Palaeont. afr., 23,21-33 (1980) ON THE AGE OF BORDER CAVE HOMINIDS 1-5 by Peter B. Beaumont McGregor Museum, Kimberley 8300. ABSTRACT Cited evidence suggests that Border Cave hominids 1-3 and 5 are associated with MSA phases 1-2 and date back to c. 90 000-110 000 yr B.P. These remains are fully modern in terms of morphology, and incipient Khoisan features are seemingly present in the case ofBCl. This evidence is taken to mean that truly ancestral forms of our own species probably range back into the late Middle Pleistocene of sub-Saharan Africa. Such a scenario implies that pre­ vious phologenetic reconstructions have been based on only the final quarter or so of modern man 's evolutionary history . CONTENTS Page INTRODUCTION. .......... ... ......... ........................... ... ..... ... .. ... .. . ... ............. .. .... .... ... ..... .............. .. .. .... ... .. .. ... .. 21 Site and setting............................................................................................................................................. 21 Excavation history. .... ........... ...... ....................... .................................... ........ .. ... .... ..... ... .... ..... .. .... .... .. .. .. ..... 21 LITHOSTRATIGRAPHY ............................................................................................................................. 24 CULTURAL SUCCESSION. ............. ..... . .. .. .. . ...... .. .... ... .. .... .... .. .. .. ... ...... .. ....... ... . .. ... .. .. .. . .. ................. ... . .. ... .. 24 CHRONOSTRATIGRAPHY ........ .. ........... .......... .. .................................. ..................... ... .. .. .. .. ....... ..... .. ........ 27 ADDITIONALANALYSES ........................................................................................................................... 29 Nitrogen analysis. ..... .... ....... ... ............ ....... .. .. ..... ... ....... ... .. ................. ... .................... ..... .. .. .. .... ... .. .. ........... .. 29 Amino acid analysis....................................................................................................................................... 30 Bone fragmentation ... ... .... .. .... .. ..... ........ ....... .... ..... ...... .. ... .... . ..... .. ... ... .. .... ... .. ... .. ... .... ..... ...... .. ......... ... .. . .... .. . 30 HOMINID REMAINS .. .. .. . ... ......... ... ...... ..... ................. .. ... .. .. ...... ... .. ... .... . .. ... .. .. .. .. .... ...................... .... ... ... .... 31 Iron Age ............... .. .... .. .... .. .... ........... ... ... ........ ......... .. .. ... .. ... .... .... .. ........ ......... ..... ........ .... ... ... . .... ... ..... . ...... . 31 Middle Stone Age .. .. .. .. ........... ... .......... ... ......... ...... .. .. .. .. ... ... . ...... ... ................. .. .. ... .... .... .......... ............ ...... .. . 31 Unprovenanced. ... .. . ..... ......... .. . ..... ........ ..... .............. ... .. ... .. .. ...... .. .. .. .... ... ....... .. .. ... .. .. ..... ................... .. ..... ... . 32 CONCLUSIONS. ...... .... .. .... ........................... .. . .... .. .. .. .. .. .. .. .......................................... ... ... ... ..... .. ...... .. ......... 32 ACKNOWLEDGEMENTS. .... .. . .. .......... ... .. .. .. .. .... .. .. .. .. ... . .... ..... .. .. .. ... .. .... .. .. .. ... .. .. ... .. ..... .. ......... . ... . ... .......... . 32 REFERENCES ........ .. ..... ............ .... .. ... ... ....... ............ .. ...... .... . ...... . ... ... .. ... ... .. .. ... .. .. ... ..... .............. .. ..... .. ..... ... . 33 21 INTRODUCTION Excavations at Border Cave in 1970-5 were partly directed at obtaining all pertinent data bearing on the age of the human bones found then (BC 4 and 5) and previously (BC 1-3). The pur­ pose of this paper is to detail the various data sets that have been accumulated to date in connection with this specific aspect of the investigation. maximum width at mouth and depth from drip­ line of c. 50 and 30 m respectively (fig. 2). Roof height is very variable with a greatest established reading from bedrock of about 7,5 m (fig. 3). De­ posit surface slopes down fairly regularly towards the talus edge at a mean angle of 13-15° (fig. 2). Site and setting Border Cave is situated in northern KwaZulu about 365 m from the Swaziland border at 27°1'19"S, 31°39'24"E (Cooke et al., 1945) . It cuts back into a cliff-face high up on the steep western scarp flank of the Lebombo Mountains over 400 m above the Swaziland lowveld and some 2 km north of the Ngwavuma River gorge. By direct line the site is 5 km east of the small agricul­ tural settlement of Nsoko, 12 km north of the dis­ trict centre of Ingwavuma and 82 km west of the Indian Ocean (fig. 1). Ongoing cave development is due to the prefer­ ential weathering of an agglomeratic zone within the local Stormberg System rhyolites (Cooke et al., 1945). Plan-form is roughly semicircular with a Excavation history Border Cave has been excavated on three occa­ sions by persons variously associated with the Uni­ versity of the Witwatersrand. Pre-1970 information is based on details recorded in the Archaeological Survey file B20/l/2, unless otherwise stated. The site was first investigated in July 1934 by R .A. Dart assisted by A. Galloway, J.H. Gear and G.F. Berry. A fortnight was spent lowering a strip of ten yard squares down to bedrock at a maxi­ mum depth of 168 em (Exc. 1 of fig. 2). This trench yielded MSA throughout except for super­ ficial Iron Age (lA) but no account of that material was ever published (Cooke et al., 1945). During 1940, W.E. Horton of Nsoko com­ menced digging in the cave with the object of ex­ tracting the fine ash-rich fraction of the sediments for sale as agricultural fertilizer (Horton's Pit in 22 NELSPRUIT • / KOMATIPOOR~ j TRANSVAAL ! I SWAZILAND NGWENYAe MINE /.// /. ( ( i VRYHEID • MBABANE • MANZINI • SITEKI e 0 CD ~ 0 CD w ..J ( \ "-- i ~- \ ~ \ .._···'-· ··-·-- ···-~-~:~.!... .. .J NATAL Figure 1. Locali ty Map. MOCfAMBIQUE SORDWANA BAY :.:· · INDIAN OCEAN ~CAVE WALL • HOMINID GRID lOCALITIES II CAVE WALL & BEDROCK 0 OEPOSir 23 ~ M M M M IM Figure 2. Border Cave: Plan. -==--==::JM Figure 3. Border Cave: Section along strip T. 24 fig. 2). Fossil human remains found in the course of these operations, limb-bones and a nearly com­ plete frontal (BC 1), eventually found their way to Dart in January 1941. T.R. Jones subsequently visited the site at Dart's request and found a parie­ tal fragment that articulated with the skull portion in the rubble dump left by Horton. This informa­ tion clearly indicated the possibility of finding fur­ ther human remains through an orderly explora­ tion of the cave deposits. Dart therefore proposed that an official University party investigate the site in July 1941. This was agreed to by the then Prin­ cipal, H. Raikes, and the support of the former Archaeological Survey was obtained through its Director, C. van Riet ~owe. This second expedition comprised H.B.S. Cooke, B.D. Malan, L.H. Wells, E.W. Williams and senior students in the Departments of Anat­ omy and Geology. Work in the cave occupied 18 days in July 1941 and a further 17 days in July 1942. The heap of debris constituting Horton's Pit was removed and sieved (fig. 2). This yielded a num­ ber of cranial vault pieces that joined with those on hand, a nearly complete lower jaw (BC2) and a right zygomatic. Also found were various skull, trunk and limb portions belonging to at least two individuals which were dubiously regarded on the basis of physical condition as being more recent in age. Excavation commenced in the purported initial hominid find area on the north-east side of Hor­ ton's Pit, and this was subsequently extended via trenches to the south and west (Exc. 2 of fig. 2). During these operations in 1941, an intact infant burial (BC3) associated with a perforated Conus shell was encountered at a depth of 114-137 em in squares F12 and F12 (Cooke et al., 1945). The cul­ tural sequence within these deposits from bedrock upwards was established as being: an "extremely simple" MSA industry (Cooke et al. , 1945) ; a "normal" Pietersburg (Cooke et al., 1945), subse­ quently termed the "Upper" Pietersburg (Malan 1950); a "refined" Pietersburg (Cooke et al., 1945) , later named the "Epi-Pietersburg" (Malan 1949); "indeterminate" material, i.e. "Early LSA" , re­ stricted to th@ south side of Horton's Pit/s trip Q; and "Bantu" /IA debris represented by a few sherds (Cooke et al., 1945). Almost thirty years later I decided to re-investi­ gate the site as part of the Swaziland Project under the general direction of R.A. Dart. Excavations were carried out during seven weeks in December 1970 and January 1971 and for a further month on various occasions until mid -1 9 7 5. The area Q-T 16-24, except for the test-pit enclave S-T 16-17 , was entirely removed in strata, with a further 7,5 em spit subdivision in the case of the relatively thick and mainly unstratified superficial layer (Exc. 3A Rear of fig. 2). T 4-15 and S 4-5 were lowered in strata to the MSA surface to determine the precise extent of post-1942 disturbance in areas immediately north and west of lines T /U and 15/16 (Exc. 3A Front of fig. 2). OZ-D 7-9 was probed to check on the stratigraphy in that area as outlined by Cooke et al. (1947). Strips C-D were taken down in 7,5 em spits to bedrock, and OZ-B was then sunk in strata to the base of the "Epi­ Pietersburg" (Exc. 3B of fig. 2). Human remains found at this time comprise a skeleton minus ca­ lotte (BC4) from the southern extremity of the cave, and the major portion of a lower jaw (BC5) from a depth of about 180 em in the north-west corner of square T20 (fig. 2). LITHOSTRATIGRAPHY Figures 4-6 detail the cave-fill sequence along the outside face of Exc. 2 and Exc. 3A Rear. Sec­ tions pertaining to the former dig are based on un­ published profiles by Cooke, which were slightly modified in terms of a further examination by me in 1973. Ten members with an accumulative thick­ ness of close on 4 m have been identified in those portions of the cave. These were sometimes found to be divisible on the basis of often extensive marker lenses of ash and/ or vegetation (table 1). Full member designations are given in Table 1 col­ umn 1, but only the acronyms listed in colums 2 and 3 are referred to henceforth. Correlations are by way of members, e.g. 10RS in Exc. 3A Front equates with 2BS rather than with 2BS.UP. A question mark indicates that the unit cannot be positively linked to the type sequence in Exc. 3A Rear. 1BS.UP "IA'', and 1BS.UP "Sterile" were only separable in terms of cultural content and are therefore not formal lithological entities. Disposition of the deposits as a whole was pri­ marily determined by the fact that bedrock slopes down towards the cave mouth at an angle of 13-15° (Beaumont 1973a). This has resulted in a tendency for successive members to be banked upon each other in rearward order of decreasing age (fig. 5, 6). Post-1969 observations show that the cave roof is entirely impervious to seepage, and the slope of the bedrock precludes rearward pen­ etration of capillary moisture from the drip-line beyond BACO.D in Strip 7 (fig. 3). The cave is extremely dry as a consequence, and it is to this that we owe the unique preservation of diverse or­ ganic categories, as exemplified by the occurrence of plant material in levels dating back to beyond the limits of radiocarbon (Beaumont 1973a, 1978). CULTURAL SUCCESSION Precisely 64 962 lithic artefacts were classified in terms of raw material and typology, and all intact unmodified flakes were also subjected to a detailed morphometric study. The information was then variously manipulated to isolate serial variations with time in or between major categories (Beau­ mont 1978) Results reveal that contiguous stone age strata were linked to a remarkable extent by overlapping formal tool spans and confluent morphometric shifts. For instance, the 3BS and 2WA aggregates differ markedly in some respects, yet each is 1f 1G ~ BLACK ASH ~ WHITE ASH ~ PINK ASH 0 BROWN SAND ~ GREYISH• BROWN SAND E] ROOF SPALLS E3 INTERFACES bl BEDROCK 15 14 AFTER H. B. S. COOKE 1H 1K 1l 1M ---M Figure 4. Border Cave: Section mainly following east wall of strip 16. 13 12 11 10 Figure 5. Border Cave: Section along north wall of strip E. IN 10 AFTER H. B. S. COOKE 18S 1WA 2 8S 2WA 3 8S 3WA lGBS 185 HtBS lWA? IR:GBS 1GBS 4WA BACO 25 26 24 23 22 20 19 18 185 lWA 2BS 2WA 3 BS 3WA lGBS 4WA BACO Figure 6. Border Cave: Section along south walls of strips S and T. clearly related to the other by the persistence of Backed pieces 2 in the latter and by the advent of Modified butts in the former (fig. 7) . Similarly, Ir­ regular 2 flake subclass mean length changes pat­ ently show a patterning in which the values per­ taining to any given level are related to those of its neighbours (fig. 8). The essential inference that can be drawn from these findings is that the entire lithic sequence at this site may be best regarded as representing major portions of an entirely un­ broken cultural trajectory in which no portion can be construed as being "intrusive" ( cf. Wymer 1979) . Many of the observed frequency shifts, such as the apparent partial replacement of Trimmed points 1 by Backed pieces 2 between lRGBS.B and 3BS (fig. 7) or the decrease in artefact length between BACO.D and 3WA (fig. 8), are reflec­ tions of very widespread trends (e.g. Mason 1957, Sampson 1974). It would seem rather unlikely on the basis of minimum hypothesis that these changes were not synchronous on a subcontinental level, a conclusion that is at least partly confirmed by the compatible dates for comparable industries at Border Cave and Klasies River (Beaumont 1978). It is possible to distinguish the following six site specific cultural entities at Border Cave (fig. 7) on the basis of typological criteria that are widely ap­ plicable (Beaumont 1978, 1979): Late Iron Age. A low density of lA objects in the upper reaches of lBS.UP suggests sporadic visits only to the site. Pottery consists of unincised black matt and burnished sherds that do not differ per­ ceptibly from modern Swazi wares. Early LSA. Phase I. Artefacts from lBS.LR and 1 W AI 1 BES fall within a revised definition of the "Early LSA" (Beaumont and Vogel 1972, Beau­ mont 1978) , characterized by the first appearance of ground bone points and ostrich eggshell beads and by the absence of Trimmed points, Backed pieces and Modified butt classes . MSA.Phase 3b . Aggregates from 2BS.UP (base) to 2BS.LR.B equate with the late " Post-Bowie­ son's Poort" (Sampson 1974), typified by the ab­ sence of Backed piece 2 segments and Modified butts and by low values for Trimmed points. MSA. Phase 3a. Implements from 2BS.LR.C and 2WA compare with the early "Post-Howieson's Poort'' (Sampson 1974), typified by the absence of Backed piece 2 segments, the presence of Modified butts and by low values for Trimmed points. 27 TABLE 1 Lithostratigraphic Successions at Border Cave Units in Units in Member Member Exc. 3A Exc. 3A Units in Units in name acronym Rear Front Exc. 3B Exc. 2 First Brown Sand lBS lBS.UP "lA" lBS.UP lBS "Bantu" lBS.UP "Sterile" lBS.LR lBS.LR First White Ash lWA lWA.UP lBES.UP - lWA.LR lBE.S.LR Second Brown Sand 2BS 2BS.UP lORS lRBS? Red earth layer 2BS.LR.A 2BS.LR.B 2BS.LR.C I I I I I I I I I Second White Ash 2WA 2WA - Third Brown Sand 3BS 3BS.UP lRBS? Roof collapse zone 3BS.LR.A 3BS.LR.B Third White Ash 3WA 3WA 3WA? First Rubbly lRGBS - lRGBS.A Ash band Grey-Brown Sand lRGBS.B above BC3 First Grey-Brown lGBS lGBS.UP lGBS Dark Earth Sand lGBS.LR Chocolate band and BC 1? Fourth White Ash 4WA 4WA 4WA Basal Complex BACO BACO.A BACO.A BACO.B I I I I I I I I I I BACO.C BACO.D Abbreviations: B = brown; G = grey; W = white; BE = beige; OR= orange; A =: ash; S = sand; R = rubbly; BA = basal; CO = complex; ? = uncertain correlation; mrn = excavation base/bedrock. MSA. Phase 2. Assemblages from 3BS and 3WA of Ex c. 3A Rear and 1 RG BS of Ex c. 3B refer to the "Epi-Pietersburg" variant of the old "Second In­ termediate" (Malan 1949), distinguished by the obtrusive presence of Backed pieces 2 (mainly seg­ ments) and by low values for Trimmed points. MSA. Phase 1. Samples from lGBS.UP to BA­ CO.D were originally described as grading be­ tween "simple" and "normal" Pietersburg (Cooke et al., 1945) and equate broadly with those from Beds 5-8 at the Cave of Hearths (Sampson 1974). They are epitomised by the absence of Backed pieces 2- and by very high values for Trimmed points 2. A progressive increase with time in the relative proportion of formal tools in those levels apparently constitutes a major basis on which "middle" and "later" stages of the Pietersburg, etc. are often identified, but it is evident from the data (fig. 7) that this separation cannot be sus­ tained in terms of formal tool class typology alone (Beaumont 1978) . CHRONOSTRATIGRAPHY Dating of the lBS.UP- 2BS.UP sequence is based on a suite of 37 broadly conformable C-14 measurements that were mainly processed by J.C. Vogel at the NPRL of the CSIR (table 2). The age BP-C of earlier levels was deduced by correlating data sets reflecting sequential temperature shifts with a temporal framework derived from calibrated global ice-volume fluctuations (Beaumont 1978, Beaumont et al., 1978). The total evidence derived from those sources is taken ·to indicate (Beaumont 1979): lBS.UP (Iron Age) = 100-600 B.P.; lBS.UP (sterile) = 600-30 000 B .P.; lBS.LR-lWA (Early LSA 1) = >30 000-38 000 B.P.; 2BS.UP-2BS.LR.B. (MSA 3b) = c. 55 000-70 000 B.P.; 2BS.LR.C.- 2WA (MSA 3a) = c. 75 000-85 000 B.P.; 3BS.UP-1RGBS.B. (MSA 2) = c. 85 000-95 000 B.P.; 1GBS.UP-4WA (MSA 1 "late") = c. 100 000-110 000 B.P.; BACO. A-D (MSA 1 "mid­ dle") = 110 000-130 000 B.P. Ofparticular note is the close correspondence that these estimates show with the ages deduced for levels with comparable aggregates at Klasies River (Butzer 1978), thereby suggesting that similar industries were probably more or less coeval over wide areas of southern Africa. Comments concerning the antiquity of the 1 WA have been largely founded on the three readings from square T21 (table 3). However, it may be that lBS.LR dates to 38 600 B.P. (Pta-704) and that lWA has an age of up to 45 000 B.P. (Pta- 1190) (Beaumont 1978). 28 SCRAPER TRIMMED POINT BACKED PIECE 1 BACKED PIECE 2 MODIFIED BUTT SCALED PIECE IBS .LR+IWA+IBIES-~-------,r--------+---------r--------1---------r---------~------~---------+------------------w 2BS.UP+ LR.A+B 2BS.LR.C + 2WA 3BS+3WA 1RGBS.A+B 1GBS.UP+LR BACO.A + B BACO. C+ D 50 ALL SEGMENT • 50 %OF TOTAL 0 TRAPEZE 0 50 50 Figure 7. Formal tool class proportion changes with time. Border Cave excavations 3A and 3B. 1 BS.LR. + 1WA +1 BES 2BS.UP+LR.A+B 2BS. LR . C 2WA 3BS+3WA 1RGBS.A 1RGBS. B 1GBS. UP 1GBS. LR BACO A BACO. B BACO. C BACO. D ·. . ·· ... - - A RHYOLITE 1 +QUARTZITE 1 .... e CHALCEDONY ·. ·. ·. ' ' ' 2R ' ~ ... ,,'~- -- + ' ' ' ' ' ' '~ ' ' ~ ' ' ---,?;:p + c:::P, ·. ·. ··. 30 12 LENGTH IN MM ' 'c:P ' ' ~ ' ' Figure 8 . Twelve subclass length changes with time. Border Cave excavations 3A and 3B. 50 55 29 TABLE 2 Radiocarbon Dates from Border Cave Lab. no. Material Square Fraction Depth (em) Pta-1728 Maize cob. Acid T21 8-15 Pta-870 Vegetation. Acid T21 8-15 Pta-703 Vegetation T22 8-15 LJ-2889 Vegetation . T22 30-38 LJ-2890 Vegetation T22 38-46 Pta-715 Vegetation Tl9 30-38 Pta-777 Bone collagen (BC4) 2S 13 Pta-1318 Bone collagen (BC4) 2S 13 Pta-506 Vegetation. Acid Sl9 38-46 Pta-721 Vegetation Sl9 46-53 LJ-2891 Vegetation T22 46-53 LJ-? Charcoal T22 61-69 LJ-2892 Charcoal T22 69-76 Pta-704 Charcoal S21 69-IWA Pta-1964 Grass. Unidentified Ql8 (Mini counter) Pta-1995 Leaves. Linociera S21 (Mini counter) UCLA-1754 C Bone collagen Ql9 UCLA-1754D Bone collagen Ql9, Rl9 Pta-446 Charcoal. Acid T21 Pta-422 Charcoal. Residue T21 Pta-423 Charcoal T21 Pta-424 Charcoal T21 Pta-1190 Twigs Rl9 Pta-1274 Charcoal Q20 Pta-1275 Charcoal Q21 Pta-877 Charcoal Q22 Lj-? Charcoal Q23 Pta-1244 Charcoal R21 Pta-872 Charcoal T23 UCLA-1754 E Bone collagen R20, S22 Pta-489 Charcoal A7 Pta-421 Charcoal. Acid oz 7 Pta-447 Charcoal. Acid oz 7 Pta-459 Charcoal. Residue oz 7 Pta-463 Charcoal. Extract PZ 7 Pta-719 Charcoal. Acid oz 7, oz 9 Pta-488 Charcoal. Acid A8 Abbreviations: Acid = acid pretreatment only. Butzer has presented sedimentological data which he interprets to mean that the basal strata range back to the onset of 6 180 stage 6 (Butzer et al., 1978). This inference is apparently incompat­ ible with other evidence which would rather sug­ gest that those levels have a maximum age of 130 000 B.P. (Beaumont 1978, Beaumont et al. , 1978). Additional information and particularly ad­ equate macrofauna! samples from BACO would seem to be necessary in order to resolve this par­ ticular problem (Klein 1977). Stratum Industry Age B.P. IBS.UP "IA" lA 90 ± 105 IBS.UP "lA" lA 170 ± 45 IBS.UP "IA" IA 500 ± 45 IBS.UP "lA" IA 500 ± 70 IBS.UP "IA" (base) IA 590 ± 70 IBS.UP "lA" (base) IA 440 ±55 I BS. UP "IA" IA 340 ± 45 IBS.UP "IA" IA 480 ± 45 IBS.UP "Sterile" (top) - 2010 ± 50 IBS.UP "Sterile" (base) - 13,300 ± 150 IBS.UP "Sterile" (top) - 650 ± 70 I BS. UP "Sterile" (base) 28,500 ± 1800 IBS.LR (top) Early LSA I 33,000 ± 2000 IBS.LR (base) Early LSA I 38,600 ± 1500 IWA Early LSA I >25,500 IWA Early LSA I >25,500 IWA (top) Early LSA I 33,000 ± 2000 IWA (base) Early LSA I 34,800 ± 2500 IWA (top) Early LSA I 37,500 ± 1200 IWA (top) Early LSA I 36,800 ± I 000 IWA (mid) Early LSA I 36,100 ± 900 IWA (base) Early LSA I 35,700 ± 1100 IWA (base) Early LSA I 45,000 + 2750 - 2200 2BS.UP MSA3b 47,200 ± 4200 - 2750 2BS.UP MSA3b >49,100 2BS.UP MSA3b 45,400 + 3000 - 2000 2BS.UP MSA3b >41,000 2BS.UP MSA3b >48,800 2BS.UP MSA3b >42,300 2WA MSA3a >45,000 IRGBS.A MSA2 >48,700 IRGBS.B MSA2 36,000 ± I 000 IRGBS.B MSA2 >47,500 IRGBS.B MSA2 >48,350 IRGBS.B MSA2 >42.600 IRGBS.B MSA2 42,000 + 3000 - 2000 IRGBS.B MSA2 >48,500 ADDITIONAL ANALYSES Other data sources that require consideration to provide a proper perspective for the results pre­ sented in Tables 3 and 4 are nitrogen analysis, amino acid analysis and bone fragmentation. Nitrogen analysis Nitrogen analysis depends on the fact that the proportion of nitrogen in any given bone tends to decrease with time as protein is removed by way of various agencies (e.g. Oakley 1963). However, that 30 effect only holds true and comparisons between readings are only justified if various potential sources of distortion can be shown to have been in­ operative. First, anomalous values will occur if the sample undergoes heating by overlying and/or contiguous hearths (de Graaff 1961), resulting in the perferen­ tial release of low molecular weight hydrocarbons (Silverman 1964). Amino acid printout configura­ tions strongly suggest that temperature extremes have not affected BC 1-3 Q. Bada pers. comm.). The other human remains all came from shallow graves or depressions, and sediment infill would have tended to protect them from the effects of contiguous fires. These two lines of evidence are taken to indicate that the readings on BC 1-5 are probably free of heat induced aberrations. Secondly, anomalous values will occur unless vertical and lateral variations in the physicochemi­ cal environment being sampled are minimal (Protsch 1973). Comparison of Exc. 3A Rear and Exc. 3B macrofauna! readings reveal that there are indeed marked differences between areas deep within the cave and those near the drip-line. A markedly higher nitrogen attrition rate was found to be applicable to the latter excavation, presum­ ably as a result of backward moisture movement from the drip-line (Beaumont 1978). It follows that samples drawn from that region must be re­ garded as minimum values if compared with those from the cave interior. Thirdly, anomalous values may occur in view of variations between some species in initial bone ni­ trogen content (Protsch 1973). This possibility has been offset here by confining comparison to the hominid remains (table 3). Amino acid analysis Much work has recently been carried out on amino acid enantiomer shifts with time as a means of dating bone in relative or absolute terms (e.g. Bada 1972, Bada and Protsch 1973). Since poss­ ible limitations in the availability of the hydroxide ion in the cave interior could have led to a reduc­ tion in the racemization rate there relative to the chemical situation near the drip-line (Bada and Schroeder 19 7 5), sampling was confined to un­ charred bone fragments from the former area. A major limitation of the method is that racemiza­ tion reaction is a chemical process and thus inher­ ently a function of both time and temperature. Sampling was confined to the microfauna in an at­ tempt to obviate the latter variable during the con­ struction of an aspartic acid temporal framework for the site. The assumption underlying this ap­ proach was that the original encapsulating owl pellets would have provided optimal protection of the fragments from hearth-derived heating. Resul­ tant readings that conformed with the sediment se­ quence were regarded as undistorted by tempera­ ture damage, and only these have been retained for presentation in Table 4. Bone fragmentation A coarse analysis was undertaken of the extent of bone damage in the various levels represented in Exc. 3A Rear (Beaumont 1978). Values obtained in terms of fragments per kg were found to vary widely with extremes of-780 and -1410 but with a site mean of -1 160. Equivalent readings from other localities indicate that the Border Cave av.erage may be regarded as indicative of extreme fragmentation. The frequent recovery during exca- TABLE3 Nitrogen Readings on Hominid Bones from Border Cave Grid N% Location Stratum Culture Lab. no. Material (washed) 2S 13 IBS.UP "lA" lA BM-SA 166 BC 4 0,93 Exc. 3A 3WA MSA2 BM-SA 167 BC 5 0,48 Exc. 2 IGBS.UP MSA I (late) BM-SA 151 BC 3 0,44 Exc. 2 IGBS.UP MSA I (late) UCLA-1754A BC 3 0,44 Horton's Pit ? ? BM-SA 164 BC I 0,28 Horton's Pit ? ? UCLA-1754B BC 1 0,41 Horton's Pit ? ' ? BM-SA 165 BC 2 0,29 TABLE4 Aspartic Acid Readings on Bone from Border Cave Grid DIL Location Stratum Culture Lab. no. Material value Exc. 3A Rear 1BS.UP "lA" lA (lower) LJ-AA? Microfauna 0,26 Exc. 3A Rear IBS.UP "lA" lA (base) LJ-AA? Microfauna 0,29 Exc. 3A Rear IBS.LR LSA I LJ-AA? Microfauna 0,55 Exc. 3A Rear 3BS MSA2 LJ-AA? Microfauna 0,61 Exc. 2 IGBS.UP MSA I (late) LJ-AA? BC 3 0,72 Horton's Pit ? ? LJ-AA? BC I 0,77 Horton's Pit ? ? LJ-AA? BC 2 0,72 vation and analysis of contiguous JOining bone fragments suggests that a substantial proportion of that breakage postdates deposition. This fraction is considered to have been largely induced by human trampling in a setting where low sediment accu­ mulation rates were the norm. HOMINID REMAINS Border Cave human remains have been num­ bered according to the scheme adopted by de Vil­ liers (1975). Stratigraphic data for BC l-3 are based largely on information drawn from Archaeo­ logical Survey file B20/l/2. Iron Age Anatomical description of lA hominid BC 4 was undertaken by de Villiers and is recorded in Beau­ mont ( 1978). The skeleton, minus calotte, belongs to a slightly built individual of uncertain sex and with an age at death of c. 38-45 yrs. Measure­ ments, indices, and non-metrical characters of the mandible fall well within the corresponding ranges of the male South African Negro. The vertebrae and limbs indicate nutritional deficiencies and ex­ treme osteoarthritis, and the teeth show severe at­ trition with dentine exposure. Artificial damage to the shaft of the right humerus may represent a witchdoctor's attempt at curing a "frozen" shoul­ der. Low 13C bone values point to an exclusive C4 intake which probably reflects a diet largely based on cultigens such as millet and/or sorghum (Vogel 1978). The remains were found in the southernmost portion of the cave and close to the back wall while clearing surface debris prior to setting up a sleep­ ing area at the onset of fieldwork in December 1970 (fig. 2). It came from a very shallow grave that had been cut through the locally very rubbly lBS from at or just below the present surface to bedrock at a maximum depth of -15 em. The body was largely surrounded and partly covered by casually stacked rock slabs. Apparently the pre­ vious removal of some of these had partly exposed the skull, thereby leading to its disappearance. BC 4 was found lying contracted on its left side with back to the west, head to the north and arms crossed over the chest, the right under the left. The right hand was over the left shoulder and on it rested the mandible. The ribs had collapsed in­ wards and the pelvis was cracked, possibly be­ cause of the overlying rocks. Almost all of the post­ cranial portions were present despite the bone being generally friable. Shells of carrion 'beetle pu­ pae surrounded the body, and a small piece of leather was found immediately east of the pelvis. A very rusted iron bracelet 8 em broad encircled the left wrist, from which a date within the past two millennia may be inferred (Dart and Beau­ mont 1969). Two corrected and calibrated radio­ carbon determinations of the bones indicate age of A.D. 1460 Q .C. Vogel, pers. comm.). This is quite close to the lower limits of the lA occupation at this site as documented by the readings in Table 2. 31 Middle Stone Age BC 3 is an incomplete skeleton representing an infant aged between four and six months (de Vil­ liers 1973). Various metrical and morphological features align it with both Khoisan and South Af­ rican Negro (de Villiers 1973). The remains were found straddling squares Fl2 and Gl2 during sys­ tematic excavations in 1941 (fig. 2). It came from a shallow but indubitable grave, 38 em long and 30 em wide with the long axis north-east to south­ west. The shaft walls sloped inwards except in the north where they had been undercut. Base depth increased southwards with a maximum of -13 7 em in the north-west quadrant of square Gl2. The first human fragments were found at just less than 114 em, and the grave lip appears to have been marginally higher. Remains seem to have been rather scattered, and the original pos­ ture cannot be deduced. The associat~d Conus shell was perforated and presumably an ornament or amulet and indicates some form of contact with the coast, 82 km or more to the east (fig. l). Some of the bones show reddish-brown stains (de Vil­ liers 1973) that may represent an application of red ochre (Beaumont l973b) before or after disso­ lution of the flesh (Beaumont 1978). The base of an undisturbed 8 em thick layer of deep brown sand overlying black ash is recorded as lying 5 em above the apparent grave rim at a depth of 107-109 em in squares El2 and Gl2 (table l) . The brown sand and black ash mark the lower limits of the "Epi-Pietersburg" industry in Exc. 2. Malan's field diary also notes an increase in Achatina shell from square E 12 between l 07 and 114 em, which strongly suggests that the lower portion of that spit was cutting into the typically mollusc-rich lGBS.UP (Beaumont 1978). It fol­ lows from these stratigraphic details that the burial amost certainly refers to the surface reaches of my late MSA 1. Further confirmation of this in­ terpretation is provided by the nitrogen and aspar­ tic acid readings as recorded in Tables 3 and 4. BC3 has thus a 14C age in excess of 49 000 B.P. (tables 1 and 2) and an inferred dating of c. 105 000 ± 5 000 B.P. (Beaumont 1979). Anatomical description of BC5 was undertaken by de Villiers ( 1976, see also Beaumont 1978). That specimen comprises the major portion of a lower jaw, probably belonging to a male, with an estimated age at death of 25- 35 yrs. Wear surfaces of all intact teeth show severe attrition with pulp cavities being exposed in some instances. The met­ rical and morphological features as a whole appear to favour a link with the South African Negro. BC5 was recovered in my presence by Colleen Powell on lOth April 1974 (fig. 9). It came from just above the base of the intact 3WA in the north­ west corner of square 20 immediately adjacent to a marked depression to the east. This stratigraphic anomaly, perhaps a grave, extends into the face of the trench and would require the lowering of strip U for its further investigation. From these data it is evident that the mandible relates to artefacts 32 representing a middle stage of the "Epi-Pieters­ burg" MSA 2. This phase has a C-14 date of greater than 49 000 B.P. (table 2) and an inferred age of90 000 ± 5 000 B.P. (Beaumont et al., 1978). Figure 9. C. Powell holds Border Cave hominid 5 in the precise position where it was found by her on April 10, 1974. It is located just above the base of the 3WA and some 25 em below the visibly intact surface of that stratum. Photograph by courtesy of Dr. G. Baker. Unprovenanced BC l and 2 comprise an incomplete adult male cranial vault and a partial adult female mandible respectively. Multivariate statistics indicate that BC l is "not markedly dissimilar" to Tuinplaas (de Villiers 1973) and that it has affinities with the Khoisan (Rightmire 1979). Discriminant analysis shows that separation from modern indigenous populations is mainly a result of greater size with particular respect to breadth (de Villiers 1973). This difference is possibly significant in view of the established reduction in that parameter between Upper Palaeolithic and Neolithic crania from Western Europe (Frayer 1972). BC2 metrical in­ dices, on the other hand, correspond closely with the norms applicable to San females according to de Villiers ( 1973, 1976). The hominid bones as a whole contrast strongly with mean macrofauna! fragment mass values ( -1,0 g), which suggests that burial may have protected the former group from post-depositional damage (Beaumont 1978). BC l is said to have come from the north-west sector of Horton's Pit, and the preservation state alone supports a derivation from backward of strip 7 (fig. 2). Guano digger accounts differed as to its precise position within the c. 170 em depth of sedi­ ments exposed in that portion of the cave (figs. 5, 6). The 1942 observations revealed that the 4WA surface extended undisturbed over the floor of Horton's Pit, thereby providing a firm lower limit to the strata potentially linked to BC l and 2. On the basis of soil adhesions in the small interstices of the skull, Cooke et al. ( 1945) concluded that it had come from a distinctive chocolate brown layer lying immediately above the 4WA, corresponding to our IGBS.LR (table 1). That claim is supported by the nitrogen and as­ partic acid readings, which indicate that BC l and 2 are perhaps somewhat older and certainly not younger than the 3WA (tables 3, 4). Any possibil­ ity of an lA ascription is clearly precluded by the nitrogen value for BC4 which is about 2-3 times higher than the others despite the fact that it came from an area where a relatively rapid nitrogen de­ pletion rate prevails. The likelihood of a post-4 WA but pre-3W A age for BC 1 and 2 would clearly be in good accord with the F and N-verified (Protsch 1973) association of the comparable Tuinplaas skull with an "advanced" Pietersburg assemblage (van Riet Lowe 1929). The overall conclusion regarding BC 1 and 2 is, thus, that all available data exclude an lA ascrip­ tion, which would thereby imply a pre-lBS.UP "Sterile" dating of at least 30 000 B.P. (table 2). Other concordant evidence supports the original contention of Cooke et al. (1945) that the remains refer to the lGBS.LR with an inferred age of about 110 000 B.P. (Beaumont et al., 1978). CONCLUSIONS Hominid remains from the 3WA and lGBS at Border Cave (Beaumont et al., 1978) and from the MSA I and II levels at Klasies River Mouth (Singer and Smith 1969, Rightmire 1978) demon­ strate the presence in southern Africa of anatom­ ically modern Homo sapiens in Last Interglacial de­ posits dating to c. 90 000-130 000 B.P. Most of the specimens are fully modern in morphology, and archaic features are not obtrusive. It is deduced that those humans ultimately derive from an as yet undiscovered primal form which is perhaps linked to aggregates near the MSA/ESA interface at something like 160 000 ± 40 000 B.P. If this scenario is correct, then it would follow that Eurasian prehistory may in fact document only various terminal developments in the unex­ pectedly protracted physical evolution and cultural development of our own kind in Africa south of the Sahara (Beaumont et al., 1978). The further inves­ tigation of modern man's formative history in ap­ propriate areas of the subcontinent should consti­ tute one of the major objectives of archaeological endeavour during the coming decade, for it is only by placing that record in true time perspective that we can hope to comprehend the interacting envi­ ronmental and cultural forces which have led to the world as we see it today.' ACKNOWLEDGEMENTS I wish to record my indebtedness to the many individuals and institutions that contributed to the recent investigations at Border Cave, the Anglo American Corporation and the Swaziland Iron Ore Development Company for a decade of financial support prior to 1978 and the Archaeological Re­ search Unit of the University of the Witwatersrand for per­ mission to cite from file B 20/1/2. 33 REFERENCES ARCHAEOLOGICAL SURVEY file B 20/l/2: BORDER FRAYER, D.W. (1972). Changes in the shape and form of the CAVE. Now housed at the Archaeological Research Unit, skull from the Upper Palaeolithic to the Neolithic. Seminar University of the Witwatersrand. on human evolution. Ann Arbor, University of Michigan. BADA,J.L. (1972). The dating of fossil bones using the race- KLEIN, R.G. (1973). Geological antiquity of Rhodesian mization of isoleucine. Earth and Planet. Sci. Letters,15, Man. Nature Lond., 244, 311-312. 223-231. ---- (1977). The mammalian fauna from the Middle ---- and PROTSCH, R. ( 1973) · Racemization reac- and Later Stone Age (late Pleistocene) levels of Border tion of aspartic acid and its use in dating fossil bones. Proc. Cave, Natal Province, South Africa. S. Afr. archaeol. Bull., Nat. Acad. Sci., U.S.A., 70, 1331-1334(. ) A . 'd 32, 14-27. ----and SCHROEDER, R.A. 1975 . mmo aCI ra- cemization reactions and their geochemical implications. MALAN, B.D. (1949). Magosian and Howieson's Poort. S. Naturwissenschaften, 62, 71-79. Afr. archaeol. Bull., 4, 34-36. BEAUMONT, P.B. (1973a). Border Cave-a progress re- ---- (1950). The Middle Stone Age in the eastern S A,r, j S · 69 41-46 Transvaal and Swaziland. S. Afr.J. Sci., 47, 146-150. port. . 'Jr. . cz., , . ---- (1973b). The ancient pigment mines of southern MASON, R.J. (1957). The Transvaal Middle Stone Age and Africa. S. Afr.J. Sci., 69, 140-146. statistical analysis. S. Afr. archaeol. Bull., 12, 119-143. ---- (1978). Border Cave. Unpublished M .A. thesis, OAKLEY, K.P. (1963). Analytica:l methods ofdating bones. University of Cape Town. In: Brothwell, D .R. and Higgs, E.S., Eds., Science in archae- ---- ( 1979) . The stone age cultural stratigraphy of ology. London, Praeger. Border Cave. Submitted for publication. PROTSCH, R.R. (1973). The dating of Upper Pleistocene ----and VOGEL, J.C. (1972). On a new radiocarbon sub-Saharan fossil hominids and their place in human evo- chronology for Africa south of the equator. Afr. Stud., 31 , lution: with morphological and archaeological implica- 69-89 and 155-182. tions. Unpublished Ph.D. thesis, University of California ---- DE VILLIERS, H. and VOGEL, J.C. (1978). at Los Angeles. Modern man in sub-Saharan Africa prior to 49 000 B.P.: a RIGHTMIRE, G.P. (1978) . Florisbad and human population review and evaluation with particular reference to Border succession in southern Africa. Am. j. phys. Anthrop., 48, Cave. S. Afr.J. Sci., 74, 409-419. 479-486. BUTZER, K.W. (1978). Sediment stratigraphy of the Middle Stone Age sequence at Klasies River Mouth, Tsitsikama coast, South Africa. S . Afr. archaeol. Bull. , 33, 141-151. ----BEAUMONT, P.B. and VOGEL, J.C. (1978). Lithostratigraphy of Border Cave, KwaZulu, South Africa: a Middle Stone Age sequence beginning c. 195 000 B.P.]. Archaeol. Sci., 5 , 317- 341. COOKE, H.B.S., MALAN, B.D. and WELLS, L.H. (1945). Fossil man in the Lebombo Mountains, South Africa: the "Border Cave", lngwavuma district, Zululand. Man 45 (3), 6-13. DART, R.A. and BEAUMONT, P.B. ( 1969). Iron Age radio­ carbon dates from western Swaziland. S. Afr. archaeol. Bull. , 24, 71. . DE GRAAFF, G . ( 1961 ). Gross effects of a primitive hearth on bones. S. Afr. archaeol. Bull., 16, 25-26. DE VILLIERS, H. (1973). Human skeletal remains from Border Cave, lngwavuma district, KwaZulu, South Africa. Ann. Transv. Mus. , 28, 229-256. ---- (1976). A second adult human mandible from Border Cave, lngwavuma district, KwaZulu, South Africa. S. Afr.J. Sci., 72, 212-215. ---- (1979). Implications of Border Cave skeletal re­ mains for later Pleistocene human evolution. Curr. Anthrop., 20, 23-26. SAMPSON, C. G. ( 1974). The Stone Age Archaeology of Southern Africa. New York, Academic Press. SILVERMAN, S.R. (1964). Investigations of petroleum ori­ gin and evolution mechanisms by carbon isotope studies. In: Craig, HJ. , Miller, S.L. and Wasserburg, G.T. Eds., Isotopic and cosmic chemistry 92-102. Amsterda m. North Hol­ land. SINGER, R. and SMITH, P. (1969). Some human remains associated with the Middle Stone Age deposits at Klasies River, South Africa. Am.]. phys. Anthrop., 31 , 256. VAN RIET LOWE, C . (1929). Notes on some stone imple­ ments from Tuinplaats, Springbok Flats. S. Afr.]. Sci., 26, 623-630. VOGEL, J .C. (1978). Isotopic assessment of the dietary ha b­ its of ungulates. S. Afr.]. Sci., 74, 298-30 I. WYMER, J.J. (1979). Comments on " Implications of Border Cave skeletal remains for later Pleistocene human evolu­ tion" . Curr. Anthrop., 20, 32-33.