Gagnon 2004.2 44 45 The Moche of north coastal Perú were among the earliest New World societies to develop a bureaucratic state organization (Moseley, 1992; Bawden, 1996). The Moche State (AD 200–800) was a centralized hierarchical society that controlled the entire Moche Valley and perhaps valleys to the north and south (Fig. 1). The Moche elite marshaled their economic resources to build large public works, such as roads and monumental ceremonial structures (Hastings and Moseley, 1975; Moseley, 1975), and to dramatically increase arable land through canal construction (Moseley and Deeds, 1982). The elite also amassed great personal wealth, as indicated by archaeological excavations of wealthy tombs (Donnan and Castillo, 1992; Alva and Donnan, 1993). To exert their influence, the elite used ideological power manifested in public rituals held at large monuments, and iconography that supported state ideologies. Physical power, in the form of warfare, conquest, and sacrifice, was also central to elite control (Shimada, 1978; Bawden, 1996; Billman, 1996; Bourget, 1996. 2001; Verano, 2001). Prior to the establishment of the state, societies in ABSTRACT The Moche of north coastal Perú were among the earliest New World societies to develop state socio-political organization. The Moche State (AD 200- 800) was a centralized hierarchical society that controlled the Moche Valley as well as valleys to the north and south. Prior to the establishment of the state, a series of less hierarchical organizations were present in the valley. Irrigation agriculture has often been cited as central to development of the Moche State. To test this assertion I examined 750 individuals recovered from the largest cemetery at the site of Cerro Oreja. Although the most important occupation of Cerro Oreja was during the Gallinazo phase (AD 1-200), many individuals were interred here during the earlier Salinar period (400 -1 BC). Consequently, the Cerro Oreja collection holds a key to understanding the development of one of the earliest and most extensive states in the Americas. The teeth and/or alveoli of each individual were examined for the presence of dental caries, periodontal disease, abscesses, and antemortem tooth loss. My analysis suggests women and children did increasingly focus their diet on agricultural products. These findings seem to support the hypothesis that increased irrigation and reliance on agricultural production was fundamental to the development of the Moche state. However, men’s diets remained consistent through time. Status seems to have been of little import in determining diet before and during early periods of state development, in dramatic contrast to what we know of its importance during the zenith of the state’s power. I suggest that increasing differentiation of gender roles was important to the development of the state, and that gender differences may have been the most salient force in the transition to political hierarchy and social stratification in the Moche valley. Dental Anthropology 2004;17(2):45-54. Food and the State: Bioarchaeological Investigations of Diet in the Moche Valley of Perú Celeste Marie Gagnon* Department of Anthropology, University of North Carolina. Chapel Hill, NC 27599 *Address for correspondence: C. M. Gagnon, Department of Anthropology, University of North Carolina at Chapel Hill,108 Alumni Building CB#3120, Chapel Hill, NC 27599-3120 Email:bioarchcel@yahoo.com Editor’s note: Ms. Gagnon’s paper was awarded ‘First Runner Up’ for 2004 in the Albert A. Dahlberg student research competition sponsored by the Dental Anthropology association. Celeste Marie Gagnon 46 47C.M. GAGNON the valley were organized in less hierarchical political structures (Topic and Topic, 1978; Brennan, 1980; Topic, 1982; Billman, 1997, 1999). It was the people of these societies who first opened the desert lands of the Moche valley to agriculture (Table 1). Construction of the valley-wide canal system that enabled agriculture production began during the Cupisnique phase (1800–400 BC), when approximately 4200m3 of canals were built, irrigating 4100 hectares. This system was expanded by approximately 60,000 m3 during the Salinar phase (400–1 BC), allowing for the cultivation of 6750 to 7300 hectares. During the Gallinazo and Early Moche phases (AD 1–400) no new land appears to have been brought under cultivation. Later, the Moche State doubled agricultural production through the irrigation of 12,550–13,200 hectares (Billman, 2002). This dramatic increase in agricultural production has led researchers to suggest that in Perú, as elsewhere, irrigation played a central role in state development (Steward, 1949; Rowe, 1963; Moseley, 1974; Haas, 1987). This is because staple crops could be produced on a grand scale in irrigated fields creating storable surpluses. By controlling these stores, elites financed their state building activities (D’Altroy and Earle, 1985; Earle, 1997). Because increased agricultural production is reflected in an increase in the consumption of agricultural products, the link between irrigation and state development can be tested by tracking prehistoric changes in diet. To this end, I examined individuals who lived during the Salinar and Gallinazo phases, just prior to and during the beginnings of state formation, for evidence of increased prevalence of dental pathological conditions indicative of the increased consumption of starchy and/or sugary agricultural products. An advantage of using biological rather then ethnobotanical data to chart consumption is that these data are linked to specific individuals for whom sex, age-at-death, and status information is known. This allowed me to examine not only changes in agricultural production, but who these changes affected, and thus link changes in social roles, particularly gender roles, to changes in political organization. Fig. 2. Map of the lower and middle Moche valley locating Cerro Oreja, from Billman 1999.Fig. 1. Perú with north coast inset, from Moseley 1992. New Canals Ceremonial Phase Estimated Dates Cultural Horizon1 Excavated Architecture Middle Moche AD 400–700 Early Intermediate Period 312,000 m3 416,000 m3 Gallinazo–Early Moche AD 1–400 Early Intermediate Period 0 m3 15,000 m3 Salinar 400–1 BC Early Intermediate Period 60,000 m3 67,000 m3 Cupisnique 1800–400 BC Initial Period–Early Horizon 42,000 m3 1,291,000 m3 Table 1. Moche Valley cultural periods and work projects 1Following Moseley (1992) and Billman (2002). 46 47FOOD AND THE STATE THE SAMPLE The remains I examined in this study were excavated by the Instituto Nacional de Cultura (INC) from the site of Cerro Oreja (Fig. 2). Cerro Oreja, located at the neck of the Moche valley, was the largest urban center in the valley during the Gallinazo phase (Billman, 1999). Downstream from this location the Moche valley widens as the river drains into the Pacific. At the neck and upstream, the Río Moche cuts into the Andean foothills. Here bottom land is limited and valley slopes rise steeply. It is in this area that people, both in the past and present, have built irrigation canal intakes. Although the most important prehistoric occupation of Cerro Oreja was during the Gallinazo phase, many individuals were also interred here during the Salinar phase. Of the 909 burials excavated from Cerro Oreja by the INC, I examined 681. Several burials contained the remains of more than one individual, and as a result the study sample represents 750 individuals. The INC made phase designations based on burial goods and stratigraphic location. The sample I analyzed for this study included the remains of 61 Salinar, 142 Early Gallinazo, 109 Middle Gallinazo and 69 Late Gallinazo individuals (Table 2). Although the mortuary analysis of the Cerro Oreja cemetery is in its preliminary stages, information about the presence of grave goods is available. Using these data, I divided individuals into two status categories, those with goods and those without. Age-at-death and sex identifications were made as a joint effort by the members of the Cerro Oreja Bioarchaeology Project, which is co-directed by Dr. Patricia Lambert of Utah State University and Dr. Brian Billman of the University of North Carolina at Chapel Hill. During the summer field seasons of 1999, 2000, and 2001, I worked with Dr. Lambert and Bonnie Yoshida, a graduate student at the University of California at Santa Barbara, to make age-at-death and sex estimations for 243 individuals. During an extended research season, Yoshida identified the age-at-death and sex of an additional 183 individuals. In 2003, I examined these individuals as well as the remains of 324 additional individuals. Our age-at-death and sex identifications were made following Standards (Buikstra and Ubelaker, 1994). We based subadult age estimates primarily on tooth formation and eruption (White, 1991). Skeletal development and fusion (Johnston, 1962; Fazekas and Kósa, 1978; Buikstra and Ubelaker, 1994) were used in estimating the age-at-death of fetuses and infants, as well as of individuals for whom the dentition was not preserved. Adult ages were estimated based on combined morphological changes at the pubic symphysis and auricular surface, and also on cranial suture closure (as presented in Buikstra and Ubelaker, 1994). Occasionally sternal rib ends were well-enough preserved to be included in our age assessments (see Bass, 1987). We assigned individuals a mean age and an error estimate. Errors ranged from several months for well-preserved children to as much as 15 years for fragmentary adults. For this analysis, I grouped individuals in five-year categories based on their mean age. Adult remains too fragmentary to be assigned a mean age were not included. Sex identification of adults was established using the Phenice Method and qualitative observations of pelvic morphology, such as relative size of greater Age in years Salinar Early Gallinazo Middle Gallinazo Late Gallinazo fetal 0 5 8 0 birth - 4.9 24 110 90 49 5 - 9.9 11 21 12 9 10 - 14.9 3 12 4 3 15 - 19.9 5 10 11 7 20 - 24.9 2 9 4 5 25 - 29.9 0 9 7 11 30 - 34.9 6 6 10 7 35 - 39.9 8 10 10 3 40 - 44.9 5 10 8 4 45 - 49.9 4 5 5 2 50 - 54.9 0 0 2 0 55 - 59.9 1 0 0 0 > 18 2 5 1 5 > 20 3 20 28 26 > 30 2 11 10 13 > 40 0 2 5 1 50 + 0 0 0 1 Table 2. Study sample 48 49 sciatic notch, length of the pubic ramus and width of the subpubic angle. We also considered cranial morphology and robusticity when sexing adults (White, 1991; Buikstra and Ubelaker, 1994). If the os pubis was extremely fragmentary or absent, we used metric data from the femora, tibiae, and humeri to support cranial sex identifications (Iscan and Miller-Shaivitz, 1984; Dittrick and Suchey, 1986). Not all individuals could be assigned a sex with the same degree of certainty. To incorporate our varying error, we employed a four- tier system to rank our identifications: female/male, probable female/male, possible female/male, and unidentified. Individuals in the last two categories were excluded from this analysis. Approximately 56% of the individuals in this study had preserved teeth and/or alveoli that I could examine for the presence of the following dental pathological conditions: caries, periodontal disease, abscesses, and antemortem tooth loss. METHODOLOGICAL CONSIDERATIONS Biological anthropologists have effectively used the frequency of dental pathological conditions to chart changes in diet, particularly with regard to the consumption of agricultural products (see Kelley and Larsen, 1991; Hillson, 1996; Larsen, 1997). Because of the nature of human dentitions and the vagaries of skeletal preservation, researchers have struggled with several sampling issues. The primary issue is scalar: Should individuals or teeth be the unit of analysis? In an analysis at the individual level, individuals are diagnosed as having or not having a particular condition. Such an analysis can provide researchers with useful information; but it can also obscure variation. At this level, an individual with 32 observable teeth, only one of which is carious, cannot be distinguished from an individual with 32 carious teeth, though these two individuals certainly had different diets. Characterizing individuals by the percentage of affected teeth addresses this issue, but raises questions about sample comparability. An individual with 32 observable teeth, all of which are carious, and an individual with only two teeth, also carious, would both be classified as 100% affected. To mitigate the effects of individual preservation differences, researchers often exclude individuals who do not have some minimum number of observable teeth. In addition to these issues of comparability, substantial data loss occurs when each individual is characterized by only one datum point. In the comparison described above, 32 teeth were observed. However, when these data are analyzed at the individual level, there is only one datum point—the presence (or absence) of carious lesions, or the percentage of affected teeth. This loss can result in sample sizes that are insufficient to address research questions. To maximize data and address comparability difficulties, many researchers analyze their data at a higher scale (following Turner, 1979). All the teeth of many individuals are pooled into groups, the boundaries of which are defined by the questions to be addressed. These types of studies yield some interesting information about diet. However, statistical analyses of data grouped in this way assume that each tooth in the group is independent of all others. This assumption does not hold, as the teeth of an individual are affected by her/his diet and are therefore more likely to resemble each other than the teeth of other individuals. Additionally, the pooling of samples can increase group heterogeneity, leading to spurious results. To test for diet change over time, all individuals from each period can be grouped to create a lesion rate for each. Statistical analysis of these rates may not identify significant differences among groups because the differences that do exist may average out. Summary measures of a bimodal distribution (e.g., female teeth are always affected while male teeth are not) can be very similar to those of a standard distribution (e.g., both female and male teeth are often affected). To address this issue, groups can be more narrowly defined, creating a larger number of groups each of which includes fewer individuals. Doing so, however, decreases sample size and the power of statistical tests to identify differences among the groups. Two other factors complicate both individual and group level analysis: the differing susceptibility of different tooth types to dental pathological conditions, and the varying ages-at-death of sample individuals. Teeth vary in their overall size and the complexity of their shape. Both of these factors affect a tooth’s probability of developing pathological lesions. At the individual level of analysis for example, a person with eight observable anterior teeth is much less likely to display carious lesions than someone with eight observable molars. At the group level, samples pooled for comparison might contain substantially different distributions of tooth types. Dental pathological conditions are also age-dependent. For this reason, researchers segregate subadults and adults in both individual and group level analyses. In some cases the adult sample is further divided into young, middle, and old categories. As with other attempts to mitigate sample heterogeneity, the level of error must to be balanced against decreasing the sample size. To address these persistent problems, I analyzed these data using logistic regression. A log-linear model simultaneously explores the complex relationships between a categorical independent variable, in this case the presence or absence of a dental condition, and any number of numeric and/or categorical dependent variables (e.g, age-at-death, sex). Secondly, it solves the dilemma of analytical scale. Model estimates C.M. GAGNON 48 49 of population parameters are calculated using individual teeth, allowing the largest possible sample size. However, the teeth of an individual remained linked, preserving individual-level information. This nesting sampling strategy adjusts for sample-effects by weighting the value of the independent variable for each tooth according to how much additional information it provides about the individual. Finally, interactions among dependent variables can be examined for significance. If two dependent variables vary simultaneously their interaction will be more significantly associated with the independent variable than either would when independently analyzed. The model I used in this analysis was programmed in SAS with the help of Chris Wiesen, staff statistician for the Odum Institute at the University of North Carolina at Chapel Hill. In this model age-at-death was the only numeric, dependent variable. We treated this variable as numeric because it behaved linearly, even when described in 5-year intervals. The categorical variables we included were period, status, sex, and tooth-type. Period, status, and sex identifications have been discussed above. I defined several tooth- types based on differing susceptibility of each type to various dental conditions. All teeth were assigned to either one of four permanent types (anterior, premolar, first or second molar, third molar), or to one of three deciduous types (anterior, first molar, second molar). This classification allowed us to modify the model so that it accounted for variation in the number of teeth from each type that were observable in each individual. Additionally, tooth-types were weighted based on their expected occurrence in the population (0.375, 0.25, 0.25, and 0.125, 0.6, 0.2, 0.2, respectively). The inclusion of these dependent variables in the model controlled for variation in the analysis without breaking the sample down into multiple subgroups, thus preserving sample size. RESULTS Permanent and deciduous teeth were considered separately in all analyses. The first analytical step was to calculate rates of dental pathological conditions at both individual and tooth levels. I grouped all individuals who had at least one observable tooth by their status, and compared the percentage of people affected by each condition across periods. No significant status differences were identified (Table 3). Similarly, no significant status differences were found when teeth were grouped by status and compared across time (Table 4). Differing patterns were found when adult individuals, and the teeth of adults, were divided into female and male samples and compared across periods (Table 5 and 6). Therefore, a sex-by-period interaction was included in the log-linear model, and individuals of unknown sex were excluded from further analyses. As part of the logistic regression, the Wald statistic was calculated to test the null hypothesis that groups are characterized by similar rates of dental pathological conditions. To create a visual representation of these data, rates of dental conditions, adjusted for sample variations in size, age-at-death, and tooth-type, were plotted. Deciduous teeth affected by carious lesions generally increased through time (Fig. 3), although this pattern is not statistically significant (P < 0.2680). Female adult carious lesions rates show a similar trend (Fig. 4), which is significant (P < 0.0077). Temporal variation in male carious lesion rates are not significant (P = 0.4903), and shows no pattern (Fig. 4). Significant differences were identified between female and males in the Middle and Late Gallinazo phases (P = 0.0051 and 0.0286, Deciduous Permanent Permanent Period Carious Carious Periodontal Permanent Permanent Status Lesions Lesions Disease Abscesses Tooth Loss n % n % n % n % n % Salinar high 28 25 31 74 14 86 17 47 19 68 low 1 0 5 100 3 100 3 67 3 33 Early Gallinazo high 27 37 36 64 6 83 21 43 22 59 low 35 29 42 50 7 71 15 47 16 69 Middle Gallinazo high 10 30 21 71 7 86 15 53 16 63 low 32 44 36 61 12 67 20 35 24 58 Late Gallinazo high 4 25 11 73 4 100 6 17 6 67 low 21 29 18 67 7 43 11 45 11 55 Table 3. Individuals affected by dental pathological conditions FOOD AND THE STATE 50 51 respectively). Adult periodontal disease rates show changes over time for both females (P = 0.2383) and males (P = 0.0871), but the pattern of change is different than that seen in carious lesion rates (Fig. 5). Among both females and males periodontal disease rates increased from the Salinar to the Early Gallinazo phase and then fell from the Early to the Middle Gallinazo. Throughout these periods, female rates were higher than males, but these differences are only marginally significant during the Early Gallinazo phase (P = 0.0961). The female and male patterns dramatically change in the Late Gallinazo phase, when there is a significant increase in periodontal disease among males (P = 0.0286), but female rates continue to fall. Dental abscessing and antemortem tooth loss among females and males display no clear temporal patterns. Females are more often affected by these conditions than males (Figs. 6 and 7). Differences in the rates of abscessing are only statistically significant during the Middle Gallinazo phase (P = 0.0536). Antemortem tooth loss differences between females and males are significant during the Salinar (P = 0.0257) and Middle Gallinazo (P = 0.0286). DISCUSSION What do the various measure of dental pathological conditions tell us about agricultural consumption at the site of Cerro Oreja during the periods preceding the development of the state? Carious lesion rates suggest that adult females, and to a lesser extent children, carbohydrate consumption steadily increase through time (Figs. 3 and 4). This seems to support the hypothesis that an intensification of agricultural C.M. GAGNON Period Deciduous Permanent Permanent Status Carious Carious Periodontal Permanent Permanent Lesions Lesions Disease Abscesses Tooth Loss n % n % n % n % n % Salinar high 263 20 350 18 236 39 360 7 616 13 low 15 0 81 12 69 30 113 7 159 3 Early Gallinazo high 153 16 287 21 62 24 281 12 683 11 low 217 10 293 19 116 45 310 13 717 12 Middle Gallinazo high 62 13 159 30 92 20 170 12 409 11 low 262 15 323 21 47 45 260 7 627 7 Late Gallinazo high 98 11 69 35 54 41 75 7 196 23 low 29 10 154 23 22 73 217 8 478 7 Table 4. Teeth affected by dental pathological conditions Fig. 3. Deciduous carious lesion rates by period. Fig. 4. Carious lesion rates for females and males by period. 50 51 Period Carious Periodontal Permanent Permanent Sex Lesions Disease Abscesses Tooth Loss n % n % n % n % Salinar female 9 87 7 86 9 56 9 89 male 7 86 7 86 7 29 7 43 Early Gallinazo female 12 83 6 100 14 43 14 86 male 16 63 7 86 19 58 20 65 Middle Gallinazo female 15 93 10 80 17 53 20 75 male 7 57 7 86 10 30 11 64 Late Gallinazo female 7 86 8 75 10 40 10 70 male 4 50 3 67 6 17 6 50 Table 5. Individuals affected by dental pathological conditions production, and a correlated increased in staple consumption, was central to the development of the Moche state. However, adult male consumption does not follow this pattern (Fig. 4). Rather, it appears that males consumed fewer starchy and/or sugary staple agricultural products than did children and females, in amounts that did not change over time. An increasingly different diet among males, compared to females and children suggests increasingly differentiated gender roles in society. What kinds of changes in gender roles might have resulted in such different diets? As a point of comparison I offer the Inka state’s policy of mit’a labor, in which men were required to work on large-scale, state-sponsored projects. While taking part in such work parties laborers were supplied with specialized food-stuffs (D’Altroy and Earle, 1985; Hastorf, 1990, 1991, 1993). Similarly, the substantial investment in public construction in the Moche valley throughout the study period would certainly have required elites to marshal and supply a sizable work force (Table 1). Therefore, I suggest that the men of Cerro Oreja were being increasingly drafted by the elite into similar work parties where they were provisioned with, or offered as an enticement, meat and/or marine resources, while women and children continued to tend agricultural fields and consume the staple crops they produced. Periodontal disease rates do not follow the same pattern as carious lesion rates, for either females or males (Fig. 5). This suggests that periodontal disease in the Cerro Oreja sample is not as closely linked to consumption as carious lesion rates. To understand this pattern I examined how non-food items that people put into their mouths can affect the oral environment. FOOD AND THE STATE Period Carious Periodontal Permanent Permanent Sex Lesions Disease Abscesses Tooth Loss n % n % n % n % Salinar female 107 17 82 54 150 1 233 13 male 111 22 90 46 130 6 199 7 Early Gallinazo female 109 32 45 76 148 14 284 18 male 123 28 45 64 161 23 367 17 Middle Gallinazo female 126 43 45 44 152 13 351 15 male 54 15 31 42 83 6 176 10 Late Gallinazo female 59 53 32 34 85 8 184 17 male 54 13 13 77 44 2 96 6 Table 6. Teeth affected by dental pathological conditions 52 53 Coca leaf chewing is a common activity in the Andes with a very long history (Rostworoski, 1988; Plowman, 1985; Allen, 1985, 1988). Because of the stimulant qualities of coca and the corrosive nature of the lime with which coca leaves are chewed, this activity is associated with alveolar resorption, periodontal disease, and the development of carious lesions in the subsequently exposed tooth roots (Langsjoen, 1996; Indriati, 1997; Indriati and Buikstra, 2001). Unfortunately, the poor curation of the Cerro Oreja skeletal collection resulted in the fragmentation of many tooth roots, thus the rate of root lesions could not be compared to that of crown lesions. Such a comparison could have provided support for my interpretation. Archaeological investigations in the Moche valley provide information relevant to the question of coca use at Cerro Oreja. Billman (1996, 1997) has identified an in-migration of highland people into the Moche valley during the beginning of the Gallinazo phase, based on the appearance of sites dominated by highland ceramics. Of particular interest is the highlander occupation of the limited coca growing areas located in the upper portion of the middle valley. The occupation of these areas by highlanders may have resulted in a reduced access to and use of coca by local residents, as indicated by the decrease in periodontal disease from the Early to the Middle Gallinazo period. Billman proposed that these distinctive highland sites were abandoned as highlanders were displaced, absorbed, and/or eliminated by the end of the Gallinazo phase. In the Late Gallinazo coca may again have been available to valley residents, but increasing gender role differentiation resulted in its use by men, not women. Since coca chewing increases work capacity (Allen, 1985; Plowman, 1985), this may again be an example of elites provisioning men as they labor on irrigation and ceremonial construction projects. Beyond coca’s dramatic biological effects, it has, and has had, important religious and social significance to Andeans (Isbell, 1978; Allen, 1985, 1988; Weismantel, 1988). Thus I would suggest, that its effects on work capacity is only part of the story. The Moche valley elite may have offered coca as payment to common men for their labor, and in this way engaged them as willing participants in their state-building projects. Female and male rates of dental abscesses and antemortem tooth loss show little difference and no patterned change over time (Figs. 6 and 7). Because these are not primary conditions, but are the result of untreated dental caries (crown or root), periodontal disease and/or dental wear, this lack of temporally patterned variation may be the result of an “averaging” C.M. GAGNON Fig. 5. Periodontal disease lesion rates for females and males by period. Fig. 6. Abscess rates for females and males by period. Fig. 7. Antemortem tooth loss rates for females and males by period. 52 53FOOD AND THE STATE of the effects of these primary pathological conditions. CONCLUSION In this analysis I tested the hypothesis that the development of the Moche state (AD 200–800), one of the earliest and most hierarchical in the Americas, was based on intensification of irrigation agriculture. This hypothesis is based on the dramatic increase of arable land in the Moche Valley during the Salinar and Gallinazo phases, prior to the development of the state and during the early period of state development (Table 1). As people increased agricultural production, they would also have increased their consumption of starchy and/or sugary agricultural products. My analysis suggests women and children did increasingly focus their diet on agricultural products (Figs. 3 and 4). These findings seem to confirm the hypothesis that increased irrigation and reliance on agricultural production was fundamental to the development of the Moche state. However, the story is more complex. Men’s diets remained consistent through time (Fig. 4). Additionally the data suggest that women and men’s use of coca varied temporally, in significantly different ways (Fig. 5). Status seems to have been of little import in determining diet before and during early periods of the state, in dramatic contrast to what we know of its importance during the zenith of the state’s power. Together these data suggest that increasing differentiation of gender roles was important to the development of the state. Gender differences may have been the most salient force in the transition to political hierarchy and social stratification in the Moche valley. ACKNOWLEDGMENTS Major support for this research was provided by the National Science Foundation (0225011). The University of North Carolina at Chapel Hill Institute for Latin American Studies and Graduate College, Sigma Xi, and The Wenner-Gren Foundation (6623) provided additional funding. I would like to thank Dr. Brian Billman and Dr. Patricia Lambert for the opportunity to work with the Cerro Oreja skeletal material, the Danger Girls for their guidance in producing this document, and Michael Scholl for his unfailing support, which made this project possible. LITERATURE CITED Allen CJ. 1985. Coca and cultural identity in Andean communities. In: Pacini D, Franquemont C, editors. Coca and cocain: effects on people and policy in Latin America. Cambridge: Cultural Survival Inc, p 35-48. Allen CJ. 1988. The hold life has: coca and cultural identity in an Andean community. Washington, DC: Smithsonian Institution Press. Alva W, Donnan CB. 1993. Royal tombs of Sipán. Los Angles: Fowler Museum of Culture History, University of California. Bass WM. 1987. Human osteology: a laboratory and field manual. Columbia: Missouri Archaeological Society. Special Publications, no. 2. Bawden G. 1996. The Moche. Cambridge: Blackwell Publishers. Billman BR. 1996. The evolution of prehistoric political organization in the Moche valley, Peru. Ph.D. dissertation. Santa Barbara: University of California. Billman BR. 1997. Population pressure and the origins of warfare in the Moche valley, Peru. In: Paine RR, editor. Integrating archaeological demography: multidisciplinary approaches to prehistoric population. Carbondale: Center for Archaeological Investigations, Southern Illinois University, p 285- 310. Billman BR. 1999. Reconstructing prehistoric political economies and cycles of political power in the Moche valley, Peru. In: Billman BR, Feinman GM, editors. Settlement pattern studies in the Americas: fifty years since Virú. Washington: Smithsonian Institution Press, p 131-159. Billman BR. 2002. Irrigation and the origins of the southern Moche state on the north coast of Peru. Latin Am Ant 13:371-400. Bourget S. 1996. Excavaciones en la plaza 3a y en la plataforma II de la Huaca de la Luna durante 1966. In: Uceda S, Mujica E, Morales R, editors. Investigaciones en la Huanca de la Luna 1996. Trujillo: Facultad de Ciencias Sociales, Universidad Nacional de La Libertad, p 43-64. Bourget S. 2001. Children and ancestors: Ritual practices at the Moche site of Huaca de la Luna, north coast Peru. In: Benson EP, Cook AG, editors. Ritual sacrifice in ancient Peru. Austin: University of Texas Press, p 93-118. Brennan CT. 1980. Cerro Arena: early cultural complexity and nucleation in north coastal Peru. J Field Archaeol 7:1-22. Buikstra JE, Ubelaker DH, editors. 1994. Standards for data collection from human skeletal remains. Fayetteville: Arkansas Archeological Survey. Research Series, No. 44. D’Altroy T, Earle TK. 1985. Staple finance, wealth finance, and storage in the Inka political economy. Current Anthrop 26:187-206. Dittrick J, Suchey JM. 1986. Sex determination of prehistoric central California skeletal remains using discriminate analysis of the femur and humerus. J Am Anthrop 70:3-9. Donnan CB, Castillo LJ. 1992. Finding the tomb of a Moche priestess. Archaeol 45:38-42. Earle T. 1997. How chiefs come to power. Stanford, CA: Stanford University Press. 54 55 Fazekas IG, Kósa F. 1978. Forensic fetal osteology. Budapest: Akadémiai Kiadó. Haas J. 1987. The exercise of power in early Andean state development. In: Haas J, Pozorski S, Pozorski T, editors. The development and origins of the Andean state. Cambridge: Cambridge University Press, p 31-35. Hastings CM, Moseley ME. 1975. The adobes of Huaca del Sol and Huaca de la Luna. Am Ant 40:196-203. Hastorf CA. 1990. The effect of the Inka state on Sausa agricultural production and crop consumption. Am Ant 55:262-290. Hastorf CA. 1991. Gender, space, and food in prehistory. In: Gero JM, Conkey MW, editors. Engendering archaeology, women and prehistory. Cambridge: Blackwell Publishers, p 132-159 Hastorf CA. 1993. Agriculture and the onset of political inequality before the Inka. Cambridge: University Press. Hillson S. 1996. Dental anthropology. Cambridge: Cambridge University Press. Indriati E. 1997. The effects of coca chewing on teeth and the Prehispanic distribution of coca chewing. Am J Phys Anthropol suppl. 24:134-135. Indriati E, Buikstra JE. 2001. Coca chewing in prehistoric coastal Peru: dental evidence. Am J Phys Anthropol 114:242-257. Isbell BJ. 1978. To defend ourselves: ecology and ritual in an Andean village. Prospect Heights: Waveland Press. Iscan MY, Miller-Shaivitz P. 1984. Determination of sex from the tibia. Am J Phys Anthropol 64:53-57. Kelley MA, Larsen CS, eds. 1991. Advances in dental anthropology. New York: Wiley-Liss. Johnston FE. 1962. Growth of the long bones of infants and young children at Indian Knoll. Am J Phys Anthropol 20:249-253. Langsjoen OM. 1996. Dental effects of diet and coca-leaf chewing on two prehistoric cultures of northern Chile. Am J Phys Anthropol 101:475-489. Larsen CS. 1997. Bioarchaeology: interpreting behavior from the human skeleton. Cambridge: Cambridge University Press. Moseley ME. 1975. The maritime foundations of Andean civilization. Menlo Park: Cummings Publishing Company. Moseley ME. 1974. Organizational preadaptation to irrigation: The evolution of early water- management systems in coastal Peru. In: Downing TE, Gibson M, editors. Irrigation’s impact on society. Tucson: University of Arizona. Anthropological Papers of the University of Arizona, No. 25, p 77-82. Moseley ME. 1992. The Incas and their ancestors. London: Thames and Hudson. Moseley ME, Deeds EE. 1982. The land in front of Chan Chan: agrarian expansion, reform, and collapse in the Moche valley. In: Moseley ME, Deeds EE, editors. Chan Chan: Andean desert city. Albuquerque: University of New Mexico Press, p 25-53. Plowman T. 1985. Coca chewing and the botanical origins of coca (Erythroxylum spp.) in South America. In: Pacini D, Franquemont C, editors. Coca and cocaine: effects on people and policy in Latin America. Cambridge: Cultural Survival Inc, p 5-33. Rostworowski MDC. 1988. Conflicts over coca fields on Xvth-Century Perú. Ann Arbor: Memoirs of the Museum of Anthropology, University of Michigan, No. 1. Rowe J. 1963. Urban settlements in ancient Peru. Ñawpa Pacha 1:1–27. Shimada I. 1978. Economy of a prehistoric urban context: commodity and labor flow at Moche V Pampa Grande, Peru. Am Ant 43:569-592. Steward JH. 1949. Cultural causality and law: a trial formulation of the development of early civilizations. Am Anthropol 51:1–27. Topic JR, Topic TL. 1978. Prehistoric fortification systems of northern Perú. Current Anthropol 19:618-619 Topic TL. 1982. The early intermediate period and its legacy. In: Moseley ME, Deeds EE, editors. Chan Chan: andean desert city. Albuquerque: University of New Mexico Press, p 255-284. Turner CG. 1979. Dental anthropological indications of agriculture among the Jomon people of Central Japan. Am J Phys Anthropol 51:619-636. Verano JW. 2001. The physical evidence of human sacrifice in ancient Peru. In: Benson EP, Cook AG, editors. Ritual sacrifice in ancient Peru. Austin: University of Texas Press, p 165-184. Weismantel MJ. 1988. Food, gender, and poverty in the Ecuadorian Andes. Prospect Heights: Waveland Press. White TD. 1991. Human osteology. New York: Academic Press. C.M. GAGNON