Journal of Archaeological Science 40 (2013) 2384e2392

Contents lists available at SciVerse ScienceDirect

Journal of Archaeological Science journal homepage: http://www.elsevier.com/locate/jas

Levallois economics: an examination of ‘waste’ production in experimentally produced Levallois reduction sequences Stephen J. Lycett*, Metin I. Eren Department of Anthropology, University of Kent, Marlowe Building, Canterbury, Kent CT2 7NR, United Kingdom

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 December 2012 Received in revised form 18 January 2013 Accepted 19 January 2013

Mathematical modelling has suggested that Levallois core morphology represents a reduction strategy driven by economic considerations; particularly the minimization of ‘waste’ while aiming to maximize cutting edge length of flakes obtained from cores of a given size. Such models are elegant in that they facilitate formal modelling of economic considerations that potentially motivate patterns seen in prehistoric data. However, the abstract nature of such models means that they do not take full account of all the practical difficulties and material challenges involved in reproducing Levallois-style reductions in stone. In particular, such models have only examined nodule morphology in two-dimensions, and did not take account of the fact that in the case of classic (lineal) Levallois reduction, core surfaces must be reprepared between successive stages of flake removal. Hence, the potential economic implications of these factors are currently unknown, potentially undermining the significance of models that assume specific economic conditions. Here, we undertook to examine these factors using a series of experimentally produced Levallois reduction sequences. A total of 3957 flaking events were considered in our analyses, and we used six specific measures of economy to examine Levallois reduction across successive phases. Our analyses found that key assumptions of mathematical models suggesting that Levallois core morphology was driven by economic considerations (i.e. conservation of raw material when attempting to remove flakes with long cutting edges) can be upheld under the practical challenges of replicating Levallois-style reduction in stone. In supporting the notion that Levallois reduction has advantageous economic properties, our results emphasize the importance of considering why Levallois reduction did not emerge earlier in the archaeological record, and indeed, why even during the later Pleistocene the temporal and geographic distribution of Levallois technology varies. Our results also re-emphasize the value of formally modelling lithic reduction strategies in specific economic terms. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: Levallois Core reduction Economy of reduction Waste Experiment

1. Introduction Discussions of Levallois cores and flakes continue to hold a prominent place in Palaeolithic archaeology, often as a result of their putative connection to pertinent issues in prehistoric research and hominin evolution. Indeed, in recent years, Levallois technology has played a part in debates regarding topics such as behavioural evolution (e.g., Tryon et al., 2006; Wilkins et al., 2010; Moncel et al., 2011; Scott, 2011), the evolution of particular hominin species (Hublin, 2009), skill learning (Eren et al., 2011a, 2011b), technological convergence (Lycett, 2009; Sharon, 2009), and issues of predetermination, planning, and cognition (Wynn and Coolidge, 2004, 2010; Pelegrin, 2005; Eren and Lycett, 2012). Such recent

* Corresponding author. Tel.: þ44 1227 827739. E-mail address: [email protected] (S.J. Lycett). 0305-4403/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jas.2013.01.016

debates are part of long tradition, with the earliest scholastic discussions of Levallois now dating back to over a century ago (e.g., Commont, 1909; Smith, 1911). As reviewed by Van Peer (1992) and Schlanger (1996), ‘Levallois’ has, at various times in its (scholastic) history, been considered variously as a set of specific ‘end-products’ (i.e. Levallois flakes and cores) and/or a specific reduction strategy or ‘technique’ for producing those end-products, especially with regard to ‘predetermined’ flakes. In a more recent study, Brantingham and Kuhn (2001) presented a novel view of Levallois, which invoked a specific sense of economic strategy with regard to the maximization of cutting edge productivity that could be obtained from cores, while also simultaneously minimizing raw material wastage. In their study, Brantingham and Kuhn (2001) modelled initial starting nodules as ellipses of varying dimensions. By varying factors such as the initial platform position for a series of longitudinally removed flakes, and the acuteness of the angle of such cores, they

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

were able to model how much ‘waste’ would be generated for nodules of a given size for varying core shapes (Fig. 1). Brantingham and Kuhn found that steep-angled cores, conforming broadly to Boëda’s (1994, 1995) ‘volumetric’ definition of Levallois core morphology, minimized both the amount of material that would be wasted while also maximizing the amount of cumulative cutting edge that would be produced for the longitudinally removed flakes (Fig. 1). As with all models, Brantingham and Kuhn’s (2001) simulations were schematic representations of a process that in reality is inevitably more complex and dynamic. Indeed, this is something that Brantingham and Kuhn (2001: 752) themselves were keen to stress. These authors noted some potential limitations in specific terms, in particular, that the ‘nodules’ they modelled were twodimensional in form (Brantingham and Kuhn, 2001: 749). Moreover, as noted above, the optimal cores modelled by Brantingham and Kuhn (2001), conformed broadly to Boëda’s (1995) ‘volumetric’ definition of Levallois core form. That is, the core is essentially organized bifacially and hierarchically, with one surface of the core providing a series of steep-angled platforms from which flakes are removed longitudinally from the opposing surface, broadly parallel to a plane of intersection between these two hierarchically organized surfaces (Fig. 1). A distinct element of Boëda’s (1995) volumetric definition, however, at least in the case of classic (lineal) Levallois, is that between each phase of longitudinal flaking, the primary flaking surface of the core is re-prepared in order to rebuild properties of distal and lateral convexity, which control the termination and margins of the flake products (Van Peer, 1992; Boëda, 1995). The Brantingham and Kuhn model does not, however, incorporate this re-preparation stage between production of Levallois flakes (Fig. 1). Hence, the potential economy of this phase relative to others factors (i.e. relative to the initial production of the volumetric Levallois core form, and the extent of waste required to produce subsequent Levallois flakes following initial removal) is unknown. Given this factor, an investigation of Levallois reduction in these terms appears warranted. As Brantingham and Kuhn (2001: 758) noted, their models are “realistic only where organized flaking surfaces can be achieved with a minimum of preparation”. Hence, Brantingham and Kuhn’s (2001) model would imply that the initial stages of Levallois core preparation are the most wasteful and that, thereafter, the relationship between cutting edge length and ‘waste’ is e at the very least e stabilized. Here, we undertake to test this proposition using an experimental framework, assessing specifically, the relative economy of raw material usage at multiple stages of Levallois

2385

reduction, and the relationship between cutting edge of Levallois flakes versus the ‘waste’ that is required to produce them at each stage of reduction. 2. Materials and methods 2.1. Experimental Levallois reductions Brantingham and Kuhn (2001: 752) noted that although their mathematical models of various core morphologies enabled specific, quantitative examination of essential aspects of core reduction, they could not possibly incorporate many key features of genuine knapping (such as the 3-dimensionality of nodule form or dynamic factors such as the correction of errors that may occur). However, actualistic experiments that attempt to replicate Levallois reduction, provide a viable means of assessing the internal economy of the relative elements of Levallois reduction and the products that it produces at multiple stages of such reductions. In order to assess the economy of Levallois reduction in a more actualistic context, one of us (MIE) undertook a series of multiple Levallois reductions (total n ¼ 75) using 25 cortical nodules of Texas Chert (Fredericksburg variety). Table 1 describes summary statistics for the dimensions and mass of these nodules, and Fig. 3 shows a range of examples. Direct, hard hammer percussion was used exclusively throughout each reduction. The aim in each reduction sequence was to produce classically-shaped lineal (i.e. ‘tortoise’) Levallois cores with flakes (i.e. ‘preferential Levallois flakes’ or simply, ‘PLFs’) that removed a large proportion of the prepared surface, as per Brantingham and Kuhn’s (2001) model. As such, these cores also conformed fully to Boëda’s (1995) ‘volumetric’ criteria for Levallois, again consistent with the assumptions of Brantingham and Kuhn’s (2001) conception of overall Levallois core morphology as depicted by their model. Following Bradley (1977); see also, Eren and Bradley (2009) and Eren and Lycett (2012), we defined ‘ventral’ flakes as those removed from the face from which the putative PLFs are removed, and refer to flakes removed from the non-PLF surface as ‘dorsal’ flakes (Fig. 2). This is potentially confusing as Levallois cores are often depicted with the Levallois flaking surface facing upward (i.e. superiorly). However, it should be noted that when PLFs are eventually removed, the cores are orientated with the Levallois surface facing downward (i.e. ventrally), thus establishing the terminology used here (Fig. 2). For each separate Levallois reduction stage, all debitage flakes from the dorsal and ventral surfaces, as well as each PLF, were bagged separately and labelled. The experimental assemblage consisted of 3957 total flakes. No archaeological experiment can re-run prehistory. However, some steps can be taken to ensure that there are genuine physical and conceptual links between experimental results and the archaeological phenomenon concerned. A series of 3D geometric morphometric analyses previously undertaken by the authors

Table 1 Basic properties of chert nodules (n ¼ 25) used in the experiments. Dimensions are maxima taken orthogonal to length (where, ‘length’ ¼ maximum dimension of nodule through the centre of its mass). These nodules produced a total of 3957 flakes, including 75 PLFs.

Fig. 1. Brantingham and Kuhn’s (2001) model of Levallois reduction in relation to raw material waste. Levallois cores are modelled as a series of flakes removed sequentially (longitudinally) from steep-angled platform core, in accordance with Boëda’s (1995) ‘volumetric’ concept of Levallois core form.

Mean Standard deviation Minimum Maximum

Mass (g)

Length (mm)

Width (mm)

Thickness (mm)

Width/ thickness ratio

Length/ width ratio

2553 632

197 24

131 17

69 10

1.91 0.29

1.52 0.25

1512 3747

155 237

110 188

53 91

1.32 2.56

1.14 2.03

2386

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

2.2. Analyses We examined Levallois reduction across different stages of PLF production in six different economic terms:

Fig. 2. Schematic view of Levallois core morphology showing the terminology applied here to describe the hierarchically related ‘dorsal’ versus ‘ventral’ surfaces of the bifacial core.

(Eren and Lycett, 2012) have already determined that Eren’s experimental Levallois cores accurately replicate the form of genuine archaeological examples of Levallois core, thus providing a justifiable link between the experimental products and the prehistoric behavioural phenomenon of interest. However, as Brantingham and Kuhn (2001: 752) noted in the case of their study “mathematical models do not attempt to completely ‘replicate’ Levallois, or any other core reduction strategy. They do, however, allow exploration of essential economic and technical features”. Similar caveats must be added to our experiments: we are testing the bounds of what is practically achievable in order to understand potential prehistoric motivating factors underlying patterns of lithic variability, not necessarily replicating what was actually done in all cases.

1. The average debitage mass (g) that was lost at each stage of PLF production, described as a proportion of the starting core mass at each individual stage. 2. The average number of debitage flakes removed at each stage of PLF production. 3. The average debitage mass (g) lost (described as a proportion of the starting core mass at each stage of PLF production) from the ventral versus the dorsal surfaces of the core. 4. The average number of debitage flakes removed from ventral versus dorsal surfaces at each stage of PLF production. 5. The average ratio of PLF cutting edge length (mm) to debitage mass lost (g) at each reduction stage. ‘Cutting edge length’ was measured using methods described in Eren and Lycett (2012). 6. The average ratio of PLF cutting edge (mm) to the average number of debitage flakes produced at each reduction stage. Essentially, these six measures of reduction economy enable us to examine three factors: (1) how much ‘waste’ (on average) is produced at each stage of PLF production relative to others; (2) how the ventral versus dorsal surfaces of the core contribute (relatively) to this ‘waste’ across different stages, and; (3) how much PLF cutting edge is generated (on average) at each reduction stage relative to the mass lost in ‘waste’ for that particular stage. Each stage was

Fig. 3. aed: Examples of chert nodules used in the experiments. Note the range of forms and their deviation from the schematic ellipse shown in Fig. 1.

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

defined as ending when a PLF was removed successfully. Reduction of the core was terminated when the knapper deemed that any future reduction would be unlikely to result in the successful (or safe) production of a PLF from the remaining core or ‘slug’ (Fig. 1). We do not assume that another knapper (i.e. in prehistory) would necessarily have been unable to remove further PLFs. However, since the mass of the slug is not counted as debitage cost against the final flake, the analyses are unaffected by these alternate scenarios. A ManneWhitney U test was used to detect statistically significant (a ¼ 0.05) differences between the six economy measures across the successive reduction stages. The ManneWhitney U test is a conservative non-parametric procedure (i.e. minimizes type I errors), requiring no assumptions of the data in terms of normality of distribution or homogeneity of variances (Dytham, 2003: 101), which is important here given that we are frequently comparing ratio data. These tests were undertaken using SPSSv.20, and exact p-values were reported in all cases. 3. Results: economy of reduction in the experimental Levallois cores The 25 chert nodules produced a total of 3957 flakes comprised of 75 PLFs and 3882 debitage (or ‘waste’) flakes. Figs. 4e8 and Tables 2e7 describe the economy of the Levallois reductions in terms of the six different measures that we employed. Taking each measure of reduction economy in turn, the results can be described as follows: 3.1. Average debitage mass (g) lost at each PLF stage Fig. 4 and Table 2 report the average debitage mass (g) that was lost at each stage of PLF production, described as a proportion of the starting core mass at each individual stage. Fig. 1 graphically illustrates that production of the first PLF from nodules resulted in the highest levels of ‘waste’ (i.e. debitage mass), which is to be expected given that in this stage the initial ‘volumetric’ morphology of the Levallois core is established. However, once the initial Levallois core morphology is in place, removing a second PLF is, on average, significantly (Table 2) less costly than the first in terms of debitage mass lost. Indeed, removing a third PLF requires, on

Fig. 4. Average debitage mass (g) lost described as a proportion of starting core mass at each stage of PLF production. Plot shows that reduction stages 2 and 3 are progressively less wasteful (in terms of debitage flake mass loss). Total debitage flakes (n) ¼ 3882.

2387

Table 2 Average debitage mass (g) lost described as a proportion of starting core mass at each individual stage of PLF production, and results of ManneWhitney U test comparisons across successive stages. Stage

N (cores)

Average ratio of debitage mass per stage (s.d. ¼ standard deviation)

ManneWhitney U test comparisons

1 2

25 24

0.604 (s.d. ¼ 0.0797) 0.4224 (s.d. ¼ 0.1191)

3

18

0.3223 (s.d. ¼ 0.1357)

4

5

0.3401 (s.d. ¼ 0.1117)

e Stage 1 vs. Stage 2 U ¼ 59, p < 0.001 Stage 2 vs. Stage 3 U ¼ 137, p ¼ 0.045 Stage 3 vs. Stage 4 U ¼ 46, p ¼ 0.626

average, even less waste than production of the second PLF, to a statistically significant extent (Table 2). Fig. 4 and Table 2 indicate that, on average, a slightly larger amount of mass was lost in core preparation for removal of a fourth PLF compared to the third PLF. However, this increase in mass loss was not statistically significant (Table 2). 3.2. Average number of number of debitage flakes removed at each stage of PLF production Fig. 5 and Table 3 report the average number of number of debitage flakes removed at each stage of PLF production. These results corroborate those produced from examination of mass loss. That is, each subsequent PLF required fewer flakes to be removed in preparation compared to the preceding PLF. Indeed, as with mass loss, production of a third PLF was, on average, significantly less costly than production of a second PLF (Table 3). Again, production costs in producing a fourth PLF were not, on average, significantly different from those involved in producing the preceding flake. 3.3. Average debitage mass (g) lost from the ventral versus the dorsal surfaces of the core at each stage of PLF production Fig. 6 and Table 4 report the average debitage mass (g) lost (described as a proportion of the starting core mass at each stage of PLF production) from the ventral versus the dorsal surfaces of the core. This analysis thus enables examination of the relative contribution of ventral versus dorsal flakes to the overall economy measures seen in previous analyses. Fig. 6 demonstrates that

Fig. 5. Average number of debitage flakes produced at each stage of PLF production. Plot shows that with each reduction stage, numbers of debitage flakes required decreases with each subsequent PLF stage. Total debitage flakes (n) ¼ 3882.

2388

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

Table 3 Average number of debitage flakes produced at each stage of PLF production, and results of ManneWhitney U test comparisons across successive stages. Stage

N (cores)

Average flake removal (s.d. ¼ standard deviation)

ManneWhitney U test comparisons

1 2

25 24

68.04 (s.d. ¼ 12.25) 51.25 (s.d. ¼ 13.76)

3

18

36.83 (s.d. ¼ 17.29)

4

5

33.50 (s.d. ¼ 10.76)

e Stage 1 vs. Stage 2 total counts U ¼ 109.5, p < 0.001 Stage 2 vs. Stage 3 total counts U ¼ 107, p < 0.005 Stage 3 vs. Stage 4 total counts U ¼ 50, p ¼ 0.807

production. These results show that in terms of raw flake count removals, subsequent stages of PLF production are generally less costly than production of the previous PLF, both for dorsal and ventral surfaces. Consistent with the results of the previous economy measure (i.e. mass loss from each surface), management of the dorsal surface was less costly than for the ventral surface. Moreover, at no stage does management of the ventral (PLF) surface require a statistically greater amount of flaking compared to the preceding stage. 3.5. Average ratio of PLF cutting edge length (mm) to debitage mass lost (g) at each reduction stage

3.4. Average number of debitage flakes removed from ventral versus dorsal surfaces at each stage of PLF production

Fig. 8 and Table 6 report the average ratio of PLF cutting edge length (mm) produced to debitage mass lost (g) at each reduction stage. The results show that PLF edge length subsequently increased relative to debitage mass lost for stages 2 (versus 1) and 3 (versus 2) (Fig. 8), and that in both cases this relative increase in cutting edge is statistically significant (Table 6). Producing a fourth PLF was just as productive (in terms of cutting edge length to mass loss) as the preceding flake (Table 6). This result shows that once the initial Levallois core morphology was established during stage 1, the relative length of cutting edge produced (compared to mass loss) was, at the very least, stable across all subsequent stages of PLF production. Where significant differences were found (Table 6) this was only ever toward a trend of significantly increased cutting edge compared to mass loss. Given the analysis comparing dorsal mass loss versus ventral surface mass loss (Section 3.3 and Table 4), it is important to note that such increases in relative cutting edge appear to be driven by the fact that by the time a second PLF is removed, the ‘volumetric’ construction of the core is established such that removal of a third PLF required relatively little mass removal from the dorsal surface (Table 4, column 5). This latter fact ensured that removal of subsequent PLFs sometimes resulted in statistically greater ratios of cutting edge to mass loss, even though mass lost through ventral re-preparation remained statistically stable across stages.

Fig. 7 and Table 5 report the average number of debitage flakes removed from ventral versus dorsal surfaces at each stage of PLF

3.6. Average ratio of PLF cutting edge (mm) to the average number of debitage flakes produced at each reduction stage

following the initial reduction stage, the rate of mass lost from the dorsal (non-PLF) surface of the core was less than that lost from the ventral (i.e. PLF) surface. Fig. 6 also indicates that the proportion of core mass lost via ventral flakes remained relatively stable across stages, compared to the rate of mass loss from dorsal flakes. This pattern is further supported by the statistical comparison across stages (Table 4). In the case of dorsal mass, once the initial volumetric form of the Levallois core had been established, removing a second or even third flake from the core was significantly less costly than that of the preceding flake (Table 4). Even removing a fourth flake was not significantly more costly in terms of dorsal mass loss compared to the preceding flake (Table 4). These results contrast with those seen for ventral mass lost, where costs of mass loss remain statistically stable across subsequent PLF production stages (Table 4). This result thus shows that once the ‘volumetric’ Levallois core morphology has been established, relatively small amounts of adjustment to the dorsal surface of the core are necessary. Furthermore, although the ventral surface required, on average, a greater amount of mass loss across stages, this mass loss was stable (i.e. statistically indistinguishable) across all subsequent stages of PLF production.

Fig. 9 and Table 7 report the average ratio of PLF cutting edge (mm) to the average number of debitage flakes produced at each reduction stage. These results corroborate those produced from examination of mass loss versus cutting edge seen in the preceding economy measure. The results show that, on average, PLF edge length subsequently increases relative to the number of flakes removed across stages 2 (versus 1) and 3 (versus 2) (Fig. 9). Again, this result shows that once the initial Levallois core morphology was established, the relative length of cutting edge produced (relative to the number of debitage or ‘waste’ flakes produced) was, at the very least, statistically stable across all subsequent stages of PLF production. Where significant differences were found (Table 7) this was only ever toward a trend of significantly increased cutting edge length compared to number of debitage flakes required at each stage. 4. Discussion

Fig. 6. Average debitage mass (g) lost described as a proportion of starting core mass at each stage of PLF production for ventral flakes (triangles) versus dorsal flakes (squares). Plot shows that following the initial reduction stage, the rate of mass loss is less for dorsal as opposed to ventral flakes. Note also, that proportion of core mass lost via ventral flakes remains relatively stable across stages, compared to the rate of mass loss from dorsal flakes.

The solution to obtaining simple flake cutting tools from stone raw material that exhibits the properties of conchoidal fracture would appear to be relatively simple: strike a suitably stable surface of the core, which exhibits an angle of less than 90 , at a point of depth (and with sufficient force) in an area exhibiting a specific level of mass in order to initiate a flake of suitable size. Indeed,

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

2389

Table 4 Average debitage mass (g) lost described as a proportion of starting core mass at each stage of PLF production for ventral flakes and dorsal flakes, and results of ManneWhitney U test comparisons across successive stages. Stage

N (cores)

Average dorsal mass (s.d. ¼ standard deviation)

Average ventral mass (s.d. ¼ standard deviation)

ManneWhitney U tests: Dorsal mass

ManneWhitney U tests: Ventral mass

1 2

25 24

0.318 (s.d. ¼ 0.07) 0.142 (s.d. ¼ 0.08)

0.249 (s.d. ¼ 0.05) 0.232 (s.d. ¼ 0.08)

3

18

0.085 (s.d. ¼ 0.04)

0.191 (s.d. ¼ 0.12)

4

6

0.082 (s.d. ¼ 0.05)

0.209 (s.d. ¼ 0.10)

e Stage 1 vs. Stage 2 U ¼ 31, p < 0.001 Stage 2 vs. Stage 3 U ¼ 127, p ¼ 0.023 Stage 3 vs. Stage 4 U ¼ 50, p ¼ 0.820

e Stage 1 vs. Stage 2 U ¼ 263, p ¼ 0.469 Stage 2 vs. Stage 3 U ¼ 166, p ¼ 0.211 Stage 3 vs. Stage 4 U ¼ 48, p ¼ 0.721

migrating platform cores, involving the consecutive recognition of suitable platforms, angles, areas of mass, etc., offered hominins a reliable means of reducing cores into utilizable flakes or blanks for much of prehistory (Toth, 1985; Delagnes and Roche, 2005). However, hominins could also potentially configure flaking episodes in a more strategic manner, thus introducing concepts of shaping into core reduction and, in turn, controlling angles, points of mass, etc. in a more deliberate manner. Such strategies may, in principle, have led to hominins being able to exert a level of control over variables such as raw material economy (e.g., time, effort, wastage) or factors such as the amount of cutting edge produced (Baumler, 1995; Brantingham and Kuhn, 2001; Brantingham, 2010). On the basis of theoretical mathematical modelling, Brantingham and Kuhn (2001) proposed that Levallois reduction represented one such strategy on the part of hominins; specifically, in terms of economization of waste and the generation of flakes with high amounts of cutting edge. Inevitably, however, it is important that such models and their key assumptions are tested under conditions that more closely approximate the material reality and challenges of prehistoric knapping. Experiments in this regard may form an important ‘middle range’ link (sensu Binford,1983), enabling both the modelling of what was potentially achievable during prehistory, and the testing of key assumptions in more abstract models, such as those derived from mathematical modelling. This is especially important in the case of Brantingham and Kuhn’s (2001) suggestions for Levallois reduction, given that their models did not account for the phase(s) of core preparation that must take place between subsequent phases of longitudinal flaking, at least in the case of ‘classic’ (lineal) Levallois.

Fig. 7. Average number of ventral (triangles) versus dorsal (squares) debitage flakes produced at each stage of PLF production. Plot shows that with each subsequent reduction stage, the number of debitage flakes required to produce a PLF is less for dorsal as opposed to ventral flakes.

Here, we examined the economy of reduction in a series of experimentally produced Levallois reduction sequences. As per the assumptions of Brantingham and Kuhn’s (2001) model, our cores conformed to Boëda’s (1995) ‘volumetric’ concept of Levallois core morphology. Our analysis of economy (i.e. core mass lost through preparation) in the experimental Levallois reductions demonstrated that removing subsequent PLFs was never statistically more costly than establishing the initial core morphology. Indeed, wherever statistical differences could be detected in our economy measures, it was always in the direction of significantly greater levels of economic efficiency in terms of reducing core mass ‘waste’ and/or increased PLF cutting edge length to mass loss. The results also demonstrated that this was because following the removal of the initial PLF, relatively little modification of the cores’ dorsal surfaces was, on average, required in order to remove subsequent PLFs. Indeed, removing a third PLF was significantly less costly than removal of a second PLF, precisely because at that point in the reduction sequence the ‘volumetric’ construction of the core is so well established such that only relatively minor amounts of mass were required to be removed from the cores’ dorsal surfaces. Although, on average, comparable amounts of modification to the ventral surface of the core was required across all subsequent stages of removal, this was found never to be significantly greater than that of the initial stage. Importantly, therefore, our results found that a key assumption of Brantingham and Kuhn’s (2001) mathematical model of Levallois reduction could be upheld, even in the case of preferential Levallois. That is, once the initial ‘volumetric’ construction of the preferential Levallois core has been established, the economic cost of core re-preparation (i.e. non-PLF mass loss) is, at the very least, stabilized across all subsequent stages. Overall, our experimental results support, therefore, Brantingham and Kuhn’s (2001) conclusions that economic considerations may logically have played a part in motivating hominins to produce Levallois-style core reduction sequences. This is further supported given that maximizing the surface area of flakes obtained from a core of a given size can logically be expected to have been an important consideration to producers of flake-based toolkits, since flakes of a larger surface area are likely to have extended use-life (Kuhn, 1994). Indeed, it is important to note that these results are not mutually exclusive to Eren and Lycett’s (2012) analysis, which indicated that Levallois flakes removed from ‘classic’ tortoise-style cores possess specific shape properties (independent of size) that may also have logically made them preferable items in terms of functional properties (i.e. greater capacity for retouch, robustness of working edge, and an evenness of weight distribution during use). Indeed, Brantingham and Kuhn (2001: 758e759) noted that within ecologically competitive situations, where evolutionary theory predicts that potential benefits must always be weighed against incurred costs, instances of behaviour resulting in “coinciding optima” (i.e. the cooccurrence of multiple independent benefits) would represent a distinct “adaptive peak” when compared against alternative strategies.

2390

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

Table 5 Average number of ventral versus dorsal debitage flakes produced at each stage of PLF production, and results of ManneWhitney U test comparisons across stages. Stage

N (cores)

Average ventral removal (s.d. ¼ standard deviation)

Average dorsal removal (s.d. ¼ standard deviation)

ManneWhitney U tests: Dorsal flake counts

ManneWhitney U tests: ventral flake counts

1 2

25 24

33.3 (s.d. ¼ 7.71) 27.6 (s.d. ¼ 10.31)

34.7 (s.d. ¼ 9.52) 23.5 (s.d. ¼ 10.37)

3

18

20.6 (s.d. ¼ 11.6)

16.2 (s.d. ¼ 9.63)

4

5

20.8 (s.d. ¼ 9.6)

12.6 (s.d. ¼ 5.2)

e Stage 1 vs. Stage 2 U ¼ 122.5, p ¼ 0.001 Stage 2 vs. Stage 3 U ¼ 117.5, p ¼ 0.011 Stage 3 vs. Stage 4 U ¼ 47, p ¼ 0.660

e Stage 1 vs. Stage 2 U ¼ 197, p ¼ 0.039 Stage 2 vs. Stage 3 U ¼ 144, p ¼ 0.068 Stage 3 vs. Stage 4 U ¼ 53.5, p ¼ 0.987

In other words, if the combined and interacting factors of economy and flake utility were important factors in the lives of prehistoric hominins, then based on the combined implications of the results we present in this paper, the conclusions of Brantingham and Kuhn’s (2001) modelling work, and those of Eren and Lycett (2012), then the widespread adoption of Levallois is not merely to be expected, but should even be predicted. This, therefore, emphasizes all the more forcefully the importance of examining potential factors surrounding why Levallois did not emerge earlier in the record (Wynn and Coolidge, 2004; Lycett and von Cramon-Taubadel, 2013), or indeed, why Levallois is geographically and temporally discontinuous, even in the later Pleistocene (cf. Movius, 1949; Schick, 1998; Gao and Norton, 2002; Lycett and Norton, 2010; Bar-Yosef et al., 2012). It is important to note that other commentators, in contrast to Brantingham and Kuhn (2001), have frequently described Levallois as a ‘wasteful’ method of flake production, at least if the aim is only to produce PLFs (e.g., Wymer, 1982: 117; Noble and Davidson, 1996: 200; Sheppard and Kleindienst, 1996: 180). However, what the results we discuss here emphasize (in combination with Brantingham and Kuhn’s (2001) original consideration of Levallois in these terms) is that concepts of ‘waste’ and by implication, ‘economy’ can only be conceived and modelled under strictly defined concepts of costs versus potential benefits. Indeed, over four decades ago, Clarke (1968: 28) emphasized that in the absence of appropriate comparative standards, where the necessary parameters required to give an otherwise relative entity (e.g. ‘rare’, ‘few’, ‘thick’, ‘thin’) its required specificity, such “non-specific generalizations” are essentially meaningless. Hence, more explicit mathematical modelling of various reduction sequences and terms such as ‘debitage’, ‘waste’, etc., under formally parameterized economic terms, is potentially profitable for lithic analysis, especially in terms of modelling factors that might motivate particular

reduction sequences under specific given situational parameters (e.g. Kuhn, 1994; Brantingham, 2010). As emphasized here, this is especially so when the parameters underlying such models can be further verified using an experimental methodology. However, these points aside, it must be noted that the ‘waste’ produced during Levallois production need not necessarily have been ignored as a further source of potential cutting tools, even under circumstances where hominins might, for various functional reasons (Eren and Lycett, 2012), have rationally preferred PLFs over other categories of flake. It should be noted that our results also suggest the underlying reason why Levallois reduction (sensu stricto) results in cores that display morphologically distinctive patterns of shape variation compared to so-called ‘Victoria West’ style cores (Lycett et al., 2010). For many years, comparisons between Lower Palaeolithic prepared cores from central South Africa and associated with the production of Acheulean cleavers and handaxes, had drawn direct comparison with later Levallois cores of the African MSA and Eurasian Middle Palaeolithic (e.g., Goodwin, 1934; Leakey, 1936; Van Riet Lowe, 1945; Clark, 1959; Rolland, 1995; Mitchell, 2002). However, more recent examinations of Victoria West artefacts have tended to emphasize the distinctiveness of Victoria West (e.g., Lycett, 2009; Sharon, 2009). Indeed, direct 3D geometric morphometric comparisons of Levallois core surfaces with Victoria West core surfaces found that Levallois cores tended to show comparable (i.e. isometric) patterns of shape variation despite variability in raw size (Lycett et al., 2010). In contrast, Victoria West cores appeared to display some aspects of shape variation that differed from those of Levallois, even compared with Levallois examples to which they were statistically comparable in terms of size (Lycett et al., 2010). The consistent pattern of shape variation seen in the Levallois core surfaces, independent of size, would be consistent with them displaying an isometric pattern of reduction. That is, as the reduction sequence proceeds through repeated phases of PLF production and cores get smaller, certain aspects of shape are maintained. Conversely, given that Victoria West cores were used to produce large (10 cm long) blanks for cleavers and handaxes, this leads to differences emerging in their shape variation, distinct from that of classic Levallois. Hence, in the light of these results, differences in terms of differing economic considerations governing raw material Table 6 Average ratio of PLF cutting edge (mm) to debitage mass loss (g) at each reduction stage, and results of ManneWhitney U test comparisons across stages.

Fig. 8. Average ratio of PLF cutting edge (mm) to debitage mass loss (g) at each reduction stage. Note that PLF cutting edge length increases relative to debitage mass in stages 2 and 3.

Stage

N (cores)

PLF cutting edge/debitage mass (s.d. ¼ standard deviation)

ManneWhitney U test comparisons

1 2

25 24

0.165 (s.d. ¼ 0.05) 0.579 (s.d. ¼ 0.29)

3

18

1.560 (s.d. ¼ 1.28)

4

5

1.559 (s.d. ¼ 0.98)

e Stage 1 vs. Stage 2 U ¼ 5, p < 0.001 Stage 2 vs. Stage 3 U ¼ 74, p < 0.001 Stage 3 vs. Stage 4 U ¼ 51, p ¼ 0.871

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

Fig. 9. Average ratio of PLF cutting edge (mm) to average number of debitage flakes produced at each reduction stage. Note that PLF cutting edge length increases relative to number of debitage flakes in stages 2 and 3.

Table 7 Average ratio of PLF cutting edge (mm) to average number of debitage flakes produced at each reduction stage, and results of ManneWhitney U test comparisons across stages.

2391

noted that our experiments, as with Brantingham and Kuhn’s (2001) original formulation, represent an optimization model for Levallois reduction under specific economic criteria. However, in so doing, these results reinforce that modelling alternative reduction sequences formally, and in terms of the specific economic parameters (costs versus benefits) that might potentially motivate patterns observable in lithic data is potentially useful (Torrence, 1989), just as optimality models in ecology lead to an understanding of the selective parameters that potentially influenced the evolution of biological adaptive features (Hecht and Sober, 1994). Indeed, in the light of our results, it appears that several distinct sets of features relating both to the economy of the reduction and its products, could logically have motivated the widespread occurrence of Levallois-style reduction during the later Pleistocene. In this sense, although on a previous occasion we have asked the question, “why Levallois” (Eren and Lycett (2012); at this point, the more appropriate question might be, why not Levallois? Acknowledgements We thank Richard Klein, Noreen von Cramon-Taubadel and three anonymous reviewers for helpful comments. We are also grateful to the Leverhulme Trust for financial support.

Stage

N (cores)

PLF cutting edge/Average flake (s.d. ¼ standard deviation)

ManneWhitney U test comparisons

References

1 2

25 24

3.55 (s.d. ¼ 0.88) 3.80 (s.d. ¼ 1.59)

3

18

6.04 (s.d. ¼ 4.78)

4

5

5.15 (s.d. ¼ 4.16)

e Stage 1 vs. Stage 2 U ¼ 295, p ¼ 0.925 Stage 2 vs. Stage 3 U ¼ 138, p ¼ 0.048 Stage 3 vs. Stage 4 U ¼ 41, p ¼ 0.415

Bar-Yosef, O., Eren, M.I., Yuan, J., Cohen, D.J., Li, Y., 2012. Were bamboo tools made in prehistoric Southeast Asia? an experimental view from South China. Quaternary International 269, 9e21. Baumler, M.F., 1995. Principles and properties of lithic core reduction: implications for Levallois technology. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 11e23. Binford, L.R., 1983. In Pursuit of the Past. Thames & Hudson, London. Boëda, E., 1994. Le Concept Levallois: Variabilité des Méthodes. Centre de la Recherche Scientifique (CNRS), Paris. Boëda, E., 1995. Levallois: a volumetric construction, methods, a technique. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 41e68. Bradley, B., 1977. Experimental Lithic Technology with Special Reference to the Middle Palaeolithic. Unpublished Ph.D. dissertation, Department of Archaeology, Cambridge University. Brantingham, P.J., Kuhn, S.L., 2001. Constraints on Levallois core technology: a mathematical model. Journal of Archaeological Science 28, 747e761. Brantingham, P.J., 2010. The mathematics of chaînes opératoires. In: Lycett, S.J., Chauhan, P.R. (Eds.), New Perspectives on Old Stones: Analytical Approaches to Palaeolithic Technologies. Springer, New York, pp. 183e206. Clark, J.D., 1959. The Prehistory of Southern Africa. Penguin, London. Clarke, D.L., 1968. Analytical Archaeology. Methuen, London. Commont, V., 1909. L’industrie moustérianne dans la région du Nord de la France. Congès Préhistorique de France 5ème session. Bureaux de la Société Préhistorique de France, Paris, pp. 115e157. Delagnes, A., Roche, H., 2005. Late Pliocene hominid knapping skills: the case of Lokalalei 2C, West Turkana, Kenya. Journal of Human Evolution 48, 435e472. Dytham, C., 2003. Choosing and Using Statistics: a Biologist’s Guide, second ed. Blackwell Science, Oxford. Eren, M.I., Bradley, B., 2009. Experimental evaluation of the Levallois “core shape maintenance” hypothesis. Lithic Technology 34, 119e125. Eren, M.I., Lycett, S.J., 2012. Why Levallois? A morphometric comparison of experimental ‘preferential’ Levallois flakes versus debitage flakes. PLoS ONE 7 (e29273), 1e10. Eren, M.I., Bradley, B., Sampson, C.G., 2011a. Middle Paleolithic skill-level and the Individual knapper: an experiment. American Antiquity 76, 229e251. Eren, M.I., Lycett, S.J., Roos, C., Sampson, C.G., 2011b. Toolstone constraints on knapping skill: Levallois reduction with two different raw materials. Journal of Archaeological Science 38, 2731e2739. Gao, X., Norton, C.J., 2002. A critique of the Chinese ‘Middle Palaeolithic’. Antiquity 76, 397e412. Goodwin, A.J.H., 1934. Some developments of technique during the earlier Stone Age. Transactions of the Royal Society of South Africa 21, 109e123. Hecht, S., Sober, E., 1994. Optimality models and the test of adaptationism. American Naturalist 143, 361e380. Hublin, J.-J., 2009. The origin of Neandertals. Proceedings of the National Academy of Sciences USA 106, 16022e16027. Kuhn, S.L., 1994. A formal approach to the design and assembly of mobile toolkits. American Antiquity 59, 426e442.

usage and the pattern of the overall reduction sequence, would appear to explain the distinctive patterns of shape variation displayed in Levallois (sensu stricto) versus Victoria West cores. 5. Conclusions On the basis of theoretical and mathematical modelling, Brantingham and Kuhn (2001) proposed that Levallois core morphology was motivated by economic considerations relating to raw material usage and the maximization of cutting edge produced for cores of a given size. However, despite logical appeal, the abstract character of this modelling work did not take account of certain practical problems associated with the fundamental material character of Levallois reduction in actual stone nodules. In particular, the re-preparation of core surfaces between Levallois flake removals, and the fact that with “each succeeding Levallois routine, the core becomes progressively smaller, presenting a new set of problems” (Wynn and Coolidge, 2010: 91). Our experimental analyses demonstrated, however, that a fundamental assumption of Brantingham and Kuhn’s (2001) model can, at least in principle, be upheld under the practical challenges of replicating Levallois reduction in stone. That is, once the initial Levallois core morphology has been established, the economic costs of mass loss can, at the very least, be stabilized across all subsequent stages. Indeed, in some cases, removing a subsequent Levallois flake can be statistically more economical in terms of mass loss through preparation than for the preceding flake. Hence, although new challenges may arise as Levallois reduction proceeds, these need not necessarily ever be large enough that it results in significantly greater levels of raw material usage than in preceding stages. It must be

2392

S.J. Lycett, M.I. Eren / Journal of Archaeological Science 40 (2013) 2384e2392

Leakey, L.S.B., 1936. Stone Age Africa: an Outline of Prehistory in Africa. Humphrey Milford, London. Lycett, S.J., Norton, C.J., 2010. A demographic model for Palaeolithic technological evolution: the case of East Asia and the Movius Line. Quaternary International 211, 55e65. Lycett, S.J., von Cramon-Taubadel, N., 2013. A 3D morphometric analysis of surface geometry in Levallois cores: patterns of stability and variability across regions and their implications. Journal of Archaeological Science 40 (3), 1508e1517. Lycett, S.J., von Cramon-Taubadel, N., Gowlett, J.A.J., 2010. A comparative 3D geometric morphometric analysis of Victoria West cores: implications for the origins of Levallois technology. Journal of Archaeological Science 37, 1110e1117. Lycett, S.J., 2009. Are Victoria West cores “proto-Levallois”? A phylogenetic assessment. Journal of Human Evolution 56, 175e191. Mitchell, P., 2002. The Archaeology of Southern Africa. Cambridge University Press, Cambridge. Moncel, M.-H., Moigne, A.-M., Sam, Y., Combier, J., 2011. The emergence of Neanderthal technical behavior: new evidence from Orgnac 3 (Level 1, MIS 8), southeastern France. Current Anthropology 52, 37e75. Movius, H.L., 1949. The Lower Palaeolithic cultures of Southern and Eastern Asia. Transactions of the American Philosophical Society 38 (4), 329e426. Noble, W., Davidson, I., 1996. Human Evolution, Language and Mind: a Psychological and Archaeological Enquiry. Cambridge University Press, Cambridge. Pelegrin, J., 2005. Remarks about archaeological techniques and methods of knapping. In: Roux, V., Bril, B. (Eds.), Stone Knapping: the Necessary Conditions for a Uniquely Hominin Behaviour. McDonald Institute Monographs, Cambridge, pp. 23e33. Rolland, N., 1995. Levallois technique emergence: single or multiple? A review of the Euro-African record. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 333e359.

Schick, K., 1998. A comparative perspective on Paleolithic cultural patterns. In: Akazawa, T. (Ed.), Neandertals and Modern Humans in Western Asia. Plenum Press, New York, pp. 449e460. Schlanger, N., 1996. Understanding Levallois: lithic technology and cognitive archaeology. Cambridge Archaeological Journal 6, 231e254. Scott, B., 2011. Becoming Neanderthals: the Earlier British Middle Palaeolithic. Oxbow Books, Oxford. Sharon, G., 2009. Acheulian giant-core technology. Current Anthropology 50, 335e367. Sheppard, P.J., Kleindienst, M.R., 1996. Technological change in the Earlier and Middle Stone Age of Kalambo Falls. African Archaeological Review 13, 171e196. Smith, R.A.,1911. A Palaeolithic industry at Northfleet, Kent. Archaeologica 62, 512e532. Torrence, R. (Ed.), 1989. Time, Energy and Stone Tools. Cambridge University Press, Cambridge. Toth, N., 1985. The Oldowan reassessed: a close look at early stone artefacts. Journal of Archaeological Science 12, 101e120. Tryon, C.A., McBrearty, S., Texier, J.-P., 2006. Levallois lithic technology from the Kapthurin Formation, Kenya: Acheulian origin and Middle Stone Age diversity. African Archaeological Review 22, 199e229. Van Peer, P., 1992. The Levallois Reduction Strategy. Prehistory Press, Madison, Wisconsin. Van Riet Lowe, C., 1945. The evolution of the Levallois technique in South Africa. Man 45, 49e59. Wilkins, J., Pollarolo, L., Kuman, K., 2010. Prepared core reduction at the site of Kudu Koppie in northern South Africa: temporal patterns across the Earlier and Middle Stone Age boundary. Journal of Archaeological Science 37, 1279e1292. Wymer, J., 1982. The Palaeolithic Age. Crom Helm, London. Wynn, T., Coolidge, F.L., 2004. The expert Neandertal mind. Journal of Human Evolution 46, 467e487. Wynn, T., Coolidge, F.L., 2010. How Levallois reduction is similar to, and not similar to, playing chess. In: Nowell, A., Davidson, I. (Eds.), Stone Tools and the Evolution of Human Cognition. University Press of Colorado, Boulder, pp. 83e103.