First technological comparison of Southern African Howiesons Poort and South Asian Microlithic...

lable at ScienceDirect

Quaternary International xxx (2014) 1e19

Contents lists avai

Quaternary International

journal homepage: www.elsevier .com/locate/quaint

First technological comparison of Southern African HowiesonsPoort and South Asian Microlithic industries: An explorationof inter-regional variability in microlithic assemblages

Laura Lewis a, *, Nimal Perera b, Michael Petraglia a

a School of Archaeology, University of Oxford, Oxford OX1 2PG, UKb Postgraduate Institute of Archaeology, 407 Bauddhaloka Mawatha, Colombo 7, Sri Lanka

a r t i c l e i n f o

Article history:Available online xxx

Keywords:MicrolithicHowiesons PoortLithic technology

* Corresponding author.E-mail address: [email protected] (L. Lewi

http://dx.doi.org/10.1016/j.quaint.2014.09.0131040-6182/© 2014 Elsevier Ltd and INQUA. All rights

Please cite this article in press as: Lewis, LMicrolithic industries: An exploration ofdx.doi.org/10.1016/j.quaint.2014.09.013

a b s t r a c t

Here we conduct the first direct metric examination of two early regional manifestations of microlithicindustries e the Howiesons Poort of southern Africa (c. 65e60 ka) and the Microlithic industry of SouthAsia (c. 38e12 ka). Inter-regional comparative analysis of microlithic industries is rare, but can contributemuch to our understanding of technological systems in the past. Metric and qualitative variables wererecorded on cores, debitage, and tools from Rose Cottage Cave and Umhlatuzana, South Africa, andBatadomba-lena, Sri Lanka, with the aim of conducting a first-stage technological assessment of thedegree of technological homogeneity and diversity within these rich microlithic assemblages. The lithicmethodology employed here uses the full range of lithic by-products, as opposed to an approach basedon tool typology alone. Preliminary analyses reveal areas of significant variation in inter-regional tech-nological strategies. These include differences in blade production and blank selection, variation inmicrolith typology and morphology, disparate quartz reduction processes designed to produce similartool types, varying degrees of utilisation of bipolar technology, and the existence of distinct reductiontrajectories within sites. The examination of the diversity of microlithic assemblages through the use ofdetailed technological attribute analyses demonstrates a useful alternative methodology for the way weexamine behavioural variability, and is a first step towards a thorough assessment of the place of mi-croliths in models of human dispersals.

© 2014 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction

Microliths are small retouched stone tools, often backed andgenerally considered to have been hafted as part of composite tools,particularly projectile hunting weapons (Elston and Kuhn, 2002).Despite their diminutive size, they are a substantial topic of dis-cussion in stone tool studies. Numerous hypotheses have beenproposed to explain their appearance. Some emphasise their role inhunting, including the use of microliths in composite toolsdesigned for hunting particular prey species (McCall and Thomas,2012), and their possible use as vehicles for poison (Lombard andPargeter, 2008). Others focus on technological aspects of micro-lith production, such as their relationship with an increased use offine-grained raw materials (Neeley, 2002) and the benefits of using

s).

reserved.

., et al., First technologicalinter-regional variability in

composite tools with standardized replaceable parts in risky envi-ronments (Elston and Brantingham, 2002; Hiscock, 2002). Thepotential role of microliths in exchange networks and their statusas group or ethnic symbols have also been assessed (Deacon, 1992;Wurz, 1999; Ambrose, 2002; Close, 2002).

An accepted standard definition of microliths has failed tomaterialise despite attempts spanning several decades (e.g. Gloverand Lampert, 1969; Gould, 1969; Clark, 1985; Ballin, 2000; Elstonand Kuhn, 2002; Hiscock et al., 2011). Existing definitions oftenuse arbitrary size limits, but can also require the presence ofbacking retouch or of the use of geometric shapes (Leplongeon, inpress). The collection of definitions in one journal volume alone(Elston and Kuhn, 2002) reveals the large degree of variation be-tween geographic areas of study. The definition used in the currentstudy is designed to be general and inclusive, making use of toolsize profiles from individual sites rather than imposing arbitraryover-arching size limits, discussed further below. Microlithic in-dustries are defined here as those where the overall lithic reduction

comparison of Southern African Howiesons Poort and South Asianmicrolithic assemblages, Quaternary International (2014), http://

mailto:[email protected]

www.sciencedirect.com/science/journal/10406182

http://www.elsevier.com/locate/quaint

http://dx.doi.org/10.1016/j.quaint.2014.09.013



L. Lewis et al. / Quaternary International xxx (2014) 1e192

process can be shown to be oriented around the production ofblanks subsequently retouched into microlithic tools, as opposed toassemblages that happen to include a handful of tools that aresmaller than arbitrary size limits.

Inter-regional comparative studies of microlithic industries arerare, and are often conducted on a very broad scale (e.g.Deraniyagala, 1988, 1992; Mellars, 2006a; Hiscock et al., 2011). As aresult, these studies tend to treat microlithic industries as mono-lithic entities, ignoring the variety of different methods of pro-duction and use. This is particularly true in cases that placemicrolithic technologies in theoretical frameworks revolvingaround the evolution of modern human behaviour (McBrearty andBrooks, 2000) and human dispersals (Mellars, 2006a; Mellars et al.,2013).

Investigation and classification of different lithic industriesformed the foundation of Palaeolithic archaeology. Assessment ofthe diversity of microlithic technologies that were utilisedthroughout the later Pleistocene (and into the Holocene in manyareas) is therefore an important part of our understanding ofhunteregatherer lifeways. Exploration of choices made and stra-tegies used in stone tool manufacture can tell us about behaviourand adaptations to social and environmental circumstances.Changes in lithic technology, typology and raw materials over timeare a valuable record of how humans adapted to environmental anddemographic changes. Although typological assessments ofmicrolithic assemblages may reveal superficial similarities, onlydetailed technological analyses can comprehensively address theseissues.

To this end, the technological systems of two early microlithicindustries e the Howiesons Poort (HP) of southern Africa and theMicrolithic industry of South Asia e will be compared here. Theseindustries are cited explicitly in Mellars' (2006a, Mellars et al.,2013) microlithic dispersal hypothesis, one of the key argumentsinvoking comparisons between microlithic industries. The HPrepresents the earliest truly microlithic industry in the world, i.e.with lithic reduction organised around the production of micro-lithic tools. The Microlithic industry of South Asia provides avaluable counterpoint to the HP. Whereas the HP is notable forbeing so short-lived (Jacobs et al., 2008, cf. Gu�erin et al., 2013;Tribolo et al., 2013), the South Asian Microlithic overlaps a periodof considerable demographic and environmental changes(Deraniyagala, 1992; Atkinson et al., 2008; Clarkson et al., 2009;Petraglia et al., 2009; Perera, 2010; Perera et al., 2011) whileremaining dominated by microliths throughout.

Here we first discuss previous attempts to compare microlithicindustries between the two regions. We then lay out our methodsfor dealing with such comparisons, and provide an introduction tothe HP of southern Africa and theMicrolithic industry of South Asia,with particular reference to potential areas of variability. The re-sults of our analyses from each site will then be presented, beforebeing compared directly for the first time. Finally, we will use theseresults to highlight areas of potential diversity within and betweenmicrolithic industries, which demonstrate that microlithic pro-duction is a complex and variable technology.

2. Previous comparative work on microlithic industries

Previous inter-regional comparative work on microlithic tech-nologies is limited, and is no doubt exacerbated by the considerabledifferences evident in definitions between regions (e.g. Elston andKuhn, 2002). This is particularly problematic given that existingcomparative reviews of microlithic technology between multipleregions almost always make use of secondary data, rather thancollecting new data from different areas using the same method-ology. For example, a comparative chronological approach has been

Please cite this article in press as: Lewis, L., et al., First technologicalMicrolithic industries: An exploration of inter-regional variability indx.doi.org/10.1016/j.quaint.2014.09.013

taken by Deraniyagala (1992), and a comparative typologicalapproach by Mellars (2006a, Mellars et al., 2013). An exception isthe comparative theoretical approach taken by Hiscock et al.,(2011), which was written in collaboration by specialists in threeseparate geographic regions. Additionally, some comparisons be-tween sites within specific regions have also been conducted suc-cessfully (e.g. Mackay, 2011).

Two case studies of comparative microlithic work serve todemonstrate the reasons why such over-arching studies havecontributed towards a homogenising view of microlithic technol-ogy. The first concerns the ‘modern human behaviour’ debate,which has formed one of the cornerstones of Palaeolithic archae-ological enquiry of the last three decades. It concerns the charac-terisation and archaeological correlates of behavioural traits andabilities associated with modern humans. The majority of archae-ological attention has been on disputes between proponents of the‘short range’ and the ‘long range’ models (Brumm and Moore,2005). The ‘short range’ or ‘human revolution’ model interpretsthe archaeological record c. 60e40 ka as demonstrating the rapiddevelopment of a package of modern human behavioural traits (e.g.Mellars, 1990, 2006a, 2006b; Klein, 1992, 1995; Bar-Yosef, 2002). Incontrast, the ‘long range’ model highlights the earlier and moregradual appearance of these traits in Africa from around 280 ka(McBrearty and Brooks, 2000; McBrearty, 2007).

These contrasting models provide very different interpretationsof the archaeological record in relation to understanding thebehaviour of Homo sapiens. However, what both models have incommon is recourse to a trait list of behaviours as a starting point,which researchers from both camps then use to assess the‘modernity’ of archaeological assemblages. Originally constructedas a list of traits characteristic of the Upper Palaeolithic of Europe,this template was later applied to Africa (McBrearty and Brooks,2000; McBrearty, 2007) and elsewhere (Brumm and Moore,2005; Franklin and Habgood, 2007; Habgood and Franklin, 2008).

Microliths have been included in such lists, as truly microlithicindustries are associated only with H. sapiens. They have also beenargued to “clearly signal a modern approach to technology”(McBrearty and Brooks, 2000, pp. 500) and to be demonstrative ofthe shift to modern cognition and behaviour (Lombard and Haidle,2012). For example, it can be argued that they are related to many‘archaeological signatures of modern human behaviour’ (as listedby McBrearty and Brooks, 2000, pp. 492), including the hafting andstandardisation of composite tools, the use of projectile weaponry,increased dietary breadth and/or specialized hunting, complexmanufacturing sequences including the procurement of raw ma-terials and the control of fire for the production and use of mastics,exchange networks, and even symbolism in the form of regionalartefact styles.

As a result of the inclusion of microliths in list of modern humanbehavioural traits, the unit of analysis has been the simple assess-ment of the presence or absence of microlithic technology; whatWadley (2001, pp. 207) terms “a ‘shopping list’ approach”. This hasthe effect of homogenising microlithic industries through its dis-cussion largely in terms of the appearance dates of certain micro-lithic typologies, while ignoring any indications of underlyingtechnological diversity.

A second related comparative framework that has resulted inthe amalgamation of multiple distinct microlithic industries is thatconcerning human dispersals. The timing and route/s of thedispersal of modern humans out of Africa are still hotly debated.One particularly contentious area concerns the ‘southern arc’ ofdispersal out of Africa and into Asia (Oppenheimer, 2009; Petragliaet al., 2010; Dennell and Petraglia, 2012). This movement ishypothesised to have occurred before the dispersal into Europe,taking H. sapiens out of north-east Africa and along the Indian


Table 1Attributes measured on cores, debitage and tools.

Attributes measured

CoresRaw material variability:

Raw materialEvidence for core recyclingMaximum length (measured perpendicular to length)Maximum width (measured perpendicular to length)Maximum thickness (measured perpendicular to length)Core reduction techniques:

Core typeNumber of striking platformsEvidence for bipolar percussionCore reduction intensity:

Number of flake removal scarsNumber of blade removal scarsLength of each removal scarNumber of scars per 1 cm3 of core volumeAverage scar length per 1 cm3 of core volumeTotal scar length per 1 cm3 of core volume

DebitageRaw material variability:

Raw materialEvidence of recyclingOriented length (measured from the point of percussion)Maximum width (measured perpendicular to length)Maximum thickness (measured perpendicular to length)Reduction trajectories:

Debitage type: flake, blade (at least a 2:1 lengthewidth ratio and having oneor more parallel arrises) or flake-blade (exhibiting only one of the two bladerequirements). If flake-blade, whether it is a squat blade (parallel arrise/sbut less than a 2:1 lengthewidth ratio) or an elongated flake (2:1lengthewidth radio but no parallel arrises)

Condition (intact or otherwise)Number of dorsal scarsNumber of parallel arrisesOrientation of dorsal scarsNumber of dorsal scars per 1 cm2 of dorsal surface areaDorsal surface cortex coverStriking platform cortex coverTermination typeEvidence for core rejuvenationEvidence for platform preparation

ToolsRaw material variability:

Raw materialOriented length (measured from the point of percussion)Maximum width (measured perpendicular to length)Maximum thickness (measured perpendicular to length)Blank selection:

Blank type (flake or blade)Condition (intact or otherwise)Retouch choices:

Typology (recorded in terms of morphology, e.g. ‘side retouched’ ratherthan ‘side scraper’)

Microlith morphology e geometric or non-geometricRetouch typeRetouch location

L. Lewis et al. / Quaternary International xxx (2014) 1e19 3

Ocean rim, eventually ending up in Australia by around 40e50 ka(Bowler et al., 2003; Olley et al., 2006). This dispersal has beenassessed through archaeological (e.g. Balme et al., 2009; Armitageet al., 2011), genetic (e.g. Macaulay et al., 2005), geographic (e.g.Field and Lahr, 2005; Field et al., 2007) and environmental means(e.g. Boivin et al., 2013). Models have often favoured a coastal route(e.g. Macaulay et al., 2005; Bulbeck, 2007), but inland routesmaking use of river valleys have also been proposed (e.g. Korisettar,2007; Boivin et al., 2013).

One model for this southern dispersal route concerns micro-lithic industries. Mellars (2006a, Mellars et al., 2013) draws paral-lels between the morphologies of the small, haftable ‘crescentic’stone tools of the southern African HP and those of the Indian andSri Lankan Microlithic. On the basis of published illustrations ofbacked tools and symbolic material culture (beads and engravedpieces), he argues that typological similarities between these twoindustries are sufficient to support a hypothesis of direct popula-tion dispersal between southern Africa and the Indian subcontinentaround 60e55 ka, via East Africa, even in the face of a substantialchronological gap (c. 20,000 years).

The use of microlithic technology as an archaeological correlateof the earliest dispersals of modern humans out of Africa and intoAsia has not yet been rigorously tested. Equifinality or similarities instone tool typology could be due to convergence, as the result ofmany potential factors such as the hunting of similar sized prey orthe degree of group mobility. The conflation of South Asian micro-liths with those of the HP in order to map a model of populationdispersals negates the opportunity to explore underlying lithicreduction processes, whichwould bemuchmore revealing in termsof our understanding of the technological systems of differentpopulations (e.g. Clarkson, 2013). This can only be achieved througha consideration of microlithic assemblages as unique technologicalstrategies through the use of technological attribute analysis, withthe aim of highlighting and describing areas of similarity and dif-ference in order to demonstrate the degree of homogeneity or di-versity evident within and between microlithic industries.

3. Sampling and methodology

This study investigates entire production trajectories by usingdetailed technological attribute analyses and making use of all theproducts and by-products of lithic manufacture. In order to main-tain suitable numbers for analysis, all cores and microliths wereincluded from each layer studied. For the purposes of the analysis,the broadest possible definition of microliths was used, i.e. as smallhaftable tools made on blade or flake blanks, retouched using anytechnique (backing or otherwise) into any shape (geometric orotherwise). An arbitrary over-arching size limit for microliths wasnot imposed. Instead, the size profiles of tools were investigated ona site-by-site basis for bimodality, which was found at all sites inboth length and width measurements. Size measurements of mi-croliths and non-microlithic tools are discussed below.

A sample of 400 other lithics (debitage plus non-microlithicretouched pieces) with a maximum oriented length of over10 mmwas taken from each layer (or groups of spits, in the case ofRose Cottage Cave). For population sizes of more than a few thou-sand, the sample size required to be representative of the popula-tion increases very little as the population size increases (Krejcieand Morgan, 1970; Guadagnoli and Velicer, 1988). A sample sizeof 400 per layer and 1200þ per site therefore meets the standardsused by the majority of social science disciplines at the 95% confi-dence level (Agresti and Finlay, 2009). Stratified sampling was usedwhere artefacts in museum collections were already bagged ac-cording to type or spit, which has been shown to produce morerepresentative samples than simple random sampling alone


(Agresti and Finlay, 2009). In the case of bags containing largenumbers of lithics, these were set out on a table and selectedthrough systematic random sampling in order to avoid the effectsof weight sorting within bags.

The data collection strategy was designed with two purposes inmind. Firstly, the focus was on variables that would help elucidatetechnological strategies in raw material preferences, reductionstrategies and techniques, blank selection and retouch choices.Secondly, comparability and replicability were key, in order tofacilitate ongoing comparisons with recoded databases of mea-surements previously made on Indian microlithic assemblages(Clarkson et al., 2009; James, 2011). The attributes measured arelisted in Table 1.


Fig. 1. Map of sites mentioned in the text (grey) and those included in the presentstudy (black).


4. Early microlithic industries in southern Africa

4.1. The Howiesons Poort of southern Africa

The HP is probably the best-known and most thoroughlyresearched part of the African MSA (Lombard, 2005; Henshilwood,2012; Lombard et al., 2012; Cochrane et al., 2013). Our current stateof knowledge defines it as a distinctive stone tool industry foundwidely in South Africa, Lesotho and Namibia. Originally believed tobe a stage within the Still Bay industry (Malan, 1952) or transitionalbetween the MSA and LSA (Clark, 1959; Hole, 1959), new OSL datesplace the HP within a tight chronological boundary between 64.8and 59.5 ka, at the end of MIS 4 (Jacobs et al., 2008). However, thereis some disagreement concerning the reliability and accuracy ofthese dates (Gu�erin et al., 2013; Tribolo et al., 2013), and bothyounger and older dates exist at several sites (see Henshilwood,2012 for a discussion of this point).

The HP is characterised by small (typically less than 40 mm)backed tools, especially on blades (Lombard, 2005; Henshilwood,2012). These backed tools are often retouched into geometricshapes such as segments, triangles and trapezes. Some researchersprefer to avoid the term ‘microlithic’ and instead refer to the HP asbeing typified by backed ‘crescentic’ stone tools or segments(Mellars, 2006a; Wadley and Mohapi, 2008; Mellars et al., 2013), insome cases to distinguish these tool forms from their much smallerLSA equivalents (Wadley, 2008). However, as HP microliths fitwithin the size range of those documented elsewhere (Elston andKuhn, 2002; Hiscock et al., 2011) most researchers refer to the HPas a microlithic industry (e.g. Ambrose, 2002; Lombard, 2005;Cochrane et al., 2013). Furthermore, referring to the HP as beingcharacterised primarily by crescentic stone tools ignores the exis-tence of non-geometric microliths, which at many sites make upthe majority of the microlith assemblage (e.g. Kaplan, 1990;Mitchell and Steinberg, 1992; Villa et al., 2010; Porraz et al., 2013).

Explanations for the origins of the HP tend to revolve aroundchanges in resource exploitation and social organisation as aresponse to changing environmental conditions (e.g. Ambrose andLorenz, 1990; Ambrose, 2002; McCall, 2007; McCall and Thomas,2012) and more specifically the use of symbolic material culturein the resultant exchange networks (e.g. Deacon, 1989, 1995; Wurz,1999). They also include models of population replacement (e.g.Singer andWymer, 1982; McBrearty and Brooks, 2000), invocationsof cognitive changes (Lombard and Haidle, 2012) and spiritualconsiderations (Lewis-Williams and Pearce, 2004).

As our understanding of HP chronology and stratigraphy im-proves, researchers have become increasingly aware of both thetemporal and the geographic variability within the industry(Cochrane et al., 2013). The occurrence of distinctive ‘phases’withinHP sequences has been suggested at Rose Cottage Cave (Wadleyand Harper, 1989), Sibudu (Wadley and Mohapi, 2008) and Die-pkloof (Porraz et al., 2013). There is also evidence for variability inlithic technology between sites. Wadley and Harper (1989)demonstrate several areas of difference between the HP at RoseCottage Cave and elsewhere in terms of proportions of stone tooltypes. Lombard (2005) notes further differences in stone tool pro-portions at other HP sites.

Another potential area of technological variability in the HP is inthe realm of hafting and adhesives. A recent molecular study ofresidue found on a quartz flake at the site of Diepkloof resulted inthe identification of Podocarpus elongatus (Yellowwood) resin(Charri�e-Duhaut et al., 2013), in contrast to the ochre-based haftingresidues found at Sibudu Cave (Lombard, 2006, 2007, 2008), RoseCottage Cave (Gibson et al., 2004) and Umhlatuzana (Lombard,2007). These glimpses of potential differences between sites indi-cate that the HP was a variable and dynamic industrial complex,


and the investigation of technological variability within this in-dustry is a potentially highly fruitful avenue for research.

4.2. Rose Cottage Cave, South Africa

Rose Cottage Cave (29�130S, 27�280E) is one of the best knownHP sites in southern Africa, containing 6 m of MSA and LSA de-posits. It is located 3 km east of Ladybrand in the eastern Free Stateof South Africa, near the Platberg River in the Caledon Rivercorridor (Wadley and Harper, 1989;Wadley, 1991,1997) (Fig. 1). Thesite has been subject to three periods of excavation; by B.D. Malanin 1943e46, by Peter Beaumont in 1982, and most recently by LynWadley in 1987e98 (Wadley, 1991; Soriano et al., 2007). The cave isapproximately 20 m long and 10 m wide, eroded into the Karoosandstone. The local environment at the site is currently predom-inantly grassland (Mitchell, 2002) with some scrubland patches(Wadley, 1997). Unfortunately there is no bone preservation inlayers older than c. 20 ka (Plug and Engela, 1992; Jacobs et al.,2008), and organic preservation below this level is rare or absent,with the exception of charcoal. Little is therefore known about thelocal environment of the site during the MSA.

HP layers have been OSL dated; two dates from the lower part ofthe HP sequence gave ages of 63.3 ± 2.3 ka and 65.0 ± 3 ka, and adate from the upper part of the sequence gave an age of63.0 ± 2.3 ka (Jacobs et al., 2008). A date from the oldest post-HPassemblage gave an age of 56.0 ± 2.3 ka. An alternative OSLdating program gave similar end dates for the HP, of 62.5 ± 2.9 ka or64.1 ± 3.0 ka, depending on the source of the U and Th dosemeasurements (Pienaar et al., 2008). However, it indicated that theearliest HP at the site may date slightly earlier, at 66.9 ± 2.6 ka or68.7 ± 2.7 ka. The OSL dates roughly correspond with those ach-ieved through TL dating of burnt lithics at the site (Valladas et al.,2005). TL ages between 56.3 ± 4.5 and 60.4 ± 4.6 ka were givenfor the top of the HP sequence, with a weighted mean age of58.2 ± 4.2 ka. However, the earlier HP layers and the uppermostpre-HP layer were not dated.

Malan only published a short note on his excavations (Malan,1952), with no analysis of the lithics recovered. These were latercounted and published by Wadley and Harper (1989), using a ty-pology based on that of Singer and Wymer (1982) and Thackerayand Kelly (1988). They suggested there may be two or threephases within the HP at Rose Cottage Cave, an argument also



advanced for Sibudu Cave (Wadley and Mohapi, 2008). They alsohighlighted differences between the Rose Cottage Cave HP and HPassemblages at other sites, including relatively small proportionsof segments and trapezes and relatively large proportions ofbacked and obliquely backed blades, noting that “[m]uch vari-ability is thus evident in Howiesons Poort assemblages, evenwhere sites are geographically close” (Wadley and Harper, 1989,pp. 31).

The lithics from Beaumont's excavations in the 1960s wereanalysed by Kohary in an attempt to investigate aspects of pro-duction techniques. However, she mis-read the stratigraphy andmixed the assemblages (Wadley and Harper, 1989), and the firstreliable technological analysis at the site did not occur until almosttwo decades later (Soriano et al., 2007). The HP lithics from Wad-ley's excavations in the 1980s and 90s were first published byHarper (1997), whose typological analysis suggested a considerabledegree of continuity between the different MSA assemblages at thesite, as well as some variability within the HP sequence in terms oftool frequencies.

A technological approach to the lithics excavated byWadley andHarper was taken by Soriano et al. (2007), who looked at the chaîneop�eratoire of blades in HP and post-HP assemblages. Their resultsconfirmed Wadley and Harper's (1989, Harper, 1997) findingsregarding variability within the HP sequence at Rose Cottage Cave,and gradual change between the HP and post-HP. Additionally,residue analysis has been conducted on backed tools from HPlevels, revealing ochre and plant residues on the backed portions ofthe tools and adding further support to the argument that thesemicroliths were hafted (Gibson et al., 2004).

A total of 1538 lithics was analysed from the site for the pur-poses of the current study. The sample was taken from Malan'sexcavations rather than the more recent excavations, as the latterdid not contain enough lithics to meet the sampling criteria of thisstudy. Analysis focussed on the ‘classic’ HP as documented byWadley and Harper (1989), between 144 and 204 inches (Fig. 2A),excluding potentially mixed levels either side.

Little is known about the history of Malan's excavation, and it istherefore possible that the integrity of the assemblage is unreli-able. However, comparisons with data from Wadley's later exca-vations at the site (Harper, 1997) indicate that debitageproportions are very similar (61% blades in Harper's reportcompared to 59% reported in the current study), and that a greaterproportion of broken lithics exists in the current study (18.5%compared to 5%), demonstrating that neither flakes nor brokenlithics were selectively thrown away. The discrepancy in brokenlithic proportions in fact indicates the only area where selectivecuration may have been utilised, namely in the preservation ofwhat are generally referred to in the South African literature as‘chips’ and ‘chunks’. The former tend to be smaller than the 10 mmcut-off point used in this study, and the latter are generallydefined as neither being complete cores nor displaying a clearventral surface, which are also not included in this study. There-fore any selective curation was minimal, and has not influencedthe results of this study or made it incomparable to the otherassemblages analysed here.

4.3. Umhlatuzana, South Africa

Umhlatuzana (29�4802800S, 30�4502200E) is a MSA site near Dur-ban in KwaZulu-Natal, South Africa, spanning from over 70 ka tothe late Holocene and including Still Bay and HP industries (Kaplan,1990). It is a large rockshelter, c. 43 m wide, located in the ortho-quartzite horizon of the Natal Supergroup, about 100 m abovethe Umhlatuzana River (Fig. 1). It is in a mixed woodland savannahbiome (Mitchell, 2002), and today is in an outlying patch of coastal


forest (Kaplan, 1989, 1990). Excavation was conducted in 1985 aspart of a rescue project, and reached a depth of 2.6 m divided into28 stratigraphic levels.

As Mohapi (2013) notes, Umhlatuzana was long neglected dueto confusions over the dating and stratigraphy at the site. However,new OSL dates have helped to clarify these issues in the earlier partof the sequence (Jacobs et al., 2008; Lombard et al., 2010, 2012).Although the dates confirmed that there had indeed been somepost-depositional mixing between layers, they demonstrated thatthis was confined to discrete, small-scale events. The majority ofgrains dated were shown to be in their primary context and so thedates are considered to be accurate, although some mixing ofsediments and artefacts is anticipated (Lombard et al., 2010).Further mixing is expected due to the excavation of the lag depositsin Layers 5e28 in spits, although Lombard et al. (2010) argue thatthere is still a good degree of internal consistency between thelayers.

Due to this small-scale mixing, there is some disagreementconcerning which layers should be attributed to the HP. Layer 24,dated to 60.0 ± 3.5 ka, is accepted as HP (Kaplan, 1990; Lombardet al., 2012; Mohapi, 2013, c.f. Lombard et al., 2010 who incor-rectly report this as Layer 22) (Fig. 2C). However, whereas theexcavator states that the HP extends to Layer 26 (Kaplan, 1990),Lombard et al. (2010) argue that it only extends to Layer 24, on thebasis of an OSL date of 70.5 ± 4.7 ka that they attribute to Layer 25.They instead assign Layers 25e27 to the Still Bay industry. However,this date actually originates from a unit which is equated to Layer27 (Kaplan, 1990).

Given recent arguments that the OSL dates of Jacobs et al.,(2008) may actually contain a much greater degree of uncertainty(Gu�erin et al., 2013; Tribolo et al., 2013), we cannot rule out theattribution of Layers 25e27 to the HP on the basis of a single dateenor should we ascribe industries or cultural traditions on the solebasis of dates. In fact, when the archaeological evidence isconsidered, the frequencies of segments, trapezoids and backedpieces remain high below Layer 24 (Fig. 3). Although it is notpossible to determine from Kaplan's data which of these backedpieces may fit within the size range of microliths at the site, thesetechnological and morphometric features are typical of HP as-semblages. Therefore, allowing for some post-depositional mixingof artefacts, it is reasonable to assert that HP occupation at the siteextends below Layer 24.

Umhlatuzana is rich in lithic artefacts, with over 1 millionlithics recovered (Kaplan, 1990). Until very recently, lithic analysisat the site consisted solely of the typological classifications con-ducted by the excavator (Kaplan, 1989, 1990), with some lengthmeasurements. Subsequent analysis has focussed on the pointswithin the Still Bay layers and on morphological comparisons withpoints found later in the sequence (Lombard et al., 2010; Mohapi,2013). However, the HP technology at the site remains poorlyunderstood.

The lithics for the present sample were taken from Layers 23,25 and 27 of squares J2 and K2, comprising a total of 1543 lithics(Fig. 2C). This selection of layers allows for an assessment ofwhether the latter two should be included as part of the HPsequence at the site, on the basis of lithic technological identifi-cations rather than solely typological assertions (cf. Lombardet al., 2010). The analysis of alternate layers was chosen in or-der to maximise stratigraphic integrity, given the possibility ofmixing at the site. Sediments slope towards the dripline inUmhlatuzana rockshelter. Therefore two adjacent squares at thesame point on the slope were chosen from the four excavated tobedrock in order to minimise the potential compounding effect ofthis sloping, given that this section of the stratigraphy wasexcavated in spits.


Fig. 2. Stratigraphic profiles of the sites. Highlighted layers are those from which lithics were sampled. A: Rose Cottage Cave (redrawn from Wadley and Harper, 1989). B:Batadomba-lena (redrawn from Perera et al., 2011). C: Umhlatuzana (redrawn from Mohapi, 2013, with dates from Lombard et al., 2010).


Please cite this article in press as: Lewis, L., et al., First technological comparison of Southern African Howiesons Poort and South AsianMicrolithic industries: An exploration of inter-regional variability in microlithic assemblages, Quaternary International (2014), http://dx.doi.org/10.1016/j.quaint.2014.09.013

Fig. 3. Diagnostic tool type proportions in lower MSA levels at Umhlatuzana (datafrom Kaplan, 1990), demonstrating that frequencies of typical Howiesons Poort arte-facts (segments, trapezoids and backed pieces) remain high below layer 24, and thatthese layers could therefore be considered part of the Howiesons Poort sequence at thesite.


5. Early microlithic industries in South Asia

5.1. The Microlithic industry of South Asia

Microlithic technology in South Asia was originally considered aHolocene phenomenon (Misra, 2001). However, in the 1980s andearly 1990s new chronometric ages indicated that microlithic in-dustries dated back to 28 ka in Sri Lanka at the sites of Batadomba-lena, Bundala Site 49, Bundala-Patirajawela Site 50 and KitulgalaBeli-lena (Deraniyagala, 1984, 1986, 1988, 1992, 2007), and at least25 ka in India at the site of Patne (Sali, 1985).

More recently the discovery of microliths at the site of Fa Hienbelow a layer dated to c. 38 ka, and recalibrated and new dates fromBatadomba-lena going back to c. 36 ka, confirmed the age of theearliest microlithic industries in Sri Lanka (Perera, 2010; Pereraet al., 2011). In India, the excavation of the site of Jwalapuram 9pushed the earliest dates for microlithic technology in the countryback to a similar age as in Sri Lanka (Clarkson et al., 2009; Petragliaet al., 2009). Microlithic industries continue into the Holocene inboth India and Sri Lanka. In the latter they occur up until the IronAge in the 2nd millennium BC (Deraniyagala, 1992).

The development of microlithic technology in South Asia fol-lowed on from the appearance of a variety of blade andmicro-bladeassemblages from around 45 ka (James and Petraglia, 2005). Thisincludes micro-blades at the site of Mehtakheri, India, dating fromc. 44e35 ka (Mishra et al., 2013), and a blade-based industry thatthe authors consider to contain elements that are microlithic in sizeat Site 55, Riwat, Pakistan, which may date to over 45 ka (Dennellet al., 1992). Variously described in the literature as ‘Mesolithic’,‘Microlithic’, ‘Upper Palaeolithic’, ‘Late Stone Age’ or, in Sri Lanka,‘Balangoda Culture’, this paper follows the lead of James andPetraglia (2005) in referring to this assortment of blade- andmicrolith-based industries as Late Palaeolithic. The termMicrolithicrefers to microlithic industries occurring within the LatePalaeolithic.

Although the origins of these microlithic technologies arecurrently poorly understood, technological analysis of lithics fromPatne, India, one of the few long stratified sequences in South Asiaspanning the Middle and Late Palaeolithic, suggests that theyemerged locally and gradually out of preceding Middle Palaeolithicindustries (James, 2011, originally hypothesised by Deraniyagala,1988, 1992). This is also suggested in Jwalapuram (Petraglia et al.,2009). Blade use appears to be a continuation of Middle Palae-olithic technological practices elsewhere in South Asia (Misra and


Bellwood, 1985). Furthermore, characteristic ‘Upper Palaeolithic’tool types (e.g. burins, end-scrapers and micro-blades) also appearin Middle Palaeolithic contexts, including at Bhimbetka rockshelterin north-central India (Misra, 1985).

In Sri Lanka, the Microlithic is characterised by very small mi-croliths made on quartz flakes, often less than 20 mm in length.Both geometric and non-geometric forms are found, including‘classic’ segment, trapeze and triangle forms as well as the distinctbifacially-retouched tanged ‘Balangoda point’, unique to Sri Lankaand southern India (Zeuner and Allchin, 1956; Deraniyagala, 1992).Bipolar blunting or backing of microliths along lateral margins wasoften used (Perera et al., 2011). In contrast, microliths from LatePalaeolithic sites in India appear to be primarily made on bladesand are longer, ranging up to ~40 mm in length (Clarkson et al.,2009; Petraglia et al., 2009; James, 2011). Although inter-regionalcomparisons have not previously been conducted on Late Palae-olithic microliths, some differences are already evident from thepublished literature.

5.2. Batadomba-lena, Sri Lanka

Batadomba-lena (6�460N, 80�120E) is a small north-east facingrockshelter located in the wet zone lowland rainforest in south-west Sri Lanka, formed in the gneiss bedrock of the region(Deraniyagala, 1992; Perera, 2010; Perera et al., 2011) (Fig. 1). It isthe richest, best known and most thoroughly researched prehis-toric site in Sri Lanka, and has yielded over 400,000 lithics, inaddition to some of the earliest Homo sapiens skeletal remains inSouth Asia (Kennedy et al., 1986, 1987; Kennedy and Deraniyagala,1989; Kennedy, 2000; Perera et al., 2011).

The site was first excavated by P.E.P. Deraniyagala in 1937e40,who interpreted it as belonging to a Mesolithic ‘Balangoda’ culturebetween the Pleistocene and Holocene. It was later excavated moreextensively by his son S.U. Deraniyagala in 1979e1986 (P.E.P.Deraniyagala, 1940, 1963; S.U. Deraniyagala, 1992). The systematicexcavations of S.U. Deraniyagala e reaching a depth of 2.8 m e

included a comprehensive dating program (radiocarbon, ESR, TLand OSL) which dated the earliest archaeological materials at thesite to the Late Pleistocene at around 28,000 BP (Deraniyagala,1984, 1992; Agrawal et al., 1985; Abeyratne et al., 1997). The mostrecent excavations at the site, conducted by Nimal Perera in 2005,pushed the earliest dates back further, with a calibrated radio-carbon date from the earliest layer of 36,283e34,609 cal BP (Wk-19963: 30,603 ± 400 BP, calibrated and reported to the 2s level inPerera et al., 2011). Microlithic occupation at the site continues upto the Pleistocene-Holocene boundary (with a calibrated radio-carbon date of 12,091e11,626 cal BP (Wk-19965: 10,193 ± 57 BP)),with no obvious temporal breaks in habitation. Equatorial rain-forest persisted in the region throughout this period, as can be seenin the floral record as well as in the consistency of the dominance ofmonkey species in the faunal record (Perera et al., 2011).

Previous lithic analysis at Batadomba-lena was conducted byS.U. Deraniyagala (1992) and Perera (2010, Perera et al., 2011).These analyses were largely typological in nature, with someadditional debitage and use-wear analysis conducted by Perera(2010) on smaller samples. A wide range of very small (<28 mm)geometric and non-geometric microliths were found from theearliest layers onwards, including lunates, triangles, trapezoids andtwo complete bifacial microlithic ‘Balangoda points’. Microlithsmake up a very small percentage of the total assemblage (<0.1%),and were found to be more abundant and morphologically diversein pre-LGM layers, becoming less abundant during intensified useof the rockshelter after the LGM.

Lithics were sampled from the 2005 excavations at Batadomba-lena for the purposes of this study, as these are the best


Fig. 4. Raw material proportions of major lithic categories at Rose Cottage Cave,demonstrating the preferential use of particular raw materials for particular lithicproducts. CCS varieties are dominant, and were preferred for blade and blade tools,especially jasper. Hornfels was used preferentially for non-microlithic tools.


provenanced. Microliths from the 1980s excavations were alsoincluded, as the numbers from the 2005 excavation alonewere low.A total of 1680 lithics was sampled from the site. The analysis waslimited to Layers 4, 5, 6, 7a, 7b and 7c (Fig. 2B), as later layers werebelieved to be disturbed in modern times and were not dated(Perera, 2010).

6. Preliminary results of lithic analyses

6.1. Rose Cottage Cave

6.1.1. Raw materialsA wide variety of raw materials were worked at the site, with a

dominance of crypto-crystalline silicas (CCS) (Table 2). Wadley andHarper (1989) prefer to group these under the category ‘opalines’,but this paper follows Malan's original definitions which dividethese into chalcedonies, cherts and agates (Malan, 1952), with theaddition of a jasper as an opaque form of CCS on the advice of localgeologists. This allows for further comparisons to be made withinsub-groups of CCS. The most likely source of the raw materialsknapped at Rose Cottage Cave is the Caledon River, 8 km away(Wadley and Harper, 1989).

One additional source of raw materials at the site was throughthe re-working of much older lithics from earlier MSA sites in thearea. Many cores (40.2%) exhibit worn and patinated surfaces withancient removal scars alongside fresher scars. These ‘recycled’lithics were an important source of rawmaterials at the site, both ascores, with much older flakes having new flakes removed for use,and as blanks, with flakes removed and the reworked original piecethen utilised. The “abundance of twice-worked specimens” wasalso noted by Malan (1952, pp. 192). There are known earlier MSAsites in the area, including one 100m away on the hillside above thecave (Wadley, pers. comm.) which are likely sources of the mate-rials. Alternatively, these lithic clasts may have been encountered inthe nearby Caledon River, as other rawmaterials are known to havebeen (Wadley and Harper, 1989; Wadley, 1991).

6.1.2. CoresA total of 204 cores was analysed from HP layers at the site

(Table 2). Most cores (91.3%) were made on CCS, with jasperparticularly favoured for blade cores (Fig. 4). Amorphous multi-directional/multi-platform cores (those worked from more thantwo directions) weremade on awider variety of rawmaterials thanother core types. Uni- and bi-directional blade cores are the mostcommon types, together making up 39.7% of the core assemblage(Table 3). Blade cores are defined here as those where over half ofterminal removals are of blade dimensions. If bipolar cores areincluded as blade cores the proportion rises to around half of the

Table 2Raw material frequencies in each major lithic category in the sample from Rose Cottage Cother numbers are sample sizes.

Raw material Blade Flake-blade Flake Microlith on blade

Jasper 285 93 42 61Chalcedony 142 64 38 28Agate 82 35 37 12Silcrete 41 20 20 1Hornfels 29 16 10 2Claystone 35 12 12 1Siltstone 17 19 12 0Chert 20 9 7 2Quartzite 15 8 5 1Impure clear quartz 0 0 0 0Milky quartz 0 0 2 0Opaque quartz 0 1 0 0Total 666 277 185 108


core assemblage (49.5%). Bipolar cores can be considered a type ofblade core in this assemblage as the mean number of blade scars onbipolar cores is identical to that of uni-directional blade cores (3.9),while the mean overall number of removals per core is almostidentical to that of bi-directional blade cores (7.2 compared to 7.1)(Table 4).

Cores of other morphologies are also evident in the sample insignificant numbers, with at least 20 examples of each type. Thisincludes 20 cores displaying potential evidence of bipolar percus-sion, including crushing, blunting and step fractures on the distalportion where the core is in contact with the anvil, as well aswedging initiations, flat fracture planes with perpendicular strikingplatforms, and fissuring from the proximal and distal ends. Bipolarpercussion is often considered to be a LSA phenomenon in sub-Saharan Africa (Ambrose, 1998; Mercader and Brooks, 2001; Villaet al., 2012; Eren et al., 2013). At the same time, elements of atypically MSA technological strategy can be seen in the use of radialcores (Ambrose, 1998; Wurz, 2013).

Core reduction intensity can be assessed through comparisonsof the numbers and sizes of removal scars to the size of cores(Table 5). There are both more removals overall on discarded

ave. Microlith and core numbers are absolute numbers from the layers analysed. All

Microlith on flake Retouched blade Retouched flake Core Total

7 15 8 86 5979 6 1 53 3416 5 2 42 2211 3 3 1 902 6 6 6 771 2 1 4 680 5 2 3 580 1 0 5 440 3 3 2 370 0 0 2 20 0 0 0 20 0 0 0 1

26 46 26 204 1538


Table 3Core type frequencies at Rose Cottage Cave, Umhlatuzana and Batadomba-lena.

Core type Rose Cottage Cave Umhlatuzana Batadomba-lena

No. % No. % No. %

Uni-directional: Blade 46 22.5 14 8.6 2 9.5Flake 23 11.3 35 21.6 2 9.5Bipolar 20 9.8 54 33.3 7 33.3

Bi-directional: Blade 35 17.2 6 3.7 0 0Flake 28 13.7 20 12.3 2 9.5

Multi-directional: Radial 26 12.7 10 6.2 2 9.5Amorphous 26 12.7 23 14.2 6 28.6

Total 204 162 21

Table 4Mean numbers of flake, blade and total terminal removals visible on each type of core at Rose Cottage Cave, Umhlatuzana and Batadomba-lena.

Mean removals per core: Rose Cottage Cave Umhlatuzana Batadomba-lena

Core type Blades Flakes Total Blades Flakes Total Blades Flakes Total

Uni-directional: Blade 3.9 1.5 5.4 4.0 1.2 5.2 4.5 0.5 5.0Flake 1.0 3.7 4.7 1.0 5.0 5.9 0.0 3.5 3.5Bipolar 3.9 3.3 7.2 1.8 3.6 5.4 1.7 3.9 5.6

Bi-directional: Blade 4.6 2.5 7.1 4.5 2.3 6.8 e e e

Flake 1.6 5.1 6.7 1.6 6.7 8.2 0.5 6.0 6.5Multi-directional: Radial 1.2 6.8 8.0 0.4 10.5 10.9 0.0 9.0 9.0

Amorphous 1.4 8.1 9.5 0.9 6.3 7.2 0.2 8.2 8.3


bipolar cores in relation to their size, and removals of a larger size inrelative terms. Bipolar cores are the most heavily worked core type,suggesting that bipolar reduction was not simply used as a ‘lastchance’ reduction method for exhausted cores, but that it waspreferentially used to produce a large number of flakes and, inparticular, blades from small cores.

6.1.3. DebitageAnalysis of the debitage sample is consistent with previous re-

ports that the assemblage at Rose Cottage Cave is predominantlyblade-based (Wadley and Harper, 1989; Soriano et al., 2007).Overall, 59.0% of the debitage sample can be considered true blades,with only 16.4% of debitage being flakes (Table 2). The remaining24.6% fall into the category of flake-blades. Of these, the vast ma-jority (93.5%) are squat blades (with one or more parallel arrises) asopposed to elongated flakes (no parallel arrises). Therefore flake-blades are mostly blades that do not meet the artificial distinc-tion of being twice as long as they are wide, adding furtheremphasis to the blade-dominated nature of the assemblage.

CCS types were preferred overall for the production of blades(Table 2, Fig. 4). Opaque jasper was particularly preferred for bladescompared to more translucent varieties of CCS. Preferences may

Table 5Threemeasures of core reduction intensity on different types of core at Rose Cottage Cave,core volumemeasured as longest axis�maximumwidth�maximum thickness: the meacore volume, and total scar length per 1 cm3 of core volume.

Mean per core per cm3 of corevolume:

Rose Cottage Cave Umh

Core type No. scars Average scarlength (mm)

Total scarlength (mm)

No. s

Uni-directional: Blade 0.9 3.4 1.8 1.5Flake 0.6 1.8 0.9 0.8Bipolar 2.4 5.4 3.8 2.3

Bi-directional: Blade 1.2 3.0 2.1 1.3Flake 0.8 2.1 1.3 0.9

Multi-directional: Radial 1.1 1.6 1.2 0.8Amorphous 1.1 1.6 1.3 0.9


have been for particular colours, qualities or opacities of CCS.Alternatively, there may have been size differences in availablenodules of different CCS varieties resulting in their use for differentpurposes. This interpretation might also explain the clear sizegroupings of CCS and claystone debitage compared to other rawmaterials (hornfels, quartzite, quartz, silcrete and siltstone). CCSand claystone debitage is both smaller and more standardised(Fig. 5).

Size groupings are particularly robust in the case of blades,where CCS varieties and claystonewere utilised in order to producesmall blades of relatively standardised dimensions. This group ofblades have average maximum dimensions (intact only) of31.2 � 11.3 � 4.3 mm. An array of variables (Table 6) providesfurther evidence that these small blades were the desired outcomeof the knapping process, by demonstrating that they were removedlater in the reduction sequence than other forms of debitage,namely flakes. Blades exhibit less dorsal surface and striking plat-form cortex, denser concentrations of dorsal scars, more frequentplunging terminations and greater evidence for late-stage corerejuvenation, while flakes are more likely to exhibit dorsal scarsoriented from three directions, and evidence for platform prepa-ration. While none of these factors alone is conclusive, together

Umhlatuzana and Batadomba-lena, recorded as ratios of removal frequency or size ton number of removal scars per 1 cm3 of core volume, average scar length per 1 cm3 of

latuzana Batadomba-lena

cars Average scarlength (mm)


No. scars Average scarlength (mm)


5.7 2.9 0.7 3.9 1.92.6 1.4 0.5 2.5 1.06.5 3.2 1.1 2.7 1.53.6 2.5 e e e

1.9 1.5 0.5 1.3 0.81.1 1.1 0.8 0.9 0.82.9 1.7 0.6 1.0 0.8


Fig. 5. Histogram of the volumes of intact CCS and claystone v. hornfels, quartz,quartzite, silcrete and siltstone debitage at Rose Cottage Cave, calculated as orientedlength�maximum width�maximum thickness. CCS and claystone debitage issmaller on average and more tightly grouped around an ideal size, whereas that ofother raw materials is larger on average and more variable.

Table 7Frequencies of microlith and non-microlithic tool typologies at Rose Cottage Cave,Umhlatuzana and Batadomba-lena.

Typology RoseCottageCave

Umhlatuzana Batadomba-lena

MICROLITHS

Backed non-geometric microliths:Backed 28 20 12Backed and side retouched 0 0 3Double backed 0 0 1

Backed non-geometric microlith points:Backed 44 70 39Backed and side retouched 0 0 1Double backed 1 2 12

Bifacial/Balangoda point 0 0 1

Crescent microliths:Crescent 16 54 26


they indicate that, on average, flakes were removed before blades inthe core reduction process. Correspondingly, a reasonable conclu-sion would be that they were primarily removed in the process ofpreparing cores for the production of blades. Microlith blanks werethen selected primarily from this population of small CCS blades, aswill be demonstrated below.

Crescent with notch 0 0 2

Microlith preform 9 0 15

Non-geometric microliths:Double side retouched 1 0 0Side retouched 1 1 0Other 9 2 12

Non-geometric microlith points:Concave point (‘awl’) 0 1 1Other 10 9 9

6.1.4. ToolsA total of 134microliths and 72 non-microlithic tools (retouched

blades and retouched flakes) was analysed (Table 2). Blade tools aremore common than flake tools, with 80.6% of microliths and 63.9%of other tools made on blade blanks. CCS varieties make up 75.7% ofthe debitage assemblage. However, they were used in much largerproportions for the manufacture of microliths (93.3%), and in muchsmaller proportions for the manufacture of other tools (52.8%)(Fig. 4). Jasper was especially preferred for microliths made on

Table 6Debitage variables at Rose Cottage Cave. Blades exhibit less cortex, greater dorsalscar densities, more frequent plunging terminations and greater evidence of corerejuvenation. Flakes exhibit more varied dorsal scar orientations and greater evi-dence of platform preparation. Overall the evidence suggests that, on average, flakeswere removed earlier in the reduction process than blades, and that blades were thetarget product.

Blades Flake-blades Flakes

Mean dorsal cortex (%) 7.2 18.3 22.5% with cortical striking platformsa 4.1 12.7 17.7Mean no. scars per 1 cm2 dorsal surface 1.0 0.7 0.6% with plunging terminationsa 23.0 12.1 7.2% core rejuvenation flakes 3.0 0.4 1.6% with dorsal scars from three directions 0.9 2.4 7.0% platform preparation flakes 2.1 2.9 4.9

a As a percentage of debitage with intact striking platforms or terminations.


blade blanks (56.5%).While the reduction of CCS nodules appears tohave been primarily for the purpose of producing blanks suitablefor retouch into microliths, the higher proportions of non-microlithic retouched tools on other raw materials, especiallyhornfels and quartzite, explains why these materials were alsoworked at Rose Cottage Cave. Two different tool production tra-jectories were utilised at the site.

The majority of non-microlithic tools (56.9%) are side ordouble side retouched, including those that also exhibit points,burins, notches or denticulation (Table 7). Elements of a moregeneral MSA technology can be seen in the existence of 13unifacial points. These are found throughout the HP sequence atRose Cottage Cave, and so cannot be attributed solely to mixingbetween layers.

A variety of microlith forms are evident (Table 7, Fig. 6A). Non-geometric forms are more common than geometric (crescent,rectangle and trapeze) shapes (81.3% non-geometric). Wadley

Notched microlith 0 0 3

Rectangle microliths:Backed 3 1 0Double backed 1 1 0Other 1 0 3

Shouldered/tanged microliths:Shouldered 1 0 3Shouldered and backed 2 0 0Shouldered and backed point 1 5 1Shouldered point 1 1 0Shouldered with notch 0 0 1Tanged point 1 0 1Tanged and backed point 0 0 1

Trapeze microlith 4 9 1

Triangle microlith 0 5 7


Table 7 (continued )

Typology RoseCottageCave

Umhlatuzana Batadomba-lena

Total 134 181 155

NON-MICROLITHIC TOOLS

Burins:Burin 1 0 1Double side retouched with burin 1 0 0Side retouched with burin 0 1 0

End retouched 5 0 0

Misc 2 0 1

Notched/denticulated:Denticulated 0 1 0Double notched 1 0 0Notched 0 0 2Side retouched and denticulated 1 0 0Side retouched with notch 1 0 0Side retouched with notches 0 1 0

Points:Backed 3 0 0Bifacial 0 1 0Double side retouched 1 0 0Side retouched 4 2 0Unifacial 13 1 0

Side and end retouched:Circumference retouched 2 0 0Double side and end retouched 2 0 0Side and double end retouched 1 0 0Side and end retouched 1 0 0

Side retouched:Double side retouched 10 6 1Side retouched 23 1 3

Total 72 14 8


and Harper (1989) similarly noted that the Rose Cottage Cave HPcontains fewer segments and trapezes and more non-geometricbacked microliths on blade blanks in comparison to other HPsites. Of the geometric microliths, crescents are most common(11.9% of the total microlith assemblage). Rectangle and trapezeshapes are also present, as are small numbers of shouldered andtanged microliths. There is a clear preference for backed andpointed microlith forms at the site, with only 21 microliths, or15.7% of the microlithic tools, being neither backed nor pointed.

The existence of separate microlithic and non-microlithic toolstrategies is emphasised by their different size profiles (Table 8). Ingeneral, the size profiles of microlithic tools match those of thegeneral debitage population, while non-microlithic tools areconsistently and considerably larger in every metric. Whereasgeneral lithic reduction at the site was geared towards the pro-duction of blanks of a suitable size for the manufacture of mi-croliths, only the largest flakes and blades were selected forretouch into other tool forms. Overall, it is clear that microlithictools were a distinct and significant tool type at Rose Cottage Cave,for which the majority of the lithic reduction process at the sitewas designed.

6.2. Umhlatuzana

6.2.1. Raw materialsHP site occupants at Umhlatuzana used a wide range of raw

materials. Quartz of various qualities and hornfels are dominant(51.2% and 28.7%, respectively) (Table 9). Hornfels was preferred


for blades and blade tools, and quartz and quartzite for flakesand flake tools, with claystone blanks also used preferentially forblade tools (Fig. 7). Hornfels and quartz were also the mainmaterials used for the production of microliths (Table 9, Fig. 6B).However, different production strategies appear to have beenused for these different raw materials, as will be demonstratedbelow.

6.2.2. CoresA total of 162 cores was analysed from HP layers at the site

(Table 3). The majority of these are on quartz or quartzite (95.7%)(Table 9). There is a clear preference for working cores from a singleplatform (63.6% of cores), with a third of cores (33.3%) containingevidence for the use of bipolar percussion. Bipolar percussion isalso evident on 8.9% of debitage, including distal crushing as well aswedging initiations and flat fracture planes with perpendicularstriking platforms. This is common in quartz assemblages (Driscoll,2011). Bipolar cores at Umhlatuzana were predominantly used forthe production of flakes, as determined by the size proportions ofterminal removals (Table 4).

The prevalence of flake production at the site can be seen in thefact that blade cores make up only 12.3% of the core assemblage.Therefore the core assemblage alone does not allow Umhlatuzanato fit within the argument advanced elsewhere that the HP ispredominantly a blade industry (Soriano et al., 2007;Wadley, 2008;Wadley and Mohapi, 2008). However, these blade cores were moreintensively worked in comparison to other core types (with theexception of bipolar cores) when core size is taken into account(Table 5). Although the absolute number of blade cores is lowcompared to flake cores, they were worked more intensively toproduce a greater number of relatively large removals.

6.2.3. DebitageIn agreement with the core assemblage, debitage at Umhlatu-

zana exhibits a mix of flake and blade technologies and cannot beconsidered to be primarily a blade industry. Only 19.6% of thedebitage sampled are true blades, with 39.0% flakes and 41.4%falling into the intermediate flake-blade category. However, 97.4%of the latter category are squat blades rather than elongated flakes,and so it is most likely that these pieces were still removed in alaminar fashion.

The dominant raw material is quartz, which makes up 46.0% ofthe debitage assemblage (Table 9). As noted above, hornfels waspreferred for blades and quartz for flakes. Interestingly, althoughthere are only six hornfels cores in the core assemblage (3.7%),hornfelsmakes up 31.6% of debitage. Theremay therefore have beendifferential transport of hornfels lithics, possibly with cores beingprimarily worked off-site. There are also considerable differencesbetween the sizes of hornfels and quartz debitage (Fig. 8). Themeansize of intact hornfels debitage is 30.5� 20.7�6.0mm, compared to21.0� 16.1�6.1 mm for quartz. Possible explanations for these sizedifferences include potential differences in the available sizes of rawmaterial clasts, and the difficultly of removing large flakes fromcores of low quality quartz (Knight, 1991). However, as will bedemonstrated below, a more compelling argument revolves aroundthe use of hornfels and quartz to intentionally produce differentsizes and types of blanks for retouch into distinct microlith forms.

6.2.4. ToolsA total of 181 microliths and 14 non-microlithic tools was ana-

lysed (Table 9). Overall, 51.9% of themicroliths and 71.4% of the non-microlithic tools were made on blade blanks. Roughly half of toolsmade on blade blanks (microliths and others) (48.1%) were man-ufactured on hornfels, compared to 17.8% of tools produced on flakeblanks. In the debitage sample, hornfels makes up 42.5% of the


Fig. 6. Examples of microliths from the sites studied. A: Rose Cottage Cave. B: Umhlatuzana (top row: hornfels, bottom row: quartz). C: Batadomba-lena, including Balangoda point(top left).


blade assemblage and 26.4% of the flake assemblage. In contrast,76.7% of tools manufactured on flake blanks were made on quartz,compared to 26.0% of tools made on blade blanks. In total, 54.8% offlake debitage is quartz, as is 25.8% of blade debitage. These rawmaterial differences are statistically significant at the 0.05 level (for


blades c2 ¼ 32.087, p < 0.001, for flakes c2 ¼ 26.875, p < 0.001).Overall, there is clear evidence for the selection of hornfels bladesand quartz flakes as tool blanks.

Average tool sizes are summarised in Table 8. Differences can beobserved between raw materials. Hornfels microliths tend to be


Table 8Length, width and thicknessmeasurements of microlith and non-microlithic tools atRose Cottage Cave, Umhlatuzana and Batadomba-lena.

Measurement Min Max Mean Std. dev.

Rose Cottage Cave:Microliths Oriented length (mm) 12.5 46.8 28.4 6.34

Maximum width (mm) 6.5 34.9 12.5 4.46Maximum thickness (mm) 2.1 8.4 4.1 1.41

Non-microlithic tools Oriented length (mm) 23.9 72.9 43.0 10.89Maximum width (mm) 7.5 42.5 21.9 7.09Maximum thickness (mm) 2.8 18.1 7.6 2.86

Umhlatuzana:Microliths Oriented length (mm) 8.0 53.3 22.2 8.86



Batadomba-lena:Microliths Oriented length (mm) 6.8 35.9 17.0 4.73



TabRaw material frequencies in each major lithic category in the sample from Umhlatuzananalysed. All other numbers are sample sizes.

Raw material Blade Flake-blade Flake Microlith on blade Micro

Hornfels 99 149 122 44 15Milky quartz 29 129 158 14 31Opaque quartz 14 64 51 1 13Quartzite 47 65 53 3 0Impure clear quartz 8 28 29 4 17Siltstone 7 21 13 0 0Clear quartz 6 4 5 6 8Claystone 6 7 3 10 2Jasper 8 3 6 9 1Silcrete 4 10 11 0 0Smokey quartz 1 4 7 0 0Granular quartz 2 2 3 1 0Chalcedony 2 4 1 2 0Rose quartz 0 1 0 0 0Total 233 491 462 94 87

Fig. 7. Raw material proportions of major lithic categories at Umhlatuzana. Hornfelsand various quartz qualities are dominant. Hornfels was used preferentially for bladesand blade tools, and quartz and quartzite were preferred for flake and flake tools.Claystone blanks were also used preferentially for blade tools.



made on blade blanks and to have awide range of sizes. These bladeblanks were selected from awide range of the size array of potentialhornfels blanks (Fig. 8). Hornfels microliths range up to 53 mm, buteven at this size are retouched using the samemethods, to the sameextent and into the sameshapes as smallerexamples. Large examplesofmicrolithmorphologies are alsoobservedonvarious rawmaterialsat the site of Klasies River (Villa et al., 2010). In contrast, quartz mi-croliths tend to be made on flake blanks and were preferentiallymadeonsmallerflake sizeswithin the rangeofpotential quartzblanksizes (Fig. 8). One potential conclusion is that there was reducedpressure on the size of haftablemicrolithsmadeonhornfels, possiblybecause a range of sizeswas useful for the task/s forwhich theywereemployed. Microliths made on quartz may have been used for adifferent purpose, for which a particular size was most useful.

In terms of tool typology, the non-microlithic tool sample issmall, but includes a unifacial point and a bifacial point, consideredtypical of MSA assemblages (previously noted by Kaplan, 1990;Lombard et al., 2010). Within the microlith assemblage there is adominance of backed pointed and crescent microliths (40.9% and29.8%, respectively). Other geometric shapes, particularly trapezes,are also relatively common, and geometric morphologies make up38.7% of the microlith assemblage overall (Table 7). With referenceto the debate concerning which layers should be attributed to theHP at the site, it should be noted that microliths are foundthroughout the sampled layers.

In addition to size variations between hornfels and quartz mi-croliths (Fig. 8), as discussed above, there are also morphologicaldifferences. Hornfels was primarily used for backed pointed micro-liths (54.2% of hornfels microliths compared to 31.6% of quartz mi-croliths), and quartz was more frequently used to manufacturegeometricmicroliths, especially crescents (38.9% of quartzmicrolithsare crescents compared to 16.9% of hornfels microliths) as well astrapezes and triangles, which are both more common on quartz.Wadley andMohapi (2008) note a similar pattern in rawmaterial useat thenearbysite of SibuduCave, i.e. that there arediscretemicrolithicsize categories for separate rock types, including quartz and hornfels,and that these may also have had discrete functions. The evidencediscussed here from the site of Umhlatuzana provides further evi-dence that the HP was not a monolithic entity, and that variability inmicrolith production strategies exists evenwithin individual sites.

6.3. Batadomba-lena

6.3.1. Raw materialsThe vast majority (98.5%) of the lithic assemblage is quartz of

varying quality, which is locally available in large quantities, with

le 9a. Microlith and core numbers are absolute numbers from the layers and squares

lith on flake Retouched blade Retouched flake Biface Core Total

6 1 1 6 4430 0 0 75 4360 0 0 58 2011 2 0 6 1770 0 0 6 920 0 0 1 420 0 0 1 301 0 0 0 291 0 0 0 280 0 0 0 250 0 0 5 171 0 0 2 110 0 0 0 90 0 0 2 3

10 3 1 162 1543


Fig. 8. Lengths of hornfels and quartz debitage and microliths at Umhlatuzana,demonstrating the different size profiles for the two raw materials. All microliths areincluded due to the difficulty in discerning whether a tool was intentionally manu-factured on a broken blank, or whether the tool was later broken. Only intact debitageis included.


1.5% chert, which is not (Fig. 9, Table 10). The source of the chert isnot currently known. Although chert is found in both the core anddebitage samples, there is no evidence of the retouching of chertpieces. However, given the small size of the chert sample it is notpossible to draw further conclusions at this point.

As will be demonstrated below, there was a strong degree ofselection for tool blanks of finer quartz qualities of quartz, i.e. clearquartz (rock crystal) and impure clear quartz (rock crystal withminor internal fractures or discolouration) (Fig. 9). Quartz nodulesat the site, washed in by a river at the front of the rockshelter, often

Fig. 9. Raw material proportions of major lithic categories at Batadomba-lena, indi-cating the selection of flakes and blades made from the finer qualities of quartz asblanks for the production of both microlithic and non-microlithic tools.


consist of a variety of quartz qualities, with the clearest quartzgenerally towards the centre. Debitage of coarser quartz qualitieswas most likely made during the extraction of higher quality quartzfrom the internal parts of these nodules, and was therefore unlikelyto be intended for use as blanks.

6.3.2. CoresA total of 21 cores was analysed from Batadomba-lena (Table 3).

The small number of cores recovered is most likely the result of thesmashing of small quartz cores into small blocky pieces of quartz‘shatter’ that havenodiscernible ventral ordorsal faces, andwerenotincluded in this study as they tell us little about core reductionprocesses and are not present at the other sites studied. Bipolar coresare the most common type, followed by amorphous or polyhedralcores, which is neither particularly surprising given the quartz-dominant nature of the assemblage (Driscoll, 2011), nor particu-larly revealing given the small population size. Bipolar cores wereprimarilyused for the removal offlakes, andonly two cores exhibitedgreater numbers of terminal blade removals than flake removals.

Uni-directional blade and bipolar cores were the most inten-sively worked core types (Table 6). Preferential working of finerqualities of quartz can also be seen in the degree of core exhaustion.Discarded cores of high quality quartz (clear, impure clear andcoloured quartz) had amode of 8 removal scars, with amean lengthof 11.9 mm (SD 4.06), while those of low quality quartz (granular,opaque and milky/cloudy varieties) most commonly bore 4removal scars, with a mean length of 18.2 mm (SD 4.86). Betterqualities of quartz (or quartz cores containing sections of higherquality quartz) were worked for longer, with more flakes removeduntil the core was fully exhausted.

6.3.3. DebitageThe lithic industry at Batadomba-lena is almost entirely flake

based (75.9% of debitage), with true blades uncommon (7.7%) andflake-blades almost as scarce (16.4%). 91.0% of the latter categoryare squat blades, indicating some laminar working of cores, albeitnot one involving a particular degree of elongation of the flakesremoved.

Higher qualities of quartz were preferred for the production ofblades. However, although the relationship is statistically signifi-cant at the 0.05 level it is of negligible strength (c2 ¼ 14.161,p ¼ 0.003, Cram�er's V ¼ 0.098). Therefore it appears that bladeswere probably not produced as part of any special process at thesite, at least in terms of raw material choices or stages within thereduction process. Overall, the debitage assemblage is very small,with mean dimensions for intact debitage of 17.5 � 14.0 � 4.8 mm.There is no significant degree of patterning in size between quali-ties of quartz or between flakes and blades, further indicating that asingle main reduction strategy was utilised at the site.

6.3.4. ToolsA total of 155 microliths and 8 non-microlithic tools was ana-

lysed (Table 10). Blanks for retouching into microliths or other toolswere intentionally selected from the better qualities of quartz. Forexample, 87.1% of microliths and 50.0% of non-microlithic toolswere made on clear quartz blanks, whereas these formed only34.8% of the debitage sample (Table 10, Fig. 9). Additionally, bladeswere selected disproportionately for tool blanks. 51.6% of microlithsand 37.5% of non-microlithic tools were made on blade blanks,whereas these form only 7.7% of the debitage assemblage. Althoughblades do not appear to have been produced as part of a distinctmanufacturing process at the site (discussed above), they weremore likely to be selected for further retouch.

Non-microlithic tools are very rare at the site. None of those inthe current sample are in an intact condition, problematising a


Fig. 10. Volumes of quartz debitage and microliths at Umhlatuzana and Batadomba-lena, demonstrating the different size profiles of lithics made from the same rawmaterial at both sites. All microliths are included, plus intact debitage.

Table 10Raw material frequencies in each major lithic category in the sample from Batadomba-lena. Microlith and core numbers are absolute numbers from the layers analysed. Allother numbers are sample sizes.

Raw material Blade Flake-blade Flake Tabular piece Microlith on blade Microlith on flake Retouched blade Retouched flake Core Total

Clear quartz 58 82 381 0 74 61 1 3 3 663Impure clear quartz 17 65 308 0 6 13 1 1 2 413Milky quartz 30 64 283 0 0 1 1 1 7 387Opaque quartz 7 24 114 1 0 0 0 0 8 154Granular quartz 1 5 19 7 0 0 0 0 0 32Chert 2 5 17 0 0 0 0 0 1 25Smokey quartz 0 0 5 0 0 0 0 0 0 5Rose quartz 0 0 1 0 0 0 0 0 0 1Total 115 245 1128 8 80 75 3 5 21 1680


clear understanding of their size profiles in relation to bothmicrolithic tools and the debitage sample. Non-microlithic toolswere produced in small numbers and most likely opportunisticallyat the site. Overall, it appears that lithic reduction at Batadomba-lena was conducted with the sole aim of producing microlithictools (setting aside any potential uses of non-retouched lithics, e.g.Elston and Brantingham, 2002; Lombard and Parsons, 2008).

Awide variety of microlith formswere produced at the site, withbacked pointed and crescent microliths dominant (33.5% and 18.1%,respectively) (Table 7). Many other microlith types are found insmall numbers, including several varieties of microliths retouchedseparately along both lateral margins, as well as various geometricand non-geometric forms sometimes exhibiting notches, shouldersor tangs (Fig. 6C). Overall, geometric morphologies make up 25.2%of the microlith assemblage. One example was also found of thetanged and bifacially worked ‘Balangoda point’.

The average size of microliths at the site is very small (Table 8).For example, the smallest intact microlith measures only7.4 � 3.3 � 1.8 mm, yet was also backed along most of the extent ofboth lateral edges. However, as previously noted by Perera et al.(2011), the small size of microliths accords with the small size ofdebitage produced at the site (Fig. 10). As will be discussed furtherbelow in comparison to Umhlatuzana, rather than intentionallyselecting only the smallest potential tool blanks for retouch intomicroliths, the entire system of lithic reduction at Batadomba-lenawas geared towards the production of very small flakes and blades.This had the effect of widening the base population of lithics fromwhich blanks could be selected, thus providing a greater range ofblank choices.

7. Comparison of microlithic assemblages

Each of the lithic assemblages from the three sites describedabove is unique, and important areas of difference can be seenwhen they are compared. One major area of difference is in flakeand blade proportions, including within the HP. Whereas on thebasis of the lithic assemblage from Rose Cottage Cave alone the HPmight be considered primarily a blade industry, the evidence fromUmhlatuzana suggests that this is not always the case and thatgeographic variation exists within the industry. This can be seen incore and debitage proportions, as well as in blank selection. Addi-tionally, it appears that blades were the desired outcome of thereduction process at Rose Cottage Cave, with flakes removed pri-marily in order to prepare cores for the production of blades. AtUmhlatuzana, on the other hand, a mix of flakes and blades wereproduced and selected as tool blanks, and raw material appears tohave been a major factor in choosing between the two.

Batadomba-lena is heavily flake-based compared to the HP sites.Here, blades do not appear to have been produced as part of adistinct or separate process during core reduction. There is alsolittle discrimination in the selection of flake or blade blanks for


retouch. For example, microlith blanks are approximately evenlysplit between flakes and blades. However, because blades make upa small minority of the debitage produced, blades were selecteddisproportionately out of the available population of potentialblanks. Possibly unretouched flakes were utilised in other ways,including as parts of other composite tools (e.g. Elston andBrantingham, 2002; Lombard and Parsons, 2008), although res-idue and use-wear analysis would be required to investigate thisfurther. Regardless, different selection pressures for choices of toolblankwere in operation at the three sites, resulting in differences inboth flake and blade production and in their selection for retouch.

Related to the issue of blade v. flake production is the diversityobserved in the use of bipolar cores. At Rose Cottage Cave, bipolarcores were primarily used for the production of blades (as far as canbe discerned from terminal removals alone), whereas at the othertwo sites they were mostly used for the removal of flakes. Some ofthis variationmay be attributed to the general preference for bladesat Rose Cottage Cave, as well as the fact that bipolar cores are CCS atRose Cottage Cave and quartz at the other sites.

Differences in the use of bipolar technology can also be observedwhen core size is taken into account, thereby reducing (althoughnot entirely negating) the effect that differences in raw materials



might have. At Rose Cottage Cave bipolar cores are the least com-mon core type, but were worked more heavily than other types toproduce large numbers of removals, especially blades (Table 5). Incontrast, at both Umhlatuzana and Batadomba-lena they are themost common core type, constituting a third of the core assem-blage at both. However, while at Umhlatuzana bipolar cores are themost heavily worked type e as at Rose Cottage Cave e those atBatadomba-lena are not significantly more heavily worked thanother types, and have similar measures of reduction intensity toother uni-directional cores at the site. At Batadomba-lena, bipolarcores were used to remove relatively small numbers of relativelysmall flakes, compared to the HP sites. In summary, bipolar tech-nology was made use of in different ways at each of the three sitesin order to produce what would otherwise be considered to be thesame form of microlithic technology if only tool typology wasconsidered. This aspect of variability can only be illuminatedthrough the use of a technological attribute approach to microlithictechnology.

Yet even tool typology varies between sites, despite the fact thatthis is the areawhere previous comparative work on microliths hastended to seek out similarities, viewing tool forms in variousmicrolithic industries as largely indistinguishable (e.g. Mellars,2006a; Mellars et al., 2013). On the most basic level, we demon-strate here that geometric microliths (including various forms ofcrescents, rectangles, trapezes and triangles) do not in fact make upthe majority of the microlithic assemblages at any of the sitesstudied, contrary to conventional assumptions. Non-geometricmicroliths are frequently overlooked in the literature, yet non-geometric backed and pointed microliths are the most commonmorphology at all three sites.

Moreover, proportions of geometric microliths vary consider-ably between sites, especially within the HP. A quarter (25.2%) ofmicroliths at Batadomba-lena are classified as geometric. AtUmhlatuzana this proportion is 38.7%, but at Rose Cottage Cave it isonly 18.7%. Additionally, proportions of ‘classic’ HP ‘crescentic’stone tools are 29.8% and 11.9%, respectively, and triangles aretotally absent from the latter. There appears to be considerablevariation in microlith typology between the two HP sites, whichaccords with previous findings at Rose Cottage Cave (Wadley andHarper, 1989; Wadley and Mohapi, 2008).

Differences can also be observed between the HP sites andBatadomba-lena in this regard. Although backed pointedmicrolithsare the most common microlithic type at Batadomba-lena as atRose Cottage Cave and Umhlatuzana, followed by crescent micro-liths as at Umhlatuzana, there are a wider variety of microlithictypes at Batadomba-lena. While the two most common types atUmhlatuzana account for 68.5% of the microlith assemblage, andthose at Rose Cottage Cave account for 53.7%, at Batadomba-lenathey account for only 41.9%.

Additionally, eight other microlith morphologies occur atBatadomba-lena but are not found at the HP sites (Table 7). Incomparison, two microlithic forms are unique to Rose CottageCave, none at Umhlatuzana, and four are found only at the HPsites. Unique microlithic forms at Batadomba-lena include severalvarieties of microliths retouched separately along both lateralmargins, notched, shouldered and tanged microliths, and abifacially-worked point (‘Balangoda’ point). The shoulder and tanginnovations may be related to the functional requirements ofhafting such small microliths (Table 8). It could be furtherhypothesised that the small dimensions of the microliths atBatadomba-lena were an adaptation to the hunting of arborealprey in a dense rainforest environment (Perera et al., 2011), whichwould provide a further example of innovations related to localcircumstances or requirements. This hypothesis requires testingthrough residue analysis.


Alternatively, differences in microlith size may be at least partlyrelated to differences in the availability of raw materials. Forexample, quartz can be very difficult to knap into larger flakes as ithas a tendency to shatter along internal fissures (Driscoll, 2011),whichmight explainwhy themicrolithic tools at quartz-dominatedBatadomba-lena are so small compared to the other sites. If thiswere the sole explanation, the quartz lithics at Umhlatuzana shouldbe equally small. However, quartz debitage at Umhlatuzana islarger on average, with mean dimensions of intact lithics of21.0 � 16.1 � 6.1 mm compared to 17.4 � 14.0 � 4.7 mm atBatadomba-lena. The size profiles at the two sites are very different,with both debitage and microliths being more variable in size atUmhlatuzana (Fig. 10). At Batadomba-lena, in contrast, the sizeprofiles of both debitage andmicroliths are less diverse, and closelymatch each other. The same raw material was being used indifferent ways to manufacture a microlithic toolkit at the two sites,indicating that factors other than raw material were influencingmicrolith size.

Finally, the non-microlithic tool assemblages also reveal areas ofdifference between the sites. Non-microlithic tools are much morecommon at Rose Cottage Cave, although the desired outcome of thelithic reduction sequence still appears to be the production of smallblades suitable for retouch into microliths, and much of the pro-duction of non-microlithic tools may have made use of thebyproducts of microlith-oriented core reduction. Nevertheless, it isinteresting that unifacial and bifacial points were produced intandem with microlithic tools at the HP sites (which has also beenhighlighted at Klasies River by Wurz (1999)). These points aretypically considered part of a very different MSA tool productionstrategy to the production of microliths. In contrast, the lithicmanufacturing system at Batadomba-lena was entirely aimed atthe production of small microlithic tools, with no clear focus on anyspecific non-microlithic tool forms.

Overall, there are indications of different chaîne op�eratoires bothwithin and between sites. At Rose Cottage Cave one main reductiontrajectory was utilised in order to produce small blades suitable forretouch into microliths, especially on CCS, with the addition of asecond process of selection of larger flakes and blades for retouchinto non-microlithic tools, including different working of non-CCSmaterials. Two distinct trajectories exist at Umhlatuzana, withhornfels primarily used to produce blades and non-geometric mi-croliths, and quartz preferred for the production of flakes andgeometric microliths. Only one reduction process was used atBatadomba-lena, where quartz was knapped more or less for thesole production of microliths. The quartz utilised here was alsoused in a different manner to that at Umhlatuzana. The samecategory of tool emicroliths ewas manufactured in different waysat the various sites, using different selection and manufacturingstrategies to produce a variety of different microlithic formsdepending on the needs and requirements of people in differentecological and social circumstances, in three distinct lithictraditions.

8. Discussion and conclusions

Microlith production is an important and distinctive lithictechnology, but has tended to be homogenised in previouscomparative studies. By applying detailed technological attributeanalyses to our understanding of microlithic technology, we canbetter investigate past technological and behavioural variabilityand adaptive strategies. This is a different approach to that taken inprevious inter-regional comparative studies of microlithic in-dustries (e.g. Mellars, 2006a; Mellars et al., 2013). Our initial resultsfrom comparisons of microlithic assemblages at the sites of RoseCottage Cave and Umhlatuzana, South Africa, and Batadomba-lena,



Sri Lanka, indicate potentially fruitful avenues of enquiry con-cerning differences in microlithic technologies and typologies.

We have demonstrated that considerable typological and tech-nological differences can exist between HP sites. For example, theHP is not always a blade-based industry, as is often assumed (e.g.Lombard, 2005; Henshilwood, 2012). The co-existence of what areconsidered to be typically MSA features (radial cores and unifacialand bifacial points) and bipolar percussion, which is often consid-ered to be indicative of LSA lithic technology, has been highlighted.Similarly, there are hints of different chaîne op�eratoires within theHP sites, with the main factors being the use of different raw ma-terials (hornfels v. quartz at Umhlatuzana, and CCS v. non-CCS atRose Cottage Cave), the choice of flake or blade blanks, and stra-tegies concerning the production of microlithic v. non-microlithictools. This discovery accords with similar findings at Sibudu Caveby Wadley and Mohapi (2008) with reference to microlith rawmaterial choices. As an additional note concerning Umhlatuzana,evidence of HP microlithic technology in Layer 27 has been pro-vided to support the claimmade here that the HP extends to at leastthis point in the sequence, dated to 70.5 ± 4.7 ka, contra Lombardet al. (2010).

In the first comparative study of its type, we have also high-lighted some of the areas of difference between microlithic in-dustries in southern Africa and those in South Asia. For example, ithas been shown that there are differences even in the use of thesame raw material to produce the same tool type, as evidenced inthe distinct size profiles of quartz debitage and microliths atUmhlatuzana and Batadomba-lena. The current analysis has alsohighlighted differences between micro-blade (dominated by smallunretouched blades) and microlithic (dominated by smallretouched flake or blade tools) industries. Micro-blades are oftenviewed as a necessary requirement for the production of micro-lithic tools (e.g. Mishra et al., 2013). However, micro-blades arerelatively rare in the microlithic assemblage at Batadomba-lena e

especially in comparison to the HP sites e and microliths weremanufactured on a mix of flake and blade blanks. Such differencesare important in understanding lithic production processes.

The dispersal model proposed by Mellars (2006a, Mellars et al.,2013) has not been comprehensively tested at this stage. However,the data examined in this study disprove a key part of this hy-pothesis, i.e. that “both the African and the Indian industries aredominated by a range of carefully shaped microlithic or larger‘backed-segment’ forms of precisely the same range of shapes asthose documented in the South Asian industries” (Mellars et al.,2013, pp. 10702). Not only have geometric shapes such as seg-ments shown to be in the minority at all of the sites analysed here,but proportions of microlith morphologies vary considerably be-tween sites.

With reference to the ‘modern human behaviour’ debate, thepresent study contributes to its methodology by demonstrating thebenefits of using a technological attribute approach to our under-standing of technological and behavioural variability in the past. Byshifting the unit of analysis from the presence or absence of mi-croliths at specific sites to the actual microliths and their associatedby-products, areas of variability are revealed that would otherwisebe overlooked. For example, different uses of raw materials and ofbipolar percussion, issues of blank selection with regards to mi-croliths as well as non-microlithic tools, and a more nuancedconsideration of variation in microlith morphology are all madepossible by a technology-based comparison of microlithicassemblages.

Previous studies of both southern African HP and South AsianMicrolithic industries indicate areas of inter-industry variation, forexample in tool morphology and hafting techniques in the HP, andin flake and blade preferences in the South Asian Microlithic. The


current study has endeavoured to highlight and clarify aspects ofintra- and inter-site variability in microlithic technology. The smallnumber of sites discussed here does not of course represent the fullrange of variability we should expect to see in either of these in-dustries. The focus of future more in-depth analyses will be theplacing of these assemblages into their environmental, de-mographic and socio-cultural contexts. By examining the variabilitywithin lithic industries, and considering microliths as regional ad-aptations in their own right, we can gain a much better idea of thelifeways of the people that produced this highly variable andnuanced technological strategy.

Acknowledgements

We are grateful to the Department of Archaeology, Sri Lanka, forpermission to study the materials from Batadomba-lena. Weparticularly wish to thank the Director General Senarath Dis-sanayake, Siran Deraniyagala and Oshan Manjula for all their help.Access to the South African materials was provided by the Uni-versity of the Witwatersrand, Johannesburg, and the KwaZulu-Natal Museum, Pietermaritzburg, with many thanks to PeterMitchell, Lyn Wadley, Karim Sadr, Amanda Esterhuysen, NicciSherwood, Carolyn Thorp and Gavin Whitelaw. We also wish tothank the four reviewers who helped improve and clarify thismanuscript. LL was funded by the Arts and Humanities ResearchCouncil (award no. 513691), the Wenner-Gren Foundation (grantno. 8684), The Boise Fund, the Faith Ivens-Franklin Travel Fund, theMeyerstein Fund, the Keble Graduate Study Support Fund and theKeble Association. MP is grateful to the European Research Councilfor support (grant no. 295719).

References

Abeyratne, M., Spooner, N.A., Grun, R., Head, J., 1997. Multidating studies of Bata-domba cave, Sri Lanka. Quaternary Science Reviews 16, 243e255.

Agrawal, D.P., Krishnamurthy, R.V., Kusumgar, S., 1985. Physical research laboratoryradiocarbon date list V. Radiocarbon 27, 95e110.

Agresti, A., Finlay, B., 2009. Statistical Methods for the Social Sciences, fourth ed.Pearson.

Ambrose, S.H., 1998. Chronology of the Later Stone Age and food production in EastAfrica. Journal of Archaeological Science 25, 377e392.

Ambrose, S.H., 2002. Small things remembered: origins of early microlithic in-dustries in Sub-Saharan Africa. Archeological Papers of the American Anthro-pological Association 12 (1), 9e29.

Ambrose, S.H., Lorenz, K.G., 1990. Social and ecological models for the Middle StoneAge in southern Africa. In: Mellars, P. (Ed.), The Emergence of Modern Humans:an Archaeological Perspective. Cornell University Press, Ithaca, pp. 3e33.

Armitage, S.J., Jasim, S.A., Marks, A.E., Parker, A.G., Usik, V.I., Uerpmann, H.P., 2011.The southern route ‘out of Africa’: evidence for an early expansion of modernhumans into Arabia. Science 331 (6016), 453e456.

Atkinson, Q.D., Gray, R.D., Drummond, A.J., 2008. mtDNA variation predicts popu-lation size in humans and reveals a major southern Asian chapter in humanprehistory. Molecular Biology and Evolution 25 (2), 468e474.

Ballin, T.B., 2000. Classification and description of lithic artefacts: a discussion of thebasic lithic terminology. Lithics 21, 9e15.

Balme, J., Davidson, I., McDonald, J., Stern, N., Veth, P., 2009. Symbolic behaviour andthe peopling of the southern arc route to Australia. Quaternary International202 (1), 59e68.

Bar-Yosef, O., 2002. The Upper Paleolithic revolution. Annual Review of Anthro-pology 31, 363e393.

Boivin, N., Fuller, D.Q., Dennell, R., Allaby, R., Petraglia, M.D., 2013. Human dispersalacross diverse environments of Asia during the Upper Pleistocene. QuaternaryInternational 300, 32e47.

Bowler, J.M., Johnston, H., Olley, J.M., Prescott, J.R., Roberts, R.G., Shawcross, W.,Spooner, N.A., 2003. New ages for human occupation and climatic change atLake Mungo, Australia. Nature 421 (6925), 837e840.

Brumm, A., Moore, M.W., 2005. Symbolic revolutions and the Australian archaeo-logical record. Cambridge Archaeological Journal 15 (2), 157e175.

Bulbeck, D., 2007. Where river meets sea: a parsimonious model for Homo sapienscolonization of the Indian Ocean Rim and Sahul. Current Anthropology 48 (2),315e321.

Charri�e-Duhaut, A., Porraz, G., Cartwright, C.R., Igreja, M., Connan, J., Poggenpoel, C.,Texier, P., 2013. First molecular identification of a hafting adhesive in the LateHowiesons Poort at Diepkloof Rock Shelter (Western Cape, South Africa).Journal of Archaeological Science 40, 3506e3518.


http://refhub.elsevier.com/S1040-6182(14)00660-0/sref1



























































Clark, J.D., 1959. The Prehistory of Southern Africa. Penguin Press, Harmondsworth.Clark, J.D., 1985. The microlithic industries of Africa: their antiquity and possible

economic implications. In: Misra, V.N., Bellwood, P.S. (Eds.), Recent Advances inIndo-Pacific Prehistory: Proceedings of the International Symposium Held atPoona, December 19-21, 1978. Oxford & IBH Publishing Co., New Delhi,pp. 95e103.

Clarkson, C., 2013. Regional diversity within the core technology of the HowiesonsPoort Techno-Complex. In: Lycett, S., Chauhan, P. (Eds.), New Perspectives onOld Stones: Analytical Approaches to Palaeolithic Technologies. Springer Press,New York, pp. 43e59.

Clarkson, C., Petraglia, M.D., Korisettar, R., Haslam, M., Boivin, N., Crowther, A.,Ditchfield, P., Fuller, D., Miracle, P., Harris, C., Connell, K., James, H., Koshy, J.,2009. The oldest and longest enduring microlithic sequence in India: 35,000years of modern human occupation and change at the Jwalapuram Locality 9rockshelter. Antiquity 83 (320), 326e348.

Close, A.E., 2002. Backed bladelets are a foreign country. Archeological Papers of theAmerican Anthropological Association 12 (1), 31e44.

Cochrane, G.W.G., Doelman, T., Wadley, L., 2013. Another dating revolution forprehistoric archaeology? Journal of Archaeological Method and Theory 20,42e60.

Deacon, H.J., 1989. Late Pleistocene palaeoecology and archaeology in the southernCape, South Africa. In: Mellars, P.A., Stringer, C.B. (Eds.), The Human Revolution:Behavioural and Biological Perspectives on the Origin of Modern Humans.Edinburgh University Press, Edinburgh, pp. 547e564.

Deacon, H.J., 1992. Southern Africa and modern human origins. PhilosophicalTransactions of the Royal Society of London. Series B: Biological Sciences 337(1280), 177e183.

Deacon, H.J., 1995. Two Late Pleistocene-Holocene archaeological depositories fromthe Southern Cape, South Africa. The South African Archaeological Bulletin 50(162), 121e131.

Dennell, R.W., Petraglia, M.D., 2012. The dispersal of Homo sapiens across southernAsia: how early, how often, how complex? Quaternary Science Reviews 47,15e22.

Dennell, R.W., Rendell, H.M., Halim, M., Moth, E., 1992. A 45,000-year-old open-airPaleolithic site at Riwat, northern Pakistan. Journal of Field Archaeology 19 (1),17e33.

Deraniyagala, P.E.P., 1940. The Stone Age and cave men of Ceylon. Part I. Journal ofthe Royal Asiatic Society, Ceylon Branch 34 (92), 351e373.

Deraniyagala, P.E.P., 1963. Prehistoric archaeology in Ceylon. Asian Perspectives 7(1e2), 189e192.

Deraniyagala, S.U., 1984. Mesolithic stone tool technology at 28,000 BP in Sri Lanka.Ancient Ceylon 5, 105e108.

Deraniyagala, S.U., 1986. Pleistocene coastal sediments in the Dry Zone of Sri Lanka:chronology, palaeo-environment and technology. Ancient Ceylon 6, 49e62.

Deraniyagala, S.U., 1988. The Prehistory of Sri Lanka: an Ecological Perspective (PhDthesis). Harvard University Microfilm, publication no. 8820579, Ann Arbor.

Deraniyagala, S.U., 1992. The Prehistory of Sri Lanka: an Ecological Perspective.Department of Archaeological Survey, Government of Sri Lanka, Colombo.

Deraniyagala, S.U., 2007. The prehistory and protohistory of Sri Lanka. In:Prematilleka, P.L., Bandaranayake, S., Deraniyagala, S.U., Silva, R. (Eds.), The Artand Archaeology of Sri Lanka, Part I. Central Cultural Fund, Colombo, pp. 1e96.

Driscoll, K., 2011. Vein quartz in lithic traditions: an analysis based on experimentalarchaeology. Journal of Archaeological Science 38, 734e745.

Elston, R.G., Brantingham, P.J., 2002. Microlithic technology in northern Asia: a risk-minimizing strategy of the Late Paleolithic and Early Holocene. ArcheologicalPapers of the American Anthropological Association 12 (1), 103e116.

Elston, R., Kuhn, S. (Eds.), 2002. Thinking Small: Global Perspectives on Micro-lithization. American Anthropological Association, Arlington, VA.

Eren, M.I., Diez-Martin, F., Dominguez-Rodrigo, M., 2013. An empirical test of therelative frequency of bipolar reduction in Beds VI, V, and III at Mumba Rock-shelter, Tanzania: implications for the East African Middle to Late Stone Agetransition. Journal of Archaeological Science 40 (1), 248e256.

Field, J.S., Lahr, M.M., 2005. Assessment of the southern dispersal: GIS-based ana-lyses of potential routes at Oxygen Isotopic Stage 4. Journal of World Prehistory19 (1), 1e45.

Field, J.S., Petraglia, M.D., Lahr, M.M., 2007. The southern dispersal hypothesis andthe South Asian archaeological record: examination of dispersal routes throughGIS analysis. Journal of Anthropological Archaeology 26, 88e108.

Franklin, N.R., Habgood, P.J., 2007. Modern human behaviour and Pleistocene Sahulin review. Australian Archaeology 65, 1e16.

Gibson, N.E., Wadley, L., Williamson, B.S., 2004. Microscopic residues as evi-dence of hafting on backed tools from the 60,000 to 68,000 year-oldHowiesons Poort layers of Rose Cottage Cave, South Africa. Southern Afri-can Humanities 16, 1e11.

Glover, I.C., Lampert, R.J., 1969. Puntutjarpa rockshelter excavations by R.A. Gould: acritical review. Archaeology and Physical Anthropology in Oceania 4 (3),222e228.

Gould, R.A., 1969. Puntutjarpa rockshelter: a reply to Messrs. Glover and Lamper.Archaeology and Physical Anthropology in Oceania 4 (3), 229e237.

Guadagnoli, E., Velicer, W.F., 1988. Relation of sample size to the stability ofcomponent patterns. Psychological Bulletin 103 (2), 265e275.

Gu�erin, G., Murray, A.S., Jain, M., Thomsen, K.J., Mercier, N., 2013. How confident arewe in the chronology of the transition between Howieson's Poort and Still Bay.Journal of Human Evolution 64, 314e317.


Habgood, P.J., Franklin, N.R., 2008. The revolution that didn't arrive: a review ofPleistocene Sahul. Journal of Human Evolution 55 (2), 187e222.

Harper, P., 1997. The Middle Stone Age sequences at Rose Cottage Cave: a search forcontinuity and discontinuity. South African Journal of Science 93 (10), 470e475.

Henshilwood, C.S., 2012. Late Pleistocene techno-traditions in Southern Africa: areview of the Still Bay and Howiesons Poort, c. 75-59 ka. Journal of WorldPrehistory 25, 205e237.

Hiscock, P., 2002. Pattern and context in the Holocene proliferation of backed ar-tifacts in Australia. Archeological Papers of the American Anthropological As-sociation 12 (1), 163e177.

Hiscock, P., Clarkson, C., Mackay, A., 2011. Big debates over little tools: ongoingdisputes over microliths on three continents. World Archaeology 43 (4),660e671.

Hole, F., 1959. A critical analysis of the Magosian. The South African ArchaeologicalBulletin 14 (56), 126e134.

Jacobs, Z., Roberts, R.G., Galbraith, R.F., Deacon, H.J., Grün, R., Mackay, A., Mitchell, P.,Vogelsang, R., Wadley, L., 2008. Ages for the Middle Stone Age of southernAfrica: implications for human behavior and dispersal. Science 322 (5902),733e735.

James, H.V.A., 2011. Becoming Human: the Emergence of Modern Human BehaviourWithin South Asia (Unpublished PhD thesis). University of Cambridge, UK.

James, H.V.A., Petraglia, M.D., 2005. Modern human origins and the evolution ofbehavior in the later Pleistocene record of South Asia. Current Anthropology 46,S3eS27.

Kaplan, J., 1989. 45,000 years of hunter-gatherer history in Natal as seen fromUmhlatuzana Rock Shelter. South African Archaeological Society Goodwin Se-ries 6, 7e16.

Kaplan, J., 1990. The Umhlatuzana Rock Shelter sequence: 100,000 years of StoneAge history. Natal Museum Journal of Humanities 2, 1e94.

Kennedy, K.A.R., 2000. God-Apes and Fossil Men. University of Michigan Press, AnnArbor.

Kennedy, K.A.R., Deraniyagala, S.U., 1989. Fossil remains of 28,000-year-old homi-nids from Sri Lanka. Current Anthropology 30 (3), 394e399.

Kennedy, K.A.R., Disotell, T., Roertgen, W.J., Chiment, J., Sherry, J., 1986. Biologicalanthropology of Upper Pleistocene hominids from Sri Lanka: Batadomba Lenaand Beli Lena caves. Ancient Ceylon 6, 165e265.

Kennedy, K.A.R., Deraniyagala, S.U., Roertgen, W.J., Chiment, J., Disotell, T., 1987.Upper Pleistocene fossil hominids from Sri Lanka. American Journal of PhysicalAnthropology 72 (4), 441e461.

Klein, R.G., 1992. The archeology of modern human origins. Evolutionary Anthro-pology 1 (1), 5e14.

Klein, R.G., 1995. Anatomy, behavior, and modern human origins. Journal of WorldPrehistory 9 (2), 167e198.

Knight, J., 1991. Vein quartz. Lithics 12, 37e56.Korisettar, R., 2007. Toward developing a basin model for Palaeolithic settlement of

the Indian subcontinent: geodynamics, monsoon dynamics, habitat diversityand dispersal routes. In: Petraglia, M.D., Allchin, B. (Eds.), The Evolution andHistory of Human Populations in South Asia: Inter-disciplinary Studies inArchaeology, Biological Anthropology, Linguistics and Genetics. Springer, Dor-drecht, pp. 69e96.

Krejcie, R.V., Morgan, D.W., 1970. Determining sample size for research activities.Educational and Psychological Measurement 30, 607e610.

Leplongeon, A., 2014. Microliths in the Middle and Later Stone Age of eastern Africa:new data from Porc-Epic and Goda Buticha cave sites, Ethiopia. QuaternaryInternational 343, 100e116.

Lewis-Williams, J.D., Pearce, D.G., 2004. San Spirituality: Roots, Expressions andSocial Consequences. Double Storey Books, Cape Town.

Lombard, M., 2005. The Howiesons Poort of South Africa: what we know, what wethink we know, what we need to know. Southern African Humanities 17,33e55.

Lombard, M., 2006. Direct evidence for the use of ochre in the hafting technology ofMiddle Stone Age tools from Sibudu Cave. Southern African Humanities 18 (1),57e67.

Lombard, M., 2007. The gripping nature of ochre: the association of ochre withHowiesons Poort adhesives and Later Stone Age mastics from South Africa.Journal of Human Evolution 53, 406e419.

Lombard, M., 2008. Finding resolution for the Howiesons Poort through the mi-croscope: micro-residue analysis of segments from Sibudu Cave, South Africa.Journal of Archaeological Science 35, 26e41.

Lombard, M., Pargeter, J., 2008. Hunting with Howiesons Poort segments: pilotexperimental study and the functional interpretation of archaeological tools.Journal of Archaeological Science 35 (9), 2523e2531.

Lombard, M., Parsons, I., 2008. Blade and bladelet function and variability in riskmanagement during the last 2000 years in the Northern Cape. South AfricanArchaeological Bulletin 63 (187), 18e27.

Lombard, M., Haidle, M.N., 2012. Thinking a bow-and-arrow set: cognitive impli-cations of Middle Stone Age bow and stone-tipped arrow technology. Cam-bridge Archaeological Journal 22 (2), 237e264.

Lombard, M., Wadley, L., Jacobs, Z., Mohapi, M., Roberts, R.G., 2010. Still Bay andserrated points from Umhlatuzana Rock Shelter, Kwazulu-Natal, South Africa.Journal of Archaeological Science 37, 1773e1784.

Lombard, M., Wadley, L., Deacon, J., Wurz, S., Parsons, I., Mohapi, M., Swart, J.,Mitchell, P., 2012. South African and Lesotho Stone Age sequence updated (I).South African Archaeological Bulletin 67, 120e144.




































































































































































































































Macaulay, V., Hill, C., Achilli, A., Rengo, C., Clarke, D., Meehan, W., Blackburn, J.,Semino, O., Scozzari, R., Cruciani, F., Taha, A., Shaari, N.K., Raja, J.M., Ismail, P.,Zainuddin, Z., Goodwin, W., Bulbeck, D., Bandelt, H., Oppenheimer, S.,Torroni, A., Richards, M., 2005. Single, rapid coastal settlement of Asia revealedby analysis of complete mitochondrial genomes. Science 308 (5724),1034e1036.

Mackay, A., 2011. Potentially stylistic differences between backed artefacts from twonearby sites occupied ~60,000 years before present in South Africa. Journal ofAnthropological Archaeology 30, 235e245.

Malan, B.D., 1952. The final phase of the Middle Stone Age in South Africa. In:Leakey, L.S.B., Cole, S. (Eds.), Proceedings of the Pan-African Congress on Pre-history, 1947. Blackwell, Oxford, pp. 188e194.

McBrearty, S., 2007. Downwith the revolution. In: Mellars, P., Boyle, K., Bar-Yosef, O.,Stringer, C. (Eds.), Rethinking the Human Revolution. McDonald Institute forArchaeological Research, Cambridge, pp. 133e151.

McBrearty, S., Brooks, A.S., 2000. The revolution that wasn't: a new interpretation ofthe origin of modern human behavior. Journal of Human Evolution 39 (5),453e563.

McCall, G.S., 2007. Behavioral ecological models of lithic technological changeduring the later Middle Stone Age of South Africa. Journal of ArchaeologicalScience 34, 1738e1751.

McCall, G.S., Thomas, J.T., 2012. Still Bay and Howiesons Poort foraging strategies:recent research and models of culture change. African Archaeological Review 29(1), 1e44.

Mellars, P. (Ed.), 1990. The Emergence of Modern Humans: an ArchaeologicalPerspective. Cornell University Press, New York.

Mellars, P., 2006a. Going East: new genetic and archaeological perspectives on themodern human colonization of Eurasia. Science 313 (5788), 796e800.

Mellars, P., 2006b. Why did modern human populations disperse from Africa ca.60,000 years ago? A new model. Proceedings of the National Academy of Sci-ences 103 (25), 9381e9386.

Mellars, P., Gorí, K.C., Carr, M., Soares, P.A., Richards, M.B., 2013. Genetic andarchaeological perspectives on the initial modern human colonization ofsouthern Asia. Proceedings of the National Academy of Sciences 110 (26),10699e10704.

Mercader, J., Brooks, A.S., 2001. Across forests and savannas: Later Stone Age as-semblages from Ituri and Semliki, Democratic Republic of Congo. Journal ofAnthropological Research 58 (2), 197e217.

Mishra, S., Chauhan, N., Singhvi, A.K., 2013. Continuity of microblade technology inthe Indian subcontinent since 45 ka: implications for the dispersal of modernhumans. PLoS One 8 (7), e69280.

Misra, V.N., 1985. The Acheulean succession at Bhimbetka, central India. In:Misra, V.N., Bellwood, P.S. (Eds.), Recent Advances in Indo-Pacific Prehistory:Proceedings of the International Symposium Held at Poona, December 19-21,1978. Oxford & IBH Publishing Co., New Delhi, pp. 35e47.

Misra, V.N., 2001. Prehistoric human colonization of India. Journal of Biosciences 26(4), 491e531.

Misra, V.N., Bellwood, P. (Eds.), 1985. Recent Advances in Indo-Pacific Prehistory.Oxford & IBH Publishing Co., New Delhi.

Mitchell, P., 2002. The Archaeology of Southern Africa. Cambridge University Press,Cambridge.

Mitchell, P.J., Steinberg, J.M., 1992. Ntloana Tsoana: a Middle Stone Age sequencefrom western Lesotho. The South African Archaeological Bulletin 47 (155),26e33.

Mohapi, M., 2013. The Middle Stone Age point assemblage from Umhlatuzana RockShelter: a morphometric study. Southern African Humanities 25, 25e51.

Neeley, M.P., 2002. Going microlithic: a Levantine perspective on the adoption ofmicrolithic technologies. Archeological Papers of the American AnthropologicalAssociation 12 (1), 45e55.

Olley, J.M., Roberts, R.G., Yoshida, H., Bowler, J.M., 2006. Single-grain optical datingof grave-infill associated with human burials at Lake Mungo, Australia. Qua-ternary Science Reviews 25, 2469e2474.

Oppenheimer, S., 2009. The great arc of dispersal of modern humans: Africa toAustralia. Quaternary International 202, 2e13.

Perera, N., 2010. Prehistoric Sri Lanka: Late Pleistocene Rockshelters and an Open-air Site. In: British Archaeological Reports International Series (Oxford).


Perera, N., Kourampas, N., Simpson, I.A., Deraniyagala, S.U., Bulbeck, D.,Kamminga, J., Perera, J., Fuller, D.Q., Szabo, K., Oliveira, N.V., 2011. People of theancient rainforest: Late Pleistocene foragers at the Batadomba-lena rockshelter,Sri Lanka. Journal of Human Evolution 61 (3), 254e269.

Petraglia, M.D., Clarkson, C., Boivin, N., Haslam, M., Korisettar, R., Chaubey, G.,Ditchfield, P., Fuller, D., James, H., Jones, S., Kivisild, T., Koshy, J., Lahr, M.M.,Metspalu, M., Roberts, R., Arnold, L., 2009. Population increase and environ-mental deterioration correspond with microlithic innovations in South Asia ca.35,000 years ago. Proceedings of the National Academy of Sciences 106 (30),12261e12266.

Petraglia, M.D., Haslam, M., Fuller, D.Q., Boivin, N., Clarkson, C., 2010. Out of Africa:new hypotheses and evidence for the dispersal of Homo sapiens along theIndian Ocean rim. Annals of Human Biology 37 (3), 288e311.

Pienaar, M., Woodborne, S., Wadley, L., 2008. Optically stimulated luminescencedating at Rose Cottage Cave. South African Journal of Science 104, 65e70.

Plug, I., Engela, R., 1992. The macrofaunal remains from recent excavations at RoseCottage Cave, Orange Free State. The South African Archaeological Bulletin 47,16e25.

Porraz, G., Texier, P., Archer, W., Piboule, M., Rigaud, J., Tribolo, C., 2013. Techno-logical successions in the Middle Stone Age sequence of Diepkloof Rock Shelter,Western Cape, South Africa. Journal of Archaeological Science 40, 3376e3400.

Sali, S.A., 1985. The Upper Palaeolithic culture at Patne, District Jalgaon, Mahara-shtra. In: Misra, V.N., Bellwood, P. (Eds.), Recent Advances in Indo-Pacific Pre-history. Oxford & IBH Publishing Co., New Delhi, pp. 137e146.

Singer, R., Wymer, J., 1982. The Middle Stone Age at Klasies River Mouth in SouthAfrica. University of Chicago Press, Chicago.

Soriano, S., Villa, P., Wadley, L., 2007. Blade technology and tool forms in the MiddleStone Age of South Africa: the Howiesons Poort and post-Howiesons Poort atRose Cottage Cave. Journal of Archaeological Science 34, 681e703.

Thackeray, A.I., Kelly, A.J., 1988. A technological and typological analysis of MiddleStone Age assemblages antecedent to the Howiesons Poort at Klasies RiverMouth. South African Archaeological Bulletin 43, 15e26.

Tribolo, C., Mercier, N., Douville, E., Joron, J.-L., Reyss, J.-L., Dufer, D., Cantin, N.,Lefrais, Y., Miller, C.E., Porraz, G., Parkington, J., Rigaud, J.-P., Texier, P.-J., 2013.OSL and TL dating of the Middle Stone Age sequence at Diepkloof Rock Shelter(South Africa): a clarification. Journal of Archaeological Science 40, 3401e3411.

Valladas, H., Wadley, L., Mercier, N., Froget, L., Tribolo, C., Reyss, J.L., Joron, J.L., 2005.Thermoluminescence dating on burnt lithics from Middle Stone Age layers atRose Cottage Cave. South African Journal of Science 101, 169e174.

Villa, P., Soriano, S., Teyssandier, N., Wurz, S., 2010. The Howiesons Poort and MSA IIIat Klasies River main site, Cave 1A. Journal of Archaeological Science 37,630e655.

Villa, P., Soriano, S., Tsanova, T., Degano, I., Higham, T.F.G., d'Errico, F., Backwell, L.,Lucejko, J.J., Colombini, M.P., Beaumont, P.B., 2012. Border Cave and thebeginning of the Later Stone Age in South Africa. Proceedings of the NationalAcademy of Sciences 109 (33), 13208e13213.

Wadley, L., 1991. Rose Cottage Cave: background and a preliminary report on therecent excavations. The South African Archaeological Bulletin 46 (154), 125e130.

Wadley, L., 1997. Rose Cottage Cave: archaeological work 1987 to 1997. South Af-rican Journal of Science 93 (10), 439e444.

Wadley, L., 2001. What is cultural modernity? A general view and a South Africanperspective from Rose Cottage Cave. Cambridge Archaeological Journal 11 (2),201e221.

Wadley, L., 2008. The Howieson's Poort industry of Sibudu Cave. South AfricanArchaeological Society Goodwin Series 10, 122e132.

Wadley, L., Harper, P., 1989. Rose Cottage Cave revisited: Malan's Middle Stone Agecollection. The South African Archaeological Bulletin 44 (149), 23e32.

Wadley, L., Mohapi, M., 2008. A segment is not a monolith: evidence from theHowiesons Poort of Sibudu, South Africa. Journal of Archaeological Science 35(9), 2594e2605.

Wurz, S.,1999. TheHowiesonsPoort backed artefacts fromKlasies River: an argumentfor symbolic behaviour. The South African Archaeological Bulletin 54, 38e50.

Wurz, S., 2013. Technological trends in the Middle Stone Age of South Africa be-tween MIS 7 and MIS 3. Current Anthropology 54 (S8), S305eS319.

Zeuner, F.F., Allchin, B., 1956. The microlithic sites of Tinnevelly District, MadrasState. Ancient India 12, 4e20.

















































































































































































First technological comparison of Southern African Howiesons Poort and South Asian Microlithic...

Documents

Transcript of First technological comparison of Southern African Howiesons Poort and South Asian Microlithic...