Politywide Analysis and Imperial Political Economy: The Relationship between Valley Political Complexity and Administrative Centers in the Wari Empire of the Central Andes

Politywide Analysis and Imperial Political Economy: The Relationshipbetween Valley Political Complexity and Administrative Centers in the

Wari Empire of the Central Andes

Justin Jennings and Nathan Craig

Department of Anthropology, University of California, Santa Barbara, California 93106

E-mail: [email protected]

Received August 25, 2000; revision received January 17, 2001; accepted February 15, 2001;

published online October 26, 2001

Journal of Anthropological Archaeology 20, 479–502 (2001)

doi:10.1006/jaar.2001.0385, available online at http://www.idealibrary.com on

This article tests a model for the political economy of the Wari Empire (AD 600–1000) of Peru. This

model divides the empire into core and periphery zones. In the core, Wari political economy was or-

ganized to extract surplus agricultural production to feed the capital. In the periphery, the Wari strove

to extract prestige goods. We suggest that there is a strong relationship between where the empire

chose to locate its centers in the periphery and the political complexity of the local population in

which the center was placed. We argue that in areas of low political organization sites should be lo-

cated near the geographic center of a population. These sites will tend to function as local administra-

tive centers geared toward the organization and exploitation of the area’s wealth potential. In areas

with more complex political organization sites should be located on the margins of a population.

These sites should have functioned as gateway centers controlling, or at least profiting from, interre-

gional exchange. Our model was systematically tested through the use of Geographic Information

Systems (GIS). The results suggest that much of the variability found in Wari site placement in the pe-

riphery can be explained by differences in local sociopolitical complexity. © 2001 Academic Press

INTRODUCTION

Empires are massive, complex, political

entities that present formidable challenges

to traditional scales of archaeological analy-

sis. Since empires extend across political

boundaries, exchange systems, and envi-

ronmental zones, the organization of popu-

lations before they fall under imperial con-

trol will undoubtedly vary from place to

place. Relying on different strategies to take

advantage of these local conditions, an em-

pire will tend to form a “mosaic of control”

(Schreiber 1992:69). In this mosaic, imperial

actions taken in one region may often differ

radically from those taken in an another.

These actions, however, are usually part of

a “grand strategy” of wealth extraction

(after Luttwark 1976). The traditional field

archaeology methods of settlement survey

and site excavation are alone insufficient

for understanding this grand strategy—a

site or region cannot serve as an imperial
479
microcosm. Politywide analysis that sys-

tematically synthesizes the results of these

smaller scale analyses is also necessary.

This article seeks to demonstrate the value

of such an analysis.

In this article, we present a model for the

political economy of the Wari Empire (AD

600–1000) of Peru. This model argues that

Wari, outside of its imperial core, strove to

extract prestige goods. Based on this model,

we suggest that there is a strong relation-

ship between where the empire chose to lo-

cate its peripheral centers and the political

organization of the local population in

which the center was placed. Sites located

near the geographic center of a population

are argued to function as local administra-

tive centers geared toward the organization

and exploitation of the area’s wealth poten-

tial. These sites should be found in areas of

low political organization. We argue that

sites located on the margins of a popula-

tion, on the other hand, functioned as gate-

0278-4165/01 $35.00Copyright © 2001 by Academic Press

All rights of reproduction in any form reserved.

A

way centers controlling, or at least profiting

from, interregional exchange. These types

of sites would tend to be located in areas

with more complex political organization.

Archaeological data on the Wari Empire

remain sparse. A small fraction of the capi-

tal city has been studied and the size and

the structure of the site are still not well un-

derstood. Only a handful of extensive stud-

ies of provincial Wari sites are available.

Our understanding of ancient travel corri-

dors and the location of many important

raw materials remains poor or, occasionally,

nonexistent. We feel, however, that suffi-

cient data exist to begin to understand the

dynamics of Wari political economy. As the

old saying goes, it is better to light a candle

than to curse the darkness. A bit of light can

be shed on the empire by offering tentative

economic models and devising empirical

means of testing them on a politywide

level. In so doing, we hope, Wari studies

480 JENNINGS

can provide insight into the political econ-

omy of other ancient empires.

A MODEL OF WARI POLITICALECONOMY

An empire can be defined as an expan-

sionist state that assumes effective control

over other polities of varying scope and

complexity (D’Altroy 1992:9). The political

economy of an empire is shaped by how

it produces, distributes, and exchanges

wealth (Stanish 1992:11). Although empires

vary tremendously in the ways that they

extract wealth (Claessen and Skalnik

1978; Sinopoli 1994a:165–166), the imperial

strategies employed are usually used to

meet two essential needs. The first is the

subsistence demands of the urbanized cen-

ter of the empire (Johnson and Earle

1987:247). In most cases, a capital develops

to such a degree that it is unable to easily

provide for the subsistence needs of the res-

idents without help from outside producers

(e.g., Schwartz and Falconer 1994:3). Since

some agricultural products spoil quickly

and most are bulky to transport, the area

around a capital often serves as the primary

breadbasket. This is clearly seen in the agri-

cultural core of such empires as the Aztec

(Hassig 1985:133) and Vijayanagara (Sinop-

oli 1994b:226).

We suggest that the Wari Empire also

contained an agricultural core that fed the

capital’s burgeoning, increasingly special-

ized population. By the middle of the Mid-

dle Horizon, the population of the capital

city had risen to at least 10,000 people (pop-

ulation estimates for the city vary between

10,000 and 70,000; see Isbell 1997a:186).

Areas of craft specialization (Gonzalez

Carre 1981:94; Spickard 1983:153–154; von

Hagen and Morris 1998:130) and elite resi-

dences (Isbell 1997a:206) in the city suggest

that at least some of the residents were not

full-time farmers. The capital lies in a valley

whose agricultural productivity, already

massively terraced and irrigated by the

smaller population of the previous period

(Lumbreras 1974b; Isbell 1988:74–75), may

not have been able to sustain a large urban

center (Lumbreras 1974a:163, Gonzalez

Carre 1981:88). It seems likely that agricul-

tural products must have been brought

from outside the valley. Research in areas

surrounding the valley supports this idea

by demonstrating a Wari focus on the inten-

sification of agricultural exploitation in

these localities (Raymond and Isbell 1969;

Isbell 1977; Vivanco and Valdez 1990; Ray-

mond 1992; Browman 1999). We believe

that this region, the area of Ayacuchoan in-

fluence before imperial expansion (Lumbr-

eras 1981:23), provided the surplus needed

to maintain the city.

The second basic need of an empire is the

support and legitimization of the imperial

hierarchy (Brumfiel and Earle 1987:3; Earle

and D’Altroy 1989:187; Brumfiel 1994:1).

This need can be met by a number of means,

several of which are often found operating

within the same society (Blanton

1998:141–148). One of the most prevalent

ways is through the extraction of prestige

ND CRAIG

items or the conversion of staple goods into

wealth. Since surplus production in the core

n

of the empire is often primarily geared to-

ward sustaining the urban center, these

goods must be largely collected in the pe-

riphery. As a heuristic, imperial extraction

strategies can be divided into wealth and

staple finance systems (D’Altroy and Earle

1985). Staple finance involves “obligatory

payments in kind to the state of subsistence

goods such as grains, livestock, or clothing”

(D’Altroy and Earle 1985:188). In a staple fi-

nance system, the bulk and potential

spoilage of staple goods (D’Altroy and Earle

1985:188) necessitates a substantial invest-

ment in storage facilities in peripheral ex-

traction centers (e.g., LeVine 1992). Despite

earlier convictions (Rowe 1963:14), scholars

have yet to find compelling evidence for the

use of massive storage facilities in the Wari

Empire.1,* Although it is possible that Wari

storage systems have yet to be identified,

present evidence suggests that the empire

did not use a staple finance system.

Wealth finance, on the other hand, in-

volves “the manufacture and procurement

of special products (valuables, primitive

money, and currency)” (D’Altroy and Earle

1985:188). An empire will generally benefit

by adopting a wealth finance system for sup-

porting the imperial hierarchy. This switch

typically allows for a reduction in transporta-

dle Horizon Peru (p. 174), by K. Schreiber, 1992, An

gan. Adapted with permission.

tion and storage costs since wealth goods

are often less bulky and more durable than

*See Notes section at end of article for all footnotes.

staple goods. (D’Altroy and Earle 1985:

193–194). Exchange of metals, obsidian, dec-

orated ceramics, textiles, turquoise, and spe-

cific marine shell (spondylus and strombus)

increase during the Middle Horizon (Burger

and Asaro 1977; Lechtman 1980; Shady Solís

1988). These items, found both in ritual con-

text in Wari administrative centers (Cook

1985, Cook 1992:359; Williams et al. 2000:73)

and at the site of Wari itself (Cabera Romero

1996:88–91, Pèrez Calderón 1995: 85–86,

Gonzalez Carre et al. 1996: 100–102;

Isbell 1985:70), were prestige items for at

least 2 millennia in the Andes. The

increased exchange of these goods, com-

bined with an absence of evidence for state

storage system of staple surplus, suggests

that the Wari organized its peripheral politi-

cal economy around the production and ex-

change of prestige objects.

The means by which the empire ex-

tracted goods from the core and periphery

were undoubtedly variable. As Schreiber

has noted (1992:267), the empire adapted its

strategies to local conditions. A region’s dis-

tance from the core, political organization,

wealth potential, and tolerance to outside

rule largely determined the amount of in-

vestment that the empire would make in

the area. In many regions, the empire chose

to construct imperial sites.

Over 30 sites have been found that con-

Arbor, Univ. of Michigan. © University of Michi-

POLITYWIDE ANALYSIS OF WARI POLITICAL ECONOMY 481

FIG. 1. Wari administrative architecture from the site of Jincamocco From Wari Imperialism in Mid-

tain buildings that follow a rigid architec-

tural canon (Schreiber 1978; Spickard 1983;

A

Isbell 1991b). Aspects shared by these cen-

ters (see Fig. 1) include (a) a high, rectangu-

lar enclosure wall; (b) limited access—usu-

ally only one or two entrances were made

into the site; (c) interior divided into rectan-

gular cells that are composed of an open

courtyard flanked by corridors on at least

two sides; and (d) restricted access through

the structure and between cells. The centers

are imposing structures that may have been

designed to appear invincible and bureau-

cratically efficient (Spickard 1983:141). The

centers, however, were not military gar-

risons. They were neither built in defensive

locations nor built or organized to defend

any kind of attack. Instead, excavations at

the centers indicate that the centers housed

482 JENNINGS

a group of elites whose tastes ran more for

feasting and ritual than for battle (Schreiber

1992, McEwan 1998).

A HYPOTHESIS FOR THE WARIPERIPHERY

This article tests if the locations chosen

for administrative centers fits with the ex-

pectations of our model. According to our

model of Wari political economy, the impe-

rial strategy in the periphery was to extract

prestige goods. This strategy will cause ad-

ministrative sites, if they are established at

all, to be located in significantly different

geographic locations in regard to the local

populations that these sites are situated to

control.

The degree of complexity will largely de-

termine what strategy the empire must pur-

sue when entering a valley. We argue that

Wari sites built in valleys of low political

complexity will be found near the geo-

graphic center of the valley’s population. In

independent villages dominated by a kin-

based mode of production, village leaders

will tend to have difficulty coercing surplus

production on a regular basis (Cobb

1996:254). Another level of political hierar-

chy would therefore be needed to extract

wealth from the region (Schreiber 1992:24).

One of the strategies used to build a politi-

cal organization is to construct a regional

administrative center (Schreiber 1992:24).

According to central-place theory, the posi-

tion of service centers, even in nonmarketbased societies, will tend to be located cen-

trally to the people that use them (Smith

1976:7). Since an administrative center of a

valley functions to exercise political and

economic control over the people of the val-

ley, the optimal location of this site is where

the costs of maintaining control would be

the lowest. The administrative site will

therefore tend to be located near the center

of the valley’s population.

Another imperial control strategy is to

place sites on the margins of valley as gate-

way administration centers. Sites located at

strategic locations on the edges of regions

are ideally suited to the control of interre-

gional exchange (Hirth 1978:37). An admin-

istrative center in this location would tend

to focus more on controlling a valley’s ex-

ternal exchange than the political control of

the valley itself. These “gateway” centers

(after Hirth 1978) should be found in val-

leys of high political complexity. In more

complex societies, elites often actively par-

ticipate in the long distance exchange of

prestige goods (Cobb 1996:258). These

goods are often the major means by which

elite power is maintained (Cobb 1996; Earle

1997:7; Baines and Yoffe 1998:253). In con-

trolling the exchange routes of these goods,

an empire can siphon off a portion of these

resources to feed the appetite of the core. By

adopting these strategies, Wari could

cheaply and effectively control the areas of

high political complexity under its influ-

ence. Administrative centers in areas of

high political complexity therefore will

tend to be found near the margins of valley

as gateway centers controlling long-dis-

tance exchange.

In archaeology, we rarely have the luxury

ND CRAIG

of studying variables in the way that we

would like. In this case, directly analyzing

the relationship between site location and

W

local political organization proved to be im-

possible. We would ideally measure the

distance of each Wari site from the center of

a valley’s population. Unfortunately, we sim-

ply do not know where populations were

during this period for most of the valleys in

which Wari centers are found. Valleys in the

Andes, however, are sharply bound oases

of arable land and permanent settlement in-

terspersed within high puna, mountains,

and desert. Although these nonvalley areas,

especially the high puna, were core compo-

nents of the Andean economy, population

size was very small in these regions and

usually directly associated with valley set-

tlements. The geographic boundaries of these

valleys were therefore used as proxy mea-

surements for the extent of local popula-

tions in our analysis.

In this article, we test the relationship be-

tween administrative site location and local

political organization. In valleys of simple

political complexity, Wari administrative

sites should tend be placed near the center

of the valley. In valleys occupied by more

complex polities, administrative sites

should tend to be located toward the mar-

gins of the valley—further from the valley’s

POLITYWIDE ANALYSIS OF

50 3 25 m rectangular enclosure sur-

center. The remainder of this article demon-

strates how this relationship was tested and

what the results of this test were.

DATA: VALLEY POLITICALORGANIZATION AND

ADMINISTRATIVE CENTERS

At least, 20 Wari peripheral administra-

tive centers are found in 11 valleys across

much of modern-day Peru2 (Fig. 2). To pro-

vide an adequate sample to test our site lo-

cation model, all centers, even those that

are disputed, are included in this analysis.

Those centers, however, conclusively

shown to be non-Wari design, such as

Pampa de las Llamas (Pozorski and Po-

zorski 1987:32), El Purgatorio (Pozorski and

Pozorski 1987:38; T. Pozoski personal com-

munication and 1999), and Chimu Capac

(Valkenier 1995:279), were not included.

Further, numerous sites attributed to the

empire, such as San Nicolás in the valley of

Supe (Lumbreras 1974b:155), Yanahuanca

in the Pasco valley (Isbell 1988:186), and

Wisajirca (MacNeish, Patterson, and Brow-

man 1975:60) in Huanaco, could not be in-

cluded in this analysis because insufficient

data exist to consider the nature of these

sites.

Data on pre-Wari political organization

were also compiled for each of the valleys

in which administrative centers have been

found. Archaeological literature pertaining

to each valley was examined for this pre-

ceding period, the Early Intermediate Pe-

riod (AD 0–700), and the valley’s political

organization was ranked as either simple or

complex. For our purposes, if a valley’s in-

habitants were organized in a political hier-

archy above the village level, the valley was

labeled complex. If the inhabitants lived as

in independent villages and/or as scattered

agropastoralists, the valley was labeled as

simple. The list below provides a valley-by-

valley account of the data used to assess

valley political organization and identify

Wari administrative centers. The data, in

part, are summarized in Table 1.

Cajamarca–Huamacucho Valley

Early Intermediate Organization. During

the Early Intermediate period, this large

valley was divided into two polities that

were probably organized as paramount

chiefdoms with wide-ranging trade rela-

tions (Ravines 1985:15). The first polity in

the north was centered on the site of Coyor

(Julien 1988:240). The second polity, in the

south, was centered on the site of Marc-

ahuamacucho (Thatcher 1975; Topic and

Topic 1984, 1987).

Huacaloma de Sulluscocha (Julien 1988:265).A 2.45-hectare site, Huacaloma consists of a

ARI POLITICAL ECONOMY 483

rounded by local architecture. In his settle-

ment survey, Julien associates the structure

The site of Wari is shown with a pentagon, while Wari Administrative sites are shown as white triangles. The

GTOPO30 digital elevation model is shown with analytical hillshading to illustrate topography. These data are

Lambert-Equal Azimuth Projected with a longitude of origin at 260° and a latitude of origin at 15°.

with Wari by its architecture. The surround-

ing settlement may postdate the enclosure.

El Palacio (Julien 1988:262). A 0.3-hectare

site, El Palacio consists of a large rectangu-

lar compound measuring 70 3 50 m. The

structure, based on its architectural ele-

ments, is Wari.

Ichabamba (Williams and Pineda 1985:59). A0.9-hectare site, Ichabamba is a 120 3 70 m

rectangular enclosure that is identified as

Wari by its architecture. The identification

of the site is based solely on air photos.

Santa Delia (Julien 1988:270). The largest

site found in Julien’s survey, Santa Delia

covers some 75 hectares. Within the site,

several medium- to large-sized rectangular

enclosures are found that, based on their ar-

484 JENNINGS AND CRAIG

FIG. 2. Locator Map of the study area showing the 20 Wari administrative sites included in this analysis.

chitectural elements, are Wari. Julien does

not specify the number or exact size of

these enclosures.

Viracochapampa Cajamarca-Huamacucho Complex Valley edge

Yamobamba Cajamarca-Huamacucho Complex Valley edge

Viracochapampa (McCown 1945; Topic andTopic 1987; Topic 1991). A massive rectan-

gular enclosure, Viracochapampa covers

an area of 32 hectares. The site, however,

may never have been completed or occu-

pied. Ceramic sherds associated with

temporary labor camps date the construc-

tion of Viracochapampa to Middle Hori-

zon 2A.

Pampa de Yamobamba (Williams and Pineda1985:59; Julien 1988:292). A 1.3-hectare site,

Yamobamba is a rectangular compound

measuring 130 3 210 m. The site’s architec-

ture is Wari.

Callejon de Huaylas Valley

Early Intermediate Organization. In the

1960s, Gary Vescelius conducted an exten-

sive project of survey and excavation in the

region. The notes for the project are fa-

mously lost and only brief summaries re-

Socos Chillon

garding this work exist (Lanning 1965:140;

Buse 1965:317–333). These summaries sug-

gest that stratified societies populated the

region and the presence of monumental

tombs (Isbell 1997b:195–205), both rich and

poor burials (Larco Hoyle 1966:106; Grieder

1978:45) and large sites with monumental

architecture (Greider 1978; Bennet 1944)

support these claims. Current research in

the region (George Lau, personal communi-

cation and 2000) demonstrates that chief-

dom level societies existed in at least some

areas of the valley.

Honcopampa (Isbell 1989, 1991a). A

densely occupied 3-hectare site is made up

primarily of individual patio groups (at

least 18) and two D-shaped enclosures.

The site is clearly Wari but its architecture

is more reminiscent of that seen in the

Capital City than the rectangular enclo-

sures typical of domestic architecture. Pot-

tery from the site appears to date to Mid-

dle Horizon 2A.

Pariamarca (Williams and Pineda 1985:60).This 0.96-hectare site is made up of a 120 3

Complex Valley edge


TABLE 1Pre-Wari Valley Political Organization and Our Predictions for Wari Site Placement

Site Name Valley Name Valley Organization Wari Site Placement

Acachiwa Colca Simple Valley center

Cerro Baul Upper Moquequa Simple Valley center

Cerro Mejia Upper Moquequa Simple Valley center

Collota Cotahuasi Simple Valley center

El Palacio Cajamarca-Huamacucho Complex Valley edge

Honcopampa Callejon de Huallyas Complex Valley edge

Huacaloma Cajamarca-Huamacucho Complex Valley edge

Huaro Cuzco Simple Valley center

Ichabamba Cajamarca-Huamacucho Complex Valley edge

Jincamocco Carhuarazo Simple Valley center

Numero 8 Chuquibamba Simple Valley center

Pariamarca Callejon de Huallyas Complex Valley edge

Pataraya Nasca Complex Valley edge

Pikillacta Cuzco Simple Valley center

Santa Delia Cajamarca-Huamacucho Complex Valley edge

Sonay Camana Simple Valley center

Tocroc Callejon de Huallyas Complex Valley edge

80 m rectangular enclosure. The architec-

ture has been identified as Wari by air pho-

tographs.

Tocroc (Williams and Pineda 1985:60). This

0.96-hectare site is made up of a 120 3 80 m

rectangular enclosure. The architecture has

been identified as Wari by air photographs.

Carhuarazo Valley

Early Intermediate Organization. Before the

Middle Horizon, settlers in the valley lived

in 7 small villages and 17 seasonally occu-

pied hamlets (Schreiber 1987:95). With

Wari, the valley became centrally organized

and agriculture was intensified (Schreiber

1987, 1992:143–144). The valley was not po-

litically centralized and no status differenti-

ation within or between villages appears to

have existed (Schreiber 1992:144).

Jincamocco. A 15-hectare rectangular en-

closure, this site dates from the MH1B to 2A

(Schreiber 1992). The valley also contains

three smaller sites from the Middle Horizon

that probably assisted imperial actions.

Cuzco Valley

Early Intermediate Organization. Survey in

the valley dates all major sites to the Middle

Horizon or later (McEwan 1987). The

tremendous rate of natural erosion in the

valley may have buried some early sites

(McEwan 1987:7), and Early Intermediate

Period ceramics have been found in the

soundings and surface collections (McE-

wan 1984:12). No evidence for complex so-

cial organization has yet been found for the

region.

Huaro (Glowacki and Zapata 1998). In 1995,

Julinho Zapata found what appears to be a

massive Wari complex scattered over ap-

proximately 150 hectares. Evidence from

preliminary excavations suggest that a few

of the buildings contain aspects of adminis-

trative architecture. Ceramics date from

486 JENNINGS A

Middle Horizon 1B to 2A.

Pikillacta (McEwan 1987, 1989, 1991, 1996,1998). A massive 47-hectare rectangular en-

closure measuring 630 3 745 m, Pikillacta

was never completed. Unlike Viracocha-

pampa, a portion of the site was occupied.

Ceramics found at the site date from the Mid-

dle Horizon 1B to 2A (Glowacki 1996:479).

Chillon Valley

Early Intermediate Organization. During

much of the Early Intermediate Period, the

valley was made up of politically indepen-

dent villages that participated in interre-

gional trade (Gonzales Carbajal 1998:56) By

the end of the period, however, the Chillon

valley became part of the Lima polity

(Shady Solís 1982:18). Differences in settle-

ment size and monumental architecture

suggest that the political complexity of this

polity was at least two-tiered (Arguto

Calvo 1984:84; Shady Solís 1989:6; Bruhns

1994:199).

Socos (Isla and Guerrero 1987). The site

contains a number of different structures

over an area of at least 13 hectares. The

largest building appears to be a rectangular

enclosure measuring 180 3 86 m. Other

buildings also exhibit Wari-derived archi-

tectural traits. The enclosure dates to Mid-

dle Horizon 2B. It is possible that the bulk

of the site postdates the enclosure.

Nasca Valley

Early Intermediate Organization. Survey

work in the Nasca valley suggests that the

valley was organized into a series of small

polities by the end of the Early Intermedi-

ate Period (Schreiber 1999:168). Although

not a centralized state, political organiza-

tion in Nasca was clearly above the village

level (Schreiber 1999).

Pataraya (Schreiber 1999:169). A 0.2-

hectare site, Pataraya is a rectangular enclo-

sure measuring 40 3 50 m. Ceramics associ-

ND CRAIG

ated with the site date to Middle Horizon 2.

A small area of domestic architecture lies

immediately adjacent to the enclosure and

may be part of the site.

W

Camana Valley

Early Intermediate Organization. Although

the area has not been intensively surveyed,

archeological work in the valley suggests

that it was occupied by the Early Intermedi-

ate Period (Manrique and Cornejo 1990). As

far as current research indicates, however,

there is nothing to suggest anything but vil-

lage-level political organization in the re-

gion during this period (Malpass personal

communication, 1999).

Sonay (Malpass et al. 1997). This 0.45-

hectare site is made up of a rectangular en-

closure measuring 90 3 50 m. Excavations

at the site uncovered Wari derived sherds

that have not yet been fully analyzed.

Cotahuasi Valley

Early Intermediate Organization. Accord-

ing to survey work on the valley, the valley

was organized into a number of small vil-

lages by the beginning of the Middle Hori-

zon (Chávez Chávez and Salas Hinijoza

1990:18, Trawick 1994:71, and Jennings in

preparation). No site size hierarchy is ap-

parent in the valley and no evidence for so-

cial stratification is known.

Collota (Chávez Chávez 1982:86–89; Traw-ick 1994:72). The site is made up of two

rectangular enclosures covering approxi-

mately 0.75 hectares. The first enclosure

measuring 90 3 50 m and the second 30 320 m. An associated Wari site contains ce-

ramics from MH 2 and 2A (Jennings, in

preparation.).

Colca Valley

Early Intermediate Organization. The ar-

chaeological work done in the valley sug-

gests that the Early Intermediate occupa-


tion in the valley was most likely confined

to pastoralists and small, independent vil-

lages (de la Vera Cruz 1987:115, de la Vera

Cruz 1988:121–122, Brooks 1998:299).

Acachiwa (de la Vera Cruz 1987:97–98).This 35-hectare site is made up of a rectan-

gular enclosure measuring 500 3 700 m. Al-

though the internal subdivisions are poorly

preserved, the ceramics are derived Middle

Horizon 2. Some scholars (Schreiber

1992:104), however, dispute the authentic-

ity of the site.

Chuquibamba Valley

Early Intermediate Organization. Survey in

the valley suggests low population densi-

ties before the Middle Horizon (Sciscento

1989, Cardones Rosas 1993). No site size hi-

erarchy is apparent in the sites that may

date to directly before the introduction of

Wari occupation in the valley (Sciscento

1989:275) and no indication for social strati-

fication has yet been found. The existing ev-

idence suggests that political organization

in the valley before the Middle Horizon

was simple.

Numero 8: (Sciscento 1989). This 6.5-

hectare site is one of the largest in the valley

and dates from the Middle Horizon

through the Inka period (Sciscento

1989:206). Among the site’s local architec-

ture, a 40 3 50 rectangular enclosure stands

out. Although the investigator felt it prema-

ture to classify this enclosure as a Wari cen-

ter (Sciscento 1989:266), architectural de-

tails of the structure are sufficient enough to

include the site in our analysis.

Upper Moquequa Valley

Early Intermediate Organization. Research

in the valley indicates that population den-

sities were low until the advent of the Mid-

dle Horizon (Rice and Wantanabe 1990:26).

During this period, a colony of the Ti-

wanaku State moved in the area to central-


ize the middle portion of the valley and sys-

tematize agriculture (Goldstein 1990:36).

The Wari Empire also established a site in

the upper portion of the valley and sub-

A

stantially modified the landscape. Al-

though the relative timing of each polity’s

intrusion into the valley during the Middle

Horizon and their extent of interaction is

still unclear (Williams 1999), political com-

plexity of the region during the Early Inter-

mediate Period did not appear to rise above

the village level.

Cerro Baul (Moseley et al. 1991; Williams1997). On the summit of an imposing mesa,

this 5-hectare site dates from the MH1B-

MH2A. The site’s architecture consists of

rectilinear structures with corridors and pa-

tios and occasionally structures of circular

or D-shaped form.

Cerro Mejia (Moseley et al. 1991; Williams1997). Situated on a hill adjacent to Cerro

Baul, this 7 hectare site also dates from the

MH1B-2A. The architecture on the site is

sparser and more poorly preserved that

that on Cerro Baul. Surface remains suggest

488 JENNINGS

ml). While the data are not at an optimal

that some of the buildings on the site were

rectangular compounds in the Wari style

(Moseley et al. 1991:135).

DATABASE CONSTRUCTION

The imposition and maintenance of im-

perial control, be it by Wari or by any

other polity, is an explicitly spatial prob-

lem. Descriptions and explanations of

whole systems, like polities, should at-

tempt to take into account the activities of

each of the system’s component parts

(Mayr 1984). Politywide analysis must in-

tegrate both local and regional scales of

analysis from data collected by multiple

researchers working on different projects.

Geographic Information Systems (GIS)

provides an information infrastructure

permitting organization of data collected

by independent research efforts and al-

lows systematic analysis at varying spa-

tial scales (Zubrow 1990; Crumley and

Marquardt 1990:73–74). GIS was therefore

used to test the relationship between site

location and valley complexity in our data

set.

The data on administrative centers and

valley political complexity was integrated

into an ArcView 3.0a GIS database. The fol-

lowing steps were taken to construct the

database used to test our ideas about Wari

site placement. First, an elevation model

was obtained and cropped to the study

area. Second, site locations were obtained

and represented in the computer database

as a set of points. Third, valleys were de-

fined and modeled as polygons. Fourth, a

method of determining valley center was

chosen and applied to each valley. Last,

varying measurements of distance from

center to site were determined and com-

pared. Each of these steps is covered in turn.

Map Database

An empire, like Wari, covers an enor-

mous spatial extent. A model of the earth,

covering the entire extent of the polity in a

common resolution and projection, was

needed for this analysis. Fine-resolution

topographic maps covering this expansive

region have not yet been digitized and or-

ganized into a single coregistered database.

Coarse resolution digital elevation data,

however, do exist for the entire Earth and

are available for free over the World Wide

Web [from Earth Resources Observation

System Data Center (EDC) Distributed Ac-

tive Archive Center (DAAC) (http://

edcwww.cr.usgs.gov/landdaac/glcc/glcc.

html)]. These 1-km, or 30 arc sec, data are

part of the GTOPO30 global digital eleva-

tion model. Using a cropped portion of the

GTOPO30 data set fulfills the basic need of

having a consistent Earth model into which

sites can be placed for executing spatial

analysis. The GTOPO30 data have been

widely used for a variety of modeling

purposes including hydrologic analysis

and drainage basin delineation (http://

edcsnw3.cr.usgs.gov/topo/hydro/apps.ht

ND CRAIG

resolution (Kvamme 1990:112), they do pro-

vide the basis for proceeding with a polity-

wide analysis of Wari administrative center

site selection until finer resolution data can

be constructed.

Wari Site Location

Accurate site location information was

crucial for this analysis. In most cases, site

coordinates were obtained directly from re-

searchers who visited or worked at these

sites. When this was not possible, coordi-

nates have been taken from published

maps onto which site locations have been

plotted. Once values for site locations were

obtained these were entered into the GIS

and represented as a layer, or theme, of

points. Additional attributes about each site

(size, type, previous occupation, and infor-

mation references) were added to the

theme’s attribute table prior to analysis.

Wari peripheral administrative centers that

are included in the analysis are reported in

Table 1.

Defining Valleys

Examining possible variation in how

Wari invested in the construction and

maintenance of control in different valley

systems forms a core question of this re-

search. In order to pursue the problem, it

was necessary to define the spatial extent

of each valley system in which we exam-

ine Wari administrative site selection. A

variety of objective methods for defining

valley limits were considered. However,

objective valley definition was quite diffi-

cult given the broad spatial setting and

the features we wished to model.3 Valley

extents were determined by examining

the physical structure of each valley repre-

sented in the Earth model. An isopleth

map of elevation was constructed such

that isobars representing contour lines

were drawn for the entire region (John-

POLITYWIDE ANALYSIS OF W

ston 1998:80; DeMers 1997:258). Contour

lines, or isobars, were then displayed on

top of the Digital Elevation Model. Inspec-

tion of the contour lines provided a good

way of representing where valley edges

met with flatter intervalley puna high-

lands.

On an isopleth map, the flatter highlands

are represented by isobars that are further

apart due to the fact that there is less

change across distance (DeMers 1997:260).

Likewise isopleth mapping of steep valley

edges produces closely spaced isobars due

to greater slope. Knowing these facts of iso-

pleth mapping, it was possible to identify

individual contours where the slope within

a polygon was high while the slope outside

of the polygon was much lower. For each of

the valleys, higher resolution topographic

and land cover maps were examined. On

the coast these higher resolution maps were

particularly important sources of informa-

tion on valley extents, since valleys tend not

to produce the deep drainage channels seen

in the highlands. The extent of arable land

as represented on these maps became an

important source of information for defin-

ing coastal valley limits.

Transitional isolines and additional infor-

mation about valley structure were used for

defining valley system boundaries on a val-

ley by valley basis (Fig. 3). This subjective

method of feature definition has precedents

in geography (Clarke personal communica-

tion and 1999). In order to proceed with

analysis it was necessary to make a deci-

sion, and occasionally those decisions must

be arbitrary. We hope to have made our val-

ley definition decision rules explicit, and

we encourage researchers to comment on

how valley extents were defined. If it is pos-

sible to improve the model by modifying

the shape and extent of a particular valley

we are interested in incorporating those

changes.

Defining Valley Center


Once valley systems were defined and

modeled as polygon’s, it was necessary to

develop a spatial point of reference that

units are in hours walking and the contour interval is 1h. The valley’s centroid is indicated by the cir-

cle with a cross. The Wari administrative center of Collota is represented as a triangle.

could be used to test theoretical predictions

about where Wari sites are located with re-

spect to the center of each of those systems.

Defining the center of a polygon is a com-

plex geographic problem, and there are a

variety of solutions to the question of

“where is the center of this polygon.” For

this analysis, we computed a polygons cen-

troid using the trapezoid rule with a

weighted mean center (DeMers 1997:195;

Clarke 1995). Using this method it is possi-

ble to have a centroid that is outside of its re-

spective polygon. Since we wanted to

model centrality within valley systems, cen-

troids were forced to occur within polygons.

Distance from Center

Since our theoretical expectations relate

to the distance Wari sites are predicted to

fall from a valley’s center, we computed

two kinds of “distance from” layers that


FIG. 3. (a) The topography of the Cotahuasi Valley and the Wari administrative site of Collota. Con-

tour interval of the isolines is 100 m. (b) The boundary defined for the Cotahuasi Valley. The valley’s

centroid is indicated by the circle with a cross. The Wari administrative site of Collota is represented as

a triangle. (c) Dotted contour lines represent distance from the centroid of the Cotahuasi Valley. The

cover the study area using the GTOPO30 el-

evation data. First, we computed a raster

theme such that each cell contains a value

representing the isotropic Euclidean dis-

tance from each valley’s centroid in units of

meters. Second, applying Tobler’s hiking

function (Tobler 1993; Kantner 1997) we

computed a cost surface that represents

space nonisotropically in terms of walking

time.

Aldenderfer (1998:12) has tested Toblers

hiking function by comparing computer

generated travel times to actual walking

trips in the Andean high-sierra and puna

rim region of the Río Osmore drainage.

Aldenderfer’s test walkers included both

indigenous Aymara speakers and North

American researchers. Ten trips were timed

and compared to the same trips modeled

on the computer using Torlber’s (1993) hik-

ing function. Walking trips ranged in dis-

tance from about a half a kilometer to 16

km. Results of Aldenderfer’s tests showed

that the model slightly underestimates

travel times, but that “it appears to be a

reasonably good estimator of travel times

in mountainous terrain” (Aldenderfer

1998:15).

For execution in the Arc/Info GIS pack-

age, Tobler’s transformation can be ex-

pressed using the following formula pro-

vided by Kantner (1997):

T 5 D/(6 exp(23.5 * abs(S 1 .6))),

where T 5 traveltime across a given dem

cell, D 5 euclidean distance across a given

dem cell, and S 5 slope of a given cell.

This nonisotropic friction surface was ap-

plied in computing a cost distance grid

theme using each valley’s centroid as the

source cells. This resulted in a grid such


that cell values represented distances from

valley centers in units of hours walking.

ANALYTICAL METHODS

After constructing the database, we

needed to develop a statistic that would

allow us to test our expectations about the

relationships between early Intermediate

Period political complexity and the place-

ment of Wari administrative centers during

the Middle Horizon. Recall that when Wari

expanded into politically complex valleys,

we expect sites to be located in places that

facilitate the movement of goods back to

the core since the empire is co-opting exist-

ing prestige good exchange networks.

When this is the case, on average we expect

sites to be located further away from the

valley’s center. Likewise, we expect Wari

would have had to develop an administra-

tive infrastructure in those valleys where

the preexisting political complexity was

low. In these cases, we expect Wari sites to

be closer to the center of the valley since

there is a need to establish political and eco-

nomic control over unorganized settle-

ments within the valley.

The metric we developed treats each ad-

ministrative center as an independent case

and relies on differences between ob-

served and expected values in Euclidean

distances from the centroid. For each val-

ley, we randomly generated a 20% sample

of points (R) such that no two points could

be closer than a kilometer. We forced

points to be further than 1 km apart so

that no grid cell within the valley will be

sampled more than once. These randomly

generated points represent expected site

locations if no factors were influencing

Wari site selection.

The random points layer for each valley

was used to query the Euclidean distance

grid theme. The mean values returned

from each (R) set of points was calculated

independently on a valley-by-valley basis.

This produced an isotropic, Euclidean, av-

erage expected distance from the center for

each valley (mR). The points representing

Wari sites were also used to query the Eu-

clidean grid theme returning an isotropic

measure of observed distance from the cen-

troid for each site (W). Each Wari site is

treated as an independent case regardless

of whether or not it was the only site in the

valley. Once isotropic observed and ex-


pected values were computed we sub-

tracted observed Wari distances from cen-

5 distance from center statistic; R 5 distance from

center for a %20 random sample based on area; W 5

ter from the expected distance from center

(mR 2 W), producing a residual. This resid-

ual, in units of meters, describes how aver-

age expected distances from the centroid

differ from an observed distance from the

centroid within each valley. Positive residu-

als indicate sites that are closer to the center

than expected, while negative residuals rep-

resent sites that are further from the center

than the mean of our 20% random sample.

In order to place all of the sites, dispersed

between 12 different valleys of varying

size, on a comparable axis we divided the

residual (mR 2 W) by the square root of val-

ley area (A), which was measured in units

of square meters. Division by the square

root of a valley’s area cancels out any units

leaving a unitless numerical descriptor

(Fig. 4).

Two key data constraints needed to be

taken into account when developing this

spatial metric. First, valley definition was

judgmental and subject to revision. How-

ever, the metric should be relatively resis-

tant to most changes in valley shape. It is

clear that if corrections need to be made to

a valley’s boundaries then location of the

polygon’s centroid will move. This is po-

tentially significant since the centroid

which is a higher order feature forms a

reference point for analysis. However,

changing the shape of a valley system will

also modify where the randomly selected

points can be placed. The distance of

every point to the center will change, but

where a particular Wari site falls with re-

spect to the full range of possibilities

should remain resistant to all except the

most drastic changes in valley shape. Sec-

ond, valley area is not constant some val-

leys are very large and others are quite

small. Any distance metric used to com-

pare valleys of differing sizes must ac-

count for the fact that there is great vari-

ability in the spatial extent of the valleys.

492 JENNINGS A

While large valleys will have a much

greater average distance from the center,

dividing by valley area standardizes the

metric controlling for this source of vari-

ability.

It is not logical to apply the distance met-

ric to the nonisotropic cost surface pro-

duced using Toblers (1993) hiking function

because it does not make sense to divide

units of hours by square meters. Therefore

we did not attempt to control for valley

area with the nonisotropic measure of dis-

tance. Differences in walking times could

be influenced by valley area, but expressing

distance of a Wari site; A 5 valley area.

ND CRAIG

FIG. 4. The distance statistic formula is shown on

the left and a schematic example of a valley is shown

on the right. In the diagram on the right the valley

centroid is represented by a cross, randomly sampled

points are shown as dots, and the location of a Wari

site in this hypothetical case is shown as a triangle. D

distances in units of hours walking should

provide a useful basis for comparing sites’

distances from center in different valleys.

RESULTS AND DISCUSSION

Results from all distance measures show

a trend for Wari administrative sites in po-

litically complex valleys to be further from

the center than one would expect on aver-

age. Likewise, Wari administrative sites

placed in politically simple valleys tend to

be closer to the center than the mean of our

20% sample. A rank order plot of our dis-

tance metric, whose computation is de-

scribed above, illustrates these trends (Fig.

6a). The cost distance from a Wari site to the

geographic center of a valley further sup-

ports our predictions (Fig. 6b). Application

of Tobler’s hiking algorithm shows that ad-

ministrative sites in politically complex val-

leys are a greater walking time away from

the center than all but one of the sites found

W

in valleys exhibiting simple Early Interme-

diate Period political complexity.

Differences between Wari sites found in

politically simple and complex valleys is

significant in three measures of distance: (1)

raw untransformed Euclidean distance is

significant at t 5 5.77, p . 0.0001, while (2)

our distance (D) metric is also significant

(t 5 23.21, p . 0.0048.), and (3) Tobler’s

cost–distance function is significant at t 55.35, p . 0.0001. All three measures indicate

that there is a significant separability in the

distance from center between politically

complex and simple valleys.

These results are promising, but we

would expect valley size to exert influence

on measures of distance. In this population,

four of five politically complex valleys have

larger areas than simple ones (Fig. 5), and

we would expect area to correlate with

measures of distance. We attempted to deal

with the problem of valley size in the devel-

opment of our distance (D) metric dis-

cussed above, and regression analysis

shows that valley area is not the sole source

of variability (Table 2). Holding area as the


FIG. 5. Rank order plot of valley area. The units

complex pre-Wari political organization are represen

pre-Wari political organization are shown with diago

Our distance metric (D) factors area out

(Fig. 4), and as we would expect regression

demonstrates that valley area accounts for a

very small fraction of variability when

using this distance statistic (R2 5 0.06). Area

appears to account for a good portion of To-

bler’s cost–distance formula (R2 5 0.59),

but the measure has the significant advan-

tage of being in easily interpretable units.

Area and complexity are directionally cor-

related as are distance from center and

complexity. Simple valleys are small and

sites in complex valleys should be further

from the center. Our distance (D) metric,

which standardizes valley area yet discrim-

inates cases well, shows that the greater

distances from center seen in large and

complex valley systems have more to do

with placement of administrative centers

within the valley rather than by the effects

of a valley’s size alone.

Four of the 20 sites do not follow trends

that we would expect. Two sites in complex

valleys are closer to the center than an

isotropically measured 20% random sam-

ple, and two sites located in politically sim-


independent measure it accounts for about

half of the variability of raw untransformed

Euclidean distance from center (R2 5 0.52).

ple valley systems are further from the cen-

ter than the 20% sample. Sonay, located in

the coastal valley of Camana, and Jin-

for the y axis are square kilometers. Valleys with

ted by horizontal lines while valleys with simple

nal hatching.

RMS error 3.32

camocco, located in the highland valley of

Carhuarazo, are both further from the cen-

ter than the mean for a 20% random sam-

ple. However, both are about a days’ walk

from the center. Pariamarca and Honco-

pampa are closer to the center than the

mean for a 20% random sample for the

Callejon de Huaylas, where they are lo-

cated. However, they appear to have travel

costs to the center that are similar to other

Administrative sites located in complex

valley systems. It is interesting to note that

the relative positions of Honcopampa and

Pariamarca do not necessarily run counter

to the proposed model. Both sites are lo-

cated along the boundary of the long nar-

row elliptically shaped Callejon de Huaylas

valley. These sites are located in a position

where travel time from the center, up the

steep flanks of the valley, is high while Eu-

clidean distance from center is low. It is sig-

nificant to note that both Honcopampa and

Pariamarca are located on the eastern flanks

of the valley, closer to the Wari core. More

extensive research in these two regions is

needed to determine what other local con-

ditions impacted site placement.

Despite the four exceptions described

above, results strongly matched our predic-

tions. The technique allowed us to test pre-

dictions in 12 local systems that vary

greatly in size over a vast horizontal extent.

Valley area ranges from less than 200 km2 to

over 2500 km2, and the northern most val-

ley is about 1380 km away from the south-

ern most valley. Early Intermediate Political

complexity tends to have occurred more

often in larger valley systems. t- Tests

showed significant differences in Wari Ad-

ministrative site placement depending on

whether they were located in valleys ex-

hibiting simple or complex Early Interme-

diate Period social complexity. Area and

complexity are correlated, and so it is diffi-

cult to tell from the distance measures alone

whether Early Intermediate Period com-

plexity or valley area is a better predictor for

Wari Administrative site placement. Regres-

sion shows that our distance metric was not

correlated with valley area. The combination

of significant differences in Wari site place-

ment between simple and complex valleys

and the lack of correlation between valley

area and our distance metric indicates that

Early Intermediate Period political complex-

ity is the likely source of variation in Wari

site placement with respect to the centrality

of site location in a valley system.

From a methodological and analytical

perspective these GIS based techniques of

spatial analysis have the advantage that

they permit additional analysis of and com-

parisons between the survey results from

numerous independent research efforts.

This permits examination and modeling of

systems that are much larger than any one

person could collect the data to test. These

ND CRAIG
494 JENNINGS A
TABLE 2Results of Regression Analysis Holding Area as the

Independent Variable

Linear Fit Euclidean distance by

valley area

Euclidean distance 5 Valley area

Summary of fit

R2 0.52R2 adj. 0.49

RMS error 12822.51

(D) Statistic by valley area(D) statistic 5 Valley area

Summary of fit

R2 0.06R2 adj. 0.008

RMS error 431.15

Linear fit Cost distance by valley areaCost distance 5 Valley area

Summary of fit

R2 0.59R2 adj. 0.56

same techniques should be applicable to

other archaeological contexts.

CONCLUSIONS

Imperial conquest and consolidation is a

complex process that is strongly shaped by

5
FIG. 6. (a) Isotropic distance measure in a rank order plot where the y axis is produced by applying

the distance metric described in the analytical methods section. In this figure sites that are located in

complex valleys have horizontal hatching while sites located in simple valleys have diagonal hatch

marks. (b) Nonisotropic costdistances in a rank order plot. The y axis is in units of hours walking time

established using Tobler’s (1993) hiking function with the GTOPO30 data set. Sites located in complex

valleys have horizontal hatching while sites located in simple valleys have diagonal hatch marks.

this article has demonstrated the value of

A

the local conditions that are encountered

during expansion. Scholars studying partic-

ular sites and areas can only hope to eluci-

date the impact of these local conditions on

their study area. The overall importance of

these conditions in imperial strategies,

however, must be left to conjecture. We sug-

gest that politywide analysis can build on

these local analyses to measure the impact

varying local conditions throughout an em-

pire.

In this case, we considered the effect of

an area’s political organization on Wari’s

imperial strategies. We suggest that the em-

pire developed a wealth finance system an-

chored by a number of administrative cen-

ters. At the time of Wari expansion, a

number of complex societies flourished in

the northern Highlands and coastal Peru.

The long distance exchange of prestige

goods appears to have formed a fundamen-

tal aspect of the political economies of these

polities (Shady Solís 1978:27, Topic and

Topic 1983; Arguto Calvo 1984:84–87). In-

stead of destroying these exchange relation-

ships, the empire co-opted them. Wari ad-

ministrative centers were placed at the

margins of these polities to control their ex-

change networks. The end result was an in-

crease in both the scale of trade and re-

gional economic prosperity that may have

benefited local groups as well (Shady Solís

1982, 1988). Outside of these areas of high

political complexity, the empire encoun-

tered areas settled by independent villagers

and, more rarely, pastoralists. In these

cases, Wari control focused on the develop-

ment and then control of these areas and

their local wealth potential (de la Vera Cruz

1996:150–151). Administrative centers for

these areas tended to be located toward the

center of the population in order to orga-

nize these populations cheaply and effec-

tively.

496 JENNINGS

Over the past 2 decades, research into an-

cient empires has underscored the incredi-

ble variety of imperial practices. Under-

neath and running though these practices,

we sincerely hope, lie patterns. Tantalizing

parallels to the processes described in this

model may exist, for example, in Egyptian

imperial practices in Nubia (Smith

1998:80–81); Assyrian strategies on their

northern frontiers (Parker 1998:379–381);

and the mixed economy of trade and trib-

ute of the Turks, Tibetans, and Uighurs of

Central Asia (Di Cosmo 1999:30–32). With-

out systematic spatial analysis at a polity-

wide level, however, the applicability of

this model to the understanding of other

empires can only be tentative. We hope that

ND CRAIG

such a level of analysis in the study of an-

cient empires and other macroregional

processes.

ACKNOWLEDGMENTS

We thank David Browman, Michael Malpass, Gor-

don McEwan, William Isbell, J. Scott Raymond, Katha-

rina Schreiber, John Topic, Teresa Topic, and Ryan

Williams for their assistance in the identification and

location of many of the Wari sites discussed in the text.

We also thank Keith Clarke for his feedback on the de-

velopment of our distance metric. This article bene-

fited enormously from the editorial advice of Mark

Aldenderfer, Karen Anderson, Christina Conlee, John

O’Shea, Stuart Smith, Kevin Vaughn, and four anony-

mous reviewers. We thank Katharina Schreiber for al-

lowing us to revise a figure from the site of Jincamocco.

NOTES

1 Evidence for storage capacity above the basic sub-

sistence needs of a peripheral center was long thought

to come from the enclosures of Viracochapampa and

Pikillacta. In these centers, there are massive blocks of

small rooms with raised doorways that have been in-

terpreted as storage facilities (Menzel 1964). The prin-

cipal investigators of these sites, however, now sug-

gest that these facilities never operated as such in any

significant way. Viracochapampa was abandoned be-

fore it was finished and put to use (Topic 1991:151),

while only a portion of Pikillacta’s room blocks were

in use during the site’s occupation (McEwan 1991,

1996). Even the rooms put into use at Pikillacta were

used for a variety of functions and only occasionally, if

at all, used for storage (McEwan 1996:183). While it is

possible that storage facilities of the empire remain

unidentified in the record, we suggest that the massive

storage and control of perishable, nonprecious goods

may not have served an important aspect in the Wari

political economy.2 In this article, administrative centers are differenti-

ated from offering deposits and temple centers. A com-

mon practice in Wari studies (Isbell and Schreiber

1977, Schreiber 1992, Cook 1985), this distinction is

based on the different functions of the three site types.

The offering deposits, consisting of intentionally

smashed decorative wares, are thought to have had a

ritual function. Temple centers, such as Pachacamac

and Wari Willka, most likely functioned as pilgrimage

centers and did not, strictly speaking, have an admin-

istrative use.

All sites included in the analysis are treated

equally regardless of variations in site size. In 1978,

William Isbell and Katharina Schreiber suggested

that the Wari empire was arranged into a three level

administrative hierarchy with the site Wari the first

order site, Viracochapampa and Pikkillacta as second

order centers, and the other enclosures as third order

sites (379). Recent work however suggests that Vira-

cochapampa was still under construction when it was

abandoned and contains only limited occupation de-

bris (Topic 1991:152). Pikillacta was also under con-

struction at the time of abandonment and only a

small portion of the site was ever occupied (McEwan

1996:181). We believe that it is likely that these cen-

ters never rose to become true second-order sites in

the Wari administrative hierarchy.3We struggled for several months in an attempt to

find more objective means to determine valley bound-

aries. A more objective definition does not seem to be

possible for several reasons. First, the vertical extent of

vegetation zones is not consistent through out the study

area. In general, the northern portion of the study area is

wetter while the southern extent tents to be more arid.

Furthermore, valleys included in analysis are found on

the coast, the highlands, and the eastern flanks of the

Andes. Each radically different environmental zones.

Therefore, horizontal variation in the nature of vertical

ecology would not permit uniform objective valley defi-

nition based on a consistent upper elevation threshold.

Second, another common approach is to delineate

drainage basins through down slope accumulation in

order to define valleys (Johnston 1998:91). Drainage

basin delineation has been explored to some extent with

this earth model. However, given the entities we want to

examine, downslope accumulation modeling is not ap-

propriate. Application of the technique produces polyg-

onal objects that define the limits of basin catchments

rather than valley extents (Johnston 1998:95). For exam-


ple, when the method was applied to our data set catch-

ment basins often included large sections of puna high-

lands. However, we wish to define valleys based on the

limits of arable agricultural land and habitable space.

REFERENCES CITED

Aldenderfer, Mark

1998 Montane Foragers: Asana and the south-centralAndean Archaic. Univ. of Iowa Press, Iowa City.

Agurto Calvo, Santiago

1984 Lima Prehispánica. Municipalidad de Lima,

Lima.

Baines, John, and Norman Yofee

1998 Order, Legitimacy, and Wealth in Ancient

Egypt and Mesopotamia. In Archaic states,

edited by Gary Feinman and Joyce Marcus,

pp. 199–260. School of American Research

Press, Sante Fe, NM.

Bennett, Wendell

1944 The north highlands of Peru: Excavations in theCallejon de Huaylas and at Chavin de Huantar,

Vol. 39, Part 1. Anthropological papers of the

American Museum of Natural History, New

York.

Blanton, Richard

1998 Beyond centralization: Steps towards a theory

of egalitarian behavior in archaic states. In Ar-chaic states, edited by Gary Feinman and Joyce

Marcus, pp. 135–172. School of American Re-

search Press, Sante Fe, NM.

Bradley, James Parker

1998 The mechanics of empire: The northern frontier ofAssyria as a case study in imperial dynamics. Un-

published Ph.D. dissertation, University of

California, Los Angeles.

Brooks, Sarah Osgood

1998 Prehistoric agricultural terraces in the Rio JapoBasin, Colca Valley, Peru. Unpublished Ph.D.

dissertation, University of Wisconsin.

Browman, David

1999 Wari impact on the Upper Mantaro Basin. Paper

presented at the 64th annual meeting of the So-

ciety for American Archaeology.

Bruhns, Karen

1994 Ancient South America. Cambridge Univ. Press,

New York.

Brumfiel, Elizabeth

1994 The Economic anthropology of the state: An

introduction. In The economic anthropology of thestate, edited by Elizabeth Brumfiel, pp. 1–12.

Univ. Press of America, Lanham.

Brumfiel, Elizabeth, and Timothy Earle

1987 Specialization, exchange, and complex soci-


eties: An introduction. In Specialization, ex-change, and complex societies, edited by Eliza-

beth Brumfiel and Timothy Earle, pp. 1–9.

Cambridge Univ. Press, Cambridge, UK.

Burger, Richard, and Frank Asaro

1977 Trace element analysis of obsidian artifacts fromthe Andes. Department of Anthropology,

Berkeley.

Buse, Herman

1965 Introducción al Perú. Lima, Perú.

Cabera Romero, Martha

1996 Unidades habitacionales, iconografía, y rituales enun poblado rural de la época Huari. Unpublished

thesis presented for the license of archaeology

to the Univerdidad Nacional de San Critobal

de Huamanga, Ayacucho.

Cardona Rosas, Augusto

1993 Características geogáficas del Patrón de asem-tamiento para el valle de Chuquibamba-Arequipadurante el Período Comprendido entre el HorizanteMedio, y el Horizonte Tardio. Unpublished thesis

presented for the license of archeology. Uni-

versdid Católica Santa Maria, Arequipa.

Chávez Chávez, José

1982 Evidencias arqueológicas en las cuencas de los RiosCotahuasi y Ocoña. Unpublished thesis pre-

sented for the license of archaeology. Universi-

dad Católica Santa Maria, Arequipa.

Chávez Chávez, José, and Ruth Salas Hinijoza

1990 Catastro arqueológico de la cuenca del

río Ocona. Gaceta Arqueológica AndinaV(18/19):15–20.

Claessen, Henri, and Peter Skalnik

1978 The early state. Mouton, The Hague.

Clarke, Keith

1995 Analytical and computer cartography. Prentice

Hall, Englewood Cliffs, NJ.

Cook, Anita

1985 The politico-religious implications of the

Hauri offering traditions. Dialogo Andino4:203–222.

1992 The stone ancestors: Idioms of imperial attire

and rank among Huari figurines. Latin Ameri-can Antiquity 3(4):341–364.

Cobb, Charles

1996 Specialization, exchange, and power in small-

scale societies and chiefdoms. Research in Eco-nomic Anthropology 17:251–294.

Crumley, Carol, and Marquardt, William

1990 Landscape: A unifying concept in regional

analysis. In Interpreting space: GIS and Archaeol-ogy, edited by Kathleen Allen, Stanton Green,

and Ezra Zubrow, pp. 73–79. Taylor & Francis,

London.

498 JENNINGS A

D’Altroy, Terence N.

1992 Provincial power in the Inka empire. Smithsonian

Institution Press, Washington, DC.

D’Altroy, Terence, and Timothy Earle

1985 Staple finance, wealth finance, and storage in

the Inka political economy Current Anthropol-ogy 26(2):187–206.

De la Vera Cruz, Pablo

1987 Cambio en los patrones de asentamiento y el

uso y abandono de los andenes en Cabana-

conde, Valle del Colca, Perú. In Pre-Hispanicagricultural fields in the Andean region, edited

by William Denevan, Kent Mathewson, and

Gregory Knapp. BAR International Series

398(1), BAR Publications, Oxford, UK.

1988 Estudio Arqueológico en el Valle de Cabanaconde,Arequipa. Unpublished thesis for the license in

archaeology, Universidad Católica Santa

Maria, Arequipa.

1996 El papel de la subregión norte de los valles oc-

cidentales en la articulación entre los Andes

centrales y los Andes centro sur. In La In-egración Surandina: Cinco siglos después, com-

plied by Xavier Albó et al, pp. 135–156. Estu-

dios y Debates Regionales Andinos 91, Centro

de Estudios Regionales Andinos ‘Bartolomé

de Las Casas’, Lima.

DeMers, Michael

1997 Fundamentals of geographic information systems.

Wiley, New York.

Di Cosmo, Nicola

1999 State formation and periodization in Inner

Asia History. Journal of World History10(1):1–40.

Drennan, Robert

1996 Statistics for archaeologists: A commonsense ap-proach. Plenum, New York.

Earle, Timothy

1997 How chiefs come to power: The political economyof prehistory. Stanford Univ. Press, Stanford,

CA.

Earle, Timothy, and Terrence D’Altroy

1989 Political economy of the Inka empire: The ar-

chaeology of power and finance. In Archaeolog-ical thought in America, edited by C. C. Lam-

berg-Karlovsky, pp. 183–204. Cambridge Univ.

Press, Cambridge, UK.

Goldstein, Paul

1990 La cultura Tiwanaku y las relaciones de sus

fases cerámicas en Moquequa. In Trabajos Ar-queológicos en Moquequa, compiled by Luis

Wantanabe, Michael Moseley, and Fernando

Cabieses, Vol. 2, pp. 31–58. Southern Peru

Copper Company, Peru.

ND CRAIG

Gonzales Carbajal, Gabriela

1998 Lima Prehispánica. Universidad Nacional

Mayor de San Marcos, Lima.

tion, Washington, DC.

Gonzalez Carre, Enrique et al.

1981 La Antigua Ciudad de Wari en Ayacucho. Bo-letin de Lima 16/17/18:83–97.

1996 El Templo Mayor en la Ciudad de Wari: Estudiosarqueológicos en Vegachayoq Moqo-Ayacucho.

Facultad de Ciencias Sociales, Universidad

Nacional de San Cristobal de Huamanga.

Glowacki, Mary

1996 The Wari occupation of the southern highlands ofPeru: A ceramic perspective from the site of Pikil-lacta. Unpublished Ph.D. dissertation, Bran-

deis University.

Glowacki, Mary, and Julinho Zapata

1998 The Wari occupation of Cuzco: Recent discoveriesfrom the Huaro Valley. Paper presented at the

38th annual meeting of the Institute of Andean

Studies, Berkeley, California.

Grieder, Terrence

1978 The art and archaeology of Pashash. Univ. of

Texas Press, Austin.

Hassig, Ross

1985 Trade, tribute, and transportation: The sixteenth-century political economy of the Valley of Mexico.

Univ. of Oklahoma Press, Norman.

Hirth, Kenneth

1978 Interregional trade and the formation of pre-

historic gateway communities. American An-tiquity 43(1):35–45.

Isbell, William H.

1977 The rural foundations of urbanism. Illinois Stud-

ies in Anthropology 10, University of Illinois,

Chicago.

1985 El Origen del Estado en el Valle de Ayacucho.

Revista Andina 3(1):57–106.

1988 City and state in Middle Horizon Wari. In Pe-ruvian prehistory, edited by Richard W. Keat-

ings, pp. 164–189. Cambridge Univ. Press,

Cambridge, UK.

1989 Honcopampa: Was it a Huari administrative

center? In The nature of Wari, edited by R. M.

Czwarno, F. M. Meddens, and A. Morgan, pp.

98–114. BAR International Series 525, Oxford,

UK.

1991a Honcopampa: Monumental ruins in Peru’s

North Highlands. Expedition: The UniversityMuseum Magazine of Archaeology and Anthro-pology 33(3):27–35.

1991b Huari Administration and the Orthogonal

Cellular Architecture Horizon. In Huari ad-ministrative structures, edited by William Is-

bell and Gordon McEwan, pp. 293–316.


Dumbarton Oaks, Washington, DC.

1997a Reconstructing Huari: A cultural chronology

for the Capital City. In Emergence and change

in early urban societies, edited by Linda Man-

zanilla. Plenum, New York, pp. 181–227.

1997b Mummies and mortuary monuments: A post-processual prehistroy of Central Andean social or-ganization. Univ. of Texas Press, Austin.

Isbell, William H., and Katharina J. Schreiber

1978 Was Huari a state? American Antiquity43(3):372–389.

Isla, Elizabeth, and Daniel Guerrero

1987 Socos: Un sitio Wari en el valle de Chillón. Gac-eta Arqueológica Andina 14(4):23–28.

Johnson, Allen, and Timothy Earle

1987 The evolution of human societies: From foraginggroup to agrarian state. Stanford Univ. Press,

Stanford, CA.

Johnston, Carol

1998 Geographic information systems in ecology. Black-

well Science, Singapore.

Julien, Daniel

1988 Ancient Cuismancu: Settlement and cultural dy-namics in the Cajamarca region of the North High-lands of Peru, 200 B.C.–A.D. 1532. Unpublished

Ph.D. dissertation, Univ. of Texas at Austin.

Kantner, John

1997 Ancient roads, modern mapping: Evaluating

Chaco Anasazi roadways using GIS technol-

ogy. Expedition 39(3):49–62.

Kvamme, Kenneth

1990 GIS algorithims and their effects on regional

archaeological analysis. In Interpreting space:GIS and archaeology, edited by Kathleen Allen,

Stanton Green, and Ezra Zubrow, pp. 112–126.

Taylor & Francis, London.

Lanning, Edward P.

1965 Highland South America. American Antiquity31(1):139–140.

Larco Hoyle, Rafael

1966 Peru, translated from the French by James

Hogarth. World, New York.

LeVine, Terry (Ed.)

1992 Inka storage systems. Univ. of Oklahoma Press,

Norman.

Lechtman, Heather

1980 The central Andes: Metallurgy without iron. In

The coming of the age of iron, edited by Theodore

A. Wertime and James D. Muhly, pp. 267–334.

Yale Univ. Press, New Haven, CT.

Lumbreras, Luis

1974a The peoples and cultures of South America, trans-

lated by Meggers, Betty. Smithsonian Institu-


1974b Las Fundaciones de Huamanga. Club Hua-

manga, Lima.

A

1981 Wari. In Historia Del Perú. Editorial Juan Mejia

Baca, Vol. 2, pp. 11–91.

Luttwak, Edward

1976 The grand strategy of the Roman Empire from thefirst century A.D. to the third. John Hopkins

Univ. Press, Baltimore, MD.

MacNeish, Richard, Thomas Patterson, and David

Browman

1975 The central Peruvian prehistoric interactionsphere. Robert S. Peabody Foundation for Ar-

chaeology Papers, Andover.

Manrique, Julio, and Manuel Cornejo

1990 Visión sobre la arqueología del valle

de Camana. Gaceta Arqueológica AndinaV(18–19):21–24.

Malpass, Michael, Anthony Puglisi, and Jason Ker-

schner

1997 Wari on the Far South Coast of Peru: The site ofSonay in the Camana Valley. Paper presented at

the 62nd Annual Meeting of the Society for

American Archaeology.

Mayr, Ernst

1982 The growth of biological thought. The Belknap

Press of Harvard Univ. Press, London.

McCown, Theodore

1945 Pre-Incaic Huamachuco: Survey and excava-

tions in the region of Huamachuco and

Cajabamba. California University Publicationsin American Archaeology and Ethnology39:223–400.

McEwan, Gordon

1984 Investigaciones en la cuenca del Lucre, Cuzco.

Gaceta Arqueológica Andina 3(9):12–14.

1987 The Middle Horizon in the Valley of Cuzco, Peru:The impact of the Wari occupation of Pikillacta inthe Lucre Basin. Bar International Series 372,

BAR, Oxford, UK.

1989 The Wari empire in the southern Peruvian

highlands: A view from the provinces. In TheNature of Wari, edited by R. M. Czwarno, F. M.

Meddens, and A. Morgan, pp. 53–71. BAR In-

ternational Series 525, Oxford, UK.

1991 Investigations at the Pikillacta site: A provin-

cial Huari center in the valley of Cuzco. In

Huari administrative structures, edited by

William Isbell and Gordon McEwan, pp.

93–120. Dumbarton Oaks, Washington, DC.

1996 Archaeological investigations at Pikillacta, a

Wari site in Peru. Journal of Field Archaeology

500 JENNINGS

(23)2:169–186.

1998 The function of niched halls in Wari architec-

ture. Latin American Antiquity 9(1):68–86.

Menzel, Dorothy

1964 Style and time in the Middle Horizon. NawpaPacha 2:1–106.

Moseley, Michael, Robert Feldman, Paul Goldstein,

and Luis Watanabe

1991 Colonies and conquest: Tihuanaco and Huari

in Moquequa. In Huari administrative struc-tures, edited by William Isbell and Gordon

McEwan. Dumbarton Oaks, Washington, DC.

Bradley, James Parker

1998 The mechanics of empire: The northern frontier ofAssyria as a case study in imperial dynamics. Un-

published Ph.D. dissertation, University of

California, Los Angeles.

Pérez Calderón, Ismael

1995 Excavaciones de salvataje y propuesta sobre conser-vación de estructuras descubiertas, area museo desitio en Wari. Convenio Instituto Nacional de

Cultura—Universidad Nacional San Cristobal

de Huamanga.

Pozorski, Shiela, and Thomas Pozorski

1987 Early settlement and subsistence in the CasmaValley, Peru. Univ. of Iowa Press, Iowa City.

Ravines, Rogger

1985 Cajamarca Prehispánica: Inventario de Monumen-tos Arqueológicos. Instituto Nacional de Cultura

de Cajamarca.

Raymond, J. Scott

1992 Highland colonization of the Peruvian mon-

tana in relation to the political economy of the

Huari empire Journal of the Steward Anthropo-logical Society 20(1-2):17–36.

Raymond, J. Scott, and William H. Isbell

1969 Cultural remains in the Pampas River Valley,Peru. Paper presented at the 34th Annual

Meeting of the Society for American Archae-

ology.

Rice, Prudence, and Luis Wantanabe

1990 La arqueología de Moquequa: Un resumen. In

Trabajos Arqueologicos en Moquequa, compiled

by Luis Wantanabe, Michael Moseley, and Fer-

nando Cabieses, Vol. 1, pp. 23–35. Southern

Peru Copper Company, Peru.

Rowe, John

1963 Urban settlement in ancient Peru. Nawpa Pacha1:1–27.

Schreiber, Katharina

1978 Planned architecture of Middle Horizon Peru:Implications for social and political organization.

ND CRAIG

Unpublished Ph.D. dissertation, SUNY-Bing-

hamton.

1987 From State to Empire: The expansion of Wari

outside of the Ayacucho Basin. In Origins and

development of the Andean State, edited by J.

Haas, S. Pozorski, and T. Pozorski, pp. 91–96.

Cambridge University Press, New York.

1992 Wari imperialism in Middle Horizon Peru. An-

thropological Papers 87, Univ. of Michigan,

Ann Arbor.

1999 Regional approaches to the study of prehistoric

empires: examples from Ayacucho and Nasca,

Peru. In Settlement pattern studies in the Ameri-cas: Fifty years since Viru, edited by Brian Bill-

man and Gary Feinman, pp. 160–171. Smith-

sonian Institution Press, Washington, DC.

Schwartz, Glenn, and Steven Falconer

1994 Rural approaches to social complexity. In Ar-chaeological views from the countryside, edited by

Glenn Schwartz and Steven Falconer, pp. 1–9.

Smithsonian Institution Press, Washington, DC.

Sciscento, Margaret

1989 Inka and Wari mastery of Chuquibamba. Unpub-

lished Ph.D. dissertation, Univ. of California at

Santa Barbara.

Shady Solís, Ruth

1978 Enterramientos en chullpas de Chota. Museo Na-

cional de Anthropología y Arqueología, Inves-

tigaciones de Campo 1.

1982 Cultura Nieveria y la interacción social en el

mundo andino en la época Huari. Arqueologi-cas 1:5–108.

1988 Epoca Huari como interacción de las so-

ciedades regionales. Revista Andina 6(1):67–99.

1989 Cambios significativos ocurridos en el mundo

Andino durante al Horizante Medio. In The na-ture of Wari: A reappraisal of the Middle HorizonPeriod in Peru, edited by R.M. Czwarno, F.M.

Meddens, and A. Morgan, pp. 3–15. BAR In-

ternational Series 525, Oxford, UK.

Sinopoli, Carla

1994a Archaeology of empires. Annual Review of An-thropology 23:159–180.

1994b Political choices and economic strategies in

the Vijayanagara empire. In The economic an-thropology of the state edited by Elizabeth M.

Brumfiel. Univ. Press of America, Lanham.

Smith, Carol

1976 Regional economic systems: Linking geo-

graphical models and socioeconomic prob-

lems. In Regional analysis, edited by Carol

Smith, Vol .1, pp. 3–63. Academic Press, New

York.

Smith, Stuart


1998 State and empire in the Middle and New King-

doms. In Monographs in Mediterranean archaeol-ogy 8, edited by A. Bernard Knapp, pp. 66–89.

Sheffield Academic Press, Sheffield, UK.

Spickard, Linda

1983 The development of the Huari administrative

system. In Investigations of the Andean past,edited by Daniel Sandweiss, pp. 136–160.

Cornell Latin American Studies Program,

Cornell.

Stanish, Charles

1992 Ancient Andean political economy. University of

Texas Press, Austin.

Thatcher, John

1975 Early Intermediate Period and Middle Hori-

zon 1B ceramic assemblages from Huamacu-

cho, North Highlands. Nawpa Pacha 17:91–106.

Tobler, Waldo

1993 Nonisopic modeling. In Three presentations ongeographical analysis and modeling. National

Center for Geographic Information and Analy-

sis. Technical Report 93-1, University of Cali-

fornia, Santa Barbara.

Topic, John

1991 Huari and Huamachucho. In Huari administra-tive structures, edited by William Isbell and

Gordon McEwan, pp. 141–164. Dumbarton

Oaks, Washington, DC.

Topic, John, and Theresa Topic

1983 Coast-highland relations in Northern Peru:

some observations on routes, networks, and

scales of interaction. In Prehistoric settlementpatterns: Essays in honor of Gordon Willey,

edited by Evon Z. Vogt and Richard M. Lev-

enthal, pp. 237–259. Presidents and Fellows

of Harvard College, Cambridge, MA.

Topic, Theresa, and John Topic

1984 Huamacucho archaeological project: Preliminaryreport on the third season, June–August 1983.

Trent University occasional papers in anthro-

pology 1, Trent Univ. Peterbourough.

1987 Huamacucho archaeological project: Preliminaryreport on the 1986 field season. Trent University

occasional papers in anthropology 4, Trent

Univ. Peterbourough.

Trawick, Paul

1994 The struggle for water in the Andes: A study oftechnological change and social decline in the Co-tahuasi Valley of Peru. Unpublished Ph.D. dis-

sertation, Yale University.

Valkenier, Lisa


1995 New evidence for Chimu Capa and the Early

Horizon period in the Supe Valley, Peru. Jour-nal of the Steward Anthropological Society23(1/2):269–286.

Vivanco, Cirilo, and Lidio Valdez

1993 Poblados Wari en la cuenca del pampas—

1999 Wari imperialism on the southern periphery: Re-sults of the excavations at Cerro Baul. Paper pre-


Qaracha, Ayacucho. Gaceta Arqueológica AndinaVII(23):83–102.

Williams, Carlos, and José Pineda

1985 Desde Ayacucho hasta Cajamarca: Formas ar-

quitectónicas con filiación Wari, unidad del

espacio andino. Boletin de Lima 7(40):55–61.

Williams, Patrick Ryan

1997 The role of disaster in the development of agricul-ture and the evolution of social complexity in thesouth-central Andes. Unpublished Ph.D. disser-

tation, University of Florida.

sented at the 39th Annual Meeting of the Insti-

tute of Andean Studies, Berkeley.

Williams, Patrick Ryan, Michael Moseley, and Donna

Nash

2000 Empires of the Andes: A majestic frontier out-

post chose cooperation over war. DiscoveringArcaeology 2(2):68–73.

Zubrow, Ezra

1990 Contemplating space:Acommentary on theory.

In Interpreting space: GIS and archaeology, edited

by Kathleen Allen, Stanton Green, and Ezra

Zubrow, pp. 67–72. Taylor & Francis, London.

Documents

Politywide Analysis and Imperial Political Economy: The Relationship between Valley Political Complexity and Administrative Centers in the Wari Empire of the Central Andes