«Item type text; Dissertation-Reproduction (electronic) Authors Munro, Natalie Dawn Publisher The University of Arizona. Rights Copyright © is held ...»
percussion marks and frequencies and fragmentation indices will be positively correlated with independent estimates of marrow content. First, the frequency of cone fractures among prey taxa ranked by total marrow yield (body size) are compared. Next, completeness indices for long bone shafts originating from taxa with significantly different marrow yields are presented (gazelle, hares, and partridges). Third, the relationship between marrow content and the frequency of cone fractures and fragmentation indices for the Hayonim Cave gazelles are examined using a Spearman's rank-order correlation coefficient. Fourth, the fragmentation (NISPiMNE) of gazelle marrow-rich long bone shafts is compared to their marrow-poor articular ends. A test for the intensity of marrow extraction is presented last and examines the relative completeness of compact gazelle elements containing small to no marrow yields.
Results for Marrow Extraction The frequency and location of cone fractures on archaeological bone provide general rather than absolute indications of the intensity of human processing. Overall, the frequency of cone fractures in the Hayonim Cave and Hilazon Tachtit assemblages are low (see Table 5.15) and increase directly with prey body size, as do bulk marrow yields and bone thickness; big animals that store greater quantities of bone marrow therefore display more evidence for percussion. The only exception is the tortoise, whose shells often exhibit cone fractures. In this case people were after meat not marrow. Impact fractures form on tortoise shell due to its relatively thick, sandwich structure, not unlike
Table 5.15: Frequency of taxa bearing cone fractures in the Natufian layer from Hayonim Cave and Hilazon Tachtit.
Taxa are ranked in descending order of body size. Value in parentheses is the percentage of long bone shafts bearing cone fractures of that taxa.
Completeness indices for long bone shafts for gazelle, hare, and partridge are compared in Table 5.16. Partridges produce no marrow, hares store small quantities, and gazelles provide rich sources of bone marrow. The results indicate that the intensity of shaft fi'agmentation is positively correlated with marrow concentration. Despite their fragility, partridge long bone shafts from Hayonim Cave have consistently high completeness indices. In contrast, hare and gazelle long bones, have reduced completeness percentages, indicating greater incidences of breakage. Interestingly the degree of completeness is negatively correlated with bone fragility (i.e., mineral density).
Although, partridges have thin-walled, delicate bones, at least 50% of the shafts of each of their long bones are complete. Hares and gazelles have higher bone densities and much lower completeness indices; the completeness of gazelle shafts ranges from 16.7%
Table 5.17: Frequency of cone fractures, marrow utility indices and fragmentation indices (NISP:MNE ratios) for gazelle marrow bearing elements from the Natufian layer at Hayonim Cave.
Values outside of parentheses in the cone fracture column, represent the proportion of that portion bearing cone fractures.
Numbers in parentheses are the NISP of fractured specimens. Only elements capable of receiving cone fractures are listed here. Non-marrow bearing elements are included to fill out the variation in marrow content. Marrow utility values are based on Binford's (1981) index for domestic sheep.
Next, fragmentation indices and the frequency of cone fractures in the Hayonim gazelle assemblage are compared against bone marrow yields derived from Binford's (1978) values for domestic sheep using a Spearman's rank-order correlation coefficient (see Table 5.17 for data). Fragmentation indices are calculated as NISP to MNE ratios
the case of long bones. All elements that store bone marrow are evaluated, as are a few non-marrow bearing elements (e.g., astragalus and terminal phalanx) to round out the range of the sample. Both cone fractures and fragmentation indices are strongly and positively correlated with bone marrow yields (for cone fractures r, = 0.686, p.01, n = 12; for fragmentation r, = 0.688, p.01, n = 12). Mammal bones with high marrow contents are significantly more fragmented and contain cone fractures more often than those of other animals with minimal or no marrow stores (Figure 5.9).
Figure 5.9: Scatterplots depicting the relationship between marrow utility and frequency of cone fractures (graph a), and marrow utility and fragmentation indices (graph b) for gazelle long bones from the Natufian layer at Hayonim Cave.
Data is presented in Table 5.17. A few non-marrow bearing bones are included to fill out the range of variation in marrow content. Spearman's rank-order correlation coefficient is significant at the.001 confidence level for both relationships (cone fractures and marrow: r, = 0.685, p.001, n = 12, fragmentation and marrow; r, = 0.688, p.001, n = 12).
The next test compares the fragmentation (NISP:MNE) indices of marrow-rich long bone shafts versus marrow-free articular ends from the same element. This
as trampling, which preferentially fragment low density bone portions, including some of the marrow-poor articular ends (Nicholson 1992). Table 5.18 shows that gazelle long bone shafts from Hayonim Cave are consistently more fragmented than their corresponding articular ends, except in the case of the proximal humerus, the lowest density portion of the gazelle long bones (0.12 g/cm^). Fragmentation and medullary marrow content are thus positively correlated in this example.
Table 5.18: NISP:MNE ratios of gazelle long bone portions in the Natufian layer at Hayonim Cave.
Though long bones provide the primary source of marrow, ungulate mandibles and toes also house concentrated fat stores. In the mandible, marrow is stored in a hollow in the base of the horizontal ramus, which can be accessed by breaking the ramus into transverse sections and splitting off the base. Stiner (1994: 140) noted this breakage pattern in ungulate mandibles processed by Paleolithic hominids in Italy, and similar patterns are found in the gazelle assemblage from the Natufian layer at Hayonim Cave.
The mandibles from Hayonim are highly fragmented, with no examples of complete or nearly complete specimens. Gazelle horizontal rami (NISP = 91) are consistently split transversely and vertically (67.0%), providing easy access to marrow from three exposed sides. Finally, the ascending ramus is nearly always detached from the horizontal ramus
140), though these breaks are also encouraged by weak points in the mandibular structure.
The smaller bones of gazelle hind legs require more effort to crack despite low returns, thus providing a simple gauge of extraction intensity. Of the three ungulate phalanges, the first and second contain marrow, though the first phalanx holds about twice as much as the second (Binford 1978). The calcaneum also contains small marrow stores, similar in weight to the second phalanx. A rough measure of extraction intensity is provided by comparing the percent completeness of the first and second phalanges, and the calcaneum against the non-marrow bearing compact bones of the lower leg (the astragalus and third phalanx). In the Hayonim Cave gazelle assemblage, the first phalanx has the lowest completeness index (30.5%). The index for the second phalanx is much higher (56.9%), and also exceeds the calcaneum though the latter contains slightly less marrow (48.3%). Both the astragalus and third phalanx have significantly higher completeness ratios than the other elements (see Table 5.19). In general, the completeness ratios for the compact elements correspond directly to bone marrow yields.
The low rate of completeness for the first phalanx indicates that though small, this source of bone marrow was usually tapped, and effort consistent with lower medullary content was invested into cracking the second phalanx and calcaneum less than half the time.
The Natufians intensively exploited marrow stores in gazelle carcasses, as attested to by the fragmentation of nearly all marrow bearing bones including the first phalanx, they
Table 5.19: Completeness indices for select gazelle foot bones including the astragalus, calcaneum and phalanges from the Natufian layer at Hayonim Cave.
Elements are ranked in order of marrow content.
The completeness index is the percentage of each element which is complete or nearly complete.
To summarize, the results of the preceding tests for marrow extraction are in strong agreement. In every case bone fragmentation and/or cone fractures are more common on portions associated with abundant, concentrated marrow stores. This pattem holds up convincingly across taxa, elements, and bone portions within the same element.
The strong correlation between marrow content and fragmentation contradicts, to some extent, the signature of density-mediated attrition pointed out in Chapter 4. In two of the tests presented above bone density and marrow content were negatively correlated across taxa. In both tests fragmentation correlated strongly with marrow content, but inversely with bone density across different taxa (gazelles, hares and partridges). For example, fragile partridge long bone shafts which contain no marrow were complete more than 50% of the time although they have extremely low mineral density, while high density gazelle long bone shafts were complete less than 10% of the time. The shafts of gazelle long bones were also fragmented more often than their epiphyses though the shafts usually have higher bone mineral densities (see Table 5.18). It can thus be concluded that fragmentation and density-mediated attrition at least in marrow rich areas
marrow rather than secondary causes such as trampling or chemical break-down, factors also expected to preferentially destroy bones with low mineral density (e.g., Lyman 1994;
Grease and Trabecular Marrow Extraction Bone grease and trabecular marrow is dispersed in tiny pockets within the microstructure of bone (Brink 1997) and must be rendered using more elaborate techniques than medullary marrow extraction. Grease is stored in both cancellous and cortical bone, but it is easier to extract from cancellous bone, since the latter's porous structure facilitates its escape when heated and also reduces the energy required to crush bones when processed cold. This point is underscored by studies of modem huntergatherers, who process cancellous bones for grease more often than cortical bone (Binford 1978; Lupo and Schmitt 1997; Vehik 1977; Yellen 1991a). Grease is usually freed in a multi-step process by crushing cancellous bone using stone hammers or groundstones, immersing the pulp in boiling water to release the grease, and skimming the fat from the surface (Brink 1997; Vehik 1977).
Evidence for boiling of animal products during the Natufian period is inconclusive. The Natufians did not manufacture ceramics, and though they are as likely or even more likely to have made watertight animal skin or vegetal containers suited for stone boiling, fire cracked rock is rare in most Natufian sites. The possibility that the Natufians boiled water is not excluded here in the absence of evidence, but signs of "cold" methods for grease extraction are easier to find in the Natufian record. Grease
particularly cancellous parts, may be crushed into a fatty pulp and consunied as is, or mixed with edible plant or animal products. The mortars, pestles, and grinding slabs common at Hayonim Cave and other Natufian sites are massive and can easily smash cancellous elements (Laure Dubreuil, personal communication 2001).
Ethnographic sources indicate that grinding, smashing, and chopping activities are central to grease extraction if one is to reduce cancellous grease-rich bones to a mixture of small digestible fragments and bone meal powder (Binford 1978; Lupo and Schmitt 1997; Vehik 1977). Modem ethnographically documented groups limit grease extraction primarily to low-density, easy-to-grind bone structures. Grease extraction is thus expected to be expressed archaeologically as a density-mediated bias, or more specifically by low survivorship of cancellous bone. The bone portions with low survivorship are also expected to have high fragmentation indices.
Results for Grease Extraction To test for grease extraction, a Spearman's rank-order correlation coefficient is used to examine the relationship between survivorship and fragmentation. Bone shafts are eliminated from the comparison, since their breakage patterns and survivorship was argued to be the result of medullary marrow extraction. The articular epiphyses of long bones encompass a substantial range of bone densities, thus the omission of shafts should not be a biasing factor. Bone portions, associated fragmentation indices, and survivorship values (% MNE) are listed in Table 5.20, while the relationship between survivorship and
level of probability (r^ = -0.511, n = 13, p.001), indicating that articular ends with low survivorship potentials, are also more intensively fragmented than those that normally survive better in assemblages. The loss and fragmentation of low density bone at Hayonim Cave is restricted to gazelle, which is the only well represented mammal that yields a significant quantity of bone grease. The cancellous articular ends of gazelle long bones have consistently higher NISPrMNE ratios than hare bones despite similar bone mineral density ranges (see Table 5.21). Overall these results, and the substantial density-mediated bias in the Hayonim gazelle assemblage, meet the expectations for grease extraction. Other mechanical processes, namely trampling, may produce similar results, but should affect different taxa equally, which is not the case here. This complex issue will be taken up again below.
Figure 5.10: Percent survivorship versus fragmentation (NISP/MNE) of gazelle long bone articular ends from the Natufian layer at Hayonim Cave (r, = -0.
847, n = 17, P.001).