles of Gemsbok Pan. The diversity of these specimens could possibly indicate that the grey nata of this area (actually reynoldsi) and the reddish deserti have already achieved some measure of reproductive isolation, since hybrid populations showing greatly increased variabi- lity are said to be evidence of secondary intergradation, thus suggesting a breakdown of previous isolating mechanisms. A few of these odd-looking birds with pinky vinaceous upper- parts were given the racial name kalaharicus by Roberts and described from the Gemsbok Pan. The fact that deserti and outlying populations of nata (some better grouped with reynoldsi) occur in absolutely unchanged and typical form within 60 — 70 miles of each other in the country immediately south and south-west of the Makarikari pans, could mean that here reproductive isolation is virtually established. Certainly one would expect that to bridge the wide difference in appearance between the russet deserti and the ashy pale nata one should encounter a whole array of intermediate races, collectively forming a cline and occupying an extensive belt of territory. It appears best to omit damarensis from our list of the known valid races. When the present collecting lacunae have been filled it might be found that birds from the Ondangua area and other extreme northern parts of South-West Africa and Botswana are inseparable from reynoldsi , in which case damarensis would have to be reinstated and reynoldsi fall away. Another possibility is that such specimens could prove to be distinct from reynoldsi. A single specimen from Namutoni, near Etosha Pan is unusual in having a distinct pale yellowish tinge to the entire upper-side grey and lacking reddish barring. From a similar latitude in the north-western corner of Botswana there is a single specimen from the Tsodilo Hill which has a somewhat similar appearance, and as suggested, could indicate the possible existence of a distinct race in these northern parts which could utilize the name damarensis . Unfor- tunately the Tsodilo Hill bird is juvenile so comparisons are mere conjecture at this stage. Geographical variation within the grey-backed group of Clapper Larks affects the colouring of the sub-apical bar to the mantle and rump feathers, the basal portion of the crown feathers and the general tone of the upper-side plumage. Thus reynoldsi is usually ashy-grey over the upper-parts, with the feather sub-apices and crown feather bases light reddish brown. M.a. nata differs from reynoldsi in that the barring on the feathers is not red- dish, but instead very pale yellowish-brown, sometimes hardly noticeable against the general ashy-grey ground colour. There is also a greater suppression of reddish-brown over the crown feathers. Finally jappi is similar to reynoldsi but the upper-side grey is richer and darker. There is among all the grey-backed races of M. apiata, variation in the colour of the sub-apical 29 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 3, DECEMBER 1970 portion of the back feathers. In reynoldsi and jappi , where these parts are usually reddish- brown, a few individuals will lack this feature so that such specimens of reynoldsi become indistinguishable from nata. Thus it is that nata in its relatively pure form exists only in limited areas about the north- east corners of the Makarikari Pan complex. From this point there is a gradual shift towards the incorporation of reddish colours in the feathering of the upper-parts in successive popu- lations found to the south-west, west and north-west. Birds from Lake Dow and other western fringes of Makarikari clearly begin to show this. The nearest known north-western popula- tion to nata is in southern Barotseland where reynoldsi shows similar changes from typical nata as those displayed by Lake Dow birds. There is thus no alternative but to unite these two populations under the name reynoldsi. To dismiss western Makarikari birds as nata showing the influence of deserti would be no better than dismissing reynoldsi as nata showing the influence of jappi. Proceeding still farther westwards takes one closer to the Gemsbok Pan area, which as mentioned, is an area of intergradation with deserti. Unfortunately there are no specimens available from the 150 — 200 miles separating the Gemsbok area from Lake Dow, nor are there any specimens forthcoming from the large land mass separating topo- typical reynoldsi from the range of nata. Only when these gaps have been filled will the true extent of the ranges of reynoldsi and/or damarensis be known, and whether the latter is a valid race. The range pattern of the Clapper Lark is interesting. Although occurring densely and evenly enough in the areas it does inhabit, it is strikingly apparent that large tracts of country are completely avoided, as is clearly shown in the distribution map. The Clapper appears to avoid both the very dry and wet parts of southern Africa, thus restricting itself largely to the central regions, though avoiding the bulk of the Great Karoo. It is also absent from Bushmanland and Great Namaqualand in the west. Those parts which are inhabited fall mainly within the 16- to 32-inch isohyets. It is only the winter rainfall group of populations that would appear to tolerate greater extremes, but even here it is clear that it becomes very sparse towards the northern reaches of Little Namaqualand; in fact it seems that the species has disappeared from these parts, as much recent collecting has failed to reveal its presence there. In the north-eastern sectors of South-West Africa, the Caprivi Strip, and the northern and eastern lowlands of the Transvaal, the Clapper Lark is largely replaced by the Flappet Lark Mirafra rufocinnamomea ; in fact there would not appear to be any areas where the two occur sympatrically. More recently, with M. A. Traylor’s description of M.a. jappi (1962), it would appear that on the Luiwa Plain in Barotseland this race of the Clapper does occur with the Flappet Lark, although segregated ecologically to a certain extent. These two larks are also very closely related, virtually sibling species, and appear to vary geographically in exactly the same way where their ranges approach one another. Thus the race of the Flappet Lark entering the north-eastern parts of South-West Africa resembles M.a. reynoldsi , while in the Transvaal M. rufocinnamomea is as richly rufous as M.a. hewitti. Were it not for the differences in the flight and tail-feather patterns, little would remain to separate them as good species on morphological grounds. NOTE The hind neck and upper mantle shows up as a collar in the Clapper Lark because of the extreme reduction of the black and russet dorsal patterning, revealing a ground colour of clear light grey or buff. Frequently these neck feathers stand out in contrast to the more russet and black-flecked crown and mantle. For these reasons the crown and nape coloration may or may not show up as a distinct cap depending on the extent of contrast provided by the neck-band. Thus the “capped” effect often referred to in the literature is of no value as a character in racial determination. 30 QUICKELBERGE: VARIATIONAL PATTERNS AND RACES OF THE CLAPPER LARK The proximal portion of this cap, or nuchal collar, has also been used as a diagnostic character, distinguishing the nominate race from other contiguous races. This nuchal collar is said to be rufous, flecked with black (vide: Ostrich 27: 156 — 7). However, it was found that in all specimens of the nominate race, including M.a. adendorffi , there was every variation not only of this nuchal collar (when visible as a collar), but also of the entire cap and mantle etc. Such variation invariably affects the entire dorsal surface, so that the more black flecking and barring there is over the mantle the more will also normally occur over the crown, nape, etc. Examples of such variation in the cap in specimens of the nominate race (as previously understood) include those in which only the crown centre is black-flecked surrounded by russet, and others where the cap is evidently russet and black-flecked except the nuchal region which is clear russet. Others again have the entire region of the cap russet and black-flecked. It actually emerges that most of the specimens which have been attributed to the nominate race are without a clear-cut rufous black-flecked collar, whereas more than half the specimens attributable to M.a. adendorffi do have such a nuchal collar. From this it can be seen that the russet half-cap which has been said to distinguish M.a. adendorffi must also fall away as a taxonomic character. Certainly specimens of M.a. marjoriae from Riviersonderend through Swellendam, Riversdale, Mossel Bay and up to Knysna (formerly classified as M.a. algoensis ) could be separated from the nominate race by a less rufous nuchal collar, but the rest of the dorsal surface is also less rufous (see description). Furthermore, no substantial differences exist in the extent of russet over the cap between topotypes of M. apiata apiata and birds from Little Namaqualand, i.e., within the range of M.a. adendorffi. In fact a comparison between topotypes of the nominate race and virtual topotypes of M.a. adendorffi (i.e., from Van Rhynsdorp and vicinity) actually reveal that the latter are mostly more heavily black-flecked over the crown and nape than most topotypical specimens of the nominate race. Then, too, it is significant that the reddest bird with the least amount of black dorsal markings over crown, nape, mantle etc., is actually a topotype of the nominate race. It is seen, therefore, that individual variation is so rampant that any attempts to reflect the meagre, incipient geographical variational trends that possibly exist, within the concept of the sub-species would lead to more confusion than clearer understanding. This, at any rate, is seen to be the position of our present state of knowledge dependent as it is on the current bird skin series at our disposal. Another character which has been used by workers but which will not be taken into account here is the amount of russet colouring in the flight-feathers. This was found to be just another highly variable character, but can be said to vary roughly in direct relation to the general dorsal colour facies in that the greater the extent of rufous development over the upper-parts, including also scapulars, coverts and secondaries, the more extensively will this colour invade the vanes of the primaries. It is thus sufficient only to take note of the degree of rufous development over the dorsal surface. This trend is only fairly general and exceptions occur. For example it is interesting to note that in some specimens from Malmesbury and even Van Rhynsdorp and Namaqualand, which show decidedly more rufous above than do specimens of M.a. marjoriae, the flight-feathers are as sparingly or even less tinged with russet than those from the Cape Flats and nearer the Peninsula. As will be apparent from these deliberations, M.a. adendorffi and M.a. algoensis cannot be maintained. Most of the specimens identified by authors as M.a. algoensis have been here reclassified as M.a. marjoriae, i.e., those from Riviersonderend to Knysna. This leaves only a very few unaccounted for, namely the specimens from Port Elizabeth and Grahamstown. As these cannot be separated satisfactorily from the nominate race the two forms must be united under the latter taxon. This arrangement gives the appearance of M.a. marjoriae being interposed between west and east Cape populations of M.a. apiata. This, however, is not strictly true as M.a. marjoriae is really confined to the extreme southern limits of the 31 ANN, CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 3, DECEMBER 1970 southern and south-west Cape with the nominate race occurring to many points north and east of this race’s range. It is, in fact, only between the 21 and 23 degree lines of longitude that the nominate race has, as yet, not been found to occur north of the range of M.a. mar- joriae and which gives the appearance of this race falling between the mentioned east and west populations of M.a. apiata. However, it seems likely that specimens connecting up these apparently sundered populations will be found to occur locally along points in the Little Karoo. There would appear to be no other route by which M.a. apiata could have spread to Port Elizabeth and Grahamstown or vice versa , considering that the coastal forests east of Knysna would appear to be ecologically unsuitable to the Clapper Lark and because the Great Karoo likewise appears to be largely avoided. For reasons stated, it is unlikely that these east Cape birds could be treated as inter- grades between M.a. marjoriae and M.a. hewitti, but this possibility cannot be ruled out entirely until more specimens are available. Another somewhat remoter possibility is that more specimens from the Port Elizabeth and Grahamstown areas will show these as being only the eastern terminal populations of M.a. marjoriae. In the light of the foregoing discussion, geographical variation in Mirafra apiata may be interpreted by the recognition of the following racial taxa: (a) Mirafra apiata marjoriae Winterbottom Mirafra apiata marjoriae Winterbottom, 1956. Ostrich , 27(4): 156. Zoetendalsvlei, Bredasdorp district, south-western Cape. Description: Sexes alike in plumage pattern and coloration. Over the upper-parts the feathers are blackish fringed with grey, coloured apically with rather narrow, dark reddish-brown bars or blotches imparting to the whole upper-surface a dark mottled appearance with the minimum of russet intrusion. What there is of this russet component of the dorsal colours may even be further reduced by plumage abrasion, so that birds collected towards mid and late summer are almost completely without any such reddish-brown markings. Ventrally this race is as variable as the other races, especially as regards the degree of vinaceous-brown or buff generally colouring these parts. This may vary from a dark brown or chestnut to a rather pale fawn, although most tend to the darker extreme. The throat is i considerably paler. The dark brown breast spotting is conspicuous and bold, often covering most, if not all, of the throat and also not infrequently extending along the flanks. With but few exceptions similar spots or streaks also adorn the under-tail coverts. Measurements Males: Females: wing 77 •5- -88-5 (83-4) mm. tail 54' ■0- -60*0 (57-4) 55 bill 16' 4- -18-5 (17-4) 55 wing 79- 0- -84-0 (81-5) 55 tail 54- 0 55 bill 16- 6- —17 * 0 (16-8) 55 14 specimens 5 13 2 55 55 55 1 specimen 2 specimens Material: Twenty specimens, comprising 7 from near the Peninsula and adjacent areas, 5 v from Zoetendalsvlei, 6 from Riviersonderend to Mossel Bay and 2 from Knysna. Range: Occupies only the extreme southern limits of the African continent, not being found much farther north of the 34th parallel and up to about Knysna, southern Cape. Remarks: Although most of the material of this sub-species is in a rather worn condition, it seems reasonably clear that this entity constitutes a valid race. Further material in fresh ij plumage is needed, however, to define its range and characters more accurately. Northwards, M.a. marjoriae grades rather gradually into M.a. apiata. I U 32 QUICKELBERGE: VARTATIONAL PATTERNS AND RACES OF THE CLAPPER LARK (b) Mira fra apiata apiata (Vieillot) Alauda apiata Vieillot , 1816. Nouv. Diet. d'Hist. Nat., 1:342. Swartland — Malmesbury distr. south-western Cape Province (ex Levaillant). Mirafra apiata adendorffi Roberts, 1919. Ann. Transv. Mus., 6(3) : 1 1 7. Olifants R., at Klaver, western Cape Province. Megalophonus apiatus a/goensis Roberts, 1926. Ann. Trans. Mus., 11(4) :222. Port Eliza- beth, east Cape Province. Description: Similar in most respects to M.a. marjoriae, differing mainly in the extent of the dorsal rufous, which in M.a. apiata has increased in area, making it almost the dominant dorsal colour. It also displays a brighter or clearer reddish hue, less drab and brownish tinged, as in the preceding race. The crown and nape are also largely russet, often invaded medially, laterally or completely by black streaks and bars. Averaging larger in size. Measurements Males : wing 75 • 5 — 93 -0 (86-6) mm. 30 specimens tail 52-7— 64-3 (59-2) „ 14 bill 16-7— 19-5 (18-1) „ 27 Females: wing 78 -5 — 82-0 (79-9) ,, 6 ,, tail 55-0 ,, bill 16-2— 18-0 (17-4) „ 6 Material: Thirty-eight specimens: 2 from the eastern Cape, 1 from Swellendam, 20 from the western Cape, 9 from the borders of southern Little Namaqualand and the Karoo and 6 from Little Namaqualand. Range: Occurs mainly north of the 24th parallel in the western Cape, extending up the western seaboard in a fairly broad belt, but apparently not penetrating the Great Karoo or southern South-West Africa. Reaches the eastern Cape at Port Elizabeth and Grahamstown. Remarks: Although as yet no specimens of this race have been found in the intervening terri- tory between the western and eastern Cape, the kinship of the eastern birds with specimens of M.a. apiata is quite clear. There is for example no indication that the birds from Port Elizabeth and Grahamstown are more russet above when compared with Malmesbury specimens; indeed a Port Elizabeth bird is actually darker above than certain such specimens from within the district of the type-locality of M.a. apiata. Nor can the birds from Riversonderend to Knysna be grouped with the Port Elizabeth and Grahamstown birds. These birds from Knysna, etc., definitely show less russet above and are inseparable from topotypes of M.a. marjoriae. As yet there does not appear to be any sign of the usual racial intergradation between the most easterly members of M.a. apiata and the most southerly representations of M.a. hewitti , in spite of their ranges approaching each other rather closely just at this point. (c) Mirafra apiata hewitti (Roberts) Megalophonus hewitti Roberts, 1926. Ann. Transv. Mus., 1 1(4) :223. Rooiberg, Transvaal. Description: Compared to the previous race there is a marked increase in the overall size in specimens of M.a. hewitti. Although spotting of the under tail-coverts is a conspicuous feature of the preceding races, it is only very rarely encountered in the present race. Also the breast spotting is less extensive, covering no more than the lower throat. These spots are also clouded and paler than in races (a) and (b). Dorsal rufous reaches peak development, and although there is as a rule still a variable, albeit reduced, amount of black barring over the upper parts, some specimens would be immaculate reddish-brown were it not for the whitish 33 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 3, DECEMBER 1970 terminal fringes to dorsal feathers in fresh moulted condition which is, incidentally, typical of all the races. On the other hand, a few individuals of hewitti are quite extensively black- barred, and although they show an approach to the nominate race in this respect, the con- spicuous lateral grey fringing to the dorsal feathers of the latter race at once distinguishes it from hewitti which is by comparison warmer and more buffy about these parts. Specimens of M.a. hewitti showing a comparatively well-developed black-barred back emanate from the eastern Cape, the Kuruman-Vryburg area and Pretoria and in the series from these parts constitute very roughly about 50 per cent of individuals. From such in- between areas as the Griquatown district, Kimberley and the western Orange Free State there appears to be a greater incidence of clearer rufous-backed birds. Measurements Males: wing 88-0 — 97-5 (92-3) mm. 50 specimens tail 58-8— 68-6 (63-4) „ 29 bill 16-5 — 20-5 (18*6) „ 45 Females: wing 79 -5 — 87-5 (83-7) ,, 5 ,, tail 54-6—60-8 (57-7) „ 2 bill 15-7—18-6(17-1) „ 6 Material: Sixty-three specimens: 5 from the eastern Cape, 9 from the Orange Free State, 30 from the northern Cape, 16 from the Transvaal, 2 from Lesotho, and 1 from Botswana. Range: Central and north-eastern Cape Province through most of the Orange Free State and adjacent edges of Lesotho to the southern, western and central parts of the Transvaal; also over parts of the northern Cape and just penetrating the south-western sector of Botswana. Remarks: Apart from some dorsal variation this is a well-marked and compact sub-species, typical of the highveld but penetrating other various bush-types of vegetation about the edges of its range. It appears to merge with only one other race, i.e., M.a. deserti along the north- western limits of its distribution. (d) Mirafra apiata deserti (Roberts) Megalophonus hewitti deserti Roberts, 1926. Atm. Transv. Mas., 11(4) :223. Omutako ) Flats, north of Okahandja, Damaraland, South-West Africa. Description: Similar to M.a. hewitti differing only in being generally paler above and below. There is some variation in the shade of the dorsal rufous in that certain specimens are a clearer i brick-red, while others are more vinaceous-tinged but hewitti also varies in the tone of these ; parts. Measurements Males: wing 88 -5 — 95-0 (92-4) mm. 12 specimens tail 62-3—65-5 (63-5) „ 3 bill 17 -3 — 20 -4 (18 -86) „ 12 Females: wing 77-0 — 86-5 (81-5) ,, 7 tail 51-6— 58-6 (55-0) „ 5 bill 17-9 — 18-2 (18-05) ,, 4 Material: Thirty specimens: 7 from the more northern parts of South-West Africa (Quick- born, Gobabis, Omatako Flats etc.) and the rest are from various parts of Botswana south of parallel 21° S. 34 QU1CKELBERGE: VARIATIONAL PATTERNS AND RACES OF THE CLAPPER LARK Range: M.a. deserti has a rather attenuated range, stretching from Botswana to South-West Africa in a north-westerly direction, with another thin arm extending northwards towards the eastern parts of Botswana, approaching hereabouts the range of the very much paler M.a. nata. Remarks: There is a rather gradual paling and thus merging of deserti with M.a. reynoldsi in Botswana along a line from the Murumusa Pan to the Gemsbok Pan in a north-easterly direction. However, M.a. deserti from the Lothlekane/Tsepe area, although not occurring far from the very ashy-grey birds of Lake Dow and Makarikari, shows no signs of becoming paler. It is possible that in these parts M.a. nata and M.a. deserti restrict themselves severely to the colour of the sub-stratum which matches their dorsal colour. On a priori grounds these soil colours must apparently also be rather sharply defined within this area. It is also possible that an already advanced state of reproductive isolation between the red and grey-backed forms of this region could be the main reason for this apparent lack of intergradation. There is confusion in the names applicable to races in Botswana, and the bird figured in Smithers’ check list (1964) as M.a. kalaharica is much too rufous for this race if compared I with topotypes and Roberts’ original description. The specimen figured does, however, match perfectly specimens from South-West Africa applicable to M.a. deserti. (e) Mirafra apiata reynoldsi Benson & Irwin Mirafra apiata reynoldsi Benson & Irwin, 1965. Arnoldia, 1(37) : 1 — 3. Near Nasionga, southern Barotseland. Mirafra damarensis Sharpe, 1874. Proc. Zool. Soc. London , 1874:650, pi. 75, fig. 2. Ondongua, Ovamboland, northern South-West Africa. (Possibly M.a. deserti inter- grading with M.a. reynoldsi). Corytpha [s/c!] hewitti kalaharica Roberts, 1932. Ann. Transv. Mas ., 35(1) :27. Gemsbok Pan, Ghanzi District, Botswana. (M.a. deserti intergrading with M.a. reynoldsi). Description: Ashy-grey over the upper-parts, sullied to a variable extent with pale sandy- yellowish, especially about the regions of the neck and upper mantle. Sub-apices to mantle feathers tinged variably with reddish-brown, some only faintly while others resemble nata in this respect. Specimens in fresher plumage would, no doubt, be purer grey dorsally. Basal portion of crown feathers rufous, the amount varying individually. With regard to underside coloration, there is a wide range of individual variation, most specimens being light creamy- white, unevenly tinged with buff-brown, some darker and more extensively brown-tinged than others. The dusky breast spotting tends to be reduced by wear. Measurements Males: wing 84 • 0 — 88 • 5 (85 • 8) mm. 8 specimens bill 17-0— 19-1 (18-1) „ 6 Females: wing 77-5 — 83-5 (80-5) „ 2 bill 16-4— 16-8 (16-6) „ 2 Material: Sixteen specimens: 9 from southern Barotseland (including 3 juveniles), and the rest from Botswana, i.e., from the Gemsbok Pan in the west to Lake Dow and Makarikari towards the east. Range: Described from a small area and limited localities about the south-eastern corners of Zambia between the Zambezi and Kwando rivers. The similar specimens from Botswana, i.e., the Gemsbok Pan, Lake Dow and Makarikari probably indicate a linking-up with the Zambian population in the intervening country where ecologically suitable to this lark. 35 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 3, DECEMBER 1970 (/) Mirafra apiata nata Smithers Mirafra damarensis nata Smithers, 1955. Bull. Brit. Orn. Club, 75(3) :29. 10 miles west of Nata, north-eastern Botswana. Description: Similar to reynoldsi but with the sub-apices of the mantle feathering much paler, having lost the reddish tones and being instead greyish to very pale yellowish-brown, a colour not easy to describe and not contrasting much with the grey ground colour. Oddly, the red- dish colouring to the crown feathers is retained to a variable extent in many specimens with only a few having lost it entirely. It is, however, more suppressed than in reynoldsi. M.a. nata in fresh dress is of course much purer ash-grey dorsally than those in worn dress, and the neck and upper mantle more delicately suffused with pinkish instead of yellowish. Under- side similar to reynoldsi. Measurements Males : wing 8 1 • 0 — 90 -5 (87-2) mm. 9 specimens tail 57-2— 62-3 (59-1) „ 11 bill 17-6 — 19- 1 (18-2) „ 11 Females: wing 77-0 — 82-5 (79-7) ,, 2 ,, tail 49-5— 55-5 (52-5) „ 2 bill 17-4— 17-9 (17-6) „ 2 Material: Some sixty specimens, the majority of which being in worn plumage render racial allocation uncertain. Range: The vicinity of the Makarikari Pan complex in north-eastern Botswana, presumably concentrated mainly about its north-eastern edges. Remarks: This form’s limited range and extremely pallid dorsal appearance gives the im- pression of this being an example of a substrate race. (g) Mirafra apiata jappi Traylor Mirafra apiata jappi Traylor, 1962. Fie/diana Zoology, 44:113 — 115. Luiwa Plain, Kalabo ) district, Zambia. Description: Similar to M.a. reynoldsi but darker over upper-parts with the reddish barring \ to feather sub-apices a trifle darker, more brownish. Underside more brownish suffused in i one or two specimens. Measurements Males: Females: wing 88 • 0 — 89 • 0 (88 • 5) mm tail 55-7 ,, bill 17-0— 19-0 (18-2) „ wing 76 -5— 78-0 (77-2) „ tail 52*7 ,, bill 16-3— 16-9 (16-6) „ 3 specimens 1 specimen 5 specimens 2 1 specimen 2 specimens Material: Seven specimens from a small area north-west of Kalabo and from the vicinity of the Luanginga River, Zambia. Range: As under Material. Remarks: This interesting race appears quite distinctive but its discreteness from reynoldsi can only be properly assessed when specimens of the latter in fresh plumage are available. 36 QUICKELBERGE: VARIATIONAL PATTERNS AND RACES OF THE CLAPPER LARK ACKNOWLEDGMENTS I am grateful to the Directors of the following institutions for the loan of specimens: Transvaal Museum (through Mr O. P. M. Prozesky), Durban Museum, South African Museum (through Prof. J. M. Winterbottom), National Museum of Rhodesia, Bulawayo (through Mr M. P. Stuart Irwin) and the National Museums of Zambia, Livingstone (through Mr C. Cross). For criticism of the manuscript, I wish to thank Mr P. A. Clancey, Director of the Durban Museum. LITERATURE CITED Clancey, P. A. 1965. “Subspeciation in the southern African populations of the Sabota Lark Mirafra sabota Smith”, Ostrich , 37(4) :207— 13. Clancey, P. A., 1966. “A subspecific arrangement of the austral populations of the Fawn-coloured Lark Mirafra africanoides Smith”, Arnoldia, 2(2): l — 8. Smithers, R. H. N., 1964. A Check List of the Birds of the Bechuanaland Protectorate. Trustees of the National Museums of Southern Rhodesia. Traylor, M. A., 1948. “New birds from Barotseland”, Fieldiana Zoology , 44:113 — 15. 37 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mas. (Nat. Hist.) VOLUME 8 * PART 4 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Taxonomic notes on South African marine moliusca, I R. N. KILBURN *East London Museum The work of the late Dr K. H. Barnard has provided a firm basis for research into the taxonomy of South Africa marine molluscs. There remains, however, a vast number of species whose status, relationships and nomenclature are in urgent need of revision. The present paper is the first in a series dealing with some of these problematic species. Fam. Turridae Drillia (Cerodrillia) burnupi (Sowerby) (Fig. 1) Pleurotoma burnupi Sowerby, 1897:3, pi. 8, figs. 1 — 2. Drillia burnupi; Turton, 1932:22. Radula prototypic (formula 1-1-R-l-l); rhachidian small, with a single weak cusp; lateral plates strongly curved, with 13 sharp cusps, inner ones strong and claw-like, becoming weaker laterally and obsolete on outer side of plate; marginals long, slender and pointed; 32 rows. Resembles the radula of Drillia ( Cerodrillia ) thea Dali (Powell, 1966:74, fig. D90). Operculum leaf-shaped, rather thick, nucleus apical. Protoconch corroded in all specimens examined, apparently two smooth whorls, 0-8 x 0-73 mm. in dimensions. Distribution: Durban (type locality) and Umhlali (Nat. Mus.); Bazaruto Island, Mozam- bique (Nat. Mus., leg. Mrs K. Eastwood); Embotyi, Pondoland (E.L. Mus., leg. R.K., living). Also Port Alfred (Turton; requires confirmation). * Present address: Natal Museum, Pietermaritzburg. 39 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 4, DECEMBER 1970 This species was synonymized with “ Drillia” hottentota (Smith) by Barnard (1958: 120), but the radula of the latter (see below) is very different. Conchologically, D. burnupi differs in being narrower, with the axial ribs reaching the suture but becoming abruptly obsolete below the periphery on the body whorl. Furthermore, the periphery of each whorl is situated towards its base, instead of in the middle as in hottentota , and the labral sinus is deeper and narrower. Both Drillia burnupi and D. falsa Barnard (1958:94) appear to be referable to the sub- genus Cerodrillia Bartsch & Rehder, 1939, which was accorded full generic rank by Powell (1966:73). Crassispira hottentota (E. A. Smith) comb. nov. (Figs. 4, 5) Pleurotoma ( Clavus ) hottentota E. A. Smith, 1882: 208. Clavatula hottentota; Barnard, 1958: 101. Drillia hottentota; Barnard, 1958: 120 (synonymy) Radula formula 1 -0-0-0- 1. Marginal teeth long and sharply pointed, each with a well- developed basal limb; no rhachidian, but basement membrane with a series of delicate, deltoid-shaped thickenings, arranged in pairs, one on either side of the midline, and simu- lating degenerate lateral plates. They are here regarded as homologous to the basal plate reported by Powell (1966:55) as supporting the rhachidian in Clionella. These thickenings were in fact mistaken by Barnard (1958: 101) for a central tooth (rhachidian), but are shown by suitable staining to be two rows of discrete structures, without cusps or cutting edges, and so diaphanous as to be probably functionless. Operculum leaf-shaped, with apical nucleus; inner surface with thickened margin on columella side. Penis long, slender and blunt, sperm duct slightly coiled, ending in a small swollen bulb, , whose opening projects slightly as a terminal papilla. The present species is here referred to Crassispira Swainson, 1840, on the grounds of a i resemblance in both shell form, and structure of the marginal radula teeth ( vide Powell, op. cit.: figs. CTO — 74, D100, 103). Although C. hottentota appears to be atypical in the absence of a subsutural cord, and in the presence of the basal plates described above, these characters ;■ are scarcely of generic status. I am indebted to Mrs. C. M. Connolly for making available dried material of this ; species. Fam. Nassariidae Bullia ancillaeformis E. A Smith (Fig. 2) Bullia ancillaeformis Smith, 1906:37, pi. 7, fig. 8. ? Ancilla bulloides (sic., non bullioides Reeve); Turton, 1932:32. Although this species is by no means rare in dead but fresh condition on the Natal south coast, it has never been reported since its original description, and Barnard (1959:61, 123) suggested that it was really an Ancilla. The absence of a basal groove shows that view to be incorrect, and the species appears to be a typical Bullia. A detailed description and figure is here given. Spire 1 to l^x length of aperture; sides of spire almost straight, each whorl being only slightly convex; aperture subtriangular in outline, with greatest width at base. Surface smooth, except for faint growth lines and very obscure traces of 3 — 4 spiral sulci on base (Smith: “obsolete spiraliter striata”). Parietal callus a thin glaze, covering rostrum and most of basal part of adapertural surface of body whorl, and extending up around spire, where it covers ' KILBURN : TAXONOMIC NOTES ON SOUTH AFRICAN MARINE MOLLUSC A, I the basal half (sometimes f) of each whorl; this callus first appears at end of second whorl. Colour milk-white, sometimes with a pale orange-brown band below suture, callus opaque white. Total number of whorls 6; limits of protoconch not defined, obtusely domed, diameter 1 -1 — 1 *5 mm. Adult size 19 X 7-5 mm. (Smith); 20 X 9*6 mm. (apex broken). Fig. 2. Btillia ancillaeformis (E. A. Smith). Type locality Port Shepstone (also Nat. Mus.); Mzamba, Pondoland (E.L. Mus.); Jeffreys Bay (in coll. Mrs C. M. Connolly). It is probable that Turton’s record of the deep-water And //a bullioides Reeve from Port Alfred is based on Bullia andliaeformis. There is, as pointed out by Barnard, a super- ficial resemblance between the two. Farm Fasciolariidae Latirus fihncrae (Sower by) comb, now 1 Euthria filmerae Sowerby, 1900: 1, pi. 1, fig. 3; Barnard, 1959: 172, Although the radula of this species is unfortunately still unknown, the shell has few features in common with members of the buccinid genus Euthria. The general form is that of a Latirus (probably subgenus Latirulus Cossman, 1889), and confirmation is provided by the presence of three small, but distinct transverse folds on the base of the columella. These are only visible when the aperture is viewed from an angle, and were apparently overlooked by Sowerby and Barnard. Dead specimens have been examined from numerous localities between Bulugha (about 12 miles north-east of East London) and Palm Beach (Natal south coast). Fam. Buccinidae Cantharus subcostatus (Krauss) comb. nov. (Fig. 7) 1 Buccinum rubiginosum var. sub cost at a Krauss, 1848: 120. Pollia subcostata Tomlin, 1948:355 (synonymy) Cantharus carinifera (Kiister); Barnard, 1959: 150; Barnard, 1969:629. The radula of this species is very similar to that of C. undosus (Linn.) and C. fumosus (Dillwyn), as figured by Robertson (1957: figs. 16, 17). Common in Natal under rocks in sheltered low-tide pools. Occurs living on the Pondo- land/Transkei coast at Umzikaba (Barnard, 1959), and Embotyi, LJmgazi and Xora River mouth (E.L. Mus., leg. R.K.). 41 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 4, DECEMBER 1970 Fam. Columbellidae Pyrene (Mitrella) natalensis (Tomlin) comb. nov. (Fig. 8) Mitrella natalensis Tomlin, 1926:291, pi. 16, fig. 5. Columbella natalensis; Turton, 1932:71, pi. 17, fig. 512. Although this species was treated as a synonym of P. albuginosa (Reeve) by Barnard (1959: 173), the material on hand serves to confirm its validity. In both species the rhachidian is almost square, very delicate and diaphanous, but the lateral teeth differ markedly in shape, those of P. natalensis being proportionately much broader, with two small but distinct denticles between the proximal cusp and the distal ones. The protoconch, consisting of two wine-red coloured whorls, is much smaller than in P. albuginosa , and measures only 0*32 x 0-3 mm., as opposed to 0*75 x 0*5 mm. The colour pattern (Tomlin, loc. cit.) on the six teleoconch whorls is relatively constant, and has not been observed in P. albuginosa. A further distin- guishing feature is the total absence of a callus deposit on the columella of P. natalensis , even in fully grown specimens. Finally, it is a much smaller species, e.g. 7x2*3 mm. (Tom- lin), 6*6 X 2*6 mm. (paratype, Nat. Mus.). Distribution: Port Shepstone, Scottburgh, Durban (Tomlin, Nat. Mus.); Ballito Bay (Umhlali area, north coast of Natal) and Port Alfred (Nat. Mus., E.L. Mus., leg. R.K.). Five paratypes from Port Shepstone are in the Nat. Mus. collection (No. 3734; T488). P. natalensis lives chiefly among coralline seaweeds, generally associated with Pyrene kraussi (Sow.), P. burnupi (Smith) and Tricolia capensis (Dunker); also occurs occasionally on the undersides of submerged rocks along the infratidal fringe. Pyrene floccata (Reeve) (Figs. 3, 9) Columbella floecata Reeve, 1859: sp. 160; Sowerby, 1892:22; Turton, 1932:72; Barnard, , 1959:174. This species, which does not appear to have ever been adequately described, is very distinct t from P. albuginosa. Protoconch somewhat mamillate, \\ whorls, 0*71 x 0*74 mm., smooth. Teleoconch i whorls 6, smooth except for growth lines and 8 — 9 weak spiral lirae on base and rostrum. Columella callus ending in a distinct terminal pleat, rostrum occasionally bearing a few feeble : granules; outer margin of callus free basally. Outer lip smooth or plicate within (see below). . Dimensions: 16*9 x 7 mm.; 9*2 x 4*3 mm. Operculum pyriform in outline, nucleus apical. Radula: Rhachidian plate delicate, semilunate in shape; lateral plate with a well-developed 1 basal limb and a flange-shaped proximal cusp, apex bifalcate; about 150 — 160 rows. In struc- KILBURN: TAXONOMIC NOTES ON SOUTH AFRICAN MARINE MOLLUSCA, I Figs. 4-12. Fig. 4. Crassispira hottentotci (E. A. Smith): radula. 5. C. hottentotci: penis, sperm duct shown by transparency. 6. Pteropurpura incurvispina, nom. nov. : radula. 7. Cantharus subcostatus (Krauss): radula. 8. Pyrene natalensis (Tomlin): radula. 9. Pyrene floccata (Reeve): radula. 10. Cinysca granulosa (Krauss): rhachidian plate. 11. Natica forata Reeve: rhachidian and lateral plates. 12. Cymatium durbanense (E. A. Smith): rhachidian and lateral plates. 43 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 4, DECEMBER 1970 ture the lateral plates closely resemble those of NitideUa nitida (Lamarck) (vide Troschel, 1869, pi. 9, fig. 4). The present species will probably prove to be congeneric, but is best retained in Pyrene (s.l.) until the classification of the Columbellidae has been finalized. Egg-capsules: Numerous individuals were found spawning at Embotyi (Pondoland) towards the beginning of August 1968. The spawn was deposited in communal sheets, some- times consisting of several hundred capsules, on the undersides of rocks. The capsules are almost colourless and shaped like short cylinders, the average size being 1*3 X 2-5 mm.; each is joined to its neighbours by a thin basal expansion. In structure each capsule consists of an inner oothecum, surrounded by a protective outer wall; the dorsal surface of the oothecum bears a thin spongy layer, and its periphery forms an eave-like collar, which barely touches the edge of the outer wall, and which bears several thin buttresses on its inner surface. Some of the older capsules contained 1 — 3 well-developed veliconchas. In structure these capsules are basically similar to those figured by Amio (1955:235, fig. 3) as “ Columbella (? versicolor Sow.)”, but have a more extensive outer wall and collar, and the dorsal surface is spongy instead of setose. Range: Port Alfred to Durban; type locality East London. Sowerby’s 1892 record from Port Elizabeth needs confirmation. P. floccata shows a certain amount of geographic variation. In shells from the eastern Cape and Transkei regions the outer lip is smooth inside. From the Coffee Bay area north- wards, however, there is a tendency towards the development of 8 — 10 labral plicae. Although the typical colour pattern (as described by Barnard, 1959: 174) occurs throughout the range, , correlated with the development of the denticulate form is the occasional appearance of pink, orange or crimson individuals, while others are strikingly marked with chestnut-brown. A pinkish-orange peripheral band does occasionally occur in East London shells, but appears to be a product of beach-wear. However, as there is no clear geographic separation of the j two forms, it does not appear advisable to grant subspecific status to the eastern population. Earn. Muricidae Pteropurpura (Poropteron) incurvispina nom. nov. (Fig. 6) Murex mitraeformis (non Brocchi, 1814) Sowerby, 1841: fig. 75; Barnard, 1959: 200, fig. 42b. j Pteropurpura ( Poropteron ) mitraeformis; Vokes, 1964: 27; Barnard, 1969: 638, fig. 19f (in- correctly listed under P. uncinarius (Lam). Radula: Rhachidian with prominent median cusp and two major and 2 — 3 minor cusps :) on each side; lateral plate slender, falcate; at least 214 rows of teeth. Somewhat similar toq that of Ocinebra erinaeeus (Linn.) (vide Troschel, 1869, pi. 11, figs. 11 — 42). The figure given : by Barnard (1969) for P. incurvispina evidently shows a very worn rhachidian tooth, as the r whole median cusp is omitted. Operculum : Ovate, with nucleus near anterior end of outer margin, growth lines distinct,! sometimes rather lamellose. Vokes (1964) transferred this species and the allied uncinarius Lam. (type species of) Poropteron Jousseaume, 1880) to the genus Pteropurpura Jousseaume, 1880, on conchologicali grounds. The presence of an ocinebrid radula and operculum provides support for this action, i; Vokes, however, and apparently Cernohorsky (1967: 127), have incorrectly synonymized P. incurvispina with P. uncinarius. Some of the characters separating the two have been pointed out by Barnard (1959), but in addition to these, P. incurvispina differs in the align- 1. ment of the main spines, which are strongly recurved from their bases, while in P. uncinarius they stand out almost at right angles to the main axis, with only the distal half recurved. Furthermore, in the latter species the minor spines, of which there are four on the body whorl, 44 KILBURN: TAXONOMIC NOTES ON SOUTH AFRICAN MARINE MOLLUSCA, I are short, erect and more or less foliated, but in P. incurvispina they are simple, distinctly incurved towards the aperture and number five or six. P. incurvispina is fairly common in the East London — Umhlali area, living on the under- sides of rocks in lower balanoid zone pools. No specimens have been seen from west of Port Alfred, although Barnard records it from Still Bay. P. uncinarius , on the other hand, seems to live entirely infratidally, and has not been examined by the author from east of Port Alfred. Fam. Cymatiidae Cymatium (Monoplex) parthenopeum (von Sails) Cymatium parthenopeum; Clench & Turner, 1957: 228, pi. 110, fig. 4, pi. 112, figs. 7 — 8, pi. 113, figs. 9 — 10, pi. 128, figs. 1 — 3 (synonymy). Cymatium olearium {non Murex olearium Linn.); Barnard, 1963a: 26, fig. 3f. After comparing specimens from widely separated localities, Clench & Turner concluded that a single almost cosmopolitan species was involved. Comparison of South African shells with examples from Brazil, the West Indies, Japan, Australia and New Zealand certainly seems to confirm this view. It should, however, be noted that radulae of examples from Durban and Embotyi agree more closely with Powell's figure (1933: 169, fig. 7) of a New Zealand specimen, than with those of examples from Japan and Brazil (Clench & Turner, pi. 1 1 3, figs. 9 and 10). C. parthenopeum is common in Natal, and fresh shells are not infrequently found as far as Jeffreys Bay. In addition, the East London Museum has a specimen taken alive at Pletten- berg Bay (Mrs H. Boswell). The species lives shiefly in sheltered, sometimes muddy, crevices around the low-tide mark, and may also be found almost completely buried in sand among rocks. Cymatium (Cabestana) durbanense (E. A. Smith) (Fig. 12) Lotorium durbanense E. A. Smith, 1899: 248, pi. 5, fig. 4. Cymatium durbanense ; Barnard, 1963a: 27, fig. 3e. Barnard’s suggestion that this might prove to be a dwarf form of C. parthenopeum cannot be maintained. Apart from differences in size and in dentition (see below), the following characters serve to distinguish the two species. C. parthenopeum C. durbanense Nucleus of operculum: Apical. Lateral. Columella callus: Strongly plicate, with chocolate-brown interstices. Feebly plicate, uniform white. Spiral cords: Without regular transverse notches With shallow, but regular trans- verse notches. Colour of animal: Greenish-yellow, head and sides of foot covered with black spots, each of which is ringed with bright yellow. Yellowish, head and sides of foot profusely spotted with liver- brown. Radula: Rhachidian approximately § as long as wide, median cusp well developed, laterals and marginals typical for genus; about 50 rows of teeth. 45 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 4, DECEMBER 1970 Range: Fresh dead shells have been examined from many localities between Jeffreys Bay (E. L. Mus., Nat. Mus.) and Santa Carolina Island, Mozambique (E. L. Mus., leg. Mrs T. Trow). Living specimens have been collected by the author at Durban and Chaka’s Rock, Umhlali district, on the undersides of rocks covered with a surface layer of muddy silt, along the infratidal fringe. Cymatium (Cabestana) labiosum (Wood) Murex labiosus Wood, 1828: 15, pi. 5, fig. 18. Triton labiosus; Reeve, 1844, sp. 52; Sowerby, 1892: 8. Cymatium labiosum; Turton, 1932: 110, pi. 24, no. 796; Cernohorsky, 1967: 317, pi. 143, fig. 8. The specimen figured by Turton is very beach worn, but is quite distinctive, and is certainly not referable to C. durbanense as suggested by Barnard (1963a: 27). Fresh specimens of C. labiosum are not uncommon in Natal, although as yet only one living specimen seems to have been collected in South African waters (Durban; Nat. Mus., leg. Burnup). The East London Museum has shells from Embotyi (Pondoland), East London and Jeffreys Bay. Fam. Naticidae Natica forata Reeve (Fig. 11) Natica forata Reeve, 1855: sp. 129; Turton, 1932: 158. Natica queketti Sowerby, 1894: 371 ; 1897, pi. 6, fig. 6. Syn. nov. Natica africana Bartsch, 1915: 138, pi. 13, figs. 13 — 15; Turton, 1932: 158. Syn. nov. Natica forata adjacens Turton, 1932: 158, pi. 35, fig. 1127. Syn. nov. This variable species is characterized by the presence of two thin spiral lirae entering the umbilicus, which also has a weak marginal ridge and a series of low, incremental periostracal lamellae within. The protoconch is low, of \\ whorls, and measures 0-79 mm. in diameter. Operculum calcareous, surface somewhat concave, smooth, except for three slender, flattened spiral threads along the outer margin, the outermost forming the actual edge. Radula with tricuspid rhachidian, laterals with weak cutting edges and small cusps; marginals simple (non-bifurcate). Although typical examples of N. forata and N. queketti appear quite distinct, the numerous intermediate specimens that have been examined show these to represent the extreme forms of a single species. N. africana Bartsch is, in fact, based on such an intermediate. Typical forata has a low, very obtuse spire, and a wide umbilicus, while in queketti the spire is distinctly higher and the umbilicus very narrow. Both forms occur in the Jeffreys Bay — East London area, while the form queketti ranges north through Natal to Bazaruto Island, Mozambique; the latter was also recorded from the Persian Gulf and Karachi by Melvill & Standen (1901 : 358). N. forata is unusual among South African naticids in that it appears to live chiefly in crevices in the undersides of rocks, generally in lower balanoid zone pools, rather than on sand or mud. Fam. Turbinidae Genus Cinysca nom. nov. Cynisca Adams, H. & A., 1954: 406 (non Gray, 1844), type-species: (original designation) Cyclostrema granulata A. Adams, 1853 [= Delphinula granulosa Krauss, 1848], Knight et al (1960: 270) have correctly pointed out that Cynisca is a junior homonym. However their suggestion that this genus might be a synonym of Leptocollonia Powell (1951 : 105, type-species L. thielei Powell) cannot be supported, as its members have a liotiid 46 KILBURN : TAXONOMIC NOTES ON SOUTH AFRICAN MARINE MOLLUSCA, I operculum, with spiral rows of small calcareous beads, while in Leptocollonia the operculum is calcareous and smooth, except for a deep spiral groove (i.e., it is homalopomatid). The radula of Ciny sea granulosa { Krauss) has been figured by Barnard (1963b: 226, fig. 8a); the rhachidian, however, is there incorrectly drawn, for staining reveals the presence of alate lateral expansions (fig. 10) such as occur in Leptocollonia thielei (Powell, 1951, fig. G13) and Homalopoma carpenter i Pilsbry (Pilsbry, 1888, pi. 60, fig. 73). Fam. Trochidae Monilea (Priotrochus) poosonbyi (Sower by) Trochus ( Gibbula ?) ponsonbyi Sowerby, 1888: 209, pi. 11, fig. 5. Priotrochus alexandri Tomlin, 1926: 295, pi. 16, fig. 8; Barnard, 1956b: 251, fig. 13. Syn. nov. Sowerby’s description of Trochus ponsonbyi was evidently overlooked by Tomlin. Although apparently extinct, this species is abundant in raised Pleistocene beaches around the mouth of the Zwartkops River in Algoa Bay, and specimens are occasionally washed up on the neighbouring shores. This was presumably the origin of both Sowerby’s and Tomlin’s examples. Trochus iiigropunctatus Reeve Trochus hanley anus {non Reeve); Krauss, 1848: 100. Trochus nigropunctatus Reeve, 1861 : sp. 71 ; Barnard, 1963b, 253, fig. 14b. Trochus textilis Reeve, 1861 : sp. 82; Smith, 1903: 388. Syn. nov. Trochus flammulatus ( non Lamarck); Braga, 1956: 32, pi. 6, fig. 6. Reeve’s figure of 7 . textilis from the “Cape of Good Hope’’ clearly shows a slightly worn example of T. nigropunctatus. The type locality of the latter was given as Natal, here restricted to Durban. In Natal this species is abundant in sheltered mid-tidal pools, both among marine growths and under rocks. The western limit of its range appears to be Bonza Bay, about three miles north-east of East London (living, in coll. Mrs M. Rix). Fam. Limopsidae Philobrya limoides (E. A. Smith) Hochstetteria limoides Smith, 1904: 42, pi. 3, fig. 25. Philobrya smithi (nom. nov.) Barnard, 1964: 386, 388, fig. 5b (synonymy). Smith’s name must be reinstated. The existence of a junior homonym (P. limoides Smith, 1907) does not necessitate renaming of the senior homonym. REFERENCES Adams, H. & A., 1853 — 1858. The genera of recent Moll use a: arranged according to their organization. London: Van Voorst. Amio, M., 1955. “On the eggs and early life-histories of Pyrenidae (Columbellidae) in marine gastropods”, J. Shimonoseki Coll. Fish , 4:231 — 238. Barnard, K. H., 1958. “Contributions to the knowledge of South African marine Moilusca. Part I. Gastro- poda: Prosobranchiata: Toxoglossa”, Ann. S. Afr. Mas., 44:73 — 163. Barnard, K. H., 1959. “Contributions to the knowledge of South African marine Moilusca. Part II. Gastropoda: Prosobranchiata: Rhachiglossa”, Ann. S. Afr. Mas., 45:1 — -237. Barnard, K. H., 1963a. “Contributions to the knowledge of South African marine Moilusca. Part HI. Gastropoda: Prosobranchiata: Taenioglossa”, Ann. S. Afr. Mas., 47:1 — 199. 47 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 4, DECEMBER 1970 Barnard, K. H., 1963b. “Contributions to the knowledge of South African Marine Mollusca. Part IV. Gastropoda: Prosobranchiata: Rhipidoglossa, Docoglossa. Tectibranchiata. Polyplacophora. Solenogastres. Scaphopoda”, Ann. S. Afr. Mus., 47:201 — 360. Barnard, K. H., 1964. “Contributions to the knowledge of South African marine Mollusca. Part V. Lammellibranchiata”, Ann. S. Afr. Mus., 47:361 — 593. Barnard, K. H., 1969. “Contributions to the knowledge of South African marine Mollusca. Part VI”, Supplement. Ann. S. Afr. Mus., 47:595 — 661. Bartsch, P., 1915. “Report on the Turton Collection of South African marine molluscs, with additional notes on other South African shells contained in the United States National Museum”, Bull. U.S. nat. Mus., 9!:i — xii, 1 — -305. Braga, J. M., 1952. “Materials para o estudo de fauna malacologica de Mocambique”, An Jta. Invest. Ultramar, 7:5 — 67, pis. 1 — 44. Cernohorsky, W. O., 1967. “The Bursidae, Cymatiidae and Colubrariidae of Fiji (Mollusca: Gastropoda)”, Veliger , 9:310—329. Cernohorsky, W. O., 1967. “The Muricidae of Fiji. Part I — subfamilies Muricinae and Tritonaliinae”, Veliger, 10:111—132. Clench, W. j. & Turner, R. D., 1 957. “The family Cymatiidae in the Western Atlantic”, Johnsonia, 3:189 — 244. Knight, B. j., et al., 1960. In Moore, R. C. ed Treatise on Invertebrate Palaeontology. Part I, Mollusca, 1:169 — 331. New York: Geol. Soc. Amer. Krauss, F., 1948. Die siidafrikanischen Mollusken. Stuttgart: Ebner & Seubert. Melvill, J. C. & Standen, R. M., 1901. “The Mollusca of the Persian Gulf, Gulf of Oman, and Arabian Sea, as evidenced mainly through the collections of Mr. F. W. Townsend, 1893 — 4900, with de- scriptions of new species”, Proc. zool. Soc. Lond., 1901 327' — 4-60. Pilsbry, H. A., 1888. In Tryon. G. W. Manual of Conchology 10. Philadelphia. Powell, A. W. B., 1933. “Notes on the taxonomy of the recent Cymatiidae and Naticidae of New Zealand”, Trans. New Zealand Inst., 63:154 — -168. Powell, A. W. B., 1951. “Antarctic and subantarctic Mollusca: Pelecypoda and Gastropoda”, Discovery Rep., 26:47—196. Powell, A. W. B., 1966. “The molluscan families Speightiidae and Turridae”, Bull. Auckland Inst. Mus., No. 5:5—184, pis. 1—23. Reeve, L. A., 1843 — 1865. Conchologia Iconica. London. Robertson, R., 1957. “A study of Cantharus multangulus (Philippi), with notes on Cantharus and Pseudo- neptunea (Gastropoda: Buccinidae)”, Not. nat. Acad. Philad., No. 300:1 — 40. Smith, E. A., 1882. “Diagnoses of new species of Pleurotomidae in the British Museum”, Ann. Mag. nat. Hist., (5)10:206—218. Smith, E. A. 1899. “Descriptions of new species of South African marine shells”, J. Conch., 9:247—252. Smith E. A., 1903. “A list of species of Mollusca from South Africa, forming an appendix to G. B. Sowerby’s ‘Marine shells of South Africa’”, Proc. malac. Soc. Lond., 5:354 — 402. Smith, E. A., 1904. “On a collection of marine shells from Port Alfred, Cape Colony”, J. Malacol., 11 :21— 44. Smith, E. A., 1906. “On South African marine Mollusca with descriptions of new species”, Ann. Natal Mus., 1:19—71. Sowerby, G. B., 1834 — 4841. Conchological Illustrations. London: Sowerby. Sowerby, G. B., 1888. “Descriptions of sixteen new species of shells”, Proc. zool. Soc. Lond., 1888,207 — 213. Sowerby, G. B., 1892. Marine shells of South Africa, i — iv, 1 — 89. London: Sowerby. Sowerby, G. B. 1894. “Marine shells of South Africa”, J. Conch., 7:368 — 378. Sowerby, G. B., 1897. Appendix to marine shells of South Africa, i, 1 — 42. London. Sowerby, G. B., 1900. “On some marine shells from Pondoland and the Kowie, with descriptions of seven- teen new species”, Proc. malac. Soc. Lond.. 4:1 — 7. Tomlin, J. R. le B., 1926. “On South African marine Mollusca with descriptions of new species”, Ann. Natal Mus., 5:283—301. Tomlin, J. R. le B., 1948. “Reports on the marine Mollusca in the collections of the South African Museum. XL Family Buccinidae”, Ann. S. Afr. Mus., 36:333 — 339. Troschel, F. H., 1866 — 4879. Das Gebiss der Schnecken zur Begriindung einer naturlichen Classification, 2:1—246. Berlin: Nicolaische Verlagsbuchhandlung. Turton, W. H., 1932. The marine shells of Port Alfred, S. Africa: i — xvi, 1 — 331. Oxford: University Press. Vokes, E. H., 1964. “Supraspecific groups in the subfamilies Muricinae and Tritonaliinae (Gastropoda: Muricidae)”, Malacologia, 2:1 — -41. Wood, W., 1828. Supplement to Index Testaceologicus: 1 — 59. London. i PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. {Nat. Hist.) PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Four new forest Millipedes from Lesotho and the Eastern Cape by R. F. LAWRENCE Albany Museum, Grahams! own, South Africa INTRODUCTION The genus Gnomeskelus is one of the largest and most widespread of the Dipfopod genera in Southern Africa; it ranges over almost the entire coastal margin of southern Africa from north of Lourengo Marques in the east to the Cape Peninsula in the west. It is an indicator of indigenous forest and is thus completely absent from the western coastline of southern Africa north of Cape Town. Four new species of this genus are described here, three from the eastern Cape, one from Lesotho, and a list is given of the species known at the present time. The types of all four species are deposited in the Albany Museum, Grahamstown. CONTENTS (i) Gnomeskelus outeniqua spec. nov. ....... Gnomeskelus basuticus spec, nov Gnomeskelus graemi spec. nov. Gnomeskelus montifelis spec, nov (ii) A list of the species of Gnomeskelus arranged under provinces (iii) The genus Gnomeskelus Attems Bibliography Page 49 52 51 52 54 53 54 Genus Gnomeskelus Attems Gnomeskelus outeniqua spec. nov. (Fig. la, b) Holotype 1 G, paratypes 2 3$, Natures Valley, mouth of the Grootrivier, Cape Province, collected R. F. Lawrence, 20th September 1969. Colour in general dirty white to light cream, collum with a narrow light violet anterior and posterior border, body segments with a similar border but only posteriorly; legs in middle of the body with the basal segments mottled violet, the remainder dirty white. Head. Antennae long and slender, the penultimate segment subequal to the two preceding ones together. Body with very fine, microscopically small setae, collum with two transverse rows of a few setae each, one along anterior margin, the second in the middle of the segment. Body segments with one row posterior to the transverse suture. Keels of all segments very weak, almost completely obsolete. 49 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 5, DECEMBER 1970 Gnomeskelus outeniqua nov. spec.: a , gonopod aboral view, b, the parsolanomerit in oral view; Gnomeskelus graemi nov. spec.: c, gonopod, aboral view: Gnomeskelus basuticus nov. spec. : d, gonopod, oral view ; e, leg XVIII ; Gnomeskelus montifelis nov. spec. /, gonopod, aboral view. 50 LAWRENCE: NEW FOREST MILLIPEDES FROM LESOTHO AND EASTERN CAPE Legs rather short, with abbreviated segments; those in the middle of the body with the whole of the inferior surface of tarsus (except for a small area at the base and apex) with transversely striated spherical nodules crowned with a short, smooth, thick cone, the nodules in general resembling those figured for G. tembulicus Lawrence (1963, p. 301, fig. 2/); tibia with a row of 4, patella with 3, femur with 2 — 4 near the distal apex only; segments of the legs without processes or tubercles. Gonopods as in fig. 1 a (aboral) and 1 b (oral) view; no tibio-tarsus; parsolanomerit large and partly divided at its apex, the distal fork small and inconspicuous; in one specimen there is no division, and the structure ends in a simple large triangle which is sharply pointed. On the oral side of the femur there is a bluntly rounded projection opposite the origin of the parsolanomerit. Dimensions. Total length 1 1 mm, width about 9 mm. The species apparently most closely resembles G. bicornis Schubart from Van Staden’s Pass, near Port Elizabeth. It differs in having the processes at the apex of the parsolanomerit only feebly developed; there is no marked constriction dividing the femur from the prefemur of the gonopod as in bicornis. Gnomeskelus graemi nov. spec. (Fig. 1 c) Holotype 1 paratype 1 <$, Dassiekrans, Grahamstown, Cape Province, collected R. F. Lawrence, 4th October 1967. Colour uniformly pale off-white, without markings (spirit specimens). Head. Antennae moderately long, the penultimate segment a little shorter than the two preceding ones together and somewhat incrassate. Body. Collum and the remaining segments with 1 transverse row of very fine short setae; the keels visible at the sides but small and sharp, not produced but ending at the postero- lateral angles in a small sharp tooth which is however very short. Legs with neither tubercles (swellings) nor processes on their ventral surfaces; the modified setae (spherical nodules) of the inferior surfaces of the segments more triangular in shape, the pointed cone at the apex of each, long, almost as long as the basal striated portion; the legs in the middle of the body with the basal and apical fourth of the tarsi free of nodules, tibia with nodules only in the distal half, patella and femur without any. Gonopods rather simple, fig. 1 c, resembling in general those of burins Verhoeff and arcuatus Verhoeff from Natal; no tibio-tarsus; parsolanomerit absent or represented by a small tri- angular tooth, solanomerit with a very slender lateral branch, femur and prefemur not divided by a constriction, sperm canal clearly visible. Dimensions. Total length about 12-7, width 1-3 mm. Additional material: 24 S S and $ $ from the type locality, collected R. F. Lawrence, 13th October 1969. The colouring of these forms is as follows: Antennae, collum and body segments dorsally varying from light brown with a reddish tinge to dirty yellow and off-white, a narrow blackish stripe down the middle of the dorsum; head, sides of body and legs uniformly pale. This small form occurs in forest humus side by side with G . hewitti, a much larger un- related species. 51 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 5, DECEMBER 1970 Gnomeskelus basuticus nov. spec. (Fig. Id, e) Ho/otype 1 <$, Masite Mountain, altitude ±5,500, near Morija, Maseru district, Lesotho, collected by J. Hewitt. Colour. Dirty white without darker markings, probably faded in alcohol; antennae a little infuscated with violet. Head. Antennae moderately long, the penultimate segment subequal to the two preceding ones together. Collum and body. Collum with some fairly long fine setae scattered irregularly in anterior half of segment; all body segments dorsally quite naked, even the last one, but telson below with 2 — 3 pairs of distinct setae on the valves, pygidium with a single pair; keels fine and distinct but not at all prominent. Legs moderate in length but distinctly longer than in the two preceding species, the seg- ments swollen at their apices where, especially ventrally, there are rounded projections. No spherical nodules except on a low rounded swelling near the distal end of the tarsus ventrally (leg XII), and then only a few, fig. le; proximally to these a few along the tarsus with the basal swollen portion elongate, not spherical, and surmounted by a thick pointed spine which is longer than the swollen base of the nodule. Gonopods as in fig. 1 d seen in oral view, the femur slender, long, with almost parallel sides, a distinct constriction separating the solanomerit from the femur, as in G. armiger Schubart from Wellington, Cape (1956, p. 69 fig. 42); the tibio-tarsus in the form of a short sharp tooth. Dimensions. Total length 11-5 mm, width 1 mm. In the opinion of the writer, as also expressed in a previous paper (1966, p. 249), Schubart’s subgenus Pristomeskelus , should be reserved for the following six species only: penicillatus Attems, cere sinus Attems armiger Schubart, clavatus Attems, krugeri Lawrence and the above species. They are all distinguished by having the femur of the gonopods separated from the distal apex of the appendage (solanomerit) by a distinct constriction. In the detailed con- struction of the gonopods this species from Lesotho does not appear to resemble any of the remaining species of Pristomeskelus Schubart. Gnomeskelus montifelis nov. spec. (Fig. 1 /) Holotype 1 $, paratypes 2 2 $ §, Katberg, Winterberg Mountains, collected by J. Hewitt, January 1927. Colour , probably faded in alcohol, a uniform dirty white or pale yellow without darker markings. Head . Antennae not elongate, penultimate segment shorter than the two preceding ones together. Body. Collum with several transverse rows of minute setae, body segments dorsally with only one row posterior to the transverse suture, last segment with fairly numerous setae. Keels almost obsolete but visible. Legs moderate to short, tarsus however fairly long and slender, with spherical nodules ventrally except on basal third or apical fifth of the segment; in addition with a series of fairly long and strong curved setae on the ventral surface, these setae shorter and less numerous on the remaining segments of the legs; tibia with spherical nodules only in apical half, patella 52 LAWRENCE: NEW FOREST MILLIPEDES FROM LESOTHO AND EASTERN CAPE and femur without these; patella-femur without blunt apical processes or only slightly enlarged distally, femur a little more so than the remaining segments. Gonopods as in fig. 1 / seen in aboral view, base of the femur apparently with only one large setose-spine, parsolanomerit short, slender, bacilliform, the whole gonopod unlike that of any other species except for the solanomerit which resembles that of auriculatus Lawrence from Natal. Dimensions. Total length about 13 mm; width 1 *2 mm. The Genus Gnomeskelus Attems The genus consists of a large number of species, the great majority of which are limited to the litoral or sublitoral forests of southern Africa. When the genus was first proposed in 1928 only 13 species were described while at the present time 82 are known. It affords a good example of one genus of a family greatly exceeding all others in numbers of species, the great proliferation of forms taking place at the extreme south of the African continent. A parallel case exists among the Spirostreptomorph millipedes of the family Odontopygidae where the genus Spinotarsus with 96 species far outstrips in numbers any of its congeners in southern Africa; while Gnomeskelus is strictly limited to indigenous forest however, most of the species occurring in coastal forest or in forests not more than 50 miles inland, Spinotarsus is less dependent on humidity and shade, flourishing in regions of prairie, bushveld or thorn- veld such as are found in the lowveld of the Kruger National Park; its penetration of the indigenous montane and coastal forests is probably secondary and incidental. With respect to Gnomeskelus on the other hand, as can be seen from the list of species, only eight of the 82 species are found in the Transvaal and of these only two occur in the lowveld, the remaining species living at fairly high altitudes in the mist-belt forests which remain on the slopes of the northern extension of the Drakensberg Range. Both these genera, in spite of their numerous species, are still incompletely known and the number of forms listed will probably eventually reach a half again of the present total. More species than have already been described may be expected to occur between Port Elizabeth and the Cape Peninsula. In most cases, and this is also true of the Odontopygid Spinotarsus , the various species of Gnomeskelus are localized, occupying very limited geographical areas and easily distinguish- able from each other on the basis of clearcut differences in the S gonopods. An exception is to be found in the case of G. tuberosus Attems, a large, mainly litoral species, in which a number of forms have been described, with the gonopods conforming to a single general pattern. A dozen such forms, deviating from the normal structure of the gonopod in minor details have already been described as subspecies; these range from Richards Bay to Port St. Johns and no doubt others will be found. Although it seems likely that according to P. M. Johns (personal communication) Gnomeskelus is a synonym of Stenauchenius Attems, described from a single female from Port Elizabeth, it would be premature to accept this before it is definitely known that the female type of Stenauchenius has been compared with various females of Gnomeskelus. The author of the two genera, C. von Attems, must have had females of both before him for comparison when erecting his new genus Gnomeskelus in 1 928, and it may therefore be supposed that he considered them to be different. 53 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 5, DECEMBER 1970 A LIST OF THE SPECIES OF GNOMESKELUS ARRANGED UNDER PROVINCES Natal — Zululand and Mozambique Transvaal Cape ( including the Transkei) Lesotho arator Lawrence 1962 cyniceps Lawrence 1958 krugeri Lawrence 1966 rudebecki Lawrence 1959 arcuatus VerhoefF 1939 attemsii Verhoeff 1939 auriculatus Lawrence 1953 bi fur cat us Lawrence 1953 brevipes Lawrence 1953 brincki Schubart 1956 burins Verhoeff 1939 circulipes Verhoeff 1939 cyclocanthus Lawrence 1958 skukuzae Lawrence 1966 dentipes Attems 1928 stuckenbergi Lawrence 1958 edentulus Lawrence 1953 f/uvialis Lawrence 1958 forcipifer Lawrence 1953 glaber Lawrence 1958 gonarthrodus Lawrence 1962 harpagonifer Verhoeff 1939 jaeulator Lawrence 1958 krausi Lawrence 1962 lawrencei Verhoeff 1939 larvatus Lawrence 1962 latzeii Verhoeff 1939 laevigatus Lawrence 1953 maritimus Lawrence 1962 medius Verhoeff 1939 minor Lawrence 1958 montivagus Verhoeff 1939 multident atus Verhoeff 1940 natalicus Attems 1928 origensis Lawrence 1953 petersii Verhoeff 1940 processiger Lawrence 1953 pugnifer Lawrence 1953 retrusus Schubart 1956 setosus Verhoeff 1939 spectabilis Verhoeff 1940 spinifer Attems 1928 subterraneus Lawrence 1962 tenuipes Lawrence 1953 tereticornis Lawrence 1962 tuber osus tuber osus Attems 1927 tuberosus hamuliger 1939 tuber osus ur bonus Lawrence 1962 tuberosus clivicolus Lawrence 1962 tuberosus falcifer Verhoeff 1939 tuberosus microdens Lawrence 1962 tuberosus serratus Verhoeff 1939 tuberosus tristriatus Attems 1938 transvaalicus Lawrence 1958 trichar dti Lawrence 1962 armiger Schubart 1956 bacillifer Attems 1944 bicornis Schubart 1956 ceresinus Attems 1928 clavatus Attems 1928 e/izabethae Lawrence 1963 fitzsimonsi Lawr. 1959 globifer Attems 1928 graemi Lawrence 1969 hewitti Lawrence 1963 inermis Lawrence 1963 kambianus Lawrence 1963 mixtus Attems 1944 montifelis Lawrence 1969 outeniqua Lawrence 1969 penicillatus Attems 1927 puteinus Attems 1928 repandus Attems 1928 rhodobates Attems 1928 silvaticus Attems 1928 spiculifer Lawrence 1953 tembulicus Lawrence 1963 tuberosus tridens Lawrence 1962 tuberosus eweri Lawrence 1962 tuberosus furculatus Verhoeff 1937 tuberosus globulatus Attems 1 927 basuticus Lawrence 1969 REFERENCES Attems, C., 1927. “Wissenschaftliche Ergebnisse R. Grauer in Zentral Afrika”, Ann. Naturh. Mus. Wien., 41:58. Attems, C., 1928. “The Myriopoda of South Africa”, Ann. S. Afr. Mus., 26. Attems, C., 1934. “The Myriopoda of Natal and Zululand”, Ann. Natal Mus., 7:459 — 522. Attems, C, 1944. “Neue Polydesmoidea”, Zool. Anz., 144:223 — 251. Kraus, O., 1966. “Phylogenie, Chorologie and Systematik der Odontopygoideen”, Abh. Senckenb. naturf. Ges. N. S., 12:1—143. 54 LAWRENCE: NEW FOREST MILLIPEDES FROM LESOTHO AND EASTERN CAPE Lawrence, R. F., 1953. “A revision of the Polydesmoidea (Diplopoda) of Natal and Zululand”, Ann. Natal Mus., 12:292—349. Lawrence, R. F., 1958. “Contributions to the Myriopoda of Natal and Zululand”, Ibid. 14(2) :279 — 301. Lawrence, R. F., 1959. “A collection of Arachnida and Myriopoda from the Transvaal Museum”. Ann. Transvaal Mus. 23(4) :363 — 386. Lawrence, R. F., 1962. “New Polydesmoidea (Diplopoda) from South Africa”, Ann. Natal Mus., 15(14): 141—165. Lawrence, R. F., 1963. “New Myriopoda from Southern Africa”, Ibid, 15(23) :297 — 316. Lawrence, R. F., 1966. “The Myriopoda of the Kruger National Park”, Zool. Africana, 2(2):225— 262. Schubart, O., 1956. “Diplopoda I. Proterospermophora”, S. African Animal Life, 3:12 — 86. Verhoeff, K. W., 1939. “Uber Sudafrikanische Polydesmoideen”, Ann. Natal Mus., 9:183 — 201. Verhoeff, K. W., 1939. “Polydesmoideen, Colobognathen und Geophilomorphen aus Sudafrika, besonders den Drakensbergen, Natal”, Ibid., 9:203 — 224. Verhoeff, K. W., 1940. “Aliquid novi ex Afrika, I. Polydesmoidea und Colobognatha”, Zool. Anz. 130: 104—119. Verhoeff, K. W., 1940. “Aliquid novi ex Afrika, II. Zur Kenntnis der Siidafrikanischen gnomeskelus.” Ibid., 131:10 — -18. 55 • ' PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (Nat. Hist.) k VOLUME 8 • PART 6 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Aspects of adaptive radiation in Southern African Accipiters by R. A. R. BLACK AND G. J. B. ROSS Port Elizabeth Museum, Snake Park and Oceanarium ABSTRACT Some aspects of the adaptive radiation of African accipitrine hawks are discussed in this study, based on the wing and culmen measurements of seven species; Accipiter minullus (Daudin), A. badius (Gmelin), A. ovampensis (Gurney), A. rufiventris (Smith), A. taehiro {Daudin) and A. melanoleucus (Smith), and Melierax gabar (Daudin). These measurements are discussed in relation to observations on habitats and hunting methods. The species were found to be spread over an even size-gradient, though they form three groups in size, habitat and hunting methods. In all seven species the female is larger than the male. A. ovampensis and A. rufiventris are morphologically similar but appear to be allopatric. M. gabar appears to have occupied an available position in the Accipiter series. INTRODUCTION There has been little work on the adaptive radiation of African accipitrine hawks. Recent work on the three North American species of Accipiter show size differences between the sexes and species forming a size-gradient. The males are smaller than the females and tend to take smaller prey (Storer, 1966). The situation in the African sub-continent has not yet been studied with respect to size-gradient, inter- and intra-specific competition and habitat preferences. This study is based on the wing and culmen measurements of seven species: Accipiter minullus (Daudin), A. badius (Gmelin), A. ovampensis (Gurney), A. rufiventris (Smith), A. taehiro (Daudin), A. melanoleucus (Smith) and Melierax gabar (Daudin), supported by obser- vations of habitat and hunting methods of each species both in the field and in captive speci- mens. Melierax gabar was included in the study owing to its similar appearance and habits. These measurements are then interpreted in terms of the adaptive radiation of the two genera. METHODS AND MATERIAL Measurements were taken of adult specimens in the collection housed in the National Museum, Bulawayo, Rhodesia. Though the number of specimens could have been increased by the addition of immature birds, it was found that the proportion of mis-sexed birds was high, while some specimens had obviously not reached full size. The high proportion of mis- sexed birds probably resulted from the mis-identification of the paired ovaries (usual in these genera) for testes. A few adult specimens were also mis-sexed and were either re-sexed correctly or omitted. The specimens used in this study were: 57 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 6, DECEMBER 1970 A. minullus 15 8 9 9; A. badius 20 $ (J, 38 9 9; A. ovampensis 5 c? <5, 6 9 9; A. rufiventris 2 3 a. 4 9 9; A. tachiro 10 tJcJ, 7 9 9; A. melanoleucus l 9 9; M. gabar 26 rj :j , 29 9 9. Wing and culmen measurements were chosen as two features best reflecting adaptation to hunting. Though body length could have been used, it was found that specimens differed considerably as a result of skin preparation. The wing measurements were taken from the wrist to the tip of the longest primary when pressed flat on a metre rule, while the culmen was measured with calipers from the tip to the naso-frontal suture at the base of the culmen. The means and standard deviations of these measurements were calculated (Table I) and a plot of wing length against culmen length is shown in Figure I, in which the range of each measurement is represented by the standard deviation. The regression lines for wing/ culmen length (1) and culmen/wing (2) were calculated by the method of least squares (Simpson and Roe, 1939). Table I. Wing and culmen measurements of six Accipiter spp. and Melierax gabar Species 1 $$ No. of Specimens Culmen Wing No. of Specimens Culmen Wing Mean St. Dev. Mean St. Dev. Mean St. Dev. Mean St. Dev. A. minullus 15 15 2 0-37 141 3-8 8 16 9 0-39 162 4-3 A. badius 20 16 6 0-60 176 4-4 38 18-1 0-74 195 4-8 A. ovampensis 5 18 4 0-48 221 5-2 6 220 0-92 248 6-3 A. rufiventris 2 18-7 0-99 219 1-4 4 21 -9 1 06 240 3-9 A. tachiro 10 23-5 0-77 215 6-9 7 27-9 1 • 56 251 5-2 A. melanoleucus 2 28-3 1-40 295 10 7 1 32-8 — 329 M. gabar 26 18-4 0-48 186 4-1 29 20-7 0-67 200 6-3 \ || HABITATS AND HUNTING METHODS The separation of the species in Figure 1 into relatively shorter or longer winged forms suggested a relationship between wing length, habitat preference and hunting methods. However, as a review of the literature on the African Accipiters showed considerable con- fusion as to the habitat preferences and hunting methods of these hawks, it was felt that a fuller discussion of these aspects was necessary for the purposes of this study. All the accipitrine hawks make use of several well-known hunting methods, and the extent to which each is used varies with the species. The most acceptable terminology for these different methods is that used by falconers, and these are used here as the most de- scriptive terms available. 58 BLACK AND ROSS: ADAPTIVE RADIATION IN SOUTHERN AFRICAN ACCIPITERS The Accipiters “prospect” for concentrations of prey by flying high over the area at no great speed; the characteristic pattern is a succession of glides interrupted by a series of fast wing-beats. The prospect flight is particularly well-known in the African Goshawk A. tachiro. Their success in hunting, however, depends on more secretive or surprise tactics. Quiet waiting or “still-hunting” on an exposed branch, or behind screening foliage for suitable prey, followed by a swift plunge through the trees, with wings partially folded, is a method most suited to heavily wooded areas (Grossman and Hamlet, 1964). A second, more active, method of hunting is known as “speculative hunting” which is distinct from the prospect flight. The bird flies at considerable speed a few feet above the ground in open areas, or just below tree-top level in wooded areas, taking advantage of any irregularities of terrain. A low glide is followed by a lightning swerve over the top of any likely thicket to surprise any prey on the other side. This opportunistic method had been described for A. nisus in Europe by Mavrogordato (1960) and for A. gentilis in North America by Beebe (1964). These methods may be considered typical of the Accipiters. Some species clearly prefer one method, while others are sufficiently adaptable to use both methods with equal success. Beebe states that of the three North American species, A. velox is the most aerial, and does less still-hunting than either A. gentilis or A. cooperi. He states further that it shows a definite preference for open or sparsely-treed country rather than thick woods. A. minullus This species inhabits continuous or riverine forest where the cover is thick enough for still-hunting to be used to the full. A hawk this size is able to find cover even in relatively thin forest, and this secretive hunting habit is suggested as the reason for its apparent scarcity in areas where it may well be common. The captive bird, once tamed, “bates” (flying off the hand, often through restlessness) less frequently than any other Accipiter except A. tachiro , which may be considered the most relaxed of the African bird hawks. A. badius The Little Banded Goshawk occupies a similar habitat to A. minullus , but it appears to prefer more open woodland, and the range of habitats is less broad. While A. minullus is a fairly common breeding resident of the eastern Cape coastal belt, A. badius has only rarely been recorded on the fringe of this region, in dry, open country (Paterson, 1958). In captivity, this is a more restless hawk than either A. minullus or A. tachiro. It is a fairly fast flier, though not strongly orientated to bird prey, and much of its diet may consist of insects and small reptiles. A. ovampensis This species inhabits the open savannah veld (open woodland) and is also common in almost treeless grassland such as certain areas of the Transvaal highveld. In a similar way to A. rufiventris, it may often be found nesting in a small grove of trees in an otherwise treeless area (Black, personal observations). In captivity, it has been found to be a very swift flier, strongly orientated to bird prey. It is a restless, fierce hawk, not disdaining to chase the most swift of quarry such as quail (Savory, personal communication). In most areas inhabited by A. ovampensis there is little opportunity to use the method of still-hunting, and this species is almost entirely a speculative hunter. 59 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 6, DECEMBER 1970 Fig. 1. A plot of the means and standard deviations of the wing and culmen lengths in the seven species: a = A. minullus\ b — A. badius\ c = M. gabar\ d = A. rufiventris ; e = A. ovampensis’, f = A. tachiro ; g = A. melanoleucus. (Line 1 = regression of wing length/culmen length) (Line 2 = regression of culmen length/wing length) BLACK AND ROSS: ADAPTIVE RADIATION IN SOUTHERN AFRICAN ACCIPITERS A. rufiventris Contrary to the majority of literature on A. rufiventris this hawk is not strictly a forest form, and observations have shown that it spends a good deal of time on the move, and is almost invariably seen on the fringe of, or in open country. It is common in certain areas of the Karoo, such as Graaff-Reinet (Black, personal observations), where it breeds in a few acres of poplars separated from the next grove by several miles, in otherwise treeless country. Its occurrence in the eastern highlands of Rhodesia (Rushworth, personal communication) and in Basutoland (Jacot-Guillarmod, 1963), where the wooded areas are separated by large tracts of open country, also show its preference for less wooded country. The distribution, hunting habits and size suggests that it is the ecological replacement of A. ovampensis in mountainous regions and to the south of the latter’s range. Observations on the trained bird indicate that it relies on speculative rather than still-hunting and is strongly orientated to bird prey; this is further reflected in its restlessness in captivity. A. tachiro Dense forest and thick riverine forest form the optimum for this species, which becomes increasingly scarce as the cover thins out towards the west. It is a very secretive hawk, re- lying on still-hunting more than any other Accipiter sp. (Black, 1968). Its great disadvantage from the falconer’s point of view is its habit of attacking only when it senses an overwhelming advantage over the prey. It shuns a long, hard chase, and immediately recognizes certain slow-flying species such as loeries and coucals. A. melanoleucus This species appears to be equally common in open woodland as in thick forest. It is an extremely fast and relentless bird in the chase, and has been seen chasing homing pigeons for long distances. In general hunting habits, it uses both speculative and still-hunting with equal success. Melierax gabar The Gabar Goshawk is absent from the dense coastal forest, but inhabits a wide variety of habitats ranging from dry woodland through lightly wooded savannah into the more arid areas of South-West Africa, where there is suitable cover (dense riverine scrub). The captive bird is fast and relentless, and is strongly orientated to bird prey, although less so than A. rufiventris or A. ovampensis. Observations in the field indicate that it uses a modified form of still-hunting. It is often found waiting on an exposed branch for its prey, but the attack may well include a fairly long chase. DISCUSSION Figure 1 shows two points clearly; the female of all seven species is always larger than the male, with no overlap, and the species are spread over an even size-gradient. While it is agreed by Amadon (1959), Cade (1960) and Selander (1966) that the “reversed" sexual dimorphism of raptors is correlated with predatory habits, there is some disagreement as to the origin of this dimorphism. Amadon and Cade suggest that the larger size of the female is related to the difficulty of pair formation in predatory birds, while Selander suggests “the basic adaptive function of the dimorphism is related to differential niche utilization". 61 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 6, DECEMBER 1970 y 10 cm. Fig. 2. The outlines of the wings of juvenile ? A. rufiventris and A A. tachiro, illustrating differences in the proportions of the wing. 62 BLACK AND ROSS: ADAPTIVE RADIATION IN SOUTHERN AFRICAN ACCIPITERS Storer (1966) has shown that the males do take smaller prey in three North American Accipiter spp. and it is assumed that the same holds true for the African species discussed here. Storer has also shown that the size of prey is directly related to the size of the hawk; this reduces inter-specific competition to a minimum, and results in an even exploitation of the available prey (cf., Crombie, 1947). The American Accipiter spp. however, are sufficiently separated in size to allow a considerable overlap in habitat. There is a considerable overlap in size in the African species, as is shown in Figure 1. However, it is found that the species fall into three groups separated by the regression lines 1 and 2, and that these groupings can be related to habitat preferences and hunting methods. Those species with a shorter wing relative to beak length, appearing above the regression line 1, are A. minnllus , A. tachiro and M. gabar. These are all species preferring dense cover, and using still-hunting predominantly. M. gabar , however, tends to be less separated than either A. minnllus or A. tachiro , as reflected in both Figure 1 and in its habitat and hunting methods. The second group, appearing below the regression line 2, is composed of A. badius, A. rufiventris and A. ovampensis, whose wings are longer relative to beak length. These are all species inhabiting more open country, and in which speculative hunting predominates. In fact, A. rufiventris and A. ovampensis are the longest winged and most aerial of all the African Accipiter spp. The third group, containing A. melanoleucus, lies between the regression lines 1 and 2, and its intermediate position relates well to its habitat preferences and inter- mediate hunting habits. The speed of this species is a reflection of its size rather than its wing length, for the larger a bird, the faster it must fly to stay airborne (Storer, 1966). At first, these groupings suggest that a shorter wing has some advantage in dense cover, such as aiding in manoeuvrability. At this point, however, it is important to remember that the wing length referred to in the proceeding discussion is that from the wrist to the tip of the wing (hand) only, and not of the whole wing. The primaries provide the propulsive force in flight, while the inner part of the wing acts as a relatively stationary aerofoil, providing lift. Thus a fast-flying bird requires a larger area of wing for propulsion, and a correspondingly smaller area for lift than a slower flier, e.g., swifts (Harrison, 1964). Figure 2 shows the wing patterns of perhaps the most extreme examples of this effect in the African Accipiter series, taken from two specimens soon after death. Figure 2a is the wing of a juvenile $ A. tachiro weighing 205 gm. (7^ oz.) while Figure 2b shows that of a juvenile $ A. rufiventris weighing 220 gm. (7f oz.). While these two birds are of comparable weight and size, and the wings are almost identical in length, the proportions of the hand length and the forearm length differ considerably. The greater length of the forearm in A. tachiro is clearly an adaptation providing lift and manoeuvrability while flying relatively slowly in dense cover. Similarly the longer hand of A. rufiventris is an adaptation to fast flight. An experiment by Chapeau on the flight of doves emphasizes the importance of the primaries in the propulsion of the bird. The removal of a small portion of the tips of the primaries prevented a dove from flying, while the reduction of 55% of the total wing area by removing secondaries still permitted flight (Welty, 1962). This is an indication of the importance of the differences in the hand length of these Accipiter spp. A further important factor as yet undiscussed is the breadth of the wing, in particular in the region of the forearm, where the breadth is the same as the length of the secondaries. This length and that of the forearm determine the wing area of the inner part of the wing and thus the lifting capacity of the wing. The preparation of specimens precluded the measure- ment of wing breadth and area, but some indication of these can be seen in Figure 3, where each species has been traced from photographs. The length of primaries showing beyond the secondaries when the wing is folded back varies from species to species. This length reflects the proportion between the length of the hand to the length of the secondaries and to a lesser extent that of the forearm as well, as is shown in the partially open wing of A. ovampensis 63 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 6, DECEMBER 1970 F Fig. 3. Outline tracings from photographs illustrating the estimation of inner-wing area from the length of primaries extending beyond the secondaries (shaded black): A = A. tachiro ; B = A. minullus ; C = M. gabar\ D — A. melanoleucus; E — A. badius\ F A. rufiventris; G = A. ovampensis 64 BLACK AND ROSS: ADAPTIVE RADIATION IN SOUTHERN AFRICAN ACCIPITERS in Figure 3g. It is interesting to note that those species demonstrated on this basis to have the largest inner-wing area are the dense cover species A. tachiro , A. minullus and M. gabar , while those with the smaller area are A. rufiventris and A. ovampensis. The almost complete overlap in size and proportions of A. rufiventris and A. ovampensis indicates competition for the same prey. The two species may be considered allopatric, for their ranges only overlap slightly in the eastern highlands of Rhodesia and north-eastern Transvaal, where A. ovampensis is reported to be sparse (Rushworth, personal communi- cation). Judging from the numbers of species of the range of the genus Accipiter, it may well be considered the most successful of the sub-family Accipitrinae. It is generally considered that the genus Melierax is an offshoot within the main stem of the group. Of the three species M. musicus and M. metabates are extremely successful in their own niches; they are Buteo- like in habits, with relatively little resemblance to the typical bird-hawks. M. gabar , however, differs considerably in habits and habitat from M. musicus and M. metabates. Its position in Figure 1 suggests that it has adapted as a typical bird hawk, filling an available gap in the Accipiter series. ACKNOWLEDGEMENTS The authors wish to thank Mr M. P. S. Irwin for the use of specimens, and for much help during the initial part of the study. We also wish to thank D. Rushworth and A. Savory for their useful comments on distributions and habits. REFERENCES Amodon, D., 1959. “The significance of sexual differences in size among birds”, Proc. amer. Philos. Soc. 103:531—536. Beebe, F. L., 1964. North American Falconry and Hunting Hawks. World Press Inc., Denver. Black, R. A. R., 1968. “The African Goshawk”, East. Cape Nat. No. 34. Cade, T. J., 1960. “Ecology of the Peregrine and Gyrfalcon Populations in Alaska”, Univ. Calif. Publ. Zool. 63:151 — 290. Crombie, A. C., 1947. “Interspecific competition”, Jour. Animal Ecology. 16:44 — 73. Grossman, M. L. & Hamlet, J., 1964. Birds of Prey of the World. Cassell, London. Harrison, J., 1964. A New Dictionary of Birds. (Landsborough Thompson ed.) Nelson, London. Jacot-Guillarmod, C., 1963. “Catalogue of the Birds of Basutoland”, The South African Avifauna Series. No. 8. Mavrogordato, J. G., 1960. Hawk for the Bush. H. F. & G. Witherby, London. Paterson, J. M., 1958. Check List of Birds of the Eastern Cape Province. East Cape Wild Bird Soc. Selander, R. K., 1966. “Sexual Dimorphism and differential niche utilization in birds”. Condor, 60:1 13 — 151. Simpson, G. G. & Roe, A., 1939. Quantitative Zoology. McGraw-Hill Book Co., New York and London. Storer, R. W., 1966. “Sexual dimorphism and food habits in three North American Accipiters”, The Auk, 83:423—436. Welty, J. C., 1962. The Life of Birds. W.B. Saunders Co. 65 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (Nat. Hist.) b MAR 1 STSS71 ) VOLUME 8 • PART 7 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA : Tilapia Mossambica Peters, from Australia R. A. JUBB Albany Museum, C.S.I.R. Bursar F. O. PETRICK Provincial Fisheries Institute, Lydenburg In June 1969 three specimens of Tilapia were received from Dr W. J. R. Lanzing of the University of Sydney, Australia. These were identified as Tilapia mossambica Peters, but, in view of the lack of knowledge regarding the origin of their ancestors, a detailed study of these specimens was carried out by Mr F. O. Petrick at the Provincial Fisheries Institute, Lydenburg. This morphological study included direct comparison with specimens of Tilapia mossambica from Transvaal waters. As stated by Atz (1954), just how and when Tilapia left Africa and reached the East Indies is a mystery. Its presence in East Java was discovered in 1939 by an overseer of fisheries, Pak Mudjair by name, who explained that the new fish was collected in a small lagoon of the Serang River on the south coast of Java. This strange fish, subsequently identified as Tilapia mossambica Peters, became very popular with Javanese fish culturists and fingerlings were transported to various places all over the island. In honour of its discoverer the fish was named “Ikan Mudjair”, which in Indonesian means Mudjair’s Fish. This new introduction soon made itself at home in a wide variety of habitats, both natural and artificial, and it thrived in both brackish and fresh waters. During World War II the Japanese army of occupation distributed Tilapia mossambica widely, and by the end of the war this species was established in the islands of the East Indies and parts of south-east Asia. Tilapia are on the list of forbidden imports of live fish into Aus- tralia but Dr Lanzing was able to obtain stock for research purposes from Singapore. It is reasonable to assume that the Singapore stock originated from Java. Petrick’s detailed morphological comparison of the Australian specimens with Transvaal material follows: 67 ANN. CAPE PROV. MUS. (NAT. HIS.) YOL. 8, PT 7, DECEMBER 1970 BODY RATIOS Far East T. mossambica Transvaal T. mossambica Range Mean Range Mean Mouth width/Head width 0 641- -0- 8917 0 7777 0- 65- -0 83 0- 76 Body length/Body depth 2 538- -2- 875 2 682 2- 59- -3 0 2- 73 Body length/Head length 2 511- -2- 933 2 773 3- 0 - -3 2 3- 08 Head length/Snout length 2 420—2- 969 2 655 2- 2 - -2 8 2 46 Head length/Eye diameter 4 63 - -5- 232 4 954 5 5 - -6 1 5 8 Head length/Postocular part of head .... 2 262- -2- 5 2 421 2 2 - -2 5 2 31 Head length/Interorbital width 3 0 - -3- 358 3 149 2 6 - -2 64 2 6 Head length/Head width 1 753- -1- 953 1 860 1 7 - -1 9 1 8 Postocular part of head/Eye diameter .... 2 0 - -2- 093 2 048 2 3 - -2 6 2 47 Interorbital width/Eye diameter 1 5 - -1- 667 1 575 1 9 - -2 3 2 14 Mouth width/Eye diameter 1 707- -2- 367 2 142 2 0 - -2 8 2 4 Postocular part of head/Snout length .... 0 968- -1- 3125 1 •103 0 9 - -1 30 1 06 Snout length/Eye diameter 1 561- -2- 163 1 •889 2 0 - -2 7 2 36 Snout length/Body length 0 117- -0 151 0 •1335 0 11- -0 148 0 1315 Mouth width/Head length 0 368- -0 475 0 •422 0 36- -0 43 0 41 Gut length/Body length 5 •9 4 42- -8 6 7 0 Lateral length lower pharyngeal/Width of lower pharyngeal 1 •342 1 29- -1- 34 1 •31 Length of premaxillary pedicle/diagonal length of premaxilla 50% SCALE COUNTS Lateral line 30—32 30—31 Lateral line to middle of back 3 3i— 4 4 Lateral line to middle of belly 13—15 14—16 Round caudal peduncle 16 16 Predorsal scales 9—12 10—12 GILL— RAKER COUNT Number of gill-rakers on lower portion of anterior arch 18 18 68 JUBB AND PETRICK: TILAPIA MOSSAMBICA PETERS, FROM AUSTRALIA INTERNAL ANATOMY I. Vertebral Column Precaudal vertebrae Caudal vertebrae Total number of vertebrae Number of pleural ribs Number of epineurals Number of epipleurals Ventral vertebral apothysis Anterior articular facets on first vertebra . II. Ray counts on Fins Dorsal: spines branched rays pterygiophores Anal: spines branched rays pterygiophores Caudal: dorsal half : spines branched rays ventral half : spines branched rays dorsal and ventral caudal apophysis . Pelvic: spines branched rays Pectoral : spines branched rays Central caudal fin skeleton: epurals hypurals uroneurals III. Teeth on Jaws {a) on upper jaw: Outer series Second series Third series Fourth series Fifth series Sixth series ( b ) on lower jaw: Outer series Second series Third series Fourth series Fifth series Sixth series IV. Pharyngeal Teeth (a) on upper pharyneals: Each anterior pharyngeal Each posterior pharyngeal . . . . ( b ) on lower pharyngeals : Total number on both bones together. Far East T. mossanibica Transvaal T. mossambica 15 15 15 15—16 30 30—31 12 — 13 pairs 13 pairs 2 pairs 2 pairs 6 pairs (last two very 6 pairs thin) on third vertebra on third vertebra 5 16—17 15—16 1 1 (eleventh double with 11—12 small second) 27—28 25—26 3 3 10 9—10 12 11 6 4 and few more 7 7 6 4 and few more 7 7 present present 1 1 5 5 1 1 11—12 11—12 3 3 5 5 2 pairs 2 pairs 62 60—64—90 64 70—75—88 60 46—60—66 14 44—56—66 up to 43 up to 20 48 60—80—94 40 48—52—62 24 32—40—52 20 24—32—46 up to 28 up to 26 9—12 12—16 circa 300 250—300 250—300 360—400 69 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 7, DECEMBER 1970 The teeth on upper and lower jaws: The outer series is bicuspid with the mesial cusp large and the lateral cusp short and small. The more posterior teeth are unicuspid, as the lateral cusp has not yet been developed. (In many Transvaal T. mossambica many or even most of the teeth of the outer series are unicuspid and remain so in large mature specimens). Inner series have all tricuspid teeth with central cusp the longest and largest. The pharyngeal teeth are all bicuspid with large outer and very small inner cusp (according to the bent shape of the cusp part of the tooth). Only the lateral ones are small and slender. The posterior ones are the largest and stoutest. The medial and even anterior ones are fairly stout and large. In comparison with Transvaal T. mossambica the teeth are wider spaced over the pharyngeal bones and proportionally a little coarser, even the anterior ones. REMARKS From a study of the body ratios it is evident that the Australian specimens have a slightly shorter snout length. In relation to the size of the fish the eye diameter is a constant factor for both samples. Regarding the other ratios the differences are slight. These may be attributed to the differences in sizes of the samples (Australian 11 — 13 cm, Transvaal 18*7 — 22*5 cm), length of time in preservative, and the fact that the Australian specimens were aquarium bred. For the Transvaal specimens the ratio of gut length to body length was found to be from 4-4 — 8-6, thus the ratio for one Australian specimen of 5-9 fits in quite well. The ventral vertebral apophysis on the third vertebra of the Australian specimens con- forms to that found in the Transvaal specimens, in that the right and left apophysis have met distally and united to form a foramen through which the dorsal aorta passes. A short, but stout spine is formed on this arch. The retractor pharyngeal muscles and the swim-bladder are attached to this arch and spine. All the internal organs: gonads, swimbladder, liver and kidneys are the same as have been described (Petrick, 1967, unpublished manuscript) for Transvaal T. mossambica. Regardless of the slight difference in body ratios the Australian specimens are considered to represent the species Tilapia mossambica Peters. ACKNOWLEDGEMENTS The authors wish to thank Dr S. S. du Plessis, Director, Transvaal Provincial Department of Nature Conservation, for permission to publish Mr F. O. Petrick’s detailed report; Dr P. H. Greenwood of the British Museum (Natural History), London, very kindly studied this paper and provided valuable comments; Mr F. L. Farquharson of the Albany Museum prepared X-ray photographs of the Australian specimens forwarded by Dr W. J. R. Lanzing. This paper is published by courtesy of the Director of the Albany Museum, Mr C. F. Jacot-Guillarmod. REFERENCES Atz, J. W., 1954. “The peregrinating Tilapia ”, Bull. New York Zool. Soc., 57(5): 148 — 155. Dineen, C. F. & Stokely, P. S., 1956. “The Osteology of the Sacramento Perch Archoplites interruptus (Girard)”, Copeia, 4:217 — 229. Ford, E., 1937. “Vertebral variation in Teleostean fishes”, Jour. Mar. Biol. Assoc., 22. Goodrich, E. S., 1930. “ Studies on the structure and development of vertebrates. The Macmillan Company, London. Gosline, W. A., 1961. “The Perciform caudal skeleton”, Copeia, 3:267 — 270. Harder, W., 1964. “Anatomie der Fische”, Handb. Binnenfisch. Mitteleurop., II A: 1 — 308. Textteil. 70 JUBB AND PET RICK: TILAPIA MOSSAMBICA PETERS, FROM AUSTRALIA Harder, W., 1964. “Anatomic der Fische”, Handb. Binnenfisch. Mitteleurop ., II A:1 — 96. Abbildungsteil. Harrington, R. W., 1955. “The osteocranium of the American Cyprinid fish, Notropis bifrenatus, with an annotated synonomy of Teleost skull bones, Copeia, 4:267 — 290. Korschelt, E., 1940. “Ober Besonderheiten im Aufbau des Knochenfischskeletts”, Z. wiss. Zool., (A) 152:507—546. Lagler, F. K., Bardach, J. E. & Miller, R. R., 1962. Ichthyology. Wiley & Sons, New York. Nursall, J. R., 1963. “The hypurapophysis, an important element of the caudal skeleton”, Copeia, 2:458 — 9. Ramaswami, L. S., 1955. “Skeleton of cyprinoid fishes in relation to phylogenetic studies. 7: The skull and Weberian apparatus of Cyprininae (Cyprinidae)”, Acta zool. Stockh ., 36:199 — 242. Trewavas, E., 1964. “A revision of the genus Serranochromis Regan (Pisces, Cichlidae)”, Ann. Mas. roy. Afr. centr., ser. 8, Sci. Zool., 125, 1 — 58. 71 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prow Mus. (Nat. Hist) VOLUME 8 • PART 8 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA i The juvenile stadia of Anampses caeruleopunctatus Ruppell (1829) by G. J. B. ROSS Port Elizabeth Museum, Snake Park and Oceanarium ABSTRACT Anampses caeruleopunctatus has been recorded in South Africa at Cape Vidal (1 specimen) at Durban (5 specimens) and at Port Elizabeth (2 specimens). Five of these showed the un- described juvenile colour pattern. This pattern and the intermediate stages are described, with a brief discussion on distribution and habits of the species. INTRODUCTION In late February 1969, a labrid fish approximately 30 mm in length was taken in a dip-net off the Port Elizabeth harbour wall. This specimen remained unidentified until a second fish approximately 50 mm in length was taken in the same way off Humewood, Port Elizabeth two weeks later. Both specimens were kept under observation in the Port Elizabeth Oceanarium for several weeks. On the basis of photographs and drawings of the living fish, they were identified as juvenile Anampses caerulopunctatus Ruppell. In January 1970 a third specimen approximately 30 mm in length was collected at Cape Vidal, northern Natal by Mr R. van Els. This species has previously been collected in South Africa at Durban by Bell-Marley (Smith, 1946) (one adult) and four juveniles in different stages (30 — 50 mm), taken by Dr A. Wright. The strikingly different colour patterns of the juvenile, intermediate and adult forms are described with comments on the species’ distribution, and a brief discussion of observations on habits and habitat. COLOUR DESCRIPTION The juvenile colour pattern is shown by specimens up to approximately 40 mm in length. Three juveniles taken at Durban are shown in Plate \a, b and c. The 30 mm specimen from Port Elizabeth is not shown as it is identical to that in Plate I a. The Cape Vidal specimen is similar to that in Plate Ic. The colour in the living juvenile may be pale green, emerald-green or brownish-green. The centre of almost every scale is marked with a faint whitish spot or ocellus. These ocelli are regular over most of the body, becoming smaller on the breast and the belly. Radiating in all directions from the eye are several brownish-green or brown lines which are more distinct anteriorly and ventrally. Not easily visible in the colour photographs, but readily seen in the Port Elizabeth specimen, is a single bluish bar connecting each eye across the interorbital. Characteristic of the species at this stage are six white and three brown blotches on the 73 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 8, DECEMBER 1970 sides of the body. These occurred in fixed positions in all five juvenile specimens. Two white blotches occur at the level of the sixth dorsal spine, one scale row above and two rows below the lateral line respectively. The latter is adjacent and anterior to a brown blotch on the third scale row ventral to the lateral line. A single white blotch occurs at the level of the fifth dorsal ray, two scale rows ventral to the lateral line, anterior to the two brown blotches 3- and 6- scale rows ventral to the lateral line. The remaining two white blotches occur at the level of the ninth dorsal ray, on the second scale row on each side of the lateral line. At this point, the lateral line is midlateral. A smaller white blotch occurs in the dorsal midline of the caudal peduncle. This last spot is more distinct in the brownish-green specimens. The dorsal fin is pale to dark brown, with unpigmented areas between the third and sixth dorsal spines and the fifth and eighth dorsal rays. The tips of the fin are edged with white, and the base of the fin is marked with an interrupted white line on the membrane. A similar, but more variable line marks the middle of the fin. A dark brown patch occurs on the last few rays. The anal fin is darker than the dorsal, and the white markings on the edge, middle and base of the fin are more distinct. The caudal fin is predominantly unpigmented posteriorly. Anteriorly, a variable yellowish-white band partially surrounds a darker, brown patch on the caudal peduncle. The pelvic fin is dark brown, and the pectorals are unpigmented except for a brown band across the base. The change from the juvenile to the adult colour pattern occurs at a length of 40 — 50 mm, though these larger specimens do not show the full adult pattern as yet. The main changes are as follows: The colour darkens to a dark olive-green or reddish-green. The lines radiating from the eye change to a more distinct ice-blue, as do the ocelli on the scales. The white lines, spots and edges of the fins also change to ice-blue. The distinctive six white and three brown blotches are obliterated by the change in background coloration, and the unpigmented areas on the dorsal fin disappear. These are masked by a progressive spread anteriorly of the dark brown patch on the last dorsal rays. The yellowish-white band on the caudal fin becomes bluish and is moved posteriorly by the spread of the dark brown on the caudal peduncle, until the entire fin is brown with a narrow bluish tip and upper and lower edges. The pectoral fin becomes slightly bluish, and the pelvic fin develops a blue outer margin with several blue spots or stripes on the fin membrane. The intermediate form’s colour pattern shown in Plate Id may be compared with the adult specimen in Plate II. DISCUSSION Labrid fish are typically tropical, and in most areas confined to a fairly narrow equatorial belt. However, along the coast of East and South Africa, the distribution of these and other fish is considerably extended owing to the effect of the warm and powerful westward Equatorial Current, part of which flows southward as the Mozambique Current, and finally, the Agulhas Current. Consequently, tropical and, in this case, Indo-Pacific forms occasionally occur as far south as Algoa Bay and Knysna. A. caeruleopunctatus has been recorded from the Seychelles (Smith and Smith, 1963), Mauritius, and the East Indies as far as Tahiti (Fowler and Bean, 1928). It also occurs from the Red Sea down the east coast of Africa as far as Durban, which may be considered the normal limit of Indo-Pacific forms. It is almost certain then that the two Port Elizabeth specimens are strays carried several hundred miles from their normal limits by the Agulhas Current. Similar cases have been reported for juveniles of Thalassoma purpureum Forskal at 34° S, 24° E (Smith, 1957a), Chrysoblephus puniceus (Gilchrist and Thompson) at Knysna (Smith, 1943), Porcostoma dentata (Gilchrist and Thompson) at Natures Valley (Smith, M. M. 74 ROSS: JUVENILE STADIA OF ANAMPSES CAERULEOPUNCTATUS personal communication) and Anthias squamipinnis (Peters) at Port Elizabeth (personal observations) where the adults are unknown even as strays. As very little is known of the habits and habitats of these fish, the few observations made seem worthwhile recording here. Both Port Elizabeth specimens were taken at a depth of ten feet, the smaller amongst concrete blocks on the breakwater, and the larger at the base of rocks adjoining open sand. While neither of these areas was well-covered with algal growth, the coloration and swimming habits of the juvenile suggest that it normally inhabits this type of growth. The juvenile, in particular, swam with its head slightly down, with slow un- dulating body movements reminiscent of a falling leaf. It always remained within a few inches of the sides or bottom of the tank, and moved comparatively little. Subsequently, observations were made on two live fish in the field. In July 1969, a pale green specimen was seen on several occasions over the same section of coral reef at Santa Carolina (21° 30' S, 35° 20' E) in 10 feet of water. This reef, though flat, had a large number of crevices in which the fish hid when approached, and the greenish algal growth on the coral provided camouflage. A brownish- green specimen was seen at Black Rock, Northern Natal (27° 10' S, 32° 50' E) in January 1970 in six feet of water, perfectly camouflaged in a bed of Sargassum sp. This fish retreated into the weed when a diver approached within a few feet. Both of these specimens swam in a similar way to those observed in captivity. The Port Elizabeth intermediate specimen was much more active, exploring the whole tank and swimming in a manner more typical of adult labrids. A similar case of camouflage in juvenile labrids is found in the brilliantly coloured young of Coris gaimardi Quoy, which resemble pieces of broken shell when they are at rest on the bottom (Smith, 1957b). Owing to their striking colour patterns, labrid fish have been collected and described for many years and even in the comparatively poorly known western Indian Ocean, a new labrid species would be an event. It is not unknown, however, for juveniles of a known species to be given specific status (c.f., Smith, 1957b). Problems of this sort can only be solved effectively by the collection of large series of specimens or by observations of the living fish during growth. Again very little is known of sexual dimorphism in western Indian Ocean labrids. In view of the rarity of many of these fish the collection of large series or the rearing of captive fish is often difficult, and it is likely to be many years before these problems are solved. ACKNOWLEDGEMENTS I wish to thank Dr A. Wright of Durban for the use of the photographs in the colour plate, and I am especially grateful to Mrs M. M. Smith (J. L. B. Smith Institute of Ichthyology, Grahamstown) for the initial identification, the use of specimens and the photograph of the adult, and for much constructive advice and criticism in the preparation of this report. REFERENCES Fowler, H. W. & Bean, B. A., 1928. “Contributions to the biology Philippine Archipelago and adjacent regions’’, Bull. U.S. natn. Mus., 100(7) :228. Smith, J. L. B., 1943. “Juvenile stadia of S.A. fishes”. Tram. R. Soc. S. Afr., 30:49. Smith, J. L. B., 1946. “New species and new records of fishes from South Africa”, Ann. Mag. nat. Hist., (2)13:801. Smith, J. L. B., 1957a. “List of the fishes of the family Labridae in the western Indian Ocean with new records and five new species”, Ichthyol. Bull. Rhodes Univ., 7:103. Smith, J. L. B., 1957b. “The Labrid fishes of the subgenus Julis Cuvier, 1814, from South and East Africa”, Ichthyol. Bull. Rhodes Univ., 8:118. Smith, J. L. B. & Smith, M. M., 1963. The Fishes of Seychelles: 38, pi. 85 B. Department of Ichthyology, Rhodes University, Grahamstown. 75 •V.. ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 8, DECEMBER 1970 Plate I. The colour patterns of juvenile and intermediate Anampses caeruleopunctatus Ruppell, 30 — 50 mm in length. Photo: A. Wright. 76 ROSS: JUVENILE STADIA OF ANAMPSES CAERULEOPUNCTATUS 77 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 8, DECEMBER 1970 Plate II. The adult colour pattern of Anampses caeruleopunctatus Ruppell. Photo: Smith and Smith, 1963. PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov . Mus. (Nat. Hist.) » K PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA The southern limits of distribution of commercially important penaeid prawns in South Africa by D. A. HUGHES* Port Elizabeth Museum, Hume wood, Port Elizabeth Seven species of commercially important penaeid prawn are commonly found along parts of the eastern coast of South Africa. Five species are recorded in the commercial catches from the St Lucia lake system (Joubert and Davies 1966) and all seven in Durban Bay (Joubert 1965). Prior to this study five of these species (Penaeus japonicus Bate, P. canaliculatus Olivier, P. monodon Fabricius, P. indicus Milne-Edwards and Metapenaeus monoceros Fabricius, had been recorded from almost as far south as East London; three (P. japonicus, P. monodon and P. indicus ) as far south as the Swartkops River estuary, Algoa Bay, and two species (P. japonicus and P. indicus) have been recorded as far south and west as Knysna (Day et al 1952). The occurrence of shrimp in these inshore waters at what appears to be the limits of their range raises some interesting questions concerning (i) their origin (in terms of spawning and nursery sites), and (ii) the factors limiting their distribution southwards. It was hoped that a survey of potential nursery areas towards the limits of distribution, together with an examina- tion of records of occurrence in these waters would shed light on these questions. METHODS Field sampling was conducted at intervals between early October 1967 and late January 1968. The techniques used were similar to those previously used successfully in Mozambique (Hughes 1966) and in the U.S.A. (Hughes 1969). A “Discovery-type” plankton net (1 metre mouth) was used to sample for both postlarvae and juveniles moving into or away from the estuaries. Sampling was conducted during the day and at night at the entrance to the Swartkops, Keurbooms-Bitou, Piesang and Knysna river estuaries. Flood tides were sampled at all the above-mentioned estuaries and ebb tides at Swartkops and Keurbooms-Bitou river estuaries. A hand-operated dredge net was used to sample for juveniles within the estuaries. The mouth of the net was semicircular (diameter 3 ft.), and from the mouth a 6-foot-long net (18 mesh/inch, 1 mm aperture size) tapered to a point where the collecting bottle was attached. The base of the net was reinforced with canvas as protection from the substrate. The net was generally pulled by hand, but in the Knysna estuary it was, in addition, towed behind a boat. A variety of shallow-water habitats were sampled in each estuary. In the Swartkops estuary numerous prawns were collected from the grids protecting the water intake system of the Swartkops power station. This proved to be the best source of material. A number of sport and bait fishermen were questioned and a survey was made of the literature for any mention of these species. The collections of several museums were examined. * Present address: Institute of Marine Sciences, University of Miami, Florida, U.S.A. Contribution No. 1123 from the Institute of Marine Sciences, University of Miami. 79 2 0 2 5 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 9, DECEMBER 1970 Fig. 1 80 HUGHES: SOUTHERN LIMITS OF PENAEID PRAWNS IN SOUTH AFRICA RESULTS Sampling. No penaeid prawns were collected in any of the four estuaries using either of the two nets. The sampling techniques employed were similar to those used successfully in com- parable situations elsewhere, and the habitats sampled were similar to those in which prawns are usually found. It was therefore assumed that these results were a fair indication of the paucity of both juvenile prawn within these estuaries and of postlarvae entering them. The following species were taken from the grids of the Swartkops power station: P. japonicus P. latisulcatus P. canaliculatus P. monodon P. indicus 6 specimens 11 -4 — 15-0 cm (total length, as measured from the post- orbital margin to the tip of the telson). 2 specimens 12-5 and 13*4 cm. 4 specimens 12-2 — 14-8 cm. 1 specimen 13-2 cm. 13 specimens 3-0 — 5- 2 cm. Examination of Museum Material and Records of Occurrence. Very few penaeid prawns from eastern Cape waters were available in museum collections. A few were present in the collections of the Albany and Port Elizabeth museums. The Durban museum had none, while the South African Museum had acquired no further penaeid material since the publica- tion of Dr Barnard’s (1950) catalogue of the decapod Crustacea of South Africa. Of the prawns examined a number which had been labelled P. japonicus were clearly P. latisulcatus and P. canaliculatus. Only P. latisulcatus from Port Alfred extended farther southward than previously known distribution records. It had previously only been recorded from Durban (Joubert 1965). Of the museum material examined the smallest specimen was an individual of P. monodon (11-0 cm) and the largest was a $ of the same species from the Swartkops estuary (19-5 cm). The significance of this is mentioned in the discussion. In their report of an ecological survey of the Knysna estuary, Day et al (1952) record finding P. indicus and P. japonicus. They give no indication of the size of these prawns. This represents the most westerly record of distribution of a commercially important penaeid along the South African coast. As it has been shown (Hughes 1966) that juvenile prawns are most abundant in the shallow fringes of the water of “nursery areas”, it is noteworthy that Macnae (1957) does not record the presence of any penaeids in his survey of the Swartkops estuary, despite the fact that his study was limited to the intertidal zone. Van Wyk (personal communication and a series of papers in the Annual Reports of the Department of Nature Conservation) examined numerous estuaries along the east coast of South Africa. He records no penaeids south of the Bushmans River. Interviews of Fishermen. A number of sport and bait fishermen operating on the Swartkops River were questioned concerning the seasonal occurrence, abundance and size of “swimming prawns”. The information was consistent in so far as all stated that the prawns “came into” the estuaries in late summer, March and April being particularly mentioned; that they were in these months most readily available to dipnetters at night, and that it was sometimes possible to accumulate a bucket or more of them. These were described as being usually about 12 cm in length. Although individuals half this size were mentioned it was not always possible to know whether these smaller individuals were in fact penaeids or the palaemonid, Palaemon paciftcus which are most abundant. 81 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 9, DECEMBER 1970 DISCUSSION Several noteworthy points emerge from the results. With the exception of P. indicus (3 • 0 — 5 • 2 cm) the size of the prawns collected within these estuaries is invariably greater than the maximum size of penaeid prawns occurring in inshore waters elsewhere. Tabb (1962) records a maximum carapace length of 3*2 cm for P. duorarum in Florida, and Hughes (unpublished) only exceptionally found individuals greater than 9*5 cm (total length) in “nursery areas” in Mozambique. No postlarvae and virtually no small juvenile shrimp were found or have been recorded within these estuaries. If the shrimp caught by fishermen in March and April were reared within these estuaries one would expect, at the time of this sampling, to have encountered fair numbers of juveniles. From this one must conclude that to explain the nature of the occurrence of penaeid shrimp in these waters it is necessary to postulate that these adults (or sub-adults) are spawned and reared farther north and subsequently move down the coast into the southern estuaries. Macnae (1962) has pointed out the possible role played by the current systems off the eastern coasts of South Africa in effecting the distribution of marine species. He points out that the south-westerly flowing Mogambique-Agulhas current which flows strongly along the edge of the continental shelf may transport southwards the larval forms of warm water species not normally found there. He specifically mentions the larvae of penaeids. It is probably true, as the presence of juvenile P. indicus suggests that shrimp larvae are washed south but from the evidence available it appears that this current brings with it adults and sub-adults, which have probably already left their more northern nursery areas, as well as larval stages. The intervening waters between the Mogambique-Agulhas current and the coast are much colder than the Mogambique-Agulhas current itself. Macnae (1962) suggests that this barrier is overcome when the influence of the trade winds sets up eddy systems along the margins of the current, bringing the warm water closer inshore. However as one moves south- wards down the coast of South Africa the width of the continental shelf increases until off Knysna it is approximately 60 miles wide. The greater width of the intervening colder waters westward from Knysna may be the principal cause of the decrease in numbers of prawns found within estuaries south-west of Port Elizabeth and may be the barrier to their coloniza- tion of estuaries west of Knysna. It is therefore probable that the presence of almost all penaeid prawn species in the waters of the eastern Cape can be accounted for in terms of their transport southwards either immediately after spawning (as in the case of P. indicus of this survey) or at later stages as sub- adults or adults. The absence of postlarval prawns and the paucity of juveniles, found during the investigation, gives further support to Macnae’s (1962) contention that penaeid stocks along most of the southern part of the African coast are not self-perpetuating, but rely entirely on replenishment by southward moving individuals, spawned farther north. ACKNOWLEDGEMENTS This work forms a part of a research programme on South African estuaries under the direction of Dr J. R. Grindley. This work was supported by the South African Council for Scientific and Industrial Research as part of the National Programme of Oceanographic Research, and is a contribution to the International Biological Programme. I 82 HUGHES: SOUTHERN LIMITS OF PENAEID PRAWNS IN SOUTH AFRICA REFERENCES Barnard, K. H., 1950. “Descriptive catalogue of the South African decapod Crustacea”, Ann. S. Afr. Mus ., 38:1—837. Day, J. H., Millard, N. A. H. and Harrison, A. D., 1952. “The ecology of South African estuaries. Part III Knysna, a clear open estuary”, Trans. Roy. Soc. S. Afr., 33:367 — 413. Hughes, D. A., 1966. “Investigations of the ‘nursery areas' and habitat preferences of juvenile penaeid prawns in Mozambique”, J. appl. Ecol., 3:349 — 354. Hughes, D. A., 1969. “Responses to salinity change as tidal transport mechanisms of pink shrimp Penaeus duorarum”, Biol. Bull. 136(1 ):43 — 53. Joubert, L. S., 1965. “A preliminary report on the penaeid prawns of Durban bay”. South African Oceano- graphic Research Institute. Investigational Report No. 11, pp. 32. Joubert, L. S. and Davies, D. H., 1966. “The penaeid prawns of the St. Lucia lake system”, Oceanographic Research Institute. Investigational Report No. 13, pp. 40. Macnae, W., 1957. “The ecology of the plants and animals in the intertidal regions of the Swartkops estuary near Port Elizabeth, South Africa”, Parts I and II. J. Ecol., 45:113 — 131 and 361 — 387. Macnae, W., 1962. “The fauna and flora of the eastern coasts of Southern Africa in relation to ocean cur- rents”, 5. Afr. J. Sci., 58(7) :208— 212. Tabb, Durbin D., David L. Du brow and Andrew E. Jones, 1962. “Studies on the biology of the pink shrimp, Penaeus duorarum Burkenroad, in Everglades National Park, Florida”, Fla. St. Bd. Cons., Univ. Miami Mar. Lab., Techn. Ser., 37:1 — 30. Whitmore, J. S., 1967. “Effects of catchment management on river flow characteristics”, Abstract. Pretoria Congress of the South African Association for the Advancement of Science. Supplement to S. Afr. J. Sci.: July 1967. 83 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mas. {Nat. Hist.) K VOLUME 8 • PART 10 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA .3 A new freshwater gobi (Pisces: Gobiidae) from Lake Sibayi, Zululand, South Africa F. L. FARQUHARSON Albany Museum Lake Sibayi, some 40,000 hectares in extent, is situated on the coastal plain of Zululand and is separated from the sea by forested dunes rising to a height of 180 metres. It has a depth of 40 metres, however, the surface is 22 metres above sea level. Although the lake is at present a freshwater system, past association with the sea is evinced by the marine affinities of much of its fauna. (Allanson et al , 1966). Three gobies have been found in the lake: the ubiquitous Glossogobius giuris (Ham.- Buch.); Croilia mossambica Smith, a new locality record; and a species of Silhouetted Smith, 1959, here described as new. Silhouettea sibayi sp.n. (Figure 1). Diagnosis. Both sexes may be distinguished from Silhouettea insinuans Smith by the naked breast. In addition, the males are characterized by the height of the dorsal fin, in excess of 2 times the body depth. Holotype: male, 27 mm standard length, from eastern shore of Lake Sibayi (27°21'S, 32°47'E) 18.1.1966, Albany Museum No. P.F. 1100 (collectors No. SIB 13B). Paratypes: males, 23 mm s.l., collected together with holotype, A.M. No. P.F. 1101; female, 20 mm s.l., 15.7.1965, A.M. No. P.F. 1102 (Coll. No. SIB 8 Q); female, 20 mm s.l., 14.7.1967, A.M. No. P.F. 1103 (Coll. No. SIB 1 1 B); juvenile, 12 mm s.l., 20.1.1966, A.M. No. P.F. 1104 (Coll. No. SIB 26Gb). Additional Specimens. Colour notes were made from four additional specimens (died on route to Grahamstown, suffered delayed preservation and therefore excluded from type series) collected 20.7.1968, A.M. No. P.F. 1105 (Coll. No. SIB 104A). Description. (Paratypes in parenthesis). Head broad, naked, somewhat depressed, 3-3. (3-1 — 3-5) in standard length. Eyes large, dorsal, adjacent, 4-3 (3-4 — 4-0) in head, exceed snout. Anterior nostril a short raised tube above lip, posterior nostril a simple pore against eye. Sensory pores and papillae of head as illustrated. Mouth terminal, oblique, reaching to below anterior margin of eye, lips pronounced, lower jaw protruding, tongue truncate. Teeth on lower jaw recurved, in inner and outer series, separated by band of minute teeth, outer series with enlarged canine-like teeth at sides of jaw (females lacking canine-like teeth), teeth on upper jaw smaller, similarly arranged. Gill-opening to below mid-operculum. Gillrakers low knobs, 1+1 + 10. Dorsal VI + I 11 (juvenile V + I 11). First dorsal: height 2-4 (other male 2-2) times body depth sub-equal to snout-to-dorsal distance, first spine normal, sub-equal to body depth, 2nd, 3rd & 4th spines filamentous, tending to run together distally giving the appearance of a single filament, 2nd longest (in one additional specimen, 3rd is longest), 5th and 6th spines progressively shorter. (In females, first dorsal height 1-1 times body depth, 1st spine 1-4 in body depth, 2nd, 3rd & 4th spines normal sub-equal, 5th & 6th spines progressively shorter.) Second dorsal separate, rays sub-equal, 1 - 2 (1 *5 — 1 -2) in body depth. 85 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 10, DECEMBER 1970 Anal I 13, inserted slightly in advance of the second dorsal, rays sub-equal 1-3(1- 6 — 1 * 3) in body depth. Pectoral 16 (15 — 16), 1-4 (1-2 — 1-3) times body depth, reaches beyond anal origin. Pelvics fused, 6 + 6, 1-4 (1*3 — 1-4) times body depth, reach beyond anal origin, fraenum strong. Caudal rounded, shorter than head. Scales ctenoid, 29 (28 — 29, undeveloped in juvenile), transverse 8, caudal peduncle 12, head, nape, breast and pectoral base naked. Coloration: in life cryptic, pale opalescent below, above flecked with reds and browns >j matching the sandy substrate. Caudal and second dorsal transparent but similarly flecked. Behind the pectoral, four or five narrow, transverse rows of spots across body. Pectorals i transparent, pelvics similar but with darkly pigmented base. Anal opalescent with darkly /I pigmented distal margin. The first dorsal in males is very darkly pigmented with a bright l| silvery-white sub-marginal stripe on the lower posterior half of the fin. First dorsal in females > similar to the second dorsal, not distinctive as in males. In preserved specimens colours fade i but darkly pigmented areas remain, as well as silvery-white stripe of males. Habits . Collected on the sandy bottom from shallows to a depth of 5 metres, but silvery flashes of males were observed down to 1 1 metres. Both in aquaria and on the lake bottom, specimens were observed to bury themselves in the sand with the aid of the pectorals, the eyes remaining uncovered. In this position the dorsal fin may be erected, perhaps serving to attract prey. Remarks . This new species runs down to Silhouettea Smith in Smith’s (1959) key to the sub-family Gobiinae in the western Indian Ocean. It does not run out in his later (1960) key to the Gobiidae in South Africa (from which Silhouettea is excluded). 86 FARQUHARSON: A NEW FRESHWATER GOBI FROM LAKE SIBAYJ ACKNOWLEDGEMENTS. All the specimens examined were obtained by Rhodes University Ecological Survey, and I am indebted to Professor B. R. Allanson for the opportunity and facilities provided in visiting the lake, and also to Mr R. E. Boltt and his colleagues, who, using SCUBA gear, obtained live specimens for observation. REFERENCES Allanson, B. R., Hill, B. J., Boltt, R. E. & Schultz, V., 1966. “Estuarine Fauna in a Freshwater Lake in South Africa,” Nature. London. 209(5022) :532 — 533. Smith, J. L. B., 1959. “Gobioid Fishes of the families Gobiidae, Periophthalmidae, Trypauchenidae, Tae- nioidae, and Kraemeriidae of the Western Indian Ocean”, Ichthyol. Bull. Grahamstown. No. 13. Smith, J. L. B., 1960. “Fishes of the family Gobiidae in South Africa”, Ichthyol. Bull. Grahamstown. No. 18. 87 ( © PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mas. (Nat. Hist.) i PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA The Acheulian occupation at Amanzi Springs Uitenhage district, Cape Province H. J. DEACON (Albany Museum, Grahamstown, South Africa) “There was a secret and a fascination in the mysterious ‘craters’ that appealed to the adventurous. The rugged ironstone throats, choked with treacherous black mud or with blue and yellow clays, whence came the ooze and the hot mineral springs and the faint smell of sulphur, might lead anywhere and mean anything”. Sir Percy FitzPatrick, “Amanzi: A private record of the first decade”, circa 1924. INDEX TO FIGURES Fig. No. Page 1 Location Map 121 2 Amanzi Hill 122 3 Plan of the spring centers 123 4 Plan of Area 1 124 5 Plan of Area 2 125 6 Section : Area 1 Facing 1 26 7 Section Area 1, Cutting I 126 8 Plan of 1963 Excavation: Area 1, Cutting 1 127 9 Plan of 1964-5 Excavation: Area 1, Cutting 1 128 10 Plot of finds in frontal section: Area 1, Cutting 1 129 11 Plot of wood occurrence: Area 1, Cutting 1 130 12 Plan and Section: Area 1, Cutting 1 extension ........... 131 13 Section: Area 2 Facing 132 14 Section across deep sounding : Area 2 132 15 Location of samples 1 — 3 : Area 1, Cutting 10 133 16 Distribution of lithic material in depth: Area 1, Cutting 10 . ....... 134 17 Bar charts of finds: Area 1, Cutting 10 135 18 Bar charts of finds: Area 2, Surfaces 1 — 3 ............ 136 19 Bar charts of finds: Area 2, Cuttings 5, 6 and 8 137 20 Plot of finds: Area 2, Cutting 6 Facing 138 89 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 INDEX TO PLATES Plate No. Page 1 Amanzi Hill: Position of excavation 138 2 Area 1: Cutting 1 and 10 139 3 Area 1: Cutting 10, North wall 140 4 Area 1: Cutting 10, Frontal Portion (6 — 15/15 — 30) 141 5 Area 1 : Wood in situ , Cutting 1 Extension . 142 6 Area 1 : Wood from Cutting 1 Extension 143 7 Area 1: Block 10 — 15/35 — 40: Artefacts in situ 144 8 Area 2: General view of excavation cultural material on Surface 1 145 9 Area 2: Deep Sounding: Wall of cutting 146 10 Area 2: Deep Sounding: Surface 3 147 11 Area 2: Section along furrow 148 12 Area 2: Wood and artefacts in situ , Cutting 6 ........ 149 13 Photomicrographs of 4 transverse sections of wood ....... 150 14 — 15 Macroscopic plant remains and wood 151-153 16—17 Cores. 153-154 18 — 19 Cobble tools 155-156 20 Half cobble — handaxe technique ............. 157 21 — -22 Small handaxes 158-159 23 — 29 Handaxes 160-166 30—35 Cleavers 167-172 36 — 41 Large bifaces 173-178 42—43 Other tools 179-180 44 — 52 Flake tools 181-189 1. Introduction The Acheulian site is on the north flank of a hill rising to 610' above sea level and over- looking the incised valley of the Coega River in the Uitenhage District (25° 29' E., 35° 42' S.). The northern portion of the hill and the slopes below which include the total extent of the spring deposits with which the Acheulian occupation is associated, are on the farm Amanzi Estates, formerly Balmoral and originally Rietheuvel (Uit. Q. 1 .41). The occurrence of artefacts in the spring deposits was recorded as early as 1924 in a private report (FitzPatrick, 1924) and their discovery was the result of the development of the springs for an irrigation scheme during the previous decade. However, it was not until 1963 that a preliminary archaeological investigation of the springs was carried out by Inskeep and a report published (Inskeep, 1965). This present paper outlines the results of more extensive excavations made in 1964 and 1965 as a follow up on the findings of Inskeep’s investigation. The main result of the 1963 excavation showed a series of spring deposits relating to two separate phases of accumulation with a quantity of “Early Stone Age” material included in the lower deposits. In these same lower deposits finds of reasonably well preserved wood were made, pointing to unusual conditions for preservation. 90 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS The sealed context of the cultural material and the preservation of wood were factors of sufficient importance to warrant further investigation. The preliminary excavation was too limited to establish the extent of the occupation, to indicate whether there was any relationship between the wood found and the cultural material, to allow firm conclusions as to the mode of occurrence of the cultural material and its typology or to give a detailed picture of the structure of the deposits. These then were the objects of the follow up excavations. Field work was carried out over two seasons, August to November, 1964, and from September 1965 to January 1966. 2. The Springs A number of springs occur in the general region but with the exception of the Amanzi springs these are “pure” water springs (Bond, 1946, “Type E”; du Toit, 1928, p. 38) issuing from the margins of the Uitenhage artesian basin (vide Haughton, 1928; Engelbrecht et al.t 1962) with waters characteristic of the underground waters in the Cape System. They are not associated with large scale deposition. The Amanzi waters are more concentrated, thermal (32°C.) and chalybeate (Hall, 1938, p. 495; sample no. 3549) and are related in composition to bore-hole waters drawn from the base of the Cretaceous System (Kent, 1950 p. 252; Rogers, 1909; Engelbrecht et al., 1962, p. 41). These waters probably derive from the basal, arenaceous, pyrite rich zone, fed by fissures in the underlying Cape System, a conclusion supported by du Toit (1928) and Kent (1950). From the occurrence of the springs, it is clear that during the Middle-Upper Pleistocene and Recent times the hydrostatic pressure in the artesian basin was sufficient to force the waters up through the flanking Variegated Marl (Cretaceous) on the margin of the hill. The occurrence at this locality is probably explained by a complex interplay of a number of factors, including the pre-Cretaceous topography, the location of deep seated fissures and shatter zones in the underlying Table Mountain Series quartzites (Cape System) and perhaps not least, the proximity to the Uitenhage or Coega Fault, a major fault in the artesian basin. The Coega fault with a throw of some 3,800 feet to the south west (Marais 1964) is the boundary of the Coega Compartment, a discrete hydraulic unit in the basin. Bore-hole records (J. P. M. Niven, pers. comm.) show the quartzites of the core of the hill to form a cliff-like margin on the north side, but the contributory importance of the other factors is more difficult to evaluate. That conditions here are unusual is supported by the fact that springs of the Amanzi type are not repeated on the same scale elsewhere in the artesian basin. The springs have ceased to flow as the result of bore-hole pumping in the basin and the water table is now some 60 feet below the natural outlets on the hill. Originally the waters issued from a number of centers located in a cluster on a shoulder of the hill with two other centers at the foot of the hill. Spring deposits cover this whole hill side over an area some half a mile by half a mile and to a maximum thickness, indicated by bore-hole records, of about 22 feet. There is thus a considerable volume of deposit worked by the spring action and re-distributed by hill wash. As far as can be gauged by included cultural material from the spring centers that have been excavated or disturbed by the cutting of irrigation furrows, all are of similar age. The deposits are referred to the Amanzi Springs Formation. The dating of the spring deposits at present depends on the included cultural material and not vice versa. Although a time sequence in the cluster of springs on the north flank is probable as they are unlikely to have been simultaneously formed and active, the chronology is not fine enough to detect this and again not all the spring depressions have been excavated. On a slope such as this a tendency for the active focus to migrate up slope in time might be expected and the higher centers in the cluster may prove to be younger. On the exposures available it is impossible to correlate the spring deposits with the terrace deposits in the Coega valley with assurance. However the spring deposits, occurring on a level below that 91 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 TABLE 1 Formation Member Sub-Units Cultural Associations Dating AMANZI SPRINGS Balmoral Member Poorly sorted sands with con- siderable lateral variation. In- cludes well stratified facies, grey mottled sediments and channel fill deposits. Unknown 31,000 ± 1,200, 1,100 years B.P. FORMATION Rietheuvel Member Enqhura Member Grey black silts Brown humic sands Marginal clays; White sands; Basal clays. Acheulian Acheulian 60,600 ± 1,100 years B.P. Provisional nomenclature adopted for the spring deposits. of the end-Tertiary peneplain, date to the Pleistocene and the inception of the present valley system. Two Pleistocene terraces are recognized in the Coega Valley, terraces 1 and 2 (Engelbrecht et a/., 1962), and there is some evidence to suggest the spring deposits in the main are younger than these aggradations. Terrace 2 is a well marked feature some 50 feet above the present incised Coega River and consists of poorly sorted coarse gravels. The spring centers occur as horse-shoe shaped or circular depressions. However, a number have been modified by banks and furrows in modern times for irrigation purposes. They appear consistent with the structure of typical mound springs occupying spring eroded hollows, in this instance in the soft Cretaceous sediments. The horse-shoe form appears to be a reflection of slope, the open end being the down slope or outlet end. The spring deposits consist of a series of sands, silts and clays with great lithological variation resulting from the sorting and transporting action of the waters. The build up of the deposits within the depressions takes the form of a lens or mound-like core of sands washed free of the finer clay and silt grades and flanked by sands, silts and clays on the margins. The longitudinal section of the horse-shoe depressions is complicated by regrading of the outlets. There is ample demonstration in the cuttings and exposures for localized and widespread natural disturbance of the deposits. Stringers of sand form upshoots into overlying deposits and in places structures resembling flow structures are the result of actual movement and churning. Probably the most important feature of this nature noted is a disconformity which is of major proportions. It was seen in the initial cutting made by Inskeep and traced in two adjacent depressions in the subsequent work. These centers show clear evidence of being re-activated following a period of quiescence when they became choked with vegetation and silted up. Radiocarbon dating suggests this second phase of activity took place in the Upper Pleistocene. The chemical deposition of iron is marked in the channels leading off bore-hole waters from the Edwards bore-hole within the area of the main cluster of spring centers and the spring waters are recorded to have had a similarly high iron content (Hall, 1938). The develop- ment of ironstone or iron-cemented horizons in the spring area is thus easily explained. In some areas, a hard sheet of irreversibly dehydrated iron oxides has formed as a lateritic- like surface deposit. Iron has also been deposited in concretionary form occurring in medium to large sized blocks in the deposits or again as zones of ferruginization within the sequence. 92 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Where ferruginization is marked, all macroscopic structures have been destroyed. The iron- stones do not relate directly to the black boggy earth described by Barrow (1801) as suggested by Inskeep (op. cit.) this black earth being more reasonably correlated with the organic rich silts that formed at the springs during periods of diminished activity. In the present cycle of hill slope development, the sheets of iron-cemented material are a major factor in inhibiting the erosion of the unconsolidated spring deposits. In view of the artesian origin of the spring waters, it is perhaps tempting to relate the periodicity of spring activity reflected in the spring deposits to long term fluctuations in the intake of water into the basin and thus to regional precipitation (see also Inskeep, op. cit. p. 242). In this way the springs may evidence pluvial conditions. Some support may derive from the botanical evidence of the macroscopic plant remains (Wells, 1970) that suggests the earlier main phase of spring activity was associated with a vegetation more mesic than developed at present. This, however, relates specifically to the micro-environment of the springs and at present there is no direct evidence such conditions were regional. Periodic activity at any one center or group of spring centers may be explained simply by a shift in the main outlet. This can result from openings in the aquifer collapsing with the escape of water, effecting a decrease in pressure in time and possible cessation of discharge. It may also be linked to flush flow phenomena (Dixey, 1966) that are due to the elastic effect of the aquifer. The concept of flush flows is useful in explaining the initial excavation of the spring depressions, thereafter sedimentation and the regrading of sediments becoming the dominant processes. One feature of this artesian system that is evident from the modern bore-hole records (vide: Engelbrecht et al., op. cit.) is the limited capacity of the reservoir and new bore-holes on the farm Sutton Vallance some ten miles from Amanzi caused a marked diminution in the discharge from the Amanzi bore-holes within a week. For the present any correlation between spring activity and regional precipitation must be regarded as tentative. To some degree cycles of activity and quiescence can be regarded as normal to a spring complex. Even the best evidence for changed conditions at the site, which is in the humification of organic material in a prominent silt bed indicating a lowering of the water table prior to the second phase of activity, is capable of a non-climatic explanation. The present day vegetation of the site and the surrounds is typical Addo Bushveld (Acocks Type 23d (1); Acocks, 1953, p. 82), a special form of sub-tropical valley bushveld developed in the coastal valleys in the eastern Cape and Natal. This is a climax vegetation of the drier valleys (Amanzi, average rainfall 15*1" p.a.) and it grades by easy transition into the dry coastal forest vegetation represented in this region by a relict area at Alexandria (Acocks, op. cit ., p. 31). On the mountainous margins of the Uitenhage Basin, the original woody cover has been replaced in modern times by a False Macchia and some extension of the Addo Bushveld area may have taken place. As relief exerts a strong control on the distribution of rainfall and vegetation, the relationship of drier valleys to wetter uplands would have remained through the Pleistocene. Conditions may have alternately favoured the invasion of forest or Karoid species in the Bushveld zone but its general character has probably been maintained over a long time period. 3. The Excavation In all thirteen cuttings and seven pits were made in the spring deposits on the north flank of the hill. Some of these were primarily for stratigraphic information and the main excavations of archaeological interest were located at two adjacent spring centers, labelled Area 1 and Area 2 (Fig. 3). Area 1 included the original cutting of Inskeep which was deepened (Cutting 1) and extended (Squares 1 and 2 and Cutting 10) (Fig. 4). From an irrigation furrow in the floor of the other depression, Area 2, a number of Acheulian artefacts had been discarded on a dump and this site was chosen for excavation to amplify the results from Area 1 (Fig. 5). 93 ANN. CAPE PROV. MUS. (NAT, HIS.) VOL. 8, PT 11, DECEMBER 1970 As far as could be seen, the irrigation furrow was the only modern disturbance Area 2 had suffered. These are not the only spring centers with known associations of cultural material. Artefacts can be seen eroding out of the deposits in the spring center immediately down slope of Area 2, the sinking of the pump sump of the Edwards bore-hole produced further finds and a number of dumps from furrows show some disturbed artefact material. There is thus ample scope for further excavation should this be warranted at some future time. Area 1 This depression was used at one time as a reservoir and the softer deposits in the floor were scraped out to deepen it and the outlet end blocked off. The original form was horse- shoe shaped. Iron cemented sediments and a compact grey black silt bed limited the scraping operations on the sides of the depression and in places in the floor a tacky clayey sand was likewise a restricting factor. The excavation of Cutting 1 was into this artificial bank of the depression exposing the truncated upper deposits lying against the sloping surface of the main disconformity. The bulk of the deposits relating to the second phase of spring activity had thus been removed over the center of the depression. Exposed in Cutting 12, Square 1 and 2 and the deep sounding of Cutting 1 was a clean white sand, the basal member of the sequence (Figs. 6 and 7). The margin of this sand body dipped steeply beneath flanking sediments in Cutting 1 where it was traced and a similar dip is inferred from exposures in Cuttings 11 and 12. The structure of the sand body is possibly a lens or mound. Because of the unconsolidated nature of the white sands, excavation within them was restricted. Limited cultural material occurred in and on the sands in Square 2 as well as isolated pieces of wood and the stems of a buried herbaceous plant community in botanical context. The burial of the herbaceous community which could be seen rooted in clayey partings in one exposure in Square 2, evidences rapid deposition locally (Plate 14). The only diagnostic cultural material found in Square 2 was a group of large tools that included a handaxe, a cleaver and two narrow bladed cleavers (Plate 35) lying on the surface of the sands and covered by overlying deposits of the spring succession. As the probability of such a group occurring by chance at this site is very low, cultural association is an acceptable assumption for these tools. A brown sand overlies the white sands in Squares 1 and 2 and in Cutting 1 and it includes abundant stems of the herbaceous plant community. In part these stems form a matted mass and a woody plant with a marked pith is another important constituent of the remains in this horizon (Plate 13). Away from the center of the depression seen in Cuttings 1 and 10, the brown sands grade into a light greenish clayey sand in which there are no plant remains. This grada- tional change may approximate closely to the outer line of the fringing vegetation around the spring center although aeration and oxidation in the greenish clayey sands have been obvious factors in differential preservation. The mode of deposition of the brown plant rich sands is indicated by two thin black clay bands traced in Cutting 1. While not extending over any great area and thus of limited value as markers, they show the brown and greenish sand facies to have been deposited with a low dip away from the center of the spring (Fig. 7). Although wood remains occur throughout the brown sands, near the top of this deposit and between the two black clay bands noted above, a wood rich zone of localized extent was traced in this excavation (Fig. 1 1). This wood occurrence was intersected in the 1963 excavation and a sample for dating was collected from it. One object of the present investigation was to establish the nature of this wood occurrence and its relationship if any to the cultural materiah In the limited exposure of the original cutting (Inskeep op. cit .) it was not apparent that the deposits in this section were truncated by the disconformity with pot-holing down into the level of the brown sands at the wood rich zone. Some cultural material had been redeposited 94 DEACON: THE ACHEULTAN OCCUPATION AT AMANZI SPRINGS in the pot hole fill and with churning up of the interface between the fill and the brown sands, this part of the cutting was unsatisfactory for testing any associations. A cutting into the west wall. Cutting 1 extension, however, proved adequate in this respect and here the wood was found to be lying on a definite surface. The only cultural material found on the same surface between clay band markers was a single untrimmed flake. No cultural material occurred below the wood rich zone in Cutting 1 extension (Plate 5). No direct relationship between the wood occurrence and the cultural material can be suggested and it is evident that the wood is a natural accumulation. Root bases with a diameter of up to 12" and small twigs 1" in diameter are included in the wood sample recovered and these represent the more resistent or easily transported elements of woody vegetation washed or blown from the immediate surrounds and accumulated by drift against the herbaceous plant community fringing the vent of the spring. While there is disturbance of the brown sands and greenish clayey sands where truncated there is no evidence of more widespread disturbance in Cuttings 1 and 10 in this horizon. The black clay bands support this. This is important in that the main artefact sample from Area 1 was excavated from the top of this horizon and the contact with the overlying deposit. The base of the artefact occurrence in Cutting 1 is the wood zone and in Cutting 10 at an equivalent depth in the greenish clayey sands there is a diminution in finds. The deep sounding in Cutting 1 provided a test of the lower and apparently sterile part of this horizon. The artefacts are thus distributed through a thickness of some four feet. It is considered that under the conditions of deposition at the site, material in the top of the white sands as found in Square 2 is possibly of equivalent age. TABLE 2 AMANZI SPRINGS FORMATION; AREA 1, CUTTING 10. GRADING ANALYSIS SAMPLE No. AMS 31 AMS 30 Location 10—15/30—35 10—15/30—35 Height below datum — 1 1 feet — 7 feet Cultural relationship .... Base of artefact zone Top of artefact zone Grading Analysis Sand (0—40) 58 % Sand 16% Silt (4—80) 28 % Silt 62% Clay (8—12*) 14% Clay 22% Field Description Light greenish sand Grey Black Silt Rietheuvel Member Rietheuvel Member The overlying grey black silts (Plates 3 and 4) are developed to a thickness of more than 5 feet in the north wall of Cutting 10. The dark colour is due to humified organic matter which is readily leached in alkali. Apart from a few flecks of carbonized stems, no macroscopic plant material is preserved. The contact with the underlying sands is locally sharp but more frequent- ly gradational on colour and in places there is some interdigitation. The deposits are conform- able and do not evidence a break in deposition. In excavation no distinction could be made between cultural material in the base of the silts and the top of the underlying sands because of the nature of the contact. Although now a compact hard deposit, the original form of the grey black silts would have been a soft organic earth and they would not have formed a very effective seal over the artefact zone they blanket. This offers an explanation for a few cultural finds (Plate 52) which on typological grounds must be considered as intrusives into the main Early Stone Age artefact sample that occur in the top of the artefact zone in Areas 1 and 2. The contamination 95 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 in Cutting 10, for example, is not localized and could not be explained by root penetration or similar mechanism. However, the number of intrusives is apparently low and they do not invalidate the conclusions reached in the analysis of the artefacts. The grey black silts at the later stage formed a sufficiently cohesive body to be smoothly truncated by the erosional surface (Plate 4) on which a few cobbles and pebbles were found. The intrusives are not of a later date than the compaction and humification of the original organic earth and the re- shaping of the original deposit is good evidence of a dry period at the end of the earlier phase of spring activity. The brown sands and grey silts are referred to the Rietheuvel Member of the Amanzi Springs Formation. The disconformity marks what is evidently the re-activation of the spring as renewed deposition gave rise to a different series of deposits. These sediments of the Balmoral Member are poorly sorted sands. The lower horizon of the member is referred to as the “pothole fill” which is unbedded and shows some flow-like structures. These deposits occupy irregularities in the unconformable surface and include derived cultural material. A series of conformable, bedded, fine sediments overlie the pothole fill and show a shallow dip outwards from the center of the depression. They are intercalated with bands of what have been termed “mottled grey sediments” in the field. In Cuttings 1, 10, 11 and 12, the development of concretions or sheets of ironstone in the younger sediments has been pronounced. The “coals” and hearth mentioned by Inskeep {op. cit ., p. 236) in Cutting 1 are features related to concretion development. Little organic macroscopic material is preserved in these sediments, although by systematically breaking open clods of sediment a small sample of leaf fragments was obtained. A single piece of wood shrunk within its original cavity from the layered sediments in Cutting 10, provided the only sample for dating (1-2241). A few small flakes were the only cultural finds made, but the sample is too limited to have any typological significance. From the succession, a reasonable picture of the history of the spring can be deduced. The main phase of activity of the spring resulted in the build-up of a central core or mound of sands with flanking sands and clayey sands anchored by vegetation. This was followed by a period of diminished activity possibly as the discharge reached a steady state and the humi- fication of the grey black silt bed evidences a drop in the water-table and a period of quiescence. At a later stage the spring depression was re-excavated and activity marked by the deposition of poorly sorted sediments. The Acheulian occupation occurred during the final stages of the initial activity or build-up of the spring deposits. Area 2 Modern disturbance in this depression is limited to a silted-up furrow following an irregu- lar course up one side for the whole length, a branch of this crossing in the center to the op- posite side, and an old water hole, also silted up, of which there is no local knowledge. The stratigraphy and structure of the deposits is very similar to that described for Area 1. Test pits on the hill slope above (Fig. 3, Pit no’s 1, 2 and 5) show this depression to be cut into Tertiary-Cretaceous sediments, although bedrock was not reached in the excavations within the depression. The initial excavation in Area 2 was Square 3 which was stopped in a clayey horizon that overlies white sands. The sands were intersected in a pit subsequently dug in the floor of Square 3 and were shown to dip steeply on this edge of the depression. The surface of the sands was exposed in a trench including Cuttings 6 and 7 extending across the depression, and Cutting 5 exposed the flanking deposits on the opposite side of the depression. These exca- vations gave a section across the full width of Area 2. Cuttings 8 and 9 were started with the object of obtaining a longitudinal section of the deposits; however, the quantity of cultural material found on the surface of the white sands in Cuttings 6 and 8 made it necessary to extend Cutting 8 as an area between the furrow and Cutting 6. Extension of Cutting 8 beyond 96 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS the furrow was restricted by an ironstone sheet. Finally a deep sounding was sunk in the white sand body and other cuttings were made for stratigraphic controls (Figs. 5 and 13). Correlations have been made between the deposits in Area I and Area 2 on the basis of lithology. In Area 2 the white sands in the center of the depression are underlain by basal clays in Cutting 5 and in the Deep Sounding. The marginal clays in Square 3 and Cutting 7 overlie the white sands and form a bank against which the deposit of the Rietheuvel Member were laid down unconformably. The recognition of this disconformity is important as it is one line of evidence supporting the distinction drawn between the basal Enqhura Member and the Rietheuvel Member. Differences in the lithology of the two units are also marked (Butzer, in preparation). The basal clays are light yellow to green with sand stringers or inclusions. As they appear to extend beneath the edge of the white sand body in Cutting 5, they pre-date at least the later build up of portions of the sand core. The clays where tested in a cutting below the irrigation furrow in Cutting 5 were sterile and here as in further exposures in the furrow 60 feet from Cutting 5 (Plate 1 1) the Acheulian artefacts were found on top of the clays and in the over- lying deposit. The white sands are more variable than the equivalent horizon in Area 1 and include bands or discrete segregations of clayey material within the sand body. Ferruginization is locally pronounced. This facilitated excavation within the sands and the Deep Sounding intersected two sloping disconformable surfaces on which natural stone and artefacts were found (Plates 9 and 10). For convenience the surface of the sand body which shows a number of shallow irregularities is labelled Surface 1 and the disconformities within the sands, Surfaces 2 and 3. The latter surfaces (Fig. 13 and Plate 14) dip towards the outlet and in view of the concentration of material on the surfaces they may represent a channel or line of wash. The surfaces slope up towards Cutting 5 although they could not be traced in this cutting and hence the relationship between the artefacts in Cutting 5 and on Surfaces 2 and 3 is not conclusively established. A possible interpretation is given in Fig. 13 which suggests these surfaces were formed by reworking of the sands at the onset of conditions leading to the deposition of the sediments of the Rietheuvel Member. On this basis the artefacts on Surfaces 1 — 3 would be of equivalent age to the artefact accumulation in Cutting 5. A vent structure, defined as a vertical passage for ascending waters, cuts through the sands in Cutting 6 and dates to the age of Surface 1 as it is seen to truncate Surfaces 2 and 3 in the Deep Sounding. There are upshoots of sands injected into the overlying deposits and these mark the terminal activity after the blocking up of the vent by the brown sands. This is the only example of this type of structure re- encountered during the excavation. There is some concentration of artefacts around the vent structure on Surface 1 and wood from this locality included one piece in which the carapace of a woodboring beetle was found. Finds of macroscopic plant remains in the white sands were limited but a sample of round seeds were recovered from Surface 3 {vide Wells 1970) and wood from between Surfaces 1 and 2 (Plate 15). Individual stems of the same herbaceous aquatic plant as found in Area 1 were noted. As in Area 1, the sub-units of the Rietheuvel Member are brown sands and grey black silts. The brown sands are of variable thickness and wedge out to a thin band in Cuttings 7 and 8, being best developed in Cutting 6. A natural channel (Balmoral Member) cuts through the grey black silts and brown sands to the surface of the white sands at the junction of Cuttings 5 and 6. The brown sands have not been completely stripped off the bottom of the channel and they can be traced from Cutting 5 across Surface 1. The overlying grey black silts are typically developed in Cutting 5 however over Surface 1 they are apparently represented by a less compacted dark organic rich deposit of similar character (B1 of Fig. 13). The grey black silts (B) and the dark organic rich deposit (B1) are lithologically similar albeit the latter shows an increase in organic material from 2% to 8% over the former (Butzer pers. comm.). Artefacts occur on the basal clays in Cutting 5, on the white sands in Cuttings 6, 7 and 8 97 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 and in the overlying brown sands. As in Area 1 the top of the artefact zone is the base of the grey black silts and the contact between these silts and the brown sands is conformable. The Cutting 5 occurrence has a thickness of some two feet. The scatter over the white sands includes some concentration of artefacts within irregularities on this surface and this material and that in the overlying brown sands has been labelled as from Surface 1. For discussion purposes a distinction is made between the artefact samples drawn from Surfaces 1, 2 and 3 and Cutting 5. The artefact density is considerably higher in Cutting 5. Three sections cleaned out along the existing irrigation furrow between Cutting 5 and the mouth of the depression showed the sediments of the Balmoral Member to occupy a series of irregular washouts in the earlier deposits. They appear to represent sedimentation associated with wash across the floor of the depression from the most recent and more limited phase of spring activity. There is no evidence of any occupation associated with these youngest deposits. The channel-like structure that forms a break in the deposits and the artefact scatter between Cuttings 5 and 6 bottoms on the white sands. Stringers of brown sands have been caught up in the channel fill and there is local displacement of artefacts upwards. The structure is the result of waters escaping from the sand aquifer or core of the spring at this point. A series of mottled grey sediments are developed in Cutting 8 W 1 and 2 (Fig. 5) and a clayey facies in Square 3. A very similar history of spring activity to that deduced for Area 1 can be suggested in Area 2. This may indicate the springs were simultaneously active and perhaps linked through the same fissure system. The sequence in Area 2 is more condensed and perhaps more complex. 4. The Lithic Industry General Stone artefacts are the only evidence of human occupation of the Amanzi Springs in prehistoric times. No certain artefacts other than in stone have been found, although wood and other botanical remains are preserved in sections of the deposit. The latter, together with the detailed picture of the stratigraphy of the deposit, contribute to a knowledge of the setting in which the occupation took place, but in large part what can be deduced of the occupants rests on the analysis of the sample of lithic cultural material. All lithic material from the excavation has been kept, including natural stone occurring in association with the artefacts. This has proved of value in the investigation of this site where the primary context of finds cannot be assumed. The excavations were directed at those areas which were expected to produce evidence of the earlier occupation of Amanzi Hill and the main samples are referable to the Acheulian Culture Complex. They include a range of arte- facts : handaxes, cleavers and other large bifacial tools, trimmed flakes, some retouched as scrapers, anvils, tools on cobbles, and other material that could be described as artefact waste, cores, untrimmed flakes and irregular pieces — indeed the range of tools and waste products that would be expected on or in the surrounds of an occupation site. While the industry as a whole has a relatively heavy and unstandardized appearance, there is no evidence contrary to the assumption that all the cultural material relates broadly to an advanced Acheulian industry. The limited sample of later cultural material excavated from the site is not discussed here. There is no extensive later occupation at the two springs excavated, but such might be found at other springs in the same cluster. Raw Material At Amanzi, two types of lithic raw material suitable and used for artefact manufacture are abundantly available in the present day environment. It is not clear whether these were 98 DEACON: THE ACHEULTAN OCCUPATION AT AMANZI SPRINGS equally available in Early Stone Age times, however, although both were used. These materials are quartzite and silcrete, the former being used with greater frequency. In Area 2 on Surface 3, for example, 12-6% of the natural stone in the pebble grade is silcrete, the percentage for the cobble grade being only 2*6%, and of the order of 1% silcrete for flaked stone and tools, (sample size 1324 pieces with natural stone making up 886 of this total). The figures for other excavated samples show comparably low percentages of silcrete used. All large tools such as the bifaces are made in quartzite. The quartzite is derived ultimately from the rocks of the Cape System; its occurrence in the form of gravels is widespread and it is in this form that the quartzite has been used. Ac- cumulations of gravels are to be found on old land surfaces dating from pre-Cretaceous times to the present and much of the material in the Quaternary deposits has been re-worked from earlier gravel deposits. The multi-cycle history of the boulders, cobbles and pebbles in the latter deposits is reflected in the high degree of rounding and the chatter marking of the surfaces. A feature initially somewhat puzzling at Amanzi was the high frequency of split cobbles and pebbles in the deposit, and this is certainly another inherited character. Haughton (1963, p. 274) refers to the sliced appearance of many pebbles produced by pressures within the Enon rocks (basal Cretaceous). The occurrence of sliced cobbles has had an interesting effect on the technique of handaxe manufacture (Plate 20). Extensive terrace gravels (Fig. 2) occur at the foot of Amanzi Hill. These are poorly sorted and include material in the large boulder grade through to the small pebble grade. Cultural material, including a few handaxes, has been found on these gravels, but none in situ in the gravels. These gravels are one obvious source of raw material and there is a widespread, spatially ill-defined scatter of artefact “debitage” along the length of the terrace. Although finds of finished tools are extremely rare, much of the debitage would not be out of place in an Acheulian sample and this probably evidences the working of these gravels and the results of primary blocking-out of tools by the spring inhabitants. A further local source of quartzite may have been available at the springs themselves. The northern flank of the hill forms a step or terrace about 100 feet below the top of the hill and although there is now no visible remnant of any accumulation of gravel that may have rested on this surface, its former existence can be inferred from the quantity of natural quartzitic stones included in the spring deposits. The silcrete outcrops as large rounded blocks on the hillside and there is scree on the slopes. There is in fact no evidence that the outcrops were worked for raw material at this period and many silcrete artefacts retain the cortex of an original cobble form. It is possible that in large measure the technology of the Acheulian populations in this Cape Folded Belt region was adapted to the working of the more generally available cobble quartzites. For large tools at least, the detachment of suitably sized pieces of silcrete of sufficiently homogenous quality was not warranted when quartzite as an alternative occurred in the same environment and even small tools show no preferential selection of silcrete. Quartzites and silcrete have very different fracture properties. Both are siliceous rocks, the former a mass of quartz grains with interstitial silica, and the latter a mass of colloidal (chalcedonic) silica in which quartz and other mineral grains are dispersed. Silcrete is the more isotropic, has a conchoidal fracture, and has seemingly superior fracture properties over the coarse irregular fracture of the granular, partially recrystallized and sheared quartzites. Whatever advantages as a raw material silcrete has over quartzite, these did not weigh in its favour with the Acheulian artificers and it is only with the appearance of the flake-blade industries of the Howieson’s Poort culture dated to 18,000 years B.P. at the type site, that full potential of silcrete is exploited in the Eastern Cape. This later dominance of silcrete at Amanzi is reflected in an artefact sample excavated from Pit 3 and in defined scatters of arte- facts on the top of the hill which include a high flake-blade frequency. Even in these flake- blade industries, quartzite remains important for the heavier and larger tools. 99 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Typology and Techniques In addition to gaining information on the structure of the deposits, the study of the mode of occurrence and the typology of the cultural material were part of the aims of this inves- tigation. These aims are inter-related in that the typological classes were defined with the form of analysis used to study the mode of occurrence or context in mind. The description of the tool classes recognized is given below: Natural Stone This class includes rounded gravel which shows no flaking or obvious signs of use and, grouped with the gravel are the few slab-like joint blocks, spalls and exfoliated stone found, as well as split or sliced stone. The class has been sub-divided on size according to the Went- worth grade scale into Cobbles (256 — 64 mm.) and Pebbles (64 — 4 mm.). Pebbles less than 20 mm. in size were not recorded. The classification according to size is some indication of the relative proportion of the natural stone in respective samples outside that which could be expected to be of value as raw material for artefact manufacture. This includes all stone in the pebble grade. It is obviously impossible to distinguish manuports from natural accumu- lations and the quantity of natural stone associated with artefact occurrences is large in all samples. There would seem to be a parallel in the accumulation of rubble on artefact floors noted at some East African Acheulian sites (Howell, Cole & Kleindienst, 1962, p. 65), but at Amanzi no assumption can be made as to the association of the artefacts and natural stone because the context may be variously disturbed. An increase in the quantity of natural stone at the more obviously disturbed occurrences is apparent. Inherited features such as chatter-marking makes identification of used cobbles as hammerstones difficult. Another inherited feature is the splitting or slicing of the natural stone and this would seem to have no cultural significance. A few cobbles and pebbles show exfoliation which could be the result of insolation or even fire. Flaked Stone Includes cobbles and pebbles which have been struck and one or two but seldom more flakes removed, but the original form retained. Here the proportions of cobbles to pebbles is a good measure of the relative importance of the two grade sizes to the artificers. In all samples flaked cobbles make up the greater portion of the flaked stone class (Figs. 17 — 19). By defi- nition the flaked stone does not include any apparently utilized or trimmed pieces. Quantita- tively the flaked stone class is small in relation to the natural stone in all samples. Irregular Pieces This class includes the irregular flaked material of different sizes that is not recognizably utilized or trimmed. Such material is usually considered as artefact waste and would be equivalent to the chips and chunks of the Kleindienst classification. Sub-division is on size and three grades are recognized: (1) Maximum dimension (or length) greater than 64 mm., (2) length less than 64 mm. and thickness greater than 20 mm., (3) length less than 64 mm. and thickness less than 20 mm. In sample counts, broken flakes are included with flakes, but flake fragments are included in the Irregular Pieces. The introduction of thickness into the grading of the Irregular Pieces was simply in order to include flake fragments in the third sub-division. Cores {Plates 16—17) Cores are defined as artefacts for the systematic production of flakes. Two sub-classes are recognized: (1) radial cores with flaking from the perimeter on one or both faces tending to produce a conical or bi-conical shape, and (2) irregular cores which are struck in two or 100 DEACON: THE ACHEULIAN OCCUPATION AT AMANZl SPRINGS more directions. This classification is comparable to that of Kleindienst except that several forms such as spindle, bi-conical, large, discoidal and pyramidal which show essentially the same radial flaking, are grouped in one sub-class. The sample of cores is not large enough to show the full range of variation and the class is considered to be under-represented. The cores for the production of flakes for large tool manufacture are not represented in the sample TABLE 3 CORE SUB-CLASS FREQUENCIES IN AREA 2 RADIAL SMALL RADIAL IRREGULAR (Discoidal) Cutting 5 9 8 7 6 11 10 3 7 1 . — — 8 2 3 5 9 6 3 4 with the exception of one on Surface 2, Area 2. It is possible that such cores as are included in the samples from the spring sites represent the use of local raw material and much of the primary manufacture was carried out at the major source of raw material on the flats below. The difficulty in defining cores as elements for flake production is that it pre-supposes that a clear distinction can be made between cores and core-like artefacts that may be tools, carefully flaked pieces which on size would appear to be too small to produce usable flakes, or on general morphology where thinness of cross-section and edge would seem to be de- liberately shaped, i.e. the equivalents of the discs and discoids of Kleindienst’s analysis. In the accompanying bar charts (Figs. 17 — 19) discs are shown as a separate class and discoids are grouped with cores. In that the discoids are distinguishable on size from the radial cores — the class into which they grade — and show no apparent edge trimming as opposed to platform preparation, they are classified as small radial cores. Trimming, unless bold, is not readily recognizable in quartzite. Statistically, the frequency of both discs and discoids is too low to have significance in these samples. The radial cores are of interest in the flaking technique they illustrate. The cores in general can be orientated with either cortex or flaking on the under surface, the flaking being platform preparation and distinctive only when steep and stepped. The bi-conical and conical forms would seem only to be a stage in the reduction of the core. Other examples show a flatter surface over which the flakes were struck. The cores can be considered in conjunction with the flakes produced. It is a fair assumption that the bulk of the flakes showing radial dorsal negative scars have been produced from the radial cores. These flakes on an average have 3 to 5 scars and rarely more. These resemble more closely the flaking products from a disc-type core (vide McBurney 1960, p. 134). The predominance of cobble raw material in itself probably explains the absence of any large Victoria West cores and a single struck core of similar type is illustrated (Plate 17, No. 3). Anvils Anvils are all on halved cobbles and show characteristically coarse, steep step flaking along the cord of the cobble fragment. This term is used by Mason (1962, p. 141, Fig. 81, No. 2) for the same artefacts. Signs of battering are not very apparent. 101 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT II, DECEMBER 1970 Cobble Tools ( Plate 18) This class includes flaked cobbles that show trimming and/or utilization as tools. A sub- class is a series of elongated cobbles with a single deep flake removed at one end giving a scoop-like end to the artefact and the tools show damage along this end (No. 5/261). Other cobble tools include heavy unifacial or bifacial cobbles with heavy damage along the working edge. Choppers would also be a suitable term to describe these tools, but is a term generally used with a wider meaning (No. 5/333). Handaxes ( Plates 19 — 29) Handaxes are defined as symmetrical, pointed tools with sharp lateral edges. The range of large bifacial tools in the sample makes the limiting of the handaxe class desirable for adequate description. The limits of the handaxe class are arbitrarily set and there are inter- mediate forms between handaxes and other large bifacial tools. This is not simply a product of the latter being unfinished handaxes or rough-outs as they show features such as the absence of emphasis on the point which by definition excludes them from the handaxe class although they show ample evidence of having been used as tools. This subdivision cannot be taken to imply functional differences. Malan (1939, p. 247), in relation to a sample of handaxes from near Wellington, considered the standardization of shape in handaxes made on cobbles was due to the selection of raw material and that the shape with the highest frequency, the pear-shape, was a product of this selection. The range in plan form at Amanzi is similarly limited and with flaking minimal in general, the cobble form of the raw material is a controlling factor. The common shape is the pear-shape or the long ovate to lanceolate form. Plate 24 illustrates the nearest approach to an ovate form and the cordiform is not represented. Cortex, frequently removed from only one face, controls the shape of the butt on many specimens. The largest homogeneous sample of handaxes numbers 63 from Area 2 and the data for this sample are given as typical for this artefact class at the site. The handaxes show con- siderable variability in the primary measurements (length, breadth, thickness and weight). For example, length, measured as the longest axis of the tool, gives the following results in this sample: SAMPLE OF 63 HANDAXES FROM AREA 2: MEASUREMENT OF LENGTH Number Range Mean (X) Standard Deviation (s) Coefficient of Variation 100 s (V = -) X 63 81 — 265 mm. 146 mm. 38 mm. 21 • 3 mm. The length frequency distribution differs significantly from the normal and it is bimodal with peaks in the 111 — 120 mm. (f = 13) and 151 — 160 mm. (f = 9) length classes. Weight is another measure of size and the range in this sample is from 6-5 — 70 oz. The frequently quoted index in the description of handaxe samples is the breadth-length B index (— X 100). This index is a simple measure of shape expressing it as a relative broadness -I—/ or narrowness. It does not take into account thickness, weight, taper, pointedness or other features which make up the whole form of the tool. In a sample such as this from Amanzi, 102 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS with a limited range in plan forms, the breadth-length index is a useful statistic. The frequency distribution approximates to normality (P = 0*8 — 0-9) and the coefficient of variation (V = 11 -75) is lower than for either of the component measurements. The sample data are tabled below. The estimate of the mean at the 95% confidence level is between 58-7 and 62-2. SAMPLE OF 63 HANDAXES FROM AREA 2: BREADTH-LENGTH INDEX Index Number Mean (X ± s.e.) Standard Deviation Coefficient of Variation (s) (V) B — x 100 L 63 60-4 ± 0-89 710 11-75 Mason {op. cit., p. 221) gives the mean values of the breadth-length index of a series of handaxe samples of comparable size from six South African sites. The mean value for this index in the Amanzi sample falls midway in the range of values cited by Mason (Riverview Estates = 54, Cave of Hearths = 65). Apart from variations in size and shape in the handaxe class there are other general characters of note. The flaking on many of the tools is minimal and is primary blocking out rather than trimming or shaping of the edges and point which are the functional parts of the tool. There are exceptions showing flat flaking and a high degree of refinement or finish. It is becoming increasingly evident that the assumptions cannot be made on the general refinement in a handaxe sample regarding its position in time (Leakey, M. D., quoted in Clark and Bishop, 1966). The points are variously acute or obtuse but not attenuated. This is partly a function of technique and a number of tools show a five flake pattern on the point. This indicates a more precise technique than Goodwin’s analogy to sharpening a pencil (Goodwin, 1933). The type of point resulting depends on how far the pointing flakes run. The point flaking is done on both cobbles and half cobbles, but not on large flakes made into handaxes, Plate 19 shows the typical undersurface preparation and Plate 25 the three upper surface pointing flake scars on a finished tool. Quartzite is relatively brittle and this is shown in the number of broken handaxes and handaxe tips found. The trimming shown on the point of some handaxes (Plate 21) may indicate retouching of a snapped point rather than a design for a specialized function. By definition, handaxes have sharp lateral edges, but this is difficult to measure. The edge trimming is variable as is the resulting fineness of this edge. As a whole the handaxes do not exhibit the same fine edge as large samples from Geelhoutboom (Albany Museum accession 2946) in a similar material. Damage on some handaxe edges, again a subjective observation, is heavy and suggests a wide range for the use of these tools. Cortex is frequently on the butt of specimens, either over the whole butt or at the butt and extending down one face or surface. This again is in part a product of the technique and the raw material. A high proportion of the handaxes are on cobbles or half cobbles and a lower proportion are on end-struck or side-struck flakes. The technique reflects not only in the points and the butt, but also in the thickness in cross-section, symmetry and other features and accounts for variation in the class. A technique evident at this site is the use of longitudinally split cobbles for biface manufacture (Plate 20). The tool is made in a different plane from that of the long axis of the split cobble and this allows the production of a tool symmetrical in minor section, keeled and with cortex on the butt and extending down half of one surface. This technique 103 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 is interesting in that it shows practical appreciation of solid geometry and also, in conjunction with point flaking, the repeated use of seemingly rigidly standardized procedures in tool making. Cleavers ( Plates 30-35) This is a class of transverse-edged tools classified as in Mason (1962) and Kleindienst (1962). Numerically the class is poorly represented. There is some variation not only in plan form but also in the damage on the transverse edge. The cleavers in the Area 2 samples are not typical in that several show notching of the transverse edge (Plate 30 and 31) due to use and the other examples are heavy tools or show heavy damage on the transverse edge. These could be classified as transverse-edged tools rather than cleavers. From Area 1, more typical examples of parallelogram-trapezoid forms (defined as in Mason, op. cit.) were found and in addition two ultra-convergent cleavers (Kleindienst, op. cit.) came from Square 2 (Plate 35). Other Bifaces ( Plates 36-41) The range of unstandardized tools made in the biface tradition have been grouped in this class. Some subdivision is possible on edge and point. With the possible exception of the few limande-shaped symmetrical, “round-bitted” bifaces, no formal tool types are included in this class. The heavy limande (Plate 36) with extensive trimming over both faces would seem to fall into the round-bitted biface sub-class of Kleindienst’s classification, but the frequency in the total sample is too low for recognizing these as separate from other elongated bifaces. The class has been divided into 4 sub-classes: (1) Elongated bifaces (Plate 36, 38 — 40) include those tools lacking any emphasis on the point, but with trimming and not infrequently heavy damage on the lateral edge. In some instances the point has been modified by steep retouch. (2) Other pointed bifaces are a group showing no emphasis on the lateral edges (Plate 37). (3) Pointed cobbles shown the absence of lateral edge retouch and flaking is confined to the point. (4) Unpointed edge trimmed bifaces are included here although they lack the symmetry about the long axis, have a single edge trimmed and no point (Plate 41). These would be excluded in a more general description from the handaxe-like forms, but it is con- venient to include them here. The accompanying table shows the frequency of these sub-classes in the Cutting 5 — 9 samples. Notably rare are any pick-like forms. There are two tools from Surface 3 which could be included in the pick class as defined for the East African Acheulian and a further example was found in Cutting 1 during the 1963 excavation. TABLE 4 FREQUENCY OF OTHER TOOLS AND OTHER LARGE BIFACES IN CUTTINGS 5—9 (a) Other Tools Cutting Stone Balls Hammerstones “Wedges” “Adzes” 5 1 1 1 2 6 — — 4 3 7 o — — 2 o 9 1 — 3 j 1 Totals 2 1 10 9 l 104 DEACON: THE ACHEULTAN OCCUPATION AT AMANZI SPRINGS (, b ) Other Large Bifaces Cutting Elongated Bifaces Other Pointed Bifaces Pointed Cobbles Unpointed Edge Trimmed 5 9 6 — 9 6 9 5 1 5 7 1 1 — — 8 4 4 — — 9 2 1 1 4 Totals 25 17 2 18 Discs: The frequency in the class is low in all samples, which has necessitated the grouping of this with another class in any analysis ( vide Cores). Other tools (Plates 42 — 43) Again, for analysis purposes, it has been necessary to group in this class several tool types which are unimportant numerically, but may have typological importance. The fre- quencies in the samples from Cuttings 5 — 9 are given in the accompanying Table. The class includes: (1) Stone balls, typical small polyhedral stone balls which require little comment. (2) Hammerstones. The recognition of hammerstones is difficult on chatter-marked cobbles, but all natural stone was examined for signs of use. A single example in the Cutting 5 sample is in a hard, clear quartzite and the damage is clearly visible. Only two other hammerstones were recorded from the whole excavation. (3) Transverse-ended tools or “Wedges”: A series of tools with a wedge-shaped major section have been grouped in this sub-class. The frequency in the samples is too low to confirm whether this sub-group is a typological entity (Plate 42). The primary form is a flake and the edge is transverse being formed by the intersection of two flake surfaces, or cortex and a flake surface, as in a cleaver. Trimming, in some examples as steep as scraper retouch, extends around the transverse end and up one or both sides. The trimming distinguishes them from the cleaver form and they are best described as flat- based, high-backed, transverse edged tools. Apart from a single example in Cutting 10, all come from Cuttings 5 — 9. (4) Small edge-trimmed tools or “Adzes”: This is again a group which has doubtful validity because of low frequency. They are small bifacially trimmed tools with shallow trimming on one face and step flaking on the under surface (Plate 43). The trimming is not scraper retouch and in some examples is more adze-like. All examples from the excavation came from the samples tabled. Flakes (Plates 44 — 5 1 ) Flakes have been classified as large and small on the basis of the length of the longest axis (> 100 mm. and < 100 mm. respectively) and as modified (miscellaneous trimmed and scraper re-touched) and unmodified. Snapped flakes have been included in the counts, but flake fragments (< 10 mm.) have been classified as small irregular pieces. There is no apparent difference in the flake length frequencies for modified and unmodified flakes, except that larger flakes show a higher incidence of trimming. Tables 5 and 6 give details of size range and the trimming in typical samples. On these figures it would not seem that there is any selection in size and the larger flakes form a “tail” in the length frequency distribution. This would seem to be against any assump- tion that the small and large flakes have specific functional differences (compare Kleindienst 105 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 op. cit.). The larger flakes in general are thicker and retouch is coarser and this may be an explanation for some variation in trimming on the flakes. Trimming on the flakes is minimal in many examples and may amount to little more than utilization damage: in quartzite as stressed elsewhere, the recognition of trimming, where minimal, is difficult and makes for a possible source of error in classification. Where trimming is steep, this has been classed as scraper retouch. The flake scrapers are for the most part informal (compare Clark, 1954, p. 95) tools with the retouch along the longest axis (side scrapers). Scraper is used here as a descriptive term without any implication of function. A range of flakes is illustrated showing characteristic shape, dorsal preparation and trimming (Plates 44 — 51). TABLE 5 AREA 2 (a) Flake Frequencies (sub classes) Cutting L; arge Flakes (100 mm.) Sm all Flakes (10C mm.) Misc. Trimmed Scraper Unmodified Misc. Trimmed Scraper Unmodified 5 11 (undifferentiated) 2 41 20 53 6 11 8 — 32 28 43 7 — — — 2 4 1 8 4 4 — 9 20 18 9 7 2 2 10 8 34 Surface 2 4 6 — 23 65 106 Surface 3 14 20 2 9 30 65 ( b ) Position of Scraper Edge on Small Flake Scrapers Cutting Side Double Side Side and End End Circular Notching of Edge (all categories) 5 11 3 — 3 3 1 6 18 1 7 — 2 2 7 1 — 1 2 — 1 8 12 2 3 — 3 8 9 5 — 2 — 1 1 Surface 2 44 2 3 9 7 11 Surface 3 21 1 2 5 1 4 Total 112 9 18 19 17 — 1 1 1 ! 1 Miscellaneous Trimmed Pieces This class includes those artefacts that show some trimming but do not fall into any one of the other classes. For the most part the trimming is on irregularly flaked pieces or chunks. 106 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS TABLE 6 FLAKE LENGTH FREQUENCIES Class (Length in mm.) Area 1 Cutting 10 Area 2 Sample 1 Sample 2 Cutting 6 Cutting 8 Surface 2 Surface 3 A B A B A B A B A B A B 10—19 — — 1 — — — — — — — — — 20—29 1 1 9 2 5 3 — — 3 — 2 — 30—39 12 5 28 1 14 7 6 4 11 3 9 I 40—49 15 8 26 8 13 15 4 4 18 15 10 3 50—59 6 5 31 8 8 15 3 5 1 1 7 11 5 60—69 2 3 17 3 2 9 3 8 6 7 7 5 70—79 1 3 11 5 1 7 2 4 6 7 3 5 80—89 — 3 2 4 — 3 — 2 — 5 2 4 90—99 1 2 4 3 — 1 — — — 2 — 5 100—109 — 1 — 4 — 4 — 1 1 — — 8 110—119 — — — 3 — 4 — 3 — — 1 6 120—129 1 1 2 2 — 2 — 3 — 1 — 6 130—139 — — — 2 — 1 — — — — — 1 140—149 — 2 1 1 — 1 — — — 1 — 1 150—159 1 — — 160—169 1 — — 2 — 4 — — — — — — 170 plus — — — — — 2 — 1 — — — — Total Measured 40 34 132 48 43 78 18 35 56 49 45 50 A — untrimmed B — trimmed. 5. Sampling and Context At Amanzi the overburden covering the Acheulian artefact occurrences is everywhere in excess of several feet and thus in each instance the cuttings were initially exploratory. With the exception of some pits (Fig. 3) all cuttings were located within two spring depressions and no Acheulian horizons are known associated with hill slope deposits. In view of obvious local disturbance within the deposits, samples of artefacts drawn from different parts of the depressions do not have equal value for interpretation. In the absence of evidence of hearths or structures, areas of occupation can only be defined in terms of areas of intense activity reflected by high density artefact scatters. As high density scatters could result from concentra- tion through transport in a wholly geological context it is important to distinguish samples in primary or semi-primary archaeological contexts from those including in the main derived and possibly selectively sorted material. It is perhaps to be expected that any main occupation areas would lie on the margins of the springs within or outside the depressions. However, some activities may have resulted in artefact scatters extending to the central portions of the spring. Here foci such as the spring vent in Area 2 may have been of import. Again it is unlikely that within an environment such as a spring depression artefact occurrences will be interpre- table in terms of simple discrete occupations. Acheulian cultural material in the youngest deposits is clearly derived as evidenced by the concentrations within erosion irregularities. This is best seen in Cuttings 1 and 10 of Area 1. The artefacts occur at the base of the pothole fill and some concentrations appear to fit what has been described elsewhere as “swarms” (Caton-Thompson, 1952). The context here is wholly disturbed and selective sorting evident to some degree. 107 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 In Area 2 there has been a measure of disturbance and selective sorting on the steep sloping Surfaces 2 and 3 within the white sand body forming the central portion of the de- pression. To a lesser degree this is again the case in the sample designated as from Surface 1. The Surface 3 sample (Plate 10) includes a high proportion of natural stone in the cobble grade and the occurrence was analogous to a cobble pavement in appearance. The Surface 2 sample again shows a high frequency of natural stone but in the pebble grade and in the Surface 1 sample there is a marked reduction in the relative proportion of natural stone to the total frequency of artefacts plus natural stone. TABLE 7 % Natural Stone Pebble Grade 64 mm. Cobble Grade 64 mm. Surface 1 44% 36% 18% Surface 2 75% 61% 14% Surface 3 67% 32% 35% The different grades of natural stone in the Surface 2 and 3 samples is evidence of selective sorting or differential transport and the anomolously high proportion of natural stone in these samples indicates the disturbed context. The Surface 1 sample (Cuttings 6, 7, 8 and 9) was derived from the sand surface and the base of the brown humic sands immediately overlying it. There are a number of small irregu- larities on the surface of Cutting 6. There is some concentration of cultural material in these irregularities and around the vent (Fig. 20) with the result that in the total sample selective sorting of larger and smaller artefacts is not pronounced. The composition of the Surface 1 sample is not significantly different from the Cutting 5 sample from the margin of the de- pression and more plausibly in a semi-primary archaeological context (x2 test P = 0-8 — 0-9; 10x2 contingency table based on Figs. 18 — 19). The scatter on Surface 1 could be traced as continuous with that in Cutting 5 in the excavation and is in part or wholly contemporary. On the excavated exposures it could not be determined whether Surfaces 2 and 3 were also an extension of the Cutting 5 occurrence but this appears possible (Fig. 13). The distribution of large tools in Cutting 6 (Surface 1) was examined to test whether any groupings of these tools could reflect direct evidence of activities in the central portions of the spring. This proved negative and the observed frequencies show a close correspondence to the expected frequencies, calculated from the Poisson probabilities. The data for the handaxe sample are given as an example ( vide Fig. 20). TABLE 8 Number of handaxes per 2' x 2' square Observed frequency Expected frequency (Poisson distribution) Poisson's probabilities 0 15 15 0-50 1 9 10 5 0-33 2 3 and over 6 0 3-6 0-9 012 003 There is a limit to the artefact distribution on Surface 1 seen in the decrease in finds in Cutting 7. This probably finds an explanation in the distribution of a clayey facies, the marginal clays on this side of the depression. There is some basis on this distributional limit to exclude the possibility of the cultural material in the Surface 1 sample being derived from outside the depression. The cultural material from Surfaces 1 — 3 is thus considered to be from disturbed 108 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS archaeological contexts. The selective sorting evident in the samples from Surfaces 2 and 3, however, is less pronounced in the Surface 1 sample, which is comparable in composition to the sample from Cutting 5. The Cutting 10 sample from Area 1 and the Cutting 5 sample from Area 2 are drawn from occurrences of similar form. Acheulian artefacts occur as a dispersed accumulation through a thickness of some 4 feet in Cutting 10 and 2 feet in Cutting 5. In the latter, a limited exposure 20' by 6' and cut by a modern furrow, the density of finds was uniformly high. In Cutting 10 it is evident that there is variation in the density of finds in both the vertical and horizontal dimension. Occurring in the base of the grey black silt sub-unit and extending into the under- lying deposits there is no clear basis for partitioning these samples. This mode of occurrence as vertically dispersed artefact accumulations, is more difficult to interpret than an occurrence of single tool thickness on a coherent temporary land surface. Two factors that are not mutually exclusive are involved in the vertical distribution of artefacts in the Cutting 10 and 5 occur- rences ; these are settling and time. In Cutting 1 0 there is a general vertical fall off in the density of finds as measured by weight in the sampled blocks (Fig. 16) which strongly suggests settling as a major factor. Again the ratio of large (>100 mm.) to small artefacts is higher in the top 1 ' 6" of the Cutting 10 accumulation than in the underlying 2' 6" of the zone sampled. This evidence in addition to varying attitudes of elongated tools found in the deposit supports the contention that some measure of settling has taken place. Evidence that a time factor is involved in addition to settling is given for example by the occurrence of three large cleavers in one sample block of Cutting 10 (O' = 6 '/0' — 5') at a depth of 10' 6" below datum (Fig. 16). On probablistic grounds, in Area 1 where the frequency of cleavers is low, this association in isolation can be regarded as significant and indicative of primary context. Mention has been made of a similar grouping of this type in Square 2, but related to a surface on the white sands towards the center of the spring. More direct evidence that time is involved in these vertically dispersed accumulations, is provided by the limited sample of cultural material found below the impersistent upper clay band in Cutting 1. The horizontal variation in the artefact density in Cutting 10 (Area 1) is shown by an abrupt increase in artefacts in the upper 18" of the artefact accumulation in sample block 10' — 15'/30' — 35' (Figs. 15 and 16). There is a two-fold increase in the density of artefacts per unit volume of deposit in this section and the denser scatter traced for some feet is not localized or related to any line of erosion or wash. For analysis this portion of the Cutting 10 sample has been designated as Sample 1. Samples 2 and 3 from Cutting 10 have been arbitrarily defined as the upper 18" and lower 30" of the artefact zone in the rest of the area of the Cutting (Fig. 15). Although Samples 1 and 2 are directly comparable. Sample 3 in view of the effect of settling is perhaps of more limited interest. The Cutting 5 sample (Area 2) has been con- sidered here as a single unit. In future investigations at the site, extensions to the eastern portion of Cutting 10 and Cutting 5 will deserve consideration. The quantity of natural stone in the samples from the Surfaces 1 — 3 in Area 2 has been used as a measure of disturbance. In Cutting 5 Area 2, the percentage is comparable to that on Surface 1, i.e. between 40 and 50%. This figure is reduced to below 40% in the Cutting 10 samples in Area 1. There is some variation in the grade of natural stone and noteworthy is the high proportion of stone in the cobble grade in Sample 1 of Cutting 10 (Fig. 17). This could be explained by selection and concentration of material in a more usable size range in this locality. The occurrence of natural stone in the deposits is in itself interesting as it is associated with the artefact occurrences and is outside the grade of normal sediments deposited by the springs. Cobbles and pebbles are not found randomly dispersed throughout the spring deposits. The ultimate source of much of this material is probably local, possibly a remnant of terrace gravel on the hill flank. On the present evidence, it appears unlikely that all the natural stone and likewise the artefacts, have been derived from the spring surrounds by hillwash and accumu- 109 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 lation on temporary stable surfaces within the depressions. However such an explanation is possible for unworked stone in the smaller pebble grade. Inskeep ( op . cit .) classified the Cutting 1 sample of cultural material on varying degrees of abrasion which would seem to suggest transport. However, this sample included artefacts from the post-unconformity member which has suffered disturbance. Any measure of sharp- ness or dullness of artefact edges is difficult to apply as differential weathering related to the shistosity and felspathic composition of some quartzites is involved in addition to any physical effects due to abrasion. In the sample as a whole the supposed effects of “abrasion” are not high. The occurrence of typologically foreign elements in the Acheulian samples is a result of natural conditions peculiar to the site and outside the control of sampling. The recognition of such artefacts lies in that they fall outside the range of variation in equivalent classes in the main sample and they exhibit different techniques of manufacture. The two micro cores mentioned by Inskeep {op. cit.) are examples. In the Cutting 10 sample there are similar small silcrete elements, one triangular point and several fragments of snapped triangular or long quadrilaterial flakes with dorsal surface preparation atypical of the rest of the flake sample. These “intrusives” are in the top of the accumulation and are not dispersed throughout. In Area 2 on Surface 1 and Cutting 5, the artefacts considered to be intrusives are illustrated in Plate 52. The proportion of such apparently foreign elements is too small to bias the samp- ling. Their occurrence demonstrates the poor seal the grey black silt bed provided at a stage prior to the humification of organic material in this bed and its truncation as a compact horizon by the erosional unconformity. 6. Analysis The excavations at the two spring depressions, although adjacent, are in effect at two sites. Each relates to a different occupation or series of occupations which may be broadly contemporary. The ridge separating the depressions is mantled by spring deposits or soil but is presumed to overlie a dividing spur of Cretaceous rock. The object of this analysis is to examine the class frequency data given in the accompanying bar charts. In this the chi squared statistic has been used as a significance test between samples. The power of the test is reduced in some applications by the number of variables considered. Again there is some loss of information in the grouping of the frequencies in certain classes to fit the requirements of this test. Area 1 Samples 1 and 2 from Cutting 10 are directly comparable as they represent different portions of the upper section of a continuous artefact zone. A major difference lies in there being a two-fold increase in the number of finds per unit volume of deposit in Sample 1 relative to Sample 2. The well marked limit to the Sample 1 scatter suggests that this may represent a distinct occurrence. The samples include the same range of artefacts; however, there is a significant difference (P = -001) in the relative frequencies of artefacts in the different classes (a 9 x 2 table based on Fig. 17). The overall difference at a general level is an increase in the large elements (natural cobbles, flaked cobbles, anvils and cores and large bifaces) and a decrease in untrimmed or unmodified flakes in Sample 1 relative to Sample 2. The frequencies of small (< 100 mm.) trimmed and untrimmed flakes was tested in a 2 x 2 con- tingency table and found to be significantly different at the 1% level for the two samples. The archaeological inference drawn from these observations is that the spatial pattern in the distribution of artefacts in Area 1 would appear to be the result of activities attendant on occupation rather than the result of selective sorting by geological agencies. The eastern or Sample 1 portion of Cutting 10 cannot be considered to be a simple extension of the artefact 110 DEACON : THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS scatter represented by Sample 2. The value of extending Cutting 10 to the east in future exploration of the site is apparent. Area 2 As noted, the artefact class frequencies in the samples from Cutting 5 and Surface 1 were not significantly different. Although the Surface 1 occurrence offered the possibility of partitioning this sample, the obviously partly disturbed context of the material did not warrant this. The combined samples from Cutting 5 and Surface 1 present the largest single sample unit from the springs for typological comparisons. In Area 2 in the fall off in artefact density towards Cutting 7 on Surface 1 there is a distributional limit which may in part be culturally determined. The occupation of the Area 2 depression is more clearly multiple as evidenced by the horizons excavated below Surface 1. The samples from Surfaces 2 and 3 have not been compared with those from Surface 1 and Cutting 5 as the former are biased due to selective sorting by natural agencies. Areas 1 and 2 The two areas excavated have provided artefact samples classified within the Acheulian complex. There are some tool sub-classes of possible typological significance found only in Area 2, but these have low frequencies even in this larger sample. There are differences between the combined samples from each depression in the relative proportions of trimmed and untrimmed flakes and in the incidence of cleavers to handaxes and other large bifaces ( x 2 test; P = -001), (data given below). Cleavers Handaxes and large bifaces Total Area 1 12 59 71 Area 2 8 196 204 Total 20 255 275 The cleavers from Area 2 are atypical in showing heavily damaged transverse edges or notching in this edge. At a purely subjective level the biface element from Area 2 is heavier and edge damage more marked. Although other differences exist and are not brought out by this analysis, the broad range of activities in as far as they are indicated by the lithic samples, would appear to have been essentially similar. 7. Dating On the basis of their geological setting the spring deposits can be dated to the Pleistocene and the phase of river incision that has resulted in the modern Coega valley. The considerable volume of deposits suggests a long period of spring activity even if the deposits form a some- what patchier cover on the hill flank than the few bore-hole sections indicate. Estimates of the age of two other spring complexes, Caledon at 300,000 years and Warmwaterberg at 850,000 years (Kent, 1950, p. 247) based on calculations of the rate of build up of iron precipitates by the spring waters, are indicative of the order of age of some spring deposits. This method of dating, however, is incapable of great accuracy or refinement. Although some members of the Amanzi Springs Formation may be of greater age, the deposits in the excavated areas near the Edwards bore-hole appear on the included cultural material to date to the Middle- Upper Pleistocene. Ill ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Following the 1963 excavation, a sample of wood from the natural drift accumulation in Cutting 1 was submitted to the Groningen Laboratory for radiocarbon dating. The apparent age 60,600 d= 1,100 years B.P. (GrN — 4407 and 4546; Vogel and Waterbolk, 1967) is a terminus post quern for the main artefact occurrence higher in the sequence. The stratigraphic position of the sample is fixed in relation to the two dark clay bands exposed in Cutting 1. No cultural material was found in the Cutting 1 extension below the surface on which the wood had accumulated. Two further wood samples from the Rietheuvel Member had been dated. They include a sample from Area 1 Square 1 from a stratigraphic position in the brown humic sands below that of clay markers in Cutting 1 (SR — 103: 32,900 ± 600 years B.P.) and a sample from Area 2 Cutting 6 from close to the edge of the spring vent and on Surface 1 (SR — 107 : 38,100 ± 2,000/1,600 years B.P.). The apparent age of these samples is lower than the expected age and neither was given any specific pretreatment. The dating of these two samples is of the same order as the result for a single sample from the Balmoral Member. While the stratigraphic position of the samples is not in doubt, it is tempting to see in this correspondence a relationship to some common factor possibly relating to the second phase of spring activity. The two dates SR — 103 and 107 must be considered minimum dates for the Acheulian occupation. The dating of the Balmoral Member rests on the age determined for a wood fragment from Cutting 10, Area 1. This result (I — 2241 : 31,000 i 1,200/1,100 years B.P.) is notin conflict with other dating evidence. The wood occurred one foot above the unconformable surface in the frontal section of Cutting 10 (co-ordinates 15/15 Fig. 15) shrunk within an isolated cavity. 8. Springs as Archaeological Sites The Amanzi Springs in themselves are of interest as a number of similar deposits have been excavated elsewhere in southern Africa and farther afield. Possibly the best known and the best published excavation of this type of deposit is the study of the Kharga Mound Springs, United Arab Republic (Caton-Thompson, 1952). The Florisbad Springs, in the Orange Free State, have been the subject of investigation by a number of workers (Dreyer, 1938; Meiring, 1966) between 1930 and 1952 and a more recent example of archaeological interest in spring deposits is the papers that have appeared on the Gwisho Springs, on the margin the Kafue Flats in Zambia (Fagan and van Noten, 1966; Gabel, 1965; Van Noten, 1965). There are some structural and depositional features at these sites mentioned which find parallels at the Amanzi Springs. As a result of the scarcity of surface waters over much of South Africa, springs played an important part in the historic settlement pattern prior to the advent of the drilling machine. Spring localities appear to have had an equally important role in prehistoric times and they are features at which conscious archaeological search can be directed. The unusual conditions for preservation at some sites may offset the limitations imposed by the complexity of the depositional processes operating in such environments. Again in South Africa where uplift and peneplanation have been the dominant processes operating throughout the Quaternary, spring deposits are an important class of deposit of potential archaeological interest in that they may include sealed artefact occurrences. Large numbers of springs are known in the dolomite areas of the Transvaal, northern Cape and South West Africa and may be associated with the build up of calcareous tufa deposits. Plant impressions are known from tufa deposits (e.g. Windhoek Springs) and the Taung find site is an example of an archaeological site associated with tufa deposits. Of more interest here are spring deposits formed by mech- anical sorting of suspended solids derived at depth and from the re-working of the bed rock. These give rise to typical mound springs. Artesian areas are rare in South Africa and most springs are fed by re-cycled meteoric waters held in fissure or fracture zones in rocks of low porosity; the location of the springs being fixed by structural controls such as dykes or faults. 112 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Interest has thus far centered on the composition of spring waters (Rundl, 1915, quoted in Hall, 1938) and the occurrence of thermal springs (Kent, 1950). Deposits other than chemical precipitates have received little attention although relatively considerable thicknesses have been built up at some localities. The thickness of the spring deposits at Amanzi (22 feet, 6-8 metres) is matched by the known thickness of deposits at Florisbad (21*5 feet, 6-8 metres, Oakley, 1954; Meiring, 1956) and the springs at Aliwal North (over 10 metres, more than 30 feet; Coetzee, 1968). Cultural and faunal material have been recovered from the Cradock Springs and they are of possible interest although the thickness of the associated deposits is unknown. Springs issuing from physically and chemically resistant rocks such as the quartzites of the Cape System may be associated with little build up of spring deposits as at the Uitenhage Springs although iron-manganese sinters, at Caledon for example, are formed by a number of springs in rocks of this system (Kent, 1950). Reliable information on the aerial extent and thickness of deposits associated with the majority of spring occurrences in South Africa is lacking. While many factors are involved, as a guide to conscious search, three of the spring deposits noted above occur in the Beaufort Series of the Karroo System (Aliwal North, Florisbad and Cradock). Selection could be directed at thermal springs as these may be associated with larger scale geological structures and the flow may be less dependent on minor fluctuation in the reservoir intake. The dating of the deposits, where known, ranges from Upper Pleistocene (Amanzi), Upper Pleistocene — Recent (Florisbad, Meiring, 1956; van Zinderen Bakker in Bishop and Clarke, 1967, p. 133) to post Pleistocene (Aliwal North, Coetzee, 1968). Outside South Africa, numbers of springs are known in Rhodesia associated with post-Karroo faulting (du Toit, 1939) but have yet to receive archaeological attention and in Zambia in addition to the Gwisho springs, a fairly extensive spring deposit with associated cultural material has been reported from near Chingola (Sampson, 1965). The mound springs at Kharga Oasis occupy spring eroded hollows in Cretaceous clays and are fed by artesian waters. The springs for the most part form discrete, strongly linear clusters of vents and the associated deposits may extend over a distance upwards of one kilometre (0-6 miles) (Caton-Thompson, 1952, Fig. 7). The sorting action of the waters has produced typical sand cores to the springs and associated finer sediments. The vent structure at KO 8 A {op. cit. Fig. 14) is the type of structure excavated in Cutting 6 Area 2 at Amanzi. The nature of the contact between the white sands and overlying green clays, the segregation of clays within the sands and stringers of sand in the overlying clays at KO 10 and again the disconformity apparent in the KO 6N trench face (op. cit. Plate 6 No. 1 and 4 respectively) are strongly reminiscent of features in the Amanzi deposits. The Acheulian handaxe clusters {op. cit. Plate 6 No. 2) resemble the mode of occurrence of cultural material in the base of the pothole fill in Cutting 1 Area 1 at Amanzi and it is tentatively suggested that a cultural explana- tion is not demanded for the occurrence of such swarms. The organic material reported from the Kharga site includes plant stems, seeds and as at KO 5B bands of organic rich sediments {op. cit. pp. 84, 153). Plant material described as carbonized probably represents de-humified organic material. In such spring environments only the skeletal structure of the plant remains is preserved. The macro and micro plant fossils at the Kharga springs may repay further study. The Florisbad Springs are best known for the find of a human skull in the lower layers of the deposit. A dolerite dyke exposed in the base of the 1930-2 excavations appears to be the structural control for the spring occurrence and the two lines of vents (Oakley, 1954; van Zinderen Bakker, 1957) may lie on either side of this barrier. The linear arrangement of vents is a reflection of the migration of vent outlets eastwards in the complex (Dreyer, 1938; Oakley, 1954). Piaget {in van Zinderen Bakker, 1966) has described the grading of the spring deposits as due to the elutriation effect of the rising waters and a succession of sand and dark organic clayey layers (van Zinderen Bakker, 1957), the latter termed Peats I — IV by Dreyer (1938), is recognized in the main section excavated. Several vent structures and minor 113 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 upshoots of sands have been noted in the deposits (Dreyer, 1938) and that disconformities in the sequence may be represented by erosional features at the top of successive dark organic layers, has been suggested by van Zinderen Bakker (1957, p. 59). Present dating evidence is somewhat unsatisfactory as a base against which to interpret the discontinuous pollen spectrum from the site. A major time break has been suggested between the lower (Peat I) and upper parts (Peats II — IV) of the deposit (van Zinderen Bakker, 1967) on the basis of available radiocarbon determinations and this requires confirmation from field studies. If confirmed, the time break would be of the same order as that recognized between the deposition of the Rietheuvel and Balmoral Members at Amanzi. It can be argued that the high counts of Chenopodiaceae, Cyperaceae and Zygophyllum in the pollen profile represent vegetation stabilized at the springs during periods of diminishing activity and not “wetter spring cycles” (van Zinderen Bakker, 1957, Fig. 2). This would reduce the apparent corres- pondence between cycles of spring activity and warm dry regional climatic conditions inferred from the relative percentages of Compositae and Gramineae. Rapid sedimentation and flush flow conditions appear more clearly evidenced in the apparently coarser facies in which stems of drowned plant communities have been preserved in their context (Dreyer, 1938) and in which, incidentally, the pollen counts are low. In the general structure of the Florisbad deposit, however, there are many features that parallel those described from Amanzi and Kharga. Emphasis in excavation has been placed on the apparently richer accumulation of fauna and artefacts in vent structures where from considerations of context, the results are least likely to be meaningful. Areas marginal to such structures may prove more productive in yields of demonstrably associated cultural and faunal samples. For the present no adequate description of the cultural, faunal or macroscopic plant remains from the site exists. Some published information is available on two of the three Gwisho Spring Mounds. Gwisho A and Gwisho B are about a quarter of a mile apart and each mound is some 120 feet (40 m.) across and seven to eight feet high. The outlets of these thermal springs are at present on the margins of the mounds. No detailed description of the lithology exists for either site and the suggestion that the Gwisho A mound represents an accumulation of wind blown alluvium (Gabel, 1965, p. 26) is unsupported. Vent structures termed spring eyes or spring heads are recorded at both mounds and they are apparently true mound springs. The dating of the Gwisho B site, 4,785 T 70 years B.P. (GrN — 4307) for the lower levels and 3,660 d= 70 years B.P. (GrN — 4305) (Fagan and van Noten, 1966) is in close agreement with the dating of the Gwisho A site (Gabel, 1965) and the mounds are post-Pleistocene in age. Archaeological interest in this group of springs centers on the occurrence of numbers of burials in the mounds, wood and lithic artefacts of the local expression of the Wilton complex. The associated faunal and plant remains are a potentially important source of information on the ecology of the spring dwellers. 9. Discussion of Botanical Remains The preservation of plant remains at the Amanzi site is the result of the damp and acid conditions prevailing there. Soil samples tested show pH values from 5-5 to less than 4-0. These values are somewhat lower than that of the bore-hole waters which have a pH of less than 7-0. Conditions for preservation are not uniformly good over the whole site and even where most favourable only the more durable elements have survived. The partial drying out and the oxidation of the upper part of the deposits, accelerated in modern times by bore- holes lowering the water-table, is a major factor that has controlled the degree of preservation. Macroscopic Plant Remains Wood is preserved in the Rietheuvel Member and most of the finds are from Area 1. The wood appears to represent natural accumulations and the source would have been woody 114 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS plants growing in the immediate spring surrounds. It is typically black in colour and it is assumed that of the original composition only the lignin of the cell walls has been preserved. The Forest Research Institute undertook to section a limited number of samples ( 1 8 specimens) and for comparison 23 species (collection MJW 3160 — 83, Albany Museum Herbarium) of woody plants growing at the present day on the north flank of the hill were sectioned. The wood from the deposits was sectioned after impregnation with celloidin. No positive identifications resulted from this work and a limiting factor to eventual identification is the detail of the cell structure preserved. In this condition there is considerable variation but at least some sections are potentially identifiable. There is no key to the identification of commercially unimportant indigenous woody plants in South Africa. At this stage the only conclusions that can be made are that the local forest species such as Podocarpus and Widringtonia are absent in the small sample sectioned, as is Acacia karroo. Seed and thorns were found in the brown humic sands in Area 1 and again seeds were recovered from the equivalent horizon in Area 2. These were isolated chance finds made by the breaking up of blocks of deposit and the quantities recovered are small. In the layered sediments of the Balmoral Member in Area 1 a number of leaves were found representing several morphologically distinct types. The sample represents leaf fall along one margin of the Area 1 depression during the deposition of the post-unconformity sediments. It has not proved possible to identify the principle component of the fringing vegetation of the springs represented by the branching stems buried in the deposits. A fuller description of these remains is given in Wells (1970). The check list of woody plants collected from the immediate surrounds of the springs by M. J. Wells of the Botanical Research Institute and sectioned by the Forest Research Institute is given here. MJW 3160 Rhus longispina E. + Z. 3161 Pappea capensis E. + Z. 3162 Maerua caffra (DC) Pax. 3163 Scutia myrtina (Burm. F.) Kurz. 3165 Hippobromus pauciflorus (L.f.) Radik 3166 Acacia karroo Hayne. 3167 Ptaeroxylon obliquum (Thunb.) Radik 3168 Grewia occidentalis L. 3169 Pterocealastrus tricuspidatus Sond. 3170 Azima tetracantha Lam. 3172 Euclea undulata Thunb. Pollens A series of samples taken from the west wall of Cutting 1 in 1963 were studied by Seagrief (Inskeep, 1965) and pollens of Cyperaceae, Gramineae, Compositae and Leguminosae were identified in these samples. Subsequently further samples have been collected from the site for pollen analysis and the study of these is in progress. Pollens are not preserved in the upper deposits where oxidized and counts are low in the sandy facies. Samples from clayey partings in the brown humic sands and the grey black silts show relatively high counts. Bacterial attack is evident on some grains. Pollens of waters-edge plants dominate the counts in samples examined with sedges showing high frequencies in addition to fern spores. Counts of arboreal pollens are low although Compositae (a number of morphologically distinct groups) and Gramineae are well represented. 10. General Discussion This report is intended as a statement on the present understanding of the Amanzi Springs Acheulian occurrences. Firm conclusions on the occupants and the occupation at MJW 3173 Schotia afra (L.) Bodin. 3174 Cassine sphaerophylla O. Ktze. 3175 Olea africana Mill. 3176 Sideroxylon inerme L. 3177 Aloe africana Mill. 3178 Scolopia zeyheri (Neesapud E. + Z.) Harv. 3179 Portulacaria afra Jacq. 3180 Lycium campanulatum E. Mey. 3181 Capparis citrifolia Lam. 3182 Diospyros simii (O. Ktze.) De Winter. 3183 Allophylus decipiens (Arn.) Radik. 115 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 this stage are few in spite of fairly extensive excavations. The full potential of information to be gained from the site has yet to be realized, more data from the 1964-66 excavations are available than discussed here, and some aspects of the investigation such as the studies of the pollen profiles sampled and the detailed lithology of the deposits for example, have still to be completed. Again from the consideration of artefact sample size further excavation at the site would seem warranted notably in the extension of Cuttings 5 and 10. This would allow more extensive analysis and better definition of the Acheulian represented at the site than presented here. At present there is little direct evidence for the general ecological setting in which the Acheulian occupation took place. The suggestion that the vegetation was more mesic than at present requires confirmation (Wells, 1970). An obvious attraction of the site to Acheulian groups was the spring waters issuing from a vantage point overlooking the Coega Valley and the site is close to a large salt pan. The number of discrete artefact occurrences associated with the springs points to multiple occupations over a time period. Other Acheulian occurrences are known from the Coega and adjacent valleys including the Sundays River sites (Ruddock, 1957) and in an upland area near Groendal Dam (E. Speed pers. comm.) however these are undated surface sites and are not necessarily the temporal equivalent of the Amanzi occurrences. There is evidence supporting occupation within the two spring depressions investigated and the micro-environment presented by these is of consequent interest. Both Area 1 and Area 2 are open ended depressions with drainage outlets directed down slope. In this they differ from the circular ironstone rimmed depression adjacent to the Edwards bore-hole which has no natural directional outlet. At the latter the sunken area clearly corresponds to the extent of a former spring pool. In Area l the plant remains representing a fringing aquatic community and the accumulation of drift wood mark the position of a sometime spring pool over the domed white sand body in the centre of the depression. The extent of such a pool would have been dependent on the height of the outlet and subject to fluctuation in this unstable environ- ment. Occupation indicated by denser artefact scatters appears to have been marginal to such a pool. A change in the transporting power of the spring discharge is evidenced in the grading analyses of the deposits (Table 2) at a stage which corresponds to the termination of the Acheulian occupation within the depression. At the onset of conditions leading to the deposi- tion of the grey black silts, effective drainage of the depression was apparently not maintained and suitable surfaces for occupation no longer existed. In Area 2 there is less evidence of the position of any spring pool during the period of occupation although such may have existed in close relation to the sub-outcrop of the white sand body, central to the depression. Possibly the advantage offered by the location of occupation areas within the depression was protection from high winds which funnel up the valley in an otherwise exposed locality. The spring deposits in themselves have proved of interest not in the least because there are structural and sedimentological features to which parallels can be drawn at other mound spring occurrences of archaeological or palaeoecological import. Within the Amanzi Springs Formation, three members are recognized, the Enqhura Member, the Rietheuvel Member and the Balmoral Member. It has proved necessary to restrict the term Rietheuvel Member to the two sub-units comprising the main artefact horizon, the grey black silts and the underlying brown sands as the results of further sedimentological studies (Butzer, pers. comm.) have indicated that these sub-units are lithologically distinct from the basal sediments. The Pleistocene dating of the spring deposits expected on the grounds of the included cultural material, is confirmed by the radiocarbon dates available. The age determined for a sample of wood (GrN — 4407 and 4546) from the wood rich zone in Cutting 1 by an enrichment process is considered a minimum age for the sample (Vogel and Waterbolk, 1967). This dating refers to the base of the artefact zone and some cultural material is of equivalent age as evidenced for example by the single flake found on the same surface as the wood in Cutting 1 116 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS extension and sealed between the two clay marker bands in the Rietheuvel Member. The apparent age of the Balmoral Member (I — 2241) again falls within the Upper Pleistocene. The Acheulian phase represented at Amanzi is characterized by the relatively heavy and unstandardized form of the artefacts. It is this variability in the attributes of formal tools that suggests the whole is a late or terminal Acheulian. There is a dearth of adequately des- cribed or dated Acheulian sites in the eastern Cape region. One of the most prolific sites is that at Geelhoutboom (24° 23' E. ; 34° 7' S.) where the artefacts lie in a series of deflation platforms in a ferruginized dune area on the outer edge of a truncated marine peneplain some 700 feet above the Tsitsikama coast (Laidler, 1947). Exposure may represent partial and relatively recent destruction of the vegetation cover in the area and the artefacts are not appreciably weathered. Later cultural material is found in some deflated portions of the site and no in situ horizons have been recorded of Acheulian or later material. The range of Acheulian artefacts in quartzite shows a high degree of finish in contrast to the Amanzi finds and this site may represent an earlier mature expression of the local Acheulian. The well illustrated report on the Wagenmakers Vallei sites (Malan, 1939) indicates that occurrences with some typological similarities to the Amanzi site may be isolated within the Cape Folded Belt to the west. The dating, typology and the cultural status of what has been termed the Natal Sangoan are not well known and further investigation of the Natal occurrences would be of considerable interest to South African archaeology. A limited selective sample from a Tugela Valley occurrence (Albany Museum Accession No. 66/1) available for comparison shows an emphasis on pick forms that contrasts strongly with the Amanzi samples. It is premature to attempt wider correlations at this time when it is impossible to relate the Acheulian sites in the same and adjacent valleys on the data available. No broad picture of the development of the handaxe culture in South Africa exists outside the Transvaal where, as elsewhere in southern Africa, it is proving difficult to demonstrate the temporal relation- ships of these early sites. The Amanzi occurrences form part of the Acheulian cultural develop- ment and our understanding of them will become clearer with the general advance in studies of this cultural period in South Africa. ACKNOWLEDGEMENTS This investigation of the Acheulian occurrences at the Amanzi Springs was carried out as a part of a programme of research into the prehistory of the Eastern Cape initiated by the Albany Museum. Financial support from the Wenner-Gren Foundation, New York and additional funds to cover dating costs from the C.S.I.R., Pretoria, are gratefully acknowledged. The kind co-operation and hospitality of Mr. and Dr. J. P. M. Niven and Mr. and Mrs. P. N. F. Niven of Amanzi Estates greatly facilitated work at the site. Mr. J. P. M. Niven generously arranged for the main area of interest to be fenced and conserved. I am grateful to Professor R. R. Inskeep for an introduction to the problems posed by the site in 1963 and for encouragement in the follow-up investigation. During the course of the field work and in subsequent visits to the site, I benefited from discussions with a number of specialists, Mr. J. P. H. Acocks of the Botanical Survey, Dr. K. W. Butzer of the University of Chicago, Dr. J. D. Clark of the University of California, Berkeley, Mr. F. L. Farquharson of the Albany Museum, Dr. C. M. Keller of the University of Illinois, Dr. R. G. Klein of the University of Washington, Mr. J. A. H. Marais of the Geological Survey, Professor E. D. Mountain and Mr. A. Ruddock of the Geology Depart- ment, Rhodes University. Dr. S. C. Seagrief of the Botany Department, Rhodes University and Mr. M. J. Wells of the Botanical Survey. My thanks are due to Mr. C. Burton for assistance in the field, to Professor Mountain for arranging for grading analyses to be carried out in his department and to Dr. Seagrief 117 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 for undertaking the initial examination of the deposits for pollens. Dr. Clark kindly commented on an earlier draft of the manuscript. Further I am indebted to Mr. M. J. Wells for the analysis of the macroscopic plant remains (Wells, 1970) and Dr. K. W. Butzer for undertaking a study of the spring deposits (Butzer, in preparation). The Secretary for Forestry, through the Forest Research Institute agreed to section some finds of wood from the site. This has been of value in indicating the potential of these finds for interpretation and the help and advice of Mr. Kromhout of the Forest Research Institute, in particular, is much appreciated. The photomicrographs illustrated in Plate 13 are reproduced by permission of the Institute. Mr. R. C. Walsh provided a sample of Natal Sangoan artefacts to facilitate comparison with the Amanzi material. The figures and artefact illustrations were kindly prepared for publication by Mrs. J. Deacon and my thanks are due to Mr. Farquharson for help in the preparation of the photo- graphic illustrations. This report is based in part on an M.A. thesis submitted to the Archaeology Department of the University of Cape Town in 1966. 118 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS REFERENCES Acocks, J. P. H., 1953. Veld types of South Africa. Mem. of the Union Botanical Survey, 28. Pretoria. Barrow, John, 1801. Travels into the interior of Southern Africa in the years 1797 and 1798. London. Bishop, Walter W. and Clark, J. Desmond, 1967. Background to evolution in Africa , University of Chicago Press. Bond, G. W., 1946. A geochemical survey of the underground water supplies of the Union of South Africa. Mem. Geol. Survey, 41 Pretoria, 1 — 208. Caton-Thompsqn, G., 1952. Kharga Oasis in Prehistory. London, University of London, 213 pp. Clark, J. Desmond, 1964. “The influence of environment in inducing culture change at the Kalambo Falls prehistoric site.” S. Afr. Archaeol. Bull., 19 (76), 93—97. Clark, J. Desmond and W. Bishop, 1966. Report on Conference on the Systematic Investigation of the African Later Tertiary and Quaternary. Current Anthropology 7 (2), 253 — 256. Coetzee, J. A., 1967. Pollen analytical studies in east and southern Africa. Palaeoecology of Africa 3 1—146. Dixey, F., 1966. Water supply, use and management, in Hills, E. S. ed., Arid Lands. Methuen & Co., London, 77 — 102. Dreyer, T. F., 1938. The archaeology of the Florisbad deposits. Arg. Nav. Nas. Mus. Bloemfontein, 1 (8), 65—77. Du Toit, A. L., 1928. The underground water resources of the Uitenhage region, in Haughton, S. H. The geology of the country between Grahamstown and Port Elizabeth. Geol. Survey Publ. Govt. Printer , Pretoria, 36 — 45. Du Toit, A. L., 1939. Geology of South Africa. Oliver and Boyd, London, 2nd Edition. Engelbrecht, L. N. J. et al 1962. Die geologie van die gebied tussen Port Elizabeth en Alexandria, Kaap- provinsie. Geol. Survey Publ. Govt. Printer, Pretoria, 1 — 49. Fagan, B. M. and Van Noten, F. L., 1966. Wood implements from Later Stone Age Sites at Gwisho hotsprings, Lochinvar, Zambia. Proc. Prehist. Soc., 22, 246 — 261. FitzPatrick, Sir Percy, c. 1924. “ Amanzi: A private record of the first decade.’’'’ Unpublished memoir. Gabel, Creighton, 1965. Stone Age Hunter of the Kafue. Boston, Boston University Press, African Research Studies, No. 6. Goodwin, A. J. H., 1933. “Some developments in technique during the Earlier Stone Age.” Trans. Roy. Soc. S. Afr. 21 (2), 109—123. Hall, A. L., 1938. Analyses of rocks, minerals, cores, coal, soil and waters. Mem. Geol. Survey, 32 465 — 516. Haughton, S. H., 1928. The geology of the country between Grahamstown and Port Elizabeth — an ex- planation of Cape Sheet No. 9 (Port Elizabeth). Geol. Survey Publ., Pretoria, Govt. Printer. Haughton, S. H., 1963. The stratigraphic history of Africa South of the Sahara. London, Oliver and Boyd. Howell, F. C., Cole, C. H. and Kleindienst, M. R., 1962. Isimila an Acheulian occupation site in the Iringa Highlands, Southern Highland Province, Tanganyika, in Mortelmans, G. ed. Actes du IVe Congres Pan-Africain de Prehistoire (Leopoldville 1959), Tervuren, 43 — 80. Inskeep, R. R., 1965. Earlier Stone Age occupation at Amanzi — A preliminary investigation. S. Afr. Journ. Sci., 61 (6): 229— 242. Kent, L. E., 1950. The thermal waters of the Union of South Africa and South West Africa. Trans. Geol. Soc. S. Afr., 52 231—264. Kleindienst, M. R., 1962. Components of the East African assemblage: an analytical approach, in Mortel- mans, G. ed. Actes du IVe Congres Pan-Africain de Prehistoire (Leopoldville 1959), Tervuren, 81 — 112. Laidler, P. W., 1947. The evolution of Middle Palaeolithic technique at Geelhoutboom, near Kareedouw in the southern Cape. Trans. Roy. Soc. S. Afr. 31 (3), 283 — 313. McBurney, C. B. M., 1960. The Stone Age of Northern Africa, London, Penguin Books. Marais, J. A. H., 1964. Preliminary report on the geological-geophysical investigation of the Uitenhage Artesian Basin. Unpublished report of the Geological Survey, Pretoria. Malan, F., 1939. The Stellenbosch industry in the Wagonmakers Vallei. Trans. Roy. Soc. S. Afr., 27 (3), 241—286. Mason, R. J., 1962. Prehistory of the Transvaal, Johannesburg, Witwatersrand University Press. Meiring, A. J. D., 1956. The macrolithic culture of Florisbad. Researches of the National Museum, Bloem- fontein, 1 (9), 205 — 237. Oakley, K. P., 1957. The dating of the Broken Hill, Florisbad and Saldanha skulls, in Clark, J. D. ed. The Proceedings of the Third Pan- A frican Congress on Prehistory (Livingstone 1955), London, 76 — 79. Oakley, K. P., 1954. Study tour of early hominid sites in southern Africa. S. Afr. Archaeol. Bull., 9 (35), 75—87. Rogers, A. W., 1909. The Zwartkops Bore-hole. Fourteenth Ann. Report of the Geol. Comm., 110 — 116. Ruddock, A., 1947. Terraces in the lower part of the Sunday's River Valley, Cape Province. Trans. Roy. Soc. S. Afr., 31 (4): 347—370. 119 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Ruddock, A., 1957. Relation between Chelles-Acheul implements and Quaternary river terraces of the Coega and Sundays River. S. Afr. Journ. Sci., 53 373 — 377. Sampson, C. G., 1965. A preliminary report on the Luano Spring deposits, Northern Rhodesia. S. Afr. Archaeol. Bull. 20 (77), 29 — 33. Van Noten, F., 1965. Nouvelles fouilles a Lochinvar (Zambia) 1964. Africa-Tervuren 11 16 — 32. Van Zinderen Barker, E. M., 1964. Palynology in Africa: Eighth Report on Palynology 1962 and 1963, Bloemfontein. Van Zinderen Barker, E. M., 1967. A pollen analytical investigation of the Florisbad deposits (S. Afr.) in Clark, J. D. ed.. The Proceedings of the Third Pan- African Congress on Prehistory , Chatto and Windus, London, 156 — 167. Vogel, J. C. and Waterbolk, H. T., 1967. Groningen Radiocarbon dates VII. Radiocarbon, 9 107 — 155. Wells, M. J., 1970. Plant remains from Amanzi Springs. Ann. Cape Prov. Mus. (Nat. Hist.), In press. 120 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS LOCATION OF AMANZI SPRINGS Fig. 1. 121 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 AMANZI HILL WMHM r- " -l-.:,. J 0 1 2 3Vft* Fig. 2. 122 I I DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS AMANZI : AREA 2 Furrow course DUMP AREA FEET Fig. 5. 125 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11 DECEMBER 1970 lj_ ** -«-» -O W — O ^ O T) D +* 3 E r- bh E cn c « ■ f\i£ o 2 =3 ) i- < 126 ENOHURA MEMBER DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS AMANZI Area 1 Cutting 1 Plan of 1963 Excavation by Inskeep -4' 6" AREA OF FINDS OF CULTURAL MATERIAL DEPTHS REACHED IN FEET BELOW DATUM DATUM O' 0 521-77' a.s.l. Fig. 8. 127 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 AMANZI Plan of 1964-5 Excavation Areal Cutting 1 datum ® 0 feet 1 2 3 4 5 6 7 8 9 10 11 12 feet AREA SAMPLED AS PART OF CUTTING 10 10 ii CUTTING 1 Ext. :i r \ \ CUTTING 1 Deep Sounding Maximum depth 16 ft. )6U feet CUTTING 1 Rear Section Maximum depth 11 ft. —TRa£S o^Un 7 Oft** CUTTING 1 Frontal Section Maximum depth 11 ft. B A U L K Fig. 9. 128 E 2 SQUAF r ABOVi 523 -W 25 FEET AMANZI : AREA 1 Section across depression B A CUTTING 12 SQUARE 2 SQUARE 1 CUTTING 1 r ABOVE SEA LEVEL FEET BELOW DATUM MADE GROUND SZZZ3 WHITE SAND Fig. 6. DEACON: THE ACHEUUAN OCCUPATION AT AMANZI SPRINGS AMANZI Spatial distribution of finds in Area 1 Cutting 1 : Frontal section Boundary of disturbed ground Potholes with high concentration of finds Position of finds Fig. 10. 129 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 CUTTING 1 /I J AMANZI Cutting 1 and Extension Distribution of finds of Wood ROOT BASES LONG PIECES OTHER PIECES FEET 2 3 4 5 6 7 8 9 10 11 Deep Soun ding CUTTING 1 Rear Section w w WW^ / wj Ww Cc§> / „ ) W w W w w w w r w. ; SECTION : Vertical distribution FEET BELOW DATUM Fig. 11. 130 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS AMANZI Areal Cutting 1 Extension Plan feet -10 12 cutting i DEEP SOUNDING CUTTING 1 EXTENSION Section (b) Section (a) 7 81 9 K) 11 feet below datum LAMINATED SEDIMENTS POTHOLE FILL ((<< 20* 10- 0- 89 SAMPLE 3 n = 290 i w -Q i O Ul U <0 U A UJ— tr© < V ^ o Ul sv in UJ tr O u > z < V) _l O O u _l GO £0 O U if) UJ X < a z < i m tr u > < ui _i u UJ o (X < —I a) x< l— — O CO if) o If) If) _l o o oc UJ X 75 80 - 11 B 14 6 4 3 . [ > X 1 i 1 ..li at CO UJ U >64 m.m. CO _J A e CO XI -0 O O H CO X) X) O k 't £: o O »- _l <■ O “ o U co XI _Q UJ X < 4) XI ft 3 w c u x t _i > (0 CD < Z 4) 0. _l u. & £ <©c O u z < O U CO UJ X < 0 z < 1 CO a. ui > < UJ _J CO UJ u if uj 5g * 3 UJ w X CO u CO 5 CO 8 ce UJ x uj ^ to * E d w nil ^ o c Jq UJ £ _J □ d E Uj I? SI Fig. 18. 136 DEACON: THE ACHEULIAN OCCUPATION AT AMAN2I SPRINGS AMANZI Bar Charts of Finds in Area 2: Cuttings 5, 6,& 8 Fig. 19. 137 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 138 Plate 1. AMANZI HILL. The arrow indicates the position of the excavations on the northern flank of the hill. The vegetation is typical Addo Bushveld with the tall trees on the sky-line and the prickly pear in the foreground representing recent plantings or invasives. DEACON: THE ACHEULIAN OCCUPATION AT AMANZl SPRINGS 03 i_ d> C ao jc T3 a. c 03 C _o 35 on Vh a *o 03 -C o c r? m N 3 Z "o < a 5 .2 ^ c3 . > ™ s W X H 3 > u jy 139 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 3. AMANZI: Area 1, Cutting 10. The north wall of Cutting 10 showing the exposed sequence of 11-5 feet of deposit. Iron- stone caps the sequence which shows some ferruginization to six feet above the base. The string line above the ranging pole indicates the top of the artefact accumulation which extends to the base of the cutting. The artefact accumulation was sampled to a depth of 9 feet below datum in a series of blocks, with alternate blocks deepened to 11-5 feet. The two blocks seen here are 0-6/0-5 (—11-5 feet) and 0-6/5-10 ( — 9 feet). Ferruginization obscures the top of the grey-black silts and the contact with the underlying greenish yellow clayey sands is gradational between —8 -5 feet and —9 feet. A sand stringer runs through the greenish sands a foot above the bottom of the section. 140 MANZI : CUT' of find tUUli? apers 20 AMANZI AREA 2 : CUTTING 6 Plot of finds Section Margin of spring vent ° Natural stone - Cobbles and pebbles C Cores • Handaxes X Other tools and Trimmed pieces ' " — - Margins of hollows • Flaked cobbles and pebbles A Anvils » Cleavers - Untrimmed flakes />-' Finds of wood Z Irregular pieces ▲ Cobble tools ■ Other bifaces ♦ © Trimmed flakes / Scrapers Fig. 20 FEET BELOW DATUM H I CUTTING CS ; Ci.ii -c c: 2 S-s'S S?i8f Cl g SO.d C ??'S £ §.S s d£ . g ''Ice ,£d * -a *- ^ ^ C il§° ° * _c 1-T - D *_. a> O > <■«-> 5 ^ i> i Z.B £ ^ ^ t>0 o "°0£ -n Cl 5is-s w-,— i> a c £ O c/5 X — « • - «,£ o u oj O u o £ Cl d _ £ ‘d aj 15 d °-£>- d w ft v) 8 J-S s i’d u ^ a> <« ?>. 4> So _. y <4—1 * d -o ^ rrt o3 a 141 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 I ' 03 N Z < ^ 2 > c5 c- g S ?■ i/i rt M 3 / 142 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS 143 A piece of wood from Cutting 1, Extension in the brown sands. The twisted grain of the wood suggests this is a root stock. ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 7. AMANZI: Area 1, Cutting 10. 10-15/35-40. Within the artefact accumulation some groups such as this occur. Several pieces are on “edge”. 144 DEACON: THE ACHEULI AN OCCUPATION AT AMANZI SPRINGS 'f ? 145 Plate 8. AMANZI: Area 2. A general view of the excavation in progress. In the foreground in Cutting 6 and 8 the white sands are exposed (Surface 1) and some cultural material is seen on this surface. Small depressions in this surface are evident. Beyond the low baulk (in the channel fill) is Cutting 5 and an extension as a series of steps up the side of the depression. ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 9. AMANZI: Area 2: Deep Sounding. The deep sounding in Area 2 was excavated below Surface 1 in Cuttings 6 and 8 seen in the foreground of the previous plate. Two surfaces were intersected and the artefact occur- rence on Surface 3 can be seen. The slope of these stone lines show the context of the cultural material to be wholly disturbed. The brown humic sands thicken in one corner of the sounding on the edge of a spring vent. Acheulian artefacts were excavated from the base of the brown humic sands and on Surfaces 1-3. 146 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 10. AMANZI: Area 2: Deep Sounding. Surface 3 has the appearance of a cobble pavement with natural stone and artefacts packed on a defined surface. 147 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 148 Plate 11. AMANZI: Area 2. A section cut in the side of the cleaned out furrow, 60 feet towards the outlet from Cutting 5. The grey-black silt (Rietheuvel Member) shown as a white band in this photograph is truncated at either end by later unconformable deposits (Balmoral Member). Cultural material and natural stone can be seen in the top of the underlying deposits. DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS n o> 03 "O C/5 £3 £ d> CJ c IE 7Z N £ < D. O -*-* d> .3 > 50 .E *C D, C/5 d> 4= H-H o d> 50 T3 « &; d> h- ■s p° d o cA t-l c3 „ O ^ 03 O O V>oo fcH (N ' -2+ 1 TJ C © C3 O *o "T 8 S £ 2 149 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 13. AMANZ1: Wood. Photomicrographs of four transverse sections ( X 20) of wood from the site prepared by the Forestry Research Unit, Department of Forestry, Pretoria. The specimen on the top right is a section of a woody plant with a marked pith that is found with the herbaceous plants in Area 1 as part of the fringing vegetation of the fossil spring. The other examples can be generally classified as hardwoods (Dicotyledenous woods), but the absence of any key to the identification in section of indigenous trees and shrubs and some collapse of cell structures makes positive identification impossible at this stage. 150 DEACON: THE ACHEULLAN OCCUPATION AT AMANZ1 SPRINGS • fa T3 r- fa X C/3 o c a O X fa tJj fa .5 5h 03 >> £ X Ofl § w 3 fa fa 3 O 3 ° 2 S2 fa 8 8 8 .a x fa Jm 3 fa 3 x cr _ 3 3 O £ U fa fa O- bh o £ X £ -M* 0,1 ■fa 00 3 fa c/3 fa-i •fa o 5j fa fa fa x : J-l fa o • **— < s— > X fa fa fa o =* o fa u x -<— > fa 0 X) ° X £ O fa +-> C/3 C-- fa fa fa £ s-g fa fa X X 3 X O 60 .£ 'C a C/3 3 O fa >?x fa '3 fa K3 X 3 c C/3 O X 0 ■M 3 151 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 15. AMANZI: Area 2. Wood from the Deep Sounding between Surfaces 1 and 2; this is the only piece of wood found in the excavations which shows any features which might evidence working ( x 1 approx.). 152 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS CIO/314 E in CIO/328 5/546 Plate 16. AMANZI: Cores. 1. (C10/314E). Radial core with extensive cortex retained on the under surface, from Cutting 10, Area 1. 2. (Cl 0/328). A disc-like core, an intermediate form between the radially flaked cores and a disc tool. From Cutting 10, Area 1. 3. (5/546). Irregular core on a cobble, from Cutting 5, Area 2. 153 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 17. AMANZI: Cores and Disc. 1. (5/545). Radial core with a biconical form. Cutting 5, Area 2. 2. (Cl 0/21 1). Disc. A class considered separate from the radially flaked core sub- class only on the degree of flaking on the perimeter which may indicate these artefacts were shaped tools. The frequency of discs in the samples is low. From Cutting 10, Area 1. 3. (6/594). Struck core; from Cutting 6, Area 2. 4. (5/562). Small radial core of the type which could be alternatively classified as a discoid. The pattern of flaking on the upper and lower faces is essentially similar to that of Plate 17, No. 1 and these implements are thus grouped with the radial cores. 154 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 18. AMANZI: Cobble Tools. 1. (5/261). An example of a scoop-like cobble tool. There are a number of such cobbles showing a single or rarely more flakes struck in the line of the long axis of a cobble so as to produce a unifacial edge of the type shown. These exhibit medium damage. From Cutting 5, Area 2. 2. (5/333). Unifacial chopper-like cobble tool: bifacial examples are also found and these are essentially large heavy tools with indications of medium to very heavy edge damage. From Cutting 5, Area 2. 155 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 156 Plate 19. AMANZI: Cobble Tool. (C 10/321) approx, x f . A piece classified as a cobble tool, but showing the typical primary flaking on the under face of biface manufacture. On this specimen the face not shown is cortex and there is extensive utilization on the flaked edge. From Cutting 10, Area 1. DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS ~o c ^ 03 O T3 03 1> 3 _ o Uh -0 03 X Cl Cl rj p 0. O 0^-0 = a-c c | ° as > co .2 H ju 3 X) o o <4! 03 N o X p co ^ ~ (U !> |x3 _Q -go 3 3 v o T <*- x «* ^3 r- TtJ2“ a> *- fioj 0 ~ = a-g E co .2 >> •p -t- CO 2 03 „ - aj "o 2 o 2 p ~ OJ X O co ~ O £ £ a> 2 2 2 «* -o © 2 P -C 03.^2^ 1- X <■«-. U u (3 >x E o c Xc*H 1) 03 O X ^P-O 03 .2 E O ^ C3 X co -*- ^°4h £ a. o p W Oj a x p 0 ^ p * >'S ’S 1 “ > 0 • X fc — ' x ^ x <2 ^ - 53 m -o p ro, p .~ O <*S w> U * 5 c3 co X 5 03 157 primary flaking. From Cutting 10, Area 1. ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 6/293 Plates 21-22- AMANZI: Handaxes. 158 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS 5/310 Plates 21-22. AMANZI: Handaxes. (8/306, 16/250, 6/292, 6/293, 5/310). A series of five small handaxes which are on the lower limit of the size range in the class. These tools show a wider range of point trimming than in the handaxe class as a whole. The point in 6/293 has been retrimmed, presumably after breaking in manufacture or use. From Cutting 8, 6 and 5, Area 2. 159 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 23. AMANZI: Handaxe. (5/802). A common long ovate or pear-shaped tool showing a thick minor section. The flaking on both faces is extensive and the tool is symmetrical about both axes. Cutting 5, Area 2. DEACON: THE ACHEULIAN OCCUPATION AT AMANZ1 SPRINGS 5/179 = ^ Plate 24. AMANZI: Handaxe. (5/179). In shape, this tool is on the extreme edge of the range towards the ovate form. The tip is broken and there is some emphasis on the lateral trimming along one edge. From cutting 5, Area 2, 161 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 25. AMANZI: Handaxe. (5/476d) x f. This is a common shape of handaxe and the flaking, though minimal, is over the whole of both faces. The point shows the three-flake scar pattern of the upper surface. The edges show little or no fine retouch and are formed by the intersection of the primary, shaping scars. The edges are thus not fine and they show medium damage. From Cutting 5, Area 2. 162 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 26. AMANZI: Handaxe. (CIO/337) x f. A handaxe made by the half-cobble technique showing the typical patch of cortex on the one side of the upper surface that is rarely removed. The keel to the too! is also characteristic. From Cutting 10, Area 1. 163 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11 DECEMBER 1970 Plate 27. AMANZI: Handaxe. (6/463) x approx, f. A lanceolate handaze, one of two similarly well finished specimens from Surface 1, Area 2, it is on the extreme end of the range towards the lanceolate form. 164 DEACON: THE ACHEUL1AN OCCUPATION AT AMANZI SPRINGS Plate 28. AMANZI: Handaxe, (5/406) xf. Ao uncommon form of handaxe made on a side-struck flake. From Cutting 5, Area 2. 165 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 , * Mf«»| - V‘ Plate 29. AMANZI: Handaxe. (5/547) xf. Handaxe on a side-struck flake, there is little retouch on the edges of the tool. From Cutting 5, Area 2. 166 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS ^ns Plate 30. AMANZI: Cleavers. 1. (5/307). Classified as a cleaver and showing a basic cleaver shape with the transverse edge notched, this is one of several similar tools from Area 2 which show this feature and suggesting use of these tools for some purpose outside the range of cleavers generally. From Cutting 5, Area 2. 2. (00/309). Trapezoid plan form cleaver from Cutting 10, Area 1. 167 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 31. AMANZI: Cleaver. (5/313). One face of an end-notched cleaver, the notching here being due to the mode of usage which would appear to have been chopping rather than cutting. From Cutting 5, Area 2. 168 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 32. AMANZI: Cleaver. (CIO/323) x f . An unstandardized form on a side-struck flake. The transverse end shows some trimming and unifacial trimming extends up the right lateral. From Cutting 10, Area 1. 169 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 33. AMANZI: Cleaver. (C10/305)xf. From Cutting 10, Area 1. 170 DEACON: THE AC HEULIAN OCCUPATION AT AMANZI SPRINGS 1 1 ■/ ■ Plate 34. AMANZI: Cleaver. (6/199) xf approx. A large, heavy transverse-edged tool included in the cleaver class. From Cutting 6, Area 2. 171 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 35. AMANZI: Cleavers. (2/10, 2/12). These two tools were found together with a larger cleaver and a handaxe on the white sands in Area 1, Square 2. The context was not obviously disturbed and a functional association is possible. The probability of such an association occurring by chance is very low. The two cleavers are of interest in their extremely narrow blades. 172 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 36. AMANZI: Large Biface. (5/527) xf approx. Round bitted biface — a limande-shaped symmetrical tool with a thick minor section, a definite butt and a broad rounded “point” and well trimmed lateral edges. The point is more specialized than indicated by the term “bit”, being slightly beaked. From Cutting 5, Area 2. 173 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 37. AMANZI: Large Biface. (CIO/238) X f . A high-backed tool with the base a flat main flake surface, it lacks the symmetry of a handaxe and also the sharp lateral edges. The tool has been pointed by a series of flakes on the dorsal surface. From Cutting 10, Area 1. 174 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 38. AMANZI: Large Biface, (6/179) x This tool shows no trimming on the point, but extensive step flaking down the left lateral edge which would seem to be marked by medium to heavy damage. From Cutting 6, Area 2, 175 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 39. AMANZI: Large Biface. (16/292) X f. An elongated biface with an obtuse, thin-edged point and utilization damage extending around the point and along both laterals. From Cutting 6, Area 2. 176 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 40, AMANZI: Large Biface. (8/320) xf. An elongated biface with a steep, curved, trimmed point and fine lateral edge trimming. From Cutting 8, Area 2. 177 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 41. AMANZI: Large Biface. (5/593) x f. An edge-trimmed biface, asymmetrical and unpointed. From Cutting 5, Area 2. 178 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 42. AMANZI: Other Tools. (6/692, 7/19). Transverse-edged, flat based, highbacked tools or “wedges”. These illustrations show the primary flake form of this subclass, the trimming around the transverse edge and on the laterals. From Cuttings 6 and 7, Area 2. 179 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 43. AMANZI: Other Tools. (8/450) x f. An example of a small bifacial tool with trimming down one side which is more adze- like than what is classified here as scraper retouch. The term “adze” used to describe these tools is provisional and some other term with less functional overtones would be preferable. As used here the term has no functional implications. From Cutting 8, Area 2. 180 DEACON: THE ACHEULI AN OCCUPATION AT AMANZI SPRINGS 1. (6/221). Two faces of a large (greater than 100 mm.) modified flake. From Cutting 6, Area 2 . 2. (6/371). Two faces of a large primary modified flake. 181 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 A2 S3/5B Plate 45. AMANZI: Flakes. (A2S3/4A, A2S3/5B). Two large flakes from the Deep Sounding, Area 2, Surface 3, showing scraper retouch on the working edges. 182 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS Plate 46. AMANZI: Flakes. 1. (A2S3/5A). A large flake showing extensive shallow trimming on the lateral edge. From Surface 3, Deep Sounding, Area 2. 2. (A2S3/8A). Large flake with scraper retouched margin. 183 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 Plate 47. AMANZI: Flakes. (A2S3/7A, 4A and 8B). Three small scrapers on primary flakes from Surface 3 in the Deep Sounding of Area 2. 184 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS EEEzi r - — ~i ~r in 5/757 Plate 48. AMANZI: Flake Scrapers. 1. (5/630). Small flake side scraper from Cutting 5, Area 2. 2. (5/757). Small flake side scraper showing notching. From Cutting 5, Area 2, 185 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 5/582 Plate 49. AMANZi: Flake Scrapers. 1. (5/582). Small flake double-side scraper from Cutting 5, Area 2. 2. (5/652). Small flake double-side scraper from Cutting 5, Area 2. 186 DEACON: THE ACHEULIAN OCCUPATION AT AMAN2I SPRINGS CIO/238 Plate 50. AMANZI: Flakes. (C10/238, 337, 238). Three small flakes from Area 1, Cutting 10 showing scraper retouch and what is considered to be the typical flaking on the dorsal surface of flakes struck from a radial core. 187 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 11, DECEMBER 1970 CIO/337 2 CIO/238 C/10/337 Plate 51. AMANZI: Flakes. (Cl 0/337, 238, 337). Two small trimmed flakes (top and center) showing the variation in size and extent of trimming, and (bottom) an irregular flake with scraper retouch. From Cutting 10, Area 1. 188 DEACON: THE ACHEULIAN OCCUPATION AT AMANZI SPRINGS 6/389 8/324 Plate 52. AMANZI: Flakes. (5/509, 6/112, 6/165, 6/177, 6/389, 8/324). These are the range of flakes which, on typological grounds, are foreign to the Early Stone Age arte- fact samples and are considered intrusives. The samples are all from Area 2. 189 ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. {Nat. Hist) k VOLUME 8 • PART 12 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Plant remains from Amanzi Springs M. J. WELLS Botanical Research Institute, Department of Agricultural Technical Services The main plant remains from the Amanzi Springs Formation (Deacon, 1970) can be summarized as follows: — Enqhura Member 1. Six fruits or fruit segments, 5 — 7 mm. in diameter, flattened on one side, this side bearing an aril-like crown close to the point of attachment, and six portions of the same kind of fruit, all found in a localized area on Surface 3 in Area 2 (Fig. 2, No. 3). These are un- identified and not morphologically indicative of any particular vegetation type. The fruits are similar in size and shape, but not in detail, to fruit segments of Grewia occidentalis L. 2. Numerous, dichotomously branching herbaceous plants were found in Area 1 in the white sands and in the brown humic or herbaceous sands. In Square 2 (Fig. 1) a single com- munity was exposed in a face of white sands. The upper branched portion of this community appeared to be cut off against a minor parting near the top of the sands. All the stems and roots of the herbaceous plants in the white sands are “carbonised”, very fragile and disintegrate on removal from the matrix. The condition of the remains in the brown sands is somewhat better. No finer parts such as leaves or flowers are preserved in any horizon. This is possibly because of the coarseness of the matrix. The stems range in thickness from about 5 mm. at the base to 1 — 2 mm. at the top, branched section of the plants. As the plants are herbaceous and the aerial stems reach heights of 1 • 5 m., it is suggested that they could not have supported their spreading much branched tops unless surrounded by water, and that they represent an aquatic community. In the community in Square 2 the stems (Fig. 1) are not vertical, but lean at an angle as might be expected if growing in water. The community has been preserved by deposition around the individual stems and for this to have taken place, sedimentation would have been extremely rapid, probably within a single growing season. Rietheuvel Member 1. Numerous, disconnected, woody stems and root fragments, unidentified, but morpho- logically similar to those of dicotyledonous trees and shrubs at present growing near the springs. From the wide range of material found, including fragile twigs, it is construed that there must have been a strong woody element present in the vegetation at the time the Rietheuvel sediments were laid down. 2. Three prickles bases with raised rims, 8 — 12 mm. long and 5 — 7 mm. wide, one re- taining its prickle, which are morphologically indistinguishable from prickle bases which persist on corky outgrowths of the trunks of arborescent Erythrina spp. such as E. caffrci Thunb. and E. lysistemon Hutch, found in Cutting 1 Extension in the brown humic or her- baceous sands (Fig. 2, No. 4). Erythrina spp. do not occur in the immediate vicinity of the springs at present and are not characteristic of the bushveld community of the Amanzi area. E. caffra is an important con- 191 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 12, DECEMBER 1970 \ \ \ /iST VNV ' I N WV. /V' X Fig. 1 192 WELLS: PLANT REMAINS FROM AMANZI SPRINGS stituent of the dry coastal forest (Alexandria) into which the bushveld grades, and trees of the species occur in the bushveld where extra water is available, typically near rock outcrops and in stream bank bush. Cultivated specimens flourish in the homestead garden. No tree species of this genus grows in mountain forests of Southern Africa and most are limited in distribution to frost free areas (an exception is E. latissima E. Mey). There is thus some basis for suggesting that the climate of Amanzi was much as it is at present, but perhaps wetter, when these remains were included in the deposit. Balmoral Member 1. Thirteen portions of leaf or stem epidermis with individual pieces up to 2 sq. cm. in area, were found in Cutting 1 Extension in the pothole fill horizon. These sections of epidermis are made up of regularly arranged octagonal cells with thickened lateral walls. No stomata are present and no features suggesting the presence of specialised tissues beneath the epidermis which covered soft, sparingly veined plant tissues such as are present in the leaves or stems of succulents and hydrophytes. 2. Seven leaflets, 5 — 9 mm. long and 3 — 4 mm. wide, morphologically similar to those of Schotia afra (L) Bodin. were found in Cutting 1 Extension in the pothole fill horizon and in Area 2, Square 3, Quadrant 1, in material overlying the marginal clays (Fig. 2, No. 2). Leaflets of this type, hard and small, are characteristic of the bushveld and savanna woody species, such as Schotia spp. and Acacia spp. and are only occasionally found in forest margin species such as the climbers Entada spp., Dalbergia spp. and Acacia spp. Therefore, the finds of this leaf type may suggest that the vegetation in the vicinity of the springs included xero- phytic woody species as found in the savanna, bushveld or dry forest when these leaflets were incorporated in the sequence. 3. Three leaf portions of a fairly large-leaved dicotyledonous species, up to 1 sq. cm. in area, found in Cutting 1 Extension, Area 1, in the pothole fill. Unidentified. (Fig. 2, No. 1). Leaves of this sort are usually borne on woody trees and shrubs and are found in bushveld, although commoner in forest and hygrophilous communities. Since the area adjacent to the springs may well have supported a vegetation more hygrophilous than that of the area as a whole, no conclusions regarding the nature of the surrounding vegetation may be drawn from the presence of these leaf fragments. Discussion Scanty though the identified plant remains from the Amanzi site are, they allow some general statements on the Pleistocene vegetation to be made. There is good evidence of at least the periodic presence of an hygrophilous vegetation in the springs themselves. The finds of prickle bases of an arborescent Erythrina species similar to, if not identical with E. caffra , in the deposits of the Rietheuvel Member could indicate a local development more mesic than that at present in the area, or this could have been a more general condition. The leaf remains in the depositional units of the Balmoral Member suggest the presence of a woody community incorporating xerophytic components in the vicinity of the springs. In as far as these few remains can be interpreted as evidence of general climatic conditions, it would appear that during the Acheulian occupation and the deposition of the lower horizons, conditions were not drier than at present and during the later period of spring activity, the climate was perhaps as dry as at present. REFERENCES Deacon, H. L, 1970. The Acheulian occupation at Amanzi Springs, Uitenhage District, Cape Province. Ann. Cape Prov. Mus. 193 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 12, DECEMBER 1970 Fig. 2 194 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN Afcl 'V ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prow Mas. (Nat. Hist.) H VOLUME 8 • PART 13 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA The Occurrence of Hector’s Beaked Whale Mesoplodon hectori (Gray) in South African Waters G. J. B. ROSS (Port Elizabeth Museum) Abstract: Two skulls of immature Mesoplodon hectori (Gray) found at Tottering River mouth, South Africa (34° 00' S, 23° 45' E) are the first records of this species in South African waters. They are the fifth and sixth known specimens of the species. Generic and specific characters are discussed, together with a review of the species’ distribution. INTRODUCTION On 12 March 1967 the remains of two small beaked whales, consisting of the skull and broken mandibles and the broken skull and a single rib were found at Lottering River mouth. Cape Province, South Africa latitude 34° 00' S, longitude 23° 45' E. Of the soft parts, only the flukes of each and a single flipper remained. The skeletal material was collected for the Port Elizabeth Museum by Mr C. K. Tayler (PEM 1511/15, PEM 1511/16 respectively). As the apical portions of the mandibles containing the teeth were missing, identification on tooth shape and position was not possible, and the specimens were subsequently identi- fied by Dr Joseph Curtis Moore by means of measurements and photographs provided by the author, as juvenile Hector’s beaked whales. This is the first record of the species for South African waters, bringing the number of Mesoplodon species known to South Africa to five. The four species previously recorded are M. layardi , grayi and densirostris (Barnard, 1954) and M. mirus (McCann and Talbot, 1963; Ross, 1969). EXTERNAL FEATURES Though the one flipper and two flukes found with the specimens were not collected, photographs were taken of each with their respective skulls. As there are apparently no published records of fluke form in M. hectori , one photograph has been reproduced (Plate III). GENERIC CHARACTERS Since Gray’s (1871) description of the type of M. hectori , there has been considerable confusion as to the species’ taxonomic status, culminating in McCann’s (1962) paper, in which he considered M. hectori to be the juvenile of Berardius arnuxi Duvernoy. Moore (1968) however, in evaluating McCann’s argument, has presented evidence that M. hectori is in fact a distinct species in the genus Mesoplodon. The Lottering Mouth specimens confirm this view in the following points. 195 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 13, DECEMBER 1970 Moore states that the asymmetry of the skull of M. hectori is as great as that of M. mirus and M. grayi. The Lottering Mouth specimens conform in this respect in comparison with this museum’s three specimens of mirus and one grayi. Ness (1967) has measured skull asymmetry in the toothed cetacea in degree of “skew”, as the maximum deviation in the “mid-line” suture from the rostral tip to foramen magnum axis in relation to skull length. He published no measurement for M. hectori , but (in a letter dated 3 November 1969 to the author) reports figures for one specimen of hectori comparable to those he obtained for M. bidens , densirostris , europaeus , grayi and mirus. In his generic diagnoses, Moore (1968) listed nine characteristics of the genus Mesoplodon. The Lottering Mouth specimens conform on all of the six not restricted to adult specimens. pit Figure I. The superior nares of M. hectori (PEM 1511/15) in antero-dorsal view showing the position of the nasal pits, me mesethmoid; n nasal; p premaxilla; pc premaxillary crest; pit nasal pit. 196 ROSS: HECTOR’S BEAKED WHALE SPECIFIC CHARACTERS In dorsal view, the posterior extremities of the premaxillae are expanded laterally in the Loitering Mouth specimens as much as those of the Falkland Island specimen (Fraser, 1950). The impression of a greater lateral expansion in these two new specimens, however, is due to the narrowness of the vertex posterior to the premaxillary crests (see Plates 1 and II). In anterior view the greatest width across the crests is equal to, or slightly less than that across the premaxiliaries at the level of the anterior border of the superior nares. The three illus- trated juveniles (Flower, 1878, pi. 71, fig. 4; Fraser, 1950, pi. 3; McCann, 1962, pi. 2), the Adventure Bay, Tasmania, specimen (Moore, in litt.) and the two present specimens conform in this character. This character thus distinguishes all known specimens of hectori from the specimens of M. mirus (3), M. gray i (1), M. densirostris (1 ), and M. layardi (1) in this museum, and M, layardi (8) in the South African Museum, and from the 4 M. europaeus illustrated in Raven (1937), Fraser (1955), Rankin (1956), Moore (1960), the 1 M. bidens illustrated by True (1910) and Moore (1966), the 5 M. stejnegeri illustrated by True (1910), Nishiwaki (1962) and Moore (1966), the 1 M, carlhubbsi illustrated by Moore (1966), the 1 M, ginkgodens illustrated by Nishiwaki and Kamiya (1958), and the 2 M . bowdoini illustrated by Andrews (1908) and Oliver (1922). Because the six known specimens of M. hectori are so distinguished from all of these small samples of the other species, the present author suggests that the relatively narrow width of the premaxillary crest is diagnostic for the species M, hectori . McCann (1962) noted a blind pit on the posterior wall of each narial passage in the New Zealand M. hectori. This pit is present in both of the Lottering Mouth specimens (Fig. I). It is situated on the lateral expansion of the mesethmoid bone contributing to the posterior narial wall, and consists of the covered apex of a depression that deepens and narrows dorsally. No function could be assigned to it. Though McCann stated that this pit is absent in other Mesoplodon species, a similar, and perhaps homologous structure was found in specimens of M. densirostris (1), M. layardi (1) and M. grayi (1) in this museum. In M. densirostris the apex of the pit was directed postero-ventrally and contains a single foramen. In M. layardi the apex of the pit was directed posteriorly, while in M. grayi the pit was a hemispherical concavity with an indistinct apex. In three specimens of M. mirus there was a very shallow groove directed dorso-ventrally in this region. Though no other species were examined, the shape of the pit and the position of its apex in M. hectori may yet be shown to be a specific character for this species. The measurements of the Lottering Mouth specimens shown in Table l, agree fairly closely with those of the two New Zealand specimens (Titahi and Plimmerton) and the Falk- land Island skull, and the skull features used by Fraser (1950) are similarly shown by the two present specimens. The sizes of these five specimens suggest that they are all less than eighteen months of age, a period of rapid growth in cetaceans. With the assumption that sexual and individual variations in measurements are minimal at this age, the differences in total skull length may be used to group the skulls in order of increasing age. On this basis the youngest is the Plim- merton specimen and the oldest the Falkland Island specimen, while the Titahi and the two present specimens are of intermediate age. While the close similarity in measurements (Table I) and the circumstances of the stranding of the present specimens suggest that they are the same age, the Titahi and Plimmerton skulls, though both stranded in January, may well be of different ages, since the time and length of the breeding season is unknown in this species. If these groupings are correct, then a comparison of measurements 1 and 2 in Table I indicate that an increase in the total length of the skull is primarily an increase in the length of the rostrum. The length of the skull posterior to the rostrum changes relatively little. In addition, measurements 18, 19 and 20 as percentages of total skull length show a decrease 197 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 13, DECEMBER 1970 Table I. Skull and mandible measurements of immature M. heetori (Gray) (in millimetres) NO. P.E.M. 1511/15 P.E.M. 1511/16 TITAHI PLIM MERTON FALKLANDS mm % mm % mm %* mm %• mm O/ * /o 1 587 100% 278 + — 587 ± 100% 505 100% 601 100% 2 340 57-9 — — 343 58-4 280 55-4 358 59-6 3 — — — — 443 75-5 383 75-8 469 78-0 4 — — — — — — — — — — 5 280 47-7 — — — — — — — — 6 360 61 - 3 — — — — 7 535 91 • 1 — — — — — — — — 8 399 67-9 — — — — — — — — 9 435 74-1 — — — — — — — — 10 535 91 - 1 — — — — — — — — 11 460 78-3 — — — — — 12 305 51 -9 — — — — — — — — 13 85 14-4 110 — — — 84 16 - 6 — — 14 91 15-5 93 — — ■ — - — — — — 15 42 7-1 37 — — — — — — — 16 ±61 10-2 + 50 — — — — — 17 265 45-1 260 — — — — — — — 18 260 44-2 264 — 258 440 238 47-0 242 40-6 19 238 40-2 245 — 228 38-8 207 41 0 228 360 20 195 33-2 193 — — — 197 40-0 — — 21 93 15-4 95 90 15 3 93 18-4 91 15 2 22 32 30 38 — — — — — — — 23 67 11-4 65 — — — 59 11-7 — — 24 39 6-6 43 — — — 38 7-5 — — 25 225 38-3 230 — — — — — — — 26 38 6-4 35 — — — 27 15 2-5 25 — — — — — — — 28 8 1 -4 19 — — — — — — — 29 112 190 114 — — — — — — — 30 (109) 18-5 (114) — — — — — — — 31 112 190 118 116 230 32 32 30 - — - — — — — — — — 33 149 25-3 136 — — — 140 27-7 — — 34 105 17-8 100 — — — — — — — 35 47 80 — — 37 6-3 46 91 35 5-7 36 37 6-3 37 7-3 37 48 81 50 — 49 8-5 49 9-7 50 8-3 38 67 11-4 68 — — — — — — — 39 ±250 42-5 ±250 — 235 400 252 44-6 240 400 40 ±55 9-3 ±55 — — — — — — — 41 80 13-6 70 „ 42 35 5-9 40 — — — — — — 43 38 6-4 30 — — — — - — - — — 44 45 135 90 22-9 — = = - — = — — 46 505 860 481 85-5 440 87-1 498 82-9 47 89± — — — 155 26-4 125 24-8 161 26-8 48 82 13-9 — — 83 14- 1 92 18-2 91 15-2 49 350 59-9 — — — — — — — — 50 90 *Measurements from McCann (1962). ROSS: HECTOR’S BEAKED WHALE DESCRIPTIONS OF MEASUREMENTS PROVIDED IN TABLE I (AFTER MOORE, 1963) 1. Greatest length of skull. 2. Greatest length of rostrum, tip of beak to line connecting apices of antorbital notches. 3. Tip of rostrum to posterior margin of pterygoid nearest mid-sagittal plane. 4. Tip of rostrum to most posterior extension of wing of pterygoid. 5. Tip of rostrum to most anterior extension of pterygoid. 6. Tip of rostrum to most posterior extension of maxillaries between the pterygoids on the palate. 7. Tip of rostrum to most posterior extension of maxillary plate. 8. Tip of rostrum to anterior margin of superior nares. 9. Tip of rostrum to most anterior point on premaxillary crest. 10. Tip of rostrum to most posterior extension of temporal fossa. 1 1. Tip of rostrum to most posterior extension of lateral tip of premaxillary crest. 12. Tip of rostrum to most anterior extension of pterygoid sinus. 13. Greatest length of temporal fossa. 14. Greatest length of orbit. 15. Greatest length of right nasal on vertex of skull. 16. Length of nasal suture. 17. Greatest breadth of skull across post-orbital process of frontals. 18. Greatest breadth of skull across zygomatic processes of squamosals. 19. Greatest breadth of skull across centers of orbits. 20. Least breadth of skull across posterior margins of temporal fossae. 21. Greatest span of occipital condyles. 22. Greatest width of an occipital condyle. 23. Greatest length of an occipital condyle. 24. Greatest breadth of foramen magnum. 25. Greatest breadth of skull across exoccipitals. 26. Greatest breadth of nasals on vertex. 27. Least distance between premaxillary crests. 28. Greatest extension of right premaxillary posterior of right nasal on vertex of skull. 29. Greatest span of premaxillary crests. 30. Least width (strictly transverse) of premaxillae where (and if) they narrow opposite superior nares. 31. Greatest width of premaxillae anterior to place of measurement No. 30. 32. Width of premaxillae at mid-length of rostrum. 33. Width of rostrum in apices of antorbital notches. 34. Width of rostrum in apices of prominential notches (if any). 35. Greatest width of rostrum at mid-length of rostrum. 36. Greatest depth of rostrum at mid-length of rostrum. 37. Greatest transverse width of superior nares. 38. Greatest inside width of inferior nares, at apices of pterygoid notches, on the pterygoids. 39. Height of skull. Distance between vertex of skull and most ventral point on pterygoids. 40. Greatest width of temporal fossa approximately at right angles to greatest length. 41. Least distance between (main or anterior) maxillary foramina. 42. Least distance between premaxillary foramina. 43. Distance from posterior margin of left maxillary foramen to most anterior extension of left maxillary prominence. 44. Greatest length of vomer visible at surface of palate. 45. Amount added to beak because of breakage. 46. Length of mandible (condyle to apex). 47. Greatest length of mandibular symphysis. 48. Greatest depth of mandible. 49. Length from most posterior extension of symphysis to most posterior extension of condyle. 50. Estimated amount added to mandible length because of breakage. with increasing skull length, indicating that there is little broadening of the skull in proportion to the increase in total length. Measurements 21, 31, 33, 37 and 39 are very similar in all five specimens, suggesting that growth in these dimensions is almost static. A lateral basirostral groove is absent, the mesorostral canal has not become reduced by increase of the vomer in size within it, and the elements of the occipital region are almost completely fused and coalesced with the basisphenoid. The premaxillary foramina differ in being slightly caudal (PEM 1511/16) or distinctly caudal (PEM 1511/15) to the maxillary foramina, while they are in the same transverse plane in the type and the Falkland Island skull, and rostral in the Tasmanian specimen (Guiler, 1967). This variation strengthens Moore’s (1960) findings in M. gervaisi that this feature is no longer suitable as a taxonomic character for all species of this genus. The Lottering Mouth specimens show slight individual differences which may be attri- 199 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 13, DECEMBER 1970 buted to individual variation, though it is possible they are sexual differences. It is unlikely however, that these are age differences. PEM 1511/16 appears more robust than PEM 1511/15 as a result of the anterior extension of the maxillary plate to form the dorsal part of the distinct antorbital tubercle. The jugal forms the antero-mesial, and the lacrimal the antero-ventral surfaces of the tubercle in 1511/16. However, in 1511/15, the lacrimal forms almost the entire tubercle, while the jugal is restricted to the apex of the antorbital notch (Plates I and II). In this latter skull the lacrimal curls around the anterior edge of the frontal, but does not in 1511/16. GEOGRAPHICAL DISTRIBUTION Four specimens of M. hectori have been recorded previously; two from New Zealand (McCann, 1962), one from the Falkland Island (Fraser, 1950) and one from Tasmania (Guiler, 1967). The present specimens greatly extend the known range of the species, support- ing Fraser’s suggestion that the species is circumpolar in distribution, and extends the northern limits of its range by some 10 of latitude. While the South African records do not support Guiler’s suggestion that the species is rare to the north of 45° S, it is possible that these two are strays from the south, as other sub-antarctic species are not uncommon on the Cape coast, particularly in the first months of the year. ACKNOWLEDGEMENTS I wish to thank Mr C. K. Tayler (Port Elizabeth Oceanarium) for collecting the two skulls from Lottering River mouth. I am especially grateful to Dr Joseph Curtis Moore (Field Museum of Natural History, Chicago) for his initial identification, his advice throughout the preparation of this report and his constructive criticism of the manuscript. REFERENCES Andrews, R. C., 1908. Description of a new species of Mesoplodon from Canterbury Province, New Zealand. I Am. Mus. nat. Hist., 24 (13): 203 — 215. Barnard, K. H. 1954. A guide book to South African whales and dolphins. Cape Town: South African i Museum. Flower, W. H., 1 878. A further contribution to the knowledge of existing Ziphioid whales, genus Mesoplodon. i Trans, zool. Soc. London, 10: 415—437. Fraser, F. C., 1950. Notes on a Skull of Hector’s Beaked Whale Mesoplodon hectori (Gray) from the Falk- land Islands. Proc. Linn. Soc. London {zool.), 162: 50 — 52. Fraser, F. C., 1955. A skull of Mesoplodon gervaisi (Deslongchamps) from Trinidad, West Indies. Ann. Mag. nat. Hist., Ser. 12, 8: 624 — 630. Guiler, E. R., 1967. Strandings of three species of Mesoplodon in Tasmania. J. Mammal, 48 (4): 650 — 652. McCann, C., 1962. The taxonomic status of the beaked-whale, Mesoplodon hectori (Gray) — Cetacea. Rec. Dom. Mus., Wellington, 4 (9): 83 — 94. McCann, C., and F. H. Talbot, 1963. The occurrence of True’s beaked whale {Mesoplodon minis True) in South African waters, with a key to South African species of the genus. Proc. Linn. Soc. London, 175 (2): 137—145, Moore, Joseph Curtis, 1960. New records of the Gulf-stream beaked whale, Mesoplodon gervaisi, and some taxonomic considerations. Am. Mus. Novit., (1993): 1—35. Moore, Joseph Curtis, 1963. Recognizing certain species of Beaked Whales of the Pacific Ocean. Am. Midi. Nat., 70 (2): 396—428. Moore, Joseph Curtis, 1966. Diagnoses and distributions of beaked whales of the genus Mesoplodon known from North American waters. In: Norris, K. ed Whales, dolphins and porpoises. Los Angeles: University of California Press. Moore, Joseph Curtis, 1968. Relationships among the living genera of beaked whales, with classifications, diagnoses and keys. Fieldiana, Zool. 53 (4): 209 — 298. 200 ROSS: HECTOR’S BEAKED WHALE Ness, A. R., 1967. A measure of asymmetry of the skulls of whales. J. Zool., London, 153: 209 — 221. Nishiwaki M., 1962. Mesoplodon bowdoini stranded at Akita Beach, Sea of Japan, Scient. Repts. Whales Res. Inst., Tokyo, 16: 61 — 78. Nishiwaki, M. and T. Kamiya, 1958. A beaked whale Mesoplodon stranded at Oiso Beach, Japan. Scient. Repts. Whale Res. Inst., 13: 53 — 83. Oliver, W. R. B., 1922. A review of the Cetacea of New Zealand seas. Proc. Zool. soc., London, 1922: 557 585. Rankin, Jessie J., 1956. The structure of the skull of the beaked whale, Mesoplodon gervaisi Deslongchamps. J. Morph., 99 (2): 229—258. Raven, H. G., 1937. Notes on the taxonomy and osteology of two species of Mesoplodon (M. europaeus Gervais, M. minis True). Am. Mus. Novit., 905: 1 — 30. Ross G. J. B., 1969. Evidence for a southern breeding population of True's Beaked Whale. Nature, 222: 585. True Frederick, W., 1910. An account of the beaked whales of the family Ziphiidae in the collection of the United States National Museum, with remarks on some specimens in other American museums. Bull. U.S. natn. Mus. 73: 1 — 89. 201 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 13, DECEMBER 1970 Plate I. Skull of Mesoplodon hectori (PEM 1511/15) in (a) posterior, ( b ) anterior, (c) lateral, {d) ventral and (e) dorsal views. 202 ROSS: HECTOR’S BEAKED WHALE Plate II. Skull of Mesoplodon hectori (PEM 1511/16) in ( a ) posterior, ( b ) dorsal, ( c ) anterior, ( d ) ventral and (e) lateral views. 203 a b Plate III. (a) Dorsal and ( b ) median lateral views of the mandibles of Mesoplodon hectori (PEM 1511/15). (c) The dried tail-flukes of M. hectori (PEM 1511/16). 204 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (Nat. Hist.) VOLUME 8 • PART 14 31st DECEMBER 1970 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA A new species of House Snake from Swaziland, with notes on the status of the two Genera Lamprophis and Boaedon by N. SCHAEFER Port Elizabeth Museum, Port Elizabeth Lamprophis swazicus new species INTRODUCTION In October 1968, two snakes were sent to me for identification by Mr J. Culverwell who found them at Forbes Reef, Swaziland. A third snake of the same kind, originally collected by Mr A. Schaefer at Havelock, Swaziland in January 1968, was sent to Mr W. Haacke, Transvaal Museum, who then loaned it to me for study. These three specimens were recognized as being some type of House Snake, but could not be classified to any known South African species using the early key of Boulenger 1893 or the recent key of FitzSimons 1966. Furthermore, their characteristics did not fit in with those of any known African House Snake species as given by Broadley (1969), nor with the characteristics of species in other genera, such as Lycodonomorphus, Bothrolycus, Bothrophthal- mus and Pseudoboadon which Boulenger (1893) and Dowling (1969) consider to be closely related to the House Snakes. The Swaziland snakes therefore, have apparently not been described before and can be considered as a new species. Although there is some doubt as to which of the two house snake genera ( Lamprophis or Boaedon ) the new species belongs, it has been placed in the genera Lamprophis for reasons discussed later in the paper. MATERIAL Three specimens: unsexed. Holotype: PEM 1514/81. Paratypes: PEM 1514/82. Type locality: Forbes Reef, Swaziland (26° 42' S, 31° 05' E). DIAGNOSIS A colubrid snake of the Lamprophis genus having one apical pit per scale and 17 scale rows at midbody. The ventrals exceed 200, and the subcaudals exceed 80 in number. There are nine maxillary teeth. The colour is light beige dorsally fading to creamy/ white ventrally. DESCRIPTION Holotype: PEM 1514/81. The lepidosis of the head is shown in Figure 1. The drawings were made directly from photographs. 205 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 14, DECEMBER 1970 Rostral twice as wide as deep and barely visible from above; nasal shield semi-divided; frontal shield only 1^ times longer than broad; one pre-ocular and two post-oculars with temporals 1 + 2. Eight upper labials of which 3rd, 4th and 5th enter the orbit; 1st upper labial not in contact with loreal. Ten lower labials on the right and nine on the left. The different labial count on left and right sides is probably an abnormality and the true number, judging by the other specimens, is 10 on both sides. The head is about 1 J times broader than the neck. Pupil brown and vertically elliptical. The body scales are smooth and have one apical pit each. There are 206 ventrals (counted according to the system proposed by Dowling 1951a) and 88 paired subcaudals. The anal is entire. There are nine maxillary teeth present increasing in size to the fourth and then diminishing again posteriorly. The snake is light beige in colour and 568 mm long (Body 430 and tail 138). Paratypes: (PEM 1514/82; TM 34836). Both paratypes are similar to the holotype but there are slight differences. PEM 1514/82. The lower labial count differs on left and right sides — on the left there are 11 and on the right 10. On both sides there is an extremely shallow labial — 7th on the left, 6th on the right. The labial arrangement however, is probably an abnormality and therefore, of little systematic importance. Ventrals 207, subcaudals 80. There are also nine maxillary teeth but there is a distinct gap between the 4th and 5th tooth. The snake is slightly lighter beige than the holotype. In addition each dorsal scale has a slightly darker edge which gives a very lightly reticulated effect. This animal was the smallest of the three at 192 mm (Body 126, tail 66). It was dissected after it died and a feather was found in the gut. TM 34836, Ventrals 208, subcaudals 91. The nine maxillary teeth are also separated by a gap between the 4th and 5th tooth, but this gap is narrow and hardly wider than the spaces between the other teeth. Fig. 1. Dorsal, ventral and side views of the head of the holotype Lamprophis swazicus showing head scaling. SCHAEFER: A NEW SPECIES OF HOUSE SNAKE FROM SWAZILAND It is darker beige, almost brown, in colour and yellow/brown ventrally. This snake however, had been in preservative for a year and a half which may have caused the darkening. It is 730 mm long (Body 545, tail 185). REMARKS The South African house snakes are separated into two genera, Lamprophis and Boaedon. FitzSimons (1962), mentions that their separation hardly seems justifiable because they appear so closely related. He does however, distinguish Lamprophis from Boaedon using the charac- teristics shown in Table 1. Table 1 Lamprophis Boaedon Apical pits absent present Ventrals 170—198 186—237 Mid-body scales 19—25 21—35 Frontal shield short, broad long, narrow Maximum teeth number 15—19 18—24 Maximum teeth size .... shorter ant. longer ant. The new snake however, does not have all the characteristics common to any one par- ticular genus. Its characteristics are shown in Table 2. Table 2 L. swazicus Apical pits Ventrals Mid-body scales Frontal shield Maximum teeth number Maximum teeth size present 206—208 17 short, broad 9 shorter ant. Some features (short, broad frontal, shortest maxillary teeth anteriorly) are charac- teristic of Lamprophis , while some (apical pits, ventral count of over 200) are characteristic of Boaedon. Nevertheless, despite the fact that the new snake has some features in common with Boaedon , it has been assigned to the genus Lamprophis. The reasons for this are twofold. Firstly, FitzSimons also separates Lamprophis from Boaedon by using the average mid- body scale-count and the average maxillary tooth-count because these differ between the two genera. The new species, although not having a mid-body scale and maxillary tooth count falling within the limits given for Lamprophis , is closer to it than to the limits given for Boaedon. Secondly, Broadley (1969) has pointed out that these average differences and in fact all the characteristics formerly used to define the two genera, are very vague and do not definitely separate them. It seems as if the species of African house snakes form a continuous series rather than two complete and separate groups. Broadley draws attention to this fact and suggests that the two genera may need to be merged. This suggestion is now further substantiated with the discovery of the new species, for it has characteristics of both genera. If in the future more 207 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 8, PT 14, DECEMBER 1970 evidence comes to hand and the genera are merged, then Boaedon will become a synonym of the earlier name, Lamprophis. Thus by naming the new species Lamprophis and not Boaedon , a future synonomy will be avoided. ACKNOWLEDGEMENTS I would like to thank Mr Culverwell for sending me two of the specimens and Mr Haacke for loaning me the third. REFERENCES Boulenger, G. A., 1893. Catalogue of the snakes in the British Museum , 1:180, 320 and 327. Broadley, D. G., 1969. “The African house snakes — How many genera?”, J. herp. Assoc. Afr. March 1969. p. 6—8. Dowling, H. G., 1951b. “A proposed standard system of counting ventrals in snakes”, Brit. J. Herpet., 1(5) :97— 99. Dowling, H. G., 1969. “Relations of some African Colubrid Snakes”, Copeia, 1969(2) :234 — 242. FitzSimons, V., 1962. Snakes of Southern Africa. Purnell & Sons, Johannesburg, pp. 1 — 423. FitzSimons, V., 1966. “A check-list; with synoptic keys, to the snakes of Southern Africa”, Ann. Transv. Mus., 25(3) :35— 79. PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (nat. Hist.) VOLUME 9 • PARTS 1—15 1972—4 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Cape Provincial Museums South Africa Albany Museum, Grahamstown Alexander McGregor Memorial Museum P.O. Box 316, Kimberley East London Museum East London Kaffrarian Museum King William’s Town Port Elizabeth Museum, Snake Park and Oceanarium, Humewood, Port Elizabeth Editor C. F. Jacot-Guillarmod Albany Museum, Grahamstown Assisted by R. M. Tietz Printed by Cape & Transvaal Printers Ltd., Cape Town LIST OF CONTENTS 1 . Records of Cuvier's Beaked Whale, Zipkius cavirostris from Southern Africa. G. J. B. Ross and R. M. Tietz 1 2. The social organization and behaviour of dolphins ( Tursiops aduncus ) and baboons ( Papio ursinus ): some comparisons and assessments. C. K. Taylor and G. S. Saayman 11 3. A preliminary study on fruit production in certain plants. R. Liversidge ... 51 4. Dress, personal decoration and ornament among the Ndlambe. E. H. Bigalke . 65 5. The status of the South African galaxiid (Pisces, Galaxiidae) R. M. McDowall . 91 6. A new species of Handlirschia Kohl (Hymenoptera : Sphecidae), a very poorly known genus from South Africa. F. W. Gess 103 7. Third contribution to the knowledge of the South African species of the genus Ceramius Latreille (Hymenoptera: Masaridae). F. W. Gess 109 8. A report on excavations carried out on a Type R Settlement Unit (Khartoum 1) in the Jacobsdal district, O.F.S. A. J. B. Humphreys 123 9. The exploitation of shell fish by coastal tribesmen of the Transkei, E. H. Bigalke 1 59 10. Report on some collections of middle stone age artefacts from Riverton, Kimberley district, South Africa. A. J. B. Humphreys 177 11. An ethological study of Dichragenia pulchricoma (Arnold) (Hymenoptera: Pompili- dae), a southern African spider-hunting wasp which builds a turreted, subterranean nest. F. W. Gess and S. K. Gess 187 12. On the life colours and nuptial tubercles of Oreodaimon quathlambae (Barnard, 1938) (Pisces, Cyprinidae). P. H. Skelton 215 13. The Trichoptera of the Sundays and Fish Rivers, Eastern Cape Province, South Africa. K. M. F. Scott 223 14. New and interesting Trichoptera collected by Dr. H. Bertrand in Southern Africa in 1959. K. M. F. Scott 237 15. Comments on the occurrence of core-axe-like artefacts in the Northern Cape. A. J. B. Humphreys 249 INDEX TO AUTHORS BIGALKE, E. H. Dress personal decoration and ornament among the Ndlambe. (4) Oct. 1972 65-90 BIGALKE, E. H. The exploitation of shellfish by coastal tribesmen of the Transkei. (9) Dec. 1973 159-75 GESS, F. W. and S. K. GESS. An ethological study of Dichragenia pulchricoma (Arnold) (Hymenoptera : Pompilidae), a southern African spider-hunting wasp which builds a turreted, subterranean nest. (11) May 1974. . . 187-214 GESS, F. W. A new species of Handlirschia Kohl (Hymenoptera: Sphecidae), a very poorly known genus from South Africa. (6) Dec. 1973 103-7 GESS, F. W. Third contribution to the knowledge of the South African species of the genus Ceramius Latreille (Hymenoptera: Masaridae). (7) Dec. 1973. 109-22 GESS, S. K. see GESS, F. W. and S. K. Gees. HUMPHREYS, A. J. B. A report on excavations carried out on a Type R. Settlement Unit (Khartoum 1) in the Jacobsdal district O.F.S. (8) Dec. 1973 123-57 HUMPHREYS, A. J. B. Report on some collections of middle stone age artefacts from Riverton, Kimberley district, South Africa. (10) Dec. 1973 177-85 HUMPHREYS, A. J. B. Comments on the occurrence of core-axe-like artefacts in the Northern Cape. (15) Dec. 1974. . . 249-59 LIVERSIDGE, R. A preliminary study on fruit production in certain plants. (3) Oct. 1972 51-63 McDOWALL, R. M. The status of the South African galaxiid (Pisces, Galaxiidae). (5) Jan. 1973 91-101 ROSS, G. J. B. and R. M. TIETZ. Records of Cuiver’s Beaked Whale, Ziphius cavirostis from Southern Africa. (1) Oct. 1972 . . 1-10 SAAYMAN, G. S. ^ TAYLER, C. K.and G. S. SAAYMAN. SCOTT, K. M. F. New and interesting Trichoptera collected by Dr. H. Bertrand in Southern Africa in 1959. (14) May 1974 237-48 SCOTT, K. M. F. The Trichoptera of the Sundays and Fish Rivers, Eastern Cane Province, South Africa. (13) May 1974 223-35 SKELTON, P. H. On the life colours and nuptial tubercles of Oreodciimon quathlambae (Barnard, 1938) (Pisces, Cyprini- dae). (12) May 1974 215-22 TAYLER, C. K. and G. S. SAAYMAN The social organisation and behaviour of dolphins ( Titrsiops aduncits) and baboons ( Papio ursinus): Some comparisons and assessments. (2) Oct 1972 11-49 TIETZ, R. M. .«? Fig. 2. A school of slowly progressing bottlenose dolphins comprising groups of subgroups with scattered lone individuals. Counts were made as members of each subgroup surfaced in synchrony as they passed in clear water beneath the observer. 34 TAYLER AND SAAYMAN : ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS consecutive days. Two bulls were subsequently netted from the rear of this group as the dolphins moved parallel to the shore in the shallows in line-astern formation. The larger of the two (Dolfie) had an inclined notched dorsal fin, closely resembling that seen in the dolphin at sea and it is likely that they were the same animal. Using the shape of the dorsal fin and the relative proportions of the head as criteria, close-range observations of this group indicated that it consisted predominantly, if not entirely of bulls. Similarly, in an earlier capture operation when Haig was taken, she was closely associated with another cow. Out of a total of five capture operations the following animals accompanied each other and were netted together: (i) an adult cow and a subadult cow, (ii) two adult cows and a subadult cow, (iii) two adult cows, two subadult bulls and a juvenile bull, (iv) two adult bulls, (v) two adult bulls. In both cases (iv) and (v) the bulls were believed to have been taken from the rear of their groups. These findings supported the subjective impression that of the many possible combinations of the four age classes (adult, subadult, juvenile and calf) the following assoc- iations did not occur: immature dolphins only and immature dolphins accompanied by bulls. The spatial organization of dolphin schools varied in relation to time of day, number of animals and form of activity. A school of dolphins in transit, comprising several distinct i / 2 KILOMETRES Fig. 3. A school of sixty to one hundred dolphins, feeding in and out of the surf, progresses rapidly along the coast with no distinct or permanent subgrouping; accurate counts of individuals in such schools were not possible. Dolphins sometimes formed up temporarily abreast to follow a swell, simultaneously disengaging as it broke. 35 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 2, OCTOBER 1972 “groups” within which there were further independent stable divisions of two to ten individuals termed “subgroups”, is illustrated in Fig. 2. Tail slapping, chasing and biting behaviour — frequently associated with sexual and dominance disputes — was conspicuous in one disorderly group of twenty individuals. Another group of ten, swimming in close proximity in line-astern formation, had no distinct subgroups. Lone dolphins were scattered on the periphery of the school and between groups. This type of school was seen slowly progressing in the early afternoon when dolphin groups were normally engaged in play or sexual activity. Formations of this kind were rarely sighted from coastal vantage points; the more frequently seen progression was a scattering of dolphins with no very clear organization except for occasional subgroups (Fig. 3). Further examples of schooling formation, also seen infrequently from the coast, are shown in Fig. 4: the horseshoe formation (Fig. 4A) was associated with fish herding in deeper waters (20 metres) whereas the assembly of subgroups (Fig. 4B) was seen in schools proceeding rapidly close to the coast in deep water. Dolphin formations seen most commonly from the coast were either the isolated group (Fig. 4C) or subgroup (Fig. 4D). Subgroups in free-ranging and captive dolphins varied both in numerical composition and in spatial deployment. Frontal, dorsal and lateral views for a trio of dolphins are shown in Fig. 5. It was not always clear whether specific formations were related to definite functions but the following observations provided a clue to the nature of some of the patterns of deploy- ed Cd Cd Cd Cdcd Cd °o° Cd o Cd Cd o o Cd> ^ c c Cd Cd Cd Cd Cd Cd Cd Cd Cd Cd dd Cd <^Cd B Fig. 4. Schooling, grouping and subgrouping in bottlenose dolphins. A. A horseshoe formation of four groups without distinct subgroups. B. A wing formation of subgroups only. C. A group — more than six dolphins — without subgrouping. D. A subgroup. 36 Fig. 5. I, II and III. Frontal, dorsal and lateral views of variations in spatial deployment of bottlenose dolphins. For convenience, only a trio is illustrated. IVa and b. Positions adopted by a young calf. IVc. Change in position of calf in uncertain situations. ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 2, OCTOBER 1972 ment: the spearhead was commonly seen when dolphins investigated a source of alarm near the surface (Fig. 5, Ih, Ilf), or near the bottom (Fig. 5 Ig, Ilf) of the pool. In both cases the leader, the dominant dolphin, was deployed on the vertical plane closest to the alarm. Similarly, the presence of a diver in the tank stimulated a change in the position of the calf from beneath the belly of her mother to her far side (Fig. 5 IVc). A functional formation, seen only in cows, was the linking of two individuals by the extended flipper of an apparently dozing cow which maintained contact with the abdomen or flank of the leading, vigilant animal (Fig. 5 Illk, Illn). The close co-ordination was remarkable as the dolphins progressed, disengaging contact momentarily while surfacing for air, apparently communicating through- out by tactile stimuli. Indeed, the importance of tactile stimuli in establishing and maintaining mother-calf formations was clearly evident from birth. In free-ranging dolphins comparable re-arrangements of formation occurred but in the majority of cases the causes were not discernible. Inspection of Fig. 5 demonstrates that many combinations were possible when sex, age class, number of individuals and circumstances varied. Inter-specific encounters influenced group progressions, subgrouping and formations. Two examples of co-ordinated avoidance responses to potentially dangerous hammerhead sharks (, Sphyrna zygaena) and a lone blue pointer ( Carcharodon carcharias ) are illustrated in Fig. 6. As the dolphin group divided to skirt the hammerhead sharks, a subgroup of two 38 TAYLER AND SAAYMAN: ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS individuals formed on one flank, increased speed and passed at close range; the group reformed with the sharks in its wake (Fig. 6A). The dolphins gave the larger, more formidable blue pointer shark a wider berth by sounding to a depth greater than ten metres (Fig. 6B). Shark attacks on dolphins are known to occur (Lineaweaver, 1967) but the incidence, severity or cause of the attacks have not been fully evaluated. It is possible that the exploratory approaches of the younger animals and the policing manoeuvres of the dominant dolphins provoked retaliatory or predatory responses by sharks. It was perhaps of significance that Dolfie, considered to be a dominant bull of a free-ranging group, bore the scars of a recent shark bite (Plate 16) and that an attack on an apparently immature male dolphin ( Tursiops truncatus has been reported (Caldwell, Caldwell and Siebenaler, 1965). Encounters with killer whales ( Orcinus orca ), probably the only true predator of dolphins, were never witnessed in detail. Tape recordings of killer whale sounds played to our captive dolphins stimulated rapid swimming, tight formations and fear responses whereas recordings of other marine species did not have these effects. Subgroups of dolphins of unknown age or sex classes made exploratory approaches to Cape fur seals ( Arctocephalus pusilus) sleeping on the surface. In captivity Flaig played ex- tensively with Tommy, a young bull seal (Plates 17 and 18). This play activity, which had no aggressive overtones, developed after several years into obvious sexual approaches by both animals and culminated in attempted copulation. It was initiated when exploratory burst of high speed swimming close to the sleeping seal developed gradually into games of chase. Plate 16. Dolfie, an adult bull, was captured from a free-ranging group bearing the scars of a recent shark bite in the motor muscles below the dorsal fin. Note the extensive “raking” by the teeth of other dolphins. 39 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 2, OCTOBER 1972 Plate 17. Haig, playing with Tommy, a subadult Cape Fur Seal. In marked contrast to the orderly formation maintained by dolphins in the vicinity of sharks, precipitate flight was shown by a group when a shot from a light calibre rifle was fired into the water fifty metres ahead of an approaching group swimming parallel to the shore: all animals turned simultaneously for the open sea and fled with the distance separating them increasing in the line of flight as the fastest dolphins outpaced the others (Fig. 6C). In another incident, when one of a subgroup of two individuals was inadvertently foul-hooked on light fishing tackle, both turned for the open sea but the hooked animal was left far behind leaping high into the air to free the line which had already parted (Fig. 6D). In both of these encounters with extraordinary and alarming stimuli, therefore, the organized and co-ordinated mode of progression had broken down. The availability of favoured food fish (shad — Pomatomus saltator) partially determined the pattern of movement of bottlenose dolphins along the coast in the early morning and in the late afternoon, at which times feeding activity was most prominent; the capture of squid, a known dietary component of dolphins was never witnessed. Due to the popularity of shad angling, it was well known that this species of fish frequented certain feeding places, mainly in turbulent water just beyond breaking surf and rocky outcrops. It was often noticeable that shad came on the bite when bottlenose dolphins appeared two to five hundred metres distant. 40 TAYLER AND SAAYMAN: ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS Plate 18. Tommy, hanging limp, permitted Haig to gently grasp and tow him during extensive periods of play. Dolphins progressed in scattered groups, sometimes spread out about 800 metres in length along the coast. Group cohesion was not a marked feature of such groups and dolphins cap- tured fish on an individual basis. However, some splashing seawards of the surf by animals leaping clear of the water with body bent concave downward and landing on their sides with a resounding report was generally present; although this behaviour in captive animals was associated with the later stages of pregnancy and in known cases of gastronomical discomfort, it was thought, because of the orderly fashion in which the leaping was effected by several animals in line-astern formation, that this procedure in the present instance related to fish herding. In contrast, organized herding by bottlenose dolphins of shoals of fish was seen, both in the open sea and at the confluence of rocks and beach. An incident, observed in clear water from a vantage point on Robbeberg at Plettenberg Bay, is illustrated in Fig. 7. A group of dolphins was sighted containing a shoal of shad at the intersection of a rocky promontory and a sandy beach. A large dolphin (A), judged by overall size, shape of head and dorsal fin, to be a bull, circled a prominent route equidistant between rocks and surf: lactating cows, entering the surf to feed, left their calves in the company of other cows in the close vicinity of this animal. Dolphins fed from the periphery of the shoal along the escape routes of the fish but at no time attempted to feed directly into the trapped main shoal which would probably have been dispersed. Dolphins continually arrived from and departed to the open sea, but at all times there were sufficient numbers to keep the fish tightly herded. A second large dolphin (B), also thought to be a bull, appeared from time to time in the position indicated in Fig. 7 ; the impression was gained that this dolphin patrolled a route to and from the open sea as 41 \ Fig. 7 Organized herding, retention and capture by bottlenose dolphins of a shoal of shad. ROCKS FOUR FEET ABOVE HIGH TIDE 200 YARDS TAYLER AND SAAYMAN: ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS indicated by the dotted line. The routes followed by dolphins A and B, the fact that they returned to specific areas after intermittently capturing a fish, as well as their apparent control over the central area in the proximity of the immature dolphins, gave the strong impression that they were performing a supervisory function over what appeared to be a highly organized herding and capturing manouevre. A similar incident was witnessed from its inception in the open sea. Bottlenose dolphins, leaping and splashing as described above, converged slowly from opposing flanks and gradually a tightly packed shoal of fish became discernible as a dark-coloured mass beneath the surface between the encircling dolphins. Simultaneously, dolphins could be seen darting under the shoal and thus preventing it from sounding. The shoal was consumed from the sides and underneath while the whole ensemble progressed slowly out of sight at about 7 k.p.h. These observations suggested that the two distinct types of feeding activity seen in bottlenose dolphins were part of the same process: the scattered groups of feeding dolphins, extending over long stretches of coast generally contained some leaping and splashing individuals, and these activities might well have driven fish ahead until they were herded into a suitable position where they could be contained and captured from the periphery as in the above examples. J/ ZWARTKLIP MUIZENBERG PREVAILING WIND SOUTH EASTERLY 1 NAUTICAL MILE T. aduncus y L. obscurus Abrupt terr ip. change X x. Splashing D istribution of food © Observation points Fig. 8. The influence of water temperature upon the day ranging of two delphinid species. 43 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 2, OCTOBER 1972 A school of dusky dolphins ( Lagenorhynchus obscuvus) was observed feeding at Hout Bay in the Cape. A single dolphin found a shoal of pilchards ( Sardinops ocellata) stationed behind an outcrop of rocks in deep calm water. As the dolphin turned towards the pilchards, the school of approximately two hundred dolphins, scattered over an area of about eight square kilo- metres, immediately converged on the area and commenced feeding. This suggested that the presence of the fish had been communicated to the other dolphins. Experiments at the Oceanarium have shown that all of the dolphins are aware of the information reflected when sonar is employed by any one dolphin, and thus no intentional transfer of information need be posited as an explanation of this incident. The feeding pattern of humpback dolphins ( Sotalia lentiginosa ) differed markedly from that of bottlenose and dusky dolphins: humpbacks were comparatively slow-moving animals, generally seen in compact groups ranging in number between three and twenty individuals. They seldom leaped out of the water, and when this happened it appeared to be associated with courtship in adults, or play in juveniles. In areas where the sea bed was sandy with outcrops of isolated reefs, humpbacks moved systematically from one outcrop to the next, feeding leisurely and remaining submerged for comparatively long periods. Groups, sometimes separated in time by several hours, usually inspected the same reefs, following similar routes and it was noticeable that they never entered the surf. A group sometimes divided into two or more subgroups at widely separated reefs but, because they later reunited, had apparently maintained acoustical contact. Lone individuals were also seen. Water temperature appeared to limit the distribution of various species of dolphins. Seasonal fluctuations in warm and cold ocean currents were associated with mass movements of dolphins apparently over great distances. In addition, sudden changes in climatic conditions affecting water temperature influenced dolphin ranging within specific areas; the presence of dolphins in sheltered bays in spring often indicated an impending wind change. In captivity sudden fluctuations in both the atmospheric humidity and the water temperature of the pool strikingly influenced the temperament and appetite of the dolphins. Free-ranging bottlenose dolphins feeding on shad habitually followed certain routes; in the example shown in Fig. 8 they on occasions turned sharply from the normal route at the thermocline in spite of there being abundant shad beyond. The only detectable factor correlated with this deviation was the sudden fall in water temperature at this point. Similarly, dusky dolphins, normally frequenting colder waters, turned at the opposite side of this thermocline. Again, their known food fish (pilchards) were distributed throughout. DISCUSSION The primary purpose of this comparative review was to assess our current knowledge of the naturalistic behaviour and social organization of Indian Ocean dolphins. It is clear that, in contrast to the more easily accessible and identifiable terrestrial baboons, the study of ranging and migratory behaviour, group composition and population dynamics of free-ranging dolphins is greatly handicapped by the lack of suitable marking techniques. A further limitation is our incomplete understanding of the acoustical perceptual systems employed by cetaceans. A useful feature for day to day identification, apart from differences in physical size of individual dolphins, is the pattern of atrophying skin on the posterior edge of the dorsal fin in association with the characteristic inclination of the fin in certain individuals; notches in the anterior peduncle occur less frequently and are less conspicuous. Such natural markings are of short-term value only, since in two captive animals these features underwent many changes in seven years. Similarly, deep scars lose their prominence and continuous skin replace- ment in dolphins (an algae growth-inhibiting mechanism) appears, therefore, to militate 44 TAYLER AND SAAYMAN : ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS against branding as a marking technique. Tagging was considered in 1962, but experience with reactions to physical marking or tagging was necessary. Violent physiological rejection of foreign bodies occurs: fish spines piercing the blubber become rapidly encapsulated and large subcutaneous abscesses erupt after several weeks with gross tissue damage. It is likely that rapid repair to damaged surface tissue is an adaptation in these deep-diving marine mammals, with consequent slow replacement of underlying tissue, from beneath, resulting ultimately in the rejection of the foreign body together with the entire surrounding affected area. Placing either loose or tight-fitting loops as markers around the peduncle are likely to injure the animal, since captive dolphins exhibit abrasions as the result of exerting mild pressure on a towing loop for only four minutes each day; apart from streamline-spoiling in these fast swimming animals, bottlenose dolphins show great stress in response to the attachment of any device, however small. A large number of unsuccessful tagging experiments on four species of dolphins have been conducted in Japan (Nishiwaki, Nakajima and Tobayama, 1966). The identification of animals inhabiting an aquatic three-dimensional environment is limited by conditions of water clarity: dolphins, marked or unmarked, are partially visible only every ten to sixty seconds when they surface to breathe in the normally unstable surface plane. Apart from the prohibitive cost of a systematic tagging programme on our coast — assuming the development of a suitable marker — the problems are even further compounded by the nomadic nature of the dolphin and by its reluctance to return to areas where it has previously been molested. Underwater chambers have been used to observe cetaceans in their own element (Evans and Bastian, 1969). However, unfamiliar objects are likely to stimulate apprehension and to restrict behaviour: observation of dolphins in clear water from elevated coastal observation points has frequently shown that their behaviour is markedly influenced by the approach of a skin diver or boat. Groups of bottlenose dolphins often deploy on a vertical plane (Fig. 5 Ie) in order to obtain a clearer view when investigating a skin diver underwater: Tayler and others have on occasion been threatened with open mouth by large individuals while swimming amongst free-ranging dolphins. On the other hand, underwater devices, although provoking caution, may hold groups of dolphins fascinated for considerable periods but only within the inevitably limited range of view: multiple underwater television devices would suffer, to a lesser extent, from the same limitations and also from inherent sound propagation. Until technology has evolved a suitable vehicle for keeping pace continuously with known groups of dolphins, so that the animals become habituated to a neutral observer, as occurs in groups of non-human primates (see under “The Animals,” p. xx), the most rewarding compromise appears to be to observe undetected from high vantage points overlooking the sea. Extensive work on dolphin phonation over seven years has revealed that the basic frequency sweep of the communicating whistle of an individual dolphin remains character- istic of that animal (Tietz and Tayler, 1964). Further, the whistle of the recently acquired bull Daan has also retained an individualistic frequency/time curve; when phonation commenced, Dolly likewise developed and retained this identifiable characteristic. This relationship probably remains reasonably constant for the lifespan of the dolphin. Similar constant relationships have been reported in Tursiops truncatus (Caldwell and Caldwell, 1965; 1968b) and were evident in the phonation of humpback dolphins at the Port Elizabeth Oceanarium. We are examining the possibility of applying these findings to determine the ranging and distribution of individual groups of dolphins by recording phonation in free-ranging animals from a motor launch. This method of recording is unsatisfactory: apart from the limitations set by the sea conditions, the dolphins are alerted by the presence of the boat and phonate infrequently, if at all, as is the case in alarmed captive animals. We intend to develop this line of investigation to a transportable system, incorporating waterborne transmitters, incon- spicuously sited where possible, at strategic coastal observation points to record or relay 45 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 2, OCTOBER 1972 data to a receiving station in the acoustic laboratory in the Oceanarium where the data would be processed, ideally by computer. A closer investigation of the components of dolphin sonar may reveal similar individual characteristics. Should this method of acoustical identification of individual dolphins prove successful — and sound transmission in water is not limited by poor water clarity and nocturnal conditions — it should in addition prove possible to correlate phonation of free-ranging dolphins with that known to occur in experimentally controlled nocturnal and diurnal situations in captivity. In drawing comparisons between animals differing as greatly in physiological and anatom- ical structure as do cetaceans and primates, it is as well not to lose sight of the substantial behavioural differences occurring between species even within the same genus. Attention has been focussed in this report upon a number of similarities between baboons and dolphins, but it is not intended to imply that the few, although important, common factors mentioned place them on a definitive behavioural plane. Indeed, if comparisons between non-human primates and dolphins are to be drawn, it is possible that the social organization of bottlenose dolphins will ultimately be shown to have more in common with that described for the more highly evolved chimpanzee (Goodall, 1965). Nevertheless, this review has indicated that the social organization of bottlenose dolphins and baboons are comparable with regard to: dominance hierarchies with co-operative functional behaviours of dominant males co- ordinating group activities ; a mating system of rotating consort relationships without permanent sexual pair bonds; a slow maturation process with strong and extended mother-infant ties; close affinitive ties between females and infants other than their own offspring; highly developed investigatory and exploratory tendencies; play behaviour at times utilizing inanimate objects; play behaviour directed towards other species — for dolphins see Plates 17 and 18 and under “The Animals”, p. 00 in this report and Alpers (1963, pp. 206 — 21) — for baboons see Plates 12 and 13 in this report and Goodall (1965, p. 436); evidence of functional deployment group co-ordinated responses to potential and natural predators. It may be surprising that two forms whose terrestrial links date back 65 million years should have evolved independently, yet share these important behavioural features. A sig- nificant deviation from similar evolutionary behavioural patterns is the perhaps inevitable acoustical adaptation by cetaceans to their aquatic environment. The velocity of sound in water is some four times that in air with a corresponding fourfold increase in wavelength; the cetacean auditory system has therefore adapted to higher frequencies where wavelengths are comparable to those evident in the phonation of almost all terrestrial forms. It is remarkable that our dolphins have not learned to receive or to repeat — and the latter is possible by means of their sonar system, there being no vocal chords — the complex harmonic composition of human speech despite the very close human contact and frequent presentation of selected English and Afrikaans words played repetitively to the animals both in and out of water over many years. Hence, we postulate (in preparation) that the cetacean auditory system has developed without good frequency discrimination and consequently with poor tone perception — not characteristic of most terrestrial forms — but has developed great sensitivity to, and fine discrimination of, relative sound intensities in contrast to, for example, the more logarithmic response of the human auditory system to sound intensity. The sophisticated system of dolphin echolocation must depend largely upon exceptional perception of relative sound levels, the perception of very small time separations in pulsed sounds — as does electronic sonar — and an extremely low threshold of hearing, as yet not measurable. The small delphinid cochlea has relatively few turns, but a very wide frequency response: 100 herz (hz) to 110 kiloherz (khz) (Tietz and Tayler, 1964). This suggests either that only poor frequency discrimination can occur across such a wide frequency spectrum, or that fine frequency discrimination occurs only over a small part of it. Whereas terrestrial animals rely almost entirely upon generated sounds, 46 TAYLER AND SAAYMAN: ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS reflected sound is apparently central to acoustical perception in dolphins. Since echolocation, rather than vision, has been emphasized by evolutionary processes in dolphins it would be logical to assume that acoustical communication would develop in association with the properties of reflected sound as in echolocation. By this reasoning, terrestrial communication would, in contrast, develop on an onomatopoeic basis. Indeed, we are fully aware of the latter in teaching our offspring to speak. These considerations reinforce the interpretation (Tietz and Tayler, 1964) that amplitude modulation of a simple frequency sweep within 2 khz to 14 khz is the means of acoustical communication in bottlenose dolphins. Further, there is evidence to show that the form of some of these dolphins whistle modulations can be correlated with dolphin sonar echoes associated with specific objects, but in '‘summary” presentation. Some of our observations suggest that the communicating whistle is produced by the two independent nasal systems functioning simultaneously in phase-locked mode, and that mod- ulation is effected by phase changes in the sounds emanating from the large, more central right-hand system — considered to be responsible for sonar generation (Tietz and Tayler, 1964) — in relation to the smaller left-hand system. The resulting modulation in phase changes from 0° to 180° may give rise to the typical wave envelope resembling that of an electronic balanced modulator. Our dolphins respond to human gestures within their limited visual range above the water, but the development of two-way communication is impaired by their anatomical structure. However, abnormal fluke, flipper and head movements made at appropriate times imply attempts at communication. It is significant in this connection that a chimpanzee had learned to employ American Sign Language to communicate more than thirty signs, although it has not been possible to develop vocal communication (English) in the same species (Gardner and Gardner, 1969). Acoustical communication in baboons may be divided into two categories employing a probable total of some twenty to thirty differently constituted sounds ranging from grunts to staccato barks. One category forms the means of long-range communication necessary for troop vigilance and cohesion (Stoltz and Saayman, 1970, pp. 130 — 131). The second category relates to the intergroup expression of aggression, fear, friendly and sexual behaviour. The system of dolphin communication, their sophisticated behavioural co-ordination and their extensive cortical development indicate that their level of social integration may well be far more complex than that found in the catarrhine monkeys. The following projects for future research on dolphins are suggested by this comparative review: 1. Acoustical detection, identification and ranging studies of Indian Ocean dolphins. 2. Ecological studies of free-ranging dolphins. 3. Establishment of a larger group of captive dolphins containing representative age and sex classes in order to investigate more fully: dominance — with special reference to functional co-operation between bulls, the nature and influence of pair bonding upon intragroup relations, hormonal factors in reproduction, and the maturation and development of behaviour. 4. Continuation of the investigations of dolphin phonation, sensory perception and cognitive functions. ACKNOWLEDGEMENTS The baboon research was supported by grants from the Department of Higher Education and the South African Council for Scientific and Industrial Research which are gratefully acknowledged. Thanks are due to Prof. J. Meester, Director of the Mammal Research Unit, Department of Zoology, University of Pretoria, the Nature Conservation Branch, Transvaal 47 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 2, OCTOBER 1972 Provincial Administration and to the National Parks Board of Trustees for their co-operation and assistance. The dolphin research was supported by grants from Messrs. John Haig and Company and by donations of Akai recording equipment by Glens (Pty) Ltd. We thank Dr J. R. Grindley, Mr H. F. B. Champion and Mr G. J, B. Ross for their critical reading of the manuscript and Mr P. G. Cury for preparing the drawings. REFERENCES Alpers, A., 1963. Dolphins. (John Murray, London) Altmann, S. A., 1967. Preface. In: Altmann, S. A., Social communication among primates , pp. ix — xii (The University of Chicago Press, Chicago & London). Altmann, S. A. and Altmann, J., 1970. Baboon ecology. Bibliotheca Primatologica. No. 12. S. Karger AG, Basel. Caldwell, M. C. and Caldwell, D. K., 1965. Individualised whistle contours in bottlenosed dolphins ( Tursiops truncatus). Nature, 207: 434 — 35. Caldwell, D. K. and Caldwell, M. C., 1968a. The dolphin observed. Nat. Hist., N. Y., 77: 58 — 65. Caldwell, M. C. and Caldwell, D. K., 1968b. Vocalisation of naive captive dolphins in small groups. Science , 159: 1121 — 3 Caldwell, M. C., Caldwell, D. K. and Siebenaler, J. B., 1965. Observations on captive and wild Atlantic bottlenosed dolphins, Tursiops truncatus , in the north eastern Gulf of Mexico. Contri. Sci., 91: 1 — 10. Carpenter, C. R., 1942. Societies of monkeys and apes. Biol. Symp., 8: 177 — 204. Devore, I., 1965. Male dominance and mating behavior in baboons. In: Beach, F. A., Sex and behavior, pp. 266 — 89 (John Wiley and Sons, Inc., New York). Devore, I. and Washburn, S. L., 1963. Baboon ecology and human evolution. In: Howell, F. C. and Bourliere, F., African ecology and human evolution, pp. 335 — 67 (Chicago: Aldine Publishing Company). Evans, W. E. and Bastian, J., 1969. Marine mammal communication: social and ecological factors. In: Andersen, H. T., The biology of marine mammals, pp. 425 — 75 (Academic Press, New York & London). Gardner, R. A. and Gardner, B. T., 1969. Teaching sign language to a chimpanzee. Science, 165: 664 — 72. Goodall, J., 1965. Chimpanzees of the Gombe Stream Reserve. In: DeVore, I., Primate Behavior, pp. 425 — 73 (Holt, Rinehart & Winston, New York & London). Hall, K. R. L., 1962a. The sexual, agonistic and derived social behaviour patterns of the wild chacma baboon, Papio ursinus. Proc. zool. Soc. Lond., 139: 283 — 327. Hall, K. R. L., 1962b. Numerical data, maintenance activities and locomotion of the wild chacma baboon, Papio ursinus. Proc. zool. Soc. Lond., 139: 181 — 220. Hall, K. R. L. and Devore, L, 1965. Baboon social behavior. In: DeVore, I., Primate Behavior, pp. 53 — 110 (Holt, Rinehart and Winston, New York & London). Kellogg, W. N., 1959. Auditory perception of submerged objects by porpoises. J. acoust. Soc. Amer., 31:1 — 6. Kummer, H., 1968a. Two variations in the social organisation of baboons. In: Jay, P. C., Primates: Studies in adaptation and variability, pp. 293 — 312 Holt, Rinehart and Winston, New York). Kummer, H., 1968b. Social organisation of hamadryas baboons. (The University of Chicago Press, Chicago & London). Lilly, j. C., 1961. Man and dolphin. (Doubleday and Co. Inc., New York). Lineaweaver, T. H., 1967. Of sharks and porpoises. Sea Front., 13: 43 — 6. McBride, A. F. and Hebb, D. O., 1948. Behavior of the captive bottlenose dolphin, Tursiops truncatus. J. comp, physiol. Psychol., 41: 111 — 23. McBride, A. F. and Kritzler, H., 1951. Observations on pregnancy, parturition and postnatal behavior in the bottlenose dolphin. J. Mammal., 32: 251 — 66. Morgan, S. M., 1968. The sagacious dolphin. Nat. Hist., N. Y., 77: 32 — 9. Nishiwaki, M., Nakajima, M. and Tobayama, T., 1966. Preliminary experiments for dolphin marking. Sci. Rep. Whales Res. Inst., 20: 101 — 7. Rowell, T. E., 1966. Forest living baboons in Uganda. J. Zool. Lond., 149: 344 — 64. Rowell, T. E., 1967. Female reproductive cycles and the behavior of baboons and rhesus macaques. In: Altmann, S. A., Social communication among primates, pp. 15 — 32 (The University of Chicago Press, Chicago & London). Saayman, G. S., 1968. Oestrogen, behaviour and permeability of a troop of chacma baboons. Nature, London, 220: 1339—40. Saayman, G. S., 1969. Endocrine factors and behaviour in the old-world monkeys. S. Afr. J. Sci., 65: 121 — 6. 48 TAYLER AND SAAYMAN: ORGANISATION AND BEHAVIOUR OF DOLPHINS AND BABOONS Saayman, G. S., J970a. The menstrual cycle and sexual behaviour in a troop of free ranging chacma baboons ( Papio ursinus). Folia primat., 12: 81 — 110. Saayman, G. S., 1970b. Baboon’s brother jackal. Anim. Int. Wildlife Mag., 13: 442—3. Saayman, G. S., 1971. Behaviour of the adult males in a troop of free-ranging chacma baboons ( Papio ursinus ). Folia primat ., 15: 36 — 57. Stoltz, L. P. and Saayman, G. S., 1970. Ecology and behaviour of baboons in the Northern Transvaal. Annals Trans v. Mus., 26: 99 — 143. Tavolga, M. C. and Essapian, F. S., 1957. The behavior of the bottlenosed dolphin ( Tursiops truncatus ): Mating, pregnancy, parturition, and mother-infant behavidr. Zoologica, 42: 11 — 31. Tietz, R. M. and Tayler, C. K., 1964. Dolphin talk. Scientific South Africa. 1: 385 — 90. Washburn, S. L. and Devore, I., 1961. The social life of baboons. Scient. Am., 204: 62 — 71. 49 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD, CAPE TOWN r ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov . Mus. {Nat. Hist.) VOLUME 9 • PART 3 31st OCTOBER 1972 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA A preliminary study on fruit production in certain plants by R. LIVERSIDGE Alexander McGregor Memorial Museum, Kimberley INTRODUCTION During the course of an ecological study on a fruit-feeding bird, the Cape Bulbul, Pycnonotus capensis , it was necessary to establish, quantitatively and qualitatively, the natural fruit and seed production of the food plants used by the bird. This paper presents the results of a large number of field records taken in an attempt to establish this fruit and seed production in coastal bush near Port Elizabeth. The concept of an orthodox annual cycle of spring, summer, autumn and winter originated in the northern hemisphere under cool-temperature conditions. There, because of the short and definite seasons, it is assumed that all plants flower and produce seed or fruit regularly. In South Africa where drought is a phenomenon that may exist at all times at one place or another (Wellington 1955) the annual cycle is often disrupted. Consequently the ecological scene is one of irregular flowering, early, normal, late or entirely absent. It is unpredictable from one year to the next and seed or fruit may be in extraordinary abundance or sometimes entirely lacking. When seed or fruit is lacking it is not only because of insect or animal preda- tion, as has been established for cool-temperate regions (Weaver and Clements 1938). Such a state of affairs in southern Africa is probably brought about through environmental conditions disrupting the normal reproductive rhythm which, in some trees at least may have a two or four year cycle (Phillips 1931). THE VEGETATION TYPE The area studied is situated halfway between the town of Port Elizabeth and Cape Reciefe at the end of a promontory that lies to the south-east of the town on the coastal dunes (Figure 1). According to Acocks (1953), this area falls under Alexandria Forest complex in subsection called Dune Forest. However, the two indicator species used by Acocks for Dune Forest are absent in the area of study. In fact the “coastal woodland” of Martin and Noel (1960 p. viii) is very much nearer to the vegetation of this area than any other vegetation type described. The plant species mentioned by Martin and Noel under coastal heath are all present. In addition there is a heavy intrusion by the exotic wattle, Acacia cy clops, dominant in much of the area. The main woody plants include Sideroxylon inerme , Rhus crenata , Maytenus procumbens , Euclea racemosa , and Cassine maritima. Elements of fynbos such as Agathosma apiculata and Coleonema pu/chrum, dominate patches of the shorter coastal heath. THE AREA OF STUDY The study area is a narrow strip approximately 20 hectares in extent. From the coastal dunes it runs back 100 to 200 metres to the fence of the Humewood Golf Course and approxi- mately a kilometre long. The dunes are from three to ten metres high with a main ridge 51 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 3, OCTOBER 1972 Fig. 1 . Showing the actual study area of wasted bush and heath with the suburb of Summerstrand in background. parallel to the coast above and behind high tide level. Several valleys run at right-angles to this ridge, extending inland. This break-up of the dunes and lack of regularity affords numerous slopes which are protected from the fairly strong prevailing westerly or easterly winds. A feature of the vegetation around the promontory is the severe wind-blown appearance of the coastal bush. In the study area the prevailing westerly wind comes over land and the on-shore easterly wind has apparently little effect on the vegetation. Consequently the life-form is normal for the species except perhaps that Sideroxylon inerme does not develop into a tree and, in fact, on some dunes has a prostrate growth, keeping it between thirty and sixty centi- metres high though it may fruit prolifically under such conditions. Analysis from aerial survey photographs indicates that 35% of the area is covered by woodland/scrub; 61 % covered by coastal heath and 4% is road and roadworks. The whole study area is on a loamy sand — indeed it forms part of several square kilometres of what was referred to as the Witsands Forestry Reserve. This indicates the extent of the loamy sandsoil in this area. Rainfall is evenly distributed throughout the year; 48% of the precipitation has been recorded between October and March, 52% in the other half year. The lowest rainfall figures for the year are generally in January, with the SE. and SW. winds predominating, and 28^% calm days in July with a westerly wind dominant. 52 LIVERSIDGE: A PRELIMINARY STUDY ON FRUIT PRODUCTION Fig. 2. Showing graphically the natural fruit production for the main plant species studied. (Note: The scale of the ordinate is not the same for all species). Climatic conditions also shown. 1961 1962 1963 53 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 3, OCTOBER 1972 Fig. 2. ( Continuation ) LIVERSIDGE: A PRELIMINARY STUDY ON FRUIT PRODUCTION Fig. 2. ( Continuation ) 1961 1962 1963 FIELD METHOD The aims of this survey were to establish the availability of fruits eaten by the Cape Bulbul both in respect to seasons and quantities at all times of the year. After trials of various methods, a simple field procedure was adopted which is unrefined but has the merit of being practical and of giving a fair comparative result. Three selected paths were taken to cover different areas — one on the edge of bush and fynbos (120 metres), one from the top of a dune down into the dune valley (110 metres) and the third in a protected valley (100 metres). Detailed notes were made at monthly intervals recording for each bush whether flowers, green fruits or ripe fruits were present. The number of fruits from every bush along the route that lay within 80 centimetres (arms length) of the path, on both sides of the path, was counted or estimated when numerous. The height of each bush was recorded approximately so that 55 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 3, OCTOBER 1972 the individual bush was recognizable from year to year. During peak fruiting times a number of ripe fruits were taken and weighed on a triple beam balance. This system would have been more satisfactory if a cyclostyled form had been used indicating each plant on a route, and giving height; presence, numbers or absence of flowers; green or ripe fruits; also presence or absence of young leaves. From a fruit production point of view it is not so important to know the actual number of flowers produced by a plant. Very often if the flowers are “out of season” the correct insect for pollination may not be available (see under Chrysanthemoides monilifera) and this will influence fruit production. Another factor is the predation by birds, which often prevents fruits reaching maturity. Colpoon compressum always had a few red fruits and only once was a fully ripe, purplish fruit seen on a bush. This was when the fruit-eating birds were attracted elsewhere by the abundance of food on another plant. While the work was concerned only with the production on those food plants utilized by the bird under study, details were kept of all larger plants. The results concerning all plants are given in Appendix 1 . RESULTS The results of the observations on the fruit-bearing plants utilized by the Cape Bulbul are given in tabulated (Table 1) and graphical (Figure 2) form with the details for each species. Viewing these results, it is apparent that flowering and fruiting of at least some of the twenty- four species discussed occurs throughout the year. Some species show a regular annual cycle, Table 1 . Showing the flowering and fruiting seasons of plants under observation for three years at Cape Receife. The maximum and minumum number of fruits, green and ripe, are given, as well as number of bushes under observation. f = flower b = fruit brackets used for irregular incidence of either Species January February March April May J June July August September October November December Nos. of green fruits recorded Nos. of ripe fruits recorded No. of months flowering No. of months to ripen fruits No. of plants counted Brunsvigia sp / / / 3 1-2 Vi scum cape use .... / fb fb fb b b fb fb fb 20-350 30-53 3-4 i 4 Viscum sp b 1 i 1 Colpoon compressum f.b fb fb b fb fb fb fb fb fb f.b fb 25-539 1-105 11 3 11 Acacia cy clops .... f.b fb b (f)b (/) b (/) (/) If) fb fb fb 20-680 3-235 5 11 9 Carpobrotus edulis . ) / fb b b b 3 J — Agathosma apiculata f f fb fb fb 4 1 — Coleonema pulchrum f / f f f fb fb b 7 4-5 6 Mundia spinosa .... b fb fb f f f f.b fb b b 60-840 5-420 7 3 11 Rhus crenata .... fb f fb b b fb fb f fb b b b 60-2 600 40-9 100 3-4 3-4 48 Rhus glauca f b lb) lb) b b 14-400 300 1 2-3 1 Maytenus procumbens . f fb fb fb b b b b 450-4 980 345-1 320 4 3 17 Cassine maritima f b b b b b 50-1 140 30 1 3-4 8 Cassine tetragona fb f.b f.b fb b b f 20-330 40-200 1 1 3 Rhiocissus digitata . fb fb fb b b b b b b b fb f.b 8-179 1-143 5 3-6 11 Passerina rigida .... fb b f fb fb fb 5 1 — Rapanea gilliana b b b fb fb fb b b b b b 25-200 25-220 3 4-6 6 Sideroxvlon inerme . fb b b b b b fb fb fb fb fb 48-350 5-50 6 9 6 F.uclea racemosa b b f.b b b b f f b b 9-600 1-4500 1-2 1-2 3 Olea exasperata fb f.b fb fb f fb f fb 3-440 8 1 2 Chironia decumbens f f fb b b b b f 20-470 23-160 4 4 9 Salvia africana-lutea f f f f 4 — Solanum quadrangulare f f f f f f f.b 7 1 — Chrysanthemoides monilifera fb fb fb b b b fb fb fb f f 17-1 600 6-290 8 3-5 16 Total no. species flowering each month .... 11 11 12 8 9 7 10 12 13 10 6 9 Total no. species in fruit each month .... 11 8 15 15 12 13 10 9 16 15 11 10 56 LIVERSIDGE: A PRELIMINARY STUDY ON FRUIT PRODUCTION the remainder are irregular in their fruiting. The irregularly fruiting species may have a half-yearly, yearly or biennial cycle which varies slightly according to seasonal changes. Though records were kept of individual bushes the results are not presented in detail. This is mainly because of lack of regular monthly counts but partly because of uncertainty as to what took place between counts. A minimum of regular fortnightly counts is required to give adequate figures for detailed information. Mention is made under the species of points taken from these chronological records of individual bushes. Some bushes appeared to be heavier bearers than others nearby but the data are too few to prove this point. Indeed, two seasons may show a reversal between two bushes — what was formerly the heavy bearer may the next season have a poor crop compared with the neighbouring bush. Flowering It is assumed that the interval between budding and the ripening of the fruit is more or less constant in individual species. It is self-evident that the fruiting period depends for the Table 2. The weight of fruits recorded at different times as well as dry weights (oven dried at 10°C for 48 hours) of some fruits. Date Wt of 100 fruits Dry wt of 100 fruits % increase in weight over lowest grammes grammes recorded wt Viscum capense 9. 3.64 12,5 Acacia Cyclops — 17,3 4,69 — 6.12.69 8,9 5,12 — Acacia cy clops: aril only — 7,3 1,27 — 6.12.69 7 1,63 — • Mundia spinosa 30.11.62 57,3 — — 6.12.69 49,5 11,36 16 Rhus crenata 31. 7.61 1,37 0,66 — 30.10.62 2,10 — 53 Maytenus procumbens 27. 3.61 13,4 8,29 — 3. 5.63 39,4 — 194 Cassine maritima 31. 7.61 18,2 4,67 — Rhiocissus digitata 30. 3.62 28,7 — — Rapanea gilliana 31. 7.61 43,6 5,87 — 3. 5.63 72,2 — 66 3. 6.63 76,2 — 75 Sideroxylon inenne 27. 3.61 7,6 — — Euclea racemosa 30. 1.63 9,6 2,59 — 30. 3.62 13,1 — 37 9. 3.64 14,0 — 46 27. 3.61 17,1 2,59 78 26. 4.63 22,0 — 130 3. 6.63 28,8 — 300 Olea exasperata 27. 3.61 22,0 8,05 — 3. 6.63 28,8 — 31 Chironia decumbens 27. 3.61 10,3 — — 30. 3.62 13,4 — 30 9. 3.64 13,7 — 33 Chrysanthemoides monilifera .... 30. 3.62 15,0 — — 26. 4.63 19,4 — 29 Solarium sp 6.12.69 61,9 8,37 — 57 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 3, OCTOBER 1972 most part on the period of flowering. Thus if flowering is early it is likely that ripe fruit will also occur early. Exceptions occur when flowering is out of phase with the pollinating agent (see Chrysanthemoides), or when fruit fails to mature (see Rapanea) for various reasons mainly connected with unseasonable weather periods. Flowering in the area occurs throughout the year (Table 1). Of the 24 species of plants reported the number flowering each month varied from a minimum of 7 to a maximum of 13. The highest numbers per month fell into two periods from July to September and January to March. The length of the flowering period for each species varied from one month to eleven months. Eight species flowered for only one or two months and nine species flowered for more than six months of the year. Fruit Production With the variation in time between budding and ripe fruit from two weeks to eleven months depending on the species, it is interesting to find two peaks for the number of species fruiting in a year. Autumn, March to May, is the most prolific period of fruiting and a secondary peak occurs in spring, from September to November. The average number of plants fruiting agrees closely with the flowering — namely 10,0 per month. The least number of species fruiting in one month was seven and the greatest thirteen. The variation in fruit production is considerable, both quantitatively and qualitatively. Reference to Table 1 indicates the extremes in number of unripe fruits in any one species along the routes taken to be 60 minimum to 2 600 maximum for Rhus crenata and 450 to 4 980 in Maytenus procumbens. For ripe fruits the extremes counted were 40 to 9 100 for Rhus crenata. The quality of the fruits varied from season to season. While most fell between 34 to 1 19% difference of fruit weight from year to year, the greatest difference recorded was in Maytenus procumbens. In this species 13,4 g for 100 heads was recorded one year and 37,4 g for the next crop two years later (Table 2). DISCUSSION Phillips (1931 : 245) pointed out that there was practically no information then concerning the biology of the flowers and fruits of the more important trees and shrubs. Little progress has been made since 1931. Phillips himself gives details of 63 species of trees and woody shrubs occuring at Knysna. The only species in his work in Knysna and the coastal-bush under study are Euelea racemosa and Sideroxylon inerme. It is shown above (Table 2) that the timing, quantity and quality of fruits varies from year to year. Phillips (loc. cit. p. 246) has shown how complex these matters are: “Individual plants near the sea flower and fruit several weeks to months before their relatives in the inland plateau or mountain kloof (forest) patches”. Coastal conditions are usually slightly warmer and rather drier than inland. Flowers are sometimes further advanced and more prolific on individual plants on the warmer northern or north-western aspects. Species near the coast flower more prolifically than those inland, a point which can be correlated with the observation that trees in the drier forests usually bear richer crops than those in moist forests — bearing in mind that this is relatively speaking, since all the forests are moist forests in the region dis- cussed. These differences would explain several of the differences in flowering time as recorded in the study area from those of the Albany and Bathurst division (taken from Martin & Noel 1960). Details of fruit production in relation to climate are given in Figure 2. If, in fact, these two can be correlated it becomes possible to predict fruit crops in advance. The problem is not so much correlating peak of crop with peak in weather but why peak in weather does 58 LIVERSIDGE: A PRELIMINARY STUDY ON FRUIT PRODUCTION not always result in peak in crop. A rest period for the plant (which need not be required at all times of the year) would probably explain some observations. Flower production at periods when pollinating agents are unavailable would also influence fruit production. The difference in size of fruit available is not so easily correlated with climatic factors. If we refer to Figure 2 it can be seen that actual rainfall one month prior to fruit production in general was very much less in 1961 than 1962. Also temperature were ascending at this time in 1962 and warmer weather may have helped. It is unlikely that degree of pollination differed markedly. Rhus laevigata is poorly pollinated by Apis mellifera and Apis caffra. Ants are active on Rhus crenata throughout the year. The two months of probable pollination had good rainfall but the minor peak of fruit production in the autumn of 1962 occurred when the rainfall was rather low. Predation on buds and young ovaries in these plants is known to take place in the area of study, particularly by members of the seedeater family, Fringillidae. This, however, is unlikely to have a significant influence on the fruit crops of such heavy producers as were studied. The observed facts are too limited to be able to suggest reasons for seasonal variations in fruit production. They do, however, show the irregularity of season and production from year to year. From an ecological point of view it is desirable to determine the variables that affect fruit production, and the time of fruit ripening. ACKNOWLEDGEMENTS For identification of plants my thanks are due to Mrs. N. Urton, Miss G. V. Britten, Mr D. Comins and Mr M. Wells. Discussions with Prof. E. A. Schelpe have assisted in formulating a method of determining ecological variables. REFERENCES Acocks, J. P. H., 1953. Veld types of South Africa. Bot. Surv. S. Afr. Mem. 28, Pretoria. Martin, A. R. H. and Noel, A. R. A., 1960. The Flora of Albany ami Bathurst , Grahamstown. Phillips, J. F. V., 1931. Forest-succession and Ecology in the Knysna Region. Bot. Surv. S. Afr. Mem. 14. Pretoria. Weaver, J. E. and Clements, F. E., 1938. Plant Ecology, New York. Wellington, J. EL, 1955. Southern Africa: A Geographical Study, Cambridge. APPENDIX Details of flowering and fruiting of plants presented in order according to that used by Martin and Noel 1960. AMAR YLLIDACEAE Brunsvigia sp.: Common in open areas of coastal heath, this spectacular flower appears from February to April, the leaves appearing from June to August. LORANTHACEAE Viseum capense L.f. : The Cape Mistletoe is a fairly common parasite in the area. Flowering and fruiting occurs in the spring and autumn; the fruit is ripe in a couple of weeks. Flowers and berries were not produced in spring, 1961 and autumn 1963. Viseum (species unknown): This is a larger leafed plant with larger fruit. Recorded in fruit in September 1962, only. 59 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 3, OCTOBER 1972 SANTALACEAE Colpoon compression Berg: A fairly common tall shrub usually occurring in clumps within the bush. Recorded flowering throughout the year in the study area, though only from April to October in the Albany and Bathurst flora. There is a period of about three months before fruit is ripe and the main fruiting season is from April to July. No green fruits were recorded in May. The only occasion fully ripe purple fruits were recorded was on 26th April 1963. This month, Chrysanthemoides , the favourite food of the Cape Bulbul and Maytenus, and the favourite of the Sombre Bulbul, Andropadus importunus, were at peak fruit production. Both birds feed upon Colpoon normally but during this time they fed elsewhere, this allowing the Colpoon fruits to ripen. This is the only period in three years when this was noted. There was no major production of fruit in the autumn of 1961 such as there was in 1962 and 1963. LEGUMINOSAE Acacia cyclops Cunn.: This alien wattle ( Rooikrans ) was introduced into the area to bind the sand dunes. It has become dominant in the general area and several patches occurred within the study area. Flowering occurs prolifically and simultaneously throughout the area at the time the pods ripen. This is from October through to February. A few flowers may be present at other times, though none were recorded for March and June. Young pods are noticeable from January though more were recorded from March onward. By June the pods are almost full- sized and green. The pods become brown in Spring and by October they begin to crack open, showing a red fleshy aril. The dry pods open with quite a loud crackling noise during January and February, ejecting the dried red aril and seed. The peak period of abundance of ripe pods varies; January in 1961, December in 1961 whilst there was an extended period from October 1962 to January 1963 — evidently a “good” year. This plant is frequently cut for firewood, and most trees under study were eliminated so that individual bush records were not available. AIZOACEAE Carpobrotus edulis (L) N.E.Br. : The “hottentot fig” is a common ground cover on the sandy areas where it sometimes forms solid patches. Flowering occurs from late July through to September, with fruits from September onwards. This species appears to have a more regular annual cycle than many other local plants. RUTACEAE Agathosma apiculata G. F. W. M ey apud Bartl. & Wendl. : Common on the flat areas of coastal heath. Flowering begins in late June and extends through to October in the study ar#a, though it is recorded until January in the Albany and Bathurst areas. Fruits do not remain much later than the flowers, being recorded from August to October. Coleonema pulchrum Hook.f. : Present in patches, this attractive little shrub is in flower most of the year, though few flowers were present from November to March. Seed heads were recorded from September to November. POLYGALACEAE Mundia spinosa DC: The “waxberry” is a common low bush on the sand dunes and was locally abundant enough to form the supply for a small wax industry in the early days of Port Elizabeth. Flowering is recorded from April to October, with the height of blooming in 60 LIVERSIDGE: A PRELIMINARY STUDY ON FRUIT PRODUCTION July in each year. The waxy fruit ripens about three months after flowering, for reasons which are not apparent, for the facts do not fit in with flowering times. The berry crop has two distinct seasons, March to May, and the more important spring crop October to December. Timing of flowering and berry production is uniform and simultaneous in the study area. Crops were produced in the spring of 1961, 1962 (no observation — spring 1963) and autumn 1964. One bush which produced a good spring 1962 crop produced a poor autumn 1964 crop whilst two which had poor and good crops respectively in spring 1962 had good crops in autumn 1964. Individual variation thus occurs. ANACARDIACEAE Rhus crenata Thunb.: Perhaps the commonest indigenous shrub, it is widely distributed throughout the study area. Though not clearly shown in Table 1, this species has two flowering seasons each year and consequently two fruiting seasons. The fruit ripens in three to four months. Berry production does not occur every spring and autumn. Very poor berry crops were produced in the spring of 1961 and autumn of 1963. Furthermore, the autumn crops of 1961 and 1962 were very much lower than those of the springs of 1962 and 1963. Not only was there a quantitative difference but also a qualitative difference between crops. Thus 100 berries in the autumn of 1961 weighed 1,37 g whereas the same number of berries weighed 2,10 g in the spring of 1962, an increase of 53% by weight. Individual bushes may flower twice in a year whilst another bush may miss out a flowering period and thus have three crops in two years. One bush had only one good crop in two years. There is thus much individual variation between plants. This results in a wider flowering and fruiting period than other plants. Rhus glauca Desf. : This larger-leafed Rhus is less common than the previous species and tends to occur in the coastal heath areas. Only one specimen was subject to observation. This individual flowered in May 1961 and had fruits again two years later in June and September 1963. This indicates the possibility of a species which flowers every second year. In view of the variability shown by Rhus crenata this needs confirmation. Fruits ripen about three months after flowering. CELASTRACEAE Maytenus procumbens (L.f.) Loes. : Common but not dominant except in localized patches of growth. Flowering was noted from February to May, green fruits from March to June and ripe fruits from April to September. In the area under study, flowering and fruiting occurred in 1959, 1961 and 1963 indicating a two-year cycle. Of some interest is that all the plants in the surrounding areas had the same season. In other words fruits were only produced alternative years and as a food source for berry-eating birds this would have disadvantages. Fruit production in 1963 was more prolific than in 1961. It should also be noted that the fruits too were larger and weighed nearly three times as much in 1963 — 100 fruits being 13,4 g in 1961 and 37,4 g in 1963, an increase of 193%. Cassine maritima (Bol.) L.Bol.: Not uncommon tall shrub with small edible fruits. The flowers were recorded in January with fruits from March to July. This species only produced fruits in 1961 and 1963 and it is therefore believed to have a two-year cycle. Cassine tetragona Loes.: There were only three plants along the paths taken for these counts; these three plants produced flowers and fruits only during the one summer, December to April 1962. It is probable that this species has a cycle longer than one year but the exact period is not determined. It should be noted that flowering is recorded in July for Albany and Bathurst. 61 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 3, OCTOBER 1972 VITACEAE Rhiocissus digitata Gilg et Brandt: This creeper is commoner growing on vegetation with a sheltered northern aspect, thus it occurs in patches. The flower is short lived and for this reason flowering was recorded only from November to March. Berries were recorded every month of the year but these may take several months to develop and ripen. On the one path where this species was common there was a peak production in the summer 1961-2 and the next peak was in winter 1963. THYMELAEACEAE Passerina rigida Wikstrom: This tall shrub is widely scattered in the survey area. Flowering occurs from September through to January with fruit recorded from October to January. It flowers annually regularly, with flowers and buds recorded 21st September 1961 and buds only on 21st September 1962. MYRSINACEAE Rapanea gilliana Mez. : This little bush is subject to considerable variations in its fruit produc- tion. The flower is short-lived and though flowers were only recorded May to July there appears to be a double crop of fruit each year. The fruit apparently requires rain for final development. During one dry period the fruits remained on the bushes for two months longer than normal, during which time they became sunburnt. One bush had 100 fruits March 1961 and its second crop in September was only 16 fruits. In a stand of this species beyond the route recorded, prolific flowering was noted in August but no fruit was set, possibly because this flowering was abnormally late. Yet this same patch produced unusually numerous and large fruit seven months later. The fruit itself varies from crop to crop. Thus on 31st July 1961, 50 fruits weighed 21,8 g whilst on 3rd June 1963 the same number weighed 38,1 g, an increase of 75%. The size of a ripe fruit varied from 6-8 mm in diameter to 12 to 15 mm, on the same dates mentioned above. SAPOTACEAE Sideroxylon inerme L. : Locally a wind-formed low shrub, not uncommon. Buds and flowers were recorded from late August to late January. This plant would seem to flower annually with fruits taking about nine months to develop from small and green to ripe. A peak pro- duction occurred in March 1961 and again in November 1963, and although the figures are inadequate to draw any conclusions there appears to be a difference quantitatively from year to year in peak fruit production. EBENACEAE Euclea racemosci Murr. : A fairly common shrub which grows, in this area, up to 2,20 m high. The same plant carries two crops of fruit in one year. Flowering occurs in August and again in March and the fruits ripen within four to six weeks. The seasons may vary from year to year by a month forward or a month later. There is also a considerable difference in quality of fruit from season to season — thus 50 ripe fruit weighed 6,34 g in March 1962; 11,0 g in April 1963 and 14,0 g in June 1963 — a total increase of 119%. OLEACEAE Olea exasperata Jacq. : This shrub, which grows up to 2,20 m high, is patchy in its distribution. In exposed positions it remains only 6 cm tall. Recorded to flower in Albany from August to November, under local conditions it seems to flower most months of the year. Despite this prolific flowering fruits were recorded only in March and May 1961 and again in December- January 1961-2. 62 LIVERSIDGE: A PRELIMINARY STUDY ON FRUIT PRODUCTION GENTIANACEAE Chironia decumbens Levyns: This attractive low bush grows up to 60 cm high in patches in semi-open sandy spots. Though a definite flowering period occurs from December to March, the odd flower was recorded in other months. The flowering period is four months earlier than that recorded for Albany and Bathurst. The fruits take about four months to mature and thus ripe fruits occur for the most part from March to June. There is some difference in quantity and quality of fruit crop from year to year. Thus a bigger crop of smaller fruits appeared in March 1961 and a smaller crop of bigger fruits was produced in April 1962, while in 1963 the crop was even later. Weight of 100 fruits on 27th March 1961 was 10,3 g and 35 on 30th March 1962 weighed 4,70 g, which represents an increase of 33%. LABI AT AE Salvia africana-lutea L. : Locally this shrub occurs within patches of mixed species of shrubs about 1,20 m high. Flowering occurred from July to September in 1961 and 1963 and from September to October in 1962. SOLANACEAE Solarium quadrangulare Thunb. : This creeper was not common and its appearance seems short-lived. Flowering occurs over a long period from February through to August. Ripe berries were only recorded in August probably because they are quickly eaten by berry- feeding creatures and have little opportunity to ripen on the plant. COMBO SITAE Chrysanthemoides monilifera (L.) T.Norl.: Perhaps the commonest indigenous shrub, this plant grows up to 2,40 m high and is dominant in the vegetation. Flowering occurred annually from September (as early as August in 1963) through to March. The fruit took from three to five months to ripen. Differences in season and quantity of flowers and fruit varied enormously from year to year. The heaviest flowering occurred in March 1961, when no insects were seen on the flowers and the subsequent fruit crop was the poorest of the three years. In 1962 the peak fruit production was recorded in April, while in 1963 this extended from April to June. One hundred fruits on 30th March 1962 weighed 13,7 g, while on 24th April 1962 the same number weighed 19,4 g, representing an increase of 34% by weight. It is interesting to note that a beetle, Melyris interstitialis, occurred on the flowers from October to December, and a green Melyris sp. from January and February. Individual bushes may not produce fruit each season. 63 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN tot C23¥ SJ ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mas. (Nat. Hist.) * K VOLUME 9 • PART 4 31st OCTOBER 1972 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Dress, personal decoration and ornament among the Ndlambe by E. H. BIGALKE Ethnologist, East London Museum The Ndlambe tribe is part of the Xhosa tribal cluster of the Transkei and Ciskei. Though the Ndlambe previously occupied a larger area, they are today confined to the East London district, where some 30 000 of them live in five discrete areas lying between the Kwelera and Chalumna rivers. This paper relates to those Red (traditionally-oriented) Ndlambe who live in an admini- strative area to the east of East London. The data to be presented concern the greater pro- portion of the population of this administrative area, Tshabo. In April 1968, 252 (70,18%) of the total number of households belonged to Red people. This accounts for over 1 600 people. The other section of the population is the School (Christianized or westernized) people, whose way of life differs radically from that of the Reds. Despite this, Red and School home- steads are intermingled and individuals of these two groups have many contacts in everyday life. The research data presented here were obtained between late 1967 and mid- 1970, as part of a larger project. Despite the fact that they relate specifically to only a small proportion of the Ndlambe population of the East London district, these data will be found to have wider relevance. The reason for this is the cultural uniformity of the Ndlambe, indeed of the Xhosa tribal cluster as a whole, at least in material culture. No attempt has been made, either while questioning informants or in committing research findings to paper, to reconstruct the traditional patterns of dress, personal decoration or ornament as they existed at some earlier time. This is rather a picture of the situation between 1967 and 1970. Any comments about earlier conventions are made to illustrate changes in taste and fashion. It is felt that the contemporary situation is worth recording for its own sake, especially as few explicit accounts of Southern Nguni material culture have been published and as, with the increasing industrialization of the Ciskei, social change will be accelerated, A general feature of the life of all the tribal groups in the Ciskei and Transkei is that women and young girls, but to a much lesser extent youths and young men, have adhered to non-western dress. All Red women habitually wear at least the traditional, red-ochred skirts; many of them wear the items to be described below. Girls of all ages also wear mainly tradi- tional dress. Youths and young men, except the few who are not migrant labourers, wear traditional dress only at weekends, when feasts and rituals are held. As regards older men, it seems that the habit of wearing western dress acquired during their long and frequent so- journs as migrant labourers in the towns, has effectively displaced traditional forms. Even those older men who have never worked in towns, or those who no longer go away to work are seldom seen in non-western dress. An aspect of dress habits in Tshabo is that those people who habitually wear non-western dress, generally speaking, have broadly the same kind of outfit for ordinary and special occasions, but for the latter there is greater attention to detail and much more use of ornament. The dress, decoration and ornament of each group is characterized by its general uniformity. Virtually all the objects described here are in the collections of the East London Museum, 65 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 Four-digit numbers in the text refer to these specimens. Sufficient incidental reference to items of dress exists in the literature on the Southern Nguni to make it unnecessary to describe these in minute detail. In describing articles of clothing here, the Xhosa word bhayi is used in preference to “kaffir sheeting” for which a less offensive description could not be found. There is virtually no published material about beadwork and other accessories, thus full details of these objects, including vernacular names and measurements, will be set down. BOYS AND YOUTHS The first distinction is that between the dress of pre-adolescent males, on the one hand, and that of adolescent and other uncircumcised males. At an early age the importance of the principle of seniority is stressed, in the formation of age-grades. All boys up to about the age of 12 or 13 fall into one broad group. It has no specially distinctive dress. Its members wear western clothing. Apart from the wearing of short trousers, this dress is distinguished only by its obviously cast-off origin. Between about 13 or 14 and circumcision, long trousers are worn on weekdays. Among uncircumcised males over about 14 there are three distinct age-grades. In order of seniority, these are the first grade ( umbhutho wokuqala) or “big boys” ( amakhwenkhwe amakhulu), the second grade ( umbhutho owesibini) or “young oxen” ( amadyongo ) and the third grade ( umbhutho owesithathu). On Saturdays and Sundays the first and second grades — and, to a much lesser extent, the third — dress in a special fashion. Each youth wears a pair of shorts ( ibhu/ukhwe ) ending two or three inches above the knee. These shorts are generally one of two varieties — navy melton cloth (5503, ibhulukhwe yesitafu inevi), or one or more colours in a thinner cotton or man-made fibre (5504, 5505; ibhulukhwe yelaphu). The whole garment may be of a single colour or be parti-coloured. If plain coloured, it is ornamented with rows of small glass beads in white, orange and navy; it may be further ornamented by having patches of a contrasting colour applied or rows of lace sewn on the legs. This is the only garment. It is usually held up by one of two types of beaded belt. The ibhrinks or ifigar (5510, 5511) consists of three or four narrow (13 mm. wide) leather straps placed horizontally one above the other and sewn together with rows of small glass beads in white, navy and turquiose between them. It is fastened by means of a buckle attached to each of the straps. The other variety of belt is the unoncyiwana (5512, 5513), a Boy-Scout-type leather belt of three sections held together by means of two metal rings. The upper and lower edges of the two shorter sections are fringed with strips of beadwork in the same colours described above, sometimes with orange beads added. The upper edge of the central section has the same kind of fringe but from its lower edge hangs a wide rectangle of beadwork; this consists largely of white beads, with geometric motifs in turquoise and navy-blue beads. It is often fringed with pink wool. The torso is not covered, but is elaborately ornamented. This ornamentation will be described below, under the appropriate heading. The head may be covered by a green or navy melton peaked cap (5547; ikepsi ), decorated with rows of white and orange beads or by a green cotton square or triangle (5075; iqhiya — see below) similarly ornamented with beads. The feet are either bare or covered by tennis shoes ( amatekisi ) with the optional addition of rugby socks (5548-50; amakawuse) in a variety of colours. Rectangles of angora goat-skin (5550; isybok ), with hair attached, may be worn about the calf. This is said to have been copied from Mfengu youths. The skins are bought in town, as this breed of goat is not kept in Tshabo. When they go to be circumcised, youths are naked, except for a waistband of bark fibre ( uluzi ), to which the bandaged penis will be tied after the operation, and blankets. There will be either two grey, shop-bought blankets ( iingubo ) or a woolly sheepskin blanket ( ingubo 66 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE yegusha ) ordered for the purpose and made in the location by a part-time specialist. These blankets are used throughout the seclusion period as clothing and for sleeping. Once the cut has healed, a plain sheep- or goat-skin penis sheath ( isidla ) is worn. This article is burned with all others the umkhxvetha (circumcised young man in seclusion) has used during seclusion. In theory, the blanket or blankets are also burned but in practice nowadays they are removed from the seclusion hut. The woollen ones are washed and given to a younger brother or other small boy. YOUNG MEN ( abafana ) On coming out of seclusion, a new young man (irwala) is given a new, plain penis sheath ( isidla ) to put on before returning to his home, and wears a new woollen blanket ( ingubo ), usually grey, sometimes white. This form of dress is maintained for only a day or two after the coming-out. After this he wears long khaki trousers, a khaki shirt, a jacket and a hat, all of which are worn until he leaves the reserve to find work, either in town or on the mines, when there is free choice of items of western dress. Traditional dress is worn only on four types of occasion — abafana s dances ( iintlombe ), at bridewealth discussions when ikhazi cattle have been delivered, at the traditional wedding ( umdudo ) and when an umfana accompanies the diviner presiding over the umhlwayelelo ceremony at a river. The most frequent of these occasions is the dance ( intlombe ) held within Tshabo or in a neighbouring reserve most weekends. All abafana wears a pair of shorts, but most of the time this is covered by a blanket; this is basically a white blanket bought in a shop. It is treated with a strong or very weak solution (depending on personal preference) of red ochre ( umakaba ), either beaten in dry or as a solution in water. When walking, the wearer has the blanket draped about the shoulders, completely covering the body, or arranged so that one corner of the blanket can be drawn forward to cover the left shoulder, while another corner passes beneath the right arm, being held by the left hand and leaving the right shoulder and arm ex- posed. For dancing, the blanket is draped about the waist, either in tubular fashion, with one end tucked in at the waist, to secure the blanket, or by folding it so that the corners are free and can be tucked in at the waist, giving the appearance of a flared skirt. Apart from differences in bead ornaments, which will be described below, all abafana dress in this fashion for dances. Marriage does not affect the types of articles worn. However, at about the age of 45 an umfana , sometimes at a specially arranged beer drink, is promoted to the ranks of the senior men ( amadoda , sing, indoda) and ceases to attend dances, otherwise this promotion occurs on an ad hoc basis. Both abafana and amadoda can serve as cattle drivers and go-betweens ( onozakuzaku ) in bridewealth and marriage negotiations. When an acceptable bridewealth has been settled, the future bridegroom’s cattle are driven to his prospective wife’s home. On this occasion two or three abafana don traditional dress before they drive the cattle along. The version of traditional dress on this occasion may omit the shorts, a plain, short-tasselled penis sheath being sub- stituted but the blanket is still worn. Essentially the same dress is worn at the week-long series of events during the traditional wedding (umdudo). Amadoda wear the same, their costume differing in the type and number of bead ornaments worn (see below). Some, though not all, senior abafana and amadoda , when at home, wear isampulu, a blanket made of small rectangles of men’s suiting cloth, lined with a blanket or other warm fabric. Most adult men at some time wear a melton headcloth ( iqhiya ) in blue, brown, green or black. It may be plain or ornamented with machine stitching in white. Older men, when in western dress, often wear a crocheted woollen cap ( isinkwane ) of brown, green and cerise, in preference to a hat. 67 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 GIRLS As among males, there are age-grades among females. Uniformity of dress is marked amongst all prepubertal girls. They all wear western clothing, especially those Red girls of school-going age who wear uniforms to school. The inciyo, a cloth or beaded cloth waistband with a fibre fringe used to cover the genitals, is worn especially by baby girls and toddlers. It is, in theory, worn by all other unmarried females and may, in fact, be worn by a proportion of them but it is generally thought by Tshabo people that bloomers ( ibhulumasi ) are becoming increasingly popular. Women, especially, attribute the large number of illegitimate births among unmarried girls to the decline of the inciyo. In the past, it was considered an effective mechanical contraceptive, being tucked between the thighs when a girl had intercrural inter- course. Unlike their adolescent male counterparts, Red girls who do not attend school habitually wear partly-traditional clothing. This consists of a plain cream or red-ochred underskirt (unomtidili wangaphantsi) of bhayi worn with or without an overskirt ( unomtidili wangaphezu/u ) of the same fabric. The underskirt is worn about two inches above the knee, the overskirt slightly shorter. The inciyo may be worn with this. The torso is sometimes left uncovered but more frequently, a vest, T-shirt or jersey is worn. It is not thought essential for unmarried girls to cover their heads, so the use of a brightly-coloured square or triangle of fabric as a head- cloth is optional. The feet are usually left bare. This kind of outfit is worn both during the week and at weekends, when the girls accompany their boys to dances or march in groups corres- ponding to those of the boys. The only time a variation on this outfit is seen is at the unmgqungqo dance, which is part of the feast held shortly before an intonjane emerges from seclusion. Then, the adolescent girls wear a long-fringed inciyo in front. This garment has an overlay of beads above the fibre fringe and the waistband consists of a cloth core strung through a number of small-diameter brass rings. Instead of a skirt, a number of brightly-coloured cotton cloth squares are worn about the buttocks and extend part-way around the waist. They do not obscure the fringe of the inciyo. The cloths are held in place by means of a belt-shaped bead waistband ( inyilingo ) of motifs in black glass beads on a white glass bead background. This is often fringed with short strings of white glass beads. When the youths with whom senior adolescent girls consort have been circumcised, these girls are promoted to a more senior group who are now eligible for marriage and no longer go about with youths. Their everyday dress is similar to that of the stage out of which they have just passed, except that the skirt is lengthened so that it falls just on or just below the knee. Some girls of this stage wear red-ochred bhayi skirts but the majority leave the fabric undyed, especially for wearing to dances. The torso is covered, as before, with a jersey or T-shirt. For the weekend intlombe these abafana's girls wear only the skirt and go bare-breasted. Apart from the ornaments to be described below, they like all other unmarried girls and married women, wear a large rectangle of plain or ochred ibhayi about their shoulders as a cloak (ibhayi) when walking out in cool weather. This article may be unornamented but often has the lower edge bordered with narrow black braid and geometric motifs in black braid applied towards the bottom centre of it. Groups of mother-of-pearl buttons are applied for orna- mental purposes; this is said to be a usage adopted from theMfengu, who were formerly the only people who used buttons in this way. Abafana's girls wear a distinctive, brightly-coloured headcloth ( iqhiya ) of cotton fabric tied about the head in such a way that a wide band stands high above the crown of the head, while pieces of brightly-coloured fabric of contrasting colours are tucked into the band. Although different, coloured cloths are used for everyday and intlombe wear, the style remains the same. The feet are bare at dances but black, lace-up shoes are often worn for walking when an intlombe is held at some distance away from the home village. 68 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE MARRIED WOMEN The hallmark of the married woman is the flared, ankle-length skirt ( umbhaco ) worn from the time when, with her entourage, she sets out from her own home for that of her husband, in the case of a traditional wedding, or from when she is brought there by force, in the case of marriage by capture ( ukutwala ). This skirt is made of ibhayi or, less frequently, of a thicker, woolly fabric called ifelani. The latter may be preferred by younger women but is more expensive than ibhayi. All skirts are bordered at the bottom and sides, though not at the top, with thin black braid ( ibleyiti ) sewn on by hand. There are a number of rows of thicker black braid ( ibleyiti ) applied by hand or machine-sewn on to the lower part of the skirt. The rows are usually slightly separated from each other, but they may be so close together as to convey the impression of being a continuous black panel. The characteristic fullness of the skirt is achieved by cutting a large square of fabric on the cross, then folding it diagonally in half and cutting out arcs for the waist and hem. An alterna- tive method is to cut the skirt in three panels, all on the cross, and then join them. The upper edge, which rests just above the buttocks, may be given either a pleat or a dart and reinforced in this spot with an extra piece of fabric. A pleat may be put in below the dart and it is then often held in position by means of a rectangular patch of cloth. Some women cut out and sew their own skirts. Others employ semi-professional seam- stresses in the location, some of whom possess hand or treadle sewing-machines. The customer supplies the cloth and is charged between R3 and R6 for the making, according to the pattern she wants and the elaborateness of the decoration (black braid and pearl buttons) which she also supplies. For everyday wear the skirt is dyed with a red variety of ochre ( umakaba ) bought in shops. Those of new brides appear particularly red because they are new and the excess ochre powder has not yet been worn off. With age, though clothing is washed, skirts take on a browner shade of red. This dye has a low degree of colour-fastness and has to be re-applied at intervals, after the garment has been washed. Most women have two skirts, one for home and field work, another for special occasions. Increasing use is, however, being made of undyed bhayi skirts for special occasions. Though the inciyo is worn mainly by unmarried females, some married ones are said to wear it when going to feasts in case they get drunk and inadvertently expose themselves. A small number of women possess the isikhaka (pi. izikhaka), a skirt made of home- tanned cowhide. A Ndlambe specimen (5004), like other Xhosa and Bomvana ones in the East London Museum collection, is cut from a single hide. Apart from being cut in the shape of a wide-based, flattened cone with the point cut off, it is not shaped, nor is there a tuck or pleat, as in the umbhaco described above. The hair is not removed from the skin. The hairless side is worn outside. The isikhaka is worn only at imigidi, the feasts held to celebrate the coming-out of young men after circumcision, shortly before an intonjane emerges from seclusion, or at the traditional wedding ( umdudo ). The incebeta is a narrow rectangle (approximately 30 cm. wide) of ibhayi , with a length of thick string or two thin strips of bhayi attached to one of the narrow ends. By these it is tied at the back. This breast cover hangs down from just above the breasts to the level of the knees or lower. Though often plain, it is sometimes ornamented with rows or geometric motifs of black braid and edged at the bottom with large, milky-white beads ( amaso ). This garment is worn by a large majority of women, especially those of child-bearing age. An alternative is a waist-length shirt, ihempe , with short sleeves, worn by older married women, especially old women. This may be plain, though some are decorated at the ends of the sleeves, on the yoke and at the bottom with rows of black, green and/or orange braid. The sides are usually slit for a few inches up from the waist, to provide greater freedom of movement. It is not 69 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 tucked into the skirt. For everyday wear a sleeveless vest or cardigan is often substituted for both these items, except by nursing mothers. An article of clothing used by all women, especially newly-married ones, is ibinqelo , a length of bhayi , left plain or dyed red in an ochre solution, and wrapped tightly about the chest at the level of the breasts. One end is tucked in at the top to hold it in position. It hangs down to the level of the thighs. This is worn at all times, as part of the intlonipho observances (see next paragraph). A large rectangle of the same fabric, left plain or ochred red, is worn as a cloak ( ibhayi ) particularly for use when attending a ritual or feast but also at home against the cold. Women’s cloaks are distinctly larger than those of adolescent girls. Though few Red women (mainly older ones) attend funerals, some wear a black cotton or rayon shawl ( ilema ) fringed at the edges, instead of the bhayi cloak The headdress of married women is also connected with intlonipho observances. This is a complex of speech patterns and modes of behaviour incumbent particularly on young married women vis-a-vis their husbands’ male agnates. When a new bride goes to her husband’s home- stead, her entire head is tied in a rectangle of black cotton cloth (iqhiya yelaphu), leaving only the mouth, nose and eyes open. Once her entourage arrives there, her whole face is covered as well, until she and her attendants have reached the privacy of the hut assigned to them. During the early months of her marriage she wears this cloth much like a nun’s wimple but gradually reveals more and more of her face, until the black cloth is used to cover only the top of the head. It must still, however, be worn low over the eyes. There is no exact set period of time which must elapse before this cloth is exchanged for a melton headcloth ( iqhiya yesitafu ); it depends very much on the whim of the husband’s mother. Usually, a wife is allowed to adopt the melton headcloth when she has borne a child. Generally this is black for everyday wear and blue, brown or green for attendance at rituals and feasts but there is no rule. It is square and much larger than the headcloth of a new bride ( umtshakazi ). Most women’s headcloths are decorated with machine-sewn patterns of white cotton, done by local seamstresses who charge up to R1 for the decoration. Especially for feasts, women often wear two of these large melton cloths, one inside the other, to add bulk to their headdresses. Brightly-coloured cotton cloths are inserted into the folds as decoration. The manner of tying the headcloth used in Tshabo is as follows: the melton rectangle is folded into two intersecting triangles, the apices of which lie next to each other. With the head bent forward, the base of the cloth is placed at the base of the skull and the points allowed to fall forward over the front of the head. The corners are then brought round to the forehead, crossed over and tucked into the wide band formed by the fold. The cloth now presents an appearance similar to a toque. At home a woman seldom wears shoes. Flat-heeled brown or black (more often the latter) lace-up shoes are sometimes worn to rituals and feasts, but usually only when a Red woman goes to town. Plastic sandals and shoes are also worn. DIVINERS While diviners among the Southern Nguni as a whole appear to be mainly female, the number of male and female diviners in Tshabo between 1967 and 1970 was very nearly equal. Though they undergo some years of instruction by a qualified diviner, after the disposition towards divinership has been diagnosed, they are not full-time specialists when qualified. Except when attending gatherings of diviners, they dress in the fashion of their respective sexes. Nevertheless a diviner is always identifiable by the wearing of some article made of white beads (these will be described under the heading of Ornament) because white is the colour identified with diviners. The special dress worn by diviners at diviners’ dances ( iintlombe yamagqira) is distinctive. Male diviners wear a mid-calf-length “skirt” (unofali) of white ibhayi , decorated with rows and geometric motifs of thin black braid. This braid is not applied as 70 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE liberally as on women’s skirts. The unofali is not sewn up, so the wearer merely winds the rect- angle of cloth about the waist and tucks one end in. Over the unofali , a kilt of long strips of animal skins ( imithika ) with the hair on is worn specifically when a diviner is presiding at a ritual. The strips of skin in this kilt are not sewn together, except at the waistband. They may be of goatskin or cowhide, but usually include strips of the skins of wild animals (pi. izilo ) but not of the wild animal (sing, isilo) to which the diviner stands in a special relationship (Bigalke, 1969, 100-2). A vest may be worn on the torso. Many diviners wear a hat made of animal skin ( isidlokolo ) somewhat like a mitre in shape. The kinds of skin most commonly used are those of the baboon, the vervet monkey and the black-backed jackal, but here also, the prohibition against the diviner’s special animal applies. Female diviners wear a white skirt ( umbhaco ) similar in shape and decoration to the red- ochred ones of ordinary married women. They also wear the breast cover ( incebeta ) in white or, less frequently, a blouse ( ihenipe ) of ibhayi. The ibinqelo is also worn by female diviners. A white bhayi cloak, with or without black braid decorations, is worn for going out. A notable difference between women diviners and ordinary married women is that the former are not required by custom to keep their heads covered at all times. While women diviners are subject to the normal rule at their husbands’ homes, they wear no head covering while dealing with people who consult them professionally or when they join the dances of other diviners. Skin hats ( izidlokolo ) or headcloths ( amaqhiya ) may be worn. PERSONAL DECORATION The Ndiambe make frequent use of some types of cosmetics and ornaments. Informants expressed the reason for this as the wish to appear attractive and well-dressed, particularly on public occasions. Few bead ornaments are worn in everyday life about the homestead. How- ever, all age groups take great pains with personal toilet, dress and ornamentation for rituals and festive occasions. Informants spoke of the personal satisfaction derived from knowing that they were better dressed and groomed than others. It was also felt that a well-dressed person has dignity ( unesidima ), is respected ( uyahlonipheka ), loveable ( uyathandeka ) and an object of attention ( uyakhangeleka ). One girl believed that a well turned-out person could compensate for unfortunate physical attributes: ‘'Many ugly abafana and boys have a lot of girl friends because they dress nicely.” A person who is well-dressed and elaborately ornamented may well use this display as a means of showing off wealth and status, though conspicuous consumption in terms of personal possessions is rarely practised among the Ndiambe. Elaborate ornamentation, especially in a youth ( inkhwenkhwe ) or young man ( umfana ), is almost certainly a means of displaying the results of his success in attracting girls. As Schoeman ( African Studies , 27, (2), 1968, p. 59) has observed for one area of Zululand, it is males from this age of approximately 14 to 40-5 years who are most elaborately beaded. Youths do not, as a rule, begin to wear beadwork ornaments until they strike up a relationship with one or more girls who then present them with bead ornaments. Considerable prestige is attached to the ability to attract girls, so much so that unsuccessful youths may get their sisters or mother to make a few ornaments for them to wear in public. As there is very little privacy in peasant life, the subterfuge is easily discovered and the perpetrator teased about it. Married men have extra-marital relationships and may be given beadwork by their girls. Generally, a girl begins to make beadwork articles ( ukuhlohla ) in the late winter, in order to be able to give her boy friend a present ( udisemba ; cf. “December”) at Christmas (this is the practice amongst ancestor-worshipping Red people, not the Christians). The recipient makes a return present to the donor, who may, however, have to wait until the following December for it. If it is a large present she has given, for example, a complete set of beadwork ornaments (as 71 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 described below), she may receive as much as R24 made up of cash and gifts in kind. One youth, whose set of beadwork and ornament included 49 articles, reciprocated as follows to the girl who had given him most of it: to the girl: 1 table cloth, 1 white towel (for tying about the waist), 3 coloured cloths (for the head), 1 towel for use as a headcloth, a pair of sun-glasses, 8 silver-coloured metal bangles, 1 mirror, 3 combs, 5 strings of large glass beads ( amaso ) in assorted colours, a pair of plastic shoes, a tin of face powder, a mouth organ, 4 large (4 oz.) packets of tobacco, 3 small (1 oz.) packets of tobacco, 30c worth each of sugar, flour and sweets and 2^-c worth of yeast. To her mother he gave 2 coloured cotton cloths, a small mirror, a pair of earrings and 3 pairs of anklets ( amowotshi — see below). Decoration and ornament are also a means of conveying socially significant facts about the wearer’s social status and about particular states, during which appropriate forms of beha- viour must be adopted towards people in these states. In this context cosmetic decoration may constitute a code, as indicated by de Lange (Ann. Cape Prov. Mus ., 3, 1963, 85-95), whose data apply equally to the Ndlambe. In another respect, the wearing of bead and other ornament, important information is conveyed. With certain exceptions, each kind of ornament is appropriate to only one particular age-grade. The wearing of an inappropriate ornament by a member of a certain age-grade is discouraged by ridicule or even by forcible removal. Old women have been known to remove iwotshi , a set of graduated brass bangles, from the arm of a woman considered too young to wear it. Senior youths give similar treatment to their juniors who wear ornaments unsuited to their age-grade. Ornament is frequently believed to have a protective function. The ubulunga necklet (illustration: Bigalke, Ann. Cape Prov. Mus ., 6, 1, 1966, Plate 1), with or without the addition of beads, is made of the tail hairs of a beast specially kept in the herd for this purpose ( inkomo yokuxwitha; ukuxwitha — to pluck). Whenever a person, regardless of sex or age, is affected by ill-health, particularly if it is thought to be the result of ancestral anger, an ubulunga necklet is made for that person to wear. It must be worn until it falls to pieces, then buried in the cattle kraal. Sometimes the necklet is encased in a stiip of cloth, which makes it easier to wear and less obvious, especially if the wearer works in town. Babies are given waistbands of string bearing an even number of cowrie shells ( Cypraea moneta and C. annulus). These are believed to assist teething even though children do not chew on them. Waistbands of amatantyisi (Job’s tears, Coix laeryma), either alone, or combined with beads, are used for the same purpose. Also connected with suckling children are amakhubalo and isixhoxho necklets (see below) worn by nursing mothers. A type of necklet often worn by youths, their girls, abafana and their girls, the groups of people most concerned to attract and keep lovers, is ibotile (see below), a small glass bottle, often of the type used for stock vaccine, covered with beads and attached to a string of beads for wearing around the neck. Though often claimed to be merely for decoration, these are widely believed to contain medicines obtained from herbalists. Their purpose, when applied to the wearer’s face or body, is to attract the affections of someone of the opposite sex. Diviners and herbalists often wear small horns ( imiphondo ), containing medicines, on a necklet of white beads, the colour associated with diviners. These are also intended as protection against medicines used by their enemies (diviners are always thought to have many enemies) and “to advertise their profession” as one cynical Ndlambe observed. The questions of preferred styles and changing fashions require some comment. There is no doubt that an increasing use of western clothing, and its use together with items of tradi- tional dress, has had a particularly marked effect on dress in Tshabo during the past decade. Professor P. Mayer (pers. comm.), who conducted research there in the middle and later fifties, has remarked on the changes in respect of clothing. Mr E. L. Xotyeni (pers. comm.), who has worked as research assistant to both Professor Mayer and the present writer, remembers that many families who are now School were then Red, and that much more traditional clothing 72 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE was seen at that time. In the fifties, youths wore penis-sheaths, not beaded shorts, and their beaded ornaments did not include the wide range of articles based on shop-bought leather straps (see below). Women’s clothing did not include the jerseys and German-print aprons so widely worn nowadays. During the course of collecting specimens for East London M useum, it became clear that certain colour conventions in beadwork were or are characteristically worn by people of certain age-groups. Black and white, or black, white and saxe blue or turquoise are the bead colours worn by senior men and women at the present time. It was possible to obtain a little information about the relative age of some of the specimens. From this, the type of bead used and the evi- dence of wear, a tentative conclusion appears possible, namely that for this century, at least, black and white or black, white and saxe blue are the “traditional” colours of the beadwork of older people. The exception is the use of turquoise beads for the isidanga necklet (see below). One specimen of isidanga (5101), dating from about 1906, is made of turquoise beads somewhat larger than the kind used at present. It is not known whether the use of this type of bead arose from individual preference or was dictated by the limited availability of beads in trading stores. There are two distinct ranges of colour combinations in youths’ beads. At present, the predominant colours are white, turquoise, orange, with the limited use, at times, of leaf green and transparent dark blue. A colour combination no longer fashionable, according to infor- mants, is white and light turquoise or white and light saxe blue. At one stage of the fieldwork it was noticeable that beadwork in the latter combinations was being freely offered for sale but very little in white, turquoise and orange combination. Fashion appears to operate also in the types of article worn. Among youths the harness- type of torso ornament ( amapasi ) entirely of beads has fallen out of favour and been replaced by the type made up of leather bands. Ukotso of both types (pp. 75 and 79) are seldom worn nowadays. Icanci (p. 78) has, with the addition of a fringe and a change of colour, become ithawuza (p. 78) and is being replaced by it. TYPES OF ORNAMENT AND ACCESSORY All the types of ornament and accessory seen or collected in Tshabo location will be listed below. Inclusion of an article does not imply that it is a standard item in every individual’s outfit. The commonness or otherwise of each item will be noted, as well as its use for any special occasion or reason. Beads are of glass (made in Italy or Czechoslovakia) unless other- wise specified. They are strung on sinew from cattle or on cotton, sometimes on a combination of the two. Cotton is much more widely used than sinew. YOUTHS Armbands: Urwashu (pi. orwashu ) (5532, 5533), two leather straps, each 2,4 cm. wide, covered with rows of white, turquoise, orange and navy beads. The straps are joined by being sewn on one long side to a rectangle of beadwork of equal length (24,5 cm.), backed with bhayi. The beadwork panel is of white beads, with geometric motifs, based on the chevron, in transparent dark blue, turquoise and orange beads. Two leather straps (95 cm. long) are attached in the middle of each of the short sides of the beadwork panel. These are bordered all round with rows of white, transparent dark blue and turquoise-blue beads. This object is attached to the upper arm by twisting the threads attached to the central straps around small mother-of-pearl buttons attached to the ends of these straps. A pair is worn. This is an unusual type of article, a new fashion. The central beadwork panel and straps are usually found without the long pendant straps. These armbands are worn to youths’ dances ( imitshotsho ). 73 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 Umxhaka (pi. imixhaka). This is a type of armband consisting of two or more leather straps. The straps may be beaded or be attached to each other by means of a rectangle of beadwork. Two types were collected: (a) (5530, 5531) a 22,5 x9 cm. panel of white beads, with geometric motifs in turquoise and transparent navy beads based on the chevron motif, is attached on each long side to a 25,5 X 2,6 cm. leather strap. Each strap is decorated with rows of white, orange, turquoise and transparent dark blue beads. At each end of both straps are two mother- of-pearl buttons. The buttons at one end of each of the straps have long pieces of white cotton attached for winding around the buttons at the other end, to attach the object to the upper arm. A pair is worn, one on each arm. This type of article is widely worn. It is seen at youths’ dances {Unit shot sho), also when the youths of various age groups march from village to village, and at stick-fights. ( b ) (5528, 5529) a leather strap 25x2,7 cm., covered lengthwise by rows of white, turquoise and transparent dark blue beads. At each end of the strap are two mother-of-pearl buttons. Long white cotton threads are attached to the buttons at one end of the strap, for winding around the buttons on the other end, to attach the object to the arm just above the elbow. A pair is worn on the same occasions as those described above. Commonly found. Bags: (a) ingxowa (pi. iingxowa ) (5508) a rectangular (51 x30,5 cm.) bag of bhayi , with drawstring at mouth, and a handle of the same fabric. Each end of the drawstring bears a cerise woollen pompom. The bag is decorated on both sides with horizontal rows of thin black braid (ib/eyiti), three adjacent rows of which are sewn on zigzag. Also on the outside of the handle are rows of black braid applied along the length. The top of the bag and the handle are edged with a row of white beads. At the bottom of the bag is a fringed strip of beadwork made up of alternating rows of white, transparent blue, turquoise and orange beads. The short fringe is of turquoise beads, edged with larger white beads ( amaso amhlophe) and a few small white and navy beads. To one side of the bag are pinned two small white towels and one small handkerchief. To each of the towels two circular discs of beadwork (white, turquoise and transparent blue) are attached. Two of the four discs have a few orange beads in addition. The discs are done around a large curtain ring and attached to the towel by means of a large safety pin. To the handker- chief is attached a large silver-coloured safety pin with a short, rectangular strip of beadwork (white background, with chevron motifs in transparent dark blue and orange; lower edge, same colours and turquoise, ending in a cerise woollen fringe). On the other side of the bag are two folded handkerchiefs and five beaded safety pins with beaded panels, one much smaller than the other four. This kind of bag is common, though smaller examples are more usual. Bags are carried by the strap or with the strap over the arm. Another example of cloth bag for a youth is the following one ( ingxowa yamakhwenkhwe , 5211), 115 x13 cm., made of bhayi and de- corated on the lower part of each side with seven rows of thin black braid. Just below the mouth is a drawstring, to the ends of which are attached cerise and blue woollen tassels. It has a handle of three strings of beads, two white, one turquoise blue, strung on sinew. This is a common type of bag for keeping tobacco. ( b ) ingxowa yenyhwagi (5507) a bag, approximately 38 cm. long, one side being a whole genet ( inyhwagi ) skin, the other a blue duiker (iphuthi) skin, both with the hair on and sewn together with leather thongs. A head skin of the genet forms a flap over the mouth of the bag. A shop- bought leather strap is sewn on to the top to form a handle. This is beaded at the edges with rows of white, transparent dark blue and turquoise beads and decorated at intervals with single mother-of-pearl buttons. The genet skin side is decorated towards the bottom with a beadwork panel (24x10,8 cm.) consisting of a background of white beads, the geometric motifs in dark blue, turquoise and orange beads. Above this is another beadwork panel (16,3x7,5 cm.) 74 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE in the same style and colours, with the addition of transparent green beads. On the duiker- skin side is a beadwork panel (23x9 cm.) in the same style and colour as above. Two genet tails and the skin of the hind legs of the duiker and genet hang down from the bottom of the bag. The style of the bag is usual, though the example described is uncommonly large. Earrings: aniacici , small silver-coloured metal rings, hinged in the centre and with a clip for fastening. These are bought from shops and are very widely worn. Harnesses : ukotso (5216), a kind of waistcoat or corset, entirely of beads. Weight 2,55 kg. (5 lb. 9J oz.). The main part of this article consists of a large rectangle (95 X 46 cm.) made up of four panels of closely-strung beads arranged in wide stripes of white alternating with narrow stripes of turquoise and dark blue. The four panels are joined, by means of numerous short strings of white beads, along the 46 cm. side, to form a large rectangle. At each end of the large rectangle is a strip of goat or sheep leather sewn onto the beadwork with sinew and fitted with hooks and eyes for attaching the “corset” to the torso. This fastening is worn at the front. Each of the two leather strips is decorated with a row of small white buttons, with a few beads decora- ting the sinew used for sewing on the buttons. As this “corset” is not held on securely enough by means of the fastening at the front, a wide band of beadwork in the form of braces is worn over the shoulders ; this is affixed to the upper side of the large rectangle in four places by means of 12 bead loops attached to the ends of the braces and 12 small buttons attached to the large rectangle. This article is worn alone with no other torso ornament but necklaces are worn with it. It is not unusual wear for dances but is scarce due to the large quantity of beads used in its making. Because of this, ukotso seems to be less fashionable than other torso ornaments and is being replaced by various kinds of decorated leather bands, as described below. Another variety of ukotso is a wide cummerbund of beads worn as a waistband (see below). Amapasi is an ornament made up entirely of strings of beads or of leather straps decorated with, and joined to each other by means of, beads. The all-bead type (a), is seldom seen worn nowadays, being replaced by the leather strap type ( b ). (a) (5210), a torso ornament resembling a harness. It consists of four single strings (each 1 1 1 cm. long) of beads, two white, two blue. From each shoulder, one of the white and one of the blue strings hangs down, the bottom of the loop resting against the side. At this point an 8 cm.- diameter disc of beadwork, built around a large curtain ring, is sewn onto the two strands of beads, joining them together. The beadwork of the discs consists of concentric circles of turquoise, black and white beads, the centre being of white beads with three triangular motifs in dark colours. The loops hanging on either side of the body would swing free but for the fact that they are joined together at pectoral muscle level, front and back, by means of two separate beadwork discs, similar to those described above, except for the addition of two rows of orange beads. This is a form relatively rarely worn by youths but common as a decoration for abafana (see below). It resembles the amaselwa in form and construction (described below). (b) (5516), a harness-like ornament consisting of two wide bands (9,5 cm.), each 133 cm. long, each of which is hung over one shoulder, reaching to the buttocks at the back and to the thigh in front. Each of these wide bands is made up of two 2,5 cm. wide shop-bought leather straps joined to each other along the length by means of rows of white, transparent dark blue and turquoise beads sewn on with cotton and bordered along the outer edge with rows of the same coloured beads. The straps are ornamented at intervals with groups of two small white buttons. At the level of the pectoral muscles, the bands are joined to each other, in front and at the 75 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 back, by means of a panel of beadwork (14,5 X 13,2 cm.), sewn on with cotton. The beadwork on these panels has a background of white beads, with continuous chevron motifs and diamond motifs in transparent dark blue, turquoise and orange. The panels are bordered above and below with rows of transparent dark blue beads and turquoise beads. The front and back of each wide band is joined at the side, at waist-level, by means of three bead panels of approxi- mately the same size, motif and colouring as those described above. This type of object is becoming increasingly common, either in the form described above or with more or fewer bead panels. It appears to be a lighter and less expensive modification of ukotso , described above. It is worn to dances. amaselwa (5514), a harness-type ornament, similar in general shape and function to amapasi. Instead of the leather straps joined to form bands, this article consists of four beaded cords (each 151 cm. long), two being worn on either side of the neck. As in amapasi , they hang down in front of and behind the body. On each side of the neck is one white-beaded and one tur- quoise-beaded cord, made by winding a continuous string of beads around a length of thick string. The two ends of each cord, where they rest on the hips, are joined onto a small, bead- covered calabash ( iselwa ) from which the whole object takes its name. The colouring of the beads on the calabash is a background of white, with horizontal stripes of turquoise and transparent dark blue, with motifs in the same colours. The beadwork does not cover the entire calabash but ends at the widest part of the bulb, with a fringe of the two shades of blue beads and a few larger milky-white beads ( a/naso ). At the level of the pectoral muscles, the bead-covered cords are joined together by means of a panel (15,4 x 10 cm.) of white bead- work with chevron and diamond motifs in the two shades of blue, together with orange. This type of object is fairly common, being worn to dances as an alternative to amapasi. ineshinal , (5517) a kind of sash. Three lengths (169 cm.) of shop-bought leather strap are joined side-by-side by means of rows of white, transparent dark blue and turquoise beads sewn on with cotton and also edged on the outside with rows of the same coloured beads. On the upper, polished yellow side, the leather straps are decorated at regular intervals with mother-of-pearl buttons. The ends of each of the three straps are sewn together, the whole forming a continuous, wide loop. Attached to this is another loop, made of two narrow leather straps attached to each other with beads in the same way as the straps in the wide band. The ends of the straps of this small loop are buckled together. It can move freely along the length of the large loop. When the large loop is being worn as a sash, the small loop rests at the bottom of the large one, on one hip. To the small loop are attached two nylon chiffon squares, one cerise, the other white with pink and blue designs. Only one example of this type of ornament was seen in Tshabo, though another example was seen in an Ndlambe area not far away. Headdress : ingqaza (5524), a narrow band (49,2x2,5 cm.) of white beads. This band is made by joining a large number of single, short strands of white beads along the tops and bottoms to form the edges of the band. On these edges transparent dark blue beads are used alternately with the white. This band is worn on the forehead. Where it rests at the base of the skull are four small white buttons. Onto these are buttoned the four loops of the other part of this object. It consists of eight single strands of white beads, each 162 cm. long. Four of the strands are passed under each arm, brought round in front of the body and tied loosely at chest level, each loose end being thrown over a shoulder. This is an unusual type of ornament. inkedama (5527), a band (61 x2,5 cm.) of beads, consisting of four small rectangles of closely- sewn beads, horizontally joined to each other by eight single strands of beads. One end of the 76 BIGALKE: DRESS AND ORNAMENTATION AMONG THENDLAMBE band has four mother-of-pearl buttons, the other four loops for attaching to the buttons. The small rectangles are of white beads, the single strings of turquoise beads and of white beads mixed with navy beads. These strings are arranged alternately. The band is worn at forehead level, being put on after the ingqaza is already in place. It is strung on sinew. This type of article, sometimes wider than the one described, is part of every senior youth’s outfit. iyokoza (5523) a wide (11 cm.) bead panel (worn above the forehead) joined to a narrower (3,9 cm.) bead panel (worn at the back of the head). The entire band has a circumference of 53,5 cm. Both parts have a back-ground of white beads, bearing chevron motifs in transparent dark blue, turquoise and orange in the front, with red substituted for orange at the back. The narrower band has a 38 cm. long fringe of 91 single strands of white beads terminating in a few transparent dark blue and turquoise beads, with one larger, milky white bead (iliso) at the end. This fringe is worn hanging down at the back of the head. A common object. iqhiya, (5075) a triangular headcloth of emerald green cotton fabric. It actually consists of a square of cloth, folded diagonally, with a 12 cm. band folded up from the base of the triangle and sewn into position with white cotton. This band is decorated with machine-sewn horizontal lines, diamonds and leaf shapes. The edges of the band and the entire edge of the headcloth are decorated with single rows of white and orange beads. To tie the headcloth, the undecorated (inner) side of the band is placed just above the base of the skull, the pointed ends of the base of the triangle brought round to the front of the head, crossed over and then tucked in. The point of the apex of the triangle is also tucked in. Knife sheaths: iskeyi, (5509) a 74 cm. long x 5,1 cm. wide shop-bought leather belt (with buckle removed), whose upper surface has a smooth, orange-coloured finish. For 22 cm. of its length, the belt is doubled over, this portion being sewn with loops of green plastic-covered wire, to form a sheath with a long, decorative tail. This embellished with chromed metal studs, white- painted metal eyes, and three small stud-like reflectors such as are used on bicycles. Approxi- mately 26 cm. at the lower end of the pendant is beaded at the edge of the belt with rows of turquoise, transparent dark blue and white beads. In this area, there are three pairs of small bead triangles, placed apex to apex, the same colours being used, with the addition of orange, transparent green and transparent yellow beads and with one mother-of-pearl button to each pair of triangles where the apices meet. A folded white handkerchief decorated with red flowers and green leaves is attached to the reflector nearest the lower end of the belt. Variations on this type are common. Leggings: umkhwelo ( pl.imikhwelo ) (5499, 5500) a shop-bought leather strap (46,5x1,4 cm.) with buckle. The glossy, orange upper surface of the strap is decorated in the centre with continuous Y-pattern of small white beads, while the sides are decorated with rows of white and turquoise beads. This strap is buckled onto the calf just below the knee. From each of the three points on the length of the strap hang two coils, one of white, the other of turquoise beads. These three groups of coils hang down to about two inches above the ankle. At mid-calf and again at the ends of the three groups of coils, these coils are joined by means of two other sets of coils. These lie horizontally on the calf, forming a ring at mid-calf and again two inches above the ankle. These horizontal ring-coils are in white and turquoise beads, the lower of the two having, in addition, a coil of vermilion beads. A pair of these ornaments is worn, one to each leg. Unusual. 77 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 Necklets: umkinxo (5129), a 29,8x4,6 cm. beadwork strip, with a 3 cm. wide strip of pigskin sewn on with sinew at each end. These leather strips are fitted with six hooks and eyes for securing the article to the back of the neck. The bead strip has a background of small opaque white beads, with a cross motif in turquoise and transparent dark blue beads in the centre. At each end it is flanked by two vertically arranged rectangles in transparent dark blue and orange beads, and a vertical bar of turquoise beads edged with rows of transparent dark blue beads. A common item, it is the first one put on when dressing up for dances. icanci , (5077) 38 X 9,6 cm., an openwork, lacy-patterned beadwork strip, shaped to fit the neck. It is made of horizontal double rows of black and dark saxe-blue beads, alternating with single short strings of white beads placed vertically. The lower edge consists of intertwined loops of white beads, forming a shallow scalloped edge. The upper edge is a single row of white beads. It is attached to the neck by means of a mother-of-pearl button and a bead loop. This is a common item, in a colour combination and pattern said to be exclusive to youths. Amacanci (pi.) are often made in a series of two or three, each progressively longer and wider than the last, and worn one on top of the other, the largest at the bottom. This is a medium-sized one. They are put on after the umkinxo and worn to dances. ithawuza, (5521) 63,5x9 cm., a beadwork strip superficially similar in appearance to icanci and also shaped to fit the neck. The lacy pattern is based on an “X” of beads, all four legs of which intersect with some part of another “X” conveying an impression of continually intersecting motifs. The beads are arranged in alternate rows of white, transparent dark blue, turquoise, transparent dark blue, white and so on. This order is interrupted in the centre, where there are two rows of orange beads. The lower edge is decorated with a 4 cm. fringe of cerise wool. The necklet fastens by means of a mother-of-pearl button and bead loop. The whole is strung on cotton. An alternative pattern of stringing the beads for this necklet is that described for icanci. Aniathawuza (pi.) are most often fringed with cerise wool but an emerald- green fringe is sometimes seen. Like amacanci , amathawuza are also made in sets and worn to dances with the largest underneath and the others above, giving the impression of a number of woollen fringes attached to one bead necklet. ithumbu , (5525) 39x2,4 cm., a close-fitting strip of beadwork, shaped similarly to icanci and ithawuza. It differs from them in texture, the main pattern being that of two horizontal, tightly intersecting bead “Y’s” in white, placed one above each other. Above, below and between these two patterns is a two-row strip of transparent dark blue beads strung horizon- tally. The upper edge of the necklet is a single row of white beads. The lower edge is a row of tightly-intersecting loops, giving the effect of shallow scallops. In the centre of the lower edge of the necklet two mother-of-pearl buttons have been sewn; to each loop is attached, and from it hang, two 19,5 cm. single strings of white beads beginning and ending with a few transparent dark blue and turquoise, with a single larger, milky white bead (Hi so) at the end. The necklet is strung on cotton. This is a common type of necklet for dances, usually found in larger sizes than the example described. iqhina, (5526) a narrow (38x1,3 cm.) strip of beadwork in transparent yellow, edged with white beads, to which is attached a mother-of-pearl button and bead loop. To this strip is affixed a 33x6,8 cm., tie-shaped piece of close-textured beadwork in transparent yellow. This is edged with two rows of white beads and decorated in the centre with two “X” motifs in white and transparent green, an inverted “V” in white and transparent dark blue and a hori- zontal bar in transparent dark blue and white beads. Not unusual, but seldom widely worn, it is put on over ithumbu or amathawuza. 78 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE usaliwe yamakhwenkhwe, (5518) two (46 x 1,3 cm.) shop-bought straps with buckles are placed horizontally, one above the other, and joined by means of rows of white, transparent dark blue and turquoise beads sewn on. The upper edge of the top strap is similarly edged. Both straps are decorated at intervals with a line of widely-spaced mother-of-pearl buttons. From the centre of the lower edge of the bottom strap, hangs a 9 cm. wide panel of beadwork 30 cm. long. This is of closely sewn white beads, with chevron-based motifs of transparent dark blue and turquoise beads and diamond motifs of transparent dark blue and orange. The lower edge of this panel extends into a 64 cm. fringe of 31 single strands of beads. Flanking this central panel and fringe from top to bottom is a full-length (96 cm.) fringe of beads. Like the panel, this is mainly white, turning at the end of the panel into a wide band of orange beads, then becoming white again. The end of the fringe has an alternation of transparent dark blue, turquoise, white and orange beads, ending with a larger translucent milky bead ( iliso ), held on by two or three of the smaller beads. The whole is strung on sinew. This item may be worn alone or with other necklets. If the latter, it is put on first. Other examples of the usaliwe differ according to the width of the central panel, the number and length of strands composing the fringe, the motifs (although all are based on the chevron and its combinations) and colour. Some have coral-coloured beads (amasomi) in addition to, or instead of, orange. Some may be strung on cotton. The weight of the specimen described above is 0,9 kg. (2 lb.). It is worn to dances. Waistband: ukotso, approximately 66x20 cm., a band of beadwork, closing by means of loops and mother-of-pearl buttons worn at the front. It resembles a wide cummerbund. When worn the beads appear as vertical stripes, pink and saxe-blue alternating, with a few stripes of large, milky white beads ( amaso ). This item is scarce but not unusual. Wrist ornament: uphondo, (5194) two coloured plastic bangles, joined by means of a partial cylinder of bead- work sewn onto holes drilled along one edge of each bangle. The beadwork has a background of white beads, with chevron motifs in transparent dark blue, turquoise and orange beads. The sides of the bangles not attached to the partial cylinder of beadwork are edged with rows of white, transparent dark blue and turquoise beads. It is strung on cotton. Total length, approximately 14 cm. Unusual. ABAKHWETHA Headdress : umqhele , a wreath-like headdress, made of pieces of skin of animals or feathers of birds killed when the abakhwetha hunt to pass the time during their seclusion. East London Museum collections include some non-Ndlambe specimens, one each made of hare skin, red-winged starling wings and Cape canary feathers. Ones made of hare skin and of Egyptian goose feathers and wings were seen in use in Tshabo. These articles are usually burned at the con- clusion of seclusion. AMARWALA (newly circumcised young men) Anklets : amangqashela , (5097) a pair of anklets each consisting of a 34 cm. string of beads. This is made up of large milky-white beads (amaso), each alternating with one small black, two small dark saxe-blue and one small black bead. Each end of the string is bifurcated, the one part being a loop of beads or a mother-of-pearl button (depending on the side), the other a very short double string of beads ending in a small tassel of cerise wool. These anklets are worn daily from the time the irwa/a comes out of seclusion until he goes to work in town. 79 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 Headdress : ingxashe wamarwala , (5095) a 54,4 x 1,5 cm. band of beads, formed of three horizontal rows of large milky-white beads ( amaso ) placed one on top of the other. The upper and lower edge of the band consists of a row of small turquoise beads. Within the row of white beads are four insets of three large black beads {amaso) each. Along the lower edge of the band a double length of thin, brass chain is attached to two points 15,5 cm. apart, forming an arc when the band is in place on the head. Another, short length of double chain is attached to a point midway between the points to which the other chain is attached. This object is worn as a forehead band, the bead band being arranged so that it lies parallel with the eyebrows, about 2,5 cm. above them. It is worn daily during the period between emergence from seclusion and departure for work. iqhiya , (5098) an 84x46 cm. rectangle of black cotton cloth, folded into a triangle. The base of the triangle is laid at the base of the skull; the points at the base of the triangle are brought round to the forehead, where the points are passed through a curtain ring (5099) and then tucked into the folds of cloth. This cloth is tied on only after the ingxashe (described above) is in place, and worn daily, like the other two articles for amarwala. Necklet: amaso amnyama wamarwala , (5096) a 45 cm, necklet of two single strings of large black beads {amaso), with a few large milky white beads {amaso) at the ends and in the middle. It closes by means of a bead loop and button. From each end a narrow, 72 cm. thong of leather hangs down. As the loop and button rest on the back of the neck, the thongs hang down the back. It is worn daily, like the other articles for amarwala. ABAFANA (young men) Anklets: isitsaba , (pi. izitsaba), (5070/1, 5072/3, 5191, 5458/9) approximately 20x7,5 — 8 cm., a rect- angular panel of beadwork, always with a strip of sheep- or goatskin sewn onto each narrow end. These strips may be perforated to allow for the insertion of a thin leather thong on the one side, which is passed through the holes on the other side to tie the anklet in position. More frequent is the use of small hooks and eyes, with mother-of-pearl buttons sewn at intervals along the leather strip as decoration. The beadwork is invariably done in a combination of white and a dark colour, arranged in geometric patterns. In older specimens, the combination is black and white, while more recent specimens have navy substituted for black, or the black/white combination with the addition of turquoise, either in the form of vertical bars, or as edging to the upper and lower edges. Izitsaba are worn by abafana specifically on occasions when full traditional dress is required, that is, at weddings, when abafana drive the bridewealth cattle to a future bride’s home and when they are asked to accompany a diviner to the river for the performance of the umhlwayelelo ceremony. Normally, izitsaba are more fittingly worn by senior men {amadoda). ingqashela (pi. amangqashela), (5089), a single string of beads, attached to the ankle by means of a bead loop and mother-of-pearl button, or two or three single strings, similarly attached. An anklet may be of one colour (usually the case when worn by abafana) or of mixed strings of beads. The colours generally used are white, pink, saxe-blue, turquoise, coral, orange and dark emerald green. Red is not worn with orange, nor saxe-blue and turquoise combined. Abafana wear these anklets only when in traditional dress for dances at weekends. 80 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE Armbands: umxhaka (pi. imixhaka). Three varieties of armband are called by the same name. Each of the types may be worn on either or both arms, above the elbow. The different types are not worn together. (a) ( 5157 — 9) approximately 9 cm. diameter, a length of thick string, fastened at the ends to form a circle, then covered by winding over it a continuous string of small beads, either plain white, dusty pink or dark saxe-blue or one of these colours with the addition of a few stripes of one or more of the remaining colours already mentioned. If the single string of beads is to be of one colour only, the technique is to wind it around the core in a continuous coil. When contrasting stripes are to be added, each ring of beads is sewn onto the core separately but joined to a continuous thread, to prevent individual rings moving. This is the variety of armband probably most frequently encountered. It is worn by abafana on all occasions when traditional dress is worn. (. b ) (5062/3) approximately 10 — 12 cm. diameter, a variation on the variety described above. The difference is that a thick rubber ring or a rolled piece of cloth is used as a core, resulting in a much thicker armband. The technique for applying the single string of beads is the same as in (a). This item is less usual than the first variety but worn on the same occasions. (c) (5090, 5094), approximately 27x2 cm. a strip of shop-bought leather strap, sometimes with the two ends sewn together to form a circle. The outer surface of the strap is beaded with rows of white, dark saxe-blue and black beads, or white, turquoise and navy beads. The upper and lower rows of beads are applied along the length of the strap, while the central space is occupied by short rows of beads at right angles to the rest. When the ends of the strap are not sewn together, the wearer secures the armband by means of cotton attached to the ends. This is as common as the first variety of umxhaka, worn on the same occasions. Bags: ingxowa , similar in all respects to the cloth ingxowa of youths, except that no orange beads are used for decoration, nor are towels or handkerchiefs applied to it. Bead discs and beaded safety pins are, however, found. A common item, this is carried with the strap over the arm, or often with the strap attached to a stick carried over one shoulder. It is used when traditional dress is worn. Dance Staffs: isitafu, approximately 75 — 80 cm. long, a long, thin wooden stick, or a wider length of plank, with sections cut out in a decorative pattern and the wood covered with single strings of beads wrapped about it. The beads are of the small variety, in white, turquoise, transparent dark blue and/or orange. This object is frequently found in use at iintlombe , the dances of abafana. Though it has been given to one specific man by a girl friend, it is often borrowed from the owner by other abafana , who dance out from among the circle or row of dancers, with the staff held in one hand high above the head. Earrings : amacici (sing, icici), plain silver-coloured rings bought in shops and inserted through holes pierced in the ears. Many abafana wear these earrings constantly. Once put on, they are not removed. Harnesses : amapasi, (5112) a torso ornament, similar in form to the amapasi of youths, as described on pp. 75 — 76. Instead of the bead discs of the youths’ version, that of the abafana has two small (5 x 3,9 cm.) beadwork rectangles, fringed with larger beads, at chest level in front and on the 81 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 back, while two larger (9,5 X 6 cm.) rectangles are attached to the side strings and rest on the hips. Two of the four single bead strings attached to the bead rectangles are in saxe-blue, the other two of white and black beads interspersed. The bead rectangles are mainly of white beads, with triangles in black and coral, “X” motifs in black, filled in with coral or turquoise and with a black zigzag motif. Each rectangle is decorated at the four corners with a mother- of-pearl button. In another specimen (5092) of this kind of harness, the colour combination is identical, except that the saxe-blue is omitted in the single strings and black and white substi- tuted. This article is worn only to dances. Headdress : ingcaca (pi. izingcaca), a narrow leather band, approximately 50 cm. in circumference, with the outer surface covered with a row of Cypraea moneta shells sewn on. Some examples of this type have a string of saxe-blue or turquoise beads spanning the diameter. Some also have a small tuft of black feather barbs, joined together, standing up from the band. Ingcaca is worn towards the back of the head with the tuft facing the front. It is worn only for rituals such as wnhlwayelelo , traditional weddings and when abafana drive ikhazi cattle to a future bride’s homestead. inkedama , (pi. amankedama) (5120) a 55,5 cm., 8-strand bead band, with small rectangles of solid-texture beadwork at the ends and four mother-of-pearl buttons and bead loops, for securing the object round the bead. Of the 8 bead strands, 4 are of saxe-blue, the other of pink and black beads interspersed. The same colours are used in the small rectangles and the loops. This is a common item of ornament, worn immediately above the level of the eyebrows, on any occasion when traditional dress is required. iintsimbi yent/oko. This is a headband, of which two types were collected. As far as is known, only the two types are worn. Both consist of bead bands, but as the descriptions indicate, they are worn differently. (a) (5123, 5126), a 50,5 cm. diameter beadwork band, 3,5 cm. wide, worn just above the inkedama. All specimens have a background of white beads, with motifs in black, or black and coral, or black and turquoise beads. The upper and lower edges of the band are trimmed with saxe-blue or turquoise beads. There may be mother-of-pearl buttons sewn on at intervals. This is the most commonly-worn headband on any occasion when traditional dress is worn. (b) (5133), a 50,5 X 5 cm. band similar in shape and colour combinations to the above variety but usually with a very short, fringe-like trimming, sometimes with mother-of-pearl buttons sewn on at intervals. This short-fringed variety is worn towards the back of the head, and is stiff enough to stand up, somewhat like a small coronet. This article is worn as part of tradi- tional dress. ingqaza, (5078) 52 cm. diameter x 1,5 cm. wide, similar to the identically-named headband of youths, as described on p. 76. The pendant, of four single strands of white beads, is wound about the torso in the same way as that of youths. The band is of short, vertical rows of white beads, edged and joined above and below with saxe-blue and black beads threaded alternately. The ends of the single strands of the pendant are short, three-pronged bead tassels in the same colours as described above. Uncommon, it is worn to dances. ip/ioco, (pi. amaphoco ) (5193) a 10,6x8,5 cm. rectangle of beadwork, with three 43 cm. single strands of white beads attached to the two top corners of the rectangle. The rectangle is predominantly of white beads, with diamond motifs outlined in black and filled with saxe- blue and white or coral and white, the upper and lower edges being trimmed with saxe-blue. 82 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE This type of object is normally worn as a necklet (see below) but the specimen described was made specifically as a headband. The bead rectangle is worn hanging at the back of the head or over one ear. Necklets: icanci (pi. amacanci), similar in appearance and shape to the youths’ amacanci described on p. 78, but the bead-stringing technique is always based on the “X” shape, as in the ithawuza , also described on p. 78. The icanci for abafana is also often made in a number of different lengths, the longest one being worn next to the skin, with the shorter ones above it. It may be as long as 20 cm., so that it has the appearance of a short cape, covering the shoulders and the chest to below the level of the pectoral muscles. The icanci is always in saxe-blue, dusty pink and white; the white appears in the form of a single horizontal stripe and is flanked, above and below, with alternate rows of the blue and pink. Informants said that abafana were not allowed to wear amacanci unless they had been given beads ( ukuxhonyxva ) by their girl friends; they would not be allowed to wear those made for them by other people. Icanci is worn at dances and weddings. ithumbu (pi. amathumbu , intestines), also similar in general shape to icanci but differing in width, bead-stringing technique and colour combination. The widest specimen in East London Museum collections is 9,8 cm., 4 — -5 cm. being more usual. The stringing technique is one of closely-interlocking bead “Y’s,” the whole presenting a somewhat lacy appearance due to the fact of the “Y’s” forming separate rows of widely-spaced beads. The upper edge of the necklet is a single row of beads, while the lower edge is a series of closely-interlocking loops. Colour combination is of alternating horizontal rows of saxe-blue, black, dusty pink, or saxe-blue and dusty pink without black, or turquoise, black, pink and white. Ithumbu is worn over icanci. It may have a long pendant of 4 — 6 single strands of beads which are attached to the lower edge of the ithumbu in the centre. isibamba valo (pi. izibamba va/o) (5184) basically an ithumbu , in all respects similar to those described, but with a number of mother-of-pearl buttons sewn on in the centre of the lower edge. To these buttons is attached a bead strip. This consists of a rectangular panel of close- textured beadwork at either end but joined to each other by means of 18 single strands of beads to form one long strip. The ends of the rectangles are buttoned onto the ithumbu by means of loops attached to the mother-of-pearl buttons. When it is being put on, the wearer takes hold of the area of single bead strands and parts them into two groups of nine. The head is inserted into this new loop and the strings of beads pulled down over the torso, so that they form a band about the torso at stomach level. The ithumbu is then fastened about the neck by means of its button and loop. It is worn at dances and weddings. umphotho (pi. imiphotho) (5106) 61,5 cm., a continuous coil of three single strings of beads twisted together. This specimen is of turquoise beads. Other colours used are white or saxe- blue, always singly. This necklet is put on by being pulled over the head once icanci and ithumbu are in place. It is worn on any occasion when traditional dress is appropriate. umdudo, (pi. imidudo) (51 19) three 52 cm. coils of beads, of identical type to umphotho , joined together by tying in two places, each of which is ornamented with a mother-of-pearl button. Two of the coils are of small white beads, the other of saxe-blue. It is worn in similar fashion to umphotho. ibotile (pi. amaboti/e) (5069) a 6,5 cm. long bottle, covered with alternate rows of light blue, black and white beads and hung about the neck by means of a string of large milky-white 83 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 beads ( amaso ), alternating with light blue and black beads. This object is put on after icanci and ithumbu when traditional dress is worn. iphoco (pi. amaphoco ), consists of a single bead rectangle, attached by the two top corners to a single or double string of beads, by which it is attached to the neck. The rectangles vary considerably in size — from about 5,8 x 4,5 cm. to 8,5 x 8,5 or larger. When two or more rectangles or squares appear on the same necklet, it is referred to in the plural, as amaphoco. The great majority of the bead rectangles or squares have a background of white beads, with a variety of types of motif, usually in black. These range from vertical or horizontal bars, to diamonds, single triangles, two triangles meeting at their apices, to stylised human figures, usually of women in long skirts and with arms in various attitudes. The latter type of motif appears usually only on an iphoco , that is, a necklet with a single rectangle. Plain geometric motifs are more common on amaphoco. This is a very common type of ornament, worn with traditional or western dress though the latter is not frequent. Waistband : ukotso (5131) (70x6,8 cm.) a band of beadwork, fastened at the front by means of eleven mother-of-pearl buttons and bead loops. When in position the beadwork consists of relatively wide vertical stripes of dark saxe-blue beads separated from each other by one row of larger milky-white beads (amaso) flanked on either side by two rows of transparent dark blue beads. There are a few rows of pink beads in the centre and at the ends of the band. This is a common article, worn by more senior young men at feasts and dances. AMADODA (senior men, from about 45 upwards) Senior men wear traditional dress only when they officiate or are guests at feasts and weddings. At sacrifices it is usually only the officiator who wears traditional dress. Anklets : isitsaba (pi. izitsaba ), exactly the same type as described for abafana. Worn on the same occasions, also at sacrifices when a man has to officiate. Armbands : umxhaka (pi. imixhaka). The same type of beaded armband, of beads wound about a central core, or a beaded strap, are worn as by abafana. Some amadoda possess but do not wear, ivory armbands, also called umxhaka (sing.). Ivory imixhaka (pi.) were in the past owned by wealthy or important older men and worn on the occasion of rituals, feasts and public meetings. The price of any ivory arm-ring is said to have been a beast. Informants now in their forties said that even in their childhood they seldom saw men wearing these arm-rings. ingxoxo (pi. amangxoxo) . (5062, 5204), approximately 30 — 35 cm. long, a band of Nerita aibicilla shells. One method of stringing these together is to perforate the body whorl, then, working with two strings, insert one through the operculum, the other through the body whorl perforation, bringing each out through the hole opposite it and continue thus until the band is completed. The alternative method is to sew individual shells to a leather band by passing a narrow thong through the operculum and out through a perforation in the apex, thus affixing it to the band. The specimen with shells sewn on to a leather band is also orna- mented with transparent dark blue and white beads, and small triangles of turquoise, black, coral and white beads arranged in concentric triangles. This ornament is worn by senior men (and women) on the biceps of the left arm, on occasions such as weddings, intonjane and important tribal meetings. Being rare, there are few specimens available. 84 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE amaso, a single string of large glass beads, worn next to the skin. The beads are in single strings of green, red, orange and pink. All abafana wear this article when traditionally dressed. Bags: ingxowa yebhokhwe (pi. iingxowa yebhokhwe), a whole goatskin, up to 108 cm. long, depending on the size of the animal. The skin is removed, starting from the rear, without cutting the skin along the abdomen or that of the legs. The skin is tanned, then turned inside out so that the hairy surface is on the inside. The opening at the head is sewn up, and the skin of the front legs is tied to the back legs to form loops. The opening is at the rear, the widest part of the skin. For carrying, the loops so formed are slipped over a shoulder, usually the left. The outer surface may be ochred or left plain. This bag is used when traditional clothes are worn; it contains a pipe, tobacco and a pipe-cleaner (isilanda). Only older men, from about 50 upwards, have this type of bag. itasi, (pi. amatasi ), a pouch-type container, often approximately 30x20 cm., of goat, blue duiker, genet or the skin of any wild animal available. One itasi of springbok skin was seen; this skin has been purchased in a town. This article may be rectangular or semi-circular. It always has a flap. The hair is left on and used on the outside. Sewing is done with thin twine or ox sinew. It is used, especially when the owner is wearing western dress, to contain a pipe, tobacco and a pipe-cleaner ( isilanda ). Earrings : amacici (sing, icici), the same shop-bought silver coloured metal rings are worn as by abafana. Headdress : amanquma, a band, approximately 50 — 52 cm., made up of small pieces of the skin of the vervet monkey, with hair still attached, each individual piece being bound with thin string and the whole lot sewn together with string. The tufts of hair extend above and below the band of string. This is worn towards the back of the head, usually without the accompaniment of any other head ornament, by older men at feasts, rituals and traditional weddings. ingcaca (pi. amangcaca, cowries), as described for abafana (p. 82). It is worn for rituals and traditional weddings. Necklets: isidanga (pi. izidanga ) (5101) made up of as many as nineteen single strings of turquoise beads, more rarely, saxe-blue ones, bound together at one point with string, sinew or a string of beads. The length of the necklet varies according to individual wish; it may be as short as navel-level or so long as to reach almost to the knees. This is worn only at weddings or sacrifices, especially when the wearer is the head of a lineage and has to preside and invoke the ancestors. isitokfele (pi. izitokfele ) (5061, 5091) consisting of a tight neck band, approximately 34 cm. long, made up of 10 — 12 single strings of beads joined together horizontally in the middle and at the sides and fastening by means of four or five bead loops and mother-of-pearl buttons. From the centre of the neck band, a varied number (6 — 10) of single strings of beads hangs down the front to thigh or knee-level. Informants said that the isitokfele was originally black and white, these colours being present in both the neck band and the pendant, with white predominating. One such example (5091) was collected. More recent specimens are in white and red beads, with red predominating. This article is worn by men of about 50 and over. It is fairly often seen at feasts. 85 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 ip/wcolamaphoco , as described for abafana on p. 84 is worn. Amadoda , however, tend to wear iphoco, the single beaded rectangle, with a motif in black on a white background. amazinyo (sing, izinyo ), a necklet of animal teeth, usually of hyrax, baboon or dog, strung together, interspersed with beads. This is worn to feasts by senior men but also on certain occasions by other age groups: girls (at the umnungqo dance), young men (at weddings, at the umhlwayelelo ritual and when ikhazi cattle are driven to a future bride’s home). Waistbands: isaziso (pi. izaziso ) (5374), 109,5 x7 cm., a band of beadwork with a 5,5 cm. fringe of single strings of beads. The predominant colour bead is white, with motifs of black or dark saxe-blue (vertical and horizontal lines, bars, triangles and diamonds). It fastens either by means of buttons attached to the strips of sheep- or goatskin sewn onto the ends, or by means of lengths of sinew or cotton wound around the mother-of-pearl buttons sewn onto the ends. Its purpose is more ornamental than functional; it is arranged to cover the top of the blanket when this has been arranged skirt-like. igqesha (pi. amagqesha) (4979), an 87 x 4 cm. band, composed of rows of large ( amaso ) and small beads and thick string, arranged horizontally. Colours: large milky white, large black, small dark saxe-blue and small black. The two ends contain no string but are entirely of beads. These two pieces of beadwork have apparently been added on to increase the length. They are joined to the main band by means of small rectangles of bhayi. To each end are attached two buttons, each with two short lengths of cord, for securing the band to the waist. Near the end worn on the left is an 81 cm. pendant of four lengths of thick string, wound round with pink, saxe-blue and black beads at the top and middle of the pendant and ending in two 13 cm. tassels each made up of 13 short strings of saxe-blue beads. This article is worn about the waist, not over a blanket worn skirt-fashion. When the wearer wraps his blanket about his body, the pendant of the igqesha can be seen hanging out below the bottom of the blanket. GIRLS ( iintombe , sing, intombe ) Anklets: ingqashela (pi. amangqashe/a), as described for abafana p. 80. These anklets are very widely found in everyday use. isitsaba (pi. izitsaba ), as described for abafana on p. 80 are worn by girls only when they perform the umngqungqo dance near the end of the seclusion period of the intonjane , also by a bride when she goes to her future husband’s home for the first time. Armbands: umxhaka (pi. imixhaka). The beaded core or beaded strap varieties, as described for abafana (p. 81), are worn by adolescent and unmarried girls, both those who consort with youths and those who consort with abafana. They are not reserved only for dances. This is a widely-worn item. Earrings : amacici (sing, icici). This is the same variety as described on p. 75 for youths, p. 81 for abafana and p. 85 for amadoda. (511 1), 4 cm. diameter (total). A 2,6 cm. diameter curtain ring, pierced in two places for the insertion of a loop of thin brass wire, by which the ring is attached to the ear, through holes pierced in the earlobes. The ring is beaded within and around the circle, giving the appearance of a flat disc of beads. Colours used are white, turquoise, black and coral. This is not so common as the plain silver-coloured rings. 86 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE Headbands: ingcaca , as described for abafana (p. 82), are worn at the umnungqo dance during the intonjane s coming-out feast. Necklets: During the period of fieldwork, the only necklets seen being worn by girls, of both the amakhwenkhwe and abafana groups, were strings of large plastic beads in white, orange, red and pink, also necklets made up of small, silvered glass balls, similar to Christmas-tree orna- ments. Informants said, however, that the following types of necklet were worn by abafana girls: icanci, as described for abafana , on p. 83. ithumbu , as described for abafana , on p. 83. umphotho, as described for abafana , on p. 83. umdudo , as described for abafana , on p. 83. Waistbands: inyilingo (pi. amanyilingo ), approximately 76x4 cm., a band of solid beadwork, or backed with a strip of bhayi or sheep- or goatskin, with two thin thongs attached to each end for tying. This band may be plain, or have two or more single strands of beads hanging down to calf-level from each side of the central ties. Colours: white (background), black, dark saxe- blue, turquoise, orange. MARRIED WOMEN ( abafazi ) Anklets: ingqashela (pi. amangqashe/a ) (5138), as described for girls on p. 86. isitsaba (pi. izitsaba), as described for abafana on p. 80. The only occasion on which a pair of these anklets is worn by an adult woman is during the period between leaving her own home as a bride with her entourage and the culmination of the traditional wedding ceremony, when she bares her torso before the assembled old men in the cattle kraal. iwoshi (pi. amawoshi), approximately 31 cm., a length of fine white-metal or brass chain, with the two ends tied together with bark fibre or a link of the chain. A pair of these is commonly worn by the majority of older married women, in preference, or in addition, to amangqashe/a. Armbands: umtseke (pi. imitseke) (5107), a metal bangle, worn singly or in great profusion on either or both arms. It consists of a continuous, thin strip of brass, wound about a core or thin brass wire or of horsehair. It may be worn at the wrist, as well as above and below the elbow. Further details of this type of article are given by Hechter-Schulz, ( S.A.J.S. , 59, pp. 51 — 3). ingxoxo (pi. amangxoxo ) (5062, 5204), as described for amadoda on p. 84. Earrings: amacici (sing, icici). Women wear the same variety of silver coloured metal earrings as de- scribed for youths (p. 75), abafana , amadoda and girls. Headbands: ingcaca , the same type as described for abafana on p. 82. These are worn at weddings. 87 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 Necklets: unogetye (pi. onogetye ) (5128), 53,5 cm. long, two single strings of beads, each closing by means of a bead loop and a large, milky-white bead. One string turquoise, the other leaf green. Worn by a young married woman. isixhoxho (pi. izixhoxho ) (5143, 5146), a fairly close-fitting necklet composed of short (6 — 7 cm.) lengths of wood, being thin branches of ichakata ( Tecomaria capensis). From 45 to 65 of these are strung together, side by side, with two narrow grooves cut into one end of each stick, to take the binding of threads of sinew or cotton. There may be a few beads between each pair of sticks or a string of beads secured to the outer surface. These are usually saxe-blue or turquoise. This type of necklet is worn by all pregnant women as a protective amulet. umkinxo (pi. imikinxo ) (5116), 33x2,7 cm., a band of bhayi, decorated with rows of well- spaced white and saxe-blue beads, also a few coral beads, with a central pendant of two short strings of red and white beads attached to a mother-of-pearl button, and fastening by means of two bead loops and two mother-of-pearl buttons. Worn by abadlezana (women suckling babies), traditionally for a period of about two years after the birth of the child. abakhubalo (sing, ikhubalo is applied to the individual pieces of root in this necklet), approximately 48 cm., a necklet composed of a number of wedge-shaped lengths of umfingwane ( Stangeria sp.) or indawa ( Cyperus sp.) roots strung on one or two single strings of beads, usually turquoise alone, or with a few white and black beads. This is worn by abadlezana (suckling mothers) as an amulet. If her baby should suddenly become ill, the mother will chew or scrape a little of one of the pieces of root, mix it with water and administer it to the child. A modification of this type of necklet is becoming popular at present. Instead of pieces of root, similarly shaped bottle tops of plastic are strung together. Though worn by nursing mothers for the same reason (ostensibly) as the real thing, the modern version is merely ornamental. intsimbi yomqhala (5181), a 44,5 cm. long necklet made up of a single row of pink beads, below which is attached a row of closely-interlocking loops of saxe-blue and black beads. At the bottom of each loop is a white bead, to which is attached a short tassel of threads of bhayi , the whole row of tassels forming a fringe. This necklet is attached to the neck by means of a bead loop and mother-of-pearl button. It is worn by older married women. incyiwana (pi. iincyiwana ) (5103), approximately 37 cm. long. One type consists of a narrow leather thong, beaded on its outer surface, with a short fringe of single strings of small beads, each string terminating in a single large black bead (i/iso amnyama). The other type consists of a narrow band of rows of beads, with a short fringe similar to the first type described. One specimen (5103), worn for a few decades up to 1952, has the band made of turquoise, white and black beads, with a fringe of coral, black and white beads. Those seen in use during the past three years have the band of transparent dark blue, white and turquoise beads, with a fringe of black and white beads. It is worn by older married women. izingqombo (5093), 52,5 cm. long, consists of two lengths of bead-covered string, sewn together one above the other, by the ends. One length is of white beads, the other of dark blue beads. It is attached to the neck by means of a piece of thin string, attached to one end, wound about a mother-of-pearl button attached to the other end. It is worn by married women. iphoco/amaphoco , similar to those described for abafana (p. 84) and amadoda (p. 86). Married women up to about the mid-fifties tend to wear amaphoco, that is, a necklet with two or more bead squares or rectangles. Women older than this, wearing this type of ornament, wear iphoco , in plain black and white. 88 BIGALKE: DRESS AND ORNAMENTATION AMONG THE NDLAMBE isidanga , as described for amadoda (p. 85), is worn by a bride on the day she goes in procession from her own home to that of her future husband. Waistbands: igqesha , as described for amadoda (p. 86), is worn by mature married women to feasts and rituals. inyilingo (pi. amanyilingo) (5121), 76,5x3,8 cm., a band of beadwork, strung in vertical rows, and with a lower border of separate loops of beads tightly packed next to each other. The beadwork of the band is white, with three groups of triangular motifs and two vertical bands in black, the border being of dark saxe-blue beads. The band is mounted on a strip of sheep- or goatskin, each end of which projects beyond the band and has three mother-of-pearl buttons. One end of the leather strip has two short, thin leather thongs which are wound around the mother-of-pearl buttons on the other end, to fix the waistband in position at the top of the skirt. It is worn mainly by older married women. Waist ornaments: isipaji (pi. izipaji ) (5083, 5156), a purse, approximately 1 1 cm. long. It is a flat, wedge-shaped object with rounded corners, hand-made of cowhide. There are two pockets, each covered by a flap. The tops of the flaps are sewn together and attached to a loop of hide. This loop is slipped on to one of the two types of waistband described above. The flap which is uppermost has its surface decorated with rows of flat brass studs and white-painted metal eyes. To the sides of the purse are attached a strip of leather, cut into streamers. These are ornamented with small pieces of thin brass strip, pinched on with pliers. This type of ornament is rarely used nowadays but has been seen worn by mature married women. DIVINERS A diviner or diviner’s novice is always readily identifiable, even in everyday dress, by the wearing of one or more articles made exclusively of white beads. In a culture apparently poor in the exegesis of symbolism, diviners to a limited extent provide explanations of the significance of white. They say that white is the colour chosen by their ancestors for them because it is associated with light and that diviners need light to clarify the problems brought to them by their clients. For this reason also, their clothing and ornament must be predominantly white. Anklets: ingqashela (pi. amangqashe/a ), a single string of white beads closing with a loop and mother- of-pearl button, or one long string of white beads wound round and round the ankle to form a wide band. The loose ends of the long string are tucked into the coil. This is worn constantly, both in everyday life, at rituals and at diviners’ dances, whether the wearer is in traditional or western dress. A pair is worn. Armbands : iwotshi ( iwatsha ) yamagqira, is a long string of white beads, wound about the right wrist in the same way such a string is wound about the ankle, it is worn on similar occasions. This item is worn every day, with traditional or western dress. 89 ANN. CAPE PROV. MUS. (NAT. HIS.) VOL. 9, PT 4, OCTOBER 1972 Headdress : iyokoza , similar to the iyokoza of youths, as described on p. 77, except that a diviner’s is of white beads only and may have a short fringe of beads in the front also. It is worn at diviners’ dances and at the rituals over which they preside. Necklets: icamagu, a necklet composed of one or two single strings of white beads, closing with a loop and mother-of-pearl button. It is worn with both western and traditional dress. usaliwe yamagqira (4956), a band (37 cm. long) of white beads with a single row of navy beads incorporated. This band is strengthened at the back with a strip of goat- or sheepskin. From the band a long (82 cm.) fringe of 120 single strings of white beads hangs down to below waist level. This article weighs approximately 1,35 kg. (2 lb. 15^ oz.). It is put on by tying the thongs at the end of the leather strip. It is worn by male and female diviners to diviners’ rituals and dances. Waistband : inyi/ingo , similar to that described for women on p. 89 but only in white beads with motifs of black beads. It is worn by male and female diviners to diviners’ rituals and dances. isaziso , similar to that described for amadoda on p. 86 but only in white with black motifs. It is worn by male and female diviners to diviners’ rituals and dances. REFERENCES Bigalke, E. H., 1966. Notes on the place of domestic and indigenous animals in Cape Nguni life, Ann. Cape Prov. Mas., 6 (1): 1 — 16. Bigalke, E. H., 1969. The Religious System of the Ndlambe of East London District, un- published M.A. Thesis, Rhodes University, Grahamstown. de Lange, M., 1963. Some Traditional Cosmetic Practices of the Xhosa, Ann. Cape Prov. Mus., 3: 85 — 95. Hechter-Schulz, K., 1963. Wire Bangles, a Record of a Bantu Craft, S. Afr. J. Science , 59 (2): 51—3. Schoeman, H. S., 1968. A Preliminary Report on Traditional Beadwork in the Mkhwanazi area of the Mtunzini district, Zululand, African Studies , 27 (2): 57 — 81. 90 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. {Nat. Hist) K VOLUME 9 • PART 5 31st JANUARY 1973 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA The status of the South African galaxiid (Pisces, Galaxiidae) R. M. McDowall Fisheries Research Division, Box 19062, Wellington, New Zealand ABSTRACT Study of general morphology and osteology shows that the South African galaxiid belongs in the genus GaJaxias. Agalaxis Scott is a synonym of Galaxias. Data also support earlier opinions that there is only one variable species of Galaxias in South Africa — G. zebratus (Castelnau) — which is described and figured. INTRODUCTION Four species of Galaxias have been described from South Africa, viz., G. zebratus (Castel- nau), G. punctifer (Castelnau), G. capensis Steindachner and G. clubius Gilchrist and Thompson. Castelnau (1861), in error, described G. zebratus and G. punctifer in Cobitis (Ostariophysi, Cobitidae — the loaches), and Steindachner (1894 — G. capensis) first placed the South African form in Galaxias. Boulenger (1905) and Regan (1905) both recognized that Castelnau’s two species belonged in Galaxias , Regan also showing that G. capensis Steindachner is a junior synonym of G. zebratus (Castelnau). Gilchrist and Thompson (1917) described a fourth species, G. clubius , but remarked that “It is not improbable that the three species described [G. zebratus , G. punctifer and G. clubius ] may on examination of more extensive collections prove to be merely varieties of one species”. This proved to be true. Barnard (1943) studied extensive collections (4 700 specimens) from diverse localities throughout the range of the genus Galaxias in South Africa, and reached the conclusion that there is only one variable species. The two names published by Castelnau (1861) had equal taxonomic status before Barnard’s revision, but as a result of Barnard’s choice of G. zebratus , this name becomes the senior synonym (Barnard, 1943, first reviser; International Code of Zoological Nomenclature 1964, Article 24 (a) (1) ). Barnard’s view has become generally accepted, although some workers have listed two (Scott, 1936, 1966) or even four species (Whitley, 1956). However, those workers familiar with the South African galaxiid seem satisfied that there is but a single species of galaxiid in South Africa, e.g., Harrison, I960, 1967; Jubb, 1965, 1967; Stokell, 1950, 1953. Nothing in the present study suggests that this is open to serious question. Despite this earlier work of Barnard’s, a full description of the species has not been prepared — all that exists is a four-line diagnosis followed by an extensive but general discussion of variability in several characters. Barnard's diagnosis has been repeated by Jubb (1965, 1967). One of the objectives of this paper is to publish a full description of the South African galaxiid, to facilitate its comparison with other galaxiids. The description conforms to those already published for New Caledonian, New Zealand and South American galaxiids (McDowall, 1968a, 1970a, 1971). 91 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 5, JANUARY 1973 A fifth galaxiid name appears in the literature (Weber, 1895) which quite obviously refers to South African galaxiids, viz., G. africanus. This name occurs in a brief communica- tion, which is quoted here because of its obscurity: De Heer Weber deelt mede, dat de door hem op de Vergadering van 30 Maart 1895 beschrevene n. sp. Galaxias africanus kort de voren door Steindachner als Galaxias capensis in de wetenschap was ingevoerd. De door hem gegeven naam moet dus vervallen. In brief, it is pointed out that Galaxias africanus described by Weber at a meeting on 30 March 1895, is the same as G. capensis Steindachner. The note states that the name G. africanus must therefore be dropped. As far as I can determine, the name G. africanus does not appear previously in the literature, and there does not seem to be a published description. Galaxias africanus is thus a nomen nudum. A more serious concern of the present paper is the generic status of the South African galaxiid. Apart from the obvious erroneous inclusion of the species in Cobitis by Castelnau (1861), the following genera have been used for G. zebratus: Galaxias Cuvier: Regan, 1905; Barnard, 1943; Jubb 1965, 1967; McDowall, 1969; and others. Galaxias subgenus Agalaxis Scott: Scott, 1936. Paragalaxias Scott: Stokell, 1950. Agalaxis Scott: Scott, 1966. That G. zebratus is a galaxiid is clear; its lack of scales, dentition, fin positions, distribu- tion of head pores, number of pelvic and caudal fin rays, and the general osteology of G. zebratus all conform closely in most details with other galaxiids (McDowall, 1969). Scott (1936) suggested that G. zebratus should be placed in a new subgenus Agalaxis within Galaxias because of its six rayed pelvic fins. Barnard (1943), on account of variation in the number of pelvic fin rays demonstrated by Stokell (1940), concluded that Scott’s sub- generic division could not be sustained, although he also pointed out that the low vertebral number in G. zebratus “40 (occasionally 39 or 41) . . . might be considered to justify generic rank” (Barnard, 1943, 231 — 2). In spite of this, Barnard retained G. zebratus within Galaxias. Stokell (1945) rejected the subgenus Agalaxis on the same grounds as Barnard, viz., intra- specific variation in pelvic fin ray counts. However, Stokell later (1950) re-examined G. zebratus. He then decided that it belongs in the genus Paragalaxias Scott, but did not indicate the diag- nostic characters that prompted this decision. He noted that, “The genus Paragalaxias is now much less distinct than it appeared when first described [Stokell, 1945, excluded Para- galaxias from the Galaxiidae but reinstated it in 1950] and it seems possible that it may ultimately intergrade with Galaxias , but in the present state of knowledge, the species dissimilis and zebratus form a natural group which requires to be distinguished in some way. The pro- visional retention of the genus Paragalaxias appears to be the course least likely to cause further complications” (Stokell, 1950:3). Low vertebral number and a short predorsal dimension seem to be the characters which Stokell used to distinguish Paragalaxias from Galaxias. Scott (1966) further dealt with the galaxiid genera. He restored G. zebratus to Agalaxis , and elevated Agalaxis from subgeneric to generic rank, following the tentative suggestion of Barnard (1943). He distinguished Agalaxis from Galaxias , Saxilaga Scott and Neochanna Gunther by its low vertebral number, from Lepidogalaxias Mees (not a galaxiid — McDowall, 1969) by absence of scales and other differences, from Br achy galaxias Eigenmann by the position of the dorsal origin and the absence of a pelvic-anal keel, and from Paragalaxias by the relative positions of the dorsal and anal fins, the size of the dorsal fin, and the disposition 92 McDOWALL: THE STATUS OF THE SOUTH AFRICAN GALAXIID and number of cephalic lateral line pores. Scott (1966) discarded pelvic fin ray number (Scott, 1936) as a generic character for Agalaxis. In a recent paper (McDowall, 1969), in which the broad relationships of the galaxioid fishes were examined, I included G. zebratus in Galaxias , making brief comments on its osteology, but did not consider in detail the generic taxonomy of Scott (1966). This considera- tion follows here. MATERIAL AND METHODS The methods of study adopted here are the same as I have used in previous galaxiid studies (McDowall, 1970a, 1971) and in general, follow Hubbs and Lagler (1958). Vertebral counts, taken from both alizarin preparations and radiographs, exclude hypural centra. Fin ray counts include all segmented rays, except in the caudal, where a standard principal ray count is given. Gill raker counts are difficult as the dorsal and ventral rakers are very small and often embedded in mucus. Thus the counts are not especially accurate and those taken from alizarin preparations tend to be higher than those taken from formalin or alcohol preserved specimens. Material studied comprised four small samples generously provided by Mr R. A. Jubb, Albany Museum, Grahamstown (AMG), South Africa. These samples came from the follow- ing localities: Olifants River system (AMG PF 1356); Eerste River system (AMG no cata- logue number); Witels River, near George (AMG PF 217); a small stream east of George (AMG no catalogue number). These samples encompass the extremes of the range of G. zebratus in South Africa, from the north-west (Olifants) to the south-east (George). GENERIC POSITION OF THE SOUTH AFRICAN GALAXIID A comprehensive description of G. zebratus is given below and data discussed are, in part, presented in this description and Tables 1 and 2. The tidiest way of considering the problem of G. zebratus and the status of Agalaxis would seem to be to consider each of the purported diagnostic characters of Agalaxis in turn, in relation both to the principal galaxiid genus Galaxias , and to the other derived genera. Of the diagnostic characters listed by Scott for Agalaxis (Scott, 1966) some are diagnostic of the family (McDowall, 1969) (reworded from Scott, 1966): scales absent; teeth in jaws and on mesopterygoids (entopterygoids) uniserial, conical, lingual teeth biserial, decumbent teeth associated with all tooth rows. Other characters come within the range of variation of genus Galaxias , viz., anal base longer than or equal to length of dorsal base, the number of dorsal fin rays equal or subequal to the number of anal rays, dorsal fin largely over anal fin; pelvic fin rays modally six (usually seven in Galaxias , but sometimes six, e.g. G. divergens Stokell); branchiostegals six or seven; gill rakers 2 — 3 and 9—10; size small. Characters remaining from Scott’s diagnosis which may be regarded as diagnostic for Agalaxis are number of vertebrae (38 — 41); number of branched caudal rays (modally 12); position of dorsal fin (0,59 — 0,67 of standard length); disposition of cephalic lateral line pores; size at sexual maturity (about 40 mm (T.L. ?)). Number of vertebrae: Scott (1966) placed great emphasis on vertebral number in the Gala- xiidae, suggesting subdivision of the family into two groups, one including forms with more than 50 vertebrae and the other those with less than 50 vertebrae. But some species, e.g., G. divergens Stokell (47 — 53 vertebrae), G. prognathus Stokell (50 — 56). G. gracilis McDowall (47 — 53) (McDowall, 1970a, 1972), have vertebral counts spanning the point of division, but are closely related to other species which fall entirely within one of such subfamilial divisions and use of vertebral number in the classification of the Galaxiidae has little value. 93 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 5, JANUARY 1973 G. zebratus has 36 to 42 vertebrae (Table 1). Nesogalaxias neocaledonicus (Weber and de Beaufort) has 41 — 43 (McDowall, 1968a); Br achy gal axias bullocki (Regan) has 38 — 42 (McDowall 1971); Paragalaxias shannonensis (Scott) (redescribed as P. dissimilis (Regan) by Stokell, 1950, a synonymy rejected by Scott (1966) who I tentatively follow) has 37 — 43 vertebrae (Stokell, 1950). Thus G. zebratus differs from all other Galaxias in having fewer vertebrae, but has about the same number as each species in the monotypic genera Neso- galaxias Whitley, Br achy gal axias Eigenmann and Paragalaxias Scott. Number of caudal fin rays: Scott gave the branched caudal fin ray count as “modally 12”. The number of principal rays is modally 14 with a range of 13 to 15 (Table 1). A usual count for Galaxias is 16, but G. divergens has a modal count of 15 (range 14 — 16); Neochanna burrowsius (Phillipps) has 13 (range 11 — 14) and N. apoda Gunther has a modal count of 16 but a wide range (13 — 17). Br achy galaxias bullocki has 15 rays (range 14 — 16). Thus the low caudal fin ray count in Galaxias zebratus overlaps counts in other Galaxias , Br achy galaxias and some Neochanna. Position of dorsal fin insertion: Scott (1966) gave the dorsal fin insertion of G. zebratus as 0,59 — 0,67 of standard length (i.e. 59 — 67%). My measurements gave figures of 59,3 to 67,0% (Table 2). Many Galaxias have this dimension 70% or more of standard length, a few New Zealand species have it 65 — 70% (McDowall, 1970a) these being species in which Table 1. Meristic variation in Galaxias zebratus Dorsal fin rays 9 10 1 1 Mean Oliphants R. 1 4 4 10,33 Eerste R. 1 7 2 10,10 Witels R. 4 3 3 9,90 “George” 5 5 10,50 Anal fin rays 9 10 11 12 13 Oliphants R. 2 3 4 11,22 Eerste R. 1 2 7 10,60 Witels R. 2 3 4 1 11,40 “George” 3 L 6 1 11,80 Pectoral fin rays 14 15 16 17 Oliphants R. 3 1 4 1 15,33 Eerste R. 1 5 3 1 15,40 Witels R. 7 3 14,30 “George” 3 5 2 14,90 Vertebrae 36 37 38 39 40 41 42 Oliphants R. 1 1 7 37,66 Eerste R. 1 5 6 38,42 Witels R. 1 6 4 41,27 “George” 3 7 4 41,07 Caudal fin rays 13 14 15 All samples 2 36 1 13,99 Pelvic fin rays 6 7 All samples 35 4 6,10 Branchiostegals 5 6 7 All samples 3 24 12 6,23 Gill rakers 8 9 10 11 12 All samples 8 10 13 7 1 9,56 94 Table 2: Variation in body proportions in Galaxias zebratus (figures given as percentages of denominator in ratio). X i^oot^ooor'ONoo- (lAcomoor^O'tMO'n^ONOoo^ o3 r-' cn — , o' r-' O — i rn oo on 33 Oi n — CTi O n 'O, I/-, rr, 1/-, rr-, ri rr, rr, rrt a C for'--0'0-'(3't«a'’tr^or^-i-H(ri-H — h-ooovoooo 5-1 CD •t co ri \D \C D- co n D- O rl Ti O O'vC — ^O0NromO|o>-Mrr rj M n ro ri £ \C o O 'Oi cr> co — C'XOr'n rl ri O O rt — oorooo rf co' — i r4 m' -h o’ n' — cb »o <£ o' r-* rr £ 6 £ — £ od £ oo ro CO— 1-^-^OC^ — <05— n vo O - < r-~ — oo >rn m rj- x ri ir iTi m in ro M ro rr M ^ C3 •rrT)- — iriovD-— oi rt m' o* oo’ oo ri ri^r.rir, nri d m io 3 0 ^ rl - ri in r r, o o O r, o O O r o ■^t* o' oo' o( — •rt- — on «-£ r~' r-* >oi r-* r~--* — ri r~* oo* doVoi cri cri cri i oo — — m \q q\ — i ioi — m3 "3“ cm — rr rj- rl -D- rj- ri ioi n — » n rj cm X -OOMirO'-"DO>nO'-H'00>rrrvoooooroooirO cD rL r-' ri — ri M3* M3* ri m’ ai O O on* o’ m3* m3* oo' >X o< o* on — onooOM30noom3c->o — >r O r cn ri on ri oo r o ri o •*— • rr Tt ri >r Tt* on m3 r~’ rr rr iri -rt* on <£ r5 ri rt — — (D oo — i — >r no on — r — r-rfri — >nori NO Tt o O rn ir! rr* O cd* On — on m* >£ oo «ri s ^ ri r^* ri o Tt- — — ri on* Tt* rt no no rd iri o ri - vi r*i o oo* >ri oo — : — Tfoo — m3 — M3 rf c i — -t in ri D- ir ri in ri ri r, ri ri CO DA JO S c r- T3 — >- o DA'S- .£ £ CX DA 5 1> c l7'£T~ u o"2 w 3 3-0 co T3 "TO C £ £ CD — Q. CX — § ^ > cD cD 33 "O T3 — ^33® ctf cD G 33 o o _£ <4— I 4— 'd CO S-H O c O c <43 33 *3, DA DA C C w a,’« o o 5 <5 ^dIlT c3 C/0 GO -t-J ctf cs p ^ § £ £ ^ f f 23 Mi CD 0A C C C o3 o3 a) sx i— J r- 4 Cu DO £3 _o -5 o: DA > 2? c or: *- I 73 03 2 m DA £3 da M-1 £3> — G.£ - cD q3 DA C _ C S3 cn 15 M3 JD £ o.£t3 £ g u o (U3 CL Qh a, X DA DA C £3 o ju "O T3 03 03 -O J A ms ‘r~ a j X7 •— J-j n o iz L 33 33 _ 95 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 5, JANUARY 1973 Fig. 1. Disposition of laterosensory pores on head of a generalized galaxiid. A. Lateral head; B. dorsal head. Arrows indicate pores lacking in Galaxias zebratus. the caudal peduncle is especially long; thus, when the caudal peduncle is particularly long, the dorsal fin is further forward in the standard length. This is true, for instance, in G. paucispondy- lus Stokell (Me Do wall, 1970a) — predorsal length/standard length is 65,8 — 72,5% and length of caudal peduncle/standard length 16,8 — 21,4%. G. zebratus also has a particularly long caudal peduncle — length of caudal peduncle/standard length is 17,8 — 23,1%. The apparently for- ward position of the dorsal fin origin in both these species is a direct consequence of the long caudal peduncle, and if any character should be regarded as distinctive for G. zebratus it is therefore the length of the caudal peduncle. However, the peduncle is not so long or so different from other galaxiids as to justify use as a generic character. Disposition of cephalic lateral line pores : The positions of lateral line pores on the head of Galaxias are fairly constant (Fig. 1). Although occasionally a fish may differ from this pore pattern, there is considerable intra- and inter-specific stability. The South African species lacks the two pores present beneath the lower jaw in Galaxias (those indicated by an arrow in Fig. 1). Curiously, Br achy galaxias bullocki lacks these same two pores (McDowall, 1971). Thus, G. zebratus differs from all galaxiids for which pore patterns have been described, except B. bullocki , in its laterosensory pore patterns. Sexual maturity: Scott (1966) noted that G. zebratus may be sexually mature at 40 mm (T.L. ?). Little is known of size at maturity in galaxiids, but G. maculatus may mature at 54 mm L.C.F. (McDowall, 1968b), G. divergens Stokell at 50 mm L.C.F. (Hopkins, 1971), G. gracilis McDowall at as little as 30 mm, although more usually at 50 mm or more (McDowall, 1972). Nesogalaxias neocaledonicus grows to 76 mm but is rarely more than 50 mm and most adults and sub-adults in collections I have examined have been between 40 and 50 mm (McDowall, 1968a). Br achy galaxias bullocki reaches 60 mm, but rarely exceeds 50 mm (McDowall, 1971), so probably matures at 40 — 50 mm. It is possibly true that G. zebratus reaches maturity at a smaller size than many galaxiids, but the difference, if any, is slight, and this character is of no consequence to the generic taxonomy of galaxiids. Osteology: Neither Scott (1936, 1966), nor others, have discussed osteology in relation to the generic classification of G. zebratus. Previously (McDowall, 1969:805) I noted that “the supracleithra are somewhat expanded to form triangular plates; the palatine spur is lacking, and in fishes examined there was no ethmoid ossification”. In fact it is the post-temporal, not the supracleithrum which is expanded. Study of further specimens otherwise confirmed these characteristics. 96 McDOWALL: THE STATUS OF THE SOUTH AFRICAN GALAXIID Variability in the caudal skeleton noted for other galaxiids (McDowall, 1969) in the occurrence and form of neural and haemal arches and spines is exemplified in G. zebratus , e.g., in two specimens cleared and stained one had one full neural arch and spine, one some- what reduced arch and spine, and one full haemal arch and spine on the urostylar vertebra, in addition to the normal parhypural and five hypurals. The other specimen had only one neural arch and spine and no haemal arch and spine on the urostylar vertebra. Clearly there is variable fusion of terminal vertebrae in Galaxias zebratus and other galaxiids. Fusion of hypural elements is evident. In one specimen, the parhypural and hypurals one and two are fused, hypural three is free, and hypurals four and five are fused. In the other, hypural three is proximally fused to but distally free from hypurals four and five. Study of further material showed that hypurals four and five are not always completely joined. Similar fusion is evident in Br achy galaxias bullocki (Greenwood et al. 1966, fig. 4; McDowall, 1969). There is nothing unique in the osteology of Galaxias zebratus that is oi sufficient significance to justify generic separation of this species. Reviewing this discussion, G. zebratus differs from all Galaxias chiefly in its low vertebral number and the absence of two laterosensory pores beneath the lower jaw on each side. It also has a long caudal peduncle. It differs from most Galaxias in its reduced number of caudal fin rays, although there is overlap. It clearly does not have affinities with the Neochanna- Saxilaga group of Tasmanian — New Zealand mudfishes with their adaptations to swamp dwelling and aestivation during drought. Paragalaxias is ill-defined and requires study, but it seems to be based on the fact that the dorsal fin is above the pelvic fins (above anal in Galaxias and G. zebratus) and that the dorsal fin has more rays than the anal (subequal in Galaxias and G. zebratus). There is no evidence of relationship between G. zebratus and Nesogalaxias , the status of which depends on the latter’s unique loss of pleural ribs. Although G. zebratus is similar to Br achy galaxias in its low vertebral count, its loss of laterosensory pores beneath the lower jaw, and its small size at maturity, it does not possess the characters that quite clearly distinguish Br achy galaxias from all other galaxiids, i.e., two distinct ural centra and an elongate alveolar process on the premaxilla which entirely excludes the maxilla from the gape (McDowall, 1969, 1971). These differences exclude G. zebratus from Brachy- galaxias. Thus G. zebratus may be considered to be excluded from all the other minor galaxiid genera that have been formally described ( Lyragalaxias Whitley and Querigalaxias Whitley await formal description, although they do have taxonomic status, having been applied to stated galaxiid species). The question for which an answer is now required is thus whether G. zebratus belongs in Galaxias , or in a separate genus, which would be Agalaxis Scott, 1936. It is important and appropriate, at this point, to discuss the circumstances surrounding the evolution of life history patterns in the Galaxiidae, and the morphological changes which have accompanied such life history changes. From osteological study it appears to me that the most primitive, unspecialized galaxiids are the diadromous species that have larvae which spend the winter in the sea (McDowall, 1969). In these species, spawning is in autumn, the eggs are relatively small and numerous, the larvae go to sea, probably soon after hatching, and they return from the sea as very characteristic, elongate, slender, transparent juveniles. Such a pattern is known in G. maculatus (Jenyns) (Australia, New Zealand, South America), G. truttaceus Valenciennes (Australia), G. brevipinnis Gunther (Tasmania, New Zealand), G. fasciatus (Gray), G. postvectis Clarke and G. argenteus (Gmelin) (all New Zealand) and is suspected in G. platei Steindachner (South America) (McDowall, 1970a, b, 1971). Many of these species have lacustrine populations in which the life history pattern is structurally similar, the lake replacing the sea. Lake populations are similar to diadromous ones, in most characters, except that vertebral number may be re- duced in the lake fishes. The diadromous life history pattern appears to be primitive for the family Galaxiidae, and this view is supported by the widespread occurrence of diadromy of 97 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 5, JANUARY 1973 various types throughout the galaxioid fishes (sub-order Galaxioidei, families Galaxiidae, Aplochitonidae, Retropinnidae and Prototroctidae — McDowall, 1969). Apart from Gallaxias, all the galaxiid genera consist wholly of non-diadromous and completely fresh water species. The fact that they are placed in different genera (mostly monotypic genera) points to specializa- tion, and the fact that they are all confined to fresh water and have dispensed with migration suggests, to me, that their specializations are connected with their adoption of a purely fresh- water life. Scott (1966) listed morphological peculiarities of Brachygalaxias , Paragalaxias and Agalaxis in which they depart from Galaxias— small size, reduction in the number of vertebrae, number of ova, number of pelvic fin rays, number of cephalic lateral line pores with some instability, movement of the dorsal, anal, and pelvic fins along the trunk, large eye. All of these are changes that occur within diadromous and non-diadromous species of Galaxias. These changes in G. zebratus can therefore be identified as those which may occur in Galaxias when a species discards the migratory and marine phases of its life history. This is seen in G. zebratus; and its morphological peculiarities — limited to reduced num- ber of vertebrae, loss of two cephalic laterosensory pores, elongate caudal peduncle, fusion of hypurals in the caudal skeleton, etc. — do not seem to justify generic recognition. Although G. zebratus has diverged from the primitive galaxiid pattern in a few characters, it is a “very ordinary” freshwater limited galaxiid. From consideration of both general morphology and osteology there seems little reason for retaining the genus Agalaxis Scott which is therefore placed in synonymy of Galaxias Cuvier. GALAXIAS CUVIER A full diagnosis of Galaxias was published by McDowall (1970a), and the reader is referred to that account. Inclusion of G. zebratus in Galaxias does not require modification of this generic diagnosis. Galaxias zebratus (Castelnau), 1872 Cobitis zebratus Castelnau, 1861: 56 (holotype: unknown; reported lost by Jubb, 1965: 16; locality: Cape Town, South Africa). Cobitis punctifer Castelnau, 1861 : 56 (holotype: unknown; also reported lost by Jubb, 1965: 16; locality: Cape Town, South Africa). Galaxias capensis Steindachner, 1894: 460 (syntypes (3): Naturhistoriches Museum, Vienna No. 6 — 466, not seen); locality: Lourens River, south-west Cape; Weber, 1895; XXXIX; Gill, 1896: 366; Weber, 1897: 154; Whitley, 1956: 34. Galaxias zebratus: Regan, 1905: 367; Boulenger, 1905: 51; Waite, 1909: 586; Boulenger, 1915: 12; Gilchrist and Thompson, 1917: 471; Barnard, 1943: 236; Harrison, 1952: 50; 1953: 128; Whitley, 1956: 34; Harrison, 1960: 14; Jubb, 1963: 8; 1964: 18; 1965: 15; Harrison, 1966: 23; 1967: 29; Jubb, 1967: 77; Breder and Rosen, 1966: 132; Anon. 1968a: 47; 1968b: 100; McDowall, 1969: 810. Galaxias punctifer: Regan, 1905: 367; Boulenger, 1905: 51; Waite, 1909: 586; Boulenger, 1915: 12; Gilchrist and Thompson, 1917: 472; Barnard, 1943: 231; Whitley, 1956: 34; Jubb, 1963: 8; 1964: 18; Breder and Rosen, 1966: 132. Galaxias dubius Gilchrist and Thompson, 1917: 472 (holotype: unknown; paratypes (3); Museum National d’Histoire Naturelle, Paris, France 23.35, not seen; locality: George River, Cape Province, South Africa); Whitley, 1956: 34. 98 McDOWALL: THE STATUS OF THE SOUTH AFRICAN GALAXIID v Fig. 2. Galaxias zebratus (Castelnau), 504- mm L.C.F., Eerste R., Cape Province, South Africa. Galaxias {Agalaxis) zebratus: Scott, 1936: 105. Galaxias {Agalaxis) punctifer: Scott, 1936: 105. Paragalaxias zebratus: Stokell, 1950: 3; 1953: 49. Agalaxis zebratus: Scott, 1966: 253. Agalaxis punctifer: Scott, 1966: 253. Description: Trunk cylindrical, moderately stout, laterally compressed on caudal peduncle, which is very long; dorsal and ventral trunk profiles about parallel; lateral line a definite mid- lateral crease. Head obtuse, of moderate size, about as broad as deep, snout rather blunt, rounded. Eye of moderate size, set high and well forward in head, interorbital more or less flat. Mouth terminal, jaws equal, cleft slightly oblique reaching to, or a little beyond anterior eye margin. Jaws without canines; mesopterygoidal teeth moderately well developed; an- terior, tubular nostril strongly developed and projecting horizontally forward almost to tip of snout; pyloric caeca lacking; gill rakers short. Fins slightly fleshy at bases and small. Dorsal fin origin well forward in standard length (due largely to great length of caudal peduncle), anal fin originates below or a little behind dorsal fin origin. Both fins short based, middle rays longest, with distal margins rounded. Pelvic fins inserted at about midpoint of standard length, pelvic-anal interval quite short, fins small. Pectoral fins short and fan-shaped, middle rays a little longer than marginal ones, fin inserted moderately high behind opercular apertures. Caudal fin long, generally truncated, sometimes slightly emarginate or rounded; peduncle flanges low but extending well along peduncle towards bases of dorsal and anal fins. Variation: Meristic and morphometric data from the four populations studied are given in Tables 1 and 2. These data show G. zebratus to be a rather variable species in a few characters, e.g., vertebral number (Table 1), but none of the populations is strikingly different from the others, and the variation seems to represent mosaic evolution in a series of allopatric popu- lations. The south-eastern populations (Witels and George) have fewer vertebrae than the western (Eerste) and northern (Olifants) populations, but the variation is no greater than is evident in comparable, widespread freshwater limited species like G. divergens Stokell and Br achy galaxias bullocki (Regan) (McDowall, 1970a, 1971). Consideration of the data in Tables 1 and 2 does not suggest that more than one species is present. Thus, Barnard’s (1943) analysis of a few characters in a great number of samples, and my analysis of many characters in a few small samples reach the same conclusion. Colouration: Highly variable, according to the nature of the habitat, varying from heavily barred to almost colourless, with all intermediates (see Barnard, 1943). Fishes I have seen 99 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 5, JANUARY 1973 usually have had somewhat regular rows of darker blotches across the back and extending down each side to varying degrees. Some have no more pigment than a general covering of tiny melanophores. Size: G. zebratus is reported by both Barnard (1943) and Jubb (1967) to reach a length of 75 mm. Habitat and life history : Little is recorded of either habitats or life history of G. zebratus. Literature sources show that it occurs in both clear and brown-stained waters, and in streams and lakes, which may be either acid (pH 5 — 6,5) or alkaline (pH 8 — 9) (Barnard, 1943; Jubb, 1967). Barnard found G. zebratus mostly in small streams, not the main rivers, on either muddy or gravelly substrates. G. zebratus clearly occurs in diverse conditions in lakes and streams. Barnard (1943) summarized knowledge of the life history of G. zebratus , and little new information has since been published. Barnard’s information is summarized briefly here. Maturity is reached at a small size, about 40 mm, the ripe eggs are moderately large, demersal, and number 30 — 40, rarely 50 or more. Spawning, and the entire life history are freshwater, although brackish water may be tolerated. Breeding is suspected to occur throughout much of the year. G. zebratus thus has a life history similar in essential details to freshwater limited galaxiids from other areas (McDowall, 1969, 1970a, 1971). Jubb (1967) reported that G. zebratus is “omnivorous . . . feeding primarily on small aquatic animals”. Distribution, zoogeography and relationships: Jubb (1967) summarized the range of G. zebratus as the “south coastal drainage basin from the Olifants River system eastwards to the Kaimans River east of George”. From Barnard’s (1943: 123 — 4) list of geographical localities, his map, (p. 233), and from subsequent reports by other workers (Anon. 1968a, b; Harrison, 1960, 1966, 1967) it is clear that G. zebratus is widespread and generally distributed in lakes and streams between the two river systems named by Jubb. It tends to be coastal and lowland, although it penetrates far inland in some rivers, e.g., tributaries of the Olifants and Gouritz rivers. Zoographicaily, G. zebratus is something of a novelty in the African freshwater fish fauna, as it is one of very few members of that group of fishes not derived from the fauna of the great tropical African rivers, and the only species of southern-temperate relationships. Galaxiids are temperate and cold-temperate fishes so that the restriction of G. zebratus to the far southern tip of South Africa is not surprising. However, it could perhaps have been expected to occur in the mountains. In view of the known salt tolerances of a good many galaxiids (see McDowall, 1964, 1970a) it seems reasonable to propose that Galaxias reached Africa across oceanic gaps, presumably from South America, since this is both the nearest land mass with Galaxias present, and is in a favourable direction if oceanic currents were instrumental in bringing Galaxias to the African continent. It is therefore, perhaps, of interest that G. zebratus shares with Br achy galaxias bullocki in Chile, the loss of laterosensory pores beneath the lower jaw. The disposition of these pores is a conservative character throughout the Galaxiidae, varying little within or between species. The similar divergence from the usual pattern in both G. zebratus and B. bullocki , may therefore be an indication of phylogenetic relationship. However, the possibility of convergence is sufficiently real to preclude any further speculation on the relationships of G. zebratus. ACKNOWLEDGEMENTS I am grateful to Mr R. A. Jubb, Albany Museum, Grahamstown, South Africa for material examined, and to Dr P. H. J. Castle, Zoology Department, Victoria University of Wellington, New Zealand for reading the manuscript. 100 McDOWALL: THE STATUS OF THE SOUTH AFRICAN GALAXIID REFERENCES Anonymous 1968(a): Galaxias in Steenbras Reservoir, Piscator, 22 (72), 47. Anonymous 1968(b): Little Princess Vlei, Heathfield, Cape, Piscator , 22 (73), 100. Barnard, K. H., 1943 : Revision of the indigenous freshwater fishes of the S.W. Cape region, Ann. S. Afr. Mas ., 36 (2), 101—262. Boulenger, G. A., 1905: A list of the freshwater fishes of Africa, Ann. Mag. nat. Hist., (7), 16, 36 — 60. Boulenger, G. A., 1909 — 16: Catalogue of the fresh-water fishes of Africa in the British Museum ( Natural History), 1 — 4, British Museum, London. Breder, C. M. & Rosen, D. E., 1966: Modes of reproduction in fishes. Natural History Press, New York. Castelnau, F. L. de L. de, 1861 : Memoire sur les poissons de T Afrique australe. Bailliere et Fils, Paris. Gilchrist, J. D. F. & Thompson, W. W., 1917: The freshwater fishes of South Africa, Ann. S. Afr. Mus., 11 (6), 5, 465—575. Gill, T., 1896: The former northward extension of the Antarctic continent, Nature, Lond., 53, 366. Greenwood, P. H., Rosen, D. E., Weitzman, S. H. & Myers, G. S., 1966: Phyletic studies of teleostean fishes, with a provisional classification of living forms, Bull. Am. Mus. nat. Hist., 131 (4), 339 — 456. Harrison, A. C., 1952: Cape Galaxias, Piscator, 6 (22), 50 — 4. Harrison, A. C., 1953: The story of Paarde Vlei. Part I: Early history and water conditions, Piscator, 7 (28), 126—8. Harrison, A. C., 1961: Longevity of Galaxias zebratus, Piscator, 14 (50), 114. Harrison, A. C., 1966: Trout in the peat-stained waters of the Southern Cape, Piscator, 20 (66), 22 — 3. Harrison, A. C., 1967: Study of a trout stream, Piscator, 21 (69), 20 — 33. Hopkins, C. L., 1971: Life history of Galaxias divergens (Salmonoidea: Galaxiidae), N.Z. Jl mar. Freshwat. Res., 5 (1), 41—57. Hubbs, C. L. & Lagler, K. F. 1958 : Fishes of the Great Lakes Region. University of Michigan Press, Ann Arbor. Jubb, R. A., 1963: A revised list of the freshwater fishes of southern Africa, Ann. Cape Prov. Mus., 3, 5 — 39. Jubb, R. A., 1964: Freshwater fishes and drainage basins in southern Africa. I. The Orange and South Coastal drainage basins, S. Afr. J. Sci., 60 (1), 17 — 21. Jubb, R. A., 1965: Freshwater fishes of the Cape Province, Ann. Cape Prov. Mus., 4, 1 — 72. Jubb, R. A., 1967: Freshwater fishes of southern Africa. Balkema, Cape Town. McDowall, R. M., 1964: The affinities and derivation of the New Zealand fresh-water fish fauna, Tuatara, 12 (2), 59—67. McDowall, R. M., 1968(a): The status of Nesogalaxias neocaledonicus (Weber and de Beaufort) (Pisces, Galaxiidae), Breviora, 286, 1 — 8. McDowall, R. M., 1968(b): Galaxias maculatus (Jenyns), the New Zealand whitebait, Fish. Res. Bull. (N.Z.) 2. McDowall, R. M., 1969: Relationships of galaxioid fishes with a further discussion of salmoniform classifi- cation, Copeia, 1969 (4), 796 — 824. McDowall, R. M., 1970(a): The galaxiid fishes of New Zealand, Bull. Mus. comp. Zool. Harv, 139 (7), 341—431. McDowall, R. M., 1970(b): A second species of Galaxias common to Tasmania and New Zealand (Pisces: Galaxiidae), Rec. Dom. Mus., Wellington, 7 (2), 13 — 19. McDowall, R. M., 1971 : The galaxiid fishes of South America, Zool. J. Linn. Soc., 50 (1), 33 — 73. McDowall, R. M., 1972: The species problem in freshwater fishes and the taxonomy of diadromous and lacustrine populations of Galaxias maculatus (Jenyns). Jl. R. Soc. N. Z. 3 (2): 325 — 367. Regan, C. T., 1905: A revision of the fishes of the family Galaxiidae, Proc. zool. Soc. Lond., 2, 363 — 84. Scott, E. O. G., 1936: Observations on fishes of the family Galaxiidae. Part I, Pap. Proc. R. Soc. Tasm., 1935, 85—112. Scott, E. O. G., 1966: The genera of the Galaxiidae, Aust. Zool., 13 (3), 244 — 58. Steindachner, F., 1894: Ichthyologische Beitrage (XVII), Sber. Akad. Wiss. Wien, 103, (1), 443 — 64. Stokell, G., 1940: A new species of Galaxias, Trans. R. Soc. N.Z., 69 (4), 422 — 24. Stokell, G., 1945: The systematic arrangement of the New Zealand Galaxiidae. Part I: Generic and sub- generic classification, Trans. R. Soc. N.Z., 75 (2), 124 — 37. Stokell, G., 1950: A revision of the genus Paragalaxias , Rec. Queen Viet. Mus. Launceston, 3 (1), 1 — 4. Stokell, G., 1953: The distribution of the family Galaxiidae, Proc. 1th Pacif. Sci. Congr., 4, 48 — 52. Waite, E. R., 1909: Vertebrata of the Subantarctic Islands of New Zealand: Pisces. In Chilton, C., (ed.) The Subantarctic Islands of New Zealand. Vol. II, pp. 585 — 98. Philosophical Institute of Canterbury, Christchurch. Weber, M., 1895: Communication on Galaxias africanus, Tijdschr. Ned. Dierk. Vereen., 2, ser. 5, XXXIX. Weber, M., 1897: Beitrage zur Kenntniss der Fauna von Siid-Afrika. Ergebnisse einer Reise von Prof. Max Weber im Jahre 1894. I. Zur Kenntniss der Siisswasser-Fauna von Siid-Afrika, Zool. Jb. Abt . Syst. Okol. Georgr., 10 (2), 135 — 99. Whitley, G. P., 1956: The story of Galaxias, Aust. Mus. Mag., 12 (1), 30 — 4. 101 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD, CAPE TOWN "S ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (nat. Hist.) k VOLUME 9 • PART 6 13th DECEMBER 1973 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA A new species of Handlirschia Kohl (Hymenoptera : Sphecidae), a very poorly known genus from South Africa by F. W. GESS Albany Museum, Grahamstown The genus Handlirschia was erected by Kohl (1896:425) for the reception of a single species, aethiops Handlirsch. This species, described from a unique male from South Africa (“Caffraria, Mus. Vindob. Coll. Winthem”), was included by Handlirsch (1889:467) in the genus Sphecius Dahlbom, though obviously with reservations. Thus Handlirsch drew attention to the differences exhibited by this species with respect to both Palaearctic and American Sphecius species and stated that it definitely represented a group of its own. It should be noted that at that date the two Sphecius species now known from the Ethiopian Region had yet to be described. As far as can be ascertained, the holotype of Handlirschia aethiops (Handlirsch). stated by Arnold (1929:260) to be in the Vienna Museum, has until the present remained the only known representative of the genus. Characters which Handlirschia shares with Sphecius and which in their aggregate delimit these two genera from the rest of the Sphecidae (South African genera at least) are the following. Head with the labrum large, in greater part exposed, wider than long but shorter than the clypeus ; ocelli convex ; mandibles with a tooth before the apex : abdomen not petiolate : legs with the middle tibiae with two spurs and with all the claws unarmed: wings with three cubital cells; the second cubital cell not stalked; stigma small; radial cell not appendiculate, distinctly longer than its distance from the point where the basal vein meets the subcosta. Characters possessed by Handlirschia which separate this genus from Sphecius are the following. Thorax lacking the episternal suture and the epicnemium (both present in Sphecius ); propodeum compressed laterally so that the declivity is transversely concave (in Sphecius not compressed, declivity rounded or only feebly concave in its middle portion); hind wings with cubital vein originating shortly before end of submedial cell (originating far in front of end in Sphecius). An unknown species represented in the Albany Museum collection by a small series of specimens from the north-eastern Transvaal Lowveld would, judged on the above delimiting characters, appear to be referable to the genus Handlirschia. Allowing for the fact that Kohl’s generic description is based upon a single species and may consequently include a few characters of specific rather than generic status, the present species fits remarkably well. It therefore seems warranted to treat it as a species of Handlirschia and to describe it as such, as is done below. Handlirschia tricolor sp.n. Male. Labrum, clypeus, frons (except for black supra-antennal spots and ferruginous streaks connecting these with the similarly coloured ocellar region), temples broadly, mandibles (except for dark ferruginous apical third), underside of scapes, upper portion of pronotum 103 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 6, DECEMBER 1973 Handlirschia tricolor sp.n.(d) Fig. 1. Head, thorax and abdomen (dorsal view). Fig. 2. Head (frontal view). Fig. 3. Sternites 6-8 (ventral view). Fig. 4. Wings. Scale a = 1 mm Fig. 1 b = 1 mm Figs. 2-4 104 GESS: A NEW SPECIES OF HAND LI RSCHIA KOHL (HYMENOPTERA : SPHECIDAE) including tubercles, anterior aspects of fore coxae, anterior aspect of mesopleura, lemon- yellow, sides of mesonotum, mesopleura medially, scutellum and metanotum, greater part of discs of tergites one to three and markings on four and five, a darker more orange-yellow. Disc of mesonotum, a narrow subvertical streak on mesopleura, propodeum (except for median area of declivity and broad anterior and ventral margins laterally), anterior (declivous) face of first tergite, sides and roughly circular marking on disc of same, sides of tergite two, apex of tergite seven, sternites one and two and apex of sternite six, reddish-ferruginous’, hind margins of tergites one to four, anterior bands or markings on tergites two to four, sides of tergites three and four, almost all of tergite five, whole of tergite six and basal part of tergite seven, sternites three to five and all but apex of sternite six, darker ferruginous, in parts very dark, occasionally black. Supra-antennal spots, sometimes spots between the ocelli, underside of head, maxillae and labium basally, groove of pronotum, prosternum, central third of anterior margin of mesonotum (sometimes expanded posteriorly along midline), mesopleura dorsally immediately ventral to tegulae and also beneath and behind pronotal tubercle as well as posteriorly along suture (marking sometimes expanded onto disc) and sometimes ventrally in front of coxal cavities, metapleura, broad anterior and ventral margins of propodeum laterally and median area of declivity of same, black. Antennae (except for lemon-yellow underside of scape) and legs (except for black at extreme base of coxae) varying from orange-yellow to light ferruginous. Wings slightly smoky, veins brown. Length 12 — 13 mm, length of fore wing 9 mm, hamuli 15 — 20. Whole body clothed with fine, decumbent, silvery-white pubescence, densest on labrum, clypeus and frons and on mesopleura (giving these parts a silvery sheen in oblique light), sparcer on rest of thorax and on legs, sparcest on tergites. Coarser vestiture consisting of fine silvery-white erect pilosity most noticeable on frons above antennal sockets and on head below, on mesopleura and particularly on abdomen. Head with puncturation (most obvious on frons and vertex) very fine and dense; pro- notum appearing almost impunctate; mesonotum, scutellum, metanotum, mesopleura as well as dorsal and posterior declivous aspects of propodeum with fairly dense, large, ill-defined, shallow, slightly elongate punctures; metapleura and sides of propodeum very finely punctured with in addition an extremely sparce scattering of larger punctures ; abdomen with puncturation of roughly similar size to that of mesonotum; puncturation of tergite one sparcest, that of following tergites progressively closer, that of tergite seven and also of sternites very like that of mesonotum and possibly somewhat coarser on terminal sternites. Head narrower than thorax, in frontal view (Fig. 2) about 1,3 times wider across the eyes than long (from vertex to distal edge of clypeus); inner margins of eyes subparallel, divergent below, closest together a little above level of antennal sockets; frons at this level 1,5 times as wide as one eye; antennal sockets closer to each other than to eyes; ocelli round and convex; anterior ocellus distinctly larger than posterior pair; posterior ocelli separated from the eyes and from the anterior ocellus by their own diameter and from each other by twice this distance; clypeus with discs evenly convex and with anterior and posterior margins concave, about 2,5 times wider than long at midline; labrum with disc evenly convex and with anterior margin evenly curved, about twice wider than long but only about 0,7 times as long in midline as clypeus; mandibles fairly robust with a small inner tooth before apex; maxillary palps 6-segmented and labial palps 4-segmented; antennal scape robust, its length (without radicle) a little less than double apical breadth; segment 2 short; segments 2 and 3 together of same length as scape (without radicle); segments 4 — 13 shorter than 3; 4 — 10 becoming progressively shorter but wider; 1 1 equal to 10; 12 and 13 (in that order) longer and narrower; none of the segments excavated beneath. 105 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 6, DECEMBER 1973 Handlirschia tricolor sp.n. (<39 Fig. 5. Genitalia (dorsal view). Fig. 6. Genitalia (ventral view). Pronotum short, transverse, not quite linear above, sloping down steeply anteriorly; pronotal tubercles prominent, extending posteriorly to just beyond the anterior edge of the tegulae; mesonotum 1,75 times wider (above pronotal tubercles) than long in midline, with well developed lamina laterally expanded over the bases of the tegulae and posteriorly obliquely truncate; scutellum with disc almost flat, almost three times longer in midline than metanotum; mesopleura and metapleura together very convex causing the thorax to bulge out laterally at this level; mesopleura without any indication of episternal suture and epicnemium; pro- podeum with sides subparallel, with dorsal and declivous parts rounded at junction, laterally at least almost smoothly arcuate; median triangular area very clearly defined by deeply impressed lines and furnished in the midline over its apical third with a wide, deeply-impressed, smooth-edged longitudinal fossa; sides of propodeum somewhat compressed and the declivity thus somewhat concave; median area of declivity longitudinally (i.e. subvertically) bicarinate; the carinae nearest together in the upper quarter of declivity, abruptly divergent above this point and gradually divergent below. Abdomen widest at middle of second segment; tergite one a little more than half as long as wide; tergite seven with sides converging distally, its apex narrowly rounded; sternite two in basal half slightly swollen medially and slightly depressed postero-laterally to swelling; sternite six (Fig. 3) with sides converging posteriorly, its apex narrowly rounded, its disc slightly depressed; sternite seven hidden under the sixth, narrow, only one third as wide as sternite eight, lamellate, its sides subparallel, its posterior angles rounded and its hind margin transverse ; 106 GESS: A NEW SPECIES OF HANDLIRSCHIA KOHL (HYMENOPTERA : SPHECIDAE) sternite eight wide, its sides slightly convergent posteriorly, its posterior angles rounded and the hind margin transverse on either side of a long, robust, downwardly curved, pointed spine projecting posteriorly in the midline and extending beyond the end of the abdomen. Genitalia (Figs. 5 and 6) with parameres large and of more or less even width; cuspis shorter than digitus, rounded apically and furnished with fine setae in distal half beneath; digitus heavily sclerotised, strongly downwardly curved, sickle-shaped in side view, inwardly curved at apex. Bases of coxae of middle and hind legs covered by paired posteriorly directed processes of the meso- and metasterna; processes of mesosternum wider, more or less lamellate and triangular; those of metasternum narrower, more or less finger-like; hind coxae approximate, much larger than separated middle coxae; legs generally unmodified; tibiae and tarsi sparsely spinose; middle tibiae with two spurs; the outer spur wider than the inner and sharply bent shortly before the apex; spurs of hind legs long and spatulate; all claws simple. Wings as in Fig. 4. Female Very similar to male in colouration, general appearance and size, differing however in the following respects: fore legs, especially tarsi, more robust; comb present on fore tarsi and composed of six stiff spines on first tarsomere, two on second tarsomere and one on both third and fourth tarsomeres (spines 1 — 5 of first tarsomere short, rest longer); terminal tarsomere of fore leg, its claws and pul villus greatly enlarged; outer spur of middle tibia the same as the inner, not bent shortly before apex; tergite six with a pygidial area indicated on apical half. Described from five specimens, all with the same data: Transvaal: Gravelotte, Beacon Ranch, i. 1966 (D. J. Brothers), Holotype Allotype ? and 3 Paratype TT- (Albany Museum collection). Judging from the description of //. aethiops (Handlirsch) the most noticeable differences between that species and tricolor sp. n. are those of puncturation, pilosity and colouration. Thus aethiops has almost the whole body very coarsely punctured and covered with a dense fuscous pilosity; in colouration it is predominantly black with a violaceous lustre on the abdomen, while yellow markings are restricted to parts of the head. Morphologically the outstanding points of difference are that aethiops has some of the antennal segments excavated beneath, has the propodeum smooth and shiny and without a median area, and has the tibiae of the middle legs furnished with a pointed tubercle below near the apex which is also produced into a point. In addition the wings in relation to the body are very small. REFERENCES Arnold, G. 1929. The Sphegidae of South Africa. Part 12. Ann. Transv. Mus. 13: 217—319. Handlirsch, A. 1889. Monographic der mit Nysson und Bembex verwandten Grabwespen. IV. Sber. Akad. Wiss. Wien , Mathem.-naturw. Classe 98: 440 — 517. Kohl, F. F. 1896. Die Gattungen der Sphegiden. Annin, naturh. Mus. Wien 11: 233 — 516. 107 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (nat. Hist) VOLUME 9 • PART 7 13th DECEMBER 1973 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA PRINTED BY CAPE & TRANSVAAL PRINTERS LTD* CAPE TOWN Third contribution to the knowledge of the South African species of the genus Ceramius Latreille (Hymenoptera : Masaridae) by F. W. GESS Albany Museum, Grahamstown CONTENTS Introduction Description, synonymy and distribution records .... Table listing flowers visited by Ceramius species in South Africa Description of nesting sites and nest superstructures Parasites Key to the South African species of Ceramius Latreille Acknowledgements References Page 109 110 115 117 118 119 122 122 INTRODUCTION The South African species of the genus Ceramius Latreille are fairly well known, at least from a taxonomic aspect, having in recent years been studied by Richards (1962) and by Gess (1965 and 1968). Nevertheless, Ceramius material in the Albany Museum, in the private collection of Mr. H. N. Empey and in the collections made in South Africa by Dr. J. Rozen (1966 and 1968) and by Dr. and Mrs. H. Townes (1970), combined with continued collecting and field observations by the author and by Mr. J. G. H. Londt, has brought to light some interesting additional information. In the present paper the hitherto unknown male of the very rare species C. rex de Saussure is described, C. sehulthessi Brauns is sunk into synonymy with C. cerceriformis de Saussure, and additional locality records of several other species are listed. A table listing flowers visited by Ceramius species in South Africa shows a five-fold increase in the number of observed wasp-forage flower associations, the preferred flowers belonging to the Mesembryanthemaceae and Compositae. Also resulting from field observations are descriptions of the nesting sites and nest superstructures of C. lichtensteinii (Klug) and C. eapicola Brauns, and the identifica- tion of a hitherto unrecorded parasite. A key restricted to the South African species of the genus but incorporating recent changes and additions is given. 109 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 7, DECEMBER 1973 DESCRIPTION, SYNONYMY AND DISTRIBUTION RECORDS Ceramius rex de Saussure Ceramius rex de Saussure, 1855:75, $; Turner, 1935:290; Gess, 1965:225, $. Ceramius lichtensteinii ( non Klug), Richards, 1962:102. Described from the Cape Colony without precise locality, this species does not appear to have been collected again until over eighty years later when a single specimen believed to be conspecific was obtained in Namaqualand. This specimen, like the type a female, labelled “Klipvlei, Garies, xi. 1931 (S. A. M. Staff)” was stated by Turner (1935: 290 — 1) to correspond to the description of rex de Saussure, an opinion shared by Gess (1965 : 225) who gave a detailed description of the specimen in question. In an unsuccessful attempt to obtain further specimens, the author collected in the vicinity of Garies during early October, 1966 and again at the same time in 1967. Since then, however, a further specimen has come to hand — collected entirely by chance by Dr. and Mrs. H. Townes at Garies (25.ix.1970). This specimen, a male, clearly conspecific with the Garies female of 1931 and therefore believed to be the long unknown male of rex, is described below. + Black; mandibles (except teeth at apex), clypeus, broad spot between antennal sockets and above clypeus (from which it is separated by a narrow black line at suture), a narrow streak on inner margin of eyes below extending as far as centre of ocular sinus, underside of scape and of second antennal segment as well as extreme base of third below, spots behind eyes (not joined along occipital margin), pronotal band produced onto humerus and to tegula (leaving a triangular black area on side), anterior corner of tegula, small streak margining postero-lateral corner of mesoscutum, small spot at apex of scutellum, flap of posttegula, small spots on posterior angles of propodeum, a large spot on mesopleuron, spots on meso- sternum behind adjoining coxae, legs (except for black upper surface of coxae and trochanters and streak on femora, distal parts of third to fifth tarsomeres of middle legs, and to a less extent hind legs, and also pulvilli and claws of all legs), posterior bands widening markedly on sides on gastral tergites 1 — 6, baso-lateral spots on tergite 7, greater part of sternites 2,4 and 5 and lateral areas of 3,6 and 7, yellow. Antennal flagellum except for black dorsal suffusion on all segments bar the last (which however has a little black basally), and extreme sides of tergite 7 ferruginous. Wings faintly brownish, veins brown, subcosta and median vein almost black. Length about 20 mm, length of fore wing 13 mm, hamuli 24. Head, thorax, propodeum, first two gastral segments and seventh sternite with fairly sparse, long, whitish hairs; coxae, trochanters and femora posteriorly with similar but shorter hairs; concave portion of disc of sternite 3 with exceedingly dense, short, light ferruginous hairs; rest of gaster with very fine tomentum-like pubescence. With the exception of the secondary sexual structural characters described below, very like the female in most respects including the structure and proportions of the clypeus, scutellum and acarinarium. Antennal sockets separated by slightly less than twice their diameter (4,1 times in female); interocular distance at level of sockets one and a third times the length of scape (without radicle) (twice in female). Scape curved, strongly widened, twice as long (without radicle) as greatest width (almost 3 times in female); segment 2 very short, broader than long, largely concealed in end of scape; segment 3 flattened in side view, narrow except at apex, two thirds as long as scape (without radicle) and slightly longer than 4+5+6; 4 — 11 becoming progressively wider; 12 forming a powerful, long, flattened and fairly wide hook, inwardly curved at both base and apex; segments 8 — 11 modified beneath (Fig. 1). Fore trochanter (Fig. 2) with a very large, outwardly curved, crescentic lobe whose distal half is more or less 110 GESS: SOUTH AFRICAN SPECIES OF THE GENUS CERAMIUS LATREILLE Ceramius rex de Saussure <$. Fig. 1. Antennal flagellum. Fig. 2. Fore trochanter, femur and tibia. Fig. 4. Sternite 3 (ventral view). Fig. 5. Sternite 3 (profile, from left side). Fig. 6. Genitalia dorsal view). Ceramius metanotalis Richards <$. Fig. 3. Fore trochanter and portion of femur. Ill ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 7, DECEMBER 1973 evenly curved and of even width (in metanotalis (Fig. 3) abruptly curved and widened apically). Gaster with tergite 7 elongate, apically widely, very shallowly and angularly emarginate; sternite 3 (Figs. 4 and 5) with disc raised anteriorly and laterally and especially antero-laterally to form a carinate rim of a concave, wider than long, saucer-like depression; sternite 4 unmodi- fied in structure; sternite 7 with a prominence. Genitalia (Fig. 6). Specimen examined: Cape Province (Namaqualand) : Garies, 25.ix.1970 (Dr. and Mrs. H. Townes) Metallotype (In the Albany Museum collection). The acarinaria both contain mites, several of which are also present amongst the hairs on the propodeum. The male of rex may readily be distinguished from the males of both caffer de Saussure and metanotalis Richards, which are the most closely related species, by the much more strikingly modified third sternite; from caffer by the absence of preapical lateral keels on sternite 4; and from metanotalis by the different shape of the lobe of the fore trochanter. Cera mi us cerceriformis de Saussure Ceramius cerceriformis de Saussure, 1853: xxi, <3; Richards, 1962:97, ?. Ceramius ( Ceramioides ) cerceriformis de Saussure, 1854: pi. 4, fig. 1, 1855:72, Cerceris vespiformis de Saussure, 1855:79, $. Ceramius schulthessi Brauns, 1902:182, ?; Brauns, 1913:196, pi. 2, fig. 6, <£, $; Richards, 1962:99; Gess, 1965:220; Gess, 1968: 10, syn.n. Richards (1962) who examined the types of cerceriformis , vespiformis and schulthessi was “not altogether convinced44 that schulthessi , which was “scarcely distinguishable” from cerceriformis , was distinct from the former, but the females which were compared “seemed to be distinct”. According to Richards’ descriptions and his key the only differences between cerceri- formis and schulthessi are in their puncturation and colouring. Thus both sexes of cerceriformis have the gaster impunctate while in schulthessi it is slightly shining and distinctly though finely punctured, at least on the first two tergites. In the female furthermore, the pronotum and mesonotum are dull with denser punctures in cerceriformis ; more shiny with coarser, sparcer punctures in schulthessi. While sternite 2 is punctured throughout in the former, it is shining with a large almost unpunctured area on each side of the disc in the latter. Their colouration differs in that cerceriformis has very extensive, entirely yellow markings, while schulthessi is less extensively yellow-marked but has some red areas. The amount of red on gastral tergite 2 and yellow on 3 — 6 is however very variable. Two pairs of wasps from Namaqualand (Swart Doringrivier and 6 miles South of Garies), representatives in the Albany Museum collection of longer series in the South African Museum, which were recorded as C. schulthessi Brauns (Gess, 1968), appear upon re-examination to fit the descriptions of cerceriformis reasonably well. Compared with these specimens, a series from Willowmore, fitting the description of schulthessi indicates some further minor differ- ences: in the latter there seems to be a tendency for the disc of the scutellum to be somewhat more clearly margined, while the spine-like projections on the propodeum are generally more developed. However, the degree of development of these spines seems to vary also within a population. While these small differences in the degree of puncturation, the degree of development of the propodeal spines, and the colouration do certainly exist, the similarity otherwise between cerceriformis de Saussure and schulthessi Brauns is such that it is impossible to separate the two using any other criteria. Are these characters of any real value? As has been noted by both Brauns and Richards, the distribution and amount of red and of yellow in schulthessi is very variable. The degree of 112 GESS: SOUTH AFRICAN SPECIES OF THE GENUS CERAMIUS LATREILLE development of the propodeal spines is similarly variable, the variability in this case being present in both cerceriformis and sehulthessi. It is believed that the remaining point of differ- ence, namely that of the degree of puncturation, will with the examination of more material likewise be found to be variable and of questionable value. On balance, the small differences in detail, of limited or questionable taxonomic validity, are far outweighed by the striking overall similarities, and there seems little if any justification in maintaining the specific integrity of the two forms on such slender evidence. C. cerceriformis de Saussure and C. sehulthessi Brauns may best be considered as opposite extremes of a single widespread and very variable species ; only much more extensive collecting can indicate whether they are sufficiently distinct to be accorded subspecific rank. C. cerceriformis de Saussure has priority and C. sehulthessi Brauns must therefore sink into synonymy. Ceramius cerceriformis de Saussure Cape Province: Willowmore, 15.xi.1905 (Dr. Brauns) 43$, 3 1.x. 1967 (C. Jacot-Guillarmod) 9 $$, (?. Ceramius richardsi Gess Cape Province: Citrusdal, 2.xi.l966 (J. G. Rozen) 2 Ceramius micheneri Gess Cape Province: Citrusdal, 2.xi.l966 (J. G. Rozen) 2 Ceramius nigripennis de Saussure Cape Province: Garies, 23. ix. 1970, <$, 24. ix. 1970, 3 $$, <3\ 25. ix. 1970, 27. ix. 1970, 11 8 (J(J, 28. ix. 1970, 4 ?$, <$, 29.ix.1970, 4 ??, J (all Dr. and Mrs. H. Townes). Ceramius braunsi Turner Cape Province: Worcester, 16.ix.1972 (R. D. A. Bayliss) This new locality is of interest as it to some extent links the previously known distributional areas — the region between Citrusdal and Vanrhynsdorp on the one hand and Willowmore on the other. Ceramius jacoti Richards Cape Province: Brandrivier road, 2 miles from junction with Ladismith— Riversdale road, 30. ix. 1972 (C. F. Jacot-Guillarmod) 3 Ceramius beyeri Brauns Cape Province: Clanwilliam, 2 — 5.xi.l966 (J. G. Rozen) 2 Grahamstown, Bible Monu- ment, 16. i. 1969 (F. W. Gess) Willowmore, no date (Dr. Brauns) 3 $$. Ceramius lichtensteinii (Klug) Cape Province: Alicedale, New Year’s Dam, 2.xii.l970 (F. W. Gess) same date (J. G. H. Londt) 4 $<$; Dunbrody, no other data, Ecca Pass, 22.xi.1964 (D. J. Brothers) <$; Fort Beaufort, 20. i. 1960 (C. Jacot-Guillarmod) J; Fort Willshire near Alice, 21. i. 1959 (C. Jacot- Guillarmod) $, S; Grahamstown, Bible Monument, 16. i. 1969 (F. W. Gess) $; Grahamstown, Clifton, 7.xi.l972 (F. W. and S. K. Gess) 7 ?$, S, 9.xi.l972 (F. W. and S. K. Gess) 3 ??; Grahamstown, Hilton, 1 — 4.xii. 1 970 (F. W. Gess — Malaise Trap) $; Grahamstown, Kranz- drif, 5.xi.l967 (C. Jacot-Guillarmod) Grahamstown, Plutos Vale, 8.xi. 1 964 (C. Jacot- Guillarmod) Grahamstown, Strowan, 9.xii.l968, 2 $$, <$, 1 1 .xii. 1 968, $, 8.L1969, $, 30. xi. 1970, ?, 2 (f(f (all F. W. Gess), 20.xii.1970 (C. Jacot-Guillarmod) <$; 4 miles NE of 113 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 7, DECEMBER 1973 Steytlerville, 1 2.xi. 1 968 (J. G. Rozen and E. Martinez) 9, 3‘, Thorngrove, 29. i. 1960 (L. Naerly) 9; Victoria West, 1 0.i. 1 965 (H. N. Empey) 19 99, 3', Waterford, 13. i. 1965 (H. N. Empey) 8 99, 3', 4 miles E of Waterford, 29.x. 1967 (C. Jacot-Guillarmod) $; Willowmore, 12. i. 1965 (H. N. Empey) 3- Orange Free State: Kroonstad, 10. i. 1965, 3 99, 2 33, 16. i. 1965, 3, 23. i. 1965, 9, 27. i. 1965, 9, xii.1965, 9, 3, 3 1 .xii. 1 965, 3, 21.xii.1966, 3 (all D. J. Brothers). Ceramius bicolor (Thunberg) Cape Province: Fullarton, Willowmore, 30.x. 1967, 6 99, 3', near Fullarton, Willowmore, 30.x. 1967, 3', Willowmore, 31.x. 1967, 9 (all C. Jacot-Guillarmod); 18 miles SE of Touwsrivier, 12. xi. 1966 (J. G. Rozen) 2 ??. Ceramius linearis Klug Cape Province: Alicedale, New Year’s Dam, 22.xi.1970 (J. G. H. Londt) 10 33, 2.xii.l970 (J. G. H. Londt) 4 99, 4 33, same date (F. W. Gess) 3 99, 10 33, 16.xii.1971 (F. W. Gess) 2 ??, 2 33', Carlisle Bridge, xii. 1971 (R. Bayliss) ?; Grahamstown, 13.vi.1959 (E. McC. Callan) 9, 15. xii. 1959 (E. McC. Callan) ?, 12.xi.1960 (E. McC. Callan) 9, 27.xii.1960 (E. McC. Callan) 3, 8. xii. 1964 (D. J. Brothers) 2 33, 18. xii. 1969 (J. G. H. Londt) $; Grahamstown, Belmont Valley, 4.xii. 1 969 (J. G. H. Londt) $, 15. xii. 1971 (J. G. H. Londt) 9, 3', Grahamstown, Boskey Dell, 24. ix. 1967 (C. Jacot-Guillarmod) $; Grahamstown, Clifton, 17.x. 1972 (F. W. and S. K. Gess) 11 99, 10 33, 26.x. 1972 (F. W. and S. K. Gess) 24 99, 46 33, 27.X.1972 (F. W. and S. K. Gess)2(d(d, 7.xi.l972(F. W. and S. K. Gess) 8 99, \0 33, 9.xi.l972(F. W. and S. K. Gess) 7 99, 13 (dcd; Grahamstown, Hilton, 22.x. 1967 (D. J. Brothers) 3 33, same date (C. Jacot- Guillarmod) 9, 5.xi.l969 (C. Jacot-Guillarmod) 3, 2 — 5.xi.l970, 9, 5 — 9.xi.l970, 9, 12 — 30.xi.1970, 9, 3 33, 1 — 4. xii. 1970, 9, 3 (all F. W. Gess — Malaise Trap); Grahamstown, Hounslow, 22. xii. 1966 (C. Jacot-Guillarmod) $; Grahamstown, Plutos Vale, 8.xi.l964 (C. Jacot-Guillarmod) \ % @ soil sample W// p sherd /;"////„ '/////., D B Fig. 7. Plan of Structure J on Khartoum 1 showing position of excavation. 134 CO© HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT KHARTOUM 2, STRUCTURE A, EXCAVATION 1 Following the results of the investigations of the small enclosure Structure J on Khartoum 1, it was decided to excavate one of the small enclosures on Khartoum 2. Structure A was selected for investigation. This Structure is located close to the base of the dolerite ridge, and on a slight slope, so that there had been deposition in the enclosure up against the lower wall rather than erosion which is such a feature of Type R Settlement Units. An area of 6 square metres was marked off on the down-slope side of the Structure (see Fig. 4 for the precise position of the excavation and Fig. 8 for detail). The excavation pro- ceeded in 15 cm spits. The first spit was removed from the entire area and proved to be com- pletely sterile. The deposit consisted of a brownish soil which, apart from the top few cm which were loose, was compacted but not as hard as the deposit in Structure L of Khartoum 1. A second 15 cm spit was excavated from the 1,5 m area nearest to the wall and at the base of this spit bedrock was exposed. The bedrock was very friable shale which flaked away as it was brushed. The base of the wall was also exposed in this spit and it was clear that the wall had been built on a surface only 3 — 4 cm above bedrock. In view of the slope on which the Structure had been built, this was not a surprising discovery because prior to the con- struction of the wall (which held back about 30 cm of deposit) the surface soil could not have been very thick on account of the rain water run-off down the slope which would have washed most of it away. 135 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 The second spit was also sterile and in view of the fact that any materials which may have been in the enclosure would almost certainly have been accumulated within the 30 cm of deposit down-slope banked against the wall, it was decided not to proceed any further with the excavation. The upper half of the excavation was therefore not taken any lower than the first spit level. THE FINDS The Khartoum 1 finds to be described here were all derived either from the excavations in Structure L or from surface collecting; the excavation in Structure J produced only 2 body sherds while the excavation in Khartoum 2, Structure A, was sterile. The sample of finds from Khartoum 1 is small but interesting and can be complimented by consideration of surface finds from several other Settlement Units as well as from the excavations undertaken by Maggs (1971) at OFD 1. 1. POTTERY Form. The excavations in Structure L only produced some 78 sherds of which 10 were rim sherds; despite much time spent, it was not possible to reconstruct any pots to the extent of being able to define shape and size although several of the rim pieces fitted together (Fig. 9). Several other Settlement Units have, however, provided evidence on the shapes and sizes of pots, among these being those on the farm Pramberg only 5 km to the north of Khartoum. The pottery from these other Settlement Units is illustrated in Figs. 10 — 13. The evidence available is somewhat fragmentary and so a detailed analysis of shapes and sizes is not possible; one pot and two bowls can, however be reconstructed sufficiently for the shape and size to be defined in detail. The pot is a large specimen with a rim diameter of 26 cm and a height of 22 cm (Fig. 1 1). The bowls are from Pramberg and Mierkraal and measure 23 cm x 13 cm and 17 cm x9 cm respectively (Fig. 10 and 13). Some 12 other pots can be reconstructed enough for estimates to be made of their rim diameters. These pots have, for convenience, been divided into 2 classes : “Pots” and “bowls”, — on the basis of the fragmentary evidence available a more detailed breakdown does not seem justified. “Pots” are those vessels that have the sides curving inwards, i.e. the rim diameter is less than the maximum diameter of the pot. “Bowls”, on the other hand, are those vessels which have the rim diameter as their maximum dimension; the sides can vary from almost vertical to very widely flared. Of the total 15 vessels than can have their rim diameters reconstructed, only 4 can be rated as “bowls”. These “bowls” range in rim diameter from 16 to 24 cm while the “pots” have a much wider range from 13 to 36 cm. The sample of vessels is too small for any definite conclusion to be drawn, but “pots” seem to outnumber “bowls” by almost 4:1 — “bowls” were also relatively rare in Maggs’ (1971) analysis. Several bases of pots could be reconstructed and all of these were rounded; there is no evidence to suggest that any of these bases were flattened or pointed. Rims. The rim fragments available for study consist of 10 excavated from Structure L and some 33 from other assemblages; clearly a sample too small for detailed analysis. Maggs (1971 : 52), on the basis of his large OFD 1 sample, has suggested that rims tend generally to be rounded, with a few flattened on their upper surfaces. He also refers to some that are “rolled over but not to the extent of being significantly thickened”. Of the OFD 1 sample about 1 in 6 rims were found to be accentuated by a projecting step or ridge a few mm below the lip and Maggs suggested that this may be a distinctive feature of the Type R pottery. The sample collected in the course of this project exhibits most of the rim features described by Maggs but the occurrence of accentuated rims is not such as would be expected if this were a “distinctive feature” of the pottery. 136 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT KHARTOUM 1 STRUCTURE L EXCAVATIONS Fig, 9. Rim sherds recovered from Excavations 1-3 in Structure L on Khartoum 1. 137 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 Fig. 10. Rim sherds recovered from Settlement Units on Pramberg, north of Khartoum. Many of the rim sherds recovered show an unevenness in finish which is noticable even on these small fragments and this may be evidence that rims were not made to a distinctive or consistent shape; even and well made rim sherds constitute a very small proportion of the total sample. Decoration. Of the several hundred sherds collected from Settlement Units, only 3 were found to have been decorated and none of these were from Khartoum 1. This lack of decora- tion is accentuated even more when we consider Maggs’ sample of 991 from OFD 1 alone which produced only another 4 decorated sherds. As Maggs (1971: 52) has pointed out, this lack of decoration is something of a diagnostic feature in itself for most Iron Age and even “Later Stone Age” pottery is decorated in a distinctive way. Maggs described the decorated sherds from OFD 1 as follows: “One has a herringbone motif in shallow grooves on a rounded rim, the others have one or more rows of small triangular or ‘D’-shaped impressions. The impressions seem to have been made by a comb rather than a stylus, but they differ from the normal comb-stamping of the Orange Free State, which shows square or rectangular impressions.” In this project decorated sherds were recovered from Weltevreden 1 (Fig. 11) and the Poortjie Settlement Locale (Fig. 12). One of the Weltevreden specimens is a rim sherd. The decoration consists of triangular comb-stamping running diagonally just below the rim; below this is a horizontal line followed by a blank area 1 cm across, with the same pattern repeated in reverse below. The second Weltevreden sherd again shows diagonal comb-stamp- 138 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT Fig. 11. Pot, rim sherds and decorated sherds from Weltevreden 1, near Plooysburg. 139 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT Fig. 13. Pottery from Mierkraal 1 and Goede Hoop 1. 141 ANN. CAPE PROV. MUS. (NAT. HIST.) YOL. 9 PT 8, DECEMBER 1973 ng, but this time the impressions are rectangular in shape. The specimen from Poortjie is also comb-stamped but in this case the impressions are “D”-shaped. The sherd is too small to tell, but the impression is that the comb-stamping itself may have followed a zig-zag pattern. The non-stamped areas of the sherd seem to have a red ochre burnish but it is very indistinct. It is not clear whether or not these decorated sherds are intrusive in the assemblages but the character of the pottery, apart from the decoration, is not noticably distinct from all the other Type R pottery. The fact that Maggs recovered triangular or “D”-shaped comb-stamped pottery from his excavations would seem to suggest that rare decorated pottery is in true association with Type R pottery assemblages. Temper. Maggs (1971 : 52) has shown that the tempering of Type R pottery consists mainly of grit and has suggested that the rare grass tempered sherds which he found belonged to the “Later Stone Age”. All the pottery collected from Settlement Units by the writer had similar grit tempering (excepting for two sherds from Weltevreden l — see Fig. 11). The excavations in Khartoum 1, Structure L, however revealed a large number of grass tempered sherds (42 out of 78). Apart from the grass tempering there was little to distinguish the two groups of pottery and there was no significant difference in thickness distributions. The two groups are regarded as being associated on the basis of observations made during excavations and have therefore been combined as one sample of pottery from Khartoum 1. It must be concluded that grass tempered pottery cannot automatically be dismissed as “Later Stone Age”. The Type R people may well have made or obtained and used some grass tempered pottery. The colour of the pottery is predominantly buff, varying to a reddish brown or grey. Some sherds (including some from Khartoum 1) seem to have been blackened on the outside by fire. A characteristic of the Type R pottery seems to be its thickness. The thickness distribution and the mean thickness for 5 samples are shown in Fig. 14. The mean thickness ranges from 9,7 mm to 13,3 mm. The thickness distribution of Type R pottery contrasts in remarkable fashion with that of 6 samples of “Later Stone Age” Phase 6 pottery from the Orange River (Fig. 14) where the mean thickness is about 7 mm. Maggs (1971 : 52) has already drawn attention to the fact that Type R pottery is distinct from any other known from this area. This fact can be reiterated here; the Type R pottery is clearly distinct from the highly decorated “Later Stone Age” pottery described by Sampson (1967 b) — similar sherds have been found near OFD 1 (Maggs 1971 : 53) and on Khartoum 1 (unillustrated). Orange Free State Iron Age pottery is distinct from the Type R material (Maggs 1971: 53) and so is pottery today associated with the Tswana and Sotho peoples (Lawton 1967). The Type R pottery seems to be as distinct as the other cultural features and Structures with which it is associated. 2. METAL WORK A total of 3 metal objects was recovered from Khartoum 1, all were found in Structure L, Excavation 3 (see Fig. 6 for the precise locations). The largest and most spectacular piece can best be described as a “spear-head” (Fig. 15 No. 1). The spear-head is 144 mm in length. The front two-thirds of the spear is round in notices but the rear end is square, tapering to a point at the very end. The mid-point of the square section is 5,0 mm. The spear-head is very heavily corroded but the shape and size can still be clearly distinguished. According to Maggs (pers. comm.) similar specimens, but with a flat blade-like point, have been recovered from Iron Age sites to the north. He also remarks, “If the end is not broken off then I suppose it could be a rod-like spear-head without a blade, but this type seems rarer in the ethnographic record.” The end of the spear-head is rounded off and, despite the corrosion, seems to be undamaged. The specimen must therefore be com- plete and represent a round rod-like spear-head. 142 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT POTTERY THICKNESS GOEDE HOOP 1 WELTEVREDEN 1 n 1 2 0 8 5 Fig. 14. Graphs showing the relative thicknesses of 5 Type R pottery samples and 6 “Later Stone Age” Phase 6 samples from the middle Orange River. The second metal object is a thin rod-like fragment (Fig. 15, No. 2). It is 31 mm long and 3,5 mm thick at the mid-point. Not much can be said about it other than mentioning the possi- bility of it being part of the tang of a spear- or arrow-head. The third specimen is a flat piece of metal (Fig. 15, No. 3). It is 36 mm long by 16 mm wide and 3,5 mm thick. The edges are more or less parallel; in cross-section one face is flat while the other has a ridge in the middle. This fragment may well be part of a knife or spear blade, in view of the central ridge. Iron work is as yet unknown at any of the other Settlement Units (except for “a small, shapeless piece of highly corroded iron” which “could well be of modern origin” from OFD 1 — Maggs 1971: 55), but three copper objects were recovered in an excavation while a bangle and part of a band were found on the surface. The rarity of any kind of metal objects on Type R Settlement Units would suggest that they were not common artefacts to the people concerned and that they may have been ob- tained from other groups — probably Iron Age peoples to the north. 3. BONE OBJECTS Three formal bone artefacts were obtained from Khartoum 1. These consist of three bone point fragments (Fig. 15, No. 4). The largest of these is perhaps the most interesting; it is 37 mm in length and its maximum thickness (at the base) is 7 mm. It has been well rounded and smoothed but “facets” created during shaping are clearly visible in the right light. About 143 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 Fig. 15. Small finds from Excavations 1-3 in Structure L on Khartoum 1. la. Bone fragment with incised line; 1. Iron spear-head; 2. Iron fragment; 3. Iron fragment — possibly part of the blade of a knife or spear; 4. Three bone points; 5. Two fragments of bored ostrich egg-shell; 6. Striated shale slab reconstructed from four joining fragments and four other fragments. 144 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT 4 mm above the tip of the point is an incised spiral line which circles the point three times. The incision is not very deep, but is clearly visible on the otherwise smooth surface of the point. The spiral seems to have been deliberately applied rather than being due to wear from some “drilling” activity. It is not clear if the point is complete or not; the base is flat and at right angles to the length of the point and although showing no tooling, does not appear to have been broken naturally, or subsequently to its incorporation in the deposit. The second and third points, on the other hand, both seem to be fragments of longer points. The second specimen is 28 mm in length and 4 mm in maximum thickness. The “facets” created during manufacture are more pronounced on this specimen than on the first and several show striations from grinding or rubbing. There is no sign of any decoration. The third point is clearly only the tip of a much longer artefact; the break at the base is jagged and runs more or less diagonally across the specimen. The fragment is 17,5 mm long and 4,5 mm in thickness. It has been carefully smoothed but some tooling marks are visible. It is not possible to say if these objects were worked with stone or iron tools, but an object with a very sharp edge was clearly needed to cut the spiral into the largest specimen. A fourth bone object was also recovered. This consists of a fragment of bone some 23 mm in length with a very fine incision or rather series of incisions running at right angles to the length (Fig. 15, No. la). The incision appears as a relatively wide cut, but closer examination shows that this cut has been created by a series of very fine incisions all made in more or less the same place. The blade of the cutting instrument was clearly very sharp indeed — it is doubtful if the incisions could have been made with a stone artefact. In view of the fact that the fragment recovered is only a small portion of the original bone it is not possible to guess at the purpose of the incisions. Maggs (1971 : 55) reported the discovery at OFD 1, of a natural bone splinter which had become polished at the tip through use. Nothing comparable has been found at Khartoum 1 but one of the functions of the bone points may have been to do the job done by the natural bone splinter. 4. OSTRICH EGG-SHELL A large number of ostrich egg-shell fragments was recovered from Khartoum 1 Structure L — the excavations yielded 1 068 fragments ranging in size from about 40 mm to 4 mm in length. It is not clear why there should be this large concentration of ostrich egg-shell frag- ments in Structure L (when none of the other excavations produced any) and although one of the activities would undoubtedly have been bead-making, only one complete bead and three broken fragments of semi-bored pieces of shell were recovered (Fig. 15, No. 5). One possible explanation for the ostrich egg-shells is that they represent smashed ostrich egg-shell containers. The fragments from Structure L are generally rather small but the writer recovered several large fragments of ostrich egg-shell which were eroding out from the base of the wall of the large enclosure on the Settlement Unit designated Waterval West 3 (Humphreys 1972). None of the pieces appeared to preserve traces of the container hole but they seemed to be too large to be simple raw material for the manufacture of beads (unless, of course, the ultimate pur- pose had been to fragment them still further). However, the idea of ostrich egg-shell con- tainers is strengthened when we consider the decorated fragments found by Maggs (1971 : 55) on OFD 1, which he considered to be from ostrich egg-shell containers. Maggs (1971 : 55) recovered only four ostrich egg-shell beads from OFD 1, although he found many more fragments of ostrich egg-shell (82 from Trench 3 alone). There is little reason to doubt that ostrich egg-shell beads were made and used by the inhabitants of the Type R Settlement Units. 145 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 5. STONE WORK Fragments of only one stone object (apart from “Later Stone Age” artefacts) were recovered from the excavations at Khartoum 1 ; several other objects of considerable interest were, however, found on the surface of both Khartoum 1 and other Settlement Units. The excavated fragments consist of 8 flat pieces of shale; four of these can be joined together (Fig. 15, No. 6). The four fragments together form a more or less oval shale slab some 55 x53 mm and 3 mm thick; the thickness is very uniform giving the impression that it may have been worked to this thickness. This impression is strengthened by the existence of literally hundreds of fine striations which run in all directions over both surfaces of the slab. All edges of the slab appear to have been snapped off, with the exception of the north-east edge in the sketch, which has been carefully ground down from both sides to form a sharp edge. On the basis of thickness and the existence of the striations, the other four separate fragments all seem to belong to the original object of which the large reconstructed piece may be the major portion; one of the separate fragments has a sharpened edge exactly similar to that on the main piece. If the four separate fragments are considered, as well as the fact that they cannot be joined directly to the main piece and that therefore several more pieces must be missing, then the original size of the shale slab must have been at least twice that of the re- constructed piece. Maggs (1971 : Plate VIIB) also recovered a striated shale slab from OFD 1 but this speci- men was about twice the size of the reconstructed piece from Khartoum, and about 10 mm thick; the shape and edges were irregular with no signs of modification or rounding (Maggs, pers. comm.). It is not clear what the purpose of either of these slabs was. A large shale slab was recovered from the surface of Khartoum 1, Structure L. This slab which has finely incised lines with cross hatching worked in a localised area on it has been described in detail elsewhere (Humphreys and Humphreys, in press). It is not clear what the cultural associations of this object are but in view of the very fine incisions it is possibly more likely to have been made by people with access to sharp metal tools than by pure stone users. The slab itself is 168 x 157 mm and about 25 mm in average thickness; it has two more or less flat surfaces and the edges have been weathered round. In addition to the incised lines, it shows pitting in the centre of one face suggesting use as some type of delicate “anvil”. A second surface find in Structure L was a beautifully made stone pipe which was re- covered in four joining fragments (Fig. 16, No. 1). The length of the pipe is 58 mm and its maximum diameter 44 mm; the diameter of the top of the hole is 26 mm and the bottom 15 mm while the narrow “waist” about three-quarter way down is ±6 mm. The pipe is per- fectly symmetrical except for one side which is flat and unpolished; the impression is that the stone from which it was made was just too small on that side for the pipe to be completely round. The rounded areas were carefully polished to a very smooth finish. The pipe is quite unlike anything described by Walton (1953), but is similar in shape to several described by Baard (1967), notably some from the Bloemfontein and Bethlehem areas. A fragment of what appears to be a second pipe was found on the surface just north-east of Structure C on Khartoum 1 (Fig. 16, No. 2). It is, however, much larger than the first, being about 53 mm in diameter (on the basis of the top which is the only part surviving) while the hole is 16 mm in diameter. The specimen could be part of a bored stone rather than a pipe but the hole runs straight through the stone rather than presenting an hour glass shape which is so common in many bored stones. The outer sides of the “pipe” are also straight suggesting that it was elongated rather than round in cross-section. A curious cylindrical stone was also found on the surface in Structure L (Fig. 16, No. 3). It is 61 mm long and 28 mm in diameter. It is almost perfectly round in section; one end has been snapped off but was later battered in some places while the other end has been rounded 146 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT Fig. 16. Surface finds from Khartoum 1: 1. Stone pipe; 2. Portion of stone pipe; 3. Cylindrical stone. 147 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 off, although it is not symmetrical. The sides seem to have been ground smooth but it was later hammered over most of the outer surface with the result that there is rough pitting over most of the surface. No explanation for this object can be advanced. Upper and lower grindstones abound on Khartoum 1 and all the other Settlements Units studied. No attempt has been made to accumulate detailed data on the occurrence of grindstones but the distribution of lower grindstones on Khartoum 1 has been plotted on the ground plan (Fig. 3). Upper grindstones range from single faceted to multi-faceted speci- mens; most lower grindstones have only one working surface and few of these have been ground to any great depth, the grinding surface appears as a slightly polished indentation. A notable feature on the lower grindstones is the great proportion of them that is broken. It would be difficult to say which grindstones were used by the inhabitants of the Settlement Units and which by “Later Stone Age” peoples, but the number and occurrence of grind- stones on Settlement Units suggests that at least some of them were used by the Type R inhabitants. The excavations on Khartoum 1 yielded a few “Later Stone Age” artefacts. They were never common and seemed to be at random in the deposit. It is important to note that the vast majority of them were undiagnostic flakes and that they were heavily weathered. From the point of view of etat physique they are comparable with the weathered (and undescribed) series collected in Structure L Collection 1 along with the Khartoum “Later Stone Age” Phase 6 surface material described in Appendix 1. The Phase 6 material is unweathered and so on the basis of etat physique and on stratigraphic evidence the “Later Stone Age” Phase 6 in the Riet River Valley must post-date, in part, the occupation of the Khartoum 1 Settlement Unit. 6. COLOURING MATTER Excavations in Khartoum 1, Structure L yielded a total of 41 fragments of red ochre weighing some 88 grams altogether. Ochre was certainly ground by the inhabitants of the Settlement Units for traces of ochre have been found on some grindstones collected by the writer and by Maggs (1971 : 55). 7. FAUNAL REMAINS The excavations in Khartoum 1, Structure L, produced a total of 2 121 bone remains excluding ostrich egg-shell fragments. As mentioned in the descriptions of the actual exca- vations, the bones were in a poor state of preservation and the hard nature of the deposit, which made excavation difficult, added further to the problem of recovering all the faunal remains in relatively good condition. The bones were recovered throughout the deposit, but a large percentage of them were located in a very fragmentary state on the surface of the hard floor layer, giving the impression of having been “tramped” on the floor. Mrs Elizabeth Voigt of the Transvaal Museum undertook an analysis of the faunal remains from Khartoum 1 (Voigt 1972) and as a result two faunal assemblages from Type R Settlement Units are now available for study; the second assemblage is that recovered from OFD 1 by Maggs (1971). 148 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT The list of fauna represented in these assemblages is given below together with their presence at or absence from the two Settlement Units: (The figures quoted represent the mini- mum number individuals present.) OFD 1 Khartoum 1 Cattle 4 Wildebeest/hartebeest . 1 1 Bovid (medium). 1 1 Sheep/goat .... 3 Springbok .... 3 3 Antelope (small) 1 2 Unidentified bovid . . — 3 Non-bo vid .... — 1 Rock rabbit . Viverrid (medium) . Viverrid (small) . Rodent (small) . Elephant shrew . Bird (medium) . Ostrich egg . Lizard . . . . Frog Fish Fresh-water mussel . 1 1 2 2 1 1 several many 2 1 1 6 1 Maggs (1971 : 56) divided the fauna from OFD 1 up in 4 groups in terms of human activity: (1) herding of cattle and small stock, (2) hunting of springbok and larger and smaller antelope, (3) collecting of tortoises and ostrich egg-shell, and (4) exploitation of the riverine fauna in the form of frogs, fish and fresh-water mussels. These four categories seem to provide a useful analysis of the faunal exploitation habits of the Type R peoples and so we may consider them with reference to Khartoum 1. Clearly the most important group missing from Khartoum 1 is the “domesticated” group; there was no direct evidence of domestic animals at Khartoum 1 although the highly fragmentary nature of the faunal remains may well disguise their presence — Voigt did remark that some of the teeth had been compared with comparative goat and sheep material, but were most similar to a comparative springbok dental series. The scond group — hunting of antelope — is well represented at Khartoum 1. There is clear evidence of the exploitation of tortoises and ostrich egg-shell at Khartoum 1 ; the large quantities of the latter have been remarked upon above. The rodent-viverrid group is not represented at Khartoum 1 although there is indirect evidence of their existence (even if not their exploitation) in the fact that one of the bones recovered had been heavily gnawed by a rodent (Voigt 1972); despite the lack of direct evidence it is possible that these animals were exploited by the inhabitants. It is also conceivable, of course, that their presence at OFD 1 is overemphasised by the fact that these burrowing animals may have died in the deposit after the abandonment of the Settlement Unit. Khartoum 1 provides scant evidence of the exploitation of the riverine fauna in the single mollusc recovered; Maggs (1971 : 56) found more extensive evidence of the use of river foods 149 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 and there is no reason to doubt that these exploitation patterns were broadly similar on the two Settlement Units. The writer has noted the occurrence of fresh-water mussels as far away as Driekopseiland and so their distribution does not seem to have been confined to any particular section of the Riet River. Maggs (1971: 56) suggested, on the basis of his OFD 1 evidence, that “The herding of cattle and small stock must have been one of the main activities of the settlement and it seems to have supplied the majority of the protein food.” The structure and nature of the Settlement Units seem to support the first part of Maggs’ contention, but the lack of domestic faunal remains at Khartoum would suggest that he has overemphasised the importance of domestic stock in the diet of the inhabitants; had Khartoum 1 been excavated before, or instead of OFD 1 there would only have been indirect evidence of access to domestic stock in the stone Structures. It is also possible to argue against Maggs’ emphasis on domestic stock as the major source of protein from a purely theoretical point of view. It is well known that in many African societies the possession of domestic stock is a symbol of wealth (Wilson and Thomp- son 1969) and so it would be somewhat anomalous to find a group in a relatively poor economic state relying on domestic stock as their main source or protein. It would seem therefore that we do have direct evidence of domestic stock on Type R Settlement Units and that the inhabitants did sometimes slaughter these animals, but to regard their proportion of occurrence in the two assemblages as being a reflection of their part in the diet of the people would seem to be an over-exaggeration; clearly, by the same token the lack of rodents-viverrids at Khartoum 1 cannot be regarded as evidence that they were not exploited on that Settlement Unit. On the other hand, the occurrence of springbok and other antelope at both Settlement Units may be a reflection not only of their abundance in the area, but also of some aspects of the hunting habits of the people. Voigt (1972) found that “at least two of the seven identified bovid individuals were juveniles,’’ and remarks elsewhere that three very fragmentary metapodials could either have come from a bovid smaller than a springbok or from a juvenile of the springbok-hartebeest sized category. We thus have clear evidence of the exploitation of juvenile as well as adult animals at Khartoum 1. At OFD 1 Maggs (1971 : 55) says, “Detailed work on the age of the animals has not been undertaken but it is very noticeable that the majority of the bovids, both domestic and wild, were juveniles.’’ According to Liversidge (pers. comm.) springbok do not seem to have a definite lambing season, but all other bovids do. Although nothing concrete can be inferred from the springbok juveniles, the existence of other bovid juveniles at both Khartoum 1 and OFD 1 would tend to suggest that at whatever seasons the Settlement Units may have been occupied, there is at least a broad coincidence between the occupations of Khartoum 1 and OFD 1. In the absence of more detail on ages and larger samples it is not possible to say more; certainly a detailed isolation of season, like that achieved by Parkington and Poggenpoel (1971) is not possible. 8. FLORAL REMAINS No direct evidence of floral remains was recovered from Khartoum 1. This was also the case at OFD 1 but here Maggs was able to infer that the row of post-holes probably at one time supported wooden posts, so this does provide some indirect evidence of the use of floral materials. Enough charcoal for C14 dating was recovered from Structure L and this provides evi- dence of the use of floral remains as fuel for fires. The charcoal fragments were far too small for any identification to be attempted. 150 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT DATING The charcoal sample recovered from the excavations in Structure L on Khartoum 1 was submitted to Dr J. C. Vogel of the National Physical Research Laboratory in Pretoria for C14 dating. Dr Vogel has communicated the following date for the sample: Pta 717. Khartoum lc 170±50 B.P. A.D. 1780. This result would seem to suggest that the Settlement Unit Khartoum 1 dates roughly to the end of the eighteenth century or early nineteenth century, and is in good accord with the available historical evidence. The writer has undertaken a detailed study of references to the Riet River area contained in the journals of the early travellers (Humphreys 1972) and several interesting points emerged as a result. The full details cannot be repeated here but a brief summary of some of the points is necessary. Perhaps the most important single event in the recorded history of the Riet River was the Difaqane of the 1820’s. There is a scant information on the actual impact of the Difaqane on the peoples in the Riet area but a consideration of the experiences of travellers who visited the area before and after the upheavals can throw light on some of the changes that took place as a result of the Difaqane. The most important direct information on conditions along the Riet River is to be found in the writings of Andrew Smith (1939) who travelled along the river from roughly the present day Kalkfontein Dam to the confluence of the Riet and Vaal Rivers in 1834-5. Smith seems to have passed very close to the spot where Khartoum 1 and 2 are located, but he makes no reference to people living there. On the basis of Smith’s evidence therefore it would seem that the Settlement Units were abandoned before 1834. Several early travellers make reference to the Riet River area in the pre-Difaqane-days (Burchell in 1811, and Campbell in 1 8 1 3) or the period during which the effects on the Difaqane were being felt in adjacent areas (Moffat in 1823 and Hodgson in 1826). Burchell, Campbell and Hodgson all make reference to the existance of “Stock-keeping” people along the Riet River in those days. Their comments are too vague to allow any clear identifications to be made, but the fact that this way of life existed before the Difaqane, but apparently not after- wards is suggestive. Moffat refers in 1823 to “an immense body of Mantatees . . . coming down the Yellow (Vaal) and Mud (Riet) Rivers . . .” (Schapera 1959 : 99) and this may well reflect part of the upheavels that were to change conditions along the Riet River. A date of about 1780 for one of the Type R Settlement Units would therefore fit in well with the historical evidence. These people clearly had access to stock and may therefore relate in some way to the pre-Difaqane stock-keeping activities mentioned by some of the early travellers. On the other hand, the fact that, on the basis of Smith’s evidence, the Type R Settlement Units had ceased to function by the 1830’s would suggest that this way of life was destroyed by the “immense body of Mantatees” which swept down the Riet River in 1823. A second point of interest in this radiocarbon date lies in the light it throws on the age of the Khartoum “Later Stone Age” Phase 6 assemblage. As has been shown the “Later Stone Age” artefacts lying on the surface of the deposit in Structure L must, on stratigraphic grounds, post-date the underlying deposit and so they must also post-date any C14 date derived from within that deposit. The Khartoum “Later Stone Age” Phase 6 assemblage must therefore be younger than 1704: 50 B.P. This date falls comfortably within the time-range for Phase 6 in the Orange River Basin as defined by Sampson (1970) — the earliest date being 235 ±80 B.P. — and so the two groups of assemblages in the Orange River Basin and in the Riet River Valley are not only typologically similar, but also relate to a similar date. 151 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 Smith (1939) records the existence of San hunter/gatherers along the Riet River in great detail and it is not impossible that these people may have been responsible for the Phase 6 artefacts, seeing that they must also have post-dated the existence of the Type R Settlement Units. DISCUSSION The excavations carried out on Khartoum 1 were limited in size, but some interesting data were nevertheless recovered. One of the aims of the project was to investigate the nature of the cultural material associated with the Settlement Units, with emphasis on the extent to which it reflected the external environment. An analysis of the factors determining the location of Settlement Units within their environment has been presented elsewhere (Humphreys 1972) and so the discussion here will be confined to a few remarks on the cultural material itself. The establishment of a settlement of sufficient permanence to justify the construction of some Structures presupposes the availability of certain basic commodities. A constant water supply is, for example, clearly a priority. This need is, however, something more than the existence of a river a kilometre away: there must be some provision for a water supply within the Settlement Unit. This need was probably fulfilled by the use, certainly, of pots and, possibly, of ostrich egg-shell containers. The availability of large reliable water containers must have gone hand in hand with the establishment of permanent stone-built Settlement Units. Pottery seems to have appeared in the “Later Stone Age” around 1200 (Sampson 1970) and a knowledge of pot-making was probably a prerequisite for the development of permanent Settlement Units. Although not “cultural” in the strict sense of the word, a constant food supply would also have been a necessity and this, too, is reflected in the remains recovered from Khartoum 1. The four groups into which the faunal remains were divided above are a good reflection of human activity within the prevailing environmental situation: The inhabitants clearly kept or had access to cattle and sheep and/or goats and their remains are reflected in the faunal assemblage from OFD 1. The existence of various Structures on the Settlement Units which could be interpreted as kraals are a further reflection of the herding of stock (see below). Wild animals such as springbok and others were hunted as a meat supply. Smaller animals such as tortoises were collected in the veld as an additional food supply; ostrich eggs, in addition to the use of the shells, could have been a good supply of food. The riverine fauna, in the form of fish, frogs and mussels, may also have been exploited, but there is but scant evidence of this. No evidence of plant or vegetable remains was found but we may assume that the floral resources of the area were also exploited. The existence of bone points, incised bone and ostrich egg-shell testifies to the further use of some of the “food waste products”. The rare metal objects, as well as the red ochre, are evidence of contact, through trade, with peoples living other ways of life. The metal objects are likely to have been obtained from Iron Age peoples to the north and so the existence of these objects of trade on Settlement Units may be interpreted as a reflection of the cultural environment with which the Type R peoples came into contact. The limited nature of the excavations and the low incidence of cultural objects in the deposits have made it impossible to define the parameters of any artefact making habits in detail and such definition will have to await much more extensive excavation. The material yielded nevertheless does give some insight into the relationship of the peoples to their en- vironment. 152 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT A second aim of the excavations on Khartoum 1 was to throw some light on the function of the various Structures on a Settlement Unit, and so we must now consider what the results of the excavations tell us. During the excavations soil samples were collected from those levels thought to have been associated with the occupation of the Settlement Unit; the hard floor layer in Structure L and the levels on which the base of the walling in the excavated Structures was built. As a control, samples were collected from unexcavated Structures and the open ground surrounding the Settlement Unit. These soil samples were then tested for the percentage P205 in them in the hope that differences in the P205 level would give some clue as to the different activities carried on in the various Structures. Full details on the samples and the analysis are given in Appendix 2. We may summarise the result by saying that there was not a great variation in P205 level from sample to sample but where differences did exist, the higher values were usually from within Structures or from occupation levels. Maggs (1971: 61) found a similar situation on OFD 1 but the samples from within the enclosures showed a very much higher level of P205 than did any of those from Khartoum 1. This difference in P205 level may be due in some measure to different techniques of analysis rather than being related to variations in the intensity of occu- pation between the two Settlement Units. The relatively higher levels of P2Os within the Structures on Khartoum 1 suggest that the areas in the enclosures were subjected to more intensive habitation than the areas outside. A soil sample from next to the wall Structure F did not show a higher P2Or> level than the sur- rounding area and so there is no evidence from this point of view to show a great concentration of activity adjacent to this Structure. Two of the small enclosures were excavated (one each from Khartoum 1 and 2) but neither excavation produced any evidence to show what activities might have been carried out in the enclosures. Apart from the virtual absence of cultural material it was not possible even to identify the floor on which the occupation would have taken place; its position could only be established on the basis of the level on which the wall appeared to have been built. Maggs (1971 : 49) excavated one of the small enclosures on OFD 1, and he too was unable to find any occupation floor; he did, however, find many sherds and some bone and tentatively suggested that, “the material suggests that domestic activity connected with food took place in the enclosure”. There is no evidence to support such an interpretation for the two Khartoum enclosures; on the contrary, the very absence of cultural material would suggest that what- ever activities were carried out, they were not such as to produce much debitage. The fact that small enclosures on some Settlement Units have produced some cultural remains while those on others have been virtually sterile shows that the enclosures may have been used for a variety of activities and that any one specific function cannot be determined or laid down to explain their existence. The wide range in diameters of these enclosures seems to support this view. The large central enclosure, however, presents a different picture; both on Khartoum 1 and OFD 1 it produced most of the cultural material. From the range of materials found in the large enclosure, there seems to be little reason to doubt that it represented a focus of attention to the inhabitants: the existence of pottery fragments, tools, ornaments and food remains, as well as a concentration of lower grindstones testifies to a large range of activities having been carried out within the enclosure. Maggs (1971 : 48) found a series of post-holes running within the large enclosure of OFD 1 and suggests that, “on present evidence the holes seem to have held uprights to form a wall”. Although the entire large enclosure on Khartoum 1 was not excavated it does seem highly unlikely that a similar line of post-holes should exist — the soil overlying bedrock when the Structures were built was not deep enough to allow the digging of holes to support uprights. If the purpose of the OFD 1 post-holes was to form some type of screen then a comparable structure on Khartoum 1 would probably have had to have been 153 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 8, DECEMBER 1973 propped up rather than planted in the ground. However, the possibility of the existence of some type of “partition” in the large enclosure would seem to provide further evidence to suggest that the enclosure formed a focal point in the lives of the inhabitants. The central position of the large enclosure, of course, also supports this idea. Maggs (1971 : 42) remarked about the large enclosures that, “It is difficult to think of any purpose for them other than livestock pens”. This may well have been one of their functions but clearly they were used for a variety of other activities as well. Recently David (1971) drew attention to the multitude of problems involved in trying to reconstruct the social and domestic activities undertaken in a settlement in terms of archaeologi- cal remains. He tested several of the approaches in current use in archaeology and found them generally to be of limited value and concluded that, “The difficulties are obvious enough and . . . sufficient to dissuade the most sanguine archaeologist from common-sensical re- constructions of social organisation. But the balance must not be allowed to swing so far as to inhibit all attempts at inference. Rigorous methods to test propositions about prehistoric social life exist . . . though it should not be forgotten that such methods cannot prove an hypothesis, but only fail to reject it. The most urgent requirement at present is for detailed case studies that mediate between the ethnographers’ structural models and the technologists’ model of structures.” In absence of well-tested hypotheses and in view of the obvious complexity of the social patterns which must have underlain the occupation of the Settlement Units it seems safest to leave speculation on the social patterns prevailing on one of these Settlement Units until much more detailed evidence is available. CONCLUSION The purpose of this paper has been to report on excavations carried out on a Type R Settlement Unit. The results of the excavations are limited in scope, but they do serve to compliment, and in some cases amplify, the results obtained by Maggs (1971) on OFD 1. The Type R Settlement Units represent a relatively unique occurrence in the archaeological record and a full understanding of how they functioned and what their relationship was to “hunter- gatherers” or “farmers” will have to await the large-scale excavation of several of the Settle- ment Units — an undertaking which was beyond the scope of the project described in full in Humphreys (1972), of which these excavations were a part. ACKNOWLEDGEMENTS These excavations were undertaken during 1972 as part of a project on the Type R Settlements initiated by the Alexander McGregor Memorial Museum, Kimberley, and I am grateful to the director of the museum, Dr Richard Liversidge for his support and assistance. Mr I. D. Strauss of the farm Langhoek kindly allowed access to the Settlement Units. My thanks are also due to the following for their assistance: Mr O. E. Bergh of the Kimberley Phosphate Co. (Ltd) (KIMFOS) who undertook the P205 analyses; Mrs Elizabeth Voigt of the Transvaal Museum who analysed the faunal remains; Dr J. C. Vogel of the National Physical Research Laboratory in Pretoria who did the C14 dating. Thanks are also due to Mr Tim Maggs of the Natal Museum who introduced me to the problems of the Riet River Settlements and who commented on various aspects of the in- vestigation. This paper is an extract from an M.A. Thesis submitted in the Department of Archaeology at the University of Cape Town and I am most grateful to Mr Hilary Deacon, Mrs Janette Deacon and Mr John Parkington for their comments and assistance. Finally my thanks are due to my wife, Carol, who took part in all the work carried out at Khartoum. 154 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT APPENDIX 1 The analysis of the fresh series of artefacts recovered from with Structure L on Khartoum 1 followed the scheme presented in Humphreys and Maggs (1970) derived from Sampson’s work in the Orange River area. Details of the composition of the assemblage are given in Table 1. The assemblage shows a close resemblance to the one recovered from Oudefontein (Humphreys and Maggs 1970; Humphreys 1972) and is probably related to the Phase 6 identified by Sampson (1970) in the Orange River area. Table 1 ANALYSIS OF “LATER STONE AGE” ASSEMBLAGE FROM KHARTOUM LITHIC Shaped : End scrapers Side and end scrapers . Small end scrapers Core hammers . Convex Scrapers Adzes Outils ecailles Small convex scrapers . Backed blades . Notched scrapers Burins Borers Hollow scrapers Utilised : Utilised flakes (whole) . Utilised flakes (broken) Grindstones (upper) (lower) Grooved stones . Hammer stones . Slabs Waste: Waste flakes Cores Chips and Chunks . NON-LITHIC Pottery Ostrich egg-shell beads . Specularite 12 18 19 11 25 4 3 17 9 5 2 1 126 17,4% 75 69 5 3 1 3 2 158 22,4% 149 19 246 414 58,0% 19 19 2,8% TOTAL 717 155 ANN. CAPE PROV. MUS. (NAT. HIST.) YOL. 9 PT 8, DECEMBER 1973 APPENDIX 2 SOIL ANALYSIS A series of 13 soil samples was collected from Khartoum 1 and 2 for analysis. The primary object of the proposed analysis was to determine whether or not there was a higher P205 value for samples taken from within the enclosures than for samples from the areas surrounding the Structures. Mr O. E. Bergh of the Kimberley Phosphate Co. (Pty) Ltd. (KIMFOS) kindly undertook to have the samples processed in his laboratory. The samples were numbered at random and submitted to Mr Bergh who reported as follows: Sample No. %P %P/). 1 0,05 0,12 2 0,075 0,17 3 0,08 0,18 4 0,062 0,14 5 0,075 0,17 6 0,07 0,16 7 0,038 0,087 8 0,062 0,14 9 0,045 0,103 10 0,04 0,092 11 0,055 0,13 12 0,038 0,087 13 0,055 0,13 The precise locations of the soil samples on Khartoum 1 can be seen in Fig. 3 for those not obtained within excavations, and in Fig. 6, 7 and 8 for those from within excavations, or associated with excavated areas. The positions may be summarised as follows: Sample 1 Khartoum 2 2 Khartoum 2 3 Khartoum 1 4 Khartoum 1 5 Khartoum 1 6 Khartoum 1 7 Khartoum 1 8 Khartoum 1 9 Khartoum 1 10 Khartoum 1 11 Khartoum 1 12 Khartoum 1 13 Khartoum 1 Structure A, Exc 1. Just outside wall below excavation. Structure A, Exc 1. At level of base of wall. Structure J, Exc 1. At level of base of wall. Structure J, Exc 1. Just outside wall below exvavation. Structure L, Exc 1. Within hard floor layer. Structure L, Exc 1. Just above hard floor layer. Structure L, Exc 1. Just outside wall. Structure B. Structure D. Open area just north of Datum Dl. Structure G. Adjacent to Structure F. Open area between Structures E, F, G and L. A glance at the results obtained in the soil analysis will show that there is very little difference in the P205 value between individual samples. However, if the differences that do exist are studied in detail, some correlation between P205 value and position on the Settlement Units can be seen. The following groupings show this correlation: Position on Settlement Unit Area outside Structures: On suspected living floors: Within other Structures: %P205 0,12; 0,14; 0,087; 0,092; 0,087; 0,13 0,17; 0,18; 0,17; 0,16 0,14; 0,103; 0,13 156 HUMPHREYS: EXCAVATIONS CARRIED OUT ON A TYPE R SETTLEMENT UNIT From this grouping it is clear that the highest values (0, 1 6 — 0, 1 8) were recorded for samples from established or suspected living floors or levels. Moderately high levels were recorded from within the other Structures samples (0,103 — 0,14). The samples from the “open” areas have, however, yielded a wide range of values but this can be broken into two groups: the first consists of values ranging from 0,087 — 0,092 (which are the lowest recorded), while the second group ranges from 0,12 — 0,14 and overlaps the values obtained from samples from un- excavated Structures. It is interesting to note, however, that all the “high” values in the second group were obtained from samples that were collected immediately down-slope from Structures while the “low” values were from samples not directly in line with any Structure. A possible explanation for these “high” values may therefore be found in the fact that the areas imme- diately below Structures may have an abnormally high level of P205 (compared to the other areas) because of percolation down from the Structure level just above. Whatever the explanation for the anomalies in the samples from the areas outside the Structures, there can be no denying that the highest values obtained are from living levels. This relatively high level may well be a reflection of the fact that the Structures concerned were inhabited by man or animals. We may therefore conclude that there is some evidence to show that the Structures were probably inhabited, but that there was apparently no difference in the intensity of habitation in either the large or small enclosures, if the interpretation of the values of P205 can be stretched to this extent. REFERENCES Baard, E. 1967. Dagga Stone Pipes in the Collection of the National Museum. Res. Nat. Mus., 2(7) :2 1 6 — 33. David, N. 1971. The Fulani Compound and the Archaeologist. World Archaeology , 3(2) : 1 11 — 131. Humphreys, A. J. B. and Maggs, T. M. O’C. 1970. Further Graves and Cultural Material form the Banks of the Riet River. S. Afr. Archaeol. Bull. 25(99 — - 1 00) : 1 1 6 — 126. Humphreys, A. J. B. 1972. The Type R Settlements in the Context of the Later Prehistory and Early History of the Riet River Valley. Unpublished M.A. thesis. University of Cape Town. Humphreys, A. J. B. and Humphreys, C. in press. Note on an Engraved Shale Slab from the Jacobsdal District, Orange Free State. S.Afr. Archaeol. Bull. Lawton, A. 1967. Bantu Pottery of Southern Africa. Ann. S.Afr. Mus., 49(1). Maggs, T. M. O’C. 1971. Pastoral Settlements on the Riet River. S.Afr. Archaeol. Bull. 26(10 1):37 — 63. Parkington. J. E. and Poggenpoel, C. 1971. Excavations at De Hangen, 1968. S.Afr. Archaeol. Bull. 26(101): 3—36. Piaget, J. E. H. 1963. Coarse Fraction Mineralogy and Granulometry of Selected Soils of the Western Orange Free State. Unpublished D.Sc. Thesis. University of the Orange Free State. Sampson, C. G. 1967. Excavations at Zaayfontein Shelter, Norvalspont, Northern Cape. Res. Nat. Mus. Bloetn. 2(4) :41 — 124. Sampson, C. G. 1970. The Smithfield Industrial Complex: Further Field Results. Nat. Mus. Bloem. Mem. 5. Schapera, I. 1951. Apprenticeship at Kuruman . . . London, Chatto & Windus. Smith, A. 1939. The Diary of Dr. Andrew Smith, ed. Kirby, P.R. Cape Town, Van Riebeeck Society. Voigt, E. A. 1972. The Riet River Ruins Fauna (Khartoum 1). in Humphreys, A. J. B., 1972. Walton, J. 1953. The Dagga Pipes of Southern Africa. Res. Nat. Mus. Bloem. 1(4) :85 — 113. Wilson, M. and Thompson, L. (ed) 1969. The Oxford History of South Africa. Oxford. 157 i PRINTED BY CAPE & TRANSVAAL PRINTERS LTD CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. {nat. Hist.) i. VOLUME 9 • PART 9 13th DECEMBER 1973 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA The exploitation of shellfish by coastal tribesmen of the Transkei by E. H. BIGALKE East London Museum The consumption of shellfish by non-Bantu groups who lived along the south and south- east coast of South Africa has been extensively documented in archaeological literature. Travellers’ accounts and early ethnographic descriptions of the Bantu-speaking peoples of the present-day Transkei and Ciskei with few exceptions stress the absence of a tradition of fishing (Alberti, 1968, p. 25; Barrow, 1801, p. 211; Brownlee, 1827, 2, p. 216; Kropf, 1889, p. 102, Vanderkemp, 1803, p. 436), often ascribing it to a prohibition against eating fish. Only two of the nineteenth-century writers specifically mention the use of shellfish as food, namely Lichtenstein (1928, 1, p. 335) and Kay (1833, p. 354), the latter in reference to the Tshezi, living near the mouth of the Umtata River. In the twentieth century, only Hunter (1936, p. 96), in her monograph on the Mpondo, devotes any attention (two paragraphs) to fishing and shellfish gathering. Holt’s recent account of the Tshezi, who are culturally assimilated to the Bomvana, makes passing mention of the use of fish and shellfish. Thus, there appears to be no detailed treatment of this subject in the published literature on the Southern Nguni. The present study was undertaken with two objectives in view. One was to fill the gap in the ethnography of the Southern Nguni living in the coastal districts of the Transkei; the other was to provide comparative data for an archaeologist colleague engaged in the analysis of shell material from prehistoric coastal middens. It presents data collected between 1969 and 1972 in the Transkeian districts of Kentani, Willowvale, Elliotdale, Mqanduli, Ngqeleni, Lusikisiki and Bizana. The groups studied in the first two districts were Xhosa, in the third and fourth, Bomvana and in the last three, Mpondo. “Shellfish” in this context is taken to include crayfish, red bait and barnacles, in addition to mollusca proper. The material culture of the tribes making up the Southern Nguni is basically very similar. Techniques for the collection, preparation and uses of shellfish all along the Transkei coast are sufficiently alike for the three tribes to be treated as one in the present context but tribal differences will be indicated where they occur. Collecting (Xhosa and Bomvana: ukuxeza; Mpondo: ukuxoza) is done almost exclu- sively by women. At all the places visited, the range of females collecting included anyone from small girls to grey-headed old women. At least one pregnant woman was observed collecting shellfish. Only one adult male was seen gathering them, at Mbotyi. While a few small boys were occasionally seen accompanying parties of women shellfish collectors, their contributions to the day’s pickings were small. Usually they merely played on the rocks. Women and girls may work individually or in parties. Such parties vary in size between one 159 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 and as many as fifteen. The largest party found to be from one household consisted of three married women, two of them the wives of brothers, the third being the wife of a husband’s brother’s son. At Mzamba, in eastern Pondoland, married women were seen catching crayfish but elsewhere among the Mpondo (at Mbotyi) and at all points westward, only youths catch crayfish. Line fishing is the concern of men and youths alone. This practice, seldom observed during the research, appears to be restricted and was not studied. While women and girls on their way to the collecting areas wear the dress appropriate to their age and status, clothing is changed before the work commences. Old clothing brought along for the purpose is put on in the shelter of rocks or nearby bushes. In contrast to the normal custom that married women keep their breasts and heads covered at all times, especially when in their husbands’ homesteads, collectors are often seen on the shore with breasts exposed and heads uncovered, Prepubertal and adolescent girls wear only some form of skirt, often very brief. Neither women nor girls use footwear while collecting. When the work has been completed they change back into normal dress before returning home. As the details of types of shellfish collected will show, all the species occur in the balanoid and cochlear zones of the inter-tidal area. Coastal Africans are aware of tidal movements and know that spring tides occur at full moon and new moon. Informants repeatedly said that these were the best times for collecting and that it was then possible to go far out on the rocks and obtain the largest shellfish. From the fact that collectors were observed arriving at the shore later on successive days following the spring tide at full moon, it appears that they are aware also of the time difference between tides each day. According to Hunter (1936, 96), knowledge of tidal alternation predated the arrival of the white man. Information about the regularity with which women go to the rocks to remove shellfish was difficult to obtain. Informants said that not all women from areas close to the sea went to collect each day when the weather and tide were favourable. It depended on whether they and their families particularly liked or were hungry for shellfish. A young woman at Ntlonyane said that she went to collect perhaps once a month, both because she was lazy and she feared the sea. At Ntlonyane and Mbotyi, during a continuous spell of good weather at spring low tide, the same women were encountered on the rocks for three days running. Informants said that they would sometimes go to fetch shellfish during other phases of the moon, depending on the demand for this kind of food. At all places visited, informants said that little collecting was done in winter because the shellfish were seldom accessible and their condition not as good as at other times of year, particularly summer. They reported that the flavour was better in summer than in winter but that shellfish were eaten at all seasons. Although the Transkei is within the red tide zone, informants claimed that mussels were not known to cause illness at any time of the year; in fact, they had never been known to cause illness. The Africans conceded that people sometimes became ill after eating shellfish but attributed this to over-indulgence, especially in the case of children. It is possible that these coastal Africans do not associate the consumption of certain foods with subsequent illness. This is claimed to be the case with plants used as food or medicine (Dr. E. Rose, pers. comm.). Irrespective of the time of low tide, all collecting appears to be confined to the morning and the early afternoon, though at Mbotyi an informant claimed that people did sometimes collect during the afternoon. In no case was any collecting observed before 8 a.m. or after 2.30 p.m. When shellfish were collected from the cochlear zone, as observed at Kobonqaba, the time spent on the rocks did not exceed two hours. The party of women who were kept under obser- vation arrived on the rocks at 8.10 a.m. and stopped collecting at 10.10 a.m., when they had decided that the tide was turning and that it would no longer be possible to continue. 160 BIGALKE : THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI The mode of collecting can be divided into two phases, gradually merging into each other and dictated by the falling of the tide. The first phase saw the women picking what they could from the rocks in the upper balanoid zone as the water receded. Though the work of removing shellfish was at all times done quickly, this first phase was not characterised by the urgency of the one which followed. During the second phase, the women moved onto the exposed rocks furthest out; such rocks were frequently still in the surf. Here the work was done with all possible speed, the women constantly throwing glances at the sea and withdrawing to higher rocks when they noticed a dangerous wave approaching. At Shixini, where women spent three hours collecting, their movements on the rocks followed the same pattern. The equipment for collecting shellfish is simple. Those molluscs which are easy to dis- lodge (e.g. Oxystele, Turbo , Charonia , Thais and Burnupena) are removed from the rocks or the seaweed by hand. Patellidae, Haliotis and clumps of Perna perna are removed from the rocks by means of a narrow, flat iron or steel bar ( ulugxa ; at Mbotyi, also umbutu) resembling a tyre lever, held in the right hand (no left-handers were observed), while the left hand grasps the dislodged shell. Patellidae are removed singly but Perna in clusters, together with any sea- weed attached. The shells are then placed in a receptacle which is held in the hand or stands nearby, or in a sack tied round the waist and forming a kind of pouch, which is emptied into a basin or billycan from time to time. The work is done quickly, particularly during the second phase, the main object being to gather as much from the fringe of the surf as time permits. Shell damage occurs, as when the edges of the shells are chipped or broken. A variety of con- tainers is used, including tin basins, small tin billycans ( iibekile ), plastic buckets, traditional grain baskets ( iingobozi ) and small sacks, either of jute or polythene. These receptacles may be held in the hand or tied to a belt, which permits easier movement when collecting. When the receptacle has been filled — the rate at which this is done depends on whether collecting is in the first or second phase — it is taken to the rocks a little distance from the immediate collecting area and emptied. Each collector makes her own pile, with the exception that children add their pickings to their mothers’ piles. Shortly before leaving the rocks, the women sort perfunctorily through their piles of shellfish, picking off' bits of seaweed and discarding mussels they consider too small to be worthwhile taking home. There are few discards. To remove sea sand, sea water is thrown over the pile a few times. Afterwards all the shellfish are packed into the collector’s receptacle and placed on the head, to be carried home. Coastal informants remarked that they took the consumption of shellfish for granted but that people from further inland were unfamiliar with this food and expressed revulsion when offered it. As regards the distance travelled to collect shellfish, at Kobonqaba, where homesteads are not built closer than two miles from the coast, one party of women travelled three miles to collect shellfish. At Shixini, women were encountered returning from shellfish collecting at a spot approximately two miles from the coast. At Hluleka homesteads two miles from the coast had middens containing shells or had shells scattered about their gardens. At Dwessa and Mbotyi informants claimed that people living about five or six miles distant from the coast came there to obtain shellfish. An Mbotyi informant said that these people brought donkeys to carry back what they had collected and would even stay overnight in order to be able to take advantage of a morning low tide. This information seems to bear out Hunter’s observation that six or seven miles inland is the limit beyond which people do not go fishing or shell-fishing. Allowing for the differential abundances of species, depending on natural distribution, a broadly similar range of shellfish is eaten all along the coast. The following species were present in modern middens or in meal remains (as indicated below) at the following places: 161 ANN. CAPE PROV. MLS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 Species list: x = present; — = absent Locality Species Xora Mouth Kobo- nqaba Shixini Dwessa Ntlo- nyane Hole- in-the- Wall Hluleka Mbotyi Mzamba Panulirus sp. — — — — — — X X X Balanus sp. X X X X X X X X — Dinoplax sp. — X X X X X X X X Fissurella natalensis X — X — X — X X X Haliotis midae — X X X X — X X X H. spadicea — X X X X X X X — Turbo coronatus X — — — X — X — X Turbo natalensis X — X — X X — — — Turbo sarmaticus — X X X X X — — X Charonia lampas pustulata — X X X X X X X X urnupena papyracea lagenaria X — X — X X — X X Burnupena sp. — — X — X — — — - — Oxystele sinensis — X X — X — — — — Oxystele tabularis Oxystele tigrina — — X X X — — X — Oxystele variegata — — — — X X — — — Oxystele sp. — X — X X — • — — — Thais capensis X — — — X X X X — Thais sp. — — X — — — — — — Nerita albicilla X — — — — — — — — Nerita plicata — — — X — — — — — Nerita sp. — — — — X — — — — Monodonta australis X — — — — — — — — Patella barbara — — X — X — — X — Patella cochlear — X X X X X X X X Patella granularis X X X — X — — X X 162 BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI Species Locality Xora Mouth Kobo- nqaba Shixini Dwessa Ntlo- nyane Hole- in-the- Wall Hluleka Mbotyi Mzamba Patella longicosta X X X — X — X X — Patella miniata* — — X — X — X X — Patella oculus — — X — X — — X — Patella tabularis X — — — X — X X — Patella concolorf X — X — X X X X X Patella sp. — — — X — X — — — ■ Siphonaria capensis — — — — X — — X — Helcion pectunculus — — — — X — — X — Helcion pruinosis — — ■ — — X — — X — Cellana capensis X — X — X X — X — Perna perna X X X X X X X X X Drupa sp. — — — — — — — X — Septifer bilocularis — — — — X — — X — Crassostrea cucullata — — X — X — — X X Crassostrea — — X X X X X X X margaritacae Octopus — X X X X X X X X Pyura stolonifera X X X X X X X X X Echinodea — — — — — — — X X * P. miniata includes P. miniata miniata and P. miniata sanguinans f Formerly P. variabilis Although women were seen to eat uncooked shellfish such as limpets ( Patellidae and Cellana) and redbait ( Pyura ) while collecting on the rocks, the bulk of what was collected was taken home in shells to be prepared there. Only at Mbotyi was a woman seen taking the meat of rock oysters ( Margaritacae cucullata) and leaving the shells on the rocks. The analysis of empty shells resulting from the meals indicates that some people carry rock oysters in their shells to the homesteads. Pyura stolonifera (redbait) is always removed from its casing at the beach. The basic method of preparation of shellfish does not vary. A three-legged cast-iron pot is filled with a mixture of shellfish and a small quantity of water is poured over it. Some informants specified that seawater should be used because boiling with fresh water makes the flesh taste insipid. After the water has boiled and the foam risen, the pot is removed from the fire, though some women leave it to boil for about five minutes more. It was said that mussel shells open in boiling water, thus making the removal of the meat easy, and that those 163 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 shellfish with spiral shells come out easily when boiled. Informants said that this boiled meat, after washing, could be fried in vegetable oil or animal fat. At Hole-in-the-Wall (Mtonjane) the frying method was said to have been learned from white people. Except at Ntlonyane and Mzamba, all informants questioned about the method of consuming shellfish said that this food is eaten alone, never mixed with other foodstuffs. This was borne out on a number of field expeditions, when men, women and children were observed eating shellfish without accompaniments, either sitting down to a dish of this food alone or merely taking a few boiled shellfish to eat as they moved about the homestead. At Ntlonyane an informant said that the liquid from boiled shellfish would be saved and mixed with boiled maize. At Mzamba, an Mpondo area in close proximity to the Natal border, a School informant described a mixture of fried shellfish and onions, added to iphuthu (stiff porridge). The addition of shellfish in Natal to other food was confirmed at Mbotyi by a visiting Natal Nguni woman from a coastal area. At Mbotyi a white and a black informant both independently reported the eating of sea urchins; the contents of the shell are eaten, raw. No accurate indication of the proportion of shellfish meat to other types of food in the diet of coastal people is available. Information was, however, obtained about the quantities of shellfish gathered and the approximate weight of their contents consumed. This was done by weighing the takings of a number of women as they left the beaches at Ntlonyane and Mbotyi, and subsequently weighing the empty shells after the completion of the meal. The figures are as follows (weights in kilograms) (See bottom of following page). These figures should be taken as a guide only; the weights of shellfish meat given above are not claimed to be absolutely accurate. When the raw shellfish is packed into a container for transfer to a homestead, it contains a certain amount of sea sand and there are usually bits of seaweed attached to many of the shells, particularly Patellidae gathered in the cochlear zone. The empty shells returned after meals were often found to include a small proportion of un- eaten shellfish (mussels too small to open) and Neritidae which, though sometimes collected, are not eaten. Not all operculi of Turbo shells are returned and not all chiton plates are in- cluded among the empty shells. Octopus and crayfish are represented only in the volume of shells and raw flesh but leave no remains in the samples of empty shells. Thus the deduction of the weight of empty shells from the weight of total takings does not give the true figure for meat content. Further, Pyura flesh is removed from its casing on the beach, thus weights of meat given do not correspond exactly with the numbers of shells in each set of meal remains. However, it would be difficult to achieve more accurate results without exercising close control over the actual eating of the shellfish and causing chaos in the homesteads. No reason can be offered for the high average weight of meat per person in the meals numbered 2, 24, 25, 26 and 27, except that informants may not have given correct information about the numbers of people sharing their meal. Nevertheless, the figures indicate that the quantity of shellfish meat available in a number of routine collections at spring tide would contribute a significant amount of animal protein to the homestead diet. Disposal of empty shells after a meal is effected in a number of ways. They may all be thrown onto one heap a little removed from the main outdoor living area where they will not get underfoot and possibly cause cuts. In the hilly coastal areas such middens are usually on sloping ground. One large modern midden at Ntlonyane was sited at the head of a donga and was thus subject to water action during heavy rains. Voigt’s paper (in preparation) deals with an excavation and analysis of this midden. At Hole-in-the-Wall one homestead threw all its empty shells into a donga so that they would be removed by storm water. At other places, the remains of each shellfish-collecting expedition are thrown in a separate small heap, sometimes in the same area, so that there is a series of small heaps in a midden area. Another method is to scatter shells thinly over the garden usually found adjacent to every coastal homestead, in order to enrich the soil with calcium when the shells decompose. 164 BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI The coastal people considered shellfish a substitute for meat, a point of view explicitly expressed by woman informants at Kobonqaba, Shixini, Dwessa and Mbotyi. A Shixini informant said that there is never enough shellfish available for it to be considered a replace- ment for meat. All people asked agreed that shellfish is a good food; this was frequently ex- pressed as “ isondeza igazV\ meaning, figuratively, that it acts as a tonic. The collection and eating of shellfish is remarkably free of restrictions imposed by cus- tom and belief. At Kobonqaba a woman said she no longer went to collect since she had had twins because such people should not go to the sea. This was qualified by another informant in the same area who said that women who had borne twins could go collecting, provided that they first washed themselves with sea water on arrival at the shore. Even if the sea were rough, it would then become calm. At Mtonjane informants related the belief than one’s luck is likely to be poor if one has had a long break from collecting; to counter this, women wash their faces in seawater before beginning to collect. Only two categories of people are customarily forbidden to eat certain species of shellfish in general. Abakhwetha (young men in seclusion after circumcision) are prohibited from eating Meal No. Total catch Empty weight Approx. weight meat Number people sharing Ave. weight meat per person Kobonqaba I 6,75 4,30 2,45 7 ,350 2 4,45 1,84 2,61 2 1,300 Shixini 3 ,65 ,45 ,20 2 ,100 4 1,95 1,15 ,80 6 ,166 5 2,35 2,23 ,12 5 ,240 6 ,89 ,55 ,34 4 ,085 7 5,30 4,20 1,10 3 ,336 Ntlonyane 8 2,78 2,06 ,72 6 ,120 9 4,31 2,05 2,26 4 ,565 10 5,556 2,825 2,731 5 ,546 11 2,891 1,470 1,421 4 ,355 12 2,268 1,375 ,839 4 ,223 13 4,224 2,400 1,824 4 ,456 14 3,061 1,520 1,547 4 ,385 15 1,927 ,780 1,147 4 ,286 16 1,757 ,560 1,197 3 ,399 17 1,672 ,700 ,972 2 ,486 18 . 4,337 2,260 2,077 4 ,517 19 4,195 2,25 1,945 3 ,648 20 2,551 ,950 1,601 2 ,800 21 3,799 1,650 2,149 4 ,537 22 3,175 1,410 1,765 7 ,252 23 2,324 ,900 1,424 4 ,356 Mbotyi 24 17 9,1 7,900 6 1,310 25 17,9 9,6 8,300 10 ,830 26 17,1 9,6 7,500 5 1,500 27 12,85 7,200 5,650 6 ,940 28 2 1,200 ,800 6 ,133 29 1,65 1,54 ,600 4 ,150 30 3,90 2.17 1,87 6 .310 31 6 2,05 3,95 4 ,870 32 6,10 4,21 1,91 7 ,272 33 1,50 ,80 ,70 4 ,175 165 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 shellfish. Informants explained this prohibition by saying that since the abakhwetha should have no contact with post-pubertal females, they must also avoid the shellfish collected by them. In fact, however, such women do prepare and, often, bring their food to them. At Mtonjane it was said that newly-married women are not supposed to eat shellfish. Married women, except in Pondoland, are not permitted to eat oysters ( iimbatyisa ) or crayfish ( ikolo - fish, sometimes isikhuphathi) because they are believed to have aphrodisiac properties. At Ntlonyane, however, a post-menopausal woman whose shellfish harvest was found to contain oysters said that old women could eat them. Seafoods or sea objects can be used as tools, ornaments and as medicine. A woman at Mbotyi was using limpet shells as spoons for feeding babies and as pot-scrapers. When tin caps are not used on the tops of the conical hut roofs, oyster shells are widely used to keep the earth cap in place. The shells of Nerita textilis are made into armbands and necklets by the Xhosa and Bomvana. Among the Bomvana and Mpondo, the tentacles of the octopus are used by young men as a love potion. The stomach (liver?) of the crayfish (considered by informants to be the brains) is used at Mbotyi to calm troublesome, crying children. This part is boiled in water and the child made to drink the water. At Mzamba it was claimed that a barren woman who takes to a diet containing a large proportion of shellfish will soon con- ceive. Crayfish eggs, mixed with herbs, are fed to cows, ewes and hens to promote fertility. Informants at Hluleka and Mzamba said that umopu, probably the “sea hare” ( Aplysia sp.), is used as a medicine to stop the vomiting of blood. This creature is used because, when dis- turbed, it emits what informants called “blood” (possibly a substance used protectively to obscure the animal’s movements); its use seems to fall into the category of sympathetic magic. Vomiting of blood is said to be stopped also by using the ground spines of the sea urchin, mixed with other medicines. Cuttlefish, scraped to a powder, is used in cases of sore eyes in humans and animals. Powdered oyster shell is used to whiten protective necklets worn by nursing mothers. It is interesting to note that there is some awareness among the shellfish collectors of the conservation of resources. At every place visited along the coast, informants claimed that they did not take immature molluscs because they wished these to grow big so that they could be used at a later date. It was also claimed that big molluscs tasted better than immature ones. Possibly this is the ideal state of affairs which occurs under optimum collecting conditions, i.e. spring low tides during calm weather. In practice, during unfavourable collecting conditions, women were observed taking hundreds of small limpets with a maximum length of about 2,5 cm. Small girls collecting during fine weather at Mboyi confined themselves largely to limpets of this size. The suspicion could not be avoided that they did this in order to take advantage of the small inducement offered to those who allowed their collections to be weighed. In the absence of measures to acquaint the coastal people with the maximum numbers and sizes of Perna perna , Turbo and Haliotis permitted by legislation (of the species under dis- cussion, these are the only ones so controlled) the gathering of undersized specimens must continue. This raises the question of the effect of shellfish collecting on inter-tidal fauna. Although informants did state that uncooked shellfish could be kept overnight, except during very hot weather, on most of the occasions when the cooking process was observed, all the meat was eaten the same day. There was little wastage, this being confined to a few shells whose contents were overlooked and a small number of Neritidae , which are small enough shells to be gathered by accident if they are attached to the shells of species eaten. Collecting was selective in that people looked for and, usually, took only the species they were going to eat. Although most of the visits to gather data were arranged to coincide with spring low tides, visits have also been made to the Transkei coast at other times. It was only on the former occasion that substantial numbers of women were observed collecting. When the tides were normal, people were not 166 BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI seen on the rocks. It appears also that little, if any, collecting is done during winter; informants state that results of winter collecting do not justify the effort. While Voigt’s account of the transects (systematic surveys conducted by counting all examples of inter-tidal life in the balanoid and cochlear zones) of inter-tidal rocks carried out by her and her helpers will be useful and interesting, they will have to be supplemented by and correlated with a study of both the availability of shellfish and the degree of exploitation in particular spots along the coast. Only in this way can the effect of African dietary habits on the inter-tidal life be ob- jectively established. During five years of visits to a number of places along the Transkei coast, the impression was gained that the frequency of collecting and the volume of shellfish taken does not have an adverse effect on inter-tidal life. Reasons such as lack of waste, selec- tive and periodical picking have been advanced above. This impression is confirmed by some white residents of the Transkei with much longer experience of its coast. On the other hand, demand by white holiday makers must effect inter-tidal life. African youths are given orders for crayfish, and women for oysters, which are purchased in large quantities, particularly during school holidays. Informants at one place reported that a regular white visitor would buy all the available crayfish and freeze them. How important a dietary component is shellfish? The traditional role of cattle in the Southern Nguni economy is axiomatic. Government control measures have resulted in stock limitation. Cattle have always been valued beyond their importance as a mere economic asset; nowadays they are slaughtered only for sacrifices to the ancestors. As these occasions are attended by large numbers of people, they provide each individual with the opportunity of eating only small quantities of animal protein. Even then, the men receive the larger share and women have to be satisfied with inferior cuts and small portions, though smaller children may receive many choice morsels from the men. Nor do other types of stock or poultry regularly provide significant amounts of animal protein. Goats, in Pondoland, sheep, are slaughtered for ritual purposes. Occasionally, also, a sheep is slaughtered to welcome an honoured guest. The hosts partake of its meat. Pigs and fowls are sometimes killed for meat but not regularly enough to provide a significant protein intake. Women customarily avoid eating eggs, so this source of protein is denied them. With reduced herd sizes, relatively little milk is available. Shellfish is therefore the only source of easily accessible animal protein available in quantity; it plays a highly important role in the diet of coastal African communi- ties in the Transkei. ACKNOWLEDGEMENTS I wish to record my appreciation to Mrs. E. Voigt for her cooperation in this study, and to her and Miss E. M. Shaw for critical reading of the manuscript. REFERENCES Alberti, L., 1968. Account of the Tribal Life and Customs of the Xhosa in 1807, Cape Town, Struik. Barrow, J., 1801. An Account of Travels into the Interior of Southern Africa 1797-98, London, Cadell & W. Davies. Brownlee, J., 1968. Account of the Amakosae in Thompson, G.: Travels and Adventures in Southern Africa, Cape Town, Van Riebeeck Society. Holt, B., 1969. The Tshezi of the Transkei, Johannesburg, the author. Hunter, M., 1963. Reaction to Conquest, London, O.U.P. Kay, S., 1833. Travel and Research in Caffraria, London, John Mason. Kilburn, R. N., 1972. Taxonomic notes on South African marine Mollusca (2), with the description of new species and subspecies of Conus, Nassarius, Vexillum and Demoulia, Annals of the Natal Museum, 2, 21, Pietermaritzburg. Kropf, A., 1889. Das Volk der Xosa-Kaffern, Berlin, Buchhandlung der Berliner evangelischen Missions- Gesellschaft. Lichtenstein, H., 1928. Travels in Southern Africa, Cape Town, Van Riebeeck Society. Vanderkemp, J., 1803. Account of the Religion, Customs, Population, Government, Language, History and Natural Productions of Caffraria, in Transactions of the Missionary Society, 1, London (?). 167 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 Appendix 1 : Species present in meal samples + = also present Locality: Shixini Meal number 1 2 3 4 5 6 Panulirus sp. Balanus sp. fragments 155 217 6 13 Dinoplax sp. plates 26 8 4 Fissurella natalensis 2 1 4 3 28 Haliotis midae Haliotis spadicea 1 1 H. species Turbo coronatus Turbo natalensis 1 1 Turbo sarmaticus 2 1 2 Charonia lampas pustulata Burnupena papyracea lagenaria 5 4 1 26 Oxystele sinensis 3 2 Oxystele tabularis 1 Oxystele tigrina Oxystele variegata Oxystele sp. Thais capensis 2 Thais sp. 1 2 Nerita albicilla Nerita plicata Nerita sp. Patella barbara 6 2 21 Patella cochlear 1 6 Patella granularis 2 13 2 1 35 Patella longicosta 19 2 22 7 9 139 168 BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI Locality: Shixini Meal number 1 2 3 4 5 6 Patella miniata 2 5 Patella oculus 11 2 8 4 9 24 Patella tabularis Patella concolor Patella sp. Siphonaria capensis Helcion pectunculus Helcion pruinosis Cellana capensis 46 60 131 63 37 44 Perna perna 17 185 347 280 27 772 Septifer bilocularis Crassostrea cucullata 1 Crassostrea margaritacae 1 Pyura stolonifera + Octopus + Total Number of shells 123 270 682 580 88 1 123 Total Weight of full shells (kg) ,65 1,95 3,55 2,3 ,89 5,3 169 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 Locality: Ntlonyane Meal number 7 8 9 10 11 12 Panulirus sp. Balanus sp. 4 1 Dinoplax sp. 8 174 1 34 Fissurella natalensis 52 22 6 13 21 7 Haliotis midae Haliotis spadicea 2 H. species Turbo coronatus 42 18 4 Turbo natalensis 18 2 Turbo sarmaticus 14 Charonia lampas pustulata Burnupena papyracea lagenaria 6 17 28 3 5 2 Oxystele sinensis Oxystele tabularis Oxystele tigrina Oxystele variegata 2 4 Oxystele sp. Thais capensis 2 1 2 Thais sp. Nerita albicilla Nerita plicata Nerita sp. + + Patella barbara 17 18 3 3 2 Patella cochlear Patella granularis 81 38 34 78 8 8 Patella longicosta 18 3 8 6 99 2 Patella miniata 94 2 10 5 Patella oculus 342 405 207 177 1 117 170 BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI Locality: Ntlonyane Meal number 7 8 9 10 11 12 Patella tabularis 45 35 1 95 3 Patella concolor 85 55 19 40 253 Patella sp. Siphonaria capensis 6 3 Helcion pectunculus 2 2 Helcion pruinosis 11 7 Cellana capensis 2 362 602 1 309 1 562 268 391 Perna perna 1 2 Septifer bilocularis Crassostrea cucullata 4 Crassostrea margaritacea Pyura stolonifera + + + Octopus Total number of shells 2 891 1 583 1 723 1 876 588 803 Total Weight of full shells (kg) 5,5 2,78 2,89 2,6 3,34 2,32 371 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 Locality: Mbotyi Meal number 13 14 15 16 17 18 19 20 21 22 Panulirus sp. Balanus sp. fragments 3 8 2 49 8 Dinoplax sp. plates 7 6 1 21 184 Fissurella natalensis 19 2 50 3 3 23 13 Haliotis midae Haliotis spadicea H. species Turbo coronatus Turbo natalensis Turbo sarmaticus Charonia lampas pustulata Burnupena papyracea lagenaria 23 2 3 3 3 8 7 Oxystele sinensis Oxystele tabularis Oxystele tigrina 4 Oxystele variegata Oxystele sp. Thais capensis 1 6 2 Thais sp. Drupa sp. 1 Nerita albicilla Nerita plicata Nerita sp. + + + Patella barbara 1 6 24 1 11 2 Patella cochlear 1 + I Patella granularis 63 11 80 26 16 174 29 Patella longicosta 4 3 15 6 3 Patella miniata 15 3 39 22 1 23 34 32 3 172 BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI Locality: Mbotyi Meal number 13 14 15 16 17 18 19 20 21 22 Patella oculus 1 17 6 4 4 9 1 Patella tabularis 12 1 1 2 19 Patella concolor 61 5 52 36 7 13 50 45 29 Patella sp. Siphonaria capensis 8 1 6 14 6 Helcion pectunculus 6 Helcion pruinosis 1 12 25 1 Cellana capensis 454 1 382 244 1 1 097 1 268 1 095 224 Perna perna 109 4 18 315 782 1 846 3 2 Septifer bilocularis 2 9 13 Crassostrea cucullata 75 28 31 9 57 Crassostrea margaritacae Pyura stolonifera Octopus Total number of shells 765 90 583 835 800 1 895 1 174 1 402 1 509 585 Total Weight of full shells (kg) 2 1,65 1,25 3,9 6 6,1 1,5 2,3 3,15 2,15 173 Appendix ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 9, DECEMBER 1973 174 Appendix 2 — continued. BIGALKE: THE EXPLOITATION OF SHELLFISH BY COASTAL TRIBESMEN OF THE TRANSKEI 175 All species of Patellidae all along the coast are isigwegwe or isagwegwe. PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mas. ( nat . Hist.) VOLUME 9 • PART 10 13th DECEMBER 1973 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA Report on some collections of middle stone age artefacts from Riverton, Kimberley district, South Africa by A. J. B. HUMPHREYS Alexander McGregor Memorial Museum, Kimberley INTRODUCTION A series of assemblages consisting mainly of “Middle Stone Age” type artefacts was collected by Prof. K. W. Butzer and Dr. G. J. Fock during their study of the “Riverton Formation” near Riverton-on-Vaal (28° 33'S, 20° 46'E) in the Kimberley area (Butzer, Helgren, Fock and Stukenrath, 1973). (Fig. 1). The assemblages consist of groups of artefacts collected from confined areas that had been exposed by recent erosion; some are clearly in secondary context while others are considered to have probably been in semi-primary context before exposure (Butzer et ah , 1973). The assemblages were derived from Member III of the Riverton Formation which relates to the late Upper Pleistocene; Member III is considered, on the basis of radio-carbon dating, to have terminated no later than 17 000 BP. Four collections of artefacts were given to the writer for analysis in February 1973, by Mr. David M. Helgren. The stratigraphic positions of the collections relative to the Riverton Formation are shown in Fig. 2 and 3. METHOD OF ANALYSIS As already mentioned, the majority of the artefacts can be related on typological grounds to the “Middle Stone Age.” The dating of Member III of the Riverton Formation confirms this impression (Butzer et a/., 1973). Perhaps the most clearly defined typology devised for the analysis of “Middle Stone Age” artefacts is that employed by Garth Sampson in the Middle Orange River area (Sampson 1968). The Middle Orange River is only some 240 km SSE of Riverton and so it was decided to use this scheme in the analysis of the Riverton collections. Full details of the types are to be found in Sampson’s monograph on “The Middle Stone Age Industries of the Orange River Scheme Area” (Sampson 1968: 11-13). PHYSICAL STATE OF THE COLLECTIONS A wide range of weathering variation is apparent in each collection: some artefacts are very heavily patinated or weathered while others are completely fresh. From the point of view of etat physique , therefore, the collections would not appear to be homogeneous in terms of internal age variation. However, all of the unpatinated artefacts are made on cherty raw materials (chert, jasper, agate, etc.) that do not normally produce patinas (“desert varnish”) at the same rate, nor of the same type, as do andesite and lydianite (on which most of the other artefacts are made), 177 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 10, DECEMBER 1973 Fig. 1 . Map showing the location of Riverton-on-Vaal and the sites of the various Collections. 178 HUMPHREYS: COLLECTIONS OF MIDDLE STONE AGE ARTEFACTS FROM RIVERTON Fig. 2. Schematic location of four artefact Collections near Riverton-on-Vaal. Not to scale. Data based on K. W. Butzer (pers. comm.). which quickly produce a ferromanganese patina, or quartzite which forms a siliceous sheen relatively rapidly. Little is known about the mechanisms of patination except that the process can be quite rapid in a subarid to semiarid environment, proceeding in as little as a few decades (see Engel and Sharp, 1958). Thus much of the patination apparent in Collection No. 1 below presumably reflects on the duration of exposure of individual artefacts that have been exhumed by sheet wash or gullying during the present century. A second aspect of variation concerns the white, greenish, or reddish weathering rinds apparent in each of the collections. The white to greenish tones are restricted to artefacts coming from white, calcareous sands (Collection No. 2-4) and reflect on surficial alteration in an alkaline environment, e.g. of the olivine of the andesites. In general, Collection No. 2 has thin rinds or cortices of this type, compared with Collection No. 3, suggesting the possi- bility that No. 3 was already weathered prior to being deposited in Member III of the Riverton Formation or, alternatively, that it is substantially older within the time range reflected by Member III. Furthermore, Collection No. 2 shows a variable degree of patination super- imposed on such a whitish cortex, presumably reflecting on its attenuated exposure whereas Collection No. 3 was still largely in place within the sediment body. By contrast, Collection No. 1 shows a variable degree of development of reddish weathering rinds, all overlaid by recent patinas. These particular differences can be attributed to the contact of a noncalcareous silt (Member IV), resting over a lag horizon of calcareous sands (Member III), that are marked by a variety of reddish yellow oxidation bands and mottles. Those artefacts completely embedded within the lag and subjacent calcareous sands developed noticeable reddish cortices 179 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 10, DECEMBER 1973 Fig. 3. The relative stratigraphic positions of the Collections within the Riverton Formation. Figure adapted from Butzer et al. (1973). prior to exposure, whereas artefacts surrounded by the silts were essentially fresh in appearance until later patinated on the surface. Thus a group of artefacts all originally found in a 3 cm stratum along this lithological contact would develop quite a range of variability in terms of etat physique. Consequently, etat physique offers no reliable grounds for chronological differentiation of artefacts within the collections. So, for example, the collections include a variable number of unretouched flakes, mainly small, that lack cortices or patina, but which at the same time comprise essentially the sum total of the siliceous/cherty raw materials. These flakes are typologically undiagnostic, and could fit in either a “Middle” or “Later Stope Age” context. Whereas the possibility of some younger, intrusive artefacts among the surface scatters of Collections No. 1 and 2 cannot be entirely excluded, all the materials of Collections No. 3 and 4 were buried by sediment at the same time. It appears reasonable, therefore, to include these seemingly “fresh” artefacts within the respective assemblages among which they were found. THE STRATIGRAPHY OF THE RIVERTON FORMATION Full details on the stratigraphy of the Riverton Formation are presented in Butzer et al. (1973) but a brief outline of the sedimentary units is necessary here to provide a geological background to the assemblages to be described. 180 HUMPHREYS: COLLECTIONS OF MIDDLE STONE AGE ARTEFACTS FROM RIVERTON The Riverton Formation consists of a complex stratigraphic sequence of deposits which span all of the late Pleistocene and Holocene and which postdate the Younger Gravels of the Vaal Sequence. Five stratigraphic units (or “Members”) have been recognised and may be summarised as follows: Member I: Over 6 m, base not exposed; light coloured, generally massive clayey silt, interdigitated with lenses of sandy or gravelly detritus. A Vaal floodplain deposit, 8,5 m above modern flood water. Member II: 7 m White silty sands to sandy silts with despersed basal pebbles, followed by 60 cm light gray loam vertisol. A Vaal flood silt to + 10,5 m. Member III: 9 m Local alluvia grading into + 13 m Vaal terrace. Includes basal gravels up to more than 3 m thick, followed by white calcareous silty sands with gravelly lenses. Member IV: 9 m. Vaal terrace at + 8,5 m, embanked against Member III, and consisting of brownish clayey to sandy silts. Member V: 6 m. Tributary alluvia interdigited with +4,5 m Vaal alluvial terrace. Member III is the only unit to have yielded archaeological material and the Collections to be described here are representative of this material. Radiocarbon dating based on Achatina shells suggests that “alluviation of Member III terminated no later than 17 000 B.P.” (Butzer et ah, 1973). There are no absolute dates for the other Members. DESCRIPTIONS OF THE COLLECTIONS No. I Donga West of Elizabeth Conradie School These artefacts were collected from three semi-contiguous concentrations within an area some 10 m in diameter. The nature of these three surface concentrations suggests the possibility of a semi-primary association. The types present are as follows: Trimmed/Utilised Blades T/U Blade Fragments T/U Flakes Frontal Scrapers Convex Scrapers Burins Trimmed Points Untrimmed Blades . Untrimmed Blade Frags Untrimmed Flakes . Cores 14 10,0% 12 8,6% 22 15,8% 2 1 .4% 8 5,7% 4 2,8% 13 9,3% 7 5,0% 55 39,5% 2 1,4% 39 The artefacts all conform well to the types defined by Sampson and perhaps the only ones worthy of special comment are the four unifacial “Trimmed Points”. The first is a neatly made “lanceolate” shaped point. There is minimal retouch and the shape was achieved by the form of the original flake. The dorsal surface consists largely of cortex and the form of the flake may consequently have been achieved more by accident than by design. The point is 74 mm long, 29 mm wide and 10 mm thick. The second point was made on a triangular levallois point 181 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 10, DECEMBER 1973 and again retouch is minimal. The dimensions are 63 x 37 x 12 mm. The third specimen was also made on a triangular levallois point; in this case the retouch along the edges has created a serrated or “oak-leaf” effect. The dimensions are 69x44x15 mm. The fourth point is extensively retouched, but cortex has been preserved on the butt half of one edge. The point measures 77 X 36 X 13 mm. The “Convex Scrapers” include one very large specimen; it is 132x 132 mm and some 67 mm thick. It may be associated with the “Middle Stone Age” artefacts, but the fact that it was found in isolation at the periphery of the collecting area and is very much fresher than the other artefacts suggests the possibility of derivation from another (later?) context. The large Convex Scraper is made from andesite. The lengths of the whole “Trimmed/Utilised Blades” may be summarised as follows: Max: 78,0 mm Min : 30,0 mm Mean: 48,6 mm These measurements have been recorded so that they can be compared with data presented by Sampson, as will be seen later. The raw materials represented are as follows: Lydianite Quartzite Andesite Chert , etc. '93 29 5 12 The nature of the exposed artefact scatters and the shallow over-burden suggest that this occurrence may be worth further archaeological investigation under controlled conditions. No. 2 Main Donga , East Bank near Bridge This assemblage was collected from an area 3 x 8 m, on a recently eroded spur of Member III sediment. The types present are as follows: T/U Blades 14 T/U Blade Frags 6 T/U Flakes 24 Frontal Scrapers — Convex Scrapers 10 Burins — Trimmed Points 1 Untrimmed Blades — Untrimmed Blade Frags 3 Untrimmed Flakes 16 Cores 2 76 The artefacts are again “typical” of the period. The single “Trimmed Point” was made on a triangular Levallois flake; the minimal retouch has produced a serrated or “oak-leaf” effect. The dimensions are 60 x 36 X 13 mm. The size range of the whole “Trimmed/Utilised Blades” is as follows: Max: 57,0 mm Min: 29,0 mm Mean: 40,5 mm 182 HUMPHREYS: COLLECTIONS OF MIDDLE STONE AGE ARTEFACTS FROM RIVERTON The raw materials represented are : Lydianite Quartzite Andesite Chert , etc, '48 6 5 17 No. 3. Eastern Tributary Donga , near Location These artefacts were collected from an area of 1 square m, reflecting a high concentration at the base of Member III, and what in terms of disposition is probably a former scatter of artefacts reworked by colluvial action. The types present are as follows: T/U Blades 12 T/U Blade Frags 2 T/U Flakes 1 Frontal Scrapers — Convex Scrapers 1 Burins — Trimmed Points — Untrimmed Blades 2 Untrimmed Blade Frags 1 Untrimmed Flakes 23 Cores 6 48 In addition: 2 “Early Stone Age” artefacts: (1 cleaver; 1 trimmed fragment). The “Middle Stone Age” assemblage is small and not particularly noteworthy. The size range for the whole “Trimmed/Utilised Blades” is as follows: Max: 83,0 mm Min: 32,0 mm Mean: 55,3 mm The raw materials are: Lydianite Quartzite Andesite 11 5 32 The “Early Stone Age” cleaver is made from andesite; it is 149 x 108x37 mm, and was made on a “side-struck flake.” The andesite cleaver is both more weathered and waterworn than the andesite artefacts assigned to the “Middle Stone Age.” This suggests that, although the “Middle Stone Age” artefacts are themselves rather undiagnostic, they may be correctly assigned. On the other hand, it is interesting to note how, as a result, the raw material is dominated by andesite in contrast to the other assemblages where lydianite was the preferred raw material. From this point of view the possibility also exists that this is a mixed assemblage (for it was not in primary context) but with no typological grounds for separating some of the less “characteristic” artefacts. The artefacts were strongly concentrated at one level and were picked out in situ from the base of Member III. They could possibly have originally been eroded from Member II deposits or be substantially older than the other “Middle Stone Age” assemblages within the time range reflected by Member III, as was suggested above. 183 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 10, DECEMBER 1973 No. 4. Riverton West Donga , Basal Gravel , Member III at “ Waterfall ” This assemblage was taken from the intact gravel body immediately above the andesite boss 900 mm upstream in the west donga. The artefacts were found distributed through a considerable thickness of sediment and lacked any horizontal association or vertical con- centration; they clearly came from a secondary context. The following types are present: T/U Blades 1 T/U Blade Frags — T/U Flakes 3 Frontal Scrapers — Convex Scrapers 1 Burins — Trimmed points — Untrimmed Blades 1 Untrimmed Blade Frags 1 Untrimmed Flakes 5 Cores 1 13 This assemblage is so small that no meaningful comments can be made on it. The raw materials present are: Lvdianite Quartzite Andesite Chert , etc. 4 3 2 4 DISCUSSION Of the four assemblages, only Collection No. 1 from the Donga West of Elizabeth Conradie School is large enough for percentages of tool types to be calculated and even in this one over a third of the artefacts are untrimmed flakes. It would therefore seem that none of the assemblages is of any real diagnostic value — apart from establishing the fact they are all ‘'Middle Stone Age.” Apart from the limited sizes of the assemblages, it is also impossible to relate them to Sampson’s sequence because many of his important diagnostic types are not represented. Sampson (1968:101) makes extensive use of variations in core types in the determination of his “Phases” but because the few cores recovered in the Riverton assemblages are undiagnostic, this method cannot be used here. Perhaps the most remarkable feature in the assemblages is the relative shortness of the “Trimmed/Utilised Blades.” The mean lengths for the Riverton assemblages are well below those recorded for the Middle Orange River area where the lowest was 64,0 mm and the highest 145,0 mm (Sampson 1968:105). Here again it is impossible to be sure whether the Riverton area “Middle Stone Age” produced shorter blades as a result of cultural or raw material differences or whether this impression is merely a function of small, statistically insignificant samples. If, however, there is some truth in the impression that the blades from these assemblages are “short” rather than “long” then it is possible that the assemblages as a group may relate to the later “Middle Stone Age” for both Sampson (1968:99) and Mason (1962:263) suggest that there was a gradual reduction in blade length through time during the “Middle Stone Age.” 184 HUMPHREYS: COLLECTIONS OF MIDDLE STONE AGE ARTEFACTS FROM RIVERTON If this is in fact the case, the assemblages may have been produced nearer the 17 000 year limit for Member III than at the 35-40 thousand dates obtained for some other “Middle Stone Age” sites. The evidence available at present is, however, too scant for any real statements to be made one way or the other. In conclusion, therefore, it can be said that the assemblages from Member III of the Riverton Formation can comfortably be accommodated within the group called the “Middle Stone Age” and that their presence in these deposits is entirely compatible with a date “greater than 17 000” for the formation of Member III. ACKNOWLEDGEMENTS 1 am greatly indebted to Prof. Karl W. Butzer of the University of Chicago not only for the opportunity to study the collections, but also for detailed information on their contexts and on the various rates of patination of different rock types. Thanks are also due to David M. Helgren of the University of Chicago who commented on various aspects of the study, and to Mr. J. B. Hawthorne of the De Beers Geology Department, Kimberley, for map enlargements. REFERENCES Butzer, K. W., Helgren, D. M., Fock, G. J., and Stukenrath, R. 1973. “Alluvial Terraces of the Lower Vaal River, South Africa. A Reappraisal and Reinvestigation.” Jour. Geol. 81. In press. Engel, C. G. and Sharp, R. D. 1958. “Chemical Data on Desert Varnish.” Bull. Geol. Soc. Amer., 69:487 — 518. Mason, R. J. 1962. The Prehistory of the Transvaal. Witwatersrand University Press. Sampson, C. G. 1968. “The Middle Stone Age Industries of the Orange River Scheme Area.” Nat. Mus. Bloem , Mem. 4. 185 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN rti i /a f ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. ( nat . Hist.) K VOLUME 9 • PART 11 15th MAY 1974 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA An ethological study of Dichragenia pulchricoma (Arnold) (Hymenoptera : Pompilidae), a southern African spider-hunting wasp which builds a turreted, subterranean nest* by F. W. GESS (Albany Museum, Grahamstown) and S. K, GESS CONTENTS Page Introduction 188 Techniques used in examining nests 188 Systematic and taxonomic considerations 189 Geographic distribution 191 Locality and description of nesting sites with particular reference to climate, nature of vegetation, soil type and vicinity of water . . .191 General comments concerning nesting and an account of the behaviour associated with burrow excavation and turret construction 192 Description of the nest turret .195 Description of the subterranean burrow and an account of the sequence of cell con- struction and provisioning . 197 Identification, composition and physical condition of the prey used for provisioning the cells .199 Mode of transport of the prey and the positioning of the latter in the cell 200 Oviposition and immature stages . 202 Parasites and other associated organisms 202 Flowers visited by adult wasps 206 Discussion 206 Summary 208 Acknowledgements 208 References 208 187 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 INTRODUCTION It is estimated that at present between five hundred and six hundred species of the family Pompilidae have been described from the Ethiopian Region excluding Madagascar. As far as can be ascertained, not a single African species has been studied with respect to its ethology; indeed, the only published biological information pertaining to these nearly six hundred species appears to consist of about half a dozen prey records involving as many species of wasps, a list of species visiting a plant attractive to Hymenoptera, and a few other fragments in which the identity of the wasp concerned is not clearly stated. Whatever the causes for this shocking hiatus in our knowledge of the African Pompilidae, it is patently obvious that a study of their biology is a potentially most rewarding field of endeavour, where virtually any observations made and recorded will be new. An inherent difficulty associated with the gathering and recording of information of this nature, however, is the unpredictability in terms of time and place of the opportunity of observing this or that species doing something. The approach towards the study must be an opportunistic one in which whatever presents itself is followed up — an approach practised by the present authors. Thus, though the authors’ curiosity had been aroused by the dis- covery on rare occasions in previous years of the nest turrets of Dichragenia pulchricoma (Arnold), the present study of the ethology of this wasp was not so much the outcome of a prior decision to investigate this species as of an unexpected opportunity which presented itself. This opportunity of making a relatively detailed study of what appeared to be a most unusual wasp arose from the chance discovery, during routine collecting in 1972, of two nesting sites, conveniently near Grahamstown, where this species occurred in large numbers. As an understanding of the basic form of the nest of D. pulchricoma is required to appreciate fully not only the import of various statements made but also the relevance of techniques listed in the initial sections of this paper and as a detailed description of the nest follows only later, a brief statement concerning the nature of the latter is given at this juncture. The nest of D. pulchricoma consists of two main sections — a subterranean portion con- sisting of a vertical burrow or shaft with at or near its end a variable number of cells and an aerial portion consisting of a mud turret of definite form surmounting the burrow entrance which latter is situated at ground level. TECHNIQUES USED IN EXAMINING NESTS After a nest turret had been located in the field, it was, if possible, covered with an inverted drinking glass, the aim being to capture the female nest-building wasp should she be in the nest at the time. Capture and subsequent examination of the female is essential, if verification of the species is required. The behaviour of the wasp is such that, having emerged from the nest on an outward-bound flight, it will only return to the nest after this has been carried out. Thus, if the wasp is in its burrow when a glass is placed over its turret, the wasp on emerging will be effectively trapped within the glass and can then be collected easily by introducing some ethyl-acetate impregnated cotton wool beneath the glass. A watch may be maintained at the covered turret until the wasp appears. If the wasp was away, when the glass was placed in position, the latter may be removed long enough to allow the wasp into its nest or an attempt may be made to capture the wasp then and there with a net. Sometimes the wasp remains in the nest until the turret is sprayed as described below, when it emerges very rapidly, and quick action must be taken to prevent its escape. If, as in the present study, it is wished to preserve the nest turret which is of an extremely delicate and fragile nature, it is essential to impregnate the dried mud pellets of which it is made with a penetrating fluid which on drying hardens the individual pellets and glues 188 GESS AND GESS: AN ETHOLOGiCAL STUDY OF DICHRAGENIA PULCHRICOMA neighbouring pellets together. In the present study an aerosol clear lacquer was found most effective. The lacquer was sprayed fairly liberally onto the turret and particularly onto the ground for a distance of about three to five centimetres around its base. After allowing the spray to dry (about \ hour in hot sunshine) the sprayed ground around the base of the turret was undercut with a strong penknife. The so-produced consolidated earthen disc bearing the turret itself was then lifted intact and placed on cotton wool in a small unit-tray in which it was readily incorporated into the collection. As soon as the turret had been removed, the top of the vertical shaft then exposed was cleared of any material that might have blocked it, when the former was undercut and lifted. Flour was then blown down the shaft coating the walls with a white tracer which was indis- pensable in the subsequent excavation. In the present investigation a plastic mustard-dispenser as seen on the tables of some restaurants was used — the tip of the nozzle fitted to the screw-off top was directed into the shaft-opening and the container was then squeezed to blow out the flour. In order to expose and draw the vertical shaft and the cells at or near its end, it was essential to excavate the nest from the side. A pit about fifteen centimetres deep and at least as long and wide was, therefore, dug to one side of the assumed position of the nest, the nearer side of the pit being about seven centimetres away from the shaft opening. As the ground was usually extremely hard, a geologist’s pick was used to make this initial excavation. Working from the pit towards the nest, the ground was then carefully broken away using initially a cold chisel and club-hammer, later the point of a penknife. Final exposure of the shaft (marked with flour) and of the cells (if sealed not marked with flour) called for precise and careful work. A plan was drawn and measurements of the nest taken as the latter was exposed, notes were taken of the nature of the cells, the prey and the wasp young. Prey and wasp young of each cell were individually collected in labelled glass vials or gelatin capsules and kept alive for subsequent closer examination. SYSTEMATIC AND TAXONOMIC CONSIDERATIONS In the key to genera and species of the tribe Macromerini (as subfamily Macromerinae) published by Arnold (1934: 289 — 95), the present species runs down easily to Pseudagenia pulchricoma Arnold, with the description of which (1934: 337 — 9) it agrees perfectly. No authoritatively determined specimens were available to confirm the identification but com- parison with specimens of the subspecies sordida Arnold from Lesotho, identified as such by Arnold himself, supported the determination, allowance being made for the stated differences. The present specimens are characteristically coloured: the greater part of the head and thorax (including the pronotum) is black, the wings are pale fuscous, while the abdomen is mostly ferruginous as are the legs with the exception of the coxae. In the female, the clypeus and antennae are ferruginous (the latter becoming gradually darker from the third or fourth joint onwards), the face, pro-mesonotum and scutellum have a pale greyish golden pubescence, while the second to fifth tergites are marked medially with black maculae which do not reach the posterior margins of the segments but which together give the impression of a somewhat diffuse, wide, median streak on the dorsum of the abdomen. This black streak on the other- wise ferruginous dorsum of the abdomen, taken in conjunction with the black thorax and fuscous wings readily identifies the wasp in the field. The male is similarly coloured but lacks the black streak on the abdomen. Since the time of Arnold's revision of the Pompilidae of the Ethiopian Region (1932 — 7), the very large cosmopolitan genus Pseudagenia Kohl, 1884 (= Auplopus Spinola, 1841) has been split up into a number of smaller genera. 189 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 Banks (1934) in his account of the Psammocharidae of the Philippines removed from Pseudagenia ( senso lato) one new subgenus and seven new genera, not all of which have been recognized by subsequent workers. The present species, pulchricoma Arnold, would, according to the generic key given by Banks (1934: 39 — 40) be referable to the genus Phanagenia Banks (1933: 18). This genus, recognized as valid by subsequent workers, was erected “for certain American forms possessing spines on the underside of the last tarsal joint” (Banks, 1934: 78). The generic description (of the female only) of Phanagenia (type species: Phanagenia osceola Banks, 1933 = bombycina (Cresson), 1867) is short, but the characters indicated are all exhibited by the present species, pulchricoma Arnold. In the key to the Nearctic genera of the Macromerini published by Townes (1957: 140) pulchricoma Arnold once more runs down to Phanagenia Banks, which genus according to Townes differs from Auplopus Spinola, 1841 (= Pseudagenia Kohl, 1884) and two other genera (Ageniella Banks and Priocnemella Banks) in the following combination of characters: (1) first tergite with a fine lateral crease that separates off the epipleuron; (2) propodeum without long erect hairs; (3) mentum of female with a brush of about 20 long stout bristles which are not divided into right and left groups; and (4) underside of last tarsal segment of female with preapical bristles. The present species, pulchricoma Arnold, differs in having the mental bristles of the female rather sparse and in having the clypeus of the male simple (clypeus of male with specialized apical margin in Phanagenia bombycina (Cresson)). However, the same differences have been accepted as possibly merely specific ones by Townes (1957: 141), when considering the Mada- gascan Agenia macula Saussure as a possible Old World member of the genus Phanagenia. Agenia macula Saussure has also been studied by Haupt who, however, named it as the type-species of Dichragenia (1950: 25 and 1957: 14), one of several new Ethiopian genera split off by him from Pseudagenia as understood by Arnold. Dichragenia, according to Haupt, is represented on the African continent itself by pulchricoma Arnold, the subject of this paper. Without being able to compare the type-species of Phanagenia and Dichragenia , it is not possible to assess fully the morphological differences between the two, differences which undoubtedly are small. While the ethology of Dichragenia macula (Saussure) is unfortunately unknown, that of pulchricoma Arnold differs in important aspects from that of Phanagenia bombycina (Cresson) and lends support to generic separation. In the present paper Haupt’s view is adopted and the wasp is referred to as Dichragenia pulchricoma (Arnold). Arnold (1934: 337 — 40) as well as describing pulchricoma (senso stricto ) from “Southern Rhodesia and British East Africa”, described two subspecies (“races” of Arnold): sordida from Harrismith (Orange Free State), Willowmore and Aliwal North (both Cape Province); laeta from Umtata (Cape Province), Delareyville (“De la Rey” of Arnold) and Lichtenburg (both Transvaal). In the Albany Museum collection there are fair-sized series of both these “races” — 9 females and 6 males of sordida , and 8 females and 15 males of laeta , the determinations being by Arnold himself. Unfortunately, all these specimens are from Lesotho, all but one from the same locality, Mamathes; months of capture for both are October — February/March, in quite a number of cases in the same year! From the above it is clear that sordida and laeta cannot both be considered as subspecies of pulchricoma. Some aspects of the ethology of one species of “a certain group of spider hunters” in Lesotho have been observed by Jacot-Guillarmod (1945: 43). He refers to this species as “the Zimbabwe builder” and from personal discussions in 1972 it emerges that the wasp in question was sordida. As far as the information goes, no differences can be found with respect to the ethology of pulchricoma ( senso stricto) as studied in Grahamstown. This together with 190 GESS AND GESS: AN ETHOLOG1CAL STUDY OF DICHRAGENIA PULCHRICOMA the great general morphological similarity leads to the view that pu/chricoma and sordida are indeed conspecific. On the other hand, despite the fact that laeta was more common than sordida in Lesotho, it was, unlike the latter, never observed nesting (Jacot-Guillarmod, pers. comm., 1972), leading to the belief that it may well do so in a different ecological situation. This taken in conjunction with the general morphological differences leads to the view that laeta is not conspecific with pu/chricoma but is better regarded as a separate species. This view agrees with that expressed by Arnold in 1961 when, during discussions with Jacot- Guillarmod, his attention was drawn to the fact that sordida and laeta are sympatric in Lesotho (Jacot-Guillarmod, pers. comm., 1972). GEOGRAPHIC DISTRIBUTION Dichragcnia pu/chricoma (Arnold) appears to enjoy a very wide distribution in Africa. Published locality records indicate a distribution from the equator (Kisumu, Kenya, at the north-eastern extremity of Lake Victoria) to the southern parts of the Cape Province (33 S.), ranging through several major vegetational zones. In addition to the localities given by Arnold (1934: 339 — 40) — namely, Rhodesia (where “a common species”), Kisumu (“British East Africa”), Harrismith (Orange Free State), Aliwal North and Willowmore (both Cape Province), the species has been recorded from a number of widely separated localities in Zaire (the former Belgian Congo) by de Saeger (1945: 99) and by Haupt (1957: 15). The Albany Museum collection contains specimens from Lesotho (Mamathes, Teyate- yaneng and Henley’s Dam, Leribe) collected by C. F. Jacot-Guillarmod (October — February) and from the Eastern Cape Province (various localities near Grahamstown). A nest turret, now in the collection, obtained by R. A. Jubb at Kenton-on-Sea, 44 kilometres SSE of Grahams- town, indicates the species’ presence there. It is probable that the species occurs also in the western part of the Cape Province for a single female associated with a nest turret of typical form and believed to be D. pu/chricoma was observed but not caught by one of the authors (F.W.G.) at the Olifants River between Klawer and Clanwilliam during October 1967. LOCALITY AND DESCRIPTION OF NESTING SITES WITH PARTICULAR REFERENCE TO CLIMATE, NATURE OF VEGETATION, SOIL TYPE AND VICINITY OF WATER Field observations in the present study were centred on Grahamstown (33 19' S, 26 32' E) in the Albany Division of the Eastern Cape Province of South Africa. The greater part of the field work, including the study of 55 nests was carried out at Hilton, a farm situated about 18 kilometres WNW of Grahamstown; a smaller number of nests (6) was studied at a second farm, Clifton, situated at about the same distance from the town but to the NW; one nest and its builder were studied in the authors’ garden in Grahamstown itself. The greater part of the field work was spread over a period of six weeks from 26.x. 1972 to 6.xii.l972; an additional three mornings’ field work was carried out at Hilton later in the summer (9 and 1 4.ii. 1 97 3 and 1 .iii. 1973). The Albany Division, situated between the winter and summer rainfall regions, receives rain in moderate amounts throughout the year, the wettest periods being spring and autumn, the dryest mid-winter. Grahamstown itself has a mean annual rainfall of 697,2 mm (27,45 inches), however, the farms Hilton and Clifton, for reasons of topography, receive consider- ably less — at Hilton the mean annual rainfall is in the region of 356 — 381 mm (14 to 15 inches). During 1972, a drought year, Grahamstown received 532 mm (20,94 inches) and Hilton received only 229 mm (9 inches). 191 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 The vegetation of Hilton and Clifton is karroid in nature and consists largely of an open community of small shrubs, few of which are strongly succulent but many of which are xerophytic. Acacia karroo forms much of the taller scrub, and grasses have largely been eaten out by stock. In both localities there has been some erosion of the soil resulting in some localized denuded areas. Nesting sites at both Hilton and Clifton as well as in Grahamstown itself are in low-lying situations (river valleys) in which the soil is derived from the Dwyka Series and is of a reddish- brown clayey nature. All nests found during the present investigation were built in places where the ground had been disturbed and had been partially or completely denuded of vegetation as a direct or indirect result of Man’s activities. Thus at Hilton, nests were found on the slightly raised edges of a shallow earthen dam; on the raised bank of earth running parallel to a water furrow and derived from the excavation thereof; on the denuded ground on the opposite side of the furrow; on slightly sloping areas denuded of vegetation and topsoil by sheet erosion, and on the sides of small erosion gullies intersecting such bare areas (Plates 1—4). At Clifton nests were found in essentially similar situations; the nest found in Grahamstown was situated on bare earth fringing a vegetable bed. Mud turrets surmounting the burrow entrances were sometimes found completely exposed in the middle of totally bare areas but were sometimes in somewhat more protected situations such as against small banks or steps in the ground level or next to the base of dwarf shrubs (Plates 5 — 8). In all cases the nests were situated in close proximity to temporary sources of water. This water, a result of rain, had collected in muddy pools, erosion gullies, shallow furrows and, in the case of the garden, a concave cabbage leaf. Similarly the previously mentioned nest turret observed at the Olifants River and believed to be that of D. pulchricoma was sited on a fairly steep, bare, sheet-eroded clayey slope situated not far from two water sources — the Olifants River itself, and, closer at hand, muddy pools below a leaking concrete irrigation furrow. Turrets of nesting D. pulchricoma at Mamathes (Lesotho) were reported by Jacot- Guillarmod (pers. comm.) to have been built on denuded patches of black clay near a spring. GENERAL COMMENTS CONCERNING NESTING AND AN ACCOUNT OF THE BEHAVIOUR ASSOCIATED WITH BURROW EXCAVATION AND TURRET CONSTRUCTION Dichragenia pulchricoma (Arnold) appears, at least in the localities studied (around Grahamstown), to be an opportunist — emerging in response to rain and nesting during the short period during which standing water, essential for nest construction, is available in pools and puddles. However, it appears as if only a fraction of the total population responds to any one fall of rain, for during summer, that is during the period of October to March, each shower of rain substantial enough to wet the ground and to cause run-off to form puddles in depressions, is followed by a flush of wasps and a spate of nest-building. As the puddles dry up again, the number of wasps observed falls off and nesting ceases. Thus, the potential nesting season is divided up into a number of short nesting bouts by apparently successive waves of wasps; the period between the emergence of the wasps and the completion of their nesting is very short; and, at any one time, virtually all nests being worked upon are at a similar stage of development. Comparison, with respect to stage of development and size, of wasp young present in a number of nests examined on a day soon after the beginning of a nesting bout with young present in other nest samples examined on subsequent days (Fig. 1), 192 100 90 ' 80 ' 70 ' 60 - 50 - 40 30 20 10 2 nests (4) (3) (0) (0) 15. xi. 1972 4 nests 3 nests 16. xi .1972 20 . xi . 1972 100 90 80 70 60 % 50 40 30 20 10 ° (1) (0) (3) (5) 23. xi. 1972 6 nests (1) (0) (1) (12) 24. xi. 1972 Egg Small larva feeding on prey- abdomen Large larva feeding on prey cephalo- t ho rax Larva in cocoon Fig. 1. Composition (expressed as a %) of wasp young (egg, small larva, large larva and larva in cocoon) col- lected at Hilton during the period 15 — 24. xi. 1972. The actual number of each stage is indicated within brackets. ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 illustrates not only the synchrony of nesting within any one nesting bout but also the rapidity of development from egg to mature larva. No evidence was found that might suggest that there was more than one generation of wasps per year. Indeed, in all the nests examined the most advanced young were fully-grown larvae within silken cocoons; no pupae were found. It seems likely that these larvae do not immediately pupate but rather enter a state of diapause in which the rest of the nesting season, the autumn and the dry winter months, are spent and that the stimulus for the breaking of the diapause is at least in part provided by the rains of the following spring and early summer. Presumably, development thereafter is rapid as is the previous development from egg to mature larva. The behaviour associated with burrow excavation and turret construction consists of a cycle of events which is repeated many times over. Each cycle may be considered to consist of three major activities as indicated below. (1) Water carrying Flying from the nest under construction to the nearby water source, filling the crop with water and returning to the nest. (2) Shaft excavation ( sinking ) and turret building Regurgitation of some of the water onto the earth at the working face of the vertical shaft being sunk, the excavation with the aid of the mandibles of the resultant mud and the moulding thereof into a pellet (building block), the transport of this pellet up the shaft and its addition to the free edge of the turret rising up above the mouth of the shaft. This major activity is repeated a number of times within each cycle. (3) Grooming Grooming of the head in general and the antennae in particular before setting out to fetch the next crop-full of water — that is, before initiating the next cycle. It will be seen that the excavation of the subterranean portion of the nest is intimately linked with the building of the turret above ground and that the size of the turret (provided it has not been damaged) provides a clear indication of the extent of the underground workings. A wasp occupied in nest building in the Grahamstown garden on 1 6.xii. 1 972 was observed for a period of 60 minutes. During this time 13 of the above defined cycles were completed. About 12 minutes were devoted to water carrying and about 48 minutes were spent on activities at the nest (shaft excavation and turret building as well as grooming). Within the 60 minutes, 68 mud pellets were formed from excavated materials (derived from the excavation of the vertical shaft) and placed in position on the free edge of the turret. Similarly, a wasp, observed at Hilton on l.iii.1973, completed 11 of the above defined cycles in a period of 39 minutes, of which 6 minutes were devoted to water carrying and 33 minutes to activities at the nest. Within the 39 minutes, 52 mud pellets were formed and posi- tioned, resulting in an increase of turret length of between two and three centimetres. The nests of both these wasps were situated close to the water source being utilised— the former about 7 metres distant, the latter 2,3 metres (Plate 2). When fetching water D. puichricoma alights near the water’s edge, walks the last few centimetres and drinks from the edge of the puddle. Similarly, when returning to the nest, she commonly alights 15 to 30 centimetres away from the turret and walks the rest of the way but occasionally alights much closer to the turret or even lands upon the latter. Flights to and from the water is generally direct and rapid. On returning to the turret the wasp enters it head first. Shortly thereafter her head re- appears at the free edge of the turret holding between her mandibles a glistening pellet of 194 GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENIA PULCHRICOMA very wet mud. This is placed in position on the free edge of the turret by the wasp which builds from within the latter, only her head and prothorax emerging. Having positioned the pellet, the head disappears again only to reappear a short while later with the next pellet. The number of pellets formed by the removal of material from the shaft face and positioned on the turret edge varies from 2 to 8 (with an average of 5) per single water load. Though it could not be observed, it is believed that the very wet mud pellet, in addition to being held by the mandibles, is supported from below by the stiff mental bristles which would act in the manner of a pitch- fork. The last pellet of a batch having been formed and positioned, the wasp grooms thoroughly before departing for more water. This grooming, in which particular attention is given to the antennae, may take place in the turret entrance or on the ground next to it. Nest construction appears to take place during the hottest period of the day — the building activities described were observed between about 10.40 a.m. and 1.30 p.m. although already begun before and continued after these times. Building is carried out only in strong sunshine; when the sun is obscured by clouds work ceases. During the period of nest construction, when the females are frequent visitors to the puddles and pools where they obtain their water, their presence or absence at such water sources is a reliable guide as to whether nesting is taking place in any given locality. Males, which in the present study were never observed at the actual nesting sites, are found near these watering places. At Hilton, males, if not on the wing, could frequently be flushed from tussocks of coarse grass and sedges fringing pools visited by the females (Plate 2 and 4). These circumstances suggest that mating may take place at the pools, when the females come to these for water. DESCRIPTION OF THE NEST TURRET The turret built above the entrance to the subterranean part of the nest is more or less circular in cross section, is of variable diameter along its length and is curved in at least the vertical plane. For any given nest the extreme base or foundation of the turret wall is at a fixed distance from the nearest edge of the vertical shaft opening and is obviously a function of the size of the wasp builder. Therefore, any deviation from the circular in the shape of the shaft opening is mirrored in the shape of the turret foundation and large wasps build turrets of greater bore than do small ones. Generally the internal diameter of the turret at its base is about three times that of the vertical shaft, with the consequence that there is incorporated within the structure of the nest a platform-like disc of the ground surface, bounded at its circumference by the turret walls and pierced at its centre by the entrance to the vertical shaft. As the turret rises the bore generally increases slightly, then decreases again to form basally a weakly developed bulb-shaped bulge beyond which the bore continues to decrease gradually towards the distal turret opening. For 42 turrets measured, the outside diameter at the foun- dation varied from 14 mm to 26 mm, the most common diameter was 19 mm and the average 20,5 mm; the outside diameter at the distal end of 24 of these turrets (those which were con- sidered complete or near-complete) varied from 10 mm to 18 mm, the most common diameter was 13 mm and the average 13,5 mm. Turrets generally start off subvertically but almost immediately begin to curve over to one side to form an arch over the ground and to bring the turret entrance down close to the latter. If the turret is lengthened by the addition of further excavated material, it then levels out horizontally and runs parallel to the ground to which it may or may not be attached. This section of the turret may curve laterally or become sinuous. Only one of a total of 64 turrets studied did not initially rise up above the ground — instead, for its entire length it ran along the ground which latter formed the floor of the domed runway. 195 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 The height above the ground attained by the top of the turret arch was found to vary considerably in the 25 turrets suitable for measuring — from 13 mm to 65 mm; 50% of these turrets fell in the 24 — 27 mm range and the average height of all 25 turrets was 28 mm. The length (measured along the outside of the curve) of 15 intact turrets belonging to nests containing at least one cell varied from 42 mm to 140 mm; 50% of these turrets fell in the 60 — 80 mm range and the average length of all 15 turrets was 77 mm. The pellets used as building bricks in turret construction are roughly elongate-oval in shape with a slight constriction at the middle of the long axis and approximate best to the form of a two-seeded peanut pod. Each pellet is laid with its long axis orientated across the long axis of the turret. The pellets vary somewhat in size between different turrets and some- times also within different sections of any one turret. This variation is probably due not only to differing wasp sizes but also to varying physical conditions of the soil. The pellets forming the walls of the basal bulge are generally closely packed and leave but few interstices of small size between them. Beyond the bulge the pellets tend to be more loosely laid and leave between them large spaces giving this section of the turret a lacy appearance. No smoothing or plaster- ing of the bore of the turret is practised by the wasp and the individual pellets are as distinct on the inside surface of the turret walls as on the outside. A well-built turret of typical form surmounting underground workings consisting of a single, sealed cell situated at the end of a 118 mm long vertical shaft (Fig. 2) was photographed in the field at Hilton (23.xi.1972) and is shown in Plate 6. Its dimensions are: outside diameters at base and at end 18 and 12 mm respectively; height of top of arch above ground 29 mm; total length 70 mm. The turrets have little resistance to wetting and even a very light shower of rain will cause the pellets to melt and the turret to collapse. A total of 22 rain-damaged turrets were found at various times during the season, 20 of which had been subject to subsequent building activity. A new turret is frequently built out from what remains intact of the original one but is often inferior in construction and size to the latter and usually consists of little more than a crudely- added, narrow, lacy tunnel akin to the narrow distal portion of a normal extended turret. In one instance, where the original turret had only reached half the height of the basal bulge before being damaged by rain, the replacement of very inferior construction was built within the walls of the old structure. Thus it is believed that the additions built subsequent to rain damage are not repairs but rather a continuation of normal building activity which would have taken place in any case at the free distal end of the turret but which activity has been shifted back in position to a new distal end — the point beyond which the original structure was destroyed. The inferior quality of the additions is thereby explained : what is more difficult to explain, however, is an instance in which a turret replacing a large collapsed one was of normal form though rather smaller than average. The function of the turret is unclear. It seems unlikely that it would prove much of a deterrent to a determined parasite or even a casual one — it certainly did not prevent five separate instances of nest parasitism by leaf-cutting Megachilidae, though it must be stated that, perhaps significantly, in all five cases the turrets were not very extensive and consisted only of that portion up to the top of the arch. As already stated, the turret is destroyed by even a light shower but it is possible that its collapse may be of advantage to the wasp in that the debris plugs the entrance to the vertical shaft preventing flooding of the underground portion of the nest. Finally it may be of some advantage to the wasp that on coming up the vertical shaft it does not immediately emerge into the open but rather does so only after passing through the turret through the walls of which it is able to see the surroundings. 196 GESS AND GESS: AN ETHOLOGICAL STUDY OF D I CHRAGENI A PULCHRICOMA DESCRIPTION OF THE SUBTERRANEAN BURROW AND AN ACCOUNT OF THE SEQUENCE OF CELL CONSTRUCTION AND PROVISIONING A total of sixty nests in various stages of construction was excavated during the present study. In Table I, the underground workings of these nests are grouped into nine categories according to the number of cells associated with each nest. The subterranean portion of the nest consists of a vertical or near-vertical shaft near the lower end of which are situated a variable number of cells. The vertical shafts of the 42 nests included in categories B — I of Table I (i.e. those shafts which were complete as indicated by the presence near their ends of at least one cell) penetrated into the ground to depths ranging from 55 mm to 180 mm, with an average penetration of 113 mm. Of the total of these 42 shafts, 76% were between 90 mm and 130 mm deep (long) and with this latter category of shafts was associated 71% of the total of 108 cells. The bore of the vertical shafts varied from 5 mm (a single abandoned shaft) to 1 1 mm (a single shaft) with 53 of the total of 60 shafts (i.e., 88%) having a bore of between 6,5 mm and 8,0 mm. Table I. Analysis of the form of the underground workings of 60 nests of Dichragenia pulehricoma (Arnold) excavated near Grahamstown (26.x. 1 972-1 4.ii. 1973). Category Description of underground workings Number of nests A Shaft with no cells 18 B Shaft with 1 cell (unprovisioned) 12 C Shaft with 1 cell (provisioned) 6 D Shaft with 2 cells (at least 1 provisioned) 6+ (2) E Shaft with 3 cells (at least 2 provisioned) 5 F Shaft with 4 cells (at least 3 provisioned) 3 G Shaft with 5 cells (at least 4 provisioned) . 2 H Shaft with 6 cells (at least 5 provisioned) 4+(l) I Shaft with 7 cells (at least 6 provisioned) 1 Total: 60 Note: Figures in brackets ( ) pertain to three nests in which for some unknown reason more than 1 cell was unprovisioned — in the former instance two nests each containing 2 cells, neither provisioned, in the latter instance one nest of 6 cells, only four of which were pro- visioned. The cells which vary in length from 14 to 19 mm are formed by the sealing off by means of earthen plugs of the distal portion of short (15 — 25 mm), usually downwardly-inclined sideshafts which branch off from the vertical shaft at or near its terminal (bottom) end. These side shafts (and therefore also the cells) are excavated at roughly a common depth and are arranged around the end of the vertical shaft in a more or less radiating pattern (like the spokes of a horizontally orientated cart wheel). A cell may, however, lie directly beneath the end of the vertical shaft and may represent the sealed off original terminal portion thereof. (Figs. 2 — 4.) In one nest with an exceptionally long vertical shaft, cells were found to have been 197 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 2 Figs. 2—5. Plans of the underground workings of four nests of Dichragenia pulchricoma (Arnold) excavated at Hilton during November, 1972, showing various degrees of complexity and also the sequence of cell con- struction and provisioning. For details of the condition of the prey spider and of the developmental stage of the wasp young within each cell see Table II. 198 GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENIA PULCHRICOMA constructed at two distinct levels — two from the end of the vertical shaft 180 mm below the surface of the ground, and a group of four radiating out from the shaft at a point 130 mm below the surface. (Fig. 5.) The earthen plug between cell and vertical shaft is flush with the wall of the latter and once in position leaves little indication of the proximal part of the side shaft. One cell at a time is excavated and as a rule only a single open cell is found at any one time in any one nest. As soon as the cell has been provisioned and oviposition has occurred, it is sealed, after which work commences on a second cell. It was not established where the earth utilized for plugging the cell comes from, but as it does not differ in colour or other appearance from the earth through which the vertical shaft passes, it is believed that it is obtained from within the nest. Possibly the earth is quarried near the end of the vertical shaft at a point close to the cell being closed, and possibly this quarry may subsequently form a convenient starting point for the excavation of the next cell. The sequence of cell construction and provisioning in multicellular nests may readily be established by virtue of the degree of development attained, at the time of examination, by the wasp young within each of several cells. Clearly, the most informative nest in this respect is one which is still being worked upon by the adult female and which shows a wide range of cell conditions — from a newly excavated, open, and as yet unprovisioned cell to one (but preferably not more than one) containing a mature larva within a silken cocoon. Figs. 2, 3 and 4 show plans of the underground workings of three progressively more complex nests in which is indicated the sequence of cell construction, as inferred from the wasp young within them. Fig. 5, pertaining to an already mentioned nest, shows an unusual nest plan but the typical sequence of cell construction. Table II shows the condition of the provi- sion (prey spider) and the developmental stage of the wasp young found within each of the cells of these four nests. In those nests in which it was present, the cell situated immediately beneath the end of the vertical shaft was found to have been the first to be constructed and provisioned. No particular order of construction could be established for the cells radiating out around the end of the vertical shaft but, in those nests in which these cells were numerous, a tendency was noted for later cells to be constructed at a level slightly above initial ones without, however, altering the general clumped arrangement of the cells. In the underground workings shown in Fig. 5 the marked separation between the initial (lower) two cells and the later (upper) four is seen as the consequence solely of the excessively and abnormally long vertical shaft and the abandonment of the lower level after the construction of only two cells there in favour of the more normal depth for cell construction. IDENTIFICATION, COMPOSITION AND PHYSICAL CONDITION OF THE PREY USED FOR PROVISIONING THE CELLS From the 72 provisioned cells of D. pulchricoma excavated during the present study 38 identifiable prey spiders were obtained. These spiders ranged in condition from some indi- viduals on which feeding by the wasp young had not yet commenced to others which had had the entire abdomen and much of the cephalothorax consumed. The prey spiders of the re- maining 34 cells had been completely devoured with the exception of the cheliceral fangs which were always left uneaten by the wasp larvae. With the addition of one specimen taken from a wasp in the act of transporting it to its nest, a total of 39 prey spiders was available for examination. These were identified as follows: Lycosidae 21 specimens (15 females, 5 males and one specimen of indeterminable sex) (One female, a subadult) 199 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 Pisauridae Euphrosthenops sp. 12 specimens (9 females and 3 males) (One female, a subadult) Sparassidae (formerly included in Clubionidae) Olios sp. 3 specimens (2 females and 1 male) Pseudomicrommata vittigerum (Simon) 2 specimens (both females) Salticidae (= Attidae) 1 specimen (a female) Immediately apparent from the above prey identifications is the preponderance of adult spiders over subadults, females over males, and members of the Lycosidae over those of other families. While a nest of two or more provisioned cells may contain prey spiders of one family only, it more frequently contains representatives of more than one family. Thus individual nests were found that were provisioned with Lycosidae and Sparassidae, Lycosidae and Pisauridae, Pisauridae and Sparassidae, and Pisauridae and Salticidae. The four spider families involved are all composed of strongly built, fast moving, wander- ing and predaceous ground-living forms which stalk and run down their prey on foot. Of the total of 39 prey spiders examined, 32 had had all the legs amputated at the coxal- trochantai joint, three each had one leg remaining, and one had two legs remaining. The other three spiders were extensively eaten but between them possessed ten intact coxae from which the distal parts of the leg had been amputated. It is clear that D. pulchricoma habitually ampu- tates the legs of its prey: that this amputation is carried out prior to the removal of the prey to the nest was demonstrated by the legless state of the spider, previously mentioned, which was taken from a wasp in the act of transporting it from the point of capture to the nest being provisioned. The pedipalps, by contrast, are not removed. In size, the total length of the legless spider bodies varied from 8 — 13 mm, the most common length being 12 mm; the maximum width (measured across the cephalothorax and the leg stumps for all specimens except Olios sp. which was measured across the abdomen) varied from 4 — 6 mm, the most common width (58%) being 5 mm. The “dressed weight” of only one prey spider was established. This specimen, a freshly caught adult female lycosid of average size (total length 1 1 mm, maximum width 5 mm) and with all the legs amputated weighed 76 mg., somewhat less than double the weight of an average sized female wasp. MODE OF TRANSPORT OF THE PREY AND THE POSITIONING OF THE LATTER IN THE CELL Only once was D. pulchricoma observed transporting its prey. This was at Hilton (23.xi.1972) where a female wasp was seen carrying a legless female lycosid across open ground in an area in which were located several nest turrets. Progression, with the wasp facing the direction in which it was going, was by a series of short hopping flights, the wasp alighting every few centimetres and running a short distance before trying to take off again. The spider, held beneath the wasp’s body and straddled by the wasp’s legs, appeared to be positioned dorsum up and head forward. Unfortunately, the point at which the spider was grasped by the wasp’s mandibles was not noted but, on the basis of the otherwise identical mode of prey transport observed in a related wasp species, it is believed to have been the base of one of the chelicerae. The related wasp, “ Pseudagenia’ ’ spilocephala Cameron (? another species of Dichragenia ), whose transport of prey was closely observed on three separate occasions, carried its prey (legless female Lycosidae and Olios sp.) beneath it, dorsum up, head forward (as in D. pulchricoma ), grasped by the base of a chelicera. The actual hunting 200 Table II. Condition of the provision (prey spider) and stage of development of Dichragenia pulehricoma (Arnold) young within each of the cells (listed in order of their construction) of four nests (see Figs 2 — 5) excavated at Hilton during November 1972. Nest (Fig. No.) Cell Condition of cell Condition of provision (prey spider) Stage of development of wasp young 2 1 Sealed Not fed upon Egg 3 1 Sealed Abdomen fed upon Small larva 2 Sealed Not fed upon Egg 4 1 Sealed Completely devoured Larva within silken cocoon 2 Sealed Completely devoured Larva within silken cocoon 3 Sealed Completely devoured Larva within 3 -spun silken cocoon 4 Sealed Completely devoured Very large larva 5 Sealed Abdomen fed upon Small larva 6 Sealed Not fed upon Egg 7 Open (None present) (None present) 5 1 Sealed Completely devoured Larva within silken cocoon 2 Sealed Completely devoured Very large larva (15 mm. long) 3 Sealed All of abdomen and part of cephalothorax eaten Medium sized larva (11 mm. long) 4 Sealed Abdomen fed upon Small larva (9 mm. long) 5 Sealed Not fed upon Egg 6 Open (Wasp builder found in this cell) (None present) (None present) 201 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 and stinging of the spider by the wasp and the process of leg amputation was not observed in either of the species. The position in which the prey spider was placed in the cell by D. pulchricoma was without exception such that it rested with its ventral surface on the floor of the cell with its head to- wards that end of the cell nearer the vertical shaft. OVIPOSITION AND IMMATURE STAGES Oviposition upon the prey spider takes place only after the latter has been positioned within the cell. Great consistency with regard to orientation and site of attachment of the egg is shown — thus in the observed cases, the egg was without exception found to be securely attached in a somewhat oblique position to the side of the spider’s abdomen, near the base of the latter. The morphological anterior end of the egg was uppermost (Figs. 6 and 7). There was no observable preference for either the right or the left side of the spider’s abdomen, equal numbers of eggs (and also larvae) being found on both sides. The young larva on hatching from the egg was found not to move away from the position in which the latter was laid but to commence feeding at the site thus determined for it. The point at which the spider’s integument was punctured and where feeding began was on the dorsum of the abdomen to one side of the midline, in the region occupied by the spider’s digestive gland (Plate 9). Young larvae engaged in feeding, during which the head was kept firmly applied to the puncture, were observed to engage in pulsating or pumping movements over the entire body. It was found that, when all the fluids had been thus imbibed and the larva had grown stronger, chewing of the harder portions of the abdomen began and the larva’s fixed position upon the prey was relinquished. When the abdomen had been completely eaten, the cephalo- thorax was started upon, beginning at the posterior end and progressing forwards until the whole of the prey including the coxae and the pedipalps but excluding the cheliceral fangs was consumed. When all the provision has been eaten, the mature larva spins its pale brownish-yellow, partly translucent, parchment-like cocoon which is attached to the cell’s walls by fine silken threads. A small opening is left at that end of the cocoon that is situated at the distal end of the cell (i.e., that end away from the vertical shaft) and the head of the mature larva may be seen at this opening at the time cocoon spinning is completed. Thereafter the larva reverses its position within the cocoon thereby bringing the head toward that end through which the adult will emerge. Meconium is released at the cocoon’s opening at which the hind end of the larva is now situated, to form a hard, dark-coloured plug sealing the cocoon and, by sticking onto the cell wall, further anchoring the former to the latter. The uneaten cheliceral fangs of the prey are to be found adhering to the outside of the cocoon. Measurements pertaining to the egg, to the larva at various stages of its existence (with reference to the site of its feeding upon the prey) and to the cocoon are given in Table III. PARASITES AND OTHER ASSOCIATED ORGANISMS In the sixty nests excavated during the present study remarkably few parasites and other associated organisms were present. Two cocoons obtained from separate nests of D . pulchricoma at Hilton on 24.xi.1972, though similar to those of that species in general shape, appearance and size (12 mm X 6 mm), differed noticeably from them with respect to detail. Thus these two cocoons were harder, darker in colour and were constructed of a greater thickness of silk spinnings, the silk thread itself being coarser. Adhering to the outside of the cocoons were the spider’s cheliceral fangs, 202 GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENIA PULCHRICOMA Fig. 7. The same (lateral view). Fig. 6. Egg of Dichragenia pulchricoma (Arnold) in position on the abdomen of a lycosid spider (dorsal view). 203 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 Table III. Measurements pertaining to the egg, to the larva at various stages of its existence (with reference to the site of its feeding upon the prey), to the cocoon and to the adult of Dichragenia pulchricoma (Arnold). Stage Size range (mm.) Number measured Egg 1,9 x 0,6— 2,3 X 0,8 (average: 2,1 x0,7) 6 Larva feeding on spider’s abdomen 2—9 7 Larva feeding on spider’s cephalothorax 9,5—15 4 Larva which has completed feeding but has not commenced cocoon-spinning 15—16 2 Cocoon 9,5x4—12,5x5,5 (average: 10,8x4,8) 20 Adult {a) female 10—16 (average: 13,1) (74% 12-14) 39 (b) male 8—11 (average: 9,6) (81% 9—10) 21 as is the case with pulchricoma cocoons. One of the two cocoons, damaged in the excavation, was opened (on the date of excavation) and within was found a developing pupa with legs, wing buds and head clearly distinguisable. The second cocoon was kept in a geletin capsule and 39 days later yielded a male Ceropales punctulatus Cam. (Pompilidae : Ceropalinae) which had emerged after cutting olf one of the ends of the cocoon. C. punctulatus appears to be a common, widespread species and is represented in the Albany Museum by specimens from Mamathes (Lesotho), Kenton-on-Sea and various localities around Grahamstown. Dates of capture range from October to May. Ceropales species are known to be cleptoparasites on other Pompi- lidae but as far as can be ascertained no “host” wasp has hitherto been recorded for this species. Five nests examined at Hilton during the period 16.xi.1972 to l.xii.1972 were found to have been invaded for the purpose of nesting by leaf-cutting bees. In all cases the nest excava- tions were incomplete and consisted of a vertical shaft without any cells (Category A in Table I), surmounted by a turret that had barely attained the top of its arch. In Fig. 8 are shown, diagrammatically, the lengths of the leaf-nests constructed by these bees and their positions within the vertical shafts. A further pulchricoma nest, examined at Hilton during the same period, and consisting of 5 sealed cells and the beginning of a sixth cell, was found to have two disc-shaped pieces of bee-cut leaf in the vertical shaft. The bee responsible for the leaf-nest in one of the five former nests was captured, when emerging from the turret of the latter, and was identified as Megachile ( Eutricharaea ) stellarum Ckll. (Megachilidae: Apoidea). This species 204 Depth in GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENIA PULCHRICOMA Fig. 8. Positions and lengths of megachilid leafnests within five vertical shafts of Dichragenia pulchricoma (Arnold). 205 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 has a wide distribution in East and South-East Africa (Pasteels, 1965:266) and is one of the most common megachilids around Grahamstown, where its flight period, as indicated by specimens in the Albany Museum collection, is from November to March. Little is known concerning its biology but it probably nests in any suitable cavity. While it is clear that in these cases D. pulchricoma was superseded by M. ( E .) stellarum in its nests, it is not evident whether or not competition for the burrows arose — whether the bee ousted the rightful owner or whether it merely occupied the burrow left untenented following its abandonment by, or the death of, the pompilid builder. Three multicellular nests excavated at Hilton during the period l.xii.1972 to 6.xii.l972 each included one sealed cell whose contents was overrun by pale-coloured mites. Two of these cells contained prey spiders in a very poor state of preservation on which wasp eggs or larvae could not be found; the third cell contained the rotten remains of a mature larva in its cocoon. The remains of the mature larva and of one of the prey spiders were also covered with the hyphae and the small white fruiting-bodies of a fungus. FLOWERS VISITED BY ADULT WASPS Adult D. pulchricoma of both sexes have been observed visiting flowers for the purpose of imbibing nectar. The following associations of wasps and their forage flowers have been recorded: one female on Foeniculum vulgare Mill. (Umbelliferae) at Belmont Valley, Grahams- town on 24.i.l970 (C. F. Jacot-Guillarmod); three males on Maytenus linearis (L.f.) Marais (Celastraceae) at Hilton, Grahamstown on 6.xii. 1 972 (F. W. & S. K. Gess); and three females on Zizyphus mucronata Willd. (Rhamnaceae) at the Koonap River near Adelaide (C.P.) on 20-22.xii.1972 (C. F. Jacot-Guillarmod). The flowers of all three plant species are small and inconspicuous, pale yellow or greenish-yellow in colour, with nectar, secreted on a freely exposed disc, easily accessible to short-tongued wasps such as pulchricoma. DISCUSSION The ethology of Dichragenia pulchricoma (Arnold) exhibits two outstanding features which mark this species as unique, firstly within its tribe, secondly within its family. Both these features pertain to the nest — to the subterranean position of the cells and to the presence of a turret rising above the burrow entrance. As already shown, D. pulchricoma is, judged on morphological grounds, a member of the tribe Macromerini ( == Auplopodini) of the subfamily Pepsinae, and is closely related to species of the genera Phanagenia , Auplopus and Ageniella , with which it has in common certain aspects of behaviour with respect to prey selection, prey mutilation and prey transport. The four spider families (Lycosidae, Pisauridae, Sparassidae and Salticidae), species of which are, in the present paper, recorded as being the prey of D. pulchricoma are amongst the eleven families recorded as prey of the genera Phanagenia , Auplopus and Ageniella by various authors: Evans and Yoshimoto (1955:17), Kurczewski (1961 :23-4), Kurczewski and Kurczew- ski (1968a: 6-8, and 1968b: 369), Peckham and Peckham (1898:164), Richards and Hamm (1939:73), Townes (1957:143-219) and Wasbauer and Powell (1962:395). Amputation of the legs of their prey spiders at the coxal-trochantal joint, prior to the removal of the prey to the nest is a characteristic behavioural feature of the Macromerini and is recorded in Phanagenia , Auplopus and Ageniella by the above listed authors. The method of transporting the prey to the nest practised by D. pulchricoma is the same as that reported for Phanagenia by Kurczewski (1961 : 23—4), a method apparently adopted also by Ageniella but not by Auplopus (Kurczewski, 1961: 24, and 1968a: 6 — 8). The latter, though also straddling the prey, transports it venter up, grasped by the spinnerets and not dorsum up, grasped by the base of a chelicera, as is the case in the other genera. 206 GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENIA PULCHRICOMA Although there is a strong similarity between D. pulchricoma and members of the other genera in various aspects pertaining to the prey, there is a remarkable dissimilarity in the form and situation of the nest. The Macromerini are characterized by their construction using wholly foreign materials, primarily mud, of aerial nests consisting of separate but adjacent cells. In the nature of their construction these nests parallel those built by some Eumenidae and some Stenogastrinae (Vespidae). Among the species whose nesting is better known is the widespread North American Phanagenia bombycina (Cresson). An early account, based upon examination of a nest found in New Hampshire was published by the Peckhams (1898: 164 — 5). The nest in question was built under a stone and consisted of sixteen small mud cells about 15x8 mm. Townes (1957: 143) lists various records and confirms that “its nest of mud cells is placed usually under stones”. Concerning the very large genus Auplopus , Townes (1957: 145) states that probably all species nest in mud cells. Two instances of the nest-building of the North American Auplopus architectus (Say) were reported by the Peckhams (1898: 165 — 6): one set of three cells (each 8 mm long by 5 mm wide) being constructed in the folds of an undisturbed furled flag upon a porch, another set of two cells being fastened to the inside of a boat house. Concerning the same species, Townes (1957: 165) states that the cells which are made of hard clay are always “under a stone in the open, in an irregular group of usually three to five, plastered to the stone and against one another, in a place the stone happened to be raised above the soil enough to give the female space”. Other North American species of Auplopus are recorded by Townes as building their mud cells in a variety of locations such as under logs, under loose bark or in various holes and crevices including old mud nests of species of Sceliphron and Trypoxylon. Evans and Eberhard (1970: Fig. 67) published a drawing of the nest of a Philippine species. Among the few species of the genera Phanagenia and Auplopus whose nesting has been noted, some degree of variation in the placement of the mud cells thus occurs but in all cases the nests have been aerial. No record of subterranean nesting by a member of the Macromerini has been found in the literature available to the authors for consultation, and D. pulchricoma which does construct its cells in the ground thus appears to be unique within its tribe. The form of the subterranean portion of the nest of D. pulchricoma seems to approximate most to the nest of the North American Priocnemis minorata Banks which has been closely described and figured by Yoshimoto (1954). Priocnemis minorata belonging to the tribe Pepsini of the subfamily Pepsinae (which includes as its other tribe the Macromerini), constructs an open nest in heavy clay-loam soil, containing from one to seven cells arranged more or less spirally in ascending order around the common vertical shaft excavated by the wasp. Several other species of the tribe Pepsini, notably Priocnemis exaltatus (Fabr.) in Sweden and two species of Priocnemioides in Chile have likewise been observed by various authors (see Evans, 1953) to construct several lateral cells from a common burrow. However, the nest of D. pulchricoma differs from those of these behaviourally advanced Pepsini and indeed from the nests of all other Pompilidae (at least those species whose nesting has been studied and re- corded) in having the entrance to the subterranean burrow surmounted by a mud turret. The ethology of D. pulchricoma thus appears to be in some respects intermediate between that of the ethologically advanced species of the tribe Pepsini (such as P. minorata) which construct multicellular subterranean nests and that of those Macromerini (such as Phanagenia and Auplopus species) which construct aerial mud nests. However, rather than forming a connecting link between these two ethological groups, it seems likely that the ethology of D. pulchricoma represents a side branch arising at this point from the main stream of the ethological evolution of the Pompilidae. As indicated by Evans (1953) in his account of the comparative ethology of the Pompilidae, the Macromerini which build aerial nests belong to a special group arising from ground 207 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 11, MAY 1974 nesting forms which like themselves belong to an advanced ethological type characterized by the fact that nest building precedes hunting, a sequence not widespread in the Pompilidae but shown by all species constructing multicellular nests. It is clear that ethologically D. pulchricoma shared the same origin with these other Macromerini with which, as shown, it has in common the behaviour with respect to prey selection, mutilation and transport, while at the same time retaining the more conservative nesting situation. Like them, D. pulchricoma is a worker in mud but, whereas Phanagenia and Auplopus species which freed themselves of the ground use this medium to build aerial cells, D. pulchricoma uses it to build a turret with which to surmount the burrow entrance. In this connection it is of interest that the use of the pygidium as a “trowel” for smoothing the surface of the mud structure as reported for Auplopus by Evans and Eberhard (1970: 100 and Fig. 67) is not seen in D. pulchricoma which as already stated does no smoothing of its turret walls. Finally, it is interesting to note the similarity between the multicellular subterranean nests surmounted by mud entrance turrets as built by Dichragenia pulchricoma (Arnold) of the Pompilidae and those built by some Eumenidae and some Masaridae ( Ceramius species) and that in all three families the ethological evolution with respect to the situation and form of the nest is from simple nests in the ground to free aerial mud nests, a trend paralleled also in the Sphecidae. SUMMARY ] The ethology of Dichragenia pulchricoma (Arnold) (Hymenoptera: Pompilidae: Macro- merini) in the Eastern Cape Province of South Africa is described. Various facets of the nesting behaviour are dealt with but particular attention is given to the description of the form of the nest, which by possessing a mud entrance turret appears to represent a nest-type previously unknown within the Pompilidae. The nest situation and type as well as various aspects of behaviour relating to nest provisioning are compared with those exhibited by related genera and the position of the present species within the ethological evolution of the Pompilidae is suggested. ACKNOWLEDGEMENTS The authors wish to thank Mr T. C. White of the farm Hilton for his much appreciated kindness over the years in allowing them free access to his land. Similarly, they wish to thank Mr H. H. Norton of the farm Clifton for permission to investigate insects there. The success of the present investigation carried out on these two farms owes much to their owners’ co- operation. Thanks are due also to Mr C. F. Jacot-Guillarmod for helpful comments and for provid- ing useful literature, to Mr M. A. Meyer for taking the photograph reproduced on Plate 9, and to Mr R. E. Stobbs for processing many photographs. Finally, the senior author is grateful to the C.S.I.R. for a grant partly utilized to cover running expenses in connection with the field work involved in the investigation. REFERENCES Arnold, G., 1934. The Psammocharidae of the Ethiopian Region. Part 3. Ann. Transv. Mus. 15: 283 — 400. Banks, N., 1933. New Psammocharidae from the United States. Psyche, Camb. 40 (1): 1 — 19. Banks, N., 1934. The Psammocharidae of the Philippines. Proc. Am. Acad. Arts Sci. 69 (1): 1 — 117. Evans, H. E., 1953. Comparative ethology and the systematics of spider wasps. Syst. Zool. 2 (4): 155 — 72. Evans, H. E. and Yoshimoto, C. M., 1955. An annotated list of pompilid wasps taken at Blackjack Creek, Pottawatomie County, Kansas (Hymenoptera). J. Kans. ent. Soc. 28 (1) : 16 — 19. Evans, H. E. and Eberhard, M. J. W., 1970. The Wasps. Ann Arbor. University of Michigan Press. 208 GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENXA PULCHRICQMA Haupt, H., 1950. Pompilidae (Hymenoptera Sphecoidea). Explor. Parc natn. Albert Miss. G. F. de Witte Fasci- cule 69: 1 — 63. Haupt, H., 1957. Pompilidae (Hymenoptera Sphecoidea) II Teil. Explor. Parc natn. Albert Miss. G. F. de Witte Fascicule 89: 1 — 37. Jacot-Guillarmod, C., 1945. Wasps and their ways. S. Afr. Insect Life 1(1): 41 — 3. Kurczewski, F. E., 1961. Some observations and prey records of Pompilidae (Hymenoptera) from North- Eastern United States. Bull. Brooklyn ent. Soc. 56 (1): 23 — 4. Kurczewski, F. E. and Kurczewski, E. J., 1968a. Host records for some North American Pompilidae (Hyme- noptera) with a discussion of factors in prey selection. J. Kans. ent. Soc. 41 (1): 1 — 33. Kurczewski, F. E. and Kurczewski, E. J., 1968b. Host records for some North American Pompilidae (Hymenoptera). First supplement. J. Kans. ent. Soc. 41 (3): 367 — 82. Pasteels, J. J., 1965. Revision des Megachilidae (Hymenoptera Apoidea) de TAfrique Noire. I. Les genres Creightoniella , Chalicodoma et Megachile (s.str.). Ann Is. Mus. r.Afr. cent. Ser. 8 vo. (Sciences zoo- logiques) 137: 1 — 579. Peckham, G. W. and Peckham, E. G., 1898. On the instincts and habits of the solitary wasps. Bull. Wis. geol. nat. Hist. Surv. 2: 1 — 245. Richards, O. W. and Hamm, A. H., 1939. The biology of the British Pompilidae (Hymenoptera). Trans. Soc. Br. Ent. 6 (4): 51—114. Saeger, H. de, 1945. Contribution a 1'etude des Hymenopteres du Congo beige: Pompilidae. Revue Zool. Bot. afr. 39 (1): 78—114. Townes, H., 1957. Nearctic wasps of the subfamilies Pepsinae and Ceropalinae. Bull. U.S. natn. Mus. 209: 1—286. Wasbauer, M. S. and Powell, J. A., 1962. Host records for some North American spider wasps, with notes on prey selection (Hymenoptera: Pompilidae). J. Kans. ent. Soc. 35 (4): 393 — 401. Yoshimoto, C. M., 1954. A study of the biology of Priocnemis minorata Banks (Hymenoptera, Pompilidae). Bull. Brooklyn ent. Soc. 49 (5): 130 — 8. 209 Plate 1. Hilton, 15.xi.1972. Excavation of a nest situated on the side of a small erosion gulley. Plate 2. Hilton, l.iii. 1 973. Situation of a nest (marked by arrow) on raised bank of earth. Note rain water pool, sedges and grasses. 210 Plate 3. Hilton, 9.H.I973. Situation of a nest (marked by spray-can in lower left corner) at a step in the ground level of an eroded area. Note rain water pool. Plate 4. Hilton, 12.xi.1972. Situations of nests (marked by figures) on bare earth path running parallel to water furrow. 211 Plate 5. Hilton, 23.xi.1972. Newly begun turret in bare sheet-eroded area. (Burrow opening at lower right is that of an eumenid.) (x 1) Plate 6. Hilton, 23. xi. 1972. Well-built turret of typical form surmounting the underground workings shown in Fig. 2. ( x 1.2) 212 GESS AND GESS: AN ETHOLOGICAL STUDY OF DICHRAGENIA PULCHRICOM A Plate 7. Hilton, 9.ii. 1 973. Turret of nest, the situation of which is shown in Plate 3. (x 1) Plate 8, Hilton, 9.ii.l973. Turret of nest constructed next to base of dwarf-shrub, (x 0,7) 213 ANN. CAPE PROV. MUS. (NAT. HIST.) VOIJ9 PT 11, MAY 1974 Plate 9. Young larva of Dichragenia pulchricoma (Arnold) in feeding position on abdomen of a lycosid spider, (x 6) 214 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. (nat. Hist.) « VOLUME 9 • PART 12 15th MAY 1974 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA On the life colours and nuptial tubercles of Oreodaimon quathlambae (Barnard, 1938) (Pisces, Cyprinidae) by P, H. SKELTON Albany Museum INTRODUCTION There have been several papers in recent years concerning the rare and unique cyprinid fish Oreodaimon quathlambae. Although Greenwood and Jubb (1967) have given a detailed account of the morphology and osteology of the species, certain aspects remain outstanding. In the original description, Barnard (1938) noted the colour of the preserved fish and also the fact that conical tubercles occurred on the head in both sexes. Both these aspects were poorly described and have not since been elaborated on. In January 1973, several specimens of O. quathlambae were collected by the writer whilst on a trip to the Sehlaba-Thebe Wildlife Sanctuary and National Park, Lesotho. A colour description has been prepared from colour transparencies, and notes taken from the living and freshly killed fish. Some of the specimens exhibited relatively well-developed nuptial tubercles, and in view of the doubtful relationships of O. quathlambae (Greenwood and Jubb 1967) a record of the tuberculation is therefore presented here. MATERIALS AND METHODS All the specimens are from the Tsoelikana River, south-east Lesotho, where the only known extant population of O. quathlambae exists. (Recent searches for the species in the type locality, the Umkomazana River, Natal, have not yielded specimens). Six specimens were sent to the Albany Museum by Mr T. Pike of the Natal Parks Board, a co-collector with Mr A. J. Tedder the Fisheries Officer for Lesotho. One of these specimens was collected in November 1970, and the other five in April 1972. A further seven were caught by the writer in January 1973. The fish were collected either by means of an electric shocker (Pike and Tedder) or with a 6 meter “anchovy mesh” minnow seine. Colour transparencies and notes were made of the living and freshly-killed fish for the colour description. The nuptial tubercles were studied by means of a conventional M5 Wild Stereo Microscope and a JEOL JSM U3 Scanning Electron Microscope. LIFE COLOURS Figure 1 shows four specimens of O. quathlambae shortly after being killed in 10% formalin. The top three fish are females and the fourth is a male. Degree of pigmentation is the only noticeable sexual difference in respect of colouration. Viewed from above, the dorsal surface is of a uniform dirty olive-brown (bluish-olive in breeding males, Figure 2), with scattered dark spots, of about 1,5 mm diameter, on the body. In certain fish these spots are arranged more regularly, in two or three, longitudinal rows. 215 Fig. 1. From the top, three females and a male O. quathlamhae . Tsoelikana River, Lesotho, January 1973. Fig. 2. Dorsal surface of male with nuptial tubercles. SKELTON: LIFE COLOURS AND NUPTIAL TUBERCLES OF OREODAIMON QUATHLAMBAE There is considerable variation in the number of spots per individual — from about 15 to 57 with an average of between 30 to 35. In some fish small whitish patches occur at the anterior and posterior edges of the dorsal fin base. In males with tubercles, those on the head and scales produce a fine spotted effect, emphasized by the deep pigmentation of these individuals (Figure 2). The dark lateral band evident in preserved material (Barnard 1938) is seldom so evident in living fish. In only one of the seven live fish observed by the writer was this an obvious feature. The band, obvious or faint, was found to be broken into a series of dashes, with no posterior expansion on the caudal peduncle. Thus, in this respect, the fishes studied here appear to differ from the description given by Barnard (1938) for this band to “end as a spot” on the caudal peduncle. The lateral line forms an approximate limit between counter-shaded dorsal and the lighter lower surface (Figure 1). Proceeding from the countershade the colour passes through a pale golden tinged olive to a blemished silvery white and finally a brilliant white on the belly. On the head the dorsal countershading encompasses the eye and the dorsal half of the opercular area. The operculum has an underlying golden green iridescence. The ventral sur- face of the head is brilliant white from below a line extending from above the angle of the mouth around and below the eye and across the operculum. Bright orange-red flashes occur in the axils of the pectoral and ventral fins and laterally along the base of the anal fin. Figure 1 shows the extent of development of these flashes. The flashes are rather sharply demarcated and are not placed on the fins themselves. The fins are however, faintly flushed towards the anterior basal angles. The flash in the region of the ventral fins dorsally circumscribes the base of each fin and crosses the ventral surface of the body just posterior to the bases of the fins. The anal fin flash is a narrow lateral band on either side of the base of the fin. The dorsal and caudal fins are pigmented a chocolate brown, darker at their bases and lighter at their outer edges. The eye has a bright silver iris with a round black pupil. NUPTIAL TUBERCLES Barnard (1938) noted that mature specimens of both sexes of O. quathlambae , in ripe condition, had warts or tubercles on the dorsal surface of the head. Examination of the present material showed that the tubercles were not limited to the dorsal surface of the head and as the distribution of nuptial tubercles is of systematic importance (Wiley and Collette 1970) a more detailed record of their distribution and development has been prepared. The classification of developmental stages of nuptial tubercles applied as by Lachner and Jenkins (1971) to the genus Nocomis is used in this description. The developmental stages are categorized by these authors as follows: spots: small, round, light to grey areas in the deeper epidermis; buds: small pimples having round to pointed tips; tubercles: the completed growth stage following budding, i.e., hard, cornified structures that attain their greatest development more or less synchronously with the (spring) reproductive period of the mature males; tubercle scars: shallow to moderately deep, discoloured cavities resulting from the loss of the nuptial tubercle. Two of the seven specimens caught in January 1973 were males showing relatively well developed nuptial tubercles. Males caught in April 1972 showed development of spots and buds on the posterior dorsal surface. In these latter specimens there is no tubercle development on the scales or fins. Females caught in April 1972 show no tubercle development; however, those caught in January 1973 minute tubercles are present on the head and scales. These are perhaps best classified as buds for although barely visible to the naked eye, microscopical 217 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 12, MAY 1974 examination shows them to have a conical structure. There is no tubercle development on the fins of any of the females examined. The following account of the distribution and development of nuptial tubercles in O. quathlambae is taken from the two tubercule-bearing males caught in January 1973. Figure 2 shows a dorsal view of the larger of the two specimens. The distribution of tubercles for these specimens is as follows: Head: A group of four to five well-developed tubercles is present anterior to each nostril in the larger specimen. On the dorsal area enclosed by a line between the nostrils, the supra- orbital canal and postocular commissure of the cephalic lateral line system and the posterior margin of the head, the tubercles are scattered more or less in groups. There are also a few scattered tubercles along the posterior and postero-ventral border of the preoperculum and latero-ventrally on the operculum. Body: Tubercles are found on all the scales (Figures 2, 4). There is usually only one tubercle per scale but two or three may also be encountered, particularly in the area just anterior to the dorsal fin. Tubercles also occur on the scales of females. Fins: On the pectoral fins there are bands of tubercles (figure 5) overlying all except the leading fin ray. In depth each band contains up to six tubercles but the numbers diminish towards the ends of the bands with a corresponding tapering effect. The bands begin soon after the ray emerges from the axil and do not reach the outer border of the fin. The tubercle bands are on the dorsal side of the fin only. Tubercles also occur on the dorsal, ventral and caudal fins. Both surfaces of these fins have tubercles but in the paired ventral fins development is more pronounced on the dorsal surface. Generally, a single row of tubercles is found overlying each finray, this row dividing with the ray. Tubercle development on the caudal fin is relatively minor. Size: Size was measured from the electron micrographs and in the case of the larger tubercles with fine calipers under the stereo microscope. The largest tubercles measured are those on the antero-dorsal surface of the head in the larger of the two male specimens. These are up to 0,4 mm across the base with an equal height. Figure 3 shows an average-sized tubercle 0. 12 mm across the base. Scale tubercles (figure 4) are smaller, averaging 0,07 mm in basal diameter. On the pectoral fins the bands of tubercles measure up to 0,25 mm in width (Figure 5). Those nearest the leading edge of the fin in each band are the largest in that band. Band tubercles are laterally compressed for more efficient accommodation in rows (Figure 5). Along the long axis of the base the largest band tubercles measure approximately 0,12 mm. The tubercles along the trailing edge of the band are a third to a half the size of the leading edge tubercles, 1. e., approximately 0,05 mm along the long basal axis. Development: Males collected in April 1972 show minor tubercle development. Spots and buds are present on the posterior dorsal part of the head. There are no obvious scars, no sign of tubercle development on the scales or fins. Of two females collected at the same time one has spots on the dorsal head surface. A female collected in November 1970 has buds or small tubercles on the head and there is microscopic tubercle development on the scales. All five females collected in January 1973 have buds or small tubercles on the head and three of the five have scale tubercles. The development of the tubercles of the two males caught in January 1973 varies considerably. The two specimens measure 85 and 72 mm (standard length) respectively, indicative of at least a year class difference. The pattern and development of the tubercles of the larger specimen is shown in Figure 2. In this specimen the head tubercles are best developed anteriorly. In the second the head tubercles are located on the posterior dorsal surface, and there are no tubercles on the snout. 218 Fig. 3. Scanning Electron micrograph of male head tubercle (partially collapsed in preparation) Fig. 4. Scale tubercle (male) Fig. 6. Male head tubercle scar. SKELTON: LIFE COLOURS AND NUPTIAL TUBERCLES OF OREODAIMON QUATHLAMBAE The head tubercles of the smaller male measure up to 0,12 mm in diameter of the base — con- siderable less than that of the larger fish (cf. above). This indicates an agreement with the find- ings of Lachner and Jenkins (1971) where in certain species of Nocomis there is an increase in tubercle number and size with increased length of fish. An interesting finding is that scars (Figure 6) are more common on the smaller male. The gonads of the larger male are relatively better developed than those of the smaller. Catching operations may be responsible for the greater number of scars of the smaller fish but it is also possible that this specimen, caught about four kilometres upstream from the larger one, might have already partially milted. The implications being that the seasonal, but not absolute, development of tubercles in the smaller individual was more advanced. GENERAL Wiley and Collette (1970) have pointed out how functionally significant structures like nuptial tubercles can be of value to systematic studies. Similarities and differences in behaviour patterns from which relationships can be inferred. The three primary functions attributed to nuptial tubercles as listed by Wiley and Collette are: maintenance of body contact between the sexes during spawning; defence of nests and territories; and stimulation of the females in breeding. No doubt the spawning behaviour of O. quathlambae will when known, give meaning to the development described above and, in conjunction with similar studies on other species, shed more light on the relationships of this interesting species. Although the degree of development or patterns of distribution of nuptial tubercles have not been considered in southern African cyprinid taxonomy, certain species have been separated using as a major character the presence or absence of such tubercles (Barnard, 1943; Jubb, 1967). Because nuptial tubercles develop concurrently with the breeding season caution is necessary when considering the use of tubercles in taxonomic studies. For example the difference between Barbus anoplus and B. motebensis rests largely on the absence of nuptial tubercles in breeding males of the former species and their presence in the latter. However recent indications suggest that only a single polymorphic species is involved and that a specific distinction on the basis of nuptial tubercles is doubtful (Jubb. pers. comm.). Similarity of red fin flashes in O. quathlambae with the “red fin” group of Barbus species from the southern and south-western Cape Province of South African might suggest some relationship. However the red flashes of O. quathlambae are unlike those of the “red fin” Barbus species in that they do not impinge on the fin membranes. Red fin flashes are known in many other cyprinid species from widely separated geographical areas — e.g., Notropis effusus in North America, Barbus tetteyi in Sri Lanka (Ceylon), suggesting that the character has some functional significance. Thus it is likely that the red fin flashes have evolved separately on different occasions. However this does not preclude a relationship between O. quathlambae and the “red fin” Barbus species, and further investigations particularly to elucidate the sig- nificance of the red fins, is necessary. ACKNOWLEDGEMENTS The assistance of the following persons is gratefully acknowledged: Mr C. F. Jacot- Guillarmod, Director of the Albany Museum and Editor of the Annals of the Cape Provincial Museums for reading and publishing this paper; Dr F. Getliffe, from University of Natal, Durban, for providing the colour transparencies of O. quathlambae ; Mr R. C. Cross, Rhodes University, for assistance in the electron microscopic studies; Mr T. Pike, Natal Parks Board, for details concerning the rediscovery of O. quathlambae and for supplying study specimens and colour transparencies; Mr A. J. Tedder, Fisheries Officer for Lesotho, for assistance with 221 ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 12, MAY 1974 regard to the collecting trip to the Sehlaba-Thebe Wildlife Sanctuary and National Park in Lesotho; Dr R. A. Jubb Associate Research Ichthyologist, Albany Museum, for numerous valuable discussions and for reading the manuscript; Mr J. C. Greig, Research Officer, De- partment of Nature Conservation, Cape, for reading the manuscript; Mrs B. Mostert for typing the manuscript. REFERENCES Barnard, K. H., 1938 A. Description of a new species of fresh-water fish from Natal. Ann. Natal Mus., 8 (3): 525—8. Barnard, K. H., 1943. “Revision of the indigenous freshwater fishes of the S.W. Cape region”. Ann. S. Afr. Mus. 36 (2): 101—262. Greenwood, P. H. and Jubb, R. A., 1967. “The generic identity of Labeo quathlambae Barnard (Pisces, Cyprinidae)”. Ann. Cape Prov. Mus. ( nat . Hist), 6 (2): 17 — 37. Jubb, R. A., 1967. Freshwater fishes of Southern Africa. A.A. Balkema, Cape Town/ Amsterdam. Lachner, E. A. and Jenkins, R. E., 1971. “Systematics, distribution and evolution of the Nocomis biguttatus species group (Family Cyprinidae :Pisces) with a description of a new species from the Ozark upland.” Smithsonian Contrib. Zoo/., 91: 1 — 28. Pike, T. and Tedder, A. J., 1973. “Rediscovery of Oreodaimon quathlambae (Barnard)”. Lammergeyer 19: 9—15. Wiley, M. L. and Collette, B. B., 1970. “Breeding tubercles and contact organs in fishes: their occurrence, structure and significance”. Bull. Am. Mus. nat. Hist., 143 (3): 143 — 216. PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN (o r> { \ i » 101 3'3'i I ANNALS OF THE CAPE PROVINCIAL MUSEUMS NATURAL HISTORY Ann. Cape Prov. Mus. ( nat . Hist.) VOLUME 9 • PART 13 15th MAY 1974 PUBLISHED JOINTLY BY THE CAPE PROVINCIAL MUSEUMS AT THE ALBANY MUSEUM, GRAHAMSTOWN SOUTH AFRICA The Trichoptera of the Sundays and Fish Rivers, Eastern Cape Province, South Africa by K. M . F. SCOTT National Institute for Water Research (Albany Museum, Grahamstown, South Africa) ABSTRACT The impoverished Trichoptera faunas of the Sundays and Fish rivers (Eastern Cape Province, South Africa) are described and discussed in relation to the harsh environment and climatic conditions prevailing in the area. This work formed part of the Orange River Project, with particular relevance to the effects of the future inflow of Orange River water into these two rivers, which at present are highly mineralized and largely or entirely seasonal. INTRODUCTION The work described in this paper formed part of several more general surveys of the Sundays and Fish rivers, and was carried out under the auspices of the Orange River Project (see Forbes 1968, Forbes & Allanson 1970a & b, Scott, Allanson & Chutter 1972). Its aims were, firstly, to determine the present Trichoptera (caddisfly) faunas of the Sundays and Fish rivers, and to establish whether or not there are any differences between them. Secondly, it was hoped to gauge the effects of brackish water and exposure to alternate droughts and floods upon the Trichoptera, since both these rivers are subject to such pheno- mena. In this they are typical of the greater part of the Eastern Cape Province. At present both rivers only flow intermittently after rain, excepting for that part of the Sundays which lies between Kirkwood and the sea (see Fig. 1), where there is a very small permanent flow. Once the first stage of the Orange River Development Scheme has been completed, some 1 460 cusecs (41,3 cumec) of water are scheduled to flow through the tunnel from the Hendrik Verwoerd Dam into the Fish and Sundays rivers, providing much-needed water for their irrigation schemes and enabling these to be expanded. This inflow of water should change both rivers below the intake to perennial rivers, though the quality of the water flowing through them will depend to a large extent upon the way in which the irrigation water is distributed. The Orange River water is likely also to bring with it members of its own fauna and flora, including Trichoptera, and it is expected that this investigation will be continued at a later date to discover what changes in the Trichoptera fauna have been brought about both by the influx of other species and by the changes in water quality. The survey was commenced in August 1967 and ended in March 1970, during most of which period there was a severe drought. Shortly after the termination of the work the drought was broken by heavy rains which caused floods in the Fish River. The Sundays received less rain at that time, but a year later, in August 1971, catastrophic floods occurred in the Sundays, Fish, Gamtoos and Swartkops rivers, during which much scouring of the river beds took place 223 SCOTT: THE TRICHOPTERA OF THE SUNDAYS AND FISH RIVERS and adjacent lands were deeply inundated. Later visits were therefore paid to the Sundays River in October and November of that year to evaluate the effects of the floods on the Trichoptera population. GENERAL DESCRIPTION OF THE SUNDAYS AND FISH RIVERS The Sundays and Fish rivers lie in adjacent catchments in the Eastern Cape, in an area bounded to the north by an escarpment formed from soft sedimentary rocks of the Karoo System. Much of the coastal plain and other mountain ranges involved is composed of the same soft rocks. Both rivers enter the ocean to the east of Port Elizabeth (see Figs. 1 & 2). Both catchments lie in the rain shadow of the coastal ranges and are therefore arid with low precipitation (125-500 mm per annum) and very high evaporation rates (1 250-2 000 mm p.a.). In addition, the rainfall is seasonal, with long rainless periods even when there is no drought. As a result of these factors the upper reaches of the Sundays and Fish and their tributaries seldom flow except in spate after heavy rain or snow, and when they do their waters are heavily laden with silt. Several irrigation dams have been built on their courses, and these tend to trap much of the silt, so that at times of low flow the river water may be clear. Most of the underlying rock formations are salt-bearing to a greater or less degree, producing brackish or saline ground waters, which together with the arid climate, result in correspondingly brackish river waters, which are further mineralized by irrigation water draining from adjacent farm lands. The waters of the entire Fish River, and of the Sundays as far down as Kirkwood, are brackish but potable; the lower reaches of the Sundays below Kirkwood, however, flow over marine beds, becoming increasingly saline and finally unusable for domestic or farming purposes. The geology and geography of this area and the physico- chemical condition of these rivers have been described in various papers (see Bond 1946, Forbes 1968, Forbes & Allanson 1970a, Scott et al. 1972), to which reference should be made for further information. The collecting stations used are shown in Figures 1 and 2, and briefly described in Appendix I; months in which collecting trips were made are given in Appendix II. Many of those expeditions were fruitless, as has been indicated in the text. THE TRICHOPTERA POPULATION OF THE SUNDAYS RIVER (1967-70) DISTRIBUTION OF THE TRICHOPTERA In the upper part of the catchment, above Lake Mentz, it was evident that the streams were too ephemeral and the country too arid for Trichoptera to survive, the only ones found being a few Hydroptilidae at Station 10, where there was water from apparently permanent springs. Even there they were not common and were found only once, in February 1968. The major obstacle faced at Station 10 was probably that of isolation from any better populated water body, because temporary waters in the northern hemisphere usually support large populations of Trichoptera. Those lie, however, in much better watered country, so that dispersal is easy. There was also permanent water at Station 9, immediately below the town of Graaff- Reinet, but during the drought years this appeared to consist mainly of sewage effluent and no Trichoptera were found in it, although it supported a fair population of more tolerant insects such as Hemiptera, Ephemeroptera, Odonata and Coleoptera. The water level in Lake Mentz was very low during the early part of the study period, and the lake dried up completely in February 1968; no Trichoptera were found there at any time. Below Lake Mentz, at Stations 7 and 6, a few Trichoptera were collected in 1967 but 225 TABLE 1. WATER ANALYSES: SUNDAYS RIVER (From Forbes 1968) Except for pH and bottom line, all values are in mg/£. ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 13, MAY 1974 * * 226 Trichoptera were collected from these stations on these dates. In the case of station 3, however, the single larva caught was un- doubtedly drift. Trichoptera were taken at these stations, but not on the same date as the chemical sample. SCOTT: THE TRICHOPTERA OF THE SUNDAYS AND FISH RIVERS none subsequently, as water in those stony runs was dependent on irrigation flows, which ceased early in 1968. Those found were larvae of Cheumatopsyche afra (Mosely) and Ecnomus thomasseti Mosely. Stations 6, 5, 4A and 4 lay on the lower course of the river where the water became very saline, the TDS (total dissolved solids) values from samples taken ranging between 800 and 1 000 mg/f at Station 6, 1 800 to 4 600 mg /£ at Station 5, and 5 500 to 9 300 mg /£ at Station 4A during the earlier part of the survey (Forbes 1968). Supplementary analyses made in May 1970 gave TDS values of 815 mg/£ at Station 6, 5 403 at Station 5 and 12 370 at Station 4A (Scott et a/. 1972) (see table 1). Station 5 was the uppermost sampling point at which permanent water was present, and the only one which appeared to remain habitable by Trichoptera throughout the period. Caddisflies evidently persisted there as a small resident population even though they could not always be found when samples were taken; this might well have been due, in part at least, to emergence of imagos, but the mineralization of the water could also have been implicated, as, for example, in March 1970 when none was found and flow in the river was almost non-existent. Unfortunately no chemical analyses were carried out at that time, but two months later, in May 1970, TDS values had risen sharply to 5 403 mg/f, some 800 mg/f higher than the highest value at which healthy living Trichoptera were found at any station on the Sundays or Fish rivers. The Trichoptera at Station 5 included four species, of which Ecnomus thomasseti was present in fair numbers in the years 1967 to 1969, numbers being higher in summer (February) than in mid-winter (July); in late winter (August) none was found. Numbers dropped con- siderably in 1969 and the species was not encountered subsequently. E. forbesi Scott, a new species described from this river (Scott 1968) was rare, only one male being found at this station in August 1967. Cheumatopsyche afra was also present in moderate numbers in the years 1967 to 1969; it too was more plentiful in summer than in winter. The next station downstream, 4A, produced a very small collection of Trichoptera in the winter of 1967, comprising a single male of the new species, E. forbesi, and two Ecnomus larvae, probably of the same species, and a single larva of C. afra. Thereafter conditions rapidly worsened as the drought continued, and no more caddis were found. At the stations below 4A TDS values rose steeply, and the sole caddis larva found lower down, at Station 3 (upper estuary), was undoubtedly drift. The more obvious reasons for the limitation of any sizeable population of Trichoptera to the vicinity of Station 5 seem to be twofold. In the upper reaches, in fact above Station 5, the primary reason appears to be the transitory flow of the water, due in part to the drought and in part to the farming methods practised and abstraction of water for irrigation. In the lower reaches, the very high TDS values (arising from the geology and aridity of the area but intensified locally by the irrigation of citrus orchards) are evidently the limiting factors. Even at Station 5 itself, it is possible that the water might have been too highly mineralized to support a population of Trichoptera had it not been for the seasonal inflow of sweet water from the Witrivier tributary shortly above it. No Trichoptera were, however, found in the lower part of the Wit (Station 5A), probably due to its ephemeral nature. Its upper reaches in the Suurberg mountains were somewhat inaccessible and were not visited. When apparent disappearance of the population took place, survival may have been by means of diapausing eggs, or diapausing larvae, or recolonisation may have taken place from elsewhere, the only possible source being the upper reaches of the Wit. There is at presence no evidence as to which of these might be concerned. In the few Trichoptera from cold or temperate climates so far studied, diapause occurs in late instar larvae (Hynes 1970). Hynes also found that many insect larvae, including some Trichoptera, take refuge in the substratum, down to depths of several feet. In the Sundays River, however, there are no 227 r « i2*ii*7i SCOTT: THE TRICHOPTERA OF THE SUNDAYS AND FISH RIVERS depths of gravelly well-aerated substrata, spaces between stones frequently being clogged with sand or mud which may be anaerobic a few centimetres below the surface. Harrison (1966) on the other hand, working on the recovery of Rhodesian streams treated with mol- luscicides, found that in the case of Cheumcitopsyche species re-invasion by adults from neighbouring streams took place; he found no evidence of diapausing eggs. SPECIES COMPOSITION OF THE POPULATION The Trichoptera of the Sundays River comprised only four species in three genera, of which one, the Hydroptilid Hydroptila sp., was only found alive at Station 10, and another, Ecnomus forbesi , was rare. The other two species, Ecnomus thomasseti and Cheumatopsyche afra , both occurred at Station 5 in moderate numbers, and were evidently able to grow well and reproduce even at relatively high TDS values (up to 4 600 mg/f TDS, 1 480 mg/f Na and 1 748 mg/f chloride). Both are eurytopic species which have been recorded from very diverse habitats in both alkaline and acid rivers in Southern Africa. E. thomasseti has been recorded from the Western Cape Province to South West Africa and Rhodesia, and C. afra from the Western Cape through to Ethiopia in the east and Sierra Leone in the West. C. afra is, however, variable, and may eventually prove to include a complex of closely related species. The Hydroptilid found was a species of Hydroptila whose larva makes a purse-like case ornamented with algal threads. Two South African species are known to do so, El. capensis Barnard and H. cruciata Ulmer, but at present it is impossible to distinguish between their larvae. It might have been expected that the low water levels and minimal currents of summer would affect Cheumatopsyche afra adversely, as Hydropsychid larvae tend to prefer areas of rapid water movement where they spin their nets in the current. This did not, however, appear to be the case, though numbers were never high. Ecnomus species (family Psychomyiidae) were usually found out of the main current under stones or in vegetation. All Trichoptera collected were normal and healthy in appearance. THE TRICHOPTERA POPULATION OF THE FISH RIVER (1964—1970) The Trichoptera collected in the Fish River included Cheumatopsyche thomasseti Ulmer, Ecnomus thomasseti and a single larva of a species of Cheumatopsyche near maculata. (C. maculata Mosely occurs fairly frequently in some of the other streams in this region.) Only the first of these was found in any quantity. C. thomasseti larvae were numerous at times at Stations 3, 4 and 5 early on in the survey, much less so at Stations 1, 2, 6, 7 and 12. Later, as the drought increased in severity and the river dried up, they disappeared except at Station 6, at which they were still present in August 1969. A few Elydroptila sp. larvae, similar to those found at Letskraal on the Sundays River, were collected at Station 3 on the Fish, also in February 1968, as were a few E. thomasseti , but no caddis were found there subsequently. As was the case in the Sundays River, these Trichoptera proved able to survive under remarkably poor conditions, living caddis larvae being found in water with TDS values ranging from 511 mg/f (Station 9) to 3 368 mg/f (Station 3), and calcium values ranging from 27,7 mg/f (Station 9) to 115 mg/f (Station 4). Calcium values have been particularly mentioned because Trichoptera larvae from all the Fish River stations except 1 and 10 were more or less heavily encrusted with a white casing of calcium carbonate; the Carlisle Bridge and Lake Arthur specimens were only lightly powdered with it. Larvae found at Stations 3, 4 and 5 were particularly badly affected, often with head and eyes completely covered and gills matted with calcareous material. Trichoptera were abundant at the time, and most were heavily coated, only a few showing little or no trace of the deposit. Ca values when the collections were made ranged from 89—115 mg/f. 229 TABLE 2. WATER ANALYSES: FISH RIVER (Allanson: pers. comm.) Except for pH and temperature, values are in mg/£. ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 13, MAY 1974 Cl Feb.* 1965 22,0 8,8 1 1 1 1 o July* 1964 ITi 8,8 877 46,8 230 3,5 ON July* 3964 1 9,0 m 27,7 0 1— « 3,3 NO July* 1964 iry c-*' 8,5 2 906 47,2 810 3,8 Feb.* 1965 27,0 8,6 3 038 102 735 00" July* 1964 10,5 o6 3 029 58,5 830 NO^ in Feb.* 1965 26,3 00" 2 857 m 445 in c-' July* 1964 13,5 8,5 2 717 89,3 580 5,6 ro Feb.* 1965 24,0 8,3 1 1 1 1 July* 1964 11,5 00' 3 368 103,7 655 00 no” Cl July* 1964 i 1 3 311 0 00 0 OO on C-' • • d o os Q U o C/0 - c - B - A SECTION LOOSE RED SAND SLIGHTLY CALCIFIED RED SAND CALCIFIED RED SAND - PLATEY AT TOP CALCIFIED RED SAND - BLACK NODULES ARTEFACTS SILICEOUS LAYER - RUBBLE AT BASE BEDROCK from B. H. RELLY Figure 3 CULTURAL MATERIAL The artefact types represented in the assemblage are listed, with their frequencies, in Table L The terms used follow current general usage or are self-explanatory; the definition of “core- axe” is as given by Clark (1963: 50). TABLE I LIST OF ARTEFACT TYPES Core-axes 6 Handaxes 3 Cleavers 1 Lanceolate biface 1 Core-choppers 3 Core scrapers 2 252 Figure 4 1 and 2: Handaxes. 3: Cleaver. 4: Lanceolate biface. 5: Core-axe. ANN. CAPE PROV. MUS. (NAT. HIST.) VOL. 9 PT 15, DECEMBER 1974 Bifacial tools 3 Discoids 1 Scrapers 2 Bifacial fragments 3 Flakes 9 Cores 3 37 Cove-axes (6) (Fig. 4:5; Fig. 5: 1—4; Fig. 6:1) Of the six core-axes, one can be classed as “divergent edged" (Fig. 5: 2) while the rest are “convergent edged”. The divergent core-axe has been worked unifacially while the others are all bifacial. The convergent bifacial core-axes all have distinct chisel shaped working edges carefully prepared by bifacial flaking; these working edges are slightly curved when viewed end on. Five of the core-axes are fresh; the sixth is slightly rolled. All the core-axes are made from quartzite. The dimensions are as follows (in mm): L. B. Th Divergent 120 63 34 (flake) Convergent 141 63 33 (core) 128 73 29 (flake) 112 53 34 (flake) 113 58 28 (flake) 124 80 33 (flake) Handaxes (3) (Fig, 4: 1 and 2) Three handaxes were recovered: two fresh and one heavily rolled. All are made from quartzite. The dimensions are as follows (in mm): L. B. Th 116 67 33 98 58 29 94 58 32 Cleavers (1) (Fig. 4: 3) One diabase cleaver was recovered in the assemblage. It is in a fresh condition but the cleaver edge is slightly damaged ; this damage appears to be recent and so a dotted line has been added to the Fig. to show the probable original outline of the implement. The cleaver was made on a thick side-struck flake. The dimensions are 153 x 81 x 38 mm. Lanceolate Biface (1) (Fig. 4: 4) One very rolled lanceolate-shaped bifacially worked artefact was recovered. Details of the flaking are largely obscured by the rolling, but it appears to have been relatively crudely made. The biface is made from quartzite and measures 118 x 48 x 29 mm. 254 Figure 5 J - — 4 : Core-axes Figure 6 1 : Core-axe. 2 and 3: Flakes. 4: Scraper. HUMPHREYS: COMMENTS ON THE OCCURRENCE OF CORE-AXE-LIKE ARTEFACTS Core-choppers (3) (Fig. 7: 3 and 4) The core-choppers are core tools with a few flakes removed bifacially so as to produce a zig-zag “chopping” edge. The edge opposite to the chopping edge is flat and can comfortably be held in the hand. The core-choppers are made from quartzite and their lengths taken at right angles to the chopping edges are 91, 75 and 71 mm. Core scrapers (2) These consist of thick cores with one edge showing steep scraper retouch; they are made from quartzite. The dimensions are as follows (in mm): L. B. Th 91 62 41 91 66 36 Bifacial tools (3) (Fig. 7 : 2) Three artefacts showing bifacial working were recovered. Two of these are so heavily weathered that details of flaking cannot easily be distinguished. The third artefact is fresh and shows crude flaking along one edge from one surface and a few flakes along the edges of the other surface; it has a hand-axe-like shape, but no effort seems to have been made to produce a pointed end. All three are made from quartzite. The dimensions are as follows (in mm): L. B. Th 138 70 30 115 67 30 114 56 20 Discoids (1) One bifacially worked discoid was recovered; it is made from quartzite and is 96 mm in length. Scrapers (2) (Fig. 6: 4) Two scrapers made on side-struck flakes were recovered ; they have convex scraping edges opposite to the flat butt of the flake on which they were made. One is made from diabase while the other is from quartzite. The lengths are 131 and 125 mm. Bifacial fragments ( 3) These are bifacially worked artefacts representing broken implements-— two are possibly portions of handaxes. One is made from diabase and the other two from quartzite. Flakes (9) (Fig. 6: 2 and 3) A total of nine quartzite flakes was collected. All but three are heavily rolled. Striking platforms can be distinguished on six of the flakes and all are plain. The flakes range in length from 47 to 94 mm. Cores (3) (Fig. 7:1) One diabase and two quartzite cores were recovered. Two of the cores can be classed as disc cores while the third is a single platform core. 257 Figure 7 1 : Core. 2: Bifacial Tool. 3 and 4: Core Choppers. HUMPHREYS: COMMENTS ON THE OCCURRENCE OF CORE-AXE-LIKE ARTEFACTS DISCUSSION The assemblage described above is small and is composed of artefacts selected from the exposed walls of a diamond prospecting pit. The assemblage is therefore of no significance as a representative archaeological sample. However, as mentioned earlier, the assemblage is of interest because it includes several artefacts which can be classified as core-axes, and it is for this reason that it has been described here. Apart from the added interest of the core-axes, there is no doubt that Cyrus I is a site with considerable potential in the study of the archaeology and palaeo-ecology of the “Early Stone Age” in the interior of South Africa, and it is therefore worth placing its existence on record. J. D. Clark (1963: 50) seems to have introduced the term “core-axe” to Southern Africa, In his definition of core-axe, Clark put forward the suggestion that these artefacts were as- sociated with woodworking. The occurrence of core-axes in the Sangoan and Lupemban Industrial Complexes and their apparent distribution in the forest and woodland areas of equatorial and south Central Africa seemed to support this interpretation of the function of these artefacts. Environmental change at Kalambo Falls and the related cultural change from Acheulean to Sangoan provided further support for the interpretation of core-axes as wood- working tools (Clark 1964). In 1967, however, MacCalman and Viereck (1967) reported a site with core-axes and lanceolate points from the middle of South West Africa. They described the site as being of Lupemban affinities possibly related to the Angolan sites some 1 600 km to the north; they suggested that their site might represent “an isolated outlier of the Angolan Lupemban”. The discovery of core-axes in the Northern Cape is not quite so easy to explain. Whatever climatic conditions might have been prevailing when the Cyrus I artefacts were made the environment would hardly have been comparable with that in the Sangoan-Lupemban areas to the north. Even during “wet” conditions where there might have been 150% of the present rainfall the area would only have supported temperate mixed grassland, according to Cooke (1964). Although Cyrus I is at present an isolated occurrence, it is clear that it will necessitate some rethinking on the identification and definition of core-axes and the interpretation of their possible function. It seems that the correlation between core-axes and particular industries and environments will have to be reviewed in the light of Cyrus 1 and even of the Peperkorrel site in South West Africa. ACKNOWLEDGEMENTS 1 Thanks are due to Ray Inskeep of the University of Oxford for sending the assemblage to the Alexander McGregor Memorial Museum, and to Dr. Bruce Relly who discovered the artefacts and who also provided notes on the geology of the site from which some of the information contained in this paper was drawn. REFERENCES Clark, J. D., 1963. Prehistoric Cultures of Northeast Angola and their Significance in Tropical Africa. Diamag Pub/. Cult. 62 Clark, J. D., 1964. The Influence of Environment in Inducing Culture Change at the Kalambo Falls Pre- historic Site. S. Afr. Archaeol. Bull. 19 (76): 93 — 101. Cooke, H. B. S., 1964. The Pleistocene Environment in Southern Africa, in Davies, D. H. S. (ed) Ecological Studies in Southern Africa. Junk, The Hague. MacCalman, H. R. and Viereck, A., 1967. Peperkorrel, A Factory Site of Lupemban affinities from Central South West Africa. S. Afr. Archaeol. Bull. 22 (86): 41 — 50. 259 PRINTED BY CAPE & TRANSVAAL PRINTERS LTD. CAPE TOWN < CO •• 2 to 2 CO * z ^ saiavdan libraries Smithsonian institution NouniusNi nvinoshiiws S3i*m x m co — m CO o 2 INSTITUTION NOUfUllSNI NVINOSHUINS 33IHVHSS1 z o LIBRARIES' SMITHSONIAN INSTITU S3 1 H¥H a n £0 z av O’ — ^ 0 L1BRARI ES SMITHSONIAN INSTITUTION NOIlfUllSNI NVIN0SH1IWS S3 1 a Vi § s S3iavaan libraries” ^Vosv^ SMITHSONIAN INSTITU' SMITHSONIAN INSTITUTION NOUfUllSNI NVINQSHill^S S3IUV5 INSTITUTION NOUfUllSNI MVINQSHIIIAJS S3IUVH8I1 LIBRARIES SMITHSONIAN avaa li'brar ies Smithsonian institution NouniusNi nvinoshiua/s^ss i bv: 2 CO institution o 2 2 o b~ co 2 avaan libraries Smithsonian institution Nou.niu.SNi nvinoshiiws saiavi z < 2 O CO X m INSTITUTION co M©finiIISNi_NVINOSHim$ S3 S avae II LS BRAR I ES SMITHSONIAN INSTITU' Vw — (UJ 2 ft CO » . a > "W* 5 CO > z w ARIES SMITHSONIAN INSTITUTION NOliniUSNI NVINOSH1IIAIS SSUJVHSn LIBRARIES — tn. <=*■ > ““ UJ i\ — q; < p G£ £0 If} UJ cr wg —1 < p>>/ Of p/ CD W O Z HUSNU NVIN0SH1I1NS S3 S UVU 8 11 LIB RAR I ES SMITHSONIAN INSTITUTION ARIES SMITHSONIAN INSTITUTION NOIlfUIiSNI NV1NOSH1IINS S3 m avaan UIJLSNI NVINOSHISINS S3StJVHai1 LIBRARIES SMITHSONIAN INSTITUTION NOIlfUIiSNI z w o 2 ■HI C o 2 {ARIES SMITHSONIAN INSTITUTION NOIXfllllSNI NVINOSHXIWS S3 I ava a 11 LI B RAR I ES S RAR I ES SMITHSONIAN INSTITUTION NOlilUllSNI NVINOSH1IWS S3 I SVU 9 IT LIBRARIES S ;VASO*v7 00 yj £t\ c=© %4\ oe ; ''SB H 1 < VS 0£ ^ w m a } yj 5\ «— LTV £K >-£{i h | < 4/ £Q iniusN LIBRARIES SMITHSONIAN 2 W INSTITUTION NOlifUllSNI N r- RAR I ES SMITHSONIAN INSTITUTION NOUfUIlSNS NVINOSH1SSAIS S3f aVHS LIBRARIES o m x h= 2 OSHXIWS S3 ! B VB 8 H LIBRARIES SMITHSONIAN INSTITUTION NOIxnxiXSNi_h m 5 = CO ^ 2 S! ^ w w