The Zapotec‐Imperialism Argument Insights from the Oaxaca Coast

5/11/2018 The Zapotec Imperialism Argument Insights from the Oaxaca Coast - slidepdf.com

http://slidepdf.com/reader/full/the-zapotecimperialism-argument-insights-from-the-oaxaca-co

The Zapotec‐Imperialism Argument: Insights from the Oaxaca Coast

Author(s): Robert N. Zeitlin and Arthur A. JoyceSource: Current Anthropology, Vol. 40, No. 3 (June 1999), pp. 383-391Published by: The University of Chicago Press on behalf of Wenner-Gren Foundation for AnthropologicaResearchStable URL: http://www.jstor.org/stable/10.1086/200029 .

Accessed: 04/02/2011 14:09

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and yo

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=ucpress. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range o

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new form

of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press and Wenner-Gren Foundation for Anthropological Research are collabora

with JSTOR to digitize, preserve and extend access to Current Anthropology.

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/action/showPublisher?publisherCode=wennergren


http://www.jstor.org/stable/10.1086/200029?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp



http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/stable/10.1086/200029?origin=JSTOR-pdf






Reports

Group Size, Memory, andInteraction Rate in theEvolution of Cooperation1

s . j . c o x , t . j . s l u c k i n , a n d j . s t e e l eDepartment of Electronics and Computer Science,University of Southampton, Highfield, Southampton,SO17 1BJ /Faculty of Mathematical Studies,University of Southampton, Highfield, SouthamptonSO17 1BJ/Department of Archaeology, University of Southampton, Highfield, Southampton SO17 1BJ,England. 13 vii 98

Human societies are characterised by high degrees of re-ciprocal altruism between unrelated individuals. It haseven been suggested that humans have evolved a cogni-tive capacity for effective reasoning about social ex-change transactions which does not readily generaliseto other, non-social reasoning tasks (Cosmides 1989).Explaining the emergence of this capacity for coopera-tion is one of the fundamental goals of evolutionary an-thropology. Some theorists have suggested that re-peated interactions in small, stable social groups couldhave led to such high levels of cooperation in the course

of human evolution (Trivers 1971, 1985). Criteria forstable cooperation in such scenarios include repeatedinteraction, recognition of the identity of individualparticipants, and memory of the outcomes of previousencounters. However, recent arguments that human so-cial evolution was characterised by a trend towards in-creased group sizes (Dunbar 1993, Aiello and Dunbar1993) have complicated matters, since it is generally be-lieved that large group sizes make cooperation unstable.Clearly, our ancestors cannot have been simultaneouslyundergoing selection for adaptations for cooperative ex-change in a small social group context (as argued byTrivers 1971, 1985) and selection for adaptations tomaintaining affiliative networks in the context of a very

large social group (as argued by Dunbar 1993). Thesetwo approaches to the evolution of human social behav-iour would seem to be mutually contradictory.

In this paper we will show, through computer simula-tion, that cooperation can evolve as a stable strategy inlarge social groups, subject to certain constraints. Our

1. © 1999 by The Wenner-Gren Foundation for Anthropological Re-search. All rights reserved 0011-3204/99/4003-0005$1.00. Wethank Alan Ingham and Denis Nicole for helpful discussion of thiswork and Rob Boyd, Robin Dunbar, and Brian Molyneaux for com-ments on drafts of the paper.

369

results therefore resolve an apparent contradiction inthe recent literature on human cognitive evolution. Ad-ditionally, we believe that they shed light on the condi-tions for the emergence of cooperation in any humansocial group in which there are many individuals andno hierarchical structures of authority.

There have been many attempts to understand theevolution of cooperation in groups (many of which arereviewed by Argyle 1991). Ethnographic studies by par-ticipant observers of cooperative exchanges have thevirtue of realism but are often hampered by a lack ofadequate longitudinal data (e.g., Dwyer and Minnegal1997). Experimental studies by social psychologists ofpeople’s behaviour in cooperative games have the virtue

of controlled experimental conditions, but the artificialsituations may be unrealistic and the research designmay place a constraint on experimental group sizes andon the number of bouts of play. Computer simulations,by contrast, enable us to study the behaviour of individ-uals in large groups over very long series of iterative in-teractions, although they are limited by the simplifyingassumptions which have to be made about players’ be-haviour when formulating decision rules.

In this paper we introduce a new class of simulationmodels for the study of cooperation in groups based onthe repeated prisoner’s dilemma game (Axelrod 1984).The prisoner’s dilemma game focuses attention on thedilemma posed by the contrast between the immediate

benefits of exploitation and the long-term benefits ofcooperation. Using large-scale computational resourceswe have been able to simulate the detailed microscopicbehaviour of identifiable individuals in the group whointeract and make decisions according to a simple setof rules. From this we examine the macroscopic behav-iour of the group. We show that cooperation can be sus-tained in large groups when there are sufficient interac-tions on each round and players are able to base futureplay on their observations of other players’ past actions.This study focuses on the manner in which (a) groupsize and (b) the number of interactions allowed betweenplayers affect the ability of a group to sustain global co-operative behaviour. Players compensate for the limited

number of interactions they make on each round byhaving a memory for past interactions, which they useto assess the likely behaviour of other players.

modelling cooperation

The prisoner’s dilemma game is widely used in thestudy of cooperation. It is used in the interpretation ofethnographic data (Dwyer and Minnegal 1997), in exper-imental research design by social psychologists (Argyle1991), and in computer simulations. It is a two-player



370 c u r r e n t a n t h r o p o l o g y

t a b l e 1Payoffs in the Prisoner’s Dilemma Game (Axelrod 1984)

i’s Action j’s Action i Receives j Receives

Cooperates Cooperates Reward 3 points Reward 3 points

Cooperates Defects Sucker

0 points Temptation

5 pointsDefects Cooperates Temptation 5 points Sucker 0 pointsDefects Defects Punishment 1 point Punishment 1 point

non-zero-sum game. Each player chooses whether to co-operate or defect; the payoff to each player dependsupon their joint actions (table 1). Whilst the reward ishigh if both players cooperate (R points), the temptationto defect is higher (T points). The spectre of only receiv-ing the sucker’s payoff (S points) may lead both playersto defect, for which they receive the punishment payoff(P points). The points are arranged such that T R

P S and 2R T S. We use T 5, R 3, P 1, and

S 0 throughout (Axelrod 1984).Conventionally, computer simulation studies have

focused on how different strategies might compete witheach other, without necessarily attaching strategies toidentifiable players (Axelrod 1984; Nowak and Sigmund1993a, b; Glance and Huberman 1993, 1994; Nowakand May 1992; Pollock 1989; Herz 1994). In an isolatedprisoner’s dilemma game, unconditional defection is anevolutionarily stable strategy (Axelrod 1984). In the re-peated game, when strategies are able to use past experi-ence of previous rounds, cooperation sometimes pre-vails. In a series of computer tournaments, Axelrod(1984) found that a winning strategy is the simple tit fortat, which repeats the co-player’s previous move. No-

wak and Sigmund (1993a, b) have studied populationsconsisting of probabilistic strategies which respond tothe payoff in the previous round. They report the suc-cess of Pavlov, which cooperates when both playersopted for the same choice in the previous round. Pavlovhas two crucial advantages over tit for tat: it can correctfor occasional mistakes between players and can exploitunconditional cooperators. In such studies, the strate-gies respond stochastically to the payoff in the previousround. On every round, all the strategies compete witheach other in a series of pairwise games.

We believe that modelling group behaviour requiresa more subtle approach. We should not consider thegroup as a mixture of proportions of strategies but

should attach strategies to individual players. The rea-son for this is that in large groups, in general, individu-als interact with only a proportion of the rest of thegroup at any one time. Players should be free to mix andneed not interact with all the other players on everyround. Consequently, players need to remember infor-mation about interactions that have occurred sometime in the past.

The class of models that we introduce is based on thepremise that collaboration in a group has three distinctparts: (1) pairs are formed, (2) the decision to cooperate

is made, and (3) the fruits of collaboration are divideup (Noe 1990). In order to model such behaviour, wneed to allow players first to make a choice about theco-players on a given round and secondly to decide hothey are to interact. Interaction between two playemust be by mutual consent. In our simulations, playehave a memory of others’ behaviour over the previoum rounds, which they use to estimate a co-playerlikely behaviour on the next round. We shall examin

the composition of the group when we change both thnumber of interactions allowed on each round and thgroup size. Our simulations demonstrate that a simpmemory for past encounters is sufficient to allow stabcooperation to evolve even where the number (or frquency) of interactions is relatively low.

the structure of the simulation model

Our simulations consist of a group of N players which players may choose up to v c partners on eacround (though they may choose to make fewer than thmaximum number of allowed interactions on any giveround). On each round, player i makes three asses

ments. First, an empirical estimate is made of the proability E ij that a co-player, j, will cooperate on the neround. Secondly, a value judgement is made whether interact on the basis of the magnitude of E ij. Thirdly,strategic judgement is made whether to cooperate, alson the basis of the magnitude of E ij.

Player types. For every potential interaction, eacplayer, i, compares E ij with a threshold for playing ancooperating, denoted by γ ( i) and φ( i) respectively (fig. 1The possible responses to these thresholds determinfour different player types caricatured by their responto their preferred partners. The Reciprocator, Policman, Sucker, and Thief respectively seek to maximiinteractions giving the Reward, Punishment, Sucker’

and Temptation payoffs. Interactions between playeoccur by mutual consent, and once both parties have dcided whether to cooperate or defect, the payoff to eacis determined according to the rules of the game (tab1). At the end of each round, any player receiving lethan 1/(3N ) of the average points per player is replaceby a new player with a randomly chosen strategy. This equivalent to the original player’s spontaneouschanging to a new strategy. Thus the simulation prceeds according to a very simple evolutionary rule.

Parameters of the model and its evolutionary mech



Volume 40, Number 3, June 1999 371

Fig. 1. For each possible pair the players decide whether to interact and whether to cooperate. Each player, i,will interact with another player, j, depending on a threshold response to i’s estimate of the probability that jwill cooperate (Eij). The decision to cooperate is made in a similar way. The response to the threshold for

playing, φ(i), and the threshold for cooperating, γ (i), depends on whether the player is a Reciprocator, a Thief, aSucker, or a Policeman.

nism. Each player is characterised by four parameters:m, γ , φ, and a player type. The memory length, m, is aparameter representing the capacity for extrapolating

from past to present behaviour; the threshold φ repre-sents a confidence threshold above (or below) which theplayer is prepared to interact. The threshold γ deter-mines whether expectation of a co-player’s cooperationinitiates cooperation or defection. The thresholds, φ( i)and γ ( i), are chosen independently for each player froma uniform random distribution in the interval [0,1].

Each player’s memory is chosen from a uniform ran-dom distribution [0,50]. The memory of players is notdriven by any selection mechanism in our simplemodel. We assume that all players see each other as po-tential partners on their first encounter. Thus E ij is ini-tialised to 1 for the Reciprocator and the Thief and to 0for the Policeman and the Sucker. The simulation starts

with N players of randomly chosen type, each with ran-domly chosen thresholds φ( i) and γ ( i) and memory m.When a player is replaced, a new player is generated inthe same way as the starting players for the game. Wedo not alter the distribution from which players’ mem-ory or the thresholds φ( i) and γ ( i) are chosen as the simu-lation progresses.

Players’ use of memory. Players use their memory tocompensate for the fact that they do not necessarily in-teract with every other player on each round. Player iwith memory m makes an empirical estimate of the

probability that a co-player, j, will cooperate on thenext round, E ij, on the basis of a weighted average ofthe results of the interactions with that player over the

previous (m 1) rounds. Players implement this usinga simple update rule for their current assessment, E ij.Numbering the rounds, k, from 0 to m, players weighteach round by a factor w k, and E ij is based on a weightedaverage of the outcomes of the previous (m 1) rounds.The update rule is:

E′ ij d

m

k0

w k

w E ij, (1)

where d is 0, 1, or 0.5 if the co-player respectively de-fected, cooperated, or did not interact on the previousround. The sum normalises the calculation so that the

assessed probabilities remain in the range 0 to 1. In oursimulations, players observe only their own interac-tions, though in principle we could include observationof other interactions in the group. We fix

w (m1 ) ε, (2)

where ε is a small parameter for the weighting of the(m 1)th round, which results in this round’s being ef-fectively forgotten. Throughout our simulations, ε 0.01. A pair of players considers no interaction on around as if the co-player cooperated (or defected) with

http://www.jstor.org/action/showImage?doi=10.1086/200029&iName=master.img-000.png&w=390&h=270




Fig. 2. Percentage of cooperation in group and its standard deviation over six runs as a function of group sizeN, and maximum allowed number of interactions, vc.

probability 1 / 2. This represents players’ being in a stateof ignorance about other players whose behaviour hasnot been directly observed on a given round (an alterna-tive would be for a player to set d to be the average ofthe current assessments for other players).

Players choose their partners simultaneously. Wehave implemented this by testing each of the possibledyadic interactions in a random order on each round.We ensure that each possible pairing is considered atmost once on each round and that players make up tov c interactions per round.

results

The results of the simulations demonstrate that thelevel of cooperation in a group varies with the sizeof the group, the maximum number of interactionswhich are permitted to a player on each round, andthe memory of players for the outcomes of past in-teractions. These results can be expressed in terms ofcharacteristics of the group at the end of a simulationrun. These characteristics include the percentage of co-operative interactions in the group, the average pay-off per player, the proportion of the group made up byeach player type, and the average memory length foreach player type. Our results were derived from aver-aging six independent samples at each {N, v c} pair over1 105 rounds. This involved over a thousand hours

of computation on an eight-node cluster of DEC AlphaWorkstations. Long simulations are important to en-sure that the system has reached equilibrium. Somefluctuations in the final state of the system occur be-cause of the stochastic nature of the simulations andbecause of a bifurcation in the underlying dynamicalmodel for the system.

Group cooperation. In figure 2 we show the percent-age of cooperation as a function of group size, N, andthe maximum number of interactions allowed perplayer per round, v c. We also show the standard devia-tion of the percentage of cooperation over the six inde-

pendent samples. For groups above size N 40 we maidentify three regions of interest:

1) v 10. In all of the simulation runs there is littcooperation. When players can make only a few interations per round, the games resemble a set of one-off prioner’s dilemma games, and the only rational strategy to defect. Our model reproduces this well-known phnomenon (Axelrod 1984).

2) 10 v 2N /3. In this region cooperation is stabin the majority of the simulation runs, though occasioally a group may become stuck in a defecting regimPlayers are able to use their memories to compensafor the fact that they do not interact with all of thother players on each round.

3) 2N /3 v N 1. When players are able to inteact with most of the other players on each round, defetors can evade effective detection and sanctions. Thfluctuations in the percentage of cooperation from ruto run are a result of a bifurcation in the model. Somtimes a simulation will yield a cooperating group; other times defection is rife. In figure 3 we demonstra

Fig. 3. Group dynamics for N 90 players makingup to vc 68 interactions per round as each of thesix independent simulations progresses.






Fig. 4. Average points payoff per player per interaction and its standard deviation over six runs as a functionof {N, vc}.

the bifurcation in the model by showing the run-timedynamics for a group of size N 90 making up to v 68 interactions per round. The results at round 1 105

persist for at least another 2 105 rounds. Two of thesimulation runs yield a stable cooperating group; the re-maining four are stuck in a defecting regime.

Figure 4 shows that the average payoff to each player,according to the rules of the game and its fluctuations,is directly proportional to the amount of cooperation inthe group. When most of the interactions between play-ers are cooperative, the players each receive the Rewardpayoff (3 points).

Large groups. Our simulations have focused on thedetailed microscopic behaviour of groups up to size N

100 with the number of interactions allowed per playervarying from 1 to (N 1). Repeated detailed simula-tions for groups of larger sizes is possible but very time-consuming. The clear trend in figure 2 shows a stableregime of cooperation in a group of size N when thenumber of interactions, v, is 10 v 2N /3. This trendis likely to be stable for larger groups up to at least N 150, and we have performed a very small number ofsimulation tests in this region which confirm this.

Composition of group. The percentage compositionof the group measures the relative probability of findingthe different player types at a later time in the simula-tion. Our simulations often resulted in a stable mix ofcooperators and defectors, with some role for each of

the player types. All too often results from simulationsof this kind become a discussion about which singlestrategy is the best, disregarding more complex interac-tions between ensembles of different players. For exam-ple, the Reciprocator is able to sustain cooperation in agroup but can be invaded by defecting strategies notonly when there are very few interactions on eachround but also when there are many interactions oneach round (in which case defectors can evade effectivesanctions).

Comparing figure 2 and figure 5, we see that the maincompetition is between the Reciprocators and the

Thieves. Both of these player types are willing to inter-act with players whom they identify as cooperators.The Reciprocators reward cooperation with coopera-tion. In contrast, a co-player’s cooperation initiates de-fection in the Thieves, who attempt to maximise thenumber of interactions in which they receive the Temp-tation payoff. When the group consists mainly of Recip-rocators, nearly all of the interactions are cooperative,and the payoff per player is high. However, in the re-gimes of defection, the Thieves are able to exploit theReciprocators and come to dominate the group. Theother strategies, Policeman and Sucker, who prefer tointeract with defectors, do not fare well and seldommake up more than about 5–15% of the group, since

they are too easily exploited.Players’ memory. Memory length (the number of

rounds that a player takes into account when assessinga co-player) is initially chosen for each player at randomusing a uniform distribution between 0 and 50 rounds.Since we do not change the distribution as the simula-tion progresses, the only selection pressure on this vari-able is that players who receive relatively few points arereplaced by new players with a different memory length(and other parameters). In figure 6 we examine the aver-age memory lengths for each of the different playertypes in the simulation, averaged over the whole runand over each of our independent samples. If there wereno selection pressure, then the average player memory

would remain at 25.5. However, there appears to besome selection pressure favouring an increased memoryin (1) Policeman, Sucker, and Thief strategies in the re-gime of stable cooperation and (2) Reciprocators at theonset of the group cooperation/defection bifurcation. Inboth of these regions the respective strategies face ad-versity. That the trend should be an increased memoryfollows from the fact that a longer memory allows amore accurate assessment of a co-player. For the Police-man, Sucker, and Thief strategies, this assessment en-ables them to interact with other players but exploitthem only when they are sure about those players’ be-





Fig. 5. Group composition: Percentage of each of the four player types averaged over whole run as a functionof {N, vc}.

haviour. Similarly, to resist the invasion of the de-fecting Thief strategy, the Reciprocators attempt to ver-ify the pedigrees of their partners to ensure that they areinteracting with other cooperators, and therefore theyneed to base their assessments on a larger number of in-teractions. In the other cases, a memory of averagelength is sufficient to compensate for the limited inter-actions made between players on each round.

discussion

Our results appear to resolve the contradiction identi-fied at the start of this paper. Cooperation can evolve asa dominant and stable strategy in very large socialgroups provided that certain conditions are met. Wehave shown that cooperation can be sustained in a largegroup with a limited number of interactions on eachround if players are able to base decisions about futureplay on the results of a number of rounds of past playand are permitted to refuse to play with others whennecessary. In evolutionary terms the rewards for this co-

operation would give a significant advantage to indiviuals in the group, who would, however, need to stomore information to coordinate relationships wiother players and would require greater processinpower to do this effectively.

These results are consistent with work by Dunb(1992, 1993, 1996), who explains primate brain evoltion as a response to changes in social group sizes, givethe need to maintain productive social relationshi

with a constant fraction of the other group memberDunbar’s model assumes stable, densely interconnectesocial groups. Our simulations indicate that it is, fact, possible to sustain cooperation in relatively larggroups and that this depends on two factors. These arfirst, that sufficient interactions should occur betweeindividuals to allow a cooperative relationship to be stble and, secondly, that individuals should assess otherintentions using a memory function to compensate fthe limited number of interactions possible on angiven round.

We therefore propose—on the basis of our findings





Fig. 6. Average memory of the different player types as a function of {N, vc}.

the simulations—that when social exchanges are domi-nated by large numbers of pairwise interactions whichare initiated by mutual consent, the level of cooperationis constrained by an interaction of group size, interac-tion frequency, and efficient use of memory of past out-comes. If, empirically, we find that cooperation regu-larly breaks down in human groups above a certainsize, then we must assume that this reflects an innatemental-capacity constraint on the numbers of partnerswhose histories of interactions can be efficiently pro-

cessed by an individual to predict future outcomes. Itwould be reasonable to expect that in groups whosesizes risk evoking such mental-capacity constraints,there will be a tendency to find social exchanges pat-terned on those versions of the simulation model inwhich we found a stable regime of cooperation but inwhich the values of v c and m (or their product) were alsorelatively low.

In contrast with our work, Nettles and Dunbar (1997)have considered the potential effects of social markingon the evolution of cooperation in dispersed social sys-tems with interaction probabilities constrained by spa-

tial proximity in one dimension and with very low val-ues of v c/N. They did not explore the effects of varyinggroup size and memory capacity on cooperation in con-ditions where many interactions are possible on eachround and where participants can select their interac-tion partners.

For our findings to have anthropological relevance,we must demonstrate that the parameters of the modelare analogous to key parameters determining the out-comes of human social exchanges. Despite the formal

simplicity of our model, we believe that it is reasonableto expect some of the constraints on cooperation seen inour simulations to find their analogue in human socialinteractions. These include the necessity of repeated in-teractions with a sufficient fraction of the group on eachround and the requirement that each individual have amemory for previous outcomes which discounts thosefrom the more distant past. We should also expect tofind some analogue for the model’s player types; thatis, we should expect human groups to be composed ofindividuals who have preferred, or dominant, strategicresponses to interactions when these have a reward





structure similar to that of the prisoner’s dilemmagame.

Of course, it is hard to identify the manifestation ofa ‘‘round’’ in an anthropological context. But we predictthat in cooperative human groups people will find—ifnecessary, by contrivance—opportunities for such re-peated sampling of one another’s intentions. Such op-

portunities could include food sharing, labour ex-changes, and conversational exchanges of adaptiveinformation. We further speculate that the prototypicalexamples of such pairwise interactions in human soci-eties are conversational exchanges, in which adaptiveinformation is exchanged which can then be evaluatedby its recipients. Our results therefore predict that incooperative human groups, individuals will attempt tosample the intentions of a sufficient fraction of othergroup members by repeatedly engaging selected part-ners in conversational exchange and in evaluating thepayoffs.

In our model, an individual assesses the probable out-comes of an interaction with a player by using a mem-

ory function in which the outcomes of more recent pre-vious interactions carry greater weight. As an empiricalcorrelate of our memory-discounting function, we notethe existence of extensive data on retention and forget-ting rates in humans. Anderson and Schooler (1991)have proposed that retention and forgetting rates arefunctions of the cost of considering a memory, the gainassociated with successful retrieval, and the probabilitythat we will need a particular memory trace now. Em-pirical curves of human ‘‘forgetting rates’’ over vari-ous time scales (e.g., Pollmann 1998) can be comparedwith theoretical curves derived from ecological modelsof adaptive information-processing in uncertain andchanging environments (e.g., Mangel 1990). Mangel pro-

poses that selection may act to enable the evolution of‘‘optimal forgetting rules’’ in environments with differ-ent fluctuation frequencies. In our model, memory de-cays exponentially with time, and McNamara andHouston (1985) suggest that this may be the optimalweighting rule.

The different personality profiles that we have carica-tured in the model as player types correspond to the dif-ferent strategies that can be employed in the prisoner’sdilemma game. Tentatively, we propose that the levelof cooperation in a group may be reflected in the fre-quencies of personality correlates of different playerstrategies; such basic personality traits may even begeneralizable across primate species boundaries. Argyle

(1991) identifies a suite of traits which tend to be associ-ated with cooperative behaviour in humans, includingextraversion and empathy, while Wilson, Near, andMiller (1996) have proposed that the Machiavellian per-sonality corresponds to the ‘‘always defect’’ strategyfrom the prisoner’s dilemma game. Other variables cor-relating in some way with cooperative behaviour in-clude gender, tendency to self-deception, and attitude torisk (e.g., Surbey and McNally 1997, Raub and Snijders1997). We propose as a testable hypothesis that suchpersonality traits (seen as indices of preferred strategy)

are maintained by their frequency-dependent advatages in human societies as a result either of natural oof cultural selection. This hypothesis could be tested bexperimental studies of the effects of such traits on individual performance in cooperative games.

Future work in this class of models may enable us texplore other aspects of individuals’ behaviour

groups. In this paper we have considered a general implementation based on a minimal number of costraints. In addition to varying the rules by which plaers make their assessments of the possible outcome an interaction, we might incorporate the effects of plaers’ making errors in their decisions by preventingplayer from carrying out a preferred action with a smaprobability, υ. We might set the value of d, in caswhere players did not interact on a round, to be the average of a player’s current assessments of other playerWe might also allow players to share information aboutheir assessments of each other and to monitor the oucomes of third-party interactions as well as of theown. It would be possible to choose new players not

random but with characteristics similar to those in thcurrent group (to mimic genetic behaviour). By varyinthe model in such ways, we may be able to make thsimulations approximate even more closely to the atual pattern of human cooperative exchanges.

References Cited

a i e l l o , l . c . , a n d r . i . m . d u n b a r . 1 9 9 3 . Neocortex sizgroup size, and the evolution of language. current anthropoogy 34:184–93.

a n d e r s o n , j . r . , a n d l . j . s c h o o l e r . 1 9 9 1 . Reflectionsof the environment in memory. Psychological Science 2:396–408.

argyle, m. 1991. Cooperation: The basis of sociability. Lon-don: Routledge.axelrod, r. 1984. The evolution of cooperation. New York:

Basic Books.cosmides, l. 1989. The logic of social exchange: Has natural

selection shaped how humans reason? Studies with the WasoSelection Task. Cognition 31:187–276.

dunb a r , r . i . m . 1 9 9 2 . Neocortex size as a constraint ongroup size in primates. Journal of Human Evolution 20:469–93.

———. 1993. Co-evolution of neocortex size, group size, and language in humans. Behavioral and Brain Sciences 16:681–734

———. 1996. Grooming, gossip, and the evolution of language.London: Faber and Faber.

d w y e r , p . d . , a n d m . m i n n e g a l . 1997. Sago games: Co-operation and change among Sago producers of Papua NewGuinea. Evolution and Human Behavior 18(2):89–108.

g l a n c e , n . s . , a n d b . a . h u b e r m a n . 1 9 9 3 . The outbreaof cooperation. Journal of Mathematical Sociology 17:281–30———. 1994. The dynamics of social dilemmas. Scientific Ame

can 270 (3):58–63.herz, a. v. m. 1994. Collective phenomena in spatially ex-

tended evolutionary games. Journal of Theoretical Biology 1665–87.

m c na m a r a , j . m ., a nd a . i . houst on. 1 9 8 5 . Optimal foaging and learning. Journal of Theoretical Biology 117:231–49

mangel, m. 1990. Dynamic information in uncertain andchanging worlds. Journal of Theoretical Biology 146:317–32.

n e t t l e s , d . a n d r . i . m . d u n b a r . 1997. Social markersand the evolution of reciprocal exchange. current anthropoogy 38:93–99.




n oe, r. 1991. A veto game played by baboons: A challenge tothe use of the prisoner’s dilemma as a paradigm for reciprocityand cooperation. Animal Behaviour 39(1):78–90.

now a k , m . a ., a nd r . m . m a y . 1 9 9 2 . Tit-for-tat in heteroge-neous populations. Nature (London) 359:826–29.

n o w a k , m . , a n d k . s i g m u n d . 1993a. A strategy of win-stay, lose-shift that outperforms tit-for-tat in the prisoner’s di-lemma game. Nature (London) 364:56–58.

———. 1993b. Chaos and the evolution of cooperation. Proceed-

ings of the National Academy of Science, U.S.A. 90:5091–94.pollmann, t. 1998. On forgetting the historical past. Memory

and Cognition 26(2):30–39.poll oc k , g. b . 1989. Evolutionary stability of reciprocity in a

viscous lattice. Social Networks 11:175–212.r a ub , w ., a nd c . sni j der s. 1 9 9 7 . Gains, losses, and cooper-

ation in social dilemmas and collective action: The effects ofrisk preference. Journal of Mathematical Sociology 22:263–302.

s u r b e y , m . k . , a n d j . j . m c n a l l y . 1997. Self-deception asa mediator of cooperation and defection in varying social con-texts described in the iterated prisoner’s dilemma. Evolutionand Human Behavior 18:417–35.

trivers, r. l. 1971. The evolution of reciprocal altruism.Quarterly Review of Biology 46:35–57.

———. 1985. Social evolution. Menlo Park, Calif.: Benjamin/Cummings.

w i l s o n , d . s . , d . n e a r , a n d r . r . m i l l e r . 1 9 9 6 . Machia-vellianism: A synthesis of the evolutionary and psychologicalliteratures. Psychological Bulletin 119:285–99.

Competing Strategies forModernization in theEcuadorean Andes1

m a r c e l o c r u zUrban and Regional Studies/Geography, University of Wisconsin at Green Bay, Green Bay, Wis. 54311-7001, U.S.A. 31 viii 98

This paper explores the nexus between ethnic iden-tity, place, and survival strategies among the Colta ofChimborazo Province in the central Ecuadorean Andes.It argues that demographic pressure, changes in landtenure, and increased integration of the region into na-tional and international economies have challenged thereproduction of Colta culture and identity. In response,Colta have selectively internalized ‘‘Colta-ized’’ aspects

of outside intervention (modernization) and thus recon-stituted Colta spaces and ethnic identity. The paper ex-plores the modernizing role of nongovernmental organi-zations, local indigenous organizations, and the state inthis context.

1. © 1999 by The Wenner-Gren Foundation for Anthropological Re-search. All rights reserved 0011-3204/99/4003-0006$1.00. I thankLeo Estrada, Sanchez-Parga, Lynn Walters, William Laatsch, andseveral anonymous referees for their assistance in articulating thearguments of this report.

colta ethnic territoriality

The Colta Lake region begins about 14 miles southwestof Riobamba, the provincial capital of Chimborazo, andis 3 miles south of Cajabamba, the county seat (canto n).Politically, the area is located in the parish (township)of Santiago de Quito, the urban center of the micro-

region. The majority of the communities of the area areconsidered anejos. In an anejo, the houses are locatedon more or less regular streets, and the main street isusually a highway or road passing through the townalong which the settlement’s few commercial or indus-trial enterprises are found. On the western side of thelake are the settlements of Colta Majipamba (ColtaMonjas) and Yanacocha, which are the focus of thisstudy. Both communities are situated on a hillside over-looking Colta Lake at 3,200 m above sea level, with thePan American Highway separating them from the lake.Except for the communities located in the higher eleva-tions of the paramo, the microregion cannot be consid-ered isolated. It has transportation and communication

links to the provincial and national areas of production,being located on the main Pan American Highway thatconnects Riobamba with the large urban centers ofCuenca to the south, Guayaquil on the coast, and Quitoto the north.

According to the census of 1990, the Colta region hada total population of 47,568 inhabitants. The parish ofSantiago de Quito had a population of 8,422 inhabi-tants, of which the vast majority (86%) lived in the ruralareas in hamlets and annexes. The Colta Lake region isone of the most densely populated Native American re-gions of Ecuador (Aguilo 1987). Approximately 90% ofthe inhabitants are Puruha-Quichua, also called Coltaafter the lake that dominates the region (Botero 1990).

The primary language of the Colta is Quichua, but 80–90% of the men speak Spanish with varying degrees offluency. Many of the women understand Spanish but‘‘don’t know how to answer.’’ In a representative sam-ple of Yanacocha2 87% of the adult males spoke Spanishin contrast to 4 percent of the adult females (Cruz 1994).(More women than this may in fact speak at least someSpanish even though they and their husbands prefer tosay otherwise.) Demographically, the microregion hasshown a steady increase since 1962.3 Between 1962 and1974, the urban population increased by 9.6% and therural population by about 11%. Between 1974 and 1982,the urban population declined by 7% while the ruralpopulation increased by 12%.4

Between 1989 and 1992, Yanacocha had a population

2. The sample consisted of 63 individuals, 45 male and 18 female.An attempt was made to assemble a sample that reflected the gen-der and age composition of the community, but men’s suspicionmade it difficult to interview women.3. The data from the first systematic census, taken in 1950, are in-complete because of Colta resistance.4. Because of the political subdivision of Cajabamba and the cre-ation of Santiago de Quito after 1982, growth rates for the periodbetween 1982 and 1990 are unobtainable. However, adding the pop-ulations of the two parishes that formerly made up Cajabamba to-gether, one finds that the microregion has continued to grow.




of approximately 1,000.5 Basically, 200 families wereclustered together in a dense area. More than three-fourths (77.4%) of the households were based on the nu-clear family (Cruz 1994). The high incidence of nuclear-family households was associated with landownershipcharacterized by small parcels; the average size of a plotwas a mere .33 of a hectare. A similar situation was

found in the neighboring community of Colta Monjas,which had been part of a hacienda of the same namethat was broken up in the course of agrarian reform inthe 1970s. Colta Monjas had a population of approxi-mately 700, and the 140 households that made up thecommunity were also predominantly nuclear families.Here again, landholdings were small.

Household size in Yanacocha was 4.9 and the averagenumber of adults (16–65 years of age) per household2.96. Similarly, in Colta Monjas, household size was 5.4and the average number of economically active adults2.7. What these statistics show is a workforce that ex-ceeds the capacity of its landholdings to support it. Aresponse to this demographic pressure has been a sharp

break with the Andean custom of using the paramostrictly for pasture and fuel gathering and a shift to ag-ricultural use. These soils require a fallow time of sevenyears for soil replenishment, and the reduction of fallowtime to three or four years has exacerbated soil degrada-tion.

The principal crops grown in the microregion are bar-ley, potatoes, broad beans ( habas), oats, and wheat. Be-cause of the degradation of the soils and the lack of wa-ter resources and arable land, there is a growing annualdeficit of production of subsistence crops. This cropdeficit has to be filled by buying basic products in themarket, and cash has become a necessity for survival.Survival strategies using communal mechanisms have

proven unsustainable (Cruz 1994).The lack of land and work makes temporary migra-

tion a viable strategy for reproducing the household.Generally, males between the ages of 15 and 30 andhousehold heads migrate. Because transportation to themajor urban centers of the country is readily accessible,the cost of migration is relatively low. The average timeaway from home is just under three months. Most ofthis migration is to Guayaquil (Abya-Yala 1985). It hasled to profound changes in the patterns of production/consumption in these communities. Increased con-sumption of industrial products has, in turn, influencedchanges in diet, dress, and housing. Concern about theloss of cultural identity among the young who spend

months in the coastal urban centers is a common topicof conversation in Yanacocha. Young men come backrefusing to wear the traditional dress and mixing Qui-chua with coastal Spanish. Many elders are concernedabout the threat to Colta social organization.

This migratory process follows a logic linked to deci-sions made within the individual household. The mi-grant, distanced from the community, participates onlymarginally in the network of relations of reciprocity

5. It is difficult to obtain a precise figure because of Colta suspicionof census taking and the extent of migration.

within it. This break in reciprocal relations may lead a distancing of migrant families even though the wiand children stay behind to work the parcel of land.

Demographic pressure has led to acute parcelizatioof communal lands in the paramo, and increased socidifferentiation has led to internal and external conflicwithin and among communities. With the objective

making production marketable, one sector of the community is obliged to increase its political relations ouside of the community, thus embarking on a process modernization. These changes directly challenge tradtional Colta survival strategies.

changes in the local and nationalpolitical economy

The 1990s have presented a series of challenges to Colagriculture and livelihood, some of them new and oters new only in that analysts and policy makers have last begun to recognize them. Among the most impotant changes are the increasing crisis in the surviv

strategies of the popular sectors as a result of interntional debt, inflation, and the national adjustment polcies of the 1980s. Currency devaluations, which havled to rapid price increases in fossil-fuel-based agrchemical inputs, have made it essential that the use Green Revolution technologies6 be efficient and effetive (Brush 1988). Trade liberalization and the new ethusiasm for revitalizing the regional trading block (thAndean Pact)7 have led to the removal of tariff and othbarriers and the opening up of agriculture to compettive pressures. These changes have increased the presure on small farmers to increase productivity, lowcosts, increase competitiveness, and use all inputs, icluding labor, much more efficiently in both technic

and economic terms.As a result of these macrolevel changes, indigenou

technical knowledge is being abandoned. This is not say that Colta knowledge of their land and crops doenot make important contributions to their technical rsponses to these challenges. It does—especially in thidentification of low-input agricultural options. Nontheless, low-input agriculture has seldom proven ecnomically viable. Economic and technical efficienrequires capacities for numeracy, economic abstration, market research, and the identification of coscontrolling high-yield seeds that the rural poor rarepossess (Chambers 1983).

The Colta are firmly integrated into a capricious an

changing market, and their well-being and survival dpend on how they handle this integration. The incorpration of rural areas into the wider economy producesseries of lifestyle changes. Modernization arrives in thform of fertilizers, radios, new textiles, bicycles, vanschool notebooks, school uniforms, and the clothes an

6. These technologies were introduced into the region in the 196by the Andean Mission Program of the United Nations, the UPeace Corps, and the Cornell Development Project.7. New regional agreements within the Andean Pact went into efect January 1, 1992, and January 1, 1993, opening up agriculturmarkets.




cars of nongovernmental and governmental extensionagents. With these come new aspirations that require ahigher income, and poor farmers look for technologiesthat may provide this income.

Yet integration and modernization have also pro-duced growing concern among the popular sectors forasserting Native American identities. This assertive-

ness is manifest in the rise of the national indigenousmovement under the umbrella of the Confederation ofIndigenous Nationalities of Ecuador (CONAIE). How-ever, because of the rapid change occurring in the coun-tryside, the goals of the movement are ambiguous andcontradictory. For example, in June 1990 CONAIEcalled on Ecuador’s Native Americans to support a na-tional uprising protesting government apathy with re-gard to indigenous peoples’ needs and demanding recog-nition of their cultural difference (Macas 1991)8 whileat the same time demanding their full and fair incorpo-ration into the nation’s development process. Anotherindigenous uprising took place in August 1997 pro-testing rural residents’ loss of social security benefits

and unfavorable constitutional reforms passed by theEcuadorean Congress. CONAIE speaks of the recoveryof indigenous crops, technologies, crop-environmenttheories, and cosmologies within larger strategies ofethnic self-determination and cultural revalorizationwhile also demanding that Native Americans haveequal rights as citizens and be allowed fairer access tomarkets, credit, research, and agricultural extension(Macas 1991). In short, on the one hand CONAIE sup-ports the perpetuation and recovery of cultural tradi-tions, and on the other it demands access to the meansof rural modernization and to the technologies and in-stitutions of the cultural other. The latter demands arenot only incongruent with the ‘‘traditional’’ practices

and ideas in which Native American identities are be-ing sought but also might be expected to contribute di-rectly to changing them.

These apparent contradictions point to the difficultyof defining and sustaining different Native Americanidentities in a modernizing economy. A possible resolu-tion can be found in CONAIE’s claim that, as bothEcuadoreans and Native Americans, indigenous peoplesare entitled to both community self-determination andaccess to state resources (Macas 1991:25–26), the impli-cation being that communities themselves should de-cide the balance between traditional and modern as-pects of their ethnic identity.

We have, then, a resurgence of concern for identities

with roots in past practices and beliefs. At the sametime, we are faced with a ‘‘new technological agenda’’(Brush 1988) that requires rapid modernization of indig-enous resource management strategies, more efficientGreen Revolution strategies, and the identification ofnew nonagrarian livelihoods that are linked to thebroader regional, national, and international econo-mies. Thus, the possibility of securing the material ba-sis for any cohesive cultural identity at all is linked to

8. Luis Macas, a Saraguro from southern Ecuador, was president ofCONAIE in 1991.

the responses to the contemporary political economiccontext. But is an ethnically distinct identity compati-ble with these modernized livelihood strategies?

Prior to land reform, the largest hacienda in the Coltaregion, Colta Monjas, devoted a relatively small area ofland to crop cultivation and left the remainder in pas-ture. This allowed a rotation with long fallow periods,

although by the 1960s some agrochemicals and newcrops had been introduced. Many Colta first experi-enced modern technology on the haciendas or in share-cropping arrangements. The Colta production systemwas based upon small plots of land devoted to a widerange of traditional crops grown primarily for householdconsumption. Agrochemicals were rarely used. The sus-tainability of the Colta system depended, however, onheavy inputs of manure from animals kept on the haci-enda’s pasture lands (IERAC 1987, 1988). This agro-ecologically sound indigenous agriculture ended withland redistribution. Hacienda pastures were dividedamong Colta farmers and planted in crops. As popula-tion increased, land was subdivided and fallow periods

were reduced. No intensification, such as stall feedingof cattle or terracing, took place, and soils were rapidlydepleted.9 The use of chemical fertilizers and pesticidesincreased with their greater availability, guaranteeingproduction from crops weakened by poorer soils (IERAC1991).

These changes were accompanied by socioeconomictransformations. Increased market orientation discour-aged the cultivation of traditional Andean crops in favorof marketable ones such as potatoes and broad beans aswell as horticultural crops such as onions, garlic, car-rots, and beetroot. Marketable crops, however, weremore easily grown on the better soils found below 3,500m and on land that was gently sloping (less than 20°)

and had access to irrigation. Thus, some households hada competitive edge. The need for capital made for a cer-tain dependency on local mestizos10 and whites. Landsubdivision in the context of local joblessness led to in-tercommunal conflicts and/or increased seasonal mi-gration to urban and coastal areas. As a consequence ofmigration, participation in community activities de-clined, health problems, petty theft, and violence in-creased, and respect for elders deteriorated.

With the hacienda system practically gone, there hasbeen a rupture of the social relations between haciendaand Colta communities that were the basis of the repro-duction of community life. Thus, these postreformchanges suggest an increasing reproduction squeeze on

the Colta economy. This has become increasingly se-vere since the mid-1980s, given rapid inflation and the

9. The reasons for this are unclear. The lack of terracing apparentlyreflects farmers’ calculations that they can earn more by migratingthan by investing time in terrace construction. The changing cul-tural valorization of wage income over agriculture for householdproduction is another possible reason; that the former is associatedwith males and the latter with females also helps to reduce the cul-tural status of agriculture.10. A particular relationship is established between Colta commu-nities and speculative capital controlled by mestizos called chul-queros who act as intermediaries in Balbanera and Cajabamba.




austerity measures implemented to control it. Thesemeasures have increased the cost of agrochemicals andhave been prejudicial to the terms of trade for mostColta products, further increasing pressure to migrateeven though urban labor markets are themselves tight-ening. The effect of the most recent policy change—free-trade blocks—is less clear. Ecuador has only re-

cently opened its doors to free trade with Colombia andother countries within the Andean Pact. It is clear, how-ever, that in order for Colta farmers to survive in thesenew policy environments and to take advantage of op-portunities as they arise, they will require increasinglyefficient production strategies and knowledge of pro-cesses in distant markets.

The Colta are fully aware of these changes. The newchallenges that emerge from increased commercial in-tegration have devalorized Colta products that have tra-ditionally been tied to Colta ethnic identity. This de-valorization is not limited to capital circulation but alsoreaches the productive sphere by giving priority to com-mercial over subsistence crops. Indigenous technical

knowledge has been associated with subsistence cropcultivation, which Colta farmers now consider insuffi-cient for the reproduction of community life.

alternative strategies

The debate over survival strategies has been conductedamong a range of institutions, both internal and exter-nal to the region, each promoting its own particularconception of what it means to be indigenous and ofthe relationships between identity, technology, and in-digenous social organization. These conceptions havechanged over time—partly on the basis of experiences,

partly in response to a continuously changing context,and partly as a result of the borrowing of elements ofothers’ approaches. Models of agrarian developmentpursued by the state, by Catholic and Protestant reli-gious institutions linked with nongovernmental organi-zations (NGOs), and by Puruha-Quichua regional feder-ations have all had an impact on Colta communities.

State agencies, to varying degrees, promote the use ofGreen Revolution techniques. This approach was ini-tially based on the idea that Colta who had just receivedland through agrarian reform required technical assis-tance to make that land productive. In 1973 the militarygovernment made the province a priority zone for themodernization of the agricultural sector. This was also

a program of cultural assimilation, aiming to turn ‘‘Indi-ans’’ into Ecuadoreans.

This governmental approach to agrarian developmenthas become the model to which NGOs and more radicalPuruha-Quichua federations have contrasted their ownwork, but it has had an impact on Colta perceptions ofthe future of indigenous agriculture. A generation ofyoung Colta adults (now in leading positions) has grownup with a state presence in their communities, andmany now believe that the reproduction of Colta agri-culture and society will necessarily be mediated bystate resources. A belief also prevails that it is the

state’s responsibility to give this support and that it the Colta’s right to demand it.

Concurrently, the NGOs and the local indigenoufederations have laid aside purist conceptions of indignous agrarian development and increasingly incorprated elements of Green Revolution technology, reconizing that in certain instances traditional technologi

are no longer viable. However, they are far from adoping the same conception of modernization as that of thstate. The essential difference between the two is selmanagement. The nongovernmental institutions blieve that local space and modernization within should be administered by Colta organizations at thcommunal and supracommunal levels. The emphason the community, a more traditional unit of selmanagement, builds on historical identities and is itsean element of cultural revalorization, even if that community is managing modernization.

The Catholic church’s previously conservative role Chimborazo was undermined by land reform and thrise of liberation theology. Taking the side of the indig

nous poor, the church under Monsignor Proano becamdirectly involved in promoting social change. It hbeen politically active, supporting land invasions andwide array of popular educational activities which havencouraged Puruha-Quichuas to claim political righand make demands on the state. More important, thchurch has become directly involved in grassroots dvelopment, fostering a model of rural development foindigenous peoples that involves the revalorization Colta culture as a first step toward a heightened politcal awareness and stronger indigenous political organzation. As part of this revalorization, church programhave as one of their goals the recovery of traditioncrops and land use practices. Yet at the same time, in i

political conscientization work, the church encouragColta to demand resources for rural modernization fromthe state.

This mismatch does not exist in the Protestanchurches, which are far more consistently modernizinin their outlook. They foster the use of good seeds anagrochemicals as means of controlling the enviroment, and they are strongly committed to a basicalWestern model of improved lifestyles, health care, anhousing. These groups are linked to Lutheran churchein Norway and Germany and to nondenominationgroups in the United States. Although some of the critcism aimed at the Protestant churches is well foundemuch of the critique that addresses their modernizin

influence and the eradication of Colta culture is exagerated and misplaced. Critics cite, for example, thshift in housing construction from the traditional chozor thatched-roof hut to Western-style housing and fromthe traditional dress toward synthetic materials such polyester and orlon. Yet changes in the diet have rduced illness and malnutrition as milk and dairy proucts as well as meat are added to the diet. Alongsidecommitment to modernization is a concern that Colcontinue asserting their cultural difference through laguage, dress, strong local organizations, and, more important, indigenous control of rural space and social lif




One of the primary differences between Catholics andProtestant in Chimborazo is that Protestant ministers,originally foreigners (European or North American), arenow Colta. Under Catholicism, the priest acts as an in-termediary between God and Colta in ritual. Protes-tants teach a one-to-one relationship with the creatorthat is more in line with traditional Colta spiritual be-

liefs. Thus, Protestantism breaks a dependency onwhite spiritual paternalism. There is a general aware-ness among Colta that the Catholic religious festivitiesconstituted a type of exploitation by whites. A Protes-tant informant remembers the festivals (Cruz 1994):

The festivals I remember . . . nothing good. Backthen give money to the merchants, to the mestizos,who have all the things. Back then, buy meat, flour,buy rice, buy sugar, buy all the things. To whom dowe give money? To the merchants, the mestizos, tothose that do commerce, to them only we givemoney. . . . there is nothing for us. That is why I re-jected the old ways.

Many critics lament and condemn these culturalchanges. However, the same respondent answers suchcriticism: ‘‘But this does not mean, like everyone says,that we are losing our culture. It is a lie; just becausewe do not observe the old festivals does not meanthat we are going to lose our culture.’’ Thus, Prot-estant church administrations have been deliberatelydecentralized by letting the community control them.Furthermore, church-supported agrarian projects haveinvolved high levels of decision making by Colta. Prot-estant ideology is more in line with Colta ideology inbelieving that Colta should be self-confident in takingcontrol of their own lives and seeing themselves asequal to non-Colta (Alban 1987).

Associated with the Catholic and Protestant churchesare NGOs that have been active in community de-velopment programs in Chimborazo since the early1970s. Among the most active and strongest organiza-tions are World Vision, based in the United States, theEcuadorean Center for Agricultural Services (CESA),and the Ecuadorean Fund for the Progress of the People(FEPP). World Vision is closely associated with theNorth American Protestant groups, the FEPP with theradical wing of the Catholic church, and the CESA withtraditional Catholicism and the more conservativeChristian Democratic party. All aim at increasing Coltaparticipation in and control of NGO agricultural proj-ects, and all insist that indigenous farmers claim their

right to benefit from state services and resources. Thereis variation, however, in the technologies they promotefor indigenous agriculture and the extent to which theyconsciously try to increase peasants’ propensity and ca-pacity for claiming rights and in their approach to cul-tural tradition.

Strongly influenced by liberation theology, the FEPPseeks to strengthen Colta organization, self-manage-ment, and culture. It concentrates on training Coltapromoters to work as agricultural advisers in their ownsectors. Similarly, the organization’s respect for and at-tempts to recover and revalidate ‘‘traditional’’ technolo-

gies are part of a concern to affirm Colta cultural prac-tices and promote agroecological alternatives. The FEPPdiscourages agrochemical pest control and fertilization,instead experimenting with organic and low-external-input fertilization and pest control techniques throughcompost management, green manuring, application ofash to soils, intercropping, terracing, and so on. It favors

Andean tubers and grains and the use of native crop va-rieties.

The FEPP has lost credibility among many Colta be-cause of its conception of what it means to be indige-nous and its insistence on ‘‘traditional’’ technologiesand crops that are increasingly nonviable in the chang-ing context of Ecuadorean rural economies. The farmerswith whom the FEPP still works prefer modern technol-ogies and varieties. It has responded lately by providingcredits to purchase such inputs while still encouragingfarmers to think about the causes and implications ofagrochemical use and of the loss of traditional crops.Colta ideas have also influenced the FEPP’s decision toaccept the notion that modern technologies have a role

to play in any future indigenous agricultural develop-ment. Given its Catholic ties, the shift in the directionof Protestantism among the Colta is an added motiva-tion for rethinking its approach.

The CESA is specifically concerned with creating lo-cal organizations that increase Colta’s ability to pres-sure the state for resources. A longer-term aim is forthese peasant organizations, once strengthened, to con-tinue to administer their own agricultural projects withdeclining assistance from CESA. Underlying this visionis a model of an indigenous agricultural future inte-grated into national development, using, in a scaled-down form, modern technologies generated by the pub-lic sector. Many CESA staff members accept that much

traditional indigenous technology is not viable undercontemporary conditions. They envision an indigenousalternative that combines indigenous forms of social or-ganization with Western technologies. CESA has re-cently had some success among the Colta, but its asso-ciation with the conservative Christian Democraticparty makes it the object of suspicion. Also, its trainingof young leaders has created tension in the communi-ties’ internal political structure.

World Vision’s approach is considerably different,promoting the rapid modernization of Colta agriculturewith dissemination of agrochemical technologies andnew seed varieties. While aspects of these actions arejustifiably criticized (Alban 1987), the NGO’s field-

workers argue that they give Colta access to techno-logical and income possibilities from which they werepreviously excluded. They also encourage communitycontrol and implementation of these modernizationprojects. What distinguishes their approach from that ofthe other NGOs is that almost all project field staff andpromoters are Colta. The agrarian model fostered is acombination of Western technologies and indigenousadministration. World Vision has been in Chimborazoprovince since 1979, and 80% of its projects at the na-tional level are in the province.

World Vision’s link with Protestant groups in the




United States that emphasize religious conversion inpart explains the increase in Protestantism in the re-gion. However, what is often overlooked is the roleof indigenous organizations in this conversion. Thestrengthening of communities as units of territorial ad-ministration has been one of the most significant socio-political changes in rural Chimborazo. At the same

time, a novel form of indigenous organization, thefederation of indigenous communities, has emerged.These federations engage in both political and devel-opmental activities, negotiating with public agencieson the one hand and implementing development proj-ects on the other. Some are more developmentalistin orientation and others more politically radical. Thegrowth and increasingly strong self-management ca-pacity of the federations results in part from supportfrom NGOs and church organizations, in particular,Protestant churches.

Neither the federations nor their programs are en-tirely ‘‘endogenous’’ innovations. For instance, the fed-erations’ agricultural programs, which typically involve

seed and input supply, farmer-to-farmer extension, andsome on-farm research, are modeled closely on NGOand state programs.11 However, if grassroots controlrather than technological content gives a strategy its‘‘alternative’’ character, what matters is not whetheragricultural development strategies are endogenous butwhether they are locally controlled. In the federationsthis local control is not perfectly democratic; certaingroups exercise more influence within them than oth-ers. The federations are, nonetheless, more accessibleand accountable to local people than any other develop-ment institution.

The emergence of these federations reflects a furtherstage in the Puruha-Quichua’s recovery of Andean

space. Going beyond the recovery of land as means ofproduction, the federations are slowly recovering theadministrative control of rural space in the province,taking back terrain once administered by the haciendaand questioning the very control of space by the stateand the dominant white/mestizo society. As white andmestizo presence declines in places like Colta, rural ar-eas are thus being returned to indigenous people to prac-tice their culture and agriculture. The federations’ per-spectives on the relationships between technology andethnic identity in an indigenous agricultural develop-ment can be understood in these terms. This is not tosay that the rationales stemming from this conceptiondetermine the federations’ strategies—socioeconomic

and ecological processes are equally important—butrather that they give meaning to these strategies. Theresult is a vision of indigenous agricultural develop-ment that embraces concerns for agrarian technology, astronger Puruha-Quichua cultural identity, and controlof rural space.

The main point of contention between and within thetwo main indigenous federations is the extent to which

11. The exception being that the state has not used farmer-to-farmer extension.

they should promote modern agrochemical and crotechnologies as opposed to traditional low-input tecnologies. The Asociacion de Indıgenas Evangelicos dChimborazo (AIECH) is linked with Protestantism anhas worked closely with World Vision and with Lutheran groups from Norway and Germany. The moradical federation, the Movimiento de Indıgenas d

Chimborazo (MICH), has emphasized the recovery traditional agriculture and technology and has workewith both the FEPP and the CESA and with the radicwing of the Catholic church. The MICH’s programs agricultural development promote the recovery of Adean crops, use of organic fertilizers, and replacement pesticides with supposedly traditional methods of pecontrol. The rationale underlying this strategy is thit constitutes a rejection of the agricultural technologassociated with white and capitalist culture and an afirmation and validation of indigenous identity.

These strategies seek social and cultural empowerment through agrarian development built on tradtional practices. However, promoting this alternati

among Colta farmers already producing for the markwho had little land from which to produce organic mnures has proved difficult. For instance, farmers in Ynacocha plant five varieties of barley, of which onone, chilena, is grown for domestic consumption. Simlarly of the five varieties of broad bean grown, four afor the market. The traditional Andean crops— quinomellocos, oca, and mashuas—are of secondary impotance. Pressures from their members have led the fedeations to work with agrochemicals, new varieties, ancash crops. Their focus now is on improving efficiencand farmer knowledge of the risks of such products. Imaking this shift they have approached a more ‘‘deveopmentalist’’ model more in line with the Protestan

modernist notion of self-reliance. They argue that moern technology can improve Colta income and thmodernization, far from being a cause of cultural ersion, is a means of cultural survival. Periodic migratioand the problems associated with it are seen as moof a threat to Colta cultural coherence than the use agrochemicals and new crop varieties. The principle that indigenous cultural identity hinges on sustaineand corporate rural residence rather than on retainintraditional technologies.

The implication is that indigenous economy and cuture must constantly adapt in order to survive and sutain group cohesion and self-management. At the samtime, there is a politically radical dimension to th

‘‘modernization from within.’’ Many Colta associa‘‘traditional’’ technologies with the hacienda, and fthem rejecting traditional technology is a symbol fthe rejection of dependent and exploitive social reltions. To embrace modern technologies is to makestatement about indigenous social equality and thright of indigenous people to benefits that historicalwere the privilege of whites and mestizos. This use omodern technology is thus part of the wider discourson citizenship rights (Cruz 1994).

Furthermore, the hoped-for benefits of moderniz




tion—reduced migration, increased community cohe-sion—are also intended to strengthen indigenous orga-nizations as sociopolitical vehicles for demandingsocial and political change, access to resources, and amore prominent role for Colta in rural development andrural government. The power to be able to administerrural modernization through their own organizations

directly challenges the ethnic exclusivity of state ruraldevelopment administration. Fostering technologicalmodernization is a response to the realities of marketproduction, soil degradation, and land subdivision. Ag-ricultural production alone, however, is not enough forsustained income growth. It is necessary to control thelocal marketing and processing of agricultural products.Thus Colta communities have begun to challenge theposition of intermediaries in the marketing chain by es-tablishing their own marketing program. In the micro-region of Colta, mestizos still control local commerceand transport, although their control is being seriouslychallenged by Colta market and transportation coopera-tives. A viable indigenous development will require a

restructuring of marketing and other social relation-ships to allow for the production of higher-value andprocessed products under the control of Colta that reachthe market more directly, resulting in higher farm in-comes.

conclusion

The Colta perceive a close relationship between resid-ing in rural areas and sustaining their identity as indige-nous farmers. An ‘‘indigenous’’ agrarian developmentmust, therefore, allow occupation of traditional Coltaspaces. In the current political economic and agro-ecological context, productive strategies based on tra-ditional technologies do not appear to be viable meansof achieving this objective. Ethnic identity will begrounded in social, spatial, cultural, and linguistic prac-tices and not in traditional technology.

The Colta response has been to identify with a con-cept of ‘‘Indianness’’ that goes beyond class differentia-tion by creating economic and political links with ex-ternal Western development agencies. The significantrole played by international organizations and nongov-ernmental agencies in the identity politics of indige-nous federations has been principally in helping localactors confront national secular interests. Under thisnotion, the sense of ethnic identity is being renegoti-ated, as are the power relations between different fami-lies within the community and other communities inthe microregion. The regional and national organiza-tions that have emerged in this process are now in a po-sition to negotiate with the state and nongovernmentaldevelopment agencies that are increasingly having animpact on the local level.

References Cited

aby a-y ala. 1985. Del indigenismo a las organizaciones in-dı genas. Quito.

a g u i l o, federico. 1987. El hombre del Chimborazo. Quito:Abya-Yala.

a l b a n , m a r ı a. 1987. Con Dios todo se puede: La invasio n de las sectas en el Ecuador. Quito.

b ot er o, lui s fer na ndo. 1 9 9 0 . Chimborazo de los Indios.Quito.

b r ush, st ep hen. 1 988. ‘‘Traditional agricultural strategies inthe hill lands of tropical America,’’ in Human impact onmountains. Edited by Nigel J. R. Allan, Gregory Knapp, and

Christoph Stadel. Austin: University of Texas Press.c ha m b er s, r ob er t . 1 983. Rural development: Putting

the last first. New York: Longman Scientific and Technical.cruz, marcelo. 1994. N ˜ aupaj Manpuni: Rethinking develop-

ment, community, and ethnicity in the Ecuadorean Andes.Ann Arbor: UMI.

i e r a c ( i n s t i t u t o e c u a t o r i a no d e r e f o r m aa g r a r i a y c o l o n i z ac i on). 1987. ‘‘Boletın informativo,’’in Jefatura Regional Centro Oriente: Bolı var, Cotopaxi, Chim-borazo, Pastaza, Tunguragua. Riobamba.

———. 1988. ‘‘Boletın informativo,’’ in Jefatura Regional CentroOriente: Bolı var, Cotopaxi, Chimborazo, Pastaza, Tunguragua.Riobamba.

———. 1991. Breves notas del IERAC, en sus 27 anos de vida. ElInformador Agrario (Quito), July.

macas, luis. 1991. El levantamiento indı gena visto por sus protagonistas. Quito: Instituto Cientıfico de Culturas

Indıgenas/Amauta Runacunapac Yachai.

The Zapotec-ImperialismArgument: Insights from theOaxaca Coast1

r o b e r t n . z e i t l i n a n d a r t h u r a . j o y c eDepartment of Anthropology, Brandeis University,

Waltham, Mass. 02454/Department of Anthropology,Vanderbilt University, Nashville, Tenn. 37235. 5 x 98

Over four decades ago, Angel Palerm and Eric Wolf, in anow classic paper (1957), offered an explanation for whyonly five ‘‘key areas’’ of Pre-Columbian Mesoamericahad evolved into enduring centers of what they termed‘‘massed economic and demographic power.’’ They ar-gued that aside from a necessary degree of agricultural

1. © 1999 by The Wenner-Gren Foundation for Anthropological Re-search. All rights reserved 0011-3204/99/4003-0007$1.00. Wethank John Clark, Robert D. Drennan, Kent Flannery, Joyce Mar-cus, Marcus Winter, and three anonymous referees for their valu-

able comments on an earlier version of this article. Where we didnot heed their counsel, the consequences are totally ours to ownup to. We also express our gratitude to the Instituto Nacional deAntropologıa e Historia, especially the several directors of theCentro INAH Oaxaca, Manuel Esparza Camargo, Marıa de la LuzTopete, Ernesto Gonzalez Licon, and Eduardo Lopez Calzada, fortheir support of our field research. Funding for this research wasprovided by grants from the National Geographic Society (grant3767-88), the National Science Foundation (grants GS 38475, BNS87-16332, BNS 89-18824), the Wenner-Gren Foundation (GR.4988), Brandeis University, Rutgers University, the VanderbiltUniversity Research Council and Mellon Fund, the Fulbright Foun-dation, the H. John Heinz III Charitable Trust, the Explorers Club,and Sigma Xi.




potential, the most critical factor in this developmentalprocess had been the ability of these centers to establish‘‘symbiotic’’ economic relationships with their ecologi-cally distinct hinterlands. The symbiosis between cen-ter and periphery was typically initiated through trade,but at some point relationships became asymmetricalwhen the burgeoning centers realized that their outly-

ing sources of supply could be more effectively ex-ploited through military conquest or colonization. Adate of around 500 b.c. was assigned to the beginningof this stage of development. The highland Valley of Oa-xaca was specifically mentioned as a key area, and theadjacent Pacific coast was identified as one of the areasthat provided the requisite ecological diversity.

The Palerm and Wolf reconstruction anticipated a hy-pothesis favored by most prehistorians currently work-ing in the Valley of Oaxaca to account for the growthof the Zapotec state at Monte Alban. Founded about 500b.c., Monte Alban rapidly developed into the urban cap-ital of highland Mexico’s first state polity (Blanton1978, Joyce and Winter 1996, Marcus and Flannery

1996). By the Terminal Formative period, 100 b.c.–a.d.200, the Zapotec rulers at Monte Alban are thought tohave embarked on a campaign of hinterland subjugationmotivated by a desire for greater quantities of ornamen-tal seashells, tropical fruit, and other exotic goods. Trib-ute, according to the imperialism hypothesis, wasdeemed a more effective means of extracting wealthfrom Monte Alban’s former exchange partners. We re-view this hypothesis and examine it specifically as itrelates to interaction between the Valley of Oaxaca andtwo of its lowland sources of supply along the Pacificcoast: the Lower Rıo Verde Valley and the southernIsthmus of Tehuantepec (fig. 1). Finally, drawing on par-allels from other examples of early state development,

we evaluate whether the imperialism hypothesis canfully account for Monte Alban’s relationship with itsvarious hinterland neighbors.

support for the zapotec-imperialismhypothesis

The most direct archaeological evidence that the Zapo-tec conquered or colonized places for the purpose of eco-nomic exploitation comes from the Cuicatlan Canada,located about 100 km northwest of Monte Alban. Exca-vation and surface survey conducted by Elsa Redmond(1983) and Charles Spencer (1982) produced evidence ofviolent settlement destruction, social disruption, rural

population relocation, agricultural alteration, and for-eign occupation, all of which were attributed to Zapotecconquest and direct administration of the region.

The Cuicatlan investigation was, in part, inspired byJoyce Marcus’s (1976, 1983, 1992) epigraphic study ofthe so-called conquest slabs on Building J at Monte Al-ban. Previously, Alfonso Caso (1947) had proposed thatthe approximately 50 carved stones set into the exteriorwalls of the odd, arrowhead-shaped building referred toprovinces conquered by the Zapotec. On each of thecarvings is a ‘‘hill’’ sign above which are one or more

glyphs identifying a specific place. The bottom segmenof some but not all of the carvings—the part below th‘‘hill’’ sign—incorporates the outline of an inverted human head. Caso had proposed that these upside-dowclosed-eye heads depicted the dead rulers of the coquered places. Marcus and Flannery (1996:197) suggethat the distinction between the slabs that include th

effigy head element and those that lack it is that thlatter refer to places subjugated by means of coloniztion rather than military conquest. Convinced that thplace glyphs on the slabs refer to identifiable geographlandmarks delineating the outer boundaries of the Zpotec empire, Marcus has been working on pinpointinthese places. Her approach has been to compare thcarved toponyms with ones found in the Codex Mendoza, a 16th-century Aztec tribute list. Doing so, shhas tentatively identified seven such places (Marcu1992:395–96). One she interprets as the CuicatlaCanada.

Bearing directly on the question of Zapotec presencon the Oaxaca coast is another of Marcus’s identifie

toponyms, the one she associates with Tututepec, lcated about 20 km inland (fig. 2). Unfortunately, thbase of the Tututepec slab, an important potentisource of textual information about that region’s reltionship with Monte Alban, is missing. The slab is thonly one so far associated with a coastal locality and believed by Marcus and Flannery (1996) to refer tomountain landmark in the region of Tututepec, near tharchaeological site of San Francisco de Arriba. The silies on the northeastern boundary of the Lower RVerde Valley, one of two coastal regions where we havcarried out a series of archaeological investigations.

The other Oaxaca coast region where we have conducted field research is the southern Isthmus of T

huantepec. The region lies just beyond the estimate150-km boundary of the Zapotec empire and has nobeen linked with any of the deciphered Building J toponyms. Its major settlement, Laguna Zope, has, however, been recognized as a likely supplier of ornamentseashells and possibly other goods to settlements in thValley of Oaxaca dating back to the early Formativ(Flannery and Schoenwetter 1970; Pires-Ferreira 197Zeitlin 1978, 1979, 1993).

Around 200–100 b.c., just when the Zapotec athought to have been engaged in their imperialist veture, many outlying regions of Oaxaca, among them thsouthern Isthmus of Tehuantepec and the Lower RVerde Valley, experienced the intrusion of a distinctiv

grayware ceramic style clearly related to the pottery Monte Alban (e.g., Gaxiola 1984; Joyce 1991a, 1993Spencer 1982; Zeitlin 1979, 1994). In excavations iboth the southern Isthmus and the Lower Rıo VerdValley, Monte Alban–related grayware dating to this priod accounts for nearly half the ceramic assemblage

Close ceramic relationships between the Valley Oaxaca and the Cuicatlan Canada that include not ongrayware but imitations of Monte Alban creamwahave been cited as evidence of Zapotec conquest of thregion (Redmond 1983:116–17). Indeed, Marcus an




Fig. 1. Map of Oaxaca indicating places discussed in the text. 1, Monte Alban; 2, San Francisco de Arriba;3, Rı o Viejo; 4, Cerro de la Cruz; 5, Tututepec; 6, Laguna Zope.

Flannery assert that ‘‘Zapotec expansion is confirmedby the large number of neighboring regions whose pot-tery sequences show an abrupt change to the MonteAlban style’’ (1996:206). Subjugation, they argue, isdemonstrated in ‘‘those regions whose previously au-tonomous ceramics are literally swamped or replacedby Monte Alban gray wares’’ (p. 199). This notwith-

Fig. 2. Tututepec conquest glyph (redrawn fromGarcı a Moll, Brown, and Winter 1986).

standing, they feel that Tehuantepec, despite its largeintrusive grayware component, was too distant to havehad any direct relationship with Monte Alban. Thenorthern part of the Lower Verde region they see as hav-ing been subjugated by Monte Alban, while sites on theLower Verde’s floodplain they interpret as having re-mained independent although perhaps intimidated bythe Zapotec empire (pp. 201–2).

the oaxaca coast during thelate/terminal formative

Given the above background, it became a central objec-tive of our recent archaeological research in both theLower Rıo Verde Valley and the southern Isthmus ofTehuantepec to help clarify the role Monte Alban

played in the sociopolitical development of these pe-ripheral coastal regions. In the Lower Verde region thefield component of this research included archaeologi-cal excavation at 12 sites, a regional site reconnais-sance, and a full-coverage surface survey over an area of70 km2 (Joyce 1991a, b, 1993a; Joyce, Winter, andMueller 1998; Workinger and Joyce 1996). Recent andprior work on the southern Isthmus has included exca-vation at 3 sites, regional site reconnaissance, and a full-coverage survey of over 100 km2 (Zeitlin 1979, 1990,1993, 1994; J. Zeitlin 1978).






The developmental trajectories of these two regionsdiffered somewhat, in part because of their dissimilarphysiographies. The Lower Verde region is on thestretch of Mesoamerica’s Pacific seaboard characterizedby rugged mountains descending to the sea (West 1964).While a narrow beach zone follows much of the shore-line, regions within this section of the coast tend to be

isolated from one another. Data are lacking that wouldindicate to what extent the people of the Lower Verdecommunicated with coastal groups to the west; we doknow that there is little evidence of contact with thesouthern Isthmian region to the east (Joyce 1993a). Asfor communication with the highland Valley of Oaxaca,resource complementarity may in part account for theobserved interaction that occurred between these twoenvironmentally contrasting zones during the Late/ Ter-minal Formative. River courses and mountain trails,some utilized to this day, would have been the likelyroutes of communication.

Physical conditions change markedly at the Isthmusof Tehuantepec, where the slopes of the Sierra Madre

del Sur recede and the Pacific shoreline is transformedinto a broad coastal plain. The open environment ex-tends southeast into the fertile region of the Pacific lit-toral known as the Soconusco. From the southern Isth-mus, an easily traversed depression extends north toconnect only 200 km away with the Gulf Coast and re-gions beyond. Interaction with highland regions flank-ing the Isthmus is also uncomplicated. Following river-beds and natural passageways, contact with the Valleyof Oaxaca and highland Chiapas is evidenced from EarlyFormative times onward (Zeitlin 1978, 1979). Thesouthern Isthmus was a virtual crossroads of communi-cation in Mesoamerica.

According to Marcus and Flannery, Monte Alban’s

ambition for territorial control was driven by a desire to‘‘establish a north-south ‘corridor of influence’ betweenTehuacan—the gateway to Central Mexico—and thePacific Coast, the gateway to the tropics’’ (1996:206).The most practical corridor from Monte Alban to thetropics would have been southeast, to the broad openplains of the southern Isthmus, rather than through themountainous, isolated Oaxaca coast to the west. TheAztecs, centuries later, were well aware of this, as isdemonstrated by their numerous military and diplo-matic efforts to control the Tehuantepec region.

Evidence for two- or three-tiered settlement hierar-chies, ascribed status differences, elite control of pres-tige goods, and large-scale construction activities indi-

cates that by the Late Formative period, from 400 to 100b.c., both the southern Isthmus and the Lower Verderegion were the locus of chiefdom-level polities (Joyce1991a, b, 1994; Zeitlin 1978, 1990, 1993), as was theValley of Oaxaca (Marcus and Flannery 1996). Our in-vestigations suggest that the exchange of prestige goodsplayed an important role in the political economy ofthese societies.

Prestige-goods procurement, controlled by elite mem-bers of each society, was implemented through a net-work of interregional cultural and material exchange

that had begun to take form over 1,000 years earlier, the Early Formative (Blake 1991, Flannery 1968, Hir1984, Pires-Ferreira 1975, Zeitlin 1978). By means this exchange network, elites acquired the goods thedesired both as status symbols and for local redistribution to high-ranking members of subsidiary lineagin return for their allegiance and economic suppor

Among the items known to have passed through thnetwork were ornamental marine and freshwater shellstingray spines, iron-ore mirrors, semiprecious stonemica, and fancy pottery. In addition, it is likely thmany other goods, unpreserved in the archaeologicrecord, were so distributed. Cacao, sea salt, feathertextiles, pelts, salted fish and shrimp, tropical fruipurpura dye, and cotton are just some of the coastlowland products known to have been exported to thhighlands during later periods of prehistory (Ball anBrockington 1978, Paso y Troncoso 1981, Whitecotto1992).

Both shell ornaments and unworked shell have beerecovered from Late Formative burials in the Lower R

Verde Valley, although there is currently no indicatioof any local export industry (Joyce 1991a). On the soutern Isthmus, evidence for large-scale procurement of onamental seashells and production of shell ornamenhas, as noted above, been found at Laguna Zope (Zeitli1978, 1979). By the Middle Formative period, 800–40b.c., Laguna Zope had grown into one of Oaxacalargest communities, its development undoubtedstimulated by its significant role in the shell industrDuring the subsequent Late/Terminal Formative, fro400 b.c. to a.d. 200, the industry apparently came undthe exclusive control of the settlement’s elite popultion (Zeitlin 1993:96–97). We know that during thtime ornamental shells from the Pacific coast we

reaching the Valley of Oaxaca (Whalen 1981:103; Witer 1984:204–5), the Ejutla Valley (Feinman and Nichlas 1993), the Mixteca Alta (Winter 1984:206–7), and far northwest as the Cuicatlan Canada (Spencer 198171).

There is little material evidence to reveal the kindof items that the Late Formative–period Monte Albaelite exchanged for coastal products. While not usualconsidered a prestige item, obsidian was often under thdistributional control of the elite and probably movethrough the same prestige-goods exchange network aother exotic commodities (Joyce 1991a:277; Spenc1982:167–74; Whalen 1981:85–87; Zeitlin 1978). LaFormative–period obsidian recovered from the southe

Isthmus and the Lower Verde was dominated by matrial originating from the Orizaba, Guadalupe Victoriand Zaragoza sources in Veracruz-Puebla and thParedon and Otumba sources in the Basin of MexicBoth the Veracruz-Puebla and the Basin of Mexico obsidian could have reached the Oaxaca coast via the Oxaca Valley, although alternative trade routes are alspossible (Joyce et al. 1995; Zeitlin 1978, 1979, 1982).

One item definitely reaching the Lower Rıo Verdfrom the Valley of Oaxaca was ornate gray-paste potte(fig. 3; Joyce 1991a, b). Excavations at the sites of Cer




Fig. 3. Late Formative Rı o Verde grayware imports from the Valley of Oaxaca.

de la Cruz and Rıo Viejo recovered 278 fragments of im-ported Monte Alban grayware serving vessels in con-texts associated with high-status or ritual activities.The most common grayware categories were well-

burnished bowls with fine-line incising that belong totypes G.16 and G.17 in the Valley of Oaxaca ceramictypology (Caso, Bernal, and Acosta 1967). During theTerminal Formative, at the time the Valley Zapotec arethought to have been engaged in their campaign of con-quest, the importation of this Monte Alban graywarenearly ceased (Joyce, Winter, and Mueller 1998:114). Incontrast, the southern Isthmus does not appear to havereceived any Monte Alban export pottery, despite evi-dence that the region was a major supplier of ornamen-tal seashells to the highland center. We do not knowwhat material goods, if any at all, the Zapotec weresending down to the Isthmus in exchange for the prod-ucts they were obtaining.

Despite the lack of pottery imports, by the TerminalFormative in both the Lower Verde (fig. 4) and thesouthern Isthmus (figs. 5, 6), locally made variants ofMonte Alban grayware became one of the dominant re-gional ceramic types (Joyce 1991a; Zeitlin 1979, 1994).The gray pottery manufactured on the coast does not,however, precisely replicate that of the Oaxaca Valley.Rıo Verde ceramic categories such as outleaning/out-flaring-wall bowls with incised interior rim-lines andoccasionally combed bases (Joyce 1991a) and thespouted effigy vessels from the southern Isthmus (Zeit-

lin 1979) are quite similar to Monte Alban types, butmost of the coastal graywares display regionally distinc-tive formal and decorative attributes. Indeed, a studycomparing the stylistic attributes of Monte Alban pot-tery with contemporaneous wares from the MixtecaAlta, the Mixteca Baja, the Cuicatlan Canada, theLower Rıo Verde, and the southern Isthmus indicates

that, with the exception of pottery from the CuicatlanCanada, precise similarities are limited (Joyce 1993b).

In both the Lower Verde region and the southern Isth-mus—and perhaps elsewhere in Oaxaca—grayware ap-pears to assume the role of a local fancy pottery,complementing a vernacular tradition of utilitarianbrownwares in the Lower Verde region and differen-tially fired black-and-white ware on the southern Isth-mus. On the Oaxaca coast at least, there is no evidencethat the intrusion of the Monte Alban–related graywarestyle, not withstanding its occurrence in high frequen-cies, indicates subjugation. Our survey and excavationdata do not indicate such local disruptions in sociopolit-ical organization, agricultural and/or manufacturing

production, religious practice, architectural tradition,or settlement pattern as might be expected had therebeen a Zapotec imperial takeover and direct administra-tion of these coastal Oaxaca regions similar to thatclaimed for the Cuicatlan Canada (Redmond 1983,Spencer 1982), Sola Valley (Balkansky 1997:14), andEjutla (Feinman and Nicholas 1990:231) regions.

comparative perspectives on zapotecimperialism

Comparative data from other early empires in Meso-america and elsewhere support our conclusion that sim-ilarities in ceramic styles are not a reliable indicator of

hinterland subjugation (e.g., Lind 1987:97–98; Schreiber1992:263; Stark 1990). The lack of pottery-based evi-dence for conquest does not, however, lead us to con-clude that the Oaxaca coast remained completely be-yond Zapotec influence. We realize that militaryconquest is notoriously difficult to identify archaeologi-cally, even where its occurrence is confirmed by writtenrecords (Marcus and Flannery 1996:198–99; Zeitlin1993). We cannot assume, however, that because con-quest or colonization has been identified at some Oa-xaca localities all the outlying Zapotec sources of eco-nomic support would have been similarly subjugated.Imperial rulers in other early states pursued a varietyof tactics to control or influence their peripheries (e.g.,

Algaze 1993, Blanton 1996, Schreiber 1992, Sinopoli1994, Stark 1990). These ranged from territorial con-quest and direct administration (as was said to havebeen the fate of the Cuicatlan Canada) to indirect hege-monic control achieved largely through the cooperationof local leaders (as was the stratagem within much ofthe Aztec empire) and asymmetrical alliances with lo-cal elites, often cemented by intermarriage and gift ex-change.

The Aztec example may be particularly instructive,for even with a military potential much more formida-





Fig. 4. Locally made Terminal Formative grayware from the Rı o Verde region.

ble than the Zapotec the Aztecs rarely maintained on-site control over their provinces (Berdan et al. 1996,Hassig 1988). Negotiation and threat were the primarymeans the Aztecs employed to secure the receipt of trib-ute, their main interest in hinterland regions. Warfare,usually in the form of a swift but savage assault, was alast resort. The intent was to teach a lesson to thosewith the temerity to defy their efforts at economic he-gemony. Aztec administrators sometimes were, some-

times were not, permanently posted to regions theyconquered. Normally, as long as the requisite tributewas forthcoming, outlying provinces were allowed tomaintain a degree of local autonomy. Alternatively,where conquest of peripheral regions whose goods weredesired was judged impractical or overly costly, trade re-lationships were maintained.

In a first effort to discern the variables that may havedictated a diversity of tactics on the part of Monte Al-ban in addressing its growing need for exotic goods fromthe coast, let us briefly return to our study of the LowerRıo Verde region and the southern Isthmus. On the ba-sis of current evidence we can acknowledge that in itsrelationships with at least some regions Monte Alban

displayed an element of belligerence. However, practi-cal limitations to the geographic extent and military po-tential of the Monte Alban empire must be recognized.The total Terminal Formative–period Valley of Oaxacapopulation—women, children, the elderly, and the in-firm included—has been estimated at about 40,000 per-sons, of which fewer than 15,000 resided at Monte Al-ban proper (Blanton et al. 1993). Given the relativelymodest pool of people available for military service orcolonization, it seems unlikely that the 50-odd locali-ties represented on the Building J conquest slabs, places

spread out over a 20,000-km2 territory, would have abeen targets of long-term, direct imperial administrtion. Marcus and Flannery recognize the possible exitence of multiple approaches to resource access on thpart of an expansionary Zapotec state (1996:198). Zaptec excursions for the purpose of territorial acquisitiowere probably directed toward a limited number small, weakly organized polities, such as the one in thCuicatlan Canada.

Another Zapotec option for increasing hinterland prductivity might have been through economic pressurby withholding or threatening to withhold goods fwhich the Valley of Oaxaca served as a distributionhub. The Lower Verde, located at the end of a transpotation link emanating from Monte Alban, would havbeen particularly vulnerable to the restriction of supplies moving between coast and highlands. Thus, whethe flow of obsidian and imported pottery to the LowVerde declined during the Terminal Formative (Joy1991a:524–34), we might attribute this loss to Zapoteinterference. Economic pressure of this sort would havbeen less costly than military intervention. By the Teminal Formative the regional center of Rıo Viejo ha

grown to cover at least 225 ha, making it a formidabobstacle for an invading army. Alternatively, interpoliconflict in the regions between the Valley of Oaxacand the coast may have disrupted exchange routes, rsulting in the observed decline in obsidian and importpottery reaching the Lower Verde. A Zapotec incursiointo the Sola Valley, which lies between the Valley Oaxaca and the Lower Verde region, is one such conflithat may have occurred (Balkansky 1997).

A similar Zapotec tactic of economic interference applied to the southern Isthmus would have been less su





Fig. 5. Locally made Late and Terminal Formative grayware from the southern Isthmus of Tehuantepec.

cessful, since that region had a variety of alternativesfor exchange partners. As an indication of these options,we find no evidence for a Terminal Formative decreasein the availability of obsidian on the southern Isthmusdespite the fact that by the end of the period Guadalupe

Victoria, previously the major source of supply, nolonger served the region. The cutoff of Guadalupe Victo-ria obsidian is of particular relevance, since a likelyroute of its travel to the southern Isthmus would havebeen through the Valley of Oaxaca. Replacing Guada-lupe Victoria was an alternative source at Zaragoza(Zeitlin 1978, 1982). The Terminal Formative switch toZaragoza obsidian, which would most likely have beenconveyed to the Tehuantepec region over a lowlandroute running down the Gulf Coast and across the shorttrans-isthmian depression, may reflect a response to a

curtailment of supplies passing through the Valley ofOaxaca.

some concluding thoughts

A model that recognizes greater variability in Monte Al-ban interaction with neighboring regions suggests abroader interpretation of the Building J ‘‘conquestslabs’’ than one that has them all identifying places con-quered or colonized. Some of the toponyms may havereferred to places where the Zapotec were exerting con-trol through alliances backed by military or economicthreat. Others may have indicated regions the Zapotechad raided but had not yet been successful in subduing.Still others may have served as little more than a wishlist. As Marcus has noted, ‘‘Military propaganda made





Fig. 6. Late Formative grayware stirrup-spout vessel from the southern Isthmus of Tehuantepec withupper segment in the form of an animal head,

possibly a monkey. An almost identical grayware

vessel from Monte Alba n is illustrated in Caso,Bernal, and Acosta ( 1967: fig. 90).

up an enormous part of later Mesoamerican writing andclaims of victory were often exaggerated’’ (1992:442).

One might also wonder why the hinterland neigh-bors of Monte Alban would crave pottery of a style as-sociated with a place that intended to exploit them.In reflecting on this seeming incongruity, it must beremembered that by the Terminal Formative MonteAlban–style grayware had been around for at least sev-eral hundred years. When the popularity of this potteryfirst spread beyond the Valley of Oaxaca, relationshipsbetween the Zapotec capital and its hinterland ex-change partners were probably less exploitative, andemulating the flourishing Zapotec center of civilizationwould have afforded a degree of prestige. Under thosecircumstances it is easy to imagine why the graywarehorizon expanded to the full extent of the Oaxaca inter-action sphere and in some cases far beyond.2

2. A recent study grant from the American Museum of NaturalHistory has given one of us (Zeitlin) the opportunity to examine apreviously unanalyzed collection of pottery from the archaeologi-cal site of Paredon. Located near the modern town of Tonala, incoastal Chiapas, Paredon lies over 125 km to the southeast of the

By the Terminal Formative, despite an increasinghostile sociopolitical environment, grayware variantproduced in many localities, may simply have becoma generic fine pottery throughout Oaxaca. Its greabundance and widespread occurrence in sites such aLaguna Zope on the southern Isthmus and in the LowRıo Verde region indicate that its use was not restricte

to higher-ranking members of society. Nor does ibroad geographic distribution throughout Oaxaca offa simple measure of the extent of the Terminal Formtive–period Zapotec empire.

Although the precise nature of Monte Alban’s politcal and economic relationship with its coastal neigbors at the end of the Formative era is still far froclear, there are no current indications of the impositioof the kind of territorial imperialism that has been agued for the Cuicatlan Canada and other outlying rgions of Oaxaca. Given the distance of the southeIsthmus and Lower Rıo Verde regions from Monte Aban, the difficulties of long-distance travel, the size ancomplexity of the societies involved, and the relative

modest scale of the Late/Terminal Formative Zapotepolity, it does not strike us as odd that these two coastregions appear to have remained autonomous durinthe period of reported Zapotec expansionism.

We believe that the precise nature of Late/TerminFormative–period Zapotec interaction with their manneighbors will eventually reveal itself as varied ancomplex. The full complexity and extent of this ecnomically motivated interaction is, however, unliketo be determined by information gleaned from investgations in the Monte Alban homeland. Rather, it wirequire the acquisition and thoughtful analysis of mucmore of those elusive, multifaceted archaeological suvey and excavation data from peripheral places, such

those on the Oaxaca coast, with which the Zapotec interacted. When patterns of similarities and differencin environment, natural resources, distances from thZapotec capital, population dynamics, regional ecoomy, local leadership, social organization, ceremoniaism, and other factors emerge from such studies, wwill have the basis for a better understanding of the vaious ways in which relationships between the Zapotecenter and periphery were structured.

References Cited

algaze, g. 1993. Expansionary dynamics of some early pris-tine states. American Anthropologist 95:304–33.

b a lk a nsk y , a . k . 1 997. Archaeological settlement patterns othe Sola Valley, Oaxaca, Mexico. Mexicon 19(1):12–18.

b a l l , h . g . , a n d d . l . b r o c k i n g t o n . 1978. ‘‘Trade andtravel in prehispanic Oaxaca,’’ in Mesoamerican communication routes and cultural contacts. Edited by T. A. Lee Jr. andC. Navarrete, pp. 107–26. Papers of the New World Archaeoloical Foundation 40.

b e r d a n, f . f . , r . e . b l a n t o n, e . h . b o o n e , m . g .

southern Isthmus and fully 320 km from Monte Alban. The colletion, excavated in 1960 by Matthew Wallrath, contains a sttlingly large percentage of gray pottery closely related to the LatTerminal Formative grayware of Monte Alban.

http://www.jstor.org/action/showImage?doi=10.1086/200029&iName=master.img-011.jpg&w=231&h=332




h o d g e , m . e . s m i t h , a n d e . u m b e r g e r . 1 9 9 6 . Aztec imperial strategies. Washington, D.C.: Dumbarton Oaks Re-search Library and Collection.

blake, m. 1991. ‘‘An emerging Early Formative chiefdom atPaso de la Amada, Chiapas, Mexico,’’ in The formation of com-

plex society in southeastern Mesoamerica. Edited by W. R.Fowler Jr., pp. 27–46. Boca Raton, Fla.: CRC Press.

blanton, r. e. 1978. Monte Alba n: Settlement patterns atthe ancient Zapotec capital. New York: Academic Press.

———. 1996. ‘‘A consideration of causality in the growth of em-pire: A comparative perspective,’’ in Aztec imperial strategies.Edited by F. F. Berdan, R. E. Blanton, E. H. Boone, M. G.Hodge, M. E. Smith, and E. Umberger, pp. 219–25. Washing-ton, D.C.: Dumbarton Oaks Research Library and Collection.

b l a n t o n, r . e . , s . a . k o w a l e w s k i, g . m . f e i n m a n ,a nd l. m . fi nst en. 1 9 9 3 . 2d edition. Ancient Mesoamer-

ica: A comparison of change in three regions. New York: Cam-bridge University Press.

caso, a. 1947. Calendario y escritura de las antiguas culturasde Monte Alban. Obras Completas de Miguel Otho n de Mendi-

za bal 1:116–43.c a s o , a . , i . b e r n a l , a n d j . r . a c o s t a . 1 9 6 7 . La cera -

mica de Monte Alba n. Memorias del Instituto Nacional de An-tropologıa e Historia 13.

f e i n m a n , g . , a n d l . m . n i c h o l a s . 1990. At the marginsof the Monte Alban state: Settlement patterns in the Ejutla Val-ley, Oaxaca, Mexico. Latin American Antiquity 1:216–46.

———. 1993. Shell-ornament production in Ejutla: Implicationsfor highland-coastal interaction in ancient Oaxaca. AncientMesoamerica 4:103–19.

fla n ner y , k . v . 1968. ‘‘The Olmec and the Valley of Oaxaca:A model for interregional interaction in Formative times,’’ inDumbarton Oaks Conference on the Olmec. Edited by E. P.Benson, pp. 119–30. Washington, D.C.: Dumbarton Oaks Re-search Library and Collection.

f l a n n e r y , k . v . , a n d j . s c h o e n w e t t e r . 1970. Climateand man in Formative Oaxaca. Archaeology 23:144–52.

g a r c ı a m o l l , r . , d . w . p a t t e r s o n b r o w n , a n d m .winter. 1986. Monumentos esculto ricos de Monte Alba n.Munich: Verlag C. H. Beck.

gaxiola, m. 1984. Huamelulpan: Un centro urbano de la Mix-teca Alta. Mexico, D.F.: Instituto Nacional de Antropologıa e

Historia.hassig, r. 1988. Aztec warfare: Imperial expansion and politi-cal control. Norman: University of Oklahoma Press.

h i r t h , k . g . Editor. 1984. Trade and exchange in early Meso-america. Albuquerque: University of New Mexico Press.

j oy c e, a . a . 1991a. Formative Period occupation in the LowerRı o Verde Valley, Oaxaca, Mexico: Interregional interactionand social change. Ann Arbor: University Microfilms.

———. 1991b. Formative period social change in the Lower RıoVerde Valley, Oaxaca, Mexico. Latin American Antiquity 2:126–50.

———. 1993a. Interregional interaction and social developmenton the Oaxaca coast. Ancient Mesoamerica 4:67–84.

———. 1993b. The interregional impact of state formation in Oa-xaca. Report submitted on research activities performed as a1992–93 Kalbfleisch Fellow, American Museum of Natural His-tory, New York.

———.1994.

Late Formative community organization and socialcomplexity on the Oaxaca coast. Journal of Field Archaeology 21:147–68.

———. 1998. Monte Alban, Teotihuacan, and the Lower RıoVerde Valley. Paper presented at the 5th Annual Mixtec Gate-way Conference, Las Vegas, Nev.

j o y c e, a . a . , j . m . e l a m , m . d . g l a s c o c k , h . n e f f ,a nd m . w i nt e r . 1995. Exchange implications of obsidiansource analysis from the Lower Rıo Verde Valley, Oaxaca, Mex-ico. Latin American Antiquity 6:3–15.

j oy c e, a . a ., a nd m . c . w i nt er . 1 9 9 6 . Ideology, power,and urban society in pre-Hispanic Oaxaca. current anthro-pology 37:33–86.

j o y c e , a . a . , m . w i n t e r , a n d r . g . m u e l l e r . 1 9 9 8 . Ar-

queologı a de la costa de Oaxaca: Asentamientos del periodoFormativo en el valle del Rı o Verde inferior. Centro INAH Oa-xaca Estudios de Antropologıa e Historia 40.

lind, m. 1987. The sociocultural dimensions of Mixtec ceramics. Vanderbilt University Publications in Anthropology 36.

marcus, j . 1976. ‘‘The iconography of militarism at Monte Al-ban and neighboring sites in the Valley of Oaxaca,’’ in The ori-

gins of religious art and iconography in Post-Classic Meso-america. Edited by H. B. Nicholson, pp. 123–39. Los Angeles:Latin American Center, University of California.

———. 1983. ‘‘The conquest slabs of Building J, Monte Alban,’’in The cloud people: Divergent evolution of the Zapotec andMixtec civilizations. Edited by K. V. Flannery and J. Marcus,pp. 106–8. New York: Academic Press.

———. 1992. Ancient Mesoamerican writing systems: Propa- ganda, myth, and history in four ancient civilizations.Princeton: Princeton University Press.

m a r c us, j . , a nd k . v . fla nner y . 1 9 9 6 . Zapotec civiliza-tion: How urban society evolved in Mexico’s Oaxaca Valley.London: Thames and Hudson.

pa ler m , a ., a nd e. r . w olf. 1 9 5 7 . ‘‘Ecological potentialand cultural development in Mesoamerica,’’ in Studies in hu-man ecology. Washington, D.C.: Anthropological Society ofWashington.

p a s o y t r o n c o s o , f . d e l . Editor. 1981. Relaciones geo- gra ficas de Oaxaca. Mexico, D.F.: Editorial Innovacion.

pi r es- fer r ei r a , j . w . 1 9 7 5 . Formative Mesoamerican ex-change networks with special reference to the Valley of Oa-

xaca. Memoirs of the Museum of Anthropology, University ofMichigan, 7.

redmond, e. m. 1983. A fuego y sangre: Early Zapotec imperi-alism in the Cuicatla n Can ada, Oaxaca. Memoirs of the Mu-seum of Anthropology, University of Michigan, 16.

sc hr e i b er , k . j . 1992. Wari imperialism in Middle HorizonPeru. Anthropological Papers, Museum of Anthropology, Uni-versity of Michigan, 87.

si nop oli , c . m . 1994. ‘‘Political choices and economic strate-gies in the Vijayanagara empire,’’ in The economic anthropol-ogy of the state. Edited by E. M. Brumfiel, pp. 223–42. Mono-graphs in Economic Anthropology 11.

spencer, c. s. 1982. The Cuicatla n Can ada and Monte Al-ba n. New York: Academic Press.

stark, b. l. 1990.The Gulf Coast and the Central Highlandsof Mexico: Alternative models for interaction. Research in Eco-

nomic Anthropology 12:243–85.west, r. c. 1964. ‘‘Surface configuration and associated geol-

ogy of Middle America,’’ in Handbook of Middle American In-dians, vol. 1, Natural environments and early cultures. Editedby R. Wauchope, pp. 33–83. Austin: University of Texas Press.

whalen, m. e. 1981. Excavations at Santo Domingo Tomal-tepec: Evolution of a Formative community in the Valley of Oaxaca, Mexico. (Prehistory and Human Ecology of the Valleyof Oaxaca, 6.) Memoirs of the Museum of Anthropology, Uni-versity of Michigan, 12.

w hi t ec ot t on, j . w . 1 9 9 2 . ‘‘Culture and exchange in Post-classic Oaxaca: A world-system perspective,’’ in Resources,

power, and interregional interaction. Edited by E. M. Schort-man and P. A. Urban, pp. 51–74. New York: Plenum Press.

winter, m. 1984. ‘‘Exchange in Formative highland Oaxaca,’’

inTrade and exchange in early Mesoamerica.

Edited by K. G.Hirth, pp. 179–214. Albuquerque: University of New MexicoPress.

w o r k i n g e r , a . g . , a n d a . a . j o y c e . 1997. The boundaryimplications of San Francisco de Arriba: Preliminary observa-tions. Paper presented at the 61st annual meeting of the Soci-ety for American Archaeology, Nashville, Tenn.

zeitlin, j . f. 1978. ‘‘Changing patterns of resource exploita-tion, settlement distribution, and demography on the southernIsthmus of Tehuantepec, Mexico,’’ in Prehistoric coastaladaptations: The economy and ecology of maritime Middle

America. Edited by B. L. Stark and B. Voorhies, pp. 151–77.New York: Academic Press.

zeitlin, r. n. 1978. ‘‘Long-distance exchange and the growth




of a regional center on the southern Isthmus of Tehuantepec,Mexico,’’ in Prehistoric coastal adaptations: The economy andecology of maritime Middle America. Edited by B. L. Stark andB. Voorhies, pp. 151–81. New York: Academic Press.

———. 1979. Prehistoric long-distance exchange on the southernIsthmus of Tehuantepec, Mexico. Ann Arbor: University Micro-films.

———. 1982. Toward a more comprehensive model of interre-gional commodity distribution: Political variables and prehis-

toric obsidian procurement in Mesoamerica. American Antiq-uity 47:260–75.

———. 1990. The Isthmus and the Valley of Oaxaca: Questionsabout Monte Alban imperialism in the Pacific lowlands. Ameri-can Antiquity 55:250–61.

———. 1993. Pacific coastal Laguna Zope: A regional center inthe Terminal Formative hinterlands of Monte Alban. AncientMesoamerica 4:85–101.

———. 1994. Accounting for the prehistoric long-distance move-ment of goods with a measure of style. World Archaeology 26:208–34.

Mapping Spatial and SocialPermeability1

t h o m a s w i d l o kDepartment of Anthropology, Institut fu r Ethnologie,University of Cologne, 50923 Ko ln, Germany ([email protected]). 11 ix 98

Two trends in recent research seem to me inherentlylinked: a growing awareness of the flexibility of socialgroups and a growing interest in the spatial correlatesof social organization. The preoccupation with space

seems to have grown at least in part out of the difficul-ties that arise when flexible social groups need to be de-scribed, analysed, and compared. Apart from attemptsto develop a largely metaphorical concept of ‘‘socialspace’’ (see Bourdieu 1985), researchers of social organi-zation have started to pay particular attention to settle-ment layouts. Settlement layouts and other materialspatial features may help to anchor and substantiate theanalysis of highly flexible social phenomena. It is oneof the working assumptions of much anthropologicalwork that social order and spatial order are correlated,and indeed a broad correlation has been shown to existcross-culturally (see Kent 1990). However, in manycases there is no precise congruence between social ac-

tion and spatial layouts or architecture because the set-ting, the ‘‘milieu which defines a situation’’ (Rapoport

1. © 1999 by The Wenner-Gren Foundation for Anthropological Re-search. All rights reserved 0011-3204/99/4003-0008$1.00. Re-search in Namibia was carried out between 1990 and 1992 with thehelp of a University of London Studentship and the James A. SwanFund (Pitt-Rivers Museum, Oxford) and between 1992 and 1994with the support of the Cognitive Anthropology Research Groupof the Max-Planck-Institute in Nijmegen. The writing of this reporthas been part of a research project on informal social institutionsfunded by the Deutsche Forschungsgemeinschaft (German Re-search Council) and its programme on the social aspects of globalenvironmental change.

1990:12), in which social action takes place is not dfined exclusively by spatial parameters. It is a majproblem, therefore, to distinguish the aspects of layouand built form that can be expected to be correlatewith social organization from those that are noWhereas any part of social organization may have its rflection in the organization of space, it is particular

promising to investigate the extent to which spatial fetures themselves, once established, encourage or dicourage particular kinds of social practice and social oganization. This requires new methods which msettlements in a way that allows us to see space noonly as a mirror of cultural ideas but also as a means producing such ideas in the disposition and behaviouof social actors. Since layout and built forms do not dtermine all situations and do not influence all sociactors equally across the board, such methods need taccount for the diverging perspectives of these actorThe method of constructing permeability maps oulined here attempts to satisfy both these requirement

ha i om camps

The conventional bird’s-eye view or ground plan !Gai| nas, a Hai om camp in northern Namibia (showin figure 1) was recorded during the dry season of 199and is described in more detail in Barnard and Widlo(1996) and in Widlok (n.d.). Windshields (made branches) and grass huts (made of branches and grasare represented as numbered half-circles. Other shelteand shade roofs which are open on several sides are reresented as numbers between two lines. The doublines surrounding the camp indicate low bushes whichave not been cleared away but which, consisting onof leafless branches, do not form an impenetrab

boundary. Other features of the camp include fireplac(F ), a large green tree (T ), and a large fireplace for coducting the medicine dance (M ). A comparison witwo ideal-type maps recorded among Haiom earlithis century (Fourie 1927:51; Lebzelter 1934:84) showcharacteristic recurring features, namely, the circle-likstructure of the camp with a shady tree at its centre anthe central place for the medicine dance. Howevemaps of this kind do not take us much farther than thiAs I have pointed out elsewhere (Widlok 1994:270), thtwo earlier sources themselves disagree on who occupies which hut in the settlement. Any new settlemenlayout that is collected, such as this one for !Gai| naadds even more unexplained diversity. This diversity

partly due to the fact that the composition of Hai olocal groups is more flexible than any single grounplan could allow for. Not all camps have a hut occupieby unmarried adolescent boys (4), a hut for unmarriegirls (11), one for a pregnant woman (7), and one for thwoman’s husband and children (8) as !Gai| nas had 1991. While Hai om camps may ‘‘typically’’ includhuts of married couples (5, 10, 12) and single/widoweparents or other relatives of these couples (2, 3, 9), tharrangement of these huts may vary, since some hubecome deserted (1) and some shelters made for visitomay at a given point in time be permanently occupie




Fig. 1. Ground plan of !Gai≠ nas.

(6) or simply used as an extra shady place for temporaryuse by occupants of other huts (13). To find patterns insuch diversity requires other methods of mapping. Apermeability map of !Gai| nas (fig. 2) shows the spatialcharacteristics of a Hai om dry-season camp from theperspective of a visitor (). Centrality, open access, and

proximity to the central shady tree are the properties ofthe fireplace for the medicine dance (M). Like (almost)all fireplaces it is accessible from more than one direc-tion and is therefore connected with double lines. Ac-cess to any of the huts in the camp, except for the openshelters, is controlled by the fireplace (F ) in front of it,which any visitor has to pass in order to reach the inte-rior of a hut. In some cases occupants of several hutsshare a fireplace, such as a couple (10), their unmarrieddaughters (11), and the husband’s mother (9); in othercases a family has more than one fireplace (as with huts

Fig. 2. Permeability map of !Gai≠ nas.

7 and 8). All huts, except for a deserted ‘‘fire-less’’ hut(1), at that stage used for utensils only, and the hut ofthe pregnant woman, have the same permeability. Se-niority, gender, and social position (such as late arrivalat the place) do not produce fixed spatial features in theHaiom camp—or, to put it differently, the spatial or-

der does not support a social differentiation accordingto age, sex, or social position. Visitors to the camp arenot spatially restricted as to which hut or fireplace theymay initially approach or where they may sit down fora meal or a visit. However, visitors are guided first to acentral, shady fireplace, where they are prompted (andexpected) to sit down and wait before they move on toone of the other fireplaces or before a resident sitting atone of these fireplaces turns to them and comes to greetthem. This corresponds to the activities carried out atthe various places within the camp. Open access to thecentral tree corresponds to the daylight activities (usu-ally handicrafts or conversation) as well as the night-time activities (medicine dancing and healing) that take

place there. The more restricted access to the individualfireplaces corresponds to the main activities carried outat these places (eating and chatting). Finally, access intoa hut, which contains individual belongings and restingplaces, is possible only via the fireplace in front of it.

The permeability map may be redrawn from the per-spective of any occupant of the camp. However, multi-ple permeability maps are unlike multiple conventionallayout maps. The former include the same spatial ele-ments across maps drawn from different positionswhile the latter typically juxtapose the different spatialfeatures which are particularly relevant for—and cul-






turally elaborated in—the diverging perspectives of in-dividual persons or of groups. Despite the emphasis onspatial structure that exists independent of perspective,permeability maps can also show how a single spatialstructure may have a number of different social impli-cations depending on the way in which occupants arepositioned in a settlement. Occupants of different huts

within a camp share many daylight activities as well asinvolvement in the medicine dance. Occupants of dif-ferent huts with only one fireplace share this place as ahearth where food and information are shared. Occu-pants of a single hut (couples or unmarried adolescents)share much of their time and their belongings with oneanother. In sum, permeability maps capture the form ofspatial connectedness (which does not always meet thebird’s eye) and help to show whether and how spatiallayouts guide social practice.

theoretical foundations of graphic maps

Permeability maps seem more abstract than the ground

plans ordinarily used in ethnography on which they areultimately based. However, despite their naturalisticappearance, most ground plans presented in anthropo-logical writing are ‘‘ideal-type maps,’’ abstracting fromindividual cases or presenting cases as types for the pur-poses of a generalizing description (for examples fromthe region discussed here, see Barnard 1992:207; Silber-bauer 1981:197; Valiente-Noailles 1993:118). They aretherefore also abstract, but the theoretical frameworkon which they rely for the abstraction differs from thatof permeability maps. When the theoretical underpin-nings that lead to ground plans are not clearly spelledout (as, for instance, in Williams 1991:184), it remainsunclear whether we are dealing with ‘‘an ‘ideal’ house,

a typical house, or merely a collage of common featuresnoticed by one anthropologist’’ (Moore 1986:91). Fur-thermore, the same spatial structure may be subject todifferent ideal-type perspectives. A plurality of perspec-tives on settlement structures exists not only amongmen and women (Bourdieu 1976:65) but also amongmembers of different social segments within a group(see Crandall 1996). Adopting a comparative perspectiveeither within a group or between groups, conventionalethnography would juxtapose different ideal-typeground plans which might be made up of quite differentfeatures depending on the different meanings attachedto spatial elements. Permeability maps, by contrast, areeasier to compare because they are always made up of

the same elements. A permeability map may be drawnfrom different positions within or outside the settle-ment in question, and the underlying spatial elementswill remain the same.

Permeability maps are not drawn to depict primarilycultural ideal types. Rather, they incorporate the mate-rial reality of spatial layouts in order to show to whatextent cultural ideas and practices can be traced to spa-tial conditions or at least explained by the way they aresupported by built forms. The analysis of spatial layoutsand built forms not only as a reflection of cultural ideas

but also as producing certain ideas and practices rquires a new strategy of abstracting from the materiinductively acquired. This is how permeability mapcan complement ground plans. They show how dometic space can have a relative permeability and densitfor the people who dwell in it that influence how it occupied, allocated, divided, and used in everyday lif

Permeability maps replace the bird’s-eye view with thperspective of a ‘‘carrier’’ (a visitor or resident) stridinthrough the camp. They distinguish the relation btween points in a settlement by the number of possibaccess routes and the number of the steps required reach one place (or all places) from another (or all oters). This creates a descriptive tool for comparing varitions and changes in settlement layouts. This tool is epecially useful in cases where architecture is nculturally elaborated in discourse or symbolism anmore generally when one is interested in assessing thimpact of built forms and spatial layout on the genertion of cultural ideas and practices. Permeability mapenable ethnographers to investigate the effect of spati

features on the generation of cultural ideas and pratices by concentrating on the fact that points in a settlment differ not only with regard to the cultural meaning attached to their properties but also with regard the nature of those properties, above all the ways iwhich they are connected.

Although permeability maps ultimately rely oground plans, this does not mean that the latter arcloser to the perception of the occupants. Some anthrpologists with an interest in architecture reject permability maps on the ground that they rely on purely fomal aspects which exclude the diverse meanings thare attached to architecture (Susan Kent, personal communication; see Lawrence 1990). This critique also a

plies, however, to a simplistic use of naturalistic grounplans. Taken by themselves, both ways of mapping aonly aids to an analysis of how spatial and non-spatielements feed into one developmental process. Neithcan claim to represent a spatial structure in the wathat it is perceived by its occupants. Occupants neithhave a bird’s-eye view nor is their perception necessaily or exclusively determined by the properties of a pemeability graph. One advantage of permeability maps that their focus on basic spatial relations makes it easito distinguish which cultural ideas and practices aconstituted or at least supported spatially by built formand which ones work without this spatial support.

the method

The basic spatial relations depicted in permeabilimaps have been theoretically conceptualized by Hilliand Hanson (1984) with the help of some specializeterminology that distinguishes places according to thnumber of access routes and their connectedness. Un

permeable (single-access) places are distinguished fromultipermeable (multiple-access) places (fig. 3, a anb). (For simplicity the latter are here depicted as bipemeable places [i.e., with two connecting lines].) Whe




Fig. 3. Basic spatial relations. a, unipermeable;b, multipermeable; c, distributed; d, non-distributed;e, asymmetric.

all places have the same number of access routes (fig. 3,c), space is distributed. Where some places are better-connected than others (fig. 3, d ), space is nondistrib-uted. Where no place controls access to another place—

that is, there is no need to pass through one place inorder to get to a second place without the second place’sbeing at the same time a necessary passage to the first(fig. 3, c)—space is symmetrical. With asymmetricspace, by contrast, one place controls access to a secondplace (fig. 3, e). These properties can be attributed eitherto a spatial structure as a whole or to different parts ofit. Although these are materially given distinctionswhich can be used as modules that make up the map ofa camp, homestead, or village, they depend on a posi-tion taken within or outside the settlement for the cal-culation of distributedness and symmetry. Hillier andHanson (1984:chap. 4) have developed a calculus thatallows one to compare places within a settlement for

distributedness and symmetry and their average valuesfor the whole settlement with those for another settle-ment, taking account of different sizes and levels ofcomplexity. This quantification has proved usefulwhere, as in urban space, settlement layouts are verycomplex and difficult to grasp as a graph. In most caseswith which anthropologists are concerned this will notbe the crucial issue. Rather, what matters here iswhether this mapping technique can be used to charac-terize the places and boundaries encountered in thefield as it can for representing the rooms and corridorsof urban-style housing.

The shelters in hunter-gatherer camps have no solidwalls, but they have spatial features which may be re-

garded as having a very similar function. Of central im-portance are the fireplaces, especially those in front ofhuts or windshields. Even if no fire is burning, thesefireplaces have to be considered separate ‘‘rooms’’ interms of their permeability. Fireplaces are usuallymultipermeable, but a visitor will never—under normalcircumstances—disregard the presence of a fireplaceand go into a hut or windshield without first stopping atthe fireplace in front of it. (Climbing through a windowwhen there is a doorway, in the absence of special cir-cumstances, would be a parallel behaviour with a solid

building.) The walls of Hai om windshields and of huts,which are made of wooden poles, allow people tolook—and certainly to talk and listen—through them,especially in the dry season. They are therefore differentfrom walls between rooms in most urban environ-ments, although the difference is one of degree ratherthan kind given the fact that even in urban settings not

all walls are made of stone (some having windows,hatches, curtains, glass doors, and similar ‘‘intermedi-ate’’ forms). The thinness of walls and the relative prox-imity of huts within the hunter-gatherer camp allow usto neglect absolute distance when drawing permeabilitymaps just as the size of rooms is often neglected onplans of buildings with solid walls. After all, it requiresno special effort to see and hear or to be seen and heardin all parts of the camp. This does not mean thatHaiom have no means to create boundaries in thisopen spatial set-up, either through modes of conversa-tion (calling out or conversing quietly) or through theuse of their bodies (looking at someone, looking away).It is important to note, however, that even these infor-

mally institutionalized practices enter the permeabilitymap only if they manifest themselves in spatial form.The hearth in front of a hut, the hearth in the middleof the camp, the entrance facing the centre or the out-side—these are elements which provide a spatialprompt for a certain routine of visiting and shouldtherefore be included in a permeability map. The notionof an upper or lower part of a settlement which is builton a plane surface, a feature typically included on ideal-type ground plans of villages among hierarchical socie-ties in Oceania (see Toren 1993), should not feature ona permeability map if there is nothing in the actual builtform or spatial layout that makes a difference with re-gard to access, symmetry, and distributedness. Simi-

larly, while it is in principle possible to draw a perme-ability map according to the impression that certainoccupants (or visitors) may have of a settlement, thiswould make it more difficult to see what the actual spa-tial layout adds to the social arrangement of which it ispart. It would also make it less clear what permeabilitymaps have to add to conventional ground plans. That isto say, if cultural salience attached to some parts of thecamp by some of its occupants does not show on a per-meability map, this is in itself an important source ofinformation. It suggests that other features, material orimmaterial, but not built form and spatial layout mayhelp to maintain and re-create the convention in ques-tion.

permeability maps in comparative research

The anthropological use of permeability maps may befurther illustrated by comparing the Hai om dry-seasoncamp of !Gai| nas with Botos, a wet-season camp in thesame area (figs. 4 and 5), and both with a homestead ofthe neighbouring Owambo farmers (figs. 6 and 7). Thewet-season camp is the more permanent camp becauseoccupants return to it every rainy season to work in themillet fields in the middle of which it is built. Haiom





Fig. 4. Ground plan of Botos.

may also return to the same place during the dry seasonbut will set up a new camp some distance away fromthe camp sites of earlier years. Furthermore, the hutsin the more permanent wet-season camp are sturdier,

usually constructed in the style of the neighbouringOwambo agropastoralists, and settlements have fea-tures that are culturally elaborated upon among theBantu-speaking Owambo. The visitors’ fireplace (olu-

pale in the language of the Owambo, usually pro-nounced olupare or orupare by Hai om) is one of thesefeatures. In an Owambo homestead it is a central placewhere visitors are welcomed and served a meal. Withinthe palisade structure of an Owambo homestead theolupale is highly connected (or ‘‘distributed’’) with re-gard to other parts of the settlement. In the permeabil-

ity graph for a chiefly Owambo homestead the distriutedness of this meeting place (9) is far greater than thaverage for places in this spatial structure and particularly than that of the sleeping quarters of the homeste

head (33). This makes it a privileged place for social cotrol—for observing the movements of the inhabitantFrom the visitor’s perspective the meeting place is aready far inside the homestead, which makes it moasymmetric and more significant than other distributeplaces nearer the homestead’s entrance, for instancthe boys’ sleeping huts (57–59). This relative asymmtry increases the non-interchangeability and signicance of this place. To be admitted to the meeting placis not trivial for a visitor to an Owambo homestead, bcause it implies that the visitor is welcomed as a gue





Fig. 5. Permeability map of Botos.

Fig. 6. Ground plan of an Owambo homestead (adapted from Hillier and Hanson 1984 after Loeb 1948 andWalter 1956). P, passage; 1–7, wife’s quarters; 9, meeting place (olupale); 10–11, shelters; 12, first wife’sbedroom; 14–15, male visitors’ quarters; 18, butchery; 22–23, anteroom and sitting place; 25–26, yard andstoreroom; 27–28, girl visitors’ quarters; 29–30, woman visitors’ quarters; 31–35, homestead head’s quarters;38–40, brewery; 44–49, first wife’s quarters; 53, guards’ hut; 55, entrance space; 56, cattle enclosure; 57–59,boys’ sleeping huts; 60–63, cattle enclosures; 65, cooking place; 66, waiting hut; 67–69, 73, second wife’squarters; 71–72, creamery; 74, stamping place.






Fig. 7. Permeability map of an Owambo homestead (after Hillier and Hanson 1984). , ‘‘ carrier’’ (e.g., avisitor); other features as in figure 6. Numbers of passages have not been included, and the space leading to

hut 66 has been categorized as a passage as the original source suggests. Hillier and Hanson depart from the information contained in Loeb’s drawing in that there should be a double line (indicating two access routes) linking the entrance space (55) to the passage that leads to the stamping place (74) and not to that which leads to the boys’ huts (57–59). Furthermore, they have disregarded a second access route to the cooking plac(65) and the routes that lead to the waterhole and the garden, which are situated outside the homestead.

However, this would have led to only minor changes in the map.

who can claim the attention of the occupants. Perme-ability maps show how the olupale can both implyasymmetry between visitor and inhabitant and implydistributedness between the headman and his fellow in-habitants.

Botos, a Hai om wet-season camp, also has its olupare (0), in fact two of them. They are easily recogniz-able by a short row of palisades to the west of a fireplace(as a protection from the hot afternoon sun from thewest). However, here the visitors’ fireplace has no spe-cial features with regard to its position in the layout. As

the permeability map shows, it differs from the otherfireplaces only in that it does not control access to ahut. The olupale is a cultural import that has not yetregistered with the way in which domestic space influ-ences daily practice. Hai om visitors continue to eat atthe hearths of their hosts, and they may occupy a hutand move around the circular structure of the campwithout being irritated by the cultural innovation of theolupale. However, its mere presence provides a spatialprompt which may develop into changing routines ofwelcoming visitors or of distributing occupational

fields within the settlement. To a limited degree this already the case at Botos. The fireplace at the olupais a preferred place for men and for visitors who takpart in local medicine dances. As regards Owambo vistors to the Hai om camp, it seems to prompt the agrpastoralists to keep away from individual hearths anto take a seat at the olupare.

Another element in which dry-season camps and weseason camps differ is that in the latter the thorn fencethat protect millet fields from cattle require the establishment of clearly defined gates (G ) for entering th

fields and less clearly defined passages (P) through thgarden. In contrast to the case of the olupale/oluparthese gates and passages are not culturally elaborateupon (either among Owambo or among Hai om), buthey directly influence the permeability features of thcamp. The boundary of the camp thickens as visitohave to go through a number of predefined steps beforthey approach the huts. Furthermore, activity areas uder shady trees in the field form another stopping placbefore reaching the huts. Within the settlement thhuts are no longer accessible through a central spac





as families create two clusters with fields in between.Unlike the inhabitants of a large Owambo homestead,Haiom do not cultivate a communal field; each coupleor individual has a distinct stretch of land on which togrow millet. Accordingly, Haiom wet-season campsare frequently made up of two or three clusters of huts.The dry- and the wet-season camps compared here are

occupied by the same local group; the group forms a sin-gle circle of huts under the conditions of the dry-seasoncamp and breaks up into clusters under the conditionsof the wet-season camp. There is nothing to suggestthat Hai om consider themselves as living in two cul-tural worlds as they move between the two camps. Nev-ertheless, the spatial pattern as it emerges from the per-meability maps shows that this transition does havestructuring implications.

Haiom settlements show considerable diversity. Anidealized ground plan would therefore be a highly mis-leading representation of Hai om domestic space, sinceone and the same local group may produce and occupya number of different settlements with different ground

plans and different built forms depending on the cir-cumstances. A series of permeability maps helps us tounderstand the diversity as a series of transformationsand to pinpoint the factors that trigger these transfor-mations and their social implications. However, the ex-amples also show that ( pace Hillier and Hanson) it isdifficult to conclude whether changes in layout andbuilt forms lead to lasting changes in social and cogni-tive processes. Although the wider processes of eco-nomic and political incorporation which influence all‘‘Bushman’’ groups also affect Hai om settlement lay-outs, this is by no means an automatism. The changesof layout and architecture in the settlements of the wet(gardening) season do not automatically lead to asym-

metric spatial relations in Hai om dry-season camps.Nevertheless, permeability maps show that the socialchange towards semipermanent settlements which in-clude agriculture has a spatial correlate in increasing ex-ternal closure (asymmetry). This is underlined by ananalysis of permanent Haiom settlements that involvefull-time wage labour (Barnard and Widlok 1996). Apermeability map of a farm settlement occupied byHaiom wage labourers (Widlok 1993, 1994) shows con-siderably more asymmetry and less distributednessthan both !Gai| nas and Botos. This is correlated withincreased closure and increased differentiation betweenlong-term occupants and those who have only recentlysettled at this farm. In this case the architecture fosters

the social change, as intended by the white farm owner,towards more hierarchical relations between long-termand short-term residents and away from an easy accessto fireplaces which could undermine the individualeconomizing of individual families.

Since permeability maps facilitate the comparativeanalysis of spatial layouts, they have played an impor-tant role in some archaeological arguments. Parkingtonand Mills (1991) show that the notion of a simple oscil-lation between hunting and gathering on the one handand livestock holding on the other is difficult to main-

tain when considering layout and built forms. The in-corporation of livestock has similar effects as the incor-poration of millet fields as shown above. In both casesthe permeability map shows an increasing asymmetryand the emergence of a ‘‘deeper,’’ more segregated space(Parkington and Mills 1991:358). Permeability maps aspart of archaeological reconstructions also show that

even under the conditions of colonial displacement ofgroups precolonial kinship arrangements seem to per-sist more strongly and longer than previously assumed(Kinahan 1996). A settlement layout, ‘‘society’s largestand most complex artefact’’ (Parkington and Mills 1991:365), is restricted in its flexibility. Even in a social set-ting like that of a hunter-gatherer society, governed byflexibility and informality and not by elaborate rules,spatial layouts nevertheless promote some social rela-tions while they inhibit others (see Widlok n.d.).

conclusion

I have sketched the problem of an incomplete correla-tion between spatial and social forms with the help ofan analysis of two Hai om settlements. The analysis ofthe social implications of architecture has recently as-sumed importance, and this requires innovation in theway in which ethnographers map settlements. Themethod of constructing permeability maps outlinedhere was first developed by architects, who use thesemaps for urban planning and who have suggested thatthey might be useful for cross-cultural comparison.This usefulness has been shown in archaeological re-search, where permeability maps are more readily ac-cepted than in anthropology (Kinahan 1996, Parkingtonand Mills 1991). This may be because archaeologists

lack data for the types of elements of meaningful infor-mation (‘‘semi-fixed-feature elements,’’ such as furni-ture, and ‘‘non-fixed-feature elements,’’ especially peo-ple and their activities [see Rapoport 1990:13]) thatpermeability maps disregard in order to gain a better un-derstanding of the effects of ‘‘fixed-feature elements’’such as walls and passages. In ethnography, where ac-cess to social practice and cultural ideas is more readilyavailable and material objects can be documented incontext, permeability mapping can be but one elementof the analysis. It complements the drawing of conven-tional ground plans of settlements but itself needs tobe complemented by—possibly unconventional—newmethods of documenting social practice.

References Cited

barnard, alan. 1992. Hunters and herders of southern Af- rica. Cambridge: Cambridge University Press.

b a r n a r d , a l a n , a n d t h o m a s w i d l o k . 1996. ‘‘Nharo andHaiom settlement patterns in comparative perspective,’’ inCultural diversity among twentieth century foragers: An Afri-can perspective. Edited by S. Kent, pp. 87–107. Cambridge:Cambridge University Press.

b our d i eu, pi er r e. 1 976. Entwurf einer Theorie der Praxis.Frankfurt: Suhrkamp.




———. 1985. The social space and the genesis of groups. Theory and Society 14:723–44.

c r a nd a ll, da v i d. 1996. Female over male or left over right:Solving a classificatory puzzle among the Ovahimba. Africa66:327–48.

fourie, l. 1927. Preliminary notes on certain customs of theHai -om Bushmen. Journal of the S.W.A. Scientific Society 1(1925/26):49–63.

hi lli er , b . , a nd j . ha nson. 1 9 8 4 . The social logic of space. Cambridge: Cambridge University Press.

kent, susan. 1990. ‘‘A cross-cultural study of segmentation,architecture, and the use of space,’’ in Domestic architectureand the use of space. Edited by S. Kent, pp. 127–52. Cam-bridge: Cambridge University Press.

kinahan, john. 1996. The archaeology of social rank amongnomadic pastoralists in southern Namibia. African Archaeolog-

ical Review 13:224–46.la w r enc e, r oder i c k . 1 9 9 0 . ‘‘Public collective and private

space: A study of urban housing in Switzerland,’’ in Domesticarchitecture and the use of space. Edited by S. Kent, pp. 73–91. Cambridge: Cambridge University Press.

leb z elt e r , v i k t o r . 1934. Eingeborenenkulturen von Sud-westafrika: Die Buschma nner. Leipzig: Hiersemann.

loeb, e. 1948. Transition rites of the Kuanyama Ambo. AfricanStudies 7:71–84.

m oor e, henr i et t a . 1 986. Space, text, and gender. Cam-bridge: Cambridge University Press.

p a r k i n g t o n , j o h n , a n d g l e n m i l l s . 1991. ‘‘From spaceto place: The architecture and social organization of Southern

African mobile communities,’’ in The ethnoarchaeology of mbile people. Edited by C. Gamble, pp. 355–70. Freetown Press

rapoport, amos. 1990. ‘‘Systems of activities and systems settings,’’ in Domestic architecture and the use of space. Ed-ited by S. Kent, pp. 9–20. Cambridge: Cambridge UniversityPress.

si lb er b a uer , geor ge. 1 9 8 1 . Hunter and habitat in the Cetral Kalahari Desert. Cambridge: Cambridge University Press

t or e n, c hr i s t i na . 1993. ‘‘Sign to symbol, symbol as sign:Cognitive aspects of a social process,’’ in Cognitive aspects o

religious symbolism. Edited by P. Boyer, pp. 147–64. Cam-bridge: Cambridge University Press.

v a li ent e- noa i lles, c a r los. 1 9 9 3 . The Kua: Life and souof the Central Kalahari Bushmen. Rotterdam: Balkema.

walton, j . 1956. African villages. Pretoria: van Shaik.w i dlo k , t hom a s. 1 993. Everyday space: Cognition and soci

practice. Poster presented at ‘‘Multiple Worlds: A Confer-ence of Spatial Representations,’’ Cognitive AnthropologyResearch Group, Nijmegen, The Netherlands, 29 November–December.

———. 1994. The social relationships of changing Hai omhunter-gatherers in northern Namibia, 1990–1994. Ph.D. dissUniversity of London, London, England.

———. n.d. Living on Mangetti: Hai om ‘‘Bushmen’’ autonomyand Namibian independence. Oxford: Oxford University PresIn press.

w i lli a m s, fr i eda - nela . 1 9 9 1 . Precolonial communities ofsouthwestern Africa. Windhoek: National Archives.

Documents

The Zapotec‐Imperialism Argument Insights from the Oaxaca Coast