Flake variation in relation to the application of force

Journal of Archaeological Science 46 (2014) 37e49

Contents lists available at ScienceDirect

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

Flake variation in relation to the application of force Matthew Magnani a, Zeljko Rezek a, b, Sam C. Lin a, Annie Chan c, Harold L. Dibble a, d, e, * a

Department of Anthropology, University of Pennsylvania, 3260 South Street, Philadelphia, PA 19104-6398, United States Department of Anthropology, Rutgers University, 131 George Street, New Brunswick, NJ 08901-1414, United States c Department of East Asian Languages and Civilizations, University of Pennsylvania, 255 S. 36th Street, Philadelphia, PA 19104-6305, United States d Institute for Human Origins, School of Human Evolution and Social Change, Arizona State University, Box 874101, Tempe, AZ 85282-4101, United States e Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, D-04103 Leipzig, Germany b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 22 February 2014 Accepted 24 February 2014 Available online 12 March 2014

The appearance of new force application techniques in the production of stone artifacts over the course of human evolution has been associated with the increasing technological capacity of hominin groups. Yet, the causal relationship between the knapping practice and the flake characteristics upon which these behavioral inferences rest remains largely untested under controlled settings. Here we present a recent controlled experiment examining the effect of various force application variables (hammer shape; location of force application; angle of blow; hammer displacement speed) on flake morphology. Results indicate that the independent variables interact with flake attributes in a complex way that makes simple analogies between particular attributes and specific force application techniques extremely difficult. However, trade-offs among the variables cast new light on the possible mechanisms underlying variation in force application techniques used in flintknapping. Ó 2014 Elsevier Ltd. All rights reserved.

Keywords: Lithic technology Flintknapping Force application Controlled experiment Experimental archaeology

1. Introduction An important step toward the explanation of lithic assemblage variability is to understand the flaking techniques used by prehistoric knappers. One of the major concerns is to distinguish the various kinds of force or load that were applied for flake production. Since the initial appearance of stone artifact production over two million years ago, the adoption of innovative knapping force applications (hammer type and percussion technique) have been associated with the increased cognitive complexity and technological capacity of past hominins as well as evidence for cumulative cultural evolution (Ambrose, 2001; Mellars, 2006; Stout, 2011; Schick and Toth, 1993; Weaver, 2005). Novel percussion techniques also relate to innovations in subsistence, technological, and economic practices which likely incurred both new costs and benefits that were evolutionary significant to hominin evolution (e.g., Brown et al., 2009; Mourre et al., 2010; Hayden, 1987). For instance, the use of soft hammer percussion technology and the ability to exert fine control over artifact morphology and reduction trajectory has implications for artifact

* Corresponding author. Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, D-04103 Leipzig, Germany. Tel.: þ49 215 898 7073. E-mail address: [email protected] (H.L. Dibble). http://dx.doi.org/10.1016/j.jas.2014.02.029 0305-4403/Ó 2014 Elsevier Ltd. All rights reserved.

design and use-life, and by extension was likely inter-related with technological innovations as well as mobility and subsistence strategies (Bleed, 1986; Kelly, 1988; Kelly and Todd, 1988; Nelson, 1991). Indeed, such arguments have been made concerning North American biface technology (Morrow, 1995), blades and microblades of the Upper Paleolithic and later periods (Newcomer, 1975; Desrosiers, 2012), the Mousterian of Acheulian tradition in Middle Paleolithic Europe (Newcomer, 1971), the Acheulian industry of the Lower Paleolithic (Hayden, 1987; Bergman and Roberts, 1988), and, more recently, the Still Bay and Howiesons Poort industries of Southern African Middle Stone Age (Mourre et al., 2010; Villa et al., 2010; Soriano et al., 2007). In this context, it is essential to develop a better understanding of exactly how different variables related to the application of force affect particular aspects of flake morphology. This paper presents the results of a highly controlled experiment designed to address these issues. Previous research based on the current experimental design has focused on a number of independent variables that are under the control of a knapper, including the role of platform depth, exterior platform angle, angle of blow, hammer velocity, and exterior core morphology (Dibble and Rezek, 2009; Rezek et al., 2011; Lin et al., 2013). The set of experiments described in this paper focus on the application of force, including the material and shape of the hammer and the location of the strike. However, because some flake attributes are affected by the interaction of

38

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

two or more variables, and because some of the earlier relationships were not tested using conditions as defined here, variables such as angle of blow and hammer speed are again included in these analyses. There is little doubt that direct percussion lithic technology is a complex process given the number of independent variables involved as well as the number of flake attributes that can be affected by them. This requires that a series of experiments be performed, and each of these experiments requires a different set of parameters. Additionally, since our goal here is to test under highly controlled conditions relationships that have, for the most part, already been proposed on the basis of replicative experiments, each experiment will be presented here in the context of those discussions. For this reason, the format and organization of this paper has been slightly modified both to simplify and clarify the presentation of results: while some basic parameters are common to all of the experiments, others are specific to each experiment, and each experiment will be presented in terms of its background in the archaeological literature. 2. Basic materials and methods The experimental design employed here builds on the one recently used to investigate the role of several other variables related to flake production, including exterior platform angle, platform depth, angle of blow, and exterior core morphology (Dibble and Rezek, 2009; Rezek et al., 2011; Lin et al., 2013). The flaking apparatus, as described in Dibble and Rezek (2009), consists of a pneumatic cylinder to which hammers made of different materials can be attached. A core mount is situated directly below the hammer, and cores are securely clamped in it on the sides and the back. When the cylinder extends the hammer makes contact with the cores and a flake is detached. In most of the experiments presented here, the strikes are dynamic, though one experiment (see below) includes static loads as well. The shape of the end of the hammer varies, however, given the goals of each experiment. The mount is adjustable for changes in the angle of the core platform surface relative to the angle at which the hammer strikes the platform (angle of blow) and also the distance from the point of percussion to the platform edge (platform depth, as defined by Dibble and Whittaker, 1981; Dibble and Pelcin, 1995). The exterior surface of the core is exposed to prevent potential interference from the mount during flake formation. The cores used in this study are manufactured from molded soda/lime glass with a semispherical surface (Fig. 1) (see Dibble and Rezek, 2009), and are consistent in size and shape, although core weights varied between 472 g and 695 g, depending on how much material was removed to produce a desired exterior platform angle. Major independent variables involved in this study that are relevant to most of the relationships explored here (see also Dibble and Rezek, 2009; Rezek et al., 2011; Lin et al., 2013) include:  Exterior platform angle (EPA): measured where the platform and exterior core surface intersect. Each core was cut transversely at the platform end to the appropriate EPA using a wet diamond saw and measured by a goniometer. Cores were designed in a way that there was no longitudinal curve along the exterior surface immediately behind the platform, thus allowing unambiguous measurements. EPAs of 55, 65 and 75 were used.  Platform depth (PD): measured from the point of percussion to the exterior edge of the platform. For each flake, platform depth was measured independently by more than one observer and the average of their recorded values was used for analysis to minimize inter-observer error. Platform depth varies continuously.

Fig. 1. Side and end views of the glass cores used in the experiment. Different exterior platform angles are produced by changing the angle at which the proximal end of the core is cut.

 Angle of blow (AOB): as in previous experiments (Dibble and Rezek, 2009), an AOB of zero represents the hammer aligning perpendicularly to the core platform (Fig. 2B). A positive AOB would indicate oblique blows striking “outward” to the core (Fig. 2A). Likewise, negative values indicate strikes hitting “inward” to the core (Fig. 2C).  Location of force application: Two locations of hammer percussion were employed in this study: either the hammer strikes directly on the platform surface some distance away from the core exterior (Fig. 2A and B), which are here called platform surface strikes, or the hammer strikes the exterior platform edge (on-edge strikes; Fig. 2C), similar to techniques associated with biface thinning. In our experiment, on-edge strikes on untreated edges led to many crushed edges and failed flake production. To reduce this problem (although it could not be completely eliminated, as discussed below), the exterior platform edges were lightly abraded with a diamond sander to produce a slight rounding. This is in keeping with the common flintknapping practice to make the edge “.well prepared to receive a percussion blow without shattering, and transmit the force as a fracture resulting in the removal of the intended flake” (Sheets, 1973). Other independent variables will be discussed below in the context of each specific study. For every flake produced, a series of dependent variables were recorded. Flake weight was measured with an electronic scale to

Fig. 2. Illustration of different AOBs and location of force.

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

the nearest 0.1 g. Flake dimensions consist of flake length (measured from the point of percussion to the most distal point on the flake), flake width (measured perpendicularly to flake length at the midpoint of the length axis), and flake thickness (measured at the point of intersection of length and width) (see Debénath and Dibble, 1994). Platform width was measured from one lateral platform edge to the other. Three other dependent variables were measured and calculated using a Microscribe G2X digitizer and Rhinoceros software. These are flake edge length, or the perimeter of the flake not including the platform (measured to the nearest 0.01 mm), flake surface area (measured to the nearest 0.01 mm2) as a projected two-dimensional measure of the outlined flake surface, platform area, measured in the same way as surface area, and bulb volume (measured to the nearest 0.01 mm3). A more detailed description of how this last measurement was taken can be found in Lin et al. (2013). 3. Background, description, and results of the individual experiments relating to application of force 3.1. The effects of different hammer materials, platform lipping, and location of force application It is difficult to isolate three of our independent variables since there is a certain degree of interaction among them. For example, while the different hammer materials appear to have a number of significant effects on flake attributes, these effects are different when flakes are struck on the platform surface rather than on the exterior edge of the platform. These interactions thus greatly complicate the discussion that follows. 3.1.1. Platform lipping as dependent on different hammer materials We will begin by investigating the production of lipping, which is a sharp extension of the platform along its interior edge where platform intersects with the ventral side of a flake. As discussed by various authors, the presence or absence of a lip may relate to two distinct kinds of fracture. One is a conchoidal, or Hertzian, fracture (Kerkhof and Müller-Beck, 1969; Crabtree, 1972; Witthoft, 1974; Cotterell and Kamminga, 1987), where the force propagates directly from the point of hammer impact to termination. This results in a clear point of percussion, a distinctive and well-formed bulb of percussion, and no lip. This kind of fracture is often considered to be the result of percussion with a “hard hammer” such as quartz, quartzite, flint, etc. The other type of fracture, often called a bending fracture, is normally considered to be the result of using a soft hammer, in which fracture initiation occurs some distance from the point where the hammer strikes (Cotterell and Kamminga, 1987, 1990). Supposedly because the propagated force does not initiate at the point of percussion, bending flakes possess neither pronounced bulbs nor clear points of percussion, and they are marked with smaller platforms and pronounced lipping on the interior platform edge (Cotterell and Kamminga, 1987). The relationship between platform lipping and the hardness of the hammer is based on several experimental studies. For example, Hayden and Hutchings (1989) distinguished the concepts of hammer ‘hardness’ (i.e., hard vs. soft) from ‘material’ (i.e., stone vs. bone vs. wood). The possible use of soft stone hammers, such as limestone and sandstone, for detaching flakes that carry typical soft hammer flake features has been proposed and demonstrated by flintknappers (Crabtree, 1967; Crabtree and Swanson, 1968; Bordes as cited in Hayden and Hutchings, 1989). Likewise, Patterson (1982) concluded with a small sample that lipping occurred more frequently with both soft hammerstones as well as antler, in comparison to quartzite hammerstones (see also Ohnuma and Bergman, 1982; Schindler and Koch, 2012). However, according to

39

Cotterell and Kamminga (1987), it is also possible to create bending flakes with hard hammers when the contact stresses are minimal. In particular, low exterior platform angle would favor the production of bending flakes, independent of hammer material (Cotterell and Kamminga, 1987). Similarly, previous controlled experiments by Bonnichsen (1977) and Pelcin (1997a) indicated no difference in the percentages of lipping between hard and soft hammer types under similar experimental conditions. In this study, we used hammers made of three materials; steel, copper, and synthetic bone, which were milled to identical shapes (spheres) and sizes (9.51 mm diameter) and firmly mounted to the pneumatic cylinder. The reason for not using materials that were used as hammers in archaeological contexts (e.g. various stones, antler, bone, wood) is because of the need to control for inherent variation in the hammer material as well as to allow for more precise control over hammer hardness (see Hayden and Hutchings, 1989). The synthetic bone (http://www.sawbones.com/products/ bio/testblocks/solidfoam.aspx) is a solid rigid polyurethane foam designed for testing orthopedic devices, and was chosen for these experiments due to its consistency, which would have been more difficult to obtain using actual bone or antler. These three materials differ in terms of hardness, although it is difficult to put all three on the same hardness scale. Thus, the hardness values for steel and copper, in the Rockwell hardness scale ‘B’, are 90.5 and 26.5, respectively, which means that the steel hammers are much harder than the copper ones. Unfortunately, because the synthetic bone is much softer than copper, it cannot be measured on the same scale and instead has a hardness measure of 73.5 on the Rockwell hardness scale ‘R’. In terms of tensile strength, the steel hammers have a value of approximately 91,000 psi, while the synthetic bone has a value of 27,000 psi; the copper tensile strength is somewhat less than 49,000 psi, though more accurate figures are not available. At any rate, it is clear that the synthetic bone is the softest of our materials, the copper somewhat intermediate, and steel is the hardest. The hammer tips for all of these materials are spherical with a diameter of 9.51 mm. Although various types of lipping have been suggested (e.g. Schindler and Koch, 2012), here it was recorded simply as present or absent. The main criterion used to identify the presence of lipping was that the lip extended for the entire length of the interior edge of the platform, including at the point of fracture initiation. Nonetheless, given that there is a certain amount of subjectivity in assessing the presence of this attribute, lipping was independently evaluated by each of the authors; if there was significant disagreement, it was coded as missing data. To a large extent, the results of this study support the notion that hammer material is directly tied to the presence or absence of platform lipping (see Table 1): overall, the bone hammers almost always produced flakes with noticeable lipping. As a percentage, copper hammers produced lipped flakes at a lower rate, though higher than occurred with steel hammers, a result that mirrors the findings of earlier experiments cited above. The fact that the percentage of lipped flakes decreases with the hardness of the hammer material (i.e., harder materials produce fewer lipped flakes) suggests that there is some relationship between hammer type and

Table 1 Presence or absence of lipping by hammer material. (Chi-square ¼ 51.36, df ¼ 2, p < 0.001). Includes EPA ¼ 55, 65, and 75, all values of AOB, and both locations of force (strikes both on the platform surface and the exterior edge of the platform). Hammer material

Lipped

Not lipped

% Lipped

Bone Copper Steel

53 39 25

1 28 46

98.1% 58.2% 35.2%

40

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

whether or not lipping occurs. However, it is also clear that the presence or absence of lipping in general has little or no predictive value regarding the kind of hammer that was used. In addition to the hardness of the hammer material, there are other independent variables to investigate to see if they also play a role in the production of lipped flakes. For the following analyses, only the flakes produced from the spherical copper or steel hammer tips will be used, since those produced with bone were almost always lipped. Two variables that do not appear to have any significant effect in producing lipped flakes are EPA and the location where force was applied (see Table 2 and Table 3). However, there is an association with AOB (see Table 4) in that AOB less than or equal to 0 produces more lipped flakes than does AOB greater than 0. To summarize, while the bone hammer almost always produced flakes with lipping, the harder hammers also produced them, though in increasingly lower percentages in direct relation to hardness. This is consistent with the conclusion reached in replicative experiments that lipping occurs most often with softer hammers (see for example, Henry et al., 1976; Schindler and Koch, 2012), but the fact that it also occurs more than a third of the time with the use of harder hammers makes it impossible to use lipping alone as a proxy for hammer hardness. In addition to this, and in the case of flake production with harder hammers, the use of negative AOB will increase the rate of flake lipping. Therefore, the introduction of negative AOB into an already weak relationship between flake lipping and increased hardness of the hammer introduces a confounded variable that makes it even more difficult to use the presence of a lip as a reliable indicator of a soft hammer flake production. Cotterell and Kamminga (1987) stated that conchoidal initiation requires high pressure over the contact area between the hammer and the raw material, and therefore necessitates the use of harder hammer materials. Also as proposed by them, if sufficiently high pressure is not achieved, the percussion will result in bending initiation. Based on this model, the differential occurrence of conchoidal and bending fracture relates to whether the hammer property can transfer the applied striking force to appropriate hammer pressure at the point of contact with the core e i.e., did all incoming energy get transferred to the core, or was some energy lost due to the deformation of the hammer? While this model agrees with the soft hammer outcome where almost all platforms exhibit lipping, it remains difficult to explain why harder hammer materials, under a controlled setting with uniform striking force, also produced bending fracture. This complex pattern may in fact relate to finer physical properties such as compressive and tensile strength of different hammer materials, as the ability for a given hammer to maintain its physical structure under applied force relates to the brittleness, elasticity, and other properties of the material (Braun et al., 2009; Noll, 2000). Further controlled testing is required to determine the relative effects of these mechanical properties. 3.1.2. Flake shape in relation to different hammer materials and location of force application The relationships between different hammer materials and flake dimensions are less clear. In previous replicative experiments, Table 2 Presence or absence of lipping by exterior platform angle. (Chi-square ¼ 3.9, df ¼ 2, p ¼ 0.14), including flakes struck with both copper and steel hammers, both force locations, and a full range of AOB. The lack of association with EPA holds even if AOB or location of force is controlled. EPA

Lipped

Not lipped

% Lipped

55 65 75

7 39 18

2 51 21

77.8% 43.3% 46.2%

Table 3 Presence or absence of lipping by location of force. (Chi-square ¼ 3.63, df ¼ 1, p ¼ 0.057). All flakes produced with copper hammer and EPA ¼ 65, and includes a full range of AOB. Location of force

Lipped

Not lipped

% Lipped

Platform surface Exterior platform edge

11 22

7 13

57.9% 62.9%

Patterson (1982) concluded that different hammer types do not affect flake length, width, or thickness, while Hayden and Hutchings (1989) reported that soft hammer flakes are longer, wider and thinner. Likewise, while some have suggested that the bulbs on flakes produced by antler are more diffuse than those detached by hard hammers (Crabtree, 1972; Ohnuma and Bergman, 1982), Patterson (1982) found no difference between soft and hard hammer flaking in the bulbs. Smaller overall platform size has also periodically been attributed to the use of soft hammer types, and in some replicative flintknapping studies, soft hammers have been associated with expanding flakes (Hayden and Hutchings, 1989). Hayden and Hutchings (1989) also identified two types of soft hammer struck flakes based on platform size to flake weight ratio e those that possess uniformly small platforms and relatively large flake surface area, resembling biface thinning flakes; and those that have larger and more varied platform sizes. The controlled experiment of Pelcin (1997a) has corroborated some conclusions of replicative knappers and contested others. His experiment showed that flakes produced with an antler hammer were longer and thinner on average than those created by harder hammers, though overall mass was found to be similar for both soft and hard hammer types. He postulated in some cases, instead of being related to hammer type, differences in flake attributes could be related to other variables such as angle of blow and platform preparation, as controlled unconsciously by flintknappers. Similarly, Hayden and Hutchings (1989) demonstrated that for certain flintknappers the products of soft and hard hammers could be distinguished, while for others they could not. To them this indicated the possibility that some other knapping techniques e and not hammer type e were responsible for differences in flake morphology. Table 5 presents data on several ratios relating to flake dimensions (the use of ratios helps to control for the effect of overall size, which, as shown in previous experiments, is overwhelmingly a result of EPA and PD). In this table, results are broken down by location of force, so that the effects of various hammer materials are first shown for flakes struck on their platform surfaces, then those that were struck on the exterior edge of the platform. In both cases the dimensional ratios, when significant according to hammer type, tend to show gradients from softer to harder hammers. However, not all of the dimension ratios show a significant relationship with hammer type in both the platform-struck flakes (the left-most column) and those struck on edge (the center column). In fact, the only two dependent variables that express significant differences in using different hammers at both locations of force are the ratios of length to width and bulb volume to flake weight.

Table 4 Presence or absence of lipping by angle of blow (Chi-square ¼ 14.33, df ¼ 1, p < 0.001). All flakes produced with copper hammer, EPA ¼ 65, and includes flakes produced by either location of force. Angle of blow

Lipped

Not lipped

% Lipped

0 >0

21 5

7 18

72.4% 21.7%

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

41

Table 5 Various dimension ratios for flakes produced by different hammers, broken down by location of force (all produced with EPA ¼ 65, AOB ¼ 5, and includes both lipped and unlipped flakes). Statistically significant differences are marked in red.

As also seen in Table 5, the right-most columns present tests of significance between location of force broken down by each of the hammer types, showing both inter-hammer (varying hammer type but with the same location of force) and intra-hammer (using the same hammer but varying location of force) effects on flake variation. These results show that some variables are affected by either hammer type or location of force, while some are affected simultaneously by both hammer type and location of force. For example, as seen in the topmost row, flakes produced with soft hammers are longer relative to their width than flakes produced with harder hammers, and this is true in both cases of striking the core on the platform surface and on the edge. However, it is only with hard hammers (steel) that the location of force will significantly impact the flake length/width ratio. In the case of hard hammers, striking the core on edge will produce longer flakes relative to their width than striking the core platform surface. On the other hand, some dimension ratios are affected either by hammer type or location of force only. The ratio between platform area and flake surface area, for example, seems to be independent from hammer type, but is affected by the location of force (see further below). Bulb volume (in relation to flake weight) is not affected by the location of force application, but it is by hammer type, such that harder hammers will result in more pronounced bulbs.

3.1.3. Flake size and platform area Previous controlled experiments (Speth, 1981; Dibble and Whittaker, 1981; Dibble and Pelcin, 1995; Dibble and Rezek, 2009; Pelcin, 1997b) have shown a strong relationship between platform depth and flake weight for any given EPA, and Fig. 3 shows that this relationship holds true for each of the hammer materials. There is, however, some confounding due to lipping, which as seen in the upper graphs of Fig. 4, results in a lower correlation. This is most likely due to the fact that in bending flakes the fractures begin some distance from the actual point of percussion, introducing a certain amount of noise, or error, in the relationship. Because the lower graphs of Fig. 4 also include lipped flakes, their correlation coefficients (r) are lower as well. If, for example, only unlipped platform-struck flakes are analyzed for the lower left-hand graph, the r2 value increases to 0.953. This observation suggests that, under bending fracture initiation, platform depth likely acts as a dependent variable rather than an independent variable in the process of flake formation. Another effect noted for flakes that were struck from the exterior edge of the platform is that they tended to produce smaller platform areas in relation to overall flake surface areas (see Table 6), whether or not the flakes exhibited lipping. In other words, for the same platform area, flakes struck from the exterior edge of the

42

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

Fig. 3. Scatter diagrams and correlations between flake weight and the cube of platform depth by hammer material (EPA ¼ 65, AOB ¼ 5, and including both platform surface-struck and edge-struck flakes).

Fig. 4. Correlations between flake weight and the cube of platform depth for lipped and unlipped flakes, and platform-struck and edge-struck flakes. AOB ¼ 5, EPA ¼ 65. The lower r2 values for three of the graphs reflect the fact that lipped flakes are included.

platform will have larger flake surface areas. This relationship was not apparent among the different hammer types and it was also not present in analyzing lipped vs. unlipped flakes. However, the ratio of platform area to flake area is also a function of EPA, since EPA affects overall flake area independently of platform area (Dibble and Rezek, 2009; Rezek et al., 2011; Lin et al., 2013). 3.1.4. Platform crushing Platform crushing has been demonstrated as more common with hard hammers in some cases (Hayden and Hutchings, 1989; Patterson, 1982) and not in others (Henry et al., 1976). In our data, it was observed that flakes struck on the core edge have a Table 6 Relationship between location of force and the ratio of platform area to flake area, broken down by the presence or absence of lipping (EPA ¼ 65, AOB ¼ 5, including all hammer types). Location of strike

Lipped flakes Unlipped flakes

Platform surface Edge strike Platform surface Edge strike

higher tendency to crush the platform, and that the hardest hammer has the highest percentage of occurrence (Table 7). As noted earlier, crushing can be reduced somewhat for edge-struck flakes if the exterior platform edge was abraded, which is a common technique in bifacial reduction (Crabtree and Swanson, 1968; Sheets, 1973). To summarize the results of this section, which examined the effects of different hammer materials (related to varying degrees of hardness) and the location of force (whether on the platform

Table 7 Platform crushing by hammer material and location of force. Hammer material

Location of force

Bone

Platform surface Exterior platform edge Platform surface Exterior platform edge Platform surface Exterior platform edge

Platform area/flake area Mean

N

Std

t

p

0.029 0.013 0.035 0.010

6 4 4 5

0.006 0.004 0.009 0.006

5.397

0.001

4.756

0.002

Copper

Steel

Intact platform

Crushed platform

N

%

N

%

12 37

92.3% 74.0%

1 13

7.7% 26.0%

32 49

97.0% 87.5%

1 7

3.0% 12.5%

129 25

92.1% 64.1%

11 14

7.9% 35.9%

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

43

Fig. 5. Flake weight relative to the cube of platform depth by angle of blow (EPA ¼ 65, copper hammer). Note that the means of weight/platform depth3 between edge-struck and platform-struck are not significantly different, except in the case of AOB ¼ 5 (t ¼ 2.55, df ¼ 13, p ¼ 0.024). All flakes produced with edge-shaped steel hammer tips (see Fig. 6 on the right).

surface or on the exterior edge of the platform), it is clear that there are complex interactions among these two independent variables. The different hammer types affect many aspects of flake morphology, as does the location of force. During actual flintknapping, since all of these variables are varied at the same time, it becomes extremely difficult to discern the independent effects that we see here. 3.2. Angle of blow Similar to the location of force application, angle of blow is an attribute that has been commonly overlooked in previous

experiments on force application. In some cases, the same variable is termed ‘angle of percussion’ or ‘angle of impact’ (Speth, 1972, 1981). It is also important to note that there are many instances where this specific term is used to denote EPA instead (e.g., Schick and Toth, 1993; Fuglestvedt, 2007). In normal flintknapping, the angle of blow tends to vary according to the location of force (see Fig. 2). When the core is struck on the platform surface, the angle of blow tends to be directed outward toward the exterior face of the core (what is here termed a positive AOB). For on-edge percussion, which is common in bifacial reduction, for example, it is commonly held that the angle is directed more to the interior of the core (Whittaker, 1994).

44

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

According to the model proposed by Cotterell and Kamminga (1987), fracture propagation caused by non-wedging percussion (i.e., not bipolar) is dependent on the intrinsic stiffness of the material and therefore less sensitive to the angle of blow (Bril et al., 2011). However, Speth (1972) argued that the angle of blow impacts the angle of the stress wave (‘angle of incidence’) that dictates how fracture travels through the material from the point of percussion. This directly affects the thickness, and hence, the size of the resulting flake. Under controlled settings, Dibble and Rezek (2009) suggested that the angle of blow does act as an independent variable in affecting certain flake characteristics. Yet, it is unclear how it relates to other independent variables and, more importantly, how it translates to artifact attributes that are visible archaeologically. Here we demonstrate that several variables are significantly affected by AOB, even when controlling for location of force, including flake weight relative to the cube of platform depth, and length, width and thickness relative to platform depth (Fig. 5, though the latter is not shown). As reported earlier (Dibble and Rezek, 2009), this is a similar relationship to that found with flakes produced by striking on the platform surface only, though here we include negative AOBs for both platform-struck flakes and those struck on-edge. While this finding illustrates the clear influence of AOB on flake attributes, how this empirical observation relates to theories of fracture mechanics as proposed by Speth (1972) and Cotterell and Kamminga (1987) remains less clear.

3.3. Hammer size and shape Another aspect of hammer variability that has not received much attention include shape and morphology. While some studies tend to make rough separations between large and small hammers, and several experiments have remarked on the possible effects hammer size has on flake dimension and other attributes (e.g., Amick et al., 1988), in most of those experiments hammers of different materials and sizes were utilized to perform different knapping tasks to fit the changing demands along the reduction sequence. Specifically, small hard hammers tend to be used for platform preparation and trimming, large hard hammers primarily for flake production, and soft hammers usually for artifact shaping and retouching (e.g. Amick et al., 1988; Newcomer, 1971). Therefore, how hammer shape affects flake formation remains a relatively unexplored area of study. For this experiment, five differently shaped hammers were used: three with round ends, but of different diameters, one with a flat surface, and one that is beveled so that only the edge of the hammer strikes the core (see Fig. 6). All of the hammers were made of steel, and the strikes were made with AOB ¼ 0, and EPA ¼ 65. As shown in Table 8, the differences in the shape or size of the hammer

Fig. 6. Profile views of steel tips of varying sizes and shapes used in current experiments.

did not produce significant differences in any of the variables examined. Based on these results, neither size nor shape of the hammer has an effect on flake morphology. 3.4. Displacement (hammer) speed A common distinction in hammer displacement speed is between a static load (where pressure increases slowly) and a dynamic load (i.e., a quick blow to the platform). This distinction roughly corresponds to the general separation in flintknapping between pressure flaking and direct percussion, respectively. It is said that in pressure flaking, the applied hammer force maintains a static equilibrium with the internal stress of the material until the force exceeds a critical threshold at which point the fracture initiates (Speth, 1972). In contrast, direct percussion consists of a sudden applied force that creates stress waves which travels through the material. Through high speed photography, Cotterell and Kamminga (1987) demonstrated that the velocity of fracture propagation is higher in direct percussion than pressure flaking. While force velocity does not seem to affect flake form in any significant way (Dibble and Pelcin, 1995), it is commonly noted that there are shape differences between flakes produced through direct percussion and pressure flaking, namely pressure flaked flakes are thinner and more evenly shaped. However, similar to the confounded nature of soft hammer experimental observations, such observation may relate more to hammer hardness, the location of the strike, or the exterior core morphology (Hayden and Hutchings, 1989; Bonnichsen, 1977; Pelcin, 1997a). It was previously reported in an earlier controlled experiment (Dibble and Rezek, 2009) that displacement speed did not affect any measures of flake morphology. In that experiment, speed was increased by intervals of one order of magnitude between 0.01 inches per minute (ipm)(0.00042 cm/s) to 10000 ipm (423.33 cm/ s), with a steel hammer positioned to strike the platform surface. Since the current experiment is concerned with different hammer materials, and it also introduces a difference in the location of force, a further test between static (using a displacement speed of 0.05 ipm [0.002 cm/s]) and dynamic (650 ipm [27.516 cm/s]) flaking was performed. First, in terms of differences in flake morphology associated with the different hammer materials and static pressure, the results (Table 9) mirror those presented earlier in Table 5. Using cores of EPA ¼ 65, AOB ¼ 5, spherical hammer tips of 9.51 diameter, and locating the force on the exterior platform edge, no significant differences in dimensions or dimension ratios were found between static and dynamic loads with the bone hammer, and only one (the ratio of length to platform depth) was found with the copper hammer e in this one case, flakes produced with static pressure are shorter relative to their platform depth. On the other hand, flakes produced with a steel hammer show several significant differences related to displacement speed. However, the differences found with the steel hammer probably relate to the fact that every flake produced with a steel hammer at the static loading is quite small and hinged, and our data show that hinging affects many of the dimensional measures used here (normally such flakes are excluded from analysis). Why this hinging occurred with flakes produced under static pressure with the steel hammer is not understood at this time. The results do not show any consistent patterning between flakes produced with static versus dynamic force. Again, except for the edge-struck flakes produced with steel hammer, the results conform to Dibble and Rezek’s findings (2009; see also Dibble and Pelcin, 1995) that hammer speed does not affect flake dimensions or dimension ratios to any significant degree, whether the location

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

45

Table 8 Mean values and (N) of various dimension ratios by shape and size of the hammer (EPA ¼ 65, AOB ¼ 0, steel hammers). Rounded tips

Length/width Length/thickness Width/thickness Length/platform depth Width/platform depth Thickness/platform Flake weight/platform

Flat

3.2 mm

6.97 mm

9.51 mm

1.59 16.5 9.52 10.51 6.63 0.74 0.08

1.32 11.16 8.05 9.37 7.02 0.92 0.08

1.54 15.18 8.56 8.68 6.29 0.84 0.06

(3) (3) (3) (3) (3) (3) (3)

(3) (3) (3) (3) (3) (3) (3)

of force is on the platform surface or on the platform edge. It should be noted that the fracture velocity can potentially be estimated using microscopic features of Wallner lines and fracture wings (Cotterell and Kamminga, 1979: 108; Kerkhof and Müller-Beck, 1969: 447; Tomenchuk, 1985; Hutchings, 1997), and perhaps through analysis of these features it would be possible to infer relative hammer velocity and identify percussion techniques involved in the manufacture of ancient lithic artifacts (e.g., Hutchings, 1999). Recent studies have employed these features to examine impact fractures on artifacts to infer their potential use as projectile armatures (Hutchings, 1997, 2011; Sahle et al., 2013). 4. Discussion Based on the results described earlier, Fig. 7 presents a summary of the average effects of the three major force application variables examined here (AOB, location of force application, and hammer hardness) on various flake attributes. While our experimental design allows us to break down the complex relationship by examining the effect of each independent variable in turn, it is more difficult to combine these observations into general models of independent causal relationships. This is because when multiple variables show the same effects, it is difficult to infer the relative role of each independent variable based on flake morphology alone. For example, platform lipping is associated with softer hammers, though it can also be produced with the hardest hammer. It might be tempting to conclude, then, that a lipped flake has a higher probability of being made with a soft hammer. However, lipping is also a function of AOB. Given this fact, it is impossible to determine that the presence of lipping is due solely to the use of a particular kind of hammer, by striking the core with a negative AOB, or both.

(3) (3) (3) (3) (3) (3) (3)

1.48 18.07 11.4 14.57 10.36 0.98 0.18

Edge

(3) (3) (3) (2) (2) (2) (2)

1.96 14.88 7.58 12.81 6.36 0.84 0.1

(2) (2) (2) (2) (2) (2) (3)

F

p

0.294 0.172 0.795 0.449 2.678 0.236 1.257

0.875 0.947 0.557 0.771 0.110 0.910 0.355

Likewise, the use of a negative AOB will produce flakes with higher ratios of length, width, thickness, and weight to platform depth, and smaller bulbs relative to weight. The use of softer hammers, or striking the core on the exterior edge of the platform, will affect some of these ratios in the same direction, and it will also increase the ratios of length to width, length to thickness, and width to thickness. On the other hand, hammer size and shape do not affect any of these variables in any consistent way, and while hammer speed shows some significant relationships to these variables, these are apparent only with the steel hammer. However, because almost all of the flakes produced by the steel hammer in this one experiment are hinged, the results should be viewed with caution. There are, however, trade-off relationships among these variables that have implications for knapping strategies. Specifically, manipulation of force application variables alters flake morphology in terms of the relative length of flake perimeter/cutting edge to flake mass e a measure reflecting the potential utility of a given flake (sensu Shott, 1996; also see Lin et al., 2013). Considering only flakes produced from platform edge strikes, those made by the bone hammer appear to be more elongated but steel hammer flakes are, in fact, as elongated in their overall morphology (t ¼ 1.0648, df ¼ 13, p ¼ 0.306) and are also considerably thinner relative to flake length (t ¼ 2.3991, df ¼ 13, p ¼ 0.0321) and width (t ¼ 4.6531, df ¼ 13, p ¼ 0.0005). What this suggests is that flakes possessing the greatest amount of cutting edge relative to flake mass can be detached by striking platform edge with hard hammer. However, as shown earlier, the use of hard hammer on platform edge also leads to a higher rate of platform crushing. Therefore, it appears that using soft hammer on core edge is a more secure technique for producing elongated flakes without crushing the platform.

Table 9 Tests of significance between flakes produced with dynamic vs. static loads, broken down by hammer material. Significant differences highlighted in red (EPA ¼ 65, AOB ¼ 5, location of force ¼ exterior platform edge).

46

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

Fig. 7. Summary of the effects of AOB, location of force application, and hammer material on various flake dimension ratios.

Likewise, for platform surface strikes, the use of a soft hammer will produce flakes that are more elongated, thinner, and have smaller bulb volumes, than those produced with harder hammers. As demonstrated in previous controlled experiments, similar effects of increasing relative flake length and thinness can also be achieved by a higher EPA (Dibble and Rezek, 2009; Lin et al., 2013). Therefore, in theory, the use of soft hammer on platform surfaces that have high EPAs should result in flakes that contain the highest cutting edge to mass ratio. However, the experiment here shows that higher EPA also increases the probability for unsuccessful flake detachment due to hammer crushing with soft hammer use (Table 10). This suggests that the physical properties of the hammer dictate how it is used to maximize the chances of successfully removing a product. These trade-offs also potentially explain the common co-occurrence among certain variables in replicative flintknapping experiments. It is often thought that soft hammers are more effective for on-edge strikes while the use of hard hammer requires greater platform surface contact. Instead, this may simply reflect the greater occurrence of platform crushing when striking platform edge with hard hammers (Table 6). This factor alone likely played an important role in the shaping and maintenance of bifacial edge. The goal of this experiment is to isolate the individual effects of several independent variables that are under the direct control of a knapper, although, and unfortunately, none of the independent variables investigated here are visible on flakes (except perhaps displacement speed through microscopic markers, which was not tested in this set of experiments). This has several implications. First, it is impossible to test the results from this controlled experiment with reference to samples of artifacts derived from archaeological sites since we cannot know how those flakes were struck and with what kind of hammer. Second, while the effects of location of force application and hammer type could be tested through replicative experiments, angle of blow would need to be controlled as well, which is difficult in actual knapping situations. Finally, given the fact that several “invisible” variables are acting simultaneously on various flake attributes, it would appear to be Table 10 Hammer crushing by EPA in bone hammer strikes on platform surface. EPA

Successful flake detachment

Hammer crushing

% Hammer crushing

65 75

9 4

1 3

10.0% 42.9%

difficult, if not impossible, to separate their effects through examination of archaeologically-derived flakes. The last point is further exacerbated by other confounding variables, some of which were not investigated here, including platform morphology (Clarkson and Hiscock, 2011; Van Peer et al., 2010: 45e46) and core morphology, potentially more in terms of surface longitudinal and lateral convexities but less so in regards to surface ridge configurations (Rezek et al., 2011). Again, because the effects of these variables are all expressed in the limited number of dimensional attributes, it becomes extremely difficult to discern exactly which independent variable(s) were responsible for the resulting flake morphology. A good example to illustrate this issue is pressure flaking, which has often been associated with the production of elongated and thinner flakes in a standardized fashion (Mourre et al., 2010). As demonstrated here and in previous experiments, hammer velocity does not appear to significantly affect flake morphology. On the other hand, longer and thinner flakes could be produced through a platform configuration involving platform-edge strikes, negative AOBs, and/or the use of soft hammers. It is therefore likely that pressure flaking is a technique for increasing the precision of flaking by slowly pressing the tip of the hammer on the desired platform location. This would allow the various force applications and platform variables to be carefully controlled for each flake removal e hence the resulting flakes are more standardized in morphology. It is not, however, displacement speed that is directly influencing the outcome. Furthermore, pressure flaking has long been associated with the ability to exert fine control in the knapping and shaping of stone artifacts. The use of these techniques has been further linked to greater complexity of hominin cognition as well as technological advancement in human evolution (e.g., Stout, 2011; Mourre et al., 2010). While it is tempting to draw large scale comparisons in the continuity and change of force application techniques along the trajectory of hominin evolution (e.g. Soriano et al., 2007), what this study has shown is that the type of fracture and flake morphological characteristics associated with finer controlled flaking can in fact also be achieved through other means without the involvement of these innovative force application techniques. In other words, while static loads may have been involved in producing Still Bay points in the South African Middle Stone Age (e.g., Mourre et al., 2010), the same flake attributes used to infer pressure flaking can be explained equally well by the knapper’s control of platform attributes and other independent variables through more accurate direct percussion. In this case, therefore, there is no need to invoke new knapping practices. What this example illustrates is the differences that underlie controlled experiments, such as those presented here, and replicative experiments. There is no question that the latter provide invaluable insights into possible ways of producing artifact forms resembling those found archaeologically. In this sense, they can arguably possess high ecological validity e i.e., the resemblance to empirical reality (Gibson, 1979), due to their experimental setup imitating ‘real-world’ settings in flake production (e.g., by using materials that might have been used in the past). Such experiments, however, inherently lack the ability to isolate and control several independent variables, which makes it more difficult to determine precisely and quantify the causal relationships among the variables being monitored and the measured outcomes. This lowers the internal validity of replicative experiments (Campbell and Stanley, 1963; Kirk, 2009, 2012), that is, the validity of accurately concluding that an independent variable is responsible for variation in the dependent variable of question. There is yet another concept of validity that is most often evaluated in experimental designs, namely external validity, which relates to the degree to

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

which an experimental finding can be generalized and applied across a wide array of techniques in production of stone artifacts. Such validity decreases if the experiments focus more narrowly on particularities of past lithic technology, regardless of whether the experiments are performed in a controlled fashion (e.g. production of Mesoamerican prismatic blades by Faulkner, 1972) or by flintknapping (e.g. Levallois as the object of study in Eren and Lycett, 2012). It should be noted that the definition of these different kinds of experimental validity contrasts with how these terms are used by Lycett and Eren (2013; see also Mesoudi, 2011). We consider the controlled experiments reported here to be of both high internal and external validity. This is because they allow precise detection of causal relationships between independent and dependent variables, and, due to the broad applicability of the experimental results, are relevant to the study of the production of stone artifacts across wide geographical, temporal and technological contexts. The representativeness of an experimental study is to be considered principally on the functional phenomenon or process that is of interest, and not from the ecological or the participants’ perspective alone (Araújo et al., 2007: 74). Indeed, in many experimental approaches to lithic studies, the concern for ecological validity in conducting experiments has been emphasized at the expense of both internal and external validity. When discussing novel force application techniques, the current literature unanimously considers soft hammer and pressure flaking as discrete and absolute categories in hammer material and hammer displacement speed, respectively. For example, soft hammer use has been associated consistently with specific materials such as antler, bone and soft or cortical stone. What the results of our experiment suggest is that the relationship between flake attributes and hammer material is more of gradational rather than of punctuated quality, and that hammer hardness is a property that is relative to the knapped raw material. Ethnographic accounts have documented instances where knapping activities at quarries were carried out with hammers of ordinary nodules of the same raw material used for flake production (Binford and O’Connell, 1984). Similarly, Gould et al. (1971), Hayden (1979), and White (1967) have all encountered the flexible use of wood, stone nodule, core, and flakes as hammer for retouching in their ethnographic studies. While this is not to say that all past knapping practices were performed in such opportunistic fashion, it does suggest that there is a need for a greater understanding of how fundamental variables interact in flake formation. Again, this means that we should be cautious in applying techniques developed by 21st century archaeologists directly to the interpretation of archaeological materials. This resonates with the concern that replicative experiments “can indicate at best only some way(s) in which artifacts could have been manufactured. Nevertheless, it does not necessarily indicate all other possible ways of manufacturing the same specimen.” (Hayashi, 1968: 129) Indeed, this point is exemplified by the multitude of techniques identified by modern flintknappers for arriving at similar artifact forms, including Upper Paleolithic blades (Bordes, 1967; Bordes and Crabtree, 1969; Tixier, 1972; Newcomer, 1975), Folsom points (Crabtree, 1966), and pressure flaked blades (Sollberger and Patterson, 1976; Pelegrin, 2012). This is largely an issue of equifinality, by which different configurations of the independent variables can result in the same flake morphology. As made evident by this and previous controlled experiments, we know there was a complex interaction of variables that determine how stone knappers in the past controlled their products in varied ways to achieve similar outcomes. It is unlikely that any one of these strategies can be confirmed without the presence of significant archaeological evidence. While there is evidence that potentially supports the occurrence of some of these flintknapping

47

innovations, such as the soft hammer billets found at Boxgrove (Bergman and Roberts, 1988), or similar finds within the Upper Paleolithic of Laugerie-Haute Ouest (Bordes, 1974), the results here suggests that it may be more important to consider hominins’ changing ability to combine various independent variables and manage their respective trade-offs to produce stone artifacts under a variety of contexts. So, while we acknowledge the importance of replicative experiments, we also suggest that such experimentation is merely one step in understanding ancient knapping practices, a step that must be further tested through experiments like the one presented here. Replicative studies serve an important role in hypothesis generation and help to illuminate potential relationships among certain variables. However, these potential relationships then need to be broken down further. In fact, this group has tested and confirmed many hypotheses put forward in previous replicative studies, thus demonstrating the utility of such experimental designs. In other cases, however, cause-and-effect hypotheses generated through replicative experiments have been either shown to be false or that other variables play a larger role than previously thought. Clearly, both kinds of experiments contribute to reconstructing past behavior, and it is important to acknowledge the limitations and potentials of each. 5. Conclusion Of course, archaeologists want to make inferences about the techniques used in the past, but because the independent variables investigated here are archaeologically invisible, such inferences must be based on the evidence at hand e the morphology and other attributes of the stone objects that we recover. The purpose of these experiments is to understand the effects that each of these independent variables may have on flake attributes, but it is a different problem to try to reconstruct the use of those variables based on the flaked products themselves. As we learn more about the effects of any particular variable, it is becoming increasingly apparent that many of them interact simultaneously. Furthermore, there are still many other independent variables that have not yet been analyzed, including the properties of different raw materials. What has been shown here is that our ability to unambiguously infer a particular technique of force application on the basis of particular flake attributes should be viewed with considerable caution. One the one hand, there are many interactions among various independent variables and, on the other hand, the strength of the relationship between one independent variable and one dependent variable may not be strong enough to infer the former from the latter. The findings presented here also demonstrate some of the issues associated with the traditional reliance on replicative flintknapping studies as the primary source of inference. Applying replicative work as a means of hypothesis generation is likely to be more fruitful, and clearly replicative experiments can highlight relationships to be tested further with more controlled experimentation. However, the issue with equifinality means that the ability to replicate certain artifact morphological outcomes through flintknapping experiments alone cannot justify an inductive interpretation about the past based on the limited range of modern observations. As Hayashi (1968: 129) rightly noted more than forty years ago, without proper control over fundamental principles that govern the process of flake formation, replicative experiments “may become a mere empirical exercise” that demonstrates only particular ways out of many of achieving a specific outcome. Instead, the study presented here is part of a series of effort that aims to establish the fundamental relationships and mechanics that govern basic stone flake variability, which will serve as a sound

48

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49

foundation for constructing inferences concerning the nature and change of hominin technological behavior through time and across space.

Acknowledgments Funding for these experiments is provided by National Science Foundation ( BCS-0649673 and BCS e 1153192). We thank Henry Halem (Halem Studios) for producing the glass cores and Alex Radin (Laboratory for Research on the Structure of Matter, University of Pennsylvania) for Rockwell hardness tests on experimental materials and his continued help in constructing the equipment necessary to conduct this experiment. Thanks also to the anonymous reviewers of the original submission for their helpful comments and suggestions.

References Ambrose, S.H., 2001. Paleolithic technology and human evolution. Science 291, 1748e1753. Amick, D.S., Mauldin, R.P., Tomka, S.A., 1988. An evaluation of debitage produced by experimental bifacial core reduction of a Georgetown chert nodule. Lithic Technol. 17, 26e36. Araújo, D., Davids, K., Passos, P., 2007. Ecological validity, representative design, and correspondence between experimental task constraints and behavioral setting: comment on Rogers, Kadar, and Costall (2005). Ecol. Psychol. 19, 69e78. Bergman, C.A., Roberts, M.B., 1988. Flaking technology at the Acheulean site of Boxgrove, West Sussex, England. Rev. Archéol. Picardie 2, 105e113. Binford, L.R., O’Connell, J.F., 1984. An Alyawara day: the stone quarry. J. Anthropol. Res. 40, 406e432. Bleed, P., 1986. The optimal design of hunting weapons: maintainability or reliability. Am. Antiq. 51, 737e747. Bonnichsen, R., 1977. Models for Deriving Cultural Information from Stone Tools. Ottawa, National Museum of Man Mercury Series, Archaeological Survey of Canada paper 60. Bordes, F., 1967. Considérations sur la typologie et les techniques dans le Paléolithique. Quartär 18, 25e55. Bordes, F., 1974. Percuteur en bois de renne du Solutreen supérieur de LaugerieHaute Ouest. In: Camps-Fabrer, H. (Ed.), Premier Colloque International sur L’Industrie de L’Os dans La Préhistoire. Editions de L’Université de Provence, Provence, pp. 97e100. Bordes, F., Crabtree, D., 1969. The Corbiac blade technique and other experiments. Tebiwa 12, 1e21. Braun, D.R., Plummer, T., Ferraro, J.V., Ditchfield, P., Bishop, L.C., 2009. Raw material quality and Oldowan hominin toolstone preferences: evidence from Kanjera South, Kenya. J. Archaeol. Sci. 36, 1605e1614. Bril, B., Smaers, J., Steele, J., Rein, R., Nonaka, T., Dietrich, G., Biryukova, E., Hirata, S., Roux, V., 2011. Functional mastery of percussive technology in nut-cracking and stone-flaking actions: experimental comparison and implications for the evolution of the human brain. Philos. Trans. R. Soc. B 367, 59e74. Brown, K.S., Marean, C.W., Herries, A.I., Jacobs, Z., Tribolo, C., Braun, D., Roberts, D.L., Meyer, M.C., Bernatchez, J., 2009. Fire as an engineering tool of early modern humans. Science 325, 859e862. Campbell, D.T., Stanley, J.C., 1963. Experimental and Quasi-experimental Designs for Research. Rand McNally, Chicago. Clarkson, C., Hiscock, P., 2011. Estimating original flake mass from 3D scans of platform area. J. Archaeol. Sci. 38, 1062e1068. Cotterell, B., Kamminga, J., 1979. The mechanics of flaking. In: Hayden, B. (Ed.), Lithic Use-wear Analysis. Academic Press, New York, pp. 97e112. Cotterell, B., Kamminga, J., 1987. The formation of flakes. Am. Antiq. 52, 675e708. Cotterell, B., Kamminga, J., 1990. Mechanics of Pre-industrial Technology: an Introduction to the Mechanics of Ancient and Traditional Materials Culture. Cambridge University Press, Cambridge. Crabtree, D.E., 1966. Stoneworker’s approach to analyzing and replicating the Lindenmeier Folsom. Tebiwa 9, 3e39. Crabtree, D.E., 1967. Notes on experiments in flintknapping: 4dtools used for making flaked stone artifacts. Tebiwa 10, 60e73. Crabtree, D.E., 1972. The cone fracture principle and the manufacture of lithic materials. Tebiwa 15, 29e36. Crabtree, D.E., Swanson, E.H., 1968. Edge-ground cobbles and blade-making in the Northwest. Tebiwa 11, 50e58. Debénath, A., Dibble, H.L., 1994. Handbook of Paleolithic Typology. University Museum of Archaeology and Anthropology, Philadelphia. Desrosiers, P.M. (Ed.), 2012. The Emergence of Pressure Blade Making: from Origin to Modern Experimentation. Springer, New York. Dibble, H.L., Pelcin, A.W., 1995. The effect of hammer mass and velocity on flake mass. J. Archaeol. Sci. 22, 429e439.

Dibble, H.L., Rezek, Z., 2009. Introducing a new experimental design for controlled studies of flake formation: results for exterior platform angle, platform depth, angle of blow, velocity, and force. J. Archaeol. Sci. 36, 1945e1954. Dibble, H.L., Whittaker, J., 1981. New experimental evidence on the relation between percussion flaking flake variation. J. Archaeol. Sci. 6, 283e296. Eren, M.I., Lycett, S.J., 2012. Why Levallois? A morphometric comparison of experimental ‘preferential’ Levallois flakes versus debitage flakes. PLoS One 7, e29273. http://dx.doi.org/10.1371/journal.pone.0029273. Faulkner, A., 1972. Mechanical Principles of Flintworking (Unpublished Ph.D. thesis). Department of Anthropology, Washington State University. Fuglestvedt, I., 2007. The Ahrensburgian Galta site in SW Norway. Dating, technology and cultural affinity. Acta Archaeol. 78, 87e110. Gibson, J.J., 1979. The Ecological Approach to Visual Perception. Houghton Mifflin, Boston. Gould, R.A., Koster, D.A., Sontz, A.H.L., 1971. The lithic assemblage of the Western Desert Aborigines of Australia. Am. Antiq. 36, 149e169. Hayashi, K., 1968. The Fukui microblade technology and its relationships in Northeast Asia and North America. Arctic Anthropol. 5, 128e190. Hayden, B., 1979. Paleolithic Reflections: Lithic Technology and Ethnographic Excavation Among the Australian Aborigines. Humanities, Atlantic Highlands. Hayden, B., 1987. From chopper to celt: the evolution of resharpening techniques. Lithic Technol. 16, 33e43. Hayden, B., Hutchings, W., 1989. Whither the billet flake? In: Amick, D., Mauldin, R. (Eds.), Experiments in Lithic Technology. BAR International Series, Oxford, pp. 235e257. Henry, D.O., Haynes, C.V., Bradley, B., 1976. Quantitative variations in flaked stone debitage. Plains Anthropol. 21, 57e61. Hutchings, W.K., 1997. The Paleoindian Fluted Point: Dart or Spear Armature? The Identification of Paleoindian Delivery Technology through the Analysis of Lithic Fracture Velocity (Unpublished Ph.D. dissertation). Department of Archaeology, Simon Fraser University. Hutchings, W.K., 1999. Quantification of fracture propagation velocity employing a sample of Clovis channel flakes. J. Archaeol. Sci. 26, 1437e1447. Hutchings, W.K., 2011. Measuring use-related fracture velocity in lithic armatures to identify spears, javelins, darts, and arrows. J. Archaeol. Sci. 38, 1737e1746. Kelly, R.L., 1988. The three sides of a biface. Am. Antiq. 53, 717e734. Kelly, R.L., Todd, L.C., 1988. Coming into the country: Early Paleoindian hunting and mobility. Am. Antiq. 53, 231e244. Kerkhof, F., Müller-Beck, H., 1969. Zur Bruchmechanischen Deutung der Schlagmarken an Steingeraten. Glastech. Berichte 42, 439e448. Kirk, R.E., 2009. Experimental design. In: Millsap, R.E., Maydeu-Olivares, A. (Eds.), The SAGE Handbook of Quantitative Methods in Psychology. Sage Publications, Thousand Oaks, pp. 23e45. Kirk, R.E., 2012. Experimental Design: Procedures for the Behavioral Sciences, fourth ed. Sage Publications, Thousand Oaks. Lin, S.C., Rezek, Z., Braun, D., Dibble, H.L., 2013. On the utility and economization of unretouched flakes: the effects of exterior platform angle and platform depth. Am. Antiq. 78, 724e745. Lycett, S.J., Eren, M.I., 2013. Levallois lessons: the challenge of integrating mathematical models, quantitative experiments and the archaeological record. World Prehist. 45, 519e538. Mellars, P., 2006. Why did modern human populations disperse from Africa ca. 60,000 years ago? A new model. Proc. Natl. Acad. Sci. U. S. A. 103, 9381e 9386. Mesoudi, A., 2011. Cultural Evolution: How Darwinian Theory Can Explain Culture and Synthesize the Social Sciences. Chicago University Press, Chicago. Morrow, J.E., 1995. Clovis projectile point manufacture: a perspective from the Ready/Lincoln Hills site, 11JY46, Jersey County, Illinois. Midcont. J. Archaeol. 20, 167e191. Mourre, V., Villa, P., Henshilwood, C.S., 2010. Early use of pressure flaking on lithic artifacts at Blombos Cave, South Africa. Science 330, 659e662. Nelson, M.C., 1991. The Study of technological organization. In: Schiffer, M.B. (Ed.), Archaeological Method and Theory, vol. 3. University of Arizona Press, Tucson, pp. 57e100. Newcomer, M.H., 1971. Some quantitative experiments in hand-axe manufacture. World Archaeol. 3, 85e94. Newcomer, M.H., 1975. “Punch technique” and Upper Paleolithic blades. In: Swanson, E.H. (Ed.), Lithic Technology: Making and Using Stone Tools. Mouton Publishers, The Hague, pp. 97e104. Noll, M., 2000. Components of Acheulian Lithic Assemblage Variability at Olorgesailie, Kenya (Unpublished Ph.D. dissertation). Department of Anthropology, University of Illinois at Urbana-Champaigne. Ohnuma, K., Bergman, C., 1982. Experimental studies in the determination of flaking mode. Bull. Inst. Archaeol. 19, 161e170. Patterson, L.W., 1982. Replication and classification of large size lithic debitage. Lithic Technol. 11, 50e58. Pelcin, A.W., 1997a. The effect of indentor type on flake attributes: evidence from a controlled experiment. J. Archaeol. Sci. 24, 613e621. Pelcin, A.W., 1997b. The formation of flakes: the role of platform thickness and exterior platform angle in the production of flake initiations and terminations. J. Archaeol. Sci. 24, 1107e1113. Pelegrin, J., 2012. New experimental observations for the characterization of pressure blade production techniques. In: Desrosiers, P.M. (Ed.), The Emergence of Pressure Blade Making: From Origin to Modern Experimentation. Springer, New York, pp. 465e500.

M. Magnani et al. / Journal of Archaeological Science 46 (2014) 37e49 Rezek, Z., Lin, S., Iovita, R., Dibble, H.L., 2011. The relative effects of core surface morphology on flake shape and other attributes. J. Archaeol. Sci. 38, 1346e 1359. Sahle, Y., Hutchings, W.K., Braun, D.R., Sealy, J.C., Morgan, J.E., Negash, A., Atnafu, B., 2013. Earliest stone-tipped projectiles from the Ethiopian Rift date to >279,000 years ago. PLoS One 8, e78092. http://dx.doi.org/10.1371/journal.pone.0078092. Schick, K.D., Toth, N., 1993. Making Silent Stones Speak: Human Evolution and the Dawn of Technology. Simon and Schuster, New York. Schindler, B., Koch, J., 2012. Flakes giving you lip? Let them speak: an examination of the relationship between percussor type and lipped platforms. Archaeol. East. N. Am. 40, 99e106. Sheets, P.D., 1973. Edge abrasion during biface manufacture. Am. Antiq. 38, 215e 218. Shott, M.J., 1996. An exegesis of the curation concept. J. Anthropol. Res. 52, 259e 280. Sollberger, J.B., Patterson, L.W., 1976. Prismatic blade replication. Am. Antiq. 41, 517e531. Soriano, S., Villa, P., Wadley, L., 2007. Blade technology and tool forms in the Middle Stone Age of South Africa: the Howiesons Poort and post-Howiesons Poort at rose Cottage Cave. J. Archaeol. Sci. 34, 681e703. Speth, J., 1972. Mechanical basis of percussion flaking. Am. Antiq. 37, 34e60. Speth, J., 1981. The role of platform angle and core size in hard-hammer percussion flaking. Lithic Technol. 10, 16e21.

49

Stout, D., 2011. Stone toolmaking and the evolution of human culture and cognition. Philos. Trans. R. Soc. B 366, 1050e1059. Tixier, J., 1972. Obtention de lames par débitage “sous le pied.” Bull. Soc. Préhist. Fr. 69, 134e139. Tomenchuk, J., 1985. The Development of a Wholly Parametric Use-wear Methodology and its Application to Two Selected Samples of Epipaleolithic Chipped Stone Tools from Hayonim Cave, Israel (Unpublished Ph.D. dissertation). Department of Anthropology, University of Toronto. Van Peer, P., Vermeersch, P.M., Paulissen, E., 2010. Chert Quarrying, Lithic Technology and a Modern Human Burial at the Palaeolithic Site of Taramsa 1, Upper Egypt. Leuven University Press, Leuven. Villa, P., Soriano, S., Teyssandier, N., Wurz, S., 2010. The Howiesons Poort and MSA III at Klasies River main site, cave 1A. J. Archaeol. Sci. 37, 630e655. Weaver, A.H., 2005. Reciprocal evolution of the cerebellum and neocortex in fossil humans. Proc. Natl. Acad. Sci. U. S. A. 102, 3576e3580. White, J.P., 1967. Ethnoarchaeology in New Guinea: two examples. Mankind 6, 409e 414. Whittaker, J., 1994. Flintknapping: Making and Understanding Stone Tools. University of Texas Press, Austin. Witthoft, J., 1974. The mechanics of conchoidal fracture: a bibliographic notes. Newsl. Lithic Technol. 3, 50e53.