Journal of Archaeological Science (1998) 25, 377–392

Chronology of the Later Stone Age and Food Production inEast Africa

Stanley H. Ambrose

Department of Anthropology, University of Illinois, Urbana, IL 61801, U.S.A.

(Received 26 March 1997, revised manuscript accepted 12 January 1998)

Evidence from several archaeological sites in sub-Saharan Africa suggests that the transition to modern humantechnology, marked by the change from the Middle to the Later Stone Age (LSA), occurred first in East Africa.Enkapune Ya Muto rockshelter, in the central Rift Valley of Kenya, contains the oldest known archaeological horizonsspanning this transition. Radiocarbon and obsidian hydration dates from this 5·6-m deep cultural sequence show thatthe Later Stone Age began substantially earlier than 46,000 years ago. Ostrich eggshell beads were made 40,000 yearsago. Early dates for the LSA and beads may have implications for the origin and dispersal of modern human behaviourand modern humans out of Africa.This site also contains the only known occurrences dating to the Middle Holocene dry phase in highland Kenya and

Tanzania, as well as occurrences that span the transition from hunting and gathering to food production, and from theNeolithic to the Iron Age. The adoption of domestic animals by indigenous Eburran hunter–gatherers in highland EastAfrica occurred gradually between 4900 and 3300 uncorrected radiocarbon years and the Neolithic/Later Iron Agetransition occurred around 1300 . ? 1998 Academic Press Limited

Keywords: MIDDLE STONE AGE, LATER STONE AGE, NEOLITHIC, IRON AGE, EAST AFRICA,HUMAN EVOLUTION, RADIOCARBON, OBSIDIAN HYDRATION, COSMOGENIC NUCLIDES,PALEOLITHIC ORNAMENTS.

Introduction

E nkapune Ya Muto rockshelter (EYM), locatedin the central Rift Valley of Kenya, contains a5·6-m deep sequence of archaeological occur-

rences spanning substantial portions of the LaterHolocene and the middle of the Upper Pleistocene. Itwas initially excavated in 1982 to fulfil three objectives:(1) to find archaeological occurrences dating to theMiddle Holocene dry phase in highland East Africa;(2) to test a model of forest-savanna ecotone preferencefor Holocene hunter–gatherers; and (3) to documentthe transition to food production during the laterMiddle Holocene (Ambrose, 1984a). These objectiveswere all fulfilled (Ambrose, 1984b, 1986; Ambrose &DeNiro, 1989; Ambrose & Sikes, 1991; Marean, 1992).One new Middle Stone Age (MSA) and two new earlyLater Stone Age (LSA) lithic industries were revealedbeneath an erosional unconformity. Further excava-tions were undertaken in 1987 in order to obtain amore complete Late Pleistocene cultural sequence andaccurate dates for the MSA/LSA transition. Radio-carbon and obsidian hydration dates indicate that thistransition may have occurred by 50,000 . Theyounger early LSA industry is associated with evidenceof ostrich eggshell bead manufacture, dated to 39,900. This paper describes the chronology of the cultural

3770305–4403/98/040377+16 $25.00/0/as980277

sequence at EYM and its implications for the origin ofmodern human behaviour in East Africa, the relation-ship between Mid/Late Holocene climate change andthe chronology of the transition to food production,and the dating of the transition to the Later Iron Age.

Dating the transition to modern human behaviourThe transition to ‘‘modern’’ human behaviour in sub-Saharan Africa is marked by the change from the MSAto the LSA stone tool industrial complex. This isequivalent to the Middle/Upper Paleolithic (MP/UP)transition in northern Africa and western Eurasia(Klein, 1989a, 1992; 1995). Flake-based stone toolindustries made on radial and levallois cores withprepared (faceted) platforms in the MSA and MP werereplaced by blade- and small flake-based industries,usually made on cores with plain platforms in the LSAand UP. Ground bone tools and perforated ornamentsbecame common in the LSA and UP, and resourceexploitation and socio-territorial organization patternsbegin to resemble those of recent hunter–gatherers(Binford, 1989; Klein, 1989a,b, 1992, 1995; Ambrose &Lorenz, 1990). Deacon (1989, 1992, 1995; Deacon &Wurz, 1996) has proposed that the MSA HowiesonsPoort Industry in South Africa, which dates to ap-proximately 60–80,000 , reflects fully modern humanbehaviour because it contains several characteristic

Correspondence to: S. H. Ambrose. Fax: (217) 244 3490; e-mail:Ambrose@uiuc.edu

? 1998 Academic Press Limited

378 S. H. Ambrose

features of the LSA, including exchange of formalblade-based backed tools made on exotic lithic rawmaterials and intra-site spatial organization patternsaround hearths. However, Klein (1989b) and Ambrose& Lorenz (1990) argue that the Howiesons Poort doesnot reflect fully modern behaviour because its resourceexploitation patterns seem less effective than those ofthe LSA in similar environmental contexts. Unlike theLSA, the Howiesons Poort also lacks ground boneartefacts and art.The transition to modern technology is thought to

have occurred earliest in East Africa, and may havespread by stimulus diffusion and/or migration out ofAfrica via the Sinai Peninsula to western and northernEurasia by 47–40,000 (Klein, 1989a, 1992, 1995;Lahr & Foley, 1994). Proof of an East African originremains equivocal because the transition appears to beat or beyond the practical limits of radiocarbon dating,around 40–45,000 years (Aitken, 1990; Taylor,1996), and other dating methods provide less reliableestimates of age.Radiocarbon dates in sedimentary sequences

become asymptotic with depth around 40–50,000 ,usually due to contamination with younger carbonwhile buried, or during excavation, storage and samplepreparation (Haas, Holliday & Stuckenrath, 1986;Chappel, Head and Magee, 1996). Other methods ofdating this crucial time period, including uraniumdisequilibrium series (U, Th, Pa) of carbonates andfossils (Schwarcz, 1992), amino-acid racemization(AAR) of avian eggshell and bone (King & Bada, 1979;Brooks et al., 1990; Johnson & Miller, 1997), thermo-luminescence (TL) of burned flints (Mercier, Valladas& Valladas, 1995), electron spin resonance (ESR) oftooth enamel (Grun, 1993), optically stimulated lumi-nescence (OSL) of sedimentary quartz (Aitken, 1994)and obsidian hydration (Michels, Tsong & Nelson,1983) can be used to date sites older than 40,000 years,but all require site-specific calibrations and assump-tions about geochemistry, background radiation,linearity of signal uptake, temperature history, soilmoisture and/or diagenetic processes. Dates older than40,000 produced by these techniques are thus intrin-sically questionable because they cannot be comparedto radiocarbon dates. Fission track (Wagner, 1996)and single crystal laser fusion (SCLF) 40Ar/39Ar canprovide accurate dates on stratified volcanic tephrafrom archaeological sites as young as 2000 (Huet al., 1994; Chen et al., 1996; Renne et al., 1997), buthave rarely been reported for MSA and LSA sites(Manega, 1993).Where radiometric calibration of these methods can

be applied to horizons younger than 40,000 , greaterconfidence can be placed in such dates on olderhorizons in the same sites. However, radiocarbon datesmust also be calibrated. Calibration of the LatePleistocene radiocarbon scale with high precisionU-series dating (Bard et al., 1990) demonstrates radio-carbon dates are approximately 3000 years too young

by 30,000 . If this rate of deviation was constant,dates on 42,000-year-old materials would be approxi-mately 5000 years too young (Bischoff et al., 1994).However, deviation rates were not constant. Theyreflect variations in the earth’s magnetic field strength(dipole moment), especially at c. 40–45,000 , whenthere was a substantial reduction in dipole momentduring the Laschamp geomagnetic reversal event(Meynadier et al., 1979). Cosmogenic nuclide produc-tion increased for several millennia, making radio-carbon dates on materials formed after 45,000 approximately 3500–5000 years younger than theirtrue age (Raisbeck et al., 1987; Mazaud et al., 1991;Sternberg & Damon, 1992; Laj, Mazaud & Duplessy,1996). Atmospheric radiocarbon production appar-ently doubled during this event. Gaps and inversions inradiocarbon dates should occur around 43–38,000 (Sternberg & Damon, 1992), and have been found indated MP/UP sequences in the Levant and westernEurope (Ambrose, 1997).Obsidian hydration and AAR dates have been

used to date Late Pleistocene sites in Africa. Wheredates are calculated using modern temperaturesthey should be considered minimum age estimatesbecause temperatures were on average 4–6)C lowerduring the Pleistocene (Schroeder & Bada, 1973;Bonnefille, Roeland & Guiot, 1990), which wouldhave decreased reaction rates (Miller et al., 1992;Johnson & Miller, 1997; Jones, Sheppard & Sutton,1997; Miller, Magee & Jull, 1997). Obsidian hydrationand AAR rates can be calibrated where associated withradiometric dates and this permits reconstructionof paleotemperatures (Miller, Magee & Jull, 1997).Radiocarbon production rates were higher and morevariable than at present for most of the LatePleistocene (Laj, Mazaud & Duplessy, 1996), so radio-carbon dates will be underestimated. Paleotempera-tures will be overestimated and reaction rates andracemization and hydration dates underestimated.Therefore an additional correction of radiocarbondates for variation in dipole moment should be incor-porated into hydration and racemization rates andpaleotemperature estimates.In Europe and western Asia the chronology of the

MP/UP transition is becoming better known throughradiocarbon, TL and U-series dating. It apparentlyoccurred around 47–43,000 in the Sinai Peninsulaand Levant (Marks, 1983; Phillips, 1994; Schwarcz,1994; Bar-Yosef et al., 1996; Rink et al., 1996) and43–38,000 in Europe and Siberia (Koslowski, 1988;Goebel, Derevianko & Petrin, 1993; Bischoff et al.,1994; Strauss, 1994; Goebel & Aksenov, 1995; Mercier,Valladas & Valladas, 1995; Rink et al., 1996). U-seriesand TL dates from the oldest western European sitesare several millennia older and are considered morereliable than coeval radiocarbon dates (Bischoff et al.,1994; Mercier, Valladas & Valladas, 1995). As in theLevant, the difference between radiocarbon and otherdates in western Europe is consistent with increased

Chronology of the Later Stone Age 379

cosmogenic nuclide production during the Laschampevent (Ambrose, 1997).In North Africa, the earliest Upper Paleolithic

occurrences are found in coastal Libya, at Haua Fteahand Hagfet ed Dabba caves, dated to about 40,000 (McBurney, 1967). Nazlet Khater 4, a chert mining sitein the upper Nile Valley of Egypt, is associated withnine radiocarbon dates between 31,000 and 35,000 (Vermeersch et al., 1984). The latest Middle Paleolithicoccurrence at Sodmein Cave, Red Sea Hills, Egypt, hastool types like the MP/UP transition Emiran Industryof the Levant and may be contemporary with theLevantine MP/UP transition (Van Peer et al., 1996).However, the underlying blade-based MP industry atSodmein Cave has radiocarbon dates of 30 and >30 ka.

Dating the Middle to Later Stone Age transition insub-Saharan AfricaMany radiocarbon dates on East African sites(Ambrose, 1984b; Bower et al., 1997) are on lessreliable materials such as bone apatite carbonate(Tamers & Pearson, 1965; Collett & Robertshaw,1983). Collagen is more reliable but is rarely preservedin tropical Africa for more than 3000 years in opensites and 7000 years in rock shelters (Ambrose, 1990).Reports of radiocarbon dates on ‘‘collagen’’ fromPleistocene sites in equatorial Africa should thus beviewed with scepticism. Tooth enamel (Lee Thorp &van der Merwe, 1991; Hedges, Lee Thorp & Tuross,1995; Grun et al., 1997; Koch, Tuross & Fogel, 1997)and especially ostrich eggshell (Long, Hendershott &Martin, 1983; Freundlich et al., 1989; Brooks et al.,1990; Miller et al., 1992), are more resistant to diagen-esis, but few radiocarbon dates on these materials areavailable. As noted above, obsidian hydration andAAR dates calculated using modern temperaturesshould be considered minimum age estimates becauseLate Pleistocene temperatures were lower andcosmogenic nuclide production was higher.In sub-Saharan Africa, few sites with LSA horizons

or shaped bone tools have been dated to older than40,000 . Vogel & Beaumont (1970) proposed theMSA/LSA transition in southern Africa occurredabout 40,000 years ago, based mainly on early datesfrom Border Cave. This date has been questionedbecause most final MSA occurrences seem to date to20–25,000 (Parkington, 1990; Thackeray, 1992;Mitchell & Vogel, 1994; Deacon, 1995; Wadley, 1995).The earliest LSA occurrence at Border Cave is associ-ated with a bored stone, a ground bone point andostrich eggshell beads. Five radiocarbon dates on char-coal range from 33,000 to 38,600 and one is 45,000 (Beaumont, de Villiers & Vogel, 1978). Twelveadditional charcoal radiocarbon dates averaging38,000&2000 , and concordant radiocarbon andamino acid racemization dates of similar age on ostricheggshell, confirm the early age of the early LSA atBorder Cave (Miller et al., 1992). Ostrich eggshell

beads are associated with an MSA industry dated to42,000 at Boomplaas Cave (Deacon, 1995), andBlombos Cave has two ground bone points dated to>40,000 , associated with bifacially flaked points ofthe MSA Stillbay Industry (Henshilwood & Sealy,1997). The transition to the LSA and aspects ofmodern technology may have occurred at differenttimes in different parts of southern Africa.In equatorial Africa, the MSA/LSA transition seems

substantially earlier than in southern Africa. The LSAat Matupi Cave, Zaıre, has a charcoal radiocarbondate >40,700 (Van Noten, 1977). In the SemlikiValley, Zaıre, MSA sites with bone harpoons havediscordant TL, ESR and U-series dates ranging from82,000 to 174,000 (Brooks et al., 1995; Yellen et al.,1995). These harpoons should be directly dated toconfirm their extraordinary age. Shum Laka rockshel-ter in Cameroon (de Maret, Clist & van Neer, 1987)has an LSA occurrence at the base of the sequence,associated with dates of 30,300 and 31,700 (Cornelissen, 1996).Tanzania has several sites with LSA occurrences of

substantial antiquity. At Kisese II rockshelter, a tran-sitional MSA/LSA industry (‘‘Second Intermediate’’)with small convex scrapers and ostrich eggshell beads(Inskeep, 1962), has a radiocarbon date on ostricheggshell of 31,480 (Deacon, 1966). At Mumbarockshelter, an MSA/LSA industry (the MumbaIndustry) in middle and upper Bed V, has many backedmicroliths and ostrich eggshell beads. Six radiocarbonand U-series dates on shell and bone range from 23,600to 65,700 , including a date of >37,000 on snailshell carbonate. The Bed IV/V interface is radiocarbondated on snail shell to 36,900 (Mehlman, 1989,1991). The overlying MSA/LSA Nasera Industry inlower Bed III has radiocarbon dates on ostrich eggshellof 27,000 and 33,200 (Mehlman, 1989; Brooks &Robertshaw, 1990). Backed microliths are rare, butseveral hundred ostrich eggshell beads and two boredstones were recovered. At Nasera rockshelter, Level 6,the Nasera Industry has three radiocarbon dates onbone ‘‘collagen’’, one 230Th and one AAR date onbone, ranging from 18,500 to 26,000 . The overlyingLSA Lemuta Industry in levels 5 and 4 has fourradiocarbon dates from 14,800 to 21,600 on boneapatite and ‘‘collagen’’ (Mehlman, 1989, 1991). TheLemuta Industry in the Naisiusiu Beds at OlduvaiGorge was originally dated on bone ‘‘collagen’’ to17,550 (Leakey et al., 1972). Concordant AAR andAMS radiocarbon dates on ostrich eggshell and SCLF40Ar/39Ar now suggest an age of ¢42,000 for thisindustry (Manega, 1993).Five sites at Lukenya Hill, Kenya, have radiocarbon

dates on bone apatite and ‘‘collagen’’ ranging from12,000 to 21,500 (Gramly & Rightmire, 1973;Merrick, 1975; Gramly, 1976; Miller, 1979; Ambrose,1984a). Improved methods of pretreatment (Krueger,1991) have produced apatite dates for Lukenya siteGvJm46 as old as 29,950 (C. M. Nelson and

380 S. H. Ambrose

H. W. Krueger, pers. comm.). These should be con-sidered minimum estimates of the age of the LSA untilconfirmed by charcoal radiocarbon or other datingmethods. Prospect Farm, in the central Rift Valley ofKenya, contains MSA and early LSA horizons datedby obsidian hydration (Michels, Tsong & Nelson,1983). Artefacts of phase 3 of the Prospect Industry(MSA) produced dates of 46,500 to 53,100 . Phase 4(Second Intermediate) dates between 45,700 and53,500 . The overlying early LSA dates to 21,800–32,500 . These should be considered minimum esti-mates of ages because cooler temperatures prior to12,000 (Schroeder & Bada, 1973; Bonnefille,Roeland & Guiot, 1990) would have reduced hydrationrates.This brief survey of dated sites indicates the MP/UP

boundary appears to be approximately 40–43,000 inEurope and southern Siberia, and 43–47,000 inWestern Asia and the Sinai Peninsula. The sparserecord of North Africa also suggests an age of around40–47,000 . The MSA/LSA transition is likely to beat least 40,000 at Border Cave (Beaumont, deVilliers & Vogel, 1978), Matupi (Van Noten, 1977),Mumba, Nasera (Mehlman, 1989), Olduvai Gorge(Manega, 1993) and Prospect Farm (Michels, Tsong &Nelson, 1983), but the accuracy and validity of thesedates can be questioned. Evidence from EnkapuneYa Muto rockshelter, presented below, reinforcesevidence that the LSA began in East Africa earlier than40,000 .

Dating the transition from foraging to food production inEast AfricaMany radiocarbon dates for the Neolithic Era andearlier Holocene on bone collagen and boneapatite have been rejected as unreliable (Collett &Robertshaw, 1983). The rejection of apatite dates hassome justification (Grun et al., 1997; Koch, Tuross &Fogel, 1997), but rejection of collagen dates seemsunwarranted. The transition from hunting and gather-ing to food production, marked by the presence ofcattle and caprines (sheep and goats), is dated to 4000 in northern lowland Kenya (Barthelme, 1985).Pottery traditions (Nderit and Ileret) that are usuallyassociated with domestic animals are dated on charcoalas early as 5000 in northern Kenya (Robbins, 1972),suggesting pastoralism appeared by this date.In southern highland Kenya and northern Tanzania

there is a pronounced gap in radiocarbon dates be-tween 6000 and 3300 (Ambrose, 1984a). The tran-sition to food production in this region is marked bythe appearance of pottery and domestic animals in sitesyounger than 3300 . Because of this chronologicalgap, the most basic questions about the transition tofood production in the highlands remain unanswered.Was the transition gradual, resulting from the adop-tion of domesticates by indigenous hunter–gatherers,or abrupt, by a process of assimilation by immigrant

farmers and herders? The Holocene deposits atEnkapune Ya Muto rockshelter span this gap with 16wood charcoal radiocarbon dates between 3000 and6300 (Table 1). These radiocarbon dates, combinedwith faunal analysis by Marean (1992), provide ananswer to this question.

Stratigraphy and Chronology of EnkapuneYa MutoEnkapune Ya Muto rockshelter (also known asTwilight Cave, GtJi12) is located on the MauEscarpment, west of Lake Naivasha in the Kenya RiftValley (36)09*E, 0)50*S, elevation 2400 m). Excavationin 1982 (Ambrose, 1984b) comprised a 2#2 m test pitwith a 1#4 m step trench extending to the talus slope(Figure 1); an additional 10 m2 were excavated in 1987.Bedrock was reached at a maximum depth of 5·54 mbelow the surface. Pre-Neolithic macro- and micro-faunal assemblages were analysed by Marean (1992;Marean, Mudida & Reed, 1994). The site is located inclose proximity to many major obsidian sources in theNaivasha basin (Merrick & Brown, 1984) and virtuallyall flaked stone artefacts were made of this material.In the stratigraphical description below, the acronymsfor the strata reflect their cultural affiliations, colour,texture and/or other physical properties.Charcoal for radiocarbon dating was abundant and

well-preserved in the Holocene, but rare and poorlypreserved in the Pleistocene levels. Radiocarbon datesare presented in Table 1. Decomposed charcoal fromthe Pleistocene levels was collected with sedimentarymatrix by trowel and placed directly into aluminiumfoil. The carbonate fraction of ostrich eggshell fromone stratum was also dated. Pretreatment proceduresfor dating ostrich eggshell carbonate are describedin the notes to Table 1. Obsidian hydration dates(Table 2) were determined by Michels using rate con-stants appropriate for the chemical composition of theartefacts (Michels, Tsong & Nelson, 1983).

Stratigraphy and radiocarbon datingThe IA (Iron Age) and ELM (Elmenteitan Neolithic)strata comprise approx. 100 cm of wood ash and silts.The IA levels contain pottery of the Lanet Tradition,characterized by twisted-cord roulette decoration(Bower et al., 1977). The Elmenteitan/Iron Age bound-ary, 45–55 cm below the surface, dates to 1295 .ELM contains typical Elmenteitan stone tools madeon large punch-struck blades, and predominantlyundecorated pottery with lug handles (Bower et al.,1977; Nelson, 1980; Ambrose, 1985). The base of ELMdates to 2595 . The fauna is mainly cattle andcaprines (Ambrose, 1984b).BSS comprises approximately 20 cm of aeolian

Brown fine Sandy Silts, dated to 2570 and 2610 .HTL is a compact layer (Hard and Tacky when moist)

Chronology of the Later Stone Age 381

of caprine dung ash up to 35 cm thick, with inverteddates of 2820, 2330 and 2560 . Lithic and ceramicartefacts from BSS and HTL are scarce and culturallyundiagnostic. Caprines dominate the fauna. BS1 isapproximately 6 cm of Brown fine Sandy silt withstratigraphically inverted dates of 2860, 3280 and 2710. Lithic artefacts belong to phase 5 of the EburranIndustry (Ambrose, 1985), but the pottery does notclosely resemble any known traditions. The largemammal fauna is exclusively caprines (Marean, 1992).Lithic artefact and bone densities are very low, reflect-ing low intensity occupation during a period of greataridity.RBL1 is approximately 10 cm of light Red-Brown

fine silts, dated to 3390 , with high densities ofEburran lithics and bone, but no pottery. The fauna

includes one cow, one caprine and five wild bovidspecies (Marean, 1992). Blue-grey Volcanic Ash (VA1)up to 53 cm thick, dated to 3125 , seals underlyingstrata. RBL2 is a thick bed of hearths and woodash, divided into three main strata. Wild herbivoresfrom grassland, woodland, bush and forest habitatsdominate the fauna (Marean, 1992), suggesting closeproximity to the ecotone (boundary) between montaneforest and savanna grasslands. The third of 12 subunitsof RBL2.1, dated to 3990 , contains one caprine(Marean, 1992), which is the oldest domestic animal inthe site. The earliest pottery (Figure 2), from the baseof RBL2.1, dates to 4860 . Some decorative motifsresemble the Nderit and Ileret traditions (Wandibba,1980; Barthelme, 1985); others resemble the Suswa/Salasun tradition, which is found only in highland

Table 1. Radiocarbon dates from Enkapune Ya Muto rockshelter (GtJi12), Naivasha, Kenya, listed in stratigraphical order

Stratum Age .. Lab no. Cultural associations

Calibrated age *

Min (Calib. age) Max

IA (middle) 500 145 GX-9935 Iron Age, Lanet ceramics 330 (521) 647IA (base) 1295 140 GX-9936 Iron Age, aceramic 1010 (1243) 1309ELM (top) 2355 170 GX-9937 Elmenteitan Neolithic 2152 (2347) 2714ELM (base) 2595 175 GX-9938 Elmenteitan Neolithic 2360 (2739) 2915BSS 2570 175 GX-9939 Undefined LSA 2356 (2735) 2843BSS 2610 70 ISGS-1756 Undefined LSA 2720 (2748) 2768HTL (upper) 2820 70 ISGS-1757 Undefined LSA 2784 (2907) 3016HTL 2330 235 GX-9398† Undefined LSA 2010 (2342) 2735HTL (lower) 2560 70 ISGS-1737 Undefined LSA 2397 (2720) 2748BS1A 2860 70 ISGS-1733 Eburran 5, ceramic 2853 (2957) 3137BS1B 2710 80 ISGS-1746 Eburran 5, ceramic 2747 (2777) 2918BS1 3280 190 GX-9940 Eburran 5, mainly domestic fauna 3270 (3469) 3716RBL1 3390 70 ISGS-1745 Eburran 5, first cattle 3482 (3609) 3690BVA 3125 185 GX-9941 Eburran 5, aceramic 3005 (3355) 3550RBL2.1a1 3110 70 ISGS-1946 Eburran 5, ceramic 3210 (3275) 3383RBL2.1a1 3030 70 ISGS-2040 Eburran 5, ceramic 3061 (3230) 3335RBL2.1b2 3990 70 ISGS-2308 Eburran 5, first caprine 4310 (4220) 4522RBL2.1a 4475 210 GX-9942 Eburran 5, ceramic 4839 (5187) 5451RBL2.1b 4535 170 GX-9943v Eburran 4, preceramic 4871 (5126) 5455RBL2.1b5 4860 70 ISGS-1742 Eburran 5, first Neolithic pottery 5493 (5597) 5657RBL2.2 5265 220 GX-9944 Eburran 4 5750 (5994) 6286RBL2.3 5365 235 GX-9399† Eburran 4 5908 (6174) 6404DBS1.1 5220 70 ISGS-1732 Eburran 4 5912 (5981) 6168DBS1.2 5470 215 GX-9945 Eburran 4 5992 (6281) 6481DBS1.5 5785 200 GX-9946 Eburran 4 6357 (6586) 6856RBL3.1 6350 150 ISGS-1750 Eburran 4** 7028 (7228) 7387RBL3.1 5860 200 GX-9947 Eburran 4 6412 (6688) 6936DBL1.2 16,300 1000 UCLA Sakutiek (Early LSA)#DBL1.2 29,300 750 ISGS-1751 Sakutiek (Early LSA)DBL1.2 35,800 550 ISGS-1779 Sakutiek (Early LSA)‡DBL1.3 37,000 1100 Pta-4889 F1Q Sakutiek (Early LSA)DBL1.3 39,900 1600 Pta-4889 F2 Sakutiek (Early LSA)RBL4.1 >26,000 GX-9948 Endingi (MSA)RBL4.1 41,400 700 QL-4259 Endingi (MSA)RBL4.1+4.2 29,280 540 ISGS-2300 Endingi (MSA)§

All dates are on charcoal or charcoal-bearing sediments from hearths except for Pta-4889, on ostrich eggshell carbonate. All dates have been13C-corrected for isotopic fractionation unless noted. ..: 1 standard deviation. Abbreviations for radiocarbon labs are GX: GeochronLaboratory; ISGS: Illinois State Geological Survey; UCLA: University of California, Los Angeles; Pta: Council for Scientific and IndustrialResearch, Pretoria; QL: Quaternary Isotope Laboratory, University of Washington, Seattle.*Minimum, midpoint (in parentheses) and maximum dates tree-ring calibrated for variations in solar flux of radiocarbon, using the CALIBprogram Version 3.0, with the decadal data set and one-sigma range (Stuiver & Reimer, 1993). †No 13C correction. **Best estimate of true age(discrete hearth). ‡Very large sample of ISGS-1751, providing best estimate of true age. #Poorest estimate of true age because of small samplesize (approx. 300 mg of carbon) and long storage time (3 years). §Combined samples from two levels, analysed 4 years after excavation, thusunreliable. QSurface contamination of ostrich eggshell was removed by dissolving and discarding 9·5% of the initial weight. The next 44·2% wasdated as Fraction 1 (F1), and the final 46.1% was dated as Fraction 2 (F2). vStratigraphy in the area sampled was poorly differentiated.ISGS-1742 provides a better date for Nderit, Ileret and Salasun/Suswa pottery.

382 S. H. Ambrose

Kenya. RBL2.2 and RBL2.3 date to 5265 and 5365 ,respectively. They have well-preserved lenses of uncar-bonized to partly carbonized fine grasses that probablyserved as bedding. Extremely high species and habitatdiversity is sustained in these strata (Marean, 1992).RBL1 and RBL2 reflect extremely high intensity occu-pation of the rockshelter, when climate was similar toor slightly more humid than at present, and the forest/savanna ecotone was in the site’s vicinity.

The coarse Dark Brown Sand and silt (DBS1) hasconformable dates of 5220, 5470 and 5785 . Artefactdensities decrease, and microfauna, rodent burrows androlled artefacts from the underlying deposit increasewith depth, reflecting intense vertical mixing by rodentbioturbation. The faunal assemblage includes moreclosed habitat species (Marean, 1992; Marean, Mudida& Reed, 1994). The site was occupied less intensively asclosed habitats began to encroach on the site.

Table 2. Obsidian hydration dates from Enkapune Ya Muto (GtJi12), Naivasha, Kenya

StratumAge atpresent T)C ..

Corrected age(present T "5)C) .. Lab no. Cultural association

DBL1.3 18,858 1165 35,348 2183 26-698 Sakutiek Industry (LSA)GG1.3 24,758 1471 46,406 2758 31-613 Nasampolai Industry (LSA)RBL4.2 17,316 665 32,458 1247 31-963 Endingi Industry (MSA)

Dates were determined at Mohlab Inc., State College, Pennsylvania. Atomic absorption spectroscopy analysis of minor element abundances(Na, K, Fe, Ca and Mg) show that they all belong to the B-1 group of obsidians (Michels, Tsong & Nelson, 1983) which is locatedapproximately 10 km from the rockshelter. The present effective hydration temperature (EHT) at site’s altitude is 292·15)K and the hydrationrate is 3·88 ìm2/1000 years. Temperature-corrected ages to account for lower temperatures during the last glacial were determined with an EHTof 287·15)K and a hydration rate of 2·07 ìm2/1000 years.

IA

2

ELM

BSSHTL

BS1RBL13

VA1

RBL2-1

RBL2-2RBL2-3DBS1-1

DBS1-5

4

VA2

RBL3

DBL1

5

GG1

OL1GG2

6RBL4

E6E7E8E9E10E11E12

GTJI12 E-W section at N10

Silt

Sandy silt

Silty sand

Sand

Platy silt-sand

Light brown loam

Gray-brown loam

Dark gray loam

Black ashy loam

Light brown sandy loam

Brown crumbly loam

White ash

Orange ash

Black ash

Gray ash

Brown ash

Hard silt

Red-brown loam

Red-brown sandy loam

Red-brown gritty loam

Dark brown silty sand

Dark brown sandy silt

Dark brown ash

Gray-brown ash

Brown-gray ash

Gray gravel

Orange sandy gravel

Rodent burrow

Charcoal

Grass

Dung

Volcanic ash

Rock

Dark brown loam

VA3

1 m

Figure 1. Stratigraphical section of the south wall of the 1982 excavation at EYM. The section is at right angles to the mouth of the shelter.The dripline is located between E9 and E10.

Chronology of the Later Stone Age 383

The second Volcanic Ash (VA2) seals RBL3, acoarse, sandy, Red-Brown Loamy gravel lag deposit,comprising heavily abraded obsidian artefacts of inde-terminable industrial affinities and small numbers ofcomparatively undamaged Eburran Phase 4 artefacts.Closed habitat species predominate (Marean, 1992;Marean, Mudida & Reed, 1994), suggesting forest inthe site’s vicinity. This lag deposit probably formedwhen the shelter’s dripline was active during the earlyHolocene wet phase. A date of 6350 , from anundisturbed hearth sealed directly beneath VA2, datesthe ashfall and cessation of erosion. This erosionalevent reworked sediments and artefacts that may havebeen deposited between 6350 and 35,800 . Closedhabitats surrounded the site during the formation ofRBL3 and occupation intensity increased throughDBS1 as rainfall decreased and the forest ecotoneapproached the site.The third Volcanic Ash (VA3) is dense and compact,

and protected the underlying deposits from erosion.The Dark Brown gritty Loam (DBL1) contains

extremely high artefact densities (approx. 63,900 piecesof flaked stone) and a faunal assemblage heavilyfragmented by trampling and compaction, with fewidentifiable specimens. The associated lithic industry,named the Sakutiek Industry (Figure 3), is dominatedby typical LSA tool types, including thumbnail endscrapers and outils ecailles, and has low frequencies ofbacked microliths. It also contains low frequencies ofthin, parti-bifacially flaked small knives, flattened dis-coids, discoidal cores and faceted platform flakes,which are typical of MSA and Second Intermediateindustries. Ostrich eggshell was abundant, comprising13 complete beads, 12 bead preforms and 593 shellfragments (Figure 4). A small charcoal sample(approximately 300 mg of carbon) from DBL1.2 wasdated after 3 years of storage to 16,300&1000 (Table 1). A small split of a large sample of charcoal-impregnated sediment from level DBL1.2, analysedwithin 2 months of excavation and dated to29,300&750 , was considered unreliable becauseof its low carbon content. The remainder produceda large amount of carbon, for a more reliable dateof 35,800&550 . Ostrich eggshell from DBL1.3(the base of DBL1) produced radiocarbon datesof 37,000&1100 and 39,900&1600 on the shellexterior and interior, respectively. The latter date ismore reliable because it is on the shell fraction best-protected from contamination. DBL1 reflects highintensity occupation, including evidence of on-sitemanufacture of ostrich eggshell beads, probably duringa warm phase of marine oxygen isotope stage 3.The underlying Grey Gravels (GG1, GG2) with

intercalated layers of Orange gravelly Loamy sand(OL) have an average thickness of 1·15 m and a maxi-mum thickness of 1·7 m. Artefact and bone densitiesare extremely low. Obsidian artefacts are fresh andsharp, and the gravel is poorly sorted and angular.There was insufficient charcoal or bone for radiocarbondating. A unique blade-based LSA lithic industry,named the Nasampolai Industry, occurs throughoutGG/OL. It is dominated by very large backed bladesand geometric microliths, with low frequencies of outilsecailles, scrapers and burins. Several backed bladeshave traces of red ochre opposite the unmodified edge(Figure 5), suggesting hafting parallel to the long axis.The Nasampolai Industry is not obviously transitionalfrom the MSA, and does not closely resemble theblade-based Howiesons Poort MSA industry of South-ern Africa because it lacks radial core preparation andfaceted platform flakes. Several backed blades havetraces of red ochre opposite the unmodified edge (indi-cated by light stippling in Figure 5), suggesting haftingparallel to the long axis. The deposit is apparentlywind-deflated and must have originally been sub-stantially thicker. GG/OL thus represents a very longperiod of ephemeral occupation earlier than 40,000 .The basal horizon is a dark Red-Brown to dark

brown gritty Loam (RBL4). It contains low densities ofbone and flaked stone. The flake-based MSA lithic

(a) (b)(c)

(d)(e)

(f)

0 5

cm

Figure 2. Decorated pottery from Stratum RBL2.1. (a) combstamped horizontal lines; (b, c) impressed horizontal lines; (d) combstamped herringbone, resembling Ileret Tradition pottery; (e) im-pressed decoration with inturned, milled rim, resembling NderitTradition pottery; (f) cord-wrapped stick decoration, characteristicof the Salasun/Suswa Tradition.

384 S. H. Ambrose

industry is named the Endingi Industry. Flakes withfaceted platforms and radial dorsal scar patterns pre-dominate (Figure 6). Outils ecailles and scrapers arethe dominant tool types. Three backed microliths wererecovered from the first two levels. Two flakes havetraces of red ochre and one small ochre-stained lowergrindstone was recovered. A sample of carbonizedsediment and decomposed charcoal from a hearth,submitted almost 2 years after excavation, dates to>26,000 . Two samples combined from adjacentlevels, submitted 4 years after excavation, were datedto 29,280&540 . A large sample of carbonizedsediment and decomposed charcoal from a hearth,submitted for dating 6 months after excavation, datesto 41,400&700 . The samples with younger dateshave probably absorbed modern contaminants duringstorage (Haas, Holliday & Stuckenrath, 1986). Thedate of 41,400 on RBL4.1 must be a minimum estimateof its true age because this horizon is about 1·2 mbelow DBL1, which is dated to 39,900 , and theintervening 1·2 m of wind-deflated grey gravels prob-ably represent a very long period of accumulation.RBL4 probably represents sparse occupation of therockshelter during early oxygen isotope stage 3 orstage 4.

Obsidian hydration dating of Pleistocene artefactsThe three Pleistocene lithic industries have also beendated by obsidian hydration, using rate constantspreviously determined for the obsidian sources onwhich the artefacts were made (Michels, Tsong &Nelson, 1983). Rate constants and temperatures usedfor calculating the dates are described in the legend toTable 2. Calibration of amino acid racemization dateson bone at Lukenya Hill indicates a 6)C average

paleotemperature difference (Schroeder & Bada, 1979).Since Upper Pleistocene temperatures were probably4–6)C cooler than at present (Schroeder & Bada, 1979;Bonnefille, Roeland & Guiot, 1990), the dates werecalculated at the present temperature and at 5)C lessthan at present.The Sakutiek Industry (LSA) in DBL1 dates to

18,860 with present temperature (T) and 35,350 at T "5)C. Relatively close agreement between thetemperature-adjusted hydration date and the radiocar-bon date of 35,860 suggests the average temperature atthe site over the last 40,000 years was slightly morethan 5)C cooler than at present. However, increasedcosmogenic nuclide flux at this time has probably madethe radiocarbon date a few thousand years too young,so the average Pleistocene temperature was probablyeven cooler.Obsidian of the Nasampolai Industry (LSA), from

approximately 30–45 cm below the top of GG, dates to24,760 at the present temperature and 46,410 at T"5)C. The Endingi (MSA) industry dates to 17,320 at present temperature and 32,458 at T "5)C. Thisdoes not agree with the associated radiocarbon dateof 41,400 and should be rejected. The originalhydration layer has probably spalled and reformed, aphenomenon also observed on MSA artefacts fromProspect Farm (Michels, Tsong & Nelson, 1983).The beginning of the LSA at EYM can be estimated

by calculating sediment deposition rates in the GreyGravel stratum. The maximum rate, which providesthe minimum estimate of age, can be calculated bydividing the difference in age between the oldest radio-carbon date plus 1 .. for DBL1.3 (41,500) and thehydration date minus 1 .. for GG1.3 (43,648) by theiraverage difference in depth (approximately 38 cm), andmultiplying by the average thickness of the stratum

(a) (b) (c) (d) (e) (f) (g)

(h) (i)

(j) (k)

(l)

0 5

cm

Figure 3. Ostrich eggshell beads from the base of Stratum DBL1, associated with the Sakutiek lithic industry. (a–g) complete beads; (h–l) beadsand shell fragments broken during manufacture.

Chronology of the Later Stone Age 385

(approximately 115 cm). At the absolute maximumdeposition rate of 57 years/cm, the LSA began by atleast 46,500 . However, if the obsidian hydrationdate of 46,410 for the top third of GG1 is accurate,and the base of DBL is 39,900 , then the averagedeposition rate was approximately 167 years/cm andthe LSA may have begun earlier than 55,000 . Thebest estimate of the age of the beginning of the LSA isprobably somewhere between dates calculated from themaximum and mean deposition rates, and thus around50,000 .

Discussion and Conclusions

The transitions to food production and the LaterIron Age

Radiocarbon dates from EYM provide new insightinto the timing of important economic and technologi-cal transitions during the Holocene. The date of 1295 for the Elmenteitan Neolithic/Iron Age transitionis consistent with evidence from other sites in theRift Valley (Ambrose, 1984c, 1985). Dates for thebeginning of the Elmenteitan of 2600 are younger

(a)

(b) (c) (d)

(e)(f)

0 5

cm(g)

(h) (i) (j) (k) (l)

(m)(n)

(o)

Figure 4. Shaped stone artefacts of the Sakutiek Industry from stratum DBL1. All were made on obsidian. (a–d) microlithic backed flakes andbladelets; (e–f) crescents; (g) bifacially flaked discoid/core; (h–l) thumbnail scrapers; (m–o) parti-bifacial and bifacially flaked knives.

386 S. H. Ambrose

than at Njoro River Cave, where it dates to 3100 (Merrick & Monaghan, 1984). This difference in datesshould not raise questions about the cultural affinitiesof the Njoro people. Low artefact densities and undi-agnostic lithic and ceramic traditions in BSS and HTLmake it impossible to determine the cultural affinitiesof these pastoralists. However, immigrant Elmenteitanand Savanna Pastoral Neolithic populations were evi-dently contemporary with the makers of the EburranIndustry after 3300 (Ambrose, 1984a, 1985).The earliest pastoral pottery traditions at EYM

(Salasun, Ileret and Nderit) are dated to 4860 , butare not associated with domestic animals. Dates onNderit sites in northern Kenya as old as 5000 (Robbins, 1972) have been rejected without apparentjustification (Collett & Robertshaw, 1983), but shouldbe reconsidered given the date for similar ceramics at

EYM. Marean’s (1992) faunal analysis demonstratesthat domestic caprines appear in EYM at 3990 , 870years after pottery suggests contact with pastoralistsfrom northern Kenya. The adoption of pastoralism byEburran hunter–gatherers was thus gradual and notcompleted until 3280 .

Chronological gap in the Middle HoloceneThe sequence at EYM fills a substantial gap in theradiocarbon chronology of the East African highlandsbetween 6000 and 3300 that correlates with a periodof drier climate (Richardson, 1972). Carbon isotopeanalysis of soil profiles in the Naivasha basin demon-strates the forest/savanna ecotone rose to higher alti-tudes during this dry phase (Ambrose & Sikes, 1991).Eburran hunter–gatherers may have preferred to settle

(a)

(b)

(c) (d)(e) (f)

0 5

cm

(g)

(h) (i) (j)

(k)(l)

Figure 5. Shaped stone artefacts of the Nasampolai Industry (earliest LSA) from stratum GG. (a–i) backed blades and geometric microliths;(j) end and denticulate side scraper; (k, l) double burins. Stippling on (b), (c), (f) and (i) indicates the location of traces of red ochre.

Chronology of the Later Stone Age 387

on this ecotone and followed it to higher altitudesduring the dry phase and to lower altitudes during theEarly Holocene wet phase. High intensity occupationof EYM by hunter–gatherers, associated with forestand savanna species (Marean, 1992; Marean, Mudida& Reed, 1994) during this arid period, is consistentwith the model of ecotonal settlement preference(Ambrose, 1986).The gap in chronology also spans the gradual tran-

sition to food production among indigenous hunter–gatherers who made the Eburran lithic industry. Thistransition began at 4860 , marked by the appearanceof pottery that in northern Kenya is associated withdomestic animals. The transition was completed by3280 when domestic animals finally dominate thefaunal assemblage in BS1. Drier climate during thisperiod may have rendered highland environmentsmarginal for food production, thus delaying andprolonging this transition (Ambrose, 1984a).Integrating the local record of the spread of food

production with continental chronologies requires cali-bration to an absolute time scale by adjusting radio-carbon dates for past variation in the production ofatmospheric radiocarbon (Stuiver & Kra, 1986). Treering-calibrated dates (Table 1) (Stuiver & Reimer,1993) for the earliest pottery and domestic animals atEYM show they appeared by 5597 and 4220 calibratedyears , respectively. Fully pastoral sites appear in

highland East Africa at c. 3470 calibrated years .Domestic cattle first appeared in North Africa after8000 (Clutton-Brock, 1993), but the spread ofpastoralism south of the Sahara apparently did notoccur until later than c. 4000 radiocarbon years (Smith, 1984). The first domestic animal in the CapeProvince, South Africa, is dated to c. 2060 cal. years, but domestic-dominated faunas are all youngerthan 1700 (Sealy & Yates, 1994). Pastoralism thusappeared in South Africa c. 1400 years after it wasestablished in the East African highlands. Domesticanimals apparently spread from north to south inseveral brief episodes separated by long periods ofstasis. The spread from northern to equatorial Africamay have been delayed by climate-induced shifts in theposition of tsetse-infested floral zones (Smith, 1984),but the spread to southern Africa may not beassociated with any obviously climatic changes.

The antiquity of the Later Stone AgeRadiocarbon samples from Upper Pleistocene levels atEYM with low carbon contents, and which were smalland/or stored for long periods (2–4 years) beforeanalysis, had systematically younger ages than largesamples dated soon after excavation. Previous studiesof carbonized sediments dated soon after excavationand redated 3–4 years later show significant decreases

(a)

(b) (c)

(d)(e)

(f)

0 5

cm

(g)

(h)(i)

Figure 6. Shaped stone artefacts of the Endingi (MSA) industry from stratum RBL4. (a–c) backed crescents; (d) triangular flake point withalternate lateral retouch and a burin or impact scar at the distal end; (e) oblique end scraper; (f) outil ecaille; (g, h) whole flakes, with red ochreindicated by stippling on the platform and dorsal surface of (g) and platform, dorsal and ventral surface of (h); (i) whole flake with facetedplatform from a radially prepared core.

388 S. H. Ambrose

in age after storage (Haas, Halliday & Stuckenrath,1986). Ostrich eggshell has consistently providedunambiguous radiocarbon dates (Long, Hendershott &Martin, 1983; Freundlich et al., 1989; Miller et al.,1992; Brooks et al., 1990). Radiocarbon dates onostrich eggshell overlying the Early LSA NasampolaiIndustry provide a minimum age of 39,900 for theearliest LSA. Temperature-calibrated obsidian hydra-tion dates demonstrate that the MSA/LSA transitionis probably significantly older than 46,000 inEast Africa, and is thus older than the MP/UPtransition (Emiran Industry) and the early UP(Ahmarian Industry) dated to c. 47–43,000 in theLevant (Marks, 1983; Phillips, 1994; Schwarcz, 1994;Bar-Yosef et al., 1996).Dates of 37–39,900 on ostrich eggshell from the

bead workshop in DBL1 at EYM confirm early datesfor these artefacts at Mumba-Hohle (Mehlman, 1989),Kisese (Deacon, 1966), Boomplaas (Deacon, 1995) andBorder Cave (Beaumont, de Villiers & Vogel, 1978;Miller et al., 1992). The Sakutiek Industry in DBL1may also provide the most reliable estimate of the ageof the thumbnail scraper-dominated occurrences atLukenya Hill site GvJm46, dated to between 19,000and 30,000 on bone apatite (Miller, 1979; C. M.Nelson & H. W. Krueger, pers. comm.), and GvJm22,dated to 17,000 on bone ‘‘collagen’’ (Gramly, 1976).As noted above, redating of the Naisiusiu LSA indus-try increased its age from 17,000 (Leakey et al., 1972)to 42,000 (Manega, 1993). If all published boneapatite dates from Lukenya are also systematicallyyoung, the true age of the Lukenya Homo sapienscranium at site GvJm22 (Gramly & Rightmire, 1973),may be substantially older than 17,000 .The Sakutiek Industry (second LSA) at EYM has

technological and typological features of a transitionalMSA/LSA industry like the Magosian, but is far fromthe MSA/LSA boundary. Assemblages that appeartransitional are thus not necessarily closest in time tothe MSA/LSA transition. The underlying NasampolaiIndustry is fully LSA and is not obviously transitionalfrom the MSA. It does not closely resemble the blade-based Howiesons Poort MSA industry of SouthernAfrica (Singer & Wymer, 1982) or the Mumba andNasera industries in Tanzania (Mehlman, 1989; 1991),both of which have large backed microliths, like theLSA, but also have bifacial and unifacial points andradially prepared cores and flakes, like the MSA.Although the lithic industries at EYM and Mumbadiffer, patterns of change through time are similarbecause earlier industries have more backed tools thanlater ones.

Significance of ostrich eggshell beads for the origin ofmodern human behaviourOstrich eggshell beads associated with the SakutiekIndustry are unlikely to be intrusive from higher strata.The overlying Holocene strata, which are 3·25 m thick

and exposed over 16 m2 in the excavation, contain 76whole beads, one bead preform (1·48 beads/m3) and222 unworked ostrich eggshell fragments. DBL1,which is 30–40 cm thick and exposed over only 6 m2,contains 13 complete beads, 12 bead preforms brokenduring drilling or grinding (11·9 beads and preforms/m3) and 593 shell fragments. Radiocarbon dating ofthe ostrich eggshell from which beads were madedemonstrates that beads were manufactured by39,900&1600 , confirming early dates for such beadselsewhere in Africa (Inskeep, 1962; Deacon, 1966;Beaumont, de Villiers & Vogel, 1978; Mehlman, 1989;Miller et al., 1992; Deacon, 1995). Applying a correc-tion for increased cosmogenic nuclide productioncould move this date back approximately 3500 years(Raisbeck et al., 1987; Mazaud et al., 1991; Sternberg& Damon, 1992; Laj, Mazaud & Duplessy, 1996).These beads may mark the dawn of an era of new

artefact manufacturing techniques (drilling and grind-ing) and of personal adornment, but may also mark afar more significant innovation in modern humanbehaviour. Among modern Kalahari !Kung Sanhunter–gatherers, beads are one of the most importantitems used in a system of gift-giving and exchange. Thisinstitutionalized system of delayed reciprocity, calledhxaro, functions to strengthen networks of regionalsocial and economic relationships that serve as a socialsafety net, enhancing survival in marginal environ-ments (Wiessner, 1982, 1986). Hxaro partners aresought with individuals who live in areas with comp-lementary resources: 24% of partners live 50–100 kmaway, and 9% live more than 100 km away. The wordfor sewn beadwork is synonymous with the word forhxaro gifts, and beadwork is considered an appropriategift for all occasions (Wiessner, 1986). This suggestsgreat antiquity for the use of ostrich eggshell beads inthe hxaro system of gift exchange.Deacon (1992; 1995) has proposed that standardized

backed tools on exotic raw materials in the HowiesonsPoort, 70,000 years ago, mark the initiation of tradenetworks like those of modern San. Arrows are indeedwidely traded among the !Kung at present (Wiessner,1983). Although I agree with Deacon that it is likelythat people travelled longer distances and may havehad intergroup exchange during the Howiesons Poort(Ambrose & Lorenz, 1990) it is difficult to prove orrefute the hypothesis that this class of stone artefactshad acquired such symbolic value. Inferring symbolicmeanings of non-utilitarian items such as beads is lessambiguous. Mitchell (1996) suggests that hxaro giftexchange networks were established in southeasternsouthern Africa by the early Holocene or terminalPleistocene, based on the presence of marine shell andostrich eggshell beads in areas of the interior of south-ern Africa where both materials do not naturally occur.If ostrich eggshell beads manufactured in eastern andsouthern Africa 40–45,000 years ago were made for ahxaro-like delayed reciprocity system, then they maysignify the invention of a symbolic marker for a social

Chronology of the Later Stone Age 389

security system that permitted behaviourally modernhumans to survive in more risky environments.

Significance of ostrich eggshell beads for the expansionand dispersal of modern humansMitochondrial DNA studies indicate that modernhuman populations expanded within and outside ofAfrica around 40,000–75,000 (Harpending et al.,1993; Sherry et al., 1994; Mountain & Cavalli-Sforza,1997; Watson et al., 1997). The invention of modernhuman technology in sub-Saharan Africa is consideredto have been a prime mover in this demographicexpansion (Harpending et al., 1993; Sherry et al.,1994). The appearance of the LSA at EYM well before46,000 provides support for the hypothesis of atechnologically mediated population expansion inAfrica. Equatorial East Africa may have been thesource area for this radiation because it may have beena refugium from the harsh climate of the last ice age,but direct evidence for population expansion in theearly LSA is lacking (Ambrose, in press). Sub-SaharanAfrica apparently had a larger population size thanother continents during the last ice age (Relethford &Harpending, 1994), and thus may have had the largestreservoir of knowledge on which to base new innova-tions (Ambrose, in press). One may further speculatethat if ostrich eggshell beads reflect an enhanced sym-bolic system of socially mediated risk-minimizationand social solidarity, this may have facilitated popu-lation increase in Africa, the spread of modern humansout of Africa and the replacement of archaic humanpopulations in Eurasia.

Chronology and geographical area of origin of modernhuman behaviourThe archaeological sequence of EYM seems to providesupport for Klein’s (1989a, 1992, 1995) hypothesis ofan equatorial East African origin for modern humanbehaviour. However, an origin anywhere in humidequatorial Africa cannot be discounted. Moreover, thechronology of the origin of many aspects of modernhuman behaviour is still uncertain. More sites span-ning this transition in Ethiopia, Central and WestAfrica should be excavated and dated by accuratelycalibrated methods (Renne et al., 1997) before EastAfrica is considered the Garden of Eden for modernhuman behaviour.

AcknowledgementsI thank the Government of Kenya for permission toconduct field research and the National Museums ofKenya and Charles M. Nelson for logistical support.Research was supported by grants BNS 81-18026 andBNS 87-07150 from the National Science Foundation,the L. S. B. Leakey Foundation and the Universityof Illinois Research Board. I am grateful to Jack Liu,

Wang Hong, the late Harold Krueger, CharlesSullivan, Minze Stuiver, Dave McJunkin and JohnVogel, for providing radiocarbon dates, and TedGoebel for inking the lithic illustrations. RichardKlein, Curtis Marean, Nancy Sikes, Olga Soffer,Henry Schwarcz and Fred Wendorf provided usefulcomments and corrections.

ReferencesAitken, M. J. (1990). Science-Based Dating in Archaeology. London:Longman.

Aitken, M. J. (1994). Optical dating: a non-specialist review.Quaternary Science Reviews 13, 503–508.

Ambrose, S. H. (1984a). The introduction of pastoral adaptations tothe highlands of East Africa. In (J. D. Clark & S. A. Brandt, Eds)From Hunters to Farmers. Berkeley: University of California Press,pp. 212–239.

Ambrose, S. H. (1984b). Holocene Environments and HumanAdaptations in the Central Rift Valley, Kenya. Ph.D. Thesis.University of California, Berkeley.

Ambrose, S. H. (1984c). Excavations at Deloraine, Rongai, 1978.Azania 19, 79–104.

Ambrose, S. H. (1985). Excavations at Masai Gorge rockshelter,Naivasha, Kenya. Azania 20, 29–67.

Ambrose, S. H. (1986). Hunter–gatherer adaptations to non-marginal environments: an ecological and archaeological assess-ment of the Dorobo model. Sprache und Geschichte in Africa 7(2),11–42.

Ambrose, S. H. (1990). Preparation and characterization of boneand tooth collagen for isotopic analysis. Journal of ArchaeologicalScience 17, 431–451.

Ambrose, S. H. (1997). Implications of geomagnetic field variationfor the chronology of the Middle to Later Stone Age and Middleto Upper Paleolithic transitions. Journal of Human Evolution32(4), A2.

Ambrose, S. H. (in press). Late Pleistocene human populationbottlenecks, volcanic winter, and the differentiation of modernhumans. Journal of Human Evolution.

Ambrose, S. H. & DeNiro, M. J. (1989). Climate and habitatreconstruction using stable carbon and nitrogen isotope ratios ofcollagen in prehistoric herbivore teeth from Kenya. QuaternaryResearch 31, 407–422.

Ambrose, S. H. & Lorenz, K. G. (1990). Social and ecologicalmodels for the Middle Stone Age in southern Africa. In (P.Mellars, Ed.) The Emergence of Modern Humans. An Archaeologi-cal Perspective. Edinburgh: Edinburgh University Press, pp. 3–33.

Ambrose, S. & Sikes, N. (1991). Soil carbon isotope evidence forHolocene habitat change in the Kenya Rift Valley. Science 253,1402–1405.

Bard, E., Hamelin, B., Fairbanks, R. G. & Zindler, A. (1990).Calibration of the 14C timescale over the past 30,000 years usingmass spectrometric U-Th ages from Barbados corals. Nature 345,405–410.

Barthelme, J. W. (1985). Fisher–Hunters and Neolithic Pastoralists inEast Turkana, Kenya. Oxford: BAR International Series 254.

Bar-Yosef, O., Arnold, M., Mercier, N., Belfer-Cohen, A.,Goldberg, P., Housley, R., Laville, H., Meignen, L., Vogel, J. C. &Vandermeersch, B. (1996). Dating of Upper Paleolithic layers inKebara Cave, Mt. Carmel. Journal of Archaeological Science 23,297–306.

Beaumont, P. B., de Villiers, H. & Vogel, J. C. (1978). Modern manin Sub-Saharan Africa prior to 49,000 years B.P.: a review andevaluation with particular reference to Border Cave. South AfricanJournal of Science 74, 409–419.

Binford, L. R. (1989). Isolating the transition to cultural adapta-tions: an organizational approach. In (E. Trinkaus, Ed.) TheEmergence of Modern Humans. Cambridge: Cambridge UniversityPress, pp. 18–41.

390 S. H. Ambrose

Bischoff, J. L., Ludwig, K., Garcia, J. F., Carbonell, E., Vaquero,M., Stafford, T. W., Jr & Jull, A. J. T. (1994). Dating of the basalAurignacian sandwich at Abric Romanie (Catalunya, Spain) byradiocarbon and uranium series. Journal of Archaeological Science21, 541–551.

Bonnefille, R., Roeland, J. J. & Guiot, J. (1990). Temperature andrainfall estimates for the past 40,000 years in equatorial Africa.Nature 346, 347–349.

Bower, J. R. F., Nelson, C. M., Waibel, A. F. & Wandibba, S.(1977). The University of Massachusetts’ Later Stone Age/Pastoral Neolithic comparative study in central Kenya: an over-view. Azania 12, 119–146.

Brooks, A. S. & Robertshaw, P. T. (1990). The Glacial Maximum intropical Africa: 22,000 to 12,000 BP. In (C. Gamble & O. Soffer,Eds) The World at 18,000 BP, Vol. 2: Low Latitudes. London:Unwin Hyman, pp. 121–169.

Brooks, A. S., Hare, P. E., Kokis, J. E., Miller, G. H., Ernst, R. D.& Wendorf, F. (1990). Dating Pleistocene archaeological sites byprotein diagenesis in ostrich eggshell. Science 248, 60–64.

Brooks, A. B., Helgren, D. M., Cramer, J. S., Franklin, A.,Hornyak, W., Keating, J. M., Klein, R. G., Rink, W. J., Schwarcz,H., Leith Smith, J. N., Stewart, K., Todd, N. E., Verniers, J. &Yellen, J. (1995). Dating and context of three Middle Stone Agesites with bone points in the Upper Semliki Valley, Zaire. Science268, 548–553.

Chappel, J., Head, J. & Magee, J. (1996). Beyond the radio-carbon limit in Australian archaeology and Quaternary research.Antiquity 70, 543–552.

Chen, Y., Smith, P. E., Evensen, N. M., York, D. & Lajoie, K. R.(1996). The edge of time: dating younger volcanic ash layers withthe 40Ar/39Ar laser probe. Science 274, 1176–1178.

Clutton-Brock, J. (1993). The spread of domestic animals in Africa.In (T. Shaw, P. Sinclair, B. Andah & A. Okpoko, Eds) TheArchaeology of Africa. Food, Metals and Towns. New York:Routledge, pp. 61–70.

Collett, D. & Robertshaw, P. T. (1983). Problems in the interpret-ation of radiocarbon dates: the Pastoral Neolithic of East Africa.African Archaeological Review 1, 57–74.

Cornelissen, E. (1996). Shum Laka (Cameroon): late Pleistoceneand early Holocene deposits. In (G. Pwiti & R. Soper, Eds)Aspects of African Archaeology. Harare: University of ZimbabwePublications, pp. 257–263.

Deacon, H. J. (1989). Late Pleistocene paleoecology and archaeologyin the southern Cape, South Africa. In (P. Mellars & C. Stringer,Eds) The Human Revolution: Behavioural and Biological Perspec-tives on the Origins of Modern Humans. Edinburgh: EdinburghUniversity Press, pp. 547–564.

Deacon, H. J. (1992). Southern Africa and modern human origins.Philosophical Transactions of the Royal Society, London B337,177–183.

Deacon, H. J. (1995). Two Late Pleistocene–Holocene archaeologi-cal depositories from the Southern Cape, South Africa. SouthAfrican Archaeological Bulletin 50, 121–131.

Deacon, H. J. & Wurz, S. J. D. (1996). Klasies River main site, Cave2: a Howiesons Poort occurrence. In (G. Pwiti & R. Soper, Eds)Aspects of African Archaeology. Harare: University of ZimbabwePublications, pp. 213–218.

Deacon, J. (1966). An annotated list of radiocarbon dates for sub-Saharan Africa. Annals of the Cape Provincial Museum 5, 5–84.

Freundlich, J. C., Kuper, R., Breuning, P. & Bertram, H.-G. (1989).Radiocarbon dating of ostrich eggshells. Radiocarbon 31, 1030–1034.

Gillot, P. Y., Labeyrie, J., Laj, C., Valladas, G., Guerin, G., Popeau,G. & Delibrais, G. (1970). Age of the Laschamp paleomagneticexcursion revisited. Earth and Planetary Science Letters 42, 444–450.

Goebel, T. & Aksenov, M. (1995). Accelerator radiocarbon datingthe initial Upper Paleolithic in southeast Siberia. Antiquity 69,349–357.

Goebel, T., Derevianko, A. P. & Petrin, V. T. (1993). Dating theMiddle-to-Upper-Paleolithic transition at Kara-Bom. CurrentAnthropology 34, 452–458.

Gramly, R. M. (1976). Upper Pleistocene archaeological occurrencesat site GvJm/22, Lukenya Hill, Kenya. Man 11, 319–344.

Gramly, R. M. & Rightmire, P. G. (1973). A fragmentary craniumand dated Later Stone Age assemblage from Lukenya Hill, Kenya.Man 8, 571–579.

Grun, R. (1993). Electron spin resonance dating in paleoanthropol-ogy. Evolutionary Anthropology 2, 172–181.

Grun, R., Abeyratne, M., Head, J., Tuniz, C. & Hedges, R. E. M.(1997). AMS 14C analysis of teeth from archaeological sitesshowing anomalous ESR dating results. Quaternary ScienceReviews (Quaternary Geochronology) 16, 437–444.

Haas, H., Holliday, V. & Stuckenrath, R. (1986). Dating ofHolocene stratigraphy with soluble and insoluble organic fractionsat the Lubbock Lake archaeological site, Texas: an ideal casestudy. Radiocarbon 28, 473–485.

Harpending, H. C., Sherry, S. T., Rogers, A. L. & Stoneking, M.(1993). The genetic structure of ancient human populations.Current Anthropology 34, 483–496.

Hedges, R. E. M., Lee-Thorp, J. A. & Tuross, N. (1995). Is toothenamel carbonate a suitable material for radiocarbon dating?Radiocarbon 37, 285–290.

Henshilwood, C. & Sealy, J. C. (1997). Bone artefacts from theMiddle Stone Age at Blombos Cave, Southern Cape, SouthAfrica. Current Anthropology 38, 890–895.

Hu, Q., Smith, P. E., Evensen, N. & York, D. (1994). Lasing into theHolocene: extending the 40Ar/39Ar laser probe method into the 14Ctime range. Earth and Planetary Science Letters 123, 331–336.

Inskeep, R. (1962). The age of the Kondoa rock paintings in light ofrecent excavations at Kisese II rock shelter. In (G. Mortelmans &J. Nenquin, Eds) Actes du IVe Congres Panafricain de Prehistoireet de l’etude du Quaternaire. Tervuren: Musee Royal de L’AfriqueCentral, Belgique Annales, Series IN-8o, Sciences Humaines,no 40, pp. 249–256.

Johnson, B. J. & Miller, G. H. (1997). Archaeological applications ofamino acid racemization. Archaeometry 39, 265–287.

Jones, M., Sheppard, P. J. & Sutton, D. G. (1997). Soil temperatureand obsidian hydration dating: a clarification of variables affectingaccuracy. Journal of Archaeological Science 24, 505–516.

King, K., Jr & Bada, J. L. (1979). Effects of in-situ leaching on aminoacid racemization rates in fossil bone. Nature 281, 135–137.

Klein, R. G. (1989a). The Human Career. Chicago: University ofChicago Press.

Klein, R. G. (1989b). Biological and behavioural perspectives onmodern human origins in Southern Africa. In (P. Mellars & C.Stringer, Eds) The Human Revolution: Behavioural and BiologicalPerspectives on the Origins of Modern Humans. Edinburgh:Edinburgh University Press, pp. 529–546.

Klein, R. G. (1992). The archaeology of modern human origins.Evolutionary Anthropology 1, 5–14.

Klein, R. G. (1995). Anatomy, behavior, and modern human origins.Journal of World Prehistory 9, 167–198.

Koch, P. L., Tuross, N. & Fogel, M. L. (1997). The effects of samplepretreatment and diagenesis on the isotopic integrity of carbonatein biogenic hydroxylapatite. Journal of Archaeological Science 24,417–429.

Koslowski, J. K. (1988). Transition from the Middle to the earlyUpper Paleolithic in central Europe and the Balkans. In (J. F.Hoffecker & C. A. Wolf, Eds) The Early Upper Paleolithic:Evidence from Europe and the Near East. Oxford: BAR Inter-national Series 437, pp. 529–546.

Krueger, H. W. (1991). Exchange of carbon with biological apatite.Journal of Archaeological Science 18, 355–361.

Lahr, M. & Foley, R. (1994). Multiple dispersals and modern humanorigins. Evolutionary Anthropology 3, 48–60.

Laj, C., Mazaud, A. & Duplessy, J.-C. (1996). Geomagnetic intensityand 14C abundance in the atmosphere and ocean during the past50 kyr. Geophysical Research Letters 23, 2045–2048.

Leakey, M. D., Hay, R. L., Thurber, R., Protsch, R. & Berger, R.(1972). Stratigraphy, archaeology, and age of the Ndutu andNaisiusiu beds, Olduvai Gorge, Tanzania. World Archaeology 3,328–341.

Chronology of the Later Stone Age 391

Lee-Thorp, J. A. & van der Merwe, N. J. (1991). Aspects of thechemistry of modern and fossil biological apatites. Journal ofArchaeological Science 18, 343–354.

Long, A., Hendershott, R. B. & Martin, P. S. (1983). Radiocarbondating of fossil eggshell. Radiocarbon 25, 533–539.

Manega, P. C. (1993). Geochronology, Geochemistry, and IsotopicStudy of the Plio-Pleistocene Hominid Sites and the NgorongoroVolcanic Highland in Northern Tanzania. Ph.D. Thesis. Universityof Colorado, Boulder.

Marean, C. W. (1992). Hunter to herder: large mammal remainsfrom the hunter–gatherer occupation at Enkapune Ya Mutorockshelter, central Rift Valley, Kenya. African ArchaeologicalReview 10, 65–127.

Marean, C. W., Mudida, N. & Reed, K. E. (1994). Holocenepaleoenvironmental change in the Kenyan central Rift Valley asindicated by micromammals from Enkapune Ya Muto rock-shelter. Quaternary Research 41, 376–389.

de Maret, P., Clist, B. & van Neer, W. (1987). Resultats despremieres fouilles dans les abris de Shum Laka et d’Abeke.L’Anthropologie 91, 559–584.

Marks, A. E. (1983). The Middle to Upper Paleolithic transition inthe Levant. Advances in World Archaeology 2, 51–98.

Mazaud, A., Laj, C., Bard, E., Arnold, M. & Tric, E. (1991).Geomagnetic field control of 14C production over the last 80 ky:implications for the radiocarbon time-scale. Geophysical ResearchLetters 18, 1885–1888.

McBurney, C. B. M. (1967). The Haua Fteah. Cambridge:Cambridge University Press.

Mehlman, M. J. (1989). Late Quaternary Archaeological Sequences inNorthern Tanzania. Ph.D. Thesis. University of Illinois, Urbana.

Mehlman, M. J. (1991). Context for the emergence of modern man ineastern Africa: some new Tanzanian evidence. In (J. D. Clark,Ed.) Cultural Beginnings. Bonn: Dr. Rudolf Habelt GMBH,pp. 177–196.

Mercier, N., Valladas, H. & Valladas, G. (1995). Flint thermolumi-nescence dates from the CFR laboratory at GIF: contributions tothe study of the Middle Paleolithic. Quaternary Science Reviews14, 351–364.

Merrick, H. V. (1975). Change in Later Pleistocene Lithic Industriesin Eastern Africa. Ph.D. Thesis. University of California, Berkeley.

Merrick, H. V. & Brown, F. H. (1984). Obsidian sources andpatterns of source utilization in Kenya and northern Tanzania:some initial findings. African Archaeological Review 2, 129–152.

Merrick, H. V. & Monaghan, M. C. (1984). The date of the crematedburials in the Njoro River Cave. Azania 19, 7–11.

Meynadier, L., Valet, J.-P., Weeks, R., Shackleton, N. J. & Hagee,V. L. (1992). Relative geomagnetic intensity of the field during thelast 140 ka. Earth and Planetary Science Letters 114, 39–57.

Michels, J. W., Tsong, I. S. T. & Nelson, C. M. (1983). Obsidiandating and East African archaeology. Science 219, 361–366.

Miller, G. H., Beaumont, P. B., Jull, A. J. T. & Johnson, B. (1992).Pleistocene geochronology and paleothermometry from proteindiagenesis in ostrich eggshells: implications for the evolution ofmodern humans. Philosophical Transactions of the Royal Society,London B337, 149–157.

Miller, G. H., Magee, J. W. & Jull, A. J. T. (1997). Low-latitudeglacial cooling in the southern hemisphere from amino-acidracemization in emu eggshells. Nature 385, 241–244.

Miller, S. F. (1979). Lukenya Hill, GvJm46, excavation report.Nyame Akuma 14, 31–34.

Mitchell, P. J. (1996). Prehistoric exchange and interaction insoutheastern southern Africa: marine shell and ostrich eggshell.African Archaeological Review 13, 35–76.

Mitchell, P. J. & Vogel, J. C. (1994). New radiocarbon dates fromSehonghong Rock Shelter, Lesotho. South African Journal ofScience 90, 284–288.

Mountain, J. L. & Cavalli-Sforza, L. L. (1997). Multilocus geno-types, a tree of individuals, and human evolutionary history.American Journal of Human Genetics 61, 701–718.

Nelson, C. M. (1980). The Elmenteitan lithic industry. In (R. E.Leakey & B. A. Ogot, Eds) Proceedings of the Eighth Panafrican

Congress of Prehistory and Quaternary Studies, Nairobi, 1977.Nairobi: International Louis Leakey Memorial Institute forAfrican Prehistory, pp. 275–278.

Parkington, J. E. (1990). A critique of the consensus view of the ageof the Howiesons Poort assemblages in South Africa. In (P.Mellars, Ed.) The Emergence of Modern Humans. An Archaeo-logical Perspective. Edinburgh: Edinburgh University Press,pp. 34–55.

Phillips, J. L. (1994). The Upper Paleolithic chronology of theLevant and the Nile Valley. In (O. Bar-Yosef & R. S. Kra, Eds)Late Quaternary Chronology and Paleoclimates of the EasternMediterranean. Tucson: Radiocarbon, pp. 169–176.

Raisbeck, G. M., Yiou, F., Bouries, D., Lorius, C., Jouzel, J. &Barkov, N. I. (1987). Evidence for two intervals of enhanced 10Bedeposition in Antarctic ice during the last glacial period. Nature326, 273–277.

Relethford, J. H. & Harpending, H. C. (1994). Craniometric vari-ation, genetic theory, and modern human origins. AmericanJournal of Physical Anthropology 95, 249–270.

Renne, P. R., Sharp, W. D., Deino, A. L., Orsi, G. & Civetta, L.(1997). 40Ar/39Ar dating into the historical realm: calibrationagainst Pliny the Younger. Science 277, 1279–1280.

Richardson, J. L. (1972). Paleolimnological records from Rift lakesin central Kenya. Palaeoecology of Africa 6, 131–136.

Rink, W. J., Schwarcz, H. P., Lee, H. K., Cabrera Valdes, V.,Bernaldo de Quiros, F. & Hoyos, M. (1996). Journal of Archaeo-logical Science 23, 945–951.

Robbins, L. H. (1972). Archaeology in the Turkana District, Kenya.Science 176, 359–366.

Schroeder, R. A. & Bada, J. L. (1973). Glacial–postglacial tempera-ture difference deduced from aspartic acid racemization in fossilbones. Science 182, 479–482.

Schwarcz, H. P. (1992). Uranium series dating in paleoanthropology.Evolutionary Anthropology 1, 56–62.

Schwarcz, H. P. (1994). Chronology of modern humans in theLevant. In (O. Bar-Yosef & R. S. Era, Eds) Late QuaternaryChronology and Paleoclimates of the Eastern Mediterranean.Tucson: Radiocarbon, pp. 21–31.

Sealy, J. C. & Yates, R. (1994). The chronology of the introductionof pastoralism to the Cape, South Africa. Antiquity 68, 58–67.

Singer, R. & Wymer, J.(1982). The Middle Stone Age at Klasies RiverMouth in South Africa. Chicago: University of Chicago Press.

Sherry, S., Rogers, A. R., Harpending, H., Soodyall, H., Jenkins, T.& Stoneking, M. (1994). Mismatch distributions of mtDNA revealrecent human population expansions.Human Biology 66, 761–775.

Smith, A. B. (1984). Environmental limitation on prehistoricpastoralism in Africa. African Archaeological Review 2, 99–111.

Sternberg, R. S. & Damon, P. E. (1992). Implications of dipolemoment secular variation from 50,000–10,000 years for theradiocarbon record. Radiocarbon 34, 189–198.

Strauss, L. G. (1994). Upper Paleolithic origins and radiocarboncalibration: more new evidence from Spain. Evolutionary Anthro-pology 3, 195–198.

Stuiver, M. & Kra, R. S. (1986). Proceedings of the 12th Inter-national 14C Conference. Radiocarbon 28(2B), 805–1030.

Stuiver, M. & Reimer, P. J. (1993). Extended 14C data base andrevised Calib 3.0 14C age calibration program. Radiocarbon 35,215–230.

Tamers, M. A. & Pearson, F. J. (1965). Validity of radiocarbon dateson bone. Nature 208, 1053–1055.

Taylor, R. E. (1996). Radiocarbon dating: the continuing revolution.Evolutionary Anthropology 4, 169–181.

Thackeray, A. I. (1992). The Middle Stone Age south of theLimpopo River. Journal of World Prehistory 6, 385–440.

Van Noten, F. (1977). Excavations at Matupi Cave. Antiquity 51,35–40.

Van Peer, P., Vermeersch, P. M., Moyersons, J. & Van Neer, W.(1996). Palaeolithic sequence of Sodmein Cave, Red Sea Moun-tains, Egypt. In (G. Pwiti & R. Soper, Eds) Aspects of AfricanArchaeology. Harare: University of Zimbabwe Publications,pp. 149–156.

392 S. H. Ambrose

Vermeersch, P. M., Paulissen, E., Gijselings, G., Otte, M., Thomas,A., Van Peer, P. & Lauwers, R. (1984). 33,000-yr old chert miningsite and related Homo in the Egyptian Nile Valley. Nature 309,342–344.

Vogel, J. & Beaumont, P. B. (1970). Revised radiocarbon chronologyfor the Later Stone Age in South Africa. Nature 237, 50–51.

Wadley, L. (1995). Review of dated Stone Age sites recently exca-vated in the eastern Free State, South Africa. South AfricanJournal of Science 91, 574–579.

Wagner, G. A. (1996). Fission track dating in paleoanthropology.Evolutionary Anthropology 5, 165–171.

Wandibba, S. (1980). The application of attribute analysis of thestudy of Later Stone Age/Neolithic ceramics in Kenya. In (R. E.Leakey & B. A. Ogot, Eds) Proceedings of the Eighth PanafricanCongress of Prehistory and Quaternary Studies, Nairobi, 1977.Nairobi: International Louis Leakey Memorial Institute forAfrican Prehistory, p. 283–285.

Watson, E., Forster, P., Richards, M. & Bandelt, H.-J. (1997).Mitochondrial footprints of human expansions in Africa.American Journal of Human Genetics 61, 691–704.

Wiessner, P. (1982). Risk, reciprocity and social influences on !KungSan economics. In (E. Leacock & R. Lee, Eds) Politics and Historyin Band Societies. Cambridge & Paris: Cambridge University Pressand Editions de la Maison des Sciences de l’Homme, pp. 61–84.

Wiessner, P. (1983). Style and social information in Kalahari Sanprojectile points. American Antiquity 48, 253–276.

Wiessner, P. (1986). !Kung San networks in a generational perspec-tive. In (M. Biesele, R. Gordon & R. Lee, Eds) The Past andFuture of !Kung San Ethnography. Quellen zur Khoisan-Forschung, Band 4. Hamburg: Helmut Buske Verlag, pp. 103–136.

Yellen, J. E., Brooks, A. S., Cornelissen, E., Mehlman, M. &Stewart, K. (1995). A Middle Stone Age worked bone industryfrom Katanda, upper Semliki Valley, Zaire. Science 268, 553–556.