Figure 3.26 Relationship between the thickness and surface area ofLevallois flakes recorded in a number of French Middle Palaeolithic assemblages, after Dibble 1985. Dibble suggests that the differences between the flake dimensions are related mainly to variations in the size and quality of the local raw material supplies employed on the different sites.

1988b). Explaining this variability remains one of the most challenging issues in current studies of the Middle Palaeolithic.

As discussed in Chapter 10, there is little doubt that some of these variations are related to the character and quality of the lithic raw material supplies available in different geographical contexts. Several workers have pointed out that some of the most frequent and elaborate occurrences of various Levallois techniques occur mainly in areas where local flint supplies are both relatively abundant and available in the form of relatively large, high quality nodules (Fish 1981; Bordes 1947, 1950b, 1954a, 1984: 169; Turq 1989a; Van Peer 1992). This can be seen in many of the industries from the flint-rich areas of northern France, in many parts of the Middle East and North Africa, and in some of the more localized occurrences of high quality materials in areas such as the Bergerac region of southwestern France and parts of western Provence (Fig. 3.26). But as Bordes emphasized (e.g. 1968a: 138), it is difficult to see this factor as more than a partial explanation for the variable occurrence of different flaking techniques. Bordes himself pointed out that in several contexts relatively large and elaborate forms of Levallois flakes can be shown to have been produced from fairly coarse-grained, apparently intractable raw materials, such as quartzite or even finegrained sandstone (Green 1984). There are equally frequent examples of the use of relatively sophisticated and abundant Levallois techniques in areas where the available flint supplies are either scarce or present in the form of small, irregular nodules - as for example the Grotte Vaufrey, Abri Caminade, Pech de 1' Aze and other sites in the Dordogne valley or at Le Moustier, La Rochette and Fonseigner in the valleys of the Vezere and Dronne. None of these areas are characterized by particularly high quality flint supplies (Dibble 1985; Geneste 1985; Bordes 1968a: 138). Most striking of all, perhaps, is the dramatic way in which the relative use of various Levallois techniques can often be seen to vary between different occupation levels within precisely the same occupation sites - for example between the different stratigraphic levels at Combe-Grenal, at the Roc-de-Marsal and at Pech de l'Aze sites II and IV (Rolland 1988b: 165-9). Any suggestion that these different flaking techniques were related in a simple, direct and spontaneous way to the varying availability and flaking qualities of local raw materials would therefore seem to be contradicted directly by the evidence from many Middle Palaeolithic sites. Evidently, some factors other than the simple character and accessibility of local raw materials were involved in the selection and variable use of different flaking techniques.

4. Finally, some useful insights into the character and operation of different primary flaking strategies have been provided by recent experimental approaches to the replication of Middle Palaeolithic techniques. The essential aims of these studies are to reproduce as accurately as possible strategies of flake production documented in particular archaeological contexts, and in this way to investigate specific problems encountered in dealing with different varieties of raw materials, and the characteristic forms of debitage generated during the different stages of flake and core production. In other words, these studies aim to clarify the conceptual planning strategies involved in the design and operation of different flaking techniques and to generate specific predictions as to how these different flaking strategies may be reflected in the resulting lithic residues.

At present this research is still at an early stage and virtually all experimental studies so far reported in any detail in the literature have been focused on the various forms of Levallois techniques. The most fully published results are those applied by Geneste (1985: 203-70) to approximately 40 different nodules, comprising several different varieties of flint from southwestern France. Although Geneste's results were published before the recent attempts by Boeda to redefine and reclassify major variants of Levallois techniques, it would seem that most of Geneste's experiments were oriented towards the production of centripetally prepared Levallois flakes, coinciding essentially with Boe-da's (1988a) categories of either lineal or recurrent centripetal techniques (Fig. 3.27). The studies by Shchelinskii, based on a total of around 60 experimental cores, have been published more briefly but were oriented towards the production of both radial and more elongated convergent or laminar flakes, corresponding broadly with Boeda's categories of unipolar and bipolar recurrent techniques (Shchelinskii 1974, 1983, summarized in Plisson 1988). Despite some apparent differences in the precise design and organization of the flaking experiments, a number of fairly clear and consistent patterns seem to emerge from both of these studies (see also Bradley 1977; Bradley & Sampson 1986; Van Peer 1992):

(a) One point emphasized by both Geneste and Shchelinskii is that Levallois techniques, however tightly controlled and designed, can only produce a limited component of typically Levallois flake products in relation to total quantities of flake debitage generated during different flaking sequences. Geneste (1985: 253) reports that in his experiments the overall percentages of Levallois, as opposed

Figure 3.27 Metrical parameters of a succession offtakes produced by Geneste during the experimental replication of Levallois flaking techniques. Phase 1 represents the initial phase of core preparation and decortication, while phase 2 represents the main succession of flake removals. After Geneste 1985.

Figure 3.27 Metrical parameters of a succession offtakes produced by Geneste during the experimental replication of Levallois flaking techniques. Phase 1 represents the initial phase of core preparation and decortication, while phase 2 represents the main succession of flake removals. After Geneste 1985.

to non-Levallois, flakes covered a fairly wide range but centred on an average figure of 18 percent. The results reported by Shchelinskii point broadly to the same conclusion. Assessed in terms of potentially 'usable' flakes, he reports that the different experimental sequences tended to yield an average of around 20 percent of fully Levallois flakes in the case of both radially prepared and more convergent techniques - though the ranges recorded in individual cores could vary from less than 15 percent to as high as 40 percent (see Plisson 1988). Interestingly, he reported a rather lower success rate of around 11-16 percent in attempting to produce more elongated flake blanks from longitudinally prepared cores. The implication of these studies is that no attempt at producing typically Levallois flake blanks from original, unworked nodules is likely to produce more than approximately one-fifth of technically Levallois products among the overall range of knapping debris generated on the flaking sites (Fig. 3.27).

(b) Similar regularities emerge in the production of various forms of cortical flakes in the course of different core reduction sequences. Again Geneste reports that these frequencies can vary within fairly wide limits but seem to centre on an average value of around 30 percent over the flaking experiments as a whole. He goes on to suggest that a systematic comparison of the relative frequencies of cortical flakes, compared with the frequencies of much smaller debitage flakes, can be used to predict fairly accurately the kind of nodules employed in the use of different flaking strategies (Geneste 1985: Fig. 73). One particularly interesting result reported by Geneste is that the relative frequencies of cortical as opposed to fully Levallois flakes do not seem to have been significantly influenced (in his studies) by the specific flaking qualities of the different types of flints employed in the experiments - comprising, as noted above, at least three distinct vari eties of flint from separate outcrops in the Perigord region (Geneste 1985: 254). He stresses, however, that much more extensive experiments would be needed to assess the overall potential impact of different raw material types on the quality and efficiency of production of different flaking strategies.

(c) Finally, Geneste reports some rather more predictable results relating to the overall size and character of the flakes produced during his experiments. The conclusion, not surprisingly, is that typically Levallois flakes tend to be substantially larger than those produced during the various preparatory stages of core flaking and also tend to show more complex patterns of flake scars on their dorsal surfaces (Figs 3.27, 3.28) (Geneste 1985: 259-68). Both these variables must inevitably be related to some extent to the sizes of the original nodules, as well as to the relative intensity of the deliberate preparation applied before the removal of the individual flakes. Nevertheless, his own experimental figures of an average of 7.4 dorsal flake scars per flake - rising for the largest flakes to around 11-16 scars -provides a useful guide to the patterns which can be achieved by carefully controlled use of Levallois techniques. He goes on to point out that equally complex patterns of dorsal-scar preparation are reflected in the material recovered from several Middle Palaeolithic sites in both Europe and the Middle East, for example in certain levels of the Tabun Cave in Israel, and at Le Tillet and Pech de 1'Aze IV in France (Table 3.3: Geneste 1985: 266-7). The implication is that many of the Middle Palaeolithic populations in Eurasia achieved no less highly controlled mastery of Levallois techniques than those attained in the recent experimental studies of modern flint knap-pers!

All these recent experimental approaches to the replication and analysis of Middle Palaeolithic flaking techniques could be carried much further. It would be particularly interesting to know how far variations in the

Table 3.3

Average numbers of dorsal flake scars recorded on samples of Lev alio is flakes in different Middle Palaeolithic assemblages in France and the Middle East, according to Dibble (1983): 58

Table 3.3

Average numbers of dorsal flake scars recorded on samples of Lev alio is flakes in different Middle Palaeolithic assemblages in France and the Middle East, according to Dibble (1983): 58



Average no. of

Was this article helpful?

0 0

Post a comment