Although we know that our lineage has been producing sharp-edged tools for over 2.6 Myr, our knowledge of what they were doing with these tools is far less complete. Studies of these sharp-edged stone tools show that they were most probably used as cutting implements. However, the only substantial evidence of this is the presence of cut marks on the bones of animals found in association with stone tools in ancient deposits. Numerous studies have aimed to quantify the frequency and placement of these marks. At present there is little consensus on the meaning of these marks and how the frequency relates to specific behaviours in the past. Here we investigate the possibility that mechanical properties associated with edges of stone tools as well as the properties of bones themselves may contribute to the overall morphology of these marks and ultimately their placement in the archaeological record. Standardized tests of rock mechanics (Young's modulus and Vickers hardness) indicate that the hardness of tool edges significantly affects cut-mark morphology. In addition, we show that indentation hardness of bones also impacts the overall morphology of cut marks. Our results show that rock type and bone portions influence the shape and prevalence of cut marks on animal bones.
It is well understood that many primates augment their diet through the use of extractive foraging . Although there are clear indications that phylogenetic relationships between primates do not predict propensity of tool use , it is likely that the last common ancestor between humans and chimpanzees used some type of tools to gain access to resources otherwise unavailable . It has long been known that Plio-Pleistocene hominins fashioned sharp-edged stone tools to cut meat and tissue off the bones of animals in deep antiquity [4,5]. Recent discoveries in the field of palaeoanthropology have turned increased attention towards evidence of ancient hominin (the taxonomic group that includes modern humans and extinct ancestral and related species) diet. For example, the appearance of cut-marked bones at the site of Dikika (Ethiopia) found in sediments that are probably older than 3.2 Myr represents a significant departure from previous interpretations of hominin tool use [6,7]. Evidence of changes in the diet of early hominins is further supported by evidence of increased variability in dietary signatures from isotopes in dental enamel that may be linked to trophic level changes [8,9]. The evidence from Dikika derives from a few fossil bones that exhibit marks on the surfaces of the bones that have been interpreted as evidence of butchery by a member of the genus Australopithecus. The marks on the surfaces of the bones at Dikika are substantially different in morphology from cut marks found in other archaeological contexts . This has prompted renewed interest in the sources of variation in cut-mark morphology .
These new investigations regarding cut-mark morphology have sparked debate regarding the inferential link between cut marks on the surfaces of these bones and hominin diet [10,11]. The majority of this debate involves a discussion of the shape and size of linear marks and how the variability in these forms affects the ability of archaeologists to diagnose the actor (e.g. carnivore versus hominin) and effector (e.g. stone tool edge versus carnivore tooth) that produced these marks . For example, abrasion on the surfaces of bones from sediment being dragged across the surface of a bone had been known to create mimics of cut marks . Similarly, the sharp tips of crocodile teeth create marks on bone surfaces that superficially resemble cut marks [14,15]. Marks inflicted by the edges of stone tools are usually diagnosed based on their cross-sectional shape (more V-shaped than U-shaped) and their association with internal linear striations . However, these qualitative criteria are proving to be inadequate to diagnose less conspicuous marks. Recent reviews have noted that, while archaeologists tend to classify the marks on the surfaces of bones into categories (e.g. cut mark, tooth mark) based on their assumed actor, the reality is that there is substantial variability in the form of marks made by the same or similar processes in the past. Initial discussions about the use of marks on bone surfaces as a means to interpret behaviour in the past warned about the necessity to understand the mechanism through which traces of behaviour are recorded on the surfaces of bones . This is critical to understand how hominin diet changed through time because the morphology of cut marks may be able to provide information about the timing of access to carcasses . While palaeoanthropologists have come a long way in answering this call to relate static traces of the past to the dynamic processes that created them we remain largely unaware of the mechanical factors that affect the morphology of bone surface modifications . A more detailed understanding of the nature of cut-mark formation is necessary because numerous biostratinomic factors are likely to have an influence on the mechanical properties of bone. Archaeologists interested in dietary changes must understand the interface between the mechanical properties of bone and stone and the resultant cut-mark morphology if they are to make informed inferences about behaviour from cut-mark morphology.
A recent blind test showed that even among trained zooarchaeologists correspondence between analysts when diagnosing the presence of cut marks is dramatically affected by the mechanical properties of the bone surfaces . This highlights the fact that marks on bone surfaces are extremely variable  and may not be easily attributed to a specific actor or process. In any given experimental or archaeological assemblage there are marks made by different actors that overlap substantially in their morphology . In the absence of contextual data, the morphology of cut marks and linear abrasions made by sediment can produce linear striations that can appear similar . If archaeologists are to use bone surface modifications to interpret aspects of ancient human behaviour they must begin to address the basic mechanical details of how marks on bones form.
Marks on the surfaces of bones can be caused by a sharp object being drawn across the surface of a bone, effectively creating a groove on the surface of the bone. The morphology of cut-marked surfaces has the potential to provide archaeologists with important information about the timing and nature of hominin access to high-quality resources in the past. The morphology of marks on bone surfaces is influenced by multiple factors. The elasticity and hardness of the bone surface will influence the resulting morphology of the groove. These properties will vary between and within bones . As a result the same process that occurs on bone surfaces of different mechanical properties will result in marks with substantially different morphologies. Different bone portions will have varied mechanical properties due to the differences in bone mineral density that can vary within a bone [19,20]. Furthermore, the mechanical properties of a bone are heavily influenced by how long an animal has been dead. Bones of animals that recently died react to plastic deformation differently from bones of animals that have been dead for several days . This is a distinction that zooarchaeologists refer to as ‘fresh’ versus ‘dry’ bones. The mechanical properties of bones are clearly an important factor, but this is only half of the process of cut-mark formation. The morphology of the artefact being drawn across the bone also influences the shape of the mark made on the bone surface . The edges of tools can vary dramatically, and furthermore, these edges will change their shape as they become dull through use . The attritional damage (i.e. dulling) on the edges of tools is related to the mechanical properties of these tools . Thus, any real understanding of the influences on the morphology of cut marks needs to take into account the mechanical properties of the specific portion of the bone where the mark was made as well as the mechanical properties of the stone edge producing that mark. The majority of these differences in bone portion and condition relate to hardness, which is a measure that is usually related to the resistance to deformation of a bone surface and is affected by numerous external conditions (porosity, density )
Here we develop a series of experiments designed to identify and quantify the impact of a variety of different physical properties of both bone and stone on cut-mark morphology. Our experiments are not intended to mimic any particular butchery action. Instead, we hope that by keeping certain variables constant (e.g. stone tool edge angle, bone portion) that it is possible to identify the influences of these different factors in the formation of marks on the surfaces of bones. We isolate variations in these independent variables using standardized engineering tests of natural materials. We then apply three-dimensional techniques to investigate the variation in morphology of cut marks relative to these properties. We predict that stone artefacts with different mechanical properties will exhibit different degrees of edge dulling depending on the hardness of the stone as well as the properties of the bone that is being marked. We also predict that these differences will be reflected in the three-dimensional micro-morphology of the cut marks. As archaeologists begin to understand the influences of mechanical properties on the marks made on bone surfaces, it may be possible to begin to explore the morphology of cut marks in ways that reflect important details of ancient behaviour.
2. Material and methods
This study represents a series of tests designed to understand the relationship between the mechanical properties of stone and bone and the morphology of cut marks. Subsequently, we apply these tests to an investigation into the influence of these mechanical properties on cut-mark morphology. Ultimately, we are interested in how the mechanical properties of these different materials affect the morphology of cut marks on bone surfaces. We used a standardized portion (2 cm of edge) of simple stone flakes to produce cut marks on dried cow bones. Dried cow femurs were used because of the possible effects of water content on bone mechanical properties . The resulting cut marks were analysed by developing high-resolution three-dimensional models of the cut-marked surfaces.
The mechanical properties of stone used to make chipped stone artefacts have received remarkably little attention. Early investigation focused on how the mechanical properties of stone can be affected by heating of stone to make it more amenable to artefact production [27,28]. Subsequent investigations investigated how different mechanical properties impact the fracture mechanics of stone . Recent investigations have identified certain mechanical properties that relate to the ability of the edges of tools to resist degradation during use [24,30]. This study is interested in the effect of edge dulling on tool edges and its effect on the resultant morphology of marks made on bone surfaces. As such, we were interested in a test that would focus on the robustness of tool edges. Tsobgou & Dabard  investigated a similar property on tool edges as a means for understanding the edge strength of phtanites. Their analysis used a combination of Vickers hardness and Young's modulus of elasticity to develop a measure they described as an index of edge performance and was based on work by Ashby et al. . We expect that rocks that have high values for this ratio will be less susceptible to changes in the morphology of edges as they are dulled. Edge dulling is likely to be the result of a variety of factors including wear, and chipping of the edges. We have assumed that this measure of edge performance provides an overall assessment of these different factors. We expect this change in edge morphology will be reflected in the cross-sectional shape of cut marks made by tools that have been subjected to more severe levels of edge attrition (i.e. edge dulling).
We investigated the mechanical properties (Young's modulus and Vicker's hardness) of two rock types that are frequently used in many archaeological collections: flint and basalt. Flint is a highly siliceous fine-grained stone found in many Middle Palaeolithic (300 000 years ago until 20 000 years ago) contexts throughout western Europe (e.g. ). Basalt is a fine-grained volcanic rock that has a higher frequency of mafic minerals (e.g. olivine) than most volcanic rocks used by hominins to make stone tools. Basalt was used to make stone artefacts in many archaeological sites in East Africa during the early part of the Stone Age (2.6 Ma until 400 000 years ago, e.g. ).
We measured Vicker's hardness using a Proceq Equotip 3 hardness tester on cut and polished slabs of stone. This tester uses a carbide indenter to measure the response of the rock surface to a dynamic impact. This tester then calculates standard Vicker's hardness (Hv) values using this data (we converted Hv to megapascals (MPa) values for the calculation of the ratio as described by Tsobgou & Dabard ). We analysed 20 separate samples of each stone to ensure that we calculated the full range of variation in each rock type. Samples were mounted on a 10 cm thick slate lab bench during testing. All samples were tested on polished clean surfaces. We measured Young's modulus of elasticity using an ultrasonic tester that uses the pulse-echo technique. This technique uses an ultrasonic transducer to generate an ultrasonic pulse and then a receiver registers the ‘echo’ of this pulse at a certain velocity given a known thickness of the specimen. The velocity of the pulse allows the calculation of the elastic properties of a specimen. We measured 20 samples of each rock to explore the full variation of these materials.
To simulate edge attrition, we created multiple cut marks on a bone surface under the assumption that drawing a tool edge across the surface of a bone creates damage to the tool edge that will result in changes in tool edge shape . We created 18 simple stone flakes of basalt and flint (e.g. chipped pieces of stone with a single continuous sharp edge around their circumference). The edges of the stone flakes were measured to standardize edge angle. Using the method described in Dibble & Bernard , we measured the edge angle of all flakes. Only edges that had a 40-degree edge angle were selected for the experiments.
We made four experimental cut marks on specific bone surfaces (see below for description of bone surface samples). After the production of each experimental cut mark the tool edge was drawn across a separate cortical bone surface a series of times to simulate the dulling of the edge. After the first experimental mark was made the tool edge was drawn across a cortical bone surface 10 times. After the second experimental mark was made the edge was drawn across a cortical bone surface 20 times. After the third experimental mark, the edge was drawn across a cortical bone surface 50 times. Finally, a fourth experimental mark was made. This resulted in four separate experimentally made cut marks for each experimental bone portion. The first cut mark, in each experimental set, represents the tool edge prior to any dulling. The second mark represents the edge after the dulling of the tool edge caused by 11 strokes of the tool across a bone surface. The third mark represents the tool edge after the edge was used to cut across a bone surfaces 32 times (the 30 cuts produced on a separate cortical bone surface and 2 experimental cut marks). The fourth experimental mark represents the maximum amount of tool edge dulling from 83 strokes of the tool edge across a bone surface. Multiple replicate bone portions were tested to replicate these effects with different stones. We expect that after each bout of drawing the edge of the tool across bone surfaces multiple times that the edge of the tool will dull. This should be reflected in the morphology of the marks that are left on the bone surfaces.
To test the effect of variation in mechanical properties of different bone portions, we divided the bone into three portions (figure 1). These portions included the epiphyses (the area where the femur articulates with the innominate on one end and the tibia on the other); the diaphysis or shaft of the bone (where the marrow cavity of the bone is located) and the near epiphyses (the intersection between the epiphyses and the diaphysis). The femur was cut into 18 separate fragments evenly divided into the three above categories. The bone portions were segmented both longitudinally along the long axis of the bone as well as transverse to the long axis. Epiphyseal fragments were segmented to maximize a flat portion of epiphyseal bone. All the experimental epiphyseal fragments were articular surfaces. This was necessary to standardize the mechanical tests of the bone surface. As cut marks result in the plastic deformation of the bone surface we tested each of these fragment using a Vicker's hardness (Hv) tester. Each surface was tested 10 individual times to create a composite score of the hardness of each surface. We expect that the variation in the hardness values of the different bone portions will be reflected in differences in the morphology of the marks made on the surfaces.
2.3. Cut-mark morphology
Archaeologists have previously measured the overall two-dimensional morphology of cut marks, yet the extremely varied nature of these marks makes it difficult to compare these measurements . In recent years, the use of high-resolution microscopy has been employed to create three-dimensional surface models of cut marks . These models can then be used to identify distinctive features of these marks that are not readily visible in two-dimensional representations of the plan view of cut marks. Capturing the full three-dimensional morphology of cut marks has still proved rather difficult .
Here we use a methodology that uses cross-sectional analysis of the three-dimensional models to identify generalized patterns of shape that are easily quantifiable. Standardization of cross-sectional measurements is complicated because of the large amounts of variability that exist across the surface of a cut mark. We chose to follow the methods outlined in Archer & Braun  because of the ease of calculation of these variables and the ability to consistently apply them to a range of different surface morphologies. Data were collected from all cut marks using a Nanovea ST400 white-light non-contact confocal profilometer that allows for the development of three-dimensional models of the cut-mark morphology at a resolution of 40 nm on the z-axis. Profiles were taken every 10 µm with a 5 µm sampling interval.
Measurements are taken from cross sections of the model and analysed in the image analysis software ImageJ. The cross sections of the mark were evenly spaced along the length of the mark. In some instances, the cross sections were not exactly evenly placed because complications of the surface morphology of the bone prevented clear measurements. Location of these cross sections was focused on locations that characterized the variation in the three-dimensional morphology of the marks. These measurements include four separate variables that focus on different aspects of shape variation (figure 2, ). All these measurements (except for the one angle measurement) are calculated as a ratio of two measures of shape. This was done to reduce the possible influence of size on different variables. Although size-related shape change (i.e. allometry) is an important aspect of the variation in cut marks, we focused on ratios of linear measurements to reduce the chance that subtle shape changes would be overshadowed by greater amounts of variation that are the result of size-related differences in cut-mark morphology. The first measurement is the breadth of the cut mark divided by the depth of the mark. This measurement describes the relative width of the opening of the mark relative to the depth of the mark. The second measurement is the ratio of two breadth measurements. The first is taken at 25% of the depth of the cut mark and the second is taken at 75% of the depth of the cut mark. This measure describes the degree to which the walls of the cut mark are parallel. Marks that are highly V-shaped tend to have high values for this variable. The third measurement is an angle measurement between the two walls of the cut mark. These three measurements were taken at three separate locations along the length of the cut mark. Values for each variable are averaged for each cut mark. This was done for each experimental cut mark for each bone portion.
We first report on the mechanical testing documenting the variation in mechanical properties within the different types of stone as well as bone portions. Subsequent analysis will focus on the effect of this variation on the morphology of the cut marks produced.
Results of the edge toughness measurement show that even though the values for edge toughness for flint are extremely variable (coefficient of variation: 55.1), the values for flint are significantly higher than they are for basalt (Mann–Whitney U-test U: 31; p = 0.006). Although there are some extremely tough outliers among the flint specimens the basalt specimens had consistently lower toughness values (figure 3).
The different bone portions varied substantially in their overall Hv values (figure 4). Epiphyseal fragments had the highest values and also had the highest variance in values (median: 136.4; CV: 46.9). This was followed by the diaphysis (median: 89.49; CV: 26.5) and then finally the near epiphyseal fragments which had the lowest values (median: 65.2; CV:24.8). Statistical tests showed that although epiphyseal fragments had higher values they were only significantly greater than the near epiphyseal fragments (Tukey's pairwise Q: 4.4; p = 0.018). Differences between fragments of the diaphysis and epiphyses were not significantly different (Tukey's pairwise Q: 2.99; p = 0.119). The differences between the shaft of the bone (diaphysis) and the near epiphyses were also not significant (Tukey's pairwise Q: 1.4; p = 0.591). The substantial variation in hardness values suggests these different portions of the bone should react differently to cut-mark production.
3.3. Cut-mark morphology
We investigate cut-mark morphology relative to the two main predictions based on the mechanical tests described previously. The first prediction is that bone hardness values of the different portion of the bone will correlate to some measure of the cross-sectional shape of cut marks. The second prediction is that stones with lower edge toughness values (i.e. basalt versus flint) will show greater attrition during the production of cut marks and as a result the cut marks that these tools make will look different.
Table 1 provides a summary of the variation in the cross sections of cut marks across bone portions. This summarizes the variation in the measurement of cut-mark morphology cross section for the different bone portions. There is substantial variation within each bone portion for each measurement. However, if we directly compare bone portion Hv values with the V-shape measurement (an indication of how parallel the walls of the cut mark are), we see a direct relationship (Kendall's τ: 0.55; p = 0.001; figure 5). Here we use Kendall's τ as a measure of correlation because of the non-parametric nature of the data used in this dataset. We should note that the denominator in V-shape variable is the width of the mark at the base of the mark. As such higher V-shape values in figure 5 refer to less V-shaped marks. Bone surfaces with lower hardness values have lower values for this ratio and, therefore, have walls of the cut marks that are less parallel.
Investigations into the effect of stone edge toughness require a comparison of tools at different stages of edge attrition. Although there was some variation in the width to height variable as well as the V-shape variable neither of these showed significant differences between cut marks made by basalt and flint flakes. A pairwise comparison between flakes at similar stages of edge attrition shows a significant difference in the angle of the cut mark opening between cut marks made by basalt flakes versus those made by flint flakes (Wilcoxon W: 247; p = 0.004). This is further seen in a plot of the angle variables relative to the different stages of edge attrition for the different raw materials (figure 6). The relationship between the number of strokes a tool has undergone and the angle of the cut mark opening is best explained by a logarithmic function (e.g. linear function Pearson's r = 0.49; logarithmic function Pearson's r = 0.30). As these data are not appropriate for parametric tests of association (i.e. Pearson's r), we use the non-parametric correlation coefficient of Kendall's τ (basalt: Kandall's τ = 0.46; p < 0.001; flint: Kendall's τ = 0.68; p < 0.001).
The results of this study suggest that the mechanical properties of both stone edges as well as bone surfaces influence the overall morphology of cut marks. Bone surfaces that have lower hardness values tend to have marks where the walls of the cut mark were more parallel. This may be a result of the tool edge moving laterally within the mark during the production of the cut mark, creating walls of the cut mark that are more parallel than in harder bone surfaces. Previous investigations of the variation in bone hardness found similar patterns . This suggests that we should expect to find variation in the cross-sectional morphology of cut marks at different portions of the bone. This also suggests that diagnosing which type of tool a certain cut mark was made by is going to require more detailed information about the butchery process . Comparison between cut mark morphology will need to be understood relative to which portion of a bone is being investigated.
Furthermore, it is clear that the edges of tools undergo substantial change during butchery processes. The attrition of tool edges during butchery is something that has previously been highlighted [16,23,24]. Here we show that this attrition is related to the opening angle of a cut mark. As tool edges begin to dull, the angle between the walls of a cut mark begins to open quickly. This relationship seems to level off after an initial bout of edge attrition (figure 6). The relationship between edge angle and tool attrition is meditated by the mechanical properties of the tool edge. Thus, the ability to identify which type of tool has been used to create a specific cut mark will require a detailed understanding of the materials that were used to make the mark, as well as the approximate stage of the butchery process.
This study attempted to control the various variables that may influence the morphology of cut marks. Although substantial controls were implemented (edge angle, raw material, etc.) there are probably further variables that need to be taken into account such as butchery action (skinning, defleshing and disarticulation) and the amount of flesh present during the butchery (bulk versus scrap). Previous attempts to study the morphology of cut marks have focused on the angle between the edge of a tool and the surface of a bone . The assertion is that that tools that drag along the edge of a bone at a high angle relative to the bone surface will create different marks compared with the scenario where the edge of the tool is directly perpendicular to the bone surface . In our experiments, we could not control for this variable, although it clearly influences the overall morphology of a cut mark.
Furthermore, we noted that even though the bones we were using were dry, the epiphyses still had substantial amounts of grease within them. This is probably related to the anatomy and structure of mammal bones that tend to have large amount of fats stored in the cancellous portion of the bones (usually found in the epiphyses; ). The variation in grease content of the bones may certainly have affected the mechanical properties of the bones. Our experimental dataset visually varied in the appearance of grease that appeared to be in the bone. It was not possible for us to quantify this variable but the hardness differences between epiphyseal and diaphyseal fragments may reflect this. It is clear that classifying bones as ‘fresh’ or ‘dry’ probably masks more nuanced variation that is important in the study of cut marks.
The most enduring record of human behaviour is the presence of stone artefacts that extend back millions of years . However, potential changes in the diet of early hominins are recorded in marks on bone surfaces that potentially presage the appearance of abundant chipped stone artefacts . These marks on bone surfaces reflect possible changes in the trophic level of our early ancestors. Currently our knowledge of how these marks are formed is based on personal observation and numerous experimental studies that sometimes favour realism over control of variables . However, the archaeological record continues to document the fact that the presence and location of marks on bone surfaces are extremely variable . Similarly, the morphology of cut marks appears to vary quite extensively [22,37]. A solid inferential framework about the behavioural meaning of marks on bone surfaces will require a more detailed understanding of the factors that influence the morphology of these marks.
This study provides details of how mechanical properties of stone edges as well as the surface of bones have an impact on the resulting cut-mark morphology. It was previously understood that bone surface properties impacted surface modifications [17,19]. This is easily taken into account when analysing cut marks (as long as the bone portion is often recognizable on the archaeological specimen). However, the influence of artefact mechanical properties is something that will be more difficult to ascertain from archaeological materials. Certainly, in localities where many of the stone artefacts are made from alkali volcanic materials (e.g. Koobi Fora, Kenya ), we should expect to find substantial differences in the morphology of cut marks compared with that seen in places where artefacts are made of much tougher materials (e.g. flint at Boxgrove, UK ).
Investigations of the mechanical properties of archaeological materials show great promise for providing information on the patterns of variation seen in the archaeological record. More investigation is needed as to which mechanical properties are the most appropriate for answering archaeological questions. Collaborations across disciplinary lines provide the opportunity for new insights into ancient human behaviour .
We declare we have no competing interests.
M.C.P. acknowledges generous funding provided by Ann Gill, Dean, College of Liberal Arts, Colorado State University that allowed for the purchase of the non-contact profilometer used in this study. D.R.B. acknowledges funding from the George Washington University Center for the Advanced Study of Human Paleobiology.
M.C.P. acknowledges the contributions of Trevor Keevil and Matthew Muttart in the collection of three-dimensional data used here. D.R.B. acknowledges the detailed work of Ella Beaudoin, who ensured that the data collected for these experiments were finished in timely fashion.
One contribution of 14 to a theme issue ‘Cutting science in biology and engineering’.
- © 2016 The Author(s)
Published by the Royal Society. All rights reserved.