AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 119:15–26 (2002) Differential Skeletal Preservation at Windover Pond: Causes and Consequences Christopher M. Stojanowski,1,2* Ryan M. Seidemann,3 and Glen H. Doran2 1 Department of Anthropology, University of New Mexico, Albuquerque, New Mexico 87131 Department of Anthropology, Florida State University, Tallahassee, Florida 32306 3 Paul M. Hebert Law Center, Louisiana State University, Baton Rouge, Louisiana 70803 2 KEY WORDS taphonomy; bone preservation; skeletal remains; prehistoric Florida ABSTRACT In this paper, we evaluate the causes of differential skeletal preservation in the Windover Pond skeletal series (8BR246). We collected data on sex and age for approximately 110 individuals, and calculated a preservation score for each individual based on the presence of 80 skeletal landmarks. Our research questions evaluated the relationship between bone preservation and individual age and sex, and between the presence of preserved brain material and skeletal preservation, and the effects of burial location on bone preservation. The results indicate variability in average preservation for the sample ( ⫽ 0.53, SD ⫽ 0.22) with an apparent lack of sex-specific (P ⫽ 0.79) or age-specific (P ⫽ 0.37) differences in preservation. The relationship between brain and skeletal preservation (P ⫽ 0.15) was not significant. The horizontal distribution of burials was not significantly correlated with skeletal preservation (north: r ⫽ ⫺0.10, P ⫽ 0.93; east: r ⫽ 0.09, P ⫽ 0.45); however, vertical depth was a significant predictor of preservation (r ⫽ ⫺0.31, P ⫽ 0.005), indicating that skeletal preservation decreased as burials were located closer to the ground surface. The observed variability in preservation scores may be related to the partial drying and resubmergence of the uppermost burials for the last few millennia. Comparison of Windover element-specific survival rates with previous analyses based on terrestrial samples (Galloway et al.  Forensic taphonomy, Boca Raton: CRC Press; Waldron  Death, decay and reconstruction, Manchester: Manchester University Press; Willey et al.  Am J Phys Anthropol 104:513–528) affirms the relationship between element weight or density and bone survival. The unique taphonomic context of our study sample effected little change in bone deterioration processes. Am J Phys Anthropol 119:15–26, 2002. © 2002 Wiley-Liss, Inc. The deleterious effect of marginal preservation on demographic parameters is often stated as a concern in demographic studies (Angel, 1969; Jackes, 1992; Johansson and Horowitz, 1986; Konigsberg and Frankenberg, 1992, 1994; Milner et al., 1989; Moore et al., 1975; Paine, 1989; Paine and Harpending, 1996; Roth, 1992; Weiss, 1973). Sample representativity is affected by many factors, of which only a few relate directly to bone preservation (Boddington, 1987). Boddington (1987) considered postdepositional disturbance and deterioration to be primary taphonomic concerns; however, additional factors include incomplete archaeological sampling, spatial variability, and age-, sex-, or class-specific differential burial treatment. The mechanics of differential bone preservation (both between individuals and among elements of a single individual) are fairly straightforward. Von Endt and Ortner (1984) stated that the surface area of an element is directly related to the speed at which water-leaching results in the destruction of the collagen and mineral matrix. Therefore, since spongy or cancellous bone has a greater surface area, one would expect these areas of the skeleton to deteriorate first. The porous and less dense vertebra, ribs, sternum, and hand and foot bones are generally those exhibiting marginal preservation (Boddington et al., 1987; Nawrocki, 1995; but see Waldron, 1987 for lumbar vertebrae). Dense areas of bone such as long bone shafts tend to preserve better (Galloway et al., 1997; Waldron, 1987; Willey et al., 1997). Density and porosity can also be used to explain the differential preservation of elements between individuals. Since bone mineral content increases through childhood and adolescence, peaks in midlife, and declines with senescence (Riggs and Melton, 1986; Specker et al., 1987; Stini et al., 1992), age-specific preservation biases may be expected. Pathological conditions also may affect bone preservation in a similar manner where osteoclastic processes (osteoporosis) promote bone deterioration, and osteoblastic processes (osteophyte formation) © 2002 WILEY-LISS, INC. *Correspondence to: Christopher M. Stojanowski, Department of Anthropology, Florida State University, Tallahassee FL 32306. E-mail: firstname.lastname@example.org Received 5 July 2001; accepted 10 January 2002. DOI 10.1002/ajpa.10101 Published online in Wiley InterScience (www.interscience.wiley. com). 16 C.M. STOJANOWSKI ET AL. promote bone preservation (Nawrocki, 1995). Differences in body size due to sex may exhibit similar effects on bone condition (Walker et al., 1988). Therefore, in marginally preserved assemblages, one would expect an underrepresentation of subadults, elderly adults, and females due to the diminished resiliency of less dense bone to taphonomic factors such as soil acidity (Gordon and Buikstra, 1981; Walker et al., 1988; Walker, 1995) and geological factors such as soil expansion and shrinkage (Boddington, 1987). Although clearly demonstrable in poorly preserved samples, these results may be slightly overstated for samples of marginal, but not terrible, preservation (Buikstra, 1997). While scholars agree that age effects are real and have been demonstrated in actual case studies (Boddington, 1987; Gordon and Buikstra, 1981; Henderson, 1987), the lack of subadults in a skeletal assemblage may be more the result of burial practices and poor archaeological recovery techniques (Saunders, 1992; Saunders et al., 1995). Despite the study by Walker (1995), sex and general body size effects are not supported by other case studies (e.g., Boddington, 1987; Nawrocki, 1995). In this study, we examine the causes of differential skeletal preservation observed in the Windover skeletal series, an Early Archaic period mortuary pond located near the Atlantic coast of Florida. We test a number of hypotheses regarding the causes of differential preservation in this sample. We first test the hypothesis that females are less well-preserved than males, and that subadults and elderly adults are less well-preserved than middle-aged adults. We also evaluate differential deterioration of individual elements and specific landmarks on a given element. Our hypothesis is that lighter and less dense elements will display more fragmentation and lower overall preservation than heavier, denser elements. In addition, for a given element, those features located near or on less dense or lighter areas of bone will preserve less frequently than features located near denser, heavier bone sections. Because brain preservation was a remarkable aspect of this sample, we also evaluate the relationship between presence of brain material and skeletal condition. Finally, we test two hypotheses regarding the effects of burial location on skeletal preservation. We first propose that burials physically located near the pond margins are less well-preserved than those located near the pond center. Finally, we test the hypothesis that burials located higher in the stratigraphic column are less well-preserved than those buried deeper in the stratigraphic column. MATERIALS Study materials The Windover site (8BR246) is located in Brevard County, Florida, approximately 8 km west of Cape Canaveral. The site consists of a long-standing pond which served as a cemetery for the Windover population. The pond bottom is composed of multiple strata of anaerobic peat, whose deposition resulted from the accumulation of decomposing plant material for the last 10,000 years (Doran, 2002). Burials were placed in shallow depressions in the peat and were often wrapped in fabric, covered in wood debris, and intentionally staked to the pond floor (Doran, 2002). Based on the presence of preserved brain matter, it appears that water had filled the pond at the time interments were occurring; however, it is uncertain whether the burials were placed below the water table or merely within saturated soil during seasonal minima in water elevation (Doran, 2002). Excavations were conducted by Glen Doran and David Dickel over a 17-month period between 1984 – 1986. Subsequent radiocarbon dating of human bone, wooden stakes, peat, and bottle gourd remains placed burial activities in the Early Archaic period (after Milanich, 1994), from 8,120 – 6,990 years BP (Doran and Dickel, 1988a). Most burial activity may have commenced between 7,330 –7,100 radiocarbon years BP; however, the maximum range of dates suggests an approximate 1,000-year interval (Doran, 2002). The remains of 168 individuals were excavated from the pond (Doran et al., 1990), 58 of which were identified in backhoe spoil from construction activity. The functional sample size for most analyses is therefore around 110 individuals (Doran, in press). Ages range from infant to 65⫹ years, with approximately half (52%) the sample consisting of subadults (Hauswirth et al., 1991). The ratio of males to females was approximately equal, suggesting that all demographic groups are represented in the skeletal assemblage. Ninety-one human brains were recovered from the site, and more than 37 individuals were recovered with associated fabric material (Doran et al., 1990; Hauswirth et al., 1991). The preserved brain and organic plant material seems to suggest excellent preservation of the associated human remains. However, the preservation of skeletal material has been variously described as good, excellent, or simply variable (Doran et al., 1986; Purdy, 1991; Tuross et al., 1994). Taphonomic context At the time of cemetery use, the area consisted of a wooded environment which flanked the pond margins. The pond itself is bowl-shaped, with a deeper center portion ringed by a much shallower and gently sloping ledge near the edge of the pond. It appears that burial locations were selected based on water levels at time of burial. Although much of the pond circumference had been utilized, excavations in some areas produced little human material (Doran, 2002). Because of the basin-shaped pond bottom, the depth and locations of stratigraphic levels vary across the site. The idealized stratigraphic sequence DIFFERENTIAL WINDOVER PRESERVATION 17 Fig. 1. Idealized stratigraphic sequence of Windover Pond site. (From Doran, 2002: Figure 10.1. Reprinted courtesy of the University of Florida Press.) 18 C.M. STOJANOWSKI ET AL. is presented in Figure 1; however, peat strata tended to be thinner near the pond margins and thicker and more compacted near the pond center (Doran, 2002). These thickness and compaction differences created a context in which two individuals of similar depth from the ground surface may not necessarily be located in the same stratigraphic level. Likewise, the position of each burial in relation to strata interfaces may differ by horizontal location. Vertical control over burial position, therefore, has more to do with distance from the ground surface. The majority of burials were located in the lower red/brown peat stratum, with a few burials located in the upper red/brown peat stratum. The lower stratum was characterized as woody to leafy, and was deposited in a woody marsh environment beginning around 8,000 BP and terminating around 7,000 BP. The pH of the lower red/brown peat is 5.9 (Doran and Dickel, 1988b). The upper red/brown peat consists of more woody debris and was deposited from 7,000 – 6,000 BP, with a pH of 6.6 (Doran and Dickel, 1988b). The highest stratum at the pond is a fibrous black peat with a pH of 6.1 (Doran and Dickel, 1988b). In comparison to other nearby ponds (with an average pH of 2.9), the pH of the Windover site is unusually neutral (Hauswirth et al., 1991). This, combined with high sulphur levels, highly mineralized water, and an anaerobic peat environment, is believed to be responsible for the preservation of organic remains (Hauswirth et al., 1991). In this paper we focus on natural explanations for differential preservation, and therefore must verify that postdepositional disturbance, incomplete archaeological sampling, spatial variability in burial position, and age-, sex-, or class-specific differential burial treatment are not important factors to consider for the purposes of this paper. Postdepositional disturbance was observed sporadically and was the result of two unrelated processes. The first was a general slippage of skeletal elements toward the deeper center of the pond. This process was variable, however, and tended to be more severe on the southern edge of the pond, where the margin ledge was steeper (Doran, 2002). However, it is important to note that burial density was limited in this unit of the excavation, and this downslope movement of bones did not significantly disperse or mix elements from multiple burials. The majority of burials were located on the north side of the pond where, due to a much gentler slope, burial disturbance was not observed (Doran, 2002). Rodent disturbance was rare; only 6 of 10,000 bone fragments displayed evidence of gnaw marks (Doran, 2002). Root disturbance was also limited and largely restricted to the uppermost burials. It is more difficult to verify that incomplete recovery or differential burial treatment did not affect the representation of elements in this sample. However, there was no apparent age- or sex-specific clustering of individuals within the site, such that for any given unit of recovery, all demographic groups were equally represented (Doran, 2002). Excavation procedures at Windover were designed to maximize floral and faunal recovery, and all burial fill was wet-screened. The difficulty even in wet screening the peat necessitated the use of 1/4 and 1/8 inch mesh sizes; there was no discernible difference in the type of material recovered in each screen size, suggesting representative recovery with the 1/4 inch mesh size (Doran, 2002). METHODS Age and sex were estimated by Glen Doran and Dave Dickel, following standard osteological procedures (Bass, 1995; Buikstra and Ubelaker, 1994; Ubelaker, 1989), and were verified by Christopher Stojanowski with consistent results. Horizontal provenience data were recorded based on the location of the pelvis for each individual in the analysis. In addition, each burial was located within the stratigraphic column by recording the distance of the pelvis from the mean annual meters above sea level. Despite the fact that a number of previously developed quantitative methodologies exist for recording skeletal preservation (Boddington, 1987; Gordon and Buikstra, 1981; Milner, 1983; Nawrocki, 1995; Seidemann, 1999; Walker et al., 1988, 1996; Walker and Snethkamp, 1984; Walker, 1995; White, 1992), none allowed the specificity we desired. We therefore developed a new scoring methodology specifically for this project. Rather than categorizing a particular skeleton as poor, fair, or good, we generated a list of skeletal and dental landmarks that could be converted into a proportion indicating skeletal completeness. Because one impetus for doing this research was to evaluate the demographic integrity of the Windover sample, a list of landmarks and measurements with demographic importance was well-suited to our needs. We chose 35 paired and 10 unpaired features for a total of 80 observations per individual (see Table 1). Each feature was determined to be either “scorable/present” or “unscorable/deteriorated” for each individual in the sample. For our purposes, deterioration is defined as a postdepositional, natural taphonomic modification of bone morphology that reduces the utility of an element or feature in sexing or aging the individual from whom the bone derives. Pathological and cultural modifications of an element that reduce its utility in sex and age assessment are disregarded and are assigned a “present” score. Because teeth serve an integral role in many osteological analyses of marginally preserved skeletal samples, their inclusion in this analysis was necessary, though problematic. We treated each arcade as a single, independent unit where the arcade score can vary between 0 –1. For adults, each tooth is valued as one sixteenth (1/16) of the total possible score, resulting in the subtraction of 0.0625 from the perfect score of 1 for each missing or damaged tooth. DIFFERENTIAL WINDOVER PRESERVATION 19 TABLE 1. Cranial and postcranial features scored with API methodology Feature Cranium Superciliary arch Supraorbital margin Mastoid process Temporal lines Root of zygomatic Gonial angle Orbital shape Nuchal crest Mandibular profile Frontal profile Mental eminence Palate size and shape Femur Length Head diameter Neck diameter Midshaft circumference Bicondylar breadth Popliteal length Tibia Length Midshaft circumference NF1 Proximal epiphysis breadth Distal epiphysis breadth Fibula length Humerus Length Vertical head diameter Transverse head diameter Distal epiphysis breadth Ulna length Radius length Scapula Height Breadth Glenoid cavity length Spine length Clavicle length Pelvis Greater sciatic notch Auricular surface Ischial spines Ilium blade orientation Ventral arc Pubic symphysis Sternum Manubrium length Mesosternum length Sacrum body:wing ratio Maxillary dentition Mandibular dentition Bilateral Reference Yes Yes Yes Yes Yes Yes Yes No No No No No Bass, 1995 Buikstra and Ubelaker, 1994 Buikstra and Ubelaker, 1994 White and Folkens, 1991 Morse et al., 1983 Schwartz, 1995 Schwartz, 1995 Buikstra and Ubelaker, 1994 Schwartz, 1995 Morse et al., 1983 Buikstra and Ubelaker, 1994 Schwartz, 1995 Yes Yes Yes Yes Yes Yes Thieme, 1957 Dwight, 1905; Stewart, 1979 Seidemann et al., 1998 Black, 1978; DiBennardo and Taylor, 1979 Olivier, 1969 Pearson, 1917–1919 Yes Yes Yes Yes Yes Olivier, 1969 Isçan and Miller-Shaivitz, 1984a,b Isçan et al., 1994 Isçan et al., 1994 Ubelaker, 1989 Yes Yes Yes Yes Yes Yes Thieme, 1957 Krogman, 1962 Krogman, 1962; Dittrick, 1979 France, 1983 Ubelaker, 1989 Ubelaker, 1989 Yes Yes Yes Yes Yes Krogman, 1962 Krogman, 1962 Dwight, 1894; Stewart, 1979 Olivier, 1969 Thieme, 1957 Yes Yes Yes Yes Yes Yes Buikstra and Ubelaker, 1994 Weaver, 1980 Schwartz, 1995 Schwartz, 1995 Phenice, 1969 Buikstra and Ubelaker, 1994 No No No No2 No2 Jit et al., 1980 Jit et al., 1980 Anderson, 1962 Buikstra and Ubelaker, 1994 Buikstra and Ubelaker, 1994 1 NF, nutrient foramen. Despite the fact that both sides of the dentition are scored in the methodology, the individual tooth scores are collapsed into a single unit score for each arcade; therefore, dental scores are not treated in a bilateral manner. 2 Any cultural or pathological alterations (attrition, congenital absence, premortem loss, carious destruction, calculus formation, dental mutilation, or ablation) were disregarded. Resorbed or abscessed alveolar regions were not scored as deterioration as defined above. A tooth still located in the jaw was scored as “present” if at least 50% of the crown area was intact. For loose teeth, at least one complete root and at least 50% of the occlusal surface area must be observable to be considered “present.” The dental scoring procedure for subadults is, by its very nature, more complicated and somewhat more subjective than for adults. Although the scor- ing criteria are identical, most subadult arcades contain a combination of deciduous and adult teeth, the latter of which are often partially formed and/or erupted. To score immature arcades, we determined the eruption age of each individual (see Ubelaker, 1989, his Fig. 71) and then generated a completeness score based on the expected number of teeth present for an individual of that age. For individuals less than age 5 years, the expected number of teeth would be 10; missing or damaged alveolar areas would result in the subtraction of 0.1 per tooth. For individuals in the lower end of this age range (⬍9 months), no teeth would be erupted and we simply 20 C.M. STOJANOWSKI ET AL. TABLE 2. API descriptive statistics grouped by sex and presence of brain matter1 considered the completeness of the alveolar regions. Complications arise with the eruption of the first permanent molar. For example, at 6 years, the expectation is for 12 teeth per arcade, a pattern that generally remains consistent through 11 years. For these individuals, a missing tooth would result in the subtraction of 0.08 per tooth. However, different teeth would be expected at different intervals in this age range. Furthermore, a missing deciduous tooth would not be scored as “absent” if the corresponding adult replacement were present. Because individuals in the 6 –11-year range are the most difficult to generalize, common sense guided most of our scoring decisions. After the second adult molar erupted, the effective number of expected teeth was 14 until the individual reached full adult dental status. After the entire suite of traits is scored, an individual score (IS) is computed for each burial. The IS is simply the sum of “present” features plus the individual maxillary and mandibular scores divided by 80, the total number of observations. The IS is always a positive number between 0 –1, and the average IS is the average preservation index (API) for the sample. The API, therefore, serves as a quantitative, semicontinuous measure of preservation. 1 min, minimum; max, maximum; 95%⫾, 95% confidence interval bounds; SE (m), standard error of the mean; SD, standard deviation. Statistical methods and research methodology RESULTS To test for demographic biases, we calculated the preservation scores by sex and by age groups (0 –5, 6 –10, 11–15, 16 –20, and 21–35 males; 21–35 females; 36 –50 males; 36 –50 female; 50⫹ male; and 50⫹ female). If skeletal decay is a relatively continuous process, then a significantly lower API score indicates that individuals from the left tail of that demographic group’s distribution are absent from the archaeological record. Parametric t-tests and nonparametric Mann-Whitney rank tests were used to analyze sex-specific data. To analyze data by age, we utilized ANOVA and the nonparametric equivalent Kruskal-Wallis test. We also calculated the Pearson product-moment correlation coefficient between IS and the point estimate for the age of each individual. These correlation coefficients were tested against zero, using post hoc Bonferroni alpha protection. To evaluate the relationship between element weight and density and bone preservation, we calculated the frequency of preservation for each trait listed in Table 1. This was accomplished by dividing the number of “present” features by the total number of observations (for the entire sample) for each trait. To determine an overall average completeness score for each element with multiple traits (cranium, femur, and tibia), we averaged the individual feature scores. For example, for a hypothetical element with three distinct scoring features (each with frequencies of 50%, 75%, and 100%), the overall element score would simply be the average of these, or 75%. We investigated the relationship between brain preservation and skeletal condition using a t-test The preservation scores for the Windover sample are presented in Tables 2 and 3. There is surprisingly little variation in mean API score regardless of age or sex. Ignoring subdivision by sex or age provides a measure of the average preservation score for the entire sample. The API was 0.53 (n ⫽ 109, SD ⫽ 0.22), indicating that the Windover sample is fairly well-preserved, with 53% completeness for the features we scored. However, as supported by Tuross et al. (1994), the range of scores is suggestive of the degree of variability seen in skeletal preservation in the Windover materials. The overall range of individual preservation varies from 0 –95%. Using the data presented in Table 2, we examined the relationship between estimated sex (or general body size) and API score. A two-sample t-test (t ⫽ 0.26; df ⫽ 71; P ⫽ 0.79) and the Mann-Whitney test (P ⫽ 0.77) failed to reach significance. We submitted the data in Table 3 to a one-way ANOVA and found no significant difference in preservation between age groups (F ⫽ 1.09; P ⫽ 0.37). The Pearson correlation coefficient between age group and API equals 0.31, indicating that age group explains 9% of the variability in API score. The Kruskal-Wallis analysis of variance test was also not significant (P ⫽ 0.33, assuming chi-square with 9 degrees of freedom). The correlation between age and preservation was not significant (r ⫽ 0.10, P ⫽ 0.30). Therefore, we conclude that the demographic integrity of the Windover sample is intact: individuals of both sexes and all ages display, on average, similar preservation proportions. In Table 4, we present the average preservation scores and average element weights (whole bones) Sex n Min Max Mean 95%⫹ 95%⫺ SE (m) SD Brain Male Female Unknown Yes No 38.0 0.16 0.94 0.54 0.61 0.47 0.04 0.22 29.0 0.19 0.95 0.52 0.59 0.44 0.04 0.21 34.0 0.03 0.91 0.48 0.56 0.40 0.04 0.22 86.0 0.0 0.96 0.55 0.59 0.51 0.02 0.20 23.0 0.04 0.91 0.48 0.59 0.36 0.06 0.26 and a nonparametric Mann-Whitney test. To analyze the relationship between horizontal and vertical burial location and preservation score, we calculated Pearson correlation coefficients for the following comparisons: API vs. north coordinate, API vs. east coordinate, and API vs. meters above sea level (MASL). The correlations were tested against zero with post hoc Bonferroni alpha protection. 21 DIFFERENTIAL WINDOVER PRESERVATION 1 TABLE 3. API descriptive statistics for age cohorts n Min Max Mean 95%⫹ 95%⫺ SE (m) SD 1 0–5 6–10 11–15 16–20 20–35, male 20–35, female 35–50, male 35–50, female 50⫹, male 50⫹, female 10.0 0.22 0.76 0.42 0.57 0.26 0.07 0.21 14.0 0.04 0.86 0.50 0.64 0.36 0.07 0.25 7.0 0.48 0.83 0.48 0.67 0.29 0.08 0.21 6.0 0.49 0.88 0.64 0.78 0.49 0.06 0.14 10.0 0.20 0.83 0.61 0.75 0.48 0.06 0.19 6.0 0.22 0.88 0.57 0.79 0.35 0.09 0.21 15.0 0.25 0.89 0.50 0.58 0.41 0.04 0.16 12.0 0.26 0.93 0.60 0.73 0.47 0.06 0.20 15.0 0.12 0.96 0.58 0.72 0.44 0.07 0.25 9.0 0.31 0.91 0.49 0.63 0.34 0.06 0.19 min, minimum; max, maximum; 95%⫾, 95% confidence interval bounds; SE (m), standard error of the mean; SD, standard deviation. TABLE 4. Windover API scores compared against average element weights (grams) Element Average API Average weight1 Cranium Femur Tibia Pelvis Humerus Scapula Ulna Fibula Radius Clavicle Sternum 0.84 0.58 0.49 0.39 0.47 0.22 0.55 0.27 0.58 0.30 0.11 514 255 154 113 92 41 38 35 31 15 13 1 Average weights in grams based on data published by Lowrance and Latimer (1957, their Table 1). derived from a modern reference sample (Lowrance and Latimer, 1957). The correlation between weight and preservation is statistically significant (r ⫽ 0.77; P ⫽ 0.005), indicating that the least well-preserved elements were, in general, also the lightest. For example, the cranium is the best preserved unit (84%), followed by the femur, radius, ulna, and tibia, respectively. The least well-preserved elements were the scapula (22%) and sternum (11%), two of the lightest and least dense elements in the body. A consideration of differential preservation within a single element produces similar results (Table 5). For example, the scapular body is the thinnest portion of the skeleton and presented the lowest overall preservation score in the sample (scapula height API ⫽ 2%). However, the dense glenoid cavity (64%) preserved at levels higher than most other noncranial traits, with the possible exception of long bone shaft fragments. In general, the data in Table 5 confirm previous analyses documenting a relationship between bone density and preservation (see Galloway et al., 1997; Mays, 1991; Spennemann, 1992; Waldron, 1987). In Table 2, we also present data on the relationship between brain and skeletal preservation. These data indicate little difference in preservation score between those individuals with preserved brains and those without. A t-test grouped by brain preservation (t ⫽ ⫺1.46; df ⫽ 107; P ⫽ 0.15) and the Mann-Whitney test (P ⫽ 0.22) were not significant. However, a different pattern emerged when cranial scores were tabulated independently of mandibular, maxillary, and postcranial scores. The data in Table 6 indicate a stark contrast between postcranial and cranial/dental preservation. As expected, the dental remains exhibited the highest preservation score, with the mandibular dentition exhibiting slightly higher (though not statistically significant) completeness scores than the maxillary remains. Cranial preservation was comparable to both dentitions, however, as indicated in Table 7. Only the maxillary and cranial scores were significantly correlated (r ⫽ 0.67; P ⬍ 0.0001). In terms of preservation of brain tissue, significant correlations were found between preservation of brain tissue and cranial preservation (r ⫽ 0.43; P ⬍ 0.0001), and maxillary preservation (r ⫽ 0.52; P ⬍ 0.0001). Parametric significance tests adjusted for familywise error indicated significant differences between those individuals with brains and those without when considering the cranial (P ⬍ 0.0001) and maxillary (P ⬍ 0.0001) scores; the mandibular (P ⫽ 0.02) and postcranial scores (P ⫽ 0.28) were not significantly different when Bonferroni adjustments were made. These data suggest that the remarkable preservation of brain tissue is related to how well the brain was encased by the cranial architecture, an unsurprising result. Finally, we examined variability in preservation across the site and through the stratigraphic column. As the results indicate (Table 8), north and east burial coordinates are not correlated with preservation. However, the correlation between API and meters above sea level is significant (r ⫽ ⫺0.31; P ⫽ 0.005), indicating that preservation decreased as meters above sea level increased. DISCUSSION In this paper, we tested several hypotheses regarding the causes of differential preservation in the Windover sample. The sample exhibited a wide range of preservation scores, indicative of the degree of preservation variability. As a whole, preservation for the entire sample was approximately 53%. Of the factors intrinsic to the bones themselves, we found little evidence for age- or sex-specific differences in preservation score. This result supports the notion of Buikstra (1997) that the demographic effects of poor preservation may only be experienced in the most marginal situations. Although the lack of a sex correlation is not surprising and confirms the results 22 C.M. STOJANOWSKI ET AL. TABLE 5. Percentage of observable features for Windover sample1 Feature Cranial paired Superciliary arch Supraorbital margin Mastoid process Temporal lines Root of zygomatic Gonial angle Orbital shape Postcranial paired Femur Length Head diameter Neck diameter Midshaft Circum at NF Bicondylar breadth Popliteal length Tibia Length Midshaft Circum at NF Proximal epiphysis breadth Distal epiphysis breadth Fibula length Humerus Length Vertical head diameter Transverse head diameter Distal epiphysis breadth Ulna length Radius length Scapula Height Breadth Glenoid cavity length Spine length Clavicle length Pelvis Greater sciatic notch Auricular surface Ischial spines Ilium blade orientation Ventral arc Pubic symphysis 1 Left (%) Right (%) 89 88 91 85 92 92 49 93 93 93 82 94 89 48 53 53 71 90 32 57 56 44 74 88 25 53 48 80 29 44 29 46 83 24 37 26 46 38 32 65 53 54 44 37 37 75 57 61 2 9 64 13 27 2 13 59 17 32 63 59 36 29 28 22 69 59 42 35 33 25 Feature (%) Cranial unpaired Nuchal crest Mandibular profile Frontal profile Mental eminence Palate size and shape 82 90 92 89 66 Postcranial unpaired Sternum Manubrium length Mesosternum length Sacrum Ratio of body:wings Dental traits Mandible Maxilla 13 9 32 92 86 NF, nutrient foramen; Circum, circumference. TABLE 6. API descriptive statistics for skeletal and dental subgroupings1 n Min Max Mean 95%⫹ 95%⫺ SE (m) SD Crania Maxilla Mandible Postcrania 102.0 0.0 1.0 0.77 0.83 0.71 0.03 0.29 102.0 0.0 1.0 0.80 0.87 0.74 0.03 0.33 98.0 0.0 1.0 0.90 0.94 0.86 0.02 0.20 102.0 0.0 0.95 0.41 0.46 0.36 0.02 0.24 1 Min, minimum; Max, maximum; 95%⫾, 95% confidence interval bounds; SE (m), standard error of the mean; SD, standard deviation. of independent analyses from other samples (Boddington, 1987; Buikstra, 1997; Henderson, 1987; Nawrocki, 1995), given the range of variability and the seemingly mediocre API for this sample, it is somewhat surprising that an age-specific bias was not evident. Another result from this study was the lack of correlation between overall skeletal preservation and the presence or absence of preserved brain material, suggesting that factors responsible for brain preservation were unrelated to skeletal preservation. We did, however, find a significant relationship between cranial condition and brain preservation, suggesting that the cranium protected brain material from the physical and chemical agents responsible for its deterioration in some individuals. The relationship between brain preservation and cranial condition was also evident based on the distribution of brain matter by age; in general, subadults had fewer intact brains than did adults, a fact attributable to the thinner cranial architecture of immature individuals (Doran, 2002). The relationship between preservation and burial location perhaps provides the best explanation for differential preservation at the site. We first examined the relationship between horizontal location and preservation score. There were two possible reasons why position within the cemetery could have affected skeletal condition. First, some evidence suggests that the site was utilized for brief periods of DIFFERENTIAL WINDOVER PRESERVATION TABLE 7. Pearson correlation coefficients and associated p-values for skeletal subgroups Crania Maxilla Mandible Postcrania Brain Crania Maxilla Mandible Postcrania Brain 1.0 1.0 0.68 0.00 0.02 0.85 0.20 0.05 0.43 0.00 1.0 1.0 0.19 0.08 0.05 0.66 0.52 0.00 1.0 1.0 0.19 0.06 0.07 0.48 1.0 1.0 ⫺0.16 0.13 1.0 1.0 TABLE 8. Pearson correlation coefficients and associated p-values for API vs. provenience data API North East Depth 1 API North East MASL1 1.0 1.0 ⫺0.01 0.93 0.09 0.45 ⫺0.31 0.01 1.0 1.0 0.74 0.00 0.00 0.98 1.0 1.0 ⫺0.05 0.68 1.0 1.0 MASL, meters above mean sea level. time over a span of 1,000 years. If groups maintained distinct burial locations, then the placement of a burial within the pond may be indicative of the date it was placed there. If the site was used over a long period of time, then potentially some skeletons have had an additional 1,000 years to deteriorate. A second explanation was a difference in soil chemistry or soil compaction across the site. If pH varied by location, then we expected skeletal condition to vary likewise. However, we found no indication that preservation varies in any consistent way across the site. Although unit-specific pH data are not currently available, the distribution of radiocarbon dates may explain the lack of correlation between preservation and horizontal distribution. One of the more problematic aspects of the Windover site concerns establishing a sequence of burial activities. This results from the fact that for any given segment of the pond, a very early and a very late date can be found within a few meters of one another (Doran, 2002). This may relate to the opportunistic fashion in which final burial location was chosen, a decision possibly based on kin affiliation but also seasonal variation in water depth (Doran, 2002). Assuming that the water table affected placement (for logistical reasons), then two individuals who died in different seasons of the same year could be buried at much different locations in relation to the pond edge. If the water level were low, one would expect the individual to be buried closer to the pond center; likewise, if the water level were high, one would expect a burial to be closer to the shallower pond margins. The same pattern would hold for individuals who died hundreds of years apart: due to water levels in the pond 23 at time of interment, two individuals of vastly different dates could be buried in close proximity. Differences in vertical stratigraphic position provide another potential source of variation in skeletal condition. As with horizontal distribution, there were several ways in which vertical position could have affected bone preservation. The most obvious was the change in soil pH in different stratigraphic levels. Because the soil was more neutral closer to the ground surface, the expectation is for improved preservation in burials located higher in the column (Gordon and Buikstra, 1981; Nawrocki, 1995). However, working against this is the factor of seasonal variation in water depth. If water depth varied after bodies were interred, burials closer to the ground surface would experience more frequent periods of partial drying and subsequent resubmergence, a process that hastens bone deterioration. Our analyses indicate that burials located closer to the modern-day ground surface were less well-preserved, despite the soil pH being more neutral. This supports the latter notion that preservation at Windover Pond may be related to the periodic partial drying of the burial area (Tuross et al., 1994). Despite the unique anaerobic burial context, we found a significant correlation between element weight or density and skeletal preservation, a result supported by previous analyses (Galloway et al., 1997; Mays, 1991; Spennemann, 1992; Waldron, 1987). The consistency of the deterioration process is demonstrated in Table 9, in which we present element- or feature-specific preservation rates based on three comparative samples (Galloway et al., 1997; Spennemann, 1992; Waldron, 1987). For each sample, the traits are ranked in order of completeness, and information on the absolute preservation of each feature is provided. In Table 9, we reproduce the preservation scores from Tables 4 and 5 for ease of comparison. The data of Spennemann (1992) are derived from a recently eroded historic cemetery in the Marshall Islands that had been disturbed by exceptional tidal activity and wave action. The taphonomic context is typified by a high-energy, littoral environment with a relatively short interval for deterioration (Spennemann, 1992). For both samples, the cranium was the best-preserved element, and the scapula the leastpreserved. In addition, the pattern of representation for the least well-preserved elements was identical: the clavicle is slightly better than the sacrum, which is slightly better than the scapula. Of the long bones, the femur and tibia were the best-preserved, followed by the humerus; the preservation of the fibula, radius, and ulna was somewhat variable between samples. In general, the pelvis was less wellpreserved than the long bones. The data of Waldron (1987) were derived from a Romano-British terrestrial cemetery dating to the second to fourth centuries AD. Whereas Spennemann (1992) reported whole bone representation, Waldron (1987) reported preservation frequencies 24 C.M. STOJANOWSKI ET AL. TABLE 9. Comparative preservation data from archaeological and modern reference samples1 Spennemann (1992), trait (%) [Windover %] Waldron (1987), trait (%) [Windover %] 1. Cranium (100%) [85%] 2. Tibia (75%) [49%] 3. Femur (69%) [58%] 4. Humerus (63%) [47%] 5. Fibula (50%) [27%] 6. Pelvis (44%) [39%] 7. Radius/Ulna (38%) [58%/55%] 8. Clavicle (25%) [30%] 9. Sacrum (13%) [32%] 10. Scapula (0%) [10%] 1. Sciatic notch (70%) [63%] 2. Auricular surface (64%) [59%] 3. Prox femur (52%) [60%] 4. Mastoid process (59%) [91%] 5. Dist humerus (56%) [65%] 6. Dist femur (52%) [32%] 7. Glenoid fossa (50%) [64%] 8. Dist tibia (48%) [44%] 9. Prox humerus (47%) [35%] 10. Prox tibia (44%) [29%] 11. Pubic symphysis (31%) [22%] 12. Manubrium (23%) [13%] 13. Mesosternum (22%) [9%] 14. Scapular body (11%) [6%] 1 Galloway et al. (1997), trait (density) [Windover %] 1. 2. 3. 4. 5. 6. 7. 8. Femur midshaft (0.65) [90%] Tibia midshaft (0.48) [80%] Femur neck (0.32) [71%] Dist humerus (0.27) [65%] Femur head/Dist tibia (0.24) [53%/44%] Prox humerus (0.20) [35%] Dist femur (0.18) [32%] Prox tibia (0.16) [29%] Prox, proximal; Dist, distal. for various bone segments and features. We attempted to correlate the landmarks of Waldron (1987) with our own, and only report those traits that correspond reasonably well. The percentage representation data agree well between samples, particularly for the most poorly preserved features. In fact, the last four entries in column 2 of Table 9 correspond exactly in both samples, both in terms of relative preservation and in terms of their rank within the respective sample. It is clear that the pubic region, the sternum, and the scapula are the first to deteriorate. Although there are minor differences in trait representation percentages, several comparative frequencies hold up reasonably well: distal humerus ⬎ proximal humerus; distal tibia ⬎ proximal tibia; and sciatic notch and auricular surface ⬎ pubic symphysis. However, despite the overall similarity in the lists, two features differ rather significantly in representation, the mastoid process and the distal femur. The reasons for this are difficult to discern. Finally, we compared the representation of long bone fragments from the Windover sample to density measures derived from a prepared sample of modern humans (Galloway et al., 1997). The comparison of these data produced the best correspondence and, in fact, the ranked density measures are perfectly correlated with the ranked preservation scores (Table 9, column 3). Midshaft sections of the long bones preserved the best, followed by the distal humerus and proximal femur. Interestingly, the within-element relationships seen in the comparative data of Waldron (1987) are also supported by the density measures: distal humerus ⬎ proximal humerus; distal tibia ⬎ proximal tibia; and proximal femur ⬎ distal femur. Overall, the comparative data suggest that the process of bone deterioration is relatively constant and consistent, regardless of the exact taphonomic context. The skull is the best-preserved unit, followed by long bone shafts, the denser epiphyseal long bone ends, and finally the fragile sternum, clavicle, and scapula. For the pelvis, the denser auricu- lar and sciatic region is better represented than the pubic region (Telmon et al., 1993). Despite the uniqueness of Windover’s taphonomic context, the results from this study seem applicable to most other human skeletal assemblages, including those subjected to hydrodynamic sorting processes in a saline-rich, high-energy coastal zone (Spennemann, 1992). In summary, the analysis of skeletal taphonomy for the Windover sample has accomplished several goals. We quantified the average and range of preservation scores for the entire sample, and determined that age and sex were unrelated to skeletal condition, as was the position of each burial within the pond. However, preservation did decline in burials that were located closer to the modern ground surface. As the water level in the pond fluctuated over the last 8,000 years, burials highest in the stratigraphic column suffered the deleterious effects of drying and resubmergence, which resulted in deterioration of the least dense skeletal regions. No demographic effects were discernible, and it seems likely that the Windover sample is representative of the population that contributed the burials. ACKNOWLEDGMENTS We thank Clark Spencer Larsen and three anonymous reviewers for their helpful comments on the manuscript. Any errors or omissions remain our sole responsibility. LITERATURE CITED Anderson JE. 1962. The human skeleton: a manual for archaeologists. Ottawa: National Museum of Canada. Angel JL. 1969. The bases of paleodemography. Am J Phys Anthropol 30:427– 437. Bass WM. 1995. Human osteology: a laboratory and field manual. 4th ed. Columbia: Missouri Archaeological Society. Black TK III. 1978. A new method for assessing the sex of fragmentary skeletal remains: femoral shaft circumference. Am J Phys Anthropol 48:227–232. Boddington A. 1987. Chaos, disturbance and decay in an AngloSaxon cemetery. In: Boddington A, Garland AN, Janaway RC, editors. Death, decay and reconstruction. Manchester: Manchester University Press. p 27– 42. DIFFERENTIAL WINDOVER PRESERVATION Boddington A, Garland AN, Janaway RC. 1987. Flesh, bones, dust and society. In: Boddington A, Garland AN, Janaway RC, editors. Death, decay and reconstruction. Manchester: Manchester University Press. p 3–9. Buikstra JE. 1997. Paleodemography: context and compromise. In: Paine RR, editor. Integrating archaeological demography: multidisciplinary approaches to prehistoric population. Center for Archaeological Investigations occasional paper no. 24. Carbondale: Southern Illinois University. p 367–380. Buikstra JE, Ubelaker DH. 1994. Standards for data collection from human skeletal remains. Arkansas archeological research series 44. Fayetteville: Arkansas Archaeological Survey. DiBennardo R, Taylor JV. 1979. Sex assessment of the femur: a test of a new method. Am J Phys Anthropol 50:635– 638. Dittrick J. 1979. Sexual dimorphism of the femur and humerus in prehistoric central California skeletal samples. M.A. thesis, Department of Anthropology, California State University, Fullerton. Doran GH, editor. 2002. Windover: multidisciplinary investigations of an Early Archaic Florida cemetery. Gainesville: University Press of Florida. Doran GH, Dickel DN. 1988a. Multidisciplinary investigations at the Windover site. In: Purdy B, editor. Wet site archaeology. New Jersey: Telford Press. p 263–289. Doran GH, Dickel DN. 1988b. Radiometric chronology of the archaic Windover archaeological site (8BR246). Fla Anthropol 41:365–380. Doran GH, Dickel DN, Ballinger WE, Agee OF, Laipis PJ, Hauswirth WW. 1986. Anatomical, cellular and molecular analysis of 8,000-yr-old human brain tissue from the Windover archaeological site. Nature 323:803– 806. Doran GH, Dickel DN, Newsom LA. 1990. A 7,290-year-old bottle gourd from the Windover site, Florida. Am Antiq 55:354 –360. Dwight T. 1894. The range and significance of variation in the human skeleton. Boston Med Surg J 13:73–76, 97–101. Dwight T. 1905. The size of the articular surfaces of the long bones as characteristic of sex: an anthropological study. Am J Anat 4:19 –32. France DL. 1983. Sexual dimorphism in the human humerus. Ph.D. dissertation, Department of Anthropology, University of Colorado, Boulder. Galloway A, Willey P, Snyder L. 1997. Human bone mineral densities and survival of bone elements: a contemporary sample. In: Haglund WD, Sorg MH, editors. Forensic taphonomy: the postmortem fate of human remains. Boca Raton: CRC Press. p 295–317. Gordon CC, Buikstra JE. 1981. Soil pH, bone preservation, and sampling bias at mortuary sites. Am Antiq 46:566 –571. Hauswirth WW, Dickel CD, Doran GH, Laipis PJ, Dickel DN. 1991. 8000-year-old brain tissue from the Windover site: anatomical, cellular, and molecular analysis. In: Ortner DJ, Aufderheide AC, editors. Human paleopathology: current syntheses and future options. Washington, DC: Smithsonian Institution Press. p 60 –72. Henderson J. 1987. Factors determining the state of preservation of human remains. In: Boddington A, Garland AN, Janaway RC, editors. Death, decay and reconstruction. Manchester: Manchester University Press. p 43–54. Isçan MY, Miller-Shaivitz P. 1984a. Determination of sex from the tibia. Am J Phys Anthropol 64:53–57. Isçan MY, Miller-Shaivitz P. 1984b. Discriminant function sexing of the tibia. J Forensic Sci 29:1087–1093. Isçan MY, Yoshino M, Kato S. 1994. Sex determination from the tibia: standards for contemporary Japan. J Forensic Sci 39: 785–792. Jackes M. 1992. Paleodemography: problems and techniques. In: Saunders SR, Katzenberg MA, editors. Skeletal biology of past peoples: research methods. New York: Wiley-Liss, Inc. p 189 – 224. Jit I, Jhingan V, Kulkarni M. 1980. Sexing the human sternum. Am J Phys Anthropol 53:217–224. Johansson SR, Horowitz S. 1986. Estimating mortality in skeletal populations: influence of the growth rate on the interpretation 25 of levels and trends during the transition to agriculture. Am J Phys Anthropol 71:233–250. Konigsberg LW, Frankenberg SR. 1992. Estimation of age structure in anthropological demography. Am J Phys Anthropol 89:235–256. Konigsberg LW, Frankenberg SR. 1994. Paleodemography: “not quite dead.” Evol Anthropol 3:92–105. Krogman WM. 1962. The human skeleton in forensic medicine. Springfield, IL: Charles C. Thomas. Lowrance EW, Latimer HB. 1957. Weights and linear measurements of 105 human skeletons from Asia. Am J Physiol Anat 101:445– 459. Mays S. 1991. Taphonomic factors in a human skeletal assemblage. Circaea 9:54 –58. Milanich JT. 1994. Archaeology of precolumbian Florida. Gainesville: University Press of Florida. Milner GR. 1983. The East St. Louis stone quarry site cemetery: (11-S-468). American bottom archaeology, volume 1. Urbana: University of Illinois Press. Milner GR, Humpf DA, Harpending HC. 1989. Pattern matching of age-at-death distributions in paleodemographic analysis. Am J Phys Anthropol 80:49 –58. Moore JA, Swedlund AC, Armelagos GJ. 1975. The use of life tables in paleodemography. In: Swedlund AC, editor. Population studies in archaeology and biological anthropology: a symposium. Society for American Archaeology memoir no. 30. Washington, DC: Society for American Archaeology. p 57–70. Morse D, Duncan J, Stoutamire J. 1983. Handbook of forensic archaeology and anthropology. Tallahassee: Florida State University Foundation. Nawrocki SP. 1995. Taphonomic processes in historic cemeteries. In: Grauer AL, editor. Bodies of evidence: reconstructing history through skeletal analysis. New York: Wiley-Liss, Inc. p 49 – 66. Olivier G. 1969. Practical anthropology. Springfield, IL: Charles C. Thomas. Paine RR. 1989. Model life table fitting by maximum likelihood estimation: a procedure to reconstruct paleodemographic characteristics from skeletal age distributions. Am J Phys Anthropol 79:51– 61. Paine RR, Harpending HC. 1996. Assessing the reliability of paleodemographic fertility estimators using simulated skeletal distributions. Am J Phys Anthropol 101:151–159. Pearson K. 1917–1919. A study of the long bones of the English skeleton I: the femur. Department of Applied Statistics, company research, memoirs, biometric series X, chapters 1– 4. London: University of London, University College. Phenice TW. 1969. A newly developed visual method of sexing the os pubis. Am J Phys Anthropol 30:297–302. Purdy BA. 1991. The art and archaeology of Florida’s wetlands. Boca Raton: CRC Press, Inc. Riggs BL, Melton LJ III. 1986. Involutional osteoporosis. N Engl J Med 314:1676 –1686. Roth EA. 1992. Applications of demographic models to paleodemography. In: Saunders SR, Katzenberg MA, editors. Skeletal biology of past peoples: research methods. New York: WileyLiss, Inc. p 175–188. Saunders SR. 1992. Subadult skeletons and growth related studies. In: Saunders SR, Katzenberg MA, editors. Skeletal biology of past peoples: research methods. New York: Wiley-Liss, Inc. p 1–20. Saunders SR, Herring DA, Boyce G. 1995. Can skeletal samples accurately represent the living population they come from? The St.Thomas’ cemetery site, Belleville, Ontario. In: Grauer AL, editor. Bodies of evidence: reconstructing history through skeletal analysis. New York: Wiley-Liss, Inc. p 69 – 89. Schwartz JH. 1995. Skeleton keys. Oxford: Oxford University Press. Seidemann RM. 1999. Descriptive analysis of Maya skeletal remains from Fighting Conch Mound, Wild Cane Cay, Belize. M.S. thesis. Department of Geography and Anthropology, Louisiana State University. 26 C.M. STOJANOWSKI ET AL. Seidemann RM, Stojanowski CM, Doran GH. 1998. The use of the supero-inferior femoral neck diameter as a sex assessor. Am J Phys Anthropol 107:305–313. Specker BL, Brazerol W, Tsang RC, Levin R, Searcy J, Steichen J. 1987. Bone mineral content in children 1 to 6 years of age. Am J Dis Child 141:343–344. Spennemann DHR. 1992. Differential representation of human skeletal remains in eroded and redeposited coastal deposits: a case study from the Marshall Islands. Int J Anthropol 7:1– 8. Stewart TD. 1979. Essentials of forensic anthropology. Springfield, IL: Charles C. Thomas. Stini W, Stein P, Chen Z. 1992. Bone remodeling in old age: longitudinal monitoring in Arizona. Am J Hum Biol 4:47–55. Telmon N, Rouge D, Brugna J-F, Sevin A, Larrouy G, Arbus L. 1993. Critères ostéscopiques d’exploration du vieillissement: l’example de la nécropolie médiévale de Saint-Etienne de Toulouse. Bull Mem Soc Anthropol Paris 5:293–300. Thieme FP. 1957. Sex in Negro skeletons. J Forensic Med 4:72– 81. Tuross N, Fogel ML, Newsom L, Doran GH. 1994. Subsistence in the Florida Archaic: the stable-isotope and archaeobotanical evidence from the Windover site. Am Antiq 59:288 –303. Ubelaker DH. 1989. Human skeletal remains: excavation, analysis, interpretation. Washington, DC: Taraxacum. Von Endt DW, Ortner DJ. 1984. Experimental effects of bone size and temperature on bone diagenesis. J Archeaol Sci 11:247– 253. Waldron T. 1987. The relative survival of the human skeleton: implications for paleopathology. In: Boddington A, Garland AN, Janaway RC, editors. Death, decay and reconstruction. Manchester: Manchester University Press. p 55– 64. Walker PL. 1995. Problems of preservation and sexism in sexing: some lessons from historical collections for palaeodemographers. In: Saunders SR, Herring A, editors. Grave reflections: portraying the past through cemetery studies. Toronto: Canadian Scholars’ Press. p 31– 47. Walker PL, Snethkamp P. 1984. Archaeological investigations on San Miguel Island: archaeological and physical anthropological research. Report to the National Park Service, volumes I and II. Ventura: Channel Islands National Park. Walker PL, Johnson JR, Lambert PM. 1988. Age and sex biases in the preservation of human skeletal remains. Am J Phys Anthropol 76:183–188. Walker PL, Drayer FJ, Kerr Siefkin S. 1996. Malibu human skeletal remains: a bioarchaeological analysis. Report to the Resource Management Division. Sacramento: Department of Parks and Recreation. Weaver DS. 1980. Sex differences in the ilia of a known sex and age sample of fetal and infant skeletons. Am J Phys Anthropol 52:191–195. Weiss KM. 1973. Demographic models for anthropology. Washington, DC: Memoirs of the Society for American Archaeology. Memoir no. 27. White TD. 1992. Prehistoric cannibalism at Mancos 5MTUMR2346. Princeton: Princeton University Press. White TD, Folkens PA. 1991. Human osteology. San Francisco: Academic Press. Willey P, Galloway A, Snyder L. 1997. Bone mineral density and survival of elements and element portions in the bones of the Crow Creek massacre victims. Am J Phys Anthropol 104:513– 528.