Introduction

The Acheulo-Yabrudian Cultural Complex (henceforth AYCC) is a local Levantine entity, characterized by a set of significant and innovative human cultural and biological adaptations, including the constant and systematic use of fire (Blasco et al. 2016; Shahack-Gross et al. 2014; Shimelmitz et al. 2014a, b), complex strategies of procurement and exploitation of lithic materials (Boaretto et al. 2009; Verri et al. 2004, 2005; Wilson et al. 2016), intensive and systematic flint recycling (e.g. Assaf et al. 2015; Lemorini et al. 2015; Parush et al. 2015; Shimelmitz 2015; Wojtczak 2015), technological innovations such as blade and Quina scraper production (Lemorini et al. 2016; Shimelmitz et al. 2011, 2014a; Zupancich et al. 2016a, b; Zaidner and Weinstein-Evron 2016) (Weinstein-Evron & Zaidner 2017), systematic fallow deer group hunting and butchering (Barkai and Gopher 2011; Stiner et al. 2009, 2011; Blasco et al. 2016), and the sharing of meat (Stiner et al. 2009). Within this context, the presence of handaxes in the AYCC stands out as an almost isolated element of technological continuation and preservation of Acheulian lifeways, alongside the presence of spheroids (Barkai and Gopher 2016), in an otherwise completely innovative technological system. Numbers of handaxes found in Acheulo-Yabrudian contexts vary from a few isolated artefacts (as in Zuttiyeh Cave; see Gisis and Bar-Yosef 1974) to several dozen (as in Tabun Cave; see Shimelmitz et al. 2017).

In this paper, we study the small assemblage of bifaces found at the Acheulo-Yabrudian site Qesem Cave (QC; 420–200 kya) by analysing the flint types from which they were produced and by identifying the potential geologic/geographic sources from which they could have originated. We compare the results to a flint type analysis of a large general sample of artefacts (n = 21,102) taken from 12 different assemblages from within QC, which is part of a large-scale ongoing project (Wilson et al. 2016). We also compare it with a sample taken from the Late Acheulian site of Jaljulia (Israel), which is situated approximately 6 km north of QC. Our results suggest that the manufacture of the QC bifaces did not take place at the cave, but, rather, that these artefacts were brought to the site in their current state, possibly from older Acheulian sites. Finally, we discuss the significance of these results and their implications for Levantine Late Lower Palaeolithic human behaviour.

The Acheulo-Yabrudian Lithic Industries

The Acheulo-Yabrudian Cultural Complex has been subdivided into three major lithic industries:

  • The Amudian (Pre-Aurignacian)—characterized by the production of blades.

  • The Yabrudian—a flake industry characterized by the production of Quina scrapers made on thick flakes with stepped retouch (resembling the scrapers known from the European Middle Palaeolithic Mousterian), alongside the appearance of demi-Quina scrapers.

  • The Acheulo-Yabrudian—characterized by the production of flakes, bifaces and scrapers.

Rust (1950) and Garrod (1956) suggested that each of these industries represents a different culture, or a different group of people. Copeland and Hours (1983), on the other hand, viewed these industries as reflecting different activities within the same cultural complex. The latter hypothesis is supported by recent observations made at QC, where Yabrudian (scraper dominated) and Amudian (blade dominated) assemblages show spatial differentiation within the same stratigraphic units (Gopher et al. 2016). This suggests the coexistence of Amudian and Yabrudian industries at QC (and see Barkai et al. 2009; Shimelmitz 2009). The technological similarities between scraper (Quina and demi-Quina) and blade production in the Amudian and the Yabrudian industries further support this interpretation. Indeed, it seems that the differences between the industries are mostly quantitative rather than qualitative (Assaf et al. 2015; Parush et al. 2016).

Handaxes—a Brief Overview

Handaxes are, in most cases, relatively large artefacts (often 10 cm long or longer), which were bifacially knapped and shaped (Machin 2009; Sharon 2010; Wynn and Gowlett 2018). They repeatedly appear throughout the Old World, starting from 1.8 mya, and until 200,000 years ago in the Levant, and even later in Europe. Since they continuously present the same general morphology, and show the same production technology, handaxes are often viewed as the expression of a technological stagnation (e.g., Elias 2012; Renfrew and Morley 2009). It should be noted, however, that while there are some general traits appearing in all handaxes, handaxes vary widely in terms of size, shape, technological details, the type of blank selected, and the degree of regularity (Wynn and Gowlett 2018). It should also be noted that the production of handaxes was accompanied by several technological innovations, such as the development and adoption of the Levallois method (Adler et al. 2014; Nowell and White 2010), and the production of small flakes by means of systematic lithic recycling (e.g. Agam et al. 2015; Agam and Barkai 2018a; Shimelmitz 2015), implying that the Acheulian Cultural Complex was not as stagnant as previously thought. These innovations, however, were not as widely distributed in time and space as the Acheulian handaxes (Finkel and Barkai 2018).

While the nature of the function(s) for which handaxes were used is still debated (see discussion below), Wynn and Gowlett (2018) describe the form of the handaxes as being “over-determined”, meaning that their makers invested more effort in the shaping of the handaxes than was necessary for their functionality. This suggests that there were considerations other than functionality affecting the manufacture and shaping of handaxes.

Within the AYCC, handaxes are found mostly in the Acheulo-Yabrudian industry (alongside the production of flakes). They are, however, also found, in lower proportions, in Amudian contexts at AYCC sites. Handaxes have been found, for example, at Qesem Cave, Tabun Cave, Misliya Cave, Bezez Cave, and Yabrud I (Barkai et al. 2013; Gisis and Ronen 2006; McPherron 2003; Shimelmitz et al. 2017; Zaidner et al. 2006). Handaxes disappear from the archaeological record of the Levant with the emergence of the Levantine Middle Palaeolithic Mousterian, some 200,000 years ago (Falguères et al. 2016; Mercier and Valladas 2003; Valladas et al. 2013).

Qesem Cave

Qesem Cave (QC) (Fig. 1) is located approximately 12 km east of the Mediterranean coast of Tel Aviv, Israel, 90 m a.s.l., on the western foothills of the Samaria Hills. It has been excavated since 2000 and has yielded many rich lithic and faunal assemblages (Barkai and Gopher 2013; Gopher et al. 2005). Based on the composition of these assemblages (Barkai et al. 2005, 2009; Gopher et al. 2005), and supported by a long series of radiometric dates (Barkai et al. 2003; Gopher et al. 2010; Mercier et al. 2013; Falguères et al. 2016), the QC sequence has been assigned to the Acheulo-Yabrudian Cultural Complex (AYCC) of the Lower Palaeolithic of the Levant and dated to between 420,000 and 200,000 years ago. The cave’s sediments comprise two major cycles of deposition—a lower sequence (~ 5.5 m thick), deposited while the karst chamber cave was closed, and an upper sequence (~ 4.5 m thick), deposited when the cave was more open (Parush et al. 2016).

Fig. 1
figure 1

QC map, with three related Acheulian sites: Jaljulia, Eyal 23, and Revadim

Full size image

Systematic and repetitive use of fire has been detected at the site, dated to ca. 400,000 years ago (Barkai et al. 2017; Blasco et al. 2016; Karkanas et al. 2007; Stiner et al. 2009, 2011; Shahack-Gross et al. 2014). The repeated use of a central hearth, starting at least 300,000 years ago, has also been identified (Barkai et al. 2017). The hearth is located in the centre of the cave and is associated with butchering and knapping activities. Its location, as well as the activities associated with it, implies the organization of space at the cave (Barkai et al. 2017). In the north-western part of the cave, a rock “shelf” was found; layers on top of this shelf were dated to 300,000 years or older. Further excavation exposed a deep sequence of layers under the shelf, which therefore has a minimum age of 300,000 years (Gopher et al. 2010).

The cave inhabitants exploited mainly fallow deer, in addition to other taxa, and brought selected body parts to the cave, following the application of cooperative group hunting. The animals were then butchered, cooked, and shared by the Qesem hominins (Karkanas et al. 2007; Stiner et al. 2009, 2011).

Most of the lithic assemblages from QC are assigned to the Amudian industry, dominated by blades (Barkai et al. 2005, 2009; Gopher et al. 2005; Shimelmitz et al. 2011). The Yabrudian industry, present in several spatially and stratigraphically distinct areas within the cave, is dominated by Quina and demi-Quina scrapers (Barkai et al. 2009). Only a few isolated handaxes have been found at the cave (Gopher et al. 2005; Barkai et al. 2013), making the Acheulo-Yabrudian industry practically absent from the cave’s assemblages. Systematic lithic recycling has also been recognized as a significant component within the site’s assemblages, aimed mainly at the manufacture of small sharp flakes and blades from parent flakes and blades (Assaf et al. 2015; Barkai et al. 2010; Parush et al. 2015). A few spheroids have also been found, made mainly of limestone (Barkai and Gopher 2016).

Thirteen human teeth have been discovered at QC and are described as closer to the later populations (e.g. Skhul/Qafzeh) of this region, rather than to Homo erectus, although they also bear some Neanderthal traits (Fornai et al. 2016; Hershkovitz et al. 2011, 2016; Weber et al. 2016).

The Bifaces of Qesem Cave—an Overview

In this study, the term bifaces relates to all artefacts which bear bifacial knapping; the term roughout refers to artefacts which bear only preliminary bifacial knapping; the term handaxe refers to items which were fully (or almost fully) bifacially knapped, and which are considered complete products. This study includes an assemblage of 17 artefacts—16 bifaces and one bifacial spall, found in a variety of stratigraphic contexts at QC. This small assemblage stands in strong contrast to the abundance of blades and Quina and demi-Quina scrapers found at the cave.

One giant roughout of a biface (item number 13 in this study, see Tables 1 and 4) was found in an almost horizontal position, a little to the north of the central hearth, buried under a collapse of massive blocks. The context in which the giant biface was found postdates the hearth, and it is part of an Amudian assemblage covered by the collapse (Barkai et al. 2013). The deposition of the large biface was dated to between 280,000 and 250,000 years ago. This roughout did not show any use-wear, suggesting that it was never used.

Table 1 The biface assemblage of Qesem Cave, with their stratigraphic origin, the assemblage to which they are assigned, and their sub-category
Full size table

It is as yet unclear whether handaxes were produced at the site (for more on this see the discussion below). Barkai et al. (2013) suggested, based on the presence of the bifacial roughout, that biface production was indeed practiced at the site, although only rarely. However, bifacial knapping waste is completely absent from the site’s assemblages, reducing the likelihood that this procedure took place inside the cave. New results, presented below, suggest that the QC bifaces were brought to the site in their current state.

Materials and Methods

Materials

The QC biface assemblage (Table 1; Fig. 3) can be divided into four typo-technological sub-groups: handaxes (n = 12), bifacial roughouts (n = 3), a trihedral pick (n = 1), and a single bifacial spall (n = 1). These items originate from various assemblages within the site, without any vertical or horizontal clustering, excluding items number 13 and 17, which were found in proximity to each other (Fig. 2). It should also be noted that some isolated artefacts which might be related to bifacial knapping stages were also found at the cave. However, as their assignment to this technological procedure was not secure, we excluded them from this study.

Fig. 2
figure 2

Spatial distribution of the QC bifaces (and see details in Table 3). Dotted line represents the estimated original contour of the cave

Full size image

Flint types of the biface assemblage were compared with flint types of a large sample of artefacts (n = 21,102; see Table 2 and Appendix Table 6) taken from 12 different lithic assemblages from various contexts at QC. This is a part of a large-scale ongoing project analysing the QC raw materials, for which preliminary results were previously published (Wilson et al. 2016). The bifaces are not counted as part of the general sample.

Table 2 A breakdown of the analysed assemblages by technological categories, numbered from youngest to the oldest
Full size table

Methods

Each artefact was weighed and then classified, with the help of a × 10 hand lens, to a flint type, based on visual traits such as colour, texture, traits of cortex, sub-cortical layers, any distinctive patterns, degree of translucency, degree of homogeneity, and any visible fossils. This procedure follows the methods provided in Wilson (2007) and Browne and Wilson (2011). Flint types are labelled alphabetically by order of identification, from A to CS. For each flint type, we defined the degree of homogeneity (from homogenous to heterogeneous, scaled from 1 to 3); roughness (from fine to coarse, scaled from 1 to 3); and degree of translucency (from translucent to opaque, scaled from 1 to 3). In total, in the general sample, 96 different flint types were classified and grouped into 42 groups of flint types, based on visual characteristics. Each of the bifaces was also measured for metrics (i.e. length, width, and thickness).

In addition, fieldwork was undertaken in order to locate potential sources of flint, following the flint-bearing outcrops, guided by the 1:50,000 geologic maps of Hildebrand-Mittlefehldt (2011), Yechieli (2008), and Ilani (1985). Potential sources located east of QC were not surveyed during this study, due to security and logistics constraints, but are mentioned below.

Petrographic thin sections were produced for selected archaeological and geologic samples and were analysed using a ZEISS Axio Scope.A1 polarized light microscope (n = 106). Each flint type was compared with the samples collected from the geologic sources and assigned to potential flint sources, whenever possible, using both macroscopic and petrographic data. Finally, flint types were grouped into groups of sources, based on their potential geologic origins.

For the Jaljulia sample, we classified the artefacts from area D to typo-technological categories, and then, the samples from both areas B and D were further classified into flint types, based on macroscopic traits, in the same procedure as that performed for the QC material. The Jaljulia material was not assigned to potential flint sources.

Results

Potential Geologic Flint Sources Around QC

QC is located within rich flint-bearing limestone outcrops of the Bi’na Formation, of the Upper Cretaceous Turonian (Fig. 3). In total, 21 flint-bearing localities, both primary and secondary, most likely of the Bi’na formation, were located within the immediate vicinity of the cave, in a radius of up to 8 km from the site.

Fig. 3
figure 3

Qesem Cave and the potential flint bearing sources, based on geologic age and distance. Wadi Qana (the blue line) which passes 3 km north of QC

Full size image

Flint sources of the Mishash formation (Upper Cretaceous) were found in the area of Ben-Shemen forest, ~ 15 km or more south of QC. In total, 9 potential sources were found in that area. Other, more distant sources of the Mishash formation are known to exist some 25 km and more east of QC (Sneh and Shaliv 2012).

Six flint sources were located 12–13 km north of QC, on the borderline between Cenomanian and Turonian exposures, making them either of the Sakhnin or the Bi’na Formations. Two sources of the Upper Cenomanian-Turonian Eyal Formation were found in Eyal Forest, 12–13 km north of QC. Cenomanian flint of the Beit Meir Formation can be found in outcrops located 25 km or more east of QC (Sneh and Shaliv 2012).

One Eocene source of the Adulam formation (Yechieli 2008) was found near the city of Lod, ~ 16 km south of QC. It currently contains very low quantities of flint, and we cannot say whether flint would have been available there during prehistoric times. Another low-density Eocene source was sampled at Tel-Gezer, 30 km south of Qesem. More Eocene sources, of the Timrat formation, exist ~ 25 km east of the cave (Sneh and Shaliv 2012).

The abundance of the sources located to the east of QC, as well as their extent, nature, and variety, is currently unknown. They are, however, located along the current course of Wadi Qana (Fig. 3), which passes ~ 3 km north of QC at its closest point. Therefore, the wadi could potentially have carried flint nodules from their geologic sources closer to the cave and made them more likely to be exploited by the QC hominins.

Recent surveys conducted in the segment of Wadi Qana closest to QC found no conclusively non-Turonian flint. However, the site of Jaljulia provides significant data in this respect. Jaljulia is located about 6 km north of QC, and 100 m north of the closest point of Wadi Qana. At the southeastern part of the site, an ancient stream was revealed, probably related to Wadi Qana. A preliminary survey of this ancient stream revealed a wider selection of flint than in the current wadi, including the possible presence of Campanian and Cenomanian flints, suggesting that the ancient course of Wadi Qana might have transported flint from the eastern sources. In this case, these relatively distant flint types would become more likely to be used. However, more work is necessary to confirm or refute these identifications.

The Bifaces Analysis

Turonian flint was often exploited by the cave’s inhabitants, and it strongly dominates the site’s lithic assemblages (74.0% of the general sample Table 4). The biface assemblage, on the other hand, reflects a different pattern. Out of the 17 bifacial artefacts, 13 are made of non-Turonian flint types (76.5%; non-Turonian flints within the general sample: 26.0%). Six items (35.3%) are made of Campanian flint (Campanian flint in the general sample: 8.2%); six more (35.3%) are of an undetermined origin (flint of undetermined sources in the general sample: 5.1%); four (23.5%) are made of Turonian flint (Turonian flint in the general sample: 74.0%), and one (5.9%) is made of Eocene flint (Eocene flint in the general sample: 1.2%). No Upper Cenomanian-Turonian or Cenomanian/Turonian flints were detected among the bifaces.

The six artefacts which are of Campanian origin are made of type AQ (item numbers 1, 5, 7, 13, 14, and 17; Fig. 4 and Table 3). This is a brecciated flint type, composed of clasts of light brown fine-textured opaque flint in a light brown matrix. Brecciated textures are a known component of Mishash flints (Kolodny 1969). This flint type constitutes only 1.4% of the general sample and reaches a maximum of 2.5% in any of the other typo-technological categories (the highest proportion being within the cores).

Fig. 4
figure 4

The bifaces made of type AQ. (1, 2) Roughouts (item numbers 13 and 14). (3–5) Handaxes (item numbers 1, 5, and 7). (6) Bifacial spall (a tranchet spall; item number 17)

Full size image
Table 3 Flint types and metrics of the bifaces
Full size table
Table 4 Comparison of the frequency of different geologic origins in the general sample and in the biface assemblage
Full size table

The six artefacts made of type AQ include three handaxes, two roughouts, and one bifacial spall. The average weight of these six artefacts is 1047.3 g. This result is, however, strongly influenced by the presence of the two roughouts. In the general sample (which does not include bifaces), the average weight of pieces made of type AQ is 20.4 g, while the average weight of all pieces in the entire general assemblage is 10.3 g, implying that type AQ was often used for the production of relatively large blanks. Our survey of the sources has shown that type AQ tends to be found in large nodules and beds, or remnant bed fragments. This large size of nodules may have played a role in the decision to use this flint type for the production of bifaces. Indeed, it has already been suggested that size and shape of the naturally available raw materials played an important part in determining which blank would be selected for biface production (Sharon 2008). However, large nodules of flint have also been observed in Turonian sources (such as Horashim Forest, located 5 km north of QC), and handaxes were in fact also manufactured of Turonian flint, as demonstrated below. Moreover, the existence of relatively small handaxes in many sites implies that size and shape were not necessarily significant factors in the decision as to what lithic materials are going to be used for the production of bifaces (Sharon 2008).

Five of the six bifaces of undetermined source (29.4% of the bifaces) are made of type T (item numbers 2, 3, 4, 10, and 15), which is a dark grey-brown and light brown roughly zoned heterogeneous medium-to-coarse-textured opaque flint type, with macroscopically visible sponge spicules (Fig. 5). Type T constitutes 1.5% of the general sample and reaches a maximum proportion of 2.2% of the cores and of the core-trimming elements (CTEs).

Fig. 5
figure 5

The bifaces made of type T. (1) A roughout (item number 15). (2–5) Handaxes (item numbers 2, 3, 4, and 10, respectively)

Full size image

Item number 6 is made of type AU (5.9%), which is also a flint type from an unknown source (Fig. 6). It is a rich chocolate brown homogenous fine-textured opaque flint, with a faintly striped appearance. Type AU constitutes 2.3% of the general sample. Its highest proportion within the other categories is 3.3% of the tools and of the special spalls.

Fig. 6
figure 6

Group 1: the homogenous fine-textured handaxes, four of which are made of Turonian flint (1–4), and one of a flint of an undetermined source (5). (1) Item number 11. (2) Item number 9. (3) Item number 12. (4) Item number 8. (5) Item number 6

Full size image

Four artefacts are made of Turonian flint types. Two handaxes (11.8% of the biface assemblage) are made of type W (item numbers 9 and 11; versus 2.8% of the general sample), a fine-textured brown roughly striped flint type. Its highest proportion in any other technological categories is 5.1%, in the special spalls.

One handaxe (5.9%) is made of type M (item number 8), a Turonian flint type which is distinctly striped, in beige, grey, and pink (4.8% of the general sample). Type M is a common flint type within the general sample, and its proportions within the other categories range between 1.9% (of the bladelets) and 9.9% (of the naturally backed knives).

The last biface made of a Turonian flint is a handaxe made of type AI (item number 12; 5.9%; versus 1.7% of the general sample). It is a greenish-brown fine-textured flint. The highest proportion of type AI in any of the other technological categories is 3.5% (of the recycled artefacts).

One artefact—the trihedral pick—is made of type BJ, a coarse-textured Eocene flint. It is presented in detail and discussed further below.

The Roughouts

Two of the three bifacial roughouts found at QC (item numbers 13 and 14) are made of type AQ (Campanian) and one (item number 15; Fig. 5a) of type T (from an unknown source). The heaviest roughout (item number 13) is significantly heavier than the two other roughouts. The heaviest roughout is also the longest, the widest, and the thickest (for more details on this item, see Barkai et al. 2013).

Item numbers 13 and 15 were produced from large nodules, and a significant proportion of both is still covered in cortex, on both faces. Item number 14, on the other hand, was produced on a large flake, with its ventral face and bulb of percussion still clearly preserved. Most of its dorsal face is still covered in cortex. It has several surfaces bearing patina, and some post-patina flaking, mainly on its ventral face, suggesting it was recycled for the production of flakes, or for further processing as a biface. Item numbers 14 and 15 (as well as item number 1) were analysed for 10Be (Beryllium-10) content, which is related to the measurement of cosmic radiation (Boaretto et al. 2009). They presented low levels of 10Be, suggesting that they were procured from primary geologic sources, possibly involving quarrying (for more details see Verri et al. 2004, 2005; Boaretto et al. 2009).

The Handaxes

Six of the twelve handaxes (50%; item numbers 3, 4, 6, 8, 11, and 12) were produced on nodules. Three handaxes were produced on flakes (25%; item numbers 5, 9, and 10). For the remaining three handaxes (25%; item numbers 1, 2, and 7), the blank could not be determined. Four of the handaxes (item numbers 1, 2, 3, and 11) are covered in patina, while one of them (item number 11) bears post-patina removals, indicating it was recycled after being produced.

Three handaxes (item numbers 5, 10, and 11) bear removals of large flakes from their circumference, removals which were most likely not related to their bifacial shaping. These removals probably reflect the recycling of these handaxes into cores, taking advantage of the handaxe convexities. The phenomenon of handaxes with preferential flake scars has been suggested by some scholars to reflect a possible link between Acheulian handaxes and the emergence of the Levallois method (see DeBono and Goren-Inbar 2001; Shimelmitz 2015; White et al. 2011).

Seven of the 12 handaxes (58.3%; item numbers 1, 2, 3, 4, 5, 7, and 10) are made of heterogeneous flint types—three of type AQ and four of type T. Five others (41.7%; item numbers 6, 8, 9, 11, and 12) are made of homogenous flint types—two of type W, one of type AU, one of type M, and one of type AI. This pattern implies that the degree of homogeneity did not play a role in the decision about what flint types to use for the production of these handaxes.

There is however a clear correlation between the degree of homogeneity, the texture, and the size of the handaxes. First, the five homogenous handaxes are also fine-textured, while the seven heterogeneous flint types are coarse-textured. Second, the average weight of the homogenous fine-textured handaxes is 172.8 g, while the average weight of the heterogeneous coarse-textured handaxes is 348.7 g. These results suggest that the handaxes can be divided into two groups: one (henceforth, group 1; Fig. 6) consists of handaxes made of homogenous, fine-textured flint types, which tend to be smaller, and the second (henceforth, group 2) consists of handaxes made of heterogeneous, coarse-textured flint types, which tend to be larger. Four of the five handaxes of group 1 were produced on nodules, while only two handaxes of group 2 were clearly produced on nodules, while two others were produced on flakes, and for the remaining three, the blank could not be determined. Moreover, three handaxes of group 2 are covered in patina, while only one handaxe of group 1 is covered in patina. Finally, four of the five handaxes of group 1 are made of Turonian flint types (with the fifth being from an undetermined source), while three of the seven handaxes of group 2 are made of Campanian flint, and four of flints of unknown origin. These differences (summarized in Table 5) suggest that there may have been two separate procedures for acquiring flint for each of these two groups, although the nature of these two procedures is still unclear. It is not our contention that each group represents a single distinctive accumulation process. Rather, we merely suggest that there were at least two different strategies of bringing handaxes to the cave.

Table 5 Summary of differences between the two groups of handaxes
Full size table

One handaxe is not included in this study because its whereabouts are, unfortunately, currently unknown, but it does deserve some special attention (Fig. 7). This is a patinated handaxe, produced on a homogenous flint type (R. Barkai, personal observation), which bears several post-patina blade removals, indicating that it was recycled into a blade core (Parush et al. 2015; Shimelmitz 2009). We can therefore see the Acheulian hallmark, the handaxe, and one of the main Acheulo-Yabrudian hallmarks, the systematic production of blades, on one artefact, with a clear time gap between these two stages.

Fig. 7
figure 7

A handaxe recycled into a blade core

Full size image

The Bifacial Spall

One bifacial spall has been found in the QC assemblages (item number 17). It was found in the same sub-square (1/4 m2) as part of the same assemblage as the largest roughout (item number 13), and it is made of the same flint type as the large roughout—type AQ, which is of Campanian origin. However, it was not flaked from the roughout.

This artefact is a product of a transversal blow, using the tranchet blow technique (Inizan et al. 1992: 72). Such spalls are known from several Lower Palaeolithic sites (e.g. Bergman and Roberts 1988; Roberts and Parfitt 1999; Rollefson 2016; Sharon 2010; and for more information, see Barkai 2005). These blows were aimed at shaping the active edge of handaxes, and at creating a very sharp edge. It has two ventral faces, indicating that the original biface was most likely produced on a large flake. The spall could not be directly associated with any of the bifaces found at the cave. Moreover, no biface from QC bears scars of the tranchet blow technique. Its presence does, however, imply that at least one additional biface from which this artefact was flaked exists or formerly existed at the cave, or, alternatively, that this artefact was brought from outside the cave in its current state.

The Trihedral Pick

The trihedral pick is made of type BJ, which is a semi-translucent grainy light brown flint with abundant small macroscopically visible white fossils (Fig. 8). Its thin section revealed some nummulitic debris (Fig. 9), conclusively assigning it to the uppermost Lower-to-Middle Eocene (Racey 2001). Additionally, an echinoid spine similar to ones detected in other Eocene samples was also observed (Fig. 9). Type BJ is found in low proportions in the general sample (0.3%). The trihedral pick bears some patinated surfaces. No waste related to its manufacture was found at the cave.

Fig. 8
figure 8

The trihedral pick, made of type BJ, of Eocene origin

Full size image
Fig. 9
figure 9

Type BJ—nummulitic debris (a) and an echinoid spine (b), in plane- and cross-polarized light

Full size image

Trihedral picks are well-known from several Levantine Acheulian sites (Gilead 1970; Shea and Bar-Yosef 1999; Tchernov et al. 1994), including Eyal 23, an Acheulian site located ~ 12 km north of QC (Ronen and Winter 1997). These tools are, however, usually absent from Acheulo-Yabrudian contexts. Besides Jaljulia (Shemer et al. 2018) and Eyal 23 (Ronen and Winter 1997), other Acheulian contexts could also have existed in the area of QC. It is, therefore, possible that this trihedral pick was collected from outside the cave, possibly from an older Acheulian site located somewhere in the vicinity of QC, rather than being produced in it.

Discussion

The Role of Handaxes in Lower Palaeolithic Lifeways

While many studies have tried to understand the functionality of handaxes, the nature of their use is still strongly debated. Past studies have proposed that handaxes were used during the Lower Palaeolithic for the processing of vegetal materials (e.g. Binneman and Beaumont 1992), the processing of wood (e.g. Domínguez-Rodrigo et al. 2001; Kohn and Mithen 1999), in butchering activities (e.g. Keeley 1977, 1980; Machin et al. 2007; Mitchell 1996; Solodenko et al. 2015), and even as hunting hurled/thrown weapons (e.g. Calvin 1993; O’Brien 1981, but see Whittaker and McCall 2001). Handaxes are also often referred to as general-purpose tools (e.g. Keeley 1980). Other, less common, proposals have suggested that handaxes should be viewed as cores, intended for the efficient production of flakes (e.g. Jelinek 1977). In addition, other, non-utilitarian, potential roles of handaxes are also often discussed (Gamble 1998; Kohn and Mithen 1999; Wynn and Gowlett 2018).

Most recent studies, however, suggest a relationship between handaxes and the processing of meat, including the skinning, dismembering, defleshing, and cutting of animal carcasses (e.g. Claud 2008; Machin et al. 2007, 2016; Solodenko et al. 2015). Of special note is the relationship between the presence of handaxes and the presence of the remains of very large game, mainly proboscideans, during the Lower Palaeolithic (Finkel and Barkai 2018). Indeed, several Acheulian sites have yielded elephant remains bearing cut marks (e.g. Blasco et al. 2013; Solodenko et al. 2015), as well as elephant bones which were found in direct association with bifacial tools (e.g. Goren-Inbar et al. 1994; Zutovski and Barkai 2016, and see additional references therein). The important role of elephants in the diet and adaptation of Acheulian populations has already been suggested in the past (Agam and Barkai 2016, 2018b; Ben-Dor et al. 2011), and is further supported by many Acheulian sites containing elephant remains (e.g. Anzidei et al. 2011; Goren-Inbar et al. 1994; Rabinovich et al. 2012).

Finkel and Barkai (2018) propose that handaxes were a useful tool in the processing of elephant carcasses, allowing the removal of meat and fat, as well as the disarticulation of elephant body parts in order to enable their transportation to habitation sites. Handaxes are highly suitable for massive and continuous butchering activities, enabling the application of force and leverage required in cutting and dismembering activities. Evidence of transportation of proboscidean-selected body parts is provided by Palaeolithic cave sites containing elephant remains, and especially elephant heads (Agam and Barkai 2016 and see references therein). Thus, Finkel and Barkai suggest that handaxes were an essential tool in large game processing during the Acheulian. The appearance of bifacial tools made of elephant bones further implies that elephants had major nutritional and social roles in the lives of these hominin groups (Zutovski and Barkai 2016). Additional support for the connection between handaxes and elephants is provided by the geographical and chronological synchronization between these two elements (Finkel and Barkai 2018). This set of evidences is used by Finkel and Barkai (2018) to propose that when elephants cease to be a part of early human diet, the manufacture and use of handaxes stop as well.

For our case, the above proposed scenario implies that with the demise of elephants from the Levant at the end of the Acheulian, and with the emergence of the Acheulo-Yabrudian, handaxes lose their role as essential functional and social tools. Therefore, we consider a “non-functional” explanation for the presence of these few bifaces within the cave’s assemblages.

Explaining the Presence of Bifaces at QC

While Turonian flint dominates the QC assemblages, non-Turonian flint types are predominant in the QC biface assemblage. The presence of three roughouts and 12 complete handaxes, alongside the complete absence of bifacial knapping by-products, as well as the absence of a clear spatial distribution pattern of the bifaces throughout the site’s sequence, stresses the fragmentation of the bifacial chaîne opératoire and suggests that the bifaces were not produced at the site but, rather, were brought to the cave in their current state.

Some other Acheulo-Yabrudian sites present similar patterns. In Tabun Cave Layer E, for example, by-products of biface production are also rare (Shimelmitz et al. 2017), leading the authors to suggest that the AYCC handaxes of Tabun Cave were produced outside the site. In the case of Yabrud I, Rust (1950) suggested that bifaces were not manufactured in the AYCC level in which they were found but, rather, were retrieved from older, biface-rich layers.

The two different groups identified within the QC handaxes (groups 1 and 2) may reflect two different types of life histories. The different circumstances behind the formation of each group are, however, yet unclear. It should be noted that similarly to group 1, homogenous flint types strongly dominate the general sample as well (62.6%), suggesting that artefacts from group 1 might be more related to the general pattern detected at QC than artefacts from group 2.

In any case, the extremely low quantity of bifaces at QC, compared with the rich lithic assemblages, suggests that handaxes did not play a major functional role in the QC hominins’ everyday lives. It is therefore possible that the QC bifaces originated from older contexts, most likely Acheulian sites in the vicinity of the cave, such as the Acheulian sites of Jaljulia and Eyal 23 (Ronen and Winter 1997; Shemer et al. 2018). Both sites have yielded bifaces, and Eyal 23 has also yielded trihedral picks (the lithic analysis of the Jaljulia material is still ongoing). Interestingly, handaxes from Jaljulia are dominated by brecciated flint types which resemble flint type AQ (A. Agam, personal observation), further supporting such a scenario. However, it is premature to suggest a direct relationship between the bifaces from QC and those from Jaljulia. Additional petrographic thin sections and geologic surveys are needed in order to test such a relationship. Furthermore, as Wadi Qana could have served as a source for flint for the QC inhabitants, it is plausible that the QC hominins explored this area and were familiar with features throughout it, older sites reflecting older human occupations included. It is of note that brecciated flint types also dominate the handaxes from the Late Acheulian site of Revadim (Israel) (A. Agam, personal observation), suggesting a general preference for brecciated flint types in the production of Lower Palaeolithic bifaces.

It is possible that the QC bifaces were brought to the site due to economic motivations. Prepared bifaces could have been brought in as tools already suitable for use in various tasks, including the processing of meat and wood, or as blanks suitable for further knapping. Such a suggestion was made by Gravina and Discamps (2015) concerning the recycled bifaces found at the late Middle Palaeolithic site Le Moustier (Southwestern France). However, given the low number of bifaces found at Qesem Cave, and as they most likely were not produced at the cave, it is our view that this is not the case. Furthermore, most of the QC bifaces were not knapped after their original production, reducing the likelihood of their collection as blanks for future knapping.

Some scholars argue that Acheulo-Yabrudian handaxes tend to be smaller in size than Acheulian handaxes, less refined, and that they present novel reduction sequences compared with those of the Acheulian (e.g. Jelinek 1975; Matskevich et al. 2002; Zaidner et al. 2006). While these suggestions are not generally agreed upon, and do not reflect our own view, the QC biface assemblage does not present any clear pattern in terms of shape and size, as it includes both small and large handaxes, in various shapes and different degrees of fineness. Therefore, it seems that the QC bifaces do not accord with the typology occasionally associated with Acheulo-Yabrudian handaxes, further supporting a scenario of an older origin.

The habit of prehistoric societies to collect older previously knapped artefacts is well documented in archaeological sites (e.g. Agam and Barkai 2018a; Hiscock 2015; Gravina and Discamps 2015; Vaquero et al. 2015; Whyte 2014). Similar patterns of behaviour have also been observed among recent hunter-gatherers. The aborigines of the Western Desert in Australia, for example, were documented collecting and re-fashioning prehistoric tools, while being fully aware of their old lives as tools produced by past societies (Gould 1980: 134).

At QC, the inhabitants of the cave often collected old artefacts covered by heavy patina and brought these previously knapped items to the cave (Barkai and Gopher 2016). Efrati et al. (2018) argued that 12% of all analysed assemblages at QC in general are made on patinated previously knapped artefacts, which were most likely collected as knapped artefacts from outside the cave, as indicated by the presence of patina and of post-patina removals.

Caricola et al. (2018) analysed the spheroids assemblage from QC, using both technological analysis and use-wear and residue analyses, and showed that at least some of these spheroids are covered by pre-use patina, suggesting that they were collected from outside the cave as knapped objects. Also, some side scrapers found at QC were produced from old patinated flakes, with post-patination scalar retouch reflecting the existence of a time gap between the two stages of manufacture (Parush et al. 2015). Another example of the collection of old knapped blanks from outside the cave is the trajectory aimed at the production of small blanks by means of lithic recycling from parent flakes or blades (Assaf et al. 2015; Parush et al. 2015). This trajectory includes the exploitation of both fresh blanks, without patina, and patinated blanks, with post-patina removals of flakes (for more details see Parush et al. 2015). It was suggested that the patinated blanks were collected from outside the cave, rather than being originally produced in it (Parush et al. 2015).

Finally, some of the handaxes found at QC also show evidence of a second use cycle. As mentioned above, one heavily patinated handaxe was recycled into a blade core after being covered in patina (Parush et al. 2015; Shimelmitz 2009). Although most of the bifaces were not used for the production of flakes after their original manufacture, a few were.

Given the data presented above, we suggest that the collecting of old knapped artefacts from outside the cave was a repetitive pattern of behaviour at QC. The QC hominins were most likely highly familiar with the surroundings of the cave. They must have explored the land in search of various resources, such as food, rocks for tool production, and wood for fire, and were well aware of the different features and localities around them. These included, most likely, old, yet-uncovered hominin sites. The knapped lithic artefacts spread on the ground in these sites, as well as the likely presence of fragmented animal bones, could have led the QC hominins to realize and conceptualize past human presence at these localities. Early humans had an intimate relationship with the lithic materials surrounding them, and stone tools played an important role in these early humans’ lives (Berleant 2007). Moreover, the fact that the lithic artefacts spread on the ground were part of and had meaning in the lives of earlier human groups could have enhanced the sensory effect they had over the later human groups seeing them (Berleant 2007). Therefore, and given the tendency of the QC hominins to collect old knapped artefacts (Parush et al. 2015), these encounters might have inspired them to collect some of these old knapped artefacts (Berleant 2007). Within this context, bifaces were more likely than other artefacts to raise interest, given their large size, high visibility, and high aesthetic value (Hodgson 2015; Mithen 2003; see Wynn and Gowlett 2018 for additional details). It has already been suggested that the QC hominins brought artefacts to the cave due to their aesthetic characteristics (Assaf 2019), so the possible collection of bifaces for similar reasons fits within the same behavioural model.

Conclusions

The exploitation of previously produced flint artefacts by means of lithic recycling was often practiced by the QC hominins as a regular technological trajectory (Parush et al. 2015). The QC biface assemblage, on the other hand, reflects, in our view, a different pattern of behaviour, possibly more related to the awareness and appreciation of the antiquity of these old knapped artefacts, rather than to their economic/technological value.

The QC bifaces were not produced at the site but, rather, were brought to the cave in their current condition. The very small number of bifaces in the outstandingly large lithic assemblages (many tens and hundreds of thousands of other lithic artefacts) found at the cave, and especially compared with the thousands of blades and many hundreds of scrapers found within the cave’s assemblages, implies that handaxes did not play a major functional role in the everyday lives of the QC hominins. While some technological trajectories detected at QC also involved, to a certain degree, the exploitation of old knapped artefacts (e.g. the production of Quina scrapers on old patinated blanks, the production of small flakes from old parent flakes), the scope of these trajectories was far more extensive than that implied by the small biface assemblage. Moreover, the complete absence of bifacial knapping waste at the site demonstrates that bifacial knapping was rarely performed at the site, if at all. The relationship between the Levantine Acheulian handaxes and proboscideans, discussed above, provides a possible explanation for the decay in the everyday use of handaxes. We, therefore, suggest that bifaces were not brought to the cave for their utilitarian qualities, but possibly as a result of the awareness and appreciation of their antiquity, as well as an understanding of their long life history. Future use-wear analysis will further explore this hypothesis.

Berleant (2007) wrote that “there is [a] sensory frisson that comes from handling a tool that we know some unknown person, twenty thousand or two hundred thousand years ago, made and used”. As archaeologists, we often experience that special tremor, picking up an old knapped artefact from the ground, holding it to the light, admiring the knapping skill of its creator, as well as the lithic material chosen for its production. It is, then, possible to assume that prehistoric people, professional knappers in their own right, most likely with their own sensory appreciation, felt that very same way.