Levallois

Recognizing Levallois

According to Gabriel de Mortillet (1883), the name Levallois was coined by M. Reboux, based on very large flakes he found in the gravels of Levallois–Perret near Paris. De Mortillet placed these flakes in an evolutionary position as the successor of the hand-axes characteristic of the Chellean (what subsequently came to be known as Acheulean). In 1910, Victor Commont’s discussion of Levallois flakes indicates a shift toward an understanding of the process of manufacture that made the production of these large flakes possible. In Commont’s discussion, there is direct reference to the actions of the knapper and how they resulted in the successful removal of large sharp-edged flakes. Commont recognized the importance of preparing the platform where the blow for removing the large flake would be delivered. Although he was not explicit about this point, Commont’s discussion does indicate a recognition that Levallois cores are exploited on two surfaces, one of which serves for the removal of the large flakes. Commont described the dominant form of Levallois cores as disks but subsequently the term “tortoise-shaped core” came into widespread usage, as the shape of a tortoise shell evokes the unequal convexity of the two surfaces of the Levallois core. W. J. Sollas integrated Commont’s observations into an evolutionary framework which continued to situate Levallois as the successor to biface (hand-axe) production. For Sollas, Levallois represents an improvement in design over hand-axes, as the product required retouch on only one side, had a reduced weight, and had a sharper edge. For Sollas, Levallois also represented an increased “mastery over technique” (Sollas 1911, 134).

Counting Levallois: The Typological Approach

As archaeologists working in the Middle Paleolithic of Europe, largely following Francois Bordes, adopted quantitative methods that characterize the relative frequency of different tool types within an assemblage the challenge shifted from simply recognizing the presence of Levallois within an assemblage to determining whether a particular flake or core is Levallois. Within the Bordes (1980) typology (list of types), the various Levallois detached pieces (flakes, blades, points, etc., Types 1–5) are something of an anomaly. Whereas most of his typology is based on the shape and location of retouch, the designation Levallois reflects the process of production. The frequency of Levallois within an assemblage was summarized in a Levallois index, based on the percentage of all Levallois products (flakes, blades, and points, both retouched and unretouched) within the flake assemblage. However, even Bordes recognized the distinctiveness of Levallois within his classification, noting at one point that underlying Levallois was “the philosophy of the method” (Bordes 1980, 49, author’s translation). For Bordes, as for many other archaeologists, the “philosophy” of Levallois’ was the predetermination of the shape of the detached flake, point, or blade through the manipulation of the morphology of the core. But how is one to identify whether a particularly detached piece should be categorized as Levallois if Levallois is more of a philosophy than a clear morphology? Not surprisingly, there came to be a wide divergence among archaeologists in what was counted as Levallois and, as a result, the use of a Levallois index came into question. As Lorraine Copeland wrote, in considering the early Middle Paleolithic of the Middle East, “It is becoming increasingly apparent that a wide divergence of views as to what constitutes ‘Levalloisness’ has developed and that the index is becoming virtually useless in assemblage-comparisons” (Copeland 1983, 17). A test in which three archaeologists classified the same assemblage demonstrated a lack of agreement among archeologists in the pieces identified as typologically Levallois (Perpère 1986). Some archaeologists emphasized the presence of a faceted platform as a diagnostic typological criterion for Levallois flakes, but this has not been widely accepted. A faceted platform is the result of the area of the core that will be struck (the platform) for flake removal.

The Technological (Chaîne Opératoire) Perspective on Levallois

The typological approach to Levallois led to an impasse where archaeologists were unable to arrive at a consensus on how to identify objects as Levallois versus non-Levallois, and some were led to doubt whether the designation of Levallois has any scientific validity. The technological approach to stone tool analysis, often referred to as the chaîne opératoire approach, offered a new way of approaching Levallois by shifting away from identification of particular artifacts as Levallois or non-Levallois toward an effort to develop an integration of the complete assemblage within a framework of the skill and knowledge used in tool manufacture within a particular cultural context (Chazan 1997). The technological approach accepts Bordes’ argument that Levallois is a kind of philosophy but extends this to all stone tool production. Following the ideas of authors, including André Leroi Gourhan and Gilbert Simondon, the technological approach situates stone tool production in general and Levallois in particular within the generalized realm of human technical action. The effort to quantify Levallois is rejected in favor of an inquiry into Levallois as a particular instance of human interaction with material.

The technological approach builds on three key concepts. The first is the distinction between skill and knowledge. Skill (savoir faire) is the embodied and non-verbal ability to effectively carry out a technical act. In a general way, one can think of skill as a component of the techniques du corpes discussed by Marcel Mauss, which includes all aspects of the way the enculturated body moves. Knowledge (connaissance) refers roughly to what one would consider in the context of sports as the rules of the game. These are aspects of technical action that can potentially be verbalized and that can be detached from the body in motion. The second distinction is between method and technique. Method is the conceptual underpinning of a technical action. Technique refers to the mode of transmission of energy in a technical process. For the technological approach, Levallois is a method.

The technique used in the Levallois method is hard hammer percussion, striking a blow with a rock hammer rather than a hammer made of antler or wood (soft hammer). The technological definition of Levallois method focuses on five criteria (Boëda 1995):

The Levallois method is a subset of prepared core technologies, which includes the discoid method and the trifacial method along with a range of other more particular variants. Prepared core technology is defined as a reduction sequence in which knapping is organized to allow for the continuous production of flakes with a morphology that is controlled by the morphology of the core—which is imposed by the knapper (the person engaged in making stone tools). Each prepared core method is distinguished by a distinctive conceptual underpinning relating to the organization of knapping.

The discoidal method is widespread and can easily be confounded with the Levallois method. In the discoidal method, the two surfaces are not hierarchically related and the fracture plan for the removal of predetermined flakes is at an acute angle (usually approaching 45 degrees) to the plane of intersection. Because flake removals are made at an acute angle to the plain of intersection, these removals rarely go beyond the midpoint of the core. It is important to emphasize that the discoidal method is a highly productive system that results in flakes with a controlled morphology. A critical distinction between the Levallois method and the discoid method is that the Levallois method is more suited for producing elongated flakes with a continuous cutting edge whereas the discoidal method produces shorter flakes that often (although not always) come off the edge of the core, resulting in an oblique edge. The trifacial method is in many ways similar to the discoidal method in terms of the products but differs in that the core is organized with three surfaces, with a triangular section (one surface shorter than the other two) (Chazan 2007; Boëda 1991).

A critical point is that all of the prepared core methods can produce flakes (detached pieces) that are typologically Levallois, and typologically Levallois flakes can also be produced by some blade production methods. The technological approach rejects the identification of the Levallois method based on a quantitative criterion regarding the morphology of individual artifacts. From this perspective, it is only through the analysis of the complete assemblage that one can reach an identification of the underlying method.

It is also important to note that the emphasis in the technological definition of Levallois on “predetermination” and the distinction between preparation and exploitation do not imply that only the predetermined flakes were used or that the preparation flakes were discarded without use (Turq et al. 2013). Rather, these terms refer to the logic of the system of production in which certain flake removals serve to prepare convexities and others are removals that are made possible by the process of preparation—and whose morphology is controlled by the convexities put in place by the knapper.

Although there are a series of criteria that define the Levallois, there is also a great deal of flexibility within this method in the manner in which the production face of the core is exploited. The knapper is able to control the morphology of the detached pieces by manipulating the convexities of the production face, but any shift in exploitation also requires adaptations to the platform face. The first major distinction within the Levallois method is between preferential exploitation, where a single large flake is removed from the production face before the repreparation of the convexities of the production face, and recurrent methods where a series of flakes can be removed from a production face before repreparation. The use of a preferential method does not imply that only a single flake is removed from a core, as there can be multiple phases of preparation and exploitation.

Within recurrent methods there are two key variants based on the pattern of exploitation. Unidirectional exploitation involves the removal of predetermined flakes from a single platform. Bidirectional exploitation refers to systems where predetermined flakes are struck from opposing platforms. There are also two systems, unidirectional convergent and centripetal, that blur the lines between preparation and exploitation, since in some cases, their exploitation flakes serve to reprepare convexities. In the centripetal method, predetermined flake removals are struck from around the periphery of the core. In the unidirectional convergent method, predetermined flakes are removed from one platform but the direction of flake removals converges through the midpoint, resulting in a configuration that allows for the removal of a central flake that will be triangular. Levallois methods can also be characterized by the dominant form of the exploitation removals, which can be flakes, blades (flakes that are twice as long as they are wide), or points.

Brantingham and Kuhn (2001) have successfully applied mathematical modeling to understand the efficiency and flexibility of the Levallois method. In this model, the Levallois method is simplified to refer to bifacial exploitation of a cobble with a steep angle (c. 80 degrees) between the plain of intersection between the surfaces and the striking platform surface. Based on mathematical modeling, the Levallois method is shown to be “capable of simultaneously minimizing raw material waste and maximizing blank production and the generation of usable cutting edge,” while simultaneously allowing for flexibility (Brantingham and Kuhn 2001, 758). It is important to emphasize that Brantingham and Kuhn do not provide a quantitative method for distinguishing Levallois from non-Levallois prepared core technology. Rather, it points to potential evolutionary and adaptive explanations for the widespread adoption of this method.

Refitting and the Understanding of Levallois

Refitting of stone tool assemblages allows archaeologists direct access to the process of knapping, and perhaps it is not surprising that the Levallois method, which is defined in terms of process rather than the morphology of finished artifacts, has been a particular focus of refitting studies. Already in 1909, Commont based his discussion of Levallois in part on refitting flakes onto the cores from which they had been removed. A detailed study by Schlanger (1996) of a single refit core from the Dutch early Middle Paleolithic (OIS 7) site of Maastricht Belvedere provided insight into how knappers adapted the Levallois method to the properties of a particular block of stone. Schlanger found six phases of exploitation and repreparation on this core. In this case, the Levallois method set “the knapper an objective to reach . . . [and] also acts as a multidimensional and multimodal resource that incorporates visual and tactile cues for proceeding, an image that makes it possible to appreciate and eventually integrate transformations as they occur” (Schlanger 1996, 248). Philip Van Peer (1992) carried out extensive core refitting on Middle Paleolithic sites from the Middle Nile. Van Peer’s results confirm both the essential validity of Levallois as a method as well as the variability in the types of exploitation within the Levallois method. Van Peer’s analysis also points to the difficulty posed by systems of production that appear to extend beyond the strict technological criteria that define the Levallois method. In an analysis of two cores from the site of Taramsa, Van Peer found that the exploitation wraps around the lateral edges of the core and thus the cores are not conceived of as bifacial with two surfaces divided by a plane of intersection. Yet, the products of knapping are very similar to reduction following the Levallois method, although the removals are unusually elongated.

Critiquing the Technological (Chaîne Opératoire) Approach

The initial identification of Levallois was based on an intuitive recognition of a distinctive component of the archaeological record that fit well with ideas about progress through time in prehistory. However, when archaeologists shifted away from intuitive methods toward quantification, it proved to be difficult to develop a typological definition of Levallois that would be consistently applied by different prehistorians. The technological approach comes to Levallois from a very different perspective, which stresses the shared knowledge and skill that characterizes the technological behavior of human societies. From this perspective, the distinctiveness of Levallois assemblages suggested an underlying method. The technical criteria that makes Levallois distinctive within the larger domain of prepared core technology emerged from a combination of the analysis of archaeological assemblages and experimental replication. The intuition of the analyst came into play in grasping at the method that underlay the totality of the assemblage.

A potential critique is that the chaîne opératoire is epistemologically inappropriate for the study of prehistory as it is essentially an emic approach that cannot be undertaken in the absence of a living informant. Thus, Tostevin argues that: “In contexts without extensive refitting sequences, however, the emic focus has led to the construction of abstract cognitive or volumetric rules, such as Boëda’s . . . criteria for the production of Levallois and other flaking technologies, but without the evidential support normally associated with other aspects of low-level and middle-level theory in archaeology” (2011, 356). This critique flows in part from an over-literal reading of some of the terminology used in the French literature. For example, the use of the term “desired end product” is not meant to imply anything about the state of mind of the knapper (or about which flake removals will be used) but rather reflects the logic underlying the production sequence, which in the Levallois method hinges on the morphology of particular removals. This reflects the intention of the knapper but only in the very limited sense of how the knapper constructed their project of working a core. The chaîne opératoire is only emic to the extent that it attempts to understand stone tools from the perspective of a skilled person who made and used these tools. This is a far cry from the emic approaches characteristic of ethnography, which embrace linguistics and belief systems. In fact, one can say that the technological definition of Levallois is explicitly etic, but it is built on the recognition of the skill and knowledge involved in flint knapping.

Although the technological approach has been extremely effective, there is also a degree of discomfort among archaeologists who question the epistemological validity of a methodology that is not based on quantifiable, testable criteria. A key critique is that the technological approach has ossified into a new typology that is excessively rigid. As stated by Bar Yosef and Van Peer, “our formal reduction strategies and methods are from a behavioral point of view, most likely only situational grades in a general technological system, perhaps forged and maintained through daily communication” (2009, 117). In other terms, one can think of Levallois as a preference in the behavior of the members of a particular group rather than a rigid set of rules that dictated actions (Chazan 2018).

A basic limitation of the technological definition of Levallois is that it is essentially interpretive, and analysts can legitimately disagree on the interpretation of the same assemblage. Any effort to overcome this limitation risks adding to the rigidity that Bar Yosef and Van Peer critique. The identification of the production method, including Levallois, is ultimately qualitative, and while not strictly speaking emic, these identifications are observer dependent. It is important to emphasize that the identification of the Levallois method is only a means to an end, a step in an analysis that provides a deeper understanding of the archaeological record. The classificatory obsession identified by Bar Yosef and Van Peer is ultimately a misapplication of a technological perspective. Levallois is from a technological perspective not a “thing” but rather a behavioral preference. Bordes was prescient when he described Levallois as a kind of “philosophy.” As demonstrated by Geneste’s (1985) studies of raw material procurement in the Middle Paleolithic of the Perigord, the identification of a production method provides an essential orientation toward a broader technoeconomic study of prehistoric societies.

The Functioning of Levallois Products

A distinguishing characteristic of the Levallois method is the production of flakes with a continuous sharp edge and a controlled morphology. Despite the advantage of maximizing the cutting edge, a limitation of Levallois flakes is that many do not offer a blunt edge that is easily gripped in the hand. This is not true of all products of Levallois knapping, as flakes that come off the edge of the core (éclats debordant) do have a blunt edge, but these types of flakes occur in low frequencies within Levallois assemblages when compared to the products of other prepared core methods, such as the discoid method. It therefore seems likely that the development of the Levallois method took place within the broader technological context of the development of hafting. Levallois flakes recovered from the site of Umm el Tlel (Syria) show traces of bitumen residues from hafts, and a use wear analysis of Levallois flakes from the site of Biache-Saint-Vaast (France) found evidence for hafting in the distribution of edge damage (Boëda et al. 2008; Rots 2013). There has been significant debate of the proposal that hafted Levallois points were used as spear tips (Shea 1988). The discovery at Umm el Tlel of the midsection of a Levallois point embedded in the vertebra of a wild ass (Equus africanus) seems to provide conclusive evidence that at least some Levallois points were in fact used as spear tips. However, use wear studies have also demonstrated a wide range of other uses for Levallois flakes, including scrapping hides, butchery, and working wood (Beyries 1988; Rots 2013). Use wear studies have not found evidence that particular types of flakes from a Levallois reduction sequence were used exclusively for one type of activity (Beyries 1988; Beyries and Boeda 1983).

What Levallois Tells Us About Human Evolution

Aspects of prepared core technology can be traced back as far as 2.5 million years. Refitting of cores from the site of Lokalalei, Kenya demonstrates long sequences of flake production and the sequential exploitation of surfaces from multiple striking platforms. However, at Loakalalei, there is no evidence that the convexities of the exploited surface were shaped to control the morphology of flake removals. The development of bifacial technology in the Acheulean involves directly shaping a tool rather than creating a core that produces flakes with a constrained morphology. However, it is important to note that Acheulean flake production does include prepared core technology.

Generally, it appears that prepared cores appear through much of the Acheulean, becoming more prevalent during the period between 500,000 and 200,000 years ago. At Garba XIII (Melka Kunture, Upper Awash, Ethiopia), dated between 1 million and 800,000 years ago, discoid flake production has been identified alongside hand-axes and cleavers (Gallotti et al. 2014). The most impressive case of prepared core technology in the Acheulean is the Victoria West Technique found on sites in South Africa. In the Victoria West Technique, large pieces of rock are shaped to create a convex surface which is exploited for the removal of a single large side-struck flake which is then minimally retouched to create the form of a hand-axe or cleaver. The age of the Victoria West Technique has not been determined but multiple lines of evidence point to an age earlier than 500,000 years ago (Sharon and Beaumont 2006). A similar large flake production method has been described from the later Acheulean of East Africa (Tryon, McBrearty, and Texier 2005). Thus, there is strong reason to think that prepared core technology was within the cognitive capacity of Homo erectus and that the explanation for the shift toward the dominance of prepared core technology within which the Levallois method emerges is likely broadly technoeconomic rather than a direct expression of cognitive evolution in the hominin lineage.

By 400,000 years ago, prepared core technologies for the production of flakes and blades are apparent in the Middle East and Africa. The site of Kathu Pan 1 (South Africa), dated to c. 500,000 years ago, has produced evidence of a prepared core technology for flake and blade that bears close similarity to the Levallois method (Wilkins and Chazan 2012). As in the Levallois method, the cores from Kathu Pan 1 are conceived as two surfaces that are hierarchically related, and flake removals from the exploitation face are subparallel to the plane of intersection between the two surfaces. However, the Kathu Pan 1 cores depart from the technological definition of Levallois as the exploitation wraps around the edges of the core such that the plane of intersection between the two faces is not maintained. The exploitation surface is worked using a recurrent bidirectional system. Production includes both flakes and blades, many of which are points. Multiple lines of evidence suggest that the points functioned as spear tips (Wilkins and Chazan 2012). Simpler methods of blade production that do not share characteristics with the Levallois method are documented from a similar age in the Kapthuran Formation, Kenya at c. 500,000 years ago and Qesem Cave, Israel beginning around 400,000 years ago (Tryon et al. 2005; Shimelmitz, Barkai, and Gopher 2011).

Typological studies often show a gradual increase in the frequency of typologically Levallois flakes toward the end of the Lower Paleolithic, which can be interpreted as evidence for the gradual adoption of the Levallois method. Thus, for example at the Lower Paleolithic site of Berekhet Ram on the Golan Heights, with an upper age limit of 233,000 years, sixteen out of seventy-one cores are typologically Levallois and 31 out of 752 flakes are Levallois (Goren-Inbar 1985). However, it is not clear what this typological observation means in terms of the development of Levallois as a method of production. From a technological perspective, it is plausible that cases where there is a low frequency of typologically Levallois flakes are the product of prepared core methods other than Levallois. Thus, for example, at the Lower Paleolithic site of Holon, Israel, dated to c. 200,000 years ago, the low frequency of Levallois flakes (50 out of 1,236) is associated with the dominant use of the trifacial method.

There are a number of cases where a low frequency of typologically Levallois flakes and cores is the result of the use of a rudimentary form of the Levallois method. In an analysis of the Purfleet site in England, the use of such a rudimentary Levallois method was identified in a context dated to 325,000 years ago (White and Ashton 2003). The cores are described as “technologically under developed and procedurally truncated, especially when compared with classic Levallois cores,” specifically “the flaking surface does not show the maintenance of distal and lateral convexities (criterion 3)” (White and Ashton 2003, 602). Another early site that may have a rudimentary form of the Levallois method is Nor Geghi 1, Armenia, dated to roughly 330,000 years ago (Adler et al. 2014). Typologically, Levallois cores and flakes account for about one-quarter of the overall assemblage form Nor Geghi 1, although overall flake production seems relatively informal. The Nor Geghi I cores are all on flakes and it is not clear whether lateral and distal convexities are maintained.

Based on the strict technological definition, it appears that the Levallois method emerges as one facies of a shift away from a focus on methods of bifacial tool shaping toward the use of prepared core methods to produce flakes with a controlled morphology. Complex prepared core technologies are widespread by 500,000 years ago. The adoption of the Levallois method as a dominant, although not exclusive, approach to tool manufacture appears to have taken place in parallel in both the Neanderthal and modern human lineages. In the context of European Neanderthals, the earliest sites with a clearly Levallois system of production date to Oxygen Isotope Stage 7, approximately 230,000–200,000 years ago. Notably, at Biache St. Vaast in northern France, an early Levallois lithic industry is associated with a Neanderthal fossil (Boëda et al. 1990; Guipert et al. 2011). In the Levant, the first evidence of the Levallois method as a major component of the lithic industry is in the Tabun D variant of the Mousterian from Hayonim Lower Unit E, dated to approximately 180,000–170,000 years ago, where Levallois flakes and cores co-occur with a non-Levallois blade production system (Meignen 1998; Mercier et al. 2007). It is not clear whether these contexts are associated with modern humans or Neanderthals. In subsequent phases of the Levantine Middle Paleolithic, the Levallois method becomes the dominant lithic technology in contexts associated with both Neanderthals and modern humans. In both East Africa and southern Africa, early modern human Middle Stone Age sites provide evidence of the use of the Levallois method (Tryon et al. 2005; Wurz 2013).

The replacement of Neanderthals by modern humans in Europe corresponds to a technological shift from the Levallois method to volumetric methods of core exploitation oriented toward blade production. The African record is far less clear, but in general terms, the Levallois method is associated with the Middle Stone Age and is not found in the Later Stone Age. Even during the Middle Stone Age, the Levallois method is only one component of a diversity of methods of stone tool technology that includes blade and bladelet production methods. Various claims have been made for the presence of the Levallois method in Holocene contexts, particularly in discussions of the Leilira points found in Australia (Dortch and Bordes 1977; Newman and Moore 2013). However, these cases appear to be based on the identification of Levallois flakes based on morphological traits (a typological approach) as opposed to a technological identification of the Levallois method.

Explaining the Levallois Method in the Context of Human Evolution

Multiple studies have demonstrated the cognitive complexity of the Levallois method and the need for the knapper to have the ability to adapt the knapping strategy to respond to the characteristics of a particular block of raw material (Wynn and Coolidge 2004; Schlanger 1996). Moreover, the regularity of the methods employed within a particular assemblage suggest that the ability to transmit the knowledge and skills involved between generations (see discussion of methods of transmission in Lycett, von Cramon-Taubadel, and Eren 2016 and Wilkins 2018). However, although a certain level of cognitive functioning was clearly a precondition for the development and adoption of the Levallois method, it is no longer plausible to argue that these developments are linked to novel cognitive capacities in either the modern human or Neanderthal lineage. There is no compelling evidence that producing flakes with the Levallois method is in any sense more cognitively demanding than producing a carefully thinned symmetrical hand-axe (a technology characteristic of the Late Acheulean). Both refined hand-axe production and Levallois flake production require advance planning, effective reaction to irregularities in raw material, and a lengthy sequence of technically linked gestures. Further weakening the linkage between the evolution of novel cognitive capacities and the Levallois method is the substantial evidence for complex prepared core methods that long predate the widespread adoption of the Levallois method. The widespread use of the Levallois method is a component of the move away from the shaping of a large bifacial tool as a dominant approach to lithic tool production. The Levallois method emerges out of a growing emphasis on prepared core methods and is embedded within this large trend. Both the widespread adoption and subsequent abandonment of the Levallois method is best understood from a technoeconomic perspective. Linked aspects of adaptation, including mobility (effecting both raw material procurement and transport) and tool functioning (a focus on skin processing and stone-tipped spears, both of which are linked to methods of hafting), were likely critical in favoring the adoption of the Levallois method. The development of methods of hunting with light projectiles and an extensive bone tool industry were similarly likely critical in the abandonment of the Levallois method in favor of other methods of stone tool manufacture.