Manual restrictions on Palaeolithic technological behaviours

Introduction

The production and use of flaked stone tools were likely important to the survival of Palaeolithic hominins. The potential influence of these manually demanding behaviours on the evolution of the human hand has long been recognised (Napier, 1962; Marzke, 1983; Marzke, 1997; Marzke, 2013; Williams, Gordon & Richmond, 2010; Rolian, Lieberman & Zermeno, 2011; Key & Lycett, 2011; Kivell, 2015; Almécija & Sheerwood, 2017; although see Almécija & Alba, 2014). Recent research has also demonstrated how the manual anatomy and associated biomechanical capabilities of different hominin species may have influenced the nature of the Palaeolithic archaeological record (Marzke & Shackley, 1986; Rolian, Lieberman & Zermeno, 2011; Domalain, Bertin & Daver, 2017; Key et al., 2017; Patiño et al., 2017; Key & Lycett, in press). That is, the types, forms and technological strategies of stone tool artefacts may have been limited by, or preferentially selected for, as a result of how effectively hominins could use the hand when manipulating or securing lithic objects. Research concerning how the evolution of the hominin hand may have been influenced by stone tool production and use has been reviewed in detail elsewhere (Marzke, 1997; Marzke, 2013; Kivell, 2015; Almécija & Sheerwood, 2017). The present article reciprocally focuses on the influence that hominin manual capabilities may have had on the types and forms of stone tools produced during the Lower Palaeolithic.

The earliest intentionally flaked stone tools are currently from the 3.3 million-year-old site of Lomekwi 3, West Turkana (Kenya), and appear to have been directed towards producing large flake cutting tools through passive-hammer or bipolar percussive strategies (Harmand et al., 2015; Lewis & Harmand, 2016). Subsequent to ∼2.6 million years ago (Mya) Oldowan flake and core technologies, typically characterised by the expedient production of variably sized flake cutting tools from hand-held cores using hard-hammer percussion, appear more widely across East Africa (Kimbel et al., 1996; Roche et al., 1999; Semaw et al., 2003; Rogers & Semaw, 2009; Hovers, 2012; Reti, 2016; Proffitt, 2018). Simple flake and core stone tools are thereafter ubiquitous throughout the Palaeolithic. After ∼1.75 Mya large bifacially flaked core tools (‘bifaces’) appear in the archaeological record across East Africa as part of the Acheulean techno-complex (Lepre et al., 2011; De la Torre & Mora, 2014; Diez-Martín et al., 2015). These early Acheulean tools, often characterised by handaxes and cleavers, are typically thought to have been produced using hard-hammer percussion. Bifaces go on to typify the next >1 million years of the archaeological record across the Old World (Lycett & Gowlett, 2008; Gowlett, 2015; Moncel et al., 2015) until the onset of Middle Palaeolithic technologies ∼300 Kya (Moncel et al., 2011; Tryon & Faith, 2013; Adler et al., 2014). The nature and extent of any chronological changes to stone technology during the Acheulean are debated (e.g., Vaughan, 2001; Chauhan, 2009; Gowlett, 2011; McNabb & Cole, 2015; Moncel et al., 2015; Gallotti, 2016), however, there are indications that later Acheulean bifacial tools (handaxes in particular) were at times produced using soft-hammer percussion, became thinner relative to their width (more ‘refined’), displayed greater evidence of intentional thinning, volume control (mass distribution), investment (e.g., time, skill), shaping and symmetry (Gowlett, 1986; Saragusti et al., 1998; Schick & Clark, 2003; Grosman, Goldsmith & Smilansky, 2011; Beyene et al., 2012; García-Medrano et al., 2014; Li et al., 2018; Moncel et al., 2016; Gallotti & Mussi, 2017; Iovita et al., 2017; Shimelmitz et al., 2017), and at times displayed evidence of platform preparation prior to a flake’s removal (Stout et al., 2014). Together, these technologies describe ∼3 million years of stone tool production and use during the Lower Palaeolithic.

Relationships between technological or morphological aspects of Lower Palaeolithic stone tools and hominin manual capabilities are often mentioned, but rarely tested, in archaeological literature (e.g., Crompton & Gowlett, 1993; Delagnes & Roche, 2005; Machin, 2009; Lycett & Von Cramon-Taubadel, 2015). Although paleoanthropologists frequently debate whether fossil hominin hand anatomy could facilitate stone tool related precision grips, it is rarely the case that specific technological or morphological aspects of these tools are discussed (although see Tocheri et al. (2008) for an example). Therefore, there are only a few instances where hypothesised relationships between technological or morphological features of Lower Palaeolithic stone tools and hominin manual capabilities have actually been investigated.

Regarding the origin of the first flaked stone tools, Rolian, Lieberman & Zermeno (2011) used a metal ‘simulated flake tool’ to calculate the external moments, internal flexion moments and joint stresses of tool users. Their data suggested that efficient flake tool use with low biomechanical stresses may not have been possible prior to the evolution of the derived pollical anatomy observed in later Homo (Rolian, Lieberman & Zermeno, 2011). Recently, Key & Lycett (in press) demonstrated the significant impact that tool user biometric variation can have on stone tool-use efficiency across the Lower Palaeolithic, revealing that relationships between biometric parameters and tool-use efficiency depend on the type of tool being used and the biometric variable under consideration. Their results suggest that the effective use of flakes and handaxes is not only dependent on hominins displaying relatively strong hands, but that the onset of Acheulean handaxes may have been linked to the evolution of more anatomically modern manual dimensions (Key & Lycett, in press). Williams-Hatala et al.’s (2018) investigation of manual pressure variation during flake and handaxe use may also indicate there to be differences in grip loading levels dependent on the size of the tool gripped.

These results are, in part, due to the variable grips required when securing different Lower Palaeolithic tools, as described by Marzke & Shackley (1986). Manual demands and grip choices have also been demonstrated to vary during different stone tool production sequences (Marzke & Shackley, 1986). Comparisons between flake and handaxe production, for example, identified differences in the motion of the dominant arm, with the latter requiring smaller, more precise flaking actions. The authors also suggest that a ‘lighter grip’ could be used to secure an Oldowan flake core relative to a handaxe or pick when detaching flakes (Marzke & Shackley, 1986). As cores become smaller over a reduction sequence, Marzke and colleagues (Marzke & Shackley, 1986; Marzke et al., 1998) describe how the distal aspects of digits are increasing heavily recruited and the palm is used less. An early experiment also suggested that lower thumb to finger length ratios may have precluded early hominin’s ability to firmly secure handaxes during production, in turn resulting in “very crude handaxes” during the early Acheulean (Krantz, 1960: 116). Key et al. (2017) found that experienced knappers gripped hammerstones with high pressure when detaching particularly large flakes. In turn, large stone flakes within Lower Palaeolithic archaeological sequences (Sharon, 2010; Shipton et al., 2014) plausibly indicate that hominins were capable of exerting and resisting high manual pressures during precision (hammerstone) manipulation.

Aside from manual requirements, other studies emphasise the increased cognitive demands of handaxe production relative to Oldowan flakes (Stout et al., 2008; Muller, Clarkson & Shipton, 2017), while Mateos, Terradillos-Bernal & Rodriguez (in press) have recently experimentally compared the energetic cost of soft and hard hammer handaxe production. Only Faisal et al. (2010), however, have empirically examined Lower Palaeolithic technological transitions from a manipulative perspective. Joint angles on, and abduction angles between, the digits of the non-dominant hand of a skilled flint knapper indicated that, at least for the individual under investigation, Acheulean and Oldowan stone tool production are “indistinguishable” in terms of manipulative complexity (Faisal et al., 2010: 6). Faisal et al.’s (2010) study also highlights key manipulative differences between these two reduction sequences, including the unique need to properly and securely brace a handaxe as it becomes increasing thin relative to its width.

Together, these studies emphasise the distinct manual demands required by the type and form of stone tool being used or produced. These demands must be facilitated by effective grips, which are, in turn, facilitated by anatomical adaptations. Without this anatomy it is unlikely that the respective tool forms would be found in associated archaeological deposits. Yet, there is still relatively little known about hand recruitment during the production of different types and forms of stone tool. Further, there is limited information about the effect biomechanical variation in a tool producer’s hand has on the efficacy of different stone tool production behaviours. Certainly, the onset and adoption of certain technological or morphological features in the Palaeolithic archaeological record could have been restricted by biomechanical capabilities, including the forceful precision grip capabilities of the hominin upper limb.

The non-dominant hand is known to experience high loading levels and perform complex manipulative tasks during the production of stone tools (Marzke & Shackley, 1986; Faisal et al., 2010; Key & Dunmore, 2015), perhaps to a greater extent than the dominant hand. Differences in manipulative requirements between stone tool production behaviours might, then, be more readily detected in this hand relative to the dominant hand. Here, we test the null hypothesis that the pressures experienced across the non-dominant hand of stone tool producers during a series of Lower Palaeolithic technological activities, including a range of tool types produced and percussors used, are not significantly different. Further, we assess how flake removal success is related to the pressure used to secure cores and whether manual pressures vary according to the stage of a core’s reduction, or the technique used to support a core against hammerstone impact reaction forces.

Methods

Reduction strategies and technological differences

Three Lower Palaeolithic reduction strategies are examined here: (1) the production of replica Oldowan flake tools (‘flake’), (2) bifacial flake removals while shaping an ‘Early Acheulean’ handaxe (EAH), and (3) bifacial flake removals while shaping a ‘Late Acheulean’ handaxe (LAH) (Figs. 1 and 2). Both flake and EAH tools were produced via hard hammer percussion while LAH were produced with soft hammer percussion as well. The latter strategy also employed specialist grinding stones during the preparation of flake platforms. The terms EAH and LAH used here refer to general increases in flaking extent, shaping, volume control, symmetry, the use of intentional ‘thinning’ flakes, soft-hammer percussion and prepared flake platforms in later Acheulean handaxes (Saragusti et al., 1998; Schick & Clark, 2003; Grosman, Goldsmith & Smilansky, 2011; Diez-Martín et al., 2014; Stout et al., 2014; Gallotti & Mussi, 2017; Iovita et al., 2017; Shimelmitz et al., 2017). While these differences are often clearest when tools produced >1 Mya are compared to those produced after ∼0.5 Mya, we do not mean to imply uniform linear progression of forms across regional records (Vaughan, 2001; Gowlett, 2013; Moncel et al., 2015; McNabb & Cole, 2015). Rather, we seek to investigate if handaxe forms produced using distinct techniques may be limited by biomechanical capabilities, as inferred from manual pressure records (see below).

Although the translation and rotation of cores are manually demanding behaviours (Marzke et al., 1998; Key & Dunmore, 2015), the present analysis focuses only on manual pressure while securing cores during flake removals or platform preparation activities (edge grinding, retouching and trimming). As these behaviours remove mass from a core, they shape a lithic artefact and have the potential to be identified from the archaeological record.

Nine skilled flint knappers, each with at least five years experience, took part in the study. At a minimum, all individuals were capable of consistently producing replica Acheulean handaxes of predetermined form when required. Notably, some of the participants exceeded this lower skill threshold by a considerable margin (cf. Eren et al., 2014). All had previously knapped while connected to manual pressure sensors and are familiar with producing tools within other experimental conditions (Winton, 2005; Williams, Gordon & Richmond, 2010; Key & Dunmore, 2015; Key et al., 2017). Additionally, most knap on a professional and frequent basis (e.g., academic, craftsman etc.) and likely provide the best possible sample available for providing natural, unfettered, pressure data. For these reasons, we are confident in the use of a single trial per reduction strategy for each knapper (collected within a single day) and the repeatability of the data collected. Each individual undertook the flake reduction first, followed by the EAH and then LAH sequence (Fig. 3). British flint from Suffolk and Kent was used in all reductions. All tool production sequences were recorded using a HD video camera. Ethical approval was granted by the School of Anthropology and Conservation Ethics Committee (University of Kent; Ref. Ares 19065). All individuals gave informed consent.

Each knapper used their own hammerstones and soft hammers, without restriction, although red deer (Cervus eleghus) and moose (Alces alces) billets were typically used. No wooden or copper billets were used. Knappers were free to use grinding stones during platform preparation events in the LAH reduction, although in many instances soft and hard hammers were also used for grinding and trimming (Figs. 3 and 4). Knappers produced flakes at their own pace and supported the core in whatever way they preferred (this varied between the core resting in the hand or on the leg). Every attempted flake removal was coded as successful if the flake detached or unsuccessful if it did not. In instances where a fracture had clearly propagated through the core but required additional minor taps to remove it, the original hammer strike was considered successful and the small taps were not included in the study. Small (micro) flake removals undertaken when preparing platforms for large flake’s removals are considered as distinct to ‘flake removals’ in this study.

Pressure sensors

A wireless Novel Pliance^® sensor system was used to record the pressures (kPa) experienced across the non-dominant hand of knappers during all three reductions (Fig. 5). The system was comprised of 10 17 ×17 mm² and two 10 ×10 mm² sensors. The larger sensors were attached to the distal and proximal phalanges of digits 1–4 as well as the intermediate phalanges of digits 2 and 3. The two smaller sensors were attached to the distal and proximal phalanges of digit 5 (Fig. 5). All sensors were attached to the palmar surfaces of digits using double-sided tape and Velcro straps. Latex finger cots were used to protect the sensors and help keep them in place. The sensors were ‘zeroed out’ prior to data collection starting to account for any potential pressure caused by the finger cots. In all instances data were collected at a rate of 50 Hz.

Data extraction

Reduction sequences ranged between 5 to 34 min in duration. The number of individual data points collected from sensors ranged from ∼12,000 to ∼102,000. To identify individual behavioural instances within data streams it was necessary to align the pressure data output with the video records of each reduction sequence. Knappers were asked to free their non-dominant hand of any loads prior the reduction sequence starting and forcefully pinch their thumb and index finger. This created a known behaviour that was clearly identifiable at the start of the pressure data and the video record, after which, the two outputs could be accurately aligned.

Every time one of the behaviours under investigation was performed the peak pressure (kPa) experienced on each sensor was identified and recorded. For an attempted flake removal, peak pressures were identified from 2-second-long segments of the data stream (1 second either side of the point of impact; Fig. 6). Platform preparation behaviours could occur for substantially longer periods, therefore peak pressures were extracted from across their entire duration. Every manual activity recorded here, and therefore every peak pressure value, was assigned a technological strategy (flake, EAH, LAH), an indenture type (hard hammer, soft hammer, grinding stone), a removal type (successful flake, unsuccessful flake, platform preparation), a core-support position (leg, hand), and a sequence number.

Pressure data from all 12 sensors were summed to produce a record of the digital peak pressures experienced at a whole-hand level during individual technological behaviours. For each statistical comparison the peak pressures from all nine participants were combined. Participant seven’s distal sensor on the first digit became detached during his flake reduction sequence. To make this discrepancy equal across all conditions examined here, no data for this sensor from this participant were included in the analyses.

To control for inter-knapper differences in pressure, records were normalised to a 0-1 scale by dividing the difference between each peak pressure record and minimum record of that reduction sequence, by the range of values in that reduction. Since all reductions begin in a similar manner this scaling should not preclude the identification of significant differences between groups.

Results

Descriptive data for the pressure values used in each analysis are detailed in Tables 1–5. Between the three types of tool production sequence there were substantially more mass removal events when producing LAHs (n = 1,503), relative to flakes and EAHs (n = 506 and 777 respectively; Table 1). Around twice as many flake removals were required during the production of LAHs relative to EAHs. Mean, summed peak pressure records across the non-dominant hand during the production of LAHs were also greater than the flake and EAH sequences by ∼50 kPa (Table 1; Fig. 7). The Friedman test did not reveal significant differences between median pressures used in the three types of reduction (p = 0.22138) and so post-hoc tests were not conducted. Although the production of ‘Late Acheulean Handaxes’ required greater mean pressures to be exerted and resisted by the non-dominant hand across all data collected, compared to the production of Oldowan flake tools or ‘early Acheulean handaxes’, these differences were not significant.

Ratios of successful to unsuccessful flake removals varied only slightly between the three reduction strategies (ranging between 7:2 and 9:2) (Table 2). In each strategy, successful flake removals reported pressure values ∼10–15 kPa below unsuccessful removals (Table 2). Mann–Whitney U tests identified that these differences were not significant in any of the three sequences (p = .069–.249). In turn, the success of flake removals does not seem to be a consequence of variation in pressure exerted by the non-dominant hand during stone tool production, although there is consistency in successful flake removal recording marginally lower pressure values.

Core support strategies varied between the leg and hand in all three reductions. In terms of data frequency there is a split between flake production, which reports greater use of hand support, the EAH reductions which are broadly equal between the two, and the LAH reductions where there were clear preferences for cores being supported by the leg (Table 3). While no significant pressure difference is recorded between the hand and leg support techniques during flake production (p = .060), both of the handaxe sequences report significant differences (p =   ≤ .001; Table 3). However, during the EAH reduction greater pressure values are reported during leg support while LAHs report greater values during hand support (Table 3). The technique used to support a stone core therefore appears related to the pressures required to secure it during flake removals and platform preparation events, however, differences appear dependent on the type of tool being produced.

It was only possible to compare hard hammer flake removals, soft hammer flake removals, and platform preparation events during the LAH reduction sequence. Across the nine participants there were equal numbers of hard and soft hammer flake removals (n = 617 for each removal type), suggesting that both types of percussor are equally important during LAH production sequences (Table 4). There were, however, 4.6 times as many flake removals relative to platform preparation events, indicating that only ∼one in five flakes required its platform to be prepared prior to its removal. When only soft hammer percussion is considered, where platform preparation may more normally be expected, every other flake was removed without its platform being prepared (i.e., one in two flakes had its platform prepared). Soft hammer percussion returned, on average, the lowest peak pressure records across the hand (Table 4; Fig. 7). Hard hammer percussion required an additional 33 kPa of pressure to be exerted and resisted by the non-dominant hand. An additional 59 and 92 kPa were recorded, on average, across the non-dominant hand of knappers during platform preparation events compared to hard and soft hammer percussion, respectively (Table 4; Fig. 7). The Friedman test between normalised median pressures used in the three types of mass removal was significant (p = .0001). Subsequent pairwise Wilxcoxon signed rank tests indicated that platform preparation events required significantly more pressure than both hard (p = 0.0002) and soft hammer (p = .0043) removals, while there was no significant difference between the latter two mass removal types. Platform preparation events do, therefore, appear to require significantly greater pressure to be exerted and resisted by the non-dominant hand compared to both hard and soft hammer flake removals.

The LAH data values used during the regression analyses were, on average, greater than both the flake and EAH reductions (by 34 and 45 kPa, respectively) despite the absence of platform preparation events (Table 5), demonstrating that even in the absence of this uniquely late Acheulean behaviour, the production of LAH forms requires greater manual pressures. Of the eight linear regressions undertaken all identified significant relationships between flake removal sequence numbers and manual pressure (Table 6). Flake and EAH reduction sequences displayed negative relationships, whereby pressure decreased as reduction sequences progressed. LAH sequences and LAH platform preparation events displayed positive relationships, indicating that later mass removal events required greater manual pressures (Table 6). In all but one instance R²values were ≤.090, indicating that limited (≤ 9%) pressure variation could be attributed to a core’s stage of reduction. The single exception was the regression between LAH platform preparation sequence numbers and their respective pressure values, where 42% of the observed pressure variation could be attributed to the stage of a handaxe’s production (Table 6; Fig. 8). This indicates that as late Acheulean handaxes progress further through production sequences (i.e., as they become smaller, increasingly shaped and thin relative to their thickness) the pressure required to stabilise them during platform preparation events increases significantly. The fact that this relationship is not similarly repeated in the normalised flake removal sequence numbers indicates that this relationship is unlikely to be driven by how close a handaxe is to being considered finished by the knapper, but by how long the sequence goes on for, how many flakes have been removed, and how ‘refined’ a biface becomes.

Discussion

The present work investigates the origin of technological innovation during the Lower Palaeolithic from a biomechanical and evolutionary perspective, and asks whether the onset of new stone tool forms and production techniques may have been restricted by hominin manual capabilities. Our results demonstrate that although later Acheulean handaxes (LAH) required the exertion and resistance of greater manual pressure during their production relative to either Oldowan flake and core tools or early ‘rougher’ Acheulean handaxes (EAH) (by an average of 22% and 29%, respectively, when all data were considered), these differences were not found to be significant and may have been driven by a few individuals. It is, therefore, not possible to state that manual pressure requirements during flake detachments vary significantly between the three tools examined here.

However, the preparation of LAH flake platforms, through retouching and edge grinding, elicited the greatest loads in this study. Indeed, the action of preparing a flake’s platform prior to its removal required significantly (22–40%) more pressure than soft or hard hammer flake removals in the same reduction sequences (Table 4; Fig. 7). Compared to Oldowan or EAH flake removals, mean pressures are 55–59% (>110 kPa) greater during LAH platform preparation events (Tables 1 and 4; Fig. 7). This result suggests that platform preparation techniques may only have been possible for hominins capable of performing particularly forceful precision grips. These grips would have required greater force than those needed for earlier stone tool types. Arguably, only once hominins evolved enhanced manipulative capabilities in response to selective pressure exerted by earlier manual behaviours, would the innovation of later Acheulean handaxe forms, produced using the preparation of flake platforms, have been possible. Such behaviours include flake tool use, hammerstone use, and Oldowan/EAH core manipulation (Marzke, 1997; Marzke, 2013; Kivell, 2015; Key & Dunmore, 2015; Williams-Hatala et al., 2018). As highlighted by Tocheri et al. (2008), fossil hand anatomy indicates the continued derivation of hominin manual capabilities subsequent to the onset of the Acheulean, which may have facilitated the forceful grips used for securing the core during platform preparation events, required for LAH production.

During platform preparation events edges are modified either via the removal of very small flakes when isolating as well as reshaping platforms or altering their angles, or they can be reduced, bevelled, reshaped and isolated through forceful grinding actions. In each case, these actions require the precise but forceful application of stone or antler against the handaxe’s edge. In turn, it is essential for handaxes to remain stable throughout this process so that the percussor or grinding stone is applied only to the specific area being shaped (for refined bifaces flake platforms are often <10 × 5 mm). Regarding small flake removals, it is the highly precise nature of the removals that necessitates a particularly firm and steady grip on the handaxe.

The act of grinding a handaxe’s edge in preparation for a flake removal, however, also requires the input of substantial and prolonged forces through an abrasive stone onto the biface’s edge. In addition to their extended duration, it is likely that the dominant hand at times creates forces in excess of those observed during flake detachments. Certainly, during edge grinding the palm contributes substantially to the loads transferred onto a core, something that is impossible during most hammerstone strikes (and therefore flake detachments). While previous biomechanical studies of the dominant hand have tended to overlook edge grinding events (although see: Marzke & Shackley, 1986), and thus these claims cannot yet be substantiated, our pressure data clearly identifies a requirement to oppose substantial reaction forces during platform preparation events. More specifically, these pressures are significantly greater than those observed during flake removals.

When LAHs are secured during platform preparation events up to 42% of the pressure variation recorded here can be attributed to the stage of a handaxe’s production, demonstrating proportionally greater force is required to prepare platforms for progressively refined flake removals (Fig. 8). This relationship cannot be straightforwardly attributed to participant fatigue, as platform preparation events and flake removals were undertaken throughout reductions and no fatiguing was reported or observed. Rather the form of the handaxe (core) being supported and secured is likely responsible for this result.

As any reduction sequence progresses, cores become smaller (Clarkson, 2013; Douglass et al., 2018) and handaxe size has been experimentally demonstrated to have a strong negative relationship with reduction intensity (Shipton & Clarkson, 2015a; Shipton & Clarkson, 2015b). Marzke & Shackley (1986) found that as reduction sequences progress the thumb and distal aspects of the fingers are increasingly used in isolation when gripping the core to secure it against hammerstone strikes (see also: Pouydebat et al., 2009). As a corollary, both the greater surface area of the palm and the most ulnar digits (fourth and fifth) are used progressively less (Marzke et al., 1998), which concentrates manual forces on the radial three digits. This concentration of force thereby increases the pressures required to produce, typically smaller, LAH’s.

The stage of a handaxe’s reduction also has potential to impact its volumetric distribution and shape (Crompton & Gowlett, 1993). Archer & Braun (2010) demonstrated that as reduction sequences progress, a handaxe’s centre of mass moves first to the centre of the tool and subsequent thinning flakes move it to the tool’s base. As highlighted by Faisal et al. (2010), this results in an increased requirement to properly secure and brace the tool during flake removals and platform preparation events. Certainly, during the latter stages of LAH production there is increased risk that a biface will break (i.e., fracture in an unintended way) when flake removals are instigated. This may be through the intended fracture ‘diving’ through the biface when searching for the route of least resistance, or by reaction forces propagating through the tool and creating stress enough to fracture in additional locations (often the tip). In both cases, the principle means for a knapper to prevent these mistakes (other than choosing suitable flakes to remove) is by forcefully bracing the length of the biface. While most effect sizes were small, the other regression analyses support this idea as flake and EAH regressions display negative relationships with pressure but LAH sequences show positive relationships. During flake and EAH reductions the reducing core mass requires less support and stabilisation, resulting in lower manual pressure. While this may also characterise early stages of LAH production, as sequences progress, pressure increases substantially. It is likely that the production of bifacially flaked tools with even lower thickness to width ratios, such as Solutrean or Clovis points (e.g., Smallwood, 2010; Eren et al., 2013), would require even greater pressures.

Conclusion

The Lower Palaeolithic artefact record represents the largest and most detailed record of the minimum technological capabilities of hominins during the Plio-Pleistocene. As such the Oldowan and Acheulean periods track significant shifts in the behaviour of hominins, which have been investigated in terms of cognition, social transmission, environmental factors and others. Here we investigate these transitions from a biomechanical perspective, as inferred from manual loading data. Our results demonstrate that the digital pressures required to forcefully secure later Acheulean handaxes during their production are not significantly greater than those required when knapping earlier Acheulean handaxe forms or Oldowan flakes. However, the novel LAH associated behaviour of preparing flake platforms would have required significantly stronger grips in the non-dominant hand compared to earlier stone tool production behaviours. Therefore, we contend that the behavioural shift marked by the onset of platform preparation behaviours, as observed in later Acheulean handaxe forms, may be intrinsically linked to the biomechanical capabilities of hominins, among other factors, in a co-evolutionary manner.

Supplemental Information