What Gifts Engender: Social Relations and Politics in Mendi, Highland Papua New Guinea

What gifts engender

What gifts engender Social relations and politics in Mendi, Highland Papua New Guinea RENA LEDERMAN Department of Anthropology, Princeton University

TA« right of the University of Cambridge to print and sell all manner of books was granted by Henry VIII in 1534. The University has printed and published continuously since 1584.

CAMBRIDGE UNIVERSITY PRESS Cambridge London New York New Rochelle Melbourne Sydney

CAMBRIDGE UNIVERSITY PRESS Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo, Delhi Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521104999 © Cambridge University Press 1986 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 1986 This digitally printed version 2009 A catalogue record for this publication is available from the British Library Library of Congress Cataloguing in Publication data Lederman, Rena. What gifts engender. Bibliography: p. Includes index. 1. Mendi (Papua New Guinea people) 2. Papua New Guinea - Social life and customs. I. Title. DU740.42.L43 1986 306'.0899912 85-21283 ISBN 978-0-521-26713-7 hardback ISBN 978-0-521-10499-9 paperback

Contents

List of tables, figures, and maps Preface Acknowledgments 1. Mendi coming into view Introduction The region and the population: background The region and the population: the social dynamics of land use The "subsistence bias" and another approach to Highland political economies 2. Sent relations: solidarity and its limits Introduction Individual action, collective responsibility Individual autonomy, political participation, and leadership Social relations between clan members Exogamy, antigamy, and the sem-twem distinction The structure of intergroup relations Names, categories, and the reality of groups Clan membership and the exclusion of women 3. Twem: personal exchange partnerships The significance of the Southern Highlands The articulation of twem and sem Twem and the life cycle Persons and things Twem etiquette The structure of exchange networks

page vii ix xi 1 1 5 10 14 19 19 21 26 30 37 40 52 55 62 62 65 70 81 89 103

vi

Contents

4. Gender ideology and the politics of exchange Introduction Women and exchange Comparing women and men Twem, female participation, and its limits Sent, gender hierarchy, and its limits

117 117 119 128 132 135

5. Twem and sent in context Inequality, reciprocity, and incremental gifts Prestations of marriage and death Large-scale ceremonies

141 142 154 162

6. Sai le at Senkere: the politics of a Pig Festival Introduction Anthropological background The Mendi Pig Festival Outline of Suolol political history Record of events The sai le, Highland Pig Festivals, and the structure of exchange

174 174 175 178 182 187

7. "Development" in Mendi Introduction Twem, sem, and indigenous development Perspectives on changing times

213 213 219 227

Appendix A. The research community Appendix B. The "accounts sample" and some comments on research methodology Notes Glossary References Index

238

200

243 255 273 276 283

Tables, figures, and maps

Tables 1.1 2.1 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 4.1 5.1 5.2 5.3 vii

Mendi District census divisions, land and population (1976) Clan organization in the Senkere community Never-married men: access to valuables and labor Contexts of first transactions among 43 Senkere men and women Reasons for terminating exchange partnerships among 43 Senkere men and women "Active" exchange partners and agnatic supporters of 43 men and women Total exchange networks of 43 men and women "Active" partnerships in relation to total exchange networks of 43 men and women Structure of the networks of 23 men Structure of the networks of 20 women Structure of terminated partnerships of 23 men Structure of terminated partnerships of 20 women Terminated partnerships in relation to total networks of 23 men Terminated partnerships in relation to total networks of 20 women Overlap in the exchange networks of husbands and wives ("active" partnerships) Frequency of marriages in Senkere Frequency of major deaths in Senkere Redistribution of Ol Egar ol ombu\ group gift

page 10 24 72 73 97 106 107 108 110 111 113 114 115 116 127 155 155 166

viii 6.1 6.2 7.1 7.2 7.3 A.I A.2 B.I B.2 B.3 B.4 B.5 B.6 B.7 B.8 B.9 B.10

Tables, figures, and maps Lengths of time sai le pigs were held before the December 1979 pig kill Sai le pigs: sources and repayments Comparison of pig herds: 1978, 1979, 1983 Sources of 1978 pigs Sources of 1983 pigs Village census information: Waparaga, 1979 Waparaga households, January-March 1978 Characteristics of the "accounts sample": 23 men Characteristics of the "accounts sample": 20 women Households included in "accounts sample" Details of active exchange partnerships and total networks of 23 men Details of active exchange partnerships and total networks of 20 women Details of the structure of the networks: 23 men Details of the structure of the networks: 20 women Details of the structure of terminated partnerships: 23 men Details of the structure of terminated partnerships: 20 women Comparison of survey households: 1978

204 210 221 222 223 239 240 244 244 244 245 246 247 248 249 250 253

Figures 2.1 2.2 2.3 4.1 4.2 5.1 5.2 5.3

Genealogical sketch of the Senkere community Pulumsem genealogies: Tonkpisem Pulumsem genealogies: Punginsem, Pulumsem, and Tonkpisem Alcome's pearl shell Paki's pig Direction of nopae payments The accumulation of gift-debts and gift-credits 01 Egar ol ombul, April 1978

42 49 51 133 133 152 159 168

Maps 1.1 1.2 2.1

Southern Highlands Province The Mendi Valley The Senkere community

6 8 38

Preface

This book, a description and argument concerning social relations and politics in Mendi, is meant as a contribution to the comparative ethnography of Highland Papua New Guinea. In it, I attempt to demonstrate that Mendi culture and history are structured in terms of two principles of relationship: twem and sem. Whereas sem relations create bounded groups ("clans") with an existence independent of particular individuals, twem relations generate egocentered, unbounded networks. Clans regulate access to land and are the primary organizational basis for male cooperation in ceremonial exchanges, whereas networks regulate access to fluid resources such as pigs, pearl shells, and money. While clan prestations require that the flow of wealth be dammed periodically, twem etiquette encourages the constant circulation of wealth. While clans are implicitly hierarchical, being exclusively male, twem networks are egalitarian and broadly inclusive, men and women, unmarried people, and old widows and widowers all participating in them. On the basis of these and other contrasts developed in the chapters to follow, I will argue against the common assumption that network relationships are simply a means by which individuals accumulate valuables for display during clan festivals. I will argue instead that twem relations constitute a kind of sociality distinct from clanship; moreover, they involve an ethic of exchange contradictory to the ethic of clan solidarity. Although network and clan obligations may be fulfilled simultaneously (and are, in a sense, necessary to one another), they also sometimes conflict. This structural problematic is realized in the tensions one may observe between men and women, on the one hand, and between leaders (big-men) and ordinary men, on the other. It is also expressed in a political rhetoric of gender meanings, so much a part of both everyday and public discourse. That is, the contradictory implications of twem and sem relations make certain important aspects of Mendi politics intelligible, ix

x

Preface

The general purpose of the analysis presented here is to make Mendi history - by which I mean the insider's sense of the significance of events accessible to outsiders; at least, I hope to have made an initial stab at that. The historical focus is on one Upper Mendi political community's Pig Festival (mok ink), a sequence of events similar in some respects to ceremonies that are well known in the anthropological literature on the Highlands. That literature has emphasized male leaders and formal ceremonial prestations of wealth. This study rights the balance by paying attention to ordinary men and women, as well as to the wealth transactions that take place in everyday settings, the frame of reference within which the structurally significant character of twem relations becomes visible. As the foregoing implies, I hope this study will contribute to Highland ethnography and to anthropology generally in a number of ways. It has something to say relevant to an anthropology of gender (as a study of male-female relationships and of gender ideology in a Highland society that is somewhat atypical in those respects) and to an anthropology of power and politics (as a study of the articulation of hierarchizing and equalizing structures, and of the social force both of words and of culturally constituted things). It also has something to say relevant both to an anthropology of history and, within that, to what might be termed a political economy of the "gift." By this I mean that gift exchange in Mendi and elsewhere engenders a type of sociality with differentiated forms of its own and a distinctive historical dynamic, and with particular implications for the structuring of people's understandings of, and practical action in, their world. Studies of gift relations, therefore, are not an antiquarian concern but have a direct bearing on our understanding of local resistance, both explicit and tacit, to the world of the market. In other words, and as the concluding chapter of this work suggests, a study of gift exchange in Mendi illuminates not only the shape of a given world but also the outlines of a world being made.

Acknowledgments

I write the following words knowing that as the names and thanks march by, they diminish one another by their weight of numbers. The fact is that each mention is remembered freshly, even if listing them all here cannot convey that. This is more a personal litany than an adequate appreciation, and for that I am sorry. The research on which this book is based was undertaken from September 1977 until April 1979 and again for three months in 1983. I was financially supported during the longer period, in part, by a National Institute of Mental Health predoctoral research fellowship and a National Science Foundation predoctoral research grant. In 1983, I was supported by the American Philosophical Association and by a Princeton University faculty grant. I thank all these institutions for their aid. I owe Andrew Strathern a large debt for steering me to Mendi. Both Marilyn Strathern and Andrew Strathern have encouraged and helped me frequently over the past decade; I am grateful to them both. In Papua New Guinea, thanks go to the staff of the Department of Anthropology / Sociology, University of Papua New Guinea, and particularly to Mary Jane Mountain, who chaired the department when my husband, Michael Merrill, and I arrived for the first time in Port Moresby and who was so friendly and helpful to us. Thanks also go to Marc Schiltz and Lisette Josephides and to James and Acsha Carrier for their hospitality during my stay in Port Moresby in 1983. In 1977, Gary Simpson (then of the National Planning Office in Port Moresby) was able to orient me concerning development issues in the Southern Highlands and was instrumental in putting me in touch with Dr. John Millar, Superintendent of the Provincial Hospital in Mendi town. Together with Dr. Eilene Sowerby, John provided my husband and me with aid and with hospitality throughout our first stay in Mendi. We also appreciated getting to xi

xii

Acknowledgments

know several other doctors and teachers in Mendi town, and we thank them for welcoming us into their homes. Various government officials helped us when we first arrived in town. Among these, thanks to Francis Pusal for encouraging us to settle in Wepra and to Lyn Clarke for keeping in touch with me long after I left Mendi. During my solo return voyage in 1983, the genprous help of Rob Crittenden (lately, team leader of the agricultural research wing of the Southern Highlands Rural Development Project) and Jim Dees (concerned with project financial records) was deeply appreciated. The people in Mendi to whom I owe the greatest debt are my neighbors in the Wepra community where Mike and I spent most of our time. We are still in awe of the ease with which they incorporated us into their lives. The debt I owe them can be repaid only by commitment to ongoing friendship and reciprocity, a promise provisionally fulfilled by my return visit in 1983. Both Mike and I look forward to the next time. Some people became especially good friends of ours: I am thinking of Nare and Nande and their children, of Mel and his brother Tolap, of Wange and Aku, of old Wendo (who died a week before we left Wepra), of Orpeyap, Alcome, Onge (my first twemol), Tekopiri, Andrew Ipopi, Alin, and Namba. Nare and Nande - our hosts in Wepra - and Mel, Tolap, and Wange - who helped us as field assistants - were all daily companions of ours. Once I had returned from Mendi to New York, I depended on the help and patience of my teachers at Columbia University, particularly Alexander Alland, Clive Kessler, Abraham Rosman (my doctoral thesis adviser), and Paula Rubel. Thanks also to Paula Brown Glick of SUNY-Stonybrook, who served pn my thesis committee. I cannot begin to thank the friends with whom I spent time while writing the thesis upon which this book is based. I will single out four: Nancy Lutkehaus, who shared the experience of doingfieldwork in Papua New Guinea with Mike and me (working on Manam Island); Lina Brock, who ran around the Columbia gym track with me, animatedly discussing kinship theory; and Alan Steinberg and Susan Schecter, with whom Mike and I shared a house in the mountains during the summer of 1981, where we all wrote together. My husband, Michael Merrill, is in another category. Any thanks I offer cannot repay all the practical and metaphysical help he has given me before, during, and after fieldwork in Mendi. On the other hand, he got to travel and see the world, an agreeable exchange. This book is dedicated to two families. First, it is dedicated to my host family in Wepra: to Nare, Nande, Paki, Papuan, Lonis, Wembe, and Rina. I miss them deeply. This book is also dedicated with love to my own father and mother, and to my brother and sister.

1

Mendi coming into view

. . . quite suddenly, at a spot where a large tree had recently fallen, we came to a break in the forest. And as we looked excitedly northwards, O'Malley and myself stood spellbound gazing at the scene of wild and lonely splendor. Below us, on the opposite side of the Ryan [River], a large lake lay on a platform of the divide, while about two miles to the northwards; and beyond the gorge, gold and green, reaching as far as the eye could see, lay the rolling timbered slopes and grasslands of a huge valley system. On every slope were cultivated squares while little columns of smoke rising in the still air revealed to us the homes of the people of this land. . . . Beyond all stood the heights of some mighty mountain chain that sparkled in places with the colors of the setting sun. As I looked on those green cultivated squares of such mathematical exactness, I thought of wheatfields, or the industrious areas of a colony of Chinese. . . . "My mother!," said Sergeant Orai. "People like the sand. They have plantations. What people are they?" [Hides 1936: 77-8]

Introduction In April 1935, after more than three months of marching across sparsely populated country, and after scaling the massive limestone barrier of the Karius Range, an exhausted troop of Australian government officers and coastal Papuan policemen and carriers marveled at their first sight of the land of the Huli, in the Papuan Highlands. The Hides and O'Malley patrol never entered the valley of the Mendi River, whose people - Ip Mend Ol - are the subjects of this study. The first patrol through Mendi, led by Ivan Champion, who described his experiences less poetically than did Jack Hides, occurred in November 1936. But Hides's impressions of their neighbors ring true for the Mendi as well. Champion reported that his party was met all along the way by large groups of men, who provided them with food and guided them on their way. Like Hides, Champion was impressed with the intensity of cultivation, the density of the human population, and the beauty of the casuarina pine-lined, parklike clearings at the centers of settlements. 1

2

What gifts engender

Plate 1. The Mendi Valley, looking southward along the government road to Mendi town (the airstrip is visible in the distance). Grassland, gardens, and stands of casuarina pines in the foreground. As a result of Champion's patrol, a government station was set up soon afterward (with Champion in charge), far to the south at Lake Kutubu. Most of the Mendi, however, saw white people for the first time only after the Australians moved their administrative headquarters for the Southern Highlands District (now Province) to the Mendi Valley in 1950, establishing it at a place the people call Murump. Western travelers' accounts, such as those of Hides and of Champion, remark on the unlikely familiarity of Highlanders' style of life. Hides (1936:88) exclaimed, "I thought of pretty little farming areas of Australia, scores of them, in a setting of great rolling timbered plains." Instead of sparse populations of slash-and-burn cultivators (more common in the tropics), here were large, settled populations "like the sand," which achieved densities of more than four hundred persons per square mile in places. Highlanders appeared to be masters of a sophisticated farming system; based overwhelmingly on the intensive cultivation of sweet potatoes (Ipomoea batatas), this system involved techniques of drainage, mulching, and soil conservation, required little or no fallowing, and supported large herds of domesticated pigs as well as large numbers of people. The people Hides and O'Malley met appeared to them to be well off: "We were in a land of plenty," Hides declared. This was a reference perhaps as

3

Mendi coming into view

much to the attitude of the people as to their objective circumstances. He could not get his hosts to accept the steel tools, cloth, and beads he offered in recompense for the food they had given him and his troop. Instead, they showed him their own green stone axes and their necklaces of Job's tear seeds, which they preferred. Dismayed, he learned that had he brought along cowries and mother-of-pearl shells, these would have been accepted eagerly; the Huli, like the Mendi and other Highland peoples, had a complex system of exchange in which shells, pigs, and other valuables - some produced locally and others obtained through trade - were given as gifts at marriages and funerals and on other occasions, ceremonial and mundane. Their involvement in exchange and their concern with shell wealth led some subsequent anthropological observers to call Highlanders (and other New Guineans who share these interests) "primitive capitalists" (Pospisil 1963; Epstein 1968; Finney 1973). This characterization was bolstered indirectly by anthropological studies, beginning in earnest during the 1950s and 1960s, on leadership and politics (e.g., Read 1959; P. Brown 1963; Strathern 1966; Meggitt 1967; Berndt and Lawrence 1971). These studies presented Highland societies as egalitarian, and leadership as achieved rather than as inherited. Highland leaders ("big-men") are self-made men, entrepreneurs with a flair for oratory and for political organizing, whose ambition for prestige obtainable through success at competitive ceremonial wealth distributions motivates their productive efforts. Yet, as we might suppose, the political economy of the region is quite different from that of Europe and North America. Production and the circulation of wealth are organized without the benefit of a market economy; ceremonial gift exchanges constitute one component of a decentralized system capable (at least in the central Highlands) of linking together thousands of people living in several high valleys, speaking different languages, and belonging to many tribal groups, through a largely egalitarian political process quite different from that of Western states. This economic and political structuring constitutes a challenge to the persistent assumption that the development of social scale and complexity of organization involves major concessions to hierarchy and to centralization, particularly to stratification and state forms of organization (see, for anthropological examples, Sahlins 1963; Spooner 1972; S. Polgar 1975; Cohen and Service 1978). For reasons relevant to a changing Papua New Guinea, and arguably beyond its borders as well, we may wish to understand indigenous egalitarian institutions better: how decisions are made and disputes resolved; how increasing scale is managed without fundamental transformations of structure; and how individual differences in energy, commitment, and skill are accommodated. Our empirical understanding of relatively egalitarian polities comes mostly from the study of extremely small-scale and economically unproductive for-

4

What gifts engender

aging societies (see Lee 1979; Leacock 1981; Woodburn 1982; see also Sahlins 1972, Chapter I). 1 What one might venture to call "democracy" among foragers like the Kalahari !Kung San is often said to be the outcome of the naturalistic ecological constraints of a mobile existence: an existence that is antimaterialist, underproductive, and communal by reason of necessity. Whatever the value of the conventional views about egalitarianism derived from the study of foragers (which both Lee and Sahlins criticize), Highlanders do not fit this model. Their decentralized and predominantly egalitarian political relationships are reproduced on an impressive demographic and economic scale. Moreover, there does not appear to be any simple correlation between economic intensification and the development of political centralization or inequality in the Highlands. In fact, both regional comparisons (Modjeska 1982) and archaeologically and ethnographically informed speculations concerning long-term change (Golson 1982) point in the opposite direction, as do some analyses of more recent transformations (see Strathern 1971 for comments on the "democratization" of ceremonial exchange in colonial Papua New Guinea). The experience of the Mendi and their neighbors provides us with a demonstration that a radical kind of democracy is possible on a scale not commonly found in Western experiments with political anarchism. Highlanders' life-style presents other analytical anomalies. In contrast to some other egalitarian peoples, Highlanders are not communal in spirit; their manner of organizing work and their way of motivating material transactions appear to be individualistic. Personal ambition and a competitive logic characterize wealth exchange and social relations generally. Nor do Highlanders follow what Sahlins (1972) called the "Zen road to affluence." Their intensive and pervasive involvement in gift exchange ensures a high demand for pearl shells, pigs, and other valuables. What is more, central Highland exchange systems are "competitive" (see Strathern 1971): One gift engenders a return, each creating new social obligations, which must be acquitted in order that balance - a critical objective - be achieved over the long term. The expandable demand for wealth generated by competitive gift exchanges may help account for the productivity of the economy of the region.2 It may also account for the observation that the economy of the Mendi, like that of other Highland peoples, appears to have been stimulated and not simply inflated by the influx of pearl shells and other exchange wealth accompanying the arrival of Westerners (Strathern 1971; Gregory 1982b; Lederman 1982). The recent efflorescence of ceremonial exchange may also have been attributable to the abolition of tribal warfare undertaken by the Australian administration in its effort to establish political control of the Highlands region. But the imposed peace does not entirely account for it. The response of Highland political economies to alien influences reflected a long-term indigenous trend toward expansion and intensification. At the least, garden production had been expanding and intensifying during the few centuries before

5

Mendi coming into view

contact (Golson 1977, 1981, 1982; Watson 1977; see also Lacey 1981; Modjeska 1982; Lederman 1986). Any presumptions we might have about this as a static "land that time forgot" ought to be buried. In subsequent chapters, I will consider the social contexts of having and exchanging wealth that give Mendi "materialism," "individualism," and competitiveness their particular cultural meaning. An understanding of the ways in which the Mendi structure their social relationships, and of their conception of social accountability and social value, is central to an appreciation of the dynamism of Mendi history. The region and the population: background There are many answers to Sergeant Orai's question "What people are they?" We can answer it by means of sympathetic projections like those of Hides or by means of surveys, population statistics, and other descriptive techniques designed to facilitate cross-cultural comparison. The question can also be answered by the people in question themselves, whose stories describe how their forebears discovered and settled the land, explain the reasons for present-day alliances and hostilities between clans and tribes, and account for the recent appearance of the "red-skinned" people among them. No single sort of answer yields the whole truth. As much as one's sense of space and time differs when traveling by plane moving rapidly and viewing the ground from a distance - from when walking along a mountain trail, so much the understanding afforded by a scientific survey differs from that of the people being surveyed. Demography and geography reveal the manner and dimensions of the Western discovery and use of the place, as well as a global frame of reference. From this perspective, the Southern Highlands Province, at the center of Papua New Guinea, is the least accessible and least economically developed Highland province. It was the last major populated region to be brought under Australian administrative control: Administrative headquarters for the province were set up in the southern part of the Mendi Valley only during 1950-1, and the region was classified "restricted" (meaning that only government officers with armed police could travel there) until 1965, and at least one part of the province was not derestricted until 1977. The Australian administration began at once to improve the business climate by banning tribal fighting and by organizing local labor for road work, but the Southern Highlands remained relatively inaccessible for a long time. Only a fair-weather road connection to Mount Hagen, the nearest marketing center, was available until 1976, when the all-weather Highlands Highway (the only land route linking the Highlands with coastal towns and ports) was extended through Mendi town. As a result, the region became a significant site neither of expatriate plantations nor of locally run cash-cropping enter-

EAST SEPIK PROV. SOUTHERN HIGHLANDS ETHNIC GROUPS

WEST SEPIK PROV.

WESTERN HIGHLANDS PROVINCE MENDI/WOLA WESTERN PROV.

Ethnic/language boundary District and subdistrict centers Provincial government center

ETORO m' ONABASULO

CHIMBU PROV. Erave^----

POLE *% {KWARE SA U

irigV.'Mt. \ Murra

GULF PROVINCE Map 1.1. Southern Highlands Province (after Eastburn 1978). Scale: 1 inch = approximately 50 km.

7

Mendi coming into view

prises. The colonial history of the Southern Highlands was dominated instead by government administrators and missionaries (Ballard 1978). The result was a relatively equitable but exceedingly limited distribution of services and infrastructure to the various districts and subdistricts. That situation may soon change, however (see Simpson 1976, 1978): Soon after independence in 1975, the Southern Highlands Province applied directly to the World Bank for a grant to fund agronomic and public health research and commercial development. Awarded well over $20 million, and starting up slowly during 1978 and 1979, the Southern Highlands Project will probably have profound effects on the regional political economy. Most of the province's 237,000 inhabitants live in the high montane valleys between 1,400 and 2,400 meters above sea level, whereas approximately 8 percent live in lowland regions like the territory around Lake Kutubu, which was the first site of the old Southern Highlands District headquarters briefly during the late 1930s before World War II interrupted Australian involvement there (see Map 1.1). The highland valleys are separated by limestone ridges and by five volcanic cones. The largest of these mountains, Mount Giluwe (4,368 meters), forms a part of the eastern boundary of the Mendi Valley, separating it from Ialibu. The Mendi Valley itself is about 25 miles long and wedge shaped (see Map 1.2). It expands as one moves northward toward the hills of the Kandep area in Enga Province, where the northern Mendi have relatives and, in precontact times, used to travel to exchange southern products (such as tree oil from the Lake Kutubu people and pearl shells) for indigenous salt and renowned Kandep pigs. In addition to their Kandep connections to the northwest, the northern Mendi have kinship and trading relationships with people living in the northern part of the Lai Valley to their west and with Tambul residents living to their northeast. Although Mount Giluwe appears to present an insurmountable barrier between Mendi and Ialibu, inhabitants of these two Southern Highlands districts have a long history of contact via the walking tracks that mark that mountain's slopes. Today the Highlands Highway, which goes around the south side of Mount Giluwe, greatly facilitates such contacts. The valley contracts as one moves southward past Mendi town, which is located at a place called Murump on land ceded to the Australian authorities by clans living at Umbim and nearby localities. Residents of the lower Mendi Valley have kinship and trading relationships with people living even farther to the south, east, and west, in Kagua and Nembi, and southern Ialibu and Lai valleys. Linguistic relationships between Mendi and other parts of the province are complex. Perhaps two-thirds of the people living in the Mendi Valley speak a language that has been called Wola or Angal Heneng (referred to as ngail enenk in upper Mendi, the latter dialectical variant translating as "true [real] speech"). Some linguistic work has been carried out in Nipa, two valleys to

8

What gifts engender to KANDEP (ENGA PROVINCE)

to TAMBUL and MOUNT HAGEN (WESTERN HIGHLANDS

MOUNT GILUWE

Key: Area of high relief : Government road

LAI VALLEY

heedeiload ~—* Rivei Waparaga Local place name ~r-r~^f7-f[~ Boggy land

Map 1.2. The Mendi Valley. Scale: 1 inch = approximately 6 km.

the west of Mendi, by members of the United Church (Methodist) mission (see also Sillitoe 1979) and in the lower part of the Mendi Valley south of Mendi town. At least three dialects of the language exist; the variant spoken in the northeastern part of the valley (the Upper Mendi census division, where the present study was undertaken) differs in phonology and in some grammatical features from the variants found around Nipa and south of Mendi town. Although the area including Nipa, the Lai Valley, Mendi, and probably also the Nembi Plateau constitutes a linguistic unit, it does not follow that it is a cultural or political unit too.3 Speakers of Wola/Angal Heneng do not invariably have more to do with each other than they have to do with speakers of other languages. In fact, the residents of the western part of this language area in the Nipa district (those whom Sillitoe 1979 calls the Wola) appear to have as much in common with the Huli, their neighbors to the west in the Tari Basin (see Glasse 1968), as they do with eastern members of their own language group in the Mendi Valley,4 although they also have culturally distinctive practices. Conversely, the linguistic boundary in Upper Mendi be-

9

Mendi coming into view

tween Mendi speakers and those speaking Imbonggu (the Ialibu language) appears to have little meaning, at least in relationship to socioeconomic life. Imbonggu speakers in Upper Mendi follow Mendi social rules and differ from Imbonggu speakers in Ialibu (see Wormsley 1978). Therefore, when I refer to "the Mendi" in this study, I mean residents of the Mendi Valley regardless of their linguistic preferences (and especially people living to the northwest and northeast of Mendi town, where D'Arcy Ryan and I each conducted research). I will not be referring to the whole language community; indeed, to do so would be to gloss over significant differences. Waparaga (called "Wepra" by local residents), the community in which this study was conducted, is adjacent to the language boundary between Angal Heneng and Imbonggu speakers. The Suolol tribal territory, of which the Waparaga community is a part, straddles this boundary (as do other upper Mendi tribal territories) and is made up of localities both in the Imbongguspeaking northeast and in the Angal Heneng-speaking southwest. People in Waparaga speak one or the other language, often being at least passively competent in the language with which they do not primarily identify. Members of this tribe function as a social and political unit despite their linguistic diversity. The Mendi are aware of certain cultural differences between their own customary social relationships and those of their neighbors. For example, they distinguish their exchange system from that of the Enga, and their marriage customs from those of their Ialibu relatives and Huli acquaintances. We might bear in mind that in the New Guinea Highlands, as elsewhere, language differences do not always suggest social or cultural distance, nor do language similarities necessarily imply similar social structures. Differences in mode of life (whether or not differences in language are involved) do not imply any lack of contact either; both in the precolonial past and in the postcolonial present, the Mendi have interacted with foreigners while maintaining a sense of themselves. In 1979, approximately 40,000 people lived in the Mendi and Lai Valleys, a geographical region called the Mendi Subprovince (or District) and divided into five "census divisions" (CD): Karint, Upper Mendi (both located north of Mendi town), Kambiri, Undiri (both in the lower Mendi Valley) and the Lai Valley (west of the Mendi Valley), all of which are part of one Local Government Council. Land and population figures for Mendi District and for census division appear in Table 1.1. A look at the table indicates that the district as a whole, and northern Mendi (Karint and Upper Mendi CDs) in particular, is moderately densely populated by Highland standards.5 Government analysts do not consider the area to be land short, unlike the Chimbu and Enga Provinces. However, both expatriate and indigenous observers in the Province are worried. Since the population is increasing at an annual rate

10

What gifts engender

Table 1.1. Mendi district census divisions: land and population (1976)

Census division Lai Valley CD Karint CD Upper Mendi CD Kambiri CD Undid CD Total (Mendi District)

Population 8,842 8,505 9,170 5,629 5,126 37,272

Gross area (km2) 238 146 536 363 186 [l,469f

Est. arable land (km2)

Est. population density (persons per km2 arable)

152 124 186 290 126

58 69 49 19 41

878*

42

a

This figure was obtained by totaling up the individual Census Division figures. The Village Directorate (from which all the other figures were obtained) claims that the gross area for Mendi District equals 1,374 km2. b Lower land availability estimates for Mendi District were obtained by Simpson (1976), who reported the total District area as 1,258 km2. From this figure he deducted 500 km2 (land over 2,400 m in altitude above sea level unfit for cultivation for climatic reasons) and another 300 km2 (land unfit for cultivation for topographic reasons). Thus, total arable land = 458 km2. Using these figures and the Village Directorate population figures, we would conclude that District population density (1976) = 81 people per km2, or about 1.2 hectares per capita. Source: Village Directorate, 1976.

of two percent and commercial coffee and cattle projects (which use land extensively) are ongoing or planned, the land situation is changing for the worse (Simpson 1976, 1978). The region and the population: the social dynamics of land use Population density figures are notoriously hard to interpret. Various methods of measuring population density are used by researchers working in different places. No clear standards may be possible because of the varied uses people may make of the land and the diverse methods of cultivation people may employ. A change in techniques can make certain land arable that previously was not. A change in people's notion of what they want to produce can have the same effect. The adoption of sweet potato into the Highlands about four hundred years ago appears to have had the effect of opening up land at higher altitudes than people had previously cultivated. The introduction of commercial cattle projects into the Mendi Valley during the last decade or so has taken some of the best land out of food production. The remaining gardens may eventually become overtaxed; cattle is not a substitute for older products but rather constitutes an addition, with uses that

11

Mendi coming into view

often fall into the category of social investment or luxury rather than of subsistence. Land pressure may be created or alleviated without altering the quantitative ratio of man to land, but by changing the structure of demand or of production. Looking at Table 1.1, it might be noted that, although much of Upper Mendi CD land is classified as "unarable" because of rugged topography, altitude, or boggy conditions, this does not mean that the land is unusable or unused. It is important to know what standards of usage are being employed. Upper Mendi CD (in the northeast) is different from Karint CD (in the northwest) for its forested hills and for the boggy Egari Lake/Tonk River area. From an indigenous viewpoint, the bog is excellent pig-grazing land. It meant that residents of Upper Mendi (especially those of the localities in the northeastern corner of the valley), hard hit by the devastating frost of 1972, did not have to sacrifice their pig herds even though they had accepted rations of rice from the government during that crisis. From the same point of view, the forests are a source of firewood, housebuilding, and fencing materials, as well as other things to which people living in Karint do not have as ready access. On the other hand, because of the high altitude of their gardens, inhabitants of Upper Mendi are more in danger of losing their crop of tubers during a frost, a problem that people living in Karint do not appear to have as frequently. Such differences may become a reason for social interaction. For example, in 1978 many persons in Waparaga complained of local shortfalls in sweet potato production. Quite apart from the weekly market in Mendi town, the result was an acceleration in the trade of sweet potatoes for Papua New Guinea kina notes in the northern part of the valley. In order to appreciate the dynamics of Mendi land use, it is useful to understand something of the range of environmental resources available. The choices people make as to what resources they will emphasize and how they will exploit them result in a particular pattern of land use, a social structuring of their material world. Together with rainfall, altitude creates conditions that Southern Highlanders have used in shaping a particular way of life. Differences in altitude define the three land systems, or broad ecological zones, represented in the Southern Highlands Province: the lowlands (to 1,400 meters), the highlands (1,400 to 1,800 meters), and the high-altitude (1,800 to 2,400 meters) ecosystems (as defined in the Southern Highlands Province Integrated Rural Development Project report, Appendix Three). Less than 10 percent of the province's population, or about 20,000 people, live in the lowlands. The area is characterized by heavy rainfall and high temperatures the year round, as well as by extremely forbidding limestone country. Malaria and various other diseases are endemic. Lowlanders practice shifting cultivation on a long forest-fallow cycle, moving their settlements

12

What gifts engender

every five to fifteen years. From the point of view of the Highlanders, loweraltitude regions like Erave and Lake Kutubu are sources of valued forest products as well as dangerous poisons used by sorcerers. The highland area is the most densely populated of the three zones. About 75 percent of the population of the province lives there, with a gross population density of fifty-two persons per square kilometer (about one hundred forty-four per square mile). The climate in this zone is wet (although less so than in the lowlands) and does not display marked seasonal contrasts. The Mendi Valley receives about 112 inches of rain per year, less than the 120 to 130 inches average for this zone in the province as a whole. The highlands are above the altitude at which malaria is endemic; daily temperatures are lower than in the lowlands, and malaria-carrying mosquitoes do not thrive there. Highland agriculture is based on the sweet potato cultivated on a short, grass-fallow system (and often interplanted with sugar cane and greens, and nowadays also with European vegetables, such as cabbages, pumpkins, seallions, and tomatoes). Settlements are not relocated in response to the need to shift garden sites (although populations were occasionally displaced in the context of local warfare and intragroup disputes). In contrast to the lowlands, intensive agricultural practices (including mulched mounds, drainage ditches, and short fallow cycles) have eliminated the original forest vegetation everywhere in this area except on ridge crests and in protected groves. Grasslands predominate wherever sweet potato gardens are not actively cultivated. The high-altitude zone, a subdivision of the highland zone above 1,800 meters, is the zone in which most of the Upper Mendi CD is located. The high-altitude human population is restricted to the area below 2,400 meters (i.e., 8,000 feet). This altitude marks a practical limit to habitation because above it the maturation time of sweet potato is more than twelve months long (as opposed to five to seven months at lower altitudes), and cultivation is impractical also because of a high susceptibility to frost damage. The area above 2,400 meters is used for its forest resources. The Suolol tribal territory, where this study was mostly conducted, extends from the Mendi River at 1,950 to 2,000 meters (or about 7,000 feet) in the southwest up to the forested crests of two ridges at above 2,400 meters and then down again about 200 meters to the flat, boggy pig-grazing country through which a river called Tonk meanders at the foot of Mount Giluwe in the far northeast. Most of the Suolol's gardens are located between 2,000 and 2,200 meters. About 15 percent of the province's population, or about 35,000 persons, live in this high-altitude zone. Overall, about 200,000 persons live in similar environments - mostly in the Enga and Western Highlands Provinces. Eric Waddell has described the agricultural practices of Enga high-altitude culti-

13

Mendi coming into view

Plate 2. Sweet potato garden surrounded by grass fallow. In Upper Mendi, sweet potatoes are planted in large, mulched mounds. Gardens contain mounds with sweet potatoes in various stages of growth. vators (Waddell 1972); his detailed description is relevant to the Upper Mendi. Because the area is subject to periodic frosts and droughts, the intensive cultivation techniques found in the Upper Mendi and places like it have been explained as means of buffering the sweet potato crop against frost damage. The large composted mounds in which members of Suolol and other Upper Mendi tribes plant their sweet potato vines raise the soil temperature by a few potentially critical degrees, as well as improving the quality of nutrients available to the crop. A reference to environmental exigencies, however, only partially explains such intensive cultivation practices. The ancestors of modern Highlanders probably did not occupy high-altitude zones. The environmental conditions under which Highlanders live, and with which they must deal in one way or another, are not abstract necessities but an historical consequence of choices their forebears made. Asking why some people now live under these conditions ought to lead us to wonder about the reasons for population growth and movement in the highland and high-altitude regions over the past few hundred years. It leads us to wonder about the reasons for the relatively recent (400year-old) adoption into Highland farming systems of the sweet potato - a tuber that, compared with others that were available for centuries before it

14

What gifts engender

arrived, is known to produce well at higher altitudes and to be particularly good pig fodder, making high-altitude habitation and pig raising possible (although not necessary). In fact, Highlanders have considerably altered the ecosystems in which they have lived. Their intensive methods of cultivation transformed the original montane forests into grasslands (Bowers 1968). Needless to say, the effects of their movements and productive interventions in their physical environment have been neither entirely intentional nor foreseeable. Some of their choices and behavior resulted in misfortune. Whereas high-altitude habitation was favored by some people for the access it gave them to pig-grazing land, forest resources, and trade routes, the frosts to which this geographical zone is subject are capable of destroying the crop and of causing famine and social disruption (Waddell 1975). Such facts imply that population density and other aspects of land-population relationships can be evaluated only in relation to what particular groups of people wish to produce, the patterning of their demand for the products of the land, and the choices they make about how they will use the land to meet their requirements. Like production systems elsewhere in the world, the intensive production systems of the Highlands reflect socially determined needs. With such needs shaping the goals of production, the land is a means that is sometimes used sensitively but that is frequently transformed in the course of its use, and often with unpredictable long-term effects. The "subsistence bias" and another approach to Highland political economies Although we cannot account for intensive agriculture in the Highlands in terms of environmental exigencies, we must recognize that the inhabitants have impressed a particular structure of demand upon those environments. We will not be able to say what that structure is as long as our understanding is impeded by the "subsistence bias" evident in much of the current literature on Highland economies. Studies of agricultural production have generally been conducted independently of studies of ceremonial (or gift) exchange by different types of researchers. Consequently, it has been possible for agricultural economists, ecological researchers, and agronomists to write about Highland political economies as "subsistence systems" and to ignore the crucial link between garden production and gift exchange most concretely embodied in Highland pig herds (Colwell 1982; Fisk 1962, 1964, 1971; Fisk and Shand 1969; see also Crittenden 1983 and Simpson 1980, who comment critically on the assumptions of the Southern Highlands Rural Development Project).6 This study began as an anthropological investigation of the relationship between production and exchange. When I first planned the research on which this book is based, I accepted as given that there would be a systemic rela-

15

Mendi coming into view

tionship between production and indigenous forms of exchange, and I planned to take as my subject the sociocultural contexts of production. I assumed that tacit and explicit conventions concerning proper behavior between categories of kin would more or less define the mode of "distribution," that is, the ways in which access to culturally valued resources and products for consumption and exchange are controlled (see, e.g., Gudeman 1978). Expectations concerning proper behavior among kin and other significant categories of person would also have an effect on the kind and quantity of goods required at different social occasions - weddings, funerals, and the like - and on the frequency and organization of those events. That is, I assumed that in the New Guinea Highlands, an understanding of the social relations through which production and distribution are organized would require study of exchange and social structures (Bloch 1975; Godelier 1972; Sahlins 1972).7 Once thefieldwork was under way, these assumptions focused my attention on Mendi exchange practices. To understand the system of conventions, tacit and explicit, pertaining to the ways access to culturally valued resources and products are controlled required that I supplement the existing anthropological emphasis on ' 'ceremonial" exchange - which generally refers to periodic public events sponsored by social groups and organized by leaders - with an analysis of informal exchanges of pigs, pearl shells and money that take place everyday between kin and friends. As it turned out, this shift in emphasis was productive and helped shape a framework for understanding Mendi society that I had not expected when initially planning the research. The rest of this book is devoted to demonstrating the significance, in Mendi, of two distinct structural principles8 or forms of sociality: exchange partnership relations (the twem principle) and clanship (the principle of sem relations), the first being most clearly revealed in everyday contexts of exchange and the second especially evident during large-scale ceremonial presentations. These two forms of sociality are densely articulated in the social experience of the Mendi. Their articulation, however, is not smooth; a tension between them underlies political process in Mendi, and is also central to an understanding of the structure of production. That is, first, a dual perspective on ceremonial and everyday exchanges sheds light on Mendi politics by revealing the structural basis of social inequality. Anthropologists have given Highlanders a reputation for egalitarianism, which by and large they deserve.9 But inequality between men and women has long been recognized as a theme of Highland social life (Read 1952; Meggitt 1964). The meaning and value of distinctions made in a gender idiom, and the differential powers ascribed to men and women, are tacitly and explicitly contested within Highland societies, among Highlanders themselves (M. Strathern 1972; see also Meigs 1983). Anthropological discussions of male-female antagonism and of male dominance have proliferated (for recent examples, see Herdt 1982, and especially Keesing's overview in that

16

What gifts engender

volume), but their implications are hard to evaluate, given the predominant emphasis in Highland research on big-men and large-scale ceremonial prestations. This emphasis, though important for other reasons, has produced a partial picture both of the gender system and of social inequality and political relations among men. The nature of male leadership and the significance of formal public events themselves will not be understood as long as they are inadequately contextualized with information about ordinary people and the interactions and transactions taking place in everyday settings. Second, a dual focus on ceremonial prestations and on everyday exchange also contributes to a comparative understanding of the structure and goals of production in Mendi by clarifying its particular cultural character. On the one hand, Highlanders have been called "primitive capitalists" because of their concern with wealth and apparent individualism (Pospisil 1963; Finney 1973). On the other hand, Highlanders are considered subsistence farmers. Ethnographic descriptions of redistributive exchange in Highland societies, such as Andrew Strathern's (1971) account of the moka system, or Mervyn Meggitt (1974) on the Te (or tee; see also Feil 1984), ought to have been significant counterweights to both a market interpretation and a subsistence interpretation of Highland economies. But they have not proved sufficient. Perhaps the subjects of these ethnographic accounts were too easily brushed aside as exotic ritual events by researchers concerned with quotidian economic realities. Everyday exchange behavior and other mundane details could themselves be treated as transparent - as unproblematic for cross-cultural analysis - as long as their specific cultural structuring was inadequately described. A more comprehensive picture of indigenous socioeconomic relations requires that one be able to encompass everyday practices and ceremonial events within one frame and that the culturally constituted character of each be delineated. This picture demonstrates the specificity of local notions about individual interests and their social contexts and clarifies how they differ from analogous notions in market societies. If a dual emphasis on everyday exchanges and public ceremonies highlights the fact that intensive production in Mendi can be accounted for neither by reference to subsistence needs alone, nor to capitalist market mechanisms, what can be said, in a positive way, about the social goals of production in Mendi? Only a framework for understanding can be suggested here, by way of introduction. The key to understanding the structure of demand, and consequently the level of production, is to know something about the relationship among production, pigs, and exchange. Sweet potatoes are the main agricultural product produced by the Mendi, but only about half the crop can be considered strictly as food for human consumption. The other half is fodder for herds of domesticated pigs.10 Pigs are also eaten, and most people look forward to pork meals, which occur for one reason or another throughout the year.11 But pigs are not simply good to

17

Mendi coming into view

eat; they are also a form of wealth and have value insofar as they are made to stand for social relationships. Eating together generally, but especially sharing and exchanging pork and pigs, demonstrates and constitutes social connectedness. Large pig herds are an artifact of sociopolitical relations that create a high demand for pigs.12 A number of researchers (e.g., Brookfield 1971, 1973; Watson 1977; Rubel and Rosman 1978) have noted the capacity of Highland exchange systems to create an expandable or escalating demand for pigs as wealth (if not as food). These systems are not all alike. The nature, timing, and organization of exchange ceremonies vary from society to society in the Highlands. Such facts as whether people exchange live pigs or pork, whether they have a system of credit or one that emphasizes "home production" (Strathern 1969a), and whether they organize most of their ceremonies as large, corporate clan affairs under the direction of a big-man or as smaller affairs are relevant to an analysis of the organization and level of production in particular places. These variations affect the structure of the pig herd (the numbers of pigs per capita, as well as each herd's age, sex, and size distribution), and with it the level of sweet potato production. The connection among garden production, pigs, and exchange is central to an understanding not only of contemporary variation in Highland political economies but also of long-term historical processes. These have included precolonial population movements and population growth, the dynamics of clan structure, the expansion of anthropogenic grasslands and other indications of the geographical spread of intensive production systems in the Highlands, and evidence for the recent, rapid adoption of the sweet potato. Highland production systems had been intensive for thousands of years before the introduction of sweet potatoes; throughout this period, pigs had been a factor. As Watson and others have pointed out, however, sweet potatoes are ideal pig fodder (Brookfield 1973; Watson 1977). The adoption of sweet potatoes into Highland production systems beginning about four hundred years ago most likely facilitated increased pig production, just as it made possible an expansion of populations into high-altitude zones. It did not actually necessitate such changes, however. There would likely have been nothing particularly significant about the introduction of the sweet potato had there not been a high and expandable demand for pigs as valuables in systems of exchange.13 Thus, in order better to understand variations in Highland political economies and the dynamic social processes that produced them, we must learn more about the kinds of demands that exchange systems place on production. Paula Rubel and Abraham Rosman (1978) organized information on contemporary variations in Highland and Lowland New Guinea exchange systems by means of a transformational model that also has implications for attempts to understand Highlands history. Different structures of exchange are character-

18

What gifts engender

ized by different degrees of regional integration and are found in societies that vary in terms of population density, agricultural intensity, and pigs per capita. Rubel and Rosman explicitly relate their transformational model to hypotheses concerning agricultural evolution. In particular, they recognize that exchange systems make demands on systems of production and that the elaboration of exchange systems is limited in turn by the ability of their environments to produce what they demand. The present account differs from that of Rubel and Rosman in avoiding the idiom of evolution and qualifying that of structure. That is, I do not consider social transformations a matter of "evolution" (i.e., a matter of natural or structural law), nor do I think that analyses phrased solely in terms of social "structures" are helpful in an effort to understand the transformational possibilities of particular culturally and historically bounded societies. In my view, the point of a structural analysis of social rules is to enable us better to understand people's social experience. Analytically identified structural transformations become culturally and politically significant when they correspond to a felt change in the meaning and morality of human action - to a change in what is "thinkable." In the final analysis, transformations are (whether more or less deliberately) the outcome of human agency, and not of structural necessity (Thompson 1978). They are the product of struggles over the control of the meaning of human action, and to take account of them means writing historically no less than structurally.14 It is toward a deeper understanding of these sorts of struggles that this analysis of social relations and local-level political history is offered. Chapters 2, 3, and 4 explore the character and problematic articulation of sent and twem relations. The structural contradiction between corporate clan and network relations provides a framework for understanding the organizational limits and the dynamism of the Mendi political economy. For the Mendi case, the discussion in Chapter 5 and 6 is meant to show how the articulation of network and group relations is actualized and negotiated in the practical historical experience of Mendi men and women. In Chapter 7, the relevance of this analysis for an understanding of contemporary social change is suggested.

2 Sem relations: solidarity and its limits

"In those days, no one slept well at night. Now those times are over. That's good; I am pleased," Kundapen told me. The old man had just concluded an account of the reasons for the age-old antagonism between his own tribe and their neighbors. Not everyone agreed with his sentiments, however. Mel [my thoughtful field assistant, who had been a youngster during the 1950s when the Australians imposed peace on Mendi tribes, and who had spent his early twenties in Port Moresby working in the Papua New Guinea army mess] shook his head and asserted that for his part, he wished that he still lived in the days of those "cowboys"! l

Introduction The social geography of the Highlands is a patchwork of clans - landholding groups of kin - linked to one another in political alliances or separated by enmity. In precolonial times, enduring antagonisms between particular groups, erupting into violence on occasion, were dominant facts of life (see, e.g., Meggitt 1977). Warfare inhibited travel, restricting the geographical elaboration of both personal and clan relationships and wealth exchanges. The endemic state of war - more accurately "Warre" in Sahlins's (1968, 1972: Chapter 4) sense - might have been both a means and a result of a political system in which access to force was decentralized. Still, if it was the reflection of a self-help attitude critical to the reproduction of egalitarian social relations, the readiness of tribesmen to take up arms in defense of their interests was not the only means of sustaining such a social order, nor were its modalities entirely egalitarian, as we shall see. The suppression of tribal fighting, undertaken during the 1950s and 1960s by the Australian colonial authorities, transformed the social landscape in Mendi, as it did elsewhere in the Highlands. But despite the loss of political autonomy implied by pacification and the national state, even during the early 1980s the Mendi displayed an active attitude to their rights and interests ap19

20

What gifts engender

propriate to an independent people. They accepted joint responsibility as members of clans (sent onda) for contemporary "sorcery" deaths (see the section on "Individual Action, Collective Responsibility" later in this chapter, and note 5), as well as for those that occurred years ago during the times of tribal wars. Large festivals and wealth exchanges were still organized on behalf of allies killed in battle during the 1930s and 1940s, or in order to gain back land that had been abandoned as a result of the fighting. Therefore, to the extent that the John Wayne movies Mel had seen in Port Moresby (without the benefit of dubbing into any of the many languages he speaks) conveyed to him a sense of autonomy and power, the regret he expressed in his commentary, in the epigraph above, aptly reflected an era in which young people of an age to begin taking over from Kundapen's generation are becoming aware of what they all may have lost.2 With the curtailment of tribal war-making powers during the past twenty years, the rationale for coordinated group action has been changing. Pacification and a system of roads have made large-scale organization easier but less pressing militarily. At the same time, it is unclear how an introduced social order expressed in wage work, commercial projects, provincial government, and the missions will articulate with indigenous structures of social relationship. The possibilities will be poorly understood without a sense of the organizational complexity of the indigenous society. No meaningful comparisons between John Wayne's cowboys and Kundapen's warriors can be made - much less any evaluation of the two - until one understands the Mendi social categories twem and sent. This chapter focuses on the meaning and organization of relationships between people as members of sent, a term that can be glossed as "family": sem kank ("small family": lineage and subclan) or sem onda ("large family": clan and tribe, or "clan cluster" in D'Arcy Ryan's (1961) apt but idiosyncratic terminology). The analysis of social structure in the Highlands of New Guinea has generated a large and contentious literature that has by now gone beyond the questions of the 1960s concerning the inappropriateness of patrilineal descent models (see, e.g., Glasse 1959; Barnes 1962; Brown 1962; Meggitt 1965; Strathern 1969b) to questions concerning the very existence of corporate groups there and the substitution of models of sociality based on exchange and other factors (Strathern 1973; Wagner 1974; Feil 1984; Sillitoe 1979; Leroy 1979a, b) during the 1970s. The application of classic African descent models has largely failed in Highlands research; societies like that of the Mendi wind up in a residual category labeled "loosely structured." But as Roger Keesing wrote, A society might qualify as ''loosely structured" if people went around surprising one another. When they go around surprising the ethnographer but not one another, the ethnographer must be looking for the wrong kind of structure in the wrong way. [1971:24]

21

Sem relations: solidarity and its limits

Keesing, Raymond Kelly (1977), Roy Wagner (1967, 1974, 1981), and others (Bourdieu 1978, Kuper 1983) have argued cogently that adequate analytical models are likely to involve a complex articulation of structural principles. In the following discussion I shall argue that the notion of corporateness does account for important social and cultural facts about sem: members of one sem onda constitute themselves as an enduring social group with a name, an estate, and an existence independent of its current membership. But I shall also argue that, in Mendi, clan relationships cannot be fully understood without an account of personal exchange partnerships (twem), a social category with which sem onda contrasts. Several solutions to the loose structures impasse have been proposed to account for data concerning the societies of the Highland fringe south of Mendi (e.g., Etoro, Kaluli). They identify kinds of complexity different from what I describe here (see, e.g., Wagner 1967; Kelly 1977). Whereas a synthesis of these approaches with analyses of central Highland societies (e.g., Huli, Mendi, Enga, Melpa, Chimbu) ought to be attempted, each was designed to account for specific ethnographic facts. Even central Highland social structures do not necessarily lend themselves to a unitary solution: In Mendi and other parts of the Southern Highlands, where there are many contexts in which obligations to personal networks of exchange partners have priority and where there are fewer contexts in which obligations to fellow clan members are invoked, the analytical challenge to incorporate networks into a general account of social relationships is great. In Mae Enga society, Mount Hagen, and other parts of the northern Highlands, where the corporate obligations of clan members are more frequently stressed, this analytical problem may not exist. In other words, I do not expect my analysis of Mendi social relations to apply to other Highland societies in any simple way. Different analytical emphases in the existing ethnographic literature appear to reflect empirical differences in Highland societies, and not simply a theoretical disagreement to be resolved. Nevertheless, neither the theoretical nor the empirical distinctions have been adequately drawn so far; this analysis of the Mendi case aims to inch us closer to a truly comparative understanding of the region. To present Mendi sociality as the product of a complex system of relations, I will initially foreground each of its significant parts in turn before examining their articulation. This chapter is concerned primarily with clan relations and specifies the kinds of collectivities they create. Individual action, collective responsibility As an outsider, one first encounters sem as names like "Kurelka" and "Yansup" in the context of conversations about events (deaths, disputes and festivals) and about individual affiliation and participation in events. The way in which these names are used in Mendi poses some problems of interpretation.

22

What gifts engender

It seems clear at first that "Kurelka" and other sent names refer to collectivities of people. The question is, what sort of collectivities? Sent names signify social relationships projected backward in time (see the section "Names, Categories, and the Reality of Groups" later in this chapter). By means of clan names, individuals assert an identity or equivalence with some of those people who have preceded them. People living today conceive themselves to be jointly accountable for the past actions of others (some of whom no longer alive) by virtue of their common sent onda identification.3 Moreover, at the same time as sent onda membership puts one in a structured relationship with members of one's own clan, it distinguishes one from people identified with other such groups. Relationships of another sort pertain between people belonging to different sent; however, the relationship is analogous to that which former members of different sent had with each other in the past. There are two points here. First, the rationale for coordinated action by the contemporary members of a sem onda is often represented by the Mendi as an historical product, the result of past actions by people of the "same" clan. Clans constitute objective groups, which the Mendi believe to persist despite a changing membership. Second, members of a clan have a joint responsibility not simply for the past actions of contemporary members but also for the actions of their long-dead fellow clan members: people they might never have known and actions in which they might never themselves have taken part. Being a group member means being accountable for actions by other people in the past taken (or construed to have been taken) in the group's name. Similarly, actions in the group's name today constrain the choices of future members. Because the Mendi talk about clans as if they are historically continuous despite a changing membership, and because they recognize the joint responsibility of current members for the past actions of previous members, I interpret Mendi sem onda as "corporate" groups.4 (There are other reasons as well, to which I will turn shortly.) That is not to say that the boundaries of corporate responsibility are always obvious. The Mendi debate among themselves how one determines whether someone in the past did in fact belong to the "same" sem onda as oneself, what in fact that person did, and where responsibility may rest. Arguments about present-day responsibilities and the meaning and moral value of conflicting courses of action turn on alternative interpretations of local history. But while people may disagree as to their accountability for particular acts by members of the same sem onda in the past, by and large they do not question the legitimacy of the general principle of corporate responsibility and continuity. Death is an important context for the active reproduction of corporate groups. Most deaths - all except those of infants and very old people - are attributed to human agency: sorcery (torn) or face-to-face combat. Thus deaths demand

23

Sem relations: solidarity and its limits

Plate 3. Pearl shells displayed before a war-death compensation ceremony (ol tenga). The shells, set up on banana leaves and ferns, rest against one of the long Pig Festival houses (sai anda).

the discovery of a responsible social agent, and to that end, divination rituals, dream interpretation, and extensive discussion and debate may occupy a community long after a burial. In the end, a clan or subclan (and not an individual) will admit that evidence points to someone among them and will accept collective responsibility for making a ceremonial prestation to the group of the person who died. Such corporate gifts - called ol tenga - do not exactly constitute admissions of guilt, but they do repair breaches between the groups by helping replace the loss of a member. Organizing ol tenga demonstrates a clan's corporate identity in relation to other groups and reinforces or alters intergroup alignments. In 1979 two members of Pulumsem, a subclan affiliated with the Kurelka clan (one of the two local clans in the Senkere community) made an ol tenga to members of Mesa, another Kurelka affiliate (refer to Table 2.1). The death in question had occurred a few years before. No member of Pulumsem was accused of killing the Mesa man. Rather, a member of Molsem (the other local clan) was said to have killed him. Why, then, was Pulumsem compensating Mesa for the death? Tasupae, an old Pulumsem man, explained with this interpretation of community history (what follows being a slightly condensed paraphrase of his words):

24

What gifts engender

Table 2.1. Clan organization in the Senkere community Sem onda name

Sem kank name

Kurelka (sometimes called Mesa-Kurelka)

Olsem Ipulpirisema Keyosem* Pulumsem Tonkpisem Pulumsem Punginsem Mesa Mesa Tonkpisem Kalap-Kurelka (one representative in Senkere area, others living in the northern part of Suolol territory) Toleke Anksuolc Keyosem Taikpilsem" Senkere-Molsem Pesal-Molsem Ombisem^ Saolsem* Molmanda-Molsem Molmanda-Molsem Temsem^ Kombal-Molsem Pombresem Napsem Koenpisem

Molsem

Major affiliated sem kank

Olsem

Note: Only groups in the Senkere community are included; other parts of Kurelka and Molsem live in the Kuma area, in the northern Suolol territory, along with other affiliated groups. Keyosem of Olsem and of Anksuol are two different groups of people who intermarry. Tonkpisem of Pulumsem and of Mesa appear to be the same group. "This group will not have heirs in the next generation. This group immigrated to Senkere in the present generation. This group immigrated to Senkere several generations ago.

Wane, the son of Wia [of Pulumsem], was working in a garden on the outskirts of the settlement when he heard the sounds of fighting. He picked up his bow and arrows and ran down the Pual road towards the noise. Men of Surup [a neighboring allied group] and of Kurelka were fighting one another. Wane joined in the fight to help his Kurelka brothers, and was speared by Yabi of Kalap-Kurelka, who was helping the Surup men. Yabi told his Surup friends to call out that they had killed Wane, which they did. Later on, men of Pulumsem found out that they had been deceived - that a fellow Kurelka and not a Surup had killed their brother. Wia was angry, and it is said that he obtained poison in order to avenge his dead son, convincing a Mesa man named Takamu to help him. Some time after this, a clansman of Yabi, named Puyu, died. People concluded that Mesa Takamu had killed Puyu with the poison provided by Wia. Now, both Pulumsem and Kalap had suffered a loss. All of this took place sometime before the white people came to Mendi [about forty years ago].

25

Sem relations: solidarity and its limits

Much later, only a few years ago, a leader of Mesa died. Other Mesa leaders accused the Kalap man Nande [who lives on Mesa land], thinking that he had been motivated to the deed by the desire to avenge the death of his clan father, Puyu. As a result of a divination ritual [suko], Mesa men also suspected that Nande had been helped by a Molsem man named Tumnaik. Nande's and Tumnaik's involvement and responsibility remained to be determined, and mutual accusations by members of Kalap-Kurelka, Molsem, and Mesa were unresolved even in 1979. But while they were arguing with Kalap-Kurelka and Molsem, Mesa also asked Pulumsem to make them an ol

tenga because, as Tasupae noted, "all of this trouble had started with Pulumsem, on account of Wane and Wia." Members of Pulumsem agreed that they had been responsible, and wanting to set things right between themselves and their Mesa clan brothers, they made the ol tenga payment. They were still on good terms when I visited Mendi again in 1983. As Tasupae said, "We sit and laugh and eat together." Ol tenga payments are made to compensate allies for losses they suffer helping the "root men" of a fight (ol sont te). If deaths are not compensated with wealth, the Mendi say that bad feelings will lead to further killings, and the groups involved will find it increasingly difficult to cooperate on joint projects (like pig kills - see Chapter 6) or to stand against external dangers (including both occasional disputes with allies like the Surup, and the ever present threat of war with their major enemies). A prestation breaks the cycle of deaths, in the Mendi view, by calming the angry feelings of members of the bereaved group (see Lederman 1981 and Chapter 5). The limits of this process - the limits of the Mendi political system - occur where one finds groups that do not compensate each other for deaths for which they admit responsibility. Major enemies each compensate their own allies for the deaths and injuries the latter suffer while aiding them in their battles. In Tasupae's history, there was never a suggestion that a Pulumsem man had been directly responsible for the Mesa leader's recent death, which had occasioned the ol tenga request. Rather, the alliance between Wia and Takamu, contracted more than a generation ago, was at issue. The men of Pulumsem agreed to the Mesa interpretation of local history, implicating individuals associated with Pulumsem in a chain of events leading to the recent Mesa death, and therefore they agreed to make an ol tenga in the name of Pulumsem. Although Pulumsem Wia and Mesa Takamu were represented as having acted as individuals, members of Pulumsem accepted joint responsibility to make the payment to Mesa. Two of them (one being a lineal descendant of Wia, the other being from another Pulumsem lineage) actively organized the payment and other Pulumsem members helped the two informally. The Pulumsem/Mesa ol tenga was characteristic of death-compensation prestations in Mendi. In this case and others, certain individual actions are accorded a group significance in political discourse, and those interpretations

26

What gifts engender

are legitimized by means of public prestations of wealth. Mesa Takamu might have acted to help Pulumsem Wia for personal reasons, but his actions came to have a meaning for his sent as a whole. Similarly, whereas Wia was said to have individually and secretly obtained poison, the people of Mesa deemed Pulumsem accountable for his actions.5 Conversely, Kalap Yabi was said to have been helping his Surup affines and tried with their cooperation to subsume his own act to Surup group responsibility. Other Kurelka did not agree, however; instead they held him and his subclan brothers responsible, interpreting Kalap Puyu's death as a repayment for Yabi's act. This way of thinking, wherein individual actions may come to represent those of a group (and sometimes contrary to the intentions of the actor), contrasts with everyday Mendi expressions of individual autonomy and freedom of action. Mendi social morality is very similar to that of the Mount Hagen people described by A. Strathern (1979a). If one inquires after the motivation of another person's actions, the Mendi (like the Hageners) typically answer that one cannot know what another person is thinking, or see another person's thoughts. As in Mount Hagen and elsewhere in the Highlands, the Mendi believe that thoughts or intentions (kone) are invisible things perceptible only to invisible beings. As Strathern wrote with reference to the Hageners, The ghosts see intentions, but the living do not, until overt actions are performed or until the ghosts make the living sick so as to reveal an uncompleted intention or a completed wrongdoing. Just as the ghosts themselves are unseen, so they are able to see the unseen; they belong to the element of thought and apprehension which is kept private from others in life. [1979a: 103] Another person's actions are simply ''what pleases him or her" (ipun turiom), and nobody else's business. Although much motivation is understood to be personal, private, and autonomous, the effects of an individual's actions may be construed as social, public, and interdependent to the extent that each person stands for his clan. Any member may be taken as representing his sem, despite himself, from the point of view of members of other groups.6

Individual autonomy, political participation, and leadership The internal politics of Mendi clans are organized in what we might term a "participatory" style, meshing individual and collective purposes without a mediating central authority. In contrast, Western political theory takes the State by and large as a given. Our models of democracy were developed during the period of the birth of capitalism in Europe by Thomas Hobbes, John Locke, and the theorists of the American Revolution (among others). They presume an adversarial political process in which one elects representatives and in which, in order to make decisions for action, votes are taken,

27

Sem relations: solidarity and its limits

with the majority ruling. This process rests on the assumption that "the citizen's interests are in constant conflict" (Mansbridge 1980: 3). Therefore, ideally, the political institutions of the democratic state are designed for the equal protection of individual interests. Jane Mansbridge, a political theorist who studied participatory organizations and cooperatives in the United States, argues that another form of democracy, which she terms "unitary," also exists. This form is based on faceto-face discussion that proceeds until a consensus is reached (not a secret ballot vote), and on community participation (rather than representation in a central government) as the means of political interaction. The central assumption of a "unitary democracy," according to Mansbridge, is that the citizens have interests in common. Mansbridge claims that these two forms of democracy are not generally distinguished in the literature on political types. As a result, our understanding of cooperatives has suffered. Finally, she argues that we need to learn how the two might be knit together. On the one hand, interests do not always conflict and adversary procedures are not always necessary. On the other hand, conflicting interests exist even in cooperatives, whose members ought not to expect to achieve consensus on all issues. Political action in Mendi and other Highland societies (where villages retain a greater degree of economic and political autonomy than do the cooperatives Mansbridge studied) provides an example of how these kinds of democratic process might be combined. In Mendi the interests of members of village communities are understood by them to diverge; individual differences attributable to each person's obligations to his or her own affines and other exchange partners are expressed in the ethic of individual autonomy. Factional cleavages are expressed in subclan distinctions.7 Despite these diverging interests, the Mendi come together around sem onda projects conceived of by them as in the common interest. Ideally, during meetings held to discuss large-scale clan and tribal affairs - festival plans, or an appropriate collective response to threats by other sem onda - diverging individual and factional interests are accommodated but not submerged by a common group policy. In Mansbridge's terms, both "adversary" and "unitary" political processes coexist. People come together constantly - both in formally called meetings and in informal gatherings - to plan and discuss group projects. Whether they meet in small groups in particular men's clubhouses or in larger numbers in open public spaces (koma, or ceremonial grounds), the goal is to achieve a consensus, or "one talk" (ngail pombor). Discussion is directed toward common corporate interests and away from potentially conflicting or divergent obligations of individuals to their particular exchange partners. Leaders ("big-men"; ol koma) - who are generally concerned with (he solidarity of sem onda - have a crucial role in this process. They take every opportunity, even informal gatherings when clan affairs are not the main topic

28

What gifts engender

of discussion, to emphasize the importance of group projects, to downplay the significance of individual disagreements and problems, and to condemn intraclan divisiveness. For example, during 1978 and 1979, the two years before the community's Pig Festival, many village courts were held in Senkere. The purpose of these events usually was to settle grievances between individuals. Nevertheless, during these events, Senkere's most influential big-man regularly made speeches about the importance of sent onda unity and about the perils of divisiveness, referring to the particular events at issue in the court as examples of behavior to be avoided. This is even more characteristic of formal meetings of clans, when clan affairs are the main topic (see Lederman 1980). If a consensus is to be achieved, members of the group have to agree about their common interest in a group project; they then expect to fulfill their individual obligations to exchange partners as best they can within the constraints imposed by group events. As in the model of a "unitary democracy," formal votes are not taken, and explicit group decisions are not announced definitively. Members simply demonstrate their common purpose by acting it out - by contributing to a wealth display or coordinating their efforts so as to participate in a festival on the specified day. But consensual procedures are not effective in Mendi when alternative and conflicting agendas for clan action each have stubborn advocates or when there are disagreements concerning whether there should be any sent onda action at all in particular situations. Some members may remain unconvinced of the value of particular proposals for group action, being less concerned about their standing vis-a-vis their brothers and more concerned at the time with their obligations to their own affines and other exchange partners. Each person has a limited amount of wealth at any time and may choose to withhold a pig or a pearl shell from a clan prestation so as to make other use of it. It follows from the ethic of individual autonomy that people have the right to make up their own minds about such matters. In the limiting case, some do not participate in clan events because they are simply not interested. Such people are held in low esteem and are sometimes referred to as "rubbish" men (ol ter). Nevertheless, in Mendi such men have gardens and a pig or two, and may even marry.8 But as long as they do not actively diminish their group (i.e., by killing a brother), they are not ostracized for their uncooperative behavior. Beyond these individual differences, whole factions - sem kank and sometimes closely allied sem onda - may not see eye to eye. Still, when formal public meetings are held, no vote is taken. To do so would be to give formal expression to factional divisions and to undercut an important purpose of such meetings: the demonstration of common group identification and solidarity (see also Rappaport 1968). Formally recognizing factional divisions may mean the destruction of the group, and creation of new ones in its place; such "fis-

29

Sem relations: solidarity and its limits

sioning" has been well described (Ryan 1961; Meggitt 1965; A. Strathern 1972). It is not an everyday occurrence. More often, forces external to the group mediate its internal divisions indirectly, playing a role functionally similar to (but structurally quite different from) a centralized state authority. Decisions concerning cooperative action, achieved through intensive community discussion, usually emerge as reactions to direct or indirect pressure from other clans and tribal alliances. Through their more decisive group action, other clans may pressure a divided community to act together: A Suolol group meeting was called to discuss when a display of wealth as well as its distribution should take place. It ended inconclusively with no clear sense of consensus. One minor Suolol big-man, Mol, visited his affines in the allied Surup tribe shortly thereafter and announced to them the date he had been advocating during the Suolol meeting. It is unlikely that his affines considered Mol's date to represent the sense of the meeting, since they were aware that Mol was not a particularly influential man in Suolol. Nevertheless, they had their own reasons for wanting to ensure that Suolol hold the event then. They agreed to arrive en masse on the suggested date, which Mol reported back to his group. His fellow clansmen were not pleased with this. Several of them said that if they went to the Surup to call the event off, they would reveal publicly to Surup both Suolol's lack of readiness and Mol's lack of judgment. In the end, they assembled sufficient wealth for the prestations in time for Sump's arrival - in effect implying just the opposite.

In this case, Mol's action was effective not so much because of his influence either within or outside Suolol as because he had introduced a new element into the Suolol deliberations, given a general state of rivalry between Surup and Suolol. It appeared that the Surup were unambivalent about shaming their Suolol allies. Their unanimity and decisiveness provoked a similar response in Suolol. Before that, several equally compelling but mutually incompatible arguments and plans (mostly relating to intra-Suolol concerns) had been advanced by members of different Suolol factions, and no clear plan for collective Suolol action had emerged. Big-men and others who advocate clan action engage both in vigorous politicking within their own groups and preemptive individual actions of the sort just described. They advocate particular group policies, encouraging a consensus, or they may try to shame fellow clansmen into action by putting on a public display of group-spirited hard work (like clearing grass from the ceremonial ground, where sem onda events take place). They also announce plans and dates for events before they have been agreed to by members of their groups, in an effort to draw outsiders into the discussion. Ultimately, group decisions are arrived at against a background of political competition with other clans. Although discussions concerning sem onda activities aim at consensus, individuals do not wait for unanimity before acting. With divergent extraclan involvements of clan members, a nonauthoritarian structure of power, and

30

What gifts engender

decentralized political relations between groups as social givens, to require unanimity would make social projects impossible. But this Hobbesian nightmare is not reality. Despite what appears as a "state of nature," large-scale projects involving the cooperation of hundreds of people take place regularly. Skillful participants in group discussions recognize that whereas no one has the authority or power to prescribe individual action, they can shape the meanings of the actions individuals undertake by shaping one public interpretation of events. Once matters of group policy are articulated publicly in a "unitary" fashion and for outsiders to hear, individuals are free to choose what they will do, knowing how their behavior is likely to be construed by others both inside and outside their group. An individual by himself or herself cannot control these interpretations. They are public social constructions either actively and consensuaily produced (as when people announce that they will sponsor a festival in the name of their sent onda) or passively accepted as the product of other people's making (as in the Surup/Suolol case cited earlier).9 Successful cooperative action in Mendi takes a form that presupposes the existence of a particular sort of personal autonomy. At the same time, that autonomy is reproduced in a social world constituted not only of an assortment of personal histories and relationships but also of clans and their histories. Neither the group nor the individual is dissolved into the other. Rather, the tension between them generates much of the factional politics of Mendi life, and some more subtle conflicts to which we will return later on. Social relations between clan members Two aspects of the internal organization of sent will be considered more closely in this section. They concern the "property" relations - that is, social relationships mediated by valuables (pigs, pearl shells, and money) and by land by means of which the cooperative organization of clans is socially reproduced. The continuity of the land Clan members constitute a corporate group with respect to land. Clans like Kurelka and Molsem (sem onda composed of about twenty to sixty households) are the most important territorial units for gardening, house building, grazing pigs, and collecting bush resources. Each is associated with several contiguous blocks of land.10 Since allied clans (e.g., Molsem and Kurelka in Upper Mendi) are typically neighbors, they appear to hold a territory in common. But clan alliance groups (tribes or "clan clusters" like Suolol of about 150 to 300 households) do not regulate access to land as units. Thus, a member of the Suolol tribe is not permitted to clear land anywhere on Suolol lands

31

Sem relations: solidarity and its limits

but only land associated with his own clan; its Kurelka members use Kurelka land, and its Molsem members use Molsem land. In the northeastern Mendi Valley, clan land often includes a variety of types: flat land along the Mendi River, land of moderate to steep slope in the main garden areas, and perhaps some steeper "bush" land along the forested ridges. The situation is different in the northwest (Karint CD), where many groups do not have forest resources on their own territories. They are given access to bush land by their exchange partners living in Upper Mendi. Gardening practices also vary. Thus, whereas in Upper Mendi, in the northeast, sweet potato mounds are often harvested three times before breaking them open and replanting them, in Karint fewer harvests are reported per mound. In contrast, the northeast, at a higher average altitude, is more susceptible to frost, and residents depend on people living at lower altitudes for runners with which to replant damaged gardens. However much their resources may differ, all clan territories have open, grassy ceremonial or parade grounds (koma) around which one or more men's clubhouses are grouped. These are places in which community meetings and both large and small public events are held. In the Senkere community, there were two active koma (in Wepra and in Senkere, on Kurelka and on Molsem territories, respectively). The Senkere koma had been enlarged during the period 1968-74 to accommodate long ceremonial houses, which were built in preparation for a large pig kill Suolol eventually sponsored in December, 1979. n Wepra koma had been expanded similarly during the early 1950s, when it was the site of a previous Suolol pig festival. Several other koma exist in the community, each associated with a particular subclan, but during 19779, all large-scale events held in the names of clans took place at the Senkere or Wepra koma. The social centers of dispersed Mendi communities, koma are concrete representations and continuous reminders of the social reality of groups, themselves only periodically realized in action. In the Senkere community, garden surveys indicated that the plots of people affiliated with Molsem and Kurelka were not intermixed. Molsem and Kurelka lands were marked off from each other in some places by natural boundaries such as water courses or ridge crests and in other places by conventional signs. The boundary running down the middle of the main garden slope at Wepra, separating Molsem and Kurelka gardens, was indicated only by planted cordyline shrubs, fences and ditches, markers indistinguishable from those separating the gardens of lineage brothers (compare Ryan 1961). Each clan's territory is divided into several named blocks. The gardens of members of the various subclans (e.g., Olsem, Pulumsem) are intermixed to some degree, although particular areas within blocks tend to be associated with particular subclans, each of which has garden areas in more than one named block within the clan territory. Each subclan has a number of "house

32

What gifts engender

places" in blocks, forming neighborhoods. Generally, people from closely related lineages (e.g., two brothers and their married sons) build houses close together; their houseyard fences form a continuous barrier between the gardens located around their houses and the village paths along which pigs often browse. The coherence of clan land holdings is maintained by cultural conventions that keep intergenerational land transfers within the clan almost by definition. One has the strongest type of control over gardens that one's father or one's father's father cleared and planted: Both use rights and the right to redistribute garden plots across generations. This control is usually legitimized by reference to family history: the claim that one has seen one's father or grandfather working there or that one's father had pointed out the trees he had planted, fences he had constructed, or houses he had built on it. All are the conventional signs of one's father's long-term use of the land. Old people have an important role in validating such claims in land disputes. Besides the particular garden plots one acquires from one's father, one has the right to clear and plant on land anywhere on one's clan's territory provided a spot is not currently claimed. When a man dies without sons and without himself giving his gardens to a particular heir, anyone in his subclan, and then clan, may claim it. Just as in the case of clan members' corporate identity and responsibility for each other's actions then, clan members have a corporate identity with respect to clan land. In Mendi land is not yet short, and no clansman - no matter what the nature of his participation (or lack thereof) in clan affairs - is denied access to it. For example, one member of the Senkere community was a recluse who had neither engaged in transactions with exchange partners nor participated in clan affairs for many years. During the early 1950s, he had been jailed by provincial authorities for having participated in a tribal fight. The experience of incarceration had apparently been traumatic. After he was released, he avoided public places, not even coming into society to mourn the death of his first wife. His second wife and his children participated in a normal way in the life of the community, but the man confined his activities to working in gardens at the outskirts of the settlement and to maintaining fences for the gardens that his wife planted near their house (where he returned at night to sleep). Because of his utter lack of involvement in community affairs, he was considered crazy but his garden rights were never questioned. All of this seems like a straightforward case of the patrilineal inheritance of garden plots, but things are not quite so simple. Female agnates have very strong claims on gardens, with which men sympathize saying that their sisters (daughters) would control gardens if only they did not move away from their natal place when they married. Indeed, married female agnates resident in the Senkere community controlled as many gardens, on the average, as resident married male agnates did. What is more, they could transfer those gardens to

33

Sem relations: solidarity and its limits

their children so long as the children became permanent residents there as well. The land tenure rights of the children of a clan sister's son were not distinguishable from those of people with longer "agnatic" pedigrees. The distinctive feature of those people who can transfer land across generations thus appears to be residence in the place of a parent, and not gender or agnatic descent. (Clan members may also give use rights in specific garden plots to outsiders, but these are in most instances temporary.) It would be inaccurate, however, to leave the matter there. Although its influence is subtle, gender is a critical factor in land inheritance. One has to ask why women tend to move to the homes of their husbands upon marrying, giving up their residential claims to garden land in their father's place. The matter cannot be dealt with in detail here, but certain features of the division of labor - particularly its symbolic ordering - suggest an answer. As I noted above, a man legitimizes his rights to the land his father and agnatic ancestors used by pointing out evidence of that use - particularly trees, or cordyline shrub boundary markers that those men planted or the remnants of fences they constructed and houses they built. All are culturally recognized as relatively permanent transformations of the land. And all are specifically male production responsibilities. Women's daily work in the gardens - composting, planting, weeding, harvesting, and rebuilding sweet potato mounds - is important for maintaining human life and energy, producing wealth (pigs), and creating and strengthening social relationships (by means of the generous sharing of food with guests). Whereas these achievements (and others, such as the manufacture of netbags) are recognized, nevertheless, women's work is not culturally understood to transform the land in a permanent way, whatever the "objective" facts may be. The necessary quotidian repetitiveness itself confirms the conventional judgment concerning its ephemeral character; what must be redone every day is quickly undone, leaving little trace. Women's work, as such, is not inscribed in enduring structures, material or social, in the way in which men's work is thought to be in Mendi. Not objectified in this way by convention, it is a source of transcendent power only exceptionally, through active individual effort. Normally, a women's labor entitles her to control the disposition of that which she herself has actively planted or made. In contrast, the products of men's labor endure, becoming the legitimizing signs of men's right to control the redistribution of land, the society's most important productive resource. This fact may make post-marital residence practices in Mendi intelligible. If so, one cannot take at face value the solidarity brothers feel for their sisters vis-a-vis land rights. Although it is significant in ways that will be explored later (Chapter 4), in the present context the sentiment obscures the structural role of a gender asymmetry, making the asymmetry out to be an accidental concomitant of postmarital residence instead of a central organizing relation. More can be said about the role of

34

What gifts engender

asymmetrical gender constructions in the making of sent; I will return to this theme again most explicitly in the last section of this chapter. Male labor is memorialized in its enduring products and is a part of the land as a persisting clan resource. In turn, ideas about the land are an important part of a larger conceptual system concerning how people come to be related as clan members and how clans are socially reproduced. In common with other Highlanders, the Mendi believe that the shared substance of kinship can be actively acquired and that it is not passively inherited simply by virtue of birth, "blood" or "semen." Whereas important substantive connections are acknowledged between parents and children, kinship connection to people with whom one has no such tie can be achieved by eating, living, and cooperating together. As the clan composition of the Senkere community and other facts I have related imply, the Mendi adopt individual affinal and cognatic kin into their clans (not to mention whole groups of nonagnatic kin) relatively easily, giving them blocks of land on which to plant and build houses - and otherwise to establish their children as full members - as long as they demonstrate their commitment to their new clan by participating in clan projects. Eating together is particularly significant. Food itself can be understood as another symbol of shared substance. The land and the food grown there may mediate the relationship between present and past generations of clan members. The Mendi believe that the spirits of clan ancestors (su temo) exert control over living people. They pervade the food one eats, as it was grown on clan land, where they were buried. For this reason, many believe that cases of fratricide need not be punished directly. In fact, to do so would risk escalating the conflict by drawing a wider circle of relatives into it. Instead, the angry spirit of the dead man itself is supposed to poison every mouthful of food the murderer eats from his home gardens. The only safe route for him is exile.12 Apart from the power of its products, garden work itself (like exchange) demonstrates the pervasive duality in Mendi of the autonomy of individual action and the cooperation and solidarity of the clan. On the one hand, gardening is not typically a communal activity, nor are gardens collective products. Men and women generally do not form work parties when they clear, fence and tend their plots. These tasks can be achieved by means of household labor alone since gardens in Upper Mendi are permanently cultivated or under very short-term grass fallow and the preparations for planting usually involve only clearing grass and rebuilding portions of old fences rather than felling trees and other heavy clearing. Individualized production applies even to fence construction. Fences appear to be joint endeavors, running continuously along village paths and enclosing large areas including the gardens of many people. But in fact, each man maintains those portions that abut on garden plots that he claims. Fences are, it will be recalled, important signs of land rights; there-

35

Sem relations: solidarity and its limits

fore, to let someone help build a fence on a plot one wishes to retain or to give to one's children may compromise one's claims. On the other hand, gardens in Mendi are not individually owned. The sem onda retains final rights in clan territories as a whole; if a sem kank has no heirs to inherit its gardens, other members of the same clan may claim them.13 Nor does the level of production respond merely to the subsistence requirements of households, however much the Mendi assertively insist that their garden work is their own business. The intensity of garden work, however individually accomplished, reflects a person's commitment to the demands of exchange partners and fellow clan members. Sharing and pooling wealth The ways Mendi clan members help one another in wealth exchanges also demonstrate their participatory corporate relationship. As events, many prestations appear to be individual affairs until one discovers the forms of cooperation that made them possible. Mortuary ceremonies (kowar) - gifts given publicly some time after a death to the deceased's maternal relatives - are often sponsored by individuals in their own names. Inquiry into the sources of pearl shells, money, and other items included in such prestations reveals two kinds of aid. First, kowar sponsors request valuables from their twem partners. This wealth will have to be replaced with items of equal or slightly greater value. Typically the terms of repayment are discussed explicitly by the sponsors and their partners at the time of the request. Second, sponsors of a kowar expect to receive wealth from fellow clan members without asking. Wealth received this way from fellow clan members is "free" (paeme); it is unsolicited and given