Dr. Chris Ballard, ANU
Dr. Chris Ballard is an Associate Professor at Australian National University. He is Senior Fellow of the School of Culture, History & Language, School of Culture, History & Language.

Trade networks and sacred geography in the Papua New Guinea Highlands

by Dr. Chris Ballard

The critical potential of a ‘social landscape’ approach to the past lies in its attention to questions of human agency and scale in the archaeological explanation of change. However, an ethnographic study of trade relationships between Huli-speakers of the Southern Highlands of Papua New Guinea and their neighbours suggests that landscapes are a matter of perspective. A ‘hidden’, subterranean landscape described in Huli cosmology is shown to be implicated in the structure of regional trade. Distinguishing between the landscapes of belief within which historic agents acted and the social landscapes identified by external observers is an important step towards framing questions for archaeological research that can exploit the insights of both perspectives. In a series of papers, Gosden (1986, 1989, 1993) has developed the notion of the Social landscape’ as a frame- work for archaeological enquiry and explanation. Social landscapes are primarily a means of identifying the most appropriate spatial and temporal scales of analysis within which ‘to examine how groups organize themselves on a local regional scale to meet social goals and to link these forms of organisation to the archaeological record they leave behind’ (Gosden 1989:45). By specifying principles which structure social process at particular spatial and temporal scales, changes in prehistoric society can be deduced from change identified in the archaeology of corresponding social landscapes. The real value of this approach is that it introduces issues of human agency to archaeological explanation while simultaneously tackling the problems of scale which have hampered previous efforts at relating archaeological material to anything more profound than technological change: ‘it is only by appreciating the social motives of individuals acting on a scale so as to be visible archaeologically that we will be able to view social change as a whole’ (Gosden 1986:184). A critical component of the social landscape approach is Gosden’s insistence that we reverse conventional archaeological procedure and take the recent past as our point of departure. Working from an understanding of the social processes underwriting the generation of the archaeological record in the recent past, differences identified in earlier records can then be compared witthis baseline as a means of reconstructing corresponding changes in society. As Gosden suggests, this requires the adaptation of ‘anthropological principles’ to suit the expanded frame required for archaeological rather than anthropological enquiry. The case studies to which Gosden has applied his social landscape approach have largely concerned the question of trade, as the process which most obviously produces a material residue that proclaims its extra-local origins and lends itself to regional study. The social landscapes identi- fied at Mailu and in the Arawe islands are thus primarily the networks of trade linking local centres of production. As a strategy, this focus on trade is in keeping with the general ethic underwriting the social landscape approach of working from the known to the unknown. But I want to suggest in this paper that the ‘ethnographic past’ which provides our point of departure is capable of furnishing us with a considerably broader definition and under- standing of the concepts of social landscape and of trade.

Archaeologists working in Melanesia have been surprisingly coy about addressing the wealth of ethno- graphic material available to them. In part, this reflects the very different concerns of conventional anthropology and archaeology in the region. But the possibility of fully exploiting the potential of the social landscape approach surely rests on the ability of archaeologists both to engage the ethnographic literature and, more importantly, to produce ethnography ourselves in such a way that it does address archaeological concerns. We need, in short, an archaeological ethnography which does not restrict itself to trawling the ethnographic literature in search of direct or technical analogues and their immediate social con- texts. It requires original ethnographic research that employs the methods, conditions and attention to detail of anthropological ethnography, but that pays particular attention to the relationships that obtain between the local and regional scales, between smaller and larger social units and between short-term and long-term social processes. If archaeological ethnography is more exten- sive in its scope and scale of enquiry, its models and conclusions are also likely to be framed in terms more general than those admissible in anthropology. An ethnographic approach raises, in turn, a number of problems implicit in the social landscape model. Key features of Gosden’s exposition of the use of social landscapes in archaeology ‘past systems of production were ordered through social rules, rather than in reaction to the structure of the environment’ (1989:49) and that ‘landscapes are ordered into social configurations’ (1993:131).

Although Gosden’s intention is to challenge the agentless landscapes of conventional archaeological explanation, the starkness of this opposition between society and environment perhaps obscures a more fundamental challenge for archaeology: an understanding of the ways in which constructions of society and environment are mutually implicated. As Thomas (1993) reminds us, different landscapes assume the adoption of particular perspectives. In some sense, landscapes’ are thus always already ‘social’; there are no landscapes in nature against which to compare our ‘social’ landscapes, merely other social landscapes. This then requires that we recognize an analytic distinction between the models developed by archaeologists (‘social landscapes’ for example), and those models (‘cultural’ landscapes, perhaps) employed and strategically deployed by historical agents; a distinction that remains analytic because the interpretation of such actors’ accounts is also necessarily conducted from the perspective of ex- ternal observers. Thus an ‘external’ perspective is capable of discerning patterns on a temporal or spatial scale not accessible to historical agents: characterisation studies, for example, indicate to archaeologists the precise origins of stone axe blades recovered hundreds of kilometres from their source, knowledge that was not available to the final users of those blades. ‘Internal’ models, by way of contrast, provide us with a sense of the bases upon which historical agents made decisions, reasons not necessarily matched by the ‘economic’ or ‘subsistence’ accounts proffered in external accounts. Archaeological narratives traditionally run the risk of conflating these two perspectives, by attributing to historic agents decisions that could be made only on the basis of an external perspective, and by reading external models back into the past as the internal models of historic agents. Viewed in this light, the social landscapes described by Gosden are external constructs which, while valid as accounts of the distribution of people and of materials, may have little to tell us of the cultural landscapes within which historic agents acted. Perhaps the challenge is not to identify a preference for either external or internal models, but rather to establish procedures for enquiry that work between the two perspectives as a means of exploring the blindspots of both. This paper describes some of the social and environ- mental contexts for the trade networks that have his- torically extended between Huli-speakers and their neighbours in the western region of the Papua New Guinea Highlands (Figure 1). The regional distribution of those material resources sought by Huli-speakers and their neighbours suggests an external explanation for the structure of Huli trade founded on the relative absence of these resources within Huli territory. Internal accounts of trade, however, stress the extent to which Huli trade in exotic materials served largely to finance exchanges within Huli society. Huli trade with Studies of trade in the Papua New Guinea Highlands and in Australia The ‘natural scarcity’ or ‘ecological’ model of trade adopted by early studies in Highland New Guinea viewed trade largely as an adaptive means of redistributing scarce or unevenly distributed resources, as ‘part of the process of human adaptation to a number of contrasting environments’ (Hughes 1973:125). Trade items were distinguished as either ‘utilitarian goods’, such as salt and stone axes, or ‘nonutilitarian valuables’, including brideprice axes, bird feathers and shells (Rappaport 1968:106), a rationale which identified their respective differences in relation to the process of adaptation. While not denying that one function of trade (and one that is articulated by High- lands people) is precisely to redistribute scarce resources, it is evident from a number of more recent studies that this is not the sole basis for understanding trade activity (Sillitoe 1978, McBryde 1984, Healey 1990). With the possible exception of trade for salt, com- munities in the Highlands regions are not ecologically or physiologically enjoined to engage in trade. For example, Burton’s (1984) analysis of stone axe production and trade from the Wahgi valley quarries has effectively demon- strated that there was no absolute need for most com- munities in the Highlands to import axe blades. He illustrates the point with reference to the Huli who, located at some distance from the major axe sources, were capable of clearing large areas of forest using a high proportion of poorer quality axe blades produced locally from river pebbles. Yet despite this potential for self- sufficiency, most Highlands communities engaged in trade for axes. Thus the Maring, with whom Rappaport worked, persisted in importing axes from the far side of the Jimi Valley and the distant Wahgi valley despite the presence on their own territory of the little-exploited quarry at Repeng which contained material comparable in quality to that of the imported blades (Burton 1989). This apparent lapse in ecological good sense on the part of the Maring is explained by Burton in terms of the nature of production at the Jimi and Wahgi quarries, where a primary concern for the production of larger ‘ceremonial’ blades had, as one consequence, an over- production of work blades which ‘flooded’ the surrounding market.

Figure 1 Map of Western Highlands region

Trade networks are thus situated within broader social contexts which need to be identified in conjunction with analyses of resource distribution. Other recent studies have turned towards the role of trade as one of a range of forms of exchange, looking beyond the issues of production and distribution to the question of demand, to the ways in which the desire for specific materials is created and sustained within communities. In a ground-breaking study of one of the principal trade items in the Highlands, Clark rejects the ‘ecological’ explanation for the value attached by many Highlands communities to pearlshell, which posits that the value of pearlshell derives simply from its scarcity, to ask instead, ‘[o]f all scarce things, why did pearlshells come to carry such a symbolic load . . .?’ (1991:310). By tracing pearlshell symbolism amongst the Wiru through to a core of cosmological and aesthetic principles, Clark is able to refigure the meaning of pearlshells and, by extension, of the grounds for trading in them. For the Wiru, Clark suggests, pearlshells represented not simply the tokens of contested power amongst men, but the embodiment of all social and cosmological process; physical markers of the flow of human substance mirror- ing the movement and reproduction of women and men. Although Clark speculates about possible precursors in Wiru symbolism for the crescent form in which worked.

pearlshell was traded, the difficulty of nominating causal links in such a situation is apparent: no conclusion about the relative priority of either the shells or their related symbolism is possible. At an archaeological scale of explanation, we need therefore to conceive of a form of ‘mutual constitution’ in which the historic links between Wiru symbolism and pearlshell form, for example, are established but need not themselves be placed in a particular causal relationship. Change in one is assumed to reflect, and be reflected in, change in the other. In its suggestion of a cosmological foundation or logic to the trade in shell, Clark’s work finds parallels in studies by archaeologists of the relationship between Dreaming tracks and trade routes in Australia. Thomson’s (1949) early description of the relationship between ceremonial cycles of obligation and the movement of trade items in Arnhem Land situated the regional exchange system within a landscape of horizontal quarters, or kumur, of exchange sources. As Jones and White have since stressed in their archaeological analysis of Ngilipitji, one of the principal quarry sources in Thomson’s system, an economic account of quarry production, while necessary, is only one of a range of ‘planes of understanding about Ngilipitji, its ritual role, and its mythology which transcend such utilitarian precepts’ (1988:84). McBryde (in press) has expanded the analysis of the ritual foundations of exchange to a regional scale in her study of the trade networks of the Cooper and Lake Eyre regions. Here, she describes the position not only of the trade sources but of the trade routes themselves within a series of sacred landscapes. The same cosmological myths which establish the sacred tracks of the Dreaming operated as ‘charters’ for regional trade. In explicitly recreating the myths of the Dreamtime, this trade was itself a form of ritual, distributing not only material items such as ochre, pituri, grindstones and axes along the Dreaming tracks, but also a steady flow of Information, new craft techniques, songs and ceremonies’ (McBryde in pressi). If there is an encompassing framework for the geography and the practice of trade in the region, argues McBryde, it is to be found in local cosmology, in which, ‘[t]he meanings of the goods and the significance of their source location are explained and maintained over time by the stories that also map the symbolic and geo- graphical world of these long distance transactions’ (in press: 24). As in the case of Wiru symbolism and pearlshells, it is impossible to establish priority for either the Dreaming myths or the trade networks; they are, for archaeological purposes, mutually constituted. The value of this association is the scope it provides for tracing the historical depth of the cosmologies underpinning the networks of trade and ritual performance through the archaeological evidence for trade. Cosmology, which explains the struc- ture of the universe and the nature of the relationship between people and their environment, also accounts for the ‘value’ of different materials, substances and relation- ships. It is thus cosmology which imbues the environment with significance and through which ‘cultural’ landscapes are perceived. The modelling of a social landscape for Huli trade must therefore include an understanding of the symbolic landscape of Huli sacred geography. Exchange in Huli society Huli-speakers of the Southern Highlands Province in Papua New Guinea, who number more than 60,000, constitute the second-largest language community in the Papua New Guinea Highlands. They occupy a linked series of swampy inter-montane basins at altitudes of between 1440m and 1900m above sea level, though fringe Huli communities are found as high as 2500m and as low as 1200m. To their north is the high Central Range, and to their south the lowlying Papuan Plateau (Figures 1 and 2). Sporadic contact between Huli-speakers and European miners and administrators between 1934 and 1945 was followed by the establishment in 1952 of a permanent government post and an airstrip at Tari, and the sub- sequent influx of different Christian missions. Huli society has been extensively described by several ethno- graphers (Glasse 1968, Goldman 1983, Frankel 1986) whose accounts suggest a degree of cultural homogeneity uncommon for so large a community. This reflects the presence in Huli culture and society of a number of regionally atypical features which also significantly structure the conduct of exchange and trade. The account of exchange and trade in Huli society given here reflects a period of Huli history in the 1920s, immediately prior to contact. It is based primarily upon the oral testimony of living Huli in each of the major basins but also draws upon the writings of colonial officers and other researchers in the region. Over time, a cognatic rule of residence which gives individual Huli the notional right to reside on the territories of any direct ancestor, male or female, has produced a complex pattern of variable group composition in which clan agnates are usually outnumbered by non-agnatic cognates on their own clan territories, or parishes. Men, in particular, can retain rights to residence and land in as many parishes as they are able to maintain links with by providing assistance to other parish members and actively using their land in the parish. This has had the effect, historically, of redistributing the population in relation to local and temporal variations in agricultural potential, but has also enabled individual men to traverse the full range of Huli territory, moving from one group of kin to another, and thus to trade with different neighbouring language communities.

Exchange is so fundamental a feature of Huli cosmology and society that it resists abstraction. The Huli universe is itself constructed upon a notion of exchange between living humans and ancestral and non-ancestral spirits. Much of Huli ritual performance involved the conceptual exchange of pork and pig fat with these spirits in an attempt to maintain the stability of the universe and restore fertility to the land. The concept of entropy, of a continuous process of decline in the land and in society, is a pervasive theme in Huli thought (Frankel 1986:26f). Entropy is also a property of the pool of mana, ‘knowledge’ or ‘skill’, made available to humans by the earliest ancestors. The loss through death of men and women endowed with mana and with the gamu spells that provide human purchase upon the universe is cited as evidence of this process of decline in overall mana. In an ironic echo of the ‘ecological’ trade model, mana and gamu are regarded by Huli as unequally distributed resources. Goldman refers to a ‘compartmentalization of knowledge’ in Huli society (1983:96), to the extent that specific individuals and clans are associated with the most accurate versions of particular forms of knowledge. Certain types of mana, such as dindi malu, the right to recite a clan’s origin story and its genealogical claims to land, are still reserved to the holders of hereditary clan leadership. Genealogies (malu) are also the preserve of kin groups, for whom they continue to represent a strategic resource in disputing the ownership of land. As such, genealogies are not available for public trans- action, but neither are they the subject of payments amongst kin. Other forms of knowledge and spells were formerly more widely circulated, but only in return for payment. Significantly, Huli views on exchange do not discriminate between material and immaterial transactions.

Figure 2: Language Communities in Western Highlands

Appropriate behaviour and preserving spells, known collectively as ndi tingi, were paid for by newly married couples. Spells for a wide range of functions were also regularly exchanged for payments in items such as pork, cowrie shells, red ochre and tree oil. Some of the knowledge and spells held by one’s parents required exchanges for transmission: former ritual leaders describe the sequence of payments in pigs and cowries made to their fathers in exchange for mana knowledge. Payment {yole) was also required for the performance of sorcery and rituals. Officiants in each of the various Huli rituals being performed at contact were paid in accordance with the scale of significance and relative rarity of the ritual involved. The specialist production of certain material artefacts also reflected the discontinuous distribution of mana knowledge. Only a few men within each clan held the mana and gamu spells required for the pro- duction of the human hair wigs worn by almost all adult Huli men; others held the knowledge for making shields, hafting axes or flaking stone tools. Importantly, even where a productive activity such as axe-hafting was formerly undertaken by most men, those individuals who possessed the appropriate mana were still identified as producing the ‘truest’ or finest forms. These payments for specialist knowledge and production aside, the vast bulk of exchange within Huli society occurred within three broadly distinct contexts: ritual performance, compensation and marriage, for each of which pigs and pork have been the principal items and symbols of exchange. Most of the rituals identified by Huli as ancient required the use of pork or pig fat, but seldom in any great quantity. However, the development from the latter part of the 19th century of new forms of Huli ritual saw the increasing deployment of large numbers of pigs in ritual contexts. The most significant of these ‘new’ rituals, tege, involved competitive cycles of reciprocal exchange of pigs and pork in numbers matching those of the major compensation and marriage exchanges (Ballard 1994). Compensation (abi) for physical and perceived injuries or for deaths, usually in the form of dead or live pigs, represented a massive structuring influence in Huli society, uniting kin in the assembly of wealth whilst promoting exchange between groups unrelated either affinally through marriage, or through descent (Goldman 1981, 1983). Exchanges conducted within the context marriage, identified generically as wariabu, have pro- vided not only the primary means of defining social groups, but also the model to which Huli themselves refer in accounting for the practice of trade with unrelated or non-Huli partners. The critical relationship established af finally is that between people who refer to each other as aba. Minimally employed as the reciprocal term defining the recipients of certain categories of brideprice payment, aba is also extended as a class term to all of one’s non- agnatic cognatic kin, as persons between whom reciprocal economic, ritual, and social obligations exist’ (Gold- man 1983:234). Indeed Goldman (1983:86f) has identified aba relationships as the basis for exchange behaviour amongst Huli, insofar as they cut across parish boundaries and establish wide-ranging and complex kin net- works across Huli society.

If aba relationships thus stand for Huli as a model of the conduct of exchange, normative descriptions offered by Huli of the nature of aba relationships are an important means of access to the meaning of exchange more generally. Goldman (1983:86) has shown that aba are conceived of as being centrally placed between the speaker and other non-agnatic kin, or exchange partners. Another term, dombeniali, literally ‘middle-man’, is employed to further stress the mediating role of aba. These middlemen play a critical role within Huli society, particularly in disputes between groups of agnatic kin between whom they are centrally placed, where they negotiate the terms and transfer of abi compensation payments. As we shall see, it is as dombeniali, or middle- men, that Huli conceive their role within the network of regional trade. Further, the regional network of trade relationships between Huli and their neighbours are described in terms that reflect the model of craft specialization and the compartmentalization of knowledge which Huli employ to describe their own society.

While there is no generic label for the range of forms of exchange described above that Huli conduct amongst themselves, what I refer to for the purposes of this paper as ‘trade’ is marked linguistically by Huli as yole. No clear boundary exists in Huli practice between ‘exchange’ and ‘trade’ but the distinction has value for my purposes. ‘Exchange’ is characterised by enduring obligations, whether in terms of the norms of behaviour that surround kin and affinal relationships or through the ties formed by the sequence of injury and compensation, or cycles of ceremonial tege exchange. ‘Trade’, in contrast, while exploiting kin and affinal ties, is conducted primarily with non-kin and entails no subsequent obligations beyond immediate reciprocity. Huli thus describe as yole the use of pigs, money and, formerly, shells in exchange with other Huli for items such as mineral oil, pandanus nuts, knowledge or imported items. Trade and the other forms of exchange were and still are conducted with other Huli as well as non-Huli, as the following discussion of resource geography, ethnicity and trade illustrates. Multiple geographies: ethnicity and resources Like Thomson’s Arnhem Land communities (1949:71) Huli describe the geography of trade in terms of immedi- ate extra-local sources, conceptually ordered as a series of cardinal points. The fundamental axis of orientation in Huli accounts of their visible landscape is between upland, headwater (mane) and lowland, downriver (wabi) regions. From the series of adjacent valleys in which most Huli reside, the critical topographic distinction is thus between the higher valleys of the mountain ranges to the north and east and the lowlying Papuan Plateau to the south. These two cardinal directions, upriver and down- river, correspond to the major ethnic groups with whom regional trade is conducted, known to Huli as obena and duguba respectively. The other axis, running between the area occupied by Duna-speakers to the northwest and the Foi and Fasu of the Lake Kutubu region to the southeast, is qualitatively distinct for Huli, for reasons that will surface through the course of this paper.

Figure 2 shows approximate territorial boundaries for speakers of Huli and some of the neighbouring languages. The Huli language is most closely related historically to the other members of the Enga Language Family, including Enga, Mendi, Ipili and Kewa (Franklin et al. 1978:89-90). The high percentage of cognates shared with the neighbouring Duna language (up to 32%) prob- ably reflects a considerable antiquity for contact between the two languages, which are in fact held to derive from different language stocks (Wurm et al. 1981). Cognate percentages obtaining between Huli and the Papuan Plateau (Kaluli, Onabasulu, Etoro, Bedamini, Kware, Kasua and Sonia) and Kutubu (Foi and Fasu) languages are much lower. Language community sizes vary dramati- cally, with much greater population densities found in the central valleys of the Highlands than on their lower margins. Recent gross estimates for the Huli (62,000), Duna (17,000) and Mendi (45,000) contrast strongly with the total populations estimated for the Papuan Plateau groups of about 7000, and for the Lake Kutubu groups of about 5000. Huli conceptions of ethnicity bear little resemblance to the results of lexicostatistics and rather more to local geography. A specific narrative genre details the origins of the hela peoples at a site in what is now Duna territory: four distinct groups, descended from a common putative ancestor, and each with their own language and distin tive material culture attributes (including those items historically sought by Huli in trade, such as salt and tree oil), are held to have migrated from this site to their present territories. The hela obena correspond roughly to speakers of the other languages of the Enga Language Family: the Ipili, Enga and the Mendi (including Wola). The hela duguba are those of the Papuan Plateau groups to the south formerly known to the Huli, including speakers of Kaluli, Onabasulu, Etoro and Bedamini. Duna-speakers are recognized by Huli as hela duna. Hela huli is the term which Huli use to refer to themselves. A number of other linguistic groups are known and named, but are not considered to be related as the ‘sons’ of hela: these include the hewa (Foi- and Fasu-speakers), bogaya (Bogaia-speakers), tinali (variously identified as Febi, Agaia or Tsinali) and kora (speakers of the Oksapmin and Bimin languages). Similar narratives are recounted by Duna and some Ipili and western Enga, but few of the members of the other linguistic groups thus identified share this Huli-centric view of the world. For Huli, however, the hela narratives order the geography of trade between these conceptually related groups, stipulating the distribution of scarce resources. What I want to suggest is that much of Huli concern in the past with their geographic position in respect to the northern obena and southern duguba has been related to the distribution of specific resources in these two areas. Figure 3 shows the principal sources of a limited but sig- nificant range of items traded through this region. Seven key, or ‘major’, materials dominated Huli trade interests: stone axe blades, salt, tree oil, blackpalm bow staves, shell, mineral oil and pigs. Of these, only the last two were produced by Huli themselves. The strategies of individual Huli traders appear to have been organized broadly around the ability to trade tree oil and blackpalm bows from the south against axes and salt from the north. Note that the distinction drawn between major and minor trade items is not yet supported by quantitative data, but rather reflects Huli statements about their relative significance. The major riverbed sources and quarries for stone axe blades are located far to the northwest and to the east of the Huli and there was no knowledge amongst Huli either of the locations of the sources or of the nature of produc- tion of the different blades that reached Huli through trade with intermediate communities. The general direc- tions of incoming blades together with physical charac- teristics, principally those of colour but also overall size, were employed by Huli to identify blades from different sources. Table 1 lists the more common Huli terms for axe blades, and comments on possible derivations. Huli produced their own axe blades from local cobbles classing them, together with blades recovered from abandoned house and garden sites and reworked (some of which appeared to be from exotic sources), as ‘earth-axes’ (dindi ayú)’ neither were held to be as strong as the various blades known to have been imported from the west or east. Of the four principal regions of salt production in the western part of the Papua New Guinea Highlands (the Lower Wahgi, Kaironk/Simbai, central Enga and Ipili sources – Figure 3), Huli probably only ever had access to the two regions nearest to them. The salt from the Enga and Ipili sources (Meggitt 1958, Moses 1978, Mangi 1988) was as extensively traded as quarried axe blades, moving west across the Strickland River to the Bimin and Oksap- min, south as far as the Foi, Fasu, and Kasua, and east to the Wahgi and Jimi valleys (Hughes 1977). Table 2 lists apparent cognate terms for the Enga and Ipili salt amongst some of the communities who imported it, illustrating the wide-ranging retention of the term for this important trade item. Indeed, the salt traded from and through their Ipili neighbours was sufficiently important for Huli to refer to them as the ibili, literally, the ‘salt- people’. For Huli, as for all their neighbours, salt was primary means of enhancing the ‘sweetness’ of food and was added, as supply permitted, to almost all consumables.

Figure 1: Huli terms for stone axe blades.

Most of the marine shell species known in the Huli area prior to contact probably derived from the southern coastline of New Guinea. The primary shell valuables amongst pre-contact Huli are listed in Table 3. Prior to contact, Huli claim to have had very few pearlshells; although they were highly prized by Huli, there was nothing of the degree of symbolic elaboration associated with pearlshells in Highlands communities to their east, such as the Wola or Wiru. Instead, Huli expressed their preferences from amongst a limited range of cowrie and cowrie-like shells. Of these, Cypraea moneta cowries, which Huli received from Wola traders to the east, were the most highly sought, particularly in light of the demand for them amongst neighbours to the west such as the Duna; both C. moneta and C. annulus cowries were also an important component of payments for ritual performances and, less commonly, a minor component in brideprice exchanges. But throughout the Huli region, shell was used as a mediating valuable in the acquisition of pigs, which were the ultimate token of wealth, and other items of trade or exchange. Why shell never attained amongst Huli the value or symbolic weight accorded to pearlshell by Wiru is unclear; certainly it relates, in part, to the sheer scarcity of pearlshell but it also reflects the dominant status of pigs and pork in Huli exchange.

Sources of Major Trade Routes by Ballard
Figure 3. Sources of major trade items. Mineral Oil: 1. Lebani sources; 2. Bebealia Puni; 3. Tai Yundiga; 4. Tuandaga; 5. Wola sources; 6. Bimin source. Axe Stone: 7. Mt. Stolle; 8. Wario River; 9. Honibo sources; 10. Sambe; 11. Dabiri; 12. Turnan quarries (including Kunjin and Apiamb); 13. Jimi Valley sources (Tsenga, Ganz River, Yambina, Mbukl, Pukl, Apin); 14. Repeng; 15. Dom. Salt: 16. Lower Wahgi; 17. Purari River; 18, 19. Wiru sources; 20. Simbai; 21. Kaironk; 22. Yandapo sources (Liyonggo, Kiowapare, Andepe, Nakatini, Langku); 23. Yalis; 24. Pipitaka; 25. Aukera; 26. Kyoale; 27. Aiyama River (Ipili). Tigaso Oil: 28. Febi; 29. Bedamini; 30. Sonia; 31. Onabasulu; 32. Fasu; 33. Foi. Ochre: 34. Doma River sources; 35. Samo.

Campnosperma brevipetiolata, the tree species from which tree oil or tigaso is extracted, is restricted in distribution to below about 850m in altitude but is found both the northern and southern fringes of the Range (Dornstreich 1973, Sillitoe 1979). The principal producers of tigaso oil in the Huli trade sphere were Foi, Fasu, Febi, Sonia, Bedamini and Onabasulu, Huli also received tree oil through trade with intermediary but apparently non-producing Kaluli and Etoro. all communities in the western Papua New Guinea lands region employed tigaso oil in contexts of performance and display, where the visual effect oily sheen on the skin was held to reflect the health status of the wearer (A. and M. Strathern 1971). Much as Huli expressed a preference for axe from exotic sources, so too were the qualities of lowland blackpalm woods preferred for bow staves and arrows. Although at least one Areca blackpalm (identified by Huli as ayaga) grows on Huli territory, wood from a lowland blackpalm, known to Huli danda (‘bow-wood’; possibly a Ptychoccus species, Hide pers. comm.), is greatly preferred; Sillitoe (1988:6) records a similar distinction made by the neighbouring Wola. The only sources for Huli of this wood were different southern to the south and the tinali to the southwest.

Shell species by Ballard

It should be stressed that the only sources or techniques of production of the trade items shown on Figure 3 that were known to Huli prior to contact, other than the mineral oil sources within Huli territory, were the Ipili and Enga salt sources and an often vague understanding of the nature of the production of tigaso oil, blackpalm and sago on the Papuan Plateau. The ultimate marine origins of the different shell species were unknown, Huli sharing with Duna (Ν. Modjeska pers. comm.), Ipili (F. Ingemann pers. comm.) and Wola (P. Sillitoe pers. comm.), the belief that Nassa shells were harvested from an enormous tree somewhere to the south. Minor trade items imported by the Huli are listed in Table 4, by their respective ethnic sources as identified by Huli. They are distinguished here from major items due not so much to any apparently lower volume in overall trade but because they appear not to have been traded on to other neighbouring non-Huli communities, being imported only for personal consumption or for trade with other Huli. The major trade items were each desired by those Huli neighbours along opposite borders. It is possibly significant that the bulk of the minor trade imported by Huli came from the southern lowlands rather than the higher altitude valleys to the north or east, suggesting that the major trade items from the north, axe blades and salt, were in greater demand amongst the southern duguba and Huli than the major items from the south, tree oil and blackpalm bows. This may also reflect the fact that most Enga also had access to (an admittedly smaller-scale) trade in tree oil and blackpalm from fringe Enga-speakers on the northern slopes of the Central Range (Dornstreich 1973:505) and there may not thus have been as strong a demand for duguba products amongst obena as in the reverse direction. Huli trade Huli territory appears thus to have been relatively impoverished in terms of the major trade items of salt, stone for axes, tree oil, blackpalm and shell. Given that these items were universally (though not equally) valued across the western Highlands region, it might fairly be said that Huli trade had as its objective the acquisition and redistribution of scarce resources. But the manner in which Huli engaged in trade with neighbouring com- munities and, in particular, the ways in which the few items of locally produced trade were deployed conceptually by Huli traders, suggests that such an ecological model, while necessary in describing the geography and directionality of trade, is insufficient as an account either of the structure of the ensuing trade relationships or of the cultural processes and systems of belief underwriting those structures.

Huli terms for salt by Ballard
Table 4  Minor Trade items imported by Huli prior to contact.

Of the major regional trade items listed above, Huli produced only two: pigs and mineral oil. Pigs, which play so central a role at every level of Huli exchange, also ‘fuelled’ Huli trade materially. The intensively cultivated basin floors occupied by most Huli, and in particular the central swamps in each basin, continue to be prime locations for pig breeding. The ratio of pigs to people in Huli, at an average of about 1.7 pigs per person (Wood 1984:198), is comparable to the maximum figures from other areas of the western and eastern Highlands regions (Bourke 1988:Table 2.3) but considerably greater than the average ratios obtaining amongst most immediate Huli neighbours (Table 5). This reflects, in part, the different husbandry practices of the Highlands Huli and Duna who retain boars for breeding and supplement the forage diet of their herds with cultivated sweet potato, compared with the practice on the Papuan Plateau of rearing pigs almost solely on forage, where male pigs are gelded and sows allowed to mate with wild boars. Pig breeding, which requires much higher inputs of labour to meet fodder requirements than pig rearing, is a necessary practice in densely settled areas like the Huli basins where the scope for forage is limited. But an even more telling impression of the scale of imbalance in pig production amongst the Huli and their neighbours can be gauged from the very rough estimates given in Table 5 of variations in population size for humans and pigs within the region. Gardner has previously issued a warning about the danger comparison between societies of differing social density; this applies equally to per capita data that fail to address the in the sheer scale of production.

Poorly the figures in Table 5 may be, the ‘economic’ power of Huli within their local though the figure listed for Etoro per tion is comparable to the ratios for scales of production are vastly different, affinal networks through which pigs particular projects. Etoro production thing of its nature to the proximity communities. Kelly (1988) estimates that all Etoro shoats were gained in trade with balanced against the export from Etoro of adult pigs. The net result, effectively, long-term and relatively inexpensive agistment percentage of Huli pigs. Mineral oil (yole or mbagua yole) tinguished from tree oil (mbagua) significance of the distinction for Huli than this. Natural seeps of mineral oil locations in and around Huli territory, surface ex- pressions of the oil and gas-condensate reservoirs of the Papuan Fold Belt currently being exploited by petroleum companies (Hill 1991). A number of these seeps (the locations of which are shown in Figure 2) were tapped by the Huli clans on whose territory they occurred. The most famous of these sources were the seeps in the high-altitude Lebani valley owned by three clans, Garua, Ligabi and Humiya. Access to the sources was closely controlled by the landowners, who immersed cane grass in the seeps and drained the soaked oil into bamboo or gourd (Lagenaria siceraria; Huli: mbagua) containers. Containers of oil were traded through the full extent of Huli territory, as well as to the Duna. An interesting comment, frequently recorded in the region (Sillitoe 1979:41, Mangi 1988:60), is that tree oil was universally preferred to mineral oil for the purposes of self-decoration, as the latter exuded a power- ful and unpleasant smell and lacked the ability of tree oil to induce a sheen on the skin. Moreover, Huli did not enjoy a monopoly on mineral oil seeps which are known to occur on the territories of Wola (west Mendi) (Sillitoe 1988), Bimin (Poole 1986), Bedamini, western Enga, Ipili and Febi. Yet it is mineral oil, as the use of the term yole for both the oil and the practice of trade implies, that provided the foundation of the logic of Huli relations with their neighbours.

A range of other minor trade items were produced by Huli. Of these the most important in trade terms was probably red ochre. Mangi (1988:66-71) has described the sources and the process of ochre production along the Doma River on the fans extending from the western slopes of Mt. Ambua. Wrapped sticks of ochre from these sources were traded to all of the immediate neighbouring language communities, and beyond them to the Bimin, Oksapmin and Bogaia (N.Modjeska pers.comm.). Other trade items consisted largely of artefacts, such as arrows, men’s string bags (nu), string aprons (dambale), woven legbands (ge hala), armbands (gi pagida) and headbands (heda) and women’s skirts (hura), together with dog canine teeth. Certain crops which are altitudinally restric- ted, such as marita pandanus (Pandanus conoideus) and karuka pandanus (P. julianettii and P. brosimos), or which are held to grow particularly well in certain soils, such as tobacco and sugarcane, are still important both within and even marginally beyond Huli territory as items of trade.

The principal routes for trade and the broad flow of the major items of trade are illustrated in Figure 4. I suggested earlier that the nature of Huli social structure and patterns of residence enabled individual Huli to travel distances that were unusual within the Highlands. This was also reflected in the exceptional range covered by Huli traders. In 1936, while passing from Kaluli to Onabasulu territory, Ivan Champion, the government officer leading the exploratory Bamu-Purari patrol, encountered a party of seven wig-wearing traders carrying large bundles of tobacco to trade for shell at a ‘trading centre’, possibly on the southeast slopes of Mt. Bosavi (Champion 1940). It was not until 1938, walking out from the government patrol post at Lake Kutubu, that he met more ‘wigmen’ and was able to identify them as Huli (Champion and Adamson 1937/38:4). The following year, in conversation with his Huli guide, Champion determined that Huli traded as far south as Lake Campbell (Figure 3), to the southwest of Mt. Bosavi (Champion and Timperley 1939/40:3). In other areas, Huli are known to have travelled prior to contact as far as the Strickland River to the northwest (N. Modjeska pers. comm.) and the Pipitaka and Yandapo salt sources to the northeast (Mangi 1988:34-35), Porgera and Paiela to the north and Lake Kutubu and the lower Wage valley to the southeast; direct distances of up to 70km from the borders of Huli territory.

Neither Hughes’ (1973:109) conclusion on the limited distances of Ί2 to 15 miles’ travelled by Highlands traders prior to contact nor his observation on the tendency for traders to be moving uphill rather than downhill around the margins of the Highlands appear to hold in the case of Huli traders (Ernst 1984:89). This in no way invalidates the results from Hughe’s work in the central region of the Papua New Guinea Highlands, but it does suggest that the mobility enjoyed by many individuals (and not exclusively men) within Huli territory, a function of the flexible nature of Huli social structure and residence, extended to the surrounding region.

This exceptional personal range raises an interesting questions: what guaranteed the security of Huli traders beyond Huli territory (which is to say, beyond the extent of their affinal and kin networks?) The implicit thereat of retribution from those same extensive kin, a threat visited upon the Etoro on the occasion (Kelly 1988:131-132, obviously played some part, but I hope to show that Huli also successfully fostered for themselves a reputation or a role that assisted in guaranteeing their safety abroad.

Another of Hughes’ general statements on Highlands trade also deserves closer examination: this is his qualification of the term “middlemen” in describing Highlands trade. Quite rightly, Hughes points out that trade along specific routes might more accurately be described as a series of ‘chainlike’ transactions in which traders were usually also consumers, passing on only that proportion of the imported items which they considered surplus to their own needs (1973:111). Certainly Huli traders imported and exported only those types of materials which they themselves consumed. Given the diversity of materials traded and the complexity of the chains of transaction within and between communities of the Highlands region, it would be hard to identify any community that did not, at some point, serve to link other communities in the passage of materials.

Yet, while Huli cannot thus be described as true ‘middlemen’ traders, there is a real sense, articulated by Huli, of the distinct benefits to be gained from ‘playing’ the gradients of value in the different trade items to their own advantage. More detailed analysis of this scope for ‘profit’ must await the publication of Mangi’s (1988) study of Huli trade in which he documents in considerable detail the range in exchange values of trade items along the borders with obena and duguba. Anecdotal evidence is sufficient, however, to support the impression that Huli, and certain of their neighbours, were acutely aware of the Value’ of yole trade. In 1991, Hai Hinawai, an older Huli man at Komo on the southern fringes of Huli territory, described to me his basic strategy in yole trade. On several occasions in the 1930s and 1940s, needing pigs for local exchange purposes, he travelled to a series of Etoro settle- ments, trading worn stone axe blades, string bags, string aprons and bundles of tobacco for mbagua tree oil. For each of these items, he received a bamboo internode of oil. He then traded the oil to central Huli in exchange for cowrie shells, with which he bought pigs at Komo on his return. At other times, he travelled with clan relatives resident in northern Huli to the obena salt sources. A full package or ’round’ of Ipili salt, weighing approximately 4kg, would then be split into as many as eight smaller packets, each of which he would then exchange with southern Huli or duguba for an item similar or equivalent to that with which he originally acquired the full round of salt; Kelly (1988:130), for example, records that Etoro returned fully matured pigs to the Huli in exchange for ‘a fist-sized bundle of salt’, corresponding to one of Hai’s smaller packets. That duguba communities were aware of the extent of their trade dependency on Huli, at least after contact, is recorded by Ernst, who describes a deliberate Onabasulu attempt in 1970 to bypass the Komo Huli and trade direct with northern Huli and western Enga at Margarima (Ernst 1984:91).

As Hai’s account suggests, the principal objectives of individual Huli traders were to parlay the cumulative benefits of trading in scarce materials into the potential for increase in their pig herds. This is further suggested by the wide-spread employment of sows as the standard reference in Huli reckoning of value; an issue barely addressed here, but one that begs for attention from archaeologists, is the historical process, documented in the Wahgi valley, of conversion from pigs to shells as the standard for reckoning wealth (Feil 1982). The longer- term goals of Huli strategy in exploiting the inequalities of resource distribution thus had to do largely with local projects: the raising of pigs for compensation, brideprice and ritual performance. On a regional scale, it might therefore appear that the centrality of Huli within local trade networks owed much to the sheer size of the Huli population and the corresponding scale of Huli wealth in, and demand for, pigs.

As the distribution of trade routes in Figure 4 suggests, Huli territory straddled a significant bottleneck in the chain of valleys extending roughly along an east-west axis through the Highlands. However, equally distant from both the eastern and western axe sources, there was little benefit to be gained by Huli from trading either material on in the opposite direction. The more significant trade axis for Huli was the north-south flow between obena and duguba, mane upriver and wabi downriver, which cut across ecological zones and placed Huli between lowlanders seeking salt and axes, and highlander demand for tree oil and blackpalm. The presence of Huli at Margarima, intersecting the trade routes running between north and south along the Wage valley, near the duguba at Komo in the south and at Paijaka near the Ipili obena was critical to the control of this flow between south and north. It is interesting, in this light, to note that all three areas appear to have been occupied in significant numbers by Huli only after the introduction of sweet potato during the last two or three centuries. I have argued else- where (Ballard 1994) that these migrations reflected the pressures of an expanding population within the central Huli basins, but the possibility that the precise direction of these migrations also served to compound Huli control over regional trade routes should be considered. A fourth major Huli migration of this period is equally revealing: the permanent settlement of the Lebani valley, home to the major mineral oil seeps, around which Huli sacred geography is conceptually deployed.

Huli sacred geography

The significance of mineral oil for Huli, like that of pearl- shell for Wiru (Clark 1991), derived not from its material properties alone but also from its role within a broader view of the world and its cosmological underpinnings. Oil seeps were engaged as critical nodes within Huli sacred geography, in which the oil itself represented a substance crucial to the survival of the Huli universe. The notion of ‘sacred geography’ employed by ethnographers of the Huli (Frankel 1986, Goldman 1983) has been developed within Mayan ethnography to address the means by which Mayan cosmology is formally encoded in the land- scape (Gossen 1974). Mayan ritual sites, on this view, are interpreted as the geographical points of access between distinct human and sacred universes. The co- ordination of ritual over larger areas links these in- dividual sites to form a regional sacred geography. While there is no suggestion in Huli discourse of a distinction between sacred and profane geographies, I continue to use the term ‘sacred geography’ to draw attention to the possibility of perspectives that lie outside the conven- tional oppositions between the domains of social and subsistence, or economic and cultural, and to the layers of meaning implicit in a Huli landscape that are not immedi- ately accessible to external observers. Although most Huli now profess Christianity and most of the former rituals are no longer performed, the extent to which this sacred geography informed the ways in which Huli conceived of their environment is reflected in the continu- ing knowledge of at least its superficial characteristics.

The signal feature of Huli sacred geography was dindi pongone, literally the knot or the root of the earth (Gold- man 1983:112, Frankel 1986:190- The term refers equally to the presence of a sacred and largely subterranean land- scape, and to the body of related myths and spells that constitute knowledge of this landscape. Dindi pongone is described as a root or vine which runs beneath the earth, composed of an intertwined python and cane, both bound around a fluid core of latent flame (of which one manifestation is mineral oil) and capped by a layer of stone. At a number of points, this root rises close to the surface of the earth, where the fluid core and the stone cap become visible as rivers and as gebeanda ritual sites respectively. Gebeanda (literally ‘ancestor house’) ritual sites are common to all clan territories, and were formerly the focus of fertility rituals which sought the assistance of ancestral spirits. The major gebeanda sites associated with the root of the earth have assumed a considerable regional significance and their locations are widely known. Figure 5 maps the positions of these major ritual sites and traces the paths followed between them by the root of the earth. The surface expression of these paths, in the form of a sacred river, is said to follow the beds of the visible rivers along which it runs, usually in an upriver direction. The root, in two parallel strands, is thus held to flow upstream from the tail of the python at the ritual sites of Bebealia Puni and Malaiya in the lowland south, to its head at the upland ritual sites of Tuandaga and Mbibi Baite in the north.

Rituals associated with the root of the earth, known as dindi gamu (‘earth spells’), expanded upon the theme of fertility familiar from rituals performed at a clan or even individual scale (Frankel 1986:154ff). In essence, dindi gamu rituals addressed the fertility of the entire universe and were performed at irregular intervals by the custodians at each site during periods of major regional stress, such as famine or earthquake. Climatic, geological and even social instability were all held to be instances of the process of entropy, of the continual decline and dissipation of fertile substance. Dindi gamu rituals took as their ‘charters’ a variety of myths in which the protagonists undertake long journeys along the tracks of the root of the earth, emplacing features and establishing codes for behaviour. The performance of dindi gamu rituals then sought to recreate certain features of these myths: Bedamini and Onabasulu ritual leaders travelled from the southern sites to the central Huli sites of Gelote and Bebenite, carrying ritually prescribed items available only in the lowlands such as sago and sago grubs, kangaroos (Dendrolagus matschiei spadixl), bandicoots (Echymipera kalubu), rope vines, various fish, wood from certain tree species and water from specific rivers and lakes. In the sacred enclosure at the Gelote ritual site, the assembled ritual leaders would then subsist on these lowland foods for the duration of the ritual performance, imitating the actions of the founding male and female ancestors. Gelote was also the focus of a perhaps less significant dindi pongone route from the direction of the Duna (not shown in Figure 5), along which western Huli, Duna and Oksapmin participants travelled bringing pigs, Nassa shells, karuka pandanus nuts and leaves, possums (Phalanger gymnotis), stone flakes, bamboo implements and cobbles and blue clay from named sources. Sub- sequently, the ritual leaders at Gelote and Bebenite would conduct expeditions to the northern sites of Tuandaga and Tai Yundiga.

While the quantities in which these items were assem- bled for ritual performances appear to have been relatively small within the context of overall trade, the symbolic importance of the directions in which they were moved, and their presence as tokens of the ability of Huli ritual leaders to mobilize and co-ordinate production over a massive area are critical to an understanding of the relationship between regional trade and sacred geography. The role of mineral oil in dindi gamu rituals provides a key to understanding the second of these points. Containers of mineral oil from seeps in the Lebani valley in the west and from the Bebealia Puni source to the south were brought to Gelote and Bebenite where they were poured over fires said to emanate from the root of the earth. The explicit function of this and other associated rites was the replenishment with mineral oil of the dindi pongone root as the fertile core of the universe. In its function of maintaining the earth’s fertility and in ascribing to Huli ritual leaders a central role in that function, dindi gamu placed Huli at the centre of the universe. Huli are not alone, of course, in representing themselves at the centre of their known universe, but they are unusual in terms of the extent to which they promul- gated this view of the world amongst their neighbours. Like Huli traders, ritual leaders from Gelote and Bebenite travelled widely, from Bebealia Puni to Tuandaga and Tai Yundiga, as far into Duna territory as the Strickland river (N.Modjeska pers.comm.) and, on at least one occasion in the 1920s, to the oil seeps in the Lebani valley, where they instructed the Garua landowners on the role of the seeps within the cosmological structure of dindi pongone. Amongst some Duna, at least, the centrality of the Huli in the project of maintaining the universe was taken to heart: Haley (1993) describes Duna perceptions of Huli not only as fabulously wealthy but also as the true holders of dindi gamu knowledge and as the controllers of the apocalypse of mbingi. Although the fertility rituals of dindi gamu were regularly required and performed on a variety of scales, from the local clan level through to the region-wide performances sketched above, they were regarded ultimately as a stop-gap measure in combatting the decline of the universe. Total renewal of the land’s fertility and the restoration of full mana knowledge to people required a more drastic measure altogether, and one over which Huli ritual leaders themselves felt they had only a partial control. This event was known to Huli as mbingi, literally the ‘time of darkness’, who drew for its description upon ancient memories of a volcanic tephra fall during the seventeenth century (Blong 1982).

This historic fall of ash was regarded as a beneficial event which restored to the soil its full fertility. The longer-term goal of dindi gamu, as it was conceived by Huli ritual leaders, was to induce a controlled recurrence of mbingi and obviate the necessity for perpetual acts of incremental restoration of fertility through ritual (Frankel 1986:22, Ballard 1992). Ambivalence about the prospect of mbingi stemmed from the threat of total annihiliation mbingi summoned not through the correct performance of dindi gamu by ritual leaders but by a failure in regional social order.

If the central role accorded to Huli ritual leaders in maintaining the fertility of the universe accounted for their ability to co-ordinate rituals on a regional scale, an external perspective suggests that the structure of that regional framework reflected the broad nature of Huli interests. Given the range of connections that might have been traced between different oil seeps and gebeanda ritual sites, the primarily north-south orientation of the major roots of the earth would appear to relate to the pattern of flow in items of regional trade. Specific injunctions issued from the sites of Gelote and Bebenite identified the separation of northern obena and southern duguba, of the head and tail of the dindi pongone snake, as a crucial component in the Huli burden of maintaining the universe. In a widely known Huli adage, the import of which is known at least to Ipili and to northern Bedamini, the prospect of an uncontrolled mbingi is still posed as the consequence of obena and duguba meeting directly without the mediation of Huli:

The land will turn red, the mountains will turn red, the people will turn red. Fire will destroy everything, for these two [obena and duguba] have been placed at either end of the root of the earth (Ballard, tape transcripts).

Huli ritual leaders thus extended to their culturally unrelated neighbours to the north and south a sacred geography, a vision of the landscape and its cosmological foundations, that enshrined the central position of Huli in the regional circulation of ideas and materials, and that ascribed to Huli the role of mediators, or dombenialU on behalf of the universe at large. Where an external perspective might suggest that the position of the Huli basins as a centre for regional trade simply reflected the sheer weight of numbers of Huli and the ability of individual Huli to coordinate labour and wealth on a scale denied to individuals of neighbouring groups, Huli cosmology identified the relationship between Huli, the root of the earth and the sources of mineral oil as the basis for their central position. Mineral oil, in all other respects a substance of minimal value in the regional circulation of trade materials, was thus refigured through Huli cosmology and represented abroad as the key to both the continued well-being of the universe and the regional prominence of the Huli. From an external perspective again, the apocalypse envisaged by the Huli was perhaps the prospect of an eclipse of their central position in the region.

Things fall apart Things fall apart; the centre cannot hold; Mere anarchy is loosed upon the world . . . (Yeats, The Second Coming)

Thus far, I have focused on a particular cultural land- scape largely as it might have appeared in the 1920s, prior to contact.

This has been necessary in order to establish the relationship between the regional trade network and the structure of Huli sacred geography. While neither the trade network nor Huli sacred geography can be said ever to have been static, the changes that followed upon contact with the colonial administration in the 1930s swiftly and fundamentally transformed both landscapes’. Though small numbers of steel tools were being traded inland from the coastline of the Papuan Gulf to the south during the 1920s (Crittenden 1982:2190, they did not reach the Papuan Plateau until the early 1930s (Scheiffelin 1991:67) and were totally unknown to those Huli who first saw them carried by the Fox brothers in November 1934 (Allen and Frankel 1991). A brief efflorescence of trade in pearlshells and steel goods into limited parts of the Highlands along the existing trade routes from the south during the 1920s and early 1930s was reversed following the establishment by the colonial administration of base camps at Mt Hagen in the Wahgi valley in 1933 and at Lake Kutubu in 1936. Hughes (1978) has described the dramatic effects upon central Highlands trade systems to the east of the Huli of the use by Euro- pean miners, missionaries and administrators of steel and shell in payment for labour and land, in which the massive influx of pearlshell from 1933, flown into Mt Hagen in crates, led swiftly to a localised devaluation of pearlshell as the dominant item of wealth.

Some of this material reached the Huli during the late 1930s and early 1940s, either directly as payment for assistance and food, from the handful of administration patrols that passed through Huli territory during that period or along the existing regional trade routes from the east. But the volume of this trade in colonial goods appears to have been small. Certainly steel axes were not widely available amongst Huli until the establishment of the permanent administrative post at Tari in 1952 and fringe Huli communities and Huli neighbours to the south and west were still described as poor in steel axes into the 1960s. The choice of Tari as a regional base for the colonial administration was an obvious one both for the administration, given its suitability for an airstrip and the density of the local population and, one suspects, for the Huli, for whom the selection of Tari as the primary local conduit for the new trade goods, and of the Huli as carriers and interpreters for government patrols out of Tari seemingly confirmed the cosmologically ordained centrality of the Huli. Two major developments in regional trade during the colonial period between 1952 and 1972 were local reversals in the existing flow of material and the exploitation by Huli and other Highlands traders of conditions under the Pax Australiana to trade directly with communities at the source of desired trade items. Shell and steel trade goods flown into Tari were circulated to Duna, Oksapmin, Hewa and the Papuan Plateau communities. This novel movement of steel tools and shells out of the Highlands and down to the Papuan Plateau completely refigured the trade landscape within the latter region. Where the Etoro had formerly received axes largely from the Honibo and trans-Strickland sources via the Bedamini to their west, their needs were now met by trade with Huli; more importantly for the Etoro and Onabasulu, a longterm process of conversion from dog’s teeth to pearlshell as the primary valuable used locally in bridewealth, compen- sation and mortuary exchanges saw the Huli emerge as the pre-eminent source of these vital goods (Kelly 1977: 11,13, Ernst 1984:92). The administration prohibition on warfare also enabled Huli traders literally to bypass the Etoro, Onabasulu and Kaluli and trade direct with southern Bedamini and Kasua for such items as casso- wary bone and feathers and tree oil.

But if central Huli in the immediate Tari area briefly enjoyed an enhanced colonial role as traders, mediat- ing new wealth and ideas, their more recent, post- Independence fortunes have appeared increasingly volatile. The 1988 alluvial goldrush at Mt. Kare near the Tai Yundiga ritual site, the opening of the goldmine at Porgera in 1990 and the discovery of oil and gas at Lake Kutubu and at the Hides well west of Komo have ushered in a new resource geography ringing Huli territory, a novel trade landscape which Huli have interpreted in the light of their sacred geography (Clark 1993, Ballard 1992), but over which they enjoy none of their former powers of interpretation, mediation and control. A new line of power, a snake of electricity pylons, extends along the track of the root of the earth, from the generators at the Hides well near the Bebealia Puni ritual site, up past the ritual site of Tai Yundiga and down to the mine at Porgerà. Though they ‘eat’ the rent paid for the pylons, Huli are keenly aware that the means to wealth on an altogether grander scale is passing over their heads. The head and tail of the python are now linked and obena and duguba mix freely amongst the labour forces at the new sites of power. If the universe itself has not ended, the landscapes of trade and ritual which positioned the Huli at its centre have been irrevocably transformed. Archaeological implications By employing ethnographic methods to address the broader contexts for the social landscapes of trade and ritual, I have tried to suggest that Huli-speakers have historically sustained their central role in a regional system of trade by exporting to their neighbours a particular way of looking at the land. The promise of a social landscape approach lies in the scope it offers for working with a series of overlapping constructs, different landscapes of meaning that address a variety of perspectives. Thus the juxtaposition of a conventional resource geography with Huli sacred geography provides insights into both the cosmological embeddedness of trade and exchange, and the strategic interests at play in the regional organisation of ritual. Neither perspective, shorn of the other, so enriches our understanding of the complexity of the social processes involved. This, in turn, renders problematic the separation of Huli economy and ritual. When Jones and White (1988:84) conclude that an archaeological approach to the Ngilipitji quarry that was ignorant and unenquiring of the social contexts for the site could yield only ‘an impoverished history’, I take them to be referring to a conventional archaeology wedded to an ecological model of trade. How then are we to bring the insights of an archaeological ethnography to bear on the design of archaeological research? It should be stressed that an archaeological or historical approach to ethnography that employs oral history, myth and genealogies is an important form of enquiry in its own right. Nevertheless, the possibility of extending to the deeper past insights developed through an understanding of the ethnographic past hinges upon our ability as archaeologists to model the complexity of social process Visible’ ethnographically at a scale and in such a relation to material residues as to be accessible to conventional archaeological methods.

European archaeologists, endowed with an enviable wealth of archaeological evidence, have recently begun to address questions of social process by focusing on ‘ritual landscapes’ (Barrett 1994, Patton 1993, Thomas 1993, Whitehouse 1992). A common feature of much of this work is an emphasis on the relationship between the development of fixed, often monumental ritual sites and fluctuations in the circulation of exotic trade items such as stone axe blades. There is no prospect in the immediate future of the breadth of archaeological material necessary for such interpretation becoming available in the Papua New Guinea Highlands. Nor, on the evidence presented here, could we expect a focus on durable trade items to provide more than the barest indication of the process of regional trade, let alone the relationships between trade and ritual, documented ethnographically. Of the various material items of trade described above, only small quantities of shell, bone, ochre, and axe blades or blade fragments have been recovered from excavations in the Huli, Duna and Ipili regions (White 1974, Mangi 1988, Ballard 1994). On this evidence alone, the Huli region would appear relatively impoverished and any sense of the position of Huli-speakers and the Huli basins in the regional circulation of materials and ideas would be lost too.

Two other means of direct archaeological access to the social landscapes described ethnographically include excavation at the central locations of trade and ritual: the fixed sources for materials such as axe blades, salt and ochre, and the gebeanda ritual sites along the root of the earth. Burton’s (1984) excavations at the Wahgi and Jimi valley axe quarries have already demonstrated the value of such a direct approach for establishing both a chronology of production and the sense of regional development implied in the initiation of quarrying and in changes quarry output. Studies of ochre and salt production which emulated Burton’s strategy would be of consider- able value in providing some idea of the temporal depth of regional trade in these materials, though probably less rewarding than the analysis of stone quarries due to the ephemeral nature of the production processes involved. The sensitivity still surrounding the gebeanda centres virtually precludes the possibility of excavations at these sites; the more pressing need is to map the major gebeanda while those former ritual officiants who are still alive can accurately locate and describe the functions of these sites. The various direct approaches to past trade and ritual landscapes of the Huli region thus offer no immediately practicable means of access.

One possible solution has already been identified by White (1985), in addressing the problem of the genesis in the Highlands region of social formations characterised by ‘Big-Man’ leadership. Finding it difficult enough to distinguish Big-Men in strictly material terms in the ethnographic record, White concluded that there could be little scope for their direct identification in the archaeo- logical record. Instead, drawing his inspiration from David Clarke, White proposed that the ‘greater predict- ability at more inclusive levels of analysis’ (1985:59) might allow the identification of broader changes indicating the emergence of social complexity on a scale comparable to that associated ethnographically with Big-Men. I con- clude with no more than a brief sketch of one possible solution along these lines that might provide archaeo- logical access to the issues raised in this paper.

Throughout the ethnographic descriptions of trade and ritual presented here, pigs appear as a significant material nexus between Huli cosmology and the landscapes of trade and ritual performance: as the primary valuables in all Huli exchange, as the mediatory ‘currency’ through which the benefits derived from regional trade were redirected towards local strategic goals, and as the principal material trace of the historic development from ritual exchange with dead ancestors to ceremonial exchange with living rivals. Again, there is little prospect of identifying changes in the practice of pig production through such direct means as the excavation of pig bone from cave or rocksheiter sites. However, in the Huli region, the management of increasing numbers of pigs within the most productive of local agricultural environ- ments produced a response that has left a formidable archaeological signature: deeply ditched pig droveways that transect the basins, enabling the movement of pig herds between their stalls and forage grounds while securing the surrounding gardens from unwelcome ingress (Ballard 1994). The excavation and dating of these features is potentially the most rewarding means of providing a chronological understanding of the intensification of Huli pig production, linking the landscapes of trade and ritual with the historical development of Huli sacred geography.

Acknowledgements

I am indebted to the people of the Southern Highlands and Enga Provinces of Papua New Guinea who have worked with me in the course of my research, and in particular to Aluya Mabira and David Hawiya. I thank the former Department of Prehistory, Research School of Pacific and Asian Studies, Australian National University for financial support. I am grateful also to the Provincial Government of Southern Highlands Province, the National Research Institute of Papua New Guinea, the National Museum and Art Gallery of Papua New Guinea, and the University of Papua New Guinea for their per- mission and support for my research. In the course of preparing this paper I have benefited greatly from the advice and assistance of Brigid Ballard, Jeffrey Clark, Isabel McBryde, Paul Sillitoe, Polly Wiessner and the editors of this issue, none of whom are responsible for its final form. The insight and comments of Robin Hide have been particularly valuable. I would also like to thank Nicole Haley, Frances Ingemann, Isabel McBryde, Jo Mangi, Nie Modjeska and Paul Sillitoe for their permission to cite unpublished material.

(Reprinted with permission of Dr. Chris Ballard, The centre cannot hold. Trade networks and sacred geography in the Papua New Guinea Highlands. Archaeol. Oceania 29 (1994) 130-148.)