Breaking the Mold of Natufian Basalt Mortars: Experimental Production and Archeological Implications

This paper presents an experimental study on the production of basalt mortars, common at Levantine Natufian sites (15,000 – 11,500 cal. BP). An increase in the gathering and processing of plant foods is among the Natufian cultural innovations preceding the Neolithic agricultural revolution, as documented in B agricultural ^ tools such as sickle blades and ground stone tools (GST). Studies of Natufian GST have focused on their typology and function; production methods are rarely discussed. We present 15 systematic experiments focusing on two methods — drilling and carving — and using a variety of techniques (bow/twist/pump/hand drilling, pecking, and battering) and tools. Our results demonstrate that drilling techniques are efficient only in the initial stages of depression preparation and only for a limited duration. Pecking and battering with a hand-held basalt chisel was found to be the preferable production method, allowing for longer, continuous working time, and producing deeper and larger depressions than other techniques. In addition, pecking and battering makes it possible to produce a complete mortar — from the initial depression to modification of the interior and exterior faces. We suggest that producing a Natufian basalt mortar takes only 5 h and produces specific waste products, and that such repetitive work causes severe pain, particularly to the elbow and palm of the working hand, which may be associated with the variety of physical deformations reported for male Natufian and Neolithic populations.

Introduction that accidental breakage may well occur during production. Second is the issue of craft specialization (e.g., Childe 1950;Costin 1991;Clark 1995;Flad and Hruby 2007). It has been suggested that the production of Natufian mortars is a very time-consuming process that may even require the work of craft specialists-highly trained professionals-who had to be supported by the community during the long duration of mortar production (Bar-Yosef: pers. comm.). If so, Natufian groups should be considered communities of increasing social complexity, practicing social stratification.
Other possible implications relate to our understanding of Natufian basalt assemblages. Experimental production of basalt mortars can provide information on possible work tools (e.g., chisels or drills) and waste products (e.g., basalt flakes) that can be traced within archeological assemblages. The presence of such tools and/or waste may contribute to our understanding of the production processes of Natufian GST as well as their place of production.
We suggest that experimental studies can serve as a suitable means of gaining insight into the production time, skill needed for production, and breakage of basalt mortars. The experiments described in this paper examine a variety of methods for the production of basalt mortars similar in shape and size to the medium-sized mortars found at Natufian sites ( Fig. 1).
In our attempts to produce Natufian-like mortars, we modified the mortars' depressions to a depth similar to that observed on medium-sized Natufian mortars (ca. 7-12 cm.). Previous experimental studies suggest that after working with basalt GST against a variety of materials, the surface topography scarcely change (Savage 2014). Yet, there is evidence of Natufian mortars that their depression reached and pierced the base of the mortar (e.g., Rosenberg and Nadel 2014). Clearly, we cannot determine how much of the depth of Natufian mortars is the result of its primary modification and how much represents use-wear. This issue should be studied more thoroughly in future studies.

Materials and Methods
According to Schiffer and Skibo (1987), Brecipes for action^are a component of technological knowledge: BRecipes for action are the rules that underlie the processing of raw materials into finished products … A recipe for action consists of (1) a list of raw Fig. 1 Archeological and experimental mortars. a Natufian Nahal Oren, limestone; height 16 cm; diameter 17 cm. b Experimental mortar no. 2, basalt; height 21 cm; diameter 25 cm. Scale bar = 10 cm materials, (2) a list of tools and facilities employed, (3) a description of the sequence of specific actions undertaken in the technological process, and (4) the contingent rules used to solve problems that may arise … A recipe for action, then, attempts to summarize the knowledge which, if possessed by the artisan (or artisans), would account for the technological behavior^ (Schiffer and Skibo 1987: 597).
In designing the experiments, we follow that concept and suggest a recipe for action that includes several variables: raw material for mortar production (i.e., different types of basalt), tools for modification of mortars (e.g., drills or chisels), and a list of questions that may be answered following the experiments. Below, we describe these variables, starting with raw material and tools. A description of these is necessary for understanding the third variable: our list of questions, which constitute the experimental design and are presented at the end of this section (Table 1).

Raw Material Selection
A total of 15 systematic experiments were carried out on 6 different basalt boulders. The selection of basalt boulders for the experiments followed careful observations of an archeological assemblage of Natufian GST. The assemblage, originating in Hayonim Cave (Belfer-Cohen 1970), included mortars and pestles that were studied in order to evaluate the characteristics of basalt raw material. In addition, numerous informal experimental attempts were carried out prior to the experiments described below in order to establish a set of criteria for workable basalt raw material that was likely preferred by Natufian people for the production of GST. These criteria include texture (fine-vs. coarse-grained basalt), size, density of vesicles, and durability (i.e., the presence of secondary minerals). Thus, based on observations of archeological Natufian mortars and on previous informal experimental attempts, we suggest that the most suitable raw materials for mortar production would be a fine-grained basalt with small, relatively dense vesicles (up to 0.5 cm in diameter and less than 0.5 cm or so between vesicles), preferably with minimal presence of durable secondary minerals within them (e.g., basalt A in Fig. 2). In addition, as recorded in Natufian GST assemblages, it is possible to work with coarser-grained basalts if they exhibit negligible vesicles and no secondary minerals (e.g., basalt B in Fig. 2). These criteria result from the particular function of mortars, which requires a raw material that is workable on the one hand, i.e., with a certain number of vesicles, but is also durable so as to allow intensive future use, i.e., fine-grained. In addition, it should be less durable than the tool that will be used to modify it, i.e., the hammerstone, and the tools that will eventually be used to work against it, i.e., the pestle. The experimental design took into consideration these two features of the basalt raw material to which the different techniques were applied (Table 1).
We suggest that the selected basalt nodules, whether fine or coarse, would be comparable in size and shape to the desired mortars (or slightly larger). Such nodules require less investment in the exterior design of the mortar, and their inner structure enables the craftsman to follow an Bonion-like^configuration that dictates the work and assists in the modification of the exterior part of the mortar.

Tools and Techniques
Previous studies provided the terminology of GST analysis and contributed greatly to our understanding of the variability, techniques for modification, and use of GST (e.g., Wright 1992Wright , 1994Adams 2002;Eitam 2009;Rosenberg and Gopher 2010). Our experiments focused on two methods-drilling and carving-and used a variety of techniques and tools, some of them specially designed (Fig. 3).
The assumption that drilling may have been used for GST production is based on the emergence of this method during the Levantine Epipalaeolithic (e.g., Nadel et al. 1991;Grosman and Goren-Inbar 2007). Three drilling techniques were examined, based on both archeological and ethnographic data (e.g., Martin 1934;Stocks 2013;Ayad 2014): the pump drill (composed of a 3-kg basalt weight, 0.3-cm-thick plastic strings, a 145cm-long straight wooden stick, a wooden plank, and a flint drill), the bow drill (composed of a 71 cm olive branch bow, 0.3 cm plastic string, a 145-cm straight wooden stick, a flint drill, and a small wooden hand piece with a socket), and the twistreverse-twist drill (TRTD) (e.g., Stocks 2013) composed of a 6-kg stone weight and a 108-cm curved wooden branch that was naturally split). However, unlike the technique suggested by Stocks (2013), here only a single, clockwise direction was employed, i.e., twist but not reverse, and only a single weight was used for gaining leverage for the twisting. We use the term twist-reverse-twist drill as we followed Stock's methods in the construction of the drilling equipment; however, it seems that our final tool operates differently in terms of the drilling directions and the purpose served by the weight.
In addition, a simple hand drill technique was tested, in which the drill is held and twisted by hand. Different flint drills were manufactured and used: For the pump and bow techniques, the drills were triangular flint flakes approximately 0.5-1 cm thick and 4 cm long, inserted into a slot carved at the edge of the wooden stick and then wrapped with leather straps. For the twist-reverse-twist drill, a larger drill was used (a triangular flake approximately 3-4 cm thick, 4-5 cm long, and bifacially retouched) and was forcefully inserted into the spot where the stick naturally branches out. All drilling experiments focused on the modification of the interior part (i.e., depression) of the mortars.
The second method, carving, refers to percussion of a mobile tool against both the interior and exterior of the mortar preform. Carving includes battering and pecking techniques that differ in the kind of tools used and the intensity and strength of percussion. Battering is more powerful and may produce micro flakes, while pecking is gentler and more regular in spacing and produces mainly small particles. The exterior modification of the mortars involved both battering and pecking, using three tools: a hafted flint axe (bifacial flint axe, 12-13 cm long and 6-7 cm thick, with a shaft 30 cm long and 7.5 cm thick), a basalt chisel (16-20 cm long and 5-7 cm thick), and a hafted basalt chisel (Fig. 3). The interior modification involved four tools: a hafted flint axe, a basalt chisel, and two basalt chisels, hafted in different directions (Fig. 3). While pecking was carried out using all four tool types, for the powerful battering, only the basalt chisels were used (Fig. 3). The decision to use both basalt and flint tools for carving was guided by a desire to explore the advantages and disadvantages of each raw material. Like the drilling method, bifacial axes make their appearance during the Levantine Epipalaeolithic, and their occurrence in Natufian assemblages may indicate their use in the modification of GST.

Experimental Design
After having established the criteria for raw material selection and described the techniques and tools constructed for the experiments, we can present our experimental design. Our main research question has to do with evaluating which method and/or technique is preferable for the manufacture of Natufian-like basalt mortars. Various factors may influence the suitability of different techniques, however, such as the type of basalt used (finer-vs. coarser-grained) and the tools used to modify it (flint vs. basalt). Thus, the experiments were structured to examine different techniques, using different tools, and on two different types of basalt (Table 1). Fig. 2 Schematic illustration of preferable basalt characteristics for Natufian GST production. Vesicularity and secondary minerals are shown from minimal (low) to prominent (high) presence; grain size is shown from fine (low) to coarse (high); and boulder size is shown from small (low) to large (high) Fig. 3 Methods, techniques, and tools used in the experimental production of Natufian GST. a Detailed descriptions of the apparatus for the various tools are given in the text. b Scale bar = 30 cm. c WT-M: work tool to mortar

Results
In order to determine the suitability of techniques for mortar production, a total of 15 systematic experiments were carried out on 6 different basalt boulders ( Table 2). The results are presented for each method: drilling and carving.

Drilling Experiments
Prior to the current study, informal experiments with drilling techniques on a variety of basalts suggested that drilling is feasible on dense basalts with negligible vesicles (type B basalt; Fig. 2). In contrast, when drilling in large vesicles, the drill bit becomes lodged and easily breaks (type A basalt; Fig. 2). Thus, all four drilling experiments (Table 2: experiments 1-4) focused on a single basalt type that has the former properties (basalt B in Fig. 2), and the different drilling techniques were attempted sequentially. Our work demonstrated that the pump drill is the most advantageous for producing a primary depression, as it provides the stability and speed required in the initial stages of drilling. A drilling session of 1.5 h resulted in a depression 1 cm deep with a diameter of 1 cm ( Table 2: experiment 1). However, as drilling continues and the depression expands, the energy inflicted on the drill by the depression walls increases. Eventually, this resistance leads to the termination of spinning and to the breakage of the drill tip. Thus, in the course of the 1.5-h drilling session, three drill heads were used (Table 2: experiment 1). While the pump drill is advantageous for its stability and speed, which are predetermined by the mechanics of the pump weight and structure, it does not allow for control of the drill direction or manipulation of the speed in between pumping. Once an initial depression is formed, the bow drill controls these variables better and was found to be preferable for continued drilling and for deepening the depression.
The bow drill is an excellent tool for drilling, but like the pump drill, it requires constant replacement of the drill tip. After 1 min of drilling, abrasion of the flint drill starts to show and the effectiveness of the drill decreases. Thus, regardless of the efficiency of the bow drill technique (Table 2: experiment 2), maintenance of the drill tips makes systematic, continual work difficult. Our experiments demonstrate that an hour of drilling requires an hour of maintenance, so that in a 20-h drilling session, 10 h was dedicated to the preparation and hafting of drills. Like the bow drill, the hand drill (Table 2: experiment 4) provides relative control over the drill speed and direction, but due to the abrasion inflicted on the hands while spinning it is nearly impossible to use for long bare-handed.
The fourth drill type examined was the twist-reverse-twist drill (Table 2: experiment 3), which was used alternately with the bow drill. This technique is effective only once an initial depression has been formed. It was found to be efficient as it is the only drilling technique that makes it possible to use larger drills and to enlarge the diameter of the depression. Thus, when combined with other drilling techniques, the twist-reverse-twist drill is an efficient means of creating wide and deep depressions. Clearly, larger drills were found to be more durable and to require less maintenance. The twist drill is easy to manipulate and was found to be efficient for producing larger depressions.
To sum up, it is difficult to determine a favorable drilling technique, as each of the experimental methods was found to be efficient in different stages of shaping the depression: the pump drill for forming the initial depression, the bow drill and hand drill for deepening the depression, and the twist drill for enlarging it. Thus, if only drilling is used, it seems that a combination of several drilling techniques is essential in order to successfully produce a basalt mortar.

Carving Experiments
As mentioned above, carving involves pecking and battering, and we experimented with different tools (hafted vs. hand-held) made of different raw materials (flint vs. basalt). In addition, unlike the drilling techniques, carving was used for modification of the interior depression of the mortar as well as its exterior design, and we tested it on the two different types of basalt raw material.
Two experiments were carried out to evaluate carving with flint tools, both on a single basalt boulder (mortar 2, Figs. 1 and 2). In both experiments (Table 2: experiments 5-6), the flint tools were hafted to allow for a comfortable grip and only pecking was used (Fig. 3). Flint was found to be efficient for working on the exterior of the mortar, either with a hafted flake or a hafted axe. Pecking on the exterior requires controlled blows that are not very forceful in order to carefully maintain the desired shape. However, when carving the interior of the mortar, stronger blows are essential in order to deepen the depression and the flint tends to break frequently, making the tools inefficient. Thus, it was possible to work on the exterior of the mortar for 2 h using the same hafted flint flake, but working on the interior resulted in almost immediate breakage of the flint flake. Although the hafted flint axe was also found to be more efficient for modifying the exterior part of the mortar, it seems to be more durable than the flint flake, as it made it possible to deepen the inner depression an additional 2 cm in a similar amount of time (Table 2: experiments 5-6), and the axe was broken only after 1 h.
Due to the frequent breakage of the flint tools, we experimented extensively on basalt tools for modifying basalt mortars through carving. These experiments ( Table 2: experiments 7-15) examined the benefits of hafted vs. hand-held basalt chisels and were carried out on six different basalt boulders. Boulders 1 and 2, partially modified in previous experiments (Table 2: experiments 1-6), were used to examine the carving method with a hafted basalt chisel and a hand-held chisel ( Table 2: experiments 7-8, 13-15). Two hafting modes were tested: a longitudinal (pestle-like) one and a transverse (axe-like) one (Fig. 3). In addition, six experiments were carried out to evaluate the use of a hand-held basalt chisel for carving, using six different basalt boulders (Table 2: experiments 7-12). The basalt in these boulders resembles basalt types A (boulders 2, 3, 5, 6) and B (boulders 1, 4). Several observations are noted regarding the use of hafted basalt chisels: First, the results clearly suggest that a pestle-like hafting method is preferable and requires less energy. Furthermore, when using an axe-like chisel, the depth of the depression is limited to the length of the working edge of the chisel, so a complete modification of the mortar is not feasible ( Table 2: experiments 14-15). Another observation concerns the properties of the basalt raw material: working with a pestle-like hafted basalt chisel made of a finer-grained, vesicular basalt (type A) requires less time (Table 2: experiment 14).
The final modification technique tested was the use of a hand-held basalt chisel, which was examined on six different basalt boulders, two of which had been partially modified in earlier experiments. The results of these experiments demonstrate the superiority of this technique over all other methods examined. Table 2 clearly illustrates that this technique allows for longer, successive working time and produces deeper mortar depressions that can even exceed 10 cm. Furthermore, the diameters of these  Boulders 1 and 4 are of basalt type B (Fig. 2); boulders 2, 3, 5, and 6 are of basalt type A (Fig. 2) c Volume was calculated using two different methods: volume formulas for a cone (experiments 1-4) and a cylinder (experiments 5-6), and sand filling (experiments 7-15) d The bow drill and the twist-reverse-twist drill were used alternately in experiments 2 and 3, for a total duration of 20 h e The volume of 477 ml relates to experiments 7 and 13 combined depressions are larger than those produced using other techniques-ranging from 9 to 13 cm. Experiment 11, for example, lasting 5.5 h, produced a mortar with a depression 12 cm deep and having a diameter of 13 cm, a process that reduced its weight by 15 kg (Fig. 4).

Discussion and Conclusions
Natufian mortars were used in the Levant during a significant cultural transitional stage, starting at the end of the Epipalaeolithic (Natufian: 15,000-11,500 cal. BP). The mortars, like other ground stone tools, are among a variety of cultural innovations that preceded the Neolithic agricultural transformation, making possible the processing of plant foods. Only some segments of the history of ground stone tools are preserved in the archeological record (Dubreuil and Savage 2014); their production process remains unclear. The goal of the present study was to use experimental archeology to examine possible techniques for producing Natufian mortars. The results suggested a variety of insights relating to the properties of  Table 2; mortar 5) basalt raw material, the efficiency of different production methods, the subsequent waste products, and the implications of these for understanding Natufian culture. Clearly, it is impossible to estimate the full range of basalts available during Natufian times or their geological source (but see the attempt by Weinstein-Evron et al. 1995); however, our experimental study enables us to evaluate the preferable properties of basalt intended for GST production. These refer to morphology (preferably round boulders requiring minimal exterior design) and durability (finevs. coarse-grained). As illustrated in Fig. 2, we recognize two types of basalt suitable for the modification of mortars. These can be described generally as vesiculated, fine-grained basalt (type A) and non-vesiculated, coarse-grained basalt (type B). The latter is also preferable for the modification of work tools, i.e., pestles.
Regarding the efficiency of production methods, drilling techniques were found to be efficient only for the initial stages of depression preparation and only for a limited duration, as they all required frequent renewal of the drill heads. We found these results to be somewhat surprising as the use of drills during the Epipalaeolithic is well known, and prior to our experimental work, they were promising candidates for the production of Natufian mortars. Instead, pecking and battering with a hand-held basalt chisel was found to be the preferable production method. This is the only method that allows for the production of complete mortars-from the initial preparation of the depression through modification of the interior face as well as the exterior face. Furthermore, this technique, and carving in general, are more efficient than drilling, requiring less time and investment (Table 3). To sum up, we conclude that the most efficient technique for the production of Natufian basalt mortars is pecking and battering with a hand-held basalt chisel. Clearly, a comparison with the Natufian archeological record is beyond the scope of this study. Similar chisels can be found at quarry/workshop sites (e.g., Rosenberg et al. 2008 A small number of waste products were formed as byproducts of the mortar modification process (Fig. 5). We should differentiate between the waste resulting from the modification of the interior face and that of the exterior face. The modification of the interior face, regardless of the technique used, contributes mostly powder/small particles that cannot be detected with the naked eye. Analyses of sediment samples may discover the presence of such powder, but it is difficult to determine whether this originates from the production or use of the basalt mortars. The modification of the exterior face produces small and mediumsized flakes (Figs. 4 and 5), and their presence at archeological sites can suggest that some exterior design work was carried out at the site and not at remote quarry/workshop sites. However, in the process of making a basalt mortar, the energy inflicted on the basalt chisel (during both interior and exterior modification) causes fragmentation and results in basalt flakes and fragments that may resemble those produced during the modification of the exterior face of the mortar itself (Fig. 5).
The presence of such waste products at Natufian sites is rarely described, and when waste from ground stone tools is reported, it consists mostly of broken pestles and GST fragments. The fragmentation of these tools is considered by many to result from intensive usage or from intentional ritual practices (e.g., Rosenberg and Nadel 2014;Ovadia 2016). Our experimental study provides another possibility-that a mortar can accidentally break during production (Fig. 6).
Such data regarding possible waste products have implications for our understanding of Natufian basalt assemblages, as some can be traced archeologically. Other possible implications relate to the physical load involved in the production of GST, although this subject was not included in our research questions. Our personal experience suggests that such repetitive work causes severe pain, particularly to the elbow and wrist of the working hand, to such an extent that changing hands was required. Although these are subjective effects, they may be related to Fig. 5 Waste products from experimental mortar production. a Fragments of basalt chisels. b Flakes produced during exterior design. c Powder produced during interior design. Scale bar = 15 cm the variety of physical deformations reported for Natufian (Peterson 1998;Eshed et al. 2004) and Neolithic (Eshed et al. 2004) male populations. The bilateral asymmetry in MSM scores for males, indicative of right-side dominance, is often interpreted as a result of continued hunting with spears or bows and arrows but may in fact result from GST production. The repetitive use of GST can also cause physical deformations (e.g., Molleson 2007).
The approximate production time for a medium-sized mortar (exterior diameter of approximately 30 cm) is only 5 h. Similar time estimates have been reported by Squitieri and Eitam (2016). However, we should be cautious when estimating production time for larger mortars, as it depends on a variety of factors, including raw material type, weight, size (length and diameter), depth, and the design of the rim and other decorative elements on the exterior face of the mortar (e.g., Perrot 1966;Edwards 1991). Such large, decorative mortars may suggest the next phase of experimental research on Natufian GST. It is likely that such mortars required a longer production time and may have involved skilled craftsmen. Another suggestion that stems from our results concerns analysis of Natufian archeological assemblages. Tracing basalt chisels and basalt waste products, such as those reported in the current study, may support our conclusion regarding the modification of basalt mortars by battering with a basalt chisel.
Ground stone tools are common components of ancient and modern households and exhibit a wide range of forms and production techniques. Although this study focused on Natufian mortars, our results may contribute to our understanding of the production of ground stone tools in general.