Charles P. Egeland and Ryan M. Byerly

From Cutmarks to Behavior:
The Reliability of Cutmarks for Inferring Processing Intensity

INTRODUCTION

            Cutmarks are the most obvious and abundant class of archaeologically visible results of the butchery process.  What is not so obvious is the dynamic set of circumstances that contribute to what we observe as tiny scratches on the surface of a bone.  Implicit in the study and interpretation of cutmarks is the assumption that these tiny scratches represent and are influenced by various butchery activities and the energy expended to create them.  As safe as this assumption appears, the nature of this relationship remains mysterious.  How do the frequencies and locations of cutmarks relate to the activities and energy outputs that created them?  Studies concerning patterns in the frequency and placement of cutmarks in archaeofaunal (Bunn and Kroll 1986, Frison 1970) and ethnoarchaeological (Binford 1978, 1981; Dominguez-Rodrigo 1997) assemblages have laid the groundwork for investigations asking this question.  This study is the first in a series of investigations explicitly addressing these questions by way of an analysis of a sample of experimentally butchered carcasses.  The nature of energy outputs and the influence of these outputs on cutmark frequencies and percentages are the concentration of this particular research.  This study will hopefully serve as a jump-off point for further studies focusing on the complex process of butchery and its prehistoric manifestations.  Future investigations will integrate data gained from this study with precise location data and archaeological data.  Because the ultimate purpose of any archaeologically-related experimental study is to asses its degree of applicability to real archaeological data, this study represents the first step towards a synthesis aimed at plugging primary cutmark data into a larger context of butchery processes, site function, seasonal land-use strategies, and group social organization.

Experimental Butchery

Experimental databases have for the most part been ignored in studies of this kind on the basis of the researcher’s unfamiliarity with the act of stone-tool assisted butchery.  It may be assumed that the inexperience of the researcher lends the data gleaned from such studies unhelpful in assessing functional correlates of archaeologically observed cutmarks.  We hope to demonstrate in this study that experimental butcheries can contribute extensively to our understanding of the dynamics involved in the process of reducing and modifying an animal carcass into humanly transportable and consumable parts.

In contrast to observer-based ethnographic studies, an experiment such as that presented here allows the researcher the opportunity to come directly into contact with the actual process of butchery.  In addition to granting personal insights, experiments of this sort allow the researcher to conceptualize in concrete terms the decisions encountered by prehistoric butchers.  From creating tools to cutting meat into transportable packets, this study offers a unique glimpse into the gamut of problems presented by reducing a carcass into manageable parts. 

Presented here is a preliminary study whose purpose is to critically examine the nature of the relationship between the energy expended to butcher a carcass and the cutmarks that are created as a result of that expenditure.  Namely, does a greater amount of processing intensity necessarily lead to increased numbers of cutmarks?  If not, is it possible to describe the influence of a multitude of variables on the creation of cutmarks?

METHODS

Between 4 November 1999 and 12 October 2000, fifteen individual butcheries were conducted by the authors and colleagues at the Laboratory of Public Archaeology (LOPA) in Ft. Collins, Colorado.  Experimentally butchered animals consisted of horse (Equus caballus) and cow (Bos taurus). With the exception of one case, all butcheries were limited exclusively to leg units (Table 1).  One of the episodes included the butchery of a frozen carcass (Table 1).  Among the data collected were meat weights, strokes per element, and strokes per activity.  A full list of all data collected is presented in Table 2.  During the first eight butcheries strokes were counted manually by making marks on a piece of paper.  Starting with the ninth butchery, a counter was used to document the strokes. 

All bones were completely skinned and defleshed during each episode.  Studies claiming that tendon removal motivated some butchery activities (Potts and Shipman 1981) encouraged us in episodes BE-3 through BE-9 and BE-11 to include the removal of the superficial and deep digital extensor tendons and the digital extensor tendon.  Meat weights for each element were determined by straight cuts along the articular planes of adjoining elements.  An experience point, beginning at one, was awarded after each subsequent butchery performed by a butcher.  These points were meant to supply a relative measure of experience and its potential impact on butcher performance as reflected by time, strokes, etc.  Because the purpose of this aspect of an ongoing study is to ascertain the nature of the relationship between energy expenditure and the visible cutmarks, a measure for energy expenditure is necessary.  For this study, the documentation of strokes and the recording of time are used here as measures of energy expenditure during butchery.  All tools were made by the researchers and consisted of unmodified obsidian and chert flakes (Figure 1).  Butcheries were conducted by one person only unless otherwise noted.  Following butchery, bones were placed in boiling water for approximately 24 hours (depending on the size of the specimen; smaller or younger individuals require less time), after which they were placed in a peroxide solution for another 24 hours (again depending on size). 

Cutmark detection for the experimental sample was conducted under strong direct light and, when required, with the aid of a hand lens; no microscopic techniques were implemented.  Although microscopic techniques are advocated by some workers (Potts and Shipman 1981, Shipman and Rose 1983), we did not feel it necessary to include these methods for reasons related to the fact that experimentally controlled butchery alone could account for surface markings.  In addition, studies have been published citing extremely high success rates of cutmark identification in both experimental and archaeological samples with hand lenses and only minor usage of any microscopic materials (e. g. Blumenschine et al. 1996).  Cutmarks for the sample, once identified, were documented for all elements on which they appeared.  Subsequent research will provide documentation as to the precise locations of cutmarks on individual elements.  In summary, we have created an experimental situation in which observed and recorded variables can serve as causal variables for what is the only archaeologically visible attribute, the cutmarks.

               Figure 1. Example of Chert (top) and Obsidian (bottom) tools used for this experiment.

Table 1. Descriptions of experimental butchery episodes.

RESULTS

                Table 3 presents cutmark frequency data for each butchery episode in the experimental sample.  The frequencies are a percentage of the total strokes from each episode that were manifested as visible cutmarks on the bones.  It should be immediately apparent upon examination of these data that very few of the strokes become visible as cutmarks, the average for all the episodes being .72% with a high of 2.9% (table 3).  Three of the episodes exhibited no cutmarks at all (table 3).  This rather rough quantitative sketch indicates that very little of the energy expended to butcher any carcass would actually survive be observed archaeologically.  Table 4 and figure 2 present these same percentages for each element.  Again, very little of the energy, expressed in strokes, becomes visible on any of the individual elements.  These data indicate that the tibia is the most conducive to cutmark visibility, with an average (N=8) of 1.87% of the strokes, all of which resulted from defleshing, appearing as cutmarks (table 4 and figure 2). 

Because meat weight should have the greatest influence on how much energy was needed to butcher a particular element in terms of strokes and/or time, a test of how tight the relationship between meat weight and energy expenditure was required.  Figures 3 and 4 are scatter plots presenting the defleshing strokes per element and the defleshing time per element, respectively, versus the meat weight.  Each figure has two graphs: figure 3(a) incorporates the frozen carcass (BE-14), figure 3(b) omits it; figure 4(a) does not incorporate BE-1 into either graph (defleshing times for the individual elements were not recorded during this episode), while the second graph in figure 4(b) omits the frozen carcass (BE-14).  When a regression statistic is applied to the data in figures 3(a) and 4(a), values of .38 and .36 are calculated, respectively.  In figures 3(b) and 4(b) in which the frozen carcass is removed, however, regression values of .79 and .72 are obtained.

Table 2. Data collected for each butchery episode.

The profound influence by the frozen carcass on this relationship presents an opportunity to examine the effects of “extreme” carcass conditions on the energy required to butcher them.  In this case, “extreme” refers to BE-14, a frozen carcass, and BE-9 and BE-10, individuals whose ages are estimated at one month.  To butcher the leg unit of BE-14, it required the butchers to make 11578 strokes (table 3).  In comparison, the average amount of strokes needed in episodes that included complete front limbs as in BE-14 was 3106.5, a staggering difference in energy expenditure.  The time required to butcher the frozen carcass was also much more, taking 119 minutes, compared with an average of 79.75 minutes for the other episodes including complete front limbs. 

BE-9 and BE-10 are individuals which, based on tooth eruption and bone fusion, are estimated to be one month of age.  BE-10 is the only episode in which a skull was butchered, so no comparison can be made with it at this time.  BE-9, a complete rear leg, required 1536 strokes to butcher.  In contrast, the average for other episodes involving complete rear legs was 3956.5.  The total time needed to butcher BE-9 was 27 minutes while the average for the other episodes consisting of a complete rear leg was 74 minutes. These two “extreme” cases thus provide examples as to how much of an impact that, on the one hand a frozen carcass and on the other hand especially young individuals, can have on the energy required to butcher a carcass.     

In the absence of the frozen carcass (BE-14), which requires a greater amount of energy to butcher than a fresh carcass, it seems the amount of defleshing strokes was the best measure of energy expenditure.  Once this relationship had been ascertained, the

Table 3. Total time, total strokes, total observed cutmarks and percentage cutmarks for each episode.  Time and strokes include all activities (skinning, defleshing, etc.).  % reflects the percentage of strokes that were manifest as visible cutmarks on the bone. 

Table 4. Total strokes, total observed cutmarks and cutmark percentages for each element.  Table includes only those elements that were defleshed. % reflects the percentage of strokes that were manifest as visible cutmarks on the bone.  Codes: SC-scapula, HM-humerus, RDU-radio-ulna, FM-femur, TA-tibia, EQ-Equus caballus, BO-Bos taurus. 

       Figure 2. Cutmark percentages for defleshed elements.  X-axis represents the average percentage of

       strokes that were manifest as visible cutmarks on each element.

       Bone diagrams from http://lamar.colostate.edu/~lctodd/bison.htm

next step was to divide the strokes by the meat weights to get the number of strokes needed to obtain a kilogram of meat.  According to this measure, the defleshing of the radius/ulna required the most work to obtain each kilogram of meat (figure 5).  However, because both strokes and time are tightly correlated with meat weight, a more accurate and inclusive measure of energy expenditure would be one in which both measures were included.  The second measure of energy expenditure was arrived at by dividing the time by the meat weight, thereby acquiring a measure of how many minutes were required to obtain a kilogram of meat.

Figure 6 displays this relationship by plotting the two energy measures as averages for each element on the x and y axes; the most energy intensive element, in this case the tibia, appears in the upper right of the graph.  The sizes of the bubbles represent the average percentage of strokes that were visible as cutmarks on each element.  Because we can observe archaeologically neither the strokes nor the time that went into butchering a carcass, we need to know how, if at all, the amount of observable cutmarks relates to the energy expended by an individual butchering a particular element.  Upon examination of figure 6, it seems the assumption that increased energy expenditure would be accompanied by an increased amount of that energy becoming manifest as visible cutmarks is unfounded.  Only the tibia conforms to this expectation.  However, table 4 and figure 2 inform us that the tibia is the most conducive to cutmark visibility, thus

  Figure 3. Scatterplots of the relationship between the amount of strokes required to completely deflesh an

  element and the meat weight, in kilograms, of each element.  Figure 3(a) includes the frozen carcass in   

  BE-14 (represented by the red data points); the regression statistic for these data is .38.  Figure 3(b) omits

  the frozen carcass; the regression statistic for these data is .79.

    Figure 4. Scatterplots of the relationship between the time required to completely deflesh an element

     and the meat weight, in kilograms, of each element.  Figure 4(a) includes the frozen carcass in BE-14

     (represented by the red data points); the regression statistic for these data is .36.  Figure 4(b) omits the

     frozen carcass; the regression statistic for these data is .72.  Because time data for individual elements

     was not recorded for BE-1, this episode is omitted from both plots.

         Figure 5. Average energy expenditure to completely deflesh each element, expressed in the amount

           strokes required to remove each kilogram of meat.

           Bone diagrams from http://lamar.colostate.edu/~lctodd/bison.htm

    Figure 6. This figure displays the relationship between the average energy required to completely

     deflesh each element, represented by the average strokes (x-axis) and average time, in minutes (y-axis),

     required to remove each kilogram of meat and the average percentage of cutmarks that were manifest

     as visible cutmarks on each element.  The bubble sizes represent the relative cutmark percentages.  

     Bone diagrams from http://lamar.colostate.edu/~lctodd/bison.htm

weakening the argument that increased energy expenditure contributed to the cutmark visibility. 

Stepping back for a moment, it becomes evident that the percentage of strokes that become visible as cutmarks may not be the most appropriate measure.  The primary reason for this lies in its inapplicability to archaeological assemblages.  What are archaeologically available are the cutmark frequencies.  Figure 7 again plots the average energy expenditure per element with the bubble sizes in this case representing the average amount of cutmarks observed on each element.  Although the relative sizes of the bubbles change, the general relationship remains the same revealing that the average number of cutmarks can be used as a reliable substitute for the percentage of strokes that become visible as cutmarks.

Upon inspection of both figures 6 and 7 it is obvious that the visibility of cutmarks, whether expressed as a percentage of strokes or as a frequency, is not directly related to the amount of energy expended.  As was suggested above in reference to the tibia, there are three ways in which to interpret the nature of an element towards cutmark visibility.  First, it could be reasoned that the amount of cutmarks observed is positively related to the amount of energy per kilogram of meat required to butcher the element.  The data from figures 6 and 7, however, do not support this interpretation.  If anything, they suggest that the two measures are negatively related; i. e. the less energy per kilogram of meat expended, the more cutmarks result.  Secondly, it is possible that the amount of meat on the element is negatively related to the amount of cutmarks.  This assertion relies on the assumption that the more meat attached to a particular element makes it less likely that a stone tool will be able to penetrate to mark the bone surface.  Figure 8 presents this data in a scatterplot.  Episodes in which no cutmarks were present are not included.  Figure 8 does include the frozen carcass, the removal of which has little effect on the regression statistic (with: .396; without: .394).  Either way figure 8 suggests that the meat weight has little direct influence on the amount of cutmarks.  The last interpretation involves the unique position of the meat in relation to the bone that is characteristic of each element as responsible for the amount of cutmarks created.

Discussion

In light of the data presented here, the third interpretation alluded to above seems most plausible, and its implications for the analysis and interpretation of cutmarks are substantial.  Fundamentally, these data challenge the notion that observed cutmarks are directly and positively related to the energy expended in their creation.  Instead, observed cutmark frequencies represent more accurately the nature of the meat attachments on particular elements.  In addition, because of the fact that cutmarked specimens in no way reflect the entirety of the butchered sample, the mere presence or absence of cutmarks cannot be reliably cited as evidence for butchery in all cases.

So what can we ascertain from cutmarks if not direct energy-dependent causalities?  What this study provides is a sort of “measuring stick”, a standard set consisting of known variables which can be placed against a comparable data set in order to identify and explain disparate patterning.  Suppose an archaeological sample is available for comparison in which very little of the tibiae are cutmarked.  Based on this study, tibiae are the most conducive to cutmark visibility.  Also based on this study, it is

    Figure 7. This figure displays the relationship between the average energy required to completely

     deflesh each element, represented by the average strokes (x-axis) and average time, in minutes (y-axis),

     required to remove each kilogram of meat and the average frequency of observed cutmarks on each

     element.  The bubble sizes represent the relative frequencies of observed cutmarks.  Note that although

     the relative sizes of the bubbles change, the general relationship between energy expenditure and

     cutmark creation remains the same (see figure 6).  

     Bone diagrams from http://lamar.colostate.edu/~lctodd/bison.htm

also likely that the reason for this stems from the way that the meat is positioned on the bone. What can we infer about this fictitious assemblage based on the data gleaned from this study? 

First, we could reason that the cutmarked tibiae do not represent the entirety of the butchered tibiae.  However, it is likely that the cutmarked tibiae reflect a large portion of the butchered elements because of their susceptibility to cutmark visibility.  By now it would be evident that it is indeed likely that a very small number of tibiae were butchered.  Why?  In reference to this study, energy expenditure could be cited as a principle factor.  The tibia requires the greatest amount of energy per kilogram of meat to butcher.  If the butchers could afford to ignore for the most part those elements requiring the greatest amount of energy to obtain meat, perhaps they had primary access to other elements that did not require them to compromise energy to a degree that the tibiae did. Although a completely hypothetical situation, it serves to highlight the potential of the data gained from this study. 

    Figure 8. Scatterplot displaying the relationship between the frequency of observed cutmarks and the

     meat weight of a particular element.  Episodes in which no cutmarks were observed are omitted.  The

     regression statistic for these data is .396.

Conclusions and Future Directions

The experimental sample presented and analyzed in part here hopefully demonstrates the strengths of an experimental approach.  Of course, all the results presented here, however promising, remain highly questionable in light of the small sample size.  Several aspects of this study need subsequent consideration in order to further its interpretive potential.  First and most obvious, a larger set of carcasses must be integrated into the sample.  This includes axial elements and complete carcasses in addition to more leg units.  Second, episodes focusing on incomplete meat removal in different combinations must be included.  Complete meat removal was not the sole motive behind prehistoric butchery.  Third, the butchery strategies employed in this experiment, because meat weights required the cutting along articular planes, were unsympathetic to the actual complexities involved in removing a packet of meat.  The attachments of most muscle masses are not relegated to individual bones, but various locations on adjoining bones. This may have some impact on the locations and frequencies of some cutmarks.

            Even in light of these weaknesses, some solid conclusions are possible.  Namely, the amount of cutmarks, measured either through percentages or frequencies, do not directly reflect the energy intensity put forth to butchery the element on which they appear.  Archaeological interpretations of cutmarks must take this into account when attempting to describe the intensity and duration of prehistoric butchery episodes.  Because only a mere fraction of a dynamic set of processes survives to be interpreted, extreme caution must be taken for how these small samples are integrated into larger interpretive schemes.   

Archaeologists do not study the past.  In the absence of a time machine, the second best option is to make inferences about the past based on material that exists in the present.  The purpose of experimental studies like that presented here is to build a bridge from the observable present to the unobservable past.  The multitudes of individual events that contribute to the process of a prehistoric butchery episode are survived by a smaller and unequal set of archaeologically visible results, cutmarks being the most ubiquitous.  Increasing awareness concerning the processes that contribute to the creation of cutmarks will lead to more robust interpretations based upon those cutmarks.

Acknowledgments

            The authors wish to extend sincere thanks to the following: Dr. Calvin Jennings for allowing us the use of LOPA for our experiments, Dennis Madden at the Necropsy for providing the specimens, and Dr. Robert Lee for granting us access to the bone preparation rooms.  Thanks go out to Jen, Julie and Ryan for coming out and getting bloody with us.  We are grateful to Dr. Larry Todd for granting us permission to use his bone figures in this study.  We would also like to thank Dr. Todd for providing the assistance, motivation and inspiration that make this project and others like it possible.

Works Cited

Binford, Lewis R.   Nunamiut Ethnoarchaeology. New York: Academic Press, 1978.

Binford, Lewis R.  Bones: ancient men and modern myths. New York: Academic

            Press, 1981.

Blumenschine, R. J., Marean, C. W. and Capaldo, S. D.  “Blind test of inter-analyst

            correspondence and accuracy in the identification of cut marks, percussion marks

            and carnivore tooth marks on bone surfaces.” Journal of Archaeological Science

            23 (1996): 493-507.

Bunn, H. T. and Kroll, E. M.  “Systematic butchery by plio-pleistocene hominids

            at Olduvai Gorge, Tanzania.” Current Anthropology. 27 (1986): 431-452.

Dominguez-Rodrigo, M.  “Meat-eating by early hominids at the FLK 22  Zinjanthropus

            Site, Olduvai Gorge (Tanzania): an experimental approach using cut-mark data.”

            Journal of Human Evolution. 33 (1997): 669-690.

Frison, G.  “The Glenrock Buffalo Jump, 48CO304: late prehistoric buffalo

procurement and butchering.” Plains Anthropologist Memoir 7. 15 (1970): 1-69.

Potts, R. and Shipman, P.  “Cutmarks made by stone tools on bones from Olduvai

            Gorge, Tanzania.” Nature. 291 (1981): 577-580.

Shipman, P. and Rose, J.  “Evidence of butchery activities at Torralba and

            Ambrona: an evaluation using microscopic techniques.” Journal of

            Archaeological Science. 10 (1983): 465-474.

home | cla | colorado state university | disclaimer | equal opportunity