You are on page 1of 22
The Analysis of Cutmarks on Archaeofauna: A Review and Critique of Quantification Procedures, and a New Image-Analysis GIS Approach Yoshiko Abe; Curtis W. Marean; Peter J. Nilssen; Zelalem Assefa; Elizabeth C. Stone American Antiquity, Vol. 67, No. 4. (Oct., 2002), pp. 643-663. Stable URL: http://lnks,jstor.org/sie¥sici=0002-73 16% 282002 10% 2967%3A4% 3C643%3ATAOCOA% 3E2.0.CO%3B2-H American Antiquity is currently published by Society for American Archacology. ‘Your use of the ISTOR archive indicates your acceptance of JSTOR’s Terms and Conditions of Use, available at hhup:/www.jstororg/about/terms.huml. JSTOR’s Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at hup:/www jstor.org/journalysam.huml. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the sereen or printed page of such transmission, JSTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support @jstor.org. hupulwww jstor.org/ Thu Jun 8 22:13:18 2006 REPORTS THE ANALYSIS OF CUTMARKS ON ARCHAEOFAUNA: A REVIEW AND CRITIQUE OF QUANTIFICATION PROCEDURES, AND A NEW IMAGE-ANALYSIS GIS APPROACH Yoshiko Abe, Curtis W. Marean, Peter J. Nilssen, Zelalem Assefa, and Elizabeth C. Stone ooarchacologlssulize a diverse se of approaches for quantifying cutmark frequencies. The least quantitative method for ‘uma analysis relies on composite diagrams of eumarks overlain on drawings of skeletal elements (diagramatic methods) Todate, interpretation ofthese data have generally relied on qualitative and subjective assessments of cumark frequency and placement. Many analysts count the numberof fragments that havea cumark, regardless ofthe number of eumarks onthe {fragments (Fragment count data). Others count the number of eumarks(cutmark-count data). Bot canbe expresed a sm ple counts (NISP data), or ata count of some more-drived measure of skeletal element abundance (MNE data). Al ofthese ‘approaches provide diferent yes of data and are not inercomparable. Several researchers have shown tha fragmentation of specimens impacts the frequency of ews, and we show here that fragmentation impacts all hese current approaches in ways ‘ha compromise comparative analysis when fragmentation difers between assemblages. We argue that cumark frequencies {from assemblages with difering levels of fragmentation are most effectively made comparable by correcting the frequency of ‘utmarks by he observed surface area. We presen anew method that allows this surface area correction by usin the Image ‘analysis abilities of GIS. This approach overcomes the fragmentation problem. Weilustate he power ofthis technique by com paring a highly fragmented archaeological assemblage to an unfragmented experimental collection. Los sooarqueslogos lean diverss métodos para euantifcr la frecuencia de huellas de cone. El método menos cuantitativo ‘pra el andlisis de huelas de corte wtlieadiagramas compuestos de este ipo de hullas ques sobreponen a dbus de elemen- ‘os esqueltics méodo diagramiico. Hasta el dia de hoy la itepretacién de estas observacones se ha basa en evalua- cones cualtatvas y subjetvas dela frecuencia y posicion de elas de corte. Muchos inestigadores cuentan el nimero de Fragments dseos con hulls de core, sn considerrelnimero de hellas eos mismos fragmento (metodo de conte de frag- ‘ments), Oto inestigadorescuentan sinplementee nimer de huellas de corte (método de coneo de huelas de corte) Ambos {pueden expresar ya sea como cuantfacié simple (datos de NISP), 5 como wna medida derivada de abundancia deeemen- ‘os se0s (datos de MNE), Todos estos méiodos ofrecer dsintotiposdeobseracianes, ls cuales nos Son comparables ene. Nios inestigadores han mosrado que la fragmentaciondseaafeca la frecuencia de hella de corte. Nosotros mosram en «ete atcul que la fragmentacin bea ify ntablemente en odo los miodos usados hasta el momento, ue, debido allo, se arriesgan los anlsscomparativas cuando el grado de fragmentacin bea es distin entre las colecciones a camparar Pro- ponemos que cuando el grado defragmentaci dea vara entra colecciones,lafecuencia de hulls de corte podria ser com Dparada en forma mds efectiva al corre la referida frecuencia con la medida del dra dela superficie observada, Nosotros Dresentamas un méiodo nuevo que permite estandarzar la freuencia de uel de corte por dreade superficie através del uso de andlss de imagen con GIS. Este méodo superael problema de a fagmentacion da. Agul mostramos su poten al com ‘para una coleccén sea altamentefragmentada con ora colecctn experimental no fragmentada. ‘ standard research endeavor in zooar- then are used to address more wide-ranging topics thacology and has been used to address a of greterinterest. We use the term “butchery” torefer variety of topics. Generally, studies of cutmarkshave tothe actions taken to render a carcass into usable T: analysis of cutmarks on skeletal elements been used to reconstruct butchery strategies, which a ‘Yoshiko Abe and Zalalem Asse Inierdpartnenial Doctoral Program in Anhropologial Slences, SUNY at Stony Brook, Stony Brook, NY 11794-1368 (Curtis W. Marean a Insite of Human Origins, Deparment of Anthropology, PO Box 872402, Arizona State University, ‘Tempe, AZ 85287-2402 Peter J. Nilsen w Deparment of Archaeology ko - South African Museum, PO. Box 61, Cape Twa, 8000, South Africa Elizabeth C. Stone # Department of Anthropology, SUNY at Stony Brook, Stony Brook, NY 11794-4368 “American Aniqity, 67(4), 2002, p. 643-663, Copyright© 2002 by te Society for American Archaeology 63 o AMERICAN ANTIQUITY portions (Lyman 1987), often for consumption, but the production of raw materials for tool manufacture ccan also be a single or related goal. Butchery by cutting for consumption typically involves skinning, disarticulation, defleshing, and in some cases removal of periosteum, Hammerstone percussion of skeletal elements is also technically butchery, but that process generally has the goal of fragmenting skeletal parts to access marrow or make them more easly boiled for grease rendering. Ham- ‘merstone percussion leaves a mark that, o a trained analyst, is distinct from a cutmark (Blumenschine et al. 1996). Our focus here is on cutmarks and their analysis, ‘Studies of cutmarks figure prominently in human origins research where zooarchaeologists have stud- ied patterning in cutmarks to investigate whether Plio-Pleistocene hominids were hunters or scav- cengers (Binford 1981, 1985, 1988; Bunn 1981, 1991; Bunn and Kroll 1986, 1988; Pots 1983, 1988; Potts and Shipman 1981; Shipman 1986, 1988; Shipman and Rose 1983). Over time this dichotomous approach gave way to using cutmarking tohelp iden- tify where in the sequence of carcass consumption hominids regularly fit (Capaldo 1995, 1998; Sel- vaggio 1994, 1998), Studies of cutmarks have also been used in modern human origins research to test Binford’s suggestions that even late-occurring non- ‘modern hominids were primarily scavengers of large ‘ungulates (Chase 1986, 1988; Grayson and Delpech 1994; Marean 1998; Marean et al. 2000; Marean and Assefa 1999; Marean and Kim 1998; Milo 1994, 1998; Stiner 1994), However, cutmark studies have been conducted in many other research endeavors where the recon- struction of butchery processes is seen as relevant to other behavioral traits. The optimistic view is that the butchery process varies with the intended use of carcass, and that this variation wil be expressed in the placement and frequency of cutmarks on the skeleton, allowing us to infer carcass use from cut- mark studies. For example, it has been argued that butchery should vary between contexts where the goals are immediate consumption versus storage (Binford 1978). Binford’s well-known utility model hhad differing predictions for skeletal element choice, and researchers have anticipated that cutmarking should also vary widely between, for example, an unselective or "gourmet strategy” versus more inten- sive utilization (Binford 1984). Zooarchacologists [Wot 67, No. 4, 2002) hhave hoped to identify fileting versus skinning (Bin- ford 1981; Shipman 1981; Shipman and Rose 1983; Wilson 1982), Yellen (1991) has argued that some butchery pattems have a “style” that could be cul- turally determined, thus holding out the possibility that butchery pattems may provide ways to examine ethnicity ‘This high promise has been frustrated by several factors, First is a lack of detailed observations of. butchery and its resultant patterning. Thereare many studies of butchered modem bones (Binford 1981 1984; Crader 1983; Gifford-Gonzalez 1989; Gifford and Crader 1977), butnone ofthese actually observed the act of butchery that produced marks, and then linked those specific actions to specific cutmarks. A recent study that filmed butchery actions close-up, thus providing this unambiguous linkage, found that ‘many ofthe disarticulation or deffeshing marks llus- trated in Binford (1981), and regularly used as a guide to butchery analysis, are not unambiguous indicators of specific activities (Nilssen 2000). Thus the strict causal linkage between observed specific behaviors (such as cutting for disarticulation or for defleshing) and their traces (such as cutmarks on articular ends versus shafts), called for by Gifford- Gonzalez (1991), is not yet fully developed in the literature, Another critical problem is the diversity of approaches for recording the cutmarks and then ‘quantifying ther frequency. Its safe to say that there is no accepted method for either. Our review of the literature shows that recording cutmarks can take ‘two paths. One i to record a count and description of the cutmarks onto a database. This ean be done either ata gross level (how many are on a specimen) ‘ora finer level (where on the specimen they occur along witha diagnosis oftheir character). A second approach is to draw cutmarks onto a diagram of bone, or what we will call a template. The two approaches can be easily combined, and probably often are. (Once the cutmarks are recorded the analyst must choose a way to quantify, analyze, and present the datain publication, and here there is also.a wide var ety of approaches. This step is problematic due to the potential for wide interanalyst variation in con- ‘vention. This variation is a severe problem for com- parative studies in zooarchaeology because it makes itdifficultifnotimpossible, to compare data between researchers. Just as zooarchaeologists have defined REPORTS anatomical landmarks and measurements that stan- dardize their approach to osteometrics (Driesch 1976), zooarchacologists must also strive for stan- a5 Disaticulation and Fileing 23 0001 2 Sos Bosh Libs Filleing Folin, Disaiculaion 268, 0001 08 >.0s and Flleting Hinglinb Disricultion and Fileting sis 0001 7 <.0s Fore, illeting Hindi Only dataset, closely following our expectations. We would expect that the intensity of cutting within zones would co-vary weakly between the experi- mental datasets since 40 percent of the zones (the articular ends) are being treated differently in the butchery procedure, while 60 percent should be treated similarly. In other words, there should be a weak tendency for cutmark frequencies to increase and decrease consistently between the two datasets. The Spearman’s Rank correlation coefficient between pooled forelimb and hindlimb datasets is consistent with this expectation (r, = 46, p <.05). Even though we would expect some discontinu- ities in the way cutmark frequencies increase and decrease between the experimental datasets, we ‘would expect the overall cutmark frequencies to be similar between the two datasets. This s because the butcher is using the same tools (metal knives) and butchering the carcasses under the same conditions (for consumption and the production of biltong) in bboth experimental datasets. This null hypothesis is supported by alack of significant difference between the medians ofthe two pooled forelimb and hindlimb datasets (Mann-Whitney U= 119, p = 367). Given the patterns of cutmarking associated with the two different activities, we can now examine the relation between the DKI pattern and the experi- ‘mental datasets and ask which experimental dataset is most similar to the DK dataset. To examine this question, we produced four experimental models from the two experimental datasets. Two of these experimental models precisely reflect the experi- ‘mental datasets in that they representa Flletingstrat- egy applied consistently to both limbs, and a Disarticulation-and-Filleting strategy applied to both limbs. However, at least two other strategies are pos- sible, produced by applying Flleting tothe forelimb combined with Disarticulation-and-Filletng to the hindlimb, and vice versa. Interestingly, the overall CNC frequencies in the DK1 dataset are significantly higher than all four experimental models (Table 3). Nilssen’s butchers used metal knives while the overwhelmingly domi- nant lithic raw material at DK is quartzite, and most of these are large unretouched flakes and blades (Thackeray 2000). This result would appear to con- firm untested expectations that metal butchery may result in overall fewer cutmarks than stone tool butch- ery (Fisher 1995; Lyman 1987). In our experience, stone tools become rapidly less effective as a result ‘of dulling edges and fat adhering to the surface, ther of which is as significant when using metal knives. The correction ofthe cutmark frequencies by surface area allows us to measure this difference in intensity of cutting quite precisely. Such abilities ‘may in the future hold out the possibility of directly relating tool resharpening and reduction intensity (Dibble 1995) tocutmark frequency, and measuring the impact of changes in raw material quality on butchery efficiency. ‘Our research on the DK1I faunal assemblage had a one if its goals the testing of Binford’s (1984) hypothesis that MSA people in South Africa were not behaviorally moder and tended to scavenge large antelope of the size represented here in our sample. Binford argued that the Klasies River assem: blage showed a pattern of butchery that failed to resemble that produced by modem butchers, in that ‘cutmarks were relatively rare in areas of bones that have large amounts of flesh. The DK large bovid forelimb pattern visually shows a compelling simi- larity to the Disarticulation-and-Filleting experi- ‘mental dataset (Figure 11), in that cuting is more abundant at the tight humeral-radial joint. The hindlimb visually resembles the Flleting dataset in its distribution (Figure 12). Thus, we would expect that the DK CNC frequencies would increase and ectease across the long-bone zones most comis- 0% 50% 100% 0% 50% | 100%, Filleting only _Disarticulation followed by Filleting HU proximal epiphysif] 26.3, HU proximal shat] 18.2 HU mid shan] 10.7 HU distal shan] 14.2 HU distal epiphysis] 28.5 RA proximal epiphysifEE) 48.1 A proximal shat] 18.6 DKI bovids (Sizes 3 and 4) Figure 11. Cutmark frequencies crus the length of forlimb long bons, expressed as percentages (cach skeletal element ‘adds up 10100 perent). Catmark frequencies data labels) are the CNC values. Filleting only Disarticulation followed by Filleting FE proximal ep FE proximal sa FE mid shat 24.5 distal shan] 20.6 Fr dita pints] 14.1 1 proximal epiphysif] 6.7 TH proximal shat] 14.8 11 mid shat] 18,5 TH distal shat] 15.7 Tidistal epiphysiq] 6 0% aay 50% 100% DK1 bovids (Sizes 3 and 4) Figure 12. Cutmark frequencies across the length of hindlimb long bones, expressed as percentages (ach skeletal element ‘adds up to 100 percent). Cutmark frequencies data labels) are the CNC values. tently with the Disarticulation-and-Filleting forelimb and Flletng hindlimb model, The correlation analy. sis presented in Table 3 supports this hypothesis. In summary, through the use of the CNC analy- sis we were able to directly and quantitatively com- pare the frequency of cutmarks in long-bone zones between a highly fragmented fossil bone collection (DK1) and an actualistic assemblage of cutmarked whole bones. This allowed us to discover two pat- tems with the following likely conclusions: 1) the DKI MSA people butchered size 3 and 4 animals such that the cutting intensity co-varied across long bones in a way that significantly correlates with a ‘model where modern people disarticulated and fil- leted the forelimb and filleted the hindlimb, and 2) that the use of stone tools at DKI likely caused a higher overall frequency of cutmarks per unit bone surface. Clearly, the calculation of CNC (made pos- sible by the image-analysis GIS approach), when combined with highly controlled actualistic data, allows us to significantly improve upon traditional approaches to cutmark analysis that rely on qualita: tive assessments of cutmark patterning guided by anecdotal knowledge of butchery patterns. Discussion and Conclusions Zooarchaeologists have developed a wide variety of| approaches for recording and presenting cutmark data, and they vary according to what is counted (cutmarks or cutmarked fragments) and the way the values are expressed (NISP or MNE). Our review of the literature finds five basic types of presented data: ddiggrammatic data, NISP cutmark-count data, MNE ccutmark-count data, NISP fragment-count data, and ‘MNE fragment-count data, There is no widespread ‘agreement as tothe most effective way to present the data, and there is also an undercurrent of concern as tothe effectiveness of these different approaches. To ‘our knowledge, only Bartram (1993) has provided a detailed discussion ofthe problems suffered by cul- ‘mark analysis asa result of bone fragmentation Diagrammatic data provides a useful overall view ‘of the placement of cutmarks, butts nearly useless for comparative analysis that attempts to go beyond qualitative and subjective impressions. Furthermore, diagrammatic representations can be very mislead- ing in regard to cutmark intensity if there is sub- stantial intra-bone differential survival. Cutmarks ccan be absent from locations on the bone simply because those locations are not well represented in ‘AMERICAN ANTIQUITY {Wol.67, No 4, 2002) the faunal assemblage or analysis, The GIS method we have presented here overcomes that problem with its ability to present an MINE density map across a bone surface along with a composite diagram of cut- ‘mark placement. More importantly, the GIS method ccan quantitatively correct the intensity of cutmark- ing by the preserved surface area in any way the ana- lyst desires. NISP cutmark-count data and NISP fragment- count data are extremely sensitive to fragmentation because with increasing fragmentation the total num ber of fragments rises more rapidly than the total ‘numberof either cutmarks or cutmarked fragments, ‘Comparisons of archaeological NISP cutmark data tounfragmented experimental collections simply are not effective unless the archaeological collection is completely unfragmented. Likewise, comparison of [ISP cutmark data between archaeological sites of ceven limited variable fragmentation will be spurious. MNE cutmark-count and MNE fragment-count data have been argued to be resistant to the frag- ‘mentation effect, While in some cases the MNE may mollify the fragmentation effec, it does not over ‘come it. Tis is because as processes of atrition act ‘on a bone, they tend to destroy the less-dense por- tions frst, leaving the denser portions preserved. MNEs based on these denser portions will remain high, while many cutmarks and cutmarked fragments are lost to the analysis either because they are ren- dered unidentifiable through fragmentation, or are ‘completely destroyed, ‘Accurately quantifying cutmark frequencies isa surface-area problem similar to sampling problems involving surface area in other disciplines, Forexam- ple, when ecologists sample habitats to estimate pop- ulation size, the proper correction is population divided by area, Zooarchacologists have tried to use the MNE as a proxy for area, but it does not work because bone portions do notall have an equal chance of surviving attitional processes. Rapson (1990) suggested a way to attempt an aerial estimate, but the method is fairly coarse and provides data of lim- ited analytical value. The GIS method does anexcel- Tent job of correcting for preserved surface area and ‘makes possible the meaningful comparison of cut- mark frequencies between heavily, lightly, and un- fragmented samples. ‘The GIS image-analysis approach has several other strengths. It provides a multitude of possible analyses that go beyond any other extant approach. REPORTS. ‘The GIS method allows the analyst unlimited flexi- bility in the definition of zones to calculate surface ‘area and count cutmark frequency. For example, if the analyst wants to examine the intensity of cuting fn just the femoral head, then the femoral head can be identified as the one for quantification Ifthe ana. lyst wishes to compare the intensity of slicing and hhacking marks within just the femoral head, that is easy as well because the GIS attaches a data table to cach individual cutmark, and one can restrict analy- ‘es with query functions onthe data tables. This abil- ity to stratify analyses by data characteristics applies toany possible data-table information such as taxon, size, age, provenience, or anything ese ‘The GIS method provides data archival abilities that go far beyond any other approach. Any approach that records cutmark placement with a database for- ‘mat immediately loses the exact placement ofa cut- mark, The GIS method forever records the location ‘of acutmark ina precise manner. Other diagrammatic approaches record cutmark placement inthis way as Well, but each mark must by coded to a separate data table if one wishes to record specific information on the nature of that cutmark, and this creates an extremely clumsy recording system. With the GIS ‘method, a cutmark is drawn directly on the computer template, and one mouse click opens a data table attached forever to that cutmark. We have focused our attention here on cutmark recording and analysis, but clearly this method can ‘beexpanded to almost any type of bone-surface mod- ification. For example, one could record burning ‘damage on bone surfaces to discover patterning in bbuming that could reflect differing approaches to ‘cooking. Estimates of bone surface area could reveal

You might also like