14

2007 Jefferey Rose

  • Upload
    rog

  • View
    215

  • Download
    0

Embed Size (px)

Citation preview

Page 1: 2007 Jefferey Rose
Page 2: 2007 Jefferey Rose

From the Mediterranean basin to the Portuguese Atlantic shore: Papers in Honor of Anthony Marks – Actas do IV Congresso de Arqueologia Peninsular

56

Page 3: 2007 Jefferey Rose

57

The Role of the Saharo-Arabian Arid Belt in the Modern Human Expansion

The Role of the Saharo-Arabian Arid Beltin the Modern Human Expansion

Jeffrey I. Rose

Southern Methodist UniversityDallas, TX, USA

ABSTRACTSome new lithic assemblages from Northeast Africa and South Arabia appear Upper Pleistocene and related to sub-Saharan Africa based ontechno-typological characteristics. These assemblages all contain thin bifacial foliates and exhibit a mix of façonnage , discoidal, and Levalloisreduction. It is argued that these sites are intrusive from East Africa, showing affinities to Middle Stone Age industries there. If there was anEast African MSA techno-complex extended northward and eastward into the Saharo-Arabian Arid Belt, this may provide direct archaeologicalevidence for an early human expansion out of sub-Saharan Africa sometime during OIS 5 or OIS 3.

KEYWORDSArabian Peninsula; Northeast Africa; Lower Paleolithic; Modern Humans

INTRODUCTION

J. Desmond Clark argued that palaeoenvironmental con-ditions in the Sahara – oscillating between ameliorated savan-nas and desiccated deserts – acted as a pump in dissemina-ting populations to and from sub-Saharan Africa throughoutthe Pleistocene (Clark, 1989). This idea is particularly intrigu-ing in light of recent MtDNA and Y-Chromosome genetic datathat suggest all modern humans are derived from an ances-tral population in sub-Saharan Africa dating to the Penulti-mate Glaciation (~200,000-128,000 BP). It is posited earlyhuman groups subsequently expanded out of sub-SaharanAfrica in one or more bottleneck release(s) that occurredaround 128,000 and 60,000 BP, durat the onset of OIs 5 and/or OIS 3 (Ambrose, 2003). These expansions correlate withthese wet-phases that transformed the Saharo-Arabian AridBelt into wooded and open savannas dotted by playa lakesand perennial streams, an environment that housed signifi-cantly increased biomass. In this sense, the Arid Belt servedas a physical cork sealing early human populations withinsub-Saharan Africa, a cork that blocked human movementthrough the Sahara and Arabia during hyperarid periods, andfacilitated their expansion during pluvials.

It is reasonable to assume that if any population expandedfrom East Africa into Northeast Africa and/or Arabia during awet-phase, they would have brought with them lithic technol-ogy from whence they came. To date, however, there has not

been any convincing archaeological evidence in northeastAfrica to suggest inter-regional affinities during the MiddleStone Age (henceforth MSA) between East Africa and thecontiguous Arid Belt. On the contrary, Middle Palaeolithic(henceforth MP) industries of the Sudan (e.g., Marks 1968a,1968b) are technologically and typologically distinct fromthose found in East Africa (e.g., Breuil et al., 1951; Wendorfand Schild, 1974; Merrick, 1975; Kurashina, 1978; Gresham,1984).

Furthermore, comparative analyses of Northeast Africanand Levantine MP assemblages – along a possible route outof Africa termed “The Levantine Corridor” – suggest therewere no compelling technological connections between thesetwo regions (e.g., Marks, 1990; Van Peer, 1998). Indeed, exa-mination of lithic technological trajectories in the Levant dem-onstrates a continuous lineage from the late Lower Palaeo-lithic to the Upper Palaeolithic (Monigal, 2002), implying therewas no significant influx of African technology (and pre-sumably associated human population) into the Levant dur-ing the Upper Pleistocene. This is not surprising, since a re-cent mtDNA study demonstrates that haplogroup M appearsin high frequencies among populations in East Africa, SouthArabia, India, and Southeast Asia, and is notably absentamong Levantine populations. Geneticists posit the initialbranching of haplogroup M took place in East Africa some60,000 years ago (Quintana-Murci et al., 1999).

So, archaeological and genetic data suggest modern hu-

Page 4: 2007 Jefferey Rose

From the Mediterranean basin to the Portuguese Atlantic shore: Papers in Honor of Anthony Marks – Actas do IV Congresso de Arqueologia Peninsular

58

mans expanding from East Africa did not enter the LevantineCorridor (in any archaeologically or genetically visible sense)during the Upper Pleistocene, as is often assumed. ThemtDNA study of haplogroup M indicates that the ArabianCorridor, which bridges Africa and Asia via South Arabia, mayhave served as the primary route of dispersal out of Africaduring the Penultimate Glaciation (Figure 1).

Archaeological evidence will be presented in this paperfrom Station One in northern Sudan and Bir Khasfa and JibalArdif 3 in central Oman – sites that demonstrate technologi-cal affinities with the East African MSA. It is proposed thathunter-gatherers moved into the Saharo-Arabian Arid Beltduring an early Upper Pleistocene pluvial event(s), expand-ing into this favorable, unoccupied niche. In most places, theArid Belt offered vast ameliorated savannas that were exten-sions of the East African phytogeographic zone (Patzelt, per-sonal communication). Thus, expansion into this niche did notrequire any major change in subsistence strategy.

The assemblages analyzed in this paper may belong toa MSA techno-complex that spread from East Africa (Ethio-pia, Kenya, Somalia, Djibouti, and Eritrea) into contiguousportions of the Saharo-Arabian Arid Belt to the north (Sudan)and east (Yemen and Oman). This expansion provides directarchaeological data in support of an early human migrationout of sub-Saharan Africa.

THE CORE AREA: EAST AFRICAN MSA

Prior to an examination of Station One, Bir Khasfa, andJibal Ardif 3, it is necessary to review lithic technologies ofthe East African MSA – the purported ancestral modern hu-man population. These industries are characterized by a suiteof reduction strategies that employ both core and façonnagetechnologies (these are not mutually exclusive). Façonnagerepresents a fundamentally different approach to producingtools from core reduction; the final tool is achieved via inva-sive flaking across one or both faces of a plaquette, blank,or preform. It is an important distinction to make, as there isnot one example of façonnage reduction in the Near East fromthe late Lower Palaeolithic through the Upper Palaeolithic, se-riously calling into question movement(s) from sub-SaharanAfrica through the Levantine Corridor during this time frame.

The production of diminutive bifacial foliates with flat, in-vasive retouch is a common thread throughout East Africa(Figure 2: d-g, j-k). These tools are considered by the authora fossile directeur of the region during the early Upper Pleis-tocene, and are best represented at the following sites: Mid-hishi 2 Cave in an escarpment along the northern coast ofSomalia, with radiocarbon and thermoluminescence datesplacing the basal layer sometime greater than 42,000 BP(Brandt and Brook, 1984; Gresham, 1984), the Gademotta

and Kulkuletti crater complexes in the Central Rift of Ethio-pia, yielding potassium-argon dates between 225,000 and150,000 BP; Porc Epic Cave in the southern Afar Rift of Ethio-pia, with the controversial obsidian hydration dating techniqueindicating 80,000-60,000 BP (Clark and Williamson, 1984);K’One 5, an undated findspot within the Garibaldi crater in thesouthern Afar Rift of Ethiopia (Kurashina, 1978); and ProspectFarm in the Nakuru Basin of northern Kenya, also with obsid-ian hydration dates ranging between 120,000 and 40,000 BP(Merrick, 1975; Anthony, 1978; Michels et al., 1983).

There is a range of variability among core reduction strat-egies within East Africa, including single-platform, convergent,Levallois point cores as well as simple unidirectional bladecores. Kurashina (1978) classifies one core type at K’One 5as Nubian Mousterian, Type I (though Nubian Mousterianassemblages in the Sudan and Egypt have no bifacial com-ponent). Kurashina (ibid. ) and others find thee cores fromK’One 5 as convincing evidence for north-south connectionsacross the Sahara at the same point in the MSA. Centripetalcore exploitation, falling within a continuum that includes dis-coidal and Levallois “tortoise” cores, are also an importantelement of MSA reduction strategies, particularly at Gade-motta, Kulkuletti, and Prospect Farm.

This disparity between single-platform and radial reduc-tion within East African MSA core reduction strategies hasbeen attributed to the nature of the locally available raw ma-terial (Merrick, 1975; Clark, 1980). Artifacts from Gademotta,Kulkuletti, and Prospect Farm are predominantly manufac-tured from obsidian cobbles, which are not particularly con-ducive to Levallois point production; thus, there is a greateremphasis on radial reduction (though point cores are present).The lithic assemblages from Midhishi 2 and Porc Epic aremade on tabular chert, while K’One raw material consistsprimarily of large blocks quarried from a local obsidian flow.Because of the large, angular nature of the available rawmaterial, these latter three sites exhibit a greater frequencyof Levallois points, as well as simple unidirectional blades.

INTO THE SAHARA: STATION ONE

Station One was discovered by A. Marks during the 1964season of the Combined Prehistoric Expedition to Nubia. Thesite is an open-air occurrence of chipped stone material ap-proximately 30 km east of the Second Cataract, in the East-ern Desert of Sudan (Figure 1). The lithic artifacts, primarilymanufactured from quartz pebbles, are scattered atop an in-selberg that is capped by a bed of workable ferrocrete sand-stone. The inselberg stands about 20 m above the pre-Niloticpeneplain; it is the only relief on the immediate landscape.Because the assemblage was outside the reservoir floodplainand had no obvious relationship with materials found within

Page 5: 2007 Jefferey Rose

59

The Role of the Saharo-Arabian Arid Belt in the Modern Human Expansion

the floodplain it was not included in the final report, and onlyrecently published (Rose, in press). The site is named for itsproximity to the first stop on the Sudanese railway travelingsouth from Wadi Halfa.

Station One was initially recognized by the sharp contrastof white quartz debris littering the deflated ferrocrete sand-stone surface. The assemblage consists of 1,939 pieces sys-temically collected from two distinct loci atop the inselberg. A10 m2 unit was established in both loci; all material withinthese arbitrary units, including tools, cores, debitage, and de-bris, was collected. Lithic analysis indicates there are no tech-nological or typological differences between the two scatters.

The Lithic Assemblage

The material from Station One is in relatively good con-dition (considering the brittle nature of quartz), although edgesand arêtes are slightly rounded from wind abrasion. Half theartifacts are complete, and among the broken pieces thereis a high frequency of false burins (c.f. Brézillon, 1971), likelydue to the numerous internal fractures within the availablequartz pebbles.

The assemblage was sorted into four categories: debitage(n = 1102), debris (n = 551), cores (n = 76), and tools (n = 210).All tools and cores were examined, and a random sample of25 per cent of the debitage was selected for attribute analy-sis (n = 280). Quartz comprises nearly the entire assembla-ge (91.45%), followed, in low percentages, by silicified wood(2.74%), ferrocrete sandstone (2.56%), Nile pebble (2.22%),metamorphic rock (0.68%), and quartz crystal (0.34%).

The Station One inselberg is capped by a bed of flake-able ferrocrete sandstone, ranging between 10 and 50 cmin thickness. Despite its quality and proximity, ferrocrete sand-stone represents a small portion of the total assemblage.This patterning is anomalous when compared with most otherMP/MSA sites in Nubia, which are dominated by tools manu-factured almost exclusively on ferrocrete sandstone (Soleckiet al., 1963; Guichard and Guichard, 1965; Marks, 1968a;Marks 1968b; Guichard and Guichard, 1968; Irwin et al.,1968). Rather than exploiting the immediately available (andperfectly suitable) raw material, inhabitants of Station Onechose to procure small quartz pebbles (between 23 and 63mm in maximum dimension) that are riddled with fractureplanes. Quartz gravels and found at the base of the insel-berg.

With the exception of Sai Island (Van Peer et al., 2003),MSA quartz industries are unprecedented in Nubia; thoughanalogies can be drawn with sub-Saharan MSA sites in Cen-tral and East Africa. The use of quartz at Station One repre-sents a deliberate choice not to exploit the most immediateraw material – flakeable ferrocrete sandstone occurring inlarge slabs – but rather to procure small, rounded quartz peb-

bles, the dimensions and quality of which are less conduciveto the manufacture of bifacial tools. Possible explanations forthis trend are that the toolmakers were not familiar with thehigh-quality chert ubiquitous within the Nile floodplain, did nothave access to resources within the Nile Valley, and/or weresimply more comfortable utilizing quartz. The well-madebifacial foliates made from these pebbles are a testament tothe skill of the knappers and/or familiarity with this challeng-ing raw material.

The closest source of Nile pebble would have been 30km to the west, within the Nile Valley. This fine-grained chertis relatively free of inclusions and excellent for knapping. Thenodules are heavily rolled and of similar size to the quartzpebbles, with the maximum dimension ranging between 29and 60 mm.

Known sources of silicified wood include the EasternDesert just behind Dibeira East, the Western Desert behindBuhen (Marks, 1968a: 199), and in the Batn al-Hajar (Soleckiet al., 1963). De Heinzelin and Paepe (1965) describe silicifiedwood as ubiquitous throughout the Cambrian sandstone for-mations outside the Nile Valley. This would imply a plethoraof sources in proximity to Station One. There are four coresof this variety from the site, ranging between 46 and 67 mmin length.

Technology

About half the cores are “informal”, which includes thefollowing types: single-platform (29.7%), multiple-platform(5.5%), and 90-degree cores (14.3%). Bidirectional cores arepresent, though low in frequency (7.7%). The Levallois cores(23.1%) all have centripetal preparation on the working face,while an additional number are radial (19.9%). It is most likelythat these last two core types are part of the same continuumof parametal exploitation of the raw material.

Unifacial blanks include flakes (70.5%), primary flakes(10.9%), Levallois blanks (8.1%), éclats de taille (7.6%), bla-de/bladelets (2.5%), and primary blades (0.2%). Éclats detaille are defined by a combination of attributes such as lon-gitudinal curvature; bidirectional, three-directional, or radialscar patterns; a modified striking platform; and/or thin di-mensions of the piece.

In general, the blanks from Station One are trapezoidaland ovoid in shape, only 11% are pointed and less than 3%are elongated. These traits are not surprising given the ten-dency toward centripetal core reduction. About 55% of theplatforms are unfaceted, 35% display some degree of plat-form manipulation, and the remaining 10% are missing dueto breakage and the tendency of quartz platforms to crushupon impact. Among the debitage, 63% of the dorsal scarpatterns are unidirectional/unidirectional-crossed, 19.6% areradial, 9.1% converging, and 8.3% bidirectional.

The technology observed at Station One differs from other

Page 6: 2007 Jefferey Rose

From the Mediterranean basin to the Portuguese Atlantic shore: Papers in Honor of Anthony Marks – Actas do IV Congresso de Arqueologia Peninsular

60

MP industries in Northeast Africa; Type 1 and Type 2 NubianMousterian cores are predominantly unidirectional and/or bi-directional with converging Levallois point preparation (Marks,1968). Khormusan Levallois technologies, with new U/Thmeasurements on ostrich eggshell and wood redating theindustry to OIS 4 (Marks et al., in preparation), have a corereduction strategy similar to Station One, with radial exploi-tation to maintain convexity on the working face of classicLevallois cores (Sellet, 1995). In contrast, there is no bifacialelement in the Khormusan, the Station One assemblage lacksthe numerous burins that are prevalent within Khormusantoolkits, and lacks the variability in raw material exploitationthat is characteristic of the Khormusan.

Typology

The tool assemblage is comprised primarily of typical,non-diagnostic MSA artifacts: retouched pieces (23.6%),scrapers (20.9%), notches (16.8%), and denticulates (15.5%).In addition to these ubiquitous types, bifacial (2.27%) andunifacial (1.4%) foliates/ovates with flat, invasive retouch arepresent (Figure 2: h-i), strongly suggesting affinities with theaforementioned MSA sites in East Africa.

Retouched pieces make up the largest tool category withinthe assemblage, with continuous irregular or marginal sec-ondary retouch. Almost all are on regular flakes. Within retou-ched pieces is a sub-category referred to as naturally-backedpieces. This type is noteworthy because it has been notedas a legitimate category at Prospect Farm. Anthony (1972:81) describes these pieces as: «a crescentic tool, not a truecrescent, that often bears on its curved back a small amountof marginal retouch. More often, the retouch is replaced by afalse “backing” resulting from the vertical trimming of theparent core. The flake, struck close to the trimmed edge of asmall core hardly larger than the resultant flake, picks up, soto speak, the edge and its trimming scars».

The four naturally-backed pieces at Station One are cres-cent to sub-crescent in shape, and possess invasive retouchon the working edge. The backing is not necessarily corti-cal; rather it is typically a fracture plane or remnant of pre-vious removals from the core perpendicular to the dorsalsurface. It is possible these blanks are a byproduct from thereduction of rounded quartz pebbles rather than a deliber-ate tool form.

Scrapers and denticulates are nearly as frequent as re-touched pieces, making up over a quarter of the overall toolkit.There is a low percentage of endscrapers, including nosed,ogival, and simple forms. Four specimens are bilateral con-verging scrapers, grading into retouched unifacial points.Retouch on these tools is predominantly observe and semi-steep.

The burins are simple and made from quartz flakes. Inone case, the burin edge is created by a single blow coming

from a truncated platform. Another burin is formed by a sin-gle blow on a natural fracture plane. The third, a dihedral bu-rin, has two spalls removed from a cortical platform.

The assemblage contains three unifacial points; one is ona standard flake, while the other two are on Levallois flakes.It should be noted these Levallois flakes were retouched intopoints, and are not derived from Levallois point production.Two points have obverse retouch, one is retouched inversely.Two points are triangular, and one is cordiform.

Bifacial pieces are distinguished by invasive retouch onboth faces. These tools include foliates/ovates, miscellane-ous bifacial elements, or fragmentary pieces that fall into acontinuum of bifacial reduction. Some of these miscellaneouselements exhibit few retouching blows, suggesting they maybe preforms, while others are extensively reduced, indicatingthey may be exhausted bifacial cores (c.f. Kelly, 1988). Thebifacial foliates/ovates are typically small and thin, with atransverse cross-section ranging from biconvex to slightlyplano-convex. They are all made on unifacial blanks, with flat,invasive retouch.

There are less than a handful of assemblages in North-east Africa that have bifacial tools resembling Station One,these sites include: Bir Tarfawi, Bir Sahara East, N2 and SaiIsland. The material at Bir Tarfawi and Bir Sahara East isexclusively manufactured from quartzitic sandstone. Theassemblages were found in the western desert of Egypt, insediments correlated with OIS 5 palaeolakes (Wendorf et al.,1993). At Sai Island, on the Nile south of the Second Cata-ract, Van Peer et al. (2003: 189), report a lithic assemblagemanufactured from quartz whose “most prominent typologi-cal feature is the presence of thin bifacial foliates… blankswere produced according to Levallois, Nubian, and discoidalreduction strategies.” The assemblage was found in situ withinblack Nilotic silt; excavators correlate these sediments withthe OIS 5 palaeolakes at Bir Tarfawi and Bir Sahara (Van Peeret al. , 2003). Site N2 was discovered within the Goshabi For-mation, the oldest fluviatile aggradation recognized in theDongola Reach. The assemblage is characterized by a roughLevallois core technology with a low index of faceting and thepresence of bifacial reduction. It is interesting to note that,while most artifacts at N2 are manufactured from wadi chertand quartzite, several foliates are made on large quartz peb-bles (Marks et al., 1971).

The location of Station One on an inselberg overlookingthe pre-Nilotic peneplain, combined with the presence of low-mass, high-velocity bifacial armatures, suggests the inhabit-ants of the site employed a subsistence strategy tied to largegame moving throughout the savanna. Lézine (1989) des-cribes a 500 km northward shift in Sahara vegetation zonesduring an early Holocene wet phase, which brought 200-500mm of rainfall per annum and transformed the plains outsidethe Nile Valley into a grassland environment. It is conceivable

Page 7: 2007 Jefferey Rose

61

The Role of the Saharo-Arabian Arid Belt in the Modern Human Expansion

Upper Pleistocene humid episodes resulted in a similar phy-togeographic reconfiguration. Large ungulates would havemoved into this ameliorated niche, bringing with them hunter-gatherers exploiting these resources.

INTO ARABIA: BIR KHASFA AND JIBAL ARDIF 3

Bir Khasfa is a surface scatter of chipped stone artifacts(>100,000 m2) on the Nejd Plateau in Oman (Figure 1). Thesite was first identified and collected by the Harvard Archaeo-logical Survey in Oman in 1973 and 1975 (Pullar, 1974; Pullarand Jäckli, 1978; Pullar, 1985), and more recent study hasbeen carried out by the Central Oman Pleistocene Researchproject in 2002 and 2004 (Rose, 2004).

The site was initially recorded due to the density of arti-facts found at various elevations on a rock outcrop above theleft bank of the Wadi Arah, where the wide, low-energy chan-nel enters a small localized basin, bending from a north-eastward to a northward orientation. The site is named for anearby well (bir) that today serves Beduin tribes throughoutthe Nejd.

The local relief is formed by a crescent-shaped outcropof Tertiary rock with abundant chert outcropping from withinthe inner basin, at the margins of an ancient playa lake. Theraw material is a brown, high-quality chert found in angularslabs and thin plaquettes derived from an exposed Eocenebed named as the Rus Formation. The undulating surface ofthe Tertiary bed is covered by a mantle of aeolian sand rang-ing from 0 to 20 cm in depth, which rests upon a cementedpetrogypsic horizon. Artifacts are primarily found on the sur-face and embedded no greater than five cm within the sub-surface aeolian veneer.

Even today, the water level is quite high and there ap-pears to be occasional surface runoff in the basin below thesite, attested to by the large number of trees and bushes seenflourishing in the depressed areas. Two 1 x 1 m test pits exca-vated from within this basin during the 2004 campaign re-vealed an interstratified mix of aeolian and lacustrine sedi-ments, though no artifacts were identified within the burieddeposit.

At least three different geomorphic areas of the site wereobserved, each demonstrating different lithic technologies.The area just above the depression has an extensive lithicscatter with a large number of completed bifacial tools. Aplethora of single-platform, unidirectional blades and coreswere observed further away from the basin, at a higher ele-vation on the outcrop. A third industry, exhibiting diagnosticNeolithic artifacts, was identified on the top of a nearby in-selberg. Survey and collection during the 2002 and 2004campaigns focused exclusively on the bifacial scatter.

Two findspots were systematically collected; although the

scatters are approximately 150 m apart they do not neces-sarily represent specific reduction sites, rather distinct zonesof one large homogenous scatter separated by areas of lowdensity. They were chosen based on the frequency of com-pleted tools (in most cases bifacial foliates) observed on thesurface. It is clear from technological and typological obser-vations that both areas collected at Bir Khasfa are more orless coeval, and will be treated here as such.

A 2 x 5 m grid was sampled from Locus 1 using .5 x .5 munits. The second findspot is about 150 m to the southwest,just above the edge of the basin. At Locus 2, a 2 x 7 gridwas placed over the area, again employing .5 x .5 m collec-tion units. At both loci, artifacts were collected on the surfaceand within the shallow sandy-gypsum carpet.

The Lithic Assemblage

The total sampled area of 24 m2 yielded 318 artifacts, ofwhich 65.9% are debitage, 0.5% are cores, 20.9% are tools,and 12.6% are debris. The material is in nearly pristine con-dition – arêtes are sharp, indicating minimal wind abrasionand/or rolling. The artifacts are all manufactured on chert, andhave a medium patina, ranging in color from yellowish-orangeto brown. Middle Holocene diagnostic artifacts on top of thenearby inselberg exhibit a pinkish-white patina, providing aclear differentiation from the Pleistocene artifacts.

Technology

Blanks at Bir Khasfa, which are predominantly trapezoi-dal, rectangular, and ovoid in shape, appear in the followingfrequencies: flakes (44.2%), followed by a significant numberof eclat de taille (34.9%), cortical flakes and blades (15.5%),blade/lets (3.9%), and core trimming elements (1.6%). Moststriking platforms are either unfaceted or crushed, less than10% are dihedral or faceted. The presence/absence of lip-ping on the bulb of percussion, viewed as a potential indi-cator of soft hammer percussion, occurred on 23.6% of thepieces. Combined with the high frequency of crushed plat-forms (17.8%), it is likely hard hammer percussion was usedfor the initial roughing out of bifacial preforms. The lippingpercentage rises to 47.4% when considering only éclats detaille , suggesting the Bir Khasfa flint knappers switched tosoft hammer percussion for thinning.

Dorsal scar patterns are primarily unidirectional (40.7%),unidirectional-crossed (29.6%), radial (13.9%), transverse(7.4%), with the remaining percentage divided among paral-lel, converging, and crested. Just two cores were recoveredat Bir Khasfa, of which one is an unidentifiable fragment. Theother is a small radial core (~50 mm in maximum dimension),with alternating parametal exploitation across both faces ofthe piece. One face is flat, and may represent the exhaustedworking surface of a centripetal Levallois core.

Page 8: 2007 Jefferey Rose

From the Mediterranean basin to the Portuguese Atlantic shore: Papers in Honor of Anthony Marks – Actas do IV Congresso de Arqueologia Peninsular

62

Typology

The 64 tools identified in the Bir Khasfa assemblage, in-clude bifacial pieces (37.5%), preforms (26.6%), sidescrapers(23.4%), endscrapers (7.8%), unifacial points (3.1%), and re-touched pieces (1.6%). Of the 23 tools made from blanks,over a quarter are manufactured on éclats de taille.

Bifacial tools, by far the largest category at Bir Khasfa,are primarily leaf-shaped foliates, with a small percentage ofovate and cordiforms shapes (Figure 2: a-c). These diminu-tive pieces range between 30 and 70 mm in length, and be-tween 6 and 16 mm in thickness. Nearly 90% of the piecesare biconvex, and all have flat, invasive retouch. There ap-pears to be an even distribution between bifaces reduced fromflakes versus those reduced from plaquettes. In cases whereunifacial blanks were chosen for bifacial production, corticalpieces were almost always selected.

There were 17 preforms, representing the entire con-tinuum of bifacial reduction. There are a few examples of con-tinuous marginal retouch along one preform edge, which dem-onstrates that edge grinding was carried out to prepare theplatform for bifacial flaking. Of the 17 preforms, ten are bro-ken in half, probably a result of shock during the thinning.One particular preform was quite large (~150 mm in length),exceeding the maximum dimension of locally available rawmaterial. The size is reminiscent of preforms collected at JibalArdif 3, both in terms of techno-typological production andraw material selection, and potentially provides a link betweenthe two findspots as part of the same kind of reduction se-quence.

Of the 15 sidescrapers collected at Bir Khasfa, nearly halfare made on éclat de taille . The retouch ranges from mar-ginal to Quina, equally divided between inverse and obversefaces. There were five endscrapers, sensu latu. The pieceshave continuous retouch on one narrow edge of a flake orblade; four have straight retouched edges while one end-scraper has a convex working edge.

Two unifacially retouched points were noted. One is apartly-bifacial foliate made on a blade with invasive, retouchcovering the dorsal face and proximal portion of the ventralface, perhaps for bulbar thinning. The second retouched pointis on a flake with semi-steep, inverse, converging bilateralretouch.

JIBAL ARDIF 3

Approximately 100 km southwest of Bir Khasfa, also onthe Nejd Plateau, is a series of hills and inselbergs calledJabal Ardif. The jagged landscape is mantled by a more orless continuous 40-km-long reg comprised of nodules, pla-quettes, and slabs of chert. This Early Tertiary exposure ispart of the chert-bearing Rus Formation; at its base is a 3-

-5 m thick brecciated dolomitic limestone with fine-grainedbrown chert inclusions. The original chert/dolomite beddingis thoroughly modified by recrystallization and collapse fromdissolution of evaporites, giving the present landscape itscraggy relief (Platel et al., 1992).

The hills rising above the chert regs are composed ofwhite laminated, dolomitic, chalky limestone interstratified withdolomitic limestone breccia. This disparity in density leads touneven erosion of the rock, producing frequent small, lowrockshelters within the hills ( ibid.). Lithic scatters with apalimpsest of technologies are ubiquitous on the Jibal Ardifreg surfaces.

One such scatter, Jibal Ardif 3, is situated in the hills about30 km south of Thumrait and five kilometers east of WadiDawkhah (Figure 1). The scatter is located on a flat plain be-tween hills, on a landscape lined by a series of parallel lime-stone ridges spaced ten meters apart on average (rarely ex-ceeding 10 cm in height and 20 cm in width). The ridges areformed by interstratified beds of folded, truncated limestonestrata; subsequent erosional processes have had a greatereffect on the softer chalky limestones, leaving prominentridges comprised of the harder dolomitic stratum. The terrainis covered by a thin veneer of hard-packed gypsum evapo-rites, indicating poor drainage within the local basin.

The Lithic Assemblage

The site was initially identified by the high number ofbifacial preforms and éclats de taille scattered over a rela-tively small area. This apparent homogenous technology isunique among other scatters on the Jibal Ardif reg, whichtypically display mixed industries.

Artifacts were recovered on the surface and imbedded inthe evaporitic sediments. Patina ranges from a pinkish-grayto orangish-brown, depending on their position in the gypsumsurface veneer. Lithics embedded in the gypsum had muchlighter discoloration, while material exposed on the surfacepossessed a deeper patina. The material is in pristine condi-tion, there is no rounding from wind abrasion or rolling.

The sampled assemblage consists of 286 pieces collectedin an area of 12 m2, including 71.8% debitage, 18.1% debris,6.9% tools, and 3.2% cores. Every artifact is derived from thelocal fine-grained Eocene chert. Raw material is ubiquitousthroughout the surface around the site, occurring in large, thinplaquettes that average 150-250 mm in length and 15-30 mmin thickness. These plaquettes are particularly conducive tothe production of bifacial tools – the site is clearly a work-shop for the exploitation of this outcrop.

Technology

The blanks from Jibal Ardif 3 are primarily trapezoidaland ovoid in shape. Flakes make up 57.1% of the debitage,

Page 9: 2007 Jefferey Rose

63

The Role of the Saharo-Arabian Arid Belt in the Modern Human Expansion

followed by éclat de taille (27.1%), cortical flakes and blades(11.3%), and blade/lets (4.0%). Cortical striking platforms aremost frequent (28.5%), also with a significant number of dihe-dral and faceted platforms (17.5%), and straight platforms(15.5%). Only 8.5% of all blanks exhibit any sign of lipping,indicating there was little or no soft hammer percussion usedduring reduction. Further evidence for hard hammer percus-sion, a large (261 mm) chert hammerstone was collectedfrom the findspot. Dorsal scar patterns on the debitage areprimarily unidirectional (45.2%) and unidirectional-crossed(34.5%).

There were nine cores collected at Jabal Ardif 3. Becauseof the thin, rounded shape of the raw material, all of the coresare flat, non-volumetric. Most are radial cores, either with oneor two working surfaces. There are three cores that have elon-gated point and flake scars removed via an opposed platformtechnique. In all three cases, both of the platforms are fac-eted, though one end appears to be supplementary with non-invasive blows for maintaining convexity. The cores exhibit amuch deeper patina, therefore are thought to date to an ear-lier phase of occupations of the site.

Typology

There are no completed formal tools within the Jabal Ardif3 assemblage, though 17 bifacial preforms were identified(Figure 2: l-m). They range in length from 60 to 200 mm, andbetween 12 and 28 mm in thickness. In every case, retouchquality is flat and invasive. The preforms exhibit several pha-ses of reduction, from initial plaquette thinning to nearly com-pleted tools. Though unfinished, they are primarily ovates andcordiforms, more a function of plaquette shape than a con-scious choice of the knapper. Except for the early stage pre-forms, the bifaces are all biconvex in profile. Like Bir Khasfa,there was a mid-stage preform exhibiting continuous marginalretouch across one lateral edge (Figure 2: l), indicating thatedge grinding was carried out to prepare for invasive bifacialflaking.

Nearly half of the preforms are broken in half, of whichsix have been refit; in every case they show that a final hea-vy blow snapped the preforms during thinning, which is notsurprising given the use of a hard chert hammerstone in re-duction.

When one considers the high ratio of preforms to totalnumber of pieces in the assemblage (1 in 6), combined withthe frequency of cortical platforms, cortical flakes, unidirec-tional scar patterns, and hard hammer percussion, it is clearthe site functioned as a workshop for initial plaquette reduc-tion. In every case, preforms were derived from plaquettes;unifacial blanks were not used for bifacial production, againa function of raw material availability rather than consciouschoice.

DISCUSSION

Prior to 1972, the Stillbay Culture was recognized through-out all of sub-Saharan Africa as an MSA entity with Levalloisand discoidal cores, as well as flat bifacial foliates. Anthony(1972: 82) reevaluated this entity in relation to East Africa andconcluded “the Stillbay Culture does, in fact, not exist”, basedon its wide geographic distribution and variable technologiesobserved among the assemblages. This position was adoptedby archaeologists working in Africa, and the subject has notbeen broached since.

Anthony was correct in pointing out the Stillbay classifi-cation is too broad; however, the baby was thrown out withthe bathwater. For the past 30 years, the term East AfricanMSA has been used only as a temporal designation; schol-ars have neither defined the techno-typological attributes norarticulated the geographic distribution of this archaeologicalcomplex.

Data has been presented from sites in East Africa, north-ern Sudan, and central Oman that suggest there are affini-ties among these three regions, most prominently the produc-tion of bifacial foliates. These tools are manufactured eitherfrom unifacial blanks or plaquettes, depending upon the na-ture of the locally available raw material. Bifacial thinning isalways carried via flat, invasive retouch to achieve a foliate,ovate, or cordiform shaped-tool with a biconvex profile.

In addition to the production of bifacial foliates, there isa variety of core technologies identified at these sites, inclu-ding discoids, and centripetal Levallois cores, while conver-gent levallois – the hallmark of the Levantine Mousterian –are absent.

This techno-typological package is geographically con-scripted; there is no façonnage technology to the north inthe lower Nile Valley, Levant, or Zagros Mountains, and westof the Rift Valley the Sangoan and Lupemban industries aretypologically distinct with larger and more elongated bifacialtools.

These MSA sites probably date to OIS 5 or OIS 3; a cli-matic episode in which the Saharo-Arabian Arid Belt was peri-odically transformed into open and wooded savannas thatwere extensions of the East African phytogeographic zone.Therefore, the sites fall within a continuous, homogenous re-gion that would have facilitated population expansion and con-traction during cyclical phases of amelioration and desiccation.

With this in mind, it is suggested there was a regionaltechno-complex that spanned East Africa, the eastern Sahara,and South Arabia during the last interglacial. The assembla-ges fall within a continuous phytogeographic niche and arelinked by the presence of a fossile directeur.

The proposed techno-complex originated in East Africa andexpanded into the Saharo-Arabian Arid Belt at the onset of plu-vial conditions some 128,000 or 60,000 years ago. This asser-

Page 10: 2007 Jefferey Rose

From the Mediterranean basin to the Portuguese Atlantic shore: Papers in Honor of Anthony Marks – Actas do IV Congresso de Arqueologia Peninsular

64

tion rests upon the two assumptions: 1) the sites examined inthis paper date to the Upper Pleistocene, and 2) lithic indus-tries that developed in the Arid Belt must derive from neigh-boring refugia because there was little or no human occupa-tion within the Saharo-Arabian Arid Belt during the PenultimateGlaciation due to hyperarid conditions. Further work must becarried out throughout East Africa and the Arid Belt to testthese assumptions; until then this proposition is conjecture.

ACKNOWLEDGEMENTS

First and foremost, I thank Tony Marks for allowing me to studythe Station One assemblage, as well as his encouragement and sup-port in preparing this paper. I am grateful to Nuno Bicho for invitingme to present at this symposium, despite the fact that the penin-sula discussed in this paper is thousands of kilometers from Iberia.My thanks to Lucy Addington for her illustrations of the Station Onematerial, and Vitaly Usik for his drawings from Bir Khasfa and JabalArdif 3. Finally, I would like to acknowledge the 2004 Central OmanPleistocene Research team, including Vitaly Usik, Diego Angelucci,Teresa Medici, Daniel Richter, Ali al-Mahrooqi, and Saeed al-Suqri.

WORKS CITED

AMBROSE, S.2003. Population bottleneck. In R. Robinson (Ed.), Genetics, Volu-me 3, pp. 167-171. New York: Macmillan Reference.

ANTHONY, B.1972. The Stillbay question. Actes du VIe Session du Congrès Pana-fricain de Préhistoire, Dakar, 1967. Chambéry: Les Imprimeries Réu-nis de Chaméry.

BRANDT, S. & BROOK, G. A.1984. Recent archaeological and palaeoenvironmental research inNorthern Somalia. Current Anthropology , 25: 119-121.

BREUIL, H.; TEILHARD DE CHARDIN, P. & WERNERT, P.1951. Le paléolithiques du Harar. L’Anthropologie, 55: 219-230.

BRÉZILLON, M.1971. La Dénomination des Objets de Pierre Taillée. Paris: CentreNational de la Recherche Scientifique.

CLARK, J. D.1980. Human populations and cultural adaptations in the Sahara andNile during prehistoric times. In H. Faure and M. Williams (Eds.), TheSahara and the Nile – Quaternary environments and prehistoric oc-cupation in northern Africa, pp. 527-582. Rotterdam: A. A. Balkema.1989. The origin and spread of modern humans: a broad perspec-tive on the African evidence. In P. Mellars and C. Stringer (Eds.),The Human Revolution: Behavioural and Biological Perspectives onthe Origins of Modern Humans, pp. 565-588. Edinburgh: EdinburghUniversity Press.

CLARK, J. D. & WILLIAMSON, K. W.1984. A Middle Stone Age occupation site at Porc Epic cave, DiraDawa (east-central Ethiopia), Part I. African Archaeological Review ,2: 37-71.

GRESHAM, T.1984. An investigation of an Upper Pleistocene archaeological sitein northern Somalia. M.A. Thesis. University of Georgia, Athens.

GUICHARD, J. & GUICHARD, G.1965. The Early and Middle Paleolithic of Nubia: a preliminary re-port. In F. Wendorf (Ed.), Contributions to the Prehistory of Nubia,pp. 57-166. Dallas: Fort Burgwin Research Center and SouthernMethodist University Press.1968. Contributions to the study of the Early and Middle Paleolithicof Nubia. In F. Wendorf (Ed.), The Prehistory of Nubia, Volume 1,pp. 148-193. Dallas: Fort Burgwin Research Center and SouthernMethodist University Press.

DE HEINZELIN DE BRAUCOURT, J. & PAEPE, R.1965. The geological history of the Nile Valley in Sudanese Nubia:preliminary results. In F. Wendorf (Ed.), Contributions to the Prehis-tory of Nubia , pp. 29-56. Dallas: Fort Burgwin Research Center andSouthern Methodist University Press.

IRWIN, H. T.; WHEAT, J. B. & IRWIN, L. F.1968. University of Colorado investigations of Paleolithic sites in theSudan, Africa . University of Utah Anthropological Papers, 90. SaltLake City: University of Utah Press.

KELLY, R. L.1988. The three sides of a biface. American Antiquity , 63: 717-734.

KURASHINA, H.1978. An examination of prehistoric lithic technology in east- cen-tral Ethiopia. Ph.D. Dissertation. University of California, Berkeley.

LÉZINE, A.-M.1989. Late Quaternary vegetation and climate of the Sahel. Quater-nary Research, 32: 317-334.

MARKS, A. E.1968. The Mousterian industries of Nubia. In F. Wendorf (Ed.), ThePrehistory of Nubia, Volume 1, pp. 194-314. Dallas: Fort BurgwinResearch Center and Southern Methodist University Press.

MARKS, A. E.; SHINER, J. L.; SERVELLO, F. & MUNDAY, F.1971. Flake assemblages with Levallois techniques from the DongolaReach. In J. L. Shiner (Ed.), The Prehistory and Geology of North-ern Sudan, Part 1. Report to the Nation Science Foundation, GrantGS 1192.

MARKS, A. E.1990. The Middle and Upper Palaeolithic of the Near East and theNile Valley: the problem of cultural transformations. In P. Mellars(Ed.), The Emergence of Modern Humans: An Archaeological Per-spective, pp. 56-80. Edinburgh: Edinburgh University Press.

MARKS, A. E.; MCKINNEY, C. & ROSE, J. I.In preparation. Uranium Series dating of wood, enamel, and ostricheggshell from the Khormusan.

MERRICK, H. V.1975. Change in later Pleistocene lithic industries in eastern Africa.Ph.D. Dissertation. University of California, Berkeley.

MONIGAL, K.2002. The Levantine leptolithic: blade technology from the LowerPalaeolithic to the dawn of the Upper Palaeolithic. Ph.D. Disserta-tion. Southern Methodist University.

PLATEL, J. P.; ROGER, J.; PETERS, T. J.; MERCOLLI, I.;KRAMERS, J. D. & LE MÉTOUR, J.1992. Geological Map of Salalah, Explanatory Notes. Muscat: Minis-try of Petroleum and Minerals.

Page 11: 2007 Jefferey Rose

65

The Role of the Saharo-Arabian Arid Belt in the Modern Human Expansion

PULLAR, J.1974. Harvard archaeological survey in Oman, 1973: flint sites inOman. Arabian Seminar , 4: 33-48.1985. A selection of aceramic sites in the Sultanate of Oman. Jour-nal of Oman Studies, 7: 49-88.

PULLAR, J. & JÄCKLI, B.1978. Some aceramic sites in Oman. Journal of Oman Studies, 4:53-71.

QUINTANA-MURCI, L.; SEMINO, O.; BANDELT, H.;PASSARINO, G.; MCELREAVEY, K. & SILVANASANTACHIARA-BENERECETTI, A.1999. Genetic evidence of an early exit of Homo Sapiens Sapiensfrom Africa through eastern Africa. Nature Genetics, 23: 437-441.

ROSE, J. I.2004. The Question of Upper Pleistocene Connections between EastAfrica and South Arabia. Current Anthropology , 45 (4): 551-555.In press. New evidence for the expansion of an Upper Pleistocenepopulation out of East Africa, from the site of Station One, NorthernSudan. Cambridge Archaeological Journal, 14 (2), in press.

SELLET, F.1995. Levallois or not Levallois: does it really matter? Learning froman African case. In H. L. Dibble and O. Bar-Yosef (Eds.), The Defi-

nition and Interpretation of Levallois Variability, pp. 279-292. Madi-son: Prehistory Press.

SOLECKI, R. S.; DE HEINZELIN, J.; STIGLER, R. L.;MARKS, A. E.; PAEPE, R. & GUICHARD, J.1963. Preliminary statement of the prehistoric investigations of theColumbia University Nubian expedition in Sudan, 1961-62. Kush, 11:70-92.

VAN PEER, P.1998. The Nile corridor and the out-of-Africa model: an examinationof the archaeological record. Current Anthropology, 39: 115-131.

VAN PEER, P.; FULLAGAR, R.; STOKES, S.; BAILEY, R. M.;MOEYERSONS, J.; STEENHOUDT, F.; GEERTS, A.;VANDERBEKEN, T.; DE DAPPER, M. & GEUS, F.2003. The Early to Middle Stone Age Transition and the Emergenceof Modern Human Behaviour at site 8-B-11, Sai Island, Sudan. Jour-nal of Human Evolution, 45: 187-93.

WENDORF, F. & SCHILD, R.1974. A Middle Stone Age Sequence from the Central Rift Valley,Ethiopia. Warsaw: Polska Akademia Nauk.

WENDORF, F.; SCHILD, R. & CLOSE, A. E.1993. Egypt during the Last Interglacial: the Middle Paleolithic of BirTarfawi and Bir Sahara East. New York: Plenum Press.

Page 12: 2007 Jefferey Rose

From the Mediterranean basin to the Portuguese Atlantic shore: Papers in Honor of Anthony Marks – Actas do IV Congresso de Arqueologia Peninsular

66

FIGURE 1. Map showing potential routes of expansion out of Africa during the Upper Pleistocene, and sitesmentioned in text that demonstrate the East African MSA techno-typological package.

Page 13: 2007 Jefferey Rose

67

The Role of the Saharo-Arabian Arid Belt in the Modern Human Expansion

FIGURE 2. Bifacial foliates and preforms: a-c – Bir Khasfa, Oman; d – Midhishi 2, Somalia (after Gresham, 1984: Figure 4.4);e-g – Porc Epic, Ethiopia (after Clark and Williamson, 1984: Figure 7); h-i – Station One, Sudan; j-k – Gademotta and Kulkuletti,

Ethiopia (after Wendorf and Schild, 1974: Plates 20, 51); l-m – Jabal Ardif 3, Oman.

Page 14: 2007 Jefferey Rose