From edge to mark. Investigating the relationship between cut marks and lithic raw materials

1 Introduction

Among the wealth of research on Paleolithic lifeways, the relationship between lithic tools and faunal materials remains insufficiently integrated. Carcass processing—encompassing gutting, meat removal, disarticulation, skinning, and the fracturing of bones to extract marrow—is central to understanding ecological strategies and social organization (Kelly, 2003; Vettese, 2019). These practices directly link lithic and faunal records: a range of stone tools, from retouched implements to unretouched flakes and cobbles, were used both to incise soft tissues and to fracture bones, leaving diagnostic traces such as cut marks and impact-related breakage patterns. Yet these lines of evidence have often been analyzed separately, obscuring aspects of the technological choices and behavioral dynamics involved in butchery.

One of the key unresolved questions is whether the properties of different lithic raw materials are reflected in specific cut marks characteristics (Courtenay et al., 2019c; Yravedra et al., 2019). Previous experimental research has examined how different lithic edge types—such as unmodified flakes, retouched flakes, and handaxes—produce distinct marks (Bello and Soligo, 2008; Boschin et al., 2021; de Juana et al., 2010; Domínguez-Rodrigo et al., 2009; Moretti et al., 2015; Val et al., 2017; Yravedra et al., 2019, 2017). A wide range of raw materials has been examined in comparative studies: flint and quartzite (Linares-Matás et al., 2019; Maté-González et al., 2018; Val et al., 2017; Yravedra et al., 2019), quartzite and basalt (Courtenay et al., 2019b; Yravedra et al., 2017), and flint, quartzite, and quartz (Stepanchuk and Naumenko, 2023). Comparative analyses from Lower (Bello et al., 2009; Courtenay et al., 2019a; Merritt, 2012; Stepanchuk and Naumenko, 2023; Yravedra et al., 2017), Middle (Linares-Matás et al., 2019; Moclán et al., 2018), and Upper (Yravedra et al., 2019) Paleolithic contexts support these findings.

However, few studies have explicitly addressed the interrelation between raw material properties and the formation of cut marks. Experimental research has suggested both qualitative (Moclán et al., 2018) and quantitative (Braun et al., 2008; Maté-González et al., 2018; Yravedra et al., 2017) differences in cut marks morphology related to raw material granulometry. These findings suggest a reciprocal relationship, where cut mark morphology is shaped by, and in turn reflects, how tool edges respond to both intentional modifications (e.g., retouching) and unintentional wear (e.g., edge dulling). Yet, despite significant advances in cut marks analysis, there is still little direct connection to the lithic tools that produced the marks. This gap reflects the tendency to address lithic and faunal evidence separately, combining them only at the interpretive stage rather than within shared analytical frameworks. The result is a fragmented view that calls for more integrated, interdisciplinary and systematic research.

This study seeks to move beyond this fragmentation by integrating lithic and faunal evidence within a single experimental framework. It is motivated by an interpretive problem identified in Middle Paleolithic contexts of the southern Caucasus, where widely diverse raw materials were utilized over an exploitation territory ranging a radius of 60–200 km (Frahm et al., 2025; Gasparyan, 2010; Gasparyan and Glauberman, 2022). Those raw materials comprise of different mechanical properties (Nora et al., 2025b) that reflect a complex decision-making and possible tool use strategies. A particular example is the site of Kalavan 2 (Malinsky-Buller et al., 2021), where imported distant obsidian blanks and retouched tools are highly curated based on their maintenance strategies (Frahm et al., 2025; Malinsky-Buller et al., 2021; Nora et al., submitted), while closer sources of raw materials such as dacite and flint appear mostly as early-stages of the reduction sequence (Malinsky-Buller et al., 2021; Nora et al., submitted). This raw material economy coexist with faunal evidence for secondary butchery activities (Malinsky-Buller et al., 2021; Mistri, 2025). These patterns raise the question of whether differences in raw material selection and curation are reflected in butchery practices and in the cut marks preserved on bone surfaces. Addressing this issue requires experimental baselines capable of evaluating the extent to which cut mark morphology can be linked to tool raw material under controlled conditions.

Here, we present a pilot experiment designed to establish baseline criteria for identifying the raw materials and edge morphologies of flakes used to produce cut marks. This approach isolates key variables and provides a solid foundation for understanding the relationship between lithic tools and the marks they leave. Future research will extend this baseline by incorporating more complex butchery contexts to better reflect the variability observed in natural archaeological assemblages. The experiment is part of a broader project designed to develop an experimental protocol for replicating butchery using stone tools made from diverse raw materials, with the aim of producing cut marks under controlled, replicable conditions.

To effectively integrate lithic and faunal analyses, this study is grounded in the premise that both lithic tools and cut marks constitute a complex system of interrelated factors. Lithic tools are conceptualized not as static artifacts but as dynamic, functional objects, whose use and modification are conditioned by their morphology, the specific actions performed and the mechanical properties of the raw materials (Brantingham and Kuhn, 2001; Collins, 2007; Eren et al., 2014; Goodman, 1944; Schiffer, 1976). The spatial organization of functional elements on a given lithic tool—such as prehensive areas, retouched edges, and active cutting edges (sensu Soriano, 2000)—varies in response to the technological and behavioral demands of each task. Consequently, tool configurations are context-dependent and shaped by the interaction between use and material properties (Kuhn, 1995; Nelson, 1991).

Building on this conceptual framework, the cut marks produced during the experimental session were analyzed through a methodological approach that moves beyond basic descriptive typologies. Previous research has demonstrated that qualitative observations alone are insufficient to capture the subtle variations that link specific cut mark features to particular tools or raw materials. To address this, we employed a multi-scalar analytical protocol incorporating stereomicroscopy, digital microscopy, and confocal microscopy, alongside statistical methods, to identify which cut mark features are most relevant for assessing the impact of raw material and, by extension, the nature of the cutting actions involved.

Our findings contribute to clarifying how specific raw materials influence actions related to carcass processing, helping to bridge the long-standing divide between tool and the traces they leave on bone. The controlled nature of an experimental setting allows for direct comparison of tool morphology, performance, modes of use, and the resulting cut marks characteristics. By describing these relationships and identifying the variables that most consistently link cut mark features to raw material, this approach offers substantial potential for refining archaeological interpretations of butchery practices and raw material selection in Paleolithic contexts.

2 Materials and methods

2.1 Context of the project

This study was developed in the context of the ERC TransCause project, investigating the population of Southern Caucasus, areas that can help address a significant gap in our understanding of the Middle-to-Upper Paleolithic transition(s), dated to approximately 50,000–38,000 years BP. As a crucial migration corridor for Homo sapiens moving from Africa to Europe, the region remains underexplored and demands further archaeological investigation (Goder-Goldberger and Malinsky-Buller, 2022).

Previous research in the region has revealed intriguing patterns in lithic technology and raw material exploitation. Among the ongoing studies, the most comprehensively investigated site to date is the open-air site of Kalavan 2 in Armenia, dated to between 60,000 and 45,000 BP and producing rich faunal and lithic assemblages (Frahm et al., 2025; Ghukasyan et al., 2010; Malinsky-Buller et al., 2021).

The lithic assemblage at Kalavan 2 includes initial reduction stages from dacite and flint from nearby outcrops (Malinsky-Buller et al., 2021; Nora et al., 2025a) and, in contrast, obsidian tools exhibiting extensive use, rejuvenation, and resharpening, but no evidence of early-stage reduction. Technological analysis of the obsidian tools suggests off-site manufacture and selective transport from a distance ranging 40 to 200 km (Frahm et al., 2025; Malinsky-Buller et al., 2021). This blunt contrast in raw material procurement and tool production strategies raises important questions about technological choices and their relationship to specific butchery practices. Preliminary use-wear analysis displayed patterns of striations and wear polish consistent with evidence of contact with hard animal material (Malinsky-Buller et al., 2021, Figure 10). The lithic findings complement ongoing faunal analyses, which suggest that Kalavan functioned as a secondary butchery site, primarily focused on processing large bovid limbs (Bos/Bison; Malinsky-Buller et al., 2021; Mistri, 2025).

To advance the common interdisciplinary aim of uncovering decision-making processes by investigating faunal prey and body part selection alongside lithic raw material choices, an experiment investigating the relationship between cut mark production and raw material features can significantly advance our understanding of these technological decisions.

2.2 The experimental session

The experiment was conducted at the European Center for Prehistoric Research (CERP) in Tautavel (France) over 8 days of active work during February and March 2025. The experiment was carried out in four consecutive phases: the production of lithic tools, the production of cut marks, the analysis of the functional internal organization of the tools—that is, the arrangement and interaction of the active parts of the lithic tools that determine how they function during the cutting action—and the microtopography analysis of the cut marks.

2.2.1 Lithic tools production and documentation

The production of lithic tools for the experiment took place over 4 days in the open-air knapping area at CERP (Figure 1A). The tools were crafted from three types of raw material with differing granulometric properties: dacite (coarse-grained), flint (fine-grained), and obsidian (non-granular; Figure 1B). The experimental design aimed to use raw materials sourced exclusively from Armenia. However, logistical constraints prevented the shipment of some specimens. As a result, the experiment employed dacite from Armenia (Aghavnatun), Eocene flint sourced from Israel and obsidian sourced from Greece (Milos) and Armenia (Gutansar).

An unstructured reduction sequence was applied to all raw material blocks (Supplementary Data Sheet 1), reflecting technological features observed in archaeological assemblages from late Middle Paleolithic sites in Armenia, where lithic assemblages predominantly consist of flakes from informal knapping rather than highly structured sequences. The objective of the knapping sessions was to produce thick, unbroken flakes featuring both sharp and thick edges suitable for retouching. Cortical flakes were deliberately excluded from production. Two expert knappers (MB and JV) carried out the reduction using hammerstones made of quartzite, sandstone, cinerite, and granite. The size and type of hammerstone were chosen by the knappers based on the specific characteristics of each core. The reduction of each block lasted between 10 and 15 min (Figure 1A).

At the end of each reduction sequence, a sample of flakes (ranging from three to nine from each block) was selected for use for the production of cut marks. In total, 36 flakes were selected, 12 from each raw material type. Of the selected flakes, half were left unretouched, while the remaining half were retouched by the same knappers. Retouching was carried out using sandstone retouchers of varying shapes (Figure 1C). No organic tools were used in order to limit variable factors in the retouch morphology. The objective of the retouching activity was to produce straight edges characterized by fan-shaped, low-angle retouch. Each retouching session lasted approximately 5 min per tool. All lithic tools were documented with photographs both before and after use.

In addition, high-resolution 3D digital mesh models of the tools were generated using a Revopoint 3D MetroX structured light scanner prior to and following use The digital models were then processed using the Artifact3-D software (Grosman et al., 2022) for consistently orienting the models to standard planform view based on normal distribution. As this geometric positioning does not conform to standard technological positioning in the sense of proximal-distal orientation, additional orientation manual orientation was applied but limited to rotation on the calculated planform plane.

The oriented artifacts were then used to create high resolution plates (12,00 DPI TIF) showing the artifacts from standard views (Supplementary Images 1–6). The functional internal organization of the lithic tools used in the experiment was extracted using an edge-outline segmentation procedure developed specifically for this study in the MATLAB development environment (https://github.com/gadiherz/EdgOutSeg.git). This procedure first identifies the boundary of a selected planar projection of the 3D model, with the chosen projection corresponding to the distal face shown on the generated plates. The analyst then marks specific segments on the boundary by selecting start and end point and assigning each segment to one of three categories: contact, prehensive or retouch area. Selected areas were based on observing the video recordings of the experiment. The procedure accommodates overlapping segments as well as multiple segments of each type on a single artifact.

For each plate, it outputs a separate Microsoft Excel file listing the following variables for each artifact: the overall arch length (in mm), and for each marked segment, its type, start and end positions expressed as percentage of the arch length and in absolute millimeters. The results for all plates were subsequently statistically analyzed in SAS JMP Pro v.18.

2.2.2 Experimental setup

The second phase of the experiment involved using the selected lithic tools to produce cut marks on bone surfaces. The bones were sourced from a butchery in Perpignan (France) 3 days before the experiment, and were kept been frozen until the morning of the experiment. The bone sample consisted of 14 long limb sections from adult domestic cattle (Bos taurus), specifically humeri and tibiae. All the bone sections used in the experiment were fresh, with some still containing marrow and varying amounts of grease and ligaments on their surfaces.

As a pilot, the focus of the experiment was not on replicating exact butchery procedures but on performing standardized actions designed to generate a range of cut marks. These actions were applied in a controlled, sequential setting and were predefined in order to explore systematically how variations in raw material, tool morphology and motion influence cut mark formation. Although these actions do not directly replicate specific butchery techniques, they provide a structured framework to investigate the relationship between tool attributes and the resulting modifications on bone surfaces.

Each flake is used to perform a single action on a specific portion of the bone surface. The actions are categorized based on their direction (unidirectional, bidirectional) and orientation (longitudinal, oblique, transverse; Figure 2A):

• Direction refers to the path or movement pattern of the tool that created the mark. It describes whether the tool was moved in one consistent way or in more than one direction during the cutting action.

• - Unidirectional: The cut mark was produced by a single, consistent movement in one direction, moving along the bone toward the operator.
• - Bidirectional: The cut mark resulted from back-and-forth movements, repeatedly moving the tool in two opposite directions during the action.

• Orientation refers to the angle or alignment of the cut mark relative to the long anatomical axis of the bone. It describes how the cutting action is positioned on the bone surface.

• - Longitudinal: The cut mark runs roughly parallel to the long axis of the bone.
• - Oblique: The cut mark is angled diagonally across the bone, not parallel or perpendicular to its long axis.
• - Transverse: The cut mark runs across the bone at a near right angle to its long axis.

The operator used their dominant hand to perform the action, and remissive hand to stabilize the bone. The definitions proposed by Key et al. (2018), Table 3) were adopted to describe the grip used for each flake and each movement. The working angle between the cutting edge and the bone surface is kept as close as possible to 90 °. Live feedback was provided by the operator during the cutting actions, allowing observation and comparison with the actual performance of the tools in real time. Such recording aimed at developing a system for assessing lithic tool performance based on action, raw material, and edge characteristics. To evaluate tool use during the experimental sessions, the following variables were considered: raw material (dacite, flint, obsidian), lithic edge type (retouched, unretouched), direction of action (unidirectional, bidirectional), and orientation of action (oblique, transversal, longitudinal). No predetermined performance scale with fixed factors was established; instead, the scale was derived from the operator's feedback, enabling identification of the aspects of tool performance most relevant to the action, the raw material, and the edge within the specific context of this experiment.

Each group of cut marks associated with one action and one lithic tool is defined as “cut marks series” (CMS). In each CMS, a predetermined number of 10 cut marks are produced (Figure 2B). On each bone section, three or four CMSs have been made, depending on the available surface and the anatomy of the bone. All flakes were manufactured with similar technological features and were randomly assigned to each action.

The experimental activities took place indoors on a table approximately one meter high. Each session was documented with video recordings from two primary perspectives. The first (POV1) captured the operator's point of view, using an AKASO E7000 action camera mounted on a head strap, angled downward between 60 and 90 ° toward the experimental activity. The second (POV2) was recorded by a Sony a7R IV digital camera with a 27–70 mm f2.8 GM lens positioned in front of the operator, angled at 45 ° toward the working area. Occasionally, a third camera (POV3) recorded by a triple rear camera system with a 108 MP main sensor, an 8 MP ultrawide lens, and a 2 MP macro lens, was placed on the table at a 90-degree angle to capture detailed contact between the lithic edge and the bone surface during the action (Figure 2C).

Following the experiment, the bones underwent a cleaning procedure to prepare their surfaces for microscopic analysis. Remnants of ligaments and greasy tissue were carefully cleaned by boiling the bones in tap water without detergent, a step necessary to preserve the samples and ensure clear microscopic observation of the cut marks. The cleaning protocol took a full day to complete and is detailed in the Supplementary Data Sheet 2.

2.3 Cut marks documentation and microtopography analysis

Photographs of the cut marks were taken using a Hirox HRX-01 3D digital microscope equipped with an HR-5000 triple-objective revolving lens at 20x magnification, employing a combination of ring and coaxial lighting. The microscope, housed in the Human-Environment Dynamics Laboratory (HumEnDy Lab) at the Institute of Archaeology (The Hebrew University of Jerusalem) was used in this study primarily to capture high-quality 2D images of the CMS produced during the experiment.

Microtopography analysis was conducted on a sample of 62 cut marks at the PACEA laboratory in Bordeaux (France) using a Sensofar S neox confocal microscope with a Nikon EPI 20x objective. Marks resulting from unidirectional action were primarily selected, while those produced by bidirectional movements were excluded. This decision, driven by time constraints to complete the full analysis, was based on the observation that bidirectional movements were more difficult to control during the experiment, likely producing highly irregular cut marks. This assumption was supported both by the operator's feedback during the experimental sessions and by high-resolution observations of the cut marks under the Hirox microscope.

The final dataset includes 25 dacite cut marks, 16 flint cut marks, and 21 obsidian cut marks (Table 1). These are associated with lithic tools as follows: seven dacite tools (three retouched, four unretouched), six flint tools (three retouched, three unretouched), and six obsidian tools (three retouched, three unretouched). Initial scans were performed with a 5x lens to obtain an overview of the cut marks, followed by 20x magnification scans of the central segment of each mark, which were used for profile extraction. Only surface scans with >95% measured points were retained for analysis.

Table 1. Sample of cut marks selected for the microtopography analysis.

The images were initially processed using the Sensoscan software. After generating the 3D images, the profile of the central section of each cut mark was extracted using the Digital Surf MountainsLab Premium software (Version: 11.0.11118, 2025/06/10). A standardized workflow was applied during this process. Outliers and edge-related distortions were removed using an algorithm set to medium intensity. Non-measured points were interpolated using a smooth nearest-neighbor method. To ensure consistent orientation across samples, all surfaces were leveled using a planar alignment fitted through least squares. Profiles were extracted at the midpoint of each cut mark and were subsequently trimmed to isolate only the cut mark's microtopography, excluding surrounding cortical surface features such as shoulder flaking, shoulder effects, or incidental striations.

The extracted profiles were analyzed using the R programming language (v.4.5.1; R Core Team, (2017), where statistical tests were conducted following the methodology outlined in Courtenay et al. (under review). The statistical analysis performed in this study was designed to assess whether the internal profile of cut marks varies depending on the raw material of the stone tool used. This analysis consists in automatically calculating Depth, Width at Surface, Opening Angle of each mark through a series of trigonometry calculations that establish a triangle on the deepest point of the mark, and the left and right shoulders. Since the Opening Angle is a circular variable, it was analyzed using circular statistics rather than linear models. Opening Angle is expressed in radians during calculations and then converted in degrees for convention of data presentation. To calculate asymmetry, we see how much of a difference exists by folding the profile along a point established at the maximum depth, followed by the calculation of the Square Root of the Mean Squared Distance between the two sides.

To assess morphological variation beyond what is captured by linear variables, global morphology is then analyses using Elliptical Fourier Analysis (EFA), with a total of six harmonics. Because allometric relationships were detected among the profiles, normalization of the coefficients was not performed, meaning all analyses were performed considering cut mark profile form (size + shape). Both the measurements and EFA coefficients were then analyzed multivariately using a principal component analysis (PCA).

Shapiro-Wilk tests were applied to assess whether each variable followed a normal distribution. The interpretation of the results followed a conservative threshold system (Courtenay, 2024):

• p < 0.003: Data is likely not normal; shows notable differences.
• 0.003 <p < 0.05: Data is likely not normal; shows moderately notable differences.
• p ≥ 0.5: no evidence against normality; no notable differences.

Non-parametric Kruskal–Wallis tests were applied to assess whether these variables varied as a function of raw material. A Watson-Williams-type test was conducted to assess whether opening angle distributions differed meaningfully by raw material, and the results did not suggest notable differences between groups.

3 Results

3.1 Experimental session observations

Each tool was assigned to a single type of movement, producing one CMS per tool (Supplementary Table 1). The cutting actions lasted between 8 and 46 s per series, with most lasting around 25 s. The operator selected the grip based on the size and shape of the tool, as well as the type of action to be performed. The most commonly used grip was 3.4b (Key et al., 2018, Table 3), which involves three digits (the palmar side of the second phalanges and the lateral side of the third phalanges) making contact with the tool (Supplementary Images 7–12).

It was observed that the raw materials behaved quite differently during the experiment. Unretouched dacite (coarse-grained) showed significant edge fragility, with edges crumbling during use and leaving small chips on the bone surface and inside the cut marks. Retouched dacite did not exhibit this behavior. Both retouched and unretouched flint (fine-grained) showed no visible edge damage—they neither crumbled nor became dull. However, retouched flint struggled to penetrate the bone surface and tended to slip, while unretouched flint performed slightly better but still exhibited difficulties penetrating the bone surface. Obsidian (non-granular), both retouched and unretouched, performed best, with edges that did not crumble and effortlessly penetrated the bone surface during the cutting actions.

Direction and orientation of cutting action also affected tool performance across raw materials. Unidirectional movements were easier to perform than bidirectional ones, as the latter made it harder for the operator to maintain a stable motion. Transversal movements were particularly challenging due to the awkward hand and arm position required to keep the tool's edge perpendicular while moving along the transverse axis, causing discomfort to the operator performing the action.

Overall, the operator's feedback was influenced by both the condition of the bone and the characteristics of the tool. When bones had residual grease, fatty tissue, or tendons still attached, the operator found the actions more difficult and less effective with certain tools. Tool features also played a key role; some tools were too small or uncomfortable to hold, negatively impacting edge performance and the quality of cut marks produced. A full summary of the operator's feedback for each experimental session is available in the Supplementary Data Sheet 3.

3.2 Rating scale for control

By observing tool behavior in relation to raw material, lithic edge type, direction and orientation of action, we identified four key indicators related to tool edge performance during use. These indicators assess the “control” of the action, defined here as the operator's ability to consistently manage the direction, depth, and placement of cut marks for each tool and movement. Each indicator is scored on a scale from one to five, based on the number of observable control issues during the session (Table 2). Additionally, two more indicators were included to describe factors influencing performance: the condition of the bone surface and the comfort of the tool's grip.

Table 2. Descriptions of the qualitative criteria used to evaluate tool control during the cutting actions.

By combining the first four indicators and factoring in the last two, we developed a control rating scale:

• Excellent (control score between 5 and 4.5): full control over the movement, with smooth, stable actions. Cut marks are placed precisely as intended and the edge remains stable. No significant interference from the bone surface, and the tool grasping position remains optimal without the need for adjustments.
• Good (control score between 4.4 and 3.5): the action is mostly controlled with minor slips or deviations. There may be some adaptation needed for the tool grip or bone surface shape, but it does not significantly affect the cut mark placement. The edge performs well, but slight inconsistencies in the direction and orientation may occur.
• Moderate (control score between 3.4 and 2.5): partial control over the action. Occasional adjustments in body position, tool grip, or stabilization are needed. Cut marks may be slightly misdirected or misplaced, and there may be noticeable interference from the bone surface or tool stability. There may be moderate discomfort in holding the tool.
• Poor (control score between 2.4 and 1.5): little control over the action. The edge frequently slips, the trajectory is inconsistent, and cut marks are misaligned or misplaced. The bone surface or tool shape may interfere significantly with the process, and the tool grasping is uncomfortable, often requiring adjustment.
• Very Poor (control score <1.5): no control over the action. The motion is unstable throughout, with the edge frequently slipping or failing to penetrate the bone surface. Cut marks are largely misplaced, and both the tool and bone surface are problematic, making it difficult to maintain any consistency or precision. The tool grasping may be uncomfortable.

Dacite tools display the lowest overall score, categorized as moderate, and show the greatest variability, indicating inconsistent performance across tools (Supplementary Table 2). Looking closer at individual scores, dacite rates relatively high in direction consistency (3.92), suggesting some reliability in movement orientation, but scores lower in stability and precision (both 3.00), reflecting occasional slips or less accurate cut mark placement. Flint also falls within the moderate range but slightly outperforms dacite overall. While flint has a somewhat lower stability score (2.58), it scores higher in precision (3.25) and edge performance (3.25), indicating that despite reduced stability, it compensates with better cutting accuracy and edge durability. Obsidian stands out with the highest average control score, classified as good. It achieves solid stability (3.00) and edge performance (3.33) scores, though it scores slightly lower in direction (2.75) and precision (2.83). This suggests obsidian tools provide a stable cutting edge and comfortable handling, even if there is some variability in movement direction and precision. Overall, these differences highlight the nuanced impact of raw material on tool control, with edge performance and stability emerging as critical factors for superior control.

The influence of edge type on tool performance is also evident. When examining the experimental sessions by raw material, obsidian consistently demonstrates superior performance, dacite exhibits the lowest, and flint occupies an intermediate position (Figure 3A). This trend persists when comparing retouched and unretouched edges within each raw material category: obsidian remains the highest performing material, dacite the lowest, and flint intermediate (Figure 3B). Retouched edges exhibit reduced variability in performance compared to unretouched edges, particularly for flint and dacite. In contrast, variability in obsidian remains relatively constant regardless of edge modification, although retouched obsidian edges tend to perform slightly less effectively in terms of action control than their unretouched counterparts. Comparisons between flint and dacite reveals distinct patterns in variability. Unretouched flint edges demonstrate higher average control scores but also greater variability, indicating a wider performance range. Conversely, retouched flint edges consistently yield lower control scores with minimal variability. A similar pattern is observed in dacite, where unretouched tools display the greatest variability in control scores, spanning from poor to very good, while retouched dacite tools generally afford diminished control with more limited variability.

Another important factor to consider is the condition of the bones combined with the features of the tools, specifically the bone's morphology and the presence of greasy areas on its surface. The potential relevance of these factors emerged during the experiment based on operator feedback. These conditions may influence tool performance because greasy areas can make it harder to stabilize the bone, resulting in less stable movements, or cause the edge to slip regardless of its suitability for the action. At the same time, the tool's features—particularly the size of the tool and how comfortable it is to grasp and hold—are crucial. Even if the edge is ideally suited for the action, an uncomfortable grip can make the movement difficult to perform. When applying the control rating scale to our dataset, we found that bone grease and morphology have a smaller impact on the action than grasp comfort (Figure 3C). Whether bone grease or morphology is present does not significantly affect the control rating. In contrast, tools that are uncomfortable to hold almost always receive poor or very poor control ratings, whereas tools that are comfortable to grasp generally yield higher control ratings.

3.3 Functional outline segmentation of lithic tools

The analysis of the experimental lithic tools focused on two main independent variables: the raw material (dacite, flint, or obsidian) and the edge modification (retouched or unretouched). The overall arc length of the artifacts' boundaries varies by raw material. Obsidian tools exhibit a markedly shorter circumference compared to both dacite and flint, a pattern observed in both retouched and non-retouched pieces. In non-retouched specimens, the difference is significant only between obsidian and flint, whereas in retouched tools, obsidian has a significantly shorter boundary than both dacite and flint (p < 0.05). Within each raw material group, however, there are no notable differences in circumference between retouched and non-retouched items. The proportion of the arc length that is retouched does not significantly differ between the raw materials. Retouch consistently occupies approximately 42–45% of the tool's boundary across dacite, flint, and obsidian.

When comparing the lengths of the prehensive and contact segments across raw materials, no consistent notable differences are observed. This holds true for both retouched and non-retouched tools, and applies whether measurements are in absolute terms (mm) or relative to the total circumference. The sole exception is found in the relative length of the contact segment in retouched tools: obsidian specimens display a significantly greater proportion of contact area compared to dacite (p = 0.0073), likely reflecting the overall smaller arc length of obsidian tools. Additionally, when comparing raw materials, obsidian tools show a significantly lower combined length of prehensive and retouched segments than flint, further reflecting differences in tool size and morphology. A detailed description of these results is presented in Supplementary Data Sheet 4.

3.4 Statistical analysis of cut marks' internal profile

This section summarize the results of the statistical analysis performed on the internal profiles extracted from the cut marks. Raw data and the full statistical results are presented in the Supplementary Data Sheet 5.

3.4.1 Descriptive and univariate analysis of cut marks' profiles

Based on the thresholds established in Courtenay (2024), all three linear variables were interpreted as not normally distributed at the global level:

• Depth: W = 0.759, p = 1.02e-08
• Width at Surface: W = 0.748, p = 5.68e-09
• Asymmetry: W = 0.791, p = 5.58e-08

In all cases, the results of the Kruskal–Wallis tests indicated no notable differences between groups under the adopted interpretive framework (Figure 4). These findings suggest that raw material did not play a detectable role in shaping cut mark morphology with respect to depth, width, or asymmetry. Although some tendencies are visually apparent in the boxplots—for example, slightly deeper marks produced with dacite or greater asymmetry associated with obsidian—these patterns are not supported by the statistical analyses.

3.4.2 Opening angle and circular statistics

Across the full dataset, opening angles displayed a unimodal and highly concentrated distribution, meaning that most cut marks are oriented in a consistent direction rather than being randomly dispersed. The Rayleigh test yielded a p-value below 0.001, which indicates that the cut marks show a clear directional tendency. More specifically, the majority of marks were clustered around a mean direction of approximately 110 °, with a median of 112.4 °.

The use of different raw materials did not appear to alter this behavior substantially. When broken down by material:

• Dacite: Mean = 105.7°, Median = 113.1°, Circular Dispersion = 0.146
• Flint: Mean = 108.5°, Median = 108.8°, Circular Dispersion = 0.169
• Obsidian: Mean = 116.6°, Median = 113.8°, Circular Dispersion = 0.151

3.4.3 These values reflect tight clustering around a common orientation, with only minor variations between groups. Principal component analysis

The first principal component (PC1) explained 54.1% of the total variance and was primarily driven by Depth and Opening Angle, indicating that this axis captures differences in how deeply the tool incised the bone and how steep or open the resulting cut mark profile was (Table 3). When plotted by raw material type, the PCA revealed no notable group separation, despite some apparent visual clustering—particularly a greater spread among flint samples (Figure 5). High PC1 scores correspond to deeper, steeper cuts, while lower scores reflect shallower, wider marks. The second component (PC2), accounting for 37.5% of the variance, was most influenced by the Width, and reflects variation in surface width independent of depth or angle. Asymmetry contributed minimally to both components, suggesting this parameter is not a key source of variation in this dataset. Opening Angle and Depth are the strongest contributors to PC1, Width strongly aligns with PC2, and Asymmetry has only a weak influence overall (Figure 5).

Table 3. Principal component analysis (PCA) results showing the percentage of variance explained by the first four principal components (PCs).

To verify this, a series of tests was conducted:

• A Shapiro-Wilk test showed that PC1 scores were not normally distributed (p < 0.003), indicating notable differences from normality and justifying the use of non-parametric methods.
• A Kruskal-Wallis test comparing PC1 scores across raw material groups yielded no notable differences (p > 0.05).
• A pairwise permutational MANOVA (Wilks' λ, 999 permutations) assessing group separation in multivariate space (PC1 + PC2) also returned p-values well above 0.05 for all pairwise comparisons, indicating no notable multivariate differences among the raw material groups.

Consistent with the absence of notable group separation in the PCA, non-parametric tests confirm that raw material is not a relevant factor in structuring cut mark variability.

3.4.4 Elliptic Fourier analysis of cut marks' form

Before performing Elliptic Fourier Analysis (EFA), permutation testing was conducted on centroid size and raw material to assess their effect on cut mark internal profile form. The results revealed a notable effect of centroid size (allometry), but no notable effect of raw material. Centroid size accounted for 9.6% of the variation in shape (F = 6.40, Z = 2.90, p = 0.003), while raw material explained only 3.6% of variance and was not notable (F = 1.20, Z = 0.58, p = 0.283). The vast majority of variation (~86.8%) was explained by residuals—i.e., factors not captured by raw material or size—implying a high degree of within-group variability.

The PCA supported these findings (Figure 6). The first axis (PC1) explained 77.3% of total variance and the second axis (PC2) 15.5%, together capturing 92.8% of variation. PC1 scores were strongly non-normal (Shapiro–Wilk W = 0.70, p < 0.001). A Kruskal–Wallis test showed no notable differences in PC1 scores among raw materials (χ² = 1.79, df = 2, p = 0.409). In the PCA scatterplot, convex hulls of dacite, flint, and obsidian overlapped extensively, with only a few outlying points stretching the hulls. This visual pattern is consistent with the regression results: most variation is driven by size differences along PC1, and raw materials do not form distinct clusters.

Morphological changes of the cut marks profiles along PC1 and PC2 further clarify the nature of this variation. Negative PC1 values correspond to narrower, deeper, V-shaped profiles, while positive PC1 values represent wider, flatter, U-shaped forms. PC2 captures variation in curvature and asymmetry, ranging from flatter forms to more curved, steep-walled profiles. These differences are an effect of allometry, but raw material groups showed no consistent separation along PC2.

Pairwise permutation MANOVA (Wilks' λ, 999 permutations) on the first five PCs (covering >95% of variance) confirmed the absence of notable group differences. Comparisons between dacite and flint showed no notable differences (p = 0.364). Comparisons involving obsidian (dacite–obsidian and flint–obsidian) returned marginal values (p = 0.082), suggesting a weak but non-notable trend for obsidian profiles to differ slightly.

Taken together, the results demonstrate that the dominant axis of cut-mark profile variation is size-related allometry, while raw material has at most a minor, not notable effect.

4 Discussion

Lithic and faunal datasets, when studied separately, can provide only partial insights to archaeological investigation. These two assemblages reveal two sets of decision-making. While lithics bring to light technological choices, procurement strategies, and mobility, faunal assemblages inform us about diet, site activities, and carcass transport. Yet these lines of evidence are functionally connected: the same butchery actions that shape tool use also leave traces on bone surfaces. If studied in isolation, the dynamics that unfold between tools and bones—such as how raw material properties and edge morphologies affect butchery efficiency and the resulting marks—are overlooked. Integrating the two perspectives has the potential to move beyond describing what tools look like or what cut marks look like, toward understanding how they relate to each other in the context of an intricate feedback loop of decision-making. This is the behavioral link that makes lithic and faunal records inseparable.

4.1 Tool performance and raw materials properties

Our experiment demonstrates that raw material had an effect in tool performance. While these performance differences were qualitatively observed during use, the lithic analysis revealed no notable differences between raw materials in tool shape, segmentation, or edge wear distribution. This lack of pattern indicates that the experimental flakes were consistent across raw materials in their overall morphology and edge configurations. As a result, the lithic dataset in its current resolution confirms that no shape- or segmentation-based variables were introduced that could bias the production or interpretation of cut marks.

Obsidian consistently outperformed both flint and dacite, producing sharp, stable edges that penetrated bone surfaces easily and without noticeable deterioration. This reflects the mechanical advantage of obsidian's homogenous, non-granular structure. These results are based on a small sample involving only 10 strokes, and contrasting outcomes have been reported in the literature for experiments involving longer use sequences—for example, after 250 strokes on pine wood (Nora et al., 2025b). A low number of strokes was appropriate for the present experiment, which focused on cut mark morphology and butchery performance rather than prolonged fatigue test on tool efficiency. On the other hand, future studies aimed at evaluating the performance and durability of stone tools would be highly valuable in this field.

Dacite, by contrast, was the most brittle raw material. Unretouched dacite edges crumbled during use, leaving chips embedded in both the bone surface and inside cut marks. This behavior underscores the challenges of coarse-grained materials for tasks requiring precision and stability. The dacite used in this study was crystalline, non-glassy variety classified as coarse-grained, which likely contributed to its fragility. Flint performed more reliably than dacite but did not match the performance of obsidian. Although its edges remained intact, penetration was often less effective, particularly for retouched flakes.

Retouching introduced additional complexity. In dacite and flint, retouched flakes produced more consistent results than unretouched ones, reducing variability in performance. However, this came at the expense of overall control: retouching often blunted cutting edges or changed the ergonomics of flakes, making precise movements more difficult. Obsidian was an exception to this pattern, maintaining high performance regardless of retouch. These results demonstrate that raw material and edge morphology together shaped not only the durability of edges but also the experience of their utilization to perform cutting actions.

4.2 Assessing “control” as a methodological bridge

The type of motion also had a clear influence on both performance and cut mark production. Unidirectional movements proved easiest to control and consistently yielded stable marks. Bidirectional movements introduced instability, while transversal actions were the most difficult to execute. The latter forced the operator into awkward positions that caused discomfort and resulted in less consistent cut marks. These findings suggest that variability in cut marks cannot be attributed solely to raw material or edge type; the ergonomics of motion and body-tool interaction must also be considered.

Equally important was the role of grasp comfort. Across the experiment, the stability of cutting actions depended less on bone condition (grease, curvature, irregularity) than on the ease of holding the tool securely. Comfortable flakes allowed smooth, precise movements, whereas poorly shaped flakes amplified instability regardless of raw material.

The concept of “control” provides a valuable framework for linking lithic ergonomics with cut mark variability. In our experiment, control reflects the interaction of raw material, tool shape, and operator handling, and it determines the stability, precision, and consistency of cutting actions. While control is shaped by variables usually analyzed in lithic studies—raw material, edge morphology, and tool overall form—its outcomes are visible in the faunal record as well. When control is high, actions are stable and cut marks are regular, reducing variability. When control is low—for example, with fragile raw materials, blunted edges, or uncomfortable grips—actions become unstable, marks are misplaced, and variability increases.

This excess variability can mask the subtle influence of raw material on cut mark morphology, making it harder to detect patterns statistically. At the same time, it suggests that measuring control may provide a way to bridge lithic ergonomics and cut mark variability, offering a means for understanding how human–tool interactions shape bone surface modifications. In turn, this framework is crucial for future studies linking experimental results to archaeological butchery practices, where action stability and control would have been central to efficiency and success. These insights also carry implications for interpreting lithic archaeological assemblages: the choice of raw material may not have been neutral or dictated solely by availability, but could instead have reflected an awareness of its impact on performance and control.

4.3 Cut marks morphology and statistical analysis

The statistical analysis of cut mark profiles revealed a striking contrast between observed tool performance and measurable cut mark morphology. Differences between raw materials that were clear during use—affecting performance, stability, and operator experience—did not translate into statistically diagnostic differences in the morphology of cut marks. Standard morphometric variables (depth, width, asymmetry) and circular measures of opening angle showed no notable variation by raw material. Multivariate analyses (PCA, EFA) confirmed that variability was dominated by size-related allometry, while raw material explained only a negligible fraction. Nearly 87% of variance was residual, pointing to the influence of uncontrolled factors. Although per-group sample sizes are modest, the use of non-parametric and circular tests provides a robust assessment of differences between raw materials, and expanding the dataset in future work will allow detection of subtler morphological variations.

At the same time, the absence of notable variation in the lithic dataset strengthens the interpretive value of the cut marks analysis. Since tool morphology and segmentation did not vary systematically between raw materials, we can rule out tool-related variables as confounding factors. In this sense, the lithic analysis serves a control function: it confirms that the observed homogeneity in cut mark morphology cannot be attributed to differences in flake form, grip area, or retouch placement, but rather reflects a genuine lack of raw material signature at the level of mark morphology—at least under current methods.

This paradox highlights both methodological and interpretive challenges. The limited sample size may have reduced sensitivity, but more fundamentally, existing analytical variables may be insufficient to capture subtle raw-material-related differences. More pronounced differences may emerge when comparing stone tools made of materials with greater contrasts in fracture and edge-wear behavior, such as quartz, quartzite, basalt, or silcrete (Courtenay et al., 2019a; Linares-Matás et al., 2019; Maté-González et al., 2018; Stepanchuk and Naumenko, 2023; Val et al., 2017; Yravedra et al., 2019, 2017). Additionally, two-dimensional cross-sectional profiles may miss surface features or three-dimensional curvatures that preserve signatures of edge performance. High-resolution 3D modeling may offer one possible solution, though it risks introducing noise that could obscure rather than clarify patterns. These results caution against over-reliance on cut mark morphology alone and reinforce the need for integrated analytical frameworks.

5 Future integration with archaeological perspectives

While the lithic and faunal results address different aspects of the experiment, together they form a coherent methodological framework. The lithic analysis confirms that the tools used in the experiment were morphologically consistent across raw materials, providing a controlled baseline free from ergonomic or segmentation bias. Raw material shaped cutting performance and control, yet these influences were not detectable in cut mark morphology under current analytical methods.

An integrated approach can address this gap. Lithic tools and cut marks are not independent traces but two facets within the same decision-making processes, and this relationship must be considered in future studies addressing the full sequence of carcass processing, from gutting, to skinning, to meat and tendon removal. Such integration should also involve systematic comparison between lithic edge angles and the angles and orientations of cut marks, as the geometry of the cutting edge and the angle of tool application might directly condition cut-mark formation and therefore reflect specific butchery gestures. By connecting tool ergonomics, motion control, and edge performance to the modifications they produce on bone, we could potentially reconstruct the technological and subsistence strategies that structured Paleolithic lifeways.

At sites such as Kalavan 2, where imported obsidian was used alongside locally procured dacite and flint, integration could prove particularly powerful. Faunal evidence indicates secondary butchery, while lithic evidence reveals deliberate raw material selection and transport. Our experimental results suggest that a strategy like this was likely informed not only by availability but by the practical performance advantages of certain materials. However, only through a full-scale experiment aimed at replicating all the stages of carcass processing using lithics made from the three raw materials, in line with Kalavan's lithic production, can this hypothesis be confirmed and provide tangible information.

In this light, raw material procurement has the potential to emerge not simply as a logistical or environmental deterministic factor imposed upon the hunter-gatherers, but as part of a broader technological and subsistence strategies. The integration of lithic and faunal analyses could allow us to move beyond parallel narratives of technology and diet, and toward reconstructing the full behavioral chain of butchery: from raw material acquisition, to carcass processing, to site organization.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fearc.2026.1738188/full#supplementary-material

Plate illustrating 3D scans of dacite unretouched tools. Supplementary material at the source

Plate illustrating 3D scans of dacite retouched tools. Supplementary material at the source

Plate illustrating 3D scans of flint unretouched tools. Supplementary material at the source

Plate illustrating 3D scans of flint retouched tools. Supplementary material at the source

Plate illustrating 3D scans of obsidian unretouched tools. Supplementary material at the source

Plate illustrating 3D scans of obsidian retouched tools. Supplementary material at the source

Plate illustrating 3D scans of dacite unretouched tools. The colors represent the functional internal organization of the tools: pink=prehensive part, blue=contact part. Supplementary material at the source

Plate illustrating 3D scans of dacite retouched tools. The colors represent the functional internal organization of the tools: pink=prehensive part, blue=contact part, orange retouch. Supplementary material at the source

Plate illustrating 3D scans of flint unretouched tools. The colors represent the functional internal organization of the tools: pink=prehensive part, blue=contact part. Supplementary material at the source

Plate illustrating 3D scans of flint retouched tools. The colors represent the functional internal organization of the tools: pink=prehensive part, blue=contact part, orange=retouch. Supplementary material at the source

Plate illustrating 3D scans of obsidian unretouched tools. The colors represent the functional internal organization of the tools: pink=prehensive part, blue=contact part. Supplementary material at the source

Plate illustrating 3D scans of obsidian retouched tools. The colors represent the functional internal organization of the tools: pink=prehensive part, blue=contact part, orange=retouch. Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Alan Outram, University of Exeter, United Kingdom

Reviewed by: Juan Marín, National University of Distance Education (UNED), Spain

Antonio Pineda, Institut de Paléontologie Humaine, France