Introduction

The use of bone as raw material for toolmaking in the Eurasian Middle Palaeolithic is primarily represented by bone retouchers. These tools were first identified in the early twentieth century (Henri-Martin 1907–1910; Patou-Mathis 2002) and have since been recognised in numerous Palaeolithic contexts across Europe, Africa, and Asia (for a synthesis, see contributions in Hutson et al. 2018).

From a technological standpoint, bone tools throughout Prehistory can be categorised into three main groups: (1) knapped bone fragments mimicking existing stone tools; (2) deeply worked bone fragments resulting in fully shaped, repetitive, and standardised tool types with specialised functions (both categorised as ‘formal’ tools); and (3) bone fragments exhibiting minimal modification (defined as ‘informal’ tools) (Klein 1989; Christensen and Goutas 2018). Bone retouchers fall into this last category, as they are not heavily modified and are often considered by-products of the breaking of long bones for marrow consumption. However, recurring technological characteristics are observed among these tools, including consistently elongated morphologies and convex surfaces actively used against the lithic edge during retouching activities (Bourguignon 2001; Mozota Holgueras 2012; Daujeard et al. 2014; Alonso-García et al. 2020; Martellotta et al. 2020, 2021). These features are necessary for the success of retouching and indicate a close relationship with the retouched lithic edge. This implies that criteria grounded on morphological, metrical, technological, and ergonomic features likely influenced bone retouchers manufacturing.

This work aims at identifying the decision-making processes involved in the intentional production of bone retouchers. We designed an experimental protocol targeting the breaking of long bones from Cervus elaphus as the primary method for manufacturing bone retouchers. Bone breaking has been extensively researched (Morlan 1984; Johnson 1985; Todd and Rapson 1988; Villa and Mahieu 1991; Bridault 1994; Alhaique 1997; Outram 2001, 2002), and it has concentrated especially on traceological identification of percussion marks on osseous remains (e.g. Blumenschine and Selvaggio 1988; Haynes 1988; Capaldo and Blumenschine 1994; Fisher 1995; Vettese et al. 2022), later enriched with an experimental component (Cáceres et al. 2002; Outram 2004; Wheatley 2008; Mozota Holgueras 2009, 2012, 2013; Jin and Mills 2011; Karr and Outram 2012a, b; Blasco et al. 2013b, 2014; Grunwald 2016; Merritt and Davis 2017; Christensen et al. 2018; Moclán and Domínguez-Rodrigo 2018; Stavrova et al. 2019; Vettese et al. 2021). Previous experimental studies have suggested that bone breakage patterns, involved in bone retoucher manufacturing, were characterised by intuitive practices (Stavrova et al. 2019; Vettese et al. 2021). While this is reasonable and supported by evidence, it is unlikely that hominins relied exclusively on intuition. Instead, they likely had a deeper understanding of the mechanical properties of bone materials in response to fractures and the role of bone morphology in the manufacturing process. This dichotomy leads to a discussion about the intentionality behind bone retoucher manufacturing—whether they were a by-product of marrow extraction or a manifestation of refined technological skills.

The debate over the intentional production of bone retouchers

The decision to explore intentional decision-making in the production of bone retouchers is driven by their high representation in the Middle Palaeolithic archaeological record. The production of bone retouchers has long been a subject of debate. While it is evident that it was linked to butchering activities, the possibility that knappers planned the production from the outset cannot be ruled out (Mozota Holgueras 2007, 2012; Jéquier et al. 2012; Abrams et al. 2014; Soulier 2014; Costamagno et al. 2018; Doyon et al. 2018, 2019; Pérez et al. 2019). The debate over intentional production of bone retouchers may stem from an outdated mindset that confines the manufacturing of ‘formal bone tools’ exclusively to the Upper Palaeolithic and Homo sapiens (e.g. Chase 1990; Noble and Davidson 1996; Klein 2000; McBrearty and Brooks 2000). Bone retouchers, on the other hand, date back to the Lower Palaeolithic and are strongly associated with the Middle Palaeolithic and Neanderthal technology. Recent approaches have increasingly highlighted the technological capabilities of Neanderthals, including their proficiency in bone technology (Soressi et al. 2013; Romandini et al. 2014; Baumann et al. 2020; Tartar et al. 2022).

In the analysis of archaeological contexts, limitations arise when attempting to ascertain whether bone retouchers were intentionally manufactured. While morphological studies outlining the shape of recovered bone retouchers can offer insights into manufacturing, this information is frequently incomplete due to post-depositional fragmentation. On the contrary, percussion marks, especially their classification and distribution, have served as primary indicators to provide insights into the manufacturing of bone retouchers. Nevertheless, recent evidence suggests that these marks can also be imprecise (Vettese et al. 2022).

Through the present experiment, we aim to reassess these indicators by incorporating the decision-making component into the study of bone retoucher manufacturing. Our goal is to understand how the objective of bone breaking—whether for marrow extraction or the production of bone retouchers—affects the decision-making processes within the production of these tools. We seek to evaluate the impact of decision-making on the percussion marks observed on the experimental retouchers and to determine whether these marks can be used to infer the purpose of bone breaking from archaeological findings.

Given the significant presence of bone retouchers in Middle Palaeolithic contexts associated with Neanderthal lithic industry, it is crucial to comprehend the manufacturing process behind these ‘informal tools’. Current research privileges the study of ‘formal’ bone tools believing that they reflect economic or symbolic importance within Palaeolithic human groups. The use of ‘informal’ tools, on the other hand, is primarily driven by day-to-day necessities, making these tools more likely to reflect individual needs (Choyke 1997).

Materials and methods

Experimental materials

Bones of red deer (Cervus elaphus) were used in this experiment based on several zooarchaeological studies that have indicated a preference for Cervidae as a source of bone retouchers (e.g. Giacobini and Malerba 1998; Patou-Mathis 2002; Jéquier et al. 2012; Mallye et al. 2012; Blasco et al. 2013a; Daujeard et al. 2014, 2018; Rosell et al. 2015; Moigne et al. 2016; Costamagno et al. 2018; Thun Hohenstein et al. 2018; Alonso-García et al. 2020; Mateo-Lomba et al. 2019; Martellotta et al. 2020, 2021). The experiment involved the breaking of 25 long bones extracted from Cervus elaphus carcasses (Table 1).

Table 1 Data regarding the long bones (N = 25) of Cervus elaphus used in the experiment. NR, not recorded
Full size table

The farmhouse La Cova del Cervo (Cassacco, Udine, Italy) supplied fresh and boiled bones extracted from two sub-adult red deer. Bones in these conditions exhibit negligible levels of degradation when compared to proper fresh bones (Karr and Outram 2012a, b), and for this reason, both samples will be referred to as ‘fresh bones’. The age of the individuals was established according to the state of fusion of epiphyses (Silver 1963, 1969). The carcasses were defleshed by professional butchers using metal cutting tools.

A second sample of bones was collected from two carcasses of Cervus elaphus recovered from a vulture feeding point in the Regional Nature Reserve of the Cornino Lake areas (Forgaria nel Friuli, Udine, Italy). The carcasses were exposed to vulture feeding in an open area during the months of May and June (dry heath conditions). The two individuals were approximately 24–36 months old at the time of death, as indicated by epiphyseal fusion stages and dental wear (Habermehl 1961; Mariezkurrena 1983).

Both samples of bones were disarticulated prior to the experiment. Remaining cartilage, tendons, grease, and flesh were removed, when necessary, using metal knives and scalpels. The periosteum was not removed as part of the cleaning process, but only when it constituted an obstacle to breaking activities. The disclosure of the freshness status of bones is included here for comprehensive coverage, but it is worth emphasising that the examination of bone freshness is not the focus of this work. Additional details on the interplay between freshness status and our sample, as well as the rationale for its exclusion, are available in Online Resource 1. Due to the limited size of the sample, statistical correlations have not been conducted on the results.

Experimental bone breaking activity

Objectives and techniques

The goals of the experimental bone breaking activity have been stated before the start of the experiment, and defined as technological ‘objectives’ (Fig. 1):

  • Objective A: the breaking activity was finalised at the extraction of the bone marrow. The conclusion of the activity was determined by the accessibility of the marrow cavity, which could be achieved either manually or with the assistance of a wooden stick. The primary focus of this approach was to prioritise the preservation of the potentially consumable marrow. Subsequently, the bone fragments resulting from the marrow extraction (by-products) were examined and handpicked to serve as bone blanks for retouchers. This objective was applied to fresh bones only, owing to the assumption that no edible marrow can be extracted from dry bones. Objective A was employed for the fracturing of 12 bones.

  • Objective B: the breaking activity was finalised at the extraction of bone blanks suitable to be used as retouchers. The process was discontinued once the diaphyseal surface of the bone had been fully utilised, and no further blanks could be obtained. This approach prioritised the extraction of blanks over the preservation of the marrow, and as such, bone fragments may inadvertently contaminate the marrow. Objective B was employed for the fracturing of 13 bones.

Fig. 1
figure 1

Description of the breaking objectives applied during the experiment. Modified from Morin et al. (2017)

Full size image

The bone breaking was accomplished by applying percussion movements, using three different techniques either independently or in combination (i.e. mixed technique) (Mourre 1996; Anconetani et al. 1995; Peretto et al. 1996):

  • Anvil-rested technique (Fig. 2a–b): the bone is placed on an anvil with only the epiphyses in contact with the surface and the diaphysis arranged to form an arch-like structure, when possible (Fig. 2a). A hammerstone is then used to strike the bone. The anvil serves as a passive element, while the hammerstone is the active means of performing the percussion.

  • Batting technique (Fig. 2c): the bone is subjected to percussion by striking it against a static anvil. In this technique, the bone serves as the active agent in performing the percussion.

  • Bipolar technique (Fig. 2d–e): the bone is positioned on the anvil, and the hammerstone is struck directly above the point of contact between the bone and the anvil. Both the anvil and the hammerstone play an active role in this technique. The force applied by the hammerstone and the resulting impact against the anvil are aligned in opposite directions, yet along the same percussion axis.

Fig. 2
figure 2

Techniques applied during the experimental breaking activity: a anvil-rested technique. The bone is positioned in order to create an arch-like structure with the anvil; b anvil-rested technique. The diaphysis of the bone is in direct contact with the anvil; c batting technique; d–e bipolar technique; f flexion movement applied to finalise the breaking of the bone following the creation of an initial fracture; g marrow extraction with the aid of a wooden stick following the partial exposure of the marrow cavity

Full size image

In some instances within both objectives A and B, the breaking activity was completed with the aid of a flexion movement exploiting an initial fracture of the bone (Fig. 2f). Once the marrow cavity was exposed, it could be extracted using a wooden stick (Fig. 2g).

The experimental protocol consisted of two sessions involving three operators—one man and two women between the ages of 25 and 28—all with variable level of expertise in Palaeolithic zooarchaeology. Each operator had six/seven fresh bones and two/three dry bones. Four limestone pebbles were used as hammerstones, while three sub-rectangular stone blocks were employed as anvils (Table 2). Throughout the experimental sessions, an observer recorded various experimental variables, including processing time, number and location of blows, number of produced bone blanks per skeletal element, individual position and gesture, the position of the bone, and feedback from the operators. The operators were assigned the task of choosing bone blanks suitable for use as retouchers from the fragments produced during the breaking activity. This selection process was guided by morphometric characteristics defined based on archaeological and experimental evidence, following general rules such as ensuring that the size of the blank allowed for an adequate grip and movement of the wrist associated with retouching, and confirming the absence of any cracking or major defects (Vincent 1993; Jéquier et al. 2012; Mallye et al. 2012; Mozota Holgueras 2012; Daujeard et al. 2014; Martellotta et al. 2020).

Table 2 Metric data and lithological description of the hammerstones and the anvils used during the experimental bone breaking activity. NR, not recorded
Full size table

Fractures and resulting bone fragments

During the percussion activity, the fracture outline (i.e. initial fracture) was recorded. Following the fracture of each bone, a thorough examination was conducted to identify the specific type of fracture that provided access to the medullary cavity (i.e. fracture planes).

Both initial fractures and fracture planes were described taking into account their surface, angle, outline, and freshness (Villa and Mahieu 1991; Outram 2001). Based on these characteristics, the fractures were then categorised into four groups:

  • Longitudinal: the fracture is parallel to the bone’s longitudinal axis.

  • Transversal: the fracture is perpendicular to the bone’s longitudinal axis.

  • Oblique: a diagonal fracture that does not curve around the shaft.

  • Helical: a spiral fracture that curves around the shaft.

  • Comminuted: a fracture consisting of multiple breaks.

After breakage of each skeletal element, the resulting products were collected and grouped together in a plastic bag with an identification code. Fragments smaller than 10 mm were recorded as debris and not counted. Bone blanks longer than 10 mm were labelled with an individual ID number. Fresh and dry bones were cleaned with a 35% hydrogen peroxide (H2O2) solution in water for 10–15 days. Cleaned bones were then left to slowly air dry before examination and storage.

Percussion marks in breaking activities

The examination of technological traces (i.e. percussion marks) resulting from bone breakage was conducted at both macroscopic and microscopic levels. All identified marks were observed with the naked eye and a portable 15 × magnification lens under directed lighting. In cases where additional analysis was necessary, low-powered magnification was performed using a stereo microscope (Leica S6D Greenough; 0.75 × –70 × magnification range). The traceological analysis focused on the identification and description of percussion marks commonly associated with bone-breaking activities. To ensure accuracy and consistency, we utilised the terminology proposed by Vettese et al. (2020), whose work includes a comprehensive glossary of these percussion marks: adhering flake, percussion notch, percussion pit, pseudo-notch, crushing mark, incipient crack, and flake. Location and distribution of the percussion marks on the skeletal elements were documented, often using refitting techniques.

Results

How the objectives impacted techniques and fracturing choices

The operators were given a clear objective (A or B) and were free to choose their approach to reach it. Throughout the experiment, operators developed individual habits and preferences related to their posture, gestures, bone positioning, and the selection of percussors (Online Resource 2). The assigned objective significantly influenced the operators’ choice of technique. Initially, they intuitively applied the anvil-rested technique to all bones and objectives. However, a quick learning process occurred as operators realised that this technique was not always effective. As the experiment progressed, operators recognised that the morphology of skeletal elements could impede achieving the objectives, especially for objective B. Consequently, they reasoned about the bones’ morphological characteristics and determined that the technique could not be applied intuitively; it needed to be reevaluated based on the specific objective and skeletal element.

This led to a change in the applied technique for each objective, resulting in a higher success rate. Objective A was typically achieved using the anvil-rested technique, while objective B involved a combination of techniques: the batting technique to separate epiphyses from diaphysis, followed by either the anvil-rested or bipolar techniques to complete the diaphysis break.

In addition, the shape of each skeletal element influenced the position of the bone impact points for both objectives. Bones with cylindrical shafts, such as humeri, tibias, and femurs, were typically broken by exploiting the arch-shaped concavity between the shaft and the anvil (Fig. 2a). This prevented slipping and increased strike precision, with the most effective impact points being the more convex areas of the bone (Fig. 3b–c). The morphology of femur allowed for various ways to create an arch, resulting in different impact point locations (Fig. 3a). On the other hand, sub-quadrangular bones like radius and metapodials were determined to be fractured by placing the bone surface in contact with the anvil (Fig. 2b). Breaking the radius proved challenging owing to the fused ulna, leading to a chaotic distribution of impact points mostly on the posterior face, between the diaphysis and distal articular portions (Fig. 3d). Operators found an effective approach by tilting the radius from the proximal edge and resting the distal edge on the anvil.

Fig. 3
figure 3

Distribution of impact points on the experimental bones: impact points chosen at the beginning of (red dots) and during (blue dots) the percussion activity (red dots); impact points which successfully resulted in the creation of a fracture on the bone (black arrows). a Femur, b humerus, c tibia, d radius-ulna, e metapodials

Full size image

Finally, the experiment revealed that the most successful approach to breaking metacarpal and metatarsal bones involved the batting technique to remove the epiphyses (Fig. 2c), followed by the bipolar technique to break the diaphysis. Operators exploited the concavity of the bone and the pointy shape of the anvil to achieve this goal (Fig. 2d–e).

Impact of the objectives on the energy investment

Number and location of blows struck during the bone breaking process serves as a measure of the amount of physical energy used in the manufacturing process for producing edible marrow (objective A) or bone fragments for retouchers (objective B). This relationship allows to quantify the energy investment in each objective, as the number of blows reflects the effort put into breaking the bones and obtaining the desired products.

Both objectives A and B required a similar number of blows, averaging 22 for objective A and 19 for objective B. As mentioned earlier, both objectives displayed a gradual learning curve. Initially, operators tend to persistently hit the same point (20–30 blows), despite its ineffectiveness, resulting in the collapse of the osseous surface and the generation of a substantial amount of debris. Such an outcome is detrimental to both objective A, where it leads to marrow contamination, and objective B, where it hinders the control of quantity, size, and morphology of the extracted bone blanks (Online Resource 3). As the experiment progresses, operators begin to shift impact points more frequently (after 10–15 unsuccessful blows), demonstrating a quick learning of the bone mechanical properties peculiar of each skeletal element, and a rapid adaptation of the applied fracturing schemes aimed at maximising the outcomes while preserving energy. The morphology of the skeletal element being processed had an impact on the number of blows required (Table 3; Fig. 3). The radii presented the greatest challenge in terms of breaking, demanding a substantial number of blows ranging from 15 to 43 per element. On the opposite end of the spectrum, the femur proved to be the least resistant, requiring a minimal number of impacts—between one and 23 blows per element. Tibiae displayed a wide variability in difficulty, necessitating between seven and 89 blows, indicative of the operators’ varying levels of experience. Humeri followed suit, requiring between five and 43 blows to break. In contrast, both metatarsal and metacarpal bones were comparatively easier to break, with a range of five to 14 blows, except for one case where it took 56 blows.

Table 3 Description of the bone breaking experimental activity. The symbol (*) indicates that the breaking activity was completed through flexion of the bone after the creation of initial fracture
Full size table

How the objectives impacted the fracture patterns

For objective A, the use of the anvil-rested technique was the most commonly used, resulting in various fractures, predominantly longitudinal (N = 3) and transversal (N = 2), followed by comminuted (N = 1), oblique (N = 1), and helical (N = 1). The combination of anvil-rested and batting techniques (N = 1), as well as batting technique alone (N = 2) both produced helical fractures. Bipolar technique was applied only in one case and in combination with the other two techniques, resulting in a longitudinal fracture (N = 1).

Objective B exhibited the choice of a more varied set of techniques, often combined, resulting in a diverse range of fractures. Anvil-rested (N = 1) or batting (N = 1) technique alone produced longitudinal fractures. When anvil-rested was completed through flexion, it resulted in oblique (N = 1) and longitudinal (N = 1) fractures, whereas batting ended with flexion (N = 1) gave longitudinal fractures. A combination of anvil-rested and batting techniques resulted in comminuted (N = 1) and longitudinal (N = 1) fractures. The bipolar technique when combined with anvil-rested (N = 1) or batting (N = 2) separately resulted in helical fractures. However, when combined with anvil-rested and batting techniques on the same bone, it led to oblique (N = 1) and transversal (N = 1) fractures.

While the fractures observed (summarised in Figs. 4 and 5) provide valuable information about the techniques used, it might be challenging to confidently infer the specific objective solely from the fractures. The fractures are a result of a combination of techniques, and the same type of fracture could potentially be produced by different combinations of techniques for different objectives.

Fig. 4
figure 4

Initial fractures: a longitudinal, b transversal, c–d helical

Full size image
Fig. 5
figure 5

Fracture planes on bones fragments following the percussion activity. a Helical. b Longitudinal. c Comminuted

Full size image

How the objectives impacted the obtained bone retouchers

Following the fracturing activity, 54 bone fragments were identified as potential retouchers based on morphometric characteristics. Among these, 22 resulted from objective A and 32 from objective B (Table 4). Objective A tended to yield a higher occurrence of bone blanks retaining a significant portion of the shaft attached to an epiphysis. Retouchers produced under objective A range in length from a minimum of 33.4 mm to a maximum of 208.7 mm, with a width ranging from 15.6 to 28.6 mm. In contrast, objective B typically yields isolated epiphyses and a greater number of blanks in various sizes. Among these, the selected retouchers appear to be larger, with lengths ranging from a minimum of 47.2 mm to a maximum of 236.9 mm, and widths ranging from 16.5 to 33.7 mm.

Table 4 Metric data and description of the bone blanks selected as potential retouchers
Full size table

Regardless of the objective, each skeletal element exhibits a predominance of retouchers obtained from a specific portion of the bone (Fig. 6). Indeed, the tibiae have the most representation on the posterior side of the diaphysis, while the humeri show predominance in the proximal and medial portions. The radius-ulnae are useful in their entirety, with a slight predominance in the proximal portion, and the metapodials have the most representation on the anterior side. Additionally, the tibial fragments tend to be longer and thicker, while femurs produce the smallest blanks. Metapodial bones yield blanks with the highest length-to-width ratio, indicating elongation, and humeri produce thinner blanks compared to other bones in the sample.

Fig. 6
figure 6

Anatomical provenance of potential bone retouchers (n = 54)

Full size image

How the objectives impacted the percussion marks

As expected, most of the impact traces typically resulting from bone breaking activities have been identified during the traceological analysis of the experimentally produced bone fragments (Table 5; Fig. 7). The majority of the impact traces is evenly distributed along the shaft circumference, especially between the metaphysis and the midshaft portion, whereas marks located around the articular portions are rare (Fig. 8). Despite conducting exploratory tests to examine the connection between percussion marks and the objective behind fracturing, no clear pattern emerged in the relationship between these two variables (Online Resource 4).

Table 5 Traceological analysis of percussion marks identified on the experimental bones following the breaking activity. The symbol (*) indicates that the breaking activity was completed through flexion of the bone after the creation of initial fracture
Full size table
Fig. 7
figure 7

Percussion marks on experimental bones: a percussion notches; b percussion pits; c crushing marks; d pseudo-notch; e adhering flake. Scale represents 1 cm

Full size image
Fig. 8
figure 8

Distribution of the experimentally produced percussion marks on the skeletal element

Full size image

Examining the individual choices made by operators during experimental sessions revealed a frequent adjustment of their technique based on the morphology of the bones. This observation suggests that bone morphology significantly influences the approach taken by operators. As a result, the analysis of percussion marks should focus on single skeletal elements that have been broken with different objectives. This approach has unveiled patterns in the distribution of percussion marks on single skeletal elements concerning objectives A or B. (Fig. 9).

Fig. 9
figure 9

Bar plots illustrating the presence of percussion marks on single skeletal elements, according to the used objective (N = 132)

Full size image

The results indicate that the presence of percussion notches on the humerus is associated with objective B, whereas on the tibia, they are more likely to be linked to objective A. Similarly, on the radius, percussion notches are more indicative of objective B. However, their distribution is equal on the femur. Percussion pits are prevalent on the radius and metacarpal bones and appear to be associated with both objectives to a similar extent. When present on the tibia, they are slightly more indicative of objective A. Flakes are strongly linked to objective A when observed on the tibia and to objective B when seen on the femur. They are also present in a certain amount on the radius, but only in association with objective B. Incipient cracks are indicative of objective B when found on the radius.

Pseudo-notches are equally distributed across both objectives. Crushing marks are predominantly associated with objective B and are totally absent on the metacarpal and metatarsal bones. When present, they are more numerous on the radius and femur, with only one instance identified on the tibia and associated with objective A. Adhering flakes are equally distributed between the objectives, but it is notable that when present on the radius, they are only associated with objective A, whereas on the metatarsal, they are only indicative of objective B.

It is important to note that these results are indicative and limited by a relatively restricted number of samples and the constraint of applying objective B solely to produce bone retouchers. Nonetheless, we believe this study provides a good foundation for proposing new approaches to the investigation of percussion marks on bone retouchers, suggesting insights for future research.

Discussion

Short-term experience and decision-making according to the technological objectives

The knappers were instructed to break the bones with one objective in mind, either prioritising the extraction of marrow or the extraction of bone blanks for making retouchers. With these guidelines in place, the knappers were free to make their own decisions on the approach to take to achieve their objective. The experimental nature of this work allowed us to observe variations in the approaches taken by the operators in the processes leading to the intentional manufacturing of bone retouchers.

In the debate over whether hominins approached bone breaking intuitively or through deliberate decision-making, our experiment introduces a complementary factor—experience gained during the task and the decisions influenced by this short-term experience. At the start of the experiment, lacking specific directions, operators initially relied on intuitive approaches, employing the anvil-rested technique for both objectives and across skeletal elements. However, this intuitive method proved inefficient, as indicated by the initial energy investment.

A learning curve emerged in both objectives, evident in the quick adaptation of technique selection based on optimal outcomes for each skeletal element and specific objective. This led to a decrease in energy investment over time. When tasked with marrow extraction, operators favoured the anvil-rested technique as the preferred approach. Conversely, when prioritising the extraction of blanks for retouchers, operators chose a combination of techniques, with the most successful being the batting technique to remove epiphyses followed by the anvil-rested technique to fracture diaphyses.

This objective-based technique selection applied to all skeletal elements, with adjustments in bone positioning on the anvil (creating an arch shape—Fig. 2a—or placing the bone surface in contact with the bone—Fig. 2b). Metacarpal and metatarsal processing for objective B involved a different combination: batting followed by bipolar technique. The natural concavity of these bones was exploited by operators, guiding the fracture for clean breaking into two suitable retouchers (Fig. 2c–e).

These results suggest that defining expertise in bone breaking should not be a polarised concept, swinging between completely intuitive and highly specialised. Such polarisation is reflected in discussions on bone retoucher manufacturing, leaning towards the idea of them being by-products of marrow extraction rather than intentionally crafted tools. Our experiment reveals a more nuanced definition, allowing for variable reasoning and rapid adaptation for better outcomes with minimal energy investment. Short-term learning and the resulting decision-making process should be integral to experimental investigations into bone breaking, whether for marrow extraction or manufacturing activities. These findings extend beyond experimental relevance. They not only reshape our understanding of bone-breaking expertise but also offer a practical tool for unravelling butchery traditions—a valuable complement to lithic studies in shedding light on Palaeolithic subsistence strategies.

Morphology of bone blanks usable as retouchers

Objective B not only yielded slightly more usable blanks than objective A, but also resulted in a greater morphometric variability among the bone blanks. The approach of prioritising the production of usable bone blanks over marrow extraction proved to be effective, leading to a higher yield of usable blanks.

Objective B offered operators the opportunity to target a larger quantity and variety of bone blanks. This was accomplished by considering the morphology of the skeletal element before fracturing and strategically planning the desired outcome of the bone fragment morphology. For instance, metacarpal and metatarsal, being elongated, regular, and possessing a natural longitudinal groove, were more suitable for obtaining elongated, wide retouchers, even though they only produced six usable fragments. Tibias produced the most retouchers (N = 23), displaying considerable size variability. Conversely, the radius and humerus produced fewer retouchers (N = 9 and 10, respectively), with decreasing size. Despite the femur being relatively easy to break, it yielded a low number of bone retouchers (N = 6), which were also the smallest in size.

These findings become valuable when examining bone retouchers found at archaeological sites. In Palaeolithic contexts, there is a noticeable pattern in the choice of skeletal elements for manufacturing bone retouchers (for a comprehensive overview of bone retoucher morphometrics in the Middle Palaeolithic across Europe, refer to the discussion in Martellotta et al. 2020, 2021, and references therein). This trend often involves a significant number of bone retouchers identified as tibia fragments, while there is consistently a low count of retouchers originating from the radius, humerus, and femur.

According to the results in our experiment, the patterns observed in archaeological sites can be attributed to two likely, possibly interconnected, reasons. Firstly, the experiment demonstrated that even after selecting the optimal technique for each skeletal element to produce bone retouchers, there remains a varying level of difficulty in breaking each bone while preserving the bone surface sufficiently for extracting usable retouchers. This difficulty scale is discerned from impact points observed during the experiment, reflecting the energy invested. The radius emerges as the most challenging bone to break, followed by the tibia and femur, which are relatively easier. Metacarpal and metatarsal bones are highly breakable once the right technique combination is found, with the femur being so easy that sometimes only a few blows are needed to extract retouchers.

Secondly, the morphology of the bone plays a crucial role. The radius, being hard to break and having an angular morphology, limits the availability of usable convex surfaces for retouchers. Similarly, the femur, although easier to break than the radius, poses challenges in obtaining suitable bone blanks for retouchers due to the significant portion occupied by epiphyses and associated spongy tissue. The shaft, the only part suitable for retouchers, is relatively small, and the cylindrical shape lacks adequate convexities. Breaking the femur requires hitting the mesial part, reducing the available bone surface for extracting usable blanks. Tibia, with its numerous convex surfaces, proves to be ideal for crafting retouchers, explaining why a high number were extracted from it in our experiment. This aligns with similar observations in archaeological contexts. Despite the humerus being the easiest bone to break in our experiment, its reduced shaft size and cylindrical shape might account for the scarcity of bone retouchers from this bone in archaeological sites.

These findings highlight differences in the exploitation of long bones across various skeletal elements and support the notion that hominins may have had preferences for certain skeletal elements based on their technological objectives. Questions around metacarpal and metatarsal, however, remain unanswered. Despite the ease of extracting retouchers from these bones through the combination of batting and bipolar techniques, archaeological sites show a scarcity of identified bone retouchers originating from metacarpal and metatarsal. Exploring the relationship between archaeological bone retouchers extracted from these bones, when available, and the associated lithic tools in archaeological sites would be intriguing. Our experiment revealed that these bones produced a distinct shape of retouchers—larger in size, very elongated, and with a relatively less convex surface compared to other bone retouchers from different skeletal elements. This specific shape may not be highly functional for producing a diverse range of retouched edges, potentially making it less preferred in manufacturing. On the other hand, an argument could be made that these bones are optimally exploitable only through the bipolar technique.

Percussion marks: a new approach

Percussion marks are commonly used as a proxy to identify bone fracturing techniques linked to butchering or manufacturing, or both activities (Blasco et al. 2013b; Masset et al. 2016; Vettese et al. 2017; Moclán and Domínguez-Rodrigo 2018). When analysing percussion marks on bone retouchers, it is essential to move beyond mere mark counting and morphological description to extract the maximum amount of data regarding manufacturing techniques and intentions.

In our experiment, objective A generally exhibits fewer marks compared to objective B, which shows a higher number of marks distributed across various skeletal elements. From our observations, we propose that the presence of percussion marks on bone retouchers might indicate the intention behind the percussion—whether for marrow extraction or retoucher manufacturing—especially when considered in relation to the specific skeletal element. The morphology of the skeletal element strongly influenced the decision-making process for achieving either objective while maintaining an optimal balance with energy investment. While the anvil-rested technique is the preferred method for marrow extraction, manufacturing bone retouchers requires a combination of batting and anvil-rested techniques for most bones. However, for metacarpal and metatarsal bones, the anvil-rested technique was replaced by the bipolar technique. This difference in technique usage resulted in varying occurrences of percussion marks on different skeletal elements depending on the objective. In the archaeological analysis of bone retouchers, the occurrence of percussion marks not only indicates retouchers resulting from marrow extraction but, when analysed alongside the skeletal element they originate from, can also suggest the intention behind their production.

In future research on analysing percussion marks on bone retouchers, considering and controlling the variable of skeletal elements could enhance the conclusiveness of results. This factor strongly influenced the current experiment, shaping the choice of technique and subsequently impacting energy investment. The sole variable accessible at an archaeological level is percussion marks. Therefore, we recommend a specialised investigation of percussion marks on archaeological bone retouchers, particularly those where the skeletal element can be identified. This focused analysis aims to provide insights into the intentions behind their production, determining whether they were mere by-products of butchery activities or involved intentional technological processes. Our experiment suggests that certain marks on specific skeletal elements are more indicative than others of the intentional production of bone retouchers, as highlighted in Table 6.

Table 6 The table serves as a grid for evaluating the reliability of percussion marks in deducing the objective—either A (marrow extraction) or B (bone retouchers manufacturing). This assessment is made on each skeletal element processed with the two objectives. Green—highly reliable for inferring the objective. Yellow—moderately reliable for inferring the objective. Orange—less reliable for inferring the objective. White—highly unreliable for inferring the objective. Gray—ambiguous for inferring the objective (presence of the percussion mark results in high reliability for both objectives)
Full size table

The presence of percussion notches on metatarsal, humerus, and radius can reliably indicate the intentional production of bone retouchers, as these marks are absent when the objective is solely marrow extraction. The same holds true for percussion pits on metacarpal and humerus, as well as incipient cracks on the radius. While other marks occur less consistently, they may still signify the objective, such as crushing marks seen with intentional approaches on humerus and radius, and to a lesser extent, on the femur. Adhering flakes, appearing indicative of objective B only on metatarsal, also contribute to understanding the purpose. Notably, percussion notches are the least reliable marks for investigating the manufacturing process of bone retouchers from femur, metacarpal, and tibia, as they appear in similar amounts in both objectives A and B. However, they remain highly reliable on metatarsal, humerus, and radius.

We acknowledge that conducting the experiment with a larger sample size, coupled with statistical analyses, could enhance these proposed criteria for interpreting percussion marks on bone retouchers in Palaeolithic contexts. Nevertheless, we believe our study holds significance and contributes valuable insights to the understanding of percussion marks on bone retouchers in Palaeolithic contexts. It provides a new direction that can be further explored in future research, offering fresh insights for researchers exploring percussion marks in Palaeolithic contexts, and specifically in relation to bone retouchers.

Conclusion

The study of bone retouchers raises questions regarding their acquisition, typically achieved through three approaches: (1) opportunistic use of butchering waste products, where bone retouchers are not the primary technological objective; (2) deliberate recycling of butchering by-products, with the aim of obtaining retouchers, albeit not the primary goal of bone breaking activities; and (3) intentional production of bone blanks for retouchers, not necessarily subordinate to marrow exploitation. Our experiment delved into operators’ decision-making during bone breaking for two specific objectives: marrow extraction and retoucher production. Initially relying on intuitive approaches, operators adapted over time, revealing a learning curve in technique selection and decreased energy investment. This challenges the polarised view of bone-breaking expertise, proposing a nuanced definition that incorporates variable reasoning and rapid adaptation.

The experiment’s main result highlighted the pivotal role of skeletal element morphology in retoucher manufacturing. Operator techniques and approaches varied based on bone morphology, resulting in distinct techniques when prioritising marrow versus retoucher manufacturing. These differences are reflected in the analysis of percussion marks. While typically used to associate retouchers with marrow extraction, our experiment revealed that percussion marks are linked to the operator’s intention, whether for marrow extraction or retoucher manufacturing. Our results advocate for a specialised investigation of percussion marks in archaeological contexts, stressing the importance of considering specific skeletal elements for conclusive results.

This work underscores differences in long bone exploitation and suggests a potential link between retoucher shape and techniques used, providing insights into hominin preferences and capabilities. The experiment contributes valuable insights into bone-breaking expertise, retoucher manufacturing, and percussion mark interpretation in Palaeolithic contexts, laying the groundwork for future research and emphasising the need for a nuanced understanding of hominin decision-making in these activities.