Archaeological Approaches to Market Exchange in Ancient Societies 9781607320296, 9781607320289

Citation preview

Archaeological Approaches to Market Exchange in Ancient Societies

Archaeological Approaches to

Market Exchange in Ancient Societies

edited by

Christopher P. Garraty and Barbara L. Stark

University Press of Colorado

© 2010 by the University Press of Colorado Published by the University Press of Colorado 5589 Arapahoe Avenue, Suite 206C Boulder, Colorado 80303 All rights reserved Printed in the United States of America The University Press of Colorado is a proud member of the Association of American University Presses. The University Press of Colorado is a cooperative publishing enterprise supported, in part, by Adams State College, Colorado State University, Fort Lewis College, Mesa State College, Metropolitan State College of Denver, University of Colorado, University of Northern Colorado, and Western State College of Colorado. The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials. ANSI Z39.481992 Library of Congress Cataloging-in-Publication Data Archaeological approaches to market exchange in ancient societies / edited by Christopher P. Garraty and Barbara L. Stark. p. cm. Includes bibliographical references and index. ISBN 978-1-60732-028-9 (cloth : alk. paper) 1. Commerce, Prehistoric. 2. Economics, Prehistoric. 3. Indians—Commerce. 4. Markets—History. 5. Exchange—History. 6. Civilization, Ancient. I. Garraty, Christopher P., 1971– II. Stark, Barbara L. GN799.C45A74 2010 381—dc22 2010006823 Design by Daniel Pratt 19 18 17 16 15 14 13 12 11 10

 1 0 9 8 7 6 5 4 3 2 1

The editors, respectively, dedicate this volume to Raymond I. Garraty (1934–2009) and Martha I. Schell (1917–2009) in memoriam.

Contents

List of Contributors  |  ix List of Figures  |  xi List of Tables  |  xiii Preface  |  Christopher P. Garraty and Barbara L. Stark  |  xv Section One: Theoretical and Methodological Overviews

Chapter 1  |  Christopher P. Garraty  |  3 Investigating Market Exchange in Ancient Societies: A Theoretical Review Chapter 2  |  Barbara L. Stark and Christopher P. Garraty  |  33 Detecting Marketplace Exchange in Archaeology: A Methodological Review Section Two: Case Studies

Chapter 3  |  David R. Abbott  |  61 The Rise and Demise of Marketplace Exchange among the Prehistoric Hohokam of Arizona Chapter 4  |  Gary M. Feinman and Linda M. Nicholas  |  85 A Multiscalar Perspective on Market Exchange in the Classic-Period Valley of Oaxaca vii

Contents

Chapter 5  |  Barbara L. Stark and Alanna Ossa  |  99 Origins and Development of Mesoamerican Marketplaces: Evidence from South-Central Veracruz, Mexico Chapter 6  |  Geoffrey E. Braswell  |  127 The Rise and Fall of Market Exchange: A Dynamic Approach to Ancient Maya Economy Chapter 7  |  Jeffrey B. Fleisher  |  141 Housing the Market: Swahili Merchants and Regional Marketing on the East African Coast, Seventh to Sixteenth Centuries AD Chapter 8  |  Michael E. Smith  |  161 Regional and Local Market Systems in Aztec-Period Morelos Section Three: Comparative Contributions

Chapter 9  |  Charles Stanish  |  185 Labor Taxes, Market Systems, and Urbanization in the Prehispanic Andes: A Comparative Perspective Chapter 10  |  Richard E. Blanton and Lane F. Fargher  |  207 Evaluating Causal Factors in Market Development in Premodern States: A Comparative Study, with Critical Comments on the History of Ideas about Markets Chapter 11  |  Kenneth G. Hirth  |  227 Finding the Mark in the Marketplace: The Organization, Development, and Archaeological Identification of Market Systems References Cited  |  249 Index  |  313

viii

Contributors

David R. Abbott, School of Human Evolution and Social Change, Arizona State University, Tempe, Arizona 85287 ([email protected]) Richard E. Blanton, Department of Anthropology, Purdue University, West

Lafayette, Indiana 47907 ([email protected])

Geoffrey E. Braswell, Department of Anthropology, University of California, San

Diego, MC 0532, 9500 Gilman Drive, La Jolla, California 92093-0532 (gbraswell@ ucsd.edu) Lane F. Fargher, Department of Anthropology, Purdue University, West Lafayette,

Indiana 47907, and Departmento de Ecología Humana, Centro de Investigación y de Estudios Avanzados (Unidad Mérida), Mérida, Yucatán, Mexico, 97130 ([email protected])

Gary M. Feinman, Department of Anthropology, Field Museum, 1400 South Lake Shore Drive, Chicago, Illinois 60605 ([email protected]) Jeffrey B. Fleisher, Department of Anthropology, 582 Sewell Hall, Rice University, Houston, Texas 77251 ([email protected]) Christopher P. Garraty, School of Human Evolution and Social Change, Arizona

State University, Tempe, Arizona 85287, and Statistical Research, Inc., 6099 E. Speedway Blvd., Tucson, Arizona 85712 ([email protected])

ix

Contributors

Kenneth G. Hirth, Department of Anthropology, 409 Carpenter Building, Penn­ sylvania State University, University Park, Pennsylvania 16802 ([email protected]) Linda M. Nicholas, Department of Anthropology, Field Museum, 1400 South Lake

Shore Drive, Chicago, Illinois 60605 ([email protected])

Alanna Ossa, School of Human Evolution and Social Change, Arizona State University, Tempe, Arizona, 85287 ([email protected]) Michael E. Smith, School of Human Evolution and Social Change, Arizona State University, Tempe, Arizona 85287 ([email protected]) Charles Stanish, Department of Anthropology, University of California at Los Angeles, Los Angeles, California 90095 ([email protected]) Barbara L. Stark, School of Human Evolution and Social Change, Arizona State

University, Tempe, Arizona, 85287 ([email protected])



Figures

3.1. Ballcourt locations in southern and central Arizona  |  62 3.2. Canal systems and ballcourt villages in the Phoenix Basin  |  63 3.3. Zones of sand and rock types in the lower Salt River valley  |  65 3.4. Site map for Palo Verde Ruin  |  70 4.1. Valley of Oaxaca, Mexico  |  87 5.1. The study area in Mesoamerica  |  100 5.2. Western lower Papaloapan basin  |  101 5.3. Distance rings around the center of Cerro de las Mesas, Early Classic period  |  112 5.4. Two methods analyzing the distribution of blades from Early Classic Cerro de las Mesas  |  118 5.5. Two methods analyzing the distribution of blades from Early Classic Nopiloa  |  119 5.6. Two methods analyzing the distribution of blades from Early Classic Azuzules  |  120 5.7. Two methods analyzing the distribution of blades from Late Classic Nopiloa  |  121 xi

Figures

5.8. Two methods analyzing the distribution of blades from Late Classic Azuzules  |  122 5.9. Two methods analyzing the distribution of blades from Middle Postclassic Sauce  |  123 5.10. Two methods analyzing the distribution of blades from Late Postclassic Callejón del Horno  |  124 6.1. Maya sites and regions discussed in this chapter  |  129 6.2. Carol Smith’s ideal spatial models of different kinds of exchange  |  131 6.3. Middle Preclassic sites of the eastern Kaqchikel highlands  |  134 6.4. Economy and average polity size in the Maya area  |  139 7.1. Eastern African coast  |  143 7.2. Northern Pemba Island  |  150 8.1. Locations of Morelos and of sites discussed in this chapter  |  163 8.2. Decorated ceramic types used in the analysis  |  169 8.3. Cluster analysis dendrograms of ceramic similarity among sites  |  172 8.4. Schematic market model for the Middle Postclassic period (AD 1100–1300)  |  173 8.5. Schematic market model for the Late Postclassic A period (AD 1300–1430)  |  175 8.6. Schematic market model for the Late Postclassic B period (AD 1430–1520+)  |  175 8.7. Changing frequencies of ceramic imports at Postclassic sites  |  179 8.8. Changing values of a wealth index for six Postclassic sites with ceramic collections from more than one period  |  180 9.1. The Andes showing Inca provinces and names mentioned in the text  |  186

xii

Tables

3.1. Hohokam Chronology in the Phoenix Basin  |  73 4.1. Multiscalar Indicators of Market Exchange in the Classic-Period Valley of Oaxaca  |  93 5.1. Obsidian Distributions from Centers per Period  |  113 6.1. Kenneth G. Hirth’s (1998) Archaeological Correlates for Different Types of Exchange  |  132 8.1. Frequency of Imported Ceramics at Sites Discussed in This Chapter  |  167 8.2. Ceramic Types Used in Cluster Analyses  |  170 8.3. Ceramic Exchange between Yautepec and Western Morelos  |  178 9.1. Estimated Area and Populations of Selected Urban Centers in the Premodern World  |  198

xiii

Preface

From the serfs of the Middle Ages sprang the chartered burghers of the earliest towns. From these burgesses the first elements of the bourgeoisie were developed. The discovery of America, the rounding of the Cape, opened up fresh ground for the rising bourgeoisie. The East-Indian and Chinese markets, the colonisation of America, trade with the colonies, the increase in the means of exchange and in commodities generally, gave to commerce, to navigation, to industry, an impulse never before known, and thereby, to the revolutionary element in the tottering feudal society, a rapid development. . . . Large-scale industry has established the world market, for which the discovery of America paved the way. This market has given an immense development to commerce, to navigation, to communication by land. This development has, in turn, reacted on the extension of industry; and in proportion as industry, commerce, navigation, railways extended, in the same proportion the bourgeoisie developed, increased its capital, and pushed into the background every class handed down from the Middle Ages. . . . [Capitalist market expansion] has accomplished wonders far surpassing Egyptian pyramids, Roman aqueducts, and Gothic cathedrals; it has conducted expeditions that put in the shade all former exoduses of nations and crusades. Karl Marx and Friedrich Engels, The Communist Manifesto (1959 [1848]:15–16)

As Marx and Engels were astutely aware, the expansion and consolidation of market systems coupled with the disintegration of long-standing trade barriers is the fundamental defining process of global history over the past 500 years. As they noted, the xv

Preface

last several centuries of market development have been a central impetus for advances in technology, global population movement and growth, as well as polity development, interaction, and conflict. Yet, as we now know 160 years later, Marx and Engels saw only the tip of the iceberg. They accurately predicted but never witnessed the full extent of market globalization and multinational corporations. One need not be a card-carrying Wallersteinian to agree that few individuals (if any), no matter how remote or removed from the mass of market centers dotting the earth, live outside the purview of the modern global market system. Marx and Engels started their historical narrative in the European Middle Ages, but the market exchange has a longer history distinct from the origins of capitalism. But how far back? Do we start the account in Ancient Greece and Rome or the Middle East, as some scholars have suggested (see discussion in Morris and Manning 2005a:131–132)? Is market development deeper-seated and more geographically widespread than the Mediterranean and Eurasian world? Is market exchange an innate component of human behavior, as Adam Smith and others believed? Or is it far more variable? If so, can we gain a better perspective about the nature and history of the modern market system by comparing market developments in other, non-Western societies? As explained in Chapter 1, we believe we can. Given the weight and importance of these questions and their implications, we would expect that archaeologists would have made great strides in the investigation of early market exchange as a central focus of their research, as they have for other seminal processes in human history, such as the transition to modern humans or the origins of farming. An outsider would assume that archaeologists had long since developed a sophisticated toolkit for studying market exchange and amassed an impressive corpus of interpretations about premodern market development, especially in times and places that predate written records. This has not been the case, however. Much of the problem stems from the difficult task archaeologists face in finding unequivocal evidence for market development and exchange, as several contributors to this volume point out. In the archaeological record, market exchange can look like many other things. How do we distinguish movements of goods resulting from market exchange from movements related to gift giving, government handouts, or other forms of interpersonal or institutional exchange? A strong methodological focus on detecting market exchange in archaeology is one element of this book. Few volumes have tackled archaeological approaches to market exchange, save for Richard Hodges (1988), who focused on applications of central-place theory and derivative models to archaeological data. In the 1970s and 1980s influential volumes in archaeology addressed modes of exchange and distributional models but without a specific focus on market institutions. Markets and marketing were the subjects of a book edited by Stuart Plattner (1985, ed.) a quarter century ago that contains two contributions by archaeologists but mainly draws upon ethnographic and historical research. Few archaeologists have attempted to demonstrate the existence of marketplaces or market systems empirically, and often researchers either rely on documentary xvi

Preface

sources or simply assume their presence, with few insights regarding market structure or diachronic development. Richard Wilk (1998) decries many archaeologists’ tendency to view the problem of premodern market development through the lens of a simple “market/no-market dichotomy.” Our volume is the first to address the development, change, and organizational complexity of markets before capitalism from an archaeological perspective. The development and importance of markets in ancient societies are especially sensitive topics among modern scholars in light of the late–twentieth-century Cold War between the capitalist United States and Western Europe and the communist Soviet Union. Consequently, scholarly debate on this issue in the mid- and late twentieth century was frequently couched in what Wilk (1998:469) refers to as “polarizing rhetoric” about the role of markets in premodern societies. No doubt, many scholars of ancient markets have projected their views concerning the proper social role of modern markets into the premodern past. For example, Bruce Bartlett (1994), an advocate of free-market policies formerly at the National Center for Policy Analysis, attributes the downfall of the Roman empire to “excessive taxation” and “over-regulation” of the market by the Roman government. To be sure, as many philosophers of science have pointed out, all observation is theory-laden, but some observations are more blatantly so than others. The contributors to this volume avoid political advocacy and carefully evaluate archaeological evidence for market exchange and development in its own right. Accepting the importance of market activity in some premodern societies does not make one a champion of liberal market policies. Nor does acknowledging state control over the market in some ancient societies constitute opposition to modern privatization and free-market exchange. We began with a symposium at the 72nd Annual Meeting of the Society for American Archaeology (SAA) in Austin during April 2007 to assemble a diverse group of archaeologists to address the archaeological development and evolution of market exchange in ancient societies. We aimed for scholars working in a variety of world areas but encountered the highest level of interest among Mesoamericanists, perhaps attributable to the recent increased research on ancient markets there. Independently, another symposium at the same meeting dealt with methods of detecting marketplace exchange in the Maya region. We added some contributors following the SAA symposium to broaden the geographic coverage of the volume. Our volume draws upon Mesoamerican research (Braswell, Feinman and Nicholas, Smith, Stark and Ossa), the Hohokam in North America (Abbott), and Swahili East Africa (Fleisher), as well as on comparative work (Stanish, Blanton and Fargher). Kenneth Hirth served as the symposium discussant and continues in this capacity for the concluding chapter. We added a case that shows the benefits of examining archaeological data when markets are known from documents (Smith) and a study of a non-market system, the Andes (Stanish), to include a “negative” case that improves our understanding of how varied conditions affect market development. xvii

Preface

Our goal with this volume is to generate interest in premodern market developments among archaeologists as well as scholars in related disciplines. The volume hinges on what we see as four principal research question or themes: (1) how markets and market systems articulate with other sectors of ancient economies; (2) how to detect market exchange archaeologically; (3) how premodern markets intersected and articulated with non-market institutions, including government; and (4) how premodern markets developed and changed over the long term. This volume and the symposium that preceded it were largely inspired by Hirth’s (1998) proposed method for archaeological assessment of market exchange, which has awakened a renewed interest in market issues. It capitalizes on Hirth’s challenge with a combination of methodological and case study contributions and continues the active dialogue about how political and social institutions intersected with premodern market systems. The emphasis is on market activities as a dynamic feature in ancient societies. We extend warm thanks to a number of individuals who generously provided their thoughts and input. We first thank Darrin Pratt, director and acquiring editor at the University Press of Colorado, for his interest in our project. We also thank John Clark and an anonymous reviewer for their detailed and thoughtful comments on an earlier draft of this volume. Conversations and commentary among contributors and colleagues have been important in mutually sharpening our efforts, and we particularly thank David Abbott, Richard Blanton, Gary Feinman, Kenneth Hirth, Robert Hunt, and Michael Smith in this regard. In addition, Michael Smith, Gary Feinman, George Cowgill, and Timothy Earle kindly provided comments and suggestions on Garraty’s early prospectus for the SAA symposium, which he originally conceived for a proposal for the Southern Illinois University Visiting Scholar Conference. Although not selected for this award, that proposal set the stage for the subsequent SAA symposium and this volume. The School of Human Evolution and Social Change at Arizona State University provided support for the volume. Christopher P. Garraty Barbara L. Stark July 2009

xviii

Archaeological Approaches to Market Exchange in Ancient Societies

Section One

Theoretical and Methodological Overviews

Chapter One

Investigating Market Exchange in Ancient Societies: A Theoretical Review

Christopher P. Garraty

This volume addresses the ways archaeologists can investigate market exchange and how it developed (or not), declined, and changed in selected times and places in premodern societies. It includes three sections: (1) two introductory chapters that review ideas and methods, beginning with this chapter; (2) six case studies that address diachronic issues of how market systems develop, change, or collapse; and (3) three chapters with a broader, more comparative scope, including the final overview chapter by Kenneth Hirth. The intent of this volume is to encourage a comparative, cross-cultural dialogue and generate new approaches for studying premodern market exchange. As Leah Minc (2006:82) pointed out, archaeologists on the whole have devoted little attention to early markets compared with other issues pertaining to premodern economies, such as craft specialization, domestic consumption, and elite finance. Sole-authored or edited volumes about ancient economies with market systems have focused on specific regions, such as the Mediterranean (e.g., Greene 1986; Manning and Morris 2005; G. Storey 2004) and parts of Mesoamerica (e.g., Masson and Friedel, eds. 2002; Smith and Berdan, eds. 2003), but few archaeological volumes have focused on premodern markets and marketplace development in general (but see Hodges 1988). What is the basis of this neglect? One reason is the legacy of previous generations of scholars who undersold the importance of markets and marketplace exchange in premodern economies, no doubt in part because of the lasting influence and eloquence of scholarly heavyweights such as Karl Polanyi (1957, 2001b [1944]) and 

C h rist o p h er P . G arrat y

Moses Finley (1999 [1973]), both of whom strongly argued against the development of premodern markets. Another reason is that archaeologists have long been impeded by the difficulty of detecting unequivocal material evidence for marketplaces and market systems (Feinman and Nicholas, Chapter 4; Stark and Garraty, Chapter 2), prompting many archaeologists to adopt an “out of sight, out of mind” attitude toward the topic. Fortunately, several archaeologists have begun to make headway in this direction (Dahlin et al. 2007; Hirth 1998; Minc 2006, 2009), including the contributors in this book. The research in this volume hinges on four principal research questions or themes: (1) how do markets articulate with other sectors of ancient economies; (2) how do we recognize market exchange archaeologically; (3) what are the relationships among markets, governments, and other social institutions or groups; and (4) how and under what conditions do market systems develop and change? The editors posed these questions to each of the contributors and asked that they tailor their discussions to them. Much of this chapter is devoted to exploring these themes. This chapter and the following one by Barbara Stark and me introduce, respectively, the theoretical and methodological dimensions of studying premodern markets. I focus here on definitions and conceptual issues and defer the methodological discussion to Chapter 2. In the first section of this chapter, I define key terms and concepts for this volume (although some contributors employ somewhat different definitions), including the elusive concepts of market, marketplace, and market exchange. I then present a brief, theoretically oriented historical sketch of research on the development of premodern markets. The third section is devoted to the four themes and to framing key points of debate for each. A final section explains the geographic scope of the volume and explains why we chose to focus on areas in which premodern market development has less frequently been a subject of detailed study. Key Terms and Concepts

The concepts of market exchange, marketplace, and market system are difficult to define and heavily bound up in colloquial usage and perceptions (Hodges 1988). As sociologist John Lie (1997:342; see also Pryor 1977:31) has explained, even economists have yet to formulate a coherent definition for the market concept. In the end, social scientists must accept that some terms are historically contingent “moving targets” that require some amount of arbitrary acceptance of key defining components. Otherwise, scholars could get caught up in a debate over semantics rather than substance. That said, I propose to make some headway toward coherent definitions suitable for the chapters in this volume. The challenge is to develop a definition that, on the one hand, is not so broad that it encompasses virtually any exchange situation and, on the other hand, is not so narrow that it equates markets with modern capitalism. Polanyi (2001a [1957]:34), whose ideas are discussed in a subsequent section, provided a narrow definition of the 

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

latter sort: “All goods and services, including the use of land, labor and capital, are available for purchase in markets and have, therefore, a price.” In a capitalist system, land and labor are exchanged as market commodities, but this was not widespread in premodern societies with market institutions (M. E. Smith 2004:78–79). Polanyi (2001a [1957], 2001b [1944]) argued against the substantial development of market exchange prior to the advent of capitalism; his argument would be correct, if one were to adhere strictly his definition. As Hirth (Chapter 11) points out, however, adherence to Polanyi’s definition also would mean that in many world areas, markets developed only within the last century or so. A more appropriate definition should be broader, more flexible, and able to accommodate various levels or degrees of market integration, including situations in which land and labor exchange were not integrated into market channels. One of the more taxing obstacles in preparing this volume concerned the appropriate use of the terms “market” and “marketplace.” Colloquially, these terms are frequently used interchangeably: “market” refers to both a physical marketplace and the broader and more abstract economic institution. It is therefore vital to clarify these concepts. I define the terms “market exchange” and “markets” as structural and behavioral concepts; the terms “marketplace” and “market system” refer to the physical implications of those behaviors. Note, however, that not all of the contributors to this volume agree with my definitions. For example, Charles Stanish (Chapter 9) tends to define market exchange more narrowly than I do, whereas David Abbott (Chapter 3) employs a rather broad definition more akin to mine. We (the editors) view these slight differences as intellectually healthy and do not believe they will confuse and mislead readers or undermine the coherency of the volume. Market Exchange and Markets

In a frequently cited definition, Frederic Pryor (1977:437, see also 31–33, 104– 108) describes market exchange as “exchange transactions where the economic forces of supply and demand are highly visible.” By “highly visible” he means that “important changes in the relative prices, the quantities of goods offered or sought, or the quality of the goods can be easily traced to changes in supply and demand forces” (Pryor 1977:104). Pryor further defines price as a form of reckoning the equivalency or value of exchanged items based on supply-demand considerations. Pryor’s definition focuses on the economic dimension, but market transactions also presuppose social relationships among the parties to an exchange. The dissemination of supply-demand information may be quite complex and reactive to a variety of social matters, including prevailing notions of value and fairness, word of mouth among marketplace patrons, interpersonal (or intergroup) bargaining behavior, relationships among parties to an exchange (e.g., Uzzi 1997), and formal price setting by sellers, merchants, guilds, and trade groups or by governing and civic officials (Block and Evans 2005). Word of mouth was probably particularly vital for conveying price 

C h rist o p h er P . G arrat y

information in premodern societies with limited communication technologies. Pryor’s definition should be slightly amended to acknowledge the fundamental importance of these and other social mechanisms on market price formation (Swedberg and Granovetter 2001:13). The proposed definition of market exchange is admittedly broad. Market ex­change, by this definition, probably occurred widely in the past and long predated the advent of formal marketplaces or market systems (Hirth, Chapter 11). If so, it is fair to question the relevance of market exchange for archaeologists and historians. As isolated occurrences, dyadic market exchanges hold little interest and, when viewed against a backdrop of reciprocal exchanges, are archaeologically invisible in any case. What concerns the contributors to this volume is not market exchange per se but rather the scalar increase in market participation that prompts development of a system of rules, norms, and a physical and legal infrastructure. Market exchange is worthy of study—and archaeologically visible—once it becomes socially institutionalized as a common and prevalent practice. As Polanyi (2001a [1957]:36–37) makes clear, economic institutions are more than simply “aggregates of personal behaviors” but are embedded in competitive sociopolitical structures (Fligstein 1996) or “economic fields” (Bourdieu 2005 [2000]). Based on this perspective, I define markets as institutions predicated on the principles of market exchange of alienable commodities. The concept refers not only to the system of economic exchange and provisioning but also to the social and political contexts of those exchanges (Bestor 2001; Plattner 1989b). The social context pertains to the networks of relationships involved in market exchanges and establishment of prices. Mark Granovetter (1985) distinguishes between two forms of economic transactions: embedded (personalized, lasting) and atomized (impersonal, fleeting). Classical and neoclassical economic theory builds on the assumption of atomized transactions among rational individuals, but market exchanges in the ancient world frequently hinged on long-term interpersonal (“embedded”) relationships, which may dampen or distort the price-making effects of supply and demand. As Hirth (Chapter 11) makes clear, however, acknowledging the social embeddedness of markets does not negate the fundamental importance of self-interested economic behavior in market exchange (see also Plattner 1989a; Stanish, Chapter 9). Social concerns may inform many market decisions, but within this social framework market buyers and sellers nevertheless may seek to maximize gains or, at a minimum, “satisfice” needs and minimize risk. Beyond social networks, markets are also embedded in larger institutions, such as governments, guilds, or religious institutions (Block 1994; Block and Evans 2005; Bourdieu 2005 [2000]; Fligstein 1996). Markets and marketplace exchange implicate webs of power and rank among actors and groups engaging in transactions and take place in the context of governing and civil authorities. Central to this discussion is how people in premodern institutions and governments chose to handle burgeoning markets, if at all. 

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

The social and institutional embeddedness of markets is where my definition diverges from Polanyi’s (2001a [1957]), who took atomized economic models—disembedded from social context—as emblematic of market exchange and juxtaposed them with what he saw as premodern, non-market forms of exchange embedded in regulative social institutions. A major drawback of Polanyi’s characterization is that he did not apply his embedded perspective to the study of markets (see also Davis 1992:7; Lie 1991). He implicitly accepted classical/neoclassical economic characterizations of markets as asocial institutions and, in the end, “fail[ed] to challenge the market concept itself ” (Lie 1991:223). Market Institutionalization

By viewing premodern market development as a process of institutionalization, it logically follows that the nascent development of market institutions entailed changes in the social and political “rules” of economic exchange, both legal and normative. Markets presuppose a sufficient number of participating buyers and sellers—a scalar element of participation—to underwrite system development. They also presuppose that some (but not all) goods are available for circulation; that is, use rights to the products of one’s labor can be transferred to buyers without undermining existing social norms or practices (Polanyi 2001a [1957]), implying development of property rights and socially acceptable transfers of those rights (i.e., rights of alienation; e.g., Clark 1995; Flad 2007; Y. Li 2007). Allocation systems based on communal property or ritual mobilization of production (Spielmann 2002; Wells and Davis-Salazar 2007) could potentially stifle alienation of crafted or surplus goods, at least for some classes of goods. According to French philosopher Michel Callon (1998), market development also requires social acceptance of a “calculative agency” in which the parties to a market exchange are able to evaluate and rank alternative market decisions under conditions of uncertainty and incomplete information about market conditions. Calculative agencies can be considered a formalist aspect of market origins, but the framework of rational decisions is culturally bounded (DiMaggio 1994; Stanish, Chapter 9). For instance, calculative agency may be socially restricted and inapplicable for transfers of some inalienable or “controlled” goods. Another precondition is an adequate flow of supply, demand, and price information. The relationships between supply-demand changes and prices are never simple and respond to a variety of conditions and circumstances. Access to information is frequently asymmetrical, creating imperfect competition, but exchanges nevertheless are based on shared understandings of value and fairness (even if one party is intent on exploiting that understanding); sometimes such understandings are predicated on ritual sanctification of market practices (Bohannan 1955; Davis 1992:66–68). Social networks based, for example, on kinship or social affinity also help facilitate perceptions of trust and price fairness, a point I explore in more detail in a subsequent 

C h rist o p h er P . G arrat y

section (Granovetter 1985). On a more formal level, legal or moral institutions may authorize sanctions that prohibit (or possibly enable) opportunistic exploitation of asymmetrical information (Williamson 1975, 1985). The spread of information required for market exchange (according to Pryor’s definition) is contingent upon the speed and efficiency of information flows, which may have been poorly developed in many premodern societies that had primitive communications technologies and widespread illiteracy. In modern markets information dissemination and processing is an important component of market operation (e.g., Dow Jones and other market sensors), even though uncertainty and unequal access to information are still rampant. This is one reason why Polanyi and his students argued that prices in premodern markets were not established through supply and demand but through price setting by governing officials. Prices responded to supply-demand information in premodern markets but probably not in the same way as in modern capitalism. Richard Swedberg and Mark Granovetter (2001:13) have characterized prices in preindustrial markets as “ ‘sticky’—that is, they only respond to major shifts in demand or supply.” In premodern markets, relatively small modulations in supply and demand probably went largely unnoticed, but abrupt, sizable changes (such as severe crop shortfalls or cessation of interregional trading relationships) would have been salient and therefore more likely to have prompted a pricing response. Prices also tended to stabilize for long periods as a result of long-standing (possibly intergenerational) exchange relationships among affines, kin, friends, ethnic brethren, and trusted associates (Braudel 1985:227; Swedberg and Granovetter 2001:13). Market Exchange and Barter

Market exchange and barter overlap in the aforementioned definition. I define barter broadly as a form of exchange that does not employ media of exchange (after Humphrey 1985; Humphrey and Hugh-Jones 1992) and that responds to some extent to conditions of supply and demand among parties to the exchange. Perceptions of supply and demand in some barter transactions may be ad hoc and based on idiosyncratic perceptions of value and equivalency. Nevertheless, in Pryor’s definition of market exchange, which admittedly casts a wide net, a dyadic barter exchange based on perceptions of variable supply and demand amounts to a form of market exchange. In other words, market exchange need not be predicated on media of exchange or currency. A question remains, however, about how barter relates to marketplace exchange and market system development. To some extent, this matter can be characterized as a “chicken-and-egg” conundrum: what came first, the practice of barter or the institution of the market? In my view, absent some external and intrusive effects of already developed market systems, the dyadic form of barter exchange likely preceded the market institution. This perspective does not exclude other factors involved in nascent market development, however, such as establishment of prices and formal 

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

equivalencies by government agencies or civil groups. According to Michael Hudson (2002, 2004), for example, some early Mesopotamian governments sponsored market ventures, offered credit lines, and established formal pricing as a means of promoting market participation and lessening the risk of market loss. These conceptualizations of market exchange and barter may not be universally shared among economic scholars, but they better conform to the developmental perspective adopted in this volume. Decoupling market exchange from institutionalized market structures highlights the early precedents of market institutions. In this volume Abbott (Chapter 3) addresses nascent market development among the Hohokam of central Arizona, a middle-range society without a state government (see also Abbott 2001, 2006; Abbott, Smith, and Gallaga 2007; Abbott, Watts, and Lack 2007). Abbott posits a market system linked to the development of regular public ceremonies associated with ritual ballgames. It is unlikely that early Hohokam market exchanges were rooted in a preexisting system of formal pricing, at least not initially. Nor was there a central government apparatus that established prices or equivalencies, as Hudson (2004) argues for Mesopotamia. Early Hohokam market exchanges likely began as a series of largely ad hoc barter transactions based on idiosyncratic perceptions of value and equivalency, although over time more well-established and normative notions of value may have developed. Barter likely preceded formal pricing mechanisms based on prevailing notions of value in most “preinstitutional” market contexts, but I do not wish to imply that the practice of barter diminished or disappeared after market institutionalization. Nor did the establishment of media of exchange undermine the importance of barter. Rather, it likely continued to be important in most market settings (Humphrey and HughJones 1992:3–4). Some early states instituted formal media of exchange, such as coinage, as means of tracking government finances and debts (M. E. Smith 2004:90–91). As Stark and Garraty (Chapter 2) explain, however, media of exchange may have been scarce or inconvenient for the bulk of petty transactions in early marketplaces. In some contexts, media of exchange may have been so infrequent and poorly distributed that they too became objects of barter (see Humphrey and Hugh-Jones 1992:4). Barter has long been an important mechanism of exchange in most premodern markets and has maintained a role even in well-developed modern markets (e.g., car tradeins [Humphrey 1985]). Marketplaces and Market Systems

Marketplaces are physical locations that accommodate regularized and orderly market exchanges. Market exchanges may have taken place at or adjacent to public assemblies, such as town fairs or places of ceremonial activities (see Abbott, Smith, and Gallaga 2007; Abbott, Watts, and Lack 2007; Abbott, Chapter 3) or in multipurpose facilities, such as plazas. Marketplaces, or any regular and predictable loci of market exchange, offer a formal setting for market exchange. As Hirth (1998:454–455) 

C h rist o p h er P . G arrat y

pointed out, this distinction between marketplace exchange and market exchange is crucial to the study of premodern markets. Market exchanges can occur in any number of centralized or decentralized settings (marketplaces, workshop procurement, itinerant middlemen). Conversely, marketplace exchange refers more narrowly to market exchanges within discrete physical loci that are spatially centralized and temporally regularized (follow a predictable schedule). From this perspective we can reasonably assume that dyadic market exchanges date back many millennia, but the development of formal marketplaces is likely more recent and marks the institutionalization of market exchange as a fundamental mechanism for economic transactions. The concept of market system refers to a regional network of interconnected marketplaces, including the market hinterlands they provision (Christaller 1966 [1933]). “Mature” marketplaces rarely develop in isolation but are linked to other marketplaces, forming a regional system (C. Smith 1974, 1976b). The articulations among these marketplaces will inevitably vary and, in some cases, evolve a hierarchical arrangement in which one or several marketplaces become larger, offer more products, and provision a larger number of consumers than smaller marketplaces. A market system also presumes connections between marketplaces in different areas such that products and commodities can be readily transferred among areas at a regional, interregional, or even a global scale. These are critical distinctions for the archaeological study of ancient market exchange. Archaeological data are poorly suited to detect dyadic market exchanges but are better suited to detect the existence of marketplaces and regional market systems, although inferring these is fraught with complications (Stark and Garraty, Chapter 2). Equally important to the question of “what is a market system” is “when is a market system” (to borrow from Hodge 1997). Do market systems require a formal infrastructure (plazas, stalls, roads)? If so, who foots the bill for constructing and managing the infrastructure? Does market system development require central authority and a strong state government able to oversee construction of formal marketplaces and enforce regulations? On a related point, can market systems develop in nonstate, middle-range societies? I do not propose to answer these questions, but I introduce them as vital areas of research. Having defined the key terms and conceptual positions for the volume, I next discuss the theoretical scholarly environment for studies of premodern markets, specifically the importance of Polanyi’s influential ideas. Theoretical Approaches to Premodern Markets: Polanyi’s Lasting Influence Polanyi’s Argument

Modern scholarly debate about the origins of market exchange hinges on the work of Polanyi (1960, 2001a [1957], 2001b [1944], and elsewhere), whose eloquent writings framed the debate about premodern market exchange for generations of scholars. It is important to understand Polanyi’s influence on this topic, even though his ideas 10

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

are well over a half century old, because he is the only scholar to have developed a comprehensive body of theory about premodern market exchange (but see Swedberg 1994, 2005). Several contributors in this volume refer to his writings, and thus a brief overview is warranted. Detailed treatments of Polanyi’s ideas can be found elsewhere (Halperin 1984, 1991, 1994; Humphreys 1969); here I evaluate the extent to which they advanced or inhibited understanding of premodern market development. Polanyi’s work was largely a reaction to what he saw as the dehumanizing tendencies of classical and neoclassical economics and the positivist approaches on which they are predicated. Starting with Adam Smith (1976 [1776]), classical economists have long asserted that humans are naturally imbued with the capacity for market rationality. Polanyi, in turn, argued that a “market mentality” is solely a product of modern capitalism. These positions essentially mark the two poles of the well-known formalist-substantivist debate in economic archaeology. Briefly, formalists follow classical/neoclassical traditions in seeking a unified theory of rational economic behavior; substantivists join Polanyi in viewing non-Western economies as embedded in deeper social, political, and religious institutions (for a review, see Isaac 1993). The work of classical/neoclassical economists implies that exchange behavior among self-interested actors inevitably leads to market development and an increasingly complex division of labor (Brumfiel and Earle 1987:1–2; Wilk 1996:43–72). These approaches seek to formulate a unified theory of economic action predicated on “atomistic” principles of utility maximization and cost minimization for allocation of scarce resources. From this perspective, market exchange originated as a consequence of individuals freely pursuing their own economic interests; humans possessed an innate predilection for market exchange long before the establishment of formal marketplaces. According to Adam Smith, for example, the pursuit of the “private interests and passions of men” created a situation “which is most agreeable to the interest of the whole society” (A. Smith 1976 [1776], Book 4:630). Market competition breeds the smooth transfer of goods and commodities, but only to the extent that society sanctions and expresses a willingness to pay for those goods; market competition provides a social check on undue accumulation. In this sense, Smith and other classical/neoclassical scholars saw markets as self-regulating not just in terms of economic provisioning but also in terms of social order and stability. State and governing bodies and other social institutions are characterized as external interlopers that interfere with the innate ambitions of a free and independent population of self-interested producers and traders (Gudeman 2001:82). Like many scholars in anthropology and sociology, Polanyi (2001a [1957], 2001b [1944]) was deeply opposed to the idea of an innate tendency to engage in exchange in a competitive market setting. He viewed the pursuit of self-interested economic gain as an utterly unnatural human ambition.1 Market rationality, he argued, surfaced in conjunction with capitalism during the Industrial Revolution: “Market society was born in England. . . . Market economy, free trade, and the gold standard were English inventions” (Polanyi 2001b [1944]:30). Polanyi (2001a [1957]:37–40) was skeptical 11

C h rist o p h er P . G arrat y

that competitive market exchange had ever constituted an important component of premodern economies and formulated three lines of argument to support his contention: (1) the absence of factor markets for land and labor, (2) the social antagonisms inherent in market exchange and bargained pricing, and (3) the technical inefficiency of primitive communications, which hindered effective flows of information about market conditions. I next explore each of these lines of argument. First, Polanyi objected to premodern market development on the grounds that market exchange was not a feasible or dependable mechanism for the allocation of key resources in premodern societies, specifically land and labor. He outlined alternative “forms of economic integration” for premodern societies that involved behaviors embedded in formal and informal social institutions or “supporting structures” that served to regulate economic behavior and prevent antagonism. Reciprocity, embedded in kinship relationships, was the dominant form of economic integration among egalitarian and early village societies; redistribution, embedded in central government, was dominant in complex chiefdoms and early states (Stark and Garraty, Chapter 2). Only with the rise of European capitalism did economic allocation become disembedded from these institutions (Humphreys 1969:185–186). Polanyi (2001a [1957]:39) further distinguished market exchange from the other forms based on the means of establishing exchange equivalencies: “exchange at set rates occurs under reciprocative or redistributive forms of integration; exchange at bargained rates . . . is limited to price-making markets.” Important here is Polanyi’s emphasis on dominant forms of integration. The dominant form, in Polanyi’s scheme, is responsible for the allocation of land and labor (factor markets), but other forms may occur as “subordinate” mechanisms of allocation. He thus did not categorically deny the existence of market exchange in premodern societies but opposed the idea that market exchange was the principal mechanism for allocating land and labor. He is likely correct that market exchange was not a major mechanism for the allocation of land and labor in premodern societies (see also Dalton 1962:365–367), although some ancient documents indicate low-level real estate and wage-labor markets. Even though market exchange of “nonfactor” products and services, such as household crafts, may constitute a sizable component of the economy, the number and range of goods integrated into the market sphere are less crucial than the processes by which various goods and services do or do not become alienable and exchangeable through market channels, a point I explore in a following section. Polanyi’s second line of opposition to premodern market development concerned what he saw as inherent antagonisms in market bargaining. Although cognizant of ancient marketplaces in various world areas (Greece, Rome, Mesoamerica, China), he was skeptical that early marketplaces could have operated according to market principles without causing undue social antagonism. In direct opposition to Adam Smith and other classical/neoclassical economists, Polanyi insisted that acts of exchange based on fluctuating prices should “tend not to occur”; if they did, “a violent emotional reaction would set in, as against acts of indecency or acts of treason, since 12

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

trading behavior is never emotionally indifferent behavior and is not, therefore, tolerated by opinions outside of the approved channels” (2001a [1957]:37). His argument is worth repeating in full: Exchange at fluctuating prices aims at a gain that can be attained only by an attitude involving a distinctive antagonistic relationship between the partners. The element of antagonism, however diluted, that accompanies this variant of exchange is ineradicable. No community intent on protecting the fount of solidarity between its members can allow latent hostility to develop around a matter as vital to animal existence and, therefore, capable of arousing as tense anxieties as food. Hence the universal banning of transactions of a gainful nature in regard to food and foodstuffs in primitive and archaic societies. The very widely spread higgling-haggling over victuals automatically removes price-making markets from the realm of early institutions. (2001a [1957]:39)

Polanyi maintained that economies predicated on market principles were actively prohibited through social or government sanctions. Polanyi viewed administrative price setting as one of the chief sanctioning measures used to prohibit market practices. As Sarah Humphreys (1969:186–191) makes clear, however, his juxtaposition of fixed and bargained pricing is based on a false dichotomy. Many factors converge in setting market prices because sellers generally establish prices based on perceptions of market conditions (mainly supply availability), often in consultation with administrative or market officials responsible for gathering such information. In some premodern and peasant markets, formal organizations—such as trade groups, market or governing officials, or legal assessors—were specifically charged with gathering information regarding supplies and availability for establishment of prices (Hudson 2004; Humphreys 1969:188–189). Furthermore, haggling is sometimes a secondary component of price formation and in some peasant markets occurred only under specific conditions, for example, to sell overstock at the end of the day (Reeves 1989) or to ensure that the seller’s overhead and transport costs were covered (S. Cook 1976:160–161). The third line of argument related to Polanyi’s opposition to premodern market development concerns the inefficient means of communicating price information. Primitive communication technologies, coupled with limitations in storage and transport technology, prohibited development of a “common language” for competition among producers and traders of similar commodities. Polanyi (1960) drew on a study of marketplaces (agorae) in classical Greece to support his argument. He viewed agorae as independent establishments with few links to other agorae, thus failing to form a regionally integrated system (Polanyi 1960). The absence of linkages among agorae would have led to inefficient commodity trafficking and prevented the communication of price information necessary for a well-functioning regional market system. As Scott Cook (1976:142–144) explains in his ethnographic study of a peasant market system in Oaxaca, however, information about market conditions may be unevenly distributed on a system-wide scale but tends to be accessible on a local scale, creating a competitive 13

C h rist o p h er P . G arrat y

marketplace that responds well to local market conditions. Limited information flows did not prevent inter-marketplace trafficking and integration of the larger market system in Oaxaca. In sum, Polanyi’s views of premodern markets appear contradictory on the surface. On one hand, he leaves open the possibility of market exchange as a subordinate means of allocation, and, on the other hand, he stresses the disruptiveness and impracticality of market development. One should be mindful, however, that his larger goal was not to repudiate premodern market exchange but rather to elucidate the structural changes involved in the rise of Western capitalism and the commoditization and alienation of labor. In so doing, however, his scheme left little “wiggle room” to account for large premodern market systems other than to downplay their significance as subsidiary components of the economy or to emphasize the extent to which they operated with fixed pricing, both of which are readily refuted. In premodern market systems, many goods and services were not integrated into the market system, and land and labor were infrequently exchanged through market channels, but that does not negate the importance of markets in the broader economy. In no economies, including modern capitalist economies, are all goods and services circulated through market channels, but market systems may develop to accommodate specific economic sectors or commodity classes. More important is the mix of exchange and provisioning mechanisms and the extent to which commodity and service sectors were integrated into market systems (Hirth, Chapter 11; Stark and Garraty, Chapter 2). Polanyi’s Legacy

Many of Polanyi’s followers echoed his arguments about premodern market development (e.g., Bohannan and Dalton 1962; Dalton 1962; Sahlins 1972). Widely cited is the debate initiated among classical scholars by Finley (1999 [1973]), who posited a model that emphasized small-scale, agrarian production and relegated marketplace exchange to a relatively minor role in the broader Mediterranean regional economy. Although Finley was critical of some aspects of Polanyi’s argument, his “primitivist” views unleashed a series of critiques and counterarguments that still resonate in classical studies (I. Morris 1999; Saller 2005). Johannes Renger (1984, 1995) similarly argued against market development in the ancient Near East, citing documentary evidence that economic transactions were closely supervised by governing authorities. In Mesoamerica, Pedro Carrasco (1978, 1980, 1982, 1983) characterized the Aztec market system in highland Central Mexico as state-controlled venues for the allocation of wealth and resources. He suggested that state leaders were able to control market transactions using heavy-handed means, such as price setting, compulsory market attendance, and sumptuary laws (but cf. Berdan 1983; Offner 1981a, 1981b). In my opinion, the tremendous influence of scholars such as Polanyi and Finley has considerably impeded scholarly investigations of premodern market development (see also Blanton and Fargher, Chapter 10; Feinman and Nicholas, Chapter 4). As a 14

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

result of their legacies, much effort has been devoted to debating the existence of “true” markets rather than to empirical investigations of market development and change. Norman Yoffee (2001:303), in an insightful review of a volume on ancient Near Eastern economies, provides perhaps the best and most succinct counterargument to this tendency to continually debate the existence of premodern markets: “Nietzsche said you can only define things that have no history. Endless debates about whether there was a market and private property do not seem to advance our understanding of economic phenomena in their historic contexts.” Like Yoffee, the contributors to this volume are concerned less with whether markets should have existed based on theoretical presuppositions and narrow definitions than with empirical evidence for market development and change. The contributors here provide no blanket endorsement of classical/neoclassical economics or a unified theory of economic behavior, however. Classical/neoclassical economists tend to view economic and social and political institutions as separate spheres of social interaction, which also presents a serious fallacy (e.g., Davis 1992; Dequech 2003; Gudeman 2001; Lie 1997; various articles in Smelser and Swedberg, eds., 2005). At the level of individual transactions, market exchange is embedded in political relations in the sense that all social interaction is rooted in webs of power, rank, and inequality (Bourdieu 2005 [2000]). In a broader sense, markets are embedded because transactions occur in the contexts of social ties and of religious and governing institutions that inevitably assume some position toward market activity (including a laissez-faire stance; see Block [1994]; Block and Evans [2005]). Even some economists—the standard-bearers of neoclassical thought—now accept that economic activity is embedded in broader social and political structures (Dequech 2003). Over thirty years ago Carol Smith (1976b:44) lamented the unfortunate tendency among scholars to adopt either a classical/neoclassical (formalist) or a substantivist position in explaining market origins and development. Such thinking, she proclaimed, propagated widespread neglect of the “interrelations of social and economic forces in market evolution.” Both perspectives are valid in that premodern markets and market exchange can and should be characterized from the perspective of both the formal-economic and socially embedded contexts of economic transactions (Bestor 2001:9227; Plattner 1989c:177; Wilk 1996). Markets involve diverse groups of actors engaging in complex relationships and operating with established rules and conventions about market operations. Newer Perspectives

Two recent approaches have been formulated to bridge the gap between classical/ neoclassical (formalist) arguments and Polanyi’s substantivist position. New Economic Sociology (NES) and New Institutional Economics (NIE) largely focus on the important issue of social order, a central component of Polanyi’s argument. Both have also been applied to advance understanding of premodern market development and 15

C h rist o p h er P . G arrat y

organization (for example, I. Morris 2004; Morris and Manning 2005a; Silver 1995; Temin 2003). I briefly discuss both approaches and explain how they have been em­ployed in the study of premodern markets, including by contributors to this volume. NES scholars, most notably Granovetter, follow Polanyi in rejecting mainstream economic perspectives that emphasize rational individual action devoid of social context. Granovetter’s (1985) widely cited statement on NES hinges on Polanyi’s notion that economies are embedded in social contexts, but NES studies more frequently focus on the embeddedness in social and interpersonal networks than on institutional embeddedness, as espoused in Polanyi’s “forms of integration.” The NES, or “embeddedness,” perspective concentrates on economic interactions among firms or individuals in modern capitalist economies, but it is equally applicable to premodern markets. Granovetter (1985) follows Polanyi in emphasizing the importance of social networks as a means of inhibiting malfeasance in market exchange and promoting social order (Cumberpatch 2001). Some anthropological studies of twentieth-century nonWestern marketplaces (e.g., Geertz 1978; Plattner 1989b; Russell 1987; Schwimmer 1979) have adopted a similar perspective that focuses on ethnic affiliations or interpersonal relationships as means of facilitating exchange and combating antagonism. In this volume Jeffrey Fleisher (Chapter 7) adopts an argument couched in NES principles to characterize market exchanges among African Swahili merchant elites and foreign traders. Exchange relationships for imported foreign goods, according to Fleisher, were rooted in social networks and the establishment of trust among local and foreign merchants. In Fleisher’s words, these interpersonal connections “served a practical function for overseas merchants, who needed a place to stay during the weeks or months after their arrival as they waited for the monsoon winds to shift back north.” His chapter highlights the bonds of trust forged among foreign and Swahili merchants that sustained market development. Like NES, the influential NIE school within economics focuses fundamentally on the issue of social order but within the framework of the classical/neoclassical tradition. Rather than focus on social networks, NIE scholars model economic activity in terms of the social and legal rules (“institutions”) that regulate and enable economic action (North 1977, 1990, 1991; Williamson 1975, 1985; but see critiques in Granovetter 1985 and Swedberg and Granovetter 2001:14–18). Thus, whereas NES studies rely on what economic anthropologists would consider a substantivist concern with social networks, NIE studies address the issue of social order from what might be viewed as a formalist perspective that emphasizes cost-effectiveness and risk minimization. NIE scholars principally focus on transaction costs, that is, the costs of implementing economic transactions (especially recurring transactions), such as those involved in gathering information, monitoring performance and negotiation, and enforcing prohibitions against unchecked malfeasance or opportunism (Williamson 1975, 1985; see Chapman and Buckley 1997 for an anthropological perspective). Examples include contracts or trade agreements, but informal mechanisms such as moral proscriptions might also function to minimize transaction costs. 16

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

In this volume Stanish (Chapter 9) adopts an NIE approach to explore differences in the effectiveness of market and administrative systems for urban provisioning. He applies the concept of transaction costs to explain that the Andean system of political provisioning entailed considerably higher “overhead” per economic transaction (for storage, accounting, policing of corvée laborers), which curbed the volume and efficiency of economic provisioning and capped urban growth. In his usage, the concept of transaction costs provides a useful heuristic for comparing different urban-­provisioning systems. Robert Hunt (1987:181–184) similarly discusses the transaction costs involved in market provisioning of food to urban centers, including costs associated with transport, bureaucratic management, and policing. Finally, in a detailed study of the development of credit practices in the ancient Middle East, Hudson (2002) explains that large institutions, such as palaces and temples, promoted “entrepreneurial” risk ventures for aspiring traders by establishing lines of interestbearing credit, including formal and administered contracts. In so doing, these institutions established standardized weights and measures, price-setting practices based on available supplies, and money designations of silver and barley (Hudson 2002:13; see also Hudson 1996). Hudson does not explicitly adopt the NIE terminology, but his argument implicates the NIE scholars’ focus on institutional arrangements for minimizing risk and fostering market development. Hirth (Chapter 11) also applies the concept of transaction costs to model market development and participation at the household level in premodern societies. Market development, according to Hirth, offered householders a means of reducing the transaction costs entailed in provisioning themselves with domestic goods from a variety of sources in different locations, which otherwise would have been inefficiently procured through multiple unconnected non-market channels. He hypothesizes that market exchanges came to be associated with predictable and regularly scheduled communal gatherings (e.g., fairs or ceremonial events) as a means of reducing such transaction costs. In Hirth’s view, premodern market development offered a viable solution (albeit not the only one) to the problem of “broad-spectrum” resource provisioning at the household level. In sum, the NES and NIE approaches, although considerably different in their underlying philosophies and assumptions, provide potentially useful stances for studying the development and evolution of premodern markets. Both address many of the issues and problems of premodern market development outlined by Polanyi, such as the problem of social order, but approach these issues with different base assumptions about human behavior (I return to the issue of social order later). The Four Principal Themes

At the outset I introduced four volume themes: (1) how do markets articulate with other sectors of ancient economies, (2) how do we recognize market exchange archaeologically, (3) what is the relationship between political and market institutions, and 17

C h rist o p h er P . G arrat y

(4) how do market economies develop and evolve over time. In this section I explore each, with the exception of the second methodological theme, which is addressed in Chapter 2. Market Integration and Commoditization

Our first theme focuses on the role of markets in the wider economy. Important here is the idea that market exchange in premodern societies typically coexisted with other, non-market forms of exchange. An essential component of premodern market studies is the extent to which different sectors of the economy (land, luxury goods, and so on) were exchanged through market versus non-market channels in different times and places. Even in modern capitalism, much of the interaction among business firms is predicated on principles of cooperation and reciprocity, which may influence prices as much as market competition and supply-demand modulations (e.g., Granovetter 2005a; Uzzi 1997). A key element of the study of premodern markets concerns the process by which various goods and services were integrated into, or detached from, the market system. Ancient economies can be conceived, as Michael Smith (2004:75–76) suggests, along a continuum of economic organization from non-commercialized to low- and intermediate-­commercialized to advanced market economies, with the last incorporating many sectors of the economy into a market exchange system. But this continuum is not meant to imply a linear process of market expansion; market change is far more dynamic and involves processes in which commodities and sectors are both added to and subtracted from the market domain (i.e., commoditization and de-commoditization of goods and services). One example is governmental decisions to regulate or control distribution of certain goods, such as weapons or narcotics, by removing them from regular market channels. Paul Bohannan and George Dalton’s (1962) distinction between peripheral and integrated markets offers a useful touchstone for studying this issue. Their intention in developing these terms was to support Polanyi’s distinctions between modern market systems and “primitive” peripheral markets predicated on price fixing and poor regional integration. I see little value in their terms as categorical distinctions. If thought of as a continuum, however, their idea provides a useful heuristic for conceptualizing variability in the extent to which households rely on marketplace exchange to procure domestic provisions (Garraty 2009). In peripheral markets, households do not rely on the marketplace for most of their basic provisions. Rather, “ ‘peripheral’ markets are patronized chiefly by ‘target marketers,’ who trade in the market when they happen to have a surplus to exchange or need some item they do not produce” (Hicks 1987:91). Differential use of marketplaces may occur among elite and nonelite households and among urban and rural populations within a broader market region. In well-integrated markets, conversely, the majority of households rely on marketplace exchange to procure all or most of their everyday provisions. 18

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

A logical extension of the question of market integration is the question of why and how market systems developed and flourished in some areas of the ancient world but not others. What differences in political or social structure account for the variable development and importance of marketplaces and market systems in different world areas? The Andean region of South America may be the single best “natural laboratory” for exploring this question. This region—with its large population, diverse regional ecologies, complex transport systems, and sizable segment of nonfood producers—would seem to be an ideal setting for market system development. For reasons yet to be fully explored, market systems were absent, suppressed, or only modestly developed in the Inca empire prior to Spanish contact (Earle 1985; Stanish, Chapter 9; Stark and Garraty, Chapter 2). Also important are the consequences of marketplace development and nondevelopment. Stanish (Chapter 9) addresses the consequences of minimal marketplace development on urban scale in the Andes (see also Stanish 1997). In the Andean case the generally moderate-sized cities offer a marked contrast with the very large cities within well-developed market systems, such as the Mediterranean, China, and Mesoamerica. He makes clear that the development of ancient cities and urban systems cannot be fully understood without also exploring the transaction costs of the economic systems that provisioned those cities with food and other domestic necessities. In economies with market systems, the state did not bear the brunt of urban provisioning costs; rather, the market generated an incentive for surplus production among petty producers, and the mobilization costs of trafficking surplus goods were borne by individual market sellers and merchants. This low-cost, high-incentive situation accommodated urban growth and sustained larger populations. Market Regulation: Social and Political Institutions

This theme relates to the relationships between markets and other social institutions, a core issue of Polanyi’s intellectual project and a vital concern in any study of premodern market development. Much of the focus in this section concerns oversight and regulation of markets and the extent to which market transactions are a potential source of social contention and disorder. The “problem of order,” as Granovetter (1985:484–487) explained, has been a matter of scholarly debate since the time of Thomas Hobbes (1996 [1651]) and the social contract theorists of the 1600s and 1700s. Hobbes famously characterized the natural state of human societies as “a war . . . of every man, against every man” (1996 [1651]:85) and saw repressive political structures as a necessary precondition for the orderly operation of society. Hobbes’s argument is thus antithetical to Adam Smith’s (1976 [1776]) suggestion over a century later of atomistic social competition as the root of social order. The quintessential disagreement between the neoclassical and the substantivist schools of thought hinges on how economic transactions among individuals are (or should be) regulated or controlled. For classical/neoclassical scholars inspired by 19

C h rist o p h er P . G arrat y

Adam Smith and others, governments (or other external sources of regulation) merely interfere with a free and unencumbered populace of traders and market patrons. Other scholars inspired by Polanyi counter that producers and consumers tend to avoid participating in marketplace exchange unless governmental or other regulatory mechanisms are in place to ensure a smooth and peaceful marketplace operation (e.g., Davis 1992; Sahlins 1972:297–314). Granovetter (1985) offers a third possibility—that order is rooted in networks of social relations that promote trust and limit malfeasance. Most archaeological research on this issue has focused on governmental oversight, as a result of Polanyi’s influence, but government is not the only source. I explore political oversight in the next subsection and nongovernmental oversight in the following subsection. Governmental Institutions and Market Oversight. According to Carol Smith (1974, 1976b), the intersection of politics and markets is the principal cause of crosscultural variation in regional market organization. Economic sociologist Fred Block (1994; Block and Evans 2005) similarly has written about the inevitable role of governing agencies in market systems (Bourdieu 2005 [2000]:81; Fligstein 1996). Like these scholars, I believe we are better off thinking about political and market institutions as co-evolutionary rather than considering politics as simply one of many “external variables” that affect market development and system organization. Even in the contexts of acephalous nonstate societies, tribal leaders or Big Men may organize, oversee, and tax market transactions, as is the case among the Tiv of West Africa (Bohannan and Bohannan 1968). One important question concerns why governing agencies even take an interest in markets. Governing institutions may benefit from marketplace development in at least three ways: tax revenues, conversion, and control over product distributions. First, taxing market activities—for example, market transactions, use of market stalls, shipments (tariffs), and so on—was an important source of revenue in some premodern states (Blanton and Fargher, Chapter 10). Marketplace exchanges are spatially centralized and temporally predictable and therefore were more readily tapped for taxation than other, decentralized mechanisms of exchange (e.g., reciprocal change), albeit with substantial costs to the tax assessor (for enforcing tax payments, retaining tax collectors, bookkeeping). According to Shmuel Eisenstadt (1993:47), compared with other mechanisms of exchange, market exchange is perhaps “the least embedded and economically the most autonomous” from leaders’ realms of control and therefore more readily alienable and “maneuverable” than other sources of elite financing, such as land or corvée labor. Eisenstadt implies that competition among elites and polities over market sales and tax revenues may have been commonplace in ancient societies with market systems. Second, governing agencies may interfere in marketplace exchange to facilitate product conversion (Hirth 1998, Chapter 11). For example, markets provide a means for state agents to convert wealth goods received as tribute into bulk commodities, 20

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

such as food or domestic wares (or vice versa). The ability to convert commodities was likely invaluable for elites and governing officials in early states that needed to feed and provision armies and bureaucrats but also to retain a stock of high-value goods for elite gifts or display during feasts. Elite financing in many early state economies relied on access to both high-value wealth goods (fine ornaments, extravagant textiles) and staple goods (mainly foodstuffs; see D’Altroy and Earle 1985). The most successful elites may have been those who were able to access and control both sources of revenue. In this context, use of the market system to convert wealth and staple goods might have been a vital component of many premodern political economies. Third, governments may wish to co-opt or control distribution of certain key commodities or raw materials, thereby removing these items from market spheres of exchange. To take a modern example, no national governments would concede to free-market sales of weapons-grade plutonium; government regulation is widely viewed as beneficial in this extreme case. In ancient contexts, governing officials may have wished to control access to goods frequently used for state construction projects (such as high-value construction materials) or to equip state armies. In a similar vein, administrative officials also may have wished to control important shipping routes, transport, and production technologies or specific craft workers and laborers (for example, Inca state-sponsored craft specialists), thus diverting labor and technologies from surplus production that might have entered the market. Are administrative institutions a necessary precondition of marketplace development? Marketplaces may entail a substantial investment in infrastructure (e.g., roads and ports) and policing. If governing officials and other regulating agencies perceive the potential benefits of marketplace exchange described previously, then it behooves them to police the marketplace, enforce market rules, and ensure a smooth and peaceful process. Beyond these basic infrastructure and operational investments, however, did governing agencies assume a more prominent role in market development by actively promoting or underwriting marketplace participation? Various scholars have come to different conclusions about this question. Answering in the affirmative is John Davis (1992:65–73), who suggests that states tend to actively promote market risk and competition among market firms, which ensures more productive and commercially engaged consumers and, hence, more market transactions and taxable wealth. Davis (1992:69) concludes: “Only if government intervenes to prohibit insider manipulation can state’s men be sure that sufficient citizens will trust the market to be impartial between them, and so enter the game of wealth creation, and so generate revenues.” Among the Tiv, for example, consumers will not participate in marketplace exchange unless the marketplace overseer or sponsor can ensure peace and intervene in settling disputes (Bohannan and Bohannan 1968). It follows from Davis’s argument that market systems may never have developed without the input and support of strong governing agencies. Other scholars of premodern economies have also supported a strong government role in early market development (Sahlins 1972). Hudson (1996) argues for 21

C h rist o p h er P . G arrat y

the ancient Near East that price-making mechanisms actually sprang from state institutional practices rather than from “free exchange” and competition, a position that has important implications for the role of states in the development of market institutions. Guillermo Algaze (1993) describes the development of trade and mercantile exchange under elite auspices at Uruk and other early states. In the Aztec empire of Central Mexico, Frederic Hicks (1987:99–101) suggests that imperial officials in Tenochtitlan demanded tribute in nonlocal goods (cotton cloth, cacao) to compel agrarian families to use the marketplace to exchange domestic food surpluses and obtain tribute goods (market conversion), thus ensuring ample supplies of foodstuffs in the marketplaces of the large urban centers (see also Brumfiel 1980, 1987b). Other scholars resist this “top-down” view and see market development as largely a “bottom-up” process initiated by aspiring market traders and producers. The chapters in this volume by Abbott (Chapter 3) and Geoffrey Braswell (Chapter 6) discuss episodes of market expansion during periods marked by small and decentralized political systems, suggesting that markets flourished in the absence of large states with centralized governments. Similarly, based on a study of long-term changes in formal and decorative pottery styles in Oaxaca, Gary Feinman (1985, 1986; Feinman, Kowalewski, and Blanton 1984) posits elaborations in decorative styles during periods of decentralization and weak governmental control over the market, which facilitated economic competition among pottery producers (and other craft manufacturers); conversely, styles generally were simpler and more uniform during periods of centralized governmental control (but see Brumfiel 1987a). I am reluctant to draw broad generalizations about market-political relationships from the previous examples. In some cases market systems appear to have grown in concert with the development of powerful states, but this was not always the case. In my own study of pottery production and exchange in the heartland of the Aztec empire in the Basin of Mexico (Garraty 2006, 2007), the evidence points to increased market expansion and commercial growth during the era of imperial expansion. I argue that imperial leaders, especially at Tenochtitlan, deliberately and strategically encouraged large-scale market participation and commercial exports as a means of garnering tax revenues and prestige from operating a prosperous marketplace, resulting in a more integrated regional system (Hassig 1985; Hodge 1992; Hodge and Minc 1990; Hodge et al. 1992, 1993; Nichols et al. 2002). Although processes of market expansion likely predate the empire by several decades (Smith and Berdan 2003), the Aztec regimes in Tenochtitlan and, to a lesser extent, in Texcoco seem to have encouraged and facilitated this expansion. Indeed, Aztec imperial expansion brought about economic growth and new market opportunities for core-area goods and merchants in conquered areas (Berdan 1985, 1996:132–135). Michael Smith (Chapter 8) similarly emphasizes the importance of top-down processes of market development and change in Aztec-period Morelos, just south of the Basin of Mexico. He attributes processes of market system expansion and shifting intraregional interaction within Morelos to political machinations among competing polities. 22

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

Richard Blanton and Lane Fargher (Chapter 10) conducted a cross-cultural study of thirty premodern states with market systems and found mixed results concerning administrative involvement in those systems. Some market systems flourished in the absence of governmental involvement or oversight (e.g., Tokugawa Japan, Java, late Feudal England), but in many other premodern societies market development was supported by governing agencies. To be clear, however, their study is not expressed in terms of a “degree” or level of governmental involvement in market development, that is, as a top-down versus a bottom-up process. They focus instead on a different dimension: the extent to which premodern market development correlated with collective versus non-collective forms of government. They define collective governments as those in which “rulers are forced to strike bargains with other civil society groups” and are “strongly dependent on taxpayer-produced revenues.” More non-collective forms of government, in contrast, rely less on citizens for government revenues—in the form of taxes or tribute—and more on other forms of revenues (e.g., inter-elite exchanges or state control over key resources). Their cross-cultural study shows that premodern markets tended to develop and flourish in polities with collective forms of government. These studies ultimately indicate that governing institutions in different times and places adopted a variety of policies and approaches to regulate market systems. Some governing agencies may have attempted to curb market development to prevent undue accumulation of wealth and resources among prosperous merchants or producers. Others may have encouraged commercial development and growth or even directly supported marketplace participation and development. Still others may have adopted a laissez-faire attitude toward market activities or certain sectors of the market. This diversity of government-market relationships should not come as a surprise, as many modern governments—including the federal government of the United States—alternatively act as both facilitators (international trade agreements, for example) and regulators (tariffs, antitrust policies) of market exchange. Scholars of premodern markets need to consider whether governing agents would have had the resources or the technology required to impose control over market transactions. Governing agents cannot simply coerce consumers to participate in marketplace exchange. Nor do they readily control the wide range of actions and interpersonal negotiations that occur over a large regional system. Market exchange on a system-wide scale represents a form of what sociologist Michael Mann (1986:8) calls “diffused social power,” which “spreads in a more spontaneous, unconscious, decentered way throughout a population.” Diffused power contrasts with “authoritative power,” which is predicated on tangible command structures and self-conscious obedience to a hierarchical institution, such as a military chain of command. Diffuse sources of power and agency do not lend themselves to authoritative, centralized control at the hands of political authorities (Mann 1986:10). Mann’s characterization of market exchange as a diffused power makes clear that premodern political regimes were probably unable to dictate and police market exchanges throughout an entire 23

C h rist o p h er P . G arrat y

market system, given the thousands of transactions occurring simultaneously in dozens of regional marketplaces—an assumption implicit in many substantivist arguments inspired by Polanyi. Nongovernmental Regulation. If market systems can grow and expand in the absence of governmental oversight, as several studies suggest, what are the mechanisms (if any) for overseeing peaceful marketplace exchange? One possibility is oversight by nongovernmental agencies, about which much less has been written with regard to premodern societies. I separately discuss social and religious or moral sources of oversight and economic order. Social Networks. In an often-cited paper, Granovetter (1985) makes the case

that social order in economic transactions is primarily rooted in social networks rather than in moral codes or regulatory institutions. Market participants frequently forge personalized relationships to ensure reliable and secure buying outlets for themselves over the long term, which provides a means of avoiding the risks of overpaying or receiving poor-quality merchandise (Plattner 1989b:214). The ethnographic literature is rife with examples of embedded relationships in market contexts. Davis (1992:70–71) observed that market traders in Libya dealt almost exclusively with clients who were kin or relatives. Brian Schwimmer (1979) similarly emphasized market wholesaling in Ghana among members of the same ethnic group. Less formal relationships, such as repeat customers, provide another example of how the social establishment of trust may condition market behavior. Stuart Plattner (1989b:212) sees long-term personalized relationships as a hall­mark of peasant markets, and the same is no doubt true of premodern markets. Plattner argues that socially embedded transactions adhere to a sense of market rationality just as much as atomized transactions do but at a different temporal scale: “The goal of each actor is his or her economic self-interest, yet the maintenance of the relationship is valued over short-run profit. . . . The key element is that exchanges do not have to be balanced in the short run, since past or future shortfalls are adjusted in the continuing stream of exchanges” (Plattner 1989b:212). The establishment of social bonds in the marketplace was vital to combat the effects of uneven information distribution, which created conditions ripe for abuse and duplicity (S. Cook 1976; Forman and Riegelhaupt 1970:205; Plattner 1989b:214–217, 220). Moreover, comprehensive oversight of all marketplace exchanges was not feasible in premodern societies, as explained previously, thereby creating a heightened need for assurances of fair pricing. Embedded transactions thus provide a means of sustaining market development without incurring social disorder and conflict, as posited by Polanyi. Polanyi failed to understand embedded transactions as an alternative to social or government sanctions against market development. Nongovernmental groups or civic institutions, such as guilds or trade associations, also contributed to marketplace oversight in many premodern markets. Guilds 24

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

and guild-like organizations formed in India, China, the ancient Mediterranean, the Near East, Europe, Mesoamerica, and possibly other areas. In medieval India, during some periods merchants and guilds appear to have exercised greater power over the marketplace than the central government did (Morrison 1997). In addition, colonial records from Aztec Central Mexico attest to an organization of high-status, long-distance merchants, or pochteca, who acted as arbiters in the marketplace at TenochtitlanTlatelolco (Sahagún 1950–1982 [ca. 1577], Book 9). Closely allied with the state, the pochteca settled disputes among market patrons, created and enforced market laws, and are said to have possessed their own court system to facilitate their role as intermediaries. Moral Economies. Many economic anthropologists have focused attention on

“moral economies,” which concern the interconnections between economic behavior and cultural beliefs (Scott 1976; Thompson 1971). In premodern market systems, moral codes and unwritten rules of behavior may have been vital for marketplace development, especially where political oversight was weak or absent (Abbott, Chapter 3). In the absence of overarching governmental or legal controls, common beliefs or religious proscriptions against cheating or dishonesty may have provided moral grounds for peaceful market operations (e.g., Bohannan and Bohannan 1968). Abbott and colleagues (Abbott, Smith, and Gallaga 2007; Abbott, Watts, and Lack 2007; and Abbott, Chapter 3) suggest that marketplaces developed in the Hoho­ kam region in connection with the spread of ceremonial ballcourt villages. They argue that subregional product specialization and exchange would have necessitated cooperation and affable relationships among sellers and consumers from diverse settlements over a large and politically decentralized area. This cooperation and large-scale integration likely rested not on any sort of political authority but rather on a “moral economy of kinship with a shared consciousness of religious identity” (Abbott, Smith, and Gallaga 2007:478). This argument resembles Paul and Laura Bohannan’s (1968) observation among the Tiv that market sponsors first had to have their marketplaces ritually consecrated to ensure peace and gain the trust of consumers. In both cases ritual sanctions and consecration likely helped enforce a moral code of behavior to reduce disputes in the absence of strong political oversight. Similarly, in early historic and medieval India (AD 500–1600), the ascent of Buddhism may have paved the way for market development by removing many of the previous Hindu sanctions and restrictions on mercantile behavior and social advancement (Ray 1986), although some scholars have contested this characterization (Dale 1994). Buddhist moral precepts might have facilitated peaceful marketplace operations. Merchants and craft guilds—operating relatively independent of the state—also were major supporters of early Buddhist monasteries and were partly responsible for the expansion and spread of Buddhism throughout southern and southeastern Asia (Morrison 1997:95–96). 25

C h rist o p h er P . G arrat y

Market Development and Change

The fourth theme of this volume concerns the origins, development, and longterm evolution of premodern markets and market systems. I concentrate here on two aspects of premodern market development: market origins and market change. Market Origins. Different perspectives on the origins of market exchange and marketplaces are implicit in the neoclassical and substantivist arguments (C. Smith 1976b:44–51). For the former, the origin of market exchange is unproblematic and developed out of a natural propensity to truck and barter goods. Marketplaces and market systems develop under specific conditions to increase efficiency by centralizing commodity access and facilitating information transfer concerning commodity availability and pricing. William Sanders (1956, 1962, 1968; Sanders and Price 1968) has posited what I view as essentially a neoclassical model of market development among towns in the Basin of Mexico, although he does not couch his argument in those terms. An important component of his model is that market exchange developed under conditions of population growth and ecological diversity, largely predicated on the assumption that humans “naturally” seek to maximize efficiency under conditions of scarcity—the underlying premise of neoclassical economics (Patterson 2003:79–82). In Sanders’s model, the ecological diversity of the basin led to an efficient system of exchange in which market centers were established in different ecological zones, resulting in increased resource availability and regional buffers against localized shortfalls (Sanders 1962:40–41). In Polanyi’s view, market origins have little to do with efficiency and scarcity but are rooted in sociopolitical concerns. As Carol Smith (1976b:45) eloquently explained, for substantivists market exchange “originates in long-distance trade between ‘stranger’ groups where economic maximization by the parties will not rend the social fabric. And it is expected to remain compartmentalized in preindustrial societies—to be a carefully controlled, circumscribed activity that takes place among strangers under the watchful eye of political watchdogs.” Marketplaces thus originated as administrative devices for regulating distribution and access to important resources. In considering a number of ethnographic case studies, Carol Smith (1976b:45– 46) found credible evidence for both the neoclassical and substantivist perspectives but also a common link that neither camp had considered. For Smith the development of a centralized social stratum of nonagricultural elites and specialists concentrated demand in the centers and generated market production of craft goods and, in rural areas, agricultural production of a marketable surplus (see also Plattner 1989c:180). Normal marketplace exchange is thus an outgrowth of the development of class, social hierarchies, and urbanization in some societies. Her model contains elements of both the substantivist and neoclassical perspectives: hierarchical social institutions provide a top-down impetus for market development (more akin to Polanyi’s view). Yet individuals pursuing their own economic interests and trading with local consumers 26

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

sustained market development and growth—a bottom-up process more in line with classical/neoclassical scholars’ celebrated “invisible hand” argument. The arguments and views expressed in this volume encompass a broad conceptual middle ground among the classical/neoclassical, substantivist, and Carol Smith perspectives. Abbott (Chapter 3) characterizes marketplace origins in the Hohokam region as a region-wide development in the absence of well-defined sociopolitical hierarchies, thus disputing both Polanyi’s and Smith’s arguments. His is a bottom-up argument more akin to neoclassical perspectives. Stark and Alanna Ossa (Chapter 5) suggest that the earliest evidence of marketplace exchange in the Mexican Gulf lowlands was not associated with the paramount center but instead with the smaller, secondary centers. They suggest that marketplace exchange may have developed in part in response to top-down pressures from secondary elites residing in the smaller centers but also from the bottom-up actions of small-scale specialists seeking economic advantage. Blanton and Fargher (Chapter 10), based on their comparative study, find that market systems tend to develop in the context of premodern societies with collective political organization (as explained previously), large population sizes (in the polity and the main urban center), intensive agricultural systems, and urban development. Among these variables, they find that urbanization and agricultural intensification are particularly important preconditions of market system development. In this sense their argument resembles Carol Smith’s (1976b:45–46) position that market systems tend to develop in contexts in which rural agricultural specialists produce surplus food to provision a growing nonagrarian population of urban specialists. Elite oversight was associated with the earliest tangible evidence of marketplace exchange along the Swahili coast in the late first millennium AD (Fleisher, Chapter 7). (Coastal traders may have established informal marketplaces earlier in time, but these “seasonal fairs” left no detectable archaeological remains.) Access to early marketplace exchange appears to have been tightly regulated by powerful clan members in the larger towns, who may have “privately” sponsored overseas traders to vend their wares to a restricted group of consumers, possibly high-ranking clan members. On the surface, Fleisher’s account may seem akin to Polanyi’s argument that marketplace exchange developed “as a carefully controlled, circumscribed activity” (C. Smith 1976b:45) to accommodate long-distance exchange among socially distant trade partners. As Fleisher explains, however, these exchanges were bound by “a system of oaths and blood brotherhoods,” which is hardly indicative of exchange among distant strangers. The explanation of market origins discussed so far focuses largely on the “macrolevel” contexts in which marketplace exchange first developed. Equally vital for understanding market origins is the “micro-level” of how and under what conditions individual consumers and producers accept and engage market institutions. Plattner (1989c:180–182) explored the conditions under which “premarket” producers would accept situations in which they give up relatively high levels of economic self27

C h rist o p h er P . G arrat y

sufficiency to pursue specialized market production. He lists eight essential preconditions. The first three can be considered “base” preconditions: (1) the market must be regularly scheduled and offer a predictable source for domestic necessities, (2) the quantity and assortment of goods must be adequate to meet domestic needs, and (3) market exchange must be protected against thievery or deceptive behavior. Five others can be thought of as supporting preconditions that promote market integration over a large area: (4) the transportation system must be sufficient to handle large-scale product trafficking, (5) storage facilities must be available for one’s stock of surplus marketable products, (6) communications must be sufficient to convey information regarding supplies and pricing, (7) relations among market centers must be peaceful to ensure low-risk product transfers, and (8) a willing and capable contingent of middlemen is needed to facilitate even distributions of market goods over a large area. The appeal of Plattner’s argument is that he, like Hirth (Chapter 11), focuses on the needs and circumstances of micro-level craft and food producers rather than on the macro-level of social benefits (Sanders) and institutional arrangements (Polanyi, Carol Smith). As has been emphasized, the development of trust and amicable social relationships among parties to a market exchange was also likely a vital component of market origins. The origin of marketplace exchange is thus partially rooted in the widespread acceptance of new social arrangements and the establishment of new interpersonal ties with a degree of trust (Granovetter 1985). More archaeologically elusive are the processes concerning rights of alienation over commodities (Clark 1995:286–287; Flad 2007; Y. Li 2007) and the socially acceptable application of “calculative agency” to economic transfers of goods (Callon 1998). Because these processes relate to changes in social attitudes and standards of acceptable behavior, archaeologists will find it difficult to detect empirical implications. However, ethnographic analogies or, perhaps, historical documents may convey useful information for modeling these behaviors. This matter should be an important focus of future research on premodern market development. Market Development and Change. Few scholars have proffered long-term developmental or “evolutionary” models of premodern market exchange. Hence, this topic represents a fertile growth area for future studies of premodern markets. The bestknown example among New World archaeologists is Blanton and colleagues’ developmental market model in Mesoamerica (Blanton et al. 1993:211–217; see also Blanton 1983; Feinman, Blanton, and Kowalewski 1984). They posit that market development was corollary to the development of large centralized Mesoamerican states, such as Teotihuacan and Monte Albán, and likely helped fuel their growth by providing tax revenues and an outlet for commodity conversion, as explained previously. In the case of Monte Albán they hypothesize that agricultural intensification, which was required to support the bourgeoning state and its growing urban population of nonfood producers, led to increased market participation. Specialized food producers increasingly 28

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

found it advantageous to obtain some nonfood goods from the marketplace rather than divert valuable household labor to produce them (Blanton et al. 1999:97–98; cf. Blanton 1983). In later times, after the large states had collapsed, market institutions persisted and evolved independent of direct state control, becoming an important component of Mesoamerican culture and society. By this time regional market systems likely possessed the organizational capacity to function, even in the absence of state support or government efforts to promote market participation (cf. Sahlins 1972). In Chapter 5 Stark and Ossa evaluate Blanton and colleagues’ (1993) model based on analyses of obsidian data collected from a regional survey in the Gulf lowlands of Mexico covering a roughly 1,200-year span (ca. AD 300–1520). They posit a modified version of their model that incorporates market development in a region with modest-sized centers. They specifically reconsider their argument for a “decoupling” of marketplace exchange from political control after the collapse of the large states (see also B. Stark 2007a; Stark and Garraty 2004:139–141). Rather than a linear trajectory of decoupling, they posit “neither solely bottom-up nor top-down instituted changes but rather a complex interaction of political and economic actions,” suggesting a more dynamic and nonlinear trajectory than that proposed by Blanton and colleagues. In a related perspective, Braswell (Chapter 6) models long-term developments in both political and market system organization in the Maya region during the Preclassic, Classic, and Postclassic periods (circa 1000–800 BC to AD 1500). Based on an interregional study of obsidian distribution, he infers diachronic patterns of market system expansion and contraction in the Maya region. Braswell juxtaposes market changes with Joyce Marcus’s (1993) dynamic model of Maya state expansion and contraction and concludes that periods of political centralization and growth correspond to lessdeveloped interregional commerce. Conversely, periods of political decentralization correspond to market expansion and increased interregional competition. Braswell’s diachronic model recalls Blanton’s (1976:259–261) argument concerning the linkage between market and state development. According to Blanton, central-place arrangements and functions in powerful states respond primarily to administrative concerns (e.g., to facilitate tribute collection or control key resources) and thus distort the optimal trafficking and information-processing requirements required for an efficient market hierarchy of central places. For this reason, he argues, market systems tend to thrive during periods of weak administrative control and falter during periods of strong administrative control. Several case studies in this volume highlight changes in the relationships of market centers and surrounding territories (Smith, Chapter 8; Stark and Ossa, Chapter 5), fluctuations in the sizes and extents of market systems (Braswell, Chapter 6), and the origins and collapse of a market system (Abbott, Chapter 3). The lesson from these studies is that no singular sequence of change is pertinent to all cases, in part because of the embeddedness of market exchange and the dynamics of political, social, moral, or other institutions and practices that promote or curb market activities. Rather than 29

C h rist o p h er P . G arrat y

a single trajectory of market development and evolution, these studies underscore the variability in market system organization and change resulting from different and culturally specific articulations between markets and social, religious, and political institutions, as well as moral codes of behavior and interpersonal conduct. Geographic and Cross-cultural Scope of the Volume

The contributors to this volume include archaeologists working in a variety of world regions and socioeconomic contexts. Research on early markets in Eurasia has long drawn interest from scholars in various disciplines dating back to Karl Marx (1964 [1857–1858]) and Max Weber (1976 [1909]), both of whom speculated about ancient Mediterranean and medieval European economies as precursors to capitalism. Interdisciplinary contributions to ancient Eurasian economies have been vigorous in recent years, with economists (Hudson and Levine 1996; Hudson and van de Mieroop 2002; Temin 2001, 2002, 2004), sociologists (Granovetter 2005b; Mann 1986), and even a policy analyst formerly at the conservative National Center for Policy Analysis (Bartlett 1994) all having a say. Recently, two classical archaeologists, Ian Morris and Joseph Manning (2005a), were invited to contribute to an updated edition of The Handbook for Economic Sociology (Smelser and Swedberg, eds. 2005). Those same authors (Manning and Morris, eds. 2005) incorporated commentaries from economic historians and sociologists in their edited volume on classical economies. In addition, part of a recent issue of the Journal of Institutional and Theoretical Economics (160, 4 [2004]) was dedicated to studies of classical economies. Undoubtedly, markets in Eurasia have received the lion’s share of scholarly and interdisciplinary attention compared with early markets and complex economies in the Americas, most of Africa, and East Asia—even though these areas also developed long-standing markets and market systems. The authors in this volume focus on a selection of less frequently studied areas in which understanding premodern market development relies principally on archaeological evidence and from which we may gain knowledge that broadens debates about markets. What underlies the areal imbalances in scholarly, especially interdisciplinary, attention? To some extent, a richer documentary record in Eurasia has allowed scholars to investigate premodern market exchange at an advanced level, addressing issues such as the development of banking and lending institutions, land privatization, maritime exchange, and aspects of the economy that are rarely, if ever, feasible in other areas of the ancient world. Archaeologists working in the Mediterranean region also enjoy several methodological advantages. For example, patterns of market exchange can be reconstructed from specific artifacts and features associated with commodity trafficking, such as coinage, weights and measures, shipwrecks, and amphorae for transporting olive oil or wine. Few such items are found in the archaeological record of the New World or in nonMediterranean areas of Asia or Africa, although the archaeology of ancient China has considerable potential in some of these categories of evidence. 30

I n v estigati n g M ar k et E x c h a n ge i n A n cie n t S o cieties

At a deeper level, however, I suspect that scholars in other disciplines look to Eurasian archaeology to shed light on the “ancestral” roots of Western capitalism (Hudson 2002:9–10, 2004:101). Market developments in sub-Saharan Africa, East Asia, and the Americas are not considered part of the economic lineage of today’s markets. In contrast with this view, I posit that careful cross-cultural study of “nonancestral” premodern markets in these world areas is equally instructive about early market development. A cross-cultural perspective provides a more appropriate and rigorous empirical basis for understanding long-term processes of premodern market development than does the study of any single world area. Avoiding the temptation of assuming a singular structure or developmental trajectory in premodern markets, this cross-cultural perspective rests on the idea that the behaviors relating to commerce and exchange were widely shared, and premodern markets in many parts of the world appear to have operated similarly and followed analogous developmental processes to some extent (see also Plattner 1989a:14–15 and C. Smith 1976b:59). This point is best exemplified in the writings and chronicles of early European explorers. When Hernán Cortés and his small army landed in Mexico in 1519, they were no doubt mystified by many of the practices and behaviors they witnessed among the indigenous peoples, yet they appear to have had little trouble recognizing what was occurring in the large, bustling daily marketplace at TenochtitlanTlatelolco, the largest and most powerful of the Aztec imperial capitals. Several Spanish witnesses provide detailed descriptions of the vendors, hawkers, and the varieties of (to them) exotic goods sold in the marketplace (Anonymous Conqueror 1969 [ca. 1500s]; Cortés 1986 [1521–1525]; Díaz del Castillo 1956 [1500s]; Durán 1971 [1581]; Sahagún 1950–1982 [1577], Book 8, ch. 19). Although modern scholars cannot rule out the possibility that Spanish chroniclers misinterpreted or misunderstood Aztec market behavior and applied their own cultural preconceptions to make sense of it (Lockhart 1985, 1992:445), the evidence does not suggest this was the case. Just because marketplaces in different times and places can be recognized does not mean they all operated or developed in the same way. Over 200 years ago Adam Smith (1976 [1776]) explained the development of market exchange as an outcome of natural human ambitions and abilities to barter. He may be correct, but humans have many “natural” propensities and abilities, and the issue in all cases for understanding their outcomes is the more complex historical and interactive effects of different economic conditions and cultural practices. Only with in-depth, cross-cultural, empirical study of the development of market exchange in different times and places can we evaluate Smith’s hypothesis and explore the causes of market development within a broader social and political context. Acknowledgments. This draft has been greatly improved by comments from Richard Blanton, John Clark, Gary Feinman, Robert Hunt, Michael Smith, Barbara Stark, and an anonymous reviewer. However, I bear all responsibility for the content. 31

C h rist o p h er P . G arrat y

Notes 1. Polanyi was not the first to oppose the application of classical economic ideas to the study of non-Western economies. A quarter century before Polanyi, Weber (1968 [1922]:164– 166) characterized market exchange as an inherently conflictive and socially disruptive form of exchange that developed only with the rise of “rational capitalism.” Other scholars also beat Polanyi to the punch. Several years before publication of The Great Transformation (Polanyi 2001b), D. M. Goodfellow (1939:5) remarked about the tendency among many anthropologists to question the application of “Western exchange economics” to the study of “primitive peoples.” Raymond Firth (1939) was an early advocate of this view and later remarked that Polanyi’s argument “came as no great surprise to many anthropologists” (Firth 1972:468). Polanyi succeeded in bringing the idea to a larger, cross-disciplinary audience, however.

32

Chapter Two

Detecting Marketplace Exchange in Archaeology: A Methodological Review

Barbara L. Stark and Christopher P. Garraty

Methods appropriate to archaeological studies are so crucial to research progress that we devote this review essay to the subject (see also Hirth 1998). Several factors have conspired to slow development of archaeological methods and deflect archaeologists’ attention from the subject. First, it is a difficult endeavor for which the most compelling approaches place a high demand on the amount and structure of data (Hirth, Chapter 11). Second, there is a considerable challenge from equifinality of patterns, particularly for marketplace exchange compared with centralized redistribution (Timothy Earle’s mobilizing redistribution [1977:215]; see also Blanton and Fargher, Chapter 10). Third, we lack parallel analyses of distribution patterns in state or imperial systems in which market system development was not prominent, such as the Andes. Fourth, some endeavors that might have sparked more interest in the investigation of market systems have instead led researchers to focus on other questions. For example, models from economic geography, including central-place theory (Christaller 1966 [1933]) and locational analysis (Chorley and Haggett 1967), have been adapted to address other, non-market research issues, such as political structure (Steponaitis 1978) and human ecology (Evans 1980). In this review we emphasize the importance of considering multiple spatial scales in the study of marketplace exchange, from localized activity areas to interregional interaction (Feinman and Nicholas, Chapter 4). We note the importance of multiple working hypotheses (Chamberlin 1965 [1890]), that is, consideration of alternative economic models, which is part of the problem of equifinality. We also emphasize 33

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

the importance of multiple lines of evidence, some of which operate at distinct scales. Much of our discussion of methods and equifinality hinges on inferring the archaeological expectations for marketplace exchange versus central redistribution or command economies, which involve dissemination of goods from a central location by authorities—both are allocation mechanisms that might produce an archaeological distribution pattern resembling that of marketplace exchange. Finally, we identify a companion to Kenneth Hirth’s (1998) household distributional approach to take into account a regional scale of production and distribution, in other words, specialization in production at some locations and a wide multi-community spatial scale of distribution, with sufficient quantities of items in regular household use that consumption involves most of the population in the area served. We designate this the “regional production-distribution” approach. Background: Multiple Exchange Mechanisms, Histories, and Equivalencies

Multiple mechanisms and spheres of exchange (Bohannan 1955) operated simultaneously in ancient complex societies, complicating the study of marketplace exchange. Marketplace exchange often coexists with reciprocity among kin (or other social groups), gift giving, household auto-production, communal share-outs, tribute, and labor service (Earle 1977; Halperin 1991; Polanyi 1944:53, 1957:250–256). Where it exists, marketplace exchange is of variable importance. In complex societies it is usually contrasted with central redistribution by governmental agents as a mechanism for large-scale economic integration. Central redistribution is often discussed with respect to “middle-range” societies. A cognate form of allocation in states is command economies (Ericson 2008; La Lone 1982:294–299), in which central government institutions set production objectives for many goods and services and, in modern cases, attempt to control the circulation of products to the population (Ericson 2008:1–2). Barter (transactions enacted without media of exchange) involves exchange with an eye toward gain and loss and, at a minimum, involves an element of market exchange even if a marketplace is not established (for definition of terms, see Garraty, Chapter 1). Barter is widespread (Dalton 1982:185), and regularized contexts for barter across a region may have afforded a historical antecedent to the establishment of marketplaces and market systems. Regularization of occasions for barter is one of a variety of scenarios for growth of market exchange, however (Garraty, Chapter 1). Instead of a “bottom-up” development, governmental or central institutions may also contribute to the growth of market exchange, for example, by sponsoring merchant expeditions (Hudson 2004). John Lie (1991) indicates for sixteenth- through eighteenth-century England that market development was not a smooth, linear process of growth and expansion in which new economic goods and sectors were continually integrated to the market domain (cf. Polanyi 1944); rather, it was nonlinear, with dynamic patterns of social regulation and ethical standards regarding self-interested behavior. 34

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

Barter or marketplace exchange may articulate with the timing of community rituals or other public events, allowing households to obtain provisions with multipurpose trips (Hassig 1982; see Abbott, Chapter 3). Frequently, marketplaces operated on a periodic basis in accordance with a ceremonial calendar. Archaeologists must remain vigilant for evidence of marketplace exchanges because such activities might be ancillary to other social or ritual events or an element of an economy in which central controls are prominent. George Dalton (1982:185) argued that barter is “minor, infrequent, petty, or emergency,” but he also implied that it may occur in a marketplace context (1982:186; see also Mayer 2002:143–171). We underscore this possibility, as there is no guarantee that media of exchange are used in every marketplace transaction; even with metal weights or coinage for exchanges, these media may not be widely circulated, and denominations may be inconvenient for small exchanges, which probably constituted the bulk of marketplace transactions (e.g., Hudson 2004:109; Humphrey and HughJones 1992:4–5; Mayer 2002:165). Enrique Mayer (2002:143–171) sees barter as a minor but important security measure for modern Andean peasants operating in a predominantly cash economy, while Caroline Humphrey (1985:57) notes that barter may be prevalent under certain conditions in her Tibetan study. Equivalencies are essential issues for the elaboration of marketplace exchange, but they seldom play a direct role in strictly archaeological studies. Exchange histories and entrenched conceptions of value were undoubtedly important for establishing expectations about equivalencies (Halperin 1994:134–135; Polanyi 1957:269). Michael Hudson (2002, 2004) has argued that large public institutions (temples and palaces) in Mesopotamia played a role in establishing equivalencies (and media of exchange) through tracking debt obligations linked to seasonal agricultural cycles and tribute. The prevalence of staple and wealth finance for ancient states (D’Altroy and Earle 1985) implies that public institutions were key players in the “top-down” establishment of equivalencies and regulation of weights and measures. Nevertheless, commoner barter and long-standing practices may also have provided a bottom-up basis for establishing equivalencies in premodern market contexts (e.g., for ancient Egypt see Kemp 1989:243). Although a fascinating subject in its own right, we do not consider the establishment of equivalencies a major hurdle for the development of market exchange, marketplaces, and market systems. Enlarging spheres of market exchange usually include some market transactions of foodstuffs or items that can be produced in incremental surplus to achieve successful exchange (see Plattner 1989c:180–182). Many items that can facilitate exchange are perishable, such as maize kernels in an Andean colonial example (Cobo 1979 [1653]:31, cited in Mayer 2002:58) or cacao beans in Mesoamerica. More durable items may also become media of exchange, such as weights of metals in Mesopotamia and Egypt (Hudson 2004:109; Kemp 1989:248), but coinage is not a necessary component of market systems and was not developed in many premodern markets, such as Mesoamerica. Media of exchange are usually not 35

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

a focus of strictly archaeological (as opposed to documentary) studies. For ancient economies, therefore, we stress inferred costs and gains rather than prices involving currency. Other forms of exchange contrast with both marketplace exchange and command economies but may form part of a dynamic history with marketplace exchange. In hierarchical societies, elite reciprocal gift giving and administered exchange of preciosities were common because they were intimately linked to communication and reinforcement of status, ritual prerogatives, and esoteric knowledge. (We define “preciosities” broadly to cover a range from items considered valuable for social status displays, to items of restricted ritual use, to items that represent wealth [i.e., luxuries; but cf. Muller 1997:41, 46]; preciosities may have socially restricted access or circulate widely.) For many preciosities, however, social restrictions mean that exchanges of some goods did not routinely lead to wider production and circulation, whether through marketplace exchange or central redistribution. Considerable social effort was devoted to controlling these goods, yet the “escape” of technologies and goods into wider social access did (and does) occur (B. Stark 2007b). Over time, wider access to preciosities may occur for a variety of reasons, which may provide conditions for them to enter into marketplace exchanges or satisfy tribute demands. Archaeologists have their best opportunity to study marketplace exchange through examination of circulation and disposal patterns for durable goods. Land and labor are less likely to have been commoditized in ancient agrarian states because they form the core of traditional livelihood and provide support for various state and civic institutions and nonfood producers (but see Silver 1983 regarding Mesopotamia). One stumbling block for archaeological studies of marketplace exchange is the modern tendency to define a market system as involving broad commercialization (i.e., market exchange of all or nearly all household commodities) rather than accommodating particular commercialized spheres (Mayer 2002:162) or degrees of commercialization (M. E. Smith 2004:78–80). Foodstuffs, raw materials, and fabricated items can be exchanged without undermining traditional landholdings and labor relations (see Eisenstadt 1993). The economic sector involving portable handicrafts, such as pottery vessels and stone tools, is a primary focus of archaeological studies of market exchange. The issue of how market exchanges are negotiated leads to a fundamental handicap that pervades archaeological studies: preservation drastically affects our abilities to track the exchange process. Particular durable products may be associated with a general locale or even a place of origin (e.g., through chemical or mineralogical provenance studies), but archaeologists can rarely document any counterflows of perishables (e.g., Abbott, Watts, and Lack 2007). Many possibilities exist, but the result is often frustrating—a one- or two-sided picture of the exchange system but not the full system. As a result, for archaeologists the ancient reckoning of supply and demand or calculations of gain or equivalency through transactions are usually not accessible. Archaeologists proceed with an assumption that market exchanges are conducted 36

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

with an eye to fluctuations in values, with traditional expectations about equivalencies based on long experience, mediated by social ties in the marketplace and sometimes by a degree of administrative regulation. Many factors can moderate supply and demand, as they do in modern economies. Archaeological approaches to marketplace exchange are challenging but no more so than identification of central redistribution or command economies because studies of these two practices must also control for the possible effects of market exchanges. Often, assessments of evidence cannot be entirely conclusive, in part because different mechanisms may be at work simultaneously or because we cannot assess enough categories of artifacts for which movement can be traced from known points of production (to determine if specialists were involved or all households produced their own). Usually, there is room for uncertainty and even more room for continued work on methods (Hirth, Chapter 11). There are expectations concerning the material distribution of items through marketplace exchange that help distinguish it from centralized redistribution, however, which necessarily emphasizes storage of staples and wealth valuables, as we discuss subsequently (Hirth 1998; see Feinman and Nicholas, Chapter 4). The Current Methodological Context The Curse and Blessing of Documents

Studies of marketplace exchange using material remains have undergone substantial change during the current century. Previous investigators mainly paid attention to markets and marketplaces in places and times when they became prominent and well documented in written sources, such as classical antiquity and the ancient Near East (e.g., Hudson 1996, 2002, 2004; Manning and Morris, eds. 2005; Scheidel and von Reden, eds. 2002; Silver 1983; see Garraty, Chapter 1). This is also true of India from AD 500 to 1600 (Morrison 1997:97–102), where written records document active marketplaces along with specialized commercial occupations (such as long-distance merchants). Temple and other institutional records have been used to argue that, in some periods and societies, market exchange did not exist or was overshadowed by centralized institutional practices (Hudson 2004:101, 110, 113–114; La Lone 1982). Often, such historical approaches fail to consider an agency perspective in which the activities of people in central institutions may contrast with the endeavors pursued at household or community levels, which may be poorly represented in written records. Such divergences may create complex economic structures in ancient economies. Historical documents have been a mixed blessing because many researchers working in areas with copious records of marketplace exchange have failed to tackle the archaeological data in their own right, such as Karl Polanyi’s (1960) seminal analysis of Greek agorae. Richard Wilk (1998:469) is correct that many archaeologists tend 37

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

to study market institutions as presence-absence phenomena rather than explore the variability in the organizational complexity and diachronic histories of premodern markets. The “combination situations,” in which documents suggest that market systems existed, are an ideal circumstance to scrutinize how archaeological data behave and what new insights they reveal that are not attested in the documentary record. Michael Smith’s Mexican study (Chapter 8) provides an example. Researchers working in highland Central Mexico have evaluated pottery distributions during the Aztec era when vigorous periodic marketplaces are described—but pottery distributions are not a subject covered by Spanish chroniclers (e.g., Garraty 2006, 2007; Hodge and Minc 1990; Minc 2006, 2009; Nichols et al. 2002), just as Roman pottery is not featured in Roman written sources (Green 1986:167). Unfortunately, the effort required to investigate marketplace exchange archaeologically has often led to a stalemate, as many archaeologists have opted to take markets “for granted” because of documents and moved on to what they regarded as more troubling economic research arenas, such as craft specialization or agricultural intensification. A considerable impetus toward investigation of market development has come from investigators working in earlier Mesoamerican periods without direct documentation in colonial Spanish chronicles (Blanton 1983; Blanton et al. 1982:23–25, 55–61, 65–68, 207–208; Feinman, Blanton, and Kowalewski 1984; Feinman and Nicholas 2004a; Hirth 1998; Kowalewski et al. 1989:294; Sullivan 2007; various authors, this volume) but for which researchers could expect that market development was a legitimate issue for the economy. Most of these researchers have drawn upon regional-scale survey to varying extents or upon extensive systematic urban surface collections, such as those from Teotihuacan (Sullivan 2007) and Monte Albán (Blanton et al. 1982). One of archaeology’s great strengths—the examination of long-term change, usually in the context of regional-scale settlement pattern projects—has led to a realization that the history and development of market activities are vital topics (Blanton 1983, 1985, 1996a; Stark and Ossa, Chapter 5). Even for earlier Mesoamerican periods, however, marketplace exchange could work as an assumption, not a focus of investigation (e.g., see summary in Santley 2007:151–174). Long-standing Approaches

Central-Place Theory. Sherlock Holmes was intrigued by the “dog that didn’t bark,” and we can similarly ponder why the wave of archaeological interest in and experimentation with Walter Christaller’s (1966 [1933]) central-place theory during the 1970s and 1980s did not yield a corresponding gain in analyzing market exchange and market systems. This version of central-place modeling is explicitly concerned with the spatial implications of retail market exchange under particular (simplified) conditions (see summaries in Plattner 1975; C. Smith 1976b). Retailing of lowerorder (frequent consumption, relatively low cost) and higher-order (less frequent, more costly) goods or services is idealized with an imagined evenly distributed popu38

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

lation in an undifferentiated environment (“isotropic plain”). The result is a lattice of evenly spaced centers that provide lower-order goods. These centers articulate in regular hexagonal patterns with larger centers (functionally) that host retail of lowerand higher-order goods and services. Price competition regulates the even spacing and hierarchical relationships of retailing. Different articulations are possible, depending on whether marketplace access, product transport, or administration is optimized. A wave of interest in central-place theory resulted in struggles to account for real-world distortions (watercourses and irrigation canals [ Johnson 1972, 1975], mountains and lakes [Blanton 1996a; M. E. Smith 1979], and incomplete survey data [e.g., Marcus 1973]). Other struggles revolved around discrepancies in the size of settlements versus their lattice positions (e.g., Blanton 1996a; Johnson 1972:776; M. E. Smith 1979:esp. 116) and whether factors other than market system development—including political or ecological factors—accounted for settlement rank and position (Blanton 1996a; Evans 1980). Other scholars argued that in some contexts, such as chiefdoms, tribute was a more powerful predictor of spatial patterns than market activities were (Steponaitis 1978). Some researchers focused on determining whether the larger centers were evenly spaced and, if so, interpreted this pattern in terms of transport costs. At the same time, discussions of marketplace exchange essentially faded as researchers became caught up in whether hierarchical or spatial regularity was present (e.g., Hammond 1974; Inomata and Aoyama 1996; Marcus 1973; Marcus and Flannery 1996:174–175). In some cases analysis and discussion hinged on exchange, without indicating specific exchange mechanisms (e.g., Johnson 1975). Ian Hodder and Mark Hassall (1971), Michael Smith (1979), and Richard Blanton (1996a) are notable exceptions for employing central-place theory and settlement lattices to infer the importance of commercial exchange, although each dealt with periods for which documents indicated a market system. As Hodder and Clive Orton (1976:55, 63) have noted, the background issue for applications of central-place theory is usually a more vague form of competition (e.g., for clients, tribute receipts, or lands) coupled with transport minimization (for travel to centers for a variety of functions, not just market exchanges); the foreground issue is the spatial regularity of a settlement hierarchy. A degree of regularity in settlement spacing for different levels in the settlement hierarchy may result from factors unrelated to marketplace exchange, such as ecological conditions (e.g., access to water or prime farmland), hierarchical governmental administration and its population base, and transport costs (see Hassig 1985:11–40). The possibility of multiple bases for competition has created a problem of equifinality, leading to a shift in focus from marketing to other criteria for settlement decisions. How can researchers assess the spatial distribution of centers as a result of market system development? The primary tool for quantifying and evaluating settlement spacing has been nearest-neighbor analysis, which can, with various caveats, detect whether settlement departs from random toward an even or a clustered distribution (Hodder and Orton 1976:38–51), but this method does not address the articulation 39

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

among different levels of the idealized hierarchical lattice and their positioning relative to one another. The irregularities of actual data and the diminishing numbers of centers in the uppermost levels of the settlement hierarchy (even with regional surveys) combine to create sample size problems as well. These methodological hurdles remain, and better methods are needed for quantifying observed versus expected spatial patterning in a central-place lattice. The “eyeball method” and “regularization” to make lattices more like Christaller’s ideal for visualization purposes did not provide credible evidence supporting a central-place model. Consequently, central-place studies provided a strong impetus for researchers to continue probing settlement hierarchies and how best to model them (with a lively branch of investigation devoted to the rank-size rule [e.g., Drennan and Peterson 2004]). In sum, central-place theory failed to take hold in archaeological studies of market exchange because of the vexatious problem of equifinality as well as the restricted situations with sufficient empirical settlement data. Christaller initially designed his models as a means of representing market system structure and the principal factors affecting that structure (e.g., market access, transportation), but subsequent studies have shown that many factors other than marketing affect the spatial distributions of centers over the landscape. These other factors, including political considerations (e.g., degree of hierarchy) and ecological concerns (e.g., access to water or good farmland), largely replaced marketing as a central focus of settlement patterns and regional landscape studies. Regional Falloff Analyses. Falloff (monotonic decrement) or distance-decay models represent another contribution to exchange studies, but they also failed to spark a flurry of interest in marketing—another dog that didn’t bark. Distance-decay models were initially posited for archaeology by Colin Renfrew (1975, 1977), who focused on spatial distributions of exchanged goods in relation to distance, examining the falloff, or decline in amount, from locales of origin as transport costs increased. Renfrew (1975) identified ten distribution and trade models, from direct access to the source (supply zone behavior) to ports of trade, including a marketplace exchange model. His idealized models underscored problems of equifinality, that is, that some distributive mechanisms or processes could result in very similar spatial distributions (a subject further elaborated by Hodder and Orton [1976:126–155]). In Renfrew’s (1975:42–43) portrayals, central-place redistribution and central-place marketplace exchange exhibit similar falloff patterns (see the next section). Consequently, marketplace exchange per se did not seem to be a promising line of investigation, as it could not be distinguished from central-place redistribution. Renfrew did not model localized distributions from a marketplace, as Hirth (1998) later addressed (e.g., Stark and Ossa, Chapter 5). He did note the action of “freelance” or “middleman” traders, defined as mobile traders who extend or improve access to goods away from the area serviced by market or distributive centers (Renfrew 1975:42, 43, 48–51). This last observation also failed to spur a focus on marketplace 40

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

exchange because it did not overcome the equifinality of distinguishing between it and central redistribution in product dissemination. Renfrew (1975:52) remarked that central storehouses would be required to support redistribution but did not clarify if he sees such storehouses as the sole means of discriminating central redistribution from market systems. He also pointed to craft specialization as an indicator of central redistribution (Renfrew 1975:52), but it can be expected to relate to marketplace exchange, centralized redistribution, and other non-market exchange mechanisms. Significant surplus production is a requisite for marketplace exchange (or central redistribution) if the product is utilized by many households, in contrast to embedded specialists supplying their own social unit, specialists commissioned by elites to make prestige items, or household auto-production. The absence of obvious storehouses in the Oaxacan archaeological record (Feinman and Nicholas, Chapter 4) is one of the ingredients of the case for marketplace exchange there, just as the presence of state storehouses is a factor in arguments for a command economy (discussed later) and against the development of a market system in the Inca region (E. C. Morris 1986). In sum, Renfrew’s spatial models, like central-place theory, suffer from the problem of equifinality. That is, spatial falloff patterns relating to marketplace exchange are not readily distinguishable from other forms of product distribution, especially centralized redistribution as originally defined. Additional archaeological data are necessary, as discussed later in this chapter. As Hirth (1998:455) noted, both forms of distribution are centralized, but, unlike marketplace exchange, redistribution adheres to social and political networks and does not generate socially widespread distribution on the same scale as marketplace exchange (Garraty 2009). Hence, researchers need to be cognizant of the social and political status of consumer households, a point not accommodated by Renfrew’s generalized falloff models or central-place models. Site-Level Artifact Assemblage Similarities. As a harbinger of Hirth’s distributional approach, Robert Fry (1979, 1980; Fry and Cox 1974) used multidimensional scaling to analyze the relationships of technological and stylistic traits for excavated ceramics from different locations in the greater Tikal area in the Maya lowlands of Mesoamerica. Fry (1979:497) developed expectations for different exchange mechanisms and for varying numbers of production and distribution areas. Assemblages that were relatively similar to one another but independent of geographic distance from the Tikal center suggested operation of a centralized marketplace system. Fry (1979:500) was concerned with market versus social factors affecting artifact distributions, but he observed no noteworthy effects of social class on collection composition. Fry’s alternate models take into consideration localized production and consumption (“supply zone”), gift exchanges, decentralized marketing with graduated distributions from several production areas, and centralized redistribution. His samples are derived from residential testing along transects extending 21 km across the dispersed Tikal settlement. Fry glosses over how centralized redistribution contrasts 41

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

with centralized marketplace exchange, the primary issue addressed by Hirth (1998). Further, he does not present examples of how different distribution mechanisms would be evident in results derived from multidimensional scaling. Fry (1979:495) addresses the issue of equifinality, but the expectations for the models only partly solve the problem because it is difficult to interpret the degrees of resemblance (as presented graphically) and the stress measures for the multivariate plots in terms of the models. His data show that multidimensional scaling is effective for suggesting distinct production and distribution locales when contrasts in attributes are marked (for example, the Jimbal area during one period). Ultimately, Fry (1979:509) takes recourse in the logic of the regional production-distribution approach explained in the next section: that the scale of exchange in both plain ware and decorated serving vessels makes marketplace exchange more likely than central redistribution. Current Approaches

The aforementioned methods of detecting marketplace exchange and market systems suffered from problems of equifinality, especially concerning the spatial and artifact distribution patterns resulting from marketplace exchange versus centralized redistribution. More recent approaches have sought to combat the equifinality problem by considering multiple lines of evidence and by evaluating marketplace exchange at various spatial scales. Especially influential has been Hirth’s (1998) household distributional approach, which emphasizes the implications of marketplace exchange at the level of individual households. Building on Hirth’s approach, other scholars have recently developed new methods of detecting marketplace exchange through detailed analyses of stylistic variability in decorated ceramics, the extent of style zones (Minc 2006, 2009), and statistical measures of inter-household artifact diversity (Garraty 2009; Stark and Ossa, Chapter 5). We suggest yet another approach to detecting market system development based on regional-scale patterns of production and exchange over a large area. We refer to this as the regional production-distribution approach, which focuses on the consistency of access by ordinary households to specialist crafts over a large area. Hirth’s Household Distributional Approach

Hirth’s (1998) distributional approach addressed the archaeological implications of marketplace exchange at the household and settlement scales. He argued that marketplace exchange makes a wide range of products available to all households, tempered by differences in wealth (see also M. E. Smith 1999). Marketplace exchange thus contrasts with exchanges conducted through clientage or other social ties (e.g., centralized redistribution), for which class and prestige considerations restrict social access for certain items, resulting in patchy distributions and marked differentials in access. In addition, centralized collection and redistribution of goods presumably 42

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

required large storage facilities not normally associated with marketplace activities. In several respects Hirth’s (1998) proposed method for detecting marketplace exchange has been a turning point, sparking greater interest. Yet to be addressed, however, are the ways such factors as ethnic or occupational identities might skew distributional patterns even if marketplace exchange is important. Ethnographic studies of marketplaces are replete with evidence for exchanges along ethnic lines (e.g., Russell 1987; Schwimmer 1979; Seligman 1993; see Garraty, Chapter 1). Hirth’s analysis required systematic sampling of households throughout Xochi­ calco, Morelos, Mexico. Increasingly, archaeologists have relied on systematic surface collections and observations to record centers, not just as constellations of public buildings but as ancient cities, with a corresponding focus on residential areas and the full suite of observable settlement activities and specializations (e.g., Charlton, Nichols, and Charlton 1991; B. Stark 2003, 2007a). Thus, changing archaeological practices opened up new avenues for studying exchange and household provisioning, including marketplace exchange. A recent example in Mesoamerica draws upon the Teotihuacan Mapping Project data plus subsequent excavations and surface collections to argue for marketplace exchange of ceramic figurines and censers or censer parts (Sullivan 2007). Barbara Stark and Alanna Ossa (Chapter 5) draw upon regional residential data to examine possible marketplace service areas for various centers in south-central Veracruz. Their approach blends Hirth’s household distributional method and the regional production-distribution approach discussed in the next section. Distributional studies can be pursued within settlements or across a region and rely on provenance information to understand product origins because of the potential for replicas or alternate sources. Craft production locations also play a vital role with respect to origins. A Regional Production-Distribution Approach

Regional-scale surface and excavated collections, combined with other lines of evidence—mainly indications of specialized production of quotidian household gear—potentially reveal marketplace exchange. The combination of regional-scale data, artifact provenance studies, and stylistic information creates an effective parallel and complement to Hirth’s household- and site-level approach. Feinman and Nicholas (Chapter 4) provide an example, asking whether articles used regularly in commoner households across a region but that were produced in particular settlements or specialist households could be efficiently supplied through redistribution or other centrally administered forms of product provisioning. This is a key question, and the answer is generally “no.” A seeming exception is large temple or governmental grain stores that offer flexibility in funding central programs and mitigating shortages or famine (Hassig 1981; Hudson 2004:122). Nevertheless, famine relief is a special case for central action, and underwriting central programs is more characteristic for such storehouses. 43

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

Households might have been able to make their own pottery or procure raw material to make stone tools, for example, but archaeological data indicating that these items were produced in large quantities by specialists and reached households in a district or a region on a reliable basis clearly negate household auto-production and small-scale reciprocity or gift giving. It is unlikely that ancient states or imperial powers invested in regular household provisioning of quotidian items or possessed the logistical capabilities for such large-scale dissemination (see Stanish, Chapter 9). Ancient states had a considerable agenda of political, ritual, military, and class-related goals, with challenges to assure sufficient revenues; provisioning of quotidian goods would not have contributed to any of these objectives and, in fact, would have been exceptionally costly. Thus, either considerable community self-sufficiency (via kin or other internal reciprocities) or alternate mechanisms such as market exchange may have been encouraged or tolerated as solutions. Centrally sponsored production and centralized redistribution lead to circulation of select items, mainly wealth goods or celebratory food and drink, that generally adhere to class or status lines or support state personnel; more general access to some items, such as foods, usually occurs when workers or armies are provisioned during government-sponsored projects (Dietler and Herbich [2001] discuss food and labor issues). In the Roman empire, for example, Kevin Greene (1986:164–165) has argued that large-scale distribution of utilitarian pottery evidenced market provisioning of domestic wares. Mesoamerican regional surveys in the Basin of Mexico and the Valley of Oaxaca provided data at a sufficient spatial scale to detect areal patterns of ceramic type concentrations and distributions anchored in quantitative stylistic and material characterizations (e.g., for the Basin of Mexico, Garraty 2006, 2007; Hodge and Minc 1990; Minc 2006, 2009; Nichols et al. 2002; for Oaxaca, Appel 1986). In these regions, which encompassed multiple polities or communities, investigators have detected the circulation of pottery at a scale (distances and amounts) and social inclusiveness unlikely to be achieved through down-the-line gift giving, community reciprocity, sporadic barter, or central redistribution. A combination of information about locales of specialist production, stylistic characterizations, and geological or compositional provenance ascriptions is crucial to these studies (e.g., Hodge et al. 1993; Nichols et al. 2002). Regional-scale studies are well represented in this volume, such as David Abbott’s (Chapter 3) study of market exchange among the Hohokam in the U.S. Southwest (see also Abbott, Smith, and Gallaga 2007; Abbott, Stinson, and Van Keuren 2001; Abbott, Watts, and Lack 2007). He and his colleagues’ interpretations do not stem from regional “complete coverage” surveys, such as those mentioned previously, but rather from a considerable range of excavated sites representing different communities and different levels in the settlement hierarchy, including numerous investigations from cultural resource management projects. In this sense, their studies resemble Geoffrey Braswell’s (Chapter 6); he uses collections from Maya centers and other large sites, rather than obsidian distribution over a continuous area, to draw conclusions about the broader market system (see also Braswell and Glascock 2002). 44

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

Operating with data from a late period when markets are indicated from postconquest Spanish accounts, Smith (Chapter 8) considers artifact distributions at several sites in Morelos in highland Central Mexico. He evaluates market system structure using cluster analysis and assesses distributions of imported pottery to interpret diachronic changes in the scale of interlocking marketplace service areas before and during Aztec rule. A mix of analytical scales is featured in Stark and Ossa’s (Chapter 5) examination of variability in obsidian blade access with increasing distance from centers in south-central Veracruz, using collections from residential units to detect marketplace service areas. Information about the contexts of intensive craft production, along with other information, has figured into regional production-distribution arguments. For example, specialized production of fine and utilitarian vessels, wide distribution of distinctive pastes and vessels used by most households, lack of sizable storage facilities, and accessible plaza areas that could have accommodated periodic marketplaces together form a powerful package of evidence about the inception and duration of marketplace exchange in the Valley of Oaxaca (Feinman and Nicholas 2004a). Feinman and Nicholas (Chapter 4) characterize this range of information as a multiscalar approach to inferring marketplace exchange and add household production patterns and marketplace plaza sizes to the mix of data. The pivotal element in these analyses is a regional distribution of artifacts combined with other lines of evidence, especially specialized production loci. Such data are similar to Hirth’s household distributional approach; both share the expectation that domestic provisioning of mundane goods through marketplace exchange is intrinsically different from centralized provisioning through mechanisms skewed by political position or social rank. A regional market system allows most households to obtain everyday goods available in marketplaces, although some caveats apply: wealth differences, special activities, perhaps distinct ethnicities, and site function may skew distribution patterns. In the next section we take a closer look at how the evidence required for inferring central redistribution or command economies differs from that required for inferring marketplace exchange. Central Redistribution and Command Economies versus Marketplace Exchange

The proposition that centralized redistribution effects wide access to items within a community or a polity has a checkered history in anthropology. As first described by Polanyi (1944:43–55, 1957), redistribution involves more than one process: it includes the centralization of resources at the community level and shared out via leaders and the centralization of resources to fund larger political endeavors in chiefdoms or states (especially benefiting elites). Redistribution contrasts with the principles of reciprocity and householding Polanyi identified and was later described by Elman Service (1962, 1975) as a social benefit provided by chiefs, who moved 45

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

products among different resource zones or areas of production, assisting households generally while sustaining chiefly authority. In a pivotal study, Timothy Earle (1977, 1978) challenged Service’s ecological and beneficent view of chiefly redistribution, showing with documentary data for premodern Hawai’i that resources were mainly drawn in to support the chief, his clients, and his civic and ritual obligations. In turn, local communities practiced considerable self-sufficiency in subsistence activities.1 From Earle’s (1977, 1978) study, it is clear that mobilizing redistribution in chiefdoms (or “middle-range” societies) is not the kind of mechanism that facilitates delivery of ceramic cooking vessels, stone tools, or other implements for daily subsistence to households or that routinely patches over subsistence shortfalls; nor does it provide a reliable mechanism for evening out subsistence resources within ecologically diverse regions, as Service (1962, 1975) proposed. Rather, communities strive for a considerable level of self-sufficiency. Kin and local reciprocal exchanges or share-outs are mechanisms that are controlled at the local level and involve less risk for reliable circulation than does central redistribution, which has a different agenda. In middle-range societies the authority and functions of leaders, including as administrators of central redistribution, are highly variable (Feinman and Neitzel 1984). A less frequently discussed stepchild of centralized redistribution (among archaeologists) is the command economy, a concept adapted by Darrell La Lone (1982) for the Inca case to contrast with the idea of supply and demand in market exchanges. For ancient states a command economy was state-run, and government institutions operated with varying balances of both staple and wealth finance (D’Altroy and Earle 1985) and substantial use of labor service, but market exchange was either absent or played a relatively modest economic role. As with chiefly redistribution, resources sought by governing institutions were predominantly garnered for supporting authorities and their projects (whether ritual or military) and for political clientage. For example, the Inca drew upon tribute (foods and fabricated items) through labor service and controlled production and distribution of certain crafts to support state endeavors. John Murra (1980) described state storehouses, relocation of groups to particular ecological zones to ensure resource production for the state, control of some specialist craft activities, and allocation of resources to dependent leaders or communities, but he did not mention marketplace exchanges, a contentious issue for the Inca (Earle 1985; La Lone 1982). In other command economies, scholars recognized mechanisms of exchange besides state action, including market exchange, but argued that they played a modest role (Hudson 2004; Kemp 1989:232–260). In comparison, states with active marketplaces also relied on tribute and command of labor for government projects, as exemplified by the Aztecs (Berdan 1977, 1985, 1988; Hicks 1984). The key is the mix of different allocation mechanisms and their articulation within the broader economy. We cannot assert that any one economic mechanism is “more efficient” than another because each solves certain problems for different seg46

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

ments of society, poses certain risks, and functions within particular contexts—from household to province to state and in terms of diverse social groups as well, such as ethnic or occupational groups (Hirth, Chapter 11). Opposed and intertwined provisioning strategies among different groups and factions in complex societies are part of the basis for diversity in economic mechanisms. Whereas many mechanisms for production and distribution might operate to provision at small scales, such as within a community, settlement systems with towns and cities across a region substantially augment the overall costs of transport and information flows. Regulation of exchange operates very differently under market systems than it does under governmental control, even though in each case centrally located places are efficient in reducing transportation costs, especially with multipurpose trips (Stanish, Chapter 9). Modern command economies have shown how challenging regulation of production and distribution to the general population can be. G. William Skinner (1964, 1965) remarked on the inefficiencies, waste, and economic hardships introduced in the People’s Republic of China during governmental administration of production and distribution previously handled through a periodic market system (see Ericson 2008 for a brief Soviet summary). The Inca state, for example, likely avoided such excesses in part because of a compensating emphasis on local community self-sufficiency. Importantly, there was no single expression of economic centralization for states and empires. The Inca represented a new degree of emphasis on central controls and storage in the Andes (LeVine 1992:15–16). For ancient Mesopotamia, Susan Pollack (1999:81–148) has described changes regarding the roles of central institutions (mainly temples) in production and distribution. Temple administrators eventually became more involved in distribution than in production, but subsequently an “oikos economy” ensued in which large institutions (mainly temples as the households of the gods and wealthy kin groups with significant landholdings) provided a more decentralized redistributive economy in which each unit sought to secure both production and internal distribution to members (kin, clients, dependents, and slaves, with the lower statuses providing a large labor force)—although surpluses might have entered the market. Thus, powerful institutions played a variety of redistributive roles over the long term. We expect no single economic mechanism in ancient states but rather a range of mechanisms in dynamic relationships propelled by different interests and loci of decision making (Garraty, Chapter 1). This is particularly true for expansionist states and empires in which new provinces may have had different economic emphases from the imperial core. Tribute (or taxes) and, usually, labor service funded the ancient state, but the importance of other mechanisms, such as market exchange, was rather variable. The archaeological examples discussed next are selected to show (1) the effects of scale on redistributive economies, (2) variation according to the products in question and the social and environmental contexts of use, and (3) problems of equifinality. 47

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

Archaeological Analyses of Central Redistribution versus Market Exchange

Although chiefly redistribution and state command economies seem to contrast with marketplace exchange in respect to their goals and functions, a key issue remains: can or does central government provision households on a broad basis and, if so, under what circumstances? Scrutiny of this issue requires differentiation of artifact or resource characteristics. The idea of central redistribution in chiefdoms highlights the contrasts between preciosities and utilitarian items, but in fact a complex web of artifact characteristics warrants different expectations concerning exchange mechanisms (Carr 2005; Pires-Ferreira and Flannery 1976). Different categories of artifacts or resources contrast in uses, use-lives, depositional frequencies and contexts, costs, symbolism, weight and fragility, and circulation. A dichotomy between preciosities and utilitarian items is convenient only as an initial characterization (Smith and Berdan eds. 2003b). Christopher Carr (2005) examined exchange and procurement based on demography and distance, with a local scale defined as neighboring communities with a radius of 80 km or less, a regional scale with an 80- to 320-km radius, and an interregional scale with a radius of over 320 km (distances reflect some access to waterways for transport). Adapting concepts from Mary Helms (1988), Carr proposed that the local scale involves communities well-known to each other, the middle scale involves “close strangers,” and the largest scale is composed of “foreigners.” Reciprocities, for example, are most active at the local scale, that is, within communities and among neighboring ones, with some reciprocal exchange at a regional level. In contrast, knowledge quests and pilgrimages are likely to operate on an interregional scale. Marketplace exchange will be most active at the local scale for individual participants but may form a large regional or interregional system of interlocking periodic marketplaces. The examples selected here for comparison with marketplace exchange are rooted in householdscale data and operate with at least the scale of a marketplace service area for everyday items, circa 6–20 km. Recent work reveals considerable variation in the ways items are produced and circulated in middle-range societies. We use Carr’s (2005) distinctions to calibrate scales for a selection of research with synthetic treatments, mainly from North America and Mesoamerica (Alden 1982; Aoyama 1999; Bayman 1995; Carr and Case, eds. 2005; Mallory 1984; Milner 1998; Muller 1997; Winter and PiresFerreira 1976). In line with the revisionist argument developed by Earle (1977), a key expectation is that access to certain goods is highly differentiated by class or rank as a result of gifts and patronage from leaders or ruling cadres in a centralized redistribution system. Kazuo Aoyama (1999:51, 133–134), James Bayman (1995), John Henry (2004:91), John Mallory (1984:192–193), and Raymond Sidrys (1976:456, 458) concur with this expectation (contra Alden 1982). This expectation is compatible with the Inca emphasis on tribute in, and state-controlled production of, cloth (Murra 1962), chicha beer (E. C. Morris 1979), and decorated Inca-style vessels (D’Altroy 2001c:263– 264) as items of particular value or symbolic import expressing state power. 48

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

Archaeologically, we might encounter spotty distributions or clustering in a centralized redistribution system; gifts to clients can percolate through reciprocal kin exchanges to reach others of lesser rank but in diminishing amounts. During sponsored feasts or ceremonies, foodstuffs are likely to reach a wider segment of the population on occasion, but this organic largesse can rarely be traced archaeologically to recipients, and it is not part of a reliable daily subsistence regimen. Particularly in societies with a modest scale of central power, what is preserved archaeologically may not reflect culturally determined sumptuary practices because perishables are emphasized in social distinctions (e.g., cloth, special foodstuffs, or colorful feathers; Turkon 2004); alternatively, leaders may have been distinguished more by authority in rituals or war or by marshaling labor instead of by possession of durable preciosities. The early villages in the Valley of Oaxaca provide an instructive example of smallscale redistribution in household contexts and indicate that differentiation in access may not be striking if social ranks are not highly contrastive. At first the distribution of imported obsidian flakes and cores from different geological sources was uneven among households, but later, especially as obsidian blades were imported, the distribution became more homogeneous among households. This change suggests an increasing role for leaders in the larger village, San José Mogote, in arranging for importation and distribution of obsidian, gradually replacing the diversity in trading partners among individual households (Winter and Pires-Ferreira 1976). The imported obsidian was used for utilitarian tasks because of its particularly sharp cutting edges, and every household had access to it, but other local materials—such as cherts—formed the majority of domestic stone tool assemblages. Centralized obsidian distribution reached a smaller subsidiary village, Tierras Largas, located 10.3 km away. In Oaxaca, central arrangements made by emerging leaders accomplished distribution of a utilitarian good to all households over a distance comparable to marketplace service areas because the item was lightweight, and households maintained small inventories with modest annual attrition. Across greater distances and population scales, obsidian could reach all households through central redistribution. As we will discuss in more detail, Aoyama (1999) noted this pattern in the Copán area, where marked skewing in access is evident according to the status of households. In other instances obsidian access does not appear strongly affected by status, and wide access is indicated when obsidian is the main source of cutting tools, suggesting marketplace distribution (Stark and Ossa, Chapter 5). If redistribution or command mechanisms were to deliver pots to the kitchen or stone tools to the home on a substantial spatial and demographic scale, we would have to further examine the periodicity in which such distributions occur compared to the breakage, wear, and replacement rates, as well as their transportability and the quantities needed for each household. From the perspective of central government, official occasions involving ceremonies or work parties provide convenient opportunities for distribution to commoner households. Nevertheless, households may experience gaps 49

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

in their supply lines and, if so, would be required to stockpile or produce their own items (or exchange with reciprocating kin). Central storage on a substantial scale is a by-product of successful state redistribution because stockpiles of various sizes and kinds of vessels and other tools consume considerable space if the population served is sizable. Stockpiles and storage should be associated with central government facilities, as noted for the Inca, Mesopotamia, and Egypt (Hudson 2004; Kemp 1989:185–197; LeVine 1992; Stanish, Chapter 9). In these cases stored items appear to be political tokens or food for supporting state officials, work parties, or soldiers, however, rather than items for commoner daily life. In many respects, pottery is an ideal archaeological material for comparing exchange mechanisms because continued use invites breakage, and a considerable household toolkit (storage, cooking, and serving vessels) creates demand; pottery is considered in several chapters in this volume. In contrast, obsidian tools—also considered in several chapters—are typically present in households in small amounts, the individual items are relatively lightweight, other materials are sometimes available, and only a few tools are required each year for a household toolkit (Sanders and Santley 1983). Obsidian tools, such as prismatic blades in Mesoamerica, might be handled more readily during redistribution events than would pottery used daily (see Bayman 1995 for an example from the U.S. Southwest), although they could also be distributed through marketplace activity. Mallory (1984:253) accepted the idea that a range of mechanisms, centralized redistribution, peddlers, or something akin to peripheral marketplaces (Bohannan and Dalton 1962) could supply obsidian blades to households in the Copán Valley. Several studies show that leaders in middle-range societies do not always play a crucial economic role in the circulation of goods, whether preciosities or utilitarian items. For Hopewell societies, initial expectations about elite-sponsored connections for long-distance importation of preciosities have not been borne out by provenance studies, and more localized production and circulation are indicated (Carr 2005). For Mississippian society at Cahokia, relatively widespread production and access were discovered rather than a close or exclusive connection with elites, although persons of high rank had more and better-made articles (Milner 1998:161–162). The economic roles of Mississippian elites remain uncertain, however, with some scholars emphasizing strong central leadership (e.g., for Cahokia, see Pauketat 2004:94–95, 100–106) and others proposing alternative economic scenarios that are not dependent on central governance. At Cahokia, John Kelly (2006) has suggested that sequential production steps articulating social groups showed that specialized activities created horizontal links without direct elite control. With respect to some imported items (utilitarian stone hoes and valuable shell ornaments), frequencies declined across 500 km from source locales, with little distortion from the presence of elite centers in the overall distribution (Muller 1997:289–384). In many instances, however, we lack the quantity of household studies required to assess production and circulation in detail at a local scale. In the case of Cahokia, the general populace enjoyed slightly improved 50

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

access to imports compared with earlier periods before the apogee of the center, suggesting that the local scale must be differentiated from interregional patterns (Milner 1998:152–153). Recent studies of centralized redistribution by Aoyama (1999) for the Copán Valley in the lowland Maya area and by Bayman (1995, 1996) for the Hohokam community of Marana in southern Arizona are of particular interest because of the scale of analysis and inclusion of a variety of household contexts, including areas close to the center and outlying settlements. The Marana community extends 21 km, with the farthest-sampled households at a distance of 14.4 km from the platform mound (Bayman 1995:39). Thus, the community extends across a distance over which a market service area might operate. For Copán, the distribution of obsidian extends approximately 14 km away from the center along the Copán Valley; the more distant and at times subsidiary La Entrada area is located 40 km from Copán, and obsidian distribution extends across approximately 32 km there (Aoyama 1999:11). In the Late Classic period, the same obsidian source predominated for blades in the Copán and Entrada areas (Ixtepeque) (Aoyama 1999:131, 139–143). These two studies have a scale and context that parallel the study by Stark and Ossa (Chapter 5), which explores distribution from a single market center during some periods. Aoyama and Bayman found that the items considered (imported obsidian tools for Copán and imported obsidian points and shell artifacts for Marana) were concentrated in the central settlement and, in the case of Copán, at high-status residences. At Copán, poor households had few obsidian blades and often used an obsidian flake technology instead. In the Late Classic period (AD 600–900) in the Copán Valley, obsidian was the most common material used for cutting tools—78 percent—with chert comprising only 22 percent. Conversely, in the more distant La Entrada area, chert rose to 44 percent of the assemblage (Aoyama 1999:127–180), indicating far less obsidian access, a pattern that corroborates the more detailed contextual evidence of a link between obsidian access and social status. At Marana, shell jewelry and shell working as well as obsidian projectile points (argued to have been used ceremonially) were concentrated in residences near the platform mound, the ritual and political focus of the community. All households in the central Marana location had access to shell and obsidian, with no indications of a marked class or ranking system among households. In both cases the authors viewed redistribution as directed by either the most powerful families (Copán) or the households located close to the community’s civic-ceremonial core (Marana). Both the Marana and Copán studies demonstrate how central redistribution differs archaeologically from marketplace exchange because of the social and settlement skewing in distributions, in other words, the tendency for substantially uneven concentrations of redistributed goods according to social status or residential location (Hirth 1998). Nevertheless, the presence of imported obsidian blades as utilitarian items in most Copán and La Entrada households shows the equifinality problem inherent in comparing redistribution and marketplace exchange. Wealth differences 51

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

among households may considerably affect access to an imported item in marketplace exchange; thus, marketplace exchange of obsidian blades and cores cannot be ruled out entirely for the Copán area, even if governing officials negotiated the importation process. The consistent social differences in blade quality and quantity, combined with the small size, light weight, and low annual use rates, makes central redistribution a more reasonable mechanism for Copán, however. Overall, centralized redistribution studies in middle-range societies with moderate to strong central leadership support the revisionist views advanced by Earle (1977). Elites do not appear crucial in amassing items or foods to reallocate to the general population under their authority. They are not as active as once thought in the restricted distribution of imported items, including both utilitarian products and preciosities. The Hopewell and Mississippian cases show relatively wide participation in production or use of some items—perhaps for recurrent social purposes, such as ritual participation and rites of passage. Based on our analysis of the various case studies, we conclude that central redistribution may be viable within Carr’s (2005) local scale of neighboring communities for obsidian and perhaps other lightweight items in regular household use, provided that quantities and weights are very modest and that receiving households need small inventories of these items; this mechanism is unlikely for effective provisioning of bulky items that need to be stored in large quantities and require frequent replacement, such as pottery. We now turn to a larger-scale imperial case concerning the Inca, who have been described as exhibiting a command economy (La Lone 1982). The Inca provide a useful example of likely variation in the roles of exchange mechanisms among provinces, as well the issue of multiple exchange mechanisms in a context in which marketplace exchange was not prominent or was largely absent. An Application of the Regional Production-Distribution Approach: The Possibility of Marketplace Exchange in the Andes

The regional production-distribution approach described earlier requires a considerable body of carefully collected data, in conjunction with economic analyses such as studies of craft production, provenance and stylistic analyses, and robust site and domestic collections (survey or excavation) that encompass a large area. We apply this method in an example from the Upper Mantaro Valley in Peru, where the scale of exchange information falls between Carr’s (2005) local and regional distances. This case study illustrates the production-distribution approach but also provides a way to explore the possibility of market exchange in the context of a largely state-­managed command economy. The Mantaro Valley project focused on occupation before and after the Inca conquest, respectively the Wanka II (AD 1350–1450) and Wanka III (AD 1450–1533) periods (D’Altroy 2002; D’Altroy et al. 2001; Earle 1985). The Mantaro case lacks some information necessary for fully implementing a regional production-distribution approach; however, our broader goal is to show the potential 52

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

of this kind of analysis. In several respects, our foray into the Andean area challenges accepted thinking about the possibility of market exchange in the Andes. As noted, instead of market exchanges and market systems, economic studies in the Andes highlight the importance of Inca state tribute requirements, labor tax, state storage facilities, and state-controlled craft production. State distribution of Inca-style decorated vessels, for example, provided one mechanism for imperial leaders to coopt local elites (Costin 2001; D’Altroy 2001c). Generally, market exchange has been dismissed as an important Andean economic institution because colonial documents describe Inca economic practices in considerable detail (Murra 1980) but mention little about market institutions (Stanish, Chapter 9). Some Ecuadorian and Peruvian coastal data, however, indicate indigenous marketplaces and traders (Rostworowski 1977; Salomon 1986, 1987; Spaulding 1973). Doubt has been cast on the documentary support for market system development in these peripheral areas (La Lone 1982), but we do not focus here on the documentary record. To address marketplace exchange in the archaeological record of the Inca, Earle (1985) looked for signs of a well-developed market system—including evidence for media of exchange, exchanges divorced from other institutional relationships, and a broad spectrum of items obtained in marketplaces. These are rigorous criteria, and with them archaeological data alone would be unable to firmly establish market institutions in late prehispanic Central Mexico, even though the existence of a periodic market system there is not in dispute. Typically in the Andes the criteria brought to bear do not reflect developmental or dynamic concerns—in which market exchanges appear in some times and places or for some types of transactions but possibly shrivel at times because of state discouragement (Blanton 1976; Braswell, Chapter 6). Christopher Garraty (Chapter 1) argues against a totalizing standard for characterizing premodern economies, suggesting that economies may be complex and rely on multiple exchange mechanisms. He uses Paul Bohannan and George Dalton’s (1962) distinction between peripheral and integrated markets to conceptualize the extents to which households variably relied on market exchange for domestic provisions. These concepts can be thought of as a continuum ranging from heavy reliance on market provisioning (integrated) to use of market channels for only a limited number of commodities (peripheral). Here we explore the possibility of peripheral market exchange in the Andes. For purposes of discussion, we pose a different question for the Mantaro data than Earle’s (1985): can the documented evidence concerning production and distribution of ordinary household objects reasonably be accomplished through kin- or community-based gifts or exchanges, by local leader/ruler redistribution, or, instead, by marketplace exchanges, perhaps linked to public events? Particularly if regular household distribution for common items extends across polity or settlement borders, we have reason to doubt the suitability of local elite/ruler redistribution and gift exchanges. Kin or other reciprocities within a community are described for the Andes when community members reside in contrastive ecological zones (Murra 1980). This 53

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

reciprocity should establish reasonably even access within communities to raw materials, local crafts, and foodstuffs; but particular classes of craft objects would not be evenly distributed across communities (see also Abbott 2000:175–184). To simplify, we concentrate on pottery distribution. How might a peripheral market system work to supply pottery in the Mantaro Valley? What might households have exchanged for pots? One possibility is food, but perhaps also wool or textiles. Even if all households had some access to such items and could have produced them themselves, exchanges may still have occurred with potters because craft specialists gave up some of their time growing crops, tending animals, or engaging in other pursuits to produce vessels for extra-household exchange, creating demand for foodstuffs. Exchange does not require all households to have a distinct craft in order to exchange with specialists. For the Wanka II period before the Inca (see D’Altroy 2001b:39), residential investigations focused on three of the four largest sites among thirty-eight sites in the survey area (D’Altroy 2001a:77). The four largest sites are located in the Upper Mantaro Valley, with several smaller settlements scattered in the vicinity and in an area extending approximately 20 km downriver. Three sites were excavated, located within approximately 5 km of each other but on separate ridgetops. The two largest (Hatunmarca and Tunanmarca) appear to have headed small polities, and a third, which was not excavated, may also have commanded a small polity. If the three polities were about equal in population size, then, given the overall Wanka II estimate (D’Altroy 2001b:39), each polity may have had a population of approximately 20,000. Analyses of pottery production are well developed for the Upper Mantaro Valley, but analyses of distribution present a challenge because of the focus on excavations at three sites. Cathy Costin (2001) conducted extensive studies to suggest that locally produced jars and bowls (Base Clara, Cream Slip, and Wanka Red types) involved two production-distribution systems, one with Tunanmarca as the head center and the other with Hatunmarca as the head center.2 Another local type, Micaceous SelfSlip, was more difficult to analyze because it was not prone to wasters but may have been produced at both Tunanmarca and Hatunmarca. For the purpose of our discussion, we assume that these two local pottery production systems served the nearby smaller settlements. Nonlocal pottery mainly consisted of Andesite-Tempered Wares (Costin 2001: 209). This broadly distributed type is thought to have been produced in the southern part of the Mantaro Valley and is encountered across a radius of nearly 100 km (Costin 2001:237). It is strongly skewed toward elite households but with noteworthy commoner access as well. Other distant imports to the Upper Mantaro Valley are much less frequent and likewise more prevalent at elite households (Costin 2001:236).3 The wide distribution of externally produced Andesite-Tempered Wares to the Upper Mantaro Valley (and elsewhere) and the distribution of local community specialist wares to all households (documented only within the bigger sites) raises the 54

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

possibility of peripheral market exchanges during periodic assemblies at centers or districts within them or at producer residences (serving also as “shops”). Thus, the major centers may have been the focus of a solar market system for some household items (C. Smith 1976b), in which a single center dominated access for a subordinate population. In contrast, the movement of Andesite-Tempered Wares may reflect elitenegotiated external exchanges and gifts but also wider circulation through distribution to clients. The alternatives to market exchange of local pots are less clear. Were specialist products in the local polity stored at elite residences to fund largesse to clients or to express reciprocity for communal labor service? This seems unlikely because household pots break with some frequency and must be replenished. Did producer households maintain extensive kin ties through which they exchanged pots to reach all households? This also seems unlikely because we would expect potters’ kin and social networks to be more limited than the entirety of Wanka II polities, given their population sizes. After Inca conquest (Wanka III period) the town of Marca became the source of locally produced storage and serving bowls. Costin (2001:231) suggests that the Inca bureaucracy largely ignored the local production and distribution system. Inca involvement in pottery focused instead on Inca-style vessels likely manufactured under state authority by attached specialists outside the Upper Mantaro Valley. These vessels were skewed toward elites even more than Andesite-Tempered Wares had been during the Wanka II period (D’Altroy 2001c:263–264). Andesite-Tempered Wares continued in a diminished quantity and became more frequent at commoner than at elite households. Thus, pots produced outside of the local area (Andesite-Tempered Ware) continued to reach households in the Upper Mantaro Valley and during Wanka III were more likely to have been circulated by mechanisms similar to those for local wares. Although there is no reason to imagine that Inca vessels were trafficked through local marketplace exchange, the local wares used in everyday food preparation may have been exchanged during periodic social assemblies, if not at formal marketplaces. The exchange situation may have resembled that described by Abbott for the Hohokam, in which the market system centered on ceremonial congregations rather than formal marketplaces (Chapter 3; see also Abbott, Smith, and Gallaga 2007 and elsewhere). The imposition of Inca rule did not fundamentally alter the Upper Mantaro production and distribution system; the Inca did impose a greater emphasis on vertical state ties and reduced the latitude for elites to participate in “horizontal” external exchanges. In Ecuador, documents suggest that Inca administrators actively discouraged long-distance traders (Salomon 1986, 1987). Clearly, imperial authorities did not encourage market activities and may have actually undermined them by elaboration of hierarchical links. If so, this situation may support Blanton’s (1976:259–261) argument that increased state control over the economy tends to reduce the efficiency and 55

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

prominence of market exchange (Braswell, Chapter 6), although states vary in governmental organizational characteristics. For example, Blanton and Fargher (Chapter 10) note that a collective governmental emphasis may lead to a more favorable context for market exchange. Evaluation of the Upper Mantaro data from the regional production-distribution perspective raises the possibility of peripheral markets that boosted opportunities for provisioning ordinary households with some domestic gear and assured that specialists maintained access to consumers. This idea does not substantially alter the picture of a non-market emphasis for the Inca political economy but introduces the possibility of more complex economic arrangements. Further investigation of regional-scale distributional patterns will be required to evaluate the possibility. Importantly, our aim is to highlight a dynamic perspective for Andean economies rather than to couch all discussion in terms of the presence or absence of a well-developed market system. Mayer (2002:47–73) revisited the Andean market exchange issue and recognized the potential for barter to have been sandwiched between two critical economic arrangements: state labor service and tribute arrangements that in turn placed a premium on community self-sufficiency. Barter, if it operated in more than a sporadic, unpredictable manner, would have been oriented to local community needs, partly mediating the tension between state economic priorities and local community selfsufficiency. Nevertheless, the coexistence of a command economy and the response of community strategies toward self-sufficiency would have provided few opportunities for market exchange. Although modern command economies do not provide identical examples for the opposition of local community and state concerns, the problems of local provisioning and action versus state planning create a comparable tension in which market exchanges and marketplace development are discouraged but still exist (Ericson 2008). The subjects of central redistribution and command economies illustrate how archaeologists can begin to separate different exchange mechanisms. Frequency of household use and replacement, the scale of production and distribution, and social skewing in access are all lines of evidence in these analyses. To make sense of diverse exchange mechanisms, the possibilities of a multicentric economy (Bohannan 1955; Bohannan and Dalton 1962:3) and associated tensions and social contexts are critical parts of the analysis. Multiple Lines of Evidence

Our emphases on Hirth’s distributional approach and our cognate regional production-distribution approach for evaluating marketplace exchange do not diminish the value of additional lines of evidence. Hirth (1998:453–454) referred to a “contextual approach” that considers both settlement hierarchies with urban centers and extensive craft specialization as evidence pointing to marketing, although not unambiguously. The provisioning of large urban centers may stimulate market sys56

D e t ec t i n g M ar k e t p lace E x c h a n ge i n A rc h ae o l o g y

tem development, or state administration may promote marketplaces as a solution to urban provisioning; but not all cities, states, and empires are associated with welldeveloped market systems (Blanton and Fargher, Chapter 10; Stanish, Chapter 9). Some craft production may also be heavily administered by the state, as in the Inca case. As Feinman and Nicholas (Chapter 4) note, important information can include the presence of storage facilities and assembly areas, such as plazas. Hirth (1998) designates such physical and architectural evidence the “configurational approach.” He (1998:453) also remarks on the ambiguity of such evidence because periodic marketplaces may be cleaned, and traces of permanent stalls—such as the depressions for shade tie-downs proposed for some Oaxacan sites (Lind 2008:181–183)—may be lacking. Even where rock alignments or other evidence of marketplace stalls are present, the function of these spaces needs to be confirmed (Dahlin et al. 2007:367). Examination of chemical residues from possible marketplace activities is the latest addition to multiple lines of evidence. Bruce Dahlin and colleagues (2007) argue from the spatial configuration of roads and plazas at Chunchucmil, a Classic Maya center (AD 450–700), and from its environmental and demographic contexts that a marketplace may have been present. As a separate line of evidence, they employed chemistry to analyze soils from a plaza in Chunchucmil (examining phosphorus, carbon, and extractable metals) and compared them with samples from a modern Guatemalan marketplace. Similarities in chemical signatures suggested to them that Chunchucmil had a marketplace. In wider applications of this approach, however, soil chemistry patterns related to marketplace activities versus those pertaining to periodic feasting or ritual events in plazas will require considerable scrutiny (Wells 2003). Formally delimited open spaces like plazas may host multiple activities over a long span, leaving a variety of chemical and other residues. For wider applications, researchers will have to deal with the possibility that ritual events can involve craft endeavors as well as food preparation and consumption, and chemical signatures of proposed marketplace activity areas can be ambiguous. In another activity area analysis, Hirth (2009b) focused on traces of obsidian craft activity at a possible marketplace in Xochicalco, relying on identification of obsidian by-products to infer locations of craft manufacture within the probable marketplace. Conclusion

Archaeology has entered a period with a renewed focus on market exchange, with a range of scales of analysis, an enriched inventory of material indicators, and new methods of analysis. Major challenges lie ahead, however, because multiple kinds of economic mechanisms can operate simultaneously within a society. For example, some market exchange—even if regularly occurring during social, political, or ritual assemblies—does not necessarily reveal the prominent mechanisms of economic allocation (Polanyi 1957). For the most part, we are still on shaky ground in examining additional or alternative exchange mechanisms and their scale of operation within 57

B arbara L . S t ar k a n d C h r i s t o p h er P . G arra t y

and among societies. This remains a daunting task for future work, especially for ancient states and empires but also for cases such as the Hohokam, who occupied multiple independent settlements and operated within a regional system of 80,000 km2 (Abbott, Smith, and Gallaga 2007:461), rivaling or exceeding the extent of most ancient states. Our digression into the archaeology of the Andean realm is meant to provoke interest in the issue of diverse economic institutions operating simultaneously, at times serving different social interests (e.g., commoners versus elites, local communities versus the state) and ascending or declining in importance in different times and places. Barry Kemp’s (1989:232–260) analysis of the ancient Egyptian economy is a good example of attention to written and representational information to argue for complexity in a command economy. To examine growth in market systems, the spatial and temporal regularity of exchange, the development of marketplaces, the diversity of items regularly accessible through marketplace exchange, and the prominence of market activities in state and imperial political economies require assessments with substantial time depth. Sources of data cannot be conclusive if they are too heavily skewed to the records of central governing institutions and elites, whether they are documents, representational art, or potsherds. Archaeological methods, despite their many challenges, provide a critical rebalancing for scholars to address a wider range of social sectors and diversity of economic mechanisms. Acknowledgments. Several colleagues provided helpful commentary or suggested references but are not responsible for the outcome. We thank Richard Blanton, Christopher Carr, John Clark, Steven Falconer, Kenneth Hirth, Jennifer Jones, Michael Smith, Katherine Spielmann, and one anonymous reviewer. Notes 1. The large-scale distribution of adzes (for farming and woodworking) from a particular Hawai’i source likely under the chief ’s authority provided a seemingly troubling contradiction because of wide household access. The discovery of additional small-scale production locales and variety in adze attributes and raw materials (Bayman and Moniz Nakamura 2001; Lass 1998:24–26) indicates that not all adzes were made in production locations likely controlled by chiefs, and many probably circulated independently across distances of over 100 km. The evidence has not been examined to determine the mechanisms of circulation. 2. Production was centered on Hatunmarca and Umpamalca, a smaller site where pottery producers served Tunanmarca. Umpamalca was slightly over half the size (in area) of Tunanmarca (D’Altroy 2001a:68, 71) and was likely a political subsidiary. 3. Costin (2001:224, 240) interprets pottery production as organized through community specializations for the Upper Mantaro rather than through elite sponsorship of production. Within the production settlements, not all households were involved in pottery making. Because pottery-making households were engaged in other activities, including agriculture, she argues for a part-time craft (Costin 2001:228).

58

Section Two

Case Studies

Chapter Three

The Rise and Demise of Marketplace Exchange among the Prehistoric Hohokam of Arizona

David R. Abbott

Fixed periodic marketplaces in pre-state societies are well documented on several continents (e.g., Bohannan and Dalton, eds. 1962; Forman and Riegelhaupt 1970; Wanmali 1981). It is not surprising, therefore, that marketplace barter was probably extant in the American Southwest during prehistoric times, where barter was organized horizontally without oversight or involvement of some overarching, vertically structured institution. Southwest archaeologists have long suspected that incipient markets were associated with ritual ballgames among the ancient Hohokam of central and southern Arizona (e.g., Bayman 2002; Doyel 1979, 1985; Haury 1976:78; Heidke 2000; Wilcox 1991; Wilcox and Sternberg 1983). During the middle Sedentary period (ca. AD 1000–1070), when a network of at least 238 ballcourts at 194 villages encompassed approximately 80,000 km2 (Figure 3.1; Doyel 1991:247; Marshall 2001), crowds of spectators and participants in the ballgames presumably gathered from near and far, providing an ideal venue for barter and other exchanges. What is surprising, however, is the high level of dependence the Hohokam apparently placed on their decentralized, periodic marketing system, including regular barter for basic necessities. In fact, periodic marketing seems to have been a central feature of the Hohokam regional economy. I begin by reviewing the evidence for Hohokam marketplaces during the middle Sedentary period and their rapid demise during the late Sedentary period (ca. AD 1070–1100), when the ballcourt network collapsed (Abbott 2006; Doyel 2000). I then look farther back into Hohokam prehistory to provide a context for understanding the 61

3.1. Ballcourt locations in southern and central Arizona.

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

3.2. Canal systems and ballcourt villages in the Phoenix Basin.

rise of marketplaces by examining the relationship between specialized pottery production and marketplace exchange. This examination reveals (1) some unexpected circumstances associated with marketplace origins, (2) a short duration for the marketplaces, and (3) vulnerabilities that may have contributed to the short duration and rapid demise of the marketplace system. Evidence for Hohokam Marketplaces

Archaeologists conceptualize the Hohokam regional system as a set of geographically separate but interacting prehistoric communities spread across a vast territory and dependent on one another through the exchange of goods and services and perhaps mates (Crown and Judge, eds. 1991; Wilcox 1979). The cultural center of the Hohokam territory was the Phoenix Basin, composed of large, densely packed irrigation-based communities along the lower Salt and middle Gila rivers. In these riverine valleys, dendritically arrayed canals and ditches diverted river water onto thousands of hectares planted with corn, beans, squash, and cotton (Figure 3.2). This core area was surrounded by more sparsely inhabited upland zones to the north, east, and south and desert lowlands to the west, where the agricultural potential was marginal but 63

D a v i d R . A b b o tt

wild foods, game, and various resources were relatively abundant. One manifestation of system integration was a dense network of ballcourts, the limits of which largely demarcated the regional boundaries. Their presence implied a shared set of religious beliefs across that expanse, which provided a basis for social and economic interaction. Hohokam ballcourts were large, oval-shaped earthen constructions where a version of the Mesoamerican ritual ballgame was probably played (e.g., Haury 1937; Wilcox 1991; Wilcox and Sternberg 1983). Ceramic evidence from the lower Salt River valley and recent excavations at Palo Verde Ruin, a ballcourt village situated in the upland zone north of the Phoenix Basin, combine to support the idea that periodic marketplaces were associated with the ballcourts during the middle Sedentary period (Figure 3.1). The pottery data suggest that a reliable and efficient mechanism must have existed during the eleventh century for distributing large numbers of clay containers from concentrated production locales. Indeed, the hallmarks for marketplace exchange are evident, including (1) transactions responsive to the conditions of supply and demand, (2) buyers and sellers engaged independent of their social standing (often including strangers), and (3) the absence of direct political control (Hirth 1998:454–455; Pryor 1977:104). In addition, the Palo Verde excavations offer an unambiguous glimpse of the inner workings of the ballcourt exchange network, at least as it functioned in the northern uplands. Pottery Data

The Phoenix Basin is characterized by a diverse geological landscape from which prehistoric potters exploited numerous rock and sand types for tempering their clay bowls and jars (Figure 3.3; Miksa, Castro-Reino, and Lavayen 2004; Schaller 1994). Geological, petrographic, and clay chemistry results have demonstrated a provenance relationship between temper varieties and pottery production sources (e.g., Abbott 1995, 2000; Abbott and Schaller 1994). As a result, tens of thousands of ceramics have been linked to spatially restricted production areas, and pottery exchanges have been traced with remarkable precision. The movement of earthenware vessels between inhabitants who lived as little as 5 km apart is now recognized, demonstrating that large numbers of clay containers changed hands (e.g., Abbott 2000, 2003a; Abbott and Walsh-Anduze 1995; Walsh-Anduze 1996). One astounding finding concerns ceramic production during the middle Seden­ tary period, when the ballcourt network was at its peak. Pottery was not made in most villages in the lower Salt River valley. Instead, artisans residing in only five pottery source areas accounted for more than 90 percent of all bowls, jars, and scoops. Moreover, each of the potter groups concentrated its work on a narrow range of vessel forms that tended to complement forms made by the other artisan groups (Abbott 1998, 2001, 2003b; Abbott, Stinson, and Van Keuren 2001; Van Keuren, Stinson, and Abbott 1997). 64

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

3.3. Zones of sand and rock types in the lower Salt River valley: Ia=basalt, Phoenix Mountains; Ib=basalt, Fountain Hills; II=phyllite; III=Squaw Peak schist; IV=Camelback granite; V=quartzite; VI=quartzite and schist; VII=South Mountain granodiorite; VIII=Estrella gneiss; IX=Usery Mountains sand (based on information in Abbott 2000:figure 5.1).

The first production source has been pinpointed at Las Colinas, near the northwest corner of the lower Salt River valley (Figure 3.2). Thousands of medium-sized and large plain ware jars (probably used for cooking and storage) and lesser numbers of plain ware bowls were made with phyllite temper and distributed throughout the territory on the north side of the Salt River (Figure 3.3). The second and third supply zones were located on the south side of the Salt River at both ends of South Mountain (Figures 3.2 and 3.3). Local artisans produced large, wide-orifice, thick-walled ollas, which possibly functioned as water coolers. These vessels were widely traded across the valley, including to Las Colinas, where jars of this shape and size were not made. Because two discrete sources for these pots can be discriminated by their temper (South Mountain granodiorite and Estrella gneiss), we can infer that at least two groups of potters labored to make them. In addition, a few of the ollas found on the south side of the Salt River were made along the middle Gila River. 65

D a v i d R . A b b o tt

Middle Gila manufacturers of plain ware containers used a glittery rock type, coarse-grained mica schist, to temper their pastes, and they supplied nearly all of the medium-sized and large jars and some bowls to the households on the south side of the Salt River. Finally, those potters specializing in decorated red-on-buff bowls and small jars were also situated along the middle Gila River at Snaketown and probably elsewhere in that vicinity (Figures 3.2 and 3.3; Abbott, Smith, and Gallaga 2007). Nearly all of the bowls in use in the lower Salt River valley during the middle Sedentary period were decorated, and the small red-on-buff jars complemented the larger plain ware containers in size and aperture diameter, with little overlap. Despite the spatial concentration of buff ware production, a standard set of redon-buff vessel forms was uniformly disseminated from the middle Gila River region to domestic units throughout the lower Salt River valley. During the middle Sedentary period, buff ware sherds consistently accounted for about 20 percent of the total ceramic assemblages in all parts of the area. Moreover, the ratios of decorated bowls to decorated jars and the relative numbers of red-on-buff bowls of particular sizes were repeated from each location in the lower Salt River valley to the next (Abbott, Stinson, and Van Keuren 2001). We now recognize that ceramic production in the Phoenix Basin during the middle Sedentary period was organized across an expansive territory so that artisan groups, sometimes separated from one another by one or two days’ travel, each fabricated a restricted set of wares and vessel forms that often functionally complemented the containers manufactured by the other potter groups. In this way, each Salt River household was dependent on multiple, often distant producers for the full complement of its domestic pottery inventory. This concentration of production could not have been more dissimilar from the pattern documented immediately following the demise of the ballcourt network. The abandonment dates for relatively few Hohokam ballcourts are known, but most of them date to the end of the middle Sedentary period, and none is known to date any later (Abbott 2002, 2006; Cable 1987; Doelle and Wallace 1991; Doyel 2000; Wasley and Johnson 1965:23; see Wilcox 1991 for a different opinion). Just as the ballcourts were abandoned, buff ware frequencies at the Salt River villages plummeted, and potters in all parts of the valley began fabricating undecorated pots in a full range of vessel forms (Abbott 2006; Abbott, Watts, and Lack 2007). This shift included the replacement of all imported buff ware bowls with locally made plain ware forms (e.g., Abbott 1988). At Las Colinas, the ballcourt was abandoned near the end of the middle Sedentary period, and at that time the percentage of the total pottery assemblage accounted for by buff wares from the middle Gila River valley dropped from more than 20 percent to around 7 percent. Also, unlike the earlier contexts, plain ware ceramics in the postballcourt assemblages included substantial numbers (more than 10 percent) of pottery made elsewhere on the north side of the Salt River, implying production of plain ware pots at many new locations (Abbott 2006). Clearly, the fall of the ballcourts was 66

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

accompanied by a transformation of ceramic production, from highly concentrated manufacture of specific vessel shapes and sizes at a few locations and far-flung distribution to local production of a full range of vessel forms across the valley. A pattern of dispersed production continued throughout the subsequent Classic period (e.g., Abbott 2000). Economic Implications of the Pottery Evidence

The ceramic results have at least four obvious implications for how to conceptualize the middle Sedentary–period economy. First, if production was concentrated for one kind of commodity (i.e., pots), it stands to reason that other kinds of goods were made by specialists elsewhere for large-scale exchange. I define “specialist producers” as artisans who practiced their craft at least in part to supply their wares beyond their own households, to external consumers. Specialized production and exchange among the Hohokam have been proposed for several commodity types, including shell jewelry (A. Howard 1993), projectile points (Hoffman 1997), manos and metates (Hoffman, Doyel, and Elson 1985), tabular knives (Bernard-Shaw 1983), axes (Doyel 1991), and some kinds of plant foods and fibers (Fish, Fish, and Madsen 1992; Gasser and Kwiatkowski 1991; Hutira 1989). The one-way flow of pottery from the middle Gila River potters to the communities along the lower Salt River implies that large quantities of other goods were moved in the opposite direction. Second, the sheer volume of ceramic transactions was impressive. The distributional data show that roughly half of all ceramics along the lower Salt River were imported from middle Gila sources during the middle Sedentary period. Recent population estimates in the lower Salt River valley hover around 40,000 during the Classic period (Doyel 2008; Fish and Fish 2008), and presumably the population during the middle Sedentary period was smaller. Nevertheless, if we conservatively assume that each household required one or two new pots annually, then it is safe to say that thousands, if not tens of thousands, of pots were sent northward from the middle Gila producers each year to fulfill household needs along the lower Salt River. Third, when we consider the huge numbers of consumers relative to the much smaller number of pottery suppliers, we recognize that the clay pots must have been distributed beyond the limits of kinship networks and other social ties. Artisans must have manufactured some of their wares for people they did not know. Presumably, trade in other commodities also transcended the connections in the immediate social field. It is unlikely, therefore, that kin-based reciprocity alone accounted for the pottery allocations, an inference also supported by the great volume of pottery transactions and the spatial uniformity across a large territory with which the pots were distributed. Fourth, the ceramic results imply that the middle Sedentary period was marked by an efficient and reliable mechanism for distributing goods. One idea is a political economy articulated by dominating elites (e.g., Earle 1997:70–75; Feinman, Blanton, 67

D a v i d R . A b b o tt

and Kowalewski 1984:169; Johnson and Earle 2000), but the available evidence offers little support. In such economies an obligatory production of surplus goods and labor was rendered to a central figure for financing the activities of governing institutions. Elsewhere, however, elites had little interest in a monopoly control over domestic pottery production and distribution (Frankenstein and Rowlands 1978; Fry 1980; Haselgrove 1982; Kipp and Schortman 1989; Rice 1987), probably because it would have been especially difficult to attain (Stark and Garraty, Chapter 2). Cheaply made utilitarian items that could be, and often were, manufactured across wide areas were not the stuff from which political hierarchies were built (Costin 1991:11; cf. Earle 1987). Also, archaeologists have found no evidence for stockpiling at redistributive central places, and ceramic production was prevalent outside the most politically prominent villages (e.g., Las Colinas). What one finds is widespread regional uniformity of imported goods. What was the mechanism that facilitated the distribution of so many pots and, presumably, other supplies? A likely candidate is periodic marketplaces associated with the ritual ballgames. This idea is not new. But what is new concerns the great quantities of materials that possibly changed hands through market exchange. Marketplace barter has important advantages from the perspectives of both supply and distribution. Marketplaces and other centralized loci of exchange offer ideal conditions for producers to minimize transportation costs when multiple buyers assemble in one place (Alden 1982:86–88; Belshaw 1965; Stine 1962). We can imagine Hohokam potters in the middle Gila River valley packing up loads of their wares and walking one or two days to ballcourt events in the lower Salt River valley, where eager buyers anticipated their arrival. Many kinds of artisans can also be attracted to a single marketplace, thereby concentrating a great variety of goods at a particular location for consumers (Hassig 1982; Hirth 1998, Chapter 11). On the distribution side, a supplier’s wares can reach sparsely settled, low-demand areas when buyers are drawn from the countryside to a marketplace (Hassig 1982). This process creates a homogenizing effect on the distribution of goods produced in limited areas (Braswell and Glascock 2002; Hirth 1998; M. E. Smith 2004:90), exactly like the uniform distribution of Hohokam red-on-buff pottery in the lower Salt River valley (Abbott, Stinson, and Van Keuren 2001). Finally, participants in market exchange require no prior kinship relationship or other social tie to transact with one another (Carrasco 1983; Hirth 1998; Plattner 1989c:179). As noted, barter among strangers almost certainly characterized the distribution of Hohokam ceramics. Evidence from Palo Verde Ruin

In the uplands north of the Phoenix Basin during the early Sedentary period and before, local dwellers inhabited single-family farmsteads dispersed along the largest but still intermittent drainages, or occasionally a few households were clustered together in small hamlets (e.g., Doyel and Elson, eds. 1985; Leonard 2002; Weed 1972). 68

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

Villages with ballcourts and central plazas, such as the one at Palo Verde Ruin, were probably not established in the upland zone until the middle part of the Sedentary period, when populations became more clustered around ballcourts and the political organization took on a new and higher level of integration (Abbott 2002; Wood and Wilcox 2000). At that time ballcourts were built even farther to the north in territories belonging to the Sinagua, Cohonina, and middle Verde populations (Wilcox et al. 1996). The upland Hohokam ballcourt villages linked these groups with those in the Phoenix Basin through the expanding ballcourt network. The middle Sedentary–period ballcourt village at Palo Verde Ruin was situated along the eastern bank of the New River, about 35 km north of the Salt River. The settlement consisted of more than 100 pithouses clustered into 14 residential units, each associated with a cemetery, trash dump, and extramural features (Figure 3.4). The village was occupied for about fifty years. Assuming that pithouses had a life span of about twenty-five years (Craig 2000:151; Haury 1976:75), approximately 50 structures were present per generation, and a village population of 200–300 people seems reasonable. Interestingly, this estimate roughly equates with those pertaining to average-size contemporary settlements in the lower Salt River valley (Abbott and Foster 2003; Doelle 1995). Palo Verde Ruin may have been an important node in the regional exchange network, probably contributing large quantities of meat, hides, and sinew from largebodied game and highly valued manos and metates. Skeletal elements from bighorn sheep and deer were found in nearly every pithouse sampled for the faunal analysis (Glass 2007), and the village was proximate to a heavily exploited ground stone quarry (Hoffman, Doyel, and Elson 1985). In addition, evidence was uncovered at the site for middleman traders, who enjoyed direct access to the exchange market. Ten of the sixty-nine excavated middle Sedentary–period pithouses, spread across eight of the residential units, contained an average of nearly six large, globular storage jars as well as extreme amounts of one or more of the following: obsidian from the Vulture source in northwestern Arizona (see Shackley 2005); obsidian from geologic sources in northern Arizona; stone objects (i.e., carved stone bowls, censers, plummets, ground stone axes, donut-shaped specimens, jewelry made from stone other than turquoise); turquoise; shell ornaments; projectile points; and argillite, steatite, and galena. Additionally, two of these structures also housed the complete or nearly complete skeletal remains of bighorn sheep, the carcasses of which probably hung from the rafters when the structures caught fire and burned. In one instance the cached goods were obtained not for local consumption but for barter to a third party. An extraordinary collection of 104 shell jewelry pieces and whole shells, including specimens representing eight marine species, was recovered from a structure in Unit F near the margins of the site. The number of items was far beyond the quantity for personal adornments of an individual or a family. Also, Victoria Vargas (2007) found no evidence for shell working at Palo Verde Ruin; thus, 69

D a v i d R . A b b o tt

3.4. Site map for Palo Verde Ruin.

we can rule out the idea that the Unit F collection was a craftsman’s cache. Instead, the shell artifacts were probably obtained by trade for the purpose of trading them again. A cluster of thirty-five Vulture obsidian cores without any associated debitage in one of the Unit E structures might also be an example of middleman exchange, although obsidian core reduction was noted at other locations in the settlement. The obsidian from the northern sources is noteworthy because prior to the middle Sedentary–period expansion of the ballcourt network, volcanic glass was rare in the 70

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

lower Salt River valley (Fertelmes, Abbott, and Shackley in press). Presumably through the web of connections established during the middle Sedentary period, the southern movement of obsidian to the Phoenix Basin increased substantially (Shackley 2005), only to be reduced following the ballcourt collapse (Fertelmes, Abbott, and Shackley in press). It seems likely that Palo Verde Ruin and perhaps other ballcourt villages in the northern uplands played a key role in that circulation. The wide distribution of the “cache” structures at Palo Verde Ruin and the evidence for middleman trading probably signify that many of the Palo Verde Ruin residential groups commonly stowed and thus controlled their trade wealth individually (Figure 3.4). The members of each group enjoyed direct and unencumbered access to exchange activities without funneling their goods through a village headman or some other centralized authority. The excavators found no evidence of a centralized warehouse at the village level; communal storage instead was the prerogative of individual residential groups, with some members probably functioning as middleman traders. We see evidence at Palo Verde Ruin that residents from all parts of the settlement directly participated in marketplace activities, and their actions were apparently guided by economic motivations. Market Rings

The pottery data from the lower Salt River valley and excavation results from Palo Verde Ruin combine to imply a robust network of barter and exchange during the middle Sedentary period that transcended the limits of kinship networks. One possibility I envision for the Hohokam case is a system of market rings, in which each ring included a set of uniformly distributed marketplaces associated with ballcourt events. Trade and ritual ballgames at each location were periodic and scheduled to offset with marketing/ballgame days at other nearby marketplace sites within the ring (e.g., Hay 1971; R. Smith 1971). The Hohokam case may have fit Carol Smith’s (1976b:39–44) “network marketing system” in the sense that the markets were probably organized by a network of overlapping rings. Different commodities were available more or less equivalently at each location, where barter was horizontal, without oversight from some overarching, vertically structured institution. We see no evidence for settlement hierarchies that fit Walter Christaller’s (1966 [1933]) central-place theory, which, if present, would imply that commercial factors were of primary importance for settlement location. Instead, commercialization was low and probably did not include land and labor (M. E. Smith 2004). The specialist artisans articulated by the market transactions were probably part-time producers, and one or more media of exchange, such as raw cotton, spun thread, or cotton textiles, were likely used (Schneider 1974:157–177; M. E. Smith 2004). Does the Hohokam model with periodic and horizontally organized market rings make sense? Market rings everywhere are composed of mostly local people who 71

D a v i d R . A b b o tt

are small-scale and part-time suppliers, serving only a single center or market ring (Bromley, Symansky, and Good 1975; Hodder 1962; C. Smith 1974). Their production decisions are little influenced by the markets, which are simply outlets for occasional surpluses of crafts and foodstuffs. For these reasons, consumers are unable to obtain necessities in market rings with regularity or certainty (C. Smith 1976b:39– 44). Market rings are rooted in a set of economic conditions, namely, direct participation of small-scale producers, horizontal organization, and a considerable degree of local self-sufficiency. These conditions imply a low dependence on market exchange and a poorly developed division of labor. In contrast, what we find in the Hohokam case is an advanced division of labor evident by transactions involving products that could have been, but were not, produced by self-sufficient households (cf. Blanton 1983:53; Forman and Riegelhaupt 1970; C. Smith 1974, 1976b). Ceramic bowls, jars, and scoops were basic necessities of Hohokam life, yet households across the lower Salt River valley did not make their own pots. Instead, each family depended on multiple pottery suppliers, often situated considerable distances away. Among the Hohokam, the scale and efficiency of market distribution are more typical of hierarchically organized market systems. The challenge before us is to understand why the Hohokam example is so different from other documented cases and how a heavy dependence on marketplace transactions was possible without political control. Part of the answer may be apparent in the rise and demise of Hohokam marketplaces. To make sense of the Hohokam case, I next turn to ceramic information that reaches far back into the Hohokam past, beginning with the time when cultivators first settled in the Phoenix Basin and long before the first Hohokam ballcourts were built (Table 3.1). I seek to ascertain when ceramic specialization at the scale and level of concentration witnessed during the middle Sedentary period first developed. I will also investigate the relationship between specialized production and the origins and expansion of the ballcourt network. A longer perspective provides insights that help reconcile Hohokam marketplaces with theoretical expectations. Extensive and Long-Term Specialization

The Hohokam occupation in the Phoenix Basin began during the Red Mountain phase, sometime before AD 450, when pottery-making households probably followed a biseasonal residence pattern, planting crops in the fertile floodplains along the lower Salt and middle Gila rivers during the summer and moving to exploit non-riverine resources during the winter (Cable and Doyel 1985, 1987). By the start of the Pioneer period (ca. AD 450–750), cultivators diverted river water onto agricultural fields through short, shallow ditches in addition to floodplain farming (Henderson 1989:344). Larger canals, constructed and maintained by farmers living in multiple villages ( J. Howard 1991), were probably first constructed near the start of the Colonial period (ca. AD 750–950) and were grouped into hydraulic complexes 72

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

known as canal systems—sets of main canals with a common headgate location, Period Phase Date (AD) the branch and distribution ditches split from the primary lines, with the culti1300–1375 Classic Civano 1100–1300 Soho vated plots spread along the canal routes 1070–1100 Sedentary Late Sacaton (Turney 1929). 1000–1070 Middle Sacaton In the lower Salt River valley, Omar 950–1000 Early Sacaton Turney (1929) named four major canal 850–950 Colonial Santa Cruz 750–850 Gila Butte systems: Systems 1 and 2, the Scottsdale 650–750 Pioneer Snaketown Canal System, and the Lehi Canal System, 600–650 Sweetwater as well as several smaller ones (Figure 3.2). 500–600 Estrella Not all of these complexes were built at the 450–500 Vahki same time. Some, such as Systems 1 and ?–450 Red Mountain 2, were initiated early and expanded and Sources: Based on Abbott 2002; Dean 1991; Doyel 2000; Mabry 2000. remodeled over the centuries, whereas the Lehi Canal System was not constructed until much later, probably during the middle Sedentary period.1 The Colonial period was also the time when the first ballcourts were built. The ballcourt network probably originated along the major rivers in the heart of the Phoenix Basin. During the Colonial period and until the end of the middle Sedentary period, this network expanded to include ever more territory and a denser distribution of ballcourt villages in the heartland. By the middle Sedentary period, a ballcourt was present at nearly every village-size settlement in the Hohokam world. Ceramic assemblages during the Red Mountain phase consisted entirely of plain wares. By the beginning of the Pioneer period, known as the Vahki phase (ca. AD 450–500), a small percentage (less than 5 percent) of the pots was made with a well-polished red slip. This type, called Vahki Red, continued to be made in small numbers until the beginning of the Colonial period, when red ware pots ceased to be manufactured. Also, the start of the Estrella phase (ca. AD 500–600) during the Pioneer period was marked by the origins of a decorated ceramic tradition among the Hohokam. Red-on-gray and, later, red-on-buff pots were components of Hohokam pottery assemblages until the end of the Classic period. What is striking about the plain wares and decorated pottery is where they were made. As described in detail elsewhere (Abbott 2009), the temporal sequence of pottery assemblages from the lower Salt River valley is revealing. During the Red Mountain phase, plain ware pots were tempered with various types of sand and rock, implying that vessels were made at a variety of places. Not surprisingly, there were no indications of concentrated specialized production among the seasonally mobile and dispersed populations. In contrast, by the Vahki phase, when the valley inhabitants had settled into permanent villages and begun irrigating their crops, a majority of the plain ware vessels, including those at sites on the north side of the Salt River, were fabricated with South Mountain granodiorite temper. This composition indicates Table 3.1. Hohokam Chronology in the Phoenix Basin.

73

D a v i d R . A b b o tt

that the vessels were manufactured on the south side of the river, somewhere near the eastern slopes of South Mountain (Figure 3.3). The dominance of the eastern South Mountain potters continued for 500 years more, until the beginning of the middle Sedentary period. During those many centuries, the South Mountain artisans continually supplied a large majority of the jar forms to the rest of the inhabitants of the lower Salt River valley. Additionally, some mica schist-tempered plain ware jars were imported to the valley from sources in the south along the middle Gila River, as were some decorated jars. Except for trace amounts, the jar forms in the Salt River collections were made nowhere else. Also, from the very beginning of decorated pottery production in the Estrella phase until the end of the middle Sedentary period, middle Gila artisans supplied a majority of the bowls and scoops in the lower Salt River valley. The remainder of the bowls and scoops were made as plain ware forms tempered with mica schist from the middle Gila River area or with South Mountain granodiorite from the eastern South Mountain vicinity. In short, from the beginning of the Pioneer period and long before the first ballcourts were built, Hohokam potters at just a few locations specialized in the production of clay containers for supplying the irrigation-based communities along the lower Salt River. Ceramic production was highly concentrated for five centuries or more, signifying amazing stability for the organization of pottery production and consumer-supplier relationships, which by virtue of their extent and long duration were unparalleled in the American Southwest. How were pots disseminated prior to the middle Sedentary period? I can say with confidence that most of the bulky jars used in the lower Salt River valley before that period were made “locally” in the eastern South Mountain vicinity. Nearly all of the settlements in the valley were within 11 km (one-day round-trip) of the eastern South Mountain source, and presumably direct procurement from the jar makers was within reach of most or all of the valley inhabitants. Some mica schist-tempered plain ware jars were sent northward from the middle Gila area, but their distribution along the lower Salt River was patchy (see Abbott 2009), implying that they were distributed using a variety of transaction mechanisms—probably through uneven networks of kinship and other social ties between the two river valleys. In contrast, decorated bowls and scoops were more uniformly distributed. These vessel forms could be nested for transport, perhaps suggesting that the potters or their agents trucked the pots to far-flung consumers. On average, however, buff ware bowls were smaller than their plain ware counterparts (which were often tempered with South Mountain granodiorite), suggesting that transportation costs were a concern for distributing the bigger forms. Miriam Stark’s (2003:209) compilation of ethnographic information indicates that traditional potters, in the absence of water transport, commonly circulated their wares within a 15- to 50-km radius. Exactly how large numbers of Hohokam clay containers were transferred between producers and consumers prior to the middle Sedentary period remains to be determined. A mix of mechanisms almost certainly existed and possibly included the barter 74

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

of decorated bowls and scoops from the middle Gila River valley at occasional ritual or social events. It is apparent, nonetheless, that production and distribution arrangements changed substantially during the middle Sedentary period, a topic I explore in the next section. Marketplace Origins The Middle Sedentary Period

The centuries-long stability in the production and supply of clay containers came to a halt at the beginning of the middle Sedentary period, and with that change we see evidence implying that marketplace barter became a central element in the Hohokam economy. The production of decorated bowls and scoops for Salt River households continued unabated along the middle Gila River, but the production of the bulkier jar forms shifted away from the eastern South Mountain vicinity. South Mountain artisans, who for twenty-five generations had fabricated a full suite of jar shapes and sizes (necked and neckless, large and small, thin-walled and thick-walled), were reduced to making relatively large, thick-walled ollas at low frequency, with few exceptions. During the middle Sedentary period, the manufacture of small jars shifted mostly to decorated pots made along the middle Gila River. The large necked and neckless plain ware forms supplied to villagers on the north side of the Salt River were made mostly with phyllite temper at Las Colinas, and those vessels consumed on the south side of the Salt River were imported from the middle Gila area. Moreover, eastern South Mountain potters had to compete with artisans in the western South Mountain vicinity and along the middle Gila River to supply thick-walled ollas. Even at settlements in the eastern South Mountain area, the ceramic inventories were dominated by pots made along the middle Gila River, including small decorated jars, decorated bowls of all sizes, decorated scoops, and large plain ware jars tempered with mica schist. Based on these patterns (i.e., the decline of production near South Mountain and the ascendance of production at Las Colinas and in the middle Gila region), there are three points to be made, which I believe add up to the inception of Hohokam marketplaces during the middle Sedentary period, although possibly building on preexisting red-on-buff barter of middle Gila pots. First, the new production and distribution arrangements of the middle Sedentary period, as compared to those previous, were characterized by an increase in the average geographic distance between pottery suppliers and consumers. With the shift from the eastern South Mountain area to Las Colinas for supplying large jars to the households on the north side of the Salt River, the average geographic distance between supplier and consumer remained about the same. In contrast, the middle Gila suppliers of large jars were located considerably farther from buyers on the south side of the Salt River than the South Mountain jar makers had been. Also, during the middle Sedentary period, most of the small jars in use on either side of the Salt River were supplied by buff ware artisans along the middle 75

D a v i d R . A b b o tt

Gila River, located considerably farther from virtually all of the Salt River communities than were the previous South Mountain potters (see Figure 3.2). Unless offset by increased efficiencies in moving pots between suppliers and consumers (as we would expect at marketplaces), transportation costs would have significantly increased during middle Sedentary times. Second, the replacement of the deeply rooted connections between the eastern South Mountain artisans and their Salt River customers, which had been operating for centuries, warrants consideration. New, more geographically distant, and presumably more socially distant relationships between the Salt River buyers and the pottery makers along the Gila River dominated ceramic transactions during the middle Sedentary period. This shift would have imposed increased costs of negotiation on the average transaction. According to Brian Foster (1978:3), “[E]conomic exchange does not in itself promote social solidarity or stability, but rather is fundamentally a dissociative, conflict relation which must be carefully regulated.” Before the middle Sedentary period, the costs of negotiation probably remained low and were underwritten by the trust between the exchanging parties, with a well-established social relationship—a pattern of social and exchange ties that had existed for hundreds of years. In contrast, if barter between strangers became a common element in the regional economy during the middle Sedentary period, then some means of offsetting the increasing costs of negotiation would have been required. In marketplaces the power of an overarching authority is commonly imposed to guarantee the peace of the market. As I discuss later, however, in the absence of a powerful authority, marketplace exchanges between strangers were probably regulated through beliefs and practices connected with ballcourt ceremonialism. If indeed barter between strangers became a common element in the regional economy (as in a market system), then the middle Sedentary period is the most likely interval for such transactions to have occurred. Third, the origins of ballcourt ceremonialism during the early Colonial period had no measurable effect on the pottery evidence pertaining to production and distribution. As best I can detect, the organization of pottery production and destinations where the vessels were distributed from specific production sources did not change when the first ballcourts were built. I do not mean to imply that pots did not change hands at ballcourt events prior to the middle Sedentary period. Once crowds from near and far regularly gathered to participate in the ballcourt festivities, however, it is hard to imagine that pots and other commodities were not exchanged. But there was no notable change in the production and distribution patterns from the time before the ballcourts to the time immediately following their initial development. It is not until the middle Sedentary period, two centuries after the first ballcourts appeared, that major changes in production and distribution are evident. Those changes probably demanded a new and efficient mechanism for distribution that reduced transportation costs and included barter between strangers. Marketplaces probably first appeared during the middle Sedentary period, coupled with wider participation in the ritual ballgames. 76

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

Hohokam marketplaces probably existed for only a short period of time. By the beginning of the late Sedentary period, only two or three generations after marketplace transactions became a central feature of the regional economy, marketplaces had probably disappeared, along with the collapse of the regional ballcourt network. I suggest later that this short duration may have been related to the inherent instability in the structure of the Hohokam marketplace system, which may explain why the Hohokam case is so aberrant. It was dissimilar from other documented cases of horizontally organized market rings, possibly because it was not sustainable at its large scale and at the high level of dependence placed upon it. Before turning to that suggestion, I first consider the conditions for marketplace origins in the Arizona desert. These conditions seem in part to have been set by an expansion of the regional economy in which the relationships among populations shifted, and ballgame ceremonialism played a central role in regulating barter that transcended kinship networks. The Conditions for Marketplace Origins

As Cyril Belshaw (1965:78) noted, a market “comes into existence as a function of the division of labor, so that those who concentrate on production of one sort may obtain the produce of others.” If I am correct about the inception of Hohokam marketplaces during the middle Sedentary period, then specialized production of ceramics had been practiced in the heart of the Hohokam territory for centuries before marketplaces were established. Consequently, a central question is, what changed during the middle Sedentary period that promoted marketplace transactions? The answer probably involves a complex set of factors, not all of which are understood, but part of the equation seems to be related to the expansion of the ballcourt network and the regional sphere of interaction and trade that occurred at the beginning of the middle Sedentary period. Increased Surplus Production

One possible piece of the puzzle that will require future testing is intensification of production of agricultural surpluses during the middle Sedentary period. I suspect that increased quantities of cotton and possibly food crops, especially those grown on the broad valley floor of the lower Salt River, were added to the mix of trade items that regularly moved through the regional economy. It is unclear why agricultural production would have been expanded in the lower Salt River valley during that time, although the streamflow regime of the river was particularly reliable and predictable for decades during the tenth and eleventh centuries (Graybill 1989; Graybill et al. 2006), probably creating ideal conditions for irrigation agriculture. If so, it becomes easy to imagine that new and different associations developed among the producers and consumers who occupied diverse portions of the expansive interaction sphere. 77

D a v i d R . A b b o tt

For example, consider the dissimilar ecological conditions between the northern uplands and the irrigated lowlands and their inverse relationship regarding resource availability. The high elevations, narrow floodplains, and seasonal streamflows in the upland tracts made this zone marginal for agriculture; but the uplands were rich in wild plant foods and resources, large-bodied game (meat, hides, and sinew), and raw materials for ground stone implements. In contrast, the irrigated lowlands were rich in agricultural potential but poor in other vital resources (Crown 1991). These complementary conditions may have promoted a kind of partnership between the populations in the lower Salt River valley and in the uplands to the north. Prior to the middle Sedentary–period expansion of the ballcourt network, the upland tracts were sparsely populated by small family groups spaced out along the area’s south-flowing intermittent rivers and creeks. These little groups seem to have been closely tethered to the dense irrigation-based settlements along the lower Salt River, perhaps returning regularly to the valley lowlands for participation in community life. Presumably, these people carried goods and resources with them, and valley inhabitants also undertook resource procurement trips into the northern uplands. These arrangements probably changed with the construction of Palo Verde Ruin and other ballcourt villages in the northern uplands. At least one of these northern settlements, Palo Verde Ruin (Hackbarth 2002), was as large as those along the Salt River, and its founding established a locally seated community on par with its lowland neighbors. The evidence from Palo Verde Ruin described earlier implies that relatively formal trade ties developed between local residents and their Hohokam brethren to the south, as well as with non-Hohokam groups farther to the north. I can imagine that a particularly extreme form of trading relationship, called mutualism, developed between the inhabitants of the northern uplands and those of the lower Salt River valley, although it remains to be demonstrated. Katherine Spielmann (1986:281) defines mutualism as the interaction between populations “in which subsistence resources, which are complementary to locally procured/produced resources, are provided on a regular basis.” It is predicated on the long-term differential distribution of critical goods, such as between producers in different ecological zones. The unequal natural distribution promotes specialization for the purpose of trade, which in the presence of a reliable mechanism for exchange can “translate into a higher carrying capacity for each partner in the interaction than would be possible in the absence of the relationship” (Spielmann 1991:5). The requisite analyses have not been undertaken, but I would expect that, if mutualism did pertain to the Sedentaryperiod Hohokam, the large-scale settlement at Palo Verde Ruin and others like it in the northern uplands would not have been sustained without the influx of agricultural produce from the canal systems of the lower Salt River valley. Evidence for mutualism may also be apparent in the effects of its disappearance. By the start of the subsequent Classic period, much of the northern uplands had been vacated, as populations swelled along the Salt River (e.g., Abbott and Foster 2003; Doyel 1981; Teague 1984). Presumably, with the demographic shift many trade con78

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

nections were severed between the two areas, including those for the supply of meat from large-bodied game. The effects on the lowland people have been noted at Pueblo Grande, perhaps the most politically prominent settlement in the Phoenix Basin (see Figure 3.2). Throughout the Classic period, iron-deficiency anemia plagued the populace, increasing death rates and diminishing the standard of living, especially among women and children (Sheridan 2003; Van Gerven and Sheridan, eds. 1994). In general, the restructuring of regional alignments during the middle Sedentary period seems to have been accompanied by a rising volume of goods, an expansion of the territory over which the wares were regularly transported, and a rising number of connections and interdependencies that composed the regional network. These developments probably translated into increases in the average social distance between exchanging parties and the amount of human energy required for carrying goods. The costs of negotiation and transportation were magnified at the regional scale, and the same forces that probably promoted marketplace transactions within the irrigated lowlands likely operated at the regional scale with the same effect. Horizontal Integration

Another condition necessary for the origins of marketplaces was almost certainly created by the ballcourts themselves. According to Stuart Plattner (1989c:181), “Markets become regular, adequate, and secure when regions are integrated economically, politically, and socially. Regional integration comes from investment in infrastructure, meaning the basic technology of transportation and communication.” In the Hohokam case, the infrastructure for transportation and communication may have been the ballcourts, promoting the periodic assembly and interaction among persons from different parts of the region. Also, in the absence of an overarching authority wielding coercive power to guarantee the peace of the marketplace, the security of ritual etiquette probably substituted for political authority (Netting 1972; Rappaport 1971, 1984; Yoffee 1994) and provided an enabling factor behind the marketplace network. The context of ritual performance, wherein the interaction among individuals was regulated by custom and propriety, constituted a mantel under which lively bargaining for utilitarian commodities could take place, even among parties who had no prior social relationship with one another (see also Fash and Fash 2007; Wells and Nelson 2007). As suggested by Norman Yoffee (1994:353), similar patterns may have been common in the Southwest, where centralization remained weak even though social and religious institutions integrated relatively large geographic regions. Unfortunately for the Hohokam, when specialized production and ballcourt ceremonialism—which had persisted over long periods of time—were coupled with marketplaces, their existence ended after only a few decades. At present I cannot rule out a spurious correlation among these conditions (i.e., a relationship among occurrences that is a result of some unseen factor rather than of a causal connection among them). Nevertheless, I suggest that there are good reasons to entertain the possibility that the 79

D a v i d R . A b b o tt

coupling of ballcourt ceremonialism, specialization, and periodic marketplaces led to provisioning arrangements that were difficult to sustain. Unsustainable Arrangements

In the previous section I tried to show how it might have been possible to combine a horizontally organized, regional-scale network of periodic marketplaces with a high dependence on marketplace barter for distributing basic necessities like clay containers, other utilitarian supplies, and possibly food. Such a combination does not square well with marketplace theory, and I am unaware of any similar documented cases. The recent results noted earlier, however, imply that the Hohokam marketplaces were relatively short-lived, perhaps indicating that they were inherently unstable and thus were not the kind of configuration often found in the ethnographic literature or the archaeological record. In this section I explore the unsustainable arrangements that may have characterized the Hohokam case. I focus on the enabling conditions that underpinned the development of Hohokam marketplaces. If I am correct in my assessment, there are at least three conditions: (1) the broad-scale horizontal integration afforded by widely shared religious beliefs and participation in ballcourt ceremonies, (2) the surplus production of agricultural products in the irrigated lowlands, and (3) the specialized production of pottery and other crafts, as well as the supply of raw materials and commodities associated with specific ecological zones. I conjecture that all three of these conditions promoted a high level of dependence on periodic marketplaces for the distribution of some commodities, including basic necessities such as ceramic containers and probably food. Aspects of all three conditions were probably subject to disruption from both natural and social perturbations, however, leading to the demise of the marketplace system only a few decades after its inception. Integration through Shared Beliefs

The first enabling condition, horizontal integration through shared beliefs associated with the ballcourts, certainly contributed to incorporating diverse social and economic units into broader Hohokam society for several centuries. During the middle Sedentary period, however, the ballcourt network expanded considerably, bringing in even more diverse populations. During that expansion, which included the rise of marketplaces, the density of kinship and other close social ties was almost certainly reduced across the expanding social field. With the web of social relationships stretched thin, the burden of regional integration increasingly rested on the cooperation among socially distant parties inspired by the ritual ballgames. Without dense numbers of crosscutting social connections, the integration afforded by the ballcourt ceremonialism may have become subject to disruption if perturbed by changing social or natural conditions that challenged the shared beliefs and the existing social order 80

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

or by disputes. For instance, we can imagine the ballcourt ceremonialism overshadowed by rising political tensions in the region, as is thought to have occurred by AD 1100 (LeBlanc 1999; Wilcox, Robertson, and Wood 2001). In short, the ballcourt network, as a religious institution, may have helped connect people across the region and enabled marketplace exchange, but it was ill-equipped to deal with disruptive regional-scale pressures that may have given impetus to sociopolitical fragmentation. Surplus Production of Irrigated Crops

Also suspect is the stability of the second enabling condition: surplus production of cotton and corn. Although it remains to be shown, I suspect that surplus production of cotton and corn was substantial on the broad, irrigated valley floor of the lower Salt River. Regular and abundant harvests in that lowland area probably stimulated a high volume of economic transactions across the region. For instance, differences in the streamflow regimes between the Salt and Gila rivers, which made irrigation more risky along the latter, may help explain the heavy south-to-north passage of pots between the two valleys, which dramatically increased during the middle Sedentary period.2 A substantial production of agricultural surpluses depends, of course, on reliable technology (irrigation infrastructure), farmers’ motivation to produce beyond their own consumption, and favorable natural and social conditions. For instance, the heavy scheduling demands on irrigation may have impinged on the ability of irrigators’ households to make the craft items they required, thus creating a steady demand for specialist-made goods (Blanton 1983; Costin 1991:17; Mills 1995; Schortman and Urban 2004:197). Also, water for the irrigated fields along the Salt River was substantial and predictable during the Sedentary period (Graybill et al. 2006), providing ideal conditions for significantly expanding the productive capacity of the valley’s irrigation infrastructure. In part, new construction accommodated the subsistence needs of a growing population. The entirely new Lehi Canal System was built during the middle Sedentary period, upstream from the larger and older irrigation cooperatives in the valley, Canal Systems 1 and 2. If water was a limiting resource, then, depending on annual fluctuations in the supply, the new upstream demands may have diminished the downstream farmers’ ability to grow surpluses. In this example, one can see how even long-established and dependable providers of agricultural surpluses may have become vulnerable, which in turn undermined the volume and reliability of marketplace transactions for securing household necessities. An Emphasis on Specialized Production

Presumably, some of the supplies that moved through the marketplaces entered the regional economy from upland territories where, during the middle Sedentary period, the ballcourt network had expanded into new and different ecological settings. 81

D a v i d R . A b b o tt

As it did so, the volume and diversity of circulating goods probably increased, perhaps to risky levels. By embracing the efficiencies of concentrating production on those goods and resources that were unique or exceptional to the local area, mutualistic relationships may have developed (i.e., in which increased carrying capacity depended on a sustained exchange of resources, including agricultural products from the irrigated lowlands). If so, these economic arrangements, including the marketplaces, may have been vulnerable to changing local conditions. I have already considered, for instance, how surplus production in the canal systems along the lower Salt River may have been disrupted by environmental or demographic shifts and how the resulting diminishing supplies of iron-rich meat probably reduced the health of Hohokam lowland farmers during the Classic period. If Hohokam marketplaces did indeed depend on widespread shared beliefs, regular agricultural surpluses, and intense and concentrated specialized production, then it may not be surprising that marketplace exchange did not last long during the prehistory of southern and central Arizona. Conclusion

Hohokam marketplaces in and around the Phoenix Basin seem to present an unusual set of circumstances. Barter during horizontally organized periodic market rings associated with ritual ballgames dependably supplied Hohokam households with various commodities, including basic necessities such as pottery and probably agricultural products. These market rings played a more prominent economic role than those described ethnographically. The milieu in which those arrangements originated was one in which highly concentrated, specialized production of pottery and possibly other crafts had been practiced for centuries. It was also one in which the geographic sphere of interaction was rapidly expanded during the middle Sedentary period. At that time the potential costs of transportation and negotiation increased significantly, probably stimulating the origins of Hohokam marketplaces for offsetting those costs and widely distributing the products of spatially concentrated craftspersons and farmers. These arrangements did not last long, however, suggesting that they may have been built on unstable conditions. If so, these circumstances might explain why the case of Hohokam marketplaces is so aberrant. Hohokam marketplaces are unusual because they probably were unusual, enabled by conditions inherently subject to change. A dependence on horizontally organized market rings for regularly supplied basic necessities may have been precariously sustained by the regional integration of ritual ballgames, surplus production of agricultural goods, and spatially concentrated specialized production. All of these enabling conditions seem to have had long histories in the Hohokam world, but their considerable expansion during the middle Sedentary period in conjunction with the origins of marketplaces may have stimulated a new regional dynamic susceptible to vulnerabilities that had not existed before. Unfortunately for the regional economy, all of the enabling conditions may 82

T he R i s e a n d D em i s e o f M ar k etp l a c e E x c ha n g e

have become volatile at their expanded scale: they led to both the rise of Hohokam marketplaces and their rapid demise. Notes 1. Jerry Howard (1987) has studied the Lehi Canal System and the timing of its construction. He concluded that it was added to the valley settlement at the beginning of the Classic period. Ballcourts, however, were present at several villages in that irrigation complex, implying to me that the villages were occupied during the middle Sedentary period. 2. As discussed by Donald Graybill and others (2006), long-term differences in the flow regimes of the lower Salt and middle Gila drainages probably affected the relative reliability and potential productivity of irrigation agriculture. Specifically, annual streamflows in the Salt River, particularly during the Sedentary period, experienced low variation and few high discharges (i.e., floods) that defined optimal conditions for irrigation. In contrast, the regime of the Gila River was “characterized by high variability, including numerous prolonged periods of low flow punctuated by extreme high annual discharges” (Graybill et al. 2006:94). In short, irrigation agriculture was probably more risky along the Gila than along the Salt. Consequently, it is easy to imagine that some residents along the Gila River turned to pottery making for exchange as a way to buffer against the risks of crop failure (Arnold 1985; Graybill et al. 2006).

83

Chapter Four

A Multiscalar Perspective on Market Exchange in the Classic-Period Valley of Oaxaca

Gary M. Feinman and Linda M. Nicholas

The market, it turns out, is the hollow core at the heart of economics. (Lie 1997:342)

All economies are embedded in societies, and the economic anthropological approach to economic analysis sensitizes us to the same issues in large-scale industrialized societies. (Plattner 1989a:4)

The new paradigm begins by rejecting the idea of state intervention in the economy. It insists that state action always plays a major role in constituting economies. (Block 1994:696)

Until recently, the role, significance, and diversity of preindustrial markets (and perhaps issues surrounding markets more generally) have been insufficiently theorized and investigated by archaeologists and scholars in cognate disciplines (Minc 2006:82). Of late, however, conceptual perspectives have begun to shift (see Block 1994; Lie 1997; Plattner 1989a:4) with the growing realization by researchers in several fields that all economies (despite significant diversity) are culturally constituted and embedded in larger societal contexts, albeit in different ways (e.g., Alexander and Alexander 1991; Barber 1995; Block 2003; Dequech 2003; Gemici 2008; Granovetter 1985; Krippner 2001). As a consequence, the oft-supposed complete break in operating principles and institutions between economies in the ancient past 85

Gary M. Feinman and Linda M. Nicholas

and the industrialized present is not as revolutionarily distinctive as many earlier scholars envisioned (Garraty, Chapter 1). With the recognized narrowing of this previously presumed gulf in economic practice, some historians, economists, classicists, and economic sociologists have reframed their attention productively on ancient economies, particularly in the Mediterranean world (e.g., Manning and Morris, eds. 2005; Temin 2001), yielding new findings concerning the importance of market exchange and commercialization. With a few key exceptions (e.g., Blanton 1983; M. E. Smith 1979), anthropologists and archaeologists have yet to probe these issues as deeply, despite their great potential to broaden empirically and theoretically the investigatory scope on preindustrial economies outside the classical world. Now, with this volume and other ongoing studies, topics concerning markets and market exchange are being placed more squarely on the archaeological agenda. We begin by briefly summarizing important conceptual and analytical shifts relevant to the study of markets and preindustrial economies. Although many of these new perspectives have been formulated outside archaeology, they are relevant to anthropological perspectives on past economies. From this broadened interpretive perspective, we turn our attention to a multiscalar examination of the Classic-period (AD 200–900) economy of the Valley of Oaxaca, Mexico (Figure 4.1). Employing this multiscalar perspective, we propose that a system of marketplace exchange was a key element of Oaxaca’s economy over a millennium ago and discuss implications of this interpretive position. Marketplaces and markets are (and have been) central institutions for many societies, taking distinct forms in past and present global contexts. At the same time, historical analyses now indicate that market exchange and marketplaces, like many other core human institutions (such as state forms of government), had multiple lines of origin and development and, seemingly, deeper temporal roots than previous generations of scholars thought (Dahlin et al. 2007; Gledhill and Larsen 1982; Manning and Morris, eds. 2005; Shen 1994; Temin 2006; Abbott, Chapter 3). Like states, market systems also were and are highly variable in organization and scale (e.g., C. Smith 1974, 1976a, 1976b). Because of their significance, diversity in character, the varied contexts in which they appear to have operated, and their deep histories, the origins and long record of market activities ought to be a fertile topic for investigation by archaeologists and scholars in cognate fields. To date, we have often been hamstrung by two impediments: (1) problems concerning the traditional ways questions about market exchange, marketplaces, and their past importance were framed theoretically across the relevant disciplines, and (2) the empirical difficulty of defining and documenting market exchange, activities, and places unequivocally through archaeological research. We consider both of these issues, albeit somewhat indirectly. We briefly outline the theoretical shifts in several fields that are spurring reconsideration of issues regarding marketplaces, market exchange, and their historical significance. Then, building on this broadened perspec86

A M u lt i s c a l a r P e r s p e c t i v e o n M a r k e t E x c h a n g e

4.1. The Valley of Oaxaca, Mexico.

tive, we turn our attention to prehispanic Oaxaca. Because there is likely no single archaeological indicator to identify and define the nature of markets and marketplace exchange indisputably, we adopt a multiscalar perspective to study the economy of the Classic period in the Valley of Oaxaca. Although it may seem a bit counterintuitive, we propose that to understand systems of distribution, researchers must also probe the other aspects of the economic triad—production and consumption (see also Moreland 2000; Plattner 1985:xvi). Drawing on observations and interpretations at three analytical scales—the region, the site, and the house—we infer that marketplace exchanges were central to the economy in this pre-Aztec Mesoamerican setting. Reframing Conceptual Bases

The study of ancient economies has always been an intellectual quest that crosses disciplinary boundaries, raising issues of framework and communication. Given the scope of this enterprise, we consider key theoretical issues only briefly (but see Isaac 1993; LeClair and Schneider, eds. 1968; McCloskey 1997; Morris and Manning 2005b; North 1977; M. E. Smith 2004). Yet such consideration is needed to clear away some of the misperceptions that have historically plagued this debate in both economics and anthropology. 87

Gary M. Feinman and Linda M. Nicholas

It may seem unnecessary to consider how marketplaces and market exchange have been conceptualized, since we all more or less know what these terms mean. Yet longstanding definitional quarrels provide insight into the intellectual disconnects that have hindered the study of past markets. For example, the economic sociologist John Lie (1997:342) has observed that “the absence or ambiguity of the market concept is as old as economics itself.” As the study of past economies is by definition an enterprise that can potentially profit from dialogue across academic disciplines, we outline some conceptual tenets on which our research is built. Lie (1997) and others (e.g., Granovetter 1985; Krugman 2009; McCloskey 1997) have poked theoretical holes in narrow economic views that have defined markets as by nature perfectly competitive or always guided by hidden hands toward self-sustaining prices. A good number of scholars, especially those with comparative or historical experience, have observed that markets are highly variable in space and time (e.g., C. Smith 1976a; Swedberg 1994:255). Significantly, it is now more broadly recognized that markets are always embedded in their cultural contexts, even in capitalist systems (Alexander and Alexander 1991:493; Granovetter 1985; McCloskey 1997; North 1977:709). Social identities may not be central to specific market transactions. It does not follow, however, that economies based on market exchange exist outside or independent of social/political constraints (such as the definition of property rights, the rules of exchange, and more). Following Ronald Coase (1988:8), there seems to be increasing skepticism about narrow, restrictive definitions of market exchange and an emerging focus on markets as social institutions that facilitate exchange between producers and consumers, often involving some form of pricing and currency (see also Dilley 1996; Plattner 1985:viii; Garraty, Chapter 1). These conceptual issues are important because they expose important limitations in the theoretical framework of Karl Polanyi (1957, 1971 [1957]) and the ways his overarching perspectives have framed the subsequent analysis and interpretation of past economies. Polanyi championed ancient economies as embedded in the larger societies of which they were a part. Few today would contest Polanyi on this point, recognizing that all economies are embedded in a larger societal context to some degree (see McCloskey 1997). Polanyi (1971 [1957]:68) had a blind spot, however, when it came to his definition of market economies, arguing that they “were controlled, regulated, and directed by markets alone.” In other words, Polanyi did not extend his own embedded perspective on economies to the investigation of market systems. Instead, he adopted (and his written legacy has been largely equated with) a very narrow definition of market exchange that to a degree has stifled the recognition of marketplaces and markets in the past (Lie 1992; see also Block [2003], who argues that Polanyi himself shifted his views on the topic during his career and so was not entirely internally consistent, thereby opening the door to later misinterpretation). As noted, these views (associated with Polanyi as well as with the equally influential Moses Finley [e.g., 1999 [1973]] in regard to the classical world) are now seen 88

A M u lt i s c a l a r P e r s p e c t i v e o n M a r k e t E x c h a n g e

as having drawn overly stark distinctions between the economies of the recent and distant pasts. The definition of market systems has been widened (recognizing greater diversity in these modes of exchange), and the qualitative gulf Polanyi and his associates posited between capitalist market economies and other commercialized economies has been narrowed. With these conceptual reevaluations, market exchange and commercialization have now been convincingly recognized in the Greek and Roman worlds (Davies 2005; Greene 1986, 2000; Temin 2001, 2006), ancient Mesopotamia (Gledhill and Larsen 1982; Silver 1983; Snell 1997:145–158), and medieval Europe (Moreland 2000), among other regions. To synthesize briefly, the multidisciplinary study of ancient economies is now underpinned by the growing recognition that all economies—market, capitalist, or otherwise—are embedded, albeit in different ways and to varying degrees (Grano­ vet­ter [1985] makes this exact point). Freed of the myth that modern/capitalist market systems are devoid of social and political constraints, it becomes easier to see markets in the pre-capitalist past than was previously envisioned. It is fair to point out that the definition of market exchange employed here is somewhat broader than the one generally associated with Polanyi and Finley (Randall Collins [1990:121], for example, has argued that their definition is too narrow). Yet it is also important to recognize that notions of entirely free market systems operating exclusively on the principles of supply and demand often ascribed to those two scholars simply have not been found to exist in history, whether in the deep past or more recently. The notion of an entirely free, perfectly self-regulating market, whether advanced by the political left or right, seems more an ideological stance trumpeted in economic textbooks than a realistic portrayal of specific historical circumstances (e.g., Braudel 1986:227; Silver and Arrighi 2003; Swedberg and Granovetter 2001:12–14). As a consequence, the metric frequently employed by anthropologists, classical archaeologists, and historians for the past has often been skewed toward an unrealistically strict model or preconception that would not even apply to contemporary market economies. At the same time, we do know that the exchange of raw materials and goods extends back before the Neolithic era (e.g., Bar-Yosef 2002:367). For societies of any scale, the oft-cited alternative to market exchange, a redistribution-based economy (e.g., Polanyi 1957), is both logistically impractical and likely has been overemphasized (see Earle 1977; see also Blanton and Fargher, Chapter 10). Because most preindustrial states were politically weak and dependent on more restrictive transport technologies than is the case today, the economic control demanded by redistribution would have been hard, if not impossible, to implement and maintain. Rather than presume the importance of redistributive economies, as has happened too frequently in the past, archaeologists must empirically probe past economies more fully to reconstruct and model how these economic systems likely operated. Among the key questions to address are how and why market systems (and economic systems more broadly) develop, function, vary, and change. 89

Gary M. Feinman and Linda M. Nicholas

Archaeological Assessment of Market Exchange

To date, the study of markets in the past has been constrained not only by conceptual issues but also by the challenge of how to identify marketplaces and market transactions archaeologically. In regard to their identification in prehispanic Mesoamerica, methodological advances have been made recently by Kenneth Hirth (1998) and Leah Minc (2006). Yet the broad diversity of market systems and their different implications for household and regional patterns of distribution remain analytical hurdles, as do issues of equifinality (Stark and Garraty, Chapter 2). As Hirth (1998:452–454, Chapter 11) has elaborated, there is unlikely to be a single unequivocal indicator of marketplace activities. Even if there were, significant questions would remain regarding the relative importance and specific nature of those activities. Of course, these definitional issues become a greater challenge in contexts where archaeological findings cannot be supplemented by ample textual sources. For the Valley of Oaxaca during the Classic period, textual sources are sparse and largely undeciphered. Because complementary documentary data are lacking (in contrast, for example, to the identification of ancient Greek agorae [marketplaces] in Athens [e.g., Camp 1992]), we believe information from multiple analytical scales is necessary to define the character of prehispanic Oaxacan economic systems, including the role and nature of markets. Even historically defined markets, such as in the classical world (e.g., Camp 1992:122; Greene 1986:47), would be hard to identify unequivocally based on archaeological observations alone. As a consequence, we take a multiscalar archaeological approach. Based on our findings, we propose that marketplace exchanges were central to the economy in this region during the Classic period. Markets in Mesoamerica

Perhaps for the reasons enumerated earlier, relatively little investigation to date has been devoted to the study of markets and marketplace exchange in pre-Aztec Meso­ america. This is odd in a sense since the sixteenth-century Spanish, who came from the Mediterranean world—a hotbed of Eurasian commercial activity at that time—stated that they had never witnessed a marketplace as large and vital as the one in the center of the Aztec world at Tlatelolco (Anonymous Conquerer 1971 [1500s]:392; Cortés 1962 [1521–1525]:87–89; Diáz del Castillo 1963 [1500s]:232–234; Sahagún 1950–1982 [ca. 1577], Book 8:67–69). “On reaching the market-place . . . we were astonished at the great number of people and the quantities of merchandise, and at the orderliness and good arrangements that prevailed, for we had never seen such a thing before. . . . You could see every kind of merchandise to be found anywhere in New Spain” (Diáz del Castillo 1963 [1500s]:232). Frances Berdan (1985), Richard Blanton (1996a), and Michael Smith (1979), among others, have painstakingly documented the regional breadth of this market system and how it was interwoven with other spheres of the Late Postclassic economy (ca. AD 1300–1519) . Surely this mar90

A M u lt i s c a l a r P e r s p e c t i v e o n M a r k e t E x c h a n g e

ket system had deep antecedents, and what about the nature and variability of markets elsewhere in prehispanic Mesoamerica? In regard to the Valley of Oaxaca (Figure 4.1), sixteenth-century documentary accounts concerning markets are rarer and less detailed than those for Aztec Central Mexico. Yet despite a much less complete ethnohistoric record overall than exists for the Basin of Mexico, market systems are described for both the Valley of Oaxaca and the neighboring Mixteca Alta (e.g., Appel 1982; Spores 1965). In the latter region, sixteenth-century marketplaces were situated outside major concentrations of population (Pohl, Monaghan, and Stiver 1997). In the Valley of Oaxaca, most sixteenth-century marketplaces and the activities associated with them appear not to have been a direct or a complete consequence of Spanish meddling or influence (Cook and Diskin 1976:11). It seems likely that the roots of market activities in Mesoamerica extend well before Spanish or even Aztec times. If that were the case in the Classic period, then we might expect to see a high degree of household economic interdependence and craft specialization. Craftwork would have extended to basic goods as well as to high-value items. Alternatively, if prehispanic Mesoamerican economies were dependent on redistribution instead of marketplace exchange, we would expect an economic system in which craft production was generally rare and centered on lightweight, high-value items. Most households would have manufactured the bulk of the goods they needed, as the long-distance movement of heavy bulk items in quantity to a central place and then back out again (see Drennan 1984) seems cumbersome for a region the size of the Valley of Oaxaca, covering more than 2,000 km2. The Valley of Oaxaca during the Late Classic period was very different from what was described for the Naco Valley (Schortman and Urban 1994) during the same era in northwestern Honduras (southeastern Mesoamerica). There, a large number of seemingly specialized economic production activities were concentrated around the large central site of La Sierra. The apparent concentration of craft production is marshaled as part of an argument that Late Classic–period Naco Valley economic transactions were largely part of a redistributive system under the aegis of the rulers at La Sierra. Nevertheless, the physical scale of the Valley of Oaxaca was more than an order of magnitude larger than the Naco Valley, thereby entailing markedly different transport implications. At the same time, specialized craft manufacture was not centralized at a few large Valley of Oaxaca settlements during the Classic period; in fact, such economic activities have been noted through archaeological survey and excavation at many sites (e.g., Feinman and Nicholas 2004a, 2004b; Kowalewski et al. 1989:251–305). In addition, if redistribution were central to the Late Classic–period economy in the Valley of Oaxaca, we would expect to find large storage facilities for maize and other goods at regional centers (see also Feinman, Blanton, and Kowalewski 1984:169). No such granaries have ever been located or described in later Spanish accounts for the Valley of Oaxaca. Such features have been recorded archaeologically or described in 91

Gary M. Feinman and Linda M. Nicholas

texts for other global regions (e.g., Earle and D’Altroy 1982; Morris 1967). In the remainder of this chapter, and reliant principally on archaeological information, we take a multiscalar approach to probe the economy of the Late Classic period (ca. AD 500–900) in the Valley of Oaxaca. The Classic-Period Economy

To explore the Classic-period economy of the Valley of Oaxaca, we draw from house excavations at two sites (Ejutla and El Palmillo [e.g., Feinman and Nicholas 2004a]), intensive mapping efforts at five sites (Monte Albán [Blanton 1978], Jalieza [Finsten 1995], El Palmillo, Guirún, and the Mitla Fortress [Feinman and Nicholas 2004b]), and systematic regional surveys conducted across the central valleys of Oaxaca (Bal­ kansky 2002; Blanton et al. 1982; Feinman and Nicholas 1990; Kowalewski et al. 1989). We begin with the household, the basic unit of production and consumption in the Valley of Oaxaca during the Classic period. Based on domestic excavations conducted at El Palmillo (Feinman, Nicholas, and Haines 2002) and Ejutla (Feinman and Nicholas 1993), it is apparent that the residents of Classic-period households produced a range of goods, some of which were exchanged. At Ejutla, marine shell ornaments and a variety of ceramic objects were produced domestically; at El Palmillo, householders utilized xerophytic plants and locally available stone to make a range of goods. Most significant are not only the clear differences between what households made at contemporaneous sites across the valley but also the fact that different domestic units at El Palmillo (only one residence was excavated at Ejutla) were engaged in distinct suites of productive activities. For example, the reduction of local stone for the production of chipped stone implements was clearly a much more important economic pursuit for some El Palmillo households than for others (Haines, Feinman, and Nicholas 2004). Likewise, select households at the site engaged in lapidary work and ceramic manufacture, whereas others did not. Even in regard to the working of fiber at El Palmillo, some households were more heavily engaged in spinning; others were more involved in weaving (Feinman and Nicholas 2005). Based on the assemblage of spindle whorls recovered in different houses, some households worked thinner/ finer fiber, and others focused on coarser yarns. The production evidence alone indicates household-to-household interdependency and a significant volume of exchange within single communities (Table 4.1). With domestic consumption, the evidence for household interdependence is solidified. No El Palmillo household made the full range of goods it consumed or used. Based on house excavations, it is clear that each household had to procure various items from outside the domestic unit. These acquired goods included basic commodities such as pottery, chipped stone (chert) tools, and ground stone implements probably produced for some El Palmillo households by others at the site, as well as other items, such as obsidian blades, where the raw material was procured from much 92

A M u lt i s c a l a r P e r s p e c t i v e o n M a r k e t E x c h a n g e

Table 4.1. Multiscalar Indicators of Market Exchange in the Classic-Period Valley of Oaxaca. Household

Site

Region

Production for exchange in households Household variation in production and access within sites

Site deficiencies in basic food staples (dry years) Differences in craft production between sites

Subregional differences in frequency and specific craft goods produced

Consumption of goods (basic Site-to-site variation in access to nonlocal goods (e.g., different commodities and exotic items) not produced by that household qualitative and quantitative representation of obsidian sources)

Some of the highest population densities in areas where maize production is less certain (e.g., eastern Tlacolula)

Open plazas on periphery of large sites, corresponding to site size No known supra-household storage facilities

farther away. Only one of eight excavated houses at El Palmillo yielded evidence of ceramic production, but every household used pottery. Obsidian blade fragments were found in every excavated house at El Palmillo, with this material coming from outside the state of Oaxaca. In addition, items such as greenstone, marine shell, and some highly decorated pottery varieties were recovered in small quantities in most residences, all of which were procured from beyond the local area. The terraces at El Palmillo were residential locales occupied by rooms and patios, with little level space for farming (Feinman, Nicholas, and Haines 2002). Relatively little flat land with adequate water is found within a kilometer or two of the site. The eastern edge of the Valley of Oaxaca, where the site is located, is the driest part of this semiarid valley; annual rainfall is at the lower limits of what is needed for a reasonable maize harvest (Kirkby 1973; Nicholas 1989). Today, little maize is farmed in the vicinity of the site, although subsistence agriculture was practiced in the area three to five decades ago. Some of this twentieth-century agriculture was enacted on those flat spaces (terraces and platforms) that were occupied and abandoned in prehispanic times (these areas were not available for cropping when the site itself was inhabited). Xerophytic plants, such as maguey, yucca, and agave, were likely key subsistence resources for the prehispanic residents of El Palmillo (Feinman and Nicholas 2005; Feinman, Nicholas, and Haines 2007). Nevertheless, both carbonized maize kernels and cob fragments have been recovered in our excavations, although not in great quantities, and maize likely composed a portion of the diet. Given the limitations of maize farming, however, and an El Palmillo Classic-period population conservatively estimated in the low thousands, it seems likely that at least some of the maize consumed at the site arrived through exchange from other settlements in the valley (Feinman and Nicholas 2005; Nicholas 1989). 93

Gary M. Feinman and Linda M. Nicholas

If most El Palmillo households acquired basic commodities such as maize and pottery from other households while producing craft items for exchange, then participation in networks of exchange was economically important to them. Such widespread and basic participation in inter-household transactions, in which householders depended on both neighborly interactions and exchanges at greater distances, would be hard, if not impossible, to control by a central authority. At the scale of the site, site-to-site variation is seen in the goods procured, such as obsidian, which appears not to have been doled out proportionally by a central authority, as one might expect in a redistributive economy. For example, based on preliminary results, the obsidian sources represented at El Palmillo are different from those identified at Ejutla, and neither site mirrors the source assemblage from Monte Albán (e.g., Elam 1993; Nolin 2006). Likewise, the differences in site-to-site production specializations also indicate a reliance on exchange. For example, the marine shell ornament production evidenced in a sizable sector of the Ejutla site is matched at few, if any, other valley communities. Finished shell ornaments, however, were not widely present at the Ejutla site and in fact were more abundant elsewhere. Consequently, the inhabitants of prehispanic Ejutla were likely not the principal consumers of what they manufactured (Feinman and Nicholas 2000). Finished shell ornaments from Ejutla were likely made largely for exchange with the residents of other communities. Likewise, the fiber products from xerophytic plants such as maguey that were produced at El Palmillo were probably more rarely made in wetter valley locations. Salt was produced in large quantities at another Late Classic–period site, Lambityeco (Peterson 1976). By the Classic period, significant quantities of goods were exchanged across the valley. Nevertheless, as noted previously, large central storage facilities remain undiscovered at Classic-period sites in the region. Given the volume of excavations at Monte Albán (e.g., Caso, Bernal, and Acosta 1967) and other valley sites, it seems highly doubtful that such storage facilities existed in the valley in prehispanic times. The apparent absence of such community-wide storage at large settlements calls into question an argument for large-scale, centralized redistribution as the main mode of economic distribution in the Valley of Oaxaca during the Classic period. Furthermore, because so many of the larger valley settlements were situated on hilltops, with high-status families generally living at the apexes of the sites, redistribution would have required many energetically costly trips up and down steep hills to the residences or other domains associated with rulers. Consequently, it is not surprising that no evidence exists for large-scale or systematic redistribution involving the transfer of basic commodities, such as utilitarian pottery, chipped stone implements, or ground stone tools. In contrast, the large Classic-period settlements in the Valley of Oaxaca often have extensive open plazas situated at their peripheries that may have been marketplaces. At Monte Albán, Blanton (1978:86) identified an open area near the base of the site as a possible setting for market activities because it was adjacent to a major road and surrounded by a residential zone in which many of the terraces yielded indications 94

A M u lt i s c a l a r P e r s p e c t i v e o n M a r k e t E x c h a n g e

of craft activity. Likewise, we suspect (Feinman and Nicholas 2004b:123–124) that a similar open plaza at El Palmillo may also have been a market area. This large, flat feature is located close to the base of the site, just below the densest concentration of residential terraces. Several ancient paths or roads lead up to this plaza from the base of the site, with one of them continuing up to the site’s densest residential zone. As at Monte Albán, the surrounding residential area was associated with a range of craft activities. In their argument for possible marketplaces in the prehispanic Mixteca Alta region of Oaxaca, Thomas Pluckhahn and Stephen Kowalewski (2003) synthesized available information on ethnographic marketplaces in Oaxaca. One variable they looked at was the size of the spaces where market assemblies were held. According to Ralph Beals (1973:123), the marketplace at Ayutla, the most important in the western Mijería, measures about 2,700 m2. A secondary marketplace in Nochixtlán, in the Mixteca Alta, covers 3,440 m2 (Warner 1976:121). The possible El Palmillo marketplace has an area of approximately 2,900 m2 and thus falls within this size range. At the Mitla Fortress, a smaller plaza on the lower part of the site’s southeastern spur may also have been a market area (Feinman and Nicholas 2004b:59). This open plaza is situated adjacent to a large platform and has several small stone foundations that do not look like room or house foundations. Four ancient roads, from all directions, lead to this open area. The possible marketplace measures 476 m2, only slightly larger than a tertiary marketplace (the lowest level in the contemporary Oaxaca market system) at Mitla (395 m2; Beals 1975:133). The proposed Early Classic Maya marketplace at Chunchucmil has been estimated to be roughly 1,500 m2 in size, which falls within this overall range (Dahlin et al. 2007:369). Although the archaeological definition of these open areas as marketplaces is difficult to confirm, their presence in conjunction with surface indications of specialized production and systems of roads leading up to them provides support for marketplace exchange in Oaxaca by the Classic period. Significantly, in the Mixtec region of Oaxaca, sixteenth-century marketplaces were often located near the boundaries of communities and not at settlement centers (Pohl, Monaghan, and Stiver 1997). At the regional scale, the buildup and relatively high density of population in the dry eastern arm of the Valley of Oaxaca during the Classic and Postclassic periods (Feinman and Nicholas 1990, 2005) would also seem to have necessitated a growing reliance on exchange. As previously noted, a good number of late prehispanic sites in this dry part of the valley could not have grown sufficient corn in their immediate agricultural catchments to have met annual subsistence needs with maize (Nicholas 1989). This problem would have become particularly acute in dry years. Our recent excavations at El Palmillo (Feinman and Nicholas 2005; Feinman, Nicholas, and Haines 2007) have illustrated that the exploitation of xerophytic plants for food, alcohol, and fiber was a key to life, as it is in this part of the valley today. Even if maize was not as much a staple as once thought, it probably was still consumed and desired even in places where it would have been risky to grow. Maize was likely supplied to 95

Gary M. Feinman and Linda M. Nicholas

these drier parts of the valley from wetter areas, at least periodically. Given the populations involved, the weight of grain or flour, and the transport technologies available, marketplace transactions seem much more likely than redistribution as the principal means of exchange. In this regard, it is not surprising that the driest sectors of the valley appear to have had the greatest reliance on craftwork during both the Classic and Postclassic periods (Feinman and Nicholas 1992; Kowalewski et al. 1989). Alcohol, fiber products, fruit, and other goods could have been exchanged for maize. These spatial patterns of exchange extended into post-contact times as well (Kowalewski 2003; Murphy, Winter, and Morris 1997), with the market system remaining the key economic mode of transfer in the Valley of Oaxaca today (Beals 1975; Cook and Diskin, eds. 1976). To synthesize from this multiscalar perspective, the economy of the Valley of Oaxaca during the Classic period was grounded in household production, in part for exchange, with significant variation in the specific goods neighboring as well as distant householders produced and procured from others. Patterns of consumption also varied from house to house and site to site, which, along with the absence of centralized storage facilities and the hilltop location of most of the region’s largest communities, diminishes the efficiency and likelihood of a central or all-encompassing role for redistribution. In addition, with most, if not all, production taking place in houses, such manufacture would have been difficult or even impossible for extant governing authorities to administer centrally. The scale of the valley populace—well over 100,000 people—and the reliance on and spatial extent of exchange spheres would seem to rule out a system in which reciprocity between known individuals functioned as the sole means of economic transactions. Rather, based on evidence from multiple scales and little support for possible alternatives, we propose that institutions (markets and marketplaces) that have been considered practically iconic (e.g., Chase 1992) for understanding the people and the economy of the Valley of Oaxaca during the twentieth and twenty-first centuries are also key for the history of the region 1,200–1,500 years earlier. In fact, it is conceivable that marketplace exchange was important in the Valley of Oaxaca even before the Classic period (Feinman, Blanton, and Kowalewski 1984; Winter 1984). As early as Monte Albán I (ca. 500–300 BC), basic utilitarian goods, such as everyday ceramics and food products, likely were broadly exchanged in volume across the Valley of Oaxaca. But the volume of these intraregional exchanges intensified during the Classic period, with higher regional populations, larger and denser settlements, and greater local imbalances between available resources and estimated numbers of inhabitants (Nicholas 1989). Concluding Thoughts and Implications

We have proposed that a market-based exchange system was important in the Valley of Oaxaca by at least the Classic period and likely earlier, thus prior to the arrival of 96

A M u lt i s c a l a r P e r s p e c t i v e o n M a r k e t E x c h a n g e

the Spanish and even the Aztecs. Yet we are not suggesting that these markets were either disembedded from Classic-period Oaxacan economy and culture or subject only to the laws of supply and demand. Rather, our interpretations are underpinned on a broader conception of market systems than traditionally has been envisioned in the disciplines of economics and anthropology. In concert, we have argued that in the absence of textual information, the documentation of a market system is a challenge, with possible indicators of this economic mode likely subject to issues of equifinality and plausible alternative interpretations. For that reason, we have proposed and employed a multiscalar model that marshals data from the house through the settlement to the region. When considered jointly and in context, the examination of these multiscalar and diverse sets of empirical information allows a strong argument to be made for market-based exchange, especially as alternative exchange modes do not dovetail well with the data at hand. At the same time, by examining evidence (in a sense, test implications) at multiple spatial scales, our model has the advantage of being underpinned by a series of independent empirical findings and is more firmly supported (sensu Chamberlin 1965 [1890]). If the market was indeed a significant institution in prehispanic Oaxaca and Mesoamerica earlier than Aztec times, then future archaeologists have their work cut out for them. In Oaxaca we must learn more about the organization of household labor, specifically how households participated in broader networks of exchange. More broadly for Mesoamerica, we should investigate how the market institution interrelated with the different political institutions that composed ancient Mesoamerica. How did market-based exchange vary over time? Critically, when did markets develop, and what factors prompted their emergence? How did such exchanges interrelate with tributary demands and other economic activities? Finally, why was the ancient Mesoamerican economy so different from what the Spanish described for the Andean highlands of South America, where market institutions were apparently far less prominent throughout the prehispanic era (Stanish 1997, Chapter 9)? How similar or different were Mesoamerican markets from European or Asian markets of the same period? Although not easy to address, questions such as these provide welcome challenges that, it is hoped, will guide archaeological investigations—through continued research, analysis, and interpretation—during the decades ahead. In sum, we hope this volume and other new views of past economic systems that are emerging across the behavioral and historical sciences will help spur research on these issues and generate new perspectives on markets, their variability, histories, and change in prehispanic Mesoamerica and beyond. If marketplaces and market exchange were more widespread and diverse than anthropologists have traditionally thought, that does not make them uninteresting. In contrast, such a finding leads to a large body of questions concerning cross-cultural variation and diachronic change in markets and economies (e.g., North 2001). At the same time, it establishes a more realistic multidisciplinary bridge for probing, understanding, and explaining how contemporary economies may have differed from those deeper in the past. .

97

Gary M. Feinman and Linda M. Nicholas

Acknowledgments. We thank Chris Garraty and Barbara Stark for inviting us to participate in this volume and in the session at the 2007 Annual Meeting of the Society for American Archaeology that preceded it. Their editorial suggestions, along with those of two external reviewers, have helped elucidate our perspective. It is our honor to be included here along with some of the scholars who have advanced the issues discussed in this volume. Our fieldwork over the years has been supported generously by the National Science Foundation (BNS-89-19164, BNS-91-05780, SBR-9304258, SBR-9805288, BCS-0349668), the National Geographic Society, the Heinz Foundation, the Foundation for the Advancement of Mesoamerican Studies, the Negaunee Foundation, the Field Museum, the University of Wisconsin–Madison, and other generous parties. We owe them all a debt of gratitude. We are also grateful for the significant help we have received from the Instituto Nacional de Antropología e Historia and the Centro Regional de Oaxaca. We also thank the people and authorities of Ejutla de Crespo, Santiago Matatlán, and San Pablo Villa de Mitla for their support. We are especially indebted to our field and laboratory crews for the effort and sweat they gave to advance the projects on which our interpretations are based.

98

Chapter Five

Origins and Development of Mesoamerican Marketplaces: Evidence from South-Central Veracruz, Mexico

Barbara L. Stark and Alanna Ossa

Archaeological evaluation of long-term market activity in Mesoamerica derives primarily from investigations in the Valley of Oaxaca, located in the southern highlands of Mexico (Blanton 1983; Blanton et al. 1982:23–25, 55–61, 65–68, 207–208; Kowalewski et al. 1989:294; Feinman and Nicholas, Chapter 4). Apart from the Valley of Oaxaca, market-oriented studies typically concentrate on the Postclassic period, for which a combination of ethnohistoric and archaeological data documents a vigorous periodic market system in the central highlands of Mexico (Berdan 1977, 1985; Blanton 1996a; Garraty 2006; Hassig 1985:67–84; Minc 2006; Nichols et al. 2002). Recently, archaeological examination of marketplace exchange in the Basin of Mexico has been extended to address the capital city of Teotihuacan during the Classic period (Sullivan 2007). In contrast to these highland locations, the Gulf lowlands typically had smaller states and contrastive environmental and settlement conditions, with higher rainfall and less nucleated settlements. Given these different conditions, the inception and circumstances of market activity in the Gulf lowlands will help us understand the variety of trajectories for market development in Mesoamerica. We evaluate distributions of obsidian prismatic blades to determine how early we can detect evidence of a market system in an area of the south-central Gulf lowlands of Mesoamerica. A further concern is tracing the history and organization of market development during the prehispanic sequence, particularly the articulation of market activities and local centers. In our study area, service areas are first evident 99

B a r b a r a L . S t a r k a n d Al a n n a O ss a

5.1. Part of Mesoamerica, with a gray rectangle indicating the study area.

at secondary centers but later are associated with primary centers. Our study builds upon prior research concerning locations of blade production (B. Stark 2007a). We follow Frederic Pryor’s (1977:31) definition of market exchanges as those in which supply and demand are highly visible; our focus is marketplace exchange of products, not labor or land. We develop test expectations about marketplace service areas, solar markets, and itinerant local peddlers for analysis of obsidian blade distribution in the western lower Papaloapan basin (WLPB) of south-central Veracruz (Figure 5.1). The principal tributary is the Blanco River, with smaller flows in the Guerengo and Tlalixcoyan rivers, located south and north of the Blanco, respectively (Figure 5.2). Intensive survey in the WLPB covered 99 km2 in several separated blocks, with each block subjected to “full coverage” pedestrian survey (see B. Stark 2006; Stark and Garraty 2008). The coverage includes a related survey project (Speaker 2001). Systematic surface collections from individual residential units and structures in monumental groups provide the data used for this study. We collected all obsidian or other chipped stone (chert is exceptionally rare) and all rim sherds, decorated sherds, and unusual pottery forms (e.g., vessel supports). Because of the presence of scattered survey blocks, the representation of occupations at different distances from the monumental centers is patchy. The five centers to be examined are labeled in Figure 5.2. 100

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

5.2. Western lower Papaloapan basin, with landforms and survey areas indicated. Stars indicate the locations of centers to be analyzed as possible foci for blade distribution.

We examine distributions of obsidian prismatic blade during four periods: the Early Classic (AD 300–600), Late Classic (AD 600–900), Middle Postclassic (AD 1200–1350), and Late Postclassic (AD 1350–1521). A possible gap in the occupational sequence occurs from ca. AD 900 or 1000 through the inception of a local Middle Postclassic complex around AD 1200 or somewhat earlier. We assess declines in access to blades with greater distance in terms of possible marketplace service areas for five centers: Cerro de las Mesas, Nopiloa, Azuzules, Sauce, and Callejón del Horno, all of which fall within sequential periods. The political and economic implications of the diachronic patterns help define the relationships between states and market centers. We observe initial market exchange and specialized production focused at secondary centers instead of the primary center. In the Valley of Oaxaca, during the Monte Albán I phase (which falls within the Late Preclassic period [600–100 BC]), specialized production that could feed into marketplaces was not characteristic at administrative centers in general (Blanton 1983; Blanton et al. 1982:57), but centers may have had marketplaces. For later Gulf periods there is at least a spatial association of the foci of production, marketing, and seats of power. In accord with expectations expressed by Richard Blanton and colleagues (1993:212–214), there is continued or even increased evidence of local market activity after the breakup of a unified realm under Early Classic Cerro de las Mesas, which was succeeded by separate independent 101

B a r b a r a L . S t a r k a n d Al a n n a O ss a

polities during the Late Classic period. They suggested that weaker, more decentralized political authority might allow greater scope for mercantile activity. An important outcome of our analyses is the indication that market system growth occurred simultaneously in multiple locations across Mesoamerica, both in highland valleys where major states developed and in smaller polities in the Gulf lowlands. This finding provides guidance concerning the extent to which we should regard the long-term history of Mesoamerica as exemplifying (1) core state innovations (such as from Teotihuacan, a major Mesoamerican expansionist state) versus (2) multiple-core innovations (such as Monte Albán, Teotihuacan, and others in later periods) (cf. Kohl 1988) or (3) a non-core multiregional perspective. The last focuses on processes at a regional scale balanced with attention to extra-local processes, such as long-distance trade in obsidian products. The Gulf area we examine is not a core area that developed extensive influence over neighboring polities or that provided unique economic innovations affecting other areas (although it was part of various broader economic changes). Thus, the third alternative best exemplifies the regional economic history we examine. Blanton (2004:226–227) compared Mediterranean and Mesoamerican settlement patterns and suggested that Mesoamerica displays certain core-periphery processes better than the Mediterranean area; nevertheless, our study does not support this characterization of Mesoamerican economic change. Models and Expectations

Empirical study of the origin and development of Mesoamerican market systems has not received the attention it merits, in part because of the challenges of acquiring appropriate distributional data (Minc 2006:82). Several studies provoked our focus on the history of obsidian prismatic blade distribution. Blanton and colleagues (Blanton 1983; Blanton et al. 1982:23–25, 55–61, 65–68, 207–208) proposed a wide-ranging set of ideas concerning the origins of Mesoamerican market systems in Oaxaca, while Kenneth Hirth (1998) considered how market systems can be distinguished archaeologically from other modes of exchange in highland Central Mexico. Market Origins

Blanton (1983, 1985; Blanton et al. 1982:23) discusses the origins of Meso­ american regional marketing, not “border markets” that responded to ecological or environmental zones that had complementary products and separately organized societies. Blanton and colleagues obtained longitudinal data about political organization and economy in the Valley of Oaxaca, located in the semiarid Mexican highlands (Blanton 1978; Blanton et al. 1982; Kowalewski et al. 1989). Monte Albán became the seat of power and one of the earliest of the large urban states that grew to dominate Classic-period Mesoamerica. Like Teotihuacan, it collapsed during the Late Classic period, leading to centuries of state fragmentation and volatile political conditions 102

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

that preceded the Aztec (Triple Alliance) empire. In Blanton and colleagues’ arguments about market origins, the interplay of political and economic conditions is a preoccupation, along with fundamental economic questions, such as whether “topdown” or “bottom-up” processes were at work. Blanton (1983; Blanton et al. 1982) dates the first market system in the Valley of Oaxaca to the Monte Albán Early I period, starting around 500 BC, thus coinciding with the foundation of the capital that administered much of the Valley of Oaxaca. Blanton (1983) and Blanton and colleagues (1982) mention five categories of evidence: (1) pottery changes suggesting more specialization in production; (2) the first clear ceramic production locations; (3) pottery distributions harkening to some central-place theory precepts, that is, higher-order goods (e.g., fancier pots) produced in a few places but lower-order goods (utilitarian vessels) produced at more places spread over a wider area; (4) open areas at Monte Albán and a few other sites that were well positioned to have served as marketplaces (Blanton et al. 1982:56; Kowalewski et al. 1989:294; Feinman and Nicholas, Chapter 4); and (5) an absence of centralized storage facilities that might have played a role in a redistributive economy. In the Oaxacan study, as with our case study, we can leave open the possibility that some earlier foundations of market systems existed but are more difficult to detect (see Feinman 1982:188). In the trajectory of changes affecting the market system, Blanton (1983) proposed that agricultural intensification and central tribute demands promoted specialized production because farmers could no longer devote as much time to production of household items. Marketplaces helped solve the problem of distribution from specialists. This model does not account for why people become specialists, however. The role of the state is important in this argument, with the idea that major states like Monte Albán afforded sufficient protection for interdependent specialists and consumers to engage in market exchange. With the decline of large states during the Late Classic period, especially Monte Albán and Teotihuacan, Blanton and colleagues (1993:212–214) suggested that market systems would have continued because of the interests of producers and consumers, who could, additionally, benefit from less interference (e.g., taxation or other regulation) by powerful governments. The outcome was a flourishing periodic market system by the Late Postclassic period, as described in early colonial documents. The Oaxacan data suggest that to study market origins we should particularly examine the Late and Terminal Preclassic periods (600–100 BC and 100 BC–AD 300, respectively). Because of the nature of our obsidian data, however, we begin our longitudinal study with the Early Classic period. In parallel fashion with the Oaxacan work, we find evidence of market exchange at least as early as that period. Regional and Local Models

At a regional scale, the interrelationship of service centers hosting marketplaces for the surrounding population generates more than one possible spatial pattern 103

B a r b a r a L . S t a r k a n d Al a n n a O ss a

(Christaller 1966 [1933]). Central-place theory was developed to model retail marketing, recognizing price competition and effects from transport costs that, for a reasonably evenly distributed population, will lead to multiple retail centers in various regular lattices. The particular lattice of centers depends on whether transport, market access, or administrative convenience is emphasized (Plattner 1975). Empirical evaluation of central-place lattices requires a regional scale and a large set of centers representing different levels in the settlement hierarchy. Although we will show evidence of multiple service areas in our study locale during some periods, we lack a sufficiently spatially extensive dataset to permit evaluation of central-place lattices. Instead, we examine service areas for one to three centers, depending on the period. Expectations concerning a “solar market” guide our analysis (C. Smith 1976b:36– 39). A solar market involves a center and its surrounding service area, potentially including subsidiary settlements, without an assumption of competition from neighboring centers. The surrounding population is dependent upon a single center for service functions, whether for retail marketing or political administration. Families and individuals relatively close to the center have easier access (less travel time) than people living at a greater distance, who may travel to the center less often and are therefore less well served by a marketplace (Plattner 1976). Normally, this degree of monopoly is unlikely because there are neighboring centers, but solar market principles could predominate in marketplace activities even if neighboring competing marketplaces exist. The administrative principle in central-place theory incorporates exclusionary relationships, for example. In the Mesoamerican lowlands, dispersed occupation is common, and we can visualize the service area comprising people in or near the center plus rural farmsteads (rancherias) in the surrounding lands administered by that center. Centers are usually obvious as monumental cores, with occupation extending outward into the countryside without obvious boundaries (non-nucleated settlements) (B. Stark 1999). In addition to marketplace exchange in centers, more distant market exchanges may be accomplished through itinerant local peddlers who range out to underserved areas, visiting periodically, with demand accumulating between visits (Plattner 1976: esp. 79–80). The peddler assumes the transport costs but has the advantage of reduced competition when arriving to vend products. Areas served only by itinerant peddlers are likely to exhibit reduced access because of the greater constraints on access and the greater distance costs. The spatial effect of a solar market combined with itinerant peddling consists of a well-serviced area near the center, for which distance costs are modest per item—given multipurpose trips that include social activities—and decreasing access to products with increasing distance beyond the service area because of distance costs. For rural Java, Glen Chandler (1985:esp. 160–162) described a combination of central periodic marketplaces and mobile traders but with some vending through mobile wholesaling and in rural homes. The mobile traders form a nonhierarchical network that operates independent of a central market service area and covers a larger service area. 104

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

In our case study we examine only one product to evaluate marketplace exchange— obsidian blades, which may have been exchanged through itinerant peddlers as well as in the marketplace. Our focus is not on blade production but instead on finished blade distribution. Obsidian blades have the advantage of small size and, despite their fragility, could be transported successfully in small numbers for modest distances if padded in containers. As Kenneth Hirth and colleagues (2006:133) noted, peddlers with “knives” (blades) and other goods operated in Aztec times, as described in early Colonial-period documents. Peddling products in the “countryside” (or distant markets) is different from the itinerant obsidian craftsmen discussed by Hirth (2008; Hirth and Andrews 2002:9; Hirth et al. 2006:132–135); at Xochicalco these itinerant specialists were merchant-knappers arriving with cores to produce blades and vend them in the central marketplace (also, vending nearly exhausted cores to local knappers). As an alternative to itinerant local peddlers, we can imagine a “penumbra” distribution area around a solar market center in which distant rural families only rarely travel to the center, perhaps more for religious or political reasons than economic ones, but who occasionally access the marketplace goods in the center (Abbott, Chapter 3). These penumbra families would have access to most goods but on a less frequent basis, with the caveat that some more costly goods might become excessive for such families when transport costs are added; overall, such families might have reduced consumption of products from the central marketplace. Of course, a penumbra of poorly served customers and local peddlers could both be responsible for lower access to blades at a distance beyond the market service area. Archaeological Criteria

A solar marketplace should show fairly comparable access (abundance) throughout the market service area because consumers reach the center easily for market transactions. Blanton (1996a:59) suggested service areas extending from 4 to 8 km, and Leah Minc (2006:99) found distances of 8 to 12 km in regard to ceramics in the Postclassic Basin of Mexico. A day’s round-trip on foot would involve a conservative maximum radius of 10 km, interpolating from diverse commodity values and load sizes and allowing for time at a destination (Drennan 1984:105; Malville 2001:234). At times, individuals may stay overnight, permitting a day’s travel to circa 20 km. Access should decline with distances greater than 10 km (i.e., beyond 10 km we might observe the penumbra served by itinerant peddlers and infrequent central marketplace trips). If a different solar market area is intersected with increasing distance, access to the product should improve again, but with reference to a different center. Access can be measured by the ratio of an item to residential units or to some proxy for the residents, such as rim sherds, as used in this study. Clay resources for pottery making are widespread in Mesoamerica, including the study locale, and therefore families with poor or no access to a marketplace for other items are not necessarily deprived of vessels, as they can be produced at the household level. 105

B a r b a r a L . S t a r k a n d Al a n n a O ss a

A market system, perhaps including itinerant peddling, is not the only context in which a falloff or decline in access to products is correlated with increasing distance (see Stark and Garraty, Chapter 2). Colin Renfrew (1975:46) noted that this general pattern could reflect “supply zone” behavior, where proximity to a natural source assures a population of ready access with short travel, followed by down-theline exchanges of various sorts—possibly involving gift giving rather than a market system. In a sense the market service area compares to the supply zone in Renfrew’s terms. Access declines with distance, possibly reflecting both peddlers and down-theline exchanges, and depends on whether another market service area is encountered. Analysis of such declines requires an adequate scale. For example, we detected signs of improving access at the greatest distance(s) in our analyses, suggesting multiple market service areas. This does not mean that down-the-line exchanges did not occur but rather that any effects were likely swamped by accessibility to other marketplaces. Discrimination between market institutions and gift giving or other exchange mechanisms requires residential data. Hirth (1998) argued that we can detect marketing from residential distributional patterns, which he examined within the city of Xochicalco in Morelos, Mexico. Because markets allow buyers and sellers considerable flexibility to conduct voluntary exchanges and negotiate exchange rates without social restrictions, Hirth reasoned that the operation of markets would yield relatively comparable inventories at all residential areas or units, although he recognized that there might be differences related to the costliest items. Michael Smith (1999), elaborating on Hirth’s ideas, emphasized that households’ economic resources will affect purchasing power; thus, access will vary by socioeconomic position, especially for more costly products. Distance from the marketplace also has an effect on access to goods; acquisition of marketplace goods becomes more “expensive” as distance increases as a result of transport costs. We adapt Hirth’s distributional method to regional data with the added assumption that proximity to a center is likely to be more characteristic of wealthier or higherstatus residential units because the centers are nexuses of civil and ritual authority in addition to economic transactions. At least for the Postclassic period, Christopher Garraty and Barbara Stark (2002) showed higher percentages of finely decorated ceramics in surface collections located closer to centers in the Gulf study zone. By examining access to obsidian prismatic blades at the residential level according to distance from the center, we incorporate a range of social classes and statuses. We do not conduct a separate analysis to subset residential units by social class or status, however, which is beyond the scope of this study. Central redistribution is possible with obsidian blades because of their light weight and low household consumption rate, but central redistribution is typically strongly skewed to favor higher-ranking households (Stark and Garraty, Chapter 2). If such households cluster near the center, there would be better access to obsidian near the center, mimicking a marketplace service area. In agrarian societies, however, some high-ranking households may also be located in a spotty manner in keeping with 106

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

their landholdings; thus, obsidian distribution would likely exhibit an irregular pattern rather than solely a plateau near the center. As discussed later, for some periods in our case study other goods such as decorated bowls also suggest marketplace service areas. Pottery is more bulky, with a higher annual consumption rate than obsidian blades, and it is less likely to be distributed effectively to households on a regular basis through central redistribution (Stark and Garraty, Chapter 2). As our examination of the Early Classic period reveals, the most powerful center was not a focus of obsidian distribution, which makes central redistribution less likely as a mechanism than marketplace exchange. Although obsidian was imported to our study region, it apparently was used nearly universally in residences, and we consider it a non-luxury item. No other stone suitable for flint knapping was available for making cutting edges. Some surface collections from residential mounds may not include obsidian because it is less abundant than pottery, leading to a higher likelihood of sampling error than for pottery. Our methods create distance groupings (distance rings) to better accommodate sampling error, as discussed subsequently. The Study Context and Cultural Sequence

Our analysis of market system development and change begins with the Early Classic period because prismatic blades did not become widely disseminated in this region until perhaps the later part of the Terminal Preclassic period; they were widely used by the Early Classic period (Heller 2001; Stark et al. 1992). We have no other product for which production evidence and widespread distribution can be tracked readily through all the periods considered. Other products either have greater sampling error because they are rare (lapidary products), or they involve widespread access to raw materials (e.g., pottery clays) that make it risky to make assumptions about distribution from centers without materials science analyses. Ceramic materials analyses have not yet provided the necessary degree of resolution within the region to track production and distribution (Stark, Speakman, and Glascock 2007). Obsidian is imported from sources in the highlands and was predominantly employed to produce fine prismatic blades used by all households. Because of the reduction technology, general locations of blade production can be detected even from the surface, provided that enough blade making occurred to detect unusually high concentrations (B. Stark 2007a). Obsidian blade production was concentrated at some centers in this region (B. Stark 2007a). Thus, obsidian is particularly useful for analysis of production and distribution systems in the study area. We rely on prior studies that established a chronological pattern regarding obsidian procurement involving a strong association of color, texture, and other visual properties with the geological sources determined by instrumental neutron activation analysis (Heller 2001; Stark et al. 1992). Black to dark gray obsidian that likely derived from Zaragoza-Oyameles in Puebla-Tlaxcala predominated during the 107

B a r b a r a L . S t a r k a n d Al a n n a O ss a

Classic period; clear gray obsidian that likely derived from Pico de Orizaba, Veracruz, predominated during the Middle Postclassic period; and green obsidian, likely from Pachuca, Hidalgo, in the central highlands, was prominent during the Late Postclassic period (Heller 2001; Stark et al. 1992). We predominantly analyze fine prismatic blades but also include percussion blades and ridge blades (blades showing knapping to establish a ridge down a core to help guide blade detachment or to correct errors); the latter two blade types might have circulated as cutting tools, even if they tended to be by-products concentrated where blade making occurred. Evidence for obsidian blade production was recovered at four of the five centers we evaluate for this study (B. Stark 2007a). Pottery provides the basis for separating sets of residential mounds for each period examined, and the changes in obsidian sources used in the study region also help differentiate materials for each period. Residential mounds tend to be reoccupied because drainage is advantageous during seasonal rains in the low-lying terrain, leading to numerous multi-component collections. We concentrate on a subset of collections and materials that can be strictly associated with each period. Prior chronological and settlement studies have established the broad outlines of the cultural sequence in the WLPB (Curet, Stark, and Vásquez Z. 1994; B. Stark, ed. 2001; Stark and Curet 1994). We next briefly summarize the settlement and political history of the region and the associated changes in obsidian importation and production to provide a backdrop for the analysis. Early Classic Period

The core of Cerro de las Mesas forms the largest mapped monumental zone in the WLPB. Located in the delta of the Blanco River, it clearly dominated the region during the Early Classic period, with secondary centers at Nopiloa, Cerro de los Muertos, Tuzales, and likely Azuzules. Apart from our surface collections, Early Classic dating is best established for Nopiloa, where Medellín Zenil (1987) conducted rescue excavations. Prismatic blades dominated the obsidian industry during the Early Classic period, with black–dark gray obsidian from Zaragoza-Oyameles forming the majority of the material (Heller 2001; Stark et al. 1992). Previously, during the Late Preclassic period, at least part of the WLPB was still reliant mainly on flake technology. Access to prismatic blades was likely increasing through the Terminal Preclassic period, but we cannot track this change effectively through the survey collections because too few diagnostic ceramic traits are available to cleanly separate collections from the Terminal Preclassic and Early Classic periods. Considerable continuity exists in pottery types from the Late Preclassic to Early Classic periods (B. Stark, ed. 2001). Surface indications of concentrated craft activities for obsidian blades, pottery, or lapidary items were not detected at Cerro de las Mesas, which seems to have been only modestly advantaged in access to obsidian (Heller and Stark 1998). Black or dark 108

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

gray blade production is not concentrated at Cerro de las Mesas (B. Stark 2007a:240– 243). Consequently, we do not expect to uncover evidence of a market service area (supply zone) from this center. If blades were imported and central redistribution occurred at the capital (rather than blade production there), then we would expect them to be concentrated in the Cerro de las Mesas vicinity because of better access among high-ranking households, with a falloff in more distant locations (with peaks at secondary centers). Instead and unexpectedly, the distribution patterns led us to consider blade production and marketplaces for two secondary centers, Nopiloa and Azuzules, mainly the former. Late Classic Period

Cerro de las Mesas persisted as a settlement during the Late Classic period but lost command of its former realm, and at least four smaller centers succeeded it. One of the successor centers, Azuzules, was located to the east in the Blanco delta. Nopiloa was located along the Guerengo River south of the Blanco. Tio Perciliano was located in the mangrove swamp, and Ajitos-Pitos was situated on the paleodunes. The successor centers diverged slightly in architectural layouts but continued to exhibit some of the same canons regarding buildings and arrangements that were prominent in the Early Classic period (and earlier). Each may have dominated a realm of 300–500 km2 (we lack sufficiently extensive survey to make a precise statement). At Nopiloa, Azuzules, and Tio Perciliano, the immediate community exhibits evidence of intensive craft activities. Nopiloa appears to have continued its role in making and distributing prismatic blades from the Early Classic period. Orange bowl production was likely in the Azuzules area, which also exhibits evidence of low-level prismatic blade production (B. Stark 2007a:240–243; Stark and Garraty 2004). Ultra-fine orange paste vessels were produced in the Nacastle-Patarata community dominated by the Tio Perciliano monumental core (Stark and Garraty 2004). Because black and dark gray obsidian continued to be imported from ZaragozaOyameles during the Late Classic period, obsidian itself provides no chronological discrimination between the Early and Late Classic periods. For the distributional analyses presented in the next subsections, collections must be separated that have either Early or Late Classic ceramic diagnostics but none of the diagnostics of the other period. This requirement reduces the number of collections included in our study but offers insurance that we have eliminated residential units occupied during both periods. Middle Postclassic Period

By the Middle Postclassic period, a massive disruption of Classic-period (AD 300–900) settlement had occurred (Curet, Stark, and Vásquez Z. 1994; B. Stark 1995), and an apparently intrusive group with highland cultural affiliations appeared 109

B a r b a r a L . S t a r k a n d Al a n n a O ss a

in the Blanco delta, one of the largest “patches” of rich, well-watered farmlands in the WLPB (B. Stark 2008). The only known center is Sauce, located in the Blanco delta near Cerro de las Mesas. In addition to striking changes in the styles and forms of vessels and figurines, obsidian procurement and technology changed. Clear gray obsidian was imported from Pico de Orizaba (see Figure 5.1) and formed the primary blade material, with evidence for blade production at Sauce that now typically employed ground platforms (B. Stark 2007a:245–249). Some black–dark gray obsidian may have continued to be imported, but we analyze only the clear gray obsidian for this period. The clear gray obsidian for prismatic blades derives from mining at Pico de Orizaba, and this tunnel mining does not seem to have figured much, if at all, in earlier procurement there (Stark et al. 1992). Some Pico de Orizaba obsidian continued to be used in the subsequent Late Postclassic period but in greatly reduced amounts. The eastern edge of the Sauce region appears not only to have evidence of production of black-on-orange and black-on-red bowls (in the Lobato area, according to Stark and Garraty [2004]) but also reflects a separate obsidian distribution network, to judge from Garraty’s (2009) distributional information. Stark and Garraty (2004:139) speculated that one possibility concerning the eastern Lobato area was a separate administrative center operating somewhere in the vicinity. Late Postclassic Period

During the Late Postclassic period, Sauce and much of the Middle Postclassic cultural complex declined, with a new center founded not far upriver along the Blanco River—Callejón del Horno. Callejón del Horno is the only Late Postclassic center recorded by our survey, although documents attest to a center to the north, Tlalixcoyan along the Tlalixcoyan River, about 15 km northeast of Callejón del Horno. Other centers are known to the west and east from documentary sources, including Cuetlaxtlan and Tlacotalpan. Callejón and its subsidiary rural area appear to have been integrated into the Aztec empire, and Callejón may have been subject to the provincial capital of Cuetlaxtlan to the west along the lower Cotaxtla River. Obsidian importation patterns shifted again, with green obsidian from the Cerro de las Navajas (Pachuca) source forming a large percentage of obsidian processed and used at Callejón del Horno (B. Stark 2007a:250–253). Discussion

This settlement information implies that Sauce and Callejón del Horno might exhibit a solar market pattern because any competing centers are either unknown or relatively far away. For example, in the Late Postclassic period, pottery made in the Callejón area seldom reached the Aztec provincial head town of Cuetlaxtlan, and vice versa (Skoglund et al. 2006). During earlier periods, with more than one center to 110

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

examine (sometimes secondary centers), we may detect more than one market service area. Garraty and Stark (2002) analyzed the distribution of pottery from residential units around the two known Postclassic centers with respect to indications of social hierarchy and wealth differentiation. They considered the possible market zone of Sauce in respect to pottery. Recently, Garraty (2009) used residential units around Sauce to detect evidence of a market zone on the basis of the diversity of decorated sherds per residential assemblage, according to distance. He found evidence of a “plateau” of comparable diversity values out to a distance of approximately 9 km, in keeping with the idea of a market service zone for Sauce. He compared pottery results against patterns for obsidian blade distribution, examining access according to the ratio of obsidian blades to Middle Postclassic rims; he calculated the median ratio for the collections within each of his 1-km distance rings. Obsidian patterns conformed to the decorated pottery patterns but with a decline beyond a smaller plateau extending approximately 6 km. Garraty also noted increased access to obsidian in the most distant zone from Sauce (including the Lobato area), approximately 13 km away, possibly indicating a different supply zone or mechanism, such as middleman traders. Although Garraty analyzed the Sauce market service area, we reanalyze the Sauce case using identical techniques to those we apply to the other periods in order to establish a longitudinal evaluation of marketplaces on a comparable basis for each period. Unlike Garraty, our focus is solely on obsidian blades. Based on the combination of Mesoamerican documentary evidence for the Late Postclassic period that indicates widespread markets in Mesoamerica and the work Garraty and others have accomplished for the Middle and Late Postclassic periods in the WLPB, obsidian evidence from each of these periods is expected to show the extents of market service areas around centers in our study area. A major issue is whether, or in what ways, periods prior to the Postclassic exhibit marketplace patterns. How far back can we detect obsidian marketing, or should we entertain different distribution mechanisms such as centralized redistribution? Methods

We examine our data in two ways, one that aggregates residential information concerning obsidian access by distance rings from each center and a second devised by Ossa in connection with her dissertation research that compiles data for individual residential units in each distance ring. We use 2-kilometer distance rings to reduce the likelihood of sampling error that would result from including too few collections per ring. Even so, we combine some distance rings that have too few collections from particular periods or because they scarcely intersect our survey blocks. Although the distance rings yield increasing area as more rings are added, the results in Table 5.1 show that the number of collections per ring is often as large or larger in inner rings with a small area because of greater settlement density around centers, as well as the 111

B a r b a r a L . S t a r k a n d Al a n n a O ss a

5.3. Example of 2-kilometer rings around the center of Cerro de las Mesas in the Early Classic period. Stars indicate the locations of the five centers analyzed during the prehispanic sequence.

patchiness of the survey blocks (outer rings tend to intercept less survey coverage than do rings near monumental complexes). One advantage of combining information in rings is inclusion of both collections made over a measured area and those made where only spotty exposures allowed collections (because of vegetation). In our methods, both types of collections are employed. Figure 5.3 shows an example of rings extending from the Early Classic center of Cerro de las Mesas. For the first method, we calculate the ratio of total obsidian blades to pottery rims per ring for each period. Pottery counts help standardize our data because collection sizes vary, as do the numbers of diagnostic rims in each period. Pooling data by rings has the advantage of readily incorporating small collections and buffering against the likelihood of sampling error, which pertains more to obsidian blades than to pottery because of the greater scarcity of blades. This method still benefits from the systematic sampling of residential units, however. As a shorthand, we refer to this as the “ring-average” method. For the second method, we first establish the overall median of ratios of obsidian blades to rims for all residential collections per period. Then, for each 2-kilometer distance ring we calculate the percentage of collections above the overall median (in contrast to Garraty, who examined the median per ring). This approach focuses on individual collections, but by using the overall median value it reduces distortions resulting from outliers. Outlier values could represent, for example, a collection 112

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

reflecting prismatic blade production that enriched blade fragments compared to the amount of pottery. Likewise, a collection might represent a residential mound where pottery making occurred, enriching the amount of pottery compared to obsidian blades. With the first method, such effects could distort the pooled data for each Table 5.1. Obsidian Distributions from Centers per Period.

Distance of Ring in km

Number of Collections

Ring Number

Ratio of Blades to Rims

Percentage of Collections >= Median Ratio of Blades to Rims

Comments about Rings and Collections

Black–Dark Gray Obsidian Distribution with Reference to Early Classic Cerro de las Mesas 0 to 2

96

1

0.33

22

Sauce (Middle Postclassic)

2 to 4

54

2

0.19

13

4 to 8

45

3, 4

0.54

49

Ring 4 closer to Azuzules than to Cerro de las Mesas

8 to 10

28

5

1.89

57

Azuzules

10 to 12

35

6

0.87

72

12 to 16

19

7, 8

1.42

63

16 to 20

35

9, 10

0.76

77

20 to 30

38

11–15

4.72

92

Ring 8 Callejón del Horno (Late Postclassic) Ring 11 Nopiloa

Black–Dark Gray Obsidian Distribution with Reference to Early Classic Nopiloa 0 to 2

17

1

3.12

88

2 to 14

21

2–7

0.51

62

14 to 16

40

8

0.40

83

Ring 8 Callejón del Horno (Late Postclassic)

16 to 20

45

9, 10

0.13

29

Ring 9 also Callejón del Horno (Late Postclassic)

20 to 22

90

11

0.07

20

Cerro de las Mesas; Sauce (Middle Postclassic)

22 to 24

30

12

0.04

17

24 to 26

54

13

0.11

56

26 to 28

22

14

0.34

59

Azuzules

28 to 30

22

15

0.24

68

Also Azuzules

40 to 48

9

21–24

3.32

100

Ring 21 Ajitos-Pitos on paleodunes, Rings 22–24 NacastlePatarata and Tio Perciliano in mangroves continued on next page

113

Table 5.1—continued

Distance of Ring in km

Number of Collections

Ring Number

Ratio of Blades to Rims

Percentage of Collections >= Median Ratio of Blades to Rims

Comments about Rings and Collections

Black–Dark Gray Obsidian Distribution with Reference to Early Classic Azuzules 0 to 2

23

1

0.30

48

2 to 4

37

2

0.19

68

4 to 8

48

3, 4

0.04

15

8 to 10

92

5

0.06

19

Cerro de las Mesas; Sauce (Middle Postclassic)

10 to 16

22

6–8

0.13

23

Ring 6 also Sauce (Middle Postclassic)

16 to 20

32

9, 10

0.27

82

20 to 26

39

11–13

0.42

69

Ring 12 Callejón del Horno (Late Postclassic)

26 to 28

28

14

1.25

68

Nopiloa

28 to 32

29

15, 16

1.03

86

Ring 15 also Nopiloa

Black–Dark Gray Obsidian Distribution with Reference to Late Classic Nopiloa 0 to 2

38

1

6.24

97

2 to 16

21

3, 4, 8

1.18

75

Ring 8 Callejón del Horno (Late Postclassic)

16 to 18

15

9

0.28

40

Also Callejón del Horno (Late Postclassic)

18 to 20

23

10

0.11

30

20 to 22

120

11

0.24

43

22 to 24

61

12

0.04

10

24 to 26

50

13

0.10

28

26 to 28

67

14

0.31

51

Azuzules

28 to 30

71

15

0.34

56

Also Azuzules

30 to 32

27

16

0.24

59

32 to 38

21

17–19

1.76

81

38 to 46

29

21–23

2.53

93

Ring 21 Ajitos-Pitos on paleodunes; Rings 22-23 NacastlePatarata and Tio Perciliano in mangroves

46 to 50

10

24, 25

0.97

80

Ring 24 Nacastle-Patarata and Tio Perciliano in mangroves continued on next page

Cerro de las Mesas; Sauce (Middle Postclassic)

Table 5.1—continued

Distance of Ring in km

Number of Collections

Ring Number

Ratio of Blades to Rims

Percentage of Collections >= Median Ratio of Blades to Rims

Comments about Rings and Collections

Black–Dark Gray Obsidian Distribution with Reference to Late Classic Azuzules 0 to 2

68

1

0.37

38

2 to 4

67

2

0.26

60

4 to 6

76

3

0.13

26

6 to 8

106

4

0.25

49

8 to 10

91

5

0.42

39

Cerro de las Mesas; Sauce (Middle Postclassic)

10 to 12

21

6

0.11

29

Also Sauce (Middle Postclassic)

12 to 16

20

7, 8

3.83

65

16 to 18

70

9

0.71

65

18 to 22

22

10, 11

2.79

86

22 to 28

30

12–14

9.35

87

Ring 12 Callejón del Horno; Ring 14 Nopiloa

28 to 32

31

15, 16

13.59

94

Ring 15 Nopiloa; Ring 15 has aberrant collection (6387) with 640 blades and 7 rims

Clear Gray Obsidian Distribution with Reference to Middle Postclassic Sauce 0 to 2

173

1

0.32

39

Cerro de las Mesas

2 to 4

119

2

0.17

27

4 to 6

87

3

0.11

20

6 to 8

75

4

0.16

48

8 to 10

62

5

0.31

37

10 to 12

64

6

0.27

38

Azuzules

12 to 14

31

7

0.31

77

Lobato continued on next page

B a r b a r a L . S t a r k a n d Al a n n a O ss a

Table 5.1—continued

Distance of Ring in km

Number of Collections

Ring Number

Ratio of Blades to Rims

Percentage of Collections >= Median Ratio of Blades to Rims

Comments about Rings and Collections

Green Obsidian Distribution with Reference to Late Postclassic Callejón del Horno 0 to 2

33

1

0.26

70

2 to 4

34

2

0.12

18

4 to 14

15

3–4, 6–7

0.07

50

Ring 7 Sauce

14 to 16

59

8

0.19

31

Cerro de las Mesas & Nopiloa

16 to 18

48

9

0.12

38

Also Nopiloa

18 to 20

29

10

0.15

28

20 to 22

18

11

0.05

28

22 to 24

24

12

0.04

15

Azuzules

24 to 26

17

13

0.00

0

Also Azuzules

26 to 40

14

14, 16–17, 20

0.25

14

Ring 14 Ajtos-Pitos on paleodunes and Lobato; Ring 16 Rio Limon block; Ring 20 Nacastle-Patarata and Tio Perciliano in mangroves

ring if there are insufficient numbers of collections. We can expect more variability in values for the second method compared to the first because it responds more directly to individual collections, and the number of collections per ring varies. We refer to the second method as the “above-median” method. Both methods potentially show the edges of market service areas in terms of a “plateau” for a set of rings that reflect readier access to the marketplace, with a decline in more distant areas (perhaps served by itinerant traders or showing infrequent trips to the center). Whether a more distant area constitutes a lower plateau or a continuous decline in access with distance is likely to vary according to the scale of assessment. The decline in access with greater distance may be reversed if another market zone is intersected. Because obsidian is imported from distant regions and blade making tends to occur at centers for the periods examined (B. Stark 2007a), the first ring (which includes the center) is likely to exhibit an especially high overall ratio of blades to rims, in keeping with the direct access to producers and the effects of blade-making debris; this first ring (0 to 2 km) should have a particularly high proportion of collections above the overall median ratio. Assignment of collections to particular periods for our analysis presents different challenges for each period under consideration. The Classic-period analysis is particu116

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

larly difficult because the same obsidian source predominated in the region during both the Early and Late Classic periods—black to dark gray obsidian from ZaragozaOyameles. In the case of the Early versus Late Classic period, the collections used are only those that also lack the other period component, but for such collections all Classic-period rims and blades are counted, not just the ones diagnostic of the component being analyzed. Diagnostic types are mainly decorated serving vessels, some of which are fairly scarce. This scarcity creates problems of sampling error in individual residential collections, which are usually fewer than 100 rims (median of 68 rims). Therefore, despite the fact that occupational remains are densest for the Classic period, the datasets that can be ascribed solely to the Early versus the Late Classic periods are necessarily much reduced. Collections particular to each period are established by the absence of diagnostics of the other period as much as by the presence of diagnostics for the period in question. One further limitation for the Early and Late Classic periods is that we exclude survey blocks from Stuart Speaker’s (2001) project, although for all other periods they are included. The earlier pottery classification system he used does not discriminate the Early Classic period adequately. In the case of the Middle and Late Postclassic periods, the datasets are those identified by Garraty and Stark (2002) through a statistical “unmixing” procedure that ascribed some collections to each of these periods. Their multivariate regression analyses used a dataset for each of these periods as a reference model and assigned Postclassic collections of a sufficient size to one of the two periods or to a mixed category. These statistically assigned collections were supplemented with some smaller collections assigned by inspection to one or the other of these periods but that did not figure in the statistical analysis. Also, each of these periods has a dominant obsidian material, which further allows us to isolate each period for analytic purposes. To examine marketplace exchange diachronically, we begin by examining patterns for Cerro de las Mesas in the Early Classic period. During this period we analyze Nopiloa and Azuzules as secondary centers to Cerro de las Mesas. Nopiloa and Azuzules are also examined for the Late Classic period. Too little information is available concerning surrounding settlement for the other two Late Classic centers (AjitosPitos and Tio Perciliano) to make them viable candidates for analysis. Only Nopiloa and Azuzules show evidence of obsidian blade making in any case. Nopiloa has strong indications of blade making, with Azuzules exhibiting only modest evidence (B. Stark 2007a). Because of the suspicion that some black–dark gray or clear gray obsidian may have been available in periods different from their primary periods of use, we note in Table 5.1 the distance rings that include other centers (even if from a different period). This notation allows us to detect spikes in access that might suggest effects from a degree of production and use in different periods, but we did not observe such effects. For each period and center we occasionally group distance rings to achieve at least fifteen collections per ring, but in three instances the most distant ring does not 117

B a r b a r a L . S t a r k a n d Al a n n a O ss a

5.4. Graph of two methods analyzing the distribution of blades from Early Classic Cerro de las Mesas.

include that many collections. For rings with fewer than fifteen collections we provide the data in Table 5.1 but do not graph the rings because of their greater susceptibility to sampling error. Occasionally, rings intersect no survey area or collections for particular periods; we do not list these rings separately in Table 5.1, as they are combined with other rings. Diachronic Analysis of Obsidian Blades and Market Service Areas Early Classic Period

Although the immediate vicinity of Cerro de las Mesas has a better supply of blades than the next closest 2-kilometer ring (2–4 km), the third ring surrounding the center (4–6 km) has considerably more blades relative to rim sherds (Figure 5.4; Table 5.1). Cerro de las Mesas does not appear to have been a noteworthy focus of blade production and distribution, as Lynette Heller and Stark (1998) have remarked. The rise in proportions of blades in more distant rings provoked our analysis of possible distribution from the secondary centers of Azuzules and Nopiloa, especially because Nopiloa was such a marked locus of blade making during the subsequent Late Classic period. 118

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

5.5. Graph of two methods analyzing the distribution of blades from Early Classic Nopiloa.

During the Early Classic period, Nopiloa has a distinctly elevated supply of blades in the immediate area, as would be expected from blade production and ready access (Figure 5.5). Unfortunately, the next few rings (2–14 km) encounter little of our survey area and few collections dated to the Early Classic period. That area is reasonably well supplied with blades using the above-median method, as is the next ring, 14–16 km. After this distance a distinct drop-off occurs according to both methods of quantification, overcome only in areas beyond 26 km, near Azuzules. Thus, Nopiloa in the Early Classic period may have served an area larger than the usual standard market areas described by Blanton and Minc (see previous discussion), perhaps because other centers were minor suppliers. To examine this possibility we turn to another secondary center, Azuzules. Azuzules, like Cerro de las Mesas, does not have a remarkably elevated proportion of blades for the first ring covering its immediate vicinity (Figure 5.6). Blade production did occur there during the Classic period, although we cannot determine how much is related to the Early versus Late Classic–period occupation. If Azuzules hosted market activities, then the area served is modest, as a drop-off occurs by 4 kilometers on the basis of both methods, and the most distant rings show a rise in the proportion of blades, closer to Nopiloa. Overall, the Early Classic period does not show evidence of substantial blade making and market distribution of obsidian blades for the major center, Cerro de las Mesas. There is a reasonable case for Nopiloa, a secondary center and locus of blade 119

B a r b a r a L . S t a r k a n d Al a n n a O ss a

5.6. Graph of two methods analyzing the distribution of blades from Early Classic Azuzules.

making, with at least a 16-km radius likely supplied in a relatively regular fashion. The mechanisms may have been varied, from marketplace exchanges at Nopiloa to itinerant peddlers or secondary distribution at smaller centers. Itinerant peddlers are more likely to have operated at greater distances, between the areas served by Nopiloa and Azuzules. Conceivably, some complementarity in products and services knitted the regional system together, provided that orange bowl production at Azuzules antedated the Late Classic period (Stark and Garraty 2004:135–137). Cerro de las Mesas appears to have been a government and ritual service center, but the economic and political hierarchies only partly overlapped. Late Classic Period

The same pair of centers, Azuzules and Nopiloa, can be examined in the Late Classic period, when both rose to prominence. Nopiloa was a focus of blade making according to various indices (B. Stark 2007a), but Azuzules had only a modest indication of blade making. The Nopiloa rings again pose a problem, as several must be combined to provide a more robust number of collections near Nopiloa (Figure 5.7). In the immediate vicinity of the center, measures of accessibility are fairly high, matching indications of considerable blade making. Distribution may have occurred out to about 16 km on the basis of both methods, similar to the Early Classic–period distri120

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

5.7. Graph of two methods analyzing the distribution of blades from Late Classic Nopiloa.

butional pattern. Subsequently, there is a falloff until 24 km with both methods (but with a spike in the percentage of collections with the above-median method that corresponds to the Cerro de las Mesas area), after which blade-access measures improve again in the vicinity of Azuzules. An even more striking increase in access occurs in the most distant rings, which intersect both Ajitos-Pitos in the paleodunes and the mangrove community of Tio Perciliano. For these far-distant rings, the majority of the collections above the overall median are located in the mangrove zone (twentyfive collections from the mangroves versus seven from the paleodunes). Mangrove inhabitants had riverine access to the Papaloapan trade corridor and were likely able to tap a different distribution system. Late Classic Azuzules has scant indication of a market service area for obsidian blades, possibly supplying an area out to 4 kilometers on the basis of the above-median method, but the best access is not at the Azuzules core but slightly farther away (2–4 km) by this method. Little variation of any sort is seen out to 12 km with the ringaverage method (Figure 5.8). At greater distances amounts fluctuate, eventually rising markedly in more distant areas, from 12 to 22 km with the above-median method and beyond 22 km for the ring-average method, a distance that includes Nopiloa. In sum, for the Late Classic period, Nopiloa appears to have continued its economic activity from the Early Classic, which may have included market distribution of obsidian blades to a surrounding area somewhat larger than the usual standard market area—possibly reflecting the lack of strong competition. In contrast, Azuzules does not have a distribution pattern suggestive of much production and marketing of obsidian blades. Azuzules might have been a prominent focus of market exchange for other commodities, such as orange-slipped bowls (Stark and Garraty 2004). The mangrove area appears to have had good access to obsidian blades, indicative of an 121

B a r b a r a L . S t a r k a n d Al a n n a O ss a

5.8. Graph of two methods analyzing the distribution of blades from Late Classic Azuzules.

additional market service area (clearest in Figure 5.7). The mangrove area was also a focus of ultra-fine orange pottery production (B. Stark 2007a). By the Late Classic period, we have an indication of different products at different centers, establishing possible complementary flows of products through a regional market system (B. Stark 2007a; Stark and Garraty 2004). Middle Postclassic Period

The only identified center for this period is Sauce, which has one markedly specialized obsidian workshop (Mound 1756; Heller 2000) and other collections where clear gray obsidian was used for blade making. The Mound 1756 collections from the workshop are not included in the analysis because they are exceptional and they represent two collections made at the location, a situation not comparable to other collections. The surrounding Middle Postclassic occupation is largely confined to the Blanco delta, and distance rings with occupation do not extend as far as those for the Classic period. To obtain a sufficient sample for Sauce, we include the statistically unmixed collections and visually assigned collections as discussed in Garraty and Stark (2002:7–9). Sauce has an elevated ratio of blades in the immediate vicinity and somewhat fitful values that are nevertheless relatively consistent out to 8 km with both methods (Figure 5.9). In comparison, Garraty (2009) found an elevated proportion of obsidian blades out to 6 kilometers from Sauce, using 1-kilometer rings. His results approxi122

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

5.9. Graph of two methods analyzing the distribution of blades from Middle Postclassic Sauce.

mate ours in view of the difference in ring width employed in the analysis. A marked increase in blade access occurs in the 8- to 14-km range with the ring-average method, an indication of the intersection of a possibly distinct market service area. The abovemedian method does not show an increase until 12–14 km, however. The distance between 8 and 14 km includes the Lobato area, and Stark and Garraty (2004) speculated from ceramic data that another center might have existed beyond the survey coverage. (The use of only the statistically unmixed collections [no visual assignments of small collections] produces similar results but has no collections in ring 7, 12–14 km). Late Postclassic Period

The only center in the survey area for this period is Callejón del Horno. There are too few statistically unmixed collections for a useful distributional analysis, so both the unmixed collections and the visual assessments of small collections from Garraty and Stark (2002) are included. As is typical for centers with blade production, Callejón del Horno exhibits better green blade access in its immediate vicinity (Figure 5.10), with rings from 2 to 20 km exhibiting a fluctuating supply that vacillates within a plateau of values (the 4- to 14-km area has few collections). Beyond 20 km there is a marked drop-off with both quantitative measures. The 26- to 40-km distance is not graphed because of few collections but exhibits an increase in green blade access, possibly indicating another market zone. Part of this distant area includes the mangroves, where water routes provided different transport connections than were found in the western 123

B a r b a r a L . S t a r k a n d Al a n n a O ss a

5.10. Graph of two methods analyzing the distribution of blades from Late Postclassic Callejón del Horno.

riverine farmlands. The Callejón distributions are suggestive of a market zone that, as with Nopiloa, was more extensive than expected, likely because of the absence of any nearby competing center within the survey bounds. At a greater distance, Tlalixcoyan and Tlacotalpan (head towns mentioned in early colonial records) may have served as distribution nodes, offering consumers alternative marketplaces. Discussion

Antecedents of the Postclassic patterns indicate marketplace exchange during the Classic period when Nopiloa seems to have had a special role in producing and distributing prismatic blades, first within the Cerro de las Mesas realm as a secondary center and later as one of the four competing centers that dominated the region. Nopiloa did not have an exclusive role in blade making, however, because during both the Early and Late Classic periods, Azuzules hosted some blade making and likely supplied its immediately surrounding area. The absence of concentrated blade making at Early Classic Cerro de las Mesas (B. Stark 2007a) matches the lack of a service area in blade distribution from that center in our analysis. Instead, blade distribution patterns from Nopiloa and, to a very modest extent, Azuzules suggest a complex economic system in which Cerro de las Mesas played a religious and governmental role, with at least one of its major secondary settlements playing more of an economic role. A division of functions is not unprecedented in the Gulf area, as survey in the Tuxtla Mountains has shown that Matacapan 124

O r i g i n s a n d D evelo p me n t o f M eso a me r i c a n M a r k e t p l a ces

played a role in pottery production that served its sustaining area but Ranchoapan, located along another drainage to the west, had a role principally in obsidian blade making (Barrett 2003; Santley and Arnold 1996). This kind of division of functions is not indicative of redistribution because with that mechanism the political capital, Matacapan, would have been important in the distribution of both kinds of products. Gift giving does not lead to this kind of division of functions either, because it should yield a patchy process involving all centers. The Postclassic period provides evidence of a market service area for obsidian blade distribution for the two successive centers of Sauce and Callejón del Horno. In the case of clear gray blades and Sauce, there is an indication that the survey area intersected a different market zone to the southeast, although no center has yet been detected archaeologically in the area. For green blades and Callejón del Horno, there are indications of a separate market area far to the east in the mangrove swamp. A separate service zone for the mangrove area is indicated during the Late Classic period as well. Given the data compiled and the number of collections for each ring, it seems unlikely that an alternate distributive mechanism other than marketing was involved because we do not detect highly variable distributional patterns. Further, the distances and consistency in the obsidian distribution patterns are more in keeping with marketplace exchange than with redistribution or kin reciprocities. Nevertheless, further consideration of the effects of socioeconomic status on the distributions will assure us that market behavior is more likely than any other mechanism, such as redistributive gift giving that might focus on particular families or social strata. In terms of the broad models advanced by Blanton and colleagues (1993:210– 217), the western lower Papaloapan basin shows specialization and likely Early Classic market distributions under the aegis of a dominant center but not precisely at the overarching center. Rather, one or more secondary centers were involved in this activity. This pattern suggests that paramount authority was not principally reliant on markets and crafts for revenues but instead on landholdings, labor, tribute, control of external economic and political arrangements—such as obsidian importation and cotton exports—or some combination. Thus, government-market links were indirect at the outset of this market system but were possibly part of a larger-scale complex economy with differential product and service emphases (Garraty, Chapter 1). A secondary center, Nopiloa, appears to have hosted some market functions involving obsidian blades. This activity may have provided indirect revenue through taxes for paramount authorities, but it may also represent a lower-level specialization undertaken to improve local economic well-being. With the breakup of the Cerro de las Mesas realm, the role of Nopiloa in blade making and marketplace exchange was enhanced. Azuzules instead emphasized orange-slipped bowl production. Centers continued to be obsidian marketing foci in the Middle and Late Postclassic periods. In the Middle Postclassic, Sauce may have encountered competition to the southeast in the Lobato area. In the Late Postclassic 125

B a r b a r a L . S t a r k a n d Al a n n a O ss a

period, Callejón del Horno appears to have served a larger area than Sauce, comparable to Nopiloa earlier. Nevertheless, a separate access zone was likely located far to the east in the mangrove area. Itinerant peddlers and down-the-line reciprocal exchanges may have moved obsidian blades even more widely than exchanges at market centers. Such activities may also be responsible for service areas that exceeded 10 kilometers. An important empirical modification of Blanton and colleagues’ (1993:210– 217) ideas is that Mesoamerican market systems developed not only in areas under major states such as Monte Albán or Teotihuacan but also in distant areas with modest-sized states such as Cerro de las Mesas. In the case of Cerro de las Mesas, however, the dominant center was unable or unwilling to host obsidian marketplace exchange but did not suppress any advantage gained at Nopiloa, a secondary center, through its blade-making industry and marketplace. In this instance state organization was not so centralized that local incentives and efforts were suppressed, offering another example of how settlement and governmental hierarchies can permit or enhance growth of specialization and markets. Because obsidian was an imported material, arriving as preformed cores, our results call attention to the economic interconnections of regions in Mesoamerica. Such connections in obsidian exchange date back centuries earlier and may have laid the groundwork for economic changes in multiple regions, some of them leading to more open access to certain products through market institutions. As argued elsewhere (B. Stark 2007b), economic changes occur for a variety of reasons, including wider social access to previously more restricted technologies; such changes may occur at greater scales than individual polities through emulation, trade, gift giving, and technological changes. The WLPB record shows neither solely bottom-up nor topdown instituted changes but instead a complex interaction of political and economic actions. Nevertheless, intimate connections between political authority and marketplace locations are evident. Acknowledgments. This study was initiated as a paper by Stark at the 72nd Annual Meeting of the Society for American Archaeology, Austin, Texas, and additional contributions were provided by Ossa in connection with her doctoral research concerning the Middle Postclassic period in the WLPB. The Proyecto Arqueológico La Mixtequilla was made possible by support from the National Science Foundation (BNS 85-19167, BNS 87-41867, and SBR-9804738), the National Geographic Society, and Arizona State University and permission from the Instituto Nacional de Antropología e Historia. Any success of the project is entirely dependent on the excellent fieldwork participation of a number of students and archaeologists over the several seasons of survey. We thank the local communities for their kind welcomes and cooperation. Christopher Garraty, John Clark, and an anonymous reviewer provided valuable suggestions to improve the chapter but are not responsible for its content.

126

Chapter Six

The Rise and Fall of Market Exchange: A Dynamic Approach to Ancient Maya Economy

Geoffrey E. Braswell

Archaeology is the study of change. The most common metaphor for change that is used to describe social process is evolution. Too often, the specific metaphor of speciation is employed. But speciation is an irreversible process; once a significant change in organization or form has occurred, a return to the previous state is virtually impossible. Another potential metaphor, drawn from thermodynamics, allows for both reversible and irreversible change. The Dynamic Model of archaic states, proposed by Joyce Marcus (1992, 1993, 1998), is an example of a model that is more thermodynamic than evolutionary in the sense that change from one political type to another is reversible. Using ethnohistorical data from Ralph Roys’s (1957) work on Mayapán, Marcus depicts Maya polities as dynamic rather than static and as growing and fragmenting in a cyclical fashion. Maya states, according to Marcus’s Dynamic Model, coalesced when a province governed by a halach uinic (“true man”) annexed formerly independent provinces and incorporated their political hierarchies into that of the core. These polities might have been highly centralized, but the degree to which expansion was manifested by true incorporation or by more general hegemonic control is unclear. As the cycle declined, provinces regained their independence or formed loosely allied confederacies. These independent provinces and confederacies were decentralized. Finally, the cycle began again as an independent province absorbed the territories and administrative hierarchies of its neighbors. Marcus stresses three important aspects of her model: (1) the state-like characteristics of small, independent 127

G e o ffre y E . B ra s w e l l

polities are a result of emulation of the political cores to which they once were linked; (2) regional provinces, rather than states, are the largest stable units of political organization; and (3) throughout most of the cycle, innovation and change are more likely to occur in peripheral provinces than in the core (Lightfoot and Martinez 1995). In this chapter I argue that economic complexity rises and falls in a cyclical manner like archaic states. Just as political systems are dynamic, so, too, are economic ones. Moreover, two critical points of change in the organization of economic systems determine their size. The less complex (or first) change point occurs when economic systems become restricted by political concerns (Blanton 1976:259–261). That is, large, decentralized, and open systems become smaller, bounded, and hierarchically controlled by elites. The more complex (or second) change point occurs when economic systems significantly outgrow the size of the polity. I argue that no form of market economy can exist below the first change point, that is, when political structure is so simple that there can be no attempt to control the means of either production or exchange. But markets, broadly defined to include both partially and highly commercialized systems, may be said to exist on both sides of the second change point. Just as Marcus defines the regional province as the largest stable unit of political organization, it is the administered market—one governed by the political concerns of medium-scale polities—that is the stable form from and to which larger, more complex competitive markets may cycle. Christopher Garraty (Chapter 1) describes our mutual project as having four goals: (1) identifying the characteristics of market systems, (2) recognizing market exchange in the archaeological record, (3) interpreting the relation between economic and political systems, and (4) understanding something about the origin and evolution of market systems. In this chapter I consider all four of these issues. I rely heavily on Carol Smith’s (1976a, 1976b) models that describe the spatial characteristics of five different types of economic systems. I employ Kenneth Hirth’s (1998) archaeological correlates for recognizing different kinds of exchange in the archaeological record but add another diacritic borrowed from Smith: the boundedness of the regional economic system. I interject comments about the relationship between economic and political cycles in my conclusion and also stress that the cyclical emergence of market systems is a developmental process more analogous to changes in thermodynamic state than to evolutionary speciation. The data I consider are derived from analyses of obsidian artifacts recovered from sites in Guatemala, Honduras, and Mexico (Figure 6.1). I analyze these data from a regional perspective but because, unfortunately, most of the information I have comes only from the centers of capital cities and other major sites, I cannot employ methodologies similar to those used by Barbara Stark and Alanna Ossa (Chapter 5) or by Leah Minc (2006).

128

6.1. Maya sites and regions discussed in this chapter.

G e o ffre y E . B ra s w e l l

Spatial and Archaeological Correlates of Market Exchange and Other Types of Distribution

Following other substantivist economic anthropologists, Carol Smith (1976a:314, 321, 334–335, 353) argues that there are three fundamental types of exchange: dyadic, polyadic, and market exchange. Dyadic exchange is direct trade between two equalstatus individuals. The long-distance trade of the Kula ring and the gift giving of jade between Classic Maya kings are examples of dyadic exchange. Polyadic exchange may or may not be direct and takes place between a high-status individual and one or more subordinates. The redistribution of blankets by chiefs in the Pacific Northwest is a classic example of polyadic exchange. The spatial organization of economic systems based on dyadic exchange is open and resembles a large, decentralized network connecting more or less equal nodes (Figure 6.2a). In contrast, distribution systems in which polyadic exchange is dominant are bounded, relatively small in size, and hierarchically ordered around the central hub of the chief and the chiefly village (Figure 6.2b). Market exchange is more complex, and the relationship between producer and consumer is most often mediated by merchants or other middlemen. Carol Smith (1976b) has identified three distinct kinds of market exchange. The first, administered market exchange, implies the control of commerce by political concerns. In contrast, monopolistic market exchange—a concept I do not employ in this chapter—entails the domination of political concerns by commerce. The difference between administered and monopolistic market exchange lies in where and how the elite regulate the circulation of goods and extract surplus and where market forces govern the value of a commodity.1 Because both market forces and elite manipulation determine value, such systems are only partially commercialized. In administered market exchange, rural producers compete to supply a relatively small class of middlemen. In other words, market forces determine wholesale value. Elite administrators control the economy by regulating middlemen, such as merchants or artisans, rather than the large rural population. Surplus is extracted through the exertion of political control proscribing who, when, where, what, and how much trade takes place. Thus, retailing is the focus of elite control. In monopolistic exchange, on the other hand, the relationship between rural producers and middlemen is regulated by the elite, and retailing follows market principles. Colonialist extractive economies are classic examples of monopolistic market exchange. Administered market systems are bounded and defined by a rigid pyramidal hierarchy of nodes and hubs because elites maintain control by manipulating retail value (Figure 6.2c). Because such economies have a single price-setting hub at their political center, Carol Smith calls them solar central-place systems. In contrast, monopolistic market systems may be much larger and are focused on a dominant hub outside the regional system (Figure 6.2d). Smith calls such colonial organizations dendritic central-place systems because of their structure. An important aspect of these systems is that, from the perspective of the periphery, a region may be open to long-distance exchange yet closed to interregional interaction. Consider in modern 130

6.2. Carol Smith’s (1976b) ideal spatial models of different kinds of exchange: (a) extended network, (b) bounded network, (c) solar central-place system, (d) dendritic central-place system, (e) two forms of interlocking central-place systems.

G e o ffre y E . B ra s w e l l

Table 6.1. Kenneth G. Hirth’s (1998) Archaeological Correlates for Different Types of Exchange. Dyadic Exchange Polyadic Exchange

(1) Geological sources represented at a given household mirror sources at production locus. (2) Households show different procurement patterns if they obtain material from different production loci.



(1) Elite households have greatest quantity of obsidian and perhaps diversity of sources. (2) “Trickle down” of quantity and source diversity is related to social hierarchy.

Market Exchange

(1) Quantity at households is related to need rather than status. (2) Market homogenizes sources are represented at the community level.

Belize the presence of both Land Rovers and Guinness stout (produced or licensed by companies based within the historical bounds of the British empire) but the near absence of Volkswagen sedans (made in Mexico) and Gallo beer (bottled in nearby Guatemala) as examples of openness to some long-distance exchange yet boundedness to interregional interaction. The final type is competitive market exchange, where market forces of supply and demand determine both the wholesale and retail value of goods. Such systems are open and have a complex interlocking structure of nodes and hyper-connected hubs; the interlocking structure promotes market rather than political forces as the principal value-setting mechanisms on both the wholesale and retail levels. Such fully commercialized systems are open and may encompass vast territories (Figure 6.2e). Hirth (1998) and, more recently, Minc (2006) have provided archaeological correlates to different kinds of exchange. To Hirth’s correlates I add Carol Smith’s observations about open and bounded systems (Table 6.1). Distribution systems characterized by dyadic exchange are open; those in which polyadic redistribution is the determining practice are closed. Systems in which administered market exchange is the norm are also closed and have boundaries that approximate those of the polity, while monopolistic market distribution creates systems that are open to long-distance interaction but not necessarily to interregional exchange. Finally, interlocking distribution systems characterized by fully commercialized market exchange are open and very large. I turn now to obsidian exchange in the Maya region and look for the archaeological signatures of each type of distribution system. My argument draws upon research conducted at Preclassic-period (ca. 1000/800 BC–AD 150) sites in the highlands of Guatemala, Classic-period (ca. AD 200–800) sites in the central Maya lowlands and southeastern periphery, and Terminal Classic to Postclassic (ca. AD 800–1520) sites in the northern Maya lowlands (Figure 6.1). Preclassic Obsidian Distribution Systems

The first region considered is the central highlands of Guatemala, an area with two important obsidian sources (El Chayal and San Martín Jilotepeque) and a third 132

T h e R i s e a n d F a l l o f M ar k et E x c h a n ge

minor source (San Bartolomé Milpas Altas). Since the late 1980s, Eugenia Robinson (1990, 1993, 1994, 1998; Robinson et al. 2002) and I (Braswell 1993, 1996a, 1996b, 1998, 2002; Braswell and Robinson in press) have conducted various survey and excavation projects in the eastern Kaqchikel region, west of modern Guatemala City (Figure 6.3). I call the region after the Mayan language spoken in the area by modern inhabitants. Robinson and I have surveyed a total area of about 400 km2 and located approximately 600 sites. During both the Middle and Late Preclassic periods, the settlement hierarchy of this vast region consisted of just one level: the small agricultural hamlet or village. There is no evidence of social stratification; all structures were made of perishable materials, and we have yet to find any evidence of unequal access to goods. Exchange within the eastern Kaqchikel region was dyadic and organized as an unbounded extended network. In fact, during the early Middle Preclassic period the quantity of obsidian from both the El Chayal and San Martín Jilotepeque sources found at sites in the eastern Kaqchikel region is directly proportional to the distance to each of those sources, a falloff pattern consistent with unbounded exchange (Renfrew 1977:72). Analysis of obsidian artifacts collected at Kaminaljuyú-Miraflores II by Marion Popenoe de Hatch and Juan Antonio Valdés reveals a somewhat different pattern (Amador and Braswell 1999). The Preclassic center of Kaminaljuyú is located between the Kaqchikel region and the El Chayal obsidian source. During the second half of the Middle Preclassic period, the relative proportion of obsidian from the El Chayal source that reached the eastern Kaqchikel region decreased. Moreover, the relative proportion of San Martín Jilotepeque obsidian at Kaminaljuyú also dramatically decreased. Similarly, ceramics in these two adjoining regions began to diverge; in particular, many new and different pottery types appeared at Kaminaljuyú (Popenoe de Hatch 1997). The boundary between the eastern Kaqchikel region and the Kaminaljuyú distribution system that emerged around 600 BC is sharp and well defined, corresponding more or less to Cerro Alux and the pass between the Valley of Guatemala and modern San Lucas Sacatepéquez. In short, a Kaminaljuyú-centric distribution system with a firm western boundary developed in the second half of the Middle Preclassic, at a time when a chiefdom emerged at the site. In contrast, a very simple settlement hierarchy and extended distribution system persisted in the Kaqchikel highlands until, as I have argued, the beginning of the Early Classic period (Braswell 1996a). A reasonable interpretation is that during the Middle Preclassic period, a bounded network system of the sort in which polyadic exchange is dominant emerged at Kaminaljuyú. This is why little El Chayal obsidian circulated in the Maya region before the Late Preclassic; the economy of late Middle Classic Kaminaljuyú was closed. In contrast, the open network system of the Preclassic eastern Kaqchikel region was predicated on dyadic exchange. For this reason, and even though the economy of the eastern Kaqchikel region was less complex than that of Middle Preclassic Kaminaljuyú, obsidian from San Martín Jilotepeque was circulated (albeit in relatively small amounts) throughout the Maya area. But at this early time 133

6.3. Middle Preclassic sites of the eastern Kaqchikel highlands: triangles are Type I sites (sherd and lithic scatters lacking visible mound architecture), medium gray indicates natural obsidian outcrops, light gray indicates modern cabeceras (named in bold roman type), contour interval is 500 m.

T h e R i s e a n d F a l l o f M ar k et E x c h a n ge

I see no evidence for any sort of market exchange in either the Kaminaljuyú polity or the eastern Kaqchikel region. The development of a bounded network system at Kaminaljuyú in the second half of the Middle Preclassic period was preceded by earlier developments in Soconusco, a region that imported much of its obsidian from the central highlands of Guatemala. For Paso de la Amada, Richard Lesure (1995) and Lesure and Michael Blake (2002) have argued that there is little evidence of differential access to imported obsidian during the Early Preclassic period ( John E. Clark, in his review of this chapter, strongly disagrees with their position). For Lesure and Blake, then, the relationship between political power and economic advantage was limited until the beginning of the Middle Formative, when Soconusco experienced a great reorganization in settlement patterns and political organization. A hypothesis for further testing is that the emergence of a bounded network system at Kaminaljuyú (but not in the eastern Kaqchikel region) was stimulated by interaction with complex Middle Preclassic chiefdoms in Soconusco, such as La Blanca. Obsidian Distribution in the Central Lowlands and the Southeastern Periphery during the Classic Period

Calakmul and Tikal (Figure 6.1) were certainly the most politically influential states of the central Maya lowlands during the Classic period. Hattula Moholy-Nagy (1994, 1997, 2003) has described the excavation of literally millions of obsidian artifacts at Tikal by the University of Pennsylvania project. Edwin Shook (personal communication, 1994), director of that project, once told me “there was so much obsidian and we had no idea it would ever tell us anything, so we only collected it from important primary contexts and discarded all the rest.” In contrast to Tikal, two recent projects at Calakmul excavated only hundreds of artifacts (Braswell et al. 2004). Just 515 obsidian artifacts were excavated and recovered by the Universidad Autónoma de Campeche Proyecto Calakmul, directed by William Folan. Moreover, from 1993 to 1995, the Instituto Nacional de Antropología e Historia project directed by Ramón Carrasco recovered only 126 obsidian artifacts. In fact, much more jade than obsidian has been found at Calakmul. There are certainly differences in the intensity of archaeological investigations at Calakmul and Tikal, as well as in methods of collection and in the contexts chosen for excavation. Moreover, Calakmul is 90 km more distant from the Guatemalan obsidian sources than is Tikal. Nonetheless, these factors cannot account for the discrepancy of three to four orders of magnitude between the amount of obsidian found at Tikal compared with Calakmul. It seems highly likely, then, that the Tikal polity prevented obsidian from going to Calakmul. This suggests a regionally bounded economy in which obsidian was subject to redistributive or administered market distribution but not to fully commercialized exchange. Kazuo Aoyama (1999) and I have independently analyzed obsidian artifacts excavated from the Copán kingdom of western Honduras (Figure 6.1). My own work 135

G e o ffre y E . B ra s w e l l

has consisted of the analysis of approximately 28,515 obsidian artifacts from the royal residential group of the last Copán ruler of the Classic period and 2,526 pieces from excavations conducted in an adjacent portion of the site called El Bosque. These excavations were conducted at or just south of the site epicenter, so they tell us very little about obsidian distribution and consumption by non-elites who lived outside the palace complex. One important observation made from these data, however, is that during the Classic period, at least some obsidian was not subject to commercialized exchange. Exotic green obsidian from Central Mexico entered the Copán royal acropolis (and the kingdom as well) in the form of finished composite artifacts. Moreover, during the Classic period, green obsidian was used only in apical elite contexts. That is, obsidian tools and ornaments from the distant Pachuca source were received by royalty and were not even redistributed to lesser members of the nobility. Aoyama’s (1999) more extensive regional analysis supports this observation but, more important, notes a very sharp economic boundary in the northern extremity of the La Entrada region between sites with access to Ixtepeque obsidian (the principal source used at Copán) and sites that received most material from Honduran sources. In other words, the Copán regional economy—like that of Tikal—was tightly bounded to the north. Moreover, Aoyama argues that within the Copán system, access to obsidian and especially to prismatic blade technology was managed by the elite. My investigations at the site core support this notion. In contrast, Aoyama observes that rural households had less access to obsidian than did urban elite households and that rural dwellers also tended to use ad hoc flake tools rather than blades. Aoyama does not employ Carol Smith’s economic typology, but his detailed analysis depicts the Classic-period Copán economy as sharing some qualities with both bounded networks and solar central-place systems. We might expect, then, that Classic Maya polities either administered or redistributed obsidian or perhaps both. It is clear that in Classic-period Copán, elites rather than the forces of supply and demand monitored the value of and (to a certain extent) access to both obsidian and blade technology. The Terminal Classic and Postclassic Northern Maya Lowlands

Several publications have presented obsidian data from a host of sites in the northern Maya lowlands that date to the Terminal Classic and other periods (Figure 6.1; Braswell 1997, 2003; Braswell and Glascock 2002, 2007; Nelson 1985). Using Hirth’s (1998) two criteria of procurement homogenization and need-based access, Michael Glascock and I have argued that within the Itzá polity, obsidian was subject to two sorts of market exchange during distinct periods (Braswell and Glascock 2002). In the ninth century, the economic system of the Itzá polity was tightly bounded. Fully 71 percent of the obsidian found at Chichén Itzá and 80 percent of that collected at its port of Isla Cerritos came from Central Mexican sources. In contrast, neighboring 136

T h e R i s e a n d F a l l o f M ar k et E x c h a n ge

centers such as Ek Balam and Cobá received more than 90 percent of their obsidian from Guatemalan sources, especially El Chayal. Nowhere is this boundary so striking as at Yaxuná, only 19 km southeast of Chichén Itzá. There, 84 percent of the recovered obsidian came from El Chayal, and all the Mexican-source artifacts were found in contexts associated with the destruction of Yaxuná by Chichén Itzá. The sharply bounded nature of the ninth-century Itzá obsidian distribution system and evidence for some sort of market exchange together imply the existence of a solar central-place system and an administered market economy at Chichén Itzá during the ninth century (Braswell and Glascock 2002). Exciting data from the Puuc region strongly suggest that sometime around AD 900, Uxmal and related sites began to participate with Chichén Itzá in an open, interlocking central-place system. Extensive excavations directed by José Huchím and Alfredo Barrera (Huchím Herrera and García Ayala 2000; Kowalski et al. 1996) within the center of Uxmal and at some distance from the city revealed that contexts dating to the ninth century have very little exotic Mexican obsidian. In contrast, during the tenth and early eleventh centuries there was a three-fold increase in the relative abundance of obsidian from central Mexican sources. Moreover, the same homogenized market “mix” of Central Mexican obsidian seen at Chichén Itzá is also represented at Uxmal, Xkipché, Labná, and at many other tenth-century Maya sites where Mexican obsidian is found. In descending order of quantity, this homogenized mix of Mexican obsidian comes from Ucareo (42 percent), Pachuca (31 percent), Zaragoza (10 percent), Paredón (9 percent), Pico de Orizaba (6 percent), Zacualtipán (1 percent), and Otumba (1 percent). Thus, distribution patterns suggest the breakdown of partially commercialized, regional, and bounded distribution systems and the emergence of a fully commercialized, interregional, and open market economy by about AD 900 (Braswell and Glascock 2002). This interlocking central-place system, however, collapsed with the decline of Chichén Itzá and the Puuc region during the eleventh century. The rise of Mayapán at the beginning of the Middle Postclassic period saw the reorganization of obsidian procurement. Obsidian from the Ixtepeque, Guatemala, source replaced both El Chayal and the Mexican sources in importance. Bárbara Escamilla Ojeda (2004) analyzed more than 14,000 obsidian artifacts collected by Carlos Peraza Lope’s project in the central precinct of Mayapán. The Mayapán collection, in fact, contains more obsidian artifacts than have been recovered in recent decades from all other sites in the northern lowlands combined. Nevertheless, Tatiana Proskouriakoff (1962) and Clifford Brown (1999) have both noted that in the residential areas of Mayapán, obsidian is rather scarce. This marked difference between the great quantity of obsidian found in the elite epicenter and the scarcity of obsidian in more humble residential zones of the city suggests that during the Middle Postclassic period there were considerable class-based differences in access to this important resource. I interpret this as a return to a simpler distribution pattern based on either administered market behavior or—perhaps more likely—polyadic exchange. 137

G e o ffre y E . B ra s w e l l

Conclusions Marcus (1992, 1993) has proposed a dynamic and processual model of political complexity for the Maya region. The model is not strictly evolutionary because, unlike biological speciation, the change from one sort of political organization to another is reversible. A graph of her model contains many local peaks and valleys corresponding to oscillating cycles between complex chiefdoms and simple states and also between simple regional states and more large-scale and complex polities (Figure 6.4a). Throughout most of the Classic period (and probably from the Middle Preclassic onward), the size of economic systems, as measured by obsidian distribution patterns, was often closely related to the size of polities. This was true for periods when polities were organized as complex chiefdoms or as simple archaic states, that is, when Maya economies were structured either as bounded networks or administered markets. It was less true during the second half of the Terminal Classic period when a competitive market integrated the northern Maya lowlands with Central Mexico. A graph of the complexity of exchange systems can also be drawn. If we superimpose this second typological curve on top of Marcus’s polity size model (Figure 6.4b), two things are immediately apparent. First, the period of economic cycles is much greater than that of political cycles. There are far fewer valleys and peaks. Longer periodicity implies greater stability. Although both trade routes and the value of goods changed dramatically over time, the ways value was determined and exchange took place were far less subject to change than was the average size of polities. Second, the economic (type) and political (size) cycles do not closely correspond. This is what Christopher Garraty (Chapter 1) means in stating that political and market systems are at best “co-evolutionary.” The lack of close correspondence between these two cycles implies that in the Maya region dramatic political change was not always causally linked with significant change in the nature of exchange. This distinction appears to be true even during periods when polyadic or administered market exchange was the norm, that is, when political concerns determined the value of obsidian. Even the Classic “collapse” did not cause Maya economies to revert to simpler forms than the administered market. Nevertheless, the fragmentation of the Itzá and Puuc kingdoms in the eleventh century may have led to the end of Maya participation in a pan-Mesoamerican, highly commercialized economy. Evidence from Mayapán—specifically, the limited distribution of obsidian beyond the epicenter of the site—suggests that the breakdown of this regional collapse could have led to a system based on either administered market exchange or redistribution. What these two systems share is their bounded and centralized nature. Like the provinces of Marcus’s political model, bounded and centralized economic systems (typified by either an administered market or polyadic exchange) constitute the largest stable economic unit. Such bounded and centralized systems are the breakdown products of more complex market systems. Karl Polanyi (1957:248, 250) once wrote that for premodern societies, economy is “embedded” in other forms of social behavior, including political relationships. 138

T h e R i s e a n d F a l l o f M ar k et E x c h a n ge

6.4. Economy and average polity size in the Maya area: (a) comparison of the areal extent of economic and political systems, (b) comparison of economic type and polity size (polity size data from Marcus 1993:figure 26).

Nevertheless, the general lack of correspondence between Maya political and economic cycles, except during the late Terminal Classic when both complex states and a fully commercialized economy (sensu C. Smith) coexisted, seems to suggest the opposite. Contrary to Polanyi, the interdependence of economic and political systems becomes greater rather than less as complexity increases. 139

G e o ffre y E . B ra s w e l l

Acknowledgments. I thank the National Science Foundation Archaeometry Program (SBR-9802366, grant to Michael D. Glascock and Hector Neff ) and the Foundation for the Advancement of Mesoamerican Studies, Inc. (Gr. 95004) for their financial support. I am very thankful and owe much to my colleagues and all the directors of the archaeological projects that contributed to this work, especially Michael D. Glascock, William Swezey, Eugenia Robinson, Marion Popenoe de Hatch, Juan Antonio Valdés, William Folan, Ramón Carrasco, E. Wyllys Andrews V, Kam Manahan, Peter Schmidt, Rafael Cobos, José Huchím, Alfredo Barrera Rubio, Tomas Gallareta Negrón, David Freidel, Hanns Prem, Iken Paap, Carlos Peraza Lope, Bárbara Escamilla Ojeda, and all my friends from the Proyecto Chichén Itzá. Notes 1. In my view, price is a concept that exists only in economic systems that employ money or some other standardized and quantified measure of cost. Here, I use “value” to mean the cost of obsidian, probably exchanged in the more qualified form of goods, services, or less tangible obligations.

140

Chapter Seven

Housing the Market: Swahili Merchants and Regional Marketing on the East African Coast, Seventh to Sixteenth Centuries AD Jeffrey B. Fleisher

Ancient Swahili towns on the eastern African coast are best known for their dedication to long-distance exchange within the Indian Ocean trade system. Middlemen traders in these towns are commonly cast as savvy, entrepreneurial brokers, negotiating trade between hinterland areas and overseas merchants as well as overseeing emerging cosmopolitan city-states. Often implied but rarely demonstrated archaeologically, these merchants participated in a vast market network that stretched thousands of miles along the coastal corridor, 100 miles to the interior, and thousands of miles across the Indian Ocean to ports along the Red Sea, the Persian Gulf, and the western coast of India. In this chapter I discuss what we know about market exchange along the Swahili coast based on archaeological and historical sources. I follow this overview with a discussion of an apparent contradiction in market development in Swahili towns on Pemba Island, Tanzania: the increasing privatization of exchange relations between Swahili and overseas merchants and the growing presence of overseas trade goods in the assemblages of non-elite towns and villages. This pattern offers a contrast to normative models of how elite goods moved within Swahili town regions, where prestigious goods are often understood to have moved in restricted spheres only. With archaeological data from Pemba Island I explore whether the increasing prevalence of overseas goods is suggestive of a regional economy with market exchange and how such an economy might be affected by greater restrictions on external commerce by local merchants. This process of privatization of external commercialization—in 141

J e f f re y B . F l e i sher

which market exchanges move into the private homes of town merchants—coupled with evidence of market exchanges in the local region provides a unique example of how a market system can thrive through highly “embedded transactions” (Granovetter 1985; Garraty, Chapter 1). Rather than overt political institutions becoming involved in regulating markets, the Swahili case shows how a shared religious tradition as well as highly personalized relationships between merchants aided the emergence of market transactions. The Ancient Swahili: A Brief Introduction

The ancient Swahili inhabited a 3,500-km stretch of the eastern African coast (Figure 7.1), extending from southern Somalia in the north to Mozambique, the Comoros, and northern Madagascar in the south. Although it is difficult to determine the precise period in which a coherent group of coastal peoples emerged, the nine centuries from AD 600 to 1500 witnessed dramatic growth in the form of dozens of new coastal towns and villages. During the eleventh and twelfth centuries AD, a number of settlements began to distinguish themselves from their rural neighbors by constructing more permanent monumental architecture. The largest of these settlements—commonly known as stonetowns because of the presence of coral rag mosques, tombs, and elite houses—participated directly in the trade of long-distance goods and likely owed much of their expansive development to that trade. These were the towns the Portuguese encountered—and were impressed by—when they entered the Indian Ocean world in the late fifteenth century (Horton and Middleton 2000; Kusimba 1999; Pearson 1998). Although the most explosive growth occurred from the seventh century onward, the coast was settled much earlier, and some of these settlements were involved in long-distance exchanges. The Periplus Maris Erythraei (Casson 1989), a first-century AD Greek description of coastal settlements, describes the trade opportunities available at coastal sites, where imported metal and glass were exchanged for coastal and hinterland products such as ivory, rhinoceros horn, and coconut oil. Unfortunately, no archaeological sites mentioned in the Periplus have been located, and finds from this period are extremely sparse (Chami 1999). In fact, little archaeological evidence exists for settlements that predate the seventh century, although the presence of Roman pottery ( Juma 1996) and other imported fourth- through sixth-century artifacts suggests that modest trade continued in the centuries that followed (Smith and Wright 1988). A textual hiatus extends from the Periplus of the first century to a series of Arab geographies that date to the tenth century and later. The sparse archaeological evidence of the first half of the first millennium gives way to a rich and increasingly well-understood set of coastal and near-hinterland settlements that date from the seventh and eighth centuries and later. Archaeologists have found dozens of archaeological sites (of what had probably been hundreds) that contained a distinctive local pottery tradition, the TIW/Tana Tradition (Chami 142

7.1. Eastern African coast.

J e f f re y B . F l e i sher

1994; Horton 1996), alongside imported glass, copper, and ceramics. Unlike the disparate finds of earlier periods, these sites suggest intense and sustained contact along the entire coastal corridor as well as regular and significant exchange with overseas merchants, now focused on those from the Persian Gulf and India. These early Tana Tradition settlements were built mostly of impermanent earth and thatch structures; some of the sites continued to develop into stonetowns over the course of a few centuries in the second millennium (Chittick 1974; Horton 1996). A few of the largest of these settlements, however, were abandoned during the tenth century ( Juma 2004; LaViolette, Fleisher, and Mapunda 2004). Based on tenth- and eleventh-century Arab geographies, we know that the products most sought from eastern Africa were ivory, gold, rhinoceros horns, and leopard skins. Mangrove poles and finished goods such as iron were likely exported as well (Chami 1992, 1994). Goods coming into the coast were varied but included cotton cloth and glass beads from India as well as glass bottles and beakers, copper bowls and jewelry, and imported glazed and unglazed ceramic bowls and jars from the Persian Gulf and China. Slavery became an important part of the trade during the eighteenth and nineteenth centuries, but there are indications that enslaved people were also traded out from the coast during these earlier centuries (B. Lewis 1974). Muslim merchants’ ability to move rapidly from port to port during the course of a year facilitated long-distance exchange among the eastern African coast, the Persian Gulf, and the western coast of India. Although widely dispersed, these parts of the Indian Ocean rim are well connected by a yearly pattern of northeast/southwest monsoonal winds. These winds allowed merchants to sail from India and the Persian Gulf to eastern Africa from November through March (with the northeast monsoon) and then, when the winds shifted in April, to return to their home ports sustained by the southwest winds. Swahili towns were an important nexus for overseas merchants: they collected desired goods from interior locations and more southerly coastal ports and also provided an important location to provision their ships and a safe location to wait for the crucial monsoonal shift that would carry them back home. Swahili towns thus provided an important economic link—a “hinge” (Pearson 1998:105)—between India/ the Persian Gulf and the African hinterlands, where desired goods were available. The growth of overseas trade and traffic to coastal ports created conditions for the emergence of sociopolitical and economic inequality, with some individuals controlling access to overseas goods and thus obtaining greater prestige locally. Henry Wright (1993:671–672) has argued that the introduction of imported goods may have been used by certain individuals to reinforce elite status at coastal villages and that the rise of hierarchies was the result of interaction and competition among coastal groups. Along the coastal corridor, the emergence of hierarchical political organizations took two general forms: patrician towns and towns with kings (Allen 1993; Horton and Middleton 2000). Patrician towns are often viewed as relatively benign oligarchies, where a series of powerful clans controlled access to, and relations within, the town limits to outsiders. Many patrician towns had surrounding walls that provided a sym144

H o u s i n g t he M ar k e t

bolic barrier between the town’s Muslim inhabitants and the non-Muslim peoples of the mainland coast. The power of patrician groups extended generally to the town walls, although there were close commercial, political, and social links between them and surrounding groups often based on patron-client relationships (Horton and Middleton 2000:163–164). Powerful individuals who composed the oligarchy were members of prominent houses (Donley-Reid 1982, 1990a; Fleisher and LaViolette 2007) that were active in trade and the town’s religious life. These towns did not have a hereditary leader (although heredity played an important role in the continuation of a house) but instead selected leaders from among the prominent clans and families, which served as a primus inter pares—first among equals. This political organization was fairly prominent in the northern towns along the Kenyan coast, best known from ethnographic accounts of towns in the Lamu archipelago (Horton and Middleton 2000; Middleton 1992; A. Prins 1971). Towns with kings were much more autocratic; a hereditary king or sultan controlled the political realm. Although some contentious relationships existed between merchants and sultans in these towns (Fleisher 2004), the sultan was often an active participant in merchant activities. This is well illustrated architecturally by the storage and putative trade areas connected to Husuni Kubwa, the sprawling residence of the fourteenth-century sultan of Kilwa Kisiwani in southern Tanzania. We know from local chronicles that the sultan was not the sole merchant, however, and others in the town were active in long-distance exchanges. As explained in the discussion that follows, Kilwa is somewhat exceptional in that it was one of the most powerful coastal towns; because of that position, the sultan was able to levy taxes on many goods moving up and down the southern coast. In particular, Kilwa was likely an important hub in the gold trade, serving as a collection point for gold coming out of the Zimbabwean plateau, moving to the coast along various river courses, and then being housed at Kilwa. Gold and other products not housed and traded at Kilwa were taxed (or, at the least, taxation was attempted by the sultan of Kilwa). In the dozens of other small towns, however, there was probably little distinction between town leaders—the “kings”—and merchants. In these more modest “towns with kings,” leading families likely controlled both politics and long-distance exchanges. One general way of assessing the type of political organization in particular towns is to look at the distribution of more permanent stone architecture. Residents of towns on the northern Kenya coast, the best-known patrician towns, lived cheek by jowl in stone houses, forming a dense architectural center. In “towns with kings,” elite-owned stone houses were much rarer, set within a majority of wattle-and-daub houses. Some of these latter towns, known to be fairly prominent (such as Ras Mkumbuu on Pemba Island), had only a handful of stone houses within a much larger settlement. In both of these political organizations there appear to have been few divisions between merchants and political leaders; in most towns they were one and the same. Emerging political leaders were most likely those who achieved success in long­distance exchanges. Therefore, the emergence of a coastal economy with market 145

J e f f re y B . F l e i sher

exchange would seem to support what Christopher Garraty (Chapter 1) and Richard Blanton and Lane Fargher (Chapter 10) call “bottom-up process” of market development, “initiated by aspiring market traders and producers” (Garraty, Chapter 1). There were times when merchants and town leaders came into conflict, however, especially when merchant capital was seen as threatening the legitimacy of political rule. One such example comes from a chronicle of the town of Kilwa that describes the circumstances surrounding the rebuilding of the Great Mosque, which had collapsed during a period of poverty: The prime mover [to rebuild the mosque] was Sayyid Hajj Rush ibn Sultan Husain. . . . [H]e asked Sultan Sulaiman permission to rebuild it at his own expense. Permission was not given, but the sovereign gave him 1,000 mithkals of gold and said: Rebuild the mosque with this money. And Sayyid Hajj Rush meditated the matter and said to himself: Unless I take this money permission to rebuild the mosque will be refused. Thus it is best to accept the money, but I shall rebuild it at my own expense. So he took the money, and rebuilt it at his own expense until it was complete. When Sultan Sulaiman died, Hajj Rush returned the money to his heirs. (Freeman-Grenville 1962:40)

This story illustrates the ability of wealthy merchants to challenge the authority of the Sultanate through their control of wealth and the sultan’s acknowledgment of those challenges. What also connected these relatively different types of towns was a commitment to overseas exchange and, by the twelfth century AD, to Islamic practice. Tantalizing evidence of the earliest dates of Islam (Horton 1996) suggest that some coastal towns had modest-sized mosques by the eighth century. However, it was not until the eleventh and twelfth centuries that Islam became the majority religion at many coastal port towns. A richly elaborated, local form of mosque architecture present in most prominent coastal towns attests to the common threads of Islam that connected towns with each other. Surely the influence of Islamic practice, Koranic teaching, and visitation by Muslim merchants from across the Indian Ocean affected the nature of merchant activities in coastal towns. In fact, Wright (1993:671–672) sees Islam as a crucial factor in the development of Swahili towns: “Islamic institutions would have been made central in community life only in order to stabilize relations among the emerging elites of the larger communities, and subsequently [been] extended to bind villages and smaller centers more closely to emerging towns.” James Allen (1993) suggests that town leaders used Islam as a way to establish rank within society, through the public performance of Islamic rituals as well as the purity and morality that accrued to them. The adoption of (and conversion to) Islam not only bound villages to centers and provided leaders with moral authority but also connected coastal merchants to overseas Muslim traders. With conversion and exposure to Muslim traders came rules and regulations regarding the exchange of goods and the morality of the market. This resonates with what Garraty (Chapter 1) refers to as a “moral source of oversight” in the regulation of market behavior (see also Abbott, Chapter 3). Although no tex146

H o u s i n g t he M ar k e t

tual sources or archaeological evidence informs us about the ways Swahili merchants drew upon Islamic rules of business and exchange, they surely shaped the way market exchanges were carried out in coastal towns. Rhetoric of the Market

The archaeology of the Swahili coast has not been immune to what Richard Wilk (1998:469; see also M. E. Smith 2004:92) has called the “polarizing rhetoric about the role of the market” in pre-capitalist economies. As Wilk noted for Mesoamerican literature, “[T]he tendency has been to make too much of markets—seeing the entire economy as dominated by the institution and capable of being analyzed by formal economic methods—or to make too little of them—seeing the entire Mesoamerican past as dominated by household production and gift-giving, with only small amounts of elite trade and part-time specialization” (1998:469). In the Swahili case, archaeologists have tended to make too much of the market, leaving the impression that the lives of coastal Swahili people were dominated completely by trade and market interests. This includes the assumption that certain types of archaeological evidence, such as coinage, indicate a suite of commercial institutions that may or may not have been present in pre–sixteenth-century towns (marketplaces, banking, credit; but see Killick 2009). To be clear, I am not questioning the importance of merchant activities in the development of Swahili society, and I believe external exchanges in Swahili towns were governed by something approximating market principles, albeit in unique ways. However, few attempts have been made to use archaeological data to support claims that the ancient Swahili practiced extensive market exchange or had advanced commercial institutions, and none have explored whether market exchange governed internal exchanges. Perhaps the history of archaeological research on the Swahili is somewhat to blame for what I think is an over-interpretation of the prevalence of market forces in ancient Swahili society. Research and writing on the ancient Swahili since the early 1980s has sought to establish them as one of the premier and autochthonous complex societies in sub-Saharan Africa. Although I do not argue with this paradigmatic reorientation (my own research has been part of this sea change), I believe some current syntheses have gone too far in their interpretation of the types of market institutions that were present throughout the Swahili world before the sixteenth century. One reason for this is the conflation of historical and ancient periods; much of our image of the Swahili is taken from descriptions of them from the sixteenth century onward, when the Swahili economy was significantly altered through the intrusion of Portuguese and, later, Omani and British forces (Pearson 1998). In an effort to draw out the deep origins of the merchant Swahili, most overviews uncritically move back and forth between the pre– and post–sixteenth-century world (Horton and Middleton 2000; Kusimba 1999), using later documentary examples of commercial institutions to explain earlier economic transformations. Although 147

J e f f re y B . F l e i sher

there may be much to be learned by looking at nineteenth- and twentieth-century historical and ethnographic sources on the Swahili, the tendency has been to read these sources into the past with little archaeological support. The result, I argue, is to invest Swahili economic forces in the pre-Portuguese period with an anachronistic level of complexity. In addition, some scholars tend to extrapolate from single exceptional cases to all coastal towns. For example, the town of Kilwa Kisiwani (Chittick 1974) is often held up as a classic example of a Swahili town (Nurse and Spear 1985). However, Kilwa was one of the most powerful and exceptional towns along the coast. From AD 1100 to 1500, the larger Swahili world contained dozens of independent and variably organized polities. Kilwa was but one of these, and its unique history and ability to control the transit of gold during the twelfth through fourteenth centuries make it ill-suited as a standard bearer for other coastal towns. My own research at a midsized Swahili polity on Pemba Island, Tanzania, has sought to redress this imbalance by investigating the political economy of a more modest town and region, perhaps closer to the coastal norm than were sites like Kilwa or Mombasa. Yet I am careful not to reinstate a normative view of coastal market exchange. In the discussion that follows, data from Pemba challenge some long-standing notions about the nature of market exchange; these data are not meant to build a unitary notion of coastal exchange patterns, however, but rather to begin to expose some of the variability in coastal market relations. What We Know about Swahili Market Exchange

Archaeological evidence of ancient markets is notoriously elusive, and in many cases market behavior is inferred rather than demonstrated with archaeological evidence (Hirth 1998:453). Unlike Mesoamerica, where ample documentary sources attest to large and thriving public marketplaces at the time of Spanish conquest, in the Swahili case there are few, if any, direct references to public marketplaces—or market activity—in documents that refer to the coast. Although most documents written by external observers describe the goods obtainable in particular towns, few describe actual marketplaces with goods for public sale or how one went about acquiring them. In fact, the only reference that specifically refers to a marketplace is from the first-century AD Periplus, which, as discussed previously, does not date to the period of Swahili towns. The Periplus refers to the town of Rhapta as a “market-town” (FreemanGrenville 1962:1–2). To date, neither Rhapta nor any towns dating from this period have been located or excavated along the coast (but see Chami 1999). Of the many other references to coastal port towns, most list the many products available or consumed in the town but never refer to any public marketplaces, customs houses, or even storehouses. Even though we know about the specialized architecture at Kilwa, especially Husuni Kubwa and Husuni Ndogo (possibly a storage facility), travelers to Kilwa never mention these structures or the activities that occurred within them. This silence is especially surprising for eyewitness accounts by Arab travelers 148

H o u s i n g t he M ar k e t

and merchants, who were accustomed to these features in other trade towns along the Indian Ocean rim. For example, in Aden, a port at the mouth of the Red Sea that was one of the principle Arabian entrepôts for the western Indian Ocean, documents describe specific twelfth- to thirteenth-century locales where custom duties were paid, wholesale transactions were carried out, and retail shops were located. As Roxani Margariti (2007:98–99) described: “Aden’s waterfront thus emerges from the sources as a nucleus of commercial structures. At the center stood the triad of the customs house and two wholesale markets. The . . . smaller markets, and single shops may have stood farther back, integrated within the downtown fabric of the port.” Another interesting document is one by Ibn Battuta, the famed fourteenth-­century Moroccan traveler who visited eastern Africa in 1331. Battuta traveled from Zeila on the Red Sea down the eastern African coast, stopping at Mogadishu, Mombasa, and Kilwa Kisiwani—three of the most prominent Swahili towns of that century. Battuta described only Zeila as “a large town with an important market” (Freeman-Grenville 1962:27). His description of eastern African towns is much more flattering: he called Kilwa “one of the most beautiful and well-constructed towns in the world” (FreemanGrenville 1962:31; Zeila, however, was “one of the dirtiest towns in existence, vile and evil-smelling”). However, he made no mention of marketplaces or any other structures or spaces related to market exchange. Elusive Swahili Markets

There are a number of reasons why markets and marketplaces might be elusive in both documentary sources and archaeological research. First, it is possible that market meetings occurred seasonally, which Mark Horton (1996) has suggested may have been the initial usage of many locations that ultimately became coastal towns. These seasonal fairs in which market exchange occurred could have been temporary and carried out in neutral locations away from established towns and villages. Seasonal fairs would have made good sense along the eastern African coast, where overseas trade was regulated by the monsoonal winds, as discussed previously. Seasonal fairs may have been the norm for the earliest periods of significant trade, the seventh to tenth centuries AD. Aside from the earliest levels of coastal cities, a number of single-component settlements from this period are draped along coastal areas, often extending for a kilometer or more in length (for example, Fukuchani and Unguja Ukuu on Zanzibar, Tumbe on Pemba Island; Figure 7.2). The archaeological deposits at these sites tend to be relatively compact but with imported pottery, glass, and metals well represented (Horton 1996; Juma 2004). Unfortunately, the spatial layout of these earliest cites is poorly understood as a result of a combination of ephemeral deposits and poor preservation of materials that indicate structures. One of these early sites is Tumbe, located on the northeastern coast of Pemba Island, Tanzania. This settlement is spread over more than 35 hectares but contains deposits no more than 30 to 40 centimeters in depth. This is strikingly different than later settlements with deeply stratified, almost 149

J e f f re y B . F l e i sher

7.2. Northern Pemba Island with sites mentioned in the text.

tell-like deposits (for example, Shanga, Kilwa, Manda). Based on this type of stratigraphy, as well as the absence of permanent structures and evidence of domestic animals (Horton 1996:412), Mark Horton (personal communication, 2004) has argued that sites like Tumbe and Fukuchani functioned as seasonal fairs. Although suggestive, these data do not constitute reliable evidence of market activity at these sites, and no other data exist to confirm Horton’s hypothesis. The data from Tumbe, in fact, somewhat contradict Horton’s argument. Exca­ vations at that site exposed early eighth- through ninth-century structures of mud construction. One particularly well-preserved building excavated at Tumbe contained a surprisingly rich assemblage of imported pottery, mostly sherds of large imported jars (Sasanian Islamic and Siraf storage jars) from the Persian Gulf, with smaller numbers of small imported serving bowls. The density of imports from these house deposits far exceeds that of any other excavation from similar periods, even that from the well-excavated site of Shanga on the Kenya coast (Horton 1996). Another excavated structure at Tumbe, far removed in space from the one just discussed, contained similar densities of imported goods. The density of the most common imported pottery, Sasanian Islamic storage jars, is between five and ten times greater than that from Shanga. Yet the results of excavations at Shanga led Horton (1996:412) to declare that the “quantity, quality, and range of imported pottery, present from the beginning, seem to indicate that Shanga . . . [was a] fairly sophisticated trading centre.” 150

H o u s i n g t he M ar k e t

Although Tumbe may have been a location for trade, data gathered from excavations of contemporary villages in the region surrounding the site suggest that these exchanges may have been more restricted than what is commonly called marketplace exchange. The site of Kimimba was a small village or hamlet less than two hectares in size, located approximately one kilometer southwest of Tumbe (Fleisher 2003). Extensive excavations at Kimimba recovered a rich material assemblage of local materials including local ceramics, daub, and evidence of household production (iron slag) but turned up very few imported artifacts, including just 13 sherds of imported pottery among thousands of local sherds. This contrasts greatly with the material assemblage of Tumbe, where more than 900 sherds of imported pottery were found in excavations of two houses briefly occupied during the eighth through tenth centuries. These data suggest, therefore, that market principles did not apply to the exchange of goods among villages on the island, even though they may have structured the exchanges between local merchants and long-distance traders. Early coastal market centers may also be elusive archaeologically because they took place in relatively ephemeral locations, such as beaches. The results of two excavations hint that this may have been the case. At Manda, an early town on the northern Kenyan coast, extraordinarily high densities of imported goods have been excavated from a buried ninth- to tenth-century beach deposit (Chittick 1984); more than 28 percent of the pottery in these contexts is imported (the average along the coast is between 1 and 6 percent). Another possible indication comes from beach sand placed underneath the congregational mosque at Shanga (a common practice to provide a “purifying” underlayment). Within this fill layer, Horton (1996:376) found coins that “were probably lost on the beach during normal trading activity” in the period before construction of the mosque. Active beach marketplaces today along the coast include one fish marketplace at Tumbe Mjini on Pemba Island, held on most days. Sailors bring their boats up to the beach and throw their daily catch onto the sand; a crowd forms around it, and an auctioneer takes bids until the fish are sold. Some of these fish are purchased in bulk and then resold in city marketplaces, but many area residents come to buy fish for direct consumption. Aside from a few wooden posts where nets are hung to dry and a ramshackle snack shack, there would be little structural indication that this beach was the site of one of the most important fish marketplaces on the island. Privatization of Commercial Exchanges

Unfortunately, these contexts might be the most visible evidence of Swahili market exchange. After the eleventh or twelfth century, market exchanges appear to have become increasingly private, moving into the restricted central parts of towns and to the houses of local merchants and elites. This privatization was accomplished mostly through a system of sponsorship, but as a ninth-century Chinese document describes, it may also have been governed by a system of oaths and blood brotherhoods: “When 151

J e f f re y B . F l e i sher

Persian traders wish to enter this country, they form a caravan of several thousand men and present them with strips of cloth. All, whether old or young, draw blood and swear an oath, and then only do they trade their goods” (Freeman-Grenville 1962:8). Sponsorship of foreign merchants may initially have allowed them access to a central marketplace, one available only to clan members of the town. Horton described the physical layout of the town of Shanga as a mechanism for the control of trade. Horton’s (1994, 1996) excavations revealed a significant portion of the central part of the settlement, the section of town that would eventually include the main Friday mosque. He argued that the town was divided into clan sections, each with a gate to the central enclosure—which contained individual clan houses, a mosque, a well, and the marketplace (Horton 1996:413). Access to this central marketplace would have been restricted to members of town clans, which meant traders had to become connected to a particular clan to gain access. The archaeological evidence for a marketplace in this central area is a set of timber kiosks that date from approximately AD 900–1000. Horton (1996:238) has interpreted these as market stalls, based on ethnographically known examples. After AD 1000, however, commercial transactions seem to have moved into private spaces in merchant and elite houses. If Horton’s interpretation is correct, overseas traders before AD 1000 were sponsored by local merchants but granted access to the central marketplace where they could buy and sell goods. Documents from the thirteenth and fourteenth centuries describe a similar type of sponsorship system, but by that time overseas traders were required to trade exclusively with their local hosts. Ibn Battuta’s description is the most detailed: When a ship comes into port, it is boarded from sanbuq, that is to say, little boats. Each sanbuq carries a crowd of young men, each carrying a covered dish, containing food. Each one of them presents his dish to a merchant on board, and calls out: “This man is my guest.” And his fellows do the same. Not one of the merchants disembarks except to go to the house of his host among the young men. . . . [W]hen a merchant has settled in his host’s house, the latter sells for him what he has brought and makes his purchases for him. Buying anything from a merchant below its market price or selling him anything except in his host’s presence is disapproved of by the people of Mogadishu. (Freeman-Grenville 1962:27–28)

This transformation is archaeologically visible in many coastal towns by the emergence of a distinctive stone house form. The distribution of these houses throughout a settlement depended on a number of factors, including wealth, access to permanent materials, and, as discussed previously, the political organization of the town. Today, many of these houses exist as standing ruins, a testament to the permanence they sought to embody (Allen 1979). The evolution of the floor plan of these houses through the centuries exhibits a growing distinction between public and private spaces within houses, with an intimacy gradient evident in later houses (Allen 1979; DonleyReid 1982). Based on historical and ethnoarchaeological research in the Lamu archipelago, we know that these houses contained guest rooms and public spaces at the 152

H o u s i n g t he M ar k e t

front for housing merchants and exchanging goods (Allen 1979; Donley-Reid 1982; Fleisher and LaViolette 2007). Archaeological excavations within the houses offer less conclusive evidence, as deposits located in houses are normally not attributable to the use-life of the house. Linda Donley-Reid’s (1990b) work on seventeenth- and eighteenth-century houses has provided unique insights into the way women used imported goods as a means of protection from evil spirits. This serves as a cautionary tale for making direct and unproblematic links between the presence of imported goods and market behavior. The movement of commercial exchanges into these domestic spaces can only be understood as a further step toward the domination of overseas commerce by particular merchant families and clans. As Horton (1996:413) has described, this system allowed inhabitants of Swahili towns to “control foreign traders, and thus monopolize the goods which the traders carried with them.” The emergence of a system of sponsorship between overseas and local merchants also provides a tidy example of the way market relations can proliferate through highly embedded transactions (Granovetter 1985; Garraty, Chapter 1). Mark Granovetter’s (1985:490) embeddedness argument stresses “the role of concrete personal relations and structures (‘networks’) of such relations in generating trust and discouraging malfeasance.” Although the movement of exchanges into private homes must undoubtedly be understood in terms of the growing control Swahili merchants sought over foreign goods, it should also be recognized as a means for both Swahili and foreign merchants to embed market transactions within a more trustworthy and dependable set of personal relationships. In the Swahili case, the embedding of market relations may be represented not only by the system of sponsorship but also by the movement into the most protective and trustworthy space in towns—the house. Ibn Battuta’s insistence on the equitable nature of merchant exchanges in the fourteenth century seems to suggest that even as these exchanges became more and more restricted and highly embedded, they continued to be governed by market forces. Connecting Internal and External Commercialization

What I have described to this point is what we know about the contexts and patterns of market exchange between Swahili merchants and external traders, what Michael Smith (2004:79) would call “external commercialization.” I have devoted a lengthy discussion to this topic because this external commercial activity has been so definitive of discussions about ancient Swahili society. The next question, however, is how this type of market exchange was connected to exchanges that took place within Swahili towns and regions and whether the commercial principles that evidently governed external trade also did so for internal exchanges. Did the increasing privatization and apparent control over external trade have any effect on the way exchanges occurred within towns and the regions that surrounded them? Is there any variability in the pattern of internal exchanges among coastal towns? 153

J e f f re y B . F l e i sher

I am investigating these questions through the excavation of rural villages in the hinterlands of Swahili polities and examination of non-elite contexts in the towns. Research on Pemba Island (co-directed with Adria LaViolette and Bertram Mapunda) focused on a town in northeastern Pemba Island, Tanzania, called Chwaka and the villages that surrounded it during the early second millennium AD (Figure 7.2). Chwaka is a midsized Swahili town, generally representative of the great majority of the dozens of ancient coastal towns. At its height it covered approximately 12 hectares and contained three stone mosques surrounded by honorific pillar tombs, a handful of stone houses (the presumptive houses of elite merchants), and scores of earth-andthatch houses; the population was likely between 3,000 and 5,000 people. Based on excavations of non-elite houses and rural villages, overseas goods brought into the homes of elite merchants were exchanged in greater Chwaka and the region at large (in contrast to the earlier pattern between Tumbe and nearby villages). Excavations of earth-and-thatch houses within Chwaka uncovered a full range of imported goods, including glazed pottery from the Persian Gulf, China, and the Red Sea; glass vessels and beads; and copper bowls and jewelry. These goods were present in non-elite contexts in significant quantities, not simply in trace amounts. In most earth-and-thatch houses, imported pottery made up between 1 and 2 percent of the total pottery recovered, which is generally equal to the percentages encountered in elite houses and middens. In one well-excavated earth-and-thatch house from a single occupation deposit, 242 sherds of imported pottery were located (3.8 percent of all pottery from those levels), representing seven different imported types. The most dominant imported types, Chinese celadon bowls and Martabani storage jars, were represented in this single context by more than half a dozen different vessels. Excavations at the fifteenth- to sixteenth-century village of Kaliwa (Fleisher 2003:186–188), located 2 km north of Chwaka, also revealed a full range of imported goods. Kaliwa is 2 hectares in size and composed of a series of earth-and-thatch houses but no stone domestic or religious architecture. It is situated 200 m from the shoreline of a protected inlet. Its proximity to the beach and abundant remains of fish bones suggest that it was a small fishing village (Fleisher 2003:358–364, 383). A series of test trenches and a section of an earth-and-thatch house produced all major imported pottery types, including Chinese celadon, Persian sgraffiato and monochromes, and Chinese stoneware storage jars. In addition, imported glass vessels and beads were recovered, as well as carnelian and rock crystal beads. These imported goods occurred in lower frequency than in contemporary deposits at Chwaka (and other stonetowns); imported pottery was less than 1 percent (.008) of the pottery excavated. Despite the lower frequency of imported goods, these finds suggest that the exchange of goods within the town and the region was based on market principles. As Kenneth Hirth (1998:456) argued, one signature of market exchange might be the ability of households “to provision themselves independently of one another and without regard to broader social and political relationships.” The relative homogeneity of the material assemblages, especially of imported goods, in diversity and (to a lesser extent) fre154

H o u s i n g t he M ar k e t

quency between elite and non-elite houses suggests that goods flowing into elite houses through restricted exchanges were also freely available to people of lower social rank. This evidence conflicts with other coastal archaeologists’ ideas about the nature and use of imported goods and with reports from similar types of research on other Swahili town regions. Chapurukha Kusimba (1999:133) has argued that certain imported goods (such as ceramics) were “circulated only in very restricted circles . . . [which] caused them to become emblematic of upper class status.” This is a relatively common view of imported goods among coastal archaeologists, based on an eighteenth- through twentieth-century image of the Swahili produced by John Middleton (1992) and Linda Donley-Reid (1990b). In particular, Donley-Reid’s ethnoarchaeological research on stone houses in Kenya (discussed previously) has contributed to this notion by exploring the meaning behind imported porcelain blue-on-white bowls in elite merchant houses. According to her, because these ceramics were crucial to the purity and power of elites, they were probably associated with elite members of society (Donley-Reid 1990b). Although the data from Pemba contradict this idea, similar types of data from surveys in the region around Kilwa Kisiwani conducted by Stephanie Wynne-Jones (2007a, 2007b) seem to confirm it. At hinterland sites on the mainland opposite Kilwa, Wynne-Jones (2007a) found almost no imported pottery in village sites—a surprising finding given that Kilwa was perhaps the most powerful and wealthy coastal town, through which an enormous amount of goods passed. However, hinterland sites were not completely disconnected to the spheres of Kilwa’s influence; locally produced goods, such as earthenware pottery, moved throughout the region (even types considered to have been “elite” goods, such as Husuni Modeled Ware). Wynne-Jones sees these distributional patterns as related to nested coastal identities and evidence of restricted trade in imported goods. The only imported goods that seem to have moved out from the town, in trace amounts, were glass beads imported from India. She is careful to note, however, that some goods considered prestigious in towns may not have been desired in countryside communities. On Pemba, however, the wide distribution of imported goods in non-elite urban and village contexts, even purported luxury or prestige items, is indicative of more open access to those goods. The way these goods were distributed internally—from elite merchants to non-elite villagers or town dwellers—is generally unknown. Whatever the socioeconomic processes (distribution from patrons to clients, local markets for local consumption), however, the effect was that all members of society had access to a broad range of goods; the economic distinctions between social classes were one of degree rather than kind. Thus, on Pemba at least, the somewhat restricted exchange between elite Swahili merchants and overseas traders that characterized external commercialization contrasts sharply with the archaeological evidence of more open access for internal commercialization. This is a surprising finding for a “town with king” (a characterization that could be applied to Kilwa as well), for which the presumption 155

J e f f re y B . F l e i sher

has long been that significant differences existed between elite and non-elite members of society. (This is less the case for patrician towns like Shanga and Lamu in the Lamu archipelago, where non-elites may have been a distinct minority and the towns were often physically separated from nearby countryside inhabitants by walls). These data provide initial support for rethinking monolithic understandings of the regional economies of coastal towns. The clear differences in distributional patterns between Chwaka and Kilwa Kisiwani suggest that internal commercialization occurred in some regions but not others. Surely the towns’ locational differences were important: Chwaka and the villages that surrounded it were located on a large, offshore island that may have been geographically bounded with a closely related economic region. Kilwa, however, was located on a small island, nestled against the coast with a hinterland that stretched onto the mainland and beyond—a situation not unlike towns in the Lamu archipelago. Thus, the pattern on Pemba may be more indicative of island-based towns (such as those on Pemba, Zanzibar, and Mafia), and the pattern at Kilwa could be representative of towns located along or adjacent to the coast. Rather than simply refuting the common conception of the way imported goods moved through coastal regions, the data from Pemba offer another variety of internal commercialization. The archaeological evidence from Pemba suggests that imported pottery and other goods were moving freely within the towns and beyond, but one question remains: why did merchants seek to increase their control over overseas traders and the goods they brought with them? There are a number of possible answers. One may be linked to qualities associated with local rulers and men of renown, specifically that of generosity. One of the key qualities of powerful leaders and merchants was their ability to act generously with the local population. I have recently argued that these acts of generosity were demonstrated through public feasts, in which imported bowls played a crucial role (Fleisher 2008). Imported bowls—the dominant form class of imported pottery from the eleventh to sixteenth centuries—were often incorporated into public monuments, such as mosques and tombs, and presented for display to the community. I interpret these public acts as demonstrations of a local leader’s ability to “feed” the community. Feasts, however, were not simply held under the purview of elites but were likely a common ritual practice carried out by all town dwellers. Although the feasts of elites were likely the most elaborate and had important political consequences, the private feasts of common people emulated the public ones hosted by elites and may have incorporated imported bowls into their rituals as well. Another, more economic interpretation follows the multiple exchange relationships of Swahili merchants. Coastal merchants were middlemen and brokers, negotiating the trade among Indian Ocean merchants, local farmers and craftspeople, and hinterland groups. Each of these groups brought specific products to the coastal market locus, but it was the Swahili merchants who connected them rather than a physical marketplace. Many of the goods sought by overseas merchants were brought in from the coastal hinterlands: ivory, animal skins, and gold. The goods sought by hinterland 156

H o u s i n g t he M ar k e t

groups were often coastal resources and some overseas goods: cloth, shell, and glass beads. Thus, coastal merchants needed to obtain overseas goods, such as imported pottery, to trade with local farmers and craftspeople so they could acquire the goods necessary to trade with other hinterland groups. Coastal merchants therefore needed to tightly control the imported goods flowing in because those goods were crucial to their ability to broker trade. Some locally produced coastal goods were in demand overseas, such as mangrove wood poles, and coastal merchants therefore needed to ensure that overseas traders were not able to simply trade directly with local woodcutters, which would effectively have cut them out of their position as brokers. Therefore, rather than open marketplaces where goods were traded freely, elite merchants on Pemba participated in market exchanges in ways that allowed them to control both the in- and outflows of commodities on the coast. The distribution of imported pottery in non-elite houses in towns and villages provides a window into this process. Imports that were considered prestigious locally and were possibly necessary to carry out ritual feasts were sold by local merchants to acquire other items they needed to conduct exchanges with overseas merchants and with partners in the coastal hinterlands. In addition, these exchanges allowed them to provision their own houses and provide for the overseas merchants who stayed with them. The growing restrictions local merchants put on overseas traders were also possibly a reaction to increasing competition among coastal towns and merchants; by the thirteenth and fourteenth centuries, dozens of coastal towns were vying for a piece of the Indian Ocean trade. On Pemba alone, five large towns and nine smaller ones, all located on the island’s coast, participated in long-distance exchanges. Although the smaller towns may have served as secondary centers to the prominent towns (Chwaka, Ras Mkumbuu, Mkia wa Ngombe, Mtambwe Mkuu, and Mandani), the largest towns grew exponentially during the centuries of intense coastal trade. A process of sponsorship between overseas and local merchants made it more likely that overseas merchants would continually revisit the same local merchant. This arrangement served as a way to compete with other town and island merchants and, as Horton (1996:413) suggested, as a way to facilitate control over important overseas resources. The process of sponsorship and the increasing privatization of external commercialization are indicative of the growing embeddedness of market transactions. These transactions—highly embedded in personal relationships between foreign and local merchants, as well as in the houses of local merchants—should not be seen as contradictory to the “undersocialized” and “atomized” behavior often associated with market transactions. As Granovetter (1985) and Garraty (Chapter 1) have argued, many premodern market systems developed based on such personalized and trustworthy relationships. These personal relations provided stability and trust between exchange partners (supported as well through the moral connections of shared religious practice), but they also served a practical function for overseas merchants who needed a place to stay during the weeks or months after their arrival while they waited for the monsoon winds to shift back north. Sailing at the cusp of monsoonal shifts was not 157

J e f f re y B . F l e i sher

advised because it could leave a trade vessel stuck in the doldrums between seasons or in a situation in which it was unable to reach its destination. Thus, overseas merchants often had slack time on the coast and required the hospitality of their local sponsoring merchant; this hospitality thus served a practical purpose as well as facilitating and encouraging market transactions. Conclusion

My goal is to instigate a discussion about the presence and nature of market exchange along the Swahili coast of eastern Africa before the sixteenth century AD. During the first half of the second millennium AD, Swahili merchants became increasingly involved in long-distance exchanges within the Indian Ocean trade system. Archaeological and historical data suggest that this pattern was accompanied by the increasing privatization of exchange relationships between Swahili elites and overseas merchants, as commercial exchanges moved into the inner spaces of stone houses in Swahili towns. Many archaeologists have viewed this as an effort to control access to prestigious goods, such as imported pottery, which became emblematic of local status. Archaeological data from urban non-elite and rural houses on Pemba Island suggest that the imported goods were widely distributed throughout these island regions, possibly indicating the presence of marketplace exchange within the regional system. This pattern offers a contrast to normative views of the way imported goods were deployed by Swahili towns (Kusimba 1999) and provides an alternative model of internal commercialization to other models of economic exchange inferred from data collected during surveys in the regions surrounding other coastal towns (Wynne-Jones 2007a). Thus, on Pemba, towns were likely centers of market exchange in both external spheres (between local and overseas merchants) and internal spheres (between local merchants and rural and non-elite members of society). In this regard, these towns were like cities in other regions of the world that fostered balanced market transactions. Yet they were also different in that market exchanges may not have been transacted in the public sphere but rather in the confines of elite homes and other private spaces. These elite members of society would have dispersed the goods, perhaps in exchange for others they needed for themselves or for trade to the interior. This pattern suggests that merchant elites on Pemba maintained significant control over the distribution of goods in towns and surrounding regions but not in ways that restricted the distributional effects of the market system. Rather than view the movement to more private transactions as a hindrance to market exchange, the embeddedness of these transactions may have provided an important support for, and extension of, market transactions. Acknowledgments. The research in this chapter includes that from my dissertation and a collaborative project with Adria LaViolette and Bertram Mapunda. Both projects were supported by the National Science Foundation (INT-9906345 and BCS158

H o u s i n g t he M ar k e t

0138319) and the University of Virginia. My dissertation research was also supported by the Fulbright Foreign Scholarship Board and the Wenner-Gren Foundation; I am grateful for all this support. I thank Chris Garraty and Barbara Stark for their excellent editorial work. Ken Hirth, Adria LaViolette, Elizabeth Vann, and Stephanie WynneJones provided critical readings and comments that strengthened the chapter. Finally, I thank Maalim Hamad Omar, director, Department of Archives, Museums, and Antiquities, Zanzibar; my colleagues in archaeology in Dar es Salaam, Zanzibar, and Pemba; and the communities on Pemba that supported me throughout the years.

159

Chapter Eight

Regional and Local Market Systems in Aztec-Period Morelos

Michael E. Smith

Ancient market systems are regional in scope. Whether they take the form of isolated solar market systems or complex interlocking systems (C. Smith 1974, 1976a), market systems integrate regions economically. Ethnographers and historians have found that large regional peasant market systems—the kind Carol Smith described as complex interlocking marketing systems—are typically composed of two hierarchical levels with distinct spatial expressions. The smaller level, which I call the local system, usually consists of a weekly market that serves a town and its hinterland. In China, G. William Skinner (1964) has called this the “standard marketing community,” and in many parts of Mesoamerica the local system consists of a central town (municipio or cabecera) and its rural dependents (sujetos). These local systems then form parts of a larger regional market system based in a major urban center. For example, in Mexico the Valley of Oaxaca regional market system in the twentieth century comprised a number of smaller local systems that corresponded to municipios (Cook and Diskin, eds. 1976; Malinowski and de la Fuente 1982). The distinction between local and regional systems in a complex market system is important for understanding peasant economies and social organization, but archaeologists have had trouble distinguishing these levels in ancient societies. Regional survey projects employing systematic surface collections of artifacts often cover areas of sufficient size to model the different levels of market systems, but the typological composition of surface collections may not provide the chronological and spatial resolution to address this issue. By applying chemical provenance methods to survey 161

Michael E. Smith

collections, however, archaeologists have made headway in distinguishing local and regional systems in the Aztec-period Basin of Mexico (Garraty 2007; Minc 2006). Excavation data typically have greater chronological and contextual control than surface collections, but in most areas too few sites have been excavated to reconstruct regional exchange systems from excavated data. In the Mexican state of Morelos, however, a number of Aztec-period sites (Middle and Late Postclassic periods) have been excavated by U.S. and Mexican archaeologists. I have analyzed the ceramic collections from nine of these sites (M. E. Smith 2010) and used these data to reconstruct the local and regional levels of market systems and their changes through time within the Aztec period (ca. AD 1100–1520). The full ceramic data and an abbreviated version of the spatial analysis are presented elsewhere (M. E. Smith 2010). Ethnohistoric sources describe the importance of markets, merchants, and commercial exchange in Postclassic Morelos, and the documentary information is strengthened by a number of archaeological studies of ceramic and obsidian exchange (see the next section). These data, however, provide little information on the spatial extent or organization of the market systems. In this chapter I employ several kinds of ceramic data to generate a series of schematic maps of the likely spatial extent of regional market systems. My argument is based on analogy with ethnographic work on peasant market systems in which local and regional systems have different material culture expressions. This chapter builds on recent conceptual and methodological advances in the archaeological analysis of past economies (e.g., Earle 2002; Feinman and Nicholas, eds. 2004; Hirth 1996, 1998; M. E. Smith 2004; see also Garraty, Chapter 1). The fine-grained analysis of ceramic distributions, coupled with relatively detailed ceramic chronologies in Postclassic Morelos (Hare and Smith 1996; M. E. Smith 2010; M. E. Smith and Doershuk 1991), permit the analysis of changing spatial and economic dynamics of market systems within the Aztec period. Postclassic Economy and Society in Morelos

The modern Mexican state of Morelos is located in the Central Mexican highlands, separated from the Basin of Mexico to the north by the Ajusco mountain range (Figure 8.1a).1 The elevation of Morelos is 1,000 meters lower than the Basin of Mexico, and the environmental differences between the two regions provided a major stimulus to exchange throughout the prehispanic past (Sanders 1956). Like the Basin of Mexico, Morelos was settled by Nahuatl speakers claiming an origin in Aztlan. These peoples arrived in the Early Aztec period (ca. AD 1100–1300), a time of population growth and the founding of cities and dynasties. City-states (altepetl in Nahuatl) spread across the landscape of Morelos, the Basin of Mexico, and other parts of Central Mexico. Early Aztec polities engaged in an active program of exchange in both the Basin of Mexico (Minc 2006; Minc, Hodge, and Blackman 1994) and Morelos and also warred with one another. 162

8.1. (a) The location of Morelos within Central Mexico, (b) the locations of sites discussed in this chapter.

Michael E. Smith

The succeeding Late Aztec period (ca. AD 1300–1520) was a time of continuing population growth, economic exchange, and warfare throughout Central Mexico. The first half of the period (Late Postclassic A, ca. AD 1300–1430) saw the rise and fall of several small imperial polities centered at Azcapotzalco and Texcoco in the Basin of Mexico, Cuauhnahuac in Morelos, and perhaps Calixtlahuaca in the Toluca Valley. During the Late Postclassic B period (ca. AD 1430–1520), the Triple Alliance or Mexica empire came to power and conquered most of Central Mexico and much of northern Mesoamerica. The city-states of Morelos were incorporated into the empire as the tributary provinces of Cuauhnahuac and Huaxtepec. This political trajectory, which led from city-states through small empires to a single powerful empire, provided the backdrop for the development of commercial exchange and market systems in Postclassic Morelos. Markets in the Postclassic Economy of Morelos

Populations grew rapidly in Postclassic Morelos, and by the Late Postclassic period most of the state was settled by farming households, with an overall regional population density on the order of 150 persons per square km. Land was owned by nobles, and commoners were granted access to agricultural plots through a variety of mechanisms, including membership in a calpolli (local administrative unit) and relations of servitude to a noble or the king. River valleys were extensively irrigated, with maize and cotton as the major crops, and hillslopes were covered with terrace walls. Given the high regional population levels, the intensity of agricultural production, and the dynamic political landscape of small city-states, it is hardly surprising to find that commercial exchange and marketplaces were prominent institutions in Aztec-period Morelos. There are several types of ethnohistoric documentation for market systems in Morelos. First, local administrative documents mention marketplaces in at least twelve communities, including major capitals such as Cuauhnahuac and Yautepec, city-state capitals such as Tlayacapan and Xantetelco, and smaller subject towns like Ocotepec and Tianguistenco (M. E. Smith 1994:table 12.5). Second, the documents mention that merchants engaged in long-distance trade and operated in a number of Morelos marketplaces, including both the regional merchants discussed by Frances Berdan (1988) and the better-known pochteca (high-status, long-distance merchants). Third, a variety of general statements about the prevalence of trade in Morelos are consistent with the operation of market systems, although they do not provide direct evidence for them. Fray Diego Durán (1967 [1567–1581], vol. 2:23), for example, stated that people from all over (“toda la tierra”) traded with Morelos cities to obtain cotton. Archaeological fieldwork provides additional evidence for commercial exchange at Postclassic sites in Morleos. First, the extent of exchange—as measured by the quantities of imported ceramics, obsidian, and other goods—was fairly high (discussed later). Second, high-value imported goods at Yautepec and Cuexcomate were widely 164

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

distributed among households, both elite and commoner (Earle and Smith 2010), a pattern indicative of commercial exchange in household provisioning (Hirth 1998; M. E. Smith 1999). Third, the household patterning of obsidian imported from different geological sources conforms to the expectations of commercial exchange (M. E. Smith et al. 2007). The ethnohistoric and archaeological evidence reviewed in the previous section provides the starting point for the present analysis. Given that processes of development of markets and market systems were important in Aztec-period Morelos, what can we say about their spatial and hierarchical organization? Archaeological Sites Discussed Here

The archaeological data used in this chapter consist of counts of ceramic types from excavated deposits at nine Aztec-period sites in Morelos (Figure 8.1b). I studied many of these collections in 1980, and the sites and ceramic data are described in a recent monograph (M. E. Smith 2010). Data from sites excavated since 1980— Cuexcomate, Capilco, and Yautepec—are described in part in the same publication (M. E. Smith 2010), with more recent information available in published and unpublished reports (including this chapter). The sites are described in order from west to east. Coatetelco. Coatetelco was excavated in the 1970s by Raúl Arana (1984), who reconstructed the public architecture of a small urban epicenter. The ceramic data are from test pits Arana excavated outside of the epicenter; I analyzed these collections in 1980 (M. E. Smith 1983, 2010). Xochicalco. Xochicalco is an urban center of the Epiclassic period with a long history of fieldwork and analysis. The collections used here come from Postclassic levels of stratigraphic test pits excavated in the 1970s by Kenneth Hirth (2000; Hirth and Cyphers Guillén 1988). This material is discussed in detail in my dissertation (M. E. Smith 1983) and summarized in later publications (M. E. Smith 2000, 2010). Its context probably represents small hamlets or isolated households that settled on the lower slopes of the Xochicalco hill a century or more after the Epiclassic urban center was abandoned ca. AD 900. Cuexcomate and Capilco. Cuexcomate and Capilco are rural Aztec sites first located by Hirth (2000). I excavated houses and other structures at these sites in 1986 (M. E. Smith 1992), and the ceramics and other artifacts are described in a recent monograph (M. E. Smith in press-a). Capilco was a small village of commoners that grew from a few initial houses in the Early Aztec period to a maximum of about twenty houses in Late Postclassic B times. Cuexcomate was a larger site with a resident noble household and temple. It was founded in the Late Postclassic A period and grew larger 165

Michael E. Smith

in the Late Postclassic B period. The ceramic data used here are presented in another recent monograph (M. E. Smith 2010). Cuauhnahuac (Teopanzolco and the Palacio de Cortés site). Cuauhnahuac was the largest city and most powerful capital in Morelos throughout the Postclassic period. After the Spanish conquest, its name was changed to Cuernavaca. Two of the major Aztec-period archaeological sites excavated in Cuernavaca are Teopanzolco and the Palacio de Cortés site. Teopanzolco is an archaeological zone located in a residential neighborhood of Cuernavaca. It consists of a large public plaza with a twin-temple pyramid and other public architecture dating to the Early Aztec period. Teopanzolco has been excavated during a number of projects between 1920 and today (M. E. Smith 2008). The collections studied here are from test pits excavated in the 1970s by Jorge Angulo Villaseñor (1976); they include fill from Structure 2 and several non-structure contexts within the epicenter. The Palacio de Cortés site is located under the Museo Cuauhnahuac in downtown Cuernavaca; it was the site of the palace of the tlatoani (king) of Cuauhnahuac at the time of the Spanish conquest. The collections used here were excavated by Angulo Villaseñor (1978) in the 1970s. They include an Early Aztec deposit for which the social context is unclear, overlaid by deposits from the royal palace dating to the Late Aztec B period. I studied collections from these two sites in 1980 (M. E. Smith 2010). The most likely scenario for the relationship between the two sites is that the urban epicenter of Cuauhnahuac was moved from Teopanzolco to the Palacio de Cortés site sometime early in the Late Aztec period. Tepoztlan (Tepozteco). Tepoztlan was a large Postclassic city-state. The collections used here are from excavations at the Temple of Tepozteco (located on cliffs above the modern and Postclassic town centers) by Angulo Villaseñor in the 1970s (M. E. Smith 2010). Information about the temple can be found in several publications (Ceballos Novelo 1928; Seler 1993 [1939]). Yautepec. Yautepec was a major political capital whose king ruled several citystates in the Río Yautepec valley. I excavated houses at Yautepec in 1993 (M. E. Smith in press-b; M. E. Smith, Heath-Smith, and Montiel 1999), and the ceramic data used here are reported in my recent monograph (M. E. Smith 2010). They are from a sample of ceramics from residential middens. Tetla. Tetla is an Early Aztec site excavated by David Grove (1987) in the 1970s. The site consists of a single excavated residential compound, with a surrounding area of Postclassic ceramics on the surface. This was probably a small village in the Postclassic period. The ceramics used here are from the residential compound described by Lynette Norr (1987a, 1987b); my data are from a partial reanalysis of those data combined with use of Norr’s publications. 166

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

Table 8.1. Frequency of Imported Ceramics at Sites Discussed in This Chapter. Site

Guinda

Morelos Imports

Basin of Mexico

Western Imports

Other Areas

Total, nonGuinda

Middle Postclassic (MPC) Xochicalco

5.32

0.75

0.29

1.20

0.00

2.24

Capilco

5.95

3.26

2.61

0.12

0.00

5.99

Teopanzolco

7.20

0.14

1.30

0.20

0.00

1.64

Palacio de Cortés

5.90

0.00

0.29

0.00

0.00

0.29

Yautepec

5.05

1.06

1.52

0.19

0.02

2.79

Tepozteco

0.56

0.06

0.31

0.06

0.12

0.56

Tetla

12.74

0.00

0.64

0.64

0.00

1.27

Mean

6.10

0.75

0.99

0.35

0.02

2.11

Coatetelco

4.86

1.48

2.36

0.81

0.00

4.65

Xochicalco

2.40

0.55

0.92

0.23

0.00

1.71

Capilco

2.51

0.85

3.00

0.27

0.04

4.16

Cuexcomate

6.13

0.46

4.76

0.06

0.02

5.31

Yautepec

3.17

0.50

1.83

0.04

0.03

2.40

Mean

3.81

0.77

2.57

0.28

0.02

3.64

Coatetelco

1.82

0.55

0.91

0.91

0.00

2.36

Xochicalco

0.75

0.25

0.25

0.13

0.00

0.63

Capilco

1.84

0.14

2.57

0.17

0.01

2.88

Cuexcomate

4.80

0.11

4.37

0.04

0.00

4.52

Palacio de Cortés

9.76

0.00

5.08

0.00

0.00

5.08

Yautepec

2.85

0.31

1.47

0.06

0.01

1.84

Mean

3.64

0.22

2.44

0.22

0.00

2.88

Late Postclassic A (LPC-A)

Late Postclassic B (LPC-B)

Note: Data are expressed as percentages of all sherds. Source: Smith (in press-b).

Ceramic Imports

As noted previously, the quantity of imported ceramics and obsidian is high at Postclassic sites in Morelos. Table 8.1 shows the combined frequencies of imported ceramics (expressed as percentages of all sherds) at the nine sites listed in the previous section (for documentation, see M. E. Smith 2010). Guinda (polished red ware) ceramics are problematic because they were manufactured in both the Basin of Mexico and at least one place in Morelos (Minc in press), but the types cannot be distinguished without chemical analysis. As a rough approximation, based on Leah 167

Michael E. Smith

Minc’s (in press) chemical analysis of guindas from Yautepec, I would guess that half of the guinda sherds at most Morelos sites were produced somewhere in Morelos and the other half were imported from the Basin of Mexico. The most abundant basin of Mexico imports are the various types of Aztec black-on-orange, Xochimilco polychrome, and Texcoco fabric-marked salt vessels. Western imports are types from Guerrero, Malinalco, and the Toluca Valley. Other areas of origin for imported ceramics include eastern Morelos and the Puebla-Tlaxcala area. The places of origin for the presumed imported sherds at Morelos sites—including locations both outside and within the state—were initially identified on the basis of type frequencies (M. E. Smith 1983). Evidence accumulated since 1983 on the spatial distributions of ceramic types supports the hypothesized places of origin for most types (M. E. Smith 2010:chapter 3, in press-b:chapter C2). A series of chemical analyses (instrumental neutron activation analysis) of selected ceramic types from Yautepec also supports the type distribution data (M. E. Smith, Fauman-Fichman, and Neff in press). For example, the types Aztec II, Aztec III, and Aztec III/IV blackon-orange are traditionally assumed to originate in the Basin of Mexico based on their distributions. In our analysis, nine out of ten sherds of Aztec II black-on-orange from Yautepec match compositional groups from the Basin of Mexico, as do twenty out of twenty-two sherds of Aztec III black-on-orange and ten out of ten sherds of Aztec III/IV black-on-orange. For ceramics produced within Morelos, most of the sherds tested by chemical analysis fit into their hypothesized place of origin. For example, eight out of ten sherds of Tlahuica polychrome Type B7 and Type B8, hypothesized as local Yautepec pottery, fit with the local Yautepec compositional group; all five tested sherds of Type B4, a type from western Morelos, are in the western Morelos compositional group. The analysis did produce some complexities. For example, all five analyzed sherds of the type Tepozteco black-on-white—hypothesized to originate in Tepoztlan—were grouped with the Yautepec ceramics, perhaps because there were no reference samples from Tepoztlan. In this study I classify this type as a Morelos import, however. My reconstruction of marketing areas in the next section is based on the occurrences of eleven ceramic categories at the nine sites listed previously. These categories are listed in Table 8.2; the frequencies of each category at each site are contained in tables published elsewhere (M. E. Smith 2010). Figure 8.2 illustrates these categories. Most are individual ceramic types, although several of the categories (Teopanzolco types, Type C polychromes, and guinda) consist of groups of related types.2 Before exploring the spatial distributions of these types, I briefly examine the comparative basis for inferring market areas from material culture distributions. Ceramic Distributions and Market Areas

As noted, a key distinction in the analysis of complex interlocking marketing systems is the hierarchical relationship between the local and regional systems. Local systems 168

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

8.2. Examples of decorated ceramic types used in the analysis. See Table 8.2 for the key to the types. Drawings are from M. E. Smith (2010, in press-a).

are typically organized around a single central place (i.e., a single market center), whereas regional systems are nodal systems that include a number of levels of hierarchically organized communities. In the modern peasant market system in the Valley of Oaxaca, Martin Diskin (1976:236) noted that local systems typically consist of a market town and its surrounding villages, whereas the regional system consists of the totality of local systems and corresponds to the entire valley. A similar hierarchy is found in the market systems of highland western Guatemala (McBryde 1947; Reina and Hill 1978; C. Smith 1975). With reference to material culture, one feature of local systems in a regional framework is that they have often been found to be the primary spatial “unit of culture” in peasant societies. This was first pointed out by Skinner (1964) in regard to Ch’ing China. In Skinner’s standard marketing community or the equivalent Mesoamerican municipio, a central place held a weekly market assembly, with a number of surrounding villages and hamlets dependent upon this marketplace. Skinner (1964:32) argued that the standard marketing community is “the effective social field of the peasant” in China; as a result, it represents the primary “culture-bearing unit” (Skinner 1964:39) in that it tends to be internally homogeneous and externally variable in many identifiable domains of culture. 169

Michael E. Smith

Table 8.2. Ceramic Types Used in the Cluster Analyses.   A B C D E F G H I J K

Type Wavy-line red-on-cream Red-on-buff Morelos-Puebla black-on-orange Tepozteco bichrome Teopanzolco types Type C polychromes B4 polychrome B7 polychrome B8 polychrome Guinda Black-rim orange

  MPC  x x x x x x x x

LPC-A

x x

x x

x x x x

LPC-B

x x x x x x

Similar patterns of cultural homogeneity at the level of the local system have been observed ethnographically in other peasant societies. A number of mechanisms can contribute to the establishment and maintenance of this cultural homogeneity. For example, administrative ties can isolate local systems from one another, or variations in religious observances between local systems might lead to differentiation in some aspects of material culture. But market systems provide some of the strongest means for shaping spatial variation in the material realm. Most of the goods and services used by peasants are obtained at the local marketplace or market town. The central church or temple serves the population of the local system, which in many cases tends to be endogamous. Also, voluntary associations and other formal and informal organizations generally fit within the boundaries of the local system (O. Lewis 1951; McBryde 1947:88ff; Skinner 1964; C. Smith 1974). In China, the cultural homogeneity of the standard marketing community was expressed in such realms as the standardization of weights and measures, linguistic micro-dialects, and stylistic details of textile design elements (Skinner 1964:39–40). As Manning Nash (1967:93) reported for Mesoamerica, “[W]hole communities [i.e., municipios or local systems] have cultural traditions which vary from each other in endless small ways.” The larger regional systems that incorporate one or more local systems do not exhibit the same degree of cultural uniformity (Skinner 1964). For example, Diskin (1976:242–243) pointed out that Oaxaca has a higher degree of “cultural similarity” within local systems than is found within the valley-wide regional system. Nevertheless, the Valley of Oaxaca as a whole is culturally distinct from surrounding areas of Mexico. There are thus two levels of cultural similarity in Oaxaca. Both are manifest in some domains of material culture, and both are structured and maintained by the operation of the Valley of Oaxaca market system. Although specific information is scattered, the ethnographic sources reviewed previously suggest that over time, regional systems expand and contract in response to both local and exogenous forces. Local systems, on the other hand, seem to have more stability through time. 170

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

On the basis of the aforementioned observations, I argue that local areas within Postclassic Morelos that share closely similar assemblages of ceramic types can be interpreted as local market systems. Regional market systems can be identified where there is a degree of similarity and exchange among adjacent local systems. I assume that the largest city within a given regional system served as the major central place that integrated the system. This argument is similar to that used by Mary Hodge and Leah Minc in their analyses of Basin of Mexico market systems using ceramic data (Hodge et al. 1992; Hodge and Minc 1990; Minc 2006; Minc, Hodge, and Blackman 1994). Reconstruction of Market Areas in Postclassic Morelos Methods and Assumptions

The excavated ceramic collections from the nine sites described previously are well phased and have large sherd frequencies; they reveal regional patterns for future testing with survey data, other excavations, or both. I employ a simple graphical method to express the strength and direction of commercial relationships among sites. For each time period I construct a regional market model based primarily on similarities among ceramic collections and the presence of imported types. These data and the model are portrayed in three parallel schematic figures (Figures 8.4–8.6), one each for the Middle Postclassic, Late Postclassic A, and Late Postclassic B periods. In this section I explain the nature and derivation of the four sections of each of these figures. Sites. The top left portion of each figure shows the sites that have quantified collections for that time period. Other sites are also included, where relevant. Cluster Analysis. I employ hierarchical cluster analysis to analyze the nature of ceramic similarities among quantified collections. Similarities and differences in the overall type inventories of deposits and sites can arise from a variety of factors, ranging from ethnicity to social class to variations in religious rituals, as well as sample error. To focus on market exchange, I limit consideration to the decorated ceramic types presumed to originate in Morelos (Table 8.2). As discussed previously, these ceramics were widely exchanged within and between local regions, almost certainly through the market system. Consumers must have deliberately chosen vessels of diverse decoration, from different origins, and this practice provides the means to examine commercial relationships through ceramic type compositions. The three cluster analysis dendrograms are shown in Figure 8.3. They were produced with SPSS statistical software using the average-linkage technique, with similarities calculated based on simple Euclidean distance between standardized values (Z scores) for key ceramic variables (expressed as percentages of all local decorated sherds). The ceramic variables used for each period are listed in Table 8.2. In Figures 171

Michael E. Smith

8.3. Cluster analysis dendrograms of ceramic similarity among sites.

8.4–8.6, the results of the cluster analyses are drawn on the maps, with ellipses linking similar collections. Imported Ceramics. The presence of imported sherds is indicated in Figures 8.4–8.6 by lines connecting the sites or regions, with arrows indicating the direction of movement. In these figures I only consider the presence and absence of imported sherds, not their quantity. This allows the inclusion of a larger number of excavated contexts, including those too small or specialized for quantitative analysis. For example, Teopanzolco is represented in the cluster analysis by a single large collection from Structure 2, whereas information from other contexts yielding ceramics is used for the diagram of imported ceramic patterns. Market Model. In Figures 8.4–8.6, I present a hypothetical spatial market model for each period based on the similarity patterns and imports described previously, plus other considerations. The latter category includes information from ethnohistoric and other sources (for example, the Relación Geográfica from Tepoztlan mentions intensive trade between Tepoztlan and Yautepec in the Late Postclassic B [LPC-B] period, but there are no quantified ceramic collections from the former site [Acuña 1984–1988, vol. 6:195]). These models include three components. (1) Local market areas, denoted by circles or ellipses, are drawn to include areas with highly similar ceramic assem172

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

8.4. Schematic market model for the Middle Postclassic period (AD 1100–1300).

blages. (2) Strong commercial ties, indicated by thick connecting lines, connect sites with numerous imports and strong ceramic similarities. (3) Weak commercial ties, denoted by thin lines, are drawn to connect sites with fewer imports and less similarity among collections. The distinction between strong and weak commercial ties is a subjective judgment based on my assessment of the cluster analysis, imported sherds, and any other relevant data. Middle Postclassic Market Areas

Of the three periods under consideration here, the Middle Postclassic (MPC) period appears to have the highest degree of economic integration throughout Morelos (Figure 8.4). As a caveat, it should be noted that two factors contributing to this situation are the larger number of sites and the larger number of local decorated ceramic types in MPC relative to LPC times. Nevertheless, the differences between the MPC and LPC cluster analyses are striking. Not surprisingly, the closest similarity (in the cluster analysis) is between the two Teopanzolco-phase sites in Cuernavaca, that is, Teopanzolco, and the Palacio de Cortés site. The next closest similarity links Yautepec with Tetla, a connection that cuts across an ethnic divide that runs northsouth through Morelos (M. E. Smith 2010:chapter 16). These two clusters then join together, and Capilco and Xochicalco are added next, leaving Tepozteco as the most 173

Michael E. Smith

distinctive ceramic collection (Figure 8.4). This pattern contrasts greatly with the two LPC cluster analyses, in which all of the highly similar sites are in one region (western Morelos) and Yautepec, a more distant site, is the most divergent. The divergence of the Middle Postclassic Tepozteco ceramics may result in part from the special nature of the deposits—possible priests’ residences on the cliffs next to the Temple of Tepozteco. Tepozteco (and Tepoztlan) was not cut off from MPC exchange networks, however. Tepozteco bichrome is common at Yautepec, and Teopanzolco polychromes (as well as exotic imports such as Fine Orange from the Gulf Coast) are found in the Tepozteco collections. One of the most notable features of ceramic exchange in the MPC period is the wide extension of polychromes of the Teopanzolco ceramic complex (Figure 8.2e). These ceramics, particularly Types A1, B1, F, and G, are common at every MPC deposit except Tetla. The MPC city of Cuauhnahuac, centered on the Teopanzolco ceremonial precinct with its large twin-stair pyramid, was probably the largest settlement in Morelos at this time, and the presence of numerous imports shows that it was well connected to exterior areas (see the next section). It was likely the predominant market center in Morelos and served as a regional center for the economy of the entire area of the state. The combination of Cuauhnahuac’s economic and symbolic prominence led to the widespread distribution of its ceramics in central and western Morelos. The Teopanzolco ceramic complex remained a local phenomenon, predominant only in the immediate vicinity of Cuauhnahuac. The contemporaneous Temazcalli complex covered western Morelos, suggesting that Cuernavaca and the Xochicalco area were probably not part of the same local system or the same polity at this time. These data indicate that the regional economy of MPC Morelos had a number of separate local market systems, linked together through commercial exchanges into a single larger regional economy. Each of the local systems shown in Figure 8.4— Teopanzolco/Palacio de Cortés sites, Tepozteco, Yautepec, Xochicalco/Capilco, and Tetla—had its own distinctive ceramic complex, but all were in commercial contact with one another. The least secure link is between Yautepec and Tetla. Although there are no definite direct imports among these sites, the common presence of ceramic Types C1, C2, C3, and B5 at Yautepec, Tetla, and Las Pilas (a site near Tetla with Middle Postclassic burials) implies close ties among these areas. Late Postclassic A Market Areas

In both Late Postclassic periods (LPC-A and LPC-B), the sites in western Morelos form a tight grouping in the cluster analysis (Figures 8.5 and 8.6), and Yautepec is highly divergent from the other sites. For the LPC-A period it is difficult to assess the role of Cuauhnahuac because the Teopanzolco complex continues into the first part of that period, after which it was replaced by the Tecpan complex. Ceramics from the Teopanzolco complex are common imports in the LPC-A period in both western Morelos and Yautepec, indicating continuing commercial ties and the continued 174

8.5. Schematic market model for the Late Postclassic A period (AD 1300–1430).

8.6. Schematic market model for the Late Postclassic B period (AD 1430–1520+).

Michael E. Smith

economic prominence of Cuauhnahuac. Tepozteco bichrome ceramics also continue into the LPC-A period at Yautepec, suggesting that either the Tepozteco phase also extends into the LPC-A period or that this type continues as part of the subsequent LPC ceramic complex(es) of the Tepoztlan area. The pattern of linkages within the western Morelos cluster indicates that all four sites were part of the same local system; Coatetelco, the most distant of the four sites, joins Xochicalco in the tightest cluster. Of the four sites, Capilco is the most divergent, perhaps because it was a smaller settlement than the others. The presence of Type B4 polychromes at Yautepec and Type B7 at Cuexcomate documents continued exchange between these two areas in LPC-A times, but the great dissimilarity of Yautepec from the western sites in the cluster analysis points to an increasing division between western Morelos and the Yautepec area that continued into the LPC-B period. The nature of commercial ties between southeastern Morelos and the other areas is not clear. No LPC ceramic complexes have been defined in the Tetla/Las Pilas area, although it is almost certain that many Las Pilas burials date to LPC times (M. E. Smith 2010: chapter 15). Type C polychromes continue to be common at Yautepec in LPC times, and they may have continued to form a link between Yautepec and the east. In sum, the LPC-A period witnessed a weakening of interregional integration compared with the earlier MPC period. The strongest ties, in both similarity and imports, were between the Cuernavaca and Xochicalco areas. Late Postclassic B Market Areas

Ethnohistoric sources show that the LPC-B period in Morelos witnessed processes of political expansion. Polities centered on the major cities (Cuauhnahuac, Yautepec, Tepoztlan, and several others) conquered nearby city-states to form larger, more complex polities that I have called conquest states (M. E. Smith 1994). The ceramic data correspond with these political dynamics and suggest that these conquest states became separate regional market systems. In the Late Postclassic B period the Cuauhnahuac ceramic complex expanded from its original home in the Xochicalco area to encompass the entire region of western Morelos, including Cuauhnahuac and other sites in the Cuernavaca area. The great similarity of ceramics throughout this area, which corresponds closely to the territory of the Cuauhnahuac conquest state, suggests that it constituted a single regional market system (Figure 8.6). The tightest cluster is composed of Coatetelco, Xochicalco, and Capilco, but this cluster is joined next by the Palacio de Cortés site in Cuernavaca and then by Cuexcomate. Again, Yautepec has the most divergent ceramics among the sites with quantified collections. Ceramic trade between Yautepec and western Morelos/Cuernavaca declined in this period, although it did not stop completely (see next section). Exchange between these two areas and southeast Morelos is inferred by the presence of B4 and B7 polychromes in the Las Pilas burials (although these might also date to the LPC-A period). 176

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

Although there is little ceramic evidence for trade between Yautepec and Tepoztlan, strong commercial ties are attested in the ethnohistoric evidence (Acuña 1984–1988, vol. 6:195). Additional excavations in Tepoztlan would help illuminate the situation. As noted previously, the ceramic evidence implies that the entire area of western Morelos and Cuernavaca constituted a single regional market system in the LPC-B period. The correspondence between the distribution of the Cuauhnahuac complex and the territory of the Cuauhnahuac polity points to a role for the market system in integrating the twenty or so city-states that were part of this conquest state. The archaeological data provide support for the ethnohistorically documented military expansion of Cuauhnahuac into western Morelos and northeast Guerrero in the early fifteenth century (M. E. Smith 1986). The replacement of the Teopanzolco complex in the Cuernavaca area by the Cuauhnahuac complex is a surprising development: the ceramic types and styles from a conquered periphery replaced the ceramics of the political center. This change probably occurred in part because Cuauhnahuac/ Teopanzolco was conquered by the Tepanec empire in the early fifteenth century, leading to the destruction and abandonment of the city centered around the monumental architecture of Teopanzolco (Santamarina 2006; M. E. Smith in press-a:chapter 2, 2010:chapter 2). Change through Time

The spatial configuration of regional market systems in Morelos changed considerably during the Postclassic period. Perhaps the most notable transformation was a decline in the overall economic integration of Morelos. In the MPC period there were many ceramic complexes, marking a number of local market systems. Exchange among these systems was prevalent, and I suggest that the entire area of the state of Morelos can be considered a single regional marketing system integrated by Cuauhnahuac. By the LPC-B period there were fewer complexes, with weaker commercial ties among them. This can be illustrated with the example of exchange between Yautepec and western Morelos/Cuernavaca. Ceramics from each of these areas were never abundant in the other area, but the quantitative data suggest a decline of trade through time (Table 8.3). Teopanzolco imports at Yautepec declined between the MPC and LPC-A periods, whereas B4 polychromes stayed at the same low level in both the LPC-A and LPC-B periods. Similarly, Type B7 polychromes from Yautepec at the western sites declined between the two LPC periods. In the LPC-B period, regional market systems were smaller and appear to correspond to the boundaries of the large conquest states. The entire state may have comprised a larger regional system that integrated the various smaller regional systems, but this is difficult to establish. The joint development of a regional system with a conquest state is clearest for Cuauhnahuac. The zone of interaction as marked by ceramics expanded from a small local system in MPC times to a regional system in LPC-B times. This change in the regional configuration of ceramic complexes reflected the 177

Michael E. Smith

Table 8.3. Ceramic Exchange between Yautepec and Western Morelos. Site

MPC

LPC-A

LPC-B

1. Teopanzolco polychromes Yautepec

0.86

0.40

—

2. Western polychrome (Type B4) Yautepec —

0.04

0.04

3. Yautepec polychrome (Type B7) Capilco — Cuexcomate —

0.11 0.14

0.01 0.02

Note: Data are expressed as percentage of total sherds.

increasing political power and extent of the Cuauhnahuac polity in western Morelos. In a similar fashion the growing isolation between Yautepec and Cuauhnahuac, as revealed by the ceramic data, was almost certainly related to the growth of these two polities from local city-states into the heads of larger, competing conquest states (Hare 2004; M. E. Smith 1994). Patterns of Economic Change

The economic interpretations discussed previously point to the operation of a number of large-scale economic trends in Postclassic Morelos. The transition from the Middle Postclassic to the Late Postclassic A period was marked by two complementary trends. First, economic integration between regions within Morelos declined, whereas exchange with exterior areas increased (Figure 8.7). This probably reflects processes of political centralization and competition among the city-states of Morelos, as documented in Colonial chronicles. Competing polities exchanged ceramics and other goods at all time periods, but as competition increased, consumers and merchants may have turned increasingly to imports from distant areas—particularly the Basin of Mexico—for their decorated ceramics. I have argued elsewhere (M. E. Smith 2001, 2003a) that Morelos became drawn into the expanding Mesoamerican world system at this time (LPC-A), and this process would help explain the overall growth of longdistance exchange indicated by the ceramic data. The transition from the LPC-A to LPC-B periods was marked by further reductions of intra-Morelos ceramic exchange, coupled with declining exotic imports at most sites (Figure 8.7). Processes of political centralization and consolidation accelerated at this time. Within Morelos, the Cuauhnahuac polity expanded to cover most of western Morelos, a development probably related to the consolidation of local market areas, as shown in Figure 8.6. Formerly separate local systems merged into a larger system that probably corresponded to a small regional market system. The Aztec empire conquered Morelos, which may be one reason for the decline in imports from LPC-A to LPC-B times. The economic effects of Aztec imperial178

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

8.7. Changing frequencies of ceramic imports at Postclassic sites. Data are the means for the nine sites discussed in this chapter.

ism were diverse and included two contradictory processes (Berdan et al. 1996). On the one hand, the Mexica encouraged commercial exchange throughout the empire and carried out a series of actions to promote markets and merchants. On the other hand, the imposition of tribute in provincial areas had a negative effect on provincial well-being and may have contributed to lowered standards of living and a subsequent lowered demand for imports. Although imperial tribute was only a modest drain on provincial resources (M. E. Smith 1994), the Aztec policy of supporting local dynasties must have contributed in an indirect fashion to increased exploitation of provincial peoples at the hands of their traditional lords. These city-state kings probably increased their own tribute demands, secure in the knowledge that their position was bolstered by their cooperation with the empire. One way to monitor economic well-being uses a wealth index constructed from quantities of valuable goods (M. E. Smith 1987). I calculate such an index using the frequency of local decorated ceramics plus two times the frequency of imported ceramics.3 The results are shown in Figure 8.8. These data show two contrasting patterns. First, most sites experienced a steady decline in per capita wealth through time. Second, the two major capitals, Cuernavaca and Yautepec, show increases. The increase in wealth at the Palacio de Cortés site was quite dramatic, reflecting very high frequencies of decorated and imported ceramics during the LPC-B period when this was the location of the royal palace of Cuauhnahuac. 179

Michael E. Smith

8.8. Changing values of a wealth index for six Postclassic sites with ceramic collections from more than one period.

The ceramic data discussed here suggest that economic change in Postclassic Morelos was predominantly a local and a regional phenomenon. Processes of political growth and market system dynamics within Morelos created the patterns of archaeological ceramics presented in this chapter. This is not to deny a role for processes from outside Morelos. The effects of the Mesoamerican world system were felt in this area, and conquest by the Aztec empire certainly had an impact on local economic and political conditions. Nevertheless, one cannot argue that developments in the Basin of Mexico or elsewhere determined the fate of societies in Morelos. The dynamics of local and regional market systems and their political contexts were of primary importance. Conclusion

In this chapter I employed quantitative ceramic data to monitor the spatial configurations of local and regional systems among the three chronological phases of the Aztec period. The broadest trend is a general decline in regional economic integration through time. Local market systems persisted in most areas, and in western Morelos they increased in size, whereas the ties among separate local systems declined. The most dramatic example of the latter trend is the growing isolation of the Yautepec local system from local systems in western Morelos. 180

R e g i o n a l a n d L o c a l M a r k e t S y s t e ms i n A z t e c - P e r i o d M o r e l o s

It is no coincidence that these changes occurred at the same time the regional polities of Cuauhnahuac and Yautepec—conquest states or small empires—were growing in size and power. The most likely explanation is that administrative dynamics exerted strong influence on the spatial configuration of market system development in this area. The documentary record chronicles a situation of competition among Cuauhnahuac, Yautepec, and other large polities (including Tepoztlan) in the Late Postclassic period, while at the same time each was consolidating its control over subject city-states (M. E. Smith 1994). The declining ceramic exchange between western Morelos and Yautepec through time thus points to state dynamics and top-down processes as important elements in structuring regional market relationships. This analysis demonstrates that excavations across a region can provide ceramic data sufficiently fine-grained to model the extent and operation of hierarchical market systems. Although I have used documentary data to establish the existence of a hierarchical regional marketing system in Morelos, such sources are silent concerning the spatial configuration of market systems or the processes of change through time for exchange and marketing. My case study and the other chapters in this volume show that archaeological methods of economic analysis are now reaching the point where ancient economies can be reconstructed and analyzed in a detailed and rigorous fashion. Acknowledgments. I thank Barbara Stark and Chris Garraty for inviting me to prepare this contribution and for their helpful editing. The discussion of regional market systems is adapted from chapter 16 of Tlahuica Ceramics: The Aztec-Period Ceramics of Morelos, Mexico (M. E. Smith 2010). Archaeological fieldwork at Yautepec, Cuexcomate, and Capilco was supported by the National Science Foundation, Loyola University of Chicago, and the University at Albany (State University of New York). I thank Jorge Angulo Villaseñor, Raúl Arana, David C. Grove, and Kenneth G. Hirth for permission to study the Postclassic ceramics from their excavations. Cynthia Heath-Smith helped improve the prose and worked on some of the chapter’s graphics and tables. Notes 1. My discussion of economic and social organization in Postclassic Morelos is based on discussions by myself (1994, 2003a) and Druzo Maldonado Jiménez (1990), which contain more details and documentation. 2. This note has brief descriptions of the ceramic types listed in Table 8.2 and illustrated in Figure 8.2; the letters refer to labels in the table and figure (more complete descriptions can be found in M. E. Smith 2010). A: Wavy-line red-on-cream is a rare decorated type most likely from northeastern Guerrero/southwestern Morelos that occurs mainly in tripod bowls and tripod grater bowls. B: Red-on-buff is a rare residual category that contains a number of fragmentary sherds from bowls. C: Morelos-Puebla black-on-orange is a ceramic type originating in Tepoztlan and eastern Morelos whose forms (simple bowls and tripod plates) and 181

Michael E. Smith

decorations are identical to Aztec I black-on-orange but with local Morelos paste. D: Tepozteco bichrome is similar to Morelos-Puebla black-on-orange in form, paste, and designs, but the designs are painted in black on a background of white paint or slip. E: The category Teopanzolco types contains a variety of polychrome types associated with Teopanzolco and the Cuernavaca area; this is one of the most widely traded ceramic categories in Postclassic Morelos. The types only occur in simple bowl forms (conical and hemispherical). F: “Type C polychromes” is another composite category of bowls with several individual types of simple decoration; they most likely originate in eastern Morelos, the Yautepec Valley, or both. G: B4 polychrome is the most common Tlahuica polychrome type in Postclassic Morelos; its area of origin lies in the Cuauhna­huac polity of western Morelos. H: B7 polychrome is the most common polychrome type at Yautepec. I: B8 polychrome, which looks like a “sloppy” version of Type B7 (sloppy in form and decoration), also originated at Yautepec. J: guinda is a complex set of polished red ware ceramics, some of which were produced in Morelos and others of which were imported from the Basin of Mexico; it is the single most common decorated type at Postclassic sites in Morelos. K: Black-rim orange is a rare type found at most Postclassic sites in Morelos. 3. The frequency of imported ceramics is doubled in the wealth index because such pottery was presumably more costly relative to local wares.

182

Section Three

Comparative Contributions

Chapter Nine

Labor Taxes, Market Systems, and Urbanization in the Prehispanic Andes: A Comparative Perspective

Charles Stanish

The Inca empire represents one of the greatest political achievements in human history. By the end of the sixteenth century AD, it stretched over 1 million km2 and maintained at least nominal control over several million people (Figure 9.1). These subject populations were members of dozens of ethnic groups organized into a complex and heterogeneous state. The Inca established provinces in deserts, mountains, high-altitude plains, and forests. Their road system stretched over thousands of kilometers, and their engineers built bridges, fortresses, storehouses, and even entire towns for strategic purposes. Their priests climbed mountain peaks for elaborate rituals, feats unimagin­able a few generations earlier. Their armies marched for weeks and months in campaigns across the Andes. The organizational genius of the Inca state, known by its inhabitants as “Tawantinsuyu,” was unmatched in ancient South America and rivals the great empires of the premodern world in Asia, northern Africa, and Mesopotamia. In spite of the huge size and power of Tawantinsuyu, estimates of the size and population of the capital city of Cusco are small compared with the capitals of polities of similar size and complexity. The core of Cusco itself was perhaps no more than a few square kilometers in size. Within a 100-km2 area around the city, there were no more than 100,000 or so people at the height of empire; the core of the urban architectural zone was probably half that size (Agurto Calvo 1980; Hyslop 1984, 1990:64–65). A significant proportion of the people located away from the city center were likely fulltime farmers, an observation based on the absence of major secondary urban centers near the capital. 185

C h a r l e s S ta n i s h

9.1. The Andes showing Inca provinces and names mentioned in the text.

Analysis of demographic and geographic data from premodern cities around the world suggests that this difference is valid—Andean cities are smaller—an effect not caused by different archaeological methods or exogenous factors such as geography or ecology. I propose that the primary reason for this difference centers on the nature of the prehispanic Andean political economy—that is, how material wealth was produced, exchanged, consumed, and controlled. The economic basis of urbanism depends on the provisioning of nonagriculturists who can aggregate into a relatively 186

L a b o r T a x e s , M a r k e t S y st e ms , a n d U r b a n i z at i o n

small zone and draw from the agricultural sustaining area. This relationship is essential for successful urbanism, consistently and reliably transporting resources—particularly basic foodstuffs—from the sustaining area and beyond to a nonagricultural core. How this provisioning is accomplished is a function of the political-economic organization of the society. Marketplaces and fairs existed in the central Andes, but there is a noticeable lack of price-fixing markets in the Inca state. Instead, Andean political economies were generally based on elaborate redistributive, tribute, reciprocal, and administered trade systems understood within the broad theoretical traditions of economic anthropology and economic history. Local fairs flourished, and there was a brisk trade in many goods. Long-distance interregional exchange of many kinds of items was also robust and historically deep. María Rostworowski de Diez Canseco (1970, 1975) has demonstrated that substantial quantities of goods traveled up and down the Pacific Coast, produced by full- and part-time craft specialists. There is some evidence for pricefixing market exchange on the periphery of the empire, but the bulk of production and exchange in the state and imperial economies of the Andes did not rely on such exchange. It was within this cultural context that the Inca built their empire. Unlike their counterparts in Central Mexico, the people of the Andes created imperial systems based on an elaborate corvée labor–tax system, avoiding or possibly suppressing market trade. The Inca were unusual in the history of premodern empires in eschewing market systems for corvée mechanisms, but they were successful nonetheless in conquering their known world in just over a century. This system was a brilliant solution in this context, and it was better than market and tribute systems for many tasks— most notably, raising troops for the military. But such a system had costs as well. In this chapter I argue that the lack of extensive price-fixing market systems in the Inca state precluded the concentration of large numbers of peoples in urban areas, a direct result of the relatively large transaction costs in operating their imperial economy (e.g., North 1981, 1990; Williamson 1975). Central Mexico and the Central Andes: A Contrast in Imperial Economic Organization

In 1519 the army of Hernán Cortés entered Mexico and in the course of a few months militarily conquered the Aztec empire. Almost immediately, Spanish observers began to record their views of the Aztec economy, political structure, religion, and other aspects of life. A central theme of these documents is the complex market system operating in Central Mexico in the early sixteenth century. Donald Kurtz (1974:689) noted that “[l]arge markets existed in both Tlatelolco and Tenochtitlan. . . . The great market in Tlatelolco served as many as sixty thousand persons daily.” He added that Texcoco and Tlaxcala had huge marketplaces as well. As Christopher Garraty aptly notes in Chapter 1, the Spaniards “appear to have had little trouble recognizing what 187

C h a r l e s S ta n i s h

was occurring in the large, bustling daily marketplace at Tenochtitlan-Tlatelolco, the largest and most powerful of the Aztec imperial capitals.” Numerous independent observers have described marketplaces for a variety of cities. Regional fairs or marketplaces with regular meeting times were commonplace. These chroniclers described elaborate legal systems to adjudicate disputes, the existence of private merchants, private property, and long-distance traders. They even described media of exchange, such as cacao, feathers, and copper objects. The documents also tell us how artisans were paid: “Montezuma richly rewarded the stone masons and stone cutters [with] blankets, loads of salt, ten loads of fruit, beans, bales of chile, a canoe of maize, two loads of cacao and cotton” (Tezozomoc 1878 [1598]:662–668, cited in Kurtz 1974:692). These items, in turn, were used to purchase goods and services at marketplaces. In short, the earliest historical documents from Central Mexico describe a macro-regional exchange system almost immediately recognizable to the Europeans. Central to this political economy was the institution of the market in which people produced, bought, and sold products. In this system, prices were created by supply and demand. Even labor was sold for convertible commodities. Tribute from the provinces flowed through exchange mechanisms that long predated the rise of Tenochtitlan. As Michael Smith and Frances Berdan (1992:356) have noted: “The incorporation of these [tributary] regions into the empire opened large areas for reliable trade and market networks, again facilitating a sustained and quite predictable flow of goods to the imperial cities. Indeed, in some cases tribute demands encouraged increased trading, in situations where tribute goods entered a province through long-established trade and market networks.” A dozen years after the fall of Tenochtitlan, Spanish conquerors from the same class and background moved against the Inca empire. In a pattern strikingly similar to that of Cortés’s victory, Francisco Pizarro and his army famously conquered the Inca forces at the town and military camp of Cajamarca in the central sierras. Within two years they had established strategic control of the empire and within thirty years consolidated their victory by eliminating any potential military threat from the old regime. Like their counterparts twelve years earlier, they described the economy, political structure, religion, and other aspects of Andean life. Yet in the intervening months and years following the initial conquest, something quite different from the situation in Central Mexico developed in the Andean historical literature—markets and marketplaces were barely mentioned. We have no descriptions of large marketplaces, few descriptions of independent traders, and no discussion of media of exchange, market equivalencies, and the like. Regional fairs were occasionally mentioned, but compared with Central Mexico such remarks are incidental and almost invisible. The Inca economy was enormous and profoundly complex. We do get descriptions of massive production and movement of goods across the landscape. Craft specialization and exchange were cornerstones of both the local and imperial economies, but this economic activity took place in a very different way from that of the Aztec 188

L a b o r T a x e s , M a r k e t S y st e ms , a n d U r b a n i z at i o n

empire. The Inca, like their counterparts in Central Mexico, were able to extract an impressive amount of resources from their provinces. In place of the institution of markets and complex tribute rolls in the Andes is the theme of labor taxation, known as corvée in the Western feudal world—unpaid labor conscripted on a regular basis by a king or other political authority. The equivalent for corvée in the Andes was known as the mit’a, an indigenous term meaning “turn,” in the sense of fulfilling an obligation. That is, in each community every taxpayer had to take his or her turn in fulfilling the labor tax imposed by the authorities. A sixteenth-century dictionary of Aymara, the second language of the Inca realm (along with Quechua), describes a mittani as “the obligated one that does his turn for community matters.”1 In the same way the market institution dominated the observations of the Europeans in Central Mexico, the institution of the mit’a dominated the literature of the Andean peoples. Writers from every political and cultural persuasion independently describe the elaborate labor arrangements for producing everything from buildings to maize beer, for supplying the army with troops, and for filling the ubiquitous storehouses (tampu) that lined the great road system. The early chronicler Juan Polo de Ondegardo (1916 [1575]), “one of the best-informed administrators . . . wrote repeatedly that no one under the Incas was forced to contribute anything from their own personal estate, that subjects contributed only their labor” ( Julien 1988:261). Catherine Julien (1988:264), echoing a generation of Andean scholars, is emphatic about this point: “The Inca system of exactions was unlike the Spanish system in that all that was assessed from local people was their labor. . . . Products might be elaborated with this labor donation, but the resources that were converted into products were held by the state.” There is some evidence of pre-European marketplace exchange in the northern boundaries of the Inca empire. John Murra (1995:62) has noted that some Colonial chroniclers describe a marketplace in Quito with the Nahuatl term tiangues, a case in which the exception perhaps proves the rule. An institution like the pochteca also existed in the north, known there as mindala, but it was not widespread in the rest of the Andes. The historian Rostworowski de Diez Canseco (1970, 1975) has argued for mercaderes on the south coast who traded up and down the Pacific Ocean from Quito into the Collao, the southern quarter of Collasuyu and home of the Aymara and other non–Quechua-speaking peoples. In a 1612 Aymara-Spanish dictionary by Ludovico Bertonio, it is clear that a distinction exists between European and indigenous “modes” of marketing. In Book I (Spanish to Aymara), one entry is “mercader a nuestro modo: mircatori ves Tintani,” and immediately below this entry is “mercader a modo de indios, Hanrucu, Alasiri.” The word for the former is obviously an Aymarization of the Spanish “mercado” (Aymara today still occasionally pronounces the letter “e” like the letter “i.”) In other words, two generations after the conquest there was no indigenous word for market in the European sense; to express the concept, they had to borrow from the Spanish. These two modest dictionary entries comprise one of the earliest anthropological descriptions on record for Andean economies: two cultures, coexisting in the same landscape, 189

C h a r l e s S ta n i s h

structured their economic lives in such different ways that a different vocabulary was needed to describe and understand them. In the same way the Aztecs constructed their imperial economy in a historical context that included indigenous markets, the Inca created their economy in an Andean context in which price-fixing markets were largely absent. The mit’a can be seen as a logical outgrowth of this historical context. It required an elaborate bureaucracy, and the entire population of taxpayers throughout the empire was theoretically ranked into a decimal organization used to extract that labor. Taxpayers belonged to a nested hierarchy of 10,000, 5,000, 1,000, 500, 100, 50, and 10 people. At the head of each of these groups was a tax-exempt official. When central government personnel needed goods or services, they assigned a number of laborers to that task, and the various administrators called on each level below them to collect the necessary labor. While the mit’a laborers were away fulfilling their tax, the other members of the community were required to work their fields and otherwise maintain their households. At first glance, there may seem to be little difference in directly taxing labor or taxing goods because both policies can be reduced to increased labor inputs by tribute-paying populations. But in fact the difference was profound. If the Aztec emperor needed deer, he taxed a province to provide the animals. If the Inca emperor needed deer, he ordered his administration to find hunters with unfulfilled mit’a obligations to conduct a royal hunt. To collect maize, no set tribute demands were given to individual villages or villagers. Rather, administrators called for the appropriate number of workers to till the imperial lands set aside for state or church use. The Aztec emperor could employ labor drafts (Hicks 1984) but could also pay skilled workers with various products or media of exchange. When the Inca emperor required something as elaborate, he called on his administrators to draft the appropriate number of skilled workers to fulfill their labor tax to execute his orders. From an economic perspective, both drew off of labor from the commoner population, and both had similar costs because they tapped a limited resource (tribute versus obligated labor). Although the end result was the same, the process by which the two empires collected resources was worlds apart and had a profound effect on the imperial political economy and the nature of urbanization in the two imperial domains. Beyond the Antimarket and Anti-Antimarket Divide

A century of economic anthropology and economic history has provided us with a set of theoretical tools to understand an economy such as that of the Inca. Beginning with Bronislaw Malinowski, three generations of scholars have helped us define the various ways by which people produce and exchange the goods and services needed to reproduce their biological and social lives. Starting in the late 1950s, the so-called substantivist versus formalist debate raged for over twenty years. Echoing perhaps a sentiment that pervades many of the chapters in this volume, I believe the substantivists set economic anthropology back two decades with their ideologically driven the190

L a b o r T a x e s , M a r k e t S y st e ms , a n d U r b a n i z at i o n

ory and interpretations. In the words of Richard Blanton and Lane Fargher (Chapter 10), they led many to “an unfortunate and mistaken anthropological consensus that rational action and market exchange could not have been important aspects of premodern complex societies.” Indeed, the substantivists based their views on a rejection of Western economics and its philosophical underpinnings of Homo economicus, preferring instead to reify nineteenth-century utopian views of non-Western peoples. As Blanton and Fargher further point out, this theory was developed within politically charged frameworks that maintained an outdated evolutionary model of simple progress, with reciprocal exchange replaced by redistribution, in turn replaced by market exchange. The inherent Spencerian/Engels evolutionary underpinnings and the idea that non-Western peoples lacked “rational” instincts were two great flaws of substantivist theory. The modern empirical observation that “market exchange and markets [existed] prior to the advent of Western capitalism” (Blanton and Fargher, Chapter 10) is now, in my opinion, effectively beyond dispute. Although I agree with most scholars in this volume that the substantivist approach was deeply flawed, I argue that there has been an overreaction to this school, and some archaeologists have been too quick to embrace the canonical model embodied in the concept of H. economicus and neoclassical Western economic theory. The problem begins with the word “rational.” Semantically, particularly in European languages, rational has positive moral and cognitive connotations, while irrational has the opposite connotation. When used in the colloquial sense, rational behavior is simply that which most effectively allows one to achieve a particular goal. Calling somebody irrational implies that the person lacks some basic human cognitive capacity. Using the colloquial sense for an entire culture is even worse and represents the epitome of uninformed ethnocentrism. In the technical-economic sense of the term, as used in the canonical model of H. economicus, the concept of rationality is quite different and much narrower in meaning: “This orthodox conception defines economic rationality by maximization of exclusively materialist objectives, namely profit by producers and utility by consumers” (Zafirovski 2003:1). In the canonical model, actors are ego-directed (as opposed to socially motivated) and maximize utility. The overreaction to the excesses of substantivism has led to a view that sometimes sees markets and economically rational agents in almost every culture. This overreaction is curious in light of the fact that some of the latest research in game theory, evolutionary game theory, and the relatively new branch of behavioral economics directly challenges the assumptions of rationality inherent in the canonical model of classical Western economics, embodied in the work of such classic scholars as Adam Smith, John Stuart Mill, David Ricardo, and so forth. In the words of Joseph Henrich and colleagues (2001:73), this new work has “uncovered large, consistent deviations from the textbook representation of Homo economicus” in both Western and non-Western societies.2 Concepts such as “bounded rationality,” “irrational prosocial behavior,” “strong reciprocity” (Gintis 2000:311), “altruistic punishment” (Egas 191

C h a r l e s S ta n i s h

and Riedl 2008), “satisficing behavior,” and “conditional cooperation” (Shalizi 1999) indicate that human beings of all cultures—both in the West and elsewhere in the world—often act irrationally in the narrow, technical sense of the term. Models based on these noncanonical premises actually explain and predict human behavior better than those based on assumptions of rational, utility-maximizing agents. In economic anthropology, some of the most exciting research comes from experimental games, such as the Ultimatum Game or Iterated Prisoner’s Dilemma conducted on both Western and non-Western subjects. Henrich and colleagues (2001) report on their study of fifteen small societies. In all cases, none conformed to the expectations of H. economicus. (There is such a vast literature on this topic that it is far too extensive to summarize here.) What is important for this discussion is that this body of work has provided a new model about the normative social behavior of humans called “Homo reciprocans” (Bowles and Gintis 2002). Unlike the H. economicus model, people in this model have an inclination toward social cooperation and will act “irrationally” and at cost to themselves. The H. reciprocans model accepts the underlying principle of humans as ultimately ego-directed but adds that, based on experimental observations and theoretical deductions, individuals acting in their own self-interest can create evolutionarily stable and complex cooperative systems. The discovery of these kinds of “emergent” properties in human societies along with the recognition that every society has some people willing to engage in “irrational, pro-social” behavior has, in my view, revolutionized our understanding of human sociality and, by implication, of economic anthropology and economic history. The canonical model of the economically rational actor, amorally and impersonally maximizing utilities, is as much a Western myth of the twentieth century as the “noble savage” myth is of the nineteenth century. If people are not rational in the technical sense, then what are they? Game theory research shows that people are adaptive in an evolutionary sense and, overall, fairly rational in the colloquial one. One of the flaws of rational actor theory is the assumption that people have full access to all relevant and unbiased information with which to make decisions. In reality, people—individually and collectively—do not make consistently rational decisions as much as they make well-informed ones based on the information at hand, the history of interaction, and the cultural filters by which that information is evaluated. Robert Axelrod (1997:14) described it well: “In complex situations, individuals are not fully able to analyze the situation and calculate their optimal strategy. Instead, they can be expected to adapt their strategy over time based upon what has been effective and what has not.” This is the key: individuals make decisions based on the best information available, but that information is always flawed and incomplete. The recent research teaches us that the old Hobbes-Rousseau dichotomy (Shalizi 1999) and its equivalent in economic anthropology—the substantivist-formalist divide—are dead. Gone is the dichotomy about whether people are economically rational, ego-directed agents acting in their own self-interest through markets or are 192

L a b o r T a x e s , M a r k e t S y st e ms , a n d U r b a n i z at i o n

utopian agents acting for the social or moral good through embedded institutions. Humans are adaptive, conditional cooperators—the H. reciprocans model—that, under the appropriate circumstances and in their own self-interest, can behave in intensely pro-social, cooperative ways. Just as the canonical model of Homo economicus is not universal, neither is pricefixing market behavior. People around the world today and throughout history have used a wide variety of mechanisms to structure their economic life in a way that can be understood as rational in the colloquial sense. Ethnography, evolutionary game theory, and allied disciplines give archaeologists the conceptual tools to understand a very broad range of human behavior, of which market exchange and other kinds of exchange are on a continuum of successful human strategies (see Hirth, Chapter 11). Allocation Mechanisms for Economic Exchange

As Garraty (Chapter 1) remarks, “The [market] concept refers not only to the system of economic exchange and provisioning but also to the social and political contexts of those exchanges” (see also, Bestor 2001; Plattner 1989b). It is clear from the ethnographic record that settled societies have opportunities in which people who produce or acquire goods meet and exchange those goods. It is also a reasonable deduction that the individuals at these meeting spots are ego-directed, adaptive agents acting ultimately in their own self-interest in the sense of Homo reciprocans. Nevertheless, occasions for barter do not necessarily lead to marketplaces and market systems. Other exchange mechanisms may be prominent. Importantly, “[R]ational, utility-­maximizing choices of the sort found in Polanyi’s ‘market mentality’ may exist comfortably side by side with other kinds of social actions and perhaps mentalities in premodern states” (Blanton and Fargher, Chapter 10). Of the many errors made by the substantivists, a major one was confusing an “allocation mechanism” with an “integrative institution.” Substantivists took the observations of many ethnographers regarding reciprocity and those of Malinowski and Firth regarding redistribution and retooled them into concepts about institutions that structured many non-modern societies. They then arranged them into a kind of pre-Darwinian evolutionary framework that presupposed concepts of progress and inherent evolutionary change. This conceptualization fit well with a simplistic evolutionary process in which societies went from reciprocal through redistributive to market societies, but it failed the scrutiny of empirical research. In turn, an overreactionary error the antisubstantivists made was to argue that any use of these concepts inherently results in the construction of an evolutionary sequence of “integrative mechanisms” of society. An allocation mechanism is simply a means by which people transfer goods or services. We now can show that many of the mechanisms discovered by economic anthropologists from Malinowski to the present are powerful ways to understand exchange. 193

C h a r l e s S ta n i s h

As Gary Feinman and Linda Nicholas discuss (Chapter 4), these categories—redistribution, reciprocity, non-market trade, and price-fixing market exchange—are best seen as distinct types of allocation mechanisms that coexist in most societies. It is not, as Barbara Stark and Garraty (Chapter 2) note, simply an issue of the presence or absence of markets. Along with these traditionally recognized mechanisms, we can add tribute and competitive feasting (Stanish 2003). Each allocation mechanism involves different transaction costs and benefits, none is morally superior, and almost all societies combine these mechanisms to varying degrees. They are all subject to supply and demand forces. Reciprocity involves an equal exchange between two or more producers—either directly or through mediators, either immediate or deferred, as in the continual discharge of ritualized social debts. Redistribution involves the collection of wealth by a central political authority generally mobilizing goods for the purpose of status enhancement—often involving elite, exotic, or other nonsubsistence items (Earle 1997)—rather than entailing a leader’s role as an intermediary, handling diverse products that go to commoner households. In both reciprocity and redistribution, exchange values of goods and services are assigned informally, with supply, demand, and labor input the key variables. In administered trade, exchange value is assigned in a more formal and standardized manner by a political or social authority. These values are periodically renegotiated, and an implicit or explicit social or political “contract,” or both, provides for the provisioning of goods and services between exchange partners. As with all economic transactions, supply and demand affect the relative value of the traded goods, but exchange values are essentially fixed by political bodies. The exchange obligations are intimately bound with non-economic considerations. Price-fixing market exchange developed as a means of moving goods and providing services in some of the world’s earliest states, usually with media of exchange. A price-fixing market mechanism exists when the value of a good or a labor service is determined by negotiations between traders without substantial interference from an authority. An essential characteristic of price-fixing market exchange is the existence of traders who earn a profit from price differentials between buyers and sellers (Hodges 1988). In market allocation systems, merchants make their profit by moving goods with different utility functions to different places and exchanging them accordingly. Direct producers can likewise sell goods in an established marketplace or at fairs, acting as their own traders. Political economies in which some or many items were distributed through price-fixing market mechanisms developed independently in at least the Near East, northern Africa, and Mesoamerica. Competitive feasting is an exchange mechanism that in strictly economic terms amounts to a form of deferred reciprocity. One person offers goods to another with the expectation of a supposedly equal exchange in the future. Unlike most forms of reciprocity, however, the motive in competitive feasting is not the future receipt of an equal return of wealth but rather a nonmaterial “profit” of political gain. Perhaps 194

L a b o r T a x e s , M a r k e t S y st e ms , a n d U r b a n i z at i o n

more than in any other exchange mechanism, political and economic motives merge in competitive feasting. Wealth is provided strategically to obligate the receiver to such an extent that he or she cannot reciprocate adequately. Competitive feasting is a major form of economic exchange in many nonindustrial societies (Hayden 1996:127). In the Inca empire, formalized feasting constituted an important allocation mechanism for sustaining the urban-rural dependency. The final mechanism of wealth transfer is tribute, an exploitative economic relationship in which one party benefits materially by extracting wealth or labor from another through some kind of coercive political relationship. The mechanism of tribute is an integral component of archaic states and imperial political economies. It takes many forms in a variety of historical and cultural contexts, including feudal rent, serfdom, chattel slavery, taxation, labor taxes (corvée), and the like. Using these concepts, we can examine one of the most important issues for urban places in any state or empire. Because the provisioning of cities is one of the most critical aspects of an imperial economy, I argue that the nature of that economy is intimately linked to the nature and size of those cities. The Inca Economy and Urban Growth

Tawantinsuyu was the largest and most complex state that developed in the Americas, in terms of both total population under its administrative control and total territory within its provincial structure. Tawantinsuyu means “land of the four quarters.” The quarters, or suyus, were political and geographical regions to the north (Chinchasuyu), east (Antisuyu), southwest (Contisuyu), and south (Collasuyu) (Figure 9.1). Inca organizational genius was manifest in the ability of the state to mobilize large numbers of people for specific tasks. At the time of the Spanish conquest, factions within the empire had several armies of 40,000–80,000 troops each (Hemming: 1970:36).3 With the support personnel and camp followers, the combined total of each army approached six figures, a huge number by premodern standards anywhere on the globe. The capital of the empire, Cusco, was a planned city built in a style that represented the zenith of Inca architecture. The core of the city was filled with buildings constructed of massive, shaped stone blocks, representing a huge investment of labor. Above the city stood the enormous Sacsahuaman, a fortress and ceremonial building. Historical documents suggest that 20,000 laborers were required to construct this sprawling complex. The grid pattern of the city core was replete with temples, palaces, storehouses, and residences for the elite and retainer populations. As remarked, the best population estimate for the entire Cusco valley is around 100,000 (Bauer 2004), and the core architectural settlement, Cusco, contained perhaps half that number. Cusco was therefore small compared with other premodern, nonindustrialized imperial capitals. Cities such as Rome, fourth-century BC Athens, fifteenth-century AD Vijayanagara in southern India, the capital of the Tang Dynasty in China, and 195

C h a r l e s S ta n i s h

Cusco’s New World counterpart in Central Mexico, Tenochtitlan, were significantly larger. Even earlier first-generation states, such as Mohenjo-Daro in the Indus valley and Teotihuacan in Central Mexico, were larger. The population estimate for Cusco is smaller than those of several Classic Maya centers, many Postclassic Mesoamerican ones, and numerous other state centers throughout the premodern Old World. In short, the population of the capital of the Inca empire was no larger (or even smaller) than the capitals of many much less complex polities, including Mohenjo-Daro, which some scholars claim was not even a state (e.g., Possehl 1996).4 Methodological Considerations: Is the Difference Real?

The huge literature on cities and urbanization contains a number of anthropological, sociological, and historical definitions (e.g., Blanton 1981; Branigan 2002; Fox 1977; Marcus 1983; Marcus and Sabloff 2008; Sjoberg 1960; M. L. Smith 2003; Wheatley 1972). Some scholars argue for a very broad definition of urban, including the Andeanist John Rowe (1963:3), who considers urban settlement as “an area of human habitation in which many dwellings are grouped closely together.” For Rowe, a small city had fewer than 2,000 inhabitants. This urban size is common in the Aegean, where settlements as small as 1,000 have been considered “cities” (Hansen 2008). Most other scholars, however, take a more restricted view of what constitutes a city (e.g., Marcus 1983:240–241). By these definitions a population of 2,000 would barely constitute a small town, and a population of primarily agriculturalists of fewer than 2,000 would not be considered a city. William Sanders and David Webster (1988:523) define a city as simply a central place in which social activities are concentrated. Such a broad definition is useful because it encompasses a wide range of cultural and historical circumstances, including most non-Western, nonindustrialized ones. Sanders and Webster, following Richard Fox (1977) to a large degree, adopt other criteria as well, including the centralization of administrative functions, the existence of an economic elite, and so forth. This point was elaborated by Joyce Marcus (1983:239), who noted that Mesoamerican cities had a relatively high density of ritual architecture, indicating that they functioned as the capitals of regional religious hierarchies. Urbanization is therefore defined as the permanent aggregation of people into a central place for economic, social, or ritual activities, or some combination. Perma­ nence of residence is a key factor. A significant percentage of these individuals must be economic or ritual specialists or both. Significant and contiguous architectural constructions preclude the growing of major food resources in the city core. That is, urban space is utilized largely for nonagricultural activities. Rural dependency is a key attribute of cities. Urban areas must normally rely on some kind of exchange mechanism to provision the resident population. An agricultural zone is, at least in theory, self-sufficient for the provisioning of foodstuffs, but an urban area is not. Rural, agricultural sustaining areas surround premodern cities. 196

L a b o r T a x e s , M a r k e t S y st e ms , a n d U r b a n i z at i o n

Population Estimates

Estimating the total demographic size of cities is central to this analysis. Calculating absolute population estimates for archaeological sites is an enormously difficult task (e.g., Kolb 1985 and subsequent comments; Read 1978; G. Storey 1997). Such analyses require a number of admittedly questionable assumptions, which are necessary whether one uses total site area, archaeological house counts, contemporary censuses, other historical data, or any combination of these. Documents occasionally lie, household composition shifts through time, nonhabitation areas are constructed over earlier houses, and the functions of buildings are both difficult to assess and change over time. Finally, cities are not static entities; populations constantly ebb and flow (G. Storey 1992). Research on early urban areas around the world is spotty at best. Perhaps most significant, methods and research interests vary among archaeologists who work on ancient cities. There is no uniform means of inferring site size and population or even of defining the meaning of “urban” for comparative research. In some cases the entire area of the site and contiguous settlement, including the agricultural sustaining area, is included in an urban site size definition (e.g., the 120 km2 for Tikal or 100 km2 for Chang’an). In other cases archaeologists include just the nonagricultural architecture of the core and omit the adjacent populations. In spite of these problems affecting many important questions, we have to estimate city sizes. I agree with Glenn Storey (1997), who noted that although the absolute population size of cities is difficult to infer, we can correctly determine the scale of a population in many urban areas. That is, although scholars may argue whether the population of imperial Rome was 400,000 or 1 million, we can certainly conclude that it was an order of magnitude larger than that of imperial Cusco, regardless of any correction factors utilized. We are slowly accumulating a body of evidence on the areal extent and population size of premodern cities (G. Storey 1992). In a number of cases, detailed maps have been made of the sites. In other cases, settlement surveys have been conducted in and around urban areas that permit a reconstruction of the site’s physical size and population. For the urban areas of classical antiquity, we have documentary evidence. Likewise, for many settlements in Latin America that existed at the time of the Spanish conquest, censuses were conducted in the first generations of colonial occupation. In spite of the many problems, enough data are available to allow us to make meaningful, albeit fairly gross, comparisons among urban sites. In the 1960s and 1970s William Sanders, Jeffrey Parsons, Thomas Charlton, and Richard Diehl, among others, used modern analogies to estimate population sizes in the Valley of Mexico. Their work (in particular see Charlton 1972; Sanders, Parsons, and Santley 1979) defined a continuous range of settlement densities from a low of 200 per km2 up to 13,000 per km2. The low end of this range was associated with rural hamlets, and the high end was associated with “high-density compact villages” (2,500–5,000 per km2) and the extremely dense sites located next to intensive raised 197

C h a r l e s S ta n i s h

fields known as chinampas. More recently, Storey surveyed preindustrial cities from around the ancient world and found that many city size estimates for the ancient Old World have been exaggerated. He estimated the population of Rome at around 400,000 instead of the often cited 1 million. His survey of preindustrial cities from the Old World indicates that the overall densities ranged from 10,000 to 20,000 per km2 (G. Storey 1992:175) and were rarely higher. A significant conclusion from his data is that the average density of preindustrial cities in five distinct cultural areas—Europe, Africa, West Asia, India, and East Asia—indicates that both the mean and the median were fairly consistent in space and time. Means ranged from 12,600 to 21,027 per km2, and the medians ranged from 10,000 to 15,000 per km2. As he described it, “[T]he phenomenon of preindustrial urbanism . . . is uniform throughout the world” (G. Storey 1992:175). This being the case, we can use these data for meaningful comparisons with other areas, such as the Andes and Mesoamerica. With caveats, Table 9.1 presents data for some cities in the premodern world, with the estimated area of the urban settlement and a population estimate range. I attempted to assess the literature on these estimates and provide either an average of reliable estimates or a consensus midpoint based on newer analyses or data. Occasionally, I chose Table 9.1. Estimated Area and Populations of Selected Urban Centers in the Premodern World. City

Area (km2)

Estimated Population/ Range

Population Density (per km2)

2 million

20,000

Chang-an (Tang Dynasty)

ca. 100

Imperial Rome

13.86

435,245–1 million

31,403–72,150

Constantinople (fifth to sixth centuries AD)

18.3

500,000

27,322

—

400,000

—

20 (+300)

250,000