The functional hypothesis is based on ethnographic or actualist comparativism, according to the postulate that the universal problems encountered by human populations are equivalent, regardless of the period, and that a similar spectrum of efficient solutions is available (Hayden, 2015). Functional constants and operational rationales are identified based on data from work on the mechanics of cutting movements (Atkins, 2009, Key, 2016, McGorry et al., 2003, McGorry et al., 2005 and Singh et al., 2016), specific work on the section morphology of (sub-)contemporaneous tools (Boucard, 2005 and Cabanié, 1932), experiments on the efficiency of working edges (Key and Lycett, 2015 and Key and Lycett, 2017a) and finally, ethnographic observations generally conducted on Aboriginal hunter-gatherers in Australia (Gould et al., 1971, Hayden, 1977 and Schick and Toth, 1993). Two elements are used to propose a functional hypothesis. The first is the type of opposition between the cutting edge and the prehensile edge. The second is the edge morphology in section, referred to in French as the “émouture”, or in English as the bevel (Bertrand and Bertrand, 1997 and Klaric et al., 2015).

A tool as a system necessarily has an active edge (cutting edge, hereafter CE) and a prehensile zone (prehensile edge, hereafter PE) (Boëda, 2013). These two poles of the tool are opposite each other on the blank, along an axis perpendicular to tool movement (Boucard, 2005 and Leroi-Gourhan, 1943). This is illustrated here using the example of Aboriginal bifaces, with known functions (Vayson de Pradennes, 1937) (Fig. 2C). Thus, for longitudinal cutting, the CE/PE pair lies opposite each other along an oblique axis. This is particularly apparent when the cutting edge is associated with a point. In the case of launched percussion and transversal cutting, this opposition is perpendicular (transversally or perpendicularly). However, although this opposition may be transversely or longitudinally perpendicular for transverse cutting, in the case of launched percussion the opposition is usually longitudinal (for elongated and massive tools). Nevertheless, exceptions to those operating rationales do exist (e.g., “couteau à pied”, “Ulu”). In addition, several motions can be associated with a single tool (Leroi-Gourhan, 1964 and Mansur-Franchomme, 1986). However, the aim here is to highlight recurrent operating rationales, which can then be used to infer hypotheses about the functional structuring of Lower Paleolithic artefacts.

In addition to percussive chopping actions, there are two main types of cutting movements: transverse cutting — to scrape, plane — and longitudinal cutting — to cut, saw or incise. Actualistic sources for transverse cutting show that the angle of the cutting edge can measure up to 60° and its section morphology is asymmetrical (chisel-shaped bevel). Also, the harder the worked material, the higher the angle. In contrast, the cutting edge for longitudinal cutting is less than 60° and is most effective at 30° and the morphology of the section is symmetrical, flat or concave. This latter morphology is the least common, in spite of its efficiency, due to the fragility of the cutting edge. For launched percussion actions, the angle is up to 45° (robustness), and the bevel can be symmetrical (biconvex) or asymmetrical, depending on the associated gesture (Bertrand and Bertrand, 1997, Cabanié, 1932, Geneste and Plisson, 1996, Gould et al., 1971, Siegel, 1985 and Wilmsen, 1968) (Fig. 2A and B). However, these findings are mainly based on metal tools and the angular data of lithic tools are likely to be different. Also, these functional propositions are mere hypotheses. They can only be validated on larger sets of tools.

A three-dimensional image acquisition program was carried out for the bifaces in order to optimize their conservation (Abel et al., 2011). A portable surface scanner with a maximal precision of 0.1 mm (Créaform Viewscan 3D) was used for digitization. When the 3D model includes small void zones that do not alter our perception of the edges, they are improved with the Géomagic Studio software.

The three-dimensional models are then processed with the Avizo fire 7.1 software. The 3D model is correctly oriented in an orthonormal coordinate system (x, y, and z), so that a virtual section can then be taken along the required axis and direction (Fig. 3). Here, the section is strictly perpendicular to the longitudinal axis in order to obtain a transverse section. This section can then be moved, without varying the longitudinal axis, over the whole length of the biface, in order to observe the evolution of the cutting edges from the base to the point. In the analysis carried out here, transverse cuts were made every centimetre, producing nine cross-sections per biface (because the length of the analysed bifaces was relatively similar — about 10 cm). Three longitudinal cuts were also made to characterize the angles of the proximal and distal parts of the biface.

Edges are described from a qualitative perspective: bevel morphology (simple/double bevel, plane, convex or concave) and cutting edge morphology in profile (straight, scalloped); and quantitative perspective: edge angle and depth of cut. The edge angle (Tringham et al., 1974) is measured between the cutting edge and the first morphological change in the section (which corresponds to the start of the bevel edge). This length corresponds to what we propose to call the depth of cut. This edge angle is associated with the penetration potential, which can be more or less limited, depending on the material. This length is measured in the following way: a straight line is traced from the ruptures in the angles on both surfaces of the edge. The depth of cut corresponds to the distance between the middle of this straight line and the extremity of the edge (Fig. 4).

The depth of cut and the edge angle are measured with Avizo software based on the three-dimensional models. These measurements are then projected onto a bivariate diagram linking the edge angle (on the vertical axis) to the depth of the section (on the horizontal axis). The overall volumetry of the biface is represented by a colorimetric chart, where each colour corresponds to a thickness.

The angle data, the depth of cut and the plan and profile of edge morphology are used to define the edge portions for each biface, which potentially correspond to the CE or PE. These observations then enable us to infer the functional capacities of the edges of the biface.

F13. 1409 (Fig. 5 and Table 1). On the basis of the measured edge angle and depth of cut, the distal part (between sections A–C) and the left edge (between sections 1–3) can be considered CEs. They correspond to two CEs. CE1 has a scalloped morphology in profile and a biconcave morphology in section, whereas CE2 has a chisel-shaped section and a convex profile. High angles, associated with reduced depth of cut, define two PEs. The first is located between sections 4–9 on the left edge (PE1) and the second between sections 3–6 on the right edge (PE2). Finally, an edge with bivalent functional potentialities (C/PE) is located on the proximal part of the biface (sections A–C and 7–9 of the right edge).

E13. 2546 (Fig. 6 and Table 1). This biface presents 3 CEs and 2 PEs. CE1 is located on the distal part, comprising sections 1 of the left edge and sections A and B. CE2 is on the left edge between sections 7 and 8. It is separated from CE3 on account of the high angle of section 9 on the left edge (and section A on the proximal part). CE3 is on the right edge and the proximal part of the biface, comprising sections B–C and 7–9. We identified two areas with very open angles and practically no depth of cut (inexistent or limited), corresponding to PE. PE1 is on the right edge between sections 1–6 and PE2 is on the left edge, between sections 2–6.

E14. 2807 (Fig. 7 and Table 1). A CE and a PE are identified on this biface. The CE corresponds to the right edge, comprising the distal extremity and a small part of the proximal zone (sections 1–9 of the right edge and B–C of the distal part and left edge). The present state of this CE is not homogeneous. Indeed, the distal extremity presents a burin-type fracture with no counter-bulb, indicating that it is not due to a technological action (not intentional). Consequently, this portion of the CE between sections 3–4 on the right edge presents a higher angle than the other parts of the edge, due to a resharpening retouch sequence after the fracture. On the opposite edge and the proximal part, a cortical zone (sections 8–9 of the left edge and A–B of the proximal part) and an abrupt zone constitute a PE.

E17. 4127 (Fig. 8 and Table 1). This biface has an extended CE (118 mm) on the left edge (sections 1-8), the distal extremity (sections A–C) and a small part of the right edge (sections 1–2). This CE is opposed to 2 PE. PE1 corresponds to the right edge, between sections 4–6 and PE2 corresponds to the cortical base, between sections B–C and 9 on the right edge. On the right edge, sections 7 and 8 can be considered as a PE. However, the hypothesis of a CE cannot be totally excluded even if there is no associated prehensile area. Thus, this part of the edge is named PE+.

F14. 3242 (Fig. 9 and Table 1). The distal part of this biface may be a CE. It is technologically shaped by two notches, delimiting a small point. Opposite the point, sections 7–9 of the left edge, 8–9 of the right edge and A–C of the base may correspond to a PE. For the lateral edges, it is not possible to make the distinction between CE and PE on the basis of the measurement of the edge angle and depth of cut. If the right edge is a CE, the left edge could be a PE and vice versa. They are thus bivalent, and are named C/PE1 or 2.

G15. 8084 (Fig. 10 and Table 1). This biface presents a CE on the right edge between sections 1–8, which extends to the distal extremity (sections B and C). On the left distal extremity (sections 1 and A), a small area may be a PE. The principal PE is located on the proximal part, between section 9 of the right edge and section 8 of the left edge. Lastly, the characteristics (angle and depth of cut) of the mesial part of the left edge are bivalent, but it is not associated with a PE, and is thus named PE+.

Linking the edge angles of the CE and the depth of the section on the bivariate diagram highlights the differences in the functional potential of each biface (Fig. 11). A line is plotted at 60° to make a first distinction between longitudinal and transverse sections. We chose to place the limit for the depth of section at 4 mm. For this analysis, this datum appears to distinguish between a shallow (< 4 mm) or deep (> 4 mm) cutting potential. However, it must be applied to a larger quantity of artefacts to grasp the pertinence of this measurement for other analyses.

The five “edge-point”-type CE are divided into three functional groups.

In one case — G15. 8084 — the CE mostly presents angles of less than 60° with a depth of cut of more than 4 mm and a maximum depth of 13.5 mm. The rectilinear morphology of the CE in profile, its symmetrical section, its long length and its oblique association with the PE, characterize a tool with a potential for deep longitudinal cutting.

In two cases — E17. 4127 and E14. 2807 — the CE mostly presents angles of less than 60° and a depth of cut between 1.3 and 6.6 mm. The CEs are regular, rectilinear or slightly convex in profile, with a long section and oblique opposition with the PE. Therefore, these tools were designed for longitudinal cutting. The fact that the depth of cut is often less than 4 mm and the CE section is asymmetric suggest that the edge was intended for a tangential longitudinal cutting motion.

In two cases — E13. 2546/CE3 and F13. 1409/CE1 — the CE present angles of more than 60° and a depth of cut of more than 4 mm. The two CEs present a sharpening retouch phase, creating a scalloped edge morphology in profile and a biconcave morphology in cross-section. The extension of the cutting edge is rather limited. The axis formed with the PE is oblique. Although the angle is more than 60°, the scalloped edge (in profile view), the cross-sectional symmetry and the oblique CE/PE axis are linked to longitudinal cutting motions. The cutting angle of more than 60° of these tools can be explained by the fact that they seem to result from considerable resharpening.

For three bifaces, the angles of certain CEs are more than 60° and the depth of the section ranges between 2 and 4 mm.

In the case of F13. 1409/CE2, the CE is rectilinear in top view and presents an asymmetrical chisel-shaped section. The CE/PE axis is transverse, which corresponds to the configuration of tools used for percussion. However, for this biface, on account of the small dimensions (63 × 46 × 17 mm) and light weight (63 g), we consider that it may have been used for transversal cutting. This is also in agreement with the morphological characteristics of the CE.

The CE1 and CE2 of E13. 2546 are linear or pointed in top view and present a rectilinear profile and a symmetric biplane section, on a relatively short segment of the cutting edge (respectively 34 and 27 mm). The axis with the PE is oblique in both cases. On account of the symmetry and the straightness of the cutting edge and the oblique opposite position of the CE/PE pair, we propose a longitudinal cutting potential, in spite of the high angle (however, the latter is no higher than 70°).

The CE of biface F14. 3242 mostly presents angles of less than 60° and a depth of cut of less than 2 mm. It shows a slightly pointed, short bill-hook morphology, defined by two inverse notches. The axis formed with the PE is transverse. In the same way as for the CE2 of biface F13. 1409, the small dimensions (74 × 60 × 24 mm) and weight of the artefact (112 g) do not suggest use for percussion. This CE/PE pair thus seems to have been used for transversal cutting. This biface also has two edges with bivalent status. These two edges are complementary: when one is a CE, the other is a PE and vice versa; when one is plano-convex in profile, the other is convex-plane.