- Social Sciences
- African Studies
- American Studies
- Asian Studies
- Communication Sciences
- Ethnic Studies
- European Studies
- Gender Studies
- Physical Sciences
- Life Sciences
- Animal Communications
- Cell Biology
- Evolutionary Biology
- Food Science and Technology
- Human Anatomy
Irrigation and the Origins of the Southern Moche State on the North Coast of Peru
by Brian R. Billman
Irrigation and the Origins of the Southern Moche State on the North Coast of Peru
Brian R. Billman
Latin American Antiquity
Updated: October 24th, 2012
IRRIGATION AND THE ORIGINS OF THE SOUTHERN MOCHE STATE ON THE NORTH COAST OF PERU
Brian R. Billman
In this paper, I examine the role that irrigation played in theformation ofthe Southern Moche state in the Moche Valley,Peru. Speci$cally, I attempt to test Wittfogel and Steward's hydraulic model, which postulates that in certain arid environments, the managerial requirements of construction and maintenance of irrigation systems played a crucial role in theformation of centralizedpolities. I formulate and evaluate four hypotheses concerning the role of irrigation systems in the Moche Valley. Those hypotheses are then evaluated using settlement pattern data drawnfrom two surveys that cover the entire coastal section ofthe valley and provide information on 910 archaeological sites. Based on those data, Ipresent a sequence of political development for the valleyfrom theformation ofthe jirst autonomous village in the Late Preceramic period (2500-1800 B.C.) to the zenith of the Southern Moche state. Evaluation of thefour hypotheses indicates that the managerial requirements of irrigation were relatively unimportant; rather, warfare, highland-coastal interaction, and political control of irrigation systems created opportunities for leaders toform a highly centralized, territorially expansive state sometime between A.D. 200 and 700.
En el presente articulo, examino el rol quejug6 la irrigacidn en laformacidn del estado Moche Sureiio en el valle de Moche, Peru'. EspecljTcamente,buscoprobar /as hipdtesis hidraulicas de Wittfogely Steward, /ascuales postulan que en ciertos medio ambientes ciridos, la necesidad. de una administracidn de la construccidn y el mantenimiento de 10s sistemas de irrigacidnjugaron un rol crucial en laformacidn de organizaciones centralizadas. Formulo y evalu'o cuatro hipo'tesis tomando en consideracidn el rol de que de 10s sistemas de irrigacidn jugaron en laformacidn del estado Moche Sureiio. Los datos del patrdn de asentamiento son usados para evaluar /as hipdtesis. Los datos han sido recopilados a travks de dos prospecciones, /as cuales cubren foda la seccidn costeiia del valley brindan informacidn de mcis de 910 sitios. Basado en dichos datos, presento una secuencia de desarrollo politico para el valle desde laformacio'n de la primera villa autdnoma en el periodo Precercimico Tardio (2500-1800 a.C.) hasta el apogeo del estado Moche Sureiio. Los ancilisis indican que 10s requisitos de una administracidn de la irrigacidn no tuvieron una importancia relativa en laformacidn del estado Moche Sureiio;por el contrario 10s conflictos, la interaccidn costa-sierra, y el control politico de 10s sistemas de irrigacidn crearon oportunidades para que 10s lideresformaran un estado altamente centralizado y territorialmente expansivo en algu'n momento entre 10s 200 y 700 d.C.
How does leadership change from its charis-exercised considerable economic, military, and idematic, sporadic, and personal form to become
ological power (Bawden 1996:108-168; Billman
centralized, institutionalized, and permanent?
1996:311-335; Moseley 1992:162-184; Shimada
1994;Topic 1982;Uceda 1997).Leadersof this state
n the middle part of the Early Intermediateperiod directed the construction of massive pyramids, led (EIP)(400B.C-A.D. 800),a large,regional state the conquest of a large section of the north coast of formed in the Moche Valley on the north coast Peru,and organized the productionof unprecedented of Peru. Centered at the site of Moche, this polity is amounts of finely crafted gold objects, pottery, and arguablyone of the earliest regional statesto emerge textiles. in theAndean region. Referred to here as the South-Irrigationhas often been cited as an importantfacem Moche state, it was a highly centralized, hierar-tor in the development of state-levelsocieties on the chically organized political system in which leaders central Andean coast (Haas 1987; Moseley 1974;
Brian R. Billman Department of Anthropology, 301 Alumni Bldg, CB 3115, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-31151. Email: firstname.lastname@example.org
Latin American Antiquity, 13(4), 2002, pp. 37 1400 Copyright0 2002 by the Society for American Archaeology
Approximatm limits of Arm of modern cultlvatlon
lower valley suney -Wash -A~proxlmate llmita of
A Modam sattlmm~nt m~ddlm vallmy survay
Figure 1. Location of lower- and middle-valley surveys, Moche Valley, Peru. The lower-valley survey was conducted by Michael Moseley during the Chan Chan-Moche Valley Project; the middle-valley survey was conducted by the author in
1990 and 1991.
Sanders and Marino 1970; Steward 1949, 197 1 ;cf. Lanning 1967; Rowe 1963). Because the coast lacks any significant annual rainfall, agriculture is depen- dent upon irrigation. Consequently, since its intro- duction around 1800 B.C., irrigation has been essential to the survival of sedentary populations and centralized polities. Few attempts, however, have been made to systematically evaluate the role that irrigation played in political evolution on the central Andean coastline.
I use Wittfogel and Steward's hydraulic model as a heuristic device for examining the origins of the Southern Moche state. The hydraulic hypothesis pos- tulates that the managerial requirements of con- struction and maintenance of irrigation systems played a crucial role in the formation of state-level polities in certain arid environments. I formulate a series of four testable hypotheses concerning the role of irrigation systems that are then evaluated using settlement pattern data from my survey of the mid- dle Moche Valley and Michael Moseley's survey of the lower valley (Figure 1) (Billman 1996). I present a sequence of political development from the for- mation of the first autonomous village in the Late Preceramic period to the zenith of the Southern Moche state. Analysis of the development of irriga- tion and political centralization indicates that the managerial requirements of irrigation were proba- bly relatively unimportant; rather, I propose that war- fare, highland-coastal interaction, and political control of irrigation systems created opportunities for leaders to form a highly centralized state.
A Single Moche State, Two Moche States, or Many Moche Chiefdoms?
Definitions of a state are almost as varied as defini- tions of culture. Researchers studying different aspects of the state stress different characteristics. Although definitions vary, they are in agreement that the state is a general form of social organization that is distinct from other forms in that states (1) have relatively permanent institutionalized forms of lead- ership (not ephemeral rule by charismatic leaders);
(2) have specialized, hierarchical administrative structures (states are bureaucratic); (3) encompass fixed territories (they are organized on the basis of land, not just kinship); (4) use some form of tribute collection to finance political activities; and (5) exer- cise political control through the use of positive and negative sanctions.
Studies conducted over the last three decades demonstrate that Moche political organizations probably had all five of these key attributes of the state (Bawden 1996; Billman 1996; Mosley 1992; cf. Schaedel 1985). The Southern Moche state appears to have existed for 400 to 600 years. Archaeological evidence shows considerable con- tinuity in iconography, ceremonial architecture, and administrative structure throughout its existence (Topic 1977, 1982; Uceda 1997). The hierarchical administrative structure of the polity and its terri- torial limits are indicated by the distribution of administrative centers that have at least four tiers of hierarchy (Billman 1996:3 12-3 16; Conrad 1978; Donnan 1973; Topic 1977, 1982; Willey 1953: 178-233; Wilson 1988:222-224, 1995:200-202). Studies of monumental construction show that trib- ute was collected in the form of labor for massive construction projects (Hastings and Moseley 1975; Moseley 1975b; Topic 1977, 1982). The scope of tribute collection is indicated by Huaca del Sol, which contained over 140,000,000 adobe bricks (Hastings and Moseley 1975; Moseley 1975b). Iconographic representations of elite members of society conducting human sacrifice (Alva and Don- nan 1993; Castillo and Donnan 1994; Donnan 1976) and the discovery of the physical remains of those sacrifices (Bourget 1997,2001; Verano 2001) demonstrate that leaders exercised coercive sanc- tions.
Debate has recently shifted from whether or not there was a Moche state to the question of how many Moche states there were. The long-held view that the Mochica culture was incorporated into a single state centered in the Moche Valley (e.g., Shimada 1987) has been questioned as differences between the Larn- bayeque and Moche Valleys in ceramic style, mon- umental construction, and urban organization during the Moche phase have become apparent. Recently, Shimada (1994: 1-6), Castillo and Donnan (1995), and Bawden (1996:227-262) have proposed the presence of a northern and a southern Moche polity in the Middle Moche phase (Moche I11 and IV), although Shimada still holds out the possibility of a brief period of southern hegemony over the north- ern polity in the latter part of the Middle Moche phase (Shimada 1994:2). Although resolution of the debate will require considerable additional research, current data from the Casma to the ChicamaValleys are sufficient to infer the development of at least one highly centralized, multivalley polity, the Southern Moche state, centered at the site of Moche, some- time between A.D. 200 and 800. Pampa Grande in the Lambayeque Valley probably was the center of a contemporary Northern Moche state (Shimada 1994).
Theories of Irrigation and the
Origins of the State
The notion that early states were a solution (or adap- tation) to social, economic, or environmental prob- lems created by population pressure has influenced theories of state formation in anthropology since the 1940s (Binford 1983; Harris 1977; Johnson and Earle
1987; Sanders et al. 1979; Service 1962,1975,1985; Steward 1955; White 1949). In these formulations- which are known as adaptationalist (Brumfiel and Earle 1987), integration (Haas 1982), or functional models-population growth resulted in resource shortages, thus creating the need for intensification of production. The increased need for centralized planning and management of capital investments in production in turn required the development of cen- tralized political institutions.
Perhaps one of the most influential formulations of this functional view of state formation is Wittfo- gel and Steward's hydraulic hypothesis (Steward 1949, 197 1; Wittfogel 1956, 1957, 197 1). Wittfogel proposed that in certain, although not all instances, the managerial requirements of the construction of large irrigation systems led directly to the develop- ment of centralized, bureaucratic states:
A large quantity of water can be channeled and kept within bounds only by the use of mass labor; and this mass labor must be coordinated, disci- plined, and led. Thus a number of farmers eager to conquer arid lowlands and plains are forced to invoke the organizational devices which411 the
basis of premachine technology--offer the one chance of success: they must work in cooperation with their fellows and subordinate themselves to a directing authority [Wittfogel 1957:18].
Subsequently, numerous authors have suggested that the increased need for centralized planning and management of irrigation systems or other capital investments in production selected for centralized political systems (Cohen 1978, 1981; Johnson and Earle 1987; Sanders et al. 1979; Sanders and Price 1968; Wittfogel 1957). Irrigation is frequently cited as an important factor because of the perceived need for centralized planning and management to (1) con- struct and maintain canals, (2) integrate labor input from multiple households, (3) settle disputes and allocate water, and (4) direct the defense of the sys- tem (see Earle 1978:37-50; Sidky 1996:20-25).
Many researchers have examined the relationship between the construction and maintenance of irri- gation systems and the growth of political authority in a wide range of contemporary and historic soci- eties (Earle 1978; Gray 1963; Hunt and Hunt 1974; Leach 1959; Lees 1974; Mitchell 1973, 1976; Net- ting 1974; Sidky 1996, Spooner 1974). Others have attempted to test the hydraulic hypothesis by exam- ining the temporal relationship between the occur- rence of large-scale irrigation systems and the development of archaic states in Mesoamerica, the Andes, and the Near East (Adarns 1966,1968; Butzer 1976; Lanning 1967; Rowe 1963; Sanders and Marino 1970; Sanders and Price 1968). Results of these assessments have been decidedly mixed. While many investigators have found cases where the development of canal systems did not lead to highly centralized political organizations (Earle 1978; Gray 1963; Mitchell 1976), others have foundcases where the managerial requirements of irrigation apparently did play an important role in political centralization (Emerson 1990; Hunt and Hunt 1974, Sidky 1996).
Functional theories of sociopolitical change have been justifiably criticized for (1) their focus on whole populations rather than individuals; (2) their limited consideration of the internal social dynamics of gen- der, class, and factions; and (3) their frequent use of population pressure as a prime mover (see Brumfiel 1992). However, the notion at the core of these the- ories-that chiefdom and state formation were the result of cooperative action for the benefit of the pop- ulation-remains largely untested. Although func- tional models have fallen into disfavor, the possible cooperative aspects of political formation have resur- faced in the form of dual process theory, which pro- poses that leaders in early chiefdoms and states may have pursued either a cooperative corporate strategy or a more coercive network strategy (Blanton et al. 1996; Earle 2001; Trubitt 2000). In a similar vein, Renfrew (1 974) has written about group-oriented vs. individualizing chiefdoms.
In contrast to this cooperative, functional view of irrigation and political formation, others have emphasized the political rather than the managerial aspects of irrigation (Earle 1978; Fried 1967:207-213; Haas 1981, 1982, 1987; Moseley 1974, 1975a; Stanish 1994). According to political models, differential access to basic resources resulted in social inequality and provided an emerg- ing elite stratum with a source of economic power. By denying or rewarding individuals with access to basic goods or productive facilities, leaders achieved centralized political control of a social group. In addition, control of surplus production from irriga- tion systems provided a means for aspiring leaders to finance a wide range of political economic activ- ities, such as feasting, monumental construction, public rituals, craft production, exchange, trade, or raiding. Carefully manipulated, the collection, redis- tribution, and investment of surplus agricultural pro- duction, which is known as staple finance (D' Altroy and Earle 1985; Earle 1987), allowed leaders to legitimize their elevated status, create networks of political allies, conquer adjacent areas, and make further capital investments in production (Billman 1999: 132-1 35; Earle 1997:75-89; Sidky 1996: 49-75; Stanish 1994:3 14-315). Thus, irrigation sys- tems created opportunities for leaders to expand their political power base and the political econ- omy.
For the central Andean coast, political models have been proposed by Moseley (1974, 1975a) and Haas (1982, 1987). In their formulations, the phys- ical control of irrigation canals by an emerging elite social stratum-not population pressure or manage- ment-led to increased political centralization dur- ing the Initial period (1800-900 B.C.). In short, the degree of political centralization was not related to the scale and complexity of the productive system, but to the potential for individuals to manipulate the productive system to create social inequality and political control.
The formulation of these two contradictory per- spectives-the hydraulic hypothesis and political models (Brumfiel and Earle 1987)-provides a use- ful contrastive tool for the empirical examination of the role irrigation played in the formation of cen- tralized polities. Each of these perspectives empha- size different causal variables: cooperation and the managerial requirements of production vs. social stratification and political control. Arguably, these variables may have worked in tandem to create polit- ical centralization, one of them may have dominated the process, or some other set of factors may have played a crucial role.
Defining the Managerial Requirements of
Irrigation in the Moche Valley
The hydraulic hypothesis proposes that political cen- tralization is directly related to the size of irrigation systems. The primary problem in testing the model is establishing a priori the relationship between the size of an irrigation system and the scale of central- ized polity required to construct and maintain it. Without question, small-scale imgation systems can be constructed and maintained by local cornmuni- ties without formal centralized political organiza- tions (Earle 1978; Gray 1963; Mitchell 1976; Netting 1974; Spooner 1974). At the other extreme, large regional imgation systems clearly do require cen- tralized management (Sidky 1996; Wittfogel 1957). Between these two extremes-small local systems and large regional systems-the hydraulic hypothe- sis should predict a threshold at which formal cen- tralized political structures emerged.
Several functional-oriented theorists have pro- posed or implied threshold values for levels of polit- ical centralization based on group size (Johnson and Earle 1987; Service 1962). Statistical analysis has indicated that to some degree population size and the number of administrative levels in a polity are cor- related (Feinman 1995:259-261; Feinman and Neitzel 1984). Although no theorist has proposed hard-and-fast expectations, functional-oriented the- orists generally agree that communities with popu- lations in the hundreds usually have only one level of political hierarchy, i.e., a village headman. Sim- ple chiefdoms have populations in the thousands and two levels of hierarchy. Complex chiefdoms have three levels of administration and typically have pop- ulations in the tens of thousands. States encompass large regions, control even larger populations, and have more than three levels of administration. The premise of these threshold values is that a certain degree of central management and political control is required for each population level.
By examining three specific types of manage- ment activities required for the construction and use of irrigation systems, these population and spatial thresholds can be applied to the problem of imga- tion and the development of political centralization.
Groups using irrigation systems face three types of management problems: (1) constructing and main- taining canals, (2) integrating households that use a particular canal, and (3) settling disputes and allo- cating water among canal systems (Earle 1978). Analysis of the irrigation system in the Moche Val- ley in terms of these three management activities should indicate the scale of centralized management required during each stage in the development of imgation. Predictions can be derived concerning the development of political centralization and then tested against settlement pattern data.
Examining the role of irrigation in the formation of early states requires estimates of parameters such as labor requirements of canal construction, agri- cultural productivity, size of social groups, and the frequency of environmental perturbations. These are difficult tasks, to say the least. I have used three prin- ciples to cope with uncertainty. First, I have tried to be clear about which aspects of the analysis are least certain. Second, I present arange of estimates, rather than one estimate. Finally, in narrowing the range of estimates, I have deliberately favored estimates that make it more difficult to disprove the hydraulic model, to avoid constructing a deliberately weak or straw-man argument.
Measuring Labor Mobilization
The first set of management activities concerns the size of the labor force mobilized to construct a canal system. The basic premise of the hydraulic hypoth- esis is that when canal construction required small groups, the need for central leadership was minimal and short term. However, when large groups were required for long periods, long-term centralized lead- ership was needed.
The crucial question is the point at which a work- force became too large for informal, ephemeral lead- ership. The functional-oriented population thresholds can be applied to this problem. For instance, if con- structing a canal required the unified effort of no more than several hundred people, the previously described thresholds predict that only one level of administration would have been necessary (e.g., vil- lage headman). If a few thousand people were required, two levels of administration (i.e., a simple chiefdom) would have been needed. Complex chief- dom or state-level organizations would have been needed only when tens of thousands of people were involved.
Construction labor can be estimated from the total volume of canal excavation, which is calculated by multiplying cross-section area by the length of each segment of the canal (Howard and Huckleberry 199 1 ; Ortloff et al. 1985). Because canal cross sections become narrower toward the terminus, cross sec- tions must be measured in segments. In cases where the canal is elevated on an aqueduct, the cross sec- tion is multiplied by the length of the aqueduct. Per- son-days of labor required to construct the canal can then be estimated from the number of cubic meters an individual can excavate per day. 'A difficulty index can adjust excavation rates to varying soil conditions (see Table 3 in Ortloff et al. 1985).
Although simple in theory, estimating canal and aqueduct volume is in practice difficult (Howard and Huckleberry 1991; Ortloff et al. 1985). Fortunately, labor estimates have already been calculated for six lower valley canals and La Cumbre Intervalley canal based on field measurements of prehistoric and mod- em canal cross sections (Ortloff et al. 1985:Table 4; Pozorski and Pozorski 1982:86&867). Calculation of the volume of the remaining modem canals is problematic because canal cross-section data have been published for only eight of 42 canals in the modem system, and these data are limited to the length and depth of one cross section per canal (ONERN 1973:2 13-220).
A rate of 1 m3 per person per day was used in this study. This is significantly less than the rate of 8 m3 per person per eight-hour day used by Ortloff and his colleagues for labor estimates of prehistoric canal construction in the MocheValley (Ortloff et al. 1985). Based on the specific gravity of adobe clay soil, 8 m3 of soil can weigh 10,800 kg or approximately 12 tons (Pozorski 1976:454). In order to excavate 8 m3 of soil in an eight-hour day, a person would have to move 1,350 kg of soil per hour or 22.5 kg per minute, which is equivalent to throwing a 50-lb basket of dirt every minute nonstop for eight hours. Although increased productivity could be achieved by teams of diggers and basket carriers, such a rate seems unattainable.
On the basis of experiments, Erasmus (1965:285) estimated that a team of men using digging sticks could move approximately 2.6 m3 of earth per per- son per day, significantly less than 8 m3. Thomas Pozorski used a rate of I m3 per day for his calcula- tions of the labor investment in the Caballo Muerto Huacas in the Moche Valley (Pozorski 1976:454).
This rate is equal to moving 1,350 kg of dirt per day or 168.75 kg per hour (one 50-lb basket of dirt every eight minutes).
Because rates of excavation vary considerably with soil conditions, I err on the side of caution and use the lowest of the three published rates (1 m3 per person per day). A lower rate favors the hydraulic hypothesis rather than political models because it increases labor requirements for canal construction and, presumably, the need for centralized manage- ment. For instance, at 1 m3 person per day, 1,349 peo- ple could have constructed the prehistoric Vichansao canal in one season. At 8 m3 person per day, the num- ber drops to 169. The Vichansao canal was con- structed in the Middle Moche phase, during which the Southern Moche state formed.
Labor estimates were standardized in terms of person-seasons. There are two agricultural seasons in the Moche Valley (ONERN 1973:225). Like most other central Andean coastal valleys, a majority of the river flow occurs during the first season between December and May (Gillin 1945; Hatch 1976; ONERN 1973:225). During the second season only a fraction of the valley can be planted because of the low volume of river flow. Because considerable sur- plus labor is available in the second season, most canal construction probably took place then. A person-season, therefore, is a useful unit for calculating labor estimates for canal construction. A person-sea- son equals 156 working days, based on six working days per week during the six-month season of the second crop. Based on 1 m3 per person per day, the amount of earth moved in an average person-season is 156 m3.
The next crucial parameter is the size of the social group from which labor was recruited for canal con- struction. If a group of households wished to con- struct a canal, they needed a certain number of able-bodied adults. Stated in another manner, if lead- ers wished to construct one of the canals, they would have needed to unify or control a social group of a certain size. The population from which the labor force was drawn was estimated using a ratio of one adult male to five other people in the population (Hassan 1981:73). If women were involved in canal construction, the social group supplying labor could have been smaller. Once again, I chose to err on the side of caution and favored the hydraulic hypothe- sis by considering only men and thus inferring a larger source population.
Measuring Canal User-Group Size
The second set of management activities concerns the number of people integrated by a single canal. The people tilling the land irrigated by a canal are tied together by sharing that lifeline. If the canal is going to continue to sustain the group, the canal must be regularly maintained, land and water allocated, and disputes among the users mediated. The basic premise of the hydraulic model is that the larger the canal user group, the greater the need for central management.
By using published data on the area irrigated by canals (Moseley and Deeds 1982:Table 2.1; ONERN 1973:2 13-220; Ortloff et al. 1985:Table 4) and crop yields for each period (Billman 1989; Pozorski 1976, 1980; Wilson 1988:326), the maxi- mum population supported by each canal can be cal- culated. These calculations, however, should take into account fallowing. During the prehistoric era, crop rotation and fallowing were probably neces- sary to prevent soil depletion. Modem farmers in the Moche Valley apply between 160 and 600 kg of fer- tilizer per hectare of corn depending on the "eco- nomic power" of the farmer (ONERN 1973: 161-166). Traditional farmers in the town of Moche sustained yields without the use of fertilizers by fal- lowing and rotating the planting of nitrogen-fixing plants such as beans, lentils, and alfalfa (Gillin 1945:15). Fallow periods were short, usually one year or one-half years, and livestock were pastured in the fields. If Gillin's observations are correct, approximately 30 percent of the fields would have been fallow during the prehistoric era.2 Because there is much uncertainty about fallowing on the coast in the prehistoric era, I use two estimates for population supported by canals. One is based on cul- tivation of 100 percent of the land; the other is based on annually fallowing 30 percent.
Water Allocation and Mediating Disputes among Canal User Groups
The third set of management activities deals with dis- pute resolution and water allocation. Unlike the two previous activities, which concern the management of individual canals, these activities involve disputes between canal user groups over water. The annual volume of the Moche River is highly variable. Based on a modem 40-year record of river flow, sufficient water for the cultivation of corn on all 20,888 ha of arable land in the valley is available only 14 out of 40 years (Billman 1996:4 1, 1997:293). Conse- quently, only as the canal system approached the limits of available water would dispute resolution and water allocation have presented problems. Disputes inevitably would have arisen between middle valley and lower valley irrigators. Because middle-valley irrigators draw off water before it reaches the lower valley, in drought years there may have been little left for lower-valley farmers. One means of coping with frequent shortages is the creation of a central- ized administration to allocate water, monitor water distribution, and resolve disputes. The hydraulic hypothesis, therefore, predicts that as demand approached the limits of the water supply, central- ized political organizations should have formed.
The point at which agricultural demand approached the limits of the water supply in the Moche Valley can be estimated by calculating the water requirements of agricultural production at each stage in the expansion of irrigation. The first step was to estimate the area under cultivation during each stage of development. Next, the modem 40-year record of river flow and the water requirements for corn (ONERN 1973: 183,229) were used to estimate the frequency of water shortages during each stage. The use of modem river flow records is problematic; however, local prehistoric proxy records of river flow do not exist. In absence of a local proxy record, I use the modem 40-year re~ord.~
Testing Irrigation Theories of the Origins of the State in the Moche Valley
Having established methods for measuring three cru- cial aspects of the managerial requirements of irri- gation, the history of agricultural expansion in of the Moche Valley must be outlined. Agriculture was first used in the valley in the Late Preceramic period (Pozorski and Pozorski 1979a). Crops are believed to have been grown in sunken fields along the coast and on the floodplain of the river. In those areas crops may not have required canal irrigation, instead deriv- ing water from the water table (Gillin 1945: 16; Mose- ley and Deeds 1982; Pozorski and Pozorski 1979a).
By the start of the Guaiiape phase, farmers turned to intensive irrigation. The irrigation system of the valley continued to expand for the next 3,100 years, reaching a peak in the A.D. 1200s or 1300s (Mose- ley and Deeds 1982). Several studies have been undertaken of the expansion of prehistoric irrigation
Table 1. Stages of the Development of Irrigation in the Moche Valley, Local Chronological Phases, and Regional Time Periods.
Stage of Irrigation
Stage 4, far north side of the lower valley reclaimed and La Cumbre Intervalley canal constructed
Stage 3, most of the north side of the lower valley reclaimed
Stage 2, south side of the lower valley
Stage 1, middle valley and part of the north side of the lower valley reclaimed
Floodplain and sunken field
Local Phase Late and Middle Chimu
Middle Moche Early Moche Gallinazo
Late Salinar Early Salinar
Late Guaiiape Middle Guaiiape Early Guafiape
Late Preceramic period
200-1 B.C. 400-200 B.C.
800400 B.C. 1300-800 B.C. 1800-1300 B.C.
Regional Time Period
Late Intermediate period Middle horizoaate
Early Intermediate period
Early horizon Initial period
Late Preceramic period
Note: Very few radiocarbon dates are available for the Moche Valley. Consequently, dates for the phases in this table are in many cases highly speculative. For instance, no reliable radiocarbon dates exist for the Salinar or Gallinazo phases in the Moche Valley. Furthermore, recent dates from the site of Moche indicate that it was still occupied into the late A.D. 700s (Chapdelaine 1998:lll-114; Chapdelaine et al. 2001:370), which suggests the Middle Moche phase may extend as late as
A.D. 800, pushing the start of the Early Chimu phase forward as much as 200 years.
3.7 km. Canal length in the middle valley is dictated by topographic factors, such as the narrowness of the middle valley and the topography of the foothills of the Andes. Because these factors do not appear to have changed significantly, the current irrigation sys- tem serves as a useful model for estimating the labor requirements of the prehistoric canals.
Canal cross-section data have been published on only three modem middle valley canals: Quirihuac,
7.0 km; Catuay, 5.5 km; and Poroto, 3.7 km (Figure 2) (ONERN 1973:214-216). These canals are the second-, fourth-, and fifth-longest canals, respec- tively, and thus provide some insight into labor requirements for the longest canals. The second- longest canal (Quirihuac) has average dimensions of only 1.25 to 1.50 m wide by .30 to .40 m deep (ONERN 1973:216) for a total volume of excava- tion of approximately 4200 m3. Consequently, a workforce of approximately 27 men could have con- structed the canal in one season (4200 m3/156 m3 per man per season). Based on a ratio of one adult male to five people, a social group of less than 150 peo- ple could have easily mustered a workforce of that size.
Although this seems like a surprisingly small
4 towns canals
r canal floodgate A archaeolog~cal site extent of modern irrigation KM
Figure 2. Locations of modern irrigation canals and key archaeological sites in the Moche Valley, Peru (adapted from Map #6 in ONERN 1973).The middle valley currently is irrigated by 33 small canals, whereas the lower valley is irrigated by only nine canals. In the Middle Moche phase, the Vichansao canal continued to the Rio Seco; in the Late Intermediate period, it was extended onto Pampas Rio Seco and Huanchaco. Also in that period, La Cumbre Intervalley canal joined the Vichansao canal on Pampa Esperanza.
workforce, the rate of excavation expressed in terms Pozorski 1976, 1980; Wilson 1988:326). Therefore, of linear meters of canal excavation is an average of the largest canal in the middle valley had a potential only 1.7 m per person per day. The workforce could maximum population of less than 600 people or 420 have been even smaller if the canal was constructed people if fallowing is considered. The average max- over several seasons. Likewise, if the workforce imum population per canal was less than 140 (less included women and adolescents rather than just than 100 with fallowing). Many canals would have adult men, a social group with less than 150 people supported only a few families. could have constructed the canal. The longest mod- In the Middle and the Late Guaiiape phases, ini- em canal, Jesus Maria, is only slightly longer, with gation expanded into the lower valley; however, cul- a length of 7.3 km. tivation was limited to easily irrigated areas adjacent
The Poroto canal is the fifth-longest canal in the to the river (Farrington 1974; Moseley and Deeds middle valley (ONERN 1973:214-216). It could 1982; Pozorski 1976,1980). The extent of irrigation have been constructed in one season by a workforce on the north side of the river can be estimated based of only 21 men drawn from a social group of approx- on the locations of the two ceremonial centers: imately 105 people. Twenty-eight canals in the mid- Caballo Muerto and Caiia Huaca (Figure 2). Far- dle valley are even smaller than the Poroto canal. This rington (1974, 1985 :Figure 3a) and Moseley and underscores the attraction of the middle valley to Deeds (1982:35-36) proposed that two or three early farmers; small groups could have constructed canals were constructed on the north side of the lower canals in one season with very modest labor. valley in this phase. Two would have closely followed
Once completed, middle valley canals also would the initial segments of the modern Mochica and have supported relatively small user groups. The area Vichansao canals (Figure 2). They would have been imgated by modem canals ranges from 7 to 286 ha, approximately 7 krn long, and each would have irri- for an average of 68 ha. Analysis of crop productiv- gated perhaps 450 ha. In addition, Moseley and ity in the Guaiiape phase indicates that a hectare of Deeds (1982:35-36) suggest that portions of the imgated land on the coast probably supported two Moro canal may have been constructed up to Caballo people per year with single cropping (Billman 1989; Muerto. This canal would have been about 7 km long
Table 2. Estimates of the Frequency of Prehistoric Shortfalls of Irrigation Water.
|Number of||Number of|
|Hectares||Shortfalls of||Shortfalls every|
|Time||of Land under||Water every||40 years if 30%|
|Period||Cultivation||40 yearsa||of land fallowa|
|Middle Moche Phase||12,550-1 3,200||12-14||6|
|Late Intermediate Period||19,665||25||15|
"Based on the modem 40-year record of annual river flow volume (ONERN 1973:183) and the water requirements of maize
cultivation (ONERN 1973:229).
and irrigated an additional 400 ha. Although these three lower valley canals would have been similar in length to the middle valley Quirihuac canal, they each would have irrigated nearly twice as much land and, therefore, would have carried twice as much water. Consequently, labor requirements can be esti- mated by doubling the volume of excavation for the Quirihuac canal. Each of these three canals could have been constructed by a workforce of approxi- mately 54 people in one season. A social group of 270 could have easily mustered such a workforce. Each canal could have supported 80CL900 people (or 56M30 people with fallowing).
Water shortages were probably relatively rare in this period (Table 2). Most if not all of the middle valley was under cultivation, and approximately 1,300 ha were cultivated on north side of the lower valley. In addition, some areas adjacent to the south side of the river also probably were irrigated; these areas totaled perhaps less than 500 ha (Farrington 1985:Figure 3a). Thus the estimated maximum extent of irrigation in the Guafiape phase is 4,062 ha, which is rounded to 4,100 ha. Based on the 40-year history of river flow, Guaiiape fanners rarely faced water shortages, and tension between lower and mid- dle valley user groups would have been minimal.
In sum, Stage 1 canals could have been construc- ted, used, and managed by small autonomous com- munities. The largest canals could have been constructed by less than 60 people in one season and had user-groups of no more than 600 or 900 people. Consequently, the hydraulic hypothesis predicts that centralized political organizations should not have during the first stage of the development of irriga- tion in the Guaiiape phase (Table 3, Hypothesis 1).
Stage 2, Salinar Phase (400-1 B.C.)
In the Salinar phase, the extent of irrigation expanded significantly in the valley. By close of the phase, vir- tually all usable land in the middle valley was under cultivation (Billman 1996:202-205). In the lower valley, Salinar phase sites are found outside the lim- its of modem cultivation along the south side of the valley (Figure 2) (Billman 1996: 195-196; Brennan 1978). One prehistoric canal also is located beyond modem cultivation; based on the dates of associated sites, this canal appears to date to the Salinar phase (Brennan 1978; Moseley and Deeds 1982). It passes by Cerro Arena, the largest Salinar phase settlement in the valley. On the north side of the valley, the only Salinar phase sites present are a large fishing village at Huanchaco and a large hilltop settlement on Cerro La Virgen (Figure 2). The latter site is located adja- cent to Caballo Muerto, indicating that the Guaiiape phase canals were probably still in use. If other Sali- nar phase sites existed on the north side of the val- ley, they would have been destroyed by the subsequent expansion of the irrigation system.
The south side of the lower valley required more labor to irrigate than the middle valley or areas adja- cent to the river on the north side of the lower val- ley. Currently, the south side is irrigated by four canals measuring 1.6,8.0, 10.0, and 13.0 krn in length (ONERN 1973:214). The Santa Lucia de Moche canal is the largest modern canal on the south side and would have required approximately 327 people to construct in one season. The necessary social group would have been approximately 1,635 people. The maximum elevation prehistoric canal on the south side would have required even more labor to construct. Of course, if these canals were constructed gradually over the course of several seasons, the size of the workforce would have been greatly reduced. However, longer-term construction projects require longer-term centralized planning and coordination.
The maximum population supported by each canal also increased sharply in the ~Znar phase. In addi- tion to larger canals, an increase in agricultural pro-
Table 3. Hypotheses, Test Implications, and Archaeological Expectations for Testing the Hydraulic Hypothesis.
Hypothesis Hypothesis 1: Centralized political organizations should not have formed in the Guafiape phase during the first stage of the development of irriga- tion. Canals could have been con- structed, used, and managed by small autonomous communities or villages (community-level politi- cal integration).
Simple centralized political orga- nizations should have first formed in the Salinar phase during the second stage of the development of irrigation. Canal construction, use, and management involved multiple villages (multiple com- munity political integration).
Hypothesis 3: Complex, centralized political organizations should not have emerged until the Middle Moche phase during the third stage of the development of irrigation. Canal construction, use, and manage- ment required the integration of all or most of the valley (valley- wide political integration).
Test Implications Guafiape phase polities in the valley did not (1) integrate numerous villages, (2) contain several levels of administrative hierarchy, (3) extract quantities of tribute from surrounding vil- lages, and (4) legitimize politi- cal authority by the construction of public monuments.
Salinar phase polities in the val- ley should have (1) integrated several villages, (2) contained two levels of administrative hierarchy, (3) extracted quanti- ties of tribute from surrounding villages, and (4) legitimized political authority by the con- struction of medium-sized pub- lic monuments.
A polity should have developed in the Middle Moche phase that
(1) integrated most of the settle- ments in the valley, (2) con- tained more than two levels of administrative hierarchy, (3) extracted large quantities of tribute from villages throughout the valley, and (4) legitimized political authority by the con- struction of large public monu- ments.
Archaeological Expectation Archaeological Expectation 1.1, Settlement Hierarchies: Site-size hierarchies should not exist in the valley during the Guafiape phase.
Archaeological Expectation 1.2, Settlement Location: Sites located at central places should not be signifi- cantly larger than surrounding sites with similar catch- ment productivity. Small sites should not be clustered around large sites.
Archaeological Expectation 1.3, Catchment Productivity: The population of large sites should not exceed the productive resources within the site catch- ment area.
Archaeological Expectation 1.4, Public Architecture: Public architecture should be absent or limited in scope in during the Guafiape phase. If present, the architec- ture should not have required significant quantities of labor input to construct.
Archaeological Expectation 2.1, Settlement Hierarchies: Two levels of site-size hierarchy should exist in the valley during the Salinar phase.
Archaeological Expectation 2.2, Settlement Location: Sites located at central places should be significantly larger than surrounding sites with similar catchment productivity. Small sites should be clustered around large sites.
Archaeological Expectation 2.3, Catchment Productivity: The population of large sites should exceed the productive resources within the site catch- ment area.
Archaeological Expectation 2.4, Public Architecture: Medium-sized public architecture should be present in the valley during the Salinar phase.
Archaeological Expectation 3.1, Settlement Hierarchies: A site-size hierarchy with three or more levels should not exist in the Moche Valley until the Middle Moche phase.
Archaeological Expectation 3.2, Settlement Location: During the Middle Moche phase, large sites should have formed at central places in the valley. Smaller sites should be clustered around the large sites.
Archaeological Expectation 3.3, Catchment Productivity: The population of large sites should greatly exceed the productive resources within the site catchment area.
Table 3. Hypotheses, Test Implications, and Archaeological Expectations for Testing the Hydraulic Hypothesis (continued).
Hypothesis 4: Complex, regional political orga- nizations should not have emerged until the Late Intermediate period during the fourth stage of the development of irrigation. Canal construction, use, and manage- ment involved more than one river valley (political integration of multiple valleys).
A polity should have developed in the Late Intermediate period that (1) integrated several val- leys, (2) contained more than three levels of administrative hierarchy, (3) extracted large quantities of tribute from vil- lages throughout the valley and adjacent valleys, and (4) legit- imized political authority by the construction of massive public
ductivity is indicated by increased corncob length. Wilson estimates Salinar phase productivity at 2.6 people per ha (Wilson 1988:326). Therefore, the largest modem canal on the south side could have sup- ported a maximum population between 1,942 and 2,774, and the average per canal on the south side was between 9 18 and 1,39 1. Estimates of the area irrigated by the prehistoric maximum elevation canal on the south side of the valley range from 650 (Farrington 1985:638) to 1,200 ha (Moseley andDeeds 1982:Fig- ure 2.1), which would mean a canal user group of between 1,690 to 3,120 people or 1,183 to 2,184 peo- ple with fallowing. These figures indicate that the largest user group in the Salinar phase was 2 to 3.5 times the size of the largest Guaiiape phase user group.
An estimate of land potentially under cultivation in the Salinar phase can be obtained by totaling the middle valley area-2,262 ha (ONERN 1973: 154-155)-the maximum extent of prehistoric cul- tivation on the south side of the lower valley- between 3,200 and 3,750 ha (Moseley and Deeds 1982: Table 2.1)-and an estimate of the area of cul- tivation on the north side of the lower valley-1,300
Archaeological Expectation Archaeological Expectation 3.4, Public Architecture: Large-scale public architecture should not be present in the Moche Valley until the Middle Moche phase. Middle Moche phase public architecture should have required large quantities of labor drawn from the entire valley.
Archaeological Expectation 4.1, Settlement Hierarchies: A region-wide system of administrative centers with multiple levels of control should not have developed on the north coast until the Late Intermediate period.
Archaeological Expectation 4.2, Settlement Location: During the Late Intermediate period, large sites should have formed at central places on the north coast.
Archaeological Expectation 4.3, Catchment Productivity: The population of the paramount center should exceed the productive resources within the Moche Valley.
Archaeological Expectation 4.4, Public Architecture: Large-scale public architecture requiring labor from beyond the Moche Valley should not be present until the Late Intermediate period. Late Intermediate period public architecture should have required massive quan- tities of labor to construct.
ha (Moseley and Deeds 1982). The resulting total of 6,762 to 7,312 ha was rounded to 6,750 and 7,300 ha. The amount of water required to irrigate that much land indicates that, when the south side of the lower valley was reclaimed in the Salinar phase, water short- ages began to occur with some degree of frequency (Table 2). Several shortfalls in irrigation water would have occurred during a person's lifetime. These water shortages would have led to serious disputes between lower and middle valley canal user groups.
In short, in Stage 2, the size of construction work- forces and user groups grew significantly. Canal pro- jects would have required labor from multiple villages; single canals would have sustained multi- ple communities. Water shortages probably began to occur several times a generation, creating tensions between middle and lower valley user groups. Con- sequently, the hydraulic hypothesis predicts that irri- gation would have required two levels of political control. Simple centralized political organizations should have first developed in the Salinar phase dur- ing the second stage of the development of irrigation (Table 3, Hypothesis 2).
Stage 3, Moche Phase (A.D. 200-900)
No significant expansion of the irrigation system has been identified for the Gallinazo and Early Moche phases. The area of irrigation in the lower valley actually may have declined as population shifted to the middle valley (Billman 1996:247-250). Cerro Arena in the lower valley was completely abandoned at the close of the Salinar phase, and the population of Cerro LaVirgen declined dramatically in the Gal- linazo phase (Figure 2).
During the Middle Moche phase, the first truly large-scale canals were constructed when irrigation expanded into the far reaches of the north side of the valley, which is the most difficult area to reclaim (Fig- ure 2). Large areas of arable land were opened up on the north side on the Pampa Esperanza through the expansion of the Moro, Vichansao, and Mochica canals (Figure 2) (Moseley and Deeds 1982:3742). Because those canals carried much more water than the segments in use in the Guaiiape and Salinar phases, it would have been necessary to rebuild them. Dating their construction is particularly important, because the Southern Moche state formed in this phase. Various researchers date this burst of canal construction to the Middle Moche phase (Farring- ton 1974:85, Figure 8.3; Moseley and Deeds 1982:3942; Pozorski 1987: 113). Numerous Moche phase sites have been recorded on Pampa Esperanza along all three canals. Ceramics from those sites include only Moche I11 and IV diagnostics; no Early Moche, Gallinazo, or Salinar phase sites have been identified.
The Mochica canal has the lowest elevation of these three canals and probably was constructed first (Figure 2). In the Middle Moche phase, it was extended all the way to the Rio Seco, with a termi- nus somewhere near Huanchaco, opening up a large portion of Pampa Esperanza. Twice in the Middle Moche phase, a segment of the canal on the pampa was moved northward to progressively higher ele- vations in order to increase the area of cultivation (Moseley and Deeds 1982:42, Figure 2.2 and 2.3). In the Moche phase, the Mochica canal was unlined and approximately 3 1 km long, a truly monumental construction (Ortloff et al. 1985:82). Pozorski (1987:Table 1) cites one uncalibrated radiocarbon date of A.D. 550 + 80 for the extension of the Mochica canal onto Pampa Esperanza, which sug- gests construction sometime between A.D. 436-779 (calibrated with two sigmas4). Given the presence of Moche I11 diagnostics on the pampa, this places con- struction after A.D. 400 in the early part of the Mid- dle Moche phase.
Later in the Middle Moche phase, the Moro canal was expanded. Although the modem Moro canal does not extend all the way to Pampa Esperanza, Ort- loff, Feldman, and Moseley proposed that in the Mid- dle Moche Phase it extended 28 km all the way to Pampa Esperanza and would have irrigated a large stretch of land above the Mochica canal (Figure 2) (Ortloff et al. 1985:79). Further, they postulated that the canal was constructed in Moche IV, sometime before a major El Niiio between A.D. 450 and 600 (Ortloff et al. 1985:79,91).
Portions of the Moro canal on Pampa Esperanza were eventually replaced by the Vichansao canal, which was constructed at an even higher elevation, opening up most of the remaining arable land on the pampa (Figure 2). Moseley and Deeds note that the Vichansao canal cut through a Moche I11 cemetery, but did not cut through any of the numerous Moche IV sites along the canal, placing the construction of the canal in the later part of the Middle Moche phase (Moseley and Deeds 1982:37). The Moche phase canal would have been at least 30 km long (Ortloff et al. 1985:81).
Whereas the largest Salinar phase canals required 300-400 person-seasons to construct, the Middle Moche phase Moro, Mochica, and Vichansao canals required approximately 500, 600, and 900 person- seasons to complete, respectively (Ortloff et al. 1985:Table 4). In other words, the Vichansao canal required more than twice the labor of any other canal previously constructed in the valley. Because of its size, the Vichansao probably was constructed over several years by a workforce of hundreds of people. Leaders would have had to mobilize or control large groups for several years. For instance, if the canal was constructed in one year, the social group from which labor was drawn would have numbered approximately 4,500; if constructed over five years, the social group would have been 900 people.
Canal user groups also would have been massive, even though corn yields appear to have remained approximately the same as in the Salinar phase (Wil- son 1988:326). The Moro canal could have supported 2,243 people (1,570 with fallowing) (Moseley and Deeds 1982:Table 2.1). The modem extent of the Mochica canal is approximately equal to its extent in the Moche phase. Currently it irrigates 4,850 ha (ONERN 1973:214) and could support 12,610 peo- ple (8,827 with fallowing).
With the dramatic increase in cultivated land in the Middle Moche phase, water shortages would have occurred regularly. Altogether between 12,538 and 13,188 ha were under irrigation in the Moche Valley by the end of the Middle Moche phase.5 Based on the modern record of river volume, irrigation water shortages would have occurred 12 to 14 times in 40 years (Table 2). If 30 percent of the land was fallow, shortages would have occurred six times every 40 years. Disputes between lower and middle valley canal user groups would have been a chronic problem.
Because of dramatic increases in the managerial requirements of irrigation, the hydraulic hypothesis predicts that the first valley-wide polity should have emerged in the Middle Moche phase (Table 3, Hypothesis 3). Construction of the Mochica, Moro, and Vichansao canals would have required hundreds of laborers; once in use, each canal would have sup- ported thousands of people; and water shortages would have occurred every few years. Irrigation of the north side of the lower valley would not have been possible without some sort of centralized authority that could allocate water and mediate disputes.
Stage 4, Late Intermediate Period (A.D. 100&14 70)
During the Late Intermediate period (LIP), irrigation expanded dramatically on the far north side of the val- ley (Figure 2). Four major canal projects of consider- able magnitude were completed by A.D. 1300, before a major El Nifio event (Moseley and Deeds 1982:4243; Pozorski 1987: 1 12-1 13). First, a new version of the Moro canal was constructed adjacent to the old canal. Second, the distal end of the Vichansao canal was elevated and stone lined. Third, the Vichansao canal was extended across the Pampa Rio Seco and Pampa Huanchaco, opening those areas for cultivation. And finally, La Cumbre Intervalley canal was constructed from the Chicama to the Moche Val- ley (Ortloff et al. 1985; Pozorski 1987; Pozorski and Pozorski 1982). All four projects were constructed over 200 to 300 years and required an extraordinary amount of labor (Ortloff 1985 et al.:Table 4). In addition, several large canals were constructed above the limits of modern cultivation in the middle Moche Val- ley; however, data on those canals are not available.
By far the largest of these projects was La Cum- bre Intervalley canal, which was intended to bring water from the ChicamaValley to newly created fields on Pampas Rio Seco and Huanchaco. The canal was over 70 km long, more than twice the length of any canal ever constructed in the Moche Valley. Labor estimates for its construction vary greatly from more than 700,000,000 person-hours (or 560,900 person- seasons) (Ortloff et al. 1985:Table4) to4,143,647 per- son-days (or 26,561 person-seasons) (Pozorski and Pozorski 1982:866). Based on their analysis of an abortive canal segment located adjacent to the canal, Pozorski and Pozorski (1982:866) further proposed that the typical workforce consisted of approximately 1,000 people. A workforce of that size could have completed the canal in 26 years, or a generation. In other words, the construction of La Cumbre Inter- valley canal required nearly 30 times the amount of labor of the largest Middle Moche phase canal in the Moche Valley.
The most dramatic rise in population per canal was in the LIP when corn yields increased to 6.8 peo- ple per ha (Wilson 1988:326). The largest canal potentially could have supported between 21,807 and 33,041 people, depending on fallowing. Even the smallest canal on the north side could have supported between 1,105 and 1,579 people.
In the LIP, the area of cultivation exceeded the modern extent of irrigation on the north side of the valley (Farrington 1974; Moseley and Deeds 1982; Ortloff et al. 1985; Pozorski 1987). On the south side of the valley, 1,223 ha of land beyond the limits of modern irrigation were abandoned, possibly due to dune formation (Moseley and Deeds 1982). Conse- quently, the potential maximum extent of irrigation in the LIP was equal to the maximum possible extent of cultivation in the valley (20,888 ha) minus the area of cultivation that was abandoned on the south side (1223 ha) for approximately 19,665 ha. Extension of the Vichansao canal to the Pampa Rio Seco and Pampa Huanchaco would have added even more irrigated land.
As the far reaches of the valley were reclaimed in the LIP, water shortages increased dramatically. Agricultural demand probably frequently exceeded the supply of water. Based on the modem record of river volume, 19,665 ha of land could have been irri- gated only 15 out of 40 years or 25 out of 40 years if land was fallowed (Table 2). Irrigation of this quan- tity of land would have required regular valley-wide management of water in order to insure that lower- valley farmers received sufficient water. Further, the construction of La Cumbre Intervalley canal indi- cates that leaders controlled water allocation in both the Moche and Chicama Valleys.
In sum, another dramatic increase in managerial requirements occurred in the LIP. The size of con- struction work groups grew to 1,000 or more labor- ers, and construction would have lasted for decades on La Cumbre Intervalley canal. Some canal user groups probably exceeded 20,000, and water short- ages occurred every two or three years. For the first time, canal construction, use, and management involved more than one river valley. Consequently, the hydraulic hypothesis predicts that the first mul- tivalley polities should not have emerged until the LIP (Table 3, Hypothesis 4).
Deriving Hypotheses, Test Implications, and Archaeological Expectations
Testing each of four proposed hypotheses for evalu- ating the hydraulic model involves the recognition and measurement of political centralization (Table 3). Attributes of centralized polities include hierarchical control and administration, tribute collection, and legitimation of authority. These attributes can be iden- tified archaeologically in four ways. First, the analy- sis of settlement hierarchies can indicate the number of levels of administrative control in a polity and the number of communities controlled by a polity (Flan- nery 1998: 1621; Isbell and Schreiber 1978; John- son 1973; Sanders et al. 1979; Wilson 1988; Wright and Johnson 1975). Second, tribute collection can be measured by comparing the size of centers to their catchment productivity (Massey 1986; Steponaitis 1981). Third, because the use of centrally located facilities reduces administrative and tribute trans- portation costs, political centralization also can be detected by monitoring the growth at central sites (Earle 1976; Johnson 1973; Massey 1986; Peebles 1978; Steponaitis 1978; Upham 1982). Finally, con- struction of public architecture is another attribute of political centralization because political leaders often use public monuments to legitimize political control and disseminate their ideology (Billman 1999; DeMarrais et al. 1996). Analysis of the design, extent, and distribution of public architecture can indicate the extent of political control and the scope of power of leaders (Billman 1999; Haas 1982, 1985, 1987; Moore 1996a, 1996b; Pozorski 1976, 1980, 1982).
These four measures of centralization were used to identify the test implications of the four hypothe- ses (Table 3). Four archaeological expectations then were derived for each of the test implications (Table 3). The archaeological expectations reveal that four types of settlement pattern data are needed to recon- struct the development of centralized polities the MocheValley and to test the four hypotheses derived from management-oriented theories. Data areneeded to assess the development of site hierarchies, the location and size of site clusters, the catchment pro- ductivity of regional and local centers, and the size and layout of public architecture.
Settlement pattern data from two pedestrian sur- veys of the Moche Valley (Figure 1) (Billman 1996) cover the coastal section and provide information on over 910 archaeological sites from the Late Prece- ramic period to Spanish conquest. The first survey, directed by Michael Moseley during the Chan Chan- Moche Valley Project, covered the entire lower val- ley, and documented over 420 sites. In 1990 and 1991, I surveyed the middle valley (Figure 1) and recorded an additional 483 sites (Billman 1996). Information on early political formation is also pro- vided by excavations at the Late Preceramic site of Alto Salaverry (Pozorski and Pozorski 1979a), the Guafiape ceremonial center of Caballo Muerto (Pozorski 1976,1980,1982), Cerro Arena (Brennan 1978,1980a, 1980b), and the site of Moche (Topic 1977,1982; Uceda et al. 1997, 1998).
The Formation of Centralized Polities in the Moche Valley
A sequence of political development for the Moche Valley spanning the formation of the first autonomous village in the Late Preceramic period to the zenith of the Southern Moche state has been reconstructed (Table 1) (Billman 1996,1997,1999). In the latter part of the Late Preceramic period, the first large sedentary village formed at Alto Salaverry (Figure 3) (Billman 1996:99-103; Pozorski and Pozorski 1979a). The subsistence base of the inhab- itants was fanning and fishing. The population of Alto Salaverry numbered in the hundreds and may have had some form of social ranking as evidenced by the presence of two possible elite residences with large storage facilities (Billman 1996: 104-1 13; Pozorski and Pozorski 1979a). A small circular sunken court may have served as some type of com- munity center for village residents; however, there is no evidence that Alto Salaveny was integrated into a multivillage polity.
In the Early Guaiiape phase the pattern of settlement changed radically (Billman 1996: 13 1-1 82; Pozorski and Pozorski 1979b). Settlements shifted inland to the middle valley, imgation agriculture was intro- duced, and over 33,000 m3 of ceremonial architec- ture were constructed (Figure 4). Nearly all the ceremonial architecture was concentrated at Huaca Menocucho in the middle valley with the remainder distributed among five small centers.
Investment in ceremonial architecture grew sig- nificantly in the Middle Guaiiape phase with the con- struction of between 283,000 and 3 18,000 m3. It was distributed between one primary center at Caballo Muerto, two secondary centers, and at least three small centers (Figure 5) (Billman 1996:141-182; Pozorski 1976,1980,1982). At Caballo Muerto over 231,600 m3 of ceremonial architecture were con- structed, seven times the total for the entire Early Guaiiape phase. Most of this construction occurred in the early part of the Middle Guaiiape phase (Bill- man 1996:171-172; Pozorski 1983). By the Late Guaiiape phase construction declined to between 65,000 and 100,000 m3, and no single center domi- nated the valley (Figure 6) (Billman 1996: 155-182; Pozorski 1983).
Because of the destruction of many habitation sites by the expansion of agriculture in the prehis- toric era, demographic information on the Guaiiape phase is limited. Fortunately, the size and design of ceremonial construction permit some inferences about the nature of political organization. The huge volume of construction indicates that leaders were able to mobilize large labor groups on a consistent basis for the construction of large public monuments. By the Middle Guaiiape phase, a three-tiered hier- archy of monuments developed. This hierarchy may be amanifestation of a hierarchy of leadership: para- mount leaders with access to a valley-wide labor pool, a second level of leadership with a local labor pool, and a third level of village leaders (Billman 1996:205-219). The design of ceremonial architec- ture indicates that these monuments were used for the presentation of public displays to groups num- bering in the thousands. Not all rituals, however, were open to the public. At Caballo Muerto, Huaca Los Reyes had a three-tiered hierarchy of space with
Figure 4. Location of archaeological sites dating to the Early Guaiiape phase, Moche Valley, Peru (1800-1300 B.C.).
Figure 5. Location of archaeological sites dating to the Middle Guaiiape phase, Moche Valley, Peru (1300-800 B.C.).
Figure 6.Location of archaeological sites dating to the Late Guaiiape phase, Moche Valley, Peru (800-400 B.C.).
a large public plaza, an intermediate-sized plaza on the second level of the mound, and small ceremo- nial spaces on top of the mound. This suggests a hier- archy of public displays and restricted access to certain ritual activities (Pozorski 1976,1980,1982).
Management-oriented models predict that cen- tralized political organizations should not have devel- oped during the first stage of irrigation in the Guaiiape phase (Hypothesis 1). However, analysis of settlement hierarchies and public architecture indi- cates that centralized political organizations did develop during this phase. A multitiered settlement hierarchy developed that encompassed all or most of the valley, and the construction of public archi- tecture increased exponentially.
By implication, the managerial requirements of irrigation in the Guaiiape phase cannot explain the formation of the first valley-wide polity. Guaiiape phase canals did not require large labor forces to construct; they integrated relatively small groups of people, and water shortages were probably still uncommon. In short, imgation could have been con- ducted without centralized political organizations.
In contrast, the managerial requirements of mon- ument construction far exceeded the requirements of canal construction. A rough estimate of the volume Guaiiape phase middle valley canals was derived based on the 85 km of 33 modem canals (ONERN 1973:214), multiplied by the average cross-sectional areas (1.4 x .4 m) of the three canals with published data (ONERN 1973:214), for a total of 47,600 m2. The three lower valley Guaiiape canals probably had a total volume of less that 25,200 m3 (based on mul- tiplying the estimated volume of the largest canal by three) for an overall canal volume of 72,800 m3 in that phase.
The volume of Guaiiape phase monumental con- struction is 417,000 m3, nearly six times the volume for all of the Guaiiape canals. The managerial requirements of constructing the Guaiiape monu- ments was roughly equivalent to constructing all the prehistoric canals on the north side of the valley (504,370 m3), a task undertaken by the Chimu state some 1,400 years later. Construction of all of the Middle Guaiiape phase monuments at the paramount center of Caballo Muerto alone involved moving 23 1,000 m3 of dirt, a greater volume than excavat- ing Vichansao (210,370 m3), the largest prehistoric
Figure 7. Location of archaeological sites dating to the Early Salinar phase, Moche Valley, Peru (400-200 B.C.).
canal, excluding the massive La Cumbre Intervalley canal. Clearly, if we wish to understand the forma- tion of the first centralized polities in the valley, we must look beyond the managerial requirements of canal construction.
By the start of the Salinar phase, the early coastal tradition of monumental construction and associated political organizations appear to have come to an end. All of the Late Guaiiape phase ceremonial centers were abandoned, the population aggregated into clus- ters, and most habitation sites shifted to defensive settings (Figure 7) (Billman 1996: 187-235). A new cycle of political formation had begun.
By the Late Salinar phase, nine distinct popula- tion clusters formed in the valley. The clusters were separated by unoccupied buffer zones (Figure 8) (Billman 1996:205-215). For instance, along the south side of the lower valley, the Cerro Arena and Cerro Orejaclusters were separated by 9 km of unoc- cupied valley margin. This occupational gap is prob- ably not due to poor site preservation since the maximum limit of cultivation had been reached and Salinar sites were not destroyed by subsequent expansion of irrigation. In the middle valley, site clusters were separated by 1.8 to 3.9 km of unoccu- pied valley margin. The unoccupied zones contrast with densely packed habitation sites within the clus- ters. For example, the Puente Serrano cluster extended for less than 1 km along the valley margin and contained 20 ha of habitation area (Figure 8). Down valley from the edge of that cluster to the next cluster there was 3 km of unoccupied valley margin; up valley there was 2 km of unoccupied margin.
The population clusters have an internal two- tiered settlement hierarchy and appear to have been politically autonomous in Early Salinar phase. In the Late Salinar phase, the clusters may have formed a loose confederation dominated by the Cerro Arena cluster (Billman 1996:2 13-2 15). With a population numbering in the thousands, the Cerro Arena clus- ter could have mustered more warriors than all the others combined.
Investment in public architecture dropped dra- matically in the Salinar phase, declining by 70-90 percent from between 65,000 and 100,000 m3 in the Late Guaiiape phase to just over 15,000 m3. New cer- emonial structures accommodated small groups, had restricted access, and were often associated with elite
Figure 8. Location of archaeological sites dating to the
dwellings (Brennan 1978, 1980a, 1980b). In addi- tion, household data indicate that a small elite stra- tum existed at Cerro Arena (Brennan 1978, 1980a, 1980b).
Because the size of construction workforces, size of canal user groups, and frequency of water short- ages increased significantly, the hydraulic model pre- dicts that the first simple chiefdoms should have developed in this phase (Hypothesis 2). Political developments in the Salinar phase conform to some of the archaeological expectations derived from Hypothesis 2. In the Early Salinar phase, several small autonomous population clusters with internal two-tier settlement hierarchies formed in the valley. By the Late Salinar phase, nine clusters may have been joined together in a loose confederation dom- inated by Cerro Arena, which appears to have extracted tribute from the other clusters (Billman 1996:212). If Cerro Arena controlled other clusters, then the settlement system would have had three tiers, one more tier than predicted by the hydraulic model. Construction of public architecture dropped dramatically in the Salinar phase, although the hydraulic hypothesis predicted an increase.
Late Salinar phase, Moche Valley, Peru (200-1 B.C.).
Gallinazo and Moche Phases
Although political evolution in the Salinar phase
loosely followed predictions derived from Hypoth-
esis 2, political centralization in the Gallinazo and
Early Moche phases increased beyond those pre-
dictions without any significant expansion in irriga-
tion. At the start of the Gallinazo phase, the middle
valley population apparently was unified into a sin-
gle polity centered at Cerro Oreja (Figure 9). A sep-
arate polity may have been located at Huanchaco at
Pampa Cruz. The Cerro Oreja polity had a three-tier
site hierarchy with a paramount center and eight
moderately sized secondary centers (Billman
1996:25&253). Approximately 67,000 m3 of cere-
monial architecture were constructed in the Gallinazo
and Early Moche phases compared to 15,000 m3 in
the Salinar phase (Billman 1996:Tables 17.1 1 and
18.4), an increase of over 400 percent. The Cerro
Oreja polity was much more complex and central-
ized than expected if only the managerial require-
ments of irrigation are considered.
Perhaps the most surprising result of the survey
of the middle valley was the identification of 1 14 sites
Figure 9. Location of archaeological sites dating to the Gallinazo and Early Moche phases, Moche Valley, Peru (A.D. 1-400).
that have large quantities of nonlocal Highland Early Intermediate period ceramics (HEIP) (Billman 1996:264-289, 1997). Over 90 percent of the ceramic artifacts at HEIP sites are made in highland styles rather than in coastal styles. In comparison to coastal ceramics from the same period, the ceram- ics from the HEIP sites are distinct in paste, vessel form, and decoration (Billman 1996:264-268). These ceramic assemblages are similar to EIP ceram- ics from the highland areas of the Moche, Vini, and Chicama drainages (Topic and Topic 1982). Based on excavation and survey data from the highlands (Topic and Topic 1982) and Vini Valley (Bennett 1950; Collier 1955; Ford and Willey 1949; Strong and Evans 1952), these ceramics in the Moche Val- ley date to the Gallinazo through Early Moche phases (Billman 1996:264-268).
HEIP sites are clustered (Figure 10) in the Sins- icap Valley (Sinsicap cluster), in the upper middle Moche Valley above the confluence of the Sinsicap River (Cruz Blanca cluster), and on the south side of the lower middle valley (Cerro Ledn cluster) (Bill- man 1996:250-253, 1997), but not in the lower val- ley. These clusters are separated by uninhabited buffer zones of 1.1 to 3.4 km. These gaps apparently were not a function of the distribution of arable land because it is evenly distributed along the river.
The population clusters have a two- or three-tier hierarchy of habitation sites (Billman 1996:250- 253). Each is dominated by a paramount site, which has large elaborate residences, suggesting the pres- ence of an elite social stratum. The settlement hier- archy, empty spaces between clusters, and the existence of an elite stratum strongly suggest that each HEIP cluster was an autonomous centralized polity.
The large number of HEIP sites in the middle Moche Valley implies a high degree of highland- coastal interaction just prior to the formation of the Southern Moche state. A similar pattern of intrusion of highland ceramic sites in the EIP has been docu- mented in the Vini, Nepefia, Casma, and Lurin Val- leys (Daggett 1984, 1987; Patterson et al. 1982; Pozorski and Pozorski 1987; Topic 1982:259-260; Topic and Topic 1982).
The form of highland-coastal interaction remains open to question. HEIP sites in the middle valley may represent (1) specialized production and exchange
Figure 10. Location of archaeological sites with highland ceramic assemblages dating to the Gallinazo and Early Moche phases, Moche Valley, Peru (A.D. 1400). At these sites, more than 90 percent of the ceramics are highland in style.
of goods between highland and coastal groups (Ros- tworowski 1977, 1978), (2) direct peaceful colo- nization of the middle valley by highland groups seeking access to lowland goods, i.e., vertical arch- ipelagos (Murra 1968,1972), or (3) violent conquest and colonization of the middle valley by highlander invaders (Billman 1996:275-276, 288-289, 1997). Each of these scenarios has profound implications for the origins of the Southern Moche state. Survey data are consistent with the third model of interac- tion (Billman 1996:264-289); however, excavation is needed to address the ethnic identity of HEIP res- idents, the nature of highland-coastal interaction, and the role that interaction played in the formation of the Southern Moche state.
Ceramic data indicate that the HEIP sites were abandoned before the start of the Middle Moche phase (Figure 11) (Billman 1996:264-268) when the Southern Moche state emerged, unifying all of the middle and lower valley. The capital of the val- ley shifted from Cerro Oreja to the site of Moche. Once control of the valley was achieved, the South- em Moche state expanded south and north, inte- grating adjacent valleys into the Moche political and economic system (Billman 1996:329-333, Conrad 1978; Topic 1982:27&273; Willey 1953:397; Wil- son 1988:222-224,1995).
Leaders of the Southern Moche state directed the construction of large public monuments in the Moche and neighboring valleys at which they conducted elaborate public rituals for large audiences (Billman 1996:3 17-3 19; Uceda 1997). One of the largest pre- historic monuments in the New World-Huaca del Sol-was constructed at Moche (Hastings and Mose- ley 1975; Moseley 1975b). The total volume of Mid- dle Moche phase ceremonial architecture dwarfs construction in the preceding Gallinazo and Early Moche phases. The volume of ceremonial architec- ture constructed in the MocheValley during the Mid- dle Moche phase was approximately 1,291,000 m3, nearly 20 times that of the Gallinazo and Early Moche phases (Billman 1996:3 17-3 18).
Leaders also directed the construction of large reclamation projects in Moche, Chicama, and Santa Valleys (see for example Wilson 1988: 198-223). Craft goods, especially fineware ceramics, were produced
Contour intawal: 200 m DEFENSE SITES
Fort Arm of modern cultivation -Fortification --Wash -CANAL
in amounts never before seen on the coast (Alva and Donnan 1993;Larco 2001; Russell et al. 1994,1998; Shimada 1994:191-215; Uceda and Armas 1997, 1998). Leaders amassed large quantities of these wealth goods for use in their households and burials (Alva and Donnan 1993;Castillo and Donnan 1994; Uceda 1997).Leadersalsomobilized laborersto constructlargewell-made,elaboratelyfinishedresidences (Topic 1977,1982;Tello 1998;Van Gijseghem2001).
Thehydraulicmodel predictsthe firstvalley-wide polity would have emerged in the Middle Moche phase (Hypothesis3). However, the SouthernMoche state is significantly more complex than predicted. Management-oriented models predict that this type of regional polity should not have occurred until the LIP when the far-north side of the valley was reclaimed and La Cumbre Intervalley canal constructed (Hypothesis 4). Although Moche leaders apparently did organize the construction of large canals in the MocheValley, construction appears to have occurred after leaders had consolidated their control over the entire valley and created a centralized polity at the site of Moche.
In the Guaiiape phase, during the first cycle of political centralization in the MocheValley, the managerial requirementsof canal constructiondo not appear to have been a major catalyst of political change. Canals were short and relatively easy to construct; canal user groups were small and self-sufficient and water shortages were probably rare. Canals could have been constructed, used, and managed at the villageor community level.Warfare also apparently was not a factor in Guaiiapephase social or political life. If armed conflict occurred, it was not of sufficient severity to become evident in the pattern of settlement (Billman 1996:181-183, 1997, 1999). Population did not aggregate or shift to defensive locations, and no fortifications were constructed. If we are to understand what factors were responsible forthe formationof Guaiiapephasepolities, we must look elsewhere.
The answermay be found in the opportunitiesfor development of socialstratificationand politicalcontrol created by changes in the subsistence economy at the beginning of the Guaiiape phase. The produc- tive system during the Late Precerarnic period pro- vided few if any opportunities for economic control by leaders. Simple technologies providedeasy access to near-shore resources for any fit individual. Agri- cultural land was still abundant, and floodplain agri- culture required slash-and-burn or simple irrigation techniques. The absence of intensive armed conflict provided few opportunities for leaders to exercise coercive power. In order to maintain leader-follower relationships, leaders probably had to rely on per- sonal charisma, ideology, and redistribution of sur- pluses produced by their own households and the households of their followers.
The rich marine environment had the potential to yield considerable surplus in this period. Several fea- tures at Alto Salaverry may have been used for the storage of large quantities of dried fish (Billman 1996: 104-1 13,1999). In the hands of aspiring lead- ers, surpluses of fish and other marine goods could have been used to enhance their status and influence. Surpluses could have been used to finance political activities-such as dances, feasts, and other cere- monies-that created obligatory relationships between leader-donors and follower-recipients. In this scenario, leaders did not control any part of the means of production, but they were capable of gen- erating surpluses that enhanced their status and helped them to form coalitions of followers.
With the shift to irrigation in the Guaiiape phase, emerging leaders had new opportunities for politi- cal control and the accumulation of goods. Although the initial cost of constructing irrigation canals was higher than the costs of fishing and floodplain agri- culture, once in production, irrigation systems dra- matically increased yields. Consequently, by organizing and financing the construction of irriga- tion systems, leaders potentially could have then extracted surpluses from canal users to fund further political activity, such as monument construction, and to further elevate the socioeconomic status of their own households. Irrigation systems also created opportunities for leaders to control land and the flow of water. The political consequence was that in the Early Guaiiape phase political leaders not only could have controlled increased agricultural production, but also the distribution of land and water. With these newfound sources of economic power, leaders were able to finance the creation of centralized, hierar- chical political organizations.
In short, although the managerial requirements of irrigation were minimal in the Guaiiape phase, oppor- tunities for political control and finance were great. Therefore, political models may be more effective in explaining the centralized polities in the Guaiiape phase than the managerial models proposed by Wittfogel, Steward, and others. Construction of three moderate-sized lower-valley canals in Middle and Late Guaiiape may have been particularly important in increasing the economic power of leaders. With the investment of relatively small quantities of labor+ach of those canals could have been con- structed by less than 60 people working for one sea- son-leaders could have reclaimed perhaps 1,300 ha of land, which could have supported 1,800-2,600 people. Leaders may have been in a position to grant use-rights to commoners in exchange for annual pay- ments in labor or a percentage of the harvest. This sort of patron-client relationship apparently existed between lords and commoner farmers on the north coast before Spanish conquest and was the founda- tion of the political economy of the Chimu state (Netherly 1984). Such relationships are also a rela- tively common form of political control and finance in a wide range of historic chiefdoms and states (see Earle 1997:67-89; Sidky 1996).
Labor service payments perhaps were used to construct the mounds at Caballo Muerto. As noted previously, during the Guaiiape phase nearly six times more earth was excavated for mound con- struction than in the construction of all of the Guaiiape phase canals. By investing labor service in the ideological infrastructure, leaders would have been able to greatly expand their ideological power base. Control of large monuments would have enabled leaders to materialize and disseminate an ideological message that both legitimized their ele- vated political status and the annual payment of goods or services by farming households. In essence, their control of ceremonial centers and the public and private rituals conducted at them allowed leaders to control the means of ideological production. Consequently, the creation of centralized polities by lead- ers in the Guaiiape phase primarily may have been based on two mutually reinforcing sources of power: control of irrigation systems and control of ceremo- nial centers.
The pattern of political development in the EIP differed significantly from that of the Guaiiape phase. Warfare and highland-coastal interaction appear to have been important factors influencing the forma- tion of polities in that period (Billman 1996, 1999). Guaiiape phase polities and their associated ideol- ogy apparently were destroyed by armed conflict by the start of the Salinar phase, and a new political order was created. In the Salinar phase, the population of the valley aggregated into several clusters for self- defense. As violent conflict increased in the subse- quent Gallinazo and Early Moche phases, autonomous population clusters coalesced into a val- ley-wide polity centered at Cerro Oreja. Rising lev- els of violent conflict in the EIP may have created opportunities for leaders to organize and direct defense, plan offensive actions, and negotiate alliances. Control of military and diplomatic activi- ties may have constituted a new power base in this period. Escalation of conflict was paralleled by increased highland-coastal interaction, which also may have been violent in nature. Highland-coastal interaction peaked just prior to the formation of the Southern Moche state with the development of three HEIP site clusters in the middle valley.
Once they consolidated their military control of the Moche Valley and expelled or assimilated high- landers in the middle valley, leaders appear to have directed a major expansion of the irrigation system. More than 7,000 ha were reclaimed by the con- struction of three canals on the north side of the val- ley, supporting between 12,700 and 18,200 people. Annual payments of goods or services by tenant farmers on newly reclaimed land would have vastly increased the economic power base of Moche lead- ers, as dramatically illustrated by the construction 1,200,000 m3 of monumental architecture in the Moche Valley during the Middle Moche phase. Annual tribute payments by thousands of households on newly reclaimed land on the north side of the val- ley provides a key to solving the puzzle of how Moche leaders were able to consistently mobilize vast quantities of labor, whereas previous political leaders could not. Leaders had never before created such a large group of tribute-paying farmers.
Agricultural tribute apparently also was used to support craft specialists, who created large quanti- ties of wealth goods (Attarian 1996; Russell et al. 1994, 1998; Uceda and Armas 1997, 1998). Lead- ers may have used the distribution of those wealth goods to disseminate state ideology and maintain political alliances, creating an effective system of wealth finance and political propaganda from a sys- tem of staple finance. After consolidation of control of the Moche Valley, they led the conquest of adja- cent valleys. Following the same strategy employed in the Moche Valley, Moche leaders undertook major reclamation projects in some of those valleys and constructed provincial centers with large monuments and craft workshops.
In sum, evaluation of the hydraulic hypothesis indicates that the managerial requirements of pro- duction (specifically canal construction, integration of households, and disputes resolution) were rela- tively unimportant in the formation of centralized polities in the Moche Valley. Currently available data suggest that the expansion of irrigation in the Guaiiape phase may have allowed leaders to gain control of crucial aspects of the means of produc- tion. Control of land, labor, and water permitted them to create the first centralized valley-wide polity. Cen- tralized institutions of rule were not required to con- struct and manage Guaiiape phase irrigation canals; neither warfare nor the managerial requirements of irrigation were important factors. The managerial requirements of irrigation also were probably rela- tively unimportant in the formation of the Southern Moche state. Warfare, highland-coastal interaction, and political control of irrigation systems created opportunities for leaders to form a highly centralized, territorially expansive state.
Acknowledgments. Funding for fieldwork and analysis was provided by an NSF Dissertation Improvement Grant, Fulbright Fellowship, Sigma Xi Grant-in-Aid of Research, UCSB Humanities-Social Science Grant, Dumbarton Oaks Pre-Columbian Fellowship, and UNC Research Council Publication grant. I thank the Instituto Nacional de Cultura for permission to conduct research in the MocheValley and for the support of Anna Maria Hoyle, Santiago Uceda, Jesus Bricefio, and Ctsar Gilaz at the INC office in Tmjillo. I am deeply indebted to Carlos Ramirez, Jorge Sechun, Alfredo Melly, Lucia Medina, Flor Diez, and Theresa Vasallo for their hard work in field and lab. My thanks also go to Michael Moseley and Carol Mackey, who provided access to the Chan Chan- Moche Valley Project's site files. Ellen O'Brian, Colleen Winchell, Chris Rodning, and Jennifer Ringberg drafted the illustrations. The Spanish abstract was translated by Julio Rucabado. I would like to thank Gary Feinman, Jonathan Haas, Charles Hastings, Joyce Marcus, Mintcy Maxham, Michael Moseley, Chris Rodning, Katharina Schreiber, Tiffiny Tung, Henry Wright, and seven anonymous reviewers for their comments on the paper. I greatly appreciate the patience and guidance of Gary Feinman, Kathy Schreiber, and Suzanne Fish during their editorship of Latin American Antiquity.
Adams, Robert McC. 1966 The Evolution of Urban Society: Early Mesopotamia and Pre-Hispanic Mexico. Aldine, Chicago. 1968 Early Civilization, Subsistence, and Environment. In Man in Adaptation: The Biological Background, edited by
Y. Cohen, pp. 363-377. Aldine, Chicago. Alva, Walter, and Christopher B. Donnan 1993 Royal Tombs of Sipdn. Fowler Museum of Culture His- tory, University of California, Los Angeles. Attarian, Christopher J.
1996 Plant Foods and Ceramic Production: A Case Study of Mochica Ceramic Production Specialists in the Chicama Valley, Peru. Unpublished M.A. thesis, Department of Anthropology, University of California, Los Angeles.
Bawden, Garth 1996 The Moche. Blackwell Publishers, Cambridge, Massa- chusetts. Bennett, Wendell C.
1950 The Gallinazo Group Kru' Valley, Peru. Yale University Publications in Anthropology, No 43. Yale University Press, New Haven, Connecticut.
Billman, Brian R. 1989 Land, Water, andkchitecture: The Economic and Polit- ical Organization of an Early Andean State. Unpublished
M.A. paper, Department of Anthropology, University of Cal- ifornia, Santa Barbara.
1996 The Evolution of Prehistoric Political Organizations in the Moche Valley, Peru. Unpublished Ph.D. dissertation, Department ofAnthropology, University of California, Santa Barbara.
1997 Population Pressure and the Origins of Warfare in the Moche Valley, Peru. In Integrating Archaeological Demog- raphy: Multidisciplinary Approaches to Prehistoric Popu- lations,edited by R. R. Paine, pp. 285-3 10. Occasional Paper No. 24, Center for Archaeological Investigations, Univer- sity of Southern Illinois at Carbondale.
1999 Reconstructing Prehistoric Political Economies and Cycles of Political Power in the Moche Valley, Peru. In Settlement Pattern Studies in the Americas: Fifty Years Since Viri, edited by B. R. Billman and G. M. Feinman, pp. 131-159. Smithsonian Institution Press, Washington, DC.
Binford, Lewis 1983 In Pursuit of the Past. Thames and Hudson, New York. Blanton, Richard E., Gary M. Feinman, Stephen A. Kowalewsky, and Peter N. Peregrine 1996 A Dual-Process Theory of the Evolution of Mesoamer- ican Civilization. Current Anthropology 26: 1-14. Bourget, Steve 1997 Las excavaciones en la Plaza 3A de la Huaca de Luna. In lnvestigaciones en la Huaca de la Luna 1995, edited by
S. Uceda, E. Mujica, and R. Morales, pp. 51-60. Facultad de Ciencias Sociales, Universidad Nacional de la Libertad- Trujillo, Peru.
2001 Children and Ancestors: Ritual Practices at the Moche Site of Huaca de la Luna, North Coast of Peru. In Ritual Sac- rifice in Ancient Peru, edited by E. P. Benson and A. G. Cook, pp. 93-1 19. University of Texas Press, Austin.
Brennan, Curtiss T.
1978 Investigations at Cerro Arena, Peru: Incipient Urban- ism on the North Coast of Peru. Unpublished Ph.D. disser- tation, Department of Anthropology, University of Arizona, Tucson.
1980a Cerro Arena: Early Cultural Complexity and Nucle- ation in North Coastal Peru. Journal of Field Archaeology 7:1-22.
1980b Cerro Arena: Rise of the Andean Elite. Archaeology 33:6-13. Brumfiel, Elizabeth M.
1992 Distinguished Lecture in Archaeology: Breaking and Entering the Ecosystem: Gender, Class, and Faction Steal the Show. American Anthropologist 94:55 1-567.
Brumfiel, Elizabeth M., and Timothy K. Earle 1987 Specialization, Exchange, and Complex Societies: An Introduction. In Specialization, Exchange, and Complex Societies, edited by E. M. Brumfiel andT. K. Earle, pp. 1-9. Cambridge University Press, Cambridge.
Butzer, Karl W. 1976 Early Hydraulic Civilization in Egypt: A Study in Cul- tural Ecology. University of Chicago Press, Chicago. Castillo, Luis Jaime, and Christopher B. Donnan 1994 La ocupacion Moche de San Jose Moro, Jequetepeque. In Moche:propuestas yperspectivas, edited by S. Uceda and
E. Mujica, pp. 93-146. Actas de Primer Coloquio sobre la Cultura Moche. Travaux de 1'Institut Franqais d'Etudes Andines 79. IFEA, FOMCIENCIAS, and UNT, Lima.
1995 Los Mochicas del norte y 10s Mochicas del sur. In Kcus, edited by K. Makowski, C. B. Donnan, I. Amaro Bullon, L.
J. Castillo, M. Diez Canseco, 0. ElCspuru Revoredo, and J.
A. Mumo Mena, pp. 143-176. Banco de CrCdito del Peni, Lima. Chapdelaine, Claude
1998 Excavaciones en la zona urbana de Moche durante 1996. In Investigaciones en la Huaca de la Luna 1996, edited by
S. Uceda, E. Mujica, and R. Morales, pp. 85-1 16. Facultad de Ciencias Sociales, Universidad Nacional de la Libertad- Trujillo, Peru.
Chapdelaine, Claude, Victor Pimentel, and HClkne Bernier 2001 A Glimpse of Moche Phase 111 Occupation at the Hua- cas of Moche Site, Northern Peru. Antiquity 75:361-372. Cohen, Mark N.
1978 Population Pressure and the Origins of Agriculture: An Archaeological Example from the Coast of Peru. In Advances in Andean Archaeology, edited by D. L. Browman, pp. 91-132. Mouton Publishers, The Hague.
1981 The Ecological Basis of New World State Formation: General and Local Model Building. In The Transition to Statehood in the New World, edited by G. D. Jones and R.
R. Kautz, pp. 105-122. Cambridge University Press, Cam- bridge. Collier, Donald
1955 Cultural Chronology and Change as Rejected in the Ceramics of the Kru' Valley, Peru. Fieldiana: Anthropology Vol. 43. Chicago Natural History Museum, Chicago.
Conrad, Goeffrey W.
1978 Models of Compromise in Settlement Pattern Studies: An Example from Coastal Peru. World Archaeology 9(3):280-298.
1984 The Early Horizon Occupation of the Nepefia Valley, Peru. Unpublished Ph.D. dissertation, Department of Anthro- pology, University of Massachusetts, Amherst.
1987 Toward the Development of the State on the North Cen- tral Coast of Peru. In The Origins and Development of the Andean State, edited by J. Haas. S. Pozorski, andT. Pozorski, pp. 70-82. Cambridge University Press, Cambridge.
D'Altroy, Terence N., and Timothy K. Earle 1985 Staple Finance, Wealth Finance, and Storage in the Inka Political Economy. Current Anthropology 26: 187-206. DeMarrais, Elizabeth, Luis Jaime Castillo, and Timothy K. Earle 1996 Ideology, Materialization, and Power Strategies. Current Anthropology 37: 15-3 l.
Donnan, Christopher B.
1973 Moche Occupation of the Santa Valley, Peru. University of California Publications in Anthropology, Vol. 8. Uni- versity of California, Berkeley.
1976 Moche Art and Iconography. UCLA Latin American Center Publications, University of California, Los Angeles. Earle, Timothy K. 1976 Nearest Neighbor Analysis of Two Formative Settlement Systems. In The Early Mesoamerican Village, edited by K.
V. Flannery, pp. 196223. Academic Press, New York.
1978 The Economic and Social Organization of a Complex Chiefdom: The Halelea District, Kaua'i, Hawaii. Museum ofAnthropology Anthropological Papers No. 63. University of Michigan, Ann Arbor.
1987 Specialization and the Production of Wealth: Hawaiian Chiefdoms and the Inka Empire. In Specialization, Exchange, and Complex Societies, edited by E. M. Brum- fie1 and T. K. Earle, pp. 6475. Cambridge University Press, Cambridge.
1997 How Chiefs Come to Power: The Political Economy in Prehistory. Stanford University Press, Stanford. 2001 Economic Support of Chaco Society. American Antiq- uity 66:2&35. Emerson, Robert M.
1990 Charismatic Kingship: A Study of State-Formation and Authority in Baltistan. In Pakistan: The Social Sciences' Perspective, edited by A. S. Ahmed, pp. 1W145. Oxford Press, Karachi.
Erasmus, Charles J. 1965 Monument Construction: Some Field Experiments. Southwest Journal of Anthropology 2 1 :277-301. Fanington, Ian
1974 Irrigation and Settlement Pattern: Preliminary Research Results from the North Coast of Peru. In Irrigation's Impact on Society, edited by T. E. Downing and M. Gibson, pp. 83-94. Anthropological Papers of the University of Arizona No. 25. University of Arizona Press, Tucson.
1985 Operational Strategies, Expansion, and Intensification within the Prehistoric Irrigation System of the Moche Val- ley, Peru. In Prehistoric Intensive Agriculture in the Tmpics, edited by I. S. Farrington, pp. 621452. BAR International Series 232, Oxford, England.
Feinman, Gary M. 1995 The Emergence of Inequality: A Focus on Strategies and Process. In Foundations ofsocial Inequality, edited by T. D. Price and G. M. Feinman, pp. 255-280. Plenum Publishing Corporation, New York.
Feinman, Gary M., and Jill Neitzel 1984 Too Many Types: An Overview of Sedentary Prestate Societies in the Americas. In Advances in Archaeological Method and Theory, Volume 7,edited by M. B. Schiffer, pp. 39-102. Academic Press, New York.
Flannery, Kent V. 1998 The Ground Plans of Archaic States. In Archaic States, edited by Gary M. Feinman and J. Marcus, pp. 15-58. School of American Research Press, Santa Fe.
Ford, James A., and Gordon R. Willey 1949 Surface Survey of the Viru Valley, Peru. Anthropological Papers of the American Museum of Natural History Vol
43: Part 1. American Museum of Natural History, New York. Fried, Morton H. 1967 The Evolution of Political Society. Random House, New York. Gillin, John P.
1945 Moche: A Coastal Peruvian Community. Institute of Social Anthropology Publications No. 3. Smithsonian Insti- tution, Washington, DC.
Gray, Robert F. 1963 The Sonjo of Tanganyika: An Anthropological Study of an Irrigation-Based Society. Oxford University Press, London. Haas, Jonathan
1981 Class Conflict and the State in the New World. In The Transition to Statehood in the New World, edited by G. D. Jones and R. R. Kautz, pp. 8C102. Cambridge University Press, Cambridge.
1982 The Evolution of the Prehistoric State. Columbia Uni- versity Press, New York.
1985 Excavations on Huaca Grande: An Initial View of the Elite of Pampa Grande, Peru. Journal of FieldArchaeology 12:391409.
1987 The Exercise of Power in Early Andean State Devel- opment. In The Development and Origins of the Andean State, edited by J. Haas, S. Pozorski, and T. Pozorski, pp. 31-35. Cambridge University Press, Cambridge.
Harris, Marvin 1977 Cannibals and Kings: The Origins of Culture. Vintage, New York. Hassan, Fekri 198 1 DemographicArchaeology.Academic Press, NewYork. Hastings, Charles M., and Michael E. Moseley 1975 The Adobes of Huaca del Sol and Huaca de la Luna. American Antiquity 40: 196203. Hatch, John K.
1976 The Corn Farmers of Motupe: A Study of Traditional Farming Practices in Northern Coastal Peru. Land Tenure Center Monographs No. 1. University of Wisconsin, Madi- son.
Howard, Jerry B., and Gary Huckleberry 1991 The Operation and Evolution of an Irrigation System: The East Papago Canal Study. Soil Systems Publications in Archaeology No. 18. Soil Systems, Inc., Phoenix.
Hunt Eva, and Robert C. Hunt 1974 Inigation, Conflict, and Politics: A Mexican Case. In Irrigation's Impact on Society, edited by T. E. Downing and
M. Gibson, pp. 129-158.Anthropological Papers of the Uni- versity of Arizona, No. 25. University of Arizona, Tucson. Isbell, William H., and Katharina Schreiber 1978 Was Huari a State? American Antiquity 43:372-389. Johnson, Allen W., and Timothy K. Earle 1987 The Evolution of Human Society: From Foraging Group to Agrarian State. Stanford University Press, Stanford. Johnson, Gregory A.
1973 Local Exchange and Early State Development in South- western Iran. Museum of Anthropology Anthropological Papers No. 5 1. University of Michigan, Ann Arbor.
Lanning, Edward 1967 Peru Before the Incas. Prentice-Hall, Englewoods Cliffs, New Jersey. Larco, Rafael 200 1 Los Mochicas: Tomos ly 11. Museo ArqueoMgico Rafael Larco Herrera, Lima. Leach, Edmund R. 1959 Hydraulic Society inCeylon. Pastand Present 15:2-26. Lees, Susan H. 1974 The State's Use of Irrigation in Changing Peasant Society. In Irrigation's Impact on Society, edited by T. E. Downing and
M. Gibson, pp. 123-128. Anthropological Papers of the Uni- versity of Arizona No. 25. University of Arizona, Tucson. Massey, Sarah A.
1986 Sociopolitical Change in the Upper IcaValley, 400 B.C. t0A.D. 400. Unpublished Ph.D. dissertation, Department of Anthropology, University of California, Los Angeles.
Mitchell, William P. 1973 The Hydraulic Hypothesis Revisited: A Reappraisal.
Current Anthmpology 14:532-534. 1976 Irrigation and Community in the Central Peruvian High- lands. American Anthropologist 78:2544. Moore, Jeny
1996a The Archaeology of Plazas and the Proxemics of Rit- ual: Three Andean Traditions. American Anthropologist 98:789-802.
1996b Architecture and Power in the Prehispanic Andes. Cambridge University Press, Cambridge. Moseley, Michael E.
1974 Organizational Preadaptation to Irrigation: The Evolu- tion of Early Water-Management Systems in Coastal Peru. In Irrigation's Impact on Society, edited by T. E. Downing and M. Gibson, pp. 77-82. Anthropological Papers of the University of Arizona No. 25. University of Arizona, Tuc- son.
1975a The Maritime Foundations of Andean Civilizations. Cummings Publishing, Menlo Park, California. 1975b Prehistoric Principles of Labor Organization in the Moche Valley, Peru. American Antiquity 40: 191-196. 1992 The Znkas and Their Ancestors. Thames and Hudson, London. Moseley, Michael E., and Eric E. Deeds
1982 The Land in Front of Chan Chan: Agrarian Expansion, Reform, and Collapse in the Moche Valley. In Chan Chan: Andean Desert City, edited by M. E. Moseley and K. C. Day, pp. 25-54. University of New Mexico Press, Albuquerque.
Murra, John V. 1968 An Aymara Kingdom in 1567. Ethnohistory 15:115-151.
1972 El "control vertical" de un mbimo de pisos ecologicos en la economia de las sociedades andinas. In Visita de la Provincia de Ledn de Hudnuco en 1562. Documentos por la Historia y Etnologia de Huanuco y la Selva Central 1 :427-476. Universidad Nacional Hermilio Valdizan, Huanuco.
Netherly, Patricia J. 1984 The Management of Late Andean Irrigation Systems on the North Coast of Peru. American Antiquity 49:227-254. Netting, Robert M. 1974 The System Nobody Knows: Village Irrigation in the Swiss Alps. In Irrigation's Impact on Society, edited by T.
E. Downing and M. Gibson, pp. 67-76. Anthropological Papers of the University of Arizona No. 25. University of Arizona, Tucson.
ONERN (Oficina Nacional de Evaluation de Recursos Naturales) 1973 Inventario, evaluacidn, y uso racional de 10s recursos naturales de la costa: cuenca del Rio Moche, Vol. I and IZ.
Oficina Nacional de Evaluaci6n de Recursos Naturales, Lima. Ortloff, Charles R., Robert A. Feldman, and Michael E. Moseley
1985 Hydraulic Engineering and Historical Aspects of the Pre-Columbian Intravalley Canal Systems of the Moche Val- ley, Peru. Journal of Field Archaeology 12:77-98.
Patterson, Thomas C., John P. McCarthy, and Robert A. Dunn 1982 Polities in the Lunn Valley during the Early Intermedi- ate Period. flawpa Pacha 20:61-82. Peebles, Christopher S.
1978 Determinants of Site Size and Location in the Moundville Phase. In Mississippian Settlement Patterns, edited by B. D. Smith, pp. 369417. Academic Press, New York.
Pozorski, Shelia, and Thomas Pozorski 1979a Alto Salaveny: A Peruvian Coastal Cotton Preceramic Site.Annals ofthe Carnegie Museum 49:337-375. Carnegie Museum of Natural History, Pittsburgh.
1979b An Early Subsistence Exchange System in the Moche Valley, Peru. Journal of Field Archaeology 6:413432. 1987 Early Settlement and Subsistence in the Casma Valley Peru. University of Iowa Press, Iowa City. Pozorski, Thomas
1976 Caballo Muerto: A Complex of Early Ceramic Sites in the Moche Valley, Peru. Unpublished Ph.D. dissertation, Department of Anthropology, University of Texas, Austin.
1980 The Early Horizon Site of Huaca de los Reyes: Soci- etal Implications. American Antiquity 45: lW110.
1982 Early Social Stratification and Subsistence Systems: The Caballo Muerto Complex. In Chan Chan: Andean Desert City, edited by M. E. Moseley and K. C. Day, pp. 225-253. University of New Mexico Press, Albuquerque.
1983 The Caballo Muerto Complex and Its Place in the Andean Chronological Sequence. Annals of the Carnegie Museum 53: 1-39. Carnegie Museum of Natural History, Pittsburgh.
1987 Changing Priorities in the Chimu State: The Role of Irri- gation Agriculture. In The Origins and Development of the Andean State, edited by J. Haas, S. Pozorski, andT. Pozorski, pp. 11 1-120. Cambridge University Press, Cambridge.
Pozorski, Thomas, and Shelia Pozorski 1982 Reassessing the Chicama-Moche Intervalley Canal: Comments on "Hydraulic Engineering Aspects of the Chimu Chicama-Moche Intervalley Canal." American Antiquity 47:851-868.
Renfrew, Colin 1974 Beyond a Subsistence Economy: The Evolution of Social Organization in Prehistoric Europe. In Reconstmcting Complex Societies: An Archaeological Colloquium,
edited by C. B. Moore, pp. 69-95. Bulletin of the American School of Oriental Research 20. Rostworowski de Diez Canseco, Maria 1977 Etnia y sociedad: costaperuanaprehispcinica. Instituto de Estudios Peruanos, Lima. 1978 SerZorios indigenas de Lima y Canta. Instituto de Estu- dios Peruanos, Lima. Rowe, John 1963 Urban Settlements in Ancient Peru. flawpa Pacha 1:l-27. Russell, Glen S., Banks L. Leonard, and JCsus Briceiio
1994 Cerro Mayal: nuevos datos sobre production de cerimica Moche en el Valle Chicama. In Moche: propues- tas y perspectivas, edited by S. Uceda and E. Mujica, pp. 18 1-206. Actas de Primer Coloqu!o sobre IaCultura Moche. Travaux de I'Institut Franqais d'Etudes Andines 79. IFEA, FOMCIENCIAS, and UNT, Lima.
1998 The Cerro Mayal Workshop: Addressing Issues of Craft Specialization in Moche Society. In Andean Ceramics: Tech- nology, Organization, and Approaches, edited by I. Shi- mada, pp. 63-91. MASCA Research Papers in Science and Archaeology, Supplement tovolume 15. University of Penn- sylvania Museum of Anthropology and Archaeology, Philadelphia.
Sanders, William T., and Jeffrey Marino 1970 New World Prehistory: Archaeology of the American Indian. Prentice-Hall, Englewood Cliffs, New Jersey. Sanders, William T., Jeffrey Parsons, and Robert S. Santley 1979 The Basin of Mexico. Academic Press, New York. Sanders, William T., and Barbara J. Price 1968 Mesoamerica: The Evolution of Civilization. Random House, New York. Schaedel, Richard P. 1985 The Transition from Chiefdom to State on the Northern Peru. In The Evolution ofSociopolitica1 Organization, edited
by H. J. M. Claessen, P. van der Velde, and E. Smith, pp. 156169. Bergin and Garvey, South Hadley, Massachusetts. Service, Elman R. 1962 Primitive Social Organization: An Evolutionary Per- spective. Random House, New York. 1975 The Origins ofthe State and Civilization: The Process of Cultural Evolution. Norton Press, New York. 1985 A Century of Controversy: Ethnological Issues from 1860 to 1960. Academic Press, New York. Shimada, Izumi
1987 Horizontal andvertical Dimensions of Prehistoric States in North Peru. In The Origins and Development of the Andean State, edited by J. Haas, S. Pozorski, and T. Pozorski, pp. 13C144. Cambridge University Press, Cambridge.
1994 Pampa Grande and the Mochica Culture. University of Texas Press, Austin. Sidky, Homayun
1996 Irrigation and State Formation in Hunza: The Anthro- pology of a Hydraulic Kingdom. University Press of Amer- ica, Inc., New York.
Spooner, Brian 1974 Inigation and Society: The Iranian Plateau. In Irrigation's Impact on Society, edited by T. E. Downing and M. Gibson, pp. 43-58. Anthropological Papers of the Univer- sity of Arizona No. 25. University of Arizona, Tucson.
Stanish, Charles 1994 The Hydraulic Hypothesis Revisited: Lake Titicaca Basin Raised Fields in Theoretical Perspective. Latin Amer- ican Antiquity 5:312-332.
Steponaitis, Vincas P. 1978 Location Theory and Complex Chiefdoms: A Missis- sippian Example. In Mississippian Settlement Patterns, edited by B. D. Smith, pp. 417453. Academic Press, New York. 198 1 Settlement Hierarchies and Political Complexity in Non- Market Societies: The Formative Period of the Valley of Mexico. American Anthropologist 83:32&363.
Steward, Julian H. 1949 Cultural Causality and Law: ATrial Formulation of the Development of Early Civilizations. American Anthropolo- gist 5 1: 1-27. 1955 Theory of Culture Change: The Methodology ofMulti- linear Evolution. University of Illinois Press, Urbana. 1971 [I9551 Some Implications of the Symposium. In PrehistoricAgriculture edited by S. Struever, pp. 614-648. The Natural History Press, Garden City, New York. Reprinted. Originally published 1955, in Irrigation Civilizations: A Comparative Study, edited by J. H. Steward, R. M. Adams,
D. Collier, A. Palerm, Karl A. Wittfogel, and R. L. Beals. Pan American Union Press, Washington, D.C. Strong, William D., and Clifford Evans, Jr.
1952 Cultural Stratigraphy in the Viru Valley, Northern Peru. Columbia Studies in Archaeology and Ethnology 4. Colum- bia University Press, New York.
Tello, Ricardo 1998 Los conjuntos arquitect~jnicos 8, 17, 18, y 19 del cen- tro urbano Moche. In Investigaciones en la Huaca de la Luna 1996, edited by S. Uceda, E. Mujica, and R. Morales, pp. 117-138. Facultad de Ciencias Sociales, Universidad Nacional de la Libertad-Trujillo, Peru.
Topic, Theresa 1977 Excavations at Moche. Unpublished Ph.D. dissertation, Department of Anthropology, Harvard University, Cam- bridge.
1982 The Early Intermediate Period and Its Legacy. In Chan Chan: Andean Desert City, edited by M. E. Moseley and K.
C. Day, pp. 255-284. University of New Mexico Press, Albu- querque. Topic, Theresa, and John Topic
1982 Prehistoric Fortifications of Northern Peru: Preliminary Report of the Final Season January-December 1980. Unpub- lished manuscript on file at the Department of Anthropol- ogy, Trent University, Peterborough, Ontario.
Trubitt, Mary Beth D. 2000 Mound Building and Prestige Good Exchange: Chang- ing Strategies in the CahokiaChiefdom. American Antiquity 65:669490.
Uceda, Santiago 1997 El poder y la muerte in la sociedad de Moche. In Investigaciones en la Huaca de la Luna 1995, edited by S. Uceda,
E. Mujica, and R. Morales, pp. 177-188. Facultad de Cien- cias Sociales, Universidad Nacional de la Libertad-Trujillo, Peru.
Uceda, Santiago, and Jo3 Armas 1997 Los talleres alfareros en el centro urbano Moche . In Investigaciones en la Huaca de la Luna 1995, edited by S. Uceda, E. Mujica, and R. Morales, pp. 93-104. Facultad de Ciencias Sociales, Universidad Nacional de la Libertad-Tru- jillo, Peru.
1998 An Urban Pottery Workshop at the Site of Moche. In
Andean Ceramics: Technology, Organization, and Approaches, edited by I. Shimada, pp. 91-1 10. MASCA Research Papers in Science and Archaeology, Supplement to Volume 15. University of Pennsylvania Museum of Anthropology and Archaeology, Philadelphia.
Uceda, Santiago, Elias Mujica, and Ricardo Morales (editors) 1997 Investigaciones en la Huaca de la Luna 1995. Facultad de Ciencias Sociales, Universidad Nacional de la Libertad- Trujillo, Peru. 1998 Investigaciones en la Huaca de la Luna 1996. Facultad de Ciencias Sociales, Universidad Nacional de la Libertad- Trujillo, Peru.
Upham, Steadman 1982 Politics and Power: An Economic and Political History ofthe Western Pueblo Indians. Academic Press, New York. Van Gijseghem, Hendrik
2001 Household and Family at Moche, Peru: An Analysis of Building and Residence Patterns in a Prehispanic Urban Center. Latin American Antiquity 12:257-273.
Verano, John W. 2001 The Physical Evidence of Human Sacrifice in Ancient Peru. In Ritual Sacrifice in Ancient Peru, edited by E. P. Ben- son and A. G. Cook, pp. 165-184. University of Texas Press, Austin.
Waylen, Peter R., and Cesar N. Caviedes 1986 El Niiio and Annual Floods on the North Peruvian Lit- toral. Journal ofHydrology 89:141-156. White, Leslie 1949 The Science of Culture. Farrar, Straus, & Giroux, New York. Willey, Gordon R.
1953 Prehistoric Settlement Patterns in the Viru Valley, Peru'. Bulletin 155, Smithsonian Institution Bureau of American Ethnology. United States Government Printing Office, Wash- ington, D.C.
Wilson, David J. 1988 PrehispanicSettlement Patterns in the LowerSanta Val- ley, Peru. Smithsonian Institution Press, Washington, D.C. 1995 Prehistoric Settlement Patterns in the Casma Valley, North Coast of Peru: Preliminary Results to Date. Journal ofthe Steward Anthropological Society 23: 189-227.
Wittfogel, Karl A.
1956 The Hydraulic Civilizations. In Man's Role in Chang- ing the Face of the Earth, edited by W. Thomas, pp. 152-164. Wenner-Gren Foundation, Chicago.
1957 Oriental Despotism. Yale University Press, New Haven, Connecticut.
197 1 Developmental Aspects of Hydraulic Societies. In PrehistoricAgriculture,edited by S. Stmever, pp. 557-57 1. The Natural History Press, Garden City, New York. Reprinted. Originally published 1955, in Irrigation Civilizations: A Comparative Study, edited by J. H. Steward, R. M. Adams,
D. Collier, A. Palerm, Karl A. Wittfogel, and R. L. Beals. Pan American Union Press, Washington, D.C. Wright, Henry T., and Gregory A. Johnson
1975 Population, Exchange, and Early State Evolution in Southwestern Iran. American Anthropologist 77:267-289.
1.Among cultures and even individuals in the same culture, the length of a workday varies. An eight-hour day is used here as a standardized convention for comparing labor investment in canals.
2. An alternative explanation is that the farmers observed by Gillin were leaving certain fields unplanted because of water shortages. According to records complied by ONERN (1973:183), during the years of Gillin's study (1938-1944), only one year was a drought year (1938), two years had low volume but still could support over 11,000 ha of corn, and four years had high volumes. Surprisingly, planting occurred during the three lowest volume years, and fallowing occurred in high volume years. Therefore, fallowing practices recorded by Gillin do not appear to be the result of water shortages.
- Use of proxy records from the eastern slope of the south- ern Andes, such as the ice cores from the Quelccaya Ice Cap, may also be problematic. Within the north coast, river flow vol- ume in any given year varies significantly among rivers. For instance, Waylen and Caviedes (1986) have demonstrated that droughts and El Niiios do not affect all valleys equally. Some valleys, such as the Moche Valley, are much more resistant to regional fluctuations in river volume. Given the problems of correlating river volume within the region, it is unclear whether one can infer a long-distance teleconnection between the Moche watershed and proxy records of precipitation on the eastern slope of the southern Andes.
- Calibrated using M. Stuiver, P. J. Reimer, and R. Reimer's calibration program. Executed online at http://depts.washington.edu/qil/calib/ .
- Based on 7,076 ha on the north side of the lower valley (Moseley and Deeds 1982: Table 2. I), between 3,200 and 3,750 ha on the south side (Moseley and Deeds 1982:Table 2. I), and 2,262 ha in the middle valley (ONERN 1973:154-155). The resulting figure is rounded to 12,550 and 13,200 ha.
Received January 21, 1998; accepted September 30, 2000; revised January 24, 2002.
Figure 11. Location of archaeologicalsites dating to the Middle Moche phase, Moche Valley, Peru (A.D. 400-800). The location of 27 Middle Moche phase sites recorded by the Chan Chan-Moche Project on Pampa Esperanza could not be deter
in the Moche Valley using data from pedestrian sur- veys and test excavation of prehistoric canal seg- ments (Billman 1996; Farrington 1974; Moseley and Deeds 1982; Pozorski 1987). These studies indicate that irrigation passed through four distinct stages of development (Table 1). Each of these stages repre- sented a significant increase in managerial require- ments. Changes in managerial requirements in each of the four stages can be used to test the hydraulic hypothesis in the Moche Valley.
Stage I' GuaAape Phase (ca' 1800-400 B'C')
Irrigation in the Early Guaiiape phase began in the middle Moche Valley-as indicated by a dramatic shift in population there from the coast (Billman 1996: 164-167)-and then expanded into the lower valley in the Middle and Late Guaiiape phases (Bill- man 1996: 158; Farrington 1985:Figure 3a; Moseley and Deeds 1982:35-36; Pozorski and Pozorski 1979a). The middle valley is by far the easiest sec- tion to irrigate. It is currently irrigated by 33 short canals, ranging from .6 to 7.3 km in length (Figure 2) (ONERN 1973:214). Only four are longer than