A comparative 3D geometric morphometric analysis of Victoria West cores:implications for the origins of Levallois technology

Stephen J. Lycett a,b,*, Noreen von Cramon-Taubadel a,b, John A.J. Gowlett b

aDepartment of Anthropology, University of Kent, Canterbury, CT2 7NR, UKbBritish Academy Centenary Research Project, SACE, University of Liverpool, Liverpool L69 3BX, UK

a r t i c l e i n f o

Article history:Received 11 September 2009Received in revised form30 November 2009Accepted 4 December 2009

Keywords:Geometric morphometricsLevalloisAcheulean handaxesVictoria WestPrepared core technology

a b s t r a c t

The ‘Victoria West’ is a Lower Paleolithic industry from South Africa, which includes prepared cores andhas previously been noted to bear strong morphological resemblances with later Middle Paleolithicprepared core technologies (i.e. Levallois cores). Indeed, from the earliest commentaries on the VictoriaWest, it has frequently been thought of as a ‘large Levallois’ variant. The hypothesis that VW cores areaccurately characterised as ‘large Levallois’ is tested here using a comparative 3D geometric morpho-metric (GM) methodology. GM methods are powerful statistical tools for shape analysis that offer manyadvantages over traditional means of shape quantification and comparison. The use of landmarks tocapture shape variation allows for the preservation of the full geometry, as well as enabling the moreprecise description of shape versus size. Moreover, biological studies have shown that the use of land-marks allows for a flexible approach to comparing specific aspects of overall morphology. Here, weemploy GM to analyse differences in core surface morphology in a range of Lower and Middle Palae-olithic artefacts, including Victoria West examples (total n¼ 639 artefacts). In comparison with coresfrom non-handaxe Mode 1, Acheulean handaxes, and Levallois cores, the Victoria West share shapeaffinities with both Acheulean handaxes and Levallois cores. However, when compared directly witha group of large Middle Palaeolithic Levallois cores from Baker’s Hole (UK), the Victoria West were foundto more closely resemble handaxes, while the Baker’s Hole set are simply isometrically-scaled Levalloiscores. These analyses show that, despite broad technological and qualitative morphological similaritieswith Levallois cores, Victoria West cores are morphologically more similar to Lower Palaeolithic artefactforms, such as handaxes, and are in some respects distinct from Middle Palaeolithic Levallois cores. Inline with other recent analyses, our results support suggestions that the Victoria West technique is anextension of longstanding Acheulean traditions for the preparation of biface blanks, but with its owndistinct characteristics.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. The Victoria West

The Victoria West, as originally defined, is a Lower Palaeolithicprepared core industry from South Africa, named after the Karoo-region town where it was originally discovered during the earlypart of the twentieth century (Smith, 1919; Van Riet Lowe, 1929;Goodwin, 1926, 1929). Cores attributed to this industry were firstdiscovered ca.1915 by the then local magistrate F.J. Jansen (seeSmith, 1919), who later wrote a short paper on his finds (Jansen,

1926). Subsequently, the term ‘Victoria West’ has been attributedto cores from a series of assemblages from sites in central SouthAfrica, especially along the Vaal River (Goodwin, 1934; Van RietLowe, 1945; Rolland, 1995; Clark, 2001; Sharon and Beaumont,2006).

From their initial discovery, cores attributed to the ‘VictoriaWest’ phenomenon drew comparisons with Middle Palaeolithic‘Levallois’ prepared cores from Europe and the African MiddleStone Age (MSA) (Smith, 1919; Van Riet Lowe, 1929; Breuil, 1930;Goodwin, 1934; Leakey, 1936). As Goodwin (1934, p. 120) noted, ina manner similar to Levallois cores, they exhibit ‘‘preparation ofupper-face and the removal of a large flake from the upper surface’’.However, due to the large size of the cores, as well as their frequentassociation with Acheulean handaxes and cleavers (manufacturedfrom their large flake products) the Victoria West was attributed tothe Lower Palaeolithic (Goodwin, 1934; Van Riet Lowe, 1945).

* Corresponding author. Department of Anthropology, University of Kent, Mar-lowe Building, Canterbury, CT2 7NR, UK. Tel.: þ44 779 11 33 593.

E-mail address: S.J.Lycett@Kent.ac.uk (S.J. Lycett).

Contents lists available at ScienceDirect

Journal of Archaeological Science

journal homepage: ht tp: / /www.elsevier .com/locate/ jas

0305-4403/$ – see front matter � 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.jas.2009.12.011

Journal of Archaeological Science 37 (2010) 1110–1117

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–1117 1111

Clark (1959, p. 125) provided an informative description of thereduction procedure resulting in the cores at Victoria West, which‘‘consisted of roughing out from a small boulder what at first sightappears to be a crude handaxe with one face much flatter than theother’’. He further describes the preparation of a striking platformon one side, and the release of the ‘‘large, comparatively flat, flake’’which was subsequently worked into ‘‘a handaxe, cleaver, or side-scraper’’ (Clark, 1959, pp. 125–126).

Despite the general comparability with Levallois cores, thisconstructional link with large Acheulean handaxes was also notedby others (Goodwin, 1934; Leakey, 1936). This similarity wasperhaps most explicitly stated by Leakey (1936, p. 85), albeit usingcertain terminology that is now obsolete: ‘‘it is decidedly noticeablethat among the forms of unstruck Victoria West cores there aremany that have more than a resemblance to large, clumsy Chellean-type hand-axes’’. However, it is their comparability with largeLevallois cores that has drawn most attention in the literature, tothe extent that they have often been regarded as a ‘proto-Levallois’core form (Breuil, 1930; Goodwin, 1934; Van Riet Lowe, 1945; Clark,1959; Bordes, 1968). As Van Riet Lowe (1929, p. 389) put it, theshape of Victoria West cores is ‘‘not unlike a magnified and slightlydistorted Levallois of Europe’’.

In some instances, direct comparisons between the SouthAfrican ‘Victoria West’ and artefacts as far away as India have beenmade (Cammiade and Burkitt, 1930). However, while it is generallyaccepted that the Levallois technique can occur anywhere, withoutany connection to the original findspot, in the case of Victoria Westthe case for reapplying an original industry label to a widespreadtechnique is not nearly so clear-cut. Indeed, Sharon (2009) hasrecently argued that the Victoria West may represent one of severaldifferent Lower Palaeolithic industries involving the production oflarge flake blanks, each of which may be convergent. Meanwhile,phylogenetic analyses have suggested that properties sharedbetween Middle Palaeolithic Levallois cores and those fromVictoria West are the product of technological convergence (Lycett,2009a).

1.2. Geometric morphometrics framework

Here, the hypothesis that Victoria West cores are accuratelycharacterised as ‘large Levallois’ is tested directly using a compar-ative 3D geometric morphometric (GM) methodology. GM (land-mark-based) methods are particularly appropriate for addressingthese questions due to the mathematical capability to separate outthe properties of ‘shape’ and ‘size’. ‘Shape’ in this context refersexplicitly to the geometric properties of an object, excluding theeffects of isometric scale (or ‘size’) (Slice, 2005). Isometric scalingrefers to the uniform magnification or reduction of an objectwithout altering its shape. Therefore, in the comparative analysis ofshape, a size parameter must first be identified such that the effectsof isometric scaling can be removed from the analysis. It should beemphasised that adjusting for scale in this manner does not auto-matically assume that size is not an important aspect of variation instone tools. Indeed, analyses of linear dimensions in Acheuleanhandaxes have shown allometric relationships between length andthickness, which has been interpreted as a design reconciliationbetween increasing size and weight (Crompton and Gowlett, 1993;Gowlett and Crompton, 1994). Meanwhile, Buchanan’s (2006)study of Folsom projectile points, and Shott and Weedman’s (2007)study of scrapers, demonstrate that allometric analyses are alsoa useful means of investigating models of technological organiza-tion in wider contexts.

Here, we wish to emphasise shape parameters and so controlfor isometric scale. In geometric morphometrics, the mostcommonly used ‘size’ measure is centroid size, which is defined as

the square root of the summed squared Euclidean distances fromeach landmark to the centroid (Niewoehner, 2005). By scaling alllandmark configurations to the same (unit) centroid size, theeffects of scaling are removed, allowing for the direct comparisonof differences in shape. Moreover, studies in physical anthropologyhave shown that the use of landmark-based methods allows fora highly flexible approach to morphometric analysis, and canaccommodate the analysis of specific localised regions such asindividual cranial bones (e.g. Lockwood et al., 2002; von Cramon-Taubadel, 2009).

In the analyses undertaken here, we deliberately focus on dorsalsurface morphology. It is often suggested that aspects of dorsal coresurface morphology (e.g., distal and lateral convexity, removal ofprepared flakes) are key components of Levallois cores (e.g., Boeda,1995; Van Peer, 1992). They are properties that are shared bothwith Victoria West cores (e.g., Van Riet Lowe, 1929, 1945; Rolland,1995) and Acheulean handaxes, thus facilitating a comparativeanalysis, and suggestive of technological links in these differentartefact forms (e.g. Schick, 1998; DeBono and Goren-Inbar, 2001).Thus, a study of dorsal surface morphology is particularly relevantto the morphometric comparison of Victoria West cores, Levalloiscores, and potentially related technological forms such ashandaxes.

2. Materials and methods

2.1. Materials

Table 1 lists the assemblages used in the comparativemorphometric analyses. Here, the artefacts are assigned to theextremely broad taxonomic category of Clark’s (1969) ‘Modes’(where broadly Mode 1¼ a simple core and flake industry; Mode2¼ a biface industry, and Mode 3¼ a prepared core industry).We are aware of the limitations of this scheme, but here itconstitutes a conservative approach to the morphological varia-tion represented in these artefacts for the purposes of thecomparative analyses (Lycett, 2007). It takes no account of thechronological or regional ‘variants’ that may emerge within thesebroadly defined technological distinctions (Lycett and Gowlett,2008). Thus, applying this scheme, the sample here includesexamples of Mode 1 cores, Mode 2 handaxes and Mode 3(Middle Palaeolithic) Levallois cores (total n¼ 639 artefacts).Included alongside this material are 36 Victoria West cores, all ofwhich represent struck examples. These latter artefacts haveparticular characterization value as they were collected by F.J.Jansen, the original discoverer of the Victoria West industry(Jansen, 1926), from a hillside adjacent to the town of VictoriaWest in the Karoo region of South Africa (Smith, 1919; Goodwin,1926, 1929; Van Riet Lowe, 1929, 1945). Hence, it is this materialthat subsequently lent its name to the entire ‘Victoria West’phenomenon.

Included within the Mode 3 (Levallois) sample is a group ofcores (n¼ 23) from Baker’s Hole (Northfleet), United Kingdom(Table 1). The Levallois cores from Baker’s Hole have drawn specificattention due to their relatively large size in comparison with manytypical Middle Palaeolithic Levallois assemblages (Smith, 1911, p.523; Roe, 1981, p. 80; Robinson, 1986, p. 20). Moreover, both Smith(1919, p. 102) and Goodwin (1929, p. 66) drew direct comparisonsbetween Victoria West cores and the Levallois cores from Baker’sHole (Northfleet) in their early descriptions of the Victoria West.Indeed, Goodwin (1929, p. 66) went as far as stating specificallythat Victoria West cores represented ‘‘a ‘tortoise core’ similar intype to those discovered at Northfleet (England)’’. Hence, inclusionof material from Baker’s Hole allows us to test directly the

Table 1Taxonomic units employed in analyses (total n¼ 639 nuclei).

Taxonomicunit number

Locality n Raw material Nuclei type/mode

1 Barnfield Pit, Kent, UK 22 Chert Ml2 Barnham St. Gregory, Suffolk, UK 30 Chert Ml3 Lion Point, Clacton, Essex, UK 18 Chert Ml4 Olduvai Gorge (Lower Bed II), Tanzania 11 Lava, chert, quartz Ml5 Olduvai Gorge (Middle/Upper Bed II), Tanzania 26 Lava, chert, quartz Ml6 Soan Valley, Pakistan 25 Quartzite Ml7 Zhoukoudian, Locality 1, China 14 Sandstone, quartz, limestone Ml8 Zhoukoudian, Locality 15, China 11 Sandstone, quartz Ml9 Attirampakkam, India 30 Quartzite M2 (Handaxe)10 Bezez Cave (Level C), Adlun, Lebanon 30 Chert M2 (Handaxe)11 Elveden, Suffolk, UK 24 Chert M2 (Handaxe)12 Kariandusi, Kenya 30 Lava M2 (Handaxe)13 Kharga Oasis (KOl0c), Egypt 17 Chert M2 (Handaxe)14 Lewa, Kenya 30 Lava M2 (Handaxe)15 Olduvai Gorge (Bed II), Tanzania 13 Quartz, lava M2 (Handaxe)16 Morgah, Pakistan 21 Quartzite M2 (Handaxe)17 St. Acheul, France 30 Chert M2 (Handaxe)18 Tabun Cave (Ed), Israel 30 Chert M2 (Handaxe)21 Baker’s Hole, Kent, UK 23 Chert M322 Bezez Cave (Level B), Adlun, Lebanon 28 Chert M323 El Arabah, Abydos, Egypt 16 Chert M324 El Wad (Level F), Israel 27 Chert M325 Fitz James, Oise, France 11 Chert M326 Kamagambo, Kenya 13 Quartzite, chert M327 Kharga Oasis (KO6e), Egypt 11 Chert M328 Muguruk, Kenya 12 Lava M329 Soan Valley, Pakistan 11 Quartzite M330 Victoria West 36 Lava (Dolerite) Para-Levallois

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–11171112

hypothesis that Victoria West cores are accurately characterised asa ‘large Levallois’ variant.

2.2. Methods

2.2.1. Landmark configurationThe basis of geometric morphometrics is the identification and

quantification of ‘homologous landmarks’, defined as ‘‘a point ofcorrespondence on an object that matches between and withinpopulations’’ (Dryden and Mardia, 1998, p. 3; O’Higgins, 2000).However, in the case of stone tools, large numbers of readilydefined points of correspondence (i.e. homologous landmarks) arenot easily located (Lycett et al., 2006; Lycett, 2009b). One wayaround this problem is to use what are termed ‘semilandmarks’(Bookstein, 1997; Buchanan, 2006; Lycett et al., 2006).

Terminologically, Bookstein (1991, pp. 63–66) originally identi-fied three categories of landmark. Type I landmarks were thosereadily identifiable points (e.g., cranial suture junctions) thatrequired no geometric definition in relation to other aspects of thespecimen. Type II landmarks were identified as morphologicallyisolated points or extremities (e.g., the tips of extrusions or invag-inations). Type III landmarks were regarded as geometricallydefined points, and thus are identified instrumentally. An impor-tant point here is that ‘homology’ is not necessarily an inherent orconveniently identifiable property, but something that may emergefrom a clear but operationally specified definition (O’Higgins,2000). Subsequently, Bookstein (1997) thus renamed Type IIIlandmarks as ‘semilandmarks’. Semilandmarks can conceptually bethought of as homologous in the sense of being geometricallycorrespondent across forms. Hence, via the use of explicitgeometric protocols for their identification, the locations of semi-landmarks are driven by the observed morphology, thus effectivelycapturing morphological similarities and disparities acrossspecimens.

In the case of the present study, 51 geometrically defined 3D co-ordinates (i.e. semilandmarks) were recorded using a CrossbeamCo-ordinate Caliper (Lycett et al., 2006). The resulting landmarkconfiguration is shown in Fig. 1. Full details regarding the semi-landmarking protocol, orientation of artefacts, and definitions of alllandmarks can be found in Lycett et al. (2006). In the case of bifaces,the ‘dorsal surface’ was defined as that exhibiting the mostextensively worked face. In ambiguous cases this was identified asthat exhibiting least amount of cortex and/or the face exhibiting thelargest number of flake scars (�1 cm in length x� 0.5 cm in width).The logic underlying this protocol is that the most intensivelyflaked surface will be that more extensively modified by homininagency.

2.2.2. Shape analysisFor each analysis undertaken, landmark configurations were

subjected to generalized Procrustes analysis (GPA) using themorphometrics software Morphologika 2.3.1. (O’Higgins and Jones,2006). GPA proceeds by removing variation between landmarkconfigurations due to isometric scale by reducing all configura-tions to unit centroid size, and then implements least-squarescriteria to minimize residual differences between configurationsdue to translation and rotation (Gower, 1975; Chapman, 1990).Any remaining variation between homologous landmark posi-tions (Procrustes residuals) can be interpreted as shape differ-ences between configurations. Thereafter, the Procrustes residualsare projected into a linear shape space tangent to the non-Euclidean shape space and subjected to principal componentsanalysis (PCA). PCA allows the major shape variation betweenindividual objects (in this case lithic nuclei) to be examined ina hierarchical fashion, whereby the first PC describes the majoraxis of shape variation (size having already been controlled for),the second PC describes the second major axis of variation, and soon. PCA was therefore employed here to examine the shapeaffinities of the Victoria West cores relative to the broad sample

Table 2Centroid size variation among assemblages.

Taxonomic Unit Average centroidsize

CV

Mode 1 Barnfield 253.56 23.45Olduvai B1 211.37 21.24Zoukoudian L1 205.62 30.38Olduvai B2 201.46 20.91Barnham 197.12 29.17Zoukoudian L15 196.34 42.36Lion Point 174.67 17.72Soan 170.61 23.90

Average CV 26.14

Mode 2 Lewa 305.62 13.85Olduvai B2 294.08 19.46Kariandusi 287.35 10.51Morgah 227.99 19.98Kharga 226.51 17.60ATPKM 220.65 19.35Elveden 212.89 17.42Bezez 202.74 20.31St. Acheul 193.50 18.51Tabun 166.14 12.60

Average CV 16.96

Mode 3 Baker’s Hole 289.47 20.36Muguruk 229.38 23.23Fitz James 212.54 24.56El Arabah 162.90 21.55Soan 161.06 16.85Kamagambo 144.48 21.54Bezez 141.94 18.37Kharga 131.53 9.39El Wad 123.00 25.15

Average CV 20.11

Victoria West 288.57 15.12

Fig. 1. Configuration of 51 landmarks used in the 3D geometric morphometricanalyses.

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–1117 1113

of core assemblages from the Lower and Middle Paleolithicdescribed in Table 1.

2.3. Analyses conducted

2.3.1. Comparison of centroid sizes in Victoria West and Baker’sHole cores

Differences in size between the different assemblages wereexamined via the statistical comparison of centroid sizes. To reit-erate, centroid size is defined as the square root of the summedsquared Euclidean distances from each landmark to the configu-ration centroid (Niewoehner, 2005). Centroid sizes for each indi-vidual artefact were calculated as part of the morphometricanalysis in Morphologika. These data were subsequently exported toSPSS v.16 for statistical comparison. As the centroid size data werenot normally distributed (Kolmogorov–Smirnov test, p¼ 0.20),a non-parametric test of mean size differences was employed(Mann–Whitney U test). Given the qualitative assessment that theVictoria West might be considered a ‘large’ Levallois assemblagesimilar to the Baker’s Hole assemblage, we predict that if thishypothesis is correct, the centroid sizes of Victoria West and Baker’sHole cores will be larger on average than other Levallois cores, butnot statistically different from each other.

2.3.2. Comparison of core shapesUsing PCA, we examine the shape affinities of the Victoria West

in relation to a large sample of Paleolithic nuclei. First, we examinedthe shape of Victoria West cores in relation to nuclei from Mode 1,Mode 2 and Mode 3. The prediction in this analysis is that if theVictoria West is accurately characterised as a ‘large’ Levallois form,

these cores will occupy the same shape space as the Levallois(Mode 3) cores.

The second analysis compares Victoria West and Baker’s Holecores directly. This second PCA analysis thus aims to examine therelationship between Victoria West cores, Mode 2 handaxes andMode 3 Levallois cores, including those from Baker’s Hole. Theprediction in this analysis is that if Victoria West is simply a largeisometric variant of Levallois – in a manner analogous to that ofBaker’s Hole – then the Victoria West cores should occupy the sameshape space as Baker’s Hole, and both should occupy the sameshape space as the smaller Levallois cores.

3. Results

3.1. Comparison of centroid sizes in Victoria West and Baker’sHole cores

Table 2 shows the average centroid size for each taxonomic unitanalysed, including the Victoria West assemblage. Co-efficient ofvariation (CV) statistics are also provided to illustrate that ingeneral, the Mode 1 cores showed greater variability in size thanMode 2 handaxes and Mode 3 cores (Table 2). In accordance withpredictions made, the Baker’s Hole and the Victoria West speci-mens were on average larger than all other Mode 3 assemblagesassessed. Mann–Whitney U tests also showed that there is nostatistically significant difference in size between the Victoria West(288.57) and the Baker’s Hole (289.47) cores (p¼ 0.852). Indeed,the Victoria West were found to be significantly larger than allother Mode 3 assemblages compared, except for those from Baker’s

Table 3Comparison of centroid sizes between Victoria West and all Mode 3 assemblages.

Taxonomic unit comparison Mann-Whitney U

Baker’s Hole 402 (p¼ 0.852)Muguruk 82 (p¼ 0.001)Fitz James 59 (p< 0.0001)El Arabah 11 (p< 0.0001)Soan 4 (p< 0.0001)Kamagambo 3 (p< 0.0001)Bezez 5 (p <0.0001)Kharga 0 (p< 0.0001)El Wad 4 (p< 0.0001)

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–11171114

Hole (Table 3). Hence, this comparison highlights not just the largesize of the Victoria West, but the highly distinctive nature of Baker’sHole amongst European Levallois assemblages (Roe, 1981, p. 80;Robinson, 1986; Wenban-Smith, 1995).

3.2. Comparison of core shapes

Fig. 2 plots the first two principal components (PCs) from theshape analysis of all 30 assemblages including the Victoria West.PC1 (29.2% of variance) separates the Mode 1 cores from the Mode2, Mode 3 and Victoria West cores. The dorsal- and lateral-viewwireframe diagrams illustrate the major shape changes associatedwith PC1. Mode 1 cores (negative PC1) are associated with anirregular outline in dorsal view and a highly domed morphologywhen viewed from the lateral aspect. Conversely, positive values ofPC1 are associated with a highly symmetrical teardrop plan formshape and a much flattened shape in lateral aspect. This combi-nation of shape variables is predominantly associated with VictoriaWest cores, which occupy the shape space at the extreme positiveend of PC1. PC2 (18.1% variance) separates the Mode 2 and Mode 3cores. Mode 2 handaxes (negative PC2) are associated with a highlysymmetrical teardrop shape in plan form and a relatively domedsurface morphology. In contrast, the Mode 3 cores are characterisedby a symmetrical round outline shape and a much flattened surface

Fig. 2. Plot of the first two principal components of shape variation among the total sampleshape changes associated with PC1 (29.2% of total variation) and PC2 (18.1% of total variati

morphology, relative to Mode 2. Victoria West cores share shapespace with both Mode 2 and Mode 3 cores, but do not overlap inshape variation with Mode 1 core forms. However, the results ofthis PCA analysis also indicate that Victoria West cores are some-what distinct in shape from both Mode 2 and Mode 3 cores, withsome cores occupying a unique position in shape space, not sharedwith other core forms. Hence, in sum, this first shape analysis doesnot support the contention that Victoria West cores are simplya large (isometrically scaled) Levallois variant.

Fig. 3 plots the first two principal components (PCs) from theshape analysis of Mode 2, Mode 3 and Victoria West assemblages.PC1 (36.3% variance) separates Mode 3 cores from Mode 2 han-daxes. The shape differences associated with this distinction areillustrated by the wireframe diagrams, which show the Mode 3cores are rounder in outline shape and flatter in lateral aspect,compared with the more pointed outline shape and domed surfacemorphology of the Mode 2 handaxes. PC2 (17.1% variance) capturesthe shape variance within Mode 2 and Mode 3 assemblages, whichranges from more domed and elongated forms (positive PC2) toflatter, less elongated forms (negative PC2). The Baker’s Hole coresshare the same shape space as other Mode 3 cores (Fig. 3). Indeed,the Baker’s Hole cores occupy all of the shape space occupied byMode 3 cores, as opposed to being limited to a single portion ofMode 3 shape space. Moreover, the Baker’s Hole do not overlap withthe Mode 2 handaxes to any greater extent than the other (smaller)Mode 3 cores. Therefore, on the basis of this analysis, we can beconfident that the large size of the Baker’s Hole cores has not causeda differentiation in their shape, relative to other Levallois coreassemblages. In contrast, the Victoria West cores share shape spacewith both the Mode 2 and Mode 3 cores. Indeed, they share moreshape affinities with Mode 2 handaxes, as over half of the VictoriaWest cores (22 out of 36) occupy positive positions on PC1. Hence,this second shape analysis also does not support the contention thatVictoria West cores are simply an isometrically-scaled large variantof Levallois core equivalent to those from Baker’s Hole. Rather, theVictoria West cores share greater shape affinities with Acheuleanhandaxes than do any of the Levallois assemblages.

of cores analysed. The dorsal- and lateral-view wireframe diagrams illustrate the majoron).

Fig. 3. PCA plot showing the shape affinities of the Baker’s Hole and the Victoria West assemblages relative to a sample of Mode 2 handaxes and Mode 3 Levallois cores. PC1 moststrongly differentiates between Mode 2 and Mode 3 cores. Baker’s Hole cores overlap predominantly with the Mode 3 Levallois cores on PC1, while the Victoria Cores overlappredominantly with Mode 2 handaxes. The dorsal- and lateral-view wireframe diagrams illustrate the major shape changes associated with PC1 (36.3% of total variation) and PC2(17.1% of total variation).

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–1117 1115

4. Discussion

4.1. Is the Victoria West a ‘large Levallois’?

The Victoria West is a Lower Palaeolithic prepared core industryfound in central South Africa (Northern Cape region) (Rolland,1995; McNabb, 2001; Kuman, 2001; Sharon and Beaumont, 2006;Lycett, 2009a; Sharon, 2009). From its initial discovery, thisphenomenon was regarded as a large and somewhat crude ‘proto-Levallois’ technology (Smith, 1919; Van Riet Lowe, 1929, 1945;Breuil, 1930; Goodwin, 1929, 1934). Some even drew directcomparisons between the Victoria West and examples of largeMiddle Palaeolithic Levallois cores from Baker’s Hole (Northfleet),United Kingdom (e.g., Smith, 1919, p. 102; Goodwin, 1929, p. 66).Here, the contention that Victoria West cores are accurately char-acterised as a ‘large Levallois’ variant was tested directly usinga series of 3D geometric morphometric (GM) analyses.

Two sets of analyses were conducted. In our first analysis, weexamined size variation in Victoria West cores statistically, againsta comparative sample of Middle Palaeolithic Levallois assemblages.This included a series of Levallois cores from the Middle Palaeolithiclocality of Baker’s Hole (UK). Given the qualitative assertions thatthe Victoria West might be considered a ‘large’ Levallois assem-blage similar to the Baker’s Hole assemblage, we predicted that ifthis hypothesis is correct, the centroid sizes of Victoria West andBaker’s Hole cores will be larger on average than other Levalloiscores, but not statistically different from each other. In congruencewith these predictions, it was found that the Baker’s Hole and theVictoria West were significantly (p� 0.001) larger than all otherLevallois assemblages assessed, yet the Baker’s Hole and theVictoria West assemblages were statistically indistinguishable onthe basis of size. Hence, this first analysis supported the contentionthat – in terms of size alone – the Victoria West are larger than themajority of Middle Palaeolithic Levallois cores, but comparable tothose from Baker’s Hole.

Our second set of analyses examined shape variation specifi-cally. The first shape analysis compared Victoria West cores againsta series of Mode 1 cores, Mode 2 handaxes and Mode 3 Levalloiscores. The prediction in this analysis was that if the Victoria Westis accurately characterised as a ‘large Levallois’ form, these coreswill occupy the same shape space as the Levallois cores. Thesecond shape analysis compared Victoria West and Baker’s Holecores directly. The prediction in this analysis was that if VictoriaWest is simply a large (isometrically-scaled) variant of Levallois, ina manner analogous to that of Baker’s Hole, then the Victoria Westcores should occupy the same shape space as Baker’s Hole, andboth should occupy the same shape space as the smaller Levalloiscores.

The results of these two shape analyses did not support thecontention that Victoria West cores are accurately characterised asa ‘large Levallois’ variant. The Victoria West cores were found toshare shape space with both Mode 2 and Mode 3 cores. However,the results of this analysis also indicated that Victoria West coresare somewhat distinct in shape from both Mode 2 and Mode 3cores, with some cores occupying a unique position in shape space,not shared with other core forms. Moreover, in our second shapeanalysis, the Baker’s Hole cores were found to share the same shapespace as other Mode 3 (Levallois) suggesting that they can accu-rately be characterised as a large (isometrically scaled) variant ofLevallois technology. However, the Baker’s Hole were found not tooverlap with the Mode 2 handaxes to any greater extent than theother (statistically smaller) Mode 3 cores. Conversely, the VictoriaWest cores were found to share markedly more shape space withMode 2 (handaxe) assemblages compared with those from Baker’sHole, and indeed, the other Levallois cores. Hence, this secondshape analysis also did not support the hypothesis that VictoriaWest cores are simply an isometrically-scaled large variant ofLevallois core, equivalent to those from Baker’s Hole.

In sum, our 3D geometric morphometric analyses demonstratedthat although the Victoria West are broadly comparable in terms of

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–11171116

size to a comparative sample of large Levallois cores (i.e. those fromBaker’s Hole), on average, their upper surface morphology ismarkedly different in terms of shape. Indeed, in contrast to theLevallois cores used in the comparative analyses, it appears that theVictoria West share, on average, a greater shape affinity with Mode2 Acheulean handaxes. Hence, these analyses suggest thatconceptualizing the Victoria West as a ‘large Levallois’ variant, is notan accurate characterization. Indeed, our analyses quantitativelyconfirm Sharon’s (2009, p. 349) recent statement that ‘‘[t]o describethe Victoria West . as nothing more than Levallois cores wouldresult in the loss of many significant details.’’ Of course, it remainspossible that other aspects of Victoria West and Levallois coremorphology not measured here (e.g. the underside of the cores)share a much greater morphometric affinity; however, sucha finding would not contradict our main conclusion that the dorsalsurface of Levallois cores and Victoria West cores exhibit quantifi-able differences.

4.2. Rethinking the Victoria West

The finding that Victoria West cores have a distinctive aspect totheir shape has important implications for debates surroundingthis prepared core industry, especially when considered alongsideother recent analyses.

Qualitatively, there are many similarities between VictoriaWest cores and Middle Palaeolithic Levallois cores. Indeed, asnoted by some authors (e.g., Kuman, 2001; McNabb, 2001, Sharon,2009) these cores are broadly congruent with Boeda’s (1995)‘volumetric’ definition of Levallois cores, which states that thevolume of the core is comprised of two distinct asymmetricalfaces, one of which serves as the source of predetermined flakes,and another that provides a series of striking platforms for theremoval of such flakes. Recognition of such similarities by earlycommentators provided a source for suggestions that the VictoriaWest constituted an ancestral proto-type for Middle Palaeolithicprepared cores, or ‘proto-Levallois’ technology (Breuil, 1930;Goodwin, 1934; Van Riet Lowe, 1945; Bordes, 1968, p. 69). Sucha role for Victoria West core industries was perhaps most elabo-rately detailed by Van Riet Lowe (1945, pp. 57–58) who, in thestyle of the culture-historical approach of the day, outlineda unilinear evolutionary scheme going from ‘proto-Levallois’ cores,through to African Levallois cores, right up to ‘‘comparable Euro-pean forms’’.

Despite such frequent reference to the idea that Victoria Westcores represented a ‘proto-Levallois’ industry, an early word ofcaution came from Louis Leakey (1936, p. 85) who stated thatalthough ‘‘there was definitely a resemblance to the Levalloistechnique . The question is whether this was a case of parallelevolution or whether the techniques are more closely related.’’Interestingly, in more recent years, two distinct origins for MiddlePalaeolithic Levallois core techniques have been suggested. One ofthese lies in the so-called ‘giant-core’ traditions of large flakemanufacture, the flake products of which formed blanks forAcheulean bifaces such as cleavers and handaxes (Gowlett, 1996;Madsen and Goren-Inbar, 2004). A range of different Acheulean‘giant-core’ techniques have been identified of which the VictoriaWest represents a localised southern African example (seeSharon, 2007, 2009 for reviews). Sharon (2009) has recentlysuggested that several of these large core techniques mayrepresent convergent solutions to the technological problem ofproducing large flake blanks. The alternative putative origin forMiddle Palaeolithic Levallois core techniques lies in Acheuleanhandaxes, whereby it is suggested that Levallois core productionhas close technological affinities with techniques involved in theprocess of Acheulean biface production and refinement (e.g.,

Leroi-Gourhan, 1966; Copeland, 1995; Rolland, 1995; Tuffreau,1995; Tuffreau and Antoine, 1995; Schick, 1998; DeBono andGoren-Inbar, 2001).

Recently, Lycett (2009a) conducted cladistic (parsimony) anal-yses designed to assess Louis Leakey’s (1936) question of whethermorphological similarities evident in Victoria West cores andLevallois cores are the result of ancestor–descendant relationships(i.e. that such similarities are phylogenetically homologous) orwhether they could, more parsimoniously, be interpreted asa result of convergent evolution within the Acheulean. Lycett(2009a) tested this assertion using a series of Mode 3 Levalloiscores and Mode 2 Acheulean handaxe assemblages, as well asa series of Victoria West cores. Results of the cladistic analysesdemonstrated that, most parsimoniously, similarities betweenVictoria West cores and Levallois cores are not due to directancestor-descendant relationships, but are the product of conver-gent technological evolution. The analyses further found somesupport for the hypothesis that the origins of Levallois flakeproduction lay in the bifacial knapping routines of handaxeproduction. Hence, (contra Kuman, 2009) Lycett’s (2009a) phylo-genetic analyses do not contradict an Acheulean origin for MiddlePalaeolithic Levallois core traditions. Far from it. Rather, what theseanalyses clarify is the nature of the connectivity between the Lowerand Middle Palaeolithic.

Hughs and Chapman (2001, p. 29) have stated that whenapproaching questions concerning ancestry in biological settings:‘‘Judicious combinations of morphometric and phylogeneticapproaches permit exploration of the context of patterns ofmorphological variation’’. We contend that similar insights can begained in the case of the Victoria West when consideration ofLycett’s (2009a) phylogenetic analyses are combined with themorphometric analyses reported in the present study. That is, our3D geometric morphometric analyses provide further evidence thatVictoria West cores are distinctive from Middle Palaeolithic Leval-lois cores, even Levallois cores that are statistically comparable interms of size. Interestingly, Sharon’s (2009, p. 348) recent exami-nation of a series of large boulder core assemblages (includingexamples from the Victoria West) led him to conclude that flakesremoved from such cores are ‘‘derived from debitage surfaces verydifferent from those for the relatively small Levallois points orblades’’. Our morphometric analyses concur with these conclu-sions. Lycett’s (2009a) cladistic analyses meanwhile, suggest thatsuch similarities which Victoria West cores do share with Levalloiscores are, most parsimoniously, the product of convergentevolution.

It thus appears that although the Victoria West is part of a seriesof Acheulean large ‘boulder-core’ technological traditions – some ofwhich may themselves be convergent innovations (Sharon, 2009) –it is neither simply a ‘large Levallois’ variant’ nor a ‘proto-Levallois’tradition. The manufacturers of Acheulean handaxes, Levalloiscores and Victoria West cores all demonstrate a clear capability toorganise operations as strategically controlled series of events (i.e.complex operational chains). However, different needs could leadto these being combined in different ways. The Victoria West isa flexible way of making large flakes, in which the outcome has thepotential to be a handaxe, cleaver, or large scraper, as required. Incontrast, later Levallois is directed toward an alternative set of endproducts (Rolland, 1995). We thus suggest that the Victoria Westshould be seen as a derived variant of the deeply-rooted ‘largeboulder core’ traditions of the Acheulean itself (Gowlett, 1996),rather than the ‘origin’ of Middle Palaeolithic Levallois core tradi-tions. As a result of these combined considerations, we also proposethat Bordes’ (1961, p. 16) term of ‘para-Levallois’ is more appro-priate when referring to Victoria West cores, rather than the term‘proto-Levallois’.

S.J. Lycett et al. / Journal of Archaeological Science 37 (2010) 1110–1117 1117

Acknowledgements

For helpful comments we thank Metin Eren, Mike O’Brien andtwo further anonymous reviewers. We gratefully acknowledgefinancial support from the British Academy Centenary ResearchProject, Lucy to Language. We are also grateful to staff at the Cam-bridge University Museum of Archaeology and Anthropology, andthe British Museum for their hospitality and assistance.

References

Boeda, E., 1995. Levallois: a volumetric construction, methods, a technique. In:Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of LevalloisTechnology. Prehistory Press, Madison, Wisconsin, pp. 41–68.

Bookstein, F.L., 1991. Morphometric Tools for Landmark Data: Geometry andBiology. Cambridge University Press, Cambridge.

Bookstein, F.L., 1997. Landmark methods for forms without landmarks: morpho-metrics of group differences in outline shape. Medical Image Analysis 1,225–243.

Bordes, F., 1961. Typologie de Paleolithique Ancien et Moyen. Delmas, de l’Universitede Bordeaux 1, Bordeaux.

Bordes, F., 1968. The Old Stone Age. Weidenfield and Nicolson, London.Breuil, H., 1930. Premieres impressions de voyage sur la prehistoire Sud-Africaine.

L’Anthopologie 40, 209–223.Buchanan, B., 2006. An analysis of Folsom projectile point resharpening using

quantitative comparisons of form and allometry. Journal of ArchaeologicalScience 33, 185–199.

Chapman, R.E., 1990. Conventional Procrustes approaches. In: Rohlf, F.J.,Bookstein, F.L. (Eds.), Proceedings of the Michigan Morphometrics WorkshopSpecial Publication No. 2. University of Michigan Museum of Zoology, AnnArbor, pp. 251–267.

Clark, G., 1969. In: World Prehistory: A New Outline, second ed. CambridgeUniversity Press, Cambridge.

Clark, J.D., 1959. The Prehistory of Southern Africa. Penguin, London.Clark, J.D., 2001. Variability in primary and secondary technologies of the Later

Acheulian in Africa. In. In: Milliken, S., Cook, J. (Eds.), A Very Remote PeriodIndeed: Papers on the Palaeolithic Presented to Derek Roe. Oxbow Books,Oxford, pp. 1–18.

Copeland, L., 1995. Are Levallois flakes in the Levantine Acheulian the result ofbiface preparation? In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition andInterpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin,pp. 171–183.

Crompton, R.H., Gowlett, J.A.J., 1993. Allometry and multidimensional form inAcheulean bifaces from Kilombe, Kenya. Journal of Human Evolution 25,175–199.

DeBono, H., Goren-Inbar, N., 2001. Note of a link between Acheulian handaxes andthe Levallois method. Journal of the Israel Prehistoric Society 31, 9–23.

Dryden, I.L., Mardia, K.V., 1998. Statistical Shape Analysis. Wiley, New York.Goodwin, A.J.H., 1926. South African stone implement industries. South African

Journal of Science 23, 784–788.Goodwin, A.J.H., 1929. The Victoria West industry. In: Goodwin, A.J.H., Lowe, V.R.

(Eds.), The Stone Age Cultures of South Africa. Annals of the South AfricanMuseum, pp. 53–69.

Goodwin, A.J.H., 1934. Some developments of technique during the earlier StoneAge. Transactions of the Royal Society of South Africa 21, 109–123.

Gower, J.C., 1975. Generalised Procrustes analysis. Psychometrika 40, 33–50.Gowlett, J.A.J., 1996. Mental abilities of early Homo: elements of constraint and

choice in rule systems. In: Mellars, P., Gibson, K. (Eds.), Modelling the EarlyHuman Mind. McDonald Institute, Cambridge, pp. 191–215.

Gowlett, J.A.J., Crompton, R.H., 1994. Kariandusi: Acheulean morphology and thequestion of allometry. The African Archaeological Review 12, 3–42.

Hughs, N.C., Chapman, R.E., 2001. Morphometry and phylogeny in the resolution ofpaleobiological problems – unlocking the evolutionary significance of anassemblage of silurian trilobites. In: Adrain, J.M., Edgecombe, G.D.,Lieberman, B.S. (Eds.), Fossils, Phylogeny and Form: An Analytical Approach.Kluwer/Plenum, New York, pp. 29–54.

Jansen, F.J., 1926. A new type of stone implement from Victoria West. South AfricanJournal of Science 23, 818–825.

Kuman, K., 2001. An Acheulean factory site with prepared core technology nearTaung, South Africa. South African Archaeological Bulletin 56, 8–22.

Kuman, K., 2009. Comment on ‘‘Acheulian giant-core technology’’. CurrentAnthropology 50, 359.

Leakey, L.S.B., 1936. Stone Age Africa: An Outline of Prehistory in Africa. HumphreyMilford, London.

Leroi-Gourhan, A., 1966. La Prehistoire. Universitaires de France, Paris.

Lycett, S.J., 2007. Why is there a lack of Mode 3 Levallois technologies in East Asia? Aphylogenetic test of the Movius–Schick hypothesis. Journal of AnthropologicalArchaeology 26, 541–575.

Lycett, S.J., 2009a. Are Victoria West cores ‘proto-Levallois’? A phylogeneticassessment. Journal of Human Evolution 56, 175–191.

Lycett, S.J., 2009b. Quantifying transitions: morphometric approaches to Palae-olithic variability and technological change. In: Camps, M., Chauhan, P.R. (Eds.),Sourcebook of Palaeolithic Transitions: Methods, Theories, and Interpretations.Springer, New York, pp. 79–92.

Lycett, S.J., von Cramon-Taubadel, N., Foley, R.A., 2006. A crossbeam co-ordinatecaliper for the morphometric analysis of lithic nuclei: a description, test andempirical examples of application. Journal of Archaeological Science 33,847–861.

Lycett, S.J., Gowlett, J.A.J., 2008. On questions surrounding the Acheulean ‘tradition’.World Archaeology 40, 295–315.

Lockwood, C.A., Lynch, J.M., Kimbel, W.H., 2002. Quantifying temporal bonemorphology of great apes and humans: an approach using geometricmorphometrics. Journal of Anatomy 201, 447–464.

Madsen, B., Goren-Inbar, N., 2004. Acheulian giant core technology and beyond: anarchaeological and experimental case study. Eurasian Prehistory 2, 3–52.

McNabb, J., 2001. The shape of things to come. A speculative essay on the role of theVictoria West phenomenon at Canteen Koppie, during the South African EarlierStone Age. In: Milliken, S., Cook, J. (Eds.), A Very Remote Period Indeed: Paperson the Palaeolithic Presented to Derek Roe. Oxbow Books, Oxford, pp. 37–46.

Niewoehner, W.A., 2005. A geometric morphometric analysis of late Pleistocenehuman metacarpel 1 base shape. In: Slice, D.E. (Ed.), Modern Morphometrics inPhysical Anthropology. Kluwer, New York, pp. 285–298.

O’Higgins, P., 2000. The study of morphological variation in the hominid fossilrecord: biology, landmarks and geometry. Journal of Anatomy 197, 103–120.

O’Higgins, P., Jones, N., 2006. Tools for Statistical Shape Analysis. Hull York MedicalSchool.

Robinson, P., 1986. An introduction to the Levallois Industry at Baker’s Hole (Kent),with a description of two flake cleavers. In: Collcutt, S.N. (Ed.), The Palaeolithicof Britain and Its Nearest Neighbours: Recent Trends. Department of Archae-ology & Prehistory, University of Sheffield, Sheffield, pp. 20–22.

Roe, D.A., 1981. The Lower and Middle Palaeolithic Periods in Britain. Routledge &Kegan Paul, London.

Rolland, N., 1995. Levallois technique emergence: single or multiple? A review ofthe Euro-African record. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition andInterpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin,pp. 333–359.

Schick, K., 1998. A comparative perspective on Paleolithic cultural patterns. In:Akazawa, T., Aoki, K., Bar-Yosef, O. (Eds.), Neandertals and Modern Humans inWestern Asia. Plenum Press, New York, pp. 449–460.

Sharon, G., 2007. Acheulian large flake industries: technology, chronology, andsignificance. British Archaeological Reports. In: International Series 1701.Archaeopress, Oxford.

Sharon, G., 2009. Acheulian giant-core technology. Current Anthropology 50, 335–367.Sharon, G., Beaumont, P., 2006. Victoria West – a highly standardized prepared core

technology. In: Goren-Inbar, N., Sharon, G. (Eds.), Acheulian Tool-making fromQuarry to Discard. Equinox, London, pp. 181–199. Axe Age.

Shott, M.J., Weedman, J.J., 2007. Measuring reduction in stone tools: an ethno-archaeological study of Gamo hidescrapers from Ethiopia. Journal of Archaeo-logical Science 34, 1016–1035.

Slice, D.E., 2005. Modern morphometrics. In: Slice, D.E. (Ed.), Modern Morpho-metrics in Physical Anthropology. Kluwer, New York, pp. 1–45.

Smith, R.A., 1911. A Palaeolithic industry at Northfleet, Kent. Archaeologica 62,512–532.

Smith, R.A., 1919. Recent finds of the stone age in Africa. Man 19, 100–106.Tuffreau, A., 1995. The variability of Levallois technology in northern France and

neighboring areas. In: In, Dibble H.L., Bar-Yosef, O. (Eds.), The Definition andInterpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin,pp. 413–427.

Tuffreau, A., Antoine, P., 1995. The earliest occupation of Europe: continentalnorthwestern Europe. In: Roebroeks, W., van Kolfschoten, T. (Eds.), The EarliestOccupation of Europe. University of Leiden and European Science Foundation,Leiden, pp. 147–163.

Van Peer, P., 1992. The Levallois Reduction Strategy. Prehistory Press, Madison,Wisconsin.

Van Riet Lowe, C., 1929. Fresh light on the Prehistoric archaeology of South Africa.Bantu Studies 3, 385–393.

Van Riet Lowe, C., 1945. The evolution of the Levallois technique in South Africa.Man 45, 49–59.

von Cramon-Taubadel, N., 2009. Congruence of individual cranial bone morphologyand neutral molecular affinity patterns in modern humans. American Journal ofPhysical Anthropology 140, 205–215.

Wenban-Smith, F.F., 1995. The Ebbsfleet Valley, Northfleet (Baker’s Hole) (TQ615735). In: Bridgland, D.R., Allen, P., Haggart, B.A. (Eds.), The Quaternary of theLower Reaches of the Thames. Quaternary Research Association, Durham, pp.147–164.