Transcript
Understanding Levallois: Lithic Technology and CognitiveArchaeology Nathan Schlanger The Levallois technique has attracted much ‘cognitive’ attention in the past decades. Many archaeologists argue that both the products and the procedure of this Palaeolithictechnique have been clearly predetermined by the prehistoric flintknappers. Attempts haverecently been made to challenge this notion of predetermination by reference to rawmaterial and ‘technological’ constraints. The aim of this article is to assess the grounds onwhich these claims have been advanced, and then work towards a better establishment of the cognitive implications of Levallois manufacture. Latest developments in the techno-logical understanding of Levallois are presented in their context, and then put to workthrough a detailed case study: the analysis, in quantitative and qualitative terms, of acomprehensively refitted Levallois core from the 250,000 year-old site of Maastricht-Belvédère, in the Netherlands. By reconstructing and following the sequence of work onthis highly productive core, it can be shown that its knapping did not simply entail theexecution of a pre-set program, nor did it respond in an adventitious manner to externalconstraints. Rather, it is argued that the course of action was a structured and goal-oriented one, a generative interplay between the mental and material activities of theancient flintknapper. Cambridge Archaeological Journal 6:2 (1996), pp. 231–54indeed many feel a certain embarrassment in dis-cussing them. How they came to acquire such a con-ventional status is an interesting question in its ownright, but there is no doubting the consequence; be-cause these cognitive claims are frequently iteratedand upheld as incontrovertible truisms (or well-wornclichés), there is a natural tendency to take theirspecific establishment for granted, and concentrateinstead on their wider implications. In doing so, how-ever, there may be a risk of putting the cart ahead of the horse. Profitable and gratifying as it is to engagein broad-scale speculations, it might be worthwhileto start by assessing the grounds on which suchclaims are advanced and adhered to. A critical-his-torical examination of this kind could help us cor-roborate or confirm those claims, and also lead us toa better-grounded understanding of technical andintellectual aspects of human prehistory. With this but there is no denying the prominent place occu-pied by this flintknapping technique in the archaeo-logical imagination. Innumerable indeed are theprehistorians and palaeoanthropologists who believethat the Levallois technique, whose desired producthas been ‘predetermined by special preparation priorto its detachment from the core’ (Bordes 1950, 21;1961), is one that required of its makers — and there-fore can reveal to us — exceptional cognitive abili-ties. On the force of such conviction, this Lower andMiddle Palaeolithic flintknapping technique has beenrepeatedly enlisted in reconstructions of human in-tellectual evolution, and discussed as a possible in-dicator of conceptualization, abstraction, intelligence,language, etc., among its pre- Homo sapiens practi-tioners.These propositions sound familiar enough, and T alk of a Levallois mystique may be exaggerated,231 232Nathan Schlangeraim in mind, I propose in this article to highlight adifficulty surrounding cognitive claims for Levallois,and then work towards a solution.The difficulty, succinctly put, is that many suchcognitive claims effectively imply or presuppose aseparation between thought and action in flint-knapping. This is the case with what I will call here‘standard’ claims, where the Levallois knapper isattributed a clear mental image of the product andthe procedure to be realized. It also underpins morerecent ‘reactionary’ claims, where the Levalloisknapper is considered to be responding, case by case,to changing external circumstances and constraints.An examination of the literature shows that thesecontrasting claims are very often promoted in thewake of the classic definition of Levallois, and infunction of the evolutionary scenario in which theyare enlisted. Having presented the difficulty in theseterms, a solution becomes more readily perceptible: by setting aside traditional archaeological preoccu-pations, even if temporarily, room should be madefor a dedicated technological understanding of Levallois, one which could in turn be brought to bear on the cognitive activities involved in the proc-ess of its manufacture.This is not to say that such flintknapping proc-esses were not known and described in the litera-ture. It is rather that they often failed to be adequatelyproblematized and disseminated, and thus remainedsomewhat under-valued and auxiliary in archaeo-logical research. In recent years various practical andtheoretical developments have contributed to changethis state of affairs, and our knowledge of theLevallois technique has considerably increased, no-tably with the work of E. Boëda. These technologicaladvances are discussed here in some detail, and thenput to work through an informative case-study: theanalysis of a comprehensively refitted Levallois corefrom the early Middle Palaeolithic site of Maastricht-Belvédère (Netherlands). Besides illustrating aspectsof controlled predetermination, variability and pro-ductivity in Levallois, this analysis makes it pos-sible to follow the actual sequence of work on thecore and then identify an overarching principle bywhich it has been repeatedly prepared. Without be-ing scripted in advance or oblivious to external cir-cumstances, this principle acts as an enabling imagethat structures the emergent course of action: its iden-tification in a concrete archaeological example makesit possible to overcome the above noted separation between action and thought, and thus help us un-derstand the interplay between mental and materialactivities in flintknapping Levallois. ‘Standard’ and ‘reactionary’ claims The earliest and by far most common cognitive claimsto be made with regard to Levallois are those I havelabelled ‘standard’: they are indeed those that springto mind with the very mention of the term ‘Levallois’.In assessing these claims it should be recalled thatwhile they are thoroughly conventional nowadays,they are actually of fairly recent srcin. The Levalloistechnique has been described at least since the laterhalf of the nineteenth century, 1 and yet for long,seekers of prehistoric intellectual phenomena con-centrated almost exclusively on aesthetically pleas-ing handaxes, as well as evidence of fire, and, of course, parietal and mobiliary art ( cf. Bergson 1907;Boule 1923; Schmidt 1936; Oakley 1949). It can besaid that the Levallois technique acquired its presentcognitive connotations only with the work of FrançoisBordes. In comparison with previous discussions(Commont 1909; Breuil 1937; Breuil & Lantier 1951;Burkitt 1933; Van Riet Lowe 1945) Bordes’ contributionwas double. He defined the Levallois technique —‘predetermination by special preparation’ — on the basis of his own flintknapping experience, and, inthe same 1950 article, he included Levallois productsin his Lower and Middle Palaeolithic type list. Theimplications of these steps will be presently dis-cussed, but so far as Bordes is concerned it was onlyin the late 1960s that he turned to flesh out the ‘stand-ard’ cognitive claim on Levallois.‘There can be no doubt’, he asserted in his popu-lar book The Old Stone Age , ‘that the men of this [pre-Neanderthal] time had a perfectly clear mental imageof the object to be made before they set about mak-ing it’. All the more so in the case of Levallois, whichpresupposes ‘not only a conception of the tool’s finalform but also of the various successive stages re-quired and the difficulties to be encountered’ (1968,137). Indeed the preparation of the core is but apreliminary stage, and what is desired is actually theshape of the flake to be detached (1970, 199–200).‘Not only is there a cutting edge in the stone; notonly can one shape at will this cutting edge; but onecan predetermine its shape before striking it out of the flint’ (1971, 3). The shapes thus predeterminedare both standardized and variable, being — accord-ing to the preparation of the core — either flakes inthe strict sense, points, or blades (e.g. 1961; 1980).With or without explicit reference to Bordes, argu-ments whereby Levallois flintknappers had a clearmental image of the product and of the procedurehave been reiterated in numerous Anglo-Saxon pub-lications from the 1970s onwards. Serving to address Understanding Levallois233such issues as conceptual abilities, language and in-telligence, they have become increasingly well-knownand indeed commonsensical. 2 More recently, however, this ‘standard’ con-ception has generated what might be called ‘reac-tionary’ claims, where attempts are made toundermine the idea that the Levallois technique re-quired, and can reveal, particular cognitive abilities.Davidson & Noble, for example, rejected the ideathat Levallois flintknappers took ‘decisions aboutthe finished form of a core envisaged from the first blow on an unstruck nodule’, and argued insteadthat in Levallois ‘the time depth of intentionality isreduced to decisions about the next flake’ (Davidson& Noble 1993, 376). Struck by the fact than manyLevallois flakes were abandoned near their cores,they ventured the startling proposition whereby theseflakes are actually nothing but core rejuvenationflakes, designed to renew the edge angle so thatmore flakes could be removed: in this scenario, asthey put it, ‘the “classic” tortoise core and flakesrepresents failure, not predetermined products’(Davidson & Noble 1993, 376). Dibble proceededwith a somewhat different argument. If the Levalloistechnique did involve predetermination, ‘this mayimply that certain flake shapes were more desirablethan others, and perhaps their production may belinked to linguistic categories’ (Dibble 1989, 424). Hethen goes on to reject this proposition: ‘If by “prede-termination” it is meant that there is an element of standardisation within Levallois flakes, then wewould expect to find less variability in size and/orshape of Levallois flakes than exists for “normal”and biface retouch flakes’ (Dibble 1989, 424). Indeed,the broad-scale statistical study he conducted failedto confirm this expectation: according to averageCoefficients of Variation recorded on Lower and Mid-dle Palaeolithic artefacts sampled in Southern France,the Levallois flakes are often as variable as the ‘nor-mal’ ones, and in any case no less variable than the‘biface retouch’ flakes (Dibble 1989, 424; Dibble 1985).Thus Dibble proposed that Levallois is not a methodfor ‘the production of a single flake with a predeter-mined size and shape’, but rather (with reference toBoëda 1986; 1988a) ‘a specific method or techniquefor core reduction that leads to the production of many flakes from a single core — a reductive strat-egy’ (Dibble 1989, 424). Grounds and implications These summarily presented ‘standard’ and ‘reaction-ary’ claims are admittedly somewhat caricatured, but they serve to raise two issues: the grounds onwhich they are promoted and used, and their impli-cations regarding the cognitive activities of Levalloisflintknappers. On the first point, it is striking to real-ize how much both claims owe to Bordes’ famousformulation. As already noted, Bordes proposed hisLevallois definition on the basis of his extensiveflintknapping experience. At the same time, it is thecase that the near universal appeal of this definition,indeed its ‘classic’ status, was not primarily due toits intrinsic explanatory value but rather to its typo-logical usage. 3 It was in this highly standardized andconstraining context — when products of a tech-nique were identified, counted and considered as if they were a type — that archaeologists were com-pelled to take account of the notion of ‘predetermi-nation’. The use of this mentalistic term was neverreally clarified, but its automatic and exclusive asso-ciation with Levallois was rapidly enshrined. 4 Theauthority of Bordes’ flintknapping experience andthe near-compulsory usage of his definition haveapparently sufficed for many to endorse ‘standard’cognitive claims, and for others to reject them with‘reactionary’ claims.In addition to this linguistic ‘after-effect’, bothclaims appear strongly influenced by the evolution-ary scenarios in which Levallois is implicated. In-deed the Levallois techniques has been associatedfrom early on not only with the Mousterian culture(e.g. de Mortillet & de Mortillet 1881; Reinach 1889), but also with those controversial Neanderthals —considered as worthy precursors by some, and asreplaceable dead ends by others (see recent discus-sions in Hayden 1993; Belfer-Cohen & Goren-Inbar1994). Generally speaking, advocates of ‘standard’claims uphold a gradualist and cumulative scenario.As a consequence, the Levallois technique is almostexclusively considered in view of Acheuleanhandaxes and Upper Palaeolithic blades, and — moreoften than not — found to be better than what pre-ceded and yet bound to be bettered by what super-sedes it. 5 Proponents of ‘reactionary’ claims, for theirpart, favour a rubicon-like scenario where cognitivecapacities of any significance are restricted to ‘mod-ern’ Homo sapiens sapiens . Levallois flakes are pre-sented as failures so as to confirm that only with theUpper Palaeolithic (and the colonization of Australia)do we see a shift from a ‘responsiveness to the un-folding contingency of the here-and-now environ-ment’ towards ‘planning that entails a capacity forconsciousness’ and language (Davidson & Noble1993, 382). Likewise, Dibble’s concluding argumentis that ‘standardization, or a reduction of variability 234Nathan Schlanger[in Levallois] can be explained on the basis of rawmaterial and technology, without having to supposethe presence of linguistic rules, structures, or catego-ries’ (Dibble 1989, 425). This makes it possible torestrict these later capacities to the advent of ‘mod-ern’ humans ( cf. Byrne 1995 for a recent endorsement).The contents and implications of these diverg-ing cognitive claims can be better appreciated now.In the ‘standard’ claims, strong emphasis is put onthe pre-existence of a ‘clear mental image’. Theknapper is said to have possessed a precise abstractrepresentation of the product, which precedes itsrealization and ensures its standardization. Theknapper also had a definite conception of the proce-dure to be followed; a procedure systematicallyplanned in advance and abided by throughout therigorously chained stages of Levallois manufacture(Bordes 1968, 137; Mellars 1991, 71). The flintknappereffectively appears to be a calculating planner who begins with a ‘predetermined’ set and sequence of operations which is then implemented or indeed ex-ecuted. In terms of the preferred evolutionary sce-nario, the resulting image is that of precursor inwhich we can readily identify the rationality, antici-pation, and thinking-things-through we tend to val-orize in ourselves. This a-priorist conception of Levallois flintknapping is rejected by advocates of ‘reactionary’ claims, only to be replaced with an em-piricist one: the flintknapper proceeds on a flake-by-flake basis, responds in function of contingent stimuliof the here and now environment, and achieves somestandardization of the products owing to raw mate-rial and technological constrains (Dibble 1989;Davidson & Noble 1993). 6 Reflection in Levalloismanufacture, if at all present, follows from externalaction and in function of it; the resulting image of the precursor is such that planning, abstraction andconceptualizing capacities can effectively be restrictedto the ‘moderns’ that we are.In sum, the conventional status of cognitiveclaims on Levallois directs attention to their widerimplications (regarding the evolution of language,intelligence, etc.), rather than, indeed at the expenseof, their specific establishment — itself taken forgranted or glossed over. A closer look at both ‘stand-ard’ and ‘reactionary’ claims — with their impliedsplit between doing and thinking, and their respec-tively a-priorist or empiricist emphases — showsmoreover that the grounds on which they are ad-vanced are insufficient, insofar as they follow sucharchaeological considerations as the classic defini-tion of Levallois (with its typological setting), andthe antiquity, taxonomy, and evolutionary status of Levallois flintknappers. What appears to be missing,and what could do much to help us overcome this stateof affairs, is a technological understanding of Levallois. Technological readings The technological understanding pursued here is of relatively recent srcin, and it is worthwhile to brieflyoutline some moments of it development ( cf. alsoSchlanger 1994a,b; in press). The Levallois techniquehas long been replicated by numerous lithic experi-menters (e.g. Reid Moir 1919; Coutier 1929; Leakey1934; Knowles 1953). A number of archaeologicalLevallois artefacts have likewise been comprehen-sively refitted (Commont 1909; Kelley 1954; 1957).These scholars reached important assessments towhich I will refer, but overall their often disparateand inconsistent reports did not promote the sys-tematization and dissemination of the acquiredknowledge ( cf. Johnson 1978; Cziesla 1990). So far asthe French research tradition is concerned, an im-portant if often overlooked role in changing this stateof affairs was actually played by Bordes himself.Before his typological system gained such overwhelm-ing predominance, his experimental work (1947; 1950)seemed to herald a real revolution in the study of techniques. Bordes’ main aim in these publicationswas to characterize cultures through their techniques.Readers such as Leroi-Gourhan, however, gatheredthat it is possible to ‘untangle the series of actionsthat lead to such and such form of tool’, and thenposited that ‘to follow the gestures, flake by flake, isto reconstruct with certainty an important part of the mental structure of the maker’ (Leroi-Gourhan1952, 82). Balout for his part recognized that theinterest in Bordes’ experiments ‘is not in the knappedand retouched object but in the manner by which itis produced’ (Balout 1953, 244). Balout then went onto advocate a ‘dynamic morphology’ which wouldconcentrate on ‘the gestures accomplished by theprehistoric artisan, from the raw material to the fin-ished product’, and seek to identify the ‘ technique ’ —‘the means put into action’ — and the ‘ méthode ’ —‘the reasoned succession of technical gestures’ — bywhich knapping proceeds (Balout 1967, 704ff.).These suggestive indications were considerablyexpanded in the ‘prehistoric technology’ researchgroup led by Jacques Tixier, where experimental re-sults were used to promote the ‘technological reading’and ‘mental reconstruction’ of prehistoric flint-knapping, and more generally to recover the humandimension behind stone artefacts (Tixier 1967; 1980;Tixier et al. 1980; 1992; Pelegrin 1990). In combination Understanding Levallois235with an innovative system of graphic representation(the schéma diacritique of Dauvois 1976; 1981, wherenumbered arrows on the negatives of removals con-vey both the direction and temporal sequence of flake detachments, see Fig. 3 below), and the spreadof the concept of chaîne opératoire from anthropolo-gists of techniques to Palaeolithic archaeologists(Audouze 1985; Geneste 1985; Pelegrin et al. 1988;Karlin et al. 1991), there resulted throughout the 1980sa substantial renewal of interest in prehistoric lithictechnology in general and in the Levallois techniquein particular (Boëda 1986; 1988a; 1994; Geneste 1985). Predetermination and the Levallois ‘concept’ The notion of ‘predetermination’, we recall, has beena key factor in the adoption or propagation of ‘stand-ard’ and ‘reactionary’ claims alike. Most scholarsconcentrated on the classic ‘tortoise’ flakes to theexclusion of other ‘candidate’ exemplars, but theysaid too little about the way — or indeed ways — bywhich the hallmark ‘special preparation’ enabled theknapper to secure the ‘predetermination’ of the prod-uct. It was generally agreed that the flint nodule wassuccessively prepared on each side so as to ‘block’ init a flake struck off ‘in one blow’, but the actualsequence of these stages was generally too brieflyand inconsistently discussed. 7 In these conditions, itis not surprising that discussions on ‘predetermina-tion’ were kept at a superficial level, and that theprocess of ‘special preparation’ by which Levalloisproducts are predetermined was portrayed as a pre-figured and immutable sequence in its own right.From a technological standpoint, however, thisfixation on ‘predetermination’ is both redundant andmisleading. In effect, the physical and mechanicalproperties of flint and other rocks are such that allmaterial flaked from them has characteristics — suchas outline, dimensions, curvature, dorsal pattern —determined by the prior state of the nodule (seeformal discussions in Cotterell & Kamminga 1987;Van Peer 1992; Luedtke 1992; Whittaker 1994). Awell delivered (ballistic) blow with a percussor onan adequate piece of flint results in the initiation of afracture or compression wave which propagates inthe material and leads to the detachment of a flake.The morphological characteristics of this flake de-pend on a number of factors, and notably on theinitiation, unfolding and termination of the fracturewave. In turn, the initiation of the fracture wavedepends on the location, depth and incidence bywhich the blow is delivered, while its unfolding andtermination depend on the mass and disposition of the raw material through which the fracture wavetravels. These particular factors will be detailed pres-ently, but it is already clear now why all debitageproducts (not only Levallois) have some of their char-acteristics determined by the state of the core priorto their detachment. Predetermination alone cannotserve as an exclusive basis for drawing cognitiveconclusions about Levallois.At the same time, this basic ‘knapping theory’also indicates how such predetermination can even-tually be controlled by the knapper, for example toproduce large flakes that are longest along theirknapping axis, gradually thinner towards the edges,regular in their outline and profile, etc.. This controlis achieved by affecting — in specific ways and inrelation to each other — various factors pertaining tothe initiation, unfolding and termination of the frac-ture wave which leads to the detachment of flakes.The concept of Levallois ‘ concept ’ developed byBoëda on archaeological and experimental groundscan help us visualize these possibilities for control.Briefly put, the volume or mass of the nodule is‘initialized’ and prepared so as to have two asym-metrical and non-interchangeable intersecting sur-faces (Fig. 1a). This morphological distinctioncorresponds to and enables a functional differentia-tion — each surface serves to control the removal of future Levallois products in a different way. Onesurface — the surface de débitage or striking surface —provides the products with what becomes, on thedetached flakes, their dorsal side (Fig. 1b). The othersurface — the surface des plans de frappe or strikingplatform — provides the products with a point of im-pact which becomes, on the detached flakes, the butt( talon ) (Fig. 1b) ( cf. Boëda 1986; 1988a; 1994, 12ff.; and below). The striking surface and the striking platformare thus the two areas of the core where the controlledpredetermination of the products can be secured. On striking surfaces, striking platforms and thequestion of faceting For a flake to be successfuly detached from the core,there needs to be some raw material overlying thepath of the fracture wave. Precise control of the shapeof future Levallois flakes can be secured by prepara-tory detachments which remove some of this overly-ing material from the striking surface in specific ways.The negatives of these preparatory removals leaveon the core ‘guiding ridges’ (Bourgon 1957; Boëda 1986;Van Peer 1992), and result in localized ‘convexities’on the surface (Kelley 1954, 151; Boëda 1986). Onesuch convexity — called by Boëda distal convexity —
